Pronominal Clitics in Tocharian

UNIVERSITY OF CALIFORNIA
Los Angeles
Pronominal Clitics in Tocharian:

A Study in the Morphology-Syntax Interface
A dissertation submitted in partial satisfaction

of the requirements for the degree
Doctor of Philosophy in Indo-European Studies
by
Teigo Onishi
2022
© Copyright by
Teigo Onishi
2022
ABSTRACT OF THE DISSERTATION
Pronominal Clitics in Tocharian:

A Study in the Morphology-Syntax Interface
by
Teigo Onishi
Doctor of Philosophy in Indo-European Studies
University of California, Los Angeles, 2022
Professor Hilda Koopman, Co-Chair
Professor David Michael Goldstein, Co-Chair
This dissertation examines the pronominal clitics of Tocharian A and B and develops a model
that best accounts for their distribution. After reviewing the phonological, morphological, and
syntactic characteristics and chief uses of the Tocharian pronominal clitics, a morphosyntactic
model is developed, which accounts for the attested uses and predicts gaps in their distribution.
It predicts that the Tocharian pronominal clitics cannot represent the possessor associated with
a transitive subject or the complement of the adposition contained in another nominal expres-
sion. The model also suggests that when a pronominal clitic represents the possessor associated
with the subject of an intransitive verb, the verb belongs to the so-called unaccusative verbs.
Furthermore, it accounts for the restricted distribution of PCs in the sentences where multiple
arguments are pronominal. When the indirect and direct objects of a ditransitive predicate are
pronominal, pronominal clitics consistently represent the indirect object. The morphosyntactic
analysis advanced in this dissertation derives this distribution since a licensor, which looks for a
pronominal argument, finds the indirect object before the direct object. Tocharian pronominal
clitics sometimes co-occur with the overt nominal expression it corefers with. In such cases, the
doubling pronominal clitic indicates the doubled associate to be topical.
ii
The dissertation of Teigo Onishi is approved.
Brent Harmon Vine
Stephanie J. Watkins
Olav Hackstein
David Michael Goldstein, Committee Co-Chair
Hilda Koopman, Committee Co-Chair
University of California, Los Angeles
2022
iii
To May and Ray
iv
TABLE OF CONTENTS
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Goals of this dissertation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Tocharian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Tocharian morphosyntax and pronominal clitics . . . . . . . . . . . . . . . . . . . . 3
1.4 Structure of this dissertation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2 Pronominal clitics in Tocharian A and B . . . . . . . . . . . . . . . . . . . . . . . . . . 9

2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2 The pronominal system of Tocharian A and B . . . . . . . . . . . . . . . . . . . . . . 9
2.2.1 Overview of the pronominal system of Tocharian A and B . . . . . . . . . . 10
2.2.2 Hosts of the Tocharian PCs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2.3 The first-person singular PC -ñi . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.2.4 Secondary case markers and Tocharian PCs . . . . . . . . . . . . . . . . . . . 17
2.2.5 Gerundives followed by anäṣ/anac . . . . . . . . . . . . . . . . . . . . . . . . 22
2.2.6 Phonological characteristics of the Tocharian PCs . . . . . . . . . . . . . . . 30
2.3 Tocharian PCs and the typology of clitics and affixes . . . . . . . . . . . . . . . . . . 33
2.4 Etymology of the Tocharian PCs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.5 Interim summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3 Descriptive data of the Tocharian pronominal clitics . . . . . . . . . . . . . . . . . . 42

3.1 Chief usages of the Tocharian pronominal clitics . . . . . . . . . . . . . . . . . . . . 42
3.1.1 Theme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.1.2 Addressee . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.1.3 Recipient . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.1.4 Goal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.1.5 Beneficiary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3.1.6 Source . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.1.7 Location . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
v
3.1.8 Experiencer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.1.9 Stimulus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
3.1.10 Argument of a complex predicate . . . . . . . . . . . . . . . . . . . . . . . . . 51
3.1.11 Experiencer of a complex predicate . . . . . . . . . . . . . . . . . . . . . . . 55
3.1.12 Oblique possessor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
3.1.13 Possessor associated with a direct object (theme) . . . . . . . . . . . . . . . 60
3.1.14 Possessor associated with an indirect object . . . . . . . . . . . . . . . . . . 64
3.1.15 Possessor associated with an argument of an adverb/postposition and a
verb . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
3.1.16 Possessor associated with a subject . . . . . . . . . . . . . . . . . . . . . . . . 65
3.1.17 Possessor associated with a noun connected by a copula . . . . . . . . . . . 66
3.1.18 Subject of a non-finite verb . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
3.1.19 Agent of a mediopassive . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
3.1.20 Causee . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
3.1.21 Doubling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
3.2 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
3.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
4 Theoretical premises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
4.2 Merge and Transfer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
4.3 Distributed Morphology and Vocabulary Insertion . . . . . . . . . . . . . . . . . . . 82
4.4 Linearization and antisymmetry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
4.5 Structure of a pronominal clitic, Agree, and defective goals . . . . . . . . . . . . . . 87
4.6 Example of derivation of a canonical SOV structure . . . . . . . . . . . . . . . . . . . 91
4.7 Interim summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
5 Split intransitivity and clitic distribution in Tocharian . . . . . . . . . . . . . . . . 98

5.1 Formal Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
5.1.1 PC representing a direct object (theme) of a transitive verb . . . . . . . . . 98
vi
5.1.2 PC representing an indirect object with various thematic roles . . . . . . . 106
5.1.3 PC representing a possessor . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
5.1.4 Interim summary and predictions . . . . . . . . . . . . . . . . . . . . . . . . 121
5.2 Implications for split intransitivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
5.3 Antigrundverbs and unaccusatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
5.4 External possession as a diagnostic for unaccusativity . . . . . . . . . . . . . . . . . 142
5.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
6 Pronominal clitics and multiple pronominal arguments . . . . . . . . . . . . . . . . 149

6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
6.2 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
6.2.1 Examples (TB) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
6.2.2 Examples (TA) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
6.2.3 Interim summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
6.3 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
6.4 Implications and further questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
6.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
7 Clitic Doubling in Tocharian A and B . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182

7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
7.2 Topic and the Question Under Discussion model of discourse . . . . . . . . . . . . . 184
7.2.1 What is a topic? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
7.2.2 Discourse-new referents may be a topic . . . . . . . . . . . . . . . . . . . . . 186
7.2.3 The Question Under Discussion model of discourse . . . . . . . . . . . . . . 188
7.2.4 Secondary topic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
7.3 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
7.4 Hypothesis and predictions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
7.5 Clitic Left Dislocation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
7.5.1 Clitic Left Dislocation in Tocharian A . . . . . . . . . . . . . . . . . . . . . . . 197
7.5.2 Clitic Left Dislocation in Tocharian B . . . . . . . . . . . . . . . . . . . . . . . 203
vii
7.6 Clitic Right Dislocation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
7.6.1 Clitic Right Dislocation in Tocharian A . . . . . . . . . . . . . . . . . . . . . . 208
7.6.2 Clitic Right Dislocation in Tocharian B . . . . . . . . . . . . . . . . . . . . . . 210
7.7 Interim summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
7.8 Clitic Doubling Proper . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
7.8.1 Doubling of a theme of a transitive verb . . . . . . . . . . . . . . . . . . . . . 218
7.8.2 Doubling of a possessor of a direct object . . . . . . . . . . . . . . . . . . . . 226
7.8.3 Doubling of a possessor of an intransitive subject . . . . . . . . . . . . . . . 235
7.8.4 Miscellanea . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
7.8.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252
7.9 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252
8 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254
8.1 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254
8.2 Further questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255
9 Appendix: Representative unaccusative verbs in Tocharian A and B . . . . . . . . 258

9.1 Representative unaccusative verbs in TA and TB . . . . . . . . . . . . . . . . . . . . . 258
9.1.1 TB √ār(ā)- ‘cease, come to an end’ . . . . . . . . . . . . . . . . . . . . . . . . 258
9.1.2 TA √ār(ā)- ‘cease, come to an end’ . . . . . . . . . . . . . . . . . . . . . . . . . 260
9.1.3 TA √i-|kälkā- ‘go’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
9.1.4 TB √kän- ‘come about, occur, be fulfilled’ . . . . . . . . . . . . . . . . . . . . 264
9.1.5 TA √kän- ‘come about, occur’ . . . . . . . . . . . . . . . . . . . . . . . . . . . 266
9.1.6 TB √käly-|stäm(ā)- ‘stand’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268
9.1.7 TB √käsk(ā)- ‘be scattered’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
9.1.8 TB √kulā- ‘recede’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
9.1.9 TB √klänts(ā)- ‘sleep’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273
9.1.10 TB √klautk(ā)- ‘turn, become’ . . . . . . . . . . . . . . . . . . . . . . . . . . . 274
9.1.11 TA √trik(ā)- ‘be confused; faint’ . . . . . . . . . . . . . . . . . . . . . . . . . . 275
9.1.12 TB √nes-|tāk(ā)- ‘be, become’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276
viii
9.1.13 TA √nas-|tāk(ā)- ‘be, become’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276
9.1.14 TA √pränk? - ‘restrain oneself ’ . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
9.1.15 TA √prutk(ā)- ‘be shut; be filled’ . . . . . . . . . . . . . . . . . . . . . . . . . . 278
9.1.16 TB √plätk- ‘overflow, develop, arise’ . . . . . . . . . . . . . . . . . . . . . . . . 278
9.1.17 TB √plu- ‘float, fly, soar’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
9.1.18 TB √mäsk- ‘be’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 280
9.1.19 TA √mäsk- ‘be’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281
9.1.20 TA √lotkā- ‘turn, become’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
9.1.21 TB √wāk(ā)- ‘differ’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
9.1.22 TB √ṣäm-|läm(ā)- ‘sit’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
9.1.23 TA √sätk(ā)- ‘spread out’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
9.1.24 TB √si-n- ‘satiate oneself, be depressed’ . . . . . . . . . . . . . . . . . . . . . 284
9.1.25 TB √spālkā? - ‘±strive actively/forcefully for’ . . . . . . . . . . . . . . . . . . 284
9.1.26 TB √tsäm(ā)- ‘grow, increase, come into being’ . . . . . . . . . . . . . . . . . 287
9.1.27 TB √tsälp(ā)- ‘pass away, be released, be redeemed’ . . . . . . . . . . . . . . 289
9.2 Limitation 1: Verbs of appearance/disappearance . . . . . . . . . . . . . . . . . . . . 289
9.2.1 TA/TB √näk-mid ‘fall into ruin, disappear’ . . . . . . . . . . . . . . . . . . . . 290
9.2.2 TA/TB √wik(ā)- ‘disappear’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292
9.2.3 TA/TB √spärk(ā)- ‘disappear, perish’ . . . . . . . . . . . . . . . . . . . . . . . 293
9.2.4 TB √naut(ā)- ‘disappear’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293
9.2.5 TB √musk(ā)- ‘disappear, perish’ . . . . . . . . . . . . . . . . . . . . . . . . . . 297
9.2.6 TA √pyutk-act ‘come into being’ . . . . . . . . . . . . . . . . . . . . . . . . . . 297
9.2.7 TA √kātkā- ‘(a)rise’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 298
9.2.8 TA √pärkā- ‘(a)rise; become clear’ . . . . . . . . . . . . . . . . . . . . . . . . . 298
9.2.9 TA/TB √läk(ā)-mid ‘appear; be seen’ . . . . . . . . . . . . . . . . . . . . . . . . 299
9.2.10 TA/TB √täm- ‘be born, come into being’ . . . . . . . . . . . . . . . . . . . . . 303
9.2.11 TB √tsäṅkā- ‘(a)rise’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 304
9.2.12 TA/TB √länt- ‘go out, emerge’ . . . . . . . . . . . . . . . . . . . . . . . . . . . 308
9.3 Limitation 2: Further ambiguous examples . . . . . . . . . . . . . . . . . . . . . . . . 310
ix
9.3.1 TB √kärstā- ‘cut off’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310
9.3.2 TA √klawā- ‘fall’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311
9.3.3 TA √tkälā? - ‘illuminate’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312
9.3.4 TB √mänk(ā)- ‘be inferior, lack, be deprived of ’ . . . . . . . . . . . . . . . . . 312
9.3.5 TB √re(-sk)- ‘flow’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313
9.3.6 TB √lik(ā)- ‘wash’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313
Index of passages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315
x
LIST OF FIGURES
2.1 〈ṣarpau me〉 in PKAS 6C a4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
6.1 Valency of host predicates in TA and TB . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150

6.2 Frequently attested ditransitive predicates in TA . . . . . . . . . . . . . . . . . . . . . . 151
6.3 Frequently attested ditransitive predicates in TB . . . . . . . . . . . . . . . . . . . . . . 151
xi
LIST OF TABLES
1.1 Pronominal clitics in Tocharian A and B . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.1 Independent personal pronouns in Tocharian A . . . . . . . . . . . . . . . . . . . . . . . 10

2.2 Independent personal pronouns in Tocharian B . . . . . . . . . . . . . . . . . . . . . . . 11
2.3 Demonstrative (anaphoric) pronouns in Tocharian A and B . . . . . . . . . . . . . . . . 11
2.4 Pronominal clitics in Tocharian A and B . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.5 Secondary case markers in Tocharian A and B . . . . . . . . . . . . . . . . . . . . . . . . 18
3.1 Summary of the chief uses of the Tocharian PCs . . . . . . . . . . . . . . . . . . . . . . . 75
5.1 Labile verbs in TA and TB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140

5.2 The Auxiliary Selection Hierarchy (Sorace 2000) . . . . . . . . . . . . . . . . . . . . . . . 145
6.1 Hypotheses and predictions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149

6.2 Major ditransitive predicates in TB and TA . . . . . . . . . . . . . . . . . . . . . . . . . . 152
6.3 Summary of the examples in TB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
6.4 Summary of the examples in TA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
6.5 Summary of (6.27) and (6.28) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
6.6 Summary of (6.29) and (6.30) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
6.7 Summary of Prediction 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
6.8 Varieties of the Person Case Constraint (based on table 1 of Compton 2019: 595) . . . . 179
7.1 Summary of (7.17ii) and (7.17iii) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191

7.2 Hypotheses and predictions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
7.3 Clitic doubling the theme of a transitive verb in TA . . . . . . . . . . . . . . . . . . . . . 223
7.4 Clitic doubling of the theme of a transitive verb in TB . . . . . . . . . . . . . . . . . . . . 227
7.5 Clitic doubling of a possessor of an intransitive subject in TB and TA . . . . . . . . . . . 241
7.6 Clitic doubling of an IO or a possessor of an IO in TB and TA . . . . . . . . . . . . . . . . 251
9.1 Representative unaccusative verbs in TA and TB . . . . . . . . . . . . . . . . . . . . . . . 259

9.2 Attestations of TB √spālkā? - and TA √spāltkā? - . . . . . . . . . . . . . . . . . . . . . . . . . 286
xii
9.3 Verbs of appearance/disappearance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
9.4 Ambiguous intransitives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310
xiii
Glossary
dim diminutive
1 first person
du dual
2 second person
emp emphatic
3 third person
f feminine
abl ablative
gdv gerundive
abs absolutive
gen genitive
acc accusative
imp imperative
act active
impf imperfect
adj adjective
indf indefinite
adjz adjectivizer
inf infinitive
all allative
ins instrumental
caus causative
loc locative
com comitative
m masculine
comp complementizer
mid middle
conj conjunction
n neuter
cop copula
neg negative
dat dative
nmlz nominalizer
dem demonstrative
nom nominative
det determiner
xiv
npst nonpast ptcp participle
opt optative refl reflexive
perl perlative rel relative
pl plural sg singular
prev preverb subj subjunctive
pst past top topic
ptcl particle voc vocative
xv
ACKNOWLEDGMENTS
First, I would like to thank my dissertation committee members, especially my co-chairs: David
Goldstein and Hilda Koopman. Without their continuous generous support, I wouldn’t have been
able to finish this dissertation. I would also like to thank Brent Vine, Stephanie Jamison, and
Olav Hackstein, not just for their helpful feedback and encouragement but also for their support
throughout the stages of my graduate study. Brent has always been supportive since I was an
undergraduate student in Kyoto who took his course on the historical grammar of Greek and
Latin, the course which attracted me to the world of Indo-European Studies. Reading Vedic with
Stephanie was one of my favorite moments at UCLA. I am also grateful for her encouragement
to present my ideas at the AOS meetings. I sincerely appreciate Olav, who hosted my visit to
Munich during the summer of 2018, read Tocharian texts with me, and was kind enough to serve
as my dissertation committee member. I would also like to thank the LMU colleagues who made
my stay in Munich unforgettable: Dieter Gunkel, Ryan Sandell, Chiara Bozzone, Sergio Neri, Ville
Leppänen, Albert Zasada, Christoph Bross, Tao Pan, Giulio Imberciadori, and Benedikt Peschl.
I owe so much to Kazuhiro Yoshida, who first introduced the world of Indo-European Studies to
me. I was able to continue thanks to his generous support in all stages of my academic life. I
am also grateful to my teachers in Kyoto: Yutaka Yoshida, who I really enjoyed reading Avestan
with; Yukinori Takubo, from who I learned theoretical tools; and Haruyuki Saito, who was so
kind to spare time to read Tocharian texts with me. I would also like to thank my colleagues
in Kyoto, including, but not limited to, Kanehiro Nishimura, Yusuke Kanazawa, Adam Catt, Chie
Nakamura, Yoko Yamazaki, and Junichi Takahashi. Adam kindly read the first three chapters of
my dissertation and gave me extensive feedback.
I was fortunate to present some of my graduate research in the workshops held in Vienna in
2020 and online (hosted by Leiden) in 2021. I thank Melanie Malzahn, Hannes Fellner, Bernhard
Koller, Georges-Jean Pinault, Gerd Carling, Michaël Peyrot, Abel Warries, and Ron Kim for their
valuable comments. Bernhard kindly helped me collect data and develop my idea when it was
very preliminary.
xvi
I am sincerely grateful to Tony Yates and Jesse Lundquist, who gave me various pieces of advice in
many stages of my doctoral study at UCLA. Tim Hunter and Dominique Sportiche kindly shared
their thoughts on clitics with me. Toru Minamimoto has always been very accommodating and
helpful with any problems I have encountered. I would also like to thank all of the colleagues and
friends of PIES, including, but not limited to, Jessica DeLisi, Mattyas Huggard, Kaspars Ozoliņš,
Éloïse Lemay, Thomas Motter, Maggie Eismeier, Corey Barnes, Benjamin Niedzielski, Ian Hollen-
baugh, Anahita Hoose, Canaan Breiss, Chengzhi Zhang, John Clayton, Alex Roy, Valentina Lunardi,
Paolo Sabattini, Denys Cennet Planchard, Elisa Migliaretti, Nadège Rollet, Milena Anfosso, and
Véronique Kremmar.
Finally, I wish to thank all of my family members who have always understood and supported me
in my journey. Especially, I would like to thank my wife May and my son Ray for always being
there for me.
xvii
VITA
2009 B.A. in Linguistics, Kyoto University
2011 M.A. in Linguistics, Kyoto University
2015–2017 Fulbright Fellowship, UCLA, the United States of America
2016 Teaching Assistant, Department of Classics, UCLA
2016 Graduate Summer Research Mentorship (under Prof. Brent Vine), UCLA
2016–2021 Teaching Assistant, Department of Linguistics, UCLA
2017 Graduate Summer Research Mentorship (under Prof. Brent Vine), UCLA
2021 Harry and Yvonne Lenart Graduate Travel Fellowships, UCLA
2021–2022 Dissertation Year Fellowship, UCLA
PUBLICATIONS
forthcoming. “Clitic Doubling in Tocharian B” In David M. Goldstein, Stephanie W. Jamison, and

Brent Vine (eds.). Proceedings of the 32nd Annual UCLA Indo-European Conference, Bremen: Hempen.
2021. “Umbrian <rs> and <rf >: a synchronic and diachronic puzzle” Indo-European Linguistics
9:203–33.
& Kanehiro Nishimura. 2019. “How to Handle the Proto-Indo-European Suffix *-eh2 - in Latin
xviii
Compounds: Tībīcen-type and Armiger-type Treatments” Historische Sprachforschung 132:208–24.
2018. “Latin hībernus and the Development of *(-)mr- in Latin” In David M. Goldstein, Stephanie
W. Jamison, and Brent Vine (eds.). Proceedings of the 29th Annual UCLA Indo-European Conference,
203–21. Bremen: Hempen.
& Kanehiro Nishimura. 2016. “Inseparable Etymology: Latin crīnis, Greek κορέω, and Related
forms in Germanic” In David M. Goldstein, Stephanie W. Jamison, and Brent Vine (eds.). Proceed-
ings of the 27th Annual UCLA Indo-European Conference, 131–42. Bremen: Hempen.
2015. “On the Indo-European suffix *-ti- in Latin: An analysis with a focus on the relationship be-
tween vowel grades and root structures” (in Japanese). Kyoto University Linguistic Research 33:135–65.
xix
CHAPTER 1
Introduction
1.1 Goals of this dissertation
There are two goals of this dissertation: one is to collect descriptive data on the distribution of
pronominal clitics in Tocharian, and the other is to formulate an explanatorily adequate theory
that not only accounts for all of the descriptive facts but also predicts the existence and absence of
certain types of constructions and serves as a basis for subsequent comparative morphosyntactic
reconstruction of Proto-Indo-European.
In reconstructing the morphosyntactic system of a proto-language from attested languages X and

Y, we need to answer the following questions either implicitly or explicitly: How much is X’s mor-
phosyntactic pattern similar to or different from that of Y? Is the similarity due to inheritance,
shared innovation, or parallel innovation? How do morphosyntactic innovations occur? How do
they develop diachronically? To answer these questions, one needs to have a good understanding
of the synchronic morphosyntactic system of X and Y (cf. “you’ve got to know what to compare”;
Watkins 1976: 249). Just as one needs to have a good knowledge of the synchronic phonological
system of a target language in order to identify the diachronic sound changes it has undergone,
one also needs to understand the synchronic morphosyntactic system of that language in order to
postulate the diachronic morphological or syntactic changes that account for the empirical data.
This is in the same spirit of Watkins (1976: 249), who emphasized the importance of a “more-
highly bound or restricted construction” when reconstructing a morphosyntactic pattern for a
proto-language. To achieve the goal of historical-comparative reconstruction, that is, to explain
“how a set of attested languages came to be the way they are” (Melchert, forthcoming), first we
1
need a better understanding of the morphosyntactic system of the attested languages. That be-
ing said, it is not an easy task to analyze the synchronic morphosyntactic system of an extinct
language since unlike living languages, negative evidence is not available: what is available to us
is only the language as presented by the writers of the source documents. Therefore, to build a
hypothesis we need to focus on the distribution, context and variation of those restricted con-
structions emphasized by Watkins. Just as a successful synchronic theory of morphosyntax may
explain all of the descriptive facts and make falsifiable predictions based on an independently
motivated model, a successful theory of diachronic morphosyntactic change should also be able
to explain all of the synchronic models that we postulate for the attested languages and make
falsifiable predictions about what morphosyntactic changes should and should not be found.
This dissertation aims to lay out the descriptive facts of Tocharian and develop a synchronic
morphosyntactic model which best explains the empirical data, and makes predictions about the
presence and absence of certain types of data. This in turn will form a basis for subsequent re-
search on reconstructing the morphosyntactic system of Proto-Tocharian, Proto-Nuclear-Indo-
European (PNIE), and Proto-Indo-European (PIE).1
1.2 Tocharian
Tocharian is one of the subgroups of the Indo-European language family and comprises two at-
tested languages: Tocharian A and B (TA and TB). Manuscripts written in these languages were
found in the present-day Xinjiang Uyghur Autonomous Region of the People’s Republic of China.
They were collected and published in the nineteenth and the twentieth century, and digitized in
the twenty-first century.2 The majority of scholars consider Tocharian to be the second clade
to branch off from PIE since it displays (morphological) innovations that are shared with the core
Indo-European languages but are absent in Anatolian and since it lacks some of the innovations
1. In this dissertation I assume that Anatolian was the first group to branch off from PIE. The rest from which
Anatolian was separated is labelled as Proto-Nuclear-Indo-European. I also use the term “core Indo-European” to
refer to the IE languages excluding Anatolian and Tocharian.
2. The digital corpus of Tocharian is accessible online (A Comprehensive Edition of Tocharian Manuscripts [CE-
ToM]: https://www.univie.ac.at/tocharian/?home).
2
that are ubiquitously found in the core Indo-European languages.3 Regarding the former, for
example, unlike Anatolian, TA and TB have *so-/*to-pronouns as in the core Indo-European lan-
guages. As for the latter, Tocharian class III preterites have a preterite-stem-building suffix -s-
in the middle paradigm and the third-person singular of the active paradigm. In contrast, this
sibilant is found in all cells in the active paradigm of the core Indo-European languages. The com-
munis opinio is that after Tocharian branched off, the core IE languages underwent the leveling of
*s throughout the active paradigm (Jasanoff 2003).
Tocharian morphosyntax is still less well understood compared to other ancient IE languages
such as Greek and Vedic Sanskrit. It is without doubt that a better understanding of the syn-
chronic morphosyntactic system of TA and TB and the reconstructed morphosyntactic system of
Proto-Tocharian will contribute significantly to the reconstruction of PIE, PNIE and its descen-
dants. At the same time, however, we face challenges since Tocharian is attested much later than
the earliest attestation of the ancient Indo-European languages such as Hittite or Vedic Sanskrit.
The earliest document of Tocharian B is dated to around the 4th or 5th century CE, and we do
not know, for example, what the phonological or morphological system of its ancestral language
was like in 1,000 BCE. Therefore, while we can pinpoint important archaisms in Tocharian that
were lost in the other Indo-European languages, we should also expect to see a number of in-
novations.4 It is therefore crucial for us to attempt to separate archaisms from innovations.
From this perspective also, a better understanding of the synchronic system of Tocharian is a
desideratum in the field of Indo-European studies.
1.3 Tocharian morphosyntax and pronominal clitics
It is fair to say that there has been much more research in Tocharian on historical sound changes
and synchronic phonology than there has been on morphosyntactic change.5 While the histori-
cal phonology of Tocharian has been continuously investigated since the twentieth century (e.g.,
3. See, e.g., Schmidt (1992), Winter (1997), Carling (2005), Jasanoff (2020), and Weiss (2018); for a different view,
see Malzahn (2016b).
4. For example, Melchert (1978) points out that Tocharian acquired several verbal roots ending in -tk- from the
prehistoric agglutination of the present-stem marking suffix *-sḱe/o- with the root.
3
Pedersen 1941; Couvreur 1947; Adams 1988; Ringe 1996), it is only recently that research on
Tocharian morphosyntax has begun (summarized in Adams 2015). Although several studies
which focus on the Tocharian verbal system from a synchronic and diachronic perspective are
available (Krause 1952; Schmidt 1974; Hackstein 1995; Malzahn 2010; Peyrot 2013b), an extensive
survey on the Tocharian nominal system has yet to be carried out.6
Moreover, many aspects of Tocharian morphosyntax still remain to be worked out, for example,
derivational morphology, morphophonological alternation (ablaut, suppletion) and its interac-
tion with locality, causatives (cf. Seržant 2014; Malzahn 2016a), second-position clitics (cf. Winter
1959; Malzahn 2012), left periphery (Koller 2013), topic/focus constructions (e.g., verb-fronting:
Hackstein 2013, 2015), wh-movement (e.g., Adams 2015; Hearn 2017), relative clauses (e.g., Pinault
1997), and discontinuous constituents, to name a few.
Pronominal clitics (PCs) are one of these important topics that remain to be investigated. In all
cases in TB and in most cases in TA, pronominal clitics attach to a finite verb and express various
syntactic relations to the verb, including direct and indirect object, and possessor of an argument.
For example, in (1.1), the third-person singular PC -ne marks the indirect object of the verb śilāre
‘brought’.7
5. Several handbooks on synchronic Tocharian grammar are now available: TEB I, Pinault (1992, 2008) and Adams
(2015); Schulze, Sieg, and Siegling (1931) for TA. on the geographic and diachronic variation in TB, see Peyrot (2008).
6. For studies on the Tocharian locative and genitive-dative case, see Carling (2000) and Meunier (2015), respec-
tively. Preverbs and adpositions are discussed in Hackstein (1997) and Kuritsyna (2016). On the use of demonstrative
pronouns in Tocharian, see Stumpf (1971) and Kümmel (2015).
7. Peyrot (2017, 2019) states that pronominal clitics mark the object of the verb. However, marking the direct or
indirect object is only one of the several attested usages (Chapter 7). In many cases, PCs do not represent the direct
or indirect object but the possessor associated with the object as in (1.a), where the second-person singular PC -c
represents the possessor associated with the theme argument ersna ‘form = Skt. rūpa.’ If the pronominal clitic was
merely marking the object of the verb we would expect to find 3sg -ne rather than -c in this case.
(1.a) -c = possessor associated with the direct object ersna ‘shape’
tot yam-c ñakta śaran-ne ¦ asta-n= eṣke mrestiwe-śc :
so go.npst.act.1sg-2sg lord.voc refuge-loc bone-loc into marrow-all
indri-nta-ṣṣeṃ semen-sa ¦ yoku-c ersna snai [b3] (so)ylyñe :
sense-pl-adjz.acc.pl ladle.pl-perl drink.subj.npst.act.1sg-2sg form without satisfaction
‘… so often I go into your protection, o god, up to the marrow in the bones. [23b] With the ladles of the senses,
I will drink your (beautiful) form without being satiated. [23c] ’
(B241b2; verse [7¦7]×4)
4
(1.1) [TB] -ne = indirect object (goal) of śilāre ‘brought’
śi[a8]lāre-ne oṅkarñai ¦ wñār-ne purwar
bring.pst.act.3pl-3sg porridge.acc.sg speak.pst.act.3pl-3sg accept.imp.mid.2sg
wesan-meṃ ¦ pinwāt rṣāka : 1 ||
1pl-abl alms sage.voc
‘(Nānda and Nandābala) brought the porridge to him (= Indra) and said to him: “Receive
the alms from us, o sage!”’
(B107a7-8; verse; [7¦7¦4]×4)
Previous studies and handbooks have also noted these various uses and have attempted to make
a generalization from the data. For example, Krause (1952: 204–6) and TEB I: 162–3 state that
PCs may stand for the accusative or genitive-dative in all its syntactic relations.8 Pinault (1992:
113; 2008: 537) describes PCs as being reserved for the functions of “complément”.9 Carling
(2006: 44) briefly reviews the attested examples of PCs and concludes that they can be used in all
“syntactic core positions” except for the subject (A or S).10 Adams (2015: 149–51) describes that
one of the functions is “as alternatives to genitive pronouns (whose head noun may or may not
be a part of the verb phrase)”.
These studies did not aim to discover the underlying principle that regulates the various uses
of PCs but simply seek to make some generalizations from the empirical data. Therefore, these
descriptions have no predictive power and we are still in need of an explanation as to why the
Tocharian PCs behave the way they do. For example, while Carling (2006: 44) notes that the
subject of an intransitive verb (S) cannot be represented by a PC, she also observes that a PC can
be used as a dative subject in the so-called mihi est construction. It is not immediately clear what
8. “Zur Syntax der suffigierten Pronomina ist in Kürze folgendes zu bemerken: 1. Das Pron. suff. steht für den
Akkusativ. […] 2. Das Pron. suff. steht im Sinn eines Genetiv-Dativs mit all seinen syntaktischen Beziehungen […]”
(Krause 1952: 204); “Zur Syntax ist zu bemerken: 1. Das Pron. suff. steht für den Akkusativ […] 2. Im Sinne eines
Genitiv-Dativs mit allen seinen syntaktischen Beziehungen” (TEB I: 162–3)
9. “Ces pronoms enclitiques sont réservés aux fonctions de complément : complément d’objet direct, complé-
ment d’attribution, génitif possessif portant sur l’objet ou le sujet du verbe, génitif-datif d’agent.” (Pinault 2008:
537)
10. A and S represent the subject of a transitive and an intransitive verb, respectively.
5
licenses PCs in the latter but not in the former. Likewise, Adams (2015: 21) notes that if both the
direct and indirect objects are pronouns, “it seems that the indirect object is favored” to surface
as a PC, but none of the generalizations formulated predicts his observation to be true or false. It
is now clear that we need to develop a model that not only explains the distributional facts but
also makes predictions as to what kind of constructions should or should not be allowed. Such a
model will reveal a more fine-grained distribution of the PCs in TA and TB than those which are
obtained by generalizing from the data.
In addition, a better understanding of PCs in Tocharian will be helpful for a better understanding
of their diachrony. The Tocharian PCs do not resemble those of ancient Indo-European languages
such as Hittite, Homeric Greek, Vedic Sanskrit, and Avestan in the following three respects. First,
while a single sentence may have more than one pronominal clitic in these languages, there is
no sentence in which a host carries multiple PCs in Tocharian (Chapter 2). Second, pronominal
clitics in the ancient Indo-European languages usually have an IO and DO distinction. In contrast,
the Tocharian PCs do not make a distinction in case (table 2.4). Third, pronominal clitics in the
core-IE languages target the so-called second position in the clause (Wackernagel’s Law; Delbrück
1878: 47–8, Wackernagel 1892). The Tocharian PCs, however, do not target second position but
mostly appear immediately after the finite verb (Chapter 2).
1sg 2sg 3sg pl

TA -ñi -ci -ṃ -m
TB -ñ -c -ne -me
Table 1.1: Pronominal clitics in Tocharian A and B
Are these characteristics innovations? If so, how did they develop? First we need to consider if
there is any diachronic model which may account for such morphosyntactic change. Only after
this task has been fulfilled can we assess if language contact with non-Indo-European languages,
such as Uralic, played a role in shaping the Tocharian languages (Peyrot 2019),
As of preparing this dissertation, only a brief descriptive survey of the Tocharian pronominal
clitics has been carried out (Carling 2006). An extensive study as to what their precise usage is,
what principle lies behind it, and how they are similar to or different from weak or clitic pronouns
6
found in other ancient Indo-European languages is a desideratum.
In researching this topic, there are a number of questions that may come to mind which can be
roughly divided into two categories: those related to the synchronic aspect and those related to
the diachronic aspect.
1. (The synchronic aspect): What are the pronominal clitics? When can we use/not use them?
Why do they behave in the way they do? What principle(s) govern their usage? Are there
any typological parallels in which suffixed pronouns are used in a similar manner?
2. (The diachronic aspect): How did the pronominal clitics develop? How are they similar
to or different from weak/clitic pronouns in other IE languages? Did (contact-induced)
grammaticalization play a role in their development?
We may answer the second group of questions only by answering the first. This dissertation
tackles the first set of questions, with the hope that subsequent research can clarify the second.
1.4 Structure of this dissertation
The structure of this dissertation is as follows. The following chapter reviews the pronominal
system and the pronominal clitics of TA and TB. I survey the phonological, morphological and
syntactic behavior of pronominal clitics in this chapter. Chapter 3 reviews the attested usages of
PCs. It shows that they are multifunctional and may represent the direct object, indirect object
with various thematic roles, argument of a predicate consisting of a verb and an adverb, agent
of a non-finite verb, causee, possessor of a direct or indirect object, possessor of a subject of an
intransitive verb, and so on. Chapter 4 introduces the theoretical premises on which my subse-
quent morphosyntactic model is built. I assume that a sentence has a hierarchical structure built
in the syntax, and that PCs spell out person and number features. Chapter 5 develops a mor-
phosyntactic model that accounts for the multifunctionality of the Tocharian PCs and predicts
some gaps in the data. The analysis I develop has implications for our understanding of split in-
transitivity in Tocharian: it is argued that by focusing on the distribution of PCs, we can single
out unaccusative verbs from unergatives in Tocharian. This analysis opens up a way to under-
7
stand split intransitivity in Tocharian, the criteria for which have been hitherto unknown. Using
the proposed diagnostic, representative unaccusative verbs are collected (Appendix I). Chapter 6
discusses predictions extracted from the analysis in Chapter 5. I show that when an IO and a DO
are both pronominal, the PC always refers to the IO. My model predicts this since the licensor of
a PC first finds an IO before a DO. Chapter 7 treats cases in which a PC co-occurs with the nominal
expression that it refers to and hence appears to be redundant. We show that the doubling of a
nominal expression with a PC indicates the doubled nominal expression to be topical. In other
words, discourse participants presuppose the existence of the doubled nominal expression prior
to the utterance of the speaker. Chapter 8 summarizes the preceding chapters and addresses
remaining questions and problems.
8
CHAPTER 2
Pronominal clitics in Tocharian A and B
2.1 Introduction
As a preliminary to the analysis in the following chapters, this chapter reviews basic facts re-
garding the pronominal clitics (PCs) of Tocharian A (TA) and B (TB). In addition to personal and
demonstrative pronouns, TA and TB have a set of PCs. These lack case distinctions and, in the
plural, also person distinctions. While PCs are consistently hosted by a finite verb in TB, partici-
ples, gerundives, and nouns in a nominal clause may also host a PC in TA. PCs may be followed by
an allative and ablative marker, and in this case TA attests peculiar allomorphs. PCs in TB form a
single phonological word with their host, while the host-PC connection seems to be weaker when
the host of a PC is a non-finite form in TA. Tocharian PCs show a mixed behavior with respect to
the typology of clitics and affixes.
The structure of this chapter is as follows. Section 2.2 surveys the pronominal system of TA and
TB and reviews the characteristics of the Tocharian PCs. Section 2.3 considers the nature of the
Tocharian PCs with respect to the typology of clitics and affixes. It is argued that Tocharian
PCs show a mixed behavior. Section 2.4 reviews the proposed etymologies of the Tocharian PCs.
Finally, Section 2.5 summarizes the chapter.
2.2 The pronominal system of Tocharian A and B
This section reviews the pronominal system and the phonological, morphological, and syntac-
tic characteristics of the pronominal clitics of Tocharian A and B. The following section (§2.2.1)
outlines the pronominal system of TA and TB. TA and TB have independent personal pronouns
9
for first and second person, and demonstrative pronouns for third person. In addition, they have
pronominal clitics. Section 2.2.2 discusses the host of the Tocharian PCs. PCs in TB consistently
follow a finite verb. In TA, they mostly follow a finite verb, but they may also follow a participle,
gerundive, and a noun in a nominal predicate. Section 2.2.3 reviews the first-person singular PC -
ñi in TA, which is homophonous with the genitive-dative of the first-person singular independent
personal pronoun. Section 2.2.4 treats cases in which a PC is followed by a so-called secondary
case marker. PCs may be followed by a secondary ablative or allative case marker. TA attests pe-
culiar allomorphs (i.e., anäṣ and anac) of the allative and ablative markers. Section 2.2.5 discusses
cases in which gerundives accompany these allomorphs. Section 2.2.6 reviews the phonological
properties of the Tocharian PCs. The PCs in TB form a single phonological word with their host,
affecting the stress calculation of the host.
2.2.1 Overview of the pronominal system of Tocharian A and B
Tocharian A and B have free and bound forms of personal pronouns. I will call the former inde-
pendent personal pronouns and the latter pronominal clitics. Tables 2.1 (TA) and 2.2 (TB) list the
first- and second-person independent personal pronouns. TA distinguishes masculine and fem-
inine only in the first-person, which is typologically rare (Jasanoff 1989). There are three cases:
nominative, accusative, and genitive-dative.1 In the personal pronouns of TA, this three-way case
distinction is only discernible in the second-person singular. In the first-person singular and the
first- and second-person plural, nominative and accusative are syncretic. In contrast, the late
and colloquial TB texts attest the first- and second-person accusative plural wesäṃ, yesäṃ and
genitive-dative plural wesi, yesi (Stumpf 1990: 91–3; Pinault 2008: 535; Peyrot 2008: 120–1), giving
rise to a three-way case distinction in these person-number combinations as well.
TA 1sg.m 1sg.f 2sg 1pl 2pl

nom tu
näṣ ñuk was yas
acc cu
gen-dat ñi nāñi tñi wasäṃ yasäṃ
Table 2.1: Independent personal pronouns in Tocharian A
1. The accusative is traditionally referred to as the oblique.
10
TB 1sg 2sg 1pl 2pl
nom twe wes yes
ñäś, ñiś
acc ci wes, wesäṃ yes, yesäṃ
gen-dat ñi tañ wesäñ, wesäṃ, wesi yesäñ, yesäṃ, yesi
Table 2.2: Independent personal pronouns in Tocharian B
As for the third-person, demonstrative (anaphoric) pronouns serve as personal pronouns (Table
2.3). Tocharian demonstrative pronouns distinguish masculine and feminine. The neuter forms
are used to refer to an immediately preceding or following utterance, and do not participate in
any gender concord.
TA 3sg.m 3sg.f 3sg.n 3pl.m 3pl.f

nom säm sām täm cem tom
acc cam tām täm cesäm tosäm
gen-dat cami temi tmis cesmi tosmāśśi
TB 3sg.m 3sg.f 3sg.n 3pl.m 3pl.f

nom su sāu tu cey, cai toṃ
acc ceu , cau tāu tu ceṃ toṃ
gen-dat cwi, cpi tāy tuntse ceṃts toṃts
Table 2.3: Demonstrative (anaphoric) pronouns in Tocharian A and B
In addition to these personal and demonstrative pronouns, TA and TB have pronominal clitics
(PCs; Table 2.4). In contrast to the independent personal pronouns and the demonstrative pro-
nouns, Tocharian PCs do not have a distinction in case. They do not have a person distinction
in the plural either. For example, TB -me may be used for the first-person plural, second-person
plural, or third-person plural reference (e.g., 2.1, 2.2, and 2.3, respectively).
1sg 2sg 3sg pl

TA -ñi -ci -ṃ -m
TB -ñ -c2 -ne -me
Table 2.4: Pronominal clitics in Tocharian A and B
2. Word-final -c develops to -ś in late and colloquial texts (Schmidt 1986: 642, Peyrot 2008: 77).
11
(2.1) [TB] -me = first-person plural
rṣāki ṣerpar-me ¦ twe ke śpālmeṃ rṣāke nes :
sage.nom.pl indicate.pst.act.3pl-pl nom.2sg ptcl excellent sage cop.npst.3sg
[Nandā and Nandābala speaking to the god Brahma:] “The sages pointed out to us that
yousg are the [most] splendid sage. [1a] ”
(B107a10; verse; [5¦7]×4)
(2.2) [TB] -me = second-person plural

ṣer-śkana se nomiye-ṣṣe bhājaṃ rerinu
sister-dim.voc.pl dem.m.nom.sg jewel-adjz.m.sg bowl.sg give.up.ptcp.m.nom.sg
star-me epe mā •
cop.npst.3sg-pl or neg
[The Bodhisattva speaking to Nandā and Nandābala:] “Little sisters, is this jeweled bowl
given up by youpl or not?”
(B107b8–9; prose)
(2.3) [TB] -me = third-person plural

|| tane araṇemi walo brāhmaṇeṃ wratsai tsäṅkor-meṃ
then Araṇemi king.nom.sg brahmin.acc.pl towards rise-abs
käṣṣī-ññe yäkne-sa asān-ne lyāmate-me ||

teacher-adjz.m.sg manner.sg-perl throne.sg-loc sit.down.caus.pst.mid.3sg-pl
‘Thereupon the king Araṇemi rose towards the brahmins [and] let them sit on the throne
in the manner of teachers.’
(B81b6; trans. by CEToM; prose)
The Tocharian PCs appear in a different place from where other pronouns usually appear. As
these examples show, they appear immediately after finite verbs. I will review the hosts of the
Tocharian PCs in the following section (§2.2.2).
2.2.2 Hosts of the Tocharian PCs
Pronominal clitics consistently appear immediately after a finite verb in TB. For example, the
preterite palātai ‘(yousg ) praised X’ hosts a third-person singular PC in (2.4).
(2.4) [TB] Finite verb hosting a PC
12
mantaṃtā pa-si marsasta ¦
never uphold.moral.behavior-inf forget.pst.2sg
palātai-ne ṣu [a5] komt-sa ¦ ṣeme ślok-tsa (:)

praise.pst.mid.2sg-3sg 7 day-perl one strophe-perl
‘Yousg have never forgotten to uphold (moral behavior). Yousg have praised him (= the
Buddha) for seven days with a single strophe. [1c] ’ (B297a.a4; verse; [7¦7¦4]×4)
In TA and TB, it seems that a host may carry at most one PC: there is no example in which multi-
ple PCs accompany a single host. While TA and TB are head-final languages in which the direct
object generally precedes the finite verb, PCs always follow their host, and hence they violate
Greenberg’s (1963) Universal 25 (“If the pronominal object follows the verb, so does the nominal
object”).3 Non-finite verbs never host a PC in TB, and when PCs represent an argument or a pos-
sessor semantically associated with the argument of a non-finite verb, they “climb” to the matrix
clause to be hosted by the finite verb (Carling 2006, Adams 2015: 151). For example, the matrix
finite verb of (2.5), auntsante ‘(they) began,’ hosts a third-person singular PC that represents the
argument of the causativized infinitive makästsi ‘to make X run; chase X’. In other words, the PC in
this example does not attach to the infinitive makästsi but to the finite verb in the matrix clause.
(2.5) [TB] Climbing of a PC

auntsante-ne ścīre makäs-tsi
begin.pst.mid.3pl-3sg hard run.caus-inf
‘(They) began to chase him [= Prince Uttara] hard.’
(B88a2; trans. based on CEToM; prose)
PCs in many cases attach to the finite verb in TA. However, in some cases, participles, gerun-
dives, and nouns in nominal clauses host a first- or second-person singular PC (Schulze, Sieg, and
Siegling 1931: 166).4 For example, a preterite participle hosts a PC in (2.6) and so does a gerundive
in (2.7).
(2.6) [TA] PC hosted by a preterite participle
3. See Ashton (2011) for a syntactic analysis of PCs in TB based on the Minimalist Grammar formalism (Stabler
1997). His analysis posits that PCs move to the specifier of some projection while finite verbs (and other full nominal
expressions) move to a higher position, preceding the PC.
13
tämyo pūk kärsnā-l wram knānmune-yo
therefore all know.npst-gdv thing knowledge-ins
lyalyku-ci :
be.illuminated.caus.ptcp.m.nom.sg-2sg
‘... therefore, everything (one) should know (is) illuminated by yousg [= the Buddha]
with (yoursg ) knowledge.’
(A249b1; verse)
The second-person singular PC -ci in example (2.7) represents the subject of the gerundive.
(2.7) [TA] PC hosted by a gerundive

[a1] /// (māski kälkā-)lyāṃ tkan-ac yä-l-ci kᵤyalte ||
difficult go.subj-gdv.f.acc.sg earth-all go.npst-gdv-2sg because
meneklin-aṃ || cämpamo träṅktsi yme
M-loc be.able speak-inf way
‘Yousg [= Bodhisattva] should go to a (hard-)to-reach place, because [In the M-tune:] (You
are) able to speak ... the way …’
(A56a1; trans. by CEToM; prose)
In a nominal clause where a finite copula is missing, nouns may host a PC as in (2.8). In this
example, tiri- ‘way, manner’ hosts the second-person singular PC -ci (Schulze, Sieg, and Siegling
1931: 166–7).5
4. A preterite participle hosts a PC in the following examples: raryu-ci (A56b1), laltuṣ-ci (A125b2 [2.45]), n(ā)ṃ(tsu)-
ñi (A147a4), lyalyku-ci (A249b1 [2.6]), śaśälpu-ñi (A258b5), kärṣto(nt)-ñ(i) (A269 and 290b1), and kaklyuṣu-ñi (YQ II.1b2).
As for worpūs-[s]kam-ci (A248a1), see (2.9). An m-participle hosts a PC in the following example: rsunāmāṃ-ñi (A67b1).
A gerundive hosts a PC in the following examples: yäl-ci (A56a1 [2.7], A56a2, A255b5), mäskal-ci (A115b3), yal-ñi (A71a1,
A189b6), yal-ci (YQ II.4a4), pärskāl-anäṣ (A456b3), präskāll-anäṣ (A155b3 [2.17]; A179a1), späntāll-anac (A61a4; A169a2
[2.41]), and (lkā)laṃ-anäṣ (YQ III.2b8). A noun hosts a PC in the following examples: tiri-ci ‘manner-2sg’ (A106a6 [2.8]),
puklā-ci ‘year.pl-2sg’ (A255b3), pratsak-ci ‘breast-2sg’ (A378a5), kāruṃ-c(i) ‘compassion-2sg’ (A260a4), and wärc(e)-ci
‘fault-2sg’ (A326a3). In A168 b5 (2.b), although its context is limited, an adjective (kā)lyāṇī- ‘good, excellent’ (cf. Skt.
kalyāṇaka- ‘[morally] good, virtuous’) seems to host a second-person singular PC -ci.
(2.b) [TA] PC hosted by an adjective?
/// (klo)pa-nt wärpnānträ kᵤyalte kos-ne wram kā /// [b5] /// (kā)lyāṇī-ci puk
suffering-pl receive.npst.mid.3pl because how.much-comp thing virtuous-2sg all
napeṃ ///
human
‘… (they) receive the sufferings, because how much … thing … yoursg … (is/are) virtuous … all human … ’
(A168b5; prose?)
5. See Hackstein (2012) on the prehistory of these copula-less clauses in Tocharian and other Indo-European
14
(2.8) [TA] PC hosted by a noun
[a6] /// ·-yo hetutwati träṅkäṣ mā tiri-ci nāśi ––––
-ins Hetutwati speak.npst.act.3sg neg manner-2sg mistress
‘… with … Hetutwati speaks: “(It is) not yoursg way, o (my) lady, …”’
(A106a6; prose?)
There is even an example in which an adverb appears to intervene between a hosting preterite
participle and a PC (2.9). In this example, the preterite participle worpūs ‘surrounded’ [ptcp.f.nom.sg]
and the second-person singular PC -ci appear to be separated by the adverb skam ‘always’ (Schulze,
Sieg, and Siegling 1931: 166).
(2.9) [TA] skam ‘always’ intervenes between a preterite participle and a PC?
lukśanuntsāṃ kaṣ-swāñcen-yo ¦ worpūs (s)kam ci kapśañi :
bright.f.acc.sg fathom-beam-ins surround.ptcp.f.nom.sg always gen.2sg body.sg
Yoursg body [is] always surrounded by a bright ray which is a fathom wide. [20a] ’
(A248a1; verse; [5¦5¦8¦7]×4)
However, among the 551 attestations that I have examined for TA, this is the only example in
which a host-PC connection appears to be interrupted by an adverb. Therefore, one might won-
der whether there is any alternative explanation available. This passage is a translation of Mātr̥ceṭa’s
Varṇārhavarṇastotra II.39 (cf. Pinault 2008: 286).
(2.10) Varṇārhavarṇastotra II.39 (Hartmann 1987: 112–3)

(asecanakarūpāya pra)bhābhāsuramūrtaye /
na(mo ’stu) sarvadr̥śyānā(m) darśanī(ya)tamāya (t)e //
‘To him, who has a form that cannot be satiated, whose body shines with light, may
there be veneration to you, the most beautiful of all visible objects!’
TA lukśanuntsāṃ kaṣswāñcenyo worpūs (s)kam ci kapśañi ‘your body (is) always surrounded by a
fathom-wide bright ray’ translates Sanskrit prabhā-bhāsura-mūrti- ‘whose body is shining with
light,’ suggesting that ci in this example serves as the possessor of kapśañi ‘body.’ Since an in-
dependent personal pronoun that represents a possessor generally precedes its possessum (e.g.,
tñi krant arämpāt ‘your beautiful shape’ in 2.9), it seems possible to take ci in this example not
languages.
15
as a PC, but as a (grammatical) error for tñi, the genitive-dative of the second-person singular
personal pronoun (table 2.1). It is probably motivated by the first-person singular personal pro-
noun ñi [gen.1sg], which is homophonous with the first-person singular PC (Section 2.2.3), and
also immediately precedes its possessum when it serves as a possessor. In addition, it is worth
pointing out that the TA translation of the Varṇārhavarṇastotra shows some peculiar syntax. For
example, we find two out of three rare examples of clitic right dislocation in A249 (see Chapter 7
for a discussion on this topic). Furthermore, the verse number “13” is misplaced in A245 a3: it is
found after pāda 14b, not after 13d (Pinault 2008: 282–3). These facts also seem to suggest that ci
in this example could be attributed to some kind of error.
2.2.3 The first-person singular PC -ñi
A word is in order regarding the TA first-person singular PC -ñi, which is homophonous with
the genitive-dative of the first-person singular personal pronoun (table 2.1).6 As reviewed in the
previous subsection, when the genitive-dative of an independent personal pronoun functions as
a possessor, it generally precedes the noun that it possesses (e.g., 2.11).
(2.11) [TA] ñi = Independent personal pronoun preceding a possessum

klopasu nāṃtsu yaṃtrācāre träṅkäṣ
depressed.m.nom.sg cop.ptcp.m.nom.sg mechanic.nom.sg speak.npst.act.3sg
pälkāc nācki ñi klop caṣ

look.imp.mid.2pl lord.nom.pl m.gen.1sg suffering dem.m.acc.sg
“Depressed, the mechanic says: behold, youpl gentlemen, this misfortune of mine!”
(A8b5; trans. based on CEToM; prose)
Therefore, the grammatical property of ñi is ambiguous when it immediately follows a finite verb,
participle, or gerundive and simultaneously precedes the noun it possesses. For example, the
first-person singular PC ñi in (2.12) may be taken as a PC (translation 1) or as an independent
6. Poucha (1955: 110–1) lists 57 examples that contain a PC -ñi, and 130 examples of the first-person singular
personal pronoun ñi ‘my, to me’. However, his tsäkse-ñi in 101b2 does not exist, “126 b5 mā lipñät ñi” is a typo for “126
b2 mā lipñät ñi”, and “knaṣtär ñi 72a2” for “knaṣtär ñi 71a2”. ñi in 65 b4, 67 a4, and 92 a5 is not an independent personal
pronoun but a PC (65 b4: (pākṣi)ññā-ñi kucne tu wsār pälkoräṣ weñāṣt ‘Explain to me what you said concerning the
wheat!’; 67 a4: bodhisattu träṅkäṣ mā ontaṃ tāka-ñi kᵤyalte ñareṣinäśśi klopant opyāc källā[a5](mām) ‘The bodhisattva
says: “By no means has there been (satisfaction) for me, because (I) remember the sufferings of those (who are) in
the hells ….’; 92 a5: /// cam śkaṃ lo psumār-ñi kᵤyal lykäly lykäly tuṣt-ñi : ‘… and take this … away from me! Why do
you burn me finer (and) finer?’).
16
personal pronoun (translation 2).
(2.12) [TA] ñi = Independent personal pronoun or a PC?

pekäntā-śśi käṣṣi-śśi ¦ käṣṣi pekant ñi naṣu :
painter-gen.pl master-gen.pl master painter m.gen.1sg friend
lokit yeṣ ñi yantär-ṣi ¦ śomiṃ cam-i spa(ktāṃ) [a1]

guest go.impf.act.3sg gen.1sg mechanism-adjz girl dem.m.sg-gen service
ypā :
do.impf.act.3sg
[The mechanic is speaking to his neighbors:7 ]
(1) “The painter, master of the master painters, my friend, [1a] came to me as a guest,
(and) a mechanical girl attended to him (lit. ‘did a service for him’). [1b] ”
or
(2) “The painter, master of the master painters, my friend, [1a] came as a guest, (and) my
mechanical girl attended to him (lit. ‘did a service for him’). [1b] ”
(A8b6; trans. based on CEToM; verse [7¦7]×4)
In this example, the function of ñi, referring to yaṃtrācāre ‘mechanic,’ is ambiguous. While it
is possible to take it as a PC hosted by yeṣ ‘(he) went,’ representing the goal of the motion verb
(interpretation 1), we may also take it as an independent personal pronoun, representing the
possessor of yantärṣi śomiṃ ‘a mechanical girl’ (interpretation 2). Therefore, for each attestation
of the first-person singular PC in TA, I checked for this type of ambiguity and excluded such
passages.
2.2.4 Secondary case markers and Tocharian PCs
Tocharian A and B have a set of so-called secondary case markers (table 2.5). They are phrasal
clitics/affixes attached to a nominal phrase ending in the accusative. Interestingly, ablative and
allative case markers may follow a PC in TA and TB.8
7. In this passage, the addressees, the mechanic’s neighbors, look at a mechanical girl that has fallen apart into
pieces (cf. A7a5–6: (śo[a6]miṃ) mā śkaṃ tāk ‘[she] was no longer a girl’).
8. The following forms are attested: 1sg-all/abl: kmeñ-ñä-ścä (Or8212.163b5); 2sg-all/abl: ñäskau-c-meṃ
(B100a1), yaskaskemar-c-me(ṃ) (THT1112a5), and preksau-ś-meṃ (IOLToch258b1); 3sg-all/abl: (wñā)-ne-ś (B22a1),
wñā-ne-ś (B22b8; PKAS6Db3), wñā-ne-ś«†ä» (PKAS6Aa5), wñā-n(e-ś)«†ä» (PKAS6Aa5), weña-ne-ś (PKAS6Aa6; b6),
weñāre-ne-ś (B107a5), śem-ne-ś (B49a7; B63a1; B417b2), śem-ne-(ś) (THT1573ab6), weṣän-ne-ścä (B85a2), weṣṣan-n(e)-
17
TA TB Major function
Instrumental -yo — instrument
Perlative -ā -sa instrument, cause
Comitative -aśśäl -mpa accompaniment
Allative -ac -ś(c) goal, addressee
Ablative -äṣ -meṃ source
Locative -aṃ -ne location, goal
Causal — -ñ cause
Table 2.5: Secondary case markers in Tocharian A and B
The following two examples show the secondary case markers (allative in [2.13] and ablative in
[2.14]) following a PC in TB.
(2.13) [TB] Allative marker -ś following a PC

ñakti arjuṃ ¦ stām nemar-ne-ś ¦ cau eṅksate :
god.nom.pl Arjuna tree bend.pst.act.3pl-3sg-all dem.m.acc.sg seize.pst.mid.3sg
‘The gods bent the Arjuna-tree toward him (= Bodhisattva). (He) took it.’
(B107b4; verse; [4¦4]×4)
(2.14) [TB] Ablative marker -meṃ following a PC

upanande ceᵤ kampāl [b2] yaṣṣāte-ne-meṃ mā
Upananda dem.m.acc.sg cloak beg.pst.mid.3sg-3sg-abl neg
wsā-ne •
give.pst.act.3sg-3sg
‘Upanandaj asked for this cloak from himk (= an Ājīvika ascetic), [but hek ] did not give
[it to] himj .’
(PKAS18Ab2; prose)
Likewise, we find an allative case marker following a PC in TA (2.15). TA lacks an example of an

ablative marker following a PC, which is probably due to chance. The allative marker shows a
peculiar allomorph when it follows a PC: instead of the expected allative marker -ac (table 2.5),
TA attests anac (e.g., 2.15).9
śca (B93a1), weskeṃ-ne-ś (B107a10), träñcā-ne-ś (THT1507b5), nemar-ne-ś (B107b4), kālat-ne-śco (B127a3), śänman-
ne-ś (PKAS6Cb1), yaṣṣāte-ne-meṃ (PKAS18Ab2), yaskemtär-ne-meṃ (PKNS32a6), and tsäṅkā-ne-ścä (IOLToch33b3);
pl-all/abl: west-me-śca (B273b3), weṣäṃn-me-śc (B81a1), w(e)nt-me-ścä (IOLToch285a3), weñā-me-ś (B107a5, a8, a9,
b1, b8; B108a2, b9; B375b2), weñār-me-ś (B107b3; B108a3), sälkāte-me-ś (B108b4), and …-me-ś (THT1576fa1).
9. The following forms are attested: anac [all]: träṅkṣ-änn anac (A144b4), träṅkṣ-äṃn anac (A184b4; A200b3), weñā-
nn anac (A95b4; A113a4), weñā-nn «an»ac (A313a5), (we)ñā-ṃ anac (A431b5), (weñā-n) anac (A432b3), (we)ñār-äṃn atac
18
(2.15) [TA] Allative marker anac following a PC
(bārāṇas riy-ac kā)[a3]tse rṣivadaṃ wärt-aṃ cesäm sä(ksäk ārāntās
Vārāṇasī city-all near Rṣivada forest-loc dem.m.acc.pl sixty arhat.acc.pl
kākku-r-äṣ we)ñā-mm anac || ṣera(śi-niṣkramānt-aṃ ||) ///
call.ptcp-nmlz-abs speak.pst.act.3sg-pl all Ṣ-loc
‘Having called these sixty arhatsj in the Ṛṣivadana-wood near the city of Vārāṇasī, (he)
said to themj [in the S-tune:] …’
(A269+A290a3; prose)
This anac allomorph may also appear immediately after a gerundive (e.g., 2.16).
(2.16) [TA] Allative marker anac following a gerundive

/// prasky aräṣ ¦ lw-āśśi okāk pācarr oki ¦ späntā-ll
fear evoke.npst.act.3sg animal-gen.pl including father as trust.subj-gdv
anac mäskatär :
all be.npst.mid.3sg
‘... even among the animals (hej ) does (not) evoke fear; (one) could trust himj as if (he
were) a father. [1c] ’
(A61a4; verse; [5¦5¦8¦7]×4)
Furthermore, TA attests an ablative marker that appears immediately after a gerundive (2.17). It
shows anäṣ, rather than -äṣ (table 2.5).
(2.17) [TA] Ablative marker anäṣ following a gerundive

[b3] /// ntsi wsomiṃ ārṣall oki anahāl wäss oki
poisonous.f.nom.sg snake.nom.sg like halāhala poison like
präskā-ll anäṣ nāṃtsu ///
be.feared.subj-gdv.nom.sg abl cop.ptcp.m.nom.sg
‘(One) could be afraid of (it? ) like a poisonous snake (or) like a halāhala poison.’
(A155b3; prose?)
Peyrot (2017) discusses the origin of these allomorphs and proposes that *anac arose via misseg-
mentation of a verbal complex consisting of PT *-a, which was a part of the inflectional ending
(for weñār-äṃn anac A95a5), yeṣ-äṃn anac (A222b6), sämse(ñc-änn a)nac (for nämseñc-änn anac A13b5-6), ...-nn anac
(A177b1), (we)ñā-mm anac (A269and290a3 [2.15]), weññā-mm anac (A436b6 [2.18]), späntāll anac (A61a4; A169a2 [2.41]),
and ///ṣ(-)ñy oky anac (A108a3 [2.20]). anäṣ [abl]: präskāll anäṣ (A155b3 [2.17]; A179a1), pärskāl anäṣ (A456b3), and
(lkā)laṃ anäṣ (YQ 1.22[III.2]b8).
19
(e.g., act.3pl *-āræ > TB -are, TA -ār), the third-person singular PC (PT *-næ >) Pre-TA *-na, and
the allative marker *-c (i.e., Pre-TA *...a-na-c > TA ...-anac). According to him, “[w]hen anac was
no longer understood as already containing the 3sg. -äṃ, the final step in the development was
the addition of anac after -äṃ” (Peyrot 2017: 638, citing Pedersen 1941). His account neatly ex-
plains why this allomorph is absent in TB: apocope of -a did not take place in TB (e.g., act.3pl
*-āræ > TB -are, TA -ār). However, it is not clear to me why gerundives such as späntāll anac did
not reintroduce the third-person singular PC -äṃ before anac (†späntāl-äṃn-anac).
Example (2.17) attests präskā-ll anäṣ ‘to be feared’ [subj-gdv abl] instead of †präskā-l-ä(ṃ)n anäṣ
[subj-gdv-3sg abl] with a PC. In fact, there is not a single example of a gerundive carrying both
a PC and anac/anäṣ. Therefore, TEB II: 78 separate anac and anäṣ as an-ac and an-äṣ, respectively,
and list -an- as “Pron. suff. der 3. Pers., nur in sekundären Kasus”. Likewise, Carling, Pinault,
and Winter (2009: 8) list -an- as a “pronominal element 3rd person singular (only with secondary
cases)”. See also Poucha (1955: 4): “anac … dat. sg. anäṣ … abl. sg.”. However, it is not clear to me
why -mm anac in (2.15), which clearly contains the plural PC -(ä)m, also contains a third-person
“singular” pronominal element.
At any rate, Thomas and Krause’s description captures the fact that there is no example in which
anac or anäṣ follows a first- or second-person singular PC. When these forms follow a PC, the pre-
ceding PC is either third-person singular (-(ä)ṃ) or plural (-(ä)m). If their morphological segmen-
tation is correct, we should expect the person-indifferent plural PC -(ä)m to show third-person
reference consistently when it is followed by anac or anäṣ. There are two occurrences of anac or
anäṣ following a plural PC. Although the person-reference of weññā-mm anac in A436b6 is at best
uncertain (2.18), (we)ñā-mm anac (A269+A290a3) undoubtedly has third-person reference (2.15,
repeated here as 2.19).
(2.18) [TA] -mm- = 1pl, 2pl, or 3pl?

[b6] /// m weññā-mm-anac lāñci tiri kä ///
speak.pst.act.3sg-pl-all royal manner
‘… (s/he) spoke to { us? /youpl ? /them? }: “… royal manner …”’
(A436b6; prose?)
20
(2.19) (= 2.15) [TA] -mm- = 3pl
(bārāṇas riy-ac kā)[a3]tse rṣivadaṃ wärt-aṃ cesäm sä(ksäk ārāntās
Vārāṇasī city-all near Rṣivada forest-loc dem.m.acc.pl sixty arhat.acc.pl
kākku-r-äṣ we)ñā-mm-anac || ṣera(śi-niṣkramānt-aṃ ||) ///
call.ptcp-nmlz-abs speak.pst.act.3sg-pl-all Ṣ-loc
‘Having called these sixty arhatsj in the Ṛṣivadana-wood near the city of Vārāṇasī, (he)
said to themj [in the S-tune:] …’
(A269+A290a3; prose)
Therefore, it appears that anac [all] and anäṣ [abl] were synchronically segmentable as an-ac
[3-all] and an-äṣ [3-abl], respectively. In other words, finite verbs with a PC and a secondary
case marker appear to mark person twice: by a PC and -an- (e.g., [A184b4 and A200b3] träṅkṣ-äṃn
an-ac ‘(he) speaks to him/her’ [speak.npst.act.3sg-3sg 3-all]).
However, there is one puzzling example (2.20), in which anac immediately follows the particle oky
‘like, as if, as it were’ (cf. Skt. iva). For the sequence before this particle, Sieg and Siegling (1921:
61) read ṣñy ‘(one’s) own’. However, this reading would amount to claim that neither a PC nor
a gerundive cooccur with anac in this example. Schulze, Sieg, and Siegling (1931: 305), instead,
read ///ṣ-ñy, that is, the third-person singular ending -ṣ, followed by the first-person singular PC
-ñi.
(2.20) [TA] Allative marker anac following a particle oki ‘like, as’
[a3] ///ṣ-ñy oky anac : śra-l-une-yo papälykus
-1sg? like all be.separated.npst-gdv-nmlz-ins torment.ptcp.f.nom.sg
sundari pra
Sundari
‘… (s/he) …s as if … to me. Having been tormented by separation, Sundari …’
(A108a3; verse?)
Although uncertainty remains, this example potentially shows that anac, which is separable from
a PC, is not limited to third-person reference. Therefore, I take anac and anäṣ to be markers that
represent all and abl, rather than 3-all and 3-abl.
21
2.2.5 Gerundives followed by anäṣ/anac
The allative and ablative allomorphs anac and anäṣ may follow a gerundive. It seems that they
are synchronically monomorphemic in this case also.
Tocharian A and B have a gerundive built on the present or the subjunctive stem. The former
indicates a deontic interpretation, while the latter an epistemic interpretation (Thomas 1952).
For example, the root √i-|kälkā- ‘to go’ forms a present stem yä- and a subjunctive stem kälkā-, and
the gerundive yä-l built to the present stem means ‘(one) should go,’ while kälkā-l, built on the
subjunctive stem means ‘(one) could go.’ Gerundives reduce valency, suppressing a nominative
subject or demoting it into a genitive-dative oblique. In (2.22), for example, the agent of the
gerundive is suppressed. In (2.7), repeated here as (2.23), the demoted agent is represented by
the second-person singular PC -ci.
(2.21) Gerundives built on a transitive/intransitive stem
i. Transitive stem: e.g., TA kropna- ‘X collects Y’

→ kropna-l ‘Y should be collected; (one) should collect Y’ (e.g., 2.22)
ii. Intransitive stem: e.g., TA yä- ‘X goes (to Y)’

→ yä-l ‘(one) should go (to Y)’ (e.g., 2.23)
(2.22) [TA] Gerundive built on a transitive stem

/// eṣä-l el wras-s-aṃ ortune ya-l
give.npst-gdv.m.nom.sg gift people-pl-loc friend do.npst-gdv.m.nom.sg
kropna-l pñi-ntu ///
collect-gdv.m.nom.sg virtue-pl
‘Gifts have to be given, friendship has to be made with the beings, [and] virtues have to
be collected.’
(A280a6; prose)
22
(2.23) [TA] (= 2.7) Gerundive built on an intransitive stem
meneklin-aṃ ||
M-loc
‘Yousg [= Bodhisattva] should go to a (hard-)to-reach place, because [In the M-tune:] …
Gerundives may be used attributively and predicatively as shown in (2.24).
(2.24) (= 2.23) [TA] Predicative and attributive use of a gerundive

meneklin-aṃ ||
M-loc
‘Yousg [= Bodhisattva] should go to a (hard-)to-reach place, because [In the M-tune:] …
In this example, the attributively used kälkālyāṃ shows gender, number, and case concord with
the feminine accusative singular tkan ‘earth,’ to which an allative marker attaches.10 In contrast,
yä-l is used predicatively, and its subject is demoted and represented by a PC.
There are 6 attestations of a gerundive followed by anäṣ/anac. One of them is severely damaged
(2.25).
(2.25) [TA] Gerundive followed by anäṣ/anac (1/6)

/// l· kār pärskā-l anäṣ ///
be.afraid.subj-gdv abl
‘… (one) could be afraid of …’
(A456.bb3; prose?)
In the following two examples, gerundives agree with a (suppressed) experiencer, that is, the
subject of mäskatär and nāṃtsu, respectively.
10. Secondary case markers usually attach to a phrase-final accusative with which attributive adjectives concord
in the accusative (the so-called Gruppenflexion; TEB I: 91-2).
23
(2.26) (= 2.16) [TA] Gerundive followed by anäṣ/anac (2/6)
/// prasky aräṣ ¦ lw-āśśi okāk pācarr oki ¦ späntā-ll
fear evoke.npst.act.3sg animal-gen.pl including father as trust.subj-gdv
anac mäskatär :
all be.npst.mid.3sg
‘... even among the animals hej does (not) evoke fear; (one) could trust himj as if (hej
were) a father.’
(A61a4; verse)

///ntsi wsomiṃ ārṣall oki anahāl wäss oki präskā-ll
poisonous.f.nom.sg snake as halāhala-poison poison as be.afraid.subj-gdv
anäṣ nāṃtsu ///
abl cop.ptcp.m.nom.sg
‘(Hej ) was afraid of himk as if (hek were) a poisonous snake (or) like a halāhala poison, …’
(A155b3; prose?)11
In these examples, the suppressed argument is the subject of the copula. TA and TB may have a
suppressed argument as the subject of a copula as example (2.28) shows.
(2.28) Gerundive agreeing with a suppressed argument

mā tañ kc= āyor ai-lle nesau ¦
neg gen.2sg indf gift give.subj-gdv cop.npst.1sg
‘I cannot give any gift to you.’ (B23b5; verse; [5¦8]×4+[8¦8¦5])
This example contains the gerundive aille, built on the subjunctive stem ai- ‘Xagent:nom gives
Ytheme:acc to Zrecipient:gen .’ This gerundive alone would mean ‘Y could be given to Z; (one) could
give Y to Z,’ but it cooccurs with a copula that agrees with the suppressed agent and means ‘(X)
could give Y to Z.’ In view of this example, we may understand the argument structures of späntāll
in (2.26) and präskāll in (2.27) as follows.
11. Cf. Thomas (1952: 30): “Vor ihm furchterfüllt wie vor einer giftigen Schlange [oder] Halāhala-Gift.” In contrast,
Carling, Pinault, and Winter (2009: 11), followed by CEToM, seem to take anäṣ as representing the experiencer of
präskāll: “/// ... to be feared by him/her like a poisonous snake [or] the halāhala poison”
24
(2.29) Argument structure of späntāl and präskāl (to be revised)
Xexperiencer:nom Ytheme:ins/all 12 √spänt(ā)- ‘X relies on Y’
i.
→ Ytheme:ins/all späntā-l ‘(one) could rely on Y’
ii. (2.26 =) späntā-ll anac mäskatär ‘(one) could rely on him’
Xexperiencer:nom Ystimulus:abl 13 √pärsk(ā)- ‘X is afraid of Y’

iii.
→ Ystimulus:abl präskā-l ‘(one) could be afraid of Y’
iv. (2.27 =) präskā-ll anäṣ nāṃtsuj ‘(hej ) could be afraid of him’
In these cases, we could understand a referential null object (pro) as the IO of the gerundives and
analyze anac/anäṣ as simply an allative/ablative marker, without any person reference. This is in
line with the analysis I developed in Section 2.2.4, where I considered anac/anäṣ to be monomor-
phemic, rather than bimorphemic.
(2.30) Argument structures of späntāl and präskāl (revised)
i. (2.26 =) pro3sg k späntā-ll anac mäskatär ‘(one) could rely on himk ’
ii. (2.27 =) pro3sg k präskā-ll anäṣ nāṃtsuj ‘(hej ) could be afraid of himk ’
In the following example (2.31), akmal ‘face’ seems to be the subject of lotäk ‘turned, became.’

/// p(e)nu akmal präskā-ll anäṣ (krā)ṣiññäl lotäk
also face be.afraid.subj-gdv abl be.angry.subj-gdv become.pst.act.3sg
‘… also … (his/her) face became (that which one) could be afraid of, (and) (that which
one) could be annoyed (with) …’
(A179a1; prose?)
The TA root √lotkā- ‘turn, become’ connects a nominal expression with an adjective, meaning ‘X
became Yadj ’ (see, e.g., 2.32).
12. Cf. Y is com or perl in TB.
13. Cf. Y is gen or perl in TB.
25
(2.32) tām praṣṭ-aṃ cam suryodgam prātīhāri-yo wrasañ pu-k
dem.f.acc.sg time-loc dem.m.acc.sg sunrise miracle-ins people.nom.pl all-emp
wlyepe sākre l(ot)k(a)r ||
gentle.m.nom.pl happy.m.nom.pl become.pst.act.3pl
‘At that time, all beings became gentle (and) happy through the marvel of the sunrise.’
(A313a3; prose)
In (2.31), there are two complement adjectives : präskāll anäṣ and (krā)ṣiññäl. Therefore, we could
translate (2.31) as: ‘(His/Her) face became präskāll anäṣ (and) (krā)ṣiññäl.’
Malzahn (2010: 613) lists TA krāṣiññäl as a gerundive built on the class XII subjunctive stem of
a transitive root (√krās- ‘annoy’). However, this is the only verbal form attested for this root,
and there is no reason that this root must be transitive. The corresponding TB root √krās- ‘[act]
annoy, vex; [mid] be angry, feel irritated’ is both transitive and intransitive (2.33).
(2.33) [TB] √krās- ‘[act] annoy, vex; [mid] be angry, feel irritated’
i. [act] = transitive
(ajātaśatru) ¦ retke ṣālla ¦ kausal-ṣets (:)
Ajātaśatru army destroy.pst.act.3sg Kosala-adjz.gen.pl
räskre krāsa tu ¦ pra(saṃnakeṃ) ¦ ///

bitterly annoy.pst.act.3sg dem.n Prasannaka
‘Ajātaśatru destroyed the army of the Kosalans. [55b] It tormented Prasannaka very
much. … [55c] ’
(B21a7; verse [5¦4¦3]×4)
ii. [mid] = intransitive

tumeṃ tanapāte kraṣiyate
then patron be/get.angry.pst.mid.3sg
Then the patron got angry.
(IOLToch248b4; prose)
In addition, there are very few present III/IV forms that are transitive (Malzahn 2010: 372–4).
The present IV krosotär in (2.34) is also intransitive, as expected.
26
(2.34) [TB] krosotär = intransitive (with an accusative stimulus)
[a1] – – – – – r«†ä» (:) k(a)marttikets wakītse ¦ mā ynāñmäññe śuketse
ruler.gen.pl distinguished.nom.sg neg honor sweet.nom.sg
¦ su krosotär (1)
dem.m.nom.sg be.annoyed.npst.mid.3sg
‘… The most excellent of the rulers, this one, the sweet [one], is not angry about honor.
[1d] ’
(PKNS29a1; verse [7¦7¦4]×4)
In this example, ynāñmäññe ‘honor, (act of) veneration’ is not the theme (i.e., *‘this one does
not annoy ynāñmäññe’) but the (accusative) stimulus of √krās- (‘this one is not annoyed about
ynāñmäññe’).
In view of these examples, it is tempting to take TA √krās- as both transitive (‘Xstimulus:nom an-
noys Yexperiencer:acc ’) and intransitive (‘Xexperiencer:nom is angry [about Ystimulus:acc/perl ]’). If a
gerundive is built on the former, it will mean krāṣiññäl ‘Yexperiencer could be annoyed; (one) could
annoy Yexperiencer ’, while a gerundive built on the latter would mean krāṣiññäl ‘(one) could be an-
gry (about Ystimulus:acc/perl ).’14
Since akmal is likely to be the stimulus of krāṣiññäl in (2.31), I analyze that krāṣiññäl in (2.31) is not
built on a transitive but on an intransitive stem, meaning ‘(one) could be angry (about Ystimulus );
(Ystimulus ) whom/which (one) could be angry about; anger-inducing.’
Likewise, TA √pärsk(ā)- ‘be afraid’ forms an intransitive stem, taking a stimulus in the ablative.
(2.35) TA √pärsk(ā)- ‘be afraid’ taking an ablative stimulus
14. TA krāṣiññäl may take a stimulus in the perlative as in (2.c).

(2.c) krāṣiññäl taking a stimulus in the perlative
/// (mā penu tu sewā)s-aśśäl wärt-ac kälkā-l-une-yā krāṣiññä-l
neg also nom.2sg son.pl-com forest-all go.subj-nmlz-perl vex.subj-gdv
‘Also, you will not be vexed about going into the forest together with the children.’
(A70b1; prose?)
27
[b6] (ṣñi w)l(al)uney-äṣṣ oki ālyakäṃ koluney-äṣ praskanträ •
own death-abl like other.m.acc.sg killing-abl be.afraid.npst.mid.3pl
‘(They) are afraid of killing another as much as they are of (their own death).’
(A262b6; Schmidt 1974: 156; prose?)
Therefore, the gerundive präskāl built on an intransitive stem would mean ‘(one) could be afraid
(of Ystimulus ); Ystimulus whom/which (one) could be afraid of; fear-evoking.’
In example (2.31), repeated here as (2.36), akmal ‘face’ is the stimulus of √präskā- ‘be afraid.’ How-
ever, akmal ‘face,’ is not marked by the ablative but by the nominative(-accusative) since it is
the subject of lotäk. Instead, the ablative marker appears on the gerundive that functions as the
complement adjective of the stimulus, connected by lotäk.

/// p(e)nu akmal präskā-ll anäṣ (krā)ṣiññä-l lotäk pākṣiṃ
also face be.afraid.subj-gdv abl be.angry.subj-gdv become.pst.act.3sg
–
‘ … also (his/her) face became (1) that which one could be afraid of (and) (2) that which
one could be angry about.’
(A179a1; prose?)

/// (lkā)-laṃ anäṣ • ṣäpnont wäṣta-käntwāñ mäskanträ • ymār
see?.subj-gdv.f.pl abl creepy double-tongued.nom.pl be.npst.mid.3pl quickly
‘… (theyf:pl ) could be seen (as terrible) for him. (They) are creepy, having two-tongues.
Quickly …’
(YQ1.22[III.2]b8; prose?)
This interpretation sheds light on the problematic form ///laṃ anäṣ (YQ 1.22[III.2]b8). The restora-
tion (lkā)laṃ anäṣ by Ji, Winter, and Pinault (1998: 152) is not compelling, since, as rightly pointed
out by Peyrot (2017: 636 n. 11), it does not explain the function of the ablative anäṣ.
28
“The feminine plural subject must be venomous snakes, but the function of the ab-
lative object anäṣ is not clear to me. A restoration to (präskā)laṃ anäṣ is also difficult,
because the snakes cause fear, but are not in fear [sic] themselves.” (Peyrot 2017: 636
n. 11)
the root √läk(ā)- ‘to see’ does not take an experiencer or a theme in the ablative. In A358a3 (2.38),
√läk(ā)- ‘to see’ takes a theme in the allative.
(2.38) [TA] √läk(ā)- ‘to see’ taking a theme in the allative

pā«†ṃ»cyā(s kä)lytär-ci ¦ vajrapā·i – ¦ (śā)lyās pracar ānant ṣāmaṃ ¦
right stand.npst.mid.3sg-2sg Vajrapāṇi left brother Ānanda monk
wsokone-yo lkeñ=-cy akml-ac :
joy-ins see.act.npst.3pl-2sg face-all
‘Vajrapāṇi is standing to your right. (Your) brother Ānanda, the monk, is (standing to
your) left. They are looking at your face with joy.’
(A358a3; verse; [5¦5¦8¦7]×4(?) )
Taking the nominative-accusative akmal to be the stimulus of präskāll anäṣ offers an alternative
interpretation of (2.37). That is, the allative marker anäṣ is semantically associated with a (nom-
inative) stimulus, although it appears immediately after a gerundive.
Peyrot’s restoration (präskā)-laṃ anäṣ ‘be.afraid.subj-gdv.f.pl abl’ is therefore more likely since
the stimulus with this root is usually in the ablative in TA (e.g., 2.35). Also, it nicely fits the Old
Uyghur translation körgäli qorqïnčïγ bolur ‘they are horrible to look at’ (Geng et al. 1988: 178–9;
Peyrot 2017: 636 n. 11). The idea is that snakes are fear-evoking.
Therefore, we may restore/understand ārṣalāñ ‘snakes’ [f.nom.pl] as the stimulus, and the gerun-
dive (präskā)laṃ [f.nom/acc.pl] predicatively modifies it and shows agreement with it.
(2.39) YQIII.2b8: … (†ārṣalāñstimulus:f.nom.pl ) … (präskā)laṃf.nom/acc.pl anäṣ

‘ … (snakes) … (they are) those which one could be afraid of.’
(2.40) Cf. A179a1: … akmalstimulus:m.nom.sg präskāllm.nom.sg anäṣ … lotäk

‘(his/her) face became that which one could be afraid of …’
29
Finally, in (2.41), späntāll anac attributively modifies the theme wlyepa-rake ‘a sweet word.’ But
in this case, the theme wlyepa-rake ‘sweet word’ is in the nominative-accusative, and the allative
marker appears immediately after the gerundive.

