THE STRUCTURE OF ENGLISH ORTHOGRAPHY

JANUA LINGUARUM
STUDIA M E M O R I A E
N I C O L A I VAN WIJK DEDICATA

edenda curat

C. H. VAN SCHOONEVELD

I N D I A N A UNIVERSITY

SERIES MINOR

NR. 82

1970
MOUTON
THE HAGUE · PARIS
THE STRUCTURE OF
ENGLISH
ORTHOGRAPHY
by

RICHARD L. VENEZKY

U N I V E R S I T Y OF W I S C O N S I N

1970

MOUTON
THE HAGUE · PARIS
 © Copyright 1970 in The Netherlands.
Mouton & Co., N.V. Publishers, The Hague.

No part of this book may be translated or reproduced in any form

by print, photoprint, microfilm, or any other means, without written
permission from the publishers.

LIBRARY OF CONGRESS CATALOG CARD NUMBER: 73-98472

Printed in The Netherlands by Mouton & Co., Printers, The Hague.

To my wife
 5

PREFACE

The work presented here is one result of a continuing analysis of the

reading process, an analysis which originated at Cornell University
as a component of what later became Project Literacy, was pursued
at Stanford University as a dissertation project, and is continuing
now at the Wisconsin Research and Development Center for
Cognitive Learning. Central to all this work has been an attempt
to understand what the reading process is and what is lacking in
those children who have trouble learning to read. The first phase
of this work, represented in part by the material presented here, was
linguistic in nature, aimed towards an exhaustive description of the
relationship between spelling and sound. The present work is
derived mainly from my doctoral dissertation (1965), with additional
material added from studies done in 1966. The second phase, in
progress now, is more psychologically oriented. It began with the
linguistic data obtained in phase one and proceeded to ask which of
the posited letter-sound patterns are employed by competent
readers, and how they are acquired. From this basis, the research
has expanded into almost all areas of the reading process. Its
original goal, however, has remained the same: to improve the
teaching of reading.
This work, like most extensive research efforts, is not solely the
product of one person, but drew upon the advice and criticism of a
number of others. Foremost among these is Ruth H. Weir, whose
tragic and untimely passing in November of 1965 deprived me of a
friend and colleague and the academic world of an esteemed scholar.
Her contribution to the study of English orthography is far in
6

PREFACE

excess of what I can express here. Among the others to whom I am

indebted are: Professors Harry Levin and E. J. Gibson of Cornell
University whose encouragements helped sustain my interest in
reading and whose own works have been models for my studies;
Professor Charles Hockett of Cornell University who started me
on the present course; and my wife whose contribution is much less
specific than those listed above, but no less important.
R. L. V.
Tel Aviv, 1969
 TABLE OF CONTENTS

Preface 5

I. Purpose, Plan, and Scope 9

1. Introduction 9
2. Purpose 11
3. Source Materials 11
4. Plan and Scope 13
II. Attitudes toward English Orthography 16
1. Introduction 16
2. Grammarians 17
3. Contemporary Linguists 25
4. Spelling Reformers 30
III. Preliminaries to Orthographic Analysis 34
1. Graphemic Features 34
2. Correspondences 39
3. Morphemic Features 41
4. Phonotactical Influences 44
5. Descriptive Model 45
IV. The Graphemic System 47
1. Selection of Units 47
2. Markers 55
3. Graphemic Alternations . 59
V. Consonant Distributions, Correspondences, and Al-
ternations 63
8 TABLE OF CONTENTS

1. Introduction 63
2. The Correspondences 64

VI. Fricative Alternations 92

1. Palatalization 92
2. Voiced-Voiceless Alternations 93

VII. Vowel Distributions, Correspondences, and Alterna-

tions 101
1. Primary Vowel Patterns 101
2. Alternations Based Upon Primary Vowel Spellings 108
3. Consonant Influences 109
4. w Influence 112
5. Miscellaneous Consonant Influences 113
6. Secondary Vowel Patterns 114

VIII. Conclusions 120

1. The Nature of the Current Orthography 120
2. Spelling Reform 122
3. The Teaching of Reading 123

Bibliography 130

Index of Terms 134

Word Index 136

Index of Names 158

 I

PURPOSE, PLAN, AND SCOPE

... everyone ... has to admit that of all languages

of culture English has the most antiquated, in-
consistent, and illogical spelling.
R. E. Zachrisson, 1930, p. 10.

1. INTRODUCTION

Educators, orthoepists, philologists and spelling reformers have

from the darkest periods of the Middle Ages joined in the assault
upon the 'antiquated', 'inconsistent', and 'illogical' spelling with
which the English speaking world is burdened.1 The Anglo-
Saxons, working with an Irish-flavored Latin alphabet, established
by the middle of the eleventh century orthographic practices con-
sistent enough not only to serve their own needs, but also to serve
as a basis for the Norwegian and Icelandic orthographies.2 Nor-
man invaders, however, carrying the orthography of the Continent,
took immediately to reforming English spelling. New letters were
introduced; old letters were discarded, and the entire system was
endowed with an Anglo-Norman flavor. In time the Anglo-Nor-
man language disappeared, but Anglo-Norman graphic practices
had already become a permanent part of English spelling.
While the Anglo-Norman attack waned, a Latin assault on
English orthography waxed. New words were imported with
1
These descriptions of English spelling can be found in, among others, Hart
(1551), Zachrisson (1930), Bloomfield (1933), and Lounsbury (1909).
a
See George T. Flom, "Studies in Scandinavian Paleography", JEGP, 14
(1915), 530-43.
10 PURPOSE, PLAN, AND SCOPE

clean, Latin spellings; old words were redressed to parade their

classical origins. But even though t was substituted for quite a
few c's and s's, and some Vs and Λ's were inserted where they
didn't belong, the old system was still visible through the pale.3
Sixteenth and seventeenth century grammarians, aroused by "the
gros and disgrac'ful barbarismes" of the existing orthography,
labored religiously to convert their countrymen to new spelling
systems, but the ravages of conservatism and the relentless move-
ment of time laid waste their proposals, leaving the banner for the
Ben Franklins and the G. B. Shaws of the coming centuries.4
With the rise of comparative philology in the early nineteenth
century, a new era of tranquility arose. Instead of scorn and con-
tempt, dispassionate analysis was in vogue for the orthography.
This romance, however, was short-lived. The Rasks and Grimms
soon learned that the letters weren't really the sounds, and the
orthography was relegated to the backporch of the new linguistic
science. With structural linguistics came even harsher treatment.
"Writing is not language" claimed one of the dicta of the new
science, "but merely a way of recording language by means of
visible marks", and the dust grew thicker, stirred into motion only
by an occasional spelling reformer who sought rapid condem-
nation and execution for the prevailing orthography.5 In recent
years a new scientific interest in English spelling has arisen. Lin-
guists, educators, and psychologists joined forces several years ago
to investigate the reading process, and through their efforts a
thorough study of English spelling-to-sound correspondences was
begun.6 In the resulting investigation, evidence was found to show
that English spelling is more complex and contains a higher degree
of patterning than was ever assumed before. From an interest in
the nature and origins of the spelling patterns revealed by that
study, the present work evolved.

8
On the Latin influence in English orthography, see Venezky (1965), Chap-
ter VII.
4
Albert Eichler, Charles Butler's English Grammar (Halle, 1910), p. 16.
6
Leonard Bloomfield, Language (New York, 1933), p. 21.
• See pp. 11-12.
 PURPOSE, PLAN, AND SCOPE 11

2. PURPOSE

The function of this book is to show the patterning which exists

in the present orthography — not just in terms of regular spelling-
to-sound rules, but in terms of the more general phonemic and
morphemic elements which characterize the system. For centuries
philologists have approached the study of English orthography
with the purblind attitude that writing serves only to mirror speech,
and that deviations from a perfect letter-sound relationship are
irregularities. Even so astute a scholar as Leonard Bloomfield
added his authority to this view.
Although our writing is alphabetic, it contains so many deviations
from the alphabetic principle as to present a real problem, whose solu-
tion has been indefinitely postponed by our educators' ignorance of
the relation of writing and speech.7 ... The difficulty of our spelling
greatly delays elementary education, and wastes even much time of
adults. When one sees the admirably consistent orthographies of
Spanish, Bohemian, and Finnish, one naturally wishes that a similar
system might be adopted for English.8
Whatever may have been the relationship between writing and
sound when the first Old English writings were inscribed in Latin
script, and whatever may have been the reasons for the subsequent
development of this system, be they due to random choice or to
an all-pervading National Orthographic Character, the simple fact
is that the present orthography is not merely a letter-to-sound
system riddled with imperfections, but instead, a more complex
and more regular relationship wherein phoneme and morpheme
share leading roles. The synchronic study described here shows
these different levels of patterning in the current orthography.

3. SOURCE MATERIALS

The main research upon which this book is based was begun at
Cornell University in 1961. As part of the inter-disciplinary study
7
Bloomfield, 1933, pp. 500-01.
8
Ibid.
12 PURPOSE, PLAN, AND SCOPE

mentioned earlier a study of spelling-to-sound correspondence was

undertaken by a group of linguists under the direction of C. F.
Hockett. After an initial study of monosyllables (Hockett, 1961),
a computer program was written to derive and tabulate spelling-
to-sound correspondence in the 20,000 most common English
words (Venezky, 1962).9 This program was used by the author
to obtain, for a corpus of the 20,000 most common words in Eng-
lish, the following information:
(1) A complete tabulation of the spelling-to-sound correlations
found in the corpus, based upon the position of consonant and
vowel clusters within the printed words. Thus, for any con-
tinuous string of vowels or consonants found in a printed word,
the tabulations include all of the pronunciations found for that
string, along with the totals and percentages for each pronun-
ciation in each word-position (initial, medial, and final).
(2) Complete word lists for each correspondence found in (1)
above. For example, if the cluster gh with the pronunciation
/g/ occurred in 1 above, then the word list would contain all
of the words in the corpus in which gh was pronounced /g/,
arranged into separate, alphabetized lists for the three word
positions.
(3) The same as (1) and (2) above for the 5,000 most common
words in the corpus and for the graphic monosyllables.
(4) A dictionary of the corpus in which spellings were reversed and
then alphabetized. This list was used extensively for studying
suffixes and other word endings.10

Foremost among the published works used for this study was the
Oxford English Dictionary (OED), the source for which the English
language has yet to create an adequate superlative. The extent to
which I am indebted to Messrs. Craige, Murray, et ah, for infor-

• This work was supported by U. S. Office of Education grants 639, OE-4-

10-206 and OE-4-10-213.
10
All of these data were compiled on the CDC 1604-A computer in Palo
Alto, California.
 PURPOSE, PLAN, AND SCOPE 13

mation on spellings, etymologies, and historical phonology is far

in excess of the number of references to the OED in this book.

4. PLAN AND SCOPE

Previous analyses of English orthography, beginning with the six-

teenth century spelling reform tracts, are surveyed in Chapter II. In
Chapters III and IV the basis for the present analysis is presented,
including a model for mapping from spelling into sound (Chapter
III), and an analysis of the graphemic system (Chapter IV). In
Chapters V and VI the model is applied to the consonant spellings
(Chapter V) and the vowel spellings (Chapter VII); a short sketch
of the morphophonemic system is presented in Chapter VI. In
the final chapter I have outlined, among other conclusions, the
implications of this study for the teaching of reading and for
spelling reform.
The synchronic material presented here is based upon the
spelling-to-sound correspondences which occur in the 20,000 most
common English words, although words outside of this corpus are
cited occasionally to illustrate interesting or unusual patterns.11
Proper nouns, contractions, hyphenated words, and variant pro-
nunciations of the same spellings are not included in this corpus.
All pronunciations are based upon Kenyon and Knott, A Pro-
nouncing Dictionary of American English (Springfield, Mass., 1951).
Symbols employed for the pronunciation of Modern English are
shown below.12 Except where non-phonemic contrasts are dis-
cussed, these symbols are enclosed in slant lines, e.g., /e/. The
same symbols are used between braces, e.g., {d}, to represent mor-
phophonemic forms. Graphemic units are given in italics, e.g., e.
11
In the Thomdike-Century Senior Dictionary the most common 20,000 words
according to the Thorndike frequency count are identified. Many low-fre-
quency words from that list, especially proper nouns, were omitted by the
present writer, and a number of words not included in the original Thorndike
list were included.
" See Hans Kurath, A Phonology and Prosody of Modern English (Ann Arbor,
1964), Chapter 1, for a discussion of the transcription system used here.
14 PURPOSE, PLAN, AND SCOPE

Vowels
13
checked vowels free vowels
ι b/t i beat
ε bet e bait
ae bat 0 ball
a bop ο home
ο some u boot
υ fall ai b/te
au how
01 bo.y

Consonants and Glides

Ρ /»η s sin
b bai ζ zebra
V
t rip s shin
V
d dim ζ rouge
δ chin m mat
ν *

J JaT η «et
k &ick ring
g gate 1 /amp
f /in Γ rip
ν vat w wet
θ thin j yet
ö then h hit
13
The terms LONG and SHORT as applied in educational literature and in
dictionaries to English vowel sounds are neither historically accurate nor
mnemonically useful. While some of the Modern English long-short pairs
have developed from Middle English vowels which supposedly differed only
in quantity (commonly called length), this is certainly not true of all such
pairs. Modern English long u C/ju/), for example, has developed primarily
from /ü/ in French borrowings. Middle English long u (/u/) has become MnE
/au/ as in house. Anglo-Norman scribes imported the French spelling ou for
ME /ü/ in the twelfth century, but left untouched the spelling u for short u
(Ju/), thus destroying for posterity a graphemic identity between the long and
short u sounds and their reflexes in MnE. (The MnE sounds derived from ME
long and short u are /au/ < / ü / and /u/ or /β/ </u/). In addition, MnE long
and short ο are not derived from their ME namesakes, but from ME /a/ and /o/.
Long and short vowels are defined today in terms of spelling — they are
the primary alternate pronunciations of the spellings a, e, i, ο and u. The other
 PURPOSE, PLAN, AND SCOPE 15

Schwa is used both for the stressed vowel as in come and the un-
stressed neutral vowel, as in the first syllable of away. To indicate
the correspondences between graphemic, morphophonemic, and
phonemic units, the abbreviations χ -»> {y}, {x} {y}, and {x} -»· /y/
are written. The arrow indicates that the unit on the left corre-
sponds to the unit on the right. This is always to be interpreted
as a one-way correspondence from left to right; correspondences
in the other direction, that is, sound-to-spelling correspondences,
are beyond the scope of this study. The urge to introduce more
symbolic notation in the name of brevity or economy has been
checked by the force of the following statement, made by Einar
Haugen at a time when mathematics was still relatively foreign
to linguistic disquisition.
Present day descriptions bristle like a page of symbolic logic and lack
entirely the leisurely, even charming quality of the traditional gram-
mars. I would not go back to those grammars, but only suggest that
economy may not always be a virtue. ...14

English vowels, however, seem to have no place in this scheme. How, for
example, are /0/ and /u/ to be classed?
To answer this question by appeal to phonetic length is not justified by the
classifications of the other vowel sounds. For two spellings, a and o, the long
vowels are on the average phonetically shorter than the short ones (see Gordon
E. Peterson and Ilse Lehiste, "Duration of Syllable Nuclei", JASA 32 (1960),
693-703). Furthermore, since vowel length varies according to the following
sound, almost all long vowels are pronounced in some environments with
durations which are shorter than the maximum durations of their correspond-
ing short vowels.
A more theoretically accurate and educationally practicable nomenclature
for English vowels is FREE and CHECKED. All stressed MnE vowels are either
CHECKED or FREE; the unstressed vowel /A/ forms a class by itself. "Checked
vowels", according to Kurath, "do not occur at the end of morphemes; they
are always followed by one or two consonants." (Kuruth, 1964, p. 17). Free
vowels occur both in morpheme-final position and before consonants. (This
distribution does not hold for final [a] forms like ma and pa, nor for the forms
like city and candy when pronounced with a final [1]). Furthermore, some free
vowels are commonly articulated with prominent off-glides — /i, e, o, u/ (pho-
netically [ij, ej, ow, uw]) — or are diphthongal — /ai, au, ai/. Checked vowels
are most often articulated as monophthongs; they may be in-gliding (pho-
netically [ι®, ε*]), but never off-gliding.
14
Einar Haugen, "Directions in Modern Linguistics", Lg. 27 (1951), 222.
 II

ATTITUDES TOWARD ENGLISH ORTHOGRAPHY

1. INTRODUCTION

The earliest writings OD English orthography are based upon an

alphabetic principle derived from the fourth and sixth century
Roman grammarians. Each letter of the alphabet has, besides its
name (nomen) and appearance (figura), a power (potestas) or sound,
and a description of the orthography involves simply a classifica-
tion of the letters according to their powers.1 Thus, orthography
from the time of Alfred to the present day has been delimited by
the letters and their powers. So ingrained has this principle become
that some contemporary linguists have attempted, by substituting
GRAPHEME for LETTER, to sanctify it with the countenance of lin-
guistic science without examining how unsound it is.2
Some critical exceptions to this view are found in the writings
of Francis (1958), Hockett (1958), and Vachek (1959). While all
three of these have recognized that the orthography is more com-
plex than an irregular letter-to-sound system, only Francis has
attempted, with considerable success, to analyze the relationship
between spelling and sound and to enumerate the non-phonemic
elements which enter into this relationship. Most other linguists
have paid little attention to the orthography. Bloomfield, for ex-
ample, held that writing was not a part of language, but simply
an imperfect image of speech, and even though he wrote at length
1
See Einar Haugen, The First Grammatical Treatise: The Earliest Germanic
Philology (Baltimore, 1950), pp. 41-2.
2
The so-called grapheme-phoneme parallel is discussed on pp. 47-49.
 ATTITUDES TOWARD ENGLISH ORTHOGRAPHY 17

on the teaching of reading, he maintained that English orthogra-

phy was simply a grossly irregular alphabetic system.8
Spelling reformers, who have contributed an enormous quantity
of literature on English orthography, have seldom analyzed the
object of their scorn beyond the more common examples of scribal
pedantry. Their arguments were, and are still, based upon the
a priori assumption that alphabets should, and by some right ought,
to be perfectly phonetic. It is no surprise, therefore, that most
spelling reformers have concerned themselves with direct letter-
to-sound relationships and have ignored all other facets of the
writing system.
Grammarians, like the spelling reformers, have also viewed the
orthography as a mirror for speech. The earliest grammarians
concerned themselves primarily with correct pronunciation and,
therefore, directed their attention towards relating spelling directly
to sound. As spelling became regularized, they turned more and
more toward establishing spelling rules, but still retained a direct
spelling-to-sound standpoint. From the time of the earliest English
grammars, however, a small number of grammarians have dis-
cerned non-phonetic features in the orthography, although few
carried out comprehensive analyses of such features.

2. GRAMMARIANS

a. Introduction

Grammarians from the time of the earliest English grammars to

the present time have generally shown little interest in orthogra-
phy. The earliest treatments of spelling contain enumerations of
the letters, based mostly upon the tri-partite system of Donatus
(name, appearance, and force), and occasional statements on how
the letters are to be pronounced. Prior to the sixteenth century,
the only grammars written in English were restatements of Latin
grammars based upon Donatus and Priscian, with little native
3
Seep. 11.
18 ATTITUDES TOWARD ENGLISH ORTHOGRAPHY

originality. 4 The revival of learning in the sixteenth century, how-

ever, brought a flood of literature on English grammar and con-
sequently a greater interest in orthography. Coincidental with this
phenomenon, the rise of printing hastened the regularization of
spelling, and for the first time, spelling rules for English were
formulated. By the eighteenth century spelling rules had become
the center of discussions about orthography.
Most English grammarians prior to the twentieth century viewed
spelling as a record of speech sounds, although some divergent
views were expressed from time to time. Direct spelling-to-sound
relationships form the basis of even the most enlightened treat-
ments of the orthography, like those by Jespersen and Sweet. 5 Of
these treatments of the orthography, three are especially significant
for the present work; these are the works of Alexander Hume
(c. 1617), James Douglas (c. 1740), and Goold Brown (1859).
Hume was one of the first grammarians to discuss orthographic
practices and to explain such early Modern English innovations
as the substitution of t for c and s in words of Latin origin like
nation and congregation. Douglas was one of the few grammarians
ever to attempt a complete description of English spelling-to-
sound correspondences, and while his rules generally neglect non-
phonetic features like morpheme identity, his results are as success-
ful as any of the modern letter-to-sound descriptions. Brown,
who published at least six different English grammars between
1820 and 1850, compiled in the Grammar of English Grammars
the opinions of several hundred grammarians on various topics
including the orthography. In what appear to be Brown's own
contributions to the topic of orthography, the notion that spelling
relates to something more than sound is advanced. Brown cer-
tainly did not view spelling as a direct reflection of speech, yet it
4
See George L. Kittredge, "Some landmarks in the history of English gram-
mars", Text-book bulletin for schools and colleges (N.Y., 1906), passim.
5
Henry Sweet discusses the orthography in several sections of his History
of English Sounds From the Earliest Period (Oxford, 1888). Jespersen was
especially interested in the English orthography, yet approached it only from
the sound-to-letter standpoint. See especially A Modern English Grammar on
Historical Principles (Heidelberg, 1909-1931), I, 3.
 ATTITUDES TOWARD ENGLISH ORTHOGRAPHY 19

is not clear what relation he postulated for spelling, sound and

meaning. His writings are, nevertheless, significant for a survey
of nineteenth century attitudes towards spelling, just as the works
of Douglas and Hume are significant for attitudes of the previous
centuries.

b. Alexander Hume

Alexander Hume's Of the Orthographie and Congruitie of the Britan

Tongue, written around 1617, is one of the earliest English works
in which spelling patterns are discussed.® Although born a Scots-
man, Hume spent sixteen years in England, studying, teaching,
and serving as schoolmaster at Bath. Upon returning to Scotland,
he became rector of a high school and later, principal master of
the grammar school of Dunbar, where he wrote the Orthographie
and Congruitie, dedicating it to James I. This book was apparently
designed as a spelling-book and grammar for use in the Dunbar
schools. Hume also wrote a Latin grammar which was, by decla-
ration of Parliament and the Privy Council, to be used in all the
schools of the Kingdom. There is no evidence, however, that the
injunction was carried out.
In the dedication to the Orthographie and Congruitie Hume
noted the "uncertentie in our men's wryting", and claimed to have
devised a "remedie for that maladie", which he said he put aside
when he learned of Sir Thomas Smith's spelling-reform proposal
(see page 30). The book itself is divided into two sections: "Of
the orthographie of the Britan tongue" and "Of the congruitie of
our Britan tongue". Spelling matters occupy approximately one-
third of the first section, the remainder of this section being de-
voted to the sounds of the "Latine" and "Britan vouales" and
"consonantes", and the syllable.

β
The text has been edited by Henry B. Wheatley and published in the EETS
Publication No. 5 (London, 1865). Some interesting remarks similar to Hume's
can be found in Mulcaster's Elementarie, ed. Ε. Τ. Campagnac (Oxford, 1925),
pp. 115ff. See also Eric J. Dobson, English Pronunciation, 1500-1700 (Oxford,
1957), I, 316-21.
20 ATTITUDES TOWARD ENGLISH ORTHOGRAPHY

The basis of orthography according to Hume consists of "...the

symbol, the thing symbolized, and their congruence". Congruence
is "... the instrument of the mouth, quhelk, when the eie sees the
symbol, utteres the sound". As for the symbols, he recommended
that i and u be used only for vowel sounds and j and ν only for
consonants, and assigned the names jod and jau to the latter two.
This separation was not widely adopted until almost 100 years
later. It was not, for example, advocated in Ben Jonson's grammar
which was published at least 15 years after the Orthographie and
Congruitie was published.7
Two short chapters in the first Section of the Orthographie and
Congruitie are devoted to spelling rules. The first, "of rules from
the Latin", is interesting because it states a rationale for the Latin
spellings which were adopted into English orthography during
the sixteenth century.
Heer, seeing we borrow miqkle from the latin, it is reason that we either
follow them in symbolizing their's or deduce from them the groundes
of our orthographie.8
Rule 2 of this chapter states that Latin derivatives written in Latin
with c, s, or sc for /s/ should retain their original spellings. The
distinction between c and s is especially important since "wordes
of one sound and diverse signification are many tymes distinguished
be these symboles". Hume cited such pairs as council·, counsil
and eel (cell): sei (sell). Latin verbals in tio, like oration, visitation,
and vocation are to be written with the t, but "wordes deryved
from the latin in tia and tium we wryte with ce; as justice, from
justitia". Hume noted also that although the final e in justice "be
idle, yet use hes made it tollerable to noat the breaking of the c."
The other rules for Latin words deal with Latin χ and with the
vowels.
In the second spelling chapter, entitled "Of some idioms in our
orthographie", Hume discussed, among less interesting topics, the
symbolization of the syllables /l/ and /p/, and final e. Since "the
7
Ben Jonson, The English Grammar (London, 1634).
8
Hume, p. 19.
 ATTITUDES TOWARD ENGLISH ORTHOGRAPHY 21

ear can hardlie judge quhither their intervenes a voual or noe"

in the syllabic endings of '7/i/e, mikle, muttne, eatne", he left to
the "wil of the wryter" whether to write the e before the final
consonant or after it. Hume rejected the idea that final e marks
the quantity of a preceding vowel, because he did not believe that
one vowel could change the sound of another if they were sep-
arated by a consonant.
... but it is as untrue that the voual e behind the consonant doth change
the sound of the voual before it. A voual devyded from a voual be a
consonant can be noe possible means return through the consonant
into the former voual. ... Nothing can change the sound of a voual but
an other voual coalescing with it into one sound. ...·
He did, however, approve of the final e to "break the sound" of
c and g as in peace and savage, and after s as in false and case, for
which he gave no reason. The e here is probably to show that s
is pronounced /s/ rather than /z/, as it would be in false and case
if the e were not present. As Wheatley pointed out in his notes
to the Ε. Ε. T. S. edition of Hume's Orthographie and Congruitie,
Hume's spelling itself is occasionally inconsistent. Judge, for ex-
ample, is rendered alternately as judge (page 8), juge (page 18),
and judg (page 33). Nevertheless, Hume's notes are valuable for
what they reveal about the development of orthographic practices.

c. James Douglas
One of the more thorough treatises now in existence on English
spelling-to-sound correspondences was written by James Douglas
around 1740. Douglas, physician, anatomist, and fellow of the
Royal Society, wrote, but never published, drafts of Latin, Greek,
French and English grammars. The manuscripts are now housed
in the Hunterion Museum, Glasgow, and a fragment on English
orthography (Η. M. 586), has recently been edited by B. Holmberg
and published in Lund Studies in English.10 In writing his treatise
9
Ibid., p. 21.
10
Borje Holmberg, James Douglas on English Pronunciation, c. 1740 (Lund,
1956). The most complete summary of direct spelling-to-sound correspond-
ences is Axel Wijk's Rules of Pronunciation for the English Language (London,
1966).
22 ATTITUDES TOWARD ENGLISH ORTHOGRAPHY

on spelling-to-sound correspondences, Douglas attempted to re-

cord the upper-class speech of London. As a phonetician he was
competent, but more important for the interests of this paper,
he gave hundreds of rules for predicting sound from spelling and
illustrated his rules with over 6,500 different words. The rules are
based mostly upon the syllable position in the word and the letter
position in the syllable. More sophisticated ideas, like those based
upon accent position and morpheme identity are also used, but
not frequently. Typical of his rules is "When the vowel A makes
a compleat syllable in the beginning of a word, & is not a pre-
position, it is sounded long & slender, as, ... Ä-BLE. . . . ' N I
Douglas attempted to use morpheme identity in some of his
rules, but usually resorted to his favorite stand-by, syllable divi-
sion, when nothing else worked. For example, I is long "in mono-
syllables before Gn, as S I G N " , and "in the first syllable of deriva-
tives from monosyllables that are long, as ... SIGNED." If DERIVA-
TIVES is limited to a certain class of derivational and inflectional
forms, this rule is valid. But for the final cluster gn, whose pat-
terning can be derived from the rules for the I before it, Douglas
resorted to syllable division, which in this case makes his rules
either circular or useless.
When gn closes the syllable in the middle of a word, the consonant G
is not sounded, as ASSIGNEE. ... Whe G & Ν are divided into two syl-
lables the consonant G is sounded hard, as ASSIGNATION.12
Douglas's rules are primitive. His selection of graphemic units
is not based upon any discernible criteria; ya, ye, wa, wi, for ex-
ample, are classed as diphthongs, and awe, ewe, eye, way, with
numerous others, are classed as triphthongs. Furthermore, Dou-
glas did not recognize the existence of /η/, but rather, interpreted
it as /n/ plus /g/. He also claimed that certain consonants were
sounded double: "... the consonant Β is sometimes sounded
double, as, ... C A B B I N . " 1 3 The effects of certain consonant groups
like Id and ft on preceding vowels are recognized, although the
11
Ibid., p . 128.
12
Ibid., p . 264.
18
Ibid., p . 233.
 ATTITUDES TOWARD ENGLISH ORTHOGRAPHY 23

effect of accent on the pronunciation of intervocalic χ is not. The

merit of his work lies more in the intent than the results, although
the 358 rules he presented, along with the lists of examples and
exceptions, represent an extremely thorough attempt to analyze
English spelling-to-sound correspondences.

d. Goold Brown

Goold Brown's Grammar of English Grammars, first published in

1850, is important both for what Brown himself says about or-
thography in remarks scattered over one thousand pages, and for
the statements of the grammarians he cites.14 The introduction
of the Grammar contains the names of 548 grammatical works
by 452 authors, including Aristotle's Poetics and the works of
Bullokar (1586), Gill (1621), and B. Jonson (1634). An entire
chapter is devoted to William Lily and Lindley Murray whose
grammars were exceedingly popular in the middle of the nineteenth
century. The section on orthography follows the traditional or-
ganization of letters, syllables and words, but includes a final
section on spelling where such nineteenth century conflicts as those
over final c vs. final ck and our vs. or are aired in full. To the graph-
emic triumvirate of NAME, APPEARANCE and FORCE Brown added
CLASS, a distinction used by Donatus, but not placed on the same
level as the other three properties. The two major classes of graph-
emic units are consonant and vowel, and consonants are further
divided into semivowels and mutes. From this traditional intro-
duction, Brown, in defining the powers of the letters, suggested
that letters may relate to more than sounds.
The deaf and the dumb, also, to whom none of the letters express or
represent sounds, may be taught to read and write understanding^. ...
Hence it would appear that the powers of the letters are not, of neces-
sity, identified with their sounds; the things being in some respect dis-
tinguishable, though the terms are commonly taken as synonymous.15
Brown claimed to have had no interest in spelling reform as such,
11
Goold Brown, Grammar of English Grammars (4th ed.; New York, 1859).
15
Ibid., p. 15.
24 ATTITUDES TOWARD ENGLISH ORTHOGRAPHY

but rather, in regularizing spelling. He summarized his observa-

tions on spelling with a plea for basing spelling on pronunciation,
etymology, and "the analogy of the particular class of words to
which it belongs". Some of his statements express quite sophisti-
cated views on orthography, but there is insufficient evidence in
his writings to determine what overall view he held about the or-
thography and whether the views he presented were his own or
were summarized from other grammarians.
In his most progressive mood, Brown wrote that "words are
not mere sounds, and in their orthography more is implied than
in PHONETICS or PHONOGRAPHY. Ideographic forms have, in gen-
eral, the advantage of preserving the identity, history, and lineage
of words...." 16 Brown's view is not unique; many of the early
spelling reformers like Gill and Bullokar made similar statements.
Unfortunately, Brown did not develop this idea further, and the
only statement he made which bears directly on the question is
shrouded in mysticism and retreat.

