Untitled

THE PROMINENCE OF
TENSE, ASPECT AND MOOD

STUDIES IN LANGUAGE COMPANION SERIES (SLCS)
The SLCS series has been established as a companion series to STUDIES
IN LANGUAGE, International Journal,
sponsored by the Foundation “Foundations of language”.
Series Editors
Werner Abraham Michael Noonan
University of Groningen University of Wisconsin-Milwaukee
The Netherlands USA
Editorial Board
Joan Bybee (University of New Mexico)
Ulrike Claudi (University of Cologne)
Bernard Comrie (Max Planck Institute, Leipzig)
William Croft (University of Manchester)
Östen Dahl (University of Stockholm)
Gerrit Dimmendaal (University of Leiden)
Martin Haspelmath (Max Planck Institute, Leipzig)
Ekkehard König (Free University of Berlin)
Christian Lehmann (University of Bielefeld)
Robert Longacre (University of Texas, Arlington)
Brian MacWhinney (Carnegie-Mellon University)
Marianne Mithun (University of California, Santa Barbara)
Edith Moravcsik (University of Wisconsin, Milwaukee)
Masayoshi Shibatani (Kobe University)
Russell Tomlin (University of Oregon)
John Verhaar (The Hague)
Volume 49
D.N.S. Bhat
The Prominence of Tense, Aspect and Mood

THE PROMINENCE
OF TENSE,
ASPECT AND MOOD
D.N.S. BHAT
Central Institute of Indian Languages, Mysore
JOHN BENJAMINS PUBLISHING COMPANY

AMSTERDAM/PHILADELPHIA
TM The paper used in this publication meets the minimum requirements
8
of American National Standard for Information Sciences — Permanence of

Paper for Printed Library Materials, ANSI Z39.48–1984.
Library of Congress Cataloging-in-Publication Data

Shankara Bhat, D. N., 1935-
The prominence of tense, aspect, and mood / D.N.S. Bhat.
p. cm. -- (Studies in language companion series, ISSN 0165-7763 ; v. 49)
Includes bibliographical references and index.
1. Grammar, Comparative and general--Verb. 2. Typology (Linguistics) I. Title. II. Series.
P281.S49 1999
415--dc21 99-11172
ISBN 90 272 3052 8 (Eur.) / 1 55619 935 X (US) (alk. paper) CIP
© 1999 – John Benjamins B.V.
No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other
means, without written permission from the publisher.
John Benjamins Publishing Co. • P.O.Box 75577 • 1070 AN Amsterdam • The Netherlands
John Benjamins North America • P.O.Box 27519 • Philadelphia PA 19118-0519 • USA
Table of Contents
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
List of Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi
C 1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Universalistic and Differentiating Approaches . . . . . . . . . . . . . . . . . 1
1.2 Nature of the present study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.3 Organisation of the monograph . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
P 
A Descriptive Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
C 2
Category of Tense . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2 Deictic tense . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.2.1 Constraint on present tense . . . . . . . . . . . . . . . . . . . . . . . . 16
2.3 Non-deictic tense . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.3.1 Justifying the distinction in Kannada . . . . . . . . . . . . . . . . . . 22
2.3.2 Need for the distinction . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.4 Distance from the reference point . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.5 Use of temporal adverbials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.5.1 Deictic/non-deictic distinction . . . . . . . . . . . . . . . . . . . . . . . 36
2.5.2 Positional distinctions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.5.3 Remoteness distinctions . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
2.5.4 Conflict with tense . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
vi TABLE OF CONTENTS
C 3
Category of Aspect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.2 Perfective and imperfective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.3 Phasal aspects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.4 Quantificational aspects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
3.5 Situational and viewpoint aspects . . . . . . . . . . . . . . . . . . . . . . . . . 58
3.6 Use of aspectual adverbials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
C 4
Category of Mood . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
4.2 Epistemic mood . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
4.2.1 Realis and irrealis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
4.2.2 Judgements and evidentials . . . . . . . . . . . . . . . . . . . . . . . . 70
4.3 Deontic Mood . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
4.4 Epistemic moods and interrogatives . . . . . . . . . . . . . . . . . . . . . . . . 79
4.5 Deontic moods and imperatives . . . . . . . . . . . . . . . . . . . . . . . . . . 82
4.6 Use of modal adverbials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
P 
A Typological Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
C 5
Basis of the Typology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
5.2 Basis of tense-aspect-mood variation . . . . . . . . . . . . . . . . . . . . . . . 93
5.3 Possibility of using alternative categories . . . . . . . . . . . . . . . . . . . . 94
5.4 Criteria for prominence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
5.5 Nature of generalizations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
5.6 Bias in grammars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
5.7 Need for diachronic considerations . . . . . . . . . . . . . . . . . . . . . . . . 100
5.8 Correlatable characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
TABLE OF CONTENTS vii
C 6
Classification of Languages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
6.2 Tense-prominent languages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
6.2.1 Grammaticalization of tense . . . . . . . . . . . . . . . . . . . . . . . . 104
6.2.2 Obligatoriness of tense . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
6.2.3 Systematicity of tense . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
6.2.4 Pervasiveness of tense . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
6.2.5 Constraints on the study . . . . . . . . . . . . . . . . . . . . . . . . . . 120
6.3 Aspect-prominent languages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
6.3.1 Grammaticalization of aspect . . . . . . . . . . . . . . . . . . . . . . . 121
6.3.2 Obligatoriness of aspect . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
6.3.3 Systematicity of aspect . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
6.3.4 Pervasiveness of aspect . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
6.4 Mood-prominent languages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
6.4.1 Grammaticalization of mood . . . . . . . . . . . . . . . . . . . . . . . 133
6.4.2 Obligatoriness of mood . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
6.4.3 Systematicity of mood . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
6.4.4 Pervasiveness of mood . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
6.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
C 7
Correlatable Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
7.2 Effects of decategorization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
7.3 Ergativity split . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
7.4 Tensedness parameter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
7.5 Absence of state verbs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
7.6 Variations in the mode of encoding the categories . . . . . . . . . . . . . . 155
7.6.1 Tense-prominent languages . . . . . . . . . . . . . . . . . . . . . . . . 156
7.6.2 Mood-prominent languages . . . . . . . . . . . . . . . . . . . . . . . . 157
7.6.3 Relative order of category markers . . . . . . . . . . . . . . . . . . . 158
7.7 Differing points of view . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
7.7.1 The concept of perfect . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
7.7.2 The concept of future . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
7.7.3 The concept of habitual . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
7.7.4 The concept of negation . . . . . . . . . . . . . . . . . . . . . . . . . . 178
viii TABLE OF CONTENTS
7.8 Foregrounding sequential events . . . . . . . . . . . . . . . . . . . . . . . . . . 179

7.9 Paths of grammaticalization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
7.10 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
Preface
This is a revised and enlarged version of a report that I submitted to the Univer-
sity Grants Commission, New Delhi, in 1994. The report, entitled Tense, Aspect
and Mood in Indian Languages, was restricted to languages that are prevalent in
India. An invitation from Prof. Johan van der Auwera to spend six months in
Antwerp, Belgium as a Visiting Scholar in 1997 allowed me to expand its scope
and to cover languages spoken outside India. I was also able to gather facts on
several additional languages with the help of books and journals that were
available at the Antwerp University Library, and also in Prof. van der Auwera’s
personal collection.
I had ventured to establish a typology of languages, based upon the relative
prominence of tense, aspect and mood, in my earlier report itself, but I was not
very sure whether this typology could be extended to languages spoken outside
India. The main problem was getting access to grammars of at least some
representative languages. Antwerp University provided this opportunity for me.
Thanks to an invitation from Prof. Frans Plank, I was also able to spend a week
in Konstanz (I wish I had more time to study in the enormous library of
Konstanz University!), and discover some additional data that supported the
typology. I was also able to present these ideas to linguists in the University of
Konstanz and also in the Amsterdam University. I am thankful to all these
linguists for their comments and helpful suggestions.
The present study may be considered as using a functional-typological
approach, as it attempts to establish generalizations regarding a particular
grammatical feature (verbal category) on the basis of a functional perspective.
It is also basically a “differentiating” approach, in the sense that it tries to find
out ways in which languages differ from one another in their use of a given
grammatical feature. Typological studies have been generally based upon the
universalistic approach. That is, they have been attempting to find generaliza-
tions that are applicable to all languages. Even while dealing with features that
x PREFACE
are applicable to only some languages, the commonly used approach is to

establish statements (called “implications”) which could cover all languages.
Differentiating approach is an alternative, and according to me, a more exciting,
way of establishing typological generalizations about language.
I have received help from several persons while preparing this monograph.
The Central Institute of Indian Languages has probably the best library for
linguistics in India. I have always found the staff members of this Library most
helpful. I have also received generous help from the former Director, Prof.
Annamalai, and the present Acting Director, Prof. Ramaswami. I wish to
acknowledge my indebtedness to all these people, and also to the linguists of the
Institute with whom I was able discuss topics connected with my study.
I prepared the revised, final draft of this monograph in Antwerp University.
I wish to acknowledge my indebtedness to Prof. Johan van der Auwera, and also
to the Research Council, and to the India Study Centre, of Antwerp University
for supporting this work.
D. N. S. Bhat
List of Abbreviations
1 first person  instrumental

2 second person  intransitive
3 third person  locative
 ablative  masculine
 absolutive  neuter
 accusative  negative
 auxiliary . non-future
 conditional  nominative
 counterfactual . non-past
 dative  object
 definite  participle
 direct case  perfective
 different subject  plural
 durative  posterior
 emphatic  potential
 ergative  predicate
 feminine  present
 future  progressive
 genitive  propriative
 habitual  question
 honorific  relative
 imperative  singular
 imperfective  simultaneous
 infinitive  subject
C 1
Introduction
1.1 Universalistic and Differentiating Approaches
There are two different ways in which one can carry out typological research on
language. We may call these Universalistic and Differentiating. It would be more
correct, perhaps, to regard the former as a Language Universal study and to
restrict the term typology to the latter. Because, as we can see from the follow-
ing, languages are viewed as forming different “types” only in the latter type of
study. Further, the latter type of study, even though regarded as part of the study
of Language Universals, is more restrictive in its scope. It generally allows some
languages to remain outside the types of languages that it establishes.
In the case of the “Universalistic” approach, a researcher tries to establish
grammatical elements, processes and strategies, general principles and con-
straints, which are applicable to all natural languages. For example, the claims
about grammatical relations like subject and direct object, or the more basic
relations called A, S and O, claims about the centrality of transitivity or the
universality of word-class distinctions like nouns and verbs, claims about
hierarchies of different kinds such as the accessibility hierarchy of noun phrases
for relativization or reflexivization, claims about the universality of semantic
distinctions like objects, events and properties, and so on are of this type. They
are Universalistic in the sense that linguists consider them to be applicable to all
natural languages.
In the case of the “Differentiating” approach, on the other hand, attempts
are made to divide languages into two or more different types on the basis of the
various contrasting characteristics that are shown by them and to find explana-
tions for those characteristics. The division of languages into isolating, agglutina-
tive and synthetic (or into analytic and synthetic) is probably the earliest study
of this nature. Word order based divisions of languages into SOV, SVO and
VSO and into configurational and non-configurational are also studies of the
2 THE PROMINENCE OF TENSE, ASPECT AND MOOD
same type. The occurrence of differences in case-marking and verbal agreement

has given rise to another division of languages into accusatives and ergatives
(and also into actives — i.e. languages that are neither accusatives nor ergatives).
Similarly, the study of adjectives has given rise to a classification of languages
into those in which property words occur as a distinct category of adjectives on
the one hand, and the ones in which they merge with nouns, verbs or both on the
other. All such studies deal with certain strategies of sentence structure in the
case of which languages are perceived as making alternative choices instead of
using one and the same strategy.
It is evident that the two types of studies can complement one another. The
generalisations put forth by the Universalistic approach forms the starting point
for the Differentiating one as it tries to find strategies at a deeper or more
fundamental level that can allow alternative choices at the higher level. Further,
cross-linguistic studies can only be based upon language-universal elements and
concepts which the Universalistic approach tries to define and codify, and hence
the Differentiating approach can only build upon the foundation that the Univers-
alistic approach constructs. In the actual practice, however, the two approaches
conflict with one another, since the Differentiating approach has the effect of
casting doubts upon the generalisations and conclusions that have been estab-
lished painfully by the Universalistic approach. For example, the division of
languages into configurational and non-configurational has made it necessary to
discard some of the Universalistic claims about the structure of sentences, like
subjacency and movement. The division of languages into active and non-active
has cast doubts upon certain other Universalistic claims such as the centrality of
transitivity and the universality of the notion of subject. The division based upon
the occurrence of word-class distinctions has cast doubts upon the universality of
word-class distinctions themselves and also upon other claims like the one which
underlies x-bar theory.
It is apparently in this sense that Stassen (1985: 4) found the Differentiating
approach to have come up with “some of the more exciting results” of typolog-
ical study. These results have been exciting because they tried to disconfirm
some of the familiar and long-standing notions and assumptions. It is rather
unfortunate, however, that the general tendency of the Universalist, when
confronted by such findings of the Differentiating approach, is to stick to his
claims at any cost, even to the extent of making the claims unfalsifiable. For
example, grammatical relations like Subject and Direct Object have been
postulated for languages like English because of the fact that the relation between
INTRODUCTION 3
case markers and semantic relations is too complex in these languages to

describe through a direct linkage between the two. The question that needs to be
raised here is therefore whether such a complexity in the relationship between
case markers and semantic relations exists in all languages such that the postula-
tion of abstract grammatical relations is necessary for all of them. It was
apparently assumed that this would be the case, but when it was shown that in
the case of some languages the use of such abstract relations would only make
the description unnecessarily complex and also less explanatory (for example, see
Bhat 1991), many of the Universalists did not give up their claim as one would
have expected, but instead, persisted with it. Their contention has been that
certain marginal cases of morphosyntax do occur in these latter languages as well,
in which the relation between case markers and semantic relations is not direct.
Notice, however, that arguments of this type, if perused further, can lead to
other alternative possibilities that may not be very welcome. For example, it is
possible to claim, on the basis of these latter type of languages, that the relation
between case markers and semantic relations is direct in all languages. Even in
the case of languages like English, in which it is indirect in most of the contexts,
there do occur instances in which it can be shown to be direct. Pursuing this line
of thought would, no doubt, make the grammar of languages like English very
complex, but this would be in no way different from the complexity that has
been imposed upon the grammars of other languages by the assumption that
grammatical relations may be established for all languages. The crucial point, I
think, is that languages do show a basic difference on this point. The difference
may be graded with some showing a greater degree of direct linkage than others,
but when languages belonging to the two extreme ends of this gradation are
compared, it becomes evident that the difference cannot be neglected.
The fate of noun-verb distinction is similar to that of grammatical relations
discussed above. The crucial question that needs to be examined here is whether
there is a categorial distinction between nouns and verbs in all languages. It is
not sufficient if one finds, in all languages, syntactic or semantic distinctions of
one type or the other that can be correlated with the familiar noun-verb distinc-
tion, because in spite of the occurrence of such distinctions between, for
example, active and stative verbs, or between animate and inanimate nouns, we
do not regard them as categorial in the languages concerned. The distinctions
must have a fundamental place in the functioning of all languages in order to
claim that they are categorial distinctions in all of them. This crucial point is
generally missed by the Universalists who try to maintain their claim about the
universality of noun-verb distinction in spite of the adverse findings of the

Differentiating approach.
There are, on the other hand, some contexts in which the conflict between
the Universalistic and Differentiating approaches appears to be unresolvable. For
example, there are languages, like Sanskrit, in which adjectives are not differenti-
ated from nouns. Can we claim, in the case of such languages, that there would
still be a semantic distinction between objects and properties? The Universalistic
approach would apparently assume that we can, in spite of the absence of any
formal distinction between the two, whereas the Differentiating approach would
lead us to the conclusion that we cannot make any such claims. It is true that
persons who use such a language may, for various reasons (like exposure to a
language which does make an adjective-noun distinction), become aware of that
distinction and also of the problem of expressing it in the language concerned,
but the language as such will have to be regarded as devoid of that distinction.
In Sanskrit, for example, the exponents of New Logic (navya-nya:ya) found it
necessary to develop an extremely complex style of writing in order to express
the distinction between objects and their properties, a fact which may support the
claim of the Differentiating approach; however, since these logicians did succeed,
at the end, to come up with a set of expressions “in the Sanskrit language” for
this purpose, the Universalist can also persist with his claim.
Another interesting point about this “contest” between the Universalistic and
Differentiating approaches is that it is being carried out, unfortunately, with
rather unequal terms, which are highly favourable for the former and are
distinctly unfavourable for the latter. Both of them have to depend upon pub-
lished grammars, whose authors generally assume, either consciously or uncon-
sciously, several familiar concepts, categories and strategies to be occurring in all
languages. This is actually the way in which we, as human beings, respond to
our environment; when exposed to a strange situation, we try to understand it in
terms of familiar elements and processes. Similarly, when exposed to a “strange”
language, we, as linguists, try to find equivalents to familiar elements like nouns,
verbs and adjectives, subjects and objects, tenses and aspects, and familiar
processes like causativization, reflexivization, complementation, agreement, case-
marking, and so on. It is only when the language under consideration utterly fails
to provide equivalents to these familiar elements and processes that we look for
alternative elements and processes for describing that language.
In view of this general human tendency which affects our descriptions of
grammars to a great extent, a typologist has to be careful in accepting statements
INTRODUCTION 5
about similarities between familiar and non-familiar languages. He should give

greater importance to differences that have been noticed, because these have been
consciously observed and recorded; the similarities, on the other hand, might
have been simply assumed rather than enquired into. For example, the relevance
of grammatical relations for the description of languages like Kannada and
Manipuri were only assumed by earlier scholars. Some doubts were raised
regarding this assumption when the languages did not show complete similarity
with familiar languages, as for example, in the case of the so-called dative-
subject sentences in Kannada, and the use of nominative suffix in Manipuri.
However, when examined in detail, the assumption itself turned out to be
untenable (see Bhat 1991). Similarly, the notion of transitivity was assumed to
be efficient for the description of Hindi causatives, or of distinctions in the forms
of Tamil verbal stems, by earlier scholars. But when detailed studies of these
topics were undertaken, it was found that an entirely different notion, namely
affectivity, was actually needed for that purpose (see Saksena 1982 on Hindi, and
Paramasivam 1979 on Tamil).
The Universalistic approach tends to give greater importance to features
which occur in the vast majority of languages, with the ones that are restricted
to a few isolated languages being put aside as “unexplained” exceptions, or as
cases that need to be accounted for. This is much the same as regarding them as
irrelevant because when the Universalist’s generalisations are being referred to
by other authors, these exceptional cases generally tend to be left out and
forgotten. The Differentiating approach, on the other hand, is in a position to
provide prominence to such features by establishing additional language types.
The establishment of such language types can help to overcome the masking effect
of familiar language types mentioned above because, in some cases at least, it can
lead to a re-examination of the existing descriptions (grammars) of languages.
As I had suggested earlier, the sets of language types that a typologist
establishes under the Differentiating approach need not necessarily be a lan-
guage-universal one. Its emphasis is upon the differences that occur between sets
of languages. It tries to find correlatable characteristics among such differences,
which may ultimately lead to interesting explanations, as one of these correlat-
able differences may turn out to be more fundamental than the remaining
differences, and may therefore be viewed as “causing” those differences. The
study may, however, be restricted to a particular area, as for example, in the case
of the Eurotype study, and the groupings that it establishes may or may not be
extendable to other areas (see van der Auwera, forthcoming). The advantage of
such a restrictive typological study is that the data being more limited and the
languages being better known to the linguist concerned, the topic would be more
manageable. Correlations that would be missed by a wider type of study would
be more readily recognised by a restrictive study. This advantage offsets the
disadvantage of some of the conclusions turning out to be relevant only for the
region under study.
It is possible to use both the Universalistic as well as the Differentiating
type of approaches in our typological study of verbal categories like tense, aspect
and mood. Languages manifest an enormous amount of variation in their
encoding and use of these verbal categories; a Universalistic approach would try
to find common elements and tendencies that occur at the base of these varia-
tions. Dahl’s (1985) attempt to establish a universal set of basic elements from
which languages make a selection and Bybee’s (1985) and Bybee, Perkins and
Pagliuca’s (1994) attempt to establish universal paths along which such elements
evolve and get established as grammatical elements in different languages can be
regarded as primarily involving such an approach. A Differentiating approach, on
the other hand, would try to find a basis for the variations by establishing
idealised language types such that sets of correlatable distinctions can be
associated with each language type. Individual languages can then be assigned to
one or the other of these idealised language types.
In the case of verbal categories, however, variations appear to occur along
different dimensions. It might therefore be necessary, for the time being at least,
to establish different sets of idealised language types with each set accounting for
a particular dimension. For example, we find variations occurring along the
dimension of grammaticalization with some languages grammaticalizing the
verbal categories to a very high degree, some showing practically no grammatic-
alization whatsoever, and others showing grammaticalization to different degrees
along this dimension. This variation can form the basis of a typology of languag-
es into isolating and synthetic (or inflectional). We also find variations occurring
in the actual types of categories that are grammaticalized, with some languages
grammaticalizing only one or the other of the three major categories, namely
tense, aspect and mood. The two dimensions cut across one another and therefore
the typology that can be established on the basis of this latter dimension would
not be the same as the former typology.
INTRODUCTION 7
1.2 Nature of the present study
The present monograph makes use of the second dimension mentioned above in
order to establish a typology of languages for the study of verbal categories. Its
emphasis is upon the differences that occur among languages in their encoding
of verbal categories. It is based upon a typological study, carried out earlier, of
the verbal categories occurring in the languages of India (Bhat 1994b). During
the course of that study, it was discovered that a classification of languages on
the basis of the prominence that the languages attach to tense, aspect and mood
could lead to certain interesting generalizations. Most importantly, it was found
that languages that give greater prominence to one of these categories appeared
to view concepts belonging to the other two categories in terms of their promi-
nent category. For example, mood-prominent languages appeared to view
temporal and aspectual notions in terms of the modal category, whereas aspect-
prominent and tense-prominent languages appeared to view modal (and other)
notions in terms of the category of aspect and tense respectively.
Consider, for example, the notion of perfect, which is generally described
as a “past event with present relevance”. This is apparently a temporal way of
viewing this concept. There are languages like Mao Naga in which the category
of mood is more prominent than tense or aspect. In such languages, perfect (i.e.
the verbal form that translates as perfect in English) represents something rather
different: it denotes a realis event about which something more needs to be done.
That is, instead of combining the notion of past with present, these languages
join together the notion of realis with that of irrealis in order to establish a
concept that is comparable to the perfect of English. Aspect-prominent languag-
es, on the other hand, appear to view this notion from an aspectual point of view,
i.e. as involving a combination of perfective with imperfective.
In order to bring out correlatable differences of the above type that occur
among languages, it is apparently useful to classify languages into tense-promi-
nent, aspect-prominent and mood-prominent types. However, natural languages
show a lot of variation among themselves concerning the degree of prominence
that they attach to one or the other of these categories. For example, among the
languages that give greater prominence to mood than to tense or aspect, some
use the modal category almost exclusively in their verbal system, leading to the
contention that they are “tenseless” (or “aspectless”). Others, however, provide
some representations to tense and aspect as well, even though the category of
mood continues to be the most prominent one among them. The existence of
such cross-linguistic variations makes it rather difficult to characterise the three

language-types directly on the basis of such varying languages.
In order to overcome this problem, we may resort to the notion of “idealised
languages”. We can assign sets of characteristics to these idealised languages
such that they are maximally different from one another; we can then group the
actual languages under one or the other of these idealised languages depending
upon the kind of similarity that they show in sharing characteristics with them.
Such a classification is helpful in that it provides us with a basis for making
further inquiry regarding characteristics that languages share when grouped under
a single language type and the ones that the languages do not share when they
are grouped in different types.
1.3 Organisation of the monograph
I have organised the present monograph into two different parts, of which the
first one provides a detailed description of the three verbal categories of tense,
aspect and mood, and the second one establishes a typology of languages based
upon their representation of these three categories. It can be claimed that the
three categories are quite distinct from one another, even though there do occur
interconnections between them. We can describe tense as indicating the location
of an event on a linear time scale (as before, simultaneously or after a particular
reference point which may be deictic or non-deictic), aspect as denoting the
temporal structure of the event (as complete or on-going, beginning or ending,
occurring once or several times, etc.) and mood as denoting the actuality of the
event (as real or not real, seen, heard, or inferred, possible, probable or certain,
necessary or unnecessary, etc.). Since these three categories denote different facets
of one and the same event, there is the possibility, mentioned above, of giving
prominence to one of them, and viewing the others in terms of that category.
In order to emphasise the differences that occur between these three
categories, I have described them separately in three different chapters. The
general tendency among grammarians is to club them together (especially the
categories of tense and aspect) apparently because the language that they
describe do not allow them to make sharp and clear-cut distinctions between
them. However, as I point out in the following chapters, the categories them-
selves are not indistinguishable. It is quite possible to establish an idealised
situation in which the three are quite different from one another. Individual
INTRODUCTION 9
languages can be viewed as representing one or the other of these categories in

detail and others through indirect means. Such an idealised differentiation
between tense, aspect and mood has the effect of making our description of these
categories less complicated and therefore less confusing. For example, as I point
out in the second chapter (see 2.6), the failure to differentiate between tense and
aspect (i.e. between temporal location and temporal structure) has made the
description of tense, provided by some linguists, to be extremely complex.
Further complexity has been added to these descriptions by the failure of these
linguists to differentiate between information provided by temporal (and also
aspectual) adverbials on the one hand and tense markers on the other. The two
are also quite different from one another, even though they replicate one another
to a certain extent.
In the second part of this monograph, I establish a typology of languages on
the basis of the prominence that languages give to tense, aspect or mood. That
is, I assume that the three categories form the basis for three different language
types, namely tense-prominent, aspect-prominent and mood-prominent. Ideally,
the category that has been chosen as the most prominent one would be the only
category that gets grammaticalized; concepts belonging to the other two catego-
ries would be viewed as different facets of this chosen category. Actual languag-
es that can be assigned to these idealised language types, however, show a
gradation, with some being closer to one of the idealised types than others. I will
try to show how these actual (natural) languages exemplify the language type by
placing greater prominence to one of the categories as compared to others. This
classification would be justified in a following chapter by showing how languag-
es belonging to one or the other of these different language types share several
other characteristics which are apparently derivable from the fact that they give
prominence to one or the other of these three categories.
P 
A Descriptive Study
C 2
Category of Tense
2.1 Introduction
Tense is an inflectional marker of the verb used for denoting the temporal
location of an event (or situation). Since time itself does not have any distin-
guishable marks on it, tense has to make use of some other event which occurs
before, simultaneously or after the event under consideration as the reference
point for indicating its temporal location. Languages differentiate between two
main types of events that they use as reference points in this fashion; the event
of uttering the sentence in which the tensed verb occurs is one of them; the
second one is any other type of event. Consider, for example, the following
Kannada sentences:
(1) a. nañu manege ho˜-d-e
I home go--1
‘I went home’
b. nañu manege ho˜g-i malagi-d-e
I home go- lie--1
‘Having gone home, I lied down’
Notice that in (1a), the event of going has the event of uttering that sentence as
its reference point (it occurred before the point of time at which that sentence
was uttered by the speaker) whereas in (1b), it has the event of lying down as its
reference point (going occurred before lying down). It is possible for this latter
reference point to be past as in (1b) or future (non-past) as in (1c), given below,
but the former reference point (utterance time) can only be present.
(1) c. nañu manege ho˜g-i malagu-tt-eñe
I home go- lie-.-1
‘Having gone home, I will lie down’
We may use the terms “deictic” and “non-deictic” in order to differentiate

between (i) tenses which have the utterance time as the reference point and (ii)
the ones which have some other event as the reference point, respectively.
Traditionally, these are called “absolute” and “relative” tenses (cf. Comrie
1985: 36) but the difference between the two does not depend upon one of them
being relative and the other one non-relative (or absolute); both are relative to a
reference point; the difference between the two is only that the former uses a
deictic event (an event which is connected with the speech act) as the reference
point whereas the latter uses some other event for that purpose. As I will be
pointing out later, this deictic/non-deictic distinction plays an important role in
the case of aspect and mood as well.
In addition to this distinction (i) regarding the kind of reference point used
(deictic versus non-deictic), and (ii) regarding the position of the event with
reference to that reference point (before, simultaneously or after it), the category
of tense also makes use of the notion of relative distance from the reference
point as a parameter of differentiation. Consider, for example, the following
verbal forms of Nkore-Kiga, a Bantu language of South-Western Uganda, in
which there is a three-fold distinction of temporal distance in the past tense, with
today past and remote past being represented by the suffixes aa and ka respec-
tively and yesterday past being represented by a modification of the verbal stem
(Taylor 1985):
(2) a. n-aa-gyenda
I-today:-go
‘I went today’
b. n-gyenzire
I-went (yesterday)
‘I went yesterday’
c. n-ka-gyenda
I--go
‘I went earlier than yesterday’
Nkore-Kiga also makes a two-fold distinction of temporal distance in the
future tense. It also has an additional present tense form, and an unmarked form
of the verb (called universal tense) which is used for denoting permanent or
habitual events or states, but it shows distinctions of temporal distance only in
the case of past and future tenses (Taylor 1985).
The category of tense provides only a skeletal view of the temporal location
CATEGORY OF TENSE 15
of a given event. Additional information regarding its temporal location is

provided by temporal adverbials that occur, optionally, as modifiers of the verb
in the sentences concerned. These adverbials replicate tense by establishing a
parallel temporal structure, which, however, is generally related to the one
established by tense. The adverbials also make use of the same set of parameters
that are used by tense, but they may use certain additional parameters as well.
Even while using the former parameters, they are able to provide more detail
than tense about the temporal location of the event.
Languages may use different types of devices for representing the temporal
location of events. The occurrence of tense markers is only one of them. For
example, there can be ‘tenseless’ languages, i.e. languages in which the notion
of temporal location does not get grammaticalized. In these languages, the notion
is represented rather indirectly by aspectual or modal distinctions. It is also
possible to have tense distinctions occurring in the auxiliary verb only and not in
the main verb, or in the temporal adverbials only and not in the verbal system.
For example, distinctions of temporal distance are represented by temporal
adverbials rather than by tense markers in English, as shown by the translations
of (2a–c) given above.
2.2 Deictic tense
There can be a three-fold distinction in the case of deictic tense, namely between
past, present and future, depending upon whether the event under consideration
occurs before, simultaneously or after the time of uttering the sentence through
which the event is being described. Consider, for example, the following three
Kurukh (Dravidian) sentences:
(3) a. eñ ij-d-an
I stand--1
‘I stand’
b. eñ ij-k-an
I stand--1
‘I stood’
c. eñ ij‘-on
I stand-()-1
‘I will stand’
The tense markers occurring in the verbal forms of these sentences have the
function of indicating the temporal location of the event of standing. According
to (3a), this event is simultaneous with the time at which that sentence was
uttered, whereas it is prior to the utterance time in (3b); and according to (3c),
it is later than the time at which that sentence was uttered.
Notice that there is always a need to have some other event that functions
as the reference point in order to specify the location in time of a given event.
This is because, as I have mentioned earlier, time itself does not contain any
distinguishable area or point with reference to which an event can be stated as
occurring before, simultaneously or after that point. However, deictic tenses
differ from non-deictic tenses in that their reference point (utterance time) is
generally considered to be the unmarked one and hence it need not be specified
in the sentence; only the reference point of non-deictic tenses needs to be
specified. Hence, when no reference point is specified in a given sentence, it is
generally assumed that the reference point is the utterance time. For example, in
all the sentences given above (3a–c), the act of uttering the sentence under
consideration functions as the reference point even though none has been
specified in any of them.
It would be useful to differentiate between deictic and non-deictic tenses by
using distinct sets of terms for denoting them. I propose to use the traditional
terms, past, present and future, and also the ones that I would be introducing
later for two different combinations of these, namely non-past (future and
present) and non-future (past and present), for referring to deictic tenses, and the
terms prior, simultaneous and posterior for referring to non-deictic tenses. There
are several interesting differences between these two types of tenses, and the use
of distinct sets of terms for representing them would be helpful in avoiding
confusions in their description.
2.2.1 Constraint on present tense
The fact that an event can be thought of as occurring before, simultaneously, or

after the point of time at which a speaker is uttering a sentence which describes
it, makes it possible for us to think of three different deictic tenses, namely past,
present and future. The three Kurukh sentences given above (3a–c) exemplify the
occurrence of this three-fold deictic tense distinction. There is, however, an
interesting constraint in the use of present tense, namely that an event which is
simultaneous to utterance time needs to be durative or progressive, or habitual
(see Bennett and Partee 1978: 13). This constraint apparently results from the fact
that a speaker needs some amount of time for producing his statement about an
observed event, and hence, by the time he has produced his statement, the
observed non-durative or non-habitual event would have turned out to be a “past”
event. Thus in Kurukh, the verb in present tense given above (3a) represents
mainly a habitual event.
Another way that has been used by languages to resolve this problem is to
have only a two-fold tense distinction; this may be between past and non-past, as
in English, Kannada, and several other familiar languages, or between future and
non-future as in Manipuri (Tibeto-Burman) and several other languages. The
notion of present tense gets combined with future in the former case and with
past in the latter case. The latter system, however, appears to have developed
from an earlier modal system involving a distinction between realis and irrealis
moods; that is, the so-called non-future represents basically the realis mood.
The following Kannada sentences exemplify the use of the first alternative
mentioned above:
(4) a. avanu manege ho˜-d-a
he home go--3:
‘He went home’
b. avanu manege ho˜gu-tt-añe
he home go-.-3:
(i) ‘He goes home (habitual)’
(ii) ‘He will go home’
c. avanu manege ho˜gu-tta˜ id-d-añe
he home- go- be--3:
‘He is (in the process of) going home’
Notice that the verb ho˜gu ‘to go’ shows only a two-fold deictic tense distinction
(past versus non-past), as seen in (4a) and (4b); this is true of all other verbs in
Kannada, excepting iru ‘to be’ (see below); (4c), used for denoting the present
tense meaning, involves a periphrastic construction, in which the main verb
occurs in the non-deictic simultaneous tense (which also has a durative or
progressive meaning in this usage), and the verb iru ‘to be’ occurs in the deictic
present tense. The verb iru ‘to be’ is exceptional in showing a three-fold (past-
present-future) deictic tense; this has apparently been facilitated by the fact that
iru ‘to be’ is a stative verb.
The verb iru ‘to be’ shows this three-fold deictic tense distinction while
indicating the location of objects and characteristics (see 5a–c below) also, but
when used for denoting the location of events, it only shows a two-fold past/non-
past deictic tense distinction (see 6a–c):
(5) a. avanu illi id-d-a
he here be--3:
‘He was here (when I came)’
b. avanu illi id-d-añe
he here be--3:
‘He is here (now)’
c. avanu illi iru-tt-añe
he here be--3:
(i) ‘He will be here (when you come)’
(ii) ‘He is (usually) here’
(6) a. ivattu ondu sabhe it-t-u
today one meeting be--3:
‘There was a meeting today’
b. ivattu ondu sabhe i-d-e
today one meeting be-.-3:
(i) ‘There is a meeting today’
(ii) ‘There will be a meeting today’
c. ivattu sabhe iru-tt-ade
today meeting be--3:
‘There is (usually) a meeting today (say, on Mondays)’
Notice that (5c) is ambiguous between future and habitual meanings whereas (6c)
has only the habitual meaning; (6b), on the other hand, has both present and
future (but not habitual) meanings (i.e. it denotes the non-past tense). The point
to be noted here is that events are generally viewed as momentary (or rather are
unspecified for duration) in Kannada and hence they can only occur in past or
non-past tenses. Verbs require the simultaneous suffix (which also has the
progressive meaning associated with it) to be attached to them in order to
indicate events as non-momentary.
The use of a future/non-future distinction, instead of this past/non-past
distinction, can be exemplified with the help of the following sentences of
Manipuri, a Tibeto-Burman language. This language has two different tense
forms, derived by adding the suffix li ‘non-future’ and k6ni ‘future’ to the verb;
the non-future form has past and present habitual meanings in the case of verbs
denoting an event, and past and present (also present habitual) meanings in the
case of verbs denoting a state; verbs denoting an event have an additional form,
derived by adding the suffix lì ‘durative’ for denoting the present meaning.
Examples:
(i) Verbs denoting an event:
(7) a. m6hak ci]-d6 c6t-li
he hill- go-.
(i) ‘He went to the hill’
(ii) ‘He usually goes to the hill’
b. m6hak ci]-d6 c6t-lì
he hill- go-
‘He is going to the hill’
c. m6hak ci]-d6 c6t-k6ni
he hill- go-
‘He will go to the hill’
(ii) Verbs denoting states:
(8) a. ]6si no] m6]-]i
today rain cloudy-.
‘It is cloudy today’
b. ]6ra] no] m6]-]i
yesterday rain cloudy-.
‘It was cloudy yesterday’
c. julay-d6 no] m6]-]i
July- rain cloudy-.
‘It is generally cloudy in July’
d. nu]da]wayr6md6 no] m6]-g6ni
evening() rain cloudy-
‘It will be cloudy in the evening’
Notice that the non-future suffix has the past and habitual meanings in (7) and
also the present meaning in (8); the future suffix, on the other hand, has only the
future meaning in both these cases.
Several other Tibeto-Burman languages are similar to Manipuri in showing
a future/non-future tense distinction, but in some of them at least the distinction
needs to be regarded as primarily one of mood (i.e. between realis and irrealis)
rather than that of tense. As I point out later (see 4.2.1) Mao Naga has basically
a modal distinction of this type with the so-called past (realis) forms being used
even for denoting future events (see also 9.4.2).
2.3 Non-deictic tense
Tense suffixes can also indicate the temporal location of an event by using some
other event as the reference point, and in such cases, we may regard them as
non-deictic. As in the case of deictic tenses, this non-deictic temporal location
can also be before, simultaneous with or after the event that has been chosen as
the reference point. As I have suggested earlier, we may use a different set of
terms, namely prior, simultaneous and posterior for denoting these three non-
deictic tenses. These can be exemplified with the help of the three converbs that
occur in Kannada, as shown in the following sentences:
(9) a. nañu haNN-annu be˜yis-i kattaris-id-e
I fruit- cook- cut--1
‘I cut the fruit after cooking it’
b. nañu haNN-annu be˜yisu-tta˜ kattaris-id-e
I fruit- cook- cut--1
‘I cut the fruit while cooking it’
c. nañu haNN-annu be˜yis-alu kattaris-id-e
I fruit- cook- cut--1
‘I cut the fruit for cooking (it) later’
The verb be˜yisu ‘to cook’ occurs in its three different converbal forms in the
sentences given above, and the temporal location of the event of cooking is
indicated in these sentences as being before (9a), simultaneous with (9b) and
after the event of cutting, with the latter event functioning as its reference point.
Non-deictic tenses are different from deictic tenses in that the reference
point must necessarily be specified in the sentence itself. This is apparently
facilitated by the fact that verbs that show this non-deictic tense generally occur
in their non-finite form; they occur in a subordinate clause which is dependent
upon the clause that denotes the event which functions as its reference point.
Notice, however, that this constraint affects only the non-deictic tense. Deictic
tenses can be represented by both finite as well as non-finite verbal forms. For
example, relative participles occurring in Dravidian languages can have deictic
tenses, but they are non-finite in form. They are, however, subordinated to nouns
and not to verbs or clauses. The following Tamil clauses illustrate this point:
(10) a. uLLe iru-kkira tiruTan
inside be-. thief
‘the thief who is inside’
b. uLLe iru-nta tiruTan
inside be- thief
‘the thief who was inside’
Notice that the two relative participles occurring in (10a) and (10b) represent
non-past and past deictic tenses respectively.
Verbal forms occurring in counterfactual conditionals are also in a subordi-
nated form, but they denote deictic (past) tense, as can be seen from the follow-
ing Tamil example:
(11) avan i]ke va-ntu iru-nt-a˜l
he here come- be--
‘if he had come here…’
Events in non-deictic tenses can also have non-tensed events as their
reference points; i.e., non-deictic tenses need not necessarily be related to deictic
tense forms, as has been suggested by some linguists (see Enc 1987). The
following Kannada sentences exemplify this point:
(12) a. kuDi-du baruv-avri-ge kelasa illa
drink- come-them- work not
‘There is no work for those who come after drinking’
b. kuDiyu-tta˜ baruva-avari-ge kelasa illa
drink- come-them- work not
‘There is no work for those who come drinking’
c. kuDiy-alu baruv-avari-ge kelasa illa
drink- come-them- work not
‘There is no work for those who come for drinking’
Notice that the relative participle baruva ‘coming’ occurring in these sentences
can have a habitual interpretation, and in such a usage, it does not represent any
deictic tense. The verbal participles of the root kuDi ‘drink’ occurring before it
in different non-deictic tenses (prior, simultaneous and posterior respectively) are
therefore unrelated to any deictic tense.
The claim that the habitual is not in any deictic tense in the sentences given
above is supported by the fact that it can occur in a matrix clause that contains
a past tense verb as well. Example:
(13) kuDi-du baruv-avari-ge kelasa iral-illa
eat- come-them- work was-not
‘There was no work for those who (used to) come after drinking’
Another interesting point that may be noted here is that the constraint that
affects the deictic present tense, namely that it must necessarily be durative (see
2.2.1) does not affect the non-deictic simultaneous tense; the latter can be
durative or non-durative. Since the reference point of this non-deictic tense can
be an event that is not directly associated with the utterance time, it is free of the
above-mentioned constraint.
2.3.1 Justifying the distinction in Kannada
Grammarians of Kannada do not differentiate between deictic and non-deictic

tense forms; nor do they consider the non-deictic tense forms given above to be
constituting a single paradigm. The prior and simultaneous tense forms are
generally considered to be past and present participles respectively (see Kittel
1903, Schiffman 1979: 67, Sridhar 1990: 71), whereas the posterior tense form is
considered to be an infinitive; the latter is also regarded as showing no “tense”
distinction as such. It is evident from the examples given above, however, that
the three forms do form a paradigm in Kannada; all three of them occur with
deictic tense forms in two main types of constructions, namely (i) with various
other verbs and (ii) with the auxiliary verb iru ‘to be’. In all these usages, they
indicate the three non-deictic tenses mentioned above. For example, in the
sentences given above (9a–c), the three participial forms be˜yisi ‘having cooked’,
be˜yisutta˜ ‘while cooking’ and be˜yisalu ‘for cooking later’ indicate only non-
deictic tenses, as their point of time is only relative to the point of time of the
event of cutting and not that of the utterance as such. It would be incorrect,
therefore, to regard them as “past” or “present” participles.
The three non-deictic tense forms can occur with non-past forms of verbs
also, as can be seen from the following examples:
(14) a. nañu haNN-annu kattaris-i be˜yisu-tt-eñe
I fruit- cut- cook-.-1
‘I will cook the fruit after cutting it’
b. nañu haNN-annu kattarisu-tta˜ be˜yisu-tt-eñe