– späntā-ll anac wlyepa-ra(k)e – – – – – – – – – cami cmol ko ///
trust.subj-gdv all soft-word dem.gen.sg birth
‘… a sweet word, (which one) could (not? ) trust … his birth …’
(A169a2; prose?)
To summarize, we saw that the TA allative and ablative allomorphs anac/anäṣ do not have any
person reference synchronically, no matter whether they follow a PC (e.g., 2.15) or a gerundive
(e.g., 2.16 and 2.17). They are not limited to the third-person singular as -mm anac in (2.15) and
-ṣ-ny oky anac in (2.20) suggest. We may explain the fact that anac/anäṣ following a gerundive
appears to show third-person reference by assuming a referential null object (2.16 and 2.17).
2.2.6 Phonological characteristics of the Tocharian PCs
Pronominal clitics form a single prosodic unit (phonological word) with their host in TB. Classi-
cal TB writes stressed /ə/ and unstressed /a/ with 〈a〉 and stressed /a/ with 〈ā〉, which enables
us to infer that a PC and its host constitute a single prosodic unit when phonology calculates
the accentuation of the host-PC complex (Krause 1952: 203). The basic stress pattern of TB is
as follows. The primary stress of a word mostly falls on the second syllable unless the word is
mono- or disyllabic, in such cases the primary stress falls on the first syllable (cf. Hackstein 2017:
1306–7). The accent in TB is morphophonemic since some morphosyntactic categories such as
class V subjunctives (Jasanoff 2015) and derivatives in -äṣṣä-/-äske- with causative interpretation
(so-called “Kausativa”) show constant initial accent. Example (2.42i) shows that a phonological
word consisting of two syllables has primary stress on the first syllable. However, when a PC -
ne (3sg) or -me (pl) attaches to a disyllabic host, it makes the prosodic unit trisyllabic, and as a
result, the primary stress falls on the second syllable (2.42ii).
30
(2.42) PCs affect the stress calculation of a verb (ω = phonological word)
i. (PWd σ́ σ ): /akṣa/ → [ákṣə] ākṣa ‘hear.pst.act.3sg’ (B18a1)
ii. (PWd σ σ́ σ ): /akṣa-ne/ → [əkṣáne] akṣā-ne ‘hear.pst.act.3sg-3sg’ (B18a1)
Adams (2015: 74 n. 91) points out that of the roughly 500 examples of verb-PC combinations reg-
istered by Krause (1952), ṣarpau-me ‘indicate’ [subj.act.1sg-pl] is the only certain case in which
a PC does not affect stress calculation (cf. ṣärpau-me with expected peninitial accent). Krause
(1952: 297) lists “ṣarpau-me F, A2a4” next to “ṣärpau-me 33 (Š)b6.” Adams (2013: 718) also regis-
ters ṣarpau-me, citing PK AS 6C (= Krause’s F, A2) a4. However, PK AS 6C a4 undoubtedly attests
ṣärpau-me, rather than †ṣarpau-me (see Figure 2.1, where the first akṣara is 〈ṣa〉 rather than 〈ṣa〉). I
was not able to find an attestation of †ṣarpau-me.15 Malzahn (2010: 929) also registers “ṣarpau-me”
with “(sic)”, but without citation.
Figure 2.1: 〈ṣa rpau me〉 in PKAS 6C a4
Tocharian PCs are typologically classified as internal clitics, which attach to a host to project a
phonological word.16 In contrast, other clitics and the secondary case markers do not affect the
stress calculation of their host and constitute a larger prosodic unit than a phonological word
(identified as a Clitic Group by Koller 2015).17 The secondary case markers in TA also behave in the
same ways as clitics in that the ablative and allative allomorphs (anac/anäṣ) trigger gemination
of a preceding consonant just as oki ‘as, just’ does (Koller 2015).
(2.43) (ClG (PWd weñā-mm ) anac ) ‘(s/he) spoke to them’ (A269+A290a3; [2.15])
15. Adam Catt (p.c.) pointed out to me that Krause’s ṣarpau-me stems from the poor-quality transliteration by Lévi
(1933: 73).
16. This is comparable to the Latin enclitic conjunction -que which is also known to trigger stress-shift (see Weiss
2020: 121 and Hackstein 2017: 1307; e.g., plḗra ‘very many.f’ vs. plērá-que; cf. x plḗra-que). For different types of
prosodic incorporation, see Goldstein (2016: ch. 3).
17. See Section 2.2.4 on the secondary case markers. Causal -ñ is exceptional in that it affects stress calculation
(e.g., /kawa-ñə/ → [kəwá-ñ] kawā-ñ ‘out of desire’ (PKAS7Lb3); cf. /kawa/ → [káwə] kāwa ‘desire’ [acc.sg]). Ablative
-meṃ also affects stress occasionally.
31
Cf. (ClG (PWd pācarr ) oki ) ‘as if (he were) a father’ (A61a4; [2.16])
When a preterite participle or a gerundive hosts a PC in TA, the connection between the host and
the clitic seems less tight than that between a finite verb and a PC because word-level sandhi is
apparently absent (Schulze, Sieg, and Siegling 1931: 167). For example, underlying /-ṣ-c-/ assim-
ilates to -śś- in TA. When the finite verb tāṣ ‘(s/he) will be’ [subj.act.3sg] hosts the second-person
singular PC (-ci) as in (2.44), the outcome is (/tāṣ/+/-ci/ →) tāś-śi ‘(s/he) will be’ [subj.act.3sg-2sg]
(TEB I: 73). In contrast, in (2.45), where the preterite participle laltuṣ ‘gone out’ hosts a second-
person singular PC, the assimilation of /-ṣ-c-/ to -śś- is absent, and /-ṣ-c-/ surfaces faithfully.
(2.44) [TA] Assimilation of /-ṣ-c-/ to -śś-

///t-ac kakmuräṣ träṅkäṣ tärkor tāś-śi mäśkit
prince-all come.abs speak.npst.act.3sg permission cop.subj.act.3sg-2sg prince
pläc w· ///
go.out.imp.act.2sg
‘(The king) came to the prince and says: “You shall have permission, o prince. Become
(a monk!) (lit. Go out [from the house]!) …”
(A81a3; prose)
(2.45) [TA] Lack of assimilation of /-ṣ-c-/ to -śś-

– laltuṣ ci cesäm pälkoräṣ nu tmä///
go.out.ptcp.m.nom.pl 2sg them see.abs conj
‘[Those who] went out from you … But having seen them, … ’ (A125b2)
This difference seems to indicate that PCs and non-finite forms constitute a larger prosodic group
than a phonological word.18
(2.46) Prosodic structures of tāś-śi in (2.44) and laltuṣ-ci in (2.45)
i. (PWd tāś-śi ) ‘there will be (permission) to yousg ’
18. Also note the lack of assimilation in the non-finite forms in footnote 4 (e.g., kāruṃ-ci instead of †kāru(ṃ)ñ-ci
‘compassion-2sg’ [A260a4]). Why non-finite verbs with a PC form a larger prosodic group than a phonological word
is an open question.
32
ii. (ClG? (PWd laltuṣ ) ci ) ‘(those who) departed’
This section reviewed the pronominal system and the phonological, morphological, and syntactic
characteristics of the pronominal clitics of Tocharian A and B. The following section discusses
whether the Tocharian PCs should be taken as affixes or clitics.
2.3 Tocharian PCs and the typology of clitics and affixes
Clitics are linguistic elements that display prosodically deficient phonology, anomalous mor-
phosyntax, or both (Anderson 2005: 33). For the Tocharian PCs, we use the term “pronominal
clitics” since they are pronominal and prosodically deficient, and because they show anomalous
morphosyntax.
Scholars refer to the Tocharian PCs with various names. Previous approaches are divisible into
two camps: those who consider them clitics and those who see them as affixes. The former in-
cludes Schulze, Sieg, and Siegling (1931), Pedersen (1941), Couvreur (1947), Adams (1988, 2015),
Carling (2006), Kim (2009), Malzahn (2010), and Meunier (2015). Those who belong to the lat-
ter are: Krause (1952), TEB I, Schmidt (1974), Pinault (1992, 2008), Klingenschmitt (1994), Ringe
(1996), and Peyrot (2013b).
The question here is whether the Tocharian PCs are clitics or affixes. Scholars have long sought a
set of diagnostics by which to distinguish clitics from affixes. Perhaps one of the most influential
studies is Zwicky and Pullum (1983). They used the following six criteria and concluded that
the English contracted auxiliaries such as ’s in She’s gone are clitics, while the English contracted
negative n’t in She hasn’t gone is an affix.
33
(2.47) Zwicky and Pullum’s (1983: 503–4) criteria:
i. “Clitics can exhibit a low degree of selection with respect to their hosts, while affixes
exhibit a high degree of selection with respect to their stems.”
ii. “Arbitrary gaps in the set of combinations are more characteristic of affixed words
than of clitic groups.”
iii. “Morphophonological idiosyncrasies are more characteristic of affixed words than of

clitic groups.”
iv. “Semantic idiosyncrasies are more characteristic of affixed words than of clitic
groups.”
v. “Syntactic rules can affect affixed words, but cannot affect clitic groups.”
vi. “Clitics can attach to material already containing clitics, but affixes cannot.”
With respect to their criteria, Tocharian PCs show mixed behavior. Regarding (2.47i), PCs in TB
are selective in that only finite verbs can host a PC in TB. In contrast, those in TA are less selective,
and participles, gerundives, and nouns in a nominal clause may serve as a host (Section 2.2.2). As
for (2.47ii), it is unclear whether there is any verb that cannot host any PC in TA or TB.
Regarding (2.47iii), the PCs in TB may trigger allomorphy of the copula: ste and skente are the reg-
ular finite forms of the copula in TB, third-person singular and plural, respectively. When they
host a PC (e.g., -ne), however, we find star-ne (cop.npst.3sg-3sg) and skentar-ne (cop.npst.3pl-3sg),
instead of †ste-ne and †skente-ne. In TA, the combination of a finite copula naṣ and a PC -äṃ/-äm
is very seldom attested (Burlak and Itkin 2009),19 and instead, we find phonologically reduced
forms n-äṃ and n-äm. Moreover, the function of the PC in näṃ and näm is sometimes opaque,
and at least in some cases they seem to be frozen and no longer analyzable as containing a PC
(Schulze, Sieg, and Siegling 1931: 167–8, TEB I: 198).20 Pinault (2008: 639) considers that naṣ-äṃ,
naṣ-äm were replaced by näṃ and näm, respectively (“Les formes avec pronom suffixé, à savoir 3e
sg. naṣ-äṃ, naṣ-äm sont remplacées par les formes tronquées n-äṃ, n-äm”). However, his account
19. These forms are attested in the following manuscripts: naṣ-äṃ: A90b5; A98a4; A106b5, b6; A146b5; PKNS2b4;
naṣ-äm: A150a5; A346b2.
20. TEB I: 198: “Dabei kann näṃ, auch in der einfachen Bedeutung von naṣ stehen, also ohne Pron. suff.”
34
would amount to the claim that morphologically opaque forms (n-äṃ and n-äm) ousted morpho-
logically transparent forms, the opposite of which is usually observed. According to Burlak and
Itkin (2009), naṣ-äṃ and naṣ-äm are used only when a sentence contains an interrogative word or
a negation. This view would suggest that naṣ-äṃ/naṣ-äm and n-äṃ/n-äm are in complementary
distribution. Batke (1999: 36), following Pinault (1992: 133), considers n-äṃ and n-äm to be mere
allegro forms of naṣ-äṃ and naṣ-äm, respectively. But these views require an additional explana-
tion as to the fact that, in contrast to naṣ-äṃ and naṣ-äm, which only take a third-person singular
subject, both näṃ and näm may take a third-person singular or plural subject (Schulze, Sieg, and
Siegling 1931: 167).
As for (2.47iv), we find no example in which a Tocharian PC shows idiosyncratic semantics. As

for (2.47v), Tocharian PCs form a cluster with their host and move as a unit. For example, we find
a finite verb with a PC that precedes a direct object, seemingly undergoing some kind of fronting
operation (2.48).
(2.48) [TB] Fronting of a finite verb and a PC

spaitu ra waltsa= ñy ¦ āsta lykaśke po wnolmi •
dust like crush.subj.act.3pl gen.1pl bone.pl small all living.being.nom.pl
kärśye-ñ kektseñ wat ¦ kwä – [b5] ///

cut.subj.act.3pl-1sg body.acc.sg or
‘Even if all beings crush my bones fine like to dust, [5a] or if they chop up my body …’
(B220b4; verse; [5¦8]×4 + [8¦8¦5]; trans. by CEToM)
As for (2.47vi), we find no example in which a PC attaches to a base containing another clitic in TB.
In TA, there is only one example where an adverb (skam ‘always’) appears to intervene between
the host and the PC (2.9), but an alternative analysis is available for this example (Section 2.2.2).
As these criteria show, Tocharian PCs exhibit mixed behavior. Regarding (2.47i), one might be
tempted to conclude that the PCs in TB are affixes while those in TA are clitics. However, since
the degree of selection is gradient, and since there is no clear line which separates clitics from
affixes, we may only conclude that PCs in TB are more affix-like than those in TA. Criteria (2.47ii),
35
(2.47iii), and (2.47iv) are about the characteristics of affixed words. We have two pieces of nega-
tive evidence (2.47ii and 2.47iv) favoring clitics and one piece of positive evidence (2.47iii) favor-
ing affixes.
It should be noted, however, that the properties mentioned in (2.47ii), (2.47iii), and (2.47iv) are
“more characteristic of affixed words than of clitic groups”, and it is possible for a clitic-host
complex to show one or more of these properties. In fact, as Spencer and Luís (2012: 110–1) em-
phasize, “[t]hese criteria […] indicate tendencies and not defining characteristics that allow us to
determine with absolute certainty whether a given formative is an affix or a clitic” (Spencer and
Luís 2012: 110–1). Clitics may also show an allomorph conditioned by the host (cf. 2.47iii). For
example, subject pronouns are conditioned by a verbal inflection in Cléire and other coastal Mun-
ster varieties of Irish (Bennett, Elfner, and McCloskey 2019: 72–3; Yuan 2021), and possessive ’s in
English has a zero allomorph (Nevins 2011b; Anderson 2008, 2013; cf. Lowe 2016). Furthermore,
in contrast to (2.47iv), clitic-host complexes may show idiosyncratic semantics (e.g., French il y
a ‘there is’; Anderson 2011). Therefore, these three criteria do not allow us to conclude whether
Tocharian PCs are clitics or affixes.
As for (2.47v), Tocharian PCs form a cluster with their host and move as a unit. For example,
when a hosting verb undergoes fronting to the beginning of a sentence, it consistently carries
the PC together with it. However, this does not suggest that the Tocharian PCs are affixes be-
cause nothing rules out the possibility that a host and a clitic form a constituent and undergo
a syntactic operation, the clitic subsequently cliticizing to the host in the phonology or in the
syntax-phonology interface.
It is now clear that one cannot determine whether a Tocharian PC is an affix or a clitic based solely
on Zwicky and Pullum’s (1983) criteria. The Tocharian PCs show mixed behavior with respect to
these criteria, and the criteria themselves do not guarantee whether a given formative is an affix
or a clitic.
However, based on a particular theoretical framework, one might conclude that they are in fact
clitics. Section 2.2.2 showed PCs representing the argument of an infinitive climb to the finite
36
verb in TA and TB. If the lexicon created a finite verb-PC complex and sent it to the syntax as a
single unit, it would somehow know in advance that the finite verb will combine with an infinitive
that takes a pronominal argument. Therefore, with the assumption that (1) inflection takes place
in morphology (the Strong Lexicalist Hypothesis), and (2) syntax cannot access the internal
structure of a word (the Lexical Integrity Hypothesis; Chomsky 1970), one may conclude that
PCs are not affixes added in morphology but clitics manipulated in syntax.
In contrast, if one rejects these hypotheses and assumes that sentence-building processes may
have access to word-building processes, the difference between clitics and affixes becomes more
subtle. For example, Distributed Morphology (DM; Halle and Marantz 1993; Marantz 1997) does
not distinguish between word-building and sentence-building processes but considers them iden-
tical, operating in the same grammatical module (syntax). DM is a realizational model of mor-
phology in which “a word’s association with a particular set of morphosyntactic properties li-
censes the introduction of those properties’ inflectional exponents” (Stump 2001: 2).21 In this
model, both affixes and clitics realize some morphosyntactic feature bundles, and base-affix and
host-clitic complexes are both created post-syntactically (see Chapter 4 on this realizational pro-
cess).
Some scholars have proposed alternative criteria that distinguish between (doubling) clitics and
agreement markers (e.g., Nevins 2011a; Yuan 2021). However, one should be cautious in applying
them here because not all agreement markers are affixes (e.g., Sorani; Haig 2008; Jügel 2009),
and not all affixes are (grammatical) agreement markers either (e.g., Chichewa; Bresnan and
Mchombo 1987).
To summarize, the Tocharian PCs display mixed behavior. They show clitic climbing, suggesting
that they cannot be affixes added to a base in the morphology and dispatched to the syntax as
a single indivisible unit. Whether a Tocharian PC is a clitic or an affix is essentially a theory-
dependent question, and this study follows a (lexical-)realizational model of morphology (DM),
21. Realizational models of morphology include Distributed Morphology (Halle and Marantz 1993, 1994; Embick
2010, 2015), Paradigm Function Morphology 1 (Stump 1993, Stump 2001), Paradigm Function Morphology 2 (Stump
2002, Stewart and Stump 2007), A-morphous morphology (Anderson 1992), and Word-and-Paradigm morphology
(Blevins 2006, 2016; Blevins, Ackerman, and Malouf 2018).
37
in which the difference between clitics and affixes becomes blurred. Under the DM approach,
Tocharian PCs are both listemes that realize morphosyntactic feature bundles post-syntactically
and form a single phonological word with their host in the phonology.
2.4 Etymology of the Tocharian PCs
This section briefly reviews the etymologies of the Tocharian pronominal clitics proposed in pre-
vious literature.
The third-person singular PCs TB -ne and TA -(ä)ṃ point to Proto-Tocharian *-næ-. It is usually
compared with Lith. anàs, OCS onŭ ‘that (one)’, Hitt. anedani ‘this’ and ani- in anišiwat ‘today’,
which all point to *ono-.22 The Proto-Tocharian form seems to continue the masculine or neuter
accusative singular PIE *onom (cf. Van Windekens 1976: 276). However, as Peyrot (2017: 641)
points out, Proto-Tocharian *-næ- “could reflect *no- as well as *ono-, *eno- or *ano-” (cf. also LIPP
II: 55–6: *ano- or *no-). Van Windekens (1944: 188) connects TA -(ä)ṃ and TB -ne with Ved. nāń ā ‘in
various ways’ and Arm. na ‘that one,’ and reconstructs *ne/o-. As for Vedic nāń ā (RV+), however,
its origin is disputed (KeWA II: 153: “Die weitere Herkunft ist nicht geklärt”). Recent etymological
studies (KeWA II: 153; EWAia II: 35) favor Thieme’s (1949: 51–4) explanation, according to which it
goes back to the repetition of nā́ ‘man’ (i.e., ‘each for oneself ’, originally ‘man for man’). Never-
theless, Arm. na unambiguously points to *no-,23 and it is hard to rule out the possibility that PT
*-næ- continues this stem. Alternatively, Pedersen (1941: 137–9) sees it as an old adverb meaning
‘therein’ (cf. the secondary locative case marker -ne in TB; Lat. endo; Hitt. anda), but his idea was
rejected by Van Windekens (1944: 188 n. 10; 1976: 276).
The plural PC TB -me and TA -m go back to PT *-mæ-, which has two sources: one is *smos, a
zero-grade pronominal stem *s- with a dative plural ending *-mos, and the other is the first- and
22. On the Hittite forms, see Neu (1991: 22 n. 31; 1997: 156) and Melchert (1994a: 74–5; 1994b: 303; 2009).
Kloekhorst (2008: 767) analyzes Hitt. anišiwat differently, proposing to emend a-ni-ši-u̯ a-at (KBo 3.45 obv. 12 [OH/NS])
to e! -ni-ši-u̯ a-at, where eni- corresponds to the NH form of the neuter nominative-accusative singular ini ‘that’ (but
cf. Melchert 2009: 151 n. 1). Melchert (1991: 139 n. 17; 1994a: 75) adds to the list Lydian ãn(a)- as another possible
reflex of *óno- in Anatolian, but this analysis is abandoned in Melchert (2009).
23. See, e.g., Godel (1975: 107); Schmitt (2007: 120); Klein (2017: 1060); LIPP II: 56; and Olsen (2017: 1088).
Martirosyan (2010: 562) reconstructs “PIE *(h2 e)no-” without a comment on the parenthesis.
38
second-person plural pronouns *n̥ s-mé and us-wé > *smy ə with aphaeresis (Čop 1974, Katz 1998:
152–72).
We see the trace of *smos in the third-person plural dative enclitic pronouns in Anatolian: PA
*=smos ‘to them’ > Hitt. =šmaš, CLuw. =mmaš, HLuw. =ma-za /=mmant͡s/, Lyc. =ñne, and Lyd. =mś.
For the latter source, however, PT *-æ- points to PIE *-o- or *-ē-, and PIE *n̥ s-mé and us-wé (>
*sy my ə) would not develop to *-mæ but *-m(y) ə in PT. To account for the vocalism, Adams (2013:
502) starts from “PIE *-n̥ smó,” although independent support for the o-vocalism is rather limited
(cf. Dor. �μέ, �μέ; Aeol. ἄμμε, ὔμμε). Klingenschmitt (1994: 362), in contrast, reconstructs *n̥ smēm
́ > Ved. asmāń ; AiGr III: 467) or *n̥ smé
and *usmēm, that is, either *n̥ smé + em (cf. Pre-Ved. *asmām
+ om with a special contraction rule (for the trace of *-om in Tocharian, cf. *tuH-om > TB twe, Ved.
tvám, tuvám). Alternatively, Katz (1998: 163) suggests the tonic pronouns underwent aphaeresis
and monosyllabic lengthening (cf. Winter 1992: 99), giving rise to *ē (> PT *æ): *n̥ s-mé, *us-wé >
*smé (aphaeresis) > *smḗ (monosyllabic lengthening) > *m(y) æ (loss of stress).
Some scholars favor an analogical explanation for the *æ-vocalism in PT. Čop (1974: 34) sets up an
intermediate stage *n̥ smos and *usmos, influenced by *=smos (> *=me) and the clitic forms *nos and
*wos. Adams (1988: 155) follows him, stating “[m]uch more likely is Čop’s (1974) proposal which
would see *-me as the more or less regular phonological development of the expected Proto-Indo-
European enclitics *n̥ sme, *usme, […]. Here again the final vowel must be analogical in origin […].”
Alternatively, one could assume that *smos, which regularly develops into PT *-mæ, replaced the
vocalism of (*n̥ smé, *uswé >) *smy ə. Pisani (1941–1942) and Van Windekens (1976: 276) connect
PT *-mæ- with Ved. amá- and amú-, but their idea has not met with wide acceptance since it does
not explain the person-invariance observed in the plural PC (cf. Adams 1988: 155; 2013: 502).
The second-person singular PC ΤΒ -c does not align with TA -ci. While the former points to PT
*-cə, the latter suggests *-cəy. The communis opinio is to consider TB -c /-cə/ as continuing a PIE
atonic pronoun *te (> PT *-cə > TB -c /-cə/; Van Windekens 1976: 517, Mallory and Adams 1997:
455, Pinault 2008: 537; but cf. Kim 2009). Then, the question is how to explain the TA form.
According to Pinault (2008: 537), TA -ci is from (PIE *te >) PT *-cə, which was recharacterized in
pre-TA with a [+person] genitive marker *-i (i.e., *-cə+y > TA -ci). As per Van Windekens (1976:
39
517), TA -ci continues not PIE *te but PIE *toy (with initial *c analogical from *te [> *-cə > TB -
c]). However, Kim (2009) rightly questions this idea, as it would presuppose that Proto-Tocharian
somehow retained the reflexes of both PIE *te (> TB -c) and *toy (> TA -ci), seemingly without
any functional difference, or that Proto-Tocharian had (*te >) *-cə and (*toy >) *-cəy with some
functional difference (e.g., accusative vs. genitive-dative) which is no longer observable in the
daughter languages.
We see a TB-TA mismatch in the first-person singular also. ΤΒ -ñ points to PT *-ñə, while TA -ñi
suggests *-ñəy. Pinault (2008: 536–7) reconstructs PT *-ñə (> TB -ñ, recharacterized in TA as *-ñə+y
> TA -ñi), which continues either the genitive singular of the first-person pronoun (PIE *mene [cf.
Ved. máma; GAv. mə̄.nə̄; YAv. mana; OCS mene] > *my ny ə > *-ñə) or the atonic pronoun *me > *my ə
» *ñə with an initial consonant analogical from *my (ə)ny ə.
Alternatively, Kim (2009) starts from PT *-cəyə [2sg] and *-ñəyə [1sg], which developed into TA -ci
and -ñi with apocope (i.e., PT *-cəyə, *-ñəyə > pre-TA *-cəy, *-ñəy > TA -ci, -ñi), while the TB form
has undergone the loss of y between two unaccented schwas and subsequent contraction (i.e.,
PT *-cəyə, *-ñəyə > pre-TB *-cə.ə, *-ñə.ə > *-cə, *-ñə > TB -c /-cə/, -ñ /-ñə/). As for the source of
PT *-ñəyə and *-cəyə, Kim (2009: 57–8) tentatively suggests that they continue the dative of the
atonic pronouns *mey, *tey (cf. OCS mi, ti), which analogically obtained *-m from the third-person
singular accusative pronoun *nom (> PT *næ) (i.e., PIE *mey, *tey » *-ñəy, *-cəy [-ñ- analogical from
(*mene >) *my ny ə] » *-ñəy-əm, *-cəy-əm > PT *-ñəyə, *-cəyə). Although this analysis derives the TA
forms without appealing to the recharacterization with *i, support for the analogical change of
*-ñəy, *-cəy » *-ñəyəm, *-cəyəm is rather limited.
To summarize, this subsection briefly reviewed the etymologies proposed for the Tocharian PCs.
Although the PCs form a single pronominal paradigm, its members are heterogeneous. The third-
person singular PCs seem to continue a pronominal stem, either *eno-, *ono-, or *no-. The plural
PCs have two sources: PIE *smos and PIE *n̥ s-mé/us-wé. Although uncertainty remains as to how
to account for the æ-vocalism in PT, this analysis has an advantage over other proposals in that it
accounts for the person-invariance effect in the plural. The first- and the second-person singular
PCs show a mismatch between TA and TB. It seems that the TB outcomes are lautgesetzlich, and
40
Pre-TA recharacterized the PT forms with *i. Both the TA and TB forms might be lautgesetzlich
if there were independent evidence to support PT *-ñəyə and *-cəyə.
2.5 Interim summary
This chapter reviewed the pronominal system and the phonological, morphological, and syntac-
tic characteristics of the pronominal clitics in Tocharian A and B.
PCs in TA and TB differ from the demonstrative and independent personal pronouns in that they
lack case distinctions. They also lack person distinctions in the plural (Section 2.2.1). While PCs
in TB are consistently hosted by a finite verb, those in TA may be hosted by a participle, gerun-
dive, or a noun in a nominal clause (Section 2.2.2). There is one example in which a host-PC unit
appears interrupted by an adverb, but an alternative explanation is available for this example.
Section 2.2.3 showed that the TA first-person singular PC ñi is homophonous with the genitive-
dative of the first-person singular independent personal pronoun, and that there are ambiguous
cases as to whether ñi is an independent personal pronoun or a PC. PCs may be followed by an
allative or an ablative case marker, and in such cases, TA attests anac and anäṣ, instead of the
usual -ac and -äṣ (Section 2.2.4). While previous handbooks and dictionaries separate -an- in anac
and anäṣ, it was shown that these markers are to be treated as synchronically monomorphemic
(Sections 2.2.4 and 2.2.5). PCs in TB constitute a single phonological word with their host, while
the host-PC connection seems to be weaker in TA when the host is not a finite verb (Section
2.2.6). The Tocharian PCs show mixed behavior with respect to the typology of clitics and affixes
(Section 2.3). Even though they form a single pronominal paradigm, the origin of the PCs is het-
erogeneous, and several pronouns came to constitute the single PC paradigm (Section 2.4). The
following chapter will review the chief usages of PCs in TB and TA.
41
CHAPTER 3
Descriptive data of the Tocharian pronominal clitics
3.1 Chief usages of the Tocharian pronominal clitics
This chapter reviews the representative usages of pronominal clitics in Tocharian A and B. Tochar-
ian PCs are multifunctional: they may represent the direct object with a theme role, the indirect
object with various thematic roles such as source, goal, addressee, recipient, beneficiary, experi-
encer, and location, the possessor, the subject of a non-finite verb, the argument of a predicate
consisting of an adverb or postposition and a verb, and so on. According to the rough estimate
given by Adams (2015: 148), independent personal pronouns and demonstrative pronouns occur
about 80% of the time and PCs 20%. This chapter reviews all of the chief usages with examples
from TB and TA.
3.1.1 Theme
Tocharian PCs may represent theme of a transitive verb. The TB third-person singular PC -ne
serves as the theme of palātai ‘(yousg ) praised X’ in (2.4), repeated here as (3.1). In (3.2), the TA
first-person singular PC -ñi represents the theme of the transitive verb pälkse- ‘(they) torture X.’
(3.1) [TB] (= 2.4) -ne = Theme of palātai

never protect-inf forget.pst.2sg

‘Yousg have never forgotten to protect (the moral behavior). Yousg have praised him (=
the Buddha) for seven days with a single strophe.’ (B297a.a4; verse; [7¦7¦4]×4)
42
(3.2) [TA] -ñi = Theme of pälkse-
kuc yāmwā kᵤyal pälkse-ñi :2
what.acc.sg do.pst.act.1sg why torture.npst.act.3pl-1sg
‘… What did I do? Why do (they) torture me?’
(A101b5; verse)
In these examples, the PCs represent the theme of a transitive verb.
3.1.2 Addressee
PCs may represent the addressee of communication verbs such as TB/TA √āks- ‘announce, pro-
claim, say’, TB √we-ñ- ‘[act] say, speak; [mid] be called’, TA √träṅk-|we-ñ- ‘id.’, TA √pärk- ‘[act] ask
for, beg; [mid] ask, bring up a question’, and TB√pärk- ‘id.’ In the following examples, PCs repre-
sent the addressee of a transitive verb.
(3.3) [TB] -me (pl) = Addressee of akṣā-
(ce ślok a)kṣā-me ¦ ku ce tne wnolmi ¦ yamantär (:)
this.acc.sg strophe proclaim.pst.3sg-pl rel.acc.sg here being.pl do.pst.subj.3pl
krent yo(laiṃ yāmor) ¦ ///

good.acc.sg bad.acc.sg deed
‘(The Buddha) proclaimed this strophe to them: “Whatever living beings do here, [51a]
(whether) a good (or) bad deed …”’.
(B21a2; verse; [5¦4¦3]×4)
(3.4) [TA] -ci (2sg) = Addressee of prakäsmār-

(kä)lp(a)-s-yo cmol-m-aṃ1 ¦ mā kaklyuṣunt ñom ¦ klyoṣā tämyo
Kalpa-pl-ins birth-pl-loc neg hear.ptcp.m.acc.sg name hear.pst.act.1sg therefore
prakäsmār-ci ¦ kus täm män= täm pättāñkä(t [b4] :) 1 ¦¦
ask.npst.mid.1sg-2sg who.nom.sg dem how dem Buddha.lord
[Bṛhaddyuti speaking to Ānanda:] “I have heard a name unheard in [my] [re]births over
[many] kalpas. Therefore I ask yousg a question: who is that? What does ‘the Buddha’
mean? [1d] ”
(A20b3; trans. based on CEToM; verse; [5¦5¦8¦7]×4)
1. cmolmaṃ seems to be an error for cmolwaṃ (Sieg and Siegling 1921: 16 n. 12).
43
In example (3.3), the plural PC -me represents the addressee of akṣā- ‘(he) proclaimed X to Y.’
Likewise, example (3.4) shows that the second-person singular PC -ci represents the addressee of
prakäsmār- ‘(I) ask X a question.’
3.1.3 Recipient
PCs may represent the recipient of a verb of transaction (e.g., TA √e-|wä(s)? - ‘give’ and TB √ai-|wä(s)? -
‘[act] give; [mid] take’).2
(3.5) [TB] -ñ (1sg) = Recipient of wsasta-

(wsa)[a1]sta-ñ onwaññe • lkoym-c kru i ynemane ¦ ypauna
give.pst.act.2sg immortality see.opt.act.1sg-2sg whenever go.ptcp land.pl
kwṣain-ne ci ¦ plu[a2]ṣṣi-ñ sak-sa palskw ārañce ¦
village.pl-loc acc.2sg float.impf.act.3sg-1sg happiness-perl mind heart
‘[…] Yousg (ga)ve immortality to me. [79b] Every time I saw yousg going through lands and
villages, my mind and heart leapt for joy.’
(B246a1; trans. based on CEToM; verse; [5¦5¦8¦7]×4)
(3.6) [TA] -äṃ (3sg) = Recipient of eṣṣ-

– (p)uk cmol-w-aṃ särki yṣ-äṃ ¦ eṣṣ-äṃ sᵤk-untu
all birth-pl-loc after go.npst.act.3sg-3sg give.npst.act.3sg-3sg fortune-pl
ñäkciyās ¦ napeṃ-ṣinās :
divine.f.acc.pl human-adjz.f.acc.pl
‘… follows it in all incarnations, gives him divine and human pleasures.’
(A14b2; trans. by CEToM; verse; [7|7|4]×4)
In (3.5), the first-person singular PC -ñ [TB] represents the recipient of wsasta- ‘(yousg ) gave X to
Y’. Likewise, the third-person singular PC -äṃ [TA] represents the recipient of eṣṣ- ‘(s/he) gives X
to Y’ in (3.6).
3.1.4 Goal
In the following examples, pronominal clitics represent the goal of motion verbs (e.g., TB √i-|mä(s)? -
‘to go’ and TA √i-|kälk- ‘id.’). The motion verbs may be intransitive or transitive.
2. The superscript question mark added to the end of a root indicates that it is uncertain whether the root has a
so-called a-character or not (cf. Malzahn 2010: 24).
44
(3.7) [TB] -ñä (1sg) = Goal of kmeñ-
mäktoṃ okt no toṃ rakṣātsanā kmeñ-ñä-ścä ///
refl.f.pl 8 conj dem.f.pl female.demon.pl come.pst.act.3pl-1sg-all
‘… In these same eight [regions], however, these rākṣasas came to me.’
(Or 8212.163 b5; trans. based on CEToM; verse)
(3.8) [TA] -äṃn-anac (3sg) = Goal of yeṣ-

pkāt nuṃ kos-si ṣñi mācär ¦ yeṣ-äṃn-anac
intend.pst.mid.3sg again kill-inf own mother go.impf.act.3sg-3sg-all
‘(Sunakṣatra) again intended to kill (his) mother (and) went to her.’
(A222b6; verse; [7|7]×4)
The first-person singular PC -ñä represents the goal of the change-of-location verb kmeñ- ‘(they)
came’ in (3.7). The third-person singular PC -äṃn, referring to Sunakṣatra’s mother, represents
the goal of yeṣ- ‘(he) went’ in (3.8). Both cases contain an allative case marker, which unambigu-
ously marks the goal of a change-of-location verb.
In the following examples, PCs represent the goal of a transitive verb.
(3.9) [TB] -ne (3sg) = Goal of śilāre

śi[a8]lāre-ne oṅkarñai ¦ wñār-ne purwar
bring.pst.act.3pl-3sg porridge.acc.sg speak.pst.act.3pl-3sg accept.imp.mid.2sg
wesan-meṃ ¦ pinwāt rṣāka : 1 ||
1pl-abl alms sage.voc
‘(Nānda and Nandābala) brought the rice porridge to him (= Indra) and spoke to him.
“Receive the alms from us, o sage!” [1d] ’
(B107a7–8; verse; [7¦7¦4])
(3.10) [TA] -c(i) (2sg) = Goal of kleñc-

anaprä pe-s-ā oram pä[a3](ṣtam) /// oṅkälmāñ ñātse mā
in.front foot-pl-perl front(?) stand.imp.act.2sg elephant.nom.pl danger neg
kleñ-c(i)
bring.subj.act.3pl-2sg
[The Ṣaḍdanta Bodhisattva speaking to the hunter:] “Stand in front of my feet! … the
elephants should not bring danger to you.”
(A79a3; trans. based on CEToM; prose?)
45
Example (3.9) from TB shows that the third-person singular PC -ne represents the goal of the
ditransitive verb śilāre ‘(they) brought X to Y.’ Example (3.10) from TA contains kleñc- ‘(they) will
bring X to Y,’ which hosts the second-person singular PC -ci, representing the goal.
3.1.5 Beneficiary
In the following examples, PCs represent the beneficiary. The third-person singular PC -ne in
(3.11). seems to represent the beneficiary of yaṃ- ‘(if) she goes.’ In (3.12), we find an intransitive
verb sekaṣ- ‘X will be overflown, X will be abundant’, which hosts a plural PC. This PC, referring
to the king Brahmadatta and his attendants, represents the beneficiary of the hosting verb.
(3.11) [TB] -ne (3sg) = Benefactive of yaṃ-

ṣamāne ytāri mā aiśtär
monk.m.nom.sg path neg know.npst.subj.mid.3sg
klyiye ytāri ṣärpṣūkiññe-sa yaṃ-ne anāpatti •

woman.f.nom.sg path guide-perl go.npst.subj.act.3sg-3sg sinless
‘If a monk does not know a route and a woman goes for him as a path-guide, it is without
sin.’
(B330a2; trans. based on Ogihara 2009: 383; prose)
(3.12) [TA] -äm (pl) = Benefactive of sekaṣ-

[a2] /// sne plā wlamträ was tāloṣ
without exception(?) die.subj.mid.1pl nom.1pl miserable.nom.pl
sekaṣ-äm śwāl – – –
be.overflown.subj.act.3sg-pl flesh
“If we, the miserable ones, die without exception(?) , flesh will be overflown to youpl
(i.e., youpl will have flesh in abundance).”
(A72a2; trans. based on Peyrot 2013a: 167; prose).
In the examples above, the attested PCs represent the beneficiary of an intransitive verb. How-
ever, PCs may also represent the beneficiary of a transitive verb (e.g., 3.13 and 3.14).
46
(3.13) [TB] -ne (3sg) = Benefactive of nemar-
ñakti arjuṃ ¦ stām nemar-ne-ś ¦ cau eṅksate :
god.pl Arjuna.tree tree bend.pst.act.3pl-3sg-all dem.m.acc.sg seize.pst.mid.3sg
‘The gods bent the Arjuna-tree for him (= Bodhisattva); (He) took it.’
(B107b4; verse; [4¦4¦4]×4)
(3.14) [TA] -ṃ (3sg) = Benefactive of yāmwe-

– śkaṃ kärṣte-ṃ kāruṇ-yo :
conj cut.off.pst.mid.1sg-3sg compassion-ins
yāmwe-ṃ kluṣpe tmäṣ mā - ¦ ///

do.pst.mid.1sg-3sg rice.porridge then neg
‘… and I cut it off with compassion. I made rice porridge for him. Then, … not …’
(A321b6; verse; [7¦7]×4)
The third-person singular PC -ne functions as the beneficiary of the transitive nemar- ‘(they) bent
X (for Y)’ in (3.13). In example (3.14), the third-person singular PC -ṃ serves as the beneficiary of
yāmwe- ‘(they) made X (for Y)’.
3.1.6 Source
PCs may also represent the source of verbs of requesting such as TB √yāsk- ‘to beg’ and TB √ñäsk-
‘to demand, desire,’ and verbs of possessional deprivation such as TA √sumā- ‘to take away, deprive
of ’. The examples are as follows:
(3.15) [TB] (= 2.14) -ne (3sg) = Source of yaṣṣāte-

wsā-ne •
‘Upanandaj begged this cloak from himk (= an Ājīvika ascetic), (but hek ) did not give (it)
to himj .’ (PKAS18Ab2; prose)3
3. Cf. B337a4–5 attests the parallel passage: upanande cewmeṃ kampās yaṣāte [a5] sū mā wsā-ne • ‘Upananda begged
the cloak from him, (but he) did not give (it) to him.’
47
(3.16) [TA] -ñi (1sg) = Source of psumār-
[a5] /// cam śkaṃ lo psumār-ñi kᵤyal lykäly lykäly
dem.m.acc.sg conj far take.away.imp.mid.2sg-1sg why fine fine
tuṣt-ñi :
burn.npst.act.2sg-1sg
‘… and take that (suffering? ) away from me! Why do you burn me finer (and) finer? ...’
(A92a5; trans. based on CEToM; verse)4
The third-person singular PC -ne represents the source of the transitive verb yaṣṣāte- ‘(he) begged
X from Y’ in (2.14), repeated here as (3.15). Example (6.14) from TA shows that the first-person
singular PC -ñi represents the source of the ditransitive verb psumār- ‘Take X away from Y!’.
3.1.7 Location
PCs may represent the location, and the hosting verbs may be intransitive verbs of appearance
such as TB √tsänkā- ‘to rise, arise’ or transitive such as TB √lup(ā)- ‘to rub, smear’ and TA √tā(-s)-
‘[act] to put, set, place; [mid] place oneself.’
In example (3.17), the plural PC -me represents the location of an intransitive verb of appearance
tsäṅkā- ‘(it) arose (in X)’.
(3.17) [TB] -me (pl) = Location of tsäṅkā-

/// -ts tsaṅka śrāvasti-ne ¦ pāk= auntsante tu yām-tsi :
arise.pst.act.3sg Śrāvastī-loc part begin.mid.pst.3pl dem do-inf
tsäṅkā-me weñye ¦ käll(au)-ntse ṣarmtsa ¦ ///
arise.pst.act.3sg-3pl talk profit-gen.sg for.the.sake.of
‘... rose. In Śrāvastī they began to take part in it. The discussion arose among them: For
the sake of profit …’
(B16b3; trans. based on CEToM and Hackstein, Habata, and Bross 2014: 77; verse;
[5¦5¦8¦7]×4)
The following two examples contain a PC that represents the location of a transitive verb.
4. Hackstein (1995: 349) takes tuṣt to be an intransitive (“warum brennst mir immer feiner?”).
48
(3.18) [TB] -ne (3sg) = Location of laupoy-
/// (ṣe)me ṣar-sa ceṃ pre(re5 – wse-cce sä)lkoy-ne •
one hand-perl dem.m.acc.sg arrow poison-adjz.acc.sg pull.opt.act.3sg-3sg
wace ṣar-sa (sāṃ)tke-nta laupoy-ne āñu yamī-ne •
second hand-perl medicine-pl smear.opt.act.3sg-3sg quiet make.opt.act.3sg-3sg
‘[and] with one hand [he] may draw this poisoned arrow [i.e., the vedanāskandha] out of
it, [and] with the other (lit. second) hand he may smear medicine on it and make him
quiet.’ (IOLToch4b1; trans. based on CEToM)
(3.19) [TA] -äṃ (3sg) = Location of tsānt-

tm-äk ṣu[b2](rmaṣ) /// ñom tsānt-äṃ ||
dem-emp reason name put.pst.mid.3pl-3sg
‘Because of this very reason, (her relatives) put the name (Unmādayantī) on her.’
(A59b2; prose)6
The third-person singular PC -ne (3sg) represents the location of the ditransitive verb laupoy
‘(s/he) may smear X onto Y’ in (3.18). In example (3.19) from TA, the third-person singular PC -äṃ
represents the (metaphorical) location of the ditransitive verb tsānt- ‘(they) put X on Y’. In these
examples, the PC representing the location does not indicate the place where the action the verb
phrase describes is to be performed, but rather the location where the theme object (medicine
in [3.18]; the name in [3.19]) is placed or given.
3.1.8 Experiencer
In the following examples, PCs represent the experiencer of an object experiencer verb (e.g., TB
√läk(ā)- ‘to see, look’ [‘to be seen, appear’ with the middle endings] and TA √klāwā- ‘to fall’).
5. Broomhead (1962: 60) restores pre(re-sa) with the secondary perlative ending. However, this restoration is
unlikely as it is the DO (theme) of the verb (cf. Sieg and Siegling 1953: 83 n. 15).
6. This passage translates Sanskrit ataś ca tasyā unmādayantīty eva bāndhavā nāma cakruḥ || ‘and due to this her
relatives gave the name Unmādayantī to her’ (Kern 1891: 81). TA √tā(s)- ‘[act] ‘put, set, place; [mid] place oneself ’
is transitive even when it takes a middle inflectional ending. The existence of tāṣimār [opt.mid.1sg] and tāṣitär
[opt.mid.3sg], which Malzahn (2010: 642) lists as optatives built on the Subjunctive II stem, is debated (see Peyrot
2013b: 430). The existence of tse [pst.mid.1sg] in A159 a4 is likewise uncertain (CEToM: “The word division of nmitā
tse is not certain”). Other middle forms, including the preterites tsāte [pst.mid.3sg] (A435a2), tsānt [pst.mid.3pl]
(A434b6), tsānt-äṃ [pst.mid.3pl-3sg] (A59b2) and imperatives pätstsār [imp.mid.2sg] (A215b1=YQI.6b8) and pätstsāc
[imp.mid.2pl] (A68b5 and A57a5) are all transitive.
49
The second-person singular PC -c describes the experiencer of lkāntar ‘(they) will appear (to X)’
in (3.20). In TA, example (3.21) shows the plural PC -äm, representing the experiencer of the
intransitive verb klāṣ- ‘(it) may fall (on X); (it) may happen (to X)’.
(3.20) [TB] -c (2sg) = Experiencer of lkāntar

(śuddha)vāsä-ṣṣi ¦ ña(k)t(i) lkāntar-c kauñī ram no ¦
Śuddhāvāsa-adjz.nom.pl god.nom.pl see.subj.mid.3pl-2sg sun as conj
ompalskoñe ṣme(-mane :) ///
meditation sit.npst-ptcp
‘The (Śuddhā)vāsa gods will appear to yousg like a sun, sit(ting) in meditation (…)’
(B76a1; trans. based on Schmidt 1974: 234; verse; [5¦5¦8¦7]×4)
(3.21) [TA] -äm (pl) = Experiencer of klāṣ-

pälskānt kupre Śrāvastī riy-aṃ anne ymäs ṣakk-atsek
think.pst.mid.3pl whether Śrāvasti city-loc into go.npst.act.1pl certainly
– – – – – – – – (ñā)[b2]tse klāṣ-äm tämyo cam kausal-ṣiṃ
danger fall.subj.act.3sg-pl therefore dem.m.acc.sg Kosala-adjz.acc.sg
wärt āssuk mā katkar ||
forest pass neg cross.pst.act.3pl
‘They thought to themselves: “Should we go into the city of Śrāvastī? Certainly … danger
will fall on us!” Therefore they did not pass through this forest of Kosala …’
(A395b2; trans. by CEToM; prose)
The subject of klāṣ ‘X will fall’ in the last example is a metaphorical, and the PC in this example
represents a metaphorical location.7 It is often ambiguous whether a PC represents an experi-
encer or a location.
3.1.9 Stimulus
As the following examples show, PCs may represent the stimulus of subject experiencer verbs
such as TA/TB √pärsk(ā)- ‘be afraid’, and TB √mänt(ā)- ‘destroy’ (‘be destroyed, be stirred, angry’
with a middle ending).
7. Compare English “befall”, as in “danger will befall us.”
50
(3.22) [TB] -ñ = Stimulus of mäntantär
(mā mäntan)tär-ñ8 ptarkaso śconai mapi wase ñī käṣṣīññe
neg be.angry.npst.mid.2pl-1sg let.go.imp.act.2pl enmity ptcl poison gen.1sg teacher
i ///
‘[King Araṇemi speaking:] (“Don’t be ang)ry at me! Leave off hate [so that] the venom of
my teacher indeed ...”’ (B79a1; trans. based on CEToM; prose)
(3.23) [TA] -äṃ (3sg) = Stimulus of praskmār-
kus ne rāme-s prask-māṃ tāc (näṣ [b6] mā)
rel.nom comp Rāma-gen be.afraid-ptcp cop.subj.act.2pl nom.1sg neg
praskmār-äṃ
be.afraid.npst.mid.1sg-3sg
‘[Daśagrīva speaking to his brother Vibhīṣaṇa:] “(Even) if youpl should fear Rāma, I am
not afraid of him.”’
(A10b6; prose)
Example (3.22) shows that the first-person singular PC -ñ represents the stimulus of the intran-
sitive verb (mäntan)tär ‘(youpl ) are angry (at X).’ Likewise, example (3.23) has a third-person sin-
gular PC -äṃ, representing the stimulus of a subject experiencer verb (praskmār- ‘I am afraid [of
X]’).
3.1.10 Argument of a complex predicate
In some cases, PCs represent the argument of a transitive verbal predicate consisting of an in-
transitive verb and an adverb. In (3.24), we find the intransitive TB √i-|mä(s)? - ‘to go’, which is
used with ompostäṃ ‘after’ to form a complex predicate ompostäṃ yneṃ- ‘(they) go after X; (they)
follow X’. The third-person singular PC -ne represents an argument of this predicate.
8. This verb could be restored as (mäntana)tär-ñ (Present VI) or (mäntan)tär-ñ (Present XIIa) (Malzahn 2010: 753f.).
51
(3.24) [TB] -ñ = Argument of ompostäṃ + √i-|mä(s)? -
mäkte wassi swarenäṃ ¦ we[b4]r(eṃ)-mpa tattaṃ ksa walke
as garment sweet.m.acc.pl odor.pl-com put.subj.act.3sg indf for.a.long.time
(:)
waipte ka(rts)e weren-meṃ ¦ sū wassi ykāk swāre

apart good odor.pl-abl dem.m.nom.sg garment still sweet.nom.sg
warṣṣäṃ :
smell.npst.act.3sg
krentauna-mpa akaly(e) ¦ [b5] maṃt rano yāmträ ce

virtue.pl-com learning.nom.sg so also do.subj.mid.3sg dem.m.acc.sg
(c)mel-ne :
birth-loc
ceᵤ āklyi-sa cmel-a-ne ¦ ompostäṃ yneṃ-ne m=

dem.m.acc.sg learning-perl birth-pl-loc after go.npst.act.3pl-3sg neg
ārsen-ne 80-7
give.up.npst.act.3pl-3sg
‘[If] someone puts a garment together with sweet odors for a long time, [87a] even [when]
separated from these odors, this garment will still smell sweet. [87b] In this way also, [if] a
practice with virtues is done in this birth, [87c] because of this practice, they (= the virtues)
will follow him in (his re)births and will not abandon him. [87d] ’
(PKAS6Cb3-5; trans. based on CEToM; verse; [7¦8]×4)
In (3.25) and (3.26), the PCs -ne (3sg) and -me (pl) are semantically associated with the adverb/post-
position postäṃ ‘after’, which forms a transitive predicate postäṃ msā- ‘(s/he) went after X’ and
postäṃ ynem- ‘(we) will go after X’, respectively.
(3.25) [TB] -ne = Argument of postäṃ + √i-|mä(s)? -

nānda cāla oṅkorñai ¦ nandābala tāy ṣerśka ¦ postäṃ
Nānda carry.pst.act.3sg porridge Nandābala dem.gen little.sister after
msā-ne
go.pst.act.3sg-3sg
‘Nānda carried the porridge. Her littler sister Nandābala went after her.’
(B107a7; verse; [7¦7¦4]×4; trans. based on CEToM)
52
(3.26) [TB] -ne = Argument of postäṃ + √i-|mä(s)? -
cem ce cisso upādhyāyi mahāśra(maṇeṃ-ś)
dem.m.acc.sg dem.m.acc.sg go.imp.act.2pl master.pl great.mendicant-all
postäṃ wes ynem-me9 •

after nom.1pl go.subj.act.1pl-pl
[The disciples of Nadīkāśyapa and Gayākāśyapa speaking to their masters:] “Go to this
(or) that Great Mendicant, o masters! We will go after youpl .”
(B108a8; prose)
We find a similar verbal complex in TA also. In (3.27), the third person clitic -äṃ is semantically
associated with a transitive predicate consisting of an adverb/postposition särki ‘after’ and the
intransitive verb yṣ- ‘(s/he) goes’.
(3.27) [TA] -äṃ (3sg) = Argument of särki + √i-|kälk-

– (p)uk cmol-w-aṃ särki yṣ-äṃ ¦ eṣṣ-äṃ sᵤk-untu
all birth-pl-loc after go.npst.act.3sg-3sg give.npst.act.3sg-3sg fortune-pl
ñäkciyās ¦ napeṃ-ṣinās :
divine.f.acc.pl human-adjz.f.acc.pl
‘… goes after him? in all (re)births, (and) gives divine and human pleasures to him.’
(A14b2; trans. based on CEToM; verse; [7¦7¦4]×4)
In (3.28), we find an adverb/postposition anapär ‘before, in front’, which forms a predicate with
an intransitive √ṣäm-|läm(ā)- ‘to sit’, meaning ‘to sit in front of X’. The third-person singular PC
-ṃ represents an argument of this predicate.
(3.28) [TA] -äṃ (3sg) = Argument of anapär + √ṣäm-|läm(ā)-

täm pälko-r-äṣ säm wäl tsmont
dem see.ptcp-nmlz-abs dem.m.nom.sg king.m.nom.sg grow.ptcp.m.acc.sg
ynāñmune-yo [b6] /// anapär ly(m)ā-ṃ •
respect-ins in.front sit.down.pst.act.3sg-3sg
‘Having seen it, the king … with growing respect …. sat in front of him.’
(A147b6; verse?)
9. Restored as mahāśra(maṇeṃś) or mahāśra(maṇeś) (Thomas 1983: 129 n. 12). CEToM reads ynem-ne
‘go.subj.act.1pl-3sg’ instead of ynem-me (Thomas 1983: 129 n. 12). The reading cannot be confirmed as the original
manuscript is missing.
53
Example (3.29) attests an adverb akarte ‘near’, which construes with √ṣäm-|läm(ā)- ‘to sit’ to form
a complex predicate akarte √ṣäm-|läm(ā)- ‘to sit near X’. Likewise, example (3.30) shows a com-
plex predicate ate √nes-|tāk(ā)- ‘to be far from X’, consisting of an adverb ate ‘far’ and the copula
√nes-|tāk(ā)- ‘to be, become’. In both cases, the first-person singular PC -ñ serves as the argument
of the complex predicate.
(3.29) [TB] -ñ = Argument of akarte ‘near’ + √ṣäm-|läm(ā)-

/// ·kañ ¦ plamas-ñ akarte :
sit.imp.act.2pl-1sg near
hā larona waipecce-nta ¦ ṣañ śamñāṣ·· ///

alas dear.f.pl possession-pl own
‘... “(Youpl dear little sons), sit down near me! Alas, the dear possessions of the kinsmen,
…”’
(B46b4; verse; [5¦5¦5¦5] + [8¦7¦7] + [5¦5] + [8¦7] or [7¦8]; trans. based on CEToM)
(3.30) [TB] -ñ = Argument of ate ‘far’ + √nes-|tāk(ā)-

/// (ta)ñ koyna-meṃ reki klyauṣim || ate takāsta-ñ ///
gen.2sg mouth-abl word hear.opt.act.1sg far cop.pst.act.2sg-1sg
“May I hear a word from (you)rsg mouth! Yousg are far away from me (now).”
(B86b2; trans. based on CEToM)
In example (3.31), the second-person singular PC -ci represents the argument of a complex pred-
icate, consisting of pācyās ‘to the right’ and √käly-|ṣtäm(ā)- ‘to stand, be situated’.
(3.31) [TA] -ci (2sg) = Argument of pācyās + √käly-|ṣtäm(ā)-

pā«†ṃ»cyā(s kä)lytär-ci ¦ vajrapā·i – ¦ (śā)lyās pracar ānant ṣāmaṃ ¦
right stand.npst.mid.3sg-2sg Vajrapāṇi left brother Ānanda monk
wsokone-yo lkeñ=-cy akml-ac :
joy-ins see.act.npst.3pl-2sg face-all
‘Vajrapāṇi is standing to your right. (Your) brother Ānanda, the monk, is (standing to
your) left. They are looking at your face with joy.’
(A358a3; verse; [5¦5¦8¦7]×4(?) )
In this example, the PC does not represent a possessor but a reference point with respect to the
relative position between the addressee and the third party.
54
The transitive predicates we reviewed so far consist of an adverb/postposition and an intransitive
verb. But we also find a predicate consisting of an adverb and a transitive verb (3.32).
(3.32) [TA] -ci (2sg) = (Second) argument of posac + √pās-

posac pās(m)ār-ci ¦ riṣakune wärt-äntw-aṃ :
next.to uphold.subj.mid.1sg-2sg sagehood forest-pl-loc
‘I will uphold sagehood (i.e., practice asceticism) by your side in the forests. [4a] ’
(A99a3; verse; [5¦7]×4)
In this example, the second-person singular PC -ci construes with a complex predicate consist-
ing of a postposition posac ‘near, next to,’ which governs the genitive-dative (Schulze, Sieg, and
Siegling 1931: 290), and the transitive verb pās(m)ār- (built on the root √pās- ‘protect, obey [rules],
beware of ’), which selects riṣakune ‘sagehood’ as a direct object.
3.1.11 Experiencer of a complex predicate
PCs may represent the experiencer of a complex predicate consisting of an adjective and a copula.
In the following examples, adjectives such as TB pācar ‘clear, obvious,’ TA pākär ‘clear, visible,’ and
TB lāre ‘dear’ are used predicatively, and the PCs are hosted by a finite copula and represent the
experiencer, meaning ‘(something) is/becomes { clear / dear } to X.’
In (3.33), the first-person singular PC -ñ may construe with the adjective pācri ‘clear, obvious.’
This adjective is used predicatively, with the finite copula tāko- meaning ‘may X become clear to
Y; may X appear to Y’.
(3.33) [TB] -ñ (1sg) = Experiencer of pācri + √nes-|tāk(ā)-

(ke)[a1]ktseñi rākoyentär-ñ paine-ne po pūdñäkteṃts (:)
body.nom.pl extend.opt.mid.3pl-1sg foot.du-loc all Buddha.gen.pl
arañcä-ṣṣi uppālta [a2] pākri tāko-ñ yke postäṃ po
heart-adjz.m.nom.pl lotus.pl clear cop.opt.act.3pl-1sg place after all
saṃṣār-ne :
Saṃsāra-loc
“May the …. bodies of mine be extended to the feet of all Buddhas! May the lotuses of the
heart be visible to me one by one in the whole Saṃsāra!”
(B271a1; verse)
55
TA also attests the same combination. In example (3.34), the plural PC -äm represents the ex-
periencer of the predicative adjective pākär ‘clear, visible, evident’ (i.e., pākär tākar- ‘[they] were
visible’).
(3.34) [TA] -äm (pl) = Experiencer of pākär + √nas-|tāk(ā)-

ālās-āp klu krop-l-une-yā kalpavṛkṣ-änt(u [a5]
lazy-gen.sg rice assemble.subj-gdv-nmlz-perl wishing.tree-pl
na)känt-äm kappāñ pākär tākar-äm
disappear.pst.mid.3pl-pl cotton.bushes(?) clear cop.pst.act.3pl-3pl
‘[But] because of collecting rice by a listless person, the wishing trees disappeared from
them. The cotton plants were visible to them [in place of the wishing trees] …’
(A2a5; trans. based on CEToM; prose)
We find the combination of a PC, a predicative adjective, and TB √mäskā- ‘to be/become’ or TB
√nes-|tāk(ā)- (copula) in (3.35), (3.36), and (3.37).
(3.35) [TB] -ne = Experiencer of lāre + √mäskā-

sū cpī yāmor-ntse ¦ okosa wnolme ¦ ekñiññe-nta
dem.m.nom.sg dem.m.gen.sg deed-gen.sg fruit-perl human.being possession-pl
(makā-yäkne ¦ yänmāṣṣ-eñca mäsketra :
many-manner obtain-ptcp be.npst.mid.3sg
wa)[a4]raṣṣä-l(ñ)e-n(ts)e ¦ (m)e(ṅ)kītsñe-sa no ¦ entse lāre mäsketär-ne ¦ m=

practice-nmlz-gen.sg lack-perl conj envy dear be.npst.mid.3sg-3sg neg
āyor aitsi cäñcan-ne :
gift give-inf please.npst.act.3sg-pl
‘This man, because of the result of his deed, becomes the obtainer of possessions of every
kind. [12a] But because of [his] lack of cultivation, envy becomes dear to him, (and) he
does not give gifts happily (lit. it does not please him to give a gift). [12b] ’
(PKAS7Fa4; verse; [5¦5¦8¦7]×4)
(3.36) [TB] -ñ (1sg) = Experiencer of (pe)rn(e)w + √nes-|tāk(ā)-

(– – śa)no-ś weṣṣäṃ ṣarya kauṃ (s)ū (pe)rn(e)w
wife-all speak.act.npst.3sg beloved day dem.m.nom.sg glorious
t(a)kā-ñ ente ce śaumo(n) ///
cop.pst.act.3sg-1sg when dem.m.acc.sg man.acc
‘… (he) speaks to (his) wife: “My dear! This day was glorious to me when ... this man.”’
(B91a6)
56
(3.37) [TB] -co (2sg) = Experiencer of alecci + √nes-|tāk(ā)-
mäkte lwasā-ntso ¦ auṣuwaṃts ṣesa ¦ lyuketrä yṣīye
as animal.pl-gen dwell.ptcp.gen.pl together light.up.npst.mid.3sg night.nom.sg
¦ waiptā(yar rano :
apart conj
lwasā-ntso ton-ak auṣūwaṃts) ¦ pr(e)ntse yente käskan-me

animal.pl-gen dem.f.pl-emp dwell.ptcp.gen.pl instant wind scatter.npst.act.3sg-pl
¦ mant ṣañ śāmna keś ptes twe :
so own person.pl counting put.imp.act.2sg nom.2sg
kos twe yṣwar tāka(t)10 ¦ – – – [a8] – – (:)

as.much nom.2sg friendly cop.subj.act.2sg
śaul ka oräñ-c tā kektseño ¦ pw alecci

life emp abandon.subj.act.sg-2sg dem.f.acc.sg body.acc.sg all foreign.nom.pl
cai tākañ-co 30-4
dem.m.nom.pl cop.subj.act.pl-2sg
‘(Just) as when the animals [i.e. fireflies] are living together, the night will grow light,
but when these animals are living apart, wind scatters them instantly. In this way, pay
attention to your relatives! [34b] As long as yousg are friendly … [34c] ... (as soon as) the life
leaves this body, they will all be strange[rs] to yousg . [34d] ’
(B46a8; verse; [5¦5¦5¦5] + [8¦7¦7] + [5¦5] + [8¦7] or [7¦8])
In example (3.38), the second-person singular PC -cä, is construed with ynãñmo ‘appreciated, eval-
uated, judged’ (with an o-mobile) with a copula.
10. Cf. Sieg and Siegling (1949: 69) and Adams (2012: 23). B47 b6 attests /// ·aupūwaṃts prentse ye(nte) ///, where
aupūwaṃts is for auṣūwaṃts.
57
(3.38) [TB] -cä (2sg) = Complement of ynāñ(m)o + √nes-|tāk(ā)-
i(me) ce-k warñai ¦ kwr(i) kälpāsta kos rā
thought dem.m.acc.sg-emp beginning.with if obtain.pst.act.2sg as.much ptcl
tsa :
emp
pālka tomp ñake ¦ mäkte ynāñ(m)o tākañ-cä (:)

see.imp.act.2sg dem.f.acc.sg now how appreciated cop.subj.act.3sg-2sg
[King Supriya, looking at the princess Kañcanaprabhā, speaking to himself:] “If yousg have
obtained a (thou)ght, beginning with this one whatsoever, [2a] look at that one there now!
How much will (she) be valuable to yousg ? [2b] ”
(PKAS17Kb5; verse; [5¦7]×4)
When the indeclinable adjective ynāñm ‘appreciated, evaluated, judged’ co-occurs with a copula,
it selects a theme in the nominative and an experiencer in the genitive-dative (e.g., 3.39).
(3.39) [TB] Agent/experiencer of ynāñm ‘appreciated’ represented in the genitive-dative case

yaltse śaula-nma ra ¦ mā ñi kca ynā(ñmä :)
1000 life-pl ptcl neg gen.1sg indf appreciated
‘Even a thousand livestheme:nom (are) not valuable to meexperiencer:gen-dat .’
(B82b6; trans. based on CEToM; verse; [6¦5] or [5¦6]×4?)
In this example, the finite copula is apparently omitted.
3.1.12 Oblique possessor
PCs may also express the possessor. One such subtype is the so-called oblique possession con-
struction (or the mihi est or dative nominative construction), where the possessum is represented
in the nominative case and the possessor in the genitive-dative.11 In (3.40), for example, the pos-
sessum is in the nominative (āñme ‘wish’), and the third-person singular PC -ne represents the
possessor of āñme. In (3.41), the third-person singular PC -äṃ represents the oblique possessor
of kācke ‘joy,’ which is in the nominative-accusative.
11. Carling (2006: 37) calls this type an inverse construction.
58
(3.40) [TB] -ṃ = Oblique possessor of āñme
kṣatriye-mpa larauñe yām-tsi āñme (tā)kaṃ-ne
warrior-com friendship make-inf wish.nom cop.subj.3sg-3sg
(śā)[a7]l-ṣana arw-ā-ts koṣkīye yamaṣ-lya •

Shorea.robusta-adjz.f.pl wood-pl-gen pit.f.nom make.npst-gdv.f
‘(If) one has the wish to make friendship with a warrior, (he) should make a (fire-)pit of
branches of the Shorea robusta.’
(PKAS8Ca6; trans. based on CEToM; prose)12
(3.41) [TA] -ṃ = Oblique possessor of kācke

wät amok-«ä»ṣ tatmu kācke mäskatr-äṃ
second skill-abl be.born.ptcp.m.nom.sg joy be.npst.mid.3sg-3sg
trit wrass-äṣ ortune kälpnāträ :

third human.pl-abl friendship obtain.npst.mid.3sg
‘Secondly, he will have pleasure born from [his] skill. Thirdly, he will attain friendship
from human beings.’
(A2b6; prose)
We also find the possessum in the accusative case instead of the nominative. In (3.42), āklyi kreṃnt
yā(mor) ‘a good deed (as) a lesson,’ which is the possessum in the oblique possession construction,
is in the accusative rather than in the expected nominative (†āklye kartse yā(mor)).
(3.42) [TB] -c = possessor of āklyi kreṃnt yā(mor) ‘a good deed (as) a lesson’
/// (āklyi) krent ¦ yāmor-ne yāmtar kwri :
learning.acc.sg good.m.acc.sg deed-loc do.subj.mid.2sg if
aum no ppo lau cmel-n=13 alyek ¦ tākañ-c āklyi

then conj all far birth-loc another.acc.sg cop.subj.act.3sg-2sg learning.acc.sg
kreṃnt yā[b4](mor :) ///
good.m.acc.sg deed
[Hastāṅkuśa speaking to the king:] “… if you learn (…) in a good action, [2a] then in every
distant future birth you will have the good action (as) a lesson. [2b] ”
(PKNS34b3; verse; [7¦7]×4)
12. For the interpretation of TB koṣkīye, see Bernard and Chen (2022). I’m grateful to Adam Catt for reminding me
of this study.
13. cmel-n= for camel-n= with syncope of accented /ə/ metri causa.
59
3.1.13 Possessor associated with a direct object (theme)
Carling (2006) points out that PCs used as a possessor may represent both alienable and inalien-
able possessor. In the cases of inalienable possession, the possessum is typically a body-part term
or an abstract concept.
In (2.48), repeated here as (3.43), the first-person singular PC -ñ represents the inalienable pos-
sessor of kektseñ ‘body’, which is the theme of kärśye- ‘(if they) chop up X’. In (3.44), the transitive
verb epsā- ‘(ignorance) covered X’ takes as its DO (theme) aś-äṃ ‘two eyes’. This verb hosts a
third-person singular PC, which represents the inalienable possessor of the DO.
(3.43) [TB] (= 2.48) -ñ = Possessor of a theme (body-part term)

spaitu ra waltsa= ñy ¦ āsta lykaśke po wnolmi •
dust like crush.subj.act.3pl gen.1pl bone.pl small all living.being.nom.pl
kärśye-ñ kektseñ wat ¦ kwä – [b5] ///

cut.subj.act.3pl-1sg body.acc.sg or
‘Even if all beings crush my bones fine like to dust, [5a] or if they chop up my body …’
(B220b4; verse; [5¦8]×4 + [8¦8¦5]; trans. by CEToM)
(3.44) [TA] -ñ = Possessor of a theme (body-part term)

kle – – ñi trik pältsäk ¦ epsā-ñ aś-äṃ
gen.1sg be.confused.pst.act.3sg mind cover.pst.act.3sg-1sg eye-du
ākntsune :
ignorance
‘… has confused my mind. Ignorance has covered my eyes. [67a] ’
(A221a1; verse; [7¦7]×4)
PCs may also represent the inalienable possessor of an abstract concept. In the following ex-
ample (3.45), we find the transitive verb tsämsen- ‘(they) grow X, (they) increase X’, which takes
maiyya ‘power’ as its DO (theme). This verb hosts a third-person singular PC, which represents
the possessor of the theme. Example (3.46) contains the second-person singular PC -śi, which
refers to bodhisatvāp śäṃ ‘the wife of the Bodhisattva’. This PC represents the possessor of tosäm
krant pñintu ‘these good virtues’, which is the DO (theme) of wināsamäś ‘we praise X’.
60
(3.45) [TB] -ne = Possessor of a theme (abstract concept)
tane ksa ṣemi ¦ onolmi nraiy-ne cme-tsi-śc :
here indf one.nom.pl being.pl hell-loc be.born-inf-all
yamanträ yāmor ¦ kraup(a)ntär ṣpä po-ykne-sa :

do.subj.mid.3pl deed gather.subj.mid.3pl conj all-manner-perl
cey cew yā[b6](mor-sa) ¦ mā parskaṃ mā

dem.m.nom.pl dem.m.acc.sg deed-perl neg be.afraid.npst.act.3pl neg
ykāṃṣñenträ :
feel.disgust.npst.mid.3pl
mā kwipeññenträ ¦ mā onmi(ṃ yamaske)n(trä) 8

neg be.ashamed.npst.mid.3pl neg remorse do.npst.mid.3pl
kātkeṃ plontonträ ¦ ṣpä-kka maiyya

be.glad.npst.act.3pl rejoice.npst.mid.3pl conj-emp power
tsämsen-ne :
grow.caus.npst.act.3pl-3sg
‘[If] here some beings do a deed leading to rebirth in hell, [8a] (and) accumulate [it] in
every manner, [8b] they will not be afraid of this deed, [and] they will not be disgusted, [8c]
they [b6] will not be ashamed, they will not show remorse. [8d] They are glad, they rejoice,
[and] even more, they increase its power [i.e., of the bad deed]. [9a] ’
(PKAS7Bb6; verse; [5¦7]×4)
(3.46) [TA] -śi = Possessor of a theme (abstract concept)

tu wäṣpā krant pñi yāmte paṃ ¦ kus ne wsā-ci
nom.2sg truly good virtue do.pst.mid.2sg ptcl rel.nom comp give.pst.act.3sg-2sg
caṃ kāpñe :
dem.m.acc.sg beloved
wināsamäś-śi ¦ tosäm krant pñi-ntu ¦ wināsam-śi [b7]

praise.npst.act.1pl-2sg dem.f.acc.pl good virtue-pl praise.npst.act.1sg-2sg
(caṃ kra)nt kāpñe ¦ can-äk tuṅk was kälpīmträ :
dem.m.acc.sg good beloved dem.m.acc.sg-emp love nom.1pl obtain.opt.mid.1pl
‘”You have truly realized good merit which has given this lover to you. We praise
these good virtues of yours. I praise that good lover of yours. May we obtain that love!”
(A253b6; trans. based on Thomas 1957: 188; verse; [5¦5¦8¦7]×4)
In contrast to the PCs representing the possessor of a body-part term or an abstract concept,
those representing a kinship relation (e.g., ‘my father’, ‘your brother’ etc.) are very rare. In our
61
corpus, there is only one example (3.47), in which the third-person singular PC -äṃ seems to
represent a kinship relation to mācar ‘mother.’
(3.47) [TA] -äṃ = Kinship relation to mācar ‘mother’
ṣñi pat «lya»lyp-äntw-= ālū pat skeys-ā tm-an-äk ///

own conj deed-pl-perl other.gen.pl conj effort.pl-perl dem-loc-emp
[a2] /// (śa)lp-māṃ mācr-äṣ tāṣ :

be.released-ptcp mother-abl cop.subj.act.3sg
wiyoss oki cam klop-yo ¦ mācar nuṃ nuṃ

be.frightened.ptcp.f.nom.sg like dem.m.acc.sg suffering-ins mother again again
trekaṣ-äṃ :
be.confused.subj.act.3sg-3sg
‘By own deeds or by the others’ efforts, there … … will be released from the mother.
His/Her mother will be confused by this suffering again [and] again, as if (she were)
frightened.’
(A152a2; verse; [7¦7]×4)
The PCs representing the possessor of a (non-body-part) concrete object are also very limited.
Example (3.48) shows that a PC represents the possessor of wsāṣinäṃ kukäl ‘golden chariot,’ which
is neither a body-part term nor an abstract concept.
62
(3.48) [TA] -äṃ = Possessor of a theme (concrete object)
Context: Śodhana saw the Bodhisattva Maitreya going out of the city of Ketumatī with
wagons decorated with jewels and speaks to Upaśodhana:
[b1] (kaṣ-swāñcenās-yo) ¦ worpu puk yärśār ¦ sumanāṃ-śäl āsān-ā

fathom-ray.pl-ins surround.ptcp.m.nom.sg all around flower-com seat-perl
lmo ¦ rohiṇiṃ-śäl maññ oki :
be.seated.ptcp.m.nom.sg rohiṇī-com moon like
cindāmaṇi-ṣiṃ ¦ wtsi lap-ā sparcwṣ-äṃ ¦

cintāmaṇi-adjz.m.acc.sg umbrella head-perl turn.npst.act.3sg-3sg
yetwe-yntw-āśśi yetwe ṣñikek ¦ (ṣa[b2]ṣärku ṣñi
decoration-pl-gen decoration nevertheless surpass.ptcp.m.nom.sg own
ye)twesyo :
decoration-ins
ājānay yukañ ¦ wsā-ṣinäṃ kukäl ¦ ylaṅkann oki lyāk

of.noble.birth horse.nom.pl gold-adj.m.acc.sg chariot in.the.air like appearing
ākeñc-äṃ ¦ prutko =ki säs
lead.npst.act.3pl-3sg be.filled.ptcp.m.nom.sg like dem.m.nom.sg
wsā-ṣī ṣont :
gold-adjz.m.nom.sg street
yetuñcäs kuklas ¦ yᵤkass oṅkälmās-yo ¦ [b3]

adorn.ptcp.m.acc.pl chariot.acc.pl horse.acc.pl elephant.pl-ins
(bodhisatv-ā)p warts-yo pälkets ¦ triskäṣ rape
Bodhisattva-gen companion-ins beautiful resound.caus.npst.act.3sg music
swiñc pyāppyāñ :1
rain.npst.act.3pl flower.nom.pl
‘[He is] surrounded by the light of one fathom all around, [he is] sitting on the seat with
Sumanā, as if the moon (were sitting with) Rohiṇī. [1a] (He) turns the umbrella of cintā-
maṇi (jewels) over his head. (He) has truly surpassed the ornament of ornaments with
(his) own ornament. [1b] Horses of noble breed drive his golden chariot as if in the air.
The golden street is filled up, as it were, [1c] by adorned chariots, horses and elephants,
by the companions of the bodhisattva. Beautiful music resounds (and) flowers rain. [1d] ’
(A253b1-3; trans. based on CEToM; verse [5¦5¦8¦7]×4)
The referent of -äṃ is metrak ‘Maitreya’. The function of the PC is to represent the alienable
possessor of wsāṣinäṃ kukäl ‘a golden chariot,’ which is the DO (theme) of ākeñc- ‘(they) drive X’.
63
3.1.14 Possessor associated with an indirect object
PCs may represent the possessor associated with an indirect object. For example, (3.49) has an
intransitive verb (kälyi)tär- ‘X stood’ accompanied by kektsent-sa ‘on the body’ as an IO (location).
This verb hosts a third-person singular PC referring to devadatte ‘Devadatta,’ who represents the
possessor.
(3.49) [TB] -ne = Possessor of a location

oṅkolm= eñcwa-ñña ¦ waltsanoy-n= āsta lykaśke :
she-elephant iron-adjz.f.nom.sg crush.act.impf.3sg-3sg bone small.nom.sg
[b5] ṣale säl(pa)mo ¦ (kälyi)tär-ne kektsent-sa : 70-3

mountain blazing.nom.sg stand.mid.impf.3sg-3sg body-perl
‘An iron she-elephant crushed his bones [to] small [bits]. [73c] A glowing-hot mountain
(stood) on his body. [73d] ’
(B22b4; trans. by CEToM; verse; [5¦7]×4)
Likewise, in (3.50) from TA, we find klwawatr-äṃ ‘X falls,’ whose third-person singular PC repre-
sents the possessor of the IO (location) pakwāśayaṃ ‘in the stomach’ (cf. Skt. pakvāśaya- ‘lit. the
receptacle of digested food; the lower part of the digestive tract’).
(3.50) [TA] -äṃ = Possessor of a location

(ku)[b5]pre-ne āmiśāy-aṃ ṣtmo tāṣ wär-yo want-yo
if-comp āmāśaya-loc stand.ptcp.m.nom.sg cop.subj.act.3sg water-ins wind-ins
wipo pāpey· /// [b6] ṣi nu mā
be.wet.ptcp.m.nom.sg blow.ptcp.m.nom.sg conj neg
päknäṣtr-äṃ māmak-äk pakwāśay-aṃ klawatr-äṃ
ripen.npst.mid.3sg-3sg raw-emp pakvāśaya-loc fall.npst.mid.3sg-3sg
(?)
‘If it [= the food] is located in the āmāśaya [lit. the receptacle of raw food;
the upper part of the digestive tract], moistened by water and blown by
wind … [b6] (the food? ) is not digested by it [lit. ‘cooked by it’], (but) falls
into his pakvāśaya [the lower part of the digestive tract], still being raw.’
(A124b6; prose?)
64
3.1.15 Possessor associated with an argument of an adverb/postposition and a verb
The following two examples show that a PC may function as the possessor associated with an
argument of a complex predicate consisting of an adverb/postposition and a verb.
(3.51) [TB] -ne = Possessor of the argument of epiṅkte + √klayā-

[a3]nolma pätsilpar-ñ läklenta-meṃ ///
be.free.caus.imp.mid.2sg-1sg suffering-abl
/// (kla)yā-ne paiyn= epiṅkte carka wekä ///

fall.pst.act.3sg-3sg foot.du between emit.pst.act.3sg voice
“Make me free from suffering! … fell between his/her feet, uttered voice …’
In (3.51), the third-person singular PC -ne represents the possessor of paiyn(e) ‘two feet,’ which is
the argument of the complex predicate consisting of epiṅkte ‘between’ and √klayā- ‘to fall.’
(3.52) [TA] -äṃ (3sg) = Possessor of the argument of ane + √i-|kälk-

metrak träṅkäṣ ṣokyo prākär sām mäśkit /// [b4] /// näk
Maitreya speak.npst.act.3sg very strong dem.f.nom.sg princess
puttiśpar-ṣi āk(āl pä)lsk-a(ṃ) kᵤpār ane yiṣ-äṃ ||
buddhahood-adjz wish mind-loc deeply inside go.npst.act.3sg-3sg
‘Maitreya speaks: “The princess (is) very strong … the wish for buddhahood goes deep
inside his mind.’
(A399b4; prose?)14
Example (3.52) contains an intransitive verb yiṣ- ‘X goes’, which accompanies an adverb ane ‘in-
side’, forming a transitive predicate. The argument of this predicate is (pä)lska(ṃ) ‘mind’, and the
third-person singular PC -äṃ represents its possessor.
3.1.16 Possessor associated with a subject
PCs may represent the possessor semantically associated with a subject of an intransitive verb,
as the following examples (3.53) and (3.54) show.
14. āk(āl pä)lska(ṃ) restored by Knoll (1996: 142).
65
(3.53) [TB] -ne = Possessor of an intransitive subject
katkomñai-sa arañce pluṣā-ne ram wināṣṣa-me (takarṣkñe-sa)
joy-perl heart float.pst.act.3sg-3sg as praise.pst.act.3sg-pl faith-perl
‘As his heart floated with joy, [he] honored them (with faith).’ (B375b4-5)
(3.54) [TA] -ñi = Possessor of an intransitive subject

rake cam-äṣ pānäsmār ¦ nmäs-māṃ kapśñ-o śl= āñcālyi :
word dem-abl ask.npst.mid.1sg bow-ptcp body-ins with hands.put.together
mā nu ākāl knäṣtär-ñi ¦
neg conj wish be.fulfilled.npst.mid.3sg-1sg
[The king Brahmadatta speaks:] “… I ask a word from him, bowing with my body [and]
with hands folded. [1c] But my wish is not fulfilled. …”
(A71a2; trans. by CEToM; verse; [7¦7]×4)
In (3.53), the third-person singular PC -ne represents the inalienable possessor of arañce ‘heart’.
This noun is the subject of the intransitive verb pluṣā ‘X floated’. In example (3.54), the first-
person singular PC -ñi, referring to brahmadatte ‘Brahmadatta (king of Jambudvīpa)’, represents
the inalienable possessor of ākāl ‘wish’, which is the subject of knäṣtär ‘X is fulfilled’.
3.1.17 Possessor associated with a noun connected by a copula
When a copula connects two nominal expressions, pronominal clitics may represent the pos-
sessor of one of the nominal expressions. For example, -me in (3.55) represents the possessor
semantically associated with ekñi ‘possession’, which is connected with srūkalñe ‘death’ by the
copula star-.
(3.55) [TB] -me = Possessor of an argument of a copula

māwk soycer piś-cmel-ṣana ¦ läkle-nta /// /// srūkalñe ¦
neg be.satisfied.npst.act.2pl five-birth-adjz.f.pl suffering-pl death
ekñi star-me :
possession cop.npst.3sg-pl
‘Are you not saturated by the sufferings of the five births, (you monks)? ... “Death is our
(only certain) possession”.’
(B12b4; trans. based on CEToM; verse; [7¦4]×4)
Likewise, -ñi in (9.46) represents an inalienable relation with se ‘son,’ connected with ārānt by the
66
copula tākiṣ-.
(3.56) [TA] -ñi = Possessor of an argument of a copula

[a6] /// (lakṣa)ṇäs-yo yetu ārānt se tākiṣ-ñi •
mark.pl-ins decorate.ptcp.m.nom.sg Arhat son cop.opt.act.3sg-1sg
‘May my son be the Arhat decorated with … marks!’
(A118a6; prose?)
3.1.18 Subject of a non-finite verb
3.1.18.1 Subject of an infinitive
There is no example in which a PC represents the subject of a finite verb (Carling 2006). A pronom-
inal subject of a finite verb is either omitted (e.g., 3.57) or expressed by an independent personal
or demonstrative pronoun (e.g., 3.58 and 3.59, respectively).
(3.57) [TB] Pronominal subject = non-overt pronominal expression (pro)

teṃ epiṅkte-ne sāu u oṅko«rño» päs pyautka •
dem.n between-loc dem.f.nom.sg porridge away come.into.being.pst.act.3sg
spharīrä-ṣṣe aise-meṃ mu[a4]tkāre-ne
crystal-adjz pot-abl pour.out.pst.act.3pl-3sg
‘In the meantime, the rice porridge became ready. They (= Nānda and Nandābala)
poured it out of the crystal bowl.
(B107a3-4; prose)
(3.58) [TB] Pronominal subject = independent personal pronoun wes

kauś maitam lyakām ¦ mokoṃ protär wes ¦ śle aklaṣlyeṃ po watesa ¦
above go.pst.act.1pl see.pst.act.1pl old brother 1pl with disciple all again
osta[a4](-meṃ ltuweṣ poyśiṃ-ś)
house-abl go.ptcp.m.acc.sg all-knowing-all
“We went up [and] saw again [our] old brother (= Urubilvākāśyapa) with all [his] disci-
ple(s), who had left the house for the omniscient one.”
(B108a3-4; verse; [5¦5¦8¦7]×4)
67
(3.59) [TB] Pronominal subject = independent demonstrative pronoun cey
asān-meṃ tetkāk ṣ [b5] ¦ (n)ek(s)ate kälymiṃ ¦ läkāṣyeṃ
seat-abl immediately conj disappear.pst.mid.3sg direction see.impf.act.3pl
cey kompirko-meṃ ¦ ipprer-ne ka ṣ lyakār-ne
dem.m.nom.pl east-abl sky-loc foc conj see.pst.act.3pl-3sg
‘And (the Buddhaj ) immediately disappeared from the seat. They (= 1,003 braided monks)
looked [in every] direction. And they saw himj in the sky in the east.’
(B108b4-5; verse; [5¦5¦8¦7]×4)
However, if the verb is non-finite (i.e., an infinitive, gerundive, or participle), PCs may represent
the subject (with obligatory clitic climbing in TB).
(3.60) [TB] -me = Subject of an infinitive

amplākätte mā rittetär-me o[a3](sta-meṃ lan-tsi)
without.permission neg be.suitable.npst.mid.3sg-pl house-abl go.out-inf
‘It is not suitable for youpl to become a monk (lit. go out from a house) without permis-
sion.’
(B108a2)
(3.61) [TA] -ñi = Subject of an infinitive

kᵤpre(-ne nu caṣ-äk c)m(o)l-aṃ nätswa-tsi klintar-ñi e(l ///
if-comp conj dem.m.acc.sg-emp birth-loc starve-inf be.obliged.subj.mid.3sg gift
‘… even if it is necessary for me to starve (to death) in this birth, a gift (?) …’
(A343a4; trans. based on CEToM; verse?)
the plural PC -me represents the subject of the infinitive lantsi ‘to go out’ in (3.60). In (3.61) from
TA, the first-person singular PC -ñi, referring to wäl ‘the king,’ represents the subject of the in-
finitive nätswatsi ‘to starve’.
3.1.18.2 Agent of a gerundive
In example (3.62), the second-person singular PC -c represents the agent of sportole, which is a
gerundive built on the intransitive present-stem sportto- (cf. √spārtt(ā)- ‘turn; behave, be’). This
PC undergoes climbing and attaches to the finite copula.
68
(3.62) [TB] -c = Agent of a gerundive
tane ñake g· – – – – – – – [a2] eneśle pañikte käṣṣi-nta-ṃts yakne-ne
here now like Buddha.lord teacher-pl-gen manner-loc
watk(ä)ṣ-älyñe-ne spo(rto-)le star-c
order-nmlz-loc turn.npst-gdv.m.nom cop.npst-2sg
‘Here now, like (the sand in the river Gaṅgā), you should behave following the way (and)
command of the Buddha-teachers.’
(THT1106a2)
A PC also represents the agent of a gerundive in example (3.63).
(3.63) [TA] -ñi = Agent of a gerundive

[a1] /// ṣ träṅkäṣ penäs kraś mänt
speak.npst.act.3sg speak.imp.act.2pl good.nom.pl how
ya-l-ñi
do.npst-gdv.nom.sg-1sg
(King Brahmadatta) speaks: “Tell (me), good ones, how I should act.”
(A71a2; trans. based on CEToM; verse; [7¦7]×4)
The first-person singular PC -ñi in this example refers to brahmadatte ‘Brahmadatta (King of Jam-
budvīpa).’ A finite copula is missing in the subordinate clause introduced by penäs ‘Tell!’, and the
PC is hosted by a gerundive (yal- ‘X is to be done; (one) should do X’).
3.1.18.3 Agent of a preterite participle
PCs may represent the agent of a preterite participle. They always climb to the finite copula in TB
(e.g., 5.67). The preterite participle itself may host a PC in TA when it lacks a copula. Otherwise,
the PC also climbs to the finite copula in TA (e.g., 3.65).
(3.64) [TB] -me = Agent of a preterite participle

tane ṣemi ksa onolmi yāmor yāmoṣ
here one.m.nom.pl indf living.being.nom.pl deed do.ptcp.m.nom.pl
nraiy-ne cmelye-sa ka(krau)pau ṣpä tākan-me

hell-loc of.birth-perl assemble.ptcp.m.nom.sg and cop.subj.3sg-pl
‘[There are] some beings here who have done a deed, and by being reborn in hell it will
be further accumulated by them.’ (PKAS7Ca4-5; verse; [5¦7]×4)
69
(3.65) [TA] -äm = Agent of a preterite participle
ñy ākāl kaknu tāṣ-äm pūrpāc
gen.1sg wish come.about.ptcp.m.nom.sg-pl cop.subj.act.3sg-pl receive.imp.mid.2pl
śāsaṃ senik – – – nervān-aṃ : 3 ||
teaching care Nirvāṇa-loc
‘My wish should be fulfilled by youpl . Accept the teaching (and) care … to the Nirvāṇa!’
(A332a1; verse)15
Rizzi (1982) has observed that in Italian, clitic climbing is not possible when an infinitival clause
undergoes fronting (Cinque 2004). It is an open question whether Tocharian PCs may climb to
the matrix finite verb when an infinitive or a participle is fronted.
3.1.19 Agent of a mediopassive
Some transitive verbs, which usually take active endings, may take middle endings to show pas-
sive interpretation. We call such verbs “mediopassive”. The Tocharian PCs may represent the
agent in these mediopassive verbs, although such uses are rare and the examples are rather lim-
ited.
In (3.66), the first-person singular PC -ñ represents the agent of the mediopassive yāmṣate- ‘(X)
was made (into Y) (by Z),’ which has a middle inflectional ending and a passive interpretation.
15. The PC in this example seems to represent the agent of the preterite participle kaknu ‘fulfilled, come about,’
although Malzahn (2010: 569) lists kaknu as a preterite participle built on an intransitive root.
70
(3.66) [TB] -ñ = Agent of a mediopassive
karuṇ(ä-)ṣ(ṣe) warkṣält-sa ¦ rī pälsko-ṣṣai yū[k](āwa :)
compassion-adjz power-perl city mind-adjz.f.acc.sg conquer.pst.act.1sg
akālkä-ṣṣe retke no ¦ were [a2] te ramt yāmṣate-ñ (1)

wish-adjz army conj odor dem.n like make.pst.mid.3sg-1sg
[King Araṇemi speaking to King Candramukha:] “By the power of compassion, I con-
quered the city of thought. [1c] And the army of desire was made into so much as odor by
me. [1d] ”
(B94a2; trans. based on Schmidt 1974: 263–4;16 verse; [7¦7]×4; cf. PKNS 36 and 20,
IOLToch 69)
The corpus does not have any example from TA in which a PC unambiguously represents the
agent of a mediopassive verb. In A124 b6 (3.50), repeated here as (3.67), a PC seems to represent
the agent of a mediopassive verb, though alternative interpretations are possible (Schmidt 1974:
266).
(3.67) [TA] (= 3.50) -äṃ = Agent of a mediopassive?

