Cambridge Studies in Speech Science and Communication

Advisory Editorial Board J. Laver (Executive editor) A. J. Fourcin J. Gilbert M. Haggard

P. Ladefoged B. Lindblom J. C. Marshall

Patterns of sounds
In this series:

The phonetic bases of speaker recognition Francis Nolan

Patterns of sounds

Ian Maddieson
University of California at Los Angeles

With a chapter contributed by Sandra Ferrari Disner

The right of the

University of Cambridge
to print and sell
all manner of books
was granted by
Henry VIII in 1534.
The University has printed
and published continuously
since 1584.

Cambridge University Press

Cambridge
London New York New Rochelle
Melbourne Sydney
CAMBRIDGE UNIVERSITY PRESS
Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, Sao Paulo, Delhi

Cambridge University Press

The Edinburgh Building, Cambridge CB2 8RU, UK

Published in the United States of America by Cambridge University Press, New York

www. Cambridge. org

Information on this title: www.cambridge.org/9780521113267
© Cambridge University Press 1984

This publication is in copyright. Subject to statutory exception

and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.

First published 1984

This digitally printed version 2009

A catalogue recordfor this publication is available from the British Library

Library of Congress Catalogue Card Number: 84—7753

ISBN 978-0-521-26536-2 hardback
ISBN 978-0-521-11326-7 paperback
Contents

Preface 1

1 The size and structure of phonological inventories 5

1.1 Introduction 5
1.2 Design of the database 5
1.3 Variations in inventory size 7
1.4 Relationship between size and structure 9
1.5 Phonetic salience and the structure of inventories 13
1.6 Compensation in inventory structure 17
1.7 Segments and suprasegmentals 20
1.8 Segment inventories and syllable inventories 21
1.9 Conclusions 23
Notes to Chapter 1 23
References 24

2 StQ
P s and affricates 25
2.1 Introduction 25
2.2 Stop series 25
2.3 Summary of analysis of stop systems 31
2.4 Stop systems by place 31
2.5 Stop and affricate places 34
2.6 Voicing and place of articulation for plosives 34
2.7 Secondary articulations with plosives 37
2.8 Affricates 38
2.9 Summary of generalizations on stops and affricates 39
Notes to Chapter 2 40
References 40
 3 Fricatives 41
3.1 Introduction 41
3.2 The occurrence of fricatives 42
3.3 Number of fricatives per language 43
3.4 Implications of voicing in fricatives 47
3.5 Gamkrelidze1s implicational proposals 49
3.6 Predicting frequency from intensity 49
3.7 Estimates of perceptual salience of fricatives 51
3.8 The structure of systems of fricatives 52
3.9 The phoneme /h/ 57
3.10 Conclusion 57
References 58

4 Nasals 59
4.1 Introduction 59
4.2 Types of nasals 59
4.3 Ferguson's "Assumptions about nasals": primary nasal consonants 61
4.4 Primary nasals and obstruents 64
4.5 Secondary nasal consonants 65
4.6 Restated generalizations about nasals 69
4.7 Explanations for nasal patterns 70
References 71

5 Liquids 73
5.1 Introduction 73
5.2 Overall frequency of liquids 73
5.3 Laterals 74
5.4 R-sounds 78
5.5 Structure of liquid systems 82
5.6 Generalizations on the structure of liquid systems 87
5.7 Conclusion 89
Notes to Chapter 5 89
References 90
 6 Vocoid approximants 91
6.1 Introduction 91
6.2 Frequency of vocoid approximants 91
6.3 Approximants and related vowels 94
6.4 Approximants and related consonants 95
6.5 Other approximants 96
6.6 Summary 97
Notes to Chapter 6 97
References 97

7 Glottalic and laryngealized consonants 98

7.1 Introduction 98
7.2 Glottalic and laryngealized sounds 99
7.3 Ejectives 101
7.4 Voiceless laryngealized segments 110
7.5 Implosives and voiced laryngealized plosives 111
7.6 Languages with both ejective stops and implosives 115
7.7 Laryngealized sonorants 115
7.8 Diachronic implications 117
7.9 Phonetic explanations for the structure of glottalic systems 117
7.10 Summary of generalizations 120
Notes to Chapter 7 121
References 121

8 Vowels 123
8.1 Introduction 123
8.2 Types of vowels 123
8.3 Number of vowels per language 126
8.4 Distinctive vowel qualities 127
8.5 Properties of vowel series 128
8.6 Diphthongs 133
8.7 Summary 134
Notes to Chapter 8 135
References 135

vii
 9 Insights on vowel spacing 136
9.1 Introduction 136
9.2 Preliminaries 136
9.3 Method 139
9.4 Analysis of defective systems 141
9.5 Conclusions 153
Notes to Chapter 9 154
References 155

10 The design o£ the UCLA Phonological Segment Inventory

Database (UPSID) 156
10.1 Introduction 156
10.2 Selection of languages for UPSID 157
10.3 Determining the inventories 160
10.4 Indices and variables 163
10.5 Using UPSID 171
References 173

Appendix .A: Language lists and bibliography of data sources 174

1. Genetic listing of languages and outline classification 174
2. Alphabetic listing of languages with key to sources 177
3. Bibliography of data sources 183

Appendix 15: Phoneme charts and segment index for UPSID languages 200
The charts 200
The index 202
Vowel symbols and diacritics used in phoneme charts 204
Segment index 205
Phoneme charts 263
This book is dedicated to the memory of Lilian Ann Maddieson and
Henry Ray Maddieson, who gave me the freedom to go my own way.

ix
Preface

The discovery of generalizations concerning the content and structure of

phonological inventories has been a significant objective of recent work in
linguistics. Such generalizations have been taken into account, explicitly
or implicitly, in the formulation of phonological theories, in evaluating
competing historical reconstructions, in constructing models of language
change and language acquisition, and they have stimulated important
linguistically-oriented phonetic research. This book reports on the work
done at UCLA using a computer-accessible database containing the
phonological segment inventories of a representative sample of the world's
languages which is designed to provide a reliable basis for such
generalizations. The project has come to be referred to by the acronym
UPSID - the UCLA Phonological Segment Inventory Database.
There seem to be three types of sources for observations on
phonological inventories. The type with the longest tradition is an
essentially impressionistic account based on a linguist's experience of a
number of languages. Statements by Trubetskoy (1939), Jakobson and Halle
(1956), and Ladefoged (1971) as well as incidental remarks in the papers of
numerous authors are examples of this category. Although they may be based
on familiarity with a very large number of languages, there is some doubt
about the scope and validity of the conclusions reached, since the list of
languages represented in this experience is not given and there is no
quantification attached to the statements made.
The second type consists of explicit samples of languages compiled for
the purpose of a single study, such as Ferguson (1963), Greenberg (1970)
and Hyman (1977) on nasals, glottalic consonants and stress respectively.
In these cases the quality of the sample (cf. Bell 1978) and the
significance of the conclusions reached (cf. Hurford 1977) can be
independently assessed by the reader.
Preface

The third kind of data source is a standardized multi-purpose survey,

epitomized by the Stanford Phonology Archive (SPA), compiled at Stanford
University as part of the broad Language Universals Project under the
direction of J. H. Greenberg and C. A. Ferguson. A large proportion of
recent work on phonological universals is either directly based on the SPA
or owes an indirect debt to it. The UCLA Phonological Segment Inventory
Database (UPSID) is a source of this third kind.
There are several reasons for the superiority of this third kind of
data source which arise from the nature of the field of enquiry involved.
The data source serves, first, to generate observations, e.g. observations
concerning the frequency of segments of different types and of the phonetic
attributes of segments, as well as their co-occurrence in phonological
inventories, and secondly, to subject hypotheses concerning such matters as
segment frequency to the test of comparison with empirical observations.
The hypotheses may range from simple ones claiming that there are
significant differences in the frequency of segments of different types to
more elaborate ones positing contingent relationships between the
occurrence of (sets of) different segments, or limitations on the
distribution of phonetic attributes within inventories. The third, and
perhaps most significant, purpose behind compilation of such data sources
is as a stimulus to the generation of hypotheses which relate to other
fields of the study of language but for which such matters as segment
frequencies, inventory size, and so on, may be the point of departure. Such
hypotheses can be directed at issues of production, perception,
acquisition, linguistic change or language contact, but establish
connections between other data and observations concerning segments and
inventories.

Most of these observations and hypotheses about phonological universals

necessarily concern relative rather than absolute matters. Experience has
shown that few interesting things are to be said about phonological
inventories that are truly universal, i.e. exceptionless. Apart from
observations such as "all languages have a contrast between consonants and
vowels" most of the substantive generalizations concerning segments and
inventories are or can be expected to be of the form "a situation x occurs
more (or less) frequently than chance leads us to predict." That is, in
layman's terms, they are statistical observations. They can therefore only
be meaningful if they are drawn from, or tested with respect to, a body of
data appropriately designed for statistical analysis. In other words, one
 Patterns of sounds

which is representative, extensive and uniform in analysis as far as

possible. This requires establishment of a large and appropriately selected
sample of languages and a standardized procedure for interpreting their
phonologies. Once such a database has been established, numerous
commensurate studies on the same data can be made.
This book contains nine chapters presenting analyses of aspects of the
UPSID inventories. Chapter 9 is contributed by Sandra F. Disner, the rest
are written by me. Each of these chapters is designed to be largely
self-contained so that readers may consult a single chapter if, for
example, they are interested in some particular segment type. Chapter 10
presents a relatively full account of the design of the database, including
the principles governing the selection of languages, the criteria used in
interpretation of the descriptive sources consulted and the set of phonetic
features used to characterize segments. A full documentation of the data
itself is also contained in the appendices at the end of the book,
including phonemic charts of each language and full lists of the types of
segments that occur. Each language is assigned an identification number
which is cited whenever the language is mentioned in the text, enabling the
corresponding phoneme chart to be easily found. The principles on which the
identification numbers are assigned is explained in Appendix A.
Many people have assisted in making this book possible. The principal
work of establishing the computer database was done by Sandra F. Disner,
Vivian Flores, J. Forrest Fordyce, Jonas N. A. Nartey, Diane G. Ridley,
Vincent van Heuven and myself. Help in collecting data was provided by
Stephen R. Anderson, Peter Austin, Steve Franks, Bonnie Glover, Peter
Ladefoged, Mona Lindau-Webb, Robert Thurman, Alan Timberlake, Anne Wingate,
Andreas Wittenstein and Eric Zee. Additional assistance has come from other
linguists at UCLA and elsewhere. Mel Widawsky provided valuable services in
persuading the computer to accept the indigestible bulk of our input. A
library of the sources from which data was drawn was compiled with
assistance from Hector Javkin and Diane G. Ridley. John Crothers provided
an early copy of the final report of the Stanford Phonology Archive,
enabling the UCLA project to benefit from the experience accrued at
Stanford. Geoffrey Lindsey and Karen Weiss did the tedious work of typing
the phoneme charts and Karen Emmorey, Karen Weiss, Alice Anderton, and
Kristin Precoda assisted with the preparation of the camera-ready copy of
the remainder of the book. To all of these people I owe an enormous debt,
which I can only pay in the coin of gratitude.
Preface

I also owe thanks to those who have shown faith in the UPSID project as
it developed by making use of it, including Louis Goldstein, Pat Keating,
Peter Ladefoged, BjBrn Lindblom and the students in Linguistics 103 at
UCLA.
A considerable portion of the work reported in this book has been
funded by the National Science Foundation through grants BNS 78-07680 and
BNS 80-23110 (Peter Ladefoged, principal investigator). Neither the NSF nor
any of the individuals named above are responsible for the errors that
undoubtedly remain. If you the reader find one, please write and tell me
about it.

Ian Maddieson
University of California
Los Angeles

References
Bell, A. 1978. Language samples. In J.H. Greenberg et al. (eds.) Universals
of Human Language, Vol 1^, Method and Theory. Stanford University Press,
Stanford: 123-56.
Ferguson, C. A. 1963. Some assumptions about nasals. In J.H. Greenberg
(ed.) Universals of Language. MIT Press, Cambridge: A2-7.
Greenberg, J. H. 1970. Some generalizations concerning glottalic
consonants, especially implosives. International Journal of American
Linguistics 36: 123-45.
Hurford, J. R. 1977. The significance of linguistic generalizations.
Language 53: 574-620.
Hyman, L. M. 1977. On the nature of linguistic stress. In L.M. Hyman (ed.)
Studies on Stress and Accent. (Southern California Occasional Papers in
Linguistics 4) University of Southern California, Los Angeles: 37-82.
Jakobson, R. and Halle, M. 1956. Phonology and phonetics. (Part 1 of)
Fundamentals of Language. Mouton, The Hague: 3-51.
Ladefoged, P. 1971. Preliminaries to Linguistic Phonetics. University of
Chicago Press, Chicago.
Trubetskoy, N. 1939. Grundzttge der Phonologie (Travaux du Cercle
Linguistique de Prague 9 ) . Prague.
The size and structure of phonological inventories

1.1 Introduction
A database designed to give more reliable and more readily available
answers to questions concerning the distribution of phonological segments
in the world's languages has been created as part of the research program
of the UCLA Phonetics Laboratory. The database is known formally as the
UCLA Phonological Segment Inventory Database, and for convenience is
referred to by the acronym UPSID. UPSID has been used to investigate a
number of hypothesized phonological universals and "universal tendencies".
Principal among these have been certain ideas concerning the overall size
and structure of the phonological inventories. The design of the database
is briefly described in this chapter. A full description is given in
chapter 10, and the various appendices at the end of the book report on the
data contained in UPSID files. The remainder of the present chapter
discusses the issues involving the overall structure and size of
phonological inventories which have been examined with its use.

1.2 Design of the database

The languages included in UPSID have been chosen to approximate a properly
constructed quota sample on a genetic basis of the world's extant
languages. The quota rule is that only one language may be included from
each small family grouping, for example, among the Germanic languages, one
is included from West Germanic and one from North Germanic (East Germanic,
being extinct and insufficiently documented for a reliable phonological
analysis to be made, is not included). Each such small family grouping
should be represented by the inclusion of one language. Availability and
quality of phonological descriptions are factors in determining which
The size and structure of phonological inventories

language to include from within a group, but such factors as the number of
speakers and the phonological peculiarity of the language are not
considered. The database includes the inventories of 317 languages. In this
and subsequent chapters, every language mentioned in the text is identified
by a number that cross-refers to the list of these languages and the data
charts at the end of the book. These numbers are assigned on the basis of
the genetic affiliation of the language.
In the database each segment which is considered phonemic is
represented by its most characteristic allophone, specified in terms of a
set of 58 phonetic attributes. These are treated as variables which take
the value 1 if the segment has the attribute and 0 if the segment lacks it.
The list of attributes with the value 1 thus provides a phonetic
description of the segment concerned.
For 192 of the 317 languages included, UPSID has profited from the work
of the Stanford Phonology Archive (SPA). Our decisions on phonemic status
and phonetic description do not always coincide with the decisions reached
by the compilers of the SPA, and we have sometimes examined additional
or alternative sources, but a great deal of effort was saved by the
availability of this source of standardized analyses. It should be noted
that UPSID, unlike the SPA, makes no attempt to include information on
allophonic variation, syllable structure, or phonological rules.
In determining the segment inventories, there are two especially
problematical areas. The first involves choosing between a unit or sequence
interpretation of, for example, affricates, prenasalized stops, long
(geminate) consonants and vowels, diphthongs, labialized consonants, etc.
The available evidence which bears on the choice in each language
individually has been examined but with some prejudice in favor of treating
complex phonetic events as sequences (i.e. as combinations of more
elementary units). The second problem area involves the choice between a
segmental and a suprasegmental analysis of certain properties. Stress and
tone have always been treated as suprasegmental; that is, tonal and stress
contrasts do not by themselves add to the number of distinct segments in
the inventory of a language, but if differences in segments are found which
accompany stress or tone differences, these may be regarded as segmental
contrasts if the association does not seem a particularly natural one. For
example, if there is an unstressed vowel which is a little shorter or more
centralized than what can be seen as its stressed counterpart, these vowels
will be treated as variants of the same segment. However, larger
 Patterns of sounds

qualitative differences between the set of stressed and unstressed vowels

will lead us to enter such sets of vowels as separate segments. In all
cases, sets of vowels which are divided into vowel harmony series are all
entered separately; the factor which distinguishes the vowel harmony series
is not extracted as a suprasegmental.

1.3 Variations in inventory size

The number of segments in a language may vary widely. The smallest
inventories included in the survey have only 11 segments (Rotokas, 625;
Mura, 802) and the largest has 141 (!Xu, 918). However, it is clear that
the typical size of an inventory lies between 20 and 37 segments - 70% of
the languages in the survey fall within these limits. The mean number of
segments per language is a little over 31; the median falls between 28 and
29. These values are very close to the number 27 +_ 7 which Hockett (1955)
estimated as the most likely number of segments in a language.
The variability in segment totals can be reflected in a number of
statistical measures. These show that the curve formed by plotting the
number of languages against the segment totals is not normally distributed.
It is both positively skewed and platykurtic, that is, there is a longer
tail to the distribution at the high end of the scale, and the shape of the
curve is one with a low peak and heavy tails. This implies that the mean
number of segments is not a good way to sum up the distribution. For this
reason more attention should be paid to the range 20-37 than the mean of
31.
Whether the tendency to have from 20 to 37 segments means that this is
an optimum range is an open question. It seems likely that there is an
upper limit on the number of segments which can be efficiently
distinguished in speech, and a lower limit set by the minimum number of
segments required to build an adequate vocabulary of distinct morphemes.
But these limits would appear to lie above and below the numbers 37 and 20
respectively.
Consider the following: the Khoisan language !Xu (918) with 141
segments is related to languages which also have unusually large
inventories. Comparative study of these languages (Baucom 1974; Trail1
1978) indicates that large inventories have been a stable feature which has
persisted for a long time in the Khoisan family. If the number of
efficiently distinguished segments was substantially smaller, there would
be constant pressure to reduce the number of segments. There does not seem
to be any evidence of such pressure.1

7
The size and structure of phonological inventories

Similarly, the facts do not seem to show that languages with small
inventories (under 20 segments) suffer from problems due to lack of
contrastive possibilities at the morphemic level. The symptoms of such
difficulties would include unacceptably high incidence of homophony or
unmanageably long morphemes. Dictionaries and vocabularies of several
languages with small inventories, such as Rotokas (625, Firchow, Firchow
and Akoitai 1973), Hawaiian (424, Pukui and Elbert 1965) and Asmat (601,
Voorhoeve 1965: 293-361), do not provide evidence that there are symptoms
of stress of these kinds in languages with small phoneme inventories.
Hawaiian, for example, with 13 segments has been calculated to have an
average of just 3.5 phonemes per morpheme (Pukui and Elbert 1965: xix) ,
clearly not unacceptably long. And again, comparative evidence indicates
that small inventory size may be a phenomenon which persists over time, as,
for example, in the Polynesian language family, which includes Hawaiian
(Grace 1959).

The restrictions on inventory size may therefore not be theoretical

ones relating to message density and channel capacity in language
processing. Although such considerations have been the most widely
discussed, they are far from the only ones likely to influence the typical
language inventory. Linguistic messages do have to be sufficiently varied
to be able to deal with myriad situations and they do need to be
successfully conveyed via a noisy channel, but the design of language is
also subject to many pressures of a "non-functional" kind. Most languages
exist in a multi-lingual social context. Limits may be placed on the size
of a typical inventory through language contact, especially situations
where a language is gaining speakers who are learning the language after
early childhood. The mechanism may be one which approximates the following:
speakers acquiring a new language make substitutions for any segment that
is not matched by a closely similar segment in their own language, or is
not capable of being generated by a simple process of adding familiar
features (e.g. acquiring /g/ is easy if you already have /p, b, t, d / and
/k/ in the first language). The resulting inventory in the acquired
language contains only the segments common to both input languages, plus a
few segments "generated" by the process outlined above. The smaller the
inventory of the first language, the greater the probability that some
segments will be generated in the fashion outlined. The greater the
inventory, the smaller the probability that similar segments will coincide
in the two languages and thus the greater the probability of inventory
simplification.
 Patterns of sounds

This proposal predicts not only that upper and lower limits on
inventory size will tend to be rather flexible, as is the case, but also
that areal-genetic deviations from the central tendency should be expected.
Thus, greater than average size inventories in Khoisan or Caucasian
languages, and smaller than average in Polynesian are understandable
results: local deviations are perpetuated because primary contact is with
other languages tending in the same direction. This proposal also avoids a
difficulty; if human processing limitations are postulated as the cause of
limitations on the size of inventories, then they ought invariably to exert
pressure to conform on the deviant cases. The evidence for this is lacking.

1.4 Relationship between size and structure

The data in UPSID have been used to address the question of the
relationship between the size of an inventory and its membership. The total
number of consonants in an inventory varies between 6 and 95 with a mean of
22.8. The total number of vowels varies between 3 and 46 with a mean of
8.7. The balance between consonants and vowels within an inventory was
calculated by dividing the number of vowels by the number of consonants.
The resulting ratio varies between 0.065 and 1.308 with a mean of 0.402.
The median value of this vowel ratio is about 0.36; in other words, the
typical language has less than half as many vowels as it has consonants.
There are two important trends to observe; larger inventories tend to be
more consonant-dominated, but there is also a tendency for the absolute
number of vowels to be larger in the languages with larger inventories. The
first is shown by the fact that the vowel ratio is inversely correlated
with the number of consonants in an inventory (r = -.40, p = .0001) and the
second by the fact that the total of vowels is positively correlated with
the consonant total (r = .38, p = .0001). However, a large consonant
inventory with a small vowel inventory is certainly possible, as, for
example, in Haida (700: 46C, 3V), Jaqaru (820: 38C, 3V) or Burushaski (915:
38C, 5V). Small consonant inventories with a large number of vowels seem
the least likely to occur (cf. the findings of Hockett 1955), although
there is something of an areal/genetic tendency in this direction in New
Guinea languages such as Pawaian (612: 10C, 12V), Daribi (616: 13C, 10V)
and Fasu (617: 11C, 10V). In these cases a small number of consonants is
combined with a contrast of vowel nasality. Despite some aberrant cases,
however, there is a general though weak association between overall
inventory size and consonant/vowel balance: larger inventories tend to have
a greater proportion of consonants.
The size and structure of phonological inventories

Such an association suggests that inventory size and structure may be

related in other ways as well. A simple form of such a hypothesis would
propose that segment inventories are structured so that the smallest
inventories contain the most frequent segments, and as the size of the
inventory increases, segments are added in descending order of their
overall frequency of occurrence. If this were so, all segments could be
arranged in a single hierarchy. Such an extreme formulation is not correct,
since no single segment is found in all languages. But if we add a
corollary, that larger inventories tend to exclude some of the most common
segments, then there is an interesting set of predictions to investigate.
We may formulate these more cautiously in the following way: a smaller
inventory has a greater probability of including a given common segment
than a larger one, and a larger inventory has a greater probability of
including an unusual segment type than a smaller one.

The extent to which languages conform to the predictions can be tested

in two straightforward ways. One is to examine inventories of some given
size and see what segments they contain; the other is to examine given
segment types and see how they are distributed across inventories by size.
Using the second approach, the distribution of 13 of the most frequent
consonants was investigated in a set of UPSID languages with relatively
small inventories and in a set of languages with relatively large
inventories. For the small inventory set, languages with 20-24 segments
were chosen. Below 20 segments a language usually has fewer than 13
consonants, so that exclusions would occur simply because of the small
numbers involved. For the large inventory set, all UPSID languages with
over 40 segments were selected. These choices resulted in subsamples
containing 57 and 54 languages respectively.
The set of consonants investigated and their distribution is shown in
Table 1.1 below, together with three percentages. The first is the
percentage of the 57 small inventory languages with the given segment, the
second is the percentage of all UPSID languages which have the segment and
the third is the percentage of the large inventory languages which have the
segment. Note that consonants in the dental/alveolar region have not been
considered here because of the frequent uncertainty as to whether they are
dental or alveolar.
The consonants investigated fall into three groups. Using the overall
frequency of the segment as the expected value, the first and third groups
of these consonants show significant deviations (p < .005), while the

10
 Patterns of sounds

central group shows no significant difference from the expected value

2
(using a )( test). There is a set (especially plain voiceless plosives)
which are more common in the smaller inventories, for example, /p/ and /k/
occur in 90% or more of these languages but in less than 80% of the
languages with larger inventories. There is also a set of these frequent
consonants that are much more likely to occur in languages with larger
inventories, these being notably the voiced stops /b/ and /g/ and the
fricatives /f/ and /J/. There is a tendency for smaller inventories to have
no voicing contrast in stops and to lack fricatives apart from some kind of
/s/. Note that the common nasals in the table are divided one to each
group; / Q / is more common in smaller inventories, /m/ is equally common in
small and large, and /p/ is more common in the larger inventories.

Table 1.1 Inventory size and frequency of selected segments

r
"Small" ." j t a l "Large"
percent percent percent
More l i k e l y in
small i n v e n t o r i e s
hi 89.5% 82.6% 77.8%
hi 93.0% 89.3% 79.3%
hi 59.6% 52.7% 51.9%
Equally l i k e l y in large
or small i n v e n t o r i e s
/m/ 94.7% 94.3% 92.6%
hi 75.4% 75.1% 77.8%
More l i k e l y in
large i n v e n t o r i e s
/b/ 45.6% 62.8% 77.8%
/g/ 42.1% 55.2% 75.9%
/?/ 33.3% 30.3% 55.6%
nil 22.8% 44.5% 64.8%
hi 15.8% 42.6% 51.8%
ill 17.5% 46.1% 70.4%
ni 78.9% 85.5% 94.4%
hi 22.8% 33.8% 37.0%

From this examination, we must conclude that the relationship between

the size and the content of an inventory is a matter that concerns
individual types of segments, rather than being amenable to broad
generalizations.

11
The size and structure of phonological inventories

A second test of aspects of the relationship of inventory size and

structure was conducted by considering what kind of consonant inventory
would be formed if only the most frequent segments were included. In this
case, only the number of consonants in an inventory was considered. Recall
that the modal number of consonants in an inventory is 21. The most
frequently occurring individual consonant segment types in the UPSID data
file would form a "modal" inventory containing the 20 consonants below plus
one other:
p, b * t :, *d tj k, 0 ?
f *s J
m *n n o
w * 1 *r j h
A certain amount of "pooling" of similar segments is assumed to be valid
for this exercise, e.g. dental or alveolar segments have been pooled, and
are represented by /*t, *d, *n/ etc. The twenty-first consonant in the
inventory might be one of several with rather similar frequencies,
especially /z/ or /ts/ which are both about as frequent. A little less
probable would be /x/, /v/ or /d^/ as these are a little less common. The
h h
aspirated stops /pV, /t / and /k / are about as frequent as this last
group but they almost always occur as part of a series of aspirated stops
and so one of them alone as the twenty-first consonant is not plausible.
Because of the several possible candidates, distribution of only twenty
consonants was examined.
Languages are most likely to have between 5 and 11 stops (including
affricates but excluding clicks in this class for these purposes); 63% of
the languages fall within the range given but the scatter is quite wide
(minimum 3, maximum 36, mean 10.5). For fricatives, 1 to 4 is the most
likely (58% of languages), and from 2 to 4 is most likely for nasals (91%
of languages). Languages are most likely to have 2 liquids and 2 vocoid
approximants (41% and 72% respectively). About 63% of the languages have
the consonant / h/ which is not included in any of the categories already
named.
The inventory made up from the most frequent consonant segments does
conform to the predominant patterns concerning the numbers of stops,
fricatives, and so on reported above. For example there are 8 (or, with
/ d y , 9) stops, and 3 (or, with /z/, /x/ or /v/, 4) fricatives. By simply
considering frequency we obtain an inventory which is typologically most
plausible in its structure. This is encouraging. However, none of the 29

12
 Patterns of sounds

languages with 21 consonants contain all 20 of the segments outlined above.

Bambara (105) is very close with 19 of them, having /z/ and /d^/ but
lacking /?/. Fur (203) only deviates by having /y/ instead of /?/ and
having /d^/ rather than /tj/ and is thus also very close to the
idealization generated. But at the other extreme, Wichita (755) has only 7
of the 20 segments (although two other segments are phonemically long
counterparts of /s/ and / n / ) . Other languages with relatively few of the
most common consonants include the Australian language Kariera-Ngarluma
(363) with 10 and Arabana-Wanganura (366) and Mongolian (066) with 11. The
majority of the languages examined have between 14 and 16 of the most
frequent segments.
Leaving aside the consonants in the dental/alveolar region because of
difficulties in arriving at exact counts, a calculation was done comparing
the expected frequency of these consonants in any random subsample of 29
languages, and the observed frequency in the 29 languages examined. The
expected frequency is simply derived from the overall frequency in the
UPSID languages. For the 14 segments compared, there is only one case in
which the expected and observed frequencies differ by more than 3. The
2
difference between these frequencies is not significant (^/ = 1.505 for 13
d.f.). In general, the conclusion suggested is that at the modal inventory
size for consonants there is no greater tendency for more frequent segments
to occur than in the UPSID data file as a whole.

1.5 Phonetic salience and the structure of inventories

Although the idea of a single hierarchy cannot be sustained, there are many
strong implicational hierarchies between particular types of segments
(although very few are exceptionless). Some examples of these, validated by
the data in UPSID and discussed in more detail in the later chapters, are
given below:
(i) /k/ does not occur without /*t/. (One exception in UPSID, Hawaiian,
424.)
(ii) /p/ does not occur without / k/. (Four exceptions in UPSID,
Kirghiz, 062, with /p, "t", q/, Beembe, 123, Tzeltal, 712, and
Zuni, 748. These last two languages have an aspirated velar
plosive /k h / beside unaspirated /p/ and /t/. There are
24 languages with / k/ but no /p/; 18 of these have /b, d, g/»)
(iii) Nasal consonants do not occur unless stops (including affricates)
occur at (broadly speaking) the same place of articulation. (There

13
The size and structure of phonological inventories

are 5 exceptions in UPSID. Ewe, 114, Efik, 119, and Auca, 818,
have /p/ but no palatal or palato-alveolar stops. Hupa, 705, has
/m/ but no bilabial stops. Igbo, 116, has /mr)/ but no labial-
velar stops; it does have labialized velars. There are numerous
examples of languages with stops at particular places of
articulation with no corresponding nasal consonant.)
(iv) Voiceless nasals and approximants do not occur unless the language
has the voiced counterparts. (No exceptions in UPSID.)
(v) Mid vowels do not occur unless high and low vowels occur.
(Two exceptions in UPSID; all languages have at least one high
vowel but Cheremis, 051, and Tagalog, 414, are reported to lack
low vowels.)
(vi) Rounded front vowels do not occur unless unrounded front vowels
of the same basic height occur. (Two exceptions in UPSID, Bashkir,
063, and Khalaj, 064.)
(vii) / 0/ and /CE/ do not occur (separately or together) unless
/ y/ also occurs. (Hopi, 738, is a clear exception. Wolof, 107,
has one front rounded vowel, / 0/, but this has allophones as high
as [ y] . Akan, 115, has marginal phonemes / 0: / and /oe:/ but
no / y/ .)
Yet, as briefly illustrated in section 1.4, such observations cannot be
compiled into a single composite hierarchy. At the very least, alternate
choices must be built in at certain points. This is because equally valid
general prohibitions on the co-occurrence of segments within an inventory
can also be found. Some of these are given below:
(i) A language does not contain both (voiced) implosives and
laryngealized plosives at the same place of articulation. (No
counterexamples in UPSID.)
(ii) A language does not contain a voiceless lateral fricative and a
voiceless lateral approximant. (No counterexamples in UPSID.)
(iii) A language does not contain both /<£/ and /f / or both /p/ and
/v/. (2 counterexamples in UPSID, Tarascan, 747, and Ewe, 114.)
(iv) A language does not include a dental stop, fricative, nasal or
lateral and an alveolar stop, fricative, nasal or lateral of the
same type. (There are 22 exceptions to this observation but this
number is significantly fewer than would be anticipated if the
co-occurrence were unrestricted; 43 co-occurrences of /£/ and
/\l alone would be expected otherwise on the basis of a calculation

14
 Patterns of sounds

which partitions those stops which are unspecified as being dental

or alveolar into dental and alveolar plosives according to the
frequency with which the plosives with known place occur.)
These statements could be subsumed under a general observation that
segments do not (usually) function contrastively unless they are
sufficiently phonetically distinct. The mutual exclusions cited here are
all between phonetically similar segments; without defining what
"phonetically similar" means with any greater precision, note that the
segments referred to could be collapsed under more inclusive labels, e.g.
/j3/ and /v/ are both voiced labial fricatives. The distinctions between
these pairs of segments verge on being noncontrastive phonetic differences
of the type that have been discussed by Ladefoged (1978; 1980: 498-501).
The hypothesis referred to here is that there are measurable phonetic
differences between segments which are generally similar but which occur in
different languages. These differences are assumed to be found along
parameters that do not serve as the basis for phonemic contrast in any
language, or are of smaller magnitude than the differences which form
phonemic contrasts. In this light, the difference between, say, dental and
alveolar stops approaches membership in this class of distinctions which
are generally unavailable for meaningful contrast in a language. (A more
typical member of this class would be, say, a difference in relative timing
of the release of the oral and glottal closures in the production of
ejectives, cf. Lindau 1982.)

This interpretation of prohibitions on co-occurrence introduces a

concept of phonetic distance or phonetic salience as an explanatory factor
in the design of phonological inventories. If we can explain why certain
kinds of segments never (or rarely) occur together in an inventory on the
grounds that the distinctions between them are not salient enough, perhaps
the favoring of certain segments can be explained on the grounds that they
are the most salient, and an appropriate selection of such sounds maintains
generous phonetic distance between the segments of the language involved.
While such ideas have principally been discussed in relation to vowel
inventories (e.g. Liljencrants and Lindblom 1972; Crothers 1978; Disner
1982), they can be extended to the whole inventory. From this perspective,
implicational hierarchies can be interpreted as involving steps down in
phonetic salience, with the most salient segments at the top of any
hierarchical arrangement, and segments which are less distinct (distant)
from each other lower down. Note that this leaves open the possibility that

15
The size and structure of phonological inventories

the mean phonetic distance between the members of an inventory is

approximately constant, as an expanded inventory means the inclusion of
additional members whose distance from their closest neighbor is less, even
though the total phonetic space used by the language is being expanded.
It is far from a straightforward matter to determine appropriate
measures of salience and phonetic distance. Nevertheless, there are
probably some questions which can be answered with only an informal
characterization of these notions. For example, to the question, "is
maximization of distinctiveness the principle on which inventories are
constructed?" the answer is obviously no. Clicks are highly salient yet few
languages (about 1%) use them. Moreover, those that do, have multiple
series of clicks rather than exploiting this feature to make a highly
salient contrast between, say, a dental click and a velar plosive in a
limited series of stops. The most frequent vowel inventory is /i, e, a, o,
u/, not / I , e, a, o, u / where each vowel not only differs in quality but
is distinctively plain, nasalized, breathy, laryngealized and
pharyngealized. Yet this second set of vowels surely provides for more
salient distinctions between them and approaches maximization of contrast
more than the first set whose differences are limited to only the primary
dimensions conventionally recognized for vowel quality.

A more adequate theory of inventory structure must recognize that

certain dimensions of contrast are preferentially used before others in
ways that do not seem related to salience. For example, the world's
languages only add the additional parameters of contrast to vowels if they
include a fairly wide sample of simple contrasts on the primary vowel
quality dimensions. In a sense, then, these additional ways of contrasting
vowels are themselves involved in an implicational hierarchy whose
arrangement is not predicted by a principle of selecting maximally salient
contrasts.
Independent of the above discussion, it must be recognized that
phonetic distance cannot explain some of the prohibitions on co-occurrence
of segments. There is a class of these prohibitions that differ from those
cited above in that the distinctiveness of the segments concerned is not
really in doubt. An example of this is the co-occurrence restriction which
applies to subinventories of laterals. A language with several lateral
segments contrasts them either by manner (voiced approximant, voiceless
fricative, ejective affricate, etc.) or by place (with all the laterals
being voiced approximants). Only one language in UPSID (Dieguefio 743)

16
 Patterns of sounds

clearly violates this rule, although Irish (001) is an arguable exception

too. Even two exceptions are significantly fewer than expected. Thus, while
multiple-lateral subsystems almost invariably contain an apical or laminal
lateral approximant, which is therefore at the top of an implicational
hierarchy, at the lower end of this hierarchy there are two branches, one
permitting elaboration by place and the other permitting elaboration of
laterals sharing the same place of articulation by variation in the manner
of production.

1.6 Compensation in inventory structure

The fact that certain types of mutual exclusions occur which do not seem to
be based on principles of phonetic distance is suggestive of the position
that there is a principle of "compensation" controlling the structure of
inventories. Martinet (1955), for example, suggests that a historical
change which simplifies an inventory in one area is counterbalanced by a
compensating elaboration elsewhere. Similar ideas are discussed at length
by Hagege and Haudricourt (1978).
If diachronic changes do generally follow this pattern, then the
consequence should be measurable relationships between various facets of
inventories which follow a pattern of negative correlation. We have already
seen, though, one aspect of inventory structure in which compensation does
not occur. The tendency for vowel inventories to increase in step with
increases in consonant inventories (section 1.4) is the opposite of the
prediction made by a compensation theory. Several other inventory sectors
were investigated for general signs of the operation of a compensation
process.
The stop inventories of the languages in UPSID were examined to see if
there was a tendency for the elaboration of the number of place contrasts
to be compensated for by reduction of the number of stop manner contrasts
and vice versa. Such a compensation is suggested by the inventories of
Australian languages. These typically have a rich range of places of
articulation for stops (and nasals) but no contrasts of manner (such as
voicing differences) within the stops (Wurm 1972; Dixon 1980). Is this a
local aberration or just a particularly striking example of a basic pattern
in human language? Has the atypical language Mabuiag (365) compensated for
its reduction to 3 places of articulation by adding a voicing contrast,
creating the stop inventory /p, "t", k; b, "d", g/?

17
The size and structure of phonological inventories

There are a number of ways in which this comparison of places and

manners could be done. In this instance, it was decided to treat
doubly-articulated stops (in practice, this means labial-velars) as having
a place of articulation distinct from that of either of their components,
i.e. labial-velar is treated as a place of articulation. Secondary
articulations on the other hand, since they are more likely to appear with
a range of primary places of articulation, seem more akin to the
"series-generating" nature of the differences in initiation and phonation
type, and hence were treated as differences in manner (a different count is
given in Chapter 2 ) . So, of the two inventories given below, (a) is treated
as having 4 places of articulation and 2 manners, whereas (b) is treated as
having 3 places of articulation and 3 manners.

(a) p t k iCp
b d g gb

(b) p t k
pw kw

b d g

The correlation was obtained between the number of places out of a list
of 10 and the number of manners out of a list of 14 "series-generating"
manner components for each language2 (glottal was not included in the
calculation of places because glottal stops do not (ordinarily) have
contrasting manners). The numbers of languages involved are shown in Table
1.2. Those rows with very sparse representation, i.e. less than 3 and more
than 5 places, or more than 4 manners, have been eliminated, removing 29
languages from the calculation. There is essentially no correlation between
the numbers of places and the numbers of manners for stops, whereas the
hypothesis of compensation would predict a strong negative correlation.
A similar computation was performed for fricatives relating place to
manner with cases with over 5 places or over 4 manners dropped (resulting
in 16 languages being excluded over and above the 21 languages which have
no fricatives). The results are given in Table 1.3. The observed data are
significantly different from expected (p = .0001), and in this case a
fairly substantial positive correlation (r = .46) between the two variables
is found. Again this is counter to the predictions of a compensation
hypothesis, and more strongly so than is the case with stops.

18
 Patterns of sounds

Table 1.2 Manners and places for stops

Manners
1 2 3 4
Totals
1 19 83 35 20 1 157
Places 2 14 28 32 21 1 95
3 3 16 11 6 1 36

Totals 36 127 78 47

Table 1.3 Manners and places for fricatives

Manners
1 2 3
Totals
1 37 8 1 | 46
2 46 35 1 | 82
Places 3 21 45 8 1 74
4 6 32 8 | 46
5 4 22 6 1 32

Totals 114 142 24

The example given by Martinet (1955) of a compensatory adjustment in

segment inventories concerns elaboration of the fricative inventory by
reduction of the stop inventory. Therefore a similar comparison of
fricative and stop numbers was made. In this computation, languages with
fewer than 5 or more than 13 stops were dropped and languages with more
than 8 fricatives were dropped, resulting in 92 languages being eliminated
from the total. No tabulation of these numbers is provided because the
table requires an inconveniently large number of cells. Statistical tests
showed a weak positive correlation between the number of fricatives and the
number of stops (r = .35), but this correlation is probably not reliable as
its significance level is under .05. However, the absence of an inverse
correlation is still notable.

19
The size and structure of phonological inventories

1.7 Segments and suprasegmentals

Despite the failure to find any confirmation of a compensation hypothesis
in several tests involving segmental subinventories, it is possible that
the compensation exists at another level. One possibility was evidently in
the minds of Firchow and Firchow (1969). In their paper on Rotokas (625),
which has an inventory of only 11 segments, they remark that "as the
Rotokas segmental phonemes are simple, the suprasegmentals are
complicated".4 A similar view of a compensatory relationship between
segmental and suprasegmental complexity seems implicit in much of the
literature on the historical development of tone. For example, Hombert,
Ohala and Ewan (1979) refer to "the development of contrastive tones on
vowels because of the loss of a voicing distinction on obstruents". If this
phenomenon is part of a pervasive relationship of compensation we would
expect that, in general, languages with larger segmental inventories would
tend to have more complex suprasegmental characteristics.
In order to test this prediction, the languages in UPSID which have
less than 20 or more than 45 segments were examined to determine if the
first group had obviously more complex patterns of stress and tone than the
second. Both groups contain 28 languages. The findings on the
suprasegmental properties of these languages, as far as they can be
ascertained, are summarized in Table 1.4.

Table 1.4 Inventory size and suprasegmentals

Languages with small Languages with large

segment inventory (< 20) segment inventories (> 45)
Stress
contrastive stress 6 8
predictable stress 7 9
pitch accent (?) 2 2
no stress 5 4
inadequate data 8 5

Tone
complex tone system 2 6
simple tone system 2 4
no tones 22 15
inadequate data 2 5

20
 Patterns of sounds

Despite some considerable uncertainty of interpretation and the

incompleteness of the data, the indications are quite clear that these
suprasegmental properties are not more elaborate in the languages with
simpler segmental inventories. If anything, they tend to be more elaborate
in the languages with larger inventories.
There are more "large" languages with contrastive stress and with
complex tone systems (more than 2 tones) than "small" languages. There are
more "small" languages lacking stress and tone. The overall tendency
appears once again to be more that complexity of different kinds goes hand
in hand, rather than for complexity of one sort to be balanced by
simplicity elsewhere.

1.8 Segment inventories and syllable inventories

Another hypothesis is that the size of the segment inventory is related to
the phonotactics of the language in such a way as to limit the total number
of possible syllables that can be constructed from the segments and
suprasegmental properties that it has. Languages might then have
approximately equal numbers of syllables even though they differ
substantially in the number of segments. Rough maintenance of syllable
inventory size is envisaged as the function of cyclic historical processes
by, for example, Matisoff (1973). He outlines an imaginary language in
which, at some arbitrary starting point, "the number of possible syllables
is very large since there is a rich system of syllable-initial and -final
consonants". At a later stage of the language these initial and final
consonantal systems are found to have simplified but "the number of vowels
has increased and lexically contrastive tones have arisen" maintaining
contrasting syllabic possibilities. If tone or vowel contrasts are lost,
consonant clustering will increase at the syllable margins again.
A brief investigation of the relationship between segmental inventory
size and syllable inventory size was carried out by calculating the number
of possible syllables in 9 languages. The languages are Tsou (418), Quechua
(819), Thai (400), Rotokas (625), G3 (117), Hawaiian (424), Vietnamese
(303), Cantonese, Higi, and Yoruba (the last three are not in UPSID but
detailed data on the phonotactics are available in convenient form for
these languages). The 9 languages range from those with small segment
inventories (Rotokas, Hawaiian) to those with relatively large inventories
(Vietnamese, Higi, Quechua) and from those with relatively simple
suprasegmental properties (Tsou, Hawaiian, Quechua) to those with complex

21
The size and structure of phonological inventories

suprasegmental phenomena (Yoruba, Thai, Cantonese, Vietnamese). In

calculating the number of possible syllables, general co-occurrence
restrictions were taken into account, but the failure of a particular
combination of elements to be attested if parallel combinations were
permitted is taken only as evidence of an accidental gap, and such a
combination is counted as a possible syllable. The calculations reveal very
different numbers of possible syllables in these languages. The totals are
given in Table 1.5.

Table 1.5 Syllable inventory size

of 9 selected languages

Language Total possible syllables

Hawaiian 162
Rotokas 350
Yoruba 582
Tsou 968
2,331
Cantonese 3,456
Quechua 4,068
Vietnamese 14,430
Thai 23,638

Even with the uncertainties involved in this kind of counting, the numbers
differ markedly enough for the conclusion to be drawn that languages are
not strikingly similar in terms of the size of their syllable inventories.
In following up this study, several tests were done to see which of a
number of possible predictors correlated best with syllable inventory size.
The predictors used were the number of segments, the number of vowels, the
number of consonants, the number of permitted syllable structures (CV, CVC,
CCVC, etc.), the number of suprasegmental contrasts (e.g. number of stress
levels times number of tones), and a number representing a maximal count of
segmental differences in which the number of vowels was multiplied by the
number of suprasegmentals. Of these, the best predictor is the number of
permitted syllable types (r = -69), an indication that the phonotactic
possibilities of the language are the most important factor contributing to
the number of syllables. The next best predictor is the number of
suprasegmentals (r = .59), with the correlation with the various segmental

22
 Patterns of sounds

counts all being somewhat lower. Although all the predictors tested show a
positive simple correlation with the number of syllables, in a multiple
regression analysis only the number of vowels contributes a worthwhile
improvement to the analysis (r change = .19) beyond the number of syllable
types. Thus we can say that syllable inventory size does not depend heavily
on segment inventory size. Nonetheless, because the predictors do have
positive correlations with syllable inventory size, the picture is once
again of a tendency for complexity of different types to go together.

1.9 Conclusions
Work, with UPSID has confirmed that segment inventories have a well-defined
central tendency as far as size is concerned. Nonetheless considerable
variation in their size and structure occurs. Their structure is subject to
a hierarchical organization in many particulars but cannot be substantially
explained in terms of a single unified hierarchy of segment types. This is
partly because segments of certain types are subject to rules of mutual
exclusion. The mutual exclusions cannot all be explained as due to the
avoidance of inadequate phonetic contrasts, as some involve strongly
salient distinctions. A search for evidence that languages maintain a
balance by compensation for complexity in one phonological respect by
possessing simplicity elsewhere failed to find it in balance between
classes of segments, between segments and suprasegmental contrasts, or
between segments and phonotactic conditions. These investigations suggest
that complexity of various kinds occurs together in languages, and that
languages really do differ in their phonological complexity.

Notes
1. If languages with large phoneme inventories were approaching some kind
of limit on the ability to discriminate contrasts, it would be expected
that speakers of these languages would show higher error rates in tasks
involving phoneme recognition than speakers of languages with small
inventories. I know of no experimental data which bear on this point.
2. The manner components are: plain voiceless, plain voiced, voiceless
aspirated, breathy, preaspirated, laryngealized, implosive, ejective,
prenasalized, nasally-released, labialized, palatalized, velarized,
pharyngealized.
3. Of course, other compensations may exist between aspects of the
segmental inventories not examined here; and the failure to find
evidence for gross compensatory tendencies does not affect the validity
of any posited historical evolution in a particular case.
4. Rotokas is not really very complex in its suprasegmentals. It has a
partially predictable stress and a contrast of vowel length that seems
only partly independent of stress (Firchow, Firchow and Akoitai 1973).
Long vowels are not treated as separate segments in UPSID for this
language.

23
The size and structure of phonological inventories

5. Matisoff also suggests that the morphological complexity of the

language would evolve along with the phonological shifts.

References
Baucom, K.L. 1974. Proto-Central Khoisan. In E. Voeltz (ed.) Third Annual
Conference on African Linguistics (Indiana University Publications,
African Series, 7 ) . Indiana University, Bloomington: 3-38.
Crothers, J. 1978. Typology and universals of vowel systems. In J.H.
Greenberg et al. (eds.) Universals of Human Language, Vol. 2, Phonology.
Stanford University Press, Stanford: 93-152.
Disner, S.F. 1982. Vowel Quality: the Relationship between Universal and
Language-Specific Factors (UCLA Working Papers in Phonetics 58).
University of California, Los Angeles.
Dixon, R.M.W. 1980. The Languages of Australia. Cambridge University Press,
Cambridge.
Firchow, I. and Firchow, J. 1969. An abbreviated phoneme inventory.
Anthropological Linguistics 11: 271-6.
Firchow, I., Firchow, J. and Akoitai, D. 1973. Vocabulary of
Rotokas - Pidgin - English. Summer Institute of Linguistics, Papua New
Guinea Branch, Ukarumpa.
Grace, G. W. 1959. The position of the Polynesian languages within the
Austronesian (Malayo-Polynesian) language family (IJAL Memoir 16).
Indiana University, Bloomington.
Hagege, C. and Haudricourt, A. 1978. La Phonologie Panchronique. Presses
Universitaires de France, Paris.
Hockett, C.F. 1955. A Manual of Phonology (IJAL Memoir 11). Indiana
University, Bloomington.
Hombert, J-M. , Ohala, J.J. and Ewan, W.G. 1979. Phonetic explanations for
the development of tones. Language 55: 37-58.
Ladefoged, P. 1978. Phonetic differences within and between languages. UCLA
Working Papers in Phonetics 41: 32-40.
Ladefoged, P. 1980. What are linguistic sounds made of? Language 56:
485-502.
Liljencrants, J. and Lindblom, B. 1972. Numerical simulation of vowel
quality contrasts: the role of perceptual contrast. Language 48: 839-62.
Lindau, M. 1982. Phonetic differences in glottalic consonants. UCLA Working
Papers in Phonetics 54: 66-77.
Martinet, A. 1955. Economie des changements phonetiques (2nd ed.). Franke,
Berne.
Matisoff, J.M. 1973. Tonogenesis in Southeast Asia. In L.M. Hyman (ed.)
Consonant Types and Tone (Southern California Occasional Papers in
Linguisticsl). University of Southern California, Los Angeles.
Pukui, M.K. and Elbert, S.H. 1965. Hawaiian - English Dictionary (3rd
ed.). University of Hawaii Press, Honolulu.
Sheldon, S.N. 1974. Some morphophonemic and tone perturbation rules in
Mura-Piraha. International Journal of American Linguistics 40: 279-82.
Snyman, J.W. 1975. Zul hoasi Fonologie en Woordeboek. Balkema, Cape Town.
Traill, A. 1978. Research on the Non-Bantu African languages. In L.W.
Lanham and K.P. Prinsloo (eds.) Language and Communication Studies in
South Africa. Oxford University Press, Cape Town.
Voorhoeve, C.L. 1965. The Flamingo Bay Dialect of the Asmat Language
(Verhandelingen van het Koninklijk Instituut voor Taal-, Land- en
Volkenkunde 46). Nijhoff, The Hague.
Wurm, S.A. 1972. Languages of Australia and Tasmania (Janua Linguarum,
Series Critica 1 ) . Mouton, The Hague.

24
Stops and affricates

2.1 Introduction
Stops occur in the inventories of all known languages and have
appropriately been regarded as the optimal consonants (e.g. by Jakobson and
Halle 1956: 4 2 ) . The most frequently found types of stops are plosives,
that is, stops made with an egressive pulmonic airstream. Apart from
differences in place of articulation, these may vary in a number of ways
through variations in laryngeal settings and in the relative timing of
voice onset and offset and of velic closure or opening. In addition there
are stops made with glottalic and velaric airstreams, i.e. ejective stops,
implosives and clicks. The principal architecture of stop systems is
conveniently discussed in terms of two dimensions representing the manner
series and the places of articulation that occur. In this chapter, we will
therefore analyze the structure of stop systems in the languages in the
UPS ID database in terms of the number of series and the number of places
used. We will also examine in more detail some questions concerning the
frequency of stops, particularly plosives, at different places of
articulation. Glottalic and laryngealized stops are discussed in more
detail in a separate chapter on glottalic consonants (Chapter 7 ) . Clicks
are not the subject of any special analysis, mainly because so few of the
UPSID languages contain any, and they are not included in the totals in
this chapter. However, nonlateral affricates (except affricated clicks) are
included in some of the analyses because of the close relationship of
affricates to stops. Lateral affricates are discussed further in Chapter 5.

2.2 Stop series

A series is a set of stops (perhaps including affricates) which share in a
general sense the same "manner". That is, they share the same phonation

25
Stops and affricates

type (voiceless, voiced, breathy voiced, laryngealized), the same airstream

(pulmonic, velaric, glottalic ingressive or glottalic egressive), the same
relative timing of the onset of voicing (unaspirated, aspirated,
pre-aspirated) and the same relative timing of velic closure (nonnasal,
prenasalized, nasally-released). In this chapter, secondary articulations
accompanying the production of stops, including palatalization,
velarization and pharyngealization, will not be considered as creating
separate series of stops. The opposite choice was made in Chapter 1 for the
purposes of the computation reported in Table 1.2.
We have considered each of the "manner" differences mentioned above
which occur in the stops and/or affricates of a language as establishing a
series in that language. Thus, for our purposes it is not necessary for a
language to show contrast between series at any given place of articulation
for the series to count as distinct. For example, if a language had only
two stops /k/ and /b/ it would be considered to have two series, a
voiceless one and a voiced one, despite the fact that both series are
defective, and voicing could be predicted from place. This decision means
that the number of series contrasts is maximally represented in our counts.
In some languages there may be phonological reasons to collapse certain
partial series together. This has generally not been done since it is
considered more important to represent the phonetic heterogeneity that
exists. In general, languages with affricates have them in the same series
as they have stops, or in a subset of the series in which they have stops.
But a language with a series whose only representatives are affricates has
been counted as adding an additional series.
All languages have at least one series of stops, but two is the most
common, just over 50% having that number. Languages with more than 4 series
are quite rare, and no language in our survey has more than 6 stop series
(excluding clicks). The frequency with which languages with different
numbers of series occur in UPSID is given in Table 2.1.

Table 2.1 Number of stop series in UPSID languages

Number of stop series

1 2 3 4 5 6
No. of languages 50 162 76 25 2 2
% of languages 15.8% 51.1% 24.0% 7.9% 0.6% 0.6%

26
 Patterns of sounds

The particular types of series represented are given in Table 2.2

together with the frequency with which such series are found.

Table 2.2 Frequency of stop series

Number of languages Percent

Plain voiceless 291 91.8%
Plain voiced 212 66.9%
Aspirated voiceless 91 28.7%
Voiceless ejective 52 16.4%
Voiced implosive 35 11.0%
Prenasalized voiced 18 5.6%
Breathy voiced 7 2.2%
Laryngealized voiced 6 1.9%
Laryngealized voiceless 3 0.9%
Preaspirated voiceless 2 0.6%
Voiceless with breathy release 2 0.6%
Postnasalized voiced 1 0.3%
Prevoiced ejective 1 0.3%
Voiceless implosive 1 0.3%

Plain voiceless plosives series are the most frequently found, with almost
92% of languages having such a series. Keating, Linker and Huffman (1983)
suggest that this type of plosive is the most widespread phonetically in
languages, and argue that they are most frequent because they are the most
efficient from the aerodynamic and articulatory points of view (at least in
initial positions). A language with only one stop series almost invariably
has plain voiceless plosives (49 out of 50) and the one exception, the
Australian language Bandjalang (368), may be incorrectly reported as having
voiced plosives. (Australian languages more typically have a voiceless
unaspirated stop series.) In none of the languages with only one series is
there an aspirated voiceless series nor do any have any glottalic or
laryngealized stops. Indeed, these types of stop series do not become at
all frequent until there are at least 3 series of stops. This may be seen
from Table 2.3, which shows the percentage of languages with a given number
of series that have a series of the particular type listed. The table only
gives a partial listing of the possibilities.

27
Stops and affricates

Table 2.3 Frequency of stop series by number of series

Number of series
1 2 3 4
Plain voiceless 98 .0% 90.1% 89.5% 96.0%
Plain voiced 2.0% 81.5% 69.7% 88.0%
Aspirated voiceless 0.0% 16.0% 63.2% 52.0%
Voiceless ejective
or voiceless laryngealized 0.0% 3.7% 42.1% 56.0%
Voiced implosive or
voiced laryngealized 0.0% 1.2% 27.6% 48.0%

Languages with 2 stop series

A language which contrasts only 2 series of stops typically has a plain
voiceless/voiced contrast. This is so for 117 of the 162 languages
concerned (72.2%). A further 27 have a contrast between plosive series that
differ only along what is often conceived of as a voice onset time (VOT)
continuum (Lisker and Abramson 1964), that is, they have a contrast of
plain voiceless and aspirated voiceless, or of plain voiced and aspirated
voiceless. Altogether, then, there is a total of 88.9% of languages with 2
stop series which differentiate the series just by use of differences which
have been described as voice onset time differences. The small number of
remaining languages include either (i) a voiced series with a nasal onset
or offset - 9 cases, of which six have plain voiceless plosives contrasting
with prenasalized voiced plosives - or (ii) a series with a less usual
phonation type or airstream - 9 cases of which 6 contrast plain voiceless
plosives with voiceless ejectives or voiceless laryngealized plosives.
There are only two languages in this group which have voiced implosives,
Nyangi (207) and Maasai (204).

Languages with 3 stop series

Among the set of languages with 3 stop series, one of the series is usually
a plain voiceless one (about 90% of the cases) but apart from this there is
a considerable amount of variation in these languages. The single most
commonly found pattern is a 3-way contrast along the VOT continuum -
aspirated voiceless, plain voiceless and voiced plosives - but only 19 of
the 76 languages concerned have this pattern (25.0%). The 3 next most
frequent types are languages with plain voiceless and voiced plosives and
an ejective series (13 languages, 17.1%), plain and aspirated voiceless

28
 Patterns of sounds

plosives and an ejective series (12 languages, 15.8%), or plain voiceless

and voiced plosives and a voiced implosive series (12 languages, 15.8%).
However, note that all 3 of these patterns, and several others, can be
summed up as consisting of two series drawn from the set of VOT contrasts
plus one series with a "glottalic" element, either a glottalic air stream or
laryngealization. In total, there are 50 languages which conform to this
general framework, 31 of them with a voiceless glottalic series, and 19
with a voiced glottalic series. In other words, almost two thirds of the
languages with 3 series of stops distinguish them in this general fashion
and the "two VOT + glottalic" pattern is more common than 3 VOT contrasts.
Only two of the languages with 3 series (Maidu, 708, and K'ekchi, 714) have
two "glottalic" series (the decision to analyse K'ekchi as having 3 rather
than 2 series of stops might be challenged, since there is no 3-way
contrast at any given place of articulation; see further in Chapter 7 ) .
There are also 4 languages which have prenasalized voiced plosives plus two
series contrasting along the VOT continuum.

Languages with 4 stop series

The 25 languages in the UPSID sample which have 4 series of stops are even
more heterogeneous in their structure than the languages with 3 series.
Four general patterns, (a)-(d) below, are about equally common:
a) plain voiceless/ plain voiced/ voiced implosive/ voiceless ejective
b) plain voiceless/ aspirated voiceless/ voiced/ voiced ejective
c) plain voiceless/ plain voiced/ prenasalized voiced/ voiced implosive
or voiced laryngealized
d) plain voiceless/ aspirated voiceless/ plain voiced/ breathy voiced
There are 6 languages with the (a) pattern and 5 with each of the others.
Two other languages, Zulu (126) and S. Nambiquara (816), are rather similar
to the (a) group in that they have plain and aspirated voiceless plosives,
ejectives and implosives. This means that the pattern of two VOT contrasts
and two glottalic series is the most widespread, although it only accounts
for about one third of the languages concerned. There are some strong areal
patterns to recognize in these systems. All the languages with the (a)
pattern and all but one with (b) are from Africa (the exception is the
Austro-Asiatic language Sedang, 304). All the languages with the (c)
pattern are from North America, and all the languages with the (d) pattern
are from the Indian subcontinent. The (a) group includes both Nilo-Saharan
and Afro-Asiatic languages from Africa, while the (b) group contains

29
Stops and affricates

Niger-Kordofanian (Gbeya, 129), Nilo-Saharan (Yulu, 216, Sara, 217) and

Afro-Asiatic (Ngizim, 269) languages. The (d) group includes Indo-European,
Dravidian and Austro-Asiatic languages. I t should be noted that the (d)
grouping displays the most completely filled out 4-series patterns, as in
Kharia (301), with a l l 4 of i t s places of articulation filled in a l l 4 of
i t s 4 series of plosives, plus a set of palato-alveolar affricates:
Kharia stop inventory:
plain voiceless plosives/affricates p t tj t k
aspirated voiceless plosives/affricates ph th tjh th kh

plain voiced plosives/affricates b d d^ d g

breathy voiced plosives/affricates b d d^ d g
Compare this with the partial contrasts at each place in Hausa (266), a
language from group (a) above. In Hausa there are no more than 3 series
represented at any given place (velars with secondary articulations are
omitted).
Hausa stop inventory:
plain voiceless plosives/affricates t tj k
plain voiced plosives/affricates b d d^ g
voiceless ejective stops k'
voiced implosives b a
In Kullo (262) the 4 series are all present only at the dental/alveolar
place of articulation. At other places, one, two, or three series occur:
Kullo stop inventory:
plain voiceless plosives/affricates "t" "ts" tj k
plain voiced plosives/affricates b "d" d^ g
voiceless ejective stops/affricates "t ' " "ts'" k'
1
voiced implosives "cP
Such a system containing deficiencies at some places is quite common,
p a r t i c u l a r l y as the number of series increases.

Languages with 5 or 6 series

There i s only a small number of languages with more than 4 series of stops.
The r e l a t i v e l y closely related languages Otomi (716) and Mazahua (717) have
5 series, contrasting 3 VOT categories and 2 g l o t t a l i c s e r i e s . Igbo (116)
and !XQ (918), with 6 series, add a breathy voiced series to these.
Interestingly, both these last two languages have an unusual glottalic
s e r i e s : Igbo has voiceless implosives beside the more usual voiced ones;
!XC[ has prevoiced ejectives beside the more usual voiceless ones. There are
too few languages to decide if t h i s represents a general pattern.

30
 Patterns of sounds

2.3 Summary of analysis of stop systems

In general, therefore, languages nearly always include a plain voiceless
series of stops. If there ,is only one series it is of this kind, As the
number of series expands, a contrast along the VOT dimension is added first
(this is usually reported as a voiced/voiceless distinction). However a
third series is more likely to be a glottalic series of one kind or another
than to involve further distinctions along the VOT continuum. Systems with
larger numbers of series seem about equally likely to add a second
glottalic series or to add to the number of VOT contrasts.

2.4 Stop systems by place

The other main dimension along which stops vary is the place of
articulation. Since affricates frequently occur at places at which there
are no stops, two separate analyses of the data will be given, the first
for stops alone, the second including affricates. The number of different
places in which the languages in UPSID have stops of some type is
summarized in Table 2.4. In this tabulation glottal stops have been omitted
since this segment type differs from other stops in several ways (e.g.
glottal stops cannot vary in manner). Also the one pharyngeal stop (Iraqw,
260) is omitted because of the suspicion that the same is true of
pharyngeals.

Table 2.4 Frequency of sizes of stop systems by place

Number of places for stops

2 3 4 5 6
No. of languages 2 171 103 35 6
Percent of sample 0.3% 53.9% 32.5% 11.0% 1.9

The overwhelming majority of languages utilize at least 3 places of

articulation for stops. The 2 languages in the survey which have stops at
only 2 places are Hawaiian (424), which has no alveolar or dental stop, and
Wichita (755), which has no labials. There are only 2 other languages that
have no labial stop of any kind, Hupa (705) and Aleut (901). Hupa also has
no velar stop, as does Kirghiz (062). All other languages have bilabial,
dental or alveolar and velar stops. The number of languages in which there
is one or more stop at each of the major places of articulation is given
below in Table 2.5. In this tabulation, dental and alveolar places have
been collapsed together, partly because they are frequently not reliably

31
Stops and affricates

distinguished in the sources and partly because a contrast between these

places is unusual. Palatal and palato-alveolar places have also been
collapsed in this stop tabulation, since no language is reported with a
contrast, and the choice between the two labels seems partly to result from
different terminological traditions.

Table 2.5 Number of languages with stops at given places

Bilabial Dental Palatal Retroflex Velar Uvular Labial-velar

or or
alveolar palato-
alveolar
No. of
languages 314 316 59 36 315 47 20
Percent 99.1% 99.7% 18.6% 11.4% 99.4% 14.8% 6.3%

Note that there are 24 languages which have stops at both dental and
alveolar places, though these are not always contrastive, since the place
could be redundantly predicted from a manner difference in some languages.
For example Guahibo (830) is reported with aspirated dental /t h / but
alveolar /t, d/; Sundanese (408) has /t/ but /d/; Yulu (216) has dental
/t, i/ but its implosive and prenasalized stops are possibly produced
further back and have been entered as "cf" and " d". It follows from the
near universality of bilabial and velar places that a language with stops
at only 3 places is unlikely to contrast dental and alveolar. In fact none
of the languages with both dental and alveolar stops have less than 4
places and 5 or 6 is more typical (17 of 24 cases, or 70.8%). The contrast
of dental and alveolar places is thus particularly associated with the use
of a relatively large number of places in the stop system of the language
concerned. Typically languages with stops in the palatal area or with
retroflex, uvular or labial-velar stops also have 4 or more places; this is
true for 58 of 59 languages in the case of palatals (98.3%), for 44 of 47
languages in the case of uvulars (93.6%), and for all the languages with
retroflex or labial-velar stops (100%).
Note that the most common place system uses 3 well distinguished
articulators - the lips (bilabial), the tongue tip or blade (dental or
alveolar) and the tongue body (velar). An additional place is more likely
to be another tongue-body articulation (either palatal or uvular) rather
than another tongue tip/blade articulation (e.g. retroflex). The only other
pattern that is at all common for expanding the number of places used is to

32
 Patterns of sounds

combine two of the basic places through introduction of labial-velars. No

other double-articulation is reported apart from a "dental-palatal" in
Maung (350), which may describe an articulation with a very extensive
single longitudinal contact rather than a true double articulation. Besides
this marginal exception, the absence of double articulations involving the
tongue tip/blade and body is understandable. These two articulators are not
fully independent in the way that the lips and the tongue are, since they
are both parts of the tongue. There is no similar reason for the absence of
labial-alveolars (or labial-dentals), combining articulations of the lips
and the tongue tip or blade. Yet they do not seem to occur; although
Hoffmann (1963) has suggested that Margi (268) has labial-alveolar stops
(as well as other types of labial-alveolar segments), this description is
inaccurate, as these are instead sequences of labial and alveolar segments
(cf. Maddieson 1983).
A more detailed analysis of the way in which place contrasts are
increased shows that in a system with 4 places, a palatal (or
palato-alveolar) place is a little more frequent than a uvular one (36 to
29 cases). These two tongue body places easily outrank tongue tip or blade
places (retroflex 17, dental 7 ) . Twice as many languages (60) add a place
which is in front of the velar than add a post-velar place (30). Twelve
languages have labial-velars. So 72 of the 103 languages with stops at 4
places (69.9%) avoid any articulation further back than velar.
A language with 5 places for stops is most likely to add one tongue
blade/tip place and one tongue body place to the basic 3 (18 of the 35
languages, 51.4%). Two additional tongue body places, palatal and uvular,
are next most likely (7, 20.0%) with two tip/blade articulations (dental,
retroflex) or either a tip/blade or body place combined with a labial-velar
place being less favored (3 or 4 cases each).
Five of the 6 languages with 6 places are uniform in the places that
they use. They have bilabial, dental, alveolar, retroflex, palatal and
velar stops, but not uvulars. The absence of uvulars is probably a genetic
characteristic, rather than anything typical of inventories with 6 places
in general. The 5 languages concerned are all Australian. The use of a
larger than average number of places is a marked characteristic of this
language family: 10 of the 19 Australian languages in UPSID have 5 or 6
places and an additional 6 have 4 places. Thus 84.2% of the Australian
languages in UPSID have more than the usual number of places in their stop
inventory. None of them is reported to have uvular stops.

33
Stops and affricates

2.5 Stop and affricate places

As mentioned above, languages quite frequently have affricates at places of
articulation where they have no stops. If places are tabulated for stops
and affricates jointly there are considerable differences from the results
with stops alone. Table 2.6 gives the number and percentage of languages in
UPSID with any given number of places at which stops and/or affricates
occur.

Table 2.6 Number of places used for stops and affricates, by language

Number of places for stops and/or affricates

2 3 4 5 6 7
No. of languages 2 62 139 87 25 2
Percent 0.6% 19.6% 43.8% 27.4% 7.9% 0.6%

When affricate places are included, a handful of languages are counted with
7 places of articulation, and the proportion of languages with 4, 5 and 6
places increases. By this count, 4 rather than 3 is the most frequent
number of places to contrast (cf. Table 2.4), with the most common pattern
being for palato-alveolar affricates to be added to the near-universal
bilabial, dental/alveolar and velar stops. This pattern accounts for 86
languages - over a quarter of the total sample by itself.
The languages with 5 places in Table 2.6 are most likely to add
palato-alveolar affricates to the 3 basic positions together with either a
uvular (19 languages), palatal (12 languages) or retroflex (11 languages)
stop place. Ten of this set of languages, however, have two additional
places at which affricates but not stops appear. Of the 25 languages with 6
places, 17 add palato-alveolar affricates to a system with a smaller number
of stop places. The languages with 6 places at which stops occur tend not
to have any affricates at any place, but this too may be an Australian
peculiarity rather than a general rule. Recall that these are mostly
Australian languages. Note that Australian languages frequently have no
fricatives as well as having no affricates (see Chapter 3 ) .

2.6 Voicing and place of articulation for plosives

In this section, we will discuss some of the interactions of voicing and
place which emerge from the examination of individual types of plosives
(pulmonic egressive stops). Glottalic stops are discussed in full in

34
 Patterns of sounds

Chapter 7 and the findings there should be compared. The most common types
of plosives are naturally enough the plain voiced and voiceless types at
the 3 main places of articulation. Their frequencies are given in Table
2.7. Separate totals are given for dental and alveolar stops, as well as
for the "unspecified dental or alveolar" category. Below, the total of
these 3 categories is given. In the text, the combined category of dentals
and alveolars is represented by a phonetic symbol preceded by an asterisk,
i.e. *t , *d.

Table 2.7 Frequency of plain plosives by place

Bilabial Dental "Dental/alveolar" Alveolar Velar

Plain voiceless 263 72 135 102 283
Plain voiced 199 53 77 65 175

Plain voiceless 309

Plain voiced 195

The class of segments represented by /*t/ is the most common of these, but
the total of 309 /*t/'s should be "corrected" for the 19 languages that
have both /t/ and /t/, so that only 290 languages have a /*t/ segment.
Nonetheless all languages with a series of plain voiceless plosives, apart
from Hawaiian, have a /*t/ segment. Note that there are about 20 more /k/'s
than /p/'s. There are also about 20 fewer /g/'s than /b/'s or /*d/'s.
Similar relationships between these stops can be seen not just in the raw
totals, but also in the structure of the inventories of individual
languages. Thus there are 24 languages that have /k/ but lack /p/. Note
that these are not languages that avoid use of the bilabial place in
general - most of them (18 of the 24) have a voiced series of stops
including /b/. All of them include /*t/ in their inventories. On the other
hand, there are only 4 languages with a /p/ but no /k/• Thus, /p/ is more
likely to be "missing" than /k/ and an implicational hierarchy can be set
up such that presence of /p/ implies the strong likelihood of the presence
of /k/, which similarly implies presence of /*t/. With a plain voiced
series, the place preferences are different. Twenty-one languages have a
series of voiced stops which lacks /g/; in 6 of these /*d/ is also missing,
leaving a "series" consisting of only /b/. Two languages (Gadsup, 608, and
Cashinahua, 813) have a voiced series consisting of /*d/ alone. All of
these languages have the velar place represented by the voiceless

35
Stops and affricates

counterpart /k/. There are only 3 languages with /g/ but without /b/. Two
of them also lack /*d/. Thus although many languages have no voiced stops,
among those which do, /g/ is more likely to be "missing" than /b/ or /*d/.
The implicational hierarchy for plain voiced plosives is thus: presence of
Iql implies presence of /*d/, which implies presence of /b/.
There is thus an asymmetrical distribution to be explained. One way of
expressing the asymmetry is to say that the bilabial place is disfavored
among voiceless plosives and the velar place is disfavored among voiced
plosives. But when the ratios of voiced to voiceless plosives at the 3
places are considered, it becomes unclear if this is an appropriate way to
describe the patterns. Among dental/alveolar plosives the voiced/voiceless
ratio is .63; since plosives with this place are relatively common in both
voiced and voiceless series, we may take this as the most typical ratio.
Now, the ratio at the velar place is .62, which is quite comparable. If
voicing was particularly disfavored at the velar place a lower voicing
ratio would be expected. At the bilabial place the ratio is .76,
substantially higher than at either the dental/alveolar or velar places.
That is, there is some factor which raises the number of voiced bilabials,
or some factor which lowers the number of voiceless bilabials, or, quite
possibly, both. Note that velar and dental/alveolar places retain their
positions relative to each other on the two hierarchies, whereas bilabial
place is at the bottom of the voiceless one and at the top of the voiced
one. If we assume that two processes which affect only bilabials are
involved, or, alternatively, that voiceless bilabials have a tendency to
turn into voiced ones, then both hierarchies can be accounted for.

There is some basis for believing that, in a language with a tendency

to devoice voiced plosives, the bilabials may resist the process more than
plosives at other places. This is because air can continue flowing into the
oral cavity during a voiced stop for a longer time before oral air pressure
equals subglottal pressure if the closure is at the lips. Hence voicing is
more readily sustained in a bilabial plosive than in any other.1 There is
no particular reason for believing that this relatively greater ease in
sustaining voicing during a bilabial closure should cause voiceless
bilabial plosives to change to voiced ones, however. The frequency of
"missing /p/" may therefore be the result of a more or less accidental
convergence of several trends, perhaps including voicing but also including
other processes such as sporadic shifts of /p/ to /<£/ as in Hausa (266;
Greenberg 1958; Newman 1977), etc. Stevens (cited by Ohala 1983: 195) has

36
 Patterns of sounds

suggested that the relatively weaker release burst of /p/ compared to the
other voiceless stops contributes to explaining its (comparative) rarity.
The intra-linguistic phenomenon of a "missing" /g/ can be accounted for
partly by the lesser frequency of use of the velar place in general (in
comparison with dental/alveolar place), and partly by the increased number
of languages with /b/, creating voiced series containing only /b/.
Although processes of the kinds referred to above may explain the
patterns of occurrence, it should also be noted that there are strong
associations between the occurrence of both the "gaps" where /p/ and /g/
are missing and particular areal or genetic groupings of languages. The
majority of the languages with a missing /p/ are from the Afro-Asiatic and
Nilo-Saharan families, which are contiguous in Africa, or from languages of
New Guinea. None of the languages in UPSID from several major language
families (e.g. Indo-European, Ural-Altaic, Sino-Tibetan) lacks /p/. The
languages with missing /g/ are also predominantly only from certain
particular areas: Austro-Asiatic and Austro-Thai languages in South East
Asia and languages from the Americas account for 19 of the 21 cases. So
although it is appropriate to continue the search for the reasons why gaps
appear in these particular places in plosive inventories and to consider
these patterns in evaluating historical reconstructions (Gamkrelidze and
Ivanov 1973; Hopper 1973), the extent to which these are local aberrations
should also be borne in mind. Similarities between these patterns in plosive
inventories and those in glottalic stop systems will be discussed further
in Chapter 7.

As for plosives in other than the major places, uvular plosives are
predominantly voiceless - there are 38 languages with /q/ but only 8 with
/G/ for a voicing ratio of only .21. Labial-velars on the other hand have a
voicing ratio of 1.05 since there is one more language with /gb/ than /£p/,
Iai (422), the only non-African language with labial-velars in the
database, and Temne (109) have /gb/ but not /l<p/, whereas Efik (119) has
/l<p/ alone. In all other cases, /l<p/ and /gB/ co-occur. The voicing ratio
is .76 for palatal plosives and .82 for retroflex plosives, both of which
indicate greater than usual tendency to voicing.

^•^ Secondary articulations with plosives

In the phonetic framework adopted for UPSID, appropriate segments can have
secondary articulations of the following types: labialization,
palatalization, velarization, pharyngealization. Of these, labialization is

37
Stops and affricates

the most common with plosives, although it is largely confined to velars

and uvulars. There are 38 languages with /k w / or 13.4% of the number that
have /k/. Fourteen languages have /g w /, which is 8.0% of the number with
/g/. Nine languages have /qw/ , which is 23.7% of the number with /q/. There
are also 3 languages with /q wh / and 4 with / G W / . Only 8 of the total of 94
labialized plosives in the whole database are not velar or uvular.
Palatalization is more often found with labials (17 examples in the
database) or with dental or alveolar stops (19 examples) than with velars
(14 cases). Of course, historically velars in "palatalizing" contexts tend
to shift their place of articulation and become palatal or palato-alveolar.
Palatalization at its most frequent - with plain voiceless dental/alveolar
plosives - only reaches 4.5% of the total of nonpalatalized ones. Velarized
plosives are altogether rare, and pharyngealized ones are only reported in
Arabic (250), Tuareg (257) and Shilha (256). Shilha has velar /k ? / as well
as pharyngealized voiced and voiceless dental/alveolars. The other two
languages have only the dentals among their pharyngealized stops.
None of the retroflex, palatal, palato-alveolar or labial-velar
plosives in the UPSID languages is reported as having any secondary
articulation. This may be due to the comparative rarity of plosives at
these positions. Since stops with secondary articulations are less common
than their simple counterparts, our survey may omit them by chance.

2.8 Affricates
The most common non-lateral and non-ejective affricates are palato-alveolar
in place and sibilant in nature. The next most frequent are dental or
alveolar sibilant affricates. Frequencies in the database of the major
types are given in Table 2.8.

Table 2.8 Frequency of the most common affricates in UPSID

Dental/alveolar Palato-alveolar
Plain voiceless /*ts/ 95 /tj/ 141
h h
Aspirated voiceless /*ts / 33 /tj / 42
Plain voiced /*dz/ 30 /d^/ 80

Note that there are s u b s t a n t i a l l y fewer examples of the voiced d e n t a l s and

a l v e o l a r s / * d z / than might be expected both on general grounds ( t h e voicing
r a t i o i s .32) and on the b a s i s of comparison with the palato-alveolars
(where the voicing ratio i s . 5 7 ) . However, a voicing r a t i o of about one

38
 Patterns of sounds

third is not unusual for fricatives (see Chapter 3 ) , and is characteristic

of the pair *z/*s with which the dental/alveolar affricates obviously share
acoustic and articulatory similarities. On the other hand, the voicing
ratio of palato-alveolar affricates is closer to the values found for
plosives. This may be related to the frequent historical descent of
palato-alveolar affricates from velar or palatal stops.
Affricates at other places of articulation are relatively rare: the
most common are palatal non-sibilant and retroflex sibilant voiceless
affricates, but less than 10 languages have such segments. It is clear that
the great majority of affricates are sibilants, and affrication rarely
occurs without assibilation. Of the grand total of 522 affricate segments
considered here, 485 are sibilant, that is, 92.9%.

2.9 Summary of generalizations on stops and affricates

A number of statements about the phonological patterns of stops and
affricates have been made in this chapter. The most important of these are
recapitulated here as summary statements, together with a fraction giving
the number of cases which conform to the generalization in question over
the number of relevant cases, and the percentage of conforming cases.
(i) All languages have stops. 317/317 100%.
(ii) A language is most likely to have two series of stops. 162/317
51.1%.
(iii) A language is highly likely to have a series of plain voiceless
stops. 291/317 91.8%.
(iv) If a language has only one stop series, that series is plain
voiceless. 49/50 98.0%.
(v) If a language has two stops series, it has a voice onset time
contrast between them. 144/162 88.9%.
(vi) If a language has three stop series it is most likely to have
two series with contrasting voice onset time and one
"glottalic" series. 50/76 65.8%.
(vii) A language is most likely to have stops at 3 places of
articulation. 171/317 53.9%.
(viii) A language most typically includes stops at bilabial, dental or
alveolar, and velar places of articulation. 312/317 98.4%.
(ix) Stops at other than the 3 places mentioned in (viii) do not
occur unless the language contrasts stops at 4 or more places.
206/210 98.1%.

39
Stops and affricates

(x) A language is most likely to have stops and/or affricates at

4 places of articulation. 139/317 43.8%.
(xi) An affricate segment is most typically a sibilant. 485/522 92.9%.
(xii) Doubly-articulated stops are labial-velar. 38/39 97.4%.
(xiii) If a language has /p/ then it has /k/, and if it has /k/ then it
has /*t/ (4 counterexamples in the UPSID sample),
(xiv) If a language has /g/ then it has /*d/, and if it has /*d/ then
it has / b/ (3 counterexamples in the UPSID sample).

Notes
1. This is confirmed both by direct experimentation (e.g. Ohala 1983) and
by modeling the aerodynamic properties of stop production at different
places (e.g Keating 1983). However, if expansion of the supraglottal
cavity during the closure is as great as shown in some production data
(e.g. in Westbury 1983), no differential effects of the place of
articulation would be expected to occur in stops with closure durations
typical of nongeminate plosives.

References
Gamkrelidze, T. and Ivanov, V. 1973. Sprachtypologie und die Rekonstruktion
der gemeinindogermanischen Verschldsse. Phonetica 27: 150-56.
Greenberg, J.H. 1958. The labial-consonants of Proto-Afro-Asiatic. Word 14:
295-302.
Hoffmann, C. 1963. A Grammar of Margi. Oxford University Press for the
International African Institute, London.
Hopper, P. 1973. Glottalized and murmured occlusives in Indo-European.
Glossa 7: 141-66.
Jakobson, R. and Halle, M. 1956. Fundamentals of Language. Mouton, The
Hague.
Keating, P. 1983. Physiological effects on stop consonant voicing. Paper
presented at 105th Meeting of the Acoustical Society of America,
Cincinnati.
Keating, P., Linker, W. and Huffman, M. 1983. Patterns in allophone
distribution for voiced and voiceless stops. Journal of Phonetics 11:
277-90.
Lindblom, B. 1983. Economy of speech gestures. In P.F. McNeilage (ed.), The
Production of Speech. Springer, New York: 217-45.
Lisker, L. and Abramson, A. 1964. A cross-language study of voicing in
initial stops: acoustical measurement. Word 20: 384-422.
Maddieson, I. 1983. The analysis of complex elements in Bura and the
syllable. Studies in African Linguistics 14:285-310.
Newman, P. 1977. Chadic classification and reconstruction (Afro-Asiatic
Linguistics 5.1). Undena, Malibu.
Ohala, J.J. 1983. The origin of sound patterns in vocal tract constraints.
In P.F. McNeilage (ed.), The Production of Speech. Springer, New York:
189-216.
Westbury, J. 1983. Enlargement of the supraglottal cavity and its relation
to stop consonant voicing. Journal of the Acoustical Society of America
73: 1322-36.

40
Fricatives

3.1 Introduction
While it is true that there are fricatives in nearly all of the world's
languages, there have been relatively few studies of their precise
distribution or of the patterns of occurrence which they show. Nonetheless,
many linguists have expressed beliefs about universal tendencies affecting
fricatives: for example, Fromkin and Rodman (1978: 331) say "If a language
has fricatives (most do), it will have an / s / M , and Bright (1978: 39) says:

It is natural for a language to have at least one sibilant, namely,

a voiceless alveolar [s]. Languages like Hawaiian, which lack even
this single sibilant, are rare (cf. Hockett 1955: 108).

Statements such as these have been made on the basis of personal experience
rather than on the kind of quantifiable research which is the only secure
foundation for conclusions which depend on frequency of occurrence.
In this chapter, we will examine the fricatives in the inventories of
the languages in UPSID, describe their frequency and patterns of
co-occurrence, and suggest some generalizations which apply. Where
possible, reasons why these generalizations hold will be suggested.
We adopt a conventional definition of fricatives, namely they are those
speech sounds produced by the narrow approximation of two articulators so
as to produce a turbulent airstream (Ladefoged 1971: 4 6 ) . Note that this
definition does not include the majority of sounds represented by the
symbol /h/. Sounds transcribed with /h/ have often been labeled "glottal
fricatives", but as Pike (1943) and others have pointed out, /h/ is
normally a voiceless counterpart of an abutting voiced segment (most often
a vowel). A brief discussion of the distribution of /h/ is given in section
3.9, but it is not included as one of the fricatives in any language in the
remainder of the discussion.

41
Fricatives

In UPSID, fricatives are recognized as occurring at 10 places of

articulation. Fricatives in the dental/alveolar region which are not
described in the source as being specifically dental or alveolar are
separately identified by a special variable. Fricatives also contrast in
voicing and with respect to other aspects of phonation type (aspiration,
breathiness, etc.). They may occur as ejectives or as laryngealized sounds,
and may have a secondary articulation in addition to their primary one. A
contrast is also found between central and lateral fricatives. These
phonetic attributes are also features of other classes of consonants.
However, one classificatory feature is restricted to fricatives and
affricates. This is sibilance. Sibilance is an acoustic property, referring
to a noise spectrum with comparatively strong energy at high frequencies.
In many cases sibilant and nonsibilant fricatives can be produced at the
same place of articulation. The articulatory difference probably involves
the profile of the tongue rather than the location of the narrowest
constriction. Sibilance is treated as an independent contrastive property
of fricatives, and all fricative and affricate segments have been assigned
a value of the variable "sibilant". Of course, the sibilant/nonsibilant
distinction is only actually contrastive for segments which are articulated
with the forward portion of the tongue.

3.2 The occurrence of fricatives

The great majority of the world's languages have at least one phoneme of
which the representative allophone stored in UPSID is a fricative. Of the
317 languages in the sample, 296 or 93.4% have one or more fricatives. We
may therefore say that it is typical of human languages that they have
fricatives.
There is an important exception to this generalization. Of the 21
languages which have no fricatives, the majority are Australian. There are
19 Australian languages in UPSID and of these 15 do not have any
fricatives. In a recent synthesis of work on Australian languages, Dixon
(1980) shows that there is no reason to reconstruct any fricatives for
Proto-Australian. Following Hale (1976), he also shows that those
Australian languages with fricatives have developed them in relatively
recent times from a medial laxing of stops (sometimes coupled with a loss
of contrastive vowel length where length was part of the conditioning
environment). It must be recognized that this family of languages is marked
by an abnormal rarity of phonemic fricatives. Moreover, fricative
allophones of other phonemes do not even seem to be typical.

42
 Patterns of sounds

As an aside prompted by this observation, we may note that the

existence of language groups with sharply different characteristics from
the rest of the world's known languages raises some doubts about the
authenticity of some of the claimed universal similarities between human
languages. In particular, it raises the question of to what extent
"universals" are a reflection of the accidents of survival and propagation
of languages, rather than of basic properties of human communication
systems. A possible approach to answering this question would involve
pursuing the more distant genetic relationships of the surviving languages.
If all the "normal" languages turn out to be more closely related to each
other than to the "deviant" languages, then apparent universals might be
due to inherited similarities. As little is yet known about the remoter
relationships of the recognized families of languages, this issue cannot be
pursued further at this time.

3.3 Number of fricatives per language

The number of fricatives of all types found in each of the UPSID languages
is summarized in Table 3.1.

Table 3.1 Number of fricatives in UPSID languages

No. of Fricatives No. of Languages % in Survey

0 21 6.6%
1 37 11.7%
2 62 19.6%
3 47 14.8%
4 37 11.7%
5 26 8.2%
6 28 8.8%
7 19 6.0%
8 20 6.3%
9 5 1.6%
10 4 1.3%
11 5 1.6%
12 2 0.6%
over 12 4 1.3%

The total number of fricatives in the languages surveyed ranges between 0

and 23, but the modal number is only 2, with about 20% of the languages

43
Fricatives

having that number. The mean is a little over 4. Only a relative handful of
languages (6.3%) have more than 8 fricatives. The structure of fricative
systems with different numbers of terms will be examined in section 3.8 •
However, first the overall frequency of fricatives of different types in
the UPSID data file will be discussed.
The most frequent fricative type is a voiceless sibilant made with the
front of the tongue. About 83% of all the languages have some kind of
"s-sound". This may be dental or alveolar in its place of articulation, but
for many purposes it is convenient to collapse this distinction. This
decision seems justified not only because of the obvious phonetic
similarity of dental and alveolar sibilants, but also because of the great
rarity of contrasts between /s/ and /s/. There are 262 languages with an
s-sound, but only 4 of these (Tzeltal 712, Karok 741, Dieguefto 743, and
Guarani 828) have both I si and /s/. It may also be noted that there are no
languages in the survey with both /z/ and /z/. In the remainder of this
chapter the following notation will be used: I si represents a voiceless
dental sibilant fricative, / s/ represents a voiceless alveolar sibilant
fricative, /"s"/ represents a sibilant with an unspecified dental or
alveolar place, and /*s/ will be used to refer to all types of s-sounds
together. The same conventions will be used to talk about other groups of
fricatives where similar distinctions need to be made.
Apart from the 4 languages mentioned above which have both I si and /si,
29 languages are reliably reported to have dental I si and 98 to have an
alveolar /s/. However, there are 131 languages with an s-sound which is not
specifically identified as dental or alveolar. In all probability, alveolar
sibilants are more common than dental ones, but this cannot be definitely
determined from our sample, as there are so many cases in which specific
information on place of articulation is missing from our source.
We may therefore state that /*s/ is the most common fricative, with Is I
probably being the more common member of the group l*sl • The number of
languages with /*s/ is 88.5% of those languages with fricatives. So Fromkin
and Rodman's statement that languages with fricatives have an s-sound has
relatively few exceptions. Bright"s assertion that it is natural for
languages to have an alveolar sibilant appears to be rather less
well-founded, unless it too is construed to refer to the class /*s/ rather
than strictly to the alveolar /s/•
The two next most frequent fricatives are the voiceless palato-alveolar
sibilant /J / with 146 cases, and the voiceless labio-dental fricative If I

44
 Patterns of sounds

with 135 cases. Their common occurrence is related in part to a general

preference for voiceless fricatives which will be discussed further below.
Next most frequent is the voiced counterpart of /*s/, namely /*z/, which is
the most common voiced fricative. There are 96 languages with one or other
of the members of /*z/. Next in frequency is hi (75 cases), followed by
hi (67 cases) and / y (51 cases).
There are only a little over a third as many cases of /*z/ as there are
of /*s/. Such a proportion seems to be approximately the usual ratio
between voiceless and voiced fricatives which are otherwise of the same
type. The predominance of voiceless fricatives over voiced ones can be
conveniently indicated by the ratio of voiced to voiceless, which we will
call the "voicing ratio". For the total number of fricatives in the file
the voicing ratio is 0.43. We may examine the separate ratios which are
substantially different from this overall ratio to determine at which
places voicing is more than usually favored or avoided in fricatives. The
frequency of occurrence of 11 pairs of voiced and voiceless fricatives,
arranged in descending order of frequency of the voiceless member of the
pair, is shown in Table 3.2, together with the voicing ratio.