I fruit- cut- cook-.-1
‘I will cook the fruit while cutting it’
c. nañu haNN-annu kattaris-alu be˜yisu-tt-eñe
I fruit- cut- cook-.-1
‘I will cook the fruit for cutting it later’
The structure and meaning of auxiliary constructions in which these three
non-deictic tense forms combine with the verb iru ‘to be’ can be shown clearly
with the help of the following table in which the three rows represent non-deictic
tense forms of the verb ho˜gu ‘to go’ (Kannada also has a negative converb,
contrasting with these affirmative converbs (ho˜gade ‘without going’) which can
occur in this auxiliary construction, but it does not show any tense distinction)
and the three columns represent the deictic tense forms of the auxiliary verb iru
‘to be’:
Table 1. Auxiliary constructions
past present future

prior ho˜gidda ho˜giddañe ho˜giruttañe
simultaneous ho˜guttidda ho˜guttiddañe ho˜guttiruttañe
posterior ho˜galidda ho˜galiddañe ho˜galiruttañe
The following sets of sentences exemplify the contrastive uses of these auxiliary
constructions:
(15) a. mu˜ru gaNTe-ge a˜ta manege ho˜g-i-d-d-a
three hour- he home go--be--3:
‘He had gone home by three o’clock’
b. mu˜ru gaNTe-ge a˜ta manege ho˜g-i-d-d-añe
three hour- he home go--be--3:
‘He has gone home (*by three o’clock)’
c. mu˜ru gaNTe-ge a˜ta manege ho˜g-i-ru-tt-añe
three hour- he home go--be--3:
‘He will have gone home by three o’clock’
(16) a. mu˜ru gNTte-ge a˜ta manege ho˜gu-tt-id-d-a
three hour- he home go--be--3:
‘He was going home at three o’clock’
b. mu˜ru gaNTe-ge a˜ta manege ho˜gu-tt-id-d-añe

three hour- he home go--be--3:
‘He is going home (now) at three o’clock’
c. mu˜ru gaNTe-ge a˜ta manege ho˜gu-tt-iru-tt-añe
three hour- he home go--be--3:
‘He will be going home at three o’clock’
(17) a. mu˜ru gaNTe-ge a˜ta manege ho˜g-al-id-d-a
three hour- he home go--be--3:
‘He was to go home at three o’clock’
b. mu˜ru gaNTe-ge a˜ta manege ho˜g-al-id-d-añe
three hour- he home go--be--3:
‘He is to go home (now) at three o’clock’
c. mu˜ru gaNTe-ge a˜ta manege ho˜g-al-iru-tt-añe
three hour- he home go--be--3:
‘He will be going home at three o’clock’
The use of posterior-future forms, such as ho˜galiruttañe in third person
masculine singular, is rather infrequent, but the ones in third person neuter
singular, such as ho˜galiruttade ‘one needs to go’ is quite frequent. Notice that
each of the nine forms given above in the table represents a distinct person-
gender-number paradigm, each of which contains nine different forms. The table
gives only the third person masculine singular form.
Grammarians of Kannada use the term “infinitive” for denoting the posterior
form, but that term is unsuitable for this form because the form functions only
as an adverbial in this language; it cannot take case suffixes (excepting the dative
ke for denoting purpose); the term “purposive”, used by some scholars, is also
not very suitable as seen, for example, in the use of that form in auxiliary
constructions like baralidda ‘he was to come’ and baraliddañe ‘he is going to
come’. The notion of “purpose” is absent in either of these two usages. The
meaning that is consistently associated with this form in all its usages is that of
posterior tense.
According to Subrahmanyam (1971: 452), infinitive in Dravidian has several
meanings such as purpose, cause, and effect, but in Kannada, its use is restricted
to contexts in which the speaker indicates an event that is to occur after another
event. That is, the notion of “cause” or of being anterior (or simultaneous) to
another event is not expressed by this form in Kannada. Some grammarians do
provide instances of the latter type of usage (see (18) below), but such construc-
tions are absent in the spoken form, and are most infrequent (probably to be
found in grammar books only) even in the written form.
(18) ‘ra˜ju taraka˜ri tar-alu si˜te aDuge ma˜DidaLu
Raju vegetable bring- Site cooking did
‘While Raju brought vegetables, Site cooked (them)’
Some of the complex forms given above appear to provide translations as
perfect, pluperfect and progressive, but these meanings do not constitute their
primary connotations. The primary connotations are the combinations of non-
deictic and deictic tense distinctions listed earlier, and the aspectual or modal
connotations only occur as their implications in some of the restricted contexts.
For example, as I have mentioned above, constructions involving the posterior
form provide purposive meaning in some of the usages (especially in some of the
main verb constructions), but in others like auxiliary constructions, they provide
only the posterior meaning and not the purposive meaning. Even in main verb
constructions, they do not always provide this purposive meaning. This is
especially true of their use with non-volitional verbs. Examples:
(19) nanage ondu haNNu tinn-alu sikkitu
me one fruit eat- got
‘I got a fruit to eat’
(20) a˜ mara bi˜L-alu siddhava˜g-ide
that tree fall- ready-is
‘That tree is ready to fall’
Similarly, the prior-past and prior-present auxiliary constructions are
generally translated as pluperfect and present perfect respectively, but their usage
is quite different from that of the corresponding perfect forms of languages like
English. They behave more like complex constructions than as unified tense
forms. One interesting point that supports this claim is that the prior-present
forms can allow temporal adverbials to denote a past point of time. Examples:
(21) avanu ninne-ye˜ ban-d-id-d-añe
he yesterday- come--be--3:
*‘He has come yesterday itself’
(22) nañu ninne na˜lku gaNTe-ge ed-id-d-eñe
I yesterday four hour- rise--be--1
*‘I have got up yesterday at four o’clock’
In fact a temporal adverbial can modify either the participle or the auxiliary
verb in such constructions as can be seen from the following sentences:
(23) a. avanu ninne beLigge ban-d-id-d-a
he yesterday morning come--be--3:
‘He had come yesterday morning’
b. avanu mu˜ru divasa ban-d-id-d-a
he three day come--be--3:
‘He had been here for three days’
Notice that in (23a) the adverbial ninne beLigge ‘yesterday morning’ modifies the
participle of the main verb baru ‘to come’, whereas in (23b) the adverbial mu˜ru
divasa ‘three days’ modifies the auxiliary verb iru ‘to be’. The latter can also
ambiguously indicate that he had come on three days, i.e. the adverbial can also
modify the participle. This point indicates clearly that the prior-past form is a
complex construction in Kannada.
In English, present perfect indicates an event which is closer to the utterance
time than the one denoted by the simple past form, apparently because an
immediate event would have greater degree of present relevance than a non-
immediate one. However, the auxiliary construction containing a prior verb
followed by a present auxiliary in Kannada indicates an event that is non-
immediate as compared to the corresponding simple past form. Examples:
(24) a. nañu ban-d-e
I come-past-1
‘I came (just now)’
b. nañu ban-d-id-d-eñe
I come-prior-be--1
‘I have come’ (I came sometime back)’
This distinction between simple past and prior-present forms also gets
reflected in the fact that Kannada uses the former, but not the latter, for denoting
an action which is going to be carried out immediately by the speaker. Example:
(24) c. i˜ga ban-d-e
now come-past-1
(i) ‘I came just now’
(ii) ‘I am coming in a moment’
Another interesting point that supports the above-mentioned claim is that,

unlike the present perfect construction of English, the corresponding auxiliary
construction of Kannada can occur in the imperative and also in nominalizations
and other non-finite verbal constructions. Examples:
(25) a. niñu beliggeñe˜ ban-d-iru
you morning () come--be
‘You come and be here in the morning itself!’
b. avanu ivatte˜ ban-d-iru-v-udu a˜šcarya
he today () come--be-.-it surprise
‘It is surprising that he has come today itself’
(26) svalpa hottu o˜du-tta˜ iru
little time read- be
‘Be reading for a while!’
Constructions containing simultaneous tense forms are generally regarded as
involving progressive (or durative) meaning. But the contrast that they establish
with prior and posterior forms in the paradigms given above makes it clear that
the language places greater emphasis upon their non-deictic tense meaning than
upon their aspectual meaning. The latter meaning can therefore be regarded as an
implicational one.
Another interesting point that my be noted here is that tense suffixes
occurring in deictic finite forms (past and non-past) are somewhat different from
the ones occurring in non-deictic converbal forms (prior, simultaneous and
posterior). The two deictic suffixes are, (i)d (with several alternants like d, id, t,
T, etc.) for past, and tt or v for non-past, and the three non-deictic suffixes are i
(with several alternants of which some are identical with those of past) for prior,
tta˜ for simultaneous and alu for posterior. Examples:
(i) Deictic tenses:
past ho˜-d-e ‘I went’
go--1
non-past ho˜gu-tt-eñe ‘I go’
go-.-1
(ii) Non-deictic tenses:
prior ho˜g-i ‘after going’
go-
simultaneous ho˜gu-tta˜ ‘while going’

go-
posterior ho˜g-alu ‘before going,
go- in order to go’
The partial identity between past and prior suffixes and also between non-past
and simultaneous suffixes apparently derives from the diachronic fact that the two
types of suffixes originate from a single set of tense suffixes; however, past and prior
suffixes are formally distinct in the case of the majority of verbal bases (about 80
per cent) with the past suffix being represented by the morph (i)d and the prior
suffix by i; the two are identical only in the case of certain “irregular verbs”.
2.3.2 Need for the distinction
I have proposed in this chapter that it would be useful to differentiate between

deictic and non-deictic tenses by using two distinct sets of terms for referring to
them, namely past, present (or non-past) and future for deictic tenses and prior,
simultaneous and posterior for non-deictic tenses. I believe that the failure to
differentiate between these two types of tenses has given rise to some amount of
confusion in the description of tense forms and their connotations. For example,
Schiffman (1969: 38) writes that, in Tamil, the past tense suffix occurring in
adverbial participles is semantically empty since it is the same no matter what
the tense of the next verb is, and could be easily introduced by a transformation.
Lehmann (1989: 72, 265) also considers the verbal participle of Tamil (formed,
according to him, by adding the participle suffix, which is homophonous with the
past tense suffix, to the verb) to be “tenseless” because its time reference is
determined by the main verb.
A closer examination of the use of this participle indicates, however, that in
all its usages, it indicates an event which is prior to the event denoted by the
following (matrix or another participial) clause. Lehmann (1989: 267) apparently
considers the following sentence to be representing the use of the participle for
denoting “non-temporal” conjunction:
(27) a. kuma˜r vi˜TT-il ta]k-i pakal muzuvatum
Kumar house- stay- day whole
tu˜]k-in-añ
sleep--3:
‘Kumar stayed at home and slept the whole day’
Notice, however, that the sentence becomes unacceptable (or at least its meaning
changes) if the verb tu˜]ku ‘sleep’ is used in the participial form instead of the verb
ta]ku ‘stay’ as shown below. That is, there is a need to regard the latter event
(‘stay’, which is denoted by the past participle) as prior to the former event (‘sleep’).
(27) b. ‘kuma˜r pakal muzuvatum tu˜]ki vi˜TT-il
Kumar day whole sleep- house-
ta]k-in-añ
stay--3:
‘Kumar slept the whole day and stayed at home’
Lefebvre & Muysken (1988) introduce a formal distinction between main
tense and relative tense for Quechua. They describe the former as relating the
time of the event expressed in a proposition to the moment of speech, and the
latter as relating the time of an event to that of an event described in the matrix
clause. The latter are also called “-Main tense”, implying that they occur in
subordinate clauses. We may regard this distinction as between deictic and non-
deictic tenses. Reh (1996), on the other hand, describes Anywa, a Western
Nilotic language, to be a “relative tense language”, in which the point of
temporal reference coincides with the time of speaking only in cases in which
there is no explicit temporal reference otherwise, be it in the shape of a temporal
adverbial, or in that of another clause. Tenses of Anywa are therefore described
by Reh (a) as applying at the time point imagined (non-past), (b) as having
ceased to apply at that point (past), and (c) as applying at a time-point after the
one imagined. Reh also describes perfect as having only the results that apply at
the time-point imagined.
Linguists generally postulate two different reference points, namely (i) an
“utterance time” or speech time, i.e. the time of uttering the sentence that is
under consideration, and (ii) a “reference time”, i.e. a point of time which may
be distinct from it and is generally denoted by a temporal adverbial. The need to
differentiate between these two types of reference time was originally put forth by
Reichenbach (1947) in order to differentiate between sentences like the following:
(28) a. I lost my pen yesterday.
b. I have lost my pen.
c. I had lost my pen by the time I came here.
The first sentence (28a) is in simple past tense, in which the reference time
(yesterday) is considered to be occurring prior to utterance time, but the event
time is simultaneous with it. The difference between this sentence and the next
one (28b), which is in present perfect tense, according to Reichenbach, is that in
the latter case the reference time is identical with utterance time, and only the
event time is prior to utterance time. This latter point is supported by the fact
that, in English, a sentence in present perfect cannot take an adverbial denoting
a past reference time. Example:
(28) d. *I have lost my pen yesterday.
In the case of the third sentence (28c) given above, which is in pluperfect
(or past perfect) tense, on the other hand, all these three reference points are
considered to be distinct from one another, with the reference time (the time of
my coming here) occurring prior to utterance time, and the event time (the time of
my losing my pen) occurring prior to that reference time (i.e. of my coming here).
Comrie (1985: 78) argues, however, that this representation of the distinction
between past and present perfect in terms of time location is untenable; the two
differ, according to him, in the fact that the latter involves an additional notion
of “current relevance”. On the other hand, other scholars like Declerck (1986)
argue that the two do differ from one another in their reference time (or rather
the “time of orientation”, according to Declerck) as well.
I would like to suggest, in this connection, that the postulation of two
different types of tenses, namely deictic and non-deictic, which may or may not
be dependent upon one another, would help us to resolve this problem to a
certain extent. For example, in the case of Dravidian languages, the so-called
perfect forms generally involve a combination of non-deictic and deictic tenses
with either of them having the ability to occur with temporal adverbials of their
own. They are not, therefore, comparable to simple past tense forms as far as the
occurrence of reference time is concerned.
As I will be arguing in a following section (see 2.5), temporal adverbials
function as parallel structures to tense markers. They make use of all the
parameters of tense, and introduce certain additional parameters as well, and
further, they are able to provide greater detail in the case of all these parameters.
However, it is necessary to describe the system of tense independently of these
adverbials because the latter are only optional elements that may be used to
modify tense, if necessary; they do not form an essential part of the tense system.
Further, as I will be pointing out in the second part of this monograph
(7.7.1), there are cross-linguistic differences in the categorial position of the
notion of “perfect”, with aspect-prominent languages including it under the
category of aspect and tense-prominent languages including it under the category

of tense. We cannot therefore establish an analysis of perfect that would apply
uniformly to all languages. In the case of languages in which the notion of
“perfect” gets included under the category of tense (like Dravidian languages), it
would perhaps be possible to claim that it is a combination of non-deictic and
deictic tenses, as I have pointed out above. Aspect-prominent and mood-promi-
nent languages, on the other hand, would use other devices for representing the
notion of “relevance at a specified point of time”.
2.4 Distance from the reference point
We have so far examined the occurrence of two different parameters in the

expression of tense, namely (i) relative position on the time line (before,
simultaneous or after) and (ii) the nature of reference point used (deictic or non-
deictic). The third parameter that needs to be examined in connection with the
category of tense concerns the relative distance of an event from a given
reference point. Instead of merely indicating that the event occurred before or
after the reference point, tense markers may specify further as to whether it
occurred immediately before or after the reference point, or whether the occur-
rence was farther removed from that point.
Distances from the reference point can be specified in great detail and
accuracy with the help of temporal adverbials in the case of all languages.
Consider, for example, the following Kannada sentences:
(29) a. nañu ninne beLigge e˜Lu˜ mu˜vatt-aid-akke bande
I yesterday morning seven thirty-five- came
‘I came yesterday morning at seven thirty-five’
b. nañu na˜Le beLigge e˜Lu˜ mu˜vatt-aid-akke barutteñe
I tomorrow morning seven thirty-five- come
‘I will come tomorrow morning at seven thirty-five’
The point which interests us here, however, is that in the case of some
languages, there are also distinct tense markers which are used for denoting
remoteness distinctions in the form of a set of simple contrasts, leaving the
function of providing greater details to adverbials (as in the case of the other two
parameters). I have already given an example for this possibility from Nkore-
Kiga (see 2.1) in which there is a three-fold differentiation in the case of past
tense (between today past, yesterday past and remote past) and a two-fold
differentiation in the case of the future tense (between near future and remote
future).
Remoteness distinctions are also expressed in Mishmi, a Tibeto-Burman
language of the North-Assam group. According to Sastri (1984: 132), Mishmi
makes a recent-remote distinction in the case of its past tense suffixes, and an
immediate-distant distinction in the case of its future suffixes. In the latter case,
the suffixes make additional distinctions of person and number (of the subject)
as shown below:
recent past so
remote past liyà
immediate future
III person à
other persons de
distant future
I singular ne
I plural ke
II person yà
III person bi], biya
definite li
The use of recent and remote past tense suffixes can be exemplified with
the help of the following sentences:
(30) a. ½ tape½ thá-so
ha
I rice eat-.
‘I ate rice’
b. ½ tape½ thá-liyà
ha
I rice eat-.
‘I ate rice’
The distinction between these two tenses, according to Sastri (1984: 132), is
in the time lag between the event and the utterance. This time lag, however, is
abstract in the sense that it is more of the speaker’s attitude towards the event
than the actual interval that has elapsed after the event is over and the speaker
talks about it. There is a constraint on the use of recent past as against that of
remote past in that the former cannot be used in contexts in which it has resulted
in another event or has been followed by some other event. Remote past, on the
other hand, does not usually occur with adverbs of very recent time like tyago
‘just now’ (unless the speaker intends to implicate another event as a conse-
quence of that event).
In the case of future tense also, the immediate form is used when the
speaker expects the event to follow the utterance without the intervention of any
other event or without any time lag. Examples (Sastri 1984):
(31) a. ½ thá-de
ha
I eat-..-
‘I shall eat’
b. cyá thá-a
he eat-..
‘He will eat’
Distant future, on the other hand, is used in contexts in which the time at
which the event is expected to take place is rather vague. Examples:
(32) a. ½ tape½ thá-ne
ha
I rice eat-..
‘I shall eat rice’
b. ní] tape½ thá-re-ke
we rice eat--..
‘We shall eat rice’
c. nyú a] haná-yà
you home come-..
‘You will come home’
d. wé a] bó-bì]
he home go-..
‘He will go home’
There is an additional remote future suffix li that occurs after the suffix kõ
denoting the definitive mood, or before the negative suffix Gm; it does not make
any personal distinctions. Examples:
(33) a. ½ tape½ thá-kõ-li
ha
I rice eat--.
‘I shall certainly eat rice’
b. ½ tha-l-Gm
ha
I eat-.-
I shall not eat’
This remoteness distinction, however, is restricted to action verbs. Stative

verbs (cognitive, sensory, existential and ambient verbs), on the other hand, make
use of a three-fold past-present-future tense distinction, by using the same set of
suffixes mentioned above. In the case of these latter type of verbs, the recent
past suffix so is used to denote simple past tense, whereas the remote past suffix
de is left unused; the immediate future suffixes denote present tense, whereas the
distant future suffixes denote future tense. Examples:
(34) a. jyimcyanewè jyím kasà-so
Jimchane truth know-
‘Jimchane knew the truth’
b. hawè masyetyó-de
I hungry-
‘I am hungry’
c. hawè bútyù jyím kasà-ne
I sometime truth know-
‘Sometimes I shall learn the truth’
Verbs containing the continuous aspect suffix bGrG also show a three-fold
past-present-future tense distinction in a similar fashion; they are not inflected for
the remoteness distinction. Verbs containing the repetitive-durative suffix ma, on
the other hand, show only a past-future tense distinction (Sastri 1984: 145).
It is interesting to note, in this connection, that the proximate-remote
distinction that Tulu (a Dravidian language) shows in past tense is restricted to
deictic tenses. The distinction occurs in finite verbal forms (35a, b) and also in
relative clauses (36a, b), which involve deictic tense, but not in converbs (37), as
shown below:
(35) a. ra˜me bat-t-e
Ram come-.-3:
‘Ram came (just now)’
b. ra˜me bat-tGd-e
Ram come-.-3:
‘Ram came (earlier)’
(36) a. bat-t-i ra˜me
come-.-. Ram
‘Ram who came (just now)’
b. bat-tGd-i ra˜me
come-.-. Ram
‘Ram who came (earlier)’
(37) ra˜me bat-tGd-G pa˜terye
Ram come-- spoke
‘Ram came and spoke (to someone)’
Notice that the converbal form in (37) uses the remote past tense suffix but it
does not make any remoteness distinction. Diachronically, the remote past form
derives from a periphrastic construction containing the auxiliary verb ‘to be’.
As I would be pointing out later (see 7.9), distinctions of temporal distance,
especially when they involve several affixes as in Mishmi, are basically modal
rather than temporal. They represent evidential distinctions in the sense that one
can be more sure about what happened today as compared to what happened
yesterday or several days or years earlier.
2.5 Use of temporal adverbials
Temporal adverbials have the function of modifying the temporal character of the
verb, or rather that of providing additional information about the location in time
of the event (or state) that the verb denotes. In order to carry out this function,
the temporal adverbials have to replicate tense by establishing a parallel structure
that is related to the temporal structure that the tense system denotes.
Notice that the temporal adverbials make use of all the parameters that are
used by tense markers such as the deictic/non-deictic distinction, prior-simulta-
neous-posterior distinction, and the immediate-remote distinction. But in addition
to this, they also make use of certain additional parameters such as the location
of an event between two different points of time (see below for examples).
The fact that the temporal structure represented by temporal adverbials is
parallel to the one represented by tense markers apparently derives from the
diachronic line of development, namely that tense is the grammaticalized version
of the temporal structure that the adverbials represent. However, tense is indepen-
dent of temporal adverbials and can stand on its own without the support of the
latter; the adverbials, on the other hand, are constrained by tense even though
there do occur some contexts in which they may conflict with tense (see 2.5.4).
Linguists have generally regarded temporal adverbials as providing an
additional parameter for tense; it is generally claimed that for a proper under-
standing of tense, it needs to be interpreted in association with temporal adverb-
ials (see Smith 1978, 1981, Declerck 1986, 1991). This claim, I think, is
untenable. The temporal adverbials only replicate (and expand upon) tense. The
additional information that they provide about the temporal location of the event
concerned is not essential or indispensable. That is, the distinction is similar to
the one between personal pronouns on the one hand and personal markers that
replicate gender-number-person distinctions that the system of personal pronouns
represents on the other. It is quite possible to interpret either of these two
independently of the other, even though the two are interconnected.
The term “temporal adverbials” is also used traditionally for denoting
adverbials which modify the aspectual character of the verb, i.e., adverbials which
indicate the duration, frequency, extent (from or to a particular point of time), etc.
of an event. I propose to examine these latter type of adverbials separately in the
next chapter (3.6). The present section is concerned only with adverbials that
provide additional information about the temporal location of an event.
It may be noted in this connection that even some of the most recent
researchers have failed to differentiate between temporal and aspectual adverb-
ials, and have thereby unnecessarily made the description of tense rather
complicated. For example, Declerck (1986, 1991) postulates four different
parameters for his theory of tense, namely (i) time of utterance (TU), (ii) time of
situation (TS), (iii) time or reference (TR), and (iv) time of orientation (TO), of
which only the first one is considered to be momentary; others can be momen-
tary or durative. The complexity of this system derives from the momentary-
durative distinction which, however, is only an aspectual distinction, and is not
directly relevant for a theory of tense.
2.5.1 Deictic/non-deictic distinction
The deictic/non-deictic distinction occurring among temporal adverbials can be

exemplified with the help of the following Kannada sentences:
(38) a. avanu ninne ban-da
he yesterday come-
‘He came yesterday’
b. *avanu ninne baru-ttañe
he yesterday come-.
(39) a. avanu na˜Le baru-ttañe

He tomorrow come-.
‘He will come tomorrow’
b. *avanu na˜e ban-da
he tomorrow come-
The two adverbials, ninne ‘yesterday’ and na˜Le ‘tomorrow’, are deictically
distinct as shown by the fact that the former cannot be used with a verb in non-
past tense (38b) and the latter cannot be used with a verb in past tense (39b).
Both of them, however, can occur with any of the three non-deictic tense forms,
i.e. prior, simultaneous or posterior. Examples:
(40) a. avanu ninne ban-du ha˜Dida
he yesterday come- sang
‘He came yesterday and sang’
b. avanu ninne aLu-tta˜ u˜Ta ma˜Dida
he yesterday cry- meal did
‘He dined yesterday crying’
c. avanu ninne ha˜D-alu mareta
he yesterday sing- forgot
‘He forgot to sing yesterday’
(41) a. avanu na˜Le ban-du ha˜Duttañe
he tomorrow come- sings
‘He will come and sing tomorrow’
b. avanu na˜Le ha˜Du-tta˜ kuNiyuttañe
he tomorrow sing- dances
‘He will dance singing tomorrow’
c. avanu na˜Le ha˜Dalu ho˜guttañe
he tomorrow sing- goes
‘He will go to sing tomorrow’
Non-deictic adverbials, on the other hand, can occur with any of the deictic
tenses. Examples:
(42) a. avanu ivattu ban-da
he today come-
‘He came today’
b. avanu ivattu baru-ttañe

he today come-.
‘He comes today’
There are some contexts in which tense markers leave certain distinctions,
such as for example the one between habitual and future in Kannada, unspecified
or ambiguous, and in such contexts, the use of an adverbial can help us to
specify one or the other of the two relevant meanings and thereby to disambigu-
ate the sentence. Examples:
(43) a. kuDi-du baruv-avarige kelasa illa
drink- come-them work not
‘There is no work for those who come after drinking’
b. na˜Le kuDi-du baruv-arige kelasa illa
tomorrow drink- come-them work not
‘There is no work for those who come after drinking tomorrow’
c. ya˜va˜galu˜ kuDi-du baruv-avarige kelasa illa
always drink- come-them work not
‘There is no work for those who always come after drinking’
Notice that there is ambiguity between habitual and future in (43a), which is
removed in (43b) and (43c) through the use of deictic and habitual temporal
adverbials respectively.
2.5.2 Positional distinctions
Temporal adverbials also show a distinction between prior, simultaneous and

posterior depending upon the position of an event relative to some other event.
The following Kannada sentences exemplify this distinction:
(44) a. nañu na˜lkak-ke modalu bande
I four- before came
‘I came before four (o’clock)’
b. nañu na˜lkak-ke bande
I four- came
‘I came at four’
c. nañu na˜lka-ra anantara bande
I four- after came
‘I came after four’
Notice that the temporal distinction that the adverbials indicate in these sentences
is independent of the tense distinction that the verb denotes; all the adverbials
occur with a verb in past tense in (44a–c), but it is also possible for all of them
to occur with a verb in non-past tense. Example:
(44) d. nañu na˜lkak-ke modalu barutteñe
I four- before come (.)
‘I will come before four’
The adverbials also make use of their own reference points, which are
distinct from the ones used by the tense forms that they modify. These reference
points may be denoted by nominals as in the sentences given above, or by
adjectival participles, as in the following sentences:
(45) a. avanu baru-va modalu nañu bande
he come-. before I came
‘I came before he came/comes’
b. avanu baru-v-a˜ga nañu o˜dutt-idde
he come-.-then I reading-was
‘I was reading when he came’
c. avanu ban-da me˜le nañu bande
he come- after I came
‘I came after he came’
All these adverbials can occur with non-past verbs as well. That is, the
reference points that they make use of do not affect the tense forms of these
verbs. Examples:
(46) a. avanu baru-va modalu nañu barutteñe
he come-. before I come
‘I will come before he comes’
b. avanu ban-da me˜le nañu barutteñe
he come- after I come
‘I will come after he comes’
As I had mentioned earlier, temporal adverbials could indicate additional
complexities about the temporal position of an event; in the following sentences,
for example, the adverbials indicate that the event occurred between two different
reference points:
(47) a. nañu mu˜rar-inda na˜lka-ra oLage ho˜gutteñe

I three-from four- inside go
‘I will go between three and four’
b. nañu a˜ke banda me˜le ni˜vu baruva modale˜ horaTe
I she came after you came before started
‘I started after she came but before you came’
The point to be noted here is that these complexities expressed by temporal
adverbials do not directly affect the temporal system that is being established by
the tense suffixes.
2.5.3 Remoteness distinctions
As I had mentioned earlier (2.4), there are some languages in which the relative
distance of a given event from a reference point is expressed by distinctions of
tense. In Mishmi, a Tibeto-Burman language, for example, there are distinct past
and future suffixes for denoting immediate and non-immediate past and future
events respectively. While these tense suffixes can indicate temporal distance
only briefly, temporal adverbials can provide minute details about it, and can
establish it with precision and complexity.
Notice, however, that in the case of languages in which remoteness distinc-
tions are specified by tense markers, the occurrence of adverbials which indicate
temporal distance would be constrained by tense markers. That is, adverbials,
once again, would be dependent upon tense markers, but the latter would be
independent of adverbials.
2.5.4 Conflict with tense
Temporal adverbials are found to conflict with tense primarily in sentences in

which a speaker is reporting the statement of some other person. There would be
two different deictic reference points in such sentences, namely (i) the utterance
time of the statement that is being reported and (ii) the utterance time of the
report itself. The conflict arises due to the fact that the temporal adverbials
occurring in the original statement are allowed, in the case of some languages,
to change their deictic reference point in order to agree with the utterance time
of the report, whereas the tenses of the statement are allowed remain unchanged.
Consider, for example, the following Kannada sentences:
(48) a. nañu na˜Le baru-tt-eñe

I tomorrow come-.-1
‘I will come tomorrow’
b. avanu ninne baru-tt-eñ-endu he˜L-idda
he yesterday come-.-1-that said-was
‘He had said that he would come yesterday’
c. *nañu ninne baru-tt-eñe
I yesterday come-.-1
*‘I will come yesterday’
Notice that there are two different deictic reference points in (48b) (which is a
report of (48a)) due to the fact that (48b) was uttered two days after (48a) was
uttered and (48b) includes (48a). Notice, however, that the tense of (48a) remains
unchanged in (48b), reflecting, correctly, the existence of two different deictic
reference points in (48b), but the temporal adverbial na˜Le ‘tomorrow’ changes
to ninne ‘yesterday’ (apparently to agree with the matrix verb) and thereby
produces a conflict with the tense of the embedded clause.
However, Kannada does not appear to allow deictic adverbials to change
from past to future when the matrix verb is in non-past tense and the embedded
verb is in past tense (see Comrie 1985: 113 for the possibility of a conflict
occurring in such cases in English). Example:
(49) d. ?avanu eraDu divasa kaledu na˜Le bande endu he˜Luttañe
he two day later tomorrow came that says
?‘He will say two days later that he came tomorrow’
Hindi, on the other hand, has a set of temporal adverbs whose meaning is
made more specific by the tense markers that occur with the verb; that is, the
adverbs and tense markers can be regarded as complementing one another in this
case. Hackman (1976: 122) points out that adverbs like kal ‘one day removed’,
parso ‘two days removed’, tarso ‘three days removed’ and ha˜l me ‘recently’ can
provide more specific meanings like ‘yesterday’ or ‘tomorrow’, ‘day before
yesterday’ or ‘day after tomorrow’, ‘two days before yesterday’ or ‘day after
tomorrow’ and ‘in the recent past’ or ‘in the recent future’ respectively, depend-
ing upon the fact as to whether the verb which occurs in the sentence is in past
tense or future tense. Examples:
(50) a. vah kal bambai gaya˜ tha˜

he one-day-removed Bombay went was
‘He went to Bombay yesterday’
b. vah kal bambai ja˜ega˜
he one-day-removed Bombay goes ()
‘He will go to Bombay tomorrow’
C 3
Category of Aspect
3.1 Introduction
As we have seen in the previous chapter, tense indicates the temporal location of
an event by relating it either with the event of uttering the sentence that denotes
the event (called deictic tense), or with some other event which is specified in
the sentence itself (called non-deictic tense). Aspect, on the other hand, indicates
the temporal structure of an event, i.e. the way in which the event occurs in time
(on-going or completed, beginning, continuing or ending, iterative or semelfact-
ive, etc.). This difference between the categories of tense and aspect can be
exemplified with the help of the following pairs of Hindi sentences in which
tense distinction is shown by the auxiliary verb and aspect distinction by the
aspect suffixes occurring with the main verb itself.
(1) a. mai a˜-ta˜ hũ˜
I come- am
‘I am coming’ (Present Imperfective)
b. mai a˜-ta˜ tha˜
I come- was
‘I was coming’ (Past Imperfective)
(2) a. mai a˜-ya˜ hũ˜
I come- am
‘I have come’ or ‘I am come’ (Present Perfective)
b. mai a˜-ya˜ tha˜
I come- was
‘I had come (at some past time)’ (Past Perfective)
Notice that the two sentence pairs (1a–b) and (2a–b) given above differ from one
another in aspect, with the former denoting an on-going event (imperfective
aspect) and the latter denoting a completed event (perfective aspect). The
sentences (a) and (b) under each of these pairs, on the other hand, differ from
one another in tense, with the former (1a, 2a) being in the present (non-past)
tense and the latter (1b, 2b) in the past tense. Kellogg (1938: 234) points out in
this connection that both (la) and (2a) agree in referring to an action occurring
in the present time. They differ from one another, according to him, in that the
former denotes an action as unfinished and the latter as finished. Similarly, both
(1b) and (2b) refer to past actions, but (1b) represents it as an action in progress,
and (2b) as a completed action. He points out further that Hindi allows its
aspectual forms to be used without any auxiliaries, and in such a usage they
denote only the aspect distinction and not any tense distinction. Examples:
(1) c. mai a˜-ta˜
I come-
‘I come, I would come’
(2) c. mai a˜-ya˜
I come-
‘I came’
Among the sentences given above, (1c) indicates an on-going and unfinished
action (at some unspecified time), whereas (2c) indicates a finished action (with
its point of time unspecified). The English translations of these two sentences
given here, however, are misleading, according to Kellog (1938), because they
imply distinctions of tense as well.
The temporal (aspectual) structure of an event can show several other types
of distinctions such as, for example, that the action may be momentary or
durative, involving change (active) or not involving change (stative), occurring
once (semelfactive) or occurring several times (iterative), occurring on a specific
occasion or occurring habitually, and so on. Languages differ, however, in
grammaticalizing one or more of these distinctions in their system of aspects. It
has been suggested that these various types of aspectual distinctions can be
divided into three distinct groups, namely (i) perfectives and imperfectives, (ii)
ingressives, progressives, egressives and resultatives, and (iii) semelfactives,
iteratives, habituals and frequentatives (see Dik 1989, Siewierska 1991). The first
one is concerned with the distinction between the view of an event as a whole
from outside versus the view of an event from inside. The second one, on the
other hand, distinguishes between different phases of an event, and the third one
represents distinctions concerning the various quantificational aspects of an event.
It has been suggested further that the position of aspect markers in a verbal form
CATEGORY OF ASPECT 45
would be correlatable with this three-fold division, with the markers for the
perfective-imperfective division occurring closest to the verbal base and the
phasal and quantificational aspects occurring away from the base.
There is also a claim, made by some linguists, that it is necessary to
differentiate between two main types of “aspectual” distinctions, by using the
term “aktionsart” (a German word meaning “kind of actions”) for referring to
different kinds of events like processes and states, momentary and durative
events, telic (resultative) and atelic events, etc., and by restricting the term
“aspect” to the various ways of viewing the events, i.e. as complete or incom-
plete, specific or habitual, beginning (ingressive), continuing (progressive), or
ending (egressive) etc. (see Smith 1986, Brinton 1988, Bache 1995). We may
regard the former as non-deictic and the latter as deictic (as the latter involve a
speaker’s view of the event), but generally the distinction is considered to be one
between lexicalization versus grammaticalization, with the former being lexical
(or derivational) and the latter grammatical (or inflectional). The former are also
called “situational aspects” in contrast with the latter which are called “view-
point aspects”.
There are some difficulties, however, in maintaining this aktionsart-aspect
distinction consistently in a cross-linguistic study of languages. One interesting
question that has been raised in this connection concerns the correlatability
between lexical and grammatical representations on the one hand, and situations
and viewpoints on the other. Certain aspectual (viewpoint) distinctions may occur
as lexical or derivational distinctions in some languages and as inflectional
distinctions in others. Further, the distinction between derivation and inflection
may also not be sharp and clear-cut in some languages. In spite of these
problems, however, the distinction can be used as a helpful device for a better
understanding of the category of aspect.
3.2 Perfective and imperfective
The most important aspectual distinction that occurs in the grammars of natural
languages is the one between perfective and imperfective. It primarily indicates
two different ways of viewing or describing a given event. Perfective provides
the view of an event as a whole from outside whereas imperfective provides the
view from inside. The former is unconcerned with the internal temporal structure
of the event whereas the latter is crucially concerned with such a structure. The
former views the situation as bounded, and as forming a unified entity whereas the
latter views it as on-going or habitual (see Comrie 1976, Dahl 1985, Bache 1995).
There are several languages in which this perfective-imperfective distinction
forms the basic division of verbal forms, with other aspectual, temporal and
modal distinctions being regarded as forming different subdivisions of either the
perfective or the imperfective category, or of both of them. In Supyire (Gur
family of Niger-Congo), for example, the great majority of verbs have two
forms, (i) a base, which is perfective, and (ii) a derived form, which is imper-
fective. Most tense-aspects require one or the other of these two forms, but a few
like the habitual may take either (Carlson 1994: 130). In Kiowa of New Mexico
and Arizona (Kiowan family), on the other hand, both perfective as well as
imperfective stems can form the bases of several types of forms like past,
future/potential, imperative, etc. Only the negative form is restricted to the
perfective stem (Watkins 1984).
According to Berntsen and Nimbkar (1975, 1982), verbal participles
denoting perfective and imperfective aspects (the terms used by them are
“perfect” and “imperfect” respectively) form the basis of several tense forms in
Marathi. The meaning distinction that these two participles indicate, according to
them, is between (i) an action viewed as complete and (ii) an action viewed as
in progress or repeated. The participles are followed by auxiliary verbs for
denoting tense (and mood) distinctions. The following pairs of sentences
exemplify this contrast:
(3) a. to tikDe jat ae
he there going () is
‘He is going there (right now he is on the way)’
b. to tikDe gel(a) ae
he there gone () is
‘He has gone there’
(4) a. to tikDe jat hota
he there going () was
‘He was going there (he was on the way)’
b. to tikDe gela hota
he there gone () was
(i) ‘He had gone there’
(ii) ‘He went there’
(5) a. to tikDe jat 6sel

he there going () might
‘He is probably going there’
b. to tikDe gela 6sel
he there gone () might
‘He has probably gone there’
It may be noted, however, that the English translation is rather misleading
in these sentences. For example, Berntsen and Nimbkar (1982: 349) point out that
the verb form in (4b), even though translated with English past perfect, is also
often used to translate English simple past. Further, it is possible to take the time
frame for granted once it has been set for the discourse, and therefore to drop the
auxiliary in subsequent sentences. It is also possible to drop the auxiliary when
the time frame is denoted by the speech context. Example (Berntsen and
Nimbkar 1975: 94):
(6) mi hOtel-m6dhe b6slo hoto;
I restaurant-in sit () was;
ek maNus at yeun majhya-j6v6L b6sla
one man in came me-near sit ()
‘I was sitting in a restaurant; a man came in and sat near me’
Notice that the completive past is expressed with the help of a perfective form
(b6slo) and a past auxiliary (hoto) in the first clause, whereas in the second one
it is expressed by the perfective verb (b6sla) occurring without the following
auxiliary.
According to Jeanne (1978), most Hopi verbs occur in perfective-imperfect-
ive pairs. The perfective is simpler of the two forms in the majority of cases. A
number of these can be viewed as forming their imperfective by means of
suffixation to a base identical to the perfective. A number of them take a k-
increment and some involve reduplication. Examples:
Gloss Perfective Imperfective
‘send home’ hoona hoon-nta
‘write’ peena pen-ta
‘enter’ paki paki-wta
‘pull’ la]a la]a-ki-nta
‘exit’ yama yayma
‘give’ maqa mamqa
‘scatter’ cala calala-ta

‘drip (leak)’ cina cinana-ta
Jeanne (1978) points out that this aspect distinction is different from tense
distinction. Perfective verb denotes the completion of an action or a process in
relation to a point of time. It denotes past tense only in the simplest cases.
Similarly, imperfective denotes an on-going process only in the clearest cases.
There are also other meanings, like repetitive, when reduplication is used, or
when the suffix wta is used, for denoting it. Hopi also has a future suffix ni
which can occur with both imperfective as well as perfective forms.
There is thus a tendency for perfective verbs to indicate past events, and
also events that are punctual or resultative, and for imperfective verbs to indicate
progressive and durative events, or events that do not terminate in a result.
However, neither of these two sets of characteristics is essential for the perfec-
tive or imperfective to occur in a given context. It is quite possible, for example,
for a future event to be viewed and described as perfective and a past event as
imperfective; similarly, an event presented from a perfective viewpoint can be
durative or progressive and the one presented from an imperfective viewpoint
can be resultative.
For example, Noonan (1992: 136) reports that in Sango (Nilo-Saharan family
of East Africa) verb is inflected for three aspects, perfective, progressive and
habitual, but not for tense; out of context, perfective will be interpreted as past,
habitual as present and progressive as present or future. But these can be
assigned any tense with which they are semantically compatible; when appropri-
ate temporal adverbials are used, perfective can have past or future (but not
present) interpretations, and the other two can have past, present or future
interpretations. Melotki (1985: 625) reports the situation in Hopi (for the perfec-
tive and imperfective forms of the verb) to be similar. In Kiowa, on the other
hand, the imperfective verb denotes habitual and repeated activities and also
events in progress, even if they are past events or activities. The language has a
future perfective form which is used for denoting repeated events that are to take
place in a bonded period of time (Watkins 1984: 157).
The presence of the above-mentioned correlation between perfective-
imperfective aspect distinction on the one hand, and past/non-past tense distinc-
tion on the other, has led to the description of verbal forms, which denote
primarily an aspectual distinction, as involving the past/non-past tense distinction.
In the case of Indo-Aryan languages, for example, the grammars tend to use
familiar terms like past and present for naming verbal forms which actually
represent the corresponding aspectual distinctions (see Masica 1991). Li