(ku)[b5]pre-ne āmiśāy-aṃ ṣtmo tāṣ wär-yo want-yo
if-comp āmāśaya-loc stand.ptcp.m.nom.sg cop.subj.act.3sg water-ins wind-ins
wipo pāpey· /// [b6] ṣi nu mā
be.wet.ptcp.m.nom.sg blow.ptcp.m.nom.sg conj neg
päknäṣtr-äṃ māmak-äk pakwāśay-aṃ klawatr-äṃ
ripen.npst.mid.3sg-3sg raw-emp pakvāśaya-loc(?) fall.npst.mid.3sg-3sg
‘If [it = the food] is located in āmāśaya [lit. the receptacle of raw food; the upper part of
the digestive tract], moistened by water and blown by wind … [the food? ] is not digested
by it [lit. ‘cooked by it’], [but] falls in his pakvāśaya [lit. the receptacle of digested food;
the lower part of the digestive tract], still being raw.’
(A124b6; prose?)
This PC, referring to āmāśaya ‘the upper stomach,’ appears to represent either the agent of the
mediopassive verb (‘… is not digested by it’) or, less likely, the location of the verb (‘… does not
mature in it’). Alternatively, one could take the PC as referring to the same individual as -äṃ in
16. Schmidt (1974: 263–4): “Mit der Kraft des Mitleide habe ich die Festung des Denkens bezwungen. Die Macht
des Wunsches aber ist so etwa von mir zum [blossen] Geruch gemacht worden.”
71
klawatr-äṃ, which represents the possessor of pakwāśayaṃ ‘in the lower stomach.’ In this case,
one could interpret the PC to serve as the beneficiary of the verb (i.e., ‘… does not mature for
him’).
3.1.20 Causee
Tocharian has both analytic and synthetic causative constructions. The former uses the verb ‘to
give’ (TB √ai-|wä(s)? - and TA √e-|wä(s)? -) and an infinitive, meaning ‘X makes/lets Y do Z’ (lit. ‘X
gives Y to do Z’; Adams 2015: 111). For the latter, we find suffixation in non-past stems and suffix-
ation or ablaut in preterite (and imperative) stems. PCs may represent the causee of a synthetic
causative verb built on an intransitive or a transitive base.
In example (3.68), the plural PC -me (pl) represents the causee of the causative verb plyatstsar-
‘make X go out!’ and tsalpäṣṣar- ‘make X be free!’, built on an intransitive √länt- ‘go out, emerge’
and √tsälp(ā)- ‘pass away, be released, be redeemed’, respectively. In example (3.69) from TA, the
third-person singular PC -äṃ represents the causee of lyalymāt- ‘made X sit down,’ built on an
intransitive root √ṣäm-|läm(ā)- ‘sit’.
(3.68) [TB] -me = Causee

(ost-meṃ) plyatstsar-me ¦ tsalpäṣṣar-me (lkle-meṃ :)
house-abl go.out.caus.imp.mid.2sg-pl be.free.caus.imp.mid.2sg-pl suffering-abl
[Nadīkāśyapa and Gayākāśyapa with their 500 students speaking to the Buddha:] “Make
us go out (from the house)! Make us be free (from suffering)! [1c] ”
(B108a9; verse; [7¦7] + [5¦6]×3)
(3.69) [TA] -äṃ = Causee

/// (kä)lpāt-äṃ pkäṃntäk āsān-ā lyalymāt-äṃ ––––
obtain.pst.mid.3sg-3sg separately seat-perl be.seated.caus.pst.mid.3sg-3sg
‘… seized him (and) made him sit on the seat separately.’
(A110a2; verse?)
The following examples show a PC representing the causee of a synthetic causative verb built on
a transitive base.
72
(3.70) [TB] -ne = Causee
/// (de)vadatte procer ṣai ñi ceᵤ preke kᵤse
Devadatta brother cop.impf.act.3sg gen.1sg dem.m.acc.sg time rel.nom
su myārsā-ne ette lyowwa ///
dem.m.nom.sg forget.caus.pst.act.3sg-3sg down
‘Devadatta was my brother at that time, [and] caused him to forget ...’
(Or8212.163b3; trans. by CEToM)
(3.71) [TA] -ṃ = Causee

tmäṣ säm mahā /// [a2] /// propmahur ca(c)äl pärwatāp
then dem.m.nom.sg crown lift.up.pst.act.3sg oldest.gen.sg
se-yo lap-ā casäs lāntune kakälypā-ṃ
son-ins head-perl put.pst.act.3sg lordship obtain.caus.act.3sg-3sg
‘Then, he … picked up the crown (and) put (it) on the head of the oldest son (and) caused
him to obtain the lordship.’
(A130a2; prose?)
Example (3.70) contains the preterite II active third-person singular myārsā- ‘(he) caused X to
forget Y’, which is built on a transitive root √märsā- ‘to forget X’. The third-person singular PC
-ne in this example represents the causee of this causative verb. In (3.71) from TA, the third-
person singular PC -ṃ represents the causee of kakälypā- ‘(he) caused X to gain the lordship; (he)
bestowed the lordship on X’, built on a transitive root √kälpā- ‘to obtain X’.
3.1.21 Doubling
Tocharian PCs sometimes appear to be redundant. For example, the first-person singular PC -ñ
in (3.72) seems to mark the possessor of yakt-āñm ‘feebleness’, even though the possessor itself is
represented by the independent personal pronoun ñi. Likewise, in (3.73), the plural PC -äm seems
to represent the theme of malkam- ‘I will join X,’ although the theme of this verb is cesmäk āyäntu
‘these bones’.
73
(3.72) [TB] Doubling of a nominal expression by a PC
pūdñäktä-ññe pelai[b3](kne ¦ /// :)
buddha-adjz law
ceu -sa ñiś ñke meṅki-tse ¦ te-sa pkārsa-ñ

dem.m.sg-perl 1sg now lack-adjz.m.nom.sg dem.n.sg-perl know.imp.act.2sg-1sg
yakt-āñm ñi :1
feebleness 1sg.gen
“The law of the Buddha … [1c] I lack it now. Because of this, understand my feeble state!
[1d] ”
(B99b3; verse; [7¦7]×4)
(3.73) [TA] Doubling of a nominal expression by a PC

wät träṅkäṣ näṣ nu ce(smä)[b6]-k āy-äntu pᵤk-ā-k
second speak.npst.act.3sg nom.1sg conj dem.m.acc.pl-emp bone-pl all-perl-emp
pusk-ās-yo kaśal malkam-äm
sinew-pl-ins together put.together.subj.act.1sg-pl
‘The second artisan says: “But I will join the bones completely with the sinews.”’
We will discuss this phenomenon in detail in Chapter 5. In (3.73), it may appear that the PC is just
a marker of object-agreement (Peyrot 2017, 2019 and Adams 2015: 149). However, a pleonastic
pronominal clitic may represent the possessor of a theme as in (3.72), and in such cases, it does
not seem to be an object agreement marker because if it were we would expect to find a marker
of the third-person singular, agreeing with the object yakt-āñm. Furthermore, we find doubling
of a possessor associated with an intransitive subject. Again, this would be unexpected if the PC
were a mere object-agreement marker.
74
3.2 Summary
The previous section reviewed the various uses of pronominal clitics in TA and TB. It has shown
that PCs may represent the following categories (table 3.1).
Function Section
1 Theme §3.1.1
2 Addressee §3.1.2
3 Recipient §3.1.3
4 Goal §3.1.4
5 Beneficiary §3.1.5
6 Source §3.1.6
7 Location §3.1.7
8 Experiencer §3.1.8
9 Stimulus §3.1.9
10 Argument of a postposition/adverb and a verb §3.1.10
11 Experiencer of an adjective and a verb §3.1.11
12 Oblique possessor §3.1.12
13 Possessor associated with a direct object §3.1.13
14 Possessor associated with an indirect object §3.1.14
15 Possessor associated with an argument of an adverb/postposition and a verb §3.1.15
16 Possessor associated with a subject §3.1.16
17 Possessor associated with a noun phrase connected by a copula §3.1.17
18 Subject/Agent of a non-finite verb §3.1.18
19 Agent of a mediopassive verb §3.1.19
20 Causee §3.1.20
21 Doubling §3.1.21
Table 3.1: Summary of the chief uses of the Tocharian PCs
A successful theory should be able to derive all of the uses in table 3.1 and none of the unattested
75
uses. In the following chapter I will proceed to build a syntactic model of the Tocharian PCs which
accounts for the various uses observed and also some of the gaps.
3.3 Conclusion
This chapter summarized the phonological, morphological, and syntactic characteristics of pronom-
inal clitics in Tocharian A and B. After reviewing the characteristics of Tocharian PCs in Chapter
2, we proceeded to briefly overview the chief uses of the Tocharian PCs with a brief comment.
Section 3.2 provides a summary.
Some of the gaps will be the subject of Chapters 5, 6, and 7, where I will show that all attested
possible instances in Table 3.1 share independently motivated structural properties. No exam-
ples are attested that do not share these properties. In order to address this, I will detail the
theoretical framework and the (morpho)syntactic analysis of Tocharian which will allow me to
investigate and test specific hypotheses on the corpus of TA and TB.
Previous treatments of the Tocharian PCs have assumed that PCs may represent whatever the
genitive-dative and accusative independent forms may represent (see, e.g., Schulze, Sieg, and
Siegling 1931: 166, TEB I: 162–3 Pinault 1992: 113; 2008: 537, Carling 2006: 44). In contrast, the
model I develop in the following chapters will show that the distribution of PCs in TB and TA is
more restricted. However, to do so requires some discussion of the details, and before going into
the details I discuss some theoretical premises in the following chapter.
76
CHAPTER 4
Theoretical premises
4.1 Introduction
The previous chapter reviewed the representative uses of the pronominal clitics (PCs) of Tochar-
ian A and B (TA and TB). This chapter outlines theoretical premises based on which I will develop
a morphosyntactic model of the PCs in Chapter 5. The model will predict that the distribution of
the PCs is more restricted than the descriptive generalizations made in previous treatments, ac-
cording to which they may represent whatever the genitive-dative and accusative independent
forms may describe. Specifically, the model will predict that the Tocharian PCs cannot express
the possessor associated with the subject of a transitive verb or the complement of the postpo-
sition contained in a nominal expression.
The general idea is that a particular hierarchical relation must hold between the position where
the PC is licensed and the position where it originates. The position is not the subject but the
object domain below the external argument such as an agent or a causer. In the following, I adopt
the terms and concepts widely used in current syntactic research. However, the findings and
predictions I will discuss in the following chapters do not hinge on whether I adopt a more recent
or traditional framework and its technical implementation. This chapter will outline the more
recent syntactic framework mainly for two reasons. One is that it is less confusing to those who
do not specialize in syntax if a theory is solely based on the current minimalism rather than the
traditional X-bar theoretic framework with some new ideas introduced only in part. The other
is that those who specialize in the syntax of contemporary languages may easily understand the
contents of the chapters.
77
The internal organization of this chapter is as follows. The following section (§4.2) introduces
the assumptions regarding syntactic computation. Section 4.3 discusses how morphosyntactic
feature bundles receive a phonological exponent post-syntactically. Section 4.4 outlines the an-
tisymmetric approach to the linearization of a syntactic object. I briefly discuss the structure of
pronominal clitics, the syntactic process Agree, and its consequences in Section 4.5. Section 4.6
shows the sample derivation of a canonical SVO structure in Tocharian B. Section 4.7 provides a
summary.
4.2 Merge and Transfer
It is widely accepted that sentences we observe have an underlying hierarchical structure built
in the (core) syntax. The syntax manipulates syntactic objects (sets of morphosyntactic feature
bundles) in a bottom-up fashion by the process Merge. Merge combines two syntactic objects, X
and Y, and creates a new object K, which is an unordered set (K = {X , Y }). In this case, K contains
X and Y, and X and Y are merge-mates or sisters. K also has a label equal to one of its members
(X or Y). The set created by Merge may be represented as a tree, as shown in (4.1). As K is an
unordered set, the ordering between X and Y in (4.1) is irrelevant.
(4.1) K= or
X Y Y X
A newly created object may undergo Merge with another syntactic object, and as a result, we
obtain a new set containing another set. For example, (4.2) shows that Z and K (4.1) have un-
dergone Merge. Again, the ordering between Z and K and X and Y are irrelevant. We define A to
c-command B if B is a merge-mate of A or if B is contained in the merge-mate of A . For example,
in the hierarchical syntactic object (4.2), Z c-commands its merge-mate K. It also c-commands X
and Y that K contains. Also, K c-commands its merge-mate Z. Likewise, X c-commands Y, and vice
versa. X does not c-command Z, and in this case, Z asymmetrically c-commands X. X does not
asymmetrically c-command Y because the c-command relation goes in both directions.
78
(4.2) L=
Z K
X Y
Merge has two types. Selectional features such as [=Y] drive External Merge (EM), which takes
two syntactic objects X and Y in the workspace and merges them to form the new set K = {X , Y }.
The notation “[=Y]” means that a given syntactic object selects another syntactic object whose
category is Y. This feature is satisfied (marked by “✓”) when it merges with Y.1 In contrast,
formal features such as [uwh] drive Internal Merge (IM; previously called movement), which
takes the syntactic object K 1 and its term YK that has a corresponding [iwh] feature, and merges
them to obtain the new set K 2 (= {K 1 , YK }). The examples include traditional wh-movement where
the C(omplementizer) causes a wh-phrase to move to its specifier.2 In a tree representation, YK
appears as if it moved from one place to another. In this chapter, I will mark the syntactic object
that has undergone IM in gray.
(4.3) External Merge

EM(X[=Y] , Y) =
X Y
✓[=Y]
(4.4) Internal Merge
1. My assumption departs from that of Chomsky, Gallego, and Ott (2019), who consider Merge a free process and
not triggered by a feature.
2. In this example, “i” and “u” represent interpretable and uninterpretable features, respectively. I will come
back to this difference in Section 4.5.
79
IM( K1 , YK ) = K2
X … YK K1
[uwh] [iwh]
… YK X …
[iwh] ✓[uwh]
… YK
[iwh]
When syntactic manipulation reaches a certain point, i.e., when a derivational phase is complete,
the syntax maps the subpart of a syntactic object onto semantic and phonetic representations
(SEM and PHON). This process is called Transfer (Chomsky, Gallego, and Ott 2019), and once the
subpart undergoes Transfer, no further syntactic computation may modify its internal structure
(Phase Impenetrability Condition; Chomsky 2001b, 2007, 2008). Phases derive restrictions
on the extraction traditionally explained under the hood of islands (Ross 1967) or subjacency
(Chomsky 1973). We call the subpart that undergoes Transfer transfer domain. The syntactic
object L is a transfer domain when it merges with the phase-defining head Hm . L, the complement
of Hm , undergoes Transfer when another phase-defining head Hn merges with L, and the Hm phase
is complete.3
(4.5) L undergoes Transfer when Hn is merged (Phase-defining head: Hm and Hn )
3. Alternatively, Bošković (2016) argues that when the Hk phase is complete both the transfer domain Lk and the
phase-defining head Hk undergo Transfer.
80
Hn …
Hm L
…X…Y…
Transfer domain of Hm
Transfer
L ======⇒ … X … Y …
When the transfer domain Lk undergoes Transfer, it projects a prosodic unit (π) in PHON. The
prosodic unit π may be smaller or larger than a phonological word, and its prosodic shape may
alter when another transfer domain Lj , which contains Lk , undergoes Transfer (cf. Sande and
Jenks 2018; Sande, Jenks, and Inkelas 2020).
I follow the standard assumption that phase-defining heads consist of C(omplementizer), Voice
that introduces an external argument, D(eterminer) (Chomsky 2000, 2001b), and category-defining
heads that merge with a root and define its category (Marantz 2007; Embick and Marantz 2008;
Embick 2010). For example, “a” and “n” merge with a root to determine the category of the root
(as an adjective and a noun, respectively). The standard literature on Distributed Morphology
(DM; Halle and Marantz 1993; Embick 2010, 2015) represents category-defining heads with lower
case characters (e.g., “n”, “v”). This chapter, however, uses upper case characters (e.g., “N” for a
noun-defining head and “V” for a verb-defining head) for expository clarity. As shown in (4.6),
when C merges, VP, the transfer-domain of Voice, undergoes Transfer.
(4.6) VP undergoes Transfer when C is merged (phase-defining heads: Voice and C)
81
C
Voice VP
Transfer domain of Voice
If the VP contains a nominal expression, it will undergo Transfer together with it, and the trans-
ferred nominal will be inaccessible to further syntactic operations. To be accessible to the subse-
quent syntactic operations, it must “move” outside the VP. In contrast, the argument introduced
by Voice is outside the transfer domain and visible to the C.
4.3 Distributed Morphology and Vocabulary Insertion
Following Distributed Morphology (DM; Halle and Marantz 1993; Embick 2010, 2015), I assume
that each set of morphosyntactic feature bundles receives a phonological realization (or exponent)
when they undergo Transfer (Vocabulary Insertion; VI). I follow the standard model of VI, in
which vocabulary items compete for insertion, and the most specified rule applies first, blocking
the less specified ones (the Elsewhere Principle [or Pāṇini’s Principle]; cf. Anderson 1969;
Kiparsky 1973). For example, there are three exponents for the English present tense copula
forms: am, are, and is. Among these three, am and is are more restricted, found only in the context
of the first- and third-person singular. In contrast, are is less restricted, found in the second-
person singular, and all persons in the plural. In other words, it is the elsewhere exponent that
may target all person and number combinations in the present tense copula. This rule, however,
82
does not apply to the first- and third-person singular since the more specific vocabulary insertion
rules (4.8i) take precedence over (4.8ii).
(4.7) English present tense copula

sg pl
1 am are
2 are are
3 is are
(4.8) Vocabulary insertion rules
i. T[present, first-person, singular] ↔ am

T[present, third-person, singular] ↔ is
ii. T[present]↔ are
VI targets terminal nodes and supplies a phonological exponent to the feature bundles. The
working hypothesis that I follow is that VI first targets the most embedded terminal and goes
inside out (Embick 2015). For example, in (4.9), VI first applies to the most deeply embedded
terminal A, then B, and finally C.
(4.9) Vocabulary insertion of a complex head
B C
A B
Step 1: A[ … ] ↔ …
Step 2: B[ … ] ↔ …
Step 3: C[ … ] ↔ …
83
4.4 Linearization and antisymmetry
Syntax generates an unordered set and its members may be another unordered set of objects.
When a syntactic object undergoes Transfer, it is mapped onto PHON, and its hierarchical struc-
ture is flattened into a linear string so that it may be uttered or signed. In other words, the sym-
metric relationship (i.e., ‘X and Y are members of K’) is converted into the asymmetric precedence
relationship (i.e., ‘X precedes Y’). How a language linearizes a hierarchical syntactic object is one
of the fields that scholars actively investigate.
Traditionally, syntax directly mapped syntactic structures onto a surface word order, and the
word-order variation was handled by the headedness parameter, whose value is set during ac-
quisition. Setting this parameter on, the syntax of a target language consistently generates a hi-
erarchical structure in which a head always follows a complement, resulting in a Japanese-type
head-final language. In contrast, if this parameter is set off, the syntax generates a head-initial
structure, resulting in an English-type head-initial language. Although this approach accounts
for some morphosyntactic properties that frequently cluster (e.g., a verb following a direct ob-
ject, postpositions rather than prepositions, a possessor preceding a possessum, and so on), it
also predicts many unattested patterns in human languages (Kayne 1994). For example, it would
predict the existence of a language in which a finite verb moves to the second-to-last position in
a matrix clause (“reverse-German”), which is unattested (Kayne 1994: 50).
Therefore, instead of allowing the parametric variation, Kayne (1994) proposes that the asymmet-
ric c-command relation strictly determines the linear precedence relation. His idea is that infor-
mally speaking, if X asymmetrically c-commands Y, X’s terminal precedes Y’s terminal. Kayne
(1994: 6) defines the axiom that generates the linear order of a syntactic object (the Linear Corre-
spondence Axiom; LCA) as (4.10), where T is the set of terminals, d (A) is the set of terminals that
A dominates, and A is all pairs of non-terminal X and Y such that X asymmetrically c-commands
Y.4
4. His definition of c-command is different from ours in that it refers to categories: “X c-commands Y iff X and Y
are categories and X excludes Y and every category that dominates X dominates Y.” (Kayne 1994: 9)
84
(4.10) Linear Correspondence Axiom (Kayne 1994: 6)
d (A) is a linear ordering of T .
(4.11) K
J L
j M N
m P
In example (4.11) from Kayne (1994: 7), there are three terminals: j, m, and p (hence T = { j, m, p }).
There are also six non-terminals: K, J, L, M, N, and P. Among these non-terminals, J asymmetrically
c-commands M, N, and P (Section 4.2). Also, M asymmetrically c-commands P. Therefore, in this
example, A contains four ordered pairs: A = <J, M>, <J, N>, <J, P>, and <M, P>. The non-terminals J,
M, N, and P all dominate just one terminal. Since J dominates j and M dominates m, from <J, M>,
we obtain <j, m> meaning ‘j precedes m.’ Likewise, we derive <m, p> from <M, P> and <j, p> from
<J, N> and <J, P>. In this way, d (A), the set of terminals dominated by A , contains three ordered
pairs of terminals: d (A) = <j, m>, <j, p>, and <m, p>. It means ‘j precedes m, j precedes p, and m
precedes p.’ that is, ‘j precedes m, which in turn precedes p.’ This is the linear order of T.
This approach is advantageous as it may derive the traditional X-bar framework without stipu-
lating it (but cf. Abels and Neeleman 2012). Also, it is restrictive and reduces the four parametric
variants (4.12) to just one type: Spec(ifier)-[Head-Comp(lement)] (4.12i). Therefore, it severely
limits the space of hypotheses that are considered when a child acquires a language, and it is
preferable from the perspective of language acquisition.
(4.12) Parametric variation of the X-bar structure
85
i. ii. iii. iv.
spec spec spec spec
head comp comp head head comp comp head
Furthermore, it also derives some typological connections. Since (4.12i) is the only possible struc-
ture, head-final languages result from consistent Internal Merge (movement) of their comple-
ment, moving it to the specifier/adjunct position so that the complement may asymmetrically
c-command the head.5 It accounts for the typological observation that SOV languages generally
lack wh-movement (Greenberg 1963; Bach 1971).6 While wh-phrases target the specifier of some
functional head, the specifier position must be occupied by the complement of the head.
In addition, Cinque (2005) shows that by positing the fixed hierarchical order of a demonstrative,
numeral, adjective, and noun, one may derive all of the attested orders of the four categories and
none of the unattested patterns.7 Carstens (2002) argues for the universal antisymmetric head-
initial structure by pointing out that the serial verb constructions of OV languages do not display
the mirror image of those of VO languages. Also, Takano (2003) points out that the antisymmetric
approach correctly predicts the ungrammaticality of the heavy NP shift constructions whose
licensor in a heavy NP fails to license a negative polarity item to its left.
Although Kayne (1994) considered the Linear Correspondence Axiom to apply in all stages of
derivation, the precedence relation is a pure PHON property and does not seem to contribute
to any SEM property. Therefore, subsequent treatments exclude the LCA from the core syntax
and assume that syntax generates a symmetric syntactic object and that the LCA applies when a
syntactic object is linearized (Chomsky 1995: ch. 4; Uriagereka 1999; Richards 2008).
Furthermore, the Government and Binding theory of syntax moved from the template-based and
5. Technically speaking, the antisymmetric approach to syntax does not distinguish a specifier from an adjunct.
6. Greenberg’s (1963) Universal 12 states “[i]f a language has dominant order VSO in declarative sentences, it
always puts interrogative words or phrases first in interrogative word questions; if it has dominant order SOV in
declarative sentences, there is never such an invariant rule.”
7. Alternatively, Abels and Neeleman (2012) analyze the data with a less restrictive theory.
86
stipulative X-bar theory to Bare Phrase Structure (Chomsky 1995) which lacks the distinction be-
tween a maximal and minimal projection. Under the Bare Phrase Structure, maximal and min-
imal projections are determined contextually: a minimal category X (represented as Xmin ) may
merge with another minimal category Y when Y no longer projects. In such cases, Y is both min-
imal and maximal (represented as ‘Y min/max ’ as in [4.13]). The LCA fails to linearize these objects
since Xmin and Y min/max c-command each other.
(4.13) Xmin and Y min/max c-command each other
Zmax
Zmin Xmax
Xmin Y min/max
To salvage this problem, Guimarães (2000), followed bv Kayne (2008), proposes self-merge, which
only takes one syntactic object and creates a singleton set. Moro (2000) suggests that the Internal
Merge of Y min/max is triggered to break the symmetry. Richards (2008) reintroduces the headed-
ness parameter, which this time operates in PHON and determines the linear ordering between
X and Y that c-command each other. I adopt his approach here and assume that when two termi-
nals c-command each other, the headedness parameter orders the head to follow the non-head
in Tocharian.
4.5 Structure of a pronominal clitic, Agree, and defective goals
Tocharian PCs are special clitics linked to the pronominal reference of arguments. There is a de-
pendency between a clitic and a pronominal DP in that (1) special clitics cannot occur in positions
where full DPs appear (Zwicky 1977), and (2) often but not always, they cannot co-occur with full
DPs. Following I. G. Roberts (2010), I consider the Tocharian PCs as realizing syntactic objects
that lack an internal structure, consisting of person and number features (“φ-features”) only.
They correspond to Cardinaletti and Starke’s (1999) clitic pronouns and Déchaine and Wiltschko’s
(2002) pro-φ(P)s. I represent it as φmin/max , that is, it is a minimal category but at the same time
87
it is also maximal since it does not project its morphosyntactic or semantic features to the set in
which it is contained. It also lacks a (syntactic) Case-feature, and adjoins to a K(ase)P, to which a
syntactic Case is assigned (Nevins 2011a).8 The PC’s person and number features are shared with
the K’s sister DP, and we call this DP the PC’s associate. PCs always refer to the same individual
as their associate. The structure of the nominal expression that contains a PC is shown in (4.14).
Doubling is absent when the KP lacks φmin/max .
(4.14) Structure of a nominal expression containing a PC
KP
φmin/max KP
[ iφ: 3sg ] K(ase) DPj
associate
[ iφ: 3sg ]
I adopt the standard Probe-Goal model of Agree (Chomsky 2000, 2001a, 2001b). In this model,
features may be interpretable or uninterpretable. In order for syntactic derivation to be successful,
all uninterpretable features must be checked or removed from an object so that everything that is
present at the interfaces may be interpretable to the interfaces (Principle of Full Interpretation).
According to Chomsky (2001b: 5), “Prior to application of Agree, these (= uninterpretable fea-
tures, TO) are distinguished from interpretable features by lack of specification of value. After ap-
plication of Agree, the distinction is lost.” In this model, the functional heads T(ense) and Voice
have uninterpretable (and unvalued) person and number features (φ-features; represented as
uφ), whose values need to be filled. They serve as a probe that searches for valued interpretable
φ-features in their c-commanding domain (i.e., in their merge-mate) in a top-down manner. If
a syntactic object contains multiple nominal expressions with valued interpretable φ-features
(represented as iφ), the one that is closest to the probe serves as a goal (Minimal Link Con-
dition; Chomsky 1995: 311), and the probe copies the values of the goal’s φ-features (Agree).9
9. Chomsky (2001b) assumes that a set of interpretable φ-features may serve as a goal if it is active. In other words,
a goal must have an unvalued (abstract) structural Case-feature to undergo Agree. We depart from this assumption
88
(4.15) Agree
i.
…
X
probe …
YP
[uφ: ]
goal
copy [iφ: 3sg ]
ii.
…
X
probe …
YP
[uφ: 3sg ]
goal
[iφ: 3sg ]
Agree may create a complex head X that consists of X and Y if a functional head X agrees with
Y and if Y is non-distinct in features from X (I. G. Roberts 2010: 62). In other words, whenever a
goal Y is “defective” in the sense defined in (4.16), it undergoes incorporation to its probe X.
(4.16) Defective goal (I. G. Roberts 2010: 62)

A goal G is defective iff G ’s formal features are a proper subset of those of G ’s Probe P .
(4.17) and (4.18) illustrate this process. The unvalued φ-features of Voice look for a goal in its
c-commanding domain, find φmin/max ’s φ-features, and copy the values. Since φmin/max ’s formal
features, namely φ-features, are a proper subset of those of Voice, φmin/max is considered as the
defective goal to the probe Voice. Therefore, it undergoes incorporation to Voice, forming a
complex head consisting of Voice and φmin/max (4.18).
(4.17) Agree
since φmin/max lacks a Case feature (Section 4.5).
89
…
Voice
[uφ: ] … φmin/max
[iφ: 3sg ]
copy
(4.18) Incorporation
…
Voice
…
φmin/max Voice φmin/max
[iφ: 3sg ] [uφ: 3sg ] [iφ: 3sg ]
incorporation
This view amounts to claim that whenever φmin/max , spelled out as a PC, serves as a goal in Agree,
it undergoes incorporation. This is a desirable result because Tocharian PCs are special clitics
(in Zwicky’s 1977 sense), and they never occur in positions where full nominal phrases usually
appear. Apart from some examples that lack a finite verb in TA (discussed in Chapter 2), PCs
consistently appear immediately after a finite verb. This analysis also has another welcoming
aspect in that it dispenses with a morphological operation that creates a complex head post-
syntactically (e.g., m-merger; Matushansky 2006; Harizanov 2014 among others).
I. G. Roberts (2010) argues that head-movement happens in this mechanism: The head H j adjoins
to another head Hk forming the complex head { Hk { H j Hk } } if and only if H j is a defective goal
to Hk . For example, Roberts assigns to the tense head T following features (excluding selectional
features): uEPP, uV, iT, and uφ. He also assigns iV, and uφ to Voice.10 T’s uV feature is checked
when it undergoes Agree with Voice. Here, Voice’s formal features (i.e., V and φ) are a proper
subset of those of T. Therefore, Voice is a defective goal with respect to T, and Voice undergoes
incorporation to T.
10. To be precise, I. G. Roberts (2010) also assigns an uninterpretable T feature on Voice (his v*). This account
predicts that Voice consistently incorporates to T since it is a defective goal to T (see Chapter 5 for cases in which
90
(4.19) Agree
…
T
[ uEPP, iT, uV, uφ: ] …
Voice …
agr
ee
[ iV, iφ: 3sg ]
(4.20) Incorporation
…
T
…
Voice T Voice
[ iV, iφ: 3sg ] [ uEPP, iT, ✓uV, uφ: ] [ iV, iφ: 3sg ]
incorporation
4.6 Example of derivation of a canonical SOV structure
To illustrate the derivation of a canonical Tocharian B SOV structure, let us consider the passage
from TB (4.21) as an example. In this example, the transitive verb lkāṣṣäṃ takes toṃ läklenta ‘these
sufferings’ as an internal argument. The demonstrative pronoun sū ‘he’ is the external ar-
gument of this verb, and it represents the subject of the sentence.
(4.21) wesäñ no perne-sa sū toṃ läkle-nta lkāṣṣäṃ.

gen.1pl conj glory-perl dem.m.nom.sg dem.f.acc.pl suffering-pl see.npst.act.3sg
‘But [it is] for [all] our sakes he endures (lit. sees) such suffering.’
The internal argument of this example is a K(ase)P that consists of a K(ase) head and a DP. The K
verbs are non-finite).
91
head has the selectional feature [=D] which triggers External Merge with the DP.11 The structure
of the internal argument is shown in (4.22).
(4.22) Syntactic structure of (4.21; to be continued)
KP
K DP
✓[=D]
[iφ: 3pl ]
toṃ läklenta
these pains
This KP is selected by the category-defining head V, also called the verbalizer. This verbalizer has
two selectional features: [=root] and [=K]. It first undergoes External Merge with a root to satisfy
the [=root] feature.12 This verbalizer is a phase-defining head that defines its merge-mate as a
transfer domain. In this case, the transfer domain of the verbalizer is the root.
V root
✓[=root] √läk(ā)-
[=K] see
-ṣṣä-
The resulting syntactic object (4.23) then undergoes External Merge with the internal argument
(4.22) to satisfy the verbalizer’s [=K] feature. Contra Marantz (1997) and Harley (2009, 2014),
I analyze that a theme KP is selected by a category-defining head, rather than a root (Alexi-
adou 2014; Cuervo 2014; Lohndal 2014; Ahn 2016; Merchant 2019). Merchant (2019) points out
category-defining heads may show an idiosyncratic selectional behavior (e.g., priden [pp in X ];
prouda [pp of X ]; pridev oneself [pp on X ]), which should be impossible if a root, which does not
have any categorial feature, selected a PP complement. De Belder and Craenenbroeck (2015) also
11. This functional head also checks a syntactic Case and licenses its merge-mate DP. The detail is omitted here for
the sake of expository simplicity.
12. De Belder and Craenenbroeck (2015) argue that roots do not have any categorial feature. In this case, a category
defining head X merges with a syntactic object Y , which is an empty set (unary merge; Zwart 2009, 2011).
92
argue that roots lack not only categories but also grammatical features (cf. Borer 2005a, 2005b).
The verbalizer also assigns the theme role to the KP. The resulting object is (4.24).
KPtheme
K DP V root
✓[=D] ✓[=root] √läk(ā)-
[iφ: 3pl ]
✓[=K] see
toṃ läklenta
-ṣṣä-
these pains
Subsequently, the functional head Voice externally merges with (4.24), satisfying its [=V] feature.
(4.25) Syntactic structure of (4.21; continued)
Voice VP
✓[=V] KPtheme
[=K]
K DP V root
[iV]
✓[=D] ✓[=root] √läk(ā)-
[iEPP] [iφ: 3pl ]
✓[=K] see
[uφ: ] toṃ läklenta
-ṣṣä-
these pains
Voice is a phase-defining head and triggers Transfer of the transfer-domain of the lower-phase
defining head. Therefore, the verbalizer’s transfer domain (i.e., root) undergoes Transfer, and it
receives the phonological exponent /ləka-/.
Transfer Root
(4.26) root ======⇒
/ləka-/
The functional head Voice has unvalued person and number features. To fill the values, it looks
for valued person and number features in its c-commanding domain (VP in [4.27]). It finds the
value “third-person plural” in the theme KP, and copies the value.
93
Agr
ee
Voice VP
✓[=V] KPtheme
[=K]
[iφ: 3pl ] V root
[iV]
toṃ läklenta ✓[=root] √läk(ā)-
[iEPP]
these pains ✓[=K] see
✓[uφ: 3pl ]
-ṣṣä-
In order to satisfy the Voice’s selectional feature ([=K]), it then undergoes External Merge with
the external argument KP (sū ‘he’) that functions as the subject of the sentence. Voice assigns
an experiencer role to this KP.
KPexperiencer
[iφ: 3sg ] Voice VP
sū ✓[=V] KPtheme
he ✓[=K]
[iφ: 3pl ] V root
[iV]
[iEPP]
✓[uφ: 3pl ]
-ṣṣä-
Subsequently, another functional head T merges with (4.28), satisfying its [=Voice] feature. It also
has unvalued person and number features. It therefore finds the value “third-person singular”
in the external argument and copies it. When it is transferred later, it surfaces as the subject
agreement marker of the third-person singular (-ṃ).
94
Agr
ee
T
✓[=Voice]
KPexperiencer
[uV]
[uEPP]
[uφ: ] sū ✓[=V] KPtheme
he ✓[=K]
[iφ: 3pl ] V root
[iV]
[iEPP]
✓[uφ: 3pl ]
-ṣṣä-
Furthermore, T agrees with Voice to satisfy the uninterpretable V feature, and since the latter’s
formal features (i.e., V, EPP, φ) are a subset of those of the former, Voice undergoes incorporation
(head-movement) to T.
T VoiceP
Voice T KPexperiencer
✓[=V] ✓[=Voice]
✓[=K] ✓[uV]
sū ✓[=V] KPtheme
[iV] [uEPP]
he ✓[=K]
[iEPP] ✓[uφ: 3sg ] [iφ: 3pl ] V root
[iV]
✓[uφ: 3pl ] toṃ läklenta ✓[=root] √läk(ā)-
[iEPP]
Agree these pains ✓[=K] see
✓[uφ: 3pl ]
-ṣṣä-
Incorpor
a tion
Moreover, T’s uninterpretable EPP feature triggers Internal Merge of VoiceP. As a result, we ob-
tain (4.31).
95
VoiceP
KPexperiencer T VoiceP
[iφ: 3sg ] Voice VP Voice T KPexperiencer
sū ✓[=V] ✓[=V] ✓[=Voice]

KPtheme [iφ: 3sg ] Voice VP
he ✓[=K] ✓[=K] ✓[uV]
root sū ✓[=V] KPtheme
[iV] [iφ: 3pl ] V ✓[uEPP]
[iV]
he ✓[=K]
[iEPP] toṃ läklenta ✓[=root] √läk(ā)- ✓[uφ: 3sg ] root
[iEPP] [iφ: 3pl ] V
see [iV]
these pains ✓[=K]
✓[uφ: 3pl ] ✓[uφ: 3pl ] toṃ läklenta -ṣṣä- √läk(ā)-
[iEPP]
-ṣṣä- see
these pains
✓[uφ: 3pl ]
Subsequently, another functional head C(omplementizer) merges with (4.31). Since C is a phase-
defining head, the lower phase-defining head Voice’s transfer domain undergoes Transfer. This
domain (VP) contains the theme KP, the verbalizer, and the root. Since this KP asymmetrically
c-commands the verbalizer and the root, it linearly precedes them. The verbalizer and the root
c-command each other, so the LCA cannot linearize them. We assume that the headedness pa-
rameter handles such cases, placing a category-defining head after a root (Section 4.4).
C TP
✓[=T] VoiceP
∅
KPexperiencer T VoiceP
[iφ: 3sg ] Voice VP Voice T KPexperiencer
sū ✓[=V] ✓[=V] ✓[=Voice]

KPtheme [iφ: 3sg ] Voice VP
he ✓[=K] ✓[=K] ✓[uV]
root sū ✓[=V] KPtheme
[iV] [iφ: 3pl ] V ✓[uEPP]
[iV]
he ✓[=K]
[iEPP] toṃ läklenta ✓[=root] √läk(ā)- ✓[uφ: 3sg ] root
[iEPP] [iφ: 3pl ] V
see [iV]
these pains ✓[=K]
✓[uφ: 3pl ] ✓[uφ: 3pl ] toṃ läklenta -ṣṣä- √läk(ā)-
[iEPP]
-ṣṣä- see
these pains
✓[uφ: 3pl ]
KPtheme Root V
Transfer
VP ======⇒ toṃ läklenta /ləka- + -ṣṣə-/
lkāṣṣä-
When the CP phase is complete, C’s transfer domain (TP) undergoes Transfer. Since the exter-
nal argument (sū ‘he’) asymmetrically c-commands the VP, sū linearizes preceding the VP. The
functional heads T and Voice form a complex head. Vocabulary Insertion targets both terminals,
96
inserting /-∅/ to T and /-n/ to Voice. Alternatively, we could think of a post-syntactic opera-
tion that fuses two terminals (T and Voice) into one (fusion; Halle and Marantz 1993, 1994), and
VI targeting this fused terminal. At any rate, we obtain the present stem lkāṣṣä-, preceding the
inflectional ending of the present-stem active third-person singular -ṃ.
(4.33) Vocabulary insertion
i. Voice[act] ↔ /-∅/
ii. T[3sg, npst] ↔ /-n/ /[act]
KPexperiencer KPtheme Root V Voice-T

Transfer
(4.34) TP ======⇒ sū toṃ läklenta /ləka- + -ṣṣə-/ /-∅-n/
lkāṣṣä- -ṃ
A somewhat surprising consequence of this analysis is that a (root +) V and (Voice +) T do not
form a constituent by themselves. When they seemingly move as a unit, they move as a TP, out
of which an agent DP and a theme DP scrambled to a higher projection. For a similar approach
to ours, see Zyman and Kalivoda’s (2020) analysis on Latin verbs.
4.7 Interim summary
This section outlined some theoretical premises for the subsequent analysis. Syntax generates a
syntactic object (Section 4.2), whose terminal receives a phonological exponent post-syntactically
(Section 4.3). A subpart of a syntactic object undergoes Transfer, converting its hierarchical sym-
metric structure into an asymmetric linear string (Section 4.4). The Tocharian PCs are φmin/max ,
consisting of a set of φ-features only (Section 4.5). They start as an adjunct to a KP, and when-
ever they undergo φ-agreement, they incorporate to a probe (I. G. Roberts’ 2010 defective goal),
appearing in unique positions where full nominal phrases do not occur. Section 4.6 showed the
sample derivation of a canonical SVO structure in Tocharian B. I will develop a syntactic analysis
of the Tocharian PCs in the following chapter.
97
CHAPTER 5
Split intransitivity and clitic distribution in Tocharian
5.1 Formal Analysis
This chapter develops a syntactic analysis of the Tocharian pronominal clitics based on the the-
oretical premises discussed in Chapter 4. The syntactic model that I develop in this chapter
accounts for the wide range of thematic roles that the Tocharian PCs may display. Chapters 2
and 3 showed that the Tocharian PCs lack case distinctions. They are compatible with the ac-
cusative and genitive-dative independent forms, representing various thematic roles including
the theme, goal, source, location, experiencer, beneficiary, and possessor. I argue that this mul-
tifunctionality results from the PCs starting as part of a nominal expression to which a thematic
role is assigned, and being attracted by a licensor that looks for the values of person and number
features and copies the highest pronominal DP in the VP region in the clause.
5.1.1 PC representing a direct object (theme) of a transitive verb
The previous chapter showed that a PC may represent the direct object (DO) with a theme role as
in (2.4) and (3.1), repeated here as (5.1).
(5.1) (= 2.4) [TB] -ne (3sg) = DO (theme) of palātai ‘(Yousg ) praised X’

never uphold.moral.behavior-inf forget.pst.2sg

‘Yousg have never forgotten to uphold (moral behavior). Yousg have praised him (= the
Buddha) for seven days with a single strophe.’ (B297a.a4; verse [7¦7¦4]×4)
98
The underlying syntactic object of (5.1) is created in a bottom-up fashion by the repeated ap-
plication of Merge, which combines two morphosyntactic objects into one. Let us examine how
each derivational step works.
In Chapter 4, I assumed that PCs spell out φmin/max , consisting of (interpretable) person and num-
ber features only. In this example, the value of φmin/max is third-person singular, represented as
“[iφ: 3sg ]”. It merges with the K(ase)P that consists of K(ase) and a pronominal DP. This pronom-
inal DP shares the person and number features with φmin/max , referring to the same individual
(5.2). Importantly, the clitic starts out as part of the thematic KP/DP.
KP
φmin/max KP
[iφ: 3sg ] K DP
ne ∅ [iφ: 3sg ]
him pro
(him)
This syntactic object (5.2) serves as the internal argument selected by a verbalizer. In this exam-
ple, the verbalizer has two selectional features: [=root] and [=K]. The verbalizer first undergoes
External Merge with a root to satisfy the [=root] feature. This verbalizer is a phase-defining
head, which defines its merge-mate as the transfer domain (§4.2).1 In this example, it defines
the root (√pälā- ‘praise’) as its transfer domain.
V root
✓[=root] √pälā-
[=K] praise
1. Squared categories represent phase-defining heads in the following tree representations.
99
The resulting syntactic object (5.3) then undergoes External Merge with the internal argument
(5.2) to satisfy the verbalizer’s selectional feature [=K]. The verbalizer assigns a theme role to the
internal argument KP. The resulting syntactic object is shown in (5.4).
KPtheme
φmin/max V root
[iφ: 3sg ] K DP ✓[=root] √pälā-

ne ✓[=K] praise
∅ [iφ: 3sg ]
him pro ∅
(him)
Subsequently, the syntactic object (5.4) undergoes External Merge with the functional head Voice.
This functional head is a phase-defining head (§4.2), and it defines its merge-mate (i.e., 5.4) as the
transfer domain. It also triggers Transfer of the root, which is the transfer domain of the lower
phase-defining head (5.6). The root receives the phonological exponent (/pəla-/), though its
phonological shape may be altered later when a stem-forming morpheme undergoes Transfer.
Voice VP
✓[=V] KPtheme
[=K]
φmin/max V root
[iV]
[iEPP]
ne ✓[=K] praise
[uφ: ] ∅ [iφ: 3sg ]
him pro ∅
(him)
Transfer Root
(5.6) root ======⇒
/pəla-/
Voice has unvalued φ-features and looks for their value in its merge-mate in a top-down manner.
100
In this example, it looks for valued person and number features in VP, finds φmin/max in KP, and
copies the value “3sg” (agree). Since φmin/max consists of φ-features only, φ’s formal features
are a proper subset of those of Voice (φ, V, and EPP).
Therefore, φmin/max is a defective goal with respect to Voice (§ 4.5), and it undergoes incorpo-
ration, forming the complex head consisting of Voice and φmin/max . As a result, we acquire the
syntactic object (5.7). In other words, it is at this point the separation of the clitic and the KP/DP
takes place.
Voice VP
φmin/max Voice KPtheme
[iφ: 3sg ] ✓[=V] root

φmin/max V
ne [=K]
him [iV] praise
ne ∅ [iφ: 3sg ] ✓[=K]
[iEPP]
him ∅
e
pro
re
✓[uφ: 3sg ]
ag
incor (him)
porat
ion
However, this is an intermediate step in the derivation. The syntactic object (5.7) undergoes Ex-
ternal Merge with an external argument (pro ‘yousg ’) technically driven by the =K selectional
feature. The Voice head also assigns an agent role to this argument. The external argument in-
troduced has valued φ-features (second-person singular in this example). The resulting syntactic
object is shown in (5.8).
101
KPagent
pro KPtheme
φmin/max Voice
(yousg )
φmin/max V
ne ✓[=K]
him [iV] praise
ne ∅ [iφ: 3sg ] ✓[=K]
[iEPP]
him pro ∅
✓[uφ: 3sg ]
(him)
This syntactic object (5.8) undergoes External Merge with another functional head T(ense). T
undergoes Agree with the external argument (pro ‘yousg ’), and when T undergoes Transfer, it
surfaces as a subject-agreement marker.
agr
ee
T
✓[=Voice]
KPagent
[uV]
[iT]
[+EPP] pro KPtheme
φmin/max Voice
[uφ: ] (yousg )
φmin/max V
ne ✓[=K]
him [iV] praise
ne ∅ [iφ: 3sg ] ✓[=K]
[iEPP]
him pro ∅
✓[uφ: 3sg ]
(him)
Furthermore, T undergoes Agree with Voice, triggering incorporation (head-movement) of Voice

to T, creating the complex head consisting of Voice and T. Since Voice is now a complex head
consisting of Voice and φmin/max (5.8), as the result of head-movement, we obtain the complex
head that consists of T, Voice, and φmin/max (5.10).
102
T VoiceP
Voice T KPagent
✓[=Voice]
φmin/max Voice [iφ: 2sg ] Voice VP
✓[uV]
[iφ: 3sg ] ✓[=V] pro KPtheme
[iT] φmin/max Voice
ne ✓[=K] (yousg )
[uEPP] [iφ: 3sg ] ✓[=V] root
φmin/max V
him [iV]
ne ✓[=K]
✓[uφ: 2sg ] [iφ: 3sg ] K DP ✓[=root] √pälā-
[iEPP]
-(a)tai him [iV] praise
ne ∅ [iφ: 3sg ] ✓[=K]
✓[uφ: 3sg ]
[iEPP]
him pro ∅
✓[uφ: 3sg ]
(him)
In addition, T has the feature [uEPP], which triggers Internal Merge of VoiceP. As the result of
InternalMerge([5.10], VoiceP), we obtain the syntactic object (5.11).
VoiceP
KPagent T VoiceP
[iφ: 2sg ] Voice VP Voice T KPagent
pro ✓[=Voice]
φmin/max Voice KPtheme φmin/max Voice [iφ: 2sg ] Voice VP
(yousg ) ✓[uV]
[iφ: 3sg ] ✓[=V] [iφ: 3sg ] ✓[=V] pro
φmin/max V root φmin/max Voice KPtheme
[iT]
✓[=K] ne ✓[=K] (yousg )
[iφ: 3sg ] K DP ✓[=root] √pälā- ✓[uEPP] [iφ: 3sg ] ✓[=V]
root
[iV] φmin/max V
praise him [iV] ✓[=K]
ne ✓[=K] ne
[iEPP]
∅ [iφ: 3sg ] ✓[uφ: 2sg ] [iφ: 3sg ] K DP ✓[=root] √pälā-
[iEPP] [iV]
him pro ∅ -(a)tai praise
✓[uφ: 3sg ] ne ∅ [iφ: 3sg ] ✓[=K]
✓[uφ: 3sg ] [iEPP]
(him) him ∅
pro
✓[uφ: 3sg ]
(him)
Subsequently, the syntactic object created undergoes further External Merge with C(omplemen-
tizer). Since C is another phase-defining head, the transfer domain of a lower phase-defining
head undergoes Transfer. In our example, the lower phase-defining head is Voice, and its trans-
fer domain is VP. This transfer domain contains the theme KP, V, and root. Since this KP’s merge-
mate also contains the verbalizer and the root, it asymmetrically c-commands them. Therefore,
the theme KP linearly precedes the verbalizer and the root. The root and the verbalizer are
merge-mates, which c-command each other. Although the Linear Correspondent Axiom cannot
determine their linear order (Chapter 4), the headedness parameter determines the ordering,
103
the category-defining head V following the root. The root and the verbalizer are realized as the
preterite-stem (/pəlá- + -∅-/ →) pälā- in this example.
Transfer KPtheme Root V

(5.12) VP ======⇒
∅pro /pəla- + -∅-/
...
C TP
✓[=T] VoiceP
∅
KPagent T VoiceP
[iφ: 2sg ] Voice VP Voice T
pro ✓[=Voice]
φmin/max Voice KPtheme φmin/max Voice
(yousg ) ✓[uV]
[iφ: 3sg ] ✓[=V] [iφ: 3sg ] ✓[=V]
φmin/max V root
[iT]
✓[=K] ne ✓[=K]
[iφ: 3sg ] K DP ✓[=root] √pälā- ✓[uEPP]
[iV] him [iV]
ne ✓[=K] praise
∅ [iφ: 3sg ] [uφ: 2sg ]
[iEPP] [iEPP]
him pro ∅ -(a)tai
[uφ: 3sg ] [uφ: 3sg ]
(him)
Finally, when the CP phase is complete, C’s transfer domain (TP) undergoes Transfer. Since the
agent KP (pro ‘yousg ’) asymmetrically c-commands the VP, the LCA places it before the theme
KP (pro ‘him’), the verbalizer, and the root. As the result of incorporation, T, Voice and φ form a
complex head. There are two possible ways in which Vocabulary Insertion applies to this complex
head. One is to analyze that VI targets all of these three terminals (φ, Voice, and T). In this case,
vocabulary items are first inserted to φ, and then to Voice. The linear ordering of φ and Voice
is Voice preceding φ: /∅-ne/ Finally, a vocabulary item is inserted to T. T precedes Voice and φ,
resulting in /-atay-∅-ne/.
(5.14) Post-Transfer operations (1)

[ φmin/max _ Voice ]_ T
VI: φ[3sg] ↔ /-ne/
(1) [ /-ne/_ /-∅/ ]_ T → /-∅-ne/_ T
VI: Voice ↔ /-∅/
(2) VI: T[pst, 2sg] ↔ /-atay/ / [mid] /∅-ne/_ /-atay/ → /-atay-∅-ne/
104
The other approach is to analyze that the complex consisting of φ, Voice and T undergoes re-
bracketing, resulting in the structure [ φ [ Voice T ] ]. After this operation, Voice and T undergo
fusion (Halle and Marantz 1993, 1994) that fuses two sets of morphosyntactic feature bundles
into a single set (Voice-T). Subsequently, Vocabulary insertion applies, inserting the preterite-
stem second-person singular middle ending /-atay/ (→ -(a)tai) to this newly created terminal
(5.16). In both cases, the vocabulary item of a PC is specified as suffixal, and as a result it linearly
follows Voice or Voice-T.
(5.15) Post-Transfer operations (2)

[ φmin/max _ Voice ]_ T
(1) Rebracketing φmin/max _ [ Voice_ T ]
(2) Fusion φmin/max _ Voice-T
VI: φ[3sg] ↔ /-ne/
(3) /-ne/_ /-atay/ → /-atay-ne/
VI: Voice-T[pst, 2sg, mid] ↔ /-atay/
KPagent KPtheme Root V Voice-T φmin/max

Transfer
(5.16) TP ======⇒ ∅pro ∅pro /pəla- + -∅- -atay -ne/
palā- -tai -ne
I have built the underlying syntactic object of a Tocharian sentence through the repeated appli-
cation of Merge: External Merge of the root and the verbalizer (5.3), followed by another External
Merge of the resulting syntactic object with an internal argument (5.4), another External Merge
of the resulting object with the functional head Voice (5.7), and so on. In addition to the repeated
application of External Merge, T’s formal features trigger incorporation (head-movement) of
Voice and Internal Merge of VoiceP (cf. Haegeman 2001 and Koopman and Szabolcsi 2001). The
former creates the complex head, consisting of T and Voice, spelled out as a subject-agreement
marker, while the latter makes VoiceP that contains external and internal arguments, the root,
and the verbalizer asymmetrically c-command T and Voice.
(5.17) Derivational steps
105
Resulting
Step Operation Consequence
structure
i ExternalMerge(V, root) (5.3)
ii ExternalMerge(5.3, 5.2) (5.4)
(1) Transfer of Root
iii ExternalMerge(5.4, Voice) (5.7)
(2) Agree & Incorporation of φ
iv ExternalMerge(5.7, KPagent ) (5.8)
(1) Agree
v ExternalMerge(5.8, T) (5.10)
(2) Agree & Incorporation of Voice
vi InternalMerge(5.10, VoiceP) (5.11)
vii ExternalMerge(5.11, C) Transfer of VP (5.13)
As the result of these operations, we obtain the syntactic object that shows an SOV order when
transferred.
When PCs represent the DO (theme) of a transitive verb, their licensor (Voice) copies the values
of person and number features from φmin/max in the internal argument. In contrast, when PCs
represent the IO, the licensor finds the value of its φ-features in the non-core (i.e., non-internal)
argument. In essence, because of Agree, the PC will correspond to the highest accessible φmin/max .
The underlying syntactic object of such an example and how it is built is shown in the following
subsection.
5.1.2 PC representing an indirect object with various thematic roles
The previous subsection illustrated the derivation of the sentence with a PC representing a DO
(theme). Let us proceed to example (3.3), repeated here as (5.18), in which a PC represents the
indirect object (IO) with an addressee role.
106
(5.18) (= 3.3) TB -me (pl) = addressee of akṣā- ‘(s/he) proclaimed (this strophe) to X’
(ce ślok a)kṣā-me ¦
this.acc.sg strophe proclaim.pst.3sg-pl
ku ce tne wnolmi ¦ yamantär ¦¦ krent yo(laiṃ yāmor) ¦ ///

rel.acc.sg here being.pl do.subj.3pl good.acc.sg bad.acc.sg deed
‘(The Buddha) proclaimed this strophe to them: “Whatever good or evil deed living be-
ings do here …”’
(B21a2; verse; [5¦4¦3]×4)
In the underlying syntactic object of (5.18), there is another functional head that introduces a
non-core argument and assigns an addressee role. The PC represents an IO in this example be-
cause the licensor of a PC (Voice) finds the value of person and number features (φmin/max ) in
this non-core argument. The derivational step proceeds as follows. In the following tree repre-
sentations, I omit selectional and formal features for the sake of expository simplicity.
Firstly, the root merges with the verbalizer. The resulting syntactic object then undergoes Exter-
nal Merge with an internal argument (ce ślok ‘this strophe’), which has valued person and number
features (third-person singular). The verbalizer assigns a theme role to this internal argument.
As a result, we obtain the syntactic object (5.19).
(5.19) Syntactic structure of ce ślok akṣā-me ‘(He) proclaimed this strophe to them’ (to be con-
tinued)
KPtheme
[iφ: 3sg ] V root
ce ślok ∅ √āks-
this strophe proclaim
This time, the resulting structure (5.19) merges with the functional head Appl(icative) (Pylkkä-
nen’s 2008 high applicative).2 This Appl head introduces another argument and assigns a the-
matic role such as goal, source, addressee, experiencer, beneficiary, and location. In this example,
Appladdressee , which merges with the verb of speech, assigns an addressee role, and the argument
107
introduced by Appl contains φmin/max whose value is third-person plural. The resulting syntactic
object is shown in (5.20).
(5.20)
KPaddressee
φmin/max Appladdressee VP
[iφ: 3pl ] K ∅
DP KPtheme
me ∅ [iφ: 3pl ]
[iφ: 3sg ] V Root
to them pro
(to them)
The syntactic object (5.20) then merges with the licensor of a PC (Voice). It looks for valued φ-
features in its merge-mate (i.e., 5.20) in a top-down fashion. Therefore, it first finds the value
“3pl” of φmin/max in the non-core argument. Again, since φmin/max is a defective goal to Voice, it
undergoes incorporation, creating the complex head consisting of Voice and φmin/max (5.21).
2. Pylkkänen (2008) distinguishes two types of applicative: high applicative and low applicative. The high applica-
tive head merges with a verb phrase, introduces a non-core argument, and denotes the thematic relation between an
individual and the Neo-Davidsonian event introduced by a verb (e.g., J HighApplgoal K = λx.λe.goal(e, x)). In contrast,
a nominal expression introduced by the low applicative head does not relate to a neo-Davidsonian event but to the
internal argument, denoting transfer of possession (e.g., J LowApplrecipient K = λy.λx.to-the-possession-of(x, y)).
The high applicative head may, in theory, merge recursively, introducing different thematic roles. One might be
tempted to use these two applicative heads to explain the empirical fact that only ablative and allative case markers
follow a PC: while the allative and ablative case markers realize the low applicative head, other secondary case mark-
ers realize the high applicative head. This approach predicts that the allative and ablative markers do not follow the
IO of an intransitive verb since low applicatives are restricted to transitive verbs (Pylkkänen 2008). However, this
prediction turns out to be false, as we find TA yeṣ-äṃn-anac ‘(he) went to her’ [A222b6], an intransitive verb accom-
panying a PC and an allative marker, suggesting that the allative marker realizes a high applicative head. Since an
IO c-commands a DO in both applicative constructions, whether a non-core argument is introduced by a high or low
applicative head is not crucial for my analysis.
108
(5.21)
Voice ApplP
φmin/max Voice KPaddressee
[iφ: 3pl ] [uφ: 3pl ] Appladdressee VP

φmin/max
me ag ∅
[iφ: 3pl ] K DP KPtheme
re
to them e me ∅ [iφ: 3pl ]
[iφ: 3sg ] V Root
to them pro
incor (to them)
pora this strophe proclaim
tion
The rest of the derivation is the same as we saw in the first example. Voice introduces an external
argument (pro) and assigns an agent role (5.22).
(5.22)
KPagent
[iφ: 3sg ] Voice ApplP

pro KPaddressee
φmin/max Voice
(he)
[iφ: 3pl ] [uφ: 3pl ] φmin/max Appladdressee VP
me [iφ: 3pl ] K DP ∅ KPtheme
to them me ∅ [iφ: 3pl ]
[iφ: 3sg ] V Root
to them pro √āks-
ce ślok ∅
(to them) proclaim
this strophe
Subsequently, T merges with (5.22). T triggers head-movement of Voice (5.23) and Internal Merge
of VoiceP (5.24).
109
(5.23)
T+Voice+φ VoiceP
me
KPagent
to them
pro KPaddressee
φmin/max Voice
(he)
φmin/max
me ∅
to them
me ∅ [iφ: 3pl ]
[iφ: 3sg ] V Root
to them pro
ce ślok ∅ √aks-
(to them)
(5.24)
VoiceP
KPagent T+Voice+φ VoiceP

me
[iφ: 3sg ] Voice ApplP KPagent
to them
pro Voice KPaddressee Voice
φmin/max [iφ: 3sg ] ApplP
(he)
[iφ: 3pl ] [uφ: 3pl ] Appladdressee VP pro Voice KPaddressee
φmin/max φmin/max
(he)
[iφ: 3pl ] K DP
∅
KPtheme [iφ: 3pl ] [uφ: 3pl ] Appladdressee VP
φmin/max
me ∅ [iφ: 3pl ] me ∅
[iφ: 3sg ] Root [iφ: 3pl ] K DP KPtheme
V
to them pro to them
ce ślok √āks- me ∅ [iφ: 3pl ]
∅ V Root
(to them) [iφ: 3sg ]
this strophe proclaim to them pro
ce ślok ∅ √aks-
(to them)
When C merges with a syntactic object, it defines its merge-mate (TP) as the transfer domain
(5.25). Subsequently, when the CP phase is complete, this merge-mate undergoes Transfer.
(5.25)
…
C TP
VoiceP

me
to them
pro Voice KPaddressee
φmin/max
(he)
φmin/max
[iφ: 3pl ] K ∅
DP KPtheme
me ∅ [iφ: 3pl ]
[iφ: 3sg ] V Root
to them pro
(to them)
In this way, φmin/max , which originates in a non-core argument with an addressee role, clusters
110
with Voice-T, and and appears as -me [pl], immediately following the preterite third-person sin-
gular active ending.
KPagent KPaddressee Appl KPtheme Root V Voice-T φ

Transfer
(5.26) TP ======⇒ ∅pro ∅pro -∅ ce ślok /aks- + -∅- -a -me/
akṣ- -ā -me
This model may also derive examples in which a PC, representing an IO, accompanies a secondary
case marker (ablative TB -meṃ, TA -anäṣ or allative TB -ś, TA -anac). For example, the allative
marker -meṃ follows a PC in (2.14), repeated here as (5.27). The PC in this example represents the
source from which Upananda begged a cloak.
(5.27) (= 2.14) TB -ne (3sg) = source of yaṣṣāte- ‘(Upananda) begged (this cloak) from X’
wsā-ne •
‘Upanandai begged this cloak from himj (= an Ājīvika ascetic), (but hej ) did not give (it)
to himi .’ (PKAS18Ab2)
The underlying syntactic object of (5.27) is generated in the same way as (5.18) until the tense
head T is merged. In the previous examples (e.g., 5.25), T’s EPP feature triggers Internal Merge
of VoiceP. In this case, however, I analyze it as causing Internal Merge of VP, rather than VoiceP.
(5.28) The structure of (5.27)
111
T+Voice+φ VoiceP
-āte-ne
KPagent

upanande KPsource
φmin/max Voice
Upananda
[iφ: 3sg ] [uφ: 3sg ] Applsource VP
φmin/max
[iφ: 3sg ] K -meṃ
DP KPtheme
ne ∅ from
[iφ: 3sg ]
[iφ: 3sg ] V Root
him pro
ceu kampāl ∅ √yask-
(an ājīvika ascetic)
this cloak beg
As the result of InternalMerge(5.28, VP), we obtain (5.29).
(5.29) The structure of (5.27)
VP
KPtheme T+Voice+φ VoiceP

-āte-ne
[iφ: 3sg ] V Root KPagent

this cloak beg
upanande KPsource
φmin/max Voice
Upananda
[iφ: 3sg ] [uφ: 3sg ] Applsource VP
φmin/max
[iφ: 3sg ] K -meṃ
DP KPtheme
ne ∅ from
[iφ: 3sg ]
[iφ: 3sg ] V Root
him pro
(an ājīvika ascetic)
this cloak beg
With subsequent Internal Merge of the agent KP (upanande ‘Upananda’) and the source KP (pro ‘an
ājīvika ascetic’), we obtain the syntactic object whose Appl head surfaces as -meṃ, immediately
following a PC (5.27).
KPagent KPsource KPtheme Root V Voice-T φ Applsource

Transfer
(5.30) TP ======⇒ upananda ∅pro ceu kampāl /yask- + -∅- -ate -ne/ /-men/
yaṣṣ- -āte -ne -meṃ
112
5.1.3 PC representing a possessor
So far, I have shown that my model may derive the Tocharian examples whose PC represents a DO
or an IO. However, PCs may also express the possessor semantically associated with a DO or an IO.
The following subsection shows that the syntactic model developed in the previous chapter may
also derive such examples. For example, let us consider (5.31), which displays a PC representing
the inalienable possessor of a DO with a theme role.
(5.31) Function: inalienable possessor of a DO (theme)
ersna lkāskeṃñ-c ¦ pä(l)l(āntär-ci ¦) ///

form.f.pl see.npst.act.3pl-2sg praise.npst.mid.3pl-2sg
‘(They) see your (beautiful) form [lit. forms]. (They) praise you. […]’
(B213b2; verse; [4¦4¦4]×4)
Alexiadou (2003) argues that an alienable and inalienable possession have different underlying
syntactic structures by pointing out three pieces of evidence. One is that inalienable possessors
cannot appear after a copula in Modern Greek (5.32ii), but inalienable ones may (5.33ii).
(5.32) Inalienable possessor cannot appear after a copula in Modern Greek
i. i miti tu Jani
the nose the John.gen
‘John’s nose’ (Alexiadou 2003: 172 ex. 9a)
ii. *i miti ine tu Jani

the nose is the John.gen
(Intended) ‘The nose is John’s.’ (Alexiadou 2003: 172 ex. 10a)
(5.33) Alienable possessor may appear after a copula in Modern Greek
i. to vivlio tu Jani
the book the John.gen
‘John’s book’ (Alexiadou 2003: 172 ex. 9b)
ii. to vivlio ine tu Jani

the book is the John.gen
113
‘The book is John’s.’ (Alexiadou 2003: 172 ex. 10b)
This parallels the difference between the internal and non-internal arguments of a nominal ex-
pression. For example, tu ktiriu ‘of the building’ is the internal argument (theme) of i katastrofi ‘the
destruction’ (5.34i), and it cannot appear after a copula (5.34ii). In contrast, ja to Chomsky ‘about
Chomsky’ is an adjunct of to vivlio ‘the book’ in (5.35i). This adjunct may appear after a copula as
in (5.35ii). This suggests that inalienable possessors are more tightly combined with a possessum
than alienable ones and behave like an internal argument (theme). In contrast, alienable posses-
sors are more loosely associated with a possessum and behave like an external argument or an
adjunct.
(5.34) The internal argument of a nominal expression cannot appear after a copula
i. i katastrofi tu ktiriu
the destruction the building.gen
‘the destruction of the building’
ii. *i katastrofi itan tu ktiriu

the destruction was the building.gen
(Intended) ‘The destruction was the building’s.’ (Alexiadou 2003: 172 ex. 12)
(5.35) The non-internal argument of a nominal expression may appear after a copula
i. to vivlio ja to Chomsky
the book about the Chomsky
‘the book about Chomsky’ (Alexiadou 2003: 172 ex. 11a)
ii. to vivlio itan ja to Chomsky

The book was about the Chomsky
‘The book was about Chomsky.’ (Alexiadou 2003: 172 ex. 11b)
Based on such evidence, she concludes that alienable and inalienable possession have different
syntactic structures. She proposes that a functional head (labeled Poss) selects an alienable pos-
sessor and assigns a possessor role in the former. In contrast, an inalienable possessor is either
the complement of a possessum, or the complement of a functional head encoding the inalien-
able relation interpretation. Since I analyze a possessum (at least) to consist of an acategorical
114
root (representing a possessum) and a category-defining head N, I consider that the category-
defining head N selects an inalienable possessor and assigns inalienable possession interpreta-
tion (e.g., part-whole relation).3 Though it is not always the case that part-whole and kinship
relations, as well as states of mind are classified as belonging to the class of inalienable posses-
sion (Dixon 2010), in the absence of counterarguments, I consider them to belong to the class of
inalienable possession in Tocharian.
The underlying syntactic object of (5.31), repeated here as (5.36), is generated through the fol-
lowing steps.
(5.36) (= 5.31) Function: inalienable possessor of a DO (theme)

‘(They) see your (beautiful) form [lit. forms]. (They) praise you. […]’
(B213b2; verse; [4¦4¦4]×4)
Firstly, the noun-defining head N merges with the root (ersna). The resulting structure under-
goes External Merge with a possessor, and N assigns a part-whole interpretation to it (5.37). The
possessor contains φmin/max whose value is second-person singular.
(5.37) Syntactic structure of (5.31) (to be continued)
KPpossessor
φmin/max N root
[iφ: 2sg ] K DP [iφ: 3pl ] ersna

c ∅ form
∅ [iφ: 2sg ]
yoursg pro
(yoursg )
3. I assume that alienable possessors also move to the edge of K. However, examples of a single PC representing
either kinship or alienable possession are very rare. Why this is so is an open question.
115
One might reasonably wonder whether the root first undergoes External Merge with the inalien-
able possessor, and the resulting structure then merges with the noun-defining head N. We do
not adopt this view since it would need an additional stipulation. If the root first merged with the
inalienable possessor, the possessor would be c-commanded by the noun-defining head N. Since
N is a category-defining head, it defines its merge-mate as the transfer domain (§4.2). When an-
other phase-defining head (D) merges in the course of derivation, the possessor, contained in the
transfer domain of N, would undergo Transfer and further syntactic computation may no longer
manipulate it. Therefore, the possessor would need to undergo Internal Merge to be outside of
the transfer domain so that it may be available for further syntactic operation. This additional
operation is dispensable if we assume that an inalienable possessor starts outside of the trans-
fer domain of N. Moreover, if we consider the selectional feature [=K] to be in the noun-defining
head, then the root is consistent with the other examples we have seen so far in that it does not
have any selectional feature.
Secondly, the functional head D merges with the syntactic object (5.37). This functional head also
triggers Internal Merge of the possessor KP. Since D is a phase-defining head, the transfer domain
of the lower phase-defining head undergoes Transfer (5.39).
(5.38) Syntactic structure of (5.31) (continued)
KPpossessor
φmin/max D NP
[iφ: 2sg ] K DP ∅ KPpossessor

c ∅ [iφ: 2sg ]
φmin/max N root
yoursg pro
[iφ: 2sg ] K DP [iφ: 3pl ] ersna
(yoursg )
c ∅ form
∅ [iφ: 2sg ]
yoursg pro
(yoursg )
Transfer root
(5.39) root ======⇒
ersna
116
Thirdly, another functional head K merges with (5.38). Again, it triggers Internal Merge of the
possessor KP. The resulting structure (5.38) later serves as an internal argument in the derivation.
(5.40) Syntactic structure of (5.31) (continued)
KP
KPpossessor
φmin/max K DP
[iφ: 2sg ] K ∅
DP KPpossessor
c ∅ [iφ: 2sg ]
φmin/max D NP
yoursg pro
[iφ: 2sg ] K DP ∅ KPpossessor
(yoursg )
c ∅ [iφ: 2sg ]
N root
yoursg pro
[iφ: 3pl ] ersna
(yoursg )
∅ form
Fourthly, the verbalizer merges with the root. The resulting syntactic object then undergoes
External Merge with the internal argument (5.38). The verbalizer assigns a theme role to the
internal argument. The resulting syntactic object is shown in (5.41).
(5.41)
KPtheme
KPpossessor V Root
K DP -ske- √läk(ā)-
φmin/max
see
[iφ: 2sg ] K ∅
DP KPpossessor
c ∅ [iφ: 2sg ]
D NP
yoursg pro
∅ KPpossessor
(yoursg )
N root
[iφ: 3pl ] ersna

∅ form
The licensor of a PC (Voice) then merges with the structure (5.41). It has unvalued φ-features
that look for values in its merge-mate (i.e., 5.41) in a top-down manner.
117
(5.42)
Voice VP
[uφ: ] KPtheme
KPpossessor V Root
φmin/max
see
[iφ: 2sg ] K ∅
DP KPpossessor
c ∅ [iφ: 2sg ]
D NP
yoursg pro
∅ ersna
(yoursg )
Transfer NP Root
(5.43) NP, root ======⇒
ersna /ləka-/
The unvalued φ-features on Voice first find φmin/max and copy the value “2sg”. Furthermore,
since φmin/max is a defective goal to Voice, it adjoins to Voice and creates a complex head.
(5.44)
Voice VP
[iφ: 2sg ] [uφ: 2sg ] KPpossessor Root

V
c
φmin/max
ag
yoursg see
∅
ree
[iφ: 2sg ] K DP KPpossessor

c ∅ [iφ: 2sg ]
D NP
inc yoursg pro
orp ∅ ersna
ora (yoursg )
tio
n
The rest of the derivation is the same as we saw in the first and second examples. Voice triggers
External Merge of an external argument (5.45), and assigns an agent role to it.
118
(5.45)
KPagent
[iφ: 3pl ] Voice VP
pro KPtheme
φmin/max Voice
(they)
V
c
φmin/max
yoursg see
[iφ: 2sg ] K ∅ ersna
DP
c ∅ [iφ: 2sg ]
yoursg pro
(yoursg )
The syntactic object (5.45) undergoes External Merge with the tense head T. T undergoes φ-
agreement with the external argument (pro ‘they’). T triggers head-movement of Voice (5.46)
and Internal Merge of VoiceP. The resulting structure is (5.47).
(5.46)
T+Voice+φ VoiceP
-ṃñ-c
KPagent
pro KPtheme
φmin/max Voice
(they)
V
c
φmin/max
see
DP
c ∅ [iφ: 2sg ]
yoursg pro
(yoursg )
119
(5.47)
VoiceP

-ṃñ-c
pro Voice KPtheme

φmin/max
(they)
V
c
φmin/max
see
DP
c ∅ [iφ: 2sg ]
yoursg pro
(yoursg )
When the CP phase is complete (5.48), TP, the transfer domain of C, undergoes Transfer. In this
way, the second-person singular PC -c, representing the inalienable possessor of ersna, appears
not before the possessum, but after the present third-person plural active ending (/-n/ →) ṃñ.
(5.48)
CP
…
C TP
VoiceP

-ṃñ-c
pro Voice KPtheme

φmin/max
(they)
V
c
φmin/max
see
DP
c ∅ [iφ: 2sg ]
yoursg pro
(yoursg )
KPagent [KPtheme KPpossessor DP ] Root V Voice-T φ

Transfer
(5.49) TP ======⇒ ∅pro ∅pro ersna /ləka- + -ske- -n -cə/
lkāske- -ṃñ -c
PCs may also represent the possessor of an IO (3.49, repeated here as 5.50) and a subject. I will
come back to the latter case in the following subsection. In the former case, an underlying syn-
120
tactic object has a structure similar to (5.48).
(5.50) [TB] (= 3.49) -ne = Possessor of a location

oṅkolm= eñcwa-ñña ¦ waltsanoy-n= āsta lykaśke :
she-elephant iron-adjz.f.nom.sg crush.act.impf.3sg-3sg bone small.nom.sg
[b5] ṣale säl(pa)mo ¦ (kälyi)tär-ne kektsent-sa : 70-3

mountain blazing.nom.sg stand.mid.impf.3sg-3sg body-perl
‘An iron she-elephant crushed his bones [to] small [bits]. [73c] A glowing-hot mountain
(sto)od on his body. [73d] ’
The possessor KP that contains φmin/max is selected either by N with an inalienable relation or
by Poss with an alienable possession role, and Voice, the licensor of a PC, copies the person and
number feature of φmin/max . This defective goal (φmin/max ) undergoes incorporation to Voice,
and as a result, a PC appears not before an IO but immediately after a finite verb, semantically
representing the possessor of an IO.
5.1.4 Interim summary and predictions
I have built a model which accounts for the multifunctionality of the Tocharian PCs. They spell
out person and number features (φmin/max ) introduced at various positions to which a thematic
role is assigned. They are licensed by Voice, and when Voice copies the value of φmin/max , the
former incorporates the latter, forming a complex head. Subsequently, the tense head triggers
head-movement of this complex head. As the Voice + T complex is realized as an inflectional
ending, PCs always follow a finite verb. I have also shown that my model may derive examples
in which a secondary case marker follows a PC just by stipulating that VP, rather than VoiceP, is
the target of the Internal Merge that T triggers.
If this analysis is on the right track, we may impose a strict restriction on the distribution of the
Tocharian PCs. Suppose we follow the standard assumption that a probe searches a goal in its
c-commanding domain (i.e., in its merge-mate). In that case, it is inevitable to conclude that a
probe cannot undergo Agree with an external argument because the former does not c-command
121
the latter.4 Therefore, this analysis predicts that Tocharian PCs cannot semantically represent
the possessor associated with an external argument.
(5.51) Prediction:
Pronominal clitics cannot semantically represent the possessor associated with an exter-
nal argument in Tocharian A and B.
I have examined 608 examples of TB and 551 examples of TA that contained a PC. It turned out
that there is no example in which a PC spells out the possessor associated with the subject of
a transitive verb. It is worth highlighting that an independent personal pronoun consistently
represents the pronominal possessor associated with a transitive subject. For example, procer
‘brother’ is the subject of the transitive verb tsāka ‘(s/he) pulled out X’ in (5.52). This transitive
subject carries the independent form of the first-person pronoun ñi ‘my’ as its inalienable pos-
sessor. Likewise, the independent personal pronoun nāñi ‘my’ represents the possessor of klośäṃ
‘two ears,’ which is the subject of the transitive verb klyosnseñc ‘(they) hear X.’
(5.52) [TB] Independent personal pronoun ñi ‘my’ [gen.1sg] represents the (inalienable) pos-
sessor associated with a transitive subject
ente procer ñi tsāka eś-ne wärkṣä(lt-sa) ///
when brother my put.out.pst.3sg eye-du violence-perl
‘When my brother pulled out (my) eyes with violence…’ (Or 8212.163 b1)
(5.53) [TA] Independent personal pronoun nāñi ‘my’ [gen.1sg.f] represents the (inalienable)
possessor associated with a transitive subject
(lke)ñc pe aśäṃ krant wramäṃ ¦ swāräṃ rake
see.npst.act.3pl also eye.du good.pl thing.pl sweet.acc.sg word.sg
klyosnseñc pe ¦ klośäṃ nāñi :
hear.npst.act.3pl also ear.du gen.1sg.f
‘[My] eyes also (se)e the good things, my ears also hear the sweet word.’
(A58b3; trans. by CEToM; verse; [7¦7¦4]×4)
4. Some alternative models enable a probe to undergo φ-agreement with a goal outside the c-commanding do-
main (see, e.g., Bejár and Rezac 2009).
122
Furthermore, my analysis predicts that when the licensor of a PC is looking for valued φ-features,
if φmin/max is contained in the domain that has already undergone Transfer, the licensor should
not be able to find φmin/max . Therefore, φmin/max should never be realized as a PC. More specifi-
cally, Voice may see an internal argument and the non-core argument introduced by Appl, but
not another nominal expression contained within the argument as an adjunct. If this another
nominal expression has a possessor, it must be invisible to the licensor also. Therefore, my anal-
ysis predicts that PCs cannot represent the possessor of the nominal expression contained in
another nominal expression. The possessor would undergo Transfer before the licensor of a PC
(Voice) merges with a syntactic object. The following syntactic structure (5.55) illustrates the
unattested hypothetical nominal expression that contains another nominal expression as an ad-
junct (TA †esā wsālyo … -(ä)ṃ ‘with the coat on his shoulder’ where -(ä)ṃ represents the possessor
of esā ‘on the shoulder’; cf. A184b3 esā wsālyo ‘with the coat on the shoulder’ [5.54]).
(5.54) [TA] Nominal expression esā ‘on the shoulder’ contained in wsālyo ‘with the coat’ as an
adjunct
/// ·k· su oki : es-ā wsāl-yo yä[b4] ///
as shoulder-perl coat-ins
‘… With the coat on the shoulder …’
(A184b3; verse)
(5.55) [TA] Unattested hypothetical example (†esā wsāl … -(ä)ṃ ‘the coat on his shoulder’)
123
DP2
D2 NP2B
NP2A PP
N2 root2 P DP1
wsāl -ā KPpossessor
coat on
φmin/max D1
[iφ: 3sg ] K DP KPpossessor NP1
-(ä)ṃ
[iφ: 3sg ] N1 root1
his
pro possessum
(his) es
shoulder
In this example, DP1 (representing ‘his shoulder’) is contained in DP2 (representing ‘the coat on
his shoulder’), and DP1 contains a possessor KP. Squared categories represent phase-defining
heads that trigger Transfer of the transfer domain of a lower phase-defining head. When D2
merges with NP2B , N2 ’s transfer domain (Root2 ) and P’s transfer domain (DP1 ) undergo Transfer.
Therefore, subsequent syntactic operations may no longer manipulate DP1 ’s internal structure.
Since φmin/max is contained in DP1 , it is invisible to the licensor (Voice) that merges with the
structure later. In this way, my analysis predicts that PCs cannot represent the possessor associ-
ated with the nominal expression contained in another nominal expression as an adjunct.
(5.56) Prediction:
In Tocharian A and B, pronominal clitics cannot semantically represent the possessor as-
sociated with the nominal expression contained in another nominal expression as an ad-
junct.
This prediction is also borne out from our corpus study. It is also consistent with the observation
made by Meunier (2015), who examined the attestations of the first-person singular PC and noted
that there is no example in which a PC is used adnominally. My analysis predicts that PCs cannot
be used adnominally unless they are visible to the licensor (Voice).
Moreover, my analysis predicts that if there is a licensor that c-commands an external argument,
124
the c-commanded external argument may undergo φ-agreement with the licensor and may thus
surface as a PC.
(5.57) Prediction:
If there is a functional head with unvalued φ-features c-commanding an external argu-
ment, the c-commanded external argument may surface as a PC.
This prediction is also borne out as the following three observations show.
(5.58) Observations:
i. Tocharian PCs may represent the causee of a morphological causative (5.59).
ii. Tocharian PCs may represent the agent of a (transitive) preterite participle (5.67).
iii. Tocharian PCs may represent the agent of a (transitive) infinitive (5.75).
As example (5.59) shows, PCs may represent the causee of a causative verb. In this example, the
plural PC -me serves as the causee of lyāmate ‘(s/he) made X be seated,’ which is the morphological
causative based on the TB root √läm(ā)- ‘to sit.’
(5.59) [TB] -me = causee of lyāmate ‘(he) made X be seated’

tane araṇemi walo brāhmaṇe-ṃ wratsai tsäṅkormeṃ
then Aranemi king brahmins-acc in.front.of rise.abs
käṣṣī-ññe yäkne-sa asān-ne lyāmate-me

teacher-adjz way-perl seat-loc be.seated.caus.pst.act.3sg-pl
‘Then, King Araṇemi, having stood up in front of the brahmins, made them sit down on
the throne in the manner of a teacher.’ (B81b6; prose)
Let us look at the derivation of the syntactic object linearized as (5.59). Firstly, the verb-defining
head V merges with the root. The resulting syntactic object then undergoes External Merge with
the locative-marked phrase asān-ne ‘on the throne.’ As a result, we obtain (5.60).
125
(5.60) VP
PP VP
asān-ne V root
on the throne ∅ √läm(ā)-
sit
Secondly, the causative head (represented as vcaus ) merges with the structure, triggering the
Internal Merge of the VP. This functional head also introduces an external argument and assigns
a causee role (5.61).5 The external argument introduced is a KP, containing φmin/max whose value
is third-person plural.
(5.61)
KPcausee
φmin/max VP
[iφ: 3pl ] K DP PP VP vcaus VP

me ∅ [iφ: 3pl ] asān-ne root ∅ PP VP
V
them pro on the throne ∅ √läm(ā)- asān-ne V root
(them) sit on the throne ∅ √läm(ā)-
sit
Thirdly, another functional head (Voice) merges with (5.61). It has unvalued φ-features, and
agrees with φmin/max in the causee KP, and incorporates it (5.62).
(5.62)
Voice
agree
φmin/max Voice KPcausee
[iφ: 3pl ] [uφ: 3pl ] φmin/max VP

me [iφ: 3pl ] K DP PP VP vcaus VP
them me ∅ [iφ: 3pl ] asān-ne root ∅
V
them pro on the throne ∅ √läm(ā)-
(them) sit
incorporation
The Voice head also introduces another external argument (araṇemi walo ‘King Araṇemi’), and
assigns an agent role (5.63).
5. Strictly speaking, the external argument is introduced by a different functional head since the antisymmetric
approach dissolves the distinction between a specifier and an adjunct, and bans multiple specifiers/adjuncts (Kayne
1994).
126
(5.63)
KPagent
[iφ: 3sg ] Voice
araṇemi walo φmin/max Voice KPcausee

King Araṇemi [iφ: 3pl ] [uφ: 3pl ] VP
φmin/max
me [iφ: 3pl ] K DP PP VP vcaus VP
them me ∅ [iφ: 3pl ] asān-ne root ∅
V
(them) sit
The rest of the derivation is the same as we have seen earlier. The tense head T merges with
(5.63). It undergoes φ-agreement with the agent KP, surfacing as the subject-agreement marker
-(a)te. It also triggers the head-movement of the Voice + φ complex to T (5.64).
(5.64)
T VoiceP
Voice T KPagent
[uφ: 3sg ] Voice
φmin/max Voice [iφ: 3sg ]
-(a)te araṇemi walo KPcausee
[iφ: 3pl ] [uφ: 3pl ]
me King Araṇemi VP
φmin/max
them [iφ: 3pl ] K DP PP VP vcaus VP
me ∅ [iφ: 3pl ] asān-ne root ∅

V
(them) sit
The tense head also triggers Internal Merge of VoiceP. As a result, we acquire (5.65).
(5.65)
VoiceP
KPagent T VoiceP
[iφ: 3sg ] Voice Voice T
araṇemi walo KPcausee [uφ: 3sg ]

φmin/max Voice
King Araṇemi VP -(a)te
φmin/max [iφ: 3pl ] [uφ: 3pl ]
[iφ: 3pl ] K DP PP VP vcaus VP me
me ∅ [iφ: 3pl ] asān-ne root ∅ them

V
(them) sit
When this syntactic object undergoes Transfer, V, Root and vcaus surface as the causativized
127
preterite stem (/lyáma-/ →) lyāma- ‘to make X sit down’, and φmin/max , representing the causee
of lyāmate-, appears immediately after the Voice+T complex as the plural PC -me.
(5.66) Syntactic structure of (5.59)

Transfer KPagent … PPlocation Root V vcaus Voice-T φ
TP ======⇒
araṇemi walo asān-ne lyāma- -(a)te -me
In the following example (5.67), the plural PC -me undergoes clitic climbing to the finite auxiliary
tākan ‘(it) will be’ and serves as the agent of the preterite participle kakraupau ‘assembled.’
(5.67) [TB] -me = agent of a preterite participle, hosted by tākan ‘(it) will be’
tane ṣemi ksa onolmi yāmor yāmoṣ
here one.m.nom.pl some living.being.nom.pl deed do.ptcp.m.nom.pl
nraiy-ne cmelye-sa ka(krau)pau ṣpä tākan-me

hell-loc of.birth-perl assemble.ptcp.m.nom.sg conj cop.subj.3sg-pl
‘[There are] some beings who have done a deed, and by being reborn in hell (an evil deed)
will be accumulated by them.’ (PKAS7Ca5; verse [5¦7]×4)
This example suggests the existence of a licensor that surfaces as a finite auxiliary. It starts above
Voice and finds valued person and number features in the external argument. It also triggers
incorporation, resulting in clitic climbing.
The following sequence of processes create the underlying syntactic object of (5.67), where a PC
represents the agent of a preterite participle. Firstly, the verb-defining head merges with the
root, introducing an internal argument and assigning a theme role (5.68).
KPtheme
pro V Root
(an evil deed) ∅ √kraupa-
gather
128
Secondly, the resulting syntactic object merges with the functional head Part(iciple) that later
surfaces as a preterite participle kakraupau with the verbalizer and the root. The internal argu-
ment then undergoes Internal Merge. The resulting syntactic object is shown in (5.69).
KPtheme
pro Part
(an evil deed) KPtheme
pro V Root
(an evil deed) ∅ √kraupa-
gather
Thirdly, Voice merges with (5.69), introducing an external argument and assigning an agent role
(5.70). The external argument is a KP, containing φmin/max whose value is third-person plural.
(5.70)
KPagent
φmin/max Voice PartP

me ∅ [iφ: 3pl ] pro Part
by them pro (an evil deed) KPtheme
(by them)
V Root
∅ √kraupa-
gather
Fourthly, the resulting syntactic object (5.70) undergoes External Merge with another functional
head Pass(ive). This functional head has unvalued φ-features, and copies the value of φmin/max
in the agent KP. As the result of Agree, φmin/max undergoes incorporation to Pass (5.71).
129
(5.71)
Pass VoiceP
agree
φmin/max Pass KPagent
[iφ: 3pl ] passive PartP

φmin/max Voice
me [uφ: 3pl ]
by them me ∅ [iφ: 3pl ] pro Part
by them pro (an evil deed) KPtheme
incorporation (by them)
V Root
∅ √kraupa-
gather
The functional head Pass also triggers Internal Merge of PartP, “smuggling” (Collins 2005) the
internal argument above the external argument (5.72).
(5.72)
PartP
KPtheme Pass VoiceP

pro Part φmin/max Pass KPagent
(an evil deed) KPtheme [iφ: 3pl ] passive PartP
φmin/max Voice
Root me [uφ: 3pl ] K DP
V
∅ √kraupa- by them ∅ [iφ: 3pl ]
gather pro
(by them)
Fifthly, Aux(iliary) merges with (5.72). It triggers the head-movement of Pass and creating the
complex head consisting of Aux, Pass, and φmin/max .6
6. In the derivation in (5.73), one has to stipulate that Pass does not have φ-features and that Aux does not attract
PartP but Pass.
130
(5.73)
Aux PassP
Pass Aux PartP
Pass √taka- KPtheme Pass

φmin/max
passive will be pro Part KPagent
[iφ: 3pl ]
me [uφ: 3pl ] (an evil deed) KPtheme PartP
φmin/max Voice
by them Root K DP
V
∅ √kraupa- ∅ [iφ: 3pl ]
gather pro
(by them)
Subsequently, the tense head T then merges with the structure. This time, it triggers head-
movement of the Aux-Pass-φ complex and creates a complex head. When the resulting syntactic
object undergoes Transfer, the plural PC -me, representing the agent of the preterite participle,
clusters with the Pass-Aux-T complex, that surfaces as the finite auxiliary tākan- ‘(it) will be.’

