University of Southern California Dissertations and Theses

Children's scope of indefinite objects

(USC Thesis Other)

Children's scope of indefinite objects

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content CHILDREN’S SCOPE OF INDEFINITE OBJECTS
Copyright 2005
by
Fatma Nihan Ketrez
A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(LINGUISTICS)
August 2005
Fatma Nihan Ketrez
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UMI Number: 3196829
INFORMATION TO USERS
The quality of this reproduction is dependent upon the quality of the copy
submitted. Broken or indistinct print, colored or poor quality illustrations and
photographs, print bleed-through, substandard margins, and improper
alignment can adversely affect reproduction.
In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if unauthorized
copyright material had to be removed, a note will indicate the deletion.
®
UMI
UMI Microform 3196829
Copyright 2006 by ProQuest Information and Learning Company.
All rights reserved. This microform edition is protected against
unauthorized copying under Title 17, United States Code.
ProQuest Information and Learning Company
300 North Zeeb Road
P.O. Box 1346
Ann Arbor, Ml 48106-1346
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Dedication
For my parents
Ay§e Sevim Ketrez and Mehmet Ketrez
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Acknowledgments
I would like to express my gratitude to my committee members Hagit Borer, Nina
Hyams, Murat Kural, Toben H. Mintz, Audrey Li and Roumyana Pancheva, for their
invaluable guidance. Special thanks to my academic advisor Hagit Borer, for her
expertise, patience and the time she, very generously, devoted to my work. I am
grateful to Nina Hyams for her important contribution to my research and my way of
looking at child language. Our meetings have always been inspiring and motivating.
It was a great privilege to have Murat Kural in my committee. His comments have
been crucial especially for my discussion of the target adult grammar. Toben H.
Mintz has been an important part of this dissertation and my graduate study at USC.
I benefited greatly from all his suggestions and comments especially during the first
phase of my data collection and experiment design. I am grateful to Audrey Li and
Roumyana Pancheva for all the difficult and important theoretical questions that they
raised. I would also like to thank Ayhan Aksu-Ko?, my former advisor at Bogaziqi
University in Istanbul. She was not an official part of my dissertation committee but
she has been with me in every single phase of this dissertation and had a lot of
contribution to my research and training as a researcher. Thank you for being a great
mentor!
I am especially indebted to my professors at USC who taught me in various areas of
linguistics and helped me to develop my dissertation topic. Maria Luisa Zubizarreta
iii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
suggested me to work on scrambling almost four years ago when I was taking her
advanced syntax class. I was not interested in scrambling at that time, but finally my
dissertation went to that direction and my research benefited greatly from the
discussions I had with her. Amit Almor, Elaine Andersen, Carmen Silva-Corvalan,
Edward Finegan, Jack Hawkins, Joseph Aoun, Hajime Hoji, Barry Schein and Jean
Roger Vergnaud have contributed a lot to my way of looking at linguistic analysis,
methodology and syntactic theory. Tania Ionin has been around and always available
during the past two years. My short meeting with Elena Guerzoni was very useful in
many aspects. I was so lucky to have them here. Before I came to USC, I was a
student and a research assistant at the linguistics program at Bogazi9i University in
Istanbul. I am grateful to my professors at Bogazi9i University, who have been
always with me whenever I needed them. Ash Goksel, Eser Erguvanli Taylan,
Sabahat Sansa and Sumru Ozsoy taught me linguistics. Cevza Sevgen has always
been very supportive and encouraging.
I presented various phases of my research at the Psychobabble meetings at UCLA. I
am grateful to Nina Hyams, Carson Schutze and Susie Curtis who have been
providing me the opportunity to present and receive feedback from the members of
the Psychobabble group. I would like to thank Irene Kramer for discussions I had
with her when I was only starting to work on indefinite objects. Her dissertation,
which she, very kindly, sent to me, was the starting point of my study. I am grateful
to Jeannette Schaeffer for useful discussions of her analysis of Dutch objects and
iv
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
meetings during her stay in Los Angeles in Spring 2005.1 would also like to thank
Jeff Lidz for a short but interesting meeting I had with him during his recent visit at
USC.
Special thanks to Meltem Kelepir for a lot of linguistic discussions I had with her
during my dissertation study. My syntactic analysis of indefinites in this dissertation
benefited greatly from her analysis of indefinites in her dissertation. Bar 15 Kabak has
been one of my dearest friends that linguistics brought into my life. Many good ideas
came out of our discussions. Balkiz Oztiirk has been a great friend and colleague
since my masters study at Bogazigi, where I met her. Her recent dissertation, which
she provided to me, and analysis of referentiality in Turkish brought an interesting
perspective into my dissertation. I would like to thank Leyla Zidani-Eroglu, another
Turkish linguist, who worked on indefinites in Turkish, for discussions and exchange
of ideas. One of my starting points was the well-known LI paper of Murvet Eng. I
am grateful to her for her time to meet me at Boston and Istanbul. Many thanks also
to Karen Miller, Dina Brun, Ken Wexler and Jaklin Komfilt who shared with me
their recent unpublished work. Ana Santos, whom I recently met thanks to Nina
Hyams, is another person who had an important contribution to my analysis. I am
grateful to her for making her manuscript available to me and taking some time to
discuss my results and her hypothesis about focus in child grammar. Many thanks
also to Aylin Kiintay for her friendship and interesting discussions of language
development.
v
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
I would like to thank my friends at USC Linguistics for a lot of good and rewarding
memories that I will be taking with me: Agnieszka Lazorczyk, Alberto Esquinca,
Bella Feng, Carolina Gonzalez, Chris Long, Elena Filatova, Eunjeung Oh, Fetiye
Karabay, Fleur van der Houwen, Greg Placencia, Hatim Hassan, Hua Lin, Isabelle
Roy, Jelena Krivokapic, Justin Aronoff, Miae Lee, Monica Cabrera, Roberto
Mayoral, Shadi Ganjavi, Youssef Nouhi, Yukiko Tsuboi... Many thanks to Emily
Hinch who helped me to get rid of the Turkish accent in my dissertation and also to
Joyce Perez, our student advisor, for her help and friendship. Special thanks to my
dear friend Bella Feng, who has shared every single good thing and the bad thing
that I went through for the last six years. I am grateful to Fetiye and Davut Karabay,
who have been my family in Los Angeles. What a great coincidence it was to move
to the same city at the same time! I very often think that I would not be able to
survive without them.
Many thanks to my friends and colleagues whom I met at Bogazit^i: Ash Igsiz,
Atakan ince, Ay§e Giirel, Belma Haznedar, Ceyda Arslan, Cem £akir, Didar Akar,
Ercan Balci, Giinizi Kartal, Mine Guven, and Tugba Hacaloglu. I would like to
thank Zeynep Kulelioglu, of course, who has been a caring friend. This dissertation
would not be in its current form without her magic touch and 24-hour tech support.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Biggest thanks go to the participants of my study, of course! During the past three
years, I tested more than 300 children, who were randomly drawn from various
kindergartens in Istanbul and Izmit: Ayi§igi and Yeni £ilek in Levent, Koza in Etiler,
I§ik in Suadiye, Seymen in Izmit, Melodi in Kiigiikyah, and Bogazigi University
Preschool Education and Research Center in Bebek. I am grateful to these wonderful
little children, who shared their language and big worlds with me. I had a great time
while playing with them. I would like to thank their parents, teachers, directors and
the staff of the kindergartens who made me feel at home during my visits. I am
grateful to Ayhan Aksu-Kog, Nebile Akbulut and Esra Sel who helped me to get in
touch with these kindergartens, and obtain necessary permissions to collect data.
Many thanks are also due to the adult participants: undergraduate students at
Bogazigi University in Turkey, and graduate students at USC as well as many friends
both in Eos Angeles and Istanbul. Many thanks to Ayhan Aksu-Kog, Mine
Nakipoglu-Demiralp, Didar Akar, Beste Can Yolcu, Asli Ciger, Fethiye Ozi§ and the
research assistants at the the Psychology Department at Bogazigi University, who
helped me to find adult subjects.
I have not been doing the experiments by myself, of course. I had a chance to meet
and work with those wonderful people, who volunteered to help me in data
collection: Petek Akman, Gokge Aslan, Billur Avlar, Dilek Qelik, Derya Giirterler,
Reyan Kanyas, Deniz Tahiroglu and Ay§e Qaglar Ta§. Special thanks to Hakan
vii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Keser for his devotion, responsibility and interesting discussions during the
experiments.
A lot of people helped me in my preparation of the experiments as well. Very special
thanks to Erkan Gunal, who accepted to be the voice of Kermit. He, very patiently,
recorded all my test sentences, and also helped me with the technical matters. I really
appreciate what he has done for me. Ash Ciger, helped me during the data collection
and design of my pilot study. I benefited greatly from the discussions I had with
Ayhan Aksu-Kog, Toben H. Mintz, Nina Hyams, Carson Schiitze, Susie Curtis, Jose
Camacho, Tania Ionin, Miirvet Eng, Meltem Kelepir, Hakan Qermik, Fetiye
Karabay, Hande Tekdemir, inang Baykur, and Zeynep Kulelioglu. In Spring 2004,1
visited the language acquisition lab at the University of Maryland for a short period
of time and had a chance to attend some experiment sessions and meet faculty
members and students. Stephen Crain spared a lot of time to talk to me about my
experiments. I am grateful to him and also to Rozz Thornton for their hospitality and
invaluable help. I would also like to thank Luisa Meroni, Utako Minai and Graciela
Tesan and other members of the language acquisition lab for helpful discussions and
the great time I had during my visit. Special thanks to Nina Kazanina, ilhan (Jagn
and their families, who hosted me during my stay at College Park.
I would also to thank the members of my family Hayriye Kayagul, I§il Kayagiil,
Murat Kayagul, Ekrem Kayagul, Bekir Oguzmetin, Yelda Akm, Esra Sel, Kaan Sel,
viii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Se?kin Oguzmetin, Nebile Akbulut, Ozge Akbulut, Ekin Akbulut and especially my
brother Kamil Ilhan Ketrez and my grandmother Ltitfiye Akbulut for their emotional
support. Kaan Sel deserves special thanks for participating in my all kinds of
linguistics experiments since the day he was bom. I have also been dependent on
them for accommodation and transportation that they very generously provided
during my long stays in Istanbul. I would like to thank some very special people
outside of the field of linguistics, who have been bearing with me while I was trying
to become a linguist: Ahu Guler, Menek§e Giingor and Mtige Ta§km have been my
friends and full time emotional support for more than ten years now. Many thanks to
Antoine Karam, who has always there whenever I needed him. And, of course, I
would like to thank Alex Jalali for his love and every little thing that came along
with it.