(1991: 26) points out that, in most of the literature, Hmong (of Southern China)
is considered as having a full-fledged past-present-future tense system. But a
careful examination of the concerned forms reveals that the suffix tan, which is
regarded as a past tense marker, is actually an aspectual one, as it denotes the
“attainment” of an event, which may be past or future. This suffix, along with
the suffix lawm ‘completion’, according to Li, can be regarded as representing
the perfective aspect, whereas taabtom ‘progressive’ can be regarded as repre-
senting the imperfective aspect. Li considers the suffix yuav ‘future’ to be the
only tense marker in Green Hmong.
We also find grammarians frequently using the terms “perfect” and “imper-
fect” for denoting the perfective-imperfective aspect distinction (see, for example,
Berntsen and Nimbkar 1982, or Borg and Azzopardi-Alexander 1997) but the
term “perfect” is also used for denoting a somewhat different concept, namely a
completed event whose result persists at a specified point of time (as in the case
of the English present perfect has come). Perfective needs to be differentiated
from this latter concept because it does not specifically say anything about the
result of the event or action concerned or its relevance to the present moment. It
has therefore been suggested (Comrie 1976: 52) that the use of the term “perfect”
may be restricted to this latter concept (‘present relevance of a previous event’)
and the term “perfective” may be used exclusively for referring to the aspect that
I have been describing in detail in this section.
3.3 Phasal aspects
An event may have a beginning and an end, a middle portion (continuing or

changing), and also an ensuing result or an altered state. These are considered to
be the various “phases” of an event. A speaker may talk about an event from the
point of view of any of these individual phases, and his language may have
inflectional (or other type of) markers for representing these distinctions. Since
such markers indicate distinctions in the temporal structure of an event, we may
regard them as belonging to the category of aspect. It has been suggested (Dik
1989: 186) that these may be grouped under a subcategory (or “level”) of aspect
called “phasal aspect”.
Languages appear to differentiate between (i) the beginning, stoppage and
ending of an event on the one hand, and (ii) its middle portion (continuing or
changing part) on the other, by affiliating the former with the perfective aspect
and the latter with the imperfective aspect (see Comrie 1976: 19). The fact that
the former denote the boundaries of an event and the latter denote its middle
portion may be the basis of this correlation. As we have seen earlier (3.2),
perfective provides a view of the event from outside and hence the boundaries of
the event are in its view, whereas the imperfective provides a view of the event
from inside and hence what is in its view is only the middle portion of the event.
This correlation occurs in several of the Indo-Aryan languages in which the
imperfective has been split into two distinct paradigms, namely habitual and
continuous, by restricting the use of the imperfective for denoting habitual
meaning and by attaching the verb rah ‘remain’ to the conjunctive participle (or
to the verbal base itself) for denoting the continuous meaning (Masica
1991: 274). This latter form apparently started as a compound verb construction
but in some of the languages like Hindi, it has been grammaticalized into a
continuous (or progressive) participle, as shown in the following sentences
(Hackman 1976: 97):
(7) a. laRka˜ patr likh-ta˜ hai
boy letter write- is
‘The boy writes letters (habitual)’
b. laRka˜ patr likh raha˜ hai
boy letter write remain () is
‘The boy is writing a letter’
Hackman notes that (7a) may also have the present (progressive) meaning, but
(7b) is used more commonly than (7a) in that sense.
This distinction between the phasal aspects of the boundaries on the one
hand, and of the middle part on the other, also appears to affect the grammatical-
ization of phasal aspects. The former are generally expressed through lexical
means whereas the latter are more frequently indicated by inflectional markers.
In Manipuri, a Tibeto-Burman language, for example, there is a set of verbal
roots for denoting the beginning, stopping and ending of an event; there are actually
several types of stoppages for each of which the language has a distinct verbal root.
These aspectual verbs are attached to the infinitive form of the main verb. The
following verbs of this type have been recorded by Bhat and Ningomba (1997):
(i) Verbs of beginning
h6w ‘start’
thu ‘start quickly’
(ii) Verbs of stopping

tok ‘stop intentionally’
lep ‘stop due to interference’
let ‘almost stop’
kh6k ‘stop (bleeding)’
ph6 ‘stop being something’
kham ‘stop (causative)’
thi] ‘stop by putting obstruction’
(iii) Verbs of ending
loy ‘finish’
pha ‘complete’
The following sentences illustrate the use of some of these aspectual verbs
in Manipuri:
(8) a. m6hak c6t-p6 h6w-wi
he walk- start-.
‘He started to walk (go)’
b. m6hak c6t-p6 tok-i
he walk- stop-.
‘He stopped walking’
c. i thok-p6kh6 k-i
blood bleed- stop-.
‘The bleeding stopped’
d. m6hak saw-b6 ph6-y
he anger- stop-.
‘He stopped being angry’
e. 6yn6 m6hak c6t-p6 thi]-]i
I him walk- stop-.
‘I stopped him going’
f. m6hak ca-b6 loy
he eat- finish (.)
‘He finished eating’
For denoting the starting of an event, Manipuri may use either the verb h6w
mentioned above, or the suffix kh6t. The two differ from one another in that the
former denotes the starting of an event that may occur several times (habitual),
whereas the latter denotes the starting of an event that may occur only once.
Examples:
(9) a. m6hak-n6 phu-b6 h6w-wi
he- beat- start-.
‘He began to beat (and would continue to do so)’
b. m6hak-n6 phu-g6t-li
he- beat-start-.
‘He began to beat (but would beat once only)’
For denoting distinctions connected with the middle phase, on the other
hand, Manipuri uses two different aspectual suffixes, namely (i) lì denoting
continuation or iteration, and (ii) le denoting the change that results from an
event’s occurrence. There is also a third suffix l6m, which indicates the comple-
tion (or the cessation of a state). This third suffix is rather different from the
other two in that it can only be a medial one. The following sentences exemplify
the use of these suffixes:
(10) a. tomb6 layrik pa-rì
Tomba book read-
‘Tomba is reading the book’
b. iran-d6 numit khudingi mi yamn6 si-rì
Iran- day every man many die-
‘Many people are dying every day in Iran’
(11) a. ce 6si mu-re
paper this black-change
‘This paper has become black’
b. 6yn6 tho] thi]-]e
I door close-change
‘I have closed the door (it will not be open now)’
(12) a. m6si mu-r6m-mi
it black--.
‘It was black (but not any more)’
b. yumth6k 6du yu-r6m-mì
roof that leak--
‘That roof had been leaking (but not any more)’
c. m6hak h6wjik ca-r6m-g6ni
he now eat--
‘He would still be eating now (incomplete)’
3.4 Quantificational aspects
The third type of aspectual distinction that gets represented in natural languages
concerns the quantificational characteristics of events. A speaker may report an
event as occurring once only (semelfactive) or several times (iterative); he may
view it as a specific event or as part of a general habit of carrying out similar
events; he may also differentiate between different degrees of frequency with
which the event occurs. The markers that a given language provides for one or
more of these meaning distinctions can be grouped under a subcategory called
“quantificational aspect”, as all of them refer to the quantitative aspect of the
event concerned.
Among these quantificational aspects, habitual differs from iterative and
frequentative crucially by the fact that the former is inductive whereas the latter
are deductive. The latter can only be based upon the observation of several
occurrences of the event concerned, whereas the former can be based upon the
observation of a single occurrence. It can even be used by a speaker who has not
actually observed any of the occurrences of the event concerned, as for example
when he states the “habitual” arrival of a train by simply looking at the time table.
Notice, further, that frequency denoting adverbials like ‘once’, ‘twice’, ‘ten
times’ or even ‘several times’ are not directly relevant for the use of the habitual
aspect. As has been pointed out by Comrie (1976: 27), if someone coughed five
times, this does not lead to an observation with the habitual aspect, namely that
*the person used to cough five times. That is, the use of the habitual marker used
to cannot go with the frequentative adverbial five times; the latter is deductive in
nature, whereas the former (habitual) is inductive in its establishment. However,
the use of the former does involve quantification over a set of occasions in the
sense that the event is predicted (inductively) to be occurring on a majority of
such occasions (Dahl 1985: 96).
We may differentiate between iterative and frequentative by the fact that the
former portrays events repeated on the same occasion (like the iterative knocking
on the door) whereas the latter portrays events repeated on different occasions
(like someone climbing a hill frequently). However, such a distinction can only
be subjective in nature because it depends upon a distinction in the temporal gap
between individual occurrences of the event, and this gap can be of different
degrees of width.
I had described durative and progressive as belonging to the phasal subcate-
gory in the previous section, but they also show affinity with iterative and
habitual aspects and hence they can as well be included in the quantificational
subcategory. As I will be pointing out below, for example, when momentary
verbs are used in their durative or progressive forms, they provide iterative
meanings, and further, languages also use their durative forms for denoting
habitual meaning.
Languages that differentiate between perfective and imperfective aspects
generally express habitual and iterative meanings with the help of their imper-
fective forms. In Kiowa, for example, the imperfective verb covers a variety of
non-completed events that include general statements, habitual or repeated
activities and events in progress. Perfective forms, on the other hand, indicate a
single completed event. This perfective-imperfective distinction occurs in the
case of the imperative, future (or potential) and also hearsay statements in this
language, and in all such cases, the imperfective denotes a continuous, repeated
or habitual event and perfective denotes a bounded event (Watkins 1984).
In Indo-Aryan languages also, the imperfective form of the verb generally
expresses habitual meaning. In Marathi, for example, the imperfective participle
is used without any accompanying auxiliary verb, but with the personal markers
directly attached to it, for denoting the present habitual meaning. Examples
(Berntsen and Nimbkar 1975, 1982):
(13) a. to roj amčya k6De ye-t-o
he daily our place come--3:
‘He comes to our place daily’
b. te v6ršatn6 donda p6NDh6rpur-la ja-t-at
they year () twice Pandharpur-to go--3
‘They go to Pandharpur twice a year’
Masica (1991: 294) refers to the occurrence of several habitual paradigms in
Punjabi, which are derived by adding different auxiliary verbal forms to the
imperfective participle:
(14) a. au-nd-a˜ hai
come--3: be ()
‘He comes’ (present habitual)
b. au-nd-a˜ si
come--3: be ()
‘He used to come’ (past habitual)
c. au-nd-a˜ hove
come--3M be ()
‘Perhaps he comes’ (subjunctive habitual)
d. au-nd-a˜ hove-ga˜
come--3M be ()-
‘He probably comes’ (presumptive habitual)
e. au-nd-a˜ hu-nd-a˜
come--3M be--3:
‘had he come regularly’ (contrafactive habitual)
In addition to these, the imperfective praticiple can also be used by itself in
Punjabi to denote ambiguously unspecified habitual or contrafactive habitual.
Other quantitative aspects like iterative and frequentative are represented by
more complex constructions. In Hindi, for example, frequentative is formed by
attaching the imperfective form of the verb kar ‘do’ to the perfective form of the
main verb. There is also a repetitive form which is derived by attaching either ja˜
‘go’ or rah ‘remain’ to the imperfective form of the main verb (Kachru
1980: 48). Examples:
(15) a. vah niyamit ru˜p se yahã˜ a˜y-a˜ kar-ta˜ tha˜
he regular way by here come- do- was
‘He used to come here regularly’
b. vah kal din bhar paRh-ta˜ rah-a˜
he yesterday day whole read- remain-
‘He kept reading all day yesterday’
Several languages have been reported to use reduplication (complete or
partial) of the verbal base for denoting the iterative aspect. In Mundari (Austro-
asiatic), for example, verbal bases are partially reduplicated or the first vowel
lengthened (or both) for denoting repeated actions, or more commonly, habitual
actions (Hoffmann 1903: 182, Osada 1991: 92). The verb dal ‘hit’, for example,
when reduplicated as dadal, provides the meaning ‘hit repeatedly’; it also has the
meaning of ‘being in the habit of beating’ or of being ‘quick to beat’.
In Santali, another Austroasiatic language, repetition of the verbal base has
the function of denoting repetition or continued performance of an act; it may
also indicate repetition of the same act towards different objects or in different
places (Bodding 1929: 179). According to Deeney (1975: 58), Ho makes a
distinction between reduplication of the first syllable and lengthening of the first
vowel, with the former stressing the repetition of an action and the latter
stressing the habitual occurrence of the same. In the case of verbs that begin with
a vowel, however, reduplication would be obligatorily accompanied by vowel
lengthening. Examples:
(16) a. sen ‘walk’ sesen ‘walk repeatedly’
b. ur ‘dig’ u˜ur ‘dig repeatedly’
(17) a. iD ‘take’ i˜D ‘take habitually’
b. horo ‘guard’ ho˜ro ‘guard habitually’
Mao-Naga, a Tibeto-Burman language, uses the suffix we for denoting the
habitual aspect (Giridhar 1994: 287). Examples:
(18) kaikho ocükothuni mail kaxi tu-we
Kaikho daily mile two run-
‘Kaikho runs two miles daily’
The suffix can be preceded by Ti to denote that the habit is infrequent or
irregular, or by the marker makra to denote that its occurrence is regular. Exam-
ples:
(19) pfohi ve-Ti-we
he steal-infrequent-
‘He steals infrequently’
(20) ciThi.kopfomüi vu-makra-we
postman comes-regularly-
‘The postman comes regularly’
The frequentative meaning can be expressed by reduplicating the verb that
occurs with the habitual suffix; the reduplication is partial in the case of polysyl-
labic verbs. Examples:
(21) a. pfokrehrü pfokho he vuvu-we
Pfokruhru Pfokho to go (reduplicated)-
‘Pfokruhru keeps going to Pfokho’
b. hehi ocü irürü-we
here rain rain (reduplicated)-
‘It keeps raining here’
There is a connection between quantificational aspects occurring with verbs
on the one hand, and number distinctions occurring with nouns on the other.
Aspectual markers may denote plurality of arguments such as the agent, patient,
experiencer, location, etc., in addition to (or instead of) the plurality of actions in
some of their usages. For example, in Manipuri, the suffix m6n can denote both these
types of plurality. It can also indicate excess or intensity in connection with one of
the adverbials that occur in the sentence. Examples (Bhat and Ningomba 1997):
(22) a. ma c6y-m6l-li
he abuse-excess-.
‘He abuses a lot’
b. ma mi yamn6 c6y-m6l-li
he man many abuse-excess-.
‘He abuses too many people’
c. mabu mi yamn6 c6y-m6l-li
him man many abuse-excess-.
‘Too many people abuse him’
(23) nc] t6pn6 yon-m6l-le
you slowly sell-excess-
‘You have been too slow in selling’
Manipuri also makes use of the process of reduplication for denoting the
meaning of excess; the reduplication, however, is combined with prefixation.
There are two prefixes, pum and i, that can be attached to a verbal root, before
the root is reduplicated. The two prefixes primarily have the completive meaning,
but they can also provide quantitative connotation. Examples:
(24) a. pum-le] le]-]i
complete-throw throw-.
‘(He) throw away everything’
b. i-kaw kaw-wi
complete-forget forget-.
‘(He) forgot everything (or completely)’
The latter prefix is more frequently used with the negative marker. Example:
(25) i-nok nok-tre
complete-laugh laugh-
‘(He) laughs only occasionally’
Wintu, on the other hand, uses the generic suffix s, added to substantives,
in order to indicate meanings which are similar to that of imperfective and
durative verbs (Pitkin 1984; see also 7.6.2 below).
3.5 Situational and viewpoint aspects
As I had mentioned earlier, some scholars claim that a distinction needs to be

made between two different types of temporal structures that can be associated
with a given event or situation. The structure may be in the situation itself, and
may show distinctions such as, for example, between (i) events that have,
inherently, an ending (“telic”) and the ones which do not have any ending
(“atelic”), (ii) situations which involve change (events) and situations which do
not involve any change (states), (iii) events which involve some duration
(durative) and events which do not involve any duration (punctual), and so on.
We may regard these inherent distinctions in the temporal structure of situations
as belonging to a category called “aktionsart” “situational aspect”, “actionality”
or merely “acion” (see Platzack 1979, Bache 1982, 1994, Smith 1983, 1986,
Brinton 1988).
The second type of temporal structure that can be associated with a situation
is based upon the speaker’s viewpoint. As we have seen earlier, a speaker may
view a situation from outside as a whole (perfective) or from inside as on-going
(imperfective); he may refer to it from the point of view of its beginning,
continuity or progression, end or stoppage, or result; he may also describe it as
forming part of a habit, or of a series of events, emphasizing one or the other of
its quantificational characteristics. This latter type of non-inherent temporal
structure that a speaker can ascribe to a situation may be regarded as the
“aspect” proper, or as the “viewpoint aspect”, according to these scholars.
The question that has been raised in connection with this proposed distinc-
tion in the category of aspect (or between two entirely different categories called
aspect and aktionsart) is whether there is sufficient grammatical basis for its
postulation, i.e., whether there are languages in which a sharp and clear-cut
distinction is made between the two such that the two need to be assigned to
distinct systems of representations.
The distinction is comparable to the deictic/non-deictic distinction occurring
among tenses (see 2.4.1), and also to the distinction between judgements and
evidentials occurring among epistemic moods (see 4.2), in the sense that the
subjective-objective dichotomy gets reflected in all these three cases. The
contrast is rather clear-cut in the case of tense (in fact, the non-deictic variety of
tense has been provided with a distinct name, “taxis”, by some scholars), but in
the case of both aspect and mood, there is a lot of overlap. For example,
judgements would be based upon the type of evidence that the speaker has about
a situation, and similarly, the speaker’s view of the temporal structure of a

situation would be highly influenced by its inherent structure, and further, his
understanding of the latter (situational aspect) would also be influenced by his
own viewpoint.
Thus, Bache (1982: 65) finds it necessary to concede that (i) aktionsart is
not always “objective” as it may involve the speaker’s conception of the
situation, and (ii) that aspect is also not always “subjective” as the speaker’s
choice of an aspect may be dictated by objective considerations to a considerable
extent. He also finds it necessary to regard aktionsart (or actionality) as markedly
different from tense and aspect in that is gets expressed, very frequently, by
lexical distinctions rather than by grammatical distinctions (see Bache 1995).
There are also problems in establishing distinct systems of representation for
aspect and aktionsart. Some scholars suggest that we might view aspect as the
grammaticalization of the temporal structure and aktionsart as the lexicalization
of the same. The problem with this proposal is that it would only provide a
formal basis for the distinction, and not a functional or semantic basis that can
be consistently correlated with it. For example, the distinction between events and
states is generally considered to be an aktionsart, as it is represented only lexically
in some of the familiar languages, but there do occur several languages in which
it has a grammatical representation. For example, the distinction between perfec-
tive and imperfective forms of Hindi has been correlated by Trask (1979: 396)
with an underlying state-process distinction. It cannot therefore be consistently
regarded as an aktionsart (or aspect) on the above-mentioned formal basis.
The distinction, on the other hand, is useful in that it allows us to state
explicitly some of the constraints that languages appear to impose upon the use
of aspectual markers with different types of verbs. For example, the habitual
aspect marker we does not occur with a class of stative verbs in Mao Naga as we
have seen in the previous section; continuatives, when used with momentary
verbs, provide iterative or frequentative meaning in several languages. Bache
(1995) lists certain other possible constraints that actionsart (actionality) can
impose upon the occurrence of aspect (and also of tense) such as punctuality or
telicness being incompatible with imperfective aspect and also with present tense.
Such constraints on the use and connotation of aspect markers can be explained
as resulting from an interaction between situation and viewpoint aspects. There
is apparently a need to make a more thorough study of such interactions.
3.6 Use of aspectual adverbials
As I had mentioned in the previous chapter (2.6), temporal adverbials may be

used to modify either the temporal location (tense) of an event, or the temporal
structure (aspect) of an event. Both these types of adverbials are generally called
temporal adverbials, but it would be useful to regard only the former as “tempo-
ral adverbials” and to use the term “aspectual adverbials” for referring to the
latter type of adverbials. The contrast between these two types of adverbials can
be seen clearly in the following pair of Kannada sentences:
(26) a. avanu mu˜ru gaNTe-ge nidde ma˜Dida
he three hour- sleep did
‘He slept at three o’clock’
b. avanu mu˜ru gaNTe nidde ma˜Dida
he three hour sleep did
‘He slept for three hours’
Notice that the adverbial in (26a) is associated with the tense marker occurring
in the verb, whereas in (26b) it is associated directly with the verbal base, i.e. the
situational aspect or aktionsart denoted by it.
The following pair of sentences exemplify the occurrence of telic-atelic
distinction among aspectual adverbials:
(27) a. avanu mu˜ru gaNTe o˜dida
he three hour read
‘He read for three hours’
b. avanu mu˜ru gaNTe-yalli o˜di mugisida
he three hour- read () finished
‘He finished reading (the book) in three hours’
Aspectual adverbials may denote either the extent of duration (quantity) of
a given event as it is, as in the preceding examples (26b, 27), or the extent from
the point of its ingressive or egressive phases; they may also indicate other
aspectual distinctions like frequency or habituality as well. Examples:
(28) a. avanu na˜lka-ra varege nidde ma˜Dida
he four- until sleep did
‘He slept until four o’clock’
b. avanu mu˜ru ba˜ri nidde ma˜Dida

he three time sleep did
‘He slept three times (i.e. on three occasions)’
c. avanu ya˜vagalu˜ nidde ma˜Duttañe
he always sleep does
‘He always sleeps (is always sleepy)’
Since the situation and viewpoint aspects indicate roughly the same type of
distinctions that are denoted by aspectual adverbials (with the latter providing
more specific information) we can expect the two to be constraining one another.
A telic adverbial, for example, would require the verb to be telic; when used
with an atelic verb, it may either force a telic interpretation to the verb, or the
sentence would become unacceptable.
Grammars are generally silent on these and other characteristics of aspectual
adverbials; the topic clearly needs to be studied in detail.
C 4
Category of Mood
4.1 Introduction
Mood is concerned with the actuality of an event. There are three different
parameters that are used by languages while establishing modal distinctions;
these are the following:
(i) a speaker’s opinion or judgement regarding the actuality of an event,
(ii) kind of evidence that is available for the speaker to form this judge-
ment, and
(iii) kind of need or requirement which forces the speaker (or someone
else) to get involved in an event (or to carry out an action).
The first two parameters establish “epistemic” (knowledge-based) moods and the
third one establishes “deontic” (action-based) moods (see Palmer 1986: 51, 96).
In addition to these three types, the category of mood is generally considered to
include illocutionary forces like interrogatives (an extension of epistemic moods)
and imperatives (an extension of deontic moods).
Judgements can be of different types depending upon the confidence that the
speaker has in asserting the occurrence of an event. He may consider the event
to be real or unreal (imaginary or hypothetical) and further, he may be sure or
unsure about his own judgement in this regard. Some of the languages use
distinct mood markers in order to represent these distinctions in the speaker’s
assessment of the reality of an event. Evidentials, on the other hand, represent
the various bases that a speaker can use for specifying the reality of an event. He
might have actually observed the event or experienced it through his own senses.
Alternatively, someone else might have observed it and has reported it to the
speaker. It is also possible for the event to be inferred or deduced by the speaker,
or derived through induction (as in the case of generic or habitual statements).
These and other similar distinctions in the evidentiality of a statement can be

represented in a language with the help of different modal suffixes.
We may consider the distinction between judgements and evidentials as
correlatable with the distinction between deictic and non-deictic tenses, or
between viewpoint and situation aspects that we have encountered in the previous
two chapters. All these three involve basically a distinction between the speaker’s
own assessment of the situation on the one hand and distinctions occurring in the
situations themselves on the other.
Deontic mood is rather different from these two types of moods in that it
refers to different degrees of external or internal compulsion which forces the
event to take place. It is modal in the sense that the stronger the compulsion, the
more certain one would be that the event has taken place or is going to take
place. This relationship between deontic and epistemic (judgement) moods gets
reflected in the fact that in several languages like English, German, and also the
Dravidian languages like Kannada, the same form is used ambiguously for
denoting both deontic as well as epistemic moods. There are also languages (like
Ladakhi), however, in which these two types of moods are represented by
distinct markers (see 4.3 below).
The inclusion of interrogatives in the modal category results from the fact
that a speaker uses an interrogative sentence in order to augment or strengthen
his knowledge about an event; they are therefore closely related to epistemic
moods like doubt and uncertainty. Imperatives, on the other hand, are modal in
the sense that they form part of the external compulsions that force an event to
take place, and are therefore closely related to deontic moods. Languages with a
complex modal system show different interrogative and imperative markers
indicating different kinds of questions and commands.
4.2 Epistemic mood
Judgements and evidentials are clearly quite distinct from one another, but they
are also related in that the latter form the basis for the former. That is, one
judges an event to be real or unreal, certain, definite, probable or improbable on
the basis of the kind of evidence that he has about it, such as, for example,
whether he has actually seen or experienced it, or only heard about it, the kind
of source on the basis of which it has been reported, etc. Languages may use
different systems of markers for denoting these two types of moods (in which
CATEGORY OF MOOD 65
case they may even allow the two types of markers to occur together), or they
may include both of them under a single system of markers.
As mentioned earlier, it is also possible to regard judgements and evidentials
as two different facets of epistemic mood in the sense that judgements represent
the speaker’s own evaluation of a situation and evidentials represent the external
evidence (or basis) for an evaluation That is, we may regard judgements as
“deictic” and evidentials as ”non-deictic”. This relatedness between judgements
and evidentials makes it possible for languages to give prominence either to the
notion of judgement or to that of evidentiality in their verbal system. There are
languages like Mao Naga in which judgement (and the realis-irrealis distinction
that is based upon it) plays the most prominent role, whereas there are also
languages like Tuyuca in which evidentiality plays the central role. Languages
may combine the two together into a more complex system of epistemic mood
as well.
4.2.1 Realis and irrealis
The most important distinction in the category of mood is the epistemic one
between realis and irrealis. It represents a distinction between events that are
portrayed as actualised or as actually occurring on the one hand, and the ones
that are portrayed as still within the realm of thought, on the other (Mithun
1995). It is comparable to the past/non-past distinction in the category of tense,
and the perfective-imperfective distinction in the category of aspect. This
comparability derives primarily from the fact that there are several languages in
which the realis-irrealis distinction functions as the most fundamental distinction,
dividing the whole system of verbal forms into two different groups, just as the
past/non-past distinction, and also the perfective-imperfective distinction, do in
several other languages. There is also a correlation between these three types of
distinctions in the sense that past and perfective events tend to be associated with
realis events whereas future and imperfective events tend to be associated with
irrealis events.
We can exemplify the occurrence of the realis-irrealis distinction as the
central one in a language with the help of Chalcatongo Mixtec, described by
Macaulay (1996). Verbs of this language have two distinct stems, called realis
and potential. The former is used to describe actions that are underway at the
time of the speech event, are habitual, or have already been finished at the time
of speaking. This realis stem occurs in progressive, habitual and stative forms as
well. It is also used with the completive suffix ni to denote completed present or
future events. The latter (potential stem), on the other hand, is used to express
future time, imperative, counterfactual, conditional and various other modal
senses. We may regard the latter as the irrealis stem.
The two types of stems (realis and potential) differ (i) segmentally, (ii) by
tone, or (iii) segmentally and by tone; a few of them involve suppletion; about
half the number of verbs, however, have identical forms for these stems. The
following pairs of forms exemplify this distinction and the sentences which
follow them exemplify their usage:
Realis Potential Gloss
kaku kákú ‘be born’
xasú kásu ‘close’
xítú kútú ‘work in the fields’
xátù kuxátú ‘be spicy’
caa caa ‘write’
xí‘i kuú ‘die’
(1) a. rú‘ú kee=rí nduči
I eat ()=1 beans
‘I will eat beans’
b. rú‘ú žee=rí nduči=rí
I eat ()=1 bean=1
‘I am eating / I eat my beans’
c. rú‘ú kútú=rí=nu ba‘á …
I work ()=1= but
‘I was supposed to work, but …’
A basic distinction between realis and irrealis moods is also reported to
occur in the verbal forms of Muna, an Austronesian language, belonging to the
Western Malayo-Polinesian branch (Van der Berg 1989). The most important
difference between the two types of forms is that they take different sets of
subject markers that are prefixed to the verb. In the case of some verbs, there is
also an infix um occurring in the irrealis forms, which distinguishes them from
realis forms. There are several morphophonemic alternations that are connected
with the use of this infix. The following pairs of forms exemplify this realis-
irrealis modal distinction:
CATEGORY OF MOOD 67
Realis Irrealis Gloss (of irrealis)

a-kula a-k-um-ala ‘I will go’
no-horo a-h-um-oro ‘It will fly’
a-gholi ae-gholi ‘I will buy’
de-basa dae-basa ‘We will read’
omo-gharo omo-gharo ‘You will be hungry’
no-lodo nao-lodo ‘He will sleep’
The realis forms can refer to either past or present events. They can also
occur with the futurity suffix ho (this suffix can only occur with realis forms).
Irrealis forms, on the other hand, can refer to the future or can express a wish,
desire or intention. They are also used obligatorily in negative clauses, i.e., in the
presence of negators such as miina ‘not’, miina-ho ‘not yet’ and pa ‘will not’.
The realis-irrealis distinction is very different from the past/non-past
distinction, even though realis forms are generally translated as past (or present)
and irrealis forms as future. In fact, most of the earlier grammars (and some
recent ones too) wrongly describe these modal forms as showing a past-present-
future (or future/non-future) tense distinction. For example, verbal forms of Mao
Naga are described by Giridhar (1994) as showing the distinctions of past, present
and future tenses, but a closer examination reveals that the basic distinction
occurring in these forms is between realis and irrealis moods. We can exemplify
the basic contrast (which involves the use of the irrealis marker le contrasting
with the absence of any marker) with the help of the following sentences:
(2) a. ai idu niyi ni-e
I yesterday you see-
‘I saw you yesterday’
b. ai izo ocü vuta le
I today home go 
‘I will go home today’
(a) Verbs which do not occur with the marker le can indicate past or present
meanings, as shown in the following contexts of their usage:
(i) Verbs expressive of mental or physical states, physical or personality
attributes, and of emotional disposition are ambiguous between past and present
meanings when used without the irrealis marker le (Giridhar 1994: 285).
Example:
(3) pfo zhü-e

he good-
(i) ‘He is good’
(ii) ‘He was good’
(ii) There is a habitual marker we, which can occur with non-stative verbs,
but the habitual meaning can ambiguously be past or present; it does not occur
with the irrealis marker le (Giridhar 1994: 287, 289):
(4) a. pfano pfo pfüyi sokhro-we
he his mother help-
‘He helps his mother’
b. imemüi deni ipre.so to-we
Maos formerly elephant.meat eat-
‘Maos used to eat elephant meat a long time back’
(iii) There is also a progressive or durative marker e, which denotes past or
present progressive meaning (when not accompanied by le); further, when
preceded by ta ‘start’, it indicates the future (imminent) meaning. That is, the
future connotation is obtained even in the absence of the “future” marker le
when the speaker is sure about the occurrence of the event. Examples (Giridhar
1994: 292, 297):
(5) a. pfo vu-e
he come-
(i) ‘He is coming’
(ii) ‘He was coming’
b. ata ho ta-e
we field start-
‘We are about to go to the field’
c. alemo avu ta-e
Alemo meals start-
‘Alemo is about to take meals’
(b) The marker le, on the other hand, generally denotes future meaning, but
in the following usages it has other meanings which appear to indicate that
basically it is an irrealis marker.
(i) It can occur with the affix Ti ‘relevance’ in order to indicate a thwarted
(past) desiderative. Examples (Giridhar 1994: 310):
CATEGORY OF MOOD 69
(7) a. alemono ovo hrü le-Ti-e

Alemo pig buy --
‘Alemo wanted to buy a pig’ (but could not as there was no
money)
b. kaikhono niyi mono le-Ti-e
Kaikho you help --
‘Kaikho wanted to help you (but he wil not be here)’
(ii) There is a suffix abüi denoting individual choice which can occur with
le to denote not only a future choice but also a thwarted past choice. Examples
(Giridhar 1994: 317, 322):
(7) a. ai sho-abüi le
I drink-choice 
‘I choose to drink (if permitted)’
b. ai iduti ta-büi le ana a pfono cü he
I yesterday go-choose  but my father home not
ho-Tie
be-
‘I would have gone yesterday but my father was not at home’
(iii) The marker le can also be used to denote an inference (about a past or
present event) which is based upon non-direct or non-reliable evidence in
contrast to ahi which represents direct and infallible inference (both these occur
with the marker oTi ‘irrelevance’). Giridhar (1994: 335) considers this to be a
distinct inferential marker (in view of the fact that the marker le is considered by
him to be a “future” tense marker), but if we regard all these usages as repre-
senting an irrealis marker, this would not be necessary. Examples (Giridhar
1994: 337):
(8) a. pfono idu rü-oTi le
he yesterday wrote- 
‘He must have written yesterday’
b pfo ico avuo bu le
he now meal take 
‘He must be taking his meal now’
(9) a. pfono idu rü-oTi-ahi
he yesterday write--
‘He must have written yesterday’
b. pfono idu rü-oTi le

he yesterday write- 
‘He has probably written yesterday’
4.2.2 Judgements and evidentials
A distinction is generally made between judgements and evidentials with the

former denoting a speaker’s own assessment about the occurrence of an event
(that is, whether the occurrence is doubtful, probable, definite or certain) and the
latter denoting the basis that the speaker has for claiming that the event has
occurred (or is going to take place). In the case of familiar languages, we may
regard the so-called “indicative” or declarative sentences to be representing the
highest degree of certainty as far as the hierarchy of judgements is concerned,
but rather surprisingly, linguists who write about modal distinctions occurring in
familiar languages like English generally do not consider indicative sentences to
be “modal”.
It is generally claimed that the modal distinctions of judgement are more
numerous in the future tense (or irrealis mood) than in the past or present tense
(or realis mood). On the other hand, evidentials appear to show a greater number
of distinctions in the realis mood (especially in the past tense) than in the irrealis
mood. That is, judgements and evidentials appear to complement one another in
the future/non-future or irrealis-realis dichotomy. However, they are not opposed
to one another, as it is quite possible to combine the two in the same verbal form.
Distinctions of judgement can be exemplified with the help of the following
sentences of Mao Naga (Giridhar 1994: 308) in which the irrealis mood is further
subdivided into (i) doubt, denoted by the marker amolo, (ii) definiteness, denoted
by the marker Ti which is followed by the irrealis suffix le and (iii) certainty,
denoted by the marker li which is also followed by the irrealis marker le.
Examples:
(10) a. pfo ta-amolo-e
he go-may-
‘He may go’
b. pfohi thi-ti le
he die-sure 
‘He will surely die’
CATEGORY OF MOOD 71
c. pfohi thi-li le
he die-certain 
‘He will certainly die’
The definiteness marker differs from the certainty marker in that the former
(Ti) can be negated but the latter cannot. Examples:
d. pfohi thi-Ti le moe
he die-sure  not
‘Surely, he will not die’
According to Barnes (1984), Tuyuca, a Tucanoan language spoken in
Columbia and Brazil, has evidentials as a mendatory feature of the independent
verb. Speakers must indicate the evidence on the basis of which they obtained
the information for producing an utterance in one of the following five distinct
ways by using the relevant suffixes: (i) visually, (ii) through a sense other than
visual, (iii) through evidence of the state or event, (iv) by being told about the
state or event, or (v) by assuming what happened. The sentences given below
exemplify these five evidentiality distinctions; they have the common meaning
‘He played soccer’, but in addition to this, they also have the meanings specified
in front of them, which are denoted by the evidential suffixes (Barnes 1984):
(11) a. Visual díga apé-wi
(I saw him play)
b. Non-visual díga apé-ti
(I heard the game and him)
c. Apparent díga apé-yi
(I have evidence, like his foot-
prints)
d. Second-hand díga apé-yigi (Someone told me)
e. Assumed díga apé-hiyi (It is reasonable to assume that
he played)
The evidential suffixes given above are used in the past tense. There are two
other sets of suffixes, of which one is used in the present tense and the other one
in the future. Barnes notes, however, that in the present tense, the second-hand
evidential does not occur and the apparent evidential occurs rarely (it does not
occur in first person). Further, a speaker cannot use the assumed evidential when
referring to himself in the present tense. Another interesting point, noted by
Barnes, is that the future paradigm is distinct from past and present paradigms
only for assumed evidential, and even this appears to have derived from an
earlier compound construction.
Ladakhi, another Tibeto-Burman language (belonging to the Central Tibetan
group), also makes several distinctions in the epistemic mood, involving both
judgements as well as evidentials. It appears to give greater prominence to
evidentials than to judgements. According to Koshal (1979: 193), verbal bases of
Ladakhi can take any of the following mood suffixes:
Epistemic mood distinctions:
6t reported event
duk, ruk observed event
6r6k experienced event
ok inferred event
cen probable event
6nok generic event
Notice that the first four suffixes are primarily evidentials in nature and only the
last two can be regarded as involving judgement. Examples (Koshal 1979: 193):
(12) a. p6lld6n-ni spe-čh6 sill-6t
Paldan- book- read-
‘Paldan reads a book (a report)’
b. kho-e lč6]-m6 č6d-duk
he- tree- cut-
‘He cuts the tree (direct observation)’
c. kho čh6-6r6k
he go-
‘He goes (speaker’s feeling)’
d. kho-6 zur-mo s6nte duk ši-ok
he- pain- very be die-
‘He will die (because) he is very sick’
e. kho-6 thore ]e 6-čo thuk-cen
he- tomorrow my brother- meet-may
‘He is likely to meet my brother tomorrow’
f. ñi-m6 z6kt6] š6r-ne š6rr-6-nok
sun- daily east- rise-
‘The sun rises daily in the east (generic)’
Ladakhi also distinguishes between different types of inferential statements
by using the suffix thig which is followed by one of the following suffixes,
which specify the type of inference that is being used:
CATEGORY OF MOOD 73
Inference distinctions
r6k inferred from sounds or from habitual occurrences
yot inferred from observations not remembered correctly
so] inferred from unobserved partial or vague knowledge
duk guessed, as for example about events that occurred at a distance and
hence cannot be seen clearly
Examples:
(13) a. dolm6 yo]-thig-r6k
Dolma come--
‘Dolma is coming (a guess made by hearing footsteps, voice,
etc.)’
b khoe ]e k6ne pene khyer-thig-yot
he me from money take--
‘He might have taken money from me’
c. kho i-kh6]pe n6]]6 duk-thik-son
he this-house in live--
‘He might have lived in this house’
d. 6-pumo rdemo yot-thig-duk
that-girl beautiful be--
‘That girl might be beautiful’
There are also two narrative suffixes, kek and tshuk, which can be added to
the reportive form. The latter (tshuk), can also be used with other verbal forms
and has the narrative sense in the case of third person subjects and of surprise
and continuity in the case of second person subjects. In the case of first person
subjects, however, it denotes the speaker’s surprise at his stupidity or foolishness
in attempting to do (or doing) something. Examples (Koshal 1979: 206, 217):
(14) l6m6gunni skurim s6ll-6t-k6k
monks worship offer--
‘Monks offered worship’
(15) khyor6] w6r6n6sie ch6-6t-tshuk
you () varanasi go--
‘So! You are going to Varanasi!’
(16) khoe ši] š6ggin-yot-pin- huk

he wood split-be--
‘He was splitting wood’
(17) ]e ch6] thu]]in-yot- huk
I chang drink-be-
‘I was drinking chang (local alcohol) (without knowing that it was
wrong to do so)’
In addition to the suffixes mentioned above, Ladakhi makes use of a two-
fold modal distinction by using the suffix yin for denoting future (“definite”)
tense and do for denoting the indefinite future tense. The latter suffix can also be
used after the “definite” future form in order to convert it into an indefinite
future (Koshal 1979: 201). Examples:
(18) a. ]6 thore 6b6-6 yige Di-yin
I tomorrow father- letter write-
‘I will write a letter to (my) father tomorrow’
b. kho thore yon-Do
he tomorrow come-may
‘He may come tomorrow’
c. 6-botkh6nbo ew6] yin-Do
that-caller Tshewang be-must
‘That caller must be Tshewang’
d. ne khyor6]]6 yige Di-yin-Do
I you letter write--may
‘I may write a letter to you’
Ladakhi also has different copula verbs for denoting different epistemic
moods. Some of these copula verbs are derivationally related to the modal
suffixes described earlier. The following set of copulas may be noted here
(Koshal 1979: 185–9):
Different copula verbs for different moods:
yin speaker is making a simple statement without any modal implications
yot speaker has a definite knowledge (direct or indirect)
rek speaker is denoting an experience or feeling like pain, or taste
duk speaker is denoting something that he has seen or read
CATEGORY OF MOOD 75
Notice, once again, that the distinctions are primarily evidential in nature.
Examples:
(19) a. ]6 m6g-mi yin
I soldier- be (simple statement)
‘I am a soldier’
b. kho]-]6 pe-ne yot
he- money- be (definite knowledge)
‘he has money’
c. ]6 go-6 zur-mo r6k
I head- pain- be (experience)
‘I have a headache’
d. pu-mo rdemo duk
girl- beautiful be (seen)
‘The girl is beautiful’
Some of these copula verbs can take the past suffix pin, generic suffix nok,
narrative suffix k6k, and future (probable) suffix Do in order to indicate the
relevant additional meaning distinctions. Examples:
(20) a. su yin-Do
who be- (probable)
‘Who must he be?’
b. kho n6kpo yot-pin
he black be-
‘He was black (definite knowledge)’
4.3 Deontic Mood
The difference between epistemic and deontic moods is that the former indicates
the kind of opinion (or knowledge) that a speaker has regarding the actuality of
an event (or the basis for such an opinion or knowledge), whereas the latter indi-
cates the kind of compulsion which makes it possible or necessary for an event
to take place. This compulsion may be internal to one or more of the participants
of the event, or external to them; that is, internal notions like ability, willingness
and desire and external notions like necessity, request and order can be brought
under the deontic mood.
As I had mentioned earlier, there is an interesting correlation between
judgement (epistemic mood) and deontic mood, in that the stronger the compul-
sion for making an event to take place, the more certain that a speaker can be
about the actuality of that event. It is apparently because of this correlation that
there are several languages in which the same set of forms are used ambiguously
either as judgements or as deontic expressions. For example, English may, should
and must can be interpreted either as denoting epistemic notions of possibility,
probability and necessity respectively, or deontic notions of permission, obliga-
tion and requirement respectively, as shown in the following sentences (Palmer
1986: 18):
(21) He may come tomorrow.
(i) ‘Perhaps he will come tomorrow’
(ii) ‘He is permitted to come tomorrow’
(22) The book should be on the shelf.
(i) ‘The book probably is on the shelf’
(ii) ‘The proper place for the book is the shelf’
(23) He must be in his office.
(i) ‘I am certain that he is in his office’
(ii) ‘He is obliged to be in his office’
However, there are several other languages in which the representations of
these two concepts are quite different from one another. In Ladakhi, for example,
there are four different sets of suffixes that are used for denoting different
deontic distinctions; these are quite distinct from the epistemic and evidential
suffixes that I have described in the previous section (4.2.2). Koshal (1979: 228)
gives the following deontic suffixes (called by her as “secondary” modal
suffixes, which include certain additional ones like “completive” which are
aspectual rather than modal) for Ladakhi:
Deontic suffixes:
thub, ñ6n can or be able to do something
gos wish to do something; should or need to do something
ne]dig allowed to do something
n6]čhog allowed to do something
phog compelled (though extremely unwilling) to do something
The following sentences exemplify the use of some of these deontic suffixes
in Ladakhi (Koshal 1979: 227–37):
CATEGORY OF MOOD 77
(24) a. kho6 spečh6 Di-thubb-6t

he book write-can-
‘He can write a book’
b. n66 yigežik Di-goss-6t
I letter write-wish-
(i) ‘I want to write a letter’
(ii) ‘I should write a letter’
c. k6rgil-p6gun led.6ksl6 l6myig-metp6 čh6-n6ndigg-6t
Kargil-people Ladakh permit-without go-allow-
‘People of Kargil are allowed to go to Ladakh without a
permit’
d. n66 č6 co6-phogg-6t
I tea do--
‘I had to prepare tea’
Ladakhi also allows some of these deontic suffixes to occur with epistemic
suffixes described in the previous section. Example:
(25) rigzinl6 ri6 dz6ks-thub-gos-duk
Rigzin mountain climb-can-wish-
‘Rigzin can desire to climb the mountain’
In addition to these deontic suffixes, Ladakhi also has a benefactive form,
which, however, is formally identical with the honorific imperative form. It is
derived by adding the suffix sik to the verbal stem (Koshal 1979). Examples:
(26) a. n66 thop-šik
I get-
‘May I get (something)!’
b. khyor6]]i tshe-ri]-šik
you () life-long-
‘May you have a long life!’
Mao Naga has a set of deontic mood markers, which are also quite distinct
from the epistemic mood markers that I have described in the previous section.
It uses the particle pha (which is added to the participial form of the verb) for
denoting obligation, which is quite different from the marker for certainty,
namely li (see 4.2.2). Examples (Giridhar 1994: 306):
(27) a. ocü korü pha-e

rain rain () ought-
‘It ought to rain’
b. ni koto pha-e
you eat () ought-
‘You ought to eat’
Mao Naga uses the marker büi for denoting the meaning of permission,
which is different from the marker for possibility, namely amolo, noted earlier
(see 4.2.2). The permissive suffix also indicates that the person concerned may
carry out the action if he chooses to. When used with the first person subject, it
provides only this latter meaning. Examples (Giridhar 1994:317):
(28) a. cakho cahrano vu-büi le
Chakho afternoon come-allow 
‘Chakho is permitted to come in the afternoon, if he chooses to’
b. ai ta büi le
I go choose 
‘I choose (and determined) to go’
There is, in addition, a marker for suggestion, namely shie, which is used
only with first person inclusive subjects. Example:
(29) pfoyi da le shie
him beat  let
‘Let us beat him’
The relevance suffix Ti, which denotes a resultant state (as shown in (30a)
below), has the function of indicating the intention of doing something, when
used with the irrealis suffix (30b). Examples:
(30) a. cühi pra-Ti-e
sun rise--
‘The sun has risen’
b. lokho lokha hihi pfo-Ti le
Lokho bag this take- 
‘Lokho intends to take this bag’
There are two abilitative markers, namely losü and lozhü, which differ from
one another in that the former indicates intrinsic ability and the latter extrinsic
ability. The following sentences exemplify this distinction:
CATEGORY OF MOOD 79
(31) a. a napüihi oloso losü-e