Transfer … KPtheme … KPagent … Root V Part Pass+Aux+T φ
TP ======⇒
… ∅pro … ∅pro … kakraupau tākan -me
PCs may also spell out the agent of an infinitive when they may undergo clitic climbing. In (5.75),
the third-person singular PC -ne has the agent role to the (transitive) infinitive wentsi ‘to speak,’
which takes waike waṣe käskor wat ‘lie, untruth, or gossip’ as the internal argument.
(5.75) [TB] -ne = agent of the infinitive wentsi ‘to say X’

waike waṣe ṣpä käskor wat wen-tsi klyin-ne
lie gossip conj gossip conj say-inf be.neccesary.subj.3sg-3sg
po weṣṣäṃ śwātsi-ntse pernesa :

all say.npst.3sg food-gen for.the.sake.of
‘And if it is necessary for him (= the ignorant one) to utter a lie, gossip or senseless talk,
he says everything for the sake of food.’ (B31b4; verse [8¦7¦6]×2 + [9¦9] +[7¦6])
This example suggests that there is a functional head that serves as the licensor of a PC and that
the licensor is merged above Voice. Let us look at each of the steps that derive the underlying
131
syntactic object of (5.75). Firstly, the verbalizer merges with the root. It also triggers External
Merge of an internal argument and assigns a theme role. The resulting syntactic object is (5.76).
(5.76)
KPtheme
[iφ: 3sg ] V root
waike waṣe käskor wat -ñ- √we-
a lie, gossip or senseless talk speak
Secondly, the functional head Voice merges with (5.76). It introduces an external argument con-
taining φmin/max and assigns an agent role to it (5.77).
(5.77)
KPagent
φmin/max Voice
[iφ: 3sg ] K DP KPtheme
ne ∅ [iφ: 3sg ] [iφ: 3sg ] root
V
for him pro waike waṣe käskor wat -ñ- √we-
a lie, gossip or senseless talk speak
Thirdly, another functional head Inf(initive) undergoes External Merge with (5.77). It triggers
head-movement of Voice to Inf, creating the complex head consisting of Voice and Inf (realized
as the infinitive marker /-tsi/). It also triggers Internal Merge of VoiceP (5.78).
(5.78)
VoiceP
KPagent Inf VoiceP
φmin/max Voice Voice Inf KPagent
[iφ: 3sg ] K DP [uφ: 3sg ] KPtheme -tsi

φmin/max Voice
ne ∅ [iφ: 3sg ] root to
[iφ: 3sg ] V [iφ: 3sg ] K DP KPtheme
for him pro waike waṣe käskor wat -ñ- √we- ne ∅ [iφ: 3sg ] V root
[iφ: 3sg ]
(for him) a lie, gossip or senseless talk speak for him pro -ñ- √we-
waike waṣe käskor wat
(for him) a lie, gossip or senseless talk speak
This infinitival phrase then undergoes External Merge with the functional head Mod(al), speci-
fied as [deontic] (5.79). It has unvalued φ-features, and looks for valued φ-features in its merge-
mate.
132
(5.79)
Moddeontic
[uφ: ] VoiceP
√klin- KPagent Inf VoiceP
be necessary
φmin/max Voice Voice Inf
[iφ: 3sg ] K DP [uφ: 3sg ] KPtheme -tsi

[iφ: 3sg ] V
(for him) a lie, gossip or senseless talk speak
As a result, it copies the person and number features of φmin/max in the agent KP, and incorporates
it, forming the complex head consisting of Moddeontic and φmin/max (5.80).
(5.80)
Moddeontic
φmin/max Moddeontic VoiceP
[iφ: 3sg ] [uφ: 3sg ] KPagent Inf VoiceP

ne √klin- Voice Inf
φmin/max Voice
for him be necessary [uφ: 3sg ] -tsi
[iφ: 3sg ] K DP KPtheme
Agree [iφ: 3sg ] V
Incorporation (for him) a lie, gossip or senseless talk speak
Subsequently, another modal head specified as [irrealis] and the tense head T merge with the
structure.7 When the resulting syntactic object undergoes Transfer, the third-person singular
PC -ne, representing the agent of the infinitive, clusters with the finite modal verb klyin- ‘it is
necessary,’ appearing immediately after it.
KPagent KPtheme Root V Voice-Inf Mod+T φ

Transfer
(5.81) TP ======⇒ ∅pro waike waṣe … käskor wat / we- + -ñ- -tsi / / klin- + -n -ne /
wen- -tsi klyin- -ne
To summarize, the model I developed for the Tocharian PCs showed that the distribution of the
PCs is more restricted than previously thought. It predicted that they could not spell out the
7. This analysis predicts that if φmin/max starts as the internal argument of an infinitive, it will undergo φ-
agreement with Voice and, with Voice-to-Inf movement, form a complex head with Inf. However, this prediction
turns out to be false since only a finite verb may host a PC in TB. We assume that when a probe above Inf finds
this φmin/max , it undergoes φ-agreement with the φmin/max , and as a result, φmin/max excorporates from Voice and
incorporates to the probe (see I. G. Roberts 2010 for technical details).
133
possessor of an external argument unless there is a licensor that c-commands the external ar-
gument. PCs cannot semantically represent the possessor of the nominal expression contained
in another nominal expression as an adjunct, either. Examples such as (5.59), (5.67), and (5.75)
support our prediction since the licensor of a PC (Voice) should c-command a causee KP and be
able to find person and number features there, and since there is another licensor of a PC in the
context of clitic climbing, which c-commands an external argument, and the person and number
features contained in the external argument should be visible to the licensor.
5.2 Implications for split intransitivity
The analysis we developed in the previous section has further implications for our understand-
ing of intransitive verbs in Tocharian. The model predicted that pronominal clitics could not
represent the possessor associated with the external argument of a hosting verb. This analysis
implies that when PCs represent the possessor of an intransitive subject, the subject should not
be an external argument introduced by Voice, but an internal one.
As observed in Chapter 4, PCs may represent the possessor associated with the subject of an in-
transitive verb. For example, the first-person singular PC -ñ represents the possessor of prosko
‘fear’ in (5.82). My analysis suggests that the subject prosko is the internal argument of the verb,
c-commanded by the licensor of the PC.
(5.82) [TB] PC representing the possessor of prosko ‘fear,’ which is the subject of the intransitive
verb lāma-
tune taukau-c saim pācer ¦ lāma-ñ prosko

therein hide.subj.act.1sg-2sg protection father rest.subj.act.3sg-1sg fear.nom.sg
10-3
‘Therein I will hide in your protection, father, so that my fear may rest. [13d] ’
(IOLToch5b2; verse [7¦7¦4]×4)
It is commonly held that there are two subclasses of intransitive verbs: unergative and unac-
134
cusative verbs (Perlmutter 1978; Burzio 1986). Unergative verbs take an argument that shows
the same distribution as the external argument of a transitive verb. In contrast, unaccusative
verbs select an argument patterning with the internal argument of a transitive verb. For ex-
ample, in German, the so-called split NP construction (or split NP topicalization; Fanselow 1988;
van Riemsdijk 1989; van Hoof 2006), which separates a head noun from its satellite, is possible
for transitive objects and unaccusative subjects but not for transitive subjects and unergative
subjects (Grewendorf 1989, Alexiadou, Anagnostopoulou, and Everaert 2004: 7).
(5.83) German (examples based on Schäfer 2008: 191)
i. Kleider, hat er immer dreckige an.

clothes has he always dirty on
‘As for clothes, he always wears dirty ones.’ (transitive, object)
ii. Fehler, sind dem Hans vermeidbare unterlaufen.

mistakes are the Hans avoidable occurred
‘With respect to mistakes, only those which were avoidable occurred to Hans.’
(unaccusative, subject)
iii. *Studenten, haben fleißige das Seminar besucht.

students have hard-working the class visited
‘Concerning students, hard working ones visited the class.’ (transitive, subject)
iv. *Studenten, haben fleißige telefoniert.

students have hard-working called
‘Concerning students, hard working ones called.’ (unergative, subject)
In Georgian, the so-called Series II forms (one of the TAM categories) mark both transitive subject
and unergative subject by -ma (-m after a vowel). In contrast, transitive objects and unaccusative
subjects are both marked by -i (-∅ after a vowel).
(5.84) Georgian (Harris 1982: 293)
i. vano-m gamozarda ʒma-∅

Vano-act 3S.3DO.grow.II brother-nom
‘Vano raised his brother.’ (transitive)
135
ii. bavšv-ma it’ira
child-act 3S.cry.II
‘The child cried.’ (Active intransitive [= unergative])
iii. rezo-∅ gamoizarda

Rezo-nom 3S.grow.II
‘Rezo grew up.’ (Inactive intransitive [= unaccusative])
In Italian, cliticization of a partitive phrase by the clitic pronoun ne is possible for direct objects
and unaccusative subjects (5.85i and 5.85ii, respectively) but not for unergative subjects (5.85iii)
(Belletti and Rizzi 1981).
(5.85) ne cliticization in Italian (examples from Alexiadou, Anagnostopoulou, and Everaert 2004:
6)
i. Giovanni ne ha insulati due.

John of.them has insulted two
‘John has insulted two of them.’ (transitive)
ii. Ne arrivano molti.

of.them arrive many
‘Many of them arrive.’ (unaccusative)
iii. *Ne telefonano molti.

of.them telephone many
‘Many of them telephone.’ (unergative)
I adopt, with many others, the idea that unaccusative and unergative verbs have different syn-
tactic structures. Specifically, the argument of unergatives is an external argument introduced
by Voice (5.86), while that of unaccusatives is an internal argument introduced by the category-
defining head V (5.87).8
8. I assume that the difference between unergatives and unaccusatives is semantically determined and syntacti-
cally encoded (Levin and Rappaport Hovav 1995; Sorace 2000, among others).
136
(5.86) Unergatives
VoiceP
KPagent
subject Voice
V root
(5.87) Unaccusatives
VoiceP
Voice
KPtheme
subject V root
My analysis suggests that the first-person singular PC -ñ in example (5.82) is a part of not the
external but the internal argument of lāma-, and the verb is therefore unaccusative. The inter-
mediate syntactic structure of (5.82) is illustrated in (5.88). In this structure, φmin/max , which
surfaces as a PC, starts as a part of the internal argument (5.88). The licensor of a PC (Voice),
which c-commands the internal argument, copies the value of φmin/max and incorporates it.
(5.88) Structure of unaccusatives (lāma-ñ prosko ‘my fear may rest’)
Voice
[iφ: 1sg ] [uφ: 1sg ] KPpossessor root

V
ñ √läm(ā)-
K DP ∅
Ag φmin/max
my re sit
e [iφ: 1sg ] K ∅
DP KPpossessor (unaccusative)
ñ ∅ [iφ: 1sg ]
D NP
Inc my pro
or ∅
por KPpossessor
ati (my)
on
N root
prosko
fear
In contrast, PCs cannot represent the possessor associated with the external argument of an
137
unergative verb. I assumed that the unvalued person and number features on Voice look for
valued φ-features in its c-commanding domain (§4.5). The structure of an unergative verb is
shown in (5.89). Voice’s c-commanding domain is VP, which consists of V and Root. The exter-
nal argument is simply outside the domain.
(5.89) Structure of unergatives
KPagent
KPpossessor Voice VP
DP [uφ: ] root
φmin/max K V
[iφ: 1sg ] K ∅ ∅ (unergative)
DP
ñ ∅ [iφ: 1sg ]
my [Case: Gen ]
pro
(my) x Agree
Therefore, the analysis I developed enables us to single out unaccusatives in Tocharian.
(5.90) If a pronominal clitic represents the possessor associated with the subject of an intran-
sitive verb, the verb is unaccusative.
If this analysis is on the right track, we should find a PC associated with the subject of an intransi-
tive verb whose unaccusative behavior receives independent support. In this respect, Levin and
Rappaport Hovav’s (1995: 98) study is of importance because it provides us with independently
motivated unaccusative verbs. They classify verbs that participate in causative-inchoative alter-
nation (labile verbs) as externally caused verbs. Externally caused verbs describe eventualities
brought about by an external force.
(5.91) Externally caused verbs (Levin and Rappaport Hovav 1995: 93)
a. Change of state:
138
bake, blacken, break, close, cook, cool, dry, freeze, melt, open, shatter, thaw, thicken, whiten,
widen, …
b. Verbs of motion:
bounce, move, roll, rotate, spin, …
They argue that all externally caused verbs are unaccusatives, whereas internally caused verbs
are mostly unergatives. Alexiadou and Anagnostopoulou’s (2004) study also points in the same
direction. They divide Modern Greek labile verbs into several subclasses. Interestingly, all sub-
classes agree that they do not have any functional head introducing an external argument.9 In
other words, they are all unaccusatives.
TA and TB also have a set of labile verbs that show both middle and active inflections (Table 5.1;
cf. Malzahn 2010: 87–9). They retain the same stem shape, and they surface as an intransitive
verb with a middle ending and a transitive verb with an active ending.10 The internal argument
of an active labile verb corresponds to the sole argument of the middle counterpart.11 In other
words, the sole argument of a middle labile verb is an internal argument c-commanded by Voice.
Therefore, I predict that I should find a PC that represents a possessor associated with the subject
of a middle labile verb but not with the subject of an active labile verb.
(5.92) Prediction:
Pronominal clitics may represent the possessor associated with the subject of a middle
labile verb.
This prediction is borne out: the following example (5.93) attests siṃsantär- ‘X are satiated,’ which
is a labile verb, functioning as an intransitive verb with a middle inflection. In this example, the
9. Modern Greek uses the same non-active voice morphology for representing passive also. Angelopoulos, Collins,
and Terzi (2020) argue that the by-phrase of a passive verb is an argument introduced by the verb, rather than an
adjunct adjoined to a verb phrase.
10. The only exception is TA/TB √pyutk- ‘[mid] establish, create, accomplish; [act] come into being,’ which func-
tions as a transitive verb with a middle ending and an intransitive with an active ending (table 5.1). Why this mismatch
occurred is a topic for future research.
11. The intransitive forms denote both a process and a state, which is also found in languages other than Indo-
European, such as Austronesian and Wolof (Niger–Congo) (Hilda Koopman p.c.).
139
Lg Root Middle (intransitive) Active (transitive)
1 TB √käs- come to extinction extinguish
2 TB √krās(ā)- be angry annoy, vex
3 TA √täm- be born, come into being beget, generate
4 TB √näk- fall into ruin destroy
5 TB √näm- bow (bend oneself) bend (something)
6 TB √päk- cook, ripen cook (something), let ripen
7 TA/TB √mänt(ā)- destroy/be destroyed; be stirred, angry stir; destroy
8 TB √ru- be open open (something)
9 TA/TB √läk(ā)- be seen, appear see
10 TA √we- sprout let sprout
11 TA √si-n- satiate oneself, be depressed satiate
12 TB √tsäk- burn burn (something)
Lg Root Middle (transitive) Active (intransitive)
13 TA/TB √pyutk- establish, create, accomplish come into being
Table 5.1: Labile verbs in TA and TB
first-person singular PC -ñi represents the inalienable possessor associated with puk marmañ ‘all
of the veins,’ which is the subject of this middle labile verb.
(5.93) [TA] PC represents the possessor associated with the subject of a middle labile verb
[a1] /// klāmār siṃsantär-ñi oki cam

bring.subj.mid.1sg be.satiated.npst.mid.3pl-1sg like dem.m.acc.sg
klop-yo pu-k marmañ : ||
suffering-ins all-emp vein.nom.pl
‘(If) I bring […], all my veins will be satiated by this pain, as it were.’
(A116a1; trans. based on CEToM; verse)
In the following example, it seems that the third-person singular PC -ne, referring to the same
individual as cpi ‘his,’ represents the (inalienable) possessor of pilko ‘insight, view; look, glance,’
which is the subject of the middle labile verb mäntäṃtär-ne ‘X is destroyed.’ If this interpretation
is correct, then this example also supports my analysis.
(5.94) [TB] PC represents the possessor associated with the subject of a middle labile verb (?)
• sruka-(l)ñ(e-ṣ)ṣ(a)na cpi nmittä-nta ///

die-nmlz-adjz.f.pl dem.m.gen.sg mark-pl
140
/// [b6] ānts-ne cpi lkānträ 10 pilko
shoulder-du dem.m.gen.sg appear.npst.mid.3pl view
mäntäṃtär-ne tucya-ne e(śa-ne) ///
be.destroyed.npst.mid.3sg-3sg yellow-du eye-du
‘.. his signs of death … (his) two shoulders will appear to this one as …. [10] His view is
destroyed. (His) two yellow eyes …’
(B118b6; verse)
To summarize, this section showed that the analysis developed in Section 5.1 allows us to single
out unaccusative verbs in TA and TB. If a PC represents the possessor semantically associated
with the subject of an intransitive verb, we may consider the verb as belonging to unaccusatives.
We can now collect unaccusative verbs in TA and TB, based on the distribution of the pronomi-
nal clitics. Appendix I lists representative unaccusative verbs in TA and TB collected using this
method, further supported by comparative syntax.
5.3 Antigrundverbs and unaccusatives
The previous section showed that one may collect unaccusative verbs in Tocharian based on the
distribution of the pronominal clitics. However, this is only one possible approach, and one might
also think of collecting unaccusative verbs in Tocharian using different approaches. Several verbs
in TA and TB show causative-inchoative alternation (e.g., ‘[somebody] breaks X’ ∼ ‘X breaks’) by
alternating their stem shape. They form the class VIII present stem that functions as a transi-
tive verb, next to the class I, III or IV present stem that functions as an intransitive verb (“anti-
causative”). They have the transitive class I or II (or VII in TA) subjunctive stem and the intran-
sitive class V subjunctive stem. They also have the transitive class III preterite stem besides the
intransitive class I preterite stem. Malzahn (2010: 64) called such transitive verbs “antigrund-
verbs,” which in general provide “oppositional transitives” to unaccusative verbs. Therefore,
one might be tempted to conclude that if one finds an antigrundverb, then its corresponding
intransitive verb is unaccusative. However, this conclusion is not warranted as there is a non-
anticausative that builds an antigrundverb. According to Malzahn (2010: 65), there is at least one
antigrundverb, which appears to serve as a causative to an unergative intransitive. Example (5.95)
141
attests the infinitive śaccätsī ‘to make X cross,’ which is a transitive built on √kätk(ā)- ‘cross, pass.’
This root does not participate in causative-inchoative alternation.12
(5.95) Antigrundverb next to a non-anticausative intransitive verb?

krent yamor mā yāmoṣäṃ ¦ cen (n)o śäccä-tsī
good.acc.sg deed neg do.ptcp.m.acc.pl dem.m.acc.pl conj proceed.caus-inf
pkate 10-4
intend.pst.mid.3sg
‘But thosej who had not done a good deed, he intended to make themj proceed. [14b] ’
(B133a4; verse; [7¦7]×2)
Also, not all anticausatives form an antigrundverb. Some anticausatives have a transitive coun-
terpart with voice alternation (e.g., √si-n- in TA) or with a productive derivational suffix and a
fixed initial accent (e.g., √si-n- with -äṣṣä-/-äske- in TB). To summarize, the existence of an anti-
grundverb does not imply the existence of a corresponding anticausative, and vice versa (5.96).
(5.96) Two false propositions
i. (False:) If there is a (transitive) antigrundverb, its corresponding intransitive is anti-

causative.
ii. (False:) If there is an anticausative, there is a corresponding (transitive) antigrund-

verb.
By focusing on the antigrundverbs, we may still collect intransitive verbs. However, there is no
independent support that suggests the intransitive verbs collected are unaccusative in TA and
TB.
5.4 External possession as a diagnostic for unaccusativity
When a PC represents a possessor, it undergoes Transfer outside the nominal phrase containing
its possessum. In this sense, one may consider it to show an external possession construction,
12. One might be inclined to take a verb built on √kätk(ā)- as unaccusative, rather than unergative (cf. the verb for
‘pass’ is unaccusative in Dutch; Hilda Koopman p.c.).
142
where a nominal phrase is morphologically associated with a verb despite being semantically
associated with the verb’s argument. In Hebrew, scholars have used external possession as a
diagnostic for unaccusativity (Borer and Grodzinsky 1986, Landau 1999): a verb is unaccusative
if one may associate an external possessor with the verb’s subject semantically.13 However, this
view has recently faced challenges by a couple of studies (Linzen 2014, Gafter 2014). They argue
that in Hebrew, (1) some unaccusative verbs do not allow an external possession construction and
(2) some unergative verbs do allow an external possession construction. For our purpose, (2) is
of interest. Linzen (2014) and Gafter (2014) report that verbs of emission of sounds/lights (e.g.,
‘shine,’ ‘crack’) allow an external possession construction as in (5.97). In this example, le-xaim
‘to Chayim’ is the possessor of ha-pelefon ‘the cell phone,’ which is the subject of the intransitive
verb cilcel ‘X rang.’
(5.97) External possession construction of an (alleged) unergative subject in Hebrew

ba- pgiša ha- reviit cilcel le- xaim ha- pelefon, …
in.the date the fourth rang to Chayim the cell.phone
‘On the fourth date, Chayim’s cell phone rang.’ (Gafter 2014: 486)
However, their observation does not undermine my analysis since it is not the case that verbs of
emission are always unergative. Levin and Rappaport Hovav (1995) point out that some verbs of
emission may describe both externally and internally caused eventualities (Levin and Rappaport
Hovav 1995: 115–9).
(5.98) Internally/Externally caused eventualities
a. The doorbell rang. (Internally caused eventuality)
b. The postman rang the doorbell. (Externally caused eventuality)
Auxiliary selection in Italian also points in the same direction. In most Germanic and Romance
13. External possession is “a phenomenon where a nominal is syntactically encoded as a verbal dependent but
semantically understood as the possessor of one of its co-arguments” (Deal 2017: 391–2). Some Tocharian examples
are indeed interpretable as containing the possessor raising construction. However, one should keep in mind that
not all PCs represent a possessor.
143
languages, past participles of unaccusative verbs select an auxiliary be, while those of unergatives
select have, when they form a periphrastic construction (Perlmutter 1989).
However, while some verbs constantly choose one particular auxiliary across languages, some
verbs show gradient behavior, selecting be in one language but have in another. For example,
Sorace (2000: 875) points out that Italian correre ‘run’ selects have (5.99), while German rennen
‘run’ takes be (5.100).
(5.99) Italian (Sorace 2000: 875)

Gli atleti svedesi { hanno corso / ?sono corsi } alle Olimpiadi.
the athletes Swedish have run are run at.the Olympics
‘The Swedish athletes ran at the Olympic Games.’ (selects have)
(5.100) German (Sorace 2000: 875)

Uschi { *hat / ist } den ganzen Tag gerannt.
Uschi has is the whole day run
‘Uschi ran the whole afternoon.’ (selects be)
Moreover, for some verbs, the agentivity of a subject and the telicity of an eventuality play a role
(cf. Dowty 1991). Regarding the former, for example, Italian durare ‘to last’ prefers be, but when
it takes an animate (agentive) subject, have is also acceptable (5.101). As for the latter, Dutch
springen ‘to jump,’ which usually selects have, selects be when it accompanies a PP indicating an
endpoint of a motion (5.102).
(5.101) Italian (examples based on Sorace 2000: 867–8)
i. La guerra { e / ? ha } durato a lungo.

the war is has lasted for long
‘The war lasted a long time.’ (Inanimate subject)
ii. Il presidente { e / ha } durato in carica due anni.

the president is has lasted in post two years
‘The president held office for two years.’ (Animate subject)
(5.102) Dutch (examples based on Borer 2005b: 32)14
144
i. Jan heeft gesprongen.
Jan has jumped
‘Jan jumped.’ (Atelic eventuality)
ii. Jan is in de sloot gesprongen.

Jan is in the ditch jumped
‘Jan jumped into the ditch.’ (Telic eventuality)
Sorace (2000, 2004) proposed the following auxiliary selection hierarchy to account for this gra-
dient behavior (Table 5.2). This hierarchy contains two core classes: verbs denoting a change
of location and verbs denoting a controlled non-motional process. These classes prototypically
denote telic and atelic eventualities, respectively, and the former constantly choose be, while the
latter have, irrespective of agentivity or telicity. Between these core classes, there are non-core
classes of verbs that show gradient behavior. They are sensitive to agentivity and telicity, and
the line that separates unaccusative verbs from unergative verbs varies from one language to an-
other. For example, verbs denoting existence of state (e.g., semblare ‘seem’) prefer be in Italian,
while they select have in French.
selects be (least variation) Change of location

Change of state
Continuation of a pre-existing state
Existence of state
Uncontrolled process
Controlled process (motional)
selects have (least variation) Controlled process (non-motional)
Table 5.2: The Auxiliary Selection Hierarchy (Sorace 2000)
(5.103) Italian
i. Verbs denoting existence of state

La commedia { è sembrata / ??ha sembrato } interessante a tutti.
the play is seemed has seemed interesting to all
‘The play seemed interesting to everyone.’ (Sorace 2000: 869)
14. Koopman (2010) argues that directional PPs (such as example 5.102ii), unlike locative PPs, are complements in
Dutch.
145
(5.104) French
i. Verbs denoting existence of state

Le dinosaures { ?*sont / ont } existé il y a 65 millions d’ ans.
the dinosaurus are have existed there.are 65 millions of years
‘The dinosaurs existed 65 million years ago.’ (Sorace 2000: 869)
Sorace (2000, 2004) observed that verbs of emission (e.g., ‘rumble’), which belong to uncontrolled
process verbs, are less “agentive” than motional controlled process verbs (e.g., ‘swim,’ ‘run’) and
non-motional controlled process verbs (e.g., ‘talk,’ ‘work’). They may take both be and have
auxiliaries, while motional controlled process verbs strongly prefer have and non-motional con-
trolled process verbs may only take have in Italian.
(5.105) Italian
i. Verbs of emission
Il tuono { hanno / e } rimbombato.
the thunder has is rumbled
‘The thunder rumbled.’ (Sorace 2004: 262)
ii. Motional controlled process verbs

Gli atleti cinesi non { hanno corso / ?*sono corsi } alle Olimpiadi.
the athletes Chinese not have run are run at.the Olympic.Games
‘The Chinese athletes did not run at the Olympic Games.’ (Sorace 2004: 260)
iii. Non-motional controlled process verbs

I delegati { hanno parlato / *sono parlati } tutto il giorno.
the delegates have talked are talked whole the day
‘The delegates spoke all day. ’ (Sorace 2004: 256)
To summarize, the observation that some verbs of emission allow an external possession con-
struction in Hebrew does not argue against the analysis we developed. Some verbs of emission,
which belong to uncontrolled process verbs, may describe externally caused eventuality that is
syntactically represented as unaccusative.
146
5.5 Conclusion
This chapter developed a syntactic analysis that accounts for the multifunctionality of the pronom-
inal clitics in Tocharian A and B. Tocharian PCs appear immediately after a finite verb, outside
the nominal expression that they are semantically associated with. This is because when a PC’s
licensor finds it, it undergoes incorporation to the licensor. My analysis restricts the distribution
of the PCs in TA and TB in a falsifiable way.
(5.106) Pronominal clitics in TA and TB cannot be semantically associated with the subject of a
transitive/unergative verb unless there is an eligible host c-commanding them.
(5.107) Pronominal clitics in TA and TB cannot be semantically associated with the nominal
expression contained in another nominal expression as an adjunct.
My analysis is supported by the fact that a PC may represent the causee of a morphological
causative, and that when a PC undergoes clitic climbing, it may represent the external argument
of a non-finite verb. Furthermore, we have shown that we may single out unaccusative verbs
based on the distribution of PCs.
(5.108) If a pronominal clitic is semantically associated with the subject of an intransitive verb,
the verb is unaccusative.
We list representative roots that form unaccusative verbs in TA and TB in Appendix I. They in-
clude middle labile verbs whose unaccusative behavior is independently supported (e.g., TA √si-n-
‘[act] satiate; [mid] satiate oneself; be depressed’). The representative roots we identified as
forming unaccusatives include verbs denoting a change of location (e.g., TA √i-|kälkā- ‘go’ [used
metaphorically]), a change of state (e.g., TA/TB √ār(ā)- ‘cease, come to an end,’ TA/TB √kän- ‘come
about, occur, be fulfilled’), the continuation of a pre-existing state (e.g., TA √trik(ā)- ’be confused;
faint’), the existence of a state (e.g., TA/TB √nas-|tāk(ā)- ‘be, become’), and uncontrolled processes
(e.g., TA √prutk(ā)- ‘be shut, be filled,’ TB √plätk- ‘overflow, develop, arise,’ TB √plu- ’float, fly, soar,’
and TB √spalkā- ‘be agitated, tremble’). Notably, I did not find any controlled process verb such
147
as ‘work,’ ‘play,’ ‘talk,’ or ‘swim,’ which we expect to surface as unergatives (table 5.2).
Unaccusativity in the ancient Indo-European languages has been a challenging topic. Since no
grammaticality judgement is available, we need to build a morphosyntactic criterion that de-
termines unaccusativity of a verb without relying on semantics. There is a risk of circularity
if one solely relies on semantics of a word to determine whether a given verb is unergative or
unaccusative. My analysis provides a non-circular criterion which enables us to single out unac-
cusatives in Tocharian.
Research on unaccusativity in the ancient Indo-European languages has been advanced in Hit-
tite since Garrett (1996, 1990), who built on the observation of Watkins (1968–1969) and provided
criteria for determining unaccusativity of Hittite verbs: (1) If a verb is intransitive and accom-
panies a subject clitic, it is unaccusative. (2) Unergatives use have while unaccusatives use be
to form a periphrastic past tense in Hittite. Recently, Yates and Gluckman (2020) pointed out
that only unaccusatives switch from active inflection to middle inflection in Hittite imperfec-
tives. The morphosyntactic analysis I developed in this chapter also provides a basis based on
which one may identify whether there is any subtype of roots that continuously displays unac-
cusative/unergative behavior over course of time or unaccusativity may easily fluctuate from
one branch to another.
148
CHAPTER 6
Pronominal clitics and multiple pronominal arguments
6.1 Introduction
The previous chapter showed that the distribution of the Tocharian pronominal clitics (PCs) is
more restricted than previously thought: they never represent the possessor associated with a
transitive or unergative subject. Nor do they represent the possessor of the noun contained in
another nominal expression. However, there was only one pronominal argument in most of the
cases we examined, and we have not reviewed cases in which multiple arguments are pronominal.
Tocharian A and B (TA and TB) seem to have only one slot for a PC—there is no example in which
a single finite verb hosts multiple PCs. Therefore, one might wonder what happens if more than
one argument is pronominal. For example, if the indirect and direct object (henceforth IO and
DO) of a verb are pronominal, which of them is represented by a PC? Regarding this question, we
could think of two hypotheses that offer us two different predictions (table 6.1).
Hypothesis Prediction
PCs arbitrarily choose It is not possible to predict
1
the argument they represent. which argument PCs represent.
A syntactic structure determines We may predict which argument
2
which argument PCs represent. PCs represent based on the structure.
Table 6.1: Hypotheses and predictions
This chapter shows that a syntactic structure determines which argument PCs represent in Tochar-
ian. In other words, when the IO and the DO of a ditransitive predicate are both pronominal, PCs
hosted by the ditransitive predicate always represent the IO (with various thematic roles such as
goal, source, recipient, beneficiary, or addressee). My analysis further restricts the PCs’ distribu-
149
tion by predicting that there will be certain gaps in the data.
The structure of this chapter is as follows. Section 6.2 examines the examples where multiple ar-
guments are pronominal. In Section 6.3, I turn to the analysis, which accounts for the distribution
of the Tocharian PCs and predicts some absence of the data. Section 6.4 discusses further impli-
cations for our understanding of the Tocharian syntax and addresses the remaining questions.
Section 6.5 concludes the chapter.
6.2 Data
I have examined 551 attestations of the pronominal clitics in Tocharian A and 608 attestations in
Tocharian B. Figure 6.1 shows that more than 70% of the examples contained either an intransi-
tive or monotransitive predicate. 103x (18.7%) and 129x (23.4%) of the examples had a ditransitive
predicate in TA and TB, respectively.
Figure 6.1: Valency of host predicates in TA and TB
The eight most frequently attested roots in TA and the nine most frequent roots in TB account for
69.1 % and 72.8 % of the entire attestation of the ditransitive predicates in TA and TB, respectively
(Figures 6.2 and 6.3).
We may group the frequently attested ditransitive predicates into the following three types in
Table 6.2.
The first type selects an addressee as the IO and a theme as the DO. It is the most frequently
150
Figure 6.2: Frequently attested ditransitive predicates in TA
Figure 6.3: Frequently attested ditransitive predicates in TB
attested type of the ditransitive predicates in TA and TB. The second type prototypically selects a
theme as the DO and either a recipient, source, or beneficiary as the animate (human) IO. The last
type also selects the theme DO. It also has an animate source or beneficiary as the IO, but some
verbs in this class may also have an inanimate goal as the IO (e.g., TB√āk-|wāyā- ‘lead, guide, drive
Xanimate inanimate ’).
theme to Ygoal
Many examples show ambiguity regarding whether PCs represent the IO or DO when both are
pronominal. For example, the third-person singular PC -ne in (6.1) may, in principle, refer to
kampāl ‘cloak’ (DO; theme) or ājivike ‘an Ājīvika ascetic’ (IO; recipient) since both are third-person
151
Lgs Root Gloss
TB/TA √āks- ‘announce, proclaim, say’
TB √we-ñ- ‘[act] say, speak; [mid] be called’
Verbs of TA √träṅk-|we-ñ- ‘speak, say’
communication TB √pärk- ‘ask, bring up a question, ask for, beg’
TA √pärk- ‘[act] ask for, beg; [mid] ask, bring up a question’
TB √yāsk- ‘beg’
TB √ai-|wä(s)- ‘[act] give; [mid] take’
Verbs of
TA √e-|wä(s)? - ‘give’
transaction
TB √mäsk- ‘exchange, change’
TB/TA √kälā- ‘lead, bring’
TB √āk-|wāyā- ‘lead, guide, drive’
Verbs of TA √āk-|wā- ‘lead, guide, drive’
motion TA √pär-|kāmā- ‘carry, take’
TA √ya(p)-|yām- ‘do’
TA √tsäk(ā)- ‘pull, take (out, away)’
Table 6.2: Major ditransitive predicates in TB and TA
singular.
(6.1) [TB] -ne = IO or DO of wsā- ‘gave’?

Context: An Ājīvika ascetic tries to retrieve his cloak from Upananda.
upananden-meṃ kampāl päst ñaṣṣi

Upananda-abl cloak away demand.impf.act.3sg
sū mā wsā-ne
dem.m.nom.sg neg give.pst.act.3sg-3sg

‘(The Ājīvika ascetic) begged the cloak from Upananda, (but) he did not give -ne .’
(PKAS18Ab3; prose)
Therefore, to avoid this type of ambiguity, I must focus on the examples in which two pronominal
arguments differ in person or number (e.g., I give [do:3sg it] [io:2sg to you]). I have collected 20 such
examples (9 TB and 11 TA), and I will review them in Sections 6.2.1 (TB) and 6.2.2 (TA). I will show
that PCs always represent the IO of a ditransitive predicate in such cases.
152
6.2.1 Examples (TB)
This subsection considers the nine TB examples where two pronominal arguments differ in per-
son or number. In the first example (6.2), there is a ditransitive verb that hosts a PC (aksau ‘I
will tell [X to Y]’ + -mme). The IO and DO of this verb are both pronominal. The PC unambigu-
ously represents the IO (addressee). The DO is third-person singular, represented by the neuter
demonstrative (anaphoric) pronoun tu.
(6.2) [TB] -me = IO (addressee) of aksau ‘I will tell’
/// ¦ mā tudo ñiś ñake aksau-mme •

neg dem.n 1sg now speak.subj.act.1sg-pl
‘I will not tell it to youpl now.’
(B108b10; verse; [7¦8]×4)
Likewise, the root √āks- ‘to announce, proclaim, say’ takes two pronominal arguments in exam-
ples (6.3), (6.4), and (6.5): content of which a speaker speaks (DO; theme) and a person/people
to whom (s/he) speaks (IO; addressee). PCs represent the IO (addressee) of the verb, while overt
demonstrative pronouns (tu, tu pw, and ce, respectively) represent the DO (theme).
(6.3) [TB] -me = IO (addressee) of ākṣāwa ‘I told’
ku se ñi yesñ= āksaṣle ¦ ku se wat no

rel.nom gen.1sg gen.2pl announce.gdv.nom.sg rel.nom conj ptcl
enäṣlyi ¦ tu pwdo ākṣā[b8](wa-me :)
teach.gdv.nom.pl dem.n all speak.pst.act.3sg-pl
‘What I had to proclaim to youpl , or what had to be taught, I have told all (of) it to youpl .
[70a] ’
(B27b7-8; verse; [6¦6¦5]×4)
(6.4) [TB] -me = IO (addressee) of akṣā- ‘(he) told’
aiśaumyi ceu pä[a5]llāntär krento ¦ āstreṃ śaul

wise.nom.pl dem.m.acc.sg praise.npst.mid.3pl good.acc.sg pure.acc.sg life
153
śay-eñcai wnolme :
live-ptcp.acc.sg living.being
snai laiwo ṣpane cedo tne ¦ ṣeme ślok akṣā-me 30-8

without lassitude sleep dem.m.acc.sg here single strophe speak.pst.act.3sg-pl
‘The sages praise him as a being (who) lives a good (and) pure life [38c] , without lassitude
(or) sleep. As the first strophe, he told this to them here. [38d] ’
(B31a5; trans. based on CEToM; verse; [8¦7¦6]×2 + [9¦9] + [7¦6])
(6.5) [TB] -ñ = IO (addressee) of pokse- ‘tell!’
m(ā) alyaik nano tu-k yäkne[a4](-sa) /// /// mäskentär tūdo

neg other.nom.pl again dem.n-emp way-perl be.npst.mid.3pl dem.n
pokse-ñ ||
announce.imp.act.2sg-1sg
(The king speaking to Hastāṅkuśa): “And again, are the other ones not (seekers of good)
precisely in this way? Tell this to me!”
(PKNS34a4; trans. based on CEToM)
The following example (6.6) contains a ditransitive verb (preku ‘I will ask’), which takes two pronom-
inal arguments: the question asked (DO; theme) and the individual to whom the speaker asks a
question (IO; addressee). The second-person singular PC -c of this example unambiguously rep-
resents the IO of the verb. The DO of this verb is the neuter independent demonstrative pronoun
tu.
(6.6) [TB] -c = IO (addressee) of preku ‘I will ask’
ceᵤ pañäkte yātka wasatpāt yām-tsi •

dem.m.acc.sg Buddha.lord order.pst.act.3sg ordination make-inf
tudo ñake tane preku-c mäkte wänta[b4]re tākaṃ tu

dem.n now here ask.subj.act.1sg-2sg how matter cop.subj.act.3sg dem.n
yäkne-sa poñ ||
way-perl say.imp.act.2sg
klyauṣi aṣ(an)ī(k)e sā(ṅk)

hear.opt.act.3sg venerable community
154
‘Such a being the Buddha has commanded to be ordained. Now I will ask you about it
here. Say how the matter is, (precisely) in that way! May the venerable community hear
(it)!’
(THT1114b3; trans. based on CEToM; prose? )
In the following example (6.7), there are two PCs: -ñ [1sg] and -cä [2sg]. Both represent the IO
(beneficiary and recipient, respectively) of the verbs of transaction. This example contains a
referential null object (pro) as the DO, which is third-person singular, referring to the cloak.
(6.7) [TB] -ñ = IO (beneficiary) of myāskasta- ‘you exchanged’

-cä = IO (recipient) of aiskau ‘I give’
tumeṃ su upanandeṃnmeṃ kampāl päst«†ä» ññaṣṣi •

then dem.m.nom.sg Upananda-abl cloak away demand.impf.act.3sg
upanande mā wsā-ne te [b2] weñā-ne

pn neg give.pst.act.3sg-3sg dem.n speak.pst.act.3sg-3sg
myāskasta-ñ mā aiskau-cä
exchange.pst.act.2sg-1sg neg give.npst.act.1sg-2sg
‘Then he demanded the cloak back from Upananda. (But) Upananda didn’t give [it] back
to him. He said this to him: “You traded [it] with me. I’m not giving [it] to you.”’
(B337b2; trans. based on CEToM; prose)
Examples (6.8) and (6.9) attest a ditransitive verb hosting a third-person singular PC. The verb also
takes an overt pronominal DO in third-person singular (po tw and ceu , respectively). Although we
cannot exclude the possibility that the PC -ne doubles the DO in these examples, it is more likely
that the PC represents the IO (addressee) of the verb as there is no secure example in which a PC
doubles an inanimate nominal expression in TB.1
(6.8) [TB] -ne = IO (addressee) of akṣā- ‘(he) announced’
1. I will examine cases in which a PC doubles a nominal expression in Chapter 7.
155
klyauṣa sū śaumo ¦ mas= āmāciṃ-śco ¦ po twdo
hear.pst.act.3sg dem.m.nom.sg man go.pst.act.3sg minister-all all dem.n
akṣā-ne amāc masa ¦ lānte tw ākṣa
speak.pst.act.3sg-3sg minister go.pst.act.3sg king.gen.sg dem say.pst.act.3sg
aurtsesa :
in.detail
‘This man heard (it) (and) went to the minister [and] reported it all to him. The minister
(then) went to the king and announced this in detail.’
(B18a1; trans. based on CEToM; verse; [5¦5¦8¦7]×4)
(6.9) [TB] -me = IO (addressee) of weñā- ‘said’
ceudo mänt wälo weñā-ne ¦ ///

dem.m.acc.sg thus king.nom speak.pst.act.3sg-3sg
‘In this way, the king told that to him.’
(B133a7; verse; [7¦7]×2)
Table 6.3 summarizes the examples discussed. All of the examples contained an animate human
IO and an inanimate DO. Furthermore, the DOs were all third-person singular. In all of these
examples, PCs unambiguously represented the IO.2
6.2.2 Examples (TA)
This subsection reviews eleven TA examples and shows that when the IO and DO of a verb are
pronominal, PCs also consistently represent the IO in TA.
In the first example (6.10), the IO (recipient) of the verb is the pretas (hungry demons), which is
third-person plural. The DO of the verb, not overtly expressed, is third-person singular, referring
to śwātsi ‘food.’ The ditransitive verb paṣ- attests together with the plural PC -äm, unambiguously
representing the IO.
2. In the examples discussed, the subject is non-third-person when an IO is first- or second-person. It is an open
question whether the person of the subject is also relevant.
156
ani- hu- ani- hu-
PC IO DO
mate? man? mate? man?
TB√āks- ‘announce, proclaim, say’
(6.2) aksau-mme pl pl + + 3sg tu - -
(6.3) ākṣā(wa-me) pl pl + + 3sg tu pw - -
(6.4) akṣā-me pl pl + + 3sg ce - -
(6.5) pokse-ñ 1sg 1sg + + 3sg tū - -
TB√pärk- ‘ask, bring up a question,
ask for, beg’
(6.6) preku-c 2sg 2sg + + 3sg tu - -
TB√mäsk ‘exchange, change’
(6.7) myāskasta-ñ 1sg 1sg + + 3sg (pro) - -
TB√ai- ‘[act] give; [mid] take’
(6.7) aiskau-cä 2sg 2sg + + 3sg (pro) - -
TB√āks- ‘announce, proclaim, say’
(6.8) akṣā-ne 3sg 3sg + + 3sg po tw - -
TB√we-ñ- ‘[act] say, speak;
[mid] be called’
(6.9) weñā-ne 3sg 3sg + + 3sg ceu - -
Table 6.3: Summary of the examples in TB
(6.10) [TA] -äm = IO (recipient) of paṣ- ‘Give (something) (to X)!’
paṣ-äm śwā-tsi paṣ-ämio nātäk ¦ pwikā-m

give.imp.act.2sg-pl eat-inf give.imp.act.2sg-pl lord disappear.caus.imp.act.2sg-pl
klop caṣ k(aśśiñ was 70-8)
suffering dem.m.acc.sg hungry.nom.pl nom.1pl
‘[The Pretas speaking to Koṭikarṇa:] “… [78c] Give us food, give us, oh lord, remove this
suffering from us! (We [are] hungr)y.” [78d] ’
(A340a4; trans. based on CEToM; verse; [5¦5¦5¦5] + [8¦7¦7] + [5¦5] + [8¦7])
The following example (6.11) contains two occurrences of et- ‘(if) he gives’ followed by the first-
person singular PC -ñi. The DO of this ditransitive verb is a referential null object (pro), referring
to päñ känt tināräs ‘500 denars (of money)’ [pl]. The PC represents not the DO (theme) but the IO
(recipient), referring to the brahmin Nirdhana.
(6.11) [TA] -ñi = IO (recipient) of et- ‘(if he) gives (money) (to X)’
157
räskrä arū nirdhane trä(ṅkäṣ) /// [a6] /// (ku)pre-ne
bitterness evoke.ptcp.m.nom.sg Nirdhana speak.npst.act.3sg if-comp
et-ñiio kāsu śāwaṃ ākāläntu
give.subj.act.2sg-1sg good great.acc.pl wish.acc.pl
knāsam-ci • ku(pre-ne) nu mā et-ñiio ||
be.fulfilled.caus.subj.act.1sg-2sg if-comp conj neg give.subj.act.2sg-1sg
samakkorren-aṃ || ṣpät koṃ-s-aṃ ywārckā ///
s-loc 7 day-pl-loc in.the.middle
‘Nirdhana, greatly angered, says: “… If yousg give me (the money), I will make yoursg
great wishes come true. If, however, yousg do not give it to me, [In the S-tune:], in seven
days …”’
(A215a7; trans. based on Ji, Winter, and Pinault 1998: 45; prose)
YQ I.6 b6 (6.12) shows the same passage.
(6.12) [TA] -ñi = IO (recipient) of the transitive verb et- ‘(if yousg ) give (money) to X’
/// (ā)k(na)ts kuro mok kupre-ne et-ñiio kāsu śāwaṃ ākā[b7](läntu

ignorant aged old if-comp give.subj.act.2sg-1sg good great.acc.pl wish.pl
knāsam-ci •)
be.fulfilled.caus.subj.act.1sg-2sg
[Nirdhana speaking to Bādhari:] “Yousg ignorant, feeble old man! If yousg give me (the
money, I will make yoursg ) great wishes (come true).”
(YQ I.6b6; trans. based on Ji, Winter, and Pinault 1998: 45; prose)
YQ I.6 b4 (6.13) also attests et-, whose IO (recipient) and DO (theme) are both pronominal (IO: 1sg,
DO: pl, respectively). Again, this verb hosts the first-person singular PC, which unambiguously
represents the IO.
(6.13) [TA] -ñi = IO (recipient) of the transitive verb paṣ- ‘(if yousg ) give (500 gold coins) to X’
ku pre-ne mā et-ñiio wtāk ṣakkats dhanīke protk-aṃ

if-comp neg give.subj.act.2sg-sg later certainly rich.man.nom.sg prison-loc
prutkāṣ-ñi • ||
be.shut.caus.subj.act.3sg-1sg
158
[Nirdhana speaking to Bādhari:] “If yousg do not give me (the money), the rich man will
surely lock me up in prison.”
(YQ I.6b4; trans. based on Ji, Winter, and Pinault 1998: 45; prose)3
In the following example (6.14), Nanda laments his separation from his wife, Sundarī. This pas-
sage contains the second-person singular imperative psumār- ‘Take X away!’, whose IO (source)
and DO (theme) are pronominal. The IO refers to the speaker Nanda, while the DO’s referent is
unclear (perhaps klop ‘suffering’?). This ditransitive verb accompanies the first-person singular
PC, unambiguously indexing the IO.
(6.14) [TA] -ñi = IO (source) of psumār- ‘Take (this) away from X!’
[a5] /// camdo śkaṃ lo psumār-ñiio kᵤyal lykäly lykäly

dem.m.acc.sg conj far take.away.imp.mid.2sg-1sg why fine fine
tuṣt-ñi :
burn.act.2sg-1sg
‘[a5] … and take that away from me! Why do yousg burn me finer and finer? ...’
We could probably include the following example (6.15). In this passage, King Prasenajit requests
water from Mālikā (Schmidt 1974: 376). If the DO is not in the lacuna but a referential null object
(pro) referring to water, the first-person singular PC -ñ represents the IO (goal) rather than the
DO (theme) of the ditransitive verb.
(6.15) [TA] -ci = IO (goal) of klāte- ‘(yousg ) brought (water) to X’ (?)
[b6] /// lyīk-tsi wär prakwā-ci nu nak klāte-ñ(i)io ///

wash.npst-inf water beg.pst.act.1sg-2sg again bring.pst.mid.2sg-1sg
‘I asked water from yousg to wash .... Again, yousg brought (it) to me …’
(A431b6; prose?)
3. Cf. A215a5: /// (ku pre)ne tu mā – – – wtāk ṣakkats dhani(ke) protkaṃ prutkāṣ-ñi • || “(If) yousg (do) not (give me
the money), the rich man will surely lock me up in prison.” For a parallel in Turkish, see Geng and Klimkeit (1988:
282–3) and Tekin (1980: 48) (Peyrot 2013b: 246)
159
In the following example (6.16), tränkäm- ‘(I) say X’ hosts a PC, referring to the second-person sin-
gular. This verb’s IO (addressee) and DO (theme) are both pronominal (2sg and 3sg, respectively).
The PC unambiguously indexes the IO of the verb.
(6.16) [TA] -ci = IO (addressee) of the transitive verb träṅkäm- ‘(I) say (that) to X’
(smi-māṃ akma)l-yo wāskāñc träṅkäṣ hai paṭṭinī kuc-ne täm

smile-ptcp face-ins lay.woman speak.npst.act.3sg oh Paṭṭinī what.acc-comp dem
weñāṣt gautami lāṃts ṣñi [a7] /// can-äkk ats taṃdo
speak.pst.act.2sg Gautami queen own dem.acc.sg-emp ptcl dem
träṅkäm-ciio || capiccen-aṃ ||
say.pst.act.3sg-2sg C.tune-loc
‘with a (smiling face), the lay woman says: “Oh, Paṭṭinī, what did yousg say? Queen Gau-
tamī, (with her) own (hands) …” [Paṭṭinī speaking to the lay woman:] ”... precisely this
indeed. I am telling this to yousg . [In the C-tune] …’
(YQ III.4a7; trans. based on Ji, Winter, and Pinault 1998: 161; prose)
In the following examples (6.17), (6.18), and (6.19), the function of the PCs is unambiguous as they
accompany the secondary case marker -ac, marking allative. In these examples, the cataphoric
pronominal DOs, representing a theme, are not overtly expressed. The PCs all index the addressee
IOs rather than the theme DOs.
(6.17) [TA] -änn-anac = IO (addressee) of the transitive verb träṅkṣ- ‘(s/he) speaks (to X)’
s· [b4] /// m· mrāc tsito-r-äṣ träṅkṣ-ännan-ac pracar n· – – te

top touch.ptcp-nmlz-abs speak.npst.act.3sg-3sg-all brother
yatār himavant ṣul – – ||
do.npst.mid.2sg Himavant mountain
‘Having touched (hisi ) head, (hej ) said (this) to himi : “Oh brother, yousg do … the mount
Himavant …”
(A144b4; prose?)
160
(6.18) [TA] -nn-anac = IO (addressee) of the transitive verb weñṣ- ‘(s/he) will speak (to X)’.
|| tm-äṣ nārade riṣak weñā-nnan-ac [b5] ///

dem.abl Nārada sage say.pst.act.3sg-3sg-all
‘Then, the sage Nārada said (this) to him: “…”’
(6.19) [TA] -äṃn-anac = IO (addressee) of the transitive verb (we)ñār- ‘(they) spoke (to X)’.
śemäl-yo nu ya(-tsi) wätkäś täm-yo puk yas ña[a5](rey-aṃ ///

goat-ins conj do-inf command.npst.act.2pl dem-ins all 2pl hell-loc
we)ñār-äṃn-atac || ānä(nda)rśn-aṃ ||
speak.pst.act.3pl-3sg-all A-meter-loc
‘But youpl order to make (a sacrifice) with a goat. Therefore, youpl all … in the hell. (They)
said (this) to him/her. [In the A-tune:] “…”’
(A95a5; trans. based on CEToM; prose)4
Finally, in the following example (6.20), TA√āks- ‘announce, proclaim’ carries the second-person
singular PC -ci, representing the IO (addressee). This example does not contain any overt DO.
(6.20) [TA] -äṃ = uncertain

Function: IO (addressee) of a ditransitive
[a2] /// k waltsu(rā ā)kṣ(i)ñam-ci || taruṇadivāk(ar-aṃ ||) ///

briefly instruct.subj.act.1sg-2sg T.meter-loc
‘… I will briefly tell (this) to yousg . [In the T-tune:] …’
(A400a2; prose)
As summarized in table (6.4), when the IO and DO of a ditransitive verb are both pronominal, PCs
consistently index the IO. The examples contained an animate human IO (goal, source, recipient,
4. Cf. A96b3 śemälyo talke yatsi wätkseñc ‘They order to make a sacrifice with a goat.’ (we)ñār-äṃn-atac is to be read
as (we)ñār-äṃn-anac (Sieg and Siegling 1921: 55 n. 2).
161
or addressee) and an inanimate DO (theme). The DOs were all third-person. Some verbs of com-
munication accompanied the secondary case marker of allative, unambiguously representing the
addressee IO.
PC IO DO
TA√e|wä(s)? - ‘give’
(6.10) paṣ-äm pl pl ‘the Pretas’ 3sg (pro) ‘food’
(6.11), (6.12), (6.13) et-ñi 1sg 1sg ‘Nirdhana’ pl (pro) ‘500 denars of money’
TA√sumā- ‘take away, deprive of ’
(6.14) psumār-ñi 1sg 1sg ‘Nanda’ 3sg cam ?
TA√kälā- ‘lead, bring’
(6.15) klāte-ñi 1sg 1sg ‘King Prasenajit’ 3sg (pro) ‘water’
TA√träṅk|we-ñ- ‘say, speak’
(6.16) träṅkäm-ci 2sg 2sg ‘Paṭṭinī’ 3sg taṃ ‘that’ [n.sg]
(6.17) träṅkṣ-änn-anac 3sg-all 3sg … 3sg (pro) ‘(the following sentence)’
(6.18) weñā-nn-anac 3sg-all 3sg … 3sg (pro) ‘(the following sentence)’
(6.19) weñār-äṃn-anac 3sg-all 3sg … 3sg (pro) ‘(the following sentence)’
TA√āks- ‘announce, proclaim’
(6.20) ākṣiñam-ci 2sg 2sg … 3sg (pro) ‘(the following sentence)’
Table 6.4: Summary of the examples in TA
6.2.3 Interim summary
The examples in the two previous subsections showed that when the IO and DO of a ditransitive
verb are both pronominal, PCs consistently represent the IO in TA and TB. There was no exam-
ple such as (6.21), in which a finite verb accompanies a PC and an independent pronoun in the
genitive-dative case (TA cami, TB cwi) represents the IO.
(6.21) Unattested hypothetical examples of TA and TB
i. [TA] †… camiio … träṅkäm-äṃdo ‘I will tell it to him’.
ii. [TB] †… cwiio … aksau-nedo ‘I will tell it to him’
Adams (2015: 21) made an important observation in his handbook regarding the distribution of
the Tocharian PCs. He noticed this asymmetric distribution of the PCs. He described that if both
IO and DO are pronominal, “it seems that the IO is favored” to surface as a PC in TB (Adams 2015:
21). Unfortunately, he did not elaborate on his observation or provide any supporting example.
As I have shown in the previous subsections, the distribution of the PCs in TA and TB is more
restricted. If ditransitive verbs carry a PC, and if the verbs’ IO and DO are both pronominal, the
PC consistently indexes the IO.
162
Indeed, PCs may represent the DO (theme) of a transitive verb, as observed in Chapter 3. In such
a case, however, the host of the PC is either a monotransitive (6.22) or a ditransitive verb, whose
IO is non-pronominal (6.23 and 6.24 in TB, and 6.25 and 6.26 in TA).
(6.22) [TB] -me = DO (theme) of a monotransitive
ku se tänmästrä sässuwa ¦ piśaka wī wakicceṃ (:)

rel.m.nom.sg be.born.npst.mid.3sg son.pl 50 2 distinguished.acc.pl
mā aiśtär-medo mā lkān-medo ¦
neg know.npst.mid.3sg-pl neg see.subj.act.3sg-pl
‘Whoever begets 52 distinguished children [14a] does not know them (if) he does not see
them.’
(B255b5; verse; [7¦7]×4)
(6.23) [TB] -ñ = DO (theme) of a ditransitive with a non-pronominal IO (inanimate goal)
arai srukalyñe ¦ ci-sa nta kca mā prāskau :

oh death 2sg-perl indf any neg be.afraid.npst.act.1sg
pontas sruke-lle ¦ kā ñiś ṣeske tañ prāskau :

all.gen.pl die-gdv why 1sg alone gen.2sg be.afraid.npst.act.1sg
s= ārai ñi palsko ¦ cisa prāskau pon preken-ne :

dem.m.nom.sg oh gen.1sg thought 2sg-perl be.afraid.npst.act.1sg all time-loc
twe ṅke kalatar-ñdo ¦ apiś wärñai nrey-enta-neio :

nom.2sg then lead.subj.mid.2sg-1sg Avīci beginning.with hell-pl-loc
‘O death, I do not fear anything but you: [1a] all have to die, why would I alone fear you?
[1b] Oh, this is my thought: Because of you, I am fearing at all times, [1c] since you will
bring me to the hells, including the Avīci. [1d] ’
(B298a1; verse; [5¦7]×2 + [5¦8]×2)
163
(6.24) [TB] -me = DO (theme) of a ditransitive with a non-pronominal IO (animate goal)
ylaiñikte bramñikte-śio mant ṣerpsa-medo weñā-me-ś

Indra.god Brahma.god-all then lead.pst.act.3sg-pl speak.pst.act.3sg-pl-all
‘Then, the god Indra guided them (= Nānda and Nandābala) to the god Brahma, [and] said
to them:’
(B107a9; prose)
(6.25) [TA] -äṃ = DO (theme) of a ditransitive with a non-pronominal IO (inanimate goal)
tsopatsäṃ ske spaltäk-yo ¦ tskāt cam (wa)ṣ(t-äṣ pättāṃñkät

great.acc.sg zeal effort-ins take.out.pst.mid.3sg dem.acc.sg house-abl Buddha.lord
[a7] :)
kᵤl(e)yaṃ pältsäk cacräṅku ¦ mrosäṅkā-tsi mā

woman-loc spirit be.attached.caus.ptcp.m.nom.sg feel.disgust.npst-inf neg
nwiññāt :
bear.impf.mid.3sg. (?)
wāt-äṃ käṣṣī – – – ¦ (antuṣ wāt-äṃ ñaktas-ac :)

lead.pst.mid.3sg-3sg teacher thereupon lead.pst.mid.3sg-3sg god.acc.pl-all
wāt-äṃdo antuṣ ñarey-aṃio ¦ kuppre ontaṃ mroskat

lead.pst.mid.3sg-3sg thereupon hell-loc if ever feel.disgust.pst.mid.3sg
säm : (100-00-7)
dem.m.nom.sg
‘With great zeal (and) effort, the Buddha-lord took him (= Nanda) out of the house (i.e.,
converted him into a monk). [107a] (He who has) attached (his) spirit to a woman did not
bear renouncing. [107b] The teacher led him … then, he led him to the gods. [107c] (But)
only when he led him into hell did he renounce. [107d] ’
(A222a75 ; verse; [7¦7]×4)
(6.26) [TA] -äṃ = DO (theme) of a ditransitive with a non-pronominal IO (animate recipient)
5. A239 a5–6 attests a parallel passage: [a5] tso(patsäṃ ske spa)ltäkyo ¦ tskāt cam waṣtäṣ pättāṃñkät «:» kᵤleyaṃ
päl(tsäk) /// [a6] (¦ a)ntuṣ w(ā)t-ä(ṃ ñaktas-ac : wāt-äṃ antuṣ ña)reyaṃ ¦ k(u)pre o(ntaṃ mroskat säm :) ///
164
putti(śpar-ṣiṃ ā)kāl-yo kāpñe āriñc-ṣinäs sewās
Buddha’s.dignity-adjz.acc wish-ins love heart-adjz.m.acc.pl son.acc.pl
pr(a)mn(e el) [b2] wäs •
brāhmaṇa.gen gift give.pst.act.3sg
cem śkaṃ l(ālaṃṣk) · ñ su k(-aṃ) śāśoṣ – – – lko – – –

dem.m.nom.pl conj tender pleasure-loc live.ptcp.m.nom.pl
(ko)ṣt-l-une-yā ṣkārā luksa-māṃ triśkās käntwās-yo
strike-gdv-nmlz-perl backwards be.illuminated.caus-ptcp ? tongue.pl-ins
pācar [b3] (tä)kwāṣānt nā – – (kāp)ñ(e) āriñc pācar kᵤyal (täm ya)kṣeio śwā-(ts)i
father ?-ptcp(?) love heart father why dem Yakṣa.gen eat-inf
eṣṣ-ämdo
give.npst.act.3sg-pl
‘With the desire of (attaining) the Buddha’s dignity, (he) gave (his) dear (and) beloved sons
to the Brahmin as a gift. And they (who) lived in pleasure … tender … with striking back …
with glowing triśkās tongues, the father …. täkwāṣānt … (our) father (who is) dear (to our)
heart, why does (he) give us to the Yakṣa as a food?
(A356b3; prose?)6
Example (2.14), repeated here as (6.27), also contains a ditransitive predicate whose IO is non-
pronominal. This example is illustrative if one compares it with (6.7), repeated here as (6.28).
(6.27) (= 2.14) [TB] -ne = DO (theme) of myāskawa- ‘I exchanged X with somebodygenitive-dative ’
sū weña upanandi «†·e»io myāskawa-nedo tum(eṃ)

dem.m.nom.sg say.pst.act.3sg pn.gen exchange.pst.act.1sg-3sg then
cai ostaññi nāksa[b3]nte-ne skarāre-ne
dem.m.nom.pl housemate.pl blame.pst.mid.3pl-3sg threaten.pst.act.3pl-3sg
‘Hej said: “I traded it with Upananda.” Thereupon, these housemates blamed himj [and]
threatened himj .’
(PKAS18Ab2; prose)
(6.28) (= 6.7) [TB] -ñ = IO (beneficiary) of myāskasta- ‘yousg exchanged’
6. For (täm ya)kṣe, see Thomas (1954: 726).
165
tumeṃ su upanandeṃn-meṃ kampāl päst«†ä» ññaṣṣi •

pn neg give.pst.act.3sg-3sg dem speak.pst.act.3sg-3sg
myāskasta-ñio mā aiskau-cä
‘Then he demanded the cloak back from Upananda. (But) Upananda didn’t give [the cloak]
back to him. He said this to him: “Yousg traded [it] with me. I’m not giving [it] to yousg .”’
Both examples contain the same preterite stem myāska- of the root √mäsk- ‘(ex)change,’ carrying
a PC. While the PC in (6.27) represents the DO, the PC in (6.28) does not index the DO but the IO.
The IO of the latter is pronominal, while that of the former is not. Table 6.5 summarizes these
examples.
IO (beneficiary) DO (theme) PC
(6.27) upanandi pronominal; 3sg -ne [3sg] = DO
(6.28) pronominal; 1sg pronominal; 3sg -ñ [1sg] = IO
Table 6.5: Summary of (6.27) and (6.28)
Likewise, the ditransitive verb aiskau-c in (6.29) contrasts with (6.7), repeated here as (6.30). The
PC in (6.30) represents the IO. In contrast, the PC in (6.29) does not index the IO but the DO.7
Table 6.6 summarizes these examples.
(6.29) -cä = DO (theme) of aiskau- ‘I give X to somebodygenitive-dative ’
walo weṣṣäṃ larekka brāhmaṇetsio āyor aiskau-cdo ||

king speak.npst.act.3sg dear.voc brahmin.gen.pl gift give.npst.act.1sg-2sg
‘The king speaks: “My darling! I give yousg to the brahmins as a gift.”’
7. Peyrot (2017: 634) writes, “aiskau-c would normally mean ‘I give [it] to you’, but theoretically possible ‘I give
you [to him]’ is also attested”. However, I could not find any example of aiskau-c accompanying a pronominal (overt
or covert) DO (cf. 6.29 where a DO is non-pronominal).
166
(6.30) (= 6.7) -cä = IO (recipient) of aiskau- ‘I give’
tumeṃ su upanandeṃn-meṃ kampāl päst«†ä» ññaṣṣi •


pn neg give.pst.act.3sg-3sg dem speak.pst.act.3sg-3sg
myāskasta-ñ mā aiskau-cäio
‘Then he demanded the cloak back from Upananda. (But) Upananda didn’t give [the cloak]
back to him. He said this to him: “Yousg traded [it] with me. I’m not giving [it] to yousg .”’
IO (recipient) DO (theme) PC
(6.29) brāhmaṇets pronominal; 2sg -c [2sg] = DO
(6.30) pronominal; 2sg pronominal; 3sg -cä [2sg] = IO
Table 6.6: Summary of (6.29) and (6.30)
We have observed that PCs may index the DO when the verb is monotransitive or ditransitive with
a non-pronominal IO. In contrast, when the IO and DO are both pronominal, PCs never represent
the DO; they consistently index the IO in TA and TB.
6.3 Analysis
The previous subsections showed that when the IO and DO of a verb are both pronominal, PCs
consistently index the IO. This observation is unexpected if I follow the view that PCs arbitrarily
choose an argument it represents (hypothesis 1 of Table 6.1). Therefore, I follow hypothesis 2
and think that there is an underlying cause that produces this asymmetric pattern. There are
several ways to formalize such machinery, but I use the one developed in Chapter 5 as it finds
independent motivation.
In Chapter 5, I built a Tocharian syntactic structure in a bottom-up fashion (6.31). The root first
merges with the verbalizer and introduces a nominal expression that serves as the DO with a
theme role. The resulting structure may merge with another functional head (high-)Appl(icative),
167
introducing another nominal expression as the IO with various thematic roles such as goal, source,
recipient, beneficiary, or addressee (Marantz 1993; Harley 1995; Alexiadou 2003; Miyagawa and
Tsujioka 2004; Pylkkänen 2008; Lochbihler 2012; Hamilton 2017; Despić, Hamilton, and Murray
2019, among others). The resulting structure then merges with another functional head Voice,
which looks for the value of person and number features in the c-commanding domain, and
copies the closest one (Minimal Link Condition; Chomsky 1995). When the IO and DO of a di-
transitive verb are pronominal, Voice always agrees with the IO since it searches the features
in a top-down fashion and finds the IO before the DO. This analysis accounts for the empirical
distribution of the PCs observed in the previous section.
(6.31) Hierarchical structure
Voice
DPio
Appl
DPdo
V Root
This analysis also yields a firm prediction. Since Voice, which looks for a pronoun, always finds
the IO before the DO, PCs cannot represent or double the possessor of the DO when the IO is
pronominal.
(6.32) Prediction 1
PCs never represent the possessor of the DO when the IO is pronominal. Examples such
as (6.32i) should be absent.
i. [TB] Hypothetical example that should be absent:

†upanande tñiio kampāldo aiṣṣäṃ-neposs-of-do ‘Upananda gives his garment to yousg .’
168
(6.33) Prediction 2
PCs never double the DO when the IO is pronominal. Examples such as (6.33i) should be
absent. In contrast, we should be able to find a PC doubling the pronominal IO (6.33ii).
i. [TB] Hypothetical example that should be absent:

†upanande tñiio cwi kampāldo aiṣṣäṃ-neposs-of-do ‘Upananda gives his garment to yousg .’
ii. [TB] Hypothetical example that should be found:

†upanande tñiio cwi kampāldo aiṣṣäṃ-cio ‘Upananda gives his garment to yousg .’)
The examples that predictions (6.32i) and (6.33i) predicted not to exist were absent in our corpus.
However, since these predictions are about negative evidence, the lack of examples might be
due to mere chance. Therefore, what is important to us is the examples that prediction (6.33ii)
predicts to exist. I show below that this prediction is borne out.8 When we find doubling of a
nominal expression by a PC, if the IO of a ditransitive verb is pronominal, the doubling PC indexes
the IO. In A213 b3 (6.34), the plural PC -äm does not double the demonstrative pronoun caṣ ‘this,’
nor the DO pärklune ‘question’ (theme), but the IO yasäṃ ‘to youpl ’ (addressee).9
8. Chapter 7 will discuss the precise function of clitic doubling in TA and TB and semantic-pragmatic conditions
that restrict doubling.
9. We find examples in which a PC represents the IO, and a demonstrative pronoun attributively modifies the DO
(6.d, 6.e). One might argue that the PC choosing the IO in such examples reflects an underlying syntactic structure.
However, PCs never represent a demonstrative pronoun that attributively modifies a noun (cf. Meunier 2015: 136–9),
and therefore, we cannot use such examples to argue for an underlying hierarchical structure.
(6.d) [TB] PC = IO (source); Demonstrative pronoun ceu attributively modifies the DO kampāl ‘cloak’
upanande ceu kampāldo [b2] yaṣṣāte-ne-meṃio mā wsā-ne •
pn dem.m.acc.sg cloak beg.pst.mid.3sg-3sg-abl neg give.pst.mid.3sg-3sg
‘Upananda begged that cloak from him, [but] he did not give [it] to him.’
(PKAS18Ab1-2)
(6.e) [TA] PC = IO (source); Demonstrative pronoun caṣ attributively modifies the DO klop ‘suffering’
paṣ-äm śwā-tsi paṣ-äm nātäk ¦ pwikā-m klop
give.imp.act.2sg-pl eat-inf give.imp.act.2sg-pl lord disappear.caus.imp.act.2sg-pl suffering
caṣ k(aśśiñ was 70-8)
dem.m.acc.sg hungry.nom.pl nom.1pl
‘[The Pretas speaking to Koṭikarṇa]: “Give us food, give us, oh lord, remove this suffering from us; (we [are]
hungry”. [78d]’
(A340a4; trans. based on CEToM; verse; [5¦5¦5¦5] + [8¦7¦7] + [5¦5] + [8¦7])
169
(6.34) [TA] PC = Doubling of the IO (of a ditransitive predicate with a pronominal IO)
/// (kupre)-n(e) säm ya(s-ä)ṃio (caṣ pe)nu pärklunedo • sne

if-comp dem.m.nom.sg 2pl-gen dem.acc.sg also question without
(tä)ṅk-l-une atäṅkät wätkāṣṣ-ämio cam yas wäṣpā waṃ
hinder-gdv-nmlz unhindered answer.subj.act.3sg-pl dem.acc.sg nom.2pl truly
///
‘(Bādhari speaking to his disciples:) “If he also answers this question to youpl without
hindrance and without hesitation, youpl indeed … him.”’
(A213b3;10 prose)
In (6.35), the second-person singular clitic -c doubles the pronominal IO cī ‘yousg ,’ which is the
beneficiary of lakle + √yām- ‘to make suffering for X; torture X.’ This PC does not double the DO,
which is third-person singular.
(6.35) [TB] PC = Doubling of the IO (of a ditransitive predicate with a pronominal IO)
snai kkarūṃ cai onolmi ¦ amaukacci yolo-sa •

without compassion dem.m.nom.pl living.being.nom.pl unceasing.nom.pl evil-perl
saim pärmaṅk cī śaiṣṣe-ntseio ¦ lakle räskredo [a4]

protection hope acc.2sg world-gen suffering bitter
yāmṣiyeñ-cio •
make.impf.act.3pl-2sg
‘Without compassion, these living beings, (who are) unceasing with an evil (thought), [3a]
harshly tortured you (lit. ‘made bitter suffering [for] you’),11 the protection (and) hope
of the world. [3b] ’
(B231a4; trans. based on Thomas 1957: 65; verse; [7¦7] × 4;
10. This passage corresponds to YQ II.5 a7-8: kupre-ne säm yasäṃ caṣ penu pärklune [a8] (• sne täṅklune atäṅkät
wätkāṣṣ-äm cam yas wäṣpā wätkālts tämne) w(ä)knā kakmunt puk knānmānänt ptāñkät pkärsäs ‘If he also (answers)
this question to youpl (immediately and without hesitation, then youpl are indeed) to recognize him surely as the
Tathāgata and the all-knowing Buddha-god’ (trans. based on CEToM).
11. When TB √yām ‘to do, make’ takes lakle ‘suffering’ as the DO, the resulting predicate means ‘to do suffering to
Xacc ; torture Xacc ’. Cf. kṣānti ‘forgiveness’ with √yām ‘to do, make’, meaning ‘to do forgiveness to Xacc ; forgive Xacc ’
(e.g., B34 a5 yāmṣa cauacc kṣānti ‘[S/he] did forgiveness [to] him; [S/he] forgave him’).
170
The following example contains a pronominal possessor, semantically associated with the IO. In
(6.36), the third-person singular PC -äṃ seems to double either ka(pśi)ññaṃ ‘in the body,’ which is
the IO (location) of the verb, or, more likely, cami ‘his,’ which refers to an embryo and represents
the possessor of the IO. The PC in this example cannot double the DO as the DO is third-person
plural (wu lotas ‘two holes’).
(6.36) [TA] PC = Doubling of the possessor of the IO (of a ditransitive predicate with a pronom-
inal possessor of an IO)
kus-ne cam-i āñc ka(pśi)ññ-aṃio wu lotasdo ruseñc-äṃposs-of-io

rel-comp dem.m-gen down body-loc two hole.acc.pl open.npst.act.3pl-3sg
ṣom āsu wesis wcaṃ lyī wesi ///
one.acc dry.ptcp.m.nom.sg excrement.gen second.acc wet excrement
‘(Those) who open two holes in the lower (part of) his body, one for dry excrement (and)
the other for wet excrement …’
(A150b6; prose? )
As I predict, PCs doubling the DO are absent when the IO is pronominal (6.33i). I find doubling of
the DO and the possessor of the DO when the verb is (1) monotransitive or (2) ditransitive with
a non-pronominal IO. In examples (6.37) and (6.38), monotransitive verbs accompany a PC and
show the doubling of the DO and the possessor of the DO, respectively.
(6.37) [TA] PC = Doubling the DO (of a monotransitive verb)
kaśśi yokañi pälkāt cesäm amoktses kātse käly-māṃ

hungry thirsty see.pst.mid.3sg dem.m.acc.pl artisan.acc.pl near stand-ptcp
cesm-äk puk śtwar śälkās poke-yo wa(ltsu-r-ä)[a2]ṣ poñcäs
?
dem.m.acc.pl-emp all four altogether paw-ins crush.ptcp-nmlz-abs all.acc.pl
kosā-m tāpa-m śkaṃ lo
kill.pst.act.3sg-pl eat.pst.act.3sg-pl conj ptcl
‘being hungry and thirsty, (the lion) saw these artisans standing nearby. Crushing those
very four altogether? with (his) paw, (he) killed them and ate them all up.’
171
(6.38) [TB] PC = Doubling the possessor of the DO (of a monotransitive verb)
— — — — — ¦ — — —)[a2]-mpa tsälpāre :
…-com be.free.pst.act.3pl
pelaikne-ṣṣai tañ ¦ kektseñ wato wināskau-c 40-7 ||
gen.2sg body.acc.sg again? honor.npst.act.1sg-2sg
law-adjz.f.acc.sg
‘… were free with … [47c] I again? praise your body of the law. [47d] ’
(B244a2; trans. based on CEToM; verse; [5¦7]×4)
In examples (6.39), (6.40), and (6.41), there is a ditransitive predicate accompanying the non-
pronominal IO and a PC. These examples show the doubling of the IO, the DO, and the possessor
of the DO, respectively.12
(6.39) [TB] PC = Doubling of the IO (of a ditransitive predicate with the non-pronominal IO)?13
(tu-meṃ pūdñäkte)-śio śārsare-ne ceᵤ wäntredo :

dem-abl Buddha.lord-all know.caus.pst.act.3pl-3sg dem.m.acc.3sg matter
‘(Thereupon) they announced this matter to (the Buddha lord).’
(PKNS22b5; verse; [5¦7]×4)
(6.40) [TB] PC = Doubling of a DO (of a ditransitive predicate with a non-pronominal IO)
– – – – – (ṣa)ñ k(e)wändo ¦ śakātai-saio kalṣtär-medo ¦ ṣñār wepeṃ-ś

own cow.pl stick-perl goad.npst.mid.3sg-pl respective corral.pl-all
aśan-me :
lead.npst.act.3sg-pl
tu-yknesa ktsaitsñe srūka[a4](lñe) ¦ śaul kältsenträ wnolmen-tso ¦ ṣañ
thus old.age death life goad.npst.mid.3pl being.acc.pl-gen own
kalymi-ś aken-ne :
direction-all lead.npst.act.3pl-3sg
“(Just as a herdsman) goads (his) own cows with a stick, and leads them to their corrals,
[89a] in this way old age and death goad the life of the beings, and lead it to its destination.
12. The IO in (6.40) and (6.41) has an instrument role, and seems to be an adjunct of a monotransitive verb.
13. In view of (6.34), the third-person singular PC -ne in (6.39) is more likely to double the IO than the DO.
172
[89b] ”
(B3a3; trans. based on CEToM; verse; [8¦7¦6]×2 + [4/5¦4/5] + [7¦6])
(6.41) [TB] PC = Doubling of the possessor of the DO (of a ditransitive predicate with a non-
pronominal IO)
lyam= ānande keni-sa ¦ (a)[b5]lyine-sa antapiio :

sit.pst.act.3sg Ānanda knee.du-perl palm.du-perl both
pudñäkte-ntse kektseñodo ¦ klawāte-ne lyawā-ne :

