Professional Documents
Culture Documents
(Cambridge Books Online) Subbārāo, Kārumūri V - South Asian Languages - A Syntactic Typology-Cambridge University Press (2012)
(Cambridge Books Online) Subbārāo, Kārumūri V - South Asian Languages - A Syntactic Typology-Cambridge University Press (2012)
(Cambridge Books Online) Subbārāo, Kārumūri V - South Asian Languages - A Syntactic Typology-Cambridge University Press (2012)
South Asian languages are rich in linguistic diversity and number. This book
explores the similarities and differences of sixty languages from the four
different language families (Austro-Asiatic, Dravidian, Indo-Aryan [Indo-
European], and Tibeto-Burman [Sino-Tibetan]). It focuses on the syntactic
typology of these languages and the high degree of syntactic convergence,
with special reference to the notion of “India as a linguistic area.” Several
areas of current theoretical interest such as anaphora, control theory, case and
agreement, relative clauses, and the significance of thematic roles in grammar
are discussed. The analysis presented has significant implications for current
theories of syntax, verbal semantics, first and second language acquisition,
structural language typology, and historical linguistics. The book will be of
interest to linguists working on the description of South Asian languages, as
well as syntacticians wishing to discover more about the common structure
of languages within this region.
Kārumūri V. Subbārāo
cambridge university press
Cambridge, New York, Melbourne, Madrid, Cape Town,
Singapore, São Paulo, Delhi, Tokyo, Mexico City
Cambridge University Press
The Edinburgh Building, Cambridge CB2 8RU, UK
Published in the United States of America by Cambridge University Press, New York
www.cambridge.org
Information on this title: www.cambridge.org/9780521861489
C K. V. Subbārāo 2012
A catalogue record for this publication is available from the British Library
Preface page xi
Acknowledgements xii
List of abbreviations xvi
1 Introduction 1
1.1 Languages of South Asia 1
1.2 Aim of the book 2
1.3 Linguistic theory, language universals and language typology 3
1.4 Inductive and deductive approaches to language analysis 4
1.5 Relevance of linguistic theory: some illustrations 5
1.6 The framework 8
1.7 The data 8
1.8 Relevance of the present work 11
1.9 The richness and complexity of the data and their relevance 11
1.10 Contributions of cross-linguistic studies to our understanding of the nature
of language 12
1.11 The binding link for all chapters 16
vii
viii Contents
6 Complementation 193
6.1 Introduction 193
6.2 The position of occurrence of the COMP and its functions 196
6.3 Structural differences between an IC and an FC 204
6.4 Some crucial issues concerning FCs and ICs 212
6.5 Position of occurrence of an IC- and an FC-clause and the Case Resistance
Principle (CRP) 218
6.6 Narrow Scope and Wide Scope 223
6.7 Arguments against Rightward Extraposition 239
6.8 Conclusion 240
Appendix – the Final-over-Final Constraint (FOFC) and diachronic change 240
Notes 313
Bibliography 336
Author index 356
Language index 359
General index 362
This book takes forward the work I have been doing for the last thirty years
on the syntactic typology and convergence in South Asian languages. South
Asian languages, many of which are not even well studied and documented,
belong to different genetic groups. Though they share common structural traits,
they exhibit their own unique properties, providing a great opportunity for
investigation. In this book several new phenomena that I found in South Asian
languages have been analyzed in the Principles and Parameters approach.
The fact that there is no work available that compares and contrasts the rich
variety of data of these languages in a suitable theoretical framework was a
major motivating factor to write a book of this nature. The bulk of the empirical
evidence that I gathered, and the insights gained, further encouraged me to
pursue this work.
Writing this book – which required an enormous effort in understanding the
intricacies of the data and the structure of many South Asian languages – the
issues involved, and attempting to present them collectively within the limita-
tions of space and time, entailed an enormous amount of work and organization.
The findings presented in here will be of great utility to scholars working on
individual languages, to typologists as well as to theoretical linguists interested
in the study of language universals and parametric variation.
The range of topics chosen, the number of issues discussed and the amount
of data provided to substantiate our arguments will enable the reader to gain a
comprehensive view of the intricacies of the syntax of South Asian languages.
This book will demonstrate the relevance and significance of structural typol-
ogy that derives its insights from a formal theoretical framework, and employs
it fruitfully for typological work.
This work, I hope, will provide an incentive for further research on the
syntactic typology of each language family and on individual languages of the
subcontinent.
There is a detailed discussion of a number of issues that are relevant to the
study of the syntactic typology of South Asian languages available at the URL
www.cambridge.org/subbarao.
xi
Acknowledgements
It is my great pleasure and privilege first to thank all those who have been work-
ing on grammars of individual languages which contribute to the understanding
of human language.
The late Professor P. B. Pandit encouraged me to initiate work on syntactic
typology in 1974, at a time when the field was less known. The late Professor
James (Jim) McCawley also persuaded me to write a book on the syntactic
typology of South Asian languages. Jim would have been delighted to see the
present volume. The late Professor W. F. Lehman, too, evinced a keen interest
in my work. I am grateful to Professors Pandit, Lehman and McCawley for
their encouragement. Professor A. R. Kelkar, too, has been very enthusiastic
about the work on the typology of South Asian languages.
In 2005 Professor Peri Bhaskararao of the Institute for the Languages and
Cultures of Asia and Africa, Tokyo University of Foreign Studies (ILCAA),
invited me there as Visiting Professor for a year and provided me an intellectual
environment with unlimited infrastructural facilities in which to pursue my
research goals. It was at the Institute in Tokyo that the first draft of this book
was written. But for his generous help, active support and sincere commitment
towards an academic cause, the book would never have been completed. I
greatly benefited from the discussions on Telugu syntax with him. I owe a lot
to him and to his Institute, which cannot be expressed in words. I am deeply
indebted to Bhaskararao.
My greatest admiration and deep gratitude goes to Alice Davison, who has
been a friend, philosopher and guide throughout the writing of this book.
She went through all the chapters meticulously, making insightful comments,
offering helpful criticism and concrete suggestions, resulting in a balanced
approach between empirical data and the theoretical issues involved. Though
her professional commitments were too numerous, she took time off to read
the manuscript at various stages of its preparation, which I greatly appreciate
and admire. Her perseverance, commitment to promote an academic pursuit,
encouragement and intellectual contribution have enhanced the quality of this
work.
xii
Acknowledgements xiii
I am grateful to Rajesh Bhatt for going through the entire manuscript very
carefully during the pre-final stages and for his constructive criticisms, his help-
ful, incisive and insightful suggestions and comments, and for raising thought-
provoking questions, which have bettered the present work. His suggestions
helped me in radically reorganizing chapters 2 and 8, which has enhanced
conceptual clarity. His unassuming style of functioning during discussions has
really impressed me. It has been fun discussing syntactic issues with Rajesh. I
am deeply indebted to him for all his help.
I am grateful to Probal Dasgupta for having patiently read the chapters and
for commenting on them in terms of data and the theoretical and empirical
issues involved. His insights into Bangla syntax in particular and syntax in
general, with solid arguments, enabled me to proceed in the right direction. I
am deeply indebted to him too.
Both Alice and Probal have been a constant source of encouragement by
repeatedly telling me that this kind of typological work has an immense value.
Thanks are due to Peter Hook, who commented on the data and content
of several chapters and helped me to think more on the issues in Indo-Aryan
languages. His insights into the functioning of these languages helped me a
great deal. His sound comments proved to be very helpful.
Colin (Nick) Masica has also been a great source of encouragement. His vast
and in-depth knowledge of the structure of Indo-Aryan and Dravidian languages
and his valuable insights have been of immense value for this book. He read
some chapters and commented on them. Whenever I needed some help, Nick
was always there with advice and suggestions. It was his monumental work on
the typology of South Asian languages and Indo-Aryan that encouraged me to
follow it up further.
I am grateful to Martin Everaert for his keen interest, support and help at
various stages of the preparation of the manuscript and also for his valuable
inputs and comments on some chapters of the book.
I am thankful to John Peterson, who provided me with his published and
unpublished work on Kharia and who went through many chapters carefully
and commented on them meticulously. I greatly benefited from his knowledge
of Kharia syntax. He was always very enthusiastic to discuss issues of Kharia
syntax, which helped me enormously.
Josef Bayer, Frans Plank, Hans H. Hock and Halldor A. Sigurdsson read
some or most of the chapters and provided helpful insights and significant
inputs to the work. I am grateful to all of them for their valuable suggestions
and support in many ways.
There are many scholars who helped me in different ways by sending rel-
evant material and answering questions on data and analysis. They include:
P. J. Mistry, Bh. Krishnamurti, Annie Montaut, Kashi Wali, Hans H. Hock,
Barbara Lust, James W. Gair, Miriam Butt, K. A. Jayaseelan, R. Amritavalli,
xiv Acknowledgements
Gupta and Rama Kant Agnihotri for Punjabi; Teresa Sundi, Bir Singh Sinku
and C. B. Deogam for Ho; Revd. Naphurai Jamatia, Binoy Deb Burman, San-
jeeb Deb Burman and Atul Deb Burman for Kokborok; Bhupen Narzaree,
Birlang Narzaree, Chandan Borgoyari and Ganga Brahma for Bodo; Parichita
Langthasa and Ron Kemprai for Dimasa; Jugen Pegu, Shikha Dele and Darika
Pegu for Mising; Rosemary Achumi, P. Toiho Sema, Hekani Jakhula, Ruth
Murru, Vikali Arkha and Toli Yeptho for Sema; Pangersenla for Ao; and Matt
Shibatani for Japanese and Korean. Their help is acknowledged with great
appreciation.
I’m grateful to G. Uma Maheshwar Rao for the enlightening discussions
we have had on several issues of Telugu syntax and for providing me a good
academic environment during my tenure as Radhakrishnan Chair Professor at
the University of Hyderabad. J. Prabhakarao helped me in many ways. I am
deeply indebted to both of them. I am thankful to my fellow faculty members
and students at the Centre for Applied Linguistics who made my stay of one
year enjoyable and memorable. My special thanks are due to Mayuri Dilip,
Kenei Mepthii-o, Parameshwari Krishnamurthy and Mimi Kevichüsa for their
help during the final stages of this work.
My friends Om Prakash Arora, Harbir Arora, Amit Mitra, R. C. Cowsik,
Yugal Swaroop, Alka Goel, and Satpal and Chhaya Gupta have always provided
their unstinted support in many ways. Rajesh Kumar’s and Lalita Dhareshwar’s
help in a variety of areas is fondly acknowledged.
I’d like to thank U. Sreehari too, for his untiring effort in patiently editing
the manuscript and for helping me in many ways.
Thanks to Sunita Topo for her caring service.
I greatly benefited from descriptive grammars written on South Asian lan-
guages, and the Routledge series of grammatical descriptions of South Asian
languages.
Many thanks to Andrew Winnard of Cambridge University Press, with whom
I first discussed the project and who got it initiated. I appreciate his patience
in bearing with the delays that occurred due to a variety of factors. Thanks are
also due to Sarah Green for her keen interest and patience at various stages of
preparation of the manuscript, and to Tom O’Reilly for all his help. I wish to
express my deep sense of gratitude to Leigh Mueller, the book’s copy-editor,
for having patiently read the manuscript and for making stylistic changes with
meticulous care, paying attention to the minutest possible detail. Her insightful
editorial style is appreciable. It was fun working with Leigh.
My wife Sarala, our daughters Sudhita, Varnita and Vidita, our sons-in-law
Srini, Sandeep and Sanyam, and our grandsons Srijay, Ishaan and Sahaj have
always been a great source of inspiration. Sarala’s support throughout my
academic career deserves special mention. I record my deep appreciation and
sense of gratitude to her in particular and to our family in general. I am thankful
to my extended family for their support and encouragement.
Abbreviations
AA Austro-Asiatic
abl ablative
abs absolutive case
acc accusative
adjr adjectivalizer
adv adverb
AGR/agr agreement
cem collaborative effort marker
caus causative
CFC contrastive focus clitic
cl classifier
com comitative
COMP/comp complementizer
cond conditional
conj conjunction
corr correlative
CP Complement Phrase
cp conjunctive participle
cpm conjunctive participial marker
CRP Case Resistance Principle
dat dative
DD1 definite determiner 1
DD2 definite determiner 2
decl declarative
def definite marker
def agr default agreement
det determiner
dis disjunction
DM/dm deictic marker
DO direct object
Dom differential object marking
DP Determiner Phrase
xvi
List of abbreviations xvii
DR Dravidian
DSC dative subject construction
dub dubitative
ECM Exceptional Case Marking
EHRC Externally Headed Relative Clause
emph emphatic marker
epen epenthetic
epm emotive predicate morpheme
EPP Extended Prejection Principle
erg ergative
ev evidentiality
excl exclusive
f feminine
FC final (right peripheral) complementizer
fin finite
foc focus
FOFC Final-over-Final Constraint
fut future
gen genitive
GF Grammatical Function
gp mkr group marker
gpm generic possession marker
hon honorific
IA Indo-Aryan
IC initial (left peripheral) complementizer
IHRC Internally Headed Relative Clause
imp imperative
imperf imperfect
inch inchoative
incl inclusive
ind indicative
INFL inflection
instr instrumental
intr intransitive
IO indirect object
IP Inflection Phrase
LF Logical Form
loc locative
m masculine
midhon mid-honorific
mkr marker
neg negative
xviii List of abbreviations
rel relative
S Subject
S-O-V Subject-Object-Verb
S-V-O Subject-Verb-Object
s singular
SA South Asian / South Asia
sam subject agreement marker
self aff self-affective
self ben self-benefactive
Spec Specifier
sub subject
sup mkr superlative marker
TB Tibeto-Burman
TEC Thematic Eligibility Condition
thematic S thematic Sentence
TP Tense Phrase
tr transitive
V Verb
VP/vP Verb Phrase
VR verbal reflexive
VREC verbal reciprocal
V-S-O Verb-Subject-Object
1 Introduction
1
2 Introduction
ways, demonstrating that each language retains its identity in spite of intense
contact with other languages either due to multilingualism, or areal contact,
or both. The similarities found in SALs are due to the fact that all the lan-
guages, except for a few such as Khasi, Rymbai and Pnar (Mon-Khmer), are
verb-final. As one would expect, verb-final languages have several features
in common. Our work further demonstrates how languages that are contigu-
ously located in an area can affect each other in a radical and in a principled
manner.
demonstrate that this fact can be explained by invoking Head to Head movement
(see chapter 3 for details).
When the verbal anaphor occurs in the embedded clause, long-distance
binding is blocked in all the languages.
Thus, the occurrence of either a nominal complex anaphor or a verbal anaphor
or both blocks long-distance binding.
Leaving the case of Marathi (IA) aside for a moment, this can be neatly
formulated in terms of the following parameter.
Parameter 1:
[−/+ complex nominal anaphor] / [−/+ verbal anaphor] directly correlates
with [+/− long-distance binding].
That is, the occurrence of the complex nominal anaphor and verbal anaphor are
negatively specified for the phenomenon of long-distance binding.
(ii) The second example concerns the occurrence of the verbal reflexive
and the locative PP. A locative PP may either be a subcategorized or a non-
subcategorized argument of a predicate. When an anaphor occurs in a sub-
categorized locative PP position, the occurrence of the verbal reflexive is
obligatory – otherwise, it is not (Lust et al. 2000). Thus, the notion of
subcategorization2 helps us in providing an explanation of the occurrence
of the verbal reflexive.
This can be formulated in terms of the following parameter.
Parameter 2:
[+/− the occurrence of a verbal anaphor] directly correlates with [+/− the
occurrence of a nominal anaphor in a subcategorized locative PP position]
(see chapter 3 for details).
That is, wide-scope reading of the wh-expression is negatively specified for the
feature of initial complementizer.
(iv) The fourth example comes from relative clauses in SALs. Some Tibeto-
Burman languages (Angami [a.k.a. Tenyidie], Sema, Mizo, Manipuri, Sangtam
and Konyak) have both externally headed and internally headed relative clauses.
In externally headed relatives, using a gap rather than movement in many SALs,
a comitative PP cannot head a gap (infinitival) relative clause. In those languages
in which it can, we have demonstrated that a thematic relation needs to be
established between the predicate of the embedded clause and the comitative
head of the externally headed and internally headed relative clause, either in
terms of an overt case marker or some verbal marker/clitic in the embedded
verb to indicate accompaniment. A similar phenomenon is found in Khasi and
Pnar (Mon-Khmer) too. Though the notion of thematic relations has run into
rough weather in recent years (see Newmeyer 2007: 140–141) and it is difficult
to agree on which “thematic roles exist” and how “to independently justify the
assignment of noun phrases to thematic roles in particular sentences” (Dowty
1989: 70, as quoted in Newmeyer 2007: 140–141) in an objective manner,
our analysis demonstrates that the phenomenon of incorporation of either an
adverb such as together or a postposition or a reciprocal marker as a group
marker in the embedded verb enables the comitative PP to meet the Thematic
Eligibility Condition (TEC) proposed in this study, and a comitative PP can
consequently head a non-finite relative clause. Thus, the TEC needs to be
included in the grammar in order to explicate the occurrence or non-occurrence
of a comitative PP as head in the gap (infinitival) relative (see section 8.7 for a
detailed discussion).4
This can be formulated in terms of the following parameter.
Parameter 4:
[+/− the occurrence of a comitative adposition with the head NP / an
adverb denoting together along with a verbal clitic / an incorporated
comitative adposition in the embedded verb] directly correlates with
[+/− the occurrence of a comitative PP as head] in the gap relative.
the verb occurs in the medial position, and Kashmiri (Indo-Aryan), in which
the verb occurs in the V2 position, though it is underlyingly verb-final (see
chapter 2 for details).
A list of the languages from the four South Asian language families from
which our work draws its data is given below:
ia Indo-Aryan
Assamese
Bangla
Gujarati
Hindi-Urdu
Kashmiri
Konkani
Maithili
Marathi
Nepali
Oriya
Punjabi
Eastern Shina
Shina of Gultari
Shina of Skardu
Sinhala
Swat-Dir Kohistani
Torwali
ib Contact languages
Bhalavali Bhasha (a.k.a. Bhalawali Marathi)
Dakkhini (Hindi-Urdu)
Eastern Bangla (Sylheti)
Mangalore Konkani
Subzapuri
ii Dravidian
Kannada
Malayalam
Manda
Tamil
Telugu
Toda
iii Austro-Asiatic
iiia Mon-Khmer
Khasi
Rymbai
Pnar
10 Introduction
iiib Munda
Ho
Juang
Kharia
Mundari
Santali
Sora/Savara
iv Tibeto-Burman
Ao (Mongsen)
Bodo
Dimasa
Dumi
Garo
Hmar
Kham
Kokborok
Konyak
Ladakhi
Manipuri
Mao
Mising
Mizo
Paite
Rabha
Sangtam
Sema
Tenyidie
Thadou
Tiwa
Most of the data on Tibeto-Burman languages and on some Munda languages
were collected in field trips in the northeastern parts of India and in the states
of West Bengal and Jharkhand.
It should also be mentioned that there are very few syntactic descriptions
available for Tibeto-Burman, Munda and Mon-Khmer languages, especially
at the level of detail and comprehensiveness which are necessary for useful
analysis. Traditional grammars that are available do not address the conceptual
issues that the linguist of today is interested in, in terms of the Universal
Grammar and parametric variation.
The data collected and analyzed by the author with a team of researchers
in the University Grants Commission’s projects on syntactic typology, over a
ten-year period, were also used in this work.
1.9 Richness of data 11
1.9 The richness and complexity of the data and their relevance
Languages belonging to genetically different language families have been
in intense contact for thousands of years in the subcontinent, as a result of
which convergence at various levels (phonological, morphological, syntactic
and semantic) has been in process between them. Syntactic convergence results
in the reanalysis of existing functional categories as new ones, and we shall
show how a contact language optimally utilizes the limited set of functional cat-
egories that it has. Convergence may also result in change in the position of the
verb from verb-medial to verb-final in a sentence, as in the case of the Munda
languages and Kamti Tai, whose source language is Thai, which is SVO.
Our work shows that, though the position of the head changes from VO to OV,
the language may still retain some of its VO patterns with regard to phenomena
such as noun incorporation, or the order of the matrix verb (higher predicate)
following or preceding the embedded verb, as in desiderative constructions
with an object complement in Sora (see chapter 4).
Out of the four different language families of the subcontinent, Austro-
Asiatic (Mon-Khmer and Munda) languages are “perhaps the most divergent
12 Introduction
in the world. They are opposite in structure at every level” (Donegan and Stampe
2004: 3). While Khasi retained its SVO structure in spite of being surrounded by
verb-final languages, the Munda languages not only changed to SOV from the
ancestral SVO,6 but also have been influenced by the neighboring Indo-Aryan
and, possibly, Dravidian languages.
The number of languages spoken, with the richness of diversity in structural
patterns, and the interaction between the languages of the different language
families for a prolonged period provides a great opportunity for investigation,
and enables us to avoid “many spurious generalizations resulting from their
very small sample size, and miss[ing] many true generalizations by not looking
at enough languages” (Baker and McLoskey 2007: 290).
Just to cite an example: the subject in these languages may be nominative
case-marked or non-nominative case-marked. The non-nominative subject con-
struction includes dative, accusative, ergative, genitive, instrumental and loca-
tive case-marked subjects. The agreement patterns in these languages interact
with the case-marking of the subject.
It is generally assumed that agreement is only possible with nominative case-
marked subjects. While one set of South Asian languages permits agreement
with nominative case-marked subjects, the other set permits agreement with
lexically case-marked subjects and even oblique PPs, which are not subjects
(see chapters 4 and 5 for further elaboration and discussion).
for the structure of each language and language family; nor was it possible to
provide examples from all languages for each of the issues considered.
The binding link for all chapters will be to show how languages of the
subcontinent, which are verb-final (except Khasi, Rymbai and Pnar which are
SVO, and Kashmiri which is a V2 language), differ from each other while
sharing a number of common features.
We present data and analysis of specific constructions from SALs. We assume
that the nature of the evidence presented and the conclusions drawn on the basis
of data from the SALs chosen will be compatible with comparable data from
other SALs.
Finally, we wish to conclude by reiterating our contention that typology
is not a theory. For typologists, theory is a guiding principle that facilitates
arriving at linguistically significant generalizations about the intricate nature of
language. Thus, for them theory is a tool and a means to achieve their goals of
research, not an end itself. In contrast, for a theoretical linguist, theory building
is the primary concern, the ultimate end, to which data may be a means. Thus,
we believe that typological studies should neither be too descriptive to become
taxonomic in nature, nor should they be too abstract for linguistically significant
generalizations to be obliterated in theoretical controversies.
2 South Asian languages: a preview
2.1 Introduction
In this chapter, we present a brief survey of a few syntactic features of South
Asian languages to familiarize the reader with some specific structures found
in SALs. We shall present evidence from the verb-final languages of the four
major language families of South Asia.1 The four major language families in the
subcontinent are: (i) Austro-Asiatic (Mon-Khmer and Munda), (ii) Dravidian,
(iii) Indo-Aryan and (iv) Tibeto-Burman.2
This chapter is a radically revised and enlarged version of Subbarao (2007). I am grateful to Rajesh
Bhatt for his very insightful comments which enabled me to reorganize and restructure the chapter.
Thanks to Alice Davison too for her constructive and helpful suggestions.
18
2.2 Typology of South Asian languages 19
exhibit certain syntactic traits which are “disharmonic” (Greenberg 1966) and
are not in conformity with word universals in terms of “consistency.” We shall
show that such structures obey the FOFC (Final-over-Final Constraint) pro-
posed in Holmberg (2000) and discussed in detail in Biberauer, Newton and
Sheehan (2009).5
In this chapter, the focus is on the following aspects of the typology of
SALs.6
Section 2.3 discusses some word order universals found in SALs; section 2.4
discusses some significant syntactic features that provide evidence in support
of the notion of “South Asia as a linguistic area.”7 Section 2.5 focuses on the
parametric variation found in SALs. Section 2.6 briefly mentions the effects of
language contact in SALs. Section 2.7 focuses on the unique syntactic features
that each language family possesses, distinguishing one family from another.
Section 2.8 is the conclusion.
All SALs except Khasi, which is a verb-medial language, share common
structural characteristics at the level of the sentence, and this could mainly be
due to either or both of the following two alternative reasons.
(i) Since all SALs except Khasi (Austro-Asiatic) are SOV (Subject-Object-
Verb) languages,8 they share a number of word order universals proposed
in Greenberg (1966) for verb-final languages.
(ii) All SALs have been in intense language contact with each other for a long
period of time, thus giving rise to the creation of a “linguistic area” or
Sprachbund which literally means ‘language league’ (Hock 1991: 494).
We shall first examine the former alternative. It is generally agreed in the studies
on word order typology as well as in recent theoretical frameworks that it is
the position of the verb (head, as it is generally labeled) in a sentence that
plays a crucial role in the order of occurrence of elements in a sentence. In
the Government and Binding framework, it is referred to as “Head (Direction)
Parameter.” The complement (object) in SOV languages occurs to the left of
the head, the verb or a postposition. Hence, the direction of case assignment in
verb-final languages in a theory such as Government and Binding framework
is from right to left.
In the unmarked order in all SALs, which are SOV (except Khasi), the
complement (direct object, for example) invariably precedes the verb (head)
just as the object of postposition precedes the postposition (head). This is in
contrast to English or French or Khasi, where the complement follows the
head. Thus, there is a mirror image relationship between the complement and
the head in verb-final and non-verb-final languages. Note that a majority of
languages in the world are of the following three types: verb-final, verb-medial
and verb-initial.
We shall now examine the second alternative. In the South Asian subconti-
nent, multilingualism is not an exception but a norm (Pandit 1972; Agnihotri
20 South Asian languages: a preview
1992, 2001), and contact between different languages is not an obstacle but
a facilitator in communication (Pandit 1972). Thus, multilingualism implies
intense language contact as a result of which characteristic features of one
language family or language may be “transferred” to another language or lan-
guage family. Such a phenomenon is labeled “convergence” (see Gumperz and
Wilson 1971; Hock 1991).
We shall discuss some of the syntactic features shared by verb-final SALs.9
In most SALs, the verb occurs in the final position of a sentence as in (2).
However, in Khasi, the verb occurs in the medial position, and in Kashmiri
(IA) (B. Kachru 1973; Bhatt 1999; Raina 2002; Wali and Koul 1997), the finite
form of the verb – be it the main verb or the auxiliary – occurs in the second
position in a sentence. This is generally referred to as the V2 position.
As in all ‘well-behaved’ verb-final languages, in SALs the auxiliary verb
follows the main verb; the indirect object precedes the direct object except in
Sema (Achumi 2000) and Tenyidie (TB) (Kevichüsa 2007) where the direct
object precedes the indirect object in the unmarked word order, if the indirect
object is not overtly dative case-marked. If the dative case marker overtly occurs
with the indirect object, the indirect object precedes the direct object just as in
the other SALs. The set of ditransitive verbs in Tenyidie (TB) includes ts ‘to
give,’ petha ‘to teach,’ pesi ‘to inform’ and rəchə ‘to talk to’ (Subbarao and
Kevichüsa 1998).
do–io order
Tenyidie (TB)
(2) abunɔ- e leda puo- pie a ts
Abuno- nom book one- acc me gave
do io
‘Abuno gave a book to me.’
2.3 Typological characteristics of SALs 21
When the indirect object is dative case-marked, the order is io–do just as in
the other verb-final languages.
io–do order
(3) abunɔ- e a- ki lea puo ketse
Abuno- nom I- dat letter one send o ben
io do
‘Abuno sent a letter to me.’
All SALs, except Khasi, Rymbai and Pnar which are verb-medial, have post-
positions. Since SOV languages have postpositions, the genitive precedes the
head noun as predicted by the implicational universal for verb-final languages.
Comparative and superlative constructions in all SALs (except Khasi) use a
postposition, comparable to than in English, to mark the standard of compari-
son. They all strictly adhere to Greenberg’s word order universal 22.10
Most SALs (except some Tibeto-Burman languages) lack bound comparative
and superlative morphemes, comparable to the -er and -est of English. As
expected, in a verb-final language such as Hindi, the marker of comparison se
‘than’ follows the standard of comparison rādhā ‘Radha’ in (4).
Hindi-Urdu (IA)
(4) raghu rādhā se lambā hai
Raghu Radha than tall is
‘Raghu is taller than Radha.’
Hmar (TB)
(5) lali lala- nekin a- in-11 sāŋ- lem
Lali Lala- than 3s- VR- tall- er
‘Lali is taller than Lala.’
Thus, while some Tibeto-Burman languages and Khasi share this feature of
bound comparative and superlative markers, Dravidian and Indo-Aryan lan-
guages do not. It may be noted that Sanskrit has the markers -tar (comparative)
and -tam (superlative) occurring to the right of the adjective, as in ucca-tar
‘high-er’ and ucca-tam ‘high-est.’ Many Modern Indo-Aryan languages do
have these markers, used only in formal contexts and not productively, unlike
in Tibeto-Burman languages and Khasi.
22 South Asian languages: a preview
Time adverbials (T) precede place adverbials (P) in SOV languages. Thus,
the order of their occurrence is TP.
Time and place adverbials occur in descending order in verb-final languages
(Subbarao 1984a). By descending order, we mean the superordinate chunk
of place or time occurs first, then a subordinate chunk, and then a chunk
subordinate to that follows.12 Examples of time adverbials in Tenyidie (TB)
and place adverbials in Hindi-Urdu (IA) are given in sentences (6) and (7),
respectively.
time adverbials
Tenyidie (TB)
(6) ticie 2000 do khr tarik kerepeŋu theva keba thetha ki
year January month date fifteen night time eight time
‘At eight o’clock at night on the fifteenth of January in the year 2000’
(Kevichüsa 1996; D. Kuolie, p.c.)
place adverbials
Hindi-Urdu (IA)
(7) gāndhı̄ nagar mẽ paccı̄s nambar kı̄ kot.hı̄ kı̄ chat par
Gandhi Nagar in twenty five number of house of roof on
‘On the roof of house number twenty-five in Gandhi Nagar’
(Subbarao 1984a: 18)
The order in which the constituents in time and place adverbials occur in (6)
and (7) demonstrates that all SALs except Khasi are left-branching.
We have discussed some word order universals in the verb-final SALs. We
shall now present some features that are commonly shared by SALs, and
these are the features that provide support to the notion of “South Asia as a
linguistic area.” These include compound verbs, conjunct verbs, reduplication,
echo words, conjunctive participles and the quotative.
Hindi-Urdu (IA)
(8) vah subah-subah hamāre ghar ā- bait.hā
he early in the morning our home come- sat
Literally: ‘He came-sat our home early in the morning.’
‘He came to our home early in the morning’ where the speaker expresses
his unhappiness with the person’s coming early in the morning.
The verb ā ‘come’ may occur alone in the sentence, but alone it does not convey
the intended meaning of the speaker’s unhappiness about the event of coming.
In all Dravidian languages, and some Tibeto-Burman (Bodo and Rabha)
and Indo-Aryan (Assamese, Bangla, Oriya, Marathi, Konkani and Mangalore
Konkani) languages, the main verb is in its conjunctive participial form.
Telugu (DR)
(9) (vād.u)i appud.(u)- appud.u madhyāhnam kūd.ā vacc- (i) vēs-
(he) then- then afternoon also come- cpm DROP-
tā- d.ui
imperf- 3m,s
‘Sometimes he comes in the afternoon, too.’
In Tamil and Telugu (DR), two vector verbs can occur in a row (see sentence
(43) in chapter 9 on the Conjunctive Participle on the Website).
Hindi-Urdu (IA)
(10) mãı̃ ne rām kı̄ madad kı̄
I erg Ram gen help (f.s) did.f.s.
‘I helped Ram.’
(Masica 1991: 368)
Telugu (DR)
(11) āvid.a rādha ki/ nā.ku/ *nā sahāyam cēsin-di
she Radha dat/ I.obl.dat/ I.gen help did-3s,nm
‘She helped Radha/me.’
2.4.3 Reduplication
Reduplication and echo formation are two characteristic features associated
with SALs. Adjectives, question words, verbs and adverbs all have reduplicated
forms (see Abbi 1992). Reduplication normally provides emphasis or imparts
distributive meaning. We shall also demonstrate that reduplication has syntactic
implications.
Marathi (IA)
(12) tı̄ bol- bol bol-lı̄
she talk- talk talk-pst.3 f,s
(stem form) (stem form)
‘She talked a lot.’
(Pandharipande 1997: 534)
2.4 “South Asia as a linguistic area” 25
Konkani (IA)
(13) ti rao- rao rao-l.i
she stand- stand stand-perf.3 f,s
(stem form) (stem form)
‘She stood (waited).’ / ‘She stood (waited) for a long time.’
(Shobha Pedhnekar and Snehal Naik, p.c.)
In Kharia (Munda), it is only the monosyllabic stem that is reduplicated and the
second stem carries the tense marker (Peterson, p.c.).
Kharia (Munda)
(14) lam- o?
search- pst
‘She/he looked for it.’
(15) lam- lam- ki
search- search- pst
‘She/he looked for [it] for a long time.’
In Ho (Munda), in verb stem reduplication, either (i) the first syllable is redu-
plicated, or (ii) only the first vowel of the stem is lengthened. In the former
case, the first syllable may also be lengthened (Deeney 1979: 58).
E.g., Case 1:
jom - ‘eat’ jō-jom - ‘repeatedly eat’
sen - ‘walk’ se-sen - ‘repeatedly walk’
Case 2:
id. - take ı̄d. - ‘habitually take’
em - ‘give’ ēm - ‘habitually give’
The lengthening of the first vowel of the verb stem seems to be unique to Ho
(Munda), as it is not found elsewhere in the subcontinent in the languages that
we studied.14
In Sema (Tibeto-Burman) too, the verb stem is reduplicated.
Sema (TB)
(16) pa- ye cu- cu- ce- ni
he- nom eat- eat- imperf- pres
‘He eats and eats.’
(Achumi 2000)
26 South Asian languages: a preview
Hindi-Urdu (IA)
(17) cāy ‘tea’ cāy-wāy ‘tea and the like’
khānā ‘food’ khānā-wānā ‘food and the like’
Bangla (IA)
(18) hɔr ‘house’ ghɔr-t.ɔr ‘house and the like’
bhat ‘cooked rice’ bhat-t.at ‘cooked rice and the like’
Telugu (DR)
(19) puli ‘tiger’ puli-ili ‘tiger and the like’
ceruku ‘sugarcane’ ceruku-iruku ‘sugarcane and the like’15
Tenyidie (TB)
(20) dapheŋ- ei [proi ilɔŋ- nu vɔ - tyo- -di] aki pu
Dapheng- nom Shillong- to go- fut- that me told
‘Dapheng told me that he would go to Shillong.’
(Subbarao 2000: 102–103)
Hindi-Urdu (IA)
Speaker A:
(21) kyā āp nei us- koj cit.t.hı̄k likh- dı̄k
pol you erg he.obl-dat letter write- gave
‘Have you written a letter to him?’
Speaker B:
(22) h㘠(mãı̃ ne)i proj prok likh- dı̄k
yes (I erg) write- gave
Literally: ‘Yes, (I) wrote.’
Indian English
(23) Yes, I wrote.
(25)–(26). The perfect aspect marker, -ā, and the past tense marker, thā, together
constitute the auxiliary. We label them as Aux1 and Aux2 for convenience.
Hindi-Urdu (IA)
(24) āp ne kis- ko dekh- ā thā
you erg who.obl- dat see- perf pst
subject verb aux1 aux2
Literally: ‘You whom saw?’
‘Whom did you see?’
Telugu (DR)
(27) evari-ki eppud.u enduku
who.obl.dat when why
indirect object time adverb reason adverb
elāgu ēdi evaru istār-
how what who will give-
manner adverb direct object subject
o nāku teliyadu
comp to me not known
Literally: ‘I do not know to whom, when, why, how, what, who will give not
known.’16
30 South Asian languages: a preview
Hindi-Urdu (IA)
(28) saritā- ne kahā thā ki [mãı̃i āp se kal
Sarita.3 pn,s,f- erg said had comp I. 1pn,s,f you with tomorrow
mil.ūngı̄]17
will meet. 1pn,s,f
Literally: ‘Saritai told (me) that Ii will meet you tomorrow.’
‘Sarita had told me that she’d meet me tomorrow.’
Though the matrix verb kahā thā ‘had said’ carries the past tense marker and
the embedded verb mil.ūngı̄ ‘will meet’ carries the future tense marker, the
sentence is grammatical. In English and some other languages, the sequence-
of-tense is observed,18 and extends to time adverbs as well (as in (29) and
(30)). The tense marker and the time adverb are in italics.
English
(29) *? Sarita told me that she’ll meet me tomorrow.
(30) Sarita had told me that she’d meet me the next day.
Unlike English and many other languages, the pronominal forms of the subjects
of the matrix and the embedded clause do not have to be identical in shape in
indirect speech in almost all SALs.19 This is due to the fact that the speaker is
quoting verbatim employing the post-clausal quotative complementizer. How-
ever, significantly in Hindi-Urdu, Kashmiri, Bangla, Oriya and Marathi, the
same phenomenon of tense mismatch and retention of the pronominal forms of
the direct speech are found in spite of the fact that the complementizer occurs
in a pre-clausal position.
In Hindi-Urdu (IA) bāri ‘rain’ is the subject in (33), and the verb agrees with
bāri ‘rain’ in (33) as is the case in all SALs.20
Hindi-Urdu (IA)
(33) dillı̄ mẽ bāri ho rahı̄ hai
Delhi in rain f,s happen progr f,s pres
‘It is raining in Delhi.’
In the following section we present two instances where intense language con-
tact in the subcontinent results in the addition of a new syntactic phenomenon.
Bangla (IA)
(38) amar t.hand.a lege- che
I.gen cold feel- pres
‘I have a cold.’
Kokborok (TB)
(39) a- ni kɔngrai tɔng- ɔ
I- gen cold be- pres
‘I have a cold.’
(Naphurai Jamatia, p.c.)
The only plausible explanation for the occurrence of a sentence with a genitive
subject could be that the construction is borrowed from Bangla.
There are several other instances of convergence between Dravidian and
Indo-Aryan, between Indo-Aryan and Munda, and between Indo-Aryan and
Tibeto-Burman, some of which will be discussed in this book. Interestingly,
though Khasi (Mon-Khmer) has been in contact with some other non-Mon-
Khmer languages for centuries, our work on Khasi shows that there is no
linguistic convergence. The topic of language contact and convergence needs
an in-depth investigation which has implications for the nature of syntactic
change, which is beyond the scope of the present work.
Sandhi rules
Dravidian languages have sandhi rules operating between the constituents of
a sentence which results in the hearer’s perception of a sentence as a single
unit without any pause between words. The operation of such sandhi rules is
the result of the rich agglutinating processes that Dravidian languages have
in comparison to all the other families of the subcontinent. In Telugu, “when
the two vowels come together in the case of two separate words, the first
vowel is generally dropped if it is short” (Krishnamurti and Gwynn 1985: 72).
A similar phenomenon of elision is observed in the other literary Dravidian
languages too. For example, in (40) below, from Telugu, the final vowel in
each word except for the last one is deleted/elided, as a result of which the
entire sentence becomes one whole uninterrupted unit. We have indicated the
elided vowels in parentheses. The words perceived as a unit are enclosed in
brackets.
Telugu (DR)
(40) [atan(u)- eppud.(u)- ēm(i)- enduk(u)- an(i)- annād.(u)- ō]
he- when- what- why- quot- said- dub mkr
Literally: ‘I wonder what, when, why he said it.’
‘I wonder what he said when and why?’
It is the clitic -ō in (40) that imparts the interpretation of ‘I wonder’ in Telugu
and the other Dravidian languages.
A similar phenomenon is observed in the other Dravidian languages too. For
example:
Kannada (DR)
(41) [awan(u)- ēn(u)- anta] hēl.ida
he what quot said
‘What did he say?’
(Hemananda Bisembli, p.c.)
In Malayalam (DR) the first vowel of a word is deleted even when it is preceded
by a word ending in a long vowel (Sreekumar, p.c.), a process that Telugu does
not permit.
Malayalam (DR)
(42) [aval.(ə)- ivit.(e)- un.t.- ō- (e)nn(ə) en (i)- kk(ə) ar.(i)- illa]
she- here- be.pres dub mkr quot I- dat know- neg
‘I do not know whether she is here.’
and Telugu (DR), a dative predicate may assign a single or double dative
case-marking to a subject. Double dative-case marking is permitted in many
Dravidian languages in dative subject constructions expressing inalienable pos-
session and part–whole relationship (Subbarao and Bhaskararao 2004). This
is a feature found only in many Dravidian languages and not elsewhere in the
subcontinent. Thus, the following two types of case marking of the subject
DP are permitted: genitive-dative as in (43) and dative-dative as in (44). The
possessor is genitive case-marked and the possessed – the body part – is dative
case-marked in (43) below.
Kannada (DR)
(43) avan- ai kan.n.j - ige gāyak āgi- de∗i∗j,k
he- gen eye- dat injury happen- agr
‘He got hurt in his eye.’
In Kannada, the possessor as well as the possessed are also both dative case-
marked as in (44).
Kannada (DR)
(44) avani - ige kan.n.j - ige gāyak āgi- de∗i∗j,k
he- dat eye- dat injury happen- agr
‘He got hurt in his eye.’
(N. Varija, p.c)
Kannada (DR)
(45) avanu bara- lilla- a mād.a- lilla
he came- not conj do.pst- not
Literally: ‘He did not come; he did not do.’
‘He did not come at all.’
(Hemananda Bisembli, p.c.)
36 South Asian languages: a preview
Malayalam (DR)
(46) avan vannat- um illa (avan) ceytat- um illa
he come.pst- conj neg he do.pst- conj neg
Meaning: The same as in (45)22
(Sreekumar, p.c.)
Cognitive gerunds
Dravidian languages have a construction (the cognitive gerund)23 in which
the predicate in its gerundival form occurs to the left of the predicate, which
provides a focus interpretation to the gerund. The following examples from
Tamil and Kannada are illustrative.
Tamil (DR)
(48) avan vandadu vantān ān.āl pan.am kon.t.u varaville
he coming (gerund) came but money bring.cpm did not come
‘As for (his) coming, he came, but he did not bring the money.’
(Boologa Rambai Arpudhasamy, p.c.)
Kannada (DR)
(49) awaru baruwud(u) ēnō bandaru, ādare dud.d.u taralilla
he come.gerund as for came but money did not bring
Meaning: The same as in (48)
(Hemananda Bisembli, p.c.)
Tamil (DR)
w i t h a n a f fi r m a t i v e c p
(50) nı̄ŋa enka vı̄t.t.u- kku van.tu ren.d.u varusam akiradu
you our home- dat come.cpm two years happened
‘It has been two years since you came to our house.’
2.7 Unique features of each language family 37
w i t h a ne g a t i v e c p
(51) nı̄ŋa enka vı̄t.t.u- kku var- āma pōy- i ren.d.u
you our home- dat come- neg go- cpm two
varusam akiradu
years happened
‘It has been two years since you did not come to our house.’24
(Boologa Rambai Arpudhasamy, p.c.)
2.7.2 Tibeto-Burman
Nominalization
The characteristic feature of Tibeto-Burman languages is that in the Gap Rel-
ative, the embedded verb is [-finite] in its nominalized form (Matisoff 1972;
Herring 1991; Bickel 1999; Subbarao and Kevichüsa 1999; Lahaussois 2003).
The nominalizer is also used in the formation of non-finite complement clauses,
as one would expect. Further, it is also used productively in the formation of
compounds.
For example:
Tenyidie (TB)
(52) ke- petha- mie
nozr- teach- person
‘teacher’ (the one who teaches)
(Kevichüsa 1996)
Manipuri (TB)
(53) i- mən- na- bə
my- similar- (verb) VREC- nozr
Literally: ‘a person who is similar to me’
‘(my) friend’
Thadou (TB)
(55) thaŋboii - in25 sinema ve øj ai - ki ɔn-
Thangboi- erg cinema (in order to) see 3s- VREC go-
pi- nu- cuj ka kadɔŋmaj - nu aj - hi
with (incorporated postposition) f mkr- def my girlfriend- f mkr 3s- be
‘The girl with whom Thangboi went to see the movie was my girlfriend.’
(Subbarao and Haokip 2009)
verbal anaphor
Kharia (Munda)
(56) proi yo- d.omi − ki- kiyari
they see- VR- middle pst- dual
‘They saw themselves (e.g., in the mirror).’
(Peterson 2006)
Sora (Munda)
(57) al’len ar’- grʔ- tə- biy
we VREC- see- [-pst]- 1p
‘We see ourselves.’
(Anderson and Harrison 2007: 328)
Korku (Munda), however, lost the verbal anaphor and it borrowed the nominal
anaphor from Indo-Aryan.
spoken in Bihar, exhibits possessor agreement and it could be due to the influ-
ence of the Munda languages spoken in the area.
The absence of the nominal anaphoric device, rich verb agreement (as dis-
cussed in chapter 4) and the incorporation of various elements in the verb are
some of the distinct features of the Munda languages.
2.7.4 Indo-Aryan
Indo-Aryan languages are heterogeneous in having syntactic phenomena which
exhibit a wide range of syntactic variation. Just to cite an example, the com-
plementizer in one set of languages occurs to the left of the embedded clause,
while it occurs to the right of the embedded complement in another set. While
the right peripheral complementizer in one subset is a form of the verb say
(generally labeled as the quotative), the other subset has a complementizer
that is derived from a pronoun. While the occurrence of the right peripheral
complementizer (the quotative or the one derived from a pronoun) conforms
to a left-branching structure in uniformity with the basic word order of IA
languages, the complementizer that occurs to the left does not conform to a
left-branching structure (see chapter 6 for a detailed discussion with suitable
examples). Indo-Aryan thus is a “less well-behaved language family” in com-
parison to Dravidian and Tibeto-Burman, which are the “most well-behaved
families.”
We shall discuss some distinct features of Indo-Aryan here.26
Word order
It is only in Indo-Aryan that a V2 language such as Kashmiri is found. All the
other languages are verb-final.
Relative clauses
Indo-Aryan languages have three types of relative clauses: relative-correlative
clauses, right-adjoined clauses and VP-adjoined clauses (Dasgupta 1980;
Subbarao 1984; Dayal 1996; Bhatt 2003). Such a wide variety is not found
in either Dravidian or Tibeto-Burman.27
In the gap relative clause too, the positions that can be relativized vary from
language to language. While Bangla, Hindi-Urdu, Punjabi and Kashmiri cannot
have gap relatives in which an Indirect Object (IO) or Oblique Object (OO) is
modified, Gujarati, Konkani and Marathi permit the modification of the IO as
well as of some specific PPs of the OO.
While Bangla, Hindi-Urdu, Punjabi and Kashmiri do not permit the modi-
fication of either the subject or the theme in the non-nominative subject con-
struction, Marathi, Konkani and Gujarati do (see chapter 8 for details).
2.8 Conclusion 41
Clefts
In a cleft construction, a constituent is focused. No Indo-Aryan language except
Sinhala (see Gair 2003 for details) permits clefts. It is only in Sinhala that they
are found and it could plausibly be due to Dravidian influence. We provide an
example from Sinhala (IA) below in (59).28
The constituent that is clefted, for example ē miniha ‘that man’ in (59),
occurs to the right of the sentence as the final constituent (Gair 2003), as is the
case in Dravidian and Tibeto-Burman languages.
Sinhala (IA)
(59) ı̄ye gunəpālə- t.ə salli dunne ē miniha
yesterday Gunapala- dat money gave-focus that man
‘It was that man who gave Gunapala money yesterday.’
(Gair and Paolillo 1997: 41)
Nominal anaphors
Indo-Aryan languages predominantly exhibit a pattern in which only the nom-
inal anaphor (reflexive and reciprocal) occurs in a sentence, and the verbal
anaphor that occurs in a very few languages has functions that are limited in
comparison to the other language families of the subcontinent (see chapter 3
for details).
2.8 Conclusion
In this chapter, we have provided a brief description of the major typological
features of SALs with a view to enabling the reader to understand some of the
structures available in these languages. We have also discussed some specific
features that are either unique or significant in terms of the structures in a
particular language family. We have provided two instances to demonstrate the
crucial role that syntactic convergence plays in language contact situations.
42 South Asian languages: a preview
3.1 Introduction
The classification of nominal expressions in terms of anaphors (reflexives and
reciprocals), pronouns and referential expressions has been one of the major
areas of attention in the generative tradition, especially in the Government and
Binding framework and the Minimalist Program. Anaphora in simple terms
means ‘backward reference.’ Lust et al. (2000: 885) point out that it is “a
relation between a form and a linguistic antecedent.” The interpretation of the
anaphor is in some way determined by the interpretation of its antecedent (Lust
1986; Wasow 1986). In this chapter, we shall discuss some of the major issues
in the study of anaphors and pronouns with a view to arriving at their typology
in SALs.1 Based on evidence from SALs, we wish to demonstrate that
(i) an anaphor may have a nominative or non-nominative subject as its
c-commanding/subcommanding antecedent;
(ii) the verbal reflexive (VR, hereafter) and verbal reciprocal (VREC, here-
after), which are monomorphemic, perform a variety of functions and are
invariably “subject-oriented”;
(iii) the verbal anaphor2 in SALs provides evidence in support of the implica-
tional universal which states: if the verbal anaphor is coindexed with the
indirect object or oblique object, it should invariably be coindexed with
the direct object;
(iv) long-distance binding is permitted only in non-finite clauses, and is pro-
hibited when either the VR/VREC or a complex form of the anaphor
is present in the embedded clause as this “insulates” the anaphor from
coindexation with a long-distance antecedent;
(v) Blocking Effects are found only in languages without subject–verb
agreement;
(vi) scrambling does not affect anaphor–antecedent relations; and
(vii) all SALs except Marathi (IA) obey Principles A, B and C of the Binding
Theory.
Thanks to Frans Plank and Martin Everaert for their helpful comments on this chapter.
43
44 Lexical anaphors and pronouns in SALs
complex form may also permit such binding. The verbal device is a simplex
(monomorphemic) form in SALs and it blocks long-distance binding.
Before we proceed further, we shall present a brief discussion of some rele-
vant notions related to anaphors and pronouns. Anaphora plays a crucial role in
natural languages. In the generative tradition, the study of anaphors (reflexives
and reciprocals) and pronouns has played a major role. There are three prin-
ciples of the Binding Theory “defined over configurational representations of
sentence structure, because ‘binding’ implies ‘c-command and coindexing’ in
a local domain” (Lust et al. 2000: 3).
Binding principles:
Some of the notions that are important in the study of anaphora include: local vs.
long-distance binding, c-command, governing category, Functional Complex,
etc. (see Lust et al. 2000: appendix).
Languages that only have a nominal device are the following:
Indo-Aryan: Hindi-Urdu, Kashmiri, Punjabi, Nepali, Bangla,
Assamese and Oriya, just to name a few
Dravidian: Malayalam and Toda
Munda: Korku
Tibeto-Burman: Kokborok and Tenyidie (only the nominal reflexive)
The nominal device consists of either a simplex or a complex anaphor. The
complex form in SALs contains a bipartite structure whose constituents may
or may not be identical.
Languages that have a nominal as well as a verbal anaphor are the following:
Indo-Aryan: Marathi, Gujarati and Sinhala
Dravidian: all except Malayalam and Toda
Tibeto-Burman: Hmar, Mizo, Aimol, Thadou, Zou, Manipuri, Rabha,
Tenyidie (no verbal reflexive, only a verbal reciprocal), Ao and
Bodo
Mon-Khmer: Khasi
Munda: Kharia and Juang (only a verbal reciprocal)
Languages that have an indigenous verbal anaphor alone are the following:
Munda: Ho, Kharia, Mundari, Santali, etc.
A nominal anaphor too is found in Ho, Kharia and Juang. It is borrowed from
Hindi, Sadani and Oriya (IA), respectively.
We shall now provide a description of the form of the anaphor in the four
South Asian language families.
46 Lexical anaphors and pronouns in SALs
3.2.1 Indo-Aryan
In Indo-Aryan, the form of the anaphor is unmarked for person, gender and
number. We provide examples from Hindi-Urdu (IA).
Hindi-Urdu (IA)
The complex form of the anaphor is apne-āp ‘self’s-self + postposition.’
It consists of the oblique form apne followed by the form āp ‘self.’ Further,
the whole complex form of the anaphor apne-āp invariably needs to be followed
by a postposition, and it “has no nominative or ergative -ne form” (Davison
2000: 401). The bare form (devoid of a postposition) of the anaphor apne-āp
‘self’s-self’ stands only for the emphatic form. The emphatic has no simplex
form.
The possessive reflexive form is apnā ‘self’s.’ When followed by a post-
position, apnā has the form apne, and it is referred to as the oblique form in
Hindi-Urdu grammars.
There is a simplex form of the anaphor too, and it is apne + postposition.
Table 3.1 illustrates the complex and simplex forms in different cases.
There are other forms such as khud ‘self’ and swayam ‘self’ followed by
a postposition. The simplex anaphors swayam + PP / khud + PP ‘self + PP’
may also occur in place of apne-āp + PP ‘self + PP,’ where PP stands for a
postposition.
3.2.2 Dravidian
We provide examples from Telugu (DR). The form of the anaphor (reflexive) is
person- and number-dependent. In 1st and 2nd person, the anaphor is formed by
the reduplication of personal pronouns, and in 3rd person by the reduplication
of tanu ‘self’ in singular and tama ‘selves’ in plural. In both the singular and
plural forms, the first part of the anaphor is followed by a postposition, which
in turn is followed by a reduplicated form of the first part (Table 3.2). Thus,
3.2 Form of the anaphor in SALs 47
the form of the anaphor in the 1st person singular, for example, is na-nnu
nēnu ‘I.obl-acc I.nom,’ ni-nnu nuvvu ‘you.obl-acc you.nom’ in the 2nd person,
whereas it is tana-ni tanu ‘self.obl-acc self.nom’ in the 3rd person. Note that
tana is the oblique form of tanu when followed by a postposition. The second
part of the complex form of the reflexive in Table 3.2 is in the nominative case,
while the first part takes the case of the position in which the reflexive occurs
in a sentence. In Table 3.2, it is taken as the accusative.
The simplex forms of the reflexive in 3rd person are given in Table 3.3. In
Table 3.3 too, it is taken as the accusative.
3.2.3 Tibeto-Burman
We provide examples from Mizo (TB).
In Mizo, the form of the anaphor is person- and number-dependent. For
example, in 1st person singular it is kei-maʔ leʔ kei-maʔ ‘1s-emph and 1s-
emph.’ Table 3.4 gives 3rd person [+human] reflexive pronouns:
48 Lexical anaphors and pronouns in SALs
3.2.4 Austro-Asiatic
Munda family
For the Munda family of languages the following statements can be formulated:
(i) Mundari, Ho and Santali have the infix -n- as a verbal reflexive, and -pa-
as a verbal reciprocal. Sora has a verbal reciprocal -əl- (Donegan and
Stampe 2004). It does not have a nominal anaphor.