my daughter sing able ()-
‘My daughter can sing (i.e. she has the physical or physiologi-
cal ability to sing)’
b. a napüihi oloso lozhü-e
my daughter sing able ()-
‘My daughter can sing (i.e. correctly, effectively, before a huge
audience, etc.)’
4.4 Epistemic moods and interrogatives
As I had mentioned earlier, there is a close affinity between epistemic moods and
interrogatives, in that a speaker generally uses the latter in order to obtain
information that can help him to use a stronger (or more definite) variety of
epistemic mood. However, the two are also quite different from one another and
this difference gets reflected in the fact that languages generally have distinct
systems of representation for interrogatives and epistemic mood distinctions.
There are two main types of interrogative sentences that occur in natural
languages, called polar (or yes–no) questions and content (or wh-) questions. The
former are used for obtaining information regarding a proposition as a whole (i.e.
to find out whether a given proposition is correct or incorrect), whereas the latter
are used for obtaining information regarding a particular constituent (an argu-
ment, an adjective, an adverb, etc.) of a proposition. Examples:
(32) a. Will he come tomorrow?
b. When will he come?
Notice that (32a), a polar question, would be used in order to find out the
correctness of the proposition he will come tomorrow, whereas (32b), a content
question, would be used in order to get information about the identity of the
temporal adverbial (tomorrow) which is not known to the speaker. The wh-word
when is used in (32b) in order to indicate the entity that needs to be identified by
the addressee.
Linguists generally consider the distinction between these two types of
interrogatives to be represented by having a sentential interrogative marker (an
affix, intonation or word order change) in the former case, and by having a wh-
word in the latter case. That is, content questions are generally considered to
have a “question word” (or an “interrogative pronoun”) as the question marker

in them. However, in the case of several Tibeto-Burman languages (and also in
many others) the two types of interrogative sentences are differentiated from one
another by the use of distinct sets of sentential affixes rather than by the presence
of question words. There is therefore a need to regard the so-called interrogative
pronouns occurring in the content questions of these languages as merely “indefi-
nite” pronouns and not as “question” words, as I argue in Bhat (1989).
For example, Sema, a Naga language, has a set of three different particles
that are used for deriving polar questions (namely kesya for questioning persons
of low status, ma for questioning persons of respect, and kema used in neutral
situations). These particles are contrasted with the particle kyá, which is used for
deriving constituent questions. The so-called interrogative pronouns occurring in
these latter sentences have only the function of denoting the entity that needs to
be identified. Examples (Sridhar 1980: 180):
(33) a. noye pa ithi anì kesya
you he know  
‘Do you know him?’
b paye kyùsi anì kyá
he what do () 
‘What is he doing?’
Khezha, another Tibeto-Burman language, also differentiates between polar
questions and content questions by using two distinct sets of suffixes, each
making several modal distinctions. Content questions contain an “interrogative”
pronoun in addition to the question marker, but the pronoun has only the
function of indicating the entity that needs to be identified. Kapfo (1993: 216,
personal communication) lists the following markers that can be used in polar
questions for representing different types of polar questions in Khezha:
díre, you speaker already knows the proposition to be correct and expects
it to be confirmed
ya, nì speaker has reliable information and is asking for confirmation
níe speaker presumes the proposition to be correct but is uncertain
about it
yo speaker is amazed by the proposition and is asking for reaffir-
mation
momí, lè speaker has heard some unusual rumour and is asking for
verification
CATEGORY OF MOOD 81
(34) a. nò merí-e nì-à-díre

you Mary- love--
‘You are in love with Mary, isn’t it?’
b. nò mhech6-à-ya
you work--
‘Are you working?’
c. àpf6 wò-dà-níe
father come--
‘Father has (probably) already come, hasn’t he?’
d. ì-zò lés6-à-momí
your-mother sick--
‘Is your mother ill?’
e. nò-n6 merí-e ni-à-yo
you- Mary- love--
‘So, you are in love with Mary?’ (I can’t believe it!)
In addition to these, the negative marker mo can also be used as a question
marker to denote uncertainty. Example:
(34) f. nò mèlh6-à-mo
you hungry--
‘You seem to be hungry, aren’t you?’
None of these question markers can be used in content questions, i.e.
questions that request for information about a particular entity in a sentence.
Khezha uses an entirely different set of question markers, namely the following,
for deriving different types of content questions (Kapfo, 1993: 211, personal
communication):
ro used in any situation
lé used by adults while questioning children
šè used for asking the addressee to repeat what he said earlier
dí, día, dèi used to ask the addressee to choose from one of the alternatives
yó rhetorical question where the speaker does not expect the
addressee to give an answer
yá, lá used when the speaker is not bothered to have the answer
Examples:
(36) a. pù-n6 ditsh6ké wo ro

he- when come 
‘When did he come?’
b. pù-n6 ditsh6ké wo lé
‘When did he come?’ (used with children)
c. pù-n6 ditsh6ké wo šè
‘When did he come? (repeat)’
d. pù-n6 ditsh6ké wo dèi
‘When did he come?’ (choose the alternative)
e. pù-n6 ditsh6ké wo yó
‘When did he come?’ (I don’t care)
None of these question markers, with the exception of šè ‘repeat’, can occur
in polar questions. The difference between the two sets of question markers is
apparently that the former requests for confirmation of a statement (the suffix or
particle is attached to a statement), whereas the latter asks for information about
a specific unknown entity. Only the notion of repeating a statement can apparent-
ly be associated with either of these two types of questions. Example:
(36) pù-n6 wo šè
he- come 
‘Is it true that he came?’ (Say that again)
The question (36) may be compared with (35c) given above.
4.5 Deontic moods and imperatives
We may regard imperatives as being closely related to deontic moods in the

sense that the imperatives, like some of the deontic moods, represent the external
compulsion which forces an event to take place (see 4.3). The stronger the
external compulsion, the more likely it is that the event would take place.
However, the two are also different from one another in that the imperatives are
used directly by a speaker as a speech act in order to get something done by the
addressee, whereas the deontic moods are used only indirectly for a similar
CATEGORY OF MOOD 83
purpose. They are only statements and not commands. Both these characteristics
of imperatives find representations in their formal expression in natural languag-
es. The fact that they are partly similar to deontic moods is apparently responsi-
ble for the occurrence of both of them in the same system of affixes (i.e. in the
same paradigm) in some languages, whereas the fact that they are partly different
from deontic moods is responsible for the occurrence of the two as distinct
systems in other languages. The latter characteristic is shown by languages in
which mood is a prominent verbal category.
In the case of Indo-Aryan languages, for example, imperative is part of the
subjunctive paradigm. The second person singular form has a zero ending (except
in Sindhi) but the plural is the same as the subjunctive. The paradigm also has
third person forms, which denote deontic (permissive) meaning. There is also a
first person (singular or plural) form in some of them (like Marathi u and
Konkani yã). Some of the languages also make a distinction between present and
future imperatives (Masica 1991: 476). The following Marathi sentences exempli-
fy some of these usages (Berntsen and Nimbkar 1982: 86):
(37) a. h6Lu bol
slowly speak (1)
‘Speak slowly!’
b. h6Lu bol-a
slowly speak-2
‘Speak slowly!’
(38) a tula b6re˜ vaT-o
you () good feel- (3)
‘May you feel well!’
(39) amhi tikDe ja-u ka
we there go- (1) 
‘Should we go there?’
Dravidian languages also include deontic mood markers and imperative
markers in a single paradigm. In Tulu, for example, the suffixes la ‘singular’ and
le ‘plural’ are attached directly to the verb to derive singular and plural (or
honorific singular) imperative forms respectively, whereas the deontic notion of
permission is expressed by adding the suffix aDG to the verb. This latter form is
used both in first and third persons. Examples:
(40) a. alenG iñci barræ pan-le

her here come () tell- (2)
‘Ask her to come here!’
b. a˜lG iñci bar-aDG
she here come-
‘Let her come here!’
On the other hand, the suffix l occurs in the third (or first) person permis-
sive marker (ali) in certain other Dravidian languages like Kannada. Example:
(41) avanu mane-ge bar-ali
he home- come-
‘Let him come home!’
Tibeto-Burman languages are different from these in that they appear to
treat deontic moods and imperatives as belonging to distinct systems of suffixes
or particles. For example, as I have pointed out earlier (4.3), Mao Naga has the
following set of suffixes for denoting different deontic mood distinctions:
pha ought to
büi allowed to
shie let us
ti wish to
losü able to (intrinsic)
lozhü able to (extrinsic)
In contrast to this set of deontic mood markers, Mao Naga has an entirely
different set of imperative markers (see Giridhar 1994: 347 for details).
First of all, there is a distinction, in Mao Naga, between (i) movement or
transfer towards the speaker and (ii) movement or transfer towards some other
person. The imperative marker to be used in the former case is ka and the one
to be used in the latter case is ha. Examples:
(42) a. larübvüsü pi-ka
book give-
‘Give the book (to me)!’
b. larübvüsü pi-ha
book give-
‘Give the book (to someone other than me)!’
The imperative marker ka differs from another marker hi in being less
CATEGORY OF MOOD 85
brusque and also in being immediate; the former (and also the marker ha) cannot
occur with temporal adverbials denoting non-immediate time, whereas the latter
can. Examples:
(43) a. *capüi oca so pi-ka
later tea do give-
‘Do me tea later!’
b. *pfoyi sodu larübvü pi-ha
him tomorrow book give-
‘Give him the book tomorrow!’
c. cahrano oca sa-hi
afternoon tea do-
‘Do (some) tea in the afternoon’
However, when used without a temporal adverbial, hi also indicates an
action that is to be carried out immediately. Examples:
(44) a. oca so-hi
tea do-
‘Prepare tea (here and now)!’
b. saba bo-hi
shawl wear-
‘Wear the shawl (here and now)!’
In contrast to the marker hi, Mao Naga uses two other markers, namely o
which is neutral concerning politeness, and ō (with mid tone) which is more
polite and is more in the nature of a suggestion (Giridhar 1994: 348). Examples:
(45) a. ohi hru-o
eye open-
‘Open the eyes!’
b. ca sho-ō
tea drink-
‘Please drink tea!’
The two markers, hi and o can be combined together to form a blunt order.
Example:
(46) ni larü hru-hi-o
you book open--
‘Open your books!’
It can be made polite, however, by adding the verb pi ‘give’ to it.

(47) ovo hiniahi so pi-hio
work this do give--
‘Please do this work!’
In contrast to the preceding imperative markers, Mao Naga has the marker
lo which indicates that the action to be performed typically lasts a considerable
amount of time, or is a continual or habitual one. Examples:
(48) a. ni sa mani-lo
you cloth show-
‘Keep showing your cloths!’
b. ni sa mani-o
you cloth show-
‘show your cloths (once)!’
(49) a. cükhu khu-lo
door close-
‘Keep the door closed; keep closing it whenever you go out!’
b cükhu khu-o
door close-
‘Close the door (once)!’
(50) cu-lo
run-
‘Keep running!’
The marker lo contrasts with another marker ló in that the latter indicates
that the benefit is for the doer (addressee) only, whereas the former indicates that
the benefit is for both (the speaker as well as the addressee). Examples:
(51) a. so-ló
do-
‘Do it! (the benefit is yours)’
b. so-lo
do-
‘Do it! (the benefit is ours (inclusive))’
We may summarize the system of Mao Naga imperative markers as follows:

CATEGORY OF MOOD 87
I ka movement or transfer towards the speaker; immediate

ha movement or transfer towards some other person; immediate
hi can be non-immediate; brusque
II o neutral regarding politeness
ō polite suggestion
hi-o blunt order
pi-hi-o polite request
III lo lasting or habitual action; action for mutual benefit
ló action to be carried out for one’s own benefit
4.6 Use of modal adverbials
We may regard modal adverbials either (i) as replicating the modal information
that is provided by the mood markers or (ii) as providing additional information
regarding the modal characteristics of the verb. They may replicate mood by
establishing a parallel structure, but they can also express additional distinctions
and also additional parameters. In the case of languages in which the modal
category is not very developed (or grammaticalized) in the verbal system,
distinctions of mood will have to be expressed primarily by these modal adverbs.
The contrast (or complementarity) and conflict between moods and modal
adverbs, however, has not been studied in the way in which the contrast between
tense and temporal adverbs has been studied, and hence one can obtain very little
cross-linguistic information about this topic. It clearly needs to be studied
carefully in individual languages.
P 
A Typological Study
C 5
Basis of the Typology
5.1 Introduction
The foregoing descriptive study of tense, aspect and mood allows us to arrive at
an interesting typological observation, namely that languages generally do not
give equal prominence to all these three categories. Instead, they select one of
them as the basic category and express distinctions connected with it in great
detail; they represent the other two categories in lesser detail and further, they
use peripheral systems like the use of auxiliaries, or other indirect means, for
representing these latter categories. Because of this constraint, we had to use
different sets of languages while describing the three categories in the first part
of this monograph, namely Kannada, Tulu, Tamil, Quechua, English, Mishmi,
etc. for tense, Hindi, Marathi, Gujarati, Mundari, Kiowa, Hmong, Sango, etc. for
aspect and Mixtec, Mao Naga, Khezha, Ladakhi, Manam, Tuyuca, etc. for Mood.
This interesting tendency of languages to give prominence to one of the
three verbal categories can form the basis of a typological classification,
especially because in addition to choosing one of the categories as the most
prominent one, languages appear to represent concepts or distinctions that belong
to the other two categories in terms of their chosen category. For example, the
notion of past is represented as a facet (or variety) of realis mood by languages
that have chosen mood as the prominent category whereas languages that have
chosen aspect as the prominent category represent it as a facet of perfective
aspect. Perfect is viewed as involving a combination of realis and irrealis moods
in the former case, whereas in the latter case it is viewed as involving a combi-
nation of perfective and imperfective. A language that has chosen tense as the
prominent category, on the other hand, views it as a combination of past and
non-past or present (i.e. as a past event that has present relevance). In order to
bring out these and other similar interesting differences that co-occur with the
relative prominence that languages attach to different verbal categories, it would
be helpful to classify languages into tense-prominent, aspect-prominent and

mood-prominent language types.
Notice, however, that the typological distinction involved here is basically
an idealised one. Natural languages do not fall into sharply differentiated
language types. There are different degrees of prominence that these categories
receive in different languages; further, in the case of some languages, two or
more of these categories may receive equal prominence, and hence we would not
be able to select the most prominent category in them. Languages might also
select some other verbal category, such as, for example, the one that makes use
of the distinction between deictic and non-deictic categories, and that cuts across
the three above-mentioned categories, as the most prominent one. In spite of
these drawbacks, however, I believe that the above-mentioned typology would be
helpful in arriving at a better understanding of natural languages.
Linguists have found the postulation of idealised languages in this fashion
to be extremely helpful in establishing meaningful correlations among cross-
linguistic variations. The earliest idealisation of this type that has been highly
successful in linguistics is the one between the so-called synthetic, agglutinating
and isolating language types. Greenberg’s (1966) idealisation of word-order types
(into VSO, SVO and SOV), which was later on reduced to VO and OV types
(Lehmann 1974), has been helpful in establishing several correlations among
cross-linguistic variations. The distinction between configurational and non-con-
figurational languages (Chomsky 1981, Hale 1983) or the one between ergative
and accusative languages (Dixon 1979, Plank 1979) also involves the application
of a successful idealisation in a similar fashion.
I have put forth in Bhat (1994) an idealised distinction between languages
in which adjectives form a distinct category on the one hand, and the ones in
which they are identified with nouns, verbs or both on the other. This typological
differentiation has formed the basis for the correlation of several interesting
cross-linguistic variations for which it also provides motivated explanations. The
present typological distinction between aspect-prominent, tense-prominent and
mood-prominent languages may therefore be considered as an additional instance
of idealisation that is helpful in correlating yet another set of cross-linguistic
variations and also, hopefully, in providing explanations to such variations.
BASIS OF THE TYPOLOGY 93
5.2 Basis of tense-aspect-mood variation
The variations that have been observed among languages concerning the
representation of tense, aspect and mood derive primarily from the fact that the
three categories are closely interconnected. Both tense as well as aspect denote
temporal notions, with tense indicating the position of an event on a linear time-
scale in relation to a reference point, (i.e. whether the event occurs before, simul-
taneously or after that reference point), and aspect indicating “the internal tem-
poral structure” of an event (Comrie 1976: 6) i.e. whether the event is completed
or continuing, beginning or progressing, semelfactive (occurring once) or
iterative (or habitual), etc. The two differ from one another in that tense relates
an event with an external reference point whereas aspect provides an internal
view. The latter does not relate the event to any reference point as such. We can
also describe tense as involving a distinction in the time that contains the event
and aspect as involving a distinction in the time that is contained in the event
(Van Valin 1975: 133). However, the two are also interconnected as both of them
deal with the “temporal structure” of the event. This point gets reflected in facts
such as, for example, that a completed event (perfective aspect) tends to be past
whereas a continuing event (imperfective aspect) tends to be present or future
(see Comrie 1976, 1985; Givón 1984; Chung and Timberlake 1985).
Similarly, tense and mood are quite distinct from one another, but are still
interconnected. Mood indicates the reality of an event, i.e. the fact as to whether
the event’s occurrence is a reality or only a possibility. It also refers to the kind
of evidence that can be adduced in support of the claim that it occurred (or is
going to occur). However, its relatedness with tense is shown by the fact that
events which were observed (in the past) or the ones which are being observed
(in the present) tend to be associated with realis mood, whereas the ones which
were not observed (primarily because they are yet to take place — i.e. the future
events) tend to be associated with irrealis mood (see Palmer 1986:208).
It is apparently this interconnectedness of tense, aspect and mood which
makes it possible for some languages to choose one of them as the primary
notion of their verbal system. The notion of past tense and realis mood can be
denoted indirectly by a form which represents primarily the aspectual notion of
completion (and vice versa), whereas the notion of future tense or irrealis mood
can be denoted by a form whose primary denotation is the aspectual notion of
non-completion (and vice versa). Similarly, the temporal notion of simultaneity
can have the aspectual notion of continuity as its implication or vice versa. It is
only when these notions need to be combined together in different ways, as for
example, when past has to be combined with non-completion (or irrealis), future
with completion (or realis), or past with continuity, that the speakers of a
language find it necessary to provide distinct representations for these categories.
5.3 Possibility of using alternative categories
The situation, however, has been made somewhat more complex by the fact that
there are other related notions which cut across this aspect-tense-mood continu-
um, or add other dimensions of variability to it. For example, the notion of
viewpoint cuts across the whole continuum. It divides the notion of aspect into
aktionsart and aspect, with the former providing a set of distinctions among
verbal bases and the latter among their inflectional forms. The former is con-
sidered to be objective, and the latter subjective (see Platzack 1979: 39; Brinton
1988: 3). Similarly, viewpoint divides tense into non-deictic tense (i.e. having the
point of time of some other event as the reference point) and deictic tense (i.e.
having the point of time of the utterance or of the speaker as the reference point).
We may regard the former as “objective” and the latter “subjective”. Viewpoint
also divides mood into evidentials (fact-oriented) and judgements (speaker-
oriented). The point to be noted here is that it is quite possible for this notion of
viewpoint to assume the prominent position in the verbal system of a given
language, and to give rise, thereby, to a different set of correlatable characteristics.
Similarly, it is also possible for the spatial notions of location and direc-
tionality to assume greater prominence than those of tense, aspect and mood.
There are languages like Toba (Klein 1979), for example, in which notions like
“coming into view” and “going out of view” give rise, metaphorically, to
temporal notions of non-past and past respectively, whereas notions like “in
view” and “out of view” give rise to temporal notions of present and non-present
respectively. Other languages have also been reported to provide extended
temporal connotations to spatial markers (Palmer 1986). Languages that em-
phasise these spatial notions and derive temporal notions from them in the
formation of verbal categories can be expected to manifest an entirely different
set of correlatable characteristics.
I propose to set aside these latter possibilities in this study and concentrate
upon the characterisation of the three idealised language types mentioned above,
namely aspect-prominent, tense-prominent and mood-prominent, in order to make
the study manageable and therefore more explicit. I am assuming that the
possible occurrence of these latter type of languages would only give rise to
additional language types as far as verbal categories are concerned, and that they
would not materially affect the generalizations that I make regarding the three
above-mentioned language types.
5.4 Criteria for prominence
The prominence that a category receives in a given language can be determined

on the basis of several factors such as degree of grammaticalization, obligator-
iness, systematicity (or paradigmatization) and pervasiveness (see Newman
1954: 83, Lehmann 1985). These factors are interconnected in the sense that the
category (tense, aspect or mood) which is the most grammaticalized in a given
language would turn out to be the most obligatory, systematic and pervasive, and
therefore the most prominent in that language. It is quite possible, however, to
show that these four factors are to a certain extent independent of one another.
The crucial distinction that underlies the notion of grammaticalization is the
one between lexical elements (“contentives”) and grammatical elements (“func-
tion words” or affixes). Several linguists have pointed out the usefulness of
differentiating between these two types of elements in natural languages (see, for
example, Sapir 1921: 25; Carlson 1983: 70; Givón 1984: 48). Contentives are
independent entities that form different “parts of speech” like nouns, verbs,
adjectives and adverbs, whereas function words are dependent entities that relate
contentives with one another (case markers and agreement markers), or enhance
and modify their connotation and of the sentence as a whole (number and
gender markers, derivational affixes, tense-aspect-mood markers and other
inflectional affixes).
This distinction between lexical and grammatical elements (or between
contentives and function words), however, is not sharp and clear-cut. There is a
gradation, with some lexical items being more contentive than others, and some
function words being more grammatical than others. For example, prepositions
or postpositions are generally regarded as forming a part-of-speech (i.e. as
contentives) but they are clearly more like case markers than independent nouns
and verbs. That is, the latter (nouns and verbs) are more contentive than preposi-
tions. Similarly, inflectional affixes are considered to be more grammatical than
derivational affixes (see Kurylowicz 1964: 36), and even among the markers of
verbal categories, affixes are considered to be more grammatical than particles

or clitics.
Grammaticalized notions are clearly more important for the grammar of a
given language than notions that only get represented by isolated lexical items.
While differentiating between inflectional elements and derivational elements,
Kurylowicz (1964: 36) points out that the latter are farther removed from
sentence-structure and its fundamental oppositions than the former, and that the
latter represent marginal levels of the linguistic system; they tend to dissolve into
isolated lexical phenomena. Further, lexical elements that get grammaticalized
(i.e. changed into inflectional elements) lose their specificity and become
generalised, as they need to apply to a whole class of lexical items. And as they
become more and more grammaticalized, there is gradual erosion of their lexical
meaning into abstract grammatical meaning (Bybee 1985: 16, 137). This also
increases their prominence in the grammar.
Lehmann (1985) considers the notion of obligatoriness and also that of
paradigmatization to be concomitant notions of grammaticalization. That is,
grammaticalized concepts tend to be obligatory and get organised into paradigms;
their representations get systematised in the sense that the various sub-compo-
nents get proper representation in the form of a paradigm. Non-grammatical
concepts, on the other hand, do not generally get interrelated in this fashion, even
though we do find some amount of systematicity among them, as in the case of
kinship terms or terms for body parts. Notice that concepts that are expressed by
obligatory forms and classes carry more weight than those contained in optional
categories, as pointed out by Newman (1954). In the former case, even the non-
use of a marker would be meaningful (called “zero affix” by some linguists),
whereas in the latter case, the markers would generally have the function of
specifying the relevant meaning; their non-use would not necessarily imply that
the meaning is absent.
Pervasiveness is another criterion that can be used while establishing the
prominent category in a given language. Concepts that are restricted to a small
area in the grammar are less prominent than the ones that have scope over a
larger area. This is also the crucial difference between grammaticalized concepts
and lexicalized concepts, as pointed out above. This notion of pervasiveness,
however, leads us to certain conflicting situations. In the case of some languages,
we find the most prominent category being extended from the verbal to other
systems like the nominal one as well. For example, the modal category of the
verbal system extends to the system of nominal case marking in Kayardild, an
Australian language (Evans 1995). Modal distinctions like actual, potential and
hortative are expressed in this language not only by affixes occurring in the verb,
but also by case markers; the two show agreement to a certain extent, but the
case markers can also express modal notions on their own (see 6.4.4).
We may contrast this “extension” of the most prominent verbal category
from the verbal system to non-verbal systems, with a situation in which a
language relegates a non-prominent verbal category distinction to non-verbal
contexts. For example, some of the mood-prominent and aspect-prominent
languages have been reported to relegate the expression of tense distinctions to
nominal or adverbial systems. On the other hand, some of the tense-prominent
languages have been reported to relegate the expression of modal and aspectual
distinctions to non-verbal areas. In Finnish, for example, the perfective-imper-
fective distinction is expressed by the accusative-dative case distinctions, whereas
in Kannada, the modal distinction between internal and external compulsions (for
carrying out an event) is expressed by the nominative-dative case distinction (see
7.6.1). These alternative ways of encoding, unlike the “extensions” mentioned
earlier, represent the non-prominence of the relevant categories.
By applying these criteria to individual languages, we would be able to
classify several of them into one or the other of the three language types
mentioned above (i.e. as aspect-prominent, tense-prominent and mood-promi-
nent). However, in view of the fact that all these criteria involve gradations of
one kind or another, it would be impossible to have sharp and clear-cut divisions
between these classes. Languages would form gradations under each of these
three types, with some of them being closer to the idealised language (in each
case) than others. Further, there would also be borderline cases in which two (or
all the three) categories might appear to be of equal prominence, and in such
cases, we would not be able to determine the actual language type to which the
language under consideration belongs. That is, the classification is not expected
to group, all the languages of the world, exhaustively, into one or the other of
these three language types.
5.5 Nature of generalizations
Another drawback of typological studies, which has been frequently commented

upon, is that they only allow us to establish general tendencies of language and
not rules or formulas that can be stated in very specific terms. Since the studies
are based upon the grammars of several different languages, the data that they
make use of are also not very detailed and sometimes not very reliable either. It
appears to me, however, that this drawback is unavoidable and reflects the very
nature of typological generalizations.
We may compare the distinction between typological studies of the above
type and in-depth studies of individual languages with the distinction between
areal or satellite pictures of a countryside and an architect’s drawings of a town
or a dam. A satellite picture would only show patches of colour and vague lines
and curves that an expert can interpret as indicating the location of a possible
earthquake or deposits of mineral wealth, whereas an architect’s drawings would
show the locations of various buildings, parks, canals, etc. in very precise terms.
It would be a mistake, however, to discard the former merely on the basis
of the fact that they are not as precise and specific as the latter. The two
complement one another, with the satellite pictures giving a warning to the
builder of dams so that he can avoid certain cites as possible disaster areas.
Typological studies and in-depth studies of language can also complement one
another in a similar fashion. The general tendencies that one can perceive
through a comparison of hundreds of different languages can be helpful in
avoiding certain conclusions and in raising certain questions that might not have
been raised otherwise while carrying out in-depth studies of individual languages.
The classification of languages into tense-prominent, aspect-prominent and
mood-prominent language types helps us to establish several interesting tenden-
cies, as I point out in the following chapters. We are not in a position to state
these tendencies in very precise terms at this moment because we do not possess
sufficient data on the languages under consideration. But the fact, however, is
that even when more data are made available on these languages, it is doubtful
that the statements can be changed into very precise rules or formulas; they may
continue to be tendencies only, with more convincing data to support them.
5.6 Bias in grammars
The relevance of this typological classification for our understanding of the

nature of different languages can be perceived even in the very process of
carrying out this study. Notice that much of the grammatical information that we
will be using as the basis of this study has been elicited and described with the
help of one single language, namely English. Since this language is a tense-
prominent language, and since, as I have suggested earlier, tense-prominent

languages view concepts belonging to other verbal categories in terms of the
category of tense, much of our data on aspect and mood would be biased by a
temporal point of view. It is something like trying to understand the colour of
various objects around us while looking at them through a red-coloured glass.
The effect of the medium can sometimes cause a lot of difficulties. For
example, English forces us to select a past or non-past (present) verbal form
whenever we make a verbal statement. We have already experienced the effect
of this obligatoriness of tense while describing the categories of aspect and mood
(see 5.2). We had to be satisfied either by a statement in past tense (“aspect
indicates how an event occurred”) or one in which a statement in past tense is
combined with one in nonpast tense (“realis denotes events that were observed
or are being observed”), but neither of these is satisfactory as it cannot indicate
the fact that aspect and mood are free of the tense distinction. We frequently
encounter grammarians who indicate this problem by pointing out that the
English translations that they provide are misleading on account of this obligator-
iness of tense (see, for example, Kellog 1938, Foster 1985, Li 1991, Refsing
1994), but since descriptions written in English will have to depend upon English
for translations, this problem cannot be avoided.
In the case of our present study, on the other hand, we need to consider this
problem more seriously because it affects the very basis of our study. For
example, is the so-called inseparability of tense and aspect a result of this
dependence of our descriptions of aspect upon a tense-prominent language? I
suspect that this to be the case to a greater extent than is generally conceded.
Similarly, our inability to consider mood as a verbal category that can be as
prominent as tense or aspect (see Bybee, Perkins and Pagliuca 1994: 239) is
probably a result of our dependence, for the description of mood-prominent
languages, upon a tense-prominent language. We may not be able to avoid this
dependence, or the interference that results from it, but we can at least try to
discount for it. One way of doing this, I think, is to give less importance to
similarities that have been noticed between English and the language that has
been described in it, and to emphasise differences that have been recorded.
Another point is that we must be aware of the fact that distinctions occurring in
the translations of forms or sentences do not necessarily indicate distinctions in
the languages that are being translated. For example, when a grammarian writes
that a particular verbal form can be used to denote past as well as present tense,
the language may, in the actual fact, be leaving that distinction unspecified. It
may only be the translator who is forced by his own language to express the
meaning either as past or as present.
There are several instances in which a language that was described by a
series of grammarians as showing a primary tense distinction, like past, present
and future, has been shown to be actually making a primary aspect or mood
distinction. This is the case, for example, of several Indo-Aryan languages like
Hindi and Marathi, which are described as showing a primary distinction
between past and present (or non-past). A careful examination of the use of
relevant verbal forms in these languages has indicated that the primary distinc-
tion is one of aspect between perfective and imperfective rather than one of tense
(see 3.2 and 6.3.1). Li (1991) points out that several scholars have described
Hmong as having a past-present-future tense distinction, but a careful examina-
tion has revealed that the distinction concerned is actually between perfective and
imperfective (progressive). Burmese has been described as showing a future/non-
future tense distinction, but Comrie (1985) has shown that the distinction is
actually between realis and irrealis (moods). Foster (1985) points out that in the
case of Northern Iroquoian languages, modal distinctions have been wrongly
described as tense distinctions. Similarly, DeChicchis (1996) argues that in
Q’eqchi’, the primary distinctions are modal and not temporal as described by
earlier scholars (see 6.4 below).
What is interesting to note, in these cases, is that in the majority of such re-
interpretations, an earlier tense-based description had to be rewritten as aspect-
based or mood-based description. That is, the need to change an earlier descrip-
tion has occurred in almost all cases as a need to remove the bias that has
resulted from our use of a tense-prominent language as the language of elicita-
tion and description.
5.7 Need for diachronic considerations
When we compare languages from a typological point of view, we generally tend

to regard them as synchronic entities rather than as entities that are involved in
a continuous process of change. We therefore fail to realise that some of the
characteristics that they show are correlatable with some of their earlier charac-
teristics which they have lost as a result of this diachronic process. That is, a
comparison of languages merely from a synchronic point of view would show
some of them to be exceptions to a given generalization or tendency, but a
diachronic study would show these to be resulting from the fact that the languag-
es concerned are in a transitional stage. For example, if an aspect-prominent
language is in the process of changing into a tense-prominent one, it might show
characteristics that conflict with generalizations about both aspect-prominent as
well as tense-prominent languages. We can expect it to lose these conflicting
characteristics once it has gone out of that particular transitional stage. The fact
that we are assigning these characteristics to an idealised language allows us to
set aside such transitional stages of languages.
Consider, for example, the case of Manipuri, a Tibeto-Burman language,
which appears to be in the process of changing from a mood-prominent state to
a tense-prominent one; being a mood-prominent language, it earlier had a large
class of state verbs, and its adjectives were indistinguishable from verbs. It now
has a basic future/non-future tense distinction. The use of verbal forms showing
this distinction does not appear to make it possible to regard it as having a realis-
irrealis mood distinction (see Bhat and Ningomba 1997). However, the language
continues to have a distinct class of state verbs that includes adjectives. Will this
language change further and become a typical tense-prominent language? Or will
it remain as an “exceptional” case? We cannot answer this question, but the fact
that some of its present characteristics derive from its being a mood-prominent
language earlier can be regarded as an explanation for its being an exception to
our generalisation.
5.8 Correlatable characteristics
I will be assuming, in the following chapter, that the prominence of tense, aspect
and mood respectively are the defining characteristics of tense-prominent, aspect-
prominent and mood-prominent language types. I will try to show how languages
can be assigned to one or the other of these three language types by examining
the relative prominence that one of these categories receive in their grammars.
This would be accomplished by examining the four main criteria mentioned above,
namely grammaticalization, obligatoriness, systematicity and pervasiveness.
I will be examining other characteristics that appear to be correlatable with
this defining characteristic of the present typology in the next (seventh) chapter.
Some of these are clearly derivable from the defining characteristic, such as the
following:
(i) The tendency of languages to retain distinctions of the prominent category
to a greater extent than those of other categories when verbs are used in
functions other than that of predication (7.2).
(ii) The tendency to view distinctions belonging to other categories like perfect,
future, habitual and negation from the point of view of the prominent
category (7.7).
(iii) The tendency to regard the prominent category as the most relevant one for
the verb and to encode it as part of the verbal inflection. Also to regard
other category distinctions as less relevant to the verb and to encode them
through auxiliaries or even through more indirect means like distinctions in
case markers (7.6).
Other correlatable characteristics, on the other hand, are not so directly
connected with the defining characteristic, but still, we can see that they are
related. These include the following:
(iv) The tendency of tense-prominent languages to use non-verbal encoding for
adjectival predicates, and further, to have a very restricted and irregular set
of state verbs (7.4).
(v) The tendency of mood-prominent languages to show a pronominal ergative
split and of tense- and aspect-prominent languages to show a temporal or
aspectual ergative split (7.3).
(vi) The tendency of languages to use the prominent category (past, perfective
or realis) for foregrounding sequential events in a narration (7.8).
(vii) The tendency of tense-prominent languages to change perfect and progres-
sive forms into past and present tense forms, and of aspect-prominent
languages to change the same into perfective and imperfective aspect forms
respectively through grammaticalization (7.9).
I believe that these correlatable characteristics provide sufficient justification
for the postulation of the typology under consideration.
C 6
Classification of Languages
6.1 Introduction
As I have pointed out in the previous chapter, the prominence that languages
give to one or the other of the three major verbal categories allows us to classify
them into three different idealized types, namely tense-prominent, aspect-
prominent and mood-prominent. This classification, however, is not exhaustive,
as it does not cover all the languages of the world. It is quite possible for some
languages to remain outside this classification either because they give equal
prominence to two or more of these categories, or because they select some other
verbal category, such as location or viewpoint as the most prominent category.
I propose to leave aside these latter types of languages from my study because
I do not possess sufficient information to establish additional language types. My
assumption is that the occurrence of these additional language types would not
materially affect the generalizations that I establish on the basis of the study of
the three language types mentioned above.
It may be noted further that the generalizations and tendencies that I
postulate here are meant for characterising the three idealised languages; they can
be expected to occur in actual languages only to the extent that the languages
resemble one or the other of those idealised languages. That is, it is not neces-
sary for any given language to show all the characteristics that are being
assigned to one of the three idealised languages that are being established here.
However, the postulation itself of these generalizations and tendencies is based
upon data from actual languages.
6.2 Tense-prominent languages
The defining characteristic of languages belonging to the tense-prominent

language type is the fact that they give greater prominence to tense than to
aspect or mood in their grammatical system as a whole. Since the notion of
prominence is a relative term, we can expect these languages to form a grada-
tion, with some being closer to the idealised language than others. The criteria
that we use for determining the relative prominence of tense in these languages
are also such that they allow us to establish similar gradations in each case. For
example, tense would be grammaticalized to a greater extent in some of these
languages than in others, but in all of them it would be grammaticalized (if at
all) to a greater degree than aspect or mood. Similarly, the criteria of obligatori-
ness, systematicity and pervasiveness would also allow us to establish gradations
among these languages.
There are several languages in which tense is more prominent than aspect
or mood, as shown by the greater degree of grammaticalization, obligatoriness,
systematicity and pervasiveness that is provided by these languages to tense, as
compared to aspect or mood. However, as I have pointed out in the previous
chapter (see 5.7), descriptions of languages appear to show more number of
languages to be tense-prominent than is actually the case because of the fact that,
in several instances, aspectual and modal distinctions have been described as
tense distinctions. I will therefore only try to establish, in the following sections,
the possibility of languages being tense-prominent by showing how a particular
set of languages (namely the ones belonging to the Dravidian family, which are
the most familiar to me) make tense more prominent than aspect or mood in
their grammaticalization, obligatoriness, systematicity and pervasiveness. The
actual set of languages that belong to this type (or even to the other two types)
can only be determined after a more systematic study of the languages concerned.
6.2.1 Grammaticalization of tense
All the languages of the Dravidian family manifest grammaticalization of tense

distinctions to a very high degree. This grammaticalization is considered to have
taken place in the Proto-Dravidian itself. Subrahmanyam (1971: 237) reconstructs
a two-fold past-non-past tense distinction for the proto-language of this family,
with the morphs *nt, *tt, *kk and *i representing the past tense, and the morphs
*pp, *kk, *tt, *um and *n representing the non-past tense. (Even though two of
the morphs representing these tenses, *tt and *kk, are identical in form, the actual
verbal forms that contain them are non-identical.) The variants in each case result
partly due to phonological conditionings and partly due to the occurrence of
CLASSIFICATION OF LANGUAGES 105
different suffix forms in different subgroups. These suffixes are functional

elements to the highest degree and are purely inflectional in their usage. No
lexical source for the derivation of these suffixes has been suggested. They are
also unconstrained by the type of verbal bases with which they occur (except for
the morphophonemic changes that accompany such usages); in fact, the verbs in
Dravidian languages are defined by their ability to occur with these tense markers.
As far as the past tense marker is concerned, *kk is found primarily in the
North Dravidian languages like Kurukh and Malto. It is possible that the
remaining four variants are phonologically conditioned forms of a single past
suffix *int, but the exact nature of this conditioning is not yet clear. The suffix
occurs as *t after monosyllabic roots (with several alternants like k, c, t, t, T, etc.
in the daughter languages which clearly result from assimilation) and *nt or *tt
after other types of roots. In some of the South Dravidian languages like Tamil,
this *nt/tt distinction in the past tense suffix represents an affective/effective
distinction (Paramasivam 1979), which might be an innovation. The marker *i
occurs primarily in verbal participles, but certain classes of verbs contain it in
finite forms as well in some daughter languages. It also occurs jointly with the
dental or nasal morph in some contexts. Much of the complexity that underlies
the formation of finite and non-finite verbs in Dravidian languages, either in the
form of the suffix or in that of the verbal base, is concerned with the use of this
past tense suffix.
Aspect and mood markers that occur in some of the modern Dravidian
languages derive from earlier verbal forms that are used as auxiliaries or vectors;
they have been grammaticalized rather irregularly and to different degrees in
different daughter languages. It is possible, however, that the basic construction
involved in their formation, namely the use of a past (prior) converb for deriving
aspectual constructions and an infinitive (posterior converb) for deriving modal
constructions, was prevalent in the Proto-Dravidian stage itself (see Steever
1983). Thus the fully grammaticalized verbal category that can be assigned to the
proto-language is only the past/non-past tense distinction. (It may be noted here
that a negative suffix, *a˜/*va˜, has also been reconstructed for Proto-Dravidian.)
As I have pointed out in the second chapter, this past/non-past tense
distinction continues to be the most grammaticalized verbal category in all the
modern Dravidian languages. Most of the complex morphophonemic alternations
that are connected with the inflectional forms of the verb occur in the formation
of the past stem. In Kannada, for example, non-past stems are formed by adding
the suffix utt to the base (and the personal suffixes are attached to these stems).
The formation of these non-past stems involves very few morphophonemic

changes. The formation of past stems, on the other hand, involves several
changes as can be seen from the following Kannada examples. In these exam-
ples, I give the past forms, which contain the verbal base (which has undergone
different types of changes), followed by the past tense suffix (also showing
several changes), and by the III person masculine singular suffix a:
Verbal base Past form Verbal base Past form
he˜Lu ‘say’ he˜Li-d-a baru ‘come’ ban-d-a
ki˜Lu ‘pluck’ kit-t-a horu ‘carry’ hot-t-a
bi˜Lu ‘fall’ bid-d-a nillu ‘stand’ nin-t-a
bare ‘write’ bare-d-a gellu ‘win’ ged-d-a
mare ‘forget’ mare-t-a kollu ‘kill’ kon-d-a
kuDi ‘drink’ kuDi-d-a nagu ‘laugh’ nak-k-a
kali ‘learn’ kali-t-a tinnu ‘eat’ tin-d-a
sa˜yu ‘die’ sat-t-a uNNu ‘dine’ uN-D-a
ka˜yu ‘wait’ ka˜-d-a ho˜gu ‘go’ ho˜-d-a
The past and non-past paradigms are also differentiated from one another in
Kannada by the personal suffixes which occur in them, as can be seen from the
following paradigms of the verb ho˜gu ‘to go’:
Past paradigm Non-past paradigm
Singular Plural Singular Plural
1 ho˜d-e ho˜d-evu ho˜gutt-eñe ho˜gutt-e˜ve
2 ho˜d-i ho˜d-iri ho˜gutt-i ho˜gutt-i˜ri
3 ho˜d-a ho˜d-aru ho˜gutt-añe ho˜gutt-a˜re
 ho˜d-aLu ,, ho˜gutt-a˜Le ,,
 ho˜yi-tu ho˜d-uvu ho˜gutt-ade ho˜gutt-ave
Aspect and mood markers are less grammaticalized than these tense
markers. There is a subjunctive paradigm in some of these languages which,
however, shows less number of personal distinctions. Other modal distinctions
and also aspect distinctions are denoted by auxiliary (modal or vector) verbs,
which are added to the prior and posterior forms of the main verb respectively.
Examples from Kannada:
(1) a. i˜vattu avanu be˜ga bar-a-bahudu
today he soon come--may
‘He may come soon today’
b. ivattu avanu be˜ga bandu-biDuttañe

today he soon come--leaves
‘He will come (Completive) soon today’
6.2.2 Obligatoriness of tense
Tense markers are obligatory in the verbal forms of Dravidian languages in the
sense that both finite as well as non-finite verbs necessarily contain a tense
marker, which may be either of the deictic or of the non-deictic type. Forms that
have been described by grammarians as showing no tense distinction actually
show non-deictic tense distinction as I have pointed out in the second chapter
(see 2.4.1). Mood and aspect markers, on the other hand, are used only when the
speaker desires to specify the relevant meaning or to denote it with emphasis.
Consider, for example, the following pair of Kannada sentences:
(2) a. avanu bahuša na˜Le bara-bahudu
he probably tomorrow come-may
b. avanu bahuša na˜Le baruttañe
he probably tomorrow comes (.)
‘He will probably come tomorrow’
Both (2a) and (2b) indicate the modal notion of probability but in (2a) the
meaning is conveyed by the adverb bahuša ‘probably’ as well as the modal form
of the verb barabahudu ‘may come’, whereas in (2b) it is conveyed only by the
adverb. The use of the modal verb is not obligatory just as the use of the modal
adverb is not obligatory, as in the following sentence:
c. avanu na˜Le bara-bahudu
he tomorrow come-may
The use of aspect markers is also non-obligatory in a similar fashion; the
relevant meaning can be expressed either by an adverbial or by an aspect marker
(vector verb) or by both of them together in a sentence. Examples:
(3) a. avanu ella˜ baTTegaLann-u˜ oge-du ha˜kida
he all cloths-too wash- put ()
‘He washed off all the cloths’
b. avanu ella˜ baTTegaLann-u˜ ogeda

he all cloths-too washed
‘He washed all the cloths’
c. avanu baTTegaLannu oge-du ha˜kida
he cloths wash- put ()
‘He washed off (all) the cloths’
Notice that the vector verb ha˜ku ‘put’ denoting the aspectual notion of
‘completion’ occurs along with the adverb ella˜ ‘all’ (and the emphatic marker
u˜ ‘too’) in (3a), whereas in (3b) the adverb and the emphatic marker occur
without the vector verb; in (3c), on the other hand, the vector verb occurs
without the adverb and the emphatic marker. All the three sentences convey the
aspectual notion of completion.
Tense markers cannot be left out from a verbal form in this fashion. Their
use in finite forms is obligatory as can be seen from the following examples. They
must also agree with the adverbials that occur with them in finite sentences.
(4) a. avanu maysu˜rin-inda ninne ban-d-a
he Mysore- yesterday come--3:
‘He came from Myosre yesterday’
b. *avanu maysu˜rin-inda ninne baru-tt-añe
he Mysore- yesterday come-.-3:
*‘He comes from Mysore yesterday’
c. *avanu maysu˜rin-inda na˜Le ban-d-a
he Mysore- tomorrow come--3:
*‘He came from Mysore tomorrow’
d. avanu maysu˜rin-inda na˜Le baru-tt-añe
he Mysore- tomorrow come-.-3:
‘He comes from Mysore tomorrow’
The language also does not have any verbal forms that are unmarked for
tense, i.e. forms that can replace the past and non-past tense forms in finite
sentences like (4a–d) given above. There are a few modal verbs which do not
show any tense distinction, such as bahudu ‘may’ and be˜ku ‘want’, but these
have only non-past connotation; they take the auxiliary iru ‘to be’ (as in bar-a-
bahud-ittu come--may-be--3: ‘should have come’) in order to
denote past meaning. This is also true of the imperatives, which show tense
distinctions in some Dravidian languages like Malto (6.5.6). Even in other
languages, such forms can be regarded as temporally “unmarked” rather than
“unspecified” for tense because tense being an obligatory and systematic

category in these languages, such unmarked forms can also specify tense. In the
case of imperatives (and other forms that are unmarked for tense), we can
assume that the forms denote non-past or future tense.
6.2.3 Systematicity of tense
The use of tense markers is also systematic as compared to that of mood and
aspect markers in these tense-prominent languages. The paradigms that represent
tense markers are generally perfectly formed in Dravidian languages; they rarely
show any gaps or neutralisations of distinctions. Aspect and mood markers, on
the other hand, do not have such perfectly formed paradigms. They either do not
have any paradigms as such or have only irregularly formed paradigms. For
example, the aspectual constructions in Tamil show a gradation of grammatical-
ization with some of them being in the border area in which they can as well be
regarded as complex verbs. Even in the case of other constructions, there is
always the possibility of having an alternative interpretation for the construction
in which the vector verb functions as an independent (main) verb.
This lack of systematicity in the use of aspectual markers in Tamil can be
seen in its use of the verb viTu ‘to leave, release’ as a vector verb for denoting
perfective meaning. There is no specific “imperfective” vector verb with which
this perfective verb can be contrasted; instead, the simple, un-extended or un-
compounded verbal form has to function as the unmarked form in contrast with
this perfective form. There is, however, a different vector verb, namely muTi
‘finish’ which contrasts with viTu in denoting a “completed” action as against the
“complete” (perfective) action denoted by viTu (Annamalai 1985: 85). Examples:
(5) a. na˜]kaL ca˜ppiTTo˜m
we ate
‘We ate’
b. na˜]kaL ca˜ppiTTu-viTTo˜m
we ate-
‘We have eaten’
c. na˜]kaL ca˜ppiTTu-muTitto˜m
we ate-finished
‘We finished eating’
Annamalai refers to several interesting constraints that form the basis for a
distinction between these two vector verbs. He suggests that these constraints can
be explained as resulting from the fact that viTu indicates a complete event (i.e.
an event presented as a whole) whereas muTi indicates a completed event; the
latter emphasises the terminal portion of the event. One interesting constraint
which results from this meaning distinction, according to him, is that some other
event can be said to have been occurring at the time when the event under
consideration is being completed in the case of muTi but not in that of viTu.
Examples (Annamalai 1985: 86):
(6) a. na˜]kaL ca˜ppiTTu-muTitta-ppo˜tu malai peytukoNTiruntatu
we ate-finish-time rain raining
‘It was raining when we finished eating’
b. *na˜]kaL ca˜ppiTTu-viTTa-ppo˜tu malai peytukoNTiruntatu
we ate-complete-time rain raining
*‘It was raining when we have eaten’
Similarly, muTi can occur with temporal adverbs referring to a specific time
(i.e. the time of completion) but viTu cannot. Examples:
(7) a. ni˜]kaL inta na˜valai eppo˜tu paTittu muTitti˜rkaL
you this novel when read finished
‘When did you finish reading this novel?’
b. *ni˜]kaL inta na˜valai eppo˜tu paTittu viTTi˜rkaL
you this novel when read completed
However, viTu, even though denoting a complete event, is not constrained
by tense or mood distinctions; it can occur in the future (as well as in the past)
and also with the modal notions of probability or certainty. Examples:
(8) a. appa˜ na˜Laikku perumpa˜lum vantu-viTu-va˜r
father tomorrow most.probably come-complete-
‘Father will most probably return tomorrow’
b. appa˜ na˜Laikku niccayama˜ka vantu-viTu-va˜r
father tomorrow definitely come-complete-
‘Father will definitely come tomorrow’
When used with the present (non-past) tense suffix, however, it can only
provide a future meaning, and not a “present” meaning. Example:
(9) nañ atai ca˜ppiTTu viTu-kkireñ