Table 3.2 Relative frequency of voiced and voiceless fricatives

Voiceless Voiced "Voicing ratio"

/s/ 33) III 11) 0.33)
*s /"s'7 131} 266 *z /"z"/ 49 > 96 0.37} 0.36
hi 102) hi 36) 0.35)
III 146 /y 51 0.34
/f/ 135 hi 67 0.50
/x/ 75 hi 40 0.53
/x/ 29 /b/ 13 0.45
/<*>/ 21 /p/ 32 1.52
/4/ o) / iy o) -- )
*4 /"4"/ 17} 30 *I5 / " F f / 2} 7 0.11 } 0.23
/*/ 13) /iy 5) 0.38)
/9/ 18 /8/ 21 1.16
/s/ 17 /?/ 3 0.17
/C/ 16 /j/ 7 0.43
/V 13 /?/ 9 0.69

There is a high correlation between the voiceless and voiced frequencies

(r = .912, significant at better than .0001 level). In other words,

45
Fricatives

whether a given voiced fricative is more common than another can be largely
predicted from knowing which of the two voiceless counterparts is more
common, and vice versa. However, certain places of articulation favor
presence or absence of voicing more strongly than the general run of
fricatives. Most saliently, bilabial and dental nonsibilant fricatives
favor voicing, to the extent that there are more instances of / p/ than / <£>/
and more of / 5/ than /9/. It should also be noted that pharyngeals are more
than usually likely to be voiced, and that retroflex sibilant and lateral
fricatives (in the dental/alveolar area) are more than usually likely to be
voiceless.
It seems to be the case that / p/ and / 5/ in many languages are of
relatively recent origin and this may be related to their unexpectedly
frequent occurrence. They derive historically from 1 axing or weakening of
voiced plosives, as in Spanish (Oil), Atayal (407), and probably also in
languages such as Kaliai (421), Gadsup (608) and Diegueilo (743); and in a
number of cases result from loan phonology, as in Quechua (819) and
Mongolian (066). Compare Ferguson's findings in his study of the process
/d/ —> /5/. He concludes that this is "a highly context-sensitive
assimilatory process which is typically part of a larger schema of
spirantization (of voiced stops or all stops) which is relatively easily
diffused across languages" (Ferguson 1978: 437). Since the voiceless
counterparts are less common, such processes apparently either generate
fewer instances of the voiceless counterparts /<£/ and / 8 / , or once
generated, these sounds are rapidly transformed into something else. It is
tempting to hypothesize that /<£/ and / 9/ are changed into the very common
voiceless fricatives /f/ and /*s/. If this hypothesis is correct, we might
expect that fewer languages would have both members of the voiceless pairs
/$, f/ and /9, * s/ than the voiced pairs /p, v/ and /ft, * z/. However, there
are more languages with the voiceless pairs than the voiced ones. There are
13 languages with / 9, * s/ and 6 with /ft, * z/ of which 5 also have / 9, * s/.
With the "labial" pairs there is also a predominance of the voiceless pair,
although there are only 4 languages with both / <f>, f /. However, there are no
languages in UPSID with both /p, v/. The correct generalization to make in
this case may therefore be from a different perspective, namely, that the
absence of competing voiced fricatives may be an important factor in
facilitating the adoption of /p/ (and, less clearly /ft/) into the inventory
of a language. Gamkrelidze (1978) relates spirantization processes of this
type to the gaps created in the stop systems, but does not discuss possible

46
 Patterns of sounds

blocking of spirantization by existing fricative phonemes. Some of his

observations will be discussed further in section 3,5.

3.4 Implication of voicing in fricatives

Generally, the existence of a given voiced fricative in the inventory of a
language implies the presence of the voiceless counterpart in the
inventory. (The converse is very far from being true, as the voicing ratios
in Table 3.2 demonstrate.) This implicational relationship is subject to a
fair number of exceptions.
Table 3.3 shows the number of cases which are exceptions for each
voiced fricative in Table 3.2. There are altogether 78 cases out of 331
possible cases where the voiced member of a fricative pair occurs without
the voiceless member, a 23.6% exception rate. In other words, for these
fricatives, it is true about 76% of the time that the voiced member of the
pair is accompanied by the voiceless one in the inventory. Note that this
is not the number of languages involved, since a language may have more
than one of the voiced fricatives being examined.

Table 3.3 Voiced fricatives without voiceless fricative

Fricative pair Unpaired voiced fricatives/ Exceptions as

total voiced fricatives % of cases
* s , *z 0/96 0.0%
"h, ? 0/9 0.0%
J, 3 2/51 3.9%
*4, *I3 1/8 12.5%
f, v 11/51 21.5%
s, z 1/3 33.3%
x, b 5/14 35.7%
x, v 15/40 37.5%
9, 5 12/21 57.1%
g, J 5/7 71.4%
$, (3 24/32 75.0%

As Table 3.3 indicates, there are considerable disparities between the

various fricative pairs. Of the relatively large number of languages with
/*z/, none lack a corresponding voiceless /*s/> and the other common voiced
sibilant, /^/, only occurs very rarely without /J/. In addition to these
sibilants, voiced dental or alveolar lateral fricatives and voiced

47
Fricatives

labio-dental and pharyngeal fricatives occur only rarely or never without a

voiceless counterpart. On the other hand, bilabial, dental and palatal
nonsibilant fricatives are found to occur without a voiceless counterpart
more often than with one. Retroflex sibilant and velar and uvular
(nonsibilant) fricatives are in an intermediate group which occurs
moderately often unpaired with a voiceless counterpart.
The particularly high occurrence of voiced bilabial and dental
nonsibilant fricatives (in relation to the frequency of the voiceless
counterparts) has been commented on above. However, the occurrence of
unpaired bilabial and dental fricatives does not follow simply from the
frequency. The difference between /p/ and /#/ in frequency of occurrence is
only 11, but there are 24 languages in which /p/ occurs unpaired with / $ / .
Similarly, the difference in frequency of /5/ and /6/ is only 3 but there
are 12 languages with /5/ but without /8/. These are significantly larger
numbers than would be expected from the simple difference in frequency.
This reinforces the remarks above concerning the nature of the processes
which lead to the addition of /j3/ and /5/ to the inventory. Note that the
strength of the implication being discussed does not necessarily relate to
frequency of voicing. For pharyngeals the voicing ratio is .69 but presence
of /?/ always implies presence of / V .

Among the group of fricatives which are intermediate in Table 3.3,

voiced velar fricatives may also be derived from the same type of process
as produces voiced bilabial and dental fricatives from stops. However,
there are either more independent sources of the voiceless counterpart /x/
or more cases in which /x/ is developed in parallel with /y/> so that a
smaller percentage of exceptions to the generalization that the voiced
fricative implies the presence of the voiceless counterpart is found.
The voiced uvular fricative perhaps occurs relatively frequently
unpaired because of its relationship to the class of liquids. For example,
/b/ is the only candidate for a nonlateral "liquid" in Hebrew (253) and Sui
(403), though in the latter case the historical origin seems to be from a
velar stop (Li 1965).
The voiced palatal fricative / j / may derive from an approximant / j / and
seems unrelated in its occurrence to the voiceless fricative /c/. As for
the voiced labio-dental fricative /v/, this is perhaps found unpaired in
some cases because it derives from the common approximant /w/ rather than
from voicing of /f/.

48
 Patterns of sounds

3.5 Gamkrelidze's implicational proposals

Rather than a general implication that a voiced fricative implies the
presence of the voiceless equivalent, Gamkrelidze (1978) suggests that the
correct generalizations about velar and labial fricatives are opposites. He
writes:

...the presence of the voiceless labial fricative phoneme /f/ in a

system presupposes the simultaneous presence of the voiced labial
fricative phoneme /w - v/... and the presence of the voiced velar
fricative phoneme /y/ in a system presupposes the simultaneous
presence of the voiceless velar fricative /x/.

He defines /f/ to include both /$/ and /f/, and its "voiced counterpart" is
/w - v/ which includes /w/, /v/ and /|3/. The velar fricatives are
apparently conventionally defined. He claims that languages with /y/ but
not /x/ are "rare exceptions" (p. 30). Our finding is that over a third of
the languages with /y/ lack /x/. There are 75 languages in UPSID with /x/
or 23.7% of the languages. There are 40 languages with /y/ or 12.6%. The
percentage of languages in which these segments would co-occur if their
distribution were random is thus 3% or about 10 languages, instead of the
25 languages in which they are found together. We may therefore correctly
argue that presence of /y/ generally implies presence of /x/, although this
implication is not as strong as the implication holding between certain
other voiced fricatives and their voiceless counterparts.
On the other hand, there are 139 languages or 43.8% with /f/ or /<£/ or
both, and 288 languages with /w/ or /v/ or /|3/ or a bilabial or
labio-dental approximant, which Gamkrelidze would presumably also include
in his /w - v/ phoneme, meaning 90.9% of the languages have a
representative of his /w - v/ phoneme. Thus the percentage of languages
which do not have /w - v/ is only 9.1%. Hence only 4.0% (9.1% of 43.8%) of
the languages (some 12 or 13) would be expected to have his /f/ but not his
/w - v/ if these phonemes were randomly distributed. There are in fact 10
languages in UPSID, about 3.2%, which have his /f/ but not his /w - v/ as
construed here. Since this is close to the expected number, we must
conclude that presence of /f/ does not significantly affect the occurrence
of / w - v/ .

3.6 Predicting frequency from intensity

It might be considered plausible that the more frequent sounds in the
inventories of languages are those which have the greatest acoustic energy.
These sounds would seem to be the most desirable to incorporate in a

49
Fricatives

language if that language is going to have good transmission properties. As

we already know, the sibilant /*s/ is the most frequent fricative, and it
is the most intense. In view of this correlation it seems most appropriate
to test this theory of inventory structure in relation to fricatives. Of
course, what is being considered here is the intensity of the frication
noise, which is greater in voiceless fricatives than in voiced ones.
There is a surprising paucity of information on relative intensity of
fricatives, although some is available on the fricatives found in some
individual languages. The best language-independent study remains that by
Strevens (1960). Strevens examined a set of fricatives chosen "to provide a
wide coverage of different places of articulation and shapes of orifice".
These were [$, f, 0, s, J, g, x, )(] ([h] was also included but has been
discarded from consideration here). Two subjects produced a large number of
tokens of each fricative at various degrees of muscular effort. Intensity
readings were obtained and divided by subglottal air-pressure readings for
the same tokens obtained using a nasal catheter inserted into the
oesophagus. From this procedure a rank-order of intensity per unit
air-pressure was obtained.
In Table 3.4 the intensity ranking of these 8 fricatives is compared
with the rank-order of these fricatives by frequency in UPSID.

Table 3.4 Ranking of fricatives by intensity and frequency

Intensity ranking Frequency rankii

1 Q "s"
2 J J
3 X f
4 s X

5 X X
6 f $
7 0 0
8 Q

There is no significant rank-order correlation between these rankings

(Spearman's rho = .1429). It does not seem that there is any strong case to
be made that intensity predicts frequency. However, the tentative nature of
the ranking arrived at by Strevens should be stressed. The relatively
low rank for [s] and the high rank for [g] are counter to the intuitions of
most phoneticians.

50
 Patterns of sounds

3.7 Estimates of perceptual salience of fricatives

Several studies have reached conclusions concerning relative perceptual
salience of fricatives of various types. Perceptual salience may involve
factors other than overall intensity. It is generally measured by how
successfully a segment can be identified in a listening task. These studies
usually only involve comparatively limited inventories of fricatives from a
single language. For example, Wang and Bilger (1973) studied responses to
fricatives at the 4 places of articulation used in English. Of the 4
voiceless fricatives /f , 0, s, J/» they found that Is/ was least likely to
be misidentified as some other consonant. The overall ranking of the 4 is
/s, J, f, 9/ in terms of decreasing likelihood of being correctly
identified. However, as Goldstein (1977) has pointed out, it is not easy to
separate out what contribution is made by perceptual salience and what is
made by response bias when the data and subjects reflect real language
habits. Thus, /s/ is the most frequent of these 4 fricatives in English
texts (Roberts 1965; Carterette and Jones 1974). The other 3 segments are
all of relatively low frequency in English running text. In view of these
facts, we may attach more importance to the fact that /J/ is more reliably
perceived than /f/ and that /f/ in turn is more reliably perceived than /9/
than to the apparent salience of / s / , since this latter result may reflect
listeners' expectations or some other factor related to the greater
frequency of /s/ in English. Recall that /J/ is a little more frequent than
/f/ and both are more frequent than /9/ in the UPSID languages.

In an experiment which involved a more varied set of consonants,

including the fricatives / f , v , 0 , 5 , s , z , J , 5, ^ h /, Singh and Black
(1966) found relatively poor identification of / $ , 3, 5 h / by all 4 groups
of listeners used. Somewhat more reliably identified was the fricative /9/,
while /f, v, s, z, J/ were correctly identified on the very great majority
of occasions. As far as these results go, they do not contradict a
hypothesis that the more common fricatives are those which are more
salient, apart from the fact that /^/ is considerably more common than /9/.
However, the conditions of this experiment did not provide a means of
discriminating between those fricative types which are relatively frequent.
Moreover, there are problems in interpreting the results of this study. The
stimuli used were spoken by native speakers of English, Hindi, Japanese and
Arabic. Speakers were given a short training session to enable them to
approximate those sounds that were unfamiliar to them. These recordings

51
Fricatives

were then played to groups of listeners who were native speakers of the
same 4 languages. The listeners were given a brief training in the
transcriptional conventions they were expected to use in recording their
responses. Singh and Black do not report if subjects had to reach any
criterion as either speakers or listeners before they were included in the
study, and the design of their experiment confounds the influence of
language-specific variation in the production of segments designated by the
same phonetic symbol with language-dependent listener biases and with
inherent salience.
As yet it seems the experimental work has not been done which can
provide a sounder basis for forming a ranking of fricatives by salience.
The few indications from this literature are suggestive of possible
correlation of fricative salience with overall frequency in the world's
languages.

3.8 The structure of systems of fricatives

Languages with one fricative
Of the 37 languages with only one fricative, 31 have some kind of /*s/ (3
/s/, 14 /"s'7, 14 / s / ) . We may therefore conclude that in languages which
have one fricative, it is overwhelmingly probable that that single
fricative will be a voiceless dental or alveolar sibilant. There are
various ways in which a quantitative evaluation can be added to this
comment: one way is to compare the ratio 31/37 to the total of /*s/'s
divided by the total number of fricatives, i.e. 265/1123 (leaving aside
fricatives with other than a single primary articulation and plain
voiceless or voiced phonation). Of the fricatives in languages with only
one fricative, the percentage of /*s/'s is 83.8%; of all the fricatives in
the UPSID file the percentage of /*s/'s is 23.6%. In one sense, then, /*s/
is unusually common in these languages. However, the frequency of /*s/ in
these languages is simply a function of the overall commonness of /*s/ and
is not a result of any special association of /*s/ with small fricative
inventories. There are 261 languages with some kind of /*s/ in their
inventory, so a random selection of 37 of the 317 languages from UPSID
would yield an expected number of 261/317 x 37, that is, 30.4 as the number
of languages within this subsample that would have an /*s/ segment. Since
31 of the 37 one-fricative languages have an /*s/, we must conclude that
/*s/ is the usual fricative in languages with only one fricative simply
because /*s/ is the most common fricative.

52
 Patterns of sounds

Languages with 2 fricatives

Of the 62 languages with 2 fricatives, 56 have an /*s/ as expected. The
most frequently found pair of fricatives is /*s, f/; there are 16 languages
with this pairing. The percentage of these languages with this pair (25.8%)
is actually less than the percentage which would be expected from
consideration of the frequency of these 2 fricatives overall (35.1%). The
next most frequent pair is /*s, J7, found in 11 of the languages. This pair
is even more infrequent (17.7%) than would be expected (37.9%). One might
be tempted to argue that the pairing of these 2 voiceless sibilants is
avoided in languages with a small number of fricatives, perhaps because
they are phonetically similar. However, besides these 11, there are another
4 languages in this group with somewhat similar pairings of voiceless
sibilants and 6 with a pair of sibilants which involve a voicing or other
phonation type contrast. Altogether, then, there are 21 languages with 2
sibilants as their sole fricatives. These data provide a weak foundation
for the argument that pairs of sibilants are avoided because they are too
similar.

What can be said about the structure of these fricative inventories is

that there is a significant tendency to prefer voiceless segments and to
avoid voiced and voiceless pairs of fricatives at the same place of
articulation. Only 16 of the languages contain any voiced fricatives, and
only 3 contain a pair in a voicing contrast (/*s, * z / ) . Approximately 32%
of the plain fricatives in UPSID are voiced, so a random distribution would
lead us to expect about 39 or so of the total of 124 fricatives in the
languages with only 2 fricatives to be voiced ones. In fact only 17 are
voiced, well under half the number predicted by this calculation. As for
the avoidance of voiced/voiceless pairs, a predicted number of cases can be
obtained by multiplying together the separate probabilities that languages
will have the voiceless and the voiced member of the pair. For the 3 most
common places at which fricatives occur (dental/alveolar, labio-dental and
palato-alveolar) the probabilities of the members of the fricative pair
occurring together in a language are .249, .090 and .074. We may interpret
these probabilities to mean that we should expect some 15 languages with
the pair /*s, * z/> 5 or 6 languages with /f, v/ and 4 or 5 with /J, y from
among a set of 62 languages. In the languages with 2 fricatives, there are
only 3 languages with /*s, * z / and none with either /f, v/ or /J, 5/. It is
therefore safe to say that voicing contrasts are generally avoided in
fricative systems which contain only 2 fricatives.

53
Fricatives

Languages with 3 fricatives

In the 47 languages with 3 fricatives in UPSID the most common set is
/f , *s , J/, i.e. the grouping containing the 3 most common fricatives. It
may be noted that there are 16 languages whose set of 3 fricatives includes
the pair /f, *s/ and 22 with the pair /* s , J7. These pairs are, of course,
the pairs found most often in the languages with only 2 fricatives. In 29
of the 47 languages only voiceless fricatives are found: the number of
voiced fricatives is less than would be expected by chance but not by such
a large margin as in the 2-fricative systems. There are 8 languages in
which there is a pair of fricatives contrasting in voicing (6 /*s, * z / ; 1
/f , v/; 1 /J , 5/). This is about two fifths of the expected number,
compared with only about one eighth of the expected number among those
languages with 2 fricatives.

Languages with 4 fricatives

When the size of a fricative inventory reaches 4 terms, there is a marked
change in the extent to which voicing is used contrastively compared with
smaller fricative inventories. Of the 37 languages concerned, 24 contain at
least one pair of fricatives contrasting in voicing, and the most common
system is one which contains 2 such pairs, that is, /f, v, * s , * z / . There
are 10 languages with this inventory, plus 7 additional languages with /f,
v/ and 6 with /*s, * z / . With 4 fricatives, instead of fewer voicing pairs
than would be expected if fricatives occurred randomly according to their
overall frequencies, there are more than expected. The expected number of
languages with /*s , * z / is only about 9, rather than the attested 16, and
the expected number of languages with /f , v/ is about 3, rather than the
attested 17. The probability of all 4 erf these fricatives occurring together
in a language is approximately .022, so only 1 language with all 4 would be
expected in any random selection of 37, rather than 10.
The situation may be summed up informally as follows. There is a strong
preference for voiceless fricatives in general, so languages with only a
few fricatives are likely to have voiceless ones. However, if voicing is
introduced into the fricative system, there is a strong likelihood that it
will be extended to fricatives at more than one place of articulation. In
an<
other words, the common system /f , *s / is converted to /f , v , *s , *z / i
the language becomes one with 4 fricatives. This may be expressed as a
condition on linguistic processes in the following fashion: a process
(historical or otherwise) which voices fricatives is more likely to apply
to a class of fricatives than to a single fricative.

54
 Patterns of sounds

Note, however, that palato-alveolar fricatives are rarely involved in

pairwise contrasts in these smaller fricative inventories. There are 2
occurrences of the pair /J, ^/ in languages with 4 fricatives and only 1 in
the languages with 3 fricatives. This is despite the fact that /J/ is very
common in languages with from 2 to 4 fricatives, occurring in 54 (37.0%) of
the 146 languages concerned. Since the 2-fricative inventory /*s, J7 is
common, the 4-fricative inventory /*s, * z , J, 5/ might also have been
expected to be frequently encountered, but, as will be shown below, the
pair /J, y occurs preferentially in languages with more than 4 fricatives.
It is not clear why this should be so.

Languages with 5 fricatives

Of the 26 languages with 5 fricatives, 15 have a voicing contrast between
one or more pairs of fricatives at the same place of articulation. Of
these, 11 have /*s, * z / , 6 have /f, v/ and 3 have /J, 5/. These are all
above the number of cases expected if distribution was random (6, 2 and 2
respectively). Five languages, all from the Niger-Kordofanian family,
include both pairs /*s, * z , f, v/. Nonetheless there are still 7 of these
5-fricative languages which have no voiced fricative at all.

Languages with 6 fricatives

The most common inventory for languages with 6 fricatives is one with 3
pairs of fricatives contrasting in voicing, namely / f, v, * s , * z , J, 5 /.
There are only 5 languages with this inventory among the 29 concerned but
they come from 3 different language families, Indo-European, Ural-Altaic
and Niger-Kordofanian, so this is not just a genetically restricted
phenomenon. Five languages with 6 fricatives have no voiced fricatives in
their inventory, but the remaining 23 contain at least one pair contrasting
in voicing. In inventories of this size, there are far more cases of / J, 5 /
than would be expected - 9 rather than the predicted 2. Other pairs of
fricatives in voicing contrast are also found in greater diversity.
Altogether pairs at 8 different places of articulation are found in one or
more of these languages.

Languages with 7 fricatives

The typical fricative inventory with 7 terms contains 4 voiceless
fricatives and 3 voiced ones which are paired with 3 of the voiceless set.
Six of the 19 languages concerned contain the subset of 3 pairs /f, v,

55
Fricatives

*s, * z , J, 3/. Quileute (732) exceptionally has all of its 7 fricatives

voiceless, having the set /s, J, x> xw$ )(, x w > *'• Tllis inven
tory
demonstrates several areal/genetic trends which are common to the languages
of the Americas, These are the tendencies to have fewer voiced fricatives
than other languages with fricative inventories of comparable size, to
avoid labial (bilabial or labio-dental) fricatives and to have several back
fricatives (velar and/or uvular).
It may be noted that voiced uvular fricatives are not found at all
commonly until there are at least 7 fricatives in the inventory. Of the
total of 14 voiced uvular fricatives in the UPSID languages, 3 are in
languages with 7 fricatives, 6 in languages with 8 fricatives and 3 in
languages with more than 8.

Languages with 8 fricatives

Twelve of the 20 languages with 8 fricatives have systems in which there
are 4 voiceless and 4 voiced fricatives, although only in 4 of these is it
a simple case of 4 pairs in voicing contrast. Nevertheless, in 13 of the
20, there are at least 3 pairs of fricatives at the same place of
articulation with a voicing contrast. So voicing contrasts are often a
major part of the content of fricative inventories of this size. However,
there are 8-fricative systems which contain no voiced fricatives, for
example in Iraqw (260) and Nootka (730). Yuchi (757) has 4 pairs of
fricatives which are plain voiceless and ejective; there are no voiced
fricatives in this language. Note that 2 of these languages are Amerindian.

Languages with 9 or more fricatives

The largest fricative inventories are fairly naturally a rather
heterogeneous set, and frequently include some component such as a widely
distributed secondary articulation, or a set of ejective or contrastively
long fricatives. The largest system of fricatives in a language in UPSID
that does not have one of these factors is Margi (268). This language has
12 fricatives composed of 6 pairs in voicing contrast, including voiced and
voiceless lateral fricatives. Kabardian (911) has a total of 22 fricatives,
reaching this magnitude by use of 7 places of articulation, 3 phonation
types (voiceless, ejective, voiced) and 2 secondary articulations
(palatalization and labialization). However, the combinatory possibilities
of these various factors are very far from being exhaustively exploited.
For example, as in a considerable number of other languages, labialization

56
 Patterns of sounds

of fricatives in Kabardian is restricted to velars and uvulars. (Of the

total number of fricatives of all types with labialization, 33 are velar or
uvular in place; only 6 are reported at other places. There are also 4
w
cases of /h /.)

3.9 The phoneme /h/

A large number of languages have a phoneme /h/. Over 63% of the languages
have a segment of this type. A small number, 13, have a "voiced h". Of
these, only 2 have a contrast of I hi with /ft/. Clearly there is a strong
tendency for /h/ and /fi/ not to co-occur. It should also be noted that /n/
(phonetically usually a breathy voiced onset or offset to a normally voiced
segment) is not limited to languages which have other breathy voiced
segments (for examples, see Kashmiri, 018, Punjabi, 019, Karen, 516,
Wichita, 755, Kabardian, 911). The classification of segments such as /h/
and /fi/ has been the subject of considerable disagreement. Although they
have often been considered as members of the class of fricatives, some
linguists have preferred to put them into a special class of "laryngeals"
together with /?/, and others have emphasized their similarity to vowels
and approximants. The most extensive discussion of the issue is by
Merlingen (1977) who reviews data from about 600 languages. His not very
helpful conclusion is that there are several different types of /h/. This
conclusion is largely based on arguments that /h/ can be associated with
different kinds of consonants through examining pattern conformity and
morphophonemic alternations. Although /h/ is often associated with
fricatives such as /x, *s> if by such patterns, it is also often associated
with stops such as /k, p, ?/ as well as with nasals and other types of
sounds. It remains likely that a relatively uniform phonetic
characterization of /h/ is possible, but that it is best made in terms of
similarities to the properties of abutting segments.

3.10 Conclusion
This chapter has presented data on the occurrence of fricatives in the
languages in the UPSID sample. It has shown that patterns of frequency bear
some relationship to what is known about relative intensity and salience of
fricatives. It has documented more extensively some of the relationships
between certain types of fricatives, such as the general tendency for
voiced fricatives to occur only if the voiceless counterpart also is
present. However, important limitations on this relationship apply when

57
Fricatives

certain places of articulation are concerned. The structure of fricative

inventories of different sizes was also examined and the connections
between inventory size and types of fricatives encountered were
investigated.

References
Bright, W. 1978. Sibilants and naturalness in aboriginal California.
Journal of California Anthropology, Papers in Linguistics 1: 39-63.
Carterette, E.C. and Jones, M.H. 1974. Informal Speech: Alphabetic and
Phonemic Texts with Statistical Analyses and Tables. University of
California Press, Berkeley.
Dixon, R.M.W. 1980. The Languages of Australia. Cambridge University Press,
Cambridge.
Ferguson, C.A. 1978. Phonological processes. In J.H. Greenberg et al.
(eds.), Universals of Human Language, Vol. 2_9 Phonology. Stanford
University Press, Stanford: 403-42.
Fromkin, V.A. and Rodman, R. 1978. An Introduction to Language, 2nd ed.
Holt, Rinehart and Winston, New York.
Gamkrelidze, T.V. 1978. On the correlation of stops and fricatives in a
phonological system. In J.H. Greenberg et al. (eds.), Universals of
Human Language, Vol. 2, Phonology. Stanford University Press, Stanford:
9-46\
Goldstein, L. 1977. Three Studies in Speech Perception: Features, Relative
Salience and Bias (UCLA Working Papers in Phonetics 39). University of
California, Los Angeles.
Hale, K.L. 1976. Phonological developments in Northern Paman languages. In
P. Sutton (ed.), Languages of Cape York. Australian Institute of
Aboriginal Studies, Canberra: 7-49.
Hockett, C.F. 1955. A Manual of Phonology (IJAL Monographs 11). Indiana
University, Bloomington.
Ladefoged, P. 1971. Preliminaries to Linguistic Phonetics. University of
Chicago Press, Chicago.
Li, F-K. 1965. The Tai and the Kam-Sui languages. Lingua 14: 148-79.
Merlingen, W. 1977. Artikulation und Phonematik des H. Verband der
wissenschaftlichen Gesellschaften Osterreichs, Vienna.
Pike, K.L. 1943. Phonetics. University of Michigan Press, Ann Arbor.
Roberts, A.H. 1965. A. Statistical Linguistic Analysis o£ American English.
Mouton, The Hague.
Singh, S. and Black, J.W. 1966. A study of twenty-six intervocalic
consonants as spoken and recognized by four language groups. Journal of
the Acoustical Society o£ America 39: 372-87.
Strevens, P. 1960. Spectra of fricative noise in human speech. Language and
Speech 3: 32-49.
Wang, M.D. and Bilger, R.C. 1973. Consonant confusions in noise: a study of
perceptual features. Journal of the Acoustical Society of America 54:
1248-66.

58
Nasals

4.1 Introduction
Unlike most of the types of segments which are the topics of chapters in
this volume, nasals have been the subject of a good earlier survey. In
fact, the study by Ferguson (1963) on nasals has served in many ways as the
model of an article on universals of segment types. Ferguson's article
provided a major part of the stimulus for the organization of a conference
devoted to nasals and nasalization (Ferguson, Hyman and Ohala 1975). For
this reason, this chapter will largely take the form of a discussion of the
various "assumptions about nasals" put forward by Ferguson, checking them
against the data in UPSID to provide the quantification which is lacking in
Ferguson's article and is only partially provided by Crothers (1975)
working from an early version of the Stanford Phonology Archive. However,
before this discussion, some summary information on the types of nasal
consonants included in the UPSID data file will be presented.

4.2 Types of nasals

There are 1057 nasals in the file, of which the great majority, 934 or
88.4%, are simple plain voiced nasals. There are a further 50 nasals which
are plain voiced but have distinctive length or a secondary articulation.
Only 36 or 3.4% are voiceless, a number almost equaled by the
laryngealized nasals (34 or 3.2%). There are also 3 breathy voiced nasals
reported (from Hindi-Urdu, 016, and !Xu, 918). The distribution of nasals
by place of articulation is given in Table 4.1. (A similar convention to
that used in other chapters is used for nasals in the dental/alveolar
region, that is, /"n"/ indicates a nasal whose exact place (dental or
alveolar) is unspecified in the source, and /*n/ represents the total class
of dental and alveolar nasals.)

59
Nasals

Table 4.1 Nasal segments in UPSID

Place No, of simple plain No, of other Total

voiced nasals nasals
55
Dental
"Dent/Alv"
n
"n" 155 1 316
10 |
23 > 40
65|
178} 356
Alveolar n 106) 113)
Bilabial m 299 47 346
Velar Q 167 23 190
Palatal P 107 11 108
Retroflex n 20 1 21
Palato-alveolar n 17 0 17
Labial-velar mo 6 1 7
Labio-dental 1 0 1
Dental-palatal GP 1 0 1

Evidently, nasals in the dental/alveolar area are the most common, but
bilabials are also very frequent. Although velar nasals are far from rare,
they are much less frequent than velar stops: for example there are 283
UPSID languages with / k/ but only 167 with /Q/. On the other hand, palatal
nasals are more common than might be expected from comparison with stop
frequencies: there are 107 languages with /p/ but only 41 with /c/. Note
that no example of a phonemic uvular nasal is reported from the UPSID
languages (38 languages have / q / ) . Of course, pharyngeal and glottal places
of articulation are ruled out for nasals because of articulatory
constraints.
Among the relatively small number of nasals that are not simple plain
voiced ones, there seem to be three reliable patterns. The first is that
there is a preferential association between bilabial place and
voicelessness: of the 36 voiceless nasals, 11 are bilabial. This is a
greater proportion than is found at other places of articulation. For
example, compare the ratio of languages with voiceless nasals to those with
plain voiced nasals at the bilabial place, 11/299 (.037), with the
comparable ratio among combined dental/alveolar nasals, 8/305 (.026). The
other two observations relate to secondary articulations: palatalization is
associated more with bilabial place of articulation than with other places,
whereas labialization is associated with velar place of articulation. Of 10
voiced nasals with palatalization, 6 are bilabial; of 10 voiced nasals with
labialization, 7 are velar. In proportion to the number of plain voiced
nasals these are far more frequent than either secondary articulation is

60
 Patterns of sounds

with dental and alveolar nasals. The ratio of palatalized bilabial nasals
to plain voiced bilabials is .020, whereas the ratio of palatalized
dental/alveolar nasals to plain dental/alveolar nasals is .003. The ratio
of labialized velar nasals to plain velar nasals is .042, but the ratio of
labialized dental/alveolar nasals to plain dental/alveolar nasals is only
.003. Labialization is also most commonly combined with velar place of
articulation in other segment types, such as stops and fricatives, but
palatalization is not especially associated with bilabial place in other
segment types.

4.3 Ferguson's "Assumptions about nasals": primary nasal consonants

We now turn to a discussion of the proposals for universals about nasals
put forward by Ferguson. He suggests that "every language has at least one
primary nasal consonant (PNC) in its inventory". By a PNC, Ferguson means a
phoneme of which the most characteristic allophone is a voiced nasal
continuant. As he recognizes, there are several languages that do not have
any such consonant, but their number is small. Hockett (1955) had already
drawn attention to some of these, and others have been pointed out in the
literature (Thompson and Thompson 1972; Le Saout 1973; Bentick 1975). In
UPSID there are 10 languages without any primary nasal consonants, 4 of
which have no phonemic nasal or nasalized segments of any kind. These 4 are
Rotokas (625), Quileute (732), Puget Sound (734), and Mura (802). Of these,
two have very small inventories altogether, and the other two are from the
Northwest Coast area of North America - the area from which the first
exceptions to this generalization were reported. The remaining 6 languages
without nasals have prenasalized stops or nasalized vowels: Kpelle (103),
Barasano (832) and Tucano (834) each have a series of 6 nasalized vowels,
m 1 Q
Hakka (502) has prenasalized stops / b , " V , g/, and Apinaye (809) and
Siriono (829) have both a series of prenasalized stops and a series of
nasalized vowels. Nasal consonant phones occur in this latter group of
languages, as well as in several of the other languages without primary
nasal phonemes referred to in the literature.
Although there are exceptions, it remains true that the very great
majority of languages do have one or more nasals - almost 97% of the UPSID
sample. None of the UPSID languages has more than 6 PNC's; this is the same
number as the maximum number of place contrasts for stops (see Chapter 2 ) .
The number of languages with various numbers of PNC's is shown in Table
4.2. Since only plain voiced nasals are counted here, this is essentially
equivalent to a tabulation of the number of place contrasts for nasals.

61
Nasals

Table 4.2 Number of PNC's in UPSID languages

Number of PNC's Number of languages % of sample

0 10 3.2%
1 7 2.2%
2 101 31.9%
3 95 30.0%
4 83 26.2%
5 14 4.4%
6 7 2.2%

Ferguson's second suggestion is that if a language has only one PNC it

will be an apical nasal, i.e. one with a dental or alveolar place of
articulation. This class of segments is represented here by /*n/« Of the 7
languages listed in UPSID as having only one nasal, 5 do indeed have /*n/,
namely Tlingit (701), Chipewyan (703), Wichita (755), Yuchi (757) and S.
Nambiquara (816). Taoripi (623), with /m/, is an exception, as is Mixtec
(728) with / Q / (although this segment is arguably a surface derivative and
Mixtec could be another language with no basic nasals). Although Ferguson's
claim about languages with only one nasal is thus true of the majority of
relevant UPSID languages, it is perhaps more important to stress that
having only one nasal is an aberrant pattern. Languages with only one nasal
are rare, even rarer than languages with no nasals (7 vs 10). Moreover,
most (13 of 17) of the languages with either one or no nasals are American
languages. The relatively frequent deficiency of nasals within the American
languages seems to indicate somewhat of an areal and/or genetic trait which
makes this group of languages stand out from the rest of the world's
languages. However, too much should not be made of this as a characteristic
of American languages, as there are another 76 in UPSID which have 2 or
more nasals in their inventory. This is about 85% of this group, compared
with about 98% of the non-American languages in UPSID - a smaller
percentage but still an overwhelming majority.

Ferguson's claim can be interpreted in a slightly different sense,

namely, that every language with any nasals has an /*n/. Exceptions to this
claim are certainly rare. They include Taoripi and Mixtec, mentioned above,
as well as Wapishana (822), which has bilabial and palato-alveolar nasals
/m, n /. All other languages in UPSID with any nasals have an /*n/ , and
there are 13 with both dental and alveolar nasals. In other words, 302 or
96% of the languages in the sample have /*n/.

62
 Patterns of sounds

Ferguson's third suggestion is that a language with 2 PNC's will have

an /ml• All of the languages except Wapishana with 2 PNC's have a bilabial
and one of the members of /*n/. Specifically, 21 are reported with /m, n/,
31 with /m, n/ and 48 with /m, "n"/. Out of the total of 101 languages with
2 PNC's, 28 also have one or more nasal vowels, and 15 have additional
nasal segments which have a difference in phonation type or length or
secondary articulation. Apart from the usual plain voiced nasals, 8 have
laryngealized nasals, and 2 of these have voiceless nasals; one language
has breathy voiced nasals. In 6 cases there are length contrasts among the
nasals. In Finnish (053) a long velar nasal appears as well as long and
short /m, n/. This segment is found as the "weak" alternant of /ok/ in
consonant gradation (Austerlitz 1967: 24) and is a rather marginal member
of the phoneme inventory; consequently it could be maintained that the
primary nasal consonants in Finnish are only the bilabial and dental ones.
The relationship between primary and secondary nasals is a subject that
will be returned to later.
If Ferguson's third claim is interpreted as implying that all languages
with at least 2 PNC's have an /m/, then there is only one marginal
exception in UPSID. All of the languages with from 2 to 6 PNC's have /m/
except Irish (001), but this exception is more a result of definitional
criteria than anything else. Irish has a contrasting pair of bilabial
nasals with differing secondary articulations - palatalized opposed to
labialized and velarized. It has no plain bilabial nasal, and because of
this it violates this interpretation of Ferguson's generalization.
Crothers (1975) conflates Ferguson's first three observations into a
single general rule, namely, "nearly all languages have contrasting labial
and dental nasals" [presumably, dental or alveolar is intended]. Almost 94%
of the UPSID languages (297 out of 317) have both / m / and /*n/ in accord
with this rule.
Ferguson does not extend his generalizations of this type beyond a
second PNC. However, it is evident that the most usual third nasal is
velar, 65 of the 95 languages with 3 PNC's having /m, *n, Q/, i.e. 68.4% of
this group. However, there is a very strong minor pattern which includes a
palatal or palato-alveolar nasal instead of a velar one, there being 27
languages representative of this pattern, or 28.4%. The greater frequency
of / Q / is more evident in the UPSID sample than in the sample used by
Crothers, where / Q / and /p/ are somewhat closer to equal frequency. Of the
remaining languages, two, Pashto (014) and Telugu (902), have a retroflex

63
Nasals

nasal, and Irish (001) has no bilabial PNC but has both.palatal and velar
nasals. However, this language is perhaps best regarded as a language with
4 PNC's, including /m/, for the reasons outlined above.
In the spirit of Ferguson's observations, we may add that a language
with 4 PNC's has a palatal nasal. In a very large majority of cases, 75 of
83 or about 90%, the inventory of nasals in a language with 4 PNC's
consists of /m, *n, *p, Q/ (palato-alveolar and "pre-palatal" nasals are
treated as members of the same class as /p/ and the class is designated by
/ * p / ) . The only reasonably common alternative is /m, *n, n, Q/, with a
retroflex in place of the palatal. Five languages have this pattern.
Of the 14 languages with 5 PNC's, all have /m, Q/. Five contrast dental
and alveolar nasals, but there is no common overall pattern, with the fifth
nasal being either retroflex (3 cases) or palatal (2 cases). Four
languages, all Niger-Kordofanian, have labial-velar nasals as part of the
set /m, *n, p, Q, mo/. Four languages have /m, *n, n, *p, Q/, with a
retroflex nasal. Teke (127) is reported with a contrast of bilabial and
labiodental nasals, /m, m/, as well as /n, p, Q/. It is the only language
with a labiodental nasal in the UPSID file.
The languages with 6 PNC's are essentially uniform, having /m, n, n, n,
p, Q/ as their system, apart from Iai (422) which has /mr)/ but no contrast
between dental and alveolar. It is noteworthy that 11 of the 13 languages
in the sample with both dental and alveolar nasals are languages with 5 or
6 PNC's and that 10 of these are Australian languages. It is valid to
observe that only languages with a large number of nasals contrast dental
and alveolar places, but also important to note the genetic bias in the
distribution. Australian languages typically have more nasals than
languages of other families; of the 19 Australian languages in UPSID, 11
have 5 or 6 nasals. Less than 3% of the languages in UPSID have more than 4
nasals but the percentage of Australian languages with that many is 57.9%.

4.4 Primary nasals and obstruents

The remaining observations by Ferguson relate to the connection between
primary nasal consonants and other types of segments, specifically
obstruents and secondary nasal consonants. He suggests that the number of
PNC's in any language is never greater than the number of obstruents. There
are no counterexamples to this claim in UPSID. In the majority of
languages, plain voiced nasals are found at a subset of the places at which
obstruents occur. In other words, the presence of a nasal at a given place

64
 Patterns of sounds

usually implies the presence of an obstruent (specifically a plosive or

affricate) at the same place of articulation. A frequent discrepancy is
that obstruents occur at palato-alveolar place but the closest nasal is a
palatal. This is close enough to be considered a matching place.
In 6 languages, Ewe (114), Efik (119), Songhai (200), Javanese (409),
Chamorro (416), and Auca (818), /p/ occurs but there is no palatal or
palato-alveolar obstruent. But these 6 languages are the only exceptions to
the stronger claim that PNC's may only occur if there is at least one
obstruent with the same or similar place of articulation. That all the
exceptions to this claim concern palatal nasals may indicate that there is
some factor favoring palatal nasals such that they do not need the support
of a obstruent at the same place. In this connection, it may be noted that
a voiced nasal is the most common strictly palatal consonant after the
palatal approximant /j7.

4.5 Secondary nasal consonants

Ferguson classifies nasal consonants other than plain voiced nasals as
secondary nasal consonants (SNC's). This class includes nasals with a
secondary articulation, or with an unusual phonation type as well as pre-
or post-nasalized obstruents. He suggests that no language has SNC's unless
it also has one or more PNC's, and that the number of SNC's in any language
is never greater than the number of PNC's. Of the languages which have any
secondary nasals of any kind, a relatively large proportion violate
Fergusons's assumption, but this occurs most often because they may have
more than one type of secondary nasal. An example is Lakkia (401) which has
3 PNC's /m, "n", Q/ but has 3 voiceless nasals plus voiced velars with
palatalization and labialization for a total of 5 SNC's. It seems more
valid to interpret this suggestion as applying separately to each of the
types of secondary nasals, and for this reason the discussion will be
organized to deal with each type of secondary nasal separately.

Other phonation types

Nasals with unusual phonation types, that is, voiceless, laryngealized or
breathy voiced nasals, do not occur unless a plain voiced counterpart occurs
in the language. With few exceptions, if there is a series of any of these
types of SNC's it will have a member corresponding to each of the nasals of
the plain voiced series (PNC's). The number of such SNC's thus is never
greater in any one series than the number of PNC's but is often equal. A

65
Nasals

language with two such series thus usually has more SNC's than PNC's. The
UPSID examples are Sedang (304), Sui (403), Klamath (707), and Otomi (716).
These languages have both voiceless and laryngealized series of nasals
which together add up to more than the number of PNC's. !Xu (918) has
laryngealized and breathy voiced bilabial nasals, but has plain voiced
nasals at bilabial, alveolar and velar places. No language in UPSID has
more nasals in any one series with an unusual phonation type than it has
plain voiced nasals.

Nasals with secondary articulation

The general frequency of nasals with secondary articulations was briefly
discussed in section 4.2 above. Within a given language, there are two
general constraints on the occurrence of such nasals: no nasal with a
secondary articulation occurs unless a simple nasal occurs at the same
place of articulation, and none occurs unless consonants of another type
also occur with the same secondary articulation and in the same place of
articulation. For example, /nr / does not occur unless /m/ occurs, and a
segment such as /p / or /b J / also occurs. There are 16 languages in the
sample which are relevant to these claims of which 14 conform to both. The
other 2 languages each violate one of them only. !Xu (918) has a
pharyngealized valar nasal, but no other pharyngealized consonants,
although it has a series of pharyngealized vowels. Irish (001) has the
appropriate other secondarily-articulated consonants but has no simple /m/
although both /m / and /m<w/ occur. The interpretation of the Irish system
is complicated because of the pervasive nature of the morphophonemic
distinction between "palatal" and "velar" variants of consonants which
coincides poorly with a phonetic characterization of the segments. Only at
the bilabial place of articulation are both variants produced with
differing secondary articulations.

Long nasals
Similar conditions apply to distinctively long nasals as apply to those
with secondary articulations. In the 13 languages with long nasals, a
matching short nasal occurs with each long nasal, except for the case of
/Q:/ in Finnish (053), discussed in section 4.3 above. There are, however,
3 languages where one or more long nasals occur but matching long
consonants of other types are lacking. These are Chuvash (060), Ocaina
(805) and !Xu (918). (Chuvash does have distinctively long approximants but

66
 Patterns of sounds

has no long bilabial consonants apart from /mi/.) Despite the exceptions,
there is a very significant association between length in the nasal system
and length in other consonant subinventories. For example, there are 11
languages with /m:/ and 9 languages with either /p:/ or /b:/ or both. In
these two small groups, 8 languages are the same, i.e. have both /m:/ and a
long bilabial plosive. The probability that segments of this type would
co-occur if they were randomly distributed is about .0001. Applying this to
our sample means that we would expect about one language in 900 to have
this co-occurrence, instead of the observed frequency of about 1 in 40.
Because long consonants of all types are fairly rare in occurrence, it
is not really possible to assign a direction to the association with long
nasals. What we can observe is simply that the sort of phonetic and
distributional facts which argue for interpretation of length as a property
of unitary phonemes tend to be common to several types of segments in any
relevant language.
For each of the types of SNC's considered above, the fact that
languages have an equal or larger number of PNC's than they have in any one
series of SNC's follows from the condition that a matching simple voiced
nasal at the same place of articulation is presupposed by the presence of
any SNC.

Prenasalized obstruents
There are 19 languages in UPSID with prenasalized segments. Such segments
are all voiced obstruents. There are 66 prenasalized plosives, 7
affricates, and 3 fricatives. (There is one language, Aranda (362), with a
series of nasally-released stops.) The languages with prenasalized
obstruents would provide rather a large percentage of exceptions to the
claim that languages do not have more SNC's than PNC's, since 10 of them
have more prenasalized obstruents than PNC's. However, it seems more
appropriate to relate prenasalized obstruents to the other obstruent series
of the language, rather than considering them in relation to the nasals.
There are several reasons for this. First, there are languages such as
Hakka (502), Apinaye (809) and Siriono (829) which have no PNC's but do
have a prenasalized plosive series. These 3 languages also lack a plain
voiced plosive series; in a sense the prenasalized stops take its place.
Secondly, there are languages which have both a series of PNC's and a
prenasalized stop series in place of a simple voiced series of stops.
Besides lacking a simple voiced obstruent series, such languages may have

67
Nasals

prenasalized obstruents at places where there is no simple nasal but where

there is a voiceless obstruent. Examples include Washkuk (602), which has
PNC's at bilabial, alveolar and palatal places but has no velar nasal.
It has prenasalized voiced plosives including the velars /q, Q W / and
w
voiceless plosives, including /k, k /, but no plain voiced plosives.
Similarly, Ngizim (269) has 3 nasals but 5 prenasalized plosives, having
velar and labialized velar prenasalized plosives but no velar nasal. Paez
(804) also has 3 PNC's none of which is velar, whereas its prenasalized
plosive series includes a velar. Ngizim and Paez have no plain voiced
stops, but they have voiceless stops at places matching their prenasalized
obstruents. The third type of example is provided by Sara (217), which has
PNC's, voiced plosives and prenasalized plosives, but it has no velar PNC
while there are velars in the plosive series.
Because of the pattern of exceptions to Ferguson's suggestion
concerning SNC's, even in the modified form advanced here, prenasalized
obstruents should be excluded from the definition of SNC's. They are
instead subject to the generalization that a prenasalized obstruent does
not occur unless an obstruent of the same class (plosive, affricate,
fricative) occurs without prenasalization at the same (or similar) place of
articulation. That is, / m b/ does not occur unless, say, /p/ occurs in the
same language. There are only 2 languages which violate this
generalization. Washkuk (602) has / m b/ and / m b w / among its prenasalized
stops but has no other bilabial stops, although it does have bilabial
fricatives /<£/ and /*w/. Sara (217) has a prenasalized palatal stop, but no
simple palatal stop; however it does have a palato-alveolar affricate /d^/.
All other cases conform; in fact, the more typical language with
prenasalized stops or affricates has one at each place where a stop or
affricate of another kind occurs. It is not unusual to find languages which
have full matching series of voiced plosives, prenasalized obstruents and
PNC's, such as Gbeya (129), Yulu (216), and Alawa (354), and languages
with full matching series of PNC's and prenasalized stops but no voiced
stops, such as Nambakaengo (626) and Kaliai (421).

Nasalized clicks
Ferguson included nasalized clicks among the secondary nasal consonants. A
nasal escape of air can accompany any click production, and the nasal
component can be either voiced or voiceless. Nasalized clicks are found as
part of a set of click series and do not appear to be related to the number

68
 Patterns of sounds

or nature of the system of true nasals in the languages concerned.

Languages with clicks in their inventory seem to include nasalized ones
among them; all 3 UPSID languages with clicks, Zulu (126), Nama (913) and
!Xu (918), have one or more series of nasalized clicks.

4.6 Restated generalizations about nasals

The observations concerning nasals above can be rephrased into an updated
series of generalizations. Following each statement the number of
conforming cases is given before the slash, and the number of relevant
cases is given after the slash. These numbers may refer to segments or to
languages, depending on the form of the statement. The percentage of cases
which conform to each generalization is then given.
(i) If a segment is a nasal, it is voiced. 984/1057 93.1%.
(ii) A palatalized nasal is likely to be bilabial. 6/12 50.0%.
(iii) A labialized nasal is likely to be velar. 7/11 63.6%.
(iv) A voiceless nasal is more likely to have a bilabial place of
articulation than any other place. 11/36 30.6%.
(v) Most languages have at least one nasal. 307/317 96.8%.
(vi) A language with any nasals has /*n/. 304/307 99.0%.
(vii) The presence of /m/ in a language implies the presence of /*n/.
297/299 99.3%.
(viii) The presence of either / Q / or /*p/ in a language implies the
presence of both /m/ and /*n/. 197/200 98.5%.
(ix) The presence of both /n/ and /n/ in a language implies the
presence of both /m/ and / Q / . 12/13 92.3%.
(x) The presence of a nasal at any given place of articulation
implies the presence of an obstruent at a similar place.
(Number of counterexamples not counted.)
(xi) The presence of a voiceless, laryngealized or breathy voiced
nasal implies the presence of a plain voiced nasal with the
same place of articulation. 73/73 100%.
(xii) The presence of a nasal with a secondary articulation implies
the presence of a simple nasal at the same place of
articulation. 24/26 92.3%.
(xiii) The presence of a nasal with a secondary articulation implies
the presence of at least one other consonant of another type
at the same place of articulation and with the same secondary
articulation. 25/26 96.2%.

69
Nasals

(xiv) The presence of a prenasalized obstruent implies the presence of

a simple obstruent of the same class at a similar place of
articulation. 73/76 96.1%.

4.7 Explanations for nasal patterns

Linguists are generally more successful at observing facts of the kind
represented by (i) - (xiv) above than in explaining them, but explanation
is obviously the more important goal. An important start has been made by
Ohala (1975) toward developing explanations for nasal sound patterns where
these may have a phonetic basis, but there is much that remains to be done.
We will briefly review what seems to be the current state of research.
It is often assumed that the more distinctive speech sounds are those
that will be used more frequently in the world's languages, since this will
achieve the most successful transmission of a message. Some of the patterns
of occurrence of nasals may be accounted for on the basis of ideas of this
kind.
In the first place, the fact that so many languages have any nasals at
all fits well with the fact that nasals have been shown to be highly
distinctive. That is, they are rarely subject to confusion with other types
of consonants and are reliably identified as nasals. (Some relevant studies
are Miller and Nicely 1955; Singh and Black 1966; Shepard 1972; Goldstein
1977.) The distinctiveness of nasals as a class means that there is value
in incorporating such sounds into any language, and that they are likely to
be retained over time. Plentiful evidence is available from historical
phonology that nasals are among the most stable of sounds diachronically.
It is less easy to argue on grounds of distinctiveness for the presence
of multiple nasals in a language; although nasals as a class are distinct,
they are prone to confusion within the class. Malecot (1956) and Nord
(1976) are among those who have shown that nasals with different places of
articulation are poorly distinguished in terms of the nasal murmur itself
(as opposed to in terms of the transitions to adjoining vowels).
Nonetheless to have up to 4 nasals in a single language is a common
occurrence. As long as there are clear vowel transitions, this is perhaps
not so surprising, but lack of distinctiveness conspires with articulatory
convenience in limiting the contrastiveness of nasals in consonant clusters
in most languages, where they are required to be homorganic (Greenberg
1978: 253).

70
 Patterns of sounds

There are some indications in favor of a perceptual basis to the

hierarchy of preferred places of articulation for nasals, which runs in
descending order of popularity: dental/alveolar, bilabial, velar, palatal.
Although phonetic and perceptual studies have generally excluded palatal
nasals from the inventory of sounds studied, there is some evidence for the
relative reliability of identification of / m, n, p/• Zee (1981) found that
/n/ is most likely to be correctly identified, while House (1957) found
that /p/ is more likely to be misidentified as one of the other nasals.
More specifically, Zee reports that /p/ is especially likely to be
identified as /n/ after / j / , whereas Malecot agrees with House that /m/ is
most likely to be identified as /n / before /i/. Zee also finds that after
I \l or /e/> /m/ is likely to be heard as /n/. These various reports
indicate that there is more chance for misperceptions, perhaps aided by
coarticulation effects in real speech, to favor /n/ at the expense of other
nasals, and for /p/ to be less favored than either /m/ or /n / • These data
should be used with some caution as they may reflect the native language
habits of the subjects used in the experiments, rather than more universal
human perceptual proclivities. However, they do suggest that the most
widely occurring nasals are the most reliably identified ones.
As for the fact that nasals are rarely voiceless, Ohala (1975) has
pointed out both that voiceless nasals can barely be distinguished from
each other on the basis of their noise spectrum, and that in languages which
have voiceless nasals, there is a voiced portion before the vowel begins
and it is this portion which makes them recognizable. It seems that, if
voiceless nasals are developed in a language, they must either be lost
through indistinctness, or have their voicing restored to them in time.
Other patterns, such as those involving the secondary articulations, or
the relationships between nasals and obstruents, are likely to be related
to wider issues concerning the exploitation of possible phonetic contrasts
that go beyond a consideration of nasal consonants in isolation.

References
Austerlitz, R. 1967. The distributional identification of Finnish
morphophonemes. Language 43: 20-33.
Bentick, J. 1975. Le niaboua, langue sans consonnes nasales. Annales de
l'Universite d'Abidjan, Serie H,Linguistique 8: 5-14.
Crothers, J. 1975. Nasal consonant systems. In C.A. Ferguson, L.M. Hyman,
and J.J. Hyman (eds.), Nasalfest. Stanford University, Stanford: 153-66.
Ferguson, C.A. 1963. Assumptions about nasals: a sample study in
phonological universals. In J.H. Greenberg (ed.) Universals of Language.
M.I.T. Press, Cambridge: 53-60.

71
 Nasals

Ferguson, C.A., Hyman, L.M. and Ohala, J.J. (eds.) 1975. Nasalfest (Papers
from a symposium on nasals and nasalization). Stanford University,
Stanford.
Goldstein, L. 1977. Bias and asymmetry in speech perception. UCLA Working
Papers in Phonetics 39: 62-87.
Greenberg, J.H. 1978. Some generalizations concerning initial and final
consonant clusters. In J.H. Greenberg et al. (eds.), Universals of Human
Language, Vol. 2, Phonology. Stanford University Press, Stanford:
243-79.
Hockett, C.F. 1955. A Manual of Phonology (IJAL Monographs 11). Indiana
University, Bloomington.
House, A.S. 1957. Analog studies of nasal consonants. Journal of Speech and
Hearing Disorders 22: 190-204.
Le Saoiit, J. 1973. Langues sans consonnes nasales. Annales de l'Universite
d'Abidjan, Serie H,Linguistique 6: 179-205.
Malecot, A. 1956. Acoustic cues for nasal consonants: an experimental study
involving tape-splicing techniques. Language 32: 274-84.
Miller, G. and Nicely, P. 1955. An analysis of perceptual confusions among
English consonants. Journal of the Acoustical Society of America 27:
338-52.
Nord, L. 1976. Perceptual experiments with nasals. Speech Transmission
Laboratory, Quarterly Status and Progress Report (Royal Institute of
Technology, Stockholm) 2-3: 5-8.
Ohala, J.J. 1975. Phonetic explanations for nasal sound patterns. In C.A.
Ferguson, L.M. Hyman, and J.J. Hyman (eds.), Nasalfest. Stanford
University, Stanford: 289-316.
Shepard, R.N. 1972. Psychological representation of speech sounds. In E.E.
David and P.B. Denes (eds.) Human Communication: A Unified View.
McGraw-Hill, New York: 67-1 lY.
Singh, S. and Black, J.W. 1966. A study of twenty-six intervocalic
consonants as spoken and recognized by four language groups. Journal of
the Acoustical Society of America 39: 372-87.
Thompson, L.C. and Thompson, M.T. 1972. Language universals, nasals and the
Northwest Coast. In M.E. Smith (ed.), Studies in Linguistics in Honor of
George Trager. Mouton, The Hague: 441-56.
Zee, E. 1981. Effect of vowel quality on perception of post-vocalic nasal
consonants in noise. Journal of Phonetics 9: 35-48.

72
Liquids

5.1 Introduction
For reasons to do with both acoustic similarities and common phonological
patterning, laterals and r-sounds have been grouped together as "liquids"
in phonetic tradition. Although the similarities involve principally the
voiced non-fricative segments concerned (Goschel 1972; Bhat 1974), the term
liquid in this paper will be applied to all lateral segments except lateral
clicks and to all sounds that are included in the somewhat heterogeneous
class of r-sounds. The core membership of this latter class consists of
apical and uvular trills, taps and flaps.1 Added to this core are a
variety of fricative and approximant sounds which seem acoustically or
articulatorily similar, or which are related by diachronic processes
(Lindau-Webb 1980).

5.2 Overall frequency of liquids

Using the definition above, almost all languages in the UPSID sample of 317
languages have at least one liquid, that is 95.9% of them do. Most
languages, that is 72.6%, have more than one liquid. Details of the
distribution are given in Table 5.1. The patterns found for systems of each
size will be analyzed in Section 5.5 below, following an examination of the
occurrence of particular types of liquids. As far as the two major classes
of liquids are concerned, some 81.4% of languages have one or more lateral
segments, whereas 76.0% have one or more r-sounds. The total number of
laterals occurring in the surveyed languages is much greater than the
difference between these percentages would suggest, since there are more
languages with greater numbers of laterals. In fact, about 57% of the
liquids reported are laterals.

73
Liquids
Table 5.1 Number of liquids in UPSID languages

No. of liquids No. of ;uages Percent in survey

0 13 4.1%
1 74 23.3%
2 130 41.0%
3 46 14.5%
4 29 9.1%
5 14 4.4%
6 8 2.5%
7 2 0.6%
1U 1 0.3%

5.3 Laterals
Types of laterals
The laterals which occur may be grouped under four broad headings: lateral
approximants, taps/flaps, fricatives, and affricates. The occurrence of
these types is summarized in Tables 5.2 - 5.5 where frequencies are
expressed in terms of the percentage of the total number of laterals
counted in the survey (418). Approximant lateral types are shown in Table
5.2. Plain voiced approximant laterals are by far the most common type of
lateral. Other types of approximant laterals are rare and only occur in
inventories in which a plain type appears.

Table 5.2 Approximant lateral types

Number Percent
of laterals
Plain voiced 313 74.7%
Plain voiceless 11 2.6%
Laryngealized voiced 8 1.9%
Breathy voiced 1 0.2%

333 79.7%

There may be some doubt as to whether linguists have consistently reported

on the distinction between voiceless approximant laterals and voiceless
fricative laterals, but the distinction is an important one to attempt to

74
 Patterns of sounds

maintain. Unlike voiceless approximants, voiceless lateral fricatives are

reported in inventories that contain no voiced lateral approximant
(Tlingit, 701, Nootka, 730, Puget Sound Salish, 734, Chukchi, 908,
Kabardian, 911) so there may be an important distributional difference
between the two types of sounds. Moreover, lateral fricatives are quite
likely to have affricate allophones, as in Zulu (126), Tolowa (704), Hupa
(705), Nez Perce (706), Totonac (713), and Alabama (759). Voiceless lateral
approximants do not seem to vary in this way.
Lateral taps and flaps are reported fairly rarely. The numbers are
given in Table 5.3. There is reason to believe that segments of this type
may be more frequent than the reports indicate. Few phonetic manuals
mention their occurrence and there is some evidence (of an anecdotal
nature) that field linguists often have difficulty in recognising them for
what they are. In such cases lateral taps/flaps are likely to be reported
as r-sounds (this is perhaps so in Tiwa, 740) or as approximant laterals
(as perhaps has happened in Zoque, 711).

Table 5.3 Lateral taps and flaps

Number Percent
of laterals

Plain voiced 9 2.2%

Laryngealized voiced 1 0.2%

10 2.4%

The few flaps reported are all voiced. Lateral fricatives, on the other
hand, are far more likely to be voiceless than voiced, as Table 5.4 shows.

Table 5.4 Lateral fricative types

Number Percent
of laterals
Plain voiceless 34 8.1%
Plain voiced 9 2.2%
Ejective voiceless 2 0.5%

45 10.8%

Two languages in the survey have a voiced lateral fricative without a

corresponding voiceless fricative. In Kanakuru (270) /15/ only occurs "in a

75
Liquids

few words" and there is an approximant / I / , but in Pashto (014) a

"prepalatal" fricative lateral is the only lateral. These examples show
that it would be unsound to propose that voiced fricative laterals only
occur with voiceless fricative laterals. Nonetheless, voiced lateral
fricatives are distinctly rare. As noted in chapter 3, lateral fricatives
are even more likely to be voiceless than fricatives of other types.
The ejective lateral fricatives reported (in Tlingit 701, Yuchi 757)
are restricted to languages with glottalic consonants of other types and in
both cases a non-ejective voiceless lateral fricative also occurs. Ejective
laterals are far more likely to be affricates than fricatives, but one
3 5
language in the survey, Tlingit 701, has both /"4 '7 and /"t4 "/ in its
inventory. This language also lacks a "normal" voiced lateral approximant.
Table 5.5 reports the types of lateral affricates in the survey.

Table 5.5 Lateral affricates

Number Percent
of laterals
Ejective voiceless 14 J • J/o

Plain voiceless 9 2.2%

Plain voiced 4 1.0%
Aspirated voiceless 2 0.5%

29 6.9%

A lateral affricate is almost always voiceless. It has a very high

probability of being ejective and, in fact, is the only segment type so far
reported with such a high probability of being ejective. The ejective
lateral affricates only occur in languages which have a non-ejective
lateral fricative or affricate (or both).

Places of articulation for laterals

Laterals are almost all articulated with the tip or blade of the tongue,
but it is not possible to be very much more specific than that. Due to
inadequacy of data, no attempt was made in UPSID to distinguish between
apical and laminal articulations, and, in a very large number of instances,
it is not possible to determine if a segment is dental or alveolar in place
of articulation. A very large number of laterals is thus classified as
"unspecified dental or alveolar". The data on place of articulation is
given in Table 5.6 (percentages add to more than 100 because of rounding).

76
 Patterns of sounds

Table 5.6 Primary and secondary places of articulation for laterals

Primary place of articulation

Dental "Dental/ Alveolar Palato- Retrof. Palatal Velar Alveolar
Alveolar" alveolar velar ?
No secondary
artic. 31 178 132 8 28 15 1 3
Pal'lzed 3 1 5 0 0 0 0 0
Velarized 0 7 3 0 0 1 0 0
Phar'gzed 0 1 1 0 0 0 0 0

Totals 34 187 141 8 28 16 1 3

% of lat. 8.1% 44.7% 33.7% 1.9% 6.7% 3.8% 0.2% 0.7%

About 87% of all laterals are produced in the dental/alveolar region.

Probably alveolar laterals are more frequent than dentals, but this cannot
be determined with certainty from the available data. Retroflex laterals
are the next most frequent and, of course, these too are tip or blade
articulations. Laterals made with the body of the tongue are comparatively
unusual. Among them, palatals are most frequent. Velar laterals are
extremely rare — only one example appears in the survey, in Yagaria (609).
A few other cases are known from other languages, including the New Guinea
languages Melpa, Mid-Waghi and Kanite (Ladefoged, Cochran and Disner 1977)
and the Chadic language Kotoko (Paul Newman, personal communication). The 3
complex lateral segments reported to have both velar and dental/alveolar
articulations are all somewhat obscurely described. All 3 are voiceless and
fricative or affricate, being interpreted as /x4/, /k4/ (Ashuslay, 814) and
/k4'/ (Zulu, 126). Apart from this rather dubious instance, there are no
significant interactions between lateral manners and places of
articulations. All types of laterals are predominantly dental or alveolar,
and at all places the most common type of lateral is a plain voiced lateral
approximant.
The preference for tip or blade articulations for laterals is
presumably related to the greater opportunity to provide a free air passage
behind the front closure if the body of the tongue is not involved in the
articulation. Tongue-body laterals are probably more subject to processes
resulting in their diachronic loss, as for example in the phenomenon known
as "yeismo" in Spanish, which has transformed the palatal lateral Ikl into
a palatal approximant / j / in many dialects (Guitarte 1971).

77
Liquids

Summary statements on lateral segments

The observations above on lateral segments suggest that the following
substantive generalizations can be made. After each statement, the number
of conforming cases in the inventory is given before the number of
potentially relevant cases, together with a calculation of the percentage
of cases which conform.
(i) A lateral segment is most likely to be articulated with tongue
tip or blade. 392/418 93.8%.
(ii) A lateral segment is most likely to be voiced. 347/418 83.0%.
(iii) A lateral segment is most likely to be an approximant. 333/418
79.7%.
(iv) A fricative lateral is most likely to be voiceless. 36/45 80.0%.
(v) A voiceless lateral is likely to be fricative. 36/72 50.0%.
(vi) An ejective lateral is most likely to be an affricate. 14/16
87.5%.
(vii) A lateral affricate is likely to be ejective. 14/29 48.3%.
It should be noted that (iv) and (v) are independent observations. The
explanation for (iv) is likely to be related to the greater salience of
voiceless fricatives in general, compared to their voiced counterparts
(Goldstein 1977). The explanation for (v) is likely to be related to the
greater salience of voiceless fricatives over voiceless approximants, added
to the fact that the only places favored for affricates in general are
palatal and palato-alveolar, which are disfavored places for laterals.
There may, however, also be a special reason why diachronic processes act
to retain few voiced lateral fricatives or voiceless lateral approximants,
that is, these sounds are difficult to distinguish from non-lateral
counterparts (e.g. [15] — > [3]; [|] — > [h]).

5.4 R-sounds
Types of r-sounds
The sources used to compile UPSID fail to specify the manner of
articulation of segments represented by /r/ in 34 instances or 10.8% of the
316 r-sounds. These will be droppped from consideration in the analysis of
the types of r-sounds below. It should be remembered that ignorance of how
this group of sounds should be distributed into the various classes below
adds a measure of uncertainty to some of the conclusions reached in this
section.

78
 Patterns of sounds

Of the remaining 282 r-sounds, the largest number are reported as

trills. 2 The data is reported in Table 5.7.

Table 5.7 Trills by phonation type

Number Percent
of r-sounds
Plain voiced 130 46.1%
Plain voiceless 3 1.1%
Laryngealized voiced 1 0.4%

134 47.5%

Obviously, trills are overwhelmingly voiced. The same is true of the next
most frequent class of r-sound, consisting of taps and flaps. All reported
taps/flaps are voiced, with only 3 being other than plain. The numbers are
given in Table 5.8. Although fewer taps/flaps are reported than trills, the
difference is less than the number of r-sounds with unspecified manner and
hence it is not possible to conclude which of these types of r-sounds is
actually most common in the languages of the world. In any case, trills and
taps/flaps are closely related sound types (often both appear as allophones
of the same phoneme) both of which involve an interruption of the flow of
air through the oral cavity. It may be observed that about 86% of those
r-sounds with specified manner are "interrupted".3

Table 5.8 Taps/flaps by manner

Number Percent
of r-sounds
Plain voiced 104 36.9%
Plain voiceless 2 0.7%
Laryngealized voiced 1 0.4%
Voiced fricated 1 0.4%

108 38.3%

The numbers of continuant r-sounds are shown in Table 5.9. As with

other types of r-sounds, voicing is obviously the norm. Note that the
fricatives included in the class of r-sounds are those non-sibilants made
with the tongue tip or blade (apart from the dental /9/ and / 5 / ) . Despite

79
Liquids

the transcriptional convention which suggests that a voiced uvular

fricative (/b/) is an r-sound of some kind, it is not considered to be one
here. A voiceless uvular fricative (/%/) is never considered to be an
r-sound, and, in general, uvular fricatives seem to be most closely
connected with velar fricatives. For those interested, there are 16 voiced
uvular fricatives in UPSID languages.

Table 5.9 Continuant r-sounds

Number Percent
of r-sounds
Voiced approximant 28 9.9%
Voiced fricative 8 2.8%
Voiceless fricative 2 0.7%

38 13.5%

Substantially more approximants are reported than fricatives. However there

is considerable room for doubting the validity of the reporting of
fricative r's. The criteria for reporting a sound as "some kind of r"
rather than, say, as a voiced retroflex sibilant Izl are obscure and may
not reflect a phonetic difference between Izl and / j / , but instead be based
on phonotactic considerations or other non-phonetic characteristics
(including orthographic convention).

Places of articulation for r-sounds

As with laterals, there is a considerable number of instances where the
place of articulation is only known to be in the dental/alveolar region
somewhere. The tabulation for places of articulation for all of the 316
r-sounds in UPSID in Table 5.10 thus includes an unspecified
dental/alveolar column. Of the 34 r-sounds with an unspecified manner, 27
also fall into this underspecified place category. However, unlike with
laterals, it can be stated that the most common place of articulation for
an r-sound is alveolar. There is a larger number of alveolar r-sounds (141)
than any other category. Note that there are only a few r-sounds reported
as dental, only about one fourteenth as many as those specified as
alveolar, so that even if all the unspecified place segments were in fact
dental, there would still be fewer dental than alveolar r-sounds. And, of
course, the unspecified category is far more likely to be correctly divided

80
 Patterns of sounds

in something approaching the ratio of the segments with known dental or

alveolar place of articulation, i.e. in about a 1:14 ratio.

Table 5.10 Place of articulation, all r-sounds

Primary place of articulation

Dental "Dental/ Alveolar Palato- Retroflex Uvular
alveolar u alveolar
No secondary
articulation 9 118 135 2 38 3
Palatalized 1 3 4 0 0 0
Velarized 0 1 1 0 0 0
Pharyngealized 0 0 1 0 0 0

Totals 10 122 141 2 38 3

Percent 3.2% 38.6% 44.6% 0.6% 12.0% 0.9%

The only other reasonably frequently occurring place for r-sounds is

retroflex. Uvulars are quite rare (and mainly restricted to prestige
dialects of Western European languages). Uvular trills are included in the
survey from French (010), German (004) and Batak (413) and a uvular
approximant occurs in Eastern Armenian (022). Other places are largely
ruled out by definition.

Interaction between place and manner

There is an important interaction between place and manner which can be
seen from Table 5.11, juxtaposing frequency of alveolar and retroflex types
of r's sorted by manners. (The unspecified dental/alveolar category has a
distribution by manner similar to the alveolars.)

Table 5.11 Interaction of place and manner in r-sounds

Alveolar Retroflex
Number Percent Number Percent
Interrupted: trills 62 44.9% 5 13.9%
taps/flaps 62 44.9% 11 30.6%
Continuant: approximants 11 8.0% 15 41.7%
fricatives 3 2.2% 5 13.9%

81
Liquids

Whereas an r-sound with an alveolar place is one of the interrupted types

(trill/tap/flap) in almost 90% of cases, retroflex r-sounds are most
commonly found as approximants in the languages surveyed and, relative to
alveolars, are very rare as trills but more common as fricatives. Within
the approximant class, 15 of 28 instances are reported as retroflex.4

Summary statements on r-sounds

The analysis in the preceding section suggests that the following
substantive generalizations about r-sounds can be made:
(i) An r-sound is most likely to be voiced. 308/316 97.5%.
(ii) An r-sound is most likely to be dental or alveolar. 273/316
86.4%.
(iii) An r-sound is most likely to be interrupted. 244/282 86.5%.
(iv) A retroflex r-sound is likely to be a continuant. 20/38 52.6%.
(v) An approximant r-sound is likely to be retroflex. 15/28 53.6%.
(vi) A fricative r-sound is likely to be retroflex. 5/10 50.0%.
The explanation for (i) probably needs to be different for different types
of r-sounds. Tap/flap durations are very short and connected speech is
mostly voiced: the voicing of taps/flaps may result from an inability to
switch back and forth from voicing to lack of voicing quickly enough.
Approximants are predominantly voiced (see chapter 6 ) , probably because
voiceless approximants are poorly distinguished from each other and tend to
fall together as the undifferentiated voiceless vowel / h/. However, there
seems to be no equivalent reason for voiceless trills to be so rare. Trills
generally have two or three contacts at a rate of vibration of about 28Hz
(Ladefoged, Cochran and Disner 1977) requiring a substantial duration (on
the order of 100 msec)• They have some similarity in production to
obstruents, which are preferentially voiceless (about 60% of stops are
voiceless). But perhaps there is some factor in the aerodynamic conditions
required for trilling which leads to the preference for voicing because of
the associated reduction in air-flow.

5.5 Structure of liquid systems

Languages in the survey have up to 6 laterals and up to 4 r-sounds,
although it is most typical to have only one of each. The number of
languages with the various numbers of liquids are given in Table 5.12.
About 31% of the languages have 2 or more laterals, but only 19% have as
many r-sounds. Of the 230 languages with 2 or more liquids, 96 have more

82
 Patterns of sounds

laterals than r-sounds, whereas only 23 have more r-sounds than laterals.
The remainder have equal numbers of the 2 major types of liquids.

Table 5.12 Numbers of laterals and r-sounds per language

No. of laterals 0 1 2 3 4 5 6
No. of languages 58 157 63 23 9 3 3
Percent 18.3% 49.5% 19.9% 7.3% 2.8% 0.9% 0.9%

No. of r-sounds 0 1 2 3 4
No. of languages 74 183 51 8 1
Percent 23.3% 57.7% 16.1% 2.5% 0.3%

Languages with one liquid (74, 23.3%)

In view of the greater overall frequency of laterals, it is rather
surprising that languages having one liquid are more likely to have an
r-sound (42) rather than a lateral (32), with the most frequent type of
r-sound in these languages reported as a voiced flap (28 cases). However,
in a number of cases, both lateral and non-lateral allophones occur, e.g.
in Nasioi (624) the flap Id occurs as a lateral before / u/ or /o/ , in
Barasano (832) the alveolar flap /r/ has flapped nasal and lateral
allophones, with lateral flaps occurring before central and back vowels and
[ r] before front vowels and in all word-final environments. Similar
allophonic variation is found in Tucano (834) but in this language it is a
preceding central or back vowel which conditions the occurrence of the
lateral allophone. Apinaye (809), Japanese (071), !Xu (918) and perhaps
Bribri (801) are among other languages with lateral allophones of a flapped
r-sound. Korean (070), Dan (106) and Zande (130) are among languages with
flapped r-sounds as allophones of a lateral phoneme. It is likely that
other cases of fluctuation between lateral and nonlateral liquids are
concealed in some of the less detailed descriptions. These fluctuations
appear, not unexpectedly, to be more frequent in languages with only one
liquid.

The most frequent lateral reported as sole liquid is a voiced dental or

alveolar lateral approximant (28 of 32 cases). The only exceptions are the
alveolar lateral flaps of Luvale (125) and Zande (130), the retroflex
lateral flap of Papago (736) and the velar lateral approximant of Yagaria
(609). The only approximants reported as sole liquids are laterals;
r-sounds as sole liquids are almost always reported as flaps.

83
Liquids

Languages with 2 liquids (130, 41.0%)

The most typical language has 2 liquids, usually one lateral and one
r-sound. Table 5.13 shows how 2-liquid systems break down.

Table 5.13 Systems with 2 liquids

Number of Percent of Percent of

languages languages with languages in
2 liquids sample
1 lateral,
1 r-sound 108 83.1% 34.1%
2 laterals 18 13.8% 5.7%
2 r-sounds 3 2.3% 0.9%

The usual system consists of a trill or tap/flap and a lateral approximant,

these being the most common varieties of liquids. The systems with 2
laterals or 2 r-sounds do not contrast them by place of articulation but by
manner, voicing or secondary articulation or some combination of these. The
most frequent system of 2 laterals has a plain voiced lateral approximant
and a voiceless lateral (13 of 28 cases). S. Nambiquara (816) has plain and
laryngealized voiced retroflex lateral flaps, with non-lateral allophones
reported for /J/. Greenlandic (900) is also among the languages with
flapped r-sounds as allophones of / 1/ (Mase and Rischel 1971); it also has
a long voiceless lateral fricative.

Languages with 3 liquids (46, 14.5%)

The structure of systems with 3 liquids is outlined in Table 5.14.

Table 5.14 Systems with 3 liquids

Number of Percent of Percent of

languages languages with languages in
3 liquids sample
2 laterals,
1 r-sound 23 50.0% 7.3%
1 lateral
2 r-sounds 17 37.0% 5.4%
3 laterals 6 13.0% 1.9%

84
 Patterns of sounds

A system with 2 laterals is somewhat more common than one with 2

non-laterals. The 2-lateral systems are about evenly divided between those
with a contrast of place (dental/alveolar vs. palatal or retroflex) and
those with a contrast of voicing, manner or secondary articulation between
their laterals. No cases are reported in which the laterals contrast in
both place and another feature. Languages with 2 r-sounds are more likely
to contrast them in manner (12 out of 17) than on any other dimension (e.g.
trill vs. tap/flap). In 7 cases there is a difference in the primary place
of articulation (usually alveolar vs. retroflex) and in 5 of these this
contrast is in addition to a manner difference. The systems with 3 laterals
usually contain a contrast of voicing (5 out of 6) with one or two
voiceless fricatives or affricates (including ejectives). These systems are
only reported from American languages, predominantly from the Northwestern
coastal region of the North American continent (5 of 6 ) . Nootka (730) has 3
voiceless laterals /*, t 4 > t4'/« Note that the 3-liquid systems in the
survey do not include any in which all of the liquids are r-sounds.

Languages with 4 liquids (29, 9.1%)

The structure of systems with 4 liquids is shown in Table 5.15. A clear
majority of these systems (19 of 29) consist of an equal number of lateral
and nonlateral liquids.

Table 5.15 Systems with 4 liquids

Number of Percent of Percent of

languages languages with languages in
4 liquids sample
2 laterals,
2 r-sounds 19 65.5% 6.0%
3 laterals,
1 r-sound 6 20.7% 1.9%
1 lateral,
3 r-sounds 2 6.9% 0.6%
4 laterals 2 6.9% 0.6%

Of these, 7 languages have laterals that contrast in place of articulation

alone; the remainder contrast laterals by manner, voicing or secondary
articulations. Five languages contrast their 2 r-sounds by place of
articulation but in 4 of these cases there is also a contrast of manner. In
total, 12 of the 19 languages contrast their r-sounds by manner.

85
Liquids

Of those languages with 3 laterals, 2 contrast plain voiced lateral

approximants at 3 different places of articulation. The remaining 5
contrast their laterals by voicing and/or manner differences and include at
least 2 fricative or affricate laterals. The 2 languages with 3 r-sounds
are both unusual. E. Armenian (022) is reported with an alveolar trill, a
retroflex fricative and a uvular fricative beside a dental lateral, i.e.
/r, j, b, y . Malagasy (410) has /r, 1/ plus voiced and voiceless "trilled
retroflex affricates". The 4-lateral systems contrast their laterals by
manner and voicing, not by place. Note again, no systems occur with all
their liquids nonlateral.

Languages with 5 liquids (14, 4.4%)

The structure of systems with 5 liquids is shown in Table 5.16. The most
typical of these systems consists of 3 laterals contrasted by place and 2
r-sounds differing in manner. This kind of system is principally
represented by Australian languages (5 of 6 cases); Alawa (354) with
alveolar, palato-alveolar and retroflex voiced lateral approximants, an
alveolar trill and a retroflex approximant, i.e. /|, |, I, r, J/, is a
representative example.

Table 5.16 Systems with 5 liquids

Number of Percent of Percent of

languages languages with languages in
5 liquids sample
3 laterals,
2 r-sounds 9 64.3% 2.8%
2 laterals,
3 r-sounds 1 7.1% 0.3%
4 laterals,
1 r-sound 2 14.3% 0.6%
5 laterals 2 14.3% 0.6%

On the other hand, Ngizim (269), one of the languages with 3 laterals and 2
r-sounds, and all the languages with 4 or 5 laterals in a 5-liquid system
contrast their laterals by voicing and manner differences.

Systems with 6 or more liquids (11, 3.5%)

The structure of systems with 6 or more liquids is shown in Table 5.17.

86
 Patterns of sounds

Table 5.17 Systems with 6 or more liquids

Number of Percent of
languages languages in
sample
4 laterals,
2 r-sounds 3 0.9%
3 laterals,
3 r-sounds 2 0.6%
5 laterals,
1 r-sound 1 0.3%
6 laterals 2 0.6%
4 laterals,
3 r-sounds 2 0.6%
6 laterals
4 r-sounds 1 0.3%

Just as with the systems with 5 liquids, the laterals tend either to differ
by place (Diyari, 367, Aranda, 362) or by manner and voicing (Sedang, 304,
Chipewyan, 703, Haida, 700, Kwakw'ala, 731). There is however, one
language, Diegueflo (743), which includes intersecting contrasts of both
place and voicing/manner, having the 4 laterals / I , I, 4,47. The largest
number of laterals reported in the survey is 6. The 7-liquid languages are
2 Australian languages (Kariera-Ngarluma, 363, Arabana-Wanganura, 366) which
contrast laterals at 4 different places of articulation, trills at 2 places
and also have a retroflex approximant. Irish (001) is the language with 10
liquids, having voiceless counterparts to its voiced alveolar liquids that
have morphophonologically specialized functions. Because of its voiceless
r-sounds Irish is the only language in the sample with 4 r-sounds.

5.6 Generalizations ^ri the structure of liquid systems

The languages in UPSID show up to 4 contrasts between places of
articulation for laterals and up to 6 contrasts of manner and voicing for
laterals. They show up to 3 contrasts of place for r-sounds and up to 3
contrasts of manner. These are likely to be the maxima for these contrasts.
Although laterals are reported at 6 major places of articulation, no
language is known to contrast palato-alveolar and palatal laterals, and
velar laterals are so rare that for them to occur with laterals at 3 other
5
places of articulation would simply be improbable (Ladefoged, Cochran and
Disner 1977 report dental, alveolar and velar lateral approximants
co-occurring in Melpa). There has been relatively little work done on

87
Liquids

phonetic differences between laterals (but see Bladon 1979; Davey,

Maddieson and Moshi 1982), so it is not clear if the failure to exploit all
the places of articulation in one language could be attributed to a lack of
phonetic distinctiveness. As there are no languages reported with over 6
laterals or over 3 r-sounds (apart from Irish), these automatically set the
maxima for the other contrasts mentioned above.
The most commonly found systems containing one to 6 liquids are
reviewed in Table 5.18. These patterns suggest that an inventory of liquids
is generally expanded by adding more laterals before adding more r-sounds.

Table 5.18 Common structures of liquid inventories

No. of liquids Most common No./ total Percent

structure of cases
1 1 r-sound 42/74 56.8%
2 1 lateral,
1 r-sound 109/130 83.8%
3 2 laterals,
1 r-sound 23/46 50.0%
4 2 laterals,
2 r-sounds 19/29 65.5%
5 3 laterals,
2 r-sounds 9/14 64.3%
6 4 laterals,
2 r-sounds 3/8 37.5%

Several other generalizations concerning the structure of liquid systems

also suggest themselves and are presented below.
(i) A language with two or more liquids is most likely to have at
least one lateral. 227/230 98.7%.
(ii) A language with two or more liquids is most likely to include a
lateral/nonlateral contrast between them. 198/230 86.1%.
(iii) A language with one or more laterals usually has a voiced lateral
approximant. 255/258 98.8%.
(iv) A language with two or more laterals contrasts them either in
place or in manner and voicing but not both. 97/101 96.0%.
(v) A language with two or more r-sounds is unlikely to restrict their
contrast to place of articulation. 55/60 91.7%.
(vi) A liquid with both lateral and r-sound allophones is most likely
to be the only liquid in the language. 8/10.
 Patterns of sounds

(vii) A language most often has two liquids (usually one lateral and one
r-sound). 109/317 34.4%.
Although approximant laterals are the most common type of lateral, the
probability of (iii) being true is significantly higher than their overall
frequency as a percentage of all laterals (79.7%) would suggest. The two
observations (iv) and (v) draw attention to a quite marked difference
between laterals and r-sounds in the way that the systems are elaborated.
The data is incomplete on point (vi) so no percentage is expressed.
However, only two counterexamples are known from among the languages in the
survey.

5.7 Conclusion
The survey of liquids in UPSID has revealed patterns of occurrence of
different types of liquids which may be taken as reliable. These patterns
concern both the overall frequency of particular sound types and their
relation to the inventory in which they occur. Although such observations
have an intrinsic interest of their own, their main value is to suggest
avenues of investigation in diachronic phonology, articulatory phonetics or
speech perception designed to seek the explanation for these patterns.
Research on liquids in these fields seems to be a neglected area.

Notes
1. Provision to distinguish between taps and flaps was made in the
variables employed in UPSID, but the sources used do not seem to
distinguish them reliably. They have therefore been treated here as a
single group. For more discussion of the use of these terms see
Ladefoged (1971) and Elugbe (1978).
2. Ladefoged, Cochran and Disner (1977) claim that "very few languages
have any trills at all." The data collected for UPSID suggest either
that trills are not in fact particularly rare or that very many
erroneous reports of trills occur in the literature.
3. Including voiceless and voiced "trilled retroflex affricates" reported
in Malagasy (410) but not included in the totals given in Tables 5.7
and 5.8.
4. There may be some reporting bias reflected in this finding. A somewhat
retracted articulation of approximant /r/ is labelled "retroflex" in
some analyses of English (e.g. Kenyon 1926). This may have led to a
predisposition to label any approximant "r" as retroflex among
English-speaking linguists.
5. If the one instance of a velar lateral and the 4 instances of laterals
at 4 places of articulation in UPSID are taken as indications of the
frequency of such occurrences, then the probability that both would
occur in the same language might be estimated at less than .00004 (i.e.
in fewer than one language in 400,000).

