On Latin Adverbs Harm Pinkster download

https://ebookbell.com/product/on-latin-adverbs-harm-
pinkster-51755410

Explore and download more ebooks at ebookbell.com

Here are some recommended products that we believe you will be
interested in. You can click the link to download.

On Latin Adverbs Amsterdam University Press Amsterdam Archaeological

Studies Harm Pinkster

https://ebookbell.com/product/on-latin-adverbs-amsterdam-university-
press-amsterdam-archaeological-studies-harm-pinkster-2545600

New Perspectives On Historical Latin Syntax Volume 2 Constituent

Syntax Adverbial Phrases Adverbs Mood Tense Philip Baldi

https://ebookbell.com/product/new-perspectives-on-historical-latin-
syntax-volume-2-constituent-syntax-adverbial-phrases-adverbs-mood-
tense-philip-baldi-50266686

New Perspectives On Historical Latin Syntax Volume 2 Constituent

Syntax Adverbial Phrases Adverbs Mood Tense Philip Baldi

https://ebookbell.com/product/new-perspectives-on-historical-latin-
syntax-volume-2-constituent-syntax-adverbial-phrases-adverbs-mood-
tense-philip-baldi-4343642

Reflections On Latin American Development Roberto De Olivera Campos

Benjamin Higgins

https://ebookbell.com/product/reflections-on-latin-american-
development-roberto-de-olivera-campos-benjamin-higgins-51924710
Chapter On Latin America Guillermo A Lemarchand

https://ebookbell.com/product/chapter-on-latin-america-guillermo-a-
lemarchand-23843170

Readings On Latin America And Its People Volume 2 Since 1800 Paperback
Mark Wasserman

https://ebookbell.com/product/readings-on-latin-america-and-its-
people-volume-2-since-1800-paperback-mark-wasserman-7410390

Oxford Reds Classic Commentaries On Latin Classics John Henderson

https://ebookbell.com/product/oxford-reds-classic-commentaries-on-
latin-classics-john-henderson-37237182

Transnational Perspectives On Latin America The Entwined Histories Of

A Multistate Region Luis Roniger

https://ebookbell.com/product/transnational-perspectives-on-latin-
america-the-entwined-histories-of-a-multistate-region-luis-
roniger-38532670

Foreign Policy On Latin America 19701980 Charles W Maynes

https://ebookbell.com/product/foreign-policy-on-latin-
america-19701980-charles-w-maynes-10441686
on latin adverbs
The Amsterdam Academic Archive is an initiative of Amsterdam University Press.
The series consists of scholarly titles which were no longer available, but which
are still in demand in the Netherlands and abroad. Relevant sections of these
publications can also be found in the repository of Amsterdam University Press:
www.aup.nl/repository. At the back of this book there is a list
of all the AAA titles published in 2005.
 Harm Pinkster

On Latin Adverbs

3 a
Amsterdam Academic Archive
 On Latin Adverbs (isbn 0 7204 6186 3) by Harm Pinkster was first published
in 1972 by the North-Holland Publishing Company of Amsterdam.
It was published at the time as Volume 6 in the North-Holland Linguistic Series
(isbn 07204 6180 4), series editors S. C. Dik and J. G. Kooij.

Cover design: René Staelenberg, Amsterdam

isbn 90 5356 843 3
nur 616
© Amsterdam University Press • Amsterdam Academic Archive, 2005

All rights reserved. Without limiting the rights under copyright reserved above,
no part of this book may be reproduced, stored in or introduced into a
retrieval system, or transmitted, in any form or by any means
(electronic, mechanical, photocopying, recording or otherwise)
without the written permission of both the copyright owner
and the author of the book.
 preface to this edition

On Latin Adverbs was originally published in 1972. This is an unchanged reprint. Certain
topics have continued to interest me, notably the hierarchical structure of sentences
and the positions in that hierarchy where adverbs can be found, as well as coordina-
tion. I refer those interested in my later writings on these topics to: Harm Pinkster
‘Attitudinal and illocutionary satellites in Latin’, in: Words in Their Places, ed. by H. Aert-
sen et al., Amsterdam, Free University 2004, pp. 191-198; Harm Pinkster ‘La coordina-
tion’, L’information grammaticale 46 (1990), pp. 8-13. Much of the book was used for my
Latin Syntax and Semantics (see: http://cybergreek.uchicago.edu/lss/) and will be used in
my forthcoming Latin Syntax.

Amsterdam, July 2005

Harm Pinkster
PREFACE

This book was submitted as a doctoral thesis to the Faculty of Arts of the
University of Amsterdam under the supervision of Prof. S.C. Dik and Prof.
A.D. Leeman. It owes much to the interest and criticism of a number of
friends both in the Institute of Classical Philology and in the Institute of
General Linguistics.
Prof. Leeman stimulated the work on this thesis by his continuous
interest in both my personal and my working circumstances and greatly
contributed to style and content.
Prof. Dik left no paragraph unchanged by his merciless criticism and has
made it permanently impossible for me to repay to him what he has done
for me in this and former studies.
Prof. Ruijgh did not confine himself to advising the Faculty on this book
- actually, that would have been rather hard to imagine - but made many
pertinent remarks.
Mrs. A.M. Pool-Bolkestein's critical remarks are almost the size of a book
themselves. Miss E. Vester and Drs. A. ~~jksbaron, too, expressed their scep-
ticism abundantly. They all spent many hours in discussing problematic
sections with me. Drs. J. J. L. Smolenaars proved not to be the layman in
linguistics he would make me believe he is. Mr. N.S.H. Smith endeavoured
to make the language this study is written in at least look like English.
A number of assistants in the Institute of Classical Philology as well as
Mrs. A.M. Pool-Bolkestein, Miss E. Vester, Drs. A. Rijksbaron and Drs.
J. J. L. Smolenaars helped in reading the proofs and so did my wife. She has
for several years bravely endured the fate of being married to someone
writing a thesis. More than to anyone I owe her a debt of gratitude and an
apology.
H.PINKSTER

VII
CONTENTS

Preface VII

Introduction 1
1. Intention 1
2. Survey of the content 1
3. General framework and terminology 4

1. General problems of studying a dead language 9

1.1. Problems of the use of a closed corpus 9
1.1.1. Productivity 9
1.1.2. Grammaticality 10
1.1.3. Adequateness of the corpus 13
1.2. Choice of material 13
1.3. Universals and related problems 14

2. Problems of classification 17
2.1. Purposes of classification 17
2.2. Types of criteria 18
2.2.1. Phonology 19
2.2.2. Morphology 19
2.2.3. Syntax 22
2.2.4. Semantics 23
2.3. Classes of words, word-elasses and parts of speech 25
2.3.1. Multi-level-elassification and rank of levels 25
2.3.2. Correspondence of levels 27
2.4. Rank and number of criteria. Overlapping classes 30
2.5. Ontological, psychological and notional definition of word-elass 32
2.6. Conclusion 32

3. Adverbs in Roman grammatical theory 35

3.1. General properties of adverbs 36
3.1.1. Syntax 36

IX
 Contents

3.1.2. Semantics 37
3.1.3. Morphology 39
3.1.4. Phonology 40
3.2. Subclassification and individual problems 40
3.3. Category and function 43

4. The adverb in Latin linguistics 45

4.1. Morphology and phonology 46
4.2. Semantics 48
4.3. Syntax 55
4.3.1. Ambiguity of the definition 55
4.3.2. Adverbs modify members of other categories as well 57
4.3.3. Inadequateness of the description of adverbs 59

5. Adverbs as derived forms 63

5.1. Adverbs and adjectives are both forms of one lexeme 64
5.2. Adverbs as adjectives marked by empty morphemes 65
5.3. Adverbs as case forms 66

6. Syntactic problems 71
6.1. Modification 71
6.1.1. Modification interpreted as optionality with respect to the
verb 72
6.1.2. Problems concerning the optionality of obligatoriness of
constituents with respect to the verb 77
6.1.2.1. Optionality is linked with certain conditions 77
6.1. 2.2. ADJUNCT constituents defined as optional with
'respect to the verb 78
6.2. Other types of affinity: NUCLEUS and PERIPHERY 83
6.2.1. Ablativus causae explained as a marker of a peripheral
ADJUNCTMOTIVE 84
6.2.1.1. NUCLEUS and PERIPHERY 91
6.2.2. ADJUNCT constituents specifying place or time 93
6.3. The problem of modal adverbs and disjuncts 96

7. Subclassification ofadverbs 103

7.1. The use of question words 103
7.2. Word order 105
7.3. Coordination 108
7.3.1. Restrictions on coordination 108
7.3.1.1. Categorial equivalence 111
7.3.1.2. Functional equivalence 112
7.3.1.3. Equivalence of semantic role 113
7.3.1.4. Relations between semantic aspects of constituents 114
7.3.1.5. Violations of restrictions on coordination 116
7.3.2. Procedure 120
7.3.3. Application to some Latin adverbs 124

x
 Contents

8. Adverbs and other invariables 135

8.1. Adverbs as optionally free forms 136
8.2. Adverbs as non-obligatorily free forms as opposed to interjections 141

9. The relationship between adverbs and prepositions 145

9.1. The synchronic point of view 145
9.2. The historical point of view 147

10. Adverbs and connectors 153

10.1. Consecutive connectors and causal adverbs 155

11. Adverbs and subordinators 165

11.1. Types of connection between sentences and clauses 166
11.2. The traditional description 167
11.3. The role of the subordinate clause in the sentence 169
11.4. Subordinators 171

Bibliography 179

Indices 189

XI
INTRODUCTION

1. Intention

This study deals with a number of aspects of the words which are usually
called adverbs in Latin. It contains on the one hand a critical discussion of
their treatment in Latin grammatical studies - the characteristics attributed
to them, their relationship to other words - and on the other hand a discus-
sion of the conditions that have to be met in order to achieve a better
(sub)classification - general problems of classification as well as criteria for
effecting such classification - and a better description of the functions of
adverbs in larger constructions. The study contains, therefore, both lan-
guage-specific sections and more general ones. I have tried to write the
passages specifically dealing with Latin in such a way that they are clear
enough to the non-Latinist 1 , the more general passages in such a way that
they are understandable for Latinists who are not acquainted with recent
developments in linguistics. I can only hope that I have avoided the risk of
being either obscure or trivial to both groups of readers.

2. Surveyo[the content

The book consists of three parts. The first deals with general problems. The
second part (ch. 3-7) contains a discussion of the class of Latin adverbs, the
1 Therefore I have added English translations to the Latin examples. I have used
published translations (Loeb library, Shackleton Bailey's translation of Cicero's letters)
whenever possible.
2 Introduction

criteria used to define them and a discussion of a new means of achieving a

subclassification. In the third part (ch. 8-11) I deal with the relationship
between adverbs and other so-called uninflected words with respect to
syntactic behaviour. Clearly, a fourth part might have been added dealing
with the syntactic characteristics of adverbs and other constituents (such as
ablative case forms 'of nouns, prepositional phrases, subordinate clauses)
which are often called 'adverbial adjuncts', 'adverbials', etc. Such a study
would require to be of considerable length and exceed the bounds of a
dissertation. It is my intention, however, to devote a separate monograph to
this subject.
The general problems dealt with in part one concern the nature of the
material we have to work with (ch. 1) and classificatory problems (ch. 2). I
will expound the idea that the work of the Latin linguist is seriously
hampered by the fact that it has to be based on a closed corpus, which has
been selected to a large extent as a result of literary and other not primarily
linguistic factors. On the other hand the linguist's work may be facilitated
by his knowledge of general linguistic facts.
In ch. 2 I will point out that classification of linguistic items is useful,
whatever type of grammar one has in mind, in that it forces the linguist to
state explicitly the similarities and differences among these items. They may
be classified on the basis of phonological, morphological, syntactic and
semantic criteria, which are as a rule not directly interrelated. There is no
objection to using criteria of different levels to define particular classes (as
has been done for the class of adverbs) nor to the overlapping of classes. The
main objection that may be raised against the traditional classification of
words into the class of adverbs lies in the fact that too few criteria are given.
Most words that are .now considered adverbs in Latin grammar were so
cO,nsidered in ancient Roman grammatical theory as well. I illustrate this by
summarizing the chapters on adverbs by Priscian and to a smaller extent
Charisius (ch. 3). Both the morphological criterion of 'uninflectedness' and
the syntactic one of 'modification' (of a verb) underlie their class of
adverbs. As in modern grammars2 their subciassification is based on
semantic considerations.
The main progress in modern grammars is that the relationship between
on the one hand adverbs and on the other hand certain adverbs and adjec-
tives is mentioned in the definition of adverbs (invariable words that modify
a verb, adjective or adverb). The intuitive nature of the subclassification in

2 4Modern' as opposed to 4ancient'. They are not necessarily recent.

 Survey of the content 3

the grammars is demonstrated with regard to manner adverbs said to be used

as intensifiers (vehementer - 'vehemently', 'strongly'). I observe next that
the defmition of adverbs is ambiguous, insufficiently specific and inade-
quate for describing the characteristics of words which are nonetheless
regarded as adverbs (ch. 4).
Two chapters are devoted to the morphological and syntactic criteria
mentioned in the defmition of adverbs. As to the criterion of uninflected-
ness (ch. 5), I will show that recent suggestions to describe the relationship
between prudens ('wise') and prudenter ('wisely') as flexional overlook a
number of difficulties.
Subsequently the concept of modification will be examined (ch. 6). If
modification is equated with 'optionality' - as it is - the definition of
adverbs raises no difficulties at the .phrase level. However, it is not
unproblematic at the clause and sentence levels. Certain adverbs and other
constituents likewise functioning as ADJUNCT 3 appear to be obligatory
with certain verbs, just as OBJECT and SUBJECT constituents are
obligatory with certain verbs. In more general terms this points to
differences 'of degrees of closeness of constituents, with respect to the verb
in particular. Once this is realized, it is clear that an adverb cannot be simply
defined on the basis of its modifying a verb. As far as the structure of
sentences is concerned, this new insight into the way constituents seem to
be associated with each other should be expressed somehow. It appears,
moreover, that there are also problems in recognizing'different degrees of
closeness of relation to the verb in the case of constituents which are
optional with respect to the verb. This is illustrated in detail with regard to
the so-called ablativus causae and temporal and local adverbs (or rather
ADJUNCT constituents). Within this conception of various degrees of
closeness of relationship the so-called modal adverbs (sentence adverbials)
fmd almost satisfactory explanation. Actually, these words may be said to
be optional expansions of the whole sentence, or rather the central part of
the sentence, which is itself built up of certain peripheral constituents and a
nuclear part which comprises the verb and a number of obligatory
constituents that depend on the particular verb chosen. Not all adverbs
occur at all these levels. This fact should, of course, be mentioned explicitly
in their definition.
The discussion about the structure of sentences goes far beyond what is
necessary for a description of adverbs alone. I feel fully justified in doing
this since any defmition of a class of words, whether ancient or modern,

3 See below in section 3.

4 Introduction

presupposes explicitly or implicitly a descriptive model. It is, to my mind,

useless to discuss a class of words separately from the grammatical
framework within which these words find a place.
In ch. 7 I endeavour to show that, given the fact that our study is
necessarily corpus-oriented, coordination is the best heuristic means for
establishing similarities and differences of linguistic items. Consequently I
put forward a theory of coordination which takes into account both formal
aspects (syntactic characteristics of words and their syntactic function) and
semantic aspects (lexical semantic aspects of words and their semantic
role 3 ). This theory is subsequently applied to some Latin adverbs. A few
instructive cases are discussed.
The third part of this book is devoted to the relationship between adverbs
and words which are usually assigned to the classes· of interjections,
prepositions and conjunctions.. If adverbs are regarded as words which may,
but need not - in contrast with interjections - occur in one-word
sentences, we may conclude as far as the material allows us to, that not only
are prepositions and conjunctions excluded from independent occurrence,
but that several words regarded as adverbs in traditional grammar have also
then to be rejected from the class of adverbs (ch. 8).
As for prepositions (ch. 9), they differ from adverbs in their obligatorily
occurring with case forms. The diachronic hypothesis of a development
from adverbs to prepositions is probably too simple. It should, at any rate,
not be transferred to synchronic analysis, since this creates several
difficulties.
A subclass of coordinating conjunctions (e.g. ergo - 'so') is often
regarded as, or at any rate not clearly distinguished from, adverbs. They
appear to be different in a number of syntactic characteristics (ch. 10).
The last chapter deals with adverbs in their relation to subordinating
conjunctions. After a discussion of hypotaxis and the role of subordinating
conjunctions in this, it is observed that some or all subordinating
conjunctions in so-called adverbial clauses could be described as relative
adverbs (ch. 11).