I it ate complete-
‘I will eat eat’
Another interesting point about the use of this “perfective” vector verb viTu
in Tamil is that, even though it implies resultative meaning in some of its usages,
in the case of certain other usages, it can be followed by a clause that denies the
occurrence of the expected result (Annamalai 1985). Example:
(10) a. kuma˜rai nañ ella˜ iTattil-um pa˜rtuviTTeñ;
Kumar () I all places-also look.for ();
kaÑo˜m
found.not
‘I looked for Kumar everywhere; he was not found’
This lack of systematicity in the use of aspectual markers in Tamil has
prompted Annamalai (1983: 1) to argue that equating these vector verb construc-
tions (or “extended constructions”) with the grammatical notion of aspect (or
mood) is misguided.
The category of mood is represented more systematically than aspect in
these Dravidian languages, but still, there are several irregularities, gaps and
neutralisations which make that representation appear to be rather unsystematic
when compared with the representation of tense. The subjunctive (future)
paradigm splits into third person and rest of the persons in some languages, with
the latter showing no gender-number distinctions. The following Havyaka forms
exemplify this paradigm:
Subjunctive paradigm of the verb tinnu ‘to eat’
singular plural
I person timbe timbeyõ (exclusive)
tingu (inclusive)
II person timbe timbi
III person tingu
Notice that the subjunctive suffix is b in first and second person forms (except-
ing the exclusive form in I person plural) and g in other forms; further, the
gender and number distinctions have been neutralised in the third person.
Other modal forms occurring in Dravidian languages are also irregular as
compared to tense forms; some of them show person-gender-number distinctions
and some do not; the distinctions are also neutralised to different degrees in the
case of different modal paradigms. On the whole, they are less systematic,
indicating that the category of mood plays a less prominent role in these
languages.
6.2.4 Pervasiveness of tense
Tense distinctions are more pervasive than aspect and mood in Dravidian
languages. In addition to occurring obligatorily in finite verbal forms (with very
few exceptions), they also occur in most of the non-finite verbal forms. In these
latter usages, however, they may be deictic or non-deictic. They are generally
deictic in adjectival participles (relative clauses), whereas in converbs (which are
used as adverbials or as part of the aspectual, modal or negative constructions)
they occur in their non-deictic form. In the case of temporal adverbials, which
also make use of converbs, on the other hand, we find both deictic as well as
non-deictic tense distinctions. Thus, most of the uses of verbal forms involve the
expression of some variety of temporal distinctions in Dravidian languages.
(i) Adjectival (relative) participles used as nominal modifiers
The adjectival participles may be used, in Dravidian languages, either as nominal
modifiers or as part of temporal adverbials; in the former usage, they show a
past/non-past deictic tense distinction. They are generally derived by adding the
participial suffix a or i to the past and non-past stems respectively. The following
Konda (Krishnamurti 1969: 302) and Kannada phrases exemplify this usage:
(11) a. vañru uNs-t-i gumeNDi]
he plant-- pumpkin
‘the pumpkin that he planted’
b. ru˜-n-i guided soRad
plough-.- field went
‘(She) went into the field they plough’
(12) a be˜yis-id-a baTa˜Te
cook-- potato
‘cooked potato’
b. be˜yisu-v-a baTa˜Te
cook-.- potato
(i) ‘potato that is to be cooked’
(ii) ‘potato for cooking’
The tense distinction occurring in these adjectival participles is deictic in the

sense that it is directly related to the utterance time and not to the point of time
of the main verb. This point becomes clear in the following Kannada sentence:
(13) ga˜Li-ge bid-d-a marak-ke nañ-u˜ hatt-idde
wind- fall-- tree- I-too climb-was
‘I too had climbed the tree that fell by wind’
The point of time of the adjectival participle bidda ‘fallen’ is earlier than that of
the utterance time, but it is not necessary that it must be earlier than that of the
event denoted by the main verb in the sentence, namely hattidde ‘had climbed’.
That is, the climbing could have taken place, according to (13), either before or
after the tree had fallen.
It is possible to retain some of the aspectual and modal distinctions in these
adjectival participles, but their retention, unlike the retention of tense distinctions,
is not obligatory. In fact, the tense distinctions have to be shown both before as
well as after the aspectual and modal markers in these vector verb constructions
which, I think, clearly indicates the prominence of tense in these languages.
Examples (from Kannada):
(14) a. kelasav-annu mugis-i biT-T-a huDuga
work- finish- release- boy
‘the boy who finished off the work’
b kelasav-annu mugis-a bahud-a˜d-a huDuga
work- finish- possible-become- boy
‘the boy who might finish the work’
Notice that the completive vector verb biDu ‘release’ requires the prior suffix to
occur before it in addition to the past suffix which occurs after it. The modal
predicate bahudu ‘possible’, on the other hand, requires the posterior suffix to
occur before it in addition to the past suffix which follows it; the latter requires,
in addition, the supportive verb a˜gu ‘become’ as a tense carrier.
Dravidian languages also have a negative adjectival participle, but in most
of these languages, it does not show any tense distinction. It is derived by adding
the negative suffix to the verbal base, as in the following Tamil expression
(Subrahmanyam 1971: 337):
(15) murai ceyy-a˜ mannavan
justice do- king
‘the king who does not do justice’
South Dravidian languages as a whole, however, use an extended form of

the negative suffix (such as a˜t in Tamil) which, according to Subrahmanyam
(1971: 392), involves the addition of the past tense suffix to the negative suffix.
Example:
(16) ceyy-a˜-t-a ve˜lai
do--- work
‘work which was not done’
Further, some of the Dravidian languages like Malayalam in the southern
region and Pengo, Kui and Kuvi in the central region, make a past/non-past
distinction in these negative participles also, by using the relevant tense suffixes
with the negative suffix. The following Malayalam examples (Abraham 1978: 145)
exemplify this usage:
(17) a. innale kLa˜ss-il var-añña kuTTikaL
yesterday class- come- () students
‘students who did not come to the class yesterday’
b. na˜Le kLa˜ss-il var-a˜tta kuTTikaL
tomorrow class- come- (.) students
‘students who will not come to the class tomorrow’
The non-past negative participle, however, can also occur with a past
adverbial as shown by the following sentence:
(17) c. innale kLa˜ss-il var-a˜tta kuTTikaL
yesterday class- come- (.) students
‘students who did not come to the class yesterday’
(ii) Adjectival participles used in other contexts
As mentioned earlier, adjectival participles can also be used in the formation of
temporal adverbials. The tense distinctions that they indicate in these usages are
not as sharp and clear-cut as in the previous usage (i.e. in relative clauses). In
Kannada, for example, the past and non-past adjectival participles occur before
temporal adverbs like andu ‘on that day’, a˜ga ‘then’, divasa ‘day’, va˜ra ‘week’,
etc. to denote deictic tense distinctions. The event that is denoted by the particip-
ial verb would be prior or posterior to the utterance time respectively in these
usages. Examples:
(18) a. avanu ban-d-a˜ga ra˜tri a˜gittu

he came--then night became
‘It was night when he came’
b. avanu baru-v-a˜ga ra˜tri a˜guttade
he come-.-then night becomes
‘It will be night when he comes’
A temporal adverbial that contains the past participle can only be used to
denote a past reference point in this fashion, whereas the one which contains the
non-past participle can be used to denote a non-past as well as a past reference
point. Examples:
(18) c. avanu baru-v-a˜ga ra˜tri a˜gittu
he come-.-then night became
‘It was night when he came’
d. *avanu ban-d-a˜ga ra˜tri a˜guttade
he come--then night becomes
The two adjectival participles can also be used with certain other temporal
adverbs, but in these cases their use is constrained by the meaning of these
adverbs. That is, the past participle can occur with adverbs me˜le ‘after’ and
ku˜Dale˜ ‘immediately’, whereas the non-past participle can occur with adverbs
like modalu ‘before’ and varege ‘until’. That is, the two pairs of adverbs denote
the relevant non-deictic tense distinctions and the use of participles is constrained
by these adverbs. Examples:
(19) a. avanu kelasa ma˜D-ida me˜le ho˜da
he work do- after went
‘He went after doing the work’
b. *avanu kelasa ma˜Du-va me˜le ho˜da
he work do-. after went
(20) a. avanu kelasa ma˜Du-va modalu ho˜da
he work do-. before went
‘He went before doing the work’
b. *avanu kelasa ma˜D-ida modalu ho˜da
he work do- before went
Another interesting construction in which the adjectival participles retain
their past/non-past (deictic) tense distinction is that of action nominals or verbal
nouns which are formed by attaching the neuter singular personal marker to the
past and non-past stems, as shown in the following Kannada examples:
(21) a. avanu baru-v-udu samšaya
he come-.-3: doubtful
‘It is doubtful that he will come’
b. avanu ban-d-udu samšaya
he come--3: doubtful
‘It is doubtful that he came’
Aspect and mood distinctions are not generally expressed in these construc-
tions which contain adjectival participles.
(iii) Converbs used as verbal modifiers
In contrast to adjectival participles, we find converbs (which are also called
adverbial or verbal participles) being used for denoting non-deictic tense
distinctions. The point of time that they indicate has the point of time of the
following verb (which may be the main verb or another adverbial participle) as
its reference point. I have already described this usage in detail in the second
chapter (see 2.4.1). The following Tulu sentences exemplify this usage:
(22) a. undenG keÑ-DGtG a˜yagG santo˜S-a˜puNu
this hear- him happy-becomes
‘Having heard this, he becomes happy’
b. a˜ye i˜ kathe-nG keÑ-ontu nalitte
he this story- hear- danced
‘While hearing this story, he danced’
c. ma]ge ko˜TenG su˜-vere pida˜Dye
monkey fort see-erior started
‘The monkey departed for seeing the fort’
Most of the grammars of Dravidian languages do not recognise these
converbs as representing a single paradigm apparently because their translations
in familiar languages like English do not constitute such a paradigm; only one of
them, namely the Prior form, is generally considered to be a “verbal participle”.
An additional non-past or durative participle is also recognised in the case of
some of these languages, which actually functions as the simultaneous form.
Posterior form, however, is almost invariably described as an infinitive or
purposive. It appears to me, however, that the three forms are basically temporal
in these languages and do form a paradigm.
(iv) Other uses of converbs

Converbs are used in certain other types of constructions in which the non-
deictic tense distinctions that they indicate appear to be of indirect relevance.
This is true, for example, of their use in verbal constructions which denote aspect
and mood distinctions, and also the ones that denote distinctions of phase
(beginning, continuing, ending, completing, etc.).
As mentioned earlier, aspect and mood distinctions are indicated in Dravid-
ian languages primarily with the help of the so-called compound verb construc-
tions in which the main verb in its converbial form is attached to another verb
which has an aspectual or modal meaning. The main verb takes the form of a
prior (past) converb in the case of aspectual constructions and that of a posterior
converb (or “infinitive”) in the case of modal constructions. Examples (Tamil:
Lehmann 1989: 209, 214):
(23) a. kuma˜r inta na˜valai paTi-ttu viTTañ
Kumar this novel read- left ()
‘Kumar read this novel’ (finished off reading)
b. inta terivil naTa-kka muTiyum
this road walk- possible
‘It is possible to walk on this road’
The prior-posterior tense distinction occurring in the main verb of these
constructions is of indirect relevance here, in the sense that the prior form
facilitates the denotation of aspectual meanings which are basically past, and the
posterior form facilitates the denotation of modal meanings which are basically
non-past or future. That is, the Dravidian languages view aspectual and modal
meanings from a temporal standpoint.
This is also true of phasal distinctions. In Kannada, for example, ingressive
predicates like toDagu ‘start’ or pra˜rambhisu ‘begin’ occur with posterior
participles but not with prior ones, whereas egressive predicates like mugisu
‘finish’ occur with prior participles but not with the posterior ones. Examples:
(24) a. avanu ma˜tana˜D-alu toDagida
he speak- started
‘He started to speak’
b. *avanu ma˜tana˜D-i toDagida
he speak- started
(25) a. avanu ma˜tana˜D-i mugisida

he speak- finished
‘He finished speaking’
b. *avanu ma˜tana˜D-alu mugisida
he speak- finished
Predicates denoting stoppage or continuation, however, take neither of these
participles in Kannada; instead, they take a non-past verbal noun.
(26) a. avanu ma˜tana˜Duvud-annu nillisida
he speaking- stopped
‘He stopped speaking’
b. avanu ma˜tana˜Duvud-annu munduvarisida
he speaking- continued
‘He continued speaking’
Similar temporal distinctions in the formation of phasal constructions occur
in other Dravidian languages also.
(v) Conditional forms
Most of the Dravidian languages make a non-deictic prior-posterior tense
distinction in their conditional sentences, but the actual markers used for deriving
these forms appear to be the deictic ones (i.e. the ones occurring in the finite
forms of verbs rather than the ones occurring in adverbial participles). In the
Havyaka dialect of Kannada, for example, past and non-past stems take the
conditional suffix are to form the conditional verb which occurs in the protasis
of a conditional sentence and denotes non-deictic prior and posterior tenses
respectively. Examples:
(27) a. kelasa ma˜D-id-are payse koDutte
work do-- money give
‘I will give money (later) if you do the work (now)’
b. kelasa ma˜Du-tt-are payse koDutte
work do-.- money give
‘I will give money (now) if you (will) do the work (later)’
The non-past conditional, however, can also denote deictic tense in that the
event denoted by the conditional (protasis) can be earlier than the event denoted
by the apodosis. Example:
(28) niñu ba-tt-are añu-de batte

you come-.- I-too come
‘If you are coming I too will come’
Notice that ‘my coming’, according to (28), need not be posterior to ‘your
coming’; it must, however, be occurring after the utterance time, and is therefore
in future tense.
The non-past conditional can also be used in a counterfactual situation, but
the past conditional needs to be replaced by the corresponding prior-past form
(see 29b) in that context. Examples:
(29) a. kelasa ma˜Du-tt-are payse koDutt-itte
work do-.- money give-would
‘I would have given money if you were going to work’
b. kelasa ma˜D-i-d-d-are payse koDutt-itte
work do--be-- money give-would
‘I would have given money if you had done the work’
The situation in Tulu is rather interesting. It has the conditional suffix Da
attached to finite verbal forms. That is, the conditional shows all the person-
number-gender distinctions, in addition to the tense distinctions, that are shown
by finite forms. Verbs retain the formal distinction between past and non-past
and also between immediate past and non-immediate past, but the meaning dis-
tinction denoted by them is non-deictic (i.e. relative to the event denoted by the
apodosis as in the case of Havyaka conditionals described above). Examples:
(30) a. a˜ye bar-p-e-Da e]kG po˜-voLi
he come-.-3:- me go-can
‘If he comes, I can go’
b. unde-nG tin-d-e-Da jvara baranG
this- eat--3:- fever come ()
‘If he eats this, fever will not come’
c. und-enG tint-t-e-Da jvara baratvanG
this- eat--3:- fever came ()
‘If he had eaten this, fever would not have come’
Notice that the conditional forms retain the agreement markers (3.) of the
corresponding finite verbal forms.
6.2.5 Constraints on the study
As I had mentioned earlier, the available grammars are rather heavily biased
towards tense-prominence and hence it is difficult to decide as to whether a
given language, described as tense-prominent, actually gives prominence to tense
or to some other category like aspect or mod. I have therefore restricted myself,
in the foregoing sub-sections, to describing the possibility of a set of languages
(of the Dravidian family, which are the most familiar to me,) being tense
prominent. I have pointed out how these languages grammaticalize tense to a
greater degree than aspect or mood, and also make tense more obligatory, more
systematic, and more pervasive than aspect or mood. There are clearly several
other languages that can be regarded as belonging to this tense-prominent
language type.
For example, some of the Indo-European languages like English and
German and some of the Uralic languages like Finnish appear to be tense-
prominent. The category of tense is grammaticalized to a greater extent than
aspect or mood in these languages. In English, for example, the basic inflectional
distinction, shown in the verbal system, is between past and present (or non-
past); much of the complexity that is involved in the derivation of verbal forms
occurs in these tense forms (especially that of the past tense). Aspect and mood
distinctions, on the other hand, are not grammaticalized to the extent to which
tense distinctions are grammaticalized. They occur only in the form of auxiliary
verbs that are attached to the past or present participles. For example, progressive
aspect is denoted by using the verb be as an auxiliary (as in am writing and was
writing) whereas several modal verbs like can, may, shall, will, must etc. are used
as auxiliaries for denoting modal distinctions.
Tense is also an obligatory category in English, and as I point out elsewhere
(see 5.7) this obligatoriness makes it rather difficult to translate into English
verbal forms of other languages in which tense distinctions have been left
unspecified. In Finnish also, tense marking is obligatory as has been pointed out
by Sulkala & Karjalainen (1992: 297). Tense is also more pervasive than aspect
or mood in these languages. In Finnish, for example, tense distinctions occur not
only in simple tense forms, but also in the forms that denote perfect, conditional
and potential (see Sulkala & Karjalainen 1992: 315–6). Another language that
appears to belong to this group is Nama Hottentot. According to Hagman
(1973: 119), tense is obligatory in this language; aspect, on the other hand,
showing a perfective-imperfective distinction, is non-obligatory. Aspect is further
restricted, in this language, to active verbs, whereas tense occurs with both active
as well as state verbs.
6.3 Aspect-prominent languages
There are several languages in which the category of aspect is more prominent
than tense or mood. The grammars of some of these languages, however, fail to
reveal this point mainly because the grammatical tradition followed by their
authors emphasises the category of tense at the expense of aspect and mood. For
example, Li (1991) points out that the grammars of Hmong use terms like past,
present and future for describing its verbal markers but actually the basic
distinction that the markers represent is between the two aspectual categories of
perfective and imperfective. The former is denoted by the markers tau ‘attain-
ment’ and lawn ‘completion’, whereas the latter is denoted by the marker
taabtom ‘progressive’. Li points out further that there is only a single tense
marker in this language, namely yuav ‘future’ which, however, is non-obligatory
and can be left unused in a sentence if it contains a temporal expression indicat-
ing future time.
In the grammars of Indo-Aryan languages also, we find the names of
temporal distinctions like past, present and future being freely used even when
the actual distinctions to be named are aspectual and not temporal. This is
especially true of Sanskrit, as I will be pointing out below (see 6.3.1). Masica
(1991: 262) considers this to be true of modern Indo-Aryan languages as well
(see also Hackman 1976). Additional terminological problem has been caused by
the use of the term “perfect” in place of the term “perfective” in these and other
aspect-prominent languages. It could be seen from the following description,
however, that aspect is similar to tense in playing the role of the most prominent
verbal category. Languages that give prominence to aspect are also similar to
tense-prominent languages in that they also fall into different gradations concern-
ing the type of prominence (in grammaticalization, obligatoriness, systematicity
or pervasiveness) that they give to the category of aspect. All of them, however,
give greater prominence to aspect than to tense or mood on all these points.
6.3.1 Grammaticalization of aspect
We can exemplify the aspect-prominent language type with the help of some of
the languages of the Gur family of Niger-Congo, such as Supyire (Carlson 1994)
and Koromfe (Rennison 1966). In Supyire, for example, Carlson points out that
the great majority of verbs have two forms, a base which is the perfective form,
and a derived one which is the imperfective form. Most tense-aspects require one
or the other of these two forms, and a few like the habitual may take either. The
derivation of the imperfective is rather complex involving several suffixes, and
processes like vowel raising, initial consonant mutation, and tonal change
(Carlson 1994: 130). Examples:
Base (perfective) Imperfective Gloss
cùgò cùgùlì deep
cenme cenmì transplanted
muguro mugure smile
kwù kwùù die
bya byìì drink
ce ceni know
kanha kanre be tired
yige yìgè take out
Koromfe is similar in having a major aspectual division among its verbal
forms between perfective and imperfective but within the imperfective, it makes
a further distinction between durative and progressive. The perfective form is
unmarked whereas the imperfective has an “extended” stem whose derivation
may be (i) regular (involving only the addition of a suffix), (ii) semi-regular
(involving the addition of a suffix and also the truncation of the end of the verb
stem) or (iii) irregular. The progressive is derived from the durative through
further extensions (Rennison 1996: 277).
Compared to these aspectual distinctions, which are clearly grammaticalized
to a very high degree in these languages, tense and mood distinctions are
grammaticalized to a lesser extent. In Supyire, for example, there is no present
tense form as such. Progressive and habitual forms have present time reference
if they are not accompanied by another tense marker. There are two different
auxiliaries used for denoting past tense reference, namely ná ‘remote past’ and
nî ‘recent past’. There is also an auxiliary màha used to introduce formal
narratives like folk tales or myths. These markers of past time reference are not
used again and again in each clause, but only in the beginning; once the past
time reference has been set, it can be assumed to persist until the speaker notifies
otherwise (Carlson 1994: 328).
In Koromfe also, there is no verbal inflection for denoting present or future

reference. In order to indicate past reference, the language uses the suffix e,
which is added to the perfective form of the verb. The suffix is used, however,
only when there is a necessity to do so; past time reference can be obtained in
the absence of the suffix as well, as in the use of the unmarked (perfective) form,
and the durative form of the progressive. Modal distinctions, on the other hand,
are not grammaticalized, except for the imperative, which is the same in form (in
the singular) as the unmarked verb or the durative form (Rennison 1996: 276).
The various Indo-Aryan languages spoken in India can also be regarded as
aspect-prominent. According to some scholars, this prominence of aspect has
been inherited by these languages from the Indo-European parent language itself.
For example, Lehmann (1974) considers the aspectual distinction between
perfective and imperfective to be the basic distinction among the verbs of Proto-
Indo-European. This distinction, according to him, was indicated either by means
of affixes or by characteristic forms of the verbal root. There was also a further
contrast between momentary and durative aspects. Tense distinctions, on the
other hand, were not indicated by means of verbal affixes in the (reconstructed)
Proto-Indo-European. Instead, certain particles or adverbs were used for denoting
the time of action. It was also possible to merely allow the time of action to be
implied by the aspects of verbal forms. Kiparsky (1968: 43) claims that tense and
mood were not features on verbs, but rather adverbs in Indo-European; this claim
is supported, according to him, by the fact that in Early Indo-European, tenses
could be contrasted, negated, etc., just like temporal adverbials.
Aspect gets grammaticalized to a higher degree than tense or mood in
Sanskrit as well. The grammarians of Sanskrit have named the four main verbal
forms of Sanskrit as laT, liT, lu] and lRT. They have devised these terms in such
a way that the statement of morphophonemic rules connected with their use is
economised. These terms have been translated into English as present, perfect,
aorist and future respectively. But these translations do not reflect their actual
connotation or usage. For example, it is aorist (lu]) rather than perfect (liT)
which is translated by the perfect forms of the English language (Whitney
1888:201). Further, present (laT) can be used in contexts in which past meaning
is to be indicated, as for example, with the adverb pura˜ ‘formerly’ or the
particle sma as shown below, or even by itself:
(31) iti vakti sma pa˜rvati:
thus speak ()  Parvati
‘Thus spoke Parvati’
According to Burrow (1955: 294), the clearest distinction to be found in this

somewhat complicated system of Sanskrit tense stems is the one between the
perfect [perfective] system indicating a state on the one hand, and the rest of the
conjugation indicating a process [imperfective] on the other. The two are
distinguished from one another not only in stem-formation, but also in the fact
that they possess distinct sets of personal endings. Burrow suggests that between
the perfect and the rest of the conjugation we have clearly the most ancient and
fundamental division in the Indo-European verbal system.
The two sets of stems are quite different from one another in the type of
morphophonemic variations that they show in their formation. The grammarians
of Sanskrit have found it necessary to postulate ten different conjugational
classes in order to describe the derivation of present [imperfective] stems, but the
description of perfect stems requires no such division of verbal bases. In the
former case (present stems), the conjugational distinction involves factors like (i)
taking a thematic vowel, (ii) undergoing internal vocalic changes (ablaut), (iii)
taking an affix ya in addition to the thematic vowel, (iv) showing reduplication
(generally of the first consonant), (v) taking the stem sign no/nu/nv, o/u or
na˜/ni˜, and (vi) taking the infix na (or n) between the vowel and the final
consonant of the root (see Mayrhofer 1972).
There is no comparable complexity in the formation of perfect stems; the
root is reduplicated in all cases, and is regularly affected by ablaut (vowel
gradation). This difference in the formation of present [imperfective] and perfect
[perfective] stems can be exemplified with the help of the following sets of third
person singular forms in the two major aspectual systems:
Root Gloss Present Perfect
kR ‘to make’ karoti caka˜ra
bhu˜ ‘to be’ bhavati babhu˜va
duH ‘to milk’ dogdhi dudoha
kup ‘to be angry’ kupyati cukopa
bhR ‘to bear’ bibharti babha˜ra
rudh ‘to obstruct’ ruNaddhi rurodha
hu ‘to sacrifice’ juhoti juha˜va
su ‘to press’ sunoti suSa˜va
muc ‘to release’ muncati mumoca
car ‘to move’ carati caca˜ra
vaH ‘to carry’ vahati uva˜ha
yaj ‘to offer’ yajati iya˜ja

pac ‘to cook’ pacati papa˜ca
In addition to this distinction in stem-formation, present [imperfective] and
perfect [perfective] forms also differ from one another in the set of personal
terminations that occur after them. There are two distinct sets of terminations, of
which one is used with present stems (and also with the aorist and the future
stems), and the other one is used with perfect stems. For example, the suffix
used for denoting the first person singular subject is mi in the case of present
stems, whereas in that of perfect stems it is a. Similarly, the second singular
suffix is si in the present form and tha in the perfect form. There are several
other differences of this type between these two aspectual forms as can be seen
from the following examples:
karo˜mi ‘I do’ caka˜ra ‘I did’
karosi ‘you do’ cakartha ‘you did’
kurvati ‘they do’ cakruH ‘they did’
All these points together indicate that the primary aspectual distinction is
highly grammaticalized in Sanskrit, whereas the tense and mood distinctions
show only very little amount of grammaticalization.
6.3.2 Obligatoriness of aspect
Aspect-prominent languages show an obligatory marking of aspect distinctions,

whereas they indicate tense and mood distinctions through specific markers only
when the meaning is not derivable from the context. Noonan (1992) points out,
for example, that in Lango, a Nilotic language of Nilo-Saharan family, verbs are
inflected for perfective, progressive and habitual aspects. Out of context,
perfective will be interpreted as past, habitual as present and progressive as
future; they can, however, be assigned to any tense (except that the perfective
may not be present) through the use of appropriate temporal adverbials. There
are, however, a number of auxiliary verbs that can be used to create definite
tense interpretations. For example, the use of bínô ‘come’ in the habitual
provides future meaning when used with the infinitive form of the main verb.
The use of nwò]]ò ‘to find’ in its perfective (third person singular) form with
the perfective verb provides relative past (or past perfect) meaning; with the
progressive and habitual, on the other hand, it provides past progressive and past
habitual meanings respectively. Examples (Noonan 1992: 138):
(32) bínô c7’m

(3) come () eat ()
‘He will eat’
(33) a. án ònwò]ò ábwôt7’
I (3) find () (1) deceive (-3)
‘I had deceived him’
b. ònwò]ò lóce ác7’m
(3) find () man (3) eat ()
‘The man was eating’
c. ònwò]ò ákwánô Ì l7’b lá]ô
(3) find () (1) study () in tongue Lango
‘I used to study Lango’
Notice, however, that the use of these auxiliaries is not obligatory, as the
meaning can be expressed by the use of the relevant temporal adverbials like
cwàñ mentioned above.
6.3.3 Systematicity of aspect
We can exemplify the systematicity of aspect by examining the occurrence of the

distinction between perfective and imperfective, indicated by the use of two
distinct sets of personal agreement affixes, in Maltese. The imperfective forms
are derived by the combination of a stem with a set of imperfective agreement
prefixes, showing a three-fold person distinction and a two-fold number distinc-
tion in third person. These imperfective forms function as the basis for the
formation of the prospective and the progressive. The prospective is formed by
combining the imperfective form with the particle sa (or other particles like se,
ser, ha or the verb sejjer ‘go’). The progressive, on the other hand, is derived by
adding the particle qed to the imperfective form. Maltese makes use of a distinct
set of perfective personal (agreement) affixes for deriving perfective forms (see
Fabri 1995). In contrast to these aspectual distinctions, the denotation of tense and
mood distinctions is rather unsystematic in Maltese. Present tense may be denoted
through the use of the imperfective form, but the latter may also denote habitual
or even future events if the form is accompanied by an explicit time adverbial.
The bare perfective form may be used for denoting past events (Fabri 1995).
We can further exemplify the systematicity of aspect (and the non-systemat-
icity of tense and mood) in aspect-prominent languages with the help of Tamir,
an Austroasiatic language belonging to the Aslian group (spoken in the Malay

Peninsula). According to Benjamin (1976: 168), this language makes primarily a
distinction between perfective, simulfactive and cotinuative aspects. In general,
the perfective aspect is unmarked, consisting of the root alone. The other two are
formed by reduplicative processes, involving the initial and final consonants in
the continuative aspect and the initial consonant and the marker vowel -a- in the
simulfactive aspect. These aspect distinctions occur in nominalizations and also
in gerunds. Compared to these aspect distinctions, the denotation of tense and
mood distinctions is less systematic. Tense is indicated with the help of certain
auxiliaries like hfj ‘already’, ti‘ ‘yet, still’ and boleh ‘can’. The auxiliary hfj
‘already’ is the most common way of indicating past tense. There is also a
desiderative marker -m- that may be used for denoting the imperative (in second
person); it also has purposive or resultative connotation.
6.3.4 Pervasiveness of aspect
The perfective-imperfective distinction is very pervasive in modern Indo-Aryan

languages. In addition to occurring in finite verbal forms, it also occurs in
several other types of verbal constructions like adjectival and adverbial partici-
ples, verbal compounds, conditionals, temporal adverbials, and also in nominal-
izations. Tense and mood distinctions, on the other hand, do not appear to get
involved in any of these derived verbal constructions.
In Punjabi, for example, perfective and imperfective forms occur contrast-
ively in adjectival constructions, with the imperfective form denoting an action
and the perfective form denoting the corresponding state (Bhatia 199-:308,
Rangila, Personal communication). Examples:
(34) a. sau-ndi˜ kuRi˜
sleep- girl
‘sleeping girl’ (in the action of sleeping)
b. su-tti˜ kuRi˜
sleep- girl
‘sleeping girl’ (in the state of sleeping)
(35) a. bai-ndi˜ kuRi˜
sit- girl
‘sitting girl’ (action)
b. bai-Thi˜ kuRi˜
sit- girl
‘sitting girl’ (state)
Similar distinction is shown by Hindi also, in which the imperfective
participle denotes ‘a doer in the act of doing’ and the perfective one denotes ‘the
doer of an act’ (Scholberg 1940: 177–8). Examples:
(36) a. hãs-te hue
laugh- be
‘the laughing one’
b. soy-a˜ hua˜
sleep- be
‘the one who has slept’
Cardona (1965: 136) points out that in Gujarati, there is a perfective
infinitive derived by adding the suffix el to the perfective verbal form, which can
be used both as a nominal attribute and also as a predicate. It contrasts with the
imperfective verbal form, which can also be used as a nominal attribute.
Examples:
(37) a bolati bhaša
speak- language
‘the spoken language’
b. pidhelo maN6s
drink- man
‘the drunken man’
Punjabi uses the marker rì/rà (derived from the verb denoting ‘remain’) for
indicating the progressive aspect. The verb can show the imperfective-perfective
contrast before this aspect marker with the imperfective form denoting an action
and the perfective form denoting a state (Bhatia 1996: 275). We may regard this
as showing the retention of imperfective-perfective distinction in compound
verbs, since this construction derives from an earlier verbal compound. Examples
(Rangila, personal communication):
(38) a. mãi sonda˜ rìa
I sleep- remain
‘I kept falling asleep’ (action)
b. mãi sutta˜ rìa˜

I sleep- remain
‘I remained asleep’ (state)
Kellogg (1938: 268) points out that in Hindi also, both imperfective as well
as perfective verbal forms can be used in compound verbs (called by him as
“conjunct verbs”), as for example, with the verb ja˜ ‘to go’ in the following pair
of sentences:
(39) a. vah likhta˜ ja˜ta˜ hai
he write- go- is
‘He going on writing’
b. ek ba˜gh paRa˜ phirta˜ tha˜
one tiger fall- move.around- was
‘A tiger was prowling about’
The two usages differ from one another, according to him, in that the
subject in (39b) is represented as having completely come into a certain state, in
which state it is then represented as remaining or moving.
The distinction is also retained in some of the adverbial constructions in
Punjabi, in which the imperfective form denotes an on-going action whereas the
perfective form denotes a completed action (Bhatia 1996: 65, 68). Examples:
(40) muNDa˜ nasda˜ a˜ia˜
boy run- came
‘The boy cam running’
(41) a. ó hasda˜ boLia
he smile- spoke
‘He spoke smiling’
b. muNDa˜ mañji te baiTha˜ boLia˜
boy cot on sit- spoke
‘The boy spoke sitting on the cot’
In the case of adverbs, on the other hand, the contrast appears to show up
in Punjabi in the form of a habitual/non-habitual distinction, with the imperfec-
tive form denoting an action that had been carried out constantly on a number of
occasions and the perfective form denoting an action that was carried out on a
single occasion. Examples (Rangila, Personal communication):
(42) a. baiThde baiThde ó tang a˜ gia˜

sit- sit- he sick come went
‘He got sick of sitting continuously’ (on a number of occasions)
b. baiThe baiThe ó tang a˜ gia˜
sit- sit- he sick come went
‘He got sick of sitting continuously’ (on a single occasion)
According to Cardona (1965: 136), imperfective and perfective forms occur
contrastingly in some of the time phrases of Gujarati, as can be seen in the
following phrases:
(43) a. awti kale ‘tomorrow’ (imperfective with ‘come’)
b. g6i kale ‘yesterday’ (perfective with ‘go’)
(44) a. awte athwaDie ‘next week’ (imperfective)
b. g6yũ 6ThwaDiu ‘last week’ (perfective)
6.4 Mood-prominent languages
Mood is generally considered to be somewhat different from tense and aspect as

a prototypical verbal category. It is considered to be less closely associated with
the verb, less united as a category, and less coherent in its encoding (see Bybee,
Perkins and Pagliuca 1994: 238). It is also considered to be more speaker-
oriented than situation-oriented. Its closer association with illocutionary acts like
imperatives and interrogatives has heightened this impression that it is somehow
different from tense and aspect in its position as a verbal category. Most
importantly, I think, the fact that in familiar languages, mood is less prominent
than tense or aspect, and is grammaticalized to a lesser degree than the latter is
crucially responsible for this feeling.
If we are willing to make an unbiased comparison of the three categories,
we will find them to be similar in being coherent or incoherent, speaker-oriented
or situation-oriented, more or less closely associated with the verb, and more or
less grammaticalized, depending upon the language that we choose to study.
Since these differences occur in the representation of these categories in different
languages and not among the categories as such, we need to regard them as
representing different manifestations of the three idealised language types (tense-
prominent, aspect-prominent and mood-prominent) and not as characterising the
categories themselves.
As in the case of tense and aspect categories described earlier, we find

languages falling into different types of gradations, such as those of grammatical-
ization, obligatoriness, systematicity and pervasiveness, in the case of the modal
category as well. The typologist, however, has to deal with a comparatively more
biased data in this case. Basic modal distinctions are frequently described as
tense distinctions, and only the exotic distinctions that cannot be treated in this
fashion, such as evidential and epistemic distinctions, are regarded as modal
distinctions. The existence of a more basic realis-irrealis distinction has been
recognised only recently. There is therefore a need to make an unbiased exami-
nation of many of these languages in order to obtain a better understanding of
the exact nature of the modal category occurring in them.
The need for such a re-examination has been pointed out by some scholars.
For example, Comrie (1985: 51) argues that the description of two different sets
of particles occurring in Burmese, namely me/ma/hma and the/tha/tǎ/hta, which
are added to verbal bases in order to derive finite as well as non-finite forms, as
“future” and “non-future” respectively (Okell 1969) may not be correct. The
prime function of these particles, according to him, is not time reference but
rather the denotation of an irrealis-realis modal distinction. It is true that
sentences in which verbal forms containing the former particle occur are
generally translatable with English future sentences, and the ones in which verbal
forms containing the latter particles occur are translatable with present or past
sentences. Examples (Okell 1969: 355, 424):
(45) a. mǎne‘hpañ sá-me
tomorrow begin-
‘We shall begin tomorrow’
b. moù ywa-nei-te
sky rain-stay-
‘It is/was raining’
c. hpwíñ-htà-ta
open-put-
‘I did open it’
Comrie points out, however, that while the realis particle can only be used
in sentences that have present or past time reference (with no grammatical
distinction between past and present time reference), the irrealis particle can be
used not only in contexts with future time reference, but also in contexts with
present time reference, provided the reference is not restricted to our actual world,
i.e. provided there is modal value to the particle. Examples (Okell 1969: 355):
(46) màcìthì sà-hpù-me htiñ-te
tamarind eat-ever- think-
‘I think he must have eaten tamarinds before’
Comrie considers (46) to be of special interest in the present context
because both irrealis as well as realis forms occur together in that sentence. The
irrealis me denotes the supposition as to what the agent may have done, and the
realis te denotes the fact as to what the speaker actually thinks. Notice that the
time reference of the irrealis me is in fact prior to that of the realis te in this
sentence, indicating clearly that time reference is not basic to the opposition
between these two particles (Comrie 1985: 51). It may be noted here that Allott
(1965: 267) describes the Burmese particles te and me as “realized” and “unreal-
ized” respectively.
Another instance in which a re-examination of the use of category names
has been advocated is that of Northern Iroquoian languages. Descriptions of these
languages began with verb morphology that used a “tense-laden” terminology, as
pointed out by Foster (1985); it was only after Chafe’s (1960) “Seneca morphol-
ogy” that a shift away from the language of tense to the language of mood
occurred. Foster points out, for example, that the term aorist or past was used
earlier for denoting verbal forms that can be translated in some of the Northern
languages as present. What the forms actually indicate is that the meaning is an
“uncontestable fact”. It has now been generally replaced by the term “factual” by
the Iroquanists. The temporal bias persists, however, in the general discussions
of these languages, and also in most of the translations of sentences. The latter
apparently is unavoidable, but we should not allow it to mislead us.
DeChicchis (1996) examines the use of terms such as “habitual present”,
“recent past”, “remote past” and “definite future” for describing the verbal
category markers (n, x, k and t respectively) occurring in Q’eqchi’, a Mayan
language. He argues that the common thread of these markers is not temporal,
but rather the cognitive status of an event vis-à-vis the speaker or agent. For
example, he finds the marker n to be indicating disposition rather than present as
in the following sentences:
(47) a. hoon na-q-il
soon -.1-see
‘Soon we will see it’
b. ink’a‘ n-in-k’ayih
 -.1-sell
‘I cannot sell it’
The marker t, on the other hand, may be described as denoting predictive
meaning according to DeChicchis:
(48) t-oo-xik se‘-tenamit
-.1-go in-town
‘We can go into the town’
Further, the marker k denotes the assertion of an event which need not
necessarily be past, as can be seen from the following:
(49) hoon k-at-in-k’ux
soon -.2-.1-eat
‘Soon I will eat you’
If languages like the ones mentioned above are conceded to be making a
basic modal distinction rather than a temporal one, we would notice that these
modal distinctions are very similar to tense and aspect distinctions described
earlier in assuming the position of the most prominent verbal category. They can
be grammaticalized to a greater degree than tense or aspect distinctions in these
languages, and further, they can also be more obligatory, systematic, and
pervasive than tense or aspect. We can therefore regard these languages as
representing the third idealised language type, namely the mood-prominent one.
6.4.1 Grammaticalization of mood
According to van der Berg (1989), Muna, an Austronesian language belonging

to the Western Malayo-Polinesian branch (spoken in the Muna Island of Indone-
sia), makes a basic distinction in its verbal forms between realis and irrealis
moods. The distinction is expressed by the use of two sets of subject markers; in
the case of some verbs, there is also an infix um occurring in the irrealis forms,
which distinguishes them from realis forms. There are several morphophonemic
alternations that are connected with the use of this infix. The following pairs of
forms exemplify this realis-irrealis modal distinction:
Realis Irrealis Gloss (of irrealis form)
a-kala a-k-um-ala ‘I will go’
no-horo a-h-um-oro ‘It will fly’

a-gholi ae-gholi ‘I will buy’
de-basa dae-basa ‘We will read’
omo-gharo omo-gharo ‘You will be hungry’
no-lodo nao-lodo ‘He will sleep’
The realis forms can refer to either past or present. They can occur with the
suffix ho to denote future. Irrealis forms, on the other hand, can refer to the
future (but cannot occur with the future suffix ho) or can express a wish, desire
or intention. They are also used obligatorily in negative clauses, i.e., in the
presence of negators such as miina ‘not’, miina-ho ‘not yet’ and pa ‘will not’.
The realis-irrealis distinction, however, is not maintained in nominalizations,
participles and imperatives, as the subject marker does not occur in these forms
(van der Berg 1989: 59).
Kayardild (an Australian language of south Wellesley Island) is another
mood-prominent language in which mood is grammaticalized to a greater extent
than aspect or tense. According to Evans (1995), verbal stems of this language
can be followed by one of a set of final inflections. Most of these, such as
Actual, Suppositional, Potential, Desiderative, Hortative, Apprehensive and
Imperative, denote primarily modal meanings. There is a past suffix which,
however, is used only when the speaker wishes to emphasise the pastness of the
action. In other contexts the Actual form is used for denoting past events. There
is a suffix used for indicating that an action almost occurred, but it usually refers
to undesirable actions. There is also a suffix denoting actions that are directed
towards the speaker. Examples:
(50) jungarra bawa-tha warrngald
big blow- wind
‘The wind is blowing strong’
(51) niya bukawa-thu mungkijiwu dulku
he die- own country
‘He will die in his own country’
(52) wakatha ngukuntha yalawu-jinj
sister water fetch-
‘Sister should fetch some water’
Aspectual distinctions, on the other hand, are indicated by using certain
verbs like wirdi-ja ‘be, stay’, dii-ja ‘sit’, karrngi-ja ‘hold, grasp’ and jirrma-ja
‘pile up’ after the main verb. Examples:

