40
DYNAMICS OF TAMIL FINITE SYSTEM

DYNAMICS OF TAMIL FINITE SYSTEM

  • Upload
    others

  • View
    2

  • Download
    0

Embed Size (px)

Citation preview

DYNAMICS OF TAMIL

FINITE SYSTEM

DYNAMICS OF TAMIL

FINITE SYSTEM

R. KOTHANDARAMAN

CENTRAL INSTITUTE OF CLASSICAL TAMIL

CHENNAI

2020

DYNAMICS OF TAMIL FINITE SYSTEM© Central Institute of Classical Tamil, Chennai

CICT Publication No. 7

First published 2010Amazon KDP Edition, 2020 Reprint

Published by Central Institute of Classical Tamil

Chennai 600 113

ISBN: 978-81-908000-0-6

MEMBERS OF THE EDITORIAL BOARD

1. Editor-in-Chief Prof. R. Chandrasekaran, Director

2. Editor Dr. T. Saravanan

3. Associate Editors

Print versions Dr. R. Venkatesan

Dr. N. Periasamy

Dr. A. Arockiadoss

Dr. C. Ramachandran

Mr. K. Kannan

E-versions Dr. R. Akilan

Mr. D. Senthilkumar

Mr. S. Karthikeyan

Mr. A. Muruga Swaminathan

4. Layout Design Mr. K. Sankar

5. Typing work Mrs. S. Geetha

Mr. V. Sivaprakasam

Mr. T. G. Aruloli

Mr. S. Saravanan

Mr. M. Muniyasamy

6. Sales and Stock Mr. S. Saravanan

maintenance

M. KARUNANIDHI

Chief Minister of Tamil Nadu

Chairman Secretariat

CENTRAL INSTITUTE OF CLASSICAL TAMIL CHENNAI – 600 009

FOREWORD

The Tamils may be justly proud of the fact that Tamil has won the status of

a Classical language, the status it richly deserves and should have got long,

long ago. The Central Institute of Classical Tamil (CICT), established in

Chennai, has mapped out various plans including preparation of definitive

editions of forty one Classical Tamil texts and translation of these works

into English and other major European languages as well as into major

Indian languages and writing of a historical grammar of Tamil. Language

being the autobiography of a people, our objective is to preserve and

safeguard the invaluable treasure of the literary compositions in our

language. If only we could delve into our past and recover the riches and

wealth of the mighty treasure trove of Classical Tamil poetry, we will be

amply rewarded by its lofty poetry, the poetry that strengthens and purifies

the holiness of heart’s affection and enlarges our imagination. Apart from

these, reading the ancient Tamil texts such as Tolkāppiyam, Eṭṭuttokai,

Pattuppāṭṭu Tirukku�aḷ provides a foundation for scholarship for the present

and in this sense provides enlightened education.

It is heartening to write this foreword to the series of publications to

be brought out by CICT, which I am sure, will do full justice to the

masterpieces in Tamil without compromising on the quality of production.

The Caṅkam corpus being a repository of our glorious culture, it behoves

our present and future generations to study them and to convey their message

and the vision of life embodied in them to the public at large. Let me,

therefore, commend the series to the enlightened beings the world over.

Sd/-

(M. KARUNANIDHI)