(ii) In Ho and Santali the VR -n- also functions as an inchoative/anticausative
suffix, as well as a passive marker. Ho also has a nominal anaphor borrowed
from Hindi.
(iii) The verbal reciprocal -pa- in Ho, Mundari and Santali is infixed in the
verb after the first syllable of the underived form.
(iv) Kharia has a verbal reflexive d.om and a verbal reciprocal kol. It also
has a nominal anaphor borrowed from the Indo-Aryan language Sadani
(Peterson, p.c.).
(v) Juang has no verbal reflexive, only a verbal reciprocal. It has two nominal
reflexives, āpein and āpan.ā; āpan.ā is borrowed from Oriya (IA) (Patnaik
and Subbarao 2000: 842).
Khasi (Mon-Khmer)
Khasi has both simplex and complex forms of the nominal reflexive. The
simplex form for the object position is ya-lade ‘acc/dat-self,’ and the complex
form is da-lade ya-lade ‘instr-self acc-self.’ The reflexive is unmarked for
person, gender and number, just as in Hindi-Urdu. The nominal reciprocal is
iwεi-ya-iwεi ‘one-acc-one’ which means ‘each other.’ Khasi also has a VR -hi
that occurs to the right of the verb stem, and a verbal reciprocal -ya- that occurs
to the left of the verb stem. The occurrence of the verbal reflexive and verbal
reciprocal3 is optional, while the occurrence of the nominal form is obligatory.
As we shall see, this is in contrast to Telugu (DR), Mizo, Thadou, Paite, Zou
3.3 Occurrence of the anaphor 49
and Hmar (TB), etc., where the nominal anaphor is optional and the verbal
anaphor is obligatory (see sentence (11)).
In this section we have presented a brief description of the forms of the
anaphor and in the following section we discuss their occurrence in different
positions in a sentence.
3.3.1 Indo-Aryan
Marathi (Wali 2000: 520) and Gujarati (Mistry 2000: 345–346) have a nominal
anaphor and a verbal anaphor. Sinhala has a verbal form gannəwa ‘take’ as a VR
which “may occur in reflexive contexts to provide or reinforce the reflexivity
and with some verbs, it will enforce a reflexive interpretation on a pronominal
argument of a verb” (Gair and Karunatillake 2000: 725–727). In these three
languages, the functions that the VR performs, the occurrence of the VR in
different syntactic domains and the overt presence of the VR in terms of
its obligatoriness in occurrence are far less in comparison to the other three
language families of the subcontinent. Further, there is no verbal reciprocal in
Gujarati and Marathi. In Sinhala, the verb + gannəwa, the verbal reciprocal,
alone “would suffice to enforce limitations of coindexing to the local domain”
(2000: 748). The authors did not elaborate this statement further.
In Gujarati, the reflexive potā ‘self’ is objective case-marked in (1) and is
coindexed with the subject.
Gujarati (IA)
(1) rāji potāi - ne vagovše
Raj self objective will humiliate
‘Raj will hurt himself.’
(Mistry 2000: 351)
In sentence (2), from Hindi-Urdu (IA), the anaphor may be simplex or complex.
The pronoun use ‘her’ cannot be coindexed with the subject and it has its own
discourse referent.
Hindi-Urdu (IA)
(2) rādhāi ne apne (āp)i,∗ j - ko/ use∗ i,j šı̄še mẽ dekhā
Radha erg self’s selfi,∗ j - dat her∗ i,j mirror in saw
‘Radhai saw / looked at herselfi,∗ j/ /her∗ i,j in the mirror.’
(Davison 2000: 408)
50 Lexical anaphors and pronouns in SALs
Kashmiri (IA)
(3) mohn chu pān- as/ pannis pān- as parnāvān
Mohan is self dat self- dat teaching
‘Mohan is teaching himself.’
The complex anaphor panun pān ‘self-abs’ in direct object position occurs in
the absolute case in (4), in contrast to Hindi-Urdu and many other SALs.
3.3.2 Dravidian
Malayalam, Toda and Telugu
To the best of our knowledge, Malayalam and Toda4 are the only Dravidian
languages which have no VR/VREC, and have only the nominal anaphoric
device. Malayalam has a verbal affix -kol..luka. According to Asher and Kumari
(1997: 165): “(V)erb forms carrying kol..luka in Malayalam carry such shades of
meaning as willingness, passive acceptance, permission and the fact that they
often combine with a reflexive pronoun would support the view that reflexivity
is not a dominant feature of the verb form itself.” Hence, -kol..luka cannot be
considered as a verbal anaphor.
Malayalam has two forms of the nominal anaphor – for example, in direct
object position: tann-e tann-e ‘self-acc emph’ or awan-e tann-e ‘he.acc emph’ –
according to Jayaseelan (2000: 122). Jayaseelan (2000: 116) considers the sec-
ond part of the anaphor tann-e as ‘emphatic,’ and not an instance of reduplica-
tion, as it is in Telugu and some other Dravidian languages.
Malayalam (DR)
(5) rāmani tanni/∗ j - e tanne / awan-i/∗ j- e tanne wer.uttu
Raman self- acc emph he- acc emph hated
‘Ramani hated himselfi/∗ j .’
(Jayaseelan 2000: 141–142) (Sentences (112a) and (117a) of
Jayaseelan 2000 are combined as one sentence in (5) above.)
Jayaseelan (2000: 122) observes: “a bare personal pronoun like awan or a bare
reflexive pronoun like tān will not do [in (5) in DO position]. A null form is
also not possible in coreferential reading.”
3.3 Occurrence of the anaphor 51
Telugu (DR)
The nominal anaphor in Telugu is a complex expression formed by a process of
reduplication of the simplex form tan ‘self’ and Case Copying (Subbarao and
Lalitha Murthy 2000). The VR occurs to the right of the main verb. In Telugu
and Kannada (Amritavalli 2000: 53) the occurrence of the VR is obligatory
(barring a few exceptions) when the nominal anaphor occurs in a subcategorized
position.
Telugu (DR)
(6) rādha (tana- ni (tanu)) pogud.u-5 kon- di
Radha self- Acc self (nom) praise- VR- agr
‘Radha praised herself.’
(Subbarao and Lalitha Murthy 2000: 233)
The nominative marker in Telugu is null. The anaphor consists of two parts:
(i) tana-ni, and (ii) tanu. The first part of the anaphor tana-ni carries the
structural accusative case marker -ni. The second part of the anaphor is in the
nominative case, and we have argued elsewhere (Subbarao and Saxena 1987;
Subbarao and Lalitha Murthy 2000) that the nominative case of the subject is
copied onto the second part of the bipartite structure of the anaphor. We have
labeled this as Case Copying. Evidence in support of our claim comes from the
dative subject construction and polymorphemic reciprocals in Telugu (see 3.14).
The following options with regard to the occurrence/non-occurrence of the
nominal anaphor in the nominative subject construction are available in Telugu:
(i) the second part of the nominal anaphor can be dropped leaving the first part
intact, but the first part cannot be dropped leaving the second part alone;
and
(ii) a null form (pro) can occur in place of the nominal anaphor,6 unlike in
Malayalam (Jayaseelan 2000: 122).
Note that the occurrence of the VR is obligatory when the nominal anaphor is
in subcategorized position.
Thus, the following patterns obtain in Telugu:
(i) subject – reduplicated anaphor – Verb – VR – agr
(ii) subject – the first part of the anaphor alone – Verb – VR – agr
(iii) subject – null form – Verb – VR – agr (the reflexive interpretation is intact
without the nominal anaphor).
There are a few predicates which do not permit the VR in Telugu. These
predicates include marcipōvu ‘forget’ (see (7)), gurtu pat..tu ‘recognize,’ cirāku
pad.u ‘be irritated,’ madhana pad.u ‘worry.’
Note that the simplex form of the anaphor is not permitted in (7) as there is no
verbal reflexive. This, in our opinion, demonstrates that reduplication plays a
crucial role in anaphoric interpretation in the absence of the VR.
Interestingly, it is the same set of predicates that do not permit a verbal
reciprocal either, and a nominal bipartite reciprocal is required obligatorily, as
there is no verbal reciprocal marker.
Tamil (DR)
In Tamil, just as in Telugu, there is a nominal as well as a verbal anaphor. The
verbal reflexive is kid.u or kol. in written Tamil according to Annamalai (2000).
According to him, the occurrence of the VR is optional. In (8), the verb may
contain kit.u, the VR (in italics), as in pāttukit..tān ‘look at.pst-VR-agr,’ or it may
not, as in pātt-an ‘look at.pst-agr.’ The glossing of kit..t as VR (verbal reflexive)
is ours, not Annamalai’s. He glosses it as aux (auxiliary). Lehmann too (1989:
361) labels it as ‘the auxiliary verb.’
Tamil (DR)
(8) kumāri tannei kan.n.ād.ile pāttu-kit..t-ān/ pātt-ān
Kumar self.acc mirror.in look at.pst-VR-agr look at.pst-agr
‘Kumari looked at himselfi in the mirror.’
(Annamalai 2000: 180)
Just as with the Telugu verb marcipōvu ‘forget,’ the occurrence of the VR is
not permitted with verbs such as ‘remember’ in Tamil.7
3.3.3 Tibeto-Burman
Mizo
In Mizo, the nominal anaphor is a complex form a-maʔ leʔ a-maʔ 8 ‘himself/
herself and himself/herself.’ It is formed by conjoining two nominal forms of
a-maʔ ‘himself/herself’ with the conjunction leʔ. The VR -in- occurs to the left
of the verb stem. Just as in Telugu, the occurrence of the VR is obligatory when
an anaphor occurs in a subcategorized position (Lalitha Murthy and Subbarao
2000).
When the anaphor occurs in direct object position, the VR detransitivizes the
predicate as a result of which ‘the ergative marker [which occurs obligatorily
with transitive verbs] does not occur with the subject’ (Lalitha Murthy and
3.3 Occurrence of the anaphor 53
Subbarao 2000: 785). That is, the subject is not in ergative case even in a
transitive sentence, when the DO is an anaphor.
a n a p h o r i n direct object p o s i t i o n
Mizo (TB)
(10) zovai (a- maʔ leʔ a- maʔ)i a- in- huā
Zova he- self and he- self 3s- VR- hate
‘Zovai hates himselfi .’
(Lalitha Murthy and Subbarao 2000: 790)
Note that in (10) the nominal reflexive a-maʔ leʔ a-maʔ that occurs in the
direct object position is not morphologically accusative case-marked. However,
it is the occurrence of the reduplicated form that does not necessitate overt
morphological case marking. It is structurally case-marked as will be evident
from (11) in which the monomorphemic form requires overt morphological
accusative case-marking. Such accusative case marking, either overtly as in
(11) or structurally as in (10), creates a paradoxical situation in which, on one
hand, the ergative marker does not occur with the subject as the predicate with
the verbal reflexive is [–transitive], and, on the other hand, the [–transitive]
predicate needs to assign accusative case to the nominal anaphor, a direct
object. Such an anomalous situation poses a problem for which any theory will
find it difficult to provide an explanation.9
The first part of the nominal anaphor a-maʔ ‘himself/herself’ as a simplex
form alone is permitted, and the occurrence of the VR is obligatory in (11) just
as in (10). The occurrence of the accusative marker cu with a-maʔ ‘himself/
herself’ is obligatory in such cases.
(11) lālii (a-maʔi - cu) a- in- that
Lali she-self- acc 3s- VR- kill
‘Lalii killed herselfi .’
(Lalitha Murthy and Subbarao
2000: 790)
Note that, in spite of the occurrence of the VR with the predicate, the ergative
marker occurs with the subject. This shows that it is the occurrence of the
anaphor in direct object position alone that is instrumental in detransitivizing
the predicate, and not in non-direct object positions.
Manipuri
We shall now discuss the case of lexical anaphors in Manipuri. The occurrence
of the VR clitic je/ce in Manipuri is optional, as indicated by parentheses in
(13). The nominal anaphor in Manipuri is a complex reduplicated form having
a bipartite structure. Just as in Telugu, there is Case Copying of the Case of
the subject onto the anaphor. However, in Manipuri, it is the first part of the
reduplicated structure that carries the Case Copy of the subject. In contrast, it is
the second constituent of the reduplicated structure that carries the Case Copy
of the subject in Telugu.
Manipuri (TB)
(13) aii - na i-sai - na i-sai - bu thagat- (cei )- i
I- nom I-self- nom I-self- acc praise- VR- [−fut]
‘I praised myself.’
(Sarju Devi and Subbarao 2003: 58)
Note that the nominative case marker -na occurs with the subject. The accusative
case marker is -bu. The anaphor is a reduplicated form as in the Dravidian
languages. The first part consists of the anaphor for self in 1st person, namely
i-sa ‘I-self,’ and is followed by the nominative case marker -na. It is the
nominative case marker of the subject that is copied onto the first part of the
reduplicated structure of the anaphor. Note that i-sa in (13) is not an emphatic
form, as the emphatic form of the 1st person is isa-mak. Thus, the phenomenon
of reduplication and Case Copying are both operative in Manipuri too.
The second part of the anaphor carries the accusative case marker -bu in direct
object position. The bipartite structure of the lexical anaphor is schematically
given in (14a) and (14b) as Part I and Part II.
Tenyidie (TB)
(16) uniei - e lešə thu kei - tsə- ya
they (dual)- nom letters write VREC- give- pres
‘They (dual)i write letters to each otheri .’
(Kevichüsa 1996)
The occurrence of the verbal reciprocal and nominal reciprocal are mutually
exclusive in Tenyidie. When the verbal reciprocal occurs, the nominal reciprocal
cannot occur, as the verb is detransitivized in such cases, according to Kevichüsa
(2007).
The verbal reciprocal in Ao is təp and it also indicates collective activity
with the interpretation of together (Coupe 2003: 336).
3.3.4 Austro-Asiatic
Khasi
Khasi has a nominal simplex form ya-lade ‘acc/dat-self’ or a complex form
da-lade ya-lade ‘instr-self acc-self.’ It has a monomorphemic VR hi-. It occurs
to the right of the verb and is optional; da-lade ‘instr-self’ too is optional (17)
(Temsen and Subbarao, ms).
56 Lexical anaphors and pronouns in SALs
Khasi (Mon-Khmer)
(17) u- lemi u- pεit- (hi) (da- lade)i ya- ladei ha- ka- yit
3 m,s- Lem 3 m,s- look- VR instr- self acc- self loc- f- mirror
‘Lemi looked at himselfi in the mirror.’
(iii) The verbal reciprocal marker -ya also indicates that the action is performed
with a collaborative effort by the subject. In (22), though the verb exhibits
agreement with the subject in the singular, the interpretation of the sen-
tence is in the plural. Such interpretation, we hypothesize, is contributed
by the collaborative effort marker (cem) ya in (22).
Munda languages
In most Munda languages, it is only the verbal device that is available as an
anaphor, indigenously. Due to the influence of the superstrate IA languages,
such as Hindi, Bangla and Oriya, however, one does find a nominal anaphor
too. The occurrence of the verbal anaphor is obligatory. It is significant that the
transitive marker that occurs with a transitive verb does not occur when the
verb occurs with a verbal anaphor. It is the intransitive marker that occurs with
the transitive verb that functions as the verbal reflexive.
Thus, in Ho, the intransitive marker, the verbal reflexive and the passive
marker are identical in form, and it is /-n-/. The marker /-d/ that occurs
with transitive verbs cannot occur when the verbal reflexive/reciprocal occurs.
Instead, the intransitive marker, which is homophonous with the verbal reflex-
ive, occurs with the verb. In Mundari (Munda) too, a similar situation obtains.
The non-occurrence of the nominal anaphor is indicated by ø.
Ho (Munda)
(23) proi ar.si- re- m ø nel- ke- ni - a
you mirror- in- 2s see- pst- VR[−tr]- fin
‘You saw yourself in the mirror.’
(Koh and Subbarao, ms)
In Mundari, the VR suffix is -en after consonants, and -n after vowels. It also
functions as an intransitive suffix.
In Ho, Mundari and Santali, the verbal reciprocal -pA- is infixed in the verb,
and it is placed after the first syllable of the underived form. The letter A in
the infix -pA- stands for any vowel, and it harmonizes with the stem vowel that
precedes it. The verb form for ‘hit’ in Santali is dāl. Since the verbal reciprocal
-pA- is infixed in the verb, and the verb is split into two parts, we’ve glossed
da- as ‘hit1’ and -l as ‘hit2.’ The absence of the nominal reciprocal is indicated
by ø.
58 Lexical anaphors and pronouns in SALs
Santali (Munda)
(24) unkini - kin ø dā- pāi - l- kā- n- a
they dual- dual hit1- VREC- hit2- pst- [−tr]- fin
‘They (two of them) are fighting [with] each other.’
(Minegishi and Murmu 2001: 104)
In Kharia, d.om is also used as passive marker. Peterson 2006 (p.c.) observes:
“with the wholesale importation of the Indo-Aryan reflexive construction with
apan . . . the reflexive use of d.om has receded from this domain and is perhaps in
the process of being lost.” This impending loss is due to superstratum influence
of Hindi and Sadani (IA). Kharia also has a verbal reciprocal kol.
In the following section, we discuss the various functions that the verbal
reflexive/reciprocal performs in SALs.
3.4.2 As a detransitivizer/anticausative
(i) Absence of the ergative marker: In Mizo (TB), in the direct reflexive the
ergative marker does not occur when the direct object is coindexed with
the subject, as the VR detransitivizes the verb (see sentences (10) and
(11)).
(ii) Verbal and nominal reciprocals in complementary distribution: In
Tenyidie (TB) the occurrence of the verbal reciprocal detransitivizes the
verb, and, hence, the nominal reciprocal cannot occur when the verbal
reciprocal occurs (Kevichüsa 2007).
60 Lexical anaphors and pronouns in SALs
(iii) Homophonous form for the verbal reciprocal and anticausative: The VR
and the detransitivizer/anticausative in Dravidian, Mon-Khmer, Munda
and Tibeto-Burman languages are homophonous. In Telugu and Kannada
(DR), and in Mizo and Hmar (TB), the detransitivizer reduces the valence
of the corresponding transitive verb by one argument. This is similar to
the phenomenon of ‘valence-decrease’ in the verb observed in many other
languages, such as Russian and Polish (Haspelmath 1993; Kazenin 2001:
923; Piñón 2001). In Dyirbal (Dixon 1972: 89) too, “[A] reflexive form
functions like an intransitive stem.”
In Telugu (DR), teruc ‘open’ is a transitive verb and, hence, it takes two argu-
ments, whereas terucu-konu ‘open-VR,’ a [−tr] verb, takes only one argument.
Thus, the VR reduces the valence of the verb (see Krishnamurti and Gwynn
1985: 208 for details).
Telugu (DR)
(27) talupu terucu- kon- di
door open- inchoative-14 3s,nm
‘The door opened.’
(Subbarao and Lalitha Murthy
2000: 227)
In Kannada (DR), the intransitive verbs and their corresponding transitive verbs
in Table 3.5 show how the occurrence of the VR (in italics) ‘detransitivizes’
the verb.
Kannada (DR)
Table 3.5 VR as a detransitivizer
[–transitive] [+transitive]
In Mizo and Tangkhul (TB) and Khasi (Mon-Khmer) too, a similar phenomenon
is observed.
Mizo (TB)
(28) konkhāy a- in- hon
door 3s- inchoative- open
‘The door opened (on its own).’
(Lalitha Murthy and Subbarao
2000: 788)
3.4 Verbal reflexive/reciprocal 61
Khasi (Mon-Khmer)
We will argue that all dative predicates are [−tr] (see chapter 5 for full dis-
cussion). The fact that the detransitivizer kon occurs in the dative subject
construction with a transitive verb such as kōs ‘cut[+tr]’ in Telugu (DR) in
(32), turning it to [−tr], lends further support to our claim that kon functions
like a detransitivizer.15
Telugu (DR)
(32) [kapil- ki]i ivāl.a vēluj kōsu- kon- di∗ i,j
Kapil- dat today finger cut[+tr]- inchoative/[−tr]- 3 s,nm
Literally: ‘To Kapil (his) finger got cut today.’
‘Kapil’s finger got cut today (inadvertently).’
Further, the presence of kon in Telugu with a dative subject indicates a non-
volitional act (32).
The verbal anaphor cannot occur in the dative subject construction in
Kannada. Amritavalli (2000: 54) attributes this to the fact that the verbal
62 Lexical anaphors and pronouns in SALs
3.4.3 As a self-benefactive
The marker for the verbal reflexive/reciprocal functions as a self-benefactive in
Telugu and Kannada (DR) and Mizo (Lalitha Murthy and Subbarao 2000: 788),
and Hmar (TB). It is labeled as the ātmanēpada marker in Sanskrit grammars,
and the effect of the action is invariably ‘agent oriented.’ Hence, it (the verbal
anaphor functioning as a self-benefactive) is not found in a passive sentence.
The occurrence of the nominal anaphor is optional.
Telugu (DR)
(33) mālati annam vand.u- kon- di
Malati food cook- self ben- 3 s,nm
‘Malati cooked food for herself.’
(Subbarao and Lalitha Murthy 2000: 227)
Kannada (DR)
ogi- ‘to wash (clothes)’ ogedukol..l- ‘to wash (clothes) for oneself’
(Amritavalli 2000: 55)16
Kannada (DR)
(34) rāma øi,/∗ j (*tanage) kai tol.edu- kon.d.a
Rama ø (*self-dat) hand wash- inch-agr
‘Ramai washed (hisi,/∗ j ) hands.’
(Amritavalli 2000: 57)
Telugu (DR)
(35) pūjai moham kad.ugu- koni - di
Puja face wash- VR- 3 s,nm
‘Poojai washed (heri ) face.’
3.4 Verbal reflexive/reciprocal 63
In Ho (Munda), the verbal reflexive, the intransitive marker and the self-
benefactive are homophonous (36). -iŋ is the 1st person singular subject agree-
ment marker on the verb.
Ho (Munda)
(36) proi em- en- tan- a- iŋi
I give- [-tr]/VR- progr- fin- 1 s
‘I am giving myself.’
(Burrows [1915] 1980: 74)
The possessor of the body part does not occur in Classical Greek either.
Classical Greek
(37) lou’o- mai t- `
‘ās khe
ir- as
wash- 1s.middle.pres article- acc.p hand- acc.p
‘I wash my hands.’
(Kazenin 2001: 918)
Ho (Munda)
Direct Object as subject of the passive
In Ho, the suffix -n functions as a [–transitive] marker, a passive marker and a
verbal reflexive. Thus, sentence (38) is ambiguous between a VR reading and
a passive reading.
Ho
(38) proi siŋboŋa- taʔa- re- mi - ema- ke- n- a
you God- to- in- 2s- give- pst- VR/passive- [+fin]
‘You gave yourself to God.’ / ‘You were given to God.’
(Koh and Subbarao, ms)
In Sora, Kharia and Gorum (Munda) too, the VR functions as a passive and
intransitive marker.
64 Lexical anaphors and pronouns in SALs
Sora (Munda)
(39) ran- n- ēten
crush- [−tr]- 3.pst
‘It was crushed.’
(Ramamurti 1931)
Kharia (Munda)
(40) proi yo- d.omi - ki- kiyari
they see- pass/VR- middle pst- dual
‘They were seen by someone else.’ / ‘They saw themselves (in the mirror).’
(Peterson 2006)
Khasi (Mon-Khmer)
(41) proi ŋai - sōʔ- hi ya- ka-ne ka- sɔpti
1s- sew- emph acc- f.s.-this f.s- dress
‘I sew this dress myself.’
Kharia (Munda)
(42)
oʔ- jom- ta
eat- VR- pres.ind
‘He eats by himself.’
(Malhotra 1982: 178)17
Hmar (TB)
(43) hi naupaŋtepa hi kha naupaŋtepa kha nekin a- in- sāŋ
this small boy this that small boy that than 3s- VR- tall
‘This small boy is taller than that small boy.’
3.4 Verbal reflexive/reciprocal 65
(i) nōd. ‘to see, to look’ nōd.i kol.l. ‘to look after’
(ii) enn ‘to say’ endu kol.l. ‘to think’
(iii) ett ‘to lift’ etti kol.l. ‘to carry’
Telugu (DR)
(44) binde lō nı̄l.l.u nind.u- tunnāyi
pot in water fill[−tr]- pres progr
‘Water is filling in the pot.’
Telugu (DR)
(46) *pragatii tanai - ni tit.t.- in- di
Pragati self- acc scold- pst- 3 s,nm
‘Pragatii scolded her*i .’
Marathi (IA)
(47) lili∗ i āplyā- lā∗ i/∗ j haste
Lili self-dat laughs
*‘Lili laughs at self∗ i,∗ j .’
(Wali 2000: 538)
However, sentence (48) shows that the occurrence of the VR makes the nominal
anaphor in (46) local.
Telugu (DR)
(48) pragatii tana- nii,∗ j tit.t.u- koni - di
Pragati self- acc scold- VR- 3 s,nm
‘Pragatii scolded herselfi,∗ j .’
In Marathi (IA), the auxiliary verb ghen.e ‘to take’ functions as a self-benefactive
and, as a VR, it creates “a reflexive context” (Wali 2000: 521): “When the verb
is compounded with ghen.e, however, the use of āpan. [which is otherwise anti-
local] becomes acceptable in the otherwise forbidden local domain” (2000:
521).
Marathi (IA)
(49) lili ni āplyā- lā mārun ghet- la
Lili erg self-dat beat VR- pst
‘Lili beat herself.’
(Wali 2000: 521)
In Sinhala (IA) too, the verb gannəwa ‘take’ occurs with a verb “in reflexive
contexts to provide or reinforce the reflexivity” (Gair and Karunatillake 2000:
725). It makes the anti-local pronoun local. The verb dı̄gatta in (50) is the
compounding of the perfect participial stem with the VR gannəwa ‘take’ (Gair
and Karunatillake 2000: 725).
Sinhala (IA)
(50) gunəpaləi eyāt.əi/∗ j t ǣgi dı̄gatta
Gunapala she/he.dat presents give.pple.+take.pst
‘Gunapalai gave himselfi/∗ j presents.’
(Gair and Karunatillake 2000: 725)
68 Lexical anaphors and pronouns in SALs
In Hmar and Mizo (TB) too, the bare reflexive amaʔ ‘self’ “cannot take an
antecedent in the local domain – because the VR is not present. It cannot take
a discourse antecedent, too” (Lalitha Murthy and Subbarao 2000: 793). Thus,
(10) and (11) above are ungrammatical if the VR is not present, and only a bare
nominal reflexive occurs.
The data above demonstrate that it is the VR that determines the local domain
for an anaphor, and its presence makes an antilocal anaphor function like a local
anaphor.
Notice that languages belonging to genetically different language families
have a similar lexical item, a verbal anaphor that performs almost identical
functions. Based on this fact, Subbarao (2000) points out that it is not for-
tuitous that such a phenomenon is found in human language. One can hope
that this phenomenon plausibly reflects the mental organization of language
that reflects cognitive capabilities of the human mind. Though functions such
as inchoative, verbal reflexive/reciprocal or self-benefactive or passive appear
outwardly to be unrelated, they all have the feature of transitivity in common –
that is, they are all [–transitive]. It is, therefore, reasonable to hypothesize that
such grammatical facts are grouped together under a single lexical head in
the mental lexicon. It is probably the cognitive capability of the human mind
that enables it mentally to classify phenomena that share cognitive/semantic
commonalities and to assign a single lexical category for such seemingly
unrelated phenomena across languages transcending genetic boundaries and
barriers.22
In this section, we have discussed the various functions that the verbal
anaphor performs in SALs. In the following section we shall discuss the occur-
rence of anaphors in non-nominative subject constructions.
Kashmiri (IA)
(52) kōrii chu panuni bōy pasand
girl.dat is self’s brother like
‘The girli likes self’si brother.’
(Wali et al. 2000: 486)
Maithili (IA)
(53) hari- kẽi apani ghar nahi ch- ainh
Hari dat self’s house not be pres.3hon.non-nominative
‘Hari doesn’t have his own house.’
(Yadava 2004: 257)
Telugu (DR)
(54) rōjā- kii tanai amma ant.ē is.t.am
Roja- dat self’s mother quot is pleasing
‘Rojai likes heri mother.’
(Subbarao and Lalitha Murthy 2000: 238)
Hindi-Urdu (IA)
(55) saritā ko apne (āp) par bharosā thā
Sarita f,s dat self’s self upon confidence m,s was m,s
‘Sarita had confidence in herself.’
Marathi (IA)
(56a) lili- lai (āpan.) swatāh∗ i āvad.te/ āvd.un- ghete
Lili- dat self [a bare reflexive] likes/ likes- VR
‘Lilii likes self∗ i .’
marker can have a dative subject as its antecedent, contrary to Kashi Wali’s
claim, as in (56c).
Telugu (DR)
(57) kamalai ki [tana- mı̄da tana- ki]i kāpam waccindi
Kamala dat self- on self- dat anger came
‘Kamala got angry [with] herself.’
(Subbarao and Saxena 1987: 126)
Hindi-Urdu (IA)
(58) *rām- koi sirf apne āpi acchā lag-tā hai
Ram- dat only self’s self.nom good strike-imperf is
‘Ram likes only himself.’
(Yamabe 1990; Experiencer as subject [as quoted in Davison
2004: 155, sentence (38a)])
Kashmiri (IA)
(60) kōrii chu ləd.k/ (*pāni / *panun pāni ) pasand
girl.dat is boy.abs self.nom self’s self.nom like
‘The girl likes the boy/*herself.’
(Wali et al. 2000: 486)
However, certain verbs, such as bōzun ‘to find/see’ or khɔš karun ‘to like, make
happy’ permit a nominative case-marked complex anaphor (Wali et al. 2000:
486–487).
(61) mei chu n panun pāni bozn yivān
I.dat be not self’s self.nom find come.pres pple
‘Ii am unable to find myselfi .’
(Wali et al. 2000: 487)
Telugu (DR)
(62) kāvyai ki tanu ant.e∗ i,j cālā is.t.am
Kavya dat self.nom quot.cond a lot pleasing
‘Kavyai likes *herselfi /herj a lot.’
In contrast to most of the SALs, Bangla (IA), Malayalam, Tamil (DR) and Bodo
(TB) are the only four languages that we know of that permit an accusative
case-marked anaphor in a genitive/dative subject construction.
Bangla (IA)
(63) gopal- er nije ke bhalo lage
Gopal- gen self acc good feels
‘Gopal likes himself.’
The question that arises now is: is there any principled reason for such a
restriction on the occurrence of nominative case-marked or accusative case-
marked anaphors in Hindi-Urdu, Punjabi (IA) and Telugu (DR)? Let us first
discuss the case of the accusative case marking. Dative/genitive predicates are
[–transitive]. Hence, the predicate in such constructions cannot assign/check
the accusative case of the theme.
The fact that Tamil, Malayalam (DR) and Bangla (IA) permit an accusative
marker with the theme in the dative/genitive subject construction is not a
counterargument against the claim that the predicate is [–transitive] in such
constructions. We will argue in chapter 5 that this marker should be treated as
a specificity marker, and, therefore, it is not problematic.
72 Lexical anaphors and pronouns in SALs
Hindi-Urdu (IA)
(64) rām ko yah bāt acchı̄ lagı̄
Ram dat this thing f,s, nom good f,s appeared f,s
‘Ram liked this matter.’
Indo-Aryan
The subject in a passive sentence corresponds to the direct object of an active
sentence. In Hindi-Urdu, “the object of an active transitive verb is normally
marked with the dative-accusative postposition if it has human reference,” as in
(65a) (Davison 2000: 413). The object of a transitive verb, which becomes the
derived subject in a passive sentence, may be: (i) nominative case-marked as in
(65b), or (ii) dative-accusative case-marked retaining the original postposition
ko of the active sentence, as in (65c).
In Kashmiri (66) and Marathi (67) too, the derived passive subject can antecede
the possessive reflexive.
Kashmiri (IA)
(66) aslam- asi āv pannisi bāg-as- manz lāyn
Aslam- dat pass self.dat garden- in beat.inf
‘Aslami was beaten in self’si garden.’
(Wali et al. 2000: 487)
Marathi (IA)
(67) lili- lāi kāl (āp-lyā) swatāh-cyāi kholit d.āmbla gela
Lili- dat yesterday self’s room.in dumped pass
‘Yesterday Lili was dumped in self’s room.’
(Wali 2000: 529)
In Punjabi the passive subject cannot antecede an anaphor for one set of speakers
(Bhatia 2000: 661), as in sentence (68): “In some dialects of Punjabi all oblique
subjects including instrumentals and locative subjects are permitted to corefer
with a reflexive, and for these dialects [(69) below] is grammatical” (Bhatia
2000: 661).
74 Lexical anaphors and pronouns in SALs
p a s s i v e s u b j e c t a s a n t e c e d e n t t o a n a n a p h o r : no t p e r m i t t e d
Punjabi (IA)
(68) *mun.d.ei tõ apn.e∗ i,∗ j / o’i de kamre dı̄ ku
jı̄ gavācı̄
boy instr self/ he of room of key f,s lose.pst.f.s
*‘The boyi (inadvertently) lost the key to hisi room.’
(Bhatia 2000: 661)25
Note that (68) is a lexical passive while (69) is a syntactic passive with abilitative
meaning.
(69) mε̃- thõ26 apn.ı̄ kitāb pár.ı̄ ná ˜ı̃ gayı̄
I- by self’s book read.perf.f.s not go.pst.f.s
‘I could not read this book of mine.’
(Om Arora, p.c.)
Dravidian
The use of the passive in Telugu is rather infrequent. It is only the possessive
form of tan ‘self’ that can occur in a passive sentence.
Telugu (DR)
(70) avinı̄tiparulaina udyōgastului tamai il.l.a- ki pamp.a- bad.d.-ārui
corrupt officials selves’ houses- dat send.inf- pass-agr
‘The corrupt officialsi were sent to theiri houses.’
(Subbarao and Lalitha Murthy 2000: 241)
predicate is formed out of a [–transitive] verb, as in (73) from Telugu (DR) and
(74) Kannada (DR) (see chapter 5 for details).
Telugu (DR)
(73) karun.ai ki tana mı̄da tanai ki kōpam vacc- (*kon)- in- di
Karuna dat self on self dat anger come- VR- pst- 3s,nm
‘Karunai got mad at herselfi .’
Kannada (DR)
(74) rāman-igei tanna mēle nēi,j.∗ k kōpa bantu (anta sı̄tej hēl.idal.u)
Ram-dat self on emph anger came that (COMP) Sita said
‘(Sitaj said that) Rami was angry with selfi,j.∗ k ’
(Amritavalli 2000: 57)
Amritavalli (2000: 54) points out: “there is a distinction between the syntac-
tically generated VR that occurs as an anaphor, and the inchoative VR that
occurs as an intransitive (anticausative) marker.” She further points out: “the
experiencer stative construction is an intransitive construction, and that syn-
tactically generated kol..lu [the VR in Kannada] (as against lexical inchoative
kol..lu) cannot occur in intransitive constructions” (Amritavalli 2000: 54).
Thus, the reason for the non-permissibility of the VR is that the dative
predicate itself is [–transitive], and, hence, a VR which is [–transitive] is not
needed. This could be due to a general constraint in language on duplication of
features with the same value.
However, a set of transitive verbs, such as kōs ‘cut,’ tat..t ‘pat,’ ad.d. ‘stop’
and kot..t ‘hit,’ permit the occurrence of the VR in Telugu (DR) in the dative
subject construction. The reason for such an occurrence is that the verbal
reflexive functions as a detransitivizer and it detransitivizes the transitive verb
(see sentence (32)).
Chinese ziji and Korean casin permit long-distance binding, while their respec-
tive polymorphemic forms zibun-zisin/kare-zisin, ta ziji and caki-casin do not
(see Cole and Sung 1994 for further details). In SALs too, it is the possessive
anaphor or the monomorphemic form of the nominal anaphor that permits long-
distance binding, while a polymorphemic form does not – except in Marathi
(Wali 2000: 534). Further, the occurrence of the verbal anaphor in the embed-
ded clause blocks long-distance binding.
Some of the major features of long-distance binding in SALs are:
(i) all long-distance anaphors are subject-oriented (Chomsky 1986);
(ii) long-distance binding is not permitted from a finite clause – only non-finite
clauses permit it;27
(iii) it is the monomorphemic form that permits long-distance binding, whereas
polymorphemic anaphors do not; hence, reciprocals, which are in general
polymorphemic in SALs, do not permit long-distance binding.28
Cole and Sung (1994) proposed ‘head raising analysis’ to account for long-
distance binding in Chinese. Under this analysis, the anaphor moves to INFL
of the lower clause, and then to the higher clause, where it is coindexed with a
long-distance antecedent.
We present an example from Hindi-Urdu to exemplify the phenomenon of
long-distance binding.
Hindi-Urdu (IA)
(75) ašoki ne lalitāj se [S2 PROj apne- liyei,j cāy banā-ne
Ashok (m) erg Lalita (f) with self- for tea make- to
ko S2 ] kahā
in order said
‘Ashok asked Lalita to make some tea for himself/herself.’
(Subbarao 1971: 191)
In (75), apne-liye ‘for self’ can be coindexed with the embedded subject PRO
of S2 , which, in turn, is coindexed with the matrix subject lalitā, thus permitting
binding of an anaphor which obeys Principle A. However, apne-liye ‘for self’
can also be coindexed with the subject of the matrix sentence ašok which is
a long-distance antecedent. Such interpretation of the anaphor is an apparent
violation of Principle A of the Binding Theory which states that an anaphor must
be bound in its governing category. One of the ways to account for long-distance
binding of the simplex anaphor apne-liye ‘for self’ without violating Principle A
is to move the anaphor by the head-to-head movement rule. INFL consists of the
interpretable features of TENSE and ASPECT and uninterpretable features of
AGR (Chomsky 2001). According to Davison (2001: 57), the monomorphemic
anaphor first moves to the non-finite TENSE of its own clause, and then, by
successive cyclic movement, it moves to the finite TENSE of the next higher
clause. It is then coindexed with the subject of the higher clause. In Cole and
3.7 Long-distance binding and Principle A 77
Hmar (TB)
(79) [S1 pārii - n lalij [S2 PROj ama∗ i/j in- that- lo- diŋS2 ]- in a- tiS1 ]
Pari- erg Lali self VR- kill- not- fut- comp 3s- say
‘Parii asked Lalij not to kill herself∗ i/j .’
(Mukerjee, Subbarao and Walia 2003: 111)
Ho (Munda)
(80) [S1 somai rāmj [S2 PROj goe- n∗ i/j - tiy a- eiS2 ] kaji- ad.- aS1 ]
Soma Ram kill- VR- to- 3s tell- pst[+tr]- fin
‘Somai told Ramj to kill himself∗ i/j .’
(Koh and Subbarao, ms)
That is, when the VR is present in S2 , the anaphor cannot cross the VR to go
further up, and thus the VR is instrumental in blocking long-distance binding.
It insulates the anaphor from long-distance coindexation. This phenomenon
has been observed elsewhere too (Everaert 1991; Kazenin 2001).
3.8.1 Indo-Aryan
Hindi-Urdu (IA)
(81) pramodi [apne (āpi/∗ j )-/ khud-i/∗ j / swayam-i/∗ j ko par.hākū] nahı̃˜
Pramod self’s self- self- self- acc nerd not
samajh-tā (hai)
understand-imperf is
‘Pramodi does not consider [himselfi a nerd].’
(Davison 2000: 420)
In Marathi (IA) and Oriya (IA) too, an anaphor occurs in small clauses.
Marathi (IA)
(83) lilii (āplyā) swatāh- lāi libral samajhte
Lili herself- acc liberal considers
‘Lili considers herself liberal.’
(Wali 2000: 534)
3.8.2 Dravidian
In Telugu (DR) too, the matrix predicate that is [+transitive] permits a complex
anaphor when the subject antecedent is in the nominative case (84) and a
[–transitive] dative predicate permits “only the simplex form of the anaphor in
nominative case” as in sentence (85): “the matrix verb which is an intransitive
verb cannot assign accusative case to the embedded subject by exceptional case
marking” (Subbarao and Bhaskararao 2004: 178).
80 Lexical anaphors and pronouns in SALs
Telugu (DR)
With a [+transitive] predicate bhāvincu ‘consider’
(84) mādhuri tana- ni tanu andagatte- gā bhāvistundi
Madhuri self- acc self pretty- adjr considers.3s.nm
‘Madhuri considers herself pretty.’
(Subbarao and Bhaskararao 2004: 178)
In Malayalam, only the complex form awan-e tanne-e ‘he-acc emph’ is per-
mitted as the subject of the embedded clause; a simplex form is not permitted
(86).
Malayalam (DR)
(86) rāmani awani - e tanne-e/ *tanni - e oru mid.ukkan
Raman he- acc emph self- acc one clever person
āyi karuti
being considered
‘Ramani considered himselfi a clever man.’
(Sentence (86) is a consolidated version of sentence (48) on
page 131, and sentence (130) on page 144, of Jayaseelan 2000)
3.8.3 Tibeto-Burman
In Mizo (TB), a simplex form of the anaphor occurs in copular complements,
and co-refers with the nonlocal subject. In Mizo, verbs such as ‘consider,’
‘believe,’ etc., inherently carry a verbal reflexive, as they invariably carry the
verbal anaphor. The form a maʔ ‘self’ as the subject of the small clause
co-refers with the matrix verb only when the VR is present. Otherwise, it
does not. The form a maʔ refers to a discourse antecedent in such syntactic
constructions, as in sentence (88) (Lalitha Murthy and Subbarao 2000: 803).
In sentence (87), inti ‘consider’ in Mizo consists of -in ‘VR’ and ti ‘say.’ The
VR -in such constructions in Mizo, and kon- in Telugu in anukonu ‘consider,’
is grammaticalized and, hence, cannot be treated as a verbal reflexive.
3.8 Small Clauses and binding 81
ecm with a vr
Mizo (TB)
(87) zovai - n [a- maʔi,∗ j - cu lāmthiam] a- inti
Zova- erg 3s- self- foc dancer 3s- consider
‘Zova considers self a dancer.’
(Lalitha Murthy and Subbarao 2000: 803)
ecm without a vr
(88) zovai - n [a- maʔ∗ i,j - cu liberal] a- hria
Zova- erg 3s- self- foc liberal 3s- know
‘Zova knows self∗ i/j is a liberal.’
(Lalitha Murthy and Subbarao 2000: 803)
Manipuri (TB)
(89) tombii na masai - bu phaza- hayna lau- (jai )- i
Tombi nom self- acc pretty- comp consider- VR- pres
‘Tombi considers herself pretty.’
(90) tombii na laugij bu phaza hayna lau- (*jai )- i
Tombi nom Laugi acc pretty comp consider VR- pres
‘Tombi considers Laugi pretty.’
Khasi (Mon-Khmer)
(91) u- lεmi u- khεiñ ba pro ui - lɔŋ u- ba- bhabrēw
3 m,s- Lem 3 m,s- count that 3 m,s- be 3 m,s- adjr- beautiful
‘Lemi thinks that hei is handsome.’
3.8.5 Summary
The question that needs to be addressed is: do the cases involving ECM sub-
jects and the coindexation of the lexical anaphor with the matrix subject
involve long-distance binding? It does not appear to be the case. The fact
that the verbal anaphor occurs in Manipuri (TB) with the matrix verb when
an anaphor occurs as the subject of the small clause shows that the subject
of the small clause is the DO of the matrix clause. That is how it gets case-
marked. Therefore, such cases cannot be treated as instances of long-distance
binding.31
In this section we have shown that, in cases involving ECM:
(i) a dative case-marked subject in Hindi-Urdu (IA) does not permit either
a nominative case-marked or accusative case-marked embedded subject
(sentence (58)), while in Bangla a genitive case-marked matrix subject
permits an accusative case-marked enbedded subject as in (82);32
(ii) in Telugu (DR), a dative case-marked subject can be coindexed with only
the simplex form of the anaphor, and not the complex form, while in
Malayalam the complex form of the anaphor is permitted; and
(iii) the coindexation of a lexical anaphor with the matrix subject in ECM
constructions does not count as long-distance binding.
Hindi-Urdu (IA)
(92) rāmi ne [apnei / us∗ i - ke- nazdı̄k] ek sarp dekh-ā
Ram erg self’s 3s.gen- near one snake see-perf
‘Rami saw a snake [near selfi ].’
(Davison 2000: 409)
Malayalam (DR)
(93) rāmani tani - te munn- il oru āna- ye kan.d.u
Ram self- gen front- in one elephant- acc saw
‘Rami saw an elephant in front of selfi .’
(Jayaseelan 2000: 126)
3.10 Binding principles A, B and C 83
Mizo (TB)
(94) zova- ni rūl amaʔi,∗ j- bulaʔ a- hmu
Zova- erg snake self- near 3s- saw
‘Zovai saw a snake near selfi,∗ j .’
(Lalitha Murthy and Subbarao 2000: 794)
Kannada (DR)
(95) rāmai tannai - mēle tappu hāki- kon.d.a- nu
Rama self.gen- on blame put- VR- agr
‘Ramai put the blame on himselfi .’
(Amritavalli 2000: 63)
Telugu (DR)
(96) saritai tana mı̄da (tanui ) nēram mōpu- kon- di
Sarita self.gen on self blame put- VR- agr
‘Saritai put the blame on herselfi .’
Thus, a polymorphemic anaphor and the VR are permitted only if the locative
PP is a subcategorized argument. If it is not, neither of them is permitted.
Malayalam (DR)
(100) rāmani wicāriccu [rāmani jayikkum ennə]
Raman thought Raman will win comp
‘Raman thought that Raman would win.’
(Jayaseelan 2000: 163)
Jayaseelan (2000: 163) points out: “the repeated proper name seems to behave
like a pronoun in that it apparently obeys Principle B.” Evidence in support of
this claim comes from the fact that a proper name that is an argument of a verb
cannot be coreferential with another argument of the same verb.
Jayaseelan (2000) further points out that, just as in the case of pronouns, the
above sentence becomes acceptable by adding tanne, the emphatic.
However, according to Mahajan (1990) šer ‘tiger’ may also be the antecedent
of apne ‘self’s’ in sentence (107), though it did not move to subject position, as
his dialect does not have a strict subject condition. Dayal (1994) does not agree
with Mahajan (1990) that scrambled NPs can serve as antecedents to anaphors.
She argues that the facts concerning binding do not support Mahajan’s claim,
and her claim is that, as far as binding goes, “scrambling is an instance of A-bar
movement only, not A-movement” (Dayal 1994: 239).
Based on sentences such as (108), Mahajan (1990) argues that scrambling
moves an NP to an A-position, as the DO binds the possessive anaphor apne
baccõ-ne ‘self’s children-ERG.’
Dayal’s contention is that sentences such as (108) are not acceptable to most
of the speakers of Hindi, and that Mahajan’s analysis fails to account for the
ungrammaticality, as the scrambled NP mohan-ko ‘Mohan-acc’ should bind
the possessive reflexive, since it is in an argument position. Dayal argues that
if scrambling is treated as A-bar movement, then the scrambled NP mohan-ko
‘Mohan-acc’ cannot bind an anaphor as “the DO would be reconstructed and
the resulting violation of Principle A would account for its ungrammaticality”
(Dayal 1994: 242).
In Telugu (DR) and Mizo (TB), sentences parallel to (108) are ungrammati-
cal, and if scrambling is treated as A-bar movement, the ungrammaticality can
be accounted for.
Telugu (DR)
(109) *pillal(a)- ni tama amma kot.t.indi
children.obl- acc selves’ mother hit
‘*Selves’ mother beat children.’
Mizo (TB)
(110) *zova- cu amaʔ nupui- in a- hau
Zova- acc self’s wife- erg 3s- scolded
‘*Self’s wife scolded Zova.’
3.12 Blocking Effects 87
The interested reader may refer to Mahajan (1990), Dayal (1994) and Kidwai
(2000) for further details.34
Malayalam (DR)
(112) *rāmani wicāriccu kris.n.an nin- akkə tan∗ i - ne paricayapped.utt-um
Raman thought Krishnan you- dat self- acc introduce-will
ennə
comp
*‘Ramani thought that Krishnan would introduce self∗ i to you.’
(Jayaseelan 2000: 133)
88 Lexical anaphors and pronouns in SALs
To sum up, SALs that exhibit subject–verb agreement have no Blocking Effects
and conform to the generalization made in Cole and Sung (1994: 364): “the
typological property shared by languages that manifest the blocking effect is
the absence of agreement on verbs and other predicates: Chinese and Korean
lack verb agreement and show the blocking effect, whereas Italian and Icelandic
have agreement and lack the blocking effect.”
Telugu (DR)
(116) mamatai ki tana mı̄da tanai ki kōpam vaccindi
Mamata dat self on self dat anger came
‘Mamatai got angry with herselfi .’
nominative subject
(117) vāl..lu okal.l.a- tō okal..lu debba lād.u- kon- t.unnā- ru
they (nom) one- with one (nom) fight- VREC- progr- m.p
‘They are fighting with each other.’
dative subject
(118) vāl..la- ki okal.l.a- tō okal..la- ki sariggā pad.adu
they- dat one.gen- with one.gen- dat well get along.neg
‘They do not get along well with each other.’
anaphor in do position
Manipuri (TB)
(119) makhoi- na ama- na ama- bu thagat- na- rammi
they- nom one- nom one- acc praise- VREC- pst
‘They each praised the other.’
90 Lexical anaphors and pronouns in SALs
anaphor in io position
(120) makhoi- na ama- na ama- da khudopot pi- na- rammi
they- nom one- nom one- dat gift give- VREC- pst
‘They gave gifts to one another.’
(Sarju Devi and Subbarao 2003: 69)
It should also be mentioned that there are two other forms of the reciprocal,
ama ga ama ga ‘each other’ and makhoi masel ‘each other’, and there is no
Case Copying in them (Sarju Devi and Subbarao 2003).
Both in Manipuri (TB) and Telugu (DR), reduplication is productively used
to create a morphological complex that refers to an anaphor, and it is Case
Copying and a lexical case-marker assignment that are used to case-mark the
constituent parts of the bipartite structure of the anaphor. There is, however,
a minor difference: while Case Copying of the subject takes place on the first
part of the reduplicated structure of the anaphor in Manipuri, it is on the second
part of the reduplicated structure that Case Copying takes place in Dravidian.
In Sarju Devi and Subbarao (2003), we pointed out that Case Copying is a
phenomenon that belongs to syntax, and reduplication is a process that belongs
to compound-formation strategies of language. It is, however, the interaction of
these two phenomena that results in the formation of anaphors in two genetically
unrelated languages.
Malayalam (DR) has two forms of the complex anaphor as in (i) and
(ii) below:
Jayaseelan (2000) treats the second part of the complex anaphor as an emphatic
in Malayalam (DR), unlike the Telugu and Manipuri cases. Sentence (122) is
illustrative:
Malayalam (DR)
(122) rāmani tanni/∗ j - e tanne / awani/∗ j - e tanne wer.uttu
Raman self- acc emph he- acc emph hated
‘Ramani hated himselfi/∗ j .’
(Sentence (122) is a consolidated version of sentences (112)
and (117) on pages 141–142 of Jayaseelan 2000)
Note that the second part of the anaphor carries the emphatic marker -e.
Jayaseelan treats the second part of the anaphor as an emphatic, and not as
a reduplicated form (Jayaseelan 2000: 142).
3.15 Conclusion 91
3.15 Conclusion
This chapter first provides a brief description of the various forms of the
anaphor in each language family. SALs have a nominal simplex and complex
anaphor and/or a monomorphemic verbal anaphor (reflexive/reciprocal). The
complex form of the anaphor in some SALs is the result of reduplication
and case copying, and in others it is formed by compounding a simplex form
with an emphatic. As for the nature of the verbal reflexive, which is invariably
monomorphemic, we have shown that the VR, the passive and the anticausative
(inchoative) reduce the valence of a transitive predicate by one argument. The
verbal reciprocal, too, performs several other functions, such as group marker,
hortative marker and collaborative effort marker.
SALs present a gradation with respect to the nature of anaphoric binding.
The IA languages are on one end of the scale with their insistence on the
occurrence of the nominal anaphor, and the polysynthetic Munda languages
are on the other end of the scale with their insistence on the occurrence of the
verbal anaphor alone, while the Dravidian languages (except Malayalam and
Toda), most of the Tibeto-Burman languages and Khasi (Mon-Khmer) fall in
between with a nominal as well as a verbal device for anaphoric binding. The
verbal anaphor in Dravidian and Munda appears to be indigenous. However,
as Toda (DR) does not have a VR, and Toda retained most of the features of
Proto-Dravidian (Peri Bhaskararao, p.c.), it could be the case that the verbal
anaphor in the rest of the Dravidian languages is the result of a subsequent
grammaticalization process, and is not indigenous.
With regard to the VR in Tibeto-Burman, one might conjecture that early
Tibeto-Burman had the device, but in course of time it got lost in some lan-
guages. In-depth research is needed before any firm conclusion can be drawn.
Munda languages also use a nominal device borrowed from superstrate lan-
guages, and Korku (Munda) has totally lost the verbal device. It is to be noted
that Marathi (IA) and Sinhala (IA) have the VR, and it could be due to contact
with Dravidian. The occurrence of the VR in Gujarati functioning as a VR
and a passive marker, as in Munda languages, is hard to explain. It could be
an indigenous development. We have also shown that the Anaphor Agreement
Principle rules out the occurrence of nominal subject anaphors in Hindi-Urdu,
Kashmiri and Marathi, which have object agreement. However, the occurrence
of such anaphors in Mizo and Hmar (TB), which have subject as well as object
agreement, cannot be explained by the aforementioned Principle, and further
research needs to be done in this area.
A non-nominative subject can antecede an anaphor. However, a verbal
anaphor is not permitted in a non-nominative subject construction, as the pred-
icates in such constructions are inherently [–transitive], and, hence, the VR –
which is [–transitive] – cannot co-occur. SALs (except Marathi) obey the three
92 Lexical anaphors and pronouns in SALs
4.1 Introduction
SALs exhibit a variety of agreement patterns in the verb, ranging from no
agreement on one hand to agreement with arguments (subject, direct object,
indirect object, etc.) and non-arguments (oblique objects) on the other. The
study of agreement in these languages becomes more fascinating coupled with
the fact that there is incorporation of arguments, non-arguments, quantifiers,
intensifiers, Negative Polarity Items operating in these languages. We shall
provide two examples to demonstrate the two ends of the spectrum: one from a
language with no agreement, and the other from a language with rich agreement.
In Manipuri (TB), the verb exhibits no agreement at all with any argument
(1), while in Hmar (TB), the verb exhibits rich agreement with the subject and
incorporates the Negative Polarity Item (NPI) la ‘yet’ (2).1
Manipuri (TB)
(1) əyhak cen-ø niŋ- ŋi
I hon run-inf want- [−fut]
‘I want to run.’