(53) nyingka kurrinda warranda wirdind
you see go be
‘You are going around to see (people) a lot’
(54) niya diya-ja karrngi-j
he eat- grasp-
‘He keeps eating, he is eating all the time’
6.4.2 Obligatoriness of mood
According to Chafe (1995), Caddo makes a distinction between realis and irrealis
with the help of two distinct sets of pronominal prefixes that are attached to the
verb. These prefixes indicate, in addition to the realis-irrealis distinction, other
distinctions like person, focus and case. They may combine, while denoting first
and second person agents, with prefixes that denote first and second person
patients and beneficiaries, to form more complex prefixes. All these markers
indicate the realis-irrealis distinction. The following are some of the contrasting
prefixes that belong to these sets:
Realis Irrealis
I person agent ci- t’a-/t’i-
I person patient ku- ba-
III person beneficiary nu- ‘u-
I person agent with
II person beneficiary t’u- t’a‘u-
Defocusing agent yi- ‘a-
The occurrence of this distinction in the verbal forms, according to Chafe,
“is an obligatory, clearly marked, and unambiguous feature of every pronominal
prefix (with one minor exception) and thus of every verb”. There are a variety
of contexts that condition the use of the irrealis prefixes. They include yes-no
questions, negations, obligation, conditionals, and several others like simulative
dúy ‘as if’, infrequentative wás ‘seldom’, admirative hús ‘surprise’, etc. In all
these types of sentences, one of the pronominal prefixes belonging to the irrealis
set must be used. Prefixes belonging to the realis set, on the other hand, are to
be used, also obligatorily, in other contexts that include past and future.
The realis-irrealis distinction is obligatory in Manam (belonging to the
Oceanic subgroup of the Austronesian family and spoken in Manam Island of

Papua New Guinea) as well. Lichtenberk (1983) reports that the language possess
two sets of verbal prefixes whose function is (i) to index the person and number
of the subject, present or ellipted, and (ii) to mark the mood, realis or irrealis.
Every finite verb must be specified, by means of this subject-mood prefix, either
as realis or irrealis. The Manam verbal system can also show a number of
aspectual distinctions like completive, continuative, persistive, etc., but the
indication of these distinctions is not obligatory. It also makes use of certain
posture and motion verbs like soa‘i ‘to sit’, eno ‘to lie’, and la‘o ‘to go’ to
express aspects like progressive, persistive and interruptive respectively (Lichten-
berk 1983: 181).
6.4.3 Systematicity of mood
There do occur languages in which the expression of modal distinctions is more

systematic than that of tense or aspect. This is the case in languages like Muna
and Caddo described earlier. Achumawi, of the Shastan family, also shows a
basic modal distinction, which is more systematic than that of tense or aspect.
According to De Angulo and Freeland (1931), verbs of this language show a
primary distinction between Indicative and Volitional, represented primarily by
the sets of personal prefixes that occur before the verb. They are also differenti-
ated from one another by the type of stem alternants that occur in them. The
Indicative form provides a simple statement about an action, which is indefinite
as to time, whereas the Volitional form indicates the wilful intent to carry out an
action. The latter corresponds, to a certain extent, to the imperatives of other
languages, but it includes more than the imperative. Examples (De Angulo and
Freeland 1931: 89):
(55) a. l-ám
I ()-eat
‘Let me eat, I wish to eat right now’
b. s-a¦.m-á
I ()-eat-Ending
‘I eat now, will eat later on, ate’
The denotation of aspect and tense distinctions, on the other hand, is less
systematic in this language. There is a present-future distinction, indicated by the
use of distinct prefixes, which, however, is restricted to the Volitional paradigm.
Other temporal distinctions like distant past and future eventuality, along with
aspectual distinctions like habituative, continuative and perfective are represented
by suffixes which are added to the verb stem in a “derivational manner”. These
suffixes belong to a class of suffixes which includes directionals like into, out of,
down, up, and also the reflexive and reciprocal.
6.4.4 Pervasiveness of mood
We have already seen how the two-fold realis-irrealis modal distinction occurs
in some of the mood-prominent languages like Caddo and Manam in exactly the
same way in which the past/non-past tense distinction occurs in some of the
tense-prominent languages and the perfective-imperfective aspect distinction
occurs in some of the aspect-prominent languages. The denotation of the
distinction is pervasive, affecting both finite as well as non-finite verbal forms.
In Amele, for example, medial verbs (i.e. verbs used in clause chaining) do not
show any tense or aspect distinctions, but when the subjects of chained verbs are
different (and the actions are simultaneous), they show a realis-irrealis modal
distinction. This distinction has the additional effect of dividing the various
verbal affixes occurring in the final verb into realis and irrealis sets, with the past
and present tense categories being regarded as realis, and a range of other
categories like future, imperative, prohibitive, counterfactual, prescriptive,
hortative, apprehensive, etc. as irrealis (Roberts 1990). Examples:
(56) a. ho bu-busal-en age qo-in
pig -run.out-3::R 3 hit-3::
‘They killed the pig as it ran out’
b. ho bu-busal-eb age qo-qag-an
pig -run.out-3:: 3 hit-3-
‘They will kill the pit as it runs out’
Amele also makes a realis-irrealis distinction in its conditionals (see 7.2 for
examples).
I had pointed out earlier that the category of mood is grammaticalized to a
greater degree than tense or aspect in Kayardild, an Australian language. The
modal distinctions that this language represents occur not only in verbs function-
ing as predicates but also the ones functioning as adverbials. Examples (Evans
1995: 304):
(57) bakii-ja ngakulda warra-ju

altogether- we go-
‘We will all go’
(58) dali-ja kuliya-th
come- fill-
‘Lots of (people) are coming’
(59) ngada yulkalu-tha dana-tha ngumbanji
I eternal- leave- you
‘I am leaving you for good’
In addition to this, Kayardild uses some of its case markers for denoting
modal connotations. That is, case suffixes, whose basic function is to denote the
relation between the argument and the verb (or the one between two arguments),
can also occur as supporting the modal meaning of the verb or even, in some
contexts, for denoting such meanings by themselves. Such modal case markers
can occur with all noun phrases except subjects and the ones that are in some
way semantically oriented with the subject. Examples (Evans 1995):
(60) nyngka kurri-nang-ku niwan-ju balambi-wu
you see-- her-() morrow-
‘You will see her tomorrow’
(61) ngada warra-ja ngarn-kir
I go- beach-
‘I am going/have gone to the beach’
(62) dangkaa raa-jarra bijarrba-na wamburugu-nguni-na
man spear- dugong- spear--
‘The man speared the dugong with a spear’
The case markers Propriative in (60), Allative in (61), and Ablative in (62)
have been used as modal markers rather than as case (relation) markers. Evans
(1995) points out that some of these case markers are formally identical with the
corresponding verbal mood markers as in the case of the Porpriative (modal)
case and the Potential verbal marker, Locative case and Immediate marker, or
Oblique case and Hortative marker. There are also semantic correlations between
the two: for example, Ablative correlates with verbal suffixes such as Past,
Precondition and Almost, and Allative correlates with the verbal suffix denoting
directed action. Evans points out, however, that the modal case can also vary
independently allowing for a multiplication of subtle meanings.
Chafe (1991) reports that Caddo has a set of defocusing prefixes whose
general function is to deflect attention from referents that may be irrelevant,
unknown or defocused for some other reason. These prefixes show distinctions
between agent, patient and beneficiary, but in addition to this, the three show a
further distinction between realis and irrealis as shown below:
Defocusing Prefixes:
Realis Irrealis
Agent yi- ‘a-
Patient ya- ‘a‘a-
Beneficiary yu- ‘a‘u-
The prominence that Caddo gives to the modal distinction between realis
and irrealis through grammaticalization has already been referred to above (see
6.4.1). The indication of this distinction in these defocusing prefixes exemplifies
its pervasiveness in Caddo.
In Lewo, an Oceanic language, the realis-irrealis modal distinction affects
negation as well. Early (1994) reports that the language has two different
negative particles, ve ‘irrealis negation’ and pe ‘realis negation’. These two
particles precede the verb, and are used along with two other particles, re, which
follows the verb and poli, which follows the direct object, if there is one.
Examples:
(63) a. naga ô-vano
he 3-go()
‘He will go’
b. naga ve ô-va re
he () 3-go() 
‘He will not go’
(64) a. naga ô-pano
he 3-go()
‘He has gone’
b. naga pe ô-pa re poli
he () 3-go()  
‘He has not gone’
6.5 Summary
The foregoing study of the tense-prominent, aspect-prominent and mood-promi-

nent languages has shown that there do occur languages belonging to different
areas and genetic affiliations that exemplify the three idealised language types
that I had earlier established. The three verbal categories (tense, aspect and
mood) are also very similar to one another in assuming a greater degree of
prominence by way of grammaticalization, obligatoriness, systematicity and
pervasiveness. As I had mentioned earlier, there is a need to re-examine the
grammars of these and other languages from the point of the relative prominence
that they attach to these three categories, before we can establish the actual set
of languages that belong to one or the other of these three language types.
C 7
Correlatable Characteristics
7.1 Introduction
I have suggested, in the previous chapter, a typological classification of languag-

es into tense-prominent, aspect-prominent and mood-prominent language types on
the basis of the relative prominence that they give to tense, aspect and mood
respectively. My purpose in the present chapter is to justify this classification by
showing that it is relevant for a proper understanding of the nature of language.
I propose to do this by describing some of the characteristics and tendencies that
can be correlated with this classification. Some of these characteristics involve
long-standing disputes in linguistics; it would be shown that the classification
brings out an entirely different facet of these disputed topics. There are also
cross-linguistic variations that have remained unexplained, and the present
typology appears to provide a basis for explaining them. Other correlations that
are brought up by this study emphasise the necessity to re-examine some of the
language-universal generalizations that have been proposed.
It must be noted, in this connection, that these correlatable characteristics
are being ascribed to the three different idealised languages and not to the actual
sets of languages that may be grouped under them. My claim is that languages
can be regarded as being more or less close to one of these idealised languages
depending upon the number of characteristics that they share with it. Further, this
closeness would be correlatable with the degree of prominence that the languages
attach to the relevant verbal category (tense, aspect or mood) as described in the
previous chapter. It is especially this latter possibility of establishing a correlation
between these two variables, which makes this typology of verbal categories
worth pursuing.
7.2 Effects of decategorization
I have pointed out in Bhat (1994: 91) that when lexical items belonging to a
particular category are used in the function of some other category, they show
the characteristics of both decategorization as well as recategorization. For
example, when verbs are used in the function of nouns through nominalization,
they fail to show several of their verbal characteristics like manifesting distinc-
tions of tense, aspect or mood, taking agreement markers and showing various
types of voice distinctions. These can be regarded as characteristics of decategor-
ization. In addition to this, they also take on nominal characteristics like showing
count-mass distinction, definite-indefinite distinction, and taking plural markers
and markers for case. These can be regarded as characteristics of recategori-
zation. Similar characteristics of decategorization as well as recategorization are
shown by words belonging to other lexical categories like nouns, adjectives and
adverbs as well, when they are used in the function of other categories.
The fact that lexical items tend to get decategorized when they are used in
functions which are not their own has been recognised by several linguists. For
example, Hopper and Thompson (1984: 710) describe several types of contexts
in which nouns and verbs show only some of their categorial characteristics.
They point out that verbs used in contexts in which they do not report an event
fail to show a range of oppositions characteristic of those used in contexts in
which they do report an event, such as (i) having the markings for agreement
with the subject or object, and (ii) showing distinctions of tense, aspect and
mood. According to Comrie and Thompson (1985: 361), these decategorized
forms may involve a cline of expressibility of verbal categories: “finite verbs can
express the most such categories, non-verbal categories fewer, action nominals
still fewer, and other noun phrases fewest of all” (see also Givón 1990: 498).
Languages have also been found to differ rather markedly from one another
concerning the extent to which they decategorize verbal categories in such
extended uses of verbal bases (see Bhat 1994). They have also been found to
differ from one another concerning the type of verbal category that they retain.
The former variation can be correlated with the degree of decategorization that
the verbal forms attain in such constructions as pointed out above, but the latter
variation has generally remained unexplained.
In fact, linguists appear to have been unduly influenced by aspect-prominent
languages like some of the Indo-European ones in their understanding of this
latter phenomenon (i.e. the type of category that gets neutralised or lost in that
CORRELATABLE CHARACTERISTICS 143
process). For example, Binnick (1978: 296) claims that deverbatives are to be
regarded as deriving from underlying structures that are marked only for aspect
and not for tense. He considers tense to be primarily a category of independent
(i.e. topmost) clause; it occurs in non-topmost clauses, according to him, only in
certain privileged enclaves, such as direct quotes or that-clauses, but not
normally in non-topmost clauses like non-finite forms. According to Comrie and
Thompson (1985: 360), the category of mood is relatively rare in non-finite verbal
forms. For example, they consider it possible, in the case of action nominals, to
claim that the modal category is simply not retained. Tense also, according to
them, is retained in the form of relative tense rather than that of absolute tense in
these non-finite verbal forms. Comrie (1985: 93) suggests that loss of tense in con-
ditionals can be correlated with the degree of their hypotheticality.
A contrastive study of tense-prominent, aspect-prominent and mood-
prominent language types indicates, however, that the retention of a particular
category in non-finite forms depends upon the kind of prominence that the
category receives in the language under consideration and not necessarily upon
the nature of that category (i.e. whether it is tense, aspect or mood). Tense-
prominent languages appear to retain tense more readily than aspect or mood,
whereas aspect and mood-prominent languages appear to retain aspect and mood
respectively more readily than the other two categories. It is true that this
difference between tense-prominent, aspect-prominent and mood-prominent
languages is correlatable with the observation that the most prominent category
would be grammaticalized to a greater degree than any of the other categories in
the language concerned, and hence one can claim that the greater degree of
retention of the prominent category in decategorization results from the greater
degree of its grammaticalization in these languages. It is also correlatable with
the generalization that the most prominent category is also the most pervasive
one in a language. The fact remains, however, that the retention of these verbal
categories in decategorization does not depend upon the nature of those catego-
ries as is apparently assumed by some linguists.
I have already given examples for this phenomenon in the previous chapter
when I described the pervasiveness of different grammatical categories in
different languages. For example, tense is retained to a greater extent than aspect
or mood in the non-finite verbal forms of Dravidian languages. In Kannada, for
example, tense distinctions occur either in the deictic or non-deictic form in
almost all constructions that involve non-finite forms like converbs and relative
participles; aspect and mood distinctions do not occur in most of these construc-
tions. In aspect-prominent languages, on the other hand, aspect-distinctions are

retained to a greater extent than tense or mood distinctions in non-finite verbal
forms. For example, the participial constructions of Lezgian show the basic
distinction between imperfective and aorist (perfective), and also between other
aspectual distinctions like continuative and perfect but not past aorist, past
perfect or past future; future is indicated only in the imperfective in these
participial constructions (Haspelmath 1993). Similarly, in the case of mood-
prominent languages, we find the distinctions that get retained in non-finite
verbal forms to be modal rather than temporal or aspectual. In Amele, for
example, medial verbs with different subjects show a modal distinction between
realis and irrealis, but not any of the tense or aspect distinctions.
This contrast between tense-prominent, aspect-prominent and mood-promi-
nent languages in their retention of verbal category distinctions in non-finite
forms can be shown clearly through an examination of the conditionals that
occur in three different languages, namely Havyaka (tense-prominent), Gujarati
(aspect-prominent) and Amele (mood-prominent). In Havyaka, the conditionals
show a two-fold prior-posterior tense distinction, whereas in Gujarati the
distinction is between the perfective and the imperfective aspects; in Amele, on
the other hand, the distinction is between the realis and the irrealis moods
(Roberts 1990). Examples:
Havyaka:
(1) a. niñu ba-tt-are paysa koDutte
you come-.- money give
‘I will give you money (now) if you will come (later)’
b. niñu ban-d-are paysa koDutte
you come-- money give
‘I will give you money (after you come) if you will come’
Gujarati:
(2) a. tuN w6kh6ts6r awe to apNe b6har j6ie
you on.time come() then we out might.go
‘If you come on time, we might go out’
b. e saurN kam k6rto hOt to pr6phes6r that
he good work do() be then professor become
‘If he were doing good work, he would (have) become a
professor’
Amele:
(3) a. ho busale-gi-na fi aqo-qo-na
pig run(out)-3- if () 3()-hit-1-
‘If the pigs run out we kill them’
b. ho busalo-u-b mi a-qo-u-m
pig run (out)- if (Irrealis) 3()-hit-1-
‘If the pigs had run out we would have killed them’
7.3 Ergativity split
Variations occurring among languages in their case marking and agreement

systems have formed the basis for a typological classification of languages into
accusative and ergative (and also into languages that are neither accusative nor
ergative and called “active”). In order to provide an explicit description of the
main distinction involved here, linguists have found it convenient to use three
main “core” concepts, generally symbolised as S, A and O. Of these, S repre-
sents the single obligatory argument of intransitive sentences, whereas A and O
represent the two obligatory arguments of transitive sentences. A is typically the
controller of actions, and O is typically the affected argument (see Dixon 1979,
1994, Comrie 1981). Examples:
(4) a. The dog died
S
b. The man killed the dog
A O
Following Dixon (1994), the distinction between accusative and ergative
systems can be described as follows:
Accusative system Ergative system
Nominative A Ergative
S Absolutive
Accusative O
The crucial difference between these two systems is that in the case of the
accusative system, the core argument S (single obligatory argument of intransi-
tive sentences) is (i) case marked and also (ii) marked by verbal agreement in
exactly the same way as A, whereas in that of ergative systems, it is marked in
the same way as O. The sentences given above show that English is an accusa-
tive language since S (the dog) and A (the man) are both in the nominative and
the verb agrees with them. Basque, a language isolate, spoken in Pyrenees, on
the other hand, shows the ergative system as can be seen from the following
sentences (Eguzkitza 1987):
(5) a. gizona etorri do
man () come is
‘The has come’
b. gizona-k libura bidali du
man- book () sent has (it)
‘The man has sent the book’
Notice that in (5a) gizona ‘the man’ (S) is unmarked for case (i.e. it occurs in
the absolutive case) whereas in (5b) gizona ‘the man’ (A) occurs with the
ergative suffix k and libura ‘the book’ (O) is unmarked for case (it occurs in the
absolutive case). That is, S is similar to O rather than A in its case marking.
There are some languages in which the ergative system is restricted to some
contexts only, with the accusative system being used in other contexts. These
languages have been regarded as showing “split-ergative” systems. The actual
conditioning of this split can be of different types such as
(i) the semantic nature of the verb (controlled vs. uncontrolled),
(ii) the semantic nature of the arguments (I or II person vs. third person), and
(iii) distinctions in the tense or aspect of the verb (past vs. non-past or perfective
vs. imperfective).
The ergative system, in these split-ergative languages, will be restricted to
(i) non-controlled verbs, (ii) third persons arguments, and (iii) past or perfective
verbs, respectively (see Dixon 1994).
This ergativity split appears to be correlatable with the verbal category that
is most prominent in a given language. This is especially true of the second and
third type of ergativity splits mentioned above. That is, mood-prominent languag-
es tend to show a personal (or nominal) split whereas tense or aspect-prominent
languages tend to show a tense-based or aspect-based split respectively. For
example, in the case of Indo-Aryan languages, which are generally aspect-
prominent as I have shown in the previous chapter, the ergativity split is
conditioned by the perfective-imperfective aspectual distinction of the verb. The
personal terminations occurring in the verbal forms of these languages agree with
the subject in the case of imperfective forms, whereas in that of perfective forms
they agree with the object. The following Marathi sentences exemplify this split
(Berntsen and Nimbkar 1982):
(6) a. polis c6uk6ši k6rit ahet
police enquiry doing be (3)
‘The police are making enquiries’
b. 6mhi s6mšodh6n k6rit ahot
we research doing be (1)
‘We are doing research’
(7) a. šetk6ryan-ni kam ke-l-6 hoti
farmers- work (:) do--: had
‘The farmers had done the work’
b. šetk6ryan-ni jvari per-l-i hoti
farmers- jowar (:) plant--: had
‘The farmers had planted jowar’
Notice that in (6a–b), which has an imperfective verb, the agreement is with
subject, whereas in (7a–b), which has a perfective verb, the agreement is with the
direct object.
Dixon (1979: 95) considers this split to be resulting from the fact that in the
case of an imperfective verb, something that has not yet happened (i.e. something
that is only thought of as the propensity of the potential agent) is being indicated,
and hence the emphasis can be on the agent (subject), whereas in the case of a
perfective verb, a completed event is being indicated (i.e. an event that can be
thought of as something that has affected an object) and hence the emphasis can
be on the object. Trask (1979: 396), on the other hand, provides a somewhat
different explanation for this split, namely that an imperfective verb indicates a
process that an agent (subject) is performing and hence the agreement is with the
agent, whereas a perfective verb indicates a state which resides in the patient
(direct object) and hence the agreement is with the patient. There is no agreement
split in the case of intransitive verbs in these languages because both processes as
well as states would be residing on the agent or subject in these cases.
We may contrast this situation with that of some of the Tibeto-Burman lan-
guages in which a split in the agreement pattern of personal markers occurs on
the basis of an evidentiality distinction. Most of the Tibeto-Burman languages do
not attach any agreement markers to their verb; however, there are some
languages like Tangut and Sherpa, called “pronominalizing” languages, in which
there do occur agreement markers on verbs. According to DeLancey (1981: 631),

a number of these languages show a nominative-ergative split pattern, but the
split is conditioned by the fact as to whether or not a speech-act participant
(speaker or hearer) assumes the role in the event concerned.
In Tangut, for example, there is no agreement with third person arguments.
If either the subject or the object is third person, and the other is first or second
person, however, the verb will agree with the first or second person; if the
subject is first person and the object the second, or vice versa, agreement is with
the object. Tangut also shows agreement with the possessor of subject or object,
provided that the possessor is of first or second person (DeLancey 1983: 104). In
Nocte, another Tibeto-Burman language, agreement is always with a speech act
participant (speaker or hearer) in preference to a third person. Unlike Tangut,
however, Nocte does have a third person agreement marker that occurs when no
speech act participant is involved. Further, when the agent is second person and
the patient is first person, the agreement is with the patient, whereas when the
agent is first person and patient the second, the agreement is with the agent (first
person plural). Examples (DeLancey 1981: 641, quoted from Das Gupta 1971):
(8) a. ate-ma nga-nang hethoh-ang
he- I- teach-1
‘He will teach me’
b. nga-ma ate hetho-ang
I- he teach-1
‘I will teach him’
c. nang-ma nga hethoh-ang
you- I teach-1
‘You will teach me’
d. nga-ma nang hetho-e
I- you teach-1
‘I will teach you’
Bauman (1979: 419) argues that this split-ergative agreement pattern is
reconstructible to Proto-Tibeto-Burman itself. The idealised split-ergative mode on
which these patterns are based, according to him, is one in which the agreement
is with the object if the object is of first or second person, and with the subject
if it is of third person. The true reason for this split, according to him, involves
the evidential nature of the speaker’s assertion; i.e. whether he (or his hearer) has
actively participated or whether it is a conjecture based on some other type of
evidence (1979: 425). Bauman points out that this notion of evidentiality plays a
crucial role even in the case of Sherpa, another Tibeto-Burman language, which
has independently developed a somewhat different agreement pattern. In that
language, the agreement suffix yin is used if the speaker and the actor are the
same, and a different agreement suffix wu is used if the hearer and the actor are
the same. However, these two suffixes are used only if each has, or is expected
to have, first-hand knowledge of the proposition. The suffix yin is therefore used
only in declarative sentences and wu only in interrogative sentences; all other
propositions which lack the same measure of certainty to the speaker are marked
with disjunct role markers or have unmarked verbs (Bauman 1979: 426).
7.4 Tensedness parameter
Stassen (1997) makes a distinction between tense-oriented and aspect-oriented

languages, a distinction that is correlatable, to a certain extent, with the distinc-
tion that I make here between tense-prominent and aspect-prominent languages.
He considers languages to be “tensed” or tense-oriented if they have a grammati-
cal category of tense (i) which is bounded on verbs and (ii) which minimally
involves a distinction between past and non-past time reference. He does not
define aspect-oriented languages, but apparently regards all non-tensed languages
to be aspect-oriented. As a result, he has no place for mood-oriented languages
in his generalization. Further, his tensed (or tense-oriented) languages need not
necessarily be tense-prominent; it is sufficient if they satisfy the conditions
mentioned above (see, however, below).
Stassen finds the distinction between tensed and non-tensed languages to be
useful in predicting the type of encoding that languages use for their adjectival
predicates. He considers these to be encoded as verbs by default; his claim is that
when a language is tensed, this default option for encoding gets overruled, and
as a result, adjectival predicates are not treated on a par with verbs; instead, they
are encoded like nominal or locational predicates. For example, Supyire, a non-
tensed (aspect-prominent language), shows the default system of encoding
adjectival predicates as verbs, as can be seen from the following sentences
(Carlson 1994):
(9) kafáága a pèè
stone ()  big
‘The stone is big’
(10) sikà]a a bò
goat ()  kill
‘The goat has been killed’
(11) yire wà pyì yè
they () be (they) children
‘They are children’
Notice that the predicate pèè ‘big’ in (9) and the verbal predicate bò ‘kill’ in (10)
occur directly with the perfective marker, whereas the nominal predicate yè
‘children’ in (11) requires an auxiliary support.
English, on the other hand, is a tensed language. Its verbal core system
consists of two paradigms of finite forms, called simple present and simple past,
which have present and past time reference respectively. Accordingly, the
adjectival predicates of English are not encoded as verbs; they require an
auxiliary support, and are similar to nominal predicates on this point.
(12) a. The boy sleeps.
b. The boy is tall.
c. The boy is a singer.
Stassen’s explanation for this tensedness constraint on the encoding of adjectival
predicates is the following: Adjectives are more time-stable than verbs and hence
tense-marking is not as relevant to them as for verbs; further, such a marking is
non-iconic or may even be anti-iconic for them and is therefore avoided. (A
similar claim and also a similar explanation for the above-mentioned tensedness
constraint have been put forth by Wetzer 1996). Strassen tests his claim on the
basis of data gathered from 410 languages (mainly published sources) belonging
to different parts of the world and finds it to be supported by a majority of them.
We can regard this tensedness constraint as representing another characteristic
of tense-prominent languages. The fact that these languages give greater promi-
nence to the time of occurrence of the concepts that their verbs denote makes the
use of adjectival predicates as verbs non-iconic as pointed out above. This point
gets support in the additional fact that I point out below (see 7.5), namely that
tense-prominent languages tend to have no state verbs as such (or very few
irregular state verbs); their verbs are primarily dynamic in nature, denoting either
actions or processes. Aspect-prominent languages and mood-prominent languages,
on the other hand, tend to have substantial classes of state verbs.
While trying to account for problematic cases, i.e. instances of languages
which, though “tensed” by definition, do allow verbal encoding for adjectival

predicates, Stassen had to make use of the notion of “degree” of being tensed or
non-tensed. For example, he finds the verbal system of Berber to be non-tensed
to a higher degree than those of other Afro-Asiatic languages, a fact which is
matched by the occurrence, in Berber, of adjectival verbs as state verbs (or
qualificative verbs) which are typically non-distinct from “real” verbs
(1997: 505). The languages of this Afro-Asiatic area, according to Stassen, can
be situated on a continuum, extending from a completely “verby” encoding (as
in Akkadian or Berber) to a completely “nouny” encoding (as in Modern Hebrew
or Modern Arabic), with several in-between stages. He suggests that this
adjectival drift from verbiness towards nouniness goes hand in hand with an
increase in the tense-orientation of the verbal system (1997: 518). That is, the
concept involved here is tense-prominence rather than merely tensedness.
Stassen (1997) makes use of this notion of tensedness gradation in several
other places in his study in order to account for problematic cases. For example,
in his examination of the languages of Mexico, he finds a correlation between a
gradual shift towards tensedness (or a gradual increase in the prominence of
tense) on the one hand, and a process by which predicate adjectives gradually
separate themselves from verbs on the other (1997: 536). He finds a similar
“gradual shift from an aspectual-modal [prominence] towards a temporal
predominance” among modern Tibetan dialects which is also correlatable with a
gradual separation of predicative adjectives from verbs (for their encoding)
(1997: 550). It can therefore be concluded that tensedness parameter, on the
whole, is a parameter of tense-prominence rather than merely that of tensedness.
It must be noted here that the tensedness parameter provides the basis for
establishing a constraint against the encoding of adjectival predicates as verbs;
it does not provide any basis for claiming that the alternative encoding must be
that of nouns (or of locational predicates) as claimed by Stassen (1997). It is
quite possible for adjectives to be encoded as adjectives in the predicative
position, i.e. as modifiers of a nominal element. This element may be a noun
proper, or a dummy element, such as, for example, a pronominal “agreement”
marker. For example, Thompson (1988) finds adjectives that predicate a property
in English to be occurring either with a copular verb as in they are appropriate,
or as the modifier of a predicate nominal as in he is a good person. Kannada
shows the latter property: several of its adjectives show the constraint that they
require the addition of a pronoun or a pronominal element for predicative use,
whereas no such supporting item is necessary for their modifying use. Examples:
(13) a. doDDa pustaka me˜ji-na me˜le ide

big book table- on is
‘The big book is on the table’
b. me˜ji-na me˜le iruva pustaka doDDa-du
table- on be(. ) book big-it
‘The book which is on the table is big’
Another important point that needs to be noted here is that the occurrence
of aspect or mood as the prominent category does not force verbal encoding
upon adjectives in the way in which non-verbal encoding is forced upon them by
the prominence of tense. That is, it is quite possible, as far as the tensedness
parameter is concerned, for a tense-prominent language, which has its adjectival
predicates encoded non-verbally, to change into an aspect-prominent or mood-
prominent language and continue to have its adjectival predicates remaining with
their non-verbal encoding. There are actually languages showing aspect-promi-
nence and having non-verbal encoding for adjectives. For example, Sanskrit and
most of the Indo-Aryan languages are aspect-prominent, as I have pointed out in
the previous chapter. But they show non-verbal encoding for adjectival predi-
cates. This is especially true of Sanskrit, in which adjectives are indistinguishable
from nouns in both predicative as well as prenominal usages (see Bhat 1994).
7.5 Absence of state verbs
Another interesting characteristic that is shown by some of the tense-prominent

languages is the absence of state verbs. This characteristic is clearly related to
the tendency described in the previous section, namely that adjectival predicates,
which are prototypically state predicates, tend to have non-verbal encoding in
these tense-prominent languages. The prominence that these languages give to
the denotation of the point of time (of the occurrence of an event) in their
encoding of verbs apparently makes it necessary for them to restrict their verbal
encoding to dynamic lexical items. Aspect- and mood-prominent languages, on
the other hand, generally have fairly substantial classes of state verbs. In many
of these languages, the whole word-class of adjectives gets included in the
category of verbs, in the form of a sub-class called state verbs.
The verbal category in Dravidian languages is primarily dynamic. There are
generally only a few state verbs occurring in the peripheral area of the verbal
system in these languages. In Kannada, for example, all the regular verbs are
dynamic, denoting either an action (i.e. having volitional arguments) or a process

(having no volitional arguments). Exceptions to this are (i) the verb iru ‘to be’
which is different from other verbs in showing a three-fold past-present-future
tense distinction (as against the normal past/non-past tense distinction), and also
in being primarily an auxiliary (see 2.3), and (ii) certain modal verbs such as
be˜ku ‘want’, sa˜ku ‘enough’, bahudu ‘posible’ and bal- ‘can’ which do not show
any tense distinctions. Examples:
(i) Two-fold tense-distinction of ordinary verbs:
(14) a. ra˜ju manege ho˜g-utt-añe
Raju home go-.-3:
‘Raju goes home’
b. ra˜ju manege ho˜-d-a
Raju home go--3:
‘Raju went home’
(ii) Three-fold tense-distinction of the verb ‘be’:
(15) a. ra˜ju maney-alli ir-utt-añe
Raju home- be--3:
‘Raju will be at home (when you go there)’
b. ra˜ju maney-alli id-d-añe
Raju home- be--3:
‘Raju is (now) at home’
c. ra˜ju maney-alli id-d-a
Raju home- be--3:
‘Raju was at home’
(iii) Absence of tense distinctions in the use of modal verbs:
(16) nanage ondu pustaka be˜ku
me one book want
‘I want a book’
(17) ra˜ju i˜ mara kadiya-ball-a
Raju this tree cut-can-3:
‘Raju can cut this tree’
A similar constraint is shown by Nama Hottentot, a Khoisan language of
South West Africa, in which tense-marking is obligatory whereas aspect-marking
is optional. Hagman (1973: 160) points out that there are only few state verbs in
this language, with meanings such as ‘know’, ‘be present’, ‘able’, ‘want’,
‘believe’ and ‘feel’. Some of these can be used as auxiliaries and some, only as
auxiliaries. These state verbs can take tense markers, but they require the support
of a copula for this purpose; further, the aspect affixes do not occur with these
state verbs.
As I have mentioned earlier, aspect-prominent and mood-prominent
languages tend to have large classes of state verbs. For example, a number of
verb types in Upriver Halkomelem are stative in nature, such as adjectival verbs
and adverbial verbs; others can be derived from many verbs by adding the stative
s-, usually in combination with the resultative aspect inflection (Galloway 1993).
This is also true of Turkana in which the adjectives are indistinguishable from
stative verbs (Dimmendaal 1983). In the case of mood-prominent languages like
Mao Naga, Khezha, Muna, Chalcatongo Mixtec etc. also, the verbal category
includes large classes of state verbs. In Muna, for example, there are only two
open word classes, namely nouns and verbs, with adjectives occurring as stative
intransitive verbs (van der Berg 1989: 46).
There is also a possible diachronic tendency that is correlatable with this
characteristic. Stassen (1997: 518) points out that in the Afro-Asiatic area, the
drift from aspect-oriented verbal system in the direction of tensedness especially
affected the status of the stative form. In some languages, this form was
reanalysed as a past form, and in others it disappeared from the core system (and
was often replaced by a new, periphrastic and peripheral, nominal verb form with
a stative/perfective meaning). The common outcome of these changes, according
to him, is that the language came to lack a simple verb form for the expression
of stativity.
English is apparently an exception to this generalization in that it does
contain a class of state verbs in spite of its being a tense-prominent language.
That is, it belongs to a lower level of the gradation of tense-prominent languages
as far as this particular characteristic is concerned. Notice, however, that most
of the state verbs of English can also be used as dynamic verbs. Quirk and
Greenbaum (1973: 21) point out, for example, that the verb be of English, even
though a stative as shown by the unacceptability of (18b) given below, can also
be used dynamically, in the progressive, when its complement is dynamic as
shown in (18c):
(18) a. The girl is now a student of a large university.
b. *The girl is now being a student….
c. He is being a nuisance again.
As we have seen in the third chapter, the distinction between state verbs and
dynamic verbs is generally regarded as an aktionsart (or actionality) distinction,
and is closely associated with aspect. There are other related distinctions like
telic-atelic and punctual-durative, which also play a prominent role in the verbal
systems of aspect-prominent languages. They constrain the occurrence as well as
the meaning of aspect markers in various ways. In Navajo, for example, the
perfective-imperfective distinction does not occur in the case of stative verbs.
Among the non-stative verbs also, there is a further distinction between durative
and instantaneous verbs, with the imperfective, when used with the instantaneous,
focusing on its preliminary stages rather than the event itself (Smith 1991: 397,
419). Such aktionsart distinctions appear to have very little influence upon the
verbal systems of tense-prominent languages.
7.6 Variations in the mode of encoding the categories
Bybee (1985: 22) differentiates between tense, aspect and mood on the basis of
their relevance to the verb. Aspect is considered to be exclusively relevant to the
verb as it describes the internal temporal structure of the event denoted by the
verb, whereas tense is considered to be less directly relevant to the verb because
it places a whole situation in time. Mood, on the other hand, is considered to be
even less relevant to the verb than tense as it expresses the speaker’s attitude
towards a situation. This difference in the relevance of these three verbal
categories is considered by Bybee to affect their grammaticalization or encoding
properties. She finds lexically determined allomorphy for aspect to be more
frequent than for tense or mood. She also finds aspect markers to be occurring
closer to the verbal stem than tense and mood markers (see 1985: 37). Her
explanation of these facts is based upon the following three observations:
(i) Generally, the most relevant element occurs closest to the verb.
(ii) A morpheme cannot become fused with a verb unless it is immediately
contiguous to the verb.
(iii) Relevance also influences the actual fusion process, since the elements to be
fused must have conceptual unity.
There are, however, some exceptions to these generalizations in that there
are languages in which tense or mood markers are grammaticalized to a greater
extent than aspect markers, and further, there are also instances in which tense
or mood markers are closer to the verb than aspect markers. The possible
explanation for these exceptional cases, I believe, is that the notion of relevance
can conflict with the notion of prominence. That is, a category that receives the
greatest prominence in a language can get closer to the verb and be the basis of
the most complex set of allomorphic rules (as a result of fusion), even if it is less
relevant to the verb from a language-universal or “semantic” point of view. I will
examine this possibility below from the point of view of both tense-prominent
(7.6.1) as well as mood-prominent (7.6.2) languages. Exceptions to the ordering
hypothesis about these markers may also arise in a similar fashion (see 7.6.3).
7.6.1 Tense-prominent languages
I have already pointed out, in the previous chapter (see 6.2.1), instances of tense-
prominent languages in which the past/non-past tense distinction forms the basis
of the most complex verbal allomorphy. Markers of aspect and mood distinctions
do not show any comparable allomorphic complexity in these languages. They
actually occur in the peripheral area of the verbal system, either in the form of
truncated paradigms or as compound verb constructions. It is tense rather than
aspect that is closest to the verbal base in these languages. The specification of
tense is obligatory for the use of the verb not only in its predicative function, but
also in several other functions like the nominal, adjectival and adverbial, as I
have pointed out in the previous chapter.
The prominence that the category of tense receives in these languages has
the effect of relegating aspect and mood to areas that are away from the verb.
For example, the aspect distinction between bounded event versus unbounded
event (perfective versus imperfective) is expressed in Finnish by the distinction
in the case markers that occur with one of the arguments of the sentence. (The
distinction, however, can also be indicated in this language by the semantics of
the verb or by the use of certain derivative affixes or adverbial phrases.)
Heinämäki (1994) considers this aspect distinction of Finnish to be the semantic
property of the whole sentence rather than merely the property of the verb (see
also Sulkala and Karjalainen 1992). Consider, for example, the following
sentences given by the latter:
(19) a. outi luki kirjan
Outi read () book ()
‘Outi read a book’
b. outi luki kirjaa