CONTENTS

PREFACE xiii

AUTHOR xviii

1. INTRODUCTION 1

1.1 Redefining S–P Agreement 1

1.2 Decisively Impersonal – Further Observations 2

1.3 Manifestations of Ceyyuntu Finite System 3

1.4 Descendants of Ceyum Finite System 4

1.5 Extended Finite System 5

1.6 Forgotten Suffix Variants 9

1.7 Formal Distortion 10

1.8 Reanalysis of Verbal Suffixes 12

1.9 Finite System – A Vital Player of Syntax 17

1.10 Problems and Resolutions 18

1.11 Diphthongs in Tamil – An Impossed Burden 18

1.12 Reinterpretation of Glides in Tamil 20

1.13 The Plosive Increments in Tamil 21

1.14 Composition of Past Tense Suffix -i­ 21

1.15 First Person Marker om and ōm 22

1.16 Is a�ik-ilār a Nonpast Ability Verb 22

2. BEGINNINGS OF FINITE SYSTEM IN TAMIL 23

2.1 Finite System in General 23

2.2 Impersonal Finite System 25

2.3 Subject-Predicate Agreement – A Later Development 26

2.4 Dating of Classical Tamil 28

2.5 Tenseless Finite System – Untenable Hypothesis 29

2.6 Tensed Impersonal Finite System 30

2.7 Ceytici­ Finite System 31

2.8 Relative Participle as Impersonal Finite System 32

2.9 A Hypothesis Called into Question 34

2.10 Predicate Verb as Nominal Constituent 34

2.11 Ceytu Verbal as Nominal Constituent 37

viii

2.12 Future Negation 38

2.13 Means Differ But Goal Same 40

2.14 Source of Personal Finite System 42

2.15 Relative Participle as Finite Verb – Further Observations 49

2.16 A Nonpast Finite System Parallel to ceyt– a­+PM Type 50

2.17 Reflexes of the Increment –am 53

2.18 Impersonal Finite System of NP and PP Types 56

2.19 Extended Impersonal Finite System 58

2.20 Nonpast Impersonal Finite System 59

2.21 Ceyal Type of Finite System 59

2.22 Ceytal Type of Finite System 60

2.23 Ceyum Type of Finite System 63

2.24 Ceyum Verbs – Stages of Development 68

2.25 Finite System of Ceypu Type 70

2.26 Source of Ceypu Finite System 71

2.27 Finite System of Ceyku Type 72

2.28 Source of Ceyku Finite System 75

2.29 Evolution of Extended Personal Finite System 76

2.30 Remarks on Single Window Conjugation 79

3. SYNTAX OF COMPLEMENTIZER 84

3.1 Complers in Traditional Treatment 84

3.2 Complers – A New Perspective 85

3.3 Compler Types 87

3.4 Lexical Semantics of Compler 87

3.5 Internal Reconstruction of the Compler – um 89

3.6 Complers in Dative Constructions 91

3.7 The Complers -a­ and -i­ 92

3.8 Again Looking Back at -am 104

3.9 The Compler -*av 106

3.10 The Compler -ē 107

3.11 Consonantal Compler 109

3.12 Historical Constraint 110

3.13 The Compler -i��u 117

3.14 The Complers tā­-/ta­- and tām/tam 119

3.15 The Compler -a��u 123

3.16 Functional Divergence – A Note 125

3.17 Sentential Complers 126

3.18 Adverbial Compler 127

CONTENTS

ix

3.19 Adjectival Compler 130

3.20 Nominal Compler 133

4. INTRICACIES OF TAMIL FINITE SYSTEM 136

4.1 Verbal System – General Observations 136

4.2 Tense Implied Finite System 140

4.3 Post Nominal Negative Finite System 149

4.4 Tensed Personal / Gender Verbal Noun 152

4.5 A Digression 158

4.6 Accusative Negation 159

4.7 Nominal Finite System 161

4.8 Person / Gender Marker as Cliticized Be Verb 163

4.9 Kannada Situation 168

4.10 Person / Gender Markers – Further Observations 179

4.11 Extended Predicate System 180

4.12 Loss of a Paradigm 191

4.13 Deflection of Verbal System 196

4.14 Finite System of Ceyum Type 203

4.15 Verbal System of ceyuntu Type 208

4.16 Emergence of Past Finite System 211

4.17 Emergence of ceyyiyar Finite System 215

4.18 Finite Verbal Bases 217

4.19 The -cu / -su Factor in Telugu / Kannada 227

4.20 From Inflectional to Derivational System 233

5. SYNTAX OF NEGATION 235

5.1 Negation at First Glance 235

5.2 Nonpast Negative Paradigm 238

5.3 Personal Negation 245

5.4 Syntax of Classical Negation 247

5.5 A Digression 252

5.6 Nonpast Negation of Ceyki­�u+il– Type 253

5.7 Nonpast Negation of VB+al– Type 257

5.8 Negative Finite System – Further Development 262

5.9 Finite Verb as Nominal Construction 266

5.10 Imperative System 266

5.11 Speculating Negative System 278

5.12 Negative Nominals 280

CONTENTS

x

5.13 Reduplicative Finite System 283

5.14 Impersonal Negation 286

5.15 Tag Question 288

5.16 Short Versus Long Negation 290

5.17 Ability Phrase 291

5.18 Kil– A Newly Emerging Ability Verb 299

5.19 Double Negation 314

5.20 Optative Negation 315

5.21 A Problem to be Resolved 321

5.22 Negative Participial Nouns 322

5.23 Negative Participles 328

5.24 Negative Adverbial Participles 330

5.25 Negative Conditionals 342

5.26 Negative Adnominal Participles 345

5.27 An Effort to Resolve a Problem 349

5.28 Adnominal Participle as Functional Category 351

5.29 The Lost Paradigm 358

5.30 Tamil Negation Through Centuries 363

6. FORGOTTEN RELATIVE CLAUSE CONSTRUCTIONS 370

6.1 Forgotten Syntax of Relative Clause 370

6.2 Retrieving the Lost Finite System 372

6.3 Classification of X+Ceyal+NP Constructions 373

6.4 Moving From Inflectional to Derivational Stage 381

6.5 Relative Clause of NP1–am+NP

2 Type 384

6.6 Tol’s Treatment of –am 391

6.7 A Digression 394

6.8 Reflexes of –am 395

6.9 Plosive and Nasal Increments in External Sandhi 402

6.10 A Note on viḷa–m palam 406

6.11 Sentential Sources of NP1–am+NP

2 Type 407

7. CONCLUSION 412

7.1 What has been done? 412

7.2 What has to be done? 416

7.3 Reinterpretation of Personal Suffixes 417

7.4 Short Pronouns without Case Marking 418

CONTENTS

xi

7.5 Relative Clause in Tamil 419

7.6 Demonstrative and Interrogative Systems in Tamil 440

7.7 Roots of Subordinate Clause 443

7.8 Ceypu Type of Verbal Participle 448

7.9 Ceyyā Type of Verbal Participle 450

7.10 As If Clause in Tamil 453

7.11 Lexical and Gender Head Nouns 455

7.12 Onomatopoeic Constructions 456

7.13 Coordinate Syntax 458

7.14 Passive Syntax 460

7.15 Reflexive Syntax 462

7.16 Tensed Finite System Ending in –ay 464

7.17 Concluding Remarks 466

ABBREVIATIONS 467

REFERENCES 471

SUBJECT INDEX 477

AUTHOR INDEX 482

CONTENTS

PREFACE

New proposals and counter proposals against the existing ones are

theoretically significant in promoting the advancement of knowledge in

any discipline. While new proposals are path-breaking ventures, counter

proposals seek to evaluate the existing theories and convictions. Exponents

of new and counter proposals are trendsetters, and traditionalists are the

guardians of the existing theoretical models. It is not uncommon that

traditionalists and trendsetters are taking defensive and offensive positions

respectively. As for the study of linguistic system, there has always been a

conflict between the trendsetters and traditionalists of whom the former

are the pillars of counter proposals. Theoretically, the traditionalists are

very much in alignment with the principles of external adequacy whereas

the trendsetters who move ahead out of reach of the traditionalists adhere

to internal adequacy. Externally adequate description is very much

concerned with the goals disregarding the means, and this is not the case

with respect to the deliberations based on internal adequacy. Internally

adequate exposition is not only particular about the goals but also about

the means. The linguistic system obtained in Tamil has been traditionally

described with reference to the principles governed by external adequacy.

It is to be noted that externally adequate description is performance -

oriented without regard to the linguistic competence of the speaker. The

Chomskyan distinction between competence and performance of the speaker

is the core issue of the present monograph. The main concern and objective

of this work is to capture the linguistic competence of the speaker

marginalizing the linguistic performance. Internally adequate description

has therefore reference to linguistic competence seeking to identify the

deep-seated regularities of the speech system in the underlying

representation.

It is to be taken into consideration that linguistic system is a

manifestation of a series of diachronic layers of different stages and

periods. Since this factor is not explicit in the synchronic system, the

grammarian is constrained to resort to different descriptive strategies to

reach the surface representation. More often than not, these strategies

employed in traditional description without any diachronic perspective

are highly adhoc and arbitrary. Notice, for instance, the underlying

xiv

representations of makattu ‘son’ (oblique base), marattu ‘tree’ (obl.), and

a©attu ‘death ritual’ (obl.) are maka+attu, maram+attu, and a©a­+attu

where attu is an instance of inflectional increment according to the

phonological description obtained in Tolkāppiyam (Tol.). Tol. proposes

that attu loses a- after nouns ending in a-. Consequently, maka+attu

becomes makattu. With a view to generalizing the application of the rule

under reference, Tol. is taking for granted the deletion of the word final

labial and alveolar nasals of maram and a©a­ before attu. This follows

the deletion of a- of attu after the contrived versions mara- and a©a-.

There are several anomalies in this analysis. First, what is the justification

to propose the inflectional increment attu in the oblique constructions

under reference? Second, does the deletion of word final consonant before

attu have any phonological motivation? Third, is the deletion of the initial

vowel of attu after nouns constrained to lose the word final labial and

alveolar nasals phonologically justified? Fourth, is there no scope to

resort to internal reconstruction within the framework of synchronic

description? Traditionally, the foregoing analysis has been upheld without

its sustainability being evaluated with reference to the issues raised above.

In the counter proposal, not only attu is replaced by the inflectional

increment tu, but also the strategy of internal reconstruction is pressed

into service. The nouns maka and a©a­ are associated with the internally

reconstructed variants *makam, and *a©am. These variants are

morphologically definable occurring both in internal and external

sandhi. E.g. *a©am+tu > a©an+tu > a©at+tu, *makam+tu > makan+tu >

makattu, *a©am+kuṭam > a©aṅ+kuṭam > a©ak+kuṭam, *makam+pē�u >

makap+pē�u. This analysis is in consonance with the analysis of nouns

ending in labial nasal, and this is the reason why maka and a©a­ are

internally reconstructed into *makam and *a©am on the basis of makattu

/ makap+pē�u and a©attu / a©ak+kuṭam which are analogous to marattu /

marak+kiḷai. Contrary to the traditional analysis, the alternative proposal

meeting the requirements of internal adequacy renders justice not only

to the goals but to the means as well. This proposal maintains that the

nasal assimilates to voiceless plosive before single and double voiceless

plosives, of which the first process is an earliest phonological process

belonging to either PDr or pre PDr linguistic system. The rule N>P/-P

belonging to PDr phonological system is retained in the individual

Dravidian languages in general and Tamil in particular. On the other

hand, the rule N>P/-PP is located in the intermediate stage of derivation.

The derivation of e�ku ‘to me’ from e­akku through *e­kku is a case in

point in this regard.

DYNAMICS OF TAMIL FINITE SYSTEM

xv

Internally adequate linguistic description is an effort to capture the

diachronic factors within the synchronic net work, although this is not an

essential requirement. In this respect, descriptively sustainable structurally

relevant and semantically identical paraphrase versions are taken into

account leading to the integrated description. Consider, for instance, the

NP construction taṇ-e­+katir ‘the cool rays’, taṇ-am+tu�ai ‘the cool water

yard’ and the relative clause constructions taṇ varal vāṭai (Ku�u. 35.5)

‘the north wind that comes cool’. The semantic import of e­ and am, and

the functional import of varal figuring in these constructions are enigmatic

in the synchronic description. However, this is resolved once their

systematic paraphrase versions are identified. The semantically and

structurally related constructions of the NPs and the relative clause

constructions under illustration are taṇṇ-iya/ taṇṇit-ākiya katir, taṇṇ-iya /

taṇṇit-ākiya tu�ai, and taṇṇ-e­a / taṇṇit-āy varum vāṭai. Notice the

expression ākiya is a Be verb functioning as tense implied noun modifier,

and varum being a complementary variant of varal is functionally a non-

past noun modifier. It is significant that ākiya and –iya(<iy-a) occurring

in taṇṇ-iya are in complementary distribution. This provides the basis to

treat –iya as an instance of Be verb functioning as tense implied noun

modifier. The structurally and semantically related paraphrase

constructions make it possible to identify e­ and am occurring in taṇ-

e­+katir, and taṇ-am+tu�ai as Be verbs functioning as noun modifiers,

and varal figuring in taṇ varal vāṭai are functionally equivalent to varum

occurring in taṇṇ-e­a / taṇṇit-āy varum vāṭai. Traditionally, e­ has no

functional and semantic explanation, am is empty morph, and the functional

significance of varal (=varum) has not received proper treatment. Note

that varal in the present deliberation is not an instance of case declinable

verbal noun since it is modified by an adverb as evidenced in the

constructions vēkamāka varal. In the absence of the analysis presented

above through counter proposal, the NPs under reference cannot be

structurally accounted for particularly with respect to e­, am and varal.

The grammarian no doubt is at liberty to formulate a theory to describe

the linguistic system without straining it. Tol’s description has however

complicated the phonological system that obtained in Tamil disrupting its

genius. We consider that the theory of phonology should not offend the

integrity of the linguistic system. This cannot be achieved without

theoretically sustainable alternative models. Counter proposals being

alternative models are inevitable for deepening the knowledge base. The

present monograph entitled Dynamics of Tamil Finite System is the

PREFACE

xvi

introductory part of the Historical Syntax of Tamil assigned to me as a

major project. This is a monograph with several counter proposals

persuading reconsideration of the existing traditional as also modern

descriptive strategies. The inflectional increments am, um, and –m occurring

in the dative constructions e­akku ‘to me’, maka­ukku ‘to the son’, and

elikku ‘to the rat’ are identified in the alternative proposal of linguistic

description. E.g., e­akku < e­-am+ku, maka­ukku < maka­-um+ku, elikku

< eli-m+ku. Verbal constructions of ceytu, ceyku, ceypu, and pāṭi��u types

occurring in different syntactic contexts in early and middle Tamil are

identified to belong to the impersonal finite system. Person / Gender

markers and pronouns are treated as complementary variants. With

reference to the structurally and semantically related paraphrase versions,

the inflectional increments such as i­, a­, am, um, ē etc., are identified as

Be verbs of suffix type. The expression e­ is a homophonous form referring

to Be and Say meanings, and quite misleadingly it is always associated

with the Say meaning only. These are all counter proposals made in this

monograph, the description of which is governed by the principles of

internal adequacy. Due to intellectual limitation, no researcher will be

able to explore the intricacies of the linguistic system in its entirety. It is

but natural that knowingly or unknowingly there may be theoretically blind

spots. The acceptance of and reluctance to subscribe to the findings in this

monograph will be a reflection of the mindset of the scholars of traditional

and trendsetting streams. This we anticipate and consider natural in any

theoretical exploration.

This monograph has been made possible by the active and productive

cooperation of the Centre of Excellence for Classical Tamil (CECT), a

component of Central Institute of Indian Languages (CIIL), Mysore. The

CECT, now CICT (Central Institute of Classical Tamil), invited me to work

on the project on Historical Syntax of Tamil. The work was started on

April 10, 2006 and completed on August 08, 2007. The present work is

the first part of the research programme assigned to me. It is a matter of

great pleasure for me to express my immense gratitude and heartfelt thanks

to Prof. Udaya Narayana Singh, Director, CIIL, and Dr.K. Ramasamy,

Professor - cum - Deputy Director, CIIL and the Chief of CECT for

providing me an ideal academic environment to finish this work without

any strain and hardship. The cooperation and encouragement of these two

academic gems were excellent beyond verbal expression. Their participation

in every respect in shaping this monograph is praiseworthy and I will cherish

it until I breathe my last.