I dedicate this dissertation to my mother Ay§e Sevim Ketrez and my father Mehmet
Ketrez for their unconditional love, support and understanding.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table of Contents
D E D IC A T IO N ..........................................................................................................................................II
A C K N O W L E D G M E N T S ................................................................................................................. Ill
LIST OF T A B L E S .............................................................................................................................X IV
LIST OF F IG U R E S .......................................................................................................................... X V I
A B S T R A C T ........................................................................................................................................... X X
1 IN T R O D U C T IO N .................................................................................................................................1
1.1 T h e g o a l ..........................................................................................................................................1
1.2 Theoretical f r a m e w o r k ......................................................................................................4
1.2.1 Children's clause structure.........................................................................................5
1.2.2 Parameter (M is)settin g................................................................................................ 7
1.2.3 M odularity H ypothesis................................................................................................ 8
1.3 Or g a n iza tio n of the d iss e r t a t io n..................................................................................9
2 N O N -C A S E M A R K ED O B JE C T S ............................................................................................ 13
2.1 Tu r k ish: A n o v e r v ie w .......................................................................................................... 13
2.2 B ar e O b je c t s............................................................................................................................. 18
2.2.1 Are bare objects non-referential?..........................................................................19
2.2.1.1 A djacency................................................................................................................20
2 .2.1.2 M odification ...........................................................................................................23
2.2.1.3 Pronom inalization and obj ect d ro p .............................................................. 26
2 .2 .1 .4 Comparison w ith nom inal co m p o u n d s.......................................................29
2.2.1.5 Comparison w ith light verb con structions................................................ 32
2.2.2 A spectual P rop erties..................................................................................................36
2.2.3 Are bare objects incorporated?...............................................................................38
2.2.4 C om plex Predicate Form ation................................................................................ 49
2.3 The D P projection a n d the s y n t a x of b i r ................................................................51
2.3.1 D P P rojection ................................................................................................................ 52
2.3.2 Bir and bir N o b jects.................................................................................................. 54
2.3.3 Bir N objects, referentiality and D P projection............................................... 60
2.3 .4 A sum m ary.....................................................................................................................61
2.4 S c o pe o f n o n - c a se m a r k e d o b j e c t s..............................................................................62
2.4.1 N e g a tio n ..........................................................................................................................62
2.4.2 A d v erb s............................................................................................................................63
2.5 C h a p t e r s u m m a r y ...................................................................................................................64
3 A C C U SA T IV E -M A R K E D IN DEFIN ITE O B JEC T S........................................................66
3.1 Interpretatio n of the a c c u sa t iv e m a r k ed o b j e c t s....................................... 70
x
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.1.1 A ccusative-m arked versus non-case-m arked obj ec ts...................................71
3.1.2 Partitivity and D -lin k in g ..........................................................................................76
3.1.3 Referentiality and out-of-the-blue ind efinites..................................................78
3.1.4 Presuppositionality......................................................................................................79
3.1.5 Are accusative marked indefinites quantifiers?..............................................81
3.1.6 Sum m ary and a generalization ............................................................................... 82
3.2 A c c u sa tiv e c a se a n d s c o p e .............................................................................................. 83
3.2.1 W ide S c o p e .................................................................................................................... 83
3.2.2 Can accusative marked indefinites take narrow sco p e? ................................ 87
3.2.3 A d jacen cy........................................................................................................................94
3.3 Pu tting ev ery th in g into st r u c t u r e........................................................................... 99
3.3.1 Presuppositionality within D P ............................................................................... 99
3.3.2 Scram bling as a w a y to escape from non-presuppositionality................101
3.4 Po stv er ba l O b j e c t s............................................................................................................104
3.5 A n alt er n a t iv e a c c o u n t for the a c c u sa t iv e c a s e ......................................... 107
3.6 Su m m a r y a n d C o n c l u sio n...............................................................................................108
4 E A R L Y IN D EFIN ITES, H Y PO TH ESES A N D PR E D IC T IO N S............................... 112
4.1 Iso m o r ph ism ............................................................................................................................. 112
4.1.1 A n y and som e p u z z le ...............................................................................................113
4.1.2 E very orange, som e guys and tw o slices o f p iz za ........................................ 115
4.1.3 C h a llen g es.....................................................................................................................119
4.2 Lin ea r v e r su s h ierarchical o r d e r in g .................................................................... 120
4.3 D isc o u r se In t e g r a t io n ..................................................................................................... 123
4.3.1 The concept o f non-shared k n o w le d g e.............................................................125
4.3.1.1 Schaeffer’s D utch gram m ar.......................................................................... 125
4.3.1.2 N on-discourse related interpretation o f o b je c ts....................................127
4.3.2 N on-integration h y p o th esis....................................................................................131
4.3.2.1 Kramer’s D utch gram m ar.............................................................................. 131
4.3.2.2 The non-integration hypothesis.................................................................... 133
4 .4 Ca r d in a l r ea d in g of indefinite a r t ic l e s...............................................................137
4.4.1 C hinese y i-ge and L exical Factor A n alysis......................................................138
4.4.1.1 S u ’s Chinese G ram m ar................................................................................... 138
4.4.1.2 The L exical Factor A n a ly sis......................................................................... 139
4.4.2 Spanish una and the scope o f bare singulars................................................... 141
4.4.3 R esponse am biguity problem ................................................................................ 143
4.5 Predictio ns for Tu r k is h .................................................................................................. 146
5 INDEFINITE OBJECTS A N D THE A D V E R B T W IC E ................................................151
5.1 Ex per im en t 1: A n a c t-o ut t a s k .................................................................................151
5.1.1 Participants....................................................................................................................152
5.1.2 P roced u re...................................................................................................................... 153
5.1.3 R esu lts.............................................................................................................................154
5.1.3.1 M ean scores.......................................................................................................... 154
xi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5.1.3.2 R esponse patterns..............................................................................................159
5.1.4 D iscu ssio n and C o n clu sio n ....................................................................................164
5.2 Ex perim ent 2: Follow -u p to Ex per im en t 1 ........................................................ 165
5.2.1 Participants...................................................................................................................165
5.2.2 P ro ced u re......................................................................................................................165
5.2.3 R esu lts............................................................................................................................167
5.2.3.1 M ean scores...........................................................................................................167
5.2.3.2 R esponse patterns...............................................................................................170
5.2.4 D iscu ssion and C o n clu sio n ................................................................................... 173
6 INDEFINITE O BJECTS A N D N E G A T IO N .......................................................................176
6.1 E xperim ent 3 ............................................................................................................................177
6.1.1 Participants................................................................................................................... 177
6.1.2 P ro ced u re......................................................................................................................178
6.1.3 R esu lts............................................................................................................................182
6.1.3.1 M ean scores...........................................................................................................182
6.1.3.2 R esponse patterns............................................................................................... 186
6.1.4 D iscu ssion and C o n clu sio n ....................................................................................190
6.2 Further D isc u ssio n of the Re s u l t s .......................................................................... 191
6.3 A REDESIGN PROPOSAL........................................................................................................... 196
7 A N O T H E R LO O K A T THE IN DEFINITE OBJECTS A N D N E G A T IO N 200
7.1 Ex perim ent 4: Follow-up to Experim ent 3 .......................................................... 201
7.1.1 Participants...................................................................................................................202
7.1.2 P roced u re..................................................................................................................... 204
7.1.3 R esu lts............................................................................................................................210
7.1.3.1 M ean scores.......................................................................................................... 210
7.1.3.2 R esponse patterns...............................................................................................216
7.1.4 D iscu ssion and C on clu sion ................................................................................... 221
7.2 Experim ent 5: Follow u p to Experim ent 4 ............................................................225
7.2.1 Participants...................................................................................................................225
7.2.2 P roced u re..................................................................................................................... 226
7.2.3 R esu lts............................................................................................................................226
7.2.3.1 M ean scores.......................................................................................................... 226
7.2.3.2 R esponse patterns...............................................................................................227
7.2.4 D iscu ssion and C on clu sion ................................................................................... 228
8 E V A L U A T IO N OF THE R ESULTS A N D THE PR O PO SA L ................................... 232
8.1 S u m m a r y of m ajor r e s u l t s........................................................................................... 232
8.2 S cope of n eg a t io n? .............................................................................................................233
8.3 Im plications for la n g u a g e acquisition h y po t h e se s......................................236
8.4 G etting b a c k in f o c u s...................................................................................................... 239
8.5 Im plications for Turkish cla u se st r u c t u r e...................................................... 246
9 A C C U SA T IV E C ASE O M ISSIO NS IN A TURK ISH C H IL D ’S S P E E C H 248
xii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
9.1 In t r o d u c t io n.......................................................................................................................... 248
9.2 M e t h o d....................................................................................................................249
9.3 Re s u l t s....................................................................................................................250
9.4 A n a l y s is................................................................................................................. 255
9.4.1 Grammatical production of nominal morphology...................................256
9.4.2 Asymmetry between accusative case and other nominal morphology 258
9.4.3 Evidence from the verbal dom ain.............................................................260
9.5 D isc u ssio n a n d C o n c l u s io n ............................................................................. 261
10 CONCLUSION............................................................................................................265
10.1 Su m m a r y of fin d ing s a n d im pl ic a t io n s...................................................... 265
10.2 Fu tu r e r e se a r c h................................................................................................268
REFERENCES.................................................................................................................. 271
APPENDIX 1: TEST ITEMS IN EXPERIMENTS 1 AND 2 ....................................285
APPENDIX 2: TEST ITEMS IN EXPERIMENT 3 .................................................... 288
APPENDIX 3: ‘NO OBJECT’ CONDITION IN EXPERIMENT 3 .........................294
APPENDIX 4: TEST ITEMS IN EXPERIMENT 4 AND 5 ......................................297
x iii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
List of Tables
Table 1: Hypotheses and predictions for Turkish.........................................................149
Table 2: 1-object reading in Experiment 1 (mean scores)...........................................155
Table 3: Response Pattern in Experiment 1 (adults).................................................... 159
Table 4: Response pattern in Experiment 1 (children)................................................. 161
Table 5: 2-object reading scores in Experiment 2.........................................................168
Table 6 : Response Pattern in Experiment 2 (adults).....................................................170
Table 7: Response pattern in Experiment 2 (children)................................................. 171
Table 8 : Dutch and Turkish twice-test results compared.............................................175
Table 9: Participants in Experiment 3 (children)...........................................................178
Table 10: Results of Experiment 3 (adults, mean accept scores)...............................182
Table 11: Results of experiment 3 (children, mean accept scores)............................ 183
Table 12: Response pattern in Experiment 3 (adults)...................................................187
Table 13: Response pattern in Experiment 3 (children)...............................................188
Table 14: Languages compared (wide scope readings, percentages)......................... 192
Table 15: Participants in Experiment 4 (accusative-marked objects)........................203
Table 16: Participants in Experiment 4 (non-case-marked objects)..........................203
Table 17: Results of Experiment 4 (adults, mean scores)........................................... 211
Table 18: Results of Experiment 4 (accusative-marked, children, mean scores) ....213
Table 19: Results of Experiment 4 (non-case-marked, children, mean scores)......213
Table 20: Response Pattern in Experiment 4 (accusative-marked objects)..............219
Table 21: Response Pattern in Experiment 4 (non-case-marked objects)................. 220
xiv
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 22: Summary of the conditions in Experiment 4 and 5.....................................225
Table 23: Results of Experiment 4 and 5 compared (mean scores)...........................226
Table 24: Response Pattern in Experiment 5 (non-case-marked objects).................228
Table 25: Hypotheses and predictions for Turkish revisited......................................236
Table 26: Objects in three different positions................................................................251
Table 27: Grammatical production and omission of the accusative case.................254
Table 28: Grammatical use and omission of nominal morphology...........................258
x v
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
List of Figures
Figure 1: 1-object reading in Experiment 1 ...................................................................155
Figure 2: Response pattern in “bir N-acc twice V” .......................................................162
Figure 3: Response pattern in “twice bir N-acc V”.......................................................162
Figure 4: Response pattern in “twice bir N V” ............................................................. 163
Figure 5: Response pattern in bare objects....................................................................163
Figure 6 : 2-object reading in Experiment 2....................................................................169
Figure 7: Response pattern in Experiment 2 (two object types m erged)...................172
Figure 8 : 1- and 2-object readings in Experiments 1 and 2......................................... 172
Figure 9: Results of Experiment 3 (Wide scope reading over negation)...................184
Figure 10: Response pattern in bir N-acc XP V -neg................................................... 188
Figure 11: Response pattern in bir N-acc V -neg..........................................................188
Figure 12: Response pattern in bir N V -neg..................................................................189
Figure 13: Response pattern in bare N V-neg............................................................... 190
Figure 14: Results of Experiment 4 (Object types and age groups)..........................214
Figure 15: Response pattern in bir N-acc DP V-neg structure....................................217
Figure 16: Response pattern in DP bir N-acc V-neg structure....................................217
Figure 17: Response pattern in DP V-neg bir N-acc structure....................................217
Figure 18: Response pattern in DP bir N V-neg...........................................................220
Figure 19: Response pattern in bare objects................................................................. 221
Figure 20: Results of Experiment 4 and 5 compared (mean scores).......................... 227
xvi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 21: Morphological variety in Deniz’s speech................................................... 258
Figure 22: Omission of the accusative case and other nominal morphology 260
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Abbreviations
ABL ablative
ACC accusative
AOR aorist
CAUS causative
CH I child
COM commutative
DAT dative
DIM diminutive
GEN genitive
INS instrumental
LOC locative
M OT mother
NEG negation
OPT optative
P plural (person agreement)
PASS passive
PAST past tense
PERF perfective
PL plural (morpheme)
POSS possessive
PRO pronoun
xviii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
PROGprogressive
5 singular (person agreement)
Q question particle
6 morphological fusion
+ compound marker
morpheme boundary
0 omission of the following morpheme
* 0 ungrammatical omission
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Abstract
This is an experimental study that tested children’s (age range 3;0-6;0)
comprehension o f indefinite objects in Turkish with a particular focus on the effect
of the accusative case marker on the scope assignment to objects with respect to
negation and adverbs. The results show that, although the accusative case is one of
the earliest acquisitions, emerging before 2 ;0 , its adult-like comprehension is not
achieved until a much later age. The results are attributed to children’s late mastery
of discourse pragmatics and information structure of the language.