89
Liquids

References
Bhat, D.N.S. 1974. The phonology of liquid consonants. Working Papers in
Language Universals (Stanford University) 16: 73-104.
Bladon, R.A.W. 1979. The production of laterals: some acoustic properties
and their physiological implications. In H. and P. Hollien (eds.),
Current Issues in the Phonetic Sciences (Amsterdam Studies in the Theory
and History of Linguistic Science, Series 4, Current Issues in
Linguistic Theory 9 ) . John Benjamins, Amsterdam: 501-8.
Davey, A., Maddieson, I. and Moshi, L. 1982. Liquids in Chaga. UCLA Working
Papers in Phonetics 54: 93-108.
Elugbe, B.O. 1978. On the wider application of the term "tap". Journal o£_
Phonetics 6: 133-9.
Goldstein, L. 1977. Perceptual salience of stressed syllables. UCLA Working
Papers ^Ln Phonetics 39: 37-60.
Goschel, J. 1972. Artikulation und Distribution der sogennanten Liquida r
in den europMischen Sprachen. Indogermanische Forschungen 76: 84-126.
Guitarte, G.L. 1971. Notas para la historia del yeismo. Sprache und
Geschichte: Festschrift fUr Harri Meier zum 65 Geburtstag. Fink, Munich:
179-98.
Kenyon, J.S. 1926. Some notes on American r. American Speech 1: 329-39.
Ladefoged, P. 1971. Preliminaries to Linguistic Phonetics. University of
Chicago Press, Chicago.
Ladefoged, P., Cochran, A. and Disner, S. 1977. Laterals and trills.
Journal of the International Phonetic Association 7: 46-54.
Lindau-Webb, M. 1980. The story of r. UCLA Working Papers in Phonetics 51:
114-19.
Mase, H. and Rischel, J. 1971. A study of consonant quantity in West
Greenlandic. Annual Report of the Institute of Phonetics, University of
Copenhagen 5: 175-247.

90
Vocoid approximants

6.1 Introduction
Approximants are consonantal sounds produced with a relatively unimpeded
flow of air through the mouth. The constriction is not narrow enough to
produce local turbulence, though cavity friction may be heard if the
segment is voiceless (Catford 1977). Apart from those approximants which
have a lateral escape or belong to the family of r-sounds, the only
frequently-occurring approximants in the world's languages are those which
have vocoid characteristics (Pike 1943). They are often known as
"semi-vowels". This chapter examines the frequency of such sounds as
phonemic units in the UPSID sample and discusses certain co-occurrence
restrictions which relate to their role in phoneme inventories.
In the UPSID file, vocoid approximants have been coded as consonants if
they don't alternate with syllabic vocoid pronunciations and share
distributional properties with other consonants. Over 90% of the surveyed
languages have one or more such segments.

6.2 Frequency of vocoid approximants

The great majority of languages, 86.1%, have a voiced palatal approximant
/ j / or a closely similar segment, such as / if in Khasi (302). Substantially
fewer languages, 75.7%, have a voiced labial-velar approximant /w/ or a
closely similar segment. The frequency of these segments and their
co-occurrence is shown in Table 6.1. The occurrence of /w/ is associated
with the occurrence of / j / . From the independently calculated frequencies
of /w/ and / j / , they would be expected to occur together in only 65.2% of
the languages, not the 71.3% shown in Table 6.1. If these two segments were
independently distributed, then there would be 31 or 32 languages in the
survey with /w/ but no / j / , rather than the 14 actually found.

91
Vocoid approximants

Table 6.1 Distribution of / j / and /w/

in UPSID languages

With /j7 No /j/

With /w/ 226 (71.3%) 14 (4.4%)
No /w/ 47 (14.8%) 30 (9.5%)

In the UPSID data, therefore, the occurrence of /w/ usually implies the
occurrence of /J/ in the same language. However, the association of /w/ and
/ j / is not as strong in UPSID as that found by Stephens and Justeson (1979)
in the materials collected for the Stanford Phonology Archive. In these
materials /w/ occurs without / j / in only 1% of the languages surveyed, as
against over 4% in the UPSID languages. Stephens and Justeson also report
substantially lower overall frequencies of both /w/ and /j /, with the
percentage of languages having these segments about 15% lower in each case.
It is not clear if these differences arise from the different selection of
languages in the two surveys or from the application of different criteria
for phonemic status of approximants. Nonetheless their claim that there is
a statistically significant tendency for /w / to occur only if / j / also
occurs is confirmed by our data (significance from ^ 2 better than .001).
Other vocoid approximants are comparatively rare. They may be divided
into two groups - those which are modified variants of / j / and /y^/ and
those which have different places of articulation. Those in the second
group include the labial-palatal approximant /q / (4 instances) and the
velar approximant /y / (5 instances). These occur in less than 2% of the
languages surveyed.They are not found in modified form in any of the UPSID
languages. Palatal approximants occur voiceless, laryngealized and
nasalized. Labial-velar approximants occur voiceless and laryngealized. The
frequency of modified segments of these types is given in Table 6.2.
Laryngealized approximants / j / and /w/ occur with approximately equal
frequency and are restricted to languages which have other glottalic or
glottalized segments in their inventories and have plain voiced / j / or /w/.
Greenberg (1970) suggested that / j / fills the place of an anticipated
palatal implosive in languages with an implosive series and a palatal place
of articulation. This issue is discussed in some detail in Chapter 7 where
it is concluded that there does not seem to be support for it in the
available data. A diachronic source of this kind for /]/ would predict that
it would be more frequent than /w/, for which no parallel source is

92
 Patterns of sounds

proposed. Only 5 of the 13 instances of / j / in the survey occur in the kind

of inventory that would appear to support Greenberg's suggestion. There is
a much stronger association between the occurrence of / j / and of /w/; in 12
cases /w/ and / j / occur together. In other words, there is only one
exception to the statement that the presence of /w/ implies the presence of
/j/.

Table 6.2 Frequency of modified / j / and /w/

Segment No. Percent of languages

7 2.2%
O.C_.

/j/ 13 4.1%
IV 3 0.9%
/AA/ 11 3.5%
/w/ 12 3.8%

The voiceless approximants / j / and /AA/ differ fairly markedly in

frequency, /AA/ being 1.7 times more frequent than / j / . This is particularly
surprising when considered in comparison to the relative frequency of their
voiced counterparts. The diachronic source of these voiceless segments is
likely to be similar in both cases - documented instances seem to arise
predominantly from a cluster of a voiceless obstruent and the voiced
approximant, or from labialized or palatalized voiceless obstruents (which
may be equivalent to a source in a cluster). Thus (one source of) //v\/ in
Hupa (705) is from Proto-Athabaskan */J w / (Huld 1980), Middle English /AA/
w
is derived from Old English /x /, and *//v\/ in early Northern Tai is derived
from Proto-Tai */x w / (Li 1977). They are likely to exit from inventories in
a variety of ways, including vocalization (as in the widespread merger of
*/M/ with /w/ in many varieties of Modern English), collapse into an
undifferentiated voiceless vowel phoneme /h/ (as in the special development
of */M/ in English before /u/ and /o/ in words such as "who", "whoop" 1 ,
"whole"), or fricativization (as in the idiolectal /?/ for the initial
segments /hj/ = [jj] in English words such as "huge", "human" etc.; compare
the change of earlier German /w/ into / v / ) .
The relative frequency of / j / and /AA/ suggests that there may be some
factor which favors the development of /AA/ over / j / or favors the loss of
/ j / more than /AA/. It is probably the case that a true voiceless palatal
approximant is poorly distinguishable from /h/, which occurs in most
languages, and hence is likely to collapse together with it. If, on the

93
Vocoid approximants

other hand, / j / is articulated more forcefully to preserve the distinction

o

it would become a palatal fricative. A voiceless labial-velar approximant

may survive better because its two strictures produce two cavities with
resonances which are rather close to each other in frequency and hence
reinforce each other (cf. Ohala and Lorentz 1977).

6.3 Approximants and related vowels

The approximants / j / and /w/ are closely related to the high vowels /}/ and
/u/ respectively. The vast majority of languages have both these vowels,
but there are more cases in which /u/ is missing than /]/ - in fact, /u/ is
the most frequently missing of the major peripheral vowels (see chapters 8
and 9 ) . The greater frequency of /]/ is undoubtedly a predictor of the
greater frequency of / j / . However, for both / j / and /w/ there are a few
languages which have the approximant but lack the corresponding vowel. The
numbers are given in Table 6.3.

Table 6. 3 Common approximants occurring without

cognate vowels

No. of languages % of sample

/j/ but no IM 8 2.5%
/w/ but no lul 23 7.3%

There are about 3 times as many cases of /w/ occurring without /u/ as of
/j/ occurring without /i/. Disner in Chapter 9 suggests that the systems
without /u/ may be regarded as falling into two principal classes: those
with a "compensating" vowel which is high or back or rounded but not all
three (such as /t/, /w/, /«/ etc) and those which simply have a gap (and
whose highest back vowel is usually / o / ) . This is suggestive of a variety
of possible sources for /w/ and may predict that the class of /w/ segments
in languages may vary phonetically through a greater range than / j / .
The less frequently occurring approximants /q / and /y / were also
investigated in relation to the corresponding vowels, in this case /y / and
/w/ respectively. The numbers are given in Table 6.4. These suggest that
/ M/ is most likely to occur if /y/ also occurs in the inventory, but that
there is no such dependence of /y/ on the occurrence of An/. However since
these numbers are so small no great reliance should be placed on these
indications.

94
 Patterns of sounds

Table 6.4 Other approximants and vowels

No. of languages

/M/ and hi 3
hi but no /y/ 1
hi and /LU/ 1
hi but no / UJ / 4

Dximants and related consonants

/ j / and palatalized consonants

True palatalized consonants, that is, ones with a palatal secondary

articulation usually perceptible because of a /j/-like offglide, occur in

about 10% of the languages in the survey. Since desyllabification of high

vowels is a major process creating both / j / and palatalized consonants, it

might be expected that palatalized consonants would occur only in languages

with / j / (cf. Bhat 1978). There are, however, 3 languages in the survey

which have palatalized consonants but no /j/-phoneme. This is exactly the

number that would be predicted if there was no association between these 2

classes of sounds. Of these 3, Ocaina (805) seems straightforward, but

Muinane (806) has a voiced palatal fricative / j / with [j] as an allophone,

and Ket (906) also has / j / , albeit largely restricted to intervocalic

positions. Thus a generalization stating that palatalized consonants occur

in inventories containing / j / or / j / would have only one exception.

/w/, labial-velar stops and labialized velars

As /w/ has two strictures of equal rank it falls into a class with other

labial-velar consonants, especially /£p/ and /gB/ which are the most common

labial-velar consonants after /w/. These labial-velar stops may vary a good

deal in their initiation (Ladefoged 1968) but belong together by virtue of

their shared place of articulation. The co-occurrences between /w/ and

/£p, gB/ are shown in Table 6.5 below:

Table 6.5 Co-occurrence of /w/ and labial-velar stops

No. of languages A, of sample

/ w / , /kp/ and /gB/ 19 5.9%
/w/ and /f<p/ 1 0.3%

/w/ and /gB/ 2 0.6%

no /w/ but /l<p/ and /gB/ 1 0.3%

95
Vocoid approximants

An assumption that there is no relation betweeen the occurrence of /w/ and

/($/ or /gB/ predicts that we should expect 16 cases of /w/ with /f<p/ (20
actual) and 16 or 17 cases of /w/ with /gb/ (21 actual). The observed
numbers suggest that there is a tendency for /l<p/, /gB/ to occur in systems
with /w/ in preference to those lacking /w/. The one exception, Kpelle
(103), is also unusual in another way that is described below.
There is an obvious similarity between labialized consonants and /w/,
and there is a historically similar source for both types of sounds in
desyllabification of /u/ in many instances. We might therefore expect
labialized consonants to occur preferentially in languages which have /w/.
By far the most frequent labialized consonant types are labialized velar
stops (cf. Ohala and Lorentz 1977). We will therefore use them as
archetypes of labialized consonants. The co-occurrences of /w/ with /k w /
are shown in Table 6.6 (/gw/ only occurs if /k w / occurs, so is not
separately listed).

Table 6.6 /w/ and labialized velar stops

No. of languages % of sample

/w/ and /kw/ 35 11.0%
no /w/ but /k"/ 5 1.6%

Random co-occurrence of /w/ and /k w / would predict that there would be 30

languages in the sample which contained both of these segments. The
observed number (35) suggests that there is a weak tendency for /k w / to be
more likely to occur in languages which have /w/. The exceptions to the
trend in UPSID are Mixtec (728), Guarani (828), Wantoat (615), Chipewyan
(703), and Kpelle (103). In Chipewyan the labialized velars have a rather
marginally contrastive status, since they are largely restricted to
occurrence before back rounded vowels where plain velars do not occur.
Kpelle is unusual in being the only language in the survey which has both
labial-velar and labialized velar stops; and it also lacks /w/!

6.5 Other approximants

In addition to the 4 most common approximants / j , w, q, y/ discussed above,
it may be noted that 6 languages (1.9%) have a bilabial approximant /p/ and
6 have a labio-dental approximant / W . The remaining approximants in UPSID
are classified as liquids and are discussed in Chapter 5.

96
 Patterns of sounds

6.6 Summary
Most languages have / j'/ and /w/, with / j / being more frequent. There is a
strong tendency for the presence of /w/ to imply the presence of / j / in the
same language. The greater frequency of / j / is parallel to the greater
frequency of / i/ than /u/, but these facts are not directly related since
/ j7 may occur without / i/ and /w/ without / u/. Modified varieties of / j/
and / w/ only occur in languages with the plain voiced counterparts. There
is some association between the occurrence of palatalized consonants and
/ j7 and between labial-velar stops and labialized velars (and other
labialized consonants) and /w/.

Notes
1. This word now frequently receives a spelling pronunciation, with
initial / w/ or even /M/ supplanting the historically derived /h/.

References
Bhat, D.N.S. 1978. A general study of palatalization. In J.H. Greenberg et
al. (eds.), Universals of Human Language, Vol. 2, Phonology. Stanford
University Press, Stanford: 47-92.
Catford, J.C. 1977. Fundamental Issues in Phonetics. Indiana University
Press, Bloomington.
Greenberg, J.H. 1970. Some generalizations concerning glottalic consonants,
especially implosives. International Journal of American Linguistics 36:
123-45.
Huld, M. 1980. Tone in Proto-Athabaskan. Unpublished paper. Dept. of
Classics, UCLA.
Ladefoged, P. 1968. A Phonetic Study of West African Languages. Cambridge
University Press, London.
Li, F-K. 1977. A Handbook of Comparative Tai. University of Hawaii Press,
Honolulu.
Ohala, J.J. and Lorentz, J. 1977. The story of [w]: an exercise in the
phonetic explanation for sound patterns. Proceedings of the Third Annual
Meeting, Berkeley Linguistic Society: 577-99.
Pike, K.L. 1943. Phonetics. University of Michigan Press, Ann Arbor.
Stephens, L.D. and Justeson, J.S. 1979. Some generalizations concerning
glides. In D.I. Malsch et al. (eds.), Proceedings of the Eighth Annual
Meeting of the Western Conference on Linguistics. Linguistic Research
Inc, Carbondale and Edmonton: 1 M - 6 4 .

97
Glottalic and laryngealized consonants

7•1 Introduction
This chapter presents the results of a survey of the occurrence of
glottalic consonants and other "glottalized" consonant segments in our
sample of the world's languages, and relates their occurrence to the rest
of the segments in the phonological inventories of the languages. It draws
substantially on a previous survey of the same data by Fordyce (1980).
In an important earlier study, Greenberg (1970) discussed the
distribution of glottalic consonants cross-linguistically and
language-internally. Although there has been considerable subsequent work
addressing Greenberg's claims, much of it either suggests specific
counter-examples (e.g. Campbell 1973; Pinkerton 1980) or brings little
additional data to bear on the general validity of his conclusions (e.g.
Hamp 1970; Javkin 1977). Greenberg's main claims are summarized in the
sentence "injectives [i.e. implosives] tend to have front articulation,
ejectives to have back articulation". Greenberg acknowledges that these
conclusions were partly anticipated by Haudricourt (1950) and independently
discovered by Wang (1968).
Our goal is to determine whether these place of articulation preference
hierarchies for implosives and ejectives can be substantiated and to
discover other distributional patterns relating to glottalic and
glottalized segments in the UPSID data. The possible phonetic motivations
for the patterns found will also be discussed. Greenberg's generalizations
were based on a survey of languages for which data was available rather
than on a carefully structured sample and it is possible that his findings
reflect accidental biases in the selection of languages examined. In using
the UPSID sample we hope to avoid the likelihood of such a bias.

98
 Patterns of sounds

7.2 Glottalic and laryngealized sounds

It may be useful to offer some definitions of the classes of sounds which
are the subject of this chapter. They are any segments produced with the
glottalic airstream mechanism (i.e. ejectives and implosives) as well as
"glottalized" segments where the glottal constriction does not serve as the
airstream initiator (i.e. preglottalized and laryngealized consonants).
Only those sounds articulated using the glottalic airstream mechanism will
be referred to as glottalic. Pulmonic or velaric "glottalized" sounds will
generally be referred to as laryngealized. Ejectives are those sounds
produced by raising the larynx with the glottis closed; with a constriction
in the oral cavity, air is compressed in the space enclosed between the
oral constriction and the glottal closure. The oral occlusion or
constriction is subsequently released with outward airflow. Implosives, on
the other hand, are articulated by lowering of the larynx. In an idealized
case, air enclosed between an oral occlusion and the laryngeal constriction
is rarefied, and air flows in through the mouth when the oral closure is
released. However, the glottis is usually not closed, but rather the vocal
folds are allowed to vibrate through leakage of pulmonic air into the oral
cavity (see Catford 1939). Thus ejectives tend to be unvoiced while
implosives tend to be voiced.

Moreover, as noted by Ladefoged (1968), Pinkerton (1980) and others,

"implosives" do not always entail inward oral air flow upon release even
when the larynx lowering gesture is present. The distinction between truly
imploded consonants and those which are preglottalized or laryngealized
with minimal or zero implosion has been difficult to maintain in UPSID.
These sounds are frequently not distinguished in the literature, but where
the airstream initiator of the most typical allophone is known to be or may
be other than glottalic, the sound is classified in UPSID as laryngealized.
As both Ladefoged (1968) and Greenberg (1970) conclude, the potential
phonological contrast of these differing types is not realized in any of
the languages known to them directly or through the literature (nor in any
included in UPSID).
It should be noted that the variables available for coding glottalic
and laryngealized segments in UPSID limit the description of segments in
certain phonetically plausible ways. Thus, ejectives must be specified as
either ejective stop (glottalic egressive stops), ejective affricate
(glottalic egressive affricates), or ejective fricative (glottalic
egressive fricatives). The presumedly phonetically impossible ejective

99
Glottalic and laryngealized consonants

approximant, for instance, is therefore excluded a priori by the coding

mechanism. Likewise, implosives may only be coded as implosive stops. The
phonetically implausible implosive affricate, fricative or approximant is
therefore excluded. 1 On the other hand, the UPSID variable
"laryngealized" is not part of the set of mutually exclusive manners of
articulation. Instead it reflects the possible and actual occurrences of
preglottalized or laryngealized stops, affricates, fricatives,
approximants, and vowels. Thus as Greenberg (1970:2) states, "the
phonological opposition in individual languages between ejectives and
injectives applies effectively only to obstruents"; other types of segments
may be laryngealized but do not exhibit an ejective/implosive contrast.
Ejectives are the most widespread of the various types of segments
being considered in this chapter, both from the genetic and geographical
points of view. Although ejective systems are found in several of the
world's major language families, by far the greatest number of languages
with ejectives appear to be from the Americas. Two thirds (35 of 52) of the
languages with ejectives in UPSID are in the Amerindian family - and most
(30) of these are from North America. Almost 60% of the 51 Northern
Amerindian languages in UPSID contain ejective systems. Only 4 of these
languages also exhibit implosives or voiced laryngealized plosives. The
particularly frequent occurrence of ejectives in this group of languages
has been commented on before (e.g. Sherzer 1973), but the comprehensive
nature of the UPSID sample allows this fact to be placed in its proper
relief. These languages have an important property which sets them apart
from the remainder of the language families of the world. Nine of the
remaining languages with ejectives are in the Afro-Asiatic family, 3 are
Nilo-Saharan, 3 are Caucasian, and there is one in each of the
Indo-European, Niger-Kordofanian and Khoisan families. Ejectives are not
known to occur outside of these major language families, although it is
quite likely that they may occur in the Austro-Asiatic and/or Austro-Tai
families, since voiced "glottalized" stops (implosive, preglottalized or
laryngealized) are found there (e.g. in Vietnamese, 303; Sedang, 304; and
Sui, 403).
As Greenberg (1970:2) notes, "the typical ejective obstruent is
unvoiced". We find no exceptions to this in the languages surveyed,
although phoneticians of the caliber of Catford and Pike have suggested
that voiced ejectives are possible speech sounds. However, pre-voiced
ejectives are reported in one language in the survey, !Xu (918). This

100
 Patterns of sounds

language has the most complex consonantal system of.any of the languages in
UPSID, including six series of stops apart from 48 click consonants. One of
the stop series is described as consisting of voiced ejectives by Snyman
(1969), but his description sounds as if there is necessarily a phonetic
sequence of voicing preceding ejection in these segments. He says:

what actually happens is that the vocal cords are activated by

pulmonary air and they produce a voiced unemitted sound which we
represent [i]. (Both the nasal and oral passages are closed.) The
unemitted sound [d] is swiftly followed by the articulation of the
ejected sound ... In close sequence [d] and ... [the ejected sound]
is perceived as a vocalized sound.

Thus, as far as is known, voiced ejectives, i.e. ejectives in which voicing

continues through the stricture for the consonant, do not in fact occur in
human languages. In "normal" ejectives, the available evidence shows that
voicing begins very promptly on release of the glottal closure. Lindau
(1982) shows this for ejective stops in Hausa and Navaho.
Implosives and their close relatives, voiced laryngealized plosives,
are found in fewer of the major language families and are more limited
geographically than ejectives. Most such systems (29 of 41) are to be found
on the African continent. The distribution of implosives thus also has a
strong areal concentration, but it is one which cuts across 3 major
language families, Niger-Kordofanian (10 languages), Nilo-Saharan (10
languages), and Afro-Asiatic (9 languages). The largest number among the
remaining languages are Amerindian languages (6), but there are a few
languages from Austro-Asiatic (3), Austro-Tai (3), and Sino-Tibetan (1).
Laryngealized sonorants (nasals, liquids, and central approximants) are
found in most language families mentioned above but not elsewhere. They too
are most common in Amerindian languages.

7.3 Ejectives
Of the 317 languages in UPSID, 52 contain ejectives, making ejectives the
most common of the glottalic or laryngealized segments. Twelve of these
languages also exhibit implosive stops and 15 also exhibit some
laryngealized stops, fricatives, sonorants, and/or vowels.
The most frequently occurring type of ejective is an ejective stop;
there are twice as many ejective stops as ejective affricates in the data
file - 188 to 94. Ejective fricatives are considerably rarer; only 20 are
recorded in UPSID. Naturally, ejective stops have been the topic of more
discussion than the other types of ejectives.

101
Glottalic and laryngealized consonants

Ejective stops
Haudricourt (1950) suggested that ejective stops exhibit a strong
preference for back articulations, Greenberg (1970) supported this claim
mainly because he found that "a gap in the class of ejectives at the
bilabial point of articulation is found in a number of world areas". Javkin
(1977) made the implied hierarchy for ejective stops more explicit,
formulating it as follows: "...[a language] will only have labial ejectives
if it has alveolar and velar; it will only have alveolar if it has velar."
(In this quote, "alveolar" refers to either dental or alveolar; as in other
chapters we will use an asterisk before a phonetic symbol to designate the
class of dentals and alveolars jointly.) Javkin counted the ejective stops
in the Stanford Phonology Archive and saw the numbers as being generally
confirmatory. The count is reproduced below as Table 7.1.

Table 7.1 Ejective Stops in the Stanford Phonology Archive

Labial Dental/alveolar Palatal Velar Uvular

26 29 7 31 15

There are relatively small differences in the numbers of ejectives reported

at the 3 major places of articulation, but Javkin notes that the
implicational hierarchy he has set up holds within the languages concerned.
Javkin further notes that palatals and uvulars do not maintain the tendency
to prefer a further back articulation over a further front one "since these
places of articulation tend to disfavor stops". Indeed, as Javkin suggests,
the relative disfavoring of palatals and uvulars is not restricted to
glottalic consonants (see chapter 2 in this volume for details concerning
non-glottalic stop distributions and Gamkrelidze 1978).
The frequency count of ejective stops in UPSID is given in Table 7.2.
Plain velar ejective stops are no more frequent than plain dental or
alveolar ones, there being 49 instances of each. However, because there are
3 languages (Nez Perce, 706, Porno, 742, Wappo, 760) with both a dental and
an alveolar ejective stop and one language (Kwakw'ala, 731) with no plain
velar but with a labialized velar, there are four more languages with an
ejective stop at the velar place than languages with one at the dental and
alveolar places considered together (50 vs. 46). One language, Hupa (705),
has /*t 5 / but no /k'/ (it does have / q ' / ) . A tendency to prefer velar place
for ejective stops can be seen in the fact that presence of /*t 5 / implies
the presence of /k'/; it is significant that the 5 languages which have
only one place of articulation for their ejective stops all have velars.

102
 Patterns of sounds

Table 7.2 Ejective stops in UPSID

Labial Dent/alv Palatal Velar Uvular

Plain 33 49 7 49 19
Labialized - - - 18 8
Palatalized - - * 2 -
Prevoiced 1 1 _ 1 _

Total 34 50 7 70 27

What is far more salient is that both velar and dental/alveolar places
are preferred to bilabial. There are significantly fewer occurrences of
/p 5 / than of either /k 3 / or /*t 5 /. No language has /p'/ that does not have
a velar, whereas 17 languages have /k'/ but no /p'/. Of the 11 languages
with only two places of articulation for their ejective stops, 10 have
/*t 3 , kV (the exceptional case is Berta, 218, with /p', k 5 / ) . Thus the
UPSID sample shows principally that bilabial place is disfavored for
ejective stops. This is reminiscent of the findings with respect to
voiceless plosives, which are also disfavored at the bilabial place,
although there is some evidence that the tendency to avoid /p 5 / is stronger
than the tendency to avoid /p/. This raises the question of whether both of
these patterns should be explained in the same way. This question will be
taken up again below.

Uvular ejectives are relatively common. We may show this by comparing

the ratio of uvular ejective stops to velar ejective stops, .39, with the
ratio of plain voiceless uvular plosives to voiceless velar plosives, which
is only .13. Their frequency is largely the result of the coincidence of
two areal tendencies, use of the uvular place of articulation, and presence
of ejectives, in North American languages. Of course, uvular ejectives are
also consistent with a preference for back places for ejectives. The
frequency of palatal ejective stops, on the other hand, is not
disproportionate but is comparable to the proportion of plosives at the
palatal place.
Only one type of secondary articulation is at all frequent with
ejective stops, and that is labialization with velar or uvular ejectives.
More will be said about secondary articulation below.
The detailed structure of ejective stop systems in the languages
surveyed is shown in Table 7.3. Dental and alveolar stops have been pooled
together, except in those languages where both occur. The presence of

103
Glottalic and laryngealized consonants

labialized ejective stops is indicated in a parenthetical comment unless

they occur at a place where there is no corresponding plain ejective stop.

Table 7.3 Number of places of articulation for ejective stops

Ejective stop Number of languages

inventory
1 ejective stop k5 5 (2 also with k w? ,
1 also has kJ')
2 ejective stops *t> k' 10 (3 also have kw')
p' k5 1
3 ejective stops p? *t' k5 15 (4 also have kw')
t' k' q' 1
t' c> q> 1
4 ejective stops > *t> k» 10 (7 also have k w \ qw>)
CL CL

q'
5
*t' c' k5 2
*t ' c> k' q' 1 (also has k w> , kw')
5 5
P t t' k9 1
5 ejective stops P' t' c' k> q' 2
P' t' c5 kw 5 q5 1 (has q w ' but not k 3 )
P' t5 t5 k' q' 2

Table 7.3 shows that the most commonly encountered set of ejective stops
contains one at each of the most common places at which stops of any kind
occur (bilabial, dental or alveolar, and velar). Frequently, these are the
same places at which the language has other types of stops (glottal place
is not considered in these cases). An example of this pattern is the stop
system of Eastern Armenian (022):
p t k
ph fh [<h

p' t' k'

or Tzeltal (712), whose stop system is as follows:
P t
kh
b d g
p' t' k'
The only other reasonably common ejective stop systems are those containing
either 2 ejective stops - dental or alveolar and velar - or 4 ejective
stops - labial, dental or alveolar, velar, and uvular. Typical of one type
of ejective stop system with only 2 places of articulation is Itonama (800)
whose stop inventory is as follows:

104
 Patterns of sounds

b d
t' k5
Here there is no bilabial ejective even though the language has other types
of stops at the bilabial place. Such languages are the crucial ones in
establishing that there is a specific tendency to avoid bilabial ejective
stops (cf. Haida, 700, below). The other main type of system is one in
which there are other deficiencies among bilabial segments - several of the
Semitic and Athabaskan languages lack a bilabial stop in another series as
well as a bilabial ejective.
If there are 4 or 5 ejective stops, there is usually one at the uvular
place: this is so for 16 of the 19 languages concerned. Again, there are
usually stops in other series at the same places of articulation. Quileute
(732) is representative of stop systems with 4 ejective stops:
p t k q
kw qw

b d (g)
p' t' k' q'
kw> qw>
All the languages with ejective stops at 5 places of articulation
include those stops found in the common 4-place inventory. An example is
Jaqaru (820):
p t c k q
ph fh ch ^h qh

p' t' c' k' q'

Thus an inventory of ejective stops is usually built up this way: if there
is one, it is velar; a second ejective is dental or alveolar; a third is
bilabial; a fourth is uvular. A small minority of languages deviate from
this pattern but the great majority conform (45 out of 52).
A few of the languages which are exceptions to this pattern suggest
that it may sometimes be appropriate to recognize a single series of
"glottalic" stops whose members may be phonetically diverse. An example is
K'ekchi (714), which has three ejective stops at alveolar, velar and uvular
places. It lacks a bilabial ejective, but has a laryngealized voiced
bilabial stop. There are no other members of a voiced laryngealized series.
Hence there is one "glottalic" consonant corresponding in place to each of
the plain voiceless plosives. The K'ekchi stop inventory is:

105
Glottalic and laryngealized consonants

p t k q
t' k' q'
6
The comparative background to this situation in K'ekchi has been
investigated by Pinkerton (1980), who examined the phonetic nature of
corresponding segments in K'ekchi and 4 other languages in the Quichean
group of Mayan languages. Among these 5 languages, ejectives, voiced and
voiceless implosives, and voiced laryngealized stops interchange (and, for
the bilabials only, even plain voiced stops are involved). The
correspondences are shown in Table 7.4.

Table 7.4 Correspondences in 5 Quichean languages

K'ekchi b/b t' k' q' (varies with q< )

Pocomchi P< t< k> q*
Cakchiquel b/b t' k' q<
Quiche 6 t' k' q<
Tzutujil 6 <f k' q<

Another exceptional language is Berta (218), a language with 2 ejective

stops, one bilabial and the other velar:
b "d" g
p' k'
11
cf
The general pattern leads us to expect a dental/alveolar ejective. This
dental/alveolar slot is filled by a glottalic consonant, but it is a voiced
implosive, and this implosive stands alone in Berta. Thus the "deficient"
ejective series and the isolated alveolar implosive between them create a
full series of glottalic stops, with a counterpart to the plain voiced
stops of the language at each place. Among other languages which could be
analyzed as having a single series of "glottalic" stops even though they
are phonetically heterogeneous are Ik (208) and Hausa (266). In the case of
Hausa, Carnochan (1951) has pointed to phonotactic constraints that apply
to all consonants with a "glottalic" component (including / ? / ) .
A second class of exceptions is illustrated by Hupa (705), which seems
at first glance particularly deviant in having 3 ejective stops but neither
a velar nor a bilabial. Hupa is the only language with /c 5 / and /q5/ that
does not also have / p V and /*t'/ in its inventory.

106
 Patterns of sounds

Hupa stop system:

t c q
th ch qh
t' c' q»
In this language the ejective stop system shares the same unusual places of
articulation as the voiceless plain and the voiceless aspirated stop
series. If Hupa had the "normal" bilabial, alveolar and velar ejectives, it
would be an exception to the general rule that languages have stops of
different series at the same places of articulation (see Chapter 2 ) .
Specifically, this means that ejectives should be expected to occur only at
places where non-glottalic stops occur. Thus, although there is a hierarchy
of preferences for place of articulation for ejectives, it is outranked by
the rule governing the relationship between places for plain stops and
ejectives. In other words, as Fordyce (1980) stressed, phonological
hierarchies may themselves be hierarchically arranged. Because Hupa
conforms to the more important rule, it comes to violate the place
hierarchy for ejectives considered by itself.
The importance of the connection between these rules can also be
appreciated from an examination of Haida (700). For a language with 4
ejective stops, Haida has an unusual set - dental, palatal, velar, and
uvular. It does not have an ejective at the bilabial place, although it
does have plain voiceless and aspirated stops which are bilabial. However,
there is no sense in which the palatal ejective stop has supplanted the
more usual bilabial one. The palatal is not unexpected because Haida is
conforming to the rule that ejective stops occur where there are also plain
stops. Palatal ejectives are not common largely because palatal stops in
general are not common. As noted above, this language constitutes part of
the evidence for the conclusion that bilabial ejective stops are disfavored
over dental/alveolar or velar ones.
Haida stop inventory:
p "t" c k q
w
k qw
ph "-(-h" c h |<h qh
^wh qwh

"f" c' k' q'

w
k ' qw'
The Wappo (760) system of 4 ejective stops, although it lacks a uvular
one, is not an exception to the ejective place of articulation hierarchy.

107
Glottalic and laryngealized consonants

It contains an unusual contrast between dental and alveolar ejective stops,

but the same place contrast is also found among the plain voiceless stops.
Here again the occurrence of an unusual ejective series can be attributed
to the precedence established by the plain stop series in the language.
Two of the languages with 5 ejective stops, Nez Perce (706) and Pomo
(742), also have both dental and alveolar stops, but there is no violation
of the place of articulation preference hierarchy for ejective stops, which
does not specify what a fifth member should be. The ejectives match the
places of articulation found for the voiceless plosives in both of these
languages, with the exception of Nez Perce /q 5 / which appears despite the
absence of /q/. The uvular place of articulation in Nez Perce is, however,
represented by the affricate /qx/» whereas there are no palatal or
palato-alveolar obstruents of any kind. Given the rest of the Nez Perce
system, an ejective at the uvular place of articulation is quite natural -
certainly more so than a palatal one, for example.
On the other hand, the 4-ejective stop systems of Kefa (264) and Maidu
(708) contain palatal ejectives rather than the more common uvular
ejectives. However, these languages lack any stops in the uvular position:
Kefa stop inventory:
p t c k
b d j g
p' t5 c' k>
Maidu stop inventory:
ph i-h ch [<h

p' t5 c> k5
G cf
The Maidu case is interesting since the plosives whose places of
articulation are matched are aspirated ones rather than the voiceless
unaspirated ones found in the other languages discussed here. Thus we may
state more generally that an ejective usually occurs only if a plosive
occurs at the same place. This rule is not limited to a particular type of
plosives.

Ejective affricates and fricatives

Ejective affricates and fricatives have more limited occurrence than
ejective stops. Forty languages in UPSID (12.6%) contain ejective
affricates. In all but one of these languages, ejective stops occur as
well. The exception is Iraqw (260) which does, however, have implosives.

108
 Patterns of sounds

Thus ejective affricates occur only in those systems containing glottalic

stops (almost exclusively ejectives). The most commonly occurring ejective
affricates are the sibilants /*ts'/ and /tj'/ and the lateral affricate
/*t4'/. The figures are given in Table 7.5.

Table 7.5 Ejective affricates in UPSID

te' *t4' *ts' tj5 ts' eg' kx' k4'

Simple 1 13 34 35 3 1 1 1
Labialized 0 0 1 1 0 0 1 0
Prevoiced 0 0 1 1 0 0 0 0

Totals I 13 36 37 3 1 2 1

Every language with any ejective affricates has at least one of the common
sibilant types, /*ts'/ and /tj'/. Seven languages have only /tj'/, 5 have
only /*ts 5 /, and 11 have both /*ts'/ and /tj'/ but no other ejective
affricates. The remainder contain one or both of these sibilant ejective
affricates plus one other, most frequently the lateral /*t4'/. Concerning
/tj'/, Greenberg (1970: 17) commented that "for the palatal region in
particular, it appears that the optimal ejective is the alveopalatal
[affricate] rather than a stop". He goes on to observe that among
non-glottalic obstruents, affricates are preferred over plosives in this
articulatory region, and adds that the preference for affricates is even
stronger in the case of ejective obstruents. In fact, Greenberg found "no
example of an ejective palatal stop" in his sample. However, they not only
do occur, but occur with a contrasting affricate. Of the 7 languages in
UPSID with a palatal ejective stop, 5 also have a palato-alveolar or
palatal ejective affricate. Nonetheless, Greenberg's observation may have
some validity. While there are 5 times as many languages with /tj'/ than
with /c'/, among the plain pulmonic obstruents there are only 3.4 times as
many occurrences of /tj/ as /c/.

With regard to ejective fricatives, Greenberg noted that they are

"relatively infrequent and always imply the presence of some ejectives with
abrupt onset". In the UPSID languages, ejective fricatives imply ejective
stops without exception. Most commonly they imply ejective affricates as
well, but exceptions do occur. Ten UPSID languages contain ejective
fricatives (3.2%); only 3 of these do not contain ejective affricates. The
ejective fricatives reported in the survey are given in Table 7.6.

109
Glottalic and laryngealized consonants

Table 7.6 Ejective fricatives in UPSID

f *s 5 J' s' s' X5 X' 4'

Simple 1 1 8 4 1 1 1 2 2
Labialized 0 0 0 0 0 0 1 1 0

Total 1 1 8 4 1 1 2 3 2

Only / s 5 / and /J'/ occur in systems not containing ejective affricates.

They are also the only ejective fricatives which occur without other
ejective fricatives in the same language. Note that, among pulmonic
fricatives, /f/ is almost as frequent as /J/, but the ejective /f'/ is
quite rare. Again, a labial place appears disfavored for an ejective.

Secondary articulations with ejectives

Labialization is the only secondary articulation which is at all common
with ejective segments. It occurs most often with ejective stops: 18
languages in the survey have labialized ejective stops. Among the stops it
only occurs with velars and uvulars (see Tables 7.2 and 7.3). Labialized
uvulars only occur if there is a labialized velar in the language, but both
these types are unusually common. Well over a third of the languages with
plain velar or uvular ejective stops also have their labialized
counterparts. Compare this with the fact that only about 13% of the
languages with the plosive /k/ have the labialized counterpart /k w /. The
apparent increase in labialization for velars and uvulars is probably due
to areal factors. Most labialized ejective stops are in North American
languages, where distinctive labialization of velars and uvulars frequently
applies to several manners of consonants in the same language. Except for
w
Kwakw'ala (731), with /k '/ but no /k'/, these labialized ejectives never
occur unless the plain counterpart also appears. In addition to /k w '/,
Hausa (266) has a palatalized velar ejective stop. Labialization also
occurs occasionally with certain ejective affricates and fricatives. The
only occurrences of labialized ejective affricates are in Lak (912), which
has /ts w '/ and /tj u '/. The only labialized ejective fricatives are those
w w
found in Tlingit (701), which has /x '/ and /x '/.

7.4 Voiceless laryngealized segments

Voiceless laryngealized segments are somewhat related to ejectives. They
have a glottal stricture simultaneous with the oral stricture, but this is

110
 Patterns of sounds

not used as initiator of an airstream. Three languages in the survey have a

set of voiceless laryngealized segments, Korean (070), Ashuslay (814), and
Siona (833). The first two have /p, t, k, tJ7 in common. Hausa (266) has
the voiceless laryngealized fricative / s/, which also occurs in both Korean
and Siona. This set of segments in Korean, usually called "fortis" or
"tense" obstruents, has been quite extensively studied, and several studies
agree that they are produced with a narrow glottal aperture (e.g. Kim
1970), although there is also evidence that there is accompanying tension
in the supraglottal structures (Kim 1965; Dart 1984). Hausa speakers vary
considerably in their production of /s/. This segment is sometimes
pronounced as an ejective fricative and it may also occur as an ejective
5
affricate [ts ], as well as occurring as a pulmonic fricative with a
glottal constriction. No other voiceless laryngealized fricative is
reported, although S. Nambiquara (816) has a rather obscure segment which
is described as a laryngealized /h/.

7.5 Implosives and voiced laryngealized plosives

As noted earlier, the only kind of glottalic ingressive segments reported
are stops, that is, implosives. Thirty-two (10.1%) of the languages of the
UPSID contain implosives. They are all voiced apart from the two segments
/pK / and /t < / in Igbo (116). There are a further 10 languages which have
voiced laryngealized plosives, making a total of 42 (13.2%). Since voiced
implosives and voiced laryngealized plosives have often been discussed
together, have not always been distinguished, and do not contrast we will
discuss them together in this section. We will use the notation /?b/ to
represent both /b/ and /b/, etc., including when quoting from other
authors. Greenberg (1970) used the term "injective" to cover both these
segment types. Following Haudricourt (1950) and Wang (1968), he noted that
"injectives tend to have front articulation". He goes on to suggest that

if a language has one injective, it is /?b/; if two, they are /?b/

and /?d/ (the most common pattern); if three, they are /?b/, /?d/,
and /?j/ (the latter a palatal stop, often replaced, however, by
[the laryngealized approximant] /]/) ; and, if four, they are /?b/,
;}/, and

The general preference for front articulations was borne out in a count of
the Stanford Phonology Archive by Javkin (1977). Table 7.7 gives the count
of voiced implosives and voiced laryngealized plosives in the languages of
UPSID, which shows a similar pattern for these stops at the different
places of articulation.

Ill
Glottalic and laryngealized consonants

Table 7.7 Voiced implosives and voiced laryngealized plosives in UPSID

Labial Dental/Alveolar Retroflex Palatal Velar Uvular

Vd. implosive 30 29 0 7 5 1
Vd. laryng'd. 9 7 1 0 0 0

39 36 1

These counts suggest that a correction might need to be made to the

implicational hierarchy posited by Greenberg. In fact he notes himself that
"there are a few languages whose sole injective is /?d/". The fact that
/?b/ and /?d/ are essentially equally frequent, and that either may occur
as the sole implosive in an inventory suggests that the hierarchy is blind
to the distinction between labial and alveolar implosives.
The structure of the systems of implosives and laryngealized plosives
is given in Table 7.8.

Table 7.8 Implosives and voiced laryngealized plosives in UPSID

Number of Implosive/laryngealized Number of

terms plosive inventory languages
?b 5 (1 also has p< , t< )
?d 2
?d 1
?b ?d 25
?b ?d ?;} 4
?b ?d ?g 1
?b ?d ?i ?g 3
?b ?d 1j ?G 1

Of the languages with only one of this class of segments, 5 have /?b/
(Kpelle, 103, Igbo, 116, and Zulu, 126, have /B/; Lakkia, 401, and K'ekchi,
714, have / b / ) . Berta (218) and Kullo (262) have /J7 alone. Somali (258)
has /d/# The number of cases is small, so any interpretation of the results
should be cautious. However, the systems with a single term here are
varied, unlike ejective stops where a single term is always /k'/« The
suggested revision of the hierarchical relationships of place in this case
would go as follows: the presence of /?b/ implies the presence of /?d/ or
of no other implosives, /?d/ implies the presence of /?b/ or of no other
implosives, while /?J7 implies the presence of both /?b/ and /?d/, and /?g/
implies the presence of /?b/, /?d/, and /?;}/.

112
 Patterns of sounds

As Greenberg observed, the system with two terms, one bilabial and one
dental or alveolar, is the most common. In fact it is the only common
system. All 25 languages with this inventory have velar stops in other stop
series. This is quite strong evidence that the velar place is disfavored
for voiced implosives and for voiced laryngealized plosives. An example of
such a system is that of Doayo (128):
p t k k'p
b d g gb
6 J
A 3-term system also generally avoids use of the velar place of
articulation, having members at the bilabial, dental or alveolar, and
palatal places. An example of such a system is Yulu (216):
p t c k k'p
b d g gb
6 <f f
As in Yulu, so also in the other 3 languages concerned (Kadugli, 102,
Angas, 267, Ngizim, 269) there is at least one plosive with a palatal place
of articulation in the inventory. One language, Hamer (265), stands apart
from the others with 3 implosive terms. It has them at bilabial, alveolar
and velar places. Despite the presence of palatal plosives it does not have
a palatal implosive. This language is discussed further below.
Three of the four 4-term systems contain /?b, *?d, ?j, ?g/• An example
is Nyangi (207):
p t c k
G S f cf
The other two languages concerned, Swahili (124) and Maasai (204), do not
have palatal plosives in their inventories, although they do have
palato-alveolar affricates. The unusual 4-term language is Ik (208), which
has /G, cf, f, G< /, that is, it has a uvular rather than a velar fourth
term. This is despite the fact that the language has no other reported
uvular segments and does have velar plosives.
All 9 of the languages in UPSID with more than two implosives or voiced
laryngealized plosives are from Africa. They are drawn from 3 different
major language families, Niger-Kordofanian, Nilo-Saharan, and Afro-Asiatic.
It follows also that the only languages using the palatal and velar places
for segments of this type are African languages. Despite the small number
of cases, this seems to be an important areal trend.

113
Glottalic and laryngealized consonants

We do not find evidence to support Greenberg's suggestion that

languages often have / j / , a laryngealized palatal approximant, in place of
/?J/. The most obvious candidate language in our sample to test this claim
is Hamer (265), since it has both palatal plosives and an iraplosive series
which lacks a palatal member. The Hamer stop inventory is:
p t c k
b d j g
k'
6 S cf
However Hamer does not have / j / despite the obvious "gap" at the palatal
place of articulation in the implosive series to which it could correspond.
The segment / j / does occur in 5 of the 25 languages with the two-term
system /?b, ?d/» but in all but one of these it occurs as part of a set of
laryngealized continuants including at least /w/. Only in Hausa (266) could
a case be made for considering / j / as complementing a gap in the stop
system. Moreover, in Hausa / j / is historically derived from "palatalized"
occurrences of /d/. But Hausa is a special case. It does not outweigh the
fact that most (28 of 33) languages with a series of implosives or voiced
laryngealized stops but with no palatal member of the series lack / j / , or
that most (8 of 13) languages with / j / lack any implosives or voiced
laryngealized plosives. It is true that no language in the survey has both
/?;}/ and / j / which might be evidence for the suppletion of /?;}/ by / j / .
However, given the low frequency of these segment types only one
co-occurrence of /?;}/ and / j / could be expected in a sample 3 times the
size of the UPSID sample even if the occurrences of the two types were
unrestricted. We therefore conclude that the suggestion that / j / takes the
place of /?j/ is unfounded. For discussion of / j / in relation to other
approximants see Chapter 6.

Retraction of dental/alveolar implosives

Both Greenberg (1970) and Haudricourt (1950) noted that an implosive
corresponding to a non-implosive dental/alveolar is often retroflexed or,
at least, articulated further back than the non-implosive. In UPSID a
retroflex implosive or laryngealized voiced plosive occurs in only one
language, Somali (258), in which the plain plosives are dental. Two other
languages, Tama (210) and Yulu (216) have dental plosive / I , d/ but
alveolar implosive /cf/. Apart from these instances, there is insufficient
phonetic detail in most of the UPSID sources to determine if implosives are

114
 Patterns of sounds

typically articulated with a further back contact than other stops made
with the tongue tip and blade, A study of voiced alveolar plosives and
implosives in Shona using dynamic palatography (Hardcastle and Brasington
1978) did find a more retracted contact (with a smaller area) for the
implosive relative to the plosive for the speaker studied.

Voiceless implosives
Overwhelmingly, implosives are voiced. However voiceless implosives do
occur. Only one language in UPSID, Igbo (116), has any voiceless
implosives. Ladefoged et al. (1976) demonstrate that both voiced and
voiceless bilabial implosives occur in the Owerri dialect of Igbo, so that
/p < / is in contrast with /B/. Igbo also has the voiceless alveolar
<
implosive /t /. Pinkerton (1980) shows that the uvular ejective stop in
K'ekchi (714) can vary allophonically with a voiceless uvular implosive,
and in certain related Quichean languages the voiceless uvular implosive
occurs as the normal case (see Table 7.4 above). Other examples of
voiceless implosives are mentioned by Campbell (1973).

7.6 Languages with both ejective stops and implosives

Thirteen languages in UPSID contain both some ejectives and some implosives
or voiced laryngealized plosives. Given the numbers of languages which have
segments of these two classes, we would expect only 6 or 7 such languages
in the sample if their occurrence was unassociated. The larger number
suggests that these two classes of segments have a tendency to occur
together in a language. One might also expect, given the differing place of
articulation preferences for ejectives and implosives, that they would
rarely occur at the same place of articulation. In several cases, however,
they do. Zulu (126), Koma (220), Maidu (708) and Otomi (716) have both / p 7
and /?b/. Koma, Kullo (262), Maidu, Otomi, Mazahua (717), and Southern
Nambiquara (816) have both / * t 7 and /*?d/. Hamer (265) has / k 7 and /^/.

7.7 Laryngealized sonorants

The glottalic airstream mechanism is not used in the production of sonorant
types of segments, but sonorants do occur laryngealized. As Greenberg
(1970) noted, "the phonological opposition in individual languages between
ejectives and injectives applies effectively only to obstruents, and is
neutralized for sonants and semi-vowels". In other words, he regards
laryngealized sonorants as counterparts in some way to the glottalic

115
Glottalic and laryngealized consonants

obstruents, although with a neutralization of the airstream contrast. There

is evidence that glottalic obstruents and laryngealized sonorants are
members of the same phonological class in at least some languages. For
example, in Hausa any segment which is either glottalic or laryngealized
may not co-occur with a different glottalic or laryngealized consonant in a
word (Carnochan 1951). What is relevant here is that this rule disallows
co-occurrence of the laryngealized sonorant / j / with glottalic obstruents
/k'/, /s/ etc.
Greenberg also noted that "there is quite surely no phonological
contrast of voicing" for the laryngealized nasals and liquids, nor, we
might add, for the laryngealized approximants and vowels. All segments of
these classes in UPSID are reported as voiced. Moreover, in all cases a
plain voiced counterpart also occurs in the inventory of a language if
there is a laryngealized sonorant.
In general, laryngealized sonorants are found only in languages with
glottalic stops. Nineteen of the 20 languages in UPSID which have
laryngealized sonorants have ejective stops, implosives or voiced
laryngealized plosives in their inventories. The exception is Tiddim Chin
(513) with /I/ and /w/ but no other segments with a "glottalic" component.
Within this group of sounds, laryngealized nasals and vocoid approximants
are a little more frequent than laryngealized liquids, as Table 7.9 shows.

Table 7.9 Laryngealized nasals, liquids and

approximants in UPSID

m *n p Q "*r" *l j w
14 14 3 3 5 8 13 14

The distribution of laryngealized nasals can be seen to be parallel to that

of voiced "glottalized" stops in that both bilabial and dental/alveolar
places are more common than back articulations, but yet are not between
themselves in hierarchical relation. In most cases (13 of 17 languages
concerned) /m/ and /*n/ occur together just as /?b/ and /*?d/ occur
together. However, note that among plain voiced nasals bilabial and
dental/alveolar places are also the most common and the velar place is less
common (see Chapter 4 ) . Hence it is not certain if the prevalence of /m/
and /n/ among laryngealized nasals should be attributed to their
laryngealized nature or to their simply being nasals.

116
 Patterns of sounds

The laryngealized approximants I}/ and /w/ usually occur together (12
of 16 languages concerned). There are 5 languages with a laryngealized
trill, tap or flap (collected in the table under the symbol /"*r"/). One
language, Wapishana (822), has a laryngealized voiced retroflex fricative,
id.

7•8 Diachronic implications

Greenberg (1970: 23) suggests that "it is possible to derive the general
diachronic hypothesis that at least one source of injectives might be a
sound shift from voiced plain to voiced implosive stops". This is based on
his observation that languages with implosives tend to lack corresponding
non-implosive voiced stops. Greenberg also suggests that loss or addition
of implosives should follow the place of articulation preference hierarchy
for implosives discussed above. One reasonable prediction to read into his
diachronic suggestions is that implosives at the same place of articulation
as voiced plosives should be rather rare. This, however, is not the finding
of our survey. As noted above, languages with implosives most commonly also
have voiced plosives corresponding to each implosive. In a very few cases,
such as Swahili (124), where there is no voiced plosive series, it does
appear that the source of the implosives may be from an earlier voiced
plosive series (Guthrie 1967-70). But even this case is unclear because it
is conceivable that Swahili has merged a voiced plosive series with an
implosive series; Stewart (1972) has found reasons to posit Proto-Bantu
implosives in addition to voiced plosives. The failure to confirm
Greenberg's prediction does not completely refute the diachronic
hypothesis, since in the languages that do not conform to the prediction
there may also have been a shift in another stop series to replace the
former plain voiced series, or the original voiced plosive series may have
split into plosive and implosive sets. However, it does weaken the evidence
for positing voiced plosives as the straightforward source of implosives.

7.9 Phonetic explanations for the structure of glottalic systems

A phonetic explanation for Greenberg's (1970) place of articulation
hierarchies of ejectives and implosives has been offered by several
linguists, including Greenberg himself. Recall that some weakening of the
preference hierarchies posited by Greenberg has been suggested. For
ejectives, although a velar place of articulation is common, it is only
marginally more common than a dental/alveolar one; however, both these

117
Glottalic and laryngealized consonants

places are preferred to a bilabial one. For implosives, etc., bilabial and
dental/alveolar places of articulation are equally common, and both are
preferred to velar.
Javkin (1977) clarified the role played by Boyle's Law in explaining
implosive and ejective distributions, correcting a misinterpretation by
Greenberg. Javkin (1977) noted that the claim that back articulations
confer an advantage in compressing air in the supraglottal chamber (for
ejectives) and front articulations confer an advantage in rarefying air
(for implosives) cannot be entirely correct. This is because it takes the
same effort to produce either compression or rarefaction in a chamber of a
given size. What matters is the proportional change in the size of the
chamber. Javkin's model suggests that the ability to change the volume of
the chamber is proportionally greater for a velar closure than for a
bilabial, dental or alveolar closure. That is, the same amount of raising
or lowering of the larynx will have a greater effect on the volume of air
between a velar closure and a glottal closure than if the oral closure is
further forward. If articulatory efficiency were the explanation for the
difference in the place preferences for ejectives and implosives, and there
were nothing else to consider, then both types of sounds would show a
preference for back articulation. It is possible to maintain this kind of
explanation for the preference for velar ejectives, providing some
overriding factor can be found to explain why implosives do not share the
preference for back articulations. Javkin suggests that this factor is
voicing.

As noted above, implosive segments are almost invariably voiced. If we

grant, for the time being, that voicing is an essential part of their
nature, then anything that facilitates voicing will be favored. In a voiced
segment some volume of pulmonic air must flow into the oral cavity. The
absolute volume by which the oral cavity is expanded must be greater than
this volume of pulmonic air or there will be no rarefaction, hence no
implosion. A chamber created by a back oral closure may not permit
expansion by the required absolute volume, whereas one further forward may
allow greater absolute expansion through adjustments of tongue position and
oral cavity walls. Such oral cavity expansion has been shown to be normal
in voiced plosives in several studies of English (e.g. Kent & Moll 1969;
Smith 1971; Bell-Berti 1975; Westbury 1983), and Lindau (1982) specifically
argues that the achievement of an even or rising amplitude of voicing
throughout the closure may be a major part of the "target" in production of

118
 Patterns of sounds

an implosive. Lindau compares the bilabial implosive and plosive in the

Niger-Kordofanian language Degema and finds that whereas the amplitude of
voicing actually increases during the closure for the implosive, during the
plosive the amplitude of voicing tends to decay and the periods become
irregular. Of course, some air can flow through the glottis without any
cavity expansion being required. Again, the further forward the oral
closure is formed, the less the intraoral air pressure increases for a
given volume of transglottal air flow.
No measurements have been done to confirm the occurrence of oral cavity
expansion by tongue movement, jaw lowering or use of the cheeks in
production of implosives, although Hardcastle and Brasington (1978) show a
pattern in the occlusion for an alveolar implosive in Shona which is
consistent with some cavity expansion by tongue lowering. Nonetheless, the
theory that such expansion occurs is plausible and appealing. In addition
to accounting for a preference for bilabial implosives, it also suggests
why the patterns for voiced laryngealized plosives are similar to those for
true implosives. For, as Greenberg (1970) and Ladefoged (1968) have noted,
"implosives" do not always entail inward air flow upon release. In some
languages (e.g. Hausa, 266) some speakers use implosives while others use
laryngealized stops. If implosion, especially the contribution of the
lowering of the larynx to oral cavity expansion facilitating transglottal
air flow, is predominantly a mechanism employed in the more general aim of
achieving salient voicing in the production of a stop, then actual
achievement of inward oral air flow is a minor part of the target. This
means that the cavity expansion need only equal, not exceed, the volume of
transglottal air flow.

Note that we are not suggesting that plain voiced plosives are likely
to become implosives (pace Greenberg) as speakers endeavor to sustain
voicing through the duration of the oral closure. Instead it see"ms more
likely that, given a contrast between voiced stops of two different types
in a language, a tactic for enhancing the contrast between them by
implosion and emphasized voicing is exploited. In this view, implosives and
voiced laryngealized plosives would be expected to co-occur with plain
voiced plosives, as they do. Because the cavity expansion possibilities are
greatest when there is a front articulation, not only is a front
articulation preferred when the target is to achieve enhanced voicing, but
also when the target is for an "ordinary" degree of voicing, since if the
oral cavity is not expanded, voicing will cease when the oral air pressure

119
Glottalic and laryngealized consonants

reaches equilibrium with the subglottal air pressure. Hence place

preferences for voiced implosives and for voiced plosives are similar (see
Chapter 2 ) . The disfavored velar implosives, if they arose, would tend to
fall together with voiced velar plosives, since it would be difficult to
maintain a distinction between them on the basis of differences in their
voicing characteristics.
On the other hand, since production of voiceless plosives has nothing
to do with rarefaction of air in a closed chamber, we would expect place
preferences for ejective stops and voiceless plosives to differ, as they
do. A small chamber confers no advantage in producing voiceless plosives,
hence no preference for a velar place is found. Instead, dental or alveolar
plosives are most common, perhaps because the tongue tip/blade is the most
mobile of all the articulators. Note that this view predicts that voiceless
implosives should show the same preferences for back articulations as
ejectives, rather than those found for voiced implosives, since the
preference for front articulation in voiced implosives is attributed to the
intention to achieve voicing. The data in UPSID is insufficient to deal
with this question. The one language with voiceless implosives, Igbo (116),
looks like a counterexample since it has them at bilabial and alveolar
places, and not at the velar place, but note that several of the Mayan
languages cited in Table 7.4 have uvular implosives which are voiceless.

7.10 Summary of generalizations

In this section we recapitulate the most important observations about
glottalic and laryngealized consonants discussed in this chapter. After
each statement, a fraction is given which represents the number of cases
which conform with the statement over the number of relevant cases. The
percentage of conforming cases is also given. Note that some of the
statements are made about segments and some are made about languages.
(i) An ejective segment is voiceless. 309/312 99.0%.
(ii) An ejective segment is likely to be a stop. 188/312 60.3%.
(iii) If a language has / p V it also has /*t'/. 33/34 97.1%.
(iv) If a language has /* t 7 it also has /k'/. 45/46 97.8%.
(v) If a language has only one ejective stop, it is / k V . 5/5 100%.
(vi) If a language has / c '/ or / q '/ it also has / p 7 , /*t'/ and /k'/.
15/19. 78.9%.
(vii) If a language has / q w 7 it also has /q'/ and /k w '/. 8/8 100%.
(viii) If a language has /k w '/ it also has /k'/. 17/18 94.4%.

120
 Patterns of sounds

(ix) If a language has ejective affricates it also has ejective stops.

39/40 97.5%.
(x) An ejective affricate segment is usually sibilant. 70/88 79.5%.
(xi) If a language has any ejective fricatives, at least one of them is
sibilant. 10/10 100%.
(xii) An implosive segment is voiced. 72/74 97.3%.
(xiii) A language with any implosives or laryngealized voiced stops has
/?b/ and /?d/. 36/42 85.7%.
(xiv) If a language has /?g/ it has /?j/. 3/4 75.0%.
(xv) Laryngealized sonorants are voiced. 74/74 100%.
(xvi) If a laryngealized sonorant segment occurs, the plain voiced
counterpart occurs in the same language. 74/74 100%.
(xvii) If a language has any laryngealized sonorants it also has
glottalic or laryngealized stops. 19/20 95.0%.

Notes
1. Hoard (1978) reports, as secondary allophones of the ejective
affricates /ts 5 / and /t4'/, the voiced implosive counterparts [cfz] and
[cfl^l in the Tsimshian language Gitksan (not included in UPSID). From
the description provided it is not clear if these voiced alternates of
the voiceless ejective affricates are actually imploded or not.
However, since they contrast phonetically with corresponding plain
voiced affricates in certain environments, it is probable that their
articulation is, if not implosive, at least accompanied by laryngeal
constriction. Still, the particular allophone characterizable as "most
typical" for these segments (i.e., that which would be coded by UPSID)
is in both cases the voiceless ejective affricate and hence it is still
true that no phonemic voiced implosive affricate segment is known to
occur. A similar analysis is presumably maintainable in regard to the
voiced palatal implosives of Fula and Serer (also not in UPSID) which
are slightly affricated (Ladefoged, personal communication).

References
Bell-Berti, F. 1975. Control of pharyngeal cavity size for English voiced
and voiceless stops. Journal of the Accoustical Society of America 57:
456-64.
Campbell, L. 1973. On glottalic consonants. International Journal of
American Linguistics 39: 44-6.
Carnochan, J. 1951. Glottalization in Hausa. Transactions of the
Philological Society 1951: 78-109.
Catford, J. 1939. On the classification of stop consonants. Le Maitre
Phonetique, 3rd scries, 65: 2-5.
Dart, S. 1984. Testing an aerodynamic model with measured data from
Korean. UCLA Working Papers in Phonetics 59: 1-17.
Fordyce, J.F. 1980. On the nature of glottalic and laryngealized consonant
and vowel systems. UCLA Working Papers in Phonetics 50: 120-54.

121
Glottalic and laryngealized consonants

Gamkrelidze, T.V. 1978. On the correlation of stops and fricatives in a

phonological system. In J. H. Greenberg et al. (eds.), Universals of
Human Language, Vol. 2^ Phonology. Stanford: Stanford University Press.
Greenberg, J.H. 1970. Some generalizations concerning glottalic consonants,
especially implosives. International Journal of American Linguistics 36:
123-45.
Guthrie, M. 1967-70. Comparative Bantu. Gregg, Farnborough.
Hamp, E. 1970. Maya-Chipaya and the typology of labials. Papers from the
Sixth Regional Meeting, Chicago Linguistic Society: 20-2.
Hardcastle, W.J. and Brasington, R.W.P. 1978. Experimental study of
implosive and voiced egressive stops in Shona: an interim report. Work-
in Progress (Phonetics Laboratory, University of Reading) 2: 66-97.
Haudricourt, A-G. 1950. Les consonnes preglottalisees en Indochine.
Bulletin de la Societe Linguistique de Paris 46: 172-82.
Hoard, J.E. 1978. Obstruent voicing in Gitksan: some implications for
distinctive feature theory. In E-D. Cook and J. Kaye (eds.), Linguistic
Studies of Native Canada. University of British Columbia Press,
Vancouver.
Javkin, H. 1977. Towards a phonetic explanation for universal preferences
in implosives and ejectives. Proceedings of the Berkeley Linguistics
Society; 3 : 559-65.
Kent, R.~D. and Moll, K.L. 1969. Vocal tract characteristics of the stop
cognates. Journal of the Acoustical Society of America 46: 1549-55.
Kim, C-W. 1965. On the autonomy of the tensity feature in stop
classification (with special reference to Korean stops). Word 21:
339-59.
Kim, C-W. 1970. A theory of aspiration. Phonetica 21: 107-16.
Ladefoged, P. 1968. A Phonetic Study of West African Languages. Cambridge
University Press, Cambridge.
Ladefoged, P., Williamson, K., Elugbe, B. and Uwalaka, A.A. 1976. The stops
of Owerri Igbo. Studies in African Linguistics, Supplement 6: 147-64.
Lindau, M. 1982. Phonetic differences in glottalic consonants. UCLA Working
Papers in Phonetics 54: 66-77.
Pinkerton, S. 1980. Quichean (Mayan) glottalized and non-glottalized stops:
a phonetic study with implications for phonological universals. Report
of the Phonology Laboratory (University of California, Berkeley) 5:
106^11
Sherzer, J. 1973. Areal linguistics in North America. In T.A. Sebeok (ed.)
Current Trends in Linguistics, Vol. 10, Linguistics in North America.
Mouton, The Hague: 749-95.
Smith, T.S. 1971. A. Phonetic Study cȣ the Function ot_ the Extrinsic Tongue
Muscles (UCLA Working Papers in Phonetics f8). University of California,
Los Angeles.
Snyman, J.W. 1969. An Introduction to the !Xu (!Kung) Language. A.A.
Balkema, Cape Town.
Stewart, J.M. 1972. "Implosives" in Proto-Bantu. Paper presented at the
Tenth West African Languages Congress, Legon, Ghana.
Wang, W.S-Y. 1968. The basis of speech. Project on Linguistic Analysis
Reports, second series, No. 4. University of California, Berkeley.
Westbury, J. 1983. Enlargement of the supraglottal cavity and its relation
to stop consonant voicing. Journal of the Acoustical Society of America
73: 1322-36.

122
Vowels

8.1 Introduction
Vowels are discussed in two chapters in this book. In the present chapter
we will deal with questions of which vowels are most frequent in the
world's languages and how many vowels are typically found. A brief
discussion of diphthongs is also included. In the following chapter,
contributed by Sandra F. Disner, the focus is on the structure of vowel
systems.

8.2 Types of vowels

The entire UPSID database contains entries for 2549 monophthongal vowels,
that is, a mean of fractionally over 8 for each language. In addition there
are 83 diphthongal segments included, making a total of 2632 vocalic
segments. In the following sections, we will describe the most frequent
types of vowels and comment on some of the implications.
The three conventional parameters for vowel description are those of
vowel height, backness, and lip-rounding. In the UPSID data file, vowels
are classified as having one of five different heights, high, higher mid,
mid, lower mid, or low. Vowels described as being mid may in fact lie
between higher and lower mid positions, or they may have simply been
transcribed or labeled as mid vowels without any further specification in
the source consulted for the language in question. In either case, such
vowels are distinguished notationally from higher mid vowels by being
enclosed in double quote marks, i.e. /e/, / 0 / , and /o/ represent higher mid
vowels, but /"e'7, /'V'V, and /"o"/ represent mid vowels. We will also use
the term "in the mid range" to cover the three height positions higher mid,
mid, and lower mid as a group. Within each height category, further

123
Vowels

classification is made with the use of a variable called "nonperipheral";

for example /i/ differs from /i/ in that it is nonperipheral. The UPSID
variables permit vowels to be classified on the front/back dimension as
being front, central, or back. Again, finer differences can be represented
by defining a vowel as nonperipheral. Only two lip positions, unrounded or
rounded, are provided for in UPSID.
In terms of these three basic parameters of vowel quality, our survey
provides few surprises. It largely supports the findings of previous
studies of vowel inventories such as those of Hockett (1955), Sedlak (1969)
and Crothers (1978). It is, however, wider in scope and is able to add more
reliable quantification. The total of 2549 monophthongal vowels is broken
down according to these parameters in Table 8.1.

Table 8.1 Broad classification of types of vowels in UPSID

Front Central Back

Unround Round Unround Round Unround Round
Totals
High 452 29 55 10 31 417 | 994
Mid range 425 32 100 8 19 448 | 1032
Low 81 0 392 1 13 36 1 523

Totals 958 61 547 19 63 901 | 2549

1019 566 964

We may make the following observations. Vowels in the mid range are a
little more common than high vowels, namely 1032 to 994, or 40.5% of the
sample to 39.0%. Low vowels are substantially less common, amounting to
only 20.5%. There are slightly more front vowels than back vowels, namely
1019 to 964, or 40.0% of the population compared with 37.8%. Central vowels
are considerably less common, amounting only to 22.2%. Unrounded vowels are
considerably more frequent than rounded vowels, namely 1569 to 981 or 61.5%
to 38.5%.l
There are some interesting asymmetries when the interactions of the
three basic parameters are looked at. Front vowels are usually unrounded
(94.0%), back vowels usually rounded (93.5%). Low vowels are usually
central (75.1%) and central vowels are usually low (69.4%). High front
vowels are more frequent than high back vowels. In the mid range, vowels

124
 Patterns of sounds

are more commonly back than front if the lip position is unmarked (i.e.
unrounded if front, rounded if back), but front rounded vowels are more
frequently found than back unrounded. Nonback low vowels are extremely
unlikely to be rounded (only one case in 474).
The most common individual vowel qualities reported are given in Table
8.2. This table lists those vowels which are found to occur in at least 30%
of the languages in the survey. The number of languages shown here is the
number which have the given vowel quality either long or short; for
example, 290 is the number of languages out of the UPSID sample of 317 that
have one or both of /]/ and /i:/. Recall that /"o'7 and /"e'7 may not be
reliably distinct from other vowels in the mid range.

Table 8.2 Most common vowel qualities

Vowel Number of Percent

languages
High and low vowels
IM 290 91.5%
279 88.0%
Nl 266 83.9%
Vowels in the mid range
/"o"/ 139 43.8%
/"e'7 118 37.2%
lei 118 37.2%
/o/ 109 34.4%
lei 100 31.5%
lol 99 31.2%

The 3 vowels at the corners of the conventional vowel triangle, /i, a, u/,
are the most widespread, but note that there are 24 fewer languages with
/u/ than with /i/. These three vowels might be expected to be equally
favored, because they each lie at an acoustic extreme. The low vowel /a/
has the highest first formant, /i/ and /u/ have the lowest first formant
but I\I has the highest second (and third) formant, whereas /u/ has the
lowest second formant. However, a contributory factor to the relative
disfavoring of /u/ may be the lower amplitude typical of /u/. In the higher
part of the mid range, note that there are substantially more cases of back
lol and /"o'V than of the front vowels lei and /"e11/, but among lower mid
vowels lei is a little more common than ID/. The within-language

125
Vowels

asymmetries that contribute to these differences are analyzed further in

Chapter 9.

8.3 Number o_f_ vowels per language

The smallest number of phonemic vowels in any of the UPSID languages is 3
and the largest is 24. The modal number is, unsurprisingly, 5. The number
of languages with a given number of simple vowels is shown in Table 8.3.

Table 8.3 Number of languages with given

number of vowels

No. of vowels No. of languages Percent

3 18 5.7%
4 15 4.7%
5 68 21.5%
6 43 13.6%
7 34 10.7%
8 24 7.6%
9 28 8.8%
10 16 5.0%
11 11 3.4%
12 18 5.7%
13 4 1.3%
14 7 2.2%
15 8 2.5%
16 10 3.2%
17 and over 13 4.1%

Some languages have been analyzed in the linguistic literature as having

fewer than three phonemic vowels. The best known are Kabardian (911) and
Abaza (Allen 1965; Anderson 1978). However, more conservative analyses of
these languages can be defended in which less of the contrast between
syllables is attributed to the consonants. This approach results in an
analysis in which they have only three vowels - it is clear that these are
languages with a small number of vowel contrasts under any analysis. As
noted in Chapter 1, we do not find that having a small number of vowels in
general predicts a large number of consonant contrasts, although this is a
feature of the Kabardian inventory which has been commented on a great
deal. Note that languages with 3 vowels also include Mura (802) with only 8

126
 Patterns of sounds

consonants, Gugu-Yalanji (364) with only 13 consonants and Alabama (759)

with only 14.

8.4 Distinctive vowel qualities

Many languages have several series of vowels, for example, short and long
vowels, oral and nasalized vowels, plain voiced and laryngealized vowels.
In the greater number of such cases, the vowels in one series can be
matched with vowels that are similar in quality in other series, so the
number of vowel phonemes is greater than the number of different vowel
qualities in the languages concerned. A simple example of this is a
language such as Mazatec (727) which has oral vowels /i, e, a, o/ and
nasalized vowels /T, e, a, 5/. While there are 8 phonemic vowels here,
there are only 4 vowel qualities involved. In other languages there may be
qualities in one series which do not occur outside that series. An example
is Zande (130) which has /e, 5/ but not /e, o/. In such a language the
total number of vowel qualities may be larger than the number found in any
one given series. The count of vowel qualities in the UPSID languages is
given in Table 8.4.

Table 8.4 Number of vowel qualities

No. of vowel No. of Percent of

qualities languages languages
3 17 5.4%
4 27 8.5%
5 98 30.9%
6 60 18.9%
7 47 14.8%
8 17 5.4%
9 25 7.9%
10 15 4.7%
11 2 0.6%
12 5 1.6%
13 2 0.6%
14 0 0.0%
15 2 0.6%

In certain respects the number of distinctive vowel qualities in a language

provides a more appropriate measure for comparing vowel systems than does

127
Vowels

the total number of vocalic phonemes. This is because the number of vowel
qualities indicates how greatly the most basic parameters of vowel contrast
(height, backness, rounding) are being used. No language in UPSID has less
than 3 vowel qualities. The most common number, found in almost a third of
the languages, is 5. Almost two-thirds of the languages have between 5 and
7 vowel qualities, although up to 10 is still relatively common. The two
languages with the largest number of vowel qualities in UPSID are the two
Germanic languages in the survey, German (004) and Norwegian (006), which
have 15. Contrasts of a large number of vowel qualities seem
disproportionately common in Indo-European languages - of the 11 languages
with more than 10 vowel qualities, 5 are Indo-European. Almost 24% of the
Indo-European languages included in UPSID have over 10 vowel qualities,
whereas only 3.5% of the total sample have that many.

8.5 Properties of vowel series

The tabulation in Table 8.4 shows the numbers of vowel qualities that are
found in one or more of the series of vowels that a language may have. In
this section we will discuss the properties that distinguish vowel series
from each other. The most important of these are length and
nasalization.

Length
Vowel length contrasts (short vs. long, or short vs. overshort) are only
recorded as phonemic in UPSID if they are linked to vowel quality
differences. In other words, if all the vowel qualities found in a language
participate in a length contrast, length is treated as a suprasegmental
feature or as resulting from a juxtaposition of simple vowels rather than
as a property of individual phonemes. There are three types of situations
where vowel length is represented in the phoneme inventory. In some
languages the long and short vowel sets do not overlap in quality; such a
case is Kurdish (015) with a 9-vowel inventory consisting of the following:
short long
i + o i : u:
"a" "e*" "o:"
a:
In such a case, length could have been treated as predictable from vowel
quality. More commonly, some vowels in each set have the same quality, as
in Tonkawa (752) with /u, u'/ and /a, a*/ pairs but with short /i, e, o/

128
 Patterns of sounds

versus long /i :, e:, o:/. Tonkawa is thus considered a language with 8

vowel qualities and 10 vowel phonemes. An alternative analysis of the
language might say that the quality differences were predictable from
length, which would result in this being treated as a language with a
5-vowel system. But here as in other areas, UPSID has preferred to keep
possibly redundant phonetic information in the data.
The third situation in which length is represented is where the
qualities of the longer vowels are a subset of the qualities of the shorter
vowels or vice versa, as, for example, in Atayal (407) with 5 short vowels
/i, e, a, o, u/ and 2 long vowels /i:, u:/, or Yurak (056) with 5 short
vowels /i, e, a, o, u/ and 3 overshort vowels /T, a, u/. These two
languages have 5 vowel qualities but 7 and 8 vowel phonemes respectively.
In all three situations outlined here, the property of length (or
shortness) only applies to vowels of certain qualities in the inventory,
hence it is considered to be inherent in the phoneme. With contrastive
length defined in this way, there are 62 languages in the sample with
length contrasts, or 19.6%.
The probability of length being part of the vowel system increases with
the number of vowel quality contrasts. No language with 3 vowel qualities
includes length, only 14.1% of the languages with 4-6 vowel qualities have
some inherent length differences, whereas 24.7% of languages with 7-9 vowel
qualities have length, and 53.8% of languages with 10 or more vowel
qualities have length. We may speculate that there are two diachronic
factors responsible for this trend; languages with an originally
suprasegmental vowel length contrast may begin to add quality differences
to the quantitative difference (probably the case in Tonkawa, as in Navaho,
702, Arabic, 250, Telugu, 902, etc.), and languages with large numbers of
qualitatively distinct vowels may begin to recruit length differences to
additionally distinguish them (there are no clear cases of this in UPSID
but compare the ongoing lengthening of /ae/ in English, perhaps to
distinguish it from /e/). In either case the outcome is the same - the use
of combinations of durational and qualitative differences to mutually
reinforce the distinctiveness of vowel contrasts.
Among the major vowel qualities there are some important asymmetries
concerning length. We may summarize the most important of these as follows:
a higher-mid vowel is more likely to be long than a lower-mid vowel; a low
vowel which is either front or back is more likely to be long than a low
central vowel; a front rounded vowel is more likely to be long than a front

129
Vowels

unrounded vowel. Table 8.5 gives the "length ratio" of selected vowel
qualities, calculated by expressing the number of long vowels as a
proportion of the number of languages with vowels of the given quality
whether long or short (as in Table 8.1).

Table 8.5 Length ratios of common vowel qualities

Front Back
Mid vowels
e:/e .280 o :/o .284
"er'Ve" .088 "oi'VV .098
e:/e .103 0 :/ 0 .090
Low vowels
ae:/ae .256 a :/a .318
a:/a .129
Selected other vowels
i:/i .151 u:/u .146
y:/y .238
oe:/oe .400

The higher mid long vowels lei I and I oil are far more likely to appear in a
language without corresponding short vowels of the same quality than any of
the other vowels examined. In 18% of the languages with the vowel quality
/e(:)/ and 19.6% of the languages with the vowel quality /o(:)/ the vowel
only occurs long. For comparison, this figure is 6.6% for the vowel quality
/i(:)/, 4.9% for /u(:)/ and 2.9% for /a(:)/. This suggests that mid vowels
tend to be raised when lengthened and/or lowered when shortened, giving
rise to associations between height and length. A well-known diachronic
example of this occurred in Late Latin (Griffiths 1966), and the subsequent
loss of length contrasts left Italian with the 4 mid vowels /e, e, o, o/ in
place of Classical Latin /e:, e, o:, o/. This illustrates one specific
direction the qualitative reinforcement of quantity differences referred to
in the previous paragraph can take.

Nasalization
Vowel nasalization is more common than inherent vowel length in the UPSID
languages, 71 languages or 22.4% of the sample having a contrast of oral
and nasalized vowels. There is again a tendency for this feature to be more
likely to occur in languages with a larger number of vowel quality

130
 Patterns of sounds

contrasts but the trend is less pronounced than with length: 21.2% of
languages with 4-6 vowel qualities have contrastive nasalization, 22.5% of
languages with 7-9 vowel qualities have it, but 53.8% of languages with 10
or more vowel qualities have it. This distribution arises in part because
vowels with nasalization sometimes have different qualities from their
closest oral counterpart; for example the 3 nasalized vowels of Burmese
(509) are reported as / I , ae, 5/ but the closest counterparts among the 8
oral vowels are given as /i, a, u/. Thus Burmese arrives at a total of 11
vowel qualities. However, such cases are relatively rare; only 4 of the 14
languages with 10 or more vowel qualities that have nasalized vowels
include any vowel qualities that are found in the nasalized set alone.
These are Dan (106), Zande (130) and Sara (217) as well as Burmese. These 4
languages amount to only 15.4% of the 26 languages with 10 or more vowel
qualities.
By far the most common nasalized vowels are the 3 vowels whose oral
counterparts are also most common. There are 59 languages with /!/ or its
long counterpart, 58 with /a(:)/ and 55 with /u(:)/. These frequencies are
representative of a general pattern: nasalized vowel frequency is generally
correlated with the frequency of the oral equivalent. The number of
occurrences of a given nasalized vowel is about one-fifth the number of the
oral counterpart. However, there is one salient exception to this pattern
among the more common vowel qualities, and that is the higher mid vowel
/e/. Only 11 languages have /e(:)/ whereas 22 have /e(:)/. There is no
similar discrepancy between higher and lower mid back rounded vowels, in
fact 21 languages have /o(:)/ whereas only 19 have /5(:)/. Lowering of mid
(and high) vowels diachronically and allophonically has been frequently
commented on (e.g. Foley 1975; Wright 1980) but no asymmetry between front
and back vowels appears to have been mentioned before. A possible
interpretation of the difference between front and back mid vowels would be
that the lowering process leads to a greater lowering of front vowels than
of back vowels in the mid range under nasalization. This could cause /e/
and /e/ to merge as /e/. Wright (1980) showed how the acoustic effect of
nasalization was perceived as a lowering of vowel height in mid vowels but
his analyses of the perceptual distance between oral and nasal equivalents
of a set of vowels do not explain the asymmetry found in the UPSID data. If
anything, they suggest that there is a greater perceived lowering of back
vowels under nasalization than of front vowels.

131
Vowels

Among less frequent vowel qualities there may also be a significant

exception to the usual pattern. On the basis of the usual ratio of oral to
nasalized vowels, 4 or 5 nasalized high front rounded vowels would have
been anticipated to occur in the survey. But there are no occurrences of
/y/ (or / ? / ) . Since other front rounded vowels and unrounded high front
vowels occur nasalized with the expected frequency it is not clear why the
combination of rounding and nasalization does not occur with high front
position.

Other properties of vowel sets

The other properties which establish sets of vowels in some of the
languages in UPSID are pharyngealization, laryngealization and breathy
voice. There are 5 languages with contrastive sets of pharyngealized
vowels: Evenki (067), Neo-Aramaic (255), Hamer (265), Lak (912) and !Xu
(918). !Xu also has vowel sets resulting from contrasts of nasalization and
length. The intersection of these features with the pharyngealization
contrast produces 8 vowel sets. Two languages are analyzed as having
contrastively laryngealized vowels, Sedang (304) and S. Nambiquara (816).
In the latter language, nasalization intersects with laryngealization to
produce 4 sets of vowels. Two languages have contrastive voiceless vowels,
Ik (208) and Dafla (508), and one language has breathy vowels, Tamang
(507). The source consulted for this language treats the breathiness as an
inherent part of a set of contrasting tones rather than localizing it in
the vowel. But the two "breathy tones" have the same pitch shapes as the
two plain tones, therefore Tamang is regarded here as a language with just
two tones and a segmental contrast of breathiness.

The instances of these various properties are too few to encourage very
much generalization, although it should be noted that no language has a set
of these "marked" vowels which contains more contrasts than is found in the
set of plain voiced vowels. In this respect these features appear more like
nasalization than like length.
In the languages with laryngealized, voiceless or breathy vowels, the
vowels in these sets have the same qualities as vowels which are found in
the plain voiced vowel set. But in a number of cases pharyngealized vowels
are more centralized than the closest non-pharyngealized vowel. For
example, Evenki has /u/ but / o /, Hamer has / j , e > a, o» u/ but /i , "e ",
e , o , o /, and Lak has /i, a, u/ but /"e^", ae , "o "/. These variations
in vowel quality are reminiscent of the qualitative differences between

132
 Patterns of sounds

sets of vowels in vowel harmony languages whose harmony is based on

variable pharynx wid'th (tongue root advancement). Among languages of this
type in UPSID are Akan (115, see Lindau 1979), Igbo (116) and Luo (205, see
Jacobson 1978). 2 In the vowel harmony languages there does not seem to be a
strong percept of pharyngeal constriction for the vowels in the set with
narrow pharynx. Instead, the vowels in the two sets are usually
well-distinguished auditorily by the basic parameters of vowel quality, and
they have been represented in UPSID on this basis. The languages with
contrastive pharyngealization may help to throw some light on the origin of
vowel harmony. Pharyngealization is not a harmonizing property in these
languages, but the nature of the vowel shifts observed does suggest that
pharyngeal width differences may be causally involved in creation of the
quality differences in vowel harmony, rather than merely co-occurring with
them. It has been suggested that the origin of vowel harmony lies in
consonant voicing differences, and voiced consonants (at least the
obstruents among them) typically are produced with a wider pharynx (see,
for example, Perkell 1969).

8.6 Diphthongs
Relatively few languages are considered to have phonologically unitary
diphthongs under the criteria used in UPSID (for further details on these
see Chapter 10). Obviously a much larger number of languages permit
sequences of juxtaposed vocalic segments which might be considered
phonetically to be diphthongs, or have diphthongal sounds which arise
allophonically. Because diphthongs are so frequently derived in this way
rather than being underlying segments, UPSID does not provide a good basis
for analysis of the phonetic patterns in diphthongs. Nonetheless, we offer
the following brief comments on the diphthongs that do occur. There are a
total of 83 diphthongal segments recorded in UPSID from 23 different
languages. Fully 22 of these 83 diphthongs are from a single language, !Xu
(918). This language has 4 sets of diphthongs: plain oral, plain nasalized,
pharyngealized oral, and pharyngealized nasalized (cf. the 8 sets of
monophthongal vowels it has). Of the other languages with diphthongs, 2
(Kurdish, 015, and Acoma, 749) have 8 each, one (Dani, 613) has 5, 2
(Hindi-Urdu, 016, and Yagaria, 609) have 4. There are a further 5 languages
with 3 diphthongs, 3 languages with 2 and 8 languages with one diphthong.
The diphthongal segments reported are rather heterogeneous and do not
show much clear patterning. The only ones recorded which are at all common,

133
Vowels

i.e. occur in more than 2 languages, are those shown in Table 8.6. Though
the numbers are small, this table seems to indicate that diphthongs that
begin or end with a high vowel element are preferred over those which lack
such an element. This cannot be explained as being the result of an attempt
to maximize the distinctiveness of the diphthongs, since diphthongs with
short trajectories through the vowel space, such as /ei/, /ie/ and /ou/,
are found among the more common types as readily as those with a large
trajectory through this space, such as /ai/, /au/ and /ui/.

Table 8.6 Common diphthongs

/ei/ 6 (also Burmese, 509, with /el/ but not /ei/)

/ai/ 5 (plus 2 languages with /ae/)
/au/ 5 (plus 2 languages with /ao/)
/ou/ 4 (also Burmese, 509, with /ou/ but not /ou/)
/ui/ 4
/io/ 4 (including Evenki with / io:/)
/ie/ 3
/oi/ 3

8.7 Summary
The principal conclusions reached in this chapter can be summarized as
follows. Front vowels are usually unrounded, back vowels are usually
rounded, low vowels are usually central and central vowels are usually low.
All languages have at least 3 phonemic vowels. Nearly all languages have
/ i, a, u/, but among these the vowel /u/ is more often absent than /i/ or
/a/. The most common number of vowel phonemes in a language is 5, and the
most common number of distinctive vowel qualities in a language is also 5.
Contrastive length is associated with an increase in the number of
distinctive vowel qualities in an inventory. The higher mid vowels / e/ and
/o/ are more likely to appear long without a short counterpart than lower
mid vowels. Among nasalized vowels, /e/ is comparatively rare. No language
has more vowels in a secondary set than it has in its primary set, where
secondary means vowels with nasalization, pharyngealization or an unusual
phonation type, and primary means that set of vowels with normal voicing
and no secondary articulation which has the most members (in some languages
this is a set of long vowels)• Diphthongs show a preference for including a
high vowel element.

134
 Patterns of sounds

Notes
1. Note that the percentages are almost exactly what would obtain if all
languages had the 5 vowel system provided for by the Roman alphabet,
/i, e, a, o, u/. This system has 40% high vowels, 40% mid vowels, 20%
low vowels; 40% front vowels, 40% back vowels, 20% central vowels; 60%
unrounded vowels, 40% rounded vowels.
2. A similar basis for vowel harmony, under the name of "register", has
also been suggested for some Austro-Asiatic languages (Gregerson 1976).

References
Allen, W.S. 1965. On one-vowel systems. Lingua 13: 111-24.
Anderson, S.R. 1978. Syllables, segments and" the Northwest Caucasian
languages. In A, Bell and J.B. Hooper (eds.), Syllables and Segments.
North-Holland, Amsterdam: 47-58.
Crothers, J. 1978. Typology and universals of vowel systems. In J.H.
Greenberg et al. (eds.), Universals of Human Language, Vol. 2,
Phonology. Stanford University Press, Stanford: 93-152.
Foley, J. 1975. Nasalization as a universal phonological process. In C.A.
Ferguson, L.M. Hyman and J.J. Ohala (eds.), Nasalfest. Stanford
University, Stanford: 197-212.
Gregerson, K.J. 1976. Tongue-root and register in Mon-Khmer. In P. Jenner,
L. Thompson and S. Starosta (eds.), Austro-Asiatic Studies, Vol. 1.
University of Hawaii Press, Honolulu.
Griffiths, T.G. 1966. The Italian Language. Faber, London.
Hockett, C.F. 1955. A Manual of Phonology (UAL Memoir 11). Indiana
University, Bloomington.
Jacobson, L.C. 1978. DhoLuo Vowel Harmony (UCLA Working Papers in
Phonetics 43). University of California, Los Angeles.
Lindau, M. 1979. The feature expanded. Journal of Phonetics 7: 163-76.
Perkell, J.S. 1969. Physiology of Speech Production: Results and
Implications o£ a Quantitative Cineradiographic Study. "M.I.T. Press,
Cambridge, Massachusetts.
Sedlak, P. 1969. Typological considerations of vowel quality systems.
Working Papers in Language Universals (Stanford University) 1: 1-40.
Wright, J. 1980. The behavior of nasalized vowels in the perceptual vowel
space. Report of the Phonology Laboratory (University of California,
Berkeley) 5: 127-63.

135
Insights on vowel spacing 1

9•1 Introduction
This chapter presents the results of an analysis of the vowel systems of
the 317 languages in the UCLA Phonological Segment Inventory Database
(UPSID). It shows that deviations from the patterns predicted by a theory
which proposes that vowels are dispersed in the available phonetic space
are relatively infrequent and, for the most part, confined to matters of
small scale, falling into a few definable classes. It will be argued that
in most of these deviations from the predicted patterns there is
nonetheless evidence that vowels tend toward a balanced and wide dispersion
in the available phonetic space.