DYNAMICS OF TAMIL FINITE SYSTEM

xvii

It is a matter of great pleasure for me to seek persuasive comments

from Prof. P. Marudanayagam and Prof. M.S. Nagarajan of CICT. Their

contribution in shaping the present work is commendable. I have great

pleasure in conveying my deep debt of gratitude to them.

I must make a reference to my wife Thirumati R. Santhagunabushanam.

She helped me to carry on my research activities without any hindrance by

spending liberally from her pension resource. I am nothing without her.

My thanks are due to her.

The first draft of this monograph was prepared by Thiru P. Thiyagarajan

in Puduchery. It was really a painful and challenging task for him. I am

bound to remember his exemplary service in this respect. The final version

of this work had the benefit of the expertise of Selvi V. Gayathri and

Thiru.A. Murugesan, Data Entry Operators, CICT. I have great pleasure

in placing on record my sincere thanks to them.

The service rendered by GAAMA DTP Division in respect of typesetting

and pagemaking of this publication is commendable. I would be failing in

my duty if I do not place on record my gratitude to Thiru P. Sudhakaran,

the man behind this grand performance.

PREFACE

CHAPTER 1

INTRODUCTION

1.1 Redefining S–P Agreement

1.1.1 It is true that the Subject – Predicate agreement (henceforth S–P

agreement) in Dravidian in general and Tamil in particular had come to stay

in PDr stage itself. However, it does not mean that this agreement had its

roots in the syntax of early Dravidian that precedes PDr stage. As a matter of

fact, S–P agreement is a later development in Dravidian syntax. This is very

much in evidence that there are instances in major literary Dravidian languages

where this agreement is not fool proof. Malayalam is a good case in point

where S–P agreement is absent. The popular argument is that Malayalam

had maintained originally this agreement, and in due course it had lost this

phenomenon. No convincing evidence however has been made available to

validate this hypothesis. Tamil, Telugu, and Kannada do not maintain S–P

agreement with respect to nonpast finite system of ceyyum type in III person.

In early Tamil, ceyyum type of finite system functions as predicate of all the

gender NPs namely masculine, feminine, nonmasculine singular, and plural

except human plural. In modern Tamil, the same finite system maintains

cooccurrence relation with nonhuman singular and plural Subject NPs. As

for Telugu and Kannada, the same type of agreement obtained in early Tamil

with respect to ceyyum type functioning as predicate is noticed to have been

maintained. Another significant evidence to be taken into consideration is

that the finite system of ceyyum type functioning as predicate assumes the

function as noun modifier to the head nouns of all the three persons in Tamil

and Malayalam without any morphological disruption.

1.1.2 In early Tamil, there were subordinate clause constructions of

ceyt–e­a, ceyt–āṅku, and ceyt–a­­a types of verbal phrases in which e­a,

āṅku, and a­­a are grammatical morphemes referring to cause and

comparison. The verbal constructions of ceytu type occurring before these

grammatical morphemes belong to the past impersonal finite system

functioning as predicate. Traditional scholarship has not taken this factor

into consideration though it interprets tu�ayva­ vant–e­a ‘since the chief

of water yard came’, for instance, as tu�ayva­ vantā­ āka or tu�ayva­

2 DYNAMICS OF TAMIL FINITE SYSTEM

vantat–āl. Notice, in the illustrations under reference vant(u), vantā­, and

vantat(u) occurring before e­a, āka, and āl are in complementary

distribution, a fact which confirms that vant(u) was orignially an instance

of past impersonal finite construction. See Subramaniyam (2009: 18–23)

for further discussion. It is interesting to note that while the impersonal

finite system of ceytu, ceypu, ceyvu, pāṭi, pāṭi��u and ceyki­�u types in

Tamil cannot be declined for case, the finite system associated with person

and gender suffixes is capable of admitting case system. E.g., kāṇp+ē­

‘will see–I,’ kāṇpē�–ku (< kāṇpē­+kku) (Na��. 332.9) ‘to me who will

see.’ There are instances in early Tamil personal finite system assuming

the nominal status. E.g., nī pi©aytt+āy pōl na­i nāṇi­+ay (Pu�am. 43.17)

‘as if you had committed the offence, you felt ashamed of,’ uṇm+ar-um

ti­m+ar-um (Pati. 24.18) ‘those who will drink/eat and eat/masticate).’

The personal verbs pi©aytt+āy, uṇm+ar and ti­m+ar in these examples are

identifiable as tensed personal verbal nouns familiarly but misleadingly

known as participial nouns. This implies that not only the finite system

marked with agreement suffix is a later development, but also the so called

person and gender suffixes are nothing but the variants of person and

gender pronouns. This is the reason why the finite system with agreement

suffixes assumes the noun status capable of being declined for case.

1.2 Decisively Impersonal – Further Observations

1.2.1 In early Tamil, the negative finite system is associated with tensed

impersonal finite constructions of ceytu, ceyta­�u, pāṭi­�u, pāṭi��u, ceyku,

and ceyki­�u types. The negation is marked by the addition of the negative

constructions namely ile­, ilam, ilay, ilir, ila­, ilaḷ, ilar, i­�u, and ila of all

three persons to the types of impersonal finite constructions under reference.

It is to be noted that the past negative finite construction ceyt-ila­ ‘did not

do–he,’ for instance, is traditionally interpreted as ceytā­ alla­ providing

clinching evidence to treat constructions of ceytu type occurring before ila­,

ilaḷ, etc., to belong to past impersonal finite system. This is yet another

case in point to maintain that the impersonal finite system is chronologically

earlier than the finite system marked with agreement suffixes. Notice there

are constructions of ceyki­�-ila­ / pāṭuki­�-ila­, ceyki­�-ilaḷ / pāṭuki­�-ilaḷ,

etc., types representing nonpast finite negation are attested in middle Tamil

particularly bhakti literature of Saivism and Vaishnavism. Despite their being

a later development, ceyki­�(u) and pāṭuki­�(u) types of tensed finite verbs

occurring before ila­, ilaḷ, etc., are impersonal in character. The constructions

of pāṭi, pāṭi��u and pāṭi­�u types occuring in pāṭiy-āṅku (=pāṭiyatu pōla),

3INTRODUCTION

pāṭi��-ila­, and pāṭi­�-ila­ types are identifiable as past impersonal finite

verbs. Similarly, ceyku type of verbs occuring before ila­, ilaḷ, ilar, etc.,

are also treated to belong to nonpast impersonal finite system.

1.2.2 It is significant that the impersonal tensed finite system of pāṭi,

pāṭunnu, ceytu and ceyyunnu types in Malayalam provides the basis for the

noun modifier constructions of pāṭiy-a, pāṭunn-a, ceyt-a and ceyyunn-a

types. The situation in Tamil does not seem to be different from Malayam.

The noun modifiers of pāṭiy-a, pāṭuki­�-a, ceyt-a and ceyki­�-a types in

Tamil should have evolved from the tensed finite system of pāṭi, pāṭuki­�u,

ceytu and ceyki­�u types whose impersonal character we have already made

reference to in 1.2.1. The Kanyakumari and Tirunelveli dialects of Tamil

furnish further evidence to hold that the impersonal finite system is

historically earlier than the personal finite system. In the dialects under

reference ceyk-a and ceyyut-a types of constructions refer to nonpast noun

modifiers. E.g., varuk-a / varut-a payya­ ‘the boy who comes,’ pāṭuk-a/

pāṭut-a payyan ‘the boy who sings,’ nā­ paṭikk-a / paṭikkut-a paḷḷi ‘the

school where I study.’ The nonpast noun modifiers under reference have

their tense bases in ceyku and ceyyutu types which are to be theoretically

treated as verbal system that belongs to nonpast impersonal types. In

Kanyakumari and Tirunelveli dialects ceyku and ceyyutu types are

synchronically identifiable as nonhuman nonpast singular finite verbs. E.g.,

ma©a varuku ‘rain comes,’ kāttu aṭikki (< aṭikku) ‘wind blows,’ puli ōṭuku

‘tiger runs.’ It is to be noted that these construction types have assumed the

status of tense base before relative suffix and person/gender suffixes. E.g.,

pēcuk-ā­/pēcut-ā­ ‘speaks-he,’ kuṭikk-ā­ / kuṭikkut-ā­ ‘drinks -he’ We

consider that the tense bases are theoretically / historically identifiable as

tensed impersonal finite constructions.

1.3 Manifestations of Ceyuntu Finite System

The emergence of nonpast finite constructions of ceyuntu (<*keyuntu)

type is a remarkable innovation in Dravidian. The finite system of this

type is attested in Pu�anā­ū�u, one of the anthologies of classical Tamil.

Different types of nonpast personal finite constructions such as a�iyun-ar,

a�in-ar, a�iñ-ar, etc., have evolved from the nonpast impersonal finite sys-

tem of ceyuntu type. The nonpast bases such as i©att(u), aḷitt(u), and

ma�aytt(u) occurring in i©att-um (Ku�aḷ. 1250) ‘will lose-we,’ aḷitt-i (Kali.

95.31, Pati. 79.3) ‘will grant-you(sg),’ and ma�aytt-ir (Ku�aḷ. 1318) ‘will

hide-you(hon.sg)’ are of *keyttu type traceable to *keyutt(u) (<*keyun[t]u)

4 DYNAMICS OF TAMIL FINITE SYSTEM

type. The present finite system of ceyyunnu type in Malayalam has its

source in cey(y)untu type, and not in ceyki­�u type. In Kannada, there are

present finite constructions of geyutt-āne ‘does-he,’ geyutt-āḷe ‘does-she,’

geyutt-āre ‘do-they (hum),’ and geyutt-ade ‘does-it’ types. The tense base

of geyuttu type is traceable to *keyuntu (>Ta. ceyuntu). The Telugu situa-

tion is closely identical with that of Kannada. The nonpast finite con-

structions pāḍut-ānu ‘sing-I,’ and pāḍut-unu ‘will sing-I,’ for instance,

have their tense base in pāḍutu of *keyuntu type. The foregoing examples

provide the basis to maintain that the Dravidian finite system was imper-

sonal earlier, and from it the personal finite system has evolved, a pro-

cess that was consolidated to the extent possible in the proto-Dravidian

itself.