In their definitions of vowels and consonants, many grammarians have

resolved letters into sounds only. ... But this confounding of the visible
signs with the things which they signify, is very far from being a true
account of either. Besides, letters combined are capable of a certain
mysterious power which is independent of all sound, though speech,
doubtless, is what they properly represent.17

It is evident, nevertheless, that Brown wanted to separate writing

from speech and avoid the confusions that arose in many earlier
discussions of the subject. His spelling rules are nearly complete
by modern standards, including such notes as "Monosyllables and
English verbs end not with c, but with ck for double c"; moreover,
his hints about the relationship of spelling to meaning foreshadow
the writings of H. Bradley. Equally important are the citations
from other grammarians on orthography, and the lengthy dis-
cussions on spelling disputes of the nineteenth century. The final
c vs. ck (music vs. musick, etc.) controversy, for example, is pre-

" Ibid., p. 203.

17
Ibid., p. 149.
 ATTITUDES TOWARD ENGLISH ORTHOGRAPHY 25

sented with quotations from over twenty different British and

American grammarians, including Lowth, Murray, Walker, John-
son, and Webster.

3. CONTEMPORARY LINGUISTS

The first twentieth century linguists were so adamant in pointing

out the nineteenth century confusion of sound and spelling that
they reduced the orthography to a secondary, subservient role
from which it has infrequently emerged in the writings of con-
temporary linguists. The relationship between spelling and sound
ranked high among the problems the first Indo-European com-
parative philologists faced. The early works of Bopp, Rask, and
Grimm show classifications of sounds, based upon orthographic
rather than phonetic data. On Bopp, F. de Saussure wrote "Even
Bopp failed to distinguish clearly between letters and sounds. His
works give the impression that a language and its alphabet are
inseparable."18
Part of the early failure to distinguish clearly between phonology
and orthography stemmed from the lack of an adequate vocabulary
for discussing phonological phenomena. Grimm, for example, in
Deutsche Grammatik, titled his discussion on sound changes
"Changes of the letters". By the time of the publications of W. D.
Whitney's The Life and Growth of Language (1874) and Hermann
Paul's Prinzipien der Sprachgeschichte (1880), philologists had re-
solved most of the confusion between spelling and sound. Never-
theless, F. de Saussure inveighed against the orthography as the
obfuscation of the true language. To Saussure are attributed the
following remarks in the posthumous Cours de linguistique ginerale.

Language and writing are two distinct systems of signs; the second
exists for the sole purpose of representing the first. ... The preceding
discussion boils down to this: Writing obscures language; it is not a
guise for language but a disguise.19
18
Ferdinand de Saussure, Course in General Linguistic, trans. Wade Baskin
(New York, 1959), p. 10.
19
Ibid., pp. 23, 30.
26 ATTITUDES TOWARD ENGLISH ORTHOGRAPHY

To Saussure can also be attributed the first exposition on the so-

called grapheme-phoneme parallel, a superficial relationship that
is frequently invoked without critical comment by some contem-
porary linguists. Saussure saw four features in the letters (he did
not use the term grapheme) which paralleled features in the phone-
mic system. His features, however, are equally applicable to mor-
phemic and syllabic writing systems.
The phoneme-grapheme parallel has also been subscribed to by
linguists since Saussure. Pulgram upped the count of parallel
features to nine, including such items as the following:20
Ρ 6 The phonetic shape of an G 6 The graphic shape of an
allophone is dependent on allograph is dependent on
its producer and on its its producer and on its
phonetic surroundings graphic surroundings
Ρ 8 Dialects are subject to G 8 Alphabets are subject to
phonemic change and sub- graphemic change and
stitution substitution
Stetson in a brief note on phoneme and grapheme claimed that to
understand written forms, one must understand writing move-
ments, just as, apparently, one must understand articulatory move-
ments to understand spoken forms.21 Bazell, disagreeing with
Pulgram in particular, objected to the phoneme-grapheme par-
allel on the grounds that phonemes contain simultaneous distinc-
tive features while graphemes contain non-simultaneous ones.22
Bazell's solution is that "the letter ... answers to the morpheme...."
More recent renditions of the phoneme-grapheme parallel can be
found in the writings of Francis (1958), Gleason (1961) and Hall
(1964).
Considerations of the orthography by linguists, from the time
of Saussure through the second World War, were directed mostly
towards spelling reform. Many scholars who demonstrated adept-
20
Ernst Pulgram, "Phoneme and Grapheme: A Parallel", Word 7 (1951),
15-20.
21
Raymond H. Stetson, "The Phoneme and the Grapheme", Melanges de
linguistique et de philologie offerts ä Jacq. Van Ginneken (Paris, 1937).
22
C. E. Bazell, "The Grapheme", Litera 3 (1956), 43.
 ATTITUDES TOWARD ENGLISH ORTHOGRAPHY 27

ness and unbiased critical ability in all other phases of linguistic

investigation could, in their passion for reforming English spelling,
see nothing in the prevailing orthography except a defective al-
phabetic system, badly in need of repair. Bloomfield (see page 11)
is responsible probably more than any other contemporary linguist
for the view that writing is secondary and subservient to speech.
This notion was espoused not only in Language, but also in several
later articles. 23 Not all linguists in this century, nevertheless, held
this view, and some, like H. Bradley who was cited earlier (see
page 24) even claimed that writing does not attempt to relate directly
and solely to speech. In commenting on graphic distinctions of
homophones, Bradley stated
It is because the expression of meaning is felt to be the real purpose
of written language that these distinctions still survive, in spite of the
disasterous effects that they have had on the phonetic intelligibility of
written words.24
In summarizing his views on writing and speech, Bradley stated
further that "speech and writing are two organs for the expression
of meaning, co-ordinate and mutually independent". 25
A more succinct summary of this relationship was made by
C. F. Hockett in 1958:
The complexities of English spelling cannot be accounted for completely
on the assumption that the system is phonemic with irregularities of
the sort listed ... [above]. It is necessary to assume that the system is
partly phonemic and partly morphemic.28
This view was also expressed in 1941 by Edgerton in direct reply
to Bloomfield's idea that writing relates only to speech. "Writing
consists in the conventional use of visible symbols for the recording
or transmission of ideas, or of ideas and sounds ... or of sounds
unaccompanied by ideas." 27
23
See especially "Linguistics and Reading", The El. Eng. Rev. 19 (April,
1942), 125-130, 19 (May, 1942), 183-86.
24
Bradley, 1928, p. 176.
25
Ibid., p. 186.
28
Charles F. Hockett, A Course in Modern Linguistics (New York, 1958),
p. 542.
27
William F. Edgerton, "Ideograms in English Writing", Lg. 17 (1941), 149.
28 ATTITUDES TOWARD ENGLISH ORTHOGRAPHY

The Czech linguist Joseph Vachek has advanced, in a series of

articles over the past thirty years, one of the most critical analyses
of the English spelling-to-sound relationship. Vachek first dis-
tinguished the aims of the traditional writing system of a language
from its phonetic transcription.
While any system of phonetic transcription provides means for an op-
tical recording of the purely acoustic make-up of spoken utterances,
the traditional writing system increasingly tends to refer to the meaning
directly without necessarily taking a detour via the corresponding
spoken utterance.*8
He then went on to give definitions of the spoken and written
norms of language and to develop, in a general fashion, the hier-
archic relation of the two norms of English.
Francis, in enumerating the nonphonemic features of English
writing, pointed out not only the graphemic differentiation of
homophones, but also the use of graphemic markers to show
the phonemic correspondences of other graphemes, and the gen-
eral tendency in English spelling to preserve morphemic identity
regardless of phonemic differences.29 Francis's most important
result is the following:
We may state this in terms of a general principle which, while admitting
many exceptions, is what governs and systematizes many of the apparent
inconsistencies of our wiiting system: The English writing system tends
to employ a single combination of graphemes to represent a given mor-
pheme, disregarding for the most part all except the grossest phonemic
differences between allomorphs.80
Several twentieth century linguists have written on spelling-to-
sound correspondences with the intent of improving the teaching
of reading. Bloomfield, who condemned the orthography to eter-
nal purgatory in Language, wrote several articles on the teaching
of reading and in collaboration with Barnhart wrote an introduc-
28
Josef Vachek, "Two Chapters on Written English", Brno Studies in English,
I, 8 (Praha, 1959), see also Josef Vachek, "Some remarks on Writing and
Phonetic transcription", Acta Linguistica 5 (1945-49), 86-93.
29
W. Nelson Francis, The Structure of American English (New York, 1958),
pp. 450-69.
80
Ibid., p.m.
 ATTITUDES TOWARD ENGLISH ORTHOGRAPHY 29

tory reading text.31 Bloomfield published nothing on spelling-to-

sound correspondences themselves. What notions he had on
patterning in the orthography, judging from his published works,
were based upon direct letter-sound relationships.
Another linguist who has written on reading is C. C. Fries, who
set forth his views recently in Linguistics and Reading** Fries
brings to this task not only the tools of an eminent scholar, as did
Bloomfield, but also those of an educator who has worked for
over a half century in language teaching. His views on spelling-
to-sound correspondences, however, are adapted for classroom
application and consequently are non-technical in the linguistic
sense. Only direct spelling-to-sound relations are mentioned, al-
though the single-letter to single-sound correspondence has been
replaced by the SPELLING PATTERN that corresponds as a whole
to a sound or sequence of sounds. Forms like the final e in like
are not seen as markers, but as parts of a larger, somewhat in-
definite, spelling pattern. Morphemic elements are totally neg-
lected as are all other non-phonetic influences.
Modern English spelling is fundamentally a system of a comparatively
few arbitrary contrastive sets of spelling patterns, to which readers, to
be efficient, must, through practice, develop high-speed recognition
responses.38
In contrast to this, Francis, who acknowledged both phonemic
and morphemic elements in the orthography, considered the rec-
ognition of morphemic elements as important as the recognition
of phonemic ones.
Efficient reading is really a combination of two skills, both of which
must be taught by any satisfactory method of teaching reading. They
are, first, the ability to recognize accurately a large number of words
and word groups as wholes ... and second, the ability to work out the
pronunciation of an unrecognized graphic configuration, in terms of
its constituent phonograms and morphograms, until it is recognized
as a unit of the spoken vocabulary.84
31
Leonard Bloomfield, 1933, 1942; L. Bloomfield and C. L. Barnhart, Let's
Read, a Linguistic Approach (Detroit, 1961).
82
Charles C. Fries, Linguistic and Reading (New York, 1962).
83
Ibid., p. 183.
84
Francis, 1958, p. 558.
30 ATTITUDES TOWARD ENGLISH ORTHOGRAPHY

4. SPELLING REFORMERS

The literature on spelling reform is imbued with a revivalistic

quality; it sees the prevailing orthography as the degenerate pro-
duct of neglect, denseness, and lack of adaptability. It sees the
future of the English speaking people, if this malignancy is allowed
to remain, as continued degeneration: children failing to learn to
read, adults falling prey to the corrupted speech of the lower
classes, world trade falling off, and business suffering throughout
the land. Nevertheless, from the Ormulum to the recent Initial
Teaching Alphabet, spelling reformers, either covertly or overtly,
have advanced theories of writing and notions about the relation-
ship of spelling and sound.
The orthography of Orm, preserved in a holograph manuscript
written around 1200 (The Ormulum) is the earliest remaining ex-
ample of English spelling reform.35 The orthographic regularity
in this manuscript is unmatched until well after the introduction
of printing and shows, among other things, an attempt to indicate
vowel quantity through consonant doubling and accent marks.
As a spelling reformer, he is both clumsy and dull, but as an
indicator of concern for spelling-to-sound correspondences in the
thirteenth century, Orm is unique. We can only assume from
observing the form and regularity of Orm's spelling system, that
he had spent some time analyzing the relationship between spelling
and sound and had done so from a pure letter-sound basis. If
Orm advocated a general spelling reform, then he failed. The
text of the Ormulum is the only evidence we have of Orm's spelling
system.
From the time of the Ormulum until the middle of the sixteenth
century, we have no evidence of spelling reform attempts. Then,
suddenly, in England spelling reformers flourished. Sir Thomas
Cheke and Sir Thomas Smith, co-defendants in the great Greek
controversy of the sixteenth century, were the first, followed
closely by John Hart, the most competent by far, and then by an
85
Bodleian Ms. Junius 1. See also R. W. Burchfield, "The Language and
Orthography of the Ormulum MS", Trans. Phil. Soc. (1956), 56-87.
 ATTITUDES TOWARD ENGLISH ORTHOGRAPHY 31

endless succession of others up to the present day. Most of the

spelling reformers viewed writing as a mirror for speech, but a
few like Bullokar and Gill proposed alphabets based at least in
part upon etymology and upon the desire for a graphic distinction
of homophones, and recognized non-phonemic elements in the
orthography.
The earliest sixteenth century spelling reformers, Sir Thomas
Cheke and Sir Thomas Smith, did not wage reform campaigns.
That Cheke was interested in spelling reform is known only from
his translation into a reformed English alphabet of the Gospel
according to St. Mark (c. 1550), and from a letter he wrote to Sir
Thomas Hoby, which was published in 1561.36 Except for the
symbol ω which he substituted for the spelling oo, Cheke employed
only Latin letters, using, among other devices, geminate vowels
to show quantity, geminate consonants to show a preceding short
vowel, and sch for /s/. Although inconsistent in the employment
of his orthography, Cheke attempted to devise a phonetic tran-
scription system, as did Smith who saw writing only as the imita-
tion of speech. Smith published his views on spelling reform in
De recta & emendata linguae anglicae (1568), b u t did not advocate
a particular spelling system. Rather, he offered variant spellings
for the writer to choose from. 37 Smith worked from sound to
spelling, employing at least thirty-four unique letter-forms along
with the diaeresis, circumflex and hyphen.
John Hart was the most competent phonetician of the early
reformers and also the most evangelistic. He saw in the orthog-
raphy the vices of "diminution" (lack of enough letters), "super-
fluity" (use of superfluous letters), "usurpation by one letter of
the powers of another", and "misplacement of letters" (failure of
the order of the spelling to correspond to the order in which the
letters are pronounced). In three major works published between
1551 and 1570, Hart enumerated and then vanquished the argu-
ments against spelling reforms, set forth a reformed alphabet based
entirely upon the existing Latin alphabet, and transcribed some
34
See Dobson, 1958,1, 38-46.
37
Ibid., 46-62.
32 ATTITUDES TOWARD ENGLISH ORTHOGRAPHY

popular texts in his system.38 Although his system is more con-

sistent than those which followed him in the next two centuries,
his chief value today is in the record he left of the speech of his
time. While claiming that writing "is to leave a record of our
thoughts", Hart could see no other way to achieve this goal than
to write phonetically.
The major spelling reformers who followed Hart contributed
little to the understanding of attitudes toward writing and the
relation between writing and speech. While some of the later
reformers based their systems upon etymology and analogy (see
especially James Elphinstone) the majority advocated phonetic
systems, and cited, to establish a need for reform, the same worn-
out examples of scribal pedantry like debt, doubt, and victuals.
Beginning with the Spelling Reform Association which was
founded in Philadelphia in 1876, various societies have been
formed to further the spelling reform cause. Before the end of the
Second World War the Anglic Association, the Simplified Spelling
Society, the British Spelling Reform Association, the Simplified
Spelling League and the Simplified Spelling Board organized,
published manifestoes and occasional periodicals, and faded from
existence. Within the ranks of these ephemeral societies, never-
theless, were counted such scholars as W.A. Craigie, A.J. Ellis,
C.Grandgent, D.Jones, G.Krapp, G.Murray, and R.Zachrisson.
The twentieth century reformers have in general presented an even
more distorted picture of the orthography than their predecessors.
Their arguments, instead of being based on the true irregularities
which exist, generally are based upon non-existent patterns like the
celebrated ghoti for /fis/ or upon renderings like the following:

If one used all possible combinations, the word scissors ... might be
spelt in 596,580 different ways. Dr. G. Dewey ... has calculated that
the word foolish, if dealt with in the same way, might be spelt in 613,975
different ways.39
88
Hart's three works have been edited by Bror Danielsson in John Hart's
works on English Orthography and Pronunciation (Stockholm, 1955).
89
R. E. Zachrisson, "Four Hundred Years of English Spelling Reform",
Studio Neophilologica 4 (1931), 5.
 ATTITUDES TOWARD ENGLISH ORTHOGRAPHY 33

A recent spelling reform revival has taken place in this country

and in England, led by Sir James Pitman, the grandson of the
inventor of the Pitman shorthand system.40 While the earlier
spelling reformers advocated complete and irreversible reform,
Pitman has limited his interests to an Initial Teaching Alphabet
(ITA) for use in the teaching of reading. Once adequate skill in
reading in ITA is gained, the reader is transferred to conventional
orthography.41 The British government is sponsoring experiments
with ITA in the British primary schools, and a number of American
schools have either experimented with the system or have ex-
pressed a desire to do so. While the immediate results of experi-
ments with ITA have been successful according to the proponents
of the system, few critical analyses of the experiments or their
results have been published.

40
Isaac J. Pitman, "Learning to read: An Experiment", J. Royal Society of
Arts, 109 (1961), 149-180.
41
See John Downing, "Pitman's Initial Teaching Alphabet", Report given
to the Conference on Perceptual and Linguistic Aspects of Reading (Stanford,
1963).
 III

PRELIMINARIES TO ORTHOGRAPHIC ANALYSIS

1. GRAPHEMIC FEATURES

a. Spelling Units

Even from the direct letter-to-sound standpoint the graphemic

system is more complex than is revealed in the notion that there
are twenty-six letters, or graphemes, which through careful ma-
nipulation, are mapped into the phonemes of English. From the
enumeration of the twenty-six graphemes to the point where
correspondences to phonemes can be considered, a number of
complexities must be untangled. One is the designation of the
spelling units themselves. Obviously, there are more than twenty-
six functional units; th, ch, and oo, for example, are as basic to
the current orthography as a, b and t. But are tch, ck, and dg
primitive units, on a level with a and th, or are they in some sense
compound units, whose correspondences to sound can be pre-
dicted from their immediate constituents?
If spelling can be mapped into sound, then regardless of the
intermediate levels which are introduced, graphemic words must
be segmented into their basic graphemic units. This requires that
there be some way of handling letters like the final e in rove and
the b in debt. Is, for example, the e in rove connected to o, forming
the discontinuous unit o...e, or is it part of the unit ve, or is it a
unit by itself? And, similarly, how is the b in debt to be handled?
As part of the unit eb, or of bt, or as a separate unit? The solution
to this problem should not only be consistent with the way similar
 PRELIMINARIES TO ORTHOGRAPHIC ANALYSIS 35

graphemes are handled, but also general enough to handle new

cases which may arise. The designation of spelling units in Web-
ster's New Collegiate Dictionary,1 for example, fails to meet both
of these aims. The editors give no general rules for handling silent
letters, but instead haphazardly and quite inconsistently classify
individual cases as they arise in traversing an alphabetical list of
spelling units, gh, for example, "in aghast, ghostly and ghost is a
useless spelling for 'hard' g. ..."2 rhy however, as in rhetoric and
rhesus is not mentioned as a separate unit, ng is a single unit in
words like long, corresponding to /η/, but mb, on the other hand,
is two units, the second being silent.
These problems cannot be settled satisfactorily by simply label-
ing all unpronounced letters as silent. Consider the so-called
silent b's in subtle and bomb. One could say as Webster's does,
that the b's in these two words are silent, and let the matter rest.
But by doing so, an important difference that exists in these two
cases would be neglected. The form subtle occurs only with the
b corresponding to zero, but bomb in bombard and bombardier
has non-silent b. It is not sufficient therefore to say that the second
b in bomb is silent; the more exact statement is that it is silent
before word juncture and before certain suffixes (cf. bombing,
bombs, bombed). This is one of the forms of orthographic pat-
terning that almost all traditional treatments of spelling overlook.
Its full implications are explored in a later section of this work.
Another inherent feature in the orthography is the distinction
between functionally simple and functionally compound conso-
nant units. One of the most general, although not entirely regular,
spelling-to-sound rules is that the vowel spellings a, e, i, o, u are
mapped into one form before a single consonant unit which is
followed by a vowel and into another form in all other environ-
ments. In the vocabulary of the direct letter-to-sound school,
these forms are the LONG and SHORT pronunciations of the vowels,

1
John P. Bethel (ed.), Webster's New Collegiate Dictionary (2d ed., Spring-
field, Mass., 1956).
2
Ibid., p. xii.
36 PRELIMINARIES TO ORTHOGRAPHIC ANALYSIS

as shown in the examples below (FREE and CHECKED in the vocab-

ulary used here — see page 14).

free checked
pronunciation pronunciation
β Ν Μ
anal annals
ache ratchet
e Ii I Μ
fetal fettle
ether hedge
i /«/ Μ
hypo h/ppo
wr/the whittle
ο /ο/ Ν
phonograph sonnet
kosher noxious
u /(j)u/ Μ
swper swpper
luxury

To apply this rule, simple and compound consonant units must

be differentiated, a task which cannot be done by simply counting
the number of letters involved, sh, for example, is composed of
two letters, yet it functions as a simple unit as in kosher, x, on the
other hand, contains only one letter, yet it functions as a com-
pound unit, as in luxury and noxious. What must be formulated
is a consistent criterion for classing consonant units as simple or
compound. While the classifications of x, ch, th, ph, and rh may
be intuitively obvious, those of ck, dg, and tch are not. What is
important is first that, the rule mentioned above, and, as will be
shown soon, almost all spelling-to-sound rules, be based not upon
letters or graphemes as such, but rather upon functional spelling
units, and second, that functionally simple and functionally com-
pound units be distinguished.
 PRELIMINARIES TO ORTHOGRAPHIC ANALYSIS 37

b. Graphemic Alternations

A feature of the graphemic system which has arisen partially from

scribal necessity and partially from pedantry is the alternation of
various letters according to their graphemic environments. In
such cases two different letters which correspond to the same
sound occur in complementary (or near-complementary) distribu-
tion. The functionally simple vowel spellings i and y, for example,
alternate, y occurring generally in final position and i in all other
positions. In addition, regular rules control the alternation of
final y to i before certain suffixes (see page 59). This alternation
holds not only for the simple vowel spellings i and y, but also for
the compound spellings in which these two letters occur as the
second elements, e.g., ai, ay, ei, ey. In the compound units, the
y spellings generally appear before other vowel spellings and
in morpheme final position, and the i spellings appear in all other
positions. For example, bait'.bay, heinous: grey, boisterous: boy.
u and w, when each corresponds to /w/, also occur in comple-
mentary distribution. As a simple consonant spelling, u occurs
in a limited number of environments: after q, g, s, and a few
others — in all other cases w occurs. As second units in compound
vowel spellings, u and w alternate similarly to i and y. Thus,
w occurs as the second element before other vowel spellings and
in morpheme final position, while u occurs in all other posi-
tions, e.g., auction:awe:draw; feud:ewer: flew\ ounce:coward:vow.
Several other alternation patterns of lesser significance should
also be considered in an exhaustive analysis of the graphemic
system.

These patterns are listed below.3

(1) ous/os Word final ous becomes os before the suffix ity,
e.g., curious'.curiosity.

8
On these patterns, see further, pp. 59-62.
38 PRELIMINARIES TO ORTHOGRAPHIC ANALYSIS

(2) i/e i in the suffix ity becomes e when the suffix is

preceded by i, e.g., society, variety, sobriety.
(3) erjre With the addition of certain suffixes, word final
er becomes re and then the e is dropped; thus,
center'.central·, theater'.theatrical. In addition, er
and re are in complementary distribution in word
final position, re occurs after c and g, and er
occurs in all other positions.
(4) Consonant Gemination of a final consonant occurs before
gemination certain suffixes, e.g., run:running, hop:hopped.
(5) e/0 Final e alternates with zero under certain types
of suffixation. Thus, dive:diving.
In addition to these alternations, a number of graphemic substitu-
tions, introduced mostly between the times of Chaucer and Shake-
speare, must be treated separately. One of these is the substitution
of t for c in suffixes like tion and tial, e.g., nation, essential (cf. ME,
nation, essenciall). Early Modern English scribes, instilled with
the fervor of classical learning, brought about in these substitu-
tions one of the few true spelling reforms in English orthographic
history. Their notion of reform, however, was to restore the ap-
pearance of the Latin root in favor of a more phonemic spelling.
Another graphemic substitution was the early Modern English
replacement of u with ο in the vicinity of m, n, u (v). This sub-
stitution accounts for many of the so-called irregularities like
some, love, and ton* All of these alternations and substitutions
are parts of the current orthography and must be considered in
a description of orthographic patterns.

4
"In ME texts of a more recent date (Chaucer, etc.) we find ο used still more
extensively for /u/, namely in the neighborhood of any of the letters m, rt, and
u (v, w). The reason is that the strokes of these letters were identical, and that
a multiplication of these strokes, especially at a time when no dot or stroke
was written over i, rendered the reading extremely ambiguous and difficult ...
this accounts for the present spellings of won, wonder, worry, woman ... above,
love... and many others." Otto Jespersen, 1909, III, 482.
 PRELIMINARIES TO ORTHOGRAPHIC ANALYSIS 39

2. CORRESPONDENCES

Having pursued the graphemic labyrinth through its intra-graph-

emic complexities, the next task is to analyze the relationships of
these units to sound. The object here is to show that even if the
direct spelling-to-sound view is assumed, more types of relation-
ships must be considered than the simple regular-irregular classes
that bisect the traditional approach to this subject. Furthermore,
it will be shown that the concepts of regular and irregular are far
more complex than is generally assumed, and, indeed, require
quite sophisticated notions for adequate definition. For the pres-
ent, however, I will use 'regular* and 'irregular' in a loose sense,
meaning simply, high frequency and low frequency, without care-
ful enumeration of what objects are to be counted to arrive at such
statistics.
Regular spelling-to-sound correspondences can be classed first
as either INVARIANT or VARIANT. / , for example, corresponds
regularly to /f/. In fact, this correspondence is so regular that
only one exception, of, occurs among the 20,000 most common
English words. Several other consonant spelling units like ck,
m, y, and ζ are also invariant or nearly so. The vowel spellings,
which are described in Chapter VI, (though not from the direct
spelling-to-sound view), are rarely invariant, though not IRREG-
ULAR in most cases.
Variant correspondences are those correspondences that are
still regular, but that relate the same spelling to two or more pro-
nunciations, depending upon graphemic, phonological, or gram-
matical features. Initial c, as an example, corresponds to /s/ when
it occurs before e, i, y; in all other positions it corresponds to /k/.
The spelling k corresponds to zero in initial position before w,
e.g., knee, know, knife. In all other positions, k corresponds to
/k/. This is graphemic conditioning from the letter-sound stand-
point. As will be shown later, the silent initial k is explained more
adequately by phonotactical rules. (The cluster /kn/ does not
occur within a single morpheme in English; where such prohibited
40 PRELIMINARIES TO ORTHOGRAPHIC ANALYSIS

consonant clusters would otherwise occur in morpheme-initial

position, the first consonant is dropped. Thus knee, gnat, ptar-
migan, pneumonia, psychology).
Position alone may determine the correspondence of a spelling
unit. For example, initial gh always corresponds to /g/: ghost,
gherkin, ghoul (but never to /f/ as assumed in the spelling reform
creation ghoti), but medial and final gh have pronunciations be-
sides jgj, as is too often pointed out in spelling reform tracts.
Stress may also be a conditioning factor for regular, variant cor-
respondences. The most prominent role that stress plays in spell-
ing-to-sound correspondences is in the pronunciation of unstressed
vowels. While the reduction of unstressed vowels to schwa is not
entirely regular, it can still be predicted in many cases. The pat-
terns, however, are highly complex and are beyond the scope of
this paper. A more interesting example of stress conditioning
occurs in the correspondences for intervocalic x, which corresponds
either to /ks/ or /gz/, depending upon the position of the main
word stress. If the main stress is on the vowel preceding jc, the
pronunciation generally is /ks/ as in exit, exercise. Otherwise, the
pronunciation is /gz/ (cf. examine, exist). While this rule is similar
to Verner's Law for the voicing of the Germanic voiceless spirants,
it is not a case of pure phonological conditioning. Words like
accede and accept have the identical phonetic environments for
/gz/, yet have /ks/.
Another type of correspondence in which stress is important
is the palatalization of /sj, zj, tj, dj/ to /s, z, c, j/. This form of
palatalization occurs when /sj, zj, tj, dj/ are followed by an un-
stressed vowel, as in social, treasure, bastion, and cordial (see
Chapter VI).
The retention or deletion of medial /h/ in most cases also depends
upon the position of the main word stress. Compare prohibit:
prohibition', vehicular'.vehicle. In each pair, the first member,
which has the stress on the vowel following h, has a fully pro-
nounced /h/, while the second member, with an unstressed vowel
after h, has no /h/. This rule also holds for vehement, shepherd,
philharmonic, annihilate, rehabilitate and nihilism, all of which
 PRELIMINARIES TO ORTHOGRAPHIC ANALYSIS 41

generally have silent /h/. (Some forms like these may have /h/
occasionally preserved by over-correct pronunciations).
Irregular spelling-to-sound correspondences also show important
differences. Arcing and cello, for example, both have irregular
correspondences for c, yet there is an important distinction be-
tween these two irregularities. Arc, from which arcing is derived,
has the correct correspondence for c. When suffixes beginning
with e, i, y are added to words ending in c, a k is normally inserted
after the c, as in picnicking (cf. picnic) and trafficked (cf. traffic).
The irregularity in arcing, therefore, is in the irregular formation
of the derivative. Cello, on the other hand, contains an aberrant
correspondence for c, paralleled by only a few other Italian bor-
rowings.5 In the examination of the influence of morphemic fea-
tures upon spelling-to-sound correspondences which follows, even
more patterning appears, even though the direct letter-sound ap-
proach shows only irregularities.