3. General framework and terminology

This study is written within a functional framework. In my opinion (cf. Dik

1968; Becker 1967a,b) a grammatical description of the following sentences
should at least give four types of information.
 General !rameworA and terminology 5

(1) pater filium laudat ('the father praises his son').

(2) filius a patre laudatur ('the son is praised by his father').
(3) pater a filio laudatur ('the father is praised by his son').

In (1), pater ('father') is a noun (syntactic lexical information), which refers

to a human being (is human; semantic lexical information), it is SUBJECT
in the sentence (syntactic function) in contradistinction to [ilium .('son'),
which shares the first two characteristics, but is OBJECT and in contra-
distinction to patre in (2), which belongs to a prepositional phrase and is
syntactically ADJUNCT. Moreover, pater in (1) is AGENT (semantic role)
and is opposed in this respect to pater in (3), where pater is PATIENT.
The information about syntactic function and semantic role will be
combined in the following way: SUBJECTAGENT. Semantic roles are
indicated as subscripts of functions. In the case of ADJUNCT in particular
(comparable with the traditional 'adverbial adjunct'), a large number of
semantic roles is assumed.
Expressions in accordance with the terminology adopted would be

(1) a constituent function as ADJUNCTMOTIVE, or

(2) an ADJUNCTMOTIVE constituent, or
(3) a causal adverb functioning as ADJUNCTMOTIVE.

On various occasions, however, I will use a less formal formulation, for the
sake of simplicity. I hope that this will not have the opposite effect of
confusing the reader.
One of the arguments in favour of the type of description followed in
this study is presented by the behaviour of constituents in coordination
patterns (cf. 7.3.). I do not believe, as Chomsky (1965: 63-74) does, that
the relevant functional information can be described solely in terms of
relations between constituents. In fact, constituents may stand in various
relations to each other (cf. 7.1.). Nor--do I think that one can fuI1y dispense
with functional information and restrict oneself to the information on
semantic role (as Fillmore 1968 seems to imply). The description of Case (in
the usual sense of the term) would be complicated: it is possible to account
for Case on the basis of the function of the nominal constituent. The.
semantic role of the nominal constituent is not instructive for the
assignment of Case. Finally, I do not think that one can manage with
syntactic functions alone. In fact, it is remarkable that a constituent may
have different functions, but remain the same from the semantic· point of
view, e.g. pater and a patre in (1) and (2).
6 Introduction

As for the term 'adverb', it has been used in different ways and it will be
useful to state at the outset what is meant by it. The following uses can be
found
(a) Hermes (1968: 99) distinguishes 'Adjektivadverbien' like longe ('far'),
propius ('nearer'), eitius ('faster'), from 'Adverbien' like adhue ('thus far'),
tamen ('still'), proeul ('far'), nee ('and not'), quam ('than'). This distinction
is, given the heterogeneity of the latter class, not very illuminating.
(b) O'Brien (1965: 39) makes a sensible distinction between 'adverbs'
and 'adverbials'. By adverbs he understands the derived words ending in -e,
-(i)ter, -ies, -tim, -tus,' by adverbials the non-derived words like eras
('tomorrow'), clam ('secretly'), sic ('so'), antea ('before') as well as the
supines (p. 100).
(c) G. Lakoff (1968: 5ff.) speaks of instrumental adverbs with reference
to expressions as with a knife in

(4) Seymour sliced the salami with a knife,

that is, he uses 'adverb' instead of the traditional term 'adverbial adjunct',
or, in my own terminology, 'prepositional phrase functioning as ADJUNCT-
INSTRUMENT'·
(d) Steinitz (1969: 1) understands by 'Adverbial' wh(~.t has been called
'adverbial adjunct'. Moreover, words like trotzdem are reckoned within this
group (p. 12), for which, traditionally, the term 'adverb' is used. Apart from
these stand 'sogenannte reine Adverbien' like oben, unten, reehts.
With regard to (c) and (d) it should be realized that in transformational
grammar 'adverb' is not used for the regular or adjectival adverb, since these
are taken as transformationally related to adjectives. For a textbook
treatment see Lyons (1968: 326-7).
I will use 'adverb' in the traditional sense, e.g. as it is used by
Kiihner-Stegmann 4 • For the sake of clarity I will sometimes speak of
'traditional adverb'. By this I understand: 'a word that is traditionally called
an adverb'. This does not mean that I accept the traditional nomenclature
and classification. My own point of view will become clear, I hope, in the
following chapters. In case confusion should arise from my using the word
'adverb', I will clarify my intentions by adding epithets and the like. In
order to give an impression of what I call adverbs I have collected a number
of examples that will be discussed later in this book. I have indicated the
adverbs by using italics. The pages refer to the future discussion.

4 At I, p. 794 they use the term more loosely for 'ADJUNCT constituent'.
 General framework and terminology 7

(5) multum maius ('much more important'; p. 14).

(6) parve per eos flectitur delphinus ('the dolphin moves dimly among
them'; ch. 2, n. 17).
(7) prudens homo prudenter agit ('the wise man acts wisely'; p. 37).
(8) insignite improbus ('remarkably bad'; p. 50).
(9) hominum concordiae causa sapienter popularium ('men who wisely
favoured popular measures to preserve peace'; p. 52).
(10) amore gloriae nimis acri fortasse ('a love for honour which is
over-keen perhaps'; p. 58).
(11) homini illic nobilissimo ('to a man who was there in high esteem'; p.
60).

I use several terms that are probably unfamiliar to some readers..I will
mention here only the terms 'phrase level' (cf. p. 3) and 'slot' (cf. 6.2.1.1.
init.). I use 'phrase level' in the sense intended by Longacre (1964: 74) for a
level between the word and the clause and sentence. Examples are noun
phrases (a sharp knife). 'Slot' is a common term in tagmemic grammar. A
linguistic expression could be seen as a structure of a number of open places
occupied by certain constituents, such as words, phrases, clauses, in certain
functions. For example, in a sentence a place is reserved for a constitutent
which may function as SUBJECT in that sentence. The SUBJECT 'slot' may
be 'filled' by a noun, a noun phrase, etc.
Other terms will be explained in the text and are recorded in the
index 5 •

5 As for a number of typographical indications, by'?' I indicate questionable

grammaticalness, by ,*, ungrammaticalness. Double colon (::) indicates that another
person begins to speak. Syntactic functions are written in capitals (SUBJECT),
semantic roles in italicized capitals (A GENn. Abbreviations of Latin works are in
accordance with the New Oxford Latin Dictionary.
1. GENERAL PROBLEMS OF STUDYING A DEAD
LANGUAGE

In this chapter I will discuss a number of problems that are inherent in the
study of a dead language. Next, I will expound what consequences these
general problems have for the choice of the material. Finally, I will justify
the use of linguistic universals in studying Latin adverbs after a brief
exposition of the nature of this concept.

1.1. Problems o/the use o/a closed corpus

Some of the most important problems one has to face when undertaking a
syntactic study of some aspect of the Latin language result from the fact
that Latin is a dead language. The lack of native speakers as well as the fact
that the corpus that has come down to us is almost definitely closed gives
rise to certain difficulties, mainly of three types: (i) concerning produc-
tivity; (il) concerning grammaticality; (ill) concerning the adequateness of
the corpus as to variety of constructions. These factors are complicated
further by the uncertainty of the manuscript tradition.

1.1.1. Productivity
It is impossible to decide which of the utterances we find exemplify
productive formation types and which do not. Some of them may agree
with rules that were productive in former stages of the language, others with
rules that possibly had not yet come fully into use. A decision could be
reached by comparing stages of the language - assuming that the construc-
tion underlying the problematic utterance is attested in other stages - and'
by distinguishing different social strata. There is, of course, a large number

9
10 General problems of studying a dead language

of comparative studies, but most of them take the total system insuffi-
ciently into account and concentrate on details. The investigation of
different social strata is difficult on account of the lack of material and has
been carried out mainly from an aprioristic psycholinguistic point of view.
Cf. Happ (1967: 64-78) on Hofmann (1963).
Since a division into productive and unproductive constructions is
difficult, it is hard to decide which constructions should be descrIbed in the
grammar and which should be referred to the lexicon as being idioms (De
Groot 1956b: 22-4).
The problem of productivity is furthermore connected with the problem
of grammaticality, which will be discussed shortly. Should we consider a
construction that is not productive (assuming that we can find this out),
ungrammatical or not? In case we answer this question in the affirmative,
we must assume that language development takes place via ungrammatical
expressions. This would be rather unattractive. A notion like 'restricted
grammaticality' (i.e. 'a low degree of grammaticalness' 1) seems necessary.
Application of these remarks on productivity to a Latin example can be
found below 1.2.

1.1.2. Grammaticality
It is also difficult to decide which of the utterances we find exemplify
grammatical expressions, and which do not (Isenberg 1965: 160ff.<;Rtiwet
1968: 36ff.). Any' corpus is bound to contain ungrammatical or less
grammatical elements. This problem cannot be solved by extending the
corpus, e.g. by taking Caesar in addition to Cicero, since the number of
ungrammatical items would be extended as well. Whereas the linguist who is
working on a modem language may check his corpus with the aid of
informants, this is impossible in the case of a dead language. That is to say,
it is not only impossible to check the rules· that one sets up on the basis of
the data by constructing new expressions to be tested, it is also impossible
to decide whether the data on the basis of which the rules are to be set up
were correct in Latin. Of course, we have a number of statements by Roman
scholars on certain constructions, e.g. Quintilian's remark that in Alexan-
driam ('to Alexandria') is incorrect (Inst. 1,5,38), but generally speaking
these statements are only of moderate value, especially for a syntactic
study. Most Roman grammarians lived in a later period than that in which

1 On the question of degrees of grammaticalness seeChomsky (1957: 16, n. 2; 1965:

148-53). Criticism in lakobson (1959).
 Problems of the use of a closed corpus 11

the constructions we are interested in were in use, and they were not
particularly interested in syntax. Their greatest value is in their chapters on
style. From the very fact that they discuss certain phenomena as stylistic we
may conclude that such constructions were peculiar and ,try to determine
why they considered them peculiar. For example, the notion 'zeugma' can
be used in the description of coordination in Latin (7.3.1.5.).
The problem of grammaticality is even more serious in Latin because of
the way in which the data have been handed down to ·us, the manuscript
tradition. We cannot be sure that what we read has indeed been said or
written as it stands - whether it is correct or not. The linguistic study of a
dead language, consequently, must be firmly based on textual criticism.
The notion of grammaticality has evoked a lot of criticism since
Chomsky (1957: 14) stressed its importance 2 • I will add a few remarks about
it. Two notions are closely connected with the notion of grammaticality,
viz. acceptability to a native speaker and conformance to a rule of grammar.
Now, in my opinion, it should be clear that something can be considered
grammatical, not because it is acceptable to a native speaker (an idealized
native speaker), but because it apparently fits in with the rules that must be
assumed as underlying language usage. Usage is not accidental, but shows
systematic patterning. In other words, grammaticality of a construction is a
relative notion: grammaticality of one construction implies ungrammatic-
ality of other ones. Only a good survey of the network of constructions
warrants a reliable decision about the grammaticality of one construction in
this network. It is highly improbable that a native speaker possesses
anything resembling such a survey. Moreover, there is reason enough to
assume a feed-back from language teaching at school to judgment on
grammaticality. Hjelmslev's remark (1935: 76) seems to be fully pertinent
to this problem: 'la linguistique de nos jours a pris une forte allure de
psychologisme et a institue une idolatrie de ce sentiment linguistique des
sujets parlants qui n'est en realite que l' effet immediat de la doctrine
traditionaliste'.
Now, what the linguist is doing is setting up a system of rules that
describe the utterances he finds and predict others. These rules need not be
identical with the rules that have been postulated as underlying language
usage, but must be expected to deliver the same results (correct language
usage). Indirectly, therefore, something may be called grammatical if it
conforms with the rules that are formulated by the linguist. Thus, pater est
bonus ('father is good') is grammatical (and will probably be accepted),