Presuppositional objects obligatorily (overtly or covertly) scramble in Turkish, while
non-presuppositional ones occur adjacent to verbs, either immediately above VP, or
inside VP. When objects remain adjacent to the verb, their case can be deleted.
Scrambled objects cannot undergo case deletion. In this sense, accusative case
signals a scrambled position, thus a presuppositional reading, and wide scope with
respect to negation and adverbs, although the case marker itself does not play a role
in scope assignment.
In a series o f experiments case-marked indefinite objects were tested at various word
orders (preverbal adjacent to verb, preverbal preceding the subject and postverbal)
and their interpretation was compared to the interpretation of non case marked
objects. The children at the youngest age group did not have access to the wide scope
xx
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
in any of the object types regardless of case morphology or word order. This reading
is observed in about 95% of the cases in adults when the objects are case-marked. An
improvement is observed by age, but even at the age of 6 ;0 , children cannot reach an
adult-like level in their scope assignments.
I propose that that children’s grammar has case assignment and scrambling as formal
operations but children do not know the complex relationship between case
morphology, scrambling, presuppositionality, defocalization of constituents and
scope assignment. These findings are compatible with a grammar that separates the
formal and interpretational properties of morphemes and allows their emergence at
their own pace.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1 Introduction
1.1 The goal
The English sentence in (1) is ambiguous. It may mean that Mickey Mouse did not eat
any apples. It may also refer to a situation where Mickey Mouse did not eat one of the
apples. In the former reading the indefinite object has narrow scope with respect to
negation (i). In the latter one, it has wide scope (ii).
(1) Mickey Mouse didn’t eat an apple.
(i) It is not the case that MM ate an apple (no apples are eaten)
(ii) There is an apple such that MM didn’t eat (one of the apples was not
eaten)
In Turkish, these two meanings may be differentiated with the presence or absence of
overt accusative case morphology. The object without a case morpheme has narrow
scope reading with respect to negation and the sentence reads as ‘Mickey Mouse did not
eat any apples’(2) The object with the overt accusative marking has the wide scope
interpretation where only one of the objects was not eaten (3).
(2) Miki Fare bir elm a-0 ye-me-di.
Mickey Mouse an apple eat-neg-past
‘It is not the case that MM ate an apple.’
1
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(3) Miki Fare bir elma-yi ye-me-di.
Mickey Mouse an apple-acc eat-neg-past
‘There is an apple such that MM didn’t eat.’
This is a very simplified version of the behavior of objects in Turkish. As I will be
showing in the following chapters, the differences between case marked and non-case
marked objects are much more complicated than what is stated above and reported in
most literature. In addition to the examples above, Turkish allows postverbal and
preverbal scrambling o f overtly case-marked objects as exemplified in (4) and (5).
Moreover, it has bare objects, which, as I will argue, form complex predicates with the
verb (6 ) and always take narrow scope with respect to negation and adverbs.
(4) [Bir elma-yi] Miki Fare ye-me-di.
[an apple-acc] Mickey Mouse eat-neg-past
‘There is an apple such that MM didn’t eat it.’
(5) Miki Fare ye-me-di [bir elma-yi],
Mickey Mouse eat-neg-past [an apple-acc]
‘There is an apple such that MM didn’t eat it.’
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
( 6 ) Miki Fare [compiexPredicate elma ye]-me-di
Mickey Mouse apple eat-neg-past
‘It is not the case that MM ate an apple.’
In this study, I propose an analysis of indefinite objects in Turkish and investigate the
factor(s) that affect(s) children's scope of indefinite objects with respect to negation and
adverbs. Possible factors that are investigated include (i) accusative case morphology, (ii)
the numeral bir 'a/one' which occurs with some indefinite objects (iii) word order (iv)
adjacency to the verb. These factors were targeted based on the properties of the adult
grammar and the predictions for child language theories as based on findings in other
languages. The main focus of the study will be the interpretation of the accusative case
under various conditions.
Turkish is known to be one of those languages whose morpho-syntax is acquired very
early (Ekmekgi, 1979, Aksu-Kog & Slobin, 1985, among others). Children acquiring
Turkish have strikingly fewer errors in their speech when compared to children acquiring
other languages (Slobin, 1985). The accusative case is among the earliest to be acquired.
Children start producing the accusative case in obligatory contexts even before the age
1;6 (Ekmekgi, 1986). They also use it to distinguish objects from subjects (Slobin &
Bever, 1982). Earlier works on the acquisition of the accusative case show that children
can produce the accusative case when it is obligatory but they do not show whether
children have an adult-like differentiated interpretation of the objects with and without
3
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
overt case morphology. No study has been done on the acquisition of the accusative
marked indefinites in Turkish, although it has been one of the most widely discussed
topics in Turkish generative linguistics especially since En? (1991).
In the present study, I focus on one of the dimensions of the acquisition pattern of
indefinite objects, while paying close attention to the role of accusative case in their
interpretation. I will argue that although children start producing the accusative case as
early as 1;3, an adult-like comprehension of the case-marked versus non-case- marked
objects is mastered only after age 6;0.1 report children's behavior as evidence that formal
and discourse related properties of morphemes can be separated. I argue that morpho-
syntax emerges early but an adult-like interpretation of structures come much later
because it requires a mastery discourse pragmatics as well as an adult-like interpretation
of the relationship between prosody and information structure. The acquisition of
indefinite objects is thus interesting because it is one of those areas where morphology,
syntax, prosody and information structure all interplay important roles in the construction
of the interpretation.
1.2 Theoretical framework
The framework adopted in this dissertation is the Principles and Parameters Theory,
which crucially assumes the existence of Universal Grammar (Chomsky, 1975, 1981,
1986). Universal Grammar is the set of principles and parameters through which
variations across languages can be defined. According to this theory, children have
4
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
innate access to a rich and structured linguistic knowledge and as guided by this
knowledge they analyze the input speech they receive. In this framework, language
acquisition is a process through which a child acquires the language specific properties of
their target grammar.
1.2.1 Children's clause structure
Grammar is divided into two major categories. Lexical categories, which include noun,
verb, adjective, preposition etc. and functional categories such as complementizer,
determiner, or inflection. Whether or not children have a full and adult-like syntactic
structure has been one of the most controversial issues in child language acquisition.
According to some early clause structure theories, children lack some (or all) of the
functional structure (Radford, 1988, 1990, 1994, Guilfoyle & Noonan, 1988, Lebeaux,
1988, Platzack, 1990, 1992, among others). These hypotheses are based on the
observation that children leave out inflections or functional words, such as
complementizers that correspond to functional categories in their early speech. According
other proposals, children have some functional structure (Meisel & Mueller, 1992,
Clahsen & Penke, 1992, among others). Meisel & Mueller (1992), for example, propose
that children have the functional structure up to IP but they lack CP. Evidence for this
hypothesis is that children can produce subject-verb agreement, but they omit
complementizers and fail to do subject-aux inversion.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Yet another proposal is that functional categories are present even at the earliest stages of
language development and omission of inflections and complementizers cannot be
attributed to the lack of functional categories (Verrips & Weisseenbom, 1992, Hyams,
1992, Poeppel & Wexler, 1993, Borer & Rohrbacher, 2003, among others). Borer and
Rohrbacher (2003), for example, propose that the systematic omission of inflection, just
as their systematic use, provides evidence that the positions that hold these inflections
exist in child grammar.
Whether or not children have an adult-like functional structure is not directly addressed
in this dissertation. However, I will be crucially assuming that children have access to the
full syntactic structure during the period I study (3;0-6;0) and I will be discussing some
relevant data that support this assumption when necessary. I will be attributing children’s
non-adult-like comprehension of the particular structures that I am studying to a late
acquisition of other components of grammar such as information structure and discourse
understanding.
In the following sections I discuss two general models of language development that
address to the questions o f what is missing in child grammar and at which particular
points child grammar is different from adult grammar.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1.2.2 Parameter (Mis)setting
According to the Parameter Setting model (Hyams, 1983, 1986), children (mis)set a
parameter to a default value in their initial stage of language development and as a result,
produce structures that do not necessarily exist in their target language. In later stages,
they reset the parameters until they reach their target language. For example, according to
this hypothesis, early child languages can be analyzed as null-subject languages like
Italian. All children start with a null-subject grammar and then those acquiring non-pro
drop languages such as English reset their parameters.