Hmar (TB)
(2) pro cainā ram hmun pou pou- ah ka- la- fe- nɔ
I China land place all- loc 1s- yet-(NPI) go- not
‘I have not yet gone to all the places in the land of China.’
The study of agreement has been one of the areas of interest in descriptive
grammars. However, in recent years, agreement in relation to case has received
considerable attention, and has become one of the focal areas of the Government
and Binding framework, and its offshoot, the Minimalist Program of Chomsky
(1995a, 1995b, 2000 and 2001). Agreement can be viewed as a syntagmatic
relationship that is established between a lexical element (a noun phrase or an
adverbial PP) and a functional category, such as tense or aspect. The agreement
markers that are manifested on the functional categories are in terms of Person,
Number and Gender, which are generally termed as the PNG or “phi-features.”
93
94 Case and agreement
long-distance agreement (LDA). Section 4.6 focuses on the role of the principle
of Pronominal Strength Hierarchy in agreement. Section 4.7 deals with ECM
cases and LDA. Section 4.8 shows how a functional category such as agreement
of a language influences the patterns of agreement of a language belonging to
another family in language contact situations. Section 4.9 presents evidence
from Mizo and Hmar (TB) where the occurrence of ergative and agreement
markers is mutually dependent. The final section concludes the chapter.
theme–verb agreement
Hindi-Urdu (IA)
(3) lər.kõ nei yah kitābj par.h- ı̄∗ i/j
boys, m,p erg this book, f,s read- pst, f,s
‘The boys read this book.’
(Davison 1991: 2)
default agreement
The agent pulis ‘police’ and patient cor ‘thief’ in (5) are both postpositionally
case-marked in the passive in Hindi-Urdu.
When the subject is case-marked dative/locative, the verb agrees with the theme
as in (7) in Kannada (DR).
Kannada (DR)
(7) nanagei mūvaru hen.n.uj makkal.u iddāre∗ i/j
I.dat three female children be-pst.3 p.human
‘I have three daughters.’
(Sridhar 1990: 133)
In Tamil (DR), the verb exhibits default agreement in the dative subject con-
struction when the subject is dative case-marked, and the theme too is postposi-
tionally case-marked by a specificity marker (accusative). Default agreement is
due to the fact that the experiencer kumār ‘Kumar’ and the patient rājā ‘Raja’
in (8) are both postpositionally case-marked, and hence agreement is blocked.
Thus, the predicate pit.ikk ‘like’ exhibits default agreement; hence, there is no
agreement with any NP in the sentence. The default agreement marker in Tamil
is in the neuter.
Subset 1
In Gujarati, a postpositionally case-marked direct object triggers agreement
(Mistry 1998) as in (9). Ergative case-marked subjects cannot trigger agree-
ment. Thus, in (9), though both the ergative subject and the direct object are
postpositionally case-marked, it is the direct object (patient) that triggers agree-
ment, and not the subject.
98 Case and agreement
Gujarati (IA)
(9) kišori - e rājj ne pajav- y- o∗ i,j
Kishor- erg Raj, m acc harass- perf- m
‘Kishor harassed Raj.’
(Mistry 1998: 426)
According to Mistry (1998), Gujarati (IA) has the marker -ne, which functions
as an object marker as well as a specificity marker. The object marker acts as a
barrier to the verb agreement in (11), just as in Hindi-Urdu and Punjabi (IA),
while the specificity marker -ne does not in (9).
Subset 2
Maithili (IA) and Kurmali (IA) are spoken in Bihar, northern India, where
the verb exhibits agreement in honorificity with an NP that is postpositionally
case-marked. We shall first discuss the case of Maithili (IA). The verb exhibits
agreement with indirect object ahā˜ ‘you’ dative case-marked by ke, in (12).
postpositionally case-marked io
Maithili (IA)
(12) hami ahā˜-kẽj pāik de- l- ahũj
I you.hon-dat money give- pst- agr.hon
‘I gave you money.’
(Singh 1979)4
Unlike in most of the SALs, the verb exhibits agreement in honorificity with a
dative subject – hun-kā ‘he’ in (13) – or with a passive subject as in (14).
4.2 Agreement typology of SALs 99
In Maithili, the verb exhibits agreement with an adjunct (an ablative PP) which
carries an overt lexical case marker as in (15). The ablative PP in (15) is
hamara sɔ∼ ‘from me,’ and the verb agrees with the ablative object, which is
a subcategorized argument.
Yadava (1999: 149) comments: “This triple agreement, is, however, heavily
constrained, in that it permits only the combination of 1.nh [non-honorific] and
2.mh [mid-honorific] and 3.h [honorific] referents.”
To summarize, the following are the characteristic features of Maithili
agreement:
100 Case and agreement
(i) though there is no overt ergative marker, the verb exhibits agreement with
direct object, and an indirect object, and the possessor of direct object too;
(ii) unlike in most other SALs, the verb exhibits agreement with a dative
subject, a passive subject, an object and an ablative postpositional phrase;
(iii) the verb exhibits triple agreement markers, a phenomenon rarely found in
SALs.
The verb exhibits long-distance agreement too, with the possessor of the direct
object of the embedded clause (see section 4.5).
We shall now discuss the case of Kurmali, an Indo-Aryan language spoken
in Bihar in northern India, in which the verb exhibits multiple agreements, with
the possessive pronoun of the subject (possessor) and the subject as well. In
(17) below, the 2nd person possessor agreement marker is -un ‘2 plural,’ and
of the subject is -o ‘3 singular.’
possessor agreement
Kurmali (IA)
(17) toharj kalam-tii gir- l- oi - unj
your.gen.2p pen.def.nom.3 s.f- fall- pst- 3s- 2p
‘Your (2p) pen fell down.’
(Mahto 1989)
The verb in Kurmali also exhibits agreement with the subject and a dative
case-marked indirect object as in (18). The subject agreement marker in (18)
is -o and the indirect object marker is -u. The subject and the indirect object
trigger agreement though thay are lexically case-marked.
Magahi (IA)
(19) ham to(h)rā dekh- l- i- o
I you (h) see- pst- 1s- 2p
‘I saw you (honorific).’
(Verma 1991: 132)
4.2 Agreement typology of SALs 101
From a typlogical point of view, Indo-Aryan languages discussed thus far fall
into two major categories with regard to agreement: languages such as Hindi-
Urdu, Punjabi, Marathi, Gujarati come under Category I where the verb agrees
only with one argument, and languages such as Maithili and Kurmali come
under Category II where the verb exhibits agreement with two, or sometimes
even three, arguments.5
In Shina of Skardu, an Indo-Aryan language too, the verb agrees with the
dative case-marked experiencer.
Kashmiri (IA)
(22) tsei von- uj - thi timan apuzj
you.erg told- m,s- 2s them.dat lie.abs.m,s
‘You told them a lie.’
(Wali and Koul 1997: 250)
The ergative case-marked subject in 1st and 3rd person does not trigger agree-
ment. However, if it is pro-dropped, the verb must exhibit agreement with the
pro-dropped subject (see Wali and Koul 1997: 250). In Punjabi, “(I)n perfec-
tive tense the 1st and 2nd person subjects are not overtly marked with the
postposition ne” (Bhatia 1993: 170).
102 Case and agreement
Sora (Munda)
(24) proi jom- bɔŋj - t- εj - n- jii pɔ?
(they) eat- buffalo- [-pst]- 3s- [−tr]- 3p pol q mkr
‘Will they eat the buffalo?’ or ‘Do they eat buffalo?’
(Ramamurti 1931: 25)
Hmar (TB)
(25) lalii - n proj ai - mij - hmu
Lali- erg me 3s- 1s.DO- met
‘Lali met me.’
The occurrence of the indirect object i-koma ‘for you’ is optional in (26), and
the pronominal clitic -ce ‘2 singular’ is encoded in the verb.7
104 Case and agreement
Ho (Munda)
(27) birsa proi tam- sida- ke- d.- i ŋi - a
Birsa (me) hit- first- pst- [+tr] 1s.DO- decl
‘Birsa hit me first.’
(Burrows [1915] 1980: 89)
The pronominal clitic -iŋ of the indirect object is incorporated into the verb in
(28) in Gorum (Munda).
Sora (Munda)
(29) pro lēm- jēŋ- t- am
(I) bow- legs- [-pst]- your
‘I bow to your legs.’
(Ramamurti 1931: 43)
4.2 Agreement typology of SALs 105
Ho (Munda)
(30) jemai aye- ʔa ote ria daʔa amaj ote- ei - em- ad.-
Jema he- gen field gen water your field- 3s- give- pst perf-
mij - a
2s- fin
‘Jema had given his field’s water to your field.’
(Koh and Subbarao, ms)
Santali (Munda)
(31) kora- dɔ kuri- (y)e dal- ke- d- e- ti-
- a
boy- topic girl- 3s beat- pst- [+tr]- 3s- gen- 1s- fin
‘The boy beat my girl.’ (*‘My boy beat the girl.’)
(Macphail 1983: 41)
Hmar (TB)
(32) zova- n ka- kut a- mi- sɔ:p- pek
Zova- erg my- hands 3s- 1s- washed- o ben
‘Zova washed my hands.’
(Subbarao 1998b)
Hmar (TB)
(33a) pro án- fāk- vɔŋ
(they) 3 p- eat- everything
‘They eat everything.’
(33b) pro án- fāk- sεm/seŋ
(they) 3 p- eat- all
‘All of them ate.’
Hmar (TB)
(34) pro tājmahal i- hmu- tā- am
(you) Taj Mahal 2s- see- ever (NPI)- pol q mkr
‘Have you ever seen the Taj Mahal?’
(35) pro cainā ram hmun pou pou- ah ka- la- fe- nɔ
(I) China land place all- loc 1 s- yet (NPI)- go- not
‘I have not yet gone to all the places in the land of China.’
Note that the negative c-commands the NPI hierarchically, as it occurs to the
right of the NPI in (35).
Adverbs such as ve ve ‘together, both’ and inclusive particles such as ve
‘too, also’ occur to the right of the verb. In (36) and (37), huŋ- .tšaŋ ‘come and
participate’ is an instance of complex verb formation in Hmar (TB).
Hmar (TB)
(36) pro an- huŋ- .tšaŋ- ve ve
(they) 3 p- come- participate- together
‘They came and participated together.’
Jowai, Pnar and Rymbai (Mon-Khmer), which are also verb-medial, are con-
sidered as “dialects of Khasi” (Mon-Khmer). An adverb is incorporated into
the verb in these languages too.8 Similar incorporation of nouns, adverbs and
several other elements is found in Savara/So-ra (Munda), spoken in Orissa.
Thus, though Savara/Sora belongs to the Munda family, and Jowai, Pnar and
Rymbai belong to the Mon-Khmer family, incorporation is shared between
the languages of these two subgroups and, hence, we consider it an inherited
feature in these languages.
In Manipuri and Bodo (TB), and Khasi (Mon-Khmer), the adverb ‘together’
is incorporated in the verb (see sentences (93) from Manipuri, and (87) from
Khasi, in chapter 8).
In Mizo (TB), the emphasizer adjunct ‘certainly’ is incorporated.
Mizo (TB)
(38) vɔin.ah zova- cu sikul- (ah) a- kal- ŋei- aŋ
today.adv Zova- foc school- to 3 s- go- certainly- fut
‘Zova certainly will go to school today.’
(Subbarao and Lalitha Murthy, ms)
In Khasi (Mon-Khmer) too, the Negative Polarity Item (NPI) pat ‘yet’ is incor-
porated into the verb to the right of the negative.
108 Case and agreement
Khasi (Mon-Khmer)
(39) ka-kayt [ba- la- bām u-khnna?] ka-m- pat- iʔ
f-banana adjr- pst- eat m-child 3 f,s-neg- yet(NPI)- ripe
‘The banana that the boy ate was not yet ripe.’
Since the negative occurs to the left of the NPI in (39), it c-commands the NPI.
We have demonstrated that Munda and Tibeto-Burman languages permit
incorporation of an object and indirect object and possessor of direct and indi-
rect objects; also of quantifiers, Negative Polarity Items and inclusive particles.
Such incorporation is not permitted in Dravidian and Indo-Aryan languages.
Thus, the incorporation of various types of constituents in the verb is a special
feature of the Mon-Khmer, Munda and Tibeto-Burman language families, a
feature not found in Dravidian and Indo-Aryan languages.
Rabha (TB)
(40) aŋ/ ciŋ/ naŋ/ u naŋ- pake reŋ- a sɔyra- jɔ
I we you he/she you- with go- purposive promise- pres perf
‘I/we/you/he/she have/has promised to go with you.’
(Subbarao et al., ms)
In many of these languages there is a nominative marker that occurs with the
subject, which is generally optional.
To summarize the above discussion, it is the lexical case marker (P-case or
postpositional case) that blocks agreement in the first group of languages. It is
assigned either structurally, for example the accusative lexical case marker ko in
Hindi-Urdu or ni in Telugu, or the ergative ne in Hindi-Urdu. It may be a lexical
(inherent) case marker as the dative on indirect objects or the dative subject or
the locative subject. In cases in which both the agent/experiencer/possessor and
the theme/patient are case-marked with a postposition, the predicate exhibits
default agreement.
In the second group of languages, a postpositionally case-marked NP can
trigger an agreement.
In the third group, in the first set of languages, in a polysynthetic language
like Sora, arguments are incorporated into the verb, a phenomenon unique to
Sora among SALs. In the second set of languages, the agreement is manifested
as a clitic, and the occurrence of a postposition is of no consequence for an
agreement, as is the case in Tibeto-Burman languages. Though the subject
4.3 Adposition incorporation 109
is ergative case-marked in Mizo, Hmar and Paite (TB), it can trigger an agree-
ment, and the agreement marker is generally a clitic. In this group of lan-
guages, incorporation of an object, indirect object, possessor of direct and
indirect objects and also of quantifiers, intensifiers, Negative Polarity Items
and inclusive particles takes place.
In the fourth group of languages, there is no agreement at all.
The Tibeto-Burman family of languages exhibits varied agreement patterns –
from very rich agreement to none, a phenomenon found in no other South Asian
language family.
In this section, we have presented the various types of agreement patterns
found in SALs. In the following section, we shall consider a specific case of
incorporation of adpositions in Hmar (TB), and demonstrate that such incor-
poration affects the transitive nature of the host verb.
Hmar (TB)
(41) lali kei- le cou a- n-9 .tšɔŋ
Lali (nom) I- with only 3s- VR- sit
‘Lali sat only with me.’
In (43) and (44), the pronouns kei ‘I’ and naŋ ‘you’ do not carry any post-
position, as the postposition le ‘with’ is incorporated into the verb. The verb
instead carries the marker pui ‘with.’ Note that le and pui are allomorphs whose
position of occurrence is with the DP and the verb stem, respectively.
p o s t p o s i t i o n le ‘with’ i n c o r p o r a t e d i n t o t h e v e r b
(43) lali- n kei cou a- mi- n- .tšɔŋ- pui
Lali- erg I only 3s- 1s- VR- sit- with
‘Lali sat only with me.’
110 Case and agreement
p o s t p o s i t i o n -aʔ ‘o n’ o c c u r r i n g w i t h t h e np .tšut.na ‘c h a i r ’
Paite (TB)
(47) lali t.šut.na- aʔ a- t.šu
Lali (nom) chair- on 3s- sat
‘Lali sat on a chair.’
(51) Adjunction
If X and Y are X0 categories and X is adjoined to Y in the syntax, then X
precedes Y in linear order.
(Baker 1996: 29)
The case of Sora (Munda) incorporation becomes interesting in this case. Baker
(1996: 29) cites data from Sora – which permits noun incorporation, as we have
already seen – and the incorporated noun occurs to the right of the verb, not to
the left as the above generalization claims. Let us now examine how this can be
explained. Sentence (52) is a normal structure in which no incorporation has
taken place.
Sora (Munda)
(52) bɔŋtεl- ən- ədɔŋ jom- t- ε- ji pɔʔ
buffalo- noun suffix- acc eat- [-pst]- 3 s- 3p, sub mkr pol q mkr
‘Will they eat the buffalo?’ or ‘Do they eat buffalo?’
(Baker 1996: 32)
In (53), the direct object bɔŋtεl ‘buffalo’ is incorporated into the verb. Nouns
when incorporated have a contracted form in Sora. Thus, bɔŋtεl’s shorter form
is bɔŋ in (53).
4.4 Incorporation in polysynthetic languages 113
Sora (Munda)
(56)
εn əd- məl- jom- jεl- yɔ- aj- t- en- ay
I neg- want- eat- meat- fish- all- [-pst]- [−tr]- 1s, sub mkr
‘I don’t want to eat all the fish.’
(Baker 1996: 32)
Starosta’s comment gains support from the fact that in (56) in Sora, the
quantifier aj ‘all’ is incorporated in the verb to its right, just as in verb-final
languages such as Mizo and Hmar (TB), in which quantifiers are incorporated
to the right of the verb (see (33a and b) in Hmar). The agreement morphol-
ogy also follows the verb in Sora in (56). Lehman (1973: 57) concludes that
Proto-Austro-Asiatic was VO and non-agglutinative in morphological struc-
ture. Donnegan and Stampe (2004) provide further evidence to show that Sora
had “VO” (head-first) structure.
Based on such facts, Baker (1996) concludes that the order of morphemes
in Sora is a mirror image of that found in polysynthetic languages. Therefore,
“the direction of adjunction performed by head movement is simply reversed
in Sora” (Baker 1996: 33).
Baker (1996) assumes that most of the polysynthetic languages are head-
initial: SVO-type languages. He points out that polysynthetic languages such
as Ainu and Kiowa which permit incorporation have head-final tendencies.
Baker’s assertion gains support from Starosta’s comment that Sora, a Munda
language, which is polysynthetic in nature, was at one time non-verb-final,
and, hence, incorporation is facilitated. Though it has some head-initial ten-
dencies, it has verb-final tendencies too, which include agreement morphology
and incorporation of quantifiers to the right of the verb as in some verb-final
languages. Thus, a diachronic explanation is helpful in resolving apparent syn-
chronic anomalies.
Another issue that is of significant interest in SALs is that of long-distance
agreement which we shall discuss in the following section.
4.5.1 Indo-Aryan
Long-distance agreement has been the focus of attention in recent discussions
on Hindi-Urdu agreement (Hook 1979; Mahajan 1989, 1990; Davison 1991;
Butt 1993; Singh 1993; Subbarao 2001; Bhatt 2005; Chandra 2007; Davison
2007b). Mahajan (1990) and Davison (1991), to the best of our knowledge, are
the first two studies of long-distance agreement in Hindi-Urdu (IA).
Hindi-Urdu and Punjabi (IA) permit long-distance agreement only when the
embedded sentence contains a verb in the infinitival form. This is also the case
in Maithili (IA). The following examples from Hindi-Urdu and Punjabi are
illustrative.
l o n g - d i s t a n c e a g r e e m e n t w i t h t h e e m b e d d e d o b j e c t sāikil ‘c y c l e ’
Hindi-Urdu (IA)
(57) rām ko [S2 PRO sāikili calā- n- ı̄i S2 ] ā- tı̄i thı̄i
Ram dat cycle drive- inf- f come- imperf.f pst.f
‘Ram used to know how to ride a bicycle.’
n o l o n g - d i s t a n c e a g r e e m e n t w i t h t h e e m b e d d e d o b j e c t sāikil
‘c y c l e ’
(58) rām ko [S2 PRO sāikil calā- n- āS2 ] ā- tā thā
Ram dat cycle drive- inf- m come- imperf.m pst.m
‘Ram used to know how to ride a bicycle.’
l o n g - d i s t a n c e a g r e e m e n t w i t h t h e e m b e d d e d o b j e c t sāikil ‘c y c l e ’
Punjabi (IA)
(59) prakāš nũ˜ [S2 PRO sāikili calā- n.- ı̄i S2 ] ā- ndı̄i sı̄
Prakash dat cycle drive- inf- f come- imperf pst
‘Prakash used to know how to ride a bicycle.’
116 Case and agreement
To recapitulate, the following facts need to be taken into account for a compre-
hensive analysis of long-distance agreement in Hindi-Urdu (IA).
(i) In subject complements LDA is obligatory, while in object complements
it is optional.
(ii) In subject complements the embedded clause does not contain PRO, while
in object complements the occurrence of PRO is obligatory.
(iii) In object complements the matrix subject needs to be either ergative
or dative case-marked; the embedded subject in subject complements is
nominative case-marked.
(iv) In subject complements long-distance agreement is blocked if the embed-
ded subject is genitive case-marked.
(v) The predicate in subject complements belongs to the ‘begin’-class of
verbs, and the matrix verb may contain a predicate adjective too.
Thus, long-distance agreement in Hindi-Urdu (IA) is quite a complex phe-
nomenon, and we have presented only a brief summary of the issues involved.
We shall now discuss the case of Kashmiri (IA).
4.5 SALs and long-distance agreement 119
However, agreement in the matrix clause has the option of not taking place,
though there is agreement of the infinitive with the embedded object. In that
case, the infinitive vuchini ‘to see’ would be in feminine plural exhibiting
agreement with the embedded object kōri ‘girls,’ while the matrix verb yotšmut
chu ‘wanted m,s, be.pres. m,s’ would exhibit agreement with the matrix subject
rām ‘Ram.’ Hindi-Urdu too exhibits such a phenomenon.
infinitival agreement – no long-distance agreement
(68) rām- an chu hamēš yotšmut [pannis necivis khətr
Ram- erg be.pres m,s always wanted m,s self.dat son.dat for
kōri vuchini]
girls see-inf f,pl
‘Ram always wanted to see girls for his son.’
(Subbarao and Munshi, ms)13
In Kashmiri, infinitival agreement takes place even when the matrix subject is
nominative case-marked. Bhatt (2005: 24) comments: “Infinitival agreement
also takes place when long-distance agreement is not even a possibility.” Recall
that infinitival agreement with the embedded object is not permitted in Hindi-
Urdu when the matrix subject is nominative case-marked (62). The matrix verbs
in (69) and (70) agree with the matrix subjects rām ‘Ram’ and zoon ‘Zoon,’
respectively.
(69) rām chu yatsān [kōri vuchini]
Ram.m be.pres.m.s wants-imperf girls see-inf.f.p
‘Ram wants to see the girls.’
(70) zoon che yatsān [kōri vuchini]
Zoon.f be.pres.f wants.imperf girls see-inf.f.p
‘Zoon wants to see the girls.’
(Bhatt 2005: 24)
120 Case and agreement
According to Bhatt, in (69) and (70) the matrix verb agrees with the matrix
subject, as it is nominative case-marked: “Consequently, there is no room for
the infinitival object to display long-distance agreement with the matrix verb.
However, this does not block the object agreeing with the infinitival verb”
(Bhatt 2005: 24).
Maithili (IA)
(71) rām kẽ tohāri gār.ı̄ calaunāi abait ch- ahi
Ram dat your 2 mid hon car to drive comes- agr mid hon
‘Ram knows how to drive your car.’
(Yogendra Yadava, p.c.)14
4.5.2 Tibeto-Burman
We consider the case of long-distance agreement in Mizo (TB) first. In Mizo
predicate adjectives exhibit agreement with the noun. Adjectives in Mizo
carry person and number agreement features just as verbs do, and they are
homophonous with the agreement markers that occur with the verb. Mizo has
no gender agreement. Sentences (72) and (73) are instances of finite subject
complementation, and hence the embedded sentence occurs in the subject posi-
tion of the matrix clause. The 3rd person singular agreement marker is -a,
and the plural marker is -an. In (72), the predicate adjective in S2 carries -an
‘3rd person plural,’ as it agrees with amerikā mite ‘American people,’ the sub-
ject of S2 . The matrix verb carries -a ‘3rd person singular’ as it is a case of
subject complementation, and the entire S2 is the subject of S1 .
Sentence (72) literally means: ‘(That) Americans are very fat is the case.’
4.5 SALs and long-distance agreement 121
4.5.3 Dravidian
Dravidian languages (e.g. Kannada, Tamil and Telugu), too, exhibit long-
distance agreement. We now discuss LDA in Telugu. In Telugu (DR), adjectives
carry pronominal suffixes in predicate position, and thus exhibit agreement
with the subject (Krishnamurti and Gwynn 1987). In (74), the adjective in the
embedded clause, manci ‘good,’ carries the 3rd person pronominal suffix vād.u
‘3rd person singular,’ and it agrees with vād.u ‘he,’ the subject of the embedded
clause. Sentences (74) and (75) are instances of subject complementation. The
verb in Telugu carries phi-features. In (74), the matrix verb is un-di ‘be-3rd
person, non-masculine, singular,’ and in (75), it is unnā-d.u ‘be-3rd person, mas-
culine, singular.’ In (74), the matrix verb undi ‘be + 3 nh’ exhibits agreement
with the entire embedded sentence, and hence -di the ‘3rd person, non-human,
singular’ marker occurs with the verb stem. In (75), the matrix verb exhibits
agreement with the embedded subject vād.u ‘he,’ and hence -d.u ‘3rd person,
masculine, singular’ occurs.
no long-distance agreement
Telugu (DR)
(74) [S2 vād.ui manci- vād.ui - lāS2 ]j un- dij
he good- 3 s,m pron suffix- comp be- s,nh
‘It appears that he is a nice fellow.’
(Subbarao 2001: 482)
long-distance agreement
(75) vād.ui manci- vād.u lā unnā- d.ui
he good- 3 s,m pron suffix comp be.pres- 3 s,m
‘He appears to be a nice fellow.’
(Subbarao 2001: 483)
no long-distance agreement
(76) [S1[S2[S3 nāku vād.ui manci- vād.u- lāS3] unna-
to me he good- 3 s,m pron suffix- comp be-
t.luS2] j kanipistundiS1] j
comp appears 3 s,nh
‘He seems to appear like a nice fellow.’
(Subbarao 2001: 483)
long-distance agreement
(77) nāku vād.ui manci- vād.u- lā unna- t.lu
to me he good- 3 s,m pron suffix- comp be- comp
kanipistunnā- d.ui,∗ j
appear- 3 s,m
‘He seems to appear like a nice fellow.’
(Subbarao 2001: 483)
We do not have data on LDA in Khasi (Mon-Khmer) and the Munda languages.
Another interesting aspect of agreement in SALs is Pronominal Strength
Hierarchy found in subject–verb/adjective agreement, which we shall discuss
in the following section.
In sentences with conjoined subjects in Mizo and Hmar, the 1st person subject
has precedence in occurrence as a pronominal clitic over 2nd and 3rd person
subjects, and 2nd person over 3rd person. Hence, when 3rd person subject and
1st person subject occur as conjuncts as in (79), it is the 1st person subject
clitic kan- in the plural that occurs, and not the 3rd person plural clitic an-. The
4.7 ECM cases (Small Clauses) 123
1st person subject as a conjunct does not occur overtly at all, and there occurs
in its place only a null pro. Thus, the 1st person singular null subject pro in
(79) is recoverable only from the verbal agreement clitic kan- in the plural. The
precise details of such recoverability need to be worked out.
(79) lalii le proj naŋk - cou in- t.šuŋ- pui- kanj+plural - ti- cek
Lali and (I) you- only VR- sit- with- 1 p- fut- 2 s
‘Lali and I will sit only with you.’16
(Subbarao 2001: 467)
In (80), the conjuncts are in 3rd and 2nd person and hence, it is the 2nd person
agreement marker in the plural that occurs with the verb. The 2nd person
subject as a conjunct does not occur at all, and in its place there is a null pro.
(81) 1p > 2p > 3p where > indicates ‘stronger than’ and p indicates person.
Sentences (79) and (80) also illustrate how recoverability of the null pronoun
(pro) takes place. It is only the agreement marker in plural on the verb that
indicates whether pro in (79) refers to a 1st person pronoun or pro in (80) refers
to a 2nd person pronoun.
In Telugu (DR) and Tamil (DR) too, the principle of Pronominal Strength
Hierarchy operates in verb and adjective agreement. Adjectives in predicate
position in Telugu carry pronominal suffixes except in 3rd person masculine
singular where the entire pronoun is suffixed to the adjective. The pronominal
suffixes too exhibit Pronominal Strength Hierarchy (see Subbarao 2001 for
details). In Oriya (IA) as well, the principle of Pronominal Strength Hierarchy
operates in conjoined structures (Neukom and Patnaik 2003).
Our data show that, in the languages in which the agreement clitic occurs,
the clitic of the embedded subject shows up as a clitic on the matrix verb, thus
demonstrating that it is a case of long-distance agreement.
4.7.1 Indo-Aryan
When the matrix subject is ergative case-marked, the matrix verb agrees with
the embedded subject if it is not followed by a postpositional case marker (82),
and the verb exhibits default agreement if it is followed by a case marker, as in
(83) in Hindi-Urdu.
Hindi-Urdu (IA)
(82) atul ne [[yah cı̄z] burı̄] samajh- ı̄ thı̄
Atul erg this thing. f bad. f consider- perf. f be.pst. f.s.
‘Atul had considered this thing bad.’
(83) atul ne [[is cı̄z] ko burā] samajh- ā thā
Atul erg this thing. f acc bad.def consider- perf. m be.pst. m.s
‘Atul had considered this thing bad.’
(Mahajan 1989)
If the matrix subject is nominative case-marked, the matrix verb agrees with
the matrix subject as (84).
(84) atul [[yah cı̄z] burı̄] samajh- tā hai
Atul this thing. f bad. f consider imperf pres
‘Atul considers this thing bad.’
(Bhatt 2005: 32)
4.7.2 Dravidian
For a discussion of ECM cases in Dravidian, see chapter 5.
4.7 ECM cases (Small Clauses) 125
4.7.3 Austro-Asiatic
Khasi (Mon-Khmer)
In (85) in Khasi, the derived object tεilin ‘Teilin’ is accusative case-marked.
The embedded subject tεilin ‘Teilin’ moves to the matrix object position from
a [+tensed] COMP-clause (see chapter 6 for a discussion of this phenomenon).
Hence, the 3rd person feminine singular agreement marker ka is retained on
the embedded verb wan ‘come’ in (85) (see appendix 2 of chapter 6 on the
Website for a detailed discussion).
Munda
In Ho (Munda), the matrix verb carries the direct object agreement marker,
as (86) illustrates. The subject of the embedded clause, añ ‘I,’ occurs in the
derived object position and hence the direct object agreement marker iñ occurs
as a suffix on the matrix verb manatiŋ ‘consider.’ The fact that the embedded
subject agreement clitic occurs on the matrix verb demonstrates that it is a case
of ECM.
Ho (Munda)
(86) ami [añj concor.-] emi - manatiŋ- t- ad.- iñj - a
you me intelligent- 2s- consider- epen- pst perf- 1s- fin
‘You consider me intelligent.’
(Koh and Subbarao, ms)
4.7.4 Tibeto-Burman
We shall now discuss cases involving ECM subjects in Mizo and Hmar. Recall
that Mizo and Hmar are split ergative in person. Subject as well as object agree-
ment markers occur as pronominal clitics in these languages. The pronominal
agreement marker ce of the subject of the embedded sentence appears as a clitic
on the matrix verb in ECM cases. Hence, ce occurs to the right of the matrix
verb.
Hmar (TB)
(87) lalii - n [øj insual- pui] in a- dit- cej
Lali- erg fight- with comp 3s- want- 2s
‘Lali wants you to fight.’
126 Case and agreement
Note that (87) is also an instance of postposition incorporation where pui ‘with’
occurs with the embedded verb insual ‘fight’ (see the discussion in section 3
of this chapter for details).
In Mizo (88), it is min which is the subject of the embedded sentence showing
up as the object clitic of the matrix S. It occurs to the left of the matrix verb.
Mizo (TB)
(88) zovi- n lekhathɔn ø- ziak turin min- duh
Zovi- erg letter write comp 1s- want
‘Zovi wants me to write a letter.’
(Subbarao and Lalitha Murthy, ms)
Note that the subject of the matrix sentence is in 3rd person. Recall that
1st person DO > (‘is stronger than’) 3rd person DO (according to the Principle
of the Pronominal Strength Hierarchy discussed in 4.6).17
To summarize, in all SALs, it is the subject of the embedded clause that
triggers agreement in the matrix verb when it is in the derived object position.
All the examples discussed in (82), (86), (87) and (88) are instances of LDA.
In the South Asian subcontinent, language contact has played a dominant
role, as a result of which phonological, morphological and syntactic conver-
gence has taken place. In the following section we discuss how features of
agreement are borrowed from one language family to the other.
a g r e e m e n t o f o o (l o c a t i v e )
Manda (DR)
(89) inelin māi - dengata pı̄yu rı̄j - dai - t- ij
yesterday us- on day rain- 1 p- pst- 3 s
‘It rained on us yesterday.’
In a Dravidian language, when the subject is dative case-marked, the verb agrees
with the theme, which is nominative case-marked (see section 4.2.1 for details).
Note that in (90) the verb does not agree with the noun phrase nūmer ‘fever’
in the nominative case. Instead, it agrees with a postpositionally case-marked
dative subject. Since none of the other Dravidian languages have agreement
of a non-subject with a verb, while Munda languages do, Ramakrishna Reddy
(1992a, 2004) argues that this phenomenon is due to the Mundari (Munda)
influence on the Dravidian Manda.
Malto (DR)
(91) proi bit- a- ka- ri oy- a- ri mand- a- ri
(they) cook- epen- cpm- 3 p take- cpm- 3 p plant- pst- 3 p
‘Having cooked, having taken them, they planted them.’
(Mahapatra 1979: 223)
Relative clauses
Type I Embedded subject carrying the ergative marker, and
the verb carrying the subject agreement marker in the
embedded relative clause
Mizo (TB)
(92) [zovi- n lekahbu a- lei- tur19 cu] lali- n a- chiar- aŋ
Zovi- erg book 3 s- buy- fut def Lali- erg 3 s- read- fut
‘Lali will read the book that Zovi will buy.’
4.9 Case and agreement in SALs 129
In (92), the embedded subject Zovi ‘Zovi’ carries the ergative case marker -n,
and the embedded verb carries the agreement marker -a. In contrast, in (93),
the embedded subject Zovi ‘Zovi’ does not carry the ergative case marker -n,
and the embedded verb lei ‘buy’ does not carry the subject agreement marker
-a. In Mizo and Hmar, the ergative marker and the agreement marker are
mutually dependent; either they both occur or neither does. Such dependency
in occurrence is also found in passives and clefts in Mizo (Subbarao and Lalitha
Murthy 1994; Prakash 2006).
4.9.2 Passives
Let us look at the passive in Mizo. The passive too exhibits a similar pattern
to that in the relative clause. The passive in Mizo has a bi-clausal structure
(see Subbarao and Lalitha Murthy 1994: 403). Example (94) is a simple active
sentence.
(94) zovi- ni thei- te ai - ei
Zovi- erg fruit- p 3 s- eat
‘Zovi ate (some) fruits.’
(Subbarao and Lalitha Murthy 1994: 403)
Examples similar to (95) and (96) are found in Hmar (TB) too.
Thus, the data from relative clauses in Mizo (TB) and passives in Mizo and
Hmar (TB) demonstrate that case and agreement are intrinsically linked.
4.9.3 Clefts
In clefts too a similar pattern is found in Mizo. Clefts are bi-clausal in Mizo.
Hmar (TB)
(101) boŋ- cu [zova- n a- that] a- ni- noʔ
cow- foc Zova- erg 3 s- kill [+fin] 3 s- be- not
‘It is not the cow that Zova killed.’
(Vanlal Englien, p.c.)
We now consider the case of the passive in Thadou. Thadou (TB) also exhibits
a pattern similar to Mizo with regard to relative clauses, passives and clefts
(Pauthang Haokip, p.c.). We present below two sentences illustrating the two
patterns of the passive in Thadou (TB). The passive in Thadou too has bi-
clausal structure. Example (103) is a simple active sentence. The quantifier
themxat ‘some’ follows the head noun.
Note that the singular and plural markers are homophonous in Thadou.
Thadou (TB)
(103) lalal- ini thei themxat ai - nei
Lalal- erg fruit some 3 s- eat
‘Lalal ate some fruits.’
Similar data can be adduced from other TB languages, such as Paite and Zou.
The relative clause in Mizo, the passive and the cleft in Mizo and Hmar,
and the passive in Thadou provide support for Chomsky’s (2001) view of
linking together the uninterpretable features of goal and probe. They are either
overtly present or absent altogether. That is, case and agreement are intrinsically
linked.20
4.10 Conclusion
This chapter provides an exhaustive account of the various agreement types
available in SALs, bringing into focus several new phenomena which have not
been discussed in any work thus far. SALs can be classified into four major
agreement types:
(1) languages in which a postposition blocks agreement;
(2) languages in which a postposition does not block agreement;
(3) languages which are polysynthetic and semi-polysynthetic;
(4) languages which exhibit no agreement at all.
In the polysynthetic Sora (Munda) language, direct object, locative PP and
instrumental PP are incorporated, a phenomenon unique in SALs. Pronominal
clitics for subject, direct object, indirect object and possessor are incorporated
in Tibeto-Burman and Munda languages. In Munda and Tibeto-Burman lan-
guages, the verb carries the agreement marker of the possessor of the object
alone, not of the possessor of the subject. Thus, though the possessive marker
can potentially refer to either the subject or the object, it is coindexed only with
the direct object and not with the subject – for example, in Santali. Quantifiers,
Negative Polarity Items, intensifiers, emphasizers, group markers and adverbs,
too, are incorporated in Tibeto-Burman and the Mon-Khmer Khasi.
Adposition incorporation of the comitative postposition in Hmar (TB) and of
the locative in Paite (TB) takes place. When an adposition is incorporated, the
host intransitive verb, as a result, becomes ‘transitivized.’ Such incorporation,
as well as the ‘transitivization’ of the host intransitive verb, is a phenomenon
unique to select languages of the Tibeto-Burman family.
4.10 Conclusion 133
5.1 Introduction
Non-nominative subject constructions in the languages of the South Asian
subcontinent include ergative, dative, genitive, locative, instrumental (by-
passive) and accusative subjects. Dative subjects are a feature of the South Asian
linguistic area (Emeneau 1956; Masica 1976). They are the most widespread
in Dravidian and in some Indo-Aryan languages. Some Tibeto-Burman and
Munda languages, too, share this feature. The ergative construction is found
only in some Indo-Aryan and some Tibeto-Burman languages, and is absent in
Dravidian and Austro-Asiatic languages.
In this chapter, we discuss the nature of case marking – lexical/inherent
vs. structural, the choice of case on the subject and object in non-nominative
subject (hereafter, NNS) constructions, general trends in SALs, and the vari-
ation by genetic affiliation or subregion. We provide a brief description of
NNSs in SALs first, keeping in view the notion of subject.1 We shall then
discuss some subject properties of NNSs. We argue that (i) the predicate
in a dative subject construction (DSC) is [−transitive] and unaccusative;
(ii) all NNSs except the ergative are inherently case-marked; (iii) such inher-
ent case marking can be done not by an intransitive verb alone, but by the
whole predicate compositionally consisting of a theme or an adjective along
with the [−transitive] verb; and (iv) information concerning agreement should
be available vP-internally (in the lower thematic S) for proper assignment of
inherent case to the NNS. We shall show that the accusative/dative case mark-
ing of the theme in dative/genitive subject constructions in Bangla, Tamil and
Malayalam does not count as counter-evidence to treating the predicate in NNS
constructions as [−transitive].
This chapter is organized as follows. Section 5.1 is an introduction.
Section 5.2 provides a description of the NNS constructions in SALs.
Section 5.3 deals with the domains of occurrence of the NNS. Section 5.4
discusses some subject and non-subject properties of NNS constructions in
SALs. Section 5.5 deals with the stative and non-stative as well as the intransi-
tive nature of the predicate in dative subject constructions (DSCs). Section 5.6
134
5.2 NNS constructions in SALs 135
discusses the inherent case marking of the dative subject in SALs. Section
5.7 discusses double/multiple dative subject constructions in Telugu and Kan-
nada, which are unique to SALs. Section 5.8 concludes the chapter. Formal
representation and case checking properties of NNS constructions in SALs are
discussed in the appendix of chapter 5 on the Website.
In all SALs, the non-nominative DP occurs in the initial position of a sen-
tence, [NP of S] (Chomsky 1965) or Spec of TP. SALs have seven types of
NNS constructions. The matrix or embedded subject is (i) ergative case-marked,
(ii) dative case-marked, (iii) genitive case-marked, (iv) locative case-marked,
(v) accusative case-marked or (vi) instrumental case-marked and is in the by-
passive or (vii) the embedded subject is accusative case-marked in non-finite
embedded clauses (the so-called Exceptional Case Marking2 [ECM] cases in
Government and Binding theory).
as a local antecedent
Hindi-Urdu (IA)
(1) baccõi ne apnı̄i/∗ j billı̄j dekh- ı̄j
children erg self’s cat f, s nom see- f, s perf
‘The childreni saw / looked at self’si/∗ j cat.’
(Davison 2004: 145)
136 Non-nominative subjects
Kashmiri (IA)
(2) mohani - an vuch panun pāni ə̃˜nas manz
Mohan- erg saw self’s self mirror in
‘Mohan saw himself in the mirror.’
(Wali and Koul 1997: 125)
as a long-distance antecedent
Hindi-Urdu (IA)
(3) aoki ne lalitāj se [PROi apne liei/j cāy banā- ne ko] kahā
Ashok erg Lalita with self for tea make- inf in order said
‘Ashok asked Lalita to make tea for self.’
(Subbarao 1971: 191; the glosses have been slightly modified)
Mizo (TB)
(5) lalii - n [PROi min- hmu] ai - duʔ
Lali- erg 1 s- meet 3 s- wants
‘Lali wants to meet me.’
(Subbarao and Lalitha Murthy, ms)
in these families, it is only the verb be that occurs and not have. Some of the
Tibeto-Burman languages such as Mizo, Hmar and Thadou have two distinct
verbs for be and have, and most of the Tibeto-Burman languages, except Bodo,
Kokborok and Manipuri, do not have any NNS construction (other than the
ergative) at all. Khasi (Mon-Khmer) also has two different verbs for be and
have, and it too does not have any NNS construction at all. We shall demonstrate
later that it is the verb coupled with a noun or an adjective in the verb phrase
that is instrumental in assigning the non-nominative case to the subject.
alienable possession
Hindi-Urdu (IA)
(6) murlı̄i ko dūkān mẽ kaı̄ acchı̄ kitābẽj milı̄j thı̄˜∗ i,j
Murali m,s dat shop in several good books f,p found f,p be+pst f,p
‘Murali found several good books in the shop.’
Telugu (DR)
(7) mādhurii ki kot.- lō cālā manci pustakāluj dorikēyi∗ i,j
Madhuri dat shop- in very good books p,nh found p,nh
‘Madhuri found very good books in the store.’
Manipuri (TB)
(8) manoŋ- də layrik- əmə ləy
he- dat book- one be
‘To him there is a book.’
(Chelliah 1990: 201)
inalienable possession
Marathi (IA)
(9) malai don hātj āhet∗ i,j
I-dat two hands p are p
‘I have two hands.’
Hindi-Urdu (IA)
(11) raghui kı̄ tı̄n bahnẽj thı̄˜∗ i,j
Raghu m gen three sisters be+pst f,p
‘Raghu had three sisters.’
Kokborok (TB)
(12) kɔrmati ni bɔkhɔrɔk sā- ɔ
Kormati gen head pain (verb)- pres
‘Kormati has a headache.’
Manipuri (TB)
(13) ram- gi hui- əmə ləy
Ram- gen dog- one be
‘Ram has a dog.’
‘There exists a dog of Ram.’
(Chelliah 1990: 200)
In Assamese (IA), the subject is genitive case-marked, and the theme is nomina-
tive case-marked, as in (14). The predicate in (14) is a psychological predicate
which depicts the emotional state that the logical subject is in.
Assamese (IA)
(14) tar khɔng uth.i- sil
he/she.gen anger (nom) rise.cpm- pst
‘He/she is angry.’
5.2 NNS constructions in SALs 139
Bangla (IA)
(15) rina- ke aj pãc t.a dokane jete holo
Rina- dat today five cl shops.loc go.inf was
‘Rina had to go to five shops today.’
(Dasgupta 2004: 130)
Telugu (DR)
(16) prastutam pratāpi daggara d.abbuluj lē- vu∗ i,j
at.present Pratap near money.3 p,nm be.not- 3 p,nm
‘Pratap does not have any money at present.’
The dative too can occur in sentences where the possessed entity is [+concrete],
and the dative indicates a permanent state of affairs. Thus, (17) is interpreted as
Pratap does not have money at all, and he is poor. Hence, the occurrence of an
adverb such as prastutam ‘at present’ will render the sentence ungrammatical.
Bodo (TB) has a locative postposition with the experiencer, when a physical
ailment is expressed.
Bodo (TB)
(18) khampha- nao lɯmja- nai dɔŋ- ɔ
Khampha- gen.loc (near) sick (verb)- inf be- pres
‘Khampha has fever.’
The non-nominative marker that occurs with the possessor (logical subject) in
IA languages such as Hindi-Urdu, Punjabi and Marathi depends on the nature
of the object possessed. In Hindi-Urdu, the genitive postposition occurs with
140 Non-nominative subjects
inalienable possession of body parts, kinship and friends, while the locative
ke pās ‘near’ occurs with concrete possession. The locative mẽ ‘in’ occurs
with inherent physical qualities in Hindi-Urdu, while in Bangla the genitive
occurs in such cases (for details, see Verma and Mohanan 1990; Masica 1991;
Bhaskararao and Subbarao 2004). According to Dasgupta (2004: 132), genitive/
locative case marker -r occurs with the experiencer in Bangla (IA).
Bangla (IA)
(19) tomar jɔthet.o aho ache
you.indirect enough courage is
‘Do you have enough courage?’
(Dasgupta 2004: 132)
Bodo (TB)
(20) bibar- ao mujang mud m- nai dɔng
flower- in good fragrance (verb)- inf be
‘The flower has good fragrance.’
Literally: ‘To fragrance [as verb] well is [there] in the flower.’
Sinhala (IA)
(21) māwə yantaŋ bēruna
I.acc barely escape.pst
‘I barely escaped.’
(Gair 2003: 791)
(Subbarao and Lalitha Murthy, ms) and Hmar. In Khasi and the Munda lan-
guages, there is only the syntactic passive, and the passive marker is the same
as the verbal reflexive and anti-causative (see chapter 3).
In Hindi-Urdu (IA), the patient/theme may be nominative case-marked, and
the verb agrees with it as in (22a), or accusative case-marked as in (22b), and
the verb exhibits default agreement as the agent as well as patient/theme are
both followed by a postposition.
Hindi-Urdu (IA)
(22a) polı̄si ke.dvārā corj pakr.ej nahı̄˜ gaye∗ i,j
police by thieves m,p catch.perf m,p not pass.pst m,p
‘The thievesi werej not caught by the policej .’
Assamese (IA)
(23) ravan.- ak ram- ɔr dwara morua hol
Ravan- acc Ram- by was killed
‘Ravan was killed by Ram.’
(Kakoli Das, p.c.)
In Kashmiri and Sinhala (IA) too, the patient/theme may be dative case-marked.
In (24), in Kashmiri, the patient koryi ‘girl.dat’ “moves to the subject position
in passive” (Bhatt 1999: 215).
Kashmiri (IA)
(24) az av koryi vuch- n (temsin- dasyi)
today came girl.dat see- pass him- by
‘Today the girl was seen (by him).’
(Bhatt 1999: 215)
Colloquial Sinhala (IA) according to Gair (2003: 792) lacks a true passive. The
subject is instrumental case-marked, and the verb carries an involitive verbal
suffix that imparts the passive meaning.
Sinhala (IA)
(25) amma- geŋ sinhəla kǣmə hoňdət.ə hædenəwa
mother- instr Sinhala food well make.involitive.pres
‘Mother (always) makes Sinhala food well.’
(Gair 2003: 792)
142 Non-nominative subjects
Kharia (Munda)
(26) pothi- ki- te ter d.om- ki- Ø
book- p- obl give pass- middle pst- default
‘The books were given.’
(Peterson 2006)
Khasi (Mon-Khmer)
(27) ya- ki- sʔiar ø- la- bām da- u- ksɔw
acc- f- fowl default- pst- eat by- m- dog
‘The hen was eaten by the dog.’
(Temsen and Subbarao, ms)
Marathi (IA)
(28) mādzyā kad.ūn /*dwārā kām kela gela nāhı̄
I-obl by work do-pst-3s go-pst-3s (pass) neg
‘This work was not done by me.’ (‘I was unable to do the work.’)
(Pandharipande 1997: 302)
Hindi-Urdu (IA)
(29) rām se per. kāt..e nahı̄˜ ga.ye
Ram by trees m,p cut.perf m,p not pass.perf m,p
Literally: ‘The trees were not cut by Ram.’
‘Ram couldn’t bring himself to cut the trees.’
(Rajesh Bhatt, p.c.)
5.2 NNS constructions in SALs 143
However, if the agent is not overtly present, the passive sentence does not
impart capabilitative meaning, as in (30).
When the agent is not overtly present, the capabilitative meaning is not
imparted.
Telugu (DR)
(33) āmei valla ilāt.i panuluj ceyya- bad.- a vuj
she by such things do- pass- neg- nh,p
‘Such things cannot be done by her.’
The lexical passive in Telugu (DR), like in Hindi-Urdu (IA), does not impart
capabilitative meaning when the agent is not overtly present.
144 Non-nominative subjects
Telugu (DR)
(35) ilāt.i panuluj ı̄ ūl.l.- ō jaraga-vuj / kā-vuj
such things this town- in happen-not happen-not
‘Such things are not done in this town.’
In Telugu (DR) too, just as in Malayalam, the passive “is not usually used in
conversation. It has tended to be confined to official language and the language
of journalism” (Asher and Kumari 1997: 268).
Just as in Telugu, in Malayalam too, “passives (including nonfinite forms)
appear without an agent” (Asher and Kumari 1997: 269).
Malayalam (DR)
(37) ivid.e pat.t.usāri-kal. vilkkappet.unnu
here silk saree-p sell-pass-pres
‘Silk sarees are sold here.’
(Asher and Kumari 1997: 269)
Bangla (IA)
(39) orita gopal ke bhalo mone kɔre
Sarita Gopal acc good thinks
‘Sarita considers Gopal to be good.’
5.2 NNS constructions in SALs 145
Kashmiri (IA)
(40) me vuch- na tem- is alakh pyāvān
I.erg saw- neg he- dat beat fall-imperf prt
‘I did not see him being beaten.’
(Bhatt 1999: 198)
In Nepali (IA), the embedded subject is dative case-marked (Poudel 2007: 32)
too, and in Mizo (TB), the accusative case marker with the derived object occurs
in the Exceptional Case Marking (ECM) construction.
Mizo (TB)
(41) zovi- n [amah- cu lāmthiam] a- inti
Zovi- erg she.self- acc dancer 3 s- considers
‘Zovi considers self a dancer.’
(Lalitha Murthy and Subbarao 2000: 803)
Kashmiri (IA)
(42) tem vuch swai [S2 ti dohay bat khyav- ānS2 ]
he.erg saw.f,s she.nom daily rice.acc eat imperf
‘He saw her eating rice every day.’
(Bhatt 1999: 219)
However, if the object clitic occurs on the matrix verb, the sentence is
grammatical (43b) (Sadaf Munshi, p.c.).
146 Non-nominative subjects
Dumi (TB)
(44) aŋ- mō- liŋ- ə- m m- a a- dūkhut- ə ʔe
I- (nom)- sit- walk about- 1 s- nozr he- erg MS- see- 1 s rep
‘He said he saw me hanging around that place.’
(van Driem 1993: 192 – rep: reported speech particle, i.e., hearsay
evidential; MS: marked scenario)8
Telugu (DR)
(45) mādhuri arun.- (ni) pāripō.tund.agā cūsindi
Madhuri Arun- acc run away.imperf saw
‘Madhuri saw Arun running away.’
Thus, Kashmiri (IA), Dumi and Tiwa (TB) and Telugu (DR) are the four
languages that we know of where, in the ECM construction, the embedded
subject is nominative case-marked. Further, the case marking in an ECM
construction in Kashmiri is sensitive to the tense of the matrix clause and
the agreement clitic of the embedded subject on the matrix verb.
Hindi-Urdu (IA)
(46) pallavii ko bahut khuı̄j huı̄∗ i,j
Pallavi dat very happiness happened
‘Pallavi felt very happy.’
The predicate in Telugu (DR) as in (47), contains verb und. ‘be’ when the
psychological predicate is an adjective, and the verb exhibits default agreement.
Telugu (DR)
(47) ı̄ pāpalai - ki santōs.amj- gā undi∗ i,∗ j
these babies- dat happiness- adjr is (def agr)
‘These babies are happy.’
Bodo (TB)
(48) bi- ha jɯbɯr raga jɯng- nai dɔng- ɔ
he- gen very anger set fire- inf be- pres
Literally: ‘to him a lot of anger is to set fire / to ignite.’
‘He is very angry.’
Hindi-Urdu (IA)
(49) pratimāi ko [zukām aur khā˜sı̄]j ho gayi thı̄∗ i,j
Pratima f,s dat [cold and cough]f happen went pst f
‘Pratima caught a cold and cough.’
148 Non-nominative subjects
Telugu (DR)
(50) pratimai ki [daggu- u jalubu- u rend.u- u9 ]j
Pratima f,s dat [cough- conj mkr cold- conj mkr both- emph]
unnāyi p∗ i,j
are
‘Pratima caught both a cough and a cold.’
Kokborok (TB) and Bangla (IA) have a genitive subject in such sentences.
Kokborok (TB)
(51) a- ni kɔngrai tɔng- ɔ
I- gen cold be- pres
‘I have a cold.’
Bangla (IA)
(52) amar t.hand.a lege- che
I.gen cold feel- pres
‘I have a cold.’
Hindi-Urdu (IA)
(53) us ādmı̄i ke sāre bālj jhar. gayej
that man gen all hair fall went
‘That man lost all his hair.’
Telugu (DR)
(54) ā manis.ii ki anta juttuj - u10 ūd.- i pō- yindij
that man dat so much hair- quantifier mkr fall- cpm go- pst
‘That man lost all his hair.’
Hindi-Urdu (IA)
(55) rāt ko mahimā ko t.hı̄k dikhāı̄ nahı̄˜ detā
night at Mahima dat well visible not give
‘Mahima cannot see well at night.’
Verbs such as sun-nā ‘to hear’ and bhūlnā ‘forget’ take a nominative case-
marked subject. In present-day Hindi-Urdu, a few speakers have started using
such verbs with a dative case-marked subject, as in (56).
Peter Hook (p.c.) informs me that in Shina (IA) too, a verb such as to forget,
which takes a nominative subject, permits a dative subject to occur, just as in
Hindi-Urdu (IA).
The verb kanipincu ‘to be visible’ takes a dative subject in Telugu.