9.2 Preliminaries
A few basic vowel inventories and a few basic configurations show up time
and again in natural languages, while other, no more complex patterns are
rare or totally absent. The most prevalent patterns seem to be the
so-called "triangular" systems, particularly those of average size, and
notably the 5-vowel systems. For example, over a quarter of the 209
languages in the Stanford Phonology Archive have a triangular 5-vowel
system consisting of /i, e, a, o, u/> while less than 5% have any of the
other 5-vowel configurations; the "square" 4-vowel and 6-vowel systems
combined total less than 10% (Crothers 1978).
Several attempts to explain these patterns invoke a principle of vowel
dispersion, proposed in slightly differing versions by Liljencrants and
Lindblom (1972), Lindblom (1975), Terbeek (1977), and Maddieson (1977). 2
This principle holds that vowels tend to be evenly distributed in the
available phonetic space and also widely distributed, within the

136
 Patterns of sounds

limitations of the particular system. The proposed models for vowel

dispersion predict an optimal arrangement for any given number of vowels in
the system; such theoretical systems may then be compared with the vowel
systems of natural languages.
Just such a comparison is the starting point for the present study. The
vowel systems of 317 languages are examined for symmetry and dispersion. We
take note of those systems in which the vowels are not evenly or widely
distributed in the available space, and seek to determine whether these
vowel configurations can nevertheless be accounted for in a principled way.
There may, for example, be straightforward historical or phonetic
explanations for these "defective" vowel systems. If, however, there remain
a substantial number of vowel systems which seem to obey no apparent rule,
we should perhaps reconsider the notion of dispersion. It may be the case
that vowel spacing is not at all a principled matter, and the success of,
say, the Liljencrants and Lindblom model in predicting the balance of the
3
vowel systems may prove to be merely coincidental.

Note that the existence of at least some defective vowel systems in

natural languages does not automatically rule out a dispersion theory. The
claims made by a dispersion theory may be essentially correct, but
languages could nonetheless undergo processes which produce defective vowel
systems, e.g. vowel mergers, shifts, etc. If this is the case then we
should expect to see evidence of pressure to "correct" the vowel spacing by
compensatory shifts. This understa riding of the interaction of vowel
dispersion and other processes predicts that vowel systems, studied
synchronically, should include those which, although well-spaced, show
compensation or rotation of the vowels. Some of the ways in which defective
vowel systems could assume configurations that are basically consistent
with the dispersion hypothesis as understood here are illustrated below.
Three possibilities are illustrated in Figure 9.1. The first, 9.1 (a),
shows a system with a gap, represented by [], in the high back region, but
with a back mid vowel higher in the phonetic space than the corresponding
front mid vowel. The system appears "skewed" and gives the appearance that
one vowel has been attracted to a higher position to compensate for the
presence of a gap. The second, 9.1 (b), illustrates another possibility, in
which the entire system is rotated with respect to the typical, unmarked
configuration for a vowel system of a particular size (here, a 5-vowel
system), thereby achieving maximal dispersion with a slightly different
orientation. The third, 9.1 (c), illustrates a defective system that is

137
Insights on vowel spacing

complemented with a vowel of unexpected quality in the vicinity of the gap.

This, too, serves to balance the system to a certain degree.
(<O (b) (c)
! [ ] + [ ] u
L
O O [] 0 O
e
e
a a a

Figure 9,1 "Defective" vowel systems

However, not all conceivable defective systems are open to
interpretation as basically consistent with a theory of dispersion in the
phonetic space. For example, there may be skewed systems in which the vowel
adjacent to the gap is farther away from it than would be expected from
comparison with the paired vowel, as in Figure 9.2 (a). Or there may be
systems in which a vowel of unexpected quality is located well away from
the vicinity of the gap, increasing the imbalance even more, e.g. 9.2 (b),
Also, there may be systems such is 9.2 (c) which are not open to
interpretation as making any compensation for the imbalance in the system.
These "stationary" systems simply contain a gap.
(a) (b) (c)
i L] jy u i LJ
e e [] e o
o
a a a
Figure 9.2 Unbalanced vowel systems
These systems appear to be counter-examples to a theory of maximal
dispersion. However, such cases as 9.2 (c) are ambiguous in that the vowels
might well be phonetically underspecified representations of the vowels in
the systems in Figure 9.1 (a) or (c). Whether or not the vowels of a
particular language are represented in sufficient phonetic, detail in UPSID
impends greatly on the phonetic judgments and transcription methods of the
field linguist. Some linguists report the auditory quality of vowels in the
narrowest detail, while others simply rely on the commonest vowel symbols,
often those available on any typewriter, to make all the necessary
4
distinctions. Unfortunately, while a vowel system reported as
/i, e, a, o, u/ may be faithfully representing a perfectly balanced system,
it may also be concealing a wealth of unreported phonetic detail.

138
 Patterns of sounds

The typology of deviations described above establishes that the

question of the essential correctness of vowel dispersion hypotheses can be
addressed by a study of the frequency with which different types of
apparent exceptions to it appear. The next section discusses how a study of
this topic was carried out.

9.3 Method
Data
The UPSID sample of 317 languages was tested for vowel dispersion. The
descriptive sources were carefully examined for any details which might
possibly shed light on the true phonetic quality of the vowels under study,
and the maximum available detail which could be represented in the coding
scheme was retained.
The vowel phonemes in UPSID are represented on a height scale with 5
basic values (high, higher mid, mid, lower raid, low). In this chapter mid
and higher mid vowels are usually transcribed with the same symbols, but
where it is necessary to distinguish between them mid vowels are shown by
symbols in quotes, e.g. /"e"/ vs. /e/. Two additional heights, "lowered
high" and "raised low", are also recognized. The vowels are also
categorized on a backness scale with 3 values (front, central, back), and
on a rounding scale with 2 values (rounded, unrounded). Additional
dimensions pertaining to length, nasalization, phonation characteristics
(laryngealization, breathiness), and other features (r-coloration, lip
compression) were recorded in the archive, but for purposes of eld city .and
simplicity these distinctions are not discussed in the present chapter.
The vowels will be discussed in terms of a distinction between
"peripheral" and "interior" vowels. (Note that this distinction is not the
same as that represented by the variable "peripheral" in the UPSID
database; see Chapter 10.) The "peripheral" vowels are the front unrounded,
back rounded, and low vowels, all of which lie along the margins of the
.available phonetic space. It should be noted, however, that the high
central vowels, although they occupy one of the margins of the phonetic
space, do not fall within the peripheral category; this more restrictive
definition of peripherality is justified on phonological grounds, as the
high central vowels tend not to pattern with the true peripheral vowels in
natural languages, and they are also less common than other peripheral
vowels. Thus, high central /+/ and /«/, along with the remaining
phonetically centralized vowels, constitute the set of "interior" vowels.

139
Insights on vowel spacing

Identifying "defective" vowel systems

A basic characteristic of all maximally dispersed vowel systems is that
there are no unbalanced gaps in the primary (peripheral) vowel system. A
language with a gap is defined as one which fails to utilize a particular
region of the vowel space, while fitting one or more vowels into each of
the remaining regions. We tested the 317 languages to identify those which
contained gaps in the peripheral vowel system.
The test examined whether the 5 major regions along the periphery of
the vowel space, high front, high back, mid front, mid back, and low
central, were filled with at least one vowel. A high or mid region may,
however, be left empty without being considered a gap so long as no other
peripheral vowel in the system has a similar value on the height scale.
This qualification ensures that balanced 3-vowel systems /i, a, u/ or
/e, a, o/ will not be classed as defective. Put more formally, this test
requires that any vowel which is [ct high, p back] be matched by at least
one [a high, - (3 back] vowel in the high and mid regions of the vowel
space. There must also be at least one [+ low] vowel in the system.
It should be emphasized that this is a very weak test of dispersion,
designed simply to find out whether the framework of the vowel system, that
is, the major peripheral subdivisions of the vowel space, fulfills the
requirements of wide and even distribution in the available space. There
are many other possible violations of the dispersion theory, unevenly
distributed interior vowels, multiple vowels in a single major subdivision
of the vowel space, and the like, which are not detected by this particular
test procedure. Future investigations will have to address these more
subtle violations. For the present, however, our test will show whether or
not the basic requirements of vowel dispersion are met in the languages of
our sample.
The various formulations of the dispersion model differ in predicting a
more or less wide spacing of the vowels in the available phonetic space.
Although we v/lll not be able to resolve the question of whether maximal or
merely adequate dispersion is the correct formulation of this principle,
our results may be suggestive. Vowel systems which lack one or more of the
"point vowels" ,/i/, /a/ (or / a / ) , and /u/, that is, those vowels with the
nost extreme values for height and backness, are not exploiting the vowel
space to the maximum. Therefore, such systems ace perhaps better explained
by a theory of adequate, rather than maximal, dispersion. We should not, in
any event, allow our test procedure to inpose expectations of maximal

140
 Patterns of sounds

dispersion by classifying such systems as defective. This calls for an

exception: namely, those vowel systems which lack all high vowels or all
low vowels, but are otherwise balanced, should not be classed as defective
systems. This exception only affects a small number of languages. Only 2 of
the 317 languages were found to lack low vowels. The Cheremis (051) system
is centered rather high in the vowel space, although it does count among
its 8 non-low vowels the lower mid interior vowel /A/. Tagalog (414), on
the other hand, has a 3-vowel system / i , a, o/ that is somewhat compressed:
it descends no lower than the mid vowel /a/, and its remaining two points
fall somewhat short of maximally high /i/ and /u/. 5 This compression
suggests that the Tagalog vowel system is, indeed, only adequately
dispersed in the available space. Three other languages, Squamish (733),
Alabama (759), and Amuesha (824), lack high vowels. All of these have a
basic 3-vowel system, /e, a, o/, which is centered rather low in the vowel
space, with two higher mid vowels, but no high or lowered high vowels.
Squamish also has /e/ in the center of the vowel space. For these
languages, and for Cheremis, which are compressed along one edge of the
vowel space only, acoustic measurements are needed to determine whether
near-maximal, or only adequate, dispersion is in effect.

^•^ Analysis of defective systems

Only 43 languages (13.6% of the sample) were found to have vowel systems
with at least one major gap. These will be discussed below under various
headings.

Four-vowel systems
The test procedure does not classify vowel systems in the same way as the
dispersion models. One particular configuration of vowels which is
classified as "defective" by our criteria is, in fact, fully in accordance
with both the Liljencrants and Lindblom model and Lindblom's later
refinement of it. This is the 4-vowel system /i, e, a, u/. In this case the
basic /i, a, u/ system has been expanded by a single vowel, and the one
first chosen is the front vowel /e/. (The corresponding back vowel hi is
not predicted to appear until the inventory has reached at least 5 vowels.)
By our procedure all 4-vowel systems with one mid vowel are regarded as
defective. This classification seems justified since these systems are in
any case rare. There are 6 such cases. Of these, Shasta (746), Paez (804),
and Moxo (827) have the inventory predicted by Liljencrants and Lindblom,

141
Insights on vowel spacing

/i, e, a, u /. Bardi (357) has a 4-vowel system in which the single mid
vowel is the back rounded vowel /o/, contrary to that prediction. Two
other languages resemble the predicted system, but are rather more
symmetrical in the distribution of their 4 vowels. Wichita (755) has a back
vowel /v/ in place of /a/, yielding a system with 2 back and 2 front
vowels. Cayapa (803), a Paezan language quite closely related to Paez, has
/i, e, a, o/. Rather than occupying their expected positions in the vowel
space, both the back vowels are closer to the gap in the lower mid region
where /o/ might be expected. In this way the vowels are rather equally
spaced. Certain other 4-vowel languages are analyzed as having two gaps in
their peripheral vowel system. These are discussed in a later section.

Frequency of missing vowels

Crothers (1978: 106), reporting on the Stanford Archive languages, notes
that the missing vowels are "generally /e/, /u/, or /o/, never /i/". And
though he formulates the near universal that "all languages have /i, a, u/"
(p. 115), the counterexamples and borderline cases that he reports all have
to do with deviations from an expected high back rounded vowel. This, he
notes, reflects the /a/ > /i/ > /u/ hierarchy observed by both Greenberg
(1966) and Jakobson (1941). (> can be read as "is presupposed by".)
Our own frequency count of the missing vowels in UPSID confirms that
the high front and the low central vowels are less likely to be missing
than the high back rounded vowel, i.e. /a/ and /i/ > /u/. It further shows
that the high back vowel is more likely to be absent in natural languages
than either the front or the back mid vowels, i.e. /e/ and /o/ > /u/.
Almost half of the 35 languages which lack a single vowel lack /u/; 9
others lack /e/, 7 lack /o/, and 2 lack /a/. The implied ranking is
therefore: ,. ^ ,Q
(a) > lo! >
rather than the /a/ > /i/ > /u/ > lei > Iol which is generally assumed.
This fact does not seem to have been commented on in the literature, and it
may have implications that bear on such ideas as markedness and the choice
between maximal and adequate dispersion.
In addition to the languages which lack a single vowel, 6 lack more
than one vowel. The most common pattern (5 languages) involves a missing
high back vowel and mid front vowel, creating a double gap of positive
slope as shown in Figure 9.3 (a). None of the languages exhibits a gap of
negative slope due to the lack of a high front vowel and a mid back vowel,

142
 Patterns of sounds

as in Figure 9,3 (b). Vertical gaps are also rare: only one language lacks
back high and mid vowels, as in Figure 9.3 (c). (Recall that languages
which lack be front and back high vowels or both front and back mid
vowels are exempted from the "defective" category.)
(a) (b) (c)
i [] [] u i []
[] o e [] e []
a a a
Figure 9.3 Languages with 2 major gaps
The defective languages demonstrate that vowel systems occasionally ^p_
avoid certain regions of the space. These systems will be discussed below
in terms of the typology discussed in Section 9.2 above.

Stationary systems
Leaving aside the matter of ambiguity of transcription, our investigation
reveals that 9 or perhaps 10 languages fall under the category of
"stationary" systems, that is, systems which have a gap that they do not
appear to compensate for in any way. All of these happen to be 3 or 4-vowel
systems, although larger systems can be stationary just as well. In these
systems all the vowels are peripheral vowels of the most common types, and
the systems are otherwise balanced, with no evidence of skewing from the
front to the back. Clear examples of such stationary systems are Klamath
(707), which has the vowel system /i, "e", a, "o"/, and Bardi (357), Shasta
(746), Paez (804) and Moxo (827), which, as mentioned above, lack a mid
vowel. The Campa (825) system is similar to Klamath, except that the mid
vowels are reported as being "mid close", i.e. higher mid. Tacana (812) is
recorded in UPSID as having the same system as Klamath (although the back
vowel is recorded in one source as /u/ rather than /of). In Hupa (705), the
basic system is a not-fully-peripheral /e, o, a/. However, /t/ occurs in
the language as a surface segment. Depending on the status accorded to this
anomalous segment, the language may or may not be viewed as having a gap in
the high back region of the vowel space. Mura (802) has a 3-vowel system
/i, a, o/. This could be classed as doubly defective by our test, with /i/
implying a missing /u/ and /o/ implying a missing /e/. However, there is no
evidence in the source to indicate whether the transcription reflects
phonetic reality or orthographic convention. The system could be a rotated
system such as /i >, o, de/ , which is maximally dispersed in the vowel space.
Seneca (754) might well be added to this group of languages. It has an /u/

143
Insights on vowel spacing

vowel which occurs only rarely; if this is considered as a gap, there is no

evidence that it is being compensated for in the vowel system. This
language is discussed further below.
As we have mentioned, genuine stationary systems constitute a
counterexample to the theory of maximal dispersion. However, we cannot
dismiss the possibility that certain important phonetic details have been
obscured by a broad phonetic transcription in the examples above.
Therefore, a theory of dispersion cannot be disproven with this set of
potential counterexamples. Definitive evidence against such a theory will
have to be sought in the form of systems with unevenly spaced vowels
(systems that have been rotated into asymmetrical patterns), rather than in
systems which appear not to have compensated for the gap at all.

Complementary vowels
Some of the languages in our inventory have defective systems that are
complemented by a single vowel of unexpected phonetic quality which shares
some of the features of the missing vowel. Vowel systems of this sort can
be classed into 3 major types, namely, those in which the complementary
vowel is (i) a central vowel (9 languages), (ii) a front rounded or back
unrounded vowel (13 languages), or (iii) a peripheral vowel similar to the
missing vowel but lacking a counterpart of equal height and opposite
rounding elsewhere in the vowel system (6 languages). Stated more formally,
if the missing vowel is [a high, 3 back, y low, 6 round], then the
complementary vowel is:
(i) (ii) (iii)
[ a high ] , [ a high ] or [-a high or - low]
[+ central] [ 3 back ] [ 3 back ]
[-3 round]
We will discuss each of these types in turn.

Vowels of unexpected backness ([a high, + central])

Two principal patterns of this type are found: /a/ for an expected missing
mid or low vowel (e.g. in Tagalog, 414, Changchow, 503, and Acoma, 749), or
/t/ for an expected missing high vowel (e.g. for missing /u/ in Abipon,
815). There are several other languages with a missing mid or low vowel.
Margi (268) has /e/ but mostly in loanwords; /e/ and /o/ are the only truly
native mid vowels. Bashkir (063) has no /Me"/, but the system is
complemented in several alternate ways. In addition to /a/ there is / 0 / ,
and there is also evidence of compensation in the peripheral system. In

144
 Patterns of sounds

Cheremis (051), /a/ offsets a missing low vowel, but there is also evidence
of compensation. Among languages with a missing high vowel Cofan (836) is
very similar to Abipon, but with evidence of compensation as well. These 3
languages are discussed further below. In Chacobo (811), in addition to a
missing high back vowel, a front mid vowel is also lacking. There is no
central vowel to offset the latter. This double gap may perhaps be better
explained as a rotation of the entire system.

Vowels of unexpected lip position ([ a high, 3 back, -& round])

Systems of this type fit into the "defective" category somewhat more
marginally than the previous type. On a formant chart with Fl and F2 as the
axes, the front rounded vowels and the back unrounded vowels are more
centralized than their front unrounded and back rounded counterparts, but
less so than the true central vowels /a/ and /t/. Crothers nevertheless
classes these centralized vowels together with the true central vowels, and
does not regard any such "interior" vowels as fulfilling the requirements
for maximal dispersion of the system. Liljencrants and Lindblom take no
stand on the status of the front rounded and back unrounded vowels; their
model is not designed to generate this particular set of vowels.
Thirteen defective systems in our sample have a vowel with the same
height and backness as the missing peripheral vowel, but with the opposite
rounding. Only in Bashkir (063) and Khalaj (064) are the complementary
vowels (/0/ in each case) embedded in a series (/y, 0, x/ for Bashkir; /y,
0/ for Khalaj). In the remaining 11 cases the vowel complementing the
peripheral system is the only front rounded or back unrounded vowel in the
language. This fact suggests that the front rounded and back unrounded
vowels do not show up casually in the vowel system. Moreover, in examples
such as Gilyak (909), with /#/ for a missing /e/, and Island Carib (823),
with /*/ for a missing /o/, the complementary role of the mid interior
vowel is underscored by the lack of a high interior vowel such as /y/ or
/tu/ in the system. In most languages with front rounded or back unrounded
vowels the system "builds down" from /y/ to /</>/ to /*/, or from /w/ to /*/
to / A / , such that the lower vowel implies the presence of the higher. The
isolated /0/ and /*/ of Gilyak and Island Carib are therefore quite
unusual. Their isolation suggests that they do not represent an incipient
interior vowel system, but instead are likely to be closely associated with
the gap in the mid peripheral region.

145
Insights on vowel spacing

In 10 languages all gaps are complemented by a single front rounded

vowel or back unrounded vowel. There are 3 languages with /w/ for a missing
high back rounded vowel: Japanese (071), Nunggubuyu (353), and Alawa (354).
Nimboran (604) has /w/ and /*/ for missing high and mid back rounded
vowels. The complementation in Island Carib and Gilyak has been mentioned
in the previous paragraph. In Bashkir and Khalaj, mentioned above, the
complementary vowels form part of a front rounded system; both have /0/ for
a missing mid front unrounded vowel. In Jaqaru (820) the native vowel
system is /i, a, w/, though an additional vowel /e/ appears in a limited
number of loanwords (this vowel is excluded from the UPSID inventory). In
Ocaina (805), in addition to complementation by /w/, there is evidence of
compensation in the peripheral vowel system (see below).
In another 3 languages, Adzera (419), Nez Perce (706) and Amahuaca
(810), only one of two gaps in the peripheral vowel system is complemented.
All of these examples involve a double gap of positive slope (missing
/e, u/) rather than the other possibility, a gap of negative slope (missing
/i, o / ) . In each case, the gap where /u/ is expected is complemented by
/LU/, while the gap where /e/ is expected remains uncomplemented. It is
rather surprising that the complementary vowel in these double-gap systems
is always the high back vowel /LU/ rather than the front mid vowel /0/. Such
preferential treatment of the high back region of the vowel space is quite
unexpected in view of the fact that defective vowel systems are far more
likely to be lacking a high back vowel than a front mid vowel. We might
well expect a greater number of /w/'s overall, due to the frequent absence
of /u/; what is surprising is the complementation of /u/ instead of the
complementation of /e/ in systems which lack both. Still, in Amahuaca and
Nez Perce the gap where /e/ is expected is somewhat offset by compensation
in the high front region. In addition, Nez Perce has a peripheral vowel of
unexpected height in the low front region. Adzera, however, shows no
evidence of such compensation; there is simply a gap in the mid front
region.
It is somewhat surprising to note the predominance of /HI/ over other
complementary interior vowels (/y, 0, e, *, A / ) . While this fact obviously
relates to the frequent absence of /u/, the latter might just as well go
uncompensated, or else be compensated by some other vowel or by rotation of
the system as a whole. Moreover, while we might expect the presence of /in/
to presuppose the presence of /u/, just as, in general, /y/ presupposes
/i/, this is not true in the relatively large number of vowel systems in

146
 Patterns of sounds

which /w/ stands alone. (No defective vowel system has a complementary /y/
in place of missing /i/.)
An investigation of the acoustic, perceptual, and auditory quality of
the vowel /w/ suggests that this vowel is in fact quite centralized,
verging on the quality of central /t/. For example, Ward's (1938) auditory
analysis of Bamum /in/ places this vowel well away from cardinal /u/,
half-way between the central and back regions of the phonetic chart.
Hombert (personal communication) has conducted a perceptual test on
speakers of 3 Bamileke languages, Banjoun, Fe?Fe?, and Bangangte in which
synthetic vowels were presented to the subjects, and the formant
frequencies which corresponded to "acceptable" vowels in their languages
were noted. For the /w/ in these languages the mean acceptable Fl was 260
Hz., and F2 was 1391 Hz.; this corresponds to a high and almost central
vowel. Papcun (1976) plotted the vowels of Spanish and Japanese in a
normalized acoustic space. Spanish has a rounded /u/ and Japanese an
m
unrounded l f• He found that the difference between these high back vowels
lies mainly in F2, which is approximately 725 Hz. for the Spanish vowel but
1275 Hz. for the Japanese vowel, corresponding to a considerably
centralized location in the acoustic space. The vowel /iu/ thus appears to
be similar to the vowel /+/. If we combine the 9 vowel systems complemented
by /w/ with the 3 complemented by /+/ into a single category, we find that
these 12 cases constitute the majority of cases of complementation in our
sample; complementation by any of the remaining interior vowels /y, 0, *,
©/ occurs in only 10 languages. The high interior region seems to be a
favored area for complementary vowels. The significance of this fact is not
readily apparent, and merits further perceptual and acoustic study. It may
be the case that the corner of the vowel space in which the high back vowel
/u/ is to be found is simply smaller, in perceptual terms, than we might
otherwise expect, and hence /+/, /a/ and /LU/ are closer to each other
perceptually than the acoustic facts suggest.

Vowels of unexpected height ([-a high or -y low; 3 back])

This category is reserved for those vowel systems with two peripheral
vowels instead of one in a region vertically adjacent to the gap in the
system. Here the complementary vowel is distinguished from other peripheral
vowels by its unexpected quality. In the case of non-low vowels we shall
define the complementary vowel as the one which lacks a counterpart of
equal height and opposite rounding elsewhere in the vowel system (e.g. the

147
Insights on vowel spacing

vowel lol in a system such as /i, e, a, o, o / ) . In the case of low vowels,

the complementary vowel is always the non-central vowel, i.e. the low vowel
that is not /a/. The precise location of the gap in the system may be
inferred from the height of the matching vowel of opposite backness and
rounding. Thus, a missing high back vowel matched by /1/ will be considered
a missing /o/.
While many patterns of complementation are possible, only 3 occur in
our sample. Kunimaipa (620) has a high back /o/ for a missing mid back
vowel. Navaho (702) and Nootka (730) have a higher mid back /o/ for a
missing high back vowel. The Nootka case is discussed further below.
Taishan (501), Nez Perce (706), and Ket (906) have a raised low front /ae/
for a missing mid front vowel. There is no instance of a peripheral vowel
of unexpected quality which is several levels of vowel height away from the
gap it complements. The additional vowels appear in the immediately
adjacent vertical sector of the vowel space in all but one of these
languages. Kunimaipa has lol complementing a gap at lol (it has the front
vowel /el); Taishan, Nez Perce, and Ket all have /ae/ complementing a gap at
/£/ (each of these languages has the back vowel / ° / ) ; Nootka and Navaho
have /o(:)/ complementing a gap at /o/ or /u/ (cf. Nootka /t/ and Navaho
/ i : / ) . Only in the long vowel set in Navaho is the complementary vowel /o:/
one step further removed from the gap (specifically identified as a missing
/u:/ by our criteria), that is, the complementary vowel is not /o:/. Note
that the short vowel system of Navaho appears as a "stationary" system,
since front /£/ is matched by back /o/.

The vowel systems in which a single vowel of unexpected quality

complements a gap in the peripheral system constitute the major portion (28
of 43, or about 65%) of the defective vowel systems in our sample. We have
seen, for example, systems lacking a high back vowel being complemented by
a high central vowel /+/, a higher mid back vowel lol, or a high back
unrounded vowel /w/. Similarly, systems lacking a mid front vowel were
found to be complemented by /a/ or /ae/ or /0/, rather than by any more
remote segment. The common denominator among almost all 31 of the
complementary vowels in our sample is that they bear a close phonetic
resemblance to the expected vowel. Due to this fact, the systems containing
such complementary vowels are "not excessively deviant" from the predicted,
fully-dispersed vowel system. We may assert that they obey some sort of
dispersion rule although one whose requirements are obviously weaker than
maximal dispersion.

148
 Patterns of sounds

Complementary vowels with additional adjustments

We have seen that a few of the systems with complementary vowels seem to be
striving for a greater degree of dispersion in the vowel space. Not only is
there a complementary vowel near the site of the expected vowel, but the
system is skewed such that one or more of the remaining vowels is found
closer to this gap than we would otherwise expect, based on the height of
the matching vowel(s). For example, the Gilyak (909) system, lacking a mid
front vowel, is complemented by the nonperipheral vowel /$/; moreover, both
the high front /i/ and the low /ae/ are unexpectedly close to the gap. That
is, lowered high /i/ appears rather than /i/ and raised high fronted /ae/
appears rather than /a/. In the Cofan (836) system, which lacks a high back
vowel, the gap is flanked on two sides by a complementary /+/ and by a
higher than expected /o/ (as evidenced by the imbalance in the mid vowels,
lower mid /e/ vs. mid / o / ) . Rather than occupying their expected locations,
these vowels seem to position themselves closer to the gap. Ocaina (805)
lacks a high back rounded vowel, but has complementary /w/ and it also has
/o/ in proximity. Bashkir (063) lacks a mid front unrounded vowel, but has
complementary /0/ and /a/ and it also has a raised /ae/ in proximity.
Again we must consider the possibility of underspecification of the
phonetic vowel quality. Some of the authors undoubtedly have chosen a
broader system of phonetic transcription than others, thus possibly
obscuring some evidence of rotation of the vowel system. We therefore
cannot rule out the possibility that some other cases of vowel system
complementation in our sample are accompanied by some compensation in the
peripheral vowel system.

Compensation in the peripheral vowel system

Most of the defective vowel systems that we have examined thus far have
fallen into two basic categories: ones in which there is either no evidence
of any compensation for the gap in the system (stationary systems), and
ones in which there is a single vowel of unexpected quality in the system,
either a non-peripheral vowel, or a peripheral vowel of unexpected height,
which may be regarded as complementary (complemented systems). We have,
however, noted a few cases of complemented vowel systems whose remaining
peripheral vowels are displaced toward the gap as well; these systems
thereby achieve an even greater degree of compensation. A number of other
languages in our sample lack a complementary vowel but nevertheless show
evidence of such displacement, even to the point of rotating the entire
system.

149
Insights on vowel spacing

Compensations involving a single vowel

Vowel systems of this sort are very similar to vowel systems with a
complementary peripheral vowel of unexpected height. In both cases there is
a vowel in the system which is found closer to the gap than we would
otherwise expect, based on the height of the closest corresponding vowel at
the opposite end of the backness scale. The crucial difference, of course,
is that in the latter systems this vowel shares its region with another
peripheral vowel, while in the former it is alone in its region of the
vowel space. In both Ocaina (805) and Cofan (836), discussed in the
preceding section, there is an /o/ which is higher than the corresponding
front vowel /e/, thus compensating for the missing high back vowel to some
degree. Bashkir (063), which lacks a mid front vowel, also shows evidence
of compensation. In these examples the vowel system is also complemented by
vowels of unexpected lip position. Five other systems lack a complementary
vowel, but otherwise follow the Ocaina pattern. Malagasy (410) lacks a high
back vowel, but has /o/ in proximity to the gap, as evidenced by the
imbalance of /"e"/ and /o/. Mazatec (727) lacks a high back vowel, but has
/o/ in proximity, as evidenced by the imbalance of /e/ and /o/. Cheremis
(051) lacks a low central vowel, but has /e/ in proximity, as evidenced by
the imbalance of /e/ and /o/. Amahuaca (810) lacks a mid front vowel, but
has / i/ in proximity, as evidenced by the imbalance of high vowels /i/ and
/w/. Wichita (755) lacks a mid back vowel but has /*)/ (rather than /a/) in
proximity.

Compensations involving multiple vowels

Vowel systems with displacements at more than one point in the peripheral
system show an even greater tendency toward dispersion in the vowel space.
In the extreme case, most or all of the vowels in the language show a
displacement from the expected values (/i> u/ for high vowels, /e, o / for
mid vowels, and /a/ for low vowels) in the direction of the gap; this may
be considered a rotation of the entire system. Such a systematic
displacement may well result in a maximally dispersed vowel system, which
is oriented along a slightly different axis from that of most other vowel
systems.
In Gilyak (909), discussed above, two of the peripheral vowels, / i/ and
/ ae/, show evidence of displacement toward a gap in the front mid region of
the system. Other cases of multiple displacement are rather uncommon,
although in Cayapa (803) the vowel system as a whole appears to be

150
 Patterns of sounds

compensating for a gap in the mid back region. The high back vowel /©/ is
lower than its front counterpart /i/, and the low vowel /a/ is further back
than the expected /a/. The -quality of the mid front vowel is reported as
/e/, and though there is no mid back vowel against which to test its
height, we may perhaps infer (from the use of the symbol /e/, rather than
the unmarked /e/) that it is lower than expected, backing up the
displacement of the low vowel. From what appears to be an anchor point at
/i/, the vowels of Cayapa are displaced from their expected values
according to the pattern in Figure 9.4.

Figure 9.4 Cayapa vowels

The resulting dispersion is very nearly maximal, in spite of the absence of
one phonological category.
When a gap occurs in the peripheral vowel system, some or all of the
remaining peripheral vowels are often found to be displaced in the
direction of the gap, regardless of whether the gap is otherwise offset by
a vowel of unexpected quality elsewhere in the system. The displacement may
be interpreted as a means of establishing a wider and more even dispersion
of the peripheral vowels, in spite of the gap. Ten of the languages in our
sample (23% of the defective systems) show evidence of some such
compensation in the peripheral vowel system. However, only 2 of these
languages, or possibly 3, extend the compensation beyond a single vowel.
This is largely due to the fact, noted above, that most of the defective
vowel systems in our sample are lacking a high back vowel; in these systems
only one peripheral vowel, the mid back rounded vowel, verges on the gap.
Now, we would naturally expect that vowel(s) adjacent to any gap would show
a greater shift in vowel quality than the vowels further away, which, if
anything, need only make subtle adjustments to those adjustments which have
preceded them. For most gaps, then, we can expect to find evidence of a few
displacements. However, in the case of systems lacking a high back vowel,
it is far less likely that the peripheral vowel on the far side of the gap
(the high front vowel) would be affected from so great a distance across
the upper boundary of the vowel space. Therefore, displacement of a single
vowel might provide an adequate degree of dispersion for these languages.

151
Insights on vowel spacing

Unevenly spaced vowel systems

The languages discussed thus far do not exhaust the list of defective vowel
systems in our sample. Vowels in the remaining systems appear to be
distributed unevenly in the phonetic space, contrary to the predictions of
any dispersion theory. As we have stated, vowel systems of this sort are
the only ones which can validly be used to disprove a dispersion theory,
for stationary systems, although they run counter to the predictions of
such a theory, are indistinguishable from dispersed systems that merely
lack adequately reported phonetic detail.
Two languages, Hopi (738) and Auca (818), constitute the clearest
examples of unevenly spaced systems in our sample. They share a number of
characteristics, notably a gap in the high back region and an apparent
displacement of the remaining vowels away from this gap. A similar
phenomenon may be found in Nootka (730), but the phonetic interpretation of
the data is subject to question. Seneca (754) is another language that
might be considered in this class.
Voegelin (1956: 24) commented on Hopi: "So far as vowel placements are
concerned, Hopi is extraordinarily asymmetrical." The analysis of Hopi in
UPSID assumes the vowel inventory /i, 0, ae, a, */ plus the diphthong /ou/.
The diphthong /ou/ could be regarded as a peripheral back vowel displaced
toward the gap at / u/. Similarly, /ae/, a front vowel of unexpected height
adjacent to the gap in the mid front region, verges close upon the low
central vowel /a/, with no accommodation on the part of the latter. There
are, however, indications that the peripheral system is actually more
dispersed than this analysis would lead us to believe. Although Voegelin
classifies it as central, the vowel /a/ of the Mesa dialect of Hopi is,
according to Whorf, "as in calm". This suggests a back, or at least a
back-central quality, which is as expected if an adequate separation is to
be maintained between the two low vowels, and, simultaneously, if the
separation between the backmost low vowel and the mid back vowel is not to
become too great. An alternative analysis would view all of these
displacements together as consequences of a compensation for the missing
high back vowel, in other words, a near-complete rotation of the peripheral
system. According to such an analysis, the peripheral vowels of the
defective Hopi system are systematically displaced from the expected /i, e,
a> o/ to the existing /i, ae, a, ou/ as the presence of the gap (at /u/)
draws the vowels successively toward the gap in a counterclockwise pattern.

152
 Patterns of sounds

If the displacements described above are indeed accurate, the net

result is that the Hopi system achieves a considerable degree of dispersion
with a nearly "square" peripheral vowel system plus two interior mid vowels
(/0, */)• Such an "hourglass" pattern may in a sense be more fully
dispersed than the system with 6 peripheral vowels predicted by the
dispersion models.
The Auca system is described as /j, e, ae, a, o/» The lei is said to
range from [i] to [e], and the /o/ from [o] to [u] and inward to [w]. No
further phonetic detail is provided for /i/, /ae/ or /a/. The unexpectedly
high range of the mid vowel /o/ is suggestive of compensation for a missing
back vowel, but the juxtaposition of vowels /i, e, ae, a/, well away from
the gap at /u/, is surprising and unique among the systems in our sample.
As for Nootka (Tseshaht), whether or not this system is actually defective
is subject to some dispute. The source consulted on this language, Sapir
and Swadesh (1939), suggests that Nootka has a gap in the high back region
of the vowel space. One interpretation of their descriptions is that the
language has a low vowel /a./ that is displaced toward, rather than away
from, the unexpected vowel /°/. A relatively high position for /o/ is
indicated by the authors' statement that /o/ and /o:/ "have the tongue
position of the vowel of coat and the lip position of hoot" (p. 13.). This
gives a Nootka system consisting of /i, e, a, o, o/. However, the source
also compares the phonetic quality of the Nootka vowel /o/ to that in put,
and long /o:/ to that in food, just as /i/ is compared with pit and /i :/
with feed. On the basis of these phonetic descriptions, it may be that the
Nootka system is not defective at all, since they suggest the system
contains a high back vowel after all.
If Seneca is considered to have a gap at hi (/u/ is rare), then it is
surprising that the back vowel /"o"/ seems to be less high than the front
vowel /e/. However, this may be a case where, rather than a failure of a
gap to draw a vowel toward it in the back vowel region, there is instead a
pressure to raise /e/ in the front. This is because there is also the vowel
/ae/ in the front vowel set, making a vowel inventory which consists of
/i, e, s, a, "o", (u)/.

9.5 Conclusions
The great majority of vowel systems in our sample assume configurations
which are predictable from a theory of vowel dispersion, considered in the
light of some basic facts about the overall number of vowels, their degree

153
Insights on vowel spacing

of peripherality, and the like. At first glance there appear to be 43

languages which are exceptions to the notion of vowel dispersion, in that
one or more of the 5 major regions of the vowel space remain unfilled.
However, some measure of dispersion is in force even in many of these
"defective" systems. As a result, these systems maintain a degree of
balance in spite of the obvious gap.
The defective systems may be classed into 3 major categories. Nine of
these 43 languages simply tolerate the gap in the system, showing no
evidence of any compensatory shifts (although such shifts may in fact be
present and merely overlooked in the broad phonological accounts). Some 2
to 4 of the languages with major gaps show a displacement of the remaining
vowels away from the gap, resulting in an even more uneven distribution of
the vowels. The great majority of the defective systems, at least 30
languages, tend toward a balanced distribution of vowels in the available
space, either by complementation with a vowel of unexpected quality or by a
displacement toward the gap of some or all vowels in the system.
Thus the number of obvious exceptions to a vowel dispersion hypothesis
in the whole of the UPSID data is extremely small. About 86% of the
languages have vowel systems that are built on a basic framework of evenly
dispersed peripheral vowels. About another 10% approach this specification.
This strongly indicates that a vowel dispersion theory correctly captures a
principle governing the distribution of vowels in natural languages.

Notes
1. This chapter was contributed by Sandra Ferrari Disner.
2. The formulations differ somewhat in the degree of dispersion they
propose, but no attempt will be made in this paper to choose between
them. Except for the absence of vowels at the extreme corners of the
vowel space, the data is unsuited for this task. For the most part, we
can only look at areas within the phonetic vowel space and label the
general arrangements according to which areas are filled. In order to
investigate whether specific points in the space are filled, we would
need acoustic measurements drawn from a large number of speakers of
each language (see Disner 1980).
3. For example, the apparent success of such a model might be attributable
to heavy emphasis on a few language areas in the sample utilized by
Liljencrants and Lindblom, or to a bias in the sources in favor of
reporting apparently balanced vowel systems when adequate phonetic
detail is lacking.
4. Considerable energy was expended to seek out whatever phonetic detail
was available in the sources, regardless of the authors'
transcriptions. Cross-references were checked, footnotes examined, and
a good deal of reading between the lines was done in hopes of adding
some detail to the reported phonetic quality. Skewed vowel systems
provide us with particularly valuable information on the phonetic
quality of the vowels in question. A linguist who is little concerned

154
 Patterns of sounds

with phonetic detail, or who does all his transcriptions on the

typewriter, may well report pairs of vowels such as [e, o] simply as
/e, o/. However, it is extremely unlikely that a pair of vowels
reported as skewed (e.g. /e, o/) is actually balanced (/e, o / ) . Here
there is little doubt that the front vowel is higher than the back,
i.e. that the front vowel is closer to the /i/ region than the back
vowel is to the /u/ region. We can therefore draw firmer conclusions
about the degree of vowel dispersion in the phonetic space from these
skewed systems. We should, on the other hand, refrain from drawing
too-firm conclusions from balanced pairs of vowels.
5. The Stanford archive notes that /e/ and /of occur in a great many
loanwords in the dialect of educated Manila speakers. These have been
excluded from the UPSID inventory, since we have chosen to represent a
more conservative dialect of the language.
6. It is not clear to what extent we can consider systems with such vowels
to be defective. Certain configurations of peripheral and nonperipheral
vowels may, in fact, be more dispersed in the vowel space than the
corresponding peripheral systems.

References
Crothers, J. 1978. Typology and universals of vowel systems. In J.H.
Greenberg et al. (eds.), Universals of Human Language, Vol. 1_: Theory
and Methodology. Stanford University Press, Stanford: 93-152.
Disner, S.F. 1980. Evaluation of vowel normalization procedures. Journal of
the Acoustical Society of America 67: 253-61.
Greenberg, J.H. f966. Language Universals, with Special Reference to
Feature Hierarchies. Mouton, The Hague.
Jakobson, R. 1941. Kindersprache, Aphasie, und allgemeine Lautgesetze.
Reprinted in Selected Writings I_. Mouton, The Hague: 328-401.
Liljencrants, J. and Lindblom, B. 1972. Numerical similation of vowel
quality systems: the role of perceptual contrast. Language 48: 839-62.
Lindblom, B. 1975. Experiments in sound structure. Paper reacf at the Eighth
International Congress of Phonetic Sciences, Leeds.
Maddieson, I. 1977. Tone loans: a question concerning tone spacing and a
method of answering it. UCLA Working Papers in Phonetics 36: 49-83.
Papcun, G. 1976. How may vowel systems differ? UCLA Working Papers in
Phonetics 31; 38-46.
Sapir, E. and Swadesh, M. 1939. Nootka Texts. Linguistic Society of
America, Philadelphia.
Terbeek, D. 1977. Some constraints on the principle of maximum perceptual
contrast between vowels. Proceedings of the Thirteenth Regional Meeting
2L t n e Chicago Linguistic Society: 640-50.
Voegelin, C.F. 1956. Phonemicizing for dialect study, with reference to
Hopi. Language 32: 116-35.
Ward, I.C. 1938. The phonetic structure of Bamum. Bulletin of the School of
Oriental and African Studies 9: 423-38.

155
10

The design of the UCLA Phonological Segment Inventory Database (UPSID)

10,1 Introduction
As remarked in the preface to this volume, the discovery of generalizations
concerning the content and structure of phonological inventories has been a
significant objective of recent work in linguistics. UPSID is designed as a
tool for research into these questions, providing uniform data from a
properly balanced sample of an adequate number of languages for
statistically reliable conclusions to be reached. It differs in some major
ways from the Stanford Phonology Archive (SPA) compiled at Stanford
University, although this archive served as a model for our work.
Stanford's plan was more ambitious than the plan for UPSID, but Stanford's
team found that they had to limit the size of their language sample as
their work progressed. The final report includes information on the
phonologies of 196 languages. A principal reason for exclusion was the
scarcity of adequately detailed phonological descriptions. The variability
in detail of the sources which were used also necessarily produced entries
which vary in their completeness—from those which cover little more than a
list of phonemes to those which are able to include a lot of allophonic
details and information on phonological alternations. Thus, for retrieval
of certain information, the true sample size is smaller than 196 languages.
With each reduction, the likelihood that the sample is no longer
representative and properly balanced increases. With Stanford's experience
in mind, UPSID was designed to be narrower in the scope of information
about each language entered, but to be more comprehensive in the number of
languages covered. Users of SPA have also commented that there is a certain
inflexibility inherent in the format chosen for data entry. This is
basically a text-oriented system (for some description, see Vihman 1974).

156
 Patterns of sounds

For example, each segment is entered as an alphabetic character (p, b, m,

etc.) or string (1-retroflex, epsilon, o-open-long-nasalized, etc),
followed by a rather open-ended set of features attributed to the segment
(e.g. obstruent, bilabial, voiceless, etc.), plus comments from the source
and so on. With this format, retrieval of essentially numerical data is not
the most readily achieved operation, and manipulation of the data can be
complex and cumbersome. Stanford's aim was to maximize "accountability"
i.e. limiting their information to that available in the source grammar.
This sometimes results in essentially the same segment receiving different
descriptions in different languages. In designing UPSID, we have aimed to
maximize the ease and flexibility with which numerical data can be
manipulated, and have decided that the compilers may on occasion need to
adopt a more active role in interpreting the source to guarantee a
consistent treatment of similar sounds. We also differ somewhat from SPA on
the principles that should govern selection of languages, e.g. in rejecting
the number of speakers as a criterion for inclusion (or exclusion) of a
language. Nonetheless we have profited greatly from the example of SPA, and
from the excellent hard work that went into its compilation, by using the
final report as a secondary source of data. The following sections describe
the plan of UPSID.

10.2 Selection of_ languages for UPSID

The ideal sample for purposes of statistical evaluation is a random sample,
drawn from the total population under study. In the case of language data,
the "population" is all the world's extant languages. It is impossible to
draw a random sample from this population for two reasons. First, there are
areas of the world about whose languages we have no data or wholly
inadequate data. Second, a "language" is not a clearly demarcated object.
The common criterion of "mutual comprehensibility" used to define
linguistic similarity yields a gradient, and besides, is often not
reflexive in its characteristics (cf. Ladefoged et al. 1972: 65-8). Thus,
unlike a population of, say, registered voters, it is not possible to
enumerate and individualize the members of the population. Hence, no basis
exists for drawing a random sample.
A proper sample of languages must therefore be constructed by some
other sampling procedure. The chosen one is a variety of quota sample (in
fact, the usual linguistic sampling procedure is to draw a quota sample
based on typological, genetic or areal groupings). The principle on which

157
Design of UPSID

UPSID is based is to select one and only one language from each moderately
distant genetic grouping, so that the selected languages represent in
proper proportion the internal genetic diversity of various groupings. The
obvious difficulties in the way of such a scheme are the lack of sound
genetic classifications in certain areas (e.g. South America, New Guinea),
the difficulty of comparing genetic distances in different language
families, and lack of requisite data from some known groups. The advantages
of this procedure are that it precludes, in principle, selection of data
which represents arguably the same language in several varieties (unlike
SPA which includes, for example, both Moroccan and Egyptian dialects of
Arabic, and Maltese); also it directs a principled search for the data to
fulfill the quota design and avoids undue reliance on descriptions that
happen to be at hand (the "bibliographic convenience" factor mentioned by
Bell 1978). A genetic basis for the sample is selected in preference to any
other since it is the only classification which is, in principle, not
arbitrarily determined by the criteria chosen for the classification, but
instead aims to represent real historical relationships. In addition, it is
appropriately independent of the phonological characteristics of which it
is desired to find the frequencies. Note that the phonological uniqueness
of a language is not a basis for inclusion. Because each language included
is relatively distinct genetically from all others in the sample, each
represents the outcome of the opportunity for independent operation of
historical processes. Similarities between languages in the sample are
therefore not due solely to the effect of shared historical origin. As
noted before, the number of speakers is considered a quite inappropriate
basis for including (or excluding) a language. The size of extant
populations of speakers of languages is an accident of political and social
history that is quite irrelevant to questions relating to the structure of
human languages.

No thorough-going attempt to determine a single criterion for the

degree of genetic separation required for inclusion of a language has been
made. The procedure has been to assemble the most comprehensive and
accurate genetic classifications available and to produce, by synthesis of
several classifications where necessary, an overall classification for each
of 11 major groupings of languages, plus several smaller groups.
Intermediate levels of classification were then sampled to select the
languages for inclusion in UPSID. The density of this sampling might be
thought of as representing an intention to include no pair of languages

158
 Patterns of sounds

which had not developed within their own independent speech communities for
at least some 1000-1500 years, but to include one language from within each
group of languages which shared a closer history than that. The 11 major
groupings, plus an "others" category, are shown in Table 10.1, together
with the number of languages included in UPSID from each and the range of
identification numbers assigned to each grouping.

Table 10.1 Language families in UPSID

Language Identification No. of languages

family numbers included
Indo-European 000-049 21
Ural-Altaic 050-099 22
Niger-Kordofanian 100-199 31
Nilo-Saharan 200-249 21
Afro-Asiatic 250-299 21
Austro-Asiatic 300-349 6
Australian 350-399 19
Austro-Tai 400-499 25
Sino-Tibetan 500-599 18
Indo-Pacific 600-699 27
Amerindian 700-899 89
Others (Dravidian,
Caucasian, Khoisan
Eskimo-Aleut, etc. ) 900-999 18

An example might help to clarify the process. The existence of a

Nilo-Saharan language family was proposed by Greenberg (1966). A composite
classification was drawn up using Greenberg (1966, 1971), Tucker and Bryan
(1956, 1966), Bender (1976) and other minor sources. This included 10 major
groupings of languages whose relationship to each other is clearly
relatively remote. Seven of these (Songhai, Saharan, Maban, Fur, Berta,
Kunama and Koman) were judged to have relatively little internal diversity
and hence were represented by one language each. (The Saharan group
consists of 4 languages whose diversity might be sufficient to merit more
than one representative.) An eighth group, Gumuz, is insufficiently known
and no phonological data is available. This leaves two groupings with
considerable internal diversity, Eastern Sudanic and Central Sudanic,
consisting of 11 and 7 sub-groupings respectively. In one subgroup, namely
the Nilotic subgroup of Eastern Sudanic, the degree of internal genetic

159
Design of UPSID

diversity justified inclusion of a language of each of the Western, Eastern

and Southern branches. The target was therefore 13 Eastern Sudanic
languages. Data from 11 were actually obtained, as two groupings were found
to lack adequate data. Only 3 of the targeted 7 Central Sudanic languages
were included; in this case partly because sources known to exist were
unavailable to us. Thus an ideal figure of 28 languages to represent all of
Nilo-Saharan is reflected with some imperfection in a total of 21 actually
sampled, a 75% "response rate".
In certain instances, notably the Southern Ameridian groups of
languages and the Indo-Pacific family, it proved much more difficult to
obtain complete classifications and to resolve conflicting groupings. In
these instances it is not really possible to assess how adequately the
whole family has been sampled. However, an educated guess is that overall
the present sample contains between 70-80% of the languages that it should
include in order to completely fulfill its design specifications. An
alphabetized list and a classified list (with only a skeleton
classification) of the languages included are given in Appendix A. In these
lists, languages included in the SPA are indicated by (S). The sources
examined for the phonological data are also listed in this appendix. In
some cases these have only been consulted indirectly via Stanford's report.

10.3 Determining the inventories

For each language a list of phonologically contrastive segments was
drawn up. This procedure presupposes that such an analysis represents
significant and interesting facts about a language. We do not propose to
take space to argue for this proposition here - merely to point out that in
practice nearly all linguists use such an analysis. Our own view goes
further than some in asserting that the phonological segments can (and
should) be characterized by certain phonetic attributes. Linguists who
believe that phonology is necessarily involved with purely abstract
constructs will part company with us at this point. (However, since such
abstractions cannot be compared, they presumably are not interested in the
kind of language universal properties this archive is designed to
investigate.)
Determining the phonological inventory for each language involves two
principal aspects, determining how many contrastive units there are and
determining what phonetic characteristics should be attributed to each one.
The first aspect concerns defining what is meant by contrast and resolving

160
 Patterns of sounds

questions about the unity or otherwise of "suspect" complex phonetic

events, such as affricates, geminated consonants, prenasalized stops, etc.,
and diphthongs. These are all open to interpretation as unitary segments or
as sequences of some simpler segments in the language.
For our purposes "contrasts" are sound differences capable of
distinguishing lexemes or morphemes in the language involved, given that
data on factors that principally concern phonological properties
themselves, such as stress placement, syllabification, boundaries (word,
morpheme, etc.), and so on, can be used to predict variants but that
diacritical features, arbitrary rule types, morpheme classes, etc., cannot
be so used. These principles are applied to evaluate critically the
information in the source, and consequently the resulting analysis may
differ from the phonological inventory assumed in the source.
The suspect consonantal units/clusters have been examined as carefully
as the available information permits in order to determine their status as
units or sequences. If, for example, they can be split by a morpheme
boundary or form a part of a more general set of permitted clusters,
including for example non-homorganic clusters, they are treated as
sequences. If, on the other hand, there are no similar clusters to those
that would be created by a sequence interpretation, this is taken as
favoring a unit interpretation.
Analysis of phonetic diphthongs is one of the most problematical areas.
A phonetic diphthong might have one of 3 phonological interpretations: (1)
as a phonemic unit, (2) as a vowel and a consonant in sequence (in either
order), or (3) as two vowels in sequence. The first interpretation is
chosen for the Khoisan language !Xu (918), which is analyzed as having a
relatively large number of diphthongs among its phonemes. Their
phonological unity is indicated by the fact that they have the same
distributional patterns as the simple vowels with respect to syllables and
tones. In this language, if [au] was analyzed as /aw/ and [oa] as /wa/, and
so on, it would introduce syllable-final consonants and syllable-initial
clusters to a language that otherwise lacks them. The second interpretation
is chosen for Standard Thai (400). Thai has a set of syllable-final
consonants (nasals) before which both long and short simple vowels can
occur with any of the 5 tone patterns of the language. When the phonetic
diphthongs, which always end with an [i] or [u]-like vocoid, are analyzed
as vowel + approximant consonant they fit in with the pattern established
by these other syllable-final sonorant consonants (Gandour 1975). The third

161
Design of UPSID

interpretation is chosen for a language such as Hawaiian (424), in which

phonetic diphthongs have traditionally been viewed as resulting from
juxtaposition of separate vowels. All vowels can occur in sequence with
each other, sequences with /w/ are distinct from vowel plus /u/, and vowels
in a diphthongal sequence that are contiguous with a like vowel coalesce
(e.g. /ia/ + /a/ — > [ia]). All of these points argue for the analysis of
these diphthongs as consisting of two independent vowels. Overall,
relatively few languages are judged to have unit diphthongs.
In addition to the above considerations, certain types of contrasts
have been interpreted as suprasegmental, i.e. as not relevant to setting up
an inventory of phonological segments. These have included, by definition,
tone and stress phenomena. Nasalization and properties involved in vowel
harmony systems have always been treated as segmental. Length, however, has
been treated as suprasegmental if it applies to a whole class of segments,
such as all vowels. Otherwise it is treated as a segmental property: for
example, Yay (402) has long and short low vowels /a, a:/ but the other
vowels do not contrast in length. So there is considered to be no basis in
this language for regarding length as suprasegmental; it is an attribute of
a particular segment only.
The remaining issue concerning the size of the inventory of segments
has to do with inclusion or exclusion of segments with more or less
marginal status. Certain segments which occur only, say, in interjections
or in foreign words that are clearly not established as loans have been
excluded from the inventory altogether. In other instances, segments which
might from certain points of view be considered marginal are included
without any indication of a distinctive status. This would be the case, for
example, for segments which are restricted to loanwords if the loans appear
to be fully assimilated in the language concerned. Efforts have been made
to reach decisions on straightforward inclusion or exclusion in the great
majority of cases. However, in a minority of instances a third option has
been taken. That is to include the segments but to indicate the nature of
their marginal status by a value of a special variable. The details of this
option are discussed below in section 10.4.
Each segment that is judged to deserve inclusion in the inventory for a
given language is represented by a phonetic specification. In determining
the phonetic properties to be attributed to each segment, an allophone
which could be considered the most representative was identified. Wherever
sufficient information was available, this selection was based on 3

162
 Patterns of sounds

principal considerations: (1) which allophone has the widest distribution

(i.e. appears in the widest range of and/or most frequently occurring
environments); (2) which allophone most appropriately represents the
phonetic range of variation of all allophones; (3) which allophone is the
one from which other allophones can be most simply and naturally derived.
There are cases in which answers to these questions produce conflicting
results; in these cases an attempt was made to resolve the conflict by
considering how badly overall the conditions would be violated if the
answer given by, say, consideration (2) was preferred over that given by
(1), and to select the answer which did least violence to all 3
considerations taken together.
The degree of phonetic detail aimed at in the segment specifications is
one that is approximately equal to that attained by the traditional 3-term
label of phonetics, specifying voicing, place and manner for consonants,
height, frontness and lip-position for vowels, plus additional labels
required for features such as secondary articulations, etc. The set of
coding variables to represent the required phonetic attributes is discussed
in full below. Overall, the effort was to design a set of variables which
was as little prone to errors of interpretation and as little likely to
generate impressions of explicitness when unjustified as could reasonably
be accommodated within practical limits of convenience and economy. The
variable set is designed so that there is a minimum of appeal to redundancy
to interpret their meaning. It was also designed to accommodate some of the
major indeterminacies found in the phonological sources consulted. However,
the database cannot be used without danger of misinterpretation unless the
system of coding variables is fully understood. Moreover, the design of the
variable set forces certain choices to be made when entering data, and at
the same time limits the phonetic distinctions which may be drawn. For
these reasons, it is recommended that the following section be read
carefully by those who are interested in obtaining the clearest
understanding of the results of UPSID represented in the preceding
chapters.

10.4 Indices and variables

Each segment in the actual UPSID database is represented as a separate
record in the form of a (notional) 80-column card image. The first 7
columns are devoted to identifying indices, 10-70 contain variables for
phonetic attributes, 80 contains the ANOMALY feature. Each of the variables

163
Design of UPSID

referring to phonetic attributes takes a value of 1 or 0, with 1 indicating

that the allophone chosen to represent the segment possesses the attribute
named and 0 indicating that it lacks the attribute, (There are a few minor
exceptions to this rule which are explained below.) In the exposition which
follows, each variable is referenced by its (notional) column location,
given a definition and followed by the form of the variable name in the
actual computer database (parenthesized and in capitals). For convenience,
the variables are grouped into classes. Notes and comments on the use of
the variables are interspersed where appropriate.

Indices
1-4. Language identification number (LANGNO). This number serves to
identify the language to which a segment belongs. It consists of 3 digits,
the first and sometimes second of which indicates affiliation to one of the
major groupings used as the genetic basis of the sampling (see Table 10.1).
5-7. Segment identification number (SEGNO). Each segment within a language
is numbered sequentially. The combination of LANGNO and SEGNO thus
identifies one and only one record in the data base.

Alphanumeric segment code

8-9. Segment code (SEGCODE). For convenience in assembling the data base, a
mainly alphabetic code was devised to represent commonly occurring
segments. Variable values were then generated from this code. As these
symbols have a useful mnemonic function, they have been left in the file.

Variables indicating place of articulation

All consonants except /h/ and /fi/ are specified for at least one place of
articulation. Places are listed in the conventional front to back sequence.
Double articulations are indicated by specifying two places of
articulation, but secondary articulations are indicated by use of a
separate set of variables. /h/ and /fi/, having place determined by
environment, do not receive any place specification.
10. Bilabial (BILABIAL).
11. Labiodental (LABODENT).
12. Dental (DENTAL). This variable has the value 1 for true dental
consonants. Specific indication that a segment is dental must be provided
in the source, for example if it says "the tongue touches the' teeth" or
"/t/ is like French t, not English t ". The description "dental" is often

164
 Patterns of sounds

applied to a segment which is more accurately described as alveolar, hence,

unless such added evidence is available, segments that are merely called
dental are not assumed to be true dentals.
13. Unspecified dental or alveolar (UNSPDENT). Segments simply indicated by
transcriptions using /\, d, s/ etc. may be either dental or alveolar. In
order not to falsify the data, this variable is used to indicate such
segments with an incompletely specified place of articulation. Segments
described simply as "dentals" are also included here.
14. Alveolar (ALVEOLAR).
15. Palato-alveolar (PALATALV). Palato-alveolar and alveo-palatal are not
distinguished.
16. Retroflex (RETROFLX).
17. Palatal (PALATAL).
18. Velar (VELAR).
19. Uvular (UVULAR).
20. Pharyngeal (PHARYNGL).
21. Glottal (GLOTTAL). This variable is only used for glottal stops; the
characterization of /h/ as a "glottal fricative" is rejected.

Variables indicating manner of articulation

Conventional phonetic labels, such as "plosive", "click", "vowel" generally
combine information on aperture and airstream. This set of variables does
likewise. All segments, except /h/ and /n/, must receive the value 1 for
least one of these; rarely, a segment may have the value 1 for more than
one, e.g. a fricative trill.
22. Plosive (PLOSIVE). Pulmonic egressive stops, including glottal stops,
receive the value 1 for this variable. Note that nasals are not considered
to be stops of any sort.
23. Implosive (IMPLOSIV). Glottalic ingressive stops, whether voiced or
voiceless, have the value 1 for this variable.
24. Ejective stop (EJECTSTP).
25. Click (CLICK). Non-affricated clicks have the value 1 for this
variable.
26. Fricative (FRICATIV). Pulmonic egressive fricatives have the value 1
for this variable. Note that /h/ and /n/ are not considered fricatives.
27. Ejective fricative (EJECTFRC).
28. Affricate (AFFRICAT). Pulmonic egressive affricates have the value 1
for this variable.
29. Ejective affricate (EJECTAFF).

165
Design of UPSID

30. Affricated click (AFFCLICK).

31. Unspecified "r-sound" (UNSPECR). This variable takes the value 1 for
segments which are simply identified as some kind of "r-sound" (e.g. by
being transcribed /r/ or called a "vibrant") but which cannot be further
classified as a trill, tap, flap, approximant, etc.
32. Tap (TAP).
33. Flap (FLAP).
34. Trill (TRILL).
35. Approximant (APPROXMT). This variable has the value 1 for all
consonants produced with open approximation, e.g. "semi-vowels", r-glides,
nonfricative laterals, etc.
36. Nasal (NASAL). Nasal consonants (with complete oral closure) receive
the value 1 for this variable. Nasalized segments, including consonants
with nasal onsets, are identified by a different variable (see 45 below).
37. Simple vowel (SIMPVOWL). Monophthongal vowels have the value 1 for this
variable. A segment which is not identified as either a simple vowel or a
diphthong (38 below) is thereby classified as a consonant.
38. Diphthong (DIFTHONG). This variable takes the value 1 for unit
diphthongs. Phonetic diphthongs which are phonologically analyzed as the
result of a juxtaposition of simple vowels or a vowel and an approximant
are, of course, not units but sequences and their components will be
entered as separate phonemes.

Other consonant features

39. Lateral (LATERAL). This variable takes the value 1 for all lateral
segments, e.g. it indicates lateral release in lateral affricates.
40. Sibilant (SIBILANT). This variable serves to identify the class of
sibilants within the fricative/affricate group. In many languages this
phonetic property is functionally redundant as place of articulation will
distinguish the same class. However, it has occasional distinctive function
in differentiating between fricatives and/or affricates with similar place
of articulation but distinct acoustic characteristics (e.g. /s/ vs. / 9 / ) .

Secondary articulations
41. Labialized (LABLZED).
42. Palatalized (PALTLZED). This variable takes the value 1 only for true
palatalized consonants, i.e. those with a secondary palatal articulation.
Thus a segment /c/ which occurs in a language as part of a "palatalized"

166
 Patterns of sounds

series of stops /p\ "H/ etc., will not be coded with this feature. Instead
it will be reported as a palatal stop.
43. Velarized (VELRZED). This variable only takes the value 1 for true
velarized segments. This value was also used to characterize vowels with
velar stricture, reported in Siriono (829).
44. Pharyngealized (PHARGZED). Unlike the preceding secondary articulation
variables, which usually only have the value 1 for consonants, this
variable takes the value 1 equally for consonants and vowels.
45. Nasalized (NASLZED). This variable takes the value 1 for nasalized
consonants and vowels, i.e. those with simultaneous nasal and oral escape.
It is also used to characterize prenasalized stops (when these are clearly
units). Thus the value 1 for this variable in combination with the value 1
for any stop variable (except click or affricated click) indicates a nasal
onset to the stop.
46. Nasal release (NASRELSE). Takes the value 1 for postnasalized segments
only.

Vowel features
All simple vowels are specified by a value of 1 on one vowel height, one
vowel backness and one lip position variable. They may also have the value
1 for other variables to indicate other distinctions. Diphthong segments
are specified by assigning the value 1 to all the vowel quality variables
needed to describe both their beginning and end points. A set of diphthong
variables, discussed below, indicates the order of conflicting
specifications.
47. High (HIGH).
48. Higher mid (HIGHMID).
49. Mid (MID). This variable is used with systematic ambiguity for both
those vowels which are indicated as "mid" without further particularization
and those which are true mid vowels (i.e. lie between higher mid and lower
mid on a height scale).
50. Lower mid (LOWMID). Note that / e/ and /o/ are considered lower mid
vowels, not low vowels.
51. Low (LOW).
52. Front (FRONT).
53. Central (CENTRAL). Note that /a/ in most languages is considered a
central vowel, not a back vowel.
54. Back (BACK).

167
Design of UPSID

55. Nonperipheral (NONPERIF). This variable takes the value 1 for "laxed"
noncentral vowels which are produced away from the periphery of the vowel
space, for example, III and /o/. It may on occasion serve a mainly
diacritical function where other features fail to distinguish vowels.
56. Rounded (ROUNDED).
57. Unrounded (UNROUNDD).
58. Lip-compressed (LIPCOMP). This variable takes the value 1 for "labial"
vowels that are produced with vertical compression of the lips but no
drawing in and forward of the corners of the mouth ("rounding").
59. R-colored (RCOLORED). This variable takes the value 1 for retroflexed
or r-colored vowels.

Diphthong variables
The 3 variables for characterizing movement in diphthongs differ from most
other variables in the inventory, which indicate only presence or absence
of the attribute named by the variable. For diphthong variables a zero
value may indicate a specific property of a diphthong, namely movement in
the opposite direction to that indicated by the variable name. Also, unlike
most of the variables, they require reference to the values of other
variables for their interpretation. Their function is to indicate which
value precedes when conflicting specifications of vowel height, backness or
lip position are given to a single segment. This method of coding
diphthongs was adopted in order to avoid a very large number of variables.
60. Backing (BACKING). This variable takes the value 1 when the end point
of a diphthong is more back than the beginning, as in /ia/, /eu/, /eu/ etc.
It takes the value 0 when the end point is either more front than or has
the same degree of backness as the beginning, as in /oi/, /ae/, /ou/, etc.
Note that only 3 degrees of backness are considered (front, central, back).
61. Lowering (LOWERING). This variable takes the value 1 for diphthongs
that have an end point lower than their beginning, such as /ia/, /ea/, etc.
It takes the value 0 when the endpoint is higher than or equal to the
beginning on the 5-point vowel height scale used in UPSID, for example
/oi/, /ou/, /ai/ etc.
62. Rounding (ROUNDING). This variable takes the value 1 when the endpoint
of a diphthong is rounded but the beginning is unrounded, as in /eu/, /ao/,
etc. It takes the value 0 when the endpoint is unrounded or both the
beginning and endpoints are rounded, as in /oi/, /ai/ or /ou/.

168
 Patterns of sounds

Note that a pair of diphthongs such as /oi/ and / io/ receive the same
values on the vowel variables. This will involve having conflicting values
for height, backness and rounded; they will be specified as being both high
and mid, front and back, rounded and unrounded. The diphthong features
interpret these conflicts. Thus /oi/ will have the value 0 for all 3
diphthong variables since there is no backing, lowering or rounding
movement in this diphthong. Yet because the segment has conflicting values,
there must be movement; it has to be in the opposite direction from the
variable names, i.e. fronting, raising and unrounding. On the other hand,
/io/ receives the value 1 for all diphthong features indicating that the
mid portion of the diphthong follows the high portion, the back portion
follows the front portion and the rounded portion follows the unrounded
portion.

Variables for phonation types, etc.

63. Voiceless (VOICELES). This variable takes the value 1 for all segments
in which the vocal cords do not vibrate. The segment /h/ has the value 1
for this variable but for no others.
64. Voiced (VOICED). This variable takes the value 1 for all segments in
which the vocal cords vibrate, whether as "regular" voicing or in some
other mode (creaky, breathy etc.). The segment /f\/ has the value 1 for this
variable, but for no others.
65. Aspirated (ASPIRATD). This variable takes the value 1 for all voiceless
aspirated segments. So-called "voiced aspirates" do not have the value 1
for this feature, but instead for the variable "breathy".
66. Laryngealized (creaky, glottalized) (LARYNGD). This variable takes the
value 1 for all segments with a laryngeal constriction in which that
constriction is not serving as an airstream initiator or primary place of
articulation. Thus, "glottalized" consonants, laryngealized vowels, etc.
have value 1 for this variable. Various distinctions which have been made
in the literature between "pre-glottalized", "postglottalized", "creaky"
etc. were judged to be inconsistently applied and hence it was decided not
to represent such distinctions in the inventory. Some inaccuracies
regarding the distinction between a segment which is glottalized and one
with a glottalic airstream (an implosive or ejective) have undoubtedly
remained.
67. Long (LONG). This variable takes the value 1 for contrastively long
vowels or geminate consonants which are single units but are not generated

169
Design of UPSID

as part of a general series of such long segments either from a

suprasegmental length feature or because adjacent identical segments occur.
(See "overshort" below.)
68. Breathy (voiced aspirated, murmured) (BREATHY). This variable takes the
value 1 for all segments characterized by breathiness.
69. Overshort (OVERSHORT). In certain languages, the basic series of vowels
is longer than a subset which may occur contrastively short. The variable
"overshort" takes the value 1 for such contrastively short segments.
70. Preaspirated (PREASPRT).

The anomaly variable

80. Anomaly (ANOMALY). This variable indicates segments with a somewhat
doubtful or marginal status in an inventory, but with sufficient claim to
be included that they were not simply eliminated from consideration. Unlike
the variables related to phonetic attributes, "anomaly" may take values
greater than 1. A value of 0 indicates a segment is a normal full member of
the inventory of the language. The other values have the meanings described
below:
1 - Indicates a segment of extremely low frequency (e.g. it only
occurs in a handful of words or certain morphological markers,
but these are well entrenched parts of the language).
2 - Indicates a segment that occurs only in foreign words or
unassimilated loans but these are frequent enough to consider
including the segment in the inventory.
3 - Indicates a segment which is posited in underlying forms to
account for some phonological patternings but which is
neutralized in surface forms. (Very rarely used.)
4 - Indicates a segment which is treated as phonemic in UPSID but
which may be regarded as derived from other underlying
segments. (Very rarely used.)
5 - Indicates a segment which although apparently a genuine member
of the inventory, is described in particularly obscure or
contradictory fashion (e.g. a segment in Ashuslay, 814,
described as simultaneously a (velar) stop and a lateral).
The values 3 and 4 were provided in order to provide an escape from
unresolved questions concerning membership of inventories. In practice,
very little use has been made of these indications as a resolution has been
achieved in nearly all cases.

170
 Patterns of sounds

10.5 Using UPSID

As readers may be interested in the practicalities of working with UPSID, a
brief description of its use is included as a final section of this
chapter. UPSID consists of 9957 records of the form described in section
10.4. These constitute a SAS (Statistical Analysis System) file on a
resident disk pack at UCLA's Central Computing facility. SAS (SAS Institute
Inc. 1982) is a powerful and flexible data manipulation and statistical
analysis system. It conveniently permits selection of subsets of records,
computation of language-by-language totals or frequency counts on the
complete data set, creation of new variables for special purposes and many
other operations to be performed. Of course, it will also
straightforwardly print out the information in the file. The phonemic
charts in the data section of this book have been compiled from a printout
of the information on each language.
Without duplicating procedural information, which is more properly
sought in the SAS manual, an example of the use of UPSID will be given.
This, it is hoped, will help to clarify the conceptual steps involved in
obtaining answers from the database.
Suppose that one wishes to use the UPSID data to check the hypothesis
that no nasal consonant appears in a language unless a stop occurs at the
same place of articulation. As a first step it is necessary to define more
precisely what is intended by the terms such as "stop", "nasal", "place of
articulation" in this hypothesis and then to translate this into the
appropriate set of UPSID variable values. Let us assume that what is
intended by "stops" are plosives and central affricates, but not lateral
affricates, implosives, ejectives, clicks etc., and "nasals" are voiced
nasal consonants. Let us also assume that "place of articulation" refers to
the usual set of primary places of articulations, that double articulations
count as separate "places" but that secondary articulations will not be
taken into account (e.g. /k/ and /k w / will both count as velar).
The data to be selected for examination from the main file is now
clear. It includes any segment meeting one of the following specifications:
(i) it has the value 1 for the variable "plosive"; (ii) it has the value 1
for the variable "affricate" and the value 0 for "lateral"; (iii) it has
the value 1 for the variables "nasal" and "voiced". All other segments can
be excluded and all variables except for "nasal" and those relating to
places of articulation can be dropped from the new, more compact, data file
which is to be tested. Among places of articulation, pharyngeal and glottal
can also be dropped as nasals cannot be produced at these places.

171
Design of UPSID

Now, we may restate the hypothesis as it relates to this reduced data

set. The question to be answered is: for each segment with a given language
number and the value 1 for "nasal", is there a segment with a matching
language number, matching values for the place of articulation features and
the value 0 for the variable "nasal"?
There are a number of ways one might proceed in order to obtain the
answer to this question, but one simple way is to create an index which
represents in a single number the information that one is interested in.
There are 10 relevant place of articulation variables, plus one double
articulation (labial-velar) that is found in nasals. Since labial-velars
are to be counted separately, a new "labial-velar" variable can be
generated and the original values for "labial" and "velar" set to 0 for
labial-velar segments. Next, 11 additional variables, one for each place at
which nasals occur, can be generated and set to some conveniently large
value, say 10, for every segment which is a nasal at the given place. The
value 1 can then be changed to 0 for the original place variables for all
the nasals, leaving only the stops and affricates with non-zero values for
these variables. One can then sum the totals for each variable on a
language-by-language basis. This will produce a single line of data for
each language in which each non-zero value for an original place of
articulation variable indicates that a plosive or affricate appears at that
place in the language and non-zero values for the new nasal place variables
indicate that a nasal occurs at that place.
One can now add the summed value for each original place variable and
the corresponding nasal place variable for each language. This will produce
a two-digit number which indicates whether or not each nasal is matched by
a plosive at the same place. Any nasals remaining unmatched will be
indicated by a value of 10 (or a multiple of 10) for the relevant nasal
place variable. These cases can easily be listed and their frequency,
distribution by place, etc., examined. For example, a language with the
inventory of stops, affricates and nasals shown below would have the value
12 (1 nasal, 2 plosives) at the bilabial and dental places of articulation
and the value 24 (2 nasals, 4 plosives) at the velar place of articulation:
p b t d tj k g kw gw
m n p Q QW

The palatal nasal place variable would have the value 10, indicating that
there was no palatal plosive or affricate matching the palatal nasal. This

172
 Patterns of sounds

language would then contain an exception to the hypothesis under

investigation. The hypothesis could be modified so that a palatal nasal was
considered to be matched by a palato-alveolar affricate and the modified
hypothesis tested by adding the values for the original "palato-alveolar"
and "palatal" variables together for each record to produce a more broadly
defined palatal variable. This would produce an index value of 11 (1 nasal,
1 affricate) and no exception to the hypothesis.

References
Bell, A. 1978. Language samples. In J.H. Greenberg et al. (eds.),
Universals of Human Language, Vol. J^, Method and Theory. Stanford
University Press, Stanford, California: 123-56.
Bender, M.L. 1976. Nilo-Saharan overview. In M.L. Bender (ed.), The
Non-Semitic Languages of Ethiopia (Monograph 5, Committee on Ethiopian
Studies, African Studies Center). Michigan State University, East
Lansing: 439-83.
Ferguson, C.A. 1963. Some assumptions about nasals. In J.H. Greenberg
(ed.), Universals of Language. MIT Press, Cambridge: 42-7.
Gandour, J. 1975. On the representation of tone in Siamese. In J.G.
Harris and J.R. Chamberlain (eds.), Studies in Tai Linguistics in
Honor of William J. Gedney. Central Institute of English Language,
Bangkok: 170-95.
Greenberg, J.H. 1966. The Languages of Africa. Indiana University Press,
Bloomington.
Greenberg, J.H. 1970. Some generalizations concerning glottalic consonants,
especially implosives. International Journal of American Linguistics 36:
123-45.
Greenberg, J.H. 1971. Nilo-Saharan and Meroitic. In T. Sebeok (ed.),
Current Trends In Linguistics, Vol. 7_> Sub-Saharan Africa. Mouton, The
Hague: 421-42.
Hurford, J.R. 1977. The significance of linguistic generalizations.
Language 53: 574-620.
Hyman, L.M. 1977. On the nature of linguistic stress. In L.M. Hyman (ed.),
Studies in Stress and Accent (Southern California Occasional Papers in
Linguistics 4 ) . University of Southern California, Los Angeles: 37-82.
Jakobson, R. and Halle, M. 1956. Phonology and Phonetics. (Part 1 of)
Fundamentals of Language. Mouton, The Hague: 3-51.
Ladefoged, P. 1971. Preliminaries to Linguistic Phonetics. University of
Chicago Press, Chicago and London.
Ladefoged, P., Glick, P. and Criper, C. 1972. Language in Uganda. Oxford
University Press, London.
SAS Institute Inc. 1982. SAS User's Guide, 1982 Edition: Basics. SAS
Institute Inc, Cary, North Carolina.
Trubetskoy, N. 1939. Grundzllge der Phonologie (Travaux du Cercle
Linguistique de Prague 9 ) . Prague.
Tucker, A.N. and Bryan, M.A. 1956. The Non-Bantu Languages of Northeast
Africa (Handbook of African Languages, Part 3 ) . Oxford University Press
for International African Institute, London.
Tucker, A.N. and Bryan, M.A. 1966. Linguistic Analyses: The Non-Bantu
Languages of Northeastern Africa. Oxford University Press for
International African Institute, London.
Vihman, M.M. 1974. Excerpts from the Phonology Archive coding manual.
Working Papers on Language Universals (Stanford University) 15: 141-53.

173
Appendix A: Language lists and bib1i og raphy of data sources

The languages included in the UPSID sample are listed below, first
according to a genetic classification, and secondly in alphabetical order.
The genetic listing enables a quick check to be made on which languages are
included from a given family. The genetic classification provided is
intended only as an outline. Main subfamilies that are not listed have no
representative in UPSID. In a few cases the affiliation of a language is
uncertain (e.g. Cofan, 836). For convenience, the large Amerindian family
has been divided on a geographical basis. The alphabetical listing enables
a check to be made for inclusion of a particular language. The language
names are those used throughout the book, but in some cases a
cross-reference from an alternative name is given to assist in tracking
down a given language. In both lists the language name is followed by the
language identification number. These numbers are assigned mainly on a
genetic basis (see Chapter 10). Note that, because of late additions and
deletions to the sample, the sequence of numbers within a genetic group is
not always continuous. The phonemic charts in this book are presented in
the order of these identification numbers. The alphabetical list also
serves as the key to the data sources consulted for each language. These
sources are identified by author and date; the full references may be found
in the bibliography of sources which forms the third part of this appendix.
A language included in the Stanford Phonology Archive (SPA) is indicated by
S after the language name in both lists. It may be assumed that the SPA
report was consulted in determining the inventory for that language. For
most of these languages, the sources used by SPA have also been directly
consulted, and, for a few, additional or different sources were used. The
remaining languages were analyzed solely at UCLA.