3. MORPHEMIC FEATURES

a. Morpheme Boundaries

If the investigation of morphemic features is begun with the as-

sumption of a direct relationship between spelling and sound,
then problems appear immediately. The spelling ph, for example,
regularly corresponds to /f/ as in phase, sphere, and morpheme.
In shepherd, however, ph clearly does not correspond to /f/, but
to f p f . One way to explain this is to say that shepherd is an excep-
tion to the more general rule of ph — /f/. But if this is done, then
the same process must be repeated when faced with uphill, tophectvy
and every other form in which ph occurs across a morpheme
boundary. The most satisfactory procedure is to say that ph cor-
responds to /f/ when it lies within a single graphemic allomorph
and that across morpheme boundaries ph is treated as the separate
5
Concerto and proper nouns like Cellini are the most common examples.
42 PRELIMINARIES TO ORTHOGRAPHIC ANALYSIS

letters ρ and h. Therefore, one factor that should be considered

in the spelling-to-sound relationship is morpheme boundaries.
That this factor is not unique to ph can be seen from the following
examples.
(1) Within graphemic allomorphs geminate consonant clusters
(as in letter, add, and canned) are pronounced as single con-
sonants. Across morpheme boundaries, however, both graph-
emic consonants may correspond to separate phonemes, as in
midday and finally.
(2) All of the digraph and trigraph spellings are subject to the
same morpheme boundary problem as ph, e.g., hothead,
changeable.
(3) The spelling n, before spellings in the same morpheme which
correspond to /g/ or /k/ corresponds to /q/, as in congress,
finger, anchor. Across morpheme boundaries this generally
does not hold, e.g., ingrain, ingenious, ingratiate.
(4) Many word final clusters contain silent letters, e.g., gm, gn,
mb (paradigm, sign and bomb). Across certain morpheme
boundaries, the silent letter remains silent, as in paradigms,
signer, and bombing. As long as the morpheme boundary is re-
cognized, the correct pronunciation can be predicted. If the
morpheme boundary is not recognized, then the three forms
above would be thrown together with stigma, ignite, and
bamboo.
In some cases the discrimination of a morphemic spelling from
an identical, non-morphemic spelling is necessary for the predic-
tion of sound from spelling. Consider the following two word lists.
A Β
boys melodious
judges stylus
cats apropos
man's careless
The pronunciation of final s in any column A word can be pre-
dicted by the following rules (these rules must be applied in the
order shown here):
 PRELIMINARIES TO ORTHOGRAPHIC ANALYSIS 43

(1) jizj after /s, ζ, δ, j, s, z/.

(2) /z/ after any other voiced sound.
(3) /s/ in all other cases.
These rules, however, apply only to s when it is one of the following
morphemes:
(1) regular noun plural.
(2) third person singular, present indicative marker for the
verb.
(3) singular or plural possessive marker.
(4) any of the contractions like John's (from John is).
The past tense marker (e)d functions similarly, but is not entirely
regular. In all of these cases, nevertheless, the direct spelling-to-
sound approach fails unless it is based upon morpheme identity —
and if so, the approach is no longer a direct spelling-to-sound
approach.
Another area in which the direct correspondence approach fails
to recognize inherent patterning is in the treatment of the final
clusters gm and gn which were mentioned above. Consider the
forms autumn: autumnal, damn: damnation, paradigm: paradig-
matic, sign: signify. It is not sufficient to state that gn and gm in
final position correspond to jnf and /m/, while in medial position
to /-gn-/ and /-gm-/. Such rules fail in cases like autumns, de-
signing and signer. There is no way to avoid reference to mor-
phemes in this case, unless one simply enumerates the words for
each pronunciation. A regular pattern is present in these forms,
the most important aspect of which is the preservation of mor-
pheme identity. The alternations of /g/ and zero in these examples,
along with the alternations of the vowels preceding g, are pre-
dictable. The direct spelling-to-sound approach once again breaks
down when morpheme identity becomes important.

b. Form Class

The school-book approach to orthography, as exemplified by

Webster's New Collegiate Dictionary, recognizes ng as a spelling
for /g/ when this phoneme is not followed by /gf.
44 PRELIMINARIES TO ORTHOGRAPHIC ANALYSIS

The digraph ng, as in sing, singing, represents the voiced tongue-back

velar nasal continuant, corresponding to the voiced tongue-back stop
g, and the voiceless tongue-back stop k*
In contrast, the parallel cluster mb is analyzed phonotactically:
"b is usually silent after m in the same syllable, as in bomb, climb,
thumb, etc."7 That Webster's treats identical phenomena in con-
trastive ways is only one of the problems here. Another problem
is that an adequate description of the pronunciations of ng and
mb must be based upon both morphemic and phonotactical rela-
tions. The pronunciation of any form ending in nger or ngest can
not be predicted unless the morphemic identities of er and est are
known. If these are the comparative and superlative markers,
then ng is pronounced /rjg/ as in stronger; in most other cases the
/rjg/ cluster is leveled to /rj/, just as it is in word final position.
Morphemic identity is also important for predicting the pronun-
ciation of word final ate. In adjectives and nouns, this ending is
generally pronounced /-it/, e.g., duplicate, frigate, syndicate, while
in verbs, /-et/, e.g., deflate, duplicate, integrate.
A final example of where form class identity is necessary for
correct pronunciation is in initial th. Functors beginning with this
cluster have the voiced inter-dental spirant /δ/: the, then, this,
those, while contentives have the voiceless spirant /Θ/: theses, thin,
thumb*

4. PHONOTACTICAL INFLUENCES

a. Consonant Clusters

A knowledge of phoneme arrangements which are not allowed in

English words is a necessary prerequisite for analyzing many
spelling-to-sound correspondences. Sequences like /bp/ and /pb/
do not occur within English words — where they would occur,
as in subpoena and clapboard, the speaker drops one sound or the
β
Webster's New Collegiate Dictionary, 1953, p. xiv.
7
Ibid., p. xi.
8
Functors and contentives are defined by Hockett, 1958, pp. 264ff.
 PRELIMINARIES TO ORTHOGRAPHIC ANALYSIS 45

other (with /pb/ and /bp/, the first sound is always omitted). While
the spellings do not change, the pronunciations do. Yet, to label
the pronunciation of b in subpoena as irregular, just as one does
for the b in debt, is to ignore a pattern of English phonology. The
elision of sounds in consonant clusters can be predicted, not only
across morpheme boundaries, but also in initial and final positions,
as in knee, gnat, bomb, and sing. In all of these cases, the correct
pronunciation can be derived by first mapping all spelling units
onto a pre-phonemic level and then applying the rules for leveling
non-English clusters to obtain the phonemic forms. Thus, knee,
gnat, bomb, sing, become first {kni}, {gnaet}, {bamb}, {sigg}, and
then the non-allowed clusters are leveled, giving /ni/, /naet/, /bam/,
/sir)/·
b. Palatalization
To predict consonant cluster leveling is not the only reason for
observing the arrangements of phonemes in English words. The
palatalization of /sj, zj, tj, dj/ to /s, z, j/ and the deletion of
/jI from the cluster /ju/ also depend upon this knowledge. In ad-
dition, many spelling-to-sound patterns which can be described
only clumsily in direct spelling-to-sound terms are more adequately
described in phonological terms. A preceding /w/, for example,
tends to change /ae/ into /a/ when this vowel is not followed by
a velar consonant, e.g., swamp, assuage, quadrant, swan, quality,
quantum: wag, quack, twang, wax. To describe this process in
direct spelling-to-sound terms is difficult. The various spellings
which correspond to /w/ and to /k/, /g/, and /η/ must be enumerated,
and even if this is done, the phonological nature of the /ae/ —>/a/
shift is not revealed.

5. DESCRIPTIVE MODEL

Whatever system of rules is chosen to relate spelling to sound

must be not only accurate and as simple as possible, but also
revealing, in the sense that it allows a differentiation of the various
patterns in the system. To present the χ patterns, which depend
46 PRELIMINARIES TO ORTHOGRAPHIC ANALYSIS

upon a graphemic distinction and stress placement, as parallel to

the /w/ pattern discussed above is, as an example, an unsatisfac-
tory account of the current orthography.
In the spelling-to-sound model employed here, graphemic words
are divided into their graphemic allomorphs, and then these allo-
morphs are related to intermediate (morphophonemic) units by
an ordered set of rules. Other rules then relate the morphopho-
nemic units to phonemic forms. All rules which are based upon
non-graphemic features are applied in an ordered sequence on the
morphophonemic level, yielding various sub-levels of intermediate
forms for each word. The final morphophonemic form is then
mapped automatically onto the phonemic level. While the inter-
mediate level is not strictly a morphophonemic level, it will be
labeled as such hereafter. Its primary function is to separate
graphemically dependent rules from grammatically and phono-
logically dependent ones.
As examples of how this model organizes spelling-to-sound
rules, the processes for generating the pronunciations of social and
signing are shown below:
social would be mapped into {sosiael} by the grapheme-to-morpho-
phoneme rules for the separate units s, o, c, i, a, I (see Chapters V and
VII). On the first morphophonemic level the main word stress would
be placed on the first syllable, resulting in {sosiael}. Then through vowel
reduction {-iae-} would become {-ja-} and the resulting {-sj-} would be
palatalized to {-δ-}. The form {soSsl} would then be mapped onto the
phonemic level, giving /so§9l/.
signing would first be broken into sign and ing and then each of these
graphemic allomorphs would be mapped onto the morphophonemic
level, yielding {sign} and {ing}. Upon combination of the two forms
and the application of stress and phonotactical rules, the form {sfgnirjg}
would result. By the rules for leveling consonant clusters, final {qg}
would become {-η} and {-gn-} would become {-n-} with compensatory
alternation of {-I-} to {ai-}. These operations yield {sainirj} which is
automatically mapped into /sainirj/.9
* A more detailed discussion of this model is given in Ruth H. Weir, Formu-
lation of Grapheme-Phoneme Correspondence Rules to Aid in the Teaching of
Reading, Cooperative Research Project S-039, Final Report (Stanford, 1964);
Ruth H. Weir and Richard L. Venezky, Rules to A id in the Teaching of Reading,
Cooperative Research Project No. 2584, Final Report (Stanford, 1965).
 IV

THE GRAPHEMIC SYSTEM

1. SELECTION OF UNITS

a. The Grapheme-Phoneme Parallel

The description of spelling-to-sound relationships begins with
certain units on the spelling or graphic level and ends, for con-
venience, not with actual speech sounds, but with phonemes.
Traditional studies in this area have generally employed the term
grapheme as parallel to phoneme and various attempts have been
made in recent years to formalize graphemics or to elaborate on
the grapheme-phoneme parallel.1
Borrowing some ideas and methods from phonemics and morphemics,
for instance, we could conclude that each of the various letters has two
or more different shapes, which seem to be in complementary distribu-
tion or free variation. This in turn suggests that each different shape
can be called an allograph, and a family of allographs a grapheme ...
Taking inventory of the segmental graphemes of standard English
writing or printing, we find that there are thirty-seven of them, which
can be classified in two groups:
(a) Twenty-six letters of the alphabet <abc ... z )
(b) Eleven marks of punctuation ; : . ? ! - " ( ) —>
In addition we must include space, a sort of zero grapheme. 1
1
The earliest mention of the grapheme-phoneme parallel that I can find is
in F. de Saussure, Course in General Linguistics, trans. Wade Baskin (New
York, 1959), 23, 30. Saussure saw four features in the letters (he did not use
the term GRAPHEME) which paralleled features in the phonemic system. Pul-
gram upped the count of parallel features to nine ("Phoneme and Grapheme:
a Parallel", Word, 7 (1951), 15-20). For a summary of linguistic writings on
the grapheme-phoneme parallel, see John C. McLaughlin, A Graphemic-
Phonemic Study of a Middle English Manuscript (The Hague, 1963), 24-28.
2
W. N. Francis, Structure, p. 436.
48 THE GRAPHEMIC SYSTEM

Even with this description, the graphemic and phonemic systems

are far from parallel. While various single graphemes can be
mapped (through intermediate levels) into single phonemes, there
are also grapheme clusters that operate as single units, like th and
ch. Since one cannot deduce the behavior of these clusters from
the behavior of their constituents, it must be concluded that there
exist units higher than the grapheme on the graphic level, like ch
and th, which correspond more closely to the phoneme than does
the grapheme. An adequate description of English graphics should,
therefore, describe such units. To meet this need, Hall differen-
tiates between simple and compound graphemes.
The compound graphemes of English include a great many sequences
of vowel letters, graphic diphthongs such as ae, ai, au, ea, ei, eo, ... eu,
ie, oa, oi, ou, ue, in addition to double vowel letters like ee or oo. There
are also certain combinations of consonant letters which function as
single units and hence must be considered as compound graphemes:
e.g., ch, gh, ph, rh, sh, th, ng, and again, all the double consonant letters
such as bb, dd, etc.3
Even this division, however, does not create a workable parallel
between grapheme and phoneme. Certain graphemes function
solely as markers, that is, do not themselves enter into correspond-
ences, but rather, mark one of several possible correspondences
for other graphemes or grapheme clusters. Final e, for example,
performs at least six different marking functions (see pages 55-
58). It indicates vowel correspondences as in mate and cute (cf.
mat and cut), it marks consonant correspondences as in trace,
change, and bathe (cf. bath), and preserves graphotactical patterns
as in have, toe, and glue. The phonemic level has no parallel to
this.
One other major difference between graphemes and phonemes
is that phonemes are language dependent, functionally-defined
units, while graphemes are not necessarily language dependent
nor functionally defined. Almost all speech communities which
use the Roman alphabet have the same graphemic system, accord-
ing to current definitions of graphemics, yet their phonemic sys-
8
Robert A. Hall, Jr., Sound and Spelling in English (Philadelphia, 1961), 14.
 THE GRAPHEMIC SYSTEM 49

tems are vastly different. Furthermore, a new grapheme can be

added at will to the graphemic system, regardless of whether or
not it contrasts functionally with existing graphemes. The Anglo-
Norman scribes, for example, re-introduced q into English orthog-
raphy. Yet this grapheme performed the same function as k which
remained in the English writing system.4 The parallel situation
does not exist on the phonemic level; in fact, the opposite is true.
A new phoneme comes into existence only when a new contrast
appears.
The term 'grapheme' refers to letter classes. The individual
letters, as written, typed, or in any other way produced, are called
graphs, and each graph is classed under one of the twenty-six
graphemes in the current alphabet. The decision to label two
graphs as same or different depends in no way upon whether they
function as the same or different form. Both q and k, for example,
correspond only to /k/ (or are silent), yet k and q are not classed
as members of the same grapheme. The difference between these
two graphs, therefore, cannot depend upon the relationship of
spelling to sound or upon any other language dependent feature,
but only upon the graphic features of the two forms. This process
of classification is vastly different from the process by which two
phones are assigned to phonemes. Form alone is considered
in classing graphs into grapheme patterns, but both form and
function are essential for classing phones as members of phoneme
classes.5
b. Types of Graphemic Units

In the remainder of this report GRAPHEME will refer to one of the

4
Latin q occurred before ν as a spelling for /kw/ in some early Old English
manuscripts (see OED, s.v. q), but cw was a more common spelling for OE
/kw/. After the Conquest q was re-introduced in English orthography by Anglo-
Norman scribes and by the end of the thirteenth century qu had replaced cw
and kw as a spelling for /kw/.
6
The discrimination of graphemes has been studied by a number of psychol-
ogists in recent years and attempts have even been made to formulate distinc-
tive features for letters. See especially Eleanor J. Gibson, et al., "An Analysis
of Critical Features of Letters, Tested by a Confusion Matrix", Harry Levin,
ed. A Basic Research Program on Reading (Cornell, 1963).
50 THE GRAPHEMIC SYSTEM

alphabetic characters a through z, and the term GRAPHEMIC LEVEL

will be used as a general reference for the spelling level.6 The
actual language-dependent units on the graphemic level which are
significant for the prediction of sound are called FUNCTIONAL
UNITS and are divided into two classes: RELATIONAL UNITS and
MARKERS.
A RELATIONAL UNIT is a string of one or more graphemes which
has a MORPHOPHONEMIC CORRESPONDENT which cannot be pre-
dicted from the behavior of the unit's smaller graphemic com-
ponents.
A MARKER is a cluster of one or more graphemes whose primary
function is to indicate the correspondences of relational units or to
preserve a graphotactical or morphological pattern. It is mapped
into a zero morphophoneme.

c. Relational Units

The same grapheme or grapheme cluster can be part of more than

one relational unit. Thus, gn in cognac and poignant is a single
relational unit which corresponds to the morphophonemic cluster
{nj}, but gn in sign and malign is not a relational unit, but rather
a combination of two relational units which correspond to the
morphophonemes {g} and {n}. Morphophonemic rules map {g}
into either {0} or into {g}, depending upon allomorphic consid-
erations (cf. signal and malignant). The selection of relational
units is based upon function and composition. Any string of
graphemes that corresponds to a non-zero morphophoneme is a
potential relational unit. However, only those strings whose mor-

Β GRAPHEME is not an ideal term, either, for the discussion of spelling-to-sound

correspondences, but I find the only plausible substitute, LETTER, more objec-
tionable. The case against LETTER was summarized by Angus Mcintosh, "The
Analysis of Written Middle English", Trans. Phil. Soc. (1956), p. 43. "The
word 'letter' in everyday use is ambiguous. It may be employed in the sense
of 'grapheme' as when we say that the word swilk begins with 'the letter $';
at other times it is used of the particular allographic form of a grapheme may
have in a given context; e.g., when we speak of 'the s used in final position in
Greek'; again, it may be used of a single instance of an allograph, as when we
say 'that's a badly formed letter' ...."
 THE GRAPHEMIC SYSTEM 51

phophonemic correspondences cannot be predicted by general

rules based upon smaller units contained in the string are classed
as relational units, ch in chair, for example, is a relational unit
since the morphophoneme {δ} cannot be predicted from general
rules based upon c and h separately. Geminate consonant clusters,
however, are not single relational units since their morphopho-
nemic forms can be obtained from rules based upon their separate
constituents. (The leveling of clusters like {ff} to {f} can be ac-
counted for by a general phonotactical rule).
Consonant relational units, furthermore, are classed as function-
ally simple or functionally compound.7 This distinction is needed
for an accurate statement of a number of rules. In the sequence
VOWEL + CONSONANT + FINAL e, for example, VOWEL is generally
mapped into its free alternate if CONSONANT is a functionally simple
unit (or this type of unit plus I or r), and into its checked alternate
if CONSONANT is a functionally compound unit, or a cluster.8 Thus,
Free alternate Checked alternate
btfke tfxe
ache badge
concede edge
clothe hodge
Vowel units are classed as primary {a, e, i, y, ο, Μ), or secondary
(all others).
Examples of markers are the final e in mate and peace, the u in
guest and the k in trafficking, all of which indicate the pronuncia-
tion of a preceding grapheme, ue in plague, catalogue, etc., is a
sequence of two markers. The u after g marks the correspondence
g {g} rather than g -> {]}. Since u generally does not appear
in final position in English words (see page 62 for exceptions), a
final e is added, as in continue and blue (cf. the alternations ou/ow,
aujaw, pages 60-61). Only graphemes mapped into zero can

7
The functionally compound units are tch, dg, ck, wh, and x. The remaining
consonant units (including sh, th, and ch) are functionally simple.
8
Free and checked vowels are described on page 14 and in Hans Kurath,
A Phonology and Prosody of Modern English (Ann Arbor, 1964), 17-20.
52 THE GRAPHEMIC SYSTEM

be classed as markers (this is a necessary, but not sufficient con-

dition). Graphemes with non-zero morphophonemic correspond-
ences, however, though properly classed as relational units, can
also perform marking functions. The i in city, for example, besides
corresponding to {i}, marks the correspondence c -*• {s}. A gemi-
nate consonant cluster also performs a marking function since it
regularly indicates the correspondence of the preceding vowel.
The strongest evidence for a separate class of markers in English
orthography is found in orthographic alternation patterns. For
example, final e as a marker for the pronunciation of a preceding
c or g is dropped before a suffix which begins with a letter that
will perform the same function as the e. Therefore, notice drops
the final e before ing (noticing) since i also marks the correspond-
ence c -> {s}, but retains the e before able since noticable would
have c {k}. Similarly, the e added to an otherwise terminal u
is dropped before any suffix since the only function of the e is
to avoid having word-final u, e.g., argue, arguing.
Relational units are classed as consonants or vowels depending
upon the class of the morphophonemes into which they are mapped
(glides are classed as consonants). Some RELATIONAL UNITS are
classed as both consonants and vowels, e.g., u in language (conso-
nant) and during (vowel). Within these classes major and minor
patterns are distinguished on the basis of frequency of occurrence.
Thus, ch is classed as a major consonant unit, but kh {khaki) is
classed as a minor unit. While the major-minor classification
may appear arbitrary, it distinguishes frequently occurring, pro-
ductive patterns from infrequent patterns which generally occur
in only a small number of borrowings.
In the traditional treatment of spelling, forms like gu (guard),
ti (nation), di (soldier), and a ... e (bake) are adopted as basic
spelling units.9 Although such a course has not been adopted
here, it is interesting and informative, nevertheless, to extend this
treatment to some of its logical conclusions. In the units just
mentioned, so-called silent letters are always included as parts of
other spelling units, regardless of the functions which they per-
9
See especially Hall, 1961, pp. 17-22.
 THE GRAPHEMIC SYSTEM 53

form. Therefore, the b in debt and the final e in peace would fall
into the same basket, in that they are both silent — the units, pre-
sumably, being bt and ce. This treatment, unfortunately, obscures
a basic difference between these two silent letters, namely, that
the e at the end of peace marks the correspondence c {s} while
the b in debt is a functionless scribal insertion. The e at the end
of peace is part of the general pattern of markers in English or-
thography; the b in debt, on the other hand, is an unproductive,
isolated occurrence.
Consider, furthermore, the traditional enumeration of spelling
units as reflected in such words as clothe and pace. Both words
have, by the traditional treatment, a silent e which must be at-
tached to another grapheme, yet it is not clear which unit either
e should be attached to, since in each word it marks two separate
patterns. In clothe e marks the correspondences ο -* {ο} and
th -»· {ö}; in pace it marks a -*• {e} and c -> {s}. The traditionalist
is faced with a dilemma here; are the units ο ... e/a ... e or the I eel
Or shall we take a fine razor and split e into two parts so that both
alternatives can be taken? Without treating e as a member of a
class of markers that are separate from the relational units, no
realistic solution is possible.
The last of the major problems inherent in the traditional view
is that an unnecessarily large number of basic units are created
by the merger of markers with relational units. Besides classing
all of the geminate consonant clusters as basic units, the following,
among others, must also be recognized as basic:
the breathe ce trace
dge edge lie belle
le able gne cologne
re acre ffe giraffe
ye bye aoh pharaoh
ge change ah pariah
xe axe eh vehement
ve love
All of the units above must be classed as basic units by the tradi-
54 THE GRAPHEMIC SYSTEM

tional view, yet their behavior can be predicted from the behavior
of their components. The final e in the clusters shown above is a
marker whose various functions are explained on pages 56-58. Gemi-
nate clusters (belle, giraffe) can be handled by considering each
consonant separately and leveling the resulting geminate morpho-
phonemic clusters by a single rule, h in pharaoh and pariah is a
marker (see page 56), while in vehement it is deleted by a phono-
tactic rule (cf. vehicle; vehicular).

d. Enumeration of Relation Units

The consonant and vowel relational units are shown below. Units
followed by an asterisk are discussed in the following paragraphs.
Major Relational Units
Consonants Vowels
simple compound primary secondary
b gh η s w* ck* a ai/ay ie
y*
c h Ρ sh dg e au/aw oa
ch j ph t Ζ tch i ea oe
d k q th wh ο ee oi/oy
f 1 r u* X u ei/ey oo
g m rh ν y eu/ew ou/ow
Minor Relational Units
Consonants Vowels
simple compound secondary
kh gn ae
sch eau
eo
uy
u is a consonant unit when it corresponds to {w} as in quack,
language and assuage. It may also be a vowel unit, or part of a
vowel unit (ou), or a marker (guest, plague).
w is a consonant unit when it corresponds to {w}, e.g., warm,
beware. It also appears as part of a vowel unit (ow, aw) but never
as a vowel unit by itself.
 THE GRAPHEMIC SYSTEM 55

j is a consonant unit when it corresponds to {j}, e.g., yes, beyond.

It also appears as a vowel unit, and as part of a vowel unit, e.g.,
cycle, boy.
ck is a consonant unit in words like rack and tack. In instances
like picnicking, however, the k is a marker. That ck in picnicking
is identical to the relational unit ck is immaterial, since the base
form picnic ends in c, not ck.

e. Examples of Minor Units

kh khaki, khan eau bureau, plateau

sch schist, schwa eo jeopardy, leopard
gn cognac, poignant uy buy, guy
ae aesthetic, algae

2. MARKERS

a. Introduction

The primary functions of markers, as mentioned above, are to

indicate the correspondences of other graphemes and to preserve
graphotactical or morphological patterns. For example, ν does
not occur in final position in English; where it would, a final e has
been added, e.g., have, love. Various relational units also serve as
markers, but since they have non-zero morphophonemic corre-
spondents, they are classed as relational units rather than as mark-
ers. In the discussion below, markers are classed according to
the functions they perform.

b. Markers of Vowel Correspondences

(1) In the word-final patterns VCe and VCle, e generally indi-

cates the free pronunciation of V.10 Thus, mate : mat, mete : met,
site : sit, note : not, cute : cut.
10
See Weir, 1964, pp. 42-45.
56 THE GRAPHEMIC SYSTEM

(2) h. Post vocalic h before a consonant or juncture is classed

as a marker, even though no clear marking function can be deter-
mined from the few existing occurrences of this form: ah, hurrah,
oh, pharaoh, pariah, hallelujah.n (The alternative is to class ah,
oh, and aoh as relational units). In prohibition, vehement, etc., h
is mapped into {h} and then deleted by a general morphophonemic
rule.

c. Markers of Consonant Correspondences

(1) e. In final position after c, e marks the correspondence

c -*• {s}. In adding suffixes which do not begin with i, y, or e, the
e is retained as a marker, e.g., trace : traceable : tracing.1*
(2) e. In final position after consonant + I or consonant + r,
e marks a syllabic / or r. Thus, able, table, acre, massacre. Con-
sonant -f re spellings are rare in American English, occurring
regularly only in the patterns ere and gre where the more normal
cer and ger would indicate the wrong pronunciations for c and g.
(3) e. In final position after g, e marks the correspondence
g ->• {]}, e.g., age, college. After dg, e is a graphotactical marker,
since dg does not occur in final position. In adding suffixes which
do not begin with the i, y, or e, the e is retained as a marker, e.g.,
change : changeable : changing. In final position after th, e marks
the correspondence th {ö}. Thus, bath : bathe, breath : breathe,
teeth : teethe, cloth : clothe, wreath : wreathe.
(4) u. After g, u is occasionally used to mark the correspondence
g -»· {g} (cf. the use of e to mark g -»• {j}, above). For example,
guess, guide, guest, guise, catalogue, guilt, plague, guild.
(5) k. A k is inserted between a final c and a suffix beginning
with i, y, or e, to mark the correspondence c {k}. Thus, picnic :
picnicking, traffic : trafficking, panic : panicky, shellac : shellacked.13
11
Ohm and other German borrowings are not considered here.
18
e simultaneously performs two marking functions in trace·, it marks the
correspondences a -> {e} and c -> {s}.
13
But, irregularly, arc.arcing, disc:discing:disced. Note, however, the ten-
dency to respell disc as disk. In lyricism (cf. lyric, lyrical) the phonetic change
in Latin of [k] [s] obviated the need for the k marker.
 THE GRAPHEMIC SYSTEM 57

(6) GEMINATION. A geminate consonant cluster regularly marks

a checked correspondent for a preceding vowel. Compare, for
example, anal: annals, fetal: fetter, hypo : hippo, rotor : rotter
(<blackguard), super: supper.

d. Markers of Graphotactical Patterns

(1) e. After what would otherwise be a final ν or u, an e is

added. This practice developed during the late Middle English
period, due partially to the graphical identity of u and v. Thus,
love, glue, have, plague, glove, continue.
(2) e. To limit the two-letter words in English to a select group
of common words, a final e has been added after a vowel in what
would otherwise have been a two-letter word. Thus, see, doe, toe,
rye, etc. On the addition of e after o, Webster's 1852 Dictionary
had the following explanation:
Woe. — This word takes the final e, like doe, foe, hoe, sloe, toe, and all
similar nouns of one syllable. The termination in ο belongs among
monosyllables, to the other parts of speech, as potato, tomato, etc.14
(3) al. The cluster al is inserted between a final adjectival ic
and the adverbial ly if an adjective in ical does not exist. Thus,
basic : basically, rustic : rustically, scenic : scenically. Generally,
the al is mapped into zero in such forms. Spelling pronunciations,
however, have introduced forms in which the al is pronounced.

e. Markers of Morphemic Patterns

After a number of words which would otherwise end in non-

morphemic s, an e has been added to avoid the appearance of a
final, inflectional s. Thus, moose, goose, mouse. After the cluster
rs, e also marks the correspondence s -*• {s} in contrast to s -* {z}
14
Webster's American Dictionary of the English Language, ed. Chauncey
A. Goodrich (2d ed. rev.; Springfield, Mass., 1852), p. 81. Two-letter words
ending in a consonant were lengthened, not by adding a final e, but rather, by
doubling the final consonant; thus, ebb, add, odd, egg, inn. (These are the only
examples of final bb, dd, gg, nn in Modern English spelling).
58 THE GRAPHEMIC SYSTEM

(cf. hearse : hers). Non-morphemic s is generally avoided in final

position in nouns, adjectives (by analogy to nouns), and verbs
under the following conditions:
(a) when preceded by a consonant
collapse dense
eclipse sense
else hearse
false coarse
(b) when preceded by a compound vowel
goose praise
moose raise
house noise
mouse poise
The only true exceptions to these restrictions are summons (n.,
sg.), lens, and words ending in ous. Summons, an Anglo-French
adoption, had variant spellings in nee and nse through the seven-
teenth century (see OED, .s.v.), but apparently through analogy
(both spelling and pronunciation) with the identical form, sum-
mons became the standard spelling. Lens is a late seventeenth
century Latin adoption for which no variant spellings in se were
found in the data used for this study. If, however, this word were
a phonological rather than a spelling adoption, then lense would
give the incorrect pronunciation /lens/. (This is also true of sum-
monse). The suffix ous, if spelled ouse would be pronounced /ausj
as in mouse and house. Apparent exceptions to (1) and (2) above
are words like hydraulics, mathematics, mumps, and billiards,
which function as both singulars and plurals. In these, however,
the s can be classed as a morphemic unit.
The only situations where the e marker is not added is where
s is preceded by a simple vowel spelling, but here e cannot be
added because it would tend to mark an incorrect pronunciation
for the simple vowels. Compare us : use, his : rise, locus : recluse,
tennis: improvise.
 THE GRAPHEMIC SYSTEM 59