1.4 Descendants of ceyum Finite System

Basically verbs of ceyum type are impersonal in character, and this

type constitutes the basis to the development of not only of cey(y)untu

type, but also of personal finite system of ceym-ar type in which ceym is

the reduced version of ceyum. In early Tamil, there are such personal

verbs as uṇm-ar (Pati. 24.18), e­m-ar (Ku�u. 395.6), ti­m-ar (Pati. 24.18),

ī­m-ar (Pu�am. 74.7), kāṇm-ar (Pattup. 6.513), etc., which have their

tense bases in ceyum type. The finite verbs of ceym-ar type makes it

possible to reconstruct a paradigm which includes such finite

constructions as *ceym-e­(I sg), *ceym-am(I pl), *ceym-ay(II sg), *ceym-

ir(II pl), *ceym-a­(masc), *ceym-aḷ(fem), and *ceym-a(nh pl). In modern

Tamil, there are finite constructions of ceyv-ar type, and in early Tamil,

there are nonpast finite constructions of ceyv-e­, ceyv-am, ceyv-ay, ceyv-

ir, ceyv-a­, ceyv-aḷ, ceyv-ar, and ceyv-a types. In spoken dialects of Tamil,

we come across such finite constructions as ceyv-e­(I sg), ceyv-om(I pl),

and ceyv-e(II sg) types. All these types of nonpast finite constructions

marked with the nonpast suffix -v are traceable to the nonpast personal

constructions associated with the nonpast marker -m which is followed

by short person and gender suffixes. This implies that the finite

constructions of ceyv-aḷ and ceyv-ar types, for instance, are traceable to

the nonpast finite constructions namely of *ceym-aḷ and *ceym-ar types.

In Kannada, the future finite constructions such as kuḍiv-a­u ‘will drink–

he,’ kuḍiv-aḷu ‘will drink–she,’ kuḍiv-aru ‘will drink–they (hum),’ etc.,

are traceable to *kuḍim-a­u, *kuḍim-aḷu, and *kuḍim-aru respectively.

The Kannada nonpast relative participles of māḍuv-a type also derive

from *māḍum-a type. Notice, Kannada future constructions and nonpast

5INTRODUCTION

relative participles of the types under reference have their tense bases in

ceym (<ceyum) type historically.

1.5 Extended Finite System

1.5.1 In Tamil, the emergence of extended personal finite system is an

important development. The finite constructions of the type under reference

is associated with the inflectional increments -a­ and -av. These increments

are preceded by the tense bases of ceyt, ceyv / ceyp, and ceym types, and

followed by short person and gender suffixes namely -e­, -am, -ay, -ir,

-a­, -aḷ, -ar -tu and -a. As for the nonhuman singular suffix -tu, it is preceded

by the inflectional increment -a which has a phonologically conditioned

variant of -av. Consider the following examples marked with gender

suffixes which are preceded by the complers -a­-, -av-, and -a-. The second

column refers to nonpast, and the third one, future.

1 (1) (a) ceyt–a­+a­ (2) (a) ceyv–a­+a­ (3) (a) ceym–a­+a­

(b) ceyt–a­+aḷ (b) ceyv–a­+aḷ (b) ceym–a­+aḷ

(c) ceyt–a­+ar (c) ceyv–a­+ar (c) ceym–a­+ar

(d) ceyt–a­+�u (d) *ceyv–a­+�u (d) *ceym–a­+�u

(e) ceyt–a­+a (e) ceyv–a­+a (e) ceym–a­+a

2 (1) (a) ceyt–av+a­ (2) (a) ceyp–av+a­ (3) (a) ceym–av+a­

(b) ceyt–av+aḷ (b) ceyp–av+aḷ (b) ceym–av+aḷ

(c) ceyt–av+ar (c) ceyp–av+ar (c) ceym–av+ar

(d) ceyt–a+tu (d) ceyv–a+tu (d) ceym–a+tu

(e) ceyt–av+ay (e) ceyp–av+ay (e) ceym–av+ay

1.5.2 In early Tamil, ceyt-av+ay, ceyp-av+ay, and nall-av+ay types

attested in Paripāṭal are also II singular finite constructions. As for the

finite constructions of the type referred to in 1 (1), they are attested in

full not only in early Tamil but in modern Tamil as well. The nonhuman

singular finite constructions of ceyt-a­+�u type are lost in modern Tamil

and whose place has been occupied by the constructions of ceyt-a+tu

type which belong to the paradigm of the type illustrated in 2 (1). The

finite constructions of the types referred to in 1(2) except ceyv-a­+a /

uṇp-a­+a type are not available in modern Tamil. In early Tamil, there

are constructions of ceyv-a­+PM/GM where PM and GM are person and

gender markers respectively. E.g., pi©iv-a­+am (Pattup. 8.60) ‘will

squeeze–we’, varuv-an+aḷ (Na��. 339.6) ‘will come-she.’ The nonhuman

6 DYNAMICS OF TAMIL FINITE SYSTEM

singular finite constructions of *ceyv-a­+�u / *uṇp-a­+�u types are only

a theoretical possibility, and their occurrence is not attested in Tamil of

any period. In early Tamil, the construction types figuring in 1 (2) do not

function in many cases as predicate either in simple sentence or main

clause construction. Mostly, they occur in early Tamil functioning as

participle. The construction types figuring in 1 (3) have developed into

such constructions as those associated with the future tense marker -v

attested in type 1 (2). For instance, *ceym-a­+aḷ has developed into ceyv-

a­+aḷ in early Tamil, and it appears that *ceym-a­+aḷ belongs to pre

Tamil period. The nonhuman singular finite constructions of *ceym-a­+�u

type figuring in 1 (3) is again a theoretical possibility not attested

anywhere else in Tamil. Its place seems to have been occupied by the

constructions of ceym-a+tu type that belong to a different paradigm of

the type marked in 2 (3).

1.5.3 The extended finite constructions of the types illustrated in

2 (1) – (3) above have undergone several changes resulting in finite

constructions associated with long personal suffixes namely -āy / -ōy, -

ā­ / -ō­, -āḷ / -ōḷ, and -ār / -ōr without the occurrence of the inflectional

increment -av. The finite constructions of ceyt-ā­/ceyt-ō­ and *pāṭiy-ā­/

pāṭiy-ō­ types, for instance, are traceable to the underlying constructions

of *ceyt-av+a­ / *pāṭiy-av+a­ types. There are II person singular finite

constructions of ceyt-āy / ceyt-ōy and *pāṭiy-āy / pāṭiy-ōy types attested

in Tamil whose underlying representations are traceable to the

constructions of *ceyt–av+ay / *pāṭiy–av+ay types. Both in early and

modern Tamil, constructions of ceyt–av+ay, ceyp–av+ay, pāṭiy–av+ay,

and pāṭup–av+ay types are generally recognized as nonhuman plural

finite constructions. However, significantly constructions of these types

are also noticed to refer to II person singular as evidenced in such

examples as aṭṭ–av+ay (Pari. 21.66) ‘fought–you,’ ēnti­–av+ay (Pari.

15.58) ‘lifted–you,’ viritt–av+ay (Pari. 21.67) ‘spread–you,’ and peyarpp–

av+ay (Pari. 21.66) ‘will uproot–you.’ It is thus clear that constructions

of ceyt-āy / ceyt-ōy, ceyv-āy / ceyv-ōy, *pāṭiy-āy / pāṭiy-ōy, and pāṭuv-āy/

pāṭuv-ōy types are traceable to *ceyt-av+ay, *ceypv-av+ay, *pāṭiy-av+ay,

and *pāṭup-av+ay types respectively. An interesting development is that

the nonpast suffix -p figuring in weak conjugation of ceyp-av+a­ and

pāṭup-av+a­ types, for instance, becomes -v when they evolve into ceyv-

ā­ / ceyv-ō­ and pāṭuv-ā­ / pāṭuv-ō­ types of finite constructions. An

exception to this ruling is that such constructions as kāṇp-av+a­, uṇp-

av+a­, ti­p-av+a­, and pūṇp-av+a­, for instance, become kāṇp-ā­ /

7INTRODUCTION

kāṇp-ō­, uṇp-ā­ / uṇp-ō­, ti­p-ā­ / ti­p-ō­, and pūṇp-ā­ / pūṇp-ō­, and

not as *kāṇv-ā­ / *kāṇv-ō­, *uṇv-ā­ / uṇv-ō­, *ti­v-ā­ / *ti­v-ō­, and

*pūṇv-ā­/ *pūṇv–ō­. There are constructions such as kāṇuv-ā­, uṇṇuv-

ā­, ti­­uv-ā­, and pūṇuv-ā­ noticed to be familiarly used in Tamil. These

constructions are traceable to the underlying constructions *kāṇup-av+a­,

*uṇṇup-av+a­, *ti­­up-av+a­, and *pūṇup-av+a­ respectively, and not

to kāṇp-av+a­, uṇp-av+a­, ti­p-av+a­, and pūṇp-av+a­ although, the

latter namely kāṇp-av+a­, for instance, is a reduction of the former

namely *kāṇup-av+a­.

1.5.4 In early Tamil, there are nonpast human plural finite constructions

of ceymār type. Tolkāppiyam, the earliest extant Tamil grammar also makes

reference to this type of finite system (Col.Cē­ā. 207) Interestingly, in

most cases ceymār type of constructions function as infinitive in early Tamil

subject to the condition that the verbal predicate of the main clause is of

human plural type. Also we come across in early Tamil, finite constructions

of ceyyumō­ and ceyyumōr types. On this basis, it is possible to reconstruct

a paradigm marked with constructions of *ceyyumē­, *ceyyumām /

*ceyyumōm, *ceyyumāy / *ceyyumōy, *ceyyumīr, *ceyyumā­/ ceyyumō­,

*ceyyumāḷ / *ceyyumōḷ, and *ceyyumār / ceyyumōr types. What is the source

of these reconstructed versions? We consider at the moment that the finite

constructions of the types figuring in 2 (3) constitute the underlying system

from which the reconstructed versions under reference are derivable. As

for nonhuman singular finite constructions of ceym–a+tu type occurring

in 2 (3), they develop into ceyv-a+tu type occurring in 2 (2). Notice,

constructions of uṇp-a+tu type has nothing to do with ceym-a+tu type.