2 Hill (1961) observes that objective criteria are lacking.

12 General problems of studying a dead language

since it is in accordance with the rules regarding Case, Number, Gender, etc.,
whereas pater est· bono is not (and is probably unacceptable), because it
does not conform to the rule regarding agreement of Case. Judgments of
native speakers about grammaticality are, therefore, only useful if they can
be shown to be related to the rules of the grammar. 'L'uniformite et la
coherence des jugements portes par un grand nombre de sujets' (Ruwet
1968: 43) in itself is no proof3 •
We have seen by now that if an utterance occurs in a given corpus, it
cannot be considered a grammatical expression on that account alone. One
might suggest that frequency of occurrence should tip the scale. In my
opinion, Chomsky (1957: 15-6; 1965: 195, n. 5) has rightly pointed to the
fact that here, too, the crucial question is whether the particular utterance
agrees or disagrees with a rule of grammar. Of course, the utterance may be
an idiom which stands outside the rules of the grammar, either because it is
still unproductive or has become unproductive (see 1.1.1.), or because it is
an isolated idiom as such.
More important is the circumstance that frequency of occurrence cannot
be used in a negative sense either. The non-existence or low frequency of
parallels can never prove that something has not been used (Riemann 1879:
3-4) nor that a particular utterance is incorrect~ In not a few Latin studies
this seems to have been forgotten, Le. 'is not attested' is understood as 'is
not correct' or even 'was avoided', e.g. on stylistic grounds. On the other
hand frequency is often used as a guide for determining the correct reading
of a text. Thus Nisbet (ad Cic. Pis. 23) reads luce palam ('openly in
daylight') in conflict with the manuscripts (luce et palam), not because the
manuscript reading is ungrammatical - which it is not -, but because 'the
normal expression is luce palam' (see 7.3.3. below).
These reservations about attestation should not be interpreted to mean
that I adhere to the view advanced by R. Lakoff (1968: 2ff.), that
attestation is not necessary in order to judge the grammaticality of an
utterance. Attested utterances may be seen to form some sort of network
from Which, with the above reservations, conclusions can be drawn about
what is Latin and what not, and which types of construction words may
enter into. Of course, every grammar, not only one of a dead language, is in
the end based upon an inventory of attested utterances. The same holds
when the linguist uses his own judgment in determining whether an
utterance in a non-native, specifically a dead language, is grammatical or
not.
3 For general observations on acceptability and the use of informants see
Quirk- Svartvik (1966).
 Choice of material 13

1.1.3. Adequateness o/the corpus

It is not only impossible to establish the grammatical reliability of the
corpus, as was concluded in the preceding section, but is is also impossible
to ascertain whether all facts which are necessary for the construction ofa
grammar are taken into account and, of course, the corpus was not intended
to provide material for a linguistic study, but constitutes a rather accidental
collection.
A considerable part consists of literary texts, which were created under
stringent literary conventions. As far as poetic texts are concerned metrical
considerations have been at work. For instance, quite a lot of adverbs are
excluded on metrical grounds (priess 1909). Other works are of technical
character. Documents of everyday Latin are not long enough to guarantee
reliable results. A representative corpus (Harris 1966: 13), if obtainable at
all for living languages, can certainly not be obtained for Latin. There are
too many extralinguistic factors involved.

1.2. Choice o/material

In view of the difficulties mentioned in 1.1. it seemed· best to base a

classification of adverbs on a body of texts of considerable size, written
within a fairly limited period, dealing with a range of subjects as wide as
possible, and influenced as little as possible by stylistic considerations, that
is to say, prose, not poetry. Consequently I was driven to Cicero, whose
work, moreover, has the advantage of showing more linguistic coherence
(being more 'one dialect') than would be found in a corpus selected from
several different authors. 4 This implies that I do not intend to give a
description of the Latin spoken in Cicero's time. On the other hand,
differences between educated and everyday language are differences of
'style' only (Robins 1964: 52) and there may be a considerable parallelism
between Cicero's Latin and the Latin spoken in everyday use in his time.
Though there are arguments for choosing Cicero'sworks as a basis, a
restriction to these works only would prevent a better insight into several
problems. When Cicero's works offered no examples or an insufficient
number of examples, whereas good instances could be found in other
authors, I have used them. For example, Plautus offers a large number of
question-answer patterns. I have consulted Plautus, therefore, when

4 Of course, Cicero's works are shaped by literary principles as well and are highly
stylistic.
14 General problems of studying a dead language

examining the usefulness of question words as a criterion for classification.

However, I have not tried to use data from various authors to give a
diachronic survey of a construction. For such a study, in my opinion, the
necessary synchronic basis is lacking.
The choice of Cicero has not been dictated by overestimation of his
prose. One may well be aware of Cicero'seffort to write correct Latin,
without denying that his prose yields expressions which otherwise do not
occur in so-called classical Latin, or which only occur in the few documents
of 'Vulgar Latin'. Too many passages in Cicero, however, are emended
because a stylist like Cicero 'could not have written it'. As an instance of
such an overestimation of Ciceronian prose and basfug a wrong conclusion
on low frequency of a construction, take the textual problem which occurs
in Cic. de Orat. 3,92.- Here the manuscripts have mu/turn maius ('much more
important'). There is no other example of mu/turn ('mu'ch') used in this way
in Cicero. Mu/to is his normal usage (so-called ablativus mensurae). Manutius
acoordingly proposed mu/to. This has appeared attractive to many scholars.
Szantyr (: 136) remarks that mu/turn is hardly acceptable in Cicero. There
are few instances in other authors (Plautus, Pliny the Elder, see Szantyr
(I.e.) and Thes. VIII, 1618, 19ff.) which are accepted. I would suggest that
those who retain mu/turn (e.g. Kumaniecki in the Teubner edition) are right.
There is reason enough to suppose that mu/turn is an adverb (Woelfflin
1885: 99) and not a so-called adverbial accusative, as Szantyr has it. In
Cicero's works we find mu/turn dormire ('to sleep much'), mu/turn Athenis
esse ('to be often in Athens'), mu/turn praestare ('to excel much'; cf.
Lundstrom 1961: 73-5), multum bonus ('very good'), permultum ante
('very much before'). In this interpretation mu/turn fits in with other
constructions of adverb+comparative adjective, e.g. longe melius ('far
better'). This construction competed with a construction like mu/to maius,
i.e. an independent adjective in the ablative case+comparative adjective
(so-called ablativus mensurae), but was not yet fully productive (Krebs-
Schmalz 1905: 11, 33). Parallels with words other than mu/turn can be
found (Kiihner-Stegmann: I, 402). This particular use of mu/turn is
exceptional, but does agree with a rule of grammar.

1.3. Universa/s and related problems

In studying an unknown language the linguist may find support for his
analysis in knowledge about other languages known to him, and especially
in knowledge of so-called universals of language. This observation, of course,
 Universals and related problems 15

reflects current practice in linguistics. The value of such knowledge is

mainly heuristic. The linguist should not be tempted, of course, to project
phenomena from one language to another automatically (Jespersen 1933:
46 - 'squinting grammar').
In this study coordination is given special attention, which construction
seems to work under similar restrictions in all, or at least in all
Indo-European, languages (cf. ch. 7). Several of the criteria for classifying
words have been borrowed from or turned out to be used in studies on
modern languages. This method seemed not only respectable in itself, but is
also advisable on negative grounds:' in Latin linguistics, sufficient attention
has not been paid to criteria of any kind and the study of syntax has not yet
taken the prominent place which it has in modern language studies,
especially since the development of transformational generative grammar.
I have used 'knowledge of universals' above in the sense of 'explicit
knowledge such as a linguist may obtain about universals'. In a re'cent study
on complementation R. Lakoff (1968) has presented the view that the
linguist when working on a foreign language 'does not depend only on the
grammar of that language; he also depends on an implicit, inborn knowledge
of the nature of language, present in all human beings, a knowledge of what
is conceivable in language and what not! ' From this inborn knowledge
'general' or 'universal' grammar derives. The specific grammar of a language,
say English or Latin, has to be learned (R. Lakoff 1968: 7-8). A similar
argument can be found in Bach (1968: 113). He first remarks that if there
were no universals 'it would be a complete mystery how knowledge about
one language could throw light on another (as it does)'. This has been
argued above. He then proceeds with something similar to R. Lakoffs
words: 'the actual rules of the base are the same for every language' and 'In
this view it is almost correct to say that language learning by the child
consists in finding the transformations which will derive the surface
structures of sentences of his language from the universal set of base
structures'.
It will be clear that universals of the type I discussed above and those
meant (probably) by Bach in the first quotation are different from the kind
of universals meant by R. Lakoff and by Bach in his second quotation. In
the first interpretation universals are observed facts about natural languages,
in the second they are regarded as properties of the human mind (or faculty
of speech). In my opinion, Kiparsky (1968: 171) is right in questioning the
correspondence between rules formulated by the linguist (among these the
universals under my interpretation) and rules which a human being must
possess to produce correct utterances. There may be restrictions on concept
16 General problems of studying a dead language

formation in the human mind which make R. Lakoffs example (1968: 6)

(1) *1 devoured for John to do it,

ungrammatical in any language (at least when devour is taken literally 5 ), but
this is an hypothesis which seems to belong to the domain of psychology in
the first place; we might, in fact, have to do with a faculty of the human
mind to discern sense and nonsense. On 'inborn knowledge of the nature of
language' see the criticism by OIler-Sales (1969).
It would seem that a knowledge of universals, as assumed by R. Lakoff,
has no heuristic value in studying the structure of a foreign language.

Notice the fact that in a figurative sense devorare can be constructed with verba
('words'). See Thes. s.v. devorare 875, 62ff.
2. PROBLEMS OF CLASSIFICATION

In this chapter I will discuss some general problems of classification, viz. the
purposes 'of classification, the types of criteria on which a classification can
be based, the relation between these types and, finally, the relative
importance of the criteria and the number of criteria required.

2.1. Purposes of classification

Linguistic items can be grouped into classes on account of certain

characteristics they have in common. Word classification, such as the
classification of adverbs, is a specific instance of this general problem. The
purposes of such a classification are to obtain a higher degree of
generalization in the description of the specific language the linguist is
working on, to make it more adequate since, while classifying, the linguist is
forced to state explicitly which items are similar, and in what respects,
which items are different, and what is the relation between the various
classes to which the items are assigned. The description becomes simpler
since identical or similar phenomena can be described together in one
statement (Crystal 1967: 25-7; Van Wijk 1967: 235; Schopf 1963: 63).
Classification should, of course, be based on a careful analysis of an
adequate sample of material (cf. 1.1.).
Once the linguist has reached a satisfactory classification and has thereby
left the stage of discovery he may incorporate the classes in the grammar.
The way in which this incorporation is effected will be determined by the
type of grammar the linguist has in mind. Traditionally the incorporation
took the form of a more or less explicit defmition 1 , like Priscian's adverb

1 Explicit definition does not belong to the stage of discovery. See Robins (1966:
17-9; 1967b: 211-4) contra Bach (1964: 28-9).

17
18 Problems of classification

definition: 'the adverb is an invariable part of speech that modifies a verb'

(cf. ch. 3). One may also indicate the membership of classes in rewriting
rules, as was usual in earlier versions of transformational generative grammar
(Bach 1964: 27-9). For example,

Adverb ~ nunc ('now'), heri ('yesterday'), etc.

The (sub)classes involved may either be closed and then we can enumerate
their members, or open, and then we can only characterize them in some
way or other2 • For example, local adverbs (hic ('here'), etc.) could be
enumerated not only because there are not many of them, but also because
in Latin there seems to be no way of adding items to this subclass. On the
other hand, the subclass of so-called regular adverbs like misere ('miser-
ably'), feliciter ('happily') is open, since items may be added any time: the
formation of these adverbs is productive.
This study will contain in the first place a discussion of the criteria by
which the traditional class of words called adverbs is set up, and of the way
this class can be split up into subclasses, and does not aim at a new and
complete classification itself.

2.2. Types o/criteria

Words have characteristics of four different types and may, consequently,

be classified on the basis of four different types of criteria, viz.
phonological, morphological, syntactic and semantic criteria.
The familiar word-classes (parts of speech) were set up on the basis of
criteria of different types. For example, Priscian'sdefinition of the adverb
cited in 2.1. contains a morphological statement ('invariable') and a
syntactic one ('modifies a verb'). I will distinguish 'classes of words' and
'word-classes' ,using 'word-classes' in the traditional sense. I will return to
this difference in 2.3. The following sections will be devoted to criteria from
the four levels of analysis separately.

2 Gleason (1965: 119) distinguishes between ~list' and ~characterization'. On the

distinction ~closed' vs. 'open' see Juilland-Lieb (1968: 53-7). The distinction is not
numerical, but qualitative: Pike (1967: 201): 'An open class is one to which new items
are known to have been added in recent times, and one to which the linguist may by
chance observe a native speaker add new words by borrowing or by direct
invention ... '.
 Types of criteria 19

2.2.1. Phonology
Phonological criteria, though they may lead the linguist to the assignment of
words to phonological classes, are generally considered to be only of
moderate use in the establishment of word-classes. Kratochvil (1967: 135-6)
mentions stress as a factor in Chinese and Crystal draws attention to an
attempt to define some English word-classes by stress as well (1967: 42),
but they, too, agree with the unimportance of phonology in this matter.
Nor does phonology seem relevant in Latin. Szantyr (: 215) remarks that
the preposition is 'tonschwach' 3, in contradistinction to adverbs, with
which they are assumed to be related historically, and words of other
classes, but this has no overall validity (cf. 3.1.4. on accent in Roman
grammatical theory).