Although the Parameter Setting model has advantages, such as a continuous learning
path, which makes the language acquisition process much simpler by paralleling the adult
and child languages, it is not supported empirically. Child grammars are different from
the grammars of adult null-subject languages like Italian, which have a rich inflectional
system. During the null subject period child grammars do not have rich inflections.
Moreover, null subject languages have modals and lexical expletives, while child
grammars do not (Valian, 1990) and cross linguistically, children acquiring non-null-
subject languages behave differently from those children who are acquiring null-subject
languages. Italian children and adults drop subjects in 70% of the pro-drop contexts,
while English speaking children drop only 30% of their subjects around the same time
(Valian, 1990, 1991, 1994, Valian & Eisenberg, 1996). In addition, children have null
subjects in positions where null subjects cannot occur in Italian-like adult grammars
(Roeper & Weissenbom, 1990, Rizzi, 2000, Haegeman, 1996, among others).
7
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Due to these more recent data and arguments, the Parameter Setting model is no longer
typically assumed, and it has been replaced by the Modularity Hypothesis, which will be
presented in the next section.
1.2.3 Modularity Hypothesis
According to the Modularity Hypothesis (c.f. Fodor, 1983, Chomsky, 1986) grammar is
composed of autonomous components that are governed by distinct principles. The
components of grammar are (i) lexicon, (ii) the computational system (grammar that
includes syntax, semantics, morphology, phonology and a processor), and (iii)
pragmatics, which include the discourse / information structure. Each of these
components may develop independently with its own speed and the interaction of the
components with each other is effected by the development and status o f each component
at particular phases of language development. A non-adult-like pragmatic component,
for example, may have a significant impact on the production or comprehension of
particular morpho-syntactic structures (e.g., object displacement, valency alternations,
anaphor binding) which require an interaction of morpho-syntax and semantics with
pragmatics. Avrutin and Wexler (1992), for example, show that children have the
principles of the binding theory, but they lack a discourse related principle that could
enable them to interpret the binding relationship in Russian and the absence of this
discourse principle results in non-adult-like behavior in child speech.
8
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In this dissertation, I attribute the delay of the interpretation of a morpho-syntactic factor
(i.e., the accusative case morphology) to the late acquisition of information structure,
which is dependent, in adult Turkish, on object scrambling and prosodic variation.
1.3 Organization of the dissertation
The following chapters of this dissertation are organized as follows. In Chapter 2 ,1
provide a general description of the language and discuss the properties of non-case
marked objects. I propose that bare objects and non-case marked bir N objects are
structurally different. Bare objects do not have NumP and DP projections and they are
not referential. They are in a dependent relationship with verbs, but they do not
incorporate. Rather, they form complex predicates with verbs. On the other hand, bir N
objects have NumP and DP projections. They move out of YP to AGRoP, which hosts
the non-presuppositional object arguments. They are marked with syntactic case but the
case morpheme that they bear is deleted under adjacency to the verb. They do not
scramble because they are non-presuppositional and their semantic interpretation is
compatible with the position in which they occur. Both bare objects and bir N objects
appear in a relatively low position in the clause structure and take narrow scope with
respect to negation and adverbs.
Chapter 3 is devoted to accusative marked indefinite objects. In that chapter, I contrast
syntactic as well as interpretational properties of accusative marked objects with those of
non-case marked objects. Specifically, I propose that presuppositional objects need to
9
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
scramble to escape the non-presuppositional field of AGRoP. When the object is in a
scrambled position, i.e., when it is not adjacent to the verb, its case cannot be deleted.
Therefore, overt accusative case indicates a scrambled position and presuppositionality.
Even those accusative marked objects that appear adjacent to the verb are analyzed as
scrambled to a position above AGRoP. Under this analysis, the accusative case is only a
formal case marker, rather than a scope or specificity marker. The condition on its
deletion is sensitive to the position of the object, which in turn determines its
interpretation. Therefore overt accusative case signals the position of the object. Once the
object is outside of the non-presuppositionality field, it has wide scope with respect to
negation and adverbs regardless of its linear position with respect to the other
constituents. Overtly accusative marked objects thus contrast with the non-case marked
objects in their scope-taking properties.
In Chapter 4 ,1 discuss various studies conducted on the acquisition of indefinite objects
in typologically different languages. I focus on four main hypotheses on the acquisition
of indefinite objects: (i) isomorphism, (ii) hierarchical organization and c-command, (iii)
The Lexical Factor Hypothesis, and (iv) discourse integration models. I present these
accounts, and complete the chapter discussing their predictions for the acquisition of a
language such as Turkish.
Chapters 5, 6, and 7 are devoted to the discussion of experimental studies and their
results. Chapter 5 presents the first series of experiments that tested children’s
comprehension of accusative marked versus non-case marked objects in interaction with
10
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the adverb twice. The experiments show that children’s overtly accusative marked
indefinites do not necessarily take wide scope with respect to the adverb twice. Chapter 6
presents an experiment that tests children’s scope of indefinite objects with respect to
negation. In contrast to the experiments with twice, children have an adult-like
performance in the negation experiment. In other words, children’s overtly accusative
marked indefinite objects take wide scope with respect to negation in more than 80% of
the cases. In Chapter 7, two follow-up negation experiments were presented. These new
experiments show that children have a non-adult like comprehension of indefinite objects
and the results of the first negation experiment are due to an experimental bias that
affected the results. All the experiments presented in Chapters 5, 6, and 7, taken together,
show that children go through a developmental period where they have a non-adult like
interpretation of overtly accusative marked indefinite objects. Word order, adjacency to
verb, or the presence of a numeral does not have an impact on their treatment of
indefinite objects.
In Chapter 8 ,1 evaluate the results and their implications. I attribute this non-adult-like
behavior to their treatment of the accusative case as a formal case marker rather than a
signal for the scrambled structural position of the object and hence its
presuppositionality. The child interpretation of objects show that children have not yet
mastered the interpretation of the prosodic variations of the focus shifts and have a
tendency to interpret objects in their default Case position (AGRoP), which is also the
neutral focus position.
11
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 9 discusses the emergence of case markers in spontaneous child speech,
providing independent evidence for the claim that the accusative case has a non-adult-like
status in children’s speech. Specifically, I show that the child’s case omission errors do
not result from a syntactic deficiency, but rather they should be attributed to the late
acquisition of the discourse related interpretation of accusative case. Chapter 10 offers
some concluding comments.
12
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 Non-case marked objects
This is the first of the two chapters devoted to the description of the target adult grammar.
The sections in this chapter present a description of the two non-case marked objects:
bare objects and bir N objects. Chapter 3 will focus on the accusative-marked objects. I
start the discussion of the adult grammar with a section that presents general properties of
the language. In Section 2.2 that follows, I present the bare objects and contrast them
with bir N objects. Section 2.3 presents the internal structure of bir N object DPs with a
particular focus on the syntax of bir “a/one,” and further discusses the structural
differences between bare objects and bir N objects. Section 2.4 presents the scope-taking
properties of the non-case marked objects. Section 2.5 summarizes the chapter.
2.1 Turkish: An overview
Turkish is an agglutinative language in which inflection is mainly realized by suffixation.
Verbs carry tense-aspect-modality and subject-verb agreement markers (7). Negation is
marked by the negative suffix — mA, which occurs immediately following the verb stem.
(7) Ko§-mu-yor-um.
run-neg-prog-lS
‘I am not running’
13
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Nouns can carry possessive, plural marker and case depending on various syntactic and
semantic conditions (8). Nominals (and nominalized verbs) can be marked with
accusative, dative, locative, ablative and genitive cases. Nominative case does not have a
phonological realization. Turkish does not have grammatical gender. Suffixes undergo
vowel harmony and consonant assimilation according to the phonological properties of
the stem (8).
(8) Araba-lar-im-a / ev-ler-im-e
Car-plu-poss& 1 S-dat / house-plu-poss& 1 S-dat
‘To my cars / to my houses’
Morphologically bare nouns can appear in argument positions. Turkish does not have
articles. Definiteness/specificity is marked with case morphology in the case of the
objects1 , with a stress shift in the case of the subjects as seen in the examples in (9), (10),
(11) and (12) respectively.
(9) Ay§e elma ye-di. non-case marked object
Ay§e apple eat-past
‘Ay§e did apple eating (=Ayse apple-ate).’
1 In Chapter 3 ,1 will challenge a direct relationship between case marking and specificity.
14
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(10) Ay§e elma-yi ye-di. case-marked object
Ay§e apple-acc eat-past
‘Ayse ate the apple.’
(11) Gem! ge9-iyor. stressed subject
Ship pass+by-prog
‘There is ship-passing.’
(12) Gemi geg-iyor. non-stressed subject
Ship pass+by-prog
‘The ship is passing by.’
It is a head final, SOV language but a variety of word orders are possible as long as the
scrambled constituents are properly marked with overt case morphology (Erguvanli 1984,
Kural, 1992; Kelepir, 1997 among others). The sentence in (13), (14) and (15) are
examples for a verb, a postposition noun and head in the final position.
(13) Ay§e ^ocuk-lar-i gor-dii. verb head
Ay§e child-plu-acc see-past
‘Ay§e saw the children.’
15
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(14) Qocuk-lar i9in.
Child-plu for
‘For children.’
postposition head
(15) Ay§e-nin ara-dig-i focuk-lar. noun head
Ay§e-gen look+for-DIK-poss child-plu
‘The children that Ay§e is/was looking for.’
The three types of objects that are studied are shown in (16). In (16)a, we have the
numeral bir and the object appears in a non-case marked form2. In (16)b, the object is
marked with the accusative case and it appears adjacent to the verb. An object can
scramble freely when it is overtly case marked. Examples can be seen in (16)c and (16)d.
Turkish has bare objects, as well, which are exemplified in (16)e. Definite objects, which
bear the accusative case without bir (an example is presented in (10) above), will not be
discussed, as the scope of the study is limited to the indefinite objects.
(16) a. Ay§e bir elma ye-di. [hirN V ]
Ay§e an apple eat-PAST
‘Ay§e ate an apple.’
2 In bir N(-acc) Verb type of indefinite objects, the focal stress is never on bir, which would give it an
unambiguous cardinal number interpretation. Unless I state otherwise, bir does not have focal stress in the
objects that I discuss. I will be discussing the focal stress in section 2.3.2.
16
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
b. Ay§e bir elma-yi ye-di. [bir N-acc V]
Ay§e an apple-ACC eat-PAST
‘Ay§e ate a particular apple.’
c. Bir elma-yi Ay§e ye-di. [bir N-acc D P subj V]
an apple-ACC Ay§e eat-PAST
‘Ay§e ate a particular apple.’
d. Ay§e ye-di bir elma-yi. [V bir N-acc ]
Ay§e eat-PAST an apple-ACC
‘Ay§e ate a particular apple.’
e. Ay§e elma ye-di. [N V]
Ay§e apple eat-PAST
‘Ay§e performed apple-eating.’