This is the reason why ceyv-a+tu type is morphologically differentiated

from uṇp-a+tu type.

1.5.5 The tensed finite system in Tamil referred to in (1) and (2) of

1.5.1 is a consequence of multi layer development. It may be analyzed

historically in terms of a single tier, two tier, and three tier finite systems.

The finite construction ceyt-a­+aḷ, for instance, is marked with three

tier system marked with gender suffix. The finite constructions of the

three tier system maintain subject–predicate agreement, whereas finite

constructions of the single and the two tier system being impersonal in

character do not maintain such agreement and they are only conceiveable

realities of pre Tamil linguistic system. In the finite construction ceyt-

a­+aḷ, the single tier is represented by the impersonal finite system of

ceytu type, and ceyt-a­ type represents the two tier impersonal finite

8 DYNAMICS OF TAMIL FINITE SYSTEM

system. This analysis holds good to the finite constructions of ceyt-av+aḷ,

ceyp-av+aḷ, pāṭiy-av+aḷ, pāṭup-av+aḷ, ceym-av+aḷ, *ceym-a­+aḷ, pāṭuv-

a­+aḷ and ceyv-a­+aḷ / uṇp-a­+aḷ types as well. The list can be enriched

by adding the nonpast finite constructions of ceyki­�-a­+aḷ and ceyki­�-

av+aḷ types in which ceyki­�u represents the single tier impersonal finite

system, and the impersonal finite constructions of *ceyki­�-a­ and

*ceyki­�-av types belong to the two tier finite system. In Tamil, although

we come across occasionally two tier nonhuman finite constructions such

as tō­�um-a­ (Tol. Col. Cē­ā. 389), tiriyum-a­ (Tol, E©uttu. Iḷam. 173),

muṭiyum-a­ (Tol. Col. Cē­ā. 233), by and large as impersonal types they

are not available. As for the single tier finite system represented by the

constructions of pāṭi, pāṭuv(u), ceyt(u), ceyv(u), ceyp(u), and ceym

(<*ceyum) types, they are attested in Tamil. Notice in such construction

types as ceyt-u©i, ceyv-u©i, and ceyp-u©i, the verbal constructions ceytu,

ceyvu, and ceypu occurring before u©i are context sensitive noun

modifiers. Noun modifiers are syntactically bound impersonal finite

constructions. This is very much in evidence from such constructions of

ceyt-e­a and ceyt-āṅku types where constructions of ceytu type occurring

before e­a and āṅku belong to the impersonal finite system.

1.5.6 The inflectional increments -a­ and -av figuring in the three

tier finite system is a synchronic designation. Historically, their occurrence

is restricted to the two tier finite system from which the three tier finite

system evolves. The so called increments under reference seem to be the

reflexes of *am. If this hypothesis is sustainable then the two tier finite

system may be traced to the impersonal finite system of *ceyt-am type

whose occurrence is noticed in Tamil in such constructions of ceytam-ay

and ceytam-ā�u types. Although we consider that the impersonal two

tier finite system of *ceyt-am type has developed into ceyt-a­ and

ceyt-av types, the offshoots do not share identical developments. For

instance, constructions of ceyt-av+aḷ type not only develop into ceyt-āḷ

and ceyt-ōḷ types but also has its corresponding negative constructions of

ceyyāt-av+aḷ type. However, constructions of ceyt-a­+aḷ type for

instance, have not developed in line with ceyt-av+aḷ type. The reason

seems to be that while -v of -av is susceptible to loss, the -­ of -a­ has no

scope of suffering loss. This is confirmed by the fact that synchronically

the inflectional increment -av occurring before person and gender suffixes

has a phonologically conditioned variant -a occurring before the

nonhuman singular suffix -tu.

9INTRODUCTION

1.5.7 The nonpast personal finite constructions in early Tamil are

marked with the tense marker -k- and -t- . Of these, the tense suffix -k-

occurs before -uø (I sg),and in I pl. before the personal suffixes -um, and

-am as evidenced in the examples of cey-k+uø (I sg),cey-k+um / cey-

k+am (I pl) types. As for the nonpast suffix -t-, it occurs in early Tamil

before the personal suffixes -um (I pl), -i (II sg), and -ir (II pl) as evidenced

in the examples of ceytum (<*ceyutum), ceyti (<*ceyuti), and ceytir

(<*ceyutir) types. The full paradigm of all three persons marked with the

nonpast suffixes -k- and -t- are not available in early Tamil. In middle

Tamil, we come across first person nonpast finite constructions of ceyk-

ē­ and ceyk-ōm types. E.g., nō+k-ē­ (TV. 28.1) ‘will be pained–I’,

urayk+k-ē­ (TV.28.1) ‘will tell–I’, irak+k-ē­ (TV.22.5) ‘will beg–I’,

to©u+k-ōm (TV.20.1) ‘will salute–I’ In modern Tamil, the two suffixes

under reference are noticed to occur in present finite and relative participle

constructions of Tirunelveli and Kanyakumari dialects. E.g., varukē­ /

varutē­ ‘come–I,’ varukōm / varutōm ‘come–we,’ varuke / varute ‘come–

you(sg),’ varukīr / varutīr ‘come–you(pl),’ varukā­ / varutā­ ‘comes–he,’

varukāḷ / varutāḷ ‘comes–she,’ varukāru / varutāru ‘comes–he (hon),’

varuku / varutu ‘comes–it,’ varuka / varuta+ payya­ ‘the boy who comes.’

The extended present finite constructions in the dialects under reference

have nothing to do with those of classical dialects. Consider the examples:

pēcukava­/pēcutava­ ‘the one (masc) who speaks,’ pēcukavaḷ / pēcutavaḷ

‘the one (fem) who speaks,’ pēcukavaru / pēcutavaru ‘the one (masc.

hon) who speaks,’ pēcukatu / pēcutatu ‘the one (nh) that speaks.’ In

standard spoken Tamil, the present marker -t- becomes alveolar flap -r-,

which more often than not is hyper corrected as alveolar trill in writing.

E.g., pēcutava­ > pēcurava­ / pēcarava­, pēcuta payya­ > pēcura / pēcara

payya­ ‘the boy who speaks’. The occurrence of verbal constructions

pēcarava­ and pēcara is noticed in the northern districts of Tamilnadu.

1.6 Forgotten Suffix Variants

In Tamil, there are verbal constructions of ceyt-a, ceyv-a, ceyp-a,

and cey-a types marked with the suffix -a. In early Tamil ceyt-a, ceyv-a,

ceyp-a types are noticed to function as nonhuman plural predicates. As

for the constructions of ceyt-a type, they are also noticed to assume the

function of the relative participle. Although the constructions of the

types under reference are marked with the suffix -a, traditional

grammarians have failed to identify the suffix -av as a phonologically

conditioned allomorph. Traditionally the -v of -av is treated as a glide.

10 DYNAMICS OF TAMIL FINITE SYSTEM

Since the occurrence of -v of -av is phonologically predictable, the

question of treating it as glide does not arise. For instance, the noun

phrase construction kāytta-v avaray (Aing. 286.2) ‘matured bean(s)’ is

associated with the relative participle construction kāytta. The

occurrence of -v after kāytta is conditioned by the following constituent

that begins in a vowel. We consider that the relative participle in the

construction under reference has to be identified as kāytt-av treating -

av therein as a phonologically conditioned variant of the relative suffix

-a. Since the suffix -av occurs in unstressed position, the -v of -av is

lost, and the same is recovered in sandhi before a constituent beginning

in a vowel, and elsewhere it is lost. There are nonhuman plural suffix -a

in early Tamil, and infinitive suffix -a. These suffixes also have -av as their

phonologically conditioned variant. E.g., avay vant-a ‘they came,’ kutiray

vant-av-um ce­�-av-um āyi­a ‘horses came and went,’ avay nall-a ‘they

are good,’ avay nall-av-um tīy-av-um ākum ‘they are good and bad,’ pāṭ-a

‘to sing,’ pāṭ-av-ā vant-āy ‘did you come to sing’.

1.7 Formal Distortion

1.7.1 The verbal system in early Tamil has undergone formal distortion

resulting in telescopic version. This is particularly noticed in early Tamil.

In modern Tamil, pukuntu and pukuntā­ develop into pūntu and pūntā­ in

spoken dialect. These dialect versions are a consequence of formal

distortion. There are several such instances in early Tamil which the

traditional grammatical scholarship has failed to look into them in proper

perspective. Notice such constructions as talīi (Ku�aḷ.544), ta©ūukam (Na��.