2.2.2. Morphology
In a morphological classification words might be assigned to various classes
depending on whether they are compound or not, etc. It will be clear, that,
at least in Indo-European languages, this particular classification would
cross-cut almost all traditional word-classes. The main morphological
division which suggests itself in Indo-European languages is according to
whether words are inflected or not. In fact, this underlies Varro's division of
the word-stock of Latin in 'fruitful' and 'barren' words (L.I0,14; 8,9-10).
The class of inflected words may then be subdivided according to, the
syntactic and semantic categories they are inflected .for. Varro thus
distinguished for Latin four classes: (a) words declined 4 for Case (nomi-
nals), e.g. docilis ('docile'), (b) words declined for Tense (verbals), e.g. docet
('he teaches'), (c) words declined for both (participles), e.g. docens
('teaching') and (d) words declined for neither of these categories (adverbs),
e.g. docte ('wisely').
This classification bears close resemblance to that of Matthews (1967),
who divides the Latin lexicon into lexemes, which are characterized for
certain so-called morphosyntactic categories, on the one hand, and

3 No enclitics are attached to it.

4 Varro seems not to have distinguished in his terminology 'inflection' and
'derivation'. Here I take over his term 'declined' as a neutral term. In ch. 5 I will clarify
my standpoint with regard to the question of inflection and derivation. I regard Case
and Tense as inflectional categories, formation of adverbs as derivational.
20 Problems of classification

particles on the. others. I will use both notions in this study. The notion
'lexeme' seems to be particularly useful in the study of inflectional
languages.
From antiquity onwards hortus, horti, have been described as related
forms ('garden'-nominative case and genitive case, respectively). The very
use of the term 'case' reflects this idea. A common way of expressing the
relationship is saying that horti is the genitive form of the word hortus. The
nominative case form has often been considered the basic form of the word:
'casus rectus' (Szantyr: 22), or - in a different framework - as the
unmarked case form par excellence, e.g. by De Groot (1948: 462; 1956a:
189)6. .
However, in any sensible definition of 'word' (e.g. 'ReicWing 1969: 33)
hortus, horti, etc. have to be taken as different words. Still, there must be
some way to express the relation between them in spite of their clear
morphological and syntactic differences. This is achieved by saying that
hortus and horti are forms of the same abstract lexical entity, which can be
called lexeme. In some cases the lexeme can partially be located, so to
speak, in some aspect of the forms (e.g. hort-), in other cases it cannot be so
easily located in this way (e.g. in the case of iter and itineris: 'route' nom.
and gen., respectively). Considering different words as related in this way· to
the same lexeme is always attractive from the semantic point of view.
Moreover, the different forms can be shown to be correlated with the
function of the lexeme in the larger construction. For example, if it occurs
as SUBJECT in a sentence with a finite verb, the nominative case form
(hortus) of the lexeme is necessary, if it functions as ATTRIBUTE with
respect to another noun the genetive case form (horti) is obligatory. The
lexeme can be arbitrarily symbolized by the italicized words in capitals
(HORTUS). Instead of saying that hortus and horti are forms of the lexeme
HORTUS, which is inflected for the morphosyntactic categories Number
and Case in accordance with its function in the construction, one can also
say that hortus and horti are 'members' of the paradigm of HORTUS.
Finally, HORTUS can be viewed as the abstract symbolisation of the sum of
forms hortus, horti, etc.
Similarly, laudo, laudas ('I praise, you praise'), etc. can be said to be

Matthews, too, considers adverbs as inflected forms. I do not agree with him, but
pass over this problem here. The term "particle' is used in a rather vague way in
traditional grammar. See ch. 8, no. 2.
6 Cf. also Greenberg (1966: 95), Hjelmslev (1935: 99-100). Criticism in Fillmore
(1968: 6).
 Types of criteria 21

forms of the lexeme LA UDO, that is inflected for the morphosemantic

categories Tense, Mood, Person, Voice 7 •
Since the differences between the forms of the lexemes are correlated
with the syntactic and semantic function the words perform in larger units,
a classification of this type is useful, since it is relevant not only to
morphology, but also to syntax. Morphosyntactic categories can help to
define syntactic classes as we1l 8 , even though there are exceptions. E.g.
nequam ('bad') must be described as the only member of the paradigm of
the lexeme NEQUAM, if morphosyntactic categories (Number, Case,
Gender) are used in defining the class of adjectives. Of course, the so-called
invariable words (particles) can, by their very nature, not be dealt with in
this way. These words do not belong to a paradigm, the members of which
are chosen in relation to the function in the construction. To assign them to
syntactic classes demands purely syntactic techniques. Syntactic analysis
was not the strongest aspect of ancient grammatical theory. It is not
surprising, therefore, that the traditional definitions of inflectional cate-
gories such as noun and verb are much more useful than those of invariables
such as adverbs and conjunctions. Cf. Matthews (1967: 165).
It will be clear, however, that a debate may always arise about whether a
particular word or a particular paradigm should be considered member of a
(larger) paradigm.
I referred to one member paradigms above with respect to nequam.
Another problem concerns the fact that it is not easy to see how one can
decide how many morphosyntactic or morphosemantic categories are
necessary. For example why shouldn't we distinguish a morphosyntactic
category 'Adverbial' and say that lexemes like MISER and FELIX
('miserable', 'happy') are inflected for Adverbial on the one hand, for
Adnominal on the other, the latter category covering Number, Case and
Gender, instead of saying that there is a derivational relation between the
particles misere and feliciter on the one hand and adjectivallexemes MISER
and FELIX on the other? Evidently, a decision can be made only within

7 See Matthe\\Ts (1965) for a description of Latin verb forms. On 'lexeme' see also
Lyons (1968: 197-8) and Matthews (1970: 107-13). I use the term 'morphosemantic'
on purpose. As yet, I do not believe in such a thing as 'meaning' of a case (it is a
morphological marker of syntactic relations), but I recognize several semantic values in
the verb forms (apart from syntactic values).
·8 "------
Hockett (1958: 221) seems to draw a borderline between inflection and syntax,
judging from his definition: 'A part of speech is a form-class of stems which show
similar behaviour in inflection, in syntax, or both'.
22 Problems of classification

the total framework of the description. I come back to this particular

problem later (ch. 5).

2.2.3. Syntax
Whereas a morphological classification of words proceeds from a study of
the internal structure of words, a syntactic classification is achieved by
classifying words in accordance with their syntactic valence, Le. their ability
to enter into certain relations with other words.
Since a word A can be assigned to a certain class on the basis of its
relation to one or more other words, e.g. B, which in turn must be assigned
to a certain class precisely because of its relation to A (among other words),
the establishment of classes is always circular. There is no objection to this
circularity and, consequently, there is no reason to follow Kratochvil (1967:
148-9), who thinks that the syntactic classification of words by examining
their functional abilities and combinatory qualities 'presupposes that it is
known or that it is possible to find out in other ways than through the
word-classes themselves what are the functions which words are said to be
able to fulfil'. Evidently, there is no such possibility. Function is a relational
notion (Chomsky 1965: 68-74; Dik 1968: 154-9)9, and what functions are
needed in the description of language in general or a particular language has
to be decided on the basis of the total description (and, of course, the
particular type of grammar). Functions have no independent existence nor,
for that matter, have categories. Compare also Crystal (1967: 26): 'defining
X by Y and' Y by X is from the viewpoint of descriptive grammar quite
permissible'. For example, BONUS ('good') is an adjective since it may have
the function ATTRIBUTE with a noun, e.g. VIR ('man'), but not with a
personal pronoun, e.g. EGO ('1'). On the other hand VIR is a noun since it
may function as HEAD with respect to the adjective BONUS and differs
from the pronoun EGO, which may not fulfil that function. The reason for
postulating these functions and distinguishing them from other functions is
again that they explain the behaviour of members of these classes as
opposed to the behaviour of members o.f other classes in this particular
construction.
The syntactic valence of a word can be determined in three interdepen-
dent ways: (a) by observing the constituency of the utterances in which it
appears; (b) by omitting it, replacing it and effecting other changes in the

9 Cf. Hjelmslev (1935: 48): ~Les deux ordres de concept se superposent et se

traduisent mutuellement'.
 Types of criteria 23

expression and checking the results; or (c) by considering whether the wo"rd
can occur in certain substitution frames (on these see Garvin 1958: 60;
Crystal 1967: 43-4; and Wheatley 1970: 48-54).
The former two, especially, are relevant to the assignment of words to
classes when the linguist proceeds from a definite corpus. Following
Carvelland-Svartvik (1969: 42-3) criteria belonging to these types may be
called 'constituent criteria' and 'transformational criteria', respectively.
Examples of transformational criteria are: (1) Question. Which question
word (if any) should be asked to get this particular utterance as an answer
with this particular word? E.g., Romae habitat ('he lives in Rome') would
be the answer to a question ubi habitat? ('where does he live? '). (2)
Coordination, viz. which words could be coordinated with this particular
word? (3) Permutation. Could word A and word B be interchanged? (4)
Omission. If this particular word is left out, would the expression still be
'viable'? (see Garvin 1958: 60, with his reservations about the term 'viable',
and 6.1.1. end, below).
These methods presuppose, of course, the presence of one or more
informants to answer the questions and therefore the transformational type
seems to be inappropriate to the study of dead languages - except insofar as
our own judgment on grammaticality in such languages can be relied on (for
this problem, see Pinkster 1971: 385). The first type (constituent criteria),
however, can be used in a transformational sense, if we compare the
constituency of particular utterances. Thus, finding a sentence Romae
habitat, we cannot ask someone which question word would be appropriate
here, but we can look and see whether we find a reliable example of Romae
(habitat) in answer to a question ubi habitat? Or, more generally, whether
there is in such answers an example of a word which, tentatively, seems to
belong to the same class as Romae. This means that syntactic classification
of words in a dead language· need not be different from classification in
modern languages, but will be much more difficult, since attestation is
necessary (cf. 1.1.2.). In Latin grammars only the first transformational
criterion (question word) is used to distinguish subclasses within the class of
adverbs (see below 4.2.).

2.2.4. Semantics
A semantic classification can be conceived of in two ways. Most usual is the
setting up of categories based on features like 'quality', 'state', etc. Mostly
the assignment of a word to such categories on account of its semantic
aspect takes place on intuitive grounds, as might be expected in the present
24 Problems of classification

state of semantics. Of course, this is not to say that such a classification is

useless. These categories cross-cut the traditional word-classes (see the end
of this section).
Another type of semantic classification is the one originated by
Reichling. In this theory the vague notion 'meaning' is replaced by the cover
term 'semantic aspect'. Several types of semantic aspect are distinguished:
(independent) 'meaning', e.g. of words like cow; 'semantic value', e.g. of
elements like -s in cows; 'deixis' of words like this (Reichling 1969: 42-9;
Dik 1968: 251-8). A classification of words according to their type of
semantic aspect might cross-cut the traditional word-classes.
As for the first type of semantic classification, most linguists are very
sceptical about the possibility of classifying words into word-classes on
account of their semantic aspect (e.g. Gleason 1965: 116; V. Wijk 1967:
244; and Kratochvi11967: 151), Le. into the same classes that would also be
set up on the basis of other (e.g. syntactic) criteria. They doubt whether it is
possible to assign a word to a certain class, say adjectives, on account of its
semantic aspect, nor do they believe that it is possible to say what categorial
semantic aspect members of a certain class (say adjectives) have in common.
It will be clear that those, who, in accordance with the second alternative
formulation, assume a categorial semantic aspect for each word-class
(whether these are similar to the traditional parts of speech or not) ought to
be able to assign words to these classes on the basis of their semantic aspect
(first alternative formulation). As far as I know nobody has been able to do
this.
As an example of the belief in a categorial semantic aspect I cite the
following passage from Bally (1965: 113): 'Les categories lexicales com-
prennent ... les semantemes virtuels designant des substances, des qualites,
des proces et des modalites de la qualite et de l'action', substantives,
adjectives, verbs and adverbs respectively. Similar remarks can be found in
Latin grammars, e.g. Kiihner-Holzweissig (1912: 253); Ernout-Thomas
(1959: 1). If we assume for the sake of the argument that 'quality' is a
well-defined notion, we may agree that e.g. alba indicates the quality of
domus in domus alba ('white house'). However, the semantic notion quality
seems pertinent to albitudo ('whiteness') and albere ('to be white') as well,
which are traditionally called substantive and verb, respectively. That is, the
semantic notion 'quality' cannot be taken as characteristic of one (primarily
syntactic) class, but is shared by members of several classes. A similar
remark can be made about the distinction, clear at first sight, of a class of
numerals with a categorial semantic aspect 'number'. In fact, this class
comprises words of varying syntactic, morphological and even semantic
 Classes of words, word-classes and parts of speech 25

characteristics (Robins 1966: 17; cf. 2.3.2.). The semantic definitions of

adjective and adverb especially, such as those of BaBy above, prove, as
Hjelmslev (1939: 143) rightly observes, that 'la categorie est etablie par la
rection non par la signification; ... il y a d'autres mots que l'adjectif et
l'adverbe qui indiquent la qualite. Si par consequent on pretend definir une
categorie par la signification, on le fait en assignant une etiquette
semantique a une categorie etablie d'abord par des criteriums fonctionnels'.
Those who adhere, like Bally, to the hypothesis of a categorial semantic
aspect of word-classes usually assume - implicitly or not - that there is a
correspondence between formally distinguished classes and semantic classes.
An exponent of this view is De Groot, whose opinion I will discuss in 2.3.2.

2.3. Qasses of words, word-classes and parts of speech

In the beginning of 2.2. I, briefly, drew attention to the difference between

'classes of words' and 'word-classes'. Next, I isolated the levels on which'
statements about words can be made and accepted the possibility of
classification on all levels. On the other hand I have used word-class in the
traditional sense ('part of speech') as a class of words which share a number
of characteristics belonging to different levels, e.g. the class of adjectives
said to be inflected for Case, Number and Gender, to partake in the
derivational category 'Degrees of comparison' - two morphological state-·
ments -, to modify a noun: syntactic statement, and to denote quality (of
the noun): semantic statement. I have on several occasions discussed the use
of characteristics of a certain level to define words which had been assigned
- or at least were believed to belong to - classes established by reason of
characteristics which pertain to one or more other levels. In this section I
will concentrate on this matter in more detail. The questions to be answered
are: (1) is it feasible to define a class by characteristics belonging to
different levels and which level should be most important in this respect? ;
(2) are there correspondences between classes that are established indepen-
dently at different levels and, if so, of what type are they?

2.3.1. Multi-level-classification and rank of levels

Paul (1920: 352-5) observes that the traditional distinction into parts of
speech is insufficient, but that this is inevitable when various criteria from
such different levels ('so verschiedenartige Riicksichten') are used. I think
that this statement is only correct as long as one feels oneself obliged always
26 Problems of classification

to use criteria from all four levels, in order to assign words to a class.
However, one should make only those generalizations which it is useful to
make. If then 'the application of all four criteria will result in chaos', as Van
Wijk (1967: 247) foresees, this will be a practical problem, which does not
follow from the application of criteria from more than one level as such.
Another consequence of the application of criteria from more than one level
will be that the number of criteria and thereby the number of classes and
subclasses grows.
As such, there is no objection to classifying words by all types of criteria.
There is only an objection if one limits one's objectives, e.g. to writing a
syntax instead of a complete grammar 1 o. Another argument for allowing
for criteria from more levels is that the levels are not completely
independent from each other: a characteristic at one level may always
accompany or be connected with a characteristic from another level and
thus be predictable. An example of this type might be: 'Latin colour
adjectives (semantic subclass) cannot be coordinated with 'normal' adjec-
tives, e.g. bonus ('good') (syntactic characteristic). Adverbs are not derived
from them (morphological characteristic)1 1 •
Once criteria from more than one level are accepted the question arises as
to which level has the most important criteria for the total description of a
language. I agree with those linguists who consider syntactic characteristics
most important (e.g. Crystal 1967: 43; Robins 1959: 122-4; Van Wijk
1967: 250). As the latter author observes, words may lack morphological
structure, but they do have syntactic valence. Even in an inflectional
language like Latin invariable words exist within what are, generally
speaking, inflectional classes, e.g. the adjective nequam ('bad'), which can be
assigned to the class of adjectives only on account of its syntactic valence.
The primacy of morphological criteria for classification has been
supported by De Groot and Blatt, for different reasons. De Groot (1948:
445)12 - on whose definition of word-cla~s cf. 2.3.2. - holds that a
classification of words ought to start with morphology since words with the
same stem-meaning have different syntactic valences, e.g. pater ('father'-
nominative), patris ('father'-genitive). This objection can be met, of course,
by taking account of the notions of lexeme and paradigm. It is true that