3. GRAPHEMIC ALTERNATIONS

a. i and y

Vocalic i and y (which can be considered as members of the same

functional unit) occur in complementary distribution: y in final
position and i in initial and medial position. There are, however,
numerous exceptions to this pattern. A considerable number of
Greek and Latin borrowings retain y in medial position, e.g.,
embryo, asylum, rhythm. In addition, some medial y's have re-
sulted from inorganic final e's, e.g., rye and bye, and from scribal
pedantry, as in rhyme which was converted from the earlier rime
on the mistaken analogy with L. rhythm. Initial y (vowel) occurs
in the forms ycleped, yttrium, and ytterbium, while final i occurs in
several distinct patterns. One is the plural of Latin borrowings
whose singulars end in -us, e.g., alumni, fungi, magi, and stimuli
(alibi and quasi are singular in Latin). Similarly, it occurs as the
plural of Italian borrowings whose singulars (used in Italian only)
end in ο, e.g., broccoli (It. sg. broccolo), confetti (It. sg. confetto),
ravioli (It. sg. raviolo), and spaghetti (It. sg. spaghetto). Macaroni,
another Italian borrowing, is singular in Italian. Of the remaining
words ending in i, alkali, charivari, houri, and potpourri are French,
anti is Greek, ski is Norwegian, chilli is Mexican, khaki is Urdu,
mufti is Arabic, yogi is Hindustani, and taxi is an early twentieth
century American coining. Many of these words had alternate
final spellings in y or ee, but apparently a desire to retain the
foreign appearance favored the final i spellings.
The alternation of i and y occurs with suffixation in the following
cases (for exceptions, see Webster's New Collegiate Dictionary,
1956, pp. 1145-47):
(1) When ing is added to words ending in ie, e is dropped and
i is changed to y, e.g., die : dying, tie : tying.
(2) When a suffix not beginning with i is added to a base ending
in y, y changes to i, e.g., icy: iciest, mercy : merciless.
(3) A few words which end in vowel -+- y irregularly change y
to / before some suffixes, e.g., day : daily, lay : laid.
60 THE GRAPHEMIC SYSTEM

i and y also alternate as parts of the compound vowel units ai (ay)

ei (ey), and oi (oy), the y alternate appearing before vowels and
word juncture and the i alternate appearing in all other cases.
Exceptions to this pattern are the following:
geyser An Icelandic borrowing, taken from the name of a hot
spring in Iceland.
moiety A ME adoption of OF moite, moitie. The earlier form
moitie was replaced by moiety in the last half of the
seventeenth century. The middle vowel in the pronun-
ciation is probably excrescent.
oyster A ME adoption of OF oistre, uistre. Judging from the
OED entries for this word, oy was the only spelling
commonly employed.
paranoia A late nineteenth century Latin adoption. The oia
spelling was probably retained because of its similiarity
to other medical terms ending in ia, e.g., megalomania,
malaria, anemia, dementia and hysteria,
sequoia A Latinization (in botanical terms generally) from Se-
quoiah, a Cherokee Indian name.

b. u and w

As consonant spellings, u and w are, for the most part, in comple-

mentary distribution, u appears as a spelling for w in the environ-
ments gu, su, qu (initial and medial), ju, nu (all medial) pu and cu
(initial), w occurs in all other cases, except for the three forms
suave, suede, and suite. Following is an exhaustive listing of oc-
currences of these spellings in the corpus used for this study.
gu su ju
anguish assuage marijuana
distinguish persuade
extinguish pu nu
jaguar pueblo ennui
language
languid
 THE GRAPHEMIC SYSTEM 61

CU

lingual cuisine
linguist
penguin
sanguine
unguent
u and w also alternate as the second parts of the compound vowel
spellings which begin with a, e, and o. The w variant generally
appears before another vowel spelling and in morpheme final
positions, while the u variant occurs in all other positions. This
pattern, however, is far from regular.

c. Examples of Regular Spellings

au aw eu ew ou ow
auction awe eucharist ewe ounce coward
audit draw eulogy ewer oust however
augur gnaw feud brewer out power
applaud saw neuter shrewe compound allow
fault straw pneumatic flew south cow
trauma thaw rheumatism threw trousers vow

d. Exceptions
(a) aw in initial and medial position before a consonant
awkward dawn
awl scrawl
awning spawn
(b) ew in medial position before a consonant
newt lewd
pewter shrewd
(c) ou in final position
bayou bijou
caribou thou
you
62 THE GRAPHEMIC SYSTEM

(d) ow in initial and medial position before a consonant

owl drowse
own fowl
crowd town
u in final position in English orthography is rare. Besides the
examples given above, the only other examples are the following:
beau adieu flu
bureau lieu gnu
chateau impromptu
plateau menu
portmanteau parvenu
trousseau tabu
virtu

e. ous and ou

The suffix ous becomes os before ity. Thus, viscous: viscosity,

curious : curiosity, generous : generosity, porous: porosity, mon-
strous : monstrosity.

f. i and e

In the suffix ity, i becomes e when preceded by i. Thus society:

propriety, piety : satiety, contrariety : anxiety, variety : sobriety.
 ν

CONSONANT DISTRIBUTIONS,
CORRESPONDENCES, A N D ALTERNATIONS

1. INTRODUCTION

The descriptions which follow of the consonant units include brief

discussions of these topics:
DISTRIBUTION. The intra-word positions the unit commonly
occupies, both by itself (i.e., contiguous with vowels or word
boundaries only) and in combination with other consonant units
(i.e., in consonant clusters).1 Since only monomorphemic clusters
are considered, such clusters as -xts as in texts are not included
in this discussion.
CORRESPONDENCES. The spelling-to-morphophoneme correspond-
ences for the consonant unit, presented as a set of ordered rules.
ALTERNATIONS. The productive morphophonemic and phonotac-
tical alternations which are based upon the particular consonant
unit, presented as a set of ordered rules. Non-productive alter-
nations are listed with examples, but are not included in the
ordered rules.2
1
Since spelling units rather than letters of the alphabet form the basis of
this discussion, the term cluster could be ambiguous, tch, for example, is a
single spelling unit, yet is classed as a cluster in most treatments of the orthog-
raphy. In this paper cluster will refer to a sequence of functional units only.
The statement that a particular spelling unit appears in certain positions in a
word (initial, medial, final) precludes occurrences of that unit in clusters.
Clusters are always treated separately and, except for rare examples, only
initial and final clusters which occur wholly within single morphemes are dis-
cussed. (With a broad enough definition of WORD, any cluster could occur in
medial position).
2
The classes PRODUCTIVE and NON-PRODUCTIVE are used here to separate
patterns which we would tend to include in a description of the morphophon-
64 CONSONANT DISTRIBUTIONS

2. THE CORRESPONDENCES

a. Distribution

Initial and medial b account for almost 95 percent of all occur-

rences of this unit; final b is rare and final bb is rarer, occurring
only in ebb, derived from OE ebba. In initial and final clusters
b occurs chiefly with r and /, forming the initial clusters, br- and
bl- and the final clusters, -rb and -lb, e.g., brake, black, herb, curb,
bulb. Final -lb occurs only in bulb while final -rb occurs in about
fifteen different words.3 Two other final clusters, -bt and -mb
occur, although they represent different patterns from the -rb and
-lb clusters (see below).

b. Correspondences

(1) b in debt, doubt, and subtle corresponds to {o}. 4

(2) b elsewhere corresponds to {b}.

c. Alternations

(1) Initial {b} in {bd-} and morpheme final {b} in {-mb} are

emics and phonotactics of English from those we would not. In certain cases
this classification is obvious, in others, quite arbitrary. The palatalization of
{tj} to {c}, for example, is an active process in Modern English and therefore
is a productive pattern; the alternation of {b} with zero in number:numeric
is an isolated pattern, resulting from borrowing the same word from two
different languages (Latin and French). But whether the alternation of {i}
and {ai} in sign: signal is productive or non-productive is difficult to determine
(it is classed as non-productive here).
3
Unless otherwise qualified, WORD implies the base form only and not the
derivational and inflectional forms derived from the base. The separation of
base forms from derived forms is not discussed in this paper. For a recent
discussion on this topic, see Hans Marchand, The Categories and Types of
Present-Day English Word-Formation (Wiesbaden, 1960).
4
There is a variant pronunciation of subtile which has no /b/. This could
lead to a {b}—{0} alternation between subtile ({sitel}) and subtility ({ssb-
tiliti}).
 CONSONANT DISTRIBUTIONS 65

dropped. 5 {Bombard is classed as a non-productive alterna-

tion with bomb; see below).
(2) In the medial combination bilabial stop + bilabial stop, the
first stop is dropped. Cf., for example, cupboard, subpoena,
clapboard, raspberry.
(3) Non-productive alternations:
{b> - {0}
bombard bomb
crumble crumb
debit debt
iambic iamb
indubitable doubt
number numeric
rhombus rhomb
thimble thumb

a. Distribution

c occurs primarily in initial and medial position; in final position

it occurs in the ending -ic and in the borrowings arc, havoc, sac,
shellac, sumac, talc, and zodiac. The low frequency of final c is
probably due to the problems created by derivational and inflec-
tional suffixes which begin with the vowels e, i, y. In such cases,
c would be pronounced /s/ unless a k were added between c
and the first vowel of the suffix. This is generally what is done
after -ic, cf. traffic : trafficking, picnic : picnicked, but, irregularly,
arc: arcing. In lyric : lyricism /k/ changes to /s/ so the spelling
does not have to be altered, c also occurs in the initial clusters
cl~ {clean, clover), cr- {crawl, crust), cz- {czar), and scr- {scream,
screw), and in the final cluster -ct {abduct, sect).

6
Hypercorrect pronunciations of iamb and rhomb have a final (-mb) cluster.
66 CONSONANT DISTRIBUTIONS

b. Correspondences
(1) c in cello and concerto corresponds to {δ}.
(2) c in victual, czar, and indict corresponds to {0}.
(3) c before the spellings i, y, e (and in facade) corresponds to
{s} (except in sceptic, an alternate spelling of skeptic, where
it corresponds to {k}).
(4) c in all other positions corresponds to {k}.

c. Alternations
(1) {sj} before an unstressed vowel and in sure and sugar corre-
sponds to {s}.6
(2) Non-productive alternations
{k} - {0}
corpuscular corpuscle
muscular muscle

{k}-{s}
lyric lyricism
music musician

ch

a. Distribution
ch occurs frequently in initial and medial position and not so fre-
quently in final position. It rarely occurs in clusters, the most
common being chl-, ehr-, and -rch, e.g., chlorine, chrome, arch.
Much rarer are -cht and -chm, e.g., yacht, drachm.

b. Correspondences

(1) ch in fuchsia, yacht, and drachm corresponds to {0}.

* This process (palatalization) is discussed in Chapter VI.

 CONSONANT DISTRIBUTIONS 67

(2) ch corresponds to {k} before /, n, r, and in the following (plus

their derivatives):

ache chasm hypochondria

alchemy chemi- lichen
anarchy chiropodist machination
anchor choir mechanic
archaeology choler melancholy
archaic cholera monarchy
archangel choral orchestra
archetype chord orchid
architect chorus parochial
architecture dichotomy patriarch
archive echo pentateuch
catechism epoch psych-
chameleon eucharist stomach
chaos eunuch
character hierarchy

(3) ch corresponds to {s} in the following words:

cache chassis chivalry

chagrin chateau chute
chaise chauffeur cliche
chalet chauvinism crochet
challis chef echelon
chamois chemise machine
champagne cheroot mustache
chandelier chevalier parachute
chanty chevior ricochet
chaperon chevron sachet
charade chic stanchion
charivari chicanery
charlatan chiffon

(4) In all other cases, ch corresponds to {c}.

68 CONSONANT DISTRIBUTIONS

c. Alternations

None. (On {k} — {ö}, see Alternations under k).

a. Distributions

d occurs commonly in initial, medial, and final position and in

the clusters dr-: draw, dregs; dw-: dwell, dwindle·, -nd: land, find',
-Id: weld, bold; -rd: beard, board. It also occurs in the cluster
-dz in adz.

b. Correspondences

d regularly corresponds to {d}. (The identity of the past tense

and participle marker {d} must be retained at the morphophonemic
level so that the alternation of {d} and {t} can be handled correctly).

c. Alternations

(1) After all unvoiced final consonants except {t} the past tense
and participle marker {d} becomes {t}.7
(2) {d} corresponds to {0} between {n} and another consonant,
as in handkerchief. Modern pronunciation is too unsettled,
however, for the application of this rule to be more definitely
delimited. It applies generally to hands, pounds, stands,
handful, grandmother, but less frequently to handstroke, and
landslide (see Gimson, p. 232).

7
For a list of exceptions to this rule, see Albert H. Marckwardt, "Origin
and extension of the voiceless preterit and the past participle inflections of the
English irregular weak verb conjugation", University of Michigan Publications:
Essays and Studies in English and Comparative Literature, Vol. XII, pp. 151-
328 (Ann Arbor, 1935).
 CONSONANT DISTRIBUTIONS 69

defend defense
expend expense
respond response
{d} - {0}
sound sonant
spindle spin

a. Distribution

f which alternates in some environments with ph, occurs in initial,

medial and final position. It also occurs in the following common
clusters: fl-, fr-, -If -rf -ft, e.g., flood, free, half, serf raft.

b. Correspondences

(1) / i n of corresponds to {v}.

(2) In all other cases, / corresponds to {f}.

c. Alternations

See also Chapter VI.

(1) In the following words morpheme final {f} becomes {v} with
the addition of the regular plural morpheme: calf, elf half,
knife, leaf, life, loaf, self, sheaf, shelf, staff, thief, wife, wolf
(2) In the following words, morpheme-final {f} becomes {v}
when the noun becomes a verb: 8 (cf. house (n): house (v),
breath: breathe).
8
There is, however, an equally long list of words in which final {f} in the
noun does not alternate with a final {v} in the verb: brief, dwarf, leaf, sniff,
spoof, stuff, whiff.
70 CONSONANT DISTRIBUTIONS

belief believe
grief grieve
half halve
life live
proof prove
relief relieve
shelf shelve
thief thieve
(3) {f} in all other cases remains {f}.
(4) Non-productive alternations:
{f} - {v}
bereft bereave
cleft cleave
drift drive
fifth five
gift give
left leave
serf serve
twelfth twelve

a. Distribution

g occurs predominantly in initial and medial position in Modern

English words. A small number of words, almost all monosyllables,
have final g, e.g., bag, bug, frog, hog, twig, g also occurs in the
initial clusters gl-, gn, gr-, e.g., glide, gnat, grow, and in the final
clusters -gm and -gn, e.g., paradigm, sign. (Because of its highly
irregular correspondences, gg is treated separately).

b. Correspondences

(1) g corresponds to {0} in seraglio, intaglio, and imbroglio.

 CONSONANT DISTRIBUTIONS 71

(2) g corresponds to {z} in the following words (some of which

have alternate pronunciations {]}):
bourgeois mirage
camouflage prestige
corsage protege
garage regime
lingerie rouge
loge sabotage
massage
(3) g in gaol and margarine corresponds to {j}.
(4) g corresponds to {g} before e, i, y in the following:
anger geyser girth
auger giddy give
begin gift gizzard
bogey gig gynaecology
conger giggle hunger
eager gild lager
finger gilt linger
fogey gill malinger
fungi gimlet monger
gear gird target
geese girt tiger
geld girdle yogi
get girl
(5) g before et i, and y corresponds to {j}.
(6) g otherwise corresponds to {g}.9

c. Alternations

(1) Morpheme final {-qg} becomes {rj} except before the com-
parative and superlative morphemes, where it remains as
{-gg}. Thus (stroi)} : {strorjgar}, but {sain} : {safnar}.10
(2) Word initial {gn-} becomes {n} (cf. gnostic and agnostic).
β
For other exceptions to rules 4 and 5, see Weir, 1964, p. 35.
10
Besides strong, the only other eligible forms are long, wrong, and young.
However, some suffixed forms of long and diphthong, retain {qg} — elongate,
elongation, diphthongize, prolongate, prolongation. Note also ting: tingle.
72 CONSONANT DISTRIBUTIONS

(3) Morpheme final {-gn} and {-gm} become {n} and {m}. Thus,
sign : {sain}, but signal: {signal}, paradigm : {paeradaim}, but
paradigmatic : {paeradigmsetik}.
(4) {g} in all other cases corresponds to {g}.

gg

a. Distribution

gg has a limited distribution in English words. It occurs in final

position in one word: egg. In medial position it occurs predomi-
nately before final -y or -er, e.g., baggy, dagger. It occurs before
a in the endings -ar (e.g., beggar), -art (e.g., braggart), and -ard
(e.g., loggard), and in baggage, luggage and toboggan. Maggot is
the only example found with gg before o. In addition, gg occurs
in the medial clusters ggl and ggr, e.g., giggle, aggrandize.

b. Correspondences

(1) gg in exaggerate corresponds to {)}.

(2) gg in suggest corresponds to {gj}.
(3) gg otherwise corresponds to {gg}.

c. Alternations

None.

gh

a. Distribution

gh is uncommon in English orthography, occurring mostly in

initial and final position and in the cluster -ght, as in night, might,
nought and fought.
 CONSONANT DISTRIBUTIONS 73

b. Correspondences

(1) gh in hough corresponds to {k}.

(2) gh corresponds to {f} in the following:
clough rough
cough slough
draught sough
enough tough
laugh(ter) trough
(3) Initial gh corresponds regularly to {g}, e.g., ghastly, gherkin,
ghetto, ghost and ghoul.
(4) gh corresponds to {g} in the following words.
aghast sorghum
burgh(er) spaghetti11
dinghy
(5) In all other cases, gh corresponds to {0}.

c. Alternations

None.

a. Distribution

The consonant spelling h occurs in both initial and medial posi-

tions; in some medial positions and in all final positions it occurs
only as a marker after a vowel. "It is intended to suggest stressed
free vowels, as in the exclamations ah!, oh!, hurrah!, bah!, and in
shah and Yahweh".12 The OED suggests that besides marking a
11
In gherkin, ghetto, burgher, dinghy, and spaghetti, h could be treated as a
marker for g {g}. Gherkin, burgher, and dinghy have earlier spellings with-
out h (gerkin, burger, dingy); ghetto and spaghetti were borrowed from Italian
where the gh spelling was common. (In ghost and ghoul, an earlier g spelling
was changed in the sixteenth century to gh, apparently through the influence
of Caxton).
12
Kurath, 1965, p. 67.
74 CONSONANT DISTRIBUTIONS

'long' (free) vowel, h also marks non-native words. "After a vowel

h is regularly silent, and such a vowel being usually long. ... the
addition of Λ ... is one of the expedients which we have for indi-
cating a long vowel in foreign or dialect words.13

b. Correspondences

(1) Initial h in heir, herb, honest, honor, hour corresponds to {0}.

(2) Medial h before a consonant, and final h are markers for the
preceding vowel and therefore correspond to {0}, e.g.,fahren-
heit, hurrah, pharaoh.
(3) h in all other positions corresponds to {h}, e.g., exhale, har-
monic, vehicle.

' c. Alternations

(1) {h} preceded by a consonant corresponds to {0} in most

English dialects, cf., e.g., philharmonic : harmonic.
(2) Intervocalic {h}, when preceded by a stressed vowel, corre-
sponds to {0}, cf., e.g., vehicle: vehicular.
(3) In all other cases {h} corresponds to {h}.

a. Distribution

The letter j appears primarily in initial positions before back

vowel spellings, but occasionally in medial position before either
back or front vowel spellings, as in cajole, deject, majesty, rajah.
In a number of cases where j appears before a front-vowel spelling,
a variant spelling with g exists, e.g., jest: gest; jingle : gingle;
Serjeant'.sergeant (now differentiated in meaning). The Hebrew
loan word hallelujah has the variant (h)alleluia, while the American-

13
O E D , ί.ν. h.
 CONSONANT DISTRIBUTIONS 75

Spanish marijuana has the variant marihuana, j does not appear in

final position or in consonant clusters.

b. Correspondences

j regularly corresponds to {j}, except in bijou ({z}), hallelujah,

({j}), and marijuana ({0}).

c. Alternations

None.

a. Distribution

k occurs in initial, medial and final position, but is limited to

certain environments in each of these positions. In initial position
it appears most commonly before e and i (both alone and in the
cluster sk-) and in the cluster kn-. It does occur, however, in
several recent borrowings before α, ο, and u, e.g., kaffir, kangaroo,
kosher, kulak. In final position it appears commonly after the
vowel spellings ea, ee, oa, and 00 as in seek, soak, and took, and
in the clusters -Ik, -nk, -rk and -sk, e.g., milk, rank, hark, and ask.
Since kk is avoided in English orthography, k rarely occurs in
final position after a stressed, single-letter vowel spelling, because
in such positions it would double before suffixes which begin with
a vowel. (The only common exception is trek — a mid-nineteenth
century borrowing from African Dutch). (On the spelling ck, see
page 90). In medial position it occurs chiefly before e and 1. In
words like snake and dike where the final e marks the correspond-
ence of the preceding vowel, a c spelling for /k/ could not be used
because the final e would also mark the incorrect correspondence
for c.
76 CONSONANT DISTRIBUTIONS

b. Correspondences

k regularly corresponds to {k}.

c. Alternations

(1) Word initial {kn-} becomes {n}, as in knee, and knowledge.

NOTE. In acknowledge, the {kn-} cluster is retained. This,
however, is the only example of the retention of ME {kn-}.
For the parallel retention of {gn-} (agnosticignostic), see
Alternations under g.
(2) {k} in all other cases corresponds to {k}.
(3) Non-productive alternations.
{k} - {δ}
speak speech
leak leach
wreak wretch

a. Distribution

/ occurs in initial, medial, and final position, as in lamb, oleander,

and coal, and in a large number of monomorphemic clusters, the
most common of which are shown below.
Initial clusters Final clusters
pl- play -lp pulp
bl- black -It malt
cl- climb -Id hold
chl- chlorine -lk bulk
gi- glaze -If half
fl- flower -Ich belch
sl- slip -Im helm
spl- splash -rl curl
-rid world
 CONSONANT DISTRIBUTIONS 77

Less common clusters are -lb : bulb, -Itz : waltz, -let: mulct, and
-Ix : calx. Except in combinations with r, I always occurs next to
a vowel in a consonant cluster.

b. Correspondences
(1) The first / in colonel corresponds to {r}.
(2) In would, could and should, I corresponds to {0}.
(3) In all other cases / corresponds to {1}.

c. Alternations

(1) In morpheme-final {ae/a} + {1} -f {k/m/v/f}, {1} corresponds

to {0}, e.g., half, calm, chalk, halves.
(2) In prejunctural position, consonant + {1} corresponds to
consonant + {si} (syllabic /), e.g., simple (cf. simply).
(3) In all other cases, {1} corresponds to {1}.

a. Distribution
m occurs commonly in initial, medial, and final position: man,
moon, demon, stamen, them, bloom. It also occurs in the initial
cluster sm-, e.g., small, smooth, and in the following monomor-
phemic final clusters:
-mp limp, hemp
-mph lymph, nymph
-lm elm, helm
-rm storm, arm
-sm spasm, chasm
-thm rhythm

b. Correspondences
(1) In accompt and comptroller, m corresponds to {n}.
(2) In all other cases, m corresponds to {m}.
78 CONSONANT DISTRIBUTIONS

c. Alternations

(1) The initial cluster {mn-} (mnemonic) is leveled to {n-}.

(2) After {ö} and {z}, pre-junctural {m} becomes {-am} (syllabic
m), e.g., rhythm, spasm, schism.
(3) Non-productive alternations.
(a) com-, con-, corepresenting Latin cum 'with', 'together',
'very', e.g., complex (cf. duplex), conclude (cf. include),
coeval (cf. medieval).
(b) sym-, syn-, syl-, representing Greek syn- 'with', 'along
with', 'at the same time', e.g., symphonic, (cf. phonic),
synonym, (cf. homonym), syllogism, (cf. antilogism).

a. Distribution

η occurs in initial, medial, and final position, although more fre-

quently in medial and final than in initial position. Besides the
cluster sn- (snow, snarl), η also appears in the initial clusters kn-
and gn-. In final position it appears in the following monomor-
phemic clusters :
-nch pinch, wrench
-nd and, bend
-nk bank, drink
-nt ant, infant
-nth labyrinth
-gn sign, malign
-In kiln
-rn churn, concern

b. Correspondences

(1) η in kiln corresponds to {0}.

(2) In all other cases η corresponds to {n}.
 CONSONANT DISTRIBUTIONS 79

c. Alternations

(1) { n } before { k } and { g } b e c o m e s {rj} except in the stressed

w
prefix-morphemes con-, syn-, and in-, and in the unstressed
14
un-, where it remains as {n}.
(2) { n } in all other cases corresponds t o {n}.
(3) N o n - p r o d u c t i v e alternations.
{n} - {0}
passenger passage
messenger message

a. Distribution

ρ occurs in initial, medial and final p o s i t i o n by itself, and in a

variety o f clusters, the m o s t c o m m o n o f w h i c h are pr-, pl-, sp-,
-Ip, -rp, -sp, e.g., program, plough, speed, help, harp, asp. A less
c o m m o n group of clusters w h i c h pattern quite differently f r o m the
ones a b o v e a r e p n - : pneumonia, ps-: psychology, pt-: ptomaine,
-pt: receipt, -rsp : corps. T h e rules for handling these latter clusters
are described below.

b. Correspondences

(1) ρ in corps, coup, and receipt corresponds t o {0}.

14
This should not be taken as a rule, but as a compromise in a phonetically
unstable situation. Kurath summarizes current practice as follows: "The pre-
fixes con-, in-, syn-, when stressed have /n/ beside /η/ before a following Ik/...
Before /g/, the velar /η/ is fully established in congress, congregate, ingot, and
fairly so in congruence, congruous. Occasionally /η/ appears also in unstressed
con-, in-, syn-, as in conclude, concrete, include, increase, synchronic." (Kurath,
1964, p. 71). To state the correspondence of η directly to {η} would be cumber-
some since the forms {g} and {k} could be spelled g, k, q, χ or c. Rules for
relating c to {k} and χ to {ks} would have to be incorporated into this rule,
along with the stress placement rules for con-, in-, and syn-.
80 CONSONANT DISTRIBUTIONS

(2) The first ρ in sapphire corresponds to {0}.15

(3) ρ in all other cases corresponds to {p}.

c. Alternations
(1) Initial {p} in (pn-, ps-, pt-) corresponds to {0}.
(2) In the medial combination bilabial stop + bilabial stop, the
first stop is dropped.

ph

a. Distribution
ph is a relatively infrequent spelling in English although it occurs
in initial, medial, and final position and in the clusters phl-, phr-,
sph-, phth-, -mph, and -Iph, as in phlegm, phrase, sphere, phthisic,
triumph, and sylph.16

b. Correspondences
ph corresponds regularly to {f}.

c. Alternations
(1) {f} becomes {0} in word initial position before {θ} (cf. phthisic
and diphthong).
(2) {f} remains {f} in all other cases. (See also Alternations
under f ) .

a. Distribution
q occurs only in initial and medial position, and in these positions
15
The pph spelling in sapphire could be treated as a single compound con-
sonant unit, akin to tch and dg. Since sapphire contains the only example of
this cluster, however, the present treatment was adopted.
16
phi- and phth- are rare clusters in English.
 CONSONANT DISTRIBUTIONS 81

always before u. It also appears in the initial cluster squ-, e.g.,

squirrel, squire.

b. Correspondences
q corresponds regularly to {k}.

c. Alternations

None.

a. Distribution
r occurs in initial, medial, and final position and in a large number
of clusters, the most common of which are shown below:

Initial r clusters
br- brace gr- grade spr- spring
chr- chrome phr- phrase str- strong
cr- crash pr- prize thr- through
dr- drive scr- screw tr- train
fr- friend shr- shred wr- write
Final r clusters
-rb herb -rf surf -rn urn
-rp sharp -rth mirth -rl curl
-rd bird -rsh marsh -rpt excerpt
-rt smart -rch birch -rst first
-rg berg -rm term -rid world

b. Correspondences
r in all cases corresponds to {r}. (On vowel + r, see Chapter VII).

c. Alternations
None.
82 CONSONANT DISTRIBUTIONS

rh

a. Distribution

The spelling rh is relatively rare, occurring in initial position before

a vowel and medially and finally in the post vocalic cluster rrh,
e.g., rheostate, rhinoceros, cirrhosis, catarrh.

b. Correspondences

rh corresponds regularly to {r}.

c. Alternations

None.

a. Distribution

s occurs in initial, medial and final position by itself, in a large

number of initial consonant clusters, and in four monomorphemic
final clusters. The initial and final clusters are shown below:
Initial Clusters
ps- psychology sl- slow squ- squash
sc- scare sm- small st- stand
sch- school sn- snow sth- sthenic
schw- schwa sp- special str- strong
scl- sclerosis sph- sphere sv- svelte
scr- scream spl- splash sw- swim
sk- ski spr- spring
Final Clusters
-sp wasp -sk ask
-St test -rst thirst
 CONSONANT DISTRIBUTIONS 83

b. Correspondences

The correspondences for initial and final s are fairly regular; those
for medial s, on the other hand, are highly irregular and cannot
be predicted with any high degree of certainty. The rules which
follow are not exhaustive and in some cases have as many excep-
tions as examples.
(1) s corresponds to {0} in the following words:
aisle apropos chassis
corps bourgeois debris
demesne challis rendezvous
island chamois velours
isle
(2) Final, inflectional s corresponds to {s}. The identity of in-
flectional s, however, must be retained on the morphopho-
nemic level.
(3) Final s after a voiced consonant spelling and in as, has, his,
is, was corresponds to {z}.
(4) Final s in all other cases corresponds to {s}.
(5) Initial s corresponds to {s}.
(6) Medial s before or after a voiceless consonant spelling and
in the combinations -ease-, -vowel + Cse corresponds to {s}
except in cleanse where it is {z}.
(7) s in all other cases corresponds to {z}.

c. Alternations
See also Chapter VI.
(1) Between an unstressed vowel and a stressed vowel, {s} corre-
sponds to {z} (cf. sign : design, solve : dissolve, sound: resound).
(2) After {s, z, s, ζ, δ, j} final morphemic {s} becomes {-iz}; after
any other voiced morphophoneme it becomes {z}; otherwise,
it remains as {s}.
(3) When the nouns house and use, and the adjective close are
converted to verbs, {s} becomes {z}.
84 CONSONANT DISTRIBUTIONS

sh

a. Distribution

sh occurs alone in initial, medial, and final position, and in the

monomorphemic clusters shr- and -rsh as in shrimp and shroud,
marsh and harsh.

b. Correspondences

sh regularly corresponds to {s}.

c. Alternations

None.

a. Distribution

t occurs in initial, medial, and final position as a single consonant

and in the following monomorphemic initial and final clusters.