50.3), uṇpatūum (Ku�aḷ.166), ni�īi (Akam. 384.7), a�īi (Akam. 392.4),

tarūum (Akam. 6.4), tarūuntu (Pu�am. 24.9), tarūumār (Kali. 101.34),

tarīiyar (Pati. 52.3), ni­ayūu (Pati. 71.16), pi©iyūu (Akam. 8.16) and kāṇūu

(Pattup. 8.230) and several such verbal forms are identifiable as distorted

versions. Traditional grammarians generally consider that the occurrence

of short vowel after homorganic long vowel is a necessity to satisfy meter

in poetry. It is true that there are instances in early Tamil where the vowel

cluster of V1V

1 type where the second vowel is short is artificially created

to maintain metrical perfection. Thus pāl ‘milk’ for instance, with one

syllable is rendered into pāal with two syllables by adding a short vowel

after homorganic long vowel. Consequently, pāal is analyzable into pā–al

with two syllables although in natural language there is no such lexical

item as pāal. Where a syllable shortage is noticed in prosody then an

additional syllable is created by adding a short vowel after the homorganic

11INTRODUCTION

long vowel. This strategy known as ceyyuḷ-icay aḷapeṭay in traditional

parlance was extensively practised in early Tamil, and discontinued in due

course. However, not all constructions associated with vowel clusters of

V1V

1 type are poetic creations. The verbal constructions illustrated above

are a consequence of phonological changes and their development is briefly

stated as follows without making any reference to the intermediate stages

of phonological changes:

3 (a) * ta©uvi > … > ta©īi (Verbal Base : ta©uvu)

(b) * ta©uvukam > … > ta©ūukam (Verbal Base : ta©uvu)

(c) * uṇpatuv–um > … > uṇpatūum (–uv is a variant of –u)

(d) * ni�uvi > … > ni�īi (Verbal Base : ni�uvu)

(e) * a�ivi > … > a�īi (Verbal Base : a�ivu)

(f) * varukum > … > varūum (Verbal Base : varuku)

(g) * tarukuntu > … > tarūuntu (Verbal Base : taruku)

(h) * tarukumār > … > tarūumār (Verbal Base : taruku)

(i) * tarukiyar > … > tarīiyar (Verbal Base : taruku)

(j) * ni­ayupu > … > ni­ayūu

(k) * pi©iyupu > … > pi©iyūu (p > ø / V–V)

(l) * kāṇupu > … > kāṇūu

Notice the loss of labio dental fricative, velar fricative, and labial fricative

all occurring in the inter vocalic position results in the compensatory

lengthening of the preceding vowel. In this process vowel harmony rule

operates wherever necessary earlier than the loss of inter vocalic fricative

under reference. The output ni�īi from *niruvi is a case in point in this

respect. One of the nine verbal participles which Tol (Col. Cē­ā. 228)

refers to includes ceyyū type. In classical texts, several of the verbal

participles of this type is rendered as ceyyūu type as evidenced in nika©ūu

(Tol. Col. Cē­ā. 173), kāṇūu (Aing. 357.2), tēṭūu (Na��. 242.10), pi©iyūu

(Akam. 8.16), ni­ayūu (Pati. 71.16), vaḷayūu (Pattup. 4.145), and the

like, and not in the form of ceyyū type. It is to be noted that *ceyyupu

type of constructions will end up in ceyyūu type, and not in ceyyū form.

We consider that ceyyūu is the correct rendering, and the form referred

to as ceyyū in Tolkāppiyam either seems to be a later development or

might be an instance of the scribe’s error. Alternatively, ceyyūu and ceyyū

might be dialect variants. However, in classical texts the majority of the

forms of the verbal participles of the type under reference are represented

by ceyyūu form, and ceyyū versions might be a consequence of the

phonological rule -u > / ū-. The telescopic verbal constructions seem

12 DYNAMICS OF TAMIL FINITE SYSTEM

to be a development in a particular dialect and later they were inducted

in poetic dialect of classical Tamil.

1.7.2 We come across in early Tamil verbal constructions such as

ta©īiyi­a (Tol. Col. Cē­ā. 17), va©īiyi­um (Tol. Col. Cē­ā. 15), etc. In

akaval prosody the two constructions under reference are marked with

two poetic syllables each. Tol (Poruḷ. Iḷam. 325) maintains that there are

also instances in akaval prosody where the short vowel preceded by long

vowel has no prosodic significance. Consequently, the constructions

ta©īiyi­a, and va©īiyi­um will be analyzed into ta©īiy+i­a and va©īiy+i­um

with reference to prosodic convention. However, the fact is that these

two constructions are dialect variants traceable to *ta©uvi­a ‘embraced–

they(nh)’ and *va©uvi­um ‘even if it slips.’ This is yet another case where

prosodic and linguistic analyses of morphological constructions do not

go hand in hand.

1.8 Reanalysis of Verbal Suffixes

1.8.1 In Tamil, there are strong and weak conjugations. In many cases,

the products of strong conjugation are reanalyzed, and the results of such

analysis are extended to the verbs of weak conjugation. Consider for

instance, pa�ittal and a�ital of which the former is a product of strong

conjugation, and the latter has the verbal base a�i ‘to know’of weak

conjugation. Strictly speaking the verbal construction pa�ittal ‘plucking’

is analyzable into pa�itt+al, where pa�itt is the nonpast base traceable to

*pa�iyuttu (<*pa�iyuntu). However, synchronically pa�ittal is analyzed

into pa�i-t+tal treating the post verbal plosive as an instance of

nonmorphemic segment and -tal as a nominal suffix. The nominal suffix

thus identified is added after the verbal bases of weak conjugation, and as

a result the construction a�i-tal ‘knowing’ for instancec, comes into being.

We consider that in such constructions as a�i-ka ‘may one know’ and a�i-

tum ‘will know–we,’ the suffixes ka and tum are a consequence of reanalysis

of such strong conjugational constructions as aḷikka (<aḷikk+a) and aḷittum

(<aḷitt+um) into aḷi-k+ka and aḷi-t+tum. The reason why the reanalysis is

resorted to is that the verbal constructions under reference are believed to

have been conjugated from the primary verbal base with a post verbal

nonmorphemic plosive entity followed by a complex suffix with plosive

initial. This is the reason why Tolkāppiyar analyzes a�itum ‘will know–

we’ and a�ikum ‘will know–we’ into a�i+tum and a�i+kum instead of a�i-

t+um and a�i-k+um treating -tum and -kum as instances of I person plural

13INTRODUCTION

suffixes. A problem in this analysis is that there is no way of knowing or

identifying the tense morpheme since the nonpast tense markers -t- and -k-

occurring before the I person plural suffix -um are treated as integral part

of the personal suffix. However, wherever the results of reanalysis of strong

conjugation are imposed on the verbal bases of weak conjugation, the

structural anomaly seems to be inevitable.

1.8.2 Nonpast personal verbs of ceytum (I pl), ceyti (II sg), and

ceytir (II pl/hon.sg) types conjugated from the verbal bases of weak

type or traceable to the long versions of ceyutum, ceyuti, and ceyutir

types respectively. Consider, for instance, the nonpast personal

constructions as kāṇṭum (Pu�am. 173.9) ‘we (will) see,’ kāṇṭir (Akam.

76.9) ‘you (pl./hon.sg) ‘you (will) see,’ and the nonpast (impersonal)

verbal noun kāṇṭal (Pati. 12.10) ‘act of seeing.’ All these three

constructions derive from kāṇutum, kāṇutir, and kāṇutal respectively.

The latter constructions which constitute the source of the former is

not attested in early Tamil. However, in post sangam and middle Tamil,

the nonpast personal constructions of the types under reference, and

the nonpast impersonal constructions of ceytal type are noticed to occur.

A problem has to be resolved at this point. It is a question whether

ceytum (I pl), and ceytal types of the conjugation noticed in early Tamil

are a product obtained by copying and shifting of the suffixes -tum,

and -tal from the nonpast constructions conjugated from the verbal bases

of strong type. Historically, we consider that the nonpast personal verbs

of ceytum (I pl), ceyti (II sg), and ceytir (II pl/hon.sg) types, and nonpast

impersonal constructions of ceytal type conjugated from the verbal

bases of weak type are traceable to the long versions namely *ceyutum,

*ceyuti, and *ceyutir types, and the nonpast impersonal constructions

of *ceyutal type. This ruling does not seem to apply to nonpast personal

and impersonal constructions of pa�ittum (I pl), pa�itti (II sg), pa�ittir

(II pl/hon.sg), and pa�ittal of types conjugated from the verbal bases

of strong type. The expected long versions namely*pa�iyuttum (I pl),

*pa�iyutti (II sg), *pa�iyuttir (II pl/hon.sg), and *pa�iyuttal of types

are not attested in Tamil of any period. Significantly, in Kannada, we

come across constructions of bareyutt-āne ‘writes-he,’ bareyutt-āḷe

‘writes -she,’ bareyutt-āre ‘writes -they (hum.),’ etc., types. This provides

the basis to propose that the construction of pa�ittum, pa�itti, pa�ittir,

and pa�ittal types are theoretically traceable to the constructions of

*pa�iyuttum (I pl), *pa�iyutti (II sg), *pa�iyuttir (II pl/hon.sg), and

*pa�iyuttal of types respectively. The emergence of short version of

14 DYNAMICS OF TAMIL FINITE SYSTEM

the types referred to above from long versions involves the following

phonological rules.

4 (a) u > ø / y-C * pa�iyuttum > * pa�iyttum

4 (a) u > ø / y-C * a�iyutum > * a�iytum

(b) y > ø / i-C * pa�iyttum > * pa�ittum

4 (a) u > ø / y-C * a�iytum > * a�itum

(c) u > ø / u-C * cērutum > cērtum

4 (a) u > ø / y-C * kāṇutum > * kāṇtum

(c) u > ø / u-C * muyalutal > * muyaltal

(c) u > ø / u-C * ti­utal > * ti­tal

* uṇutal > * uṇtal

* velutal > * veltal

(c) u > ø / u * koḷutal > * koḷtal

(d) t > ṭ / ṇ- * kāṇtum > * kāṇṭum

* uṇtal > * uṇṭal

(e) t > � / [l, ­]- * muyaltal > * muyal�al

* payiltal > * payil�al

* ti­tal > ti­�al

(f) l > ø /-� * muyal�al > muya�al

* payil�al > * payi�al

(g) [l,ḷ]> ø /[ �, ṭ]-

(h) [CVḷ, CVl] > [CVḷ, CVl] / - t, where V is short vowel � / l- e.g.