The following sections, and Chapter 3 present a description of the scope properties of
three different object types and address the following issues: (i) bare objects and their
contrast with other types of objects, (ii) the internal structure of object DP with a
particular concentration on the syntax of bir “a/one,” (iii) the role of case morphology
and scrambling on scope assignment of indefinite objects.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2.2 Bare Objects
Bare nominals, which are exemplified in (17), are analyzed in the literature either as
object-incorporations (Mithun, 1984, Nilsson, 1985,1986, Komfilt, 1997, 2003,
Schroeder, 1999, Aydemir, 2004 among others), compounds (Swift 1963, Knecht 1986)
or “non-defmite” nominals (Dede, 1986) in the sense that the nominal lacks the [±
definite] feature.
(17) Ay§e elma ye-di.
Ay§e apple eat-past
'Ay§e apple-ate.' (=Ay§e performed apple-eating)
In Ketrez (2004b), I proposed that bare nominals lack both DP and NumP projections and
thus enter a complex predicate formation with verbs that they immediately precede. Due
to the lack of these functional projections, they do not have definiteness or number
features. In the present study, I pursue a version of this idea. I show that the bare object
and the verb are in a dependent relationship due to the absence of DP and NumP
projections, which prevent the movement of the object out of the VP projection. I also
show that they have a loose dependency unlike compounding and noun incorporation
examples reported in the literature and only a complex predicate analysis that will treat
the verb and the object as two distinct lexical items can capture their structural properties
(18). The bare objects do not undergo an incorporation operation through which they
18
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
become one lexical unit with the verbs. However, they do not behave like full arguments
either, due to the absence of NumP and DP.
(18) A complex predicate formation
In the following sections, I present the properties of the bare nouns and discuss various
analyses. In Section 2.2.1,1 argue that bare objects are not referential. In Section 2.2.3,1
further argue that although they are not referential, they do not get incorporated to verbs.
Section 2.2.4 summarizes the discussion and presents the structures adopted in this study.
2.2.1 Are bare objects non-referential?
The evidence for the lack o f DP, nonreferentiality and the complex predicate come from
a number of properties. These are adjacency, modification by adjectives and relative
clauses, pronominalization and aspectual properties of the verb-object complexes. I will
be comparing bare objects with bir N object when the contrast is important for the
discussion of the bare objects. Bir N will be discussed in more detail in Section 2.3.
VP
kitap ‘book’ oku‘read’
19
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2.2.1.1 Adjacency
The bare nominal occurs adjacent to the verb, i.e., scrambling is not allowed and no
constituent, except the focus particles, can intervene between the nominal and the verb, as
exemplified in (19) and (20). Bare objects share the adjacency property with bir N
objects, but they differ in other aspects as will be discussed below.
(19) *Ay§e oku-du kitap.
Ay§e read-past book
(20) *Ay§e kitap dun oku-du.
Ay§e book yesterday read-past
Focus particles such as the question particle m l and the adverb bile ‘even’ can intervene
between the bare object and the verb (see examples in (21), (22) and (23)). These are
problematic for any head incorporation analysis, which would analyze the noun and the
verb head as one single lexical unit. These examples rather suggest that the bare objects
are phrasal categories (but see the following discussion).
(21) Ay§e elma mi yi-yor?
Ay§e apple Q eat-prog
‘Is Ay§e eating (an) apple / apple-eating’?
20
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(22) Ay§e elma bile yi-yor?
Ay§e apple even eat-prog
‘Ay§e is eating even an apple / even apple-eating.’
(23) Ay§e elma da yi-yor.
Ay§e apple too eat-prog
‘Ay§e is eating (an) apple, too / apple eating, too.’
However, the status of the focus particles is controversial. They can appear within verbal
complexes as well, as seen in the examples below. In (24) the question particle appears
between the aspect and the tense particles5 . In (25) the particle DA occurs within the
abilitative marker -yAbil. The fact that the focus elements undergo vowel and consonant
harmony suggests that they are a part of the phonological word in Turkish. This
flexibility of occurrence is restricted to the question particle and the particle DA. The
other focus particle bile ‘even’ does not occur in verb complexes.
(24) Anla-ma-mi§ mi-ydi-mz?
Understand-neg-perf que-past-2P
‘Hadn’t you understood?’
3 See Besler (1999) for a detailed discussion of the positions where the question particle can appear.
21
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(25) Anla-ma-ya-da-M-ir-di-niz.
Understand-neg-abil]-dA-abil2-aor-past-2P
‘You might also not understand.’
These examples show that the occurrence is not restricted to the edge of phrases4. They
can appear within words. As I will discuss in Section 2.2.1.5 below, they can appear in
complex predicates such as light verb constructions, as well. The flexibility of the m l and
DA particles are stated as evidence to argue against the claim that bare objects and verbs
are not adjacent. (See, for example, Komfilt (2003) as the most recent discussion).
Although it is true that m l and DA have unexpected distributional properties, their
occurrence in between the bare object and the verb is still problematic for an analysis that
adopts a strict head incorporation (c.f., Baker, 1988)) or lexical compounding model.
Turkish contrasts with the other traditional incorporation languages that I will discuss in
more detail below.
4 There has not been any proposal for the syntactic position of these particles that can appear within verbs.
Goksel (1993) discusses the occurrence of DA in this position as a problematic example for Baker’s (1982)
mirror principle but does not propose an analysis. According to Erdal (2000), DA is a clitic and clitics can
appear anywhere in Turkish. Korkmaz (2003) analyzes yA bil as a complex predicate, which consists of the
adverb yA and a verb bil. Ban§ Kabak (p.c.), adopting Korkmaz (2003), suggests that the occurrence of DA
in between the two parts is not problematic because there is a ‘word boundary’ between them although they
are spelled as one single word. A property that could support this suggestion is the absence of vowel
harmony in yAbil. However, Turkish allows non-harmonic word stems, so the absence of vowel harmony
does not mean that they are not a single stem, it only means that bil was not attached to yA as a suffix,
which has not been claimed.
22
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In summary, bare objects occur adjacent to verbs. The only exceptions are the focus
particles, but their rather flexible distributional properties suggest that their occurrence
between the bare objects and the verb is not problematic for a complex predicate
formation analysis as long as the occurrence of two counterparts as two distinct lexical
units is allowed.
2.2.1.2 Modification
Another property is modification. Modification of bare objects by an object relative
clause is not possible as seen in (26). Note that the structure is grammatical when we
have an object modified by bir (27). The nominal in the example with bir is obligatorily
referential and singular as will be discussed in Section 2.3. Modification of bare objects
(as well as bir N objects) is possible by subject relative clauses in examples such as (28)
and (29). The examples (30) and (31), however show that the modification by subject
relative clauses, too, is restricted5. These restrictions, too, seem to apply to the bare
objects only, as the bir N counterpart is grammatical.
(26) *Ay§e [[herkes-in sevdigi/sev-eceg-i] kitap] ari-yor.
Ay§e [[mother-poss-gen buy-REL] book] read-past
‘Ay§e read books that her mother bought.’
5 The contrast between the examples (28) and (30) needs further investigation.
23
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(27) Ay§e [[herkes-in sevdigi/sev-eceg-i] bir kitap] an-yor.
Ay§e [[mother-poss-gen buy-REL] book] read-past
‘Ay§e read books that her mother bought.’
(28) Ay§e [Ingilizce bil-en] sekreter ari-yor.
Ay§e [English speak-REL] secretary looking for
‘Ay§e is looking for (a) secretary who can speak English.’
(29) Ay§e [Ingilizce bilen] bir sekreter ari-yor.
Ay§e [English speak-REL] secretary looking for
‘Ay§e is looking for (a) secretary who can speak English.’
(30) *Ay§e [[ba§ka ulke-ler-i anlat-an] kitap] oku-du.
Ay§e [[other country-plu-acc tell-REL] book] read-past
‘Ay§e read books that are about other countries.’
(31) Ay§e [[ba§ka ulke-ler-i anlat-an] bir kitap] oku-du.
Ay§e [[other country-plu-acc tell-REL] a/one book] read-past
‘Ay§e read a book that is about other countries.’
24
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
When a bare object is preceded by an adjective, which can have an adverb interpretation,
the adverbial reading is forced, where the adjective/adverb modifies the event (bare
object + verb complex) rather than the object alone6. Therefore in (32), the interpretation
is that Ayse is driving fast. Ayse is driving fast cars is also possible but the adverb
interpretation is more salient. With another verb, the same word fast sounds better in the
adjective sense (33).
(32) Ayse hizli araba kullan-iyor.
Ayse fast car drive-prog
‘Ayse is driving fast.’ / ‘*Ayse is driving a fast car’.
(33) Ayse hizli araba ari-yor.
Ayse fast car look-for-prog
'Ayse is looking for a fast car' / 'Ayse is looking for a car fast.'
When the adjective cannot have an adverb reading, as it is the case with yesil ‘ green’ in
(34), the only interpretation available is the adjective interpretation where the adjective
modifies the bare object.
6 In Turkish, as it is the case with some adverbs in English, such as fast, the adverb and the adjective forms
o f the words is morphologically the same.
25
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(34) Ay§e ye§il elma yi-yor.
Ay§e green apple eat-prog
‘Ay§e is eating a green apple.
The possibility of the adverb reading suggests that bare objects and verbs behave as one
single unit. The possibility of the adjective reading suggests that the nominal can be
treated as one single word independent of the verb. As far as the non-referentiality is
concerned, I still conclude that the object is non-referential but it is more than a bare
noun. I analyze it as a bare NP, rather than a bare N, which can have a modifier in [spec,
NP] position7.
2.2.1.3 Pronominalization and object drop
Restriction in pronominalization and object drop is another property that supports the
non-referentiality proposal. Only objects that are referential and that have a DP projection
can be pronominalized and dropped. The example of the ungrammaticality of
pronominalization and object drop are given in (35)a and b. The same structure is
grammatical with bir as seen in (36)a and b.
7 See Oztiirk (2004) for the same argument for bare objects in Turkish.
26
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(35) a. *Ay§e kitapi oku-du, sonra Oj-nu kiitiiphane-ye geri ver-di
Ay§e book read-past, then it-acc library-dat back give-past
'Ayse did booki-reading, then returned iti to the library.'
b. *Ay§e kitap; oku-du, sonra pro; kiitiiphane-ye geri ver-di
Ay§e book read-past, then library-dat back give-past
'Ayse did book;-reading, then returned (it;) to the library.'
(36) a. Ay§e bir kitap; oku-du, sonra 0 ;-nu kutiiphane-ye geri ver-di.
Ay§e a book read-past, then it-acc library-dat back give-past
'Ayse read a book;, then returned it; to the library.'
b. Ay§e bir kitap; oku-du, sonra pro; kiitiiphane-ye geri ver-di.
Ay§e a book read-past, then library-dat back give-past
'Ayse read a book;, then returned (it;) to the library.'