Initial Clusters
pt- ptomaine tm- tmesis
st- storm tr- train
str- strip tw- twenty

Final Clusters
-ct pact -mpt attempt -rst first
-ft left -net distinct -rt smart
-ght fought -nt rent -St must
-let mulct -ntz chintz -tz blitz
-It halt -pt apt -xt text
-ltz waltz -rpt excerpt
 CONSONANT DISTRIBUTIONS 85

b. Correspondences
(1) t corresponds to {0} in depot, debut, hautboy, savant, mortgage,
and in words ending in -et with stress on the last syllable
(e.g., buffet, valet).
(2) t in equation corresponds to {z}.
(3) In the combinations t + vowel + vowel, t corresponds to {s}
when not preceded by s or χ (cf. nation : bastion).
(4) t otherwise corresponds to {t}.

c. Alternations
(1) On the palatalization of {sj} to {s}, {zj} to {z}, {tj} to {δ},
and {dj} to {j}, see Chapter VI.
(2) In the medial clusters {-Ct-} + | ^ j , {t} corresponds to {0},
e.g., castle, often.
(3) {t} otherwise corresponds to {t}.

th

a. Distribution
th occurs in initial, medial, and final position, and in the mono-
morphemic clusters thr- and -rth, as in through, throw, birth, and
north. It also appears in the noun-clusters sth-: sthenic, thw-:
thwart, and -nth : labyrinth.

b. Correspondences
(1) th in thyme corresponds to {t}.
(2) th in isthmus and asthma corresponds to {0}.
(3) Initial th in functors corresponds to {δ}.17
(4) Medial th in the morpheme-final clusters, -the and -ther (ex-
cept for ether) corresponds to {δ}.
(5) In ether and all other cases, th corresponds to {δ}.
17
See Hocket, 1958, pp. 264-65, for a definition of functors.
86 CONSONANT DISTRIBUTIONS

c. Alternations

See also Chapter VI.

Non-productive alternations: 18
{Θ} - {0} north, northern
bath, baths south, southern
breath, breathe swath, swathe
cloth, clothe teeth, teethe
mouth, mouths worth, worthy

a. Distribution

u is a consonant spelling after q, (e.g., quack, queen, acquire), g

(e.g., guano, language, linguist), and, exceptionally, after s as in
suave, persuade. Except for qu- these spellings are rare in Modern
English.

b. Correspondences

(1) Forms ending in -que, -quet, -quette and quee have u corre-
sponding to {0}.
(2) In quay, queue, liquor, piquant, turquoise, u corresponds to {0}.
(3) In all other cases, consonantal u corresponds to {w}.

c. Alternations
None.

18
There is also a tendency towards a {Θ} — {ö} alternation in noun singular—
noun plural, e.g., bath:baths, moth:moths. path:paths.
 CONSONANT DISTRIBUTIONS 87

a. Distributions
Over 75 percent of all occurrences of ν are found in medial posi-
tions in Modern English words; the remainder occur in initial
positions, ν rarely occurs in clusters with other consonants or in
a geminate cluster (exceptions are navvy, divvy, and flivver).

b. Correspondences
ν regularly corresponds to {v}.

c. Alternations
See also Chapter VI.
Non-productive alternations:

W - {w}
vinter wine
Miscellaneous
deceive deception
poverty poor
receive reception
revolve revolution
solve solution
weave web

a. Distribution

w appears commonly in initial position before a vowel and in the

clusters wr- as in write and wrought, tw as in two and twine, and
sw as in swan and sword.
It occurs medially in answer, award, aware, bewail, beware and
reward, toward.
88 CONSONANT DISTRIBUTIONS

b. Correspondences

(1) In answer, sword, two, and toward, w corresponds to {0}.

(2) In all other cases w corresponds to {w}.

c. Alternations

For {v} — {w}, see non-productive alternations under w.

wh

a. Distribution

The spelling wh- occurs only in morpheme-initial position, as in

where, when, why. It does not occur in clusters.

b. Correspondences

(1) wh in who, whore and whole corresponds to {h}.

(2) Elsewhere, wh corresponds to {hw-}.19

c. Alternations

None.

a. Distribution

χ occurs chiefly in medial and final position, although it does

appear initially in a number of scientific terms like xenon, xeno-
phobia, and xylem. Consonant clusters with χ do not occur in
initial or final position, except for Ix'.calx.

19
The dialectal variations of {hw}, {w} and {m.} are not treated here.
 CONSONANT DISTRIBUTIONS 89

b. Correspondences

χ corresponds regularly to {ks}.

c. Alternations

(1) Initially and between voiced morphophonemes, {ks} derived

from χ becomes {gz} if the preceding vowel is unstressed.20
(2) Initial {gz-} corresponds to {z}.

a. Distribution

As a consonant spelling y is relatively rare in Modern English,

occurring in initial position in about one hundred words and in
medial position in beyond, canyon, and in the suffix -yer, as in
lawyer and bowyer (obsolete).

b. Correspondences

y corresponds regularly to {j}.

c. Alternations

None.

20
Before this rule can be applied, the main word stress must be determined
and {h} elision must take place. Otherwise, words like exhaust and exhibit
would be handled incorrectly. Leveling of geminate consonant clusters, how-
ever, must not be done until after rule 1 is applied. That this rule does not
apply to {ks} obtained from spellings other than χ can be seen from the forms
like accept, accede. On the palatalization of {sj} to {§} and {zj} to {z} as in
luxury'.luxurious, see pp. 92 ff. There are several exceptions to this rule, e.g.,
doxology, luxation, and proximity.
90 CONSONANT DISTRIBUTIONS

a. Distribution
ζ is the least frequently used letter in Modern English orthography,
occurring in initial and medial position and less frequently in final
position, as in adz, chintz, quartz, and waltz.21 The only initial
and final clusters it occurs in are cz:czar, rtz: quartz, tz: blitz,
ntz: chintz, and Itz: waltz.
b. Correspondences
(1) ζ in final tz corresponds to {s}.
(2) Otherwise, ζ corresponds to {z}.

c. Alternations
See Alternations, under s.

tch, ck, dg
These three clusters share a number of features; each appears only
after a checked vowel spelling, but in no monomorphemic final
clusters, and each has a single morphophonemic correspondent.
In addition, each was introduced in the early sixteenth century
as a replacement for a geminate consonant cluster, tch replaced
cch; ck replaced cc and kk; and dg replaced gg when it represented
0}·
Correspondences22
tch corresponds to {δ}, as in kitchen, match,
ck corresponds to {k}, as in black, neck,
dg corresponds to {]}, as in edge, midget.
21
Frequency counts for letters in English texts have appeared in Fletcher
Pratt, Secret and Urgent: the Story of Codes and Ciphers (Indianapolis, 1939)
and H. F. Gaines, Cryptanalysis (New York, 1939).
82
From a theoretical standpoint, these three clusters do not have to be
treated as separate units. Correspondences for ck can be handled by the cor-
respondence rules given under c and k. For tch and dg, the rules for t, d, ch,
and g can be applied, along with the rule for leveling geminate consonant
clusters, given that {c} is represented as {ts} and {j} as {dz}. Then tch and dg
are mapped into {tts} and {ddz} and these are later leveled to {t§} and {dz}.
 CONSONANT DISTRIBUTIONS 91

Some forms like exaggerate, account, and bacchanal have retained

Latin geminate clusters in spite of the general tendency in the six-
teenth century to replace them with tch, ck, and dg.

gn, kh, sch

These three consonant units appear in a small number of bor-

rowings; the words in which they occur and the correspondences
for each are shown below:
gn
gn corresponds to {n} in champagne
gn corresponds to {nj} in lorgnette, mignon, mignonette, poig-
nant and vignette
kh
kh corresponds to {k} in khan and khaki
sch
sch corresponds to {s} in schism
sch corresponds to {s} in schist and schwa.
 VI

FRICATIVE ALTERNATIONS

1. PALATALIZATION

a. Phonetic Process

The process of palatalization (often called assibilation) accounts

for many so-called irregular spelling-to-sound correspondences
which involve the spellings d, s, t, χ, ζ and the vowels which follow
them in certain environments.
Under this name (assibilation) it is convenient here to comprise two
changes, /sj/ and /zj/<(s, z), and /tj/ and /dj/<(ts, dz). In the fiist
we havt a sort of complete assimilation of the two sounds. In the second
change, the off-glide from the more or less palatalized stop /t, d/ had
developed into the sibilant, which has in most cases absorbed the fol-
lowing /j/.1
Kenyon (1943, p. 135), in what is probably the best summary to
be found of the phonetics of this process, describes two phonetic
tendencies which lead to palatalization. The first is the shift of
unaccented /if before another vowel to /j/, so that the number
of syllables in the word is reduced by one. Then, through assim-
ilation, the changes /sj, zj, tj, dj/ — /s, ζ, δ, jf/ occur. In the model
used in this paper, palatalization is described as a morphopho-
nemic process. Thus, {sj, zj, tj, dj} become {s, z, c, j} in certain
environments.
The palatalizing {j} is derived from several independent
sources. In creature, for example, u is mapped directly into {ju};
1
Jespersen, 1909,1, 341.
 FRICATIVE ALTERNATIONS 93

in cordial, however, i is mapped first into {i}, and then unstressed

{1} + vowel becomes {j} + vowel (with some exceptions). Palat-
alization generally takes place before an unstressed vowel; com-
pare, for example, the stress on u in the following lists:
Palatalized Unpalatalized
credulous credulity
cynosure pursuit
capitulate presume
schedule importune
assiduous assiduity
treasure institute

b. Morphophonemic rules for palatalization

(1) {sj} + (unstressed vowel) -> {s} + (vowel)

(2) {zj} + (unstressed vowel) -> {z} + (vowel)
(3) {tj} + (unstressed vowel) -> {c} -+- (vowel)
(4) {dj} + (unstressed vowel) {j} + (vowel)
Examples:
luxury {luksjuri} -> {laksuri}
azure {aezjur} -> {aezjar} -> {aezar}
creature -> {kritjur} -> {kritjar} -> {kricsr}
cordial -* {kordial} -»> {kordjal} {korjal}
Exceptions:
sure, sugar, mature, produce(n).

2. VOICED-VOICELESS ALTERNATIONS

The morphophonemic alternations {s-z}, {f-v}, {θ-δ}, {s-z} reveal

a number of problems in the construction of a model for spelling-
to-sound correspondences. These problems, though encountered
here in traversing from the orthography toward sound, also occur
in the construction of other models for English morphophonemics,
94 FRICATIVE ALTERNATIONS

so that the considerations presented below have a much wider

significance than in spelling-to-sound studies alone. In the dis-
cussion which follows it is assumed that spellings have been mapped
onto an initial morphophonemic level and the next objective is to
describe alternations in terms of regular and irregular rules. For
this section, an alternation that is both productive and frequent
is labeled REGULAR. All others are IRREGULAR. The specific prob-
lems which are discussed in relation to the voiceless-voiced alter-
nations are:
(1) generalizations which can be made about these alternations.
(2) relevant differences which exist among the non-productive
(irregular) patterns.
Whether or not an alternation is marked in the orthography is
relevant for both reading and spelling. This fact will be mentioned
for each pattern, but not discussed thoroughly.

Summary of the Alternations

(I) Noun plural (s-z(iz)} cats : dogs(horses) (unmarked)
(II) Noun-adjective
(a) {s-z} luxury : luxurious (unmarked)
(b) {s-z} louse : lousy (unmarked)
(c) {Θ—5} north : northern (unmarked)
(III) Verb-verb
{s-z} sound: resound (unmarked)
(IV) Noun-verb (infinitive)
(a) {f-v} belief: believe (marked)
(b) {θ-ö} breath : breathe (marked)
(c) {s-z}
(1) house (n.): house (v.) (unmarked)
(2) glass : glaze (marked)
(V) Noun singular-noun plural
(a) {f-v} wife : wives (marked)
(b) {θ-ö} bath : baths (unmarked)
(c) {s-z} house : houses (unmarked)
(VI) Verb (past)-verb (present)
{f-v} left: leave (marked)
 FRICATIVE ALTERNATIONS 95

(VII) Noun-noun
{θ-ö} smith : smithy (dialectal) (unmarked)
(VIII) Ordinal-cardinal
{f-v} fifth : five (marked)
Of these alternations, only a few are invariant. The noun plural
(I) certainly is. It is a phonetically conditioned alternation which
is describable by a single rule. It is generally not marked in the
spelling, as shown by the following plurals, all of which add only
s to the singular:
houses {hauziz}
boys {boiz}
cats {kaets}
(Where -es is added for the plural, the alternate {iz} is marked).
Class VIII, the ordinal-cardinal alternations, is also invariant
since the only two members which end in voiced {v} in the cardinal
have unvoiced {f} in the ordinal, viz., five : fifth, twelve : twelfth.
One rule for generation of the ordinal forms from the cardinal
forms, therefore, can be stated as:
final, cardinal {v} -*• ordinal {f}.
It is also possible to state this rule as follows:
final voiced fricative (cardinal) unvoiced fricative (or-
dinal)
because no other cardinals end in a voiced fricative. Such a gen-
eralization, however, is probably not acceptable unless there are
other major patterns in the orthography which show a voiced-
voiceless alternation for fricatives. The existence of such patterns
beyond the noun plural are explored in the remainder of this
section.
Class IV b, the {θ-δ} alternation in noun-verb, is nearly in-
variant, but since it has so few examples, there is a temptation
to base its classification upon those of the other class IV alter-
nations. The nouns which end in {Θ} and which can also be used
as verbs are the following:
96 FRICATIVE ALTERNATIONS

{θ-ö} {θ-θ>
n. v. n. v.
breath breathe froth froth
cloth clothe earth (un)earth
mouth mouth
swath swathe
teeth teethe
The {f-v} alternations for noun-verb are not nearly so regularas,
the following lists demonstrate.
{f-v} {f-f}
n. v. n. v.
belief believe brief brief
grief grieve dwarf dwarf
half halve leaf leaf
life live sniff sniff
proof prove spoof spoof
relief relieve stuff stuff
shelf shelve whiff whiff
thief thieve
It would be stretching a point considerably to state that in general
nouns ending in {f} change {f} to {v} in forming the corresponding
infinitive. At best, it can be said that there is a class of nouns
(which must be enumerated in the lexicon attached to the model)
in which the voiceless-voiced alternation occurs for this production.
The case for an {s-z} alternation in this class is even more ten-
uous. Abuse, grease (dialectically), house, and use have this alter-
nation, as do, in an extended sense, brass {braze), glass {glaze),
and grass {graze). But there is an even longer list of forms like
class, curse, dress, eclipse, glimpse, mass, nurse, pass, press, stress,
and witness in which no such alternation occurs. In general, nouns
ending in {s} do not change {s} to {z} in forming the corresponding
infinitive.
On the basis of this evidence there is no justification for estab-
lishing a regular voiceless-voiced alternation for this class. This
indicates, furthermore, that no special status should be assigned
 FRICATIVE ALTERNATIONS 97

to the {θ-δ} alternation, even though it occurs in five of the seven

possible cases. The only conclusion that can be drawn is that
nouns ending in voiceless fricatives generally retain the voiceless
quality in forming the infinitive. There is, however, a productive
subpattern in which the voiceless fricative becomes voiced.
Class II a, which has only one example in the corpus used for
this study, and probably not very many more in all of the English
language, might also be classed as regular. But this depends upon
whether the alternation {ks} {gz} is classed as regular or not.
If spelling is used as the starting point, then this is possible, but
not necessarily desirable. Intervocalic χ could be mapped into
{k x s x } and then a stress rule applied to convert {k x s x } to {gz} :
{k x s x } {gz} when the primary stress is not on the preceding
vowel, cf. execute :exist. {ks} derived from x, here labeled {kxsx},
must be distinguished from {ks}, derived from cc, since {ks} de-
rived from cc does not become {gz} under any circumstances (cf.
access : accessible, accept: acceptable). Under the classifications
of χ-derived {ks} and cc-derived {ks}, the {s-z} alternation in
luxury: luxurious is regular; the actual alternation, however, is
not {s-z}, but {s-z}. {s} and {z} result from the palatalizations
{sj} {s} and {zj} {z}. This means first that the alternation
{s-z} is, from the standpoint of a generative model, actually the
alternation {s-z} and second, that it is regular only if the identity
of λ> derived {ks} is maintained until after stress is applied. If
this identity is not maintained, then there is no reason to call
{ks} {gz} regular. In lieu of the lack of other criteria to apply,
frequency favors the invariantly voiceless cluster: accent: accen-
tuate, accept: acceptable, access : accessible, accident: accidental.
Therefore, the alternation in luxury: luxurious is not classed as
regular here.
Class III could, and probably should, be designated as regular.
The members of this class are:
sign, design (resign, etc.)
solve, dissolve (absolve, resolve, etc.)
sound, resound.
 98 FRICATIVE ALTERNATIONS

The rule is that initial {s} becomes {z} when a prefix ending in an
unstressed vowel is added.
Class Y is the largest of the remaining classes, containing the
following pairs:
{s-z} {f-v} {θ-δ}
house, houses calf, calves bath, baths
elf, elves cloth, cloths
half, halves lath, laths
knife, knives moth, moths
leaf, leaves mouth, mouths
life, lives path, paths
loaf, loaves wreath, wreaths
self, selves (Most of the plurals
sheaf, sheaves above vary between
shelf, shelves {Θ} and {δ}. The
staff, staves dialect distribution
thief, thieves for this alternation
wife, wives has not been fully
wolf, wolves determined).
The lack of a {s-z} alternation and the rarity of the {s-z} alterna-
tion eliminate the possibilty of a general voicing rule for final
fricatives in forming the noun plural. This leaves the possibility
of labeling the {f-v} and {θ-δ} alternations as regular. But the
regular label implies either that there is a phonetic or grammatical
factor that will allow the prediction of the alternation, or that this
alternation occurs with a high frequency. That the {f-v} or {θ-δ}
alternations are not phonetically or grammatically predictable can
be seen from the following items:
gulf gulfs breath breaths month months
reef reefs width widths earth earths
The frequency question also is answered in the negative, although
the {f-v} alternation does occur in almost one-third of the nouns
which end in final {f}. The remainder retain {f} in forming the
plural. This would imply that the leaf: leaves alternation should
 FRICATIVE ALTERNATIONS 99

be classed as irregular, just as is the petrified alternation represented

by smith: smithy. Our native speaker's intuition, however, is
uncomfortable with this solution.
The smith : smithy alternation is not only uncommon, but also
would be rarely invoked as a model for forming new English
words. Leaf: leaves on the other hand, appears more productive.
We know, for example, that voiced fricative plurals of such
words as dwarf and roof occur in American speech. Furthermore,
the formation of the noun plural from the noun singular is one of
the most common forms of word formation in English. Formation
of one noun from another by the addition of y is much less pro-
ductive.
The issue here is that a model that concerns itself strictly with
regular and irregular categories fails to account for some features
of the language. While much work remains to be done, the tenta-
tive decision is to subdivide the irregular class into productive and
petrified subclasses. At best, alternations like leaf: leaves fit into
the productive-irregular class while those like smith:smithy fit into
the petrified-irregular group.
Finally, it should be noted that all {f-v} alternations are marked
in the orthography by change in the functional unit, while in the
{s-z} alternations only the petrified alternation represented by
glass: glaze is marked (also by a change of functional unit), and
in the {θ-δ} alternations, only the noun-verb one is marked, and
this by addition of the final e marker which is the only possible
device for marking the voiced pronunciation of th.
A reclassification of these alternations, based upon the newly
adopted classes, is shown below:
REGULAR
(1) noun plural {s-z (-Iz)} cats : dogs
(2) verb-verb {s-z} sound: resound
(3) ordinal-cardinal {f-v} fifth: five
IRREGULAR
Productive
(1) noun-adjective {θ-δ} north : northern
(2) noun-verb {θ-δ} breath: breathe
100 FRICATIVE ALTERNATIONS

{f-v} belief: believe

{s-z} house (n.): house (v.)
(3) noun singular-noun plural {f-v} wife: wives
{θ-δ} bath : baths
{s-z} house: houses
Petrified
(1) noun-adjective {s-z} louse: lousy
luxury: luxurious
(2) noun-verb {s-z} glass: glaze
(3) verb (past)-verb (present) {f-v} left: leave
(4) noun-noun {θ-δ} smith : smithy
thief: thieve (but theft)
 VII

VOWEL DISTRIBUTIONS, CORRESPONDENCES,

AND ALTERNATIONS

1. PRIMARY VOWEL PATTERNS

a. Introduction

The vowel spellings a, e> ijy, o, u, called primary vowel spellings

in this paper, carry the major burden of vowel representation in
the current orthography. 1 They occur in all positions and have
a vast complexity of morphophonemic correspondences and alter-
nations which reflect an even more complex history. When viewed
from the direct spelling-to-sound standpoint, the patterns for
these units reveal no regularity, ο corresponds to seventeen differ-
ent sounds, a to ten, e to nine, and the combined group to forty-
eight. When the morphemic structure and consonant environ-
ments of the words in which these units appear are analyzed, how-
ever, a single major pattern emerges, from which regular sub-
patterns can be derived. Exceptions still remain, large numbers
of them in some cases, but the underlying pattern is so dominating
that the exceptions, which were once the rule, become mere od-
dities, begging for historical justification. In the discussion which
follows the major pattern for the stressed vowels is introduced
in a general form and then refined through the introduction of its
regular sub-patterns, alternations and exceptions.2
1
Henceforth i will stand for both i and y.
2
Word stress patterns play a significant role in the relationship of spelling
to sound, especially in the correspondences of the primary vowel spellings.
With a few exceptions, only stressed vowels are treated in this paper. Three
levels of stress, all introduced on the morphophonemic level, are assumed:
102 VOWEL DISTRIBUTIONS

b. Major Patterns

Each of the primary vowel units corresponds regularly to two

different morphophonemes, a checked one and a free one, accord-
ing to the morphemic structure of the word in which it occurs and
the consonant and vowel units which follow it. These correspond-
ences are shown in Table l. 8
TABLE 1

Major Pattern for Primary Vowels

Spelling Free Alternate Checked Alternate
a {e} {*>
sane sanity
mate mat
ration rattle
e {i} {ε}
athlete athletic
mete met
penal pennant
I {ai} {1}
rise risen
mal/gn malignant
site sit
ο {o} {a}
cone conic
robe rob
posy possible
u {ju}1 W
induce induction
rude rudder
lucre luxury

primary, secondary, and tertiary (unstressed). While some forms of English

word stress are "predictable", no extensive analysis of this topic has ever been
published. Two recent publications, however, indicate that work is being done
in this area. See George S. Waldo, "The Significant of Accentuation in English
Words", Proceedings of the Ninth International Congress of Linguists (The
Hague, 1964), pp. 204-10, and Roger Kingdon, The Groundwork of English
Stress (London, 1958).
3
For exceptions to this pattern, stated as direct spelling-to-sound correspond-
ences, see Wijk, pp. 26-41.
4
The retention or elision of {j} before {u} is handled as a morphophonemic
process.
 VOWEL DISTRIBUTIONS 103

In monomorphemic words a primary spelling unit corresponds to

its free alternate when it is followed by (1) a functionally simple
consonant unit which in turn is followed by another vowel unit
(including final e) or (2) a functionally simple consonant unit,
followed by I or r, and then another vowel unit (including final
e).b It corresponds to its checked alternate in the remaining cases,
i.e., when followed by (1) a functionally compound consonant
unit, e.g., JC, dg, (2) a cluster of consonant units, e.g., -nn, Ith or,
(3) a word-final consonant unit or units. Examples of these corre-
spondences are shown in Table 2. The column numbers corre-
spond to the numbered qualifications in the sentences above.

TABLE 2
Examples of Primary Vowel Correspondences for Selected
Environments

Spelling Free Alternate Checked Alternate

1 2 1 2 3
a canine ladle badge saddle sat
e median zebra exit antenna ebb
i pilot microbe chicken epistle hitch
ο vogue noble pocket cognate sod
u dubious lucre luxury supper rug

A simple vowel spelling, followed by a simple consonant spelling

and then le, corresponds to its free alternate, e.g., table. But a
geminate consonant cluster before the le marks the checked alter-
nate, e.g., apple.

6
The difference between monomorphemic and polymorphemic words, a
difference unfortunately neglected in the teaching of reading, is too complex
to be discussed adequately here. For the prediction of sound from spelling
in a large number of words, however, the distinction is crucial. What are in-
volved, primarily, are the morphophonemic alternations which occur with
suffixation, as in /arben/: /arbäniti/, /kon/: /kämk/. On this topic, see Stanley S.
Newman, "English Suffixation: A Descriptive Approach", Word4 (1948), 24-36.
For a pedagogical approach to suffixation, see Edward L. Thorndike, The
Teaching of English Suffixes (New York, 1941).
104 VOWEL DISTRIBUTIONS

TABLE 3

The le Pattern
a ladle addle
e pebble
i rifle riffle
0 noble cobble
u ruble rubble

This pattern has a limited distribution. Only b, p, t, f , g, and ng

occur more than once before final le, and of these, only b, ρ, d
t, a n d / double in this position, (5 and χ occur before le in measles
and axle, ss occurs in tussle). Treble, an exception for the pronun-
ciation of e, is the only example in Modern English of the spelling
e + C + le (final).
The correspondences for these vowel spellings in polymorphemic
words depends not only upon the graphemic environment, but
also in many cases upon the morphemic structure of the word.

c. Sub-Patterns
The two most important sub-patterns which can be derived from
the major pattern are the final e pattern and the geminate conso-
nant pattern. Examples of these are shown in Table 4. Complete
analyses are given in the two following sections.

TABLE 4

Examples of Final e and Geminate Consonant Patterns

Spelling Final e Pattern Geminate Consonant Pattern
a rate — rat anal — annals
e mete — met Peter — petter
i s/te — s/t diner — dinner
0 pope — pop coma — comma
u cwte — cut SHper — supper

d. Final e Pattern
While the final e pattern applies primarily to monosyllablic words,
it also holds for many polysyllabic words, even when the vowel
 VOWEL DISTRIBUTIONS 105

before the final e is unstressed, as in microbe, decade, schedule,

volume, placate. Besides the patterns mentioned above (vowel +
consonant + e, vowel + consonant + le, vowel + consonant -f re),
the environment vowel + ste is also part of the final e pattern, as
in baste, chaste and haste. Examples of regular correspondences
are shown below, followed by an exhaustive list of exceptions for
stressed vowels.
TABLE 5

Examples of the Final e Pattern

«-{e} i -> {ai} u {ju> ο -> {ο}
bake cycle cube accede cove
decade domic/le duke impede erode
fable f/ve mwle obsolete globe
gage prize produce scheme joke
haste profile resume serene mediocre
shake subl/me secure theme smoke

e. Irregular Correspondences for the Final e Pattern

(1) a corresponds to {ae} in bade, forbade, have, morale. a corre-
sponds to {a} in are, barrage, camouflage, corsage, facade,
garage, massage, mirage, sabotage, NOTE: a in the ending
-ate corresponds to {e} in verbs, but alternates to {i} in nouns
and adjectives. Cf. duplicate (vb.): duplicate (adj., noun).
(The stress patterns are also different in these forms).
(2) e corresponds to {ε} in allege, clientele, ere, there, treble,
where.6
e corresponds to {i} in renege,
e corresponds to {a} in were,
e corresponds to {i} in fete.
(3) i corresponds to {i} in the following words:
bas tile7 mach/ne rav/ne val/'se
6
In the following words final e is not a marker, but a relational unit, (1)
corresponding to {i}: adobe, coyote, epitome, extempore, facsimile, finale,
hyperbole, nike, recipe, sesame, simile, ukulele, (2) corresponding to {e}: cafe,
protege.
7
Bastile is an alternate spelling of bastille.
106 VOWEL DISTRIBUTIONS

caprice marine regime

elite police routine
prestige sardine
tangerine
ι corresponds to {1} in give and live, NOTE: The spelling i is
highly irregular in the ending -me when it does not receive the
primary word stress. Cf. canine, asinine : examine, famine.
(4) ο corresponds to {a} in above, come, done, dove, glove, love,
none, shove, some.
ο corresponds to {u} in lose, move, prove, whose,
ο corresponds to {0} in gone.
f. Vowel Spellings before Geminate Consonants
The primary spellings a, e, i, o, u occur frequently before geminate
consonants, the digraph (secondary) spellings, rarely.8 (The French
8
Except for the Spanish borrowing llama, geminate consonants occur only
in medial and final position in English spelling. The most common final clus-
ters are ff, 11, and ss as in fluff, mill and toss. Rare final clusters are bb, dd,
gg, nn, rr, tt, and zz, which occur only in the words, ebb, add, odd, egg, inn,
err, burr, purr, whirr, boycott, butt, putt, watt, buzz, fizz, fuzz, jazz, razz. In
medial position the following are common:
bb cabbage, lobby mm comment, summit
cc broccoli, occupy nn annals, dinner
dd addict, eddy PP copper, happy
ff coffee, giraffe rr error, surrey
gg luggage, nugget ss blossom, gossip
11 fillet, pillow tt button, clutter
vv and zz also occur medially, but only in divvy, flivver, navvy, blizzard, dizzy.
N o other geminate clusters occur, although dg, ck, and tch can be considered
as geminate substitutes.
Consonant gemination also occurs when certain suffixes are added to words
ending in a single consonant. For example, run: running, forbid:forbidding,
abet:abetted. The orthographic rule is usually stated as follows:
Monosyllables and words accented on the last syllable, when ending in a
single consonant, b, d , f , g, k, I, m, n, p, r, s, t, ν preceded by a single vowel,
double the consonant before adding a termination beginning with a vowel
or the suffix -y. Webster's, 1856, p. 1145.
Exceptions to this rule are plentiful and usage is still unsettled on whether or
not to double final /. For a colorful survey of eighteenth and nineteenth cen-
tury grammatical literature on this topic, see Goold Brown's, Grammar of
English Grammars (4th ed. New York, 1859), pp. 198-200.
The non-gemination of ν and th has led to a large number of exceptions to the
major pattern for the correspondences of the primary vowel spellings. Cover,
 VOWEL DISTRIBUTIONS 107

borrowings, braille, chauffeur, and trousseau are the only examples

from the corpus used for this study). Before geminate consonant
clusters, primary vowel spellings correspond to their checked
alternates, with the following exceptions:
(1) a in mamma corresponds to {a} and a in marshmallow cor-
responds to {ε}.
(2) ο in across, albatross, and boss and before ff corresponds to
{ο}·
(3) ο in gross, and before final 11 (except for doll), corresponds
to {o}.
(4) u in butte corresponds to {ju} and u in pudding and pussy
corresponds to {u}.