(i) t > [ṭ, �] / [ḷ, l] -

(j) [ḷ, l] > ø / - [ṭ, �]

The nonpast impersonal finite constructions of a�it-al and pa�itt-al are

structurally and semantically identical with the impersonal constructions

of a�iv-atu and pa�ipp-atu respectively. This makes it possible to treat the

suffix -al of ceyt-al (both weak and strong) types on par with the suffix -

atu of ceyv-atu (both weak and strong) types. As for the nonpast impersonal

verbal constructions kōṭal, vē�al, cē�al and kō�al, they seem to belong to

preTamil period profusely attested in early Tamil. Kumarasamy Raja,

(1973:525-529) derives these constructions in a different manner inducting

āytam in this analysis thus: veltal > vel�al > ve«�al > vē�al. This analysis

maintains that the loss of āytam leads to the lengthening of the preceding

15INTRODUCTION

short vowel. We do not however subscribe to this analysis, since we consider

that āytam is an instance of archiphoneme representing the phonetically

voiced plosives. For a discussion see Kothandaraman (2004: 208-210).

1.8.3 Historically the strong–weak conjugational dichotomy is a later

development in Dravidian in general and Tamil in particular. Before the

emergence of this conjugational dichotomy, there was a single window

conjugational system. Notice such verbal constructions as ma�iyal, kuḷiyal,

camayal etc., belong to the single window conjugational system. In Tol

and early Tamil texts we come across such constructions as ni­ayal, nilayal,

and ka­ayal which also belong to single window conjugation. As a matter

of fact these verbal forms belong to pre Tamil period or possibly still to an

earlier period. In Tamil, these forms have been replaced by ma�ittal, kuḷittal,

camayttal, ni­ayttal, nilayttal, and ka­ayttal all of which belong to strong

conjugation. In early Tamil, we come across verbal constructions belonging

to weak conjugation which subsequently shifted to strong conjugation.

Consider the following:

Verbal Base Weak Conjugation Strong Conjugation

5 (1) tay ‘be dressed/to sew’ tay-i (Na��. 204.1) tay+ttu ‘sewed’

(2) ni­ay ‘to think’ ni­ay-i­+ay (Na��. 253.4) ni­aytt-a­+ay (II sg)

(3) tuṭay ‘to wipe’ tuṭay-i­+aḷ (Na��. 120.8) tuṭaytt–a­+aḷ (fem)

(4) ma�ay ‘be hidden’ ma�ay-i­+aḷ (Na��. 113.8) ma�aynt-a­+aḷ (fem)

(5) kuṭi ‘to drink’ kuṭiy-i­+aḷ (Na��. 271.7) kuṭitt-a­+aḷ (fem)

(6) uyir ‘to sigh’ uyir-i­+ay (Na��. 253.2) uyirtt-a­+ay (II sg)

(7) nilay ‘to stand’ nilay-i­+a (Tol.Col.Cē­ā. 56) nilaytt-a­+a (nh.pl)

There are certain other verbal bases which in early Tamil share two types

of weak conjugations involving the past markers -i and -nt as evidenced

below:

6 (1) ni­ay ‘to think ni­ay-i (Aing. 486.1) ni­ay-ntu

(2) ci­ay ‘to be angry’ ci­ay-i (Na��. 247.1) ci­a-ntu

(3) pi�a© ‘to become upside down’ pi�a©-iy+a (Pu�am. 287.8) pi�a©-nt+a

(4) iruḷ ‘to become dark’ iruḷ-i (Akam. 139.1) iruṇ-ṭu

(5) payil ‘to be familiar with’ payil-iy+atu (Ku�u. 2.3) payi­-�+atu

In Telugu, the past base of ti­(u) is ti­-i, whereas in Tamil it is ti­�u. The

past base of kēḷ(u) ‘to listen to, to ask for’ in Tamil is kēṭṭu, whereas in

Kannada it is kēḷ-i and kēḷ-idu of which the latter occurs in the finite

16 DYNAMICS OF TAMIL FINITE SYSTEM

constructions kēḷ-id+anu ‘asked–he,’ kēḷ-id+aḷu ‘asked-she,’ kēḷ-id+aru

‘asked–they(hum), etc., and in relative participle construction kēḷ-id+a.

This provides the basis to treat that the first conjugation noticed in (5)

historically earlier than the second one occurring therein.

1.8.4 In early Tamil, there are constructions of ceyyiya and ceyyiyar

types both functioning as infinitive and optative predicate. Although these

construction types are noticed in weak conjugation, occasionally they are

sighted in strong conjugation as well. E.g., kuṭikkiya (Ku�u. 356.4), aḷakkiya

(CC. 2593), viḷaykkiya (CC. 2867), mēykkiya (NDP. 3917). The suffix -iya

occurring in ceyyiya type of constructions of strong conjugation has been

copied and pasted not only after the verbal bases of weak conjugation but

also in the constructions of the strong conjugation the verbal bases of which

do not end in -i, -ay, and -y. E.g., kāṇ-iya (Ku�u. 111.7), āḷ-iya (Kali.

118.7), ukukk-iya (CC. 2636), eṭukk-iya (CC. 65), kēṭk-iya (CC. 397),

kākk-iya (KR. 7555). Notice the infinitive forms kuṭikka and kuṭikkiya are

traceable to the underlying version *kuṭiykka. The derivation of the surface

versions from *kuṭiykka proceeds thus:

7 (1) *kuṭiykka > kuṭikka (y lost)

(2) (a) *kuṭiykka > kuṭikkya (y shifted forward)

(b) kuṭikkya > kuṭikkiya (insertion of i)

We come across such developments noticed in 7 (2) in the Kanyakumari

dialect of Tamil. For a full scale discussion, see Kothandaraman (2004: 120 -

122, 264. fn). We consider that the suffix -iyar occurring in ceyyiyar type of

weak conjugation also should have been copied from ceyyiyar type of strong

conjugation and pasted after verbal bases of weak conjugation. In early Tamil

we come across constructions of ceyar type functioning as infinitive and optative.

E.g., Inf. e�iyar (Pati. 52.22); Opt. a©iyar (Akam. 212.21), teḷiyar (Akam. 303.16),

tēyar (Na��. 197.4), poliyar (Pu�am. 387.14), muṭiyar (Pu�am. 171.7).

1.8.5 There are ability verbal constructions of ceyki�p+PM and

ceyki�kum types in early Tamil. E.g., taruki�pāy (Kali. 144.49) ‘you are

able to bring / give,’ kaṭakki�pār (Kali. 81.28) ‘he(hon) is able to pass

through,’ kāṇki�pi­ (Ku�aḷ. 436) ‘if one is able to consider.’ Historically

these constructions are analyzable into ceyku-i�p+PM and ceyku+i�kum

respectively where ceyku is secondary verbal base. However, these

constructions are synchronically reanalyzed into cey+ki�p-PM and

cey+ki�k-um respectively treating ki�p and ki�k have their root in kil ‘to be

able to’. Notice there is no such natural verbal base as kil, and it is a created

17INTRODUCTION

verb due to reanalysis. It is significant that kil has no scope of being

conjugated with reference to any specific tense. As a matter of fact,

taruki�pāy, for instance, has no future reference, and this testifies that this

verbal construction is not a product of natural conjugation. We consider

that the verbal construction of ceyyiya and ceyyiyar types, as also the ability

verbal constructions of the type under reference are identifiable as telescopic

constructions of a different type.

1.9 Finite System – A Vital Player of Syntax

Syntax of any natural language has to do with the finite system that

functions as predicate. Theoretically, subordinate clause constructions also

in Dravidian are identifiable as finite syntax historically where the bound

finite system functions as predicate. The bound finite system in Dravidian

is treated as participle. The moment a verbal base is associated with a

grammatical suffix the resulting morphological construction in Dravidian

is entitled to be treated as a finite verb. Such finite verbs are of different

types of which some are noticed to occur in subordinate clause constructions

and some other finite constructions have the privilege of functioning as

predicate in simple sentences and main clause constructions. In Dravidian,

the finite verbs functioning as predicate in subordinate clause constructions

are impersonal in character. In early Tamil, even personal finite

constructions are noticed to function as predicate in subordinate clause

constructions which traditional scholarship identifies as mu��eccam. E.g.,

avayal kiḷavi ma�aytta­ar kiḷattal (Tol. Col. Cē­ā. 442) ‘the expressions

which should not be used in public should be used indirectly,’ ma�anta­am

tu�anta kā© (Na��. 172.2) ‘the seed which we left forgetting.’ The finite

constructions ma�aytta­ar and ma�anta­am in these examples are

functionally equivalent to ma�ayttu and ma�antu respectively which in

traditional parlance are verbal participles. The present work has no occasion

to speak anything about Tamil syntax. This is due to the fact that the study

of syntax should precede the study of the dynamics of finite system in

natural languages. And yet we have made reference in the present work to

two types of relative clause constructions associated with the relative verbals

am and ceyal types. E.g., cēriy-am peṇṭir (Na��. 171.4) ‘the womenfolk of

the street,’ cārntu varal marapu (Tol. E©uttu. 01) ‘the nature of coming

dependent.’ We consider that the forms –am and ceyal types noticed in

these examples are equivalent to uḷḷa and varum respectively. Of these,

the form -am not only has lost its morphology synchronically but also has

been reduced to the status of empty morph. As for the ceyal type of verbs,

18 DYNAMICS OF TAMIL FINITE SYSTEM

the traditional scholarship has overlooked its function as nonpast noun

modifier. However, significantly the grammatical works Pirayōka Vivēkam

and Ilakkaṇak kottu that belong to seventeenth century have brought to

the notice of the modern scholarship that there are relative clause

constructions in Tamil associated with nonpast relative verbs of ceyal type.

In early Tamil, there are several nonpast relative clause constructions marked

with -am and ceyal type of verbs and a detailed historical study on these

constructions is a desideratum. We have deliberated to the extent possible

the relative clause constructions of the types under reference in the present

work. What we have done and discussed in the present work is nothing but

touching the tip of the iceberg.

1.10 Problems and Resolutions

The present work has been constrained to face problems relating to

phonology and morphology of which the former includes phonemic system

obtained in Tamil. Ttraditional grammarians have included twelve vowels,

eighteen consonants, and one archiphoneme i.e. āytam besides shortened

u and i in the Tamil phonemic inventory. Of these, āytam, shortened u and

i are treated as allophones since their occurrence is predictable. In the

present deliberation we are very much concerned about the vowel inventory.