Another example is seen below in (37) and (38). When the first sentence has a bare
object, what is recommended is interpreted as the action (book-reading versus swimming,
for example). When the object has the numeral bir, what is recommended is the object,
namely the book, only. In other words, when the object is in the bare form, it behaves like
one single unit with the verb. When the object has a numeral, it behaves as an
independent unit.
27
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(37) Ay§e tatilde [bir kitap]; oku-du, o;-nu herkes-e tavsiye et-ti.
Ay§e break-loc a book read-past, it-acc everybody-dat recommend-past
'Ayse read a book during the break and recommended it to everybody'
(38) Ay§e tatil-de [kitap oku];-du, o;-nu herkes-e tavsiye et-ti.
Ay§e break-loc book read-past, it-acc everybody-dat recommend-past
'Ayse read a book during the break and recommended it to everybody'
The contrast between the bare objects and bir N objects in pronominalization further
supports the idea that bare objects cannot have a referential reading. Because they are
non-referential, they cannot be replaced by pronouns. However, as it is the case in the
adjacency and modificaton, as well, the picture is not all that clear. Although
pronominalization is not possible, the wh-word ‘what’ can substitute a bare object as in
(39). Contrast it with light verb constructions such as telefon et- ‘phone’ exemplified in
(40). In (40), the question would be grammatical with Ay§e neyapiyor? ‘What’s Ay§e
doing?’ which would substitute the object+verb as one unit.
(39) Q: Ay§e ne oku-yor?
Ay§e what read-prog
‘What’s Ay§e reading?’
A: Kitap
‘book’
28
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(40) Q: *Ay§e ne ed-iyor?
Ay§e what do-prog
‘what’s Ay§e doing?
A: *telefon
‘telephone’
In summary, bare objects cannot be pronominalized. The wh-word ‘what’ is the only
exception. It is important to note, however, that in the case of ‘what,’ the object does not
have to be referential. The referential or the specific ‘what’ would be case marked: ne-yi
‘what-acc.’ As will be discussed in the next chapter, when the non-case marked form and
the accusative marked form are both possible, the non-case marked form obligatorily has
a non-specific interpretation and contrasts with the specific readings.
2.2.1.4 Comparison with nominal compounds
Bare object + verb structures share their properties that were discussed above with
nominal compounds such as bebek arabasi ‘baby carriage.’ In nominal compounds, too,
the non-head nominal must be adjacent to the verb (41), it cannot be post-posed (44) or
pronominalized (45).
(41) *Bebek [kutpiik] araba-si
Baby small car-cm
29
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(42) *Bebek mi araba-si
Baby que car-cm
(43) *Bebek bile araba-si
baby even car-cm
(44) *Araba-si bebek
car-cm baby
(45) *0 araba-si
It carriage.
As it is the case with the bare objects, the wh-word what can replace both the head and
the non-head nominal of the compound. Note that in (46), the wh-word carries the
compound marker; therefore, it is embedded into the compound.
(46) Bebek ne-yi?
Baby what-cm?
(47) Ne araba-si?
What car-cm?
30
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The same referentiality constraint holds for modification in compounds, too, that is, an
object relative clause cannot modify the non-head nominal alone (48), rather it modifies
the compound as a whole. However, as the example in (49), shows, the modification of
the non-head nominal is possible with subject relative clauses as long as the non-
referentiality is maintained8.
(48) [Ali-nin gdr-dtigu] bebek bakici-si.
[Ali-gen see-REL] baby sitter-cm.
a. ‘The baby sitter that Ali saw.’
b. *’The baby that Ali saw.’
(49) [Qok agla-yan] bebek bakici-si
[Very cry-REL] baby sitter-cm
a. (A) baby sitter that cries a lot (‘baby sitter’ is modifed)
b. (A) (Baby) sitter for babies who cry a lot (‘baby’ is modifed)
In summary, the behavior of the bare objects is similar to the behavior of non-head
nominals in nominal compounds.
8 See Goksel (1990) and Tun? (1997) for a discussion of so-called ‘bracketting paradoxes’ in Turkish.
31
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2.2.1.5 Comparison with light verb constructions
The bare objects are similar to the nominals in light verbs constructions that are formed
by et- and yap-. Nominals that occur with light verbs are non-referential. In those
constructions too, the nominals have to appear adjacent to light verbs (50) and (51),
cannot be pronominalized (52), adverbs/adjectives modify the object +verb complex,
rather than the nominal alone (53). In those rare examples, such as the one in (54), the
nominal can be modified independently. Remember that the bare objects can be modifed
by adjectives too, therefore the behavior of the nominals in examples such as (54) is
similar to the behavior of objects I discussed earlier.
(50) ? iitu bugiin yap-iyor-uz.
Iron today do-prog-IP
‘We are ironing today.’
(51) *Siz yap-iyor-sunuz iitu.
Iron do-prog-2P
‘You are ironing.’
(52) * 0 yap-iyor-sunuz.
it do-prog-2P
‘You are doing it.’
32
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(53) Hizli iitu yap-iyor-sunuz.
Fast iron do-prog-2P
‘You are ironing fast’
(54) Buharli iitu yap-iyor-sunuz.
Steam iron do-prog-2P
‘You are ironing (with a steam iron)’
Not very surprisingly, although the nominal cannot be pronominalized, it can be replaced
by ‘what’ in a question and the answer can be titii ‘iron’ alone9.
(55) Q: Ne yap-iyor-sunuz?
what do-prog-2P
‘What are you doing?’
A: iitu
‘Iron’
The other light verb et- behaves like yap in terms of its adjacency. As seen in (56) and
(57), the nominal should be adjacent to the verb. It can be pronominalized (58) and
9 If this question is asked at a factory where irons are manufactured and where the nominal can refer to the
actual physical object, it can have a referential interpretation. This is not the intended context / meaning
here.
33
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
preceding adjectives modify the N+V complex rather than the nominal alone (59). In
contrast to the bare objects andyap-constructions, though, an adjective cannot modify the
nominal (Contrast (34) and (54) with (60)). However they differ in the question formation
with ‘what’ as it is not possible to replace the nominal part with ‘what’ alone and the
answer cannot be ‘telephone’ only. All these properties show that bare objects are similar
to the bare nouns in light verb constructions as well.
(56) *Telefon bugtin ed-iyor-uz.
telephone today do-prog-2P
‘We are calling today’
(57) *Siz ed-iyor-sunuz telefon.
telephone do-prog-2P
(58) *0 ed-iyor-sunuz.
it do-prog-2P
(59) Hizli telefon ed-iyor-sunuz.
Fast telephone do-prog-2S
‘You are calling fast.’
(60) * Jetonlu telefon ed-iyor-sunuz.
Coin telephone do-prog-2S
34
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(61) Q: *Ne ed-iyor-sunuz?
what do-prog-2P
‘What are you doing?’
A: *telefon
The focus particles can appear between the nominal and the light verbs just as they do in
the case of the bare objects as seen in the examples (62), (63) and (64) below.
(62) Utu mu yap-iyor-sunuz?
Iron Q do-prog-2P
‘Are you ironing (=doing iron)’
(63) Util de yap-iyor-sunuz.
Iron too do-prog-2P
‘You are ironing too’.
(64) Utu bile yap-iyor-sunuz.
Iron even do-prog-2P
'You are even ironing.'
35
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In summary, the behavior of the bare objects is similar to the behavior of nominals in
light verb constructions. Just as the nominals in light verbs, the bare objects are non-
referential.
2.2.2 Aspectual Properties
In Aydemir (2004), the aspectual properties of verbs with bare objects and bir N objects
are compared and it is proposed that bare objects appear in events with atelic
interpretation, whereas bir N objects are parts of telic events. According to Aydemir
(2004), bare objects are incorporated to verbs and bir N objects are arguments and thus
do not undergo incorporation. This difference in telicity is presented as evidence for the
incorporation versus argumenthood analysis. Although I agree with Aydemir (2004) that
bare objects and bir N objects are structurally different, in this section I argue that the
telicity contrast is not as clear as it is described in Aydemir (2004).
The atelic interpretation of bare object structures is obvious and their interpretation can
be attributed to their non-argument status. What is not clear is the reason why bir N
objects appear in telic events. In other words, what is not clear, is whether or not telicity
can be a good test for the argumenthood of bir N objects (without case marking), as they
behave more like bare objects, as far as the intended meaning is concerned. Note the clear
contrast between the bare objects and case-marked objects, exemplified in (65)a and
36
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(65)b1 0 . The only reason why bir N objects can result in telic reading is the cardinal
interpretation of bir. So, in the example in (65)c below, the interpretation is Ayse read
one book in an hour (as opposed to two or more), rather than Ayse read a book in an
hour.
(65) a. *Ay§e bir saat-te kitap oku-du.
Ay§e an hour-loc book read-past
Ay§e in an hour book read-past.'
b. Ay§e bir saat-te bir kitab-i oku-du.
Ay§e an hour-loc a book-acc read-past
'Ay§e read a book/one of the books in an hour.'
c. Ay§e bir saat-te bir kitap oku-du.
Ay§e in an hour-loc a book read-past
‘*Ay§e read a book in an hour.’ / ‘Ayse read one book in an hour.’
Therefore, the telicity interpretation is not a good test to distinguish bir N objects from
bare objects. Bir N objects are grammatical only when bir has a cardinal reading. It is
1 0 Aydemir (2004) compares bir N objects with bare objects only and her conclusion is based on this
comparison. She does not discuss accusative marked objects. I included them in the discussion to show the
contrast between accusative marked objects and bir N objects (without accusative).
37
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ungrammatical without a cardinal reading, just like it is the case with bare objects.
However, the behavior of the bare objects in such constructions exemplified above
supports an analysis that treats them as nominals that are dependent upon the verbs.
2.2.3 Are bare objects incorporated?
As already mentioned in the introduction part of this section, bare objects are analyzed as
head incorporation cases and compounds in the literature. In this section, based on what I
presented above, I argue against the incorporation or compounding analyses.
According to Mithun (1984:873), Turkish non-case marked objects exhibit properties of
lexical compounding (her Type-1 Noun Incorporation) similar to those in relatively
analytic Oceanic or Mayan languages. In such compounding, in the languages she
analyzes, the noun does not have definiteness or number features, but rather qualifies the
verb (as in berry-picking in English). She reports that no constituent can intervene
between the object and the verb although the noun and the verb behave as separate
phonological words and written as such in orthography.
In such languages, the noun loses its argument status. In Oceanic languages for example,
Verb-Noun complex behaves like an intransitive verb. The transitivity is detected through
the ergative versus nominative case marking in Tongan or Mayan languages, for
example, and the noun-verb complexes behave like intransitives. In Tongan, the form of
38
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the relative clause is also determined by the transitivity of the verb, and the relative
clause formation signals the incorporation of the object as well (Mithun, 1984:851).
After incorporation, the noun and verb behaves as a single unit. In Samoan, for example,
the particles, which cliticize to the right of the verbs, cliticize to the right of the
Verb+Noun complex after incorporation (Mithun, 1984:850). A similar behavior is seen
in the suffixes that follow verbs in Micronesian languages, and adverb placement in
Kusaien sentences. In such languages, in some cases the meaning of the compounds is
not the equivalent to the meanings of the noun and the verb. In Kusaien, for example,
srasre ‘raise’ + po ‘hand’ means ‘surrender’. Such examples further support the lexical
compounding analysis.