TABLE 6

Examples of Regular Correspondences Before Geminate Consonants

a -»• {ae} e {ε} i ->{1} ο {a} U ->{3}
abbess appellate artillery accommodate button
accent beggar bacillus collar funnel
apple bellow blizzard college fanny
babbittry cellar cribbage comma hullabaloo
cabbage Cheddar issue commerce hummock
callow dilemma million dollar mummy
fallacy ebb shrill hobby puddle
flabby fellow symmetry hollow pummel
grammar kennel vanilla lobby rubber
happen lesson village sonnet shutter
mammal message willow toboggan supper
rattle tennis wriggle toggle tunnel

bevel, level, river, brother, mother, other, for example, have vowel spellings
corresponding to checked alternates in environments which indicate free alter-
nates. To indicate the checked alternate, ν and th would have to be geminated,
but the graphotactical patterns of English exclude the doubling of these units.
In prison, a slightly different problem exists. While ss, which Is needed to
make the correspondence i — {i}, is allowed, it generally corresponds to {s}
in medial position, as in blossom, gossip, and lasso, so it could not be employed
where s corresponds to {z}. The present use of ss is derived from old French
orthography where "intervocalic ss served to distinguish voiceless s from
voiced s ( = z)." Alfred Ewert, The French Language (London, 1933), p. 113.
108 VOWEL DISTRIBUTIONS

2. ALTERNATIONS BASED UPON PRIMARY VOWEL SPELLINGS

The major pattern for the primary vowel spellings in stressed posi-
tions depends upon two basic features: environment, which was
discussed in the previous sections, and morphemic structure, which
is discussed briefly in this and the following sections. Morphemic
structure forms the basis for describing the morphophonemic
alternations based upon the primary vowel spellings. For example,
the word sanity, if considered solely on the basis of the rules given
in the preceding section, would be an exception to the major pat-
tern since a before a simple consonant unit followed by a vowel
corresponds to its checked rather than to its free alternate. If
viewed, however, in relation to the sequences, sane: sanity, hu-
mane : humanity, and urbane : urbanity, another regular feature can
be seen. By starting with the forms sane, humane, and urbane,
regular rules can be written for changing the free alternate {e} to
the checked alternate {ae} when the suffix -ity: {-iti} is added.
This rule also holds for the spellings e, i and o, as can be seen
from the following examples:
extreme extremity
obscene obscenity
serene serenity
asinine asininity
divine divinity
malign malignity
senile senility
frivolous frivolity
mediocre mediocrity
precocious precocity
verbose verbosity
(Morphophonemic alternations based upon stressed u are rare in
Modern English, the most common being those which occur in
assume : assumption, conduce : conduction, presume : presumption,
reduce : reduction. Even with loss of stress u: {ju} tends not to
change to {a}. Thus, compute : computation, execute : execution,
usurer : usurious, utilize : utility).
 VOWEL DISTRIBUTIONS 109

While a complete survey of vowel morphophonemics is beyond

the scope of this paper, some of the more common alternations
are presented below:
(1) -ic: free — checked
a angel angelic
state static
e athlete athletic
hygiene hygienic
meter metric
i cycle cyclic
mime m/mic
paralyze paralytic
0 cone conic
neurosis neurotic
phone phonic
ion free — checked
e concede concession
convene convention
i collide collision
decide decision
provide provision

3. CONSONANT INFLUENCES

r
a. Introduction
Post-vocalic /r/ is the source of not only a wide variation of vowel
pronunciations across dialects, but also a complex and in some
places, irregular development of spelling-to-sound correspond-
ences. The selection of transcriptions of post-vocalic /r/ words
is, in some cases, arbitrary. One speaker may alternate freely
between such forms as /ziro/ and /ziro/, /meri/ and /meri/ and
/born/ and /born/. Some speakers contrast horse : hoarse, for:four,
marry: merry: Mary, and some do not. 9 In the transcriptions
• See Kenyon, 1958, pp. 110-12.
110 VOWEL DISTRIBUTIONS

used for this paper, /e/ and /i/ have coalesced with tautosyllabic
/ε/ and /i/. Intra-dialectal variation between /a/ and /o/ and
between /o/ and /o/ is indicated by a/o and a/o (either morpho-
phonemic or phonemic), as in orange, origin, forest, g/ory, moron,
orient, and the syllabic peak in words like 6m/, word, and wrge is
symbolized phonemically (and morphophonemically) as /-ar/; e.g.,
bird /bard/, word /ward/, urge /arj/.10

b. Correspondences

The correspondences of a primary vowel spelling before tauto-

syllabic r depends upon the environment following r. Three cases
must be considered:
(1) r followed by a vowel unit, which in turn is followed by a
vowel unit or juncture.
(2) r followed by a vowel unit, which in turn is followed by a
consonant, or r followed by r.
(3) r followed by a consonant or juncture.

The morphophonemic correspondences for the primary vowel

spellings are shown below. Environments 1, 2 and 3 refer to the
descriptions just given. The r columns contain the correspondences
for the vowels before r; the normal column contains the corre-
spondences normally anticipated for that environment.

TABLE 7

Vowel Correspondences Before r

Spelling Environment 1 Environment 2 Environment 3
r normal r normal r normal
a ε e ae X a ae
e I i ε ε 3 ε
i ai ai I I 3 1
ο o/o ο a/o a o/o a
u ju ju 0 3 9 3

10
See Use Lehiste, Acoustical Characteristics of Selected English Consonants
(= I J AL Publication No. 34) (Bloomington, Indiana, 1964), pp. 51-115.
 VOWEL DISTRIBUTIONS 111

EXAMPLES

Environment 1
a beware, malaria, nefarious, secretary
e adhere, cereal, exterior, sphere
i dire, enquiry, hire, wire
ο adore, glory, more, shore
u bureau, cure, mature, spurious

Environment 2
a arid, arrogate, marriage, tariff
e austerity, errand, peril, terrace
i empiric, irrigate, miracle, mirror
ο borrow, foreign, horrid, orange
u burr, current, furrier, hurry

Environment 3
a alarm, carve, gargle, star
e erb, erst, her, infer
i bird, girl, virtue, whirl
ο adorn, formula, or, storm
u cur, spur, urge, urn

EXCEPTIONS

Environment 1
a are, aria, safari
e very
i delirium
u bury

Environment 2
a alarum, catarrh, harem
e err
i iris, irony, siren, spiral, squirrel, stirrup, tirade, virus
ο borough, thorough, worry
u mural, urine
112 VOWEL DISTRIBUTIONS

Environment 3
a scarce
e concerto, sergeant
ο attorney

Correspondences

Before final -// and medial or final -/ plus another consonant, a

corresponds not to {ae}, but to {o} as in almanac, alternation, call,
chalk, mall, psalm, and walk.11 This shift does not occur before
medial 11, however. Compare, for example, the following: call:
calliper ; fall: fallacy; gall: gallow; hall: hallow; mall: mallard;
tall: tallow. In addition, ο before final -// or medial / plus con-
sonant corresponds to {o} rather than to {a}.12 Thus,
bold polka boll: bollard
folk scold poll: pollen
jolt told roll: rollick
molt yolk troll: trolley
That no other consonants are influenced by -11 can be seen from
the following examples:
bell: bellow fill: filly
fell: fellow bull: bullet
bill: billion gull: gullet

4. W INFLUENCE

With the exception of the a + / patterns discussed above, a pre-

ceded by initial w i n a checked environment corresponds to {a}
before spelling for velars ({k, g, η}) and to {a} otherwise. Com-
pare, for example, wad, waffle, wander, wasp: wax, wag, waggle,
11
Some exceptions are alkali, altitude, contralto and palmetto.
12
Exceptions are dolphin, olfactory, revolve, solve and volcano.
 VOWEL DISTRIBUTIONS 113

wangle. With many of the w + a words there is considerable

dialectal variation (e.g., want, wash, water).
o + r preceded by w corresponds to {ar}, except for worn which
has ο {ο}.

5. MISCELLANEOUS CONSONANT INFLUENCES

i before final nd, Id, and gn/gm corresponds to its free alternate
{ai} rather than its checked alternate {i}.
-nd behind grind remind
bind hind rind
blind kind wind (v.)
find mind
-Id child, mild, wild (but gild)
-gn align design
assign malign
benign resign
consign sign
-gm paradigm
ο before final Id corresponds to its free alternate {o} rather than
its checked alternate {a}.
bold old
fold scold
hold sold
mold

gh
Of the simple vowel spellings, only i occurs before gh, and in this
environment i corresponds invariably to {ai} and gh to {0}.
Examples
blight light thigh
fight right tight
flight sigh
high slight
Of the compound vowel spellings, ai, au, ei, and ou occur before
114 VOWEL DISTRIBUTIONS

gh, but the correspondences for these vowels or for gh in these

environments are not entirely predictable. Shown below are the
correspondences for each compound vowel, along with exhaustive
word listings from the research corpus for each correspondence.
ai {e} straight
au - • {ae} draught, laugh
au {0} caught naughty taught
daughter naught
fraught slaughter
Note that au -* {ae} implies gh -> {f} while au -> {0} implies gh
- w .
ei -> {ai} height, sleight
ei -*• {e} sleigh
freight neighbor
inveigh weigh
ou -*• {3} clough, enough, rough, slough,
sough, tough
ou -*• {0} borough, dough, furlough, thorough,
though
ou -*• {a} hough
ou {u} through
ou {au} bough, slough, sough
ou {0} bought ought wrought
cough sought
fought thought
nought trough

6. SECONDARY VOWEL PATTERNS

a. Introduction

The secondary vowel spellings differ from the primary vowel

spellings in several important ways. First, they occur less fre-
quently and have a more limited distribution. None appears
commonly before geminate consonant clusters; some like ai, au,
 VOWEL DISTRIBUTIONS 115

ei, and eu rarely occur in word-final position; others, like ie and

oa rarely occur in word initial position. Second, while each pri-
mary vowel spelling has two basic correspondences, according to
the graphemic environment and the morphemic composition of
the word in which it occurs, each secondary vowel spelling gen-
erally has a single major correspondent. Third, the morphophon-
emic correspondences based upon the secondary vowel spellings
tend not to alternate in quality with reduction in stress. Compare,
for example, the first vowels in neutral: neutrality, cause : causa-
tion with those in melody : melodious; potent: impotent.
Historically, the secondary spellings also differ from the primary
ones. Primary vowel spellings are found in the earliest English
records; their correspondences can be traced through a compli-
cated chain of sound changes from Old English to the present
time. Most secondary vowel spellings, on the other hand, were
introduced during the late Middle English period and consequently
have been involved in considerably fewer sound changes. Cor-
respondences for the secondary vowel spellings follow, divided
into major and minor categories on the basis of frequency of
occurrence in different English words.

b. Correspondences
ai/ay

Major Correspondences
{e} : bait, day, player, wa/t

Minor Correspondences
{ai} : aisle, aye, bayou, cayenne
{i} : plait, quay
{ε} : again, against, said
{ae} : plaid
au/aw
Major Correspondences
{0} : audience, claw, cause
116 VOWEL DISTRIBUTIONS

Minor Correspondences
{e} : gauge
{ae} : aunt, draught, laugh
{o} : chai/ffeur, chauvinist, hautboy, mauve
{au} : sauerkraut, umlaut

ea
Major Correspondences
{i} : breach, each, reach, teach
Minor Correspondences
{e} : break, great, steak, yea
{ε} : (1) before -I, e.g., health, realm, wealth
(2) before -sure, e.g., measure, pleasure, treasure
(3)in the following words:
bread spread endeavor pheasant
breadth thread feather sweat
dead tread weather sweater
dread treadle heaven threat
head breast heavy treachery
instead breath leather weapon
meadow deaf leaven weather
ready death peasant

ee
Major Correspondences
{i} : bleed, eel, absentee
Minor Correspondences
{i} : been, breeches, creek
{e} : matinee, melee

ei/ey

Major Correspondences
{e} : abeyance, obe^, re/gn, veil
 VOWEL DISTRIBUTIONS 117

Minor Correspondences
{ai} : eye, fahrenheit, geyser, gneiss, height, holstein
kale/doscope, sleight, stein, walleye
{i} : caffein, ceiling, conceit, conceive, deceit, deceive,
either, key, leisure, neither, receipt, receive, seize,
sheik
{ε} : heifer
eu/ew

Major Correspondences
{ju} : ewe, ewcharist, neutron, pewter
Minor Correspondences
{o} : sew, shew
ie

Major Correspondences
{ai} (in final position in monosyllables): die, lie, pie, tie
{1} (in final position in polysyllables): calorie, collie, eerie,
movie
{i} (medial position): achieve, diesel, niece, yield
Minor Correspondences
{e} : lingerie
{i> : sieve
{ε} : friend
NOTE: Allied, applied, dried, etc., should be treated
as ally + ed, apply + ed, dry + ed, etc. Like-
wise bodied, candied, studied, etc., should be
treated as body + ed, candy + ed, study +
ed, etc.

oa

Major Correspondences
{o} : approach, boast, goal, shoal
118 VOWEL DISTRIBUTIONS

Minor Correspondences
{0} : broad

oi/oy

Major Correspondences
{01} : boy, join, loiter, ouster

Minor Correspondences
{ai} : coyote
{1} : chamois
{9} : mademoiselle, porpoise, tortoise
{wa} : valo/s

00
Major Correspondences
{u} : boot, broom, proof, tycoon

Minor Correspondences
{9} : blood, flood
{0} : brooch
w : pooh
Μ : book foot look shook wood
brook forsook mistook soot wool
cook good nook stood
crook hook rook took

ou/ow

Major Correspondences
{au} : (initial and medial positions): abound, crown, mown-
tain, owl
{0} (final position): arrow, bow, glow, pillow
Minor Correspondences
{a} : clough double slough trouble
country enough sough young
couple joust touch
cousin rough tough
 VOWEL DISTRIBUTIONS 119

{a} : hough.
{0} : cough, trough
{0} : bowlder, bowl, cantaloup, mould, moult, owe, own,
poultice, showider, smoulder, sowl, thorough
{u} : boulevard, could, foulard, should, would
{u} : accouter caribou group routine troubadour
bayou cougar rendezvous soup uncouth
boudoir coupon rouge souvenir vermouth
boule croup roulette stoup wound
bouquet ghoul route toucan
{au} : allow, brow, cow, endow, how, mow, now, plow, sow
prow, thou, vow

ui

Major Correspondences
{(j)u) : bruise, nuisance, pursuit, suitor
Minor Correspondences
{1} : biscu/t, buzld, circuit
 VIII

CONCLUSIONS

1. THE NATURE OF THE CURRENT ORTHOGRAPHY

a. Base Forms and Compounds

The spelling of the base form of a word tends to be phonemic —

not in the one-letter one-sound system that has become the Holy
Grail of many educators and linguists, but in a more graphemically
economical fashion whereby position, environment, and overt
markers allow the same symbol to perform several distinct func-
tions, and whereby several symbols represent the same sound.
(That homo sapiens is somehow more at ease with a one-letter
one-sound system has often been assumed, but no evidence has
ever been produced to substantiate this limitation on man's mental
capacities).
The spelling of compounds and derivatives tends to be mor-
phemic; the established graphemic form of the base is retained
as much as possible, regardless of the phonemic alternations in-
volved. Once, for example, the base form melody is established
for /msbdi/, it would be highly irregular to spell /malodias/ in any
other way except with an initial segment melodi-. (The shift from
y to i has no relation to the pronunciation; it stems from purely
orthographic considerations). The orthographic preservation of
morphemic identity is predicted on the assumption that the reader
knows the phonemic alternations that accompany derivational and
inflectional formations. The form sign- is spelled the same in sign,
signing, signal, and signify, yet it has the different pronunciations
 CONCLUSIONS 121

/sain-/ and /sign-/. For the foreigner this is an obstacle towards

learning to pronounce English from its spelling, but English spell-
ing is geared for the convenience of the native speaker, not for
the foreigner.
Even with these morphemic tendencies, the orthography em-
ploys numerous phonemic aids in forming derivational and in-
flectional forms. The doubling of the final consonant in forms
like can: canning, confer: conferred, and the insertion of k in
picnicking facilitate letter-to-sound conversion. The phonemic and
the morphemic tendencies are equally strong; morphemic identity
is seldom preserved at the expense of the more general phonemic
patterns, and the general phonemic patterns, from all appearances,
tend to preserve morphemic identity.
Rules for pronouncing English monosyllables are relatively
simple; those for pronouncing polysyllabic but monomorphemic
words, more complex; and those for polymorphemic words, the
most complex. Spellings for monomorphemic words, while gen-
erally more phonemically based than those for polymorphemic
words, still are not completely phonemic. Stress, form class, and
phonotactics, among others, are conditioning factors. With poly-
morphemic words, even graphotactics becomes important. For
example, the difference between the spelling ie in believe and in
applied is readily apparent if one knows the graphotactical rules
for changing final y to i before certain suffixes. Without a knowl-
edge of this rule, however, one must view the ie spelling in applied
as irregular.

b. Loan Words and Homophones

More irregular spellings in English are due to borrowings than to

any other cause. Yet such borrowings cannot be classed as en-
tirely irregular since their spellings often mark their foreign iden-
tities. Foreign spellings have been retained in English orthography
since the Middle English period. For example, /δ/ changed to /§/
in French in the thirteenth century although French orthography
retained the ch spelling. At the time when the first French words
122 CONCLUSIONS

with ch corresponding to /s/ were borrowed in English, the native

orthography was still dynamic, yet English sh was not generally
substituted for French ch. Some French spellings were altered,
but seldom without prompting by a parallel alternation in French
orthography.
In addition to leaving intact foreign spellings that contrasted
with native ones, Middle and early Modern English scribes also
tended to alter English spellings to reflect foreign origins. Thus,
m a n y / ' s were replaced by pKs and initial Κs added to exhibit
French and Latin parentage. Occasionally an obsolete or foreign
spelling was retained to differentiate forms with the same pronun-
ciations.
This graphemic differentiation of homophones was noted as an
essential feature of English orthography as early as the beginning
of the seventeenth century. (See Hume, p. 19.) To what extent
conscious efforts have been made to enforce this orthographic
function is difficult to determine. Nevertheless, the number of
homophones that are now differentiated graphemically is con-
siderable. For an extensive list of such forms, see Robert Bridges,
"On English Homophones", SPE Tract No. 2 (London, 1919).

2. SPELLING REFORM

A rational approach to spelling reform must recognize the various

phonological, morphological and syntactical patterns in the current
orthography, and must increase either the regularity of the existing
patterns or the range of one group of patterns at the expense of
others. To base spelling reform upon the argument that orthog-
raphy should by nature be phonemic, morphemic, or anything
else is both unrealistic and unsupportable. There is no valid basis,
either diachronic or synchronic, for claiming that the current
orthography should be anything in particular other than what it is.
Some people may desire that it be phonemic or morphemic, but
this is somewhat different from the claim that the orthography,
by nature, should be that way.
 CONCLUSIONS 123

To argue that the existing orthography is irregular and then to

propose a phonemic alphabet for English as a cure is to present
a non-sequitur. The existing irregularities are in the syntactic and
morphological patterns as much as in the phonological ones, so
a phonemic alphabet, while presumably correcting the phono-
logical deviations, creates even greater irregularities in the other
patterning systems.
Certain spelling changes can be made without altering the basis
of the existing system, while others require entirely new orthog-
raphic principles. Eliminating the silent letters in debt, doubt, and
subtle, for example, would make the current system more syste-
matic. But to spell the first vowel in both sane and sanity with
different letters would change a basic morphological pattern.
Even in the attempts to institute a phonemic system, varying
degrees of change are required. One letter for one sound proposals
would completely alter the existing system. The same results,
obtained through an extension of the current phonemic patterns
like the final e pattern and consonant gemination, coupled with
a few new letters, could cause a much smaller change, and would
preserve a large number of the morphemic patterns.

3. THE TEACHING OF READING

a. Regular and Irregular Correspondences

In view of the foregoing conclusions, some notice must be given

to the concept 'regular spelling-to-sound correspondences' which
is paraded quite frequently through the current literature on read-
ing and orthography. Educators claim that the earliest reading
instruction should be built around such correspondences. Psy-
chologists report the results on children of various feedings of
these entities, and linguists have been prone to dismiss the entire
orthographic system with a single flap of the regular-irregular
banner. Since prior to the past few years no extensive analysis of
the current orthography has been available, it is difficult to im-
124 CONCLUSIONS

agine what the concept of regularity was based upon other than
upon a simple letter-to-sound view. Rather than thrash through
the few definitions which have been offered, I intend to analyze
anew this concept, based upon the material presented in this book.
Regularity implies, in some sense, a numerical predominance of
one event over others. To claim that by some reckoning the cor-
respondence a b is regular requires that clearly definable entities
be counted. Exactly what criterion level must be reached to have
regularity cannot and need not be answered. Certainly, if a-+b
occurs in 99 percent of all correspondences for a, then it would be
labeled as regular. But if it occurs in 51 percent, or 60, or 70, or
even 80 percent, an arbitrary decision would have to be made. But
this is not a problem unique to spelling-to-sound correspondences,
but to the definition of the term regular, and need not be a concern
at present.
The true problem faced here is in deciding what to count, c, for
example, corresponds to /k/ in approximately 74 percent of its
correspondences, to /s/ in approximately 22 percent, and to fsj in
approximately 4 percent. In addition, it corresponds to ft/ in two
different words. Suppose that we had arbitrarily decided on a
criterion level of 73 percent for regularity. Then, c /k/ would
be regular and all the other correspondences would be irregular.
But this is patently absurd since it is clear that c corresponds to
/k/ in one set of environments, to fsj in a different set, and to /§/
in a totally different set. That is, by considering environment
alone, three of the four correspondences for c, accounting for all
words containing c except two, become regular.
If, however, environment is accepted as a variable in the deter-
mination of regularity, the problem then becomes "what is an
admissable environment?" Consider as a starting point the prob-
lem of palatalization which was discussed earlier. To predict
palatalization, the word stress pattern must be known. For ex-
ample, medial d before u corresponds to /d/ in about 60 percent
of the words in which this sequence occurs and to /]/ in the re-
maining occurrences. Without considering stress, d-+/j/ would
 CONCLUSIONS 125

be classed as irregular. If, however, stress is considered as part of

the environment for d, then both d->/dJ and d-*•/}/ are regular.
Morphology is also significant, especially for the designation of
spelling units. Are, for example, the correspondences for medial
t and h in hothouse to be considered irregular correspondences for
the unit th or regular correspondences for the separate units t and
A? In all of these examples the question of regularity cannot be
resolved adequately without reference to the underlying patterns
of the orthography. But considerations of such patterns do not
automatically determine regularity. Before unstressed /ju/, /d/
palatalizes, as do /t/, /s/ and /z/, e.g., credulous, creature, erasure
(also -2-), and azure. Before unstressed /i/ or /j/ plus vowel, how-
ever, /d/ tends not to palatalize, but /t/, /s/ and jzj do. Thus,
custodian, medial, radiance, but bastion, appreciate, abrasion. In
relation to its immediate environment d regularly corresponds to
/d/ before unstressed i plus vowel, but in relation to the more
general pattern of palatalization, d is irregular here. These prob-
lems concern not only regularity, but also the concept of simplicity.
A description of spelling-to-sound correspondences that consid-
ers only direct letter-to-sound relationships without regard to
morphology or stress is certainly simpler than the model presented
in this paper, and is, in some sense correct, yet linguistically it is
inadequate since it fails to describe many phenomena which bear
upon spelling-to-sound relationships. In other words, simplicity
is secondary to linguistic adequacy, which implies that no con-
sideration of simplicity is meaningful until all relevant phenomena
are described. Then, simplicity enters only if more than one ex-
planation is adequate for all the data.
Regularity occupies a position similar in importance to that of
simplicity. Until we have discovered all existing patterns we cannot
be sure that any one pattern is regular or irregular; that is, until
all countable entities (patterns) are uncovered, counting is mean-
ingless. There are obviously points in an investigation where
supportable statements about regularity and simplicity can be
made, but all such statements must be tentative. Whether or not
such statements ever become definite in an investigation of natural
126 CONCLUSIONS

language is a question beyond the scope of this paper.

In the present study the primary goal has been to discover and
describe the underlying patterns of the current orthography. The
discovery of patterns is itself based upon predictability so that
ultimately the regular spelling-to-sound correspondences are those
that are predictable from orthographic patterns. In the complete
description, unpredictable spelling-to-sound correspondences must
be overtly marked so that predictable correspondences, which are
the majority, will be the residue. The alternative to this procedure
is to list for each pattern all words which fit the pattern, so that the
residue would consist of words with irregular correspondences.
But this is, by definition, more complex than the first procedure
since more words must be listed. In addition, this approach re-
quires that words added to English be listed under each regular
pattern which they contain, rather than under the irregular pat-
terns, which tend to be in the minority.
In place of the categories, REGULAR, and IRREGULAR, the follow-
ing classification scheme for spelling-to-sound patterns is offered:
I. Predictable: patterns that can be predicted upon the basis of
regular graphemic, morphemic or phonemic features of the
words or sentences in which they occur.
A. Invariant: patterns which admit no (or very few) variations
or exceptions.
B. Variant: patterns which have predictable variations or ex-
ceptions. (Variant patterns could be divided further on the
basis of the features needed to predict each pattern).
II. Unpredictable: all patterns which do not fit into I above.
A. Affix-aided: patterns which could be derived by relating
the word to one of its prefixed or suffixed forms, e.g., sign-
signal.
B. High-frequency: occurs frequently (frequent enough to al-
low an association group to be profitably employed in
teaching).
C. Low-frequency: occurs too infrequently to merit the forma-
tion of an association group.
 CONCLUSIONS 127

The importance of this classification for the teaching of reading is

that it separates patterns according to the pedagogy which can be
employed to teach them. All predictable patterns could be taught
by rules — that is, through reasoning — although this may not be
the best technique for some of them. Furthermore, the differences
between the invariant and variant predictable patterns can be
related to the teaching of a set for invariance as against teaching
a set for variance. Presenting all the invariant patterns before
introducing any of the variant ones may interfere with the teaching
of the latter through reinforcement of this set for invariance.
In the unpredictable class, the low-frequency patterns should not
be presented as letter-sound or sound-letter patterns since this may
encourage transfer to inappropriate situations. The high-frequency
patterns cannot, in a strict sense, be transferred either, since there
is no way to predict where they apply. However, by associating
the words in which a particular pattern occurs, an extra measure
of learning efficiency might be gained.
From this classification for letter-sound and sound-letter pat-
terns, a three-fold classification for reading and spelling words is
derived. (The word groups for reading will rarely be the same as
those for spelling).
Class I-Transfer words: Words which contain predictable patterns.
The patterns in these words can be transferred to the pronuncia-
tion (or spelling) of other words in which the same spellings
(or pronunciations) occur.
Class II-Association words: Words grouped according to frequent-
ly occurring, but unpredictable patterns.
Class Ill-Isolated words: Words which should be handled as whole
words to inhibit transfer of unpredictable, low frequency patterns.

b. Pedagogy

English spelling-to-sound correspondences can be described in

terms of patterns of graphemic, morphemic, syntactic, and phono-
tactical processes. With these patterns a complete model can be
constructed for relating spelling-to-sound, a model from which
128 CONCLUSIONS

the rules needed to predict the pronunciation of any word can be

found. What relationship this model has to the reading process
of literates or to the teaching of reading is at present unknown. It
is safe to predict that many of the patterns in the model have no
counterparts in reading habits, yet this does not say that there is no
relationship between the linguistic constructs and the reading
process. As a minimum it can be assumed that the classification
scheme for spelling-to-sound patterns presented above could be
used as a basis for the selection of words for the teaching of reading.
How the frequency of occurrence of various words in running text
is to be weighed in the selection of reading words is a problem for
the psychologists and educators to decide.
Another area in which the linguistic patterns could be helpful
is in the selection of the age level for introducing various corre-
spondences. If it is assumed that at the time a child is first taught
to read he has mastered all of the patterns of English which are
important for reading, then one could safely introduce any existing
spelling-to-sound correspondence. But it is doubtful that this is
true. While a first grade child may know most or all of the phono-
logy and syntax of English, he probably lacks some of the morpho-
phonemics. Since experimental evidence is lacking for determining
the age of comprehension (and production) of much of the mor-
phophonemics of English, consideration should be given to the
language features upon which various correspondences are based.
To introduce, for example, the alternation of the voiced velar stop
with zero in strong!stronger before the child recognizes or produces
such an alternation, is probably not desirable. Many of the pat-
terns discussed in this paper may have little bearing upon the
initial stages of reading instruction, but might be helpful in later
stages where morphology and syntax become more important. The
various pronunciations of final -ate and initial th- are examples
of such patterns.
In the translation from spelling-to-sound, words must in some
sense be scanned. Whether the scanning is through eye movements
or cognitive processes is immaterial. What is important is that
except where whole words are recognized, a sequence of units
 CONCLUSIONS 129

within the word is observed in the translation process. The rele-

vance of this to the teaching of reading is in the instructions given
to a child who is first learning to read. Should he be told to scan
left to right, letter by letter, pronouncing as he goes, or is there a
more efficient scheme? In the first place, a person who attempts
to scan left to right, letter by letter, pronouncing as he goes, could
not correctly read most English words, \ l a n y of the English
spelling-to-sound patterns require, at a minimum, a knowledge of
succeeding graphemic units. How, for example, is initial e- to be
pronounced if the following units are not known (cf. erb, ear,
ewer, eight)! This is just the beginning of the problem. In some
patterns, the entire word must be seen — and this is true of almost
all polysyllabic words since stress patterns are significant for vowel
quality. The implication here is that single pass left-to-right
scanning is unproductive except for some monosyllabic words.
Whether the reader experiments with several stress patterns and
then selects the most English-like one or whether he utilizes some
other scheme is not known, but should be determined.
 BIBLIOGRAPHY

Bazell, Charles E., "The Grapheme", Litera 3 (1956), 43.

Bethel, John P., ed., Webster's New Collegiate Dictionary, 2d ed. (Springfield,
Mass., 1956).
Bloomfield, Leonard, Language (New York, 1933).
, "Linguistics and Reading", The Elementary English Review XIX (1942),
125-30, 183-86.
, and Barnhart, Clarence K., Let's Read: a Linguistic Approach (Detroit,
1961).
Bopp, F., Uber das Konjugationssystem der Sanskritsprache (Frankfurt am
Main, 1916).
Bradley, Henry, "On the Relations between Spoken and Written Language
with Special Reference to English", British Academy 6 (1914), 211-32.
Bridges, Robert, "On English Homophones", SPE Tract No. 2 (London, 1919).
Brown, Goold, The Grammar of English Grammars, 4th ed. (New York, 1859).
Bullokar, William, Booke at Large, for the Amendment of Orthographie for
English Speech (London, 1580).
Burchfield, R. W., "The Language and Orthography of the Ormulum MS",
Trans. Phil. Soc. (1956), 56-87.
Butler, Charles, The English Grammar (Oxford, 1634). Re-issued with an
introduction by Albert Eichler (Halle, 1910).
Craigie, William Α., English Spelling, Its Rules and Reasons (New York, 1927).
"Some Anomalies of Spelling", SPE Tract No. 59 (London, 1942).
, "The Critique of Pure English from Caxton to Smollett", SPE Tract
No. 65 (London, 1946).
Danielson, Bror, John Hart's Works on Orthography and Pronunciation (Stock-
holm, 1955).
Dobson, Eric J., English Pronunciation 1500-1700, 2 vols. (Oxford, 1957).
Downing, John, "Pitman's Initial Teaching Alphabet", Paper read at the con-
ference on perceptual and linguistic aspects of reading (Stanford, California,
October 31, 1963).
Edgerton, William F., "Ideograms in English Writing", Lg. 17 (1941), 148-50.
Ellis, Alexander ., "On the Diphthong 'oy"\ Trans. Phil. Soc. (1867), Supp.
I, pp. 53-66.
Elphinston, James, On the English Language (London, 1796).
 BIBLIOGRAPHY 131

Ewert, Alfred, The French Language (London, 1933).