Telugu (DR)
(57) pramodi ki rātrul.l.u sariggā kanipinc- a- du∗ i
Pramod dat nights properly visible- not- 3 s,nm (def agr)
‘Pramod cannot see well at night.’
Hindi-Urdu (IA)
(58) rādhāi kı̄ tı̄n bahnẽj thı̄˜∗ i,j
Radha f,s gen three sisters f,p were f,p
‘Radha had three sisters.’
Kashmiri (IA)
(59) mohani - as chuj dukānj / bōyj
Mohan- dat has [is] shop brother
‘Mohan has a shop/brother.’
(Wali and Koul 1997: 139)
Kannada (DR)
(60) nana-i ge mūvaru hen.n.u makkad.uj iddāre∗ i,j
I- dat three female children be.nonpst.3 p.hum
‘I have three daughters.’
(Sridhar 1990: 133)
In Kokborok (TB) too, the genitive occurs with the logical subject, just as in
Hindi-Urdu, and the occurrence of the verb to be depends on the presence
of the classifier with the possessed. If the classifier is not present with the
possessed entity, the verb tɔng ‘to be’ must be overtly present as in (61), and if
the [+human] classifier is present, then the occurrence of the verb tɔng ‘to be’
is optional as in (62).
Kokborok (TB)
(61) ni- ni thakhuk b.sk tɔng
you- gen brothers how many be
‘How many brothers do you have?’
Hindi-Urdu (IA)
(63) un logō˜i ke pās bahut paisāj thā∗ i,j
those people near a lot of money m,s was m,s
‘Those people had a lot of money.’
(64) un logō˜i ke pās utnā t.āimj bilkul nahı̄˜ thā∗ i,j
those people near that much time m,s at all (NPI) not was m,s
‘Those people did not have that much time at all.’
5.3 Domains of occurrence of NNSs in SALs 151
In Telugu (DR), the locative marker daggara ‘near’ occurs with the subject
to indicate temporary possession. Recall that ki, the dative case marker, also
occurs in sentences denoting possession, as in (17) above. However, there is
a difference: “while the occurrence of the dative [case marker] denotes ‘per-
manent or long-term possession,’ the occurrence of the locative [case marker]
denotes ‘temporary possession’” (Subbarao and Bhaskararao 2004: 172).
Telugu (DR)
(65) mā.kui mugguru panivāl.l.uj unnāruj kānı̄ prastutam māi
we (excl).dat three servants are but right now our (excl)
daggara iddare- e unnāru∗ i,j
near two- emph are
‘We’ve three servants but right now we’ve only two.’
Manipuri (TB) has the locative postposition occurring with the subject in the
non-nominative subject construction. The subject is in the non-nominative
case and is marked by the postposition manak-ta ‘near at.’ The occurrence
of non-nominative subjects in Tibeto-Burman languages is rather infrequent.
Most of the languages do not have the non-nominative subject construction
at all, in contrast to Indo-Aryan and Dravidian languages, where the use of
non-nominative subject construction is a dominant one.
Manipuri (TB) also makes a distinction in terms of which postposition occurs
for temporary and permanent possession, just as in Telugu (DR).
Need or necessity
The dative subject also occurs in sentences denoting need or necessity in Hindi-
Urdu, Punjabi (both IA), Bodo, Garo (TB), Telugu and Kannada (DR).
Hindi-Urdu (IA)
(66) ramyā ko ek kitāb cāhiye
Ramya dat one book needed
‘Ramya wants a book.’
Bodo (TB)
(67) khampa n gəikher naŋɯu
Khampa dat milk want
‘Khampa wants some milk.’
Garo (TB)
(68) aŋ- na i- ko naŋ- no- a
I- dat this- acc need- fut- ?
‘I will need this.’
(Burling 2004: 122)
152 Non-nominative subjects
Kannada (DR)
(69) nana- ge idu is.t.a illa
I- dat this liking neg
‘I don’t like this.’
(Sridhar 1979: 101)
Malayalam (DR)
(70) nı̄ pōyē tı̄rū
you go-emph must
‘You really must go.’
(Asher and Kumari
1997: 307)
To denote a recipient
The dative subject also denotes a recipient.
Hindi-Urdu (IA)
(71) mujhei yah㘠ek cit.t.ı̄j milı̄∗ i,j
I.dat here one letter f,s found f,s
‘I got a letter.’
(Montaut 2004: 193)
Telugu (DR)
(72) ā abbāyilai - ki rōd.d.u mı̄da oka pustakamj dorikin-di∗ i/j
those boys- dat road on a book nm,s got-nm,s
‘Those boys found a book on the road.’
Hindi-Urdu (IA)
(73) is panjābi lar.kı̄i ko bharat nāt.yamj ātā∗ i,j thā∗ i,j
this Punjabi girl dat classical Indian dance come was
‘This Punjabi girl used to know the classical Indian dance.’
5.3 Domains of occurrence of NNSs in SALs 153
Telugu (DR)
(74) vāl.l.a ammāyi- ki sangı̄tam vaccu
their daughter- dat music comes
‘Their daughter has the knowledge of music.’
genitive–dative
Hindi-Urdu (IA)
(75) [prasād ke man]i ko gahrı̄ cot.j pahuncı̄∗ i/j
Prasad, m gen mind dat deep wound, f reached, f
Literally: ‘Prasad’s mind got hurt badly.’
‘Prasad’s feelings were hurt (deeply).’
(Om Arora, p.c.)
genitive–locative
(76) [prasād ke sar]i mẽ gahrı̄ cot.j āı̄∗ i/j
Prasad, m gen head in deep wound, f came f
‘Prasad got hurt badly in the head.’
(Om Arora, p.c.)
Kannada (DR)
(77) avani - a kan.n.j - ige gāya(v)k āgi- de∗ i∗ j,k
he- gen eye- dat injury happen- agr
‘He got hurt in his eye.’
In Kannada and Telugu (DR), the possessor, as well as the possessed, are both
dative case-marked.
154 Non-nominative subjects
Kannada (DR)
(78) avani - ige kan.n.j - ige gāya(v)k āgi- de∗ i∗ j,k
he- dat eye- dat injury happen- be
‘He got hurt in his eye.’
See section 5.7 for further details. The Double Dative Subject construction in
Dravidian is similar to the double subject construction discussed in Japanese
by Shibatani and Pardeshi (2001).
Hindi-Urdu (IA)
(79) kamzor insāni ko apne- āp par gussāj ā- tā∗ i,j hai∗ i,j
weak human being dat self+gen- self on anger come- imperf pres
‘A weak human being gets angry at himself.’
Telugu (DR)
(80) kāvyai - ki tana mı̄da *(tana-kii ) jālij put.t.in- di∗ i,j / *put.t.u-
Kavya- dat self on self-dat pity was born- 3 s,nh be born-
kon- di∗ i,j
VR- 3 s,nh
‘Kavya pitied herself.’
In Hindi-Urdu (IA), Saxena (1985) first pointed out that, in the DSC, the
possessive anaphor alternates with a simple possessive pronoun (see Gurtu
1992, too).
Hindi-Urdu (IA)
(81) mohani ko apnei / uskei/j mā˜- bāp kı̄ yād
Mohan.ms dat self’s his (3s gen) mother- father gen memory f,s
ā- yı̄
come- pst.f,s
‘Mohani remembered self’si/∗ j /hisi/j parents.’
(Saxena 1985)
Thus, dative subjects bind a pronoun and anaphor, or just an anaphor, while
nominative subjects bind an anaphor.
Mohanan and Mohanan (1994: 175) provide an example to show that the
possessive pronoun in a DSC refers to the nominative NP, and not to the dative
subject.12
Further, examples similar to (82) from Hindi-Urdu include (83), where the
possessive pronoun refers to a discourse antecedent, and not to the dative
subject.13
Hindi-Urdu (IA)
(83) us lar.kı̄i ko us ke∗ i,j bare mẽ socnā cāhiye thā
that girl dat her about to think needed was
‘That girl should have thought about her/*herself.’
(84) us lar.kei,∗ j ko apne āp pari,∗ j / us par∗ i,j bahut bharosā hai
that boy dat self-gen self on him on very trust is
‘That boyi,∗ j has confidence in himselfi,∗ j /him∗ i,j .’
156 Non-nominative subjects
Bangla (IA) has the possessive reflexive nije-r ‘self’s’ which requires a
c-commanding antecedent. In Bangla, the possessive pronoun tar ‘his/her’
is coindexed with the genitive subject just like the possessive reflexive nije-r
‘self’s,’ in specific contexts.
Bangla (IA)
(85) bou.t.ii - r nijeri / tari bap- er bar.i- r kɔtha
bride.cl- gen self’s her father- gen house- gen thought
mone por.lo
remembrance fell
‘The bride remembered self’s / her parents’ home.’
(86) oi chelei - t.a- r nijeri opor / or∗ i,j opor khub bhɔra ache
that boy- cl gen self on/ him on very trust is
‘That boy has a lot of confidence in himselfi,∗ j /him∗ i,j .’
Telugu (DR)
(87) kāminii ki tanai,∗ j / āme∗ i,j samasyalu ardham avv-
Kamini dat self’s/ her problems understanding happen-
a- vu
neg- nm,p
‘Kamini does not understand her [own] / her [someone else’s] problems.’
is a subject can bind only a reflexive or reciprocal and not a pronoun. Since the
dative DP “locally binds a pronoun” in (81) above, “it is not a subject.” Thus,
the dative DP in (81) is “either a subject binding a reflexive or not a subject,
binding a pronoun.” In the Minimalist Program, the subject moves to the Spec
position of a Tense Phrase (TP) due to the requirement of the EPP (Extended
Projection Principle).14 The EPP in simple terms means: every sentence must
have a subject. Hence, Davison (2004: 155) argues that a dative DP may or
may not move to Spec/TP depending upon whether it is an antecedent to an
anaphor or a pronoun.
Hindi-Urdu (IA)
(88) harek baccei ko [PROi tair-nā] acchā lagtā hai
every child dat swim-to good feel pres
‘Every child likes to swim.’
(89) harek baccei mẽ [PROi yah bāt sun kar] jān mẽ jān ā gayı̄
every child in this news hear cpm life in life come went
‘On hearing the news every child became cheerful.’
Bangla (IA)
(90) ɔb baccai - r- i [PROi sãtar kat.- te] bhalo lage
all child- gen- emph to swim cut- imperf pple good strikes
‘All children like to swim.’
Hindi-Urdu (IA)
(91) [PROi is bāt kā patā cal- te hı̄] lar.kiyõi ko
dat this news gen knowledge go- right after girls dat
ghabr.āhat. huı̄
panickiness happened
‘As soon as PROi (the girls) came to know of this, the girlsi got panicky.’
158 Non-nominative subjects
Telugu (DR)
(92) [PROi ı̄ sangati teliya- gānē] ammāyili-i ki kangāru
dat this news know- right after girls- dat panickiness
put.t.indi
was born
‘As soon as PROi (the girls) came to know of this, the girlsi got panicky.’
Telugu (DR)
(93) [[ammāyilai - ki ı̄ sangati teliy- a gānē] ∀i kangāru pad.d.āru]
girls- dat this news know- as soon as nom panicky fell
‘As soon as PROi (the girls) came to know of this, the girlsi got panicky.’
In contrast, in IA languages that we know of, such sentences are not permitted.
Hindi-Urdu (IA)
(94) *[lar.kiyõi ko is bāt kā patā cal- te hı̄] ∀i
girls dat this news gen knowledge go- as soon as nom
ghabr.ā gayı̄˜
panicky went
‘As soon as PROi (the girls) came to know of this, the girlsi got panicky.’
Bangla (IA)
(95) *[baccai - r jɔr aa matro- i] ∀i u- e por.lo
child- gen fever come as soon as- emph lay- cpm fell
‘As soon as the child had fever, he lay down.’
Sentences (94) and (95) are grammatical only if the matrix subject is present
and the embedded subject is not present. That is, Hindi-Urdu and Bangla
permit only Forward Control, and not Backward Control (see chapter 7 for
more discussion).
In conjunctive participial clauses too, the dative subject can be a controller
of PRO.
In Telugu (DR) and Kharia (Munda), the dative subject controls PRO.
5.4 Subject and non-subject properties 159
Telugu (DR)
(96) [sı̄ta kui [PROi ā māt.a vin- i] kōpam vaccindi]
Sita dat that matter hear- cpm anger came 3s,nm
‘Having heard that matter, Sita got angry.’
(Lalitha Murthy 1994)
Kharia (Munda)
(97) [etwa- tei [PROi u-ki ya? haleit yo- kon] lebui la?- ki]
Etwa- obl these of condition see- cpm love epm- pst
‘Etwa, seeing their condition, felt compassion.’
(Peterson 2006; PRO in (97) was added for ease of exposition)
Tibeto-Burman languages such as Bodo, Rabha and Kokborok that have been
in intense contact with Indo-Aryan languages have a non-nominative subject
construction. The only language that we know of that has no apparent con-
tact with an Indo-Aryan language but has a non-nominative construction is
Manipuri. Evidence from Khasi (Mon-Khmer) cannot be adduced, as Khasi
does not have constructions involving an NNS.
We shall now consider passive sentences. The subject of a passive sentence
is the controller of PRO in Hindi-Urdu (IA) in (98).
Hindi-Urdu (IA)
(98) baccõi se [PROi yah bāt sun kar] cup rah- ā nahı̄˜ gayā
children by nom this news hear cpm quiet be- perf not go+pst
‘The children could not keep quiet on hearing the news.’
Davison (2004: 146) points out that Hindi-Urdu “does not allow a new gram-
matical subject to be created by a process like passive,” as the ungrammaticality
of (99) indicates.
Just as a nominative subject can occur in the position of PRO, so can an NNS
in a conjunctive participle. The embedded predicate kōpam vacc ‘anger come’
in Telugu and the other Dravidian languages, cāhi-era ‘needed’ in Nepali and
tsakh khas ‘anger climb’ in Kashmiri require a dative subject. PRO is glossed
here with the case marker that would normally occur with a lexical subject.
160 Non-nominative subjects
Telugu (DR)
(100) mamatai [PROi kōpam vacc- i] vel.l.i pōyindi
Mamata dat anger come- cpm left
‘Having gotten angry, Mamata left.’
Nepali (IA)
(101) [PROi pāni cāhi- era] ui nadi tira ga- yo
dat water need- cpm he river toward go- pst
‘Needing water, he went to the river.’
(Ichihashi-Nakayama 1994 as quoted in Bickel 2004: 81)
Kashmiri (IA)
(102) [PROi tsakh khas- ith] tul lar.kani or
dat anger climb- cpm lifted boy.erg noise
‘After the boyi got angry, hei raised hell.’
(Bhatt 1999: 196)
In Kashmiri (IA) and Telugu (DR), the matrix subject (controller) can be a
dative subject too.
Kashmiri (IA)
(103) [PROi tsakh khas- ith] āv salim- as vadun
dat anger climb- cpm came Salim- dat crying
Literally: ‘Having become angry, crying came to Salim.’
‘Having become angry, Salim began to cry.’
(Aadil Kak, p.c.)
Telugu (DR)
(104) [PROi kōpam vacc- i] ānti ki ēd.upu vaccindi
dat anger come- cpm Shanti dat crying came
Literally: ‘Having become angry, crying came to Shanti.’
‘Having become angry, Shanti began to cry.’
The fact that PRO occurs in a case-marked position in (103) and (104) shows
that PRO is case-marked (see chapter 7 for details). In contrast, in Hindi-Urdu
(Davison 2004), Oriya (Beermann and Hellan 2002: 45) and Punjabi, PRO
cannot occur in the subject position of an embedded conjunctive participial
clause which has a predicate that takes a dative subject.
Hindi-Urdu (IA)
(105) *rādha [S2 PRO gussā ā karS2 ] bāhar cal- ı̄ ga- ı̄
Radha dat anger come cpm outside walk- pst go- pst
Intended meaning: ‘Having felt angry, Radha went outside.’
5.4 Subject and non-subject properties 161
Punjabi (IA)
(106) *baccā [S2 PRO pu’kkh lag keS2 ] ro rı́ā ai
child dat hunger feel cpm cry prog pres
Intended meaning: ‘Having felt hungry, the child was crying.’
(Sandeep Gupta, p.c.)
Oriya (IA)
(107) *mote jara ho- i ousadha khā- il- i
I.dat fever happen- cpm medicine eat- pst- 1,s
‘Having had fever I took medicine.’
(Beermann and Hellan 2002: 45)
In Bangla (IA), PRO cannot occur in a position where the genitive subject
occurs. Bayer (2004: 56) observes: “In perfective participial clauses in which
a PRO subject is required, the nominative is replaced by PRO [as in (108)], but
the genitive cannot be [as in (109)].”
Bangla (IA)
(108) [PRO hẽe- hẽe] ram amake bolchilo je ---
NOM laughing laughing Ram me told that ---
‘Constantly laughing, Ram told me that . . . ’
(109) *[PRO hãi peye] ram amake bolchilo je ---
GEN laugh having.gotten Ram me told that ---
Intended meaning: ‘Constantly laughing, Ram told me that . . . ’
(Bayer 2004: 56)
However, as Probal Dasgupta (p.c.) points out: “PRO can occur in subject
position of an (embedded) conjunctive participial clause provided that both
that clause and the main clause have a genitive/dative subject.”
Bangla (IA)
(110) [S2 PRO hɔt.at rege giy- eS2 ] amar khub matha dhore gælo
gen suddenly angry become- cpm I.gen very head hold.cpm went
‘Having become suddenly angry, I got a headache.’
Telugu (DR)
(111) mallika [S2 PRO krindat.i nela ı̄ t.aimu lō jvaram rā-
Mallika dat last month this time in fever come-
vad.amS2 ] gurtu cēsu kon- di
inf remember do VR- 3snm
‘Mallika remembered getting fever at this time last month.’
(Subbarao and Bhaskararao 2004: 176)
Hindi-Urdu (IA)
(112) *mallikā ne [S2 PRO pichle mahı̄ne is vaqt bukhār ā- nāS2 ]
Mallika erg dat last month this time fever come- Inf
yād kiyā
memory did
Intended meaning: ‘Mallika remembered getting fever at this time last month.’
The vector verb bait.hnā ‘to sit’ suggests something was done which had bad
consequences, perhaps unintended, not done on purpose.
t h e v e c t o r v e r b jānā ‘ t o g o ’ w i t h a non-nominative s ubject
(114b) mujhe acānak cot. lag gay- ı̄
I.dat suddenly injury strike go- perf
‘I got hurt suddenly.’
Davison (2004: 147) comments: “For the speakers who accept sentences such
as [(114a)], these auxiliaries cut across subject case possibilities.” A similar
comment holds for (114b) too.
In Telugu (DR) as well, vector verbs such as kurconu ‘to sit,’ cāvu ‘to die,’
pō ‘to go’ are nominative subject-oriented, but they may freely occur with a
non-nominative (dative and locative) subject too.
t h e v e c t o r v e r b caccu ‘ t o d i e ’ w i t h a nominative s ubject
Telugu (DR)
(115) proddunna- proddunna appulavāl.l.u int.i mundu vacc- i
morning- morning lenders (nom) house in front of come- cpm
caccēru/ kūrconnāru
died (vector verb) sat (vector verb)
‘The moneylenders are sitting in front of the house early in the morning.’
(The speaker is expressing his anger by using the vector verbs caccu
‘to die’ / kūrconu ‘to sit.’)
164 Non-nominative subjects
t h e v e c t o r v e r b caccu ‘ t o d i e ’ / kūrconu ‘ t o s i t ’ w i t h a
non-nominative s u b j e c t w i t h a d a t i v e s u b j e c t
Telugu (DR)
(116) panimanis.i ki proddunn(a)- e jvaram vacc- i
servant dat in the morning- emph fever come- cpm
caccindi/ kūrcondi
died (vector verb) sat (vector verb)
‘The servant got fever right in the morning.’ (The speaker is unhappy
about it.)
w i t h a l o c a t i v e su b j e c t
(117) vād.i- daggara inta d.abbu und.- i cacc- i
he.obl- near so much money be- cpm die (vector verb)- cpm
evari- ki ēmi lābham
who- dat what profit
‘Who gets benefited in spite of the fact that he has so much money?’
t h e v e r b ja ‘ t o g o ’ w i t h a no m i n a t i v e s u b j e c t
Bangla (IA)
(118) projit bar.i- te col- e gælo
Projit house- to go- cpm went
‘Projit went home.’
t h e v e r b ja ‘ t o g o ’ w i t h a g e n i t i v e s u b j e c t
(119) projit- er ɔukh kor- e gælo
Projit gen ill do- cpm went
‘Projit fell ill.’
(Probal Dasgupta, p.c.)
The evidence presented above clearly demonstrates that there is a set of vector
verbs that are nominative as well as non-nominative subject-oriented.
We discuss below another kind of evidence for the subject property of the
dative subjects from Marathi, in which the complementizer can be reduced.
The post-sentential COMP in Marathi (IA) has two alternative forms: asə
and -sə. The latter, a result of the elision of the initial vowel, occurs only when
the subjects of the matrix and embedded sentence are identical. In (120a), the
subject of the matrix clause is a dative subject and the notional subject of the
embedded complement is a nominative subject. Phonological attrition of a- in
the complementizer asə takes place in (120a), which demonstrates that both
the dative subject and nominative subject have identical properties of a subject.
s u b j e c t o f t h e m a t r i x a n d e m b e d d e d c l a u s e i d e n t i c a l : sə o c c u r s
(120a) ma- lāi [CP [S2 PROi mumbai- lā dzā- wa-S2 ] səCP ]
I- dat nom Mumbai- to go- subjunctive- COMP
wāt.ate/wāt.ata
feel.pres s
‘I feel like going to Mumbai.’
Note that such deletion of the vowel is not permitted if the subjects are non-
identical. In (120b) the subject of the matrix clause is ma-lā ‘I-dat’ and of the
embedded clause is tyā-ne ‘he-erg’ and they are not identical.
s u b j e c t o f t h e m a t r i x a n d e m b e d d e d c l a u s e not i d e n t i c a l
(120b) ma- lāi [CP [S2 tyā- nej mumbai- lā dzā- wa-S2 ] ase/
I- dat he- erg Mumbai- to go- subjunctive- COMP
∗
səCP ] wāt.ate/wāt.ata
COMP feel.pres s
‘I think he should go to Mumbai.’
(Kashi Wali, p.c. to Prashant Pardeshi)
Agreement
It is a well-attested fact that in most of the SALs, except in Manda (DR)
(Ramakrishna Reddy 1992b) and Maithili (IA) (Subbarao 2001), the NNS does
not trigger agreement on the verb. However, agreement cannot be considered
as a viable test for subjecthood, as verbs in SALs agree with non-subjects too
(see chapter 4).
166 Non-nominative subjects
m o d a l w i t h a no m i n a t i v e su b j e c t : possibility , a nd capability
Telugu (DR)
(121) vād.ui ı̄ sangatuluj rēpu telusu- kō- galad.ui,∗ j
he this news tomorrow know- VR- can/might
‘He might/can find out this news tomorrow.’
m o d a l w i t h a no n - n o m i n a t i v e s u b j e c t : possibility , a nd
not capability
(122) vād.i- kii ı̄ sangatuluj rēpu teliya- gala- vu∗ i,j
he.obl- dat this news (p) tomorrow known- might/*could- 3 p,nh
‘He will/might get to know this news tomorrow.’
*‘He can get to know this news tomorrow.’
While the modal gala with capability meaning in (121) has a corresponding
negative form (123a), the modal with possibility interpretation does not have a
corresponding negative form (123b).
Further, in the passive voice too, the modal gala ‘can’ behaves differently
from the sentence in active voice. In active voice the modal gala ‘can’ does
not overtly occur in Telugu when the negative occurs as in (124). It is the
agreement on the negative that imparts the capabilitative interpretation. With
5.4 Subject and non-subject properties 167
Telugu uses a lexical passive in such cases (126). A lexical passive does not carry
the overt morphology that a passive predicate carries; it imparts the meaning
of a passive, as the verb in such cases is [−transitive]. Thus, in syntactic and
lexical passives, the predicate is invariably [−transitive].
(126) vād.i- valla ilāt.i cetta panulu kā- vu/
he.obl- by such useless deeds cannot.happen- 3 p,nh
jaraga- vu
will.not.happen- 3 p,nh
‘Such useless things cannot be / will not be done by him.’
Evidence similar to Telugu (DR) can be adduced from Hindi-Urdu (IA) too,
from the difference exhibited by the modal sak-nā ‘can.’ Just like the modal
gala ‘can, might’ in Telugu, saknā ‘can, might’ in Hindi-Urdu (IA) has a capa-
bility and a possibility interpretation in nominative–accusative constructions
(127); there is no corresponding capability interpretation in the dative subject
construction (128). It has only the possibility meaning.
Hindi-Urdu (IA)
(127) ren.u yah bāt kal malūm kar saktı̄ hai
Renu this thing tomorrow learn do can pres
‘Renu can [capability] /might [possibility] find out this tomorrow.’
(128) ren.u ko yah bāt kal malūm ho saktı̄ hai
Renu dat this thing tomorrow learn be can pres
‘Renu might get to know this tomorrow.’ (possibility)
*‘Renu can find out this tomorrow.’ (capability)
From a semantic point of view too, the dative subject construction differs from
the nominative subject construction with regard to the feature of volitionality. It
is generally agreed that the predicate in the dative/genitive subject constructions
is [−volitional].
Krishnamurti (1975), McAlpin (1976), Klaiman (1979) and Pandharipande
(1990) point out that the predicate in the dative/genitive subject construction
is [−volitional]. Hence, adverbs such as kāvāli ani ‘deliberately,’ kōru-koni
‘desirously,’ i.tam gā ‘willingly,’ kutūhalam gā ‘anxiously,’ ātruta gā ‘eagerly’
168 Non-nominative subjects
and anu kōkund.ā ‘unintentionally’ in Telugu (DR), jān būjh kar ‘deliberately,’
cāh kar ‘desirously,’ utsuktā se ‘eagerly’ and samhal kar ‘carefully’ in Hindi-
Urdu (IA) and icche kore ‘deliberately’ in Bangla (IA), and jān. ke ‘intention-
ally’ in Punjabi (Bhatia 1993: 87) cannot occur in a dative, genitive or locative
subject construction (see papers in Verma and Mohanan 1990 for a discussion
on the non-volitional nature of DSCs).15
Evidence from Telugu (DR) and Hindi-Urdu (IA) clearly shows that, while
nominative subjects and dative subjects do share many structural properties,
they differ with regard to the semantic feature of volitionality.
Coordinate reduction
The issue with regard to NNS constructions and coordinate reduction is whether
it is the morphological identity in terms of case marking on the conjuncts that
counts or the grammatical function that is crucial. In Icelandic, the subject of
the second conjunct can be elided, though it might be differently case-marked
from the subject of the first conjunct. We shall demonstrate that three out of four
SALs we checked permit differently case-marked subjects in the two conjuncts.
Thus, our data support the view that “it is not the morphological identity that
counts, it is the grammatical function” (Zaenen, Maling and Thráinsson 1985).
Bayer (2004: 57–58) shows that, in German and Bangla (IA), conjunction
reduction is not permitted when the subjects of the conjuncts are not identically
case-marked, as in (129). Sentence (129) is ungrammatical because “nominative
and genitive subjects cannot mix in deletion contexts” (Bayer 2004: 57–58).
Bangla (IA)
(129) *ami bar.i theke gelam ar *(amar) kanna peyeche
I home from went and I.gen crying came
‘I left the house and I felt like crying.’
(Bayer 2004: 57)16
Sentence (130) shows that the nominative subject of the first conjunct and the
genitive subject of the second conjunct can mix in deletion contexts in Bangla
(IA) and, hence, the genitive subject can be elided.
5.4 Subject and non-subject properties 169
Hindi-Urdu (IA)
(131) ravı̄ ne khānā khāyā aur ø pikcar dekhı̄
Ravi erg meal ate and erg movie watched
‘Ravi ate his meal and watched a movie.’
(132) ravı̄ ghar gayā aur *ø khānā khāyā
Ravi home went and erg meal ate
‘Ravi went home and he ate his meal.’
(Mohanan 1994: 131)
Thus, Hindi-Urdu requires strict case identity of the subjects in both conjuncts.
In contrast, in Nepali (A), sentences of the type in (133), in which the
first conjunct has a dative subject, and the second a nominative subject, are
permitted.
Nepali (IA)
(133) us.lāı̄i jyādai tirkhā lāg-yo ra ɵi gāū-
he.dat much:focus thirsty perceptible-pst.3s and nom village-
eu-t.ā pas-yo
one-cl enter-pst.3sm
‘He was thirsty and went into a village.’
(Ichihashi-Nakayama 1994, as quoted in Bickel 2004: 81)
Telugu (DR), too, is similar to Nepali (IA) and permits sentences of the type in
(134), as does Nepali.
Telugu (DR)
(134) āvid.a ki ēd.upu rānu- u vaccindi proi kas.t.am mı̄da
she dat crying come- conjn came she.nom difficulty on
āpukonu- u āpukondi
stop- conjn stopped
Literally: ‘Crying came to her, but she even controlled it with difficulty.’
‘She was about to cry (but) she controlled herself with difficulty.’
Based on the fact that Bangla, Nepali (IA) and Telugu (DR) permit non-
identical case-marked subjects in two conjuncts, we can tentatively conclude
that coordination can be used as a piece of evidence to demonstrate that NNSs
behave like nominative subjects in this context.
170 Non-nominative subjects
[+s t a t i v e ]
Hindi-Urdu (IA)
(135) un baccõ ko yah bāt mālūm thı̄
those children dat this matter known was
‘Those children knew of this.’
Telugu (DR)
(136) ā pillala- ki ı̄ sangati telusu- at.a
those children- dat this matter known- ev mkr
‘It seems that those children knew this thing.’
[ −s t a t i v e ]
Sentence (137) is an example with a [−stative] predicate.17
Telugu (DR)
(137) ānti ki eppud.ū kōpam vastūn.ē unt.undi
Shanti dat always anger coming.emph keeps
‘Shanti keeps getting angry all the time.’
Kachru (1990: 67), discussing experiencer and other oblique subjects in Hindi
(IA), points out that the dative subject occurs in transient psychological states.
She further points out that transient psychological states, beliefs, knowledge,
want, need, etc., can be expressed with active (nominative subject) constructions
too. To summarize, the predicates that take non-nominative subjects may be
either stative or non-stative.
Case marking
In sentences involving Exceptional Case Marking in SALs, the matrix verb
assigns accusative case to the embedded subject and, hence, it carries the
accusative case marker.
Hindi-Urdu (IA)
(138) ham sab armilā ko ı̄māndār mānte rahe
we.nom all Sharmila acc trustworthy consider kept
‘We all had been considering Sharmila trustworthy.’
(139) ham sab ko armilā (*ko) ı̄māndār lagı̄
we all dat Sharmila acc trustworthy appeared
‘Sharmila appeared to be trustworthy to all of us.’
Sentence (138) is a case of Exceptional Case Marking and the verb mān-nā ‘to
consider’ is [+transitive]. Hence, the predicate assigns accusative case to the
patient armilā ‘Sharmila.’ In contrast, the dative predicate lagnā ‘to appear’ in
(139) is [−transitive], so it cannot assign accusative case to the theme/patient
armilā ‘Sharmila.’ The derived subject armilā does not permit the accusative
case marker ko to occur in (139).
172 Non-nominative subjects
Hindi-Urdu (IA)
(141) surabhi ne kriti ko nācte.hue dekhā
Surabhi erg Kriti acc dancing saw
‘Surabhi saw Kriti dancing.’
If the matrix sentence has a dative predicate, the embedded subject cannot be
case-marked accusative (142) as the dative predicate is [−transitive].
In (142), kriti ‘Kriti’ has to be in the nominative case because the dative
predicate dikhāyı̄ par.nā ‘to appear,’ which is unaccusative, cannot accusative
case-mark the embedded subject. Hence, it is nominative case-marked in (143).
w i t h a no m i n a t i v e p r e d i c a t e
Telugu (DR)
(144) nēnu prasād ni nammakastud.u- gā bhāvistunnā-nu
I.nom Prasad acc trustworthy- adjr consider.1 s
‘I consider Prasad trustworthy.’
In contrast, the verb anipincu ‘to feel’ is a dative predicate, and it too per-
mits exceptional case marking. If our hypothesis that dative predicates are
[−transitive] is correct, then the embedded subject cannot be accusative case-
marked by a dative predicate. If the embedded subject is nominative case-
marked, the sentence is grammatical. Our prediction turns out to be correct in
(145).
exceptional case marking with a dative predicate
(145) nā-ku prasād (*ni) nammakastud.u- gā anipistunnād.u
I-dat Prasad acc trustworthy- adjr appears.3 s,m
‘Prasad appears trustworthy to me.’
Thus, the evidence from case marking in Telugu (DR) and Hindi-Urdu (IA)
shows that the predicate in NNS constructions is [−transitive]. However,
there appears to be some putative counterevidence to our claim regarding the
[−transitive] nature of dative predicates. In Tamil, Malayalam (DR), a dative
predicate, and in Bangla (IA) and Bodo (TB), a genitive predicate, appear to
assign a non-nominative case to the theme.
In standard Hindi-Urdu (IA), in such cases the theme kaun ‘who’ is in the
nominative case, as cāhiye ‘needed’ is [−transitive].
174 Non-nominative subjects
Hindi-Urdu (IA)
(148) āp ko kaun cāhiye
you dat who.nom needed
‘Who do you want?’
Let us first examine the Bangla (IA) data further. In Bangla, the features animacy
and specificity play an important role in the assignment of the accusative case
marker ke to the theme, just as in many Indo-Aryan languages – for example,
in Hindi-Urdu (Mahajan 1990) and in Marwari (IA) (Magier 1987, 1990). The
accusative case marker ke does not occur when the theme is [−definite] and
[−animate].
Bangla (IA)
(149) rina- r kichu bhalo lage na
Rina- gen anything good appear neg
‘Rina does not like anything.’
(150) rina- r kono jini bhalo lage na
Rina- gen any thing good appear neg
‘Rina does not like anything.’
(Probal Dasgupta, p.c.)
Note that the marker ke is not present with the theme kichu ‘anything’ or kono
jini ‘anything,’ which clearly shows that ke is a marker that does not occur
when the theme is [−definite] and [−animate]. If the theme is under focus or
contrastive stress, the marker ke occurs, as in (151), as Probal Dasgupta (p.c.)
points out.
Dasgupta further points out that the correlation between the behavior of the
patient in the experiencer subject sentences (see (149) and (150)) and in the
agent (nominative subject) sentences (see (152) and (153)) is exact. There is
an interaction with animacy and specificity, but that interaction is identical in
the two clause types. Note that the marker ke is not present in (152) and (153),
while it is present in (154), when the theme is under contrastive focus.
t h e m e (i n i t a l i c s ) u n d e r n e u t r a l f o c u s
(152) rina kichu pɔchondo kɔre na
Rina anything liking does not
‘Rina does not like anything.’
5.5 Nature of the predicate in DSCs 175
t h e m e (i n i t a l i c s ) u n d e r c o n t r a s t i v e f o c u s
(154) rina kono jini ke- i otti-otti pɔchondo kɔre na
Rina any thing acc- emph really liking does not
‘Rina does not like really anything at all.’
(Probal Dasgupta, p.c.)
Malayalam (DR)
(155) kut..ti- k’k’ə āna- ye it.am āyi
child- dat elephant- acc liking became
‘The child liked the elephant.’
(Jayaseelan 2004: 229)
Malayalam (DR)
(156a) kut..ti āna- ye is.t.a- ppet.t.- u
child elephant- acc liking- ?- pst
‘The child liked the elephant.’
(Jayaseelan 2004: 229)
When the theme is [−animate] and [−definite], the accusative marker -ye does
not occur (156b).
Thus, the features animacy and definiteness explicate the occurrence of the
accusative marker -ye, and it is not the transitive nature of the predicate that is
instrumental in its presence.
Tamil (DR)
(157) kumār- ukku inta ūr- ai.t teri.y- um
Kumar- dat this place- acc know.fut- 3s,neuter
‘Kumar knows this place.’
(Lehmann 1989: 184)
In Malayalam too, the accusative marker -e occurs with the theme, just as in
Tamil.
Malayalam (DR)
(158) awal.a- kkə awan- e is.t.appet.um
she- dat he- acc like.fut
‘She will like him.’
(Shobha Nair and P. Sreekumar, p.c.)
Note that in Tamil (DR) too, the features [+animacy] and [+specificity] play
a crucial role in the occurrence of the specificity marker, as (159) and (160)
indicate. In (159) and (160), the theme inta id.am ‘this place’ is [+specific] and
[−animate] and no accusative marker occurs with the theme.
In Bodo (TB), the adjective mɯjaŋ ‘good,’ together with a tense marker,
imparts the meaning of ‘like,’ and this predicate assigns genitive case ha ‘of’
to its subject. Recall that adjectives behave like verbs in many Tibeto-Burman
languages (see chapter 2 for details). The patient in such cases is accusative
case-marked by khɯu.
Bodo (TB)
(163) khampha ha laogi- khɯu mɯjaŋ- mɯn
Khampha gen Laogi- acc good- pst
‘Khampha liked Laogi.’
We do not have further data to show that the accusative marker khɯu is a
specificity marker in Bodo.
In conclusion, though the phenomenon of accusative/dative case marking
of the theme in Bangla (IA), Malayalam and Tamil (DR) seems to suggest that
the predicate in DSCs is [+transitive], we have demonstrated that the marker
that occurs with the theme in such constructions is a marker of specificity and
animacy as Magier (1987,21 1990) and Mahajan (1990) have shown for Marwari
and Hindi-Urdu (IA), respectively. Hence, predicates in NNS constructions are
syntactically [−transitive]. We agree with Alice Davison (p.c.) who points out:
“The issue of whether dative subject verbs are [+transitive] is complicated . . .
the dative/ergative near minimal pairs in Hindi like dekh-nā ‘to see’ and dikh-
nā/dikhāı̄ de-nā ‘to be visible’ are both bivalent, i.e., transitive in argument
structure, though the dative subject version does take intransitive vectors.” She
points out that the fact that dative subject predicates take small clauses and
participial complement clauses shows their transitive nature. Hence, we feel
178 Non-nominative subjects
that one may, with reasonable certainty, conclude that dative predicates are
semantically transitive but syntactically intransitive.
Since the object marker is not an accusative case marker, its occurrence does
not count as counterevidence to the claim that the predicate in non-nominative
subject constructions is [−transitive].
Aissen (2003) discusses DOM, differential object marking, in which some
objects are case-marked, and some others are not, depending upon the semantic
and pragmatic features of the object. Aissen points out that DOM depends on
two features – animacy and definiteness – and they compete with each other
for dominance. While Persian prefers specificity, Hindi chooses animacy. The
case marking of theme in the DSC (dative/genitive subject construction, to be
very specific) in Bangla, Malayalam and Tamil, for example by the accusative,
we have observed, depends on animacy/specificity, independent of transitivity
in the [−NNS construction].
Recall that, in nominative–accusative constructions too, the accusative
marker is associated with transitivity and animacy/specificity, and hence should
be treated as a specificity marker, as it denotes specificity (see Magier 1987,
1990; Mahajan 1990).
Thus, with regard to differential object marking (DOM) in SALs, when
the accusative case marker denoting specificity occurs, the predicate is
[−transitive] in the DSC (dative/genitive subject construction), and the pred-
icate is [+transitive] in the nominative subject construction. Based on these
facts, we propose the following parameter to account for this variation.
The Differential Object Marking (DOM) parameter: When the noun
phrase is accusative case-marked, the object marker is associated with
either transitivity and animacy/specificity in the [+NNS construction],22
or purely animacy/specificity, independent of transitivity, in the [−NNS
construction].
Discussing accusative case in Kannada, Lidz (2006) notes that there are two
kinds of specificity: positional and inherent. Inherent specificity means case-
marked NPs independent of their position in a sentence get a specific interpreta-
tion. Non-case-marked NPs too can get specific interpretation depending on the
position of occurrence, which he labels as positional specificity. The accusative
marking in DSCs in Tamil, Malayalam (DR) and Bangla (IA) comes under
inherent specificity.
Anaphors
The second piece of evidence comes from anaphors in Telugu (DR) (Subbarao
and Bhaskararao 2004).
When the matrix sentence has a nominative predicate and the embedded sub-
ject is coindexed with the matrix subject and ECM takes place, a reduplicated
anaphor can occur as in (164).
5.5 Nature of the predicate in DSCs 179
Telugu (DR)
(164) mādhuri tana- ni tanu andagatte- gā bhāvistundi
Madhuri self- acc self pretty- adjr considers-3 s.nm
‘Madhuri considers herself pretty.’
(Subbarao and Bhaskararao 2004: 178)
However, when the matrix sentence has a dative predicate, only the simplex
form of the anaphor in nominative case can occur, and a reduplicated form in
accusative case is not permitted, as in (165).
A complex anaphor is not permitted due to the fact that the matrix verb which
is [−transitive] in (165) cannot assign accusative case to the embedded subject
by Exceptional Case Marking (ECM), unlike the [+transitive] verb in (164).
That is, NNS predicates cannot exceptionally case-mark and, hence, they are
[−transitive].
Passivization
Bhatt (1999) demonstrates that sentences with a dative subject do not passivize
in Kashmiri. This phenomenon is found in other SALs too.
Kashmiri (IA)
(166) *rām- an vuch [humis lar.k- as yi kūr khar- ni
Ram- erg saw that boy- dat this girl (nom) hate- pass
yiv- ān]
come- imperf
‘Ram saw the boy being hated by the girl.’
(Bhatt 1999: 201)
Hindi-Urdu (IA)
(167) *mādhurı̄ ko is bāt par jān būjh kar gussā ā gayā
Madhuri dat this news on intentionally anger came
Intended meaning: ‘Madhuri’s anger went up intentionally at this news.’
In Telugu and the other Dravidian languages as well, such sentences are not
permitted.
Hindi-Urdu (IA)
(168) *urmilāi [PROi dat bhūkh lagne ke liye] rasoı̄ mẽ gayı̄
Urmila (controllee) hunger feel in order to kitchen in went
Literally: ‘*Urmila went into the kitchen in order to feel hungry.’
Telugu (DR)
(169) d.ākt.aru gārui [PROi dat jvaram tagga- d.āni ki]
doctor polite mkr (controllee) fever become less- in order to
mandu tı̄sukont.unnāru
medicine is taking
‘The doctori is taking some medicine in order to bring hisi fever down.’
In (170a), the verb agrees with the subject vād.u ‘he’ in 3rd person singular
masculine. In (170b), the verb does not agree with the subject, as the latter is
dative case-marked, and there is no other DP in the nominative case; kōpam-gā
‘angry’ is an adjective, and there is no agreement of the verb with an adjective.
The verb, therefore, exhibits default agreement, which is 3rd person singular
non-masculine in Telugu.
182 Non-nominative subjects
In (171a), the verb agrees with the subject āme ‘she’ in 3rd person singu-
lar non-masculine [+animate]. In (171b), the verb does not agree with the
subject – as the latter is dative case-marked – and, hence, exhibits default
agreement.
Our claim is: when there is a predicate adjective and the verb exhibits default
agreement, the subject is inherently assigned non-nominative case (dative or
locative) by the predicate compositionally. When the verb exhibits person
agreement in masculine or non-masculine [+animate], the subject is in the
nominative case. That is, the feature [+/−animate] plays a role in case assign-
ment. Alice Davison (p.c.) suggests that default agreement in T should be the
consequence of non-nominative case, rather than the way it is suggested here.24
We leave the issue open for further research.
We observe that the verb phrases in (a) and (b) sentences are identical.
What really distinguishes (a) sentences from (b) sentences is agreement. While
(a) sentences exhibit subject–verb agreement, (b) sentences exhibit default
agreement. The verb un alone cannot assign inherent case to its subject. The
adjective + verb to be + default agreement together play a crucial role in
assigning inherent case to the subject in the (b) sentences. This implies that
information concerning agreement should be available in the lower thematic S
for the proper assignment of inherent case to the subject. Since the occurrence
of the dative case on the subject in our approach is the result of the feature
agreement found in the lower thematic S, it appears that it is not what bottom
to top MERGE/checking would derive.
Telugu (DR)
While the verb in the nominative subject construction agrees with the subject,
the verb in the non-nominative (dative subject) construction agrees with the
theme.
nominative subject construction
(172) āvid.ai piccij ekkipōyin- dii,∗ j
she craziness climb- 3 s,nm [+animate]
‘She went crazy.’
dative subject construction
(173) āvid.a kii piccij ekkipōyin- di∗ i,j
she dat craziness climb- 3 s,nm [−animate]
Literally: ‘Craziness climbed on to her.’
‘She went crazy.’
Sentences (172) and (173) differ in the feature animacy alone, with regard to
agreement.
If the agreement marker is coindexed with the subject as in (172), the case that
is assigned to the subject is nominative. If the agreement marker is coindexed
with the theme as in (173), the case that is assigned to the subject is non-
nominative. Hence, for the proper assignment of case to the subject, the complex
predicate plus information concerning animacy in the agr phrase are required.25
We have shown earlier that the predicate in a DSC in Telugu is [−transitive].
There are verbs in Telugu which are transitive in the nominative–accusative
construction, and are also used in the DSC. The set includes: pōyu ‘to pour,’ ‘to
serve,’ ‘to pass something, e.g. urine’; oppu ‘to agree’; vēyu ‘to put,’ ‘to keep,’
‘to serve,’ ‘to wear,’ ‘to take something, e.g. medicine’; pet..tu ‘to put,’ ‘to keep,’
‘to insert’; tappu ‘to miss’; tippu ‘to turn,’ ‘to take someone around’; tōyu ‘to
push’; pat..tu ‘to catch’; ād.u ‘to play’; tagulu ‘to touch’; kot..tu ‘to hit’; mottu ‘to
hit’; kat..tu ‘to tie’; vācu ‘to have a swelling’; kalugu ‘to have’; ekku ‘to climb.’
The verb is used transitively in (174) in a nominative–accusative sentence and
intransitively in (175).
(174) āmei pillala- ki cokkāj vēs- in- dii,∗ j
she.nom children- dat shirt put.on- pst- 3 s,nm [+animate]
‘She put the shirt on the children.’
Further, the verb in (174) does not agree with the theme, whereas it invariably
agrees with the theme in the nominative in (175). We have already shown that
the theme and verb together assign theta role to the subject in the DSC. We now
propose that the theme and the verb together assign inherent case to the subject
vP-internally in the lower (thematic) clause in view of the following points.
If it is only the verb that assigns inherent case to the subject in (174),
the question that needs to be answered is: why doesn’t the same verb assign
inherent case to the subject in (175)? It is because of the transitive nature of
the verb in (174) that the theme is accusative case-marked, while it needs to
be explained why the theme in (175) cannot be accusative case-marked, as the
ungrammaticality of (176) shows.
One might say that there are two different sets of transitive verbs which are
homophonous, and a specific set (call it Set A) assigns inherent case while
the other set (call it Set B) assigns nominative case. Such an assertion is
counterintuitive, and it is not clear how such a stipulation can be formulated
and implemented in the grammar of a language. Hence, we conclude that it is
the theme/adjective, together compositionally with the predicate, that assigns
dative case to the subject.
Sentence (178a) with the dative subject in the present tense is ungrammatical,
according to Jayaseelan.
dative subject i n present t e n s e – u n g r a m m a t i c a l
(178a) *awal.a- kkə awan- e is.t.appet.unnu
she- dat he- acc like-pres ind
‘She liked him.’
When the predicate is..tappet. ‘to like’ occurs in past tense, nominative subject
is questionable.
nominative subject i n past t e n s e – q u e s t i o n a b l e
(178b) ?awal. awan- e is.t.appet.t.u
she he- acc like-pst
‘She liked him.’
According to Jayaseelan (1999: 105), the dative case marker is assigned to the
subject “at the point where is..tappet. and the past tense marker are put together
[i.e. MERGE].”
In Hindi-Urdu (IA) as well, modals play a role in the assignment of dative case.
The modal sak-nā ‘to be able to’ takes a nominative subject, as in (181), while
the modal cāhiye ‘must’ requires a dative subject, as in (182).
Hindi-Urdu (IA)
(181) bacce ghar jā sakte hãı̃
children (nom) home go can pres
‘The children can go home.’
186 Non-nominative subjects
The verb ‘to come’ is used in Hindi-Urdu, Nepali (IA) and Newari (TB)
(T. K. Kansakar, p.c.), too, as a verb of motion, and to denote possession of
knowledge and skill.
Hindi-Urdu (IA)
ā ‘ t o c o m e ’ a s a v e r b o f m o t i o n
(185) pratimā daftar se āyı̄
Pratima office from came
‘Pratima came back home from the office.’
ā ‘ t o c o m e ’ a s a v e r b i n d i c a t i n g k n o w l e d g e ((73) repeated here)
(73) is panjābı̄ lar.kı̄i ko bharat nāt.yamj ātā∗ i,j thā∗ i,j
this Punjabi girl dat classical Indian dance come was
‘This Punjabi girl used to know the classical Indian dance.’
5.6 Inherent case assignment in DSCs 187
Note that the verb ā ‘to come’ in Hindi-Urdu (IA), by itself, cannot assign
dative case inherently to its subject as the verb does not have any information
about the nature of activity that it represents. Combined with the theme which
involves knowledge or a skill, the verb ā ‘to come’ forms a complex predicate
and only then is it able to assign dative case to its subject.
Thus, a predicate with subject and other constituents pro-dropped in
Telugu (DR) or Hindi-Urdu (IA) is ambiguous between the nominative sub-
ject and dative subject readings. The following dialog from Telugu (DR) is
illustrative.
Telugu (DR)
(186a) mādhuri rātri ki vastund(i)- ā
Madhuri night dat comes.3 s,nm pol q mkr
‘Will Madhuri come tonight?’
(186b) pro pro vastundi
comes.3 s,nm
‘She’ll come.’
(187a) mādhuri ki telugu vastund(i)- ā
Madhuri dat Telugu comes.3 s,nm- pol q mkr
‘Does Madhuri know Telugu?’
(187b) pro pro vastundi
comes.3 s,nm
‘She knows it.’
Malayalam (DR)
(193) avan- ə kai- kku katam parri
he- dat hand- dat injury happened
‘He got hurt on his hand.’
Telugu (DR)
(194) *vād.i- ivāl.a cēti ki debba tagil- in- di
his- today hand dat injury hit- pst- 3 s,nm
Kannada (DR)
(195) *avana indu kan.n.- ige gāyav āgide
his today eye- dat injury happened
Telugu (DR)
(196) vād.i- ki ivāl.a cēti ki debba tagil- in- di
he.obl- dat today hand dat injury hit- pst- 3 s,nm
‘He got hurt on his hand today.’
Kannada (DR)
(197) avan- ige indu kan.n.a- ige gāya(v)- āgide
his- dat today eye- dat injury- happened
‘He got hurt on his eye today.’
The third piece of evidence for this construction comes from particles in
Telugu. The focus particle aitē ‘as for,’ the inclusive particle kūd.ā ‘also’
and -ē ‘the emphatic particle’ can be inserted after the possessor. We provide
only one example with the particle kūd.ā ‘also.’
The fourth piece of evidence comes from the contrastive focus clitic (CFC) -ā27
in Telugu.
t h e ‘w h o l e ’ q u e s t i o n e d
(201) vād.i- k(i)- ā cēti- ki debba tagil- in- di
he- dat CFC hand- dat injury hit- pst- 3 s,nm
‘Is it he who got hurt on his hand?’
t h e ‘p a r t ’ q u e s t i o n e d
(202) vād.i- ki cēti- k(i)- ā debba28 tagil- in- di
he- dat hand- dat- CFC injury hit- pst- 3 s,nm
‘Is it on his hand that he got hurt?’
Thus, these arguments indicate that, in the Single Dative Subject construction,
the genitive case-marked DP (the ‘whole’) and the dative/locative case-marked
DP (the ‘part’) together form a single constituent, while in the Double Dative
Subject construction, the two inherently case-marked DPs are separate con-
stituents. Subbarao and Bhaskararao (2004) show how inherent case is assigned
in sentences (188)–(190), and (191)–(193). Since the genitive case-marked DP
and the dative case-marked DP together function as a single constituent in
sentences in (188)–(190), the predicate compositionally assigns dative case to
the whole DP vP-internally in the thematic S.
192 Non-nominative subjects
5.8 Conclusion
This chapter provides a description of the NNS constructions in SALs, and
presents a discussion of the domains of occurrence of the NNS. It also focuses
on some subject and non-subject properties of these constructions. The stative
and non-stative, as well as the intransitive, nature of the predicate in NNS
constructions are also discussed.
We have argued that: (i) the verb in the DSC is intransitive; (ii) all DSCs
are inherently case-marked; (iii) such inherent case marking cannot be done by
an intransitive verb alone, but by the whole predicate compositionally, consist-
ing of a theme or adjective along with the [−transitive] verb; (iv) information
concerning agreement should be available vP-internally (in the lower thematic
S) for proper assignment of inherent case to the non-nominative subject; and
(v) the accusative/dative marker that occurs with the patient/theme in NNS con-
structions is a specificity marker and, hence, does not count as counter-evidence
to the claim that dative predicates are [−transitive]. A brief summary of the
four major analyses that account for the NNS constructions is also presented.
Finally, the Double Dative Subject construction that occurs in sentences that
express inalienable possession was discussed.
6.1 Introduction
Complementation1 has been a topic of intense study, in SALs as well as many
other languages, starting with the work of Rosenbaum (1967). In this chapter,
we discuss the complement constructions in SALs, which are of great empirical
and theoretical significance in view of the fact that the complementizer may
occur:
(i) exclusively in clause-initial position (labeled as left peripheral or pre-
sentential or initial complementizer [IC]), as in many Indo-Aryan (Kash-
miri, Hindi-Urdu, Kurmali and Punjabi) and Munda languages (Ho,
Santali, Kharia and Korku), and the non-verb-final Mon-Khmer Khasi;
or
(ii) exclusively in clause-final position (labeled as right peripheral or post-
sentential or final complementizer [FC]), as in most of the Dravidian
languages and all Tibeto-Burman languages, the Munda language Mundari
and the Indo-Aryan Sinhala; or
(iii) in both clause-initial and clause-final positions, as in Eastern Indo-Aryan
Nepali, Assamese, Bangla and Oriya, or Western Indo-Aryan Gujarati
and Marathi, or in the transplanted language Dakkhini Hindi-Urdu in the
South of India.
In this chapter, we shall show how the heterogeneous distribution of comple-
mentizers in SALs correlates with question markers, clefts and subordinate
relative clause markers. Several issues with regard to the movement of clauses
and operator scope, too, will be discussed.