Outi read () book ()
‘Outi was reading a book’
Notice that the sentence (19a), in which the object occurs in the accusative case,
denotes a bounded (perfective) event, whereas (19b), in which the object occurs
in the partitive case, denotes an unbounded (imperfective) event.
There is a similar encoding of modal distinction by case markers, occurring
in Havyaka, a dialect of Kannada. The modal meaning of necessity can be
denoted in this dialect by adding the suffix ekku ‘must’ (which has the negative
equivalent aDa ‘do not’ or ‘need not’) directly to the verbal base. One of its
arguments can occur either in the nominative or in the dative; in the former case,
the need or compulsion is external, whereas in the latter case, it is internal.
Examples:
(20) a. añu manege ho˜y-ekku
I () home go-must
‘I must go home’ (I am being forced to go home)
b. ena-ge manege ho˜y-ekku
I- home go-must
‘I must go home’ (I desire to go home)
7.6.2 Mood-prominent languages
As I have pointed out in the previous chapter (see 6.4.1), there are mood-
prominent languages like Chalcatongo Mixtec in which the distinction between
realis and irrealis (or potential) is grammaticalized to a very high degree.
Compared to this modal distinction, tense and aspect distinctions are grammat-
icalized to a lesser extent in these languages. Further, these latter distinctions are
denoted by markers that are attached to stems that are already marked for modal
distinctions. In Chalcatongo Mixtec, for example, there is a temporal prefix a
‘already, now’, and a completive ni both of which are attached to the realis stem;
there is also a repetitive (iterative) na which is attached to the potential stem.
Examples (Macaulay 1996: 62, 74):
(21) a. ni-caà=rí be‘e=ró
-arrive()=1 house=2
‘I arrived at your house’
b. a-ni-kušíní=zó
Already--eat()=1
‘We already ate’
c. káta ‘sing’ na-kata ‘sing again’
kaka ‘walk’ na-kaka ‘walk again’
Kendall (1997) reports that Takelma has two types of verb stems, called
assertive and indeterminate, whose formation (from verbal roots) may involve
processes like reduplication, vowel lengthening, consonant alternation, suffix-
ation, or some combinations of them. By using the assertive stem, the speaker
claims first-hand knowledge of the topic, and within this setting, the assertive is
used for past, present, and immediate future. The indeterminate stem is the base
for all other forms, including nominal derivation. Generally assertive stems are
of C1VC2V pattern and the indeterminate stems are of C1VVC2 pattern. The
stems occur with certain suffixes called “petrified suffixes”, of which -c’ in most
cases denotes perfective aspect and the suffix n denotes durative aspect. There is
also the suffix -k’ which is strongly associated with the perfective. The point,
which is of interest here, is that the modal distinction gets expressed by a
classification of the stem itself, and the aspectual distinction by suffixes that are
attached to the stem. That is, in this mood-prominent language, modal distinction
is fused with the verbal base whereas aspect distinctions are only attached to it in
the form of suffixation.
Another interesting case is that of Wintu, a language spoken in northern
California, and belonging to the Penutian stock. According to Pitkin (1984: 61),
modals, involving a distinction between indicative (real or actual events) and
non-indicative (imperative or non-actual events), are central to the inflectional
system of the verb in this language. The aspectual distinction between generic
and particular, on the other hand, is denoted not in the verb but in the substan-
tives. This distinction, however, resembles aspects of verbs in function to the
extent that the “particular” aspect of nominal forms (marked by the suffix t)
implies finiteness or specificity like the perfective aspect in verbs of other
languages, while the “generic” aspect of nominal forms (marked by the suffix s)
implies an extensiveness or generality like the imperfective and durative of verbs
of other languages (see Pitkin 1984: 106).
7.6.3 Relative order of category markers
There have been several attempts, recently, to establish theoretical claims

regarding the relative order of tense, aspect and mood markers in verbal con-
structions. The general consensus among the scholars concerned is that the aspect
markers tend to occur closest to the verb (either preceding or following it), with
tense markers occurring next to them, and mood markers forming the outermost
constituent. That is, the order, according to these scholars, would be mood-tense-
aspect-verb in the case of VO languages and verb-aspect-tense-mood in the case
of OV languages (see Foley and Van Valin 1984, Bybee 1985, Hengeveld 1989,
Dik 1989, Siewierska 1991, Van Valin 1993).
Foley and Van Valin (1984) consider it possible to derive the above-
mentioned ordering tendency of verbal categories by establishing a correlation
between what they call “levels” of clause structure on the one hand, and the
markers of tense, aspect and mood on the other. They argue (1984: 208) that
there is a need to distinguish between three different levels for clause structure
as shown below:
(i) Nucleus (predicate), which is the innermost layer in the clause,
(ii) Core, consisting of the core arguments of the clause (i.e. arguments which
depend upon the valency of the verb), and
(iii) Periphery, consisting of all the remaining constituents of the clause (such
as its spatio-temporal settings).
The grammatical categories of tense, aspect and mood are considered to be
the “operators” which modify these layers, having scope over one or more of
them. According to Van Valin (1993), nucleus operators modify the action, event
or state itself, without reference to the participant, whereas core operators modify
the relation between a core argument (normally the actor) and the action.
Peripheral (or clausal) operators, on the other hand, may modify either (i) a
single clause or (ii) a sentence (that may contain one or more clauses) as a
whole; and depending upon this factor, clausal operators may be divided into two
distinct subgroups called inner and outer operators.
Foley and Van Valin (1984: 209) argue that the category of aspect is to be
regarded as a nuclear operator because it is not directly concerned with the
participants of the event; it simply expresses the temporal structure of the event,
without reference to anything else. Tense, on the other hand, is to be regarded as
a peripheral operator, because it is concerned with the grounding of the reported
event in the real world, as it expresses the temporal orientation of the event with
regard to the present act of speaking. This difference in the levels to which the
aspectual and temporal operators belong is considered to get reflected in the
relative ordering of the morphemes that represent those operators. A nuclear

operator (aspect) would occur closer to the nucleus (verbal base) as compared to
a peripheral operator (tense). Foley and Van Valin find the category of mood to
be rather problematic in this respect because it appears to divide itself into
(i) Modality (ability, obligation and intention),
(ii) Status (actuality — realis-irrealis),
(iii) Evidential, and
(iv) Illocutionary force
They consider it necessary to assign the first one to the core, and the rest to
the periphery. Van Valin (1993) regards the last two as belonging to the “outer”
layer and the second one (status), along with tense, as belonging to the “inner”
layer of the peripheral (clausal) level.
Notice, however, that similar problems can arise in the case of aspect and
tense categories as well. In the case of tense, for example, there is the distinction
between deictic and non-deictic tenses with the former, but not the latter, being
dependent upon the utterance time (i.e. the time connected with the participants
of the speech act). This distinction can affect the relative ordering of tense,
aspect and mood markers as shown by the fact that in Kannada, non-deictic tense
markers can occur between verbal bases on the one hand, and the aspect or mood
markers on the other, (with the latter being represented by vector and modal
verbs respectively) whereas the deictic tense markers (suffixes) can occur only
after the aspect and mood markers (see 2.3).
In the case of aspect also, there is the well-known distinction between
aktionsart and aspect, with the former denoting, according to some scholars (see
Smith 1986, Platzack 1979) the inherent aspectual characteristics of events or
actions, and the latter denoting the speaker’s viewpoint in this regard. Hengeveld
(1989: 134) finds this distinction to be relevant for determining the relative order
of verbal affixes; the operators which affect changes in the aktionsart distinctions
like stative, telic, momentary etc. (i.e. operators such as the ones which change
statives into dynamics) appear, according to him, to belong to the innermost
layer of the clause (called by him as “predicate operators”), whereas others like
iterative and semelfactive appear to belong, along with tense markers and
markers of objective modality (possible-impossible, realis-irrealis and acceptable-
forbidden) to an outer layer called “predication operators”. The aktionsart
distinctions proper, like stative, telic and momentary, however, are considered by
him to be defining the typology of situations.
What I wish to point out in this connection, however, is that in addition to

the factors mentioned above, the actual ordering of category markers can also be
affected by the relative prominence that languages give to those categories. That
is, even though one expects aspect markers to occur closer to the verb than tense
markers (on the basis of language-universal tendencies), one may find tense
markers to be occurring closer to the verb in the case of some of the tense-
prominent languages as a result of the greater prominence that the languages
attach to tense.
One interesting point that supports this claim is the fact that the most
prominent category in a given language would be grammaticalized to a greater
degree than other categories. That is, the most prominent category would be
represented by inflectional markers, which are closest to the verbal base, whereas
other categories would be represented by particles or clitics, or by auxiliary
verbs, which are less close to the verbal base than inflectional markers. It is
possible for diachronic changes to affect such languages in such a way that the
particles or auxiliaries get reduced to affixes and as a result, there may arise a
situation in which an inflectional affix that marks tense occurs closer to the verb
than one which marks aspect. Such a change has apparently occurred in Hav-
yaka, a dialect of Kannada. It has a suffix kku denoting completeness which
occurs after the prior tense suffix; this suffix has probably developed from an
earlier vector verb (cf. Old Kannada ikku ‘to put’), which is attached to the
converbial form of the main verb, occurring in the non-deictic prior tense.
Examples:
(22) a. avã ada-ra tinn-ali
he it- eat-let
‘Let him eat it’
b. avã ada-ra tin-d-ikk-ali
he it- eat---let
‘Let him finish eating it’
The prior tense suffix is functional in this usage in that it helps to differenti-
ate between modal and aspectual connotations; that is, prior converbs occur when
the vector verbs denote an aspectual meaning, whereas posterior converbs occur
when the they denote a modal meaning. Examples:
(23) a. avã ada-ra tin-du-biDu-gu
he it- eat---may
‘He may eat it off’
b. avã ada-ra tinb-al-akku

he it- eat--allow
‘He is allowed to eat it’
Even when two different categories are represented by the same type of
grammatical elements, the closeness of one of them to the verbal base may
depend upon the category to which the language under consideration attaches
greater prominence. For example, in Lahu (Matisoff 1973), which is a mood-
prominent language, the notion of aspect, directionality, subjective attitude
towards the verbal event, etc. are conveyed by a set of particles that follow the
verb. There are over twenty particles of this type that may be divided into four
subclasses on the basis of their syntactic and semantic properties, especially their
privileges of occurrence. The four groups represent
(i) Directionality (like motion towards or away from the centre of interest)
(ii) Subjective attitudes (like desiderative, experiential and asservative of reality)
(iii) Aspectual distinctions (like completed, durative and inchoative), and
(iv) Different grades of imperatives (suggesting, urging, ordering, etc.)
Matisoff (1973: 315) points out that particles of group I may be followed
within the same verb phrase by particles belonging to each of the other groups,
but particles of group III may also be preceded by those of group II. That is, the
general tendency is apparently for the mood particles denoting subjective attitude
to be closer to the verb than those of aspect.
There are also certain verbal roots, in Lahu, that can immediately precede
or follow the main verb (before taking the above-mentioned particles) in order to
provide additional modal and aspectual connotations. Matisoff (1973: 200) calls
these as “versatile” verbs. The relative ordering of these, however, is rather free.
That is, the versatile verbs (which may denote aspectual meanings like continuat-
ive, completive, durative, etc.) may occur in-between, before or even after other
versatile verbs which denote modal meanings like ability and potentiality (see
Matisoff 1973: 217).
Ladakhi (Tibeto-Burman) is another mood-prominent language in which
verbs take several “orientation” suffixes like reportive, observed, narrative,
experiential and inferential. These may be of different varieties like fact-based,
based on general habit, based on something that is not remembered clearly,
guessed about something that is not seen clearly, etc. Some of these can be
marked for past tense by the suffix pin occurring after them; some of them can
also be marked for continuity by the suffix yin, which precedes the orientation
suffixes. Most of these forms can also show certain modal distinctions like
abilitative, desiderative, permissive and compulsive by the use of certain
additional suffixes which, however, are placed between the verbal base and the
various tense-aspect-orientation suffixes (Koshal 1979: 184). The following
sentence illustrates the occurrence of the continuity (aspect) suffix after the
modal (abilitative) suffix in a verbal form (Koshal 1979: 228):
(24) st66 chu bin-thub-bin-yot-k6k
horse water cross-Abi---
‘The horse had been able to cross the water’
Notice that the abilitative suffix thub occurs between the verb bin ‘cross’ and the
continuative suffix bin (<yin) in (24). There is, however, a problem here,
concerning the status of orientation suffixes; they also occur (with the relevant
aspect and tense markers) in nominal sentences and are translated as different
forms of the verb ‘to be’ (Koshal 1979:185); it is not clear, therefore, whether
they need to be treated as independent auxiliaries in these verbal forms.
Ladakhi also makes a distinction in the case of its verbal forms between
honorific and non-honorific stems either by using distinct verbal roots or by
adding an honorific suffix (dz6d) to them; this suffix occurs closer to the verb
than the tense and aspect suffixes. Examples (Koshal 1979:250).
(25) a. khoe spech6 di-6t
he book read-
‘He reads a book’
b. khonni ch6kspe di-6dz6dd-6t
he () book () read--
‘He reads a book’ (honorific)
However, the so-called secondary modal suffixes (see 4.3) can occur either
before or after this honorific suffix. Examples (Koshal 1979: 254):
(26) a. khonni c6kspe di-6dz6t-thub-6t
he () book () write--able-
‘He can write a book’ (honorific)
b. khonni c6kspe di-thubb-6dz6dd-6t
he () book () write-able--
‘He can write a book’ (honorific)
According to Abraham (1985: 95), particles denoting progressive (continuing
action) and perfect (completed action) aspects in Apatani (a language that

belongs to the Abor-Miri-Dafla branch of the Tibeto-Burman family) follow the
verbal base and the tense markers, whereas the particles denoting completive
(action with a completed object), semi-completive (action with a partly complet-
ed object) and inceptive (action begun) aspects occur between the verbal base
and tense markers. Examples:
(27) a. lu-lyi-do
talk-.-
‘(He) was talking’
b. lu-ne-ku
talk- (.)-
‘(He) has talked’
c. mó lu-ja-ne
he talk-- (.)
‘He finished talking’
d. mi-]ó-ne
do-- (.)
‘(Hi) did half of it’
Hagman (1973: 129) refers to a phenomenon called “tense movement”
occurring in Nama Hottentot; the language appears to be tense-prominent as the
use of tense-markers, but not that of aspect or mood markers, is obligatory.
There are five tense markers in this language, namely kè ‘remote past’, kò
‘recent past’, nií ‘future’, kà ‘indefinite’ (in the sense that the occurrence of the
event itself is indefinite), and an unmarked ‘present’. Generally, tense markers
precede the verbal base, and the perfective aspect marker follows it, but when
tense movement occurs, the tense marker follows the verb and thereby comes in-
between the verb and the perfective aspect marker.
7.7 Differing points of view
I had suggested earlier that languages that give greater prominence to one of the
three verbal categories tend to view concepts belonging to other categories in
terms of their prominent category. They also tend to represent them as facets of
their own prominent category. Consider, for example, the way in which Kannada
encodes aspectual and modal distinctions through its verbal auxiliary system. In
the case of aspectual distinctions like completive, the auxiliary verbs (or vectors)
are attached to the prior tense forms (past converbs) of the main verb, whereas
in that of modal distinctions like ability and possibility, the auxiliary verbs are
attached to the posterior (infinitive) forms of the main verb. That is, aspectual
distinctions are viewed as facets of the non-deictic prior tense whereas modal
distinctions are viewed as facets of the non-deictic posterior tense. Some
aspectual notions like progressive and habitual, on the other hand, are encoded
by attaching the auxiliary to the simultaneous (present converb) form of the main
verb, and are apparently viewed as facets of the non-deictic simultaneous tense.
(i) Completive with prior tense:
(28) a. a˜ hakki ha˜ritu
that bird flew (.3:)
‘That bird flew’
b. a˜ hakki har-i biTTitu
that bird fly-  (.3:)
‘That bird flew off’
c. a˜ hakki ha˜r-i biTTi˜tu
that bird fly-  (.3:)
‘That bird might fly off’
(ii) Abilitative with posterior tense:
d. a˜ hakki ha˜r-a-balludu
that bird fly--can (3:)
‘That bird can fly’
(iii) Progressive with simultaneous tense:
e. a˜ hakki ha˜r-utt-ittu
that bird fly--be (.3:)
‘That bird was flying (when I looked up)’
An interesting consequence of this temporal treatment of aspect and mood
distinctions in Kannada is that the negative word illa ‘not (existential)’, being
regarded as part of the modal category, is attached to the posterior form of the
main verb. That is, the combination of posterior converb and the negative word
provides “past” negative meaning as can be seen in the following sentence:
(29) a. a˜ hakki ha˜r-al-illa

that bird fly--
‘That bird did not fly’
Kannada attaches the negative word illa ‘not’ to a derived nominal (in its
non-past form) in order to indicate the meaning of non-past (or habitual)
negation as shown below:
(29) b. a˜ hakki ha˜ru-v-ud-illa
that bird fly-.-it-not
‘That bird does not fly’
(30) nañu allige ho˜gu-v-ud-illa
I there go-.-it-not
‘I do not/will not go there’
In Ainu, a language isolate of Japan, temporal concepts are viewed from the
aspectual point of view and are represented by aspect markers as their implicat-
ional connotations. Refsing (1994) argues that there are no morphemes in this
language which have the specific function of marking tense in the verbal system;
the language deals with time as aspect or actionality, or through the employment
of other morphemes which carry out other more dominant functions. The
language marks two different aspect distinctions, namely (i) between perfective
and imperfective and (ii) between completed and incompleted. There are also
certain auxiliaries, such as a ‘durative’, moyre ‘be slow’ (or ‘late’), and tunas ‘be
fast’ (or ‘early’) and certain suffixes like pa ‘plural’ (which can denote iterative
meaning). Events, actions or states can be connected together in a sequential or
contemporal relationship with the help of conjunctions like konno ‘sequential’.
These and certain other modal auxiliaries like hortative and intentional provide
the basis for denoting temporal distinctions in this language.
Mood-Prominent languages, on the other hand, view temporal distinctions
as facets of modal distinction. Pitkin (1984) points out, for example, that in
Wintu (belonging to the Wintun family of the Penutian stock), the primary
distinction in the verbal system is the modal one between what is actual and real
(for which the “indicative” stem is used) and what is hypothetical and potential
(for which the “imperative” stem is used) (see 7.6.2). The former may be
translated with both past as well as present forms of English, but the meaning is
primarily modal and not temporal. The case of Wintu is interesting for a
different reason as well. Pitkin (1984) points out that originally, the personal
markers occurring in the verbal forms obligatorily showed only a two-way

contrast between self (first person) and non-self. That is, it was also, basically a
modal (evidential) contrast. This claim is strengthened by the fact that the first
person (indicative) suffix da functions as an evidential suffix as well, indicating
what is particularised as absolutely and reliably known to be true. It cannot occur
in forms that indicate that something is being inferred or judged by the speaker
(Pitkin 1984: 137):
(31) a. hara.da ‘I am walking’
b. coyilake.ni ‘I am drunk (they say)’
A grammatical phenomenon called “flip-flop”, by which verbal forms, when
negated, appear to completely reverse their temporal or modal connotation, has
been reported to occur in certain languages like Copala Trique (Hollenbach 1976)
and Terena (Ekdahl and Grimes 1964). In the latter language, for example,
every verb is in one of two modes, actual or potential. The actual is used for
making definite statements about past or present events or definite observations
about future, whereas the potential is used in imperatives and indefinite predica-
tions. There are two different negative markers in Terena, namely ako and hyoko.
The former is used with the potential verbal forms but the meaning that it
provides appears to be the actual, such as ‘X did not do Y’. The latter (hyoko),
on the other hand, is used with the actual verbal forms, but it appears to give
potential meanings such as that of the negative imperative. Examples (Ekdahl
and Grimes 1964: 268):
(32) a. pih-óp-o
go--
‘He went to where he came from’
b. hyoko pih-ép-o
 go--
‘Do not go where you came from’
(33) a. pih-ép-a
go--
‘Go back to where you came from’
b. ako pih-áp-a
 go--
‘He did not go where he came from’
Honda (1996: 166) suggests that there is a natural explanation for this
pattern: We can view the situations ‘X did Y’ and ‘X will not do Y’ (or ‘Don’t
do Y!’) as two different ways of describing one and the same situation in which
an event has taken place and therefore will not take place again (or need not be
performed). Similarly, we can view the situations ‘X did not do Y’ and ‘X will
do Y’ (or ‘Do Y!’) as two different ways of describing one and the same
situation in which an event has not taken place and has therefore the potential to
take place. What is interesting in the present context, however, is that these
alternative ways of viewing negation derive from the fact that the languages
under consideration, which are mood-prominent, perceive negation from a
“modal” point of view. Notice that a comparable view of “current relevance” (or
of perfect) occurs in Mao Naga as I point out below (see 7.7.1).
Cross-linguistic studies of certain concepts like perfect, future and habitual
have given rise disputes and conflicting generalizations regarding their categorial
status; I believe that the basis of these disputes is the above-mentioned character-
istic of languages, namely that they view non-prominent verbal categories in
terms of their prominent category. I propose to examine some of these disputes
below in order to show that they allow explanations on the basis of the typolog-
ical classification that is under consideration here.
7.7.1 The concept of perfect
The question as to whether the notion of perfect is to be included under the

category of tense or of aspect (or of neither of them) has given rise to a dispute
in linguistics. Traditionally, perfect is regarded as a tense, along with past,
present, future and pluperfect. In Reichenbach’s (1947) theory of tense, for
example, both perfect as well as pluperfect get interpreted, along with past,
present and future, as distinct tense forms. All of them are described in terms of
three distinct parameters, namely an event time, a reference time and a speech
time. Perfect differs from past, according to this formulation, by the fact that
reference time overlaps with event time in the case of past tense, whereas in that
of perfect tense, reference time overlaps with speech time. The three points of
time are distinct from one another in the case of pluperfect tense.
Comrie (1985: 78) argues, however, that this formulation cannot be correct.
The difference between past and perfect does not involve a difference in the
“location in time” of an event; both of them, according to him, denote an event
which is located prior to the present moment (speech time). Consider, for
example, the following two sentences:
(34) a. I lost my penknife.

b. I have lost my penknife.
The crucial difference between the two sentences given above is that the
past form in (34a) merely indicates the time of a given event (namely that it is
earlier than the time of uttering the sentence), whereas the perfect form in (34b)
has an additional meaning, namely that the effect of that event (the loss of
penknife) still persists at the time of speaking. Comrie considers this additional
meaning (‘current relevance’) to be the basic meaning of the perfect form and
hence he argues that it falls outside tense theory.
Salkie (1989: 5), on the other hand, argues that pluperfect also involves a
similar notion in that the event denoted by it needs to be relevant at the time
denoted by the reference point. Consider, for example, the following sentence:
(35) a. He died after he had been ill for a long time.
Notice that the event described by the pluperfect form in (35a) (he had been ill)
has an obvious relevance to the event in the simple past that precedes it. This is
shown by the fact that the pluperfect form cannot be replaced by a past form
without making the resultant sentence bizarre as shown below:
(35) b. *He died after he was ill for a long time.
Salkie (1989: 6) is of the opinion that this notion of relevance is only an
implicational one, and that the basic meaning of perfect or pluperfect is only one
of tense.
There are several other ways in which the notion of perfect is described by
different scholars. For example, Mourelatos (1981: 194) suggests that perfect is
the encoding of the “phase” of time reference, whereas Givón (1984: 278) argues
that perfect is by far the most complex “tense-aspect”, as it involves (i) tense
elements such as time-axis, sequentiality and precedence, (ii) aspectual elements
like perfectivity (accomplishment) and also (iii) certain categories outside the
tense-aspect-mood system, such as the passive and the notion of stativity. Brinton
(1988: 6) suggests further that perfect has aspectual meanings of result, current
relevance, and completion, and other meanings like indefinite past, embedded
past and extended now.
I would like to suggest in this connection that these divergent views about
the nature of perfect derive partly from the complexity of the notion involved,
but partly from the fact that languages differ from one another in viewing or
encoding that notion. My claim is that tense-prominent languages view and
encode it from the point of tense, aspect-prominent languages from that of

aspect, and mood-prominent languages from that of mood. Notice that the
definition of the notion of perfect that is generally given provides us only with
a “temporal” (or tense-oriented) view of this notion. We can also have aspectual
and modal definitions of this notion, as shown below:
The notion of Perfect:
(i) Temporal view: past event with current (present) relevance
(ii) Aspectual view: completed (perfective) event with continuing (imperfect-
ive) relevance
(iii) Modal view: realis event with irrealis relevance (something needs to be
done)
The question as to whether perfect is a tense, aspect or mood (or none of
the three) can therefore be answered only with reference to the language in
which it is being expressed. Depending upon the prominence that the language
attaches to tense, aspect or mood, we can expect the notion to be expressed as a
temporal, aspectual or modal category respectively.
(a) Tense-Prominent languages:
The notion of perfect, and also that of pluperfect and future perfect, is indicated
in Dravidian languages by a combination of non-deictic and deictic tense forms.
These constructions are primarily temporal rather than aspectual or modal both
in their formation as well as usage. In Kannada, for example, a temporal
adverbial can be associated with either of the two tense forms (non-deictic and
deictic) that occur in these constructions. Examples:
(36) nañu ninne mu˜ru gaNTe-ge band-iddeñe
I yesterday three hour- come ()-be ()
*‘I have come at three o’clock yesterday’
(37) a. nañu ninne mu˜ru gaNTe-ge kuLit-idde
I yesterday three hour- sit ()-be ()
‘I have sat at three o’clock yesterday’
b. nañu ninne mu˜ru gaNTe kuLit-idde
I yesterday three hour sit ()-be ()
‘I had sat for three hours yesterday’
Notice that the temporal adverbial in (36) and (37a), namely ninne mu˜ru
gaNTe-ge ‘yesterday at three o’clock’ modifies the prior form of the main verb,
whereas the one in (37b), namely ninne mu˜ru gaNTe ‘yesterday for three hours’
modifies the present form of the auxiliary verb. The claim that the constructions
are primarily temporal in nature also gets supported by the fact that they form
part of a whole paradigm of auxiliary constructions involving prior, simultaneous
and posterior forms of the main verb and past, present and future forms of the
auxiliary verb, as I have shown in the second chapter (see Table 1 in 2.4.1).
Other Dravidian languages also have auxiliary constructions of the above
type, in which the deictic tense forms of the auxiliary verb are attached to the
non-deictic tense forms of the main verb, which are used for denoting perfect
and pluperfect meanings (see, for example, Winfield 1928 on Kui, Krishnamurti
(1969) on Konda, and Hahn 1908 on Kurukh). The auxiliary form is reduced to
a suffix in some of them like Malto (Mahapatra 1979) and Parji (Burrow and
Bhattacharya 1953), but they continue to function as tense forms contrasting with
other tense forms occurring in the language under consideration.
The formation of perfect is similar in Finnish as well. According to Sulkala
and Karjarainen (1992: 297), a compound tense form, derived by joining the
present form of the verb olla ‘to be’ with the past participle of the main verb,
provides the perfect meaning whereas another compound tense form, derived by
joining the past form of the same verb with the past participle of the main verb,
provides the pluperfect meaning. Both are regarded as tense forms in Finnish.
Examples:
(38) he ovat matkusta-nee-t somaliaan
they be (.)-3 travel- ()- Somalia
‘They have travelled to Somalia’
(39) viime kevään minna oli ollut tallinnassa
last spring Minna be (, 3) be ( ) Tallinna
‘Minna had been to Tallinna last spring’
Notice that in both Kannada as well as Finnish, the most prominent concept in
these “perfect” constructions is the past/non-past tense distinction. For example,
the notion of relevance, in the case of present perfect, can only be non-past; it
cannot be past. This is also true of the perfect construction of English.
(b) Aspect-prominent languages
Some of the aspect-prominent languages, on the other hand, appear to derive
their “perfect” forms by combining verbal forms that denote perfective aspect
with the ones that denote imperfective aspect. For example, in Supyire, an
aspect-prominent language, perfect is formed by attaching the auxiliary à to the

perfective form of the verb; this perfect auxiliary, according to Carlson
(1994: 337) derives from an original *mà which seems to have developed from
the imperfective form of a verb meaning ‘come’. Examples:
(40) a. u à pa táñjáà
he  come () yesterday
‘He came yesterday (and is still here)’
b. u ná m-pá tàñjáà
he  (Remote) -come yesterday
‘He came yesterday’
Carlson points out that the typical perfect is perfective in Supyire, as it can
be derived only from a perfective form of the verb; further, there is no equiva-
lent of “perfect progressive” in this language.
In Maltese also, perfect is viewed as a facet of the perfective aspectual
form. According to Fabri (1995), when the perfective form is used on its own,
it can denote past events with or without current relevance; further, when used
with the past auxiliary kein and future auxiliary ikun, it has meanings that
resemble the past perfect and future perfect of English respectively. Examples:
(41) a. fis-s-sitta mikiel kien diga kiel
in--six Michael  already eat ()
‘At six, Michael had already eaten’
b. fis-s-sitta mikiel ikun diga kiel
in--six Michael  already eat ()
‘At six, Michael will have already eaten’
Even in the case of other aspect-prominent languages, perfect occurs as a
part of the aspectual paradigm, and is therefore to be regarded as belonging to
the aspectual category. For example, Mundari distinguishes between a set of four
perfective aspect markers and a set of two imperfective aspect markers; the
former includes the perfect ja (which also has the phasal (ingressive) meaning
‘start’) along with three completive suffixes (see 3.2). Example:
(42) senz-ja-na-ko
go----3
(a) ‘They have just gone’
(b) ‘They have started going’
Similarly, in Tongan, a Polynesian language, perfect is part of the aspectual

system. There is a three-fold contrast in this aspectual system, with perfect
forming part of this contrast as shown below (Pyne 1985):
na’e completed and non-continuing
kuo completed and continuing (perfect)
‘oku non-completed and non-continuing (habitual or future)
(c) Mood-prominent languages
In contrast to the above two types of languages, mood-prominent languages
appear to treat the notion of relevance as a modal concept. For example, Mao
Naga has two different suffixes, namely Ti and oTi, which can be translated in
some of their usages as present perfect and past perfect, as can be seen from the
following pairs of sentences: Examples (Giridhar 1994: 299):
(42) a. ole apru-Ti-e
pot break--
‘The pot has broken’ (It is still in the broken state)
b. ole apru-oTi-e
pot break--
‘The pot had broken’ (It is no longer in the broken state)
(43) a. tasoni ocü vu ta-Ti-e
Tasoni home go.away--
‘Tasoni has gone home (and is not here)’
b tasoni ocü vu ta-oTi-e
Tasoni home go.away--
‘Tasoni had gone home (but is back here)’
However, from a language-internal point of view, the two appear to denote
the notions of “current relevance” and “current irrelevance” respectively of the
event that is being stated, with the notions being not constrained by any temporal
distinctions. This claim is supported by the fact that the two suffixes are different
from the two perfect forms of English on several points. Both of them can
denote events that had occurred prior to some other event, but still, the two differ
from one another in these usages with regard to the relevance that the speaker
attaches to (the result of) the event. Examples (Giridhar 1994: 300):
(44) a. ai idu ni cü he.vue ana ni ozhi-Ti-e

I yesterday your house came but you sleep--
‘I came to your house yesterday but you had slept’
b. ai idu ni cü he.vue ana ni ozhi-oTi-e
I yesterday your house came but you sleep--
‘I came to your house yesterday but you had slept and were up’
The prominence that the language attaches to the modal facet of relevance
(as against its temporal and aspectual facets) gets reflected in the fact that a
speaker can use the suffix Ti in several types of contexts in which the event is
felt by him to be relevant, and the suffix oTi in contexts in which he regards the
event to be not very relevant. For example, oTi is used in contexts in which an
action has been done away with, and therefore nothing more need to be done
about it. Examples (Giridhar 1994: 302):
(45) a. ai pfoyi da-oTi-e
I him beat--
‘I have beaten him’ (so there is no need (for you) to think of
beating him)
b. ai pfo oko hinahi ayi pe-oTi-e
I father story this me told--
‘My father has (already) told me this story’ (so don’t narrate it
to me)
The relevant suffix Ti, on the other hand, is used in contexts in which an
action has been accomplished (contrary to earlier expectations) and is therefore
felt to be very relevant. Examples (Giridhar 1994: 326):
(46) a. kaikho ico coho colo-Ti-e
Kaikho now can hear--
‘Kaikho can hear now (he couldn’t hear earlier)’
b. kolamüinati otu to-Ti-e
plainsman beef eat--
‘The plainsman ate beef’ (he hadn’t earlier)
When used with the irrealis (future) suffix le, the irrelevant suffix oTi
provides the meaning ‘would have’ and the relevant suffix Ti indicates definite-
ness or decisive intention. Examples (Giridhar 1994: 304, 315, 323):
(47) a. pfo khivu-oTi le

he return- 
‘He would have come back’
b. pfo khivu-Ti le
he return- 
‘He will be coming back’
c. pfohi thi-Ti le
he die- 
‘He will surely die’
d. ata izho okho bo-Ti le
we today fish cook- 
‘We intend to cook fish today’
Giridhar (1994) considers these different uses of the suffixes Ti and oTi as
representing different homophonous aspectual and modal suffixes, but I think this
is not necessary. The usages indicate, instead, that the notions of relevance and
irrelevance function as part of the modal category in this language.
(d) Constraints on the occurrence
My claim that the concept of “perfect” is being viewed by languages from the
point of view of their prominent category is supported by the various constraints
shown by these languages on the occurrence of their respective perfect forms.
These constraints appear to depend upon the most prominent category. For
example, Comrie (1976: 62) points out that English allows its perfect to occur
with the progressive (as in I have been singing) but Ancient Greek does not. And
as we have seen above, Supyire also does not have a perfect progressive. On the
other hand, English does not allow the perfect construction to occur with a past
adverbial; that is, in this latter case the constraint is temporal rather than
aspectual. (Kannada allows such an adverbial to occur with its “perfect” con-
struction (see 2.4.1) because it is more phrasal as compared to that of English,
and the adverbial can directly modify the non-finite verb.
7.7.2 The concept of future
There is also a controversy concerning the categorial position of the concept of

future, which is comparable to that of perfect described in the previous section.
Since “future” has the primary function of expressing, grammatically, the tem-
poral location of a given event or state (namely that the event or state is
subsequent to speech time), it clearly forms part of the category of tense. But
several linguists have argued that the concept is at least partly modal in nature
because, according to them, future, unlike past or present, is necessarily specu-
lative. They point out that any statement that we make about future necessarily
includes an element of prediction in the sense that it might be changed by intervening
events, including our own conscious intervention (see Lyons 1977: 677, Fleischman
1982: 14, 24, Comrie 1985: 42, Chung and Timberlake 1985: 204, Dahl 1985: 103).
According to Bybee, Perkins and Pagliuca (1994: 280), future is less a temporal
category and more a category resembling agent-oriented and epistemic modality,
with important temporal implications. They derive this claim from the fact that
future markers appear to develop out of lexical items that denote desire, obliga-
tion, attempt or ability. (The markers may also develop out of constructions
involving movement verbs or temporal adverbs.) Further, they find the focal use
of the future form to be the prediction on the part of the speaker that the event
is to occur after the moment of speech.
This variability in the use of verbal forms between two or more verbal
categories, however, is not a peculiarity of the future tense form. We find similar
variability in the use of other verbal forms as well, such as for example, the
present, which is used quite frequently for denoting the aspectual notion of
habitual or progressive. Bybee, Perkins and Pagliuca (1994: 141) suggest, in fact,
that present may “accurately” be described as present imperfective, (i.e. as an
aspect rather than tense). The past, which is used for denoting completive
(perfective) actions, and the perfective, which is used for denoting past actions,
also show cross-category variations (see Wallace (1982: 202) and Bache
(1995: 267) for a similar opinion). In the case of future, however, there is an
added problem caused by the fact that grammarians have been describing modal
forms as tense forms. That is, languages which actually show a past/non-past
tense distinction, and have an additional subjunctive form, have been described
as having a past-present-future tense distinction (as in the case of some of the
Dravidian languages) leading to the problem of having a future “tense” form
which is unlike other tense forms.
What I am suggesting here is that the notion of future, if it does occur in a
language, would be temporal or modal depending upon the prominence that the
language attaches to the categories of tense and mood respectively. If a tense-
prominent language has a distinct future form, we can expect it to function
primarily as a tense form, with the modal connotations occurring only as its
implications; in the case of a mood-prominent language, on the other hand, such
a form would primarily be a modal one. For example, in a tense-prominent

language like Kannada, future is denoted by the non-past tense form, whereas in
an aspect-prominent language like Navajo, it is denoted by a marker which
occurs along with usitative, iterative, and optative, in contrast with the perfective-
imperfective aspect distinctions. That is, future and optative are part of the
aspectual system in this latter language (see Smith 1991: 428). In a mood-
prominent language like Muna, on the other hand, future is part of the modal
system, and is being represented by the irrealis suffix, contrasting with the realis
suffix, which denotes past or present events (Van der Berg 1989).
7.7.3 The concept of habitual
Another disputed concept of the above type is habitual which, according to

Comrie (1985: 40), can be (i) tense, as it involves location of a situation (event
or state) across a large slice of time, (ii) aspect, as it refers to the internal
temporal contour of a situation — a situation that occupies a large slice of time
— or (iii) mood, as it involves induction from limited observation about the
actual world to generalization about possible worlds. As a result, the question as
to whether habitual is to be regarded as tense, aspect or mood has given rise to
a dispute. According to Dahl (1985: 96), habitual is a major aspect, a claim that
is supported by data from German and Czech, whereas according to Bache
(1982: 61, 1985), habitual is an aktionsart and not an aspect. Swahili, on the
other hand, is claimed to have a habitual tense hu ‘customary, recurring situat-
ions’, contrasting with present and universal tenses, even though, as pointed out
by Comrie (1985: 40), there is no reason for treating it as a distinction in tense
rather than a distinction in aspect or mood.
The cross-linguistic variation that forms the basis of this dispute also
appears to be correlatable with the typological distinction between tense-promi-
nent, aspect-prominent and mood-prominent languages. For example, Kannada
represents the habitual meaning indirectly with the help of its tense forms. It uses
the non-past tense form for denoting the habitual meaning in general, and the
more complex auxiliary constructions for denoting further temporal distinctions
in the habitual. In this latter case, the habitual meaning is denoted by the non-
deictic simultaneous tense form of the verb, and the fact as to whether the
habitual is past, present or future is denoted by the three tense forms of the
auxiliary verb iru ‘to be’. Examples:
(48) avanu ra˜tri mane-ge ho˜guttañe

he night house- goes
‘He (usually) goes home at night’
(49) a. avanu a˜fi˜si-ge aTo˜da-lli ho˜gu-tt-iddañe
he office- auto- go--be ()
‘He goes to office by auto (nowadays)’
b. avanu a˜fi˜si-ge aTo˜da-lli ho˜gu-tt-idda
he office- auto- go--be ()
‘He (used to) go to office by auto’
c. avanu a˜fi˜si-ge aTo˜da-lli ho˜gu-tt-iruttañe
he office- auto- go--be ()
‘He would (usually) be going to office by auto’
Maltese, an aspect-prominent language, on the other hand, uses its aspectual
form for denoting the habitual meaning. Its bare imperfective is understood as
being present habitual (which can, however, express the future if it occurs with
an explicit time adverbial). Its past imperfective form expresses a habit in the
past and corresponds to the used to form of English (Fabri 1995). In contrast to
these, mood-prominent languages appear to view habitual either as realis or
irrealis, providing two different modal facets of habitual events. For example,
according to Macaulay (1996: 45), the uninflected realis stem can be used for
denoting the habitual in Chalcatongo Mixtec, whereas according to Lichtenberk
(1983), Manam uses its definite irrealis form for denoting sequences of custom-
ary, habitual events.
7.7.4 The concept of negation
Negation is generally contrasted with affirmation, providing a polarity distinction

between positive and negative. It is also viewed primarily as a clausal concept,
even though languages do differentiate between clausal and phrasal negations, as
seen, for example, in the use of English not versus un. Aikhenvald and Dixon
(1998) argue, on the basis of the assumption that negation is primarily clausal,
that it would be unaffected by verbal categories like tense, aspect and evidential-
ity (or mood). This claim, however, appears to get contradicted by the fact that
the typological distinction between tense-prominent, aspect-prominent and mood-
prominent languages does have a bearing upon the representation of negation.
For example, Miller (1965) points out that in Acoma, negative is used only
when the speaker himself is involved in the event that is being negated (as a
subject or object). That is, it is used only in contexts in which the speaker can
vouch for the non-occurrence of the event. In other contexts, he uses the
dubitative. We can consider this constraint as an evidentiality (or modal)
constraint on the occurrence of negation. Another interesting case is that of
Muna, referred to earlier (6.4.1), in which negative clauses obligatorily contain
the irrealis verbal forms (Van der Berg 1989). A similar constraint has been
reported to occur in Yimas, a Papua New Guinean language, in which the basic
distinction of the verb is between realis and irrealis. It is only the irrealis that
gets further divided into negative, potential and likely, denoted by the prefixes
ta-, ant- and ka- respectively (Foley 1991: 251). We may contrast this with a
constraint occurring in Russian, namely that the perfective is rarely used in the
negative, which is an aspect-based constraint.
There is also a correlation between the prominence of verbal categories on
the one hand, and the kind of morphemic distinction that is shown in the
representation of negation on the other. For example, Lewo (Early 1994), a
mood-prominent language, makes a two-fold distinction in its negative markers,
between realis negative ve and irrealis negative pe (see 6.4.4). This is also true
of several Tibeto-Burman languages. In Manipuri, for example, the suffix roy is
used to negate a sentence in future tense and the suffix de for negating one in
non-future tense. Examples (Bhat and Ningomba 1997: 248):
(50) a. m6hak l6phoy-du ca-roy
he banana-that eat-:
‘He will not eat that banana’
b. m6hak l6phoy-du ca-de
he banana-that eat--
‘He did not eat that banana’
The influence of mood on negation in a mood-prominent language can also
be exemplified by the occurrence of a phenomenon called “flip-flop” in languag-
es like Copala Trique and Terena, described earlier (see 7.7).
7.8 Foregrounding sequential events
An analysis of extended (especially narrative) texts has revealed that there is

generally a distinction between the language of the actual story line and the
language of supportive material. Hopper (1979a) refers to the former as the

foregrounded part of the narrative and the latter as the backgrounded part. He
points out that the former involves events that succeed one another in the same
order as their succession in the real world. The events are punctual rather than
durative or iterative, and dynamic rather than stative. The latter (backgrounded
part), on the other hand, involves events or states which are not in a sequence to
the foregrounded events. They may be concurrent with them or may involve
other temporal relations, as they provide information concerning the character-
isation, description, motivation, etc of those foregrounded events (Hopper 1979a,
b). Languages generally use one specific verbal form for encoding the fore-
grounded events, and all the other verbal forms for the purpose of encoding the
backgrounded material.
The actual type of verbal form that gets selected for encoding the fore-
grounded events, and also the category to which it belongs, appear to vary from
one language to another. It seems possible to correlate this variation with the
variation in the relative prominence that the languages give to one of the three
verbal categories. That is, languages appear to select a verbal form that belongs
to their most prominent verbal category for encoding the foregrounded material.
For example, Kannada uses a tense form (either the finite past form or the prior
converb) for denoting foregrounded events. Similarly, French uses the simple
(narrative) past form for the same purpose (Hopper 1979b). Aspect-prominent
languages, on the other hand, select an aspectual form, namely the perfective in
contrast to the imperfective, for encoding the foregrounded events. This is the
case with languages like Latin, Greek, Russian and other languages that have a
primary perfective-imperfective aspectual distinction in their verbal system, as
pointed out by Wallace (1982: 208).
In the case of mood-prominent languages also, we find a similar tendency
of selecting a form that belongs to the most prominent category, namely the
realis in contrast to the irrealis, for encoding foregrounded events. For example,
Evans (1995) points out that in Kayardild, the unmarked choice of the narrative
is the verbal form which is inflected for the “actual”, a form which stresses that
the action described has been “instantiated” (speaker knows that the action is
taking place or has taken place). This form is not concerned with locating the
action in time.
It is interesting to note, in this connection, that some of the aspect-promi-
nent languages use an entirely different marker for denoting sequential events of
a narrative that form the foregrounded part. It is possible that, in these languages,
the perfective form does not clearly indicate the sequential nature of the events
that it denotes. For example, Supyire does not use its perfective form for
encoding foregrounded material. Instead, it uses a special narrative/sequential
auxiliary, which does not have past time reference per se. The language also has
a special past tense marker that is used in the very beginning of a narrative
(Carlson 1994: 328).
In spite of the fact that languages differ in their choice of the category for
encoding the foregrounded material, a general consensus appears to have
developed among linguists, namely that the aspectual distinction between
perfective and imperfective is the most important and prototypical category
distinction which forms the basis of this choice in all languages, with the
perfective being used for foregrounding and imperfective for backgrounding. This
appears to be the general assumption of several articles in Hopper (1982) and
also other subsequent publications (see Herring 1993, Bybee, Perkins and
Pagliuca 1994: 126). Herring (1993) makes use of this mistaken assumption as
the basis of a claim that the Old Tamil past tense forms must be regarded as
perfective aspect forms. This claim is based upon a survey of the use of verbal
forms for foregrounding in Old Tamil texts. It was found that past tense forms
(and also prior converbs (or past participles) which, however, are regarded as
“lacking tense indication”) are used for this purpose; since a strong correlation
is claimed to exist between foregrounded events and grammatical marking by
perfective aspect, it is argued that the past tense forms of Old Tamil must be
aspectual and not temporal.
7.9 Paths of grammaticalization
Bybee, Perkins and Pagliuca (1994) examine the lexical sources of tense, aspect
and mood markers occurring in different languages and try to establish paths
along which these lexical items tend to get grammaticalized into inflectional (or
derivational) affixes. They find copular verbs like ‘be’ and ‘have’, dynamic verbs
like ‘finish’, movement verbs and also adverbials like ‘away’, ‘up’ and ‘into’
forming the sources from which resultatives and completives develop. They
claim that these later on give rise to constructions which denote the notion of
perfect (called by them “anterior”), which in its turn, gives rise to either the
perfective aspect or past tense. Similarly, they find auxiliaries like ‘sit’, ‘stand’,
‘lie’, ‘be at’, ‘stay’ and ‘reside’, and locational markers like ‘in’, ‘at’ and ‘on’
forming the sources of the progressive marker, which in its turn, gives rise to
either the imperfective aspect marker or the present tense marker.
Perfective
Perfect (anterior)
Past
Imperfective
Progressive
Present
Bybee, Perkins and Pagliuca find this branching out of perfect and progres-
sive into perfective-past and imperfective-present respectively to be rather
puzzling. They suggest that past can be regarded as a more grammaticalized
version of the perfective (1994: 91); present, on the other hand, is considered by
them to be a specialised imperfective, i.e. one which is restricted to the present
moment (1994: 126, 141).
I would like to suggest an alternative possibility, namely that the two
branchings represent distinct developments in languages that give greater
prominence to aspect and tense respectively. That is, we may assume that
languages that give greater prominence to aspect than to tense develop a
perfective form from an earlier perfect construction and an imperfective form
from an earlier progressive construction, whereas languages that give greater
prominence to tense than to aspect develop past and present forms directly from
their perfect and progressive constructions respectively. It would not be necessary,
according to this formulation, to assume that the latter type of languages necessari-
ly undergo a change of the former type as an earlier stage of development.
Some of the Dravidian languages appear to support this claim; for example,
Old Kannada had the suffix dapa for denoting past tense (which was formed
from an earlier periphrastic construction involving a prior converb and a non-past
form of the verb a˜gu ‘to become’); it has been replaced in Modern Kannada by
the suffix utt; this latter suffix also occurs as a non-deictic simultaneous suffix
(see 2.4.1), and probably has progressive meaning as its basis. That is, the path
of grammaticalization used by Kannada for moving from progressive to present
apparently does not involve the denotation of the imperfective aspect as an
intermediary stage.
Another interesting aspect of this grammaticalization process is that the
aspect-prominent and mood-prominent languages show distinct tendencies of
change when they develop temporal distinctions. In the case of aspect-prominent

languages, we generally find a two-way past/non-past distinction or a three-way
past-present-future distinction developing from an earlier perfective-imperfective
distinction, which correlates with the fact that from an aspectual point of view
present and future appear as imperfective and past as perfective. For example,
the Semitic languages had originally an aspectual system, which distinguished
between perfective and imperfective events, but in some of the modern languages
this has changed into a temporal system in which the distinction is between past
and present (see Stassen 1997: 493–519).
In the case of mood-prominent languages, on the other hand, the general
tendency is to develop primarily a future/non-future distinction. In Manipuri, for
example, the tense distinction is between future and non-future (see 2.2.1), which
has apparently developed from an earlier realis-irrealis modal distinction. We can
account for this tendency by the fact that from a modal point of view both past
as well as present events appear as realis events and only the future events
appear as irrealis events. Another interesting point is that in the case of mood-
prominent languages, the general tendency is to develop elaborate systems of
past and future tenses (involving distinction of temporal distance). This is also a
“modal” tendency in the sense that these distinctions of temporal distance
correspond to the various evidentiality (or judgement) distinctions in these tenses.
One can be more certain about a past event that took place today than one that
took place yesterday or last year.
7.10 Conclusion
I hope to have listed, in the foregoing sections, sufficient number of characteris-

tics that can be correlated with the classification of languages into tense-promi-
nent, aspect-prominent and mood-prominent types, to make such a typological
classification worth pursuing. The characteristics appear to indicate that the most
prominent verbal category influences not only the structure of verbal forms, but
also other areas of grammar such as nominalizations, case marking, adjectival
and adverbial systems, etc. It appears to influence even the world-view of the
speakers, and hence we might regard this typology as providing evidence in
support of the Sapir-Whorf hypothesis.
References
Abraham, P.T. 1978. Relative Clauses in Malayalam. Annamalainagar: Anna-

malai University.
Abraham, P.T. 1985. Apatani Grammar. Mysore: Central Institute of Indian
Languages.
Aikhenvald, A.Y. and Dixon, R.M.W. 1998. “Dependencies between grammati-
cal systems”. Language 74: 56–80.
Allott, A. 1965. “Categories for the description of the verbal syntagma in
Burmese”. Lingua 15: 283–309.
Annamalai, E. 1985. Dynamics of Verbal Extension in Tamil. Trivandrum:
Dravidian Linguistics Association.
Bache, C. 1982. “Aspect and aktionsart: Towards a semantic distinction”.
Journal of Linguistics 18: 57–72.
Bache, C. 1985. Verbal Aspect. Odense: Odense University.
Bache, C. 1995. The Study of Aspect, Tense and Action: Towards a Theory of the
Semantics of Grammatical Categories. Frankfurt am Main: Deter Lang.
Barnes, J. 1984. “Evidentials in Tuyuca verb”. International Journal of American
Linguistics: 50, 255–71.
Bauman, J.J. 1979. “An historical perspective on ergativity in Tibeto-Burman”.
In Ergativity, F. Plank (ed), 419–33. London: Academic Press.
Benjamin, G. 1976. “An outline of Tamir Grammar”. In Austroasiatic Studies,
Vol. I, P.N. Jenner, L.C. Thompson and S. Sarosta, (eds.), 129–187.
Hawaii: University of Hawaii.
Bennett, M. and Partee, B. 1978. Toward the Logic of Tense and Aspect in
English. Indiana University Linguistic Club.
Berntsen, M. and Nimbkar, J. 1975. A Marathi Reference Grammar. Philadelphia:
South Asia Regional Studies, University of Pennsylvania.
Berntsen, M. and Nimbkar, J. 1982. Marathi Structural Patterns. Book One.
Pune: Granthali and Marathi Abhyas Parishad.
186 REFERENCES
Bhat, D.N.S. 1989. Wh-words. A report submitted to the University Grants