1. Genetic listing of languages and outline classification

Indo-European (000-049)
Greek: Greek 000 S
Celtic: Irish 001 S, Breton 002 S
Germanic: German 004 S, Norwegian 006 S
Baltic: Lithuanian 007 S
Slavic: Russian 008 S, Bulgarian 009 S
Romance: French 010 S, Spanish 011 S, Romanian 012 S
Iranian: Farsi 013 S, Pashto 014 S, Kurdish 015
Indie: Hindi-Urdu 016 S, Bengali 017 S, Kashmiri 018 S,
Punjabi 019 S, Sinhalese 020 S
Albanian: Albanian 021 S
Armenian: Eastern Armenian 022 S

Ural-Altaic (050-099>
Finno-Ugric: Ostyak 050 S, Cheremis 051 S, Komi 052 S,
Finnish 053 S, Hungarian 054 S, Lappish 055
Samoyed: Yurak 056 S, Tavgy 057
Turkic: Osmanli (Turkish) 058 S, Azerbaijani 059 S,
Chuvash 060 S, Yakut 061 S, Kirghiz 062 S,
Bashkir 063, Khalaj 064, Tuva 065.
Mongolian: Mongolian 066 S
Tungus: Evenki 067 S, Goldi 068, Manchu 069
Korean: Korean 070 S
Japanese: Japanese 071 S

174
 Language l i s t s

Niger-Kordofanian (100-199)
Kordofanian: Katcha 100 S, Moro 101, Kadugli 102
Mande: Kpelle 103 S, Bisa 104, Bambara 105, Dan 106
West Atlantic: Wolof 107 S, Diola 108, Temne 109
Voltaic: Dagbani 110 S, Senadi 111, Tampulma 112, Bariba 113
Kwa: Ewe 114 S, Akan 115 S, Igbo 116 S, G3 117 S
Togo Remnant: Lelemi 118
Cross River: Efik 119
Plateau: Birom 120, Tarok (Yergam) 121, Amo 122
Bantoid: Beembe 123 S, Swahili 124 S, Luvale 125 S,
Zulu 126 S, Teke (Kukuya)127
Adamawa: Doayo 128
Eastern: Gbeya 129 S, Zande 130

Nilo-Saharan (200-249)
Songhai: Songhai 200 S
Saharan: Kanuri 201 S
Maban: Maba 202
Fur: Fur 203
Eastern Sudanic: Maasai 204 S, Luo 205 S, Nubian 206 S,
Nyangi 207, Ik 208, Sebei 209, Tama 210,
Temein 211, Nera 212, Tabi 213, Mursi 214
Central Sudanic: Logbara 215 S, Yulu 216, Sara 217
Berta: Berta 218
Kunama: Kunama 219
Koman: Koma 220

Afro-Asiatic (250-299)
Semitic: Arabic 250 S, Tigre 251 S, Amharic 252 S,
Hebrew 253 S, Socotri 254, Neo-Aramaic 255
Berber: Shilha 256 S, Tuareg 257
Cushitic: Somali 258 S, Awiya 259 S, Iraqw 260 S,
Beja 261
Omotic: Kullo 262, Dizi 263, Kefa 264, Hamer 265
Chadic: Hausa 266 S, Angas 267 S, Margi 268 S,
Ngizim 269, Kanakuru 270

Austro-Asiatic (300-349)
Munda: Mundari 300 S, Kharia 301 S
Khasi: Khasi 302 S
Vietmuong: Vietnamese 303 S
Bahnaric: Sedang 304 S
Khmer: Khmer 306 S

Australian (350-399)
Iwaidjan: Maung 350 S
Bureran: Burera 352
Tiwian: Tiwi 351
Nunggubuyan: Nunggubuyu 353 S
Ma ran: Alawa 354 S
Daly: Maranungku 355 S, Malakmalak 356
Nyulnyulan: Bardi 357
Pama-Nyungan: Wik-Munkan 358 S, Kunjen 359 S,
Western Desert 360 S, Nyangumata 361 S,
Aranda 362, Kariera-Ngarluma 363,
Gugu-Yalanji 364, Mabuiag 365,
Arabana-Wanganura 366, Diyari 367, Bandjalang 368

175
Language lists

Austro-Tai (400-499)
Kam-Tai: Standard Thai 400 S, Lakkia 401 S, Yay 402 S,
Sui 403, Saek 404, Po-ai 405, Lungchow 406
Atayalic: Atayal 407 S
West Indonesian: Sundanese 408 S, Javanese 409 S, Malagasy 410 S,
Cham 411 S, Malay 412 S, Batak 413 S
Philippine: Tagalog 414 S, Sa'ban 415 S, Chamorro 416 S,
Rukai 417
Formosan: Tsou 418
N.E. New Guinea: Adzera 419 S, Roro 420
New Britain: Kaliai 421 S
Loyalty Is: Iai 422 S
Polynesian: Maori 423 S, Hawaiian 424 S

Sino-Tibetan (500-599)
Sinitic: Mandarin 500 S, Taishan 501 S, Hakka 502 S,
Changchow 503 S, Amoy 504, Fuchow 505, Kan 506
Himalayish: Tamang 507
Mirish: Dafla 508 S
Lolo-Burmese: Burmese 509 S, Lahu 510 S
Kachin: Jingpho 511
Kuki-Chin: Ao 512, Tiddim Chin 513
Baric: Garo 514 S, Boro 515
Karenic: Karen 516 S
Miao-Yao: Yao 517 S

Indo-Pacific (600-699)
Andamanese: Andamanese 600
West New Guinea: Asmat 601 S
North New Guinea: Washkuk 602 S, Sentani 603 S, Nimboran 604, Iwam 605
South-East New Guinea: Telefol 606 S
Central New Guinea: Selepet 607 S, Gadsup 608 S, Yagaria 609, Kewa 610,
Chuave 611, Pawaian 612, Dani 613, Wantoat 615,
Daribi 616, Fasu 617
South New Guinea: Suena 618
North-East New Guinea: Dera 619
East New Guinea: Kunimaipa 620 S, Yareba 621, Koiari 622, Taoripi 623
Bougainville: Nasioi 624 S, Rotokas 625
Central Melanesian: Nambakaengo 626

Amerindian I (Northern) (700-799)

Haida: Haida 700 S
Tlingit: Tlingit 701
Athapaskan: Navaho 702 S, Chipewyan 703 S, Tolowa 704 S,
Hupa 705 S
Northern Penutian: Nez Perce 706 S, Klamath 707
California Penutian: Maidu 708 S, Wintu 709
Mexican Penutian: Chontal 710 S, Zoque 711 S, Tzeltal 712 S,
Totonac 713 S, K'ekchi 714, Mixe 715
Oto-Manguean: Otomi 716 S, Mazahua 717 S, Mazatec 727 S,
Mixtec 728 S, Chatino 729
Wakashan: Nootka 730 S, Kwakw'ala 731 S
Chemakuan: Quileute 732
Salish: Squamish 733 S, Puget Sound 734 S
Uto-Aztecan: Papago 736 S, LuiseTio 737 S, Hopi 738 S, Yaqui 739
Kiowa-Tanoan: Tiwa (Picuris) 740 S

176
 Language lists

Hokan: Karok 741 S, Porno 742 S, Diegueflo 743 S,

Achumawi 744, Yana 745, Shasta 746
Tarascan: Tarascan 747 S
Zuni: Zuni 748 S
Keres: Acoma 749
Macro-Algonkian: Ojibwa 750 S, Delaware 751 S, Tonkawa 752,
Wiyot 753
Macro-Siouan: Seneca 754 S, Wichita 755 S, Dakota 756 S,
Yuchi 757 S, Tunica 758 S, Alabama 759 S
Wappo: Wappo 760

Amerindian II (Southern) (800-899)

Chibchan: Itonama 800 S, Bribri 801, Mura 802
Paezan: Cayapa 803 S, Paez 804 S
Witotoan: Ocaina 805 S, Muinane 806
Carib: Carib 807 S
Macro-Ge: Apinaye 809 S
Pano-Tacanan: Amahuaca 810 S, Chacobo 811 S, Tacana 812,
Cashinahua 813
Mataco: Ashuslay 814
Guaycuru: Abipon 815
Nambiquara: Southern Nambiquara 816
Zaparoan: Arabela 817, Auca 818
Quechumaran: Quechua 819 S, Jaqaru 820 S
Chon: Gununa-Kena (Puelche) 821
Arawakan: Wapishana 822 S, Island Carib 823 S, Amuesha 824 S,
Campa 825 S, Guajiro 826 S, Moxo 827 S
Tupi: Guarani 828 S, Siriono 829 S
Guahibo-Pamigua: Guahibo 830
Tucanoan: Ticuna 831 S, Barasano 832 S, Siona 833, Tucano 834
Jivaroan: Jivaro 835 S, Cofan 836
Penutian: Araucanian 837 S

Other Families (900-999)

Eskimo-Aleut: Greenlandic 900 S, Aleut 901 S
Dravidian: Telugu 902 S, Kota 903 S, Kurukh 904 S,
Malayalam 905, Brahui 917
Paleo-Siberian: Ket 906 S, Yukaghir 907 S, Chukchi 908 S,
Gilyak 909 S
Caucasian: Georgian 910 S, Kabardian 911 S, Lak 912 S
Khoisan: Nama 913 S, !Xu 918
Basque: Basque 914 S
Burushaski: Burushaski 915 S
Ainu: Ainu 916 S
2
• Alphabetic list of languages with key to sources

Abipon 815 Najlis (1966)

Achumawi 744 Olmsted (1964, 1966)
Acoma 749 Miller (1966)
Adzera 419 S Holzknecht (1973)
Ainu 916 S Simeon (1969)
Akan 115 S Welmers (1946), Schachter and Fromkin (1968),
Stewart (1967)
Alabama 759 S Rand (1968)
Alawa 354 S Sharpe (1972)
Albanian 021 S Newmark (1957)

177
Language lists

Aleut 901 S Bergsland (1956)

Amahuaca 810 S Osborn (1948)
Amharic 252 S Leslau (1968), Klingenheben (1966), Sumner (1957)
Amo 122 di Luzio (1972)
Amoy Chinese 504 [Hanyu Fangyan Gaiyao] (1960)
Amuesha 824 S Fast (1953)
Andamanese 600 Voegelin and Voegelin (1966), Radcliffe-Brown (1914)
Angas 267 S Burquest (1971)
Ao 512 Gowda (1972)
Apinaye 809 S Burgess and Ham (1968)
Arabana-Wanganura 366 Hercus (1973)
Arabela 817 F u m e (1963)
Arabic 250 S Mitchell (1962), Tomiche (1964), Kennedy (1960)
Aranda 362 O'Grady, Voegelin and Voegelin (1966)
Araucanian 837 S Echeverria and Contreras (1965)
Armenian, Eastern 022 S Allen (1950)
Ashuslay 814 Stell (1972)
Asmat 601 S Voorhoeve (1965)
Atayal 407 Egerod (1966)
Auca 818 Saint and Pike (1962)
Awiya 259 S Hetzron (1969)
Aymara, see Jaqaru
Azande, see Zande
Azerbaijani 059 S Householder (1965)
Bambara 105 Bird, Hutchinson and Kante (1977)
Bandjalang 368 Cunningham (1969)
Barasano 832 S Stolte and Stolte (1971)
Bardi 357 Metcalfe (1971)
Bariba 113 Welmers (1952)
Bashkir 063 Poppe (1964)
Basque 914 S N'diaye (1970)
Batak 413 S van der Tuuk (1971)
Beembe 123 S Jacquot (1962)
Beja 261 Hudson (1976)
Bengali 017 S Ferguson and Chowdhury (1960)
Berta 218 Triulzi, Dafallah and Bender (1976)
Birom 120 Wolff (1959)
Bisa 104 Naden (1973)
Boro 515 Bhat (1968)
Brahui 917 Emeneau (1937), De Armond (1975)
Breton 002 S Ternes (1970)
Bribri 801 Arroyo (1972)
Bulgarian 009 S Klagstad (1958), Aronson (1968)
Burera 352 Glasgow and Glasgow (1967)
Burmese 509 S Okell (1969)
Burushaski 915 S Morgenstierne (1945)
Campa 825 S Dirks (1953)
Carib 807 S Peasgood (1972), Hoff (1968)
Cashinahua 813 Kensinger (1963)
Cayapa 803 S Lindskoog and Brend (1962)
Chacobo 811 S Prost (1967)
Cham 411 S Blood (1967)
Chamorro 416 S Topping (1973,1969), Seiden (1960)
Changchow Chinese (Wu) 503 S Chao (1970)
Chasta Costa, see Tolowa
Chatino 729 Pride (1965)
Cheremis 051 S Ristinen (1960)

178
 Language lists

Chipewyan 703 S Li (1946, 1933, 1932)

Chontal 710 S Keller (1959)
Chuave 611 Thurman (1970)
Chukchi 908 S Skorik (1968,1961)
Chuvash 060 S Andreev (1966), Krueger (1961)
Cofan 836 Bo m a n (1962)
Dafla 508 S Ray (1967)
Dagbani 110 S Wilson and Bendor-Samuel (1969)
Dakota 756 S Boas and Deloria (1939)
Dan 106 Bearth and Zemp (1967)
Dani 613 Bromley (1961), van der Stap (1966)
Daribi 616 MacDonald (1973)
Delaware 751 S Voegelin (1946)
Dera 619 Voorhoeve (1971)
Diegueflo 743 S Langdon (1970)
Diola 108 Sapir (1965)
Diyari 367 Austin (1978)
Dizi 263 Allen (1976b)
Doayo 128 Wiering (1974)
Efik 119 Cook (1969)
Evenki 067 S Novikova (1960)
Ewe 114 S Berry (no date a ) , Stahlke (1971), Ladefoged (1968)
Fante, see Akan
Farsi 013 S Obolensky, Panah and Nouri (1963)
Fasu 617 Loeweke and May (1964)
Finnish 053 S Lehtinen (1964), Harms (1964)
French 010 S Sten (1963)
Fuchow Chinese 505 [Hanyu Fangyan Gaiyao] (1960)
Fur 203 Beaton (1968), Tucker and Bryan (1966)
G3 117 S Berry (no date b ) , J.N.A. Nartey (p.c.)
Gadsup 608 S Frantz and Frantz (1966)
Garo 514 S Burling (1961)
Gbeya 129 S Samarin (1966)
Georgian 910 S Robins and Waterson (1952), Tschenkeli (1958),
Vogt (1938, 1958)
German 004 S Moulton (1962), Philipp (1974)
Gilyak (Nivkh) 909 S Panfilov (1962, 1968)
Goldi 068 Avrorin (1968)
Greek 000 S Householder, Kazazis and Koutsoudas (1964)
Greenlandic 900 S Rischel (1974), Thalbitzer (1904),
Kleinschmidt (1851)
Guahibo 830 Kondo and Kondo (1967)
Guajiro 826 S Holmer (1949)
Guarani 828 S Gregores and Suarez (1967), Uldall (1956), Lunt (1973)
Gugu-Yalanji 364 Oates and Oates (1964), Wurm (1972a), Oates (1964)
Gununa-Kena 821 Gerzenstein (1968)
Haida 700 S Sapir (1923)
Hakka Chinese 502 S Hashimoto (1973)
Hamer 265 Lydall (1976)
Hausa 266 S Hodge (1947), Kraft and Kraft (1973), Hodge and
Umaru (1963)
Hawaiian 424 S Pukui and Elbert (1965)
Hebrew 253 S Chayen (1973), Cohen and Zafrani (1968)
Hindi-Urdu 016 S Kelkar (1968), Vermeer and Sharma (1966)
Hopi 738 S Whorf (1946), Voegelin (1956)
Huambisa, see Jivaro
Hungarian 054 S Kalman (1972), Banhidi, Jokay and Szabo (1965),

179
Language lists

Hall (1938, 1944)

Hupa 705 S Woodward (1964), Golla (1970)
Iai 422 S Ozanne-Rivierre (1976), Tryon (1968)
Igbo 116 S Williamson (1969), Carnochan (1948), Swift, Ahaghota
and Ugorji (1962)
Ignaciano, see Moxo
Ik 208 Heine (1975b)
Inuit, see Greenlandic
Iraqw 260 S Whiteley (1958)
Irish 001 S Mhac an Fhailigh (1968), Sommerfelt (1964)
Island Carib 823 S Taylor (1955)
Itonama 800 S Liccardi and Grimes (1968)
Iwarn 605 Laycock (1965)
Japanese 071 S Bloch (1950), Martin (1952), Jorden (1963)
Jaqaru 820 S Hardman (1966)
Javanese 409 S H o m e (1961)
Jingpho 511 Liu (1964)
Jivaro 835 S Beasley and Pike (1957)
Kabardian 911 S Kuipers (1960)
Kadugli 102 Abdalla (1973)
Kaliai 421 S Counts (1969)
Kan Chinese 506 [Hanyu Fangyan Gaiyao] (1960)
Kanakuru 270 Newman (1974)
Kanuri 201 S Lukas (1937)
Karen (Sgaw) 516 S Jones (1961)
Kariera-Ngarluma 363 O'Grady, Voegelin and Voegelin (1966), Wurm (1972a)
Karok 741 S Bright (1957)
Kashmiri 018 S Kelkar and Trisal (1964)
Katcha 100 S Stevenson (1957) Tucker and Bryan (1966)
Kefa 264 Fleming (1976)
K'ekchi 714 Haeseriju (1966), Freeze (1975)
Ket 906 S Dul'zon (1968), Krejnovich (1968a)
Kewa 610 Franklin and Franklin (1962)
Khalaj 064 Doerfer (1971)
Khalka see Mongolian
Kharia 301 S Biligiri (1965), Pinnow (1959)
Khasi 302 S Rabel (1961)
Khmer 306 S Huffman (1970a, 1970b), Jacob (1968)
Kirghiz 062 S Herbert and Poppe (1963)
Klamath 707 Barker (1964)
Koiari 622 Dutton (1969)
Koma 220 Tucker and Bryan (1966)
Komi 052 S Lytkin (1966), Bubrikh (1949)
Korean 070 S Martin (1951), Cho (1967), Martin and Lee (1969)
Kota 903 S Emeneau (1944)
Kpelle 103 S Welmers (1962), Hyman (1973)
Kullo 262 Allen (1976a)
Kunama 219 Tucker and Bryan (1966)
Kunimaipa 620 S Pence (1966)
Kunjen 359 S Sommer (1969)
Kurdish 015 Abdulla and McCarus (1967)
Kurukh 904 S Pinnow (1964), Pfeiffer (1972)
Kwakw'ala 731 i S. R. Anderson ( p . c ) , Boas (1947) Grubb (1977)
Lahu 510 S Matisoff (1973)
Lak 912 S Murkelinskij (1967), Zhirkov (1955), Khaidakov (1966)
Lakkia 401 S Haudricourt (1967)
Lappish 055 Hasselbrink (1965)

180
 Language lists
Lelemi 118 Hoftmann (1971)
Lithuanian 007 S Senn (1966), Augustitis (1964), Ambrazas, et al. (1966)
Logbara 215 S Crazzolara (I960), Tucker and Bryan (1966)
Luiseffo 737 S Malecot (1963), Bright (1965, 1968), Kroeber and
Grace (1960)
Lungchow 406 Li (1977)
Luo 205 S Gregersen (1961)
Lushootseed, see Puget Sound Salish
Luvale 125 S Horton (1949)
Maasai 204 S Tucker and Mpaayei (1955), Tucker and Bryan (1966)
Maba 202 Tucker and Bryan (1966)
Mabuiag 365 Wurm (1972a)
Maidu 708 S Shipley (1956, 1964)
Malagasy 410 S Dahl (1952), Dyen (1971)
Malakmalak 356 Tryon (1974), Birk (1975)
Malay 412 S Verguin (1967), MacDonald and Soenyono (1967)
Malayalam 905 Kumari (1972), McAlpin (1975), Velayudhan (1971)
Manchu 069 Austin (1962)
Mandarin Chinese 500 S Dow (1972), Chao (1968), C-C. Cheng (1973)
Maori 423 S Biggs (1961), Hohepa (1967)
Maranungku 355 S Tryon (1970)
Margi 268 S Hoffmann (1963)
Maung 350 S Capell and Hinch (1970)
Mazahua 717 S Spotts (1953)
Mazatec 727 S Pike and Pike (1947)
Mixe (Totontepec) 715 Crawford (1963), Schoenhals and Schoenhals (1965)
Mixtec 728 S Hunter and Pike (1969)
Mongolian (Khalka) 066 S Hangin (1968), Street (1963), Luvsanvandan (1964)
Moro 101 Black (1971)
Moxo 827 S Ott and Ott (1967)
Muinane 806 Walton and Walton (1967)
Mundari 300 S Gumperz and Biligiri (1957)
Mura 802 Sheldon (1974)
Mursi 214 Turton and Bender (1976)
Nama 913 S Beach (1938), Ladefoged and Traill (1980)
Nambakaengo 626 Wurm (1972b)
Nambiquara, Southern 816 Price (1976)
Nasioi 624 S Hurd and Hurd (1966)
Navaho 702 S Sapir and Hoijer (1967)
Nenets, see Yurak
Neo-Aramaic 255 Garbell (1965)
Nera 212 Thompson (1976)
Nez Perce 706 S Aoki (1970, 1966)
Ngizim 269 Schuh (1972)
Nimboran 604 Anceaux (1965)
Nivkh, see Gilyak
Nootka (Tseshaht) 730 S Sapir and Swadesh (1939, 1955)
Norwegian 006 S Vanvik (1972)
Nubian (Mahas) 206 S Bell (1971)
Nunggubuyu 353 S Hughes and Leeding (1971)
Nyangi 207 Heine (1975a)
Nyangumata 361 S O'Grady (1964)
Ocaina 805 S Agnew and Pike (1957)
Ojibwa 750 S Bloomfield (1956)
Osmanli (Turkish) 058 S Swift (1963), Lees (1961)
Ostyak 050 S Gulya (1966)
Otomi 716 S Blight and Pike (1976)

181
Language lists

Paez 804 S Gerdel (1973)

Papago 736 S Hale (1959), Saxton (1963), I. Maddieson (p.c.)
Pashto 014 S Shafeev (1964)
Pawaian 612 Trefry (1972)
Persian, see Farsi
Po-Ai 405 Li (1977)
Porno 742 S Moshinsky (1974)
Puget Sound Salish 734 S Snyder (1968)
Punjabi 019 S Gill and Gleason (1963)
Quechua 819 S Bills, Vallejo and Troike (1969), Lastra (1968)
Quileute 732 Powell (1975)
Romanian 012 S Agard (1958), Ruhlen (1973)
Roro 420 Bluhme (1970), Davis (1974)
Rotokas 625 Firchow and Firchow (1969)
Rukai 417 Li (1973)
Russian 008 S Jones and Ward (1969), Halle (1959)
Sa'ban 415 S Clayre (1973)
Saek 404 Gedney (1970)
Sara 217 Caprile (1968), Thayer and Thayer (1971)
Sebei 209 Montgomery (1970)
Sedang 304 S Smith (1968)
Selepet 607 S McElhanon (1970)
Senadi 111 Welmers (1950)
Seneca 754 S Chafe (1967)
Sentani 603 S Cowan (1965)
Shasta 746 Silver (1964)
Shilha 256 S Applegate (1958)
Sinhalese 020 S Coates and da Silva (1960)
Siona 833 Wheeler and Wheeler (1962)
Siriono 829 S Priest (1968)
Socotri 254 Johnstone (1975), Leslau (1938)
Somali 258 Armstrong (1934), Andrzejewsky (1955)
Songhai 200 Prost (1956), Williamson (1967)
Spanish 011 Navarro (1961), Harris (1969), Saporta and Contreras
(1962)
Squamish 733 Kuipers (1967)
Suena 618 Wilson (1969)
Sui 403 Li (1948)
Sundanese 408 Van Syoc (1959), Robins (1953)
Swahili 124 Polome (1967)
Tabi 213 Tucker and Bryan (1966)
Tacana 812 Key (1968), Van Wynen and Van Wynen (1962)
Tagalog 414 Bloomfield (1917), Schachter and Otanes (1972)
Tahaggart, see Tuareg
Taishan Chinese 501 S T.M. Cheng (1973)
Tama 210 Tucker and Bryan (1966)
Tamang 507 Mazaudon (1973)
Tampulma 112 Bergman, Gray and Gray (1969)
Taoripi 623 Brown (1973)
Tarascan 747 S Foster (1969)
Tarok 121 Robinson (1974)
Tavgy 057 Castren (1966), Terescenko (1966)
Teke (Kukuya) 127 Paulian (1975)
Telefol 606 S Healey (1964)
Telugu 902 Lisker (1963), Krishnamurti (1961), Kelley (1963)
Kostid et al (1977)
Temein 211 Tucker and Bryan (1966)

182
 Bibliography

Temne 109 Dalby (1966), Wilson (1961)

Thai (Standard) 400 S Abramson (1962), Noss (1954, 1964)
Ticuna 831 S Anderson (1959), Anderson (1962)
Tiddim Chin 513 Henderson (1965)
Tigre 251 S Palmer (1962)
Tiwa (Picuris) 740 S Trager (1971)
Tiwi 351 Osborne (1974)
Tlingit 701 Story and Naish (1973), Swanton (1909)
Tolowa 704 S Bright (1964)
Tonkawa 752 Hoijer (1972, 1949)
Totonac 713 S Aschmann (1946)
Tseshaht, see Nootka
Tsou 418 Tung (1964)
Tuareg 257 Prasse (1972)
Tucano 834 West and Welch (1967)
Tunica 758 S Haas (1941)
Turkish, see Osmanli
Tuva 065 Sat (1966)
Tzeltal 712 S Kaufman (1971)
Vietnamese 303 S Thompson (1965)
Wantoat 615 Davis (1969)
Wapishana 822 S Tracy (1972)
Wappo 760 Sawyer (1965)
Washkuk 602 S Kooyers, Kooyers and Bee (1971)
Western Desert 360 S Douglas (1955, 1964)
Wichita 755 S Garvin (1950), Rood (1975)
Wik-Munkan 358 S Sayers and Godfrey (1964)
Wintu 709 Broadbent and Pitkin (1964)
Wiyot 753 Teeter (1964)
Wolof 107 S Sauvageot (1965)
!Xa 918 Snyman (1969, 1975)
Yagaria 609 Renck (1967, 1975)
Yakut 061 S Krueger (1962), BBhtlingk (1964)
Yana 745 Sapir and Swadesh (1960)
Yao 517 S Purnell (1965)
Yaqui 739 Johnson (1962), Crumrine (1961)
Yareba 621 Weimer and Weimer (1972)
Yay 402 S Gedney (1965)
Yuchi 757 S Crawford (1973), Ballard (1975)
Yukaghir 907 S Krejnovich (1958, 1968b)
Yulu 216 Thayer (1969), Santandrea (1970)
Yurak 056 S Hajdu (1963), Decsy (1966), Ristinen (1965, 1968)
Zande 130 Tucker and Hackett (1959)
Zoque 711 S Wonderly (1951)
Zulu 126 S Doke (1926, 1961)
Zuni 748 S Newman (1965)

3. Bibliography oj£ data sources

Abdalla, A.I. 1973. Kadugli Language and Language Usage (Salambi Prize
Series 3 ) . Khartoum University Press, Khartoum.
Abdulla, J. and McCarus, E.N. 1967. Kurdish Basic Course - Dialect £f_
Sulaimania, Iraq. University of Michigan Press, Ann Arbor.
Abramson, A.S. 1962. The Vowels and Tones of Standard Thai: Acoustical
Measurements and Experiments (IJAL 28/2 Part II, Indiana University
Publications in Anthropology, Folklore and Linguistics 20). Indiana
University, Bloomington.
Agard, F.B. 1958. Structural Sketch of Rumanian (Language monograph 26).
Linguistic Society of America, Baltimore.

183
Bibliography

Agnew, A. and Pike, E.G. 1957. Phonemes of Ocaina (Huitoto). International

Journal of American Linguistics 23: 24-7.
Allen, E.J. 1976a. Kullo. In M.L. Bender (ed.), The Non-Semitic Languages
of Ethiopia. African Studies Center, Michigan State University, East
Lansing: 324-50.
Allen, E.J. 1976b. Dizi. In M.L. Bender (ed.), The Non-Semitic Languages of
Ethiopia. African Studies Center, Michigan State University, East
Lansing: 377-92.
Allen, W.S. 1950. Notes on the phonetics of an Eastern Armenian speaker.
Transactions of the Philological Society: 180-206.
Ambrazas, V., Vajtkavichjute, V., Valjatskene, A., Morkunas, K.,
Sabaljauskas, A., and Ul'vidas, K. 1966. Litovskij jazyk. In V.V.
Vinogradov (ed.), Jazyki Narodov SSSR, Vol. 1. Nauka, Leningrad and
Moscow: 500-27.
Anceaux, J.C. 1965. The Nimboran Language: Phonology and Morphology.
Nijhoff, The Hague.
Anderson, D. 1962. Conversational Ticuna. Summer Institute of Linguistics,
University of Oklahoma, Norman.
Anderson, L. 1959. Ticuna vowels: with special regard to the system of five
tonemes. PublicacSos do Museu Nacional, Serie Lingllistica Especial 1
(Museu Nacional, Rio de Janeiro): 76-119.
Andrzejewsky, B.W. 1955. The problem of vowel representation in the Isaaq
dialect of Somali. Bulletin of the School of Oriental and African
Studies 17: 567-80.
Andreev, I.A. 1966. Chuvashskij jazyk. In V.V. Vinogradov (ed.), Jazyki
Narodov SSSR Vol. 2. Nauka, Leningrad and Moscow: 43-65.
Aoki, H. 1966. Nez Perce vowel harmony and Proto-Sahaptian vowels. Language
42: 759-67.
Aoki, H. 1970. Nez Perce Grammar. University of California Press, Berkeley
and Los Angeles.
Applegate, J.R. 1958. An Outline of the Structure of Shilha. American
Council of Learned Societies, New York.
Armstrong, L.E. 1934. The phonetic structure of Somali. Mitteilungen des
Seminars fllr Orientalische Sprachen (Berlin) 37/3: 116-61.
Aronson, H.I. 1968. Bulgarian Inflectional Morphophonology. Mouton, The
Hague.
Arroyo, V.M. 1972. Lenguas Indigenas Costarricenses, 2nd ed. Editorial
Universitaria Centroamericana, San Jose, Costa Rica.
Aschmann, H.P. 1946. Totonaco phonemes. International Journal of American
Linguistics 12: 34-43.
Augustitis, D. 1964. Das Litauische Phonationssystem. Sagner, Munich.
Austin, P.A. 1978. A Grammar of the Diyari Language of North-East South
Australia. Ph. D. Thesis. Australian National University, Canberra.
Austin, W. 1962. The phonemics and morphophonemics of Manchu. In N. Poppe
(ed.), American Studies in Altaic Linguistics (Indiana University
Publications, Uralic and Altaic Series 13). Indiana University,
Bloomington: 15-22.
Avrorin, V.A. 1968. Nanajski jazyk. In V.V. Vinogradov (ed.), Jazyki
Narodov SSSR, Vol. 5. Nauka, Leningrad and Moscow: 5: 129-48.
Ballard, W.L. 1975. Aspects of Yuchi morphonology. In J.M. Crawford (ed.),
Studies in Southeastern Indian Languages. University of Georgia Press,
Athens, Georgia: 164-87.
Banhidi, Z., Jokay, A. and Szabo, D. 1965. Lehrbuch der Ungarischen
Sprache* Publishing House for Textbooks, Budapest.
Barker, M.A.R. 1964. Klamath Grammar. University of California Press,
Berkeley and Los Angeles.

184
 Bibliography

Beach, D.M. 1938. The Phonetics of the Hottentot Language. Heffer,

Cambridge.
Bearth, T. and Zemp, H. 1967. The phonology of Dan (Santa). Journal of
African Languages 6: 9-29.
Beasley, D. and Pike, K.L. 1957. Notes on Huambisa phonemics. Lingua
Posnaniensis 6: 1-8.
Beaton, A.C. 1968. A Grammar of the Fur Language (Linguistic Monograph
Series 1). Sudan Research Unit, University of Khartoum, Khartoum.
Bell, H. 1971. The phonology of Nobiin Nubian. African Language Review 9:
115-39.
Bergman, R., Gray, I., and Gray, C. 1969. Collected Field Reports on the
Phonology of Tampulma. Institute of African Studies, University of
Ghana, Legon.
Bergsland, K. 1956. Some problems of Aleut phonology. In M. Halle (ed.),
For Roman Jakobson. Mouton, The Hague: 38-43.
Berry, J. no date a. The Pronunciation of Ewe. Heffer, Cambridge.
Berry, J. no date b. The Pronunciation of Gg. Heffer, Cambridge.
Bhat, D.N.S. 1968. Boro Vocabulary. Deccan College, Poona.
Biggs, B. 1961. The structure of New Zealand Maori. Anthropological
Linguistics 3/3: 1-54.
Biligiri, H.S. 1965. Kharia: Phonology, Grammar and Vocabulary. Deccan
College, Poona.
Bills, G.D., Vallejo C , B. and Troike, R.C. 1969. An Introduction to
Spoken Bolivian Quecha. University of Texas Press for Institute of Latin
American Studies, Austin.
Bird, C , Hutchinson, J., and Kante, M. 1977. An Ka Bamanankan Kalan:
Beginning Bambara. Indiana University Press~Bloomington.
Birk, D.B.W. 1975. The phonology of Malakmalak. Papers in Australian
Linguistics 8 (Pacific Linguistics, Series A, 39): 59-78.
Black, Mr. and Mrs. K. 1971. The Moro Language: Grammar and Dictionary
(Linguistic Monograph Series 6 ) . Sudan Research Unit, University of
Khartoum, Khartoum.
Blight, R.C. and Pike, E.V. 1976. The phonology of Tenango Otomi.
International Journal of American Linguistics 42: 51-7.
Bloch, B. 1950. Studies in colloquial Japanese, IV: Phonemics. Language
26: 86-125.
Blood, D.L. 1967. Phonological units in Cham. Anthropological Linguistics
9: 15-32.
Bloomfield, L. 1917. Tagalog Texts with Grammatical Analysis (University of
Illinois Studies in Language and Literature 3/2-3). University of
Illinois, Urbana.
Bloomfield, L. 1956. Eastern Ojibwa. University of Michigan Press, Ann
Arbor.
Bluhme, H. 1970. The phoneme system and its distribution in Roro. In S.A.
Wurm and D.C. Laycock (eds.), Pacific Linguistic Studies in Honor of
Arthur Capell (Pacific Linguistics, Series C, 13): 867-77.
Boas, F. 1947. Kwakiutl grammar. Transactions of the American Philosophical
Society 37: 201-377.
Boas, F. and Deloria, E. 1939. Dakota Grammar (Memoirs of the National
Academy of Sciences 23/2). Washington, D.C.
BBhtlingk, 0. 1964. Uber die Sprache der Jakuten (reprint of Dr. A. Th. v.
Middendorff's Reise in der kussersten Norden und Osten Sibiriens, Band
III, St. Petersburg, 1851; Indiana University Publications, Uralic and
Altaic Series 35). Indiana University, Bloomington.
Borman, M.B. 1962. Cofan phonemes. Studies in Ecuadorian Indian Languages
l_. Summer Institute of Linguistics, University of Oklahoma, Norman:
45-9.

185
Bibliography

Bright, J.O. 1964, The phonology of Smith River Athapaskan (Tolowa).

International Journal of American Linguistics 30: 101-7.
Bright,~ W. 1957. The Karok Language. University "of California Press,
Berkeley and Los Angeles.
Bright, W. 1965. Luiseflo phonemics. Interna.tional Journal of American
Linguistics 31: 342-5.
Bright, W. 1968. A Luiseflo Dictionary. University of California Press,
Berkeley and Los Angeles.
Broadbent, S.M. and Pitkin, H. 1964. A comparison of Miwok and Wintu. In W.
Bright (ed.), Studies in Californian Linguistics. University of
California Press, Berkeley and Los Angeles: 19-45.
Bromley, H.M. 1961. The Phonology of Lower Grand Valley Dani
(Verhandelingen van het Koninklijk Instituut voor Taal-, Land- en
Volkenkunde 34). Nijhoff, The Hague.
Brown, A.H. 1973. The Eleman language family. In K. Franklin (ed.), The
Linguistic Situation in the Gulf District and Adjacent Areas, Papua New
Guinea (Pacific Linguistics, Series C, 26). Australian National
University, Canberra: 279-375.
Bubrikh, D.V. 1949. Grammatika Literaturnogo Komi Jazyka. Leningrad State
University, Leningrad.
Burgess, E. and Ham, P. 1968. Multilevel conditioning of phoneme variants
in Apinaye. Linguistics 41: 5-18.
Burling, R. 1961. A Garo Grammar. Deccan College, Poona.
Burquest, D.A. 1971. A Preliminary Study of Angas Phonology. Institute of
Linguistics, Zaria.
Capell, C. and Hinch, H.E. 1970. Maung Grammar. Mouton, The Hague.
Caprile, J-P. 1968. Essai de phonologie d'un parler mbay. Bulletin de la
SELAF 8: 1-40.
Carnochan, J. 1948. A study on the phonology of an Igbo speaker. Bulletin
of the School of Oriental and African Studies 22: 416-27.
Castren, M.A. 1966. Grammatik der Samojedischen Sprachen (Indiana
University Publications, Uralic and Altaic Series 53 [Reprint of 1854
edition, St. Petersburg]). Indiana University, Bloomington.
Chafe, W.L. 1967. Seneca Morphology and Dictionary. Smithsonian
Institution, Washington, D.C.
Chao, Y-R. 1968. A Grammar of Spoken Chinese. University of California
Press, Berkeley.
Chao, Y-R. 1970. The Changchow dialect. Journal of the American Oriental
Society 90: 45-59.
Chayen, M.J. 1973. The Phonetics of Modern Hebrew. Mouton, The Hague.
Cheng, C-C. 1973. A Synchronic Phonology of Mandarin Chinese. Mouton, The
Hague.
Cheng, T.M. 1973. The phonology of Taishan. Journal of Chinese Linguistics
1: 256-322.
Cho, S-B. 1967. A Phonological Study of Korean (Acta Universitatis
Upsaliensis, Studia Uralica et Altaica Upsaliensia 2 ) . Almqvist and
Wiksells, Uppsala.
Clayre, I.F.C.S. 1973. The phonemes of Sa'ban: a language of Highland
Borneo. Linguistics 100: 26-46.
Coates, W.A. and da Silva, M.W.S. 1960. The segmental phonemes of
Sinhalese. University of Ceylon Review 18: 163-75.
Cohen, D. and Zafrani, H. 1968. Grammaire de l'Hebreu Vivant. Presses
Universitaires de France, Paris.
Cook, T.L. 1969. The Pronunciation of Efik for Speakers of English. Indiana
University, Bloomington.
Counts, D.R. 1969. A Grammar of Kaliai-Kove. University of Hawaii Press,
Honolulu.

186
 Bibliography

Cowan, H.K.J. 1965. Grammar of the Sentani Language (Verhandelingen van het
Koninklijk Instituut "voor Taal-, Land- en Volkenkunde 47). Nijhoff, The
Hague.
Crawford, J.C. 1963. Tptontepec Mixe Phonotagmemics. Summer Institute of
Linguistics, University of Oklahoma, Norman.
Crawford, J.M. 1973. Yuchi phonology. International Journal of American
Linguistics 39: 173-9.
Crazzolara, J.P. 1960. A Study of the Logbara (Ma'di) Language. Oxford
University Press, London.
Crumrine, L.S. 1961. The Phonology of Arizona Yaqui (Anthropological Papers
of the University of Arizona 5 ) . University of Arizona, Tucson.
Cunningham, M.C. 1969. A Description of the Yugumbir Dialect of Bandjalang
(University of Queensland Faculty of Arts Papers 1/8). University of
Queensland Press, Brisbane.
Dahl, O.C. 1952. Etude de phonologie et de phonetique malagache. Norsk
Tidsskrift for Sprogvidenskap 16: 148-200.
Dalby, D. 1966. Lexical analysis in Temne with an illustrative wordlist.
Journal of West African Languages 3: 5-26.
Davis, D.R. 1969. The distinctive features of Wantoat phonemes. Linguistics
47: 5-17.
Davis, M.M. 1974. The dialects of the Roro language of Papua: a preliminary
survey. Kivung 8: 3-22.
De Armond, R.C. 1975. Some rules of Brahui conjugation. In H.G. Schiffman
and C M . Eastman (eds.), Dravidian Phonological Systems. University of
Washington Press, Seattle.
Decsy, G. 1966. Yurak Chrestomathy (Indiana University Publications, Uralic
and Altaic Series 50). Indiana University, Bloomington.
Di Luzio, A. 1972. Preliminary description of the Amo language. Afrika und
Ubersee 56: 3-60.
Dirks, S. 1953. Campa (Arawak) phonemes. International Journal of American
Linguistics 19: 302-4.
Doerfer, G. 1971. Khalaj Materials (Indiana University Publications, Uralic
and Altaic Series 115). Indiana University, Bloomington.
Doke, C M . 1926. The phonetics of the Zulu language. Bantu Studies Special
number.
Doke, C M . 1961. Textbook of Zulu Grammar. Longmans, Cape Town.
Douglas, W.H. 1955. Phonology of the Australian aboriginal language spoken
at Ooldea, South Australia, 1951-1952. Oceania 25: 216-29.
Douglas, W.H. 1964. An Introduction to the Western Desert Language (Oceania
Linguistic Monographs 4, revised ed.). University of Sydney, Sydney.
Dow, F.D.M. 1972. An Outline of Mandarin Phonetics. Faculty of Asian
Studies, Australian National University, Canberra.
Dul'zon, A.P. 1968. Ketskij Jazyk. Tomsk University, Tomsk.
Dutton, T.E. 1969. The Peopling of Central Papua (Pacific Linguistics,
Series B, 9 ) . Australian National University, Canberra.
Dyen, I. 1971. Malagasy. In T.A. Sebeok (ed.), Current Trends in
Linguistics, Vol. 8, Linguistics in Oceania. Mouton, The Hague: 211-39.
Echeverria, M.S. and Contreras, H. 1965. Araucanian phonemics.
International Journal of American Linguistics 31: 132-5.
Egerod, S. 1966. A statement on Atayal phonology. Artibus Asiae, Supplement
23: 120-30
Emeneau, M.B. 1937. Phonetic observations on the Brahui language. Bulletin
oj[ the School ot_ Oriental Studies 8: 981-3.
Emeneau, M.B. 1944. Kota Texts, Vol. 1. University of California Press,
Berkeley and Los Angeles.
Fast, P.W. 1953. Amuesha (Arawak) phonemes. International Journal of
American Linguistics 19: 191-4.

187
Bibliography

Ferguson, C.A. and Chowdhury, M. I960, The phonemes of Bengali, Language

36: 22-59.
Firchow, I, and Firchow, J. 1969. An abbreviated phoneme inventory.
Anthropological Linguistics 11: 271-6.
Fleming, H.C. 1976. Kefa (Gonga) Languages. In M.L. Bender (ed.), The
Non-Semitic Languages of Ethiopia. African Studies Center, Michigan
State University, East Lansing: 351-76.
Foster, M.L. 1969. The Tarascan Language. University of California Press,
Berkeley and Los Angeles.
Franklin, K. and Franklin, J. 1962. Kewa I: phonological asymmetry.
Anthropological Linguistics 4/7: 29-37.
Frantz, C.I. and Frantz, M.E. 1966. Gadsup phoneme and toneme units.
Pacific Linguistics, Series A, 7: 1-11.
Freeze, R.A. 1975. A Fragment of an Early Kekchi Vocabulary (University of
Missouri Monographs in Anthropology 2 ) . University of Missouri,
Columbia.
F u m e , R. 1963. Arabela phonemes and high-level phonology. Studies in
Peruvian Indian Languages 1. Summer Institute of Linguistics, University
of Oklahoma, Norman: 193-206.
Garbell, I. 1965. The Jewish Neo-Aramaic Dialect of Peralan, Azerbaijan.
Mouton, The Hague.
Garvin, P.L. 1950. Wichita I: phonemics. International Journal of American
Linguistics 16: 179-84.
Gedney, W.J. 1965. Yay, a Northern Tai language in North Vietnam. Lingua
14: 180-93.
Gedney, W.J. 1970. The Saek language of Nakhon Phanom province. Journal of
the Siam Society 58: 67-87.
Gerdel, F. 1973. Paez phonemics. Linguistics 104: 28-48.
Gerzenstein, A. 1968. Fonologia de la lengua Gununa-ke*na (Cuadernos de
Linguistica Indigena 5 ) . Centro de Estudios Linguisticos, University of
Buenos Aires, Buenos Aires.
Gill, H.S. and Gleason, H.A. 1963. A Reference Grammar of Panjabi. Hartford
Seminary Foundation, Hartford.
Glasgow, D. and Glasgow, K. 1967. The phonemes of Burera. Papers in
Australian Linguistics 1 (Pacific Linguistics, Series A, 10): 1-14.
Golla, V.K. 1970. Hupa Grammar. Ph.D. Dissertation. University of
California, Berkeley.
Gowda, K.S.G. 1972. Ao-Naga Phonetic Reader. Central Institute of Indian
Languages, Mysore.
Gregersen, E.A. 1961. Luo: A Grammar. Ph. D. Dissertation. Yale University,
New Haven.
Gregores, E. and Suarez, J.A. 1967. A Description of Colloquial Guarani.
Mouton, The Hague.
Grubb, D.M. 1977. A. Practical Writing System and Short Dictionary o^
Kwakw'ala (Kwakiut1). National Museum of Man, Ottawa.
Gulya, J. 1966. Eastern Ostyak Chrestomathy (Indiana University
Publications, Uralic and Altaic Series 51). Indiana University,
Bloomington.
Gumperz, J.J. and Biligiri, H.S. 1957. Notes on the phonology of Mundari.
Indian Linguistics 17: 6-15.
Haas, M.R. 1941. Tunica. Handbook of American Indian Languages, Vol. 4.
Augustin, New York: 1-143.
Haeseriju, E.V. 1966. Ensayo de la Gramatica del K/ekchl. Suquinay,
Purulha.
Hajdu, P. 1963. The Samoyed Peoples and Languages (Indiana University
Publications, Uralic and Altaic Series 14). Indiana University,
Bloomington.

188
 Bibliography

Hale, K. 1959. A Papago Grammar. Ph.D. Dissertation. Indiana University,

Bloomington.
Hall, R.A. 1938. An Analytical Grammar of the Hungarian Language (Language
Monograph 18). Linguistic Society of America, Philadelphia.
Hall, R.A. 1944. Hungarian Grammar (Language Monograph 21). Linguistic
Society of America, Baltimore.
Halle, M. 1959. The Sound Pattern of Russian. Mouton, The Hague.
Hangin, J.G. 1968. Basic Course in Mongolian (Indiana University
Publications, Uralic and Altaic Series 73). Indiana University,
Bloomington.
(Hanyu Fangyan Gaiyao) 1960. [Outline of Chinese Dialects]. Wenzi Gaige
Chubanshe, Beijing.
Hardman, M.J. 1966. Jaqaru: Outline of Phonological and Morphological
Structure. Mouton, The Hague.
Harms, R.T. 1964. Finnish Structural Sketch (Indiana University
Publications, Uralic and Altaic Series 42). Indiana University,
Bloomington.
Harris, J.T. 1969. Spanish Phonology. MIT Press, Cambridge, Massachusetts.
Hashimoto, M.J. 1973. The Hakka Dialect. Cambridge University Press,
Cambridge.
Hasselbrink, G. 1965. Alternative Analyses of the Phonemic System in
Central-South Lappish (Indiana University Publications, Uralic and
Altaic Series 49). Indiana University, Bloomington.
Haudricourt, A-G. 1967. La langue lakkia. Bulletin de la Societe de
Linguistique de Paris 62; 165-82.
Healey, A. 1964. Telefol Phonology (Pacific Linguistics, Series B, 3 ) .
Australian National University, Canberra.
Heine, B. 1975a. Tepes und Nyangi: zwei ostafrikanische Restsprachen.
Afrika und Ubersee 58: 263-300.
Heine, B. 1975b. Ik - eine ostafrikanische Restsprache. Afrika und Ubersee
59: 31-56.
Henderson, E.J. 1965. Tiddim Chin: A Descriptive Analysis of Two Texts.
Oxford University Press, London.
Herbert, R.J. and Poppe, N. 1963. Kirghiz Manual (Indiana University
Publications, Uralic and Altaic Series 33). Indiana University,
Bloomington.
Hercus, L.A. 1973. The prestopped nasal and lateral consonants in Arabana-
Wanganura. Anthropological Linguistics 14: 293-305.
Hetzron, R. 1969. The Verbal System of Southern Agaw. University of
California Press, Berkeley and Los Angeles.
Hodge, C.T. 1947. An Outline of Hausa Grammar (Language Dissertation 41).
Linguistic Society of America, Baltimore.
Hodge, C.T. and Umaru, I. 1963. Hausa Basic Course. Foreign Service
Institute, Washington D.C.
Hoff, B.J. 1968. The Carib Language (Verhandelingen van het Koninklijk
Instituut voor Taal-, Land- en Volkenkunde 55). Nijhoff, The Hague.
Hoffmann, C. 1963. A Grammar of the Margi Language. Oxford University Press
for International African Institute, London.
Hoftmann, H. 1971. The Structure o£ the Lelemi Language. Enzyklopa*die,
Leipzig.
Hohepa, P.W. 1967. A Profile Generative Grammar of Maori (IJAL Memoir 20,
Indiana University Publications in Anthropology and Linguistics).
Indiana University, Bloomington.
Hoijer, H. 1946. Tonkawa. In H. Hoijer (ed.), Linguistic Structures of
Native America. Wenner-Gren Foundation, New York: 289-311.
Hoijer, H. 1949. An Analytic Dictionary of the Tonkawa Language. University
of California Press, Berkeley and Los Angeles.

189
Bibliography

Hoijer, H. 1972. Tonkawa Texts. University of California Press, Berkeley

and Los Angeles.
Holmer, N.M. 1949. Goajiro (Arawak) I: Phonology. International Journal of
American Linguistics 14: 45-56.
Holzknecht, K.G. 1973. The phonemes of the Adzera language. Pacific
Linguistics, Series A, 38: 1-11.
H o m e , E.C. 1961. Beginning Javanese. Yale University Press, New Haven.
Horton, A.E. 1949. A grammar of Luvale. Witwatersrand University Press,
Johannesburg.
Householder, F.W. 1965. Basic course in Azerbaijani (Indiana University
Publications, Uralic and Altaic Series 45). Indiana University,
Bloomington.
Householder, F.W., Kazazis, K. and Koutsoudas, A. 1964. Reference Grammar
of Literary Dhimotiki (IJAL 30/2 Part II, Publications of the Indiana
University Research Center in Anthropology, Folklore and Linguistics,
31). Indiana University, Bloomington.
Hudson, R.A. 1976. Beja. In M.L. Bender (ed.), The Non-Semitic Languages of
Ethiopia. African Studies Center, Michigan State University, East
Lansing: 97-132.
Huffman, F.E. 1970a. The Cambodian System of Writing and Beginning Reading.
Yale University Press, New Haven.
Huffman, F.E. 1970b. Modern Spoken Cambodian. Yale University Press, New
Haven.
Hughes, E.F. and Leeding, V.J. 1971. The phonemes of Nunggubuyu. Papers on
the Languages of Australian Aboriginals (Australian Aboriginal Studies
38):" 72-81.
Hunter, G.G. and Pike, E.V. 1969. The phonology and tone sandhi of Molinos
Mixtec. Linguistics 47: 24-40.
Hurd, C. and Hurd, P. 1966. Nasioi Language Course. Department of
Information and Extension Services, Port Moresby.
Hyman, L.M. 1973. Notes on the history of Southwestern Mande. Studies in
African Linguistics 4: 183-96.
Jacob, J.M. 1968. Introduction to Cambodian. Oxford University Press,
London.
Jacquot, A. 1962. Notes sur la phonologie du beembe (Congo). Journal of
African Languages 1: 232-42.
Johnson, J.B. 1962. El Idioma Yaqui. Instituto Nacional de Antropologia e
Historia, Mexico.
Johnstone, T.M. 1975. The modern South Arabian languages. Afroasiatic
Linguistics 1: 93-121.
Jones, D. and Ward, D. 1969. The Phonetics of Russian. Cambridge University
Press, Cambridge.
Jones, R.B. 1961. Karen Linguistic Studies. University of California Press,
Berkeley and Los Angeles.
Jorden, E.H. 1963. Beginning Japanese, Part I. Yale University Press, New
Haven.
Kalman, B. 1972. Hungarian historical phonology. In L. Benko and S. Imre
(eds.), The Hungarian Language. Mouton, The Hague.
Kaufman, T. 1971. Tzeltal Phonology and Morphology. University of
California Press, Berkeley and Los Angeles.
Kelkar, A.R. 1968. Studies in Hindi-Urdu I: Introduction and Word
Phonology. Postgraduate and Research Institute, Deccan College, Poona.
Kelkar, A.R. and Trisal, P.N. 1964. Kashmiri word phonology: a first
sketch. Anthropological Linguistics 6/1: 13-22.
Keller, K.C. 1959. The phonemes of Chontal (Mayan). International Journal
of American Linguistics 24: 44-53.

190
 Bibliography

Kelley, G. 1963. Vowel phonemes and external vocalic sandhi in Telugu.

Journal £ £ the American Oriental Society 83: 67-73.
Kennedy, N.M. 1960. Problems of Americans in Mastering the Pronunciation of
Egyptian Arabic. Center for Applied Linguistics, Washington, D.C,
Kensinger, K.M. 1963. The phonological hierarchy of Cashinahua. Studies in
Peruvian Indian Languages 1. Summer Institute of Linguistics, University
of Oklahoma, Norman: 207-17.
Key, M.R. 1968. Comparative Tacanan Phonology. Mouton, The Hague.
Khaidakov, S.M. 1966. Ocherki po Lakskoj Dialektologii. Nauka, Moscow.
Klagstad, H. 1958. The phonemic system of colloquial standard Bulgarian.
Slavic and East European Journal 16: 42-54.
Kleinschmidt, S. 1851. Grammatik der GrBniandischen Sprache. Reimer,
Berlin.
Klingenheben, A. 1966. Deutsch-Amharischer SprachfUhrer. Harrassowitz,
Wiesbaden.
Kondo, V. and Kondo, R. 1967. Guahibo phonemes. Phonemic Systems of
Colombian Languages. Summer Institute of Linguistics, University of
Oklahoma, Norman: 89-98.
Kooyers, 0., Kooyers, M., and Bee, D. 1971. The phonemes of Washkuk
(Kwoma). Te Reo 14: 36-41.
Kostid, D., Mitter, A., and Krishnamurti, B. 1977. A Short Outline of
Telugu Phonetics. Indian Statistical Institute, Calcutta.
Kraft, C.H. and Kraft, M.G. 1973. Introductory Hausa. University of
California Press, Berkeley.
Krejnovich, E.A. 1958. Jukagirskij jazyk. Academy of Sciences of the USSR,
Moscow and Leningrad.
Krejnovich, E.A. 1968a. Ketskij jazyk. In V.V. Vinogradov (ed.), Jazyki
Narodov SSSR, Vol. 5. Nauka, Leningrad and Moscow: 453-73.
Krej"novich, E.A. 1968b. Jukagirskij jazyk. In V.V. Vinogradov (ed.), Jazyki
Narodov SSSR, Vol. 5. Nauka, Leningrad and Moscow: 435-52.
Krishnamurti, B. 1961. Telugu Verbal Bases. University of California Press,
Berkeley and Los Angeles.
Kroeber, A.L. and Grace, G.W. 1960. The Sparkman Grammar of Luiseflo.
University of California Press, Berkeley and Los Angeles.
Krueger, J.R. 1961. Chuvash Manual (Indiana University Publications, Uralic
and Altaic Series 7 ) . Indiana University, Bloomington.
Krueger, J.R. 1962. Yakut Manual (Indiana University Publications, Uralic
and Altaic Series 21). Indiana University, Bloomington.
Kuipers, A.H. 1960. Phoneme and Morpheme in Kabardian. Mouton, The Hague.
Kuipers, A.H. 1967. The Squamish Language. Mouton, The Hague.
Kutnari, B.S. 1972. Malayalam Phonetic Reader. Central Institute of Indian
Languages, Mysore.
Ladefoged, P. 1968. A Phonetic Study of West African Languages, 2nd ed.
Cambridge University Press, Cambridge.
Ladefoged, P. and Traill, A. 1980. The phonetic inadequacy of phonological
specifications of clicks. UCLA Working Papers in Phonetics 49: 1-27.
Langdon, M. 1970. A Grammar of Diegueflo, Mesa Grande Dialect. University of
California Press, Berkeley and Los Angeles.
Lastra, Y. 1968. Cochabamba Quechua Syntax. Mouton, The Hague.
Laycock, D.C. 1965. Three Upper Sepik phonologies. Oceanic Linguistics 4:
113-17.
Lees, R.B. 1961. The Phonology of Modern Standard Turkish. Indiana
University, Bloomington.
Lehtinen, M. 1964. Basic Course in Finnish. Indiana University,
Bloomington.
Leslau, W. 1938. Lexique Socotri. Klincksiek, Paris.

191
Bibliography

Leslau, W. 1968. Amharic Textbook. University of California Press,

Berkeley.
Li, F-K. 1932. A list of Chipewyan stems. International Journal of American
Linguistics 7: 122-51.
Li, F-K. 1933. Chipewyan consonants. Bulletin of the Institute of History
and Philology, Academia Sinica Ts'ai Yuan P'ei Anniversary Volume,
(Supplementary Volume 1): 429-67.
Li, F-K. 1946. Chipewyan. In H. Hoijer (ed.), Linguistic Structures of
Native America. Wenner-Gren Foundation, New York: 394-423.
Li, F-K. 1948. The distribution of initials and tones in the Sui language.
Language 24: 160-7.
Li, F-K. 1977. A Handbook of Comparative Tai. University of Hawaii Press,
Honolulu.
Li, P.J-K. 1973. Rukai Structure (Special Publication 64). Institute of
History and Philology, Academia Sinica, Taipei.
Liccardi, M. and Grimes, J. 1968. Itonama intonation and phonemes.
Linguistics 38: 36-41.
Lindskoog, J.N. and Brend, J.M. 1962. Cayapa phonemics. Studies in
Ecuadorian Indian Languages 1. Summer Institute of Linguistics,
University of Oklahoma, Norman: 31-44.
Lisker, L. 1963. Introduction to Spoken Telugu. American Council of Learned
Societies, New York.
Liu, L. 1964. Ching-p'o-ytl Kai-k'uang. Chung Kuo Yu Wen 5: 407-17.
Loeweke, E. and May, J. 1964. The phonological hierarchy in Fasu.
Anthropological Linguistics 7/5: 89-97.
Lukas, J. 1937. A Study of the Kanuri Language. Oxford University Press,
London.
Lunt, H.G. 1973. Remarks on nasality: the case of Guarani. In S.R. Anderson
(ed.), A Festschrift for Morris Halle. Holt, Rinehart and Winston, New
York: 131-9.
Luvsanvandan, S. 1964. The Khalkha-Mongolian phonemic system. Acta
Orientalia (Academiae Scientiarum Hungaricae) 17: 175-85.
Lydall, J. 1976. Hamer. In M.L. Bender (ed.), The Non-Semitic Languages of
Ethiopia. African Studies Center, Michigan State University, East
Lansing: 393-438.
Lytkin, V.I. 1966. Komi-Zyrjanskij jazyk. In V.V. Vinogradov (ed.), Jazyki
Narodov SSSR, Vol.3. Nauka, Moscow and Leningrad: 281-99.
Mal6cot, A. 1963. Luiseflo, a structural analysis I: Phonology.
International Journal of American Linguistics 29: 89-95.
MacDonald, G.E. 1973. The Teberan Language Family. In K. Franklin (ed.),
The Linguistic Situation in the Gulf District and Adjacent Areas, Papua
New Guinea (Pacific Linguistics, Series C, 26): 113-48.
MacDonald, R.R. and Soenyono, D. 1967. Indonesian Reference Grammar.
Georgetown University Press, Washington, D.C.
Martin, S.E. 1951. Korean Phonemics. Language 27: 519-33.
Martin, S.E. 1952. Morphophonemics of Standard Colloquial Japanese
(Language Dissertation 47). Linguistic Society of America, Baltimore.
Martin, S.E. and Lee, Y-S.C. 1969. Beginning Korean. Yale University Press,
New Haven.
Matisoff, J.A. 1973. The Grammar of Lahu. University of California Press,
Berkeley and Los Angeles.
Mazaudon, M. 1973. Phonologie Tamang (Nepal) (Collection "Tradition Orale"
4 ) . SELAF, Paris.
McAlpin, D.W. 1975. The morphophonology of the Malayalam noun. In H.F.
Schiffman and C. Easton (eds.), Dravidian Phonological Systems.
University of Washington, Seattle: 206-23.

192
 Bibliography

McElhanon, K.A. 1970. Selepet Phonology (Pacific Linguistics, Series B,

14). Australian National University, Canberra.
Metcalfe, C D . 1971. A tentative statement of the Bardi aboriginal
language. Papers on the Languages of Australian Aboriginals (Australian
Aboriginal Studies 38): 82-92.
Mhac An Fhailigh, E. 1968. The Irish of Erris, Co. Mayo. Dublin Institute
for Advanced Studies, Dublin.
Miller, W.R. 1966. Acoma Grammar and Texts. University of California Press,
Berkeley and Los Angeles.
Mitchell, T.F. 1962. Colloquial Arabic. English Universities Press, London.
Montgomery, C. 1970. Problems in the development of an orthography for the
Sebei language of Uganda. Journal of the Language Association of East
Africa 1: 48-55.
Morgenstierne, G. 1945. Notes on Burushaski phonology. Norsk Tidsskrift for
Sprogvidenskap 13: 61-95.
Moshinsky, J. 1974. A Grammar of Southeastern Porno. University of
California Press, Berkeley and Los Angeles.
Moulton, W.G. 1962. The Sounds of English and German. University of Chicago
Press, Chicago.
Murkelinskij, G.B. 1967. Lakskij jazyk. In V.V. Vinogradov (ed.), Jazyki
Narodov SSSR, Vol. 4_. Nauka, Moscow and Leningrad: 489-507.
Naden, A.J. 1973. The Grammar of Bisa. Ph. D. Thesis. School of Oriental
and African Studies, London University, London.
Najlis, E.L. 1966. Lengua Abipona I. Centro de Estudios Linguisticos,
Universidad de Buenos Aires.
Navarro, T.T. 1961. Manual de Pronunciaci6n Espaflola, 10th ed. Consejo
Superior de Investigaciones Cientificas, Madrid.
N'diaye, G. 1970. Structure du Dialecte Basque de Maya. Mouton, The Hague.
Newman, P. 1974. The Kanakuru Language. Cambridge University Press,
Cambridge.
Newman, S. 1965. Zuni Grammar. University of New Mexico Press, Albuquerque.
Newmark, L. 1957. Structural Grammar of Albanian (IJAL 23/4 Part II,
Indiana University Research Center in Anthropology, Folklore and
Linguistics, Publication 4 ) . Indiana University, Bloomington.
Noss, R.B. 1954. An Outline of Siamese Grammar. Ph. D. Dissertation. Yale
University, New Haven.
Noss, R.B. 1964. Thai Reference Grammar. Foreign Service Institute,
Washington, D.C.
Novikova, K.A. 1960. Ocherki Dialektov Evenskogo Jazyka: 01'skij Govor 1.
Academy of Sciences of the USSR, Moscow and Leningrad.
Oates, L.F. 1964. Distribution of phonemes and syllables in Gugu-Yalanji.
Anthropological Linguistics 1: 23-6.
Oates, W. and Oates, L.F. 1964. Gugu-Yalanji and Wik-Munkan Language
Studies (Occasional Papers in Aboriginal Studies 2 ) . Australian
Institute of Aboriginal Studies, Canberra.
Obolensky, S., Panah, K.Y., and Nouri, F.K. 1963. Persian Basic Course.
Center for Applied Linguistics, Washington D.C.
O'Grady, G.N. 1964. Nyangumata Grammar (Oceania Linguistic Monographs 9 ) .
University of Sydney, Sydney.
O'Grady, G.N., Voegelin, C.F. and Voegelin, F.M. 1966. Phonological
diversity. Languages of the World: Indo-Pacific Fascicle 6_
(Anthropological Linguistics 8/2): 56-67.
Okell, J. 1969. A Reference Grammar of Colloquial Burmese. Oxford
University Press, London.
Olmsted, D.L. 1964. A History of Palaihnihan Phonology. University of
California Press, Berkeley and Los Angeles.

193
Bibliography

Olmsted, D.L. 1966, Achumawi Dictionary. University of California Press,

Berkeley and Los Angeles.
Osborn, H. 1948. Amahuaca phonemes. International Journal of American
Linguistics 14: 188-90.
Osborne, C.R. 1974. The Tiwi Language (Australian Aboriginal Studies 55).
Australian Institute for Aboriginal Studies, Canberra.
Ott, W. and Ott, R. 1967. Phonemes of the Ignaciano language. Linguistics
35: 56-60.
Ozanne-Rivierre, F. 1976. Le Iaai, SELAF, Paris.
Palmer, F.R. 1962. The Morphology of the Tigre Noun. Oxford University
Pre s s, London.
Panfilov, V.Z. 1962. Grammatika Nivkhskogo Jazyka, JU Akademia Nauk, Moscow
and Leningrad.
Panfilov, V.Z. 1968. Nivhskij jazyk. In V.V. Vinogradov (ed.), Jazyki
Narodov SSSR, Vol. 5. Nauka, Moscow and Leningrad: 408-34.
Paulian, C. 1975. Le Kukya, Langue Teke du Congo (Bibliotheque de la SELAF
49-50). SELAF, Paris.
Peasgood, E.T. 1972. Carib phonology. In J.E. Grimes (ed.), Languages of
the Guianas. Summer Institute of Linguistics, University of Oklahoma,
Norman: 35-41.
Pence, A. 1966. Kunimaipa phonology: hierarchical levels. Pacific
Linguistics, Series A, 7: 49-97.
Pfeiffer, M. 1972. Elements of Kurux Historical Phonology. E.J. Brill,
Leiden.
Philipp, M. 1974. Phonologie des Deutschen. Kohlhammer, Stuttgart.
Pike, K.L. and Pike, E.V. 1947. Immediate constituents of Mazateco
syllables. International Journal of American Linguistics 13: 78-91.
Pinnow, H-J. 1959. Versuch einer Historischen Lautlehre der Kharia-Sprache.
Harrassowitz, Wiesbaden.
Pinnow, H-J. 1964. Bemerkungen zur Phonetik und Phonemik des Kurukh.
Indo-Iranian Journal 8: 32-59.
Polome, E.C. 1967. Swahili Language Handbook. Center for Applied
Linguistics, Washington, D.C.
Poppe, N. 1964. Bashkir Manual (Indiana University Publications, Uralic and
Altaic Series 36). Indiana University, Bloomington.
Powell, J.V. 1975. Proto-Chimakuan: Materials for a Reconstruction. Working
Papers in Linguistics, University of Hawaii 7/2.
Prasse, K-G. 1972. Manuel de Grammaire Touaregue. Akademisk Forlag,
Copenhagen.
Price, P.D. 1976. Southern Nambiquara phonology. International Journal of
American Linguistics 42: 338-48.
Pride, K. 1965. Chatino Syntax. Summer Institute of Linguistics, University
of Oklahoma, Norman.
Priest, P. 1968. Phonemes of the Siriono Language. Linguistics 41: 102-8.
Prost, A. 1956. La Langue Songay et ses Dialectes (Memoires de l'Institut
francais d'Afrique Noire 47). IFAN, Dakar.
Prost, G.R. 1967. Phonemes of the Chacobo language. Linguistics 35: 61-5.
Pukui, M.K. and Elbert, S.H. 1965. Hawaiian-Eng 1 ish Dictionar~y, 6th ed.
University of Hawaii Press, Honolulu.
Purnell, H.C. 1965. Phonology of a Yao Dialect. Hartford Seminary
Foundation, Hartford.
Rabel, L. 1961. Khasi, a Language of Assam. Louisiana State University
Press, Baton Rouge.
Radcliffe-Brown, A. 1914. Notes on the languages of the Andaman Islands.
Anthropos 9: 36-52.
Rand, E. 1968. The structural phonology of Alabaman, a Muskogean language.
International Journal of American Linguistics 34: 94-103.

194
 Bibliography

Ray, P.S. 1967. Dafla phonology and morphology. Anthropological Linguistics

9: 9-14.
Renck, G.L. 1967. A tentative statement of the phonemes of Yagaria. Papers
in New Guinea Linguistics 6 (Pacific Linguistics, Series A, 12): 19-48.
Renck, G.L. 1975. A Grammar of Yagaria (Pacific Linguistics, Series B, 40).
Australian National University, Canberra.
Rischel, J. 1974. Topics in West Greenlandic Phonology. Akademisk Forlag,
Copenhagen.
Ristinen, E.K. 1960. An East Cheremis Phonology (Indiana University
Publications, Uralic and Altaic Series 1 ) . Indiana University,
Bloomington.
Ristinen, E.K. 1965. On the phonemes of Nenets. Ural-Alt ajLsche Jahrblicher
36: 154-64.
Ristinen, E.K. 1968. Problems concerning vowel length in Nenets. Ural-
Altaische Jahrblicher 40: 22-44.
Robins, R.H. 1953. The phonology of the nasalized verbal forms in
Sundanese. Bulletin of the School of Oriental and African Studies 15:
138-45.
Robins, R.H. and Waterson, N. 1952. Notes on the phonetics of the Georgian
word. Bulletin o£ the School o£ Oriental and African Studies 15: 55-72.
Robinson, J.O.S. 1974. His and hers morphology: the strange case of the
Tarok possessives. Studies in African Linguistics Supplement 6: 201-9.
Rood, D. 1975. The implications of Wichita phonology. Language 51: 315-37.
Ruhlen, M. 1973. Rumanian Phonology. Ph. D. Dissertation. Stanford
University, Palo Alto.
Saint, R. and Pike, K.L. 1962. Auca phonemics. In B. Elson (ed.), Studies
in Ecuadorian Indian Languages 1. Summer Institute of Linguistics,
University of Oklahoma, Norman: 2-30.
Samarin, W.J. 1966. The Gbeya Language. University of California Press,
Berkeley and Los Angeles.
Santandrea, S. 1970. Brief Grammar Outlines of the Yulu and Kara Languages
(Museum Combonianum 25). Nigrizia, Bologna.
Sapir, E. 1923. The phonetics of Haida. International Journal of American
Linguistics 3-4: 143-58.
Sapir, E. and Hoijer, H. 1967. The phonology and morphology of the Navaho
language. University of California Press, Berkeley and Los Angeles.
Sapir, E. and Swadesh, M. 1939. Nootka Texts. Linguistic Society of
America, Philadelphia.
Sapir, E. and Swadesh, M. 1955. Native accounts of Nootka ethnography (IJAL
21/4 Part II, Indiana University Publications in Anthropology, Folklore
and Linguistics). Indiana University, Bloomington.
Sapir, E. and Swadesh, M. 1960. Yana Dictionary. University of California
Press, Berkeley and Los Angeles.
Sapir, J.D. 1965. A Grammar of Diola-Fogny. Cambridge University Press,
Cambridge.
Saporta, S. and Contreras, H. 1962. A Phonological Grammar of Spanish.
University of Washington Press, Seattle.
Sastry, J.V. 1972. Telugu Phonetic Reader. Central Institute of Indian
Languages, Mysore.
Sat, S.C. 1966. Tuvinskij jazyk. In V.V. Vinogradov (ed.), Jazyki Narodov
SSSR, Vol 2. Nauka, Moscow and Leningrad: 387-402.
Sauvageot, S. 1965. Description Synchronique d'un Dialecte Wolof: le Parler
du Dyolof (Memoires de l'Institut Fundamental d'Afrique Noire 73). IFAN,
Dakar.
Sawyer, J.O. 1965. English-Wappo Vocabulary. University of California
Press, Berkeley and Los Angeles.

195
Bibliography

Saxton, D. 1963, Papago phonemes. International Journal of American

Linguistics 29: 29-35.
Sayers, B. and Godfrey, M. 1964. Outline Description of the Alphabet and
Grammar of £ Dialect of Wik-Munkan, Spoken at_ Coen, North Queensland
(Occasional Papers in Aboriginal Studies 2 ) . Australian Institute of
Aboriginal Studies, Canberra.
Schachter, P. and Fromkin, V.A. 1968. A Phonology of Akan;Akuapem,Asante,
Fante (UCLA Working Papers in Phonetics 9 ) . Phonetics Laboratory,
University of California, Los Angeles.
Schachter, P. and Otanes, F.T. 1972. Tagalog Reference Grammar. University
of California Press, Berkeley and Los Angeles.
Schoenhals, A. and Schoenhals, L.C. 1965. Vocabulario Mixe de Totontepec.
Summer Institute of Linguistics, Mexico.
Schuh, R. 1972. Aspects of Ngizim Syntax. Ph. D. Dissertation. University
of California, Los Angeles.
Seiden, W. 1960. Chamorro phonemes. Anthropological Linguistics 2/4: 6-33.
Senn, A. 1966. Handbuch der Litauischen Sprache, Band 1:Grammatik. Winter,
Heidelberg.
Shafeev, D.A. 1964. A Short Grammatical Outline of Pashto (translated and
edited by H.H. Paper). Indiana University, Bloomington.
Sharpe, M. 1972. Alawa Phonology and Grammar (Australian Aboriginal Studies
37). Australian Institute of Aboriginal Studies, Canberra.
Sheldon, S.N. 1974. Some morphophonemic and tone perturbation rules in
Mura-Piraha'. International Journal of_ American Linguistics 40: 279-82.
Shipley, W.F. 1956. The phonemes of Northeastern Maidu. International
Journal of American Linguistics 22: 233-7.
Shipley, W.F. 1964. Maidu Grammar. University of California Press, Berkeley
and Los Angeles.
Silver, S. 1964. Shasta and Karok: a binary comparison. In W. Bright (ed.),
Studies in California Linguistics. University of California Press,
Berkeley and Los Angeles: 170-81.
Simeon, G. 1969. Hokkaido Ainu phonemics. Journal of the American Oriental
Society 89: 751-7.
Skorik, P.I. 1961. Grammatika Chukotskogo Jazyka, 1_. Akademia Nauk, Moscow
and Leningrad.
Skorik, P.I. 1968. Chukotskij jazyk. In V.V. Vinogradov (ed.), Jazyki
Narodov SSSR, Vol. 5. Nauka, Moscow and Leningrad: 248-70.
Smith, K.D. 1968. Laryngealization and de-laryngealization in Sedang
phonemics. Linguistics 38: 52-69.
Snyder, W. 1968. Southern Puget Sound Salish: phonology and morphology.
Sacramento Anthropology Society Papers 8: 2-22.
Snyman, J.W. 1969. An Introduction to the !Xu Language. Balkema, Cape Town.
Snyman, J.W. 1975. Zu|'h5asi Fonologie en Woordeboek. Balkema, Cape Town.
Sommer, B.A. 1969. Kunjen Phonology: Synchronic and Diachronic (Pacific
Linguistics, Series B, 11). Australian National University, Canberra.
Sommerfelt, A. 1964. Consonant clusters or single phonemes in Northern
Irish? In D. Abercrombie, et al. (eds.), In Memory of Daniel Jones.
Longmans, London: 368-73.
Spotts, H. 1953. Vowel harmony and consonant sequences in Mazahua (Otomi).
International Journal of American Linguistics 19: 253-8.
Stahlke, H. 1*971. Topics in Ewe phonology. Ph. D. Dissertation. University
of California, Los Angeles.
Stell, N.N. 1972. Fonologfa de la lengua ax-fcuxiaj (Cuadernos de Linguistica
Indigena 8 ) . Centro de Estudios Linguisticos, University of Buenos
Aires, Buenos Aires.
Sten, H. 1963. Manuel de Phonetique Francaise. Munksgaard, Copenhagen.

196
 Bibliography

Stevenson, R.C. 1957. A survey of the phonetics and grammatical structure

of the Nuba mountain languages IV. Afrika und Ubersee 61: 27-65.
Stewart, J.M. 1967. Tongue root position in Akan vowel harmony. Phonetica
16: 185-204.
Stolte, J. and Stolte, N. 1971. A description of Northern Barasano
phonology. Linguistics 75: 86-92.
Story, G.L. and Naish, C M . 1973. Tlingit Verb Dictionary. Alaska Native
Language Center, University of Alaska, Fairbanks.
Street, J.C. 1963. Khalkha Structure (Indiana University Publications,
Uralic and Altaic Series 24). Indiana University, Bloomington.
Summer, C. 1957. Etude Experimentale de L/Amharique Moderne. University
College Press, Addis Ababa.
Swanton, J.R. 1909. Tlingit Myths and Texts (Bureau of American Ethnology
Bulletin 39). Smithsonian Institution, Washington, D.C.
Swanton, J.R. 1911. Tlingit. In F. Boas (ed.), Handbook of American Indian
Languages, Part 1 (Bureau of American Ethnology Bulletin 40).
Smithsonian Institution, Washington, D . C : 425-559.
Swift, L.B. 1963. A Reference Grammar of Modern Turkish (Indiana University
Publications, Uralic and Altaic Series 19). Indiana University,
Bloomington.
Swift, L.B., Ahaghota, A., and Ugorji, E. 1962. Igbo Basic Course. Foreign
Service Institute, Washington, D.C
Taylor, D. 1955. The phonemes of the Hopkins dialect of Island Carib.
International Journal of American Linguistics 21: 233-41.
Teeter, K.V. 1964. The Wiyot Language. University of California Press,
Berkeley and Los Angeles.
Terescenko, N.M. 1966. Nganasanskij jazyk. In V.V. Vinogradov (ed.), Jazyki
Narodoy SSSR, Vol. 3_. Nauka, Moscow and Leningrad: 438-57.
Ternes, E. 1970. Grammaire Structurale du Breton de 3/lie de Groix. Winter,
Heidelberg.
Thalbitzer, W. 1904. A Phonetical Study of the Eskimo Language (Meddelelser
om gr«$nland 31). Copenhagen.
Thayer, L.J. 1969. A Reconstructed History of the Chari Languages:
Comparative Bonbo-Bagirmi-Sara Segmental Phonology with Evidence from
Arabic Loanwords. Ph. D. Dissertation. University of Illinois, Urbana.
Thayer, L.J. and Thayer, J.E. 1971. Fifty Lessons in Sara-Ngambay.
University of Indiana, Bloomington.
Thompson, E.D. 1976. Nera. In M.L. Bender (ed.), The Non-Semitic Languages
of Ethiopia. African Studies Center, Michigan State University, East
Lansing: 484-94.
Thompson, L.C 1965. A Vietnamese Grammar. University of Washington Press,
Seattle.
Thurman, R.C. 1970. Chuave phonemic statement. Ms.
Tomiche, N. 1964. Le Parler Arabe du Caire. Mouton, Paris and The Hague.
Topping, D.M. 1969. Spoken Chamorro. University of Hawaii Press, Honolulu.
Topping, D.M. 1973. Chamorro Reference Grammar. University of Hawaii Press,
Honolulu.
Tracy, F.V. 1972. Wapishana phonology. In J.E. Grimes (ed.), Languages of
the Guianas. Summer Institute of Linguistics, University of Oklahoma,
Norman: 78-84.
Trager, F.H. 1971. The phonology of Picuris. International Journal of
American Linguistics 37: 29-33.
Trefry, D. 1972. Phonological Considerations of Pawaian (Oceania Linguistic
Monographs 15). University of Sydney, Sydney.
Triulzi, A., Dafallah, A.A., and Bender, M.L. 1976. Berta. In M.L. Bender
(ed.), The Non-Semitic Languages of Ethiopia. African Studies Center,
Michigan State University, East "Lansing: 513-32.

197
Bibliography

Tryon, D.T. 1968. Iai Grammar (Pacific Linguistics, Series B, 8 ) .

Australian National University, Canberra.
Tryon, D.T. 197 0. An Introduction to Maranungku (Northern Australia)
(Pacific Linguistics, Series B, 15). Australian National University,
Canberra.
Tryon, D.T. 1974. Daly Family Languages (Pacific Linguistics, Series C,
32). Australian National University, Canberra.
Tschenkeli, K. 1958. EinfUhrung in die Georgische Sprache I_. Amirani,
Zurich.
Tucker, A.N. and Bryan, M.A. 1966. Linguistic Analyses: the Non-Bantu
Languages of North-Eastern Africa. Oxford University Press for
International African Institute, London.
Tucker, A.N. and Hackett, P.E. 1959. Le Groupe Linguistique Zande (Annales
du Musee Royal de l'Afrique Centrale, Serie in 8, Sciences Humaines,
22). Mus6e Royale de l'Afrique Centrale, Tervuren.
Tucker, A.N. and Mpaayei, J.T.O. 1955. A Maasai Grammar (Publications of
the African Institute, Leyden, 2 ) . Longmans, London.
Tung, T-H. 1964. A Descriptive Study of the Tsou Language, Formosa (Special
Publication 48). Institute of History and Philology, Academia Sinica,
Taipei.
Turton, D. and Bender, M.L. 1976. Mursi. In M.L. Bender (ed.), The
Non-Semitic Languages of Ethiopia. African Studies Center, Michigan
State University, East Lansing: 533-62.
Uldall, E. 1956. Guarani sound system. International Journal of American
Linguistics 20: 341-2.
Van Der Stap, P.A.M. 1966. Outline of Dani Morphology (Verhandelingen van
het Koninklijk Instituut voor Taal-, Land- en Volkenkunde 48). Nijhoff,
The Hague.
Van Der Tuuk, H.N. 1971. A Grammar of Toba Batak. Nijhoff, The Hague.
Van Syoc, W.B. 1959. The Phonology and Morphology of the Sundanese
Language. Ph. D. Dissertation. University of Michigan, Ann Arbor.
Vanvik, A. 1972. A phonetic-phonemic analysis of Standard Eastern
Norwegian. Norwegian Journal of Linguistics 26: 119-64.
Van Wynen, D. and Van Wynen, M.G. 1962. Fonemas Tacana y Modelos de
Acentuaci6n (Notas Lingliisticas de Bolivia 6 ) . Summer Institute of
Linguistics, Cochabamba.
Velayudhan, S. 1971. Vowel Duration in Malayalam. The Dravidian Linguistic
Association of India, Trivandrum.
Verguin, J. 1967. Le Malais: Essai d'Analyse Fonctionelle et Structurale .
Mouton, Paris and The Hague.
Vermeer, H.J. and Sharma, A. 1966. Hindi-Lautlehre. Groos, Heidelberg.
Voegelin, C.F. 1946. Delaware: an Eastern Algonquian language. In H. Hoijer
(ed.), Linguistic Structures of Native America. Wenner-Gren Foundation,
New York, 130-57.
Voegelin, C.F. 1956. Phonemicizing for dialect study with reference to
Hopi. Language 32: 116-35.
Voegelin, C.F. and Voegelin, F.M. 1966. [Andamanese]. Languages of the
World: Indo-Pacific Fascicle 8 (Anthropological Linguistics 8/4): 10-13.
Vogt, H. 1938. Esquisse d'une grammaire du georgien moderne. Norsk
Tidsskrift for Sprogvidenskap 9: 5-114.
Vogt, H. 1958. Structure phonemique du georgien. Norsk Tidsskrift for
Sprogvidenskap 18: 5-90.
Voorhoeve, C.L, 1965. The Flamingo Bay Dialect of the Asmat Language.
Smits, The Hague.
Voorhoeve, C.L. 1971. Miscellaneous notes on languages in West Irian, New
Guinea. Papers in New Guinea Linguistics 14 (Pacific Linguistics, Series
A, 28): 47-114.

198
 Bibliography

Walton, J. and Walton, J. 1967. Phonemes of Muinane. Phonemic Systems of

Colombian Languages. Summer Institute of Linguistics, University of
Oklahoma, Norman: 37-47,
Weimer, H. and Weimer, N. 1972. Yareba phonemes. Te Reo 15: 52-7.
Welmers, W.E. 1946. A Descriptive Grammar of Fanti (Language Dissertation
39). Linguistic Society of America, Philadelphia.
Welmers, W.E. 1950. Notes on two languages of the Senufo group, I: Senadi.
Language 26: 126-46.
Welmers," W.E. 1952. Notes on the structure of Bariba. Language 28: 82-103.
Welmers, W.E. 1962. The phonology of Kpelle. Journal of African Languages
1: 69-93.
West, B. and Welch, B. 1967. Phonemic system of Tucano. Phonemic systems of
Colombian Languages. Summer Institute of Linguistics, University of
Oklahoma, Norman: 11-24.
Wheeler, A. and Wheeler, M. 1962. Siona phonemics. Studies in Ecuadorian
Indian Languages 1. Summer Institute of Linguistics, University of
Oklahoma, Norman: 96-111.
Whiteley, W.H. 1958. A Short Description of Item Categories in Iraqw. East
African Institute of Social Research, Kampala.
Whorf, B.L. 1946. The Hopi language, Toreva dialect. In H. Hoijer (ed.),
Linguistic Structures of Native America. Wenner-Gren Foundation, New
York: 158-83.
Wiering, E. 1974. The indicative verb in DoowaSyaayo. Linguistics 124:
33-56.
Williamson, K. 1967. Songhai Word List (Gao Dialect). Research Notes,
Department of Linguistics and Nigerian Languages, University of Ibadan,
3: 1-34.
Williamson, K. 1969. Igbo. In E. Dunstan (ed.), Twelve Nigerian Languages
Longmans, London, 85-96.
Wilson, D. 1969. Suena phonology. Papers in New Guinea Linguistics 9
(Pacific Linguistics, Series A, 18): 87-93.
Wilson, W.A.A. 1961. An Outline of the Temne Language. School of Oriental
and African Studies, University of London, London.
Wilson, W.A.A. and Bendor-Samuel, J.T. 1969. The phonology of the nominal
in Dagbani. Linguistics 52: 56-82.
Wolff, H. 1959. Subsystem typologies and area linguistics. Anthropological
Linguistics 1/7: 1-88.
Wonderly, W.L. 1951. Zoque II: phonemes and morphophonemes. International
Journal of American Linguistics 17: 105-23.
Woodward, M.F. 1964. Hupa phonemics. In W. Bright (ed.) , Studies in
California Linguistics. University of California Press, Berkeley and Los
Angeles: 199-216.
Wurm, S. 1972a. Languages of Australia and Tasmania. Mouton, The Hague.
Wurm, S. 1972b. Notes on the indication of possession with nouns in Reef
and Santa Cruz Islands languages. Papers in Linguistics of Melanesia 3
(Pacific Linguistics, Series A, 35): 85-113.
Zhirkov, L.I. 1955. Lakskij jazyk: fonetika i morfologija. Akademia Nauk,
Moscow.

199
Appendix B^: Phoneme charts and segment index for UPSID languages
The following pages contain charts showing the phoneme inventory of each of
the carefully selected sample of 317 languages which comprise the UCLA
Phonological Segment Inventory Database (UPSID). It also includes an index
of each segment type which occurs in the database. This index is arranged
according to the phonetic classification of the segments, and includes the
number of languages with each given segment type and a list of the
languages in which it occurs.
The phoneme charts and segment index make available to other users the
basic data of UPSID. With these tools, much of the information in the
database can be manipulated without the use of a computer. For example, the
question "Does /g/ only occur in languages with /k/?" can be answered for
the UPSID sample by using the index to find the list of languages with /g/
and then turning to each relevant chart to see if /k/ occurs, or
cross-checking with the list of languages with /k/ in the index. More
complex co-occurrences of segments can be examined more easily by a study
of the charts.
Publishing the data in this form serves another purpose, that of making
the interpretations of the phoneme inventories adopted in UPSID open to
independent evaluation. Scholars may reach their own conclusions on the
appropriateness of the inventory as it is represented in UPSID.

The charts
Consonants
The consonants of each language are represented by phonetic symbols located
on a chart which is fully labeled for each place of articulation, secondary
articulation, and manner of articulation which occurs in that language.
Places of articulation are arranged along the top of the chart in a
sequence from the front to the back of the oral cavity. These places
correspond to those recognized in the list of variables used in the UPSID
database (Chapter 10). The last place of articulation in this sequence is
glottal. Note that /h/ is not considered to have a glottal place of
articulation, but to have a variable place. It is placed after glottal. In
addition to dental and alveolar places, an unspecified dental or alveolar
column is included for those segments which are not precisely identified as
being one or the other in the source used for the description of a given
language. This column is labeled dental/alveolar.
Double articulations are listed after variable place in a front-to-
back sequence determined by the more forward of the two articulations
concerned. Secondary articulations are placed in the columns following the
primary place. Nasalization, including pre- and post-nasalization, is
considered a secondary articulation. It is listed first among them. Other
secondary articulations follow in a sequence from the front to the back of
the oral cavity, i.e. labialization is first, pharyngealization is last.
Manners of articulation are arranged down the chart in a sequence in
which the primary arrangement is by decreasing degrees of stricture, with
stops first and approximants last. Stops are followed by affricates,
fricatives, nasals, taps, trills, and flaps, then approximants. Within
stops, plosives precede ejective stops and implosives. The Khoisan
languages Nama and !Xu required special handling because of their large
inventories of clicks. The clicks are placed on a separate chart. Sibilant
affricates and fricatives follow nonsibilant ones, and lateral affricates
and fricatives follow both. Lateral approximants precede central ones. A
special category for segments which are simply identified as some kind of
r-sound follows taps, trills and flaps.
Differences in phonation type are nested within this sequence in a way
similar to the way that secondary articulations are nested within the

200
 Segment index

sequence of primary places. However, the sequence of phonation types varies

for different classes of consonants. For plosives, affricates and
fricatives - which are more commonly voiceless - voiceless types are listed
first. For nasals, trills, flaps and approximants - which are more commonly
voiced - voiced types are listed first. More unusual phonation types follow
the plain voiceless or voiced types, so that voiceless aspirated or
laryngealized (glottalized) voiceless categories follow the plain
voiceless, and breathy (murmured) or laryngealized voiced types follow the
plain voiced. Ejective affricates or fricatives follow their pulmonic
central or lateral counterparts.
Each phonemic consonant segment in the language is represented by a
symbol placed at the intersection of the row and column defining its place
and manner. The symbols used are orthodox I.P.A. symbols as far as
possible. However, I.P.A. conventions have been supplemented in some cases,
drawing on conventions used in Preliminaries to Linguistic Phonetics (P.
Ladefoged, University of Chicago Press, 1971) and elsewhere in phonetic
literature, or newly invented for the purpose. The result is a
transcription system in which each symbol, meaning any distinct letter or
combination of letter(s) and diacritic(s), uniquely represents one segment
type. That is to say, each symbol corresponds to one and only one
combination of values for the phonetic variables in the UPSID data file. In
particular, it may be noted that unspecified dental/alveolar segments are
distinguished from alveolar segments in transcription by the use of quote
marks, e.g. /"t"/ vs. /t/. An r-sound of unspecified manner is represented
by a doubled letter r; thus /rr/ represents an unspecified type of r-sound
with alveolar place of articulation, and /"rr"/ represents an unspecified
r-sound of undetermined dental or alveolar place of articulation.

Vowels
The vowels of each language are given below the consonant chart. The vowel
charts are not fully labeled. Instead, the choice of symbols is left to
provide more of the information on the nature of the vowels in question.
However, labels are provided for those levels of the familiar dimension of
vowel height that are represented in the given language, and the symbols
are arranged in a left to right order which corresponds to a front-to-back
dimension. The layout approximates a conventional vowel chart of triangular
shape. Lip position is indicated entirely by choice of symbol. Where
rounded and unrounded vowels occur with the same height and front/back
positions the front unrounded vowel precedes the front rounded vowel and
the back rounded vowel precedes the back unrounded vowel. The pair of
symbols is separated by a comma. In many languages there are several
separate series of vowels (short vs. long, oral vs. nasalized, etc.). These
are given in separate arrangements, with the "plainest" series first.
Transcriptions for vowels are also based on I.P.A. conventions, apart
from a few necessary additional distinctions. Overshort vowels are
indicated with a superscript micron. Mid vowels which are not further
distinguished as higher mid (half-close) or lower mid (half-open) are
represented by letters enclosed in quote marks. The same letters without
the quotes represent higher mid vowels. Thus /e/ is a higher mid front
unrounded voiced vowel, /"e"/ is a similar vowel which is only described as
mid (it might be in the middle of the mid range, or vary in quality within
the mid range, or it might just be incompletely described). Since the
vowels are not fully labeled on the charts a key to the vowel symbols and
the diacritics which may be applied to them is given on page 204.
If a language has any diphthongs which are accepted as unit phonemes by
the criteria used by UPSID, they are given to the right of the simple
vowels. No attempt is made to label the diphthongs; they are simply
represented by digraphs indicating approximate starting and ending points.

201
Segment index

Anomalous segments
Those segments in the inventory of a given language that are regarded as
somewhat anomalous (though still worthy of being included in the inventory)
are indicated by superscript numbers on the charts. These correspond to the
non-zero values of the "anomaly" variable in the UPSID data file which are
defined in Chapter 10.

Abbreviations of phonetic terms used on the charts

Some of the phonetic labels have had to be abbreviated to save space on
the charts. Below is a listing of all the abbreviations used. Many are
self-evident, but in the event that an abbreviation is unclear, its
expansion can be found here.
Abbreviation Expansion
affric, aff., af. affricate
alv. alveolar
approx. appr., app. approximant
asp. aspirated
cent., c. central
cl., c. click
eject. ejective
fric. fricative
labial-vel. labial-velar
laryngd., laryng. laryngealized
lat. lateral
nas. nasal
nonsib. nonsibilant
palato-alv. palato-alveolar
plos. plosive
preasp. preaspirated
rel. release
sib. sibilant
var. variable
vel. velar
vl. voiceless
vd. voiced
w. with

The index
The index to segment types is divided into 10 major sections:
1. plosives
2. glottalic stops
3. clicks (including affricated clicks)
4. affricates (including ejective affricates)
5. fricatives (including ejective fricatives)
6. nasals
7. trills, taps, and flaps
8. approximants
9. vowels
10. diphthongs.
These sections separate segments into major manner classes. Within these
sections, the sequence of segments is ordered in a way that is similar to
the way the charts are arranged, except that, since the index is a list
rather than a two-dimensional array, the values for manner and place
features are nested into a single sequence. Within each section, place is
given priority over manner with the following exceptions:
(a) ejectives precede implosives in section 2.
(b) affricated clicks follow plain clicks in section 3.

202
 Segment index

(c) ejective affricates and fricatives follow all the pulmonic types in
sections 4 and 5.
(d) in section 7, trills precede taps, which precede flaps. Unspecified
r-sounds are included at the end of this section.
(e) lateral segments are grouped together in the relevant sections,
following segments with central articulation in sections 3, 4, 5 and 7, but
preceding central segments in section 8. This arrangement juxtaposes the
common types of flapped and approximant laterals in sections 7 and 8. Note
that the arrangements specified in (b), (c), and (d) above take precedence
over (e).
Voiceless segments of a given place are listed before voiced ones in
sections 1, 2, 3, 4, and 5. Voiced ones precede voiceless ones of the same
place in sections 6, 7, 8, 9, and 10. Other modifications of the phonation
type are next most highly ranked, while secondary articulations (of place)
are the most superficial criterion for sequencing. Thus, a voiceless
aspirated alveolar plosive will precede any voiced alveolar plosives, and a
palatalized voiceless aspirated alveolar plosive follows one with no
secondary articulation but precedes a velarized one.
Vowels are arranged in three main classes: all front vowels are listed
before all central vowels, which in turn precede all back vowels. Within
each height/backness class (e.g. high front or lower mid back), unrounded
vowels precede rounded vowels. This arrangement applies even for back
vowels, where rounding is more common. Other modifications (such as
nasalization, pharyngealization) form the most superficial level of vowel
classification.
Diphthongs are arranged in the following way. First, the quality of the
vowel recognized as their point of origin is used to arrange them in a
similar way to that in which the simple vowels are arranged. Diphthongs
starting with a high front unrounded vowel come first and so on. Those with
the same starting point are then arranged on the basis of the vowel quality
recognized for their ending point. The manner in which this is done can be
understood by imagining an array of lines on the conventional vowel chart
from the point of origin to the point of termination. The diphthongs are
then sequenced in an "anticlockwise" manner with the closest to 12:00
o'clock as the starting point. Note that the transcriptional distinction
between mid vowels (/"e"/, /"o"/ etc.) and higher mid vowels (/e/, /o/
etc.) is not maintained for diphthongs. No quote marks are used here.
Following each phonetic definition, a list of those languages in UPSID
in which the defined segment occurs is given. The listing of language names
is in the order in which identification numbers were assigned, which
corresponds with a few exceptions to the order in which languages appear in
the genetic listing given in Appendix A. If a segment is regarded as
anomalous in a language, the language is listed at the end of the entry for
the segment following a short keyword which describes the nature of the
anomaly. The following are the keywords used in the index to correspond to
the non-zero values of the "anomaly" variable in the UPSID data file.
Value of anomaly Index keyword Brief definition
1 RARE~ " Extremely low lexical frequency
2 LOAN Occurs in unassimilated loans
3 ?ABSTRACT Posited underlying segment
4 7DERIVED Segment possibly derivable from others
5 OBSCURE Particularly vague or contradictory
description.

203
Segment index

Vowel symbols and d i a c r i t i c s used on phoneme charts

Front Central Back

High ', y f, tt U , LU

Lowered high I, Y O , LU

Higher mid e, 0 a, e O, ¥

Mid "e", "0" "a", " e " "o", "

Lower mid e, oe 3, 0, A

Raised low SB, B, D , CL

Low a, a, x), a

Diacritics (exemplified with the vowel /a/);

/a:/ long
/5/ overshort
/a/ laryngealized
/a/ breathy voiced
/a/ voiceless
/a/ nasalized
/ar/ retroflexed
Y
/a / with velar stricture
/|/ fronted
/a/ retracted
/a/ lowered

/a/ raised
/a 9 / pharyngealized

204
 Segment index

SEGMENT INDEX

1. PLOSIVES

Voiceless bilabial plosive /p/ 263 (-9)

Greek, German, Lithuanian, Russian, French, Spanish, Romanian, Pashto,
Kurdish, Hindi-Urdu, Bengali, Kashmiri, Punjabi, Sinhalese, Albanian, E.
Armenian, Ostyak, Cheremis, Komi, Finnish, Hungarian, Lappish, Yurak,
Tavgy, Azerbaijani, Chuvash, Kirghiz, Bashkir, Khalaj, Tuva, Goldi,
Korean, Japanese, Moro, Kadugli, Kpelle, Bisa, Bambara, Dan, Wolof,
Diola, Temne, Dagbani, Senadi, Tampulma, Bariba, Igbo, Birom, Tarok,
Amo, Beembe, Swahili, Luvale, Zulu, Teke, Doayo, Gbeya, Zande, Kanuri,
Fur, Maasai, Luo, Nyangi, Ik, Sebei, Temein, Tabi, Mursi, Logbara, Yulu,
Sara, Koma, Hebrew, Neo-Aramaic, Awiya, Iraqw, Kefa, Hamer, Angas,
Margi, Ngizim, Kanakuru, Mundari, Kharia, Khasi, Sedang, Khmer, Maung,
Tiwi, Burera, Nunggubuyu, Maranungku, Malakmalak, Bardi, Wik-Munkan,
Kunjen, Western Desert, Nyangumata, Aranda, Kariera-Ngarluma,
Gugu-Yalanji, Mabuiag, Arabana-Wanganura, Diyari, Standard Thai, Lakkia,
Yay, Sui, Saek, Po-ai, Lungchow, Atayal, Sundanese, Javanese, Malagasy,
Cham, Malay, Batak, Tagalog, Sa'ban, Chamorro, Rukai, Tsou, Roro,
Kaliai, Iai, Maori, Hawaiian, Mandarin, Taishan, Hakka, Changchow, Amoy,
Fuchow, Kan, Tamang, Dafla, Burmese, Lahu, Jingpho, Ao, Tiddim Chin,
Garo, Boro, Karen, Yao, Andamanese, Asmat, Sentani, Nimboran, Iwam,
Gadsup, Yagaria, Pawaian, Dani, Wantoat, Daribi, Fasu, Suena, Dera,
Kunimaipa, Taoripi, Nasioi, Rotokas, Nambakaengo, Haida, Tlingit,
Navaho, Chipewyan, Tolowa, Nez Perce, Wintu, Chontal, Zoque, Tzeltal,
Totonac, K'ekchi, Mixe, Otomi, Mazahua, Chatino, Nootka, Quileute,
Squamish, Puget Sound, Papago, Luisefio, Hopi, Yacqui, Tiwa, Karok, Pomo,
Diegueflo, Achumawi, Shasta, Tarascan, Zuni, Ojibwa, Delaware, Tonkawa,
Wiyot, Dakota, Yuchi, Alabama, Wappo, Itonama, Bribri, Mura, Cayapa,
Paez, Ocaina, Muinane, Carib, Apinaye, Amahuaca, Chacobo, Tacana,
Cashinahua, Ashuslay, Abipon, S. Nambiquara, Arabela, Auca, Quechua,
Jaqaru, Gununa-Kena, Amuesha, Campa, Guajiro, Moxo, Guarani, Siriono,
Guahibo, Ticuna, Barasano, Siona, Tucano, Jivaro, Cofan, Araucanian,
Greenlandic, Telugu, Kurukh, Malayalam, Yukaghir, Chukchi, Gilyak,
Georgian, Nama, Basque, Burushaski, Ainu, Brahui, !Xu, RARE Manchu,
Island Carib; LOAN Yakut, Evenki, Lelemi, Amharic, Telefol, Mazatec,
Mixtec.