An embedded clause may occur as a complement either in the subject or
object position of a predicate (verb/adjective) of the matrix clause, or the
next higher clause. The former is a case of subject complementation and the
latter, a case of object complementation. It is the complementizer that links
the embedded clause to the next higher clause. It may occur to the left of the
embedded clause as in (1) in Hindi-Urdu (IA), or to the right of the embedded
clause as in (2) in Kokborok (TB). Thus, (1) instantiates the case of the IC
ki ‘that’ in Hindi-Urdu, and (2) the FC hn-i ‘that’ in Kokborok (TB); hn-i
193
194 Complementation
Kokborok (TB)
(2) akuŋ [CP [S2 khumti naithɔkS2 ] hn-iCP ] sa- kha
Akung Khumti beautiful say-cpm that (FC) tell- pst
‘Akung said that Khumti is beautiful.’
Marathi and Gujarati (IA) have: (i) two FCs derived from a pronoun: in Marathi
(IA), for example, the FC is derived from asa/asə ‘like this,’ as in (3), and in
Gujarati it is evũ/em ‘like this, this way, thus’; and (ii) another FC, namely
te ‘that’ in Marathi (Wali 2006: 210) as well as Gujarati (IA) (Cardona 1965;
Masica 1999; P. J. Mistry and Trupti Nissar, p.c.).
Marathi (IA)
(3) [CP [S2 tū tithe kā dzātes?S2 ] asaCP ] ti- ne malā witšārle
you there why go.2 s,f quot she- erg I.dat ask.pst.3 s, neuter
She asked me, “Why do you go there?”
(Pandharipande 1997: 70)
IC as well as the FC. Some central and north Dravidian languages do not have
the quotative (FC) (Masica 1976: 189).
The generalization that non-verb-final languages have an initial complemen-
tizer (IC), and verb-final languages have a final complementizer (FC) holds
good only for the non-verb-final language Khasi, which has an IC, and the
literary verb-final Dravidian languages and all Tibeto-Burman languages in the
subcontinent, which have only an FC. Some Munda and Indo-Aryan languages
have both an IC and FC. Hence, they do not conform to the cross-linguistic
generalization that verb-final languages have an FC. It is quite possible that
some Munda languages lost the quotative FC as a complementizer due to the
influence of the major superstrate languages such as Hindi and Sadani. The
absence of an FC in Hindi-Urdu, Kashmiri and Punjabi, too, could be due to
the influence of Persian, which has only an IC ke.
It is worth noticing that some Munda languages, for example Ho, Santali
and Juang, have a post-sentential quotative (complementizer): some languages,
like Ho and Santali, have the quotative functioning as a purposive marker,
and some languages, like Kharia and Ho, have the quotative functioning as a
reason marker as well. These functions in different languages of the Munda
family indicate that the quotative must have been in use as subordinate linker
(complementizer, to be specific) in all Munda languages at some stage, but
subsequently must have been lost in some languages.
In this chapter, we discuss several issues that concern complement clauses
in SALs. First, we discuss the nature of the quotative, reported speech vs.
indirect speech with regard to the phenomena of person agreement and tense
harmony. We focus our attention on the position of occurrence of an FC-clause,
the nature of the left and right peripheral COMP, the position of occurrence of
an IC-clause and the Case Resistance Principle of Stowell (1981). We present a
detailed analysis of the parametric variation among SALs in the expression of
wide and narrow scope devices. Several issues that concern the problems with
the Rightward Extraposition analysis, the direction of c-command in sentences
with Negative Polarity Items, COMP agreement (Juang, Mundari) and, finally,
the reanalysis of the initial complementizer resulting in syntactic changes due
to convergence are also discussed.
We shall also focus our attention on one of the crucial issues that concern
complementation: that is, is there any correlation between the semantic nature
of the predicates – such as factive or non-factive – the position of occurrence
and the choice of the complementizer? We demonstrate that there is.
This chapter is organized as follows. Section 6.2 discusses the position of the
COMP and some crucial functions that it performs in SALs. Sections 6.3 and 6.4
deal with structural differences between an IC and an FC. Some notable issues
include: how the subject of a finite CP clause gets case-marked by the matrix
verb and whether such marking violates the Phase Impenetrability Condition
196 Complementation
of Chomsky (2001) (see the Website); and COMP agreement with the matrix
subject (see 6.3.4). Section 6.5 focuses on the position of occurrence of an
IC- and an FC-clause and the Case Resistance Principle (CRP). Section 6.6
deals with wh scope and the factors which permit and limit wide wh scope.
Section 6.7 provides arguments against Rightward Extraposition in finite CP
clauses and discusses some issues related to it. Section 6.8 concludes the
chapter. The Appendix of this chapter presents a universal constraint labelled
as the Final-over-Final Constraint (FOFC) and discusses how the diachronic
changes that took place in the position of occurrence of the complementizer in
SA language contact situations strictly obey this universal constraint.
Some other crucial issues related to complementation in SALs are also
discussed separately on the Website. These include: (i) formal syntax as a tool
for explicating a typological distinction between the English-type languages
and the Korean- and Kashmiri-type languages; (ii) the direction of c-command
and the occurrence of the Negative Polarity Items; and (iii) syntactic reanalysis
in language contact situations, to demonstrate how the IC of Hindi-Urdu (IA)
is reanalyzed in Dakkhini Hindi-Urdu (IA) due to contact with Telugu (DR).
Category 1a: the quotative FC derived from the verb ‘to say’
All the Dravidian and the Tibeto-Burman languages, IA languages – Marathi,
Nepali, Sinhala – and the Munda languages – Mundari, Ho and Juang – have
the quotative FC derived from the verb ‘to say.’
Category 2: only an IC
Indo-Aryan languages such as Hindi-Urdu, Kashmiri and Punjabi, Khasi (Mon-
Khmer) and some Munda languages have only an IC.
Category 3: an IC as well as an FC
Assamese, Bangla, Gujarati, Konkani, Marathi, Nepali, Oriya (IA) and Juang
(Munda) have an IC as well as an FC.
6.2 Position of occurrence of COMP 197
Khasi (Mon-Khmer)
(4) u- kmεn u- m- ŋεyt [CP ba [S2 ki- brēw ki- hyar patεŋ
3m- Kmen 3m- neg- believe IC p- humans p- descended
na- ki- šrē?S2 ]CP ]
abl- p- monkey
‘Kmen does not believe that human beings descended from monkeys.’
Ho (Munda)
(5) somai ur.ū- ke- d.- a [CP ci ramj ø∗ i/j suku- n- tan- aCP ]
Soma think- pst- tr- fin IC Ram self like- VR- pres- fin
‘Somai thought that Ramj likes himself∗ i/j .’
(Koh and Subbarao, ms)
In some languages (e.g., Oriya and Marathi), the IC and FC may occur in
the same sentence, where the IC precedes and the FC follows the embedded
complement.
198 Complementation
Oriya (IA)
(6) se kɔt.ak jibɔ boli bhabu thila je kintu gad.i nɔ thi.ba-
he Cuttack will go quot was thinking COMP but train not be.inf-
ru jai parila nai
due to could go not
‘He was thinking of going to Cuttack but he could not go as there was
no train.’
(P. P. Mahapatra and Prakash Patnaik, p.c.)
Marathi (IA)
(7) anū mhant.e [kı̄ [[tı̄/mı̄ hindı̄ šikel] (asa/ mhan.ūn)]]
Anu say.pres.3s IC she/I Hindi learn.fut.3s FC/ FC
‘Anu is saying [that she will learn Hindi].’
(Pandharipande 1993: 2)
Malayalam (Hany Babu and Sobha Nair, p.c.), Tamil (Arulmozhi, p.c.) and
Telugu (DR) have two FCs occurring in a row, as in (8) and (9). For example,
the FC in Tamil enru ‘having said,’ is a grammaticalized form of the verb en ‘to
say.’ The other FC that can follow it is solli ‘having told,’ a grammaticalized
form of the verb soll ‘to tell.’ We label them as FC1 and FC2.2
Tamil (DR)
(8) nı̄ vara.villai enru (solli) nān vara.villai
you came.not FC1 (REASON) FC2 I came.not
‘I did not come because you did not come.’
(Arulmozhi, p.c.)
FC1 enru has reason interpretation in (8). FC2 can occur only if FC1 has
reason interpretation; and FC2 can be optionally dropped, while FC1 cannot
be in Tamil or Telugu.
A similar situation obtains in Malayalam too.
Malayalam (DR)
(9) nı̄ varinn.illa ennu (para
u) nān vann.illa
you came.not FC1 (REASON) FC2 I came.not
‘I did not come because you did not come.’
(Sobha Nair, p.c.)
v 2 (i n i t a l i c s) i n s 2 i n a c o m p l e x s t r u c t u r e
(10b) tem dop ki [S2 rāth dyut lar.k- an
V2
he.erg said comp yesterday gave boy- erg
tswat.h dār- yithS2 ]
trash throw- cpm
‘He said that the boy threw (away) the trash yesterday.’
(Bhatt 1999: 83)
Thus, there is no asymmetry between the root and embedded clause in Kashmiri
(IA) complements,3 unlike in German, which exhibits asymmetry.
Bhatt and Yoon (1991) and Bhatt (1999) show that the FC or IC functions like
a subordinator linking the main clause and the embedded clause. In this context,
the idea that CP resolves into mood and subordination, proposed by Bhatt
and Yoon (1991) and Bhatt (1999), is worth considering, and is presented in
appendix 6.1 on the Website.
In this section, we have provided data that demonstrate: (i) the positions of
occurrence of the IC and FC; and (ii) the occurrence of double complement
clauses in SALs. We have shown that, unlike German, Kashmiri retains the
V2 order in the main as well as subordinate clauses. We show in appendix 6.1
how the separation of subordinator and mood as separate categories helps to
explicate the absence of asymmetry in root and embedded complements in
Kashmiri.
The IC and FC perform several functions other than just as complementizers,
which we present now.
200 Complementation
Kashmiri (IA)
(15) b drās ki/zi su av
I left IC he came
‘As soon as I left, he came.’
(Kak, ms)
Kashmiri (IA)
(16) bad.šah [yus ki akh jān pādšah o:s] o:s akh ke:šur
Badshah who IC one good king was was one Kashmiri
‘Badshah, who was a good king, was a Kashmiri.’
(Kak, ms)
Hindi-Urdu (IA)
(17) ašokā jo ki ek mahān rājā the bar.e dayālu the
Ashoka who IC a great king was very kind was
‘Ashoka, who was a great king, was very kind.’
In Gujarati (IA) too, the IC ke occurs in relative clauses, but it precedes the
relative pronoun (P. J. Mistry, p.c.), whereas in Hindi-Urdu, Kashmiri and
Punjabi (IA), the IC follows the relative pronoun.
Gujarati (IA)
(18) mān.as ke je letter lakhe che te ā story jān.e che
man IC who letter is writing he that story knows pres
‘The man who is writing a letter knows this story.’
(P. J. Mistry, p.c.)
We mention below a few of the functions that the quotative (FC) performs in
SALs (see Kachru 1979, Subbarao et al. 1989 and Patnaik 1998 for a detailed
description).
(i) As a complementizer
In Gujarati (IA), Malayalam (DR) and Mizo (TB), the embedded clause with
an FC may occur to the left of the matrix subject.
Gujarati (IA)
(19) tεno bhāı̄ pās thayo εm te bolyo
he younger brother pass be.pst FC he said
‘He said that his younger brother passed.’
(Lambert 1971: 205, as quoted in Masica 1991: 403)
Malayalam (DR)
(20) [CP [S2 kut.t.i uraŋŋiS2 ] ennəCP ] amma paraŋŋu
child slept say.cpm (FC) mother said
‘The mother said that the child slept.’
(Hany Babu 1997)
Mizo (TB)
(21) [CP naktuk cu colhni a- ni- aŋ tiʔCP ] zova- n
tomorrow def holiday 3 s- be- fut comp Zova- erg
a- hria
3 s- know
‘Zova knows that tomorrow will be a holiday.’
(Subbarao and Lalitha Murthy, ms)
The FC, a quotative, in Gujarati is -εm (19), in Malayalam ennə (20) and in
Mizo tiʔ (21). The FC, derived from the verb to say, is: (a) a participial form
in Dravidian, Indo-Aryan languages, Mundari, Santali (Munda) and Kokborok
(TB); (b) an infinitival form in Manipuri (TB); and (c) a stem form of the verb
in Mizo and Hmar (TB).
Sinhala (IA)
(23) tāttage upandinə kiyəla api kēk hæduwa
father’s birthday quot we cake made
‘We made cake on father’s birthday’ (i.e., ‘since it was his birthday,’ or
‘realizing it was his birthday’).
(Gair 2003: 807)
Kokborok (TB)
(25) aŋ khumti nɔ nai- na hn-i phai- kha
I Khumti acc see- inf say-cpm (quot) come- pst
‘I came in order to see Khumti.’
(Subbarao et al. 2010: 29)
Santali (Munda)
(26) i
- dɔ nɔtε- ge maraŋ hɔr- thεn guti- k’-a
I- topic here- focus big person- dat serve- middle indicative
mεnte orak’- khɔn- i
odok- hij- uk’- kan-
purpose mkr house- abl- 1s come out- come- middle- imperf-
tahεkan- a
copula.pst- ind
‘I was coming from home in this direction, to serve under a big person.’
(Neukom, ms)
(iv) As an identifier
Just as in Sanskrit (Hock 1982a), the quotative is used in naming and labeling in
Marathi, Assamese (IA), the Dravidian languages and in some Tibeto-Burman
languages (Subbarao et al. 1989).
204 Complementation
Marathi (IA)
(27) tyā gāwāt widzay mhan.ūn/ asa ek mulgā malā bhet.la
that town. loc Vijay quot (identifier) that one boy I.dat meet.pst.3sm
‘In that town I met a boy named Vijay.’
(Pandharipande 1997: 6)
Telugu (DR)
(28) indira anē maniši ikkad.i-ki vaccindi
Indira quot (identifier) person here-to came
‘A person named Indira came here.’
Out of more than thirty functions that the quotative performs in SALs
(Subbarao et al. 1989), we have discussed only a restricted set due to limi-
tations of space. Though both IC and FC function as complementizers, there
are differences in their syntactic behaviour, which we discuss in the following
section.
Telugu (DR)
(29) mādhuri ēmi (ani) andi
Madhuri what comp said
‘What did Madhuri say?’
(30) tanu pei.i.i cēsu konnānu *(ani)
self marriage do VR.pst.agr comp
‘that she got married’
In Sema (TB) and Dakkhini (IA) too, the FC, the quotative, cannot be dropped
in an elliptical answer.
6.3 Structural differences – IC and FC 205
Sema (TB)
(31) pa- ye kla- va *(pi)
he- nom wed- pst FC
‘that he got married’
(Achumi 2000; Achumi and Subbarao 2005)
Dakkhini (IA)
(32) unõ bajār ku jāte *(kate)
they market to are going FC
‘(that) they are going to the market’
(Harbir Arora, p.c.)
The data demonstrate that the FC is closely tied in or attached to the embedded
clause.
We now demonstrate that, in contrast, the IC cannot occur in sentence-initial
position in an elliptical answer in Hindi-Urdu (IA).
Hindi-Urdu (IA)
(33) mādhurı̄ ne kyā kahā
Madhuri erg what said
‘What did Madhuri say?’
(34) (*ki) mãı̃ ne šādı̄ kar lı̄
IC I erg marriage do took
‘that I got married’
In Kashmiri (IA), the elliptical answer with ki/zi is grammatical, although not
very commonly used. Further, the COMP in an elliptical answer is also used to
provide emphasis to the answer (Kak, ms).
elliptical answer
Kashmiri (IA)
(35) (ki/zi) burhān mūd
IC Burhan died
‘that Burhan died’
(Kak, ms)
Ho (Munda)
(36) soma cinā kaji- tad.- a
Soma what say- pres- [+fin]
‘What does Soma say?’
206 Complementation
Bangla (IA)
(38) (je) o bie koreche
COMP she marriage has done
‘(that) she got married’
The pronominal clitic of the matrix clause subject cliticizes onto the right of
the COMP (39) in Mundari (Munda). The post-sentential COMP in Mundari is
a quotative form of the verb to say. The first person subject agreement marker
-
occurs to the right of the FC mente.
Mundari (Munda)
(39) proi [CP somaj - hiju?- aka- n- a mente-CP ]
i,∗ j ayum-
I Soma- come- asp- [−tr]- [+fin] comp- 1 s hear-
ke - d- a
asp- pst- [+fin]
‘I heard that Soma came.’
(Osada 1992: 67; the glosses have been slightly modified)
In (39) the subject agreement marker cliticizes onto the right of the preverbal
constituent. The embedded S plus the COMP is the preverbal constituent in
(39), and hence the agreement marker occurs to the right of the COMP mente.
In Juang (a Munda language spoken in Orissa) too, the final complementizer
gamɔjɔ agrees with the subject of the matrix clause. The complementizer
gamɔjɔ is the quotative form of the verb to say.
Juang (Munda)
(40) nii amte [rabij ɔpɔmanian nei,∗ j - gamɔjɔ] nei - gamɔ
we you-acc Ravi was insulted 1 pn- comp 1 pn- told
‘We told you that Ravi was insulted.’
(Patnaik 1999)7
selected by factive verbs, attitude verbs, verbs of perception (in their epistemic
reading) as well as by nominals.” FCs on the other hand are restricted in various
ways. They do overlap in the semantic selection of verbs, but, while certain
predicates, such as to know, to think, to hear, etc., permit both an IC and an
FC, predicates such as to see, to be possible, to be likely, etc., do not (Bayer
2001: 15).
Bangla (IA)
(41) [ram kolkatta- y jacche bole] {janlam/ bhablam/ šunlam/
Ram Kolkata- loc goes quot knew-1 thought-1 heard-1
*dekhlam/ *ɔšombhob}
saw-1 impossible
‘That Ram is going to Kolkata [I knew / I thought / I heard / *I saw/*is
unlikely].’
Bangla (IA)
(43) ram ke aš- be (bole) jiggeš koreche
Ram who come- will quot question made
‘Ram asked who will come.’
(44) ram jiggeš koreche (*je) ke aš- be
Ram question made IC who come- will
‘Ram asked who will come.’
6.3 Structural differences – IC and FC 209
Hindi-Urdu (IA)
(45) tum jānte ho [ki [us ne kyā kiyā]]
you know pres comp he/she erg what did
‘You know what he/she did.’
Marathi (IA)
(46) mohan ne witšārle [kı̄ [sudhā kut.he gelı̄] {asa/ mhan.ūn}]
Mohan agentive asked IC Sudha where went FC quot
‘Mohan asked where Sudha went.’
(Pandharipande 1997: 6)
6.3.8 Linear order of the COMP (IC and FC) and yes/no (polarity)
question words: a mirror image relationship
Rizzi (1997) proposes that CP has a much finer “articulated structure,” and is
really a series of projections consisting of four possible functional heads such
as Force, Topic, Focus, etc. Davison (2007a) following Rizzi’s idea suggests
that FCs are in a different projection from ICs. She discusses the case of ICs
in languages such as Hindi, Punjabi and Kashmiri, and argues that the IC ki
‘that’ in Hindi-Urdu is a marker of subordination in the FORCE projection.
Davison (2007a) presents a correlation between the positions of occurrence of
the COMP and the yes/no (polarity) question marker in Indo-Aryan languages.
She demonstrates how the complementizer precedes the yes/no question marker
in sentences with IC, while it follows it in sentences with FC (for details, see
Rizzi 1997; Davison 2007a; and the appendix 1 at the end of this chapter).
Languages that have an FC have a sentence-final polarity question marker
and languages that have an IC have a sentence-initial question marker. Further,
while the yes/no question marker precedes the FC, it follows the IC. Thus, there
is a mirror image relationship in the order of occurrence of these, which is as
follows:
210 Complementation
A. Languages with FC
pol q mkr – FC
B. Languages with IC
IC – pol q mkr
Our analysis of the SALs shows that this correlation holds good in all the SALs
that we worked on.11
Biberauer, Newton and Sheehan (2009) argue that the changes that occurred
in contact situations in South Asian languages systematically obey a univer-
sal labeled the Final-over-Final Constraint (FOFC) proposed in Biberauer,
Holmberg and Roberts (2007, 2008). We shall present a brief discussion of this
in the appendix at the end of this chapter.
6.3.9 The semantic nature of the matrix predicate and the form of the FC
In Manipuri (TB), Gujarati and Marathi (IA), it is the semantic nature of the
matrix predicate that determines the form of the FC that occurs.
Manipuri (TB)
(47) tombə- nə lak- kəni hay-nə ta- y
Tomba- nom come- nonfut comp hear- nonfut
‘I heard that Tomba would come.’ (unplanned coming)
(48) tombə- nə lak- kəni hay-bə ta- y
Tomba- nom come- nonfut comp hear- nonfut
‘I heard that Tomba would come.’ (planned coming)
(Bhat and Ningomba 1995: 436)
Bhat and Ningomba (1995: 437) further point out: “according to [47] Tomba’s
coming might not actually have been programmed, whereas according to [48],
it is considered a settled fact.”
It appears that the complementizer hay-bə ‘say-inf’ occurs with factive
predicates such as khaŋ ‘to know,’ niŋamakta ‘to regret,’ pharakta ‘to become
odd,’ and the complementizer hay-nə ‘say-adv mkr’ occurs with non-factive
predicates such as thajə ‘to believe,’ khan ‘to think,’ khaŋ ‘to guess,’ oythok
6.3 Structural differences – IC and FC 211
‘to be possible’ (used with a negative suffix) (Bhat and Ningomba 1995: 438).
The following examples are illustrative.
w i t h t h e f a c t i v e p r e d i c a t e khaŋ ‘k n o w ’
(49) [cp məhak- nə cak ca-y hay-bəcp ] əy khəŋ- ŋi
he- nom rice eat-nonfut FC I know- nonfut
‘I know that he ate rice.’
(Bhat and Ningomba 1995: 435)
w i t h a v e r b o f c o m m u n i c a t i o n hay ‘t o s a y ’
(50) əy [cp pro cat-kani hay-nəcp ] ha-y
I go-fut say-adv say-nonfut
‘I said that I will go.’
(Bhat and Ningomba 1995: 435)
Sarju Devi (2006) also discusses the differences in the occurrence of the two
complementizers in Manipuri in some detail. However, from the data and
analysis that are provided in Bhat and Ningomba (1995) and Sarju Devi (2006),
it is not clear whether the generalizations that they made would clearly hold.
Further research needs to be done on this issue.
Gujarati (IA)
the f c -em w i t h a no n - f a c t i v e p r e d i c a t e
(53) rāj [[S2]- em] lakhe che
Raj FC writes
‘Raj writes that . . . ’
the f c -te w i t h a f a c t i v e p r e d i c a t e
(54) rāj [[S2]- te] jān.e che
Raj that knows
‘Raj knows that . . . ’
(P. J. Mistry, p.c.)
present. In many SALs and languages across the world the subject of an embed-
ded clause cannot be case-marked by the matrix verb as the complementizer
(IC/FC) blocks such case marking. However, in a few SALs, either with an
IC as in Khasi (Mon-Khmer) or an FC as in Assamese and Bangla (IA), the
embedded subject does get case-marked by the matrix verb.
Khasi (Mon-Khmer)
(55) u- lam u- la- kwa? [DP ya- ka- linDP ]i [CP ba ti ka-
3m,s- Lam 3m,s- pst- want acc- 3 f,s- Lin comp 3 f,s-
n- jɔpCP ]
fut- win
‘Lam wanted Lin to win.’
(Temsen and Subbarao 2004; Temsen 2006)
The question that arises is: how does the embedded subject get case-marked,
when the initial complementizer ba ‘that’ is overtly present? Since in (55) it
does, it is a clear violation of the ‘Tensed-S Condition’ (proposed in Chomsky
1973: 238) and the PIC (the Phase Impenetrability Constraint – Chomsky 2001:
12). Temsen and Subbarao (2004) and Temsen (2006) provide three pieces of
evidence to demonstrate that the embedded subject moves out of S2 and is in
the object position of the matrix clause (see appendix 6.2 on the Website for
further details).
w i t h a n I C je ‘t h a t ’-c l a u s e
Assamese (IA)
(56) mala- yei xopun- ot dekhile [CP je [S2 taii / *nijo- kei ud.i-
Mala- nom dream- in saw that she self- acc fly-
asilS2 ]CP ]
be.pst progr
‘Malai saw in her dream that shei was flying.’
214 Complementation
In contrast, when the FC quotative buli occurs, the subject of the embedded
clause surfaces as nijo- ke ‘self-acc,’ as the matrix subject and embedded subject
become clause-mates. The anaphor gets accusative case-marked by the matrix
verb.
w i t h a n f c buli ‘t h a t ’-c l a u s e
(57) malai - ye nijo- kei xopun- ot [CP ti plεn- ɔr ud.i-thoka buliCP ]
Mala- nom self- acc dream- in plane- in was flying quot
bhabhile
thought
‘Mala thought that she was flying in a plane in her dreams.’
(Subbarao 1984b)
Marathi (IA)
(59) anū-i lā wāt.ta [CP kı̄ [S2 madhūj ne āplyā- lai/∗ j *swatāhā-
Anu (f)- dat feels IC Madhu (m) erg self- acc self-
lā/ āplyā swatāhā- lai/∗ j phasawlaS2 ]CP ]
acc self- acc deceived
‘Anui feels that Madhuj deceived heri /*himselfj .’
(Pandharipande 1997: 210; the glosses have been slightly modified)13
In this case the PIC seems not to have been violated, as there is no movement of
a DP from the embedded clause, but the question remains whether the elements
in the CP phase are ‘visible’ to a higher antecedent in the matrix clause for
establishing antecedent–anaphor relationship. They do not seem to be.
In support of his claim, Yadava (2007: 154) provides evidence which shows that
the matrix verb must agree with the raised/derived subject. Recall that Maithili
has a rich system of verb agreement (see chapter 4 for details).
To summarize the above discussion, the PIC is violated in verb-final lan-
guages with an FC and non-verb-final languages with an IC.
the matrix clause and the embedded clause is required (Masica 1991: 403;
Subbarao 2001).
Languages with an FC
In Sinhala, just as in other SALs, the sentence can be interpreted “as either
a direct or indirect quotation with different meanings” (Gair 2003: 806–807).
The ambiguity arises due to the fact that the quotation is repeated verbatim. The
subscripts ‘-i’ and ‘-j’ refer to the subject of the matrix clause and the speaker,
respectively.
Sinhala (IA)
(62) [(mat.əi,j wad.uwædə kərannə bǣ) kiyəla] nimali kiwwa
I.dat carpentry do.inf can’t quot Nimal say.pst
‘Nimal said “I can’t do carpentry”’ or ‘Nimal said that
I [speaker] can’t do carpentry.’
(Gair 2003: 806–807)
Kannada (DR)
(63) [(nānui,j bombāige hōde) anta] ammai hēl.idaru
I Bombay.dat go.pst.1s quot mother say.pst.3hon
“‘I [= mother] went to Bombay,” mother said.’
‘Mother said that I [= speaker] went to Bombay.’
(Sridhar 1990: 45)
Sema (TB)
(64) atolii nɔ he [(nii,j ye dilli lo wu- ni-) pi]
Atoli nom wrote I nom Delhi to go- fut- quot
‘Atoli wrote that she’d go to Delhi.’/‘Atoli wrote, I [speaker]
would go to Delhi.’
(Achumi and Subbarao 2005)
In Manipuri (TB) (Bhat and Ningomba 1995: 441–442) and Angami (TB)
(Kevichüsa 2007), there is no person harmony in such cases.
Languages with an IC
Though Hindi-Urdu has an IC that is not a quotative, the direct quote is repeated
verbatim and, hence, the sentence in (65) contains the quotation with an IC
preceding it.
6.4 Some crucial issues – FCs and ICs 217
Hindi-Urdu (IA)
(65) rešmāi ne likhā thā [ki (mãı̃i,j dillı̄ jā rahı̄ hũ˜ i,j )]
Reshma erg wrote had comp I Delhi go progr am
‘Reshma wrote that I [= Reshma] would be going to Delhi’ or ‘Reshma
wrote that I [= speaker] would be going to Delhi.’
(67) ka- barii ka- la- thɔ? ba ŋa∗ i,j - n- lεit sha- delhi
3s, f- Bari 3 s, f- pst- write IC 1 s, f- fut- go to- Delhi
‘Bari wrote that I [the speaker, not Bari] will go to Delhi.’
(Temsen 2006)
In the case of direct quotes, the SVO Khasi (Mon-Khmer) differs from the other
SALs, and it is similar to English to the extent that it observes person harmony
but not tense harmony.
The embedded subject I may ambiguously refer either to the speaker of the
utterance or to Raman.
To summarize the above discussion, both IC- and FC-clauses share the
following features:
(i) case marking of the embedded subject by the matrix verb takes place in
Assamese and Bangla with an FC and Khasi with an IC;
(ii) the occurrence of the polymorphemic anaphor in a finite complement
clause is permitted with an antecedent in a higher clause, violating Prin-
ciple A of the Binding Theory;
218 Complementation
6.5.1 FC-clauses
First we consider FC-clauses. Recall that all FC-clauses have the quotative, a
grammaticalized form of the verb to say, as complementizer. In Marathi (IA)
and Gujarati (IA), there is also a different set of FCs derived from a pronoun.
In Marathi it is asə/asa ‘like this’ (literally), and in Gujarati it is em/evũ ‘this
way’ (literally) (P. J. Mistry and Trupti Nisar, p.c.).
The canonical position of the FC quotative clause in SALs is the same as the
position in which the direct object occurs. It may stay in situ in its canonical
(clause-medial) position14 in the matrix clause as in (69), or may occur to the
right of the verb of the matrix clause as in (71), or to the left of the verb of
the matrix clause as in (73), illustrated in (70), (72) and (74), respectively,
from Mizo (TB). Such occurrence in the three positions is found in Dravidian,
Indo-Aryan and Tibeto-Burman – except in Rabha – languages that we have
studied.
fc in situ
(69) [S1 - - - [CP S2 COMP] VP1 S1 ]
Mizo (TB)
(70) zova- n [CP naktuk cu colhni a- ni- aŋ tiʔCP ] a- hria- aŋ
Zova- erg tomorrow def holiday 3 s- be- fut FC 3 s- know- fut
‘Zova knows that tomorrow will be a holiday.’
6.5 Position and Case Resistance Principle 219
Rabha (TB) is the only language that we know of where an FC-clause with a
quotative as the complementizer can either remain in situ, or occur to the left
of the matrix verb, but not to the right, as in (75).
Rabha (TB)
(75) *bharoti (be) kani- bamɯn [CP te u- (be) riba- nɔ n-eCP ]
Bharati nom tell- pst today she- nom come- fut FC
‘Bharati told that she would come today.’
(Subbarao et al., ms)
Khasi (Mon-Khmer)
(76) u- kmεn u- m- ŋεyt [CP ba [S2 ki- brēw ki- hyar patεŋ
3m- Kmen 3m- neg- believe IC p- humans p- descended
na- ki- šrēʔS2 ]CP ]
abl- p- monkey
‘Kmen does not believe that human beings descended from monkeys.’
(Temsen 2006)
220 Complementation
Bangla (IA)
(78) *chele t.a [CP je [S2 or baba aš- beS2 ]CP ] šuneche
boy cl IC his father come- fut heard
Intended meaning: ‘The boy heard that his father will come.’
(Sentence extrapolated from Bayer 2001: 14)
Khasi (Mon-Khmer)
(79) *u- kmεn [CP ba [S2 ki- brēw ki- hyar patεŋ na- ki- šrēʔS2 ]CP ]
3 m- Kmen IC p- humans p- descended abl- p- monkey
u- m- ŋεyt
3 s,m- neg- believe
‘Kmen does not believe that human beings descended from monkeys.’
6.5 Position and Case Resistance Principle 221
Hindi-Urdu (IA)
(83) *[CP ki [S2 ve ā rahe hãı̃S2 ]CP ] ham ne abhı̄ abhı̄ sunā
IC they come progr pres we erg just now heard
‘We just now heard (the news) that they are coming.’
(Subbarao 1974, 1984a: 103)
Bangla (IA)
(84) *[CP je [S2 or baba aš- beS2 ]CP ] chele t.a šuneche
IC his father come fut boy cl heard
Intended meaning: ‘The boy heard that his father will come.’
(Bayer 2001: 14)
We will now consider the case of Khasi (Mon-Khmer) complement clauses.
Khasi, a non-verb-final language, also does not permit a CP clause with the
COMP ba to occur in clause-initial position. Recall that none of the Indo-Aryan
languages which are verb-final with an IC permit this option. In (85) the order
is VSO.
Khasi (Mon-Khmer)
(85) [CP *ba [S2 ka- n- lεyt ka- monS2 ]CP ] u- kmεn
verb subject
IC 3f,s- fut- go 3f, s- Mon 3m,s- Kmen
u- ɔŋ ya- ŋa
3m,s- said acc- I
‘“That Mon will go,” Kmen told me,’ ie., ‘Kmen told me that Mon will go.’
(Temsen 2006)
Change in word order to SV(O) in S2 does not render the sentence grammatical,
as (86) shows.
(86) [CP *ba [S2 ka- mon ka- n- lεytS2 ]CP ] u- kmεn
subject verb
IC 3f,s- Mon 3f,s- fut- go- 3m,s- Kmen
u- ɔŋ ya- ŋa
3m,s- said acc- I
‘“That Mon will go,” Kmen told me,’ i.e., ‘Kmen told me that Mon will go.’
We do not have any explanation as to why Khasi, a non-verb-final language,
does not permit this option of having a complement clause in clause-initial
position.
If the COMP ba is not overtly present, SVO order is permitted, as (87) shows.
Such sentences as (87) are permitted in Hindi-Urdu too, if the IC ki is not overtly
present. Subbarao (1984a: 108) treats sentences parallel to (87) in Hindi as cases
of Sentence-Lifting (Ross 1973).16
In this section we have shown that:
(i) an FC-clause can occur either to the left or right of the matrix clause
(except in Rabha [TB]);
(ii) an IC-clause cannot occur clause-internally in any of the SALs, including
the non-verb-final Khasi language, and is obligatorily moved rightwards;
and
(iii) an IC-clause cannot occur to the left of the matrix clause in any SAL,
including Khasi, which is a non-verb-final language.
We have not addressed the following questions due to lack of substantial data
and evidence from SALs: why are FC-clauses more mobile than IC-clauses,
and why are IC-clauses restricted in their mobility? Can such mobility or non-
mobility be linked to the presence or absence of the dummy NP (NP correlate)?
Bayer (2001) attributes the mobility of the FC-clauses to the absence of the
NP-correlate in the matrix clause. We wish to leave this issue open for further
study.
In the following section we discuss issues related to the scope of wh-
expressions in complement clauses in SALs.
narrow scope
Hindi-Urdu (IA)
(88) anu jānnā cāhti hai ki kis- ne kya kharı̄dā
Anu know.inf want pres that who- erg what bought
‘Anu wants to know who bought what?’
‘*What is such that Anu wants to know who bought it?’
(Dayal 1996: 3)
224 Complementation
Note that the scope of the question word is limited to the embedded clause
alone in (88). In contrast, in (89) the scope of the question word extends to the
matrix clause, and it has a Wide Scope reading.
wide scope
(89) kauni tum soc- te ho [ki ti āyegā]
who you think- imperf pres IC will come
‘Who do you think will come?’ (Wide Scope)
(Dayal 1996: 35)
There are several factors that affect the scope of a question expression. These
factors include the finite/non-finite nature of the embedded clause, the position
in which the embedded clause occurs (sentence-initial, pre-verbal or post-verbal
in the matrix clause), the nature of the COMP (IC or FC), the use of specific
scope markers, the role of intonation, etc. The ten points represent parametric
variation among languages in the expression of Wide and Narrow Scope. These
are:
(i) the finite and non-finite nature of the initial complement (IC-) clause;
(ii) finiteness of the IC-clause vs. post-verbal position in blocking Wide
Scope;
(iii) nature of the matrix predicate and Comp: whether it permits Wide Scope
or blocks it;
(iv) the role of intonation, pause and morphophonemic criteria;
(v) the position of the FC-clause;
(vi) the position of question expressions and Scope possibilities;
(vii) topicalization and Wide Scope;
(viii) the occurrence of clitics/particles as Scope Inhibitors;
(ix) the use of pleonastic question expressions as Scope markers;
(x) the position of occurrence of the complementizer: initial or final.
We shall discuss these ten parameters, providing evidence from SALs.
6.6 Narrow Scope and Wide Scope 225
w i d e s c o p e : w h - e x p r e s s i o n i n a no n - f i n i t e c l a u s e
(92) [tum [kyā karnā] jānte ho]
you what do.inf know pres
‘What do you know to do?’ NOT
‘You know what to do.’
(Dayal 1996: 23)
Wh-expressions in infinitives in Telugu (DR) too have only the Wide Scope
interpretation as in (93), and for Narrow Scope interpretation the finite com-
plementizer -ō is employed, as in (94).
w i d e s c o p e : w h - e x p r e s s i o n i n a no n - f i n i t e c l a u s e
Telugu (DR)
(93) nı̄ku [ēmi ceyy.ad.am] vaccu
you.dat what do.gerund come
‘What do you know to do?’ NOT
‘You know what to do.’
In Bangla (IA), when the embedded clause occurs to the right of the matrix
verb as in (96), it has only Narrow Scope interpretation (see 6.6.10 for further
discussion).
narrow scope
Bangla (IA)
(96) ora šune.che [ke aš- be]
they heard who come- fut
‘They have heard who will come’ (Narrow Scope of ke)
(Bayer 2001: 24)
w i d e s c o p e w i t h t h e n o n - f a c t i v e p r e d i c a t e ‘believe’ w i t h a n f c
Marathi (IA)
(98) mini-la [[lili-ni ravi-la kəy dila] asa] və.ta
Mini-dat Lili-erg Ravi-to what gave comp believes
‘What does Mini believe that Lili gave to Ravi?’ (Wide Scope)
In contrast, the factive verb māhit ‘to know’ in Marathi in (99) does not permit
Wide Scope reading when the FC te occurs.
n a r r o w sc o p e w i t h t h e f a c t i v e p r e d i c a t e ‘know’ w i t h t h e f c te
(99) mini-la [[lili-ni ravi-la kəy dila] te] māhit āhe
Mini-dat Lili-erg Ravi-to what gave that knows
‘*What does Mini know that Lili gave to Ravi?’ (Wide Scope not permitted)
‘Mini knows what Lili gave to Ravi.’ (Narrow Scope)
(Wali 2006: 210; meaning added)
The non-factive verb to think does not permit Narrow Scope reading in Bangla.
Hence, (100) is ungrammatical. According to Bayer (2001: 25), “the nar-
row scope interpretation of ke [‘who’] yields a semantic violation” as “the
matrix verb is incompatible with the [+wh] feature of the embedded clause
[in Bangla].” Bayer argues that ke ‘who’ cannot rise outside the scope of the
matrix clause’s attitude verb.
(100) *tumi bhabcho [ke bar.i kor- be]
you think who house build- will
‘*You think who will build the house?’
(Bayer 2001: 25)
228 Complementation
Factive verbs with an IC ki ‘that’ in Kashmiri (IA) also resist Wide Scope as in
(101).
Kashmiri (IA)
(101) mirāi chi khabar [ki [k’eh por mohanan]]
Mira.dat is news IC what read Mohan.erg
‘Mira knows what Mohan read,’ NOT
‘What does Mira know that Mohan read?’
(Wali 2006: 215)
In contrast, the quotative FC mhan.un in Marathi (IA) in (102) (Wali 2006: 212)
and ani in Telugu (DR) in (103) permit Wide Scope interpretation, even when
the predicate is factive.
w i d e s c o p e w i t h a f a c t i v e p r e d i c a t e w i t h mhan.un ‘having said’
(the quotative), an fc
Marathi (IA)
(102) lili ni kon.ā-lā pustak dila mhan.un tulā du’kkha vat.tā
Lili erg who-dat book gave quot (so) you.dat regret
‘Who do you regret that Lili gave a book to?’
(Wali 2006: 212)
w i d e s c o p e w i t h a f a c t i v e p r e d i c a t e w i t h ani ‘that,’ a n f c
Telugu (DR)
(103) lilli evari- ki pustakam iccindi- ani nuvvu bādhā pad.utunnāvu
Lili who- dat book gave- quot you pain fall
‘Who do you regret that Lili gave a book to?’
w i d e a n d na r r o w sc o p e r e a d i n g
(104) mı̄ru [CP [S2 pārt.ı̄ ki evaru vastāru-S2 ] aniCP ] anukont.unnāru
you party dat who will come- FC are thinking
‘Who do you think will come to the party?’ (Wide Scope)
‘You are thinking who’d be coming to the party’ (Narrow Scope)
In Narrow Scope reading, (i) there is a rise in intonation on the embedded verb
vastāru ‘will come,’ and there is a fall in intonation on the matrix verb; and (ii)
there are two distinct pauses – one after the embedded verb, and the other after
the FC-complement.
In contrast, in Wide Scope interpretation, (i) there is no pause after the embed-
ded verb; and (ii) there is a rise in intonation on the matrix verb. Hence, pitch
considerations also play an important role in interpretation (Peri Bhaskararao,
p.c.).
Thus, we observe that the position of occurrence of the CP-clause affects the
scope of the question word.
According to Mahajan (1987) and Gurtu (1992), the scrambling of the question
word kyā ‘what’ from the embedded clause to the matrix clause also changes
scope possibilities.
kyā ‘what’ o c c u r r i n g i n t h e e m b e d d e d c l a u s e – n a r r o w s c o p e
(110) tum jānte ho [ki us ne kyā kiyā]
you know pres IC he erg what do.pst
‘You know what he did,’ NOT
‘What do you know he did?’
kyā ‘what’ s c r a m b l e d t o t h e m a t r i x c l a u s e – w i d e s c o p e
(111) kyāi tum jānte ho ki us ne ti kiyā
what you know pres that he erg do.pst
‘What do you know that he did?’
(Dayal 1996: 22)
Note that in (109) and (111) the question word is moved one clause up to the
matrix clause. Alice Davison (p.c.) points out that the scope marker kyā in
Hindi-Urdu must be in the next clause up, but the topic in topicalization (see
6.6.7) can be several clauses away from its coindexed pro.
According to her, “topicalization is not subject to the finiteness constraint,”
while scope marking is. The wh-expression kaun in (112a) is in the scope of
the question marker kyā, which occurs in S2 as well as S3 . If it does not occur
in S2 (indicated by ø in S2 ), the sentence is ungrammatical. This shows that the
scope marker must be in the next clause up.
(112a) zayd ne kyā kahā ki mı̄nā kyā / *ø soctı̄ hai ki kaun tum
Zayd erg what said that Mina what think pres that who you
se nārāz hai
with angry is
‘[S1 Who did Zayd say [S2 that Mina thinks [S3 e is angry with you. S3 ] S2 ] S1 ]’
ēmi ‘what’ o c c u r r i n g i n t h e e m b e d d e d c l a u s e – na r r o w s c o p e
Telugu (DR)
(113a) prasād ēmi cēsēd.(u)- ō nı̄ku telusu
Prasad what did- comp you.dat known
‘You know what Prasad did.’ (Narrow Scope reading)
When clefted, the final vowel -i of the embedded verb is elided after it vowel-
harmonizes the complementizer -ō to -ı̄. We have labeled -ı̄ as ‘comp.’17
wide scope
(114) prasād ēmi/elāgu cēsēd.(u)- ō nı̄ku ēmi telusu
Prasad what/how did- comp you.dat what known
‘What/how do you know what Prasad did?’ (Wide Scope reading)
Option II:
The question word ēmi ‘what’ occurs in the matrix clause, and the embedded
clause is clefted with or without the question word, or with the question word to
the right of the cleft, as in (115), or with the question word ēmi ‘what’ followed
by the dubitative marker -ō, as in (116). There is no overt complementizer
in (116).
e m b e d d e d c l a u s e clefted , w i t h t h e q u e s t i o n w o r d i n t h e
matrix clause – wide scope
(115) prasād cēs- in- (a-) di- i nı̄ku ēmi telusu
Prasad do- pst- adjr- 3 nm,pron suffix- comp you.dat what known
‘What do you know what Prasad did?’
6.6 Narrow Scope and Wide Scope 233
Hindi-Urdu (IA)
(117) [tum- ne kisei kahā [ki ei / wo kal āyegā?]]
you- erg who.dat say.perf that 3p,s tomorrow come.fut
‘Whoi did you say [ei will come tomorrow?]’
Davison (1988: 183)
Thus, the direct object evari-ni ‘who-acc’ has matrix scope in (118) due to the
presence of the scope marker ēmi ‘what’ in the preverbal position in the matrix
clause. The occurrence of evari-ni ‘who-acc’ as scope marker in the matrix
clause is not permitted in (118).
To recapitulate, we have shown that there are differences between Scope
Marking and Topicalization: (i) Scope Marking can be done only one sen-
tence higher up, whereas there is no such restriction in Topicalization; and
(ii) in Topicalization, the topicalized NP is an argument subcategorized by the
matrix verb, while the in situ wh-element originates in the matrix clause and
subsequently moves to the COMP position (Davison 1988).
The Wide Scope reading is blocked, and the sentence only has Narrow Scope
reading, when the yes/no question marker clitic -ā or the dubitative marker -ō
occurs to the right of the embedded clause, as in (120).
Our claim is: it is due to the presence of these clitics that Wide Scope reading
is blocked in (120). Hence, we may label them as Scope Inhibitors.
In Sema (TB) too, Narrow Scope results from: (i) the presence in the embed-
ded sentence of the dubitative marker keno as in (121), contrasted with (122);
and (ii) the presence in the embedded sentence of the yes/no question marker
as in (123), contrasted with (124) (Achumi 2000).
6.6 Narrow Scope and Wide Scope 235
n a r r o w sc o p e w i t h t h e d u b i t a t i v e m a r k e r keno
Sema (TB)
(121) no- ye [khū hile iγe- ni keno pi] kms - a
you nom who here come- fut dub mkr FC think- progr
‘You were wondering who would come here.’ (Narrow Scope)
w i d e s c o p e w i t h o u t t h e d u b i t a t i v e m a r k e r keno
(122) no- ye [khū hile iγe- ni pi] kms- a
you- nom who here come- fut FC think- progr
‘Who do you think will come here?’ (Wide Scope)
n a r r o w s c o p e w i t h t h e yes/no q u e s t i o n m a r k e r kya
(123) panoŋu- ye [[khū iγe- ni kya] pi] celu
they- nm who come- fut pol q mkr FC heard
‘They heard who will come.’ (Narrow Scope)
w i d e s c o p e w i t h o u t t h e yes/no q u e s t i o n m a r k e r kyā
(124) panoŋu- ye [[khū iγe- ni] pi] celu
they- nm who come- fut FC heard
‘Who did they hear will come?’ (Wide Scope)
(Achumi 2000)
Sema lends further support to: (i) treating the yes/no question marker and the
dubitative marker (as in Telugu [DR]) as Scope Limiting Particles; and (ii) the
proposal that certain clitics act as Scope Inhibitors.
Kashmiri (IA)
(125) mohan- as k’ah chu bāsān [CP (ki) [S2 kamis nish chu mirā
Mohan- dat what aux believes IC whom near pres Mira
bihitS2 ]CP ]
sit
‘Who does Mohan believe Mira is sitting near?’
(Adapted from Wali 2006: 217 in consultation with O. N. Koul, p.c.)
When the question word ēmi ‘what’ in Telugu (DR) or kyā ‘what’ in Hindi-Urdu
(IA) occurs in the embedded clause, the sentence has Narrow Scope reading,
as in (126) in Telugu and (127) in Hindi-Urdu (IA).
Hindi-Urdu (IA)
(127) tum jānte ho [S2 ki usne kyā kiyāS2 ]
you know pres that he.erg what did
‘You know what he did.’
Srivastav (1989: 2)
The question words ēmi ‘what’ in Telugu (DR) and kyā ‘what’ in Hindi-Urdu
(IA) may also be used as Scope Changing devices. That is, their addition will
lead to a change from Narrow to Wide Scope. Srivastav (1989) labels such
markers as pleonastic operators. The occurrence of ēmi ‘what’ in Telugu (DR)
or kyā ‘what’ in Hindi-Urdu (IA) in the matrix clause, together with the question
expression in the embedded clause, imparts Wide Scope reading.
Telugu has the question word ēmi that is homophonous with the scope marker
ēmi ‘what,’ which is used to indicate Wide Scope. We label the question word
as ēmi1 and the scope marker as ēmi2 for clarity of exposition. Note that, just as
with k’ah in Kashmiri (IA), ēmi2 in (128) in Telugu has no semantic content.
w i d e s c o p e w i t h a n a d d i t i o n a l question word ( w h - e x p r e s s i o n )
a s an expletive
Telugu (DR)
(128) nı̄ku [S2 nası̄m ēmi1 and(i)- ōS2 ] ēmi2 telusu
you.dat Nasim what said- dub mkr FC what (expletive) known
‘What do you know what Naseem said?’ (Wide Scope)
6.6 Narrow Scope and Wide Scope 237
Hindi-Urdu (IA)
(129) tum kyā jānte ho [S2 ki usne kyā kiyāS2 ]
you what (expletive) know pres that he.erg what did
‘What do you know what he did?’ (Wide Scope)
Srivastav (1989: 2)
The scope markers ēmi2 ‘what’ in Telugu and kyā ‘what’ in Hindi-Urdu are
different from the interrogative markers. Evidence for this comes from the fact
that the interrogative word ēmi ‘what’ in (126) can be reduplicated – just as any
question word can be in Telugu and in other SALs – as in (130), but the scope
marker ēmi in (128) cannot be reduplicated as in (131).
s c o p e m a r k e r r e d u p l i c a t e d : no t p e r m i t t e d
(131) *nı̄ku [S2 nası̄m ēmi1 and(i)- ōS2 ] ēmi2 ēmi2 telusu
you.dat Nasim what said- dub mkr what what known
Similar evidence can be adduced from Hindi-Urdu too with the reduplicated
question word kyā ‘what.’
In Sema (TB) too, a question expression functions as a scope marker (see
Achumi 2000, and Achumi and Subbarao 2005).
Bangla (IA)
(132) ora [ke aš- be] šune.che WS,NS
they who come- fut heard
(i) ‘They have heard who will come.’ (Narrow Scope of ke)
(ii) ‘Who have they heard will come?’ (Wide Scope of ke)
The question that needs to be answered is: how do we explain the ambiguity in
(132) which has Narrow Scope as well as Wide Scope interpretations?
Before we proceed further, it would be helpful if the facts in Bangla (IA) can
be summarized as follows.18
238 Complementation
The question that arises now is: how to account for the Narrow Scope inter-
pretation? Probal Dasgupta (p.c.) suggests that we hypothesize that in (134)
we have the IC as a zero complementizer (ZERO COMP) and a ZERO NP
correlate too (see Dasgupta 2007 for details).20
ICs in many Indo-Aryan languages do permit an NP correlate, such as yah
‘it’ or yah bāt ‘this news’ in Hindi-Urdu (IA) (Subbarao 1984a), or he ‘this’ or
hi gos..ta ‘this story’ with factive predicates in Marathi (Wali 2006: 211).
Taking this as a cue, we can say that the zero form in (134) corresponds to
(135), the non-zero form.
It is the occurrence of šet.a ‘DEM’ that blocks Wide Scope reading in (135).
Thus, Dasgupta’s suggestion of having zero forms enables us to account for
the ambiguity in (132) in a neat fashion. We observe that the final COMP bole
permits Wide Scope reading in (133), while initial COMP je permits Narrow
Scope reading as in (135).
Bayer (2001: 29) provides another explanation in terms of harmony of direc-
tionality. He provides an explanation as to why an in situ IC-clause in (134)
does not permit Wide Scope reading. For the Wide Scope reading of the ques-
tion word, the question expression must move overtly or covertly to the matrix
clause.
According to Bayer (2001: 29), “covert wh-rising from the IC-clause to the
matrix clause is blocked by the fact that the IC-clause, although theta-marked,
falls outside the pattern of argument licensing in head-final languages.” Note
that verb-final languages have an FC. Hence, the position of occurrence of the IC
je- in Bangla, Assamese or Oriya, or ki- in Hindi-Urdu or Punjabi and Kashmiri,
6.7 Against Rightward Extraposition 239
is not ‘consistent’ from the point of view of word order universals. They are
‘exceptionally’ selected as they are not in “harmony with the directionality
pattern that is dictated by the parameter of the language” (Bayer 2001: 29). In
contrast, FC-clauses permit Wide Scope reading as “the FC-clause is selected
according to the parameter of head-finality” (Bayer 2001: 29). Hence, if there
is no other variable that blocks the movement of the wh-expression from the
FC-clause, an FC-clause does permit Wide Scope reading.21
In this section we discussed the interplay of various factors that affect Narrow
and Wide Scope. The presence or absence of the complementizer, for example,
might lead to Wide or Narrow Scope interpretations, as in Bangla. An IC-clause
that occurs to the right of the matrix clause does not permit Wide Scope reading.
Some other factors include: the nature of the matrix verb; the nature of the
complementizer; the position of occurrence of the scope marker; the presence
of clitics; clitics functioning as scope inhibitors; and question expressions as
scope markers. We distinguished between scope marking and topicalization as
scope marking strategies. Finally, we conclude by saying that it is the interplay
of various strategies that yields scope readings.
Finally, as Alice Davison (p.c.) puts it, a typological generalization that
emerges is: “in all the various scope extending methods there are strict con-
straints on wh scope in all SALs, according to finiteness or position. Languages
find different syntactic, lexical and phonological devices for overcoming these
constraints.”
6.8 Conclusion
In this chapter we discussed the differences between the ICs and FCs; direct
and indirect speech and tense and person harmony; the Case Resistance Princi-
ple and the non-occurrence of a CP clause intra-sententially; scope of question
expressions in complements; arguments against rightward extraposition; direc-
tion of c-command and the NPI; the Tensed S-Condition and the PIC in relation
to the accusative case marking of the embedded subject; COMP agreement; and
the reanalysis of the IC and FC in language contact situations. We discussed
wh-scope in finite and non-finite COMP-clauses, and the factors that permit
and restrict wh-scope. We have shown that there is a variety of syntactic, mor-
phophonemic and phonological criteria that affect Narrow and Wide Scope. We
demonstrated that it is the interplay of a variety of factors that yields Narrow
and Wide Scope readings.
We have shown that the IC ki, a functional category, of the source language
Hindi-Urdu (IA) is reanalyzed as an FC in Dakkhini. It occurs as an FC in
embedded questions, as a clausal and phrasal disjunctive marker, as a linker in
relative clauses, etc.
Appendix
The Final-over-Final Constraint (FOFC) and diachronic change
In this appendix we present a universal constraint labeled as the Final-over-
Final Constraint (FOFC), discussed recently in Biberauer et al. (2009),22 and
discuss how the diachronic changes that took place in the position of occurrence
of the complementizer in South Asian language contact situations strictly obey
this universal constraint.