Commission, New Delhi.
Bhat, D.N.S. 1991. Grammatical Relations: Evidence against Their Necessity and
Universality. London: Routledge.
Bhat, D.N.S. 1994 (a). The Adjectival Category: Differentiating and Identifying
Characteristics. Amsterdam: John Benjamins.
Bhat, D.N.S. 1994b. Tense, Aspect and Mood. A report submitted to the Universi-
ty Grants Commission, New Delhi.
Bhat, D.N.S. and Ningomba, M.S. 1997. Manipuri Grammar. Munich: Lincom
Europa.
Bhatia, T.K. 1993. Panjabi: A Cognitive-Descriptive Grammar. London: Rout-
ledge.
Binnick, R.I. 1978. “On the underlying tense of deverbatives”. Language 54:
289–299.
Bodding, P.O. 1929. Materials for a Santali Grammar. Benagaria: Santali
Mission.
Borg, A. and Azzopardi-Alexander, M 1997. Maltese. London: Routledge.
Brinton, L.J. 1988. The Development of English Aspectual Systems. Cambridge:
Cambridge University.
Burrow, T. 1955. The Sanskrit Language. London: Faber and Faber.
Burrow, T. and Bhattacharya, S. 1953. The Parji Language. Hertford: Stephen
Austen and Sons.
Bybee, J.L. 1985. Morphology. Amsterdam: John Benjamins.
Bybee, J.L., Perkins, R and Pagliuca, W. 1994. The Evolution of Grammar:
Tense, Aspect and Modality in the Languages of the World. Chicago: Univer-
sity of Chicago.
Bybee, J.L. and Fleishman, S. (eds.). 1995. Modality in Grammar and Discourse.
Amsterdam: John Benjamins.
Cardona, G. 1965. A Gujarati Reference Grammar. Philadephia: University of
Pennsylvania.
Carlson, G.N. 1983. “Marking constituents”. In Linguistic Categories: Auxiliaries
and Related Puzzles, F. Heny and B. Richards (eds.), 69–98. Dordrecht: D.
Reidel.
Carlson, R. 1994. A Grammar of Supyre. Berlin: Mouton de Gryuter.
Chafe, W. 1960 “Seneca morphology”. International Journal of American
Linguistics 26: 11–22, 123–129.
REFERENCES 187
Chafe, W. 1990. “Uses of defocusing pronominal prefixes in Caddo”. Anthropo-

logical Linguistics 32: 57–68.
Chafe, W. 1995. “The realis-irrealis distinction in Caddo, the North Iroquoian
languages and English”. In J. Bybee & S. Fleishman (eds.), 349–365.
Chomsky, N. 1981. Lectures on Government and Binding. Dordrecht: Foris.
Chung, S. and Timberlake, A. 1985. “Tense, aspect and mood”. In Language
Typology and Syntactic Description, part III, T. Shopen (ed.), 202–258.
Cambridge: Cambridge University.
Comrie, B. 1976. Aspect. Cambridge: Cambridge University.
Comrie, B. 1981. Language Universals and Linguistic Typology. Chicago:
University of Chicago.
Comrie, B. 1985. Tense. Cambridge: Cambridge University.
Comrie, B. and Thompson, S. 1985. “Lexical nominalization”. In Language
Typology and Syntactic Description, Vol. III: Grammatical Categories and
the Lexicon, T. Shopen (ed.), 349–398. Cambridge: Cambridge University.
Dahl, O. 1985. Tense and Aspect Systems. Oxford: Basil Blackwell.
DasGupta, K. 1971. An Introduction to the Nocte Language. Shillong: North East
Frontier Agency.
De Angulo, J. and Freeland, L.S. 1931. “The Achumawi language”. International
Journal of American Linguistics 6, 77–120.
Declerck, R. 1986. “From Reichenbach (1947) to Comrie (1985) and beyond”.
Lingua 70: 305–64.
Declerck, R. 1991. Tense in English: Its Structure and Use in Discourse. London:
Routledge.
Deeney, J. 1975. Ho grammar and vocabulary. Singhbhum: Xavier Ho.
DeLancey, S. 1981. “An interpretation of split ergativity and related patterns”.
Language 57: 626–57.
DeLancey, S. 1983. “Tangut and Tibeto-Burman morphology”. Linguistics of the
Tibeto-Burman Area 7 (2): 100–8.
Dik, S.C. 1989. The Theory of Functional Grammar,  I: The Structure of the
Clause. Dordrecht: Foris.
Dimmendaal, G,J. 1983. The Turkana Language. Dordrecht: Foris.
Dixon, R.M.W. 1979. “Ergativity”. Language 55: 59–138.
Dixon. R.M.W. 1994. Ergativity. Cambridge: Cambridge University.
Early, R. 1994. “Lewo”. In Typological studies in negation, P. Kahrel & R. Van
den Berg (eds.), 65–92. Amsterdam: John Benjamins.
188 REFERENCES
Eguzkitza, A. 1987. Topics on the System of Basque and Romance. Reproduced by

the Indiana University Linguistic Club.
Ekdahl, M. and Grimes, J.E. 1964. “Terena verb inflection”. International
Journal of American Linguistics 30: 261–268.
Enc, M. 1987. “Anchoring conditions for tense”. Linguistic Inquiry 18: 633–57.
Evans, N.D. 1995. A Grammar of Kayardild. Berlin: Mouton de Gruyter.
Fabri, R. 1995. “The tense and aspect system of Maltese”. In Tense systems in
European languages, Vol II, R. Thieroff (ed.), 327–343. Tubingen: Max
Niemeyer Verlag.
Fleischman, S. 1982. The Future in Thought and Language: Diachronic Evidence
from Romance. Cambridge: Cambridge University.
Foley, W.A. 1991. The Yimas Language of New Guinea. Stanford: Stanford
University.
Foley, W.A. and Van Valin, R.D. Jr. 1984. Functional Syntax and Universal
Grammar. Cambridge: Cambridge University.
Foster, M.K. 1985. “The language of tense, mood and aspect in Northern
Iroquoian descriptions”. International Journal of American Indian Languag-
es, 52: 65–72.
Galloway, B.D. 1993. A Grammar of Upriver Halkomelm. Berkeley: University
of California.
Giridhar, P.P. 1994. Mao Naga Grammar. Mysore: Central Institute of Indian
Languages.
Givón, T. 1984. Syntax, vol I. Amsterdam: John Benjamins.
Givón, T. 1990. Syntax, vol II. Amsterdam: John Benjamins.
Hackman, G.J. 1976. An Integrated Analysis of the Hindi Tense and Aspect
Systems. University of Illinois dissertation.
Hagman, R.S. 1973. Nama Hottentot Grammar. Columbia University Doctoral
dissertation.
Hahn, F. 1908. Grammar of the Kurukh Language. Reprinted by Mittal Publica-
tions, Delhi.
Hale, K. 1983. “Warlpiri and the grammar of non-configurational languages”.
Natural Language and Linguistic Theory 1: 1–43.
Haspelmath, M. 1993. A Grammar of Lezgian. Berlin: Mouton de Gruyter.
Heinämäki, D. 1994. “Aspect as boundedness in Finnish”. In Tense, aspect and
action: Emphrical and theoretical contributions to language typology, C.
Bache, Hans Basbøl and C.E.Lindberg (eds.), 207–233. Berlin: Mouton de
Gruyter.
REFERENCES 189
Hengeveld, K. 1989. “Layers and operators in functional grammar”. Journal of

linguistics 25: 127–157.
Herring, S.C. 1993. “Functional stability in language change: the evolution of
tense and aspect in Tamil”. Studies in Language 17: 3l3–341.
Hoffmann, J. 1903. Mundari Grammar. Calcutta: Bengal Secretariat.
Hollenbach, B.E. 1976. “Tense-Negation interplay in Copala Trique”. Interna-
tional Journal of American Linguistics 42, 126–32.
Honda, I. 1996. Negation: A Cross-linguistic Study. University of New york
(Buffalo) Ph.D. dissertation.
Hopper, P.J. 1979a. “Aspect and foregrounding in discourse”. In Discourse and
Syntax: Syntax and Semantics 12, T. Givón (ed.), 213–241. New York:
Academic Press.
Hopper, P.J. 1979b. “Some observations on the typology of focus and aspect in
narrative language”. Studies in Language 3: 37–60.
Hopper, P.J. (ed.) 1982. Tense-aspect: Between Semantics and Pragmatics.
Hopper, P.J. and Thompson, S.A. 1984. “The discourse base for lexical catego-
ries in universal grammar”. Language 60: 703–752.
Kachru, Y. 1980. Aspects of Hindi Grammar. New Delhi: Manohar.
Kapfo, 1993. A Descriptive Analysis of Khezha. University of Mysore disserta-
tion.
Kellogg, S.H. 1938. A Grammar of the Hindi Language. London: Routledge and
Kegan paul.
Kendall, D.L. 1997. “The Takelma verb: toward Proto-Takelma-Kalapuyan”.
International Journal of American Linguistics 63:1–17.
Kiparsky, P. 1968. “Tense and mood in Indo-European syntax”. Foundations of
Language 4: 30–57.
Kittel, F. 1903. A grammar of Kannada language. Mangalore: Basel Mission
Book and Track Depository.
Klein, H.E.M. 1979. “Noun classifiers in Toba”. In Ethnolinguistics: Boas, Sapir
and Whorf Revisited, M. Mathiot (ed), 85–95. The Hague: Mouton.
Koshal, S. 1979. Ladakhi Grammar. Delhi: Motilal Banarsidass.
Krishnamurti, Bh. 1969. Konda or Kubi: A Dravidian Language. Hyderabad:
Tribal and Cultural Research Institute.
Kurylowicz, J. 1964. The Inflectional Categories of Indo-Euro pean. Heidel-
berg: Carl Winter.
190 REFERENCES
Lefebvere, C. & Muysken, P. 1988. Mixed Categories: Nominalization in Que-

chua. Dordrecht: Kluwer.
Lehmann, C. 1985 (l995). Thoughts on Grammaticalization. Munich: Lincom
Europa.
Lehmann, W.P. 1974. Proto-Indo-European Syntax. Austin: University of Texas.
Lehmann, T. 1989. A Grammar of Modern Tamil. Pondicherry: Pondicherry
Institute of Linguistics and Culture.
Li, C.N. 1991. “The aspectual system of Hmong”. Studies in Language, 15:
25–58.
Lichtenberk, F. 1983. A Grammar of Manam. Hawaii: University of Hawaii.
Lyons, J. 1977. Semantics, vol 2. Cambridge: Cambridge University.
Macaulay, M. 1996. A Grammar of Chalcatongo Mixtec. Berkeley: University of
California.
Mahapatra, B.P. 1979. Malto: An Ethnosemantic Study. Mysore: Central Institute
of Indian Languages.
Masica, C.P. 1991. The Indo-Aryan language. Cambridge: Cambridge University.
Matisoff, J.A. 1973. The Grammar of Lahu. Berkeley: University of California.
Miller, W.R. 1965. Acoma Grammar and Texts. Berkeley: University of Califor-
nia.
Mithun, M. 1995. “On the reality of irreality”. In J.L. Bybee et al (eds.),
368–388.
Mourelatos, A.P.D. 1981. “Events, processes and states”. In Tense and aspect:
syntax and semantics 14, P. Tedeschi and A. Zaenen (eds.), 191–212. New
York: Academic Press.
Newman, S. 1954. “Semantic problems in grammatical systems and lexemes: a
search for method”. In Language in culture, H. Hoijer (ed.), 82–91. Chica-
go: Chicago University.
Noonan, M. 1992. A Grammar of Lango. Berlin: Mouton de Gruyter.
Okell, J. 1969. A Reference Grammar of Colloquial Burmese. London: Oxford
University.
Osada, J. 1992. A Reference Grammar of Mundari. Tokyo: Institute for the Study
of Languages and Cultures of Asia and Africa.
Palmer, F.R. 1986. Mood and Modality. Cambridge: Cambridge University.
Paramasivam, K. 1979. “Effectivity and causativity in Tamil”. International
Journal of Dravidian Linguistics 8: 71–151.
Pitkin, H. 1984. Wintu Grammar. Berkeley: University of California.
Plank, F. (ed.). 1979. Ergativity. London: Academic Press.
REFERENCES 191
Platzack, C. 1979. The Semantic Interpretation of Aspect and Aktionsart.

Dordrecht: Foris Publications.
Payne, J.R. (1985), “Negation”. In Shopen, T. (ed.), Language Typology and
Semantic Description, Volume 1. Cambridge: Cambridge University Press,
197–242.
Quirk, R. and Greenbaum, S. 1973. A University Grammar of English. London:
Longmans.
Refsing, K. 1994. “Tense, aspect and directionality in Ainu language”. In Tense,
aspect and action, K. Bache, H. Hasbel and C-E. Lindberg (eds.), 311–321.
Berlin: Mouton de Gruyter.
Reichenbach, H. 1947. Elements of Symbolic Logic. London: Colier-Macmillan.
Rennison, J.R. 1996. Koromfe. London: Routledge.
Roberts, J.R. 1990. “Modality in Amele and other Papuan languages”. Journal of
Linguistics 26: 363–401.
Saksena, A. 1982. Topics in the Analysis of Causatives with an Account of Hindi
Paradigms. Berkeley: University of California.
Salkie, R. 1989. “Perfect and pluperfect”. Journal of Linguistics 25: 1–34.
Sapir, E. 1921. Language. New York: Harcourt Brace.
Sastri, G.D.P. 1984. Mishmi Grammar. Mysore: Central Institute of Indian
Languages.
Schiffman, H.F. 1969. A Transformational Grammar of the Tamil Aspectual
System. University of Chicago Ph.D. dissertation.
Schiffman, H.F. 1979. A Reference Grammar of Spoken Kannada US Department
of Health, Education and Welfare.
Scholberg, H.C. 1940. Concise Grammar of the Hindi Language. Madras: Oxford
University.
Siewierska, A. 1991. Functional Grammar. London: Routledge.
Smith, C.S. 1978. “The syntax and interpretation of temporal expressions in
English”. Linguistics and Philosophy 2: 43–99.
Smith, C.S. 1983. “A theory of aspectual choice”. Language 59: 479–501.
Smith, C.S. 1986. “A speaker-based approach to aspect”. Linguistics and Philoso-
phy 9: 97–115
Smith, C.S. 1991. The parameter of Aspect. Dordrecht: Kluwer.
Sridhar, M.V. 1980. Sema Grammar. Mysore: Central Institute of Indian Lan-
guages.
Sridhar, S.N. 1990. Kannada. London: Routledge.
Stassen, L. 1985. Comparison and Universal Grammar.Oxford: Basil Blackwell.
192 REFERENCES
Stassen, L. 1997. Intransitive Predication. Oxford: The Clarendon Press.

Subrahmanyam, P.S. 1971. Dravidian verb Morphology. Annamalainagar:
Annamalai University.
Sulkala, H. and Karjalainen, M. 1992. Finnish. London: Routledge.
Taylor, C. 1985. Nkore-Kiga. London: Croom-Helm.
Thompson, S.A. 1988. “A discourse approach to the category of Adjective”. In
Explaining Language Universals, J.A. Hawkins (ed.), 167–185. Oxford:
Basil Blackwell.
Trask, R.L. 1979. “On the origins of ergativity”. In Ergativity, F. Frank (ed.),
385–404. London: Academic Press.
Unbegaum, B.D. 1957. Russian Grammar. Oxford: The Clarendon Press.
Van der Auwera, J. and O Baoil, D.P. (eds.) 1998. Adverbial Constructions in the
Languages of Europe. Berlin: Mouton de Gruyter.
Van der Berg, R. 1989. A Grammar of the Muna Language. Dordrecht: Foris.
Van Valin, R.D. Jr. 1993. “A synopsis of Role and Reference grammar”. In
Advances in Role and Reference Grammar, R.D. Van Valin Jr. (ed.).
Wallace, S. 1982. “Figure and ground: The interrelationships of linguistic
categories”. In Tense-aspect, P.J. Hoppe (ed.), 201–223. Amsterdam: John
Benjamins.
Watkins, L.J. 1984. A Grammar of Kiowa. Lincoln: University of Nebraska.
Whitney, W.D. 1888. Sanskrit Grammar. Cambridge: Harward University.
Wetzer, H. 1996. The Typology of Adjectival Predication. Berlin: Mouton de
Gruyter.
Winfield, W.W. 1928. A Grammar of the Kui Language. Calcutta: Asiatic Society
of Bengal.
Index
A Bauman, 148–9
abilitative, 78 benefactive, 77, 86
ablaut, 124 Benjamin, 127
Abraham, 114, 163 Bennett, 17
absolute/relative, 14 Berber, 151
Achumawi, 136 Berntsen, 46, 49, 83, 147
Acoma, 178 Bhat, 3, 5, 7, 50, 57, 92, 101, 142,
adjectival participle, 112–6 152, 179
adjectival predicate, 149–52 Bhatia, 127,129
adjective, 4, 92 Bhattacharya, 171
Aikhenvald, 178 Binnick, 143
Ainu, 166 Bodding, 55
Akkadian, 151 Borg, 49
aktionsart, 45, 57, 58–9, 160 Brinton, 45, 58, 94, 169
Allott, 132 Burmese, 100, 131
Amele, 137, 144 Burrow, 124, 171
Ancient Greek, 175 Bybee, 6, 96, 99, 130, 155, 159,
Annamalai, 109–11 176, 181–2
Anywa, 29
Apatani, 164
Arabic, 151 C
aspectual adverbs, 36, 60–1 Caddo, 135–7, 139
Azzopardi-Alexander, 49 Cardona, 128, 130
Carlson, 46, 95, 122, 149, 171, 181
case marker, 97, 138, 156
B Chafe, 132,135, 139
Bache, 45, 58–9, 176–7 Chalcantongo Mixtec, 65, 154, 157,
Barnes, 71 178
Basque, 146 Chomsky, 92
194 INDEX
Chung, 93, 176 E

clause-chaining, 137 Early, 139, 179
Comrie, 14, 30, 41, 46, 49, 50, 53, Eguzkitza, 146
93, 100, 131–2, 142–3, 145, Ekdahl, 167
168–9, 175–7 Enc, 21
conditional, 118 English, 2, 7, 17, 26, 29, 30, 41,
configurational, 92 44, 49, 64, 70, 76, 98–9, 116,
contentives, 95–6 120, 123, 131, 146, 150, 154,
contrafactual, 21 166, 173, 175, 178
converb, 116–8 epistemic, 63–75, 77, 79
Copala Trique, 167, 179 Eurotype, 5
copula verb, 23, 74–5 Evans, 97, 134, 137–8, 180
correlatable characteristics, 102 evidential, 35, 63–4, 65, 70–5, 148,
counterfactual, 119 179
current relevance, 169
Czech, 177
F
Fabri, 126, 172, 178
D factual, 132
Dahl, 6, 46, 53, 176–7 Finnish, 120, 156, 171, 97
Das Gupta, 148 Fleischman, 176
De Angulo, 136 flip-flop, 167, 179
decategorization, 142–3 Foley, 159, 179
DeChicchis, 100, 132–3 foregrounding, 179–81
Declerck, 30, 36 Foster, 99, 100, 132
Deeney, 55 Freeland, 136
defocusing, 139 French, 180
deictic/non-deictic, 14, 15–31, 34, function words, 95–6
36–8, 45, 58, 64–5, 92, 94, future, 121, 123
112 160, 165 future, as modal, 175–7
DeLancey, 148 future/non-future, 17–20, 67–8,
deontic, 75–9, 82–3 70–1, 132
diachronic 28, , 100–1
differentiating, 1–2, 4–6
Dik, 44, 49, 159 G
Dimmendaal, 154 Galloway, 154
Dixon, 92, 145–7, 178 German, 45, 64, 120, 177
INDEX 195
Giridhar, 56, 6–70, 77–8, 84–6, I

173–5 idealised language, 6, 8–9, 92
Givón, 93, 95, 169 illocutionary force, 63
gradation, 104 imperative, 64, 79–82
grammatical relations, 3–4 indicative, 72–3
grammaticalization, 6, 50, 95–6, infinitive 24
104–7, 120–5, 130–1, 133–5, interrogative, 64, 79–82
161, 181–2 interrogative, polar vs. content, 79,
Greek, 180 80
Greenbaum, 154
Greenberg, 92
Grimes, 167 J
Gujarati, 128, 144 Jeanne, 47–8
judgement, 63–5, 70–5
H
habitual, 21–2, 53, 65, 68, 122, K
125–6, 129, 177–8 Kachru, 55
Hackman, 41, 50, 121 Kannada, 5, 13, 17–8, 20–6, 36–42,
Hagman, 120, 153, 164 60–1, 64, 84, 97, 105–119,
Hahn, 171 143, 151–3, 160–1, 164, 170,
Hale, 92 175, 177, 180–1
Haspelmath, 144 Kapfo, 80–1
Havyaka, 111, 119, 144, 157, 161 Karjalainen, 120, 156, 171
Hebrew, 151 Kayardild, 96, 134, 137–8, 180
Hengeveld, 159–60 Kellogg, 44, 99, 129
Herring, 181 Kendall, 158
Hindi, 5, 41–4, 50, 55, 59, 100, Khezha, 80–1, 154
128–9 Kiowa, 46, 48, 54
Hmong, 49, 100, 121 Kiparsky, 123
Ho, 55 Kittel, 22
Hoffmann, 55 Klein, 94
Hollenbach, 167 Konda, 112, 171
Honda, 167–8 Konkani, 83
Hopi, 47–8, 50 Koromfe, 122–3
Hopper, 180–1, 142 Koshal, 72, 76–7, 163
Krishnamurti, 112, 171
196 INDEX
Kui, 114, 171 Mayrhofer, 124

Kurukh, 15–7, 31–2, 105, 171 Melotki, 48
Kurylowicz, 95–6 Miller, 178
Kuvi, 114 Mishmi, 32, 35
Mithun, 65
Mourelatos, 169
L Muna, 66, 133, 136, 154, 177, 179
Ladakhi, 64, 71–7, 162–3 Mundari, 55, 172
Lahu, 162 Muysken, 29
Lango, 125
Latin, 180
Lefebvre, 29 N
Lehmann, C., 95–6 Nama Hottentot, 120, 153, 164
Lehmann, T. 28, 117 narrative, 73
Lehmann,W.P, 92, 123 Navajo, 155, 177
Lewo, 139, 179 negation, 33, 67, 81, 113–4, 139,
Lezgian, 144 165, 167, 178–9
Li, 49, 99, 100, 121 New Logic, 4
Lichtenberk, 136, 178 Newman, 95
Lyons, 176 Ngore-Kiga, 14
Nimbkar, 46, 49, 83, 147
Ningomba, 50, 57, 101, 179
M Nocte, 148
Macaulay, 65, 157, 178 Noonan, 48, 125
Mahapatra, 171 noun-verb distinction, 3–4
Malayalam, 114
Maltese, 126, 172, 178
Malto, 105, 108, 171 O
Manam, 135, 137, 178 obligatoriness, 96, 107–9, 135
Manipuri, 5, 17, 18–9, 50–2, 57, Okell, 131–2
101, 179, 183 Old Kannada, 182
Mao Naga, 7, 19, 56, 59, 65, 67, Old Tamil, 181
70, 77–8, 84–7, 154, 168, operators, nuclear, 159–60
173–5 ordering tendency, 159
Marathi, 46, 54, 83, 100, 147 Osada, 55
Masica, 49–50, 54, 83, 121
Matisoff, 162
INDEX 197
P Quechua, 29
Pagliuca, 6, 99, 130, 176, 181–2 Quirk, 154
Palmer, 63, 76, 93–4
Paramasivam, 5, 105
Parji, 171 R
Partee, 17 Rangila, 127,129
past/non-past, 16–20, 114, 120, 123 realis-irrealis, 65–6, 93, 100, 105,
Pengo, 114 131–40
perfect, 7, 25, 29–31, 49, 91, 121, recategorization, 142
168–75 reduplication, 55, 57, 127, 158
perfective-imperfective, 45–9, 65, reference point, 16, 39
93, 97, 100, 120–30, 144, 155, reference time, 29 30, 168
166 Refsing, 99, 166
Perkins, 6, 99, 130, 176, 181–2 Reh, 29
permissive, 78 Reichenbach, 29–30, 168
perposive, 24 relative distance, 14, 31–5, 40, 85
person, 73, 135 relevance, 68, 78, 155
pervasiveness, 96, 112–9, 127–130, relevance-irrelevance, 173–5
137 Rennison, 122–3
phasal aspect, 49–52 restrictive, 5–6
Pitkin, 57, 158, 166–7 Roberts, 137, 144
Plank, 92 Russian, 179–80
Platzack, 58, 94, 160
polite, 85–6
potential, 65 S
prior-posterior, 117–8 Saksena, 5
progressive, 68, 122, 128 Salkie, 169
prospective, 125 Sango, 48
Proto-Dravidian, 104–5 Sanskrit, 4, 121, 123–5
Proto-Tibeto-Burman, 148 Santali, 55
Punjabi, 54, 127,129 Sapir, 95
Pyne, 173 Sastri, 32
satellite picture, 98
Schiffman, 22, 28
Q Scholberg, 128
Q’eqchi’, 100, 132 Sema, 80
quantificational aspect, 53–7 Seneca, 132
198 INDEX
Sherpa, 147, 149 Tongan, 173

Siewierska, 44, 159 translation, misleading, 44, 47, 99
Sindhi, 83 Trask, 59, 147
Smith, 36, 45, 58, 155, 160, 177 Tulu, 34–5, 83, 92, 116, 119
speaker attitude, 32 Turkana, 154
split ergative, 145–9 Tuyuca, 65, 71
Sridhar, M.V., 80
Sridhar, S.N., 22
Stassen, 2, 149–52, 154, 183 U
state verb, 152–5 Unbegaum, 179
stative, 34 universal paths, 6
Steever, 105 universalistic, 1–2, 4–6
Subrahmanyam, 24, 104, 113–4 Upriver Halkomelem, 154
Sulkala, 120, 156, 171 utterance time, 29, 30
Supyire, 46, 122, 149, 171, 175,
181
Swahili, 177 V
systematicity, 109–112, 126–7, Van der Auwera, 5
136–7 Van der Berg, 66, 133–4, 154, 177,
179
Van Valin, 93, 159–60
T vector verb, 107–8, 161
Takelma, 158 viewpoint, 58–9, 64
Tamil, 5, 21, 28–9, 91, 109–111,
113, 117
Tamir, 126 W
Tangut, 147–8 Wallace, 176, 180
Taylor, 14 Watkins, 46, 48, 54
temporal adverbs, 115, 170 Wetzer, 150
temporal bias, 132 Winfield, 171
tensedness, 149–52 Wintu, 57, 158, 166–7
tenseless language, 15 word order, 92
Terena, 167, 179
Thompson, 142–3, 151
Tibetan, 151 Y
Timberlake, 93, 176 Yimas, 179
Toba, 94
In the STUDIES IN LANGUAGE COMPANION SERIES (SLCS) the following volumes
have been published thus far or are scheduled for publication:
1. ABRAHAM, Werner (ed.): Valence, Semantic Case, and Grammatical Relations. Work-
shop studies prepared for the 12th Conference of Linguistics, Vienna, August 29th to
September 3rd, 1977. Amsterdam, 1978.
2. ANWAR, Mohamed Sami: BE and Equational Sentences in Egyptian Colloquial Arabic.
Amsterdam, 1979.
3. MALKIEL, Yakov: From Particular to General Linguistics. Selected Essays 1965-1978.
With an introd. by the author + indices. Amsterdam, 1983.
4. LLOYD, Albert L.: Anatomy of the Verb: The Gothic Verb as a Model for a Unified Theory
of Aspect, Actional Types, and Verbal Velocity. Amsterdam, 1979.
5. HAIMAN, John: Hua: A Papuan Language of the Eastern Highlands of New Guinea.
Amsterdam, 1980.
6. VAGO, Robert (ed.): Issues in Vowel Harmony. Proceedings of the CUNY Linguistics
Conference on Vowel Harmony (May 14, 1977). Amsterdam, 1980.
7. PARRET, H., J. VERSCHUEREN, M. SBISÀ (eds): Possibilities and Limitations of
Pragmatics. Proceedings of the Conference on Pragmatics, Urbino, July 8-14, 1979. Am-
sterdam, 1981.
8. BARTH, E.M. & J.L. MARTENS (eds): Argumentation: Approaches to Theory Formation.
Containing the Contributions to the Groningen Conference on the Theory of Argumenta-
tion, Groningen, October 1978. Amsterdam, 1982.
9. LANG, Ewald: The Semantics of Coordination. Amsterdam, 1984.(English transl. by John
Pheby from the German orig. edition “Semantik der koordinativen Verknüpfung”, Berlin,
1977.)
10. DRESSLER, Wolfgang U., Willi MAYERTHALER, Oswald PANAGL & Wolfgang U.
WURZEL: Leitmotifs in Natural Morphology. Amsterdam, 1987.
11. PANHUIS, Dirk G.J.: The Communicative Perspective in the Sentence: A Study of Latin
Word Order. Amsterdam, 1982.
12. PINKSTER, Harm (ed.): Latin Linguistics and Linguistic Theory. Proceedings of the 1st
Intern. Coll. on Latin Linguistics, Amsterdam, April 1981. Amsterdam, 1983.
13. REESINK, G.: Structures and their Functions in Usan. Amsterdam, 1987.
14. BENSON, Morton, Evelyn BENSON & Robert ILSON: Lexicographic Description of
English. Amsterdam, 1986.
15. JUSTICE, David: The Semantics of Form in Arabic, in the mirror of European languages.
Amsterdam, 1987.
16. CONTE, M.E., J.S. PETÖFI, and E. SÖZER (eds): Text and Discourse Connectedness.
Amsterdam/Philadelphia, 1989.
17. CALBOLI, Gualtiero (ed.): Subordination and other Topics in Latin. Proceedings of the
Third Colloquium on Latin Linguistics, Bologna, 1-5 April 1985. Amsterdam/Philadelphia,
1989.
18. WIERZBICKA, Anna: The Semantics of Grammar. Amsterdam/Philadelphia, 1988.
19. BLUST, Robert A.: Austronesian Root Theory. An Essay on the Limits of Morphology.
20. VERHAAR, John W.M. (ed.): Melanesian Pidgin and Tok Pisin. Proceedings of the First
International Conference on Pidgins and Creoles on Melanesia. Amsterdam/Philadelphia,
1990.
21. COLEMAN, Robert (ed.): New Studies in Latin Linguistics. Proceedings of the 4th
International Colloquium on Latin Linguistics, Cambridge, April 1987. Amsterdam/
Philadelphia, 1991.
22. McGREGOR, William: A Functional Grammar of Gooniyandi. Amsterdam/Philadelphia,
1990.
23. COMRIE, Bernard and Maria POLINSKY (eds): Causatives and Transitivity. Amster-
dam/Philadelphia, 1993.
24. BHAT, D.N.S. The Adjectival Category. Criteria for differentiation and identification.
25. GODDARD, Cliff and Anna WIERZBICKA (eds): Semantics and Lexical Universals.
Theory and empirical findings. Amsterdam/Philadelphia, 1994.
26. LIMA, Susan D., Roberta L. CORRIGAN and Gregory K. IVERSON (eds): The Reality of
Linguistic Rules. Amsterdam/Philadelphia, 1994.
27. ABRAHAM, Werner, T. GIVÓN and Sandra A. THOMPSON (eds): Discourse Grammar
and Typology. Amsterdam/Philadelphia, 1995.
28. HERMAN, József: Linguistic Studies on Latin: Selected papers from the 6th international
colloquium on Latin linguistics, Budapest, 2-27 March, 1991. Amsterdam/Philadelphia,
1994.
29. ENGBERG-PEDERSEN, Elisabeth et al. (eds): Content, Expression and Structure. Studies
in Danish functional grammar. Amsterdam/Philadelphia, 1996.
30. HUFFMAN, Alan: The Categories of Grammar. French lui and le. Amsterdam/Philadel-
phia, 1997.
31. WANNER, Leo (ed.): Lexical Functions in Lexicography and Natural Language Processing.
32. FRAJZYNGIER, Zygmunt: Grammaticalization of the Complex Sentence. A case study in
Chadic. Amsterdam/Philadelphia, 1996.
33. VELAZQUEZ-CASTILLO, Maura: The Grammar of Possession. Inalienability, incorpora-
tion and possessor ascension in Guaraní. Amsterdam/Philadelphia, 1996.
34. HATAV, Galia: The Semantics of Aspect and Modality. Evidence from English and Biblical
Hebrew. Amsterdam/Philadelphia, 1997.
35. MATSUMOTO, Yoshiko: Noun-Modifying Constructions in Japanese. A frame semantic
approach. Amsterdam/Philadelphia, 1997.
36. KAMIO, Akio (ed.): Directions in Functional Linguistics. Amsterdam/Philadelphia, 1997.
37. HARVEY, Mark and Nicholas REID (eds): Nominal Classification in Aboriginal Australia.
38. HACKING, Jane F.: Coding the Hypothetical. A Comparative Typology of Conditionals in
Russian and Macedonian. Amsterdam/Philadelphia, 1998.
39. WANNER, Leo (ed.): Recent Trends in Meaning-Text Theory. Amsterdam/Philadelphia,
1997.
40. BIRNER, Betty and Gregory WARD: Information Status and Noncanonical Word Order in
English. Amsterdam/Philadelphia, 1998.
41. DARNELL, Michael, Edith MORAVSCIK, Michael NOONAN, Frederick NEWMEYER
and Kathleen WHEATLY (eds): Functionalism and Formalism in Linguistics. Volume I:
General papers. Amsterdam/Philadelphia, 1999.
42. DARNELL, Michael, Edith MORAVSCIK, Michael NOONAN, Frederick NEWMEYER
and Kathleen WHEATLY (eds): Functionalism and Formalism in Linguistics. Volume II:
Case studies. Amsterdam/Philadelphia, 1999.
43. OLBERTZ, Hella, Kees HENGEVELD and Jesús Sánchez GARCÍA (eds): The Structure of
the Lexicon in Functional Grammar. Amsterdam/Philadelphia, 1998.
44. HANNAY, Mike and A. Machtelt BOLKESTEIN (eds): Functional Grammar and Verbal
Interaction. 1998.
45. COLLINS, Peter and David LEE (eds): The Clause in English. In honour of Rodney
Huddleston. 1999.
46. YAMAMOTO, Mutsumi: Animacy and Reference. A cognitive approach to corpus linguis-
tics. 1999.
47. BRINTON, Laurel J. and Minoji AKIMOTO (eds): ollocational and Idiomatic Aspects of
Composite Predicates in the History of English. 1999.
48. MANNEY, Linda Joyce: Middle Voice in Modern Greek. Meaning and function of an
inflectional category. 2000.
49. BHAT, D.N.S.: The Prominence of Tense, Aspect and Mood. 1999.
50. ABRAHAM, Werner and Leonid KULIKOV (eds): Transitivity, Causativity, and TAM.
In honour of Vladimir Nedjalkov. 1999.
51. ZIEGELER, Debra: Hypothetical Modality. Grammaticalisation in an L2 dialect. 2000.
52. TORRES CACOULLOS, Rena: Grammaticization, Synchronic Variation, and Language
Contact.A study of Spanish progressive -ndo constructions. 2000.
53. FISCHER, Olga, Anette ROSENBACH and Dieter STEIN (eds.): Pathways of Change.
Grammaticalization in English. 2000.
54. DAHL, Östen and Maria KOPTJEVSKAJA TAMM (eds.): Circum-Baltic Languages.
Volume 1: Past and Present. n.y.p.
55. DAHL, Östen and Maria KOPTJEVSKAJA TAMM (eds.): Circum-Baltic Languages.
Volume 2: Grammar and Typology. n.y.p.
56. FAARLUND, Jan Terje (ed.): Grammatical Relations in Change. 2001.
57. MEL’C UK, Igor: Communicative Organization in Natural Language. The semantic-
communicative structure of sentences. n.y.p.
58. MAYLOR, Brian Roger: Lexical Template Morphology. Change of state and the verbal
prefixes in German. n.y.p.

Untitled

Uploaded by

Copyright:

Available Formats

Untitled

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Untitled

Uploaded by

Copyright:

Available Formats

THE PROMINENCE OF

TENSE, ASPECT AND MOOD

The Prominence of Tense, Aspect and Mood

JOHN BENJAMINS PUBLISHING COMPANY

of American National Standard for Information Sciences — Permanence of

Library of Congress Cataloging-in-Publication Data

7.8 Foregrounding sequential events . . . . . . . . . . . . . . . . . . . . . . . . . . 179

are applicable to only some languages, the commonly used approach is to

1 first person  instrumental

1.1 Universalistic and Differentiating Approaches

same type. The occurrence of differences in case-marking and verbal agreement

case markers and semantic relations is too complex in these languages to

universality of noun-verb distinction in spite of the adverse findings of the

about similarities between familiar and non-familiar languages. He should give

1.2 Nature of the present study

such cross-linguistic variations makes it rather difficult to characterise the three

1.3 Organisation of the monograph

languages can be viewed as representing one or the other of these categories in

We may use the terms “deictic” and “non-deictic” in order to differentiate

of a given event. Additional information regarding its temporal location is

2.2 Deictic tense

2.2.1 Constraint on present tense

The fact that an event can be thought of as occurring before, simultaneously, or

2.3 Non-deictic tense

2.3.1 Justifying the distinction in Kannada

Grammarians of Kannada do not differentiate between deictic and non-deictic

b. na˜nu haNN-annu kattarisu-tta˜ be˜yisu-tt-e˜ne

past present future

b. mu˜ru gaNTe-ge a˜ta manege ho˜gu-tt-id-d-a˜ne

Another interesting point that supports the above-mentioned claim is that,

simultaneous ho˜gu-tta˜ ‘while going’

2.3.2 Need for the distinction

I have proposed in this chapter that it would be useful to differentiate between

category of aspect and tense-prominent languages including it under the category

2.4 Distance from the reference point

We have so far examined the occurrence of two different parameters in the

This remoteness distinction, however, is restricted to action verbs. Stative

2.5 Use of temporal adverbials

2.5.1 Deictic/non-deictic distinction

The deictic/non-deictic distinction occurring among temporal adverbials can be

(39) a. avanu na˜Le baru-tta˜ne

b. avanu ivattu baru-tta˜ne

2.5.2 Positional distinctions

Temporal adverbials also show a distinction between prior, simultaneous and

(47) a. na˜nu mu˜rar-inda na˜lka-ra oLage ho˜gutte˜ne

2.5.3 Remoteness distinctions

2.5.4 Conflict with tense

Temporal adverbials are found to conflict with tense primarily in sentences in

(48) a. na˜nu na˜Le baru-tt-e˜ne

(50) a. vah kal bambai gaya˜ tha˜

3.2 Perfective and imperfective

(5) a. to tikDe jat 6sel

‘scatter’ cala calala-ta

represent the corresponding aspectual distinctions (see Masica 1991). Li

3.3 Phasal aspects

An event may have a beginning and an end, a middle portion (continuing or

(ii) Verbs of stopping

3.4 Quantificational aspects

3.5 Situational and viewpoint aspects

As I had mentioned earlier, some scholars claim that a distinction needs to be