Long voiceless bilabial plosive /p:/ 7

Punjabi, Finnish, Yakut, Japanese, Maranungku, Delaware, Lak,

Palatalized voiceless bilabial plosive /p J / 7 (-1)

Lithuanian, Russian, Yurak, Igbo, Nambakaengo, Amuesha; LOAN Cheremis,

Labialized voiceless bilabial plosive /p / 1

Nambakaengo.

Voiceless aspirated bilabial plosive /p h / 82 (-4)

Breton, Norwegian, Bulgarian, Farsi, Hindi-Urdu, Bengali, Kashmiri,
Punjabi, E. Armenian, Osmanli, Azerbaijani, Kirghiz, Korean, Ewe, Akan,
Igbo, G3, Beembe, Swahili, Zulu, Mundari, Kharia, Khasi, Sedang, Khmer,
Kunjen, Standard Thai, Lakkia, Yay, Sui, Saek, Lungchow, Cham, Adzera,
Mandarin, Taishan, Hakka, Changchow, Amoy, Fuchow, Kan, Tamang, Dafla,
Burmese, Lahu, Jingpho, Tiddim Chin, Karen, Yao, Selepet, Daribi,
Nambakaengo, Klamath, Maidu, Wintu, Otomi, Mazahua, Kwakw'ala, Tiwa,
Yana, Tarascan, Acoma, Wiyot, Dakota, Yuchi, Tunica, S, Nambiquara,

205
Segment index

Quechua, Jaqaru, Wapishana, Cofan, Kota, Gilyak, Georgian, Kabardian,

Lak, Burushaski, !Xu; RARE Haida; LOAN Mongolian, Po-ai, Telugu.

Long voiceless aspirated bilabial plosive /ph:/ 1

Punjabi.

Labialized velarized voiceless aspirated bilabial plosive

Irish.

Palatalized voiceless aspirated bilabial plosive

Irish, Bulgarian, Igbo.

Voiceless preaspirated bilabial plosive / h p / 2 (-1)

Guajiro; OBSCURE Ojibwa.

Voiceless bilabial plosive with breathy release /p / 2

Javanese, Changchow.

Laryngealized voiceless bilabial plosive /p/ 3

Korean, Ashuslay, Siona.

Voiced bilabial plosive /b/ 198 (-11)

Greek, Breton, German, Norwegian, Lithuanian, Russian, Bulgarian,
French, Romanian, Farsi, Pashto, Kurdish, Hindi-Urdu, Bengali, Kashmiri,
Punjabi, Sinhalese, Albanian, Komi, Hungarian, Lappish, Tavgy, Osmanli,
Yakut, Bashkir, Khalaj, Tuva, Mongolian, Evenki, Goldi, Manchu,
Japanese, Katcha, Moro, Kadugli, Kpelle, Bisa, Bambara, Dan, Wolof,
Diola, Temne, Dagbani, Tampulma, Bariba, Ewe, Akan, Igbo, G3, Lelemi,
Efik, Birom, Tarok, Amo, Teke, Doayo, Gbeya, Zande, Songhai, Kanuri,
Maba, Fur, Luo, Nubian, Ik, Tama, Temein, Nera, Tabi, Mursi, Logbara,
Yulu, Sara, Berta, Kunama, Koma, Tigre, Amharic, Hebrew, Socotri,
Neo-Aramaic, Shilha, Tuareg, Somali, Awiya, Iraqw, Beja, Dizi,
Kefa, Hamer, Hausa, Angas, Margi, Ngizim, Kanakuru, Mundari, Kharia,
Khasi, Alawa, Mabuiag, Bandjalang, Standard Thai, Yay, Sui, Saek,
Sundanese, Malagasy, Cham, Malay, Batak, Tagalog, Sa'ban, Chamorro,
Rukai, Tsou, Adzera, Roro, Iai, Amoy, Dafla, Burmese, Lahu, Jingpho,
Tiddim Chin, Garo, Boro, Yao, Andamanese, Nimboran, Telefol, Yagaria,
Chuave, Suena, Dera, Kunimaipa, Yareba, Koiari, Nasioi, Tlingit,
Klamath, Wintu, Tzeltal, Chontal, Otomi, Chatino, Kwakw'ala, Quileute,
Puget Sound, Papago, Yacqui, Porno, Yana, Acoma, Dakota, Yuchi, Alabama,
Itonama, Bribri, Mura, Cayapa, Ocaina, Muinane, Carib, Tacana,
Cashinahua, Auca, Gununa-Kena, Island Carib, Siriono, Guahibo, Ticuna,
Barasano, Tucano, Cofan, Telugu, Kota, Kurukh, Malayalam, Ket, Yukaghir,
Kabardian, Lak, Basque, Burushaski, Brahui, !Xu; RARE Seneca; LOAN
Finnish, Chuvash, Zoque, Mazatec, Tiwa, Tarascan, Tunica, Wappo, Moxo;
OBSCURE Yurak.

Long voiced bilabial plosive /b: / 5

Punjabi, Wolof, Arabic, Shilha, Somali.

Prenasalized voiced bilabial plosive / m b / 18 (-1)

Luvale, Gbeya, Yulu, Sara, Berta, Ngizim, Sedang, Alawa, Hakka, Washkuk,
Selepet, Kewa, Wantoat, Nambakaengo, Paez, Apinaye, Siriono; RARE
Kaliai.

Prenasalized labialized voiced bilabial plosive / m b W / 2

Washkuk, Nambakaengo.

206
 Segment index

,m. -
Nasally-released voiced bilabial plosive / D / 1
Aranda.

Labialized voiced bilabial plosive /b / 1

Irish.

Palatalized voiced bilabial plosive /b J / 6 (-1)

Irish, Lithuanian, Russian, Bulgarian, Igbo; OBSCURE Yurak.

Breathy voiced bilabial plosive /b/ 6 (-1)

Hindi-Urdu, Bengali, Igbo, Mundari, Kharia; LOAN Telugu.

Palatalized breathy voiced bilabial plosive /b / 1

Igbo.

Laryngealized voiced bilabial plosive /b/ 9

Logbara, Ngizim, Sedang, Lakkia, Sui^ Lungchow, K'ekchi, Otomi,
Wapishana.

Voiceless dental plosive /t/ 72

Russian, French, Spanish, Kurdish, Hindi-Urdu, Sinhalese, Albanian,
Finnish, Hungarian, Azerbaijani, Khalaj, Tuva, Manchu, Katcha, Moro,
Kadugli, Temne, Gbeya, Ik, Tama, Temein, Tabi, Logbara, Yulu, Kunama,
Arabic, Tigre, Neo-Aramaic, Tuareg, Beja, Mundari, Kharia, Tiwi,
Nunggubuyu, Kunjen, Western Desert, Aranda, Kariera-Ngarluma,
Arabana-Wanganura, Diyari, Standard Thai, Sundanese, Javanese, Malagasy,
Rukai, Roro, Maori, Tamang, Garo, Yao, Nimboran, Nez Perce, Tzeltal,
Mixe, Nootka, Squamish, Papago, Luisefio, Porno, Dieguetio, Alabama, Wappo,
Gununa-Kena, Guarani, Siona, Araucanian, Greenlandic, Telugu, Malayalam,
Georgian, Nama, Brahui.

Long voiceless dental plosive /t:/ 2

Finnish, Arabic.

Palatalized voiceless dental plosive /£ / 1

Russian.

Pharyngealized voiceless dental plosive /£ / 2

Arabic, Tuareg.

Long pharyngealized voiceless dental plosive /t :/ 1

Arabic.

Voiceless aspirated Rental plosive / t h / 22

Irish, Norwegian, Farsi, Hindi-Urdu, Osmanli, Azerbaijani, Ewe, G2,
Somali, Mundari, Kharia, Kunjen, Standard Thai, Tamang, Yao, Selepet,
Yana, Acoma, Guahibo, Kota, Georgian, Kabardian.

Voiceless dental plosive with breathy release /£ / 1

Javanese.

Voiced dental plosive /d/ 53 (-3)

Irish, Norwegian, Russian, French, Farsi, Kurdish, Hindi-Urdu,
Sinhalese, Albanian, Hungarian, Osmanli, Khalaj, Tuva, Manchu, Katcha,
Kadugli, Ewe, Gbeya, Ik, Tama, Temein, Tabi, Logbara, Yulu, Kunama,
Arabic, Tigre, Neo-Aramaic, Tuareg, Somali, Beja, Mundari, Kharia,
Standard Thai, Malagasy, Rukai, Garo, Yao, Nimboran, Kunimaipa, Mixe,

207
Segment index

Papago, Yana, Acoma, Gununa-Kena, Telugu, Kota, Malayalam, Kabardian,

Brahui; LOAN Tzeltal, Wappo; ?DERIVED Finnish.

Long voiced dental plosive /d:/ 2

Arabic, Somali.

Prenasalized voiced dental plosive /"d/ 2

Gbeya, Selepet.

Nasally-released voiced dental plosive /d"/ 1

Aranda•

Palatalized voiced dental plosive /d J / 1

Russian.

Pharyngealized voiced dental plosive /d / 2

Arabic, Tuareg.
o
Long pharyngealized voiced dental plosive /d :/ 1
Arabic.
Breathy voiced dental plosive /d/ 3
Hindi-Urdu, Mundari, Kharia."

Laryngealized voiced dental plosive /d/ 1

Logbara.

Voiceless dental/alveolar plosive /"t"/ 135

Greek, German, Lithuanian, Romanian, Pashto, Bengali, Kashmiri, Punjabi,
E. Armenian, Ostyak, Cheremis, Lappish, Yurak, Tavgy, Chuvash, Yakut,
Kirghiz, Evenki, Goldi, Korean, Japanese, Wolof, Dagbani, Senadi, Igbo,
Tarok, Beembe, Swahili, Luvale, Doayo, Songhai, Kanuri, Maba, Maasai,
Nubian, Sebei, Nera, Mursi, Arabic, Amharic, Hebrew, Socotri, Shilha,
Awiya, Kullo, Dizi, Kefa, Hausa, Khasi, Sedang, Khmer, Maranungku,
Wik-Munkan, Gugu-Yalanji, Mabuiag, Lakkia, Yay, Sui, Po-ai, Lungchow,
Atayal, Cham, Malay, Batak, Tagalog, Tsou, Kaliai, Iai, Mandarin,
Taishan, Hakka, Changchow, Amoy, Fuchow, Kan, Dafla, Burmese, Lahu,
Jingpho, Karen, Andamanese, Sentani, Telefol, Pawaian, Dani, Wantoat,
Suena, Dera, Kunimaipa, Taoripi, Nasioi, Nambakaengo, Haida, Tlingit,
Chipewyan, Tolowa, Chontal, Mazahua, Mazatec, Mixtec, Chatino, Tiwa,
Karok, Shasta, Zuni, Delaware, Tonkawa, Wiyot, Seneca, Wichita, Dakota,
Yuchi, Itonama, Bribri, Mura, Carib, Amahuaca, Ashuslay, Abipon,
Arabela, Quechua, Island Carib, Guajiro, Siriono, Ticuna, Jivaro, Cofan,
Aleut, Kurukh, Ket, Yukaghir, Chukchi, Gilyak, Basque, Burushaski.

Long voiceless dental/alveolar plosive /"t:"/ 7

Punjabi, Yakut, Japanese, Shilha, Maranungku, Delaware, Lak.

Labialized voiceless dental/alveolar plosive /"t "/ 1

Nambakaengo.

Palatalized voiceless dental/alveolar plosive /"t Jl 7 8 (-2)

Lithuanian, Yurak, Chuvash, Nambakaengo, Itonama, Ket; LOAN Cheremis;
7DERIVED Songhai.
o
Pharyngealized voiceless dental/alveolar plosive /"t / 1
Shilha.

208
 Segment index

Voiceless aspirated dental/alveolar plosive /"t h "/ 48 (-2)

Breton, Bulgarian, Bengali, Kashmiri, Punjabi, E. Armenian, Kirghiz,
Mongolian, Korean, Akan, Igbo, Beerabe, Swahili, Khasi, Sedang, Khmer,
Lakkia, Yay, Lungchow, Cham, Adzera, Mandarin, Taishan, Hakka, Amoy,
Fuchow, Kan, Burmese, Lahu, Jingpho, Karen, Nambakaengo, Haida, Tolowa,
Mazahua, Kwakw'ala, Tiwa, Wiyot, Wichita, Dakota, S. Nambiquara,
Quechua, Cofan, Gilyak, Lak, Burushaski; LOAN Irish, Po-ai.

Long voiceless aspirated dental/alveolar plosive /"t":"/ 1

Punjabi.

Palatalized voiceless aspirated dental/alveolar plosive /"tvJ^1"/ 1

Bulgarian.

Velarized voiceless aspirated dental/alveolar plosive /"th'7 1

Chipewyan.

Preaspirated voiceless dental/alveolar plosive /" h t"/ 1

Guajiro.

Voiceless dental/alveolar plosive with breathy release /"t "/ 1

Changchow.

Laryngealized voiceless dental/alveolar plosive /""t11/ 2

Korean, Ashuslay.

Voiced dental/alveolar plosive /"d'7 77 (-5)

Greek, Breton, German, Lithuanian, Bulgarian, Romanian, Pashto, Bengali,
Kashmiri, Punjabi, Lappish, Tavgy, Yakut, Mongolian, Evenki, Japanese,
Wolof, Dagbani, Senadi, Akan, Igbo, Tarok, Doayo, Songhai, Kanuri, Maba,
Nubian, Nera, Mursi, Berta, Amharic, Hebrew, Socotri, Shilha, Kullo,
Kefa, Hausa, Khasi, Kariera-Ngarluma, Mabuiag, Yay, Sui, Cham, Malay,
Batak, Tagalog, Adzera, Iai, Amoy, Dafla, Burmese, Lahu, Jingpho,
Andamanese, Sentani, Suena, Dera, Tlingit, Chatino, Yuchi, Itonama,
Bribri, Carib, Island Carib, Ticuna, Cofan, Kurukh, Ket, Yukaghir, Lak,
Basque, Burushaski; RARE Chontal; LOAN Irish, Chuvash, Telefol, Tiwa.

Long voiced dental/alveolar plosive /"d:'7 3

Punjabi, Arabic, Shilha.

Prenasalized voiced dental/alveolar plosive /"ndfl/ 9

Luvale, Yulu, Berta, Sedang, Hakka, Wantoat, Nambakaengo, Mazatec,
Siriono.

Long prenasalized voiced dental/alveolar plosive /" d:"/ 1 (-1)

RARE Kaliai.

Prenasalized labialized voiced dental/alveolar plosive /" n d W > 7 1

Nambakaengo.

Palatalized voiced dental/alveolar plosive /"d Jl 7 4 (-1)

Lithuanian, Bulgarian, Ket; 7DERIVED Songhai.
o
Pharyngealized voiced dental/alveolar plosive /"d '7 1
Shilha.

209
Segment index

Breathy voiced dental/alveolar plosive /"d'V 3 (-1)

Bengali, Igbo; LOAN Telugu,

Laryngealized voiced dental/alveolar plosive /"d'V 3

Sedang, Sui, Lungchow.

Voiceless alveolar plosive /"t/ 102 (-1)

Komi, Bashkir, Katcha, Kadugli, Kpelle, Bisa, Bambara, Dan, Diola,
Temne, Tampulma, Bariba, Lelemi, Efik, Birom, Amo, Zulu, Teke, Zande,
Fur, Luo, Nyangi, Temein, Sara, Koma, Iraqw, Hamer, Angas, Margi,
Ngizim, Kanakuru, Vietnamese, Maung, Tiwi, Burera, Nunggubuyu,
Malakmalak, Bardi, Kunjen, Western Desert, Nyangumata, Aranda,
Arabana-Wanganura, Diyari, Saek, Javanese, Sa'ban, Chamorro, Ao, Tiddim
Chin, Boro, Asmat, Washkuk, Iwam, Gadsup, Yagaria, Kewa, Chuave, Daribi,
Fasu, Yareba, Koiari, Rotokas, Navaho, Hupa, Nez Perce, Wintu, Zoque,
Totonac, K'ekchi, Otomi, Quileute, Puget Sound, Hopi, Yacqui, Porno,
Diegueflo, Achumawi, Tarascan, Ojibwa, Wappo, Cayapa, Paez, Muinane,
Apinaye, Chacobo, Tacana, Cashinahua, S. Nambiquara, Auca, Jaqaru,
Amuesha, Campa, Moxo, Guahibo, Barasano, Tucano, Araucanian, Ainu,
Brahui, !Xu; OBSCURE Ojibwa.

Palatalized voiceless alveolar plosive IX I 3

Nyangumata, Paez, Ocaina.

Voiceless aspirated alveolar plosive IXh/ 19

Zulu, Vietnamese, Kunjen, Saek, Changchow, Tiddim Chin, Daribi, Navaho,
Hupa, Klamath, Maidu, Wintu, Otomi, Tarascan, Tunica, Jaqaru, Wapishana,
Kota, !Xu.

Voiceless alveolar plosive with breathy release IX I 1

Javanese.

Voiced alveolar plosive /d/ 65 (-5)

Komi, Bashkir, Katcha, Kadugli, Kpelle, Bisa, Bambara, Dan, Diola,
Temne, Tampulma, Bariba, G3, Efik, Birom, Amo, Teke, Zande, Fur, Luo,
Temein, Sara, Koma, Iraqw, Hamer, Angas, Margi, Kanakuru, Alawa, Diyari,
Bandjalang, Saek, Sundanese, Sa'ban, Chamorro, Tiddim Chin, Boro,
Gadsup, Chuave, Yareba, Koiari, Klamath, Wintu, Otomi, Kwakw'ala,
Quileute, Puget Sound, Porno, Cayapa, Muinane, Tacana, Cashinahua, Auca,
Wapishana, Guahibo, Barasano, Tucano, Kota, Brahui, !Xu; LOAN Zoque,
Yacqui, Tarascan, Tunica, Moxo.

Prenasalized voiced alveolar plosive / d/ 7

Sara, Ngizim, Alawa, Washkuk, Kewa, Paez, Apinaye.

Prenasalized palatalized voiced alveolar plosive / d / 1

Paez.

Nasally-released alveolar plosive Id I 1

Aranda.

Palatalized voiced alveolar plosive /d J / 1

Ocaina.

Velarized voiced alveolar plosive /dV 1

!Xu.

210
 Segment index

Laryngealized voiced alveolar plosive /d/ 3

Ngizim, Otomi, Wapishana.

Voiceless palato-alveolar plosive /t/ 7

Nunggubuyu,, Malakmalak, Bardi, Wik-Munkan, Aranda, Cayapa, Campa.

Voiced palato-alveolar plosive /d/ 2

Alawa, Cayapa.

Prenasalized voiced palato-alveolar plosive / d/ 1

Alawa•

Nasally-released voiced palato-alveolar plosive /d / 1

Aranda.

Voiceless retroflex plosive l\l 28 (-1)

Pashto, Hindi-Urdu, Bengali, Kashmiri, Punjabi, Sinhalese, Moro, Maba,
Beja, Mundari, Kharia, Maung, Nunggubuyu, Bardi, Western Desert,
Nyangumata, Aranda, Kariera-Ngarluma, Arabana-Wanganura, Diyari, Cham,
Rukai, Iai, Telugu, Kurukh, Malayalam, Burushaski; 7DERIVED Tiwi.

Long voiceless retroflex plosive l\:/ 1

Punjabi.

Laryngealized voiceless retroflex plosive IX/ 1

Sionai

Voiceless aspirated retroflex plosive / t h / 10 (-1)

Norwegian, Hindi-Urdu, Bengali, Kashmiri, Punjabi, Kharia, Cham, Kota,
Burushaski; LOAN Telugu.

Long voiceless aspirated retroflex plosive /th :/ 1

Punjabi.

Voiced retroflex plosive /d/ 23

Norwegian, Pashto, Hindi-Urdu, Bengali, Kashmiri, Punjabi, Sinhalese,
Lelemi, Maba, Awiya, Beja, Mundari, Kharia, Alawa, Diyari, Rukai, Iai,
Papago, Telugu, Kota, Kurukh, Malayalam, Burushaski.

Long voiced retroflex plosive /d:/ 1

Punjabi.

Prenasalized voiced retroflex plosive /*d/ 1

Alawa.

Nasally-released voiced retroflex plosive /d*/ 1

Aranda.

Breathy voiced retroflex plosive /d/ 5 (-1)

Hindi-Urdu, Bengali, Mundari, Knaria; LOAN Telugu.

Laryngealized voiced retroflex plosive /d/ 1

Somali.

Long laryngealized voiced retroflex plosive /d:/ 1

Somali.

211
Segment index

Voiceless palatal plosive /c/ 41 (-2)

Ostyak, Komi, Tavgy, Azerbaijani, Katcha, Kadugli, Wolof, Diola, Senadi,
Tampulma, Birom, Nyangi, Sebei, Tabi, Mursi, Yulu, Hamer, Angas, Margi,
Ngizim, Vietnamese, Burera, Kunjen, Kariera-Ngarluma, Gugu-Yalanji,
Arabana-Wanganura, Diyari, Yay, Cham, Malay, Yao, Kewa, Haida, Hupa,
Muinane, Jaqaru, Kurukh, Gilyak, Basque; RARE Karen, OBSCURE Cofan.

Voiceless aspirated palatal plosive / c V 16 (-3)

Breton, Azerbaijani, Kunjen, Yay, Cham, Yao, Haida, Hupa, Klamath,
Maidu, Kwakw'ala, Acoma, Jaqaru; RARE Karen; LOAN Osmanli; OBSCURE
Cofan.

Voiced palatal plosive / j / 31 (-2)

Breton, Komi, Tavgy, Katcha, Wolof, Diola, Senadi, Tampulma, Birom,
Maba, Tama, Temein, Tabi, Mursi, Tuareg, Hamer, Angas, Margi, Ngizim,
Sundanese, Malay, Yao, Klamath, Kwakw'ala, Acoma, Muinane, Kurukh,
Yukaghir, Basque; LOAN Osmanli; OBSCURE Cofan.

Long voiced palatal plosive /j:/ 1

Ngizim.

Prena sali zed voiced palatal plosive / j/ 3

Yulu, Sara, Apinaye.

Voiceless velar plosive /k/ 283 (-3)

Greek, German, Lithuanian, Russian, French, Spanish, Romanian, Pashto,
Kurdish, Hindi-Urdu, Bengali, Kashmiri, Punjabi, Sinhalese, Albanian, E.
Armenian, Ostyak, Cheremis, Komi, Finnish, Hungarian, Lappish, Yurak,
Tavgy, Chuvash, Yakut, Bashkir, Khalaj, Tuva, Evenki, Goldi, Manchu,
Korean, Japanese, Katcha, Moro, Kadugli, Kpelle, Bisa, Bambara, Dan,
Wolof, Diola, Temne, Dagbani, Senadi, Tampulma, Bariba, Igbo, Lelemi,
Efik, Birom, Tarok, Amo, Swahili, Luvale, Zulu, Teke, Doayo, Gbeya,
Zande, Songhai, Kanuri, Maba, Fur, Maasai, Luo, Nubian, Nyangi, Ik,
Sebei, Tama, Temein, Nera, Tabi, Mursi, Logbara, Yulu, Sara, Kunama,
Koma, Arabic, Amharic, Tigre, Hebrew, Socotri, Neo-Aramaic, Shilha,
Tuareg, Awiya, Iraqw, Beja, Kullo, Dizi, Kefa, Hausa, Angas, Margi,
Ngizim, Kanakuru, Mundari, Kharia, Khasi, Vietnamese, Sedang, Khmer,
Maung, Tiwi, Burera, Nunggubuyu, Maranungku, Malakmalak, Bardi,
Wik-Munkan, Kunjen, Western Desert, Nyangumata, Aranda,
Kariera-Ngarluma, Gugu-Yalanji, Mabuiag, Arabana-Wanganura, Diyari,
Standard Thai, Lakkia, Yay, Sui, Saek, Po-ai, Lungchow, Atayal,
Sundanese, Javanese, Malagasy, Cham, Malay, Batak, Tagalog, Chamorro,
Rukai, Tsou, Roro, Kaliai, Iai, Maori, Hawaiian, Mandarin, Taishan,
Hakka, Changchow, Amoy, Fuchow, Kan, Tamang, Dafla, Burmese, Lahu,
Jingpho, Ao, Tiddim Chin, Garo, Boro, Karen, Yao, Andamanese, Asmat,
Washkuk, Sentani, Nimboran, Iwam, Telefol, Gadsup, Yagaria, Chuave,
Pawaian, Dani, Wantoat, Daribi, Fasu, Suena, Dera, Kunimaipa, Yareba,
Koiari, Taoripi, Nasioi, Rotokas, Nambakaengo, Haida, Tlingit, Navaho,
Chipewyan, Tolowa, Nez Perce, Wintu, Chontal, Zoque, Totonac, K'ekchi,
Mixe, Otomi, Mazahua, Mazatec, Mixtec, Chatino, Nootka, Quileute, Puget
Sound, Papago, Luiseflo, Hopi, Yacqui, Tiwa, Karok, Porno, Diegueflo,
Achumawi, Shasta, Tarascan, Ojibwa, Delaware, Tonkawa, Wiyot, Seneca,
Wichita, Dakota, Yuchi, Alabama, Wappo, Itonama, Bribri, Mura, Cayapa,
Paez, Ocaina, Muinane, Carib, Apinaye, Amahuaca, Chacobo, Tacana,
Cashinahua, Ashuslay, Abipon, S. Nambiquara, Arabela, Auca, Quechua,
Jaqaru, Gununa-Kena, Amuesha, Campa, Guajiro, Moxo, Guarani, Siriono,
Guahibo, Ticuna, Barasano, Siona, Tucano, Jivaro, Cofan, Araucanian,

212
 Segment index

Greenlandic, Aleut, Telugu, Kurukh, Malayalam, Ket, Yukaghir, Chukchi,

Gilyak, Georgian, Nama, Basque, Burushaski, Ainu, Brahui, !Xu; RARE
Squamish, Island Carib; LOAN Azerbaijani.

Long voiceless velar plosive /k:/ 9

Punjabi, Finnish, Yakut, Japanese, Arabic, Shilha, Maranungku, Delaware,
Lak.

Labialized voiceless velar plosive /k W / 38 (-1)

Kpelle, Igbo, Amharic, Awiya, Iraqw, Beja, Hausa, Ngizim, Lakkia,
Taishan, Washkuk, Telefol, Dani, Wantoat, Nambakaengo, Haida, Tlingit,
Chipewyan, Tolowa, Mazahua, Mixtec, Nootka, Quileute, Squamish, Puget
Sound, Luiseflo, Hopi, Tiwa, Dieguefio, Tarascan, Tonkawa, Wiyot, Wichita,
S. Nambiquara, Guarani, Ticuna, Siona; 7DERIVED Otomi.

Long labialized voiceless velar plosive /k W :/ 1

Lak.

Palatalized voiceless velar plosive /k*V 5

Lithuanian, Russian, Hausa, Nambakaengo, Siriono.

Pharyngealized voiceless velar plosive /k / 1

Shilha.

Voiceless aspirated velar plosive /k h / 79 (-3)

Irish, Breton, Norwegian, Bulgarian, Farsi, Hindi-Urdu, Bengali,
Kashmiri, Punjabi, E. Armenian, Osmanli, Azerbaijani, Korean, Ewe, Akan,
Igbo, G3, Beembe, Swahili, Zulu, Somali, Mundari, Kharia, Khasi, Sedang,
Khmer, Kunjen, Standard Thai, Lakkia, Yay, Saek, Lungchow, Cham, Adzera,
Mandarin, Taishan, Hakka, Changchow, Amoy, Fuchow, Kan, Tamang, Burmese,
Lahu, Jingpho, Karen, Yao, Selepet, Daribi, Nambakaengo, Haida, Navaho,
Klamath, Maidu, Tzeltal, Otomi, Mazahua, Yana, Tarascan, Acoma, Wiyot,
Wichita, Dakota, Yuchi, Tunica, S. Nambiquara, Quechua, Jaqaru,
Wapishana, Cofan, Kota, Gilyak, Georgian, Lak, Burushaski, !Xu; LOAN
Mongolian, Po-ai, Telugu.

Long voiceless aspirated velar plosive /k h :/ 1

Punjabi.

Labialized voiceless aspirated velar plosive /k h / 14 (-1)

Igbo, Lakkia, Taishan, Haida, Mazahua, Kwakw'ala, Tarascan, Zuni, Wiyot,
Wichita, S. Nambiquara, Kabardian, Lak; RARE Navaho.

Palatalized voiceless aspirated velar plosive /k J h / 4

Irish, Bulgarian, Lakkia, Kabardian.

Preaspirated voiceless velar plosive /^k/ 2 (-1)

Guajiro; OBSCURE Ojibwa.

Voiceless velar plosive with breathy release /k /

Javanese, Changchow.

Laryngealized voiceless velar plosive /k/ 3

Korean, Ashuslay, Siona.

213
Segment index

Voiced velar plosive /q/ 175 (-14)

Greek, Irish, Breton, German, Norwegian, Lithuanian, Russian, Bulgarian,
French, Romanian, Farsi, Pashto, Kurdish, Hindi-Urdu, Bengali, Kashmiri,
Punjabi, Sinhalese, Albanian, Komi, Hungarian, Lappish, Tavgy, Osmanli,
Yakut, Bashkir, Khalaj, Tuva, Mongolian, Evenki, Goldi, Manchu, Katcha,
Moro, Kadugli, Kpelle, Bisa, Bambara, Dan, Diola, Temne, Dagbani,
Senadi, Tampulma, Bariba, Ewe, Akan, Igbo, GS, Lelemi, Birom, Tarok,
Amo, Teke, Doayo, Gbeya, Zande, Songhai, Kanuri, Maba, Fur, Luo, Nubian,
Ik, Tama, Temein, Nera, Tabi, Mursi, Logbara, Yulu, Sara, Berta, Kunama,
Koma, Tigre, Amharic, Hebrew, Socotri, Neo-Aramaic, Shilha, Tuareg,
Somali, Awiya, Iraqw, Beja, Kullo, Dizi, Kefa, Hamer, Hausa, Angas,
Margi, Ngizim, Kanakuru, Mundari, Kharia, Alawa, Mabuiag, Bandjalang,
Sundanese, Malagasy, Malay, Batak, Tagalog, Chamorro, Rukai, Adzera,
Iai, Amoy, Dafla, Burmese, Lahu, Jingpho, Tiddim Chin, Garo, Boro, Yao,
Andamanese, Nimboran, Yagaria, Kewa, Chuave, Suena, Dera, Kunimaipa,
Yareba, Koiari, Rotokas, Tlingit, Klamath, Mixe, Otomi, Mazahua,
Chatino, Papago, Yana, Acoma, Yuchi, Bribri, Mura, Cayapa, Ocaina,
Muinane, Carib, Auca, Wapishana, Island Carib, Ticuna, Barasano, Tucano,
Cofan, Telugu, Kota, Kurukh, Yukaghir, Lak, Basque, Burushaski, Brahui,
!Xu; RARE Puget Sound; LOAN Finnish, Chuvash, Sa'ban, Telefol, Zoque,
Tzeltal, Quileute, Yacqui, Tiwa, Tarascan, Tunica, Wappo; 7DERIVED
Japanese.

Long voiced velar plosive /q:/ 4

Punjabi, Arabic, Shilha, Somali.

Prenasalized voiced velar plosive /^g/ 18 (-1)

Luvale, Gbeya, Yulu, Sara, Berta, Ngizim, Sedang, Alawa, Hakka, Washkuk,
Selepet, Wantoat, Nambakaengo, Mazatec, Paez, Apinaye, Siriono; RARE
Kaliai.

Labialized prenasalized voiced velar plosive /^gW/ 4

Ngizim, Washkuk, Wantoat, Nambakaengo.

Nasally-released voiced velar plosive /g'V 1

Aranda.

Labialized voiced velar plosive /n W / 14 (_i)

Kpelle, Igbo, Amharic, Awiya, iraqw, Beja, Hausa, Ngizim, Tlingit,
Mazahua, Kwakw'ala, Puget Sound, Kabardian; 7DERIVED Otomi.

Palatalized voiced velar plosive j'gJ/ 5

Irish, Lithuanian, Bulgarian, Hausa, Kabardian.

Breathy voiced velar plosive /g/ 7 (-1)

Hindi-Urdu, Bengali, Igbo, Mundari, Kharia, !Xu; LOAN Telugu.

Labialized breathy voiced velar plosive /g / 1

Igbo.

Voiceless uvular plosive /q/ 38 (-2)

Kurdish, Kirghiz, Arabic, Neo-Aramaic, Tuareg, Awiya, Iraqw, Hamer, Sui,
Atayal, Lahu, Haida, Tlingit, Hupa, Wintu, Totonac, K'ekchi, Nootka,
Quileute, Squamish, Puget Sound, Luisefio, Hopi, Pomo, Achumawi, Abipon,
Quechua, Jaqaru, Gununa-Kena, Greenlandic, Aleut, Ket, Yukaghir,
Chukchi, Gilyak, Burushaski; RARE Diegueflo; LOAN Hindi-Urdu.

214
 Segment index

Labialized voiceless uvular plosive /q / 9 (-1)

Awiya, Iraqw, Haida, Tlingit, Quileute, Squamish, Puget Sound, Luisefio;
LOAN Nootka.

Long labialized voiceless uvular plosive /q : / 1

"~Lak.

Voiceless aspirated uvular plosive /q h / 11

Kirghiz, Sui, Lahu, Haida, Klamath, Kwakw'ala, Quechua, Jaqaru, Gilyak,
Lak, Burushaski•

Labialized voiceless aspirated uvular plosive /q V 3

Haida, Kwakw'ala, Lak.

Voicecl uvular plosive / G / 8

Farsi, Somali, Awiya, Kunimaipa, Tlingit, Klamath, Kwakw'ala, Lak.

Long voiced uvular plosive / G : / 1

Somali.

Labialized voiced uvular plosive /G / 4

Awiya, Tlingit, Kwakw'ala, Lak.

Voiced pharyngeal plosive /*?/ 1

Iraqw.

Glottal plosive /?/ 146 (-11)

Farsi, Kurdish, Yurak, Tavgy, Wolof, Kanuri, Maba, Luo, Tabi, Logbara,
Berta, Koma, Arabic, Tigre, Hebrew, Socotri, Neo-Aramaic, Somali, Iraqw,
Beja, Kullo, Kefa, Hausa, Margi, Khasi, Vietnamese, Sedang, Khmer,
Wik-Munkan, Standard Thai, Lakkia, Yay, Sui, Saek, Po-ai, Lungchow,
Atayal, Sundanese, Javanese, Cham, Tagalog, Sa'ban, Chamorro, Rukai,
Tsou, Adzera, Roro, Hawaiian, Changchow, Amoy, Burmese, Jingpho, Ao,
Tiddim Chin, Garo, Karen, Yao, Washkuk, Gadsup, Yagaria, Dani, Nasioi,
Haida, Tlingit, Chipewyan, Tolowa, Hupa, Nez Perce, Klamath, Wintu,
Chontal, Zoque, Tzeltal, Totonac, K'ekchi, Mixe, Otomi, Mazahua,
Mazatec, Mixtec, Chatino, Nootka, Kwakw'ala, Quileute, Squamish, Puget
Sound, Papago, Luiseflo, Hopi, Yacqui, Tiwa, Karok, Porno, Diegueflo,
Achumawi, Yana, Shasta, Zuni, Acoma, Ojibwa, Tonkawa, Wiyot, Seneca,
Wichita, Yuchi, Tunica, Wappo, Itonama, Mura, Cayapa, Paez, Ocaina,
Muinane, Apinaye, Chacobo, Tacana, Ashuslay, S. Nambiquara, Gununa-Kena,
Wapishana, Amuesha, Guajiro, Moxo, Guarani, Ticuna, Siona, Tucano,
Cofan, Kurukh, Ket, Chukchi, Kabardian, Lak, Nama, Brahui; RARE Temne,
Gbeya, Lahu, Tarascan, Jivaro; LOAN Hindi-Urdu, Osmanli, Bashkir;
7DERIVED Mundari, Navaho, Carib.

Long glottal plosive / ? : / 1

Arabic.

Labialized glottal plosive / ? W / 1

Kabardian.
o
Pharyngealized glottal plosive /? / 1

Nootka.

Voiceless labial-velar

215
Segment index

Kpelle, Dan, Dagbani, Senadi, Tampulma, Bariba, Ewe, G2, Lelemi, Efik,
Birom, Tarok, Amo, Doayo, Gbeya, Zande, Logtfara, Yulu.

Voiced labial-velar plosive /cJB/ 19

Kpelle, Dan, Temne, Dagbani, Senadi, Tampulma, Bariba, Ewe, G3, Lelemi,
Birom, Tarok, Amo, Doayo, Gbeya, Zande, Logbara, Yulu, Iai,

Prenasalized voiced labial-velar plosive / gB/ 1

Gbeya.

Voiceless dental-palatal plosive /fc/ 1

Maung.
2
- GLOTTAL1C STOPS
(Note that 'voiced ejectives' in !Xu are prevoiced, the release is
voiceless.)

Voiceless bilabial ejective stop / p V 33 (-1)

E. Armenian, Zulu, Berta, Koma, Kefa, Tlingit, Nez Perce, Klamath,
Maidu, Wintu, Chontal, Tzeltal, Otomi, Nootka, Kwakw'ala, Quileute,
Squamish, Puget Sound, Tiwa, Porno, Yana, Shasta, Acoma, Dakota, Wappo,
S. Nambiquara, Quechua, Jaqaru, Gununa-Kena, Georgian, Kabardian, Lak;
LOAN Amharic.

Voiced bilabial ejective stop / b V 1

!Xu.

Voiceless dental ejective stop /t'/ 11

Nez Perce, Tzeltal, Nootka, Squamish, Porno, Yana, Acoma, Wappo,
Gununa-Kena, Georgian, Kabardian.

Voiceless dental/alveolar ejective stop /"t"7 21

E. Armenian, Tigre, Amharic, Socotri, Kullo, Dizi, Kefa, Haida, Tlingit,
Chipewyan, Tolowa, Chontal, Mazahua, Kwakw'ala, Tiwa, Shasta, Dakota,
Yuchi, Itonama, Quechua, Lak.

Voiceless alveolar ejective,stop / t 5 / 17

Zulu, Koma, Navaho, Hupa, Nez Perce, Klamath, Maidu, Wintu, K'ekchi,
Otomi, Quileute, Puget Sound, Porno, Wappo, S. Nambiquara, Jaqaru, !Xu.

Voiced alveolar ejective stop /d'/ 1

!Xu.

Voiceless palatal ejective stop /c'/ 7

Haida, Hupa, Klamath, Maidu, Kwakw'ala, Acoma, Jaqaru.

Voiceless velar ejective stop / k V 49 (-1)

E. Armenian, Zulu, Ik, Berta, Koma, Tigre, Amharic, Socotri, Kullo,
Dizi, Kefa, Hamer, Hausa, Haida, Tlingit, Navaho, Chipewyan, Tolowa, Nez
Perce, Klamath, Maidu, Wintu, Chontal, Tzeltal, K'ekchi, Otomi, Mazahua,
Nootka, Quileute, Puget Sound, Tiwa, Porno, Yana, Shasta, Zuni, Acoma,
Wichita, Dakota, Yuchi, Wappo, Itonama, S. Nambiquara, Quechua, Jaqaru,
Gununa-Kena, Georgian, Lak, !Xu; RARE Squamish.
w
Labialized voiceless velar ejective stop /k '/ 18 (-2)
Amharic, Hausa, Haida, Tlingit, Chipewyan, Tolowa, Mazahua, Nootka,
Kwakw'ala, Quileute, Squamish, Puget Sound, Zuni, S. Nambiquara,
Kabardian, Lak; RARE Tiwa; ?DERIVED Otomi.

216
 Segment index

Palatalized voiceless velar ejective stop / k ' / 2

Hausa, Kabardian.

Voiceless uvular ejective stop / q V 19 (-1)

Kefa, Haida, Tlingit, Hupa, Nez Perce, Klamath, Wintu, K'ekchi,
Kwakw'ala, Quileute, Squamish, Puget Sound, Porno, Quechua, Jaqaru,
Georgian, Kabardian, Lak; LOAN Nootka.
w
Labialized voiceless uvular ejective stop /q '/ 8
Haida, Tlingit, Nootka, Quileute, Squamish, Puget Sound, Kabardian, Lak,

Voiceless bilabial implosive /p < / 1

Igbo.

Voiced bilabial implosive 11/ 30 (-1)

Katcha, Kadugli, Kpelle, Dan, Igbo, Tarok, Swahili, Zulu, Doayo, Gbeya,
Maasai, Nyangi, Ik, Tama, Mursi, Yulu, Sara, Koma, Hamer, Hausa, Angas,
Margi, Kanakuru, Vietnamese, Khmer, Karen, Maidu, Mazahua, S.
Nambiquara; RARE Iraqw.

Voiced dental implosive /d1/ 3

Kadugli, Gbeya, Ik.

Voiced dental/alveolar implosive /"cf7 13

Katcha, Tarok, Swahili, Doayo, Maasai, Mursi, Yulu, Berta, Kullo, Hausa,
Karen, Mazahua, S. Nambiquara,

Voiceless alveolar implosive /t* / 1

Igbo,

Voiced alveolar implosive /J/ 13 (-1)

Dan, Nyangi, Tama, Sara, Koma, Hamer, Angas, Margi, Kanakuru,
Vietnamese, Khmer, Maidu; RARE Iraqw.

Voiced palatal implosive /J/ 7

Kadugli, Swahili, Maasai, Nyangi, Ik, Yulu, Angas.

Voiced velar implosive /cf/ 5

Swahili, Maasai, Nyangi, Ik, Hamer.

Voiced uvular implosive /GK / 1

Ik.

3. CLICKS
(Including affricated clicks. Note that clicks may be affricated in two
ways. The front release may be affricated or the back (velar) release may
be affricated. UPSID refers only to the former as affricated clicks. A
fricated release of the back closure is referred to as velarization of the
click.)

Voiceless alveolar click l\I 2

Nama, !Xu.

217
 Segment index

Velarized voiceless alveolar click /\/ \

!Xu.

Voiceless aspirated alveolar click

!Xu.

Nasalized voiceless aspirated alveolar click /0

Nama, !Xu.

Velarized voiceless aspirated alveolar click

Nama.

Glottalized nasalized voiceless alveolar click /^\ / 2

Nama, !Xu.
7
Glottalized velarized voiceless alveolar click /$ / 1
!Xu.

Voiced alveolar click /g\/ 1

!Xu.

Nasalized voiced alveolar click /Q V 2

Nama, !Xu.

Velarized voiced alveolar click /g V 1

!Xu.

Breathy voiced alveolar click /g V 1

!Xu.

Nasalized breathy voiced alveolar click / Q V 1

!Xu.
7
Glottalized velarized voiced alveolar click /g\ / 1
!Xu.

Voiceless palato-alveolar click /]•/ 2

Zulu, Nama.

Nasalized voiceless aspirated palato-alveolar click /^Ih/ 1

Nama.

Velarized voiceless aspirated palato-alveolar click /\^/ 1

Nama.

Glottalized nasalized voiceless palato-alveolar click /^\_ / 1

Nama.

Nasalized voiced palato-alveolar click

Nama.

Voiceless palatal click /[/ 1

!Xu.

218
 Segment index

Velarized voiceless palatal click /£/ 1

!Xu.

Voiceless aspirated palatal click

!Xu.

Nasalized voiceless aspirated palatal click / o fh/ 1

!Xu.

Glottalized nasalized voiceless palatal click /o[ / 1

!Xu.

Glottalized nasalized velarized voiceless palatal click / ^ / 1

!Xu.

Voiced palatal click /gC/ 1

!Xu.

Nasalized voiced palatal click / Q C / 1

!Xu.

Velarized voiced palatal click /g C/ 1

!Xu.

Breathy voiced palatal click

!Xu.

Glottalized velarized voiced palatal click /g{ / 1

!Xu.

Nasalized breathy voiced palatal click / Q [ / 1

!Xu.

Voiceless dental affricated click /lS/ 2

Nama, !Xu.

Velarized voiceless dental affricated click /^S/ 1

!Xu.
h
Voiceless aspirated dental affricated click A / 1
!Xu.

Nasalized voiceless aspirated dental affricated click / o i h / 2

Nama, Xu.

Velarized voiceless aspirated dental affricated click /j;h/ 1

Nama.

Glottalized nasalized voiceless dental affricated click /0} / 1

Nama, !Xu•

Glottalized nasalized velarized voiceless dental affricated click /o* / 1

!Xu.

Voiced dental affricated click

!Xu.

219
Segment index

Velarized voiced dental affricated click /g*S/ 1

!XB.

Nasalized voiced dental affricated click / Q ^ S / 2

Nama, !Xu.

Glottalized velarized voiced dental affricated click

!Xu.

Nasalized breathy voiced dental affricated click /Q-1-S/ 1

!XQ.

Breathy voiced dental affricated click /g^S/ 1

!xa.
Voiceless alveolar affricated click / 1 S / 1
Zulu.

Voiceless palatal lateral affricated click /£ / 1

!Xu.

Velarized voiceless palatal lateral affricated click

!Xu.

Voiceless aspirated palatal lateral affricated click

!XQ.

Glottalized nasalized voiceless palatal lateral affricated click

!Xu\: °

Glottalized nasalized velarized voiceless palatal lateral affricated click

!XB.

Nasalized voiceless aspirated palatal lateral affricated click I IJ*h/

!xn.
Voiced palatal lateral affricated click /g£ / 1
!Xu.

Velarized voiced palatal lateral affricated click /g& / 1

!X8«v

Breathy voiced palatal lateral affricated click /gi / 1 (-1)

RARE !Xu.
A?
Glottalized velarized voiced palatal lateral affricated click /g£ / 1
!Xu.

Voiceless alveolar lateral affricated click II I 2

Zulu, Nama,

Velarized voiceless aspirated alveolar lateral affricated click /£ / 1

Nama.

Nasalized voiceless aspirated alveolar lateral affricated click / JL I 1

Nama.

220
 Segment index

Glottalized nasalized voiceless alveolar lateral affricated click /^l^/ 1

Nama •

Nasalized voiced alveolar lateral affricated click / Q G / 1

Nama.

Nasalized voiced palatal lateral affricated click / Q G / 1

!

palatal lateral affricated click

!Xu.

4. AFFRICATES
(Pulmonic and glottalic affricates are included here. Affricated clicks are
listed together with other clicks in section 3.)

Voiceless labio-dental affricate /pf/ 3

German, Beembe, Teke•

Voiceless aspirated labio-dental affricate /pf h / 1

Beembe•

Voiced labio-dental affricate /bv/ 1

Teke.

Voiceless dental affricate /t0/ 2

Luo, Chipewyan.

Voiceless aspirated dental affricate /t9 h / 1

Chipewyan.

Voiced dental affricate /d3/ 1

Luo.

Voiceless dental sibilant affricate /ts/ 10

Hungarian, Tuva, Ewe, Standard Thai, Malagasy, Tamang, Tzeltal,
Squamish, Gununa-Kena, Kabardian.

Voiceless aspirated dental sibilant affricate /ts h / 4

Standard Thai, Tamang, Acoma, Nama.

Voiced dental sibilant affricate /dz/ 5

Hungarian, Ewe, Malagasy, Acoma, Kabardian.

Voiceless dental/alveolar sibilant affricate /"ts'7 46 (-5)

German, Russian, Bulgarian, Romanian, Pashto, Kashmiri, Albanian, E.
Armenian, Lappish, Yurak, Kirghiz, Hebrew, Awiya, Kullo, Lakkia, Atayal,
Tagalog, Tsou, Mandarin, Hakka, Changchow, Amoy, Fuchow, Kan, Jingpho,
Yao, Tlingit, Chipewyan, Chontal, Mazahua, Tonkawa, Wichita, Yuchi,
Wappo, Bribri, Ashuslay, Jivaro, Greenlandic, Aleut, Basque, Burushaski;
RARE Chukchi; LOAN Lithuanian, Cheremis, Komi, Chuvash.

Long voiceless dental/alveolar affricate /"ts:"/ 3 (-1)

Wichita, Lak; 7DERIVED Japanese.

Long labialized voiceless dental/alveolar sibilant affricate / M ts W :'7 1

Lak.

221
Segment index

Palatalized voiceless dental/alveolar sibilant affricate /"ts J "/ 3 (-1)

Bulgarian, Yurak; LOAN Lithuanian.

Voiceless aspirated dental/alveolar sibilant affricate /"ts "/ 19

Kashmiri, E. Armenian, Mongolian, Lakkia, Adzera, Mandarin, Hakka,
Changchow, Amoy, Fuchow, Kan, Yao, Chipewyan, Mazahua, Zuni, Wiyot,
Yuchi, Lak, Burushaski.

Labialized voiceless aspirated dental/alveolar sibilant affricate /"ts Wh '7 1

Lak.

Voiceless dental/alveolar sibilant affricate with breathy release /"ts "/ 1

Changchow.

Laryngealized voiceless dental/alveolar sibilant affricate /"ts11/ 1

Ashuslay.

Voiced dental/alveolar sibilant affricate /"dz"/ 18 (-1)

Lithuanian, Pashto, Kashmiri, Albanian, Mongolian, Yulu, Adzera, Amoy,
Jingpho, Yao, Suena, Tlingit, Puget Sound, Seneca, Yuchi, Ocaina,
Telugu; RARE Awiya.

Voiceless alveolar affricate /tJ/ 1

Tamang.

Voiceless aspirated alveolar affricate / t ^ h / 1

Tamang.

Voiceless alveolar sibilant affricate /ts/ 39

Greek, Bashkir, Lelemi, Amo, Beembe, Ik, Hamer, Margi, Sui, Javanese,
Chamorro, Rukai, Tiddim Chin, Garo, Navaho, Hupa, Zoque, Tzeltal,
Totonac, K'ekchi, Mixe, Mazatec, Nootka, Quileute, Puget Sound, Hopi,
Porno, Tarascan, Cayapa, Paez, Ocaina, Chacobo, Cashinahua, Jaqaru,
Campa, Moxo, Guahibo, Georgian, !Xu.

Velarized voiceless alveolar sibilant affricate /£s/ 1

!Xu.

Voiceless aspirated alveolar sibilant affricate /ts h / 10

Beembe, Sui, Navaho, Hupa, Kwakw'ala, Tarascan, Jaqaru, Amuesha,
Georgian, !Xu.

Voiceless alveolar sibilant affricate with breathy release /ts / 1

Javanese.

Voiced alveolar sibilant affricate /dz/ 10

Greek, Lelemi, Ik, Margi, Chamorro, Garo, Yareba, Kwakw'ala, Puget
Sound, Ocaina.

Prenasalized voiced alveolar sibilant affricate / dz/ 1

Mazatec.

Velarized voiced alveolar sibilant affricate /dz/ 1

Breathy voiced alveolar sibilant affricate /dz/ 1

!Xu.

222
 Segment index

Voiceless palato-alveolar sibilant affricate /tJ7 141 (-5)

Irish, Lithuanian, Russian, Bulgarian, Spanish, Romanian, Farsi, Pashto,
Kurdish, Hindi-Urdu, Bengali, Punjabi, Albanian, E. Armenian, Cheremis,
Komi, Hungarian, Lappish, Azerbaijani, Chuvash, Yakut, Kirghiz, Khalaj,
Tuva, Evenki, Goldi, Manchu, Korean, Japanese, Moro, Bambara, Dagbani,
Tarok, Amo, Swahili, Luvale, Teke, Kanuri, Maasai, Luo, Nubian, Ik,
Logbara, Kunama, Tigre, Amharic, Neo-Aramaic, Somali, Awiya, Kullo,
Dizi, Hausa, Angas, Margi, Mundari, Kharia, Sedang, Khmer, Maranungku,
Sui, Po-ai, Lungchow, Sundanese, Sa'ban, Iai, Taishan, Burmese, Lahu,
Jingpho, Andamanese, Asmat, Washkuk, Haida, Tlingit, Navaho, Chipewyan,
Tolowa, Hupa, Wintu, Chontal, Totonac, K'ekchi, Mazahua, Mazatec,
Mixtec, Nootka, Quileute, Squamish, Puget Sound, Papago, LuiseTio,
Yacqui, Tiwa, Karok, Diegueflo, Shasta, Tarascan, Ojibwa, Delaware,
Wiyot, Dakota, Yuchi, Alabama, Wappo, Itonama, Bribri, Cayapa, Paez,
Ocaina, Muinane, Apinaye, Amahuaca, Chacobo, Tacana, Cashinahua,
Ashuslay, Abipon, Quechua, Jaqaru, Island Carib, Campa, Guajiro, Moxo,
Siriono, Ticuna, Siona, Jivaro, Cofan, Araucanian, Malayalam, Yukaghir,
Georgian, Basque, Burushaski, Brahui, !Xu; LOAN Goldi, Hebrew, Otomi,
Guarani; OBSCURE Kefa.

Long voiceless palato-alveolar sibilant affricate /tf:/ 5 (-1)

Punjabi, Yakut, Delaware, Lak; ?DERIVED Japanese.

Labialized voiceless palato-alveolar sibilant affricate /\Jw/ i

Ga.

Long labialized voiceless palato-alveolar sibilant affricate /tJW:/ 1

Lak.

Palatalized voiceless palato-alveolar sibilant affricate /\JJ/ l

Kashmiri.

Velarized voiceless palato-alveolar sibilant affricate

!Xu.

Voiceless aspirated palato-alveolar sibilant affricate / t j h / 43 (-2)

Hindi-Urdu, Bengali, Punjabi, E. Armenian, Osmanli, Mongolian, Korean,
Igbo, G3, Swahili, Kharia, Khmer, Sui, Lungchow, Taishan, Burmese, Lahu,
Haida, Navaho, Chipewyan, Tolowa, Mazahua, Yana, Tarascan, Zuni, Acoma,
Wiyot, Dakota, Yuchi, Tunica, Quechua, Jaqaru, Wapishana, Amuesha,
Cofan, Kota, Gilyak, Georgian, Lak, Burushaski, !Xu; RARE Jingpho; LOAN
Po-ai.

Long voiceless aspirated palato-alveolar sibilant affricate /tjh;/ 1

Punjabi.

Labialized voiceless aspirated palato-alveolar sibilant affricate /\\ h / 2

Hupa, Lak.

Palatalized voiceless aspirated palato-alveolar sibilant affricate 7tJ Jh / 2 (-1)

Kashmiri; OBSCURE Amuesha.

Preaspirated voiceless palato-alveolar sibilant affricate / h tj7 2 (-1)

Guajiro; OBSCURE Ojibwa.

Laryngealized voiceless palato-alveolar sibilant affricate /tJ7 3

Korean, Mundari, Ashuslay.

223
Segment index

Voiced palato-alveolar sibilant affricate / d y 80 (-5)

Irish, Lithuanian, Bulgarian, Romanian, Farsi, Pashto, Kurdish,
Hindi-Urdu, Bengali, Punjabi, Albanian, Komi, Osmanli, Azerbaijani,
Yakut, Mongolian, Evenki, Goldi, Manchu, Japanese, Moro, Bambara,
Dagbani, Igbo, G2, Tarok, Zulu, Teke, Kanuri, Fur, Luo, Nubian, Ik,
Nera, Logbara, Sara, Berta, Kunama, Arabic, Tigre, Amharic, Neo-Aramaic,
Awiya, Beja, Kullo, Dizi, Hausa, Angas, Margi, Mundari, Kharia, Khasi,
Bandjalang, Batak, Sa'ban, Iai, Amoy, Burmese, Lahu, Jingpho,
Andamanese, Tlingit, Wintu, Papago, Achumawi, Yana, Acoma, Yuchi,
Muinane, Ticuna, Cofan, Kota, Malayalam, Burushaski, Brahui; RARE Puget
Sound; LOAN Hungarian, Khalaj, Hebrew; OBSCURE Kefa.

Long voiced palato-alveolar sibilant affricate /d3:/ 2

Punjabi, Yakut.

Prenasalized voiced palato-alveolar sibilant affricate /-d^/ 5

Luvale, Sedang, Washkuk, Mazatec, Siriono.

Labialized voiced palato-alveolar sibilant affricate /d^ W / 1

Ga.

Palatalized voiced palato-alveolar sibilant affricate /dy'/ 1

Kashmiri.

Velarized voiced palato-alveolar sibilant affricate

!Xu.

Breathy voiced palato-alveolar sibilant affricate ^

Hindi-Urdu, Bengali, Igbo, Mundari, Kharia, !Xu.""

Voiceless retroflex affricate

Araucanian.

Voiceless retroflex sibilant affricate /ts/ 7

Ostyak, Mandarin, Mazatec, Tacana, Jaqaru, Basque, Burushaski.

Voiceless aspirated retroflex sibilant affricate /t s h / 6

Mandarin, Acoma, Wichita, Jaqaru, Amuesha, Burushaski.

Voiced retroflex sibilant affricate /dz/ 2

Acoma, Burushaski.

Prenasalized voiced retroflex sibilant affricate /^dz/ 1

Mazatec.

Voiceless retroflex affricated trill /tf/ 1

Malagasy.

Voiced retroflex affricated trill /dr/ 1

Malagasy.

Voiceless palatal affricate /eg/ 9

Sinhalese, Albanian, Komi, Hungarian, Akan, Saek, Mandarin, Kan, Ao.

Labialized voiceless palatal affricate /cg W / 1

Akan.

224
 Segment index

Voiceless aspirated palatal affricate /cgh/ 2

Mandarin, Kan.

Voiced palatal affricate /jj/ 5

Sinhalese, Albanian, Komi, Hungarian, Akan.

Labialized voiced palatal affricate /j-j / 1

Akan.

Voiceless palatal sibilant affricate /cjs/ 2 (-1)

Gununa-Kena; LOAN Telugu.

Voiceless velar affricate /kx/ 1

Tavgy.

Voiceless aspirated velar affricate /kx h / 2

Chipewyan, Nama.

Labialized voiceless aspirated velar affricate /kx h / 1

Chipewyan•

Voiceless uvular affricate /qx/ 3

Wolof, Nez Perce, Kabardian.

Labialized voiceless uvular affricate /q)( / 1

Kabardian.

Palatalized voiceless dental lateral affricate /tJJ/ 1

Kabardian.

Voiceless dental/alveolar lateral affricate /llt411/ 5

Haida, Tlingit, Chipewyan, Nootka, Squamish.

Voiceless aspirated dental/alveolar lateral affricate / M t4 h "/ 1

Chipewyan.

Voiced dental/alveolar lateral affricate /"dI 3"/ 2

Haida, Tlingit.

Voiceless alveolar lateral affricate /t4/ 3

Navaho, Wintu, Quileute.

Voiceless aspirated alveolar lateral affricate /t4 h / 1

Kwakw'ala.

Voiced alveolar lateral affricate /dlV 2

Navaho, Kwakw'ala.

Voiceless velar plosive with alveolar lateral fricative release /k4/ 1 (-1)
OBSCURE Ashuslay.

Voiceless dental ejective affricate /t0 5 / 1

Chipewyan.

Voiceless dental sibilant ejective affricate /tsV 5

Tzeltal, Squamish, Acoma, Gununa-Kena, Kabardian.

225
Segment index

Voiceless dental/alveolar sibilant ejective affricate /"ts'"/ 11

E. Armenian, Kullo, Dizi, Tlingit, Chipewyan, Mazahua, Zuni, Wichita,
Yuchi, Wappo, Lak.

Labialized voiceless dental/alveolar sibilant ejective affricate /"tsW''7 1

Lak.

Voiceless alveolar sibilant ejective affricate / t s V 18

Ik, Tigre, Iraqw, Navaho, Tolowa, Hupa, Wintu, Tzeltal, K'ekchi, Otomi,
Nootka, Kwakw'ala, Quileute, Puget Sound, Porno, Jaqaru, Georgian, !Xu.

Voiced alveolar sibilant ejective affricate / d z V 1

!Xu.

Voiceless palato-alveolar sibilant ejective affricate / t | V 35 (-2)

E. Armenian, Zulu, Tigre, Amharic, Dizi, Haida, Tlingit, Navaho,
Chipewyan, Tolowa, Hupa, Wintu, Chontal, K'ekchi, Mazahua, Nootka,
Quileute, Squamish, Puget Sound, Yana, Shasta, Zuni, Acoma, Dakota,
Yuchi, Wappo, Itonama, Quechua, Jaqaru, Gununa-Kena, Georgian, Lak, !Xu;
?DERIVED Otomi; OBSCURE Kefa.

Labialized voiceless palato-alveolar sibilant ejective affricate /tj V 1

Lak.

Voiced palato-alveolar sibilant ejective affricate / d V / 1

!Xu.

Voiceless retroflex sibilant ejective affricate / t s V 3

Tolowa, Acoma, Jaqaru.

Voiceless palatal ejective affricate / c g V 1

Gununa-Kena.

Voiceless velar ejective affricate /kx 5 / 1

Tlingit.

Labialized voiceless velar ejective affricate /kx W '/ 1

Tlingit.

Voiceless dental/alveolar lateral ejective affricate /"t4"7 6

Ik, Haida, Tlingit, Chipewyan, Nootka, Squamish.

Voiceless alveolar lateral ejective affricate /t4V 7

Iraqw, Navaho, Hupa, Wintu, Kwakw'ala, Quileute, Puget Sound.

Voiceless laterally-released velar ejective affricate /k4 9 / 1

Zulu.

5. FRICATIVES

Voiceless bilabial fricative / $ / 21 (-5)

Ewe, Hausa, Sui, Iai, Kan, Washkuk, Kewa, Fasu, Yareba, Otomi, Yuchi,
Alabama, Cayapa, Ocaina, Muinane, Araucanian; RARE Kanuri; LOAN
Sinhalese, Mongolian, Tarascan, Quechua.

Labialized voiceless bilabial fricative / $ W / 1

Washkuk.

226
 Segment index

Labialized velarized voiceless bilabial fricative /# / 1

Irish.

Palatalized voiceless bilabial fricative / $ J / 3

Irish, Hausa, Paez.

Voiced bilabial fricative /p/ 32 (-2)

Spanish, Pashto, Cheremis, Kirghiz, Evenki, Goldi, Ewe, Neo-Aramaic,
Dizi, Atayal, Kaliai, Iai, Washkuk, Gadsup, Rotokas, Mazatec, Mixtec,
Diegueflo, Paez, Ocaina, Muinane, Carib, Chacobo, Tacana, Amuesha, Campa,
Moxo, Cofan, Greenlandic, Georgian; LOAN Mongolian, Quechua.

Palatalized voiced bilabial fricative /(3J/ 1

Irish.

Voiceless labio-dental fricative /f/ 135 (-20)

Greek, Breton, German, Norwegian, Russian, Bulgarian, French, Spanish,
Romanian, Farsi, Pashto, Kurdish, Albanian, Hungarian, Lappish, Osmanli,
Azerbaijani, Kirghiz, Tuva, Manchu, Katcha, Moro, Kadugli, Kpelle, Bisa,
Bambara, Dan, Wolof, Diola, Temne, Dagbani, Senadi, Tampulma, Bariba,
Ewe, Akan, Igbo, G3, Efik, Birom, Tarok, Amo, Beembe, Swahili, Zulu,
Teke, Doayo, Gbeya, Zande, Songhai, Maba, Fur, Luo, Nubian, Ik, Tama,
Nera, Tabi, Logbara, Berta, Kunama, Arabic, Tigre, Amharic, Hebrew,
Socotri, Neo-Aramaic, Shilha, Tuareg, Somali, Awiya, Iraqw, Beja, Kullo,
Dizi, Kefa, Angas, Margi, Ngizim, Vietnamese, Kunjen, Standard Thai,
Lakkia, Yay, Po-ai, Lungchow, Malagasy, Chamorro, Tsou, Adzera, Iai,
Maori, Mandarin, Taishan, Hakka, Changchow, Lahu, Yao, Asmat, Sentani,
Telefol, Chuave, Koiari, Taoripi, Wintu, Karok, Porno, Ashuslay, Island
Carib, Guahibo, Cofan, Ket, Kabardian, Basque, Brahui; RARE Luvale,
Kanuri; LOAN Hindi-Urdu, Punjabi, E. Armenian, Cheremis, Komi, Finnish,
Chuvash, Yakut, Bashkir, Khalaj, Yacqui, Tarascan, Wappo, Moxo, Guarani,
Ticuna, Telugu, Burushaski.

Long voiceless labio-dental fricative /f;/ 3

Arabic, Shilha, Greenlandic.

Palatalized voiceless labio-dental fricative /f J / 2

Russian, Bulgarian.

Voiced labio-dental fricative /y/ 67 (-3)

Greek, Breton, German, Lithuanian, Russian, Bulgarian, French, Romanian,
Farsi, Kurdish, Albanian, E. Armenian, Komi, Finnish, Hungarian,
Osmanli, Azerbaijani, Tuva, Kpelle, Bisa, Dan, Dagbani, Senadi,
Tampulma, Ewe, Igbo, G3, Lelemi, Birom, Tarok, Amo, Beembe, Swahili,
Zulu, Doayo, Gbeya, Zande, Logbara, Hebrew, Angas, Margi, Ngizim,
Vietnamese, Yay, Saek, Po-ai, Lungchow, Malagasy, Rukai, Tsou,
Changchow, Lahu, Tiddim Chin, Yagaria, Nambakaengo, Mixe, Luisefio, Hopi,
Apinaye, Guarani, Guahibo, Kota, Kabardian, Brahui; LOAN Yakut, Bashkir,
Khalaj.

Palatalized voiced labio-dental fricative / V J / 3

Lithuanian, Russian, Bulgarian.

Voiceless dental fricative /Q/ 18 (-1)

Greek, Spanish, Albanian, Bashkir, Tabi, Mursi, Berta, Lakkia, Yay,
Rukai, Iai, Burmese, Karen, Chipewyan, Wintu, Amahuaca, Araucanian; RARE
Arabic.

227
Segment index

Voiced dental fricative /5/ 21 (-5)

Greek, Spanish, Albanian, Cheremis, Yurak, Tavgy, Moro, Tabi, Rukai,
Iai, Burmese, Koiari, Chipewyan, Mixtec, Tacana, Aleut; RARE Arabic,
Kunjen; LOAN Mazatec, Quechua, Guarani.

Palatalized voiced dental fricative / 5 J / 1

Yurak.

Pharyngealized voiced dental fricative /5 / 1 (-1)

RARE Arabic.

Voiceless dental sibilant fricative /s/ 33

Irish, Russian, French, Spanish, Kurdish, Punjabi, Sinhalese, Albanian,
Hungarian, Yakut, Tuva, Manchu, Moro, Nyangi, Arabic, Tuareg, Beja, Stan-
dard Thai, Javanese, Tamang, Nimboran, Tzeltal, Mixe, Squamish, Papago,
Luiseno, Karok, Diegueno, Yana, Acoma, Gununa-Kena, Guarani, Kabardian.

Long voiceless dental sibilant fricative /s:/ 3

Punjabi, Yakut, Arabic.

Palatalized voiceless dental sibilant fricative /s^/ 1

Russian.

Pharyngealized voiceless dental sibilant fricative /s / 1

Arabic.

Long pharyngealized voiceless dental sibilant fricative /s :/ 1

Arabic.

Voiced dental sibilant fricative /z/ 11 (-2)

Russian, French, Kurdish, Albanian, Hungarian, Tuva, Arabic, Tuareg,
Kabardian; LOAN Punjabi, Yakut.

Long voiced dental sibilant fricative /z:/ 1

Arabic.

Palatalized voiced dental sibilant fricative / z J / 1

Russian.

Pharyngealized voiced dental sibilant fricative /z / 2

Arabic, Tuareg.

Long pharyngealized voiced dental sibilant fricative /z :/ 1

Arabic.

Voiceless dental/alveolar sibilant fricative /"s'7 131 (-3)

Breton, Norwegian, Lithuanian, Bulgarian, Romanian, Farsi, Pashto,
Hindi-Urdu, Kashmiri, E. Armenian, Ostyak, Cheremis, Lappish, Yurak,
Tavgy, Chuvash, Kirghiz, Khalaj, Mongolian, Goldi, Korean, Japanese,
Katcha, Wolof, Dagbani, Senadi, Ewe, Akan, Igbo, G3, Tarok, Swahili,
Luvale, Doayo, Songhai, Kanuri, Maasai, Nubian, Sebei, Nera, Logbara,
Berta, Amharic, Hebrew, Socotri, Shilha, Awiya, Kullo, Dizi, Hausa,
Kharia, Khasi, Sedang, Khmer, Mabuiag, Yay, Sui, Atayal, Malagasy,
Malay, Batak, Tagalog, Chamorro, Tsou, Adzera, Kaliai, Iai, Mandarin,
Hakka, Changchow, Amoy, Fuchow, Kan, Dafla, Burmese, Jingpho, Karen,
Yao, Telefol, Pawaian, Dani, Wantoat, Suena, Kunimaipa, Taoripi,
Nambakaengo, Tlingit, Chipewyan, Chontal, Mazahua, Mazatec, Mixtec,

228
 Segment index

Chatino, Tiwa, Porno, Shasta, Tarascan, Zuni, Ojibwa, Delaware, Tonkawa,

Wiyot, Seneca, Wichita, Dakota, Yuchi, Wappo, Itonama, Bribri, Mura,
Carib, Amahuaca, Ashuslay, Arabela, Quechua, Island Carib, Guajiro,
Siriono, Jivaro, Cofan, Greenlandic, Kurukh, Ket, Yukaghir, Gilyak, Lak,
Basque, Burushaski; RARE Apinaye; LOAN Bengali, Ticuna.

Long voiceless dental/alveolar sibilant fricative /"s:"/ 4

Japanese, Shilha, Wichita, Lak.

Labialized voiceless dental/alveolar sibilant fricative /"s "/ 1

Amharic.

Palatalized voiceless dental/alveolar sibilant fricative /"s J.

"/ 6 (-1)
Lithuanian, Bulgarian, Yurak, Ket, Chukchi; LOAN Cheremis.

Pharyngealized voiceless dental/alveolar sibilant fricative /"s "/ 1

Shilha.

Voiceless aspirated dental/alveolar sibilant fricative /"s""/ 3

Burmese, Karen, Mazahua.

Preaspirated voiceless dental/alveolar sibilant fricative / " h s " / 1 (-1)

OBSCURE Ojibwa.

Laryngealized voiceless dental/alveolar sibilant fricative /"s'V 2

Korean, Hausa.

Voiced dental/alveolar sibilant fricative /"z'V 50 (-4)

Breton, Lithuanian, Bulgarian, Romanian, Farsi, Pashto, E. Armenian,
Cheremis, Kirghiz, Khalaj, Japanese, Katcha, Dagbani, Senadi, Ewe, Igbo,
G2, Tarok, Swahili, Doayo, Songhai, Kanuri, Logbara, Yulu, Amharic,
Hebrew, Socotri, Shilha, Tuareg, Awiya, Kullo, Dizi, Hausa, Mabuiag,
Sui, Malagasy, Changchow, Burmese, Kunimaipa, Chipewyan, Mazahua,
Dakota, Bribri, Apinaye, Lak, Burushaski; RARE Karen; LOAN Hindi-Urdu,
Chuvash, Evenki.

Long voiced dental/alveolar sibilant fricative /"z:"/ 1

Shilha.

Prenasalized voiced dental/alveolar sibilant fricative /" z'7 1

Wantoat.

Palatalized voiced dental/alveolar sibilant fricative / M z J l 7 1

Bulgarian.

Pharyngealized voiced dental/alveolar sibilant fricative / M z "/ 1

Shilha.

Voiced alveolar fricative / V 3

Azerbaijani, Sa'ban, Karen.

Voiceless alveolar sibilant fricative /s/ 112 (-3)

Greek, German, Komi, Finnish, Osmanli, Azerbaijani, Bashkir, Evenki,
Kadugli, Kpelle, Bisa, Bambara, Dan, Diola, Temne, Tampulma, Bariba,
Lelemi, Efik, Birom, Amo, Beembe, Zulu, Teke, Gbeya, Zande, Maba, Fur,
Luo, Ik, Tama, Temein, Tabi, Sara, Kunama, Koma, Tigre, Neo-Aramaic,
Somali, Iraqw, Hamer, Angas, Margi, Ngizim, Mundari, Vietnamese, Saek,

229
Segment index

Sundanese, Sa'ban, Rukai, Ao, Tiddim Chin, Garo, Boro, Asmat, Washkuk,
Iwam, Selepet, Yagaria, Kewa, Chuave, Daribi, Fasu, Yareba, Navaho,
Tolowa, Hupa, Nez Perce, Klamath, Maidu, Wintu, Zoque, Tzeltal, Totonac,
K'ekchi, Otomi, Nootka, Kwakw'ala, Quileute, Puget Sound, Hopi, Yacqui,
Karok, Dieguefio, Achumawi, Tunica, Alabama, Cayapa, Paez, Muinane,
Ocaina, Chacobo, Cashinahua, S. Nambiquara, Jaqaru, Wapishana, Amuesha,
Campa, Moxo, Guarani, Guahibo, Siona, Tucano, Malayalam, Georgian, Nama,
Brahui, Ainu, !Xu; RARE Telugu; LOAN Barasano, Araucanian.

Long voiceless alveolar sibilant affricate /s:/ 2

Finnish, Iraqw.

Pharyngealized voiceless alveolar sibilant fricative /s / 1

Kurdish.

Laryngealized voiceless alveolar sibilant fricative /s/ 2

S. Nambiquara, Siona,

Voiced alveolar sibilant fricative / z / 36 (-1)

Greek, German, Komi, Osmanli, Azerbaijani, Kpelle, Bisa, Bambara, Dan,
Tampulma, Bariba, Birom, Amo, Zulu, Gbeya, Zande, Maba, Fur, Ik, Tabi,
Tigre, Neo-Aramaic, Hamer, Angas, Margi, Ngizim,Vietnamese, Ao, Tiddim
Chin, Boro, Navaho, Otomi, Georgian, Brahui, !Xu; LOAN Bashkir.

Prenasalized voiced alveolar sibilant fricative / z/ 1

Paez.

Voiceless palato-alveolar fricative / j / 1

Kabardian.

Voiced palato-alveolar fricative / j / 1

Kabardian.

Voiceless palato-alveolar sibilant fricative /J/ 146 (-9)

Irish, Breton, German, Norwegian, Lithuanian, Bulgarian, French,
Romanian, Farsi, Pashto, Kurdish, Albanian, E. Armenian, Cheremis, Komi,
Hungarian, Lappish, Osmanli, Azerbaijani, Chuvash, Kirghiz, Bashkir,
Khalaj, Tuva, Mongolian, Manchu, Japanese, Bambara, Senadi, G3, Birom,
Tarok, Amo, Swahili, Luvale, Zulu, Kanuri, Maba, Fur, Maasai, Luo,
Nubian, Nera, Tabi, Mursi, Berta, Kunama, Koma, Arabic, Tigre, Amharic,
Hebrew, Socotri, Neo-Aramaic, Shilha, Tuareg, Somali, Awiya, Iraqw,
Beja, Kullo, Dizi, Kefa, Hamer, Hausa, Angas, Margi, Ngizim, Kanakuru,
Khasi, Sedang, Sui, Po-ai, Lungchow, Cham, Iai, Taishan, Burmese, Lahu,
Jingpho, Washkuk, Navaho, Chipewyan, Tolowa, Chontal, Totonac, K'ekchi,
Mixe, Otomi, Mazahua, Mixtec, Chatino, Nootka, Quileute, Squamish, Puget
Sound, Karok, Porno, Achumawi, Tarascan, Zuni, Acoma, Ojibwa, Delaware,
Wiyot, Dakota, Yuchi, Tunica, Wappo, Bribri, Cayapa, Paez, Ocaina,
Muinane, Chacobo, Tacana, Cashinahua, Ashuslay, Arabela, Quechua,
Jaqaru, Gununa-Kena, Wapishana, Amuesha, Campa, Guajiro, Siriono,
Jivaro, Cofan, Greenlandic, Aleut, Georgian, Kabardian, Lak, Basque,
Burushaski, !Xu; RARE Karen, Moxo; LOAN Hindi-Urdu, Sinhalese, Ostyak,
Finnish, Yakut; 7DERIVED Luiseflo; OBSCURE Ojibwa.

Long voiceless palato-alveolar sibilant fricative /J:/ 4

Japanese, Arabic, Shilha, Lak.

230
 Segment index

Labialized voiceless palato-alveolar sibilant fricative /J / 2

G3, Lak.

Long labialized palato-alveolar sibilant fricative / J : / 1

Lak.

Palatalized voiceless palato-alveolar sibilant fricative /J J / 3

Lithuanian, Kashmiri, Paez.

Velarized voiceless palato-alveolar sibilant fricative /f/ 1

Russian.

Voiced palato-alveolar sibilant fricative / y 51 (-7)

Breton, Lithuanian, Bulgarian, French, Romanian, Farsi, Pashto, Kurdish,
Albanian, E. Armenian, Cheremis, Komi, Hungarian, Osmanli, Azerbaijani,
Khalaj, Tuva, Senadi, Tarok, Luvale, Maba, Tigre, Amharic, Socotri,
Neo-Aramaic, Shilha, Tuareg, Dizi, Angas, Margi, Ngizim, Kanakuru,
Navaho, Mixe, Mazahua, Mixtec, Dakota, Ocaina, Apinaye, Aleut, Georgian,
Kabardian, Lak, !Xu; RARE Atayal; LOAN German, Hindi-Urdu, Chuvash,
Yakut, Bashkir, Hebrew.

Long voiced palato-alveolar sibilant fricative / 5 : / 1

Shilha.

Prenasalized voiced palato-alveolar sibilant fricative / 5/ 1

Paez.

Palatalized voiced palato-alveolar sibilant fricative /^ J / 1

Lithuanian.

Velarized voiced palato-alveolar sibilant fricative /^/ 1

Russian.

Voiceless retroflex fricative /^/ 1

Hopi.

Voiced retroflex fricative / V 4

E. Armenian, Araucanian, Chukchi, Burushaski.

Voiceless retroflex sibilant fricative /s/ 17

Pashto, Punjabi, Cham, Mandarin, Tolowa, Mazatec, Papago, Tarascan,
Acoma, Chacobo, Tacana, Cashinahua, Telugu, Kota, Malayalam, Basque,
Burushaski.

Voiced retroflex sibilant fricative /z/ 3

Pashto, Mandarin, Amuesha.

Laryngealized voiced retroflex sibilant fricative / z / 1

Wapishana.

Voiceless palatal fricative / $ / 11 (-1)

Irish, Norwegian, Bengali, Komi, Margi, Mandarin, Kan, Haida, Kwakw'ala,
Paez; OBSCURE Chuvash.

Long voiceless palatal fricative /g:/ 1 (-1)

?DERIVE"b Japanese.

231
Segment index

Labialized voiceless palatal fricative /g W / 1

Akan.

Voiced palatal fricative / j / 7

Komi, Khalaj, Goldi, Margi, Asmat, Greenlandic, Ket.

Voiceless palatal sibilant fricative /c/ 2

Gununa-Kena, Telugu.

Voiced palatal sibilant fricative /z/ 2

Muinane, Cofan.

Voiceless velar fricative /x/ 76 (-7)

Greek, Irish, Breton, German, Russian, Bulgarian, Spanish, Kurdish,
Yurak, Azerbaijani, Chuvash, Kirghiz, Bashkir, Mongolian, Goldi, Manchu,
Hebrew, Shilha, Iraqw, Angas, Margi, Vietnamese, Sui, Atayal, Iai,
Fuchow, Dafla, Tiddim Chin, Karen, Kewa, Haida, Tlingit, Navaho,
Chipewyan, Tolowa, Hupa, Nez Perce, Wintu, K'ekchi, Nootka, Quileute,
Luiseflo, Tiwa, Karok, Porno, Achumawi, Yana, Shasta, Tarascan, Delaware,
Tonkawa, Dakota, Bribri, Ocaina, Muinane, Amahuaca, Ashuslay, Jaqaru,
Gununa-Kena, Amuesha, Guarani, Guahibo, Jivaro, Aleut, Gilyak, Lak,
Nama, Brahui, !Xu; LOAN Lithuanian, Cheremis, Komi, Khalaj, Kanuri,
Somali; ?DERIVED K'ekchi.

Long voiceless velar fricative /x:/ 3

Shilha, Greenlandic, Lak.

Labialized voiceless velar fricative /x W / 18 (-1)

Iraqw, Haida, Tlingit, Chipewyan, Tolowa, Hupa, Nootka, Kwakw'ala,
Quileute, Squamish, Puget Sound, Luiseflo, Tiwa, Diegueflo, Tonkawa,
Kabardian, Lak; LOAN Guarani.
w
Long labialized voiceless velar fricative /x :/ 1
Lak.

Palatalized voiceless velar fricative /x J / 1

Kabardian.

Voiced velar fricative /y/ 40 (-5)

Greek, Irish, Spanish, Kurdish, Ostyak, Cheremis, Azerbaijani, Kpelle,
Igbo, Fur, Shilha, Angas, Margi, Vietnamese, Maung, Kunjen, Sui, Saek,
Atayal, Lahu, Karen, Navaho, Chipewyan, Tolowa, Dakota, Paez, Abipon,
Amuesha, Guarani, Cofan, Greenlandic, Aleut, Chukchi, Gilyak, Brahui;
RARE Tiwi; LOAN Mazatec, Quechua; ?ABSTRACT Osmanli; OBSCURE Tuva.

Long voiced velar fricative /y:/ 1

Shilha.

Labialized voiced velar fricative /y W / 2 (-1)

Chipewyan; LOAN Guarani.

Palatalized voiced velar fricative /y J / 1

Kabardian.

Laryngealized voiced velar fricative /y/ 1

Sui.

232
 Segment index

Voiceless uvular fricative /)(/ 2 7 (-4)

Farsi, Pashto, E. Armenian, Wolof, Arabic, Neo-Aramaic, Tuareg,
Mandarin, Haida, Tlingit, Wintu, Kwakw'ala, Quileute, Squamish, Puget
Sound, Porno, Achumawi, Aleut, Kurukh, Gilyak, Georgian, Kabardian, Lak;
LOAN Hindi-Urdu, Nootka, Basque, Burushaski.

Long voiceless uvular fricative lyj:J 3

Arabic, Greenlandic, Lak.

Labialized voiceless uvular fricative / Y W / 9 (-1)

Haida, Tlingit, Kwakw'ala, Quileute, Squamish, Puget Sound, Kabardian,
Lak; LOAN Nootka.
w
Long labialized voiceless uvular fricative /\ / 1
Lak.

Voiced uvular fricative /b/ 14 (-1)

Pashto, Arabic, Hebrew, Tuareg, Sui, Abipon, Greenlandic, Aleut,
Yukaghir, Gilyak, Georgian, Kabardian, Burushaski; LOAN Hindi-Urdu.

Long voiced uvular fricative / b: / 1

Arabic.

Labialized voiced uvular fricative /b W / 1

Kabardian.

Voiceless pharyngeal fricative /ft/ 13

Kurdish, Ewe, Tama, Arabic, Tigre, Socotri, Shilha, Somali, Iraqw,
Atayal, Nootka, Kabardian, Lak.

Long voiceless pharyngeal fricative /R:/ 2

Arabic", Shilha.

Voiced pharyngeal fricative / ? / 8 (-1)

Kurdish, Ewe, Arabic, Tigre, Socotri, Shilha, Somali; LOAN Kabardian.

Long voiced pharyngeal fricative / ? : / 1

Arabic.

Voiceless "h" /h/ 202 (-2)

Irish, German, Norwegian, Romanian, Farsi, Pashto, Bengali, Albanian, E.
Armenian, Finnish, Hungarian, Lappish, Osmanli, Azerbaijani, Bashkir,
Khalaj, Korean, Japanese, Bambara, Diola, Temne, Tampulma, Bariba, Akan,
Igbo, G3, Birom, Tarok, Amo, Beembe, Luvale, Zulu, Teke, Doayo, Gbeya,
Zande, Kanuri, Fur, Luo, Nubian, Ik, Tama, Nera, Tabi, Mursi, Logbara,
Sara, Berta, Kunama, Koma, Arabic, Tigre, Amharic, Hebrew, Socotri,
Neo-Aramaic, Shilha, Tuareg, Somali, Iraqw, Beja, Kullo, Dizi, Kefa,
Hamer, Hausa, Angas, Ngizim, Mundari, Khasi, Vietnamese, Sedang, Khmer,
Standard Thai, Lakkia, Yay, Sui, Saek, Po-ai, Lungchow, Sundanese,
Javanese, Malagasy, Cham, Malay, Batak, Tagalog, Sa'ban, Chamorro,
Rukai, Tsou, Adzera, Roro, Kaliai, Iai, Maori, Hawaiian, Taishan, Hakka,
Changchow, Amoy, Kan, Tamang, Burmese, Lahu, Jingpho, Tiddim Chin, Garo,
Boro, Yao, Washkuk, Sentani, Nimboran, Iwam, Selepet, Yagaria, Pawaian,
Dani, Daribi, Fasu, Koiari, Taoripi, Haida, Navaho, Chipewyan, Tolowa,
Hupa, Nez Perce, Klamath, Maidu, Wintu, Chontal, Zoque, Tzeltal,
Totonac, Otomi, Mazahua, Mazatec, Mixtec, Chatino, Nootka, Kwakw'ala,
Quileute, Squamish, Puget Sound, Papago, Luiseflo, Hopi, Yacqui, Tiwa,

233
Segment index

Karok, Porno, Achumawi, Yana, Shasta, Tarascan, Zuni, Acoma, Delaware,

Tonkawa, Seneca, Dakota, Yuchi, Tunica, Alabama, Wappo, Itonama, Mura,
Cayapa, Paez, Ocaina, Carib, Amahuaca, Chacobo, Cashinahua, Abipon, S.
Nambiquara, Arabela, Quechua, Gununa-Kena, Island Carib, Campa, GuajirOj
Moxo, Siriono, Guahibo, Barasano, Siona, Tucano, Cofan, Aleut, Kurukh,
Malayalam, Ket, Gilyak, Lak, Nama, Burushaski, Ainu, Brahui; RARE
Wapishana; LOAN Georgian.

Long voiceless "hM /h:/ 2

Arabic, Delaware,

Labialized voiceless "h" /h W / 4

Igbo, Amharic, Hupa, Siona.

Laryngealized voiceless "h" /h/ 1 (-1)

OBSCURE S. Nambiquara.

Voiced "h" /&/ 13 (-1)

Hindi-Urdu, Kashmiri, Punjabi, Sinhalese, Zulu, Kharia, Changchow,
Dafla, Wichita, Telugu, Kabardian, !Xd; RARE Karen.

Long voiceless dental lateral fricative /*:/ 1

Greenlandic.

Palatalized voiceless dental lateral fricative /& I 1

Kabardian.

Palatalized voiced dental lateral fricative I

Kabardian.

Voiceless dental/alveolar lateral fricative /"*"/ 18

Ik, Socotri, Lakkia, Po-ai, Lungchow, Taishan, Haida, Tlingit, Tolowa,
Nootka, Tiwa, Zuni, Wiyot, Yuchi, Alabama, Aleut, Chukchi, Brahui.

Voiced dental/alveolar lateral fricative /"I 5'7 2

Ik, Socotri.

Voiceless alveolar lateral fricative /i/ 13

Zulu, Nyangi, Iraqw, Margi, Ngizim, Navaho, Hupa, Nez Perce, Totonac,
Kwakw'ala, Quileute, Puget Sound, Diegueffo.

Voiced alveolar lateral fricative /1 3/ 4

Zulu, Margi, Ngizim, Kanakuru.

Voiceless palato-alveolar lateral fricative /\I 1

Diegueffo.

Voiced palato-alveolar lateral fricative I V^/ 1

Pashto.

Voiced retroflex lateral fricative l\y 1

Ao.

Voiceless velar-alveolar lateral fricative /x4/ 1 (-1)

OBSCURE Ashuslay.

234
 Segment index

Voiceless bilabial ejective fricative /<£' / 1

Yuchi.

Voiceless labio-dental ejective fricative IV I 1

Kabarcfian.

Voiceless dental/alveolar sibilant ejective fricative /"s"'/ 7

Berta, Socotri, Tlingit, Mazahua, Wichita, Dakota, Yuchi.

Voiceless alveolar sibilant ejective fricative /s' / 1

Koma.

Voiceless palato-alveolar sibilant ejective fricative /J'/ 4 (-1)

Dakota, Yuchi, Kabardian; RARE Socotri.

Voiceless retroflex sibilant ejective fricative / s V 1

Acoma.

Voiceless palatal sibilant ejective fricative /p'/ 1

Acoma.

Voiceless velar ejective fricative /x'/ 2

Tlingit, Dakota.

Labialized voiceless velar ejective fricative /x ' / 1

Tlingit.