10 Fries' objection to using more types of criteria may be due to the limited
(syntactic) scope of his study (1957: 67). Cf. Schopf (1963: 64-5).
j 1 As far as I can judge this statement may even be correct. At least I find no
counterexamples. (I take Plin. Paneg. 48,1 albi et attoniti ('pale and astonished') as a
non-literal use of the colour adjective.) On these adverbs see Lofstedt (1967: 80-1).
12 De Groot (1964) pays more attention to syntax.
 Classes of words, word-classes and parts of speech 27

pater has a different distribution from patris, if one concentrates on the

words as such. If, however, one views these words as forms of the same
lexeme (2.2.), one may explain the difference between the actual forms by
saying that the forms are determined by the function that the lexeme
PATER fulfils in the construction in accordance with its syntactic valence.
The distribution of pater and patris is part of the total distribution of the
lexeme PATER.
The reverse problem seems to be more difficult: if stem and stem-
meaning must be the basis of classification of words, amare ('to love') and
amator ('lover') might well be assigned to the same class. But it is not very
useful to consider these words to be different forms of the same lexeme, as
it is of more explanatory value to s.ay that there is a derivational relation
between them, which could be accounted for by saying that the lexeme
AMATOR is a secondary lexeme derived from the (primary) lexeme
AMARE. Each of these two lexemes has its own syntactic valence.
Blatt (1952: 27) in his excellent chapter on the parts of speech remarks
that it is preferable to base the classification of words on morphology and
only then examine the syntactic uses of the classes thus determined. 'Nous
eviterons ainsi le cercle vicieux inherent fatalement au pr'ocede inverse, qui
consiste a determiner d'abord une classe d'apres un principe syntaxique et
ensuite a en etudier les emplois'. Blatt, regrettably, gives no elucidation of
his remark. The type of circularity he is afraid of, if I understand him
correctly, is in itself no objection since one should distinguish the stage of
discovery from the stage of writing a syntax (cf. also 2.2.3. init.). The
criteria which induce a linguist to establish certain syntactic classes are
among those that will be given in a syntax as the syntactic properties of the
envisaged classes (cf. 2.1.). Of course, what is at stake here is the type of
syntax: most ancient and modern grammars concentrate more on the
properties of the constituents in a construction in relation to the general
properties of the class to which they belong (paradigmatic) than on the
properties of the construction and the role of the constituents therein
(syntagmatic). Blatt, too, is unable to maintain his principle throughout.
The invariables are to be assigned to several word-classes (1952: 33): 'ces
classes sont pareillement soumises a des criteres purement linguistiques'.
These criteria, not unexpectedly, appear to be syntactic, in fact, e.g. in the
case of prepositions, the fact of their governing certain case forms.

2.3.2. Correspondence of levels

Most modern linguists agree that a one-to-one correspondence between the
28 Problems of classification

above mentioned levels of description is exceptional and can as a rule not be

expected. For the assumption that such a correspondence must generally be
expected Bazell (1952) coined the term 'correspondence fallacy'.
Examples of partial correspondences are well-known. I give two of them.
(a) Adjectives of material, e.g. ligneus ('wooden') have no corresponding
derived adverb (LOfstedt 1967: 80-1). (b) Verbs with a 'punctual' semantic
aspect (punctual 'mode of action' or 'Aktionsart') are not constructed with
constituents expressing temporal duration (cf. 6.1.2.2. below):

(1) *1 found the book for two hours.

It seems reasonable to postulate that the semantic aspect is responsible for

these partial correspondences. Observations of the latter kind .are also at the
back of the notion 'meaning-class' in R. Lakoff (1968: 165): 'We define a
meaning-class in terms of both syntax and semantics, as a set of semantic
markers that can function in syntactic rules'. See also on 'congruence in
grammatical and semantic classification' , Lyons (1968: 166-9).
Several scholars, however, like Jensen (1949: 155) and De Groot (1948;
1964) assume that a complete correspondence between levels is normal. As I
said above (2.2.4.), this idea lies also, tacitly, behind the assumption of
something like a categorial semantic aspect or 'class-meaning' (Bloomfield
1935: 202-5; 247-51). I will discuss De Groot's opinion in more detail since
he applied his view to Latin in his 1948 a"rticle.
De Groot's positiqn is reflected in his definition of word-class (1948:
447-8): 'By word-class we understand a collection of words of a given
language which have a certain categorial element or feature of meaning of
their stem in common, in so far as the existence of this common element
appears not solely from the investigation of the meaning, but also from the
morphological valence of the stem or (and) the syntactic valence of the
word'l 3, that is, the occurrence of the stem ~n certain morphological
structures and of the word (formed on the stem) in certain syntactic
structures. There is an (indirect) relation between syntax and morphology
(and even phonology; 1964: 258) in as far as all three are determined by the
Ineaning of the stem.
De Groot bases his view on word-classes on a number of considerations

13 Hockett's remark (1958: 221): ~The part of speech of a word is that of its stem'
bears some resemblance to this in its concentration on the stem. This calls forth certain
difficulties. See 2.3.1.
 Classes of words, word-classes and parts of speech 29

(1964: 257-9). The relation between semantics and syntax 14 appears from
the valence of e.g. numerals! s. The relation between semantics and
morphology appears inter alia from the fact that verbs can be derived only
from certain stems, e.g. albus ('white'), albescere ('to become white'). The
relation between semantics and phonology stands out inter alia with the
interjections. Therefore, in De Groot's opinion, it seems a reasonable theory
to assume that words of a given language belong to word-classes which are
opposed to each other by their categorial semantic aspect. De Groot is well
aware that his thesis is, in fact, a postulate and that in practice it will be
hard to substantiate the postulate 'that in a given language all words of a
certain stem valence class really have a semantic aspect in common: (1964:
259; my translation). Yet, in De Groot's opinion, this should not worry us.
It is clear that there exists a categorial difference e.g. between verbs and
numerals, though this difference is hard to define. If we want to define it,
we may say that the meaning of the category of verbs is 'process', of numer~
als 'number', while intending these notions in a 'linguistic sense', a concept
that remains undefined 1 6. 'The terminology only presupposes that the verb
in the given language really, as such, means something that is especially
inherent to the meaning of the verb in this language' (1948: 473). In other
words, the postulate is defended by saying that it is a postulate. The postu-
late itself rests upon observations which are the most obvious that could be
cited in this connection. The summary of De Groot's argument shows that
his approach is open to the criticism by Hjelmslev quoted in 2.2.4.
The concentration on stem and stem-meaning has the effect of making
syntactic differences less important. Thus, tres ('three'), tertius ('third'), ter
('three times') and terni ('three each') are assigned to one class alongside
multus (,much', 'many'), since all undoubtedly have something to do with
number (1948: 448-9; 495; 1964: 257). Since De Groot argues from stem.
and stem-meaning it is not surprising that he holds (1948: 451; 479-80) that
'Adjectives and adverbs, however different their syntactic valence may be,

14 Cf. De Groot (1948: 445): 'equality of syntactic valence as a rule can only be
explained from equality - in some respect - of meaning'.
15 Of course, numerals are not just a collection of ordinary adjectives, adverbs, etc., as
might be suggested by the following words from Robins (1966: 17, n. 57): 'the
separate chapters on the numerals are merely a pedagogical rearrangement of certain
members of the noun, adjective, and adverb classes'. Syntactically there are distinctions
from 'normal' adjectives, etc. Compare the example *extremely five houses in Gleason
(1965: 126-7).
16 Perhaps meant in contrast with what De Groot calls 'ontological' definitions (1948:
429). On these briefly 2.5. below.
30 Problelns of classification

are not different word-classes in Latin. They are morphological categories,

for which mostly the same stem is used: longus, longe ('long' adjective and
adverb , respectively). In connection with a particular meaning a stem is
sometimes only used for an adjective' - magnus ('big'), parvus ('small') -
'or only for an adverb' - tunc ('then'), saepe ('often')l 7 • As we see, the
general properties of the adjective (that is, in De Groot's terminology,
presence of - meaningless - 'syntagmemes' for Gender, Number and Case;
derivation of an adverb by a - meaningful - morpheme) are explained on
the basis of the categorial semantic aspect of the adjective, which is not
given, but only postulated because of these properties. Exceptions, on the
other hand, are explained by reference to the specific semantic aspect of the
stem, with, however, no further indication of in what respect this semantic
aspect is specific so as to exclude for example formation of an adverb.
Obviously, in this form this explanation is a postulate just as well, and
dependent on the first one at that 1 8 •
De Groot's theory has been applied to Latin adverbs by Bos (1967). The
same arbitrariness in postulating a categorial and specific semantic aspect
can be found there. Thus, the fact that bene ('well') and other so-called
adjectival adverbs modify adjectives has to be attributed, according to Bos
(1967: 117), to a specific semantic aspect of 'intensity' (cf. 4.2. end,
below).

2.4. Rank and number of criteria. Overlapping classes

If one were to extend the number of criteria ad infinitum, one would

ultimately obtain classes with only one member. Logically, there is, of
course, no objection to this, but the purposes of classification would not be
realized in this way. This implies that somewhere a boundary must be drawn
17 I have taken the examples from p. 480 and added translations of the Latin words.
By the way, parve occurs rarely: Vitro 9.4.5. parve per eos flectitur delphinus ('the
dolphin moves dimly among them ').
18 De Groot did not regard all traditional adverbs as adverbs. For example, etiam
C'also'), quoque C'too '), fortasse ('perhaps'), certe ('certainly'), non ('not') are called
'determining particles, which may determine all kinds of words'. Nonetheless he
considers his remarks about the relation between adverbs and adjectives valid for a
variety of adverbs, which have different morphological, syntactic and, intuitively,
semantic characteristics. Though he has continued in his belief in a categorial semantic
aspect, his position with respect to adverbs seems to have changed (1964: 260): 'Thus,
under the term 'adverb', various words are grouped together which do not share either
valence or meaning' (my translation). In this connection he refers to Roose (1964).
 Rank and number of criteria. Overlapping classes 31

between categorial and individual properties. The latter will be referred to

the lexicon. It is not only problematic how far classification should be
extended, but also which criterion should be so to say the first. For
example, we might extend the analysis by taking modification by 'adverbs
of degree' such as valde ('very much') as a final criterion for defining
adjectives. In that case numerals cannot be said to be adjectives (cf. note
15). On the other hand we might set up in English a separate class of words
like run, as e.g. Wheatley (1970: 52) thinks feasible (he calls them 'nerbs').
Usually two words run are distinguished, one belonging to the class of
nouns, the other to the class of verbs. In practice a separate class of 'nerbs'
would hardly be useful (Gleason 1965: 124), since we need the two other
classes anyway and since the difference noun vs. verb seems of more
importance than the fact "that some words (or word-forms) belong to both
classes.
The question arises, therefore, which criteria are to be considered
relevant and which not. We need 'some way of ranking the criteria' (Crystal
1967: 41). One type of ranking would be the assignment of a certain order
to criteria from the four levels, as has been discussed in 2.3.1. Leaving levels
out of account, choice of criteria can be made on the basis of two
considerations: (1) those criteria are most important which apply to most
cases (Crystal 1967: 45); (2) those criteria are most important which have
most systematic impact, Le. can be shown to be connected with and predict
other phenomena. For example, inflection of the nominals is more
important than the fact that diminutives can be derived from them, since
inflection is connected with the syntactic characteristics of the nominals (cf.
Carvelland-Svartvik 1969: 37_8)19.
Since the assignment of a word to a certain class is determined by its
sharing certain characteristics with other words and since a class may be
seen as the 'extension' of a certain number of certain characteristics
(Hempel 1965: 139), it is possible that words can be assigned to one class
on the basis of some characteristics, to one or more other classes on the
basis of other characteristics. There is no objection to this since the criteria
on the basis of which the words are assigned to more than one class do not
overlap. This implies that Bloomfield's remark (1935: 196) that 'However,
it is impossible to set up a fully consistent scheme of parts of speech,
because the word-classes overlap and cross each other' is incorrect:

19 Van Wijk (1967: 224) observes that 'there seems ... to be no syntactic principle
on which the choice of criteria can be based', but could one expect a syntactic
principle for ranking syntactic criteria?
32 Problems of classification

consistency of classification and overlapping classes do not exclude each

other.
As an example one might think of the class of Latin words like intra
('inside') which share characteristics with prepositions on the one hand
(governing case forms), with certain adverbs, e.g. intus ('inside'), on the
other (cf. ch. 9). An example given by Van Wijk (1967: 260) is the infinitive
which can be assigned to the class of verbs on some grounds, to the category
of nominals on others. Crystal cites as an example adverb and adjective in
English. 20

2.5. Ontological, psychological and notional definition of word-class

I have not entered into a discussion of a number of studies in which the

authors endeavour to establish the categories of things in extralinguistic
reality of which the word-classes are supposed to be the linguistic
counterparts - 'ontological' in De Groot's terminology (1948: 429). In
these studies, e.g. in Otto (1928) notions like 'Dingwort' frequently occur.
Nor have I discussed studies in which it is held that word-classes reflect the
categorial knowledge the speaker has of the world ('psychological'), e.g.
Slotty (1929a, b), with reference to psychological investigation 2 1. These
studies are quite speculative and have no practical value for writing a
grammar or for the classification of linguistic elements. Lyons (1966) shows
some points of resemblance to these studies.

2.6. Conclusion

Words can be assigned to word-classes on the basis of certain criteria.