Appendix 241
Table 6.1
The next issue that is of interest is the order in which the complementizer and
the polarity question word occur in complement constructions.
Davison (2007a) also makes an interesting observation in this regard.
In languages in which the complementizer occurs in final position, the polar-
ity question marker precedes it, and in languages in which the complementizer
occurs initially, the polarity question marker follows the initial complementizer.
Thus, the order of the polarity question marker and the complementizer is
shown in Table 6.1.
The following examples from Marathi (IA) are illustrative:
Marathi (IA)
(5) [[to kal parat ālā kā(y) mhan.un/ asa] rām malā
he yesterday back come.pst.3m,s pol quot/ such Ram I.dat
witšārat hotā]
ask.progr be.pst.3m,s
‘Ram was asking me whether/if he came back yesterday.’
(Davison 2007a: 184)
Appendix 243
In Hindi-Urdu (IA), for example, the Polarity Question marker follows the IC
initially in the unmarked word order, as in (8). It may occur in the final position
of the embedded sentence too.
Hindi-Urdu (IA)
(8) us- ne pūchā [ki [kyā tum ā- ōgē]]
3s- erg ask.perf that pol you come- fut
‘He asked whether you would come.’
If the Polarity Question marker precedes the IC, the sentence is ungrammatical.
Thus, there are four types (Types A–D) that are possible with regard to the
order of occurrence of the complementizer and the Polarity Question marker.
Davison (2007: 183), based on data from Indo-Aryan languages, demonstrates
that there is no language in which the initial Polarity Question marker and
final complementizer co-occur – that is, Type D is not attested. The other three
combinations are permissible, as illustrated in Table 6.2.
We shall now present the universal constraint labeled as the Final-over-
Final Constraint (FOFC) discussed in Biberauer et al. (2009), and discuss how
the diachronic changes that took place in the position of occurrence of the
complementizer in South Asian language contact situations strictly obey this
universal constraint.25
244 Complementation
Table 6.2
Source: Adapted from Davison (2007a: appendix B) in Biberauer et al. (2009: 14).
t h e f i n a l - o v e r - fi n a l c o n s t r a i n t (f o f c )
Biberauer et al. (2009: 14) demonstrate that the FOFC can explicate diachronic changes
too. They observe: “There is good evidence that FOFC constrains the borrowing of final
complementisers in Indo-Aryan. While Indic languages with final Pol heads [Polarity
heads] were free to borrow/develop final complementisers, the languages with initial
Pol [Polarity] were not as the borrowing of a final C would have led to a FOFC violation.
Further evidence that this is what is at stake in Indic comes from typological trends,
which appear to behave in the same way.”
The type that FOFC predicts to be ungrammatical is in fact the type that is
precisely unattested. Table 6.2, adapted from appendix B of Davison (2007a)
in Biberauer et al. (2009: 14), is illustrative.
Note that types A, B and C are attested in Indo-Aryan while Type D is not,
as predicted by FOFC.
Dravidian and Tibeto-Burman languages too adhere to the FOFC discussed
above. The following examples are illustrative.
In Telugu (DR) and Thadou (TB) the Complementizer, a quotative, follows
the Polarity Question expression.
Appendix 245
Telugu (DR)
(11) pel.l.i- ki lalita vastund(i)- ā ani mālati nannu ad.igindi
marriage- to Lalita will come- pol q m quot Malati I.acc asked
‘Malati asked me whether Lalita would come to the wedding.’
Thadou (TB)
(12) [[kicenna- a na- huŋ- diŋ- ham]- tin] thaŋboi- in lalboi
marriage- loc 2s- come- fut- pol q m- comp Thangboi- erg Lalboi
a- tiem- me
3s- asked/invited- fut
‘Thangboi asked whether Lalboi would come to the wedding.’
(Pauthang Haokip, p.c.)
7.1 Introduction
This chapter brings to the attention of scholars working in the area of language
typology and universals how widespread the phenomenon of Backward Control
is in the languages of the South Asian subcontinent and the explanatory power
it has to explicate a variety of syntactic phenomena. We shall demonstrate that
Backward Control is a parametric choice that Universal Grammar (UG) permits
as an alternate structure or as the only structure in some specific cases in some
SALs.
We shall demonstrate that the finite or non-finite nature of the embedded
verb with case-marking properties plays a crucial role in permitting Backward
Control. On the basis of data from control structures and the conjunctive par-
ticipial construction, we demonstrate that it is the nature of the embedded verb,
and consequently the assignment/checking of either structural nominative case
or inherent case of the embedded subject that permits Backward Control. We
shall demonstrate that many case alternations found in SALs, and ‘creative
errors’ made by non-Hindi-Urdu speakers while speaking Hindi-Urdu (IA),
can be explained by invoking the phenomenon of Backward Control. We shall
further show the effects of Backward Control in language contact situations
in Dakkhini Hindi-Urdu, Bhalavali Bhasha spoken in Mangalore, and Bangla
(IA) spoken in Assam. We shall provide evidence from Subzapuri, a dialect of
Hindi-Urdu, which exhibits Backward Control and discuss the implications of
this for the language–dialect dichotomy.
This chapter is organized as follows: section 7.2 introduces the reader to
the notion of Backward Control while comparing it with Forward Control
which is generally attested in many languages. Section 7.3 presents evidence
in support of the phenomenon of Backward Control in Mizo (TB) and Telugu
(DR). Section 7.4 briefly discusses Polinsky and Potsdam’s (2003) approach
to explicate Backward Control in terms of the movement theory of control
This chapter is a revised version of the plenary lecture delivered at the South Asian Languages
Analysis (SALA) conference, University of Illinois, 2005.
246
7.2 Forward and Backward Control 247
proposed in Hornstein (1999), and section 7.5 concludes the chapter. In the
appendix at the end of the chapter we discuss the Case-marked nature of PRO.
The reader is further referred to appendixes 7.1–7.5 on the Website. The
issue discussed in each appendix is mentioned here.
Appendix 7.1 discusses case alternations that occur in constructions involv-
ing matrix verbs such as ‘to say,’ ‘to tell,’ ‘to mention,’ ‘to send a message,’
etc., in SALs, in terms of the phenomenon of Backward Control, and attempts
to explain them in terms of control theory invoking the phenomenon of Back-
ward Control. Appendix 7.2 discusses time expressions and Backward Control.
Appendix 7.3 discusses cases involving Backward Control in language contract
situations. Appendix 7.4 demonstrates how a syntactic phenomenon such as
Backward Control can be used as a heuristic tool to show whether a language
under consideration should be treated as a dialect or language, Appendix 7.5
discusses cases involving Backward Control in some SALs.
forward control
(1a) [S1 Johni wanted [S2 PROi to kill a mosquitoS2 ]S1 ]
[−finite]
248 Backward Control
Schematically:
backward control
(1b) [S1 øi want [S2 Johni kill a mosquitoS2 ]S1 ]
[+/−finite] [+/−finite]
Schematically:
t.šu t.šut
sit sit
[+fin] [−fin]
ziak ziah
write write
[+fin] [−fin]
Mizo (TB)
(2b) ∀ [zova- ø t. šutleŋ- ah a- .tšu] ø- duh
Zova- nom bench- on 3s- sit [+fin] 3s- want
‘Zova wants to sit on the bench.’
(Subbarao and Lalitha Murthy, ms)
Note that in (2b) the matrix verb duh ‘want’ does not carry any subject agree-
ment marker (indicated by ø), as the matrix subject is not overtly present. In
contrast, it is the embedded verb that carries the 3rd person singular agreement
marker a- (in italics) in (2b). The crucial point to which we wish to draw the
attention of the reader is that it is the embedded verb in Backward Control in
Mizo that is [+finite] carrying the agreement marker, and the matrix verb does
not carry agreement features.
With the matrix verb tum ‘to try,’ Forward Control is permitted (3a) for all
speakers, and Backward Control is permitted for most of the speakers (3b) –
though, for a few, it is not. The verb for to write in English has two forms in
Mizo – one for the finite (ziak ‘write’) and the other for the non-finite (ziah ‘to
write’) – just as the verb sit has.
Let us now go back to (2a) and (2b). Evidence for our claim that the DP zova
‘Zova’ in (2b) occurs as the embedded subject comes from the fact that the
embedded subject does not carry the ergative case marker in such sentences,
as the embedded verb .tšut ‘to sit’ is [−transitive]. Note that the matrix verb is
duh ‘want’ and it is [+transitive], and hence requires the subject of the matrix
verb to carry the ergative case marker -n. The occurrence of the ergative case
7.3 Evidence for Backward Control 251
marker with zova ‘Zova’ in (2a) illustrates the point. In (2b), the DP zova
‘Zova’ does not carry any ergative marker, as it is not the matrix subject and it
is the embedded subject that overtly occurs. Thus, (2b) is a case of Backward
Control in which only the embedded subject is overtly present and the matrix
subject is absent. We now provide evidence to demonstrate that zova ‘Zova’ in
(2b) is the subject of the embedded clause, and not that of the matrix clause.
forward control
(4) [S1 zovii - n [S2 PRO chiarS2 ] ai - duhS1 ]
Zovi- erg study [−fin] 3s- want
‘Zovi wants to study/read [something].’
b a c k w a r d c o n t r o l : e m b e d d e d v e r b i s [ +t r a n s i t i v e ] a n d [+fin ],
a n d he n c e e m b e d d e d s u b j e c t i s e r g a t i v e c a s e - m a r k e d
(5a) [S1 ∀ i [S2 zovii - n ai - chiarS2 ] ø- duhS1 ]
Zovi- erg 3s- study [+fin] 3s- want [−fin]
‘Zovi wants to study/read [something].’
The ungrammaticality of (5b) is due to the fact that the subject agreement
marker in Mizo can occur on the verb if, and only if, the subject is overtly
present.
Since the DP zovi ‘Zovi’ is the subject of the embedded verb, which is
[+finite], it has to be obligatorily case-marked ergative, or else the sentence is
ungrammatical as in (6).
252 Backward Control
forward control
(8) lali- n [S2 PRO lala ø- thahS2 ] a- duh
Lali- erg Lala 3s- kill [−fin] 3s- want
‘Lali wants to kill Lala.’
The embedded subject PRO in (8) is not overtly present, and hence the embed-
ded verb does not carry 3rd person agreement marker -a. Its absence is denoted
by ø in (8). The form of the verb is thah, which is [−finite], and the matrix
verb carries the 3rd person agreement marker a- as the matrix subject is overtly
present.
In contrast to (8), in (9) the embedded subject is overtly present and the
matrix subject is not, and hence:
(i) the embedded subject carries the ergative case marker -n as the embedded
verb is [+transitive];
(ii) the embedded verb carries the 3rd person agreement marker a- and the
form of the verb is that which is [+finite]; and
(iii) the matrix verb does not carry the 3rd person agreement marker (denoted
by ø in (9)) as the matrix subject is not overtly present.
backward control
(9) [S1 ∀ i [S2 lali- n lala a- thatS2 ] ø- duhS1 ]
Lali- erg Lala 3s- kill [+fin] 3s- want
‘Lali wants to kill Lala.’
Telugu (DR)
(10) [S1 raman.ai [S2 PROi kōpam vacc- iS2 ] int.i- ki vel.l.i pōyēd.uS1 ]
Ramana (nom) (dat) anger come- cpm home- to left
‘Having become angry, Ramana left for home.’
Note that, in (10), the embedded predicate kōpam vacc ‘anger come’ takes a
dative subject, and PRO occurs in the embedded subject position. An example
of Backward Control in Telugu is:
(11) [S1 [S2 raman.ai ki kōpam vacc- iS2 ] ∀ i int.i- ki vel.l.i pōyēd.uS1 ]
Ramana (nom) dat anger come- cpm home- to left
‘Having become angry, Ramana left for home.’
We have used the symbol ∀ to indicate the absence of the matrix subject
coindexed with the embedded dative subject.
254 Backward Control
Let us now look at the dative subject construction. The patient/theme in the
dative subject construction in (13) carries the locative marker mı̄da ‘on’ (see
chapter 5 for details).
When the subject is dative case-marked with ki, it is the dative case marker ki
that is copied onto the second part of the anaphor, and the first part carries the
locative marker mı̄da ‘on’ (Subbarao and Lalitha Murthy 2001). The occurrence
of the verbal reflexive kon in NNS constructions is prohibited (15), as the
predicate is [−transitive] and the occurrence of the reduplicated form of the
nominal anaphor is obligatory in the dative subject construction (14) (Subbarao
and Bhaskararao 2004).
(15) *kamala kii tana-mı̄da-tana kii eppud.ū kōpam vacc- kon- t.undi
Kamala dat self-on-self dat always anger come- VR- keeps
An anaphor that occurs in the embedded clause with the dative subject mādhuri
‘Madhuri’ as its antecedent must carry the dative case marker ki if our claim is
correct. Sentence (17) shows that it does. We’ve coindexed the two occurrences
of the dative case marker ki to show that the second occurrence is a copy of the
subject dative case marker.
(17) [S1 [S2 mādhuri kii tana-mı̄da-tana kii kōpam vacc- iS2 ] ∀ ippud.ē
Madhuri dat self-on-self dat anger come cpm just now
pad.ukondi S1 ]
slept
‘Having gotten angry at herself, Madhuri slept just now.’
Recall that a dative predicate does not permit a verbal reflexive to occur with the
verb as ‘dative predicates’ are [−transitive] (see (15) above, and also chapter 3).
If the dative subject were the subject of the matrix clause, the verbal anaphor
should not occur with the matrix verb. The fact that verbal anaphor does occur
in (18) and (19) proves that the dative subject is the subject of the embedded
clause, and not that of the matrix clause. An empty element (∀ ) is the subject
of the matrix clause.
(18) [S1 [S2 raman.a ki picci ekk- iS2 ] ∀ tana- ni- tanu tit.t.u kon-
Ramana dat mad go- cpm self- acc- self abuse VR-
t.unnād.u S1 ]
be.progr.3 m
‘Having gone crazy, Ramana is abusing himself.’
(19) [S1 [S2 salmā ki santōs.am vacc- iS2 ] ∀ tana- ni- tanu pogud.u-
Salma dat happiness come- cpm self- acc- self praise-
kon- t.un-diS1 ]
VR- progr.3 s,f
‘Having felt happy, Salma is praising herself.’
If the dative subject were the subject of the matrix clause, a verbal anaphor
would not be permitted.
Sentences (20a) and (20b) are ungrammatical as the verbal reflexive is not
present. Recall that the occurrence of the verbal reflexive is obligatory, when a
256 Backward Control
Thus, the ungrammaticality of (20a) and (20b) is a conclusive proof that the
dative case-marked DPs raman.a ki ‘Ramana to’ in (18) and salmā ki ‘Salma
to’ in (19) belong to the lower sentence and the subject position of the matrix
clause is null.
(22) [S1 [S2 madhu ki padi ēl.l.a nunci jabbu- gā und.- iS2 ] ∀ ippud.ē
Madhu dat ten years from sick- adjr be- cpm just now
pōyēd.uS1 ]
died
‘Having been ill for ten years, Madhu died just now.’
Neither of the matrix verbs vel..l ‘to go’ in (21) nor pō ‘to die’ in (22) permits
an adverb of duration, and hence it is the embedded verb that is in the scope
of the duration adverb, and the dative case-marked subject occurs to its left in
the embedded clause as its subject. This shows that the dative subject is in its
in situ position as the subject of the embedded clause.
(2003) observe that the Principles and Parameters approach (P and P approach:
Chomsky and Lasnik 1995) is not adequate to handle the cases of Backward
Control as the empty element in the matrix clause in such cases “is not bound,”
as in (23) (Polinsky and Potsdam 2003: 10). They mark the empty element as
PRO and we marked it as ∀ in our discussion so far.
Malagasy
(23) manomboka [mitondra ny fiara Rabei ] PROi
begin drive the car Rabe
‘Rabe is beginning to drive the car.’
The authors make two observations: (i) if PRO is not bound, it gets the arbitrary
PRO interpretation, which is unavailable in Malagasy; it is also the same case in
SALs; (ii) the configuration in example (23) from Malagasy “violates Condition
C of the Binding Theory, which requires that R-expressions like Rabe be
free” (Polinsky and Potsdam 2003: 10). Though the DP in (23) is not free, the
sentence in Malagasy is grammatical, they point out. They conclude: “Since the
arguments against the PRO analysis are essentially independent of the details of
Malagasy syntax, we conclude that P and P quite generally rules out Backward
Control” (Polinsky and Potsdam 2003: 10). They propose an analysis in terms
of Hornstein’s (1999) Movement analysis. We agree with their conclusion
concerning the P and P approach’s failure to account for Backward Control.
However, as we have shown elsewhere (Subbarao 2003), Hornstein’s (1999)
Movement analysis – as well as Manzini and Roussou’s (2000) analysis –
encounters problems in handling the data from SALs. Hence, we leave the
issue open for further investigation.
A significant point made by Frans Plank (p.c.) is that the available evidence
shows that Backward Control is possible only in verb-peripheral languages
(SOV and VSO) and not in verb-medial languages. It is not clear what the
implications of this are for Universal Grammar (UG).
7.5 Conclusion
To conclude, Dravidian languages such as Telugu, Kannada and Malayalam
and Indo-Aryan languages such as Assamese, Nepali, Eastern Bangla and
Subzapuri that have a tensed conjunctive participle permit Backward Control,
while standard Bangla and Kashmiri do not, though they do have a tensed
conjunctive participle. Dakkhini Hindi-Urdu, which has been in contact with
Dravidian languages, permits Backward Control, though it does not have a
tensed conjunctive participle. We have shown that allowing Backward Control is
a parametric choice that Universal Grammar permits, and while some languages
may exploit it, some may not. We have also shown that, in some cases, Backward
258 Backward Control
Nepali Indo-Aryan
Assamese Indo-Aryan
Eastern Bangla Indo-Aryan
Oriya Indo-Aryan
Gujarati Indo-Aryan
Bhalavali Bhasha Indo-Aryan
Dakkhini (Hindi-Urdu) Indo-Aryan
Swat-Dir Kohistani Indo-Aryan
Torwali Indo-Aryan
Kannada Dravidian
Malayalam Dravidian
Telugu Dravidian
Ladakhi Tibeto-Burman
Mao Naga Tibeto-Burman
Mizo Tibeto-Burman
Control is the only option permitted in a language in some structures, and thus
it may not necessarily be a marked construction, as Table 7.4 illustrates.
7.6 Postscript2
Backward Control is permitted in Mangalore Konkani (IA) while it is not in
Standard Konkani (IA). Note that, in (1) below, the subject of the embedded
clause ammi ‘we’ is nominative case-marked as the embedded predicate yev
‘to come’ does not take a non-nominative (dative) subject.
backward control
Mangalore Konkani (IA)
(1) [S1 [S2 ammii haŋga yev- nuS2 ] ∀i dha vərsəj dzallı̃jS1 ]
we here come- cpm ten years happened
Intended meaning: ‘It is ten years since we came here.’
(Lalita Dhareshwar, p.c.)
f o r w a r d c o n t r o l st r u c t u r e w i t h a p e r f e c t p a r t i c i p l e
(3) [S1 [S2 PROi haŋga ai- l εS2 ] am-kãi dha vərsəj dzallı̃j S1 ]
here come- perf pple we.dat ten years happened
‘It is ten years since we came here.’
Telugu (DR)
(1) rādha [PRO kōpam vacc- i] vel.l.- i pōyindi
Radha dat anger come- cpm go- cpm went
‘Having felt angry, Radha left.’
Nepali (IA)
(2) [PRO pāni cāhi- era] u nadi tira ga- yo
dat water need- cpm he river toward go- pst
‘Needing water, he went to the river.’
(Ichihashi-Nakayama 1994, as quoted in
Bickel 2004: 81)
Hindi-Urdu (IA)
(3) *rādhā [PRO gussā ā kar] bāhar cal- ı̄ ga- ı̄
Radha dat anger come cpm outside walk- pst go- pst
Intended meaning: ‘Having got angry, Radha went outside.’
In such cases, only a predicate such as nārāz honā ‘to be/become angry’ that
takes a nominative subject is permitted, as in (4).
Note that genitive case is inherently assigned by the predicate to the subject
in Manipuri, and hence the occurrence of PRO in (6) cannot escape being
case-marked genitive.
An experiencer subject is dative/locative case-marked (7).
8.1 Introduction
‘A relative clause is a clause that modifies a phrasal constituent, generally a
noun phrase. We call the noun phrase that is so modified the head of the relative
clause’ (Riemsdijk 2006: 338).1 A relative clause is a subordinate clause in
which the embedded predicate may be [+finite] or [−finite]. Such a non-finite
predicate may be participial or infinitival depending upon the language family
that the specific language under consideration belongs to.
In this chapter, we discuss three types of relative clauses2 that are found in
the South Asian subcontinent. They include:
(i) externally headed relative clauses (EHRCs);
(ii) Relative-Correlative clauses; and
(iii) internally headed relative clauses (IHRCs)
We shall briefly elaborate the above three types:
(i) Externally headed relative clauses (EHRCs)
EHRCs are of two subtypes:
(a) the sentential-type; and
(b) the non-finite type.
(a) The sentential-type are post-nominal relative clauses with a finite
embedded verb, as in Indo-Aryan. They do not have a null
operator.
(b) The non-finite type may either be pre-nominal or post-nominal.
They both have a ‘null (covert) operator’ in the embedded rela-
tive clause. They may either be participial as in Dravidian, Indo-
Aryan, Mon-Khmer and Munda, or infinitival (nominalized) as in
Tibeto-Burman languages.
In fact it is the absence or presence of the covert operator that distinguishes
the sentential-type from the non-finite type, in addition to other distinctive
features such as:
• the [+finite] nature of the embedded verb in the sentential-type and the
[−finite] nature of the embedded verb in the non-finite type; and
263
264 Noun modification: relative clauses
Pattern 1: EHRC
These consist of either the post-nominal sentential relative clause or the non-
finite (infinitival or participial) relative clause.
Recall that, according to our classification, Pattern 1 has the post-
nominal relative clause and the EHRC. Let us label them as Subpattern 1
and Subpattern 2, respectively.
In Subpattern 1, the embedded relative clause may also occur to the right of the
matrix VP, in which case the relative clause is away from the head and is not
adjacent, as in (2). The embedded relative clause is thus considered extraposed
and is right-adjoined to CP.
The right-adjoined relative clause together with the head in (1) forms a DP
constituent, while the CP-adjoined or extraposed relative in (2) does not form
a DP constituent, as it is away from its head.
e x t e r n a l l y h e a d e d r e l a t i v e c l a u s e (e h r c )
d i r e c t o b j e c t m o d i fi e d
Hindi-Urdu (IA)
(3) [S2 pūjā ke kal ke øi kharı̄de hueS2 ]
Pooja gen (m,p) yesterday gen (m,p) bought (perf pple, m, p)
kapr.ei bahut khūbsūrat the
clothes (m, p) very pretty were (m, p)
‘The clothes that Pooja bought yesterday were very beautiful.’
subject modified
Tenyidie (TB)
e h r c : t h e he a d khriesa ‘y o u n g m a n ’ o c c u r s i n t h e m a t r i x c l a u s e
(4) [S2 øi bulie kemerie se- ke- baS2 ] khriesai - u- e a-zemia
shirt red wear- nozr- progr young man- def- nom my-friend
‘The boy who is wearing a red shirt is my friend.’
(Kevichüsa and Subbarao 1998: 45)
Multi-Head correlative
Multi-Head relative clauses as in (6) are invariably the relative-correlative type
and not extraposed. They generally have the interpretation of a relative clause
where the head is [−definite].
multi-head correlative
(6) [jis- nei joi kar-nā cāhā]i,j [us- nei voj ki-yā]
who- erg what do-gerund/ want.perf he/she- erg that do.perf
For x, y s.t. x wanted to do y, x did y.
Literally: ‘Who wanted to do what, he/she did that.’
(Bhatt 2003: 486)
Multi-Head correlatives are found in all the Indo-Aryan languages (Bhatt 2003:
486) and also in Dravidian Languages.
IHRC TYPE B
i n d i r e c t o b j e c t m o d i f i e d – i o a s he a d o f a n i h r c w i t h no o v e r t
postposition
Hmar (TB)
(8a) [lala- n bɔŋi (IO) bu a- pek] kha øi zanikhan a- thi
Lala- erg cow food 3s- give det pro yesterday 3s- died
‘The cow which Lala gave food to died yesterday.’
(Kumar 1998)
Note that the dative postposition hnenaʔ ‘to, for’ with the internal head bɔŋ
‘cow’ is not present in (8a), while in a simple sentence (8b) its presence is
obligatory.
The examples provided above show the various types of relative clauses
found in SALs. We shall now discuss the syntax of relative clauses in SALs.
It is in Indo-Aryan languages that relative-correlative clauses are used fre-
quently and freely, “except for Sinhala, which, like neighboring Tamil, has
only the latter [the EHRC]” (Masica 1991: 408).9 In Dravidian languages,
8.3 Grammatical functions accessible 271
relative-correlative clauses are used less frequently. Korku (Munda) and most
of the TB languages do not have sentential relative clauses at all. They only
have a participial or infinitival gap relative.
One of the interesting issues that arises is: what are the grammatical functions
that are accessible to relativization in SALs? That is, what are the positions in
which a covert operator can occur?
In EHRCs (sentential relative clauses) of Pattern 1 and in the relative-
correlative construction of Pattern 2, all positions / grammatical functions
(subject, direct object, indirect object, oblique object, possessor modification
and object of comparison) of the Noun Phrase Accessibility Hierarchy (NPAH)
(Keenan and Comrie 1977) are relativizable, and thus there are no restric-
tions on these SALs. The relative-correlative construction that is found in
SALs is:
the characteristic Indo-Aryan (Old as well as New) . . . where the modifying clause,
marked by a member of the “J”-set of relative pronouns, adverbs, and other words, is
“represented” by a correlative in its role-slot (i.e. basically that of Modifier in the NP)
in the main clause. The correlative is usually identical with the remote demonstrative
in the language (except in Sindhi and Dakkhini, where the correlative so preserves the
form of an earlier IA demonstrative). (Masica 1991: 410)10
8.3.1 Indo-Aryan
Nominative subject as head
The NP mulgi ‘girl’ is not present in (9) in the right-adjoined relative clause.
Marathi (IA)
(9) ti mulgi [ji ghari geli] ti ithe rāhte
that girl which home went that here lives
‘The girl who went home lives here.’
(Wali 2006: 289)
Marathi (IA)
(10) jyā- lāi mulaj nastāt toi tyānnāj premāne vāgavto
who- to children not that (he) them-to love.with treat
‘One who does not have children treats them with love.’
(Wali 2006: 291)
An important fact about the relative clause in SALs is that the relativized NP
is generally fronted, if it is a non-subject (DO, IO, OO, etc.).
A comment about the V2 order in Kashmiri (IA) is in order. Recall that
Kashmiri is a V2 language (chapter 2). Bhatt (1999: 83) observes: “In relative
clauses and subordinate clauses V2 is blocked.” Hence, in (11) and (12) the
word order in the embedded clause is SOV and not SVO, while in the matrix
clause it is SVO. The position from which the NP is fronted is indicated by the
trace t, which is coindexed with the relativized NP in (11).
direct object
Kashmiri (IA)
(11) su d.ab [S2 yusi tse ti lobuthS2 ]
subject direct object verb
corr box rel (which) you found
chu k¯mtı̄
is precious
‘The box which you found is precious.’
(Wali and Koul 1997: 54)
8.3 Grammatical functions accessible 273
indirect object
(12) [su naphar] [S2 yas me gər ditsS2 ]
subject subject direct object verb
that person rel (whom) I.erg watch gave
gav gar
v e r b Adv
went home
‘The person (to) who(m) I gave the watch went home.’
(Wali and Koul 1997: 61)
Thus, the word order in Kashmiri (IA) in relative clauses exhibits an asymmetry
with that in the main clause.12 In contrast, there is no such asymmetry in
complement clauses (see chapter 6 for a discussion).
Oblique object
All oblique objects can be relativized in IA languages.
locative object
Marathi (IA)
(13) te ghar [dzyā ø- t mı̄nā rāhte]
corr house rel.obl (which) loc Meena live.pres.3s,f
‘The house in which Meena lives . . . ’
(Pandharipande 1997: 88–89)
Comitative
Sentence (14) is an example of the relative-correlative clause in which the
relative clause is left-adjoined to the correlate. The word order is V2 in the
matrix clause and SOV in the embedded clause.
Kashmiri (IA)
(14) [yemis naphras- sı̄¯th’ me sinema vuch]
who.dat (which) man.dat- with I.erg cinema saw
subject direct object verb
təm’ h’ot makān
he.erg bought house
s u b j e c t v e r b ( v 2) d i r e c t o b j e c t
‘The person with whom I saw the movie bought the house.’
(Sadaf Munshi, p.c.)
Sentence (15) is an example in which the relative clause occurs to the right of
the VP of the matrix clause.
274 Noun modification: relative clauses
Time adverbs
The relative adverb yelyi ‘when’ is the head of the relative clause in (16) in
Kashmiri and all types of adverbs can head a relative clause in Indo-Aryan,
Dravidian and Munda languages. A few Tibeto-Burman languages also permit
this. The word order is SOV in the embedded relative in (16).
r e l a t i v i z a t i o n o f p o s s e s s o r no u n
Marathi (IA)
(17) tı̄ mulgı̄ [dzi ø- tsa gān.a mı̄ aikla]
corr girl rel poss.3s,neut song.3s,neut I heard.pst.3s,neut
‘The girl whose song I heard . . . ’
(Pandharipande 1997: 89)
8.3.2 Dravidian
Dravidian has only the relative-correlative clauses (Pattern-2-type clauses dis-
cussed in 8.2), which are left-adjoined, and it does not have the other two
varieties, namely NP-adjoined and extraposed relative clauses, as in (2) and
(3) in Hindi-Urdu, and other Indo-Aryan languages. The question word and
the relative pronoun are homophonous, just as in English and French. Any
nominal constituent or an adverbial phrase can be the head of a relative clause
in Dravidian.
In Kannada, Sridhar (1990: 47) observes: “It [the relative-correlative clause]
is more used, however, in writing than in speech, especially to relativize minor
constituents such as location, circumstance, etc.” Asher and Kumari (1997: 58)
point out that in Malayalam too the “little used sentential relativization strategy
8.3 Grammatical functions accessible 275
Tamil (DR)
(18) yār aŋkē mutali vantu cēru- v- ār- ō avar
who there first come.cpm reach- fut- 3s,hon- dub mkr he.nom
t.iket.t.u vāŋkalam
ticket buy.permissive
‘Let whoever reaches there first buy the tickets.’
(Steever 1988: 30)
Steever (1988: 30) glosses -ō as ‘or,’ which is one of the functions that the
dubitative marker performs in several Dravidian languages.
Telugu (DR)
(19) [ēdi kāwāl(i)- ō] adi pat.t.u- ku- pō
what be-wanted- comp that take- VR go-imp-s
‘Take away what you want.’
Krishnamurti (2003: 448)
(vii) In Munda languages, the head noun may occur in the embedded as well
as matrix clauses.13
(viii) The relative pronoun in Khasi is ba-, and ba- is used as an adjectivalizer
as well as a complementizer. Following Temsen (2006) we treat ba- as
a relative pronoun. In this role, ba carries the agreement marker of the
head that it modifies (see examples (11)–(18) in appendix 8.1).
information with regard to the grammatical function of the noun phrase vis-à-
vis the embedded predicate is not overtly manifested. Hence, it may lead to an
ambiguous interpretation in some EHRCs in SALs. The EHRC is found in all
the four language families of the subcontinent, and it is the preferred strategy
for noun modification in Dravidian, Munda and Tibeto-Burman. In fact, in
almost all TB languages (for example, Hmar, Mizo, Manipuri, Zou, Thadou
and Paite), the EHRC/IHRC is the only strategy that is available. It appears
that the EHRC is the only indigenous strategy in Munda languages, and the
sentential relative is borrowed from Indo-Aryan.15
IHRCs, generally found in verb-final languages, are available exclusively
in some of the languages of the Tibeto-Burman language family. The crucial
issues that involve EHRCs and IHRCs are the following:
(i) grammatical positions accessible for relativization;
(ii) (potential) ambiguous interpretation of EHRCs and IHRCs; and
(iii) factors responsible for ambiguity and disambiguation.
We shall now discuss each of these issues.
Indo-Aryan
Subject modification
Subject modification of a [+transitive] verb is permitted in Sinhala, just as in
Dravidian, Munda, Tibeto-Burman and the Mon-Khmer Khasi. The modified
head corresponds to the subject in the participial clause. The gapped subject is
indicated by ø.
Sinhala (IA)
(21) [øi ē wæd.a kərəpu] minissui
that work do.pple men
‘The men who did the work.’
(Gair and Paolillo 1997: 54)
Hindi-Urdu (IA)
(22) [*øi kitāb par.h- ā huā] lar.kāi
book read- perf pple boy
Literally: ‘The book read boy.’
(See Subbarao 1984a for
a detailed discussion)16
There are some [–transitive] verbs kūdnā ‘to jump,’ daur.nā ‘to run,’ tairnā ‘to
swim’ in Hindi-Urdu (IA), which do not permit a perfect participle to modify
their subject “unless the predicate is telic, with an end point represented” (Alice
Davison, p.c.). For details, see Kachru (1980) and Subbarao (1984a).
280 Noun modification: relative clauses
Experiencer subject
The modified head corresponds to the experiencer subject in the participial
clause. The gapped subject which is a null (covert) operator is indicated by ø.
Hindi-Urdu (IA)
(24) [*øi bhūkh lag- ā huā] insāni
dat hunger feel- perf pple person
‘The man who felt hungry.’
In Gujarati (IA) and Marathi (IA), the modification of the dative experiencer
subject is permitted.
Gujarati (IA)
(25) [øi patthar vāgelo] vidyārthii
dat rock struck student
‘The student struck by a rock.’
(P. J. Mistry, p.c.)
Marathi (IA)
(26) ?[øi bāhuli āvd.- lel- i] mulgii
dat doll like- perf pple- 3s,f girl.3s,f
Literally: ‘The doll liked girl.’
[‘The girl who liked the doll.’]
(Wali 2004: 245)
Thus, Marathi and Gujarati differ from Hindi-Urdu in permitting the modifica-
tion of a dative subject.
corresponds to the direct object in the participial clause. The gapped direct
object which is a null (covert) operator in all these cases is indicated by ø.
The subject of a transitive verb is genitive case-marked in Bangla, Hindi-
Urdu, Punjabi, Kashmiri and Assamese, which indicates that embedded subject
NP acquires the status of a modifier by losing its nominative/ergative case.18
The modified head corresponds to direct object in the participial clause in (27).
The null (covert) operator is indicated by ø.
(27) [ram- er øi pɔr.- a] boii
Ram- gen read- perf book
‘The book that Ram read. . . ’
(Anupam Das, p.c.)
In (27), the predicate contains a perfect participle, and the embedded subject
is genitive case-marked. Ergative case marking is not permitted in (28a) in
Hindi-Urdu (IA). Though the sentence has the interpretation of a passive in
(28a), the predicate in (28a) is not in the passive.
Hindi-Urdu (IA)
(28a) [salmā- kı̄/ *ne øi likh- ı̄ huı̄] kitābi
Salma- gen erg write- perf pple book
‘The book written by Salma . . .’ (‘The book that Salma wrote . . . ’)
In Hindi-Urdu (IA), in some cases the theme can be modified (30a), if the
embedded predicate is dative case-marked, and in some cases it cannot be (30b),
282 Noun modification: relative clauses
while in Marathi (31) and Konkani (IA), this is permitted. The predicates cot.
lagnā ‘injury get [literally]’ and bhūkh lagnā ‘appetite feel [literal meaning]’ in
Hindi-Urdu (IA) take a dative case-marked subject, and the theme is cot. ‘injury’
in (30a) and bhūkh ‘appetite’ in (30b). While the former theme in (30a) can be
modified, the latter in (30b) cannot.19 As we shall see later, all the Dravidian
languages freely permit the modification of any non-nominative subject or the
theme, except the passive agent.
The modified head corresponds to the theme in the participial clause in (30a)
and (30b). The null theme is indicated by ø.
t h e m e 20
Hindi-Urdu (IA)
(30a) [rām ko øi lagı̄] cot.i bahut gahrı̄ hai
Ram dat get hurt.perf pple injury very deep is
‘The injury that Ram had is quite deep.’
(Rajesh Bhatt, p.c.)
In Marathi (IA) in (31), the predicate āvd. ‘like’ takes a dative case-marked
subject and the theme is bāhuli ‘doll’.
Marathi (IA)
(31) [mulgi- lā øi āvd.- lel- i] bāhulii
girl.3s,f- dat like- perf pple- 3s,f doll,3s,f
‘The doll which the girl liked.’
(Wali 2004: 245)
That is, while Marathi (IA) permits the modification of theme in a dative
case-marked sentence, Hindi-Urdu (IA) does it selectively.
Oriya (IA) differs from its east Indo-Aryan neighbor Bangla (IA). The subject
of the embedded sentence is nominative case-marked and it cannot be genitive
case-marked, as in (32). Recall that genitive case marking is permitted in
Bangla, Hindi-Urdu and Punjabi. We shall show that in Dravidian languages, the
embedded subject is nominative case-marked in such cases. Thus, Oriya does
not retain the Indo-Aryan pattern of the genitive case-marked subject, and it
appears to have borrowed the Dravidian pattern of the nominative case-marked
subject, as Dravidian permits such case marking. The nominative marker in
Oriya is null.
8.4 Externally Headed Relative Clause 283
Oriya (IA)
(32) [radha øi pɔd.hi thibə] bɔhii
Radha (nom) read.perf pple book
‘The book that Radha read . . . ’
(Prakash Patnaik, Subha Lakshmi Das
and Ahalya Das, p.c.)
Marathi (IA) exhibits two distinct patterns of case marking of the embedded
subject in participial EHRCs: it is either ergative case-marked or genitive case-
marked, with slightly different aspectual interpretations. The theme in (33) is
pustak ‘book.’ The embedded verb is in the perfect participial form.
The modified head corresponds to the direct object in the participial clause in
(33) and (34). The gapped direct object which is a covert operator is indicated
by ø.
Thus, Marathi shares with Bangla, Hindi-Urdu and Punjabi the feature of having
the genitive with the embedded subject (34), but it also may have the ergative
case marking on it, as in (33).
284 Noun modification: relative clauses
Nepali (IA)
(35) [məy- le hijo øi gare-ko] kāmi
I- erg yesterday do-perf pple work
Literally: ‘the (I did yesterday) work . . . ’
‘The work which I did yesterday . . . ’
(Clark 1977/1989: 185)
In Sinhala (IA), like in Dravidian, the genitive does not occur, just as in Oriya
(IA).
Sinhala (IA)
(36) [siri gunəpālat.ə øi dunnə] potəi
Siri (nom) Gunapala give.pst.adjective book
‘The book that Siri gave Gunapala.’
(Gair and Paolillo 1997: 54)
It is worth noting that in Dravidian languages the embedded subject does not
carry the genitive case marker. Synchronic evidence to show that contact may
induce such change comes from Dakkhini Hindi-Urdu (IA). Dakkhini Hindi-
Urdu, a transplanted variety of Hindi-Urdu (IA) in the southern parts of India,
which has been in intense contact with Telugu (DR) too, does not have the
genitive with the embedded subject, and nor does Telugu have the genitive in
such cases. Data from Dakkhini and Telugu are illustrative.
Dakkhini (IA)
(37) [baccā øi kar- e so] kāmi
boy do- perf pple adjr work
‘The work that the boy did . . . ’
(Arora 2004: 134)
Telugu (DR)
(38) rāmud.u øi cadiv- in- a pustakami
Ram read- pst- adjr book
‘The book that Ram read . . . ’
Thus, the non-occurrence of the genitive with embedded subject in Oriya and
Sinhala (IA) could plausibly be due to Dravidian influence, just as in Dakkhini,
as Oriya and Dakkhini have been in intense contact with Telugu (DR) and
Tamil (DR) with Sinhala.
8.4 Externally Headed Relative Clause 285
Punjabi (IA)
(39) *øi katāb dittı̄ (hoı̄) kur.ı̄i
book give.perf pple.f.s girl.f.s.
‘The girl to whom the book was given. . . ’
(Bhatia 1993: 61)
Kashmiri (IA)
(40) *rajan øi gər ditsmts kūri
Raj.erg watch give.perf.pple girl
Literally: ‘Raj watch-given girl.’
(‘The girl to whom Raj gave the watch.’)
(Wali and Koul 1997: 65)
Marathi (IA)
(41) tyā- ne øi āmantran. patrikā dilelı̄ sagl.ı̄ mān.sai
he- erg invitation card give.pst pple all.p.neuter people.p.m
lagnālā ālı̄
wedding.that come.pst.3p.m
‘All the people, whom he had sent (given) invitation cards, had come to
the wedding.’
(Pandharipande 1997: 90)
286 Noun modification: relative clauses
Sinhala (IA)
(42) [siri øi potə dunnə gunəpāləi ]
Siri book.def give.pst.adjr Gunapala
‘(the) Gunapala to whom Siri gave the book.’
(Gair and Paolillo 1997: 54)
Marathi (IA)
locative
(43) mı̄ øi rāhāt as.lel.a ghari khūp dzuna āhe
I live.pres be.pst.pple.s,n house.s.neut very old is
‘The house in which I am living is very old.’
(Pandharipande 1997: 90)
instrumental
(44) mı̄ øi patra lihilelı̄ pensı̄li mād.ı̄wartšyā kapāt.āt āhe
I letter write pst pple.s.f. pencil.s.f. upstairs of closet.loc is
‘The pencil with which I wrote the letter is in the closet upstairs.’
(Pandharipande 1997: 90)
time adverbial
(45) raghū øi dzanmalelyā wars.ı̄i wad.il㘠nnā khūp
Raghu born.pst pple.oblique year.loc father.dat a lot
paisa mil.ālā
money receive.pst.s.m.
‘The father got a lot of money the year Raghu was born.’
(Pandharipande 1997: 90)
comitative
(46) *[tsor øi tšyā barobar ālelā] polı̄s
thief poss with come.pst pple.3s. m policeman
‘The policeman who the thief came with.’
(Pandharipande 1997: 91)25
Oriya (IA) too permits the modification of the oblique object except the comi-
tative, just like Telugu, in particular, and Dravidian languages in general. The
modified head corresponds to the locative PP in the participial clause in (47)
and the comitative PP in (48). The gapped PP is indicated by ø.
Oriya (IA)
locative
(47) [mũ øi khai thiba] patrai
I eat. perf pple leaf
‘The leaf on which I ate’ (locative PP modification)
‘The leaf which I ate’ (DO modification)
comitative
(48) *[kɔmala øi asi.thiba] lokoi
Kamala come.perf pple person
‘The person with whom Kamala came.’
(Prakash Patnaik, p.c.)
Nepali (IA), too, permits the modification of the oblique object (locative, instru-
mental and ablative) (Pawan Upadhyaya, p.c.).
Nepali (IA), like the Dravidian languages, also permits the modification of
adverbial phrases, destination in (49) and time in (50). The modified head
corresponds to the PP indicating destination in the participial clause in (49),
the time adverbial phrase in (50) in Nepali. The gapped PP is indicated by ø.
Nepali (IA)
destination
(49) [timi hijo øi gae- ko] ghari
you yesterday go- perf pple home
Literally: ‘The (you went yesterday) house . . . ’
‘The house you went to yesterday . . . ’
(Clark 1977/1989: 185)
time
(50) [ma hijo øi ae- ko] belāi
I yesterday come- perf pple time
Literally: ‘The (I came yesterday) time . . . ’
‘The time at which I came yesterday . . . ’
(Clark 1977/1989: 185)
288 Noun modification: relative clauses
Thus, Marathi, Oriya and Nepali differ from Hindi-Urdu, Punjabi and Kashmiri
in permitting the modification of oblique objects.
The modification of the comitative NP is not permitted in any IA language.
We provide an explanation in 8.6, in terms of the Thematic Eligibility Con-
dition, of why the comitative PP cannot head an EHRC.
Modification of possessor
Possessors can be modified in Hindi-Urdu and Bangla (IA). The embedded
verb kat.nā in Hindi-Urdu and kat. in Bangla (‘to be cut’) is [−transitive]. In
Hindi-Urdu, as Rajesh Bhatt (p.c.) points out, only unaccusative predicates
such as kat.nā ‘to be cut,’ .tūt.nā ‘to be broken,’ can form a participle, and the
possessed NP is an argument of an unaccusative predicate.
In Bangla too a similar restriction holds (Shukla Basu, p.c.). Sentences (51)
and (52) are illustrative.
Hindi-Urdu (IA)
(51) øi dum kat.ı̄ billii
tail cut.perf pple cat
‘The cat whose tail got cut.’
(Subbarao 1984b)
Bangla (IA)
(52) øi læj kat.a ber.ali
tail cut.perf pple cat
‘The cat whose tail got cut.’
Recall that in Hindi-Urdu, while the subject – with a specific set of verbs – and
direct object can be modified, the oblique object cannot be. A similar situation
obtains in some other IA languages too. The fact that the object of the genitive
in Hindi-Urdu can be modified demonstrates that there are breaks in the NPAH,
and it is not a continuum as Keenan and Comrie (1977) claimed.26 We have
shown the positions that are accessible in IA languages and the variation that is
found within the same language family with regard to the positions accessible.
We shall now show the positions accessible for relativization in the Dravidian
language family.
Dravidian
In Dravidian languages, all the positions of the NPAH can be modified. We
provide only a few examples.
In Dravidian languages, the preferred strategy is the participial-type, and the
relative-correlative clause occurs either in formal style or in free relatives. In
most Indo-Aryan languages, both the strategies are available, and it is rather
8.4 Externally Headed Relative Clause 289
Subject modification
In Dravidian languages, there are no restrictions on the transitive nature of the
predicate of the participle in EHRCs. Thus, both [+transitive] and [–transitive]
predicates freely permit the modification of the subject.28 The modified head
corresponds to the subject in the participial clause. The gapped subject is
indicated by ø.
Telugu (DR)
(53) [ivāl.a proddunna øi gōd.a gent- in- a] pillai
today morning wall jump- pst- adjr girl
Literally: ‘Today morning wall-jumped girl . . . ’
‘The girl who jumped the wall this morning . . . ’
Malayalam (DR)
(54) [niŋŋal.ōt.ə sammεl.anatte kuriccə øi para
a innale ivit.e
you. with conference.acc about tell.pst pple yesterday here
vanna] āl.əi
come.pst pple man
‘The man who told you about the conference and who came here yesterday.’
(Asher and Kumari 1997)
Telugu (DR)
(55) [mallika øi cadiv- in- a] pustakami bāg- undi
Mallika (nom) read- pst- adjr book good- is
‘The book that Mallika read is good.’
290 Noun modification: relative clauses
Tamil (DR)
simple sentence
(56) taccan van.n.ān- ukku pan.am kut.uttān
carpenter washerman- dat money give.pst.3 s,m
‘The carpenter gave money to the washerman.’
ehrc – io modified
(57) [S2 taccan øi pan.am kut.utt- aS2 ] [van.n.āni ]
carpenter(nom) money give.pst- rp washerman
[iŋe vantān]
here came.pst.3 s,m
‘The washerman to whom the carpenter gave money [came here].’
(The VP iŋe vantān is added to the original sentence from
Asher 1985: 28)
Asher (1985: 28) uses the term ‘rp’ (relative (adjectival) particle), which is
termed an adjectivalizer in this work.
The following points concerning (57) are worth mentioning: (i) the identical
relativized noun phrase in the embedded clause is absent, and it is indicated
by ø, and (ii) it is coindexed with the head noun (Nadkarni 1970; Subbarao
1984b). The suffix -a is the adjectivalizer.
We shall now discuss the modification of oblique objects. The modified head
corresponds to the oblique object (a PP) in the participial clause in sentences
(58)–(61), (63) and (65). The gapped oblique object is indicated by ø.
Kannada (DR)
(58) avanu øi hut.t.id- a ūrui idē
he born.pst- adjr town this.emph
‘This is the town where he was born.’
(Sridhar 1990: 57)
8.4 Externally Headed Relative Clause 291
Ablative
Telugu (DR)
(59) [pragati øi nı̄l.l.u tecc- in- a] bāvii
Pragati water bring- pst- adjr well
‘The well from which Pragati brought water.’
Instrumental
Malayalam (DR)
(60) amma øi māŋŋa pūl.iya kattii
mother mango slice.pst.rp knife
‘The knife with which mother sliced the mango.’
(Asher and Kumari 1997: 60)
A time adverb can head an EHRC in Telugu (DR). The head of the EHRC in
(61) is appud.u, which literally means ‘then.’
Telugu (DR)
(61) mı̄ru øi vacc- in- a appud.ui ceppand.i
you come- pst- adjr then (time) tell.hon
‘Tell me when you come.’
Other time expressions such as tarvāta ‘after,’ dāka ‘until,’ lōpala ‘prior to’ and
mundu ‘before’ can also head an EHRC (Krishnamurti and Gwynn 1985:246) in
Telugu – for example, cēs-in-a tarvāta ‘after having done,’ cēsē mundu ‘before
doing’ (1985: 246). A negative participle may also modify a time expression
such as mundu ‘before’ (1985: 246), as in ceyya.ka mundu ‘before not doing,’
that is, ‘before doing.’ The truth value of the sentences with an affirmative CP
and negative CP is identical.30
(65) the comitative PP is the head. In (64) and (65), vastādu ‘wrestler’ is a
subcategorized argument of the conjunct verb kustı̄ pat..tu ‘to wrestle.’
Telugu (DR)
(64) mā snēhitud.u oka vastādu tō kustı̄32 pat.t.ēd.u-
our friend one wrestler with (comitative) wrestling (noun) caught
‘Our friend wrestled with a wrestler.’
to the patient / theme in the participial clause. The gapped patient is indicated
by ø.
e h r c : p a t i e n t m o d i fi e d i n a d a t i v e s u b j e c t c o n s t r u c t i o n
Malayalam (DR)
(68) [aval.kkə øi karun.a tōnniya] pāvappet.t.avati
she.dat pity feel.pst.rp poor people
‘The poor people for whom she felt pity . . . ’
(Asher and Kumari 1997: 61)
e h r c : n o m i n a t i v e t h e m e m o d i f i e d w i t h a no m i n a t i v e l o c a t i v e
sub j e c t
Telugu (DR)
(69) [kavita daggira øi unn- a] mugguru panivāl.l.ui
Kavita near be- adjr three workers
‘The three workers that Kavita has . . . ’
It is pertinent to mention here that Hindi-Urdu (IA) does not permit the modi-
fication of the subject (experiencer) in (70), or the theme in the dative subject
construction in (71).33
The modified head corresponds to the experiencer in the participial clause in
(70). The gapped experiencer is indicated by ø.
n o n - n o m i n a t i v e ( d a t i v e ) s u b j e c t m o d i f i e d – no t p e r m i t t e d
experiencer modified
Hindi-Urdu (IA)
(70) [*øi bhūk lag- ı̄ huı̄] baccı̄i
appetite feel- perf pple female child
Literally: ‘The hunger-felt female child.’
‘The female child who felt hungry.’
The modified head corresponds to the theme in the participial clause in (71).
The gapped theme is indicated by ø.
theme modified
(71) [*baccı̄i ko øi lag- ı̄ (huı̄)] bhūk
female child dat feel- perf pple appetite
‘The appetite felt by the female child.’
8.4.2 Asymmetries
In appendix 8.2 on the Website we discuss the asymmetries found in various
types of sentential relative-correlative clauses with regard to double or single
8.5 Internally Headed Relative Clause 295
occurrence of the head, the feature of definiteness on heads, and the occurrence
of bare NPs. In appendix 8.4, also on the Website, we discuss the asymme-
tries found in Tenyidie (TB) with regard to the position of occurrence of the
embedded relative when a quantifier or numeral has the head noun in its scope
in EHRCs.
Tenyidie (TB)
(72) [nɔ kutarii - pie nhasi le- ke-] c- øi u a- se
you knife- instr fruit cut- nozr- dm- def big- very
‘The knife with which you cut the fruit is very big.’
permit the definite marker cu to occur (Lalremzami Chinzah, p.c.) as the internal
head is already [+definite]. In a simple sentence, cu, the definite marker /
accusative case marker, occurs with the direct object, as in (73a).35
Mizo (TB)
(73a) zovi- n lekhabu cu a- lei
Zovi- erg book def 3s- buy
‘Zovi bought a book.’
d i r e c t o b j e c t m o d i fi c a t i o n
(73b) zova- n [S2 zovi- n lekhabu a- leiS2 ] kha a- chiar- aŋ
Zova- erg Zovi- erg book 3s- buy det 3s- read- fut
‘Zova will read the read book which Zovi bought.’
Just as in EHRCs (sentences (21) and (22) in appendix 8.3), there is another
alternative for (73b), as in (74). In (74) the embedded subject does not carry
ergative case, and consequently the embedded verb does not have agreement
markers. We have marked their absence by ø in S2 in (74).
Mizo (TB)
(74) zova- n [S2 zovi-ø lekhabu ø- leiS2 ] kha a- chiar- aŋ
Zova- erg Zovi- ø book ø- buy det 3s- read- fut
‘Zova will read the book which Zovi bought.’
Indirect object
Category I: IHRC with an overt case marker on the IO, the internal head. The
indirect object as head, and the postposition, are in italics.
i o a s h e a d o f a n i h r c (to- d a t i v e )
Sangtam (TB)
(75) [n- n nistar (IO) khaŋ šeti t.hraʔ- ba- tsə] khataŋ tšŋle
you- nom person to letter write- nozr- def very tall
‘The person you wrote a letter to is very tall.’
Category II: IHRC with no overt case marker on the IO, the internal head In
Hmar (TB) the indirect object is case-marked by the dative case marker hnenaʔ
‘to, for’ in a simple clause as in (76).
Sentences (77a) and (77b) are examples of a left-adjoined IHRC in which the
internal head is an indirect object. Note that in (77a), (i) the head NP bɔŋ ‘cow’
occurs in the embedded clause as an internal head, (ii) the dative case marker
hnenaʔ ‘to, for’ is not present with the IO, and (iii) the ergative marker -n with
the embedded subject is overtly present, as the embedded verb pek ‘give’ is
[+transitive], and hence, a-, the 3rd person singular agreement marker with the
verb too is overtly present. Note that pro in (77a) and (77b) is obligatorily null
and it is coindexed with bɔŋ ‘cow.’
In Mizo, Hmar, Thadou and Paite IHRCs, the empty position in the matrix clause
coindexed with the internal head in an IHRC is pro as the matrix verb/adjective
exhibits agreement in number and person with the pro. The 3rd person singular
agreement marker -a on the matrix verb thi ‘died’ shows that the deleted
element is pro which is obligatorily null.