Voiceless uvular ejective fricative /x'/ 1

Tlingit.

Labialized voiceless uvular ejective fricative /^W'/ 1

Tlingit.

Voiceless dental/alveolar lateral ejective fricative /'V 1 '/ 2

Tlingit, Yuchi.

6. NASALS

Voiced bilabial nasal /m/ 299 (-3)

Greek, Breton, German, Norwegian, Lithuanian, Russian, Bulgarian,
French, Spanish, Romanian, Farsi, Pashto, Kurdish, Hindi-Urdu, Bengali,
Kashmiri, Punjabi, Sinhalese, Albanian, E. Armenian, Ostyak, Cheremis,
Komi, Finnish, Hungarian, Lappish, Yurak, Tavgy, Osmanli, Azerbaijani,
Chuvash, Yakut, Kirghiz, Bashkir, Khalaj, Tuva, Mongolian, Evenki,
Goldi, Manchu, Korean, Japanese, Katcha, Moro, Kadugli, Bisa, Bambara,
Dan, Wolof, Diola, Temne, Dagbani, Senadi, Tampulma, Bariba, Ewe, Akan,
Igbo, G3, Lelemi, Efik, Birom, Tarok, Amo, Beembe, Swahili, Luvale,
Zulu, Teke, Doayo, Gbeya, Zande, Songhai, Kanuri, Maba, Fur, Maasai,
Luo, Nubian, Nyangi, Ik, Sebei, Tama, Temein, Nera, Tabi, Mursi,
Logbara, Yulu, Sara, Berta, Kunama, Koma, Arabic, Tigre, Amharic,
Hebrew, Socotri, Neo-Aramaic, Shilha, Tuareg, Somali, Awiya, Iraqw,
Beja, Kullo, Dizi, Kefa, Hamer, Hausa, Angas, Margi, Ngizim, Kanakuru,
Mundari, Kharia, Khasi, Vietnamese, Sedang, Khmer, Maung, Tiwi, Burera,
Nunggubuyu, Alawa, Maranungku, Malakmalak, Bardi, Wik-Munkan, Kunjen,
Western Desert, Nyangumata, Aranda, Kariera-Ngarluma, Gugu-Yalanji,
Mabuiag, Arabana-Wanganura, Diyari, Bandjalang, Standard Thai, Lakkia,
Yay, Sui, Saek, Po-ai, Lungchow, Atayal, Sundanese, Javanese, Malagasy,

235
Segment index

Cham, Malay, Batak, Tagalog, Sa'ban, Chamorro, Rukai, Tsou, Adzera,

Roro, Kaliai, Iai, Maori, Hawaiian, Mandarin, Taishan, Changchow, Amoy,
Fuchow, Kan, Tamang, Dafla, Burmese, Lahu, Jingpho, Ao, Tiddim Chin,
Garo, Boro, Karen, Yao, Andamanese, Asmat, Washkuk, Sentani, Nimboran,
Iwam, Telefol, Selepet, Gadsup, Yagaria, Kewa, Chuave, Pawaian, Dani,
Wantoat, Daribi, Fasu, Suena, Dera, Kunimaipa, Yareba, Taoripi, Nasioi,
Nambakaengo, Haida, Tolowa, Hupa, Nez Perce, Klamath, Maidu, Wintu,
Chontal, Zoque, Tzeltal, Totonac, K'ekchi, Mixe, Otomi, Mazahua,
Mazatec, Chatino, Nootka, Kwakw'ala, Squamish, Papago, Luiseflo, Hopi,
Yacqui, Tiwa, Karok, Porno, Diegueflo, Achumawi, Yana, Shasta, Tarascan,
Zuni, Acoma, Ojibwa, Delaware, Tonkawa, Wiyot, Tunica, Alabama, Wappo,
Itonama, Bribri, Cayapa, Paez, Ocaina, Muinane, Carib, Amahuaca,
Chacobo, Tacana, Cashinahua, Ashuslay, Abipon, Arabela, Auca, Quechua,
Jaqaru, Gununa-Kena, Wapishana, Island Carib, Amuesha, Campa, Guajiro,
Moxo, Guarani, Guahibo, Ticuna, Siona, Jivaro, Cofan, Araucanian,
Greenlandic, Aleut, Telugu, Kota, Kurukh, Malayalam, Ket, Yukaghir,
Chukchi, Gilyak, Georgian, Kabardian, Lak, Nama, Basque, Burushaski,
Ainu, Brahui, !Xu; RARE Navaho, Seneca; 7DERIVED Dakota.

Long voiced bilabial nasal /m:/ 11

Punjabi, Finnish, Chuvash, Yakut, Wolof, Arabic, Shilha, Maranungku,
Delaware, Ocaina, !Xu.

Labialized voiced bilabial nasal /m W / 2

Wa shkuk, Nambakaeng o.

Labialized velarized voiced bilabial nasal /m / 1

Irish.

Palatalized voiced bilabial nasal /m J / 6

Irish, Lithuanian, Russian, Bulgarian, Yurak, Amuesha.

Breathy voiced bilabial nasal /m/ 2

Hindi-Urdu, !Xu.

Laryngealized voiced bilabial nasal /m/ 14

Gbeya, Sedang, Sui, Haida, Tolowa,"Nez Perce, Klamath, Otomi, Mazahua,
Nootka, Kwakw'ala, Acoma, Wappo, !Xu.

Voiceless bilabial nasal /m/ 11 (-1)

Sedang, Lakkia, Sui, Iai, Burmese, Yao, Klamath, Mazahua, Hopi, Aleut;
7DERIVED Otomi.

Voiced labiodental nasal /nrj/ 1

Teke.

Voiced dental nasal /n/ 55

Irish, Norwegian, Russian, French, Spanish, Farsi, Kurdish, Hindi-Urdu,
Punjabi, Finnish, Hungarian, Yakut, Khalaj, Tuva, Manchu, Gbeya, Ik,
Temein, Arabic, Tigre, Tuareg, Beja, Mundari, Tiwi, Nunggubuyu, Kunjen,
Western Desert, Aranda, Kariera-Ngarluma, Arabana-Wanganura, Diyari,
Standard Thai, Roro, Garo, Tamang, Yao, Nimboran, Selepet, Nez Perce,
Tzeltal, Squamish, Papago, Luiseflo, Diegueflo, Yana, Acoma, Tonkawa,
Gununa-Kena, Guarani, Araucanian, Greenlandic, Malayalam, Kabardian,
Nama, Brahui.

236
 Segment index

Long voiced dental nasal /n;/ 4

Punjabi, Finnish, Yakut, Arabic.

Palatalized voiced dental nasal /n / 1

Russian.

Breathy voiced dental nasal /n/ 1

Hindi-Urdu.

Laryngealized voiced dental nasal /n/ 3

Gbeya, Nez Perce, Acoma.

Voiceless dental nasal /n/ 1

Yao.

Voiced dental/alveolar nasal /"n'7 154 (-1)

Greek, Breton, German, Lithuanian, Bulgarian, Romanian, Pashto, Bengali,
Kashmiri, Sinhalese, E. Armenian, Ostyak, Cheremis, Lappish, Yurak,
Tavgy, Osmanli, Azerbaijani, Chuvash, Kirghiz, Mongolian, Evenki, Goldi,
Korean, Japanese, Katcha, Wolof, Dagbani, Senadi, Ewe, Akan, Igbo, G3,
Beembe, Swahili, Luvale, Doayo, Songhai, Kanuri, Maasai, Nubian, Sebei,
Nera, Mursi, Logbara, Yulu, Berta, Amharic, Hebrew, Socotri,
Neo-Aramaic, Shilha, Awiya, Kullo, Dizi, Kefa, Hausa, Kharia, Khasi,
Vietnamese, Sedang, Khmer, Burera, Maranungku, Wik-Munkan,
Kariera-Ngarluma, Gugu-Yalanji, Mabuiag, Lakkia, Yay, Sui, Po-ai,
Lungchow, Atayal, Sundanese, Javanese, Malagasy, Cham, Malay, Batak,
Tagalog, Tsou, Adzera, Kaliai, Iai, Hawaiian, Taishan, Changchow, Amoy,
Fuchow, Kan, Dafla, Burmese, Lahu, Jingpho, Karen, Andamanese, Sentani,
Telefol, Gadsup, Pawaian, Dani, Wantoat, Suena, Dera, Kunimaipa, Nasioi,
Nambakaengo, Haida, Tlingit, Chipewyan, Tolowa, Chontal, Mazahua,
Chatino, Nootka, Tiwa, Karok, Porno, Achumawi, Shasta, Tarascan, Zuni,
Delaware, Wiyot, Seneca, Wichita, Yuchi, Alabama, Wappo, Itonama,
Bribri, Carib, Amahuaca, Ashuslay, Abipon, Arabela, Quechua, Island
Carib, Guajiro, Ticuna, Jivaro, Cofan, Aleut, Kurukh, Ket, Yukaghir,
Chukchi, Gilyak, Georgian, Lak, Basque, Burushaski; ?DERIVED Dakota.

Long voiced dental/alveolar nasal /"n:"/ 6

Chuvash, Wolof, Shilha, Maranungku, Delaware, Wichita.

Labialized voiced dental/alveolar nasal /"n Wf 7 1

Nambakaengo.

Palatalized voiced dental/alveolar nasal /"n J "/ 5 (-1)

Irish, Bulgarian, Yurak, Chuvash; OBSCURE Nambakaengo.

Velarized voiced dental/alveolar nasal / " R " / 1

Irish.

Laryngealized voiced dental/alveolar nasal /"n"/ 8

Sedang, Sui, Haida, Tolowa, Nootka, Yuchi,~Wappo, S. Nambiquara.

Voiceless dental/alveolar nasal /"n"/ 6

Sedang, Lakkia, Sui, Burmese, Mazahua, Aleut.

Palatalized voiceless dental/alveolar nasal /"nJ'V 1 (-1)

7DERIVED Irish.

237
Segment index

Velarized voiceless dental/alveolar nasal / V 7 1 (-1)

7DERIVED Irish.

Voiced alveolar nasal /n/ 106

Albanian, Komi, Bashkir, Moro, Kadugli, Bisa, Bambara, Dan, Diola,
Temne, Tampulma, Bariba, Lelemi, Efik, Birom, Tarok, Amo, Zulu, Teke,
Zande, Maba, Fur, Luo, Nyangi, Tama, Tabi, Sara, Kunama, Koma, Somali,
Iraqw, Hamer, Angas, Margi, Ngizim, Kanakuru, Maung, Tiwi, Nunggubuyu,
Alawa, Malakmalak, Bardi, Kunjen, Western Desert, Nyangumata, Aranda,
Arabana-Wanganura, Diyari, Bandjalang, Saek, Sa'ban, Chamorro, Rukai,
Maori, Mandarin, Ao, Tiddim Chin, Boro, Asmat, Washkuk, Iwam, Yagaria,
Kewa, Chuave, Daribi, Fasu, Yareba, Koiari, Navaho, Hupa, Wintu,
Klamath, Maidu, Zoque, Totonac, K'ekchi, Mixe, Otomi, Mazatec,
Kwakw'ala, Hopi, Yacqui, Dieguetio, Ojibwa, Tunica, Cayapa, Paez, Ocaina,
Muinane, Chacobo, Tacana, Cashinahua, S. Nambiquara, Auca, Jaqaru,
Amuesha, Campa, Moxo, Guahibo, Siona, Araucanian, Telugu, Kota,
Malayalam, Ainu, !Xu.

Long voiced alveolar nasal /n:/ 2

Somali, Ocaina.

Palatalized voiced alveolar nasal /rr/ 1

Nyangumata.

Laryngealized voiced alveolar nasal /n/ 3

Klamath, Otomi, Kwakw'ala.

Voiceless alveolar nasal /n/ 3 (-1)

Klamath, Hopi; ?DERIVED Otomi.

Voiced palato-alveolar nasal /n/ 17

Albanian, Zulu, Nunggubuyu, Alawa, Malakmalak, Bardi, Wik-Munkan,
Aranda, Bandjalang, Mazatec, Papago, Diegueflo, Cayapa, Wapishana, Campa,
Moxo, Jivaro.

Voiced retroflex nasal /n/ 20 (-1)

Norwegian, Pashto, Punjabi, Ostyak, Mundari, Maung, Nunggubuyu, Alawa,
Bardi, Western Desert, Nyangumata, Aranda, Kariera-Ngarluma,
Arabana-Wanganura, Diyari, Iai, Telugu, Kota, Malayalam; 7DERIVED Tiwi.

Voiceless retroflex nasal /n/ 1

5
Iai.

Voiced palatal nasal /p/ 107 (-3)

Irish, Breton, Lithuanian, French, Spanish, Punjabi, Ostyak, Cheremis,
Komi, Hungarian, Lappish, Tavgy, Yakut, Evenki, Goldi, Katcha, Moro,
Kadugli, Bisa, Bambara, Wolof, Diola, Dagbani, Tampulma, Ewe, Igbo, G2,
Lelemi, Efik, Amo, Swahili, Luvale, Teke, Zande, Songhai, Kanuri, Maba,
Fur, Maasai, Luo, Nubian, Nyangi, Ik, Sebei, Tama, Temein, Tabi, Mursi,
Yulu, Sara, Amharic, Tuareg, Hamer, Angas, Margi, Ngizim, Kanakuru,
Kharia, Khasi, Vietnamese, Sedang, Khmer, Burera, Maranungku, Kunjen,
Kariera-Ngarluma, Gugu-Yalanji, Arabana-Wanganura, Diyari, Yay, Sui,
Saek, Sundanese, Javanese, Cham, Malay, Sa'ban, Chamorro, Iai, Kan,
Burmese, Yao, Andamanese, Washkuk, Mazahua, Hopi, Bribri, Paez, Ocaina,
Muinane, Abipon, Auca, Quechua, Jaqaru, Amuesha, Guarani, Ticuna, Cofan,
Araucanian, Malayalam, Ket, Yukaghir, Gilyak, Basque; RARE Karen; LOAN
Sinhalese, Tarascan.

238
 Segment index

Long voiced palatal nasal /p:/ 2

Yakut, Ocaina.

Laryngealized voiced palatal nasal /p/ 3

Sedang, Sui, Mazahua.

Voiceless palatal nasal /n/ 6

Sedang, Sui, Iai, Burmese, Yao, Mazahua.

Voiced velar nasal / Q / 167 (-9)

Irish, Breton, German, Norwegian, Kurdish, Bengali, Punjabi, Ostyak,
Cheremis, Lappish, Yurak, Tavgy, Yakut, Kirghiz, Bashkir, Tuva, Evenki,
Goldi, Manchu, Korean, Japanese, Katcha, Moro, Kadugli, Bisa, Bambara,
Wolof, Diola, Dagbani, Senadi, Tampulma, Ewe, Igbo, G3, Lelemi, Efik,
Birom, Tarok, Amo, Luvale, Teke, Doayo, Gbeya, Songhai, Kanuri, Fur,
Maasai, Luo, Nyangi, Ik, Sebei, Tama, Temein, Nera, Tabi, Mursi, Yulu,
Berta, Kunama, Awiya, Iraqw, Hamer, Angas, Margi, Kanakuru, Kharia,
Khasi, Vietnamese, Sedang, Khmer, Maung, Tiwi, Burera, Nunggubuyu,
Alawa, Maranungku, Malakmalak, Bardi, Wik-Munkan, Kunjen, Western
Desert, Nyangumata, Aranda, Kariera-Ngarluma, Gugu-Yalanji, Mabuiag,
Arabana-Wanganura, Diyari, Bandjalang, Standard Thai, Lakkia, Yay, Sui,
Saek, Po-ai, Lungchow, Atayal, Sundanese, Javanese, Cham, Malay, Batak,
Tagalog, Sa'ban, Chamorro, Rukai, Tsou, Adzera, Kaliai, Iai, Maori,
Mandarin, Taishan, Changchow, Amoy, Fuchow, Kan, Tamang, Dafla, Burmese,
Lahu, Jingpho, Ao, Tiddim Chin, Garo, Boro, Yao, Andamanese, Nimboran,
Iwam, Telefol, Selepet, Wantoat, Dera, Kunimaipa, Nambakaengo, Hupa,
Zoque, Luiseflo, Hopi, Tarascan, Cayapa, Carib, Auca, Jaqaru, Guarani,
Ticuna, Araucanian, Greenlandic, Aleut, Kota, Malayalam, Ket, Yukaghir,
Chukchi, Gilyak, Burushaski, !Xu; RARE Swahili, Koma, Karen; LOAN
French, Hebrew, Moxo; ?DERIVED Sinhalese, Mundari, Mixtec.

Long voiced velar nasal / Q ; / 2 (-1)

Yakut; ?DERIVED Finnish.

Labialized voiced velar nasal / Q W / 7

Awiya, Iraqw, Lakkia, Wantoat, Nambakaengo, Hopi, Guarani.

Palatalized voiced velar nasal / Q J / 2

Irish, Lakkia.
9
Pharyngealized voiced velar nasal /Q / 1
!Xu.
Laryngealized voiced velar nasal / Q / 3
Sedang, Sui, Haida.

Voiceless velar nasal / Q / 7

Sedang, Sui, Iai, Burmese, Yao, Hopi, Aleut.

Palatalized voiceless velar nasal /g / 1

Lakkia.

Voiced labial-velar nasal /(n¥)/ 6

Dagbani, Tampulma, Igbo, G2, Gbeya, Iai.

Voiceless labial-velar nasal /ijig/ 1

Iai.

239
Segment index

Voiced dental-palatal nasal /np/ 1

Maung.

7. TRILLS, TAPS AND FLAPS

Voiced dental trill /r/ 5

Russian, Hungarian, Tuva, Tamang, Gununa-Kena.

Palatalized voiced dental trill /r J / 1

Russian.

Voiced dental/alveolar trill /"r'7 52 (-4)

Bulgarian, Farsi, Pashto, Kurdish, Hindi-Urdu, Kashmiri, Punjabi,
Albanian, E. Armenian, Ostyak, Komi, Finnish, Lappish, Yurak, Tavgy,
Chuvash, Yakut, Kirghiz, Mongolian, Evenki, Goldi, Manchu, Katcha, Akan,
Kanuri, Nubian, Nera, Sebei, Mursi, Yulu, Berta, Khasi, Sedang,
Wik-Munkan, Kariera-Ngarluma, Sundanese, Malagasy, Malay, Adzera,
Chipewyan, Luiseflo, Arabela, Guajiro, Kurukh, Gilyak, Georgian, Lak,
Basque; RARE G3, Mixtec; LOAN Mazatec, Quechua.

Long voiced dental/alveolar trill / " r : 7 2

Finnish, Chuvash.

Palatalized voiced dental/alveolar trill /" r J "/ 2

Bulgarian, Yurak.

Velarized voiced dental/alveolar trill / V 7 1

Yukaghir.

Laryngealized voiced dental/alveolar trill /"r'7 1

Sedang.

Voiceless dental/alveolar trill /r/ 3

Maasai, Sedang, Gilyak.

Voiced dental/alveolar fricative trill /"r'7 1 (-1)

LOAN Guarani.

Voiced alveolar trill /r/ 54 (-3)

Breton, Lithuanian, Spanish, Bashkir, Moro, Kadugli, Bisa, Temne,
Tampulma, Amo, Maba, Fur, Nyangi, Ik, Tama, Temein, Tabi, Kunama, Koma,
Tigre, Shilha, Tuareg, Hausa, Angas, Margi, Ngizim, Mundari, Maung,
Nunggubuyu, Alawa, Maranungku, Malakmalak, Bardi, Kunjen, Nyangumata,
Arabana-Wanganura, Diyari, Bandjalang, Javanese, Rukai, Kaliai, Maori,
Dafla, Tarascan, Tunica, Guahibo, Kota, Malayalam, Kabardian, Nama,
Brahui; RARE Diegueflo; LOAN Swahili, Zulu.

Long voiced alveolar trill /r:/ 3

Arabic, Shilha, Somali.

Palatalized voiced alveolar trill /r J / 1

Lithuanian.

Pharyngealized voiced alveolar trill /r / 1

Shilha.

240
 Segment index

Voiced retroflex trill /r/ 5 (-2)

Pashto, Kariera-Ngarluma, Arabana-Wanganura; 7DERIVED Kurukh; OBSCURE
Batak.

Voiced uvular trill /R/ 2

German, French.

Voiced dental tap /B/ 1

Acoma.

Laryngealized voiced dental tap /B/ 1

Acoma.

Voiced dental/alveolar tap /"B'7 1

Khalaj.

Voiced alveolar tap /B/ 5

Spanish, Bambara, Tiwi, Rotokas, Malayalam.

Voiced labio-dental flap /*/ 2 (-1)

Gbeya; RARE Margi.

Voiced dental flap /r/ 1

Roro.

Voiced dental/alveolar flap /"r"/ 26 (-1)

Greek, Norwegian, Kurdish, Cheremis, Japanese, Wolof, Ewe, Igbo, Sebei,
Logbara, Sara, Awiya, Kharia, Suena, Nasioi, Chontal, Itonama, Amahuaca,
Quechua, Island Carib, Amuesha, Siriono, Ticuna, Jivaro, Basque; LOAN
Porno,

Palatalized voiced dental/alveolar flap /"r Jl 7 1

Igbo.

Voiced alveolar flap /r/ 53 (-2)

Irish, Romanian, Bengali, Sinhalese, Osmanli, Kpelle, Diola, Bariba,
Gbeya, Maasai, Luo, Tama, Arabic, Amharic, Neo-Aramaic, Somali, Hamer,
Hausa, Ngizim, Burera, Western Desert, Aranda, Standard Thai, Garo,
Boro, Washkuk, Selepet, Kewa, Chuave, Daribi, Fasu, Yareba, Koiari,
Tzeltal, Otomi, Yacqui, Karok, Ocaina, Carib, Chacobo, Tacana, Jaqaru,
Wapishana, Campa, Moxo, Guarani, Barasano, Tucano, Telugu, Lak, Brahui;
RARE Asmat; LOAN Paez.

Palatalized voiced alveolar flap /r*V 1

Irish.

Voiced alveolar fricative flap /r/ 1

Tacana.

Palatalized voiceless alveolar flap /cj/ 1

Irish.

Velarized voiceless alveolar flap /$/ 1

Irish.

Voiced retroflex flap /r/ 11 (-1)

241
Segment index

Bengali, Punjabi, Kharia, Khmer, Maung, Kaliai, Pawaian, Kunimaipa,

Apinaye, Kota; LOAN Hindi-Urdu.

Voiced dental/alveolar lateral flap /"i'V 3

Luvale, Logbara, Guajiro.

Voiced alveolar lateral flap /I/ 3

Zande, Kewa, Paez.

Palatalized voiced alveolar lateral flap /-I / 1

Paez.

Voiced retroflex lateral flap /I/ 3

Moro, Papago, S. Nambiquara.*

Laryngealized voiced retroflex lateral flap lit 1

S. Nambiquara.

Voiced dental r-sound /rr/ 1

Nimboran.

Voiced dental/alveolar r-sound /"rr'7 25 (-2)

Tarok, Doayo, Songhai, Socotri, Beja, Kullo, Dizi, Kefa, Gugu-Yalanji,
Mabuiag, Yay, Atayal, Cham, Jingpho, Boro, Chatino, Achumawi, Yana,
Shasta, Bribri, Abipon, Cofan, Burushaski; LOAN Tiwa, Wappo.

Laryngealized voiced dental/alveolar r-sound /"rr"/ 2 (-1)

Wichita; LOAN Wappo.

Voiced alveolar r-sound /rr/ 5

Iraqw, Iwam, Cayapa, Muinane, Ainu.

Palatalized voiced alveolar r-sound /rr / 1

Muinane•

8. APPROX1MANTS

Voiced dental lateral approximant /J/ 26 (-1)

Lithuanian, French, Spanish, Punjabi, Finnish, Tuva, Manchu, Gbeya,
Arabic, Tuareg, Mundari, Nunggubuyu, Western Desert, Aranda,
Kariera-Ngarluma, Arabana-Wanganura, Diyari, Tamang, Dafla, Tzeltal,
Luiseflo, Tonkawa, Gununa-Kena, Araucanian, Greenlandic; LOAN
Guarani.

Long voiced dental lateral approximant / J,: / 3

Punjabi, Finnish, Arabic.

Palatalized voiced dental lateral approximant /j, / 1

Russian.

Voiceless dental lateral approximant /W 1

Gununa-Kena.

Voiced dental/alveolar lateral approximant /"l'7 122 (-3)

Greek, Farsi, Kurdish, Hindi-Urdu, Bengali, Kashmiri, E. Armenian,
Ostyak, Hungarian, Lappish, Yurak, Tavgy, Osmanli, Azerbaijani, Chuvash,
Kirghiz, Khalaj, Mongolian, Evenki, Goldi, Korean, Katcha, Wolof,

242
 Segment index

Dagbani, Senadi, Ewe, Igbo, G2, Tarok, Beembe, Swahili, Doayo, Songhai,
Kanuri, Maasai, Nubian, Ik, Sebei, Nera, Mursi, Yulu, Berta, Hebrew,
Socotri, Shilha, Awiya, Beja, Kullo, Kefa, Hamer, Kharia, Vietnamese,
Sedang, Maranungku, Wik-Munkan, Kariera-Ngarluma, Gugu-Yalanji, Mabuiag,
Standard Thai, Lakkia, Yay, Sui, Po-ai, Lungchow, Sundanese, Javanese,
Malagasy, Cham, Malay, Batak, Tagalog, Tsou, Kaliai, Hawaiian, Mandarin,
Taishan, Hakka, Changchow, Fuchow, Kan, Burmese, Lahu, Jingpho, Yao,
Andamanese, Telefol, Dani, Kunimaipa, Taoripi, Nambakaengo, Haida,
Chipewyan, Tolowa, Nez Perce, Chontal, Mazahua, Chatino, Squamish, Tiwa,
Porno, Yana, Zuni, Delaware, Wiyot, Dakota, Yuchi, Alabama, Wappo,
Itonama, Abipon, Quechua, Island Carib, Aleut, Kurukh, Yukaghir, Gilyak,
Lak, Basque, Burushaski; LOAN Tarascan, Ojibwa; ?DERIVED Mixtec.

Long voiced dental/alveolar lateral approximant /"I:"/ 4

Chuvash, Wolof, Shilha, Delaware.

Palatalized voiced dental/alveolar lateral approximant /" I / 5

Bulgarian, Yurak, Chuvash, Amuesha, Ket.

Velarized voiced dental/alveolar lateral approximant /"f V 6

Bulgarian, Yakut, Khmer, Ket, Yukaghir, Georgian.

Long velarized voiced dental/alveolar lateral approximant /"t:"/ 1

Yakut.
9
Pharyngealized voiced dental/alveolar lateral approximant /"I "/ 1
Shilha.

Breathy voiced dental/alveolar lateral approximant / " I " / 1

Hindi-Urdu.

Laryngealized voiced dental/alveolar lateral approximant /"I"/ 4

Sedang, Haida, Yuchi, Wappo.

Voiceless dental/alveolar lateral approximant / " [ " / 5

Mongolian, Sedang, Burmese, Yao, Chipewyan.

Voiced alveolar lateral approximant /I/ 93 (-3)

Breton, German, Norwegian, Romanian, Sinhalese, Albanian, Komi, Bashkir,
Moro, Kadugli, Kpelle, Bisa, Bambara, Dan, Diola, Temne, Tampulma,
Bariba, Lelemi, Amo, Zulu, Teke, Maba, Fur, Luo, Nyangi, Sebei, Tama,
Temein, Tabi, Sara, Kunama, Koma, Tigre, Amharic, Neo-Aramaic, Somali,
Iraqw, Hamer, Hausa, Angas, Margi, Ngizim, Kanakuru, Maung, Tiwi,
Burera, Nunggubuyu, Alawa, Malakmalak, Bardi, Kunjen, Western Desert,
Nyangumata, Aranda, Arabana-Wanganura, Diyari, Bandjalang, Saek, Sa'ban,
Chamorro, Rukai, Ao, Tiddim Chin, Boro, Karen, Selepet, Navaho, Hupa,
Nez Perce, Klamath, Maidu, Wintu, Zoque, Totonac, K'ekchi, Kwakw'ala,
Quileute, Yacqui, Diegueflo, Achumawi, Tunica, Cayapa, Jaqaru, Guahibo,
Araucanian, Telugu, Kota, Malayalam, Brahui; RARE Ashuslay; LOAN Otomi,
Moxo.

Long voiced alveolar lateral approximant /1 : / 1

Somali.

Palatalized voiced alveolar lateral approximant /1 V 3

Irish, Lithuanian, Nyangumata.

243
Segment index

Velarized voiced alveolar lateral approximant /+/ 2

Irish, Albanian.
<p
Pharyngealized voiced alveolar lateral approximant /I / 1
Kurdish.

Laryngealized voiced alveolar lateral approximant /I/ 4

Tiddim Chin, Nez Perce, Klamath, Kwakw'ala.

Voiceless alveolar lateral approximant /|/ 1

Klamath.

Palatalized voiceless alveolar lateral approximant l\ I 1 (-1)

7DERIVED Irish.

Velarized voiceless alveolar lateral approximant /£/ 1 ("1)

7DERIVED Irish.

Voiced palato-alveolar lateral approximant I\J 6

Alawa, Malakmalak, Bardi, Aranda, Dieguefio, Cayapa.

Voiced retroflex lateral approximant / } / 22 (-1)

Norwegian, Punjabi, Ostyak, Kanuri, Mundari, Khasi, Maung, Nunggubuyu,
Alawa, Bardi, Western Desert, Nyangumata, Kariera-Ngarluma,
Arabana-Wanganura, Diyari, Rukai, Iai, Tarascan, Telugu, Kota,
Malayalam; 7DERIVED Tiwi.

Voiceless retroflex lateral approximant /[/ 1

Iai.

Voiced palatal lateral approximant /A/ 15 (-2)

Spanish, Ostyak, Cheremis, Komi, Tavgy, Kariera-Ngarluma,
Arabana-Wanganura, Diyari, Quechua, Jaqaru, Araucanian, Malayalam,
Basque; LOAN Osmanli, Guarani.

Velarized voiced palatal lateral approximant /k/ 1

Irish.

Voiced velar lateral approximant /cj/ 1

Yagaria.

Voiced bilabial approximant /(3/ 6

Hindi-Urdu, Lappish, Lakkia, Kunimaipa, Karok, Telugu.

Long voiced bilabial approximant /{3:/ 1

Telugu.

Voiced labio-dental approximant /v/ 6

Norwegian, Sinhalese, Luvale, Khmer, Malayalam, Ket.

Voiced dental/alveolar approximant / " J " / 1

Andamanese.

Voiced alveolar approximant /J/ 11 (-1)

Kanakuru, Maung, Burera, Maranungku, Malakmalak, Saek, Iai, Wintu,
Diegueflo, Wiyot; RARE Tarascan.

244
 Segment index

Voiced retroflex approximant / j / 15

Tiwi, Nunggubuyu, Alawa, Bardi, Kunjen, Western Desert, Nyangumata,
Aranda, Kariera-Ngarluma, Gugu-Yalanji, Arabana-Wanganura, Diyari,
Chamorro, Kan, Wiyot,

Voiced palatal approximant / j / 271 (-1)

Greek, Irish, Breton, German, Norwegian, Lithuanian, Russian, Bulgarian,
French, Spanish, Romanian, Farsi, Pashto, Kurdish, Hindi-Urdu, Bengali,
Kashmiri, Punjabi, Sinhalese, Albanian, E. Armenian, Ostyak, Cheremis,
Komi, Finnish, Hungarian, Lappish, Yurak, Tavgy, Osmanli, Chuvash,
Yakut, Kirghiz, Bashkir, Tuva, Mongolian, Evenki, Manchu, Korean,
Japanese, Katcha, Moro, Kadugli, Kpelle, Bisa, Bambara, Dan, Wolof,
Diola, Temne, Dagbani, Senadi, Tampulma, Bariba, Ewe, Akan, Igbo, G3,
Lelemi, Efik, Birom, Tarok, Amo, Beembe, Swahili, Luvale, Zulu, Teke,
Doayo, Gbeya, Zande, Songhai, Kanuri, Maba, Fur, Maasai, Luo, Nyangi,
Ik, Sebei, Tama, Temein, Nera, Tabi, Mursi, Logbara, Yulu, Sara, Berta,
Kunama, Koma, Arabic, Tigre, Amharic, Hebrew, Socotri, Neo-Aramaic,
Shilha, Tuareg, Somali, Awiya, Iraqw, Beja, Kullo, Dizi, Kefa, Hamer,
Hausa, Angas, Margi, Ngizim, Kanakuru, Mundari, Vietnamese, Sedang,
Khmer, Maung, Tiwi, Burera, Nunggubuyu, Alawa, Maranungku, Malakmalak,
Bardi, Wik-Munkan, Kunjen, Western Desert, Nyangumata, Aranda,
Kariera-Ngarluma, Gugu-Yalanji, Mabuiag, Arabana-Wanganura, Diyari,
Bandjalang, Standard Thai, Lakkia, Yay, Sui, Saek, Po-ai, Lungchow,
Sundanese, Javanese, Malagasy, Cham, Batak, Tagalog, Sa'ban, Chamorro,
Rukai, Adzera, Mandarin, Taishan, Changchow, Tamang, Dafla, Burmese,
Lahu, Jingpho, Ao, Boro, Karen, Yao, Andamanese, Washkuk, Sentani, Iwam,
Telefol, Selepet, Gadsup, Yagaria, Kewa, Chuave, Pawaian, Dani, Daribi,
Fasu, Suena, Dera, Yareba, Nambakaengo, Haida, Tlingit, Navaho,
Chipewyan, Tolowa, Hupa, Nez Perce, Klamath, Maidu, Wintu, Chontal,
Zoque, Totonac, K'ekchi, Mixe, Otomi, Mazahua, Mazatec, Chatino, Nootka,
Kwakw'ala, Quileute, Squamish, Puget Sound, Papago, Luisefio, Hopi,
Yacqui, Tiwa, Karok, Porno, Dieguefio, Achumawi, Yana, Shasta, Tarascan,
Zuni, Acoma, Delaware, Tonkawa, Wiyot, Seneca, Dakota, Yuchi, Tunica,
Alabama, Wappo, Itonama, Bribri, Cayapa, Paez, Yukaghir, Amahuaca,
Tacana, Cashinahua, Ashuslay, Abipon, S, Nambiquara, Arabela, Quechua,
Jaqaru, Gununa-Kena, Island Carib, Amuesha, Campa, Guajiro, Moxo,
Siriono, Guahibo, Barasano, Siona, Tucano, Jivaro, Cofan, Araucanian,
Aleut, Telugu, Kota, Kurukh, Malayalam, Yukaghir, Chukchi, Gilyak,
Kabardian, Lak, Burushaski, Ainu, !Xu; 7DERIVED Atayal.

Long voiced palatal approximant / j : / 2

Chuvash, Arab ic.

Nasalized voiced palatal approximant / j / 2

Yakut, Kharia.

Laryngealized voiced palatal approximant / j / 13

Logbara, Hausa, Sui, Haida, Nez Perce, Klamath, Otomi, Mazahua, Nootka,
Kwakw'ala, Acoma, Yuchi, Wappo.

Voiceless palatal approximant /j/ 7

Malagasy, Yao, Klamath, Otomi,0Mazahua, Hopi, Aleut.

Lowered voiced palatal approximant A / 1

KhasiT

245
Segment index

Mid palatal approximant /"e'V 1

Bengali.

Lower mid back approximant / A / 1

Vietnamese,

Voiced velar approximant /v/ 5

Kanakuru, Aranda, Adzera, Wiyot, Cofan.

Nasalized voiced velar approximant /y/ 1

Japanese. *

Voiced uvular approximant /g/ 1

E. Armenian.

Voiced labial-palatal approximant /i|/ 4

Breton, French, GS, Mandarin.

Voiced labial-velar approximant /w/ 238 (-5)

Irish, Breton, French, Spanish, Romanian, Farsi, Pashto, Kurdish,
Bengali, Kashmiri, Punjabi, Ostyak, Hungarian, Yurak, Chuvash, Bashkir,
Korean, Katcha, Moro, Kadugli, Bisa, Bambara, Dan, Wolof, Diola, Temne,
Dagbani, Senadi, Tampulma, Bariba, Ewe, Akan, Igbo, GS, Lelemi, Efik,
Birom, Tarok, Amo, Beembe, Swahili, Luvale, Zulu, Teke, Doayo, Gbeya,
Zande, Songhai, Kanuri, Maba, Fur, Maasai, Luo, Nyangi, Ik, Sebei, Tama,
Temein, Nera, Tabi, Mursi, Logbara, Yulu, Sara, Berta, Kunama, Koma,
Arabic, Tigre, Amharic, Socotri, Shilha, Tuareg, Somali, Awiya, Iraqw,
Beja, Kullo, Dizi, Kefa, Hamer, Hausa, Angas, Margi, Ngizim, Kanakuru,
Mundari, Kharia, Vietnamese, Sedang, Tiwi, Burera, Nunggubuyu, Alawa,
Maranungku, Malakmalak, Bardi, Wik-Munkan, Kunjen, Western Desert,
Nyangumata, Aranda, Kariera-Ngarluma, Gugu-Yalanji, Mabuiag,
Arabana-Wanganura, Diyari, Bandjalang, Standard Thai, Lakkia, Sui,
Sundanese, Javanese, Malagasy, Cham, Batak, Tagalog, Sa'ban, Chamorro,
Rukai, Iai, Maori, Hawaiian, Mandarin, Taishan, Hakka, Changchow,
Tamang, Burmese, Jingpho, Ao, Garo, Karen, Yao, Andamanese, Asmat,
Washkuk, Sentani, Iwam, Telefol, Selepet, Kewa, Chuave, Pawaian, Dani,
Daribi, Fasu, Suena, Dera, Yareba, Nambakaengo, Haida, Tlingit, Tolowa,
Hupa, Nez Perce, Klamath, Maidu, Wintu, Chontal, Zoque, Tzeltal,
Totonac, K'ekchi, Otomi, Mazahua, Chatino, Nootka, Kwakw'ala, Quileute,
Squamish, Puget Sound, Papago, Luisefto, Hopi, Yacqui, Tiwa, Porno,
Diegueflo, Achumawi, Yana, Shasta, Tarascan, Zuni, Acoma, Delaware,
Tonkawa, Wiyot, Seneca, Wichita, Dakota, Yuchi, Tunica, Alabama,
Itonama, Bribri, Cayapa, Paez, Carib, Amahuaca, Chacobo, Tacana,
Cashinahua, Ashuslay, Abipon, S. Nambiquara, Arabela, Auca, Quechua,
Jaqaru, Gununa-Kena, Wapishana, Island Carib, Amuesha, Guajiro, Siriono,
Ticuna, Barasano, Siona, Tucano, Jivaro, Cofan, Araucanian, Aleut,
Kurukh, Yukaghir, Chukchi, Gilyak, Kabardian, Lak, Burushaski, Ainu,
!Xu; RARE Japanese, Navaho; LOAN Lahu; 7DERIVED Atayal, Moxo.

Long voiced labial-velar approximant /w:/ 3

Chuvash, Arabic, Delaware.

Nasalized voiced labial-velar approximant /w/ 1

Breton.

246
 Segment index

Laryngealized voiced labial-velar approximant /w/ 14 (-1)

Logbara, Sedang, Sui, Tiddim Chin, Haida, Nez Perce, Klamath, Ototni,
Mazahua, Nootka, Kwakw'ala, Acoma, Yuchi; RARE Wichita.

Voiceless labial-velar approximant /M/ 11 (-1)

Sedang, Lakkia, Kaliai, Iai, Yao, Klamath, Otomi, Mazahua, Hopi, Aleut;
RARE Burmese,

Lowered voiced labial-velar approximant /o/ 1

Khasi.

Higher mid voiced labial-velar approximant /o/ 1

Bengali. ~

9. VOWELS

High front unrounded vowel /]/ 271 (-1)

Greek, Breton, Russian, Bulgarian, French, Spanish, Romanian, Farsi,
Pashto, Kashmiri, Punjabi, Albanian, E. Armenian, Ostyak, Cheremis,
Komi, Finnish, Hungarian, Lappish, Yurak, Tavgy, Osmanli, Chuvash,
Yakut, Bashkir, Khalaj, Tuva, Mongolian, Goldi, Manchu, Korean,
Japanese, Katcha, Moro, Kadugli, Kpelle, Bisa, Bambara, Dan, Wolof,
Diola, Temne, Dagbani, Senadi, Tampulma, Bariba, Ewe, Akan, Igbo, G3,
Lelemi, Efik, Birom, Tarok, Amo, Beembe, Swahili, Luvale, Zulu, Teke,
Doayo, Gbeya, Zande, Songhai, Kanuri, Maba, Fur, Maasai, Luo, Nubian,
Nyangi, Ik, Sebei, Tama, Temein, Nera, Tabi, Mursi, Logbara, Yulu, Sara,
Berta, Kunama, Koma, Amharic, Socotri, Neo-Aramaic, Shilha, Tuareg,
Somali, Awiya, Beja, Kullo, Dizi, Kefa, Hamer, Ngizim, Kanakuru, Kharia,
Khasi, Sedang, Maung, Tiwi, Burera, Maranungku, Malakmalak, Bardi,
Wik-Munkan, Kunjen, Western Desert, Nyangumata, Aranda,
Kariera-Ngarluma, Gugu-Yalanji, Mabuiag, Arabana-Wanganura, Diyari,
Bandjalang, Standard Thai, Lakkia, Yay, Sui, Saek, Po-ai, Lungchow,
Atayal, Sundanese, Javanese, Malagasy, Cham, Malay, Batak, Sa'ban,
Chamorro, Rukai, Tsou, Adzera, Roro, Kaliai, Iai, Maori, Hawaiian,
Mandarin, Taishan, Hakka, Changchow, Fuchow, Kan, Tamang, Dafla,
Burmese, Lahu, Jingpho, Ao, Tiddim Chin, Garo, Boro, Karen, Yao,
Andamanese, Asmat, Washkuk, Sentani, Nimboran, Iwam, Telefol, Selepet,
Gadsup, Yagaria, Kewa, Chuave, Pawaian, Dani, Wantoat, Daribi, Fasu,
Suena, Dera, Kunimaipa, Yareba, Koiari, Taoripi, Nasioi, Rotokas,
Nambakaengo, Haida, Tlingit, Chipewyan, Tolowa, Klamath, Maidu, Wintu,
Chontal, Zoque, Totonac, K'ekchi, Mixe, Otomi, Mazahua, Mazatec, Mixtec,
Chatino, Nootka, Kwakw'ala, Quileute, Papago, Hopi, Yacqui, Tiwa, Karok,
Yana, Shasta, Tarascan, Zuni, Acoma, Delaware, Tonkawa, Wiyot, Seneca,
Wichita, Dakota, Yuchi, Tunica, Wappo, Itonama, Bribri, Mura, Cayapa,
Paez, Ocaina, Muinane, Carib, Apinaye, Chacobo, Tacana, Cashinahua,
Ashuslay, Abipon, S. Nambiquara, Arabela, Auca, Jaqaru, Wapishana,
Island Carib, Campa, Guajiro, Moxo, Guarani, Siriono, Guahibo, Ticuna,
Barasano, Siona, Tucano, Jivaro, Cofan, Araucanian, Greenlandic, Aleut,
Telugu, Kota, Kurukh, Malayalam, Ket, Lak, Nama, Burushaski, Ainu,
Brahui, !Xu; 7DERIVED Margi.

Long high front unrounded vowel /i :/ 41 (-1)

Irish, Breton, German, Norwegian, Lithuanian, Kurdish, Hindi-Urdu,
Sinhalese, Finnish, Evenki, Korean, Wolof, Dagbani, Fur, Arabic, Tigre,
Neo-Aramaic, Tuareg, Iraqw, Hausa, Angas, Khmer, Bardi, Lakkia, Atayal,
Adzera, Iai, Yagaria, Navaho, Chipewyan, Tolowa, Karok, Ojibwa,
Delaware, Tonkawa, Telugu, Kabardian, Lak, Brahui, !Xu; RARE Hungarian.

247
Segment index

Overshort high front unrounded vowel /i/ 3

Yurak, Po-ai, Lungchow.

Nasalized high front unrounded vowel /T/ 53 (-1)

Bulgarian, Kashmiri, Punjabi, Kpelle, Bambara, Senadi, Bariba, Ewe,
Akan, Igbo, G3, Lelemi, Beembe, Doayo, Gbeya, Zande, Sara, Kharia,
Changchow, Pawaian, Daribi, Fasu, Nambakaengo, Chipewyan, Tolowa, Otomi,
Mazahua, Mazatec, Mixtec, Chatino, Tiwa, Dakota, Paez, Ocaina, Apinaye,
Cashinahua, S. Nambiquara, Auca, Wapishana, Island Carib, Guajiro,
Guarani, Siriono, Guahibo, Ticuna, Barasano, Siona, Tucano, Jivaro,
Kurukh, Nama, !Xu; RARE Karok.

Long nasalized high front unrounded vowel /T;/ 8

Irish, Hindi-Urdu, Lakkia, Navaho, Chipewyan, Tolowa, Ojibwa, Delaware.

High front unrounded vowel with velar stricture /i'/ 1 (-1)

OBSCURE Siriono. ~^

Nasalised high front unrounded vowel with velar stricture /T^/ 1 (-1)
OBSCURE Siriono.

Pharyngealized high front unrounded vowel /i / 1

Neo-Aramaic.

Long pharyngealized high front unrounded vowel /I :/ 2

Evenki, Neo-Aramaic•

Breathy voiced high front unrounded vowel /i/ 1

Tamang.

Laryngealized high front unrounded vowel /i/ 2

Sedang, S. Nambiquara.

Laryngealized nasalized high front unrounded vowel /T/ 1

S. Nambiquara.

Voiceless high front unrounded vowel /J/ 2

Ik, Dafla.

High front rounded vowel /y/ 21

Breton, French, Albanian, Cheremis, Finnish, Hungarian, Tavgy, Osmanli,
Chuvash, Yakut, Kirghiz, Bashkir, Khalaj, Tuva, Korean, Iai, Mandarin,
Changchow, Fuchow, Kan, Tzeltal.

Long high front rounded vowel /yi/ 5 (-1)

Breton, German, Norwegian, Finnish; RARE Hungarian.

Lowered high front unrounded vowel /i/ 54 (-1)

Irish, German, Norwegian, Lithuanian, Kurdish, Hindi-Urdu, Bengali,
Punjabi, Sinhalese, Azerbaijani, Kirghiz, Evenki, Kpelle, Diola,
Tampulma, Akan, Amo, Zande, Maasai, Luo, Ik, Tama, Tabi, Logbara,
Kunama, Arabic, Hebrew, Somali, Iraqw, Hausa, Angas, Mundari,
Vietnamese, Khmer, Nunggubuyu, Alawa, Tagalog, Dani, Navaho, Nez Perce,
Tzeltal, Puget Sound, Luiseflo, Porno, Diegueflo, Ojibwa, Amahuaca,
Quechua, Yukaghir, Chukchi, Gilyak, Georgian, Basque; 7DERIVED Hupa.

248
 Segment Index

Long lowered high front unrounded vowel A s / 1

Mongolian.

Nasalized lowered high front unrounded vowel 111 9

Irish, Hindi-Urdu, Bengali, Punjabi, Kpelle, Akan, Burmese, Navaho,
Amahuaca.

Pharyngealized lowered high front unrounded vowel /i / 2

Evenki, Hamer.

Lowered high front rounded vowel /Y/ 3

German, Norwegian, Azerbaijani.

Higher mid front unrounded vowel /e/ 83 (-1)

French, Romanian, Farsi, Kashmiri, Mongolian, Goldi, Korean, Katcha,
Bisa, Bambara, Dan, Wolof, Diola, Temne, Senadi, Tampulma, Bariba, Ewe,
Akan, Igbo, G2, Lelemi, Efik, Birom, Amo, Swahili, Luvale, Doayo, Gbeya,
Maba, Fur, Maasai, Nyangi, Ik, Sebei, Tama, Temein, Tabi, Sara, Koma,
Hamer, Kharia, Vietnamese, Sedang, Standard Thai, Lakkia, Saek, Po-ai,
Javanese, Cham, Malay, Iai, Amoy, Tamang, Burmese, Lahu, Ao, Garo, Boro,
Karen, Yao, Asmat, Washkuk, Sentani, Kunimaipa, Chipewyan, Otomi,
Mazahua, Squamish, Tiwa, Seneca, Yuchi, Tunica, Alabama, Apinaye, Campa,
Guarani, Siriono, Telugu, Chukchi, Nama, Brahui; RARE Logbara.

Long higher mid front unrounded vowel lei I 23 (-1)

Irish, German, Norwegian, Lithuanian, Hindi-Urdu, Sinhalese, Hungarian,
Mongolian, Korean, Wolof, Dagbani, Arabic, Iraqw, Khmer, Iai, Gadsup,
Karok, Diegueflo, Tonkawa, Wichita, Telugu, Brahui; 7DERIVED Finnish.

Nasalized higher mid front unrounded vowel lei 9

Kashmiri, Bambara, Ewe, Igbo, Zande, Mazahua, Tiwa, Guarani, Siriono.

Long nasalized higher mid front unrounded vowel lei I 2

Irish, Hindi-Urdu.

Higher mid retracted front unrounded vowel lei 1

Karen.

Breathy voiced higher mid front unrounded vowel lei 1

Tamang.

Laryngealized higher mid front unrounded vowel lei 1

Sedang.

Higher mid front rounded vowel /#/ 15 (-1)

Breton, Cheremis, Finnish, Hungarian, Azerbaijani, Kirghiz, Bashkir,
Khalaj, Tuva, Manchu, Korean, Wolof, Hopi, Gilyak; OBSCURE Guajiro.

Long higher mid front rounded vowel l$:l 6 (-2)

Breton, German, Norwegian, Hungarian; RARE Akan; ?DERIVED Finnish.

Overshort higher mid front rounded vowel /$7 1

Chuvash.

Nasalized higher mid front rounded vowel /$/ 1 (-1)

OBSCURE Guajiro.

249
Segment index

Mid front unrounded vowel /"e'7 113 (-2)

Irish, Breton, Norwegian, Spanish, Bengali, Sinhalese, Albanian, Ostyak,
Komi, Lappish, Yurak, Tavgy, Chuvash, Tuva, Manchu, Moro, Kadugli,
Tarok, Teke, Zande, Songhai, Kanuri, Luo, Nubian, Nera, Mursi, Yulu,
Berta, Kunama, Amharic, Socotri, Neo-Aramaic, Tuareg, Somali, Awiya,
Beja, Kullo, Dizi, Kefa, Hausa, Angas, Kanakuru, Mundari, Khasi,
Malakmalak, Mabuiag, Sui, Lungchow, Malagasy, Chamorro, Tsou, Roro,
Kaliai, Lahu, Jingpho, Andamanese, Nimboran, Iwam, Kewa, Chuave, Dani,
Wantoat, Daribi, Fasu, Dera, Yareba, Koiari, Taoripi, Rotokas,
Nambakaengo, Tlingit, Tolowa, Hupa, Klamath, Maidu, Wintu, Chontal,
Tzeltal, K'ekchi, Mixe, Mixtec, Chatino, Luiseflo, Yacqui, Achumawi,
Yana, Shasta, Wiyot, Wappo, Itonama, Bribri, Tacana, Ashuslay, Abipon,
S. Nambiquara, Arabela, Auca, Gununa-Kena, Island Carib, Guahibo,
Ticuna, Barasano, Siona, Tucano, Kota, Kurukh, Malayalam, Georgian,
Basque, Burushaski, !XQ; RARE Alawa, Bandjalang.

Long mid front unrounded vowel /"e:"/ 10

Pashto, Kurdish, Tigre, Tuareg, Hausa, Ngizim, Tolowa, Hupa, Kabardian,
!Xu".

Overshort mid front unrounded vowel /"e"/ 1

Ostyak.

Nasalized front mid unrounded vowel /"§"/ 22 (-2)

Irish, Bengali, Punjabi, Dan, Zande, Sara, Daribi, Fasu, Zoque, Chatino,
Yuchi, Bribri, S. Nambiquara, Auca, Island Carib, Guahibo, Barasano,
Siona, Tucano, Kurukh; RARE Songhai, Mixtec.

Pharyngealized mid front unrounded vowel /"e "/ 3 (-1)

Neo-Aramaic, Hamer; ?DERIVED Lak.

Laryngealized mid front unrounded vowel /"e'V 1

S. Nambiquara.

Laryngealized nasalized front mid unrounded vowel /"§"/ 1

S. Nambiquara.

Voiceless mid front unrounded vowel /"§"/ 1

Ik.

Lower mid front unrounded vowel /e/ 116 (-5)

Greek, German, Russian, Bulgarian, Cheremis, Finnish, Hungarian,
Osmanli, Azerbaijani, Yakut, Kirghiz, Evenki, Japanese, Katcha, Kpelle,
Bambara, Dan, Wolof, Diola, Temne, Senadi, Tampulma, Bariba, Ewe, Igbo,
G3, Lelemi, Efik, Birom, Amo, Beembe, Zulu, Doayo, Gbeya, Maba, Fur,
Maasai, Nyangi, Ik, Tama, Temein, Tabi, Logbara, Sara, Koma, Hebrew,
Somali, Iraqw, Angas, Vietnamese, Sedang, Khmer, Maung, Burera,
Wik-Munkan, Kunjen, Lakkia, Yay, Saek, Po-ai, Atayal, Sundanese,
Javanese, Cham, Batak, Sa'ban, Maori, Hawaiian, Hakka, Fuchow, Kan,
Dafla, Burmese, Tiddim Chin, Karen, Andamanese, Washkuk, Sentani,
Selepet, Yagaria, Pawaian, Suena, Nasioi, Navaho, Zoque, Mazahua,
Mazatec, Nootka, Kwakw'ala, Porno, Tarascan, Zuni, Acoma, Delaware,
Tonkawa, Wichita, Dakota, Tunica, Cayapa, Paez, Ocaina, Muinane, Carib,
Apinaye, Guajiro, Moxo, Cofan, Araucanian, Yukaghir, Chukchi, Ainu; RARE
French, Karok; LOAN Margi, Quechua; 7DERIVED Iai.

250
 Segment index

Long lower mid front unrounded vowel /e:/ 12 (-2)

Breton, German, Evenki, Wolof, Khmer, Lakkia, Telefol, Navaho, Ojibwa,
Wichita; RARE French; 7DERIVED Iai.

Overshort lower mid front unrounded vowel /e/ 1

Po-ai.

Nasalized lower mid front unrounded vowel /e/ 20

Bulgarian, Kpelle, Bambara, Senadi, Bariba, Ewe, Igbo, G2, Lelemi,
Beembe, Doayo, Kharia, Lakkia, Pawaian, Navaho, Mazatec, Seneca, Paez,
Apinaye, Guajiro.

Long nasalized lower mid front unrounded vowel /e:/ 4 (-1)

Lakkia, Navaho, Ojibwa; LOAN Breton.

Laryngealized lower mid front unrounded vowel /e/ 1

Sedang.

Lower mid front rounded vowel /ae/ 7

German, Norwegian, French, Osmanli, Yakut, Iai, Fuchow.

Long lower mid front rounded^ vowel /oe:/ 1 (-1)

RARE Akaru

Nasalized lower mid front rounded vowel /&/ 1

French.

Raised low front unrounded vowel /ae/ 38 (-5)

Norwegian, Bengali, Punjabi, Ostyak, Finnish, Tavgy, Bashkir, Korean,
Dan, Luo, Shilha, Tuareg, Somali, Vietnamese, Maranungku, Standard Thai,
Chamorro, Iai, Taishan, Burmese, Yao, Andamanese, Wantoat, Nambakaengo,
Nez Perce, Mixe, Otomi, Hopi, Seneca, Yuchi, Auca, Ket, Gilyak; RARE
Hungarian, Quileute; 7DERIVED Lithuanian, Awiya, Hakka.

Long raised low front unrounded vowel /ae;/ 10

Norwegian, Lithuanian, Sinhalese, Finnish, Korean, Arabic, Tuareg, Iai,
Delaware, Telugu.

Nasalized raised low front unrounded vowel /ae/ 8

French, Bengali, Punjabi, Gbeya, Changchow, Nambakaengo, Otomi, Auca.

Long nasalized raised low front unrounded vowel /ae: / 1

Delaware.

Pharyngealized raised low front unrounded vowel /ae / 1

Lak.

Low front unrounded vowel /a/ 14

Breton, Russian, French, Farsi, Sinhalese, Azerbaijani, Kpelle, Khmer,
Sentani, Pawaian, Acoma, Cashinahua, Yukaghir, Georgian.

Long low front unrounded vowel /a:/ 5

Tigre, Vietnamese, Khmer, Gadsup, Kabardian.

Nasalized low front unrounded vowel /a

/$/
Dan, Pawaian, Cashinahua.

251
Segment index

High central unrounded vowel /+/ 40

Romanian, Kurdish, Kashmiri, Komi, Lappish, Tavgy, Tuva, Dan, Kanuri,
Amharic, Awiya, Standard Thai, Saek, Po-ai, Sundanese, Cham, Sa'ban,
Rukai, Lahu, Washkuk, Nimboran, Maidu, Mixe, Otomi, Papago, Acoma,
Itonama, Muinane, Chacobo, Abipon, Gununa-Kena, Wapishana, Guarani,
Ticuna, Barasano, Siona, Tucano, Cofan, Ket, Kabardian.

Long high central unrounded vowel / + ;/ 2

Angas, Khmer.
\j

Overshort high central unrounded vowel /+/ 3

Kashmiri, Sebei, Po-ai.

Nasalized high central unrounded vowel /?/ 7

Kashmiri, Wapishana, Guarani, Ticuna, Barasano, Siona, Tucano.

Overshort nasalized high central unrounded vowel /?/ l

Kashmiri.

Retroflexed high central unrounded vowel /+/ l

Tarascan.

High central rounded vowel /«/ 6

Norwegian, Ostyak, Mongolian, Yay, Tsou, Nambakaengo.

Long high central rounded vowel /tt:/ 2

Norwegian, Mongolian.

Overshort high central rounded vowel /S/ 1

Sebei.

Lowered high central unrounded vowel /+/ 3 (-1)

Kirghiz, Karen; LOAN Hindi-Urdu. "*"

Lowered high central rounded vowel /&/ 1

Somali.

Higher mid central unrounded vowel /a/ 6 (-1)

Goldi, Dan, Khmer, Standard Thai, Lahu; RARE Iraqw.

Retroflexed higher mid central unrounded vowel /a/ 1

Mandarin.

Mid central unrounded vowel /"a'7 67 (-1)

Breton, German, Norwegian, Romanian, Pashto, Kurdish, Kashmiri, Punjabi,
Albanian, E. Armenian, Cheremis, Komi, Tavgy, Bashkir, Diola, Temne,
Tarok, Fur, Tabi, Sara, Koma, Tigre, Amharic, Socotri, Tuareg, Hausa,
Margi, Kanakuru, Khmer, Maranungku, Malakmalak, Yay, Sui, Saek, Po-ai,
Sundanese, Javanese, Cham, Malay, Tagalog, Sa'ban, Changchow, Kan,
Burmese, Boro, Asmat, Sentani, Iwam, Wantoat, Dera, Nambakaengo, Tolowa,
Chontal, Mixe, Otomi, Mazahua, Kwakw'ala, Squamish, Puget Sound, Tiwa,
Dieguefio, Achumawi, Delaware, Gununa-Kena, Kurukh, Nama; ?DERIVED
Sinhalese.

Long mid central unrounded vowel /"e:"/ 2 (-1)

Khmer; LOAN Sinhalese.

252
 Segment index

Overshort mid central unrounded vowel /"§"/ 3

Sebei, Chukchi, Georgian.

Nasalized mid central unrounded vowel /"§"/ 8

Kashmiri, Punjabi, Dan, Sara, Nambakaengo, Mazahua, Tiwa, Guahibo.

Long nasalized mid central unrounded vowel /"§:"/ 1

Delaware.

Mid central rounded vowel /"e'V 5

Ostyak, Mongolian, Somali, Ket, Yukaghir.

Long mid central rounded vowel /"e:11/ 1

Mongolian.

Overshort mid central rounded vowel /"e'7 2

Ostyak, Sebei.

Lower mid central unrounded vowel /3/ 10

Hindi-Urdu, Dan, Zande, Logbara, Tigre, Iraqw, Sui, Gadsup, Telugu,
Kabardian.

Nasalized lower mid central unrounded vowel /3/ 2

Hindi-Urdu, Zande.

Raised low central unrounded vowel /g/ 6

German, Amo, Tuareg, Dizi, Mazahua, Carib.

Overshort raised low central unrounded vowel /§/ 1 ("1)

RARE Sebei.

Pharyngealized raised low central unrounded vowel /e / 1

Hamer•

Low central unrounded vowel /a/ 274

Irish, Lithuanian, Spanish, Romanian, Bengali, Kashmiri, Punjabi,
Sinhalese, Albanian, E. Armenian, Ostyak, Komi, Finnish, Lappish, Yurak,
Tavgy, Osmanli, Yakut, Kirghiz, Khalaj, Evenki, Goldi, Manchu, Korean,
Japanese, Moro, Kadugli, Kpelle, Bisa, Bambara, Dan, Wolof, Diola,
Temne, Dagbani, Senadi, Tampulma, Bariba, Ewe, Akan, Igbo, GS, Lelemi,
Efik, Birom, Tarok, Amo, Beembe, Swahili, Luvale, Zulu, Teke, Doayo,
Gbeya, Zande, Songhai, Kanuri, Maba, Fur, Maasai, Luo, Nubian, Nyangi,
Ik, Sebei, Tama, Temein, Nera, Tabi, Mursi, Logbara, Yulu, Sara, Berta,
Kunama, Koma, Arabic, Amharic, Hebrew, Socotri, Neo-Aramaic, Somali,
Awiya, Iraqw, Beja, Kullo, Dizi, Kefa, Hausa, Angas, Margi, Ngizim,
Kanakuru, Mundari, Kharia, Khasi, Sedang, Maung, Tiwi, Burera,
Nunggubuyu, Alawa, Maranungku, Malakmalak, Bardi, Wik-Munkan, Kunjen,
Western Desert, Nyangumata, Aranda, Kariera-Ngarluma, Gugu-Yalanji,
Mabuiag, Arabana-Wanganura, Diyari, Bandjalang, Standard Thai, Lakkia,
Yay, Sui, Saek, Po-ai, Lungchow, Atayal, Sundanese, Javanese, Malagasy,
Cham, Malay, Batak, Sa'ban, Rukai, Tsou, Adzera, Roro, Kaliai, Hawaiian,
Mandarin, Taishan, Hakka, Changchow, Amoy, Fuchow, Kan, Tamang, Burmese,
Lahu, Jingpho, Ao, Tiddim Chin, Garo, Boro, Karen, Yao, Andamanese,
Asmat, Washkuk, Nimboran, Iwam, Telefol, Selepet, Yagaria, Kewa, Chuave,
Dani, Wantoat, Daribi, Fasu, Suena, Dera, Kunimaipa, Yareba, Koiari,
Taoripi, Nasioi, Rotokas, Nambakaengo, Haida, Tlingit, Navaho,
Chipewyan, Tolowa, Hupa, Nez Perce, Klamath, Maidu, Wintu, Chontal,

253
Segment index

Zoque, Tzeltal, Totonac, K'ekchi, Mixe, Otomi, Mazahua, Mazatec, Mixtec,

Chatino, Nootka, Kwakw'ala, Quileute, Squamish, Puget Sound, Papago,
Luiseflo, Hopi, Yacqui, Karok, Porno, Diegueflo, Achumawi, Yana, Shasta,
Tarascan, Zuni, Ojibwa, Delaware, Tonkawa, Wiyot, Seneca, Wichita,
Dakota, Yuchi, Tunica, Alabama, Wappo, Itonama, Bribri, Mura, Paez,
Ocaina, Muinane, Apinaye, Amahuaca, Chacobo, Tacana, Ashuslay, Abipon,
S. Nambiquara, Arabela, Auca, Quechua, Jaqaru, Gununa-Kena, Wapishana,
Island Carib, Amuesha, Campa, Guajiro, Moxo, Guarani, Siriono, Guahibo,
Ticuna, Barasano, Siona, Tucano, Jivaro, Cofan, Araucanian, Greenlandic,
Aleut, Kota, Kurukh, Malayalam, Ket, Chukchi, Lak, Nama, Basque,
Burushaski, Ainu, Brahui, !Xti.

Long low central unrounded vowel /a:/ 35 (-1)

German, Lithuanian, Kurdish, Hindi-Urdu, Sinhalese, Finnish, Hungarian,
Evenki, Korean, Wolof, Dagbani, Fur, Neo-Aramaic, Angas, Ngizim, Khasi,
Bardi, Lakkia, Yay, Adzera, Yao, Telefol, Navaho, Chipewyan, Tolowa,
Hupa, Karok, Diegueflo, Ojibwa, Delaware, Tonkawa, Telugu, Lak, !Xu;
?DERIVED Sa'ban.

Overshort low central unrounded vowel /a/ 3

Yurak, Po-ai, Lungchow.

Nasalized low central unrounded vowel /a/ 55 (-3)

Irish, Bengali, Kashmiri, Punjabi, Kpelle, Bambara, Dan, Senadi, Bariba,
Ewe, Akan, Igbo, GS, Lelemi, Beembe, Doayo, Gbeya, Zande, Sara, Kharia,
Lakkia, Daribi, Fasu, Nambakaengo, Navaho, Chipewyan, Tolowa, Otomi,
Mazahua, Mazatec, Mixtec, Dakota, Bribri, Paez, Ocaina, Apinaye,
Amahuaca, S. Nambiquara, Auca, Wapishana, Island Carib, Guajiro,
Guarani, Siriono, Guahibo, Ticuna, Siona, Tucano, Jivaro, Kurukh, Nama,
!Xu; RARE Songhai, Logbara, Karok.

Long nasalized low central unrounded vowel /a:/ 8

Hindi-Urdu, Lakkia, Navaho, Chipewyan, Tolowa, Ojibwa, Delaware, !Xu.

Retroflexed low centra^ unrounded vowel /a/ 1 (-1)

7DERIVED Mandarin.

Pharyngealized low central unrounded vowel /a / 2

Neo-Aramaic, !Xu.

Long pharyngealized low central unrounded vowel /a :/ 2

Neo-Aramaic, !Xu.

Nasalized pharyngealized low central unrounded vowel /a / 1

Long nasalized pharyngealized low central unrounded vowel /a :/ 1

Breathy voiced low central unrounded vowel /a/ 1

Tamang.

Laryngealized low central unrounded vowel /a/ 2

Sedang, S. Nambiquara.

Laryngealized nasalized low central unrounded vowel /a/ 1

S. Nambiquara.

254
 Segment index

Voiceless low central unrounded vowel /a/ 1

Ik.

Overshort low central rounded vowel /*>/ 1 (-1)

RARE Sebei. """

High back unrounded vowel /w/ 20

Ostyak, Osmanli, Chuvash, Yakut, Korean, Japanese, Lungchow, Adzera,
Dafla, Ao, Nimboran, Nez Perce, Ocaina, Carib, Apinaye, Amahuaca,
Cashinahua, Jaqaru, Jivaro, Araucanian.

Long high back unrounded vowel /w:/ 1

Korean.

Overshort high back unrounded vowel /w/ 1

Lungchow.

Nasalized high back unrounded vowel /w/ 5

Apinaye, Amahuaca, Cashinahua, Abipon, Jivaro.

High back rounded vowel /u/ 254 (-1)

Greek, Breton, Norwegian, Russian, Bulgarian, French, Spanish, Romanian,
Farsi, Pashto, Kashmiri, Punjabi, Albanian, E. Armenian, Ostyak,
Cheremis, Komi, Finnish, Hungarian, Lappish, Yurak, Tavgy, Osmanli,
Chuvash, Yakut, Bashkir, Khalaj, Tuva, Evenki, Goldi, Manchu, Korean,
Katcha, Moro, Kadugli, Kpelle, Bisa, Bambara, Dan, Wolof, Temne,
Dagbani, Senadi, Tampulma, Bariba, Ewe, Akan, Igbo, G2, Lelemi, Efik,
Birom, Tarok, Amo, Beembe, Swahili, Luvale, Zulu, Teke, Doayo, Gbeya,
Zande, Songhai, Kanuri, Maba, Fur, Maasai, Luo, Nubian, Nyangi, Ik,
Tama, Temein, Nera, Tabi, Mursi, Logbara, Yulu, Sara, Berta, Kunama,
Koma, Amharic, Socotri, Neo-Aramaic, Tuareg, Awiya, Beja, Kullo, Dizi,
Kefa, Hamer, Angas, Margi, Ngizim, Kanakuru, Kharia, Khasi, Sedang,
Maung, Tiwi, Burera, Malakmalak, Bardi, Kunjen, Western Desert,
Nyangumata, Aranda, Kariera-Ngarluma, Gugu-Yalanji, Mabuiag,
Arabana-Wanganura, Diyari, Bandjalang, Standard Thai, Lakkia, Yay, Sui,
Saek, Po-ai, Lungchow, Atayal, Sundanese, Javanese, Cham, Malay, Batak,
Sa'ban, Chamorro, Rukai, Tsou, Roro, Kaliai, Iai, Maori, Hawaiian,
Mandarin, Taishan, Hakka, Changchow, Amoy, Fuchow, Kan, Tamang, Dafla,
Burmese, Lahu, Jingpho, Ao, Tiddim Chin, Garo, Boro, Karen, Yao,
Andamanese, Asmat, Washkuk, Sentani, Iwam, Telefol, Selepet, Gadsup,
Yagaria, Kewa, Chuave, Pawaian, Dani, Wantoat, Daribi, Fasu, Suena,
Dera, Kunimaipa, Yareba, Koiari, Taoripi, Nasioi, Rotokas, Nambakaengo,
Tlingit, Chipewyan, Tolowa, Maidu, Wintu, Chontal, Zoque, Totonac,
K'ekchi, Mixe, Otomi, Mazahua, Mixtec, Chatino, Kwakw'ala, Quileute,
Papago, Yacqui, Tiwa, Achumawi, Yana, Shasta, Tarascan, Zuni, Acoma,
Delaware, Tonkawa, Wiyot, Wichita, Dakota, Yuchi, Tunica, Wappo,
Itonama, Bribri, Paez, Muinane, Carib, Apinaye, Cashinahua, Ashuslay, S.
Nambiquara, Arabela, Gununa-Kena, Wapishana, Island Carib, Guajiro,
Moxo, Guarani, Siriono, Guahibo, Ticuna, Barasano, Siona, Tucano,
Jivaro, Araucanian, Greenlandic, Aleut, Telugu, Kota, Kurukh, Malayalam,
Ket, Yukaghir, Chukchi, Gilyak, Lak, Nama, Basque, Burushaski, Ainu,
Brahui, !Xu; RARE Seneca.

Long high back rounded vowel /u:/ 37 (-1)

Irish, Breton, German, Norwegian, Lithuanian, Kurdish, Hindi-Urdu,
Sinhalese, Finnish, Hungarian, Evenki, Korean, Wolof, Dagbani, Fur,
Arabic, Tigre, Neo-Aramaic, Tuareg, Iraqw, Hausa, Khmer, Bardi, Lakkia,

255
Segment index

Atayal, Iai, Telefol, Chipewyan, Tolowa, Delaware, Tonkawa, Telugu,

Kabardian, Lak, Brahui, !Xu; RARE Angas.

Overshort high back rounded vowel /u/ 4

Yurak, Sebei, Po-ai, Lungchow.

Nasalized high back rounded vowel /u/ 51 (-2)

Bulgarian, Kashmiri, Punjabi, Kpelle, Bambara, Dan, Senadi, Bariba, Ewe,
Akan, Igbo, G3, Lelemi, Beembe, Doayo, Gbeya, Zande, Sara, Kharia,
Lakkia, Pawaian, Daribi, Fasu, Nambakaengo, Chipewyan, Tolowa, Otomi,
Mazahua, Mixtec, Chatino, Dakota, Bribri, Paez, Apinaye, Cashinahua,
Wapishana, Island Carib, Guajiro, Guarani, Siriono, Guahibo, Ticuna,
Barasano, Siona, Tucano, Jivaro, Kurukh, Nama, !XQ; RARE Songhai, Tiwa.

Long nasalized high back rounded vowel /u:/ 7

Irish, Hindi-Urdu, Lakkia, Chipewyan, Hupa, Delaware, !Xu.

Pharyngealized high back rounded vowel /u / 1

Neo-Aramaic.

Long pharyngealized high back rounded vowel /u :/ 1

Neo-Aramaic•

Breathy voiced high back rounded vowel /u/ 1

Tamang.

Laryngealized high back rounded vowel /u/ 2

Sedang, S. Nambiquara.

Voiceless high back rounded vowel /y/ 2

Ik, Dafla.

Lowered high back unrounded vowel /ui/ 4

Vietnamese, Nunggubuyu, Alawa, Mandarin.

Lowered high back rounded vowel /o/ 48

Irish, German, Lithuanian, Kurdish, Hindi-Urdu, Bengali, Punjabi,
Sinhalese, Azerbaijani, Kirghiz, Mongolian, Kpelle, Diola, Tampulma,
Akan, Igbo, Amo, Zande, Maasai, Luo, Tama, Logbara, Kunama, Arabic,
Hebrew, Somali, Iraqw, Hausa, Mundari, Vietnamese, Khmer, Maranungku,
Wik-Munkan, Tagalog, Dani, Kunimaipa, Haida, Tzeltal, Puget Sound,
Luiseflo, Karok, Porno, Dieguefio, Ojibwa, Cayapa, Quechua, Georgian,
Brahui.

Long lowered high back rounded vowel /o:/ 1

Karok.

Nasalized lowered high back rounded vowel /o/ 7

Irish, Hindi-Urdu, Bengali, Punjabi, Akan, Igbo, Burmese.

Pharyngealized lowered high back rounded vowel /o / 2

Evenki, Hamer.

Long pharyngealized lowered high back rounded vowel /o : / 1

Evenki.

256
 Segment index

Higher mid back unrounded vowel /*/ 4

Vietnamese, Hopi, Apinaye, Gilyak.

Higher mid fronted back unrounded vowel / Y / 1

Azerbaijani.

Higher mid back rounded vowel /o/ 88 (-1)

Romanian, Farsi, Bengali, Kashmiri, Komi, Finnish, Hungarian,
Azerbaijani, Kirghiz, Khalaj, Goldi, Korean, Katcha, Bisa, Bambara, Dan,
Diola, Temne, Senadi, Tampulma, Bariba, Ewe, Akan, Igbo, G3, Lelemi,
Efik, Birom, Amo, Swahili, Luvale, Doayo, Gbeya, Songhai, Maba, Fur,
Maasai, Nyangi, Ik, Sebei, Tama, Temein, Tabi, Sara, Koma, Hamer,
Sedang, Standard Thai, Saek, Javanese, Malagasy, Cham, Malay, Iai, Amoy,
Tamang, Burmese, Lahu, Ao, Garo, Boro, Karen, Yao, Selepet, Nasioi,
Chipewyan, Mixe, Otomi, Mazahua, Mazatec, Nootka, Squamish, Tiwa,
Tarascan, Dakota, Yuchi, Tunica, Alabama, Ocaina, Apinaye, Amuesha,
Campa, Guarani, Siriono, Barasano, Telugu, Nama; RARE Logbara.

Long higher mid back rounded vowel /o:/ 25 (-2)

Irish, Breton, German, Lithuanian, Hindi-Urdu, Sinhalese, Hungarian,
Korean, Wolof, Dagbani, Arabic, Iraqw, Khmer, Iai, Gadsup, Navaho,
Karok, Diegueflo, Ojibwa, Tonkawa, Wichita, Telugu, Brahui; RARE French;
7DERIVED Finnish.

Overshort higher mid back rounded vowel /o/ 1

Po-ai.

Nasalized higher mid back rounded vowel /o/ 17 (-1)

French, Bengali, Kashmiri, Bambara, Ewe, Igbo, Zande, Lakkia, Mazahua,
Mazatec, Tiwa, Ocaina, Apinaye, Guarani, Siriono, Barasano; RARE
Songhai.

Long nasalized higher mid back rounded vowel /o:/ 4

Irish, Hindi-Urdu, Navaho, Ojibwa.

Breathy voiced higher mid back rounded vowel /o/ 1

Tamang.

Laryngealized higher mid back rounded vowel /o/ 1

Sedang.

Mid back unrounded vowel / ' V 7 6

Bulgarian, Bashkir, Korean, Moro, Nimboran, Island Carib.

Long mid back unrounded vowel /"*:"/ 1

Korean.

Nasalized mid back unrounded vowel /'TV 2

Bulgarian, Island Carib.

Nasalized mid fronted back unrounded vowel /"_*"/ 1

Zoque. """

Mid back rounded vowel /"o"/ 133 (-3)

Irish, Breton, Russian, Spanish, Punjabi, Sinhalese, Albanian, E.
Armenian, Ostyak, Cheremis, Lappish, Yurak, Tavgy, Osmanli, Bashkir,
Tuva, Mongolian, Evenki, Manchu, Moro, Kadugli, Tarok, Teke, Zande,

257
Segment index

Kanuri, Luo, Nubian, Nera, Mursi, Yulu, Berta, Kunama, Amharic, Hebrew,
Socotri, Neo-Aramaic, Tuareg, Awiya, Beja, Kullo, Dizi, Kefa, Hausa,
Angas, Margi, Kanakuru, Mundari, Kharia, Khasi, Vietnamese, Malakmalak,
Kunjen, Mabuiag, Sui, Lungchow, Sundanese, Batak, Chamorro, Tsou,
Adzera, Roro, Kaliai, Maori, Hawaiian, Changchow, Kan, Jingpho,
Andamanese, Washkuk, Iwam, Yagaria, Kewa, Chuave, Pawaian, Dani,
Wantoat, Daribi, Fasu, Dera, Yareba, Koiari, Taoripi, Rotokas,
Nambakaengo, Tlingit, Hupa, Klamath, Maidu, Wintu, Chontal, Zoque,
Tzeltal, K'ekchi, Mixtec, Chatino, Luiseflo, Yacqui, Achumawi, Yana,
Wiyot, Seneca, Wappo, Itonama, Bribri, Mura, Muinane, Amahuaca, Chacobo,
Tacana, Ashuslay, Abipon, S. Nambiquara, Arabela, Auca, Gununa-Kena,
Guahibo, Ticuna, Siona, Tucano, Cofan, Kota, Kurukh, Malayalam,
Yukaghir, Chukchi, Gilyak, Georgian, Basque, Burushaski, !Xu; RARE
Tigre; LOAN Chuvash; 7DERIVED Telugu.

Long mid back rounded vowel /"o:"/ 13

Norwegian, Pashto, Kurdish, Mongolian, Manchu, Tigre, Tuareg, Hausa,
Ngizim, Adzera, Hupa, Kabardian, !Xu.

Overshort mid back rounded vowel / M o"/ 1

Ostyak.

Nasalized mid back rounded vowel /"o"/ 24

Irish, Punjabi, Dan, Zande, Sara, Kharia, Changchow, Pawaian, Daribi,
Fasu, Nambakaengo, Mixtec, Chatino, Yuchi, Bribri, Amahuaca, S.
Nambiquara, Auca, Guahibo, Ticuna, Siona, Tucano, Kurukh, !Xu.

Pharyngealized mid back rounded vowel /"o "/ 3

Neo-Aramaic, Lak, !Xu.

Long pharyngealized mid back rounded vowel /"o " / 1

1X0".

Laryngealized mid back rounded vowel /"o"/ 1

S. Nambiquara,

Laryngealized nasalized mid back rounded vowel / M 6'7 1

S. Nambiquara.

Voiceless mid back rounded vowel /"o"/ 1

Ik.

Lower mid back unrounded vowel /A/ 4

Cheremis, Vietnamese, Dafla, Apinaye.

Nasalized lower mid back unrounded vowel /A/ 1

Apinaye.

Lower mid back rounded vowel /o/ 100 (-3)

Greek, German, Lithuanian, Bulgarian, French, Bengali, Yakut, Mongolian,
Evenki, Japanese, Katcha, Kpelle, Bisa, Bambara, Dan, Wolof, Diola,
Temne, Senadi, Tampulma, Bariba, Ewe, Akan, Igbo, G3, Lelemi, Efik,
Birom, Amo, Beembe, Zulu, Doayo, Gbeya, Maba, Fur, Maasai, Nyangi, Ik,
Sebei, Tama, Temein, Tabi, Logbara, Sara, Koma, Somali, Iraqw,
Vietnamese, Sedang, Khmer, Maung, Burera, Bardi, Wik-Munkan, Standard
Thai, Yay, Saek, Po-ai, Atayal, Javanese, Cham, Sa'ban, Iai, Taishan,
Hakka, Amoy, Fuchow, Dafla, Burmese, Lahu, Tiddim Chin, Karen, Yao,

258
 Segment index

Andamanese, Asmat, Sentani, Pawaian, Suena, Taoripi, Navaho, Nez Perce,

Mixe, Otomi, Mazahua, Kwakw'ala, Papago, Porno, Zuni, Delaware, Tonkawa,
Tunica, Carib, Apinaye, Guajiro, Araucanian, Ket, Ainu; RARE Diegueflo;
LOAN Nootka, Quectiua.

Long lower mid back rounded vowel /o:/ 9 (-1)

Mongolian, Evenki, Wolof, Khmer, Lakkia, Iai, Telefol, Delaware; RARE
Dieguetlo.

Overshort lower mid back rounded vowel /o/ 2

Chuvash, Po-ai.

Nasalized lower mid back rounded vowel /5/ 17

Bulgarian, Bengali, Kpelle, Bambara, Senadi, Bariba, Ewe, Igbo, G3,
Beembe, Doayo, Gbeya, Sara, Pawaian, Navaho, Seneca, Guajiro.

Long nasalized lower mid back rounded vowel /5:/ 2

Lakkia, Delaware.

Pharyngealized lower mid back rounded vowel /o / 1

Hamer.
<^
Long nasalized pharyngealized lower mid back rounded vowel /5 :/ 1
!XQ.

Laryngealized lower mid back rounded vowel /o/ 1

Sedang.

Centralized low back unrounded vowel /^/ 1

Selepet. •+•

Centralized low back rounded vowel /^/ 2

Norwegian, Kirghiz. +

Low back unrounded vowel /&/ 22 (-1)

Greek, Norwegian, Bulgarian, Farsi, Azerbaijani, Chuvash, Bashkir,
Khalaj, Tuva, Mongolian, Dan, Temne, Hamer, Khmer, Chamorro, Maori,
Dafla, Nambakaengo, Tiwa, Cayapa, Ashuslay; LOAN Hindi-Urdu.

Long low back unrounded vowel /&:/ 7

Irish, Norwegian, Pashto, Mongolian, Iraqw, Khmer, Brahui.

Nasalized low back unrounded vowel /a/ 4 (-1)

Bulgarian, Nambakaengo, Tiwa; LOAN Hindi-Urdu.

Long nasalized low back unrounded vowel /a:/ 2

Irish, Breton.

Low back rounded vowel A) / 5

French, Punjabi, Hungarian, Luo, Wichita.

Long low back rounded vowel /© •* / 2

Breton, Wichita.

Overshort low back rounded vowel /o/ 1

Ostyak.

259
Segment index

Nasalized low back rounded vowel /©/ 3

French, Punjabi, Dan,

10. DIPHTHONGS

High front unrounded to mid front unrounded diphthong /i e/ 3

E. Armenian, Evenki, Lakkia.

Nasalized high front unrounded ta mid front unrounded diphthong /Te/ 1

Lakkia.

High front unrounded to low front unrounded diphthong / i a / 1

Evenki.

High front unrounded to low central unrounded diphthong /ia/ 2

Saek, !Xfl.

High front unrounded ^o_ mid back rounded diphthong / i o/ 3

Kurd i sh, Acoma, ! XQ.

Long high front unrounded^ j £ mid back rounded diphthong /io:/ 1

Evenki.

High front unrounded to high back rounded diphthong /iu/ 1

Acoma •

Lowered high front unrounded to mid front unrounded diphthong A e / 1

Gilyak.

Mid front unrounded _to_ high front unrounded diphthong /e i / 6

Kurdish, Khmer, Dani, Nambakaengo, Acoma, !Xu.

Nasalized mid front unrounded to high front unrounded diphthong /el/ 2

Burmese, !XQ.

Mid front unrounded to low central unrounded diphthong /ea/ 1

Tsou. *

Mid front unrounded to mid back rounded diphthong /eo/ 2

Tsouy Acoma.

Nasalized mid front unrounded to_ high back rounded diphthong /eu/ 1
ixa.
Mid front unrounded to high central rounded diphthong /eu/ 1
Tsou.

Lower mid front unrounded to high front unrounded diphthong /e i/ 2

Dagbani, Yagaria.

Low front unrounded to mid front unrounded diphthong /ae/ 1

Hindi-Urdu.

Nasalized low front unrounded to lower mid front unrounded diphthong /ae/ 1
+
Hindi-Urdu.

260
 Segment index

High central unrounded to high front unrounded diphthong /j +/ 1

Acoma.

High central unrounded to low central unrounded diphthong / + a/ 1

Saek.

Mid central unrounded to high central unrounded diphthong /ef/ 1

Khmer.

Mid central unrounded to high front unrounded diphthong /ei / 1

Kurdish.

Mid central unrounded to high back rounded diphthong /eu/ 1

Kurdish.

Lower mid central unrounded to high front unrounded diphthong /3 i/ 1

Angas.

Low central unrounded to high central unrounded diphthong /a + / 1

Po-ai.

Low central unrounded to high front unrounded diphthong /a i/ 5

Kurdish, Dani, Yana, Acoma, Arabela.

Low central unrounded to mid front unrounded diphthong /ae/ 2

Yagaria, !Xu.

Pharyngealized low central unrounded to mid front unrounded diphthong /ae / 1

!Xu.

Nasalized pharyngealized low central unrounded to mid front unrounded

diphthong / a e V 1 ~~
!Xu.

Low central unrounded to mid back rounded diphthong /ao/ 2

Yagaria, !Xu.

Pharyngealized
__ low central unrounded to mid back rounded diphthong /ao / 1

Nasalized pharyngealized low central unrounded to mid back rounded

diphthong /aoV 1
!Xu.

Low central unrounded to high back rounded diphthong /au/ 5

Kurdish, Dani, Yana, Acoma, Arabela.

Low central unrounded to high back unrounded diphthong /aw/ 2

Lungchow, Island Carib.

High back rounded to higher mid back rounded diphthong /uo/ 1

Lakkia.

High back rounded to low central unrounded diphthong /ua/ 1

Saek.

261
Segment index

High back rounded to high front unrounded diphthong /ui/ 4

Kurdish, Yana, Acoma, !Xu.

Nasalized high back rounded to high front unrounded diphthong /GT/ 1

!Xu.

Mid back rounded to high back rounded diphthong /ou/ 4

Khmer, Yagaria, Dani, Hopi.

Nasalized mid back rounded to high back rounded diphthong /ou/ 1

Burmese.

Mid back rounded to low central unrounded diphthong /oa/ 1

!Xu.

Nasalized mid back rounded to low central unrounded diphthong /oa/ 1

!Xu.

Pharyngealized mid back rounded to low central unrounded diphthong /oa / 1

!Xu.

Nasalized phapyngealized mid back rounded to low central unrounded

diphthong /6a / 1
!Xu.

Mid back rounded to mid front unrounded diphthong /oe/ 1

!Xu.

Mid back rounded to high front unrounded diphthong /o i/ 3

Kurdish, Dani, !Xu.

Nasalized mid back rounded to high front unrounded diphthong /oT'/ 1

!Xu.

Pharyngealized mid back rounded to high front unrounded diphthong /oi / 1

!Xu.

Nasalized pharyngealized mid back rounded to high front unrounded diphthong

Tm ToTv i
Lower mid back rounded to higher mid back rounded diphthong /oo/ 1
Hindi-Urdu.

Nasalized lower mid back rounded to higher mid back rounded diphthong/oo/ 1
Hindi-Urdu.

Lower mid back rounded to high back rounded diphthong /ou/ 1

Dagbani.

262
 <<
a. o
bilabial
labio-dental
dental
dental/alveolar
alveolar
palatal
velar
I 2. I
t ! t :
r • • s p r g r & &r
M
^'S'ssr * ? ? a1
3 S S *° S" R- ff
& I
T3_ bilabial palatalized
-* bilabial labialized and
5- velar ized
i^. dental
dental velarized
»+ dental/alveolar
dental/alv. palatalized
dental/alv. velarized
alveolar
alveolar palatalized
CD! O
alveolar velarized
palato-alveolar
palatal
palatal palatalized
^ velar
^_ velar palatalized
variable place
labial-velar
 1
•H
3
,3 2 3 a
Tj
4J
S
>
S> Language
C r-l r-1 X I&
•S| ,_| rH
^ ^H
& 7
O
Z
IM Cd
O e d c t J O W O - U M 6 cd cd cd

ibial
ibial pal ata
• H 4 J 4 - » < u c d i - . c d c d
X ) C C > > - l 4 J i H r H
H
Norwegian (006) c d < u o ) r - < c d < u c d a ) Lithuanian (007) •H -H g g ,

lab].o-dental
labiLo-dental
1 H T 3 T 3 cd O . M O u > .Q X<

vl. aspirated plosive ph th

voiceless plosive P Pj "t" "tJ"

voiced plosive t> d

voiced plosive b bj "d" "d^"

vl. nonsibilant fricative f vl. sibilant affricate »ts" " t s j "

vl. sibilant fricative vd. sibilant affricate "dz"

voiced nasal

voiced flap vd. nonsib. fricative v

vd. lateral approximant vl. sibilant fricative J JJ

vd. central approximant vd. sibilant fricative

voiced nasal m mj

voiced trill rJ

vd. lat. approximant

long
high vd. cent, approximant

higher mid

mid

lower mid long

low high

highe r mid

lower mid
low
 fe* 8. 8. SL
s s s
6 ft s s
labio-dental
3= N= W= dental/alveolar
s s s
s
alveolar
palato-alveolar
palatal
velar
labial-palatal
labial-velar
labial-viar nasalized
a. o <<
a M o i
S S S S 31 B1 R. 2.
"o. bilabial
labio-dental
dental/al1
alveolar
palato-alveolar
palatal
velar
« .• I variable place
 vl.
vl.