Conversely no class should comprise words which are not in accordance

20 The difference between infinitives and words like run, words like intra and adverbs
and adjectives (in English) is that the infinitive has both the properties of a verb and of
a nominal at the same time. For example, in hoc ipsum nihil agere (lit.: 'this very doing
not~ing'; Cic. de Orat. 2,24), agere governs a noun phrase as its OBJECT (nihil) and is
modified by a pronoun as its ATTRIBUTE (hoc). (More examples in Kiihner-
Stegmann: I, 666.) With a word like English run, Latin intra, it is a matter of
either-or. We can distinguish two homophones run and intra, but no homophones
agere. This fact suggests, of course, that the infinitive should be assigned to an
intermediate category of its own.
21 See also in the bibliography Brinkmann (1950), Hempel (1954), Hermann (1928),
Sandmann (1939).
 Conclusion 33

with these criteria. In this respect the traditional parts of speech, especially
the class of adverbs, need revision. The criteria by which word-classes are
defined may belong to all four linguistic levels. These levels do not show, as
a rule, a one-to-one correspondence. There is no objection to assigning a
word to more than one class if it has more than one g,roup of characteristics.
Most important in the total description of a language is a classification in
accordance with syntactic characteristics. A classification of words on the
basis of their semantic aspect is difficult. The determination of the semantic
aspect of a class which has been defined by other, not semantic, criteria
seems a hopeless task. Therefore, it seems not necessary to revise the
traditional efforts to find such categorial semantic aspects.
What is most striking in the traditional class of adverbs is the (1) absence
of formulation of criteria; (2) absence of syntactic criteria; (3) assignment
of words to classes on semantic grounds, though a procedure of semantic
classification is absent.
These points will become clear, I hope, in the following chapters.
3. ADVERBS IN ROMAN GRAMMATICAL THEORY

This chapter will be devoted to what Roman grammarians have said about
the adverb. It is not my intention to present an exhaustive discussion of the
different opinions held by each grammarian, of their sources or mutual
relations. Those who are interested may consult Jeep (1893: 268-82). I will
give only what seems to be relevant to the modern view on adverbs and
confine myself almost entirely to Priscian's Institutiones grammaticae and
to Charisius' Ars grammatical..
The definition of the adverb given in most modem handbooks is based to
a large extent on what Priscian, Charisius and others have said about it. In
their turn, they were heavily indebted to their Greek predecessors. In fact,
the very term 'adverbium' is a translation of Greek epirrhema 2 • For the
1 Latin grammarians are cited after the edition supervised by Keil, Charisius also after
the edition by Barwick, henceforth abbreviated K. and B., respectively. If no further
references are given 'Priscian' means 'Priscian in his Institu tiones'.
2 When the Roman grammarians translated the Greek term by adverbium, they
apparently interpreted 'epirrhema' in a neutral sense: 'what is placed with the verb'. It'
is possible, however, that epirrhema was interpreted in a more specific sense by some
Greeks at least: 'what follows the verb'. Prof. Ruijgh suggested this to me, when we
were discussing Apollonius Dyscolus' definition of epirrhema and especially the
precise meaning of €1rLAeyop,€VoV in this definition, which runs: €1r{PPllP,a earl, P,€POf;
AD-yOV aKALTov KaTa p~p,aTof; Aeyop,€VoV ~ €1rLAeyop,€VoV PrllJ,aTL (p. 7i, ed. Uhlig).
Robins' translation (1967a: 33-4; 'part of speech without inflection, in modification of
or in addition to a verb') reflects the neutral interpretation of the prefix ~1rL-. Perhaps
€1rtAeyop,€VoV ;nl p, aT 1" a kind of gloss to explain l1rL- in €1ripPllP,a, should be translated
'following a verb'. That a more specific interpretation along these lines was current
among Greek scholars appears from the following remark of the scholiast on Dionysius
Thrax (p. 271, ed. Hilgard): €t Se Kat 1rPOTar:'€rat Kat V1rOraOO€Tat, 1rWf; U1rO P,t,Q,f;
aVVTa~€Wf; p,OVOlJ wvop,ao(Jll; ('If it (the adverb) is both prepositive and postpositive,
how could it have been called after one construction only? ').

35
36 Adverbs in Roman grammatical theory

chapter of Priscian, which I take here as a base, one may readily find almost
verbatim parallels in e.g. Apollonius Dyscolus 3 • However, I will neglect
these.
Priscian and others endeavour to distinguish between the 'parts of
speech' by giving the semantic aspects of each class (Priscian 2,4,17:
'proprietates significationum'), following Apollonius (Robins 1951: 64-5;
1967a: 57), but, in fact, most definitions are mixed (Schopf 1963: 64).
They contain information about syntax, semantics and morphology, and we
find stray remarks about accent as well. I will now first discuss the
definitions and dispersed general observations under these headings, next
discuss a few specific subclasses and individual adverbs and, finally, add a
remark about the primary importance of word-classes in Roman grammar.

3.1. General properties ofadverbs

3.1.1. Syntax
The adverb is placed with a verb (Prisc. 2,4,20), 'more suitably' (aptius)
before it, just as adjectives are placed before substantives (15,6,39;
16,3,16)~ Adverbs may be postponed, Priscian observes, with the exception
of all monosyllables such as non ('not'), ne ('that not'), dum ('while'),
temporal cum ('when'), affirmative per ('very'), and vel, when meaning
valde ('very'). There are several subclasses, too, that are usually placed
before the verb: adverbia demonstrativa like en, ecce ('look'), interrogativa
like cur, quare, quamobrem ('why'), adverbia' hortativa like heia, age ('come
on'), adverbia similitudinis like quasi, ceu, veluti ('like', 'as it were'),
adverbia vocandi like heus ('hey! '), adverbia optandi like 0, utinam ('would
that').
That it was c9nsidered norma~ for adverbs to be placed before the verb.
may also appear from Charisius (233, 27B. = 181, 17K).), who reports a
definition of the adverb as the word-class which precedes a verb ('praeposita
verbo'). The same grammarian enters into a discussion about what term
would be more suitable: 'praeverbium' or 'adverbium' (252, 21B. =
194,14K.).

3 Cf. Prisc. 17,1,1 on his relation to Apollonius.

4 Adjectives were regarded in antiquity as a subclass of the nomen. They were called
adiectiva nomina as opposed to appellativa nomina and propria nomina ('substantives'
and 'proper names'; cf. Robins 1951: 41; 1966: 16). I use modern terms in the normal
modern sense. If I want to stress in this chapter the different use of terms by Roman
grammarians, I will use inverted commas and/ or Latin terms.
 General properties of adverbs 37

The relation that exists between adverb and verb is considered the same
as the one found between adjective and substantive (15,1,1; 17,5,37). This
was a common idea. The Stoics, Priscian reports (2,4,16), called adverbs 'as
it were adjectives of verbs' (quasi adiectiva verborum). The meaning of the
adverb is added (adicere, cf. nomen adiectivum) to that of a verb, or, as we
would say with a similar vagueness, the adverb modifies a verb (cf. ch. 6).
The parallelism between prudens homo and plUdenter agit ('the wise man' -
'acts wisely') seemed to confirm this idea. This parallelism still plays an
important role. Pattier (1962: 52ff.) even uses the terms 'adjectif de 'verbe'
and 'adjectif de substantif. In the Roman theory it can be found as early as
Varro L. 8,12, within a different framework originating from Aristotle (see
Dahlmann a.1.). He observes that within the category of noun and verb we
can distinguish primary and secondary members, homo and doctus, scribit
and docte ('man' - 'wise'; 'writes' - 'wisely'), respectively.
As for the relation of adverbs to members of other categories, adverbs
differ from prepositions in that they mayor may not be constructed with
certain case forms (14,1,4), e.g. intra ('inside'). A word should be called a
preposition if it cannot occur without a case form (15,5,30). Charisius
seems to be more practical: if such words govern case forms they are
prepositions, otherwise they are adverbs (245, 19B. = 189, 15K.)5 .

3.1.2. Semantics
As has been said, the meaning of the adverb is added to that of the verb.
The adverb expresses the quality or quantity or number or time or place of
the verb (17,5,38: 'quemadmodum officio adiectivi fungunter adverbia ad
significandum verborum qualitatem vel quantitatem vel numerum vel
tempus vel locum'). The adverb has no 'complete meaning' (2,4,20:
'perfectam significationem'; 15,1,4: 'plenam sententiam'), except when
added to a verb or participle, whereas the verb can have complete meaning
without an adverb. In those cases where one uses adverbs without a verb,
information from context and/or situation is understood: 'it is necessary
that they either refer to verbs said before by another person' (e.g. bona est
superbia? :: non ('is arrogance good? :: no')) 'or that they are said
elliptically' ('per ellipsin'), e.g., if someone tells a story one may say bene

5 Elsewhere Priscian is prepared to assign a word to more than one class, e.g. ut,
quasi and si to the class of adverbs and of 'conjunctions' (15,6,34-5).
38 Adverbs in Roman grammatical theory

('well'), diserte ('clever'), eloquenter ('eloquent'). In that case dicis ('you

speak') is understood ('subauditur')6 _
Adverbs are distinct from interjections. The latter do not only occur
without a verb, but also do not need a verb (of emotion) to be understood.
The expression of a particular .emotion seems to be the very meaning of
interjections (prisc. 15, 7,40; Charis. 246, 27B. = 190, 14K.). In this respect
the Roman grammarians differed from Greek scholars, who did not
recognize interjections as a class of their own. Of course, these statements
(both the one in the preceding paragraph, and this one about interjections)
can easily be translated into a syntactic statement. We might say that
adverbs have restricted sentence valence (that is, the ability to occur in a
one-word-sentence under certain conditions), whereas interjections have
sentence valence obligatorily. I will return to this problem in ch. 8.
Though it is maintained by Priscian and others that the adverb has no
complete meaning it should be recalled that, according to Priscian (IS,S ,31 ;
14,2,12) adverbs differ from prepositions in that they have a definite
meaning of their own ('per se habent aliquam certam significationem'),
whereas if one would say de or in one does not know what is meant unless
they are constructed with some word, e.g. de partibus orationis ('about the
.parts of speech'), de loco in locum ('from place to place').
As we have seen, a distinction is made by Priscian and others between
different types of semantic aspect. Verbs (and nouns) may be said to have
an independent semantic aspect, adverbs (and adjectives) on the other hand
have a dependent semantic aspect. This bears some resemblance (if we
ignore temporal and local adverbs which have deixis) to Reichling's
treatment of the semantic aspect of noun and adjective (University lectures)
6 Another version of the completeness of meaning is attributed to Palaemon by
Charisius (241, 20B. = 186, 30K.). In his version it is claimed that with verbs adverbs
are obligatory. The definition and associated explanation run: 'adverbium est pars
orationis quae adiecta verbo significationem eius explanat atque implet (cf. Prise.
Partit. 474, 26K.). Ita nam cum dico, Palaemon docet, nondum significo satis vim
planam verbi, nisi adiecero bene aut male' ('The adverb is a part of speech which is
added to the verb and clarifies and completes its meaning. For if I say it like this,
Palaemon observes, I do not yet make the meaning of the verb sufficiently clear, unless
I add well or badly'). This is probably due to a misunderstanding of Appollonius
Dyscolus' definition "EaTw obv €1r{pprll..La Jl€V A€~L<; l1.KALTO<; KaTl1'Yopovaa TWV Jv Tot<;
PllJ..Laaw ~'YKA[aEwv Ka{)oAov ~ JlEPLKW<;. :£v llvEv o~ KaTaKAE£aEL DulvoLav. ( - 1TEpl
€1rLPPl1JlaTWv 529b, p. 119,6 ed. Schneider). ('The adverb is an invariable part of
speech, which modifies the verb forms entirely or partly without which it (the adverb)
does not complete its meaning'). The same mistake is made in Greek theory as well. Cf.
Uhlig's comment ad Dionysius Thrax 641b, p. 72, Uhlig. For modern views on
obligatoriness of adverbs cf. 6.1. 2. 2. below.
Random documents with unrelated
content Scribd suggests to you:
 SOUTHALL BROS., Bull Street, BIRMINGHAM.
Wholesale Agents: SHARP, PERRIN & Co., LONDON.
For protection against useless and injurious imitations, the label on each packet
bears the signature of the Patentees.

PEPPER’S
QUININE and
IRON TONIC.
HEALTH. STRENGTH. ENERGY.
Great bodily strength.

Great nerve strength.

Great mental strength.

Great Digestive STRENGTH follows the use of PEPPER’S QUININE AND IRON
TONIC.
Bottles, 32 doses. Sold by Chemists everywhere. Refuse Imitations. Insist on having
Pepper’s.

PEPPER’S
TARAXACUM &
PODOPHYLLIN.
A Fluid Liver Medicine, made from
DANDELION AND MANDRAKE ROOTS.
Good for Liver Disorder & Indigestion.
The best Antibilious Remedy.
Without a particle of Mercury.
Safest & surest Stomach & Liver Medicine.
Clears the Head and cures the Headache.
Regulates the Bowels.
Bottles, 12 doses. Sold by most Chemists. Decline imitations; many Chemists
professing their own to equal Pepper’s renowned Liver Preparation.

SULPHOLINE
LOTION.
THE CURE FOR SKIN DISEASES!
IN A FEW DAYS
Eruptions, Pimples, Blotches entirely
fade away.
BEAUTIFULLY FRAGRANT. PERFECTLY
HARMLESS. CURES OLD-STANDING
SKIN DISEASES. IT REMOVES EVERY
KIND OF ERUPTION, SPOT, OR BLEMISH,
AND RENDERS THE SKIN CLEAR,
SMOOTH, SUPPLE, AND HEALTHY.

Sulpholine Lotion is sold by Chemists.

Bottles, 2s. 9d.

LOCKYER’S SULPHUR
HAIR RESTORER.
The Best. The Safest. The Cheapest.
Restores the Colour to Grey Hair.
Instantly stops the Hair from fading.
Occasionally used, Greyness is impossible.
Where the Sulphur Restorer is applied scurf cannot exist, and a sense of cleanliness, coolness, &c.,
prevails, which cannot result from daily plastering the hair with grease. Sold everywhere, in large
bottles, holding almost a pint, 1s. 6d. each. Be sure to have Lockyer’s.

“THE BLOOD PURIFIER.”

Old Dr. Jacob Townsend’s

SARSAPARILLA
Has been long used by the Medical Profession in all
Skin and Blood Diseases. Pimples, Gout, Scurvy, Sores
on the Neck and Legs, Dropsy, Impaired Health, and
for cleansing the system of all impurities, which, when
suffered to remain, surely destroy life.
G. C. Kernott M.D., London, says:—“I strongly
recommend it in cutaneous diseases and all impurities
of the blood.”

Read the following letter of a distinguished Churchman:—“The Hon. the Dean

of Lismore requests Dean, Steel & Co. will send him two bottles of their Old
Dr. Jacob Townsend’s Sarsaparilla. The Dean has no objection to their
publishing that he has found their Sarsaparilla very useful in his family.”
Sold in Bottles, 2s. 6d., 4s. 6d., 7s. 6d., and 11s., by all Chemists, &c.

OLD DR. JACOB TOWNSEND’S SARSAPARILLA

PILLS
Are highly recommended for Bilious Affections, Indigestion, Liver, and Stomach Complaints.
1s. 1½d., 2s. 9d., and 4s. 6d., of all Chemists.

Chief Depot:—DEAN, STEEL & Co., 131, FLEET STREET, LONDON.

TURNER’S TAMARIND
COUGH EMULSION.
“It is a Pleasant and
Elegant
 Preparation.”
Extracts from Letters.
“I know nothing equal to it for distressing Coughs.”
“I believe the best medicine for the Throat & Lungs.”
“The Tamarind has been quite a boon to me.”
“Invaluable to Speakers and Singers.”

Thirty Drops on Lump Sugar.

1s. 1½d. and
2s. 9d.
per bottle.

Saving 7½d.
in larger
size.
All Testimonials guaranteed Truthful. A 2/9 bottle per parcel post, carriage free.
AGENTS—Barclays, 95, Farringdon Street; Hooper, London Bridge; Sangers, 489,
Oxford Street; Duncan, Flockhart & Co., Edinburgh; Apothecaries Co., Glasgow. All
Wholesale Houses, and any Pharmacist or Chemist in the kingdom.

J. A. TURNER, Pharmacist, LIVERPOOL.

BEWARE
OF IMITATIONS!