Kumar (1998) observes that, in (77b), the absence of both of them (the
ergative marker with the embedded subject and the agreement marker with the
embedded verb) in the IHRC results in ungrammaticality.
i h r c p a t t e r n i i – no t p e r m i t t e d
(77b) *[lala- ø bɔŋi bu ø- pek] kha proi zanikhan a- thi
Lala- erg cow food 3s give det yesterday 3s- died
‘The cow which Lala gave food to died yesterday.’
Thus, in Hmar (TB), the overt presence of the ergative marker and the
agreement marker in the embedded clause in an IHRC with IO as head is
required, in contrast to in Mizo, where there is an option with regard to their
occurrence.
In (80), the NP azkhikhi ‘well’ occurs with an ablative case marker lɔnɔ ‘from.’
Still it cannot head the IHRC, as the PP azkhikhi- lɔnɔ ‘well- from’ occurs only
in the embedded relative clause. Thus, it is the single occurrence of the PP in
the embedded clause alone that rules out the interpretation with the ablative PP
as head, in spite of fulfilling the requirements of both case and word order for it
to be the head. However, the DO az ‘water’ or azkhikhi ‘well’ can potentially
be the heads of the internally headed relative clause; the DO is interpreted as
the head in (80), and not the ablative PP azkhikhi lɔnɔ ‘well from.’
To make an ablative PP the head of an IHRC, there is a specific strategy
that Sema adopts. Under this strategy, the head noun is partially repeated in
the matrix clause. It occurs to the right of the definite marker -u in a position
earmarked for the head noun in an externally headed relative clause. Sentence
(81) is illustrative.
The repetition of the noun phrase azkhikhi ‘well’ as zkhikhi is only partial, and
a-, the generic possession marker, is not repeated. Thus, partial reduplication
is a syntactic strategy that Sema adopts to distinguish between IHRCs with DO
and ablative PP as head.
In this section, we have discussed IHRCs with and without an overt case
marker on the internal head. We have noted that an externally headed relative
clause is not permitted with comitative PP as head in Dravidian, Indo-Aryan,
Mon-Khmer and Tibeto-Burman languages. The modification of the comi-
tative PP as head in EHRCs and IHRCs deserves a special discussion and,
hence, it is the topic of discussion of the following section. We shall expli-
cate the reasons for the non-permissibility of sentences with comitative PP
as head in EHRCs in most of the languages, and in IHRCs in a subset of
languages.
Hierarchy (NPAH) of Keenan and Comrie (1977) amongst South Asian lan-
guages of the four language families (Indo-Aryan, Dravidian, Austro-Asiatic
[Mon-Khmer, Munda] and Tibeto-Burman), the modification of the comitative
PP in EHRCs is not permitted in almost all SALs, except for a few. In the few
languages in which it is permitted, specific syntactic criteria have to be met
to enable the comitative PP to head the EHRC. We shall first elaborate what
these criteria are and then attempt to formulate a generalization that would
neatly account for all these phenomena. We shall also propose a condition
that we shall label as the Thematic Eligibility Condition (TEC) to account for
the facts cross-linguistically, and we wish to demonstrate that if any one of
these syntactic specific criteria is fulfilled, then the requirements of the TEC
are met.
We shall now present a discussion of the modification of the comitative PP.
It is worth recalling that the head of all EHRCs and only a subset of IHRCs
does not carry any adposition that would indicate the thematic relationship of
the head with the embedded predicate, whereas a subset of IHRCs in some
Tibeto-Burman languages does carry an overt case marker. We shall show that
the overt occurrence of the postposition plays a crucial role in the subset of
languages which carries it.
There are restrictions on the occurrence of an EHRC and an IHRC in gap
strategy with a comitative PP as head, when it has the interpretation of accom-
paniment. In those languages in which a PP can head an EHRC, the modification
of the other PPs (locative, ablative, instrumental) is freely permitted, but not of
the comitative PP. The issue we wish to address is: why is there this asymmetry
with regard to the comitative PP alone? Can the reasons, if any, be pinpointed?
What are the syntactic phenomena that are involved in enabling a comitative
PP to head an EHRC?
Unlike most other PPs in the NPAH, for the comitative PP to head an EHRC
or an IHRC, specific syntactic criteria have to be met. We shall examine the
criteria below. We shall demonstrate that once any one of the criteria is fulfilled,
a thematic relationship between the head of the comitative PP and the embedded
predicate is established and that is what enables a comitative PP to head an
EHRC.
First, let us look at a few examples of EHRCs in SALs with the comitative
PP as the head.
Dravidian:
Recall that Dravidian languages do freely permit the modification of oblique
objects (locative, ablative and instrumental), but comitative PP as head is not
permitted in any Dravidian language. For example:
8.6 Comitative PP as head 301
e h r c w i t h a c o m i t a t i v e p p a s he a d
Malayalam (DR)
(82) *bābu vanna ez.ut.t.ə
Babu come.adjr letter
‘The letter with which Babu came’
[That is, Babu came bringing a letter with him.]
(Asher and Kumari 1997: 61)
Tamil (DR)
(83) *kumār vel.iye pōna umā
Kumar out go.pst.adjr Uma
‘Uma, with whom Kumar went out’
(Annamalai 1997: 78)
Telugu (DR)
(84) *nēnu vel.l.- in- a saral.a
I go- pst- adjr Sarala
‘Sarala with whom I went’
(Ramarao 2003: 79)36
Indo-Aryan:
Marathi (IA)37
(86) *karun.a gel- lel.i mulgi
Karuna go- perf pple.f girl
‘The girl with whom Karuna went’
(Lalita Dhareshwar, p.c.)
We do not have data from Munda languages on the modification of the comi-
tative PP. Khasi (Mon-Khmer) permits an EHRC with a comitative PP as head
302 Noun modification: relative clauses
Tibeto-Burman:
In Tibeto-Burman languages, an EHRC is not permitted with a comitative PP as
head in Manipuri, Tenyidie, Sangtam, Mizo, Hmar and Konyak. For example:
t e n y i d i e (t b ) – e h r c n o t p e r m i t t e d
(88) *nɔ vɔr ke- c- miepuo- u puo cha cha se
you come nozr- dm- person- def 3 s height long very
Intended meaning: ‘The person you came with is very tall.’
(Subbarao and Kevichüsa 1999: 162)
Sangtam (TB) – e h r c n o t p e r m i t t e d
(89) *n - n ro- ba- nitsar tsə khataŋ atšŋ e
you- nom come- nozr- person def very tall is
Intended meaning: ‘The person with whom you came is very tall.’
In Tibeto-Burman, there are two sets of languages: let us label them as Set A and
Set B. In Set A the internal head of an IHRC overtly carries the comitative case
marker that establishes a thematic relation with the embedded predicate. In Set
B the internal head of an IHRC does not carry the comitative case marker and,
hence, it is similar to the head in an EHRC in not having an overt postposition
with the head and, consequently, it cannot establish a thematic relation with the
embedded predicate.
Let us consider the case of languages of the Tibeto-Burman family from
Set A in which the internal head carries an overt comitative postposition. These
languages include Ao, Tenyidie, Konyak and Sangtam (TB).
Note that miepuo ‘person’ in (90) in Tenyidie (TB), for example, occurs in
the embedded relative, and is case-marked by ze ‘with.’
8.6 Comitative PP as head 303
t e n y i d i e (t b ) – ihrc w i t h a n o v e r t c o m i t a t i v e – p e r m i t t e d
(90) nɔ miepuo- ze leški- nu vɔ ke- c- u- e mhacha
you person- com school- to go nozr- dm- def- nom cook
si se
well very
‘The person (who) you went to school with cooks very well.’
In Sangtam (TB), too, the case marker te ‘with’ is overtly present with the
internal head nitsar ‘person.’
s a n g t a m (t b ) – o o a s he a d o f a n ihrc w i t h a n o v e r t c o m i t a t i v e –
permitted
(91) n- n nitsar- te ro- ba- tsə khataŋ atšŋ e
you- nom person- com come- nozr- def very tall is
‘The person with whom you came is very tall.’
Let us now consider the languages in Set B where the internal head does
not carry an overt comitative postposition. These languages include Manipuri,
Mizo, Hmar, Thadou, Zou, etc. In this set, a comitative PP cannot head an IHRC,
for example as in Manipuri (TB). The head mi ‘person’ is uncase-marked in
(92) and the sentence is ungrammatical.
m a n i p u r i ( t b ) – o o a s he a d o f a n i h r c (c o m i t a t i v e ) – n o t
permitted
(92) *nahak mi adu lak- pə pha- i
you person det come- inf good- [-fut]
‘The person with whom you came is good.’
There are some Tibeto-Burman languages, such as Hmar, Mizo, Paite and
Thadou, in which a postposition is incorporated into the verb (see chapter 4 for
details).
In Thadou, in the embedded verb, the verbal reciprocal ki and the comitative
postposition pi are incorporated in an EHRC, as in (94). The valence of the
[−transitive] verb ɔn ‘go’ increases by one and as a result it acquires the status
304 Noun modification: relative clauses
of a [+transitive] verb. Evidence for this claim comes from the fact that the
verb ɔn ‘go’ is [–transitive] and, hence, it does not permit an ergative marker
to occur with the subject in Thadou. However, once the incorporation of the
postposition takes place, the embedded subject thaŋboi ‘Thangboi’ is marked
ergative in (94), though the verb is still ɔn ‘go.’ Hence, the reason for the
acquired transitive status of the predicate is due to the incorporation of the
postposition.39 The verbal reciprocal and the incorporated postposition are in
italics.
Thadou (TB)
(94) thaŋboii - in sinema ve øj ai - kii 40 ɔn- pi- nu- cuj
thangboi- erg cinema to see 3s- VREC go- with (ip) f mkr- def
ka kadɔŋmaj - nu aj - hi
my girlfriend- f mkr 3s- be
‘The girl with whom Thangboi went to see the movie was my girlfriend.’
The abbreviation ip in (94) stands for “incorporated postposition.”
(Subbarao and Haokip 2009, 2011)
Once postposition incorporation takes place, the bare head of the comitative
PP in the EHRC gets syntactically linked to the newly derived verb with the
incorporated postposition.42 In addition, the occurrence of the verbal reciprocal
shows that the agent of the verb performs the action together with someone else,
namely the accompanied person. Thus, the incorporated postposition performs
the same function as the comitative postposition with the head in an IHRC in
some Tibeto-Burman languages such as Tenyidie and Sangtam.
Thus, it is due to the occurrence of these markers – namely,
(i) the comitative postposition with the head in an IHRC, or
(ii) the verbal reciprocal and the adverb ‘together,’ or
(iii) the comitative postposition incorporated in the verb –
that a thematic relationship between the comitative NP as head and the predicate
is established.
Based on the discussion above, the following generalization can be made:
(96) A comitative PP can head an EHRC/IHRC if and only if either:
(i) the head is overtly case-marked comitative, or
(ii) if the head is not overtly case-marked comitative, the embedded verb
carries either the verbal reciprocal, or an incorporated postposition, or
both, to establish a thematic relationship of the bare head NP with the
embedded predicate.
In Korean, too, the occurrence of the adverb hamkkey ‘together’ with the
embedded predicate is obligatory.
Korean
(98) yeongsig- i *(hamkkey)- on yeca
Yeongsig- nom together- come girl
’The girl Yeongsig came with’
(Data provided by Masayoshi Shibatani, p.c.)
308 Noun modification: relative clauses
Other intransitive verbs such as to swim, to walk, to run, to sit, to crawl, etc.,
also require the presence of the adverb ‘together’ with the embedded predicate
in Japanese and Korean (Matt Shibatani, p.c.), just as in Khasi (Subbarao and
Temsen in press), Manipuri and Thadou, etc. The occurrence of the adverb
‘together’ in Japanese and Korean clearly demonstrates that a comitative PP
can head an EHRC if and only if the predicate can get “thematically related ”
to the embedded predicate.
To summarize the above discussion, we have demonstrated that a thematic
relation has to be established between the predicate of the embedded clause and
the head of the EHRC/IHRC, in terms of either an overt case marker with the
head, or some marker in the embedded verb to indicate accompaniment. That
is, specific syntactic criteria have to be met to make the predicate “thematically
eligible” to accept a comitative PP as head. We propose that each predicate has
to meet the “Thematic Eligibility Condition” (TEC) for it to accept a comi-
tative PP as head. Such eligibility is the result of language-specific syntactic
processes such as Postposition Incorporation, the incorporation of the verbal
reciprocal as a group marker and the adverb ‘together,’ or the incorporation
of the adverb together alone in the embedded predicate, or the overt presence
of the comitative postposition with the head in an internally headed relative
clause.
It is significant that in languages belonging to different language families,
the same kind of verbal marker – namely, a verbal reciprocal – is attached to
the embedded verb to achieve the same result.43
It is also significant that in languages belonging to two different language
families, the same kind of verbal marker – namely, a verbal reciprocal – is
attached to the embedded verb to achieve the same result.
Hmar (TB)
e h r c w i t h a [+t r a n s i t i v e ] v e r b – p e r m i t t e d
(99) [loman hmu] naupaŋpa kha kan in- aʔ a- huŋ
prize get boy DD2 our house- to 3 s- came
‘The boy who got the prize came to our house.’
i h r c w i t h a [+t r a n s i t i v e ] v e r b – no t p e r m i t t e d
(100) *[loman naupaŋpa hmu] kha kan in- aʔ a- huŋ
prize boy get DD2 our house- to 3 s- came
Examples (99) and (101) show that EHRCs are permitted independently of the
transitive nature of the embedded verb. However, this is not the case for the
IHRCs. The IHRC in (100) is ungrammatical, while the EHRC in (101), in
contrast, is grammatical as the embedded verb is [–transitive].
Let us now look at Postposition Incorporation in Hmar (see chapter 4 for
discussion of incorporation). We shall now consider applicative constructions
in Hmar and the formation of IHRCs.
Applicative constructions are the ones that permit incorporation of an adpo-
sition (preposition and postposition) of a postpositional phrase (PP) in the verb.
Once the adposition is incorporated, in place of the PP we are left with a bare
NP. That is, the PP is stripped of its adposition due to incorporation that results
in the addition of a suffix to the verb.
In Hmar, we observe that: (i) only intransitive verbs permit Postposition
Incorporation, and incorporation of postposition has a transitivizing effect on
the verb; and (ii) the PP becomes the object of the verb after the adposition is
310 Noun modification: relative clauses
incorporated, as there is only a bare NP that is left. The bare NP that is left
receives accusative case from the verb.
(103) zova cauki- a a- t.šuŋ
Zova chair- on 3 s- sat
‘Zova sat on the chair.’
After incorporation
(104) zova- n cauki a- t.šuŋ- pui
Zova- erg chair 3 s- sat- loc suffix (on)
‘Zova sat on the chair.’
In (103), (i) the verb .tšuŋ is an intransitive verb and, hence, the subject Zova
does not carry the ergative case marker, and (ii) the PP carries the postposition
a ‘on.’ In contrast, in (104), incorporation of the postposition has taken place,
as a result of which the derived verb acquired the status of a transitive verb. In
(104), due to adposition incorporation: (i) the subject Zova carries an ergative
case marker -n; (ii) the NP cauki ‘chair’ is a bare NP, and is not followed
by the locative postposition; (iii) the verb carries the locative suffix pui to its
right; and (iv) the newly formed transitive verb .tšuŋ -pui assigns structural
accusative case to the object NP cauki ‘chair.’ Incorporation in Hmar, thus,
conforms to Baker’s proposal concerning applicative constructions, according
to which: “a grammatical applicative construction can only occur when the
derived verb assigns accusative case to the NP that is stranded by the movement
of preposition” (Baker 1988: 252).
We shall now demonstrate that this type of Postposition Incorporation in
Hmar has implications for the formation of the IHRCs in that language. We
observe that a [–transitive] verb in the embedded clause permits an IHRC, as
in (105), while a [+transitive] verb does not.
i h r c ( w i t h no i n c o r p o r a t i o n o f t h e p o s t p o s i t i o n i n t h e
embedded clause) – permitted
(105) [zova tabul cu- a kei- le t.šuŋ- na] cu a- lien
Zova table det- on I- with sit- loc adv mkr DD2 3 s- big
‘The table on which Zova sat with me is big.’
8.8 Conclusion48
In this chapter we provided a detailed analysis of relative clauses, participial
and nominalized EHRCs and IHRCs. With regard to the relative-correlative
clauses, we have shown that: (i) Indo-Aryan languages have relative-correlative,
312 Noun modification: relative clauses
1 I N TRO D U CTION
1. Andamanese and Nicobarese are omitted from this study because of lack of available
syntactic data.
2. Chomsky (1965) used the term subcategorization. Subcategorization, put simply,
means the number of essential arguments that a verb can take in its predicate. As
Lasnik and Uriagereka with Boeckx (2005: 3) put it: “a lexical entry must contain
syntactic information about whether, for instance, a verb is transitive, intransitive,
ditransitive, and so on (so-called subcategorization).” See Chomsky (1965) for fur-
ther details.
3. The feature [−initial complementizer (IC)] in parameter (3) corresponds to [+finite
complementizer].
4. Martin Everaert (personal communication) informs me that in Tanya Reinhart’s Theta
system, it is clearly defined (i) which thematic roles exist, and (ii) how to decide on
what thematic role an argument has (see Reinhart 2002).
See Subbarao and Temsen (in press) for a detailed discussion of Externally
Headed Relative Clauses (EHRCs) with the comitative PP as head in Khasi, and
Subbarao and Haokip (2011) for Thadou EHRCs and Internally Headed Relative
Clauses (IHRCs).
5. Thanks to Alice Davison for her help in the formulation of this parameter.
6. Based on data from noun incorporation in Sora, we demonstrate that there are some
vestiges left of ancestral non-verb-final structures in this language and point out
that only diachronic evidence can resolve the paradox in the order of occurrence of
incorporated elements (see chapter 4 for a detailed discussion).
2 S O U TH A S I A N L ANGUAGE S: A PR E VIEW
1. Due to the limitations of space, we have not focused on Khasi (Austro-Asiatic,
Mon-Khmer). For a detailed study of the features of Khasi, see Temsen (2006) and
Temsen and Subbarao (in preparation).
2. We have also not included Andamanese and Nicobarese in this study due to lack
of sufficient syntactic data and evidence. See Abbi (2006) for a description of the
endangered languages of the Andaman Islands.
3. South Asia, along with the Balkans, is a prime example of the rather rare phenomenon
known as Sprachbund, first mentioned by Trubetzkoy, or “linguistic area” (Emeneau
1956) or “convergence area” (first suggested in Weinreich 1958; see also Hock 1991:
494).
313
314 Notes to pages 18–19
Weinreich (1958) in his paper in Word (14,379) was the first to use the phrase
“convergence area.” The term convergence has been in use ever since to characterize
the phenomenon observed in language contact situations which results in changes at
the level of phonology, morphology, syntax and semantics. It is not clear, however,
who used the specific term convergence first.
4. Pnar, Jowai and Rymbai (Mon-Khmer) are verb-medial too (see Bareh 2007 for
details).
5. Thanks to Rajesh Bhatt for bringing this important paper to our attention.
6. According to Comrie, language typology is concerned with “differences among lan-
guages” and “with the study of variation” (1981: 30–31). The most commonly found
types of languages are verb-final, verb-medial and verb-initial, in which the verb
occurs in the final or medial or initial position of a sentence, respectively. In the stud-
ies on word order typology depending on the position of a verb (head) in a sentence,
certain common structural characteristics amongst languages are abstracted. These
common characteristics are generally called word order universals (see Greenberg
1966 for details). A study of word order universals of SALs demonstrates how sim-
ilar or different the languages are that belong to the four different language families
of the subcontinent.
7. The features discussed are based on the study of the syntax of these languages
and there is no psychological or functional explanation in support of the observed
similarities.
8. There is, however, an exception to this generalization: in Kashmiri, the finite form
of the verb occurs in the second position (labeled as V2 position) in a sentence, just
as in Dutch and German and, thus, it may precede the object as in (i).
Kashmiri (IA)
(i) rāman dits šāmas kitāb
Ram.erg give.pst V2 Sham.dat book
‘Ram gave Sham a book.’
In (i), the finite form of the verb (V2 ) dits ‘gave’ carries the finite past tense marker,
and in (ii), chu ‘be.pres’ carries the finite present tense marker. Sentences (i) and
(ii) would be ungrammatical if the finite form of the verb occurred in the final position
in the sentence to the right of the object, as (iii) and (iv) illustrate.
11. The verbal reflexive marker -in occurs with a set of adjectives in Hmar (TB) and
such occurrence seems to be lexically determined.
12. Time and place adverbials occur in ascending order in Khasi, a non-verb-final
language.
13. Butt and Geuder (1999) point out that vector verbs, a.k.a. light verbs, are the result
of a type of polysemy, and semantic bleaching is not an issue.
14. See the appendix to this chapter on the Website for a discussion of the reduplication
of some other constituents in SALs.
15. Though echo words are generally considered a part of morphology, they have a
role to play in terms of politeness. While in some contexts the use of an echo
expression might indicate intimacy between the speaker and hearer, in some other
contexts it might signal rudeness that hurts the sentiments of the hearer. Such use
also is an index of the social hierarchy between the speaker and hearer. The use
316 Notes to pages 29–31
Telugu (DR)
mı̄ ammāyi pel.l.i ayyindi mari pārt.ı̄
your [+honorific] daughter wedding occurred so party
adi+i/ gı̄rt.ı̄ ivvar- ā
that (literal)+emph/ echo word give- pol q mkr
and the thing like
‘Your [+honorific] daughter got married. Aren’t you giving a party and
the like?’
Hindi, Punjabi and Bangla (IA) do not permit such usage of a pronominal copy,
and it needs to be worked out whether it is permitted in the other two language
families.
16. Khasi (Mon-Khmer), which is verb-medial, is the only South Asian language that
we know of that strictly observes superiority effects.
17. While Hindi allows for “direct speech” complements as in (28) as reported in the
main text, it also allows for indirect speech complements, too. Thus, one can say:
18. It is not uncommon to find sentences which do not observe tense harmony strictly
in English (Peter Hook and Colin Masica, p.c.). Peter Hook points out that a Google
search will substantiate this point.
19. There are only two SALs that we know of which are an exception to this gener-
alization. These are Khasi (Mon-Khmer) (Subbarao 2001) and Tenyidie (Tibeto-
Burman) (Kevichüsa 2007).
20. In Khasi, which is a non-verb-final language, too, there are no expletive expressions.
Normally, pro occurs in subject position. However, the NP ka-bnεŋ ‘the heaven’
may also occur in subject position.
Notes to pages 32–43 317
Khasi (Mon-Khmer)
proi / kai - bnεŋ kai - khryat
f,s- heaven f,s- cold
Literally: ‘The heaven is cold.’ / ‘It is cold.’
(Temsen 2006)
21. Mangalore Konkani also has structures involving Backward Control as the example
below shows.
Mangalore Konkani (IA)
[S1 [S2 ammii haŋga yev- nuS2 ] ∀i dha vərsəj dzallı̃jS1 ]
we here come- cpm ten years happened
‘It is ten years, since we came here.’
(Lalita Dhareshwar, p.c.)
22. Hany Babu (p.c.) points out that such sentences are not used in his dialect of
Malayalam. This matter of dialect variation in Malayalam is worth investigating.
23. Thanks to Philippe Bourdin (p.c.) for suggesting the label “cognitive gerunds” for
such constructions. According to him, French too has a similar construction.
24. In most of the Indo-Aryan languages we checked, this construction does not exist,
the exceptions being Gujarati (Trupti Nissar and P. J. Mistry, p.c.) and Mangalore
Konkani (Lalita Dhareshwar, p.c.). We do not have any data from the other two
families.
25. Though the embedded predicate ɔn ‘go’ and the matrix predicate hi ‘be’ are
both [−transitive], the embedded subject is ergative case-marked. The reason for
such marking is due to the fact that postposition incorporation in Tibeto-Burman
languages increases the valence of the embedded predicate. Hence, the ergative
marking on the embedded subject. See chapter 4 for a detailed discussion.
26. See Masica (1991) for a brief discussion of the syntax of Indo-Aryan languages.
27. Munda languages have the relative–correlative construction in which a relative
pronoun is used, and a question expression is also used as a relative pronoun in the
formation of relative clauses, as in Dravidian (see Anderson 2008).
28. Clefts are permitted in Dravidian, in some Tibeto-Burman languages and in Khasi
(Mon-Khmer). Amongst the Munda languages, Ho does not permit clefts, and we
do not have information on the other languages.
29. Emeneau (1956) uses the phrase “India as a linguistic area.”
3 L E X I C A L A NA P H O R S A N D P RO N O U N S I N S O U T H A S I A N
LANGUAGES
1. See Lust et al. (2000) for an exhaustive description and discussion of anaphors
and pronouns in fourteen different SALs encompassing four different language
families, a detailed introduction focusing on the issues and a questionnaire. For a
typology of anaphors, see Huang (2000). This chapter greatly benefited from the
papers in Lust et al. (2000).
2. We use the term verbal anaphor as a convenient label for the reflexive or reciprocal
in the absence of a better term. Martin Everaert (p.c.) observes that the term
‘anaphor’ implies referential status, which implies being an argument / DP-status.
318 Notes to pages 48–61
The evidence presented below, as Martin Everaert observes, shows “the limitations
of the standard Binding Theory, which is solely focused on arguments, allowing no
room for the concept of a verbal reflexiviser.”
3. The verbal reciprocal -ya- also functions as a group marker. We shall see in chapter 8
that the group marker -ya-, however, is obligatory when it occurs in sentences with
the comitative PP as head. See sentence (87) in chapter 8.
4. We provide an example from Toda (DR):
5. A vector verb such as vēs ‘to drop’ can also occur with the main verb, when a verbal
anaphor occurs.
6. We shall demonstrate later that a null form ( pro) is not permitted in the non-
nominative subject construction, as the verbal anaphor does not occur in such
constructions.
7. To the best of our knowledge, E. Annamalai was the first scholar to have made this
observation regarding such predicates in any Dravidian language.
8. In Lalitha Murthy and Subbarao (2000), we have transcribed the Mizo expression
for ‘self’ as māʔ. Some of the language consultants we interacted with in recent
years feel that ‘self’ should be transcribed as maʔ. Hence, in this work, we have
transcribed it as maʔ.
9. Reinhart and Siloni (2005: 404) argue that “The selection of se [the reflexive clitic in
French] reduces the verb’s ability to check accusative Case [(i) below from French].”
In Mizo, however, the DO needs to have its accusative Case assigned/checked and,
hence, their claim cannot be accepted for Mizo. The letter ti in the example below
stands for trace.
Jeani sei lave ti
John SE washes
‘Jean washes himself.’
10. In the Jowai and Rymbai dialects of Khasi too, the marker ya- functions as a
reciprocal marker as well as a group marker (Bareh 2007).
11. Also see sentence (87) in chapter 8 where ya- the verbal reciprocal, obligatorily
occurs as a group marker in sentences with the comitative PP as head in EHRCs in
Khasi.
12. In our system of analysis, -p- is marked as ‘-pA-’.
13. Evidence for Kazenin’s claim comes from Mizo (TB) in which the ergative marker
does not occur when the anaphor occurs in the direct object position, as the verb
is detransitivized. When the anaphor occurs in a non-direct object position, the
occurrence of the ergative marker is obligatory (cf. sentence (12) from Mizo).
14. An inchoative verb denotes the beginning of an action, state or event. Verbs such
as open, close, thicken, etc., in English are a few examples.
15. Such instances as in (32) are examples of Possessor Ascension as there are corre-
sponding sentences, as below, where the possessive pronoun occurs as a modifier
Notes to pages 62–76 319
16. For more discussion of functions of the VR kol..l and its relation to nominal marking
of coreference, see Lidz (2001a, b and c).
17. As quoted in Anderson (2007: 42).
18. Adjectives behave like verbs in Tibeto-Burman languages. They can take “tense,
aspect and mood suffixes” in Manipuri (TB) (Bhat and Ningomba 1995: 66). They
can take nominalizers and causative prefixes in Tenyidie (TB) (Koulie 2006: 15)
and in Mizo and Hmar (TB). In Kokborok (TB), “the superlative marker, which
normally occurs with adjectives, also occurs with the modal man ‘can’ and is
followed by the nominalizer nai in comparative constructions” (see Subbarao et al.
2010 for details).
19. For a more elaborate list of verbs with konu, see Subbarao and Lalitha Murthy
(2000: 227).
20. These are called intrinsic/inherent reflexive verbs in Germanic/Romance (Everaert,
ms).
21. Lidz (2001c) proposes to account for the distribution of the verbal reflexive in
Kannada in terms of the interaction between lexical aspectual structure and thematic
structure. According to him, there is a mismatch in linking between the verb’s
aspectual decomposition and thematic representation. This, he claims, provides
an explanation for the absence of the VR in NNS constructions, VR blocking
long-distance binding, etc. (2001: 351).
22. Cf. Hopper and Thompson (1980), Kemmer (1993), for similar views.
23. In (56c), in place of the dative subject lili-lā ‘Lili-dat,’ a genitive subject lili-tsyā
‘Lili-gen’ can also occur (Trupti Nissar, p.c.). Lalita Dhareshwar prefers āplya-vərr
‘self-obl-on’ to svətāhā-vər ‘self-obl-on’ with the genitive subject in (56c).
24. Davison (2004: 155) proposes a theory-internal argument, and we quote: “To avoid
this constraint on nominative reflexives, the internal argument or theme [that is,
rām in our (59)] may rise to Spec/TP for the EPP [Extended Projection Principle]
feature, while the dative argument stays within the VP. The reflexive argument is
therefore dative in [(our (59)], bound by the nominative subject.”
There needs to be further investigation into why a nominative case-marked
non-anaphor (R-expression or pronoun) is permitted as theme in the DSC, while a
nominative case-marked anaphor is not.
25. Thanks to Rajesh Bhatt (p.c.) for helpful suggestions.
26. Our language consultants insist on having an aspirated -th in thõ ‘by’ in (69) in
contrast to Bhatia’s (2000: 661) transcription without an aspirated -th in tõ ‘by’ in
(68).
27. A. Usha Rani (p.c.) brought to my notice the sentence below, in which the possessive
anaphor tana ‘self’s’ in a finite complement clause is coindexed with the subject of
the matrix clause. Such coindexing is a violation of Principle A, and is similar to
the violation found in Marathi in (76).
320 Notes to pages 76–103
Telugu (DR)
amei [dongaj tanai,∗ j parsu dongilincu- kon i pōyēd.u]
she thieves self’s purse steal- self ben cpm went
ani andi
quot said
‘Shei said that the thievesj stole heri,∗ j purse.’
28. Subbarao (1971) is the first study that discussed long-distance binding in any SAL.
Davison (2001) discusses long-distance binding in Hindi-Urdu in detail. The papers
in Lust et al. (2000) discuss long-distance binding in various SALs.
29. Recall that Indo-Aryan does not have a robust verbal anaphor and, hence, no
evidence can be adduced from Indo-Aryan.
30. The permissibility of occurrence of the accusative case marker in (82) is due to the
fact that in select languages such as Bangla (IA), Bodo (TB), Tamil and Malayalam
(DR), a dative predicate may assign an accusative case marker to the theme in
DSC – cf. (146) from Bangla, (155) from Malayalam, (157) from Tamil and (163)
from Bodo in chapter 5.
31. In the Minimalist Program (Chomsky 1995b) an ECM subject raises to the Spec
Agro for Case checking (Lasnik 1999). Hence, the anaphor occurs in the governing
category of its antecedent.
32. We shall show that a genitive case-marked subject in Bangla permits an accusative
case-marked theme as in (146) in chapter 5, while Hindi-Urdu does not permit an
accusative case-marked theme when the subject is dative case-marked, as in (147)
in chapter 5. The two facts concerning (82) above and (146) in Bangla in chapter 5
might be intrinsically connected. Only further research can shed light on this issue.
33. Rajesh Bhatt (p.c.) speculates that all such cases obligatorily involve a simplex
form.
34. Kidwai (2000) attempts to analyze binding facts of Hindi-Urdu as instances of
XP-adjunction, the discussion of which is beyond the scope of the present work.
35. We did not provide a detailed discussion of reciprocals in SALs. For a detailed
discussion, see Everaert and Subbarao (2011) and Subbarao and Everaert (2011).
4 CA S E A N D AGR E E ME NT
1. An NPI is a syntactic item that strictly requires a negative licensor in the local
domain where the licensing takes place during the course of derivation. See Kumar
(2006) for a detailed discussion of the syntax of the NPI in Hindi.
The NPI is incorporated in Khasi (Mon-Khmer) and in Kokbodok (TB) too,
which we shall discuss later.
2. Thanks to Rajesh Bhatt for suggesting this classification into two subsets.
3. P. J. Mistry (p.c.) labels the marker -(n)e on the subject in (11) as an agentive marker
and points out that there is dialect variation between -ne and -e.
4. See Das (2006) for a discussion of agreement in Hindi-Urdu and its major dialects.
5. Thanks to Rajesh Bhatt for his suggestion in this regard.
6. An example of DO incorporation in Khasi is given below. It may be mentioned
that the function of the clitic -sa in the example is not fully understood. It is
not discussed in any Khasi grammar. We have labeled it as “pm,” where “pm” stands
Notes to pages 103–107 321
for “procrastination marker.” Thanks to Juanita War (p.c.) for pointing out that sa
literally means ‘by and by’ or ‘later.’ We may hypothesize that sa, originally an
adverb, is subsequently grammaticalized as a procrastination aspect marker.
Khasi (Mon-Khmer)
ŋi- n- sa- thēd do? ma- ŋi-
1p- fut- pm- buy meat nom- 1p-
aux 1 aux 2 v e r b o b j e c t subject
‘We will buy meat.’
7. For some speakers the occurrence of i-koma ‘for you’ and the pronominal clitic -ce
‘2 singular’ are in complementary distribution (Vanlal Englien, p.c.).
8. Though Rymbai, Jowai and Pnar are generally called dialects of Khasi, the work
of Bareh (2007) demonstrates that there are many syntactic differences between
these “dialects” and Khasi. The adverb yaw ‘market’ in the underived structure is
incorporated into the verb in (i) in Rymbai (Mon-Khmer). Note that it occurs to
the right of the verb stem and is followed by ko, the 3rd person feminine, singular
agreement marker.
Rymbai (Mon-Khmer)
(i) ka- daw lε- yaw- ko
3s,f fut go- market- 3s,f
‘She’ll go to the market.’
(Bareh 2007)
In Pnar (Mon-Khmer) too, incorporation of a DP into the verb takes place. Sentence
(ii) exhibits no incorporation, while sentence (iii) incorporates the DP ja-i
‘rice-
hot.’ Note that the subject agreement marker occurs to the right of the verb in
(ii) while the subject agreement marker -u occurs to the right of V-bar which is
the verb and the first conjunct of the incorporated DP. The first occurrence of the
preposed conjunction wa does not occur in (iii) due to incorporation.
no incorporation with a conjoined dp
Pnar (Mon-Khmer)
(ii) u- jɔni bam- ui [DP w(a)- i ja- i
w(a)- u-
3s- John eat- subj agr and dm rice- hot and- m,s-
sa-pe
DP ]
fruit-mango
‘John ate hot rice and a mango.’
Incorporation of a part of the DP into the verb
(iii) u- jɔni [v̄¯ bam- ja- i
- [DP ui v̄¯ ] w(a)- u- sa-pe
DP ]
3s- John eat- rice- hot- subj agr and- m,s- fruit-mango
‘John ate hot rice and a mango.’
(Curiously Bareh, p.c.)
It is significant that only one conjunct is incorporated into the verb which is a
violation of subjacency. Note that the conjunction marker in Pnar is wa and, when
followed by a vowel, sandhi takes place and the vowel -a gets elided. The particle -u
322 Notes to pages 109–127
is a diminutive marker which occurs with items which are very dear to the speaker,
a feature found in Mon-Khmer and some Tibeto-Burman languages.
9. Recall that with some verbs and stative adjectives in Hmar, the occurrence of the
verbal reflexive is obligatory. See chapter 3 for a discussion.
10. Vanlal Englien (p.c.) informs me that, in his dialect, it is not possible to have the
verbal reflexive in (44). Such prohibition could be due to the fact that adposition
incorporation into the verb makes the verb [+transitive] and the verbal reflexive/
reciprocal is [−transitive]. Hence, the verbal reflexive/reciprocal is prohibited as it
is not compatible with the transitive nature of the host verb.
11. The absence of LDA in finite clauses can be explained in recent theoretical terms as
follows: “[T]he embedded clause is a finite CP and given that CPs constitute strong
phases [in the sense of Chomsky 2001] and the Phase Impenetrability Condition
(PIC) of Chomsky (1999), the embedded finite T0 and the other TP-internal material
would be invisible to the matrix finite T0 ” (Bhatt 2005: 17). Since the T0 of S2 is
invisible, the probe–goal relationship cannot be established. A similar argument
can be extended to the antecedent–anaphor relationship too.
12. Sentences (62) and (63) are slightly modified versions of the original. We have used
the past auxiliary as it reflects phi-(PNG) features, unlike the present tense marker.
13. For some speakers in Hindi-Urdu too, agreement on the infinitive might take place
with the embedded object, and the matrix verb might exhibit default agreement. The
default agreement in Hindi-Urdu is the masculine singular form. frij ‘refrigerator’
is in feminine singular and the embedded infinitive cal-nı̄ ‘to start’ is also in feminine
singular. The matrix verb šurū ho jāegā ‘will start’ exhibits default agreement –
that is, masculine singular.
14. Recall that possessor agreement takes place in Hmar (TB) too. In Hmar, the pronom-
inal clitic of the possessor of the DO occurs to the right of the subject agreement
marker (see (32)).
It is worth mentioning here that a possessor that ‘subcommands’ an anaphor in
Hindi-Urdu (IA) can be an antecedent to an anaphor too (see the cases involving
subcommand in chapter 3).
15. Raising is normally to the next higher clause. If raising takes place to a higher
clause, skipping one sentence in-between, it is called Superraising.
16. Sentence (79) is a case of incorporation of the comitative postposition pui ‘with’
in the verb (see Subbarao 2001: 467). (See section 4.3, sentences (47) and (48), for
postposition incorporation in Paite [TB].)
17. Note that the agreement marker on the embedded verb in (87) and (88) is null (ø),
as the subject of the embedded clause is no longer present there; it moves to the
matrix clause, and shows up as object agreement clitic.
18. The glosses have been slightly modified. See Ramakrishna Reddy (1980, 1992a)
for more details.
Notes to pages 128–145 323
19. Note that -aŋ and -tur are allomorphs of the future tense marker; -aŋ occurs with
the matrix verb, while -tur occurs with the embedded verb.
20. See sentences (16) and (17) in Mizo (TB) in appendix 3 of chapter 8 on the Website.
In the modification of subject in EHRCs, the embedded subject in (16) and (17)
is not overtly present (indicated by ø). The embedded verb consequently does not
carry subject agreement marker (indicated by ø). This provides further evidence for
the Chomskyan hypothesis that case and agreement are intrinsically linked.
5 NON-NOMINATIVE S UBJECTS
1. This chapter derives its inspiration in content and spirit from the work by the author
in collaboration with Peri Bhaskararao on NNSs in Telugu at ILCAA, Tokyo
University of Foreign Studies, Tokyo, Japan. Crucial analysis and most of the
Telugu examples in this chapter have been taken from Subbarao and Bhaskararao
(2004); hence the source is not always cited. This chapter greatly benefited from
the two volumes on NNSs edited by Peri Bhaskararao and Subbarao (2004) and the
volume edited by M. K. Verma and K. P. Mohanan (1990) on Experiencer Subjects
in South Asian Languages.
2. In sentences involving Exceptional Case Marking, the subject of the embedded
sentence is accusative case-marked by the matrix verb.
3. In Rabha (TB) the accusative marker occurs with the subject to indicate need, or
want.
Rabha (TB)
(i) aŋ- a- be taka- be lagi ca- be
I- acc- emph money- emph want not- emph
‘I really do not want any money at all.’
Manipuri (TB)
(ii) əy- bu- nə barton təw- wi
I- acc- foc invitation get- progr
‘I was invited, but not others.’
4. See Mahajan (2000), Kachru (2006), for a detailed discussion of the Hindi
passive.
5. Tiwa is a Tibeto-Burman language spoken in Assam in the northeastern part of
India. The following examples are illustrative.
Embedded subject is accusative case-marked:
6. vuchān ‘see.imperf’ may move leftwards to the right of chus-an ‘pres.1 m,s’ (Sadaf
Munshi, p.c.).
7. Sadaf Munshi (p.c.) provided further data which demonstrate that the occurrence
of the 2nd person agreement clitic of the derived object as a pronominal clitic on
the matrix verb yields grammatical sentences, as (i) and (ii) illustrate.
(i) bai chusi - athj- tsj dohay bat khyav- ān
I.nom pres.1 m,s- 2 pn- you.nom daily rice.acc eat- imperf
vuchān
see.imperf
‘I see you (m,s) eating rice every day.’
8. The fact that the clitic of the embedded subject occurs on the matrix verb in Kashmiri
(43b) and Dumi (44) shows that this is a case of long-distance agreement.
9. In Telugu (DR) -u is a bound morpheme (a clitic) that performs the function of
a conjunction marker, an emphatic marker, universal quantification marker and a
negative polarity marker. The occurrence of -u is phonologically conditioned, and
its form depends on the preceding vowel of the host (see Dilip, Subbarao and
Umamaheshwar Rao 2011).
10. In Telugu (DR) -u also functions as a marker to impart quantificational interpreta-
tion.
11. In colloquial Hindi the expression mere ko ‘I (obl)-dat’ is often heard, though the
formal expression cited in grammatical descriptions is mujh ko / mujhe ‘I (obl)-dat.’
Since this sentence is a verbatim quote from a native speaker of Hindi, we did not
make any alteration in the sentence. Such usage of mere ko ‘I (obl)-dat’ in place of
the expression mujh ko / mujhe ‘I (obl)-dat’ indicates that there is a syntactic change
in progress. Example (56) was reattested by Rama Kant Agnihotri and several other
native speakers.
12. The verb exhibits agreement in number and gender with the nominative case-marked
noun phrase nı̄nā ‘Nina.’
13. The verb exhibits default agreement as there is no nominative case-marked noun
phrase in (83).
14. EPP in simple terms means that every sentence must have a subject. (See Lasnik
and Uriagereka 2005 for further details.)
15. Masica (1991) prefers the terms neutral (unmarked)/non-volitional (marked) to the
terms volitional and non-volitional that are generally used. Alice Davison (p.c.)
Notes to pages 168–192 325
agrees with Masica (1991) and points out that ergative is neutral between the
features [+/−volitional] while dative is invariably [−volitional].
16. Bayer’s sentence has been slightly modified. Our language consultant pronounces
the word for ‘went’ as gelam, and not as gεlam as Bayer transcribes it. Bayer
crosses out the word amar in (129). We have used the more familiar convention of
the parentheses marked by * to show that amar cannot be deleted.
17. Sentence (79) is a similar example from Hindi-Urdu (IA).
18. Dasgupta (2004: 131) points out that toma.r, which is generally glossed as
‘you.gen,’ should be treated as an indirect case form. For a discussion on this,
see Dasgupta, Ford and Singh (2000).
19. Šotti-šotti is a reduplicated form.
20. Even when the patient or theme is in contrastive focus or under emphasis, the
accusative marker is not needed, as below.
Tamil (DR)
kumār- ukku yār- um- e teri- yā- du
Kumar dat who npi- also know- neg- 3 s
‘Kumar does not know anybody at all.’
21. Magier (1987: 192–193) clearly articulated that ko in Hindi does not “convey rela-
tional information” when it occurs with “direct objects,” but “follows a semantic
hierarchy of specificity and animacy that contributes to the overall salience of the
marked object noun.”
22. Thanks to Alice Davison for the formulation of this parameter.
23. See the papers in Bhaskararao and Subbarao (2004) for some discussion on this
issue.
24. Rajesh Bhatt (p.c.) supports Alice Davison’s contention.
25. Though AGR phrase is dispensed with in Chomsky (1995b), our analysis shows
that agreement does play a role in dative case assignment. Further, the fact that
quantifiers, emphatics, negative and negative polarity items are incorporated as
a part of the auxiliary in Tibeto-Burman languages and Khasi (see chapter 4
for an elaborate discussion) shows that the notion of AGR phrase needs to be
incorporated in the theory, and it cannot be dispensed with, as Subbarao (1998)
argued.
26. Tamil (DR), in contrast, has the dative–locative double subject construction
(Parameshwari Krishnamurthy and Rajeshwari, p.c).
27. The marker -ā also functions as a polarity question marker in Dravidian
languages.
28. The marker -a may also occur on all the other constituents in the sentence including
the verb. See Bhaskararao and Subbarao (in preparation).
29. Dakkhini (IA) has borrowed the Double Dative Subject construction (i) from Telugu
(DR). It is not permitted in standard Hindi-Urdu (IA), as in (ii).
Dakkhini (IA)
(i) us koi hāth koi cot.j lag- ı̄∗ i,j
he dat hand dat injury hit- pst f,s
‘He got hurt on his hand.’
326 Note to page 192
Hindi-Urdu (IA)
(ii) *us koi hāth koi cot.j lag- ı̄∗ i,j
he dat hand dat injury hit- pst f,s
‘He got hurt on his hand.’
The Dakkhini case shows that the phenomenon of inherent case assignment may
be borrowed from a member of a different language family.
We shall present below data that illustrate the Single and Double Dative Subject
constructions in Mangalore Konkani (IA) (Lalita Dhareshwar, p.c.).
s i n g l e d a t i v e s u b j e c t c o n s t r u c t i o n ( w i t h t h e w h o l e , ta ‘he,’
t h e p o s s e s s o r , i n t h e g e n i t i v e c a s e , a n d t h e p a r t , hātt ‘hand,’
in the dative case)
Mangalore Konkani (IA)
(iii) tagal- ya hātt- ā- ka lāg- l- ε
he.gen- obl hand- obl- dat hurt- perf- def agr
‘He got hurt on his hand.’
d o u b l e d a t i v e s u b j e c t c o n s t r u c t i o n ( w i t h t h e w h o l e , ta ‘he,’
i n t h e d a t i v e c a s e , a n d hātt ‘ h a n d ’ i n t h e d a t i v e c a s e )
(iv) tā- -kkā hātt- a- ka lāg- l- ε
he.obl- dat hand- obl- dat hurt- perf- def agr
‘He got hurt on his hand.’
However, it is possible for an adverb to occur to the right of the possessor that is
dative-case marked as in (iv).
6 COMPLEMENTATION
1. Complementation is a process in which a clause is embedded in a matrix (higher
clause) either in its subject or object position. Such embedding depends upon the
argument structure of the matrix predicate – that is, whether the matrix predicate
permits an embedded sentential complement or not. One set of predicates permits
subject complements, while another permits object complements depending on the
nature of the predicate itself.
2. As Alice Davison (p.c.) rightly points out, the FCs in (8) and (9) are “different
from complements” as the first complementizer performs reason function. For a
discussion, see 6.2.2.
3. There is no V2 phenomenon observed in embedded relative clauses and adverbial
clauses in Kashmiri. An explanation in terms of the absence of the mood phrase
in such clauses is provided in Bhatt (1999). See also sentences (5) and (6) in
appendix 6.1 on the Website for an explanation in terms of the etymology of the
complementizer.
4. In Hindi-Urdu (IA), the IC occurs as a disjunctive marker in embedded questions.
Hindi-Urdu (IA)
rekhā ko mālūm hogā ki karan ā rahā hai ki nahı̄˜
Rekha dat known will IC Karan come progr pres or not
‘Rekha would know whether Karan is coming or not.’
5. Recall that Marathi and Gujarati (IA) have another form of the FC. In Marathi it is
asə and in Gujarati it is εm (Masica 1991: 403).
6. As we shall see later, the IC ki ‘that’ of Hindi-Urdu (IA) is reanalyzed as a post-
nominal and post-sentential marker in Dakkhini (IA) and it performs a different set
of functions, such as a disjunctive marker, and as a dubitative marker patterning the
Dravidian complementizer -ō (Subbarao and Arora 1989; Arora 2004). Such a shift
in position, we argue in appendix 6.5 on the Website, is due to syntactic reanalysis.
7. To account for the phi-features of COMP in Juang, Patnaik (1999) proposes that the
quotative construction be treated as a control structure where the complementizer is
treated as the verb of a clause with PRO as its subject. This PRO in turn is coindexed
with the subject of the matrix clause.
8. Although, in Sinhala (IA) and Telugu (DR), the quotative occurs with a verb of
perception such as see, it performs the function of a conditional in Sinhala in (i) and
that of a reason marker in Telugu in (ii). Thus, sentences (i) and (ii) do not go against
Bayer’s claim.
Sinhala (IA)
(i) gihiŋ malli tamange kāmərē innəwa də
go.perf pple brother self.gen room.loc be (animate) q mkr
kiyəla balannə
quot look.inf
‘Go and see if younger brother is in his (own) room.’
(Gair 2003: 807)
328 Notes to pages 208–214
Telugu (DR)
(ii) pūlı̄su vād.u vastunnād.u ani cūs- i donga pāripōyed.u
policeman is coming quot see- cpm thief ran away
‘Having seen that the policeman was coming, the thief ran away.’
9. In Bangla (IA), the complementizer je is compatible with a question word inside
the complement clause in specific cases. Probal Dasgupta (p.c.) observes that it
is the explicit and intricate information in the embedded clause that forces an
interpretation of the ‘fact’-type rather than the ‘indirect question’-type and this is
what triggers the je option.
lɔta nije- o janto, ama- ke janiye diyechilo je piknik-e
Lata herself- also knew me- dat informed that picnic-loc
kara ašbe ar kara ašbe na
who will come and who will come not
‘Lata herself also knew, she informed me too as to who will come and
who will not to the picnic.’
(Probal Dasgupta, p.c.)
10. In Sanskrit (IA), the embedded clause with the IC yat ‘that’ occurs to the right of
the VP of the matrix clause. For example:
vij
ān- āmy’ aham yat [prān.- o
perceive- 1s I.nom that breath- m,s,nom m,s,nom
brahma]
Brahman neuter,nom/acc
‘I understand that breath is Brahman.’
(Chandogya Upanishad 4,10,6, cited by Speijer
1886 [1980]: 359)
languages with fc
Telugu (DR)
(i) pillalu vaccēr- ā ani amma ad.igindi
children come- pol q FC mother asked
‘The mother asked whether the children had come.’
languages with ic
Hindi-Urdu (IA)
(ii) us ne mujh se pūchā ki kyā tum āoge
he erg me asked IC pol q you come.fut
‘He asked me whether you will come.’
(Davison 2008: 181)
12. Thanks to Suranjana Barua and Shukla Basu for confirming the Assamese and the
Bangla data, respectively.
Notes to pages 215–252 329
13. The anaphor swatāhā-lā ‘self–acc’ requires a strict clause-mate condition, and
hence it is permitted if and only if the “antecedent is within the same clause” in
which it occurs (Pandharipande 1997: 210).
14. The canonical position of the complement clause in verb-final languages is the
pre-verbal argument position.
15. In object complement clauses, the complement clause occurs in the ‘object position’
of the matrix clause, and it is called the ‘canonical position’ – that is, the position
in which a complement occurs before it is moved to some other position.
16. By Sentence-Lifting, we mean an embedded complement clause is moved from its
canonical clause-internal position and is placed to the left of the matrix sentence.
The complementizer does not occur with the embedded clause in such cases. It was
John Robert Ross who proposed the term ‘Sentence-Lifting’ (Ross 1973).
17. We labeled -ı̄ as ‘comp’ as it links the main clause with embedded complements.
18. Thanks to Probal Dasgupta (p.c.) for this summary. Dasgupta points out that if the
notion of zero complementizer is invoked, the facts concerning Wide and Narrow
Scope in Bangla can be neatly explained (see Dasgupta 2007).
19. Sentence (133) is not ungrammatical, though the embedded S occurs clause-
internally. The reason for the occurrence of a clause with an FC clause-internally
in languages with an FC is that the FC functions like a case marker shielding its
clause from the CRP.
20. We have slightly modified Dasgupta’s suggestion by replacing his ZERO NP cor-
relate by null DEM. The idea of having a null DEM is plausible as Bangla, like the
other SALs, is a pro-drop language. Under this analysis, there is a zero comple-
mentizer and a null DEM.
21. Taking such facts concerning scope marking in Bangla (IA) into consideration,
Simpson and Bhattacharya (2003: 127) argue that “Bangla actually has obligatory
wh-movement from a basic SVO word order,” and such movement is overt in
Bangla. We do not wish to go into the details as it is beyond the scope of the present
work. For further information, see Simpson and Bhattacharya (2003).
22. Thanks to Rajesh Bhatt for bringing this important paper to our notice.
23. According to Koul and Wali (2006: 144), Kashmiri also has the question marker
-ā which is “added to the finite predicate at the end of all inflections.” It also has a
question marker k’ā which occurs in the initial position, just as in Hindi-Urdu and
Punjabi.
24. Davison (p.c.) rightly observes that: (i) FOFC permits such sentences, as discussed
in Biberauer et al. (2009: 13); and (ii) Rajeshwari Pandharipande (1997) cites such
examples as grammatical. However, Lalita Dhareshwar (p.c.) finds such sentences
ungrammatical. We leave this open for further research.
25. See Biberauer et al. (2009) for a detailed discussion.
26. According to Biberauer et al. (2009: 5), categorial non-distinctness, it appears,
includes categories such as VP, DP, PP, C, inflected auxiliaries and verbs. They
admit that “the notion of ‘categorial distinctness’ is by no means ‘straightforward.’”
17. The nominative theme in a dative subject construction may be modified in Hindi-
Urdu (Rajesh Bhatt, p.c.) as in (i) below, and Punjabi (Om Prakash Arora,
p.c.).
Hindi-Urdu (IA)
(i) rām ko lagı̄ cot. bahut gahrı̄ hai
Ram dat get hurt.perf pple injury very deep is
‘The injury that Ram had is quite deep.’
23. Some speakers accept sentences in which the IO bacce ‘children’ is modified with
the causative (ditransitive) verb par.hānā ‘to teach’ in Hindi-Urdu (IA).
Hindi-Urdu (IA)
[shaguftā ke urdū par.hā- ye hue] baccei
Shagufta gen Urdu teach perf pple m,p children m,p
‘The children (who were) taught Urdu by Shagufta.’
24. Hindi and Bangla, Assamese and Oriya (IA), however, permit sentences such as
(i) and (iii) in which a locative PP is modified. Such occurrence of locative PP
modification is highly contextualized as the ungrammaticality of (ii) from Hindi-
Urdu and of (iv) from Bangla shows.
Hindi-Urdu (IA)
(i) øi mehandı̄ rac- e hāthi
paste made of a type of leaf put- perf pple hands
‘Hands in which the ‘mehandi’ paste is put.’
332 Notes to pages 287–289
Bangla (IA)
(iv) *thole rakh- a jayga
bag keep- perf pple place
‘The place in which the bag is kept . . . ’
(Shukla Basu, p.c.)