Mithun (1984) does not discuss the properties of Turkish in detail, but based on the
adjacency properties, she includes it in her Type-1, lexical compounding category
together with the languages mentioned above. She does not talk about the focus elements
that can intervene between the bare object or and the verb. She does not talk about other
properties either. Based on what I discussed in the sections above, it is clear that that the
type of operation that bare nouns undergo, is different from the types of languages
Mithun analyzes. Bare objects are not so bare in Turkish as they can be modified, they
can appear non-adjacent to verbs when they interact with focus elements, the meaning of
the noun and the verb does not change.
39
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Adopting a similar approach, Knecht (1986) also proposes that bare object and verb
constructions are compounds. In addition to the properties reported above (adjacency,
adverb placement and adverb scope), she evaluates the stress pattern of the noun-verb
sequences. She states that in the case of the bare nouns (as opposed to the case marked
constituents) the stress always falls on the bare noun and this can be further evidence for
compounding. However, the immediately preverbal position is the neutral focus position
in Turkish and the placement of the stress is most probably due to the focus and it can be
shifted to other constituents. As seen in the following examples, stress can be placed on
the verb, when the focus particles such as the question particle and the adverb bile ‘even’
follows the verb ((6 6 ) and (67)).
(6 6 ) Ay§e kitap oku-mu§ mu?
Ay§e book read-perf Q
‘Did Ay§e read (a) book / do book-reading?’
(67) Ay§e kitap oku-mu§ bile.
Ay§e book read-perf even
‘Ay§e even book-read.’
(6 8 ) Kim kitap oku-mu§?
Who book read-past.
‘Who book-read?’
40
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(69) Ay§e kitap oku-ma-mi§.
Ay§e book read-neg-perf
‘Ay§e didn’t book-read.’
These focal particles require the focal stress to be on the preceding word. This supports
an analysis that treats them as single complex (but not necessarily a single compound, as
focal scope can extend over multiple lexical units). In addition, when noun-verb complex
occurs with a wh-word such as ‘who,’ the stress obligatorily falls on the question word
(6 8 ). In negated sentences, too, the neutral position for the stress is the syllable preceding
the negative morpheme. Therefore, it is not true that the objects always bear stress.
Rather, the stress can be shifted to the other constituents and such a stress shift supports
that the stress falls onto the noun because it is in a focus position, not because it is a part
of the compound. Moreover, if the nominal and the verb formed a single unit, the focal
stress would shift to the left of the object.
Nilsson (1986) also analyzes the bare object+ verb constructions as noun incorporations
(70). She acknowledges that the so-called noun incorporation phenomena in Turkish is
different from the noun incorporations in Eskimo languages discussed in Sadock (1980).
She proposes a head incorporation model, which is very similar to Baker’s (1988). She
also states that the incorporation takes place in the lexicon but does not provide any
convincing argument to support her proposal.
41
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(70) Nilsson’s (1986) incorporation
VP
v°
/ \
V° N°
More recently, Komfilt (2003), adopting Baker’s head incorporation model, proposes a
similar analysis for Turkish bare nouns. According to the original head incorporation
analysis of Baker (1988), a nominal head that is in the c-command domain of a verb, can
be incorporated to the verb in such a way that the two heads become one predicate. In
Eskimo languages, after the incorporation N+V complex has intransitive morphology. In
other languages, the verb remains transitive and can assign case to another direct object.
(71) Baker’s (1988) incorporation proposal.
Before incorporation After incorporation
VP VP
Komfilt (1997, 2003), proposes an Bakerian type incorporation analysis for Turkish, and
at the same time assumes that bare objects have a DP projection and they are further
embedded in a Case Phrase (KP), where the K-head is filled with a morphological case
marker in the case of the case marked objects. In bare objects, both the head of DP and
the head of KP are not filled. The N-head, then, moves up to the head of DP and then to
42
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the head of KP and gets incorporated to the verb. In those cases where the DP and KP are
filled, the object cannot move any further and cannot get incorporated. Therefore the
structures before and after the incorporation are as follows.
(72) Komfilt’s (2003) incorporation proposal.
Before incorporation After incorporation
VP VP
[-specific] tj
Komfilt (2003) does not state why she needs the KP and DP projections in the case of the
bare objects i.e., why she does not incorporate the objects directly into verbs. Besides
this, her analysis, as well as any incorporation analysis, runs into problems, as Komfilt
herself acknowledges, because of the properties that I listed above: Bare objects are not
always adjacent to verbs. Focus particles can intervene between the object and the verb.
Even if this can be attributed to the unusual properties of focus particles in Turkish, the
43
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
head incorporation analysis is still problematic because of the modification properties of
nouns. What is incorporated in Turkish has to be more than a bare noun and thus, it is no
longer a head incorporation in the sense of Baker (1988).
One of the strongest evidence against syntactic incorporation, which is also
acknowledged by Nilsson (1986), comes from causative formations in Turkish. Note in
(73) that when a transitive verb is marked with a causative marker and becomes a
ditransitive verb, the agent of the transitive verb bears the dative case. When an
intransitive verb is causativized, however, the agent of the intransitive verb is marked
with the accusative case.
(73) Bebek bir elma-yi yi-yor.
Baby a/one apple-acc eat-prog.
‘The child is eating an apple.’
(74) Ay§e bir elma-yi bebeg-e ye-dir-iyor.
Ay§e a/one apple-acc baby-dat eat-caus-prog
‘Ay§e is making a child eat an apple.’
(75) Bebek agli-y or.
Baby cry-prog
'The baby is crying'
44
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(76) Ay§e bebeg-i/*-e agla-t-iyor.
Ay§e baby-acc/*-dat cry-caus-prog
‘Ay§e is making a child cry.’
(77) Bebek elma yi-yor.
Baby apple eat-prog
‘Baby is apple-eating.’
(78) Ay§e bebek-e/*-i elma ye-dir-iyor.
Ay§e baby-dat/*-acc apple eat-caus-prog
‘ Ay§e is making a child eat (an) apple’
Note the causativization of a verb that has a bare object. If the bare object were
incorporated to the verb, we would expect the verb+object complex to behave like an
intranstive verb, and thus the subject to appear as marked with accusative case. The
example shows that this is not borne out. The agent is marked with a dative case,
although the object is not marked with accusative. In typical compounding or
incorporation cases, the incorporation “reduces the valence of the verb by one and it
permits another argument of the clause to occupy the case role vacated by the NI.”
(Mithun, 1984:857). Such examples contrast with classical incorporation examples. In
West Greenlandic cases reported by van Geenhoven (1998), for example, verbs bear
intranstive verb morphology once the objects are incorporated. In the polysynthetic
languages, too, an incorporated object+verb complex behaves like an intransitive verb,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
can take another direct object (e.g., it is possible to book+read Anna Karanina in such
languages, whereas in Turkish, such cases are impossible)1 1 . These examples do not only
show that bare objects do not incorporate, they also show that bare objects behave as
syntactic arguments1 2 . An analysis that proposes an argumenthood contrast between the
two non-case-marked object types fails to capture these facts.
Komfilt (2003) attributes the behavior of bare objects in causative structures to the
peculiarities of the causative formation in Turkish. She states that even in lexicalized light
verb constructions such as dua et- ‘pray,’ the agent of the verb ‘pray’ is not marked with
the accusative case as seen in (79) ((40) in Komfilt, 2003). Based on such examples she
suggests that the case assignment in causative constructions should be treated differently
because even a predicate noun such as dua ‘prayer’ in (79) can block the accusative case
assignment. She does not propose a detailed analysis for the causative structures and the
noun+ef constructions as she does not consider it an important problem for her head
incorporation analysis.
1 1 It is not possible with verbs such as ‘read’ and coplex predicates such as ‘book-reading.’ However the
language allows such structures when the verb is yap- or et-, which can be considered light verbs in
Turkish. So you can say hastalari tedavi et ‘ treat patients ’ (make-treatment patients), kitaplan paketyap-
‘wrap books (literally do+parcel books)’. See the following discussion for details.
1 2 Another test for argumenthood and transitivity would be passivisation, but intransitive verbs, as well as
transitives, can be passivized in Turkish, therefore it is not a good test.
46
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(79) Hasan Ali-ye/-*yi dua et-tir-di.
Hasan Ali-dat/*-acc prayer do-caus-past
‘Hasan made Ali pray.’
Ozturk (2004), who proposes a pseudo incorporation analysis (a la Massam 2001) for
non-case marked objects in Turkish, suggests another account for the case assignments in
causatives. Following the analyses of Kuroda (1988), Miyagawa (1991) and Marantz
(1991) she treats case assignment as independent from case morphology. Adopting
Harley’s (1995) mechanical case parameter, she suggests that the realization of the dative
is due to a parametric case morphology realization mechanism in which if there are three
case features checked structurally in a clause, the second case that is realized after
nominative is dative. Accusative case is realized as the third case. That is why the dative
case is realized after the nominative and the accusative case is realized only in the third
place, if there is a third nominal.
Even if there were such a case realization mechanism in Turkish, it does not seem to
apply to all the object-verb formations. Some object+verb complexes do take accusative
marked objects. See examples (80), (81) and (82).
(80) Ay§e hasta-lar-i muayene et-ti.
Ay§e patient-plu-acc examination do-past
‘Ay§e examined the patients.’
47
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(81) Ay§e eti yahni yap-ti.
Ay§e meat-acc stew do-past
‘Ay§e made the meat stew.’
(82) Ay§e kitap-lar-i paket yap-ti.
Ay§e book-plu-acc parcel do-past
‘ Ay§e wrapped the books’
Nilsson gives another example for the accusative realization where the embedded,
nominalized sentence bears the accusative case (83). The same verb+object can be used
with a simplex object too as seen in (84).
(83) Ne zaman gel-eceg-im-i siz-e haber ver-eceg-im.
What time come-poss&lS-acc you-dat news give-fut-lS
‘I’ll tell you when I’m coming.’
(84) Toplanti-yi siz-e haber ver-eceg-im.
Meeting-acc you-dat news give-fut-lS
‘I ’ll inform you about the meeting.’
The object+verb constructions in these structures contrast with the objects described
above and the examples show that Turkish does allow accusative-marked objects with
some noun+verb complexes. Therefore, the non-realization of the accusative case cannot
48
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
be because of a mechanism that restricts accusative realization in Turkish.The objects in
these structures have the same properties that were mentioned above.
Yet another problem for an incorporation proposal is that in such constructions the verb
can undergo ellipsis by itself, without the object nominal. If they were one single lexical
unit, as has been proposed, they would be expected to undergo ellipsis as one single unit.
The sentence in (85) is an example where the verb read has undergone ellipsis.