Buddha-gen body touch.pst.mid.3sg-3sg rub.pst.act.3sg-3sg
‘Ānanda sat on the knees. With both palms he massaged the body of the Buddha and
rubbed it.’
Doubling of
Doubling of Doubling of
the possessor
the IO the DO
of the DO
Monotransitive — Yes (6.37) Yes (6.38)
Ditransitive predicate
Yes? (6.39) Yes (6.40) Yes (6.41)
with the non-pronominal IO
Ditransitive predicate
Yes (6.34, 7.88) No No
with the pronominal IO
Table 6.7: Summary of Prediction 2
Table 6.7 summarizes the examples discussed. Clitic doubling of the DO and the possessor of the
DO are limited to monotransitives and ditransitive verbs with the non-pronominal IO.
Furthermore, the following example (6.42) from the Ṣaḍdanta-jātaka nicely conforms to my anal-
ysis. If wotka-ñi, restored by Sieg (1952: 15 n. 11) to fill the gap of three akṣaras, is correct, the
first-person singular PC in this example indexes the agent of an infinitive, while the infinitive’s
IO is also pronominal.
(6.42) [TA] PC = Agent of an infinitive
173
tmäṣ cesäm āṅkaräs lāntse suknā-māṃ träṅkäṣ
then dem.m.acc.pl tusk.pl queen.gen present-ptcp speak.npst.act.3sg
cesäs śkaṃ āṅkaräs tñiio es-si (wotka-ñi ||)

dem.m.acc.pl conj tusk.pl gen.2sg give-inf order.pst.act.3sg-1sg
‘Then, presenting the tusks to the queen, (the hunter) says: “And these tusks (he [= the
elephant] ordered me) to give to you.”’
(A77a4; prose?)
PCs may undergo clitic climbing to represent the IO or DO of an infinitive. However, I found
no examples where a PC serves as the IO or DO of the infinitive while an independent pronoun
represents the agent of the infinitive (6.43).
(6.43) [TA] Hypothetical unattested example:

†cesäs śkaṃ āṅkaräs ñi es-si wotka-ci
‘And these tusks, (he) ordered me to give to you’.
My analysis predicts this asymmetry since the licensor of a PC (Voice in the matrix clause) finds
the agent of an infinitive (introduced by another Voice) before the IO and DO of the infinitive, as
shown in (6.44).
(6.44) Hierarchical structure
174
Voice …
(licensor)
…
DPagent
Voiceinfinitive
DPio
Appl
DPdo
V Root
Likewise, the first-person singular PC in the following example does not represent the IO (ad-
dressee) (not ‘orders (someone) to ask a question to me’) but the agent of the infinitive (‘orders
me to ask a question’).
(6.45) [TA] PC = Agent of an infinitive
/// tmäṣ ṣu skamat prakäs-si wätkäṣ-ñi •

then hither always ask.a.question-inf order.npst.act.3sg-1sg
‘Then, she orders me to ask a question here repeatedly.’
(YQ III.4b7; trans. based on Ji, Winter, and Pinault 1998: 161; prose)
To summarize, I have shown that the analysis developed in the previous chapter accounts for the
asymmetry in the distribution of the Tocharian PCs. In my syntactic model, the licensor of a PC
looks for a pronoun in a top-down fashion and licenses the closest one as the PC. It accounts for
the empirical distribution of the Tocharian PCs, which always represent the IO when the IO and
the DO are both pronominal. Furthermore, my analysis predicts there is a gap in the data (6.46).
175
(6.46) Predictions
i. PCs cannot represent the possessor associated with the DO when the IO is pronominal.
ii. PCs cannot double the DO or the possessor associated with the DO when the IO is
pronominal.
iii. PCs may double an IO when the IO is pronominal.
I have found that a PC doubling the DO or the possessor of the DO is limited to the cases where
the IO of a verb is non-pronominal (Table 6.7). However, there is one potential counterexample to
(6.46i): PKAS 8C a5 (6.47) contains weñau-ne, whose third-person singular PC seems to represent
the possessor of the DO upacar ‘practice, method’ (cf. Skt. upacāra-), while the verb’s IO (‘you’?)
is not overtly expressed.
(6.47) PC = Possessor of the DO (theme) of a ditransitive verb with the pronominal IO(addressee)?
se vi(j w)eṣ-le •
dem.m.nom.sg spell speak.npst-gdv.m.nom.sg
arañc empreṃ aunaṣ-le

heart true begin.npst-gdv.m.nom.sg
warke-sa po ekaññe-sa kekenu maṇḍāl

garland-perl all possession-perl be.filled.ptcp.m.nom.sg maṇḍala
yamaṣ-le •
make.npst-gdv.m.nom.sg
kurkalä tuñe • viciträ pyāpyai maṇḍāl-ne

bdellium perfume Citrullus.colocynthis flower.acc.sg maṇḍala-loc
taṣa-le •
put.npst-gdv.m.nom.sg
te-yäkne-sa se sātäṃ star-ne14 ||

dem.n-manner-perl dem.m.nom.sg success cop.npst.3sg-3sg
ñake upacar weñau-ne ||

now method speak.subj.act.1sg-3sg
‘This charm [has] to be spoken. It [has] to be started with determination [lit. (One’s) heart
(has) to be started (as) true]. (One) should make a maṇḍala provided with a garland [and]
all (sorts of) wealth. (One) should put perfume of bdellium [and] a flower of citrullus
176
colocynthis [cf. Skt. vicitra-] in the maṇḍala. In this way, this (spell) is complete [lit. this
is the accomplishment (cf. Skt. sādhana-) of it]. Now, I will speak about its use.’
(PKAS8Ca5; trans. based on CEToM; prose)
The third-person singular PC -ne in weñau-ne refers to vij ‘the charm.’ The function of the PC is
to represent an (objective) genitive associated with upacar ‘practice, method, usage’ (i.e., upacar
weñau-ne ‘I will speak its [= the charm’s] usage’). PKAS 8C is a medical/magical text, and although
the verb is in the first-person singular, no overt addressee is mentioned in the text. Therefore,
weñau- in this example may be a simple monotransitive verb rather than a ditransitive accompa-
nying a referential null object as the IO.
6.4 Implications and further questions
My analysis has some implications for our understanding of the Tocharian syntax: it enables us
to identify when TA and TB have a referential null object as the DO. In (6.48), where the IO and
DO of the verb are both pronominal, the third-person singular PC -ne represents the IO, and the
pronominal DO lacks overt phonetic realization.
(6.48) [TB] PKAS18Ab3

sū mā wsā-ne
dem.m.nom.sg neg give.pst.act.3sg-3sg
✓ ‘He did not give itnull object to him.’
× ‘He did not give it to himnull object .’
It allows us to identify the condition under which TA and TB license a referential null object. It
is a topic recently studied for some ancient Indo-European languages (cf. Keydana and Luraghi
2012 on Vedic and Greek, Inglese, Rizzo, and Pflugmacher 2019 on Hittite), but not for Tocharian.
Based on the observation made in this chapter, we may now collect the TA and TB examples in
which the DO of a verb is a null object.
14. This copula carries a third-person singular PC. However, its function is not clear to us (Cf. CEToM: “this [pro-
cess] of it is a success”).
177
Finally, there is one point characteristic of this analysis. Our analysis disregards the animacy
or humanness of an argument. In other words, no matter whether the IO is animate or not, or
whether the DO is animate or not, in our analysis, PCs always represent the IO when the IO is
pronominal.
Ditransitive predicates of TA and TB usually organize their argument structure following Silver-
stein’s (1976) animacy hierarchy (6.49). All the examples we observed represent arguments at
the higher levels of the hierarchy as IOs and those at the lower levels as DOs (Tables 6.3 and 6.4).
(6.49) Silverstein’s (1976) animacy hierarchy

1st/2nd person » 3rd person pronoun » proper noun » human » animate » inanimate
One might wonder whether PCs still represent the IO when the DO of a ditransitive is at a higher
level of the hierarchy than the IO. My analysis predicts that PCs represent the IO even if a ditran-
sitive verb has, for example, an inanimate IO and an animate DO (6.50). In contrast, examples
such as (6.51) should be absent.
(6.50) [TB] Hypothetical example that should be found (DO = animate; IO = inanimate)
†brāhmaṇi cido aken-neio
brahmin.nom.pl acc.2sg lead.npst.act.3pl-3sg
‘The brahmins guide yousg to itinanimate .’
(6.51) [TB] Hypothetical example that should be absent (DO = animate; IO = inanimate)
†brāhmaṇi cewä-śio aken-cdo
brahmin.nom.pl dem.m.sg-all lead.npst.act.3pl-2sg
‘The brahmins guide yousg to itinanimate .’
Unfortunately, my corpus did not contain any example of a ditransitive predicate where the DO is
at a higher level of the hierarchy than the IO. If PCs represent the DO in such cases, I will need to
make our model more complex by stipulating that the licensor of a PC is sensitive to the animacy
or humanness of a pronominal argument.
Also, some languages have a restriction against a particular combination of phonologically weak
178
arguments of verbs (Person Case Constraint; PCC).
(6.52) Person Case Constraint (Bonet 1991; Anagnostopoulou 2005)
i. Strong PCC: the direct object has to be third person.
ii. Weak PCC: if there is a third person it has to be the direct object.
In addition to the strong and weak versions of the PCC, scholars added the following subtypes to
the typology of the PCC.15
(6.53) Further subtypes of the PCC (Nevins 2007; Anagnostopoulou 2017; Pancheva and Zu-
bizarreta 2018)
i. Me-first PCC: the direct object has to be second or third person.
ii. Ultra-Strong PCC: the direct object has to be second or third person, and if there is
a third-person argument, it has to be the direct object.16
iii. Super-Strong PCC: the indirect object has to be first or second person and the direct
object has to be third person.
PCC varieties 1>217 1>3 2>1 2>3 3>1 3>2 3>3 Example
1. Super-Strong * ✓ * ✓ * * * Kambera (Klamer 1997)
2. Strong * ✓ * ✓ * * ✓ Modern Greek (Bonet 1991)
3. Ultra-Strong ✓ ✓ * ✓ * * ✓ Classical Arabic (Nevins 2007)
4. Me-first ✓ ✓ * ✓ * ✓ ✓ Romanian (Nevins 2007; Ciucivara 2009)
5. Weak ✓ ✓ ✓ ✓ * * ✓ Catalan (Bonet 1991)
TB ? ✓ ? ✓ ? ? ✓ (Table 6.3)
TA ? ✓ ? ✓ ? ? ✓ (Table 6.4)
Table 6.8: Varieties of the Person Case Constraint (based on table 1 of Compton 2019: 595)
In my examples, the DOs were all third-person (Tables 6.3 and 6.4). One might wonder whether
Tocharian ditransitive predicates may have a third-person singular PC (-ne) with the first- or
15. I set aside the number of an argument since, as Nevins (2007) observed, there seems to be no Number Case
Constraint.
16. The combinations allowed by the Ultra-strong PCC are the intersection of the set of combinations allowed
by the Me-first PCC and that allowed by the Weak PCC. The Ultra-strong PCC is a misnomer since it is actually less
restrictive than the Strong PCC in that the former allows the combination of the first-person IO and the second-
person DO (Table 6.8).
17. “1>2” represents the combination of the first-person IO and the second-person DO.
179
second-person DO represented by an independent personal pronoun (6.54). Unfortunately, my
corpus did not contain any such example either.
(6.54) [TB] Hypothetical example (DO = 1st/2nd; IO = 3rd)

†upanande ñäś/cido aiṣṣäṃ-neio
Upananda 1sg/2sg give.npst.act.3sg-3sg
‘Upananda gives me/yousg to him.’
However, I have found a case where a PC indexes the first-person over the second-person. Exam-
ple (6.55) is a copular sentence in which tāk(ar)- ‘(they) were X, (they) became X’ connects two
nominal expressions tñi kāswoney(ä)ntu ‘yoursg virtues’ and ṣrum ‘cause.’
(6.55) [TA] -ñi = Possessor of ṣrum ‘reason’ connected by a copula

|| wäl träṅkäṣ tñi kāswone-y(ä)ntu āṣānik ṣrum
king speak.npst.act.3sg gen.2sg virtue-pl venerable cause
tāk(ar)-ñi pälkār [a2] ///
(?)
cop.pst.act.3pl -1sg see.imp.mid.2sg
‘The king speaks: “Yoursg virtues, o venerable one, have become my reason (for this).
Look! …”’
(A147a1; trans. based on Thomas 1957: 19018 ; verse?)
The PC in this example represents the possessor of ṣrum ‘cause,’ while the possessor of kās-
woney(ä)ntu ‘virtues’ is represented by an independent personal pronoun (tñi). Since two nominal
expressions connected by the copula have the same distance from the licensor of a PC (Voice),
the first-person possessor seems to take precedence over the second-person possessor in this
example.
6.5 Conclusion
This chapter discussed cases in which multiple arguments are pronominal. I showed that PCs do
not arbitrarily determine which pronominal argument to represent. Instead, I found an asym-
18. Thomas (1957: 190): “Deine Verdienste, o Würdiger, sind mir Grund [hierfür] geworden […].”
180
metric distribution of the PCs. PCs consistently represent the IO when both IO and DO are pronom-
inal in ditransitive predicates. I utilized a hierarchical model developed in the previous chapter,
which accounted for this restricted empirical distribution and offered two predictions. (1) PCs
cannot represent the possessor of the DO when the IO is pronominal. (2) PCs cannot double the
DO or the possessor of the DO when the IO is pronominal. I also found that doubling of the DO
and the possessor of the DO are absent when the IO is pronominal. Furthermore, my analysis
explains the absence of the examples whose PC represents the IO or DO of an infinitive with an
independent pronoun describing the infinitive’s agent.
181
CHAPTER 7
Clitic Doubling in Tocharian A and B
7.1 Introduction
In the previous chapters, pronominal clitics (PCs) of Tocharian A (1sg -ñi, 2sg -ci, 3sg -(ä)ṃ, pl
-(ä)m) and Tocharian B (1sg -ñ, 2sg -c, 3sg -ne, pl -me) replaced overt nominal expressions.1
The Tocharian PCs, however, sometimes cooccur with their antecedent, and in such cases, they
appear to be redundant. In (7.1), for example, the plural PC -äm represents the direct object
(theme) of the transitive verb kaśal malkam- ‘I will put X together’, although the direct object
itself is represented by the full nominal expression ce(smä)k āyäntu ‘the bones’. Likewise, in (7.2),
the third-person singular PC -ne appears to repeat uttareṃ śamaśkeṃ ‘the boy Uttara’, which is the
direct object (theme) of the transitive verb tsopaṃ ‘(the brahmin Durmukha) pokes X’.2
(7.1) [TA] Doubling of ce(smä)k āyäntu

wät träṅkäṣ näṣ nu ce(sm-ä)[b6]k āy-äntu pᵤkāk
second speak.npst.act.3sg 1sg conj dem.m.acc.pl-emp bone-pl completely
puskās-yo kaśal malkam-äm
sinew.pl-ins together join.subj.act.1sg-pl
‘The second (artisan) says: “But I will join the bones with the sinews completely.”’
(A11b6; trans. by CEToM; prose)
(7.2) [TB] Doubling of uttareṃ śamaśkeṃ

tumeṃ durmukhe brāhmaṇe uttare-«ṃ» śamaśke-ṃ kärwā-ṣṣai witsakai-sa
thereupon Durmukha brahmin Uttara-acc boy-acc reed-adjz.f.acc.sg root-perl
1. An earlier version of this chapter appeared in Onishi (forthcoming).
2. I use the term associate to refer to the nominal expression doubled by a PC.
182
räskare tsopaṃ-ne
sharply sting.npst.act.1sg-3sg
‘Thereupon the Brahmin Durmukha harshly jabs the boy Uttara with a reed root.’
Scholars have recognized this phenomenon at least since the middle of the twentieth century
(e.g., Krause 1952: 207, TEB I: 163 n. 1, Adams 2015: 149, among others).3 For example, Meunier
(2015: 139–41) noted that doubling clitics function as an anaphor which has a focalizing effect
(“anaphore focalisante”).4 According to Pinault (2008: 537), doubling of a nominal expression
by a pronominal clitic has a pragmatic function, that is, to refer to the theme of an utterance,
and doubling is partly motivated morphologically because of the frequent lack of distinction be-
tween nominative and accusative in nouns.5 In contrast, Peyrot (2017, 2019) and Adams (2015:
149) treated doubling clitics as object agreement, that is, as markers of agreement with a (direct)
object.
However, despite these analyses, it is fair to say that many questions remain unanswered. Some
outstanding questions are: What function does clitic doubling in TA and TB have? When or why
does it occur? Does it have any semantic effect? Is it subject to any grammatical or semantic re-
striction(s)? Is there any difference between clitic doubling of TA and TB? How did clitic doubling
develop in (pre-)TA and TB? This chapter focuses on the following two questions: (1) What does
clitic doubling do in TA and TB? and (2) Does clitic doubling in TA and TB have any grammatical or
semantic restriction(s)? This chapter reveals that doubling of a nominal expression by a PC indi-
cates the nominal expression is topical. We will observe that a doubled associate that undergoes
dislocation represents the primary topic. In contrast, a non-dislocated associate may represent a
3. It seems that it was not known to Schulze, Sieg, and Siegling (1931).
4. “L’emploi « focalisant » du clitique est peut-être plus difficile à cerner, mais il reste évident que si le clitique mis
pour un génitif représente, comme on le pense, un complément d’intérêt, ou dativus sympatheticus, dans certaines
phrases ce clitique est redondant par rapport à un génitif adnominal, ou par rapport à un génitif-datif ; il y a donc
un phénomène d’« anaphore focalisante » mis en jeu dans ce clitique, que sa fonction grammaticale soit identique
ou non à celle du terme anaphorisé.” (Meunier 2015: 140–1)
5. “Un pronom suffixé peut aussi référer à un complément déjà exprimé par un constituant dans la même phrase:
cette reprise a en partie une motivation morphologique, en raison de l’absence fréquente de distinction entre nom-
inatif et oblique dans les noms. Mais elle a simultanément (et probablement à l’origine) une fonction pragmatique,
pour renvoyer au thème de l’énoncé.” (Pinault 2008: 537)
183
primary or secondary topic depending on whether or not the associate is in the subject position.
In all cases of doubling, discourse participants must presuppose the existence of the associate’s
referent.
7.2 Topic and the Question Under Discussion model of discourse
7.2.1 What is a topic?
Before we examine the data, a few words are in order regarding the term topic since many schol-
ars have used it in various contexts. I follow Reinhart (1981) and others in taking a topic to be
pragmatic aboutness (Reinhart 1981, Krifka 2008, C. Roberts 2011, Matić 2015, among others).
A topic is a part of an utterance about which the utterance is meant to give information. For
example, the utterance (7.3) concerns Mary, and the topic of this sentence is Mary. In (7.4), how-
ever, an addressee is interested in knowing about Harry regarding what Mary gave to him. The
sentence topic of (7.4) is therefore not Mary but Harry.
(7.3) What about Mary? What did she give to Harry?

— [topic Mary ] gave a shirt to Harry.
(7.4) What about Harry? What did Mary give to him?

— [topic To him ] Mary gave a shirt.
(Examples based on C. Roberts 2011 [2] and Vallduví 1993: 7 [9a])
I assume that information that is mutually known to be shared by the discourse participants is
stored in the Common Ground (CG; Stalnaker 1978). The CG also contains a set of entities that
have been introduced into the discourse before (Krifka 2008: 246). According to Krifka (2008:
265), new information is not just added to the content of the CG in the form of unstructured
propositions, “but is rather associated with entities, just like information in a file card system
is associated with individual file cards that bear a particular heading.” A sentence topic corre-
sponds to the header of a file card under which new information is stored (Reinhart 1981).
184

Sentence topic of (7.3): Mary

• She gave a shirt to Harry.
• …

As summarized by Erteschik-Shir (2007: 13–5), if a topic is what a statement is about, and if one
evaluates the truth value of a statement as true or false with respect to the topic, then a topic
constituent must have a reference. Otherwise, a statement which “is about something is really
about nothing” (Strawson 1964: 116, but see von Fintel 2004 for a different view). For example,
(7.5i) fails to assign a truth value because the interlocutors do not presuppose the existence of
the referent of the king of France in the world of their discourse. In contrast, one should intu-
itively think that the statement (7.5ii) is false as the king of France is certainly not in the set of
individuals who visited the exhibition. In other words, the interlocutors must presuppose the
existence of a topic referent in order to evaluate a statement as true or false.
(7.5) Topic expressions must be referential
i. [topic The king of France ] is bald. (No truth value)
ii. [topic The exhibition ] was visited by the king of France. (False)
This assumption entails that topic referents must be specific since specific nominal expressions
are those for which a speaker presupposes the existence of a particular referent (Lyons 1999:
173).6 For example, the speaker in (7.6i) presupposes the existence of a specific kind of car (e.g.,
blue Toyota). At the same time, it is not the case in (7.6ii): the speaker does not presuppose that
there is a specific kind of supermarket nearby.
(7.6) Specificity
i. You know what? My dad bought a new car! (Specific)
ii. Excuse me. I’m looking for a supermarket. (Non-specific)
6. Specificity is a notoriously difficult term to define. For a recent overview, see von Heusinger (2019). I follow
a somewhat informal definition according to which a speaker has “a particular individual in mind” (cf. Lyons 1999:
171).
185
A topic referent must be specific, and, furthermore, its existence has to be presupposed by both
speaker and addressee at the time of the utterance.
7.2.2 Discourse-new referents may be a topic
Although most topics are discourse-old, discourse-new referents may serve as a topic. In (7.7),
there are two topic constituents: Mary and the door. In this short discourse, a sentence topic shifts
from Mary to the door, the latter of which has not been introduced to the discourse before.
(7.7) [topic Mary ] bought a car. But [topic the door ] was broken.
In this example, an addressee may presuppose the existence of a particular door from the rele-
vant nominal expression a car. This process is called bridging (Clark 1975), which enables an ad-
dressee to find a unique referent by making inferences from something that s/he already knows.7
Bridging may license a discourse-new topic as in (7.7), and it is not limited to part-whole rela-
tionships (e.g., 7.8).
(7.8) John was murdered yesterday. [topic The knife ] lay nearby. (Clark 1975: 172)
According to Wilson and Matsui (1998), discourse relevance plays a role in licensing a bridging
inference. In contrast, Asher and Lascarides (1998) argue that rhetorical connections (i.e., dis-
course coherence) between the propositions introduced play a crucial role.8
A discourse-new referent may also serve as a sentence topic when it shows generic interpretation.
For example, Japanese has a topic marker -wa, which may mark a sentence topic (Kuno 1973).9
In example (7.9), sono hon ‘the book,’ which is discourse-old, is wa-marked and serves as a topic.
In contrast, the nominal expression kuzira ‘whale’ in (7.10) is discourse-new. Still, it has a wa-
marking and serves as the topic expression of the sentence. In such cases, a topic displays generic
7. In this case, an addressee implicitly presupposes that a car has a door. Relevant linguistic expressions do not
have to be nominal; a verbal expression, or even a proposition may serve as a bridge (Hou 2015; e.g., I travelled to
Frankfurt. [topic The train ] was very full. Why do humans collaborate? [topic The answer ] lies in …).
8. Zhao (2014) reviews previous approaches to bridging.
9. Kuno (1973: 38) calls this type “thematic” -wa.
186
interpretation, or a topic referent is identifiable from a listener’s general knowledge about the
world.
(7.9) [Japanese] Sentence topic = ‘the book (that I bought yesterday)’
i. Taro: watashi-wa kinou aru hon-wo kat-ta

I-top yesterday some book-acc buy-pst
Taro: “I bought a book yesterday.”
ii. Taro: Sono hon-wa yasuku-nakat-ta

that book-top be.cheap-neg-pst
Taro: “[topic The book ] was not cheap.” (Discourse-old topic)
By (7.9i), two interlocutors set up a file card whose header is ‘the book.’ Then, the utterance
(7.9ii) updates the file card by adding pieces of information therein.

Sentence topic: the book

• Taro bought it yesterday.
• It was not cheap.
• …

In contrast, when a speaker produces (7.10) without any preceding context, kuzira does not refer
to a specific whale but whales in general (Kuno 1972: 270).
(7.10) [Japanese] Sentence topic = ‘whales (in general)’

Context: Out of blue
Kuzira-wa honyuu-doobutu desu
whale-top mammal is
‘[topic Whales ] (in general) are a mammal.’ (Discourse-new topic)

Sentence topic: whales (in general)

• They are a mammal.
• …

This chapter will show that clitic doubling always cooccurs with a topical associate in TA and TB.
187
However, doubling a nominal expression by a PC does not make the expression topical. I assume
that topicality is defined not on expressions but referents (Lambrecht 1994; Nikolaeva 2001). It
is determined in semantics and/or pragmatics and optionally realized morphologically. In other
words, doubling is a sufficient condition for topicality: a nominal expression might represent a
topic even if it lacks doubling.10
7.2.3 The Question Under Discussion model of discourse
This chapter follows the Question Under Discussion (QUD) discourse model. This framework
models discourse as a game, organized around the questions under discussion by the interlocu-
tors (C. Roberts 2012). The advantage of this model is that it may provide an unified perspective to
semantic and pragmatic phenomena including presupposition inference and information struc-
ture such as topic and focus, which have been treated separately as different phenomena (Beaver
et al. 2017).11
This model takes the goal of discourse to share information about our world regarding what it is
like, that is, to answer the Big Question What is the way things are? (Stalnaker 1978; C. Roberts 2012).
To achieve this goal, discourse participants set up several subinquiries that help to answer the
Big Question, and answer each of the questions stated either explicitly or implicitly. Under the
QUD model, discourse is structured around such immediate questions under discussion (QUDs)
(Carlson 1983).
To achieve the goal, interlocutors may choose two types of move: setup move (i.e., question) and
payoff move (i.e., assertion). The former proffers a question, which is a contextually restricted
set of propositions that are possible answers to the question (Hamblin 1973; Karttunen 1977;
Beaver et al. 2017). The latter chooses among the set of alternative propositions proffered. If the
interlocutors accept a question, it becomes the immediate QUD. If the interlocutors accept an
assertion, the chosen alternative proposition is added to the Common Ground. The CG contains a
10. Topicality is necessary but insufficient for clitic doubling. Most of the topic expressions lack clitic doubling, as
the limited number of the examples suggests (Section 7.1).
11. Of course, it is not the only possible approach. See, for example, Yokoyama’s (1986) Transactional Discourse
Model as an alternative way to analyze the data.
188
set of propositions that the interlocutors take to be true. Taking a proposition as a set of possible
world, the CG is a set of sets of possible worlds. Adding a proposition to the CG reduces the number
of possible worlds contained in the intersection of the propositions (what is called Context set).
The goal of the discourse is to reduce the Context set to contain a single possible world, namely
the actual world (C. Roberts 2012).
(7.11) Context: John, Mary, Sam, and Lucy are in the classroom.
i. QUD: Whom does John know?
ii. J (7.11i) K = { know(j, y) | y ∈ { m, s, l } }

= { John knows Mary, John knows Sam, John knows Lucy }
(7.12) John knows [focus MARY ].
The speaker first sets up the implicit QUD (7.11i), which proffers three alternative propositions
John knows Mary, John knows Sam, and John knows Lucy (excluding the union and intersection of
individuals for the sake of simplicity). Other individuals who are not in the classroom are con-
textually excluded. The speaker then asserts (7.12), which has a focal intonation on Mary. This
focal intonation helps an addressee identify the implicit QUD (7.11i). Roughly speaking, a focus
constituent corresponds to the wh-constituent of the QUD, and in this example, Mary is the focus
of this utterance. If the assertion (7.12) is accepted by the interlocutors, the (ordinary) semantic
value of (7.12), namely know(j, m) = John knows Mary is added to the CG, reducing the number of
possible worlds in the Context Set.
In addition to (ordinary) semantic value, a focus also introduces a set of alternative propositions
(called “alternative set”; Rooth 1985, 1992, 2016). For example, the focus semantic value of (7.12)
is an alternative set of the form ‘John knows y’, where y is in the domain of individuals De . The (or-
dinary) semantic value and the focus semantic value of (7.12) are (7.13i) and (7.13ii), respectively.
(7.13) (Ordinary) semantic value and focus semantic value of (7.12)
i. (Ordinary) semantic value: J John knows [focus MARY ] Ko = know(j, m)

= John knows Mary
189
ii. Focus semantic value: J John knows [focus MARY ] Kf = { know(j, y) | y ∈ De }
= { John knows Mary, John knows Sam, John knows Lucy, John knows Hiro … }
(7.14) John [focus KNOWS ] Mary.
i. J John [focus KNOWS ] Mary Ko = know(j, m)

= John knows Mary
ii. J John [focus KNOWS ] Mary Kf = { know(j, m), see(j, m), like(j, m), hate(j, m), … }
= { John knows Mary, John sees Mary, John likes Mary, John hates Mary … }
Since the focus semantic value (7.13ii) is a proper superset of the denotation of the QUD (7.11ii),
the utterance (7.12) is congruent with the QUD (7.11). In contrast, (7.14), which has a focus on
like and a focal intonation of like, induces the set of alternative propositions (7.14ii). This is not a
proper superset of (7.11ii), making the utterance (7.14) infelicitous even though it has the same
(ordinary) semantic value as (7.13i). In this way, QUDs, equivalent to discourse topics, constrain
the felicitous flow of discourse (C. Roberts 2011).
The complement of a focus is called background. For example, John knows is the background of
(7.12). A topic constituent is a proper subpart of the background of an utterance. For example,
the topic constituent John is a proper subpart of the background John knows in (7.11).
(7.15) [background [topic John ] knows ] [focus MARY ].
Not all sentences have a topic. A topic constituent may be absent if a focus domain extends to the
entire utterance. In (7.16), an utterance provides an answer to the QUD “What happened?” and
the focus domain covers the entire sentence.
(7.16) QUD: What happened?

[focus I killed this reindeer ].
There is no background in this example, and therefore, it does not have an overt topic constituent.
190
7.2.4 Secondary topic
A single sentence may have more than one topic. In (7.17ii) and (7.17iii), John is the topic of the
sentences as they update information about what happened to John.
(7.17) i. What happened to John?
ii. He married Rosa.
iii. But he didn’t really love her.

(7.17ii and 7.17iii from Dalrymple and Nikolaeva 2011: 54)
At the same time, however, (7.17iii) “also increases the addressee’s knowledge about Rosa, namely,
the fact that she was not loved by her husband John.” (Dalrymple and Nikolaeva 2011: 54) In this
case, the utterance (7.17iii) provides information regarding the relationship between the primary
topic (John) and the referent (Rosa) introduced to the discourse in (7.17ii). Both interlocutors
pragmatically presuppose her existence at the time of (7.17iii). I define a secondary topic as
“an entity such that the utterance is construed to be about the relation that holds between it and
the primary topic” (Nikolaeva 2001: 2; Dalrymple and Nikolaeva 2011: 54–7) and analyze Rosa to
be the secondary topic of (7.17iii).12
(7.17ii) (7.17iii)
QUD What happened to John? In what relation did John stand to Rosa?
Focus married Rosa didn’t really love
Topic expression/referent (primary) he/John he/John
Topic expression/referent (secondary) — her/Rosa
Table 7.1: Summary of (7.17ii) and (7.17iii)
Table 7.1 summarizes the information structures of (7.17ii) and (7.17iii). By producing (7.18iii),
the speaker updates the addressee’s knowledge regarding the relation between John and Rosa by
asserting that the former did not love the latter.
(7.18) (= 7.17)
12. According to Lambrecht (1994: 148), “[a] sentence containing two (or more) topics, […] conveys information,
about the relation that holds between them as arguments in the proposition.”
191
i. What happened to John?
ii. [topic He ] [focus married Rosa ].
iii. But [1ry topic he ] [focus didn’t really love ][2ry topic her ].
(7.18ii and 7.18iii from Dalrymple and Nikolaeva 2011: 54)
Primary and secondary topics correspond to Erteschik-Shir’s (2007: 22–3) main and subordi-
nate topics.13 A secondary topic also roughly corresponds to Vallduví’s (1993) tail, although
the former may be a shifted topic while the latter cannot. A secondary topic constituent may
be an overt object nominal expression or a referential null element (pro). A sentence with sec-
ondary topics always has a primary topic, but a sentence with a primary topic may not have a
secondary topic. As in primary topics, secondary topics are proper subparts of the background
and the current QUD.
Primary topics differ from secondary topics in pragmatic saliency: the former is more salient
than the latter. The former is the most salient topic of the utterance, and it is equivalent to a
sentence topic. In many cases, however, it is difficult to determine whether a given nominal
expression is the primary or the secondary topic of a sentence. In such cases, we follow Givón
(1983: 22) and take a nominal expression that occupies the subject position to represent the pri-
mary topic.
In this subsection, I outlined our assumptions regarding the term topic. A topic is a part of an
utterance about which an utterance gives information, and it is a proper subpart of background.
A topic constituent must be either referential, with the interlocutors presupposing its existence,
or capable of showing the generic interpretation. A sentence may have more than one topic
constituent, and the one which denotes the most salient referent is the sentence topic.
13. Focus domain of (7.18iii) excludes the secondary topic (i.e., [focus didn’t really love ] [topic her ]). In other words, a
focus domain does not has to be a syntactic constituent (Dalrymple and Nikolaeva 2011: 86). According to Erteschik-
Shir (2007), however, a (secondary) topic (her embedded topic) may be embedded within a focus domain (i.e., [focus
didn’t really love [topic her ] ] ).
192
7.3 Data
Using the CEToM database, I collected 551 TA and 608 TB examples containing a PC. I examined
whether a PC doubles an overtly expressed associate or is used just as a pronoun which substitutes
a nominal expression. It turned out that 14x (2.5%) of the TA attestations and 20 examples (3.3%)
of the TB verbs with a PC showed clitic doubling.
As these numbers suggest, doubling is quite limited in TA and TB. The question is why it is so
rare. Many of the non-doubling cases of PCs are fragmentary, and perhaps some might actually
contain doubling. Still, most PCs do not show doubling, and I follow the working hypothesis that
doubling was never fully grammaticalized to mark a topic in Tocharian.14 Perhaps doubling was
partly motivated by metrical needs. In TB, for example, an additional third-person singular or
plural PC conveniently provides an additional syllable.15 Still, doubling is optional in TA and TB,
and not all examples are explicable in this way. Furthermore, doubling is attested in prose and
verse texts and their genres do not seem restricted.
7.4 Hypothesis and predictions
Regarding doubling of a nominal expression by a pronominal clitic in Tocharian A and Tocharian

B, one may think of two hypotheses that offer different predictions. The first hypothesis is to take
“redundant” clitics to be the manifestation of a single unitary phenomenon. It predicts the exis-
tence of a single set of restrictions that applies to all of the attested data (Table 7.2). The previous
treatments of the Tocharian pronominal clitics have implicitly adopted this hypothesis. However,
we argue against this view and show that at least two different types should be recognized, each
of which is subject to its own morphosyntactic and pragmatic restrictions (Hypothesis 2 of Table
7.2).
There are two types of doubling that we distinguish (7.19). The first type has a doubled associate
that undergoes dislocation. This type is further divisible into two subtypes: one whose associate
14. How TA and TB expressed a topic is an open question that needs to be investigated separately.
15. See, for example, (7.33).
193
Hypothesis Prediction
A morphosyntactic/pragmatic condition
Clitic doubling in TA and TB is
1 which triggers doubling should be
a unitary phenomenon.
the same for all of the manifestations.
Clitic doubling in TA and TB is not
a single phenomenon Conditioning environments can be different
2
but it is a composite depending on morphosyntactic/pragmatic configurations.
of different phenomena.
Table 7.2: Hypotheses and predictions
precedes a subject and the other in which a dislocated associate follows a finite verbal complex,
possibly separated by an intonational break. I label the former as Clitic Left Dislocation (CLLD)
and the latter Clitic Right Dislocation (CLRD). The second type has a non-dislocated associate
in argument position, following a subject. I call this type Clitic Doubling Proper (CDP).
(7.19) Different types of doubling
i. Dislocated associate
• Associate preceding a (non-topical) subject (= Clitic Left Dislocation)

[sentence … Associate … Subject … Verb+clitic ]
• Associate following a verbal complex (= Clitic Right Dislocation)

[sentence … Subject … Verb+clitic, Associate ]
ii. Non-dislocated associate

• Associate following a subject (= Clitic Doubling Proper)

[sentence … Subject … Associate … Verb+clitic ]
For the first- and second-person singular PCs, doubling is separable from apposition. The former
accompanies an associate that contains an independent form of a personal pronoun, while the
latter does not. For example, ārkiśoṣṣs krant knāṃmune(nāṃt)sunt cū ‘you (who) have become
good wisdom of the world’ is the associate of the second-person singular PC -ci in (7.20) and
contains the independent form of the second-person singular pronoun cū. In contrast, the sub-
stantivized adjective lareṃ ‘the beloved (one),’ the associate of the second-person singular clitic
-c in (7.21), does not accompany any independent personal pronoun (cf. TB ci [2sg]). Therefore,
I take (7.20) as an example of doubling, whose associate underwent dislocation from a preverbal
194
argument position to a postverbal position (CLRD). In contrast, I consider (7.21) not an example
of doubling but apposition.
(7.20) Clitic Right Dislocation [TA]

Context (Buddhastotra):
(kärsnālaṃ wramäṃ ¦ pu)k (l)otkasyo tñi ¦ käsont tämyo pūk kärsnāl wram ¦ knānmuneyo lyalyku-
ci :
‘The things to be understood with all manifestations have been understood by yousg .
Therefore, everything to be understood [is] illuminated by yoursg knowledge. [29c] ’
(puk knānmune-yis) ¦ lukśone ypant naṣt ¦

all knowledge-gen light make-nmlz.nom.sg cop.npst.act.2sg
wināsam-ci ārkiśoṣṣs ¦ krant knāṃmune [b3] (nāṃt)sunt
praise.npst.act.1sg-2sg world.gen good.acc wisdom be.ptcp.m.acc.sg
cū 20-9
acc.2sg
‘Yousg are the one who produces the illumination of all knowledge. I praise yousg , the
one who has become the good wisdom of the world. [29d] ’
(A249b2; trans. based on Pinault 2008: 289; verse; [5¦5¦8¦7]×4)
(7.21) Apposition [TB]16
Context (Buddhastotra):
śauläṃnmasa śauläṃnma ¦ śaulässontäs käryatai :
śaul r= anaiśai paṣṣatai ¦ śilä«ṣṣa»na sälyai(no :)
‘Yousg have bought the lives with lives from those who have lives. [22a]
As (if protecting your) life, yousg have carefully protected the rules of moral behavior.
[22b] ’
(kos [b1] a)räñc-n(e) klyentä«r» ñ(i) ¦ to= ta= ña«kta»

how.many heart-loc stand.npst.mid.3pl gen.1sg so.many gen.2sg lord.voc
krentauna :
virtue.pl
po śaul-äṣṣeṃ klautken-ne ¦ śault-sa lare-ṃ triśim-c mā 22

all life-adjz.acc.pl manner.pl-loc life-perl dear-acc.sg fail.opt.act.1sg-2sg neg
195
‘As many (virtues) stand in my heart, O Lord, so many virtues (stand in) yoursg heart.[22c]
In all life situations, may I not fail yousg , the beloved (one), throughout (my) life! [22d] ’
(B241b1; trans. based on Thomas 1997: 100; verse; [7¦7]×4)
I will show that CLLD/CLRD and CDP are subject to different morphosyntactic and pragmatic
conditions, and argue for Hypothesis 2 of Table 7.2. I will show that the first type (CLLD and
CLRD) consistently marks a primary topic. In contrast, the second type cooccurs with a doubled
associate, representing either a primary or a secondary topic in TA and TB.
7.5 Clitic Left Dislocation
Clitic Left Dislocation (CLLD) is a construction in which a clitic cooccurs with an associate to its
left. In most cases an associate precedes a subject as in (7.22), where the pronominal clitic ton
‘him’ cooccurs with the associate ton Kosta ‘Kosta,’ which precedes the subject i Maria ‘Mary.’
(7.22) Modern Greek

ton Kosta, [subject i Maria ] ton idhe.
det Kosta.acc det Mary.nom him saw
‘Mary saw Kosta.’ (Iatridou 1995: 11)
CLLD is not a combination of Clitic Doubling Proper (CDP) and dislocation because some languages
do not license CDP while they do CLLD (e.g., Standard Italian; Cinque 1990). The following two
subsections show that left dislocated associates consistently denote a topic referent (primary
16. This example is to be distinguished from secondary predication, which consists of two types: resultatives and
depictive predicates. While the former involves the resulting state caused as a result of an action (7.f), the latter
describes a quality that applies before the action described by a verb (7.g).
(7.f) John painted the house red. (resultative)
(7.g) John ate the meat raw. (depictive predicate)
Though it appears that (7.20) contains a depictive predicate, depictive predicates usually describe a property that
is not inherent (7.h; Rothstein 2017). An adjective or a prepositional phrase describing a person’s inherent property
is mostly infelicitous (7.i).
(7.h) Mary met John drunk. (depictive predicate; stage-level property)
(7.i) *Mary met John tall. (non-depictive predicate; inherent property)
larem ‘dear’ in (7.20) describes an inherent property, not a stage-level property such as drunk. Therefore, we think
(7.20) is an example of apposition, ārkiśoṣṣs krant knāṃmune (nāṃt)sunt cū and the second-person singular PC -c
placed next to each other.
196
topic) in TA and TB.
7.5.1 Clitic Left Dislocation in Tocharian A
First, let us look at the examples from Tocharian A. Example (7.23) shows the third-person singu-
lar pronominal clitic -äṃ doubles cam rupyāvateṃ śomäṃ ‘this boy Rupyāvata,’ an associate that
precedes the subject utpalāvatṣiñi wrasañ ‘the people of Utpalāvatī.’
(7.23) [TA] Context:

– – – – (kā)swac cam ¦ yāmär pākär oṅk ñomyo :
lāntuneṣiṃ kärparäṃ ¦ ṣärpseñc cam krant ṣo(treyā :
– – – – [a2] – – – ¦ – – – – – y)p(e)yis : 1 ||
‘… towards the (good) [thing? ] they revealed this one with a man’s name [= Rūpyāvata].
[1b] They point to the dignity of kingship (because of) the good si(gn). [1c] ... of the country.
[1d] ’
tmäṣ cam rupyāvateṃ śomäṃ [subject utpalāvat-ṣiñi

then dem.m.acc.sg Rupyāvata boy.acc.sg Utpalāvatī-adjz.m.nom.pl
wrasañ ] tsopatsäṃ abhisaṃskār-yo lāntun(e-ṣi [a3]
people.m.nom.pl great.acc.sg mental.determination-ins kingship-adjz
ynāñmune-yo yā)mr-äṃ
homage-ins make.pst.act.3pl-3sg
‘Thereupon the people from Utpalāvatī (greet)ed [topic this boy [named] Rūpyāvata ]
with careful mental preparation [and] (the homage due to) a king.’
This associate is the direct object (goal) of the verbal complex abhisaṃskāryo ynāñmuneyo yāmr-
‘(they) greeted X with mental preparation and respect’ (lit. ‘[they] made X with mental determi-
nation and veneration’).
This associate is the primary topic of (7.23), as it describes how he was welcomed by the citizens of
Utpalāvatī. The name Rūpyāvata appears in Haribhaṭṭa’s Jātakamālā (§6; Rūpyāvatī-jātaka). The
story goes as follows (see Ohnuma 2000 for a summary). There was a young woman called Rūpyā-
197
vatī (Rūpāvatī in the Divyāvadāna §32, Rukmavatī in Kṣemendra’s Bodhisattvāvadānakalpalatā §51).
She witnessed another woman who was suffering from hunger and about to eat her newborn
baby. Rūpyāvatī cut off her breasts for the woman and gave it to her as food. Śakra, the lord of
the gods, transformed himself into a brahmin and tested her, asking if she regretted her actions.
She denied it and vowed to him that if she had no regrets and offered her breasts for the sake of
buddhahood, she would become a man. Thus, she became a young male Rūpyāvata. Subsequently,
when the king of Utpalāvatī passed away, leaving no heir, ministers appointed him as a king, and
he ruled Utpalāvatī for sixty years. Example (7.23) describes that Rūpyāvata is welcomed by the
people of Utpalāvatī as a newly appointed king.
The corresponding Sanskrit passage reads as follows:
(7.24) Rūpyāvatī-jātaka in Haribhaṭṭa’s Jātakamālā (§6)

tad idam atra prāptakālam ayaṃ rūpyāvataḥ kumāraḥ sakalarājalakṣaṇopeta ābhigāmikaguṇasaṃ-
pannaś ca tad imam evādhipatyāyābhiṣekṣyāma iti |
tasyābhiṣekam atha cakrur udīrṇaharṣāḥ paurāḥ paropakṛtitatparamānasasya /
sārdhaṃ ca cāmarayugena mano ’bhirāmam uccikṣipe parijanena sitātapatram //
‘[The ministers speaking to the citizens:] “Now is the moment to act. This young man,
Rūpyāvata, bears all the marks of a king and has a king’s personal magnetism. So let it
be him we consecrate as king.” It was with intense joy that the citizens consecrated
someone devoted heart and soul to helping others. His retinue raised a delightful
white parasol over him, as well as a pair of chowries.’ (trans. by Khoroche 2017: 50)
A parallel passage that describes the citizens of Utpalāvatī celebrating Rūpyāvata is missing in
the Divyāvadāna (§32) and Kṣemendra’s (ca. 990–1066 CE) Bodhisattvāvadānakalpalatā (§51).
(7.25) Rūpāvatī-avadāna in the Divyāvadāna (§32; Vaidya 1959: 309)

tatra paṇḍatajātīyānāṃ mahāmātrāṇāmetadabhūt—yannu vayamutpalāvatyāṃ rājadhānyāṃ rājā-
naṃ sthāpayema | teṣāmetadabhūt—nānyatra rūpāvatakumārātkṛtapuṇyātkṛtakuśalāt | te rūpā-
vataṃ kumāramutpalāvatyāṃ rājadhānyāṃ rājānaṃ sthāpayanti | atha sa ṣaṣṭivarṣāṇi rājyaṁ
198
kārayati |
“The learned chief ministers had this thought about the matter: “We need to appoint a
king for the capital Utpalāvatī.” It occurred to them, “There is no one more qualified than
the young man Rūpāvata—he has performed meritorious deeds and virtuous actions.” So
they appointed the young man Rūpāvata as the king of the capital Utpalāvatī, and
there he ruled for sixty years.” (trans. by Rotman 2017: 264)
In the Bodhisattvāvadānakalpalatā (11th century CE), however, Rukmavant is promoted to the sub-
ject, suggesting that he is the primary topic. Although the TA passage is not a word-by-word
translation of a Sanskrit text, it is likely that the TA passage is also about Rupyāvata, describing
what happened to him when the king of Utpalāvatī passed away.
(7.26) Rukmavatī-avadāna in Kṣemendra’s Bodhisattvāvadānakalpalatā (§51.17–8; Straube 2009:

139)
nagaryāmutpalāvatyāmasminnavasare nṛpaḥ | utpalākṣaḥ samāptāyurvyādhiyogādvyapadyata
|| lakṣaṇajñairathābhyetya pravarairvṛddhamantribhiḥ | sadyaḥ samprāptapuṃstvo ’sau
rukmavānabhyaṣicyata ||
At that time, in the city of Utpalāvatī, the lotus-eyed king, whose lifetime had expired,
died due to illness. Then the best ministers, knowing signs, came near and consecrated
Rukmavant, who had suddenly become a man (trans. based on Straube 2009: 269).17

Sentence topic: Rūpyāvata

• People of Utpalāvatī greeted him and showed respect to him.
• …

(7.27) Summary of (7.23)
17. Straube (2009: 269): “Zu dieser Zeit starb in der Stadt Utpalāvatī der lotosäugige König, dessen Lebenszeit
abgelaufen war, infolge einer Krankheit. Da kamen die besten zeichenkundigen Minister herbei und weihen den
plötzlich zum Mann gewordenen Rukmavant.”
199
• QUD: What happened to the boy Rūpyāvata?
• Focus: utpalāvatṣiñi wrasañ tsopatsäṃ abhisaṃskāryo lāntun(eṣi ynāñmuneyo yā)mr- ‘the
people of Utpalāvatī greeted (him) with great respect and homage due to a king’
• Topic expression/referent (primary): cam rupyāvateṃ śomäṃ/Rupyāvata
This TA passage sets up a stage, introducing the citizens of Utpalāvati into the discourse. The
primary topic then shifts from Rūpyāvata to the citizens of Utpalāvati, describing how they cel-
ebrated him (7.28).
(7.28) [TA] Continuation of (7.23):

täm nu mänt wäkn-ā || ārśi-lāñcin-aṃ ||
dem.n conj how way-perl Ā–loc
rakär oplās-yo ¦ tkaṃ riy-aṃ ṣiraś ¦ prasar wrä-ntu

cover.pst.act.3pl lotus.pl-ins earth city-loc around sprinkle.pst.act.3pl water-pl
snum-ṣinä[a4](s18 – ¦ – – nāṃ)tsuṣ puk wrasañ :
fragrance-adjz.m.acc.pl cop.ptcp.m.nom.pl all being.nom.pl
“And that [they did] in which way? In [the tune] “Royal [anthem] of Ārśi”: They covered
the earth with lotus flowers, all around the city they sprinkled perfumed waters, all the
beings have been (busy (?)). [1d] …”
(A63a3-4; trans. by CEToM; verse [5¦5¦8¦7]×4)
Another example is from A395 (7.29). In this example, the PC -äṃ [3sg] doubles the associate
cam ‘him’ which refers to wasäṃ se ‘our son (i.e., Priyadatta).’ This associate precedes the subject
ñäktañ ‘the gods.’

klyomiṃ mar täpreṃ priyadatteṃ kāpñe se ṣurmaṣ klopaṃtsuts mäskatār puk wramäṃ ṣäpnā nā-
pak mā mäskanträ ṣñikek wasäṃ se ṣokyo patatam«†ä» [a2] neṣ cmolwāṣinās krant pñintwāśśi
plyāk ṣotre prant
18. Is snumṣinä(s) ‘of fragrance’ [m.acc.pl] an error for snumṣinā(s) [f.acc.pl]? Cf. A395 b4 : tmäṣ prāmnāñ preya-
datteṃ maṅkalṣināsf.acc.pl wräntuyo yāyruräṣ ‘then, the brahmins having bathed (?) Priyadatta with water of good
fortune …’
200
(The house-master is speaking to his wife about their son Priyadatta:) “O noble (woman),
don’t be so sad because of (our) dear son Priyadatta! Not all things happen according to
the dream. In contrast, our son (is) very fortunate? carrying the visible sign of the good
virtues of (his) previous births.”
ṣak-atsek cam [subject ñäktañ ] pāsaṃtr-äṃ

certainly dem.m.acc.sg god.m.nom.pl protect.npst.mid.3pl-3sg
ñātsey-ac mā kumnäṣ
danger-all neg come.npst.act.3sg
“The gods will surely protect him. [He] will not face [any] danger.” (lit. come to danger)
(A395a2; prose)
The primary topic here is Priyadatta, as this sentence describes how he will be protected by the
gods thanks to the virtues he gathered during his previous lives. Since the housemaster and his
wife are talking about him, the referent is discourse-old, both speaker and addressee presuppose
his existence.

Sentence topic: Priyadatta

• He has the signs of the good virtues he collected during his previous lives.
• The gods will undoubtedly protect him.
• He will not face any danger.
• …

In this passage the speaker answers the implicit QUD0 “Will our son Priyadatta face any danger?”
The speaker’s strategy to answer this question is to set up two another QUDs: QUD1 “Will things
(in general) happen according to the dreams?” and QUD2 “Is our son Priyadatta fortunate?” The first
QUD sets up a background reasoning for an answer to the aforementioned QUD0 . The second QUD
serves as a bridge to another QUD “What will happen to our son Priyadatta?” (QUD3 ), which in turn
provides a background reasoning to QUD0 . The speaker then answers the QUD0 with these two
background reasonings.
201
(7.30) Summary of (7.29)
• QUD: What will happen to our son Priyadatta? (= QUD3 )
• Focus: ṣak-atsek ñäktañ pāsaṃtr- ‘the gods will surely protect (him)’
• Topic expression/referent (primary): cam/Priyadatta
In (7.31), the plural PC -m doubles the associate ces śtwar mäśkitāśśi ‘to these four princes,’ namely
Vīryavān, Śilpavān, Rūpavān, and Prajñāvān. This associate again precedes the subject.

āmāśāñ ypeṣiñi wrasañ puro(hi)tāñ knānmānäṣ pᵤkāk puṇyavāṃ mäśkite parnore kärso(rä)[b6]ṣ
ārtant pālant puṇyavāni abhiṣe(k) yāmrä cam ypeyaṃ lāṃt yāmtsāṃtäṃ || ratisāyakaṃ || …
‘The ministers, people of the country, the chancellors, and the wise ones, who all had rec-
ognized prince Puṇyavān’s splendor, performed the anointment of Puṇyavān, and made
him king of the country. || In the R.-tune: || …
(tmä)ṣ ālyäkyāṃ praṣt-aṃ ces śtwar mäśkitā-śśi [subject tämne

thereupon other.f.acc.sg time-loc dem.m.acc.pl four prince-gen.pl so
wäkn-ā lānt wätkāṣ-äl wram ] kātka-m (tmäṣ) [a4]
manner-perl king be.decided.caus-gdv thing arise.pst.act.3sg-pl thereupon
po(ṃ)ś lānt-ac kakmuṣ lānt pälko-r-äṣ lāñci
all.m.nom.pl king-all come.ptcp.m.nom.pl king.acc see.ptcp-nmlz-abs royal
waṣt-is yok-m-aṃ klyānt
house-gen gate-du-loc stand.impf.mid.3pl
‘Thereupon at another time a problem (which) could (be) solved by such a king arose to
these four princes. (Then) they all came to the king, and facing the king they stood at
the gate of the royal palace.’
(A16a3; trans. based on Stumpf 1971: 28 n. 23; prose)
The associate ces śtwar mäśkitāśśi ‘to these four princes’ is the indirect object (location/experi-
encer) of the intransitive verb kātka- ‘(the problem) arose to X.’ It is the primary topic of (7.31)
as this sentence explains why they came to the palace of King Puṇyavān. These referents are all
discourse-old.
202

Sentence topic: The four princes

• A complex problem arose to them.
• They came to the palace of King Puṇyavān.
• They stood at the palace gate.
• …

This story answers the QUD1 “What happened?”. To answer this QUD, the author first sets up a
sub-QUD “What happened to the four princes?” (QUD1A ), which sets the basis for another sub-QUD
“What did they do?” (QUD1B ). The answer to the first sub-QUD is presented here.
(7.32) Summary of (7.31)

• QUD: What happened to the four princes? (= QUD1A )
• Focus: tämne wäknā lānt wätkāṣäl wram kātka- ‘a problem (that) could be solved by
such a king arose (to them)’
• Topic expression/referent (primary): ces śtwar mäśkitāśśi/Vīryavān, Śilpavān, Rū-
pavān, and Prajñāvān
To summarize, I have examined the examples whose PC doubles a nominal expression that pre-
cedes a subject and shown that the doubled associate is always discourse-old, and the interlocu-
tors presuppose its existence. In all of the examples, the left-dislocated associate represents the
primary topic of the utterance. In the following subsection, we will turn to the examples of Clitic
Left Dislocation in Tocharian B. Although evidence is somewhat limited, a left-dislocated asso-
ciate also denotes a topic referent (primary topic) in TB.
7.5.2 Clitic Left Dislocation in Tocharian B
Tocharian B also attests Clitic Left Dislocation, but the number of examples is somewhat lim-
ited. In (7.33) from the Araṇemijātaka, the tree-god is speaking to his wife about King Araṇemi’s
son Uttara. In this example, the subject of yärtten-ne ‘drag’ (aināki caimp brā(hmaṇi) ‘those mean
brahmins’) follows the DO (theme) of the verb.
203
(7.33) [TB] Context:
The tree-god is speaking to his wife after he saw that the brahmin Rudraśarma seized Ut-
tara and dragged him:
ykā-k cwi śa-māne pācer wlo ¦ vipṇu-nta-ṃts19 ra amāskai ¦

nevertheless-emp dem.m.gen.sg live-ptcp father king Viṣṇu-pl-gen ptcl difficult
yām-tsi sū erkatte •
make-inf dem.m.nom.sg unfriendly
(ya)k no cwi soṃśke lalaṃ[a6]ṣke-ṃ ¦ [subject aināki caimp

still conj dem.m.gen.sg son tender-acc mean.nom.pl dem.nom.pl
brā(hmaṇi ] ¦ yä)rt(t)en-ne śle tremeṃ :
brahmin.nom.pl drag.npst.act.3pl-3sg with anger
“His (= Uttara’s) father, the king (= Araṇemi) (is) still alive. Even for Viṣṇus, (it is) difficult
to treat him badly (lit. he is difficult to treat badly). [1a] But still, those mean brahmins
drag his tender little son with anger. [1b] ”
(B88a6; verse; [8¦7¦6]×2 + [9¦9] + [7¦6])
This example is structurally ambiguous: the associate of the clitic -ne [3sg] is either cwi soṃśke
lalaṃṣkeṃ ‘his tender little son (i.e., Prince Uttara),’ or cwi ‘his (i.e., King Araṇemi),’ as both of
them are in the third-person singular. Since the tree-god is talking to his wife regarding how
Prince Uttara is mistreated by the brahmin Rudraśarma, Uttara is likely to be the primary topic
of this utterance. Therefore, although we cannot exclude a different interpretation, this example
potentially shows that CLLD marks a sentence topic in Tocharian B.
The speaker tries to answer the QUD What happens? The speaker’s strategy to answer this question
is to set up a sub-QUD “What happens to Prince Uttara?” and answer this sub-QUD in pāda 1b. This
sub-QUD serves as a bridge to another sub-QUD What does he do?, which the speaker answers in
pāda 1c (7.34).
(7.34) [TB] B88a6-b1
19. vipṇuntaṃts is likely to be a copying error for viṣṇuntaṃts (CEToM).
204
pilko(-s= ā)ñmālaṣkeṃ lkāṣṣän-me ¦ taṅ-sa sam
look-perl sympathetic.acc look.npst.act.3sg-pl love-perl dem.m.nom.sg
mñcuṣke lareṃ pātär ramt :
prince.nom.sg dear.acc father.acc as
mā wa ksa ṣ cwimp [b1] mäsketrä ¦ waste comp

neg conj indf conj dem.gen.sg be.npst.mid.3sg protection dem.acc.sg
l(akle-ne 1)
suffering-loc
‘With a look begging for compassion that prince looks at them, full of love, as [if he were
looking at his] dear father. [1c] Nevertheless, not a single one [of them] is [any] protection
to him in that s(uffering). [1d] ’
(B88a6; trans. based on CEToM; verse; [8¦7¦6]×2 + [9¦9] + [7¦6])
(7.35) Summary of (7.33)

• QUD: What happens to Prince Uttara?
• Focus: aināki caimp brā(hmaṇi yä)rt(t)en- … śle tremeṃ ‘those common brahmins drag
(him) with anger’
• Topic expression/referent (primary): cwi soṃśke lalaṃṣkeṃ/Uttara
There is also an example of CLLD in which only a part of an antecedent precedes a subject. In
(7.37), which corresponds to the Udānavarga 4.5 (7.36), the pronominal clitic -ne [3sg] doubles
ceu … wnolme ‘that living being’ which refers to maimaṃtse ‘the intelligent one.’ This associate is
the direct object (theme) of the transitive melyan- ‘(the flood) crushes X,’ and it is discontinuous,
separated by the subject kerekauna ‘flood.’20
(7.36) Udānavarga 4.5 (Bernhard 1965: 127)

utthānenāpramādena saṃyamena damena ca /
dvīpaṃ karoti medhāvī tam ogho nābhimardati // 5
20. A hypothetical non-discontinuous alternative †ceu wnolme ˈ kerekauna ¦ mā melyan-ne would fit the meter better
because a minor caesura (indicated by ˈ) would be placed after the third syllable, as seen in other pādas whose first
colon consists of seven syllables. The motivation of this discontinuity is unclear. It appears either the violation of
a minor caesura has some pragmatic effect (drawing attention of audience?) or ceu has undergone dislocation to
align with the left edge of the pāda, perhaps imitating the Sanskrit passage (cf. tam ogho nābhimardati ‘flood does not
crush it,’ although tam ‘it’ refers to dvīpaṃ ‘island’).
205
‘With resurrection, care, self-restraint, and self-control, the intelligent (one) makes an
island which flood does not crush.’

The Buddha is speaking the Udānavarga 4.5 to the monks:
(spe)[b7]lke-sa snai ykorñe-sa ¦ wlāwalñe-sa maimaṃtse (:) /// /// – isa :

zeal-perl diligence-perl control-perl learned.m.nom.sg
ceu [subject kerekauna ] wnolme ¦ mā melyan-ne :
dem.m.acc.sg flood living.being neg crush.pst.act.3sg-3sg
(wc)e (śl)ok ///

second strophe
‘By (effort), diligence, and self-control, the learned one [31a] ... [31b] flood does not crush
such a living being. [31c] (He spoke the se)cond strophe … [31d] ’
(B12b7; trans. based on CEToM; verse; [7¦7] + [7¦4]×3)
Although the context is not available to us, the primary topic of this sentence does not seem
the subject kerekauna ‘flood’ but the associate ceu … wnolme ‘that living being,’ which refers to
maimaṃtse ‘the learned (one).’ This reference is discernible from the Udānavarga and the Pāli,
Patna, and Gāndhārī Dhammapadas, where the primary topic is ‘the intelligent’ (e.g., medhāvī
in the Udānavarga).21 The associate seems to be discourse-old, which is first introduced in 31a.
Since the preceding pāda is missing, the QUD of this passage is difficult to determine. Assistance
from the Sanskrit text is also limited because the TB passage is not a word-for-word translation.
We tentatively set What happens to the intelligent one? as the QUD, but we rather expect what does
the intelligent one do? from the Sanskrit passage.
(7.38) Summary of (7.37)

• QUD: What happens to the intelligent one?
• Focus: kerekauna … mā melyan- ‘flood (in general) does not crush (him)’
• Topic expression/referent (primary): ceu wnolme/‘the intelligent one’
21. See Ānandajoti (2020: 61) for the corresponding Pāli, Patna, and Gāndhārī versions.
206
To summarize, I have examined two examples where a pronominal clitic doubles an associate
which precedes a subject. Although both pieces of data are indecisive, it seems that the left-
dislocated associates also represent the primary topics of the utterances in TB. Therefore, these
examples in total suggest that TA and TB both use CLLD as a strategy to mark a non-subject con-
stituent as a primary topic of an utterance.
In the following section, I will turn to Clitic Right Dislocation (CLRD) in Tocharian A and B. I will
show that CLRD in TA also marks a sentence topic. Evidence is inconclusive for CLRD in TB.
7.6 Clitic Right Dislocation
Clitic Right Dislocation (CLRD) is a construction in which a clitic cooccurs with an associate to its
right. An intonational break in many cases separates the dislocated associate as in (7.39) from
Bulgarian.22
(7.39) Bulgarian
Decata ja običat, (φ Marija ).
the.kids acc.3sg.f love Maria
‘The kids love her, Maria.’ (Harizanov 2014: 1038)
The distribution of CLRD is different from that of CLLD. For example, Krapova and Cinque (2005)
point out that the so-called na-drop in colloquial Bulgarian (Vakareliyska 1994) is possible in CLLD
but not in CLRD. They also show that specific indefinite DPs may undergo CLLD but not CLRD.
CLRD is different from Clitic Doubling Proper (CDP) as some languages have the former but not
the latter (e.g., French; Kayne 1975; Anagnostopoulou 2006). Inanimate DPs may undergo CLRD
but not CDP in Standard Spanish (Anagnostopoulou 2006). It turns out that CLRD is very rare in
both Tocharian A and B, limited to verse texts. While no secure conclusion is drawable for TB due
to the limited amount of data, CLRD in TA seems to double a discourse-old associate and mark a
sentence topic.
22. CLRD does not necessarily involve an intonational break (Anagnostopoulou 2006).
207
7.6.1 Clitic Right Dislocation in Tocharian A
Clitic Right Dislocation is extremely rare in TA, and I found only three possible examples. Two
of them are from the TA translation of Mātṛceṭa’s Varṇārhavarṇastotra. The first example corre-
sponds to Varṇārhavarṇastotra II.63 (7.40).
(7.40) Mātṛceṭa’s Varṇārhavarṇastotra II.63 (Hartmann 1987)

sarvākāraparijñānāt sarvajñ(e)yāvabhāsine /
jñānaloka(karā)yāstu jñāna(bhūtā)ya te nama(ḥ) //
‘To him who, out of the complete understanding of all manifestations, illuminates all that
is knowable, who illuminates knowledge, and who has become knowledge, to you may
there be reverence!’
This passage is translated in Tocharian A as in (7.20), repeated here as (7.41):
(7.41) (= 7.20) [TA] Context:

(kärsnālaṃ wramäṃ ¦ pu)k (l)otkasyo tñi ¦ käsont tämyo pūk kärsnāl wram ¦ knānmuneyo lyalyku-
ci :
‘The things to be understood with all manifestations have been understood by you. There-
fore, everything to be understood [is] illuminated by your knowledge. [29c] ’
(puk knānmune-yis) ¦ lukśone ypant naṣt ¦

all knowledge-gen light make-nmlz.nom.sg cop.npst.act.2sg
wināsam-ci ārkiśoṣṣs ¦ krant knāṃmune [b3] (nāṃt)sunt
praise.npst.act.1sg-2sg world.gen good.acc wisdom be.ptcp.m.acc.sg
cū 20-9
acc.2sg
‘You are the one who produces the illumination of all knowledge. I praise you, the one
who has become the good wisdom of the world. [29d] ’
(A249b2; verse; [5¦5¦8¦7]×4; trans. based on Pinault 2008: 289)
In this example, the pronominal clitic -ci [2sg] doubles the associate ārkiśoṣṣs krant knāṃmune
(nāṃt)sunt cū ‘you (who have) become the good wisdom of the world,’ which translates Sanskrit
208
jñāna(bhūtā)ya te ‘to you (who) have become the wisdom.’23 This discourse-old associate is the
primary topic of the sentence.
The second example is from Mātṛceṭa’s Varṇārhavarṇastotra II.64.
(7.42) Mātṛceṭa’s Varṇārhavarṇastotra II.64 (Hartmann 1987)

te ’pi lokasya guravo yeṣām asi aguror guruḥ /
ato gurūṇām api te gu(ra)ve gu(ra)ve namaḥ //
‘Having no master, you are the master of those who (are) the masters of the world. There-
fore, may there be reverence to you (who) are the master of the masters, the master!’

bramñäktäṣṣ aci ¦ ñäktañ wras(a)śśi ¦ käṣṣiñ puk śkaṃ śaikṣy aśaikṣy ¦ pañcābhijñ(e)ñ riṣaki :
‘The gods, beginning with the god Brahmā, all the disciples, the Arhats, the sages with
five abhijñā are the masters of the beings. [30a] ’
cesm-i śkaṃ pᵤkis tu ¦ käṣṣi naṣt nātäk ¦

dem.m.pl-gen conj all.gen nom.2sg master cop.npst.act.2sg lord
wināsam-ci puk käṣṣi-śśi ¦ käṣṣi wäṣpā käṣṣiṃ (cu :)
praise.npst.act.1sg-2sg all master-gen.pl master truly master.acc acc.2sg
‘And you are the master to them all. Lord, I praise you, the master of all masters, the
true master. [30b] ’
(A249b4; verse; [5¦5¦8¦7]×4; trans. based on Pinault 2008: 289)
In (7.43), the nominal expression puk käṣṣiśśi käṣṣi wäṣpā käṣṣiṃ (cu) ‘(you), the master of all mas-
ters, the true master’ follows the matrix verb wināsam- ‘I praise X,’ which accompanies the pronom-
inal clitic -ci [2sg]. This nominal expression is the associate of the clitic and translates San-
skrit gurūṇām api te gu(ra)ve gu(ra)ve ‘to you the master of the masters, the master.’ This right-
23. In this example and (7.43), the second-person singular PC -ci occurs in Wackernagel’s position (without count-
ing a vocative). One might be tempted to consider examples such as these facilitated the change of a clitic that had
been placed in the Wackernagel position to the position immediately following the finite verb. However, a different
explanation would be required for the lack of case-distinction and the lack of co-occurrence of two or more PCs in
TA and TB.
209
dislocated associate is discourse-old and the primary topic of the sentence.
Finally, the last example of CLRD found in TA is A358a1 (7.44). Unfortunately, there is no context
available for this sentence.
(7.44) [TA] Context: N/A
rāhū swarbhaṇū ¦ asureñi lañś ¦ kaumary oki

Rāhū Svarbhāṇu Asura.nom.pl king.nom.pl prince.nom.pl like
prantär-cy opl-aṃ ¦ pūttiśpar-ṣiṃ oppal cu :
carry.npst.mid.3pl-2sg lotus-loc Buddha’s.dignity-adjz.acc.sg lotus acc.2sg
‘Rāhu (and? ) Svarbhāṇu, the Asura-kings, as if (they were) youths, carry yousg to the lotus,
yousg , the lotus of the Buddha’s dignity.’
(A358a1; verse; [5¦5¦8¦7]×4)
In (7.44), the pronominal clitic -cy [2sg] doubles the associate pūttiśparṣiṃ oppal cu ‘you, the lotus
of the Buddha’s dignity’, which is discourse-old. It is the direct object (theme) of the transitive
prantär- ‘(they) carry X (to Ylocative ).’ This associate appears to be the primary topic of the sen-
tence, but it is not secure because of the lack of context.
To summarize, I have shown that CLRD in TA marks a primary topic just as CLLD does. This paral-
lelism between CLRD and CLLD is not surprising since in some languages left and right-dislocated
associates both represent primary topics (see, e.g., De Cat 2007 for French). However, it is worth
pointing out that CLRD in TA is limited only to a couple of passages written in verse, and no
passage in prose displays CLRD.
7.6.2 Clitic Right Dislocation in Tocharian B
In Clitic Right Dislocation (CLRD), an associate follows a finite verb that hosts a pronominal clitic.
However, if a verb also precedes some other constituent than the associate, one might be in-
clined to think what is displaced is not an associate but the verb. In Tocharian B, there are three
examples in which an associate comes after a finite verb carrying a PC. In two of them, however,
more than one constituent follows the verbal complex. The verbal complex appears sentence-
210
or clause-initially, aligned with a pāda boundary or a major caesura, suggesting that it is not the
associate but the finite verb that underwent dislocation.
In (3.5), repeated here as (7.45), the finite verb with the pronominal clitic lkoym-c ‘I would see
yousg ’ precedes kru i ‘if ’ and ynemane ypauna kwṣainne ci ‘yousg going through lands (and) villages’
and appears clause-initially. In our analysis, this verbal complex underwent fronting to the be-
ginning of the subordinate clause, aligned with the beginning of Pāda 79c.
(7.45) (= 3.5) [TB] Context (Buddhastotra):

The Buddha’s mother (= Māyā) is speaking to him, recalling the happy days he still lived
among the men:
• lkoym-c kru i yne-mane ¦ ypauna kwṣain-ne ci ¦

see.opt.act.1sg-2sg whenever go-ptcp land.pl village.pl-loc acc.2sg
plu[a2]ṣṣi-ñ sak-sa palskw ārañce ¦ yapit wat no
float.impf.act.3sg-1sg happiness-perl mind heart enter.opt.act.2sg conj ptcl
wertsyai-ne (:)
assembly-loc
ñakty= āñcāl [a3] ṣar-ne ¦ keṃññi rämnoyeṃ ¦
god.pl A. hand-du knee.du bow.impf.act.3pl
‘Every time I saw you going through lands and villages, my mind (and) heart leapt for
joy, or (every time) you entered the community [79c] the gods bowed their knees with the
añjali hands ...’
(B246a1; verse; [5¦5¦8¦7]×4)
The nominal expression ynemane ypauna kwṣainne ci ‘you going through lands (and) villages’ is
the associate of the clitic -c and represents the direct object (theme) of the transitive lkoym- ‘I
used to see X.’24 This sentence seems to describe how Māyā used to feel when she saw his son.
Therefore, the primary topic seems to be Māyā, the subject of lkoym- ‘I used to see X.’25
24. According to Meunier (2015), pronominal clitics represent dativus sympatheticus. As dativus sympatheticus gen-
erally represents an individual affected by the event which a verb describes, we expect to find a doubled associate
that is affected. In this example, however, the doubled direct object does not seem to be affected by Māya’s seeing
of the Buddha.
25. In this manuscript, the <ci> akṣara is added below <nne>. It seems that the author of this manuscript omitted
211
In (7.46), the finite verb with a PC ṣärpsentär-ne ‘they indicate X’ precedes the reciprocal pronoun
ālyauce ‘each other’ and the full nominal expression nraiṣṣe wnolme tallāntä ‘the miserable hell-
being’ and appears at the beginning of the matrix clause.

The Buddha is speaking at the assembly: “If living beings do a deed, they will receive a ter-
rible fruit from that (deed) in their (re)births. The miserable beings blamed themselves,
evoked false thoughts, and blamed the good ones. […] Here, the good ones are called
those who stand in the pure moral conduct, (namely) in the discipline of the monkhood,
the novitiate, and the lay brotherhood, and eventually in the eight-fold discipline day and
night.”
ceu (-sa tattaṃ) ¦ nāki neṣmye snai ya[a7](mor 10-4

dem.m.acc.sg-perl place.subj.act.3sg rebuke evil.rumor without basis
sū temeñ sraukaṃ ¦ nrai-ne tänmastär ¦ māka)
dem.m.nom.sg therefore die.subj.act.3sg hell-loc born.npst.mid.3sg many
lykwarwa māka cmel-a ¦ māka lkāṣṣäṃ (läkle-nta :
times many rebirth-pl many see.npst.act.3sg suffering-pl
māka pudñäkti ¦ tsaṅkaṃ śaiṣṣe-ne ¦ ṣä)rpsentär-ne āly(au)c(e

many Buddha.pl rise.subj.act.3pl world-loc indicate.npst.mid.3pl-3sg each.other
ka) ¦ nrai-ṣṣe wnolme tall(āntä) :
ptcl hell-adjz being miserable.m.acc.sg
‘(If someonej ) puts blame or slander (which is) baseless on such (a good) onek , [14d] be-
cause of this, (when) this onej dies, (hej ) will be reborn in hell many times [and] see many
rebirths [and] many sufferings. [15a] If many Buddhas arise in the world, (they) will point
out the miserable hell-beingj to one another. [15b] ’26
(B15a7 = B17b1; trans. based on Hackstein, Habata, and Bross 2014: 51–3; verse;
this akṣara and added it later.
26. One might expect the reciprocal pronoun to appear as the genitive-dative †ālyaucentse rather than the ac-
cusative ālyauce found in this passage. The verb √särp- may take a non-human theme in the accusative and a human
goal in the genitive-dative (‘to explain/instruct/indicate somethingacc to somebodygen ’; e.g., THT255a5, THT255 b1
[2x], PKAS6C a4, PKAS6E b4, PKAS6K a7; cf. PKAS17H a4 that has a human goal in the locative), a human goal in the ac-
cusative and an embedded clause (‘to point out to somebodyacc that Embedded clause’; e.g., IOLToch214 a4; THT107
a10), or a human theme in the accusative and a human goal in the allative (‘to lead/guide/direct somebodyacc to
somebodyall ’; e.g., THT107 a9). I am unaware of any example in which √särp- takes two accusative arguments.
212
[5¦5¦8¦7]×4)27
Again, I analyze the verbal complex to be fronted to the beginning of the matrix clause, placed
immediately after the major caesura. The referent of the pronominal clitic -ne [3sg] is sū ‘he’
in a7, and seems to double the direct object nraiṣṣe wnolme tall(āntä) ‘miserable hell-being.’ It is
not clear whether the associate represents the primary topic of this sentence. It seems that the
primary topic is māka pudñäkti ‘many Buddhas’ and the associate is the secondary topic as this
sentence elaborates the relationship between the many Buddhas and the man who experienced
many rebirths in the hells.
Although an associate follows a finite verb with a PC in these cases, neither of them is likely
to involve CLRD since more than one constituent follows the finite verb and since the finite verb
appears clause-initially, aligned with a pāda-boundary or a major caesura. Instead, in our opinion,
they display dislocation of the finite verb. The only possible example of CLRD in TB comes from
PKAS17K a2, whose finite verb (peññan-me ‘[s/he] will pull X’) accompanies a PC and precedes the
subject (märskoṣṣäṃ ‘the frightened ones’).
(7.47) [TB] Context: N/A
sklok pr(o)sk(ai-ṣṣe or)k(a)mñe ¦ pkāte n(a)k(-t)s(i olya)po :

doubt fear-adjz darkness intend.pst.mid.3sg destroy-inf more
märkartsana pälsko-nta ¦ tākäs-si-ṃtse pelkiñä :

?
troubled.f.pl mind-pl destroy -inf-gen for.the.sake.of
(y)n(eś śai)ṣṣ(e) com= parn(a) ñke ¦ [a2] cä peññan-me

clear world dem.m.acc.sg outside now dem.m.acc.pl pull.subj.act.3sg-pl
märskoṣṣäṃ 1
be.afraid.ptcp.m.acc.pl
‘(He) intended to cer(tainly) destroy the (dark)ness of doubt [and] fe(ar). [1b] For the sake
of destroying the troubled thoughts, [1c] (he) will now pull them, (namely) the frightened
27. ‘Richtet jemand gegen einen solchen Tadel und Verleumdung, die gegenstandslos ist, und stirbt dieser deswe-
gen, so wird er in der Hölle wiedergeboren, viele Male, und erlebt viele Wiedergeburten und Leiden. Erheben sich
viele Buddhas auf der Welt, so weisen sie einander auf dieses unglückliche Höllenwesen hin.’ (Hackstein, Habata,
and Bross 2014: 51-53)
213
ones, out of this pre(sent wo)rld. [1d] ’
(PKAS17Ka2; trans. based on CEToM; verse; [7¦7]×4? )
Unfortunately no context is available, and the referent of -me [pl] remains uncertain. The func-
tion of the clitic seems to double cä … märskoṣṣäṃ28 ‘these frightened (ones),’ which is the direct
object (theme) of the transitive verb peññan- ‘(he) will pull X (out of Y).’ The primary topic seems
to be the subject of pkāte ‘(s/he) intended’ and peññan-me ‘(s/he) will pull X.’
7.7 Interim summary
To summarize, the examples whose doubled associate undergoes left- or right- dislocation dis-
played that when a PC doubles a dislocated associate, the associate always represents the primary
topic of the sentence. The associates discussed were all discourse-old, and the interlocutors pre-
supposed their existence at the time of the utterance.
So far, all of the examples discussed contained a discourse-old associate. However, a discourse-
new referent may become a topic (Section 7.2.2) as long as the discourse participants may pre-
suppose its existence by bridging (7.7) or it shows generic interpretation (7.10). In the following
example, however, the pronominal clitic -m [pl] seems to double nimittājñes brāmnāśśi ‘the nimitta-
jña Brahmins.’ This example is puzzling because the associate is discourse-new and the existence
of the nimittajña brahmins does not seem to be presupposed. There does not seem to be a linguis-
tic expression from which one can bridge the existence of the brahmins. They do not show the
generic interpretation either (i.e., “the nimittajña Brahmins [in general]”). Currently, we do not
have an explanation regarding why they may be topical.29

Priyadatta’s father and mother heard that he had been captured and taken to King Prase-
najit. They left the city of Sāketa and set out for the city of Śrāvastī. However, they could
not cross the forest of Kosala.
28. According to CEToM, cä is a copying error for ceṃ, and märskoṣṣäṃ is a misspelling for pärskoṣṣäṃ, triggered by
märkartsana in 1c.
214
mäṃt ne nimittājñes brāmnāśśi Śrāvastī ri-yā pre [subject
how comp acquainted.with.omens Brahman.gen.pl Śrāvastī city-perl outside
säm maṇḍal ] plyocksā-m tmä(-k) – – – – – – (mana)[b3]rkāṃ
dem.m.nom.sg maṇḍala arise.pst.act.3sg-pl dem-emp disciple.acc.sg
cam maṇḍl-ac kātse wānt-aṃ
dem.m.acc.sg maṇḍala-all in.front.of lead.pst.mid.3pl-3sg
‘When the maṇḍala arose outside the city of Śrāvastī to the brahmins (who are) ac-
quainted with omens, (they) … led the boy up to the maṇḍala.’
tämnek päñ känt oṅkälmās päñ känt ykas päñ kä«n»t kos nu nak maṇḍlac kātse wāworäṣ asläntwac
śarkr-äm •
‘Then, having led 500 elephants, 500 horses, (and) 500 cattle close to the maṇḍala, they
tied them to the (sacrificial) posts.’
The primary topic of this sentence is the nimittajña brahmins, and the discourse continues to
explain how they performed the ritual. That the subjects of wāworäṣ ‘having led X’ and śarkr-
‘(they) led X’, not expressed overtly, remain the same, suggests that the primary topic of (7.48) is
the brahmins.