BROWN’S PATENT
“DERMATHISTIC” CORSET
BONES, BUSKS AND SIDE STEELS PROTECTED BY
LEATHER.
 The QUEEN, November 17th, says: “‘There is nothing like
leather,’ and the novel idea of covering with kid those parts
which wear out first, seems a most practical idea. The
Dermathistics are shapely, neatly sewn, and the leather adds
but little to their weight.”
MYRA, December 1st, says: “An ingenious method for
ensuring durability. The leather adds in no way to the bulk,
while it gives a decided added support to the figure,
besides preventing wear. They are very comfortable.”
YOUNG LADIES’ JOURNAL, May 1st, says: “The Dermathistic
Corset is elegant in form, light in weight, and marvellously
strong, and is particularly adapted to Ladies who ride, play
SOLD EVERYWHERE. at Lawn Tennis, or who are fond of Boating.”
Avoid Worthless
Imitations. The WAREHOUSEMAN AND DRAPER, March 1st, says: “The
Dermathistic has been steadily growing in favour since its
introduction; its merits are not, however, even yet so well
known as they deserve to be, and in many cases Drapers,
we think, might with advantage introduce it more freely, for the very durable
appearance of this shapely and really serviceable Corset could not fail to prove an
attraction to their customers.”

BLACK AND ALL COLOURS,

No. 1. No. 2. No. 3. No. 4.
5/11. 7/11. 10/6. 15/6.

EVERY PAIR STAMPED.

For Measurement: Circumference of Abdomen and Hips.

B AILEY’S
Papers.
PATENT ABDOMINAL BELTS.—Highly commended by all the Medical
Several hundred unsolicited testimonials have been received from
Medical men and others. Undoubtedly the greatest improvement ever effected. The
hips are free. “Cannot shift or ruck up.” Self-adjusting. Price 45s., 35s., 25s.—
Address the Superintendent, the Ladies’ Department. New Catalogue free.
B AILEY’S ELASTIC STOCKINGS.—Accurately fitted, upon which the utility of these
articles entirely depends. Strong, light, and porous. Cotton, 5s., 6s. 6d.; Silk, 7s.
6d., 10s. 6d., 14s. 6d., 17s. 6d. each. For measurement send the circumference at
calf, ankle, and instep. New Catalogue free.

B AILEY’S TRUSSES.—Covered in Gum Elastic, indestructible, perfectly impervious,

and very cheap, suitable for Infants or the bath. (The necessity of wearing a
Truss, especially in a warm bath, is not generally understood.) Trusses with or
without springs. Every known description manufactured on the premises. Trusses
repaired and recovered. The most difficult cases are courted. New Catalogue free.

B AILEY’S IMPROVED CHEST-EXPANDING BRACES.—Invaluable

children. Price 12s. 6d. State age. New Catalogue free.
for growing

B AILEY’S AIR AND WATER BEDS.—On Sale or Hire. Crutches, Enema Apparatus,
&c. New Catalogue free.

W. H. BAILEY & SON, 38, Oxford Street, London,

W.

AT A GREAT SAVING TO THE

PURCHASER!
Ladies, send letter or post card, and you will
receive, POST FREE, Sample Patterns, with Prices,
of all the LEADING NOVELTIES OF

The B. M. Co., by trading direct with the public,

have effected a revolution in the Styles and Fabrics of Dress Materials. Carriage
Paid to any part of the United Kingdom on all orders over £1. The Century
Cashmeres, as exhibited at the Health Exhibition, are in ever-increasing demand.
Be particular to address in full; please write at once, mention John Bull’s
Womankind.

REGISTERED TRADE MARK.

For over Forty Years the Medical Profession have approved of this pure Solution as
the best remedy for Acidity of the Stomach, Heartburn, Headache, Gout and
Indigestion; and as the safest aperient for delicate Constitutions, Ladies, Children
and Infants.
SOLD BY CHEMISTS THROUGHOUT THE WORLD.

CAUTION—See that “DINNEFORD & Co.” is on every Bottle

and Label.

The Best, Quickest, & Most Agreeable Remedy for

RHEUMATISM, BRONCHITIS, SORE
THROAT, NEURALGIA, LUMBAGO, &c., IS

Smedley’s Chillie Paste.

IT GIVES INSTANT RELIEF AND DOES NOT BLISTER.
If applied when the first symptoms appear, one application will usually effect a
complete cure and avert what might otherwise be a severe illness.
In 1s. 6d. and 2s. 9d. Bottles of all Chemists, or post free from the Sole Proprietors,
HIRST, BROOKE & HIRST, LEEDS.

SMITH’S COLCHESTER DIGESTIVE

OR
 LIVE-LONG
CANDY.
No Household should be without this most valuable and Palatable Digestive Stimulant.
A small piece taken after a meal is an effectual Preventative to Indigestion, or taken
at bed-time, by its warmth-giving and Stomachic properties, Promotes Sleep. A small
piece dissolved in the mouth when exposed to damp and cold Warms the Chest, and
prevents those injuries which arise from Chills. It is invaluable to all Sportsmen.
The Marquis of Waterford writes (in 1884):—“I find your Candy most useful.”
At 1s. 1½d., 1s. 9d., and 4s. 6d. (post, 1s. 4d., 3s., and 5s.), of all Chemists.
Be careful to buy only that prepared by
J. C. SHENSTONE, Manufacturing Chemist, COLCHESTER.

SOLD BY THE PRINCIPAL DRUGGISTS.

JACKSON’S A sparkling means of incensing a domicile, and of exorcising
evil smells.
INCENSE
An Enchanter’s little wand, that on being fired becomes to
SPILLS. the receptive as a Medium which quickens the fancy, be its
mood grave or gay, kindly leading the captive to that ladder,
the top of which reaches through the clouds to the borders of
Fairyland.
IN BOXES AT 6d., OR BY POST SEVENPENCE.
JACKSON’S Surpasses in neatness, in strength, in cheapness, and retains
its virtues in all climates. It has stood the test of time, and in
CHINESE all quarters of the Globe.
DIAMOND
CEMENT. In Bottles at 6d. and 1s., by post 1/2.
JACKSON’S For the removal of Hair from the Arms, Neck,
RUSMA. or Face, as well as Sunburn or Tan from the
Skin.
The activity of this depilatory is notable. It works without
noise. It leaves a whole skin and a clean complexion.
In Bottles at 1s., by post for 1/2.
FROM THE LABORATORY OF
THOMAS JACKSON, Strangeways, MANCHESTER.

Window Blinds! Window Blinds!

 Use the Patent Metallic Enamelled
Venetian Window Blind.
ALWAYS CLEAN. PREVENTS INFECTION. CHEERFUL.
MORE DURABLE THAN ANY OTHER.

By Washing can be made to look Equal to New after Years of

Wear.
Send for Testimonials, Prices, and Estimates to

Hodkinson & Clarke, Limited,

WHO ARE MANUFACTURERS OF
The Latest Improved and Most Artistic Window Blinds.
Venetians, Rollers, Cane, and Stained and Painted Glass Blinds.

Canada Works, 2, Chiswell St.,

Small Heath, Finsbury,
BIRMINGHAM. LONDON.

Minster Buildings,
12, Church St.,
“WHEN GEORGE THE FOURTH WAS KING,” LIVERPOOL.
Gold
Medal,
Adelaide,
1881.

Needham’s
Polishing
Paste.
Order of Merit,
Melbourne, 1880.
 Diploma of Merit,
Vienna, 1873.
Was first
introduced to
the Public.
It is the Oldest
and Best Preparation in the Market for Cleaning and Polishing

BRASS, COPPER, TIN, BRITANNIA METAL,

ETC.
Pickering’s Furniture Polish, Plate Powder, Knife
Powder, Brunswick Black, Razor Paste.
SOLD BY CHEMISTS, GROCERS, IRONMONGERS.

Joseph Pickering & Sons, SHEFFIELD.

“JOHN BULL’S WOMANKIND”

who buy Cod Liver Oil should always have Peter Möller’s.

Peter Möller is the Inventor of the Process used, and Manufacturer of

Cod Liver Oil only. His attention is given solely to the preparation of this
invaluable medicinal food in its highest state of perfection, and the measure of his
success is shewn by the award of NINETEEN FIRST PRIZES at FIFTEEN
EXHIBITIONS. PETER MÖLLER was THE ONLY MANUFACTURER to whom TWO
GOLD MEDALS were awarded at the International Fisheries Exhibition for Cod
Liver Oil.
It is sold in Capsuled Bottles only by Chemists, Grocers, &c.
 Newest
Invention! THE Y & N PATENT
Greatest
Novelty! DIAGONAL SEAM CORSET.
Patented in England and on the Continent.
Will not split in the Seams nor tear in the
Fabric. Exquisite Model. Perfect Comfort.
Guaranteed Wear.
MADAME MARIE ROZE writes:—“I have very much pleasure
in stating that the two pairs of Corsets you have made for
me are a great success. They fit perfectly, and are far
superior in every way to all the English and French Corsets I
have tried.”—Yours truly, MARIE ROZE MAPLESON.
LE FOLLET says:—“A novel invention in Corsets admirably
calculated to prevent the very disagreeable occurrence of
split seams. The cut is very good and becoming, and may
be adapted to any figure with advantage.”
THE QUEEN says:—“These Corsets are a new departure. The material is cut on the
cross, and the component parts being also arranged diagonally, the seams have no
strain. They are admirably modelled, exquisitely neat and strong, and the
workmanship all that could be desired.”
Beware of worthless imitations.
Every genuine Y & N Corset is stamped “Y & N Patent Diagonal Seam Corset, No.
116” in oval.
Gold Medal, New Zealand Exhibition, 1882; GOLD MEDAL, HIGHEST AWARD FOR
CORSETS, LONDON INTERNATIONAL EXHIBITION, 1884.
Sold by all Drapers and Ladies’ Outfitters.

Corner’s
Patent
 Cinturon de Cuero
CORSET.
L ADIES who appreciate a GOOD FITTING CORSET,
that will bear any strain without stretching in the
waist, or becoming unshapely through wear, will find
that the “CINTURON DE CUERO,” or Leather Waist
Corset, has been admirably designed to meet both
these requirements.
The Leather is used in such a way that it forms a Band
or Belt, and, while rendering the waist perfectly
unstretchable, imparts a charming sense of comfort to
the wearer; the original size and shape of the Corset
being thus always retained, gives an elegant
appearance to the figure, and considerably increases
the durability.

BLACK AND ALL COLOURS,

5s. 6d. to 15s. 6d.
From Drapers and Ladies’ Outfitters throughout the Kingdom, through the principal
Wholesale Houses.

SWEARS & WELLS’

CELEBRATED OUTFITS FOR SCHOOLS AND COLLEGES.

M ESSRS. S. & W. (Prize Medallists Paris Exhibitions, 1867 and 1878) respectfully
draw the attention of Parents and Guardians to the unrivalled facilities their
Establishment affords for fitting out Young Gentlemen for Private and Public Schools
and Colleges, as the Stock on view comprehends every requisite first-class article of
attire and utility, and the well-known reputation of the Firm as Juvenile Outfitters is
a guarantee that the goods, specially selected and enumerated in the following list,
possess all the necessary qualifications for their various purposes:
Socks
Knickerbocker Hose
Underclothing
Bathing Drawers
Sponges and Bags
Shirts
Collars, Cuffs
 Collar Boxes
Gloves
Boots and Shoes
Travelling Rugs
Portmanteaus
Ties, Scarfs
Neckerchiefs
Scarf Pins
Studs, Links
Braces, Belts
Hand Bags
Towels, Sheets
Pillow Cases
Leggings
Pocket Handkerchiefs
Umbrellas
Knives
Suits for Best Wear
Suits for School Wear
Suits for Cricketing
Suits for Boating
Suits for Football
Suits for Riding
Pea Jackets
Overcoats
Dressing Gowns
Hats, Caps
Hat Cases
S. & W.’s Students’ Celebrated Solid Leather Dressing Cases, with Warranted
Fittings,
27s. 6d. and 34s. each.
Inventory of Clothes required for the various Public Schools, Fashion Sheets,
Patterns of Materials, and Measurement Forms forwarded Post Free.

NEW DEPARTMENT FOR BOOTS AND SHOES.

(Boots and Shoes for Infants, Children, Ladies and Gentlemen.)

MESSRS. SWEARS & WELLS, in compliance with the continuous request of many of
their Patrons, have the gratification of announcing to their Customers the opening
of a SHOW ROOM for the Sale of Boots and Shoes. This arrangement completes
their series of Departments, and thereby enables them to supply entire Outfits for
Schools and Colleges, also for their numerous connections in India, the Colonies,
the United States, and Foreign Countries.

DEPARTMENTS:
Hosiery (Children’s, Ladies’, and Gentlemen’s), Glove (Children’s, Ladies’, and
Gentlemen’s), Shirts and Collars (Children’s and Gentlemen’s), Tie and Scarf,
Parasol, and Umbrella; Juvenile Tailoring, Hat & Cap; Girls’ Costume; Infants’ and
Ladies’ Underlinen; Children’s, Ladies’, and Gentlemen’s Boots and Shoes.

192, REGENT STREET, W.

MORSONS’
PREPARATIONS OF

Highly recommended by the Medical Profession

As Wine in Bottles at 3s., 5s. & 9s.; Lozenges, 2s. 6d. and 4s.
6d.; Globules, 2s., 3s. 6d., and 6s. 6d.; and Powder as
“Medicinal Pepsine,” at 2s. 6d. and 4s., and “Porci,” a more
concentrated preparation than the Medicinal, at 4s. 6d. each.
Sold by all Chemists.
Pepsine has acquired as
HE popularity almost a specific for Chronic
Dyspepsia, Indigestion, &c., is due to the fact that it is the nearest
possible production of the active principle of the gastric juice of the stomach.
Unfortunately, like all other inventions of a like nature, Pepsine has been not
slightly discredited by the spurious Manufactures that have been issued from time
to time; it is therefore necessary as a guarantee of its efficacy to see that each
bottle bears the Makers’ name.
 THE MOST EFFICACIOUS TONIC.

TASTELESS, AND
NEITHER STAINS
NOR
INJURES THE
TEETH.
———

DOES NOT
CONSTIPATE or
DISTURB Sold by all the PRINCIPAL
CHEMISTS and DRUGGISTS,
THE DIGESTION. in Bottles in portable Card
London Medical Record,
Cases, with Drop Measure
March 15, 1877, says:
complete, 3s. and 4/6 —“‘Bravais’ Iron’ is
each. tasteless, free from
Pamphlets, with full Particulars and Testimonials, Post styptic character, and
Free on application to the appears in the most
simple state of
Agency and Wholesale Depot, 8, IDOL combination, that is to
LANE. LONDON, E.C. say, merely united with
oxygen and water,
without the presence of
acids. It is a most
energetic preparation. It
is the beau ideal of a
ferruginous tonic. We
regard it as a therapeutic
of great value.”
Invaluable in all cases of
general weakness or
debility, and is taken
 with the greatest facility
on a small piece of sugar
or bread, or in a glass of
wine before meals.