(85) Ay§e butiin gun kitap oku-du, Mehmet de gazete 0
Ay§e all day book read-past, Mehmet TOP newspaper
'Ay§e read (a) book all day today, and Mehmet (a) newspaper.1
2.2.4 Complex Predicate Formation
I suggest that the properties discussed above show that the object + verb combinations do
not qualify for noun-incorporation in Turkish. Bare objects are NPs, they are not bare N
heads, so they cannot incorporate through a head incorporation model. Bare objects do
not have NumP or DP projections and the absence of number and definiteness
interpretation comes from the absence of these projections. They remain within VP and
do not interact with quantifiers and negation. Such a loose object+verb formation is
proposed by Miner (1982,1984) for Amerindian languages under the name of ‘noun
stripping’. In noun stripping, objects are very closely related to the verbs but they do not
incorporate. As he suggests there seems to be a parametric variation among languages
49
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
based on the degree of structural dependency between complements and verbs. Turkish is
definitely on the ‘loose end’ of this parametric variation where the objects do not
incorporate but share some properties with incorporated objects1 3 . In this study, I adopt
the structure and the complex predicate definition which I originally proposed in Ketrez
(2004b).
(8 6 ) Complex predicate
A verb x and a nominal y form a complex predicate if (i) y is not a DP, (ii)
every maximal projection z that dominates x dominates y, (iii) x locally c-
commands y.
(87)
VP
In the present analysis, the properties of the bare objects are attributed to their complex
predicate formation and the absence of the NumP and DP projections. What is crucial for
the discussion here is that they are in a position lower than the lowest possible position
1 3 Orgun and Inkelas (2004) propose that Miner’s Noun Stripping analysis can account for the behavior of
bare objects in Turkish.
50
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
that an adverb and negation can occur. They do not incorporate to the verbs but they
appear in a complex predicate formation as described above and thus they are unable to
interact with other constituents. They are not assigned case because they lack a maximal
projection that can carry case.
In summary, in this section I presented the properties of bare objects in Turkish and
discussed various accounts that treated them as incorporations. I argued that their
properties do not qualify them for being incorporated heads. Rather, they behave like
independent constituents that have mobility restrictions due to the absence of DP and
NumP projections. They remain within VP, and do not interact with negation or other
quantifiers in terms of scope. For the present study, the most important property of bare
objects is that they always have narrow scope with respect to negation and adverbs that
are adjuncts above the VP. The scope properties of bare objects will be discussed in more
detail in Section 2.4.
2.3 The DP projection and the syntax of bir
In this section, I discuss the properties of bir N objects and contrast them with bare
objects. In the sections above, I have already presented some properties of bir N objects. I
showed that, in contrast to the bare objects, they can be modifed by relative clauses, they
can be pronominalized and pro-dropped. In the first section below, I discuss them in more
detail and I propose that bir N objects have NumP and DP projections and they are
referential. In the second section, I discuss the accounts that treat them alike and show
51
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
that if we assign the same syntactic structure to them, we fail to capture all their
properties.
2.3.1 DP Projection
In this section I propose that bir is not an indefinite article and although Turkish does not
have determiners, it still has a maximal projection that functions as a DP in other
languages (Abney, 1987, Ritter, 1991 among others). For the shake of simplicity I call
this projection DP.
As already described above in Section 2.1, Turkish does not have a definite article.
Definite nominals (including proper names, pronouns and wh-question words except
what) obligatorily bear overt accusative case in the object position. Subject nominals are
ambiguous with respect to definiteness1 4 .
Bir is a candidate for an indefinite article, however it cannot be the head or the spec of
DP for the following reasons: Turkish is a strictly head-final language and bir appears to
the left of the nouns, which is not a head position. If it were the head or the spec of DP,
we would expect it to appear at the edge of the phrase (see Fukui (1995) for a similar
discussion for Japanese). Rather bir appears between the modifiers (adjectives and
1 4 Stress pattern can distinguish referential versus non-referential nominals in subject position. I will not
discuss this, because I would like to concentrate on object nominals only.
52
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
relative clauses as well as the genitive marked possessors) and the noun. In (8 8 ), the
relative clause modifier and the possessor Ay§e-nin occur to the left of bir.
(8 8 ) Ay§e-nin [park-ta kaybol-an] bir kedi-si
Ay§e-gen park-loc lose-rel a cat-poss
‘A cat of Ay§e [that was lost in the park] ’
Under the present analysis, the head of DP is occupied by an abstract D. I analyze the
overt case morphology as a realization of (and an evidence for) a DP projection1516.
Tateishi (1983) proposes a similar account for Japanese. Whenever DP projects, the
nominal is referential and we also have [± specific] and [± definite] features. In the
absence of the DP projection, we do not have these features. Similarly, the presence of
NumP results in a number interpretation of the nominal. Bir N nominals have NumPs,
thus they have a number interpretation, which is [-plural].
1 5 That does not necessarily imply that only those nominals that are overtly case marked have DP
projection.Case morphology can be an evidence for a DP projection, but the absence of it does not imply
the lack of this projection. Nominals, those that appear in the subject position, for example, can appear in
bare form and still behave like DPs.
1 6 Oztiirk (2004) proposes another analysis for Turkish phrase structure. Based on the properties stated
above, she rules out the possibility of a DP projection in Turkish. However, the language specific properties
mentioned above only shows that bir or the demonstratives do not qualify as DP heads in Turkish. The
absence of a D head in the sense of the D heads in other languages does not necessarily show that Turkish
lacks such a projection which would carry the definiteness and specificity features.
53
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Bir N nominals can be pronominalized, which is an evidence that they are not bare
NumPs, but rather they have a DP projection. They can go out of VP, but cannot
scramble to a higher position due to the absence of case morphology.
2.3.2 Bir and bir N objects
I analyze bir as a numeral modifier that occurs in [spec, NumP]. It is homophonic with
the numeral “one” and appears in the same position as other numerals such as cardinal
numbers, exemplified in (89) and (90). It marks singularity because of its cardinal
number property. It is not different from other cardinal numbers in the sense that it marks
count nouns. In Aygen (1999), bir is assumed to be a numeral (cardinal number) but no
structure is proposed. Tura (1973) analyzes bir as an instance of number-marking. She
proposes that bir indicates that the indefinite phrase is singular, that it will introduce new
individuals into discourse.
(89) Ay§e-nin U 9 §arki-si.
Ay§e-gen three song-poss
‘Three songs of Ay§e.’
(90) Ay§e-nin bir §arki-si.
Ay§e-gen a/one song-poss
‘One/a song of Ay§e.’
54
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In Dede (1986) and Kiintay (2002) bir is analyzed as “an optional marker of
indefmiteness,” in the sense that indefinite objects can have bir as in the case of bir N
objects. In a more recent study, Oztiirk (2004) defends the same idea.
As I have already shown in the sections above, optionality is not the right term to
describe the occurrence of bir because it does not have a free variation pattern. As
discussed in the earlier sections, it is more plausible to accept bare nominals as non-
referential nominals that lack any number or definiteness specification. Bir N nominals
however do have a number specification (i.e., they are singular).
In Oztiirk (2004), both bir N objects and bare objects are non-referential. What marks
referentality is the accusative case. When the accusative case is absent, the nominals are
non-referential and they pseudo-incorporate to verbs, as shown in (91). Bare objects and
bir N objects share some properties such as adjaceny, but the reason why Oztiirk (2004)
treats them alike is their scopal properties.
( 9 1 ) Ali [complex predicate bir kitap oku-du].
Ali one book read-past
'Ali read a book'
She shows that both bare objects and bir N objects take narrow scope in contrast to
indefinite objects (she refers to the accusative marked indefinite objects only) that can
take both narrow and wide scope. The examples are in (92), (93) and (94). Therefore, she
55
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
concludes that only the accusative marked indefinites are ‘real’ indefinites in Turkish,
while the non-case marked ones appear in complex predicates and cannot interact with
other quantifiers.
(92) Her 9 0 cuk bir kitab-i oku-du. every > bir, bir > every
Every child one book-acc read-past
‘Every child read a book.’
(93) Her t^ocuk bir kitap oku-du. every > bir, *bir > every
Every child one book read-past
‘Every child read a book.’
(94) Her gocuk kitap oku-du. every> bir, *bir>every
Every child book read-past
‘Every child book-read.’
However, unability to take wide scope can be attributed to various causes. The reason
why both object types take narrow scope with respect to the QNP Subject can be due to
different reasons. One may take narrow scope, because it is in a complex predicate
structure with the verb, and the other may be in a narrow scope position, which is still
56
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
lower than the QNP subject, but not neccessarily at the same level or below VP.
Therefore scope similarity does not neccessarily show that they have the same structure1 7 .
Oztiirk (2004) also proposes a different account for the cardinal (stressed) reading of bir
(BIR) and the indefinite (non-stressed) counterpart bir. She shows that BIR and bir have
different ditributional properties with respect to adjectives. The unstressed bir follows
adjectives (95) and it denotes a singular indefinite book, whereas BIR, the stressed
version, precedes adjectives (96) and it denotes quantity. The distribution of BIR is
similar to that of other numerals (97).
(95) Ali kirmizi bir kitap al-di.
Ali red one book bought
‘Ali bought a red book (an indefinite, arbitrary book).’
(96) Ali BIR/ *bir kirmizi kitap al-di1 8 .
Ali one / a red book took
‘Ali bought two red books.’
1 7 Actually, the experiment results that will be discussed in the following chapters will show that bir N
objects can take wide scope and the wide scope taking property in the absence o f the case morophology
will be attributed to the contrast between bare objects and bir N objects.
1 8 Grammaticality judgments o f the examples belong to the original study.
57
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(97) Ali iki kirmizi kitap al-di.
Ali two red book bought
‘Ali bought two red books.’
Bir N and BIR N exhibit distributional differences. The example in (98) shows that,
where BIR is stressed, it is possible to strand it and delete VP. In (99) we see that a
similar kind o f deletion is not allowed with the unstressed bir. In (100), we see that when
the object is case marked it is not possible to strand the numeral. Therefore she argues
that BIR is not a modifier or a part of the pseudo-incorporated NP, but rather it is
introduced externally.
(98) Ali iki kirmizi kitap oku-du, ben de BIR [kirmizi kitap oku du m]
Ali two red book read, I too one [red book read]
‘Ali read two books, and I read one.’
(99) Ali kirmizi bir kitap oku-du, *ben de ye§il bir [kitap oku-du m]
Ali red one book read, I too green one [red book read]
‘Ali read a red book, and I read a green one.’
(100) Ali iki kitab-i oku-du, ben de bir [kitabi oku du m]
Ali two book-acc read, I too one [book-acc read]
‘Ali read two of the books, and I read one.’
58
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(101) Ali [BIR [compiexPredicate [kirmizi kitap al-di]]
Ali one red book bought
‘Ali bought one red book.’
Although it is true that BIR and bir have different properties in Turkish, the differences
can be accounted for without assigning them completely different structures. First,
although the stressed BIR can occur in a position preceeding the adjective, it is also
possible to have it in the post-adjectival position with a contrastive stress. Such a
possibility is problematic for an account that analyzes it outside the pseudo-incorporation
unit, because it will, then, be breaking into the incorporation unit. Besides, it is also
possible to have a non-stressed counterpart in the pre-adjectival position, especially in
fairy tales and narrations. So the linear ordering