Sentence topic: The nimittajña brahmins

• A maṇḍala arose to them.
• They led the boy to the maṇḍala.
• They led 500 elephants, 500 horses and 500 cows to the maṇḍala.
• They tied the animals to the sacrificial posts.
• …

The implicit QUD of (7.48) is What happened to the nimittajña brahmins?. It is a sub-QUD set up to
answer the bigger QUD What happened? This sub-QUD also serves as a bridge to the following QUD
What did they do to the boy Priyadatta?. It is unclear why the author was able to set up this sub-QUD
even though the nimittajña brahmins have not been introduced to the discourse before. It seems
215
that the author somehow considered that discourse participants are able to presuppose their
existence.
(7.49) Summary of (7.48)

• QUD: What happened to the nimittajña brahmins?
• Focus: Śrāvastī riyā pre säm maṇḍal plyocksā- ‘a maṇḍala arose (to them) outside the
city of Śrāvastī’
• Topic expression/referent (primary): nimittājñes brāmnāśśi/the nimittajña brahmins
The following section will turn to Clitic Doubling Proper (CDP) in TA and TB. I will show that the
distribution of CDP is different from that of CLRD and CLLD: CDP cooccurs with a topical associate,
29. In contrast to CLLD, there is no example of CLRD in TA whose associate is discourse-new. Although this might
be due to a mere lack of data, it might reflect some semanticopragmatic restriction. We find parallel examples in
Catalan (Vallduví 1992, 1995) and Italian (Brunetti 2009), where CLLD may topicalize a discourse-new associate while
CLRD cannot. In Catalan, for example, if an associate is discourse-old, CLLD and CLRD are both permissible as in (7.jii)
and (7.jiii).
(7.j) [Catalan] Associate (les llibres ‘the books’) = discourse-old (introduced in 7.ji)
i. A: On va posar les llibres? B: Em sembla que …
where pst.3sg put the books to.me seems that
A: ‘Where did (s)he put the books?’ B: ‘It seems to me that …’
ii. els llibres, els va posar al despatx.
the books them.m pst.3sg put in.the study.
‘(s)he put the books in the study.’ (CLLD; discourse-old)
iii. els va posar al despatx, els llibres.
them.m pst.3sg put in.the study the books.
‘(s)he put the books in the study.’ (CLRD; discourse-old)
If an associate is discourse-new, however, CLRD cannot be used. In contrast to (7.kii), where les coses ‘the things’
allows els llibres ‘the books’ to be topical (bridging; Section 7.2.2), (7.kiii) is infelicitous (Villalba 1998, Erteschik-Shir
2007).
(7.k) [Catalan] Associate (les llibres ‘the books’) = discourse-new (not introduced in 7.ki)
i. A: On va posar les coses? B: Em sembla que …
where pst.3sg put the things to.me seems that
A: ‘Where did (s)he put the things?’ B: ‘It seems to me that …’
ii. els llibres, els va posar al despatx.
the books them.m pst.3sg put in.the study.
‘(s)he put the books in the study.’ (CLLD; discourse-new)
iii. # els va posar al despatx, els llibres.
them.m pst.3sg put in.the study the books.
‘(s)he put the books in the study.’ (CLRD; discourse-new)
216
but it may represent either a primary or secondary topic.
7.8 Clitic Doubling Proper
In Clitic Doubling Proper (CDP), an associate neither precedes a subject nor is it separated by a
prosodic boundary. We observe it where a full nominal expression usually appears (e.g., 7.50). It
is challenging to distinguish CDP from CLRD in SVO languages since an associate follows a verb
in both configurations. However, Tocharian A and B, which are SOV languages, allow us to dis-
tinguish CDP from CLRD even though their intonational evidence is quite limited.
(7.50) Bulgarian
Decata ja običat neja.
the.kids acc.3sg.f love her
‘The kids love her.’ (Harizanov 2014: 1036)
Clitic doubling is known to be sensitive to semantic or pragmatic conditions such as animacy,

specificity, definiteness, and givenness. For example, Amharic may optionally double specific
indefinites, while it cannot double non-specific indefinites (Baker and Kramer 2018). A definite
direct object is obligatorily doubled, while doubling of an indefinite direct object is impossible in
Macedonian (Kochovska 2011). In Albanian, doubling of a direct object is possible only when it is
given (Kallulli 2008).
In the following subsections, I will show that the attested distribution of CDP in TA and TB is dif-
ferent from that of CLLD and CLRD: In contrast to CLLD and CLRD, CDP does not need to mark a
primary topic but may cooccur with an associate that denotes a secondary topic. Firstly, I will
examine the attestations in which a pronominal clitic doubles a direct object (theme). Secondly,
I will turn to the examples whose direct object contains a possessor. I will show that a PC consis-
tently doubles the possessor in such cases. Thirdly, I will move on to the cases in which a subject
contains a possessor doubled by a PC. Finally, I will discuss some examples in which a PC doubles
an indirect object (beneficiary) or a possessor of an indirect object (source).
217
7.8.1 Doubling of a theme of a transitive verb
7.8.1.1 Examples in Tocharian A
Unlike CLLD or CLRD, where an associate always represents a primary topic, CDP does not need to
double a primary topic in Tocharian A. Instead, when it doubles a theme argument of a transitive
verb, the doubled associate represents the secondary topic of a sentence. As defined in Section
7.2.4, a secondary topic is “an entity such that the utterance is construed to be about the rela-
tion that holds between it and the primary topic” (Nikolaeva 2001: 2; Dalrymple and Nikolaeva
2011: 54–7). Secondary topic has to be specific, and our analysis therefore predicts that PCs
never double non-specific quantified expressions such as ‘anybody,’ ‘nobody,’ and ‘who/which’
in Tocharian since they cannot be topical. Our analysis also predicts that doubling should be im-
possible in Tocharian if a focus extends to an entire utterance. In the following three examples,
a doubled associate represents a secondary topic.
The first example of CDP is from Puṇyavantajātaka (7.1 repeated here as 7.51):
(7.51) [TA] Context (Puṇyavantajātaka):

sas träṅkäṣ ñi amokyo täṣ cämplune kupre ne waluntāp ṣñi āyäntu (pkä)[b5]nt pkänt penu kāk-
loñcäs kälpāmār cesäm näṣ wnā kaśal tswāsam ||
‘The first (artisan) says: “Through my art this is my ability: If I find the bones of a de-
ceased, even [if] they have fallen apart, I will put them together again.”’
wät träṅkäṣ [subject näṣ ] nu ce(sm-ä)[b6]k āy-äntu

second speak.npst.act.3sg nom.1sg conj dem.m.acc.pl-emp bone-pl
pᵤk-ā-k puskās-yo kaśal malkam-äm
all-perl-emp sinew.pl-ins together put.together.subj.act.1sg-pl
‘The second (artisan) says: “But I will join the bones with the sinews completely.”’
In (7.51), the pronominal clitic -äm [pl] doubles the associate cesmäk āyäntu ‘the bones.’ This
associate is the direct object (theme) of the transitive verb kaśal malkam ‘I will join X together.’
Unlike CLLD and CLRD, this associate is not the primary topic: the sentence topic is näṣ ‘I’ (i.e.,
218
the second artisan) since his utterance describes how he is different from the first artisan.

Sentence topic: näṣ ‘I’ (the second artisan)

• can join the bones of a dead animal with sinews completely.
• …

This discourse-old associate cesmäk āyäntu ‘bones’ has been introduced to the discourse by the
first artisan, and it is pragmatically salient when the second artisan speaks. The utterance of the
second artisan updates the relationship between the primary topic (the second artisan himself)
and the bones. Therefore, in this example, cesmäk āyäntu ‘the bones’ is the secondary topic of the
utterance.
The QUD of this discourse is “Whose skill is the best?”. To answer this question, the first artisan sets
up a QUD “What can you do?” and answers this implicit QUD. The second artisan accepts the first
artisan’s assertion and sets up another implicit QUD “What can you do to the bones of a deceased?”.
(7.52) Summary of (7.51)

• QUD: What can you do to the bones of a deceased?
• Focus: pᵤkāk puskāsyo kaśal malkam- ‘(I) will join (them) with sinews completely’
• Topic expression/referent (primary): näṣ/speaker (= Śilpavān)
• Topic expression/referent (secondary): cesmäk āyäntu/the bones
There is another example of CDP in Puṇyavantajātaka:
(7.53) [TA] Context (Puṇyavantajātaka):

(tmä)[b6]ṣ śtärt kulmäṃtsyo wär camā eṣäk paṃpärs tmäk säm potäk pañwmāṃ śla śewiñlune
lakeyäṣ kākä[a1]tkuräṣ
‘Then, the fourth (artisan) sprinkled water over him (= a dead lion) with a blowpipe? . Im-
mediately, stretching his paw and yawning, he rose from his bed,’
219
kaśśi yokañi pälkāt cesäm amoktses kātse käly-māṃ
hungry thirsty see.pst.mid.3sg dem.m.acc.pl artisan.acc.pl near stand-ptcp
cesm-äk puk śtwar śälkās poke-yo wa(ltsu-r-ä)[a2]ṣ poñcäs
dem.m.acc.pl-emp all four altogether? paw-ins crush.ptcp-nmlz-abs all.acc.pl
kosā-m tāpa-m śkaṃ lo
kill.pst.act.3sg-pl eat.pst.act.3sg-pl conj ptcl
‘being hungry and thirsty, he saw these artisans standing nearby. Crushing those very
four altogether? with (his) paw, (he) killed them and ate them all up.’
In (7.53), the pronominal clitic -m [pl] doubles the associate poñcäs (= cesäm amoktses ‘the arti-
sans’). I consider it to be a substantivized adjective (i.e., ‘the all’), following Schulze, Sieg, and
Siegling (1931: 76) and Thomas (1997: 88). This associate is the direct object (theme) of the tran-
sitive kosā- ‘(the lion) killed X’ and tāpa- ‘(the lion) ate X,’ and has previously been introduced to
the discourse. This example is particularly illustrative because the primary topic is unambigu-
ously the lion: this sentence describes how he acted on the four artisans. The artisans are thus
the secondary topic of the sentence.

Sentence topic: The lion

• He rose.
• He was hungry and thirsty.
• He saw the four artisans standing nearby.
• He crushed the artisans with his paw, killed them, and ate them up.
• …

The big QUD of this discourse is “What happened?” (QUD1 ). The author breaks this question into
three sub-questions: QUD1A “What did the fourth artisan do?,” QUD1B “What happened to the lion?,”
and QUD1C “What did the lion do to the four artisans?.” Before the author answers QUD1C , he sets up
another QUD2 “What was the lion like?”, which offers reasoning to QUD1C .
(7.54) Summary of (7.53)

• QUD: What did the lion do to the four artisans?
220
• Focus: pokeyo wa(ltsurä)ṣ poñcäs kosā- tāpa- śkaṃ lo ‘having crushed (them), (he) killed
and ate (them) up’
• Topic expression/referent (primary): pro/the lion
• Topic expression/referent (secondary): cesmäk puk śtwar/the four artisans (= Vīryavān,
Śilpavān, Rūpavān, and Prajñāvān)
In the following example (7.55), the subordinate clause beginning with mäṃt ne … ‘when …’ in-
volves CLLD as observed in (7.48).30 In the matrix clause, the pronominal clitic -aṃ (for -äṃ [3sg];
TEB II: 34) doubles the associate manarkāṃ ‘the boy (= Priyadatta).’

Priyadatta’s father and mother heard that he had been captured and taken to King Prase-
najit. They left the city of Sāketa and set out for the city of Śrāvastī. However, they could
not cross the forest of Kosala.
mäṃt ne nimittājñes brāmnāśśi Śrāvastī riyā pre

how comp acquainted.with.omens Brahman.gen.pl Śrāvastī city-perl outside
säm maṇḍal plyocksā-m tmä(-k) – – – – – – (mana)[b3]rkāṃ
dem.m.nom.sg maṇḍala arise.pst.act.3sg-pl dem-emp disciple.acc.sg
cam maṇḍl-ac kātse wānt-aṃ
dem.m.acc.sg maṇḍala-all in.front.of lead.pst.mid.3pl-3sg
‘When the maṇḍala arose outside the city of Śrāvastī to the Brahmins (who are) acquainted
with omens, (they) … led the boy up to the maṇḍala.’
This associate is the direct object (theme) of the transitive verb kātse wānt- ‘(they) led X (to
30. Although English allows dislocation mostly in matrix clauses, some languages license it even in subordinate
clauses (e.g., Bulgarian; Krapova and Cinque 2008: 260).
(7.l) Bulgarian (Krapova and Cinque 2008: 260)
Ivan kaza če na Marija ti s ništo ne si ì pomognal.
Ivan said that to Maria you.nom with nothing not be.2sg her.dat helped.ptcp
‘Ivan said that Maria you haven’t helped her at all.’
221
Yallative )’. The primary topic of this sentence is nimittājñes brāmnāśśi ‘brahmins who decode the
omens.’ This sentence updates the relationship between the brahmins and the boy Priyadatta,
who is pragmatically salient at the time of the utterance. Therefore, (mana)rkāṃ ‘the boy’ is the
secondary topic of this sentence.

Sentence topic: The nimittajña brahmins

• A maṇḍala arose to them.
• They led the boy to the maṇḍala.
• …

The implicit QUD of (7.48) is What happened to the nimittajña brahmins?, which is a sub-QUD of the
bigger QUD What happened? This sub-QUD serves as a bridge to the following QUD What did the
nimittajña brahmins do to the boy Priyadatta?.
(7.56) Summary of (7.55)

• QUD: What did the nimittajña brahmins do to the boy Priyadatta?
• Focus: maṇḍlac kātse wānt- ‘(they) led (him) to the maṇḍala’
• Topic expression/referent (primary): nimittājñes brāmnāśśi/the nimittajña brahmins
• Topic expression/referent (secondary): (mana)rkāṃ/Priyadatta
As summarized in Table 7.3, when a clitic doubles a theme of a transitive verb in TA, the doubled
theme argument is always topical. However, it does not represent a primary but a secondary
topic. In this point CDP contrasts with CLLD and CLRD, where left or right-dislocated associates
consistently represent a primary topic.
7.8.1.2 Examples in Tocharian B
Pronominal clitics may also double a theme of a transitive verb in Tocharian B. In (7.2), repeated
here as (7.57), uttareṃ śamaśkeṃ ‘Prince Uttara’ is the internal argument (theme) of tsopaṃ- ‘(s/he)
222
(7.51) (7.53) (7.55)
Genre Prose Prose Prose
Associate cesmäk āyäntu poñcäs manarkāṃ
Gloss these bones the all the boy
Animacy [-human] [+human] [+human]
Person 3rd 3rd 3rd
Grammatical
DO DO DO
Function
Semantic Role Theme Theme Theme
Is an associate
No No No
pronominal?
Is an associate
No No No
discourse-new?
Is an associate
No No No
a primary topic?
Is an associate
Yes Yes Yes
a secondary topic?
Table 7.3: Clitic doubling the theme of a transitive verb in TA
pokes X’ and doubled by the pronominal clitic -ne [3sg].31
(7.57) (= 7.2) [TB] Context:

Prince Uttara was tormented by the brahmin Durmukha. His tongue was hanging out of
his mouth:
tumeṃ durmukhe brāhmaṇe uttare-«ṃ» śamaśke-ṃ kärwā-ṣṣai witsakai-sa

thereupon Durmukha brahmin Uttara-acc boy-acc reed-adjz.f.acc.sg root-perl
räskare tsopaṃ-ne
sharply sting.npst.act.3sg-3sg
‘Thereupon the brahmin Durmukha jabs the boy Uttara sharply with a reed root.’
(B88a1; trans. by CEToM; prose)
31. Since the instrument kärwāṣṣai witsakaisa ‘with a reed root’ is also in the third-person singular, one might
wonder whether the clitic -ne [3sg] does not double the theme but the instrument. Alternatively, one might wonder
whether the clitic does not double anything but represents a possessor of the instrument (i.e., ‘with his reed root’).
We cannot exclude these alternative interpretations in this example. However, in the following example (7.60), in
which the plural PC -me [pl] appears next to the internal argument ((ṣa)ñ k(e)wän ‘own cows’) and the instrument
(śakātaisa ‘with a stick’), the clitic is unambiguously doubling the internal argument because the instrument and
the possessor of the stick (‘a herdsman’) are both the third-person singulars, while the internal argument is the
third-person plural. Therefore, given this example, we take the PC -ne in (7.57) to double the internal argument.
223
This sentence concerns Durmukha and describes what he did to Uttara, so I take the brahmin Dur-
mukha to be the primary topic of this sentence.32 Nevertheless, Uttara is also discourse-old and
pragmatically salient when (9.109) was produced. Since this sentence updates the relationship
between Durmukha and Uttara, I take Uttara to be the secondary topic of this sentence.

Sentence topic: the brahmin Durmukha

• harshly jabs the boy Uttara with a reed root.
• …

The QUDs in this discourse shifts from QUD1 “What was his condition?” to QUD2 “What then hap-
pened?”. The author’s strategy to answer QUD2 is to set up a sub-QUD “What did the brahmin Dur-
mukha do to Uttara?”
(7.58) Summary of (7.57)

• QUD: What does the brahmin Durmukha do to Uttara?
• Focus: kärwāṣṣai witsakaisa räskare tsopaṃ- ‘(he) harshly jabs (him) with a reed root’
• Topic expression/referent (primary): durmukhe brāhmaṇe/Durmukha
• Topic expression/referent (secondary): uttare«ṃ» śamaśkeṃ/Uttara
CDP may refer to both discourse-new and discourse-old associates in TB. In the first example,
the associate (‘the boy Uttara’) is discourse-old. In contrast, the following example (7.60), which
corresponds to the Udānavarga 1.17, contains a PC (-me [pl]) which doubles a discourse-new as-
sociate ((ṣa)ñ k(e)wän ‘(his) own cows’).
(7.59) Udānavarga 1.17 (Bernhard 1965: 101)

yathā daṇḍena gopālo gāḥ prāpayati gocaram /
evaṃ rogair jarāmṛtyuḥ āyuḥ prāpayate nṛṇām // 1733
32. One might be inclined to take Uttara as the primary topic of this sentence since the discourse explains how he
is tormented by the brahmin Durmukha. Still, I think Durmukha is the sentence topic of this example because it is
Durmukha, not Uttara, that is in the subject position.
224
‘Just as a herdsman leads cows to a pasture with a stick,
in this way, old age (and) death lead the life of the human beings with sickness.’

Maudgalyāyana is explaining to Jātiśrona regarding how one’s life ends:
– – – – – (ṣa)ñ k(e)wän ¦ śakātai-sa kalṣtär-me ¦ ṣñār wepeṃ-ś

own cow.pl stick-perl goad.npst.mid.3sg-pl respective corral.pl-all
aśan-me :
lead.npst.act.3sg-pl
tu-yknesa ktsaitsñe srūka[a4](lñe) ¦ śaul kältsenträ wnolmen-tso ¦ ṣañ

thus old.age death life goad.npst.mid.3pl being.acc.pl-gen own
kalymi-ś aken-ne :
direction-all lead.npst.act.3pl-3sg
“(Just as a herdsman) goads (his) own cows with a stick, and leads them to their corrals,
[89a] in this way old age and death goad the life of the beings, and lead it to its destination.
[89b] ”
mäkte cake (ṣ)liye – – – ¦ – – – – – – (kl)autkot(rä) :

mant kättaṅkäṃ śaulanma ¦ mā ṣp wtentse klautkonträ 80-9
“Just like a mountain stream … (does not) turn back, [89c] so lives go by and do never turn
back. [89d] ”
(B3a3; trans. based on CEToM; verse; [8¦7¦6]×2 + [4/5¦4/5] + [7¦6])
Doubling in (7.60) is not imitating the corresponding Sanskrit passage since the direct object gāḥ
does not show doubling of any independent or bound pronoun in (7.59). Thus, CDP found here is
motivated by some properties of the Tocharian grammar.
Even though the nominal expression (ṣa)ñ k(e)wän ‘(his) own cows’ is discourse-new, I think the
interlocutors presuppose the existence of the referents thanks to the subject ‘a herdsman’ (cf.
33. As observed by Thomas (1983: 142), the TB passage seems to contain translation from a text which had a variant
reading evaṃ jarā (ca mr̥tyuś ca) ‘in this way old age and death …’ listed in Bernhard (1965: 101) (cf. Pāli Dhp. 135c
evaṃ jarā ca maccu ca, with the same meaning).
225
Skt. gopālo). It is an example of bridging by which a listener may presuppose the existence of a
referent from a relevant linguistic expression (Section 7.2.2). This example shows that it is not
givenness that determines whether a PC may double a nominal expression or not. A discourse-
new associate may show doubling when its existence is presupposed by bridging or when it offers
a generic interpretation. The primary topic of the subordinate clause is ‘a herdsman’ (cf. Skt.
gopālo), which offers a generic interpretation in this case (i.e., ‘Just as a herdsman, in general,
goads his cows …’). The cows represent the secondary topic of the subordinate clause.34
The QUD of this utterance is “What does a herdsman (in general) do to his livestock?” The Buddha sets
up this QUD to introduce a parallel to the following QUD “What do old age and death (in general) do
to us?”
(7.61) Summary of (7.60)

• QUD: What does a herdsman (in general) do to his livestock?
• Focus: śakātaisa kalṣtär- ‘(he) goads (them) with a stick’
• Topic expression/referent (primary): NA/a herdsman (in general; cf. Skt. gopālo)
• Topic expression/referent (secondary): (ṣa)ñ k(e)wän/his cows
Table (7.4) summarizes the examples discussed. When there is a PC that doubles a theme in TA
and TB, the associate is a secondary topic. I have shown that an associate does not have to be
discourse-old: a PC may double a discourse-new associate whose existence the interlocutors pre-
suppose via bridging or an associate which allows generic interpretation.
7.8.2 Doubling of a possessor of a direct object
So far, I have examined the examples in which a pronominal clitic doubles a theme of a transitive
verb. In those examples, theme arguments did not have a possessor. However, when the theme
argument accompanies a possessor, a PC may, in principle, refer to the theme (possessum) or the
34. In this example, the pronominal clitic does not double the goal (ṣñār wepeṃś ‘to (their) own corrals’) but the
theme argument. The choice between these arguments seems determined by the referential hierarchy (i.e.,
first/second person » human » animate » inanimate). Note that when a PC doubles an IO, the doubled IO is always
animate (e.g., 7.86, 7.88, and 7.92). In (7.60), the IO is inanimate while the DO is animate.
226
(7.57) (7.60)
Genre Prose Verse
Associate uttare«ṃ» śamaśkeṃ (ṣa)ñ k(e)wän
Gloss Prince Uttara own cows
[+animate]
Animacy [+animate]
[+human]
Person 3rd 3rd
Number singular plural
Grammatical
DO DO
Function
Semantic Role Theme Theme
Is an associate
No No
pronominal?
Is an associate
No Yes
discourse-new?
Is an associate
No No
a primary topic?
Is an associate
Yes Yes
a secondary topic?
Table 7.4: Clitic doubling of the theme of a transitive verb in TB
possessor who possesses the theme. If a possessor and a possessum are in the same person and
number, one cannot decide if a PC doubles a possessor or a possessum. In the following example
of Tocharian A (7.62), the referent of the pronominal clitic -äm [3sg] is ambiguous as the possessor
mācrik ‘of the mother’ and the possessum śol ‘life’ are both in the third-person singular.

– – – – – – – ¦ – – (śa)lpmāṃ mācräṣ tāṣ :
wiyoss oki cam klopyo ¦ mācar nuṃ nuṃ trekaṣ-äṃ :
‘[…] (S/he) will be released from the mother. [3a] As if (she were) frightened, his/her
mother will be confused by that suffering again and again. [3b] ’
täm kärso-r-äṣ ·o /// [a3] /// (:)

dem know.ptcp-nmlz-abs
227
mācr-i-k śkaṃ śol pāsantr-äṃ ¦ mā śkaṃ cam-aṃ tuṅk
mother-gen-emp conj life protect.npst.mid.3pl-3sg neg conj dem.m.sg-loc love
mskatr-äṃ :3
be.npst.mid.3sg-3sg
‘Having recognized it […]. [3c] and (they) protect the life of the mother. But she does not
have love for him. [3d] ’
(A152a3; verse; [7¦7]×4)35
Likewise, klawāte-ne ‘He touched X’ in (7.63) of Tocharian B accompanies the third-person singular
clitic (-ne [3sg]), which may refer to the theme argument kektseño ‘(the Buddha’s) body,’ or the
possessor pudñäktentse ‘of the Buddha’ since both of them are in the third-person singular.

The Buddha was in Śrāvastī. In the morning he went out of the cell. The sun had already
risen high. He sat on the seat, took off his upper garment, and held his back against the
sun.
lyam= ānande keni-sa ¦ (a)[b5]lyine-sa antapi :

pudñäkte-ntse kektseño ¦ klawāte-ne lyawā-ne :

‘Ānanda sat on [his] knees. With both palms he massaged the body of the Buddha and
rubbed it.’
(B5b5; trans. based on CEToM; verse; [7¦7]×4)36
In the following example, the matrix verb śwāṃ-ne ‘(he) eats X’ takes the internal argument
pelaikneṣṣe kektseṃ cwi ‘his body of laws.’ As the possessor (cwi) and the possessum are both third-
35. One might be inclined to take this pronominal clitic to represent the source of the transitive verb (i.e., ‘protect
X from Y’). However, we are unaware of a parallel example in which √pās- ‘to protect, obey’ takes a source argu-
ment/adjunct.
36. One might wonder whether the PC in this example is referring to ānande ‘Ānanda,’ representing the (inalien-
able) possessor of the instrument (a)lyinesa antapi (i.e., ‘Using both of his palms’). We cannot exclude this interpre-
tation (but cf. footnote 31).
228
person singular, it is ambiguous whether the clitic -ne [3sg] doubles the possessor or the posses-
sum.
(7.64) [TB] Context (Satkāravarga):

‘If someone gives up (his) possession and properties, [2c] because of faith and weariness
of the world, he will not gather (them) again. [2d] … But if he who has become a monk
and (is) eating (alms he) begged [3c] obtains possessions, (he) will be smeared with great
blame. [3d] If youpl had cared for veneration (and) flattery, you yourselves should rather
have stayed (in) the house (i.e., should not have become monks). [4a] From (your) respec-
tive male- and female-slaves, (you) would surely have obtained veneration and gathered
possessions. [4b] Now youpl have become a monk with the wish for Nirvaṇa. [4c] (But)
youpl are (still) bound by the fetters of veneration, flattery, and profit. [4d] ’
mäkte yelyi-tse ¦ ku tallāw tākaṃ ¦

comp worms-adjz.nom.sg dog.nom.sg miserable.m.nom.sg cop.subj.act.sg
śwāṃ-ne yṅkau«ṃ» [b1] kästwer ¦ yelyi pile-nta :
eat.npst.act.3pl-3sg by.day at.night worm.nom.pl wound-pl
mant källau-ṣṣi yelyi cmentär ¦ ontsoytñe-ṣṣe

thus gain-adjz.nom.pl worm.nom.pl be.born.subj.mid.3pl insatiability-adjz.sg
pīle-ne ¦ pelaikne-ṣṣe37 kektseṃ cwi :
wound-loc law-adjz body.acc.sg dem.gen.sg
śwāṃ-ne yṅkauṃ kästwer ¦ mā =ñu kälpāṣṣäṃ :

eat.npst.act.3pl-3sg by.day at.night neg peace acquire.npst.act.3sg
‘(If) a miserable dog is (infected) with worms, the worms will eat (his) wounds by day
(and) night. [5a] In this way, (if) the worms of profit are born in the wound of insatiability,
(they) will eat his body of the law by day (and) night, (and) he will not obtain any peace.
[5c] ’
(B33b1; verse; [5¦5¦5¦5] + [8¦7¦7] + [5¦5] + [8¦7])
Therefore, the examples whose possessor and possessum differ in person or number are critical
for us. In what follows, we use the attestations of Tocharian B and show that when a possessor and
37. pelaikneṣṣe ‘of the law’ [m.nom/acc.sg] is to be read as pelaikneṣṣai [f.acc.sg]
229
a possessum differ in person or number, a PC consistently doubles the possessor. The possessors
in the following examples are all discourse-old, and the existence of the referents is pragmatically
presupposed, while the possessa are all discourse-new.
In (7.65), the first-person singular PC -ñ refers to the speaker, King Subhāṣitagaveṣin. It does not
double the direct object yakt-āñm ñi ‘my feeble (state),’ which is third-person singular. Here the
possessor is discourse-old and pragmatically salient at the time of Subhāṣitagaveṣin’s utterance
while the possessum yakt-āñm is not.
(7.65) [TB] Context: Indra, who changed his appearance to a yakṣa, asks King Subhāṣitagavesin
why he is so sad. King Subhāṣitagaveṣin answers to him:
pūdñäktä-ññe pelai[b3](kne ¦ /// :)

buddha-adjz law
ceu -sa ñiś ñke meṅki-tse ¦ te-sa pkārsa-ñ

dem.m.sg-perl 1sg now lack-adjz.m.nom.sg dem.n.sg-perl know.imp.act.2sg-1sg
yakt-āñm ñi :1
feebleness 1sg.gen
“ … The law of the Buddha … [1c] I lack it now. Because of this, understand my feeble state!
[1d] ”
(B99b3; verse; [7¦7]×4)
The speaker sets up an implicit QUD “What does the speaker (King Subhāṣitagavesin) lack?” (QUD1 )
and answers it. This QUD proffers reasoning to another QUD “What then should the addressee (Indra)
do to the speaker (King Subhāṣitagavesin)?” (QUD2 ).
(7.66) Summary of (7.65)

• QUD: What should the addressee (Indra) do to the speaker (King Subhāṣitagavesin)?
• Focus: pkārsa- yakt-āñm ‘Understand (my) feebleness’
• Topic expression/referent (primary): pro/addressee (= Indra)
• Topic expression/referent (secondary): ñi/speaker (= King Subhāṣitagavesin)
230
In the following example (7.67), the second-person singular PC -c [2sg] refers to the addressee,
namely the Buddha. It doubles the possessor, not the possessum (pelaikneṣṣai kektseñ ‘the body of
the law’), which is third-person singular. Again, the possessor tañ is discourse-old and pragmat-
ically salient at the time of the utterance. At the same time, the possessum pelaikneṣṣai kektseñ
‘the body of the law’ has not been introduced to the discourse before.
(7.67) [TB] Context (Buddhastotra): N/A
ka (— up)pāl-yok ¦ weś(e)ñ(ai)-sa brahmaswar (:)

lotus-colored voice-perl Brahma.svara
śtwār= empre-nma-ṣṣe ¦ klene(nt —————:

four truth-pl-adjz resounding
— — — — — ¦ — — —)[a2]-mpa tsälpāre :
…-com be.free.pst.act.3pl
pelaikne-ṣṣai tañ ¦ kektseñ wato wināskau-c 40-7 ||

law-adjz.f.acc.sg gen.2sg body.acc.sg again? honor.npst.act.1sg-2sg
‘… with the lotus-colored Brahmasvara-voice, [47a] … resounding … of the four truths. [47b]
… were free with … [47c] I again? honor yoursg body of the law. [47d] ’
Because of the damage of the text, the QUD of this sentence is difficult to pinpoint. I tentatively
set the QUD as “What does the speaker do to the addressee (the Buddha)?”
(7.68) Summary of (7.67)

• QUD: What does the speaker do to the addressee (the Buddha)?
• Focus: pelaikneṣṣai kektseñ wato wināskau ‘praise (your) body of the law again’
• Topic expression/referent (primary): pro/speaker
• Topic expression/referent (secondary): tañ/addressee (= the Buddha)
Furthermore, it is possible to expand the list by adding the following example.
231
(7.69) [TB] Context (Commentary of the Udānavarga. 31.6 and 31.7):38
The Buddha lord saw Vajraka (i.e. one adorned with a jewel) from afar on the earth. Va-
jraka spoke to him from afar. The Buddha, in turn, spoke to him: “I have seen you from
afar.”
dharmaruci [a6] weñā-ne-ś ¦ poyśeñcai lauk(a)r olypotstse :

Dharmaruci speak.pst.act.3sg-3sg-all omniscient.voc far very
kārpa kent-sa poyśi-ntse ¦ wi(nā)ṣṣa-ne pai-(n)e

descend.pst.act.3sg earth-perl omniscient-gen praise.pst.act.3sg-3sg foot-du
l(a)laṃṣ(k)i 10-1
tender.du
‘Dharmaruci spoke to him: “O Omniscient, a long time (ago) [11c] , (Vajraka) descended on
the earth and praised the two tender feet of the omniscient one. [11d] ’”
(PKAS6Aa6; trans. based on CEToM; verse; [7¦8]×4)39
38. The example (7.69) is from a commentary which follows the translation of the Udānavarga 31.6 and 31.7.
(7.m) Udānavarga 31.6 and 31.7 (Cittavarga; Bernhard 1965: 409–10)
anekaṃ jātisaṃsāraṃ saṃdhāvitvā punaḥ punaḥ /
gr̥hakārakaiṣamāṇas tvaṃ duḥkhā jātiḥ punaḥ punaḥ // 6
‘Having experienced the cycle of countless rebirths again and again, you (were) seeking a house builder:
(re)birth (and) suffering, again and again.’
gr̥hakāraka dr̥ṣṭo ’si na punar gehaṃ kariṣyasi |

sarvo te pārśukā bhagnā gr̥hakūṭaṃ visaṃskr̥tam /
visaṃskāragate citte ihaiva kṣayam adhyagāḥ // 7
‘O housebuilder, you have been seen. You will not build a house again. All your rafters have been broken, and
the ridgepole has been destroyed. The mind has reached dissolution. Just here, you obtained disappearance.’
(7.n) [TB] Translation of the Uv. 31.6 and 31.7 (PKAS6A a1–4):
(snai) [a1] keś cmel(a)ṣṣe serke ¦ mäkorm(eṃ) nä no-nä n= ost-yāmṣeñcai :
ñ(ä)sk(e)m(a)ne tw(e) l(a)kle ¦ c(ä)m(e)lyñe nä no-nä no śl(o)k yparwe :
‘Having passed the cycle of countless rebirths again and again, you (were) longing for a house builder (i.e.,
mind influenced by desire): [9b] rebirth (and) suffering, again and again. (This is) the first strophe. [9c] ’
(o)[a2]st-yāmṣ(e)ñcai lyelyku nest ¦ mā nano ost yāmttarä 9

poṃc tañ kleñcaṃ kakautaṣ ¦ ostantse mrāce käskowä :
[a3] käskālläññe ykūweṣn(e) ¦ (pa)lskone tane a(ttsaik) :
nautalläññe yonmasta ¦ se wcepi āke ślo[a4]kantse :
‘O house-builder, yousg have been seen: yousg will not build a house again. [9d] All yoursg rafters are broken
and the top of the house is destroyed. [10a] The mind has reached annihilation. Just here, yousg have [10b]
obtained extinction. This is the end of the second strophe. [10c] ’
39. One might be inclined to take this clitic to represent the direct object (theme) and to see here the so-called
Accusative of Respect (i.e., ‘[Vajraka] praised him [with respect to] the two feet’). This account, however, would
232
In (7.69), the third-person singular PC -ne [3sg] doubles poyśintse ‘of the omniscient one’ who is
the inalienable possessor of paine ‘two feet’ [du]. This clitic does not seem to double the direct
object (poyśintse paine) because a dual noun usually triggers dual or plural agreement. In (7.70),
for example, a finite verb in the third-person plural carries a predicative adjective in the dual
(Adams 2015: 68f.). In (7.71) from TA, a dual subject (aśäṃ ‘two eyes’ and klośäṃ ‘two ears’) takes
a verb in the third-person plural (lkeñc ‘[they] see’ and klyosnseñc ‘[they] hear’, respectively).
(7.70) [TB] Plural subject taking a predicative adjective in the dual

– – – sonopälle • prakaryanedu mäskeṃtär 3pl •
‘… to be massaged, [and] (they) become3pl firmdu .’ (W26b3; Broomhead 1962: 26)
(7.71) [TB] Dual noun triggering plural agreement
(lke)ñc pe aśäṃ krant wramäṃ ¦ swāräṃ rake

see.npst.act.3pl also eye.du good.pl thing.pl sweet.acc.sg word.sg
klyosnseñc pe ¦ klośäṃ nāñi :
hear.npst.act.3pl also ear.du gen.1sg.f
‘[My] eyesdu also (se)e3pl the good things, my earsdu also hear3pl the sweet word.’
(A58b3; trans. by CEToM; verse; [7¦7¦4]×4)
In example (7.69), the primary topic is Vajraka, who is the non-overt pronominal subject of the fi-
nite verbs kārpa ‘(he) descended’ and wi(nā)ṣṣa-me ‘(he) praised X.’ The Buddha is the addressee of
Dharmaruci’s utterance, which updates the relationship between Vajraka and the Buddha. There-
fore, we analyze the Buddha as the secondary topic of the utterance. The PC’s associate (poyśintse
‘of the omniscient one’) is discourse-old, and his existence is pragmatically presupposed by the
interlocutors when Darmaruci spoke to him. At the same time, the possessum paine ‘two feet’
has not been introduced to the discourse before. The QUD of Dharmaruci’s utterance seems to
be “What did Vajraka do to the omniscient one?”.
(7.72) Summary of (7.69)
leave the genitive poyśintse ‘of the omniscient one’ unexplained (cf. in a double accusative construction, a possessor
is in the accusative). According to Luraghi (2020), Accusative of Respect is restricted to Middle/Neo Hittite, Greek
and Armenian, and not reconstructable in PIE.
233
• QUD: What did Vajraka do to the omniscient one?
• Focus: wi(nā)ṣṣa- pai(n)e l(a)laṃṣ(k)i ‘praised two tender feet’
• Topic expression/referent (primary): pro/Vajraka
• Topic expression/referent (secondary): poyśintse/the omniscient
So far, the examples discussed all had the following structure:
(7.73) … [internal argument Possessorj Possessum ] Verb-PCj .
All possessors were discourse-old, and their referents were pragmatically presupposed, whereas
the possessa were all discourse-new. Doubling consistently targeted the possessors in these ex-
amples. The examples all expanded the discourse by updating the relationship between the pri-
mary topic and the discourse-old possessor by introducing a discourse-new possessum. CDP in
TB always cooccurs with an associate that represents a secondary topic.
Returning to an ambiguous example (7.63), repeated here as (7.74), the possessor (pudñäktentse
‘of the Buddha’) is discourse-old, and his existence is pragmatically presupposed. At the same
time, the possessum (kektseño ‘the body’) is discourse-new, first introduced here. Therefore, our
analysis suggests that the third-person singular PC -ne in (7.74) does not double the direct object
but the (inalienable) possessor pudñäktentse ‘of the Buddha.’
(7.74) (= 7.63) [TB] Context:

The Buddha was in Śrāvastī. In the morning he went out of the cell. The sun had already
risen high. He sat on his seat, took off his upper garment, and held his back up to the sun.
lyam= ānande keni-sa ¦ (a)[b5]lyine-sa antapi :

pudñäkte-ntse kektseño ¦ klawāte-ne lyawā-ne :

‘Ānandaprimary topic sat on [his] knees. With both palms [of his hands] he massaged the
body of the Buddhasecondary topic and rubbed it’
234
In this example, Ānanda is the primary topic of the sentence as it concerns what he did to the
Buddha. At the same time, it expands the discourse by introducing a discourse-new possessum
kektseño and updating the relationship between him and the Buddha. Therefore, the Buddha is
the secondary topic of the sentence.
The first QUD the author sets up is “What happened?” (QUD1 ), which is answered by lyam=ānande
kenisa ‘Ānanda sat on [his] knees.’ As a result of this payoff move (assertion), the CG now contains
the proposition SatOnTheKnees(ā) = Ānanda sat on [his] knees. Subsequently, the author sets up
another QUD: “What did Ānanda do to the Buddha?” (QUD2 ). This QUD contains Ānanda, introduced
to the CG in the previous assertion.
(7.75) Summary of (7.74)

• QUD: What did Ānanda do to the Buddha?
• Focus: (a)lyinesa antapi kektseño klawāte- lyawā- ‘(He) massaged (his) body with both
palms and rubbed it’
• Topic expression/referent (primary): ānande/Ānanda
• Topic expression/referent (secondary): pudñäktentse/the Buddha
7.8.3 Doubling of a possessor of an intransitive subject
7.8.3.1 Tocharian B
In the previous subsection, I have shown that when an internal argument (possessum) accompa-
nies a possessor, it is always the possessor that is doubled by a pronominal clitic. In such cases,
possessors are always discourse-old and topical, possessing discourse-new non-topical possessa.
All possessa in the examples discussed are the direct object of a transitive verb. However, a PC may
also double a possessor of an unaccusative subject (Chapter 4). In such cases, the PC’s associate
is unambiguous—it consistently refers to the possessor. Then, if the proposed analysis is on the
right track, I expect to find a topical possessor, typically discourse-old, possessing a non-topical
discourse-new possessum. I also expect discourse participants to presuppose the existence of the
235
topical possessor.
According to Krifka (2008: 267), “There is a well-documented tendency to keep the topic con-
stant over longer stretches of discourse (so-called topic chains, cf. Givón 1983)”. If a possessor
represents a topical constituent and a pro represents a continued topic, we expect to find two
consecutive sentences that retain the same topic but have different subjects. In other words, if
an unaccusative subject carries a topical possessor and if the subject of an immediately following
sentence is a pro, I predict that the pro does not refer to the possessum but the possessor (con-
tinued topic with subject shift). In contrast, if a possessum is the topic of a sentence and if the
subject of an immediately following sentence is a pro, the possessum should be the antecedent of
the pro.
(7.76) Primary topic = possessor
i. [subject Possessorj Possessum ]k … Verb.
ii. [subject proj ] … Verb.
(7.77) Primary topic = possessum
i. [subject Possessorj Possessum ]k … Verb.
ii. [subject prok ] … Verb.
This prediction is borne out. A discourse-old possessor represents the primary topic in the fol-
lowing two examples (7.78 and 7.80). In (7.78), the third-person singular PC -ne doubles upagentse
‘of Upaga,’ the inalienable possessor mañu ‘desire,’ which is the subject of an intransitive verb.
This sentence concerns Upaga, regarding how he felt after Nānda and Nandābala rejected his
request. Therefore, we analyze this associate to be the primary topic of the sentence. The sub-
ject of a sentence shifts from upagentse mañu ‘Upaga’s desire’ to pro, which refers to Upaga. This
non-overt subject shift supports the analysis that Upaga is the primary topic of the first sentence.

Nānda and Nandābala were preparing rice porridge. An ājīvika ascetic Upaga came by and
saw it. He requested it from them, but they rejected his request. They said they would
236
give it to the most brilliant among the sages.
upage-ntse mañu kärstāte-ne ṣañ ytāri [a6] masa ||

Upaga-gen desire destroy.pst.mid.3sg-3sg own way.acc go.pst.act.3sg
‘The desire of Upaga was destroyed, (and he) set out on (his) way.’
(B107a6; trans. by CEToM; prose)
To answer the QUD “What happened?” the author sets up a sub-QUD “What happened to Upaga?”
and answers it here.
(7.79) Summary of (7.78)

• QUD: What happened to Upaga?
• Focus: mañu kärstāte- ‘(his) desire was destroyed’
• Topic expression/referent (primary): upagentse/Upaga
In the second example (7.80), the clitic -ne [3sg] doubles araṇemiñ lānte ‘of King Araṇemi’ who is
the inalienable possessor of the discourse-new referent (pit ‘gall’). This sentence concerns King
Araṇemi describing what happened to him after hearing his son Uttara. Therefore, King Araṇemi
is the primary topic. This primary topic is continued into the following sentence as the omitted
subject of klāya ‘(he) fell.’

Prince Uttara is seeking help and speaking to his father, King Araṇemi: “My father, o lord,
take me away from these Rākṣasas! You are still alive, but they will now devour me.”
te keklyau«ṣo»rmeṃ araṇemi-ñ lā[b5]nte pit maiwāte-ne

dem.n hear.abs Aranemi-gen king.gen gall tremble.pst.mid.3sg-3sg
k(eṃ)t-sa klāya •
earth-perl fall.pst.act.3sg
‘Having heard this, King Araṇemi fainted (lit. King Araṇemi’s gall trembled) [and] fell
to the ground.’
237
The QUD of this discourse is “What happened?” (QUD1 ). To answer this QUD, the author first sets
up a sub-QUD “What happened to King Araṇemi?” (QUD1A ), which serves as a bridge to another
sub-QUD: “What did King Araṇemi do?” (QUD1B ).
(7.81) Summary of (7.80)

• QUD: What happnned to King Araṇemi?
• Pragmatic presupposition: Something happened to King Araṇemi.
• Pragmatic assertion: X = (his) gall trembled
• Focus: pit maiwāte- ‘(his) gall trembled’
• Topic expression/referent (primary): araṇemiñ lānte/King Araṇemi
The following example likewise shows the doubling of a possessor of an intransitive subject. How-
ever, due to the lack of context, it is uncertain whether the possessor is topical or not.40
(7.82) [TB] Context: N/A
/// auloñ cpī sätk(e)ntär-ne lyitkwä-nmā sruke-māne ku se

blood.vessel.pl this.gen.sg spread.npst.mid.3pl-3sg tube-pl die.npst-ptcp rel
lä///
‘… his blood vessels (of the body and) tubes spread out. He (who is) dying …’
(B139a3; verse; [7¦7]×4? )
In this example, the third-person singular PC -ne doubles cpī (for cwi) ‘his,’ which is the inalienable
possessor of auloñ … lyitkwänmā ‘blood vessels and tubes.’
40. Cf. Schmidt (1974: 278): “Der einzige mediale Beleg aus Toch.B findet sich in einem durch grosse Lücken un-
durchsichtig gewordenen Kontext, der keine sichere Übersetzung erlaubt.”
238
7.8.3.2 Tocharian A
In the following example (7.83), the third-person singular pronominal clitic -äṃ seems to double
āṣānikyāp bodhisatvāp ‘of the venerable Bodhisattva.’ This associate represents the (inalienable)
possessor of puk marmañ ‘all of the veins,’ which is the subject of the intransitive verb protkar- ‘X
is filled (with sufferings)’. Although it is unclear whether he has already been introduced to the
discourse, it seems that the Bodhisattva is the primary topic of the sentence since the subject of
klā ‘(he) fell’ is likely to be pro referring to the Bodhisattva.

nā – – (kāp)ñ(e) āriñc pācar ku yal (täm ya)kṣe śwā(ts)i eṣṣ-äm
tñik anaprä ṣakk ats lyā āpsā – – m •
“O father, dear to (our) heart, why does (he) give us to the yakṣa as a food? In front of you,
surely, […] the limbs.”
cam klop-yo āṣāniky-āp bodhisatv-āp puk (ma)rmañ

dem suffering-ins venerable-gen Bodhisattva-gen all vein.nom.pl(?)
protkar-äṃ – – tkan-ā klā •
be.filled.act.3pl-3sg earth-perl fall.pst.act.3sg
‘All of the veins of the venerable Bodhisattva are filled with that suffering. […] (he) fell
on the ground.’
(A356b4; prose?)41
This sentence immediately follows the Bodhisattva’s sons’ speech. The QUD of this discourse
is “What happened?” (QUD1 ). The author sets up and answers a sub-QUD “What happened to the
Bodhisattva?” in this passage.
(7.84) Summary of (7.83)

• QUD: What happened to the Bodhisattva?
• Focus: cam klopyo puk (ma)rmañ protkar- ‘all of (his) veins are filled with the suffering’
41. A407 a4 attests a parallel passage: /// (āṣāniky)āp bodhisatvāp puk marmañ /// ‘All of the veins of the venerable
Bodhisattva …’
239
• Topic referent (primary): the Bodhisattva
• Topic expression (primary): āṣānikyāp bodhisatvāp ‘of the venerable Bodhisattva’
In the following example (7.85), the third-person singular PC -ṃ [3sg] doubles the nominal ex-
pression (wa)ṣ(t) lmontāp ‘of the householder,’ which represents an inalienable possessor of śäṃ
‘wife.’ This associate is discourse-old. It seems to be the primary topic of the sentence, but it is
not secure because of the damage on the manuscript.

/// säm waṣt lmo kuryaru kā(ma)t ā ///
‘[…] the householder seized the goods […]’
/// (wa)ṣ(t) lmont-āp śäṃ ekrots tāka-ṃ – k· pyo śo

house sit.ptcp-gen wife poor.f.nom.sg cop.pst.act.3sg-3sg
‘The householder had a poor wife (lit. there was a poor wife to the householder) […].’
(A435b3; prose? )
Table 7.5 summarizes the examples discussed. I have shown that when a PC doubles a posses-
sor of an unaccusative subject, if there is sufficient context available, the possessor consistently
represents the primary topic of the utterance. I have observed that a sentence that immediately
follows may switch its subject to a pro without introducing an overt nominal expression. This
suggests that the omitted subject (pro) represents the continued topic of the sentence and that
the topic of the preceding sentence is not the possessum but the possessor doubled by a PC.
7.8.4 Miscellanea
Finally, I find three examples of doubling an indirect object (beneficiary) and two examples of
doubling a possessor of an indirect object (source and location, respectively).
7.8.4.1 Doubling of an indirect object
In the first example (7.86), the plural clitic -me doubles an indirect object (beneficiary; 1pl) of a
copula. The primary topic of this sentence is yes upādhyāyai ‘you masters,’ and wesi ‘to us’ serves
240
(7.78) (7.80) (7.82) (7.83)
Language TB TB TB TA
Genre Prose Prose Verse Prose?
Associate upagentse araṇemiñ lānte cpi āṣānikyāp bodhisatvāp
the venerable
Gloss Upaga’s King Araṇemi’s his
Bodhisattva
[+animate] [+animate] [+animate] [+animate]
Animacy
[+human] [+human] [+human] [+human]
Person 3rd 3rd 3rd 3rd
Number singular singular singular singular
Grammatical
A part of subject A part of subject A part of subject A part of subject
Function
(inalienable) (inalienable) (inalienable) (inalienable)
Semantic Role
possessor possessor possessor possessor
Possessum mañu pit auloñ … lyitkwänmā puk (ma)rmañ
(subject) ‘desire’ ‘gall’ ‘blood vessels (and) tubes’ ‘all of the veins’
Is an associate
No No No Yes
pronominal?
Is an associate
No No No ?
discourse-new?
Is an associate
Yes Yes ? Yes
a primary topic?
Table 7.5: Clitic doubling of a possessor of an intransitive subject in TB and TA
as the secondary topic.

500 disciples of Nadīkāśyapa and Gayākāśyapa all sat down on their knees and spoke to
them:
yes upādhyāya[i] wes-i saiym waste ṣeycer-me

nom.2pl master.nom.pl 1pl-gen support protection cop.impf.act.2pl-pl
eśane klausane ṣeycer-me kartstse yolo lkā-tsi klyaus-si-sa

eye.du ear.du cop.impf.act.2pl-pl good bad see-inf hear-inf-perl
kartstse kälā-tsi yesan-meṃ mañu-sa ṣeyem

good obtain-inf 2pl-abl desire-perl cop.impf.1pl
“You masters (= Nadīkāśyapa and Gayākāśyapa) were the support (and) protection to us;
241
(Youpl ) were the eyes (and) ears for us to see (and) hear a good (and) a bad thing. We had
(lit. We were with) a desire to obtain the good from youpl .”
(B108a6; prose)
The QUD of this utterance is QUD1 “What were the addressees (Nadīkāśyapa and Gayākāśyapa) like to
the speakers (their disciples)?” This question is set up by the speakers as part of a strategy to answer
another QUD explicitly raised by Nadīkāśyapa and Gayākāśyapa in a5: ostmeṃ lantsi camñcer mā
wat (wesäṃ)mpa? ‘Can youpl leave the house (i.e., become a monk) with us?’ (QUD0 ). To answer
this question, the disciples first set up this QUD1 , which serves as a bridge to the following QUD
“What did the speakers desire to do?” (QUD2 ).
(7.87) Summary of (7.86)

• QUD: What were the addressees (Nadīkāśyapa and Gayākāśyapa) like to the speakers (their
disciples)?
• Focus: saiym waste ṣeycer- ‘(youpl ) were the support and the protection’
• Topic expression/referent (primary): yes upādhyāya[i]/addressees (= Nadīkāśyapa
and Gayākāśyapa)
• Topic expression/referent (secondary): wesi/speakers (= 500 disciples of Nadīkāśyapa
and Gayākāśyapa)
In the second example (7.88), the second-person singular clitic -c [2sg] represents an indirect
object (beneficiary) of lakle … √yām ‘to make suffering to X; torture X.’ There is a topic shift be-
tween 2d and 3a. The primary topic is twe ‘yousg ’ in 2d, but it shifts to piś cmelaṣṣeṃs ‘(the living
beings) of five births’ in 3a. The independent personal pronoun ci ‘yousg ’ seems to represent the
secondary topic of this sentence.
(7.88) [TB] Context (Buddhastotra):

piś cmelaṣṣeṃs pernesa ¦ mā we wīna kälpāṣṣit •
pelaiknentse pelkiñ no ¦ ot ṣpä ramer arṣit twe • 2
‘For the sake of (living beings) of the five births, you (the Buddha) did not obtain pleasure.
242
[2c] For the sake of the law, you immediately abandoned (it).’ [2d]
snai kkarūṃ cai onolmi ¦ amaukacci yolo-sa •

without compassion dem.m.nom.pl living.being.nom.pl unceasing.nom.pl evil-perl
saim pärmaṅk cī śaiṣṣe-ntse ¦ lakle räskre [a4] yāmṣiyeñ-c •

protection hope acc.2sg world-gen suffering bitter make.impf.act.3pl-2sg
‘Without compassion, these living beings, (who are) unceasing with an evil (thought), [3a]
harshly tortured yousg (lit. ‘did bitter suffering [to] yousg ’),42 the protection (and) hope
of the world. [3b] ’
(B231a4; verse; [7¦7] × 4; trans. based on Thomas 1957: 65 and Schmidt 1974: 129)
The QUD of this discourse shifts from QUD1 “What did you do?” to QUD2 “What did the living beings
do to you?.” QUD2 is answered in detail in pādas 3b, 3c and 3d (7.90).
(7.89) Summary of (7.88)

• QUD: What did the living beings do to yousg ?
• Focus: lakle räskre yāmṣiyeñ- ‘made the bitter suffering’
• Topic expression/referent (primary): cai onolmi/the living beings
• Topic expression/referent (secondary): saim pärmaṅk cī śaiṣṣentse/addressee (= the
Buddha)
(7.90) [TB] Continuation of (7.88)
ṣarne paine kärsnoyeñ-c ¦ inte kc= eśne tsaknoyeñ-c

hand.du foot.du cut.off.impf.act.3pl-2sg when indf eye.du pierce.impf.act.3pl-2sg
(•)
mahūr-tstsana āstäṃ toṃ ¦ empalkaicci kärsnoyeñ-c

diadem-adjz.f.pl head.acc.pl dem.f.pl unworried.nom.pl cut.off.impf.act.3pl-2sg
:3
42. Cf. kṣānti ‘forgiveness’ with √yām ‘to do, make’ means ‘to do forgiveness to X; forgive X’ (e.g., B34a5 yāmṣa
cauacc kṣānti ‘[S/he] did forgiveness to him; [S/he] forgave him’).
243
‘[They] were cutting off yoursg hands (and) feet, and then (lit. whenever) [they] pierce
yoursg eyes. [3c] Without any fear, [they] were cutting off those crowned heads of yourssg .
[3d] ’
(B231a4; verse; [7¦7] × 4; trans. based on Schmidt 1974: 129)
The following example translates Karmavibhaṅga §32 (7.91). The associate of -ne [3sg] is alyek
īkene ykuweṣepi ‘(to him) who has gone to another place,’ which translates Skt. deśāntaragatasya
‘id.’ The function of the clitic seems to double the indirect object (beneficiary/location) of the
intransitive pakṣtär- ‘(the deed) ripens for/in X.’
(7.91) Karmavibhaṅga §32 (Kudō 2004: 88f.):
i. tatra katamat karma deśāntaravipākam.

‘Herein, what is the deed that has results in another country?’
ii. ucyate.
‘It is said:’
iii. yat karma tasminn eva janmāntare vā deśāntaragatasya vipacyate.

śubham aśubhaṃ vā tat karma deśāntaravipākam […]
‘Whatever deed matures for him who has gone to another country already in this
life or in another (life), whether (it is) good or bad, that deed has results in another
country.’
iv. idaṃ karma deśāntaravipākam

‘this is the deed that has results in another country.’
(7.92) [TB] Context (cf. Karmavibhaṅga §32):

The deed which leads to a non-specific rebirth is discussed.
kᵤse no sū yāmor ¦ alyek ī«ke»-ne yāmträ :

what conj dem.m.nom.sg deed other.acc.sg place-loc do.subj.mid.3sg
alyek ī(ke)-ne ¦ pke-lñe tuntse yänmāṣṣäṃ :

other.acc.sg place-loc ripen-nmlz dem.n-gen obtain.npst.act.3sg
244
[b3] mäkcwi yāmor-ntse ¦ śäktalye āläṃ ktowä (:)
rel.gen deed-gen seed differently strew.ptcp.m.nom.sg
alyek īke-ne ¦ ykuweṣepi pakṣtär-ne 5

other.acc.sg place-loc go.ptcp.m.gen.sg ripen.npst.mid.3sg-3sg
‘But what deed [is] it, [if one] does [it] in a different place, [5a] (he) will obtain its maturity
at the different place? [5b] (Answer:) The seed of whosesoeverj deed, sown in a different
(place), [5c] will mature for himj who has gone to the different place. [5d] ’
(PKAS7Bb3; trans. based on Sieg 1938: 9; verse; [5¦7]×4)
The primary topic of this sentence seems to be yāmorntse ‘of the deed (which has results in an-
other country)’ (cf. Skt. karma deśāntaravipākam). This example is puzzling since alyek īkene
ykuweṣepi ‘(to him) who has gone to another place’ does not seem topical, but it still receives
doubling of the PC -ne.
The last example is A213 b3, corresponding to YQ II.5 a7–8.
(7.93) [TA] Context (cf. YQ II.5 a7–843 ):

/// cäṃ • bādhari brāhmaṃ kuc kotraṣ • kospreṃ puklyi ko(s)pr(eṃ) manarkāśśi śāsträntu ākläṣ •
mrāc kus mä ///
(The brahmin Bādhari is speaking to his disciples:) (“If all of these thirty-two signs are
complete in their entirety on his [= the Buddha’s] body, then you, standing in front of
him, only ask him questions without a hint in your minds as follows:) “As for Bādhari the
brahmin, what is his descent? How old is he? How many brahmin youths does he teach
the Śāstras? What is ‘the top’? (What is ‘falling from the top’?)””
/// (kupre)-n(e) [subject säm ] ya(s-ä)ṃ (caṣ pe)nu pärklune •

if-comp dem.m.nom.sg 2pl-gen dem.acc.sg also question
sne (tä)ṅk-l-une atäṅkät wätkāṣṣ-äm cam yas

without hinder-gdv-nmlz unhindered answer.subj.act.3sg-pl dem.acc.sg nom.2pl
wäṣpā wä(tkālts tämne wäkn-ā kakmunt puk knān-mānänt
truly certainly thus way-perl come.ptcp.m.acc.sg all know-ptcp.m.acc.sg
245
ptāñkät pkärsäs)
buddha.lord know.imp.act.2pl
“If he (is able to) answer these questions (lit. this question) to youpl immediately and
without hesitation, then youpl (must) surely (and certainly recognize him as Tathāgata
and the all knowing Buddha-god.)”
(A213b3; trans. based on Ji, Winter, and Pinault 1998; prose)
In this example, the plural PC -äm [pl] doubles the discourse-old associate ya(sä)ṃ ‘youpl (= Bād-
hari’s disciples).’ This associate is the indirect object (addressee) of the transitive verb wätkāṣṣ-
‘(if s/he) answers (these questions) to X.’ The primary topic of this sentence is säm ‘he,’ referring
to the Buddha, and the associate is not the primary but the secondary topic of this sentence. This
subordinate clause precedes a matrix clause whose subject is overtly marked by yas ‘youpl (= Bād-
hari’s disciples),’ suggesting that a primary topic shifted from the Buddha to Bādhari’s disciples.

Sentence topic: this one (= the Buddha)

• If he can immediately answer Bādhari’s questions without hesitation, he should be rec-
ognized as the Buddha.

• …

The QUD of this utterance is QUD1 “What should he do to us regarding these questions?,” which offers
a condition to the following QUD2 “What should we do to him?”
(7.94) Summary of (7.93)

• QUD: What should he do to us regarding these questions?
• Focus: sne (tä)ṅklune atäṅkät wätkāṣṣ- ‘(he) immediately answers (the questions to
you) without hesitation’
• Topic expression (primary): säm/the Buddha
• Topic expression (secondary): ya(sä)ṃ/addressees (= Bādhari’s disciples)
43. YQ II.5 a7–8: kupre-ne säm yasäṃ caṣ penu pärklune [a8] (• sne täṅklune atäṅkät wätkāṣṣ-äm cam yas wäṣpā wätkālts
tämne) w(ä)knā kakmunt puk knānmānänt ptāñkät pkärsäs
246
• Topic expression (secondary): (caṣ pe)nu pärklune/the series of questions asked to
the Buddha
7.8.4.2 Doubling of a possessor of an indirect object
There are two examples of doubling a possessor of an indirect object (source and location, respec-
tively). Again, there is ambiguity about whether the PC doubles a possessor or a source/location
when they are in the same person and number. The first example translates Karmavibhaṅga §76.
(7.95) [TB] Karmavibhaṅga §76
i. Katame daśa guṇā gandha-pradānasya?

‘What are the ten virtues of giving incense?’
ii. Ucyate:
‘It is said:’
iii. Gandha-bhūto bhavati lokasya.

‘One becomes like incense to the world.’ (1)
iv. Ghrāṇendriyaṁ viśudhyati.

‘One’s faculty of smell is purified.’ (2)
v. Kāya-daurgandhyam apaiti.
‘Bad odor of the body disappears.’ (3)
vi. Saugandhyam prādur-bhavati.

‘Good odor appears.’ (4)
vii. Daśa diśaḥ śīla-gandhaḥ pravāti.

‘The sweet fame of one’s virtue blows in (all) ten directions.’ (5)
viii. …
(7.96) [TB] Context (Karmavibhaṅga §76):

10 advantages of giving perfumes are listed:
tu(ñe —) ta(tākau — cmel-a)-n(e) :

perfume.nom.sg cop.ptcp.m.nom.sg birth-pl-loc
247
mele(ṃ)-ṣṣe indri cpi mā kauṣträ :
nose-adjz sense dem.m.gen.sg neg destroy.npst.mid.sg
kektseñ-meṃ c(p)i [a2] karttse were yaṃ

body-abl dem.m.gen.sg good odor go.npst.act.3sg
kektseñ-meṃ cpi yolo mā warṣä(ṃ)-ne

body-abl dem.m.gen.sg bad.nom.sg neg smell.npst.act.3sg-3sg
kälymi-kälymi ṣäp cpi papāṣṣorñe-ṣe were kartts«e»

direction-direction conj dem.m.gen.sg moral.behavior-adj odor good
ya(ṃ)
go.npst.act.3sg
‘One will be(come like) a per(fume) in (the rebirths). His sense of smell (lit. pertaining to
the nostrils) is not destroyed. Good odor comes (out) from his body. Bad (smell) does not
come out (lit. [he] does not smell bad) from his body.44 And the good odor of his virtuous
observance goes in every direction.’
(PKAS7Na2; trans. based on CEToM; prose)
The associate of the pronominal clitic -ne [3sg] in this example is ambiguous since the indirect
object (source) kektseñmeṃ ‘from the body’ and its inalienable possessor (cpi) are both the third-
person singular. The latter is likely to represent the primary topic of this sentence as this passage
discusses how he will attain benefit by giving perfume. If so, the third person singular clitic -ne
doubles the primary topic.
In the second example (7.97), the third-person singular PC -äṃ [3sg] doubles either the loca-
tion (ka(pśi)ññaṃ ‘in the body’) or, more likely, the possessor cami ‘his,’ which refers to ācoyis ‘of
the embryo.’ This referent is discourse-old while the possessum ka(pśi)ññaṃ ‘in the body’ seems
44. All of the attested examples of the present stem TB warṣṣä-/warske-act (Present II as per Malzahn 2010: 887; IX
as per Peyrot 2013b: 821; and IXa as per Krause 1952: 289) are intransitive. They usually accompany a predicative
adjective and mean ‘(something/somebody) smells Adj’ (e.g., 7.o and 7.p).
(7.o) [TB] sū wassi ykāk swāre warṣṣäṃ ‘this clothing still smells good.’ (PKAS6Cb4)
(7.p) [TB] askwaś[i] rano pupaṃñ warskeṃ ‘also the darbha-grass smells bad,’ (B308b4; Peyrot 2013b: 821 n. 850).
The example (7.96) also carries a predicative adjective (yolo ‘bad’). It seems that the indirect object cpi kektseñmeṃ
‘from his body’ in this example is an adjunct to the intransitive (i.e., lit. ‘(He) does not smell bad from his body.’).
248
discourse-new. It is not clear whether this associate is the primary or the secondary topic of the
sentence.

cesäm ṣpät koṃsā kälytär ok prākroneyaṃ /// (ca)mi ācoyis mācri kātsaṃ ṣurmä – – – räṣ
‘For these seven days (the embryo? ) still stays in the firm state. […] in the womb of this
embryo’s mother […]
tämne wäkn-ā aneñcāṣ kākätkunt [b6] /// t·eñc

thus way-perl from.inside rise.ptcp
kus-ne cam-i āñc ka(pśi)ññ-aṃ wu lotas ruseñc-äṃ

rel-comp dem.m-gen down body-loc two hole.acc.pl open.npst.act.3pl-3sg
ṣom āsu wesis wcaṃ lyī wesi ///
one.acc dry.ptcp.m.nom.sg excrement.gen second.acc wet excrement
‘In this way, […] (the embryo) stepped out (lit. risen) from the inside (of his mother’s
womb) […] (those) who open two holes in the lower (part of) his body, one for dry excre-
ment (and) the other for wet excrement …’
(A150b6; prose? )
Determining the QUD of this passage is difficult because of the damage in the manuscript. We
tentatively set the QUD as What … do to the embryo?.
(7.98) Summary of (7.97)

• QUD: What … do to the embryo?
• Focus: āñc ka(pśi)ññaṃ wu lotas ruseñc- ‘open two holes in the lower part of (his)
body’
• Topic expression (primary): kusne/?
• Topic expression (secondary): cami/the embryo
To summarize, in this subsection I have examined the examples in which a pronominal clitic
doubles an indirect object (beneficiary) or a possessor of an indirect object (source or location)
(Table 7.6). In the former cases, the associates doubled by a PC represent the secondary topic of
249
a sentence. The PCs are likely to double the discourse-old topical possessors in the latter case.
However, we cannot exclude an alternative analysis since the possessor and the possessum are in
the same person and number.
250
(7.86) (7.88) (7.92) (7.93) (7.96) (7.97)
Language TB TB TB TA TB TA
Genre Prose Verse Verse Prose Prose Prose (?)
Associate wesi saim pärmaṅk cī śaiṣṣentse alyek īkene ykuweṣepi ya(sä)ṃ cpi cami
you, the protection (him) who has gone
Gloss to/of us to youpl his his
(and) hope of the world to a different place
[+animate] [+animate] [+animate] [+animate] [+animate] [+animate]
Animacy
[+human] [+human] [+human] [+human] [+human] [+human]
Person 1st 2nd 3rd 2nd 3rd 3rd
Number plural singular singular plural singular singular
Grammatical
indirect object indirect object indirect object indirect object part of an adjunct part of an adjunct
Function
(inalienable) (inalienable)
Semantic Role beneficiary beneficiary beneficiary addressee
possessor possessor
āñc ka(pśi)ññ-aṃ
251
kektseñmeṃ
Possessum N/A N/A N/A N/A ‘on the lower (part
‘from the body’
of) the body’
Transitivity
intransitive transitive intransitive transitive intransitive transitive
of the host
Is an associate
Yes Yes No Yes Yes Yes
pronominal?
Is an associate
No No No? No No No
discourse-new?
Is an associate
No No No No Yes ?
a primary topic?
Is an associate
Yes Yes Yes? Yes No ?
a secondary topic?
Table 7.6: Clitic doubling of an IO or a possessor of an IO in TB and TA

7.8.5 Summary
In all of the examples whose pronominal clitic doubles a theme of a transitive verb, the theme
represents the secondary topic of a sentence. Its existence is pragmatically presupposed at the
time of the utterance, and the sentence updates the relationship between the primary and the
secondary topic.
When a theme of a transitive verb accompanies a possessor, a PC consistently doubles the pos-
sessor. I have shown that the possessors in the examples were all discourse-old and (inalienably)
possess discourse-new possessa. The possessors are thus topical—they are primary or secondary
topics depending on whether or not possessa sit in the subject position. When a possessum is
in the subject position, its possessor represents the primary topic, which may continue as the
pro in the following sentence. In contrast, when there is a separate external argument, the ex-
ternal argument is the primary topic, and the doubled associate represents the secondary topic.
When a PC doubles an indirect object, the indirect object is likely to be the secondary topic of the
sentence.
7.9 Conclusion
This chapter has shown that, in contrast to the previous treatments of the Tocharian pronominal
clitics, we must recognize at least two different types of doubling in TA and TB: Clitic Left/Right
Dislocation and Clitic Doubling Proper. They are used for different purposes and subject to dif-
ferent restrictions, supporting Hypothesis 2.
The two research questions that we tackled in this chapter are as follows: (i) What does clitic
doubling do in TA and TB? and (ii) Does it have any grammatical or semantic restriction(s)? For
(i), we have shown that CLLD and CLRD highlight a non-subject associate as the primary topic.
In contrast, CDP may represent a primary or secondary topic, depending on a separate external
argument. For (ii), we have seen that associates in CLRD/CLLD and CDP are topical. In order to
have this status, the referent’s existence at the time of the utterance must be presupposed.
Clitic doubling is optional in Tocharian. It contrasts with languages such as Tundra Nenets, where
252
object agreement is obligatory when an object is third-person and a primary or secondary topic
(Dalrymple and Nikolaeva 2011). Doubling of a third-person DO is also obligatory when it is top-
ical in Albanian (Kallulli 2008). Rather, it patterns with Persian indefinite specific objects, which
may optionally be marked by râ when topical (Dalrymple and Nikolaeva 2011). In Chichewa, an
object marker may cooccur with a full nominal expression only when the nominal expression is a
dislocated topic (Bresnan and Mchombo 1987). Tocharian CDP is animacy insensitive: Inanimate
DPs may show doubling in Tocharian. It contrasts with Romanian, where inanimate DPs do not
show doubling (Cornilescu and Dobrovie-Sorin 2008). Tocharian CDP is definiteness insensitive;
In Macedonian, DOs need to be definite for being doubled (Mišeka Tomić 2008); this is not the case
in Tocharian. Tocharian CDP patterns with Hittite in that it is optional and driven pragmatically
to mark a topic. It never doubles a focus constituent, as observed by Sideltsev (2011a, 2011b) for
Hittite.
253
CHAPTER 8
Conclusion
8.1 Conclusion
The goal of this dissertation was twofold. One was to produce the descriptive generalization of
the Tocharian pronominal clitics, and the other was to develop a morphosyntactic model that
best accounts for their empirical distribution. The model I developed revealed the fine-grained
distribution of the Tocharian PCs by predicting there is a gap in the data.
Chapter 2 reviewed the pronominal system of Tocharian. This chapter also showed that one
should take the ablative and allative secondary case markers in Tocharian A (anäṣ and anac) as
monomorphemic rather than bimorphemic formatives consisting of the third-person marker -
an- and -äṣ/-ac. Chapter 3 outlined the representative uses of the Tocharian PCs and reviewed
their multifunctionality. They were mostly compatible with the genitive-dative and accusative
independent forms. They may also represent (inalienable) possession relations, although they
rarely described kinship relations. Chapter 4 introduced theoretical premises on which I devel-
oped a morphosyntactic analysis. I assumed that syntax manipulates morphosyntactic feature
bundles, which receive phonological realization post-syntactically, and that pronominal clitics
only consist of person and number features, and they are defective goals when they agree with
some functional head.
Chapter 5 developed a morphosyntactic model, in which PCs realize person and number features
licensed by the functional head that introduces an external argument. This model accounts for
the multifunctionality of the Tocharian PCs in that the licensor finds person and number features
in various thematic positions. Furthermore, it predicts that the Tocharian PCs may not represent
254
the possessor associated with a transitive agent since the external argument is not in the domain
where the licensor looks for valued person and number features. The model also predicts that PCs
cannot represent the possessor of the nominal expression contained in another nominal expres-
sion since the embedded nominal expression and its possessor should be invisible to the licensor
of a PC by the time it merges with the structure. Furthermore, this model enables us to separate
unaccusative verbs in Tocharian without relying on the semantics of a verb.
Chapter 6 considered examples in which multiple arguments are pronominal. My analysis backed
up Adams’ (2015) finding in a more restrictive sense: when the IO and DO of a ditransitive verb are
both pronominal, PCs consistently refer to the IO. This distribution finds a natural explanation
since the licensor of a PC looks for valued φ-features and finds the IO before the DO. This analysis
also enables us to analyze the semantics and pragmatics of a referential null object in Tocharian,
a topic that awaits further research.
Chapter 7 treated cases in which a PC cooccurs with a nominal expression to which it refers.
Based on the Question Under Discussion framework of discourse, I identified the topic and focus
of each sentence. This chapter revealed that doubling of a nominal expression by a PC indicates
the doubled expression to be topical, either primary or secondary. In either case, discourse par-
ticipants presupposed the existence of a referent at the time of the utterance. Furthermore, I
briefly compared clitic doubling in Tocharian with that in other IE languages (e.g., Hittite) and
the languages in Balkan Sprachbund.
8.2 Further questions
Although this dissertation shed light on various aspects of the Tocharian PCs, it focused on their
synchronic status and had to set aside many questions for future research. As discussed in Chap-
ter 1, I believe it is now possible to seek answers to the questions regarding the diachronic aspect
of the Tocharian PCs. For example, scholars commonly distinguish genitive/dative (PIE *moy
[1sg], *toy [2sg]) and accusative (PIE *me [1sg], *te [2sg]) atonic personal pronouns in PIE (e.g.,
Fortson 2010: 143). In contrast, this distinction is presumably lost already in Proto-Tocharian. It
remains an open question as to what pre-Proto-Tocharian changes caused this distinction to be
255
lost.
Another outstanding question is regarding the position in which Tocharian PCs appear. Schol-
ars reconstruct PIE atonic personal pronouns to appear second after the first stressed element
(Wackernagel’s law; Delbrück 1878; Wackernagel 1892). In contrast, the Tocharian PCs almost
always appear after a finite verb, except for some limited cases in TA. There is still no agreed view
regarding how Tocharian PCs acquired their distributional pattern. Language contact could be a
cause (Peyrot 2019). One might, however, be inclined to consider whether any language-internal
mechanism could account for the change. The model I developed may explain the change, albeit
specific details need to be worked out in the future. I. G. Roberts (2010) analyzes the second-
position clitics as a D attracted to the left periphery by C’s Edge Feature. They move to the po-
sition they appear in by passing through the edge of Voice. Voice does not incorporate them
because they have other features than person and number features (e.g., prosodic and referential
features; Cardinaletti and Starke 1999), and thus, they are not defective goals to Voice (Chapter
4). However, when they somehow lose these features, they develop from D to φ and become de-
fective goals to Voice and start undergoing incorporation. In this way, diachronic loss of features
will give rise to the Tocharian pattern in which PCs cliticize to the T-Voice complex and occurs
immediately after a finite verb.
It is also an open question whether the animacy of an argument or the Person Case Constraint
(Bonet 1991; Anagnostopoulou 2005) plays a role in determining whether PCs represent the IO or
DO of a verb (Chapter 6). For example, which argument do PCs represent when the IO is inanimate
and the DO animate? My analysis predicts that PCs should represent the IO in such cases. A more
thorough corpus study will provide insight into this question.
Chapter 3 confirmed that PCs may express an inalienable possession relation, particularly ab-
stract concepts and (part-whole) body-part terms. In contrast, PCs representing kinship relations
are extremely rare, although part-whole and kinship relations usually pattern to constitute the
class of inalienable possession. Why Tocharian PCs rarely represent kinship relations is an open
question. Regarding this issue, it is worth highlighting that PCs representing alienable posses-
sions are also rare.
256
Chapter 7 considered the semantics and pragmatics of clitic doubling in Tocharian. However, it
did not answer why clitic doubling is rare in these languages. If language contact played a role,
we would find doubling more frequently. Neither TA nor TB seems to have grammaticalized clitic
doubling. Why did Tocharian not grammaticalize doubling as a morphosyntactic or pragmatic-
semantic marker? Subsequent research will provide further insight into this issue.
By examining the synchronic aspects of Tocharian PCs, I opened a way to understand their di-
achrony. The Tocharian PCs are attractive not just for Tocharologists or Indo-Europeanists but
also for general linguists specializing in any subfield. I hope this dissertation established the em-
pirical and theoretical bases and became the basis for further research that includes the various
perspectives of diverse scholars with different expertise.
257
CHAPTER 9
Appendix: Representative unaccusative verbs in Tocharian A

and B
9.1 Representative unaccusative verbs in TA and TB
This chapter lists representative roots in TA and TB that form an unaccusative verb (Table 9.1). I
will briefly review each of the roots collected.
9.1.1 TB √ār(ā)- ‘cease, come to an end’
The TB and TA root √ār(ā)- ‘cease, come to an end’ forms an unaccusative verb. The following
example (9.1) from TB contains a pronominal clitic referring to cey wnolmi ‘the human beings.’
This PC represents the possessor of yāmor śaul ṣpä ‘deed and life,’ which is the subject of aran- ‘X
will cease’.
(9.1) [TB] Function: possessor of an intransitive subject
kᵤse no cey wnolmi ¦ ket śaul

rel.nom conj dem.m.nom.pl human.being.nom.pl rel.gen life
nanautau ¦ yāmor rano pest nanautau ¦ kᵤce
disappear.ptcp.m.nom.sg deed conj ptcl disappear.ptcp.m.nom.sg what.acc
klautke-sa aiśa-(lyi :
manner-perl know-gdv.m.nom.pl
/// [a5] – ¦ tetemo)ṣä tumeṃ no pest ¦ yāmor śaul ṣpä

be.born.ptcp.m.nom.sg thereupon conj ptcl deed life conj
aran-me 1 :
cease.subj.act.3pl-pl
258
Language Root Gloss Section
1 TB √ār(ā)- cease, come to an end (§9.1.1)
2 TA √ār(ā)- cease, come to an end (§9.1.2)
3 TA √i-|kälkā- go (§9.1.3)
4 TB √kän- come about, occur, be fulfilled (§9.1.4)
5 TA √kän- come about, occur (§9.1.5)
6 TB √käly-|stäm(ā)- stand (§9.1.6)
7 TB √käsk(ā)- be scattered (§9.1.7)
8 TB √kulā- recede (§9.1.8)
9 TB √klänts- sleep (§9.1.9)
10 TB √klautk(ā)- turn, become (§9.1.10)
11 TA √trik(ā)- be confused; faint (§9.1.11)
12 TB √nas-|tāk(ā)- be, become (§9.1.12)
13 TA √nas-|tāk(ā)- be, become (§9.1.13)
14 TA √pränk? - restrain oneself (§9.1.14)
15 TA √prutk(ā)- be shut, be filled (§9.1.15)
16 TB √plätk- overflow, develop, arise (§9.1.16)
17 TB √plu- float, fly, soar (§9.1.17)
18 TB √mäsk- be (§9.1.18)
19 TA √mäsk- be (§9.1.19)
20 TA √lotkā- turn, become (§9.1.20)
21 TB √wāk(ā)- differ (§9.1.21)
22 TB √ṣäm-|läm(ā)- sit (§9.1.22)
23 TA √sätk(ā)- spread out (§9.1.23)
24 TB √si-n- (mid) satiate oneself; be depressed (§9.1.24)
(act) satiate
25 TA √si-n-
(mid) satiate oneself; be depressed
26 TB √spālkā? - ±strive actively/forcefully for (§9.1.25)
27 TB √tsäm(ā)- grow, increase, come into being (§9.1.26)
28 TB √tsälp(ā)- pass away, be released, be redeemed (§9.1.27)
Table 9.1: Representative unaccusative verbs in TA and TB
‘But who are the beings whose life has been disappeared (and whose) deed has completely
disappeared also? With what manner are they to be recognized? [19a] … having been
(re)born, thereupon, however, their deed and life will come to an end completely. [19b] ’
(PKAS7Ga5; trans. based on CEToM; verse; [5¦5¦8¦7]×4)
1. aran-me seems to be a misspelling of āran-me, which has the initial accent for Subjunctive V (Malzahn 2010:
283).
259
9.1.2 TA √ār(ā)- ‘cease, come to an end’
We also find a PC representing th possessor of the subject of the root √ār(ā)- ‘cease, come to an
end’ in TA. In (9.2), the third-person singular PC -äṃ represents the (inalienable) possessor of śol
‘life,’ which is the subject of aratr- ‘X ceases, X comes to an end.’
(9.2) [TA] Function: possessor of an intransitive subject
[a1] /// ṣ pñi-ṃtu aratr-äṃ śol ///

virtue-pl cease.npst.mid.3sg-3sg life
‘… virtues. His life comes to an end.’
(A165a1; verse?)
In example (9.3), the third-person singular PC -äṃ represents the possessor of lyalypäntu ‘deeds,’
which is the subject of the intransitive verb āreñc- ‘X will cease, X will come to an end.’
• känta-ntu-yo pukl-ā kätkeñc cam-i kapśiññ-äṣ śwāl lātäṅkā-tsi mā

100-pl-ins year-pl pass.subj.act.3pl dem.m.sg-gen body-abl flesh cut.off-inf neg
aratär cam lāt··ntaṃ lyalyp-än(tw)-ā(śś)i ///
stop.npst.mid.3sg dem.m.acc.sg deed-pl-gen
[b1]ntaṃ lyalyp-äntu āreñc-äṃ

deed-pl stop.subj.act.3pl-3sg
‘The years will pass by hundreds, (but) one does not stop cutting flesh from his body (lit.
cutting off flesh from his body will not stop). This … of the deeds … his deeds will come
to an end.’
There is another occurrence of a PC, lyalypäntu ‘deeds’ and √ār(ā)- (9.4). Although the context
of this example is missing, it seems that the PC -äm [pl] represents the possessor of lyalypäntu,
which is the subject of arantr- ‘X ceases, X comes to an end.’
260
(9.4) [TA] Function: possessor of an intransitive subject (?)
[a2] – lyalypä-ntu arantr-äm ///

deed-pl cease.npst.mid.3pl-pl
‘… their(?) deeds come to an end …’
(THT1308aa2; prose?)
Example (9.5) contains the first-person singular PC -ñi, referring to nirdhane ‘Nirdhana [a name
of a brahmin].’ It represents the (alienable) possessor of puk niṣpalntu ‘all of the possessions,’
which is the subject of the intransitive verb ārar- ‘X ceased, X came to an end.’ This example also
supports that TA √ār(a)- forms an unaccusative verb.
[a2] /// (śäk) w(e)-pi pkul wsā elant ¦

10 2-ptcl year give.pst.act.1sg alms
ārar-ñi puk niṣpal-ntu :

cease.pst.act.3pl-1sg all possession-pl
‘I have been giving alms for twelve years; all my properties are gone.’
(A215a2; trans. based on Ji, Winter, and Pinault 1998: 45; verse; [7¦7]×4)
YQ I.6 b1 (9.6) attests a parallel passage.
/// – ¦ pu kis puk ākāl knäs-si pke :

all.gen all wish be.fulfilled-inf intend.pst.mid.1sg
śäk we-pi pu kul wsā elant ¦ ārar-ñi pu[b2](k niṣpal-ntu :)

10 2-ptcl year give.pst.act.1sg alms cease.pst.act.3pl-1sg all possession-pl
[Bādhari speaking to Nirdhana:] “I intended every wish to be fulfilled for everyone. I have
been giving alms for twelve years; all (properties) of mine are gone …”
(YQ I.6b1; trans. based on Ji, Winter, and Pinault 1998: 45; verse [7¦7]×4)
261
9.1.3 TA √i-|kälkā- ‘go’
The TA root √i-|kälkā- ‘go’ forms an unaccusative verb, which may accompany a PC that represents
the possessor of the subject. For example, kalkaṣ- ‘X (will) go’ in (9.8), used metaphorically, ac-
companies the first-person singular PC -ñi that represents the inalienable possessor of the verb’s
subject ime ‘thought, mind.’ This sentence follows a passage that translates the Udānavarga 21.7.
(9.7) Uv. 21.7 (Tathāgatavarga; Bernhard 1965: 280)

na hi santaḥ prakāśyante viditvā lokaparyāyam /
ādeśayanto virajaḥ padaṃ śāntamanīṣiṇaḥ // 7
‘For, having understood the way of the world, the good ones are not illuminated, (they
are) calm (and) intelligent, indicating the dustless place.’
mā nu kraś pälkiñc ¦ ārkiśoṣṣi-s ym(e) ¦ kärso-r-äṣ

neg conj good.nom.pl appear.npst.act.3pl world-gen way know.ptcp-nmlz-abs
mā āksisa-māṃ ¦ sne twe lame knānmās [a5] – (:)2
neg instruct-ptcp without dust place knowing.nom.pl
/// tti ¦ weñam tāpärk ślokaśśi wram ¦ kos ne ime

speak.subj.act.1sg now strophe.gen.pl thing how.much comp thought
kalkaṣ-ñi :
go.subj.act.3sg-1sg
‘And having understood the way of the world, the good ones do not shine. (They are)
intelligent, not indicating the place without dust. [15a] … I will now tell (you) the sense of
the strophes as much as my memory goes. [15b]
(A218a5; verse; [5¦5¦8¦7]×4? )3
2. Sieg and Siegling (1933: 171 n. 5), followed by Bernhard (1965: 280), emend knānmāṣ as knānmānäṣ ‘knowing,
intelligent’ [nom.pl].
3. The TA passage mā āksisamāṃ ‘not teaching’ seems to translate Sanskrit adeśayanto ‘not indicating’, rather than
ādeśayanto ‘indicating’ (Sieg and Siegling 1933: 172).
262
Example (9.9) attests the same combination of īme ‘thought’ and √i-|kälkā- ‘go’ with a PC. In this
example, the first-person singular PC -ñi also represents the inalienable possessor of īme, which
is the subject of the intransitive verb kälkā- ‘X went.’
prasku lyalypur-äṣ ¦ mar märka ///

be.afraid.pst.act.1sg deed-abl neg law
/// tsopats wäl nunak «:»

great king again
raritwā kāvvi ¦ kos ne īme (kä)lkā-ñi :

translate.pst.act.1sg poem how.much comp awareness go.pst.act.3sg-1sg
kanis ā[b4]yāt(w)ā ¦
tune.gen.sg according.to
‘I was afraid of the deed. Don’t … the Law! … … O great king, again, [60b] I translated the
poem as much as my awareness went, [60c] following the tune.’
(A230b3; verse; [5¦7 + 5¦4¦6 + 5¦7 + 5¦4¦6])
The following example (9.10) contains the third-person singular PC -äṃ, which refers to an em-
bryo. This PC seems to represent the inalienable possessor of the missing subject, but this exam-
ple is not secure because of the lacuna in the manuscript.
(wikiñu)piñcinäs ṣ(pät k)oṃs-aṃ i – – [a6] /// m nas-l-une-yaṃ yiñc-äṃ ||

29th.m.acc.pl 7 day.pl-loc be-gdv-nmlz-loc go.npst.act.3pl-3sg
‘In the 29th week, … his … go into … existence.’