1.11 Diphthongs in Tamil –An Impossed Burden

1.11.1 Linguistically speaking, there are only five short vowels, namely

a, i, u,e, and o, and their corresponding long vowels. Contrary to this, the

tradition bound grammarians including Tolkāppiyar have included the two

diphthongs namely ai and au as long vowels of two units duration. These

diphthongs do not have their corresponding short vowels thus disturbing

the pattern congruity. As a matter of fact, not only these diphthongs are

against the genious of the Tamil phonemic system, but also there is no

place for them in Tamil or for that matter in Dravidian. The use of these

diphthongs in the Tamil writing system since the advent of literacy in Tamil

has been phonologically anomalous. It is noticed the presence of y between

the diphthong ai, for instance, and vowel. E.g., paiya­ ‘boy’, malaiya­

‘name of a chieftain.’ Notice the occurrance of y in these examples are

traditionally treated as glide. We consider that the diphthongs ai and au

are the graphological representations of ay and av. As such they can be

dispensed with in the Tamil phonemic system. This leads to the change of

the formal represtantion of paiya­ and malaiya­ into payya­ and malaya­

19INTRODUCTION

respectively. Wherever ay (= ai) occurs in the initial syllable the y, gets

geminated, and when it occurs in the noninitial syllable it does not geminate

as evidenced in malaya­. We have now reason to maintain that y occuring

in paiya­ and malaiya­ is not an instance of glide. We have some problems

in prosody where the morphemes are split and added before the succeeding

cīr. For instance, consider the prosodic construction kuṭiñai yiraṭṭum (Aing.

291.1) ‘the big owls (will) blare’. It is to be noted that the y that occurs

initially in the second cīr is an integral part of the expression kuṭiñay. For

prosodic purpose the y of kuṭiñay is shifted to the second cīr. Although

prosody reshuffles the morphemes, in actual language their original forms

are preserved. What we have deliberated for ay holds good to av as well.

1.11.2 There are also certain other constraints to replace the diphthongs

ai and au by ay and av respectively. The diphthong ai has such reflexes as

a, e, and i as evidenced in the following examples: talai (Ta) / tala (Te

and Ma) / tale (Ka). In spoken Tamil, ai is realized into a and i. E.g., talai

/ tala ‘head’, illaiyā / illiyā ‘is (it) not’. There are long and short

morphologically conditioned person and gender suffixes in Tamil namely,

ē­ / e­ (I sg.), ēm / em (I pl.),īr / ir (II pl.), ā­/a­ (masc.), āḷ / aḷ (fem.), and

ār / ar (hum.pl). Venkatarajulu Reddiyar (year is not known) maintains

that since every long person and gender suffix has its short counterpart,

the second person singular long suffix āy also should have a short

counterpart namely ay. With reference to this evidence he concludes that

there is no place for the diphthong ai in Tamil, and it is nothing but ay.

Kondal S. Mahadevan is reported to have argued that since the lexical

item vauvāl ‘bat’ has a long version namely vāval (Ku�u.172.1, Na��.87.1),

the diphthong au is not sustainable in Tamil writing system. Muthu (1994)

observes that if the transcription kai ‘hand’ with the diphthong ai is

maintainable, than mey ‘truth, body’ and koy ‘to pluck’ also can be

transcribed into mei and koi respectively. However, the latter transcription

namely mei and koi is not entertained in Tamil writing system. Hence, ai

and also au cannot be included in the vowel inventory of Tamil. There are

lexical items in Tamil such as cilampu ‘to sound, anklet’, ku©ampu ‘to be

confused, to be blended’, kalaṅku ‘to be perturbed, to be stirred’, tiraṅku

‘to shrink’, varampu ‘limit, bund’, alaṅku (Pati.58.18) ‘to sway’ etc., These

lexical items develop from the primary verbal bases cilay, ku©ay, kalay,

tiray, varay, and alay. If these primary verbal bases are associated with

the diphthong ai, then ai being a long vowel in traditional parlance has

no scope of becoming -a. On the other hand, if the primary verbal bases

under reference are identified to end in -y, then the loss of -y before

20 DYNAMICS OF TAMIL FINITE SYSTEM

nasal can be naturally explained. The arguments presented above clearly

testify that there is no place for ai and au in the Tamil writing system.

There are many instances in early and middle Tamil where (C)ai and

(C)au rhyme with (C)ay and (C)av respectively. Consider the following

the examples:

1. kaiyatu vēlē

meyyatu viyarē (Pu�am.100.1,2)

2. kaiyatu... meyyatu (Pu�am.69.1)

3. mai paṭu.

ney-kuṭai. (Pari.16.2,3)

4. kaiya�iyāmai

meyya�iyāmai (Ku�aḷ.925)

5. poyttalai

kaitta­aḷ (KR.7661.3,4)

6. kai-vēl

mey-vēl (Ku�aḷ.774)

7. mayyal ta©ai

mey-pe��a­a

moy-pe��u

kay-a��a­a (KR.7866.1–4)

8. vayva­a mu­ivar

koyva­a talaikaḷ

kay-vaḷai eyva­a (KR.7841)

It is to be noted that in (1) – (6) (C)ai rhymes with (C)ey and (C)oy and in

(7) and (8) (C)ai is replaced by (C)ay rhyming with (C)ey and (C)oy. This

confirms that ai and au are totally alien to Tamil phonemic system. In the

circumstances, not only we are constrained to do away with the so called

traditional long vowels ai and au from the Tamil phonemic system, but

also throughout the present work we replace ai and au by ay and av

respectively. For more details see Kothandaraman (2004) and

Kothandaraman (2001: 123–130).

1.12 Reinterpretation of Glides in Tamil

As for the morphemes ending in -i and -u and occuring before vocalic

expressions, traditional grammarians introduced y and v as glides. E.g.,

vali-y-āl ‘with pain,’ vali -y-eṅkē ‘where is the pain’, teru-v-il ‘in the

21INTRODUCTION

street’, teru-v-eṅkum ‘entire street.’ Historically speaking, glides under

reference are part of the preceding morpheme, be it free or bound.

Consequently morphemes ending in i and u will have phonologically

conditioned variants ending in -iy and -uv. The -y and -v occuring in the

unaccented final syllable are lost, and recovered through sandhi. This

provides the basis to treat the members of the sets kari / kariy and teru/

teruv as phonologically conditioned variants. Similarly, the past suffix -

i, the infinitive suffix -a, and the relative participle suffix -a have -iy, -av,

and -av respectively as phonologically conditioned variants. The

constructions pāṭ-i-y+a, āṭ-a+v-ā vant-āy ‘did you come to dance’, kāytt-

a-v+avaray ‘the beans which matured’, ceyt-a-v+utavi ‘the help which

one did.’ will be restructured into pāṭ-iy+a, āṭ-av+ā vant-āy, kāytt-

av+avaray, and ceyt-av+utavi respectively treating the members of the

sets i/iy, a/av as phonologically conditioned variants.

1.13 The Plosive Increments in Tamil

In external sandhi, the plosives k,(velar) c (palatal), t (dental), and p

(labial) are noticed to geminate E.g., tēr-k+kāl, ‘the leg in the wheel of

chariot’, ka­i-c+cā�u ‘fruit juice’, puli-t+tōl ‘skin of the tiger’, tāy-

p+pāl ‘mother’s milk’. These plosive incremnets are traceable to the labial

nasal in the underlying representation. There are nasal increments also

noticed in external sandhi. These increments also are traceable to labial

nasal. However, we have not resorted to this type of analysis here since it

concerns primarily with phonology. The plosive increment in the present

work are treated as instances of complers (abbreviated to C).

1.14 Composition of Past Tense Suffix -i­

The past tense marker -i­ occuring in Tamil and Telugu is a

problematic one. This suffix is generally considered a complex one

consisting of two past suffixes namely i and ­. For a discussion see

3.7.7-8. It is to be noted that although there are constructions such as

pāt-i­+a­, pāt-i­+a, pāt-i­+um, etc., we do not have pāt-i­ functioning

as past verbal participle analogous to past verbal participles of ceytu

type. However, pat-i­ is noticed to function as conditional participle in

Tamil. It is possible to treat the nasal segment -­- of -i­ as an instance of

compler i.e., cāriyay analogous to the compler -a­- that occures, for

instance, in ceyt-a­+ar ‘did -they (hum).’ This implies that -­ -of -i­,

and -a­- of ceyt-a­+ar are in complementary distribution. We leave this

22 DYNAMICS OF TAMIL FINITE SYSTEM

issue for further deliberation. In the present work, we treat -i­ as a

single unit referring to past tense.

1.15 First Person Markers om and ōm

Tol. makes reference to eight personal markers of first person plural

namely am, ām, em, ēm, kum, ṭum, tum, and �um. Of these, ṭum and �um

are phonologically conditioned variants of tum. Although Tol. treats the

last four as personal markers, k, t and its variants ṭ and � are identifiable as

nonpast markers, and –um that occurs after the nonpast markers as first

person plural suffix that belongs to proto Tamil – Telugu stage. In Telugu

–mu (<*um) is profusely attested as a first person singular suffix, and

however, it is restricted to early and post early Tamil. Significantly, the

first person plural suffixes om and ōm are not attested in early Tamil, and

there is no reference to them in Tol. as well. In spoken Tamil am and om

are conditioned variants, and ōm is retained in modern literary Tamil. E.g.

Sp. Tamil. vant–om ‘came – we’, vant–am+ā ‘did we come’; Lit. Tamil.

vant- ōm ‘came –we’.

1.16 Is a�ik-ilār a Nonpast Ability Verb

In early Tamil, we come across the nonpast negative finite constructions

of a�ik-ilar (Ku�u.152.1) ‘they will not know’, and a�ik-ilār (Ku�aḷ.1139)

types. It is not clear whether the latter is semantically different from the

former. In the present work, subject to reconsideration, we treat the

constructions of a�ik-ilār type referring to the sense of inability.

Consequently, a�ik-ilār will be translated into ‘they are / they will be unable

to know’.