Flom, George T., "Studies in Scandinavian Paleography", J EGΡ 14 (1915),
530-43.
Francis, W. Nelson, The Structure of American English (New York, 1958).
Franklin, Benjamin, "A Scheme for a New Alphabet and Reformed Mode of
Spelling", in John Bigelow, ed. The Works of Benjamin Franklin, 12 vols.
(New York, 1904). Vol. V, pp. 30-37.
Fries, Charles C., The Teaching of English (Ann Arbor, 1949).
, Linguistics and Reading (New York, 1963).
Gaines, H. F., Cryptanalysis (New York, 1939).
Gibson, E. J., et al., "An Analysis of Critical Features of Letters, Tested by
a Confusion Matrix" in Harry Levin, ed., A Basic Research Program on
Reading (Cornell University, 1963).
Gill, Alexander, Logonomia Anglica (London, 1619).
Gimson, A. C., An Introduction to the Pronunciation of English (New York,
1962).
Gleason, Η. Α., Jr., An Introduction to Descriptive Linguistics (New York,
1961).
Grimm, J., Deutsche Grammatik (Gottingen, 1819-1837).
Hall, Robert., r., Linguistics and your Language (Garden City, 1960).
, Sound and Spelling in English (Philadelphia, 1961).
, "Graphemics and Linguistics", Proceedings of 1962 annual spring meet-
ing of the American Ethnological Society, pp. 53-59.
Haugen, Einar, First Grammatical Treatise: the Earliest Germanic Philology,
Language monogram No. 25 (Baltimore, 1950).
, "Directions in Modern Linguistics", Lg. 27 (1951), 211-22.
Hockett, Charles F., A Course in Modern Linguistics (New York, 1958).
, "Analysis of Graphic Monosyllables", Harry Levin, et al. A Basic Re-
search Program on Reading (Cornell University, 1963).
Holmberg, Borje, James Douglas on English Pronunciation c. 1740. Lund
studies in English XXVI (Lund, 1956).
Hume, Alexander, Of the Orthographie and Congruitie of the Britian Tongue.
Edited with an introduction by Henry B. Wheatley, EETS publication No.
5 (London, 1865).
Jespersen, Otto H., A Modern English Grammar on Historical Principles.
7 vols. (Heidelberg, 1909-1949).
, Essentials of English Grammar (New York, 1933).
Jones, D., Outline of English Phonetics, 3rd ed. (Leipzig, 1932).
Jonson, Ben, The English Grammar (London, 1634).
Kenyon, John S., American Pronunciation, 9th ed. (Ann Arbor, 1943).
Kenyon, John S., and Knott, Thomas, A Pronuncing Dictionary of American
English (Springfield, Mass., 1951).
Kingdon, Roger, The Groundwork of English Stress (London, 1958).
Kittredge, George L., "Some Landmarks in the History of English Grammars",
Text-book bulletin for schools and colleges, 1906 (New York).
Krapp, G. P., The Pronunciation of Standard English in America (New York,
1919).
132 BIBLIOGRAPHY

Kurath, Hans, "The Loss of Long Consonants and the Rise of Voiced Frica-
tives in Middle English", Lg. 32 (1956), 435-45.
, A Phonology and Prosody of Modern English (Ann Arbor, 1964).
, and McDavid, Raven I., Jr., The Pronunciation of English in the Atlantic
States (Ann Arbor, 1961).
Lehiste, Ilse, Accoustical Characteristics of Selected English Consonants, IJAL
publication No. 34 (Bloomington, 1964).
Levin, Harry, et al., A Basic Research Program on Reading, Final report,
Cooperative research project No. 639 (Cornell University, 1963).
, editor, Project Literacy Reports (Ithaca, 1964-67).
Lily, William, Short Introduction of Grammar (London, 1549).
Lounsbury, Thomas R., English Spelling and Spelling Reform (New York,
1909).
Mcintosh, Angus, "The Analysis of Written Middle English" Trans. Phil.
Soc. (1956), 26-55.
McLaughlin, John C., A Graphemic-Phonemic Study of a Middle English
Manuscript (The Hague, 1963).
Marchand, Hans, The Categories and Types of Present-Day English Word-
Formation (Wiesbaden, 1960).
Marckwardt, Albert H., "Origin and Extension of the Voiceless Preterit and
the Past Participle Inflections of the English Irregular Weak Verb Conjuga-
tions", University of Michigan Publications: Essays and studies in English
and comparative literature, Vol. XIII, pp. 151-328 (Ann Arbor, 1935).
Middle English Dictionary. Edited by Hans Kurath (Ann Arbor, 1954).
Mulcaster, Richard, The First Part of the Elementarie Which Entreateth Che-
felie of the Right Writing of Our English Tung [1582] Edited with an intro-
duction by Ε. T. Campagnac (Oxford, 1925).
Newman, Stanley S., "English Suffixation: A Descriptive Approach", Word 4
(1948), 24-36.
The Oxford English Dictionary, 12 vols. Edited by James A. H. Murray, et al.
Corrected re-issue (Oxford, 1933).
Paul, H., Prinzipien der Sprachgeschichte (Halle, 1880).
Peterson, Gordon E. and Lehiste, Ilse, "Duration of Syllable Nuclei", J ASA,
32 (1960), 693-703.
Pitman, Isaac J., "Learning to Read: an Experiment", Journal of the Royal
Society of Arts, 109 (1961), 149-180.
Pratt, Fletcher, Secret and Urgent; the Story of Codes and Ciphers (Indiana-
polis, 1939).
Pulgram, Ernst, "Phoneme and Grapheme: a Parallel", Word 7 (1951), 15-20.
, "Graphic and Phonic Systems: Figurae and Signs", Word 21 (1965),
208-224.
Rask, R. K., Undersegelse om det Nordiske Eller Islandske Sprogs Oprindelse
(Copenhagen, 1818).
Saussure, Ferdinand de, Cours de linguistique generale (Paris, 1916). Trans-
lated by Wade Baskin, Course in General Linguistics (New York, 1959).
Stetson, Raymond H., "The Phoneme and the Grapheme", Melanges -de lin
guistique et de phonologie offerts ä Jacq. van Ginneken (Paris, 1937), 353-56.
 BIBLIOGRAPHY 133

Sweet, Henry, "The History of the 'th' in English", Trans. Phil. Soc. (1868-
1869), 272-88.
, A History of English Sounds from the Earliest Period (Oxford, 1888).
Thorndike, Edward L., The Teaching of English Suffixes (New York, 1941).
, Thorndike-Century Senior Dictionary (New York, 1941).
Vachek, Josef, "Some Remarks on Writing and Phonetic Transcription", Acta
Linguistica 5 (1945-1949), 86-93.
, "Two Chapters on Written English", Brno Studies in English, Vol. 1
(Praha, 1959).
, "On the Interplay of External and Internal Factors in the Development
of Language", Lingua 11 (1962), 433-48.
Venezky, Richard L., "A Computer Program for Deriving Spelling to Sound
Correlations", Masters thesis (Cornell University, 1962). Published in part
in Harry Levin, et al. A Basic Research Program on Reading (Cornell Univer-
sity, 1963).
, "A Study of English Spelling-to-Sound Correspondences on Historical
Principles". Unpublished doctoral dissertation (Stanford University, 1965).
, "English Orthography: its Graphical Structure and its Relation to
Sound", Reading Research Quarterly, II, 75-106 (Spring, 1967).
, "Reading: Grapheme-Phoneme Relationships", Education 87 (1967),
519-24.
, "The Basis of English Orthography", Acta Linguistica 10 (1967), 145-59.
, and Weir, Ruth H., A Study of Selected Spelling-to-Sound Correspond-
ence Patterns, Final report, Co-operative Research Project No. 3090 (Stan-
ford, 1966).
Waldo, George S., "The Significance of Accentuation in English Words",
Proceedings of the Ninth International Congress of Linguists (The Hague,
1965), 204-10.
Webster, Noah, An American Dictionary of the English Language. Revised and
enlarged by Chauncey A. Goodrich (Springfield, Mass., 1852).
Webster's New Collegiate Dictionary. Edited by John P. Bethel, 2nd ed.
(Springfield, Mass., 1956).
Webster's Third New International Dictionary. Edited by Phillip B. Gove
(Springfield, Mass., 1963).
Weir, Ruth H., "Formulation of Grapheme-Phoneme Correspondence Rules
to Aid in the Teaching of Reading", Final report, Co-operative Research
Project No. S-039 (Stanford, 1964).
, and Venezky, Richard L., "Rules to Aid in the Teaching of Reading",
Final report, Co-operative Research Project No. 2584 (Stanford, 1965).
Whitney, William D., Language and the Study of Language (New York, 1967).
Wijk, Axel, Rules of Pronunciation for the English Language (London, 1966).
Zachrisson, Robert E., Anglic, a New Agreed Simplified English Spelling (Upp-
sala, 1930).
, "Four Hundred Years of English Spelling Reform", Studia Neophilo·
logica 4 (1931), 1-69.
 INDEX OF TERMS

affix-aided correspondences, 126 invariant correspondences, 39

alphabetic principle, 16 irregular correspondences, 39ff.,
allograph, 47 123 ff.
assibilation, 92 isolated words, 127
association words, 127
/-influence, 112
checked vowels, 14, 35-36 letter, 50 fn.
cluster, 63 ff. long vowels, 14fn., 35-36
compound consonant units, 51, 54
consonant units, 35-36, 51 major correspondences, 115
major relational units, 54
figura, 16 marker, 50
final e pattern, 104 minor correspondences, 115
form class, 43-44 minor relational units, 54
free vowels, 14, 35-36 morphemic spelling, 42
functional units, 50 morphophonemic alternations, 63,108
functionally compound units, 35-36, morphophonemic level, 46
51
functionally simple units, 35-36, 51 nomen, 16
functors, 44 non-productive alternations, 64ff.

geminate consonant clusters, 51, 57, palatalization, 40, 92-93

106fn. petrified alternations, 99
geminate consonant pattern, 104 potestas, 16
grapheme, 49-50, 50fn. predictable correspondence, 126
grapheme-phoneme parallel, 26, 47 ff. primary vowel spellings, 54, 101
graphemic allomorph, 42 productive alternations, 64 fn.
graphemic alternations, 37-38, 59if.
graphemic substitutions, 37 regular alternations, 94ff.
graphotactical patterns, 50, 57 regular correspondences, 39ff., 123ff.
relational unit, 50, 54
homophones, 122 r-influence, 109

Initial Teaching Alphabet, 33 secondary vowels, 54, 114

 INDEX OF TERMS 135

short vowels, 14fn., 35-36 transfer words, 127

silent letters, 35, 52-53
simple consonant units, 51, 54 unpredictable correspondences, 126
spelling pattern, 29 variant correspondences, 39
spelling-to-sound model, 45-46
spelling units, 34 ff. H'-influence
 WORD INDEX

A addle, 104
abbess, 107 adhere, 111
abduct, 65 adieu, 62
abet, 106fn. adobe, 105fn.
abetted, 106fn. adore, 111
abeyance, 116 adorn, 111
able, 22, 53, 56 adz, 68, 90
abound,118 aesthetic, 55
above, 38fn., 106 again, 115
abrasion, 125 against, 115
absentee, 116 age, 56
absolve, 97 aggrandize, 72
abuse, 96 aghast, 35, 73
accede, 40, 89fn., 105 agnostic, 71, 76
accent, 97, 107 ah, 56, 73
accentuate, 97 aisle, 83, 115
accept, 40, 89fn., 96, 97 alarm, 111
acceptable, 97 alarum, 111
access, 97 albatross, 107
accessible, 97 alchemy, 67
accident, 97 algae, 55
accidental, 97 alibi, 59
accommodate, 107 align, 113
accompt, 77 alkali, 59, 112fn.
account, 91 allege, 105
accouter, 119 allied, 117
ache, 36, 51, 67 allow, 61, 119
achieve, 117 ally, 117
acknowledge, 76 almanac, 112
acquire, 86 alternation, 112
acre, 53, 56 altitude, 112fn.
across, 107 alumni, 59
add, 42, 57fn., 106fn. anal, 36, 57, 104
addict, 106fn. anarchy, 67
 WORD INDEX

anchor, 42, 67 assiduity, 93

and, 78 assiduous, 93
anemia, 60 assign, 113
angel, 109 assignation, 22
angelic, 109 assignee, 22
anger, 71 assuage, 45, 54, 60
anguish, 60 assume, 108
annals, 36, 57, 104, 106fn. assumption, 108
annihilate, 40 asthma, 85
answer, 87, 88 asylum, 59
ant, 78 athlete, 102, 109
antenna, 103 athletic, 102, 109
anti, 59 attempt, 84
antilogism, 78 attorney, 112
anxiety, 62 auction, 37, 61
appellate, 107 audience, 115
applaud, 61 audit, 61
apple, 103, 107 auger, 61, 71
applied, 117, 120 aunt, 116
apply, 117 austerity, 111
appreciate, 125 autumn, 43
approach, 117 autumnal, 43
apropos, 42, 83 autumns, 43
apt, 84 award, 87
arc, 41, 56fn., 65 aware, 87
arch, 66 away, 15
archaeology, 67 awe, 37, 61
archaic, 67 awkward, 61
archangel, 67 awl, 61
archetype, 67 awning, 61
architect, 67 axe, 51, 53
architecture, 67 axle, 104
archive, 67 aye, 115
arcing, 41, 56 fn, 65 azure, 93, 125
are, 105, 111
argue, 52 Β
arguing, 52 babbittry, 107
aria, 111 bacchanal, 91
arid, 111 bacillus, 107
arm, 77 bade, 105
arrogate, 111 badge, 51, 103
arrow, 118 bag, 70
artillery, 107 baggage, 72
as, 83 baggy, 72
asinine, 106, 108 bah, 73
asininity, 108 bait, 37, 115
ask, 75, 82 bake, 51, 52, 105
asp, 79 bamboo, 42, 78
138 WORD INDEX

bank, 78 blitz, 84, 90

barrage, 105 blizzard, 106fn., 107
basic, 57 blood, 118
basically, 57 bloom, 77
baste, 105 blossom, 106fn., 107fn.
bastile, 105, 105fn. blue, 51
bastille, 105 fn. board, 68
bastion, 40, 85, 125 boast, 117
bath, 48, 56, 86, 86fn., 94, 98, 100 bodied, 117
bathe, 48, 56 body, 117
baths, 86, 86fn., 94, 98, 100 bogey, 71
bay, 37 boisterous, 37
bayou, 61,115, 119 bold, 68, 112, 113
beard, 68 boll, 112
beau, 62 bollard, 112
been, 116 bomb, 35, 42, 44, 45, 65
beggar, 72, 107 bombard, 35, 65
begin, 71 bombardier, 35
behind, 113 bombed, 35
belch, 76 bombing 35, 42
belief, 70, 94, 96, 100 bombs, 35
believe, 70, 94, 96, 100, 120 book,118
bell, 112 boot, 118
belle, 53, 54 borough,111,114
bellow, 107, 112 borrow, 111
bend, 78 boss, 107
benign, 113 boudoir, 119
bereave, 70 bough, 114
bereft, 70 bought, 114
berg, 81 boulder, 119
bevel, 107fn. boule, 119
bewail, 87 boulevard, 119
beware, 54, 87, 111 bouquet, 119
beyond, 55, 89 bourgeois, 71, 83
bijou, 61, 75 bow, 118
bill, 112 bowl, 119
billiards, 58 bowyer, 89
billion, 112 boy, 37, 55, 118
bind, 113 boycott, 106fn.
birch, 81 boys, 42, 95
bird, 81, 110, 111 brace, 81
birth, 85 braggart, 72
biscuit, 119 braille, 106
black, 64, 76, 90 brake, 64
blackguard, 57 brass, 96
bleed, 116 braze, 96
blight, 113 breach, 116
blind, 113 bread, 116
 WORD INDEX

breadth, 116 calorie, 117

break, 116 calves, 98
breast, 116 calx, 77
breath, 56, 69, 86, 94, 96, 98, 99, 116 camouflage, 71, 105
breathe, 53, 56, 69, 86, 94, 96, 99 can, 120
breaths, 98 candied, 117
breeches, 116 candy, 15fn., 117
brewer, 61 canine, 103, 115
brief, 69 fn., 96 canned, 42
broad, 118 canning, 120
broccoli, 59, 106fn. cantaloup, 119
broccolo, 59 canyon, 89
brooch, 118 capitulate, 93
brook,118 caprice, 106
broom, 118 careless, 42
brother, 107fn. caribou, 61, 119
brow, 119 carve, 111
bruise, 119 case, 21
buffet, 85 castle, 85
bug, 70 catalogue, 51, 56
build, 119 catarrh, 82, 111
bulb, 64, 77 catechism, 67
bulk, 76 cats, 42, 94, 95, 99
bull, 112 caught, 114
bullet, 112 causation, 115
bureau, 55, 62, 111 cause, 115
burgh(er), 73, 73 fn. cayenne, 115
burr, 106fn., I l l ceiling, 117
bury, 111 cell, 18
butt, 106fn. cellar, 107
butte, 107 cellini, 41
button, 106fn., 107 cello, 41, 66
buy, 55 center, 38
buzz, 106fn. central, 38
bye, 53, 59 cereal, 111
chagrin, 67
C chair, 51
cabbage, 106fn., 107 chaise, 67
cabin, 22 chalet, 67
cache, 67 chalk, 77, 112
cafe, 105fn. challis, 67, 83
caffein, 117 chameleon, 67
cajole, 74 chamois, 67, 83, 118
calf, 69, 98 champagne, 67, 91
call, 112 chandelier, 67
calliper, 112 change, 48, 53, 56
callow, 107 changeable, 42, 56
calm, 77 changing, 56
140 WORD INDEX

chanty, 67 clean, 65
chaos, 67 cleanse, 82
chaperon, 67 cleave, 76
character, 67 cleft, 70
charade, 67 cliche, 67
charivari, 59, 67 clientele, 105
charlatan, 67 climb, 44, 76
chasm, 67, 77 close, 83
chassis, 67, 83 cloth, 56, 86, 96, 98
chaste, 105 clothe, 51, 53, 56, 86, 96
chateau, 62, 67 cloths, 98
chauffeur, 67, 106, 116 clough, 73, 114, 118
cause, 115 clover, 65
chauvinism, 67 clutter, 106fn.
chauvinist, 116 coal, 76
cheddar, 107 coarse, 58
chef, 67 cobble, 104
chemi-, 67 coeval, 78
chemise, 67 coffee, 106fn.
cheroot, 67 cognac, 50, 55
chevalier, 67 cognate, 103
chevior, 67 collapse, 58
chevron, 67 collar, 107
chic, 67 college, 56, 107
chicanery, 67 collide, 109
chicken, 103 collie, 117
chiffon, 67 collision, 109
child, 113 cologne, 53
chilli, 59 colonel, 77
chintz, 84, 90 coma, 104
chiropodist, 67 come, 15, 106
chivalry, 67 comma, 104, 107
chlorine, 66, 76 comment, 106fn.
choir, 67 commerce, 107
choler, 67 complex, 78
cholera, 67 compound, 61
choral, 67 comptroller, 77
chord, 67 computation, 108
chorus, 67 compute, 108
chrome, 66, 81 concede, 51, 109
churn, 78 conceit, 117
chute, 67 conceive, 117
circuit, 119 concern, 78
cirrhosis, 82 concerto, 41, 66, 112
city, 15fn., 52 concession, 109
clapboard, 44, 65 conclude, 78, 79 fn.
class, 96 concrete, 79fn.
claw, 115 conduce, 108
 WORD INDEX

conduction, 108 crochet, 67

cone, 102, 109 crook,118
confer, 120 croup, 119
conferred, 120 crowd, 62
confetti, 59 crown, 118
confetto, 59 crumb, 65
conger, 71 crumble, 65
congregate, 79 fn. crust, 65
congregation, 18 cube, 105
congress, 42, 79fn. cuisine, 61
congruence, 79 fn. cupboard, 65
congruous, 79 fn. cur, 111
conic, 102, 109 curb, 64
consign, 113 cure, 111
continue, 51, 57 curiosity, 37, 62
contralto, 112fn. curious, 37, 62
contrariety, 62 curl, 76, 81
council, 18 current, 111
counsil, 18 curse, 96
convene, 109 custodian, 125
convention, 109 cut, 48, 55, 104
cook,118 cute, 48, 55, 104
copper, 106 fn. cycle, 55, 105, 109
cordial, 40, 93, 99 cyclic, 109
corps, 79, 83 cynosure, 93
corpuscular, 66 czar, 66, 90
corpuscle, 66
corsage, 71, 105 D
cougar, 119 dagger, 72
cough, 73, 114, 119 daily, 59
could, 77, 119 damn, 43, 59
country, 118 damnation, 43
coup, 79 daughter, 114
couple, 119 dawn, 61
coupon, 119 day, 59, 115
cousin, 119 dead, 116
cove, 105 deaf, 116
cover, 107fn. death, 116
cow, 61, 119 debit, 65
coward, 37, 61 debris, 83
coyote, 105 fn., 118 debt, 32, 34, 45, 53, 64, 65, 123
crash, 81 debut, 85
crawl, 65 decade, 105
creature, 93, 125 deceit, 117
credulity, 93 deceive, 87, 117
credulous, 93, 125 deception, 87
creek, 116 decide, 109
cribbage, 107 decision, 109
142 WORD INDEX

defect, 78 doxology, 89fn.

defend, 69 drachm, 66
defense, 69 draught, 73, 114, 116
deflate, 44 draw, 37, 61, 68
deject, 74 dread, 116
delirium, 111 dregs, 68
dementia, 60 dress, 96
demesne, 83 dried, 117
demon, 77 drift, 70
dense, 58 drink, 78
depot, 85 drive, 70, 81
design, 83, 97, 113 drowse, 62
designing, 43 dry,117
dichotomy, 67 dubious, 103
die, 59, 117 duke, 105
diesel, 117 duplex, 78
dike, 75 duplicate, 44, 105
dilemma, 107 during, 52
diner, 104 dwarf, 69fn., 96, 99
dinghy, 73, 73 fn. dwell, 68
dinner, 104, 106fn. dwindle, 68
diphthong, 71 fn., 80 dying, 59
diphthongize, 71 fn.
Ε
τ—1

dire, 111
disc, 56 fn. each, 116
disced, 56 fn. eager, 71
discing, 56 fn. ear, 129
disk, 56fn. earth, 96, 98
dissolve, 83, 97 earths, 98
distinct, 89 ebb, 57fn., 64, 103, 106fn., 107
distinguish, 60 ebba, 64
dive, 38 echelon, 67
divine, 108 echo, 67
diving, 38 eclipse, 58, 96
divinity, 108 eddy, 106fn.
divvy, 87, 106fn. edge, 51, 53, 90
dizzy, 106fn. eel, 116
doe, 57 eerie, 117
dogs, 94, 99 egg, 57 fn, 76, 106fn.
doll, 107 eight, 129
dollar, 107 either, 117
dolphin, 112fn. elf, 69, 98
domicile, 105 elite, 106
done, 106 elm, 77
double, 118 elongate, 71 fn.
doubt, 32, 64, 65, 123 elongation, 71 fn.
dough, 114 else, 58
dove, 106 elves, 98
 WORD INDEX

embryo, 59 facade, 66, 105

empiric, 111 facsimile, 105fn.
endeavor, 116 fahrenheit, 74, 117
endow, 119 fall, 112
ennui, 60 fallacy, 107, 112
enough, 73, 114, 118 false, 21, 58
enquiry, 111 famine, 106
epistle, 103 feather, 116
epitome, 105 fn. fell, 112
epoch, 67 fellow, 107, 112
equation, 85 fetal, 36, 57
erasure, 125 fete, 105
erb, 111, 129 fetter, 57
ere, 105 fettle, 36
erode, 105 feud, 61
err, 106fn., I l l fifth, 70, 95, 99
errand, 111 fight, 113
error, 106fn. fill, 112
erst, 111 fillet, 106fn.
essential, 38 filly, 112
ether, 36, 85 finale, 105fn.
eucharist, 61, 67, 117 finally, 42
eulogy, 61 find, 68, 113
eunuch, 67 finger, 42, 71
ewe, 61, 117 first, 81, 84
ewer, 37, 61, 129 five, 70, 95, 99, 105
exaggerate, 72, 91 fizz, 106fn.
examine, 40, 106 flabby, 107
excerpt, 81, 84 flew, 37, 61
execute, 97, 108 flight, 113
execution, 108 flivver, 87, 106fn.
exercise, 40 flood, 69,118
exhale, 74 flower, 76
exhaust, 89 fn. flu, 62
exhibit, 89 fn. fluff, 106fn.
exist, 40, 97 foe, 57
exit, 40, 103 fogey, 71
expend, 69 fold, 113
expense, 69 folk, 112
extempore, 105fn. foot, 118
exterior, 111 for, 109
extinguish, 60 forbade, 105
extreme, 108 forbid, 106fn.
extremity, 108 forbidding, 106fn.
eye, 117 foreign, 111
forest, 110
F formula, 111
fable, 105 forsook, 118
144 WORD INDEX

fought, 72, 84, 114 gill, 71

foulard, 119 gilt, 71
four, 109 gimlet, 71
fowl, 62 gingle, 74
fraught, 114 giraffe, 53, 54, 106fn.
free, 69 gird, 71
freight, 114 girdle, 71
friend, 81, 117 girl, 71, 111
frigate, 44 girt, 71
frivolity, 108 girth, 71
frivolous, 108 give, 70, 71, 106
frog, 70 gizzard, 71
froth, 96 glass, 94, 99, 100
fuchsia, 66 glaze, 76, 94, 99, 100
fungi, 59, 71 glide, 70
funnel, 107 glimpse, 96
funny, 107 globe, 105
furlough, 114 glory, 110, 111
furrier, 111 glove, 57, 106
fuzz, 106fn. glow, 118
glue, 48, 57
G gnat, 40, 46, 70
gage, 105 gnaw, 61
gall, 112 gneiss, 117
gallow, 112 gnostic, 71, 76
gaol, 71 gnu, 62
garage, 71, 105, 120 goal, 117
gargle, 111 gone, 106
gauge, 116 good,118
gear, 71 goose, 57, 58
geese, 71 gossip, 106fn., 107fn.
geld, 71 grade, 81
generosity, 62 grammar, 107
generous, 62 grandmother, 68
gest, 74 grass, 96
get, 71 graze, 96
geyser, 60, 71, 117 grease, 96
ghastly, 73 great, 116
gherkin, 40, 73, 73 fn. grey, 37
ghetto, 73, 73fn. grief, 70, 96
ghost, 35, 40, 73, 73fn. grieve, 70, 96
ghostly, 35 grind, 113
ghoul, 40, 73, 73fn., 95, 119 gross, 107
giddy, 71 group,119
gift, 70, 71 grow, 70
gig, 71 guano, 86
giggle, 71, 72 guard, 52
gild, 71, 113 guess, 56
 WORD INDEX 145

guest, 51, 54, 56 hemp, 77

guide, 56 her, 111
guild, 56 herb, 64, 74, 81
guilt, 56 hers, 58
guise, 56 hierarchy, 67
gulf, 98 high, 113
gulfs, 98 hind, 113
gull, 112 hippo, 36, 57
gullet, 112 hire, 111
guy, 55 his, 58, 83
hitch, 103
gynaecology, 71
hoarse, 109
Η hobby, 107
half, 69, 70, 76, 77, 96, 98 hodge, 51, 57
hall, 112 hoe, 57
halleluia, 74 hog, 70
hallelujah, 56, 74, 75 hold, 76, 113
hallow, 112 hollow, 107
halt, 84 holstein, 117
halve, 70, 96 homonym, 78
halves, 77, 98 honest, 74
handful, 68 honor, 74
handkerchief, 68 hook, 118
hands, 68 hop, 38
handstroke, 68 hopped, 38
happen, 107 horrid, 111
happy, 106fn., 107 horse, 109
hark, 75, 85 hothead, 42
harmonic, 74 hothouse, 125
harp, 79 hough, 73, 114, 119
harsh, 84 hour, 74
has, 83 houri, 59
haste, 105 house, 14fn., 58, 69, 83, 94, 96, 100
hautboy, 85, 116 houses, 94, 95, 100
have, 48, 55, 57, 105 how, 119
havoc, 65 however, 61
head,116 hullabaloo, 107
health, 116 humane, 108
hearse, 58 humanity, 108
heaven, 116 hummock, 107
heavy, 116 hunger, 71
hedge, 36 hurrah, 56, 73, 74
heifer, 117 hurry, 111
height, 114, 117 hydraulics, 58
heinous, 37 hygiene, 109
heir, 74 hygienic, 109
helm, 76, 77 hyperbole, 105 fn.
help, 79 hypo, 36, 57
146 WORD INDEX

hypochondria, 67 John's, 43
hysteria, 60 join, 118
joke, 105
I jolt, 112
iamb, 65, 65 fn. joust, 119
iambic, 65 judge, 21
iciest, 59 judges, 42
icy, 59
ignite, 42 Κ
imbroglio, 70 kaffir, 75
impede, 105 kaleidoscope, 117
importune, 93 kangaroo, 75
impotent, 115 kennel, 107
impromptu, 62 key, 117
improvise, 58 khaki, 52, 55, 59, 91
include, 78, 79fn. khan, 55, 91
increase, 79 fn. kiln, 78
indict, 66 kind, 113
induce, 102 kitchen, 90
indubitable, 65 knee, 39, 40, 46, 76
induction, 102 knife, 39, 69, 98
infant, 78 knives, 98
infer, 111 know, 39
ingenious, 42 knowledge, 76
ingot, 79 fn. kosher, 36, 75
ingrain, 42 kulak, 75
ingratiate, 42
inn, 57fn., 106fn. L
instead, 116 labyrinth, 78, 85
institute, 93 ladle, 103, 104
intaglio, 70 lager, 71
integrate, 44 laid, 59
inveigh, 114 lamb, 76
iris, 111 land, 68
irony, 111 landslide, 68
irrigate, 111 language, 52, 54, 60, 86
is, 83 languid, 60
island, 83 lasso, 107 fn.
isle, 83 lath, 98
issue, 107 laths, 98
isthmus, 85 laugh, 114, 116
laughter, 73
J lawyer, 89
jaguar, 60 lay, 59
jazz, 106fn. leach, 76
jeopardy, 55 leaf, 69, 69fn., 96, 98, 99
jest, 74 leak, 76
jingle, 74 leather, 116
 WORD INDEX