(A151a6; prose?)4
4. Cf. A115 a1: okāk prākär nasluneya(ṃ) yäṣ ‘(it) goes into a solid existence.’
263
9.1.4 TB √kän- ‘come about, occur, be fulfilled’
Our analysis in Chapter 5 identifies the TA and TB root √kän- ‘come about, occur, be fulfilled’ as
a root that forms an unaccusative. This root frequently accompanies TB ākālk and TA ākāl ‘wish’
as the subject and a pronominal clitic as the subject’s inalienable possessor.5
In example (9.11), the plural PC -me represents the possessor of akālk ‘this wish,’ which is the
subject of the intransitive verb knetär- ‘will X be fulfilled.’
mai no knetär-me ¦ ritau akālk laukaññe : 1 ||

ptcl conj be.fulfilled.subj.mid.3sg-pl wish.ptcp.m.nom.sg wish for.a.long.time
(Nandā [and] Nandabalā speaking to God Brahma): “But will our long-wished wish still
be fulfilled?” [1d]
(B107b1; verse; [5¦7]×4)
Moreover, the following examples attest a formulaic expression { paiykalñesa X-vārg | X-vārg paiykā-
mai } akālk kñitär-ñ ‘{ By the writing of the X-varga | I have written the X-varga }, may my wish be
fulfilled!’.
(9.12) [TB] PKAS 4A b2 (verse; [7¦7¦4]×4):

paiykalñesa drohavārg«†ä» ¦ akālk kñītär-ñ serkene ¦ po cmelaṣṣe :
5. The following example (9.q) contains a PC -ñ [1sg], a gerundive knelle ‘to be fulfilled,’ and akālk ‘wish,’ which
represents the subject of the gerundive. But this example does not support the unaccusativity of √kän- since the PC
is not hosted by the gerundive but by the finite copula star-.
(9.q) [TB] Function: possessor of a subject of a gerundive
(– – – – puwa)r-ne nauṣ yopu mā ṣpä akālk kne-lle star-ñ
fire-loc earlier enter.subj.act.1sg neg conj wish be.fulfilled.npst-gdv cop.npst.3sg
pañaktä[a6](-ññe ślok-sa) /// lareṃ pelaikne klyau-tsi
Buddha-adjz stroph-perl dear law hear-inf
(King Subhāṣitagaveṣin speaking to Indra who changed his shape to a yakṣa): “(But) if I enter into the fire
first, my wish to hear the dear law with a Buddha-strophe cannot be fulfilled.”
Example (9.q) follows Thomas’s (1983: 121 n. 12) restoration pañaktä[a6](ññe śloksa); klyautsi is for klyauṣtsi (Thomas
1983: 121 n. 13).
264
‘By writing of Drohavarga, may my wish be fulfilled in the cycle of all births!’
(9.13) [TB] PKNS 27 a3 (verse; [7¦7¦4]×4):

paikalñesa droha(vārg ¦ akālk) kñītär-ñ serkeṃne ¦ po cmela(ṣṣ)e(ṃ)
‘By writing of Drohavarga, may my wish be fulfilled in the cycle of all births!’
(9.14) [TB] PKAS 4A b4 (verse; [7¦7¦4]×4):

po krentaunaṃts śmoññai ce ¦ paiykalñesa smṛtivārg«†ä» ¦ akālk kñītär-ñ :
‘By the writing of Smṛtivarga, the foundation of all virtues, may my wish be fulfilled!’
(9.15) [TB] PKNS 27 a5 (verse; [7¦7¦4]×4):

po krentaunaṃts śmoññai ce ¦ paikalñesa smṛtivārg ¦ s= akālk kñitär-ñ
‘By the writing of Smṛtivarga, the foundation of all virtues, may this wish of mine be
fulfilled!’
(9.16) [TB] PKAS 5A b3 (verse; [7¦7¦4]×4):

/// [b3] paiykāmai ¦ cwī yāmorntse «o»kosa ¦ se= kālk kñītär-ñ :
‘I have written (the Anityavarga). Through the fruit of this deed, may this wish of mine
be fulfilled!’
(9.17) [TB] PKAS 5B a6 (verse; [7¦7¦4]×4):

(krodhavā)[a6]r(g) paiykāmai ¦ cwī yāmorntse okosa ¦ s= ākālk kñītär-ñ :
‘I have written the Krodhavarga. As the fruit of this deed, may this wish of mine be
fulfilled!’
(9.18) [TB] PKAS 5C a6 (verse; [7¦7¦4]×4):

nirvāṇavārg«†ä» śpālmeṃ ce ¦ paiykalñesa akālk«†ä» se ¦ nemc(ek) [a6] kñītär-ñ :
‘By writing this excellent Nirvāṇavarga, may this wish of mine be fulfilled for sure!’
(9.19) [TB] PKNS 25 and 26 b4 (verse; [7¦7¦4]×4):

nirvāṇavārg paiykāmai ¦ cew yāmorsa akālk se ¦ ne(mc)ek (k)ñītär-ñ ¦
‘I have written the Nirvāṇavarga. By the deed, may this wish of mine be fulfilled for
sure!’
Furthermore, we may list two additional examples, although their contexts are missing.
265
(9.20) [TB] M 500.1 a1 (verse; [7¦7¦4]×4):
/// nemceksa ¦ akālk kñītär-ñ«†ä» :
‘… for sure, may my wish be fulfilled!’
(9.21) [TB] PKAS 5D a5 (verse; [7¦7¦4]×4):

/// (nemñceksa ¦ akālk kñītä)[a6]r-(ñ)«†ä» :
‘… for sure, may my wish be fulfilled!’
9.1.5 TA √kän- ‘come about, occur’
TA also attests the combination of ākāl ‘wish,’ √kän-, and a PC. Example (9.22) contains the second-
person singular PC -cy, referring to Bṛhadyuti. It represents the inalienable possessor of ākāl
‘wish,’ which is the subject of the intransitive verb knitär- ‘May X be fulfilled.’
knitär-cy ākāl (kni)[b5]tär-ci ¦ mar klopa-ṃtt-yo

be.fulfilled.opt.mid.3sg-2sg wish be.fulfilled.opt.mid.3sg-2sg neg suffering-pl-ins
si(näṣtār :)
be.depressed.npst.mid.2sg
‘Your wish shall come true, it shall be granted to you, You will not despair through suf-
fering. [2b] ’
(A25b4; trans. based on Sieg 1944: 306 ; verse [7¦7]×4)
The following example (9.23) also contains a PC that represents the inalienable possessor of ākāl
‘wish.’ This nominal expression is the subject of the intransitive verb knatr- ‘X will be fulfilled’.7
6. Sieg (1944: 30): “Dein Wunsch soll in Erfüllung gehen, er soll dir in Erfüllung gehen, du sollst durch Leiden
nicht verzagen.”
7. One might wonder whether this PC represents the agent of the preterite participle rito (i.e., ‘the wish desired
by yousg ’), rather than the possessor of ākāl ‘wish.’ Indeed, Tocharian PCs may represent the agent of a preterite
participle, as we reviewed in Chapter 4. However, preterite participles always function predicatively with a finite
copula in such cases. Tocharian PCs never represent the agent of a preterite participle used attributively. I think this
is because External Merge of Voice triggers Transfer of the complement of the lower phase-defining head D. If the
agent of an attributively-used preterite participle resides in this Transfer-domain, it cannot move out of the phase
to incorporate to Voice.
266
ke-ne kus-ne naṣ nati muk tampe ¦

rel.gen-comp rel.nom-comp cop.npst.act.3sg strength zeal(?) strength
oñi[b5](cmolṣi ṣñi āñcä)m : ārwar yāmu-r-äṣ ¦ cam koṃ
by.nature own self prepared make.ptcp-nmlz-abs dem.m.acc.sg sun
tām tkan-ā ¦ pukmäs knatr-äm
dem.f.acc.sg earth-perl come.imp.act.2pl come.about.subj.mid.3sg-pl
rito ākāl ¦ plāntac ku leñciṃ ñemi-yo [b6] (1)
desire.ptcp.m.nom.sg wish rejoice.subj.act.2pl female.m.acc.sg jewel-ins
(King Mahendrasena ordering his ministers to announce to the messengers:) “Whoever
[of you] have power, zeal(?) , strength [and] manly (nature), having made (yourselves)
ready, come to this place on this day! Yourpl cherished wish will be fulfilled. You will
take delight in the jewel of women.”
(A66b5; verse [5¦5¦8¦7]×4)
In example (9.24), ākāl ‘wish’ is the subject of knäṣtär ‘X is fulfilled.’ It accompanies the first-
person singular PC -ñi, which refers to brahmadatte ‘Brahmadatta (king of Jambudvīpa).’ This PC
also represents the inalienable possessor of the subject ākāl ‘wish.’
jambudvip-ā kākmärtik ¦ wäl näṣ pācar ñom-yo [a2] /// w·

Jambudvīpa-perl master king nom.1sg father name-ins
ypamār ṣñi se-yo : rake cam-äṣ pānäsmār ¦ nmäs-māṃ
made.npst.mid.1sg own son-ins word dem-abl ask.npst.mid.1sg bow-ptcp
kapśñ-o śl= āñcālyi : mā nu ākāl knäṣtär-ñi ¦ penä
body-ins with hands.put.together neg conj wish fulfill.npst.mid.3sg-1sg say.imp
(The king Brahmadatta speaks:) “I [am] the ruler over the Jambudvīpa, the king, father
by name (only) [a2] ... I make through my son. A word from him I ask, bowing with my
body [and] with hands folded. But my wish is not fulfilled. Say [you], …”
(A71a2; trans. by CEToM; verse [7¦7]×4)
In A309 b2 (9.25), Carling, Pinault, and Winter (2009: 62) restore (knäṣt)r-äm ‘X is fulfilled’.8 If
their restoration is correct, this example also supports that TA √kän- forms an unaccusative verb.
8. This restoration, according to Malzahn (2010: 569), in fact goes back to Siegling’s personal copy.
267
[b1] /// ·t·ne ptāñkat mā

buddha.lord neg
[b2] ///··r-äm ākāl rito [b3] ///

wish desire.ptcp.m.nom.sg
‘… the Buddha-lord … not … Yourpl cherished wish is fulfilled.’
(A309b2; trans. based on CEToM; verse?)
9.1.6 TB √käly-|stäm(ā)- ‘stand’
In the following example (9.26), the third-person singular PC -ne refers to the inalienable pos-
sessor of kwipe ike ‘shame place,’ which is the subject of kalltärr- ‘X stands (up).’ This example
suggests that the TB verb kalltärr- is unaccusative.
ṣamāne-ntse yśelmi pälsko-ne tsa[a4]ṅkaṃ kwipe ike keᵤwco

monk-gen pleasure.pl mind-loc arise.subj.act.3pl shame place upwards
kalltärr-ne tu maś-ne eṅkastär
stand.npst.subj.mid.3sg-3sg dem fist-loc seize.npst.mid.3sg
nuskaṣṣäṃn-ne [a5] tu-ne swāralyñe yamastär krāke
squeeze.npst.act.3sg-3sg dem-loc pleasure make.npst.mid.act.3sg filth
län-ne saṅghā-träṅ(k)ä kätä[a6]ṅkäṃ
go.out.subj.act.3sg-3sg s. commit.a.sin.npst.act.3sg
‘If the sexual desire arises in the mind of a monk, (and) his the shame-place stands up,
(if) he seizes it in the fist and presses it, he finds pleasure in it (and) the filth of him runs
out, he commits SA.-sin.’
(B334a4; trans. based on Ogihara 2009; prose)
This phrase appears four times in the same manuscript (9.26, 9.27, 9.28, and 9.29).
(9.27) [TB] B334 a8: ṣamāne[a8]ntse yśelmi pälskone tsaṅkaṃ kwipe īke keᵤwco kalltärr-ne
(9.28) [TB] B334 b3: ṣamānentse yśel(mi pä)[b3]lskone tsaṅkaṃ kwipe īke keᵤwco kalltärr-ne
(9.29) [TB] B334 b7: ṣamāneṃntse yśe[b7]lmi pälskone tsaṅkaṃ kwipe īke keᵤwco kalltärr-ne
268
9.1.7 TB √käsk(ā)- ‘be scattered’
Example (3.49), repeated here as (9.30), suggests that TB √käsk(ā)- ‘be scattered’ forms an unac-
cusative verb. In this example, āśce ‘head’ is the subject of the intransitive verb käskaññītär- ‘X
is scattered.’ The third-person singular PC -ne, hosted by käskaññītär-, refers to Devadatta and
represents the inalienable possessor of the subject.
(9.30) (= 3.49) [TB] Function: possessor of an intransitive subject
laur-sa eñcwaññe ¦ tarne räskre ¦ tsopyeṃ-ne :

spike-perl of.iron top violent sting.act.impf.3sg-3sg
käskaññītär-ne ¦ waiptār āśce ¦ po lykaśke «:»
be.scattered.impf.mid.3sg-3sg apart head.nom.sg all small.nom.sg
‘With an iron spike, they violently pierced him the crown of his head. [74a] His whole head
was scattered apart [into] small [pieces]. [74b] ’
9.1.8 TB √kulā- ‘recede’
Based on the following examples, we may add the TB root √kulā- ‘recede’ to the list of roots that
form unaccusative verbs. In example (9.31), maiyo ‘power, strength’ is the subject of kuletär ‘X
recedes.’ In this example, the third-person singular PC -ne represents the inalienable possessor
of the subject.
ś(r)āwa-ṃne śāmñe ayā-ṣe ṣat yamaṣäle sana-tse yoñiye-ne tsapanale maiyo

śravanā-loc manly bone-adjz piece make.gdv enemy-gen city-loc crush.gdv power
˙
kuletär-ne
recede.npst.mid.3sg-3sg
‘In [the lunar mansion] Śrāvana a splitter of human bones [is] to be made; it has to be
crushed on the passage of the enemy, [and] his strength will recede.’
(PKAS8Ba4; trans. by CEToM; prose)
B21 b5 (9.32) attests the same combination of maiyyo ‘power, strength’, √kulā- ‘recede,’ and a PC.
In this example, maiyyo is also the subject of ku lātär- ‘X will recede,’ and the plural PC -me, which
269
seems to refer to pakwāreṃts ‘the evil ones,’ represents the inalienable possessor of the subject.
pakwāreṃ-mpa nauṣ ¦ /// (:)

evil.pl-com earlier
/// ¦ (att)s(ai)k maiyyo ¦ ku lātär-me 60

completely power recede.subj.mid.3sg-3pl
‘With the evil ones, earlier, ... [60c] … their power will recede completely. [60d] ’
(B21b5; verse; [5¦4¦3]×4)9
In (9.33), the first-person PC -ñ, referring to King Araṇemi, represents the inalienable possessor
of palsko ‘mind, spirit.’ This nominal expression is the subject of ku lā- ‘X has receded, diminished.’
sū ñi yärke śpālme(ṃ) ṣai ¦ pāramit-ne āyor-ṣṣe ¦

dem.m.nom.sg gen.1sg veneration excellent cop.impf.act.3sg virtue-loc gift-adjz
mā no ku lā-ñ palsko :
neg conj recede.pst.act.3sg-1sg mind
[King Araṇemi speaking to his wife]: “This was the highest veneration for me. And my
spirit has not diminished in the virtue of (giving) a gift (= generosity). [1b] ”
(B78a1; trans. based on Thomas 1957: 218; verse; [5¦5] × 2 + [7¦7¦5] × 2)10
B231 b1 (9.34) also attests the combination of palsko ‘mind, spirit,’ √kulā- ‘recede,’ and a PC. The
first-person singular PC -ñ in this example also seems to represent the possessor of the subject
palsko.
9. CEToM: ‘With the bad ones first .... their power will recede completely. [60d]’ Cf. Hackstein, Habata, and Bross
(2015: 79): “Mit den schlechten (Kämpfern) vorne … … fürwahr wird ihnen die Kraft nachlassen.”
10. Thomas (1957: 218): “In der Vollkommenheit des Gebens [d.h. in der Freigebigkeit] hat aber mein Denken [bis
jetzt] nicht nachgelassen.”
270
mā twe ceṃ-ne krämpitar ¦ mā ra palsko

neg nom.2sg dem.acc.pl-loc disturb.impf.mid.2sg neg also thought
kulyitär-ś :
recede.impf.mid.3sg-2sg
pelaikne-ṣṣe śaul śpālmeṃ ¦ cau-k twe ñyās-sa ñäṣṣitar •4

law-adjz life excellent dem.acc.sg-emp nom.2sg desire-perl wish.impf.mid.2sg
‘You were never angry at them, nor did your spirit diminish. [4c] The excellent life of the
law, you wished this with longing. [4d] ’
(B231b1; trans. based on Thomas 1957: 65; verse; [7¦7] × 4)11
In the following example, the second-person singular PC, referring to the Buddha, represents
the (inalienable) possessor of warkṣäl ‘energy,’ which is the subject of the intransitive verb kᵤlā-
‘X receded, diminished.’
ṣeme ṣṣeme ślok-sa kᵤce ¦ yältse-nma-sa karsta-tsi ¦

one.m.acc.sg one.m.acc.sg strophe-perl since 1000-pl-perl cut.off-inf
wsāsta āstaṃ :
give.pst.act.2sg head.pl
yetse tsaṅ-tsi kektseñ-meṃ ¦ ysār-a ṣis-(s)i (mrestī)[a4]we ¦ mā

outer.skin flay-inf body-abl blood-pl drain-inf marrow neg
ku lā-c warkṣäl :
recede.pst.act.3sg-2sg energy
‘Since, for each single strophe, you allowed the heads to be cut off by the 1,000 (people), [2b]
the outer skin to be flayed from the body, and the blood (and) the marrow to be drained,
your energy did not recede. [2c] ’
(PKAS4Ba4; verse; [7¦7¦4] × 4)
M 500.1 b1 (9.36) and B104 a1 (9.37) attest the same passage.
11. Thomas (1957: 65): “Du warst nie ärgerlich [wörtl. gestört, gehemmt] auf sie, auch liess dein Denken nie nach.
Nach dem Leben des Gesetzes als dem besten, nach dem eben begehrtest du [immer wieder] mit Verlangen.”
271
(ṣeme) ṣṣeme ślok-sa kᵤce ¦ yältse-nma-sa karsta-tsi ¦

one.m.acc.sg one.m.acc.sg strophe-perl since 1000-pl-perl cut.off-inf
wsāsta āstäṃ :
give.pst.act.2sg head.pl
yetse tsaṅ-tsi kekts(e)ñ-meṃ ¦ ysār-a ṣis-s(i) mr(estīwe ¦ mā)

outer.skin flay-inf body-abl blood-pl drain-inf marrow neg
ku l(ā)-c (warkṣäl :)
recede.pst.act.3sg-2sg energy
‘Since, for each single strophe, you allowed the heads to be cut off by the 1,000 (people),
the outer skin to be flayed from the body, and the blood (and) the marrow to be drained,
your energy did not recede.’
(M500.1b1; verse; [7¦7¦4] × 4)
/// (ṣi)s-si mrestīwe ¦ mā ku lā-c warkṣäl (:) ///

drain-inf marrow neg recede.pst.act.3sg-2sg energy
‘… the marrow to be drained, your energy did not recede.’
(B104a1; verse; [7¦7¦4] × 4)
In the following example, āk ‘(my) zeal’ is the subject of the intransitive verb kᵤloytär- ‘May X
recede!’ This example attests the first-person singular PC -ñ, that represents the inalienable pos-
sessor of the subject.
śīla-ṣṣana sälyeṃno ¦ prākre ysomo eñcīmar ¦ mā āk

moral.behavior-adjz.f.pl rule.pl firm altogether seize.opt.mid.1sg neg zeal
ku loytär-ñ 20-4
recede.opt.mid.3sg-1sg
‘May I seize the rules of moral bahavior firmly (and) completely! May my zeal not recede!
[24d] ’
(PKAS4Aa3; verse; [7¦7¦4] × 4)
272
Finally, the following sentence (9.39) contains the verbal complex ku lā-ne ‘X receded.’ The subject
of this verb is pilkw añmaṣṣe ‘self-view’ (corresponding to Skt. ātmadṛṣṭi; Thomas 1983: 143). The
third-person singular PC of this example seems to refer to brāhmaṇ(i) ‘the brahmin’ and represent
the possessor of the subject.
(9.39) [TB] Function: possessor of an intransitive subject (?)
te keklyauṣ-or-meṃ brāhma[a5]ṇ(i) ¦ śaul-ne (s)kw(añ)ñ(e)

dem hear.ptcp-abs-abl Brahman.gen.sg life-loc happiness
wikā-ne ¦ pilkw añma-ṣṣe ku lā-ne :
disappear.pst.act.3sg-3sg view self-adjz recede.pst.act.3sg-3sg
‘When he had heard this, the brahmin’s pleasure in life faded and his self-view dimin-
ished. [90a] ’
(B3a5; verse; [8¦7¦6]×2 + [4/5¦4/5] + [7¦6])12
9.1.9 TB √klänts(ā)- ‘sleep’
Example (9.40) shows that lkäntsan- (for kläntsan-) is an unaccusative. The subject of this verb
is pro, referring to kektseñ ‘body,’ and the restored third-person singular PC -n(e) represents this
subject’s (inalienable) possessor.13
12. CEToM: ‘When he had heard this, the Brahman lost his (pleasure) in life, and his view on his own self disap-
peared in him.’ Thomas (1983: 143) proposes to restore brāhmaṇ(e) [nom.sg], rather than brāhmaṇ(i) [gen.sg],
13. One might wonder whether this PC refers to kektseñ ‘body’ and represents the verb’s subject (i.e., ‘it [= the body]
sleeps’). However, there is no parallel example of a PC representing the subject of an intransitive verb while being
hosted by the intransitive verb itself.
273
(ke)kts(e)ñ mā palkṣträ ¦ sak-sa ṣp aiksnar

body neg be.burned.npst.mid.3sg pleasure-perl conj (all).together
lkäntsan-n(e) :
sleep.npst.act.3sg-3sg
āksau ṣpak kāccä(n ¦ pa)pāṣṣoṣ-äṃts toṃ

wake.ptcp.m.nom.sg conj rejoice.npst.act.3sg protect.ptcp-gen.pl dem.f.pl
skwa-nma 10[b4](4)
happiness-pl
‘The body is not burned, and thanks to the pleasure his (body) sleeps (well) altogether,
[14c] [and] having awoken he is all the more pleased. These [are] the goods of happiness
for those who have obeyed the rules [14d] ’
9.1.10 TB √klautk(ā)- ‘turn, become’
As examples (9.41), (9.42), and (9.43) suggest, TB √klautk(ā)- ‘turn, become’ forms unaccusative
verbs.
(ṣañ lä)kle-nta warpa-tsi ¦ waśīr klautkoy-ñ ara(ñce) ¦

own suffering-pl endure-inf diamond become.opt.act.3sg-1sg heart
‘May my heart become a diamond to endure (my) own sufferings!’
(M500.1b5; trans. by CEToM; verse; [7¦7¦4]×4)
In this example, the first-person singular PC -ñ represents the inalienable possessor of arañce
‘(my) heart.’ As this PC represents the subject’s possessor, the intransitive verb klautkoy- ‘May X
become (a diamond)!’ is unaccusative. PKAS4B b1 attests the same passage (9.42).
274
ṣañ läkle-nta warpa-tsi ¦ waśīr klautkoy-ñ arañce ¦

tsmoytär-ñ nete :
increase.opt.mid.3sg-1sg strength
‘May my heart become a diamond to endure (my) own sufferings! May my strength in-
crease! [4a] ’
(PKAS4Bb1; trans. by CEToM; verse; [7¦7¦4]×4)
In (9.43), kektseñe ‘body’ is the subject of klautkañ- ‘X will become (heavy).’ This verb contains the
second-person singular PC -cä, representing the inalienable possessor of kektseñe ‘body.’
(wa)lo weṣṣäṃ brāhmaṇiśka mäkcepi ṅke kektseñe

king speak.npst.act.3sg little.brahmin.voc refl.gen ptcl body.f.nom.sg
krarma[b5](rtsa) klautkañ-cä – – – – · l · ntsi :
heavy.f.nom.sg become.subj.act.3sg-2sg
krᵤi (twe re)ki-meṃ mā pä – – – – – – – –

if nom.2sg word-abl neg
‘(The ki)ng speaks: “Little Brahmin! Your own body will become heavy … if you don’t …
from the word.”’
(B78b5; trans. based on CEToM14 ; verse?)
9.1.11 TA √trik(ā)- ‘be confused; faint’
We may also add TA √trik(ā)- ‘be confused; faint’ to the list of unaccusative verbs based on exam-
ple (3.47), repeated here as (9.44) This example contains the third-person singular pronominal
clitic -äṃ. It represents the inalienable possessor of mācar ‘mother,’ which is the subject of the
intransitive verb trekaṣ- ‘X will be confused.’
14. CEToM seems to consider the PC represents an IO (experiencer; “(The ki)ng speaks: ‘Little Brahmin! Your own
body will become heavy for you .... If from the (wo)rd (you) don’t ... ’”).
275
(9.44) (= 3.47) [TA] Function: possessor of an intransitive subject
wiyoss oki cam klop-yo ¦ mācar nuṃ nuṃ

be.frightened.ptcp.f.nom.sg like dem.m.acc.sg suffering-ins mother again again
trekaṣ-äṃ :
be.confused.subj.act.3sg-3sg
‘His mother will be confused for a long time (lit. again and again), as if (she were) upset
by this pain. [3b] ’
(A152a2; verse; [7¦7]×4)
9.1.12 TB √nes-|tāk(ā)- ‘be, become’
The ΤB root √nas-|tāk(ā)- ‘be, become’ forms an unaccusative verb. The following example (9.45)
contains tākoy- ‘May X be (firm and respectful)!’ with the first-person PC -ñ. This PC represents
the inalienable possessor of arañce ‘heart,’ which is the verb’s subject. Therefore, we take TB
tākoy- as unaccusative.
(rekauna-ṣṣeṃ ścirenäṃ ¦ kwri ra) [b1] yepeṃ

word-adjz.acc.pl harsh.m.acc.pl if ptcl sword.acc.pl
swāṣye-ñ tsa ¦ ainaki ra :
rain.caus.opt.act.3pl-1sg emp mean.nom.pl ptcl
kälṣamñe-ṣṣe niṣke su ¦ prākre tākoy-ñ arañce ¦ poś

patience-adjz ornament dem.m.nom.sg firm cop.opt.act.3sg-1sg heart all-all
pautarṣke :
respectful
‘Even if the mean ones should make the swords of harsh words rain on me, [27c] may my
heart, this ornament of patience, remain firm [and] respectful to everybody! [27d] ’
(PKAS5Bb1; verse; [7¦7¦4]×4)
9.1.13 TA √nas-|tāk(ā)- ‘be, become’
The ΤΑ root √nas-|tāk(ā)- ‘be, become’ also forms an unaccusative verb. Example (9.46) contains
the first-person singular PC -ñi. It represents the inalienable possessor of se ‘son,’ which serves
276
as the subject of the intransitive verb tākiṣ- ‘May X become (the arhat)!’.
pättāñäkte mācar tākim || tem-i-k noṣpeṃ [a6] ///

Buddha.lord.gen mother cop.opt.act.1sg dem.f-gen-emp ?
(lakṣa)ṇäs-yo yetu ārānt se tākiṣ-ñi •
mark.pl-ins decorate.ptcp.m.nom.sg arhat son cop.opt.act.3sg-1sg
‘May I become the mother of the Buddha-lord! … to/of her … May my son become the
arhat decorated with … marks!’
(A118a6; prose?)
9.1.14 TA √pränk? - ‘restrain oneself’
The following example (9.47) shows that the TA root √pränk? - ‘restrain oneself ’ forms an unac-
cusative verb.15 In this example, the first-person singular PC -ñi, referring to sundari ‘Sundari,’
serves as the inalienable possessor of käryāñ ‘choice, will,’ which is the subject of the intransitive
verb präṅki- ‘X restrain oneself.’
sundari träṅkäṣ /// [a4] /// oki ñi poñcäṃ truṅk-aṃ roñcäm klop
Sundari say.npst.act.3sg like gen.1sg all.acc.sg hole-loc jealousy suffering
: käryāñ präṅki-ñi {t}patär-ñy16 oki ni ///
will.nom.pl restrain.npst.act.3pl-1sg ?.npst.mid.3sg-1sg like
‘Sundari says: “… like … for me … jealousy (and) suffering … in the entire hollow. My
thoughts restrain myself, like my ... ”’
15. There seems to be only two attestations of the non-causative stems of this root in TA: A115 a4 (9.47) and A64
b1 : ote täpreṃ märkampal sälpāṣl(u)ne o(t)e (tä)preṃ t(ñ)i bodhiṣi ytār pākär yāmlune • ote täpreṃ [b1] (ñäktaśśi empelune
kus-ne p)rä(ṅ)k(i)ñc “Oh, what glowing of the Law, oh, what revelation of your path to enlightenment, oh, what
(cruelty of the gods who) restrain themselves!” (A64b1; trans. based on CEToM; prose)
16. For the reading {t}patär-ñy, see Schulze, Sieg, and Siegling (1931: 446) and Malzahn (2010: 652f.).
277
9.1.15 TA √prutk(ā)- ‘be shut; be filled’
I consider that TA √prutk(ā)- ‘be shut; be filled’ forms an unaccusative verb based on the following
example (7.83), repeated here as (9.48). In this example, the subject of protkar- ‘X were filled’ is
puk marmañ ‘all of the marman (veins?).’ This verb accompanies the third-person singular PC -äṃ,
which doubles the subject’s possessor āṣānikyāp bodhisatvāp ‘of the venerable Bodhisattva.’
(9.48) (= 7.83) [TA] Function: Doubling of the possessor of an intransitive subject
ku yal (täm ya)kṣe śwā-(ts)i eṣṣ-äm

why dem Yakṣa.gen eat-inf give.npst.act.3sg-pl
tñi-k anaprä ṣakk ats lyā āpsā [b4] – – m •

gen.2sg-emp before certainly ptcl parts.nom.pl
cam klop-yo āṣāniky-āp bodhisatv-āp puk (ma)rmañ

dem suffering-ins venerable-gen Bodhisattva-gen all vein? .nom.pl
protkar-äṃ – – tkan-ā klā •
be.filled.pst.act.3pl-3sg earth-perl fall.pst.act.3sg
‘Why does (he) give us to the yakṣa as a food? In front of you, surely, the parts […]. All veins
of the venerable Bodhisattva were filled with this pain. […] (he) fell on the ground.
(A356b4; prose?)
9.1.16 TB √plätk- ‘overflow, develop, arise’
The TB root √plätk- ‘overflow, develop, arise’ also forms an unaccusative verb. The following ex-
ample (9.49) contains pletkar-c ‘X overflowed,’ whose subject is ysāra ‘(your) blood.’ The second-
person singular PC -c hosted by pletkar- refers to the Buddha and represents the inalienable pos-
sessor of the subject. PKAS4B a4 (9.50) attests the same passage.
278
(säswa āñm plyäñca-lñe-sa ¦ klokastäṃ-meṃ o)k tmane ¦ pletkar-c

lord.voc self.acc sell-nmlz-perl pore.pl-abl 8 10,000 overflow.pst.act.3pl-2sg
ysār-a 40-2
blood-pl
‘Oh lord! By selling yourself, your blood overflowed from 80,000 pores. [42d] ’
(M500.1b2; trans. based on CEToM; verse; [7¦7¦4]×4)
säswa ā(ñm plyäñca-)lñ(e)-sa ¦ klokastäṃn-meṃ ok tmane ¦ pletkar-c

lord.voc self.acc sell-nmlz-perl pore.pl-abl 8 10,000 overflow.pst.act.3pl-2sg
ysār-a 2
blood-pl
‘Oh lord! By selling yourself, your blood overflowed from 80,000 pores. [2d] ’
(PKAS4Ba4; trans. based on CEToM;17 verse; [7¦7¦4]×4)
9.1.17 TB √plu- ‘float, fly, soar’
We may also add the TB root √plu- ‘float, fly, soar’ to the list of the representative roots that form
unaccusative verbs. In the following example, the subject of pluṣṣi-ñ ‘X leaped, floated’ is palskw
ārañce ‘mind (and) heart,’ and the first-person singular PC -ñ represents the inalienable possessor
of the subject.
• lkoym-c kru i yne-mane ¦ ypauna kwṣain-ne ci ¦

see.opt.act.1sg-2sg whenever go-ptcp land.pl village.pl-loc acc.2sg
plu[a2]ṣṣi-ñ sak-sa palskw ārañce ¦
float.impf.act.3sg-1sg happiness-perl mind heart
[Māyā, the Buddha’s mother, speaking:] “Every time I saw you (= the Buddha) going
through lands and villages, my mind (and) heart leaped for joy […]”
(B246a1-2; verse; [5¦5¦8¦7]×4)
17. CEToM: ‘O lord, by (sell)ing y(ourself) blood overflowed out from eighty thousand pores.’
279
9.1.18 TB √mäsk- ‘be’
The following examples show that the root √mäsk- ‘be’ forms an unaccusative verb in TB. In (9.52),
the third-person singular PC -n(e) represents the inalienable possessor of cämpamñe ‘power,’ which
is the subject of the intransitive verb mä(s)k(e)tär ‘X remains (superior)’ (lit. ‘X is more’).
mant se pals(k)o mā yairu ¦ tākaṃ kwri śtwer

thus dem mind neg practice.ptcp.m.nom.sg cop.subj.act.3sg if 4
warä[b1](ṣ-lyñe-sa :
practice-nmlz-perl
mä)kc(e)w ra tsa indrī-sa ¦ eṅkal-ṣe śänman-ne-ś swese 80-5

rel.acc ptcl ptcl sense-perl passion-adjz come.subj.act.3sg-3sg-all rain
su ceu palsko päst kauṣäṃ ¦ cämpamñe

dem.m.nom.sg dem.m.acc.sg mind away destroy.npst.act.3sg power
mä[b2](s)k(e)tär-n= oṃṣap :
be.npst.mid.3sg-3sg more
‘In this way, if this spiritj has not been exercised by the four exercises, [85c] to whichj the
rain of passion also comes through (the hole of the) sense(s), [85d] this (rain)k will utterly
destroy the spirit. Itsk power is superior. [86a] ’
(PKAS6Cb1; verse; [7¦8]×4)
Although the context is limited, the third-person singular PC -ne in Example (9.53) seems to show
the inalienable possessor of käṃtwo ‘tongue,’ which is the subject of mäsketär- ‘X is (white).’
280
(9.53) [TB] possessor of an intransitive subject (?)
/// [b6] ānts-ne cpi lkānträ 10 pilko

shoulder-du dem.m.gen.sg appear.npst.mid.3pl view
mäntäṃtär-ne tucya-ne e(śa-ne) ///
be.destroyed.npst.mid.3sg-3sg yellow-du eye-du
///[b7] rämer satāṣṣäṃ • ārkwi mäsketär-ne käṃtwo

quickly exhale.npst.act.3sg white.m.nom.sg be.npst.mid.3sg-3sg tongue
epe wat no ///
conj conj conj
‘(His) two shoulders will appear to this one as …. [10] His view is destroyed. (His) two
yellow eyes … (He) exhales quickly. But his tongue is white or …’
(B118b7; verse)
9.1.19 TA √mäsk- ‘be’
In example (9.54), kapśañi ‘body’ is the subject of the intransitive verb mäskatr- ‘X is (like the
golden mountain Jāmbūnada).’ The third-person singular PC -aṃ represents this subject’s in-
alienable possessor.
• (säm nape)n-äṣ lyutār penu knānmune yāt-l-une-yo

dem.m.nom.sg human-abl more also wisdom be.capable-gdv-nmlz-ins
kaknu mäskatär
come.about.ptcp.m.nom.sg be.npst.mid.3sg
lyalypä-ntw-āśśi ta(m)pe-wāts-une-yā nu mā ok täm

deed-pl-gen power-adjz-nmlz-perl conj neg yet dem
ksa-l-une-yaṃ ytsi [a7] cämpäṣ •
come.to.extinction-gdv-nmlz-loc go-inf be.able.npst.act.3sg
ja(mb)unāt wsā-ṣi ṣull oki kapśañi mäskatr-aṃ •

Jāmbūnada gold-adjz mountain like body be.npst.mid.3sg-3sg
‘(He) is more than (a [normal] human being) endowed with knowledge and power. But
because of the powerfulness of (his) deeds, (he) is not able to go to the Nirvāṇa yet. His
body is like the golden mountain Jāmbūnada.’
(A295a7; trans. based on CEToM; prose?)
281
9.1.20 TA √lotkā- ‘turn, become’
As the following example suggests, TA √lotkā- ‘turn, become’ forms an unaccusative verb. In
(9.55), the intransitive verb lotka- ‘X becomes (clean)’ accompanies the third-person singular PC
-ṃ that represents the (inalienable) possessor of the subject kapśaññi ‘body.’
(o)[a3]mäskenäṃ lyalypūr-äṣ ¦ pärsäk prākär tampewāts : 60(-8)

evil.m.acc.sg deed-abl be.afraid.pst.act.3sg strong powerful
– m-äk praṣt-äṃ18 sne wraske ¦ āṣtraṃ lo{t}ka-ṃ kapśaññi (:) ///

(?)
-emp time-loc without sickness clear turn.pst.act.3sg-3sg body
‘The powerful (person) was afraid of an evil deed very much. [68d] At that(?) time, his body
became clean (and) without sickness. [69a] ’
(A221a3; verse; [7¦7]×4)
9.1.21 TB √wāk(ā)- ‘differ’
The TB root √wāk(ā)- attests an (intransitive) present IV woko-mid ‘X become divided, X blossom’
and a (transitive) preterite IV wākäṣṣa-act ‘split X, make X blossom.’ In addition, this root also
forms a present VIII intransitive wākṣ-mid ‘X differ (from Yabl ),’ as attested in (9.56). The intran-
sitive verb wākṣtär- ‘X differs’ in this example is unaccusative as the second-person singular PC
represents the possessor of yakne ‘manner,’ which is the subject of this verb.
18. praṣt-äṃ seems to be a misspelling for praṣt-aṃ. Also, āṣtraṃ [f.nom/acc.pl] is likely to be a grammatical error
for āṣtri [f.nom.sg].
282
po-sa auṣap po-s= olypo ¦ po-meṃ wākṣtär-ś seṃ

all-perl moreover all-perl more all-abl differ.pst.mid.3sg-2sg dem.m.nom.sg
yakne •
manner
te-sa ykū ñī saim wästa ¦ cī-ne aurtse [b5]

dem-perl go.ptcp.m.nom.sg gen.1sg protection refuge 2sg-loc broad.m.nom.sg
larauññe • 6
love
‘More than all, over all, and from all, this manner of yours distinguishes itself. [6c] That
is why deep [lit. broad] love to you, o help and stay, has come to me. [6d] ’
9.1.22 TB √ṣäm-|läm(ā)- ‘sit’
Example (5.82), repeated here as (9.57), contains the first-person singular PC -ñ. It represents the
inalienable possessor of prosko ‘fear,’ which is the subject of the intransitive verb lāma- ‘X may
rest.’ Therefore, we consider lāma- to be an unaccusative verb.
(9.57) (= 5.82) [TB] Function: possessor of an intransitive subject
tune taukau-c saim pācer ¦ lāma-ñ prosko :

therein hide.subj.act.1sg-2sg protection father rest.subj.act.3sg-1sg fear.nom.sg
‘Therein I will hide in your protection, father, (so that) my fear may rest.’
(IOLToch5b2; verse [7¦7¦4]×4)
9.1.23 TA √sätk(ā)- ‘spread out’
Example (9.58) shows that TA √sätk(ā)- ‘spread out’ forms an unaccusative verb. This example
contains the third-person singular PC -ṃ, that represents the possessor of tuṅk ‘love.’ This nom-
inal expression is the subject of the intransitive verb sätkā- ‘X spread.’
283
(9.58) [TA] Function: inalienable possessor of an intransitive subject
māk cmol-w-ā /// [b6] /// m(au)dgalyāyann-aṃ cmaul-ṣi tuṅk

many birth-pl-perl Maudgalyāyana-loc birth.adjz love
sätkā-ṃ : – ///
spread.pst.act.3sg-3sg
‘Through many births ... ... his genuine love for Maudgalyāyana spread.’
(A50b6; trans. by CEToM; verse)
9.1.24 TB √si-n- ‘satiate oneself, be depressed’
Unlike the TA root √si-n-, which shows causative-inchoative alternation (i.e., sinäṣ-act ‘to satiate
X [tr.]’ vs. sinäṣ-mid ‘to satiate oneself [itr.]’), TB √si-n- does not take any active inflection to form
a transitive verb. Instead, it uses a causative stem Xb sīnäṣṣä-/sīnäske-act to represent a transitive
verb meaning ‘to satiate X.’ The following example shows that sīntsate- in TB is unaccusative. In
example (9.59), pälsko ‘mind’ is the subject of the intransitive verb sīntsate- ‘X was satisfied,’ and
the second-person singular PC -c represents the subject’s (inalienable) possessor.
śaiṣse-ntse kärtse-ṣc ¦ mā pälsko nta sīntsate-c (:)

world-gen.sg good-all neg mind emp be.satiated.pst.mid.3sg
ket no cämpämñe ¦ seṃ takoy alyek-e[a2]pi (:)

rel.gen.sg conj power dem.m.nom.sg cop.opt.npst.act.3sg another-gen
‘[…] For the good of the world, your spirit was never satisfied. [2b] Who else could have
that ability? [2c] ’ (B224a1; verse; [5¦7] × 4)
9.1.25 TB √spālkā ? - ‘±strive actively/forcefully for’
In the following example (9.60), the first-person singular PC -ñ represents the possessor of mar-
manma ‘veins,’ which is the subject of the intransitive verb spalkkaskentär. This verb is built on
the root TB √spālkā? - ‘±strive actively/forcefully for’ (Malzahn 2010: 965f.).
284
spalkkaskentär-ñ marma-nma katkauña-ṣṣe warkṣält-sa

strive.actively.for.npst.mid.3sg-1sg vein-pl joy-adjz power-perl
‘My veins forcefully strive for (it) by the strength of joy’
(PKNS 19a4; trans. by CEToM)
In view of this example, one might wonder, however, whether spalkkaskentär should be classified
as an unaccusative verb since the gloss ‘±strive actively/forcefully for’ (Malzahn 2010: 965) or
‘make an effort’ (Peyrot 2013b: 837) gives an impression that a subject has control over an action,
which is not typical for unaccusative verbs (see, e.g., Sorace 2000, 2004).
In fact, there is a debate over the the semantics of TB √spālkā? - (and TA √spāltkā? -; see Malzahn
2010: 965f. for a summary). Scholars have traditionally connected the root with a noun spelke
‘zeal’ (cf. Schulze, Sieg, and Siegling 1931: 480 ‘sich anstrengen’; Krause 1952: 302 ‘sich beeifern’;
TEB II: 259 ‘sich beeifern, sich bemühen’; Schmidt 1974: 28 ‘sich beeifern’; and Pinault 2008:
326 ‘faire effort, s’efforcer, s’appliquer à’). In contrast, Couvreur (1954: 84f.) glosses it as ‘sich
wälzen.’ Winter (1984: 120) also thinks that this verb involves a motion and thus translates it as
‘flap around, crawl.’ Malzahn (2010: 965f.) describes that according to her correspondence with
Adams, the meaning ‘act/move with force’ may explain all of the attested cases. When there is
no goal associated with it, the verb means ‘thrash about;’ when it takes an infinitive as a comple-
ment, it means ‘strive actively/forcefully.’
However, as Table (9.2) shows, the root √spālkā? - may select an animate entity and a body-part as
a subject. In the latter case, the subject is not a volitional entity capable of striving.
In (9.61), spalkāte-ne is followed by an infinitival phrase rīmeṃ lantsi ‘to go out of the city.’ In this
example, one may take the PC as representing either the possessor of arañce ‘heart,’ which is the
subject of spalkāte, or the subject of the infinitive lantsi ‘to go out,’ with clitic climbing. Again, the
subject arañce ‘heart’ is not a volitional entity that has the ability to control the process described
by spalkāte.
285
LG Attestation Subject Verb
1 TB THT100 b4 ? spalkaṣyeṃ(ntär) impf.mid.3pl
2 TB THT1573 a4 wnolmi ‘human beings’ spalkaṣyenträ impf.mid.3pl
3 TB PKAS13F a5-6 ? spalkaskema[a6](ne) ptcp
4 TB PKNS19 a4 marmanma ‘veins’ spalkkaskentär-ñ npst.mid.3pl-1sg
5 TB PKNS398 a1 arañce ‘heart’ spalkāte-ne pst.mid.3sg-3sg
6 TB IOLToch 55 b2 ara(ñce) ‘heart’ (spa)lkāte-ne pst.mid.3sg-3sg
7 TB IOLToch 5 b5 walo māga(tṣe) ‘King of Magadha’ spalkāte pst.mid.3sg
8 TA A116 b5 tāloṣ ‘miserable ones’ spāltäṅkānträ npst.mid.3sg
9 TA A237 a3 ? spāltäṅkāmāṃ ptcp
Table 9.2: Attestations of TB √spālkā? - and TA √spāltkā? -
(9.61) [TB] Function: possessor of an intransitive subject or subject of an infinitive
… (krenta) yāmor-nta ñitkāre-ne

good deed-pl push.away.pst.mid.3pl-3sg
spalkāte-ne ram no arañce rī-meṃ lan-tsi

?
√spālkā -.pst.mid.3sg-3sg as conj heart city-abl go.out-inf
‘... the (good?) deeds held him off, (but) his heart urges (him) to go out of the city, as it
were.’
(PKNS398a1; trans. based on CEToM)
The same passage is found in (9.62). Again it is not immediately clear whether -ne is the subject
of an infinitive or the possessor of ara(ñce).
(9.62) [TB] Function: possessor of an intransitive subject (?)
… (spa)lkāte-ne ram no ara(ñce) …

?
√spālkā -.pst.mid.3sg-3sg as conj heart
‘his heart urges, […] as it were.’ (IOLToch55b2; trans. by CEToM)
Therefore, it seems reasonable to consider that the verb √spālkā? - describes an uncontrolled in-
voluntary bodily function/process (such as English ‘sweat,’ ‘shiver,’ and ‘tremble’), rather than a
controlled process; such as ‘strive’ or ‘make an effort.’ We propose to gloss √spālkā? - as ‘tremble,
be agitated’ and translate (9.60) as ‘my veins tremble because of the power of joy.’
In Sorace’s (2000; 2004) Auxiliary Selection Hierarchy (Table 5.2 in §5.4), verbs denoting uncon-
286
trolled processes include involuntary bodily functions (e.g., ‘shiver,’ ‘tremble’) and emission of
substance, light, sound, or smell (e.g., ‘ring,’ ‘shine’). TB √spālkā? - (and probably TA √spālkā? -
also) belongs to this class, which surfaces as unaccusatives.
9.1.26 TB √tsäm(ā)- ‘grow, increase, come into being’
The following examples show that the TB root √tsäm(ā)- ‘grow, increase, come into being’ forms
an unaccusative verb. In (9.63), the first-person singular PC -ñ represents the inalienable posses-
sor of nete ‘strength,’ which is the subject of tsmoytär- ‘May X increase!’.
(9.63) [TB] Function: inalienable possessor of an intransitive subject
ṣañ läkle-nta warpa-tsi ¦ waśīr klautkoy-ñ arañce ¦

tsmoytär-ñ nete :
increase.opt.mid.3sg-1sg strength
‘May my heart become a diamond to endure (my) own sufferings! May my strength in-
crease!’
(PKAS4Bb1; trans. by CEToM; verse; [7¦7¦4]×4)
Likewise, PKAS4B b2 (9.64) attests tsmoytär- ‘May X increase!’ which takes nete pälskoṣṣe ‘strength
of the mind’ as its subject. The first-person singular PC -ñ hosted by tsmoytär- represents the
subject’s inalienable possessor.
287
(piś cmel-a)-ṣṣ(e-ṃ)ts pakāna ¦ apiś-ne ka kwri cmīmar ¦

five birth-pl-adjz-gen.pl for.the.sake.of Avici-loc ptcl if be.born.opt.mid.1sg
kalpa-nma-sa :
practice-pl-perl
tsmoytär-ñ nete pälsko-ṣṣe ¦

increase.opt.mid.3sg-1sg power mind.adjz
‘If, for the sake of (the beings) of the five births, I were reborn in the Avici-hell for the
kalpas, [4c] may my strength of the mind increase!
(PKAS4Bb2; trans. based on CEToM;19 verse; [7¦7¦4]×4)
(9.65) [TB] Function: Possessor associated with an intransitive subject (?)
/// (yśā)[a4]m(na) ¦ ś(a)teṃ o(st)-ne (tä)nmaske(n)trä ¦ (e)kñiññe-sa

among.men rich.acc house-loc be.born.npst.mid.3pl possession-perl
kekenoṣ :
be.fulfilled.ptcp.m.nom.pl
tākaṃ orocci ¦ ktsaitsäññe śman-me ¦

cop.subj.act.3pl great.m.nom.pl old.age come.subj.act.3sg-pl
tsmeṃtär-ne ka waipecce-nta ¦ ///
increase.npst.mid.3pl-3sg emp possession-pl
‘They are reborn (among) humans in a rich house, (and) provided with possession. [6a]
[When] they become adults [and] old age comes to them, their possessions (pl) will surely
increase. [6b] …’
(PKAS7Ea4; verse; [5¦5¦8¦7]×4)
Regarding example (9.65), CEToM considers that the third-person singular -ne refers to ktsait-
säññe ‘old age’ and represents the indirect object (benefactive) of the intransitive verb tsmeṃtär-
‘(possessions will) increase for X.’20 However, since a PC representing bene- or malefactive usu-
ally refers to a [+animate] referent, I am inclined to follow Sieg (1938: 21 n. 2) in considering -ne
to be an error for -me, referring to the subject of (tä)nmaske(n)trä and tākaṃ. This PC represents
19. CEToM: ‘If for the sake of the beings of the five births I were even be reborn in the Avīci [hell], may my spiritual
strength increase! (May I remain?) without turning back from the omniscience! 4.’
288
the possessor of waipeccenta ‘possessions,’ which is the subject of tsmeṃtär- ‘X will increase.’
9.1.27 TB √tsälp(ā)- ‘pass away, be released, be redeemed’
The following example shows that tsälpā- ‘X was released (from Yabl )’ is unaccusative. This in-
transitive verb accompanies the first-person singular PC -ñ that represents the possessor of palsko
‘mind, spirit.’ This nominal expression is the subject of the intransitive verb.
(9.66) [TB] Function: inalienable possessor of an intransitive subject
(śtwāra e)[b2]mpreṃ-nma ¦ auspa lyakāwa ¦ tsälpā-ñ palsko

four truth-pl truly see.pst.act.1sg be.relased.pst.act.3sg-1sg spirit
kleśa-nma-meṃ ¦ yonwa ik(e kekeṣos 10-3)
affliction-pl-abl obtain.pst.act.1sg place become.extinguished.ptcp.m.acc.sg
‘I have truly seen the four truths. My mind has been released from afflictions. I have
reached the place of extinction. [13d] ’
(PKAS6Eb2; trans. based on CEToM; verse; [5¦5¦8¦7]×4)
9.2 Limitation 1: Verbs of appearance/disappearance
The previous section listed representative roots in TA and TB that form unaccusative verbs. I
consider them to be unaccusatives based on the function of pronominal clitics: they may take a
PC that represents a possessor of a subject. However, my analysis has two limitations. The first
limitation is that when a PC appears next to a verb of appearance or disappearance, its function
is often ambiguous: it may represent either a possessor of the verb’s subject or an indirect object
(source, location, benefactive, or experiencer) of the verb. Table 9.3 lists representative verbs of
appearance and disappearance in TA and TB.
20. CEToM: “(Among) humans they are reborn in a rich house provided with possession. [6a] [When] they become
adults [and] old age comes to them, the richness (pl.) will surely increase for it [= old age]. [6b]”
20. CEToM: ‘I have truly seen the (four) truths; my mind is freed from passions, I have reached the place (of extinc-
tion).’
289
1 TA/TB √näk-mid ‘fall into ruin, disappear’ (§9.2.1)
2 TA/TB √wik(ā)- ‘disappear’ (§9.2.2)
3 TA/TB √spärk(ā)- ‘disappear, perish’ (§9.2.3)
4 TB √naut(ā)- ‘disappear’ (§9.2.4)
5 TB √musk(ā)- ‘disappear, perish’ (§9.2.5)
6 TA √pyutk-act ‘come into being’ (§9.2.6)
7 TA √kātkā- ‘(a)rise’ (§9.2.7)
8 TA √pärkā- ‘(a)rise; become clear’ (§9.2.8)
9 TA/TB √läk(ā)-mid ‘appear; be seen’ (§9.2.9)
10 TA/TB √täm- ‘be born, come into being’ (§9.2.10)
11 TB √tsäṅkā- ‘(a)rise’ (§9.2.11)
12 TA/TB √länt- ‘go out, emerge’ (§9.2.12)
Table 9.3: Verbs of appearance/disappearance
9.2.1 TA/TB √näk-mid ‘fall into ruin, disappear’
The function of a pronominal clitic is ambiguous when it appears next to a verb of disappearance.
For example, there are examples in which TA and TB √näk-mid ‘fall into ruin, disappear’ hosts a
PC.21 In (9.67), the third-person singular PC -ṃ, referring to rāvane ‘Rāvaṇa,’ appears to repre-
sent the inalienable possessor of kärparäṃ ‘dignity,’ which is the subject of nakt- (i.e., ‘his dignity
disappeared’). However, it is also possible to take the PC as representing the IO (source) of the
verb (i.e., ‘dignity disappeared from him’).22
21. E.g., TA: A11 a5 (9.67); A332 b1; TB: B108 a10 (9.68)
22. It might even be possible to take the PC to represent both; whether a PC may have more than one thematic
role is an open question.
290
(9.67) [TA] Function: possessor of an intransitive subject or IO (source)?
wartsi krop-lyāṃ praṣṭ-ā ¦ (wa)[a4]rtsi kot rāvane ¦

accompany gather-gdv.f.acc.sg time-perl accompany split.pst.act.3sg Rāvaṇa
sne knānmuney-ā :
without knowledge-perl
nati eṣ-lyāṃ praṣṭ-ā ¦ nati kot rākṣts-āśśi ¦

power give-gdv.f.acc.sg time-perl power split.pst.act.3sg Rakṣas-gen.pl
vibhi(ṣa)[a5]ne-ṃ – ṣt :
Vibhīṣana-acc.sg
kälymey-ā eñlune ¦ keṃpar eṃtsāt pracr-i ¦

direction-perl teaching wrongly understand.pst.mid.3sg brother-gen
nakt-äṃ kärparäṃ :
disappear.pst.mid.3sg-3sg dignity
‘At the time that company had to be gathered, Rāvaṇa split the company out of ignorance.
[1a] At the time that power had to be given, he split the power of the Rākṣasas and struck
Vibhīṣana. [1b] He misinterpreted the proper advice from his brother and { the dignity
disappeared from him | his dignity disappeared }. [1c] ’
(A11a5; trans. based on CEToM; verse; [6¦6¦5]×4)
Likewise, ambiguity remains as to whether the plural clitic -me in (9.68) represents the posses-
sor of an intransitive subject (“das Jaṭila-Gewand”; Thomas 1983: 259) or the IO (source) of the
intransitive neksate ‘(the jaṭila-robe) disappeared from X.’
(9.68) [TB] Function: possessor of an intransitive subject or IO (source)?
te tot we(weñormeṃ) /// neksate-me • kaṣār-wässa-nma ka

dem.n so.much say.abs disappear.pst.mid.3sg-pl K-robe-pl ptcl
ṣ kektseñt-sa lyakānte-me.
conj body-perl appear.pst.mid.3pl-pl
‘Having sa(id) this much […] { (the jaṭila-robe) disappeared from them | their (jaṭila-robe)
disappeared }, and kaṣār-robes appeared on their bodie[s].’
(B108a10; prose)
291
9.2.2 TA/TB √wik(ā)- ‘disappear’
We may take a PC hosted by TA/TB √wik(ā)- ‘disappear’ as representing either the possessor of
the verb’s subject or the source of the verb.23 In the following example (9.69) from TA, the func-
tion of the first-person singular PC -ñi is either the inalienable possessor of yoke ‘thirst,’ which is
the subject of wekaṣ- (i.e., ‘my thirst will disappear’) or the IO (source) of wekaṣ- (i.e., ‘thirst will
disappear from me’).
(9.69) [TA] Function: possessor of an intransitive subject or IO (source)?
[b7] /// lā ñi tsokam yoke wekaṣ-ñi tñi

gen.1sg drink.subj.act.1sg thirst disappear.subj.act.3sg-1sg gen.2sg
tampey-aṃ ///
power-loc
‘… (If) I drink my …, { thirst will disappear from me | my thirst will disappear }. In your
power …’
(A431b7; prose?)
Example (9.70) from TB is also ambiguous. The first-person singular PC -ñ in this example repre-
sents either the possessor of a missing subject or the IO (source) of wikoytär‘may … disappear!’.
kos te postäṃ saṃsār-ne ¦ kliñi-ñ walka spārta-tsi

how dem.n after Saṃsāra-loc be.necessary.opt.act3sg-1sg for.a.long.time turn-inf
:
tañ pern· - - - [b2] ñi ¦ yekte-perne wikoytär-ñ 10

gen.2sg glory gen.1sg? of.little.worth disappear.opt.mid.3sg-1sg
“How would it be necessary for me to be in the Saṃsāra for a long time after this? [10c]
Your glory … May { […], being low-rank, disappear from me | my […], being low-rank,
disappear }! [10d] ”
(B249.ab2; verse; [7¦7]×4)
23. E.g., TA: A46 b4; A340 a4; A431 b7 (9.69) TB: B249.a b2 (9.70)
292
9.2.3 TA/TB √spärk(ā)- ‘disappear, perish’
The TA and TB root √spärk(ā)- ‘disappear, perish’ also shows this type of ambiguity.24 In the
following example, ere ‘color, appearance’ is the subject of the intransitive verb sparkā- ‘X dis-
appeared, perished.’ The third-person singular PC -ne hosted by sparkā- represents either the
(inalienable) possessor of the subject or the IO (source) of the verb.
/// (subhāṣitagave)ṣ(i) walo olyapotse läklessu ere päst

S king.nom very sorrowful color away
sparkā-ne ||
disappear.pst.act.3sg-3sg
‘... King (Subhāṣitagave)ṣin [became] very sorrowful, { the color disappeared from him
| his color disappeared }.’
9.2.4 TB √naut(ā)- ‘disappear’
We find several examples where a PC appears next to TB √naut(ā)- ‘disappear.’25 Example (9.72)
contains the first-person singular -ñ. It seems to represent the inalienable possessor of the noun
yāmor ‘deed,’ which is the subject of the intransitive verb nauyto (sic) ‘May X disappear!’ Never-
theless, it is also possible to take the PC as representing the IO (source) of the verb (i.e., ‘May X
disappear from me!’).
24. E.g., TA: A11 a5, A222 a4; A239 a2 TB: B99 a1 (9.71)
25. E.g., IOL Toch 5 b3-4 (9.72), b4 (9.73), B22 a4 (9.75), B271 a3 (9.76), and PKAS7G a1 (9.74).
293
ciṣṣe saim-äś kloyomar nauyto-ñ yāmor

your protection-all fall.npst.mid.1sg disappear.opt.npst.act.3sg-1sg deed
kāntoytär-ñ, kṣ[ā]nt[i] tākoy-ñ
rub.off.opt.npst.mid.3sg-1sg forgiveness cop.opt.act.3sg-1sg
‘I fall onto your protection. May { my deed disappear | the deed disappear from me }!
May my (deed) be wiped out! May there be forgiveness for me!’
(IOL Toch 5 b3-4; trans. based on CEToM)
Examples (9.73) and (9.74) attest the same combination of a PC, √naut(ā)-, and yāmor/yāmornta
‘deed/deeds.’ In these examples, yāmor/yāmornta ‘deed/deeds’ is the subject of nautā- ‘X disap-
peared’ in (9.73), and nautan- ‘X will disappear’ in (9.74), respectively. The third-person singular
PC -ne describes either the possessor of the subject or the IO (source) of the verb.
ci-ne yāmu śrigupti yolo yāmor nautā-ne ciṣṣe-k

you-loc do.ptcp.m.nom.sg protection evil deed disappear.pst.act.3sg-3sg your-emp
saimt-sa
protection-perl
‘He who has taken (excellent) protection in you, { his evil deed disappeared | the evil
deed disappeared from him } through your protection.’
(IOL Toch 5 b4; trans. based on CEToM)
294
kᵤse no su wnolme ¦ ket śaul nanautau ¦

rel.nom conj dem.m.nom.sg human.being rel.gen life disappear.ptcp.m.nom.sg
yāmor-nta no ykāk nesaṃn-ne ¦ māw-k nautan-ne
deed-pl conj still cop.npst.act.3pl-3sg neg-emp disappear.subj.act.3pl-3sg
po-ykne-sa :
all-way-perl
‘But whoever person he (is), whosesoever life (has) disappeared, his deeds still exist. { His
(deeds) will not surely disappear | (They) will not surely disappear from him } in any
way. [17a] ’
(PKAS7Ga1; verse; [5¦5¦8¦7]×4)
The following two examples (9.75) and (9.76) have a different subject than yāmor (perne ‘dignity’
in [9.75] and empelñe ‘horror’ in [9.76]). Still, the PC used in these examples is ambiguous between
the possessor of the subject and the IO (source).
295
ṣñār ekñe-nta-sa ¦ soy-tsi lāñco mā campe(ṃ)

respective possession-pl-perl be.satisfied-inf king.nom.pl neg be.able.npst.act.3pl
(:)
[a5] (co)wai tärkan(aṃ ¦ ypauna) ku (ṣ)aino alyeṅkä-ts •

prev rob.npst.act.3pl land.pl village.pl other-gen.pl
nautaṃ-me perne ¦ tumeṃ yukseṃ ceu 26

disappear.subj.act.3sg-pl dignity then overcome.npst.act.3pl dem.acc.sg
aly(ai)k •
other.nom.pl
taiknes= erkatte ¦ lāñc mäskeṃtr ontsoytñe-sa 60[a5](-6)

thus hostile king.nom.pl be.npst.mid.3pl insatiability-perl
‘Kings cannot be satisfied with (each of) their own possessions [66a] , (and) they rob others’
(lands) [and] villages. [66b] [But] if { their dignity disappears | dignity disappears from
them }, then the others will defeat them (lit. him). [66c] In this way, kings are hostile
because of [their] insatiability. [66d] ’
tūsa tāppoṃ sai[a3]m-wästi ¦ mai no nauta-ñ

?
therefore appear .opt.act.3pl protector.pl ptcl conj disappear.subj.act.3sg-1sg
empelñe ¦ arañcä-ntse :
horror heart-gen
“Therefore, may protectors appear? , so that the horror of (my) heart may disappear
from me. [39c] ”
(B271a3; trans. based on CEToM; verse; [7¦7¦4]×4)
26. Is ceu ‘this one’ [m.acc.sg] in this example an error for ceṃ ‘them’ [m.acc.pl] (referring to the kings)?
296
9.2.5 TB √musk(ā)- ‘disappear, perish’
The following example (9.77) contains an intransitive verb musk(entä)r- ‘X disappear,’ accompa-
nying the plural PC -me. Nevertheless, it is unclear whether the plural PC -me in this example
represents the possessor of the subject of musk(entä)r- ‘X disappear,’ which is in the lacuna, or
the IO (source) of the intransitive verb (i.e., ‘… disappeared from X’).
cai krᵤi nta yśāmna ¦ cmenträ onolmi ¦

dem.m.nom.pl if indf among.men be.born.subj.mid.3pl living.being.pl
snaice ost-ne tänmaskenträ ¦ ekñiññe-sa meṅkīce :
poor.acc.sg house-loc be.born.npst.mid.3pl possession-perl lacking.acc
tāk· /// [b3] ṣpä ¦ musk(entä)r-(m)e po-ykne-sa (:)

conj disappear.npst.mid.3pl-pl all-manner-perl
When these beings are reborn somehow among humans, they are reborn in a poor house
(which) lacks possession. [8b] ... and { their … disappear | ... disappear from them } in
every way. [8c] ’
(PKAS7Eb3; verse; [5¦5¦8¦7]×4)
9.2.6 TA √pyutk-act ‘come into being’
Likewise, TA √pyutk- ‘[act] come into being; [mid] establish, create, accomplish’27 resists our
classification. We may interpret a PC used in example (9.78) to represent either the possessor of
a subject or the IO (experiencer or location) of the verb.
27. E.g., A2 b6, A372 a2, and YQ III.6 b5 (9.78).
297
(9.78) [TA] Function: possessor of an intransitive subject or IO (location)?
[b4] /// (ka)nak eṃtsāträ ptāñkät käṣyāp el eṣ

cotton.cloth take.subj.mid.3sg Buddha.lord teacher.gen gift give.npst.act.3sg
täm ṣurmaṣ ṣakkats klyomäntā(p [b5] meträkyāp) /// • taryāk we-pi
dem because.of certainly noble.m.gen.sg Maitreya.gen 30 2-ptcl
lakṣañi salu pyutkāseñc-äṃ •
mark.nom.pl entirely emerge.subj.act.3pl-3sg
‘... (if) he takes this cotton cloth, he will give it as a present to the Buddha-god the teacher.
For this reason, the ... of the noble (Metrak) certainly ... the thirty-two marks will appear
on him in their entirety.’
(YQ III.6b5; trans. based on Ji, Winter, and Pinault 1998: 161 followed by CEToM; prose?)
9.2.7 TA √kātkā- ‘(a)rise’
We find ambiguous examples in which TA √kātkā- ‘(a)rise’ hosts a PC.28 In the following example
(9.79), ākāl ‘wish’ is the subject of kātka- ‘X arose.’ This verb hosts a third-person singular PC that
refers to Bṛhadyuti ‘Bṛhadyuti.’ This PC represents either the possessor of the subject (i.e., ‘X’s
wish’) or the IO (experiencer or location) of the verb (i.e., ‘… arose in X’).
(9.79) [TA] Function: possessor of an intransitive subject or IO (experiencer/location)?
ākāl kātka-ṃ ¦ puttiśparn-ac ¦ ṣokyo wärṣṣälts : 1 ||

wish arise.pst.act.3sg-3sg Buddha’s.dignity-all very strong.nom.sg
‘A very strong wish for the rank of Buddha arose in him. [1d] ’
(A24b2; trans. based on CEToM; verse [4¦4¦4]×4)
9.2.8 TA √pärkā- ‘(a)rise; become clear’
TA √pärkā- ‘(a)rise; become clear’ also shows this type of ambiguity.29 The second-person singu-
lar PC -ci of example (9.80) describes either the possessor of prāptiñ ‘prāpti; the power of obtaining
everything (one of the eight superhuman faculties),’ which is the subject of parkar- ‘X appeared’
28. E.g., A24 b2 (9.79), A307 b6, A313 a1, A366 a2, and A394 a3
29. E.g., A372 a4 (9.80), and A412 a2 (9.81)
298
or the IO (experiencer or location) of parkar- ‘(The prāpti-s) appeared on/to X.’
[a4] /// (kā)swone-ntw-āśśi parkar-ci prāptiñ :

virtue-pl-gen arise.pst.act.3pl-2sg possession.nom.pl
tākaṣt pättāññākät puk knānmāṃ āṣānik ca ///

cop.act.2sg Buddha.lord all wisdom venerable
‘[…] of (all) virtues appeared to you (as) the prāpti (“possession”). You have become the
Buddha, all-knowing Arhat […]’
(A372a4; verse)
The same passage appears in A412 a2 (9.81).
[a2] /// pūk kāswone-ntw-āśśi parkar-ci prāptiñ ///

all virtue-pl-gen arise.pst.act.3pl-2sg possession.nom.pl
‘[…] of all virtues appeared to you (as) the prāpti (“possession”). […]’
(A412a2; verse)
9.2.9 TA/TB √läk(ā)-mid ‘appear; be seen’
When √läk(ā)- takes an active ending and hosts a PC, the PC’s function is often unambiguous. For
example, the second-person singular PC in (5.31), repeated here as (9.82), represents the posses-
sor of ersna ‘(beautiful) forms,’ the DO (theme) of the verb.
(9.82) (= 5.31) [TB] Function: inalienable possessor of a DO (theme)

‘(They) see your (beautiful) forms. (They) praise you. […]’
(B213b2; verse; [4¦4¦4]×4)
When it takes a middle ending, however, the function of a PC is often ambiguous.30 In the follow-
30. E.g., TA: A397 a7 (9.86), and b1 (9.87). TB: B5 b5 (9.83), b6 (9.84), B76 a1 (3.20), b5 (3.20), B207 b4 (9.90), B213 b2
(5.31), and B242 a4 (9.85),
299
ing two examples (9.83) and (9.84), Ānanda, who massages the Buddha, speaks to him. Both ex-
amples contain a verb of appearance (lkāntär- ‘X appear’), which accompanies the second-person
singular PC -c, referring to the Buddha. This PC seems to represent the possessor of the subject
(i.e., ‘yoursg X appear’), but it is also possible to interpret the PC as representing the IO (location)
of the verb (i.e., ‘X appear on yousg ’).
(9.83) [TB] Verb of appearance + PC = IO (location) or possessor of an intransitive subject?
weña poyśiṃ-ś ānande ¦ lkāntär-c ñakta

speak.pst.act.3sg Omniscient-all Ānanda appear.npst.mid.3pl-2sg lord.voc
(indri)[b6]-nta-ṃts •
sensory.organ-pl-gen
allek te-sa nesalyñe ¦ eśne warñai piśa-ntso 73

other dem-perl existence eye.du beginning.with five.pl-gen
‘Ānanda spoke to the Omniscient: “O God, being-another-than-this (i.e., the change) of
the five sensory organs, beginning with eyes, is seen on you.” [73d] ’
(B5b5; trans. based on CEToM; verse; [7¦7]×4)31
(9.84) [TB] Verb of appearance + PC = IO (location) or possessor of an intransitive subject?
kauta-läñe yetse-ntse ¦ misā-ṃts lkāntär-c īlār-ñe :

break-nmlz skin-gen flesh.pl-gen appear.npst.mid.3pl-2sg weak-nmlz
[Ānanda speaking to the Buddha:] “The cracking of the skin and the frailness of the flesh
are seen on you. [74a] ”
(B5b6; trans. by CEToM; verse; [7¦7]×4)32
The following example (9.85) is also ambiguous: The second person singular PC -c represents
either the IO (location/experiencer) of lkāntär- or the possessor of yärponta ‘(good) deeds.’
31. The subject of lkāntär ‘appear’ is allek tesa nesalyñe ‘the being other than this,’ which is singular. However, lkāntär
shows plural agreement, influenced by indrintaṃts ‘sensory organs’ [gen.pl] (Thomas 1983: 150).
32. In this example, location seems more likely since the possessor seems embedded under kautaläñe (i.e., ‘cracking
of your skin’).
300
(9.85) [TB] Verb of appearance + PC = IO (location/experiencer) or possessor of an intransi-
tive subject?
/// [a4] lkāntär-c yāmwa nauṣ yärpo-nta •

appear.npst.mid.3pl-2sg do.ptcp.f.pl former good.deed-pl
‘[…] the (good) deeds, accomplished earlier, will appear to you […]’
(B242a4; verse; [4¦4¦4]×4? )
Examples (9.86) and (9.87) from TA attest a third-person singular PC that represents either the
IO (location/experiencer) of lkātr- (i.e., ‘[something] appears on/to X’) or the possessor of wles
‘work, service, effort’ (i.e., ‘His X appears’).
(9.86) [TA] Verb of appearance + PC = IO (location/experiencer) or possessor of an intransitive

subject?
uddeśake träṅkäṣ tm-ann ats wles wätkāltsi lkātr-äṃ •

Uddeśake speak.npst.act.3sg dem-loc ptcl work.f certain see.npst.mid.3sg-3sg
kᵤyalte maltowinu dhyāṃ pälskālune-yo pa(päly)ku ///

because first dhyāna thinking-ins burn.ptcp.m.nom.sg
‘Uddeśaka says: “Thus, the task certainly appears to him. Because the first dhyāna has
been burnt by the thinking …”’
(A397a7; prose?)
(9.87) [TA] Verb of appearance + PC = IO (location/experiencer) or possessor of an intransitive

subject?
täm ne-k pälske – /// [b1] /// naṃ nu ṣokyo tsopats cämp-l-une
dem comp-emp think conj very great be.able-gdv-nmlz
lkātr-äṃ
see.npst.mid.3sg-3sg
‘Thereupon, … and great ability appears on him.’
(A397b1; prose?)
Still, we can find examples in which a PC unambiguously represents an IO. The second-person
singular PC -c of example (3.20), repeated here as (9.88), unambiguously represents the IO (ex-
301
periencer) of lkāntar- ‘(the Śuddhāvāsa gods) will appear to X.’ It is not likely to represent the IO
(location) (i.e., ‘the Śuddhāvāsa gods will appear on X’) nor the possessor of the subject (i.e., ‘X’s
Śuddhāvāsa gods will appear’).
(9.88) (= 3.20) [TA] Verb of appearance + PC = IO (experiencer)
/// ¦ (śuddha)vāsä-ṣṣi ¦ ña(k)t(i) lkāntar-c kauñī ram no ¦

Śuddhāvāsa-adjz.nom.pl god.nom.pl see.subj.mid.3pl-2sg sun as conj
ompalskoñe ṣme(-mane :) ///
meditation sit-ptcp
‘The (Śuddhā)vāsa gods will appear to you like a sun, sit(ting) in meditation.’
(B76a1; trans. based on Schmidt 1974: 234; verse; [5¦5¦8¦7]×4)33
Finally, the following two examples contain a nominal expression with a locative marker (e.g.,
kektsen-ne ‘on the body’). It is uncertain whether a PC hosted by lkāntar ‘appear, are seen’ rep-
resents the possessor of the subject (‘X’s … appear’) or the possessor of the IO (‘… appear on X’s
body’).
(9.89) [TB] Function: possessor of an intransitive subject or possessor of an IO (location)?
/// (- - sa)swa ¦ lkāntar-c kektsen-ne ¦ täryāka wī lakṣānä-nta ¦ mai ///

master.voc see.npst.mid.3pl-2sg body-loc 30 2 mark-pl ptcl
‘O (mas)ter, the 32 marks appear on your body …’
(B76b5; trans. based on Schmidt 1974: 234; verse; [5¦5¦8¦7]×4)34
Example (9.90) attests lkantar-c, which one should read as lkāntär-c ‘X appear’ [npst.mid.3pl-2sg].
The passage of this example is restored based on B221 a5 : /// (ysā-yo)[a5]kñana swañcaiyno po
kälymintsa cärkāsta maiytarṣṣana ‘You discharged the (gold)en rays of friendship to all directions.’
33. lkāntar-c for lkāntär-c ‘X will appear’ [subj.mid.3pl-2sg]
34. lkāntar-c is to be read as lkāntär-c ‘X appear’ [npst.mid.3pl-2sg]
302
(9.90) [TB] Function: possessor of an intransitive subject or possessor of an IO (location)?
/// [b4] kektse(ññ)e35 ¦ lkantar-c ñakta :

?
body.nom.sg see.npst.mid.3pl-2sg lord.voc
ysa-yok(ñana swañcaiyno ¦ po kälymint-sa cärkāsta ¦ maiytar-ṣṣana

golden.f.pl ray.f.acc.pl all direction-perl emit.pst.act.2sg friendship-adjz.f.pl
:) ///
‘(…) appear on your body, O Lord! You discharged the golden rays of friendship to all
directions.’
(B207b4; verse; [7¦7¦4]×4)
9.2.10 TA/TB √täm- ‘be born, come into being’
The intransitive root √täm- ‘be born, come into being’ is ambiguous in TA and TB.
Example (9.91) contains the third-person singular PC -ne, which represents either the inalienable
possessor of takarṣkäññe ‘faith’ or the IO (location) of the verb. It is likely to represents the lo-
cation since takarṣkäññe seems indefinite. Likewise, the third-person singular PC -äṃ in example
(9.92) [TA] is ambiguous: it represents either the possessor of puskāñ ‘nerves’ or, more likely, the
IO (location/benefactive) of the verb.
35. An upper part of the <ññe> akṣara is visible (cf. Sieg and Siegling 1953: 124 n. 9: “Nach Ausweis des Originals
wohl zu kektse(ññ)e sic zu erg.”), although we expect kektse(nn)e ‘body’ [loc.sg] rather than kektse(ññ)e [nom.sg].
B207 b4 〈 ke ktse (ññ)e lka nta_ rc ña kta 〉
303
(9.91) [TB] Function: inalienable possessor of an intransitive subject or IO (location)
/// sa tarya naumīye-nta kärso(r-m)e(ṃ ta)karṣkäññe

three.f.pl jewel-pl understand.abs-abl faith
tänmastär-ne •
be.born.npst.mid.3sg-3sg
‘… Having understood the three jewels, faith was born in him.’
(B159a5; prose?)
(9.92) [TA] Function: IO (location/benefactive) (?)
wiki [a3] /// s-aṃ puskāñ tmäṃsaṃtr-äṃ ||

23rd.m.acc.pl day.pl-loc nerve.nom.pl be.born.npst.mid.3pl-3sg
wiki(śtwarpiñcinäs) ṣpät koṃs-aṃ śwāl kapśiññ-ā tämnäṣtr-äṃ ||

24th.m.acc.pl 7 day.pl-loc flesh body-perl be.born.npst.mid.3sg-3sg
wikipäñpiñcinäs [a4] /// tämnäṣtr-äṃ ||

25th.m.acc.pl be.born.npst.mid.3sg-3sg
wikiṣäkpiñci(näs ṣpät koṃ)s-aṃ ya· (kapśi)ññ-ā tämnäṣtr-äṃ ||

26th.m.acc.pl 7 day.pl-loc skin body-perl be.born.npst.mid.3sg-3sg
wikiṣpätpiñcinäs ṣpät koṃs-aṃ [a5] /// ṣtr-ä(ṃ) ||

27th.m.acc.pl 7 day.pl-loc be.born.npst.mid.3sg-3sg
‘In the 23rd week, nerves are born to him. In the 24th week, flesh is born to him on the
body. In the 25th week, (inner-skin) is born to him (on the body). In the 26th week, skin
is born to him on the body. In the 27th week, … is born to him.’
(A151a3; prose?)
9.2.11 TB √tsäṅkā- ‘(a)rise’
We find examples that contain TB √tsäṅkā- ‘(a)rise’ and a PC.36 It is often not easy to determine
whether a PC represents the possessor of an intransitive subject or the IO (location/experiencer).
It seems that a PC unambiguously represents the IO of an intransitive verb when the verb’s subject
is indefinite. In example (9.93), palsko ‘thought’ is the subject of the verb tsāṅkaṃ- ‘(if) X arises.’
36. E.g., B107 a3 (9.97), B169 b3 (9.98), M 500.1 b2 (9.94), PKAS 6E b6 (9.95), b7 (9.96), and PKAS 8C b2 (9.93).
304
It contains the third-person PC, referring to kete ‘(he) to whom.’ It does not seem to represent
the possessor of the subject (palsko) but the IO (location/experiencer) of the intransitive verb.
(9.93) [TB] Function: IO (location/experiencer)
ket(e āñ)m(e) tākaṃ (śwātsi yoktsi [b2] kä)llā-tsi •

rel.gen desire cop.subj.3sg food drink obtain-inf
erkenma-meṃ śerk pra-lle •

cemetery.pl-abl rope bring.npst-gdv
cew śerkwa-meṃ wente yamaṣ-le •

dem.m.acc.sg rope-abl covering make.npst-gdv
cew wente-sa ñuwe kuntiśke taṣa-le •

dem.m.acc.sg covering-perl new little.pot set.npst-gdv
ton āyar-sa śwātsi yoktsi eneṃ taṣa-lle •

dem.f.acc.pl giving-perl food drink inside set.npst-gdv
tumeṃ ṣukt nässait yama-ṣle

thereupon 7 spell make.npst-gdv
ente palsko tsāṅkaṃ-ne ot śwātsi (yo[b3]ktsi) kälpāṣäṃ

when(ever) thought arise.subj.mid.3sg-3sg then food drink obtain.npst.act.3sg
7 ||
‘(If) one has a desire to obtain food (and) drink, (he) should bring a rope from cemeteries.
From this rope, (he) should make a covering. On this covering, (he) should put a new little
pot. (He) should put food (and) drink inside, in the manner of giving them. Then, (he)
should cast a spell 7 (times). He will obtain food (and drink) whenever a thought (of them)
arises to him. 7’
(PKAS8Cb2; trans. based on CEToM; prose)37
M 500.1 b2 (9.94) attests the same combination of palsko ‘thought,’ √tsäṅkā- ‘(a)rise,’ and a PC.
37. CEToM: ‘[If] one has the desire to obtain (food and drink) one should take rope out of cemeteries, from this rope
one should make a covering, with this covering one should cover [lit. set] a new pot. In the manner of giving those
[little pieces] one should put food [and] drink inside. Thereupon one cast [lit. make] a spell seven times. Whenever
the thought arises to anyone, then he obtains food [and] (drink). 7.’
305
(9.94) [TB] Function: IO (location/experiencer)?
rṣāke-ṃts lānte kreñc-epi ¦ tsäṅkā-ñ palsko klāwäs-si ¦ lālyi

sage.pl-gen king.gen good-gen arise.pst.act.3sg-1sg thought announce-inf striving
wrotsai :
great.f.acc.sg
‘A thought arose to me to proclaim the great striving of the good king of Riṣis. [43a] ’
(M500.1b2; verse; [7¦7¦4]×4)
In example (9.95), kᵤse ksa (…) vitark ‘whatever consideration’ is the subject of tsäṅkā- ‘… arose
(to X).’ This nominal expression contains ksa and is indefinite. The first-person singular PC of
this example, referring to the Buddha, describes the indirect object (location/experiencer) of
the intransitive verb, rather than the possessor of the verb’s subject.
– – – (lyka)[b5]śk-ñe ¦ vitarkä-nt-aṃts toṃ ¦ yneś yama-lyñe akwatsä-ññe ¦

small-nmlz consideration-pl-gen dem.f.pl manifest do-nmlz sharp-nmlz
mällarṣkä-ññe pälkor-meṃ :
pressing-nmlz see.abs
/// [b6] ¦ yneś toṃ mäskeṃträ ¦ istak campeṃ

manifest dem.f.pl become.npst.mid.3pl instantly be.able.npst.acc.3pl
putkatsi päst ¦ kᵤse ksa tsäṅkā-ñ vitark ñke 10(-4)
separate-inf away what.nom indf arise.pst.act.3sg-1sg consideration now
‘(The coarseness and) the fineness, having seen the manifestation, the sharpness and the
pressure of these considerations, [14c] … These (considerations) become manifest. They
can be immediately distinguished clearly, whatever consideration has arisen to me now.
[14d] ’
(PKAS6Eb6; verse; [5¦5¦8¦7]×4)38
38. Cf. CEToM: “b6/7 Twice attested tsäṅkā-ñ clearly has a suffixed 1. singular pronoun, and even though the
Buddha is speaking, such a reference is somewhat disturbing.” We find the content of this passage in the Udānavarga
31.33 (Cittavarga).
(9.r) Uv. 31.33 (cittavarga; Bernhard 1965: 419)
sthūlāṃ vitarkān atha vāpi sūkṣmāṃ samudgatāṃ mānasaṃplavārtham /
vitarkayaṃ vai satataṃ vitarkān etāṃ sadā dhāvati bhrāntacittaḥ // 33
‘Coarse or fine thoughts, arisen on account of flowing of the mind, [33b] considering these thoughts con-
stantly, the puzzled one always wanders. [33d] ’
306
PKAS 6E contains another instance of √tsäṅkā- ‘(a)rise’ occurring with a PC. Again, the subject
(kᵤse ksa pals(k)a-ly(ñ)e ‘whatever thought’) contains ksa and is indefinite. This suggests that the
PC represents the IO (location/experiencer).
/// ¦ [b7] ineś mäskentär ¦ mā ṣp kärsnanträ kleśä-mpa

manifest become.npst.mid.3pl neg conj know.npst.mid.3pl affliction-com
ṣe ¦ kᵤse ksa tsäṅ(k)ā-ñ pals(k)a-ly(ñ)e (:) ///
with what.nom indf arise.pst.act.3sg-1sg think-nmlz
‘… (If the considerations) do (not) become manifest, they are not recognized with afflic-
tion anymore, whatever thought has arisen to me. [15a] ’
(PKAS6Eb7; verse; [5¦5¦8¦7]×4)
The PCs used in (9.97) and (9.98) also seem to represent the IO (location/experiencer) rather than
the possessor of an (indefinite) subject.