How productive the examples such as (i) and (iii) are, needs to be worked out.
25. The right kind of example for the comitative should be
as ‘the thief’ is the head of the comitative PP and not ‘the policeman.’ Cf. 8.6 for
further discussion.
26. Based on evidence from seven IA languages, in Subbarao (1984b), it was shown
that Keenan and Comrie’s NPAH cannot be maintained to be a universal, and there
are exceptions to it. For example, in Hindi-Urdu, Punjabi, Bangla and some other
Indo-Aryan languages (except Marathi, Oriya and Nepali), the Oblique Object (OO)
cannot be relativized while the object of the genitive can be, which is lower than
the OO in the hierarchy.
27. However, in Marathi (IA), it is the participial strategy that is “preferred.” Accord-
ing to Pandharipande (1997: 74), “though Marathi allows the use of relative
participles . . . as well as relative clauses, speakers of Marathi show a clear prefer-
ence for participles over finite relative clauses for communicating the same meaning.
In this respect, Marathi is closer to the Dravidian languages as opposed to the Indo-
Aryan languages, since the speakers of the former show a preference for participles
over relative clauses.” This comment is equally applicable to standard Konkani and
Mangalore Konkani (IA) too (Lalita Dhareshwar, p.c.).
28. In Hindi-Urdu and some other IA languages, most of the intransitive verbs such
as kūdnā ‘jump,’ daur.nā ‘run,’ calnā ‘walk,’ do not permit the modification of the
subject. For example:
Hindi-Urdu (IA)
*āj subah dı̄wār kūd.ı̄ huı̄ lar.kı̄
today morning wall jump.perf pple girl
‘The girl who jumped the wall this morning . . . ’
Notes to pages 289–303 333
29. Recall that, in Indo-Aryan Hindi-Urdu, Punjabi, Bangla, etc., the subject is invari-
ably genitive case-marked, while in Sinhala and Dakkhini (IA) it is always nomina-
tive case-marked, and in Oriya (IA) it is either genitive or nominative case-marked.
30. In Telugu and other Dravidian languages, “when a time expression occurs in the
predicate of the matrix clause, the truth-value of the sentences with an affirmative
CP and negative CP is ‘nearly identical’. . . ” (Krishnamurti and Gwynn 1985: 194)
(see section 9.6.1 on the Website for a detailed discussion).
31. See Asher and Kumari (1997: 61–62) and Ramarao (2003) for a detailed discussion
of permissible sentences with other interpretations of the sociative.
32. The construction kustı̄ pat..tu, literally ‘wrestling [noun] catch,’ is a conjunct verb,
and the noun and the verb together impart the meaning of ‘wrestle.’
33. There are some exceptions to this generalization as in sentence (30a) above. The
precise characterization of the predicates permitting such cases in Hindi-Urdu and
other Indo-Aryan languages needs to be worked out.
34. The case marker does not occur when the DO or IO is modified. However, when an
OO is modified, the case marker with the internal head may occur.
35. Thanks to Rajesh Bhatt for a very helpful discussion on this point.
36. Kannada (DR) also does not permit the comitative PP as head in EHRCs.
Kannada (DR)
*hud.uga (ø-jote) banda sušı̄la
boy (ø-with) came.pst.adjr Susheela
‘Susheela, who the boy came with’
(Sridhar 1990: 58)
Note that gal ‘go’ in the perfect aspectual suppletive form of the verb stem vats is
similar to the suppletive form ga- of the verb stem jā ‘go’ in Hindi-Urdu (Lalita
Dhareshwar, p.c.).
38. An example from Dimasa (TB) is given below.
Dimasa (TB)
aŋ hathai- ha thaŋ- pha- lai- ya- ba masainjik
I market- to go- together- gp mkr / VREC adjr- nozr girl
ani besik
my daughter
‘The girl with whom I went to the market is my daughter.’
(Parichita Langthasa and Ron Kemprai, p.c.)
Note that a ŋ‘I’ in (i) does not carry an ergative marker though the verbal reciprocal /
group marker lai and the adverb pha ‘together’ are incorporated. The reason for
such non-occurrence is due to the fact a verbal anaphor “detransitivizes” a predicate
334 Notes to pages 304–308
and, hence, unlike an incorporated postposition, it does not have the capability to
increase the valence of a predicate. We can therefore label a verbal anaphor as a
“valence reducer.” See chapter 3 for further discussion.
39. A similar phenomenon of valence augmenting is observed in Hmar and Paite. See
section 4.3 on Postposition Incorporation in these languages.
40. The verbal reciprocal ki in (94) in Thadou functions as a group activity marker
just as the verbal reciprocal in Khasi and Manipuri. In Thadou, verbs such
as to sit, to walk, to swim, to crawl, to stand and to run require the occur-
rence of the verbal reciprocal / group activity marker ki when the comitative PP
heads an EHRC. Thanks to Pauthong Haokip for a helpful discussion on this
issue.
Note that it is due to incorporation of the postposition pi ‘with’ that the embed-
ded predicate ɔn ‘go’ is transitivized in spite of the presence of the group marker /
verbal reciprocal ki. Recall that the verbal reciprocal detransitivizes a predicate
(cf. chapter 3). In spite of the occurrence of the verbal reciprocal as a group
marker in the embedded predicate, the embedded subject carries the ergative
marker. We hypothesize that the effect of postposition incorporation neutralizes
the “detransitivization effect” of the verbal reciprocal and supersedes its effect.
This hypothesis, however, needs to be empirically tested with more data and
analysis.
41. The following example from Manipuri is illustrative where the subject is in the
plural. Sentence (15) from chapter 3 is repeated below.
Manipuri (TB)
(15) tomba- ga tombi- ga skūl- da cat- min- na- khre
Tomba- and Tombi- and school- to go- together- VREC- pst
‘Tomba and Tombi went to school together.’
42. Recall that postposition incorporation has a “transitivizing effect” on the host verb
that incorporates it. See the case of Hmar and Paite (TB) in chapter 4.
43. In Mising and Bodo (TB), a comitative PP can head an EHRC when the embedded
verb carries the incorporated comitative postposition, which in Mising is zon and
Bodo is phā, as in (i) and (ii), respectively.
Mising (TB)
(i) sita- kə bozar- pɔ g- zon- hom konə- dɔ ŋo- k brmə
Sita- gen bazaar- to go- com- nzr girl- def I- gen friend
‘The girl with whom Sita went to the movie is my friend.’
(Jugen Pegu, Shikha Dele and Darika Pegu, p.c.)
Bodo (TB)
(ii) sita- ya hātha- yau thaŋ- phā hinjāu- sā- yā aŋ-
Sita- gen bazaar- to go- com girl- dm- nom I-
ni binānāu
gen sister
‘The girl with whom Sita went to the movie is my sister.’
(Ganga Brahma, p.c.)
Notes to pages 308–311 335
The occurrence of the genitive postposition in (i) with the embedded subject in
Mising is due to superstratum influence of Assamese with which languages such
as Mising, Bodo, Rabha and Dimasa, spoken in Assam, have been in contact for
centuries.
44. This analysis is abstracted from Kumar and Subbarao (2005), and, hence, the source
of the text and each example is not mentioned.
45. There is another determiner DD1 which occures to left of the relative clause and it
is optional. DD1 is not relevant for our discussion here, and hence it is omitted.
46. A set of intransitive verbs and adjectives in Hmar carry the verbal reflexive marker
and such occurrence is lexically determined (see chapter 3 for details).
47. The personal pronoun kei ‘I’ has two agreement markers ka- and -min- both of
which are bound morphemes; ka- is subject agreement marker and -min- is object
agreement marker. For example:
Hmar (TB)
(i) kei- maʔ- cu- an sikul- aʔ lali- cu ka- hmu- aŋ
l- emph- def- erg school- at Lali- def 1 s- see/meet- fut
‘Only I’ll see/meet Lali at school.’
(ii) lali- n sikul- aʔ a- min- hmu- aŋ
Lali- erg school- at 3 s- me (DO) see/meet- fut
‘Lali will see/meet me at school.’
48. This section presents conclusions drawn on the basis of material presented in the
main text of the chapter and the appendices of this chapter in the web material.
Bibliography
Abbi, A., 1980. Semantic Grammar of Hindi: A Study of Reduplication (New Delhi:
Bahri Publications Private Ltd.).
1984. “The Conjunctive Participle in Hindi-Urdu,” International Journal of Dravidian
Linguistics, 13.2, 252–263.
1990. “Experiential Constructions and the ‘Subjecthood’ of the Experiencer NPs in
South Asian Languages,” in Verma and Mohanan (1990: 253–268).
1992. Reduplication in South Asian Languages: An Areal, Typological and Historical
Study (New Delhi: Allied Publishers Ltd.).
1994. Semantic Universals in Indian Languages (Shimla: Indian Institute of Advanced
Study).
2001. A Manual of Linguistic Field Work and Indian Language Structures (Munich:
Lincom Europa).
2006. Endangered Languages of the Andaman Islands (Munich: Lincom Europa).
Achumi, R., 2000. “Sentential Complementation in Sema,” M.Phil. dissertation, Delhi
University, Delhi.
Achumi, R., and K. V. Subbarao, 2005. “Sentential Complementation in Sema,”
paper presented at the International Conference on South Asian Languages and
Literatures-6, Osmania University, Hyderabad, India.
Agnihotri, R., 1992. “India: Multilingual Perspectives,” in N. D. Crawhill (ed.), Demo-
cratically Speaking (Capetown: National Language Project), 46–55.
2001. Half the Battle and a Quarter, CIIL, Foundation Day Lectures (Mysore: Central
Institute of Indian Languages).
Aissen, J., 2003. “Differential Object Marking: Iconicity vs. Economy,” Natural Lan-
guage and Linguistic Theory, 21, 435–483.
Alexiadou, A., E. Anagnostopolou and F. Schafer, 2006. “The Properties of Anti-
causatives Crosslinguistically,” in M. Frascarelli (ed.), Phases of Interpretation
(Berlin: Mouton), 187–212.
Amritavalli, R., 2000. “Lexical Anaphors and Pronouns in Kannada,” in Lust et al.
(2000: 49–112).
2004. “Experiencer Datives in Kannada,” in Bhaskararao and Subbarao (2004: I,
1–24).
Anagnostopoulou, E., and M. Everaert, 1996. “How Exceptional are Nominative
Anaphors? A Case Study of Greek,” in L. Nash, G. Tsoulas and A. Zribi-Hertz
(eds.), Actes du deuxiēme colloque langue et grammaire (Paris: Department of
Language Sciences, Paris-8), 19–32.
336
Bibliography 337
Anderson, G. D. S. (ed.), The Munda Verb: Typological Perspectives (Berlin and New
York: Mouton de Gruyter, 2007).
(ed.) 2008. The Munda Languages (London: Routledge).
Anderson, G. D. S., and K. D. Harrison, 2008. “Sora,” in Anderson (2008: 299–380).
Annamalai, E. 1997. Adjectival Clauses in Tamil (ILCAA, Tokyo University of Foreign
Studies, Tokyo, Japan).
2000. “Lexical Anaphors and Pronouns in Tamil,” in Lust et al. (2000: 169–216).
Arora, H., 2004. Syntactic Convergence: The Case of Dakkhini Hindi-Urdu (Delhi:
Publication Division, Delhi University).
Arora, H., and K. V. Subbarao, 2004. “Syntactic Change and Convergence,” in
Bhaskararao and Subbarao (2004: II, 25–48).
Asher, R. E., 1985. Tamil. Lingua Descriptive Studies 7 (Amsterdam: North-Holland).
Asher, R. E., and T. C. Kumari, 1997. Malayalam (London and New York: Routledge).
Aze, F. R., 1973. “Clause Patterns in Parengi-Gorum,” in R. L. Trail (ed.), Patterns
in Clause, Sentence, Discourse in Selected Languages of India and Nepal, vol. I
(Norman, OK: Summer Institute of Linguistics, University of Oklahoma), 235–
313.
Baker, M. C., 1988. Incorporation: A Theory of Grammatical Function Changing
(Chicago: University of Chicago Press).
1996. The Polysynthesis Parameter (New York and Oxford: Oxford University Press).
2001. Atoms of Language (New York: Basic Books).
2010. “Formal Generative Typology,” in B. Heine and H. Narrog (eds.), The
Oxford Handbook of Linguistic Analysis (Oxford: Oxford University Press),
285–312.
Baker, M. C., and J. McLoskey, 2007. “On the Relationship of Typology to Theoretical
Syntax,” Linguistic Typology, 11.1, 285–296.
Bal, B. K., 1990. “COMP and Complementation in Oriya and English,” Ph.D. thesis,
Central Institute of English and Foreign Languages, Hyderabad.
Barbora, M., 2001. “COMPs in Assamese: Implications in Second Language Acquisi-
tion,” Ph.D. thesis, Tezpur University, Tezpur.
Bareh, C., 2007. “Descriptive Analysis of the Jowai and Rymbai Dialects of Khasi,”
Ph.D. thesis, North Eastern Hill University, Shillong.
Barua, S., 2003. “Typological Characteristics of Kokborok,” Masters’ degree fieldwork
dissertation, Delhi University.
Bashir, E., 1988. “Topics in Kalasha Syntax: An Areal and Typological Perspective,”
doctoral dissertation, The University of Michigan, Ann Arbor.
2003. “Dardic,” in G. Cardona and D. Jain (eds.), Indo-Aryan (London: Routledge),
818–894.
Bayer, J., 1996. Directionality and Logical Form: On the Scope of Focusing Particles
and wh-in-situ (Dordrecht, Boston and London: Kluwer).
2001. “Two Grammars in One: Sentential Complements and Complementizers in
Bangla and other South Asian Languages,” in Bhaskararao and Subbarao (2001:
11–36).
2004. “Non-nominative Subjects in Comparison,” in Bhaskararao and Subbarao
(2004: I, 48–76).
Bayer, J., T. Bhattacharya and M. T. Hany Babu, 2007. Linguistic Theory and South
Asian Languages: Essays in Honour of K. A. Jayaseelan (Amsterdam: John
Benjamins).
338 Bibliography
1986. Knowledge of Language: Its Nature, Origins and Use (New York: Praeger).
1995a. The Minimalist Program (Cambridge, MA: MIT Press).
1995b. “A Minimalist Program for Linguistic Theory,” in Chomsky (1995a).
1999. “Derivation by Phase,” MIT Working Papers in Linguistics.
2000. “Minimalist Inquiries: The Framework,” in R. Martin, D. Michaels and
J. Uriagereka (eds.), Step by Step: Essays in Minimalist Syntax in Honour of
Howard Lasnik (Cambridge, MA: MIT Press), 89–155.
2001. “Derivation by Phase,” in M. Kenstowicz (ed.), Ken Hale: A Life in Language
(Cambridge, MA: MIT Press), 1–52.
Chomsky, N., and H. Lasnik, 1995. “The Theory of Principles and Parameters,” in
N. Chomsky, The Minimalist Program (Cambridge, MA: MIT Press).
Cinque, G., 2007. “A Note on Linguistic Theory and Typology,” Language Typology,
11.1, 93–106.
Clark, T. W., [1977] 1989. Introduction to Nepali (New Delhi: Heritage Book House).
Cole, P., 1987. “The Structure of Internally Headed Relative Clauses,” Natural Language
and Linguistic Theory, 5, 277–302.
Cole, P., and L. Sung, 1994. “Head Movement and Long-distance Reflexives,” Linguistic
Inquiry, 25, 355–385.
Cole, P., G. Hermon and C.-T. James Huang (eds.), 2001. Long-distance Reflexives.
Syntax and Semantics 33 (San Diego: Academic Press).
Comrie, B., 1978. “Ergativity,” in W. P. Lehmann (ed.), Syntactic Typology (Sussex:
The Harvester Press), 329–394.
1981. Language Universals and Linguistic Typology (Chicago: University of Chicago
Press).
Coopmans, P., 1983. “A Review of B. Comrie, Language Universals and Linguistic
Typology,” Journal of Linguistics, 19.1-2, 455–473.
Coupe, A. L., 2003. “The Mongsen Dialect of Ao: A Language of Nagaland,” Ph.D.
thesis, La Trobe University, Bundoora, Victoria.
Croft, W., 1990. Typology and Universals (Cambridge: Cambridge University
Press).
Cysouw, M., 2003. “Against Implicational Universals,” Language Typology, 7,
89–101.
Das, B., 2005. “PRO in Bangla,” term paper, M.Phil. course, Delhi University.
Das, P. K., 2006. Grammatical Agreement in Hindi-Urdu and its Major Varieties
(Munich: Lincom Europa).
Dasgupta, P., 1977. “The Internal Grammar of Compound Verbs in Bangla,” Indian
Linguistics, 38, 68–85.
1980. “Question and Relative and Complement Clauses in a Bangla Grammar,” Ph.D.
thesis, New York University, New York.
2003. “Bangla,” in Cardona and Jain (2003: 391–443).
2004. “Some Non-nominative Subjects in Bangla,” in Bhaskararao and Subbarao
(2004: I, 128–140).
2007. “The Ubiquitous Complementizer,” in Bayer et al. (2007: 163–174).
Dasgupta, P., A. Ford and R. Singh, 2000. After Etymology: Towards a Substantive
Linguistics (Munich: Lincom Europa).
Davison, A., 1981. “Syntactic and Semantic Indeterminacy Resolved,” in P. Cole (ed.),
Radical Pragmatics (New York: Academic Press), 101–128.
1988. “Operator Binding, Gaps and Pronouns,” Linguistics, 26, 181–214.
Bibliography 341
Dwivedi, V., 1994. Syntactic Dependencies and Relative Phrases in Hindi (Amherst,
MA: GLSA Publications).
Emeneau, M. B., 1956. “India as a Linguistic Area,” Language, 32.3, 3–16.
1974. “India as a Linguistic Area Revisited,” in Southworth and Apte (1974).
Everaert, M., 1991. “Contextual Determination of the Anaphor/Pronominal Distinc-
tion,” in J. Koster and E. Reuland (eds.), Long-distance Anaphora (Cambridge:
Cambridge University Press), 77–118.
1999. “The Morphosyntax of SE(-Reflexivization): Some Preliminaries,” paper pre-
sented at the seminar on Grammatical Models, Tilburg.
2001. “Paradigmatic Restrictions on Anaphors,” in K. Megerdoomian and L. A. Bar-el
(eds.), WWCFL 20 Proceedings (Somerville, MA: Cascadilla Press), 178–191.
Everaert, M., and E. Anagnostopoulou, 1997. “Thematic Hierarchies and Binding
Theory: Evidences from Greek,” in F. Corblin, D. Godard and J.-M. Marandin
(eds.), Empirical Issues in Formal Syntax and Semantics (Bern: Peter Lang),
43–60.
Everaert, M., and K. V. Subbarao, 2011. “Verbal and Nominal Reciprocals in South
Asian Languages (SALs) and Obligatory Order Reversal in the Nominal Anaphor
in Dravidian,” paper presented at the South Asian Languages Analysis Conference,
CIIL, Mysore.
Fox, B., 1987. “The Noun Phrase Accessibility Hierarchy Reinterpreted: Subject Pri-
macy or the Absolutive Hypothesis?” Language, 63.4, 856–870.
Gair, J. W., 1998. Studies in South Asian Linguistics: Sinhala and Other South Asian
Languages (New York and Oxford: Oxford University Press).
2003. “Sinhala,” in Cardona and Jain (2003).
Gair, J., and W. S. Karunatillake, 2000. “Lexical Anaphors and Pronouns in Sinhala,”
in Lust et al. (2000: 715–773).
Gair, J., and J. Paolillo, 1997. Sinhala (Munich: Lincom Europa).
Gair, J., and K. Wali, 1988. “On Distinguishing AGR from Agr: Evidence from South
Asia,” in Papers from Parasession on Agreement in Linguistic Theory (Chicago:
Chicago Linguistics Society), 87–104.
Gambhir, V., 1981. “Syntactic Restrictions and Discourse Functions of Word Order in
Standard Hindi,” doctoral dissertation, University of Pennsylvania, Philadelphia.
Geeta Devi, T., 2000. “Relative Clauses in Manipuri,” M.Phil. dissertation, Delhi Uni-
versity.
Gill, D., 2000. Riau Indonesian: A VO Language with Internally Headed Relative
Clauses (Leipzig: Max Plank Institute of Evolutionary Anthropology).
Giridhar, P. P., 1980. Angami (Mysore, India: Central Institute of Indian Languages).
1994. Mao Grammar (Mysore, India: Central Institute of Indian Languages).
Greenberg, J. H., 1966. “Some Universals of Grammar with Particular Reference to the
Order of Meaningful Elements,” in J. H. Greenberg (ed.), Universals of Language,
vol. II (Cambridge, MA: MIT Press), 73–113.
Gumperz, J. J., and R. Wilson, 1971. “Convergence and Creolization: A Case from the
Indo-Aryan/Dravidian Border,” in D. Hymes (ed.), Pidginization and Creolization
of Language (Cambridge: Cambridge University Press).
Gurtu, M., 1992. Anaphoric Relations in Hindi and English (Delhi: Munshiram
Manoharlal).
Hany Babu, M. T., 1997. “Functional Categories,” Ph.D. thesis, Central Institute of
English and Foreign Languages, Hyderabad.
Bibliography 343
Hasan, M., 2005. “Backward Control in Subzapuri,” term paper, M.Phil. course, Delhi
University.
Haspelmath, M., 1993. “More on the Typology of Inchoative/Causative Verb Alterna-
tions,” in B. Comrie and M. Polinsky (eds.), Causatives and Transitivity (Amster-
dam: John Benjamins), 87–120.
Haspelmath, M., and E. König (eds.) 1995. Converbs in Cross-linguistic Perspective:
Structure and Meaning of Adverbial Verb Forms – adverbial participles, Gerunds.
Empirical Approaches to Language Typology 13 (Berlin: Mouton de Gruyter).
Herring, S. C., 1991. “Nominalization, Relativization, and Attribution in Lotha, Angami,
and Burmese,” Linguistics of the Tibeto-Burman Area, 14.1, 55–72.
Hock, H. H., 1975. “Substratum Influence on (Rig Vedic) Sanskrit?” Studies in the
Linguistic Sciences, 5.2, 76–125.
1982a. “The Sanskrit Quotative: A Historical and Comparative Study,” Studies in the
Linguistic Sciences, 12.2, 39–85.
1982b. “Conjoined We Stand: Theoretical Implications of Sanskrit Relative Clauses,”
Studies in the Linguistic Sciences, 19.1, 93–126.
1991. Principles of Historical Linguistics (Berlin: Mouton de Gruyter).
2001. “Typology vs. Convergence: The Issue of Dravidian/Indo-Aryan Syntactic
Similarities Revisited,” in Bhaskararao and Subbarao (2001: 63–100).
2005. “How Strict is Strict OV?” in T. Bhattacharya (ed.), The Yearbook of South
Asian Languages and Linguistics (Berlin: Mouton de Gruyter), 145–164.
Holmberg, A., 2000. “Deriving OV Order in Finnish,” in P. Svenonius (ed.), The Deriva-
tion of VO and OV (Amsterdam: John Benjamins), 123–152.
Hook, P. E., 1973. Compound Verbs in Hindi (Ann Arbor, MI: University of Michigan).
1974. The Compound Verb in Hindi (Ann Arbor: Center for South and South-East
Asia Studies, The University of Michigan).
1976. “Is Kashmiri an SVO Language?” Indian Linguistics, 37, 133–142.
Hindi Structures: Intermediate Level. Michigan Papers on South and South-East Asia
16 (Ann Arbor, MI: Center for South and South-East Asia Studies, University of
Michigan).
1990. “A Note on Expressions of Involuntary Experience in the Shina of Skardu,”
BSOAS, 67, 77–82.
1991. “The Emergence of Perfective Aspect in Indo-Aryan Languages,” in E. Traugott
and B. Hein (eds.), Approaches to Grammaticalization, vol. II (Amsterdam: John
Benjamins), 59–89.
1996. “Kesar of Layul: A Central Asian Epic in the Shina of Gultari,” in W. Hanaway
and W. Heston (eds.), Studies in the Popular Cultures of Pakistan (Lahore: Sang-
e-Meel), 121–183.
1997. “Relative Clauses in Eastern Shina,” in P. Michalove, I. Hegedüs and
A. Manaster-Ramer (eds.), Indo-European, Nostratic, and Beyond: Festschrift
for Vitalij V. Shevoroshkin (Washington: Institute for the Study of Man),
140–154.
2001. “Where Do Compound Verbs Come From? (And Where Are They Going?)”
in Bhaskararao and Subbarao (2001: 101–130).
Hook, P., and O. N. Koul, 1996. “Kashmiri as a V-2 Language,” in V. S. Lakshmi
and A. Mukerjee (eds.), Word Order in Indian Languages (Hyderabad: Center of
Advanced Study in Linguistics, Osmania University), 95–105.
2004. “Case as Agreement,” in Bhaskararao and Subbarao (2004: I, 213–226).
344 Bibliography
Hopper, P., and S. Thompson, 1980. “Transitivity in Grammar and Discourse,” Lan-
guage, 56, 251–299.
Hornstein, N., 1999. “Movement and Control,” Linguistic Inquiry, 30.1, 69–96.
Huang, Y., 2000. Anaphora: A Cross-linguistic Approach (Oxford: Oxford University
Press).
Ichihashi-Nakayama, K., 1994. “On Dative Subject Construction in Nepali,” in
C. Genetti (ed.), Papers in Linguistics 6 (Santa Barbara: Department of Linguistics,
UC Santa Barbara), 41–76.
Jacquesson, F., ms. “A Dimasa Grammar.” Brahmaputra Studies database,
http://brahmaputra.vjf.cnrs.fr/bdd/MG/pdf/Dimasa Grammar-2.pdf.
Jaeggli, O., and K. J. Safir, 1989. The Null Subject Parameter. Studies in Natural
Language and Linguistic Theory (Dordrecht, Boston and London: Kluwer).
Jayaseelan, K. A., 1990. “The Dative Subject Construction and the Pro-Drop Parameter,”
in Verma and Mohanan (1990: 269–283).
1998. “Blocking Effects and the Syntax of Malayalam Taan,” in R. Singh (ed.),
Yearbook of South Asian Languages and Linguistics (New Delhi: Sage), 11–
27.
1999. Parametric Studies in Malayalam Syntax (New Delhi: Allied Publishers).
2000. “Lexical Anaphors and Pronouns in Malayalam,” in Lust et al. (2000: 113–
168).
2004. “The Possessor–experiencer Dative in Malayalam,” in. Bhaskararao and Sub-
barao (2004: I, 227–244).
2007. “The Argument Structure of the Dative Construction,” in E. Reuland,
T. Bhattacharya and G. Spathas (eds.), Argument Structure (Amsterdam and
Philadelphia: John Benjamins), 37–48.
Kachru, B., 1973. An Introduction to Spoken Kashmiri, part 1 (Urbana, IL: Department
of Linguistics, University of Illinois).
Kachru, Y., 1970. “The Syntax of ko-sentences in Hindi-Urdu,” Papers in Linguistics,
2.2, 299–316.
1979. “The Quotative in South Asian Languages,” South Asian Languages Analysis,
1, 63–77.
1980. Aspects of Hindi Grammar (New Delhi: Manohar Publications).
1981. “On the Syntax, Semantics and Pragmatics of the Conjunctive Participle in
Hindi-Urdu,” Studies in the Linguistic Sciences, 11.2, 35–49.
1990. “Experiencer and Other Oblique Subjects in Hindi,” in Verma and Mohanan
(1990: 59–75).
1993. “Verb Serialization in Syntax, Typology and Historical Change,” in M. K. Verma
(ed.), Complex Predicates in South Asian Languages (Delhi: Motilal Banarsidass),
115–133.
2006. Hindi (Amsterdam and Philadelphia: John Benjamins).
Kachru, Y., and T. K. Bhatia, 1977. “On Reflexivization in Hindi-Urdu and its Theoret-
ical Implications,” Indian Linguistics, 38.1, 21–38.
Kak, A., ms. “Complementation in Kashmiri,” University of Kashmir, Sri Nagar, India.
Kayne, R., 1994. The Antisymmetry of Syntax (Cambridge, MA: MIT Press).
Kazenin, K. I., 2001. “Verbal Reflexives and the Middle Voice,” in M. Haspelmath,
E. König, W. Oesterreicher and W. Raible (eds.), Language Typology and Language
Universals (Berlin: Mouton de Gruyter), 916–927.
Bibliography 345
Keenan, E. L., 1985. “Relative Clause,” in T. Shopen (ed.), Language Typology and
Syntactic Description, vol. II (Cambridge: Cambridge University Press), 141–
170.
Keenan, E. L., and B. Comrie, 1977. “Noun Phrase Accessibility and Universal Gram-
mar,” Linguistic Inquiry, 8.1, 63–99.
Kellogg, S. H., 1938. A Grammar of the Hindi Language (Delhi: Oriental Books Reprint
Corporation).
Kemmer, S., 1993. The Middle Voice. Typological Studies in Language 23 (Amsterdam
and Philadelphia: John Benjamins).
Kevichüsa, M., 1996. “Relative Clause Formation in Angami (Tenyidie),” M.Phil. dis-
sertation, Delhi University.
2007. “Apects of Angami (Tenyidie) Syntax,” Ph.D. thesis, Delhi University.
Kevichüsa, M., and K. V. Subbarao, 1998. “The Relative Clause in Tenyidie (Angami
Naga),” South Asian Language Review, 8.2, 40–64.
Kidwai, A., 2000. XP-Adjunction in Universal Grammar: Scrambling and Binding in
Hindi-Urdu. Oxford Studies in Comperative Syntax (Oxford: Oxford University
Press).
Klaiman, M. H., 1977. “Bengali Syntax: Possible Dravidian Influence,” International
Journal of Dravidian Linguistics, 4.2.
1979. “On the Status of Subjecthood Hierarchy in Hindi,” International Journal of
Dravidian Linguistics, 8, 17–31.
1980. “Bengali Dative Subjects,” Lingua, 51, 275–295.
1983. “Bengali Conjunctive Participle Constructions,” in Proceedings of the Berkeley
Linguistic Society, 19, 138–147.
Koh, T. J., 2003. “Agreement in Ho,” Masters’ field methods course dissertation, Delhi
University.
Koh, T. J., and K. V. Subbarao, ms. “A Grammar of Ho.”
Koul, O. N., and K. Wali, 2006. Modern Kashmiri Grammar (Springfield, VA:
Dunwoody Press).
Koshal, S., 1979. Ladakhi Grammar (Delhi: Motilal Banarasi Dass).
Krishnamurti, Bh., 1975. “Verbs of Cognition in Telugu,” Osmania Papers in Linguistics,
1.1, 1–15.
2003. The Dravidian Languages. Cambridge Language Survey (Cambridge:
Cambridge University Press).
Krishnamurti, Bh., and J. P. L. Gwynn, 1985. A Grammar of Modern Telugu (Delhi:
Oxford University Press).
Kuolie, D., 2006. A Structural Description of Tenyidie: A Tibeto-Burman Language of
Nagaland (Kohima: Ura Academy).
Kuiper, F. B. J., 1974. “The Genesis of a Linguistic Area,” in F. C. Southworth and M. L.
Apte (eds.), “Contact and Convergence in South Asian Languages,” International
Journal of Dravidian Languages, 3.1, 135–153.
Kumar, R., 1998. “Internal Relative Clauses in Hmar and Mizo,” M.Phil. dissertation,
Delhi University.
2006. Syntax of Negation and Licensing of Negative Polarity Items in Hindi (London
and New York: Routledge).
Kumar, R., and K. V. Subbarao, 2005. “Postposition Incorporation and Internally Headed
Relative Clauses in Hmar,” in Y. Yadava, G. Bhattariai, R. R. Lohani, B. Prasain
346 Bibliography
Raina, A. M., 2002. “The Verb Second Phenomenon in Kashmiri,” in O. N. Koul and
K. Wali (eds.), Topics in Kashmiri Linguistics (New Delhi: Creative Books), 113–
128.
Ramakrishna Reddy, B., 1980. “Non-Dravidian Element in Manda Syntax: A Study in
Linguistic Convergence,” Osmania Papers in Linguistics, 6, 71–88.
1992a. “Syntactic Convergence in Central Indian Languages,” in M. K. Sastri (ed.),
Sri Korada Ramakrishnaiah Centenary Festschrift (Visakhapatnam: Sita Trust),
571–577.
1992b. “Predicate Agreement in Dravidian,” PILC Journal of Dravidic Studies, 1.1,
33–44.
Ramamurti, G. V., 1931. A Manual of the Sora (or Savara) Language (Madras: Govt.
Press).
Ramanujan, A. K., and C. Masica, 1969. “Toward a Phonological Typology of the
Indian Linguistic Area,” in T. E. Sebeok (ed.), Current Trends in Linguistics, 5:
Linguistics in South Asia (The Hague: Mouton), 543–577.
Ramarao, C., 1975. Telugu vākyam [“The Telugu Sentence” – in Telugu] (Hyderabad:
A. P. Sahitya Academy).
2003. Telugu velugulu [“Telugu Brightness” – in Telugu] (Hyderabad: Navodaya
Book House).
Ramasamy, K. 1981. “Correlative Relative Clauses in Tamil,” in S. Agesthialingom and
N. R. Nair (eds.), Dravidian Syntax. Annamalai University Publications in Lin-
guistics 73 (Annamalainagar: Annamalai University Department of Linguistics),
363–380.
Ray, T., 2000. “Lexical Anaphors and Pronouns in Oriya,” in Lust et al. (2000: 575–636).
Regmi, C. U., 2002. “Relativization in Nepali,” Ph.D. thesis, Tribhuvan University.
Reinhart, T., 2002. “The Theta System: An Overview,” Theoretical Linguistics, 28.3,
229–290.
Reinhart, T., and T. Siloni, 2005. “The Lexicon–Syntax Parameter: Reflexivization and
Other Arity Operations,” Linguistic Inquiry, 36.3, 389–436.
Riemsdijk, V. H., 2006. “Free Relatives,” in M. Everaert and V. H. Riemsdijk (eds.),
The Blackwell Companion to Syntax, vol. II (Oxford: Blackwell), 338–382.
Rizzi, L., 1997. “The Fine Structure of the Left Periphery,” in L. Haegeman (ed.),
Elements of Grammar (Dordrecht: Kluwer), 281–283.
Rosenbaum, P. S., 1967. The Grammar of Predicate Complement Constructions (Boston,
MA: MIT Press).
Ross, J. R., 1973. “Slifting,” in M. Gross and M. Schützenberger (eds.), The Formal
Analysis of Natural Languages (‘s Gravenhage, Holland: Mouton and Company),
133–172.
Sarju Devi, T., 2006. “Embedded Clauses in Manipuri,” doctoral thesis, Delhi University.
Sarju Devi, T., and K. V. Subbarao, 2002. “Reduplication and Case Copying: The Case of
Lexical Anaphors in Manipuri and Telugu,” Linguistics of the Tibeto-Burman Area,
25.2 (Fall), 47–72; also in R. Laury, G. McMenamin, S. Okamoto, V. Samiian and
K. V. Subbarao (eds.), Papers in Honor of P. J. Mistry (New Delhi: Indian Institute
of Language Studies, 2003).
Savio, D., 1991. “WH-questions in Tamil,” CIEFL Occasional Papers in Linguistics, 3,
55–67.
Saxena, A., 1985. “Reflexivization in Hindi: A Reconsideration,” International Journal
of Dravidian Linguistics, 14, 224–237.
Bibliography 351
356
Author index 357
Austro-Asiatic, 1, 9, 11, 15, 18–19, 38, 48, 55, Kharia, 10, 25, 38–39, 44–45, 48, 57–59,
81, 94, 114, 125, 134, 300, 313 63–64, 81, 142, 144, 158–159, 193,
Mon-Khmer, 1–2, 7–11, 15, 18, 21, 33, 42, 195, 200–201, 294, 346, 348
44–45, 48, 56, 59–61, 64, 81, 84, 91, Mundari, 10, 45, 48, 57, 61, 126–127,
102, 106–108, 122, 125, 132, 137, 142, 193, 195–196, 202, 207, 347
144, 149, 159, 193, 196–197, 212–213, Santali, 10, 39, 45, 48, 57–58, 61,
217–220, 222, 249, 263, 265, 271, 276, 105, 132, 193, 195, 202–203,
279, 294, 299–302, 304–305, 313–314, 345–347
316–317, 320–321 Savara, 10, 39, 107, 348
Khasi, 7–9, 12, 17–22, 29, 33, 42, 44–45, Sora, 10–13, 39, 48, 63–64, 94, 102–104,
48, 55–56, 59–61, 64, 81, 84, 91, 102, 107–108, 111–114, 132–133, 313, 336,
106–108, 113, 122, 125, 132, 137, 348
141–142, 144, 149, 159, 193, 195–197, Dravidian, 1–2, 9, 11–12, 15–16, 18, 21,
212–214, 217–220, 222–223, 249, 265, 23–24, 26–28, 31, 33–37, 40–41,
276–277, 279, 294, 301–302, 304–306, 44–47, 50, 54–55, 60–62, 65–66, 74,
308, 313–318, 320–321, 325, 329, 333, 79, 90–91, 94–95, 103, 108, 114, 121,
336, 345, 347, 352–353 124, 126–128, 133–134, 136–137, 140,
Pnar, 7, 9, 17, 21, 107, 314, 321 142, 149, 151–154, 158–159, 180,
Rymbai, 9, 17, 21, 107, 314, 318, 321, 187–188, 193, 195–197, 201–203, 206,
336 218, 241, 244, 248, 257, 259–260,
Munda, 1, 10–12, 15, 18, 25, 27–28, 33, 262–264, 266–271, 274–279, 282, 284,
38–40, 42, 44–45, 48, 57–64, 66, 78, 286–289, 291–294, 299–300, 311,
81, 91, 94, 102–105, 107–108, 316–318, 325–327, 330, 332
111–114, 122, 125–128, 132–134, Kannada, 9, 34–36, 51, 55, 60–62, 65,
141–142, 144, 158–159, 188, 193–197, 74–75, 83, 88, 97, 121, 135, 137,
200–203, 205–207, 245, 249, 262–263, 150–154, 156, 178, 188–190, 216, 248,
265, 271, 274, 276–279, 294, 300–301, 257, 259, 274, 276, 290–291, 311, 319,
312, 317 332, 335, 345, 347, 350
Ho, 1–2, 5, 10–11, 13, 19–20, 22–23, 25, Malayalam, 8–9, 27, 34, 36, 44–45, 50–51,
27, 30–31, 33, 44–45, 48, 50, 53–54, 70–72, 74, 80, 82, 84–85, 87, 90–91,
57–58, 60–63, 67, 69, 71, 75–76, 78, 95, 134, 137, 139, 144, 152–153, 162,
81, 86, 91, 93, 95, 101, 103–105, 115, 171, 173, 175–178, 184–185, 187,
119–120, 125, 135, 139, 143, 145, 189–190, 198, 202, 229, 248, 257, 274,
149–151, 161–162, 165–166, 173, 179, 276, 289, 291–293, 301, 305, 317, 336,
193–197, 201–203, 205, 209–211, 213, 343, 346–347
217, 234, 237, 244, 247–248, 257, 262, Manda, 9, 126–128, 165, 348
276, 291, 299, 303–305, 309–310, 313, Tamil, 8–9, 23, 36, 52, 70–71, 97, 121,
316–317, 319, 326, 329, 331–332, 123, 134, 139, 171, 173, 175–178, 198,
335–340, 342, 344, 346, 349, 352–353 229, 248, 270, 275, 284, 290, 301, 305,
Juang, 10, 44–45, 48, 195–196, 207, 327, 311, 324–325, 336, 345, 348–349,
348 351
359
360 Language index
Swat-Dir Kohistani, 9, 248 Manipuri, 7, 10, 28, 37, 44–45, 54–55, 59,
Torwali, 9, 248 81–82, 89–90, 93, 107–108, 137–138,
Tibeto-Burman, 1, 7, 10–11, 13–16, 18, 21, 140, 151, 159, 202, 210–212, 216, 218,
23, 25, 27–28, 31–33, 37–41, 44–45, 259, 261–262, 278, 295, 302–306, 308,
47–48, 52, 59–61, 65–66, 72, 80, 91, 319, 323, 333–334, 337–338, 341,
94–95, 102–106, 108–109, 111, 114, 348–349, 352
120, 122, 125, 132–135, 137, 140, 149, Mao, 10, 248, 341
151, 159, 177, 188, 193, 195–196, 201, Mising, 10, 21, 334n.43, 348
203, 206, 218, 241, 244, 248, 263–265, Mizo, 7, 10, 15–16, 27, 45, 47–48, 52–53,
268–269, 271, 274, 276, 278–279, 290, 59–60, 62, 65, 68, 72, 80–83, 86, 91,
294–295, 298–300, 302–304, 306–307, 95, 103, 106–107, 109, 111, 114, 120,
312, 315–317, 319, 321, 323, 325, 122, 125–126, 128–133, 135–137, 140,
329 145, 149, 202, 218, 246, 248–252, 278,
Ao, 10, 44–45, 55, 59, 248, 302, 339, 348 294–297, 302–303, 305–306, 308,
Bodo, 10, 21, 23, 45, 70–71, 107–108, 137, 318–319, 322, 329, 344–345, 348, 352
139–140, 147, 151, 159, 173, 177, 248, Paite, 10, 27, 48, 95, 102–103, 106, 109,
276, 334n.43, 337 111, 122, 132, 149, 278, 297, 303, 322,
Dimasa, 10, 303, 333n.43, 334, 343 333–334
Dumi, 10, 123, 145–146, 324, 353 Rabha, 10, 23, 28, 45, 55, 108, 140, 159,
Garo, 10, 151, 248, 338 218–219, 223, 259, 276, 323, 334, 352
Hmar, 10, 16, 21, 27, 45, 49, 60, 62, 64–66, Sangtam, 7, 10, 269–270, 295–296, 298,
68, 72, 78, 88, 91, 93, 95, 102–103, 302–307
105–107, 109–111, 114, 122, 125, Sema, 7, 10, 20–21, 25, 204–205, 216,
128–130, 132, 135, 137, 141, 149, 202, 234–235, 237, 276, 295, 298–299, 308,
264–265, 270, 278, 294–297, 302–303, 335, 339–341, 343, 346–351
305–306, 308–312, 315, 319, 321–322, Tenyidie (a.k.a. Angami), 7, 10, 20–22, 28,
333, 334n.48, 344, 347 37, 44–45, 55, 59, 265, 267–268, 276,
Kham, 10, 111, 139, 151, 177, 353 295, 302, 304–308, 316, 319, 343–344,
Kokborok, 10, 13, 15, 33, 42, 45, 108, 351
137–138, 148–150, 159, 193–194, Thadou, 10, 16, 38, 45, 48, 103, 128,
202–203, 248, 315, 319, 336, 346, 352 131–132, 135, 137, 244–245, 278, 294,
Konyak, 7, 10, 295, 298, 302, 306, 347 297, 303–306, 308, 313, 333n.40, 351
Ladakhi, 10, 248, 344 Tiwa, 10, 102, 145–146, 323, 347
General index
362
General index 363
echo word(s), 9, 20, 22, 26, 199, 314, 315–316 two FCs occurring in a row, 198
ECM, 79, 81–82, 95, 123–125, 133, 135, Final Complement, 6, 154, 164, 193–195, 207,
144–146, 178–179, 320, 347, see also 241–244
Exceptional Case Marking final complementizer, 6, 154, 164, 193–195,
cases in Dravidian, 124 207, 241–243, see also FC
cases in Munda, 125 Final over Final Constraint / FOFC, 19, 27,
cases (Small Clauses) and long-distance 196, 210, 240, 243–245, 329, 337
agreement, 123 Finiteness of the IC-clause vs. post-verbal
cases (Small Clauses) in Khasi, 125 position in blocking Wide Scope, 226
subjects Forward Control, 14, 31–32, 158, 246–250,
in Mizo and Hmar, 125 252–253, 258–259
accusative case-marked, 144 functional category, 14, 93–95, 106, 126, 128,
nominative case-marked, 145 185, 194, 240, 350
EHRCs, 263–265, 269, 271, 277–278, 283,
289, 292, 294–296, 299–300, 305, genitive subject, 7–8, 33, 44, 68, 70–71, 134,
307–309, 311–313, 318, 322, 332, 351, 136, 147–148, 156–157, 161, 164,
see also Externally Headed Relative 167–168, 178, 187–188, 259–260, 319
Clause grammatical functions (GF), 106, 110, 168,
embedded relative clause, 128, 263, 266–267, 271, 273, 275, 277–278, 293, 297, 336,
269, 276, 295, 299, 309, 326 353–354
embedded subject accessible to relativization in sentential
accusative case-marked, 135 relative clauses, 271
ergative case-marked, 135, 283 grammaticalization, 65, 91, 342
ergative grammaticalized form, 194, 198, 201, 218
case marking, 281, 283, 330 grammatical relation, 264, 270
subject modification, 279
subjects, 154 Head Direction Parameter, 6, 12, 194
Ergative–Absolutive, 41, 135, 187 Head to Head Movement, 6, 12
Evidence for Backward Control, 249, 251,
253, 255 IC-clause in clause-initial position, 221
Exceptional Case Marking, 79, 123, 135, IC functioning as
144–145, 171–173, 179, 323, see also a disjunctive marker, 200, 201, 307
ECM an expletive IC in non-restrictive clauses,
existential possession, 137 201
experiencer (dative) as a yes/no question marker in Kharia
as null PRO, 180 (Munda), 201
modified, 180 a subordinating linker with the
expletive as an additional question word interpretation of then, 201
(wh-expression), 236–237 IHRC(s), 38, 263–266, 269–270, 277–278,
expletive expressions in SALs – absence of, 290, 295–300, 302–313, 329–330, 351,
30–31, 316 see also internally Headed Relative
expletive light verb, 35, 37 Clause(s)
Externally Headed Relative Clause, 263–265, Pattern I, 297
267–268, 277, 279, 281, 283, 285, 287, Pattern II, 297
289, 291, 293, 299, 313, 352, see also imperfect participle, 172, 278
EHRC implications of postposition incorporation, 308
inalienable possession, 35, 137–138, 140, 153,
factive, 59, 62–63, 67–68, 195, 208, 210–212, 189, 192
215, 227–228, 232, 238 incorporation, 7, 11–12, 14, 38–40, 93–95,
factors responsible for ambiguity and 102–114, 126, 132, 264–265, 304,
disambiguation in IHRCs, 278 306–313, 317, 320–322, 333–334, 336,
FC, 6, 19, 27, 154, 165, 191, 193–219, 344
223–225, 227–230, 234–236, 238–240, in polysynthetic languages, 111, 113
243–245, 326–329, 337, see also final into the predicate, 39
complementizer nouns, 11, 107, 112
General index 365
of a direct object, 102, 104, 320 299, 308, 313, 329, 339, 341, 344, 348,
of a DP, 321 see also IHRC(s)
of adverbs, 7, 107, 308 internal head of an IHRC with an overt case
of inclusive particles, 106, 108 marker, 269
of intensifiers, 106 nature of, 264
of locative and instrumental PP, 39, 102 with no overt postposition, 270
of locative postposition in Paite, 111 intransitive nature of the dative predicate, 192
of Negative Polarity Items / NPIs, 111, 106,
108 Kayne’s generalization, 13, 112–113, 133,
of quantifiers, 106, 108 239
of the comitative postposition, 14, 110, 132,
322 language contact, 14, 18–20, 31, 33, 41, 95,
possessor of object and direct object, 104 126–127, 196, 240, 243, 246, 312, 314
possessor of subject, 105 and agreement, 126–127
postpositions, 38, 95, 106, 111, 126, 304, language–dialect dichotomy, 246
306–311, 313, 317, 321, 323, 336 LDA, see long-distance agreement
pronominal suffixes, 105 left peripheral, 193
verbal reciprocal, 308 COMP, 193, 195, 199, 241
indirect object(s), 15, 20–21, 27, 29, 39–40, lexical case, 90, 94–95, 97, 99, 108, 181, 260,
43, 53–54, 93, 98, 100, 102–104, 295, 340
108–109, 132–133, 135, 267, 269–271, marker (postpositional), 94
273, 278, 285, 290, 294, 296–298 lexical case marking, 12, 100, 264
modification, 285, 290 lexical/inherent (case), 134
indirect speech, 30, 195, 215, 240, 316 lexical passive, 74, 96, 140, 143, 167
infinitival clauses as NPs, 116 locative, 6, 12, 39, 44, 46, 73, 82–83, 95, 97,
infinitival form of the embedded verb, 265 102, 108, 111, 126–127, 132, 134–135,
infinitival (nominalized), 263 139–140, 150–151, 153, 157, 163–164,
inherent Case, 12, 15, 94, 134–135, 181–185, 168, 181–182, 188–191, 254, 261–262,
187–188, 191–192, 246, 248, 325, 329 271, 273, 286–287, 290, 292–293, 298,
assignment in DSCs, 181, 183, 185, 187 300, 305, 310–311, 325, 331
complex predicates and verb agreement, subject(s), 73, 108, 157, 164, 168, 292–293
182 long-distance agreement, 13, 94–95, 100,
default agreement and inherent Case, 181 114–124, 126, 133, 259, 262, 321, 324,
role of aspect, 185 337–338, 340
role of tense, 184 in Hindi-Urdu, 115–118, 259, 262, 337
role of the verb “to come”, 186–187 in Kashmiri, 119
marker, 12, 15, 94, 108, 134–135, 181–185, in Mizo, 120
187–188, 192, 246, 248, 325, 329 in Telugu, 121
marking of the dative subject, 135 long-distance anaphoric relationships, 115
of the embedded subject, 246 long-distance antecedent, 43, 76, 83, 87, 92,
inherently case-marked, 15, 94, 134, 181, 135–136, 214
191–192, 261 long-distance binding, 5–6, 14, 43–45, 66,
Initial Complement, 6–7, 12, 27, 193, 195, 75–78, 82, 87, 92, 319–320
201, 209, 213, 220, 224–225, 241–242,
313 Minimalist Program, 43, 93–94, 157, 320,
Initial Complementizer (IC), 6, 27, 188, 338–339
193–201, 204–226, 228–231, 236, modals, the case of, 166
238–241, 243, 313, 317, 321, 327–329 modification
instrumental, 12, 18, 39, 46, 54, 73, 78, 102, of possessor, 278, 288
132, 134–135, 137, 141, 176, 188, of the comitative, 286, 288, 294, 299–301
286–287, 291, 298, 300, 305 of the locative phrase, 286
(by-passive), 134 of the oblique object, 287
intensifiers, 13, 93, 106, 109, 132 monomorphemic anaphor, 5, 38, 43–45, 53,
Internally Headed Relative Clause(s), 7, 15, 55, 75–77, 91
38, 263–266, 269–270, 277, 295, 297, movement of clauses, 193
366 General index
multi-head relative clauses, 268–269 146, 148, 150–152, 154, 156, 158, 160,
multiple sets of agreement features, 95 162, 164–166, 168, 170, 172, 174, 176,
178, 180, 182, 184, 186, 188, 190, 192,
Narrow and Wide Scope, 195, 223–225, 259, 278, 292, 322, 326, 336–337,
227–229, 231, 233–235, 237–240 339–340, 346, 350–351, 353–354, see
factors affecting Wide and Narrow Scope, also accusative subject(s); dative
224 subject(s); genitive subject; subject,
finite and non-finite nature of the Initial genitive case-marked; locative,
COMP (IC) clause, 225 subject(s); NNSs
finiteness of the IC-clause vs. post-verbal as head, 292
position in blocking Wide Scope, 226 subject in by-passive (instrumental subject),
nature of the matrix predicate and COMP, 140
227 non-subject properties of NNSs, 165
occurrence of clitics/particles as scope Noun Phrase Accessibility Hierarchy, 15, 264,
inhibitors, 234–5 271, 275, 276, 278, 285, 288, 294, 300,
position of occurrence of the 332, 341
complementizer: initial or final, 224 NP-adjoined relative clauses, 266
position of question expressions and Scope NPAH, see Noun Phrase Accessibility
possibilities, 230 Hierarchy
position of the FC clause, 229 null (covert) operator, 263, 277, 279–281
role of intonation, pause and
morphophonemic criteria, 228 object complementation, 193
topicalization and Wide Scope, 233 Object of the Genitive, 267, 288, 332
use of pleonastic question expressions as obligative constructions, 101, 139
scope markers, 235 Oblique Object(s) (OO), 27, 40, 43, 59, 93,
Wide Scope reading, 224, 227–230, 103, 108–109, 126, 267, 271–273, 278,
232–236, 238–240 286–288, 290, 294, 297–298, 300–301,
natural phenomena pertaining to the body – 305, 332
dative subjects, 146, 148 modification, 286, 294, 297
nature of the predicate in DSCs, 170–171, 173, operator, see null (covert) operator
175, 179 operator scope, 193
Negative Polarity Items, 13, 93, 106, 108–109,
132, 195–196, 325, 337, 344 parameter(s), 1, 5, 6–8, 12, 19, 27–28,
NNSs, see also non-nominative subjects 111–112, 178, 194, 224, 239, 257, 313,
and subject-oriented verbs, 154, 162 325
as antecedents to anaphors, 154 negatively specified, 6–7, 208
as controllers of PRO, 154, 157 positively specified, 6
domains of occurrence, 134, 146–147, 149, parametric, 1–3, 10, 12, 19, 27, 29, 42, 94–95,
151, 153, 192 195, 209, 224, 246, 257, 343
formal representation, 192 choice, 246, 257
nominal anaphor, 5–6, 16, 38–42, 44–45, variation, 1–3, 10, 12, 19, 27, 29, 195, 224
48–54, 56–59, 62, 67–69, 72, 76, 78, participial, 22–23, 26, 41, 67, 127, 158,
81, 91, 155–156, 254, 341 160–161, 172, 177, 202–203, 212, 246,
nominalization, 37–38, 265, 269, 338, 342, 259–260, 263–268, 271, 277, 279–293,
344, 346–347, 351 311–312, 330, 332, 340, 344, 353
nominalizer, 37, 65, 319 infinitival relatives, 264
nominal reflexive, 15, 48, 53, 68, 254 participial-type, 288, 312
nominative–accusative, 7, 41, 126, 135, part–whole relationship, 35, 147, 153, 189
166–167, 175, 178, 183 single and double dative case marking, 147,
construction, 135, 167, 178, 183 153
nominalized-type, 312 passive, 16, 38, 48, 50, 57–59, 62–63, 68,
non-factive, 195, 210–212, 215, 227–228 72–74, 91, 96, 98–100, 106, 110,
non-finite type, 263–264 128–135, 140–144, 159, 166–167, 171,
non-nominative subjects, 3, 15, 32, 68–69, 71, 179, 281–282, 323, 346–347, 352,
73, 94, 134, 136, 138, 140, 142, 144, 354
General index 367