leave, 70, 94, 100 lyric, 56fn., 65, 66

leaven,116 lyrical, 56fn.
leaves, 98, 99 lyricism, 56fn., 65, 66
left, 70, 84, 94, 100
leisure, 117 Μ
lens, 58 ma, 15fn.
leopard, 55 macaroni, 59
lesson, 107 machination, 67
letter, 42 machine, 67, 105
level, 107fn. mademoiselle, 118
lewd, 61 maggot, 72
lichen, 67 magi, 59
lie, 117 majesty, 74
lieu, 62 malaria, 60, 111
life, 69, 70, 96, 98 malign, 50, 78, 102, 108, 113
light, 113 malignant, 50, 102
limp, 77 malignity, 108
linger, 71 malinger, 71
lingerie, 71, 117 mall, 112
light, 113 mallard, 112
like, 29 malt, 76
lingual, 61 mamma, 106
linguist, 61, 86 mammal, 107
liquor, 86 man, 77
live, 70, 96, 106 man's, 42
lives, 98 margarine, 71
llama, 106fn. marihuanna, 75
loaf, 69, 98 marijuana, 60, 75
loaves, 98 marine, 106
lobby, 106fn., 107 marriage, 111
locus, 58 marry, 109
loge, 71 marsh, 81, 84
loggard, 72 marshmallow, 106
loiter, 118 Mary, 109
long, 35, 71 fn. mass, 96
longnette, 91 massacre, 56
look,118 massage, 71, 105
lose, 106 mat, 48, 55, 102
louse, 94, 100 match, 90
lousy, 94, 100 mate, 48, 51, 55, 102
love, 38, 38fn., 53, 55, 57, 106 mathematics, 58
lucre, 102, 103 matinee, 116
luggage, 72, 106fn. mature, 93, 111
luxation, 89fn. mauve, 116
luxurious, 89fn., 94, 97, 100 meadow, 116
luxury, 36, 89fn., 93, 94, 97, 100, measles, 104
103 measure, 116
lymph, 77 mechanic, 67
148 WORD INDEX

medial, 125 moon, 77

median, 103 moose, 57, 58
medieval, 78 morale, 105
mediocre, 105, 108 more, 111
mediocrity, 108 moron, 110
megalomania, 60 morpheme, 41
melancholy, 67 mortgage, 85
melee, 116 moth, 86 fn., 98
melody, 115, 120 mother, 107fn.
melodious, 42, 115 moths, 86fn., 98
menu, 62 mould, 119
merciless, 59 moult, 119
mercy, 59 mountain, 118
merry, 109 mouse, 57, 58
message, 79, 107 mouth, 86, 96, 98
messenger, 79 mouths, 86, 98
met, 55, 102, 104 move, 106
mete, 55, 102, 104 movie, 117
meter, 109 mow, 119
metric, 109 mufti, 59
microbe, 103, 105 mulct, 77, 84
midday, 42 mule, 105
midget, 90 mummy, 107
might, 72 mumps, 58
mignon, 91 mural, 111
mignonette, 91 muscle, 66
mild, 113 muscular, 66
milk, 75 music, 24, 66
mill, 106fn. musician, 66
million, 107 must, 84
mime, 109 mustache, 67
mimic, 109
mind, 113 Ν
miracle, 111 nation, 18, 38, 52, 85
mirage, 71, 105 naught, 114
mirror, 111 naughty, 114
mirth, 81 navvy, 87, 106fn.
mistook, 118 neck, 90
mnemonic, 78 nefarious, 111
moiety, 60 neighbor, 114
mold, 113 neither, 117
molt, 112 neurosis, 109
monarchy, 67 neurotic, 109
monger, 71 neuter, 61
monstrosity, 62 neutral, 115
monstrous, 62 neutrality, 115
month, 98 neutron, 117
months, 98 newt, 61
 WORD INDEX

niece, 117 oust, 61

night, 72 out, 61
nihilism, 40 owe, 119
nike, 105 fn. owl, 62, 118
noble, 103, 104 own, 62, 119
noise, 58 oyster, 60, 118
none, 106
nook, 118 Ρ
north, 85, 86, 94, 99 pa, 15fn.
northern, 86, 94, 99 pace, 53
not, 55 pact, 84
note, 55 palmetto, 112fn.
noticable, 52 panic, 56
notice, 52 panicky, 56
noticing, 52 parachute, 67
nought, 72, 114 paradigm, 42, 43, 70, 72, 113
now, 119 paradigmatic, 43, 72
noxious, 36 paradigms, 42
nugget, 106fn. paralytic, 109
nuisance, 119 paralyze, 109
number, 64fn., 65 paranoia, 60
numeric, 64fn., 65 pariah, 53, 54, 56
nurse, 96 parochial, 67
nymph, 77 parvenu, 62
pass, 96
Ο passage, 79
obey, 116 passenger, 79
obscene, 108 path, 86fn., 98
obscenity, 108 paths, 86fn., 98
obsolete, 105 patriarch, 67
occupy, 106fn. peace, 21, 51, 53
odd, 106fn. peasant, 116
of, 39 pebble, 104
often, 85 penal, 102
oh, 56, 73 penguin, 61
old, 113 pennant, 102
oleander, 76 pentateuch, 67
olfactory, 112fn. peril, 111
or, 111 persuade, 60, 86
orange, 110, 111 Peter, 104
oration, 18 petter, 104
orchestra, 67 pewter, 61, 117
orchid, 67 pharaoh 53, 54, 56, 74
orient, 110 phase, 41
origin, 110 pheasant, 116
other, 107 fn. philharmonic, 40, 74
ought, 114 phlegm, 80
ounce, 37, 61 phone, 109
150 WORD INDEX

phonic, 78, 109 power, 61

phonograph, 36 praise, 58
phrase, 80, 81 precocious, 108
phthisic, 80 precocity, 108
picnic, 41, 55, 56, 65 press, 96
picnicked, 65 prestige, 71, 106
picnicking, 41, 55, 56 presume, 93, 108
pie, 117 presumption, 108
piety, 62 prison, 107fn.
pillow, 106fn., 118 prize, 81, 105
pilot, 103 produce, 93, 105
pinch, 78 profile, 105
piquant, 86 program, 79
placate, 105 prohibit, 40
plague, 51, 54, 56, 57 prohibition, 40, 56
plaid, 115 prolongate, 71fn.
plait, 115 prolongation, 71 fn.
plateau, 55, 62 proof, 70, 96, 118
play, 76 propriety, 62
player, 115 protege, 71, 105fn.
pleasure, 116 prove, 70, 96, 106
plough, 79 provide, 109
plow, 119 provision, 109
pneumatic, 61 prow, 119
pneumonia, 40, 79 proximity, 89fn.
pocket, 103 psalm, 112
poignant, 50, 55, 91 psych-, 67
poise, 58 psychology, 40, 79, 82
police, 106 ptarmigan, 40
polka, 112 ptomaine, 79, 84
poll, 112 pudding, 107
pollen, 112 puddle, 107
pooh,118 pueblo, 60
poor, 87 pulp, 76
pop, 104 pummel, 107
pope, 104 purr, 106fn.
porosity, 62 pursuit, 93, 119
porous, 62 pussy, 107
porpoise, 118 putt, 106fn.
portmanteau, 62
possible, 102 Q
posy, 102 quack, 45, 54, 86
potato, 57 quadrant, 45
potent, 115 quality, 45
potpourri, 59 quantum, 45
poultice, 119 quartz, 90
pounds, 68 quasi, 59
poverty, 87 quay, 86, 115
 WORD INDEX

queen, 86 reward, 87
queue, 86 rheostate, 82
rhesus, 35
R rhetoric, 35
rack, 55 rheumatism, 61
radiance, 125 rhinoceros, 82
raft, 69 rhomb, 65, 65 fn.
raise, 58 rhombus, 65
rajah, 74 rhyme, 59
rank,75 rhythm, 59, 77, 78
raspberry, 65 richochet, 67
rat, 104 riffle, 104
ratchet, 36 rifle, 104
rate, 104 right, 113
ration, 102 rime, 59
rattle, 102, 107 rind, 113
ravine, 105 rise, 58, 102
ravioli, 59 risen, 102
raviolo, 59 river, 107 fn.
razz, 106fn. rob, 102
reach, 116 robe, 102
ready, 116 roll, 112
realm, 116 rollick, 112
receipt, 79, 117 roof, 99
receive, 87, 117 rook,118
reception, 87 rotor, 57
recipe, 105fn. rotter, 57
recluse, 58 rouge, 71, 119
reduce, 108 rough, 73, 123, 119
reduction, 108 roulette, 119
reef, 98 route, 119
reefs, 98 routine, 106, 119
regime, 71, 106 rove, 34
reign, 116 rubber, 107
relief, 70, 96 rubble, 104
relieve, 70, 96 ruble, 104
remind, 113 rudder, 102
rendezvous, 83, 127 rude, 102
renege, 105 rug, 103
rent, 84 run, 38, 106fn.
resign, 97, 113 running, 38, 106fn.
resolve, 97 rustic, 57
resound, 83, 94, 97, 99 rustically, 57
respond, 69 rye, 57, 59
response, 69
resume, 105 S
revolution, 87 sabotage, 71, 105
revolve, 87, 112fn. sac, 65
152 WORD INDEX

sachet, 67 sequoia, 60
saddle, 103 Sequoiah, 60
safari, 111 seraglio, 70
said, 115 serene, 105, 108
sane, 102, 108, 123 serenity, 108
sanguine, 61 serf, 69, 70
sanity, 102, 108, 123 sergeant, 74, 112
sapphire, 80, 80fn. serjeant, 74
sardine, 106 serve, 70
sat, 103 sesame, 105fn
satiety, 62 sew, 117
sauerkraut, 116 shah, 73
savage, 21 ski, 59, 82
savant, 85 shake, 105
saw, 61 sharp, 81
scarce, 112 sheaf, 69, 98
scare, 82 sheaves, 98
scenic, 57 sheik, 117
scenically, 57 shelf, 69, 70, 96, 98
sceptic, 66 shellac, 56, 65, 70
schedule, 93, 105 shellacked, 56
scheme, 105 shelve, 70, 96, 98
schism, 78, 91 shepherd, 40, 41
school, 82 shew, 117
schwa, 55, 82 shoal, 117
scissors, 32 shook, 118
schist, 55, 91 shore, 111
sclerosis, 82 should, 77, 119, 135
scold, 112, 113 shoulder, 119
scare, 82 shove, 106
schist, 91 shred, 81
school, 82 shrew, 61
scrawl, 61 shrewd, 61
scream, 65, 82 shrill, 107
screw, 65, 81 shrimp, 84
schwa, 82, 91 shroud, 84
secretary, 111 shutter, 107
sect, 65 sieve, 117
secure, 105 sigh, 113
see, 57 sign, 22, 42, 43, 46, 50, 64fn., 70, 72,
seek, 75 78, 83, 97,113, 120
seize, 117 signal, 50, 64fn., 72, 120, 140
self, 69, 98 signed, 22
sell, 18 signer, 42, 43
selves, 98 signify, 43, 120
senile, 108 signing, 46, 120
senility, 108 simile, 105 fn.
sense, 58 simple, 77
 WORD INDEX

simply, 77 souvenir, 119

sing, 44, 45 sow, 119
singing, 49 spaghetti, 59, 73, 73 fn.
siren, 111' spaghetto, 59
sit, 55, 102, 104 spasm, 77, 78
site, 55, 102, 104 spawn, 61
skeptic, 66 speak, 76
ski, 82 special, 82
slaughter, 114 speech, 76
sleigh, 114 speed, 79
sleight, 117 sphere, 41, 80, 82, 111
slight, 113 spin, 69
slip, 76 spindle, 69
sloe, 57 spiral, 111
slough, 73, 114, 118 splash, 76, 82
slow, 82 spoof, 69fn., 96
small, 77, 82 spread,116
smart, 81, 84 spring, 81, 82
smith, 95, 99, 100, 107 spur, 111
smithy, 95, 99, 100 spurious, 111
smoke, 105 squash, 82
smooth, 77 squire, 81
smoulder, 119 squirrel, 81, 111
snake, 75 staff, 69, 98
snarl, 78 stamen, 77
sniff, 69fn., 96 stanchion, 67
snow, 78, 82 stand, 82
soak, 75 stands, 68
sobriety, 38, 62 star, 111
social, 40, 46 state, 109
society, 38, 62 static, 109
sod,103 staves, 98
sold, 113 steak, 116
soldier, 52 stein, 117
solution, 87 sthenic, 82, 85
solve, 83, 87, 97, 112fn. stigma, 42
some, 38, 106 stimuli, 59
sonant, 69 stirrup, 111
sonnet, 36, 107 stomach, 67
soot, 118 stood, 118
sorghum, 73 storm, 77, 84, 111
sough, 73, 114, 118 stoup, 119
sought, 114 straight, 114
soul, 119 straw, 61
sound, 69, 83, 94, 97, 99 stress, 96
soup, 119 strip, 84
south, 61, 86 strong, 71 fn., 81, 82, 128
southern, 86 stronger, 44, 128
154 WORD INDEX

studied, 117 tangerine, 106

study, 117 target, 71
stuff, 69fn., 96 tariff, 111
stylus, 42 taught, 114
suave, 60, 86 taxi, 59
sublime, 105 teach, 116
subpoena, 44, 45, 65 teeth, 56, 86, 96
subtile, 64fn. teethe, 56, 86, 96
subtility, 64 fn. tennis, 58, 107
subtle, 35, 64, 94, 123 term, 81
suede, 60 terrace, 111
sugar, 66, 93 test, 82
suggest, 72 text, 84 1
suite, 60 text 8, 63
suitor, 119 thaw, 61
sumac, 65 the, 44
summit, 106fn. theater, 38
summons, 58 theatrical, 38
super, 36, 57, 104 theft, 100
supper, 36, 57, 103, 104, 107 them, 77
sure, 66, 93, 124 theme, 105
surf, 81 then, 44
surrey, 106fn. there, 105
svelte, 82 theses, 44
swamp, 45 thief, 69, 70, 96, 98, 100
swan, 45, 87 thieve, 70, 96, 100
swath, 86, 96 thieves, 98
swathe, 86, 96 thigh, 113
sweat, 116 thimble, 65
sweater, 116 thin, 44
swilk, 50 thirst, 82
swim, 82 this, 44
sword, 87, 88 thorough, 111, 114, 119
syllogism, 78 those, 44
sylph, 80 thou, 61, 119
symmetry, 107 though, 114
symphonic, 78 thought, 114
synchronic, 79fn. thumb, 65
syndicate, 44 trauma, 61
synonym, 78 thread, 116
threat, 116
Τ threw, 61
table, 56, 103 through, 81, 85
tabu, 62 throw, 85
tack, 55 thumb, 44, 65
talc, 65 thwart, 85
tall, 112 thyme, 85
tallow, 112 tie, 59, 117
 WORD INDEX

tiger, 71 two, 87, 88

tight, 113 tycoon,118
tirade, 111 tying, 59
tmesis, 84
toboggan, 72, 107 U
toe, 48, 57 ukulele, 105 fn.
toggle, 107 umlaut, 116
told, 112 uncouth, 119
tomato, 57 unearth, 96
ton, 38 unguent, 61
took, 75, 118 urbane, 108
topheavy, 41 urbanity, 108
tortoise, 118 urge, 110, 111
toss, 106fn. urine, 111
toucan, 119 urn, 81, 111
touch,119 use, 58, 83, 103
tough,73, 114,119 userer, 96, 108
toward, 87, 88 usurious, 108
town, 62 utility, 108
trace, 48, 53, 56, 56fn. utilize, 108
traceable, 56
tracing, 56 V
traffic, 41, 56, 65 valet, 85
trafficking, 41, 51, 56, 65 valise, 105
train, 81, 84 valois, 118
treachery, 116 vanilla, 107
tread, 116 variety, 38, 62
treadle, 116 vehement, 40, 53, 54, 56
treasure, 40, 93, 124 vehicle, 40, 54, 74
treble, 104, 105 vehicular, 40, 54, 74
trek, 75 veil, 116
triumph, 80 velours, 83
troll, 112 verbose, 108
trolley, 112 verbosity, 108
troubadour, 119 vermouth, 119
trouble, 118 very, 111
trough, 73, 114, 127 victual, 32, 66
trousers, 61 vignette, 91
trosseau, 62 village, 107
tunnel, 107 vinter, 87
turquoise, 86 virtu, 62
tussle, 104 virtue, 111
twang, 45 virus, 111
twelfth, 70, 95 viscosity, 62
twelve, 70, 95 viscous, 62
twenty, 84 visitation, 18
twig, 70 vocation, 18
twine, 87 vogue, 103
156 WORD INDEX

volcano, 112fn. wine, 87

volume, 105 wire, 111
vow, 37, 61, 119 witness, 96
wives, 94, 98, 100
W woe, 57
wad, 112 wolf, 69, 98
waffle, 112 wolves, 98
wag, 45, 112 woman, 38fn.
waggle, 112 won, 38 fn.
wait, 115 wonder, 38fn.
walk, 112 wood, 118
walleye, 117 wool, 118
waltz, 77, 84, 90 word, 110
wander, 112 world, 76, 81
wangle, 113 worn, 113
want, 113 worry, 38fn., I l l
warm, 54 worth, 86
was, 83 worthy, 86
wash, 113 would, 77, 119
wasp, 82, 112 wound, 119
water, 113 wreak, 76
watt, 106fn. wreath, 56, 98
wax, 45, 112 wreathe, 56
wealth, 116 wreaths, 98
weapon,116 wrench, 78
weather, 116 wretch, 76
weave, 87 wriggle, 107
web, 87 write, 87
weigh, 114 writhe, 36
weld, 68 wrong, 71 fn.
were, 105 wrought, 87, 114
when, 88
where, 88, 105 X
whiff, 69 fn., 96 xenon,88
whirl, 111 xenophobia, 88
whirr, 106fn. xylem, 88
whittle, 36
who, 88 Y
whole, 88 yacht, 66
whore, 88 Yahweh, 73
whose, 106 yea,116
why, 88 ycleped, 59
width, 98 yes, 55
widths, 98 yield, 117
wife, 69, 94, 98, 100 yogi, 59, 71
wild, 113 yolk,112
willow, 107 you, 61
wind, 113 young, 71 fn., 118
 WORD INDEX

ytterbium, 59 Ζ
yttrium, 59 zebra, 103
zodiac, 65
 INDEX OF NAMES

Barnhart, Clarence Κ., 28 Gill, Alexander, 23, 24, 31

Bazell, Charles Ε., 26 Gimson, A. C., 68
Bethel, John P., 35 Gleason, Η. Α., Jr., 26
Grimm, J., 10, 25
Bloomfield, Leonard, 9fn., 10,11,27-
29 Hall, Robert Α., Jr., 26, 48, 52fn.
Hart, John, 9fn., 30-32
Bopp, F., 25 Haugan, Einar, 15fn.
Bradley, Henry, 24, 27 Hockett; Charles F., 12,17, 27,44fn.,
Bridges, Robert, 122 85 fn.
Brown, Goold, 18, 23-25, 107fn. Holmberg, Borje, 21
Bullokar, William, 23, 24, 31 Hume, Alexander, 18-21, 122
Burchfield, R. W., 30fn.
Butler, Charles, 10 Jespersen, Otto H., 18, 38fn., 92
Jones, D., 32
Campagnac, Ε. T., 19fn. Jonson, Ben, 20, 23
Craigie, William Α., 32
Danielsson, Bror, 32fn. Kenyon, John S., 13, 92, 109fn.
Dobson, Eric J., 19fn., 31 fn. Kingdon, Roger, 102fn.
Douglas, James, 18-19, 21-23 Kittredge, George L., 18fn.
Downing, John, 33 fn. Knott, Thomas, 13
Krapp, G. P., 32
Edgerton, William F., 27 Kurath, Hans, 13fn., 15fn., 51 fn., 73,
Ellis, Alexander J., 32 79 fn.
Elphinston, James, 32
Ewert, Alfred, 107 fn. Lehiste, Ilse, 15fn., llOfn.
Levin, Harry, 49 fn.
Flom, George T., 9fn. Lily, William, 23
Francis, W. Nelson, 17, 26, 28-29, Lounsbury, Thomas R., 9fn.
47 fn.
Franklin, Benjamin, 10 Mcintosh, Angus, 50fn.
Fries, Charles C., 29 McLaughlin, John C., 47 fn.
Marchand, Hans, 64 fn.
Gaines, H. F., 90fn. Marckwardt, Albert H., 68 fn.
Gibson, E. J., 49fn. Mulcaster, Richard, 19fn.
 INDEX OF NAMES 159

Murray, James A. H., 12, 23, 25, 32 Thorndike, Edward L., 13 fn., 103fn.

Newman, Stanley S., 103fn. Vachek, Josef, 17, 28

Venezky, Richard L., 10fn., 12,46fn.
Paul, H., 25
Peterson, Gordon E., 15fn. Waldo, George S., 102fn.
Pitman, Isaac J., 33 Webster, Noah, 25
Pratt, Fletcher, 90 fn. Weir, Ruth H., 46fn., 55fn., 71 fn.
Pulgram, Ernst, 26, 47fn. Wheatley, Henry B., 19fn., 21
Whitney, William D., 25
Rask, R. K., 10, 25 Wijk, Axel, 21 fn.
Saussure, Ferdinand de, 25-26, 47 fn.
Smith, Thomas, 19, 30-31 Zachrisson, Robert E., 9fn., 32
Stetson, Raymond H., 26
Sweet, Henry, 18
160

JANUA LINGUARUM

STUDIA MEMORIAE NICOLAI VAN WIJK DEDICATA

Edited by C. H. van Schooneveld
SERIES MINOR
1. ROMAN JAKOBSON and MORRIS HALLE: Fundamentals of Language.
1956. 97 pp. Gld. 8.—
3. EMIL PETROVICI: Kann das Phonemsystem einer Sprache durch frem-
den Einfluss umgestaltet werden? Zum slavischen Einfluss auf das
rumänische Lautsystem. 1957.44 pp. Gld. 5.—
4. NOAM CHOMSKY: Syntactic Structures. Eighth printing. 1969. 117
pp. Gld. 8.—
5. N. VAN WIJK : Die baltischen und slavischen Akzent- und Intonations-
systeme: Ein Beitrag zur Erforschung der baltisch-slavischen Ver-
wandtschaftsverhältnisse. 2nd ed. 1958.160 pp. Gld. 19.—
7. ERNST PULGRAM: Introduction to the Spectrography of Speech.
Second printing 1964. 174 pp., 31 figs., 2 tables. Gld. 14 —
8. AERT H. KUIPERS: Phoneme and Morpheme in Kabardian (Eastern
Adyghe). 1960. 124 pp. Gld. 20.—
9. A. ROSETTI: Sur la theorie de la syllabe. Deuxi£me edition, refondue
et augmentee. 1963. 43 pp. Gld. 9.50
10. URIEL and BEATRICE WEINREICH: Yiddish Language and Folklore:
A Selective Bibliography for Research. 1959. 66 pp. Gld. 8.—
11. E. and κ. DELAVENAY: Bibliography of Mechanical Translation —
Bibliographie de la traduction automatique. 1960. 69 pp. Gld. 13.—
14. DWIGHT L. BOLINGER: Generality, Gradience, and the All-or-None.
1961.46 pp. Gld. 7.—
19. WALERIAN SWIECZKOWSKI: Word Order Patterning in Middle Eng-
lish: A quantitative study based on Piers Plowman and Middle
English Sermons. 1962. 114 pp. Gld. 20 —
20. FINNGEIR HIORTH: Zur formalen Charakterisierung des Satzes. 1962.
152 pp. Gld. 19.—
22. Ε. F. HADEN, M. S. HAN, and Y. W. HAN: A Resonance Theory for Lin-
guistics 1962. 51 pp. Gld. 9 —
23. SAMUEL R. LEVIN: Linguistic Structures in Poetry. Second printing.
1964. 64 pp. Gld. 9.50
24. ALPHONSE JUILLAND and JAMES MACRIS: The English Verb System.
1962. 81 pp. Gld. 10.—
25. IVAN FONAGY : Die Metaphern in der Phonetik: Ein Beitrag zur Ent-
wicklungsgeschichte des wissenschaftlichen Denkens. 1963. 132 pp.,
5 figs. Gld. 18.—
26. Η. MOL: Fundamentals of Phonetics, I: The Organ of Hearing. 1963.
70 pp., 28 figs. Gld. 11.—
27. LÄSZLO ANTAL: Questions of Meaning. 1963. 95 pp. Gld. 13.—
 161

29. PUNYA SLOKA RAY: Language Standardization: Studies in Prescrip-

tive Linguistics. 1963. 159 pp. Gld. 20 —
30. PAUL L. GARVIN: On Linguistic Method: Selected Papers. 1964.
158 pp. Gld. 16.—
31. LÄSZL0 ANTAL: Content, Meaning and, Understanding. 1964. 63 pp.
Gld. 9.50
32. GHORGES MOUNIN: La machine Ä traduire: Histoire des problemes
linguistiques. 1964. 209 pp. Gld. 32.—
33. ROBERT E. LONGACRE: Grammar Discovery Procedure: A Field
Manual. 1964. 162 pp. Gld. 12.—
34. WILLIAM s. COOPER: Set Theory and Syntactic Description. 1964.
52 pp. Gld. 9.50
35. LUIS J. PRIETO: Principes de noologie: Fondements de la theorie
fonctionnelle du signifie. Preface d'Andre Martinet. 1964. 130 pp.,
36 figs. Gld. 21.—
36. SEYMOUR CHATMAN: A Theory of Meter. 1965.229 pp., many graphs,
2 plates. Gld. 23.—
37. WAYNE TOSH: Syntactic Translation. 1965.162 pp., 58figs.Gld. 23.—
38. NOAM CHOMSKY: Current Issues in Linguistic Theory. 1964. 119 pp.
Gld. 12.—
39. D. CRYSTAL and R. QUIRK: Systems of Prosodic and Paralinguistic
Features in English. 1964. 94 pp., 16 plates. Gld. 14 —
40. FERENC PAPP: Mathematical Linguistics in the Soviet Union. 1966.
165 pp. Gld. 24.—
41. s. κ. SAUMJAN : Problems of Theoretical Phonology. 1968. 2 2 4 pp.
figs. Gld. 26.—
42. MILKA IVIC: Trends in Linguistics. Translated by Muriel Heppell.
1965. 260 pp. Gld. 28.—
43. ISTVAN FODOR: The Rate of Linguistic Change: Limits of the Ap-
plication of Mathematical Methods in Linguistics. 1965. 85 pp.,
some figs. Gld. 13.—
44. THEODORE M. DRÄNGE : Type Crossings: Sentential Meaninglessness
in the Border Area of Linguistics and Philosophy. 1966. 218 pp.
Gld. 23.—
45. WARREN H . FAY: Temporal Sequence in the Perception of Speech.
1966. 126 pp., 29 figs. Gld. 19.50
46. A. CAPELL: Studies in Socio-Linguistics. 1966. 167 pp., 2 tables.
Gld. 22.—
47. BOWMAN CLARKE: Language and Natural Theology. 1966. 181 pp.
Gld. 26.—
49. SAMUEL ABRAHAM and FERENC KIEFER: A Theory of Structural
Semantics. 1966. 98 pp., 20 figs. Gld. 15.—
50. ROBERT j. SCHOLES: Phonotactic Grammatically. 1966. 117 pp.,
many figs. Gld. 15.—
51. HOWARD R. POLLIO: The Structural Basis of Word Association
Behavior. 1966. 96 pp., 4 folding tables, 8 pp. graphs, figs. Gld. 16.—
162

52. JEFFREY ELLIS: Towards & General Comparative Linguistics. 1966.

170 pp. Gld. 22.—
54. RANDOLPH QUIRK and JAN SVARTVIK: Investigating Linguistic Ac-
ceptability. 1966. 118 pp., 14 figs., 4 tables. Gld. 15.—
55. THOMAS A. SEBEOK (ED.): Selected Writings of Gyula Laziczius. 1966.
226 pp. Gld. 26.—
56. NOAM CHOMSKY: Topics in the Theory of Generative Grammar.
1966. 96 pp. Gld. 12.—
58. LOUIS G . HELLER and JAMES MACRIS: Parametric Linguistics. 1967.
80 pp., 23 tables. Gld. 10.—
59. JOSEPH H. GREENBERG : Language Universale: With Special Reference
to Feature Hierarchies. 1966. 89 pp. Gld. 18.—
60. CHARLES F. HOCKETT: Language, Mathematics, and Linguistics. 1967.
244 pp., figs. Gld. 21.—
62. B. USPENSKY: Principles of Structural Typology. 1968. 80 pp.
Gld. 14.—
63. v. z. PANFILOV: Grammar and Logic. 1968. 106 pp. Gld. 16.—
64. JAMES c. MORRISON: Meaning and Truth in Wittgenstein's Tractatus.
1968. 148 pp. Gld. 18.—
65. ROGER L. BROWN: Wilhelm von Humboldt's Conception of Lin-
guistic Relativity. 1967. 132 pp. Gld. 16—
66. EUGENE J. BRIERE: A Psycholinguistic Study of Phonological Inter-
ference. 1968. 84 pp. Gld. 12.—
67. ROBERT L. MILLER: The Linguistic Relativity Principle and New
Humboldtian Ethnolinguistics: A History and Appraisal. 1968.
127 pp. Gld. 16—
69. i. M. SCHLESINGER: Sentence Structure and the Reading Process.
1968. 172 pp. Gld. 18 —
70. A. ORTIZ and E. ZIERER: Set Theory and Linguistics. 1968. 64 pp.
Gld. 10.—
71. HANS-HEINRICH LIEB: Communication Complexes and Their Stages.
1968. 140 pp. Gld. 17.—
72. ROMAN JAKOBSON: Child Language, Aphasia and Phonological
Universale. 1968. 104 pp. Gld. 12.—
73. CHARLES F. HOCKETT: The State of the Art. 1968.124 pp. Gld. 14.—
74. A. JUILLAND and HANS-HEINRICH LIEB: "Klasse" und "Klassifikation"
in der Sprachwissenschaft. 1968. 75 pp. Gld. 13.—
75. jifti KRAMSKY : The Word as a Linguistic Unit. 82 pp. Gld. 12.—
76. URSULA OOMEN: Automatische Syntaktische Analyse. 1968. 84 pp.
Gld. 16.—

MOUTON · PUBLISHERS · THE HAGUE