993
Vowels
sib.
sib.

voiced t

vd. late
vd. nons
vl. nons
voiced p
voiceles

voiced iu

CD o
CO T3
ive
1 osive

fricat

icative
fricat
Ericate

approx

bilabial
bilabial palatalized
labio-dental
labio-dental palatalized
dental
dental palatalized
dental velarized
dental/alveolar
pala to-alveolar
palato-alv. velarized
palatal
X (O velar
velar palatalized
< < < < < <
(X (- 1 CL (-• O H>
H n> n co 3 s tn co n> to
(U CL CL H- o o H- H- a co
3 3 o" cr >a
S S B § S
rt n
sr r s?
o- x? bilabial
a_ tj bilabial palatalized
labio-dental
labio-dental palatalized
,5 <£ 4 dental/alveolar
-P c£ -+ dental/alv. palatalized
dental/alv. velarized
r+ palato-alveolar
palatal
(o 7T velar
CQ T: velar palatalized
Language
CO T3 I <H
2 o -3 2 2 ^ 43 43
French (010)
3 2 S 3 3 | -3 -S Spanish (Oil)
2 3 .8 2. a
voiceless plosive voiceless plosive
p i
voiced plosive vl. sibilant affricate

vl. nonsibilant fricative vl. nonsibilant fricative

vd. nonsibilant fricative vd. nonsibilant fricative

vl. sibilant fricative vl. sibilant fricative

s /

vd. sibilant fricative voiced nasal

2 3
voiced nasal voiced trill
Q

voiced trill voiced tap

vd. lateral approximant vd. lateral approximant

vd. central approximant vd. central approximant

long
high I. Y high
higher mid
mid
lower mid e.oe o
low
low a v

short nasalized
higher mid

lower mid

low
 893
E ss §
ss
bilabial
labio-dental
j+ a r+ dental/alveolar
alveolar
palato-alveolar
palatal
velar
variable place
labial-velar
I t 2 t I £ I
s sr I i i k S S. 3
IIaI| | I
P. £ £ S S
^ ^ R
labio-dental
dental
dental/alveolar
palato-alveolar
palatal
velar
uvular
•o glottal
variable place
labial-velar
 Language

2
Pashto (014) •3 Kurdish (015)
a
voiceless plosive voiceless plosive

voiced plosive voiced plosive

vl. sibilant affricate "ts" tj vl. sibilant affricate
vd. sibilant affricate "dz" d^ vd. sibilant affricate
vl. nonsibilant fricative vl. nonsibilant fricative
vd. nonsibilant fricative

vl. sibilant fricative "s" J vl. sibilant fricative

vd. sibilant fricative "z" 5 vd. sibilant fricative

voiced nasal voiced nasal

voiced t r i l l voiced trill

voiced lateral fricative voiced flap

vd. central approximant vd. lateral approximant

vd. central approximant

long
long
high
high
mid
mid
low
low
oi
ui
 Q. 1-1 O. O O H O Z I t
«< 5
I- H CT
f
l-h l-h CD
i-t i-l H- H- <j Hi S, O* *
8
11 »8
labio-dental
dental
dental/alveolar
palato-alveolar
retroflex
palatal
H- P
1 velar
uvular
Oj +Q), O +Q)
O* Q)' O CD glottal
variable place
| O
1
Vl. £
breal
vl. z

Vl. c
Vl. £

voice
vd. s

vd. j

vd. n

j ET a
rt 2. cr c 13
m
I
ft> g CO ft)
S
cei
ilar
ilai
ilai
plos

» ft)
cr
0) 1 <* 8.
1 >o X (D
! "^ X)
ib.

M.
affi

affi

2" ft)
ft h O I 8
3 I 1 * ^* ft
£ ffi
^ i' (D r (0 I
ate

ox ima
ricat

-o bilabial
<£ r+ ,+ dental/alveolar
• CL Q. r+ ^+
palato-alveolar
J.Q. .Q. .^t-
retroflex
palatal
velar
variable place
labial-velar
 <r a
g n £
I 6
S ? £ 5 £
5 s S BS
dental/alveolar
alveolar
palato-alv. palatalized
-+ •<* retroflex
palatal
+J g
jr_ ^- velar
variable place
labial-velar
M < < < M < M
O O Q
O M O. M O O. Q
£ ' * ' <3 * I < 8. < < £
< H § § < & < o n.
H> H
o "O M H- H- CO
» tt (6
2 f ? ° a
p. sr s. •§
ft
P
H
§ s
0) H O* ft
p, . p> (1)
r i i 3 « "S R, R, 5 8 *
f! 2 & J? iS
s g s* -
.. H. O O O
H. <J B) » »
III
T^ -o_ -o TJ bilabial
labio-dental
13 13 IN l(/> dental
^ ^ ^ ~t dental/alveolar
palato-alveolar
.^_ . ^ .-»• .-»• retroflex
palatal
7r 7: 7T 7T velar
variable place
labial-velar
 u
o
41
SI >
u

"3 "3 -3

lato -al

veol
Sinhalese (020)
A iJ A A t Albanian (021) 3
(0

bil
den

den
2. £ §. > > 8.
voiceless plosive voiceless plosive p
p t
voiced plosive b voiced plosive b
d
vl. nonsibilant affricate vl. sibilant affricate

vd. nonsibilant affricate vd. sibilant affricate

vl. nonsibilant fricative #2 vl. nonsibilant affricate

vd. nonsibilant fricative vd. nonsibilant affricate

vl. sibilant fricative vl. nonsibilant fricative

s
voiced nasal m "n vd. nonsibilant fricative

voiced flap vl. sibilant fricative

vd. lateral approximant vd. sibilant fricative

vd. central approximant voiced nasal m

&
voiced trill

vd. lateral approximant

vd. central approximant

long
high

higher mid

mid "e" "a H o "a:" 2

high i, y
low
mid "e'

low
 t £ I f.
S 8. I S. f.
i IS ? a s ?•S, S S
8 S
& Jl n o
< < B)
t r •
labio-dental
• * • * - • dental/alveolar
palato-alveolar
retroflex
palatal
uvular
variable place
I I
2. 8. S. § &
2 I I es
s ? 5
-o bilabial
-£ dental/alveolar
palato-alveolar
retroflex
o palatal
7T velar
labial-velar
 < HZ
p a n
Vowels
vd. cen
vl. sib
vd. sib

rr
voiced i
voiced i

rt
i E 2
2 P>
vd. nonsibd

sr
H rt H
mid
higl
higl

1 3 Pi S
s*
Lve
Efricate

-icative
-. fricat

approximan
-icative
: fricat

i I 1
-o bilabial
x, bilabial palatalized
labio-dental
dental
^ ,£ dental/alveolar
J dental/alv. palatalize
^. palato-alveolar
palatal
je velar
i t i. i. % O O
(»rt
\\w\\\\\ ° S1
H, M, 3 3 S & »
p. p. MI H» o n
bilabial
labio-dental
dental/alveolar
alveolar
palatal
Language Language

Finnish (053) Hungarian (054)

voiceless plosive voiceless plosive

long voiceless plos. voiced plosive

voiced plosive vl. sibilant affric.

vl. nonsibilant fric. vd. sibilant fric.

vd. nonsibilant fric. vl. nonsibilant affric. c?

vl. sibilant fric. s J2 vd. nonsibilant affric.
jj
long vl. sib. fric. vl. nonsibilant fric.

voiced nasal vd. nonsibilant fric.

long voiced nasal vl. sibilant fric.

voiced trill vd. sibilant fric.

long voiced trill voiced nasal

vd. lateral approx. voiced trill

I
long vd. lat. approx. I: vd. lateral approx.

vd. central approx. vd. central approx.

long
long
high I, y u I:, y: hi
«h i:, y:
higher mid <f,
higher mid e .^

lower mid
lower mid
low low
 9LZ
S* S. 8. § * 8. H
I? rr p g S , 5
2 e z <?
a
If f
labio-dental
-£ dental/alveolar
palato-alveolar
palatal
^- velar
variable place
• • K1 I? * * * * J? S"
8 6 " 8 . 8 . S . § § S . 8 . I 5 .
R S c o S 1 8 S •» S
li 1 M 6) H ' H - H - H ' l - ' e o
p > o > H - o ) i » c r c r p > o
! _ • • _ • |_ B) 9 H - H - 9 (0 -O
l H r r T
03 o> ~ ' ' ' ( B ' t o ' < " o '
8 £
rt n H- t-h <
S X 2 f? H* H-
F B " O O S»
§ § ^ R R ?
a- -o bilabial
cr_ -o_ bilabial palatalized
dental
dental palatalized
r+ dental/alveolar
-+ dental/alv. palatalized
palatal
?r velar
H* glottal
labial-velar
 IE1 LLZ
o n <K
3 rt £ g g 2. d ^
o" 3 3 (b ^
H w c n ' O O ) o
8 £ £
S 8
< p)
a- -o bilabial
dental
ex -£ dental/alveolar
M.. n palatal
CQ ?c velar
-o glottal
H- H- CL
6 § S
labio-dental
dental
dental/alveolar
alveolar
palato-alveolar
palatal
velar
glottal
variable place
to
00

Language • 3 8 . 8
4J > > >

g 3 3 3
Azerbaijani (059) cfl i-l U U (fl (0 (0
rH ja C C I-I i-t i-i
Chuvash (060) -H cfl a; ai co id a)
xi rn T3 TJ a a. >

voiceless plosive P voiceless plosive p "t" "tJ" k

i
vl. aspirated plos. ph 2 2 2
i voiced plosive b "d" 9

vl. sibilant affric. 2

tj vl. sibilant affricate "ts" tj

vd. sibilant affric. vl. nonsibilant fricative f2 g5 x

vl. nonsibilant fric. f vl. sibilant fricative J

vd. nonsibilant fric. V vd. sibilant fricative

vl. sibilant fric. voiced nasal i

vd. sibilant fric. long voiced nasal

voiced nasal m "n" voiced trill

vd. lateral approx. "I" long voiced trill

vd. central approx. vd. l a t e r a l approximant

long vd. l a t . approximant

vd. central approximant

high long vd. cent, approximant

higher mid

lower mid

low

high

higher mid

"o"2

lower mid
 O O O O . O M O - M O C L O I - ' O
H 3 H 3 3 3 H
8
1 • ? • 3. &
bilabial
labio-dental
dental
5 S 5
% Z *t dental/alveolar
dental/alv. velarized
Q. CL ^+
pala to-alveolar
palatal
palatal nasalized
velar
vl.
vl.

vd.

voic
voic
voic

3 0) pi
8. 8. p. H- C O H- 01 H"
3
jl CO P P> C
3 3 1-"• H- 3 rt
(-•
3 0)
a
H. =
s
s- ft ,h H. t? <
8
ive
ate

ive

t
cative

r+ r+ rZ_ dental /a lveo lar

-+ palato-alveolar
palatal
velar
xi xi uvular
 g. SL 2 . ? 083
tr g. g.
voiceles

nons

sibi
sibi
voiced p

8 -a
I I S !
M> HI a i (S S
i <
£ £ M, h
itive
itive

*
-icative

o- -o bilabial
labio-dental
dental
a. r-t- alveolar
palato-alveolar
palatal
to 7? velar
>o glottal
variable place
labial-velar
t i- & & t i. i i
HJi I I *• i
bilabial
labio-dental
dental
dental/alveolar
palato-alveolar
palatal
velar
variable place
 8 • '. is 0 « 8 8
O H>
? I i I? •
? »»a •
l— n 13 IN w jo- v* dental
palato-alveolar
palatal
O M H S 8 » D P » » S » I I
8 8 8 S ? j? £ £ 8
S R S J J
3 T^ ^ o- -o^ bilabial
3 W
°^ -: "C ! I £ * ^~ i dental/alveolar
«—• °" Jt, palato-alveolar
<-• palatal
x ca *- velar
 o. a. o o <x M a* a. M o
0 * 0 " o"
(B » &. O- H- K O H- H- CL
s 3 s g S £ &g g .
p > p > H . t o i » P i c r i t t p i o
M M P t r T r T r r
p ) t e £ ' l <
•0 13 Kt K| ft Hi M>
O O O D Hi fi ri
°-I <-••" dental/alveolar
palat o-alveolar
vo
voi

I o5 ^. h
g 8. P. R 0 p. 8. n>
C n
n> rt 8 >o 0)
Ian
Ian
los

s s
E£ 0> 1 <
s*
ff

s "
ricate

Ericat
Ericat
native
ricate

bilabial
dental/alveolar
palato-alveolar
palatal
velar
 voiced t

vd. cent
voiced p

vd. late
voiceles

vl. nons

voiced n
vl. sibi

vd. sibi

C (to p> o
• p. e $ g p. p 3 CO "O
L

L
Lve
Losive

Lant fricat

approximan
affricate

approximan
. affricate

labio-dental
jo. y+ dental
dental/alveolar
palato-alveolar
palatal
m 7T velar
M CO 3
0> H- O S1 S IL
H 13 fl>
Z p. 8.
%I I I
' J J
5 5" 2
•o_ -o bilabial
-+ <•+ dental/alveolar
palato-alveolar
palatal
7T 7? velar
variable place
labial-velar
 I? &
8. 8. g. < g. 8.
I I nit
- !r If S
•o -o bilabial
r+ r+ dental/alveolar
palato-alveolar
palatal
7T 7T velar
velar nasalized
variable place
labial-velar
i - h h h o * O
M »-t 0 ) t 3
n
€ § 51
8
II
bilabial
labio-dental
n- dental
denta1/alveolar
r+ alveolar
palatal
 O- O O O M
P- r I ° s
ST 8. 8. 8. S. & S& -o8. 8£ IS
2
M n M
§ £ CO (B (C
H n
labio-dental
*+ dental
alveolar
palat o-alveolar
"+ retroflex
palatal
to ?r velar
labial-velar
g. g.
S fc1 8. 8. " 8. E
H S M S M
P». » H- CO F
< a oo"
labio-dental
]Q- k+ dental
°- ^ alveolar
n palatal
ID x- velar
labial-velar
 982
a
3 rt C
2 2 ST 61
5
M M "O 5
. imp Losive
iced 3losive

( U P
"o "d ••
icele 3S plOS Lve

B 8 I
g- g- c
. non sibilan t fricat
sibilan t fricat

3 3 n
cr> o- -o bilabial
labio-dental
a- r+ alveolar
palatal
to 7T velar
IQc 7c
c velar labialized
CQ-V 7T-. labial-velar
a- a. o o a. M d. M o o
r 5 fj c- a. .S DS - i Ho - wp 3- Q
s to Da ir -o '
3( o n cr D" 3 3 m
H i - t f i l U M M H - H . I - ' t O
( U B J H - C O B i t U O ' O ' O
M M M 0> 3 3 H - j J - C O ' D
M M rt rt M M H- M
(U ft) CD 03 < O
rt
^ 3 ?. n. " $•
a s
g B
s s g; ?
H» H> O fi
ft) ft) < < ft) ft)
3 3 n n rt rt
<' <
bilabial
labio-dental
alveolar
palatal
labial-velar
 Language

Bambara (105) Dan (106)

voiceless plosive
voiceless plosive
voiced plosive
voiced plosive
vd. implosive
vl. sibilant affricate

vd. sibilant affricate vl. nonsibilant fricative

vl. nonsibilant fricative vd. nonsibilant fricative

vl. sibilant fricative vl. sibilant fricative

vd. sibilant fricative vd. sibilant fricative

voiced nasal voiced nasal

voiced tap vd. lateral approximant

vd. lateral approximant vd. central approximant

vd. central approximant

high

higher mid
high
mid "e" "§" "5"
higher mid
lower mid
lower mid e 5
low
low a
0 H-
g s 883
» "
< 8.
S. S
o- -o bilabial
labio-dental
ex -+ dental/alveolar
t+- o palatal
7T velar
uvular
o glottal
labial-velar
f
s s.
labio-dental
^+ alveolar
c+. o palatal
variable place
labial-velar
to
00
VO

Language

•3 7 S
Temne (109) Dagbani (110)
.3

labic

bilat
1 2. 2.

>
voiceless plosive "t"
V

voiced plosive b voiced plosive "d"

vl. nonsibilant fricative f vl. sibilant affricate

vl. sibilant fricative vd. sibilant affricate

voiced nasal m vl. nonsibilant fricative

voiced trill vd. nonsibilant fricative

vd. lateral approximant vl. sibilant fricative

vd. central approximant vd. sibilant fricative

voiced nasal

vd. lateral approximant

vd. central approximant

high

higher mid
long
mid
high
lower mid
higher mid
low
low
 o D H a M O o 06Z
£ £
B" H Jl £ fi"
g < < o) a>
labio-dental
I j dental/alveolar
palato-alveolar
palatal
velar
labial-velar
8. 8.
MIS
-o bilabial
labio-dental
-+ alveolar
n palatal
* velar
variable place
*v labial-velar
 Language cd
g
Ewe (114)
Bariba (113)

labio-di
dental

bilabiai
voiceless plosive voiceless plosive

voiced plosive vl. aspirated plos. ph th

g*
vl. nonsibilant fricative voiced plosive b d
vl. sibilant fricative vl. sibilant affric. is
vd. sibilant fricative vd. sibilant affric. dz
voiced nasal vl. nonsibilant fric. $ f
voiced flap vd. nonsibilant fric. P V

vd. lateral approximant vl. sibilant fric.

vd. central approximant vd. sibilant fric. "z"

M
voiced nasal ID n"

voiced flap

vd. lateral approx.

high vd. central approx.

higher mid

lower mid

low high i

higher mid

lower mid

low
 |_i |_4 CL !-• O 363
i i r I
8. 8.
i e e
f &
o- -o bilabial
labio-dental
oT ^ dental/alveolar
M^ £ palatal
Q.. o p a l a t a l labialized
c c
ca x- velar
variable place
lal>ial-velar
3 f-
co 9 ce H-
I s. S §
HI a"
& 5 i o
I I o <
S R1 H Hi
O* Hi R1 »
3. P. ?
I ? !
o- -c^ -a bilabial
cr -o TI bilabial palatalized
labio-dental
ex r+_ -+ dental/alveolar
dental/alv. palatalized
alveolar
palato-alveolar
palatal
co 7^ 7r velar
<ot ^t ^t velar labialized
variable place
variable place labialized
J) labial-velar
 e- ° | * r £ r ?• ?• r ° r ° €63
3 I §
r ,° :? :? n ? a
bilabial
labio-dental
dental
dental/alveolar
alveolar
palato-alveolar
palato-alv. labialized
€ £
palatal
velar
variable place
labial-palatal
-g") labial-velar
< < < < < < < < < f
f L C L O H - ' C L C l - M O O fD
h - > n c o ? c o c o ( D ( t > B
( a a . H . O H - H - C L I - ' p.
bilabial
labio-dental
alveolar
retroflex
palatal
8 8. S.
3 s S
labio-dental
-+ alveolar
palatal
labial-velar
n co n>
r cr
g 8. to
g •o OJ
CO
lant
los:

H*
lanl

o
l-h l-h 9 § (t>
i-l i-l
J. p.
M.
2 2
t rt H-
ft) ID
S"
ft rt
bilabial
labio-dental
alveolar
pala to-alveolar
palatal
velar
variable place
labial-velar
Language

3 lar
Tarok (121) Amo (122)

labio-d
bilal3 ial
labi(
palaito-al
palai:al
velai

bilabia
dental/
palato-
palatal
cd

voiceless plosive P "t" k voiceless plosive P t k

voiced plosive b "d" g voiced plosive b d g gb
vd. implosive b "S" vl. sibilant affricate ts tj
vl. sibilant affricate tj vd. sibilant affricate d5
vd. sibilant affricate d5 vl. nonsibilant fricative f

vl. nonsibilant fricative f vd. nonsibilant fricative v

vd. nonsibilant fricative V vl. sibilant fricative s J
vl. sibilant fricative "s" J vd. sibilant fricative z
vd. sibilant fricative "z" voiced nasal m n P 0
voiced nasal m "n" n voiced trill r

voiced r-sound "rr" 1

vd. lateral approximant "I" vd. central approximant j

vd. central approximant j

high

high higher mid

mid lower mid

low low
 963
S 3 H -
•o. -o bilabial
Q| P) labio-dental
^ <-£ dental/alveolar
alveolar
palatal
JT velar
variable place
labial-velar
1 H1 i i i ' i!
3. rj. rj.
o o o
bilabial
labio-dental
dental/alveolar
alveolar
palato-alveolar
palatal
velar
labial-velar
 : a >
2 i r r § j ! t o ' £ S p ' ( £ ' o ' E
ss
K
H- If E rt ^ ^" ^
S to to rt to to
<
m
3 -o bilabial
^3 bilabial prenasalized
X •* labio-dental
£ =1 w "+ dental/alveolar
^ dental/alv. prenasalized
OJ •—• ^ palat o-alveolar
^ palato-alv. prenasalized
<-• "a palatal
velar
° <nJ *
velar prenasalized
^ variable place
* labial-velar
< < < < r^
2. 1 § O D. H a M a
O
§ r 5 H!
eject

jibil
Jibil

Jibil

o
id tr ill

s r P.
asp. lat.
ated

Lat. affrj

centr al ai
Later al aj
a
click

•a o> o *o Hi •
o o H
bilant J
ater al eject
lonsi bilant i
aspirated cli

3 R co
rt rt M
p r1'
i
•icate

click
jximant
s«
affric

Lcative
affricat.ed click

Lcative

TJ bilabial
labio-dental
^ alveolar
palato-alveolar
palatal
?r velar
variable place
labial-velar
P* O M M 0- M O. M 862
t - ' ( D O ) P 3 P ( n c n
( t t O - H - O O O H - H -
r* a* 3 3 3 o" T
^ s
-o bilabial
labio-dental
r+ alveolar
palato-alveolar
palatal
7T velar
variable place
labial-velar
M O - O
S. 8. & P. g g- 8.
xi bilabial
labio-dental
r+ dental/alveolar
palatal
variable place
labial-velar
Language

3 3
Gbeya (129) 3 3 Zande (130)

labio-dental

bilabial
bilabial pre
voiceless plosive
I!
P voiceless plosive t
m
voiced plosive b b voiced plosive d 9b
vd. implosive b vl. nonsibilant fricative

vl. nonsibilant fricative f vd. nonsibilant fricative

vd. nonsibilant fricative vl. sibilant fricative

vd. sibilant fricative

vd. sibilant fricative voiced nasal

voiced nasal m ffi, voiced lateral flap

laryngd. vd. nasal vd. central approximant

voiced flap

vd. lateral approximant

vd. central approximant

high

higher mid e 6

mid "e" "o"

high lower mid 5 e

higher mid low a

lower mid

low
 & i t 2 && ooe
' ' K* B* * *
S ST 8. 8. ?. p. K* K*
a ?? g 2 5
M 9 8 S §§
M
i : :
labio-dental
dental/alveolar
dental/alv. palatalized
palatal
velar
labial-velar
O M S S O J C D S M C O S S H fc
1
O P J O . D . H - H - O H - H - D - H H- |P>
I I I I if it I f r t I I
o o o o H » » i > i n
3
%, or -o -bilabial
"•{_, labio-dental
-= "L =C NI °l % -f dental/alveolar
*~* ^ "X palato-alveolar
retroflex
palatal
glottal
variable place
labial-velar
 O* O. O o P- M M o O toe
S f 8. S. {5. {2. 2 8. £
i - l i - I H f t l l - ' l - ' H - t - ' C n
0 ) ( U H - C O ( U P O * 0
M (-• M Pi 3 3 H - W T 3
•an HI M, 3 n> S
II I? S
labio-dental
'f dental/alveolar
alveolar
palato-alveolar
"+ retroflex
palatal
^" velar
-» glottal
labial-velar
P- ?• 3 ?• P- ?• ?• I I ? Is
5* 5*
bilabial
labio-dental
alveolar
palato-alveolar
palatal
velar
variable place
labial-velar
Language Language

Maasai (204) Luo (205) o 3 .3 i

J Z ^ 2 .£>
2« «
iJC r H
>
O )
i-H
< 0
tH
t )
f-l
r H
a
O
(
-a
1 - 1 , 0
0 c S
-a
a « a) co
voiceless plosive P "t" voiceless plosive

vd. implosive 6 "cf" vl. aspirated plos.

vl. sibilant affric. voiced plosive d g

vl. sibilant fric. vl. sibilant affric.

voiced nasal vd. sibilant affric.

voiced trill vl. nonsibilant affric.

voiced flap vd. nonsibilant affric.

vd. lateral approx. vl. nonsibilant fric.

vd. central approx. vl. sibilant fric.

voiced nasal

voiced flap

high vd. lateral approx.

higher mid vd. central approx.

lower mid

low

high

mid "e" "o"

low ae a v
Language
Language

Nubian (206)
Nyangi (207)

voiceless plosive "t"

voiceless plosive
voiced plosive "d"
vd. implosive
vl. sibilant affric.
vl. sibilant fric.
vd. sibilant affric.
vl. lateral fric.
vl. nonsibilant fric.
voiced nasal
vl. sibilant fric.
voiced trill
voiced nasal "n"

ii r i t
vd. lateral approx.
voiced trill
ii | II
vd. central approx.
vd. lateral approx.

high
high
highe r mid
mid
lower mid
low
low
 i i ?•
C O W 00 M. ft ft ft
8. 8. *rnM cr cr p- B t* (L M
ft rt 3 ft ft p. H>
p> p. cp pi (u B> (U 5 I 5* £
ft> CO H- O
O* M> p> PI
ss
& 8
ft ft
s s
labio-dental
H- dental
dental/alveolar
alveolar
palato-alveolar
palatal
7T velar
uvular
variable place
labial-velar
o. o o o
8. 8. 8.
sr P. I I? I
I I
n palatal
x- velar
labial-velar
 goc
3* 0*
£ 5
5 3
bilabial
labio-dental
dental
alveolar
palatal
velar
pharyngeal
variable place
labial-velar
It £
g sr s. s.
5 Sa g
8
!I
or -o bilabial
Q. ^ alveolar
<-!-. palatal
to 7r velar
 ?- ?- t I I 9oe
M 8
" § H. 2. P
P M E S 6
•8 £
r i:
s s
bilabial
labio-dental
dental/alveolar
palato-alveolar
palatal
velar
variable place
labial-velar
!I! I S 1
i.
labio-dental
^ dental
alveolar
palato-alveolar
° palatal
*" velar
•° glottal
variable place
labial-velar
 i ti I Is
f S. S. 8 8
I 1
to
H-
i->
O
p. p. £
3 3
dental
't dental/alveolar
p ala to-alveo1ar
o palatal
^" velar
variable place
labial-velar
0.1-1 hh Hi Q.
£ S2
S:
bilabial
labio-dental
dental
dental/alveolar
palato-alveolar
palatal
velar
glottal
variable place
labial-velar
 80C
(-• t-> O I-
» p. £ I ~I IB P
5.S g
I
cr» cr XJ bilabial
3 bilabial prenasalized
jo. ir+ dental
<4 dental/alveolar
2
s
=> dental/alv. prenasalized
<+, n palatal
^•D palatal prenasalized
co x velar
^o velar prenasalized
*o^ -o*' labial-velar
O l - ' S S ( B P » H . S S
I II
sf ! * I II I
•S -S
I •s i
8
o, cr ^ bilabial
cr3 bilabial prenasalized
o, a. ^+ alveolar
of alveolar prenasalized
palato-alveolar
palatal
t^3 palatal prenasalized
co 77 velar
(0° velar prenasalized
variable place
labial-velar
 6oe
O P - H -
i-h i-( 0)
00 f i H - M i
3 3
3 bilabial prenasalized
labio-dental
dental
<=£ dental/alveolar
-= ••=
^ dental/alv. prenasalized
palato-alveolar
palatal
7T (o velar
yf velar prenasalized
•>> glottal
variable place
labial-velar
p. H-
M S S
iu a. a.
£ PL 5
« p. « 00
S3
labio-dental
alveolar
palato-alveolar
palatal
velar
variable place
labial-velar
 M M Ou M O O oie
to s p. n n n
2. 8.
rt 3
H- O
CT4 CD
I E &
i II I
t^ o- x> bilabial
r* a. ^+ alveolar
palato-alveolar
palatal
7T (Q x- velar
-° glottal
variable place
labial-velar
O O I P O
3*
no TO
?I I
Jf o- bilabial
labio-dental
»• i°- i^- i^ dental
1
% 1 % 1 't, 1 ^o dental pharyngealized
alveolar
palato-alveolar
palatal
<? «° *" ^ velar
_Q ,• uvular
pharyngeal
7f "° glottal
variable place
labial-velar
 < < < ue
bilabial
labio-dental
dental
alveolar
palato-alveolar
palatal
velar
pharyngeal
glottal
variable place
labial-velar
03 (D co
I|
labio-dental
<-+ ex r+ dental/alveolar
alveolar
palato-alveolar
palatal
7r^ m 7T velar
TT^ ID £ TT^ velar labialized
variable place
variable place labialized
labial-velar
OJ

Language
Language

o 73
•3 Socotri (254)
Hebrew (253)
•8 £ 1 3
t-l T) CX Pu voiceless plosive "t"
voiceless plosive
voiced plosive "d"
voiced plosive "t"
vl. elective stop
vl. sibilant affric. "ts"
vl. nonsibilant fric.
vd. sibilant affric.

vl. nonsibilant fric.

*? vd. nonsibilant fric.

vl. sibilant fric. "s"

vd. nonsibilant fric.
vd. sibilant fric. "z"
vl. sibilant fric.
vl. sib. eject, fric.
vd. sibilant fric.
vl. lateral fric.
voiced nasal
vd. lateral fric.
vd. lateral approx.
voiced nasal
vd. central approx.
voiced r-sound

vd. lateral approx.

vd. central approx.

high

mid

lower mid
high
low
 O O CL a. M a. (-• o o (t> eie
. . . . p. p. o
8. 8. P. g § £ £ 8. * *"
asa

3 3 p. p. o 3 M
: :£s:: ?
1 l-» H- H-
labio-dental
dental
alveolar
palato-alveolar
TO 3 0),
palatal
velar
uvu lar
glottal
variable place
4 § I s I wS 2. § §£
8 ww S
M B> M
M> H cr* rt O*
S g* g
o- bilabial
labio-dental
% ? "\ dental/alveolar
"^ -^ dental/alv. pharyngealized
palato-alveolar
palatal
^ ?T ^ velar
^ velar pharyngealized
pharyngeal
variable place
labial-velar
Language
C
Tuareg (257) 11
a f- p. >
Somali (258)
.o o cu i i
voiceless plosive k voiceless plosive
s 3 I
voiced plosive vl. aspirated plosive

vl. nonsibilant fric. voiced plosive 9 a

vd. nonsibilant frlc. long voiced plosive g: a:
vl. sibilant fric. laryng. vd. plosive

vd. sibilant fric. long laryng. vd. plosive

voiced nasal vl. sibilant affricate

voiced trill vl. nonsibilant fricative

vd. lateral approx. vd. nonsibilant fricative

vd. central approx. vl. sibilant fricative

voiced nasal

long long voiced nasal

high voiced trill

mid voiced flap

low vd. lateral approximant

long vd. lat. approximant

vd. central approximant

high

mid

lower mid

low
 I I
8 8
I E S E S 8.
P-h -h 3 pi 09 (B
labio-dental
-+ dental/alveolar
palato-alveolar
retroflex
palatal
^ velar
^ velar labialized
-° u\ular
-° c uvular labialized
labial-velar
O O o ?. 5.
lat

a. PL < P. f Q p- a" D- M
Tar

I 3 M CO
CO » 3 £ (D a n
C H V rt
Ki f
ect.
sive

3 g. s
3
)losiv

P- £ n H P
ric
ric

tiv

t < a »
bilabial
labio-dental
r-t- CL, Q. -+ alveolar
palato-alveolar
palatal
velar
velar labialized
uvular labialized
pharyngeal
glottal
variable place
labial-velar
 JBXO9AXB/XB5U9P +- J> ^
+- ^
^J ^
+• ^

XBxqBXjq -°

0) p,
•> O

XB33OX8

JBXO9AXB-O3BX Bd

I i I II
5! S ^ -3 2 2
•H M-i JJ M O. a
- H S U ^ ^ "S S* S*
a . W C - H c ' f f l O r H r H
O c o ^ a m c o c a c d n )

S a 2 g 2 c h 2 3
316 •g | -s § -a n -s 5 s
"o "o T3 -H rl *O "^
 bilabial
labio-dental
dental/alveolar
palato-alveolar
palatal
velar
variable place
labial-velar
labio-dental
9J "t dental/alveolar
palato-alveolar
palatal
<° * velar
•° glottal
variable place
l&ial-velar
 8. |L
o- a4 a-
H g
H- |H-
If
alveolar
palato-alveolar
palatal
velar
uvular
variable place
labial-velar
to 3 r v cr -a (D n>
H 2 2 f P w H P oT cr » oT S ? o "
Lve
sive

3 sr
:op
Lve

it a cfricate
it a Efricate

ricative
ricative

. approximant
/I. sib. fr

zed vd. central

cr bilabial
bilabial palatalized
o. * dental/alveolar
alveolar
palato-alveolar
palatal
IQ pr velar
en 7>r velar labialized
c c
IQ 7T velar palatalized
•o glottal
variable place
labial-velar
 n i-h
°" ~° bilabial
labio-dental
Q- -+ alveolar
palato-alveolar
u n
" palatal
10
^ velar
variable place
labial-velar
S - & O O O O . M D . I - ' O . H ' a j - ' O . O
O M r e S S M M c n c o s a o i w H - r e
g ^ O . a ^ p ^ ^ ^ o O H ; ^ ! ^
£U Q3 03 M» CO 05 Q> P 05 CT1 O* p 03 C O O
l
M M r t r t t - ' l - r t r t < H '
BJ CU i - h i - h » S <6 <
•3 T3 h i i i t - h ^ O S O J P ) (»
bilabial
liio-dental
alveolar
pdato-alveolar
palatal
velar
glottal
labial-velar
 Language

Kanakuru (270)
Ngizim (269)

voiceless plosive p t
voiced plosive b d
voiceless plos. p vd. implosive 6 cf
voiced plosive b vd. nonsibilant fric.
laryng. vd. plos. b vl. sibilant fric.
vl. nonsib. fric. vd. sibilant fric.
vd. nonsib. fric. vd. lateral fric. I:
vl. sib. fric. voiced nasal m n
vd. sib. fric. vd. lateral approx. 1
vl. lat. fric. vd. central approx. J
vd. lat. fric.

voiced nasal m

voiced trill
high
voiced flap mid
vd. lat. approx. low

vd. cent, approx.

long
high

mid

low
 0 * 0 * 0 O M M I O . M M i - 1 O
O M
. M- H- • • rt> • • • H) H-
S J S. S. H; g l H; S £ |
ft <B rt 3
r
Br
OO
MM
]— 1=J
palato-alveolar
'..a. .o. .-+ retroflex
palatal
vela:
glottal
variable place
labial-velar
£ • '
Khar

• s s-
voiced
voiced

(I)
voicel

(D H- O 3" H- CO H- 3* O.
rt (0 M, H- CO
P CO
Di (B C
(301

Jirat

* M 3 H- °* 3 "
plos ive

s.
lant f
b. aff
losive

t affr
>ilant affr

o o t % % 0
sib. c
? voiced pic

t fricc

approj

appro:
H- n* 2 P.
;ive

£ f
-o_ -o bilabial
jr+ i-t- dental
dental/alveolar
palato-alveolar
.^ .'-H retroflex
palatal
^ ^- velar
variable place
labial-velar
 JBXO3AXB 4- +- S3

•H T3

•a -a I
" M S ! Q « c d o ) n )

3J C

HJI >

> ,3

322 > >

 M <
£ 3 ••
fD 3 O M 0? 2. H- TO 2. H. g
S £ S m M £ CO H S H H-
8. » 3. PB) S S T3" <'
H- CL
n HO"
s s *
-o bilabial
bilabial prenasalized
dental/alveolar
dental/alv. prenasalizt
palato-alveolar
palatal
velar
velar prenasalized
glottal
variable place
labial-velar
(u cr en
labio-dental
-+ dental/alveolar
dental/alv. velarized
palato-alveolar
retroflex
palatal
0) •+• 7? velar
-o glottal
variable place
 BXO3AXB +-

& 6
t 5

I "S ^5

c
xaxjoaaaj +" *

1 ti I
II III
-S I 1
324 2 ? ?
 3 8

o .

<H — "I

S I I
2 :
325
Language

•3 8
Alawa (354)
•a
Maranungku (355)
I
voiced plosive voiceless plosive

voiced nasal long vl. plosive "t:"

voiced trill vl. sibilant affric.
vd. lateral approx. voiced nasal
vd. central approx. long voiced nasal
voiced trill
vd. lateral approx.
high vd. central approx.
mid

low

high
mid
low
 f I I
S TJ

"o "o *o •« -d
> > > > >

327
 Language

Wik-Munkan (358) Kunjen (359)

voiceless plosive voiceless plosive

voiced nasal vl. aspirated plos.

voiced trill vl. nonsibilant fric.

vd. lateral approx. vd. nonsibilant fric.

vd. central approx. voiced nasal

voiced trill

vd. lateral approx.

high vd. central approx.

lower mid

low

high

mid

lower mid

low
 Language

o u u
Western Desert (360) U cfl cfl
Nyangumata (361)

voiceless plosive p t tJ t
voiceless plosive

voiced nasal
voiced nasal
voiced flap
voiced trill
vd. lateral approx.
vd. lateral approximant I IJ I
vd. central approx.

J j

high
low
high
low
I I oee
1
P.
" P.
2 «? 8S S
8
bilabial nasally-released
]— ]^ dental
dental nasally-released
alveolar
alveolar nasally-released
palato-alveolar
palato-alv. nasally-released
retroflex
retroflex nasally-released
palatal
velar
velar nasally-released
labial-velar
& I i t2
H« H« f*
O M 8 fl> S
(D (U CX CL M
3 rt n
rt (6 rt 3 tc
H H ^ BJ (0
dental/alveolar
retroflex
palatal
velar
labial-velar
 JBX0BAXB/XB3U9P

1 I 5 3 s

? 1
o o i a )
a) I-H

1 i i

JBXO9AXB/XBV9P +-

U
T3 Cd
H
g
331
 IBXO3AXB +-

§ 8

•3 "8
1 1

2 S

-a -g
332
 I I o) c >w
0) M

t t I I
5 3 -O
C w <u

2 2 £

•g 6

333 1 I
 < T 3 H -
JCO o3 3 -o_ -o bilabial
labio-dental
dental
-^ -+ dental/alveolar
palatal
7T 7T velar
7T 7T velar labialized
TT^ velar palatalized
•o glottal
variable place
labial-velar
n n n o
|_i (_. O OJ B H - H - C D r t n d
H H H- M <
o o o HI Hi o r e
XX H H CO
liio-dental
dental
Ji dental/alveolar
n palatal
*• velar
^ glottal
variable place
 ^ o b
gee
8 8 . . %
^ 3- 3 - - - ^ 3 m (u n>
? g M O. H- a q o o c n H - a o D - c n M
P- CM
sr sr
p 3
$ s
bilabial
dental/alveolar
Q) OJ 0)
alveolar
palato-alveolar
palatal
velar
uvular
glottal
variable place
labial-velar
< < rt
cr x> -o bilabial
labio-dental
-+ ^+ alveolar
velar
glottal
variable place
 9ee
sr s. sr p. g §
rt rt cr a 3
8 § 8 P S &s S
labio-dental
•^. 't dental/alveolar
palato-alveolar
palatal
^ ?r velar
•° glottal
variable place
vl. n
vl. s
vl. a

cr
laryngea

vd. nons
voiceles

cr
vd. cpnt ral
vl. sibiIan
vd. sibiIan

N" ? •a
asal

L
Lungchow (406)

3 M, 3 » » S
Ml
<
vl. 1ateral frica tive
ricat
icate

icate

ximan
vd. 1ateral approximan
ricat

ative
plos

io- -o_ T3 bilabial

labio-dental
iQ
I ^ 't dental/alveolar
palato-alveolar
palatal
7^ 7T velar
•o glottal
variable place
OJ

Language Language

Atayal (407) Sundanese (408)

Labia
ital
.a •S g s
voiceless plosive "t" voiceless plosive P t
vl. sibilant affric. "ts" voiced plosive b
vl. nonsibilant fric. vl. sibilant affric.

vd. nonsibilant fric. vd. sibilant affric.

vl. sibilant fric. vl. nonsibilant fric.

vd. sibilant fric. vl. sibilant fric.

voiced nasal voiced nasal m

voiced r-sound voiced trill

vd. central approx. vd. lateral approx.

vd. central approx.

long

high

lower mid high

low mid

lower mid

low
 O - D - O O H> M M M M
see
S » a. a ?• o "• H- "M
3 rr v 3 o* a* o
! — ' » — ' ( — ' & > 3 H> Mi p t;
5* t "O "O Mi p . pj II
| | <" £ 2 S «
< *<; to
3 S
bilabial
dental
dental/alveolar
alveolar
palatal
velar
glottal
variable place
labial-velar
DO 3 3 C 0 S S * IC
1
( D O J i - n M i D i D ' H ' H ' O Q H ' H ' a ' h - BJ lp>
p r t M i M i o ' c r s p c r ' a ' m co K5
I-1 3 3* D*
Q H- H» H-
-o bilabial
labio-dental
Jg- j - i- dental
dental/alveolar
retroflex
palatal
7T velar
variable place
labial-velar
VO

Language

Cham (All)
Malay (412)
2 -S
voiceless plosive P "t" t c k ?
voiceless plosive P "t
vl. aspirated plos. ph ..th» th ch kh
voiced plosive b "d
voiced plosive b "d"
vl. nonsibilant fric.
vl. nonsibilant fric.
vl. sibilant fric. "s
vl. sibilant fric.
voiced nasal m "n
voiced nasal
voiced trill
voiced r-sound
vd. lateral approx. "1
vd. lateral approx.
vd. central approx.

high I
higher mid
high
mid
higher mid
low
mid
lower mid
low
Language Language

Batak (413) Tagalog (414)

voiceless plosive p "t" voiceless plosive

voiced plosive b "d" voiced plosive

vd. sibilant affric. vl. sibilant affric.

vl. nonsibilant fric. vl. nonsibilant fric.

vl. sibilant fric. vl. sibilant fric.

voiced nasal voiced nasal

voiced trill vd. lateral approx.

vd. lateral approx. vd. central approx.

vd. central approx.

high

high mid

mid

lower mid

low
 M a- M
IB
S 5
B » »
r+
alveolar
of +• o*
palato-alveolar
palatal
7T velar
•° glottal
variable place
labial-velar
O O O
(B pl O . H - O H > H - a M
(B Q] p) <! O
T3 'd M i 3 j j ( D ( B C 0
*T3 ^ fi rt hh Hi H*
g s ? s1 a £
"° bilabial
labio-dental
dental/alveolar
"*• alveolar
retroflex
palatal
^ velar
*° glottal
variable place
labial-velar
 s I a

C 1-4 -H
TO X i Xi

Xi C
r-4 T3 -H O
01 CO CO (3

342 > >

 bilabial
labio-dental
dental/alveolar
palatal
velar
glottal
variable place
S. *
§ a. w*
glottal
variatie place
 r f i i I r ?• ? ° i
( t o O j P J H - W g O ' C F ' O
"d T3 i-t rt rt H-
bilabial
dental/alveolar
alveolar
retroflex
velar
variable place
labial-velar
H . O . M O . O O M a H ' O . H ' O O B^ to"
f D f D t D P M C L H ' O O H ' H ' O * ! — • J> 1 ^
ttoiesujco c o o l <o
x> bilabial
labio-dental
dental
'-f dental/alveolar
alveolar
palato-alveolar
"-+ retroflex
palatal
7T velar
variable place
labial-velar
 Language

Maori (423)
Hawaiian (424)

voiceless plosive
voiceless plosive
vl. nonsibilant fric.
vl. nonsibilant fric.
voiced nasal
voiced nasal
voiced trill
vd. lateral approx.
vd. central approx.
vd. central approx.

high
high
mid
mid
lower mid
lower mid
low
low
 9VE
to 3 to 3
e 2 « a
i - l i - l r t o ' r t i t t H i ' d
O* O* H, H! I? o*
a 1
labio-dental
dental/alveolar
palatal
velar
uvular
labial-palatal
labial-velar
< < < < < < < < < <
n M n H c a p i B c a i D n
§. s p. s j g p. s
BE n <• & to <t a
p 3 (D rt rt (D
labio-dental
-+ dental/alveolar
palato-alveolar
palatal
7T velar
-K^ velar labialized
variable place
labial-velar
CO

Language s
Language
rH CO

Hakka (502) Changchow (503)

voiceless plosive "t" voiceless plosive

vl. aspirated plos. "t h" vl. aspirated plos.
voiced plosive
vl. plos. w. breathy rel.
vl. sibilant affric.
vl. sibilant affric.
vl. asp. sib. affric.
vl. asp. sib. affric.
vl. nonsibilant fric.
vl. sib. af./breathy rel.
vl. sibilant fric.
vl. nonsibilant fric.
vd. lateral approx.
vd. nonsibilant fric.
vd. central approx.
vl. sibilant fric.

vd. sibilant fric.

voiced nasal
high
vd. lateral approx.
lower mid
vd. central approx.
low

high

mid

low
oo

Language
"3
o

I I
Amoy (504) Fuchow (505)

palat o-alve
I

bilab ial
velar

dentd
i.t t. voiceless plosive "t"
voiceless plosive p k

vl. aspirated plos. h ,.th" k' vl. aspirated plos. »th»

P
voiced plosive b "d" g vl. sibilant affric. "ts"

vl. sibilant affric. "ts" vl. asp. sib. affric. " t s h»

vl. asp. sib. affric. " t s h» vl. nonsibilant fric.

vd. sibilant affric. "dz" d5 vl. sibilant fric.

vl. nonsibilant fric. voiced nasal

vl. sibilant fric. "s" vd. lateral approx.

voiced nasal m "n" D

high 1, y

high lower mid

higher mid low

lower mid

low
 M 8 » 3 ( D 3 0 > [ »
MI 3 H< 3
j ^ r t c r r t
p> Hi
si n
T> bilabial
dental/alveolar
palatal
variable place
§ R S. S.
& &£ S B g
s* rn s.
3 3 S ft 8. «
Hi hh M
R, £ 3. S
"° bilabial
'-1- dental
alveolar
palatal
^ velar
variable place
labial-velar
 :
n
^ bilabial
dental
't dental/alveolar
alveolar
palatal
•*" velar
variable place
& I
n n 3 3 0> »
3 S
to 3
M rt
f fJ H, 6 B
8. S1
•^ is *d t(
f^ 3 3 33 H> H- O
S X X X
-o bilabial
dental
't dental/dveolar
palato-alveolar
palatal
^ velar
•° glottal
variable place
labial-velar
 1
Q . O - O I- D. M O. H1 M O M O ISC
f D ( D O * H - O O ( - " • ( » H - O . C O t - '
a r t a - a s c r - o c r ti <»
rt (D 3 F- CO W F- • p. T> H> BJ
CO Q3 O 03
§
H- § to S -o
to cu & pi • < a. o
• d i j i-h 3 3 p p > r s ca
33 J ^ ^ ^ S ; ? s* S"
-o bilabial
labio-dental
-T dental /alveolar
palato-alveolar
palatal
7T velar
-o uvular
•>>_, glottal
variable place
labial-velar
O M n > r t > C D 3 c n a ) C n ( t > » r t > "d
( B O l C L O . H - O H - C O H . a o i l - ' D*
9 r t & p V y v T3 n o
cr -o_ -o bilabial
^ ^- 1 dental/alveolar
palato-alveolar
palatal
co ^ 7T velar
•° glottal
variable place
labial-velar
Ao (512)
Tiddin Chim (513)

voiceless plosive
voiceless plosive P t
vl. nonsibilant affric.
vl. aspirated plos. ph
vl. sibilant fric.
voiced plosive b d
vd. sibilant fric.
vl. sibilant affric. ts
vd. lateral fric.
vl. nonsibilant fric.
voiced nasal
vd. nonsibilant fric. V
vd. lateral approx.
vl. sibilant fric. s
vd. central approx.
vd. sibilant fric. z
voiced nasal m n
vd. lateral approx. I
high
laryng. vd. lat. approx. 1
higher mid
laryng. vd. cent, approx.
low

high

lower mid
 ese
I I ! S" I ?
Hi M h(
jo. 1-+ dental
alveolar
velar
glottal
variable place
labial-velar
|< < < < < < < < < < w If
M M V p) 9 9 H- CD T)
8 § ? P ff
palatal
variable place
 n H M
O O

•a -a | *
£ 3 -3 °

CL Q_ CO £

> rH 0) TH ^ ^| MM
* 1
•
8
H O )
TJ «
O r
fl)
H r H
§
W ^
""*
1 4 -
•
t l
**
- l
S'lfl' § I o

354
 •H -H O O

I "8 "8
1 I

JBXO3AXB/XB3U3P

g 8
(-1 M
O. P.
•H 6 C

CO (X -H -H C <D 4J

O rH T3 O T3 TJ
355
 I
Language
Language 1 I :
•H t-t 4)
iH 0) N j;
3 8 2 Sentani (603)
Washkuk (602)
5 S. 3 -3 5
•a ^
voiceless plosive "t"
8. 2
voiced plosive "d"
voiceless plosive k
vl. nonsibilant fric.
voiced plosive
voiced nasal
vl. sibilant affric.
vd. central approx.
vd. sibilant affric.

vl. nonsib. fric.

vd. nonsib. fric.

vl. sibilant fric.

higher mid
voiced nasal
mid
voiced flap

vd. central approx. lower mid

low

high i

higher mid

mid

lower mid

low
 8 8 L "B
a. p. p. o
1-1 P H- CD
51 B. S s s§ s
t a. TO
variable place
< < < <
O M M O
§
rt H
°" 3°* H&
-
S »
CO CO
H I Pi M H- CO
M O (JJ 3 M- TJ
•° bilabial
palatal
variable place
labial-velar
Language

Telefol (606) Selepet (607) 1I

voiceless plosive p* "t"

voiced plosive b "d"

vl. aspirated plos.
vl. nonsibilant fric. f
voiced plosive
vl. sibilant fric.
vl. nonsibilant fric.
voiced nasal
vl. sibilant fric.
vd. lateral approx.
voiced nasal
vd. central approx.
voiced flap

vd. lateral approx.

long \d. central approx.

high i

lower mid

low
high

higher mid

lower mid

low
 3 TO -o bilabial
dental/alveolar
°- -*• alveolar
palatal
glottal
< < < < <
M O. M O O
S %
• »
3 H- I*
3. 3-
o- -o bilabial
°- -+ alveolar
palatal
velar
glottal
variable place
 P. M -H
8 g 8
o 5
« C co

1 ^^

•e. E

^ s 6 rH

"S "8
360 » §
Language s Language

•3 3
Pawaian (612) Dani (613)

voiceless plosive p »t" voiceless plosive

vl. nonsibilant fric. vl. nonsibilant fric.

vl. sibilant fric. vl. sibilant fric.

voiced nasal voiced nasal

voiced flap vd. lateral approx.

vd. central approx. vd. central approx.

nasalized

high i T u high

mid "6" mid

lower mid e o e 5 low

low
I
 3
C
g **
Language 5 u •i *a

ren
zed
CO CO
P. d N
8 ]j « C
p. > >
Wantoat (615) cd
Daribi (616)
Ii 3?

labi
labi
ntal
atal
1 Of
lar lat
•H -H CU cu a
£> £i T3
voiceless plosive
voiceless plosive "t" k k
vl. aspirated plos.
m
voiced plosive K " n H'
vl. nonsibilant fric.
vl. sibilant fric.
vl. sibilant fric.
vd. sibilant fric.
voiced nasal
voiced nasal
voiced flap

vd. central approx.

high
nasalized
mid "e 9 o
high T 0
low
mid "e" "o"

low
 £ o I 2. 2. i
ft M, 3 S g S
M g 3*
O Hi »
-+• alveolar
palatal
variable place
labial-velar
<
9 a. o t~* ?* o.
ft>
°, o (B
n
n to ca
3
S. o* cr n>
ral
los

p> < o
*d IT S,
H- Hi
8 ? H-
"° bilabial
dental/alveolar
palatal
labial-velar
 A Kunimaipa (620)

bil
voiceless plosive P "t" voiceless plosive
voiced plosive b "d" voiced plosive

voiced nasal m "n" vl. sibilant fric.

vd. central approx. j vd. sibilant fric.

voiced nasal

voiced flap

high vd. lateral approx.

mid "©" "o" vd. central approx.

low

high i

higher mid

low
< < <
a- o
vl.
vd.

voi
voi

CO CO f (B
8 8. s cr cr ID
o
-> cn
cr
Llan

I 1 0) B S •a
n> o
i-h
H rt H,
palatal
labial-velar
S 8 3 s 8 S
>d (a H-
labio-dental
-^ alveolar
variable place
M 8 »
CD 0. H
H« O M
labio-dental
dental/alveolar
variable place
< < <
3. 3. a
r> o o
I S. S.
3 cr XJ
"t dental/alveolar
glottal
 8. § 8.
<; a- TO
-o bilabial
•os bilabial labialized
•o_ bilabial palatalized
bilabial prenasalized
bHabial labialized & prenasalized
labio-dental
^t dental/alveolar
-^ dental/alv; labialized
-f dental/alv. palatalized
dental/alv. prenasalized
an ox
dental/alv. labialized & prenasalized
p; o]
palatal
7T velar
-^ velar labialized
^ velar palatalized
velar prenasalized
velar labialized & prenasalized
labial-velar
 > t
\ - < > - ' < y - < < < < < < < < < < < < 35 1ST1
( t i O . O J C U f U O I - ' M h - ' & . M M I - ' H ' M H ' O p> p
«3 ' 3 ' .3 l
ft l >
o a K
3 o 3 H > : 3 ( B t - 9 l - t - * H - C f l p > C f l f l > p > ( D 01 [C
s
v
i 5 t I S E 2 & '• 2 2 '• • j? S s
• H« • l - ' N ( U I - ' H » < - ^ l - » M < - J » H ' 0 < r t d
§ ^
( D H * M ( D fD a* rt (» (t> I-"
IS 3
r t
T
i
3
-
r
l
t
"
*
» i
d
a
^
.
H
O
- r t » H > i - h »
H > i
(
H
D
)
i
i
-
-
h
h
r
O
t '
I
d
- l
H
<
-
• s ,g s ; • j s. ^ s1 gj 3 g. ^ s •
I ' ? * e ? 3! ' ' S ? *°
g 8 ?
13 3
-«3. ^ bilabial
1 l 3
~ ^ =1 I •€ X — ^ "* 'f. 't dental/alveolar
- oi s = ? "
^t, CX tlX palato-alveolar
ic
- ^ "° fj, ^ ° palatal
~s ** ** velar labialized
Q
-^ •%. uvular
Q
s: -^ °s uvular labialized
•° glottal
variable place
labial-velar
en 3
H" O
a- 3
H. 8
f I S S*
rt B (u co
bilabial
-+ -+ Q. dental/alveolar
palato-alveolar
palatal
velar
velar labialized
uvular
uvular labialized
S £ «
glottal
labial-velar
 vd.
vd.

vd.
vl.
vl.

vl.

vl.
vl.

69€
CO to
voiced
voicel

la
la
vl. as

la
la
la

si
si
vd. no
vl. as
si

vd. ce
vl. noi

rt 3 rt> p-
to pi to pi to & v n cu PI n> co pi
f rt
S ^
rt
3* a" pi PI H. g » P. 8. S1
3 it i-i
sr pr rt • to
p-
X 2 rt to to iii 'U
« n »
P- Oi rt
alveolar
palato-alveolar
palatal
velar
velar labialized
glottal
variable place
labial-velar
i
3 3 h - P i M c a P i c o B P i 3
o o w c o p i p - c o p . o c o o
it in
3 p. CO O CO
p. rt CO
crp-
>
P . < O
f
3
» n T
3 rt
.. a s
f f f
dental
r+ r+_ dental/alveolar
-*2 dental/alveolar velarized
palato-alveolar
palatal
7T 7T velar
7r 7r velar labialized
c c
>o glottal
variable place
 vl.

vl.

vl.
vl.

o. 03
voiced
la
sil
sil
sil

vd. noi
vl. noi

•o
laryngea
vl. ejec
voiceles

0 (B H- CO CO
vd. cent!

03 P H- cn tu tu cr cr n cn 03* H.
cr
2
M I
ect
3lOS

03 CD
v pas

?
doq

(B
• 0
"frica

Lcativ

II
Jlosiv

-icati
. fric

(B
Llan : fricati

sal
a f f r ica
i f f r i c ate

-+ dental/alveolar
alveolar
palato-alveolar
retroflex
palatal
yr velar
7T velar labialized
•o glottal
variable place
labial-velar
0 rt o* T)
rt rt> 0
i-( i-( pa
(B 03 35 H- CIS
H, M»
palato-alveolar
palato-alv. labialized
n
palatal
velar
velar labialized
-° uvular
^ glottal
variable place
var. place labialized
labial-velar
 M M M O
§ 3 £ s
(B 93 3 (5
a* a* H-
§ § . s
D. *"- K, 3*
§ 3 ft
' 3
en C rt ro o en
]-+ ]-H dental
S 8 2. <
U3 C fl>
^ -+ alveolar
palatal
rt 3 03 n> ^ *• velar
-^ uvular
(D 03 cu sr
•>> glottal
|->- M 3 n>
rr O rt pj"
variable place
labial-velar
O M M
8
S 3
se
3 3 3 3 rt 3
i-i rt H H ii
§ •?§. s
bilabial
alveolar
palatal
velar
uvular
glottal
variable place
labial-velar
 Language
Language

Wintu (709)

Maidu (708)

voiceless plosive

vl. aspirated plos. Pn t vl. aspirated pics,

vl. ejective stop. P' t voiced plosive b d

vd. implo8ive 6 <f vl. ejective stop P'

vl. nonsibilant fric. vl. sibilant affric.

vl. sibilant fric. s vd. sibilant aifric.

voiced nasal m n vl. sib. eject, affric.

vd. lateral approx. 1 vl. lateral affric. t4

vd. central approx. vl. lat. eject, affric. t4»

vl. nonsibilant fric.

vl. sibilant fric.

high voiced nasal

mid vd. lateral approx.

low •vd. central approx.

high

mid

low
 r> n n o
drt cr 9 5* a" <p °* ST
3 " M 2 M H- ' M rt "^ CO
? eu IN
•a TJ H rt • M> rt H-
>-< M H- l-h O <J
OO O i - h O J i i T S (V
XX • r( M, p.
H- l-h O
j^- jo. i^. dental
alveolar
palato-alveolar
palatal
velar
glottal
variable place
labial-velar
I S 5 §
r+ a. r+ alveolar
palatal
velar
glottal
variable place
labial-velar
= E
 Ii r r^ I
«; oi w
S, J
-a bilabial
]-•• dental
palatal
velar
glottal
variable place
labial-velar
I 1 H i Ji
rt rt> 3
2£ 1 i "
3 3
-o bilabial
-+ alveolar
palato-alveolar
palatal
7T velar
-• uvular
•o glottal
variable place
labial-velar
Language

K'ekchi (714) Mixe (715)

I 3
voiceless plosive
voiceless plosive
voiced plosive
laryngealized vd. plos.
vl. sibilant affricate
vl. ejective stop
vd. nonsibilant fricative
vl. sibilant affric.
vl. sibilant fricative
vl. sib. eject, affric.
vd. sibilant fricative
vl. nonsibilant fric.
voiced nasal
vl. sibilant fric.
vd. central approximant
voiced nasal

vd. lateral approx.

vd. central approx.

high

higher mid
high
mid "e" "a"
mid
lover mid
low
low
5 I 6 II I | I 1 ' 1 I & I I t r J | \ "* j |
? » .* * 2 • • ti 3 £ -a ?
I I § s ? n • s*
g • r
X
°3 13 3 >e< XJICTCTXIXI bilabial
— •-> <o i a 3 N w ^ H- i a a ^ »+ alveolar
2
«—, ^ -+ palato-alveolar
ou IU c_ palatal
x^ 7c CQ 7T x- velar
c
7T CQ 7T velar labialized
•H **
•o glottal
9
p, 3- variable place
p.
* i« s labial-velar
o o
vl.
vl.
vl.

lar
voi<

H- I-1 ?
7 Ia
aspira

vl. centra

fl> n> co
vd. centra!
vd. sit>ilar

» sr
:ed plos

isib:
:eless {

IB sr g
o K ?* !? 8. S1

mid

low
g. g sr

high
: | to
t Ml
o
H
5 ive

rica

rica
fric
plos

lower m
<

higher mid
SI 5" re
cat<

vd. latera]L approx imant

approx imant
imant
nasal

to (D
IT
ica :ive

icat.ive

(0
bilabial
3 ^ dental/alveolar
i =f 0=>I = °l N: *i WI UT UT ST 4 t "i ^
palato-alveolar
palatal
velar
velar labialized
glottal
variable place
labial-velar
Language
1 i
2 -3
•* S Mixtec (728)
Mazatec (727) x S g •a -S a. "to >
0) iH rH M iH
f - l « 4 H e O p . e B

fi J « ^ ^ J voiceless plosive

voiceless plosive
s s g. 1 1 -a vl. sibilant affric.
k ?
voiced plosive vl. nonsibilant fric.

vl. sibilant affric. vd. nonsibilant fric.

? a
vd. sibilant affric. vl. sibilant fric. "s" J
vl. nonsibilant fric. vd. sibilant fric.

v d . nonsibilant fric. voiced nasal

v l . sibilant fric. voiced trill "r"

,,,,,4
voiced nasal vd. lateral approx.

voiced trill

vd. lateral approx.

vd. central approx. high

mid

low

high 1

higher mid

lower mid

low
 i i r r I
8 £ « 3 8
fa. a. H» o a.
H 3 S S -O
ci B ST 5 S 1
C M rt M "
"° bilabial
"t dental/alveolar
palato-alveolar
palatal
*• velar
° glottal
variable place
labial-velar
MM
i f. Mr
? .. . bilabial
dental
dental/alveolar
alveolar
palato-alveolar
palatal
velar
velar labialized
uvular
uvular labialized
pharyngeal
glottal
glottal pharyngealized
variable place
labial-velar
 I b O l b O P t O H ' M M O . M M M C L I - ' M O M O «! 6LZ
l
9 ( D 2 ( D S < t M C O H h - > P ) 9 c n C O P > f t < t p ) < t 3
P 3 O 0 ) H ' ( D 3 ( » H ' » ffi • (B • H- • O T 3 H - 0 ) M
p * 3 H > r t F - c n p > » < B 0 > i - « C T ' » p > C 0 H . O ( t t
» II (B »i (» M r t ( D r t | - i ( D r » O ' n H > A M ^ s
: &: &:
O-
.
0)
»o
D-
.
B>
<o
D-
•

Sss&ss&s.es*.??.
O
•
H-
o
<
it
li
F«
M
MI
< 3 K ) 3 O
» H n U
H-><
O »

2
*«•
> d " d • O 0» 1 H- CO
O f i M i - 1 3 ( U ( » l - h H - 0 > O 0 > H>
O tt> O 03 M> rt l-h o M» • Mi <
&
I ' ;' i ? S- 3-
•a
!3 3 ^ o- -o_ bilabial
J— — j D ^ - o - t n ^ - C L ^ - - + Q . J ^ - ' H - Q . ^ alveolar
JC-. <-. .0 o^ C4-. n^ palatal
velar labialized
uvular
uvular labialized
glottal
variable place
labial-velar
< < < < < < < < < < < <
O . O ' M I - ' M t - ' l - ' h - ' H I - ' O O
H- h^
>
O H - M C a 9 H ' H ' C a C a ( D ( D A
f 0 B > e > H ' O 0 > 0 ) H - H - t - u a i - '
3 r t r t o ' 9 r t r t o * c r n ) (D
u >
r t ( B ( B H - « l ' r t > ' l - n O t 0
p f i p § J u H l i S < " » " ( )
1
( a p j M i ' ^ e ' o t t O ^ ^ " o *
• O - O M M i O r t M i r t p i C O I t O )
< > r t 1
M H o " H - * i i * M i O <"
>
O O » O M i ( » h J . p ) H O (D
XX 1 Mi O Mi H-
M- Ml Ml O
bilabial
alveolar
palato-alveolar
palatal
velar labialized
uvular
uvular labialized
glottal
variable place
labial-velar
 ose
r t n i S H - C O . (0 • H - O C D
d i t u w p i o ' i v o i n p i H»
M r t M I t (D r» (D M
T 3 X ) > i r t » i t i « i - h r t H -
i-l H H* "-( Hi o <
O O O t - h p j H - O J i - l ' O n )
dental
TM lw ^»
dental/alveolar
palato-alveolar
palatal
velar
velar labialized
uvular
uvular labialized
glottal
variable place
labial-velar
< < < < < < < < < $ < < *tj It- 1
O . M H ' M I - ' I - ' O . I - ' I - ' O O C BJ
H- ^ 09 [5
( S Q > H > O Q l H > H - H > ( - i . a i - > IpJ
O r t o ' S r r a ' a ' o ' i B n> w to
rt(T>H-(n««H.H.O>T30J O |»
s H, ; g 5 j ; ; ; ? ? Q
• d H - i i r r « * H i M i r t H> w
H O H - Mi Mi O < -P»
O ' O H i W f t t H H I d (D >—'
^ * H- MJ M5 o" o"
•o^ o- T3 bilabial
1—1
Jt, vS" --^i palato-alveolar
*— palatal
^ ^ ^ velar
x
€ ^ ^g. ^ velar labialized
a a
X ^ uvular
x
£ -^ ^ ^ uvular labialized
•° glottal
=*" variable place
£
labial-velar
Language
Language •a
•a .g rH O
Papago (736) •H I
iI Luiseno (737)
•a
•a ^^ 5 22 a
2 3 1 1 i ii
v oiceless plosive
voiceless plosive
voiced plosive
vl. sib. affric.
vl. sibilant affric.
vl. nonsib. fric.
vd. sibilant affric.
vd. nonsib. fric.
vl. nonsibilant fric.
vl. sib. fric.
vl. sibilant fric.
voiced nasal
voiced nasal
voiced trill
vd. lateral flap
vd. lat. approx.
vd. central approx.
vd. cent. approx.

high
lower mid high

low old
low
 38C
n n M t x H - o o M
R S 2
i - f h l M
S 2 § 8
f 0D M H - H - C D
(a (i> ca is a* a *
M M 3 PI 9 H - H - T 3
pi pi n ' " " ' ^ O T I D ' O '
*O *d » MI 3 3 OD
! 3 3 3" ^ ^ S"
bilabial
labio-dental
alveolar
retrrflex
palatal
velar
velar labialized
uvular
glottal
variable place
labial-velar
M M M
i & I
8 £ I
£ S S
I I If § R
? Jf H-
-o bilabial
labio-dental
^- alveolar
palato-alveolar
palatal
jr velar
^j glottal
variable place
labial-velar
00

o o Language
Language

2 u u 2 -S Karok (741)
Tiwa (740)
rH rH 3 S t

voiceless plosive voiceless plosive

vl. aspirated plos. vl. sibilant affric.

voiced plosive vl. nonsibilant fric.

vl. ejective stop vd. nonsibilant fric.

vl. sibilant affric. vl. sibilant fric. S s J

vl. nonsibilant fric. voiced nasal "n"

vl. sibilant fric. voiced flap

vl. lateral fric. vd. central approx.

voiced nasal
„ ,,2 short
voiced r-sound "rr long nasalised

vd. lateral approx. high i

vd. central approx. higher mid

lower mid

low
oral
high !

higher mid e

mid "9"

low a
 S g. H
5 •& S
8 5 55
bilabial
labio-dental
dental
dental/alveolar
alveolar
palato-alveolar
palatal
velar
uvular
glottal
variable place
labial-velar
t
•o bilabial
]H- dental
^* alveolar
palato-«lveolar
palatal
7T velar
3rt velar labialized
-Q^ uvular
labial-velar
-o glottal
 §g
i j &f
•o bilabial
dental/alveolar
,+ alveolar
palato-alveolar
palatal
^ velar
x, uvular
^j glottal
variable place
labial-velar
O M r t n i M S c n c o o j n i f c t t J
• d - d H. ft • Ht p> rt "d
H> Hi O M»
O H • O
dental
dental/alveolar
palato-alveolar
palatal
velar
glottal
variable place
labial-velar
 o o o
I t "l I I I : I \
§ ? J? 8, p. ^ ro
bilabial
dental/alveolar
palato-alveolar
palatal
velar
glottal
variable place
labial-velar
M M 3*
CL O
m
8
bilabial
labio-dental
dental /alveolar
alveolar
palato-alveolar
retroflex
palatal
7? velar
^ velar labialized
•° M glottal
variable place
labial-velar
 M M h-> O
8 £ 8£
s sg 1
•8 •§ i-h
»
P
g
rt
?" • CO
8. S
(0
i-( rt pj rt >O M-
o" it. pi hh •§ O1 ffi
•-( H, li CD
bilabial
dental/alveolar
palato-alveolar
palatal
velar
velar labialized
glottal
variable place
labial-velar
w c o S
1 ; $ H - H - O
c r a ' 3
r t p > 3 H C O
1 35
r ? I 33
° 5 I
I
bilabial
dental
palato-alveolar
retroflex
palatal
velar
•° glottal
variable place
0) »
c O c labial-velar
 88€
S D 2 S 2 S 2 S
pi w >a P> •© pi n
M pi 3 S H- T3
n
pi co co " pi *o
T> H- Hi H- PI rt CO
I I "S ?" H ?* K 8. 5"
3 H, S p. ?. , •
H- n rt i-h O M
a H. p. nt pi o
Pi O < i-l rt CO
rt rt n <
^ -o bilabial
dental/alveolar
£ * alveolar
pa lato-alveolar
£ 7T velar
•o glottal
O- rt O.
• s • g
sr & g s
I P
B. I"
•a -o bilabial
— — 3 3 ;f -E dental/alveolar
0) d)
palato-alveolar
palatal
x- 7T velar
variable place
labial-velar
 68€
S1 Z S S § S 8
£ H, 31 IS
dental
J ,? dental/alveolar
palatal
^ velar
TT velar labialized
c
•o glottal
variable place
labial-velar
g » S. S
P. S S. S
•8 « >i P> H>
S 3
bilabial
dental/alveolar
alveolar
palato-alveolar
retroflex
palatal
velar
velar labialized
glottal
labial-velar
 a. o o
O fl>
n a. o- 9 o-
9 H- 9
g § Bg g
s s ?
f 5 s
bilabial
't dental/alveolar
palatal
*" velar
•>» glottal
variable place
labial-velar
OJ O * O O O H ' Q M O ' l - ' t - ' O M M M O
H < h O P C 0 < |
• r r g g ? ? r s
8 • H,
H, B B.. •" : : «• §
i-ti ii rt (» to i-h
>i
it H-
ii rt Hi I-h hh
M i H - O l ^ f l * Mi M» - •S 5*
H O 1 Mi ri h)
H. H* Mi M> H-
P P H P P
dental/alveolar
velar labialized
*° glottal
variable place
labial-velar
 Language
Language

Yuchi (757)
Dakota (756)

voiceless plosive
P "t" vl. aspirated plos.
vl. aspirated plos. Ph voiced plosive "d"
voiced plosive b
vl. ejective stop "t"'
vl. ejective stop P' "t" 1
vl. sibilant affric. "ts" tj
vl. sibilant affric. tj vl. asp. sib. affric. "ts h "
vl. asp. sib. affric. tj vd. sibilant affric. "dz"
vl. sib. eject, affric. tj vl. sib. eject, affric. "ts"'
vl. nonsibilant fric.
vl. nonsibilant fric.
vl. nonsib. eject, fric.
vl. nonsib. eject, fric.
vd. nonsibilant fric.
vl. sibilant fric.
J
vl. sibilant fric. "s" J vl. sib. eject, fric.
vd. sibilant fric. "z"
vl. lateral fric.
vl. sib. eject, fric. J' vl. lat. eject, fric.
4
voiced nasal "n"
m voiced nasal
vd. lateral approx.
laryngeallzed vd. nas.
vd. central approx.
vd. lateral approx.

oral laryng. vd. lat. approx.

Vowels high i
vd. central approx.
higher mid
laryng. vd. cent, approx.
lower mid e
low a
Vowels high

higher mid

mid "e" "o"

low
 p. H« • . . H.
2 » a o. H- o m a S p.
bilabial
alveolar
palato-alveolar
paLatal
velar
.o glottal
variable place
labial-velar
O M M I - ' M O O
2 2 S 2 g1 5= f
H H, §
H- >i rt
-o bilabial
dental
dental/alveolar
alveolar
palato-alveolar
palatal
velar
variable place
labial-velar
 §
oc
x
-S . g. 3. K ff
?
bilabial
labio-dental
dental
' =C =C «C dental/alveolar
alveolar
palato-alveolar
palatal
velar
glottal
variable place
3. I. ? ?*
If * •* ?
S. &
I" 2 £
S 8 2 S t
jB p M B g a s < O
H> O
p» H -d
Hi O
cr x. bilabial
a -+ dental/alveolar
-^ dent./alv. palatalized
palato-alveolar
palatal
7T velar
•o glottal
variable place
labial-velar
f D 0 * a . H > H > Q H * Q * M H- I Co
3 CT" O* 3 O* (D TO
H r t r f r t w
( » § ' £ ' ! ^ 5 '
T 3 D . H i l - h 3 0 > ( B C 0
•d li II rt Hi P-
H y* H- n> 4
3 o- -o bilabial
q 3_ N_ w_ J
-J- CL -+ dental/alveolar
; s
•—• j ^ palato-alveolar
*— "^ palatal
x «Q x- velar
* labial-velar
r* ? 3.
oo 3 n
I - 8
dental/alveolar
glottal
variable place
 P. P. g £8
bilabial
alveolar
-t- IQ. |-+ pal*o-alveolar
palatal
velar
glottal
variable place
labial-velar
p. p.
M M a*
O O H
-o bilabial
bilabial prenasalized
bilabial palatalized
r+ alveolar
alveolar prenasalized
'Z.. alveolar palatalized
aTu prenasalized & palatalized
palato-alveolar
palato-alv. prenasalized
pabto-alv. palatalized
palatal
yr velar
velar prenasalized
•o glottal
variable place
labial-velar
 96€
8 |
s 8 ee i »»s s. .
3 3 -a> •» o- x> bilabial
K» 3 3 w a. *+ -*• alveolar
o
*, a. -h alveolar palatalized
OJ «-, r+ palato-alveolar
•^ o palatal
x co 7r velar
3 ^> glottal
&>
^* =r variable place
8.
I
H- Hi Ml <
O M, H, H M »
if a & i
o- -o bilabial
ex r+ alveolar
alveolar palatalized
palato-alveolar
t+. o palatal
co TT velar
-o glottal
 < < «s <
?• ^ & 8.
s s 3 3 S A
|-« M ET EI f S" 8
^ rt§ rt
s 3 !?
EL n
n n
-o bilabial
't dental/alveolar
alveolar
palatal
^ velar
o glottal
variable place
labial-velar
3 SL S S
O O* O. M
g c «s
M M D* 3*
? S cT •
O O H> h* H- rt CO * d
1 1
I- M> I-
S R, ^ S
^ bilabial
bilabial prenasalized
labio-dental
dmtal/alveolar
^ alveolar
alveolar prenasalized
palato-alveolar
retroflex
palatal prenasalized
PT velar
velar prenasalized
•° glottal
 •3 2
Amahuaca (810) Chacobo (811)

voiceless plosive voiceless plosive t

vl. sibilant affric. vl. sibilant affric. •ts

vl. nonsibilant fric. vl. nonsibilant fric.

vl. sibilant fric. vd. nonsibilant fric.

voiced nasal vl. sibilant fric.

voiced flap voiced nasal

vd. central approx. voiced flap

vd. central approx.

nasalized

high i QJ

mid high

low mid

low
8 H, 8 8
alveolar
palatD-alveolar
retroflex
palatal
velar
glottal
labial-velar
< < < <
& $ H. H-> O O
MM
P. S H. S CD
o. M- alveolar
palato-alveolar
retroflex
palatal
velar
variable place
labial-velar
 i-t iH O

g 1 1 X 8
HJ| < i 1 1 i i

JBX3A-JBXO3AXB/XB5U3P

MX
o o
p P

400 31 ^
 o o> o g. e j
srs g.• ss=
O H- H-
r ft H. hh
1
i !? R
it ?
•o -o -o bilabial
,+ ^. ^. alveolar
palatal
yr velar
. velar
•>> glottal
variable place
labial-velar
n S 8 <B 9 SM
» o. a p. o
s „ a s 3 :
M a s * i » ! ! ! J
« M^ : r I
-o bilabial
,+ dental/alveolar
s
palato-alveolar
palatal
*• velar
variable place
labial-velar
 < <
zov
8 8
rt 3 XI
M 3 3*
S -a
3 £ £
palatal
labial-velar
?• p- I
O O H
hh § § rt
bilabial
dental
dental/alveolar
palato-alveolar
palatal
velar
uvular
variable place
labial-velar
 < < < < < < < < < < <
sr is1 e o v
rtCLCLa:§ o- •§ cr V •$ m'
( U B j c n p j a ' n i c n B J H - B J
M "O d> D p. i_u H- 0 < rt "O
1 1
H rt H (B C T r t O ) (0 H
(u (u n a- o
-o -o bilabial
palato-alveolar
o palatal
labial-velar
g) cr m OJ H- o
M, § ft 0) en i" S*
H- Mi O <"
3 S3
dental
palato-alveolar
palatal
velar
uvular
glottal
variable place
labial-velar
S § IT t-
CD
Q. o. no H<
S" s
bilabial
alveolar
palato-alveolar
retroflex
velar
glottal
variable place
labial-velar
l t
3 3
"° bilabial
labio-dental
't dental/alveolar
palato-alveolar
palatal
^ velar
variable place
labial-velar
 M M O
o o
5 ?•
bilabial
bilabial palatalized
dental/alveolar
dental/alv. palatalized
alveolar
palato-alveolar
palato-alv. palatalized
retroflex
glottal
labial-velar
< < < < < < < < n
a O O M O - H I - ' O B>
fD D. O. H- O O H- H1
3 cr 3 3 cr n> ~
( u o t o O J O ' a ' O ) oi
( - ' • a p j S H - H - S ' a ^
•a i-h 3 5 BJ tn
rt
"H [^. *"* [Ijj >'
3 -co XJ bilabial
h n w ^ 4 - ^ - alveolar
i ^ —, ^ i^- palato-alveolar
<-• palatal
7T velar
rr v a r i a b l e place
 p* I I r r r r r i g 90V
i - ' m S c n P ' a t o ' a S H"
f D r - t S H - a i j u H . ( u t n
M, (U CO T3 O - ^
M Hi D H- CD M Cn
(u i-i n- a* Mi o H-
f ^ •O H- Mi CO <
o MI hi . n>
T3 bilabial
^+ dental/alveolar
palato-alveolar
palatal
7T velar
•o glottal
variable place
labial-velar
< < < < < < < < < <
Pi (to KB <" °
"H "S H- ^ ^ n, <"
o o n MI MI n (D
labio-dental
alveolar
palato-alveolar
palatal
velar
glottal
variable place
labial-velar
Guarani (828)

voiceless plosive
voiceless plosive
vl. sibilant affric.
voiced plosive
vl. nonsibilant fric. w2
vl. sibilant affric.
vd. nonsibilant fric.
vd. sibilant affric.
vl. sibilant fric.
vl. nonsibilant fric.
voiced nasal
vl. sibilant fric.
voiced fricative trill
voiced flap
voiced flap
vd. central approx.
vd. lateral approx.

high

nasalized higher mid

high 7 ? u
low
higher mid

low
 80 V
voice
vl. n
voice
voice

vl. s
vl. a

Cu en
3 rt
rt (B rt 3
01 pi H- CO cr 01 O
£
2
ed

01 01 < o
Ian

Ian

i-j
cT
fri

8 8 CO
labio-dental
palatal
variable place
< <
I-1 0 3*
l-h 3 P (U (D CO
O M, t? t? (S
"° bilabial
labio-dental
^ dental/alveolar
palato-alveolar
palatal
?r velar
%• velar labialized
.o glottal
labial-velar
 I-1 M- M
P> O* O
3 H- CO
^ s S"
palatal
variable place
labial-velar
n 8 3
3 c
•o •. •
H- 09 I-1
r-t CO < I - H -
H- to ^ § H^
bilabial
dental
alveolar
palato-alveolar
retroflex
palatal
velar
velar labialized
glottal
variable place
variable place labialized
labial-velar
 01*7
0> CO 0 rt> (D
a- 3 (D
^+ alveolar
palatal
^ velar
^> glottal
variable place
labial-velar
& CL H- O H- M
W
»T m pT a 1 pT
Zs
bilabial
dental/alveolar
palato-alveolar
palatal
velar
glottal
labial-velar
o n > m c n c n 3 3 c n B ) c n n > p j n >
a
3 C L C L H ; ^ ^ O £ y £ ,$ £
W t n m t o ' B j o ' o ' i a c n w ' o ' w 0 1
M O f b 3 3 H - l - ' - 3 H - 5 c n r t - > T 3
r f r t
( U S g ) { u r f - r f < ' c L o '
labio-dental
't dental/alveolar
palato-alveolar
n
^ palatal
^ velar
•^ glottal
variable place
labial-velar
a a- o M a H M
p. . . . .
n H n> en 0 3 p
3 rt a" 3 3 3
rt n> 3 H- CD co tn
^ & I B p p5
I «M ^ B g B
palato-alveolar
retroflex
palatal
velar
labial-velar
 I
3
2 S
M sr £ S R &5
bilabial
labio-dental
dental
dental/alveolar
palato-alveolar
palatal
velar
-° uvular
dental
't dental/alveolar
palato-alveolar
palatal
7? velar
variable place
labial-velar
 i

2 i X
1 0)
tfl
"V __, ^H ( tO
1 O
d M lo

Lat
.3 2 S t
Telugu (902) H 1 <U

lat
•a 2. SI

voiceless plosive h k
k vl. aspirated plos. D ^ ih t
t
vl. aspirated plosive t h2 k d d
voiced plosive b d g

voiced plosive d g vl. sibilant affric.

breathy vd. plosive vd. sibilant affric.

f 9
vl. sibilant affricate ts t?2 vd. nonsibilant fric

vd. sibilant affricate dz vl. sibilant fric.

vl. nonsibilant fricative voiced nasal
vd. nonsibilant fricative voiced trill
vl. sibilant fricative voiced flap
voiced nasal vd. lateral approx.
voiced flap vd. central approx.
vd. lateral approximant

vd. central approximant

high

mid

long low

high
higher mid
lower raid
low
 t i £ £ t js. t i VIV
p. p. . . . p. p.
o- p cr cr in
p. co pj cr pi p> o
r iE
bilabial
dental/alveolar
palato-alveolar
retroflex
palatal
variable place
labial-velar
(B pi a o. a.
pi pi pi H- as pi a4
M M T3 H1 pi P H-
3 S
labio-dental
ID. >+ dental
palato-alveolar
•Q. •-+ retroflex
palatal
variable place
P92TJBX9A *A

JBXO9AXB/XB3U9P

I 3 I 8
^ 0 rH rH

-H -H O -H

3 I
1 $

BXBd 'AXB/XB3U9P 'Z

JBX09AXB/XB3U9P ^

S -H rH

o o
fl C

415
 ?• I r r * r 91*7
II! II I
PL S S
73
bilabial
^ 't dental/alveolar
retroflex
palatal
-^ velar
-° uvular
•° glottal
labial-velar
M ( D ( 0 ( 0 ( 0 ( n 9 9 0 >
(u M a- a. F- F- o o to
r1 i I c &; ? 1 1 1
•o. -o bilabial
labio-dental
^. ^ dental/alveolar
palato-alveolar
° palatal
7y. *• velar
-Q^ -Q uvular
variable place
labial-velar
 CO M
p. p. i_i- p. 3
B 3-
r
s s * :
:
Hi 3
H> H- Hi p< O
O O H O (D
0) (B p. 0) n
ft rt O pt (0
H ^ o> n>
•o T3 bilabial
dental
dental/alveolar
dental/alv. velarized
alveolar
palato-alveolar
variable place
< < < < < < < < < < < < < < <
a t-> |_. M O. M
cr cr cr
" 2.
l-{ p.
H. o
bilabial
labio-dental
fcT IN 6T dental
dental palatalized
alveolar
palato-alveolar
palatal
velar
velar labialized
c c c
velar palatalized
uvular
uvular labialized
pharyngeal
glottal
glottal labialized
labial-velar
 Nama (913)

I S a) Non-click consonants and

vowels
voiceless plosive

long vl. plosive p: "t:" k : q: voiceless plosive i

h h Wh h vl. asp. sib. affricate
vl. asp. plosive p "t " k q

voiced plosive b "d" vl. asp. nonsib. affricate

v l . eject, stop p' "t'" q' q ' vl. nonsiblant fricative

v l . asp. sib. af. "ts h ""ts W h "1:J h tJ W f i vl. sibilant fricative

W W voiced nasal
long v l . s i b . a f f . "ts:""ts :"tJ tJ

v l . sib. ejec. af. " t s " m t s W " 1 t J ' tJW' voiced t r i l l

vl. nonsib. fric. X X

long " x: x w
vl. sibilant fric. "s" J JW
.. u r rw high
long » s:" J: J :

vd. sibilant fric. "z" 3 higher mid

voiced nasal m "n" mid

voiced trill "r" low

vd. lat. approx. "I"

vd. cent, approx.

short long short pharyngealized

high

mid

low
 ^ o
CO 4->

JJ tO
CO fO Basque (914)
Nama (913)

b) Click, consonants voiceless plosive P "t" c k

voiced plosive b "d" j g
voiceless click
vl. sibilant affric. "ts" tj ts
voiceless aspirated click
o; h *h vl. nonsibilant fric. f
glottalized vl. click
vl. sibilant fric. "s" J s
voiced click
voiced nasal m "n" p
voiceless affric. click
voiced trill "r"
vl. asp. affric. click
voiced flap "r"
glottalized vl. affric. click
vd. lateral approx. "1" A
voiced affricated click

vl. lateral affric. click

vl. asp. lateral affr. click

high
glottalized vl. lat. aff. cl.
mid
voiced lat. affric. click

Note. Those clicks which are characterized as "velarized" in UPSID

have a fricated release of the velaric closure. The term "affricated
click" is reserved for clicks in which the front closure is released
with accompanying local friction. Clicks which are described as both
voiceless aspirated and nasalized are described in the literature as
having "delayed aspiration". See Ladefoged and Traill (UCLA Working
Papers in Phonetics 49: 1-27, 1980).
 g £ J
2 g £ i
§ 8 S . o
r t i - H ' r t C r t M ' < D M <-s
» 0 ) • < a o \ o
i - h O 3 B > P> rt> co M
!* M > l - h H > | - * ^ s ^
r o
SS S I? S1 P. J? ^ S
IB r T H - H - O H » O K
99
( B r t r t ( D r t ( D
m re
o- -o x. bilabial
-^ labio-dental
W" ^ r+ CL r+ r+ dental/alveolar
W W
= ? = = ? =
•—. o. ^+ r+ palato-alveolar
•w .»_< -a. .^- .^+ .o. .r+ .^- retroflex
•N <w -w 3-
palatal
to yr 7T velar
a x r> JQ uvular
3- variable place
labial-velar
p. p. •
S 8 o>
o. a H;
H, B
alveolar
palato-alveolar
palatal
velar
variable place
labial-velar
 & I
bilabial
labio-dental
dental
alveolar
palato-alveolar
velar
glottal
variable place
O o l —
( t o C O r t ^ r t f f i P J r o
s. s. n "So1 're* <3* •§ {T
8 ?
tD £ <D' H- O 5* B>
o » >
F - 3 O < F - ( » r T O
o - r t r t m O H - r o i - '
- < o. o
8. ^
II hh n i-h ht>
H- O O >-( l-h
< CD P> H- H
(0 rt rt o p-
H. H- 6) O
< < rt (D
:3 3 3 bilabial
Qu .Q. -+ Q. r+ r+ alveolar
alveolar velarized
palato-alveolar
palato-alveolar velariz
palatal
velar
velar pharyngealized
variable place
labial-velar
 to
8 g g
Language is i
•H TH tH cd cd
Language
CO iH iH c > >
IXu (918)
S2 % si
b) Click consonants ^jj
J J ^H !Xu (918)
8
c) Vowels
voiceless click * c short o r a l long o r a l
high i u i: i
vl. aspirated click
mid
glottalized vl. click
low
voiced click g* gC nC

breathy voiced click gC nC

Rlottalized vd. click short nasalized long nasalized

high T u 0:
vl. affricated click
mid
vl. asp. affr. click
low
glot. vl. aff. click

vd. affric. click

short long
breathy vd. aff. cl. pharyngealized oral pharyngealized oral
mid 9 9
glot. vd. aff. click
low
vl. lat. aff. click

vl. asp. lat. aff. c.

short pharyngealized long pharyngealized
gl. vl. lat. aff. cl. oC<" ,CA? nasalized nasalized
mid
vd. lat. affr. click
low
gl. vd. lat. aff. cl.

breathy vd. lateral

affricated click
§T
eu
See note on the description of click consonants on the chart for
Nama (913). oT oT
6a9