Context: An ājīvika Upaga saw Nānda and Nandābala preparing rice porridge.
akālk tsäṅkā-ne mäkte pi kca tā oṅkorñai ñiś śwā-tsi

wish arise.pst.act.3sg-3sg how ptcl indf dem.f.acc.sg porridge I eat-inf
källā-lle ṣeym •
obtain.subj-gdv cop.impf.act.1sg
‘A wish arose to him: “How on earth could I manage to eat this rice porridge?”’
(B107a3; prose)
307
Context: N/A
/// (pratītyasa)mutpāt klyauṣṣäṃ takaṣkäṃñe tsāṅka-ne keś tä

pratītyasa-mutpāda hear.subj.act.3sg faith arise.subj.act.3sg account
/// [b4]
‘(…) will hear pratītyasa-mutpāda and faith will arise in him. […]’
(B169b3; prose)39
9.2.12 TA/TB √länt- ‘go out, emerge’
Finally, when there is a PC attested next to the TA/TB root √länt- ‘go out, emerge,’ we may take
the PC to represent the possessor of the subject (i.e., ‘X’s … goes out’) or the IO (source) of the
verb (i.e., ‘… goes out from X’).40
For example, the third-person singular PC -ne in (9.99) represents either the possessor of an in-
transitive subject or the IO (source) of the verb. The same passage appears five times in this
manuscript (9.100, 9.101, 9.102, and 9.103).
(9.99) [TB] Function: possessor of an intransitive subject or IO (source)
[a1] tu-ne swāralñe yamasträ krāke läṃn-ne

dem-loc pleasure make.npst.mid.act.3sg filth go.out.subj.act.3sg-3sg
saṅghā-träṅkä kätäṅkäṃ
s. commit.a.sin.npst.act.3sg
‘(if) he finds pleasure in it (and) filth comes out of him, he commits the saṃghāvaśeṣa
offence.’
39. One might be tempted to construe the PC in this example with takaṣkäṃñe (i.e., ‘faith in him’). However, al-
though PCs may represent an alienable possessor and part-whole relationship, such complement usage is hardly
attested.
40. E.g., TA: A82 b5, A146 b4, A153 b2, A295 b1, and A298 a5; TB: B88 a1 (9.109), B334 a1 [2x] (9.99, 9.104), a5 (9.100),
a6 (9.100), a10 (9.101), b1 (9.106), b4 (9.102), b5 (9.107), b8 (9.103), and b9 (9.108).
308
(9.100) [TB] B334 a5: [a5] tune swāralyñe yamastär krāke län-ne saṅghā-träṅ(k)ä kätä[a6]ṅkäṃ
(9.101) [TB] B334 a10: tune swāralyñe yamas(tär krāke) [a10] läṃn-ne saṅgha-träṅko kätäṅkäṃ
(9.102) [TB] B334 b4: tu[b4]ne swāralyñe yamastär kr(ā)ke läṃn-ne saṅghā-träṅko kätäṅkäṃ
(9.103) [TB] B334 b8: tune swāralyñe yamastär krāke läṃn-ne saṅghā-träṅko [b9] kätäṅkäṃ
The following example also shows that a PC represents the inalienable possessor of an intransitive
subject or an IO (source).
krᵤi mā krāke läṃn-ne [a2] koss tu māka ette keᵤwcä

If neg filth go.out.subj.act.3sg-3sg as.much dem much downwards upwards
maścä-ś tot stulāṃñcana träṅkä-nta kätta[a3]ṅkäṃ po sāṅ-ne
fist-all so.much s. offense-pl commit.a.sin.npst.act.3sg all community-loc
teśitä yamaṣällona
confession make.gdv.f.pl
‘If filth does not come out of him, as much as he rubs it over the fist upward or
downward, so much he commits Sthūlātyatya-sins that should be confessed before the
whole community’
This manuscript attests a similar passage five times (9.104, 9.105, 9.106, 9.107, and 9.108), although
each example has a different apodosis.
(9.105) [TB] B334 a6: mā krāke läṃn-ne
(9.106) [TB] B334 b1: [b1] krᵤi mā krāke läṃn-ne
(9.107) [TB] B334 b5: krᵤi mā krā[b5]ke läṃn-ne
(9.108) [TB] B334 b9: krᵤi mā kkrāke läṃn-ne
Finally, when √länt- and a PC accompany a nominal expression that represents the IO (source),
the PC may represent either the possessor of the IO or the possessor of the subject (9.109).
309
(9.109) [TB] Function: possessor of an intransitive subject or possessor of an IO (source)
k(a)ntwo koyna-meṃ parna lnaṣṣi-(ne) ||

tongue mouth-abl outside go.out.impf.act.3sg-3sg
‘[His] tongue was hanging [lit. going] out of his mouth.’
(B88a1; trans. by CEToM; prose)41
9.3 Limitation 2: Further ambiguous examples
The second limitation to our analysis is that some intransitive verbs allow multiple interpreta-
tions of a PC: Table 9.4 summarizes them. In this subsection, I briefly look at each root with an
example.

1 TB √kärstā- cut off (§9.3.1)
2 TA √klawā- fall (§9.3.2)
3 TA √tkälā- illuminate (§9.3.3)
4 TB √mänk(ā)- be inferior, lack, be deprived of (§9.3.4)
5 TB √re(-sk)- flow (§9.3.5)
6 TB √lik(ā)- wash (§9.3.6)
Table 9.4: Ambiguous intransitives
9.3.1 TB √kärstā- ‘cut off’
The following example (9.110) contains karstaytär- ‘May X be cut off!’, which accompanies a PC.
The first-person singular PC -ñ of this example represents either the inalienable possessor of
yātalñe ‘ability,’ which is the verb’s subject or the IO (source) of the verb.
41. As CEToM points out, although Thomas (1983: 110) read lnaṣṣine, the ne akṣara is actually in the lacuna.
B88 a1 〈 lna ṣṣi (ne) 〉
310
(9.110) [TB] Function: possessor of an intransitive subject or an IO (source)
/// (aran)ñcä-ṣṣe vimā-ne ¦ saṃvarä-ṣṣe yāta-lñe ¦ mā

?
heart-adjz palace -loc restraint-adjz be.capable-nmlz neg
karstaytär-ñ 30-6
cut.off.opt.mid.3sg-1sg
“May the ability of restraint in the heart-palace not be cut off from me! [36d] ”
(B270a3; trans. based on Schmidt 1974: 208;42 verse; [7¦7¦4]×4)
9.3.2 TA √klawā- ‘fall’
In example (9.111), kratswañ śorkmi ṣutkmi ‘clothes, pegs, (and) strings’ is the subject of the intran-
sitive verb klār- ‘X fell.’ This example contains -äṃ [3sg], which seems to represent the possessor
of the subject. However, one could also interpret this PC as representing the IO (source) of klār-
(i.e., ‘[clothes, pegs, and strings] fell off from X’).
(9.111) [TA] Function: possessor of an intransitive subject or an IO (source)?
pkänt pkänt kratswañ śorkmi ṣutkmi klār-äṃ, (śo[a6]miṃ) mā śkaṃ

apart apart clothes pegs strings fall.pst.act.3pl-3sg girl neg conj
tāk
cop.pst.3sg
‘Her clothes, pegs, (and) strings fell apart, and there was no longer (a girl).’
42. Schmidt (1974: 208): ‘Die Fähigkeit der Zurückhaltung im Herzenspalast möge mir nicht abgeschnitten wer-
den.’ vimā-ne is for vimān-ne ‘in the palace’ and karstaytär-ñ is for karstoytär-ñ ‘may X be cut off!’
311
9.3.3 TA √tkälā? - ‘illuminate’
(9.112) [TA] Function: IO (location/experiencer)?
[b2] /// ne-yā āsäñc-ac porr oki eṅkäl santān-aṃ täklā-ṃ

-perl buttocks-all fire like passion continuity-loc illuminate.pst.act.3sg-3sg
kapśañi sākäs-si mā śkaṃ campäs ṣūraṃ lcā-ṃ ///
body restrain-inf neg conj be.able.pst.act.3sg semen come.out.pst.act.3sg-3sg
‘... passion for buttocks shone in him like a fire in [his] (birth) sequence. He could not
hold back his body; semen came out of it.’
(A153b2; trans. based on Schmidt 1997: 23543 ; prose?)
Malzahn (2010: 664) lists this verbal root as transitive, but as Peyrot (2013b: 758 n. 304) points
out, there are only three forms attested for this root: (1) preterite participle täklo,44 (2) verbal
noun tkālluneyo (see Pinault 2013: 223–4),45 and (3) preterite I täklā-ṃ attested here. If täklā-
is transitive and if eṅkäl ‘passion’ is the subject of the verb, the third-person PC -ṃ in example
(9.112) represents the direct object (theme) of the transitive täklā- (i.e., ‘[passion] illuminated
him’). However, if we consider täklā- to be an intransitive verb (i.e., ‘X shone, X illuminated one-
self ’; cf. Present VIII transitive nt-participle tkäl(ṣa)ntā|(s) ‘illuminating X’ in A273a3–4), this PC
is ambiguous. It represents either the inalienable possessor of eṅkäl ‘desire,’ which is the subject
of täklā- (i.e., ‘his desire for buttocks shone’) or the IO (location/experiencer) of the verb (i.e.,
‘desire for buttocks shone in him’).
9.3.4 TB √mänk(ā)- ‘be inferior, lack, be deprived of’
In example (9.113), ekñinta ‘possessions’ is the subject of the intransitive verb mäṅkāntär- ‘(if) X
are missing’ that hosts the plural PC -me. This PC seems to serves as the possessor of ekñinta or
the IO (source/experiencer) of the verb.
43. Schmidt (1997: 235): “… sein leidenschaftliches Verlangen nach Hinterbacken leuchtete wie ein Feuer in
[seiner] Geburtenfolge. Und er vermochte seinen Penis nicht zurückhalten; Sperma trat aus ihm heraus.”
44. A308 b6: /// [b6] yāytunt täklo ///
45. A397 b2: pälskālune tkālluneyo papälykunt pat nu maltowinuṃnt dhyāṃ ‘Or the first meditation burnt by thought
and enlightenment’
312
(9.113) [TB] Function: possessor of an intransitive subject or an IO (source/experiencer)
cai no akn(ātsa-ñ) ¦ – – po – – (mā) aiś-e(ñca)-ñ :
dem.m.nom.pl conj fool-nom.pl all neg know-ptcp-nom.pl
kektseñt-s= ekñi-nta ¦ mā (mä)ṅ(k)ān(tä)r-me ce-k warñai (:)

body-perl possession-pl neg lack.subj.mid.3pl-pl dem.acc.sg-emp including
‘And these fools ... (not) recognizing … all … If possessions on the body are not lacking to
them in any way, […]’
(B24b3; verse; [5¦8]×4 + [8¦8¦5])
9.3.5 TB √re(-sk)- ‘flow’
In example (9.114), ysāra ‘(drops of) blood’ is the subject of the intransitive verb reske«ṃ»- ‘X flow.’
This verb accompanies the first-person singular PC -ñ, which represents either the possessor of
the subject or the IO (source) of the verb.
/// wartto yne-mane ¦ reske«ṃ»-ñ ysār-a :

forest go-ptcp flow.npst.act.3pl-1sg blood-pl
‘… while walking into the forest, (drops of) blood are running from me …’
(B90a6; trans. based on CEToM; verse; [7¦7¦4] × 4? )
9.3.6 TB √lik(ā)- ‘wash’
The following example (9.115), containing laikontär- ‘may X be washed,’ is also ambiguous. The
first-person singular PC -ñ in this example represents either the inalienable possessor of the
verb’s subject käntwaṣṣana yāmornta ‘deeds related to speeches’ or the possessor of the verb’s
external argument ce yāmorsa ‘by this deed.’
313
(9.115) [TB] Function: possessor of an intransitive subject or possessor of an external argu-
ment
takarṣkñe-sa nätkausa ¦ ku ce palamai-c päla-lyu :

belief-perl hold.off.ptcp.f.nom.pl rel praise.pst.mid.1sg-2sg praise-gdv.m.voc.sg
ce yā[b6]mor-sa laikontär-ñ ¦ käntwa-ṣṣana yāmor-nta :

dem.acc.sg deed-perl wash.opt.mid.3pl-1sg tongue-adjz.f.pl deed-pl
‘As I, pressed by faith, have praised you, O praiseworthy one, [25a] may my deeds of
speeches be washed by this deed! [25b] ’ (B241b6; verse; [7¦7]×4)46
46. Thomas (1957: 172): “Weil ich, auf Gläubigkeit gestützt, dich gepriesen habe, o [du] zu Preisender, möchten
mir durch diese Tat [meine] Zungensünden abgewaschen werden” (followed by Schmidt 1974: 247).
314
Index of passages
A A 77 a4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
A 2 a5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 A 79 a3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
A 2 b6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 A 81 a3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
A 7 a5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311 A 92 a5 . . . . . . . . . . . . . . . . . . . . . . . . . . . 48, 159
A 8 b5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 A 95 a5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
A 8 b6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 A 95 b4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
A 10 b6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 A 99 a3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
A 11 a5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291 A 101 b5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
A 11 b6 . . . . . . . . . . . . . . . . . . . . . . . 74, 182, 218 A 106 a6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
A 13 a2 . . . . . . . . . . . . . . . . . . . . . . . . . . 171, 219 A 108 a3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
A 14 b2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44, 53 A 110 a2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
A 16 a3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202 A 115 a4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
A 20 b3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 A 116 a1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
A 24 b2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 298 A 118 a6 . . . . . . . . . . . . . . . . . . . . . . . . . . 67, 277
A 25 b4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266 A 124 b6 . . . . . . . . . . . . . . . . . . . . . . . . . . . 64, 71
A 50 b6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284 A 125 b2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
A 56 a1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14, 23 A 130 a2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
A 58 b3 . . . . . . . . . . . . . . . . . . . . . . . . . . 122, 233 A 144 b4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
A 59 b2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 A 147 a1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
A 61 a4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19, 24 A 147 b6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
A 63 a3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197 A 150 b6 . . . . . . . . . . . . . . . . . . . . . . . . . 171, 249
A 63 a3-4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200 A 151 a3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 304
A 64 b1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277 A 151 a6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
A 66 b5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267 A 152 a2 . . . . . . . . . . . . . . . . . . . . . . . . . . 62, 276
A 70 b1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 A 152 a3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
A 71 a2 . . . . . . . . . . . . . . . . . . . . . . . . 66, 69, 267 A 153 b2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312
A 72 a2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 A 155 b3 . . . . . . . . . . . . . . . . . . . . . . . . . . . 19, 24
315
A 165 a1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260 A 321 b6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
A 168 b5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 A 332 a1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
A 169 a2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 A 340 a4 . . . . . . . . . . . . . . . . . . . . . . . . . 157, 169
A 179 a1 . . . . . . . . . . . . . . . . . . . . . . . . . . . 25, 28 A 343 a4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
A 184 b3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123 A 356 b3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
A 213 b3 . . . . . . . . . . . . . . . . . . . . . . . . . 170, 245 A 356 b4 . . . . . . . . . . . . . . . . . . . . . . . . . 239, 278
A 215 a2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261 A 358 a1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
A 215 a5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159 A 358 a3 . . . . . . . . . . . . . . . . . . . . . . . . . . . 29, 54
A 215 a7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157 A 372 a4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299
A 218 a5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262 A 395 a2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
A 221 a1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 A 395 b2 . . . . . . . . . . . . . . . . . . . . . . 50, 214, 221
A 221 a3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282 A 395 b4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
A 222 a7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164 A 397 a7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301
A 222 b6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 A 397 b1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301
A 230 b3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263 A 399 b4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

A 400 a2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
A 239 a5–6 . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
A 407 a4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239
A 248 a1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
A 412 a2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299
A 249 b1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
A 431 b6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
A 249 b2 . . . . . . . . . . . . . . . . . . . . . . . . . 195, 208
A 431 b7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292
A 249 b4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
A 435 b3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
A 253 b2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
A 436 b6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
A 253 b6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
A 456.b b3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
A 262 b6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
A 269 + A 290 a3 . . . . . . . . . . . . . . . . . . . . 19, 21 B
A 280 a6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 B 3 a3 . . . . . . . . . . . . . . . . . . . . . . . . . . . 172, 225
A 283a a3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 B 3 a5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273
A 295 a7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281 B 5 b5 . . . . . . . . . . . . . . . . . . . 173, 228, 234, 300
A 295 b1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260 B 5 b6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300
A 309 b2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268 B 12 b4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
316
B 12 b7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206 B 88 a2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
B 14 b3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274 B 88 a6 . . . . . . . . . . . . . . . . . . . . . . . . . . 203, 204
B 15 a7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212 B 88 b5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
B 16 b3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 B 90 a6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313
B 18 a1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155 B 91 a6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
B 21 a2 . . . . . . . . . . . . . . . . . . . . . . . . . . . 43, 107 B 94 a2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
B 21 b5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270 B 99 a1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293
B 22 a4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 296 B 99 b3 . . . . . . . . . . . . . . . . . . . . . . . . . . . 74, 230
B 22 b5 . . . . . . . . . . . . . . . . . . . . . . . 64, 121, 269 B 100 a5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264
B 23 b5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 B 104 a1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
B 24 b3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313 B 107 a3-4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
B 27 b7–8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153 B 107 a7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
B 31 a5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153 B 107 a7-8 . . . . . . . . . . . . . . . . . . . . . . . . . . . 5, 45
B 31 b4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131 B 107 a10 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
B 33 b1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229 B 107 a3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
B 34 a5 . . . . . . . . . . . . . . . . . . . . . . . . . . 170, 243 B 107 a6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236
B 46 a8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 B 107 a9 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
B 46 b4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 B 107 b1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264
B 76 a1 . . . . . . . . . . . . . . . . . . . . . . . . . . . 50, 302 B 107 b4 . . . . . . . . . . . . . . . . . . . . . . . . . . . 18, 47
B 76 b5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302 B 107 b8–9 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
B 78 a1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270 B 108 a2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
B 78 b5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275 B 108 a3-4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
B 79 a1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 B 108 a6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
B 81 b6 . . . . . . . . . . . . . . . . . . . . . . . . . . . 12, 125 B 108 a8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
B 82 b6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 B 108 a9 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
B 83 a5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166 B 108 a10 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
B 85 b5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237 B 108 b4-5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
B 86 b2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 B 108 b10 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
B 88 a1 . . . . . . . . . . . . . . . . . . . . . . 182, 223, 310 B 118 b6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
317
B 118 b7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281 B 334 a1 . . . . . . . . . . . . . . . . . . . . . . . . . 308, 309
B 133 a4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142 B 334 a4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268
B 133 a7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156 B 334 a5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
B 139 a3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238 B 334 a6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
B 159 a5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 304 B 334 a8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268
B 207 b4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303 B 334 a10 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
B 213 b2 . . . . . . . . . . . . . . . . . . . . . 113, 115, 299 B 334 b1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
B 220 b4 . . . . . . . . . . . . . . . . . . . . . . . . . . . 35, 60 B 334 b3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268
B 221 a5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302 B 334 b4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
B 224 a1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284 B 334 b5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
B 231 a4 . . . . . . . . . . . . . . . . . . . . . 170, 242, 243 B 334 b7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268
B 231 b1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271 B 334 b8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
B 231 b4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283 B 334 b9 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
B 241 b1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195 B 337 b2 . . . . . . . . . . . . . . . . . . . . . 155, 165, 167
B 241 b2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 B 375 b4-5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
B 241 b6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 314 B 169 b3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308
B 242 a4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301
B 244 a2 . . . . . . . . . . . . . . . . . . . . . . . . . 172, 231 I
B 246 a1 . . . . . . . . . . . . . . . . . . . . . . . . . . 44, 211 IOLToch 4 b1 . . . . . . . . . . . . . . . . . . . . . . . . . . 49
B 246 a1-2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279 IOLToch 5 b2 . . . . . . . . . . . . . . . . . . . . . 134, 283
B 249.a b2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292 IOLToch 5 b3-4 . . . . . . . . . . . . . . . . . . . . . . . 294
B 255 b5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163 IOLToch 5 b4 . . . . . . . . . . . . . . . . . . . . . . . . . 294
B 270 a3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311 IOLToch 55 b2 . . . . . . . . . . . . . . . . . . . . . . . . 286
B 271 a1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
B 271 a3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 296 M
B 297a a4 . . . . . . . . . . . . . . . . . . . . . . . 12, 42, 98 M 500.1 a1 . . . . . . . . . . . . . . . . . . . . . . . . . . . 266
B 298 a1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163 M 500.1 b1 . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
B 308 b4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248 M 500.1 b2 . . . . . . . . . . . . . . . . . . . . . . . 279, 306
B 330 a2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 M 500.1 b5 . . . . . . . . . . . . . . . . . . . . . . . . . . . 274
318
O PKAS 7G a1 . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
Or 8262.163 b1 . . . . . . . . . . . . . . . . . . . . . . . 122 PKAS 7G a5 . . . . . . . . . . . . . . . . . . . . . . . . . . . 258
Or 8262.163 b3 . . . . . . . . . . . . . . . . . . . . . . . . . 73 PKAS 7N a2 . . . . . . . . . . . . . . . . . . . . . . . . . . 247
Or 8262.163 b5 . . . . . . . . . . . . . . . . . . . . . . . . . 45 PKAS 8B a4 . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
PKAS 8C a5 . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
P
PKAS 8C a6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
PKAS 4A a3 . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
PKAS 8C b2 . . . . . . . . . . . . . . . . . . . . . . . . . . . 305
PKAS 4A b2 . . . . . . . . . . . . . . . . . . . . . . . . . . 264
PKAS 17K a2 . . . . . . . . . . . . . . . . . . . . . . . . . 213
PKAS 4A b4 . . . . . . . . . . . . . . . . . . . . . . . . . . 265
PKAS 17K b5 . . . . . . . . . . . . . . . . . . . . . . . . . . 58
PKAS 4B a4 . . . . . . . . . . . . . . . . . . . . . . 271, 279
PKAS 18A b1–2 . . . . . . . . . . . . . . . . . . . . . . . 169
PKAS 4B b1 . . . . . . . . . . . . . . . . . . . . . . 275, 287
PKAS 18A b2 . . . . . . . . . . . . . . . 18, 47, 111, 165
PKAS 4B b2 . . . . . . . . . . . . . . . . . . . . . . . . . . 288
PKAS 18A b3 . . . . . . . . . . . . . . . . . . . . . 152, 177
PKAS 5A b3 . . . . . . . . . . . . . . . . . . . . . . . . . . 265
PKNS 19 a4 . . . . . . . . . . . . . . . . . . . . . . . . . . . 285
PKAS 5B a6 . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
PKNS 22 b5 . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
PKAS 5B b1 . . . . . . . . . . . . . . . . . . . . . . . . . . 276
PKNS 25+26 b4 . . . . . . . . . . . . . . . . . . . . . . . 265
PKAS 5C a6 . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
PKNS 27 a3 . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
PKAS 5D a5 . . . . . . . . . . . . . . . . . . . . . . . . . . . 266
PKNS 27 a5 . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
PKAS 6A a6 . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
PKNS 34 a4 . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
PKAS 6C b1 . . . . . . . . . . . . . . . . . . . . . . . . . . . 280
PKNS 34 b3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
PKAS 6C b4 . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
PKNS 398 a1 . . . . . . . . . . . . . . . . . . . . . . . . . . 286
PKAS 6C b5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
PKAS 6E b2 . . . . . . . . . . . . . . . . . . . . . . . . . . . 289 T
PKAS 6E b6 . . . . . . . . . . . . . . . . . . . . . . . . . . . 306 THT 1106 a2 . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
PKAS 6E b7 . . . . . . . . . . . . . . . . . . . . . . . . . . . 307 THT 1114 b3 . . . . . . . . . . . . . . . . . . . . . . . . . 154
PKAS 7B b3 . . . . . . . . . . . . . . . . . . . . . . . . . . 244 THT 1308a a2 . . . . . . . . . . . . . . . . . . . . . . . . . 261
PKAS 7B b6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
W
PKAS 7C a5 . . . . . . . . . . . . . . . . . . . . . . . 69, 128
W 26 b3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
PKAS 7E a4 . . . . . . . . . . . . . . . . . . . . . . . . . . . 288
PKAS 7E b3 . . . . . . . . . . . . . . . . . . . . . . . . . . . 297 Y
PKAS 7F a4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 YQ I.6 b1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
319
YQ I.6 b4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158 YQ III.4 a7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
YQ I.6 b6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158 YQ III.4 b7 . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
YQ II.5 a7-8 . . . . . . . . . . . . . . . . . . . . . . . . . . 170 YQ III.6 b5 . . . . . . . . . . . . . . . . . . . . . . . . . . . 298
YQ III.2 b8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
320
Bibliography
Abels, Klaus, and Ad Neeleman. 2012. Linear Asymmetries and the LCA. Syntax 15:25–74.
Adams, Douglas Q. 1988. Tocharian Historical Phonology and Morphology. New Haven, CN: American
Oriental Society.
. 2012. Shedding Light on *leuk- in Tocharian and Hittite and the Wider Implications of
Reconstructing its Indo-European Morphology. Tocharian and Indo-European Studies 13:21–55.
. 2013. A Dictionary of Tocharian B. 2nd ed. Amsterdam: Rodopi.
. 2015. Tocharian B: A Grammar of Syntax and Word-Formation. Innsbruck: Institut für Sprachen
und Literaturen der Universität Innsbruck.
Ahn, Byron. 2016. Severing Internal Arguments from Their Predicates: An English Case Study.
Paper presented at the 90th Annual Meeting of the Linguistic Society of America, Washington
DC, 7–10 January 2016.
Alexiadou, Artemis. 2003. Some Notes on the Structure of Alienable and Inalienable Possessors.
In From NP to DP, edited by Martine Coene and Yves D’Hulst, 2:167–88. Amsterdam: John Ben-
jamins.
. 2014. Roots Don’t Take Complements. Theoretical Linguistics 40:287–97.
Alexiadou, Artemis, and Elena Anagnostopoulou. 2004. Voice Morphology in the Causative-Inchoative
Alternation: Evidence for a Non-Unified Structural Analysis of Unaccusatives. In The Unac-
cusativity Puzzle: Explorations of the Syntax-Lexicon Interface, edited by Artemis Alexiadou, Elena
Anagnostopoulou, and Martin Everaert, 114–36. Oxford: Oxford University Press.
Alexiadou, Artemis, Elena Anagnostopoulou, and Martin Everaert, eds. 2004. The Unaccusativity
Puzzle: Explorations of the Syntax-Lexicon Interface. Oxford: Oxford University Press.
321
Anagnostopoulou, Elena. 2005. Strong and Weak Person Restrictions. A Feature Checking Analy-
sis. In Clitic and Affix Combinations. Theoretical Perspectives, edited by Lorie Heggie and Francisco
Ordóñez, 199–235. Amsterdam: John Benjamins.
. 2006. Clitic Doubling. In The Blackwell Companion to Syntax, edited by Martin Everaert and
Henk van Riemsdijk, 519–81. Oxford / Malden, MA: Blackwell.
. 2017. The Person Case Constraint. In The Wiley Blackwell Companion to Syntax, 2nd ed.,
edited by Martin Everaert and Henk C. van Riemsdijk, 3001–47. Oxford: Blackwell.
Ānandajoti, Bhikkhu. 2020. A Comparative Edition of the Dhammapada. 4th ed.
Anderson, Stephen R. 1969. West Scandinavian Vowel Systems and the Ordering of Phonological
Rules. PhD diss., Massachusetts Institute of Technology.
. 1992. A-Morphous Morphology. Cambridge: Cambridge University Press.
. 2005. Aspects of the Theory of Clitics. Oxford: Oxford University Press.
. 2008. The English “Group Genitive” Is a Special Clitic. English Linguistics 25:1–20.
. 2011. Clitics. In The Blackwell Companion to Phonology, edited by Marc van Oostendorp,
Colin J. Ewen, Elizabeth Hume, and Keren Rice, 2002–18. Oxford: Wiley-Blackwell.
. 2013. The Marker of the English “Group Genitive” Is a Special Clitic, Not an Inflection.
In Morphosyntactic Categories and the Expression of Possession, edited by Kersti Börjars, David
Denison, and Alan Scott, 193–218. Amsterdam: John Benjamins.
Angelopoulos, Nikos, Chris Collins, and Arhonto Terzi. 2020. Greek and English Passives, and the
Role of By-phrases. Glossa: a journal of general linguistics 5:90. 1–29.
Ashton, Neil. 2011. Tocharian Object Clitics and the Derivation of SOV. Paper presented at the 13th
Diachronic Generative Syntax Conference. June 2-5 2011.
Bach, Emmon. 1971. Questions. Linguistic Inquiry 2:153–66.
322
Baker, Mark, and Ruth Kramer. 2018. Doubled Clitics Are Pronouns: Amharic Objects (and Beyond).
Natural Language & Linguistic Theory 36:1035–88.
Batke, Christiane. 1999. Das Präsens und Imperfekt der Verben für ‘Sein’ im Tocharischen. Tochar-
ian and Indo-European Studies 8:1–74.
Beaver, David I., Craige Roberts, Mandy Simons, and Judith Tonhauser. 2017. Questions Under Dis-
cussion: Where Information Structure Meets Projective Content. Annual Review of Linguistics
3:265–84.
Bejár, Susana, and Milan Rezac. 2009. Cyclic Agree. Linguistic Inquiry 40:35–73.
Belletti, Adriana, and Luigi Rizzi. 1981. The Syntax of “Ne”: Some Theoretical Implications. The
Linguistic Review 1:117–54.
Bennett, Ryan, Emily Elfner, and James McCloskey. 2019. Prosody, Focus, and Ellipsis in Irish. Lan-
guage 95:66–106.
Bernard, Chams Benoît, and Ruixuan Chen. 2022. A Fall into the Pit: Remarks on Tocharian B koṣko,
koṣkīye. Indo-Iranian Journal 65:1–31.
Bernhard, Franz. 1965. Udānavarga. Band I, Einleitung, Beschreibung der Handschriften, Textausgabe,
Bibliographie. Göttingen: Vandenhoeck & Ruprecht.
Blevins, James P. 2006. Word-Based Morphology. Journal of Linguistics 42:531–73.
. 2016. Word and Paradigm Morphology. Oxford: Oxford University Press.
Blevins, James P., Farrell Ackerman, and Robert Malouf. 2018. Word-and-Paradigm Morphology.
In Oxford Handbook of Morphological Theory, edited by Jenny Audring and Francesca Masini,
265–84. Oxford: Oxford University Press.
Bonet, M. Eulélia. 1991. Morphology after Syntax. PhD diss., Massachusetts Institute of Technol-
ogy.
323
Borer, Hagit. 2005a. In Name Only. Oxford: Oxford University Press.
. 2005b. The Normal Course of Events. Oxford: Oxford University Press.
Borer, Hagit, and Yosef Grodzinsky. 1986. Syntactic Cliticization and Lexical Cliticization: The Case
of Hebrew Dative Clitics. In The Syntax of Pronominal Clitics, edited by Hagit Borer, 175–217.
Orlando, FL: Academic Press.
Bošković, Željko. 2016. What Is Sent to Spell-out Is Phases, Not Phasal Complements. Linguistica
56:25–66.
Bresnan, Joan, and Sam A. Mchombo. 1987. Topic, Pronoun, and Agreement in Chicheŵa. Language
63:741–82.
Broomhead, J. W. 1962. A Textual Edition of the British Hoernle Stein and Weber Kuchean Manuscripts,
Vol. 1: Transliteration, translation, grammatical commentary and vocabulary. PhD diss., Uni-
versity of Cambridge.
Brunetti, Lisa. 2009. On Links and Tails in Italian. Lingua 119:756–81.
Burlak, Svetlana, and Ilya B. Itkin. 2009. The Tocharian A Forms naṣ=äṃ, naṣ=äm and n=äṃ, n=äm
Revisited. Tocharian and Indo-European Studies 11:43–8.
Burzio, Luigi. 1986. Italian Auxiliaries: A Government Binding Approach. Dordrecht: Reidel.
Cardinaletti, Anna, and Michal Starke. 1999. The Typology of Structural Deficiency. In Clitics in the
Languages of Europe, edited by Henk C. van Riemsdijk, 145–233. Berlin: de Gruyter.
Carling, Gerd. 2000. Die Funktionen der lokalen Kasus im Tocharischen. Berlin: de Gruyter.
. 2005. Proto-Tocharian, Common Tocharian, and Tocharian: on the Value of Linguistic

Connections in a Reconstructed Language. In Proceedings of the Sixteenth Annual UCLA Indo-
European Conference: Los Angeles, November 5-6, 2004, edited by Karlene Jones-Bley, Martin E.
Huld, Angela Della Volpe, and Miriam Robbins Dexter, 47–70. Washington, DC: Institute for
the Study of Man.
324
Carling, Gerd. 2006. Notes on the Syntactic Functions of the Tocharian Clitic Pronouns. In GIŠ.ḪUR gul-
za-at-ta-ra: Festschrift for Folke Josephson, edited by Gerd Carling, 32–45. Göteberg: Meijerbergs
institut för svensk etymologisk forskning.
Carling, Gerd, Georges-Jean Pinault, and Werner Winter. 2009. A Dictionary and Thesaurus of Tochar-
ian A. Vol. 1. Wiesbaden: Harrassowitz.
Carlson, Lauri. 1983. Dialogue Games—An Approach to Discourse Analysis. Dordrecht: Reidel.
Carstens, Vicki. 2002. Antisymmetry and Word Order in Serial Constructions. Language 78:3–50.
Chomsky, Noam. 1970. Remarks on Nominalization. In Readings in English Transformational Gram-

mar, edited by Roderick A. Jacobs and Peter S. Rosenbaum, 184–221. Waltham, MA: Ginn &
Co.
. 1973. Conditions on Transformations. In A Festschrift for Morris Halle, edited by Stephen R.

Anderson and Paul Kiparsky, 232–86. New York: Holt, Rinehart & Winston.
. 1995. The Minimalist Program. Cambridge, MA: MIT Press.
. 2000. Minimalist Inquiries: the Framework. In Step by Step: Essays on Minimalism in Honor
of Howard Lasnik, edited by Roger Martin, David Michaels, and Juan Uriagereka, 89–155. Cam-
bridge, MA: MIT Press.
. 2001a. Beyond Explanatory Adequacy. Cambridge, MA: MIT Working Papers in Linguistics.
. 2001b. Derivation by Phase. In Ken Hale: A Life in Language, edited by Michael Kenstowicz,
1–52. Cambridge, MA: MIT Press.
. 2007. Approaching UG from Below. In Interfaces + Recursion = Language? Chomsky’s Mini-

malism and the View from Syntax-Semantics, edited by Hans-Martin Gärtner and Uli Sauerland,
1–29. Berlin: Mouton de Gruyter.
. 2008. On Phases. In Foundational Issues in Linguistic Theory, edited by Robert Freidin, Carlos
P. Otero, and Maria Luisa Zubizarreta, 133–66. Cambridge, MA: MIT Press.
325
Chomsky, Noam, Ángel J. Gallego, and Dennis Ott. 2019. Generative Grammar and the Faculty of
Language: Insights, Questions, and Challenges. Catalan Journal of Linguistics: 229–61.
Cinque, Guglielmo. 1990. Types of A’-dependencies. Cambridge, MA: MIT Press.
. 2005. Deriving Greenberg’s Universal 20 and Its Exceptions. Linguistic Inquiry 36:315–32.
Ciucivara, Oana. 2009. A Syntactic Analysis of Pronominal Clitic Clusters in Romance: The View
from Romanian. PhD diss., New York University.
Clark, Herbert H. 1975. Bridging. In Theoretical Issues in Natural Language Processing, edited by Roger
C. Schank and Bonnie L. Nash-Webber, 169–74. New York: Association for Computing Machin-
ery.
Collins, Chris. 2005. A Smuggling Approach to the Passive in English. Syntax 8:81–120.
Compton, Richard. 2019. Person Complementarity and (Pseudo) Person Case Constraint Effects:
Evidence from Inuktitut. Canadian Journal of Linguistics/Revue canadienne de linguistique 64:592–
616.
Čop, Bojan. 1974. Das tocharische Personalpronomen Suffixum B -me, A -m und sein Ursprung.
Linguistica 14:27–38.
Cornilescu, Alexandra, and Carmen Dobrovie-Sorin. 2008. Clitic Doubling, Complex Heads and
Interarboreal Operations. In Clitic Doubling in the Balkan Languages, edited by Dalina Kallulli
and Liliane Tasmowski, 289–319. Amsterdam / Philadelphia: John Benjamins.
Couvreur, Walter. 1947. Hoofdzaken van de Tochaarse Klank- en Vormleer. Leuven: Katholieke Univer-
siteit te Leuven.
. 1954. Review of Krause (1952). Göttingische Gelehrte Anzeigen 208:79–92.
Cuervo, María-Cristina. 2014. Arguments for a Root. Theoretical Linguistics 40:375–87.
326
Dalrymple, Mary, and Irina Nikolaeva. 2011. Objects and Information Structure. Cambridge: Cam-
bridge University Press.
De Belder, Marijke, and Jeroen van Craenenbroeck. 2015. How to Merge a Root. Linguistic Inquiry
46:625–55.
De Cat, Cécile. 2007. French Dislocation without Movement. Natural Language & Linguistic Theory
25:485–534.
Deal, Amy Rose. 2017. External Possession and Possessor Raising. In The Wiley Blackwell Companion
to Syntax, 2nd ed., edited by Martin Everaert and Henk C. van Riemsdijk. Wiley-Blackwell.
Debrunner, Albert, and Jacob Wackernagel. 1930. Altindische Grammatik. Vol. 3. Göttingen: Van-
denhoeck & Ruprecht.
Déchaine, Rose-Marie, and Martina Wiltschko. 2002. Decomposing Pronouns. Linguistic Inquiry
33:409–42.
Delbrück, Berthold. 1878. Syntaktische Forschungen. Vol. 3. Halle an der Saale: Waisenhaus.
Despić, Miloje, Michael David Hamilton, and Sarah E. Murray. 2019. A Cyclic and Multiple Agree
Account. Natural Language & Linguistic Theory 37:51–89.
Dixon, Robert M. W. 2010. Basic Linguistic Theory. Vol. 2. Oxford: Oxford University Press.
Dowty, David R. 1991. Thematic Proto-roles and Argument Selection. Language 67:547–619.
Dunkel, George E. 2014. Lexikon der indogermanischen Partikeln und Pronominalstämme. Vol. 2. Hei-
delberg: Winter.
Embick, David. 2010. Localism Versus Globalism in Morphology and Phonology. Cambridge, MA: MIT
Press.
. 2015. The Morpheme. Berlin: de Gruyter.
Embick, David, and Alec P. Marantz. 2008. Architecture and Blocking. Linguistic Inquiry: 1–53.
327
Erteschik-Shir, Nomi. 2007. Information Structure. Oxford: Oxford University Press.
Fanselow, Gisbert. 1988. Aufspaltung von NPn und das Problem der ‘freien’ Wortstellung. Linguis-
tische Berichte 114:91–113.
Fortson, Benjamin W. 2010. Indo-European Language and Culture. 2nd ed. Malden, MA: Blackwell.
Gafter, Roey J. 2014. The Distribution of the Hebrew Possessive Dative Construction: Guided by
Unaccusativity or Prominence? Linguistic Inquiry 45:482–500.
Garrett, Andrew. 1990. The Syntax of Anatolian Pronominal Clitics. PhD diss., Harvard University.
. 1996. Wackernagel’s Law and Unaccusativity in Hittite. In Approaching Second: Second Po-
sition Clitics and Related Phenomena, edited by Aaron Halpern and Arnold M. Zwicky, 85–133.
University of Chicago Press.
Geng, Shimin, Hans-Joachim Klimkeit, Helmut Eimer, and Jens Peter Laut. 1988. Das Zusammen-
treffen mit Maitreya. Die ersten fünf Kapitel der Hami-Version der Maitrisimit. Vol. 1. Wiesbaden:
Harrassowitz.
Givón, Talmy. 1983. Introduction. In Topic Continuity in Discourse: A Quantitative Cross-language Study,
edited by Talmy Givón, 1–42. Amsterdam: John Benjamins.
Godel, Robert. 1975. An Introduction to the Study of Classical Armenian. Wiesbaden: Reichert.
Goldstein, David M. 2016. Classical Greek Syntax: Wackernagel’s Law in Herodotus. Leiden: Brill.
Greenberg, Joseph H. 1963. Some Universals of Grammar with Particular Reference to the Or-
der of Meaningful Elements. In Universals of Language, edited by Joseph H. Greenberg, 73–113.
Cambridge, MA: MIT press.
Grewendorf, Günther. 1989. Ergativity in German. Dordrecht: Foris.
Guimarães, Maximiliano. 2000. In Defense of Vacuous Projections in Bare Phrase Structure. Uni-
versity of Maryland Working Papers in Linguistics 9:90–115.
328
Hackstein, Olav. 1995. Untersuchungen zu den sigmatischen Präsensstammbildungen des Tocharischen.
Göttingen: Vandenhoeck & Ruprecht.
. 1997. Präverb, Post- und Präposition im Tocharischen: Ein Beitrag zur Rekonstruktion
urindogermanischer Syntax. Tocharian and Indo-European Studies 7:35–60.
. 2012. Der Negative Existentialsatz Im Indogermanischen. Historische Sprachforschung 125:123–

42.
. 2013. Polar Questions and Non-headed Conditionals in a Crosslinguistic and Historical

Perspective. In Grammatica et Verba: Glamor and Verve: Studies in South Asian, Historical, and Indo-
European Linguistics in Honor of Hans Henrich Hock on the Occasion of His Seventy-Fifth Birthday,
edited by Shu-Fen Chen and Benjamin Slade, 99–116. Ann Arbor, MI: Beech Stave.
. 2015. The Word-order Patterns Troiae qui primus ab oris and summa cum dignitate in Latin
and Tocharian. In Tocharian Texts in Context: International Conference on Tocharian Manuscripts
and Silk Road Culture, Vienna, June 25-29th, 2013, edited by Melanie Malzahn, Michaël Peyrot,
Hannes Fellner, and Theresa-Susanna Illés, 73–86. Bremen: Hempen.
. 2017. The Phonology of Tocharian. In Handbook of Comparative and Historical Indo-European

Lnguistics, edited by Jared Klein, Brian Joseph, and Matthias Fritz, 2:1304–1335. Berlin / New
York: De Gruyter.
Hackstein, Olav, Hiromi Habata, and Christoph Bross. 2014. Tocharische Texte zur Buddhalegende
I: Die Geschichte von der Nonne Sundarī B15-18. Münchener Studien zur Sprachwissenschaft
68:31–99.
. 2015. Tocharische Texte zur Buddhalegende II: Die Geschichte von Devadatta und Ajā-
taśatru (B21-22). Münchener Studien zur Sprachwissenschaft 69:51–121.
Haegeman, Liliane. 2001. Antisymmetry and Verb-Final Order in West Flemish. Journal of Compar-
ative Germanic Linguistics 3:207–32.
329
Haig, Geoffery. 2008. Alignment Change in Iranian Languages: A Construction Grammar Approach. Berlin:
de Gruyter.
Halle, Morris, and Alec Marantz. 1993. Distributed Morphology and the Pieces of Inflection. In
The View from Building 20, edited by Ken Hale and Samuel Jay Keyser, 111–76. Cambridge, MA:
MIT Press.
. 1994. Some Key Features of Distributed Morphology. MIT Working Papers in Linguistics
21:275–88.
Hamblin, Charles L. 1973. Questions in Montague English. Foundations of Language 10:41–53.
Hamilton, Michael David. 2017. Ditransitives and Possessor Raising in Mi’gmaq. In Papers of the
46th Algonquian Conference, edited by Monica Macaulay and Margaret Noodin, 81–100. East
Lansing, MI: Michigan State University Press.
Harizanov, Boris. 2014. Clitic Doubling at the Syntax-Morphophonology Interface. Natural Lan-
guage & Linguistic Theory 32:1033–88.
Harley, Heidi. 1995. Subjects, Events, and Licensing. PhD diss., MIT.
. 2009. Compounding in Distributed Morphology. In The Oxford Handbook of Compounding,

edited by Rochelle Lieber and Pavol Štekauer, 129–44. Oxford: Oxford University Press.
. 2014. On the Identity of Roots. Theoretical Linguistics 40:225–76.
Harris, Alice C. 1982. Georgian and the Unaccusative Hypothesis. Language 58:290–306.
Hartmann, Jens-Uwe. 1987. Das Varṇārhavarṇa des Mātṛceṭa. Göttingen: Vandenhoeck & Ruprecht.
Hearn, Ryan. 2017. Wh-questions in Hittite and Tocharian and Implications for PIE Reconstruc-
tion. Paper presented at the 36th Annual East Coast Indo-European Conference.
Iatridou, Sabine. 1995. Clitics and Island Effects. University of Pennsylvania Working Papers in Lin-
guistics 2:11–30.
330
Inglese, Guglielmo, Giuseppe Rizzo, and Miriam Pflugmacher. 2019. Definite Referential Null Ob-
jects in Old Hittite: A Corpus Study. Indogermanische Forschungen 124:137–70.
Jasanoff, Jay H. 1989. Language and Gender in the Tarim Basin: The Tocharian 1 sg. Pronoun.
Tocharian and Indo-European Studies 3:125–47.
. 2003. Hittite and the Indo-European verb. Oxford: Oxford University Press.
. 2015. The Tocharian B Accent. In Tocharian Texts in Context: International Conference on

Tocharian Manuscripts and Silk Road Culture, edited by Melanie Malzahn, Michaël Peyrot, Hannes
Fellner, and Theresa-Susanna Illés, 87–98. Bremen: Hempen.
. 2020. Doing Things with Word Equations: A Retrospective on Hittite and the Indo-European
Verb Since HIEV. In Proceedings of the 30th Annual UCLA Indo-European Conference, edited by
David M. Goldstein, Stephanie W. Jamison, and Brent Vine, 93–122. Bremen: Hempen.
Ji, Xianlin, Werner Winter, and Georges-Jean Pinault. 1998. Fragments of the Tocharian A Maitreyasamiti-
Nāṭaka of the Xinjiang Museum, China. Berlin: de Gruyter.
Jügel, Thomas. 2009. Ergative Remnants in Sorani Kurdish? Orientalia Suecana 58:142–58.
Kallulli, Dalina. 2008. Clitic Doubling, Agreement and Information Structure: The Case of Alba-
nian. In Clitic Doubling in the Balkan Languages, edited by Dalina Kallulli and Liliane Tasmowski,
227–56. Amsterdam: John Benjamins.
Karttunen, Lauri. 1977. Syntax and Semantics of Questions. Linguistics and Philosophy 1:3–44.
Katz, Joshua T. 1998. Topics in Indo-European Personal Pronouns. PhD diss., Harvard University.
Kayne, Richard S. 1975. French Syntax. Cambridge, MA: MIT Press.
. 1994. The Antisymmetry of Syntax. Cambridge, MA: MIT Press.
. 2008. Antisymmetry and the Lexicon. Linguistic Variation Yearbook 8:1–32.
331
Kern, Hendrik. 1891. The Jātaka-Mālā. Stories of Buddha’s Former Incarnations. Boston: Harvard Uni-
versity Press.
Keydana, Götz, and Silvia Luraghi. 2012. Definite Referential Null Objects in Vedic Sanskrit and
Ancient Greek. Acta Linguistica Hafniensia 44:116–28.
Khoroche, Peter. 2017. Once a Peacock, Once an Actress: Twenty-Four Lives of the Bodhisattva from Harib-
haṭṭa’s Jātakamālā. Chicago: University of Chicago Press.
Kim, Ronald I. 2009. Tocharian B -ñ ‘(to) me’, -c ‘(to) you’, pas ‘go!’ Tocharian and Indo-European
˙
Studies 11:49–62.
Kiparsky, Paul. 1973. Elsewhere in Phonology. In A Festschrift for Morris Halle, edited by Stephen R.
Anderson and Paul Kiparsky, 93–106. New York: Holt, Rinehart and Winston.
Klamer, Marian. 1997. Spelling Out Clitics in Kambera. Linguistics 35:895–927.
Klein, Jared. 2017. The Phonology of Classical Armenian. In Handbook of Comparative and Historical
Indo-European Linguistics, edited by Jared Klein, Brian Joseph, and Matthias Fritz, 2:1037–79.
Berlin: Walter de Gruyter.
Klingenschmitt, Gert. 1994. Das Tocharische in indogermanistischer Sicht. In Tocharisch. Akten der
Fachtagung der Indogermanischen Gesellschaft, Berlin, September 1990, edited by Bernfried Schlerath,
310–411. Reykjavík: Málvísindastofnun Háskóla Íslands.
Kloekhorst, Alwin. 2008. Etymological Dictionary of the Hittite Inherited Lexicon. Leiden: Brill.
Knoll, Gabriele. 1996. Die Verwendungsweisen der Adjektive im Tocharischen. PhD diss., Univer-
sität Frankfurt am Main.
Kochovska, Slavica. 2011. Clitics and Direct Objects in Macedonian. Acta Linguistica Hungarica 58:241–
60.
Koller, Bernhard. 2013. On the Status of the Particle ne in Tocharian A. Paper presented at the
25th Annual West Coast Indo-European Conference.
332
Koller, Bernhard. 2015. Studies in Tocharian Phonology above the Word-level. PhD diss., Univer-
sity of California, Los Angeles.
Koopman, Hilda. 2010. Prepositions, Postpositions, Circumpositions, and Particles. In Mapping

Spatial PPs, edited by Guglielmo Cinque and Luigi Rizzi, 26–73. Oxford: Oxford University Press.
Koopman, Hilda, and Anna Szabolcsi. 2001. Verbal Complexes. Cambridge, MA: MIT Press.
Krapova, Iliyana, and Guglielmo Cinque. 2005. Two Asymmetries between Clitic Left and Clitic
Right Dislocation in Bulgarian. In Organizing Grammar: Linguistic Studies in Honor of Henk van
Riemsdijk, edited by Hans Broekhuis, Norbert Corver, Riny Huybregts, Ursula Kleinhenz, and
Jan Koster, 359–64. Berlin: Mouton de Gruyter.
. 2008. Clitic Reduplication Constructions in Bulgarian. In Clitic Doubling in the Balkan Lan-
guages, edited by Dalina Kallulli and Liliane Tasmowski, 257–87. Amsterdam: John Benjamins.
Krause, Wolfgang. 1952. Westtocharische Grammatik. Vol. 1. Heidelberg: Winter.
Krause, Wolfgang, and Werner Thomas. 1960. Tocharisches Elementarbuch. Vol. 1. Heidelberg: Win-
ter.
Krifka, Manfred. 2008. Basic Notions of Information Structure. Acta Linguistica Hungarica 55:243–
76.
Kümmel, Martin. 2015. About the Demonstrative System in Tocharian B. In Tocharian Texts in Con-
text. International Conference on Tocharian Manuscripts and Silk Road Culture Held June 26–28, 2013 in
Vienna, edited by Melanie Malzahn, Michaël Peyrot, Hannes A. Fellner, and Theresa-Susanna
Illés, 109–16. Bremen: Hempen.
Kuno, Susumu. 1972. Functional Sentence Perspective: A Case Study from Japanese and English.
Linguistic Inquiry 3:269–320.
. 1973. The Structure of the Japanese Language. Cambridge, MA: MIT Press.
Kuritsyna, Anna. 2016. Preposition Repetition in Tocharian. Indo-European Linguistics 4:190–208.
333
Lambrecht, Knud. 1994. Information Structure and Sentence Form. Cambridge: Cambridge University
Press.
Landau, Idan. 1999. Possessor Raising and the Structure of VP. Lingua 107:1–37.
Lévi, Sylvain. 1933. Fragments de textes koutchéens. Udānavarga, Udānastotra, Udānālaṁkāra et Kar-
mavibhaṅga, publiés et traduits avec un vocabulaire et une introduction sur le «tokharien». Paris:
Imprimerie Nationale.
Levin, Beth, and Malka Rappaport Hovav. 1995. Unaccusativity: At the Syntax-Lexical Semantics Inter-
face. Cambridge, MA: MIT Press.
Linzen, Tal. 2014. Parallels between Cross-linguistic and Language-internal Variation in Hebrew
Possessive Constructions. Linguistics 52:759–92.
Lochbihler, Bethany Christina. 2012. Aspects of Argument Licensing. PhD diss., McGill University.
Lohndal, Terje. 2014. Phrase Structure and Argument Structure: A Case Study of the Syntax-Semantics
Interface. Oxford: Oxford University Press.
Lowe, John J. 2016. English Possessive ’s: Clitic and Affix. Natural Language & Linguistic Theory 34:157–
95.
Luraghi, Silvia. 2020. External Possessor Constructions in Indo-European. In Reconstrcuting Syntax,

edited by Jóhanna Barðdal, Spike Gildea, and Eugenio R. Luján, 162–96. Leiden: Brill.
Lyons, Christopher. 1999. Definiteness. Cambridge: Cambridge University Press.
Mallory, James P., and Douglas Q. Adams, eds. 1997. Encyclopedia of Indo-European Culture. Chicago,
IL: Fitzroy Dearborn.
Malzahn, Melanie. 2010. The Tocharian Verbal System. Leiden: Brill.
. 2012. Position Matters: The Placement of Clitics in Metrical Texts of Tocharian B. Tochar-
ian and Indo-European Studies 113:153–62.
334
Malzahn, Melanie. 2016a. The Development of the Tocharian Causative System: Top-down or
Bottom-up? Indogermanische Forschungen 121:387–400.
. 2016b. The Second One to Branch Off? The Tocharian Lexicon Revisited. In Etymology
and the European Lexicon, edited by Bjarne Simmelkjær Sandgaard Hansen, Benedicte Nielsen
Whitehead, Thomas Olander, and Birgit Anette Olsen, 281–92. Wiesbaden: Reichert.
Marantz, Alec. 1993. Implications of Asymmetries in Double Object Constructions. In Theoretical

aspects of Bantu grammar 1, edited by Sam Mchombo, 113–50. Stanford, CA: CSLI.
. 1997. No Escape from Syntax: Don’t Try Morphological Analysis in the Privacy of Your
Own Lexicon. Edited by Alexis Dimitriadis. University of Pennsylvania Working Papers in Linguis-
tics 4 (2): 201–25.
. 2007. Phases and Words. In Phases in the Theory of Grammar, edited by Sook-Hee Choe, 191–
222. Seoul: Dong In.
Martirosyan, Hrach K. 2010. Etymological Dictionary of the Armenian Inherited Lexicon. Leiden: Brill.
Matić, Dejan. 2015. Information Structure in Linguistics. In International Encyclopedia of the Social
& Behavioral Sciences, 95–9. Elsevier.
Matushansky, Ora. 2006. Head Movement in Linguistic Theory. Linguistic Inquiry 37:69–109.
Mayrhofer, Manfred. 1963. Kurzgefaßtes etymologisches Wörterbuch des Altindischen. Vol. 2. Heidel-
berg: Winter.
. 1996. Etymologisches Wörterbuch des Altindoarischen. Vol. 2. Heidelberg: Winter.
Melchert, H. Craig. 1978. Tocharian Verb Stems in -tk-. Zeitschrift für vergleichende Sprachforschung
91:93–130.
. 1991. The Lydian Emphasizing and Reflexive Particle -ś/-is. Kadmos 30:131–42.
. 1994a. Anatolian Historical Phonology. Amsterdam: Rodopi.
335
Melchert, H. Craig. 1994b. Čop’s Law in Common Anatolian. In In Honorem Holger Pedersen, edited
by Jens Elmegård Rasmussen, 297–306. Wiesbaden: Reichert.
. 2009. Deictic Pronouns in Anatolian. In East and West, edited by Kazuhiko Yoshida and
Brent Vine, 151–61. Bremen: Ute Hempen.
. Forthcoming. The Role of Indo-European Studies in the XXIst Century. In Oxford Hand-
book of Diachronic and Historical Linguistics, edited by Paola Crisma and Giuseppe Longobardi.
Oxford: Oxford University Press. Forthcoming.
Merchant, Jason. 2019. Roots Don’t Select, Categorial Heads Do: Lexical-selection of PPs May Vary
by Category. The Linguistic Review 36:325–41.
Meunier, Fanny. 2015. Recherches sur le génitif en tokharien. PhD diss., École Pratique des Hautes
Études.
Mišeka Tomić, Olga. 2008. Towards Grammaticalization of Clitic Doubling: Clitic Doubling in Mace-
donian and Neighbouring Languages. In Clitic Doubling in the Balkan Languages, edited by Dalina
Kallulli and Liliane Tasmowski, 65–88. Amsterdam / Philadelphia: John Benjamins.
Miyagawa, Shigeru, and Takae Tsujioka. 2004. Argument Structure and Ditransitive Verbs in Japanese.
Journal of East Asian Linguistics 13:1–38.
Moro, Andrea. 2000. Dynamic Antisymmetry. Cambridge, MA: MIT Press.
Neu, Erich. 1991. Etruskisch - eine indogermanische Sprache Altanatoliens? Historische Sprach-
forschung 104:9–28.
. 1997. Zu einigen Pronominalformen des hethitischen. In Studies in Honor of Jaan Puhvel,

edited by Dorothy Disterheft, Martin Huld, and John Greppin, 139–69. Washington, DC: Insti-
tute for the Study of Man.
Nevins, Andrew. 2007. The Representation of Third Person and Its Consequences for Person-Case
Effects. Natural Language & Linguistic Theory 25:273–313.
336
Nevins, Andrew. 2011a. Multiple Agree with Clitics: Person Complementarity vs. Omnivorous
Number. Natural Language & Linguistic Theory 29:939–79.
. 2011b. Phonologically-Conditioned Allomorph Selection. In The Blackwell Companion to

Phonology, edited by Marc van Oostendorp, Colin J. Ewen, Elizabeth Hume, and Keren Rice,
2357–82. Oxford: Wiley-Blackwell.
Nikolaeva, Irina. 2001. Secondary Topic as a Relation in Information Structure. Linguistics 39:1–49.
Ogihara, Hirotoshi. 2009. Researches about Vinaya-texts in Tocharian A and B. PhD diss., École
Pratique des Hautes Études.
Ohnuma, Reiko. 2000. The Story of Rūpāvatī: A Female Past Birth of the Buddha. Journal of the
International Association of Buddhist Studies 23:103–46.
Olsen, Birgit Anette. 2017. The Morphology of Armenian. In Handbook of Comparative and Historical
Indo-European Linguistics, edited by Jared Klein, Brian Joseph, and Matthias Fritz, 2:1080–96.
Berlin: Walter de Gruyter.
Onishi, Teigo. Forthcoming. Clitic Doubling in Tocharian B. In Proceedings of the 32nd Annual UCLA
Indo-European Conference, edited by David M. Goldstein, Stephanie W. Jamison, and Brent Vine.
Bremen: Hempen. Forthcoming.
Pancheva, Roumyana, and Maria Luisa Zubizarreta. 2018. The Person Case Constraint: The Syn-
tactic Encoding of Perspective. Natural Language & Linguistic Theory 36:1291–337.
Pedersen, Holger. 1941. Tocharisch vom Gesichtspunkt der indoeuropäischen Sprachvergleichung. Køben-
havn: Ejnar Munksgaard.
Perlmutter, David. 1978. Impersonal Passives and the Unaccusative Hypothesis. In Proceedings of
the 4th Annual Meeting of the Berkeley Linguistics Society, 4:157–89. Berkeley: Berkeley Linguistics
Society.
337
Perlmutter, David. 1989. Multiattachment and the Unaccusative Hypothesis: the Perfect Auxiliary
in Italian. Probus 1:63–119.
Peyrot, Michaël. 2008. Variation and Change in Tocharian B. Amsterdam: Rodopi.
. 2013a. Die tocharische Daśakarmapathāvadānamālā. In Die Erforschung des Tocharischen

und die alttürkische Maitrisimit, edited by Yukiyo Kasai, Abdurishid Yakup, and Desmond Durkin-
Meisterernst, 161–82. Turnhout: Brepols.
. 2013b. The Tocharian Subjunctive: A Study in Syntax and Verbal Stem Formation. Leiden: Brill.
. 2017. Slavic onъ, Lithuanian anàs and Tocharian A anac, anäṣ. In Usque ad radices: Indo-
European Studies in Honour of Birgit Anette Olsen, edited by Bjarne Simmelkjær Sandgaard Hansen,
Adam Hyllested, Anders Richardt Jørgensen, Guus Kroonen, Jenny Helena Larsson, Benedicte
Nielsen Whitehead, Thomas Olander, and Tobias Mosbæk Søborg, 633–42. Copenhagen: Mu-
seum Tusculanum.
. 2019. The Deviant Typological Profile of the Tocharian Branch of Indo-European May Be
Due to Uralic Substrate Influence. Indo-European Linguistics 7:72–121.
Pinault, Georges-Jean. 1992. Introduction au tokharien. LALIES 11:5–224.
. 1997. Sur l’assemblage des phrases (“Satzgefüge”) en tokharien. In Berthold Delbrück y

la sintaxis indoeuropea hoy. Actas del Coloquio de la Indogermanische Gesellschaft, Madrid, 21-24 de
septiembre de 1994, edited by Emilio Crespo and José Luis García Ramón, 449–500. Wiesbaden:
Reichert.
. 2008. Chrestomathie tokharienne: textes et grammaire. Leuven/Paris: Peeters.
. 2013. Contribution de Maitrisimit à l’interprétation de textes parallèles en tokharien.

In Die Erforschung des Tocharischen und die alttürkische Maitrisimit, edited by Yukiyo Kasai, Ab-
durishid Yakup, and Desmond Durkin-Meisterernst, 17:183–234. Silk Road Studies. Turnhout:
Brepols.
338
Pisani, Vittore. 1941–1942. Glottica Parerga. Milano: Ulrico Hoepli.
Poucha, Pavel. 1955. Thesaurus Linguae Tocharicae Dialecti A. Praha: Státní Pedagogické Nakladatel-
ství.
Pylkkänen, Liina. 2008. Introducing Arguments. Cambridge, MA: MIT Press.
Reinhart, Tanya. 1981. Pragmatics and Linguistics: An Analysis of Sentence Topics. Philosophica
27:53–94.
Richards, Marc D. 2008. Desymmetrization: Parametric Variation at the PF-Interface. Canadian

Journal of Linguistics/Revue canadienne de linguistique 53:275–300.
Ringe, Donald A. 1996. On the Chronology of Sound Changes in Tocharian. Vol. 1. New Haven, CN: Amer-
ican Oriental Society.
Roberts, Craige. 2011. Topics. In Semantics, edited by Klaus von Heusinger, Claudia Maienborn,
and Paul Portner, 2:1908–34. Berlin: de Gruyter.
. 2012. Information Structure in Discourse: Towards an Integrated Formal Theory of Prag-

matics. Semantics & Pragmatics 5:1–69.
Roberts, Ian G. 2010. Agreement and Head Movement. Cambridge, MA: MIT Press.
Rooth, Mats E. 1985. Association with Focus. PhD diss., University of Massachusetts, Amherst.
. 1992. A Theory of Focus Interpretation. Natural Language Semantics 1:75–116.
. 2016. Alternative Semantics. In The Oxford Handbook of Information Structure, edited by

Caroline Fery and Shinichiro Ishihara, 19–40. Oxford: Oxford University Press.
Ross, John R. 1967. Constraints on Variables in Syntax. PhD diss., Massachusetts Institute of Tech-
nology.
Rothstein, Susan. 2017. Secondary Predication. In The Wiley Blackwell Companion to Syntax, 2nd ed.,
edited by Martin Everaert and Henk C. van Riemsdijk, 3872–902. Oxford: Blackwell.
339
Rotman, Andy. 2017. Divine Stories: Divyāvadāna, Part 2. Boston: Wisdom Publications.
Sande, Hannah, and Peter Jenks. 2018. Cophonologies by Phase. In North East Linguistic Society
(NELS) 48, edited by Sherry Hucklebridge and Max Nelson, 2:39–52.
Sande, Hannah, Peter Jenks, and Sharon Inkelas. 2020. Cophonologies by Ph(r)ase. Natural Lan-
guage & Linguistic Theory: 1211–61.
Schäfer, Florian. 2008. The Syntax of (Anti-)Causatives. Amsterdam: John Benjamins.
Schmidt, Klaus T. 1974. Die Gebrauchsweisen des Mediums im Tocharischen. PhD diss., Georg-
August-Universität zu Göttingen.
. 1986. Bemerkungen zur westtocharischen Umgangssprache. In o-o-pe-ro-si. Festschrift für

Ernst Risch zum 75. Geburtstag, edited by Annemarie Etter, 635–49. Berlin: de Gruyter.
. 1992. Archaismen des Tocharischen und ihre Bedeutung für Fragen der Rekonstruktion
und der Ausgliederung. In Rekonstruction und relative Chronologie. Akten der VIII. Fachtagung der
Indogermanischen Gesellschaft, Leiden, 31. August - 4. September 1987, edited by Robert S.P Beekes,
Alexander Lubotsky, and Jos Weitenberg, 104–14. Innsbruck: Institut für Sprachwissenschaft
der Universität Innsbruck.
. 1997. Liebe und Sexualität im Spiegel der tocharischen Sprachzeugnisse. In Eros, Liebe und
Zuneigung in der Indogermania. Akten des Symposiums zur indogermanischen Kultur- und Altertum-
skunde in Graz (29. – 30. September 1994), edited by Michaela Ofitsch, 227–62. Graz: Leykam.
Schmitt, Rüdiger. 2007. Grammatik des Klassisch-Armenischen mit sprachvergleichenden Erläuterungen.

2nd ed. Innsbruck: Institut für Sprachwissenschaft.
Schulze, Wilhelm, Emil Sieg, and Wilhelm Siegling. 1931. Tocharische Grammatik. Göttingen: Venden-
hoeck & Ruprecht.
Seržant, Ilja A. 2014. Das Kausativ im Tocharischen. Munich: Lincom Europa.
340
Sideltsev, Andrej V. 2011a. Clitic Doubling: A New Syntactic Category in Hittite. Altorientalische
Forschungen 38 (1): 81–91.
. 2011b. Hittite Parallels for Balkan Sprachbund Clitic Doubling. Зборник 54:9–26.
Sieg, Emil. 1938. Die Kutschischen Karmavibhaṅga-Texte der Bibliothèque Nationale in Paris.
Zeitschrift für Vergleichende Sprachforschung 65:1–54.
. 1944. Übersetzungen aus dem Tocharischen I. Berlin: Akad. d. Wissenschaften.
. 1952. Übersetzungen aus dem Tocharischen II, aus dem Nachlass hg. v. Werner Thomas. Berlin:
Akademie-Verlag.
Sieg, Emil, and Wilhelm Siegling. 1921. Tocharische Sprachreste, I. Band. Die Texte. A. Transcription.
Berlin: de Gruyter.
. 1933. Bruchstück eines Udānavarga-Kommentars (Udānālaṃkāra?) im Tocharischen. In

Festschrift Moriz Winternitz 1863 – 23. December – 1933, edited by Otto Stein and Wilhelm Gampert,
167–73. Leipzig: Harrassowitz.
. 1949. Tocharische Sprachreste. Sprache B, Heft 1. Die Udānālaṅkāra-Fragmente. Text, Übersetzung

und Glossar. Göttingen: Vandenhoeck & Ruprecht.
. 1953. Tocharische Sprachreste. Sprache B. Vol. 2: Fragmente Nr. 71-633. Göttingen: Vanden-
hoeck & Ruprecht.
Sorace, Antonella. 2000. Gradience in Auxiliary Selection with Intransitive Verbs. Language 76:859–
90.
. 2004. Gradience at the Lexicon-Syntax Interface: Evidence from Auxiliary Selection and
Implications for Unaccusativity. In The Unaccusativity Puzzle: Explorations of the Syntax-Lexicon
Interface, edited by Artemis Alexiadou, Elena Anagnostopoulou, and Martin Everaert, 243–68.
Oxford: Oxford University Press.
Spencer, Andrew, and Ana R. Luís. 2012. Clitics. Cambridge: Cambridge University Press.
341
Stabler, Edward. 1997. Derivational Minimalism. In Logical Aspects of Computational Linguistics, edited
by C Retoré, 68–95. Berlin: Springer.
Stalnaker, Robert C. 1978. Assertion. In Syntax and Semantics, edited by Peter Cole, 9:315–32. New
York: Academic Press.
Stewart, Thomas W., and Gregory T. Stump. 2007. Paradigm Function Morphology and the Morphology-
Syntax Interface. In The Oxford Handbook of Linguistic Interfaces, edited by Gillian Ramchand
and Charles Reiss, 383–421. Oxford: Oxford University Press.
Straube, Martin. 2009. Studien zur Bodhisattvāvadānakalpalatā. Wiesbaden: Harrassowitz.
Strawson, Peter F. 1964. Identifying Reference and Truth-values. Theoria 30:96–118.
Stump, Gregory T. 1993. On Rules of Referral. Language 69:449–79.
. 2001. Inflectional Morphology. Cambridge: Cambridge University Press.
. 2002. Morphological and Syntactic Paradigms: Arguments for a Theory of Paradigm Link-
age. In Yearbook of Morphology 2001, edited by Geert E. Booij and Jaap van Marle, 147–80. Dor-
drecht: Kluwer Academic.
Stumpf, Peter. 1971. Der Gebrauch der Demonstrativ-Pronomina im Tocharischen. Wiesbaden: Harras-
sowitz.
. 1990. Die Erscheinungsformen des Westtocharischen. Ihre Beziehungen zueinander und ihre Funk-
tionen. Reykjavík: Málvísindastofnun Háskóla Íslands.
Takano, Yuji. 2003. How Antisymmetric is Syntax? Linguistic Inquiry 34:516–26.
Thieme, Paul. 1949. Untersuchungen zur Wortkunde und Auslegung des Rigveda. Halle: Max Niemeyer.
Thomas, Werner. 1952. Die tocharischen Verbaladjektive auf -l. Berlin: Akademie-Verlag.
342
Thomas, Werner. 1954. Die Infinitive im Tocharischen. In Asiatica: Festschrift Friedrich Weller. Zum
65. Geburtstag, gewidmet von seinen Freunden, Kollegen und Schülern, edited by Johannes Schubert
and Ulrich Schneider, 701–64. Leipzig: Harrassowitz.
. 1957. Der Gebrauch der Vergangenheitstempora im Tocharischen. Wiesbaden: Harrassowitz.
. 1983. Tocharische Sprachreste. Sprache B. Teil I: Die Texte. Band 1. Fragmente Nr. 1-116 der Berliner
Sammlung. Göttingen: Vandenhoeck & Ruprecht.
. 1997. Interpretationsprobleme im Tocharischen: Unflektiertes A puk, B po “ganz, all, jeder”. Stuttgart:

Steiner.
Thomas, Werner, and Wolfgang Krause. 1964. Tocharisches Elementarbuch. Vol. 2. Heidelberg: Win-
ter.
Uriagereka, Juan. 1999. Multiple Spell-out. In Working Minimalism, edited by Norbert Hornstein
and Samuel Epstein, 251–82. Cambridge, MA: MIT Press.
Vaidya, P. L. 1959. Divyāvadāna. Darbhanga: The Mithila Institute of Post-Graduate Studies and
Research in Sanskrit Learning.
Vakareliyska, Cynthia. 1994. Na-drop in Bulgarian. Journal of Slavic Linguistics 2:121–50.
Vallduví, Enric. 1992. The Informational Component. New York: Garland.
. 1993. Information Packaging: A Survey. Edinburgh: University of Edinburgh.
. 1995. Structural Properties of Information Packaging in Catalan. In Discourse Configura-

tional Languages, edited by Katalin É Kiss, 122–52. Oxford: Oxford University Press.
Van Windekens, Albert J. 1944. Morphologie comparée du tokharien. Louvain: Muséon.
. 1976. Le tokharien confronté avec les autres langues indo-européennes. Vol. 1. Louvain: Centre
International de Dialectologie Générale.
343
Van Hoof, Hanneke. 2006. Split Topicalization. In The Blackwell Companion to Syntax, edited by Mar-
tin Everaert and Henk van Riemsdijk, 408–62. Oxford: Blackwell.
Van Riemsdijk, Henk. 1989. Movement and Regeneration. In Dialect Variation and the Theory of
Grammar, edited by Paola Benincà, 105–36. Dordrecht: Foris.
Villalba, Xavier. 1998. Right Dislocation Is Not Right Dislocation. In Studies on the Syntax of Central
Romance Languages: Proceedings of the III Symposium on the Syntax of Central Romance Languages,
edited by Olga Fullana and Francesc Roca, 227–41. Girona: Universitat de Girona.
von Fintel, Kai. 2004. Would You Believe It? The King of France Is Back! Presuppositions and Truth-
value Intuitions. In Descriptions and Beyond: An Interdisciplinary Collection of Essays on Definite and
Indefinite Descriptions and Other Related Phenomena, edited by Anne Bezuidenhout and Marga
Reimer, 315–41. Oxford: Oxford University Press.
Von Heusinger, Klaus. 2019. Indefinites and Specificity. In The Oxford Handbook of Reference, edited
by Barbara Abbott and Jeanette Gundel, 146–67. Oxford: Oxford University Press.
Wackernagel, Jacob. 1892. Über ein Gesetz der indogermanischen Wortstellung. Indogermanische
Forschungen 1:333–436.
Watkins, Calvert. 1968–1969. The Celtic Masculine and Neuter Enclitic Pronouns. Études Celtiques
12:92–5.
. 1976. Towards Proto-Indo-European Syntax: Problems and Pseudo-problems. In Papers

from the Parasession on Diachronic Syntax, edited by Sanford B. Steever, Salikoko S. Mufwene,
and Carol C. Walker, 305–26. Chicago: Chicago Linguistic Society.
Weiss, Michael. 2018. Tocharian and the West. In Priscis Libentius et Liberius Novis: Indogermanische
und sprachwissenschaftliche Studien. Festschrift für Gerhard Meiser zum 65. Geburtstag, edited by
Olav Hackstein and Andreas Opfermann, 373–81. Hamburg: Baar.
. 2020. Outline of the Historical and Comparative Grammar of Latin. 2nd ed. Ann Arbor, MI: Beech
Stave.
344
Wilson, Deirdre, and Tomoko Matsui. 1998. Recent Approaches to Bridging: Truth, Coherence,
and Relevance. UCL Working Papers in Linguistics 10:173–200.
Winter, Werner. 1959. Zur „tocharischen“ Metrik. In Akten des XXIV. Internationalen Orientalistenkon-
gresses, München 28. August bis 4. September 1957, edited by Herbert Franke, 520–1. Wiesbaden:
Steiner.
. 1984. B śka, A śkā; B e; B ecce, A aci. In Studia Tocharica: Selected writings, 117–23. Poznán:
Uniwersytet Im. Adama Mickiewicza.
. 1992. [Numerals in] Tocharian. In Indo-European Numerals, edited by Jadranka Gvozdanović,

97–161. Berlin: de Gruyter.
. 1997. Lexical Archaisms in the Tocharian Languages. In Historical, Indo-European, and Lex-
icographical Studies. A Festschrift for Ladislav Zgusta on the Occasion of His 70th Birthday, edited by
Hans Henrich Hock, 183–93. Berlin: de Gruyter.
Yates, Anthony D., and John Gluckman. 2020. Voice Reversals and Syntactic Structure: Evidence
from Hittite. Glossa: a journal of general linguistics 5:120. 1–39.
Yokoyama, Olga T. 1986. Discourse and Word Order. Amsterdam: John Benjamins.
Yuan, Michelle. 2021. Diagnosing Object Agreement vs. Clitic Doubling: An Inuit Case Study. Lin-
guistic Inquiry 52:153–79.
Zhao, Wei. 2014. A Survey of Studies of Bridging Anaphora. Canadian Social Science 10:130–9.
Zwart, C. Jan-Wouter. 2009. Uncharted Territory? Towards a Non-Cartographic Account of Ger-

manic Syntax. In Advances in Comparative Germanic Syntax, edited by Artemis Alexiadou, Jorge
Hankamer, Thomas McFadden, Justin Nuger, and Florian Schäfer, 59–84. Amsterdam: John
Benjamins.
. 2011. Structure and Order: Asymmetric Merge. In Oxford Handbook of Linguistic Minimalism,
edited by Cedric Boeckx, 96–118. Oxford: Oxford University Press.
345
Zwicky, Arnold M. 1977. On Clitics. Bloomington: Indiana University Linguistics Club.
Zwicky, Arnold M., and Geoffrey K. Pullum. 1983. Cliticization vs. Inflection. Language 59:502–13.
Zyman, Erik, and Nick Kalivoda. 2020. XP- and X0 -movement in the Latin Verb: Evidence from
Mirroring and Anti-mirroring. Glossa: a journal of general linguistics 5:20. 1–38.
346

Pronominal Clitics in Tocharian

Uploaded by

Copyright:

Available Formats

Pronominal Clitics in Tocharian

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Pronominal Clitics in Tocharian

Uploaded by

Copyright:

Available Formats

UNIVERSITY OF CALIFORNIA

Pronominal Clitics in Tocharian:

A dissertation submitted in partial satisfaction

Pronominal Clitics in Tocharian:

Brent Harmon Vine

David Michael Goldstein, Committee Co-Chair

Hilda Koopman, Committee Co-Chair

University of California, Los Angeles

2 Pronominal clitics in Tocharian A and B . . . . . . . . . . . . . . . . . . . . . . . . . . 9

3 Descriptive data of the Tocharian pronominal clitics . . . . . . . . . . . . . . . . . . 42

5 Split intransitivity and clitic distribution in Tocharian . . . . . . . . . . . . . . . . 98

6 Pronominal clitics and multiple pronominal arguments . . . . . . . . . . . . . . . . 149

7 Clitic Doubling in Tocharian A and B . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182

9 Appendix: Representative unaccusative verbs in Tocharian A and B . . . . . . . . 258

Index of passages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315

2.1 〈ṣarpau me〉 in PKAS 6C a4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

6.1 Valency of host predicates in TA and TB . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150

1.1 Pronominal clitics in Tocharian A and B . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2.1 Independent personal pronouns in Tocharian A . . . . . . . . . . . . . . . . . . . . . . . 10

3.1 Summary of the chief uses of the Tocharian PCs . . . . . . . . . . . . . . . . . . . . . . . 75

5.1 Labile verbs in TA and TB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140

6.1 Hypotheses and predictions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149

7.1 Summary of (7.17ii) and (7.17iii) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191

9.1 Representative unaccusative verbs in TA and TB . . . . . . . . . . . . . . . . . . . . . . . 259

opt optative refl reflexive

perl perlative rel relative

prev preverb subj subjunctive

pst past top topic

ptcl particle voc vocative

2009 B.A. in Linguistics, Kyoto University

2011 M.A. in Linguistics, Kyoto University

2015–2017 Fulbright Fellowship, UCLA, the United States of America

2016 Teaching Assistant, Department of Classics, UCLA

2016–2021 Teaching Assistant, Department of Linguistics, UCLA

2021 Harry and Yvonne Lenart Graduate Travel Fellowships, UCLA

2021–2022 Dissertation Year Fellowship, UCLA

forthcoming. “Clitic Doubling in Tocharian B” In David M. Goldstein, Stephanie W. Jamison, and

1.1 Goals of this dissertation

In reconstructing the morphosyntactic system of a proto-language from attested languages X and

1.3 Tocharian morphosyntax and pronominal clitics

1sg 2sg 3sg pl

Table 1.1: Pronominal clitics in Tocharian A and B

1.4 Structure of this dissertation

Pronominal clitics in Tocharian A and B

2.2 The pronominal system of Tocharian A and B

2.2.1 Overview of the pronominal system of Tocharian A and B

TA 1sg.m 1sg.f 2sg 1pl 2pl

Table 2.1: Independent personal pronouns in Tocharian A

1. The accusative is traditionally referred to as the oblique.

Table 2.2: Independent personal pronouns in Tocharian B

TA 3sg.m 3sg.f 3sg.n 3pl.m 3pl.f

TB 3sg.m 3sg.f 3sg.n 3pl.m 3pl.f

Table 2.3: Demonstrative (anaphoric) pronouns in Tocharian A and B

1sg 2sg 3sg pl

Table 2.4: Pronominal clitics in Tocharian A and B

(2.2) [TB] -me = second-person plural

(2.3) [TB] -me = third-person plural

käṣṣī-ññe yäkne-sa asān-ne lyāmate-me ||

2.2.2 Hosts of the Tocharian PCs

(2.4) [TB] Finite verb hosting a PC