“Is as nearly tasteless as Cod Liver Oil can be.”—Lancet.

“Has almost the delicacy of salad oil.”—British Medical Journal.
“No nauseous eructations follow after it is swallowed.”—Medical Press.
It can be borne and digested by the most delicate; is the only oil which does not
“repeat,” and for these reasons the most efficacious kind in use.

In Capsuled Bottles only, at 1/4, 2/6, 4/9, and 9/. SOLD

EVERYWHERE.

ALLEN & HANBURYS, Plough Court, Lombard

Street, LONDON.
The
“Nonpareil”
is the richest,
softest, and
most becoming
Fabric ever
produced, and is
pre-eminently
suited for
Ladies’ indoor
and outdoor
Costumes, Boys’
Suits and
Children’s Dress.
Of all Drapers
everywhere.
Every yard is
stamped on the
back

The finer
Can be purchased of all leading retailers at from qualities are
2s. to 6s. per yard. equal in
appearance and
Wholesale Agents: J. H. Fuller, 92, Watling Street, London. wear better than
John R. Taylor, 51, Miller Street, Glasgow. the very best
Lyons Silk
Velvet, and cost
 only a quarter of
the price.

By Special Royal Appointment.

Spearman’s Devon Serges

Thousands of Customers testify that no other article woven equals this in general
utility.
According to the Queen “It has no rival.”

On Sale all the Year round. Pure Wool only. New Colours, Checks
& Mixtures.
For Ladies’ wear, beautiful qualities, 1/6 to 4/6 the yard. For Children’s wear,
capitally strong, 1/3 to 2/ the yard. For Gentlemen’s wear, double width, 2/6 to
10/6 the yard.
The Navy Blues and the Blacks are fast dyes. On receipt of instructions samples will
be sent post free.
N.B.—Any length cut, and Carriage Paid to principal Railway Stations.

SPEARMAN & SPEARMAN, PLYMOUTH.

ONLY ADDRESS. NO AGENTS.
 14, Holborn Viaduct and 5, Lisle Street, Leicester Square,
LONDON.

All the great Road Races for either Bicycles or Tricycles have
been won on Machines of our make.
One Stamp for Catalogues, List of Patrons and Testimonials.

BAYLISS, JONES & BAYLISS.

 SAMPLES AND LOW PRICES
ON APPLICATION.

Catalogues of Solid & Tubular Bar Fencing, Iron Hurdles, Gates, Wire Fencing, Rick
Stands, Chain Harrows, Dog Kennel Railing, Galvanized Wire Netting, &c., &c., free
on application.

BAYLISS, JONES & BAYLISS,

VICTORIA WORKS, WOLVERHAMPTON,
And 3, CROOKED LANE, KING WILLIAM STREET, London, E.C.
Please mention John Bull’s Womankind.

Wm. POLSON’S
CORN FLOUR.
The Original and First Manufactured in Great Britain.

UNRIVALLED FOOD OF HEALTH FOR CHILDREN

AND INVALIDS.
 The Best Known Material for
PUDDINGS, CUSTARDS, CAKES, BLANC MANGE.

Ask for MOIR’S

TABLE JELLIES.
SOUPS IN GLASSES.
CURRY POWDER.
POTTED MEATS, &c.
Purveyors to H.R.H. The Prince of Wales.

TOUGHENED GLASS
CELEBRATED FOR ITS EXTRAORDINARY
STRENGTH.
Proprietors:—
THE TOUGHENED GLASS COMPANY, Limited, 75, LEADENHALL STREET, E.C.

TO THE DAUGHTERS OF JOHN BULL!

Save Time, Labour, Temper, and Money by Using
BRUCE’S OIL COOKING STOVES
 The Cheapest and quickest method of Cooking known.
Absolutely Safe. No Smoke, Smell, Dirt, or Danger.
Portable. No Flues or Fixing. The “Household Friend,”
the latest invention, and most successful Oil Stove ever
offered to the Public. It cooks a joint 10 lbs., dinner for
6, with 3 courses in 3 hours, cost 2½d. Complete,
with ½-gall. Kettle, Saucepan, Steamer, Fry-pan, Meat
Tray and Grid, Funnel and pair Scissors, fitted with 4
large moveable Burners, Indicators, and the new Plate
Warmer to heat 1 doz. Plates, securely packed in
strong Box, 35s. The well-known BAZAAR says: “We
have no hesitation in recommending it; trustworthy, well finished, a marvel of
cheapness, and The Best Oil Stove of its kind in the Market. The Maker, with
great fairness, offers to change it or return the money if not approved of.” Send for
Descriptive Illustrated Price Lists of all kinds of Petroleum, Cooking and Heating
Stoves, and Lamps specially suited for residents in the Country and Abroad, where
the difficulty of obtaining a trustworthy article is so much felt, post free, to any part
of the World. Buy direct of the Maker, and save 30 per cent.

J. B. BRUCE, Wholesale, Retail, and Export,

90, BLACKMAN STREET & 74, LANT STREET, LONDON, S.E.

Annual Sale over Half-a-Million.

SOLD BY

McCall’s ALL
GROCERS.

Paysandu
Ox Tongues.
In various sizes, 1½ to 3½ lbs. in Tins.

Delicious for Breakfasts, Luncheons & Suppers.

 Liebig COMPANY’S
EXTRACT OF MEAT.
The ONLY BRAND
WARRANTED GENUINE
by BARON LIEBIG.
CAUTION!
Numerous inferior and low-priced
substitutes being in the market with
misleading titles, labels, and portraits of
the late Baron Liebig, purchasers must
insist upon having the
Liebig COMPANY’S
Extract of Meat.

THE FINEST
MEAT-FLAVOURING
INGREDIENT.
Invaluable & efficient
Tonic for Invalids.
N.B.—Genuine ONLY with facsimile of Baron Liebig’s Signature, in Blue Ink across
Label.

Cooper, Cooper & Co.

SELL THE FINEST
TEA
THE WORLD PRODUCES AT

Per 3s. Pound.

 And Magnificent TEAS at 2/6 and 2/- a Pound, as supplied to
Princes, Dukes, Marquises, Earls, Viscounts, Barons, Bishops,
and the County Families of the United Kingdom.
Samples and Book about TEA post free on application to
Cooper, Cooper & Co.,
CHIEF OFFICE—
50, KING WILLIAM STREET, LONDON BRIDGE.
BRANCH ESTABLISHMENTS—
63, BISHOPSGATE STREET WITHIN, E.C.
268, REGENT CIRCUS, W.
35, STRAND (near Charing Cross), W.C.
7, WESTBOURNE GROVE, W.
334, HIGH HOLBORN, W.C.
LONDON.

HEAL & SON.

BEDSTEADS.
3ft. IRON FRENCH from 10s. 6d.
3ft. BRASS FRENCH from 48s.
200 FIXED FOR INSPECTION.

BEDDING.
MATTRESSES, 3ft., from 11s.
A NEW SPRING MATTRESS, warranted good and serviceable, at
a very Moderate Price.
3ft., 28s.; 3ft. 6in., 32s.; 4ft., 36s.; 4ft. 6in., 40s.
 THIS WITH A TOP MATTRESS—
3ft., 20s.; 3ft. 6in., 23s.; 4ft., 26s.; 4ft. 6in., 29s.
Makes a most comfortable Bed and cannot be surpassed at the price.

HEAL’S PATENT SOMNIER ELASTIQUE PORTATIF of which

30,000 have been sold, is the best Spring Mattress yet invented,
3ft., 40s.; 5ft., 63s.
GOOSE DOWN QUILT, 1 by 1¼ yds., 10s.
BLANKETS, 2 by 2½ yds., 9s. 6d. per pair.
BEDDING CLEANED & REMADE.

BEDROOM FURNITURE.
PLAIN SUITES from £3.
DECORATED SUITES from £8 10s.
ASH and WALNUT SUITES from £12 12s.
300 SUITES ON VIEW.

SCREENS, suitable for Bedrooms, 21s.

EASY CHAIRS from 35s.

COUCHES from 75s.
DINING TABLES from 70s.
DINING-ROOM CHAIRS in Leather from 24s.
BOOKCASES from 38s. & BOOKSHELVES from 7/6.
WRITING TABLES from 25s.
OCCASIONAL TABLES from 10s. 6d.
CHAIRS & SOFAS RESTUFFED & RECOVERED.
English and Foreign Carpets. A Bordered Seamless
Carpet from 26s.
Illustrated Catalogue of Bedsteads and Bedroom Furniture with Price List of
Bedding, free by post.

HEAL & SON, 195 to 198, TOTTENHAM

COURT ROAD

New Patterns, Post Free, with other Fashionable

Fabrics in all the Newest Tints.

Under the Direct Patronage of the Courts of

Great Britain, Germany, Russia, France,
Austria & Italy.
EGERTON BURNETT’S
ROYAL SERGES
Include the Best Makes of this Indispensable Material, and can be relied on to
stand Wind and Weather on Land and Sea, in Summer or Winter,
for LADIES’, GENTLEMEN’S, or CHILDREN’S WEAR.
They can be had in any Colour or Quality, from the finest and lightest, suitable for
Tropical Climates, to the warm heavy makes capable of Resisting an Intense Degree
of Cold.
 Prices for Ladies, 1/2½ to 4/6 per yard.
Extra Strong for Gentlemen and Boys’ Wear.
(54 in.) from 2/11 per yard.
The QUEEN says:—“It is pre-eminently useful, and recommends it to practical
minds and purses of all lengths.”

Carriage Paid on Orders over 20s. to any Railway Station in

ENGLAND, IRELAND, or SCOTLAND.

EGERTON BURNETT, No. 12, Wellington,

SOMERSET.

Dr. J. COLLIS BROWNE’S

Is the Best Remedy known for Coughs, Consumption, Bronchitis, Asthma.
Effectually checks and arrests those too often fatal Disease known as Diptheria,
Fever, Croup, Ague.
Acts like a charm in Diarrhœa, and is the only specific in Cholera and Dysentery.
Effectually cuts short all attacks of Epilepsy, Hysteria, Palpitation and Spasms.
Is the only palliative in Neuralgia, Rheumatism, Gout, Cancers, Toothache, Meningitis,
&c.
The Right Hon. the EARL RUSSELL has graciously favoured J. T. DAVENPORT with
the following:—“Earl Russell communicated to the College of Physicians that he
received a despatch from Her Majesty’s Consul at Manilla, to the effect that Cholera
had been raging fearfully, and that the ONLY remedy of any service was
CHLORODYNE.”—See Lancet, December 1st, 1864.
CAUTION.—The extraordinary medical reports on the efficacy of Chlorodyne, render
it of vital importance that the public should obtain the genuine, which bears the
words “Dr. J. Collis Browne’s Chlorodyne.”
Sole Manufacturer—J. T. DAVENPORT, 33, Great Russell Street, London, W.C.

AT HOME OR ABROAD,
I ALWAYS HAVE WITH ME
Lamplough’s Pyretic Saline,
Which forms a most Invigorating, Vitalising, and Refreshing Draught.
Drs. PROUT, MORGAN, TURLEY, GIBBON, SPARKS, DOWSING, STEVENS, and many
other Medical Men, have given unqualified Testimony to the importance of the
 discovery and the immense value of
THIS GREAT REMEDY
As possessing elements most essential to the restoration and
maintenance of health, with perfect vigour of Body and Mind.

Headache, Sea or Bilious Sickness,

It gives instant relief in
Constipation, Indigestion, Lassitude, Heartburn, and
Feverish Colds; and prevents and quickly relieves or cures the worst form of
Typhus, Scarlet, and other Fevers, Smallpox, Measles, and Eruptive or
Skin Complaints, and various other altered conditions of the Blood. It is the
cure for Cholera. “It

‘SAVED MY LIFE,’
For the Fever had obtained a strong hold on me. In a few days I was quite well.”—
Extract from Letter of C. Fitzgerald, Esq., formerly Correspondent of the Manchester
Guardian in Albania.

CAUTION.—Dr. Wilson writes:—“We all know how much rubbish is put into the
market in imitation of it.”
In Patent Glass-stoppered Bottles, 2s. 6d., 4s. 6d., 11s., and 21s. each.
To be obtained of any Chemist or Patent Medicine Dealer, and of
H. LAMPLOUGH, 113, HOLBORN, LONDON, E.C.

By
Special Appointments
to H. M. the Queen,

CAMBRIC and H. I. and R. H.

the Crown Princess
POCKET HANDKERCHIEFS of Germany.

ALL PURE FLAX. SAMPLES AND PRICE LISTS POST FREE.

Children’s, hemmed for use, 1/8 per dozen.

Ladies’ 2/11 per dozen.
Gents’ 3/11 per dozen.
 Hemstitched.
Ladies’ 5/6 per dozen.
Gents’ 7/3 per dozen.

“The Cambrics of Robinson & Cleaver have a world-wide fame.”—Queen.

IRISH LINENS.
Real Irish Linen Sheeting, fully bleached, 2 yards wide, 1/11 per yard; 2½ yards
wide, 2/4½d. per yard (the most durable article made, and far superior to any
foreign manufactured goods).
Roller Towelling, 18 in. wide, 3½d. per yd.
Surplice Linen, 8½d. per yard.
Linen Dusters, 3/3; Glass Cloths, 4/6 doz.
Fine Linens and Linen Diaper, 10d. yard.
Samples post free.

IRISH DAMASK TABLE LINEN.

Fish Napkins 2/11 per dozen.
Dinner Napkins 5/6 per dozen.
Table Cloths, 2 yards square, 2/11½d. each.
Table Cloths, 2½ yds. by 3 yds., 6/11 each.
Kitchen Table Cloths 11½d. each.
Strong Huckaback Towels, 4/6 per dozen.

Monograms, Crests, Coats of Arms, Initials, &c., woven & embroidered. Samples
post free.

Robinson & Cleaver, Belfast.

HIGHEST AWARDS WHEREVER EXHIBITED.

Welcome to our website – the perfect destination for book lovers and
knowledge seekers. We believe that every book holds a new world,
offering opportunities for learning, discovery, and personal growth.
That’s why we are dedicated to bringing you a diverse collection of
books, ranging from classic literature and specialized publications to
self-development guides and children's books.

More than just a book-buying platform, we strive to be a bridge

connecting you with timeless cultural and intellectual values. With an
elegant, user-friendly interface and a smart search system, you can
quickly find the books that best suit your interests. Additionally,
our special promotions and home delivery services help you save time
and fully enjoy the joy of reading.

Join us on a journey of knowledge exploration, passion nurturing, and

personal growth every day!

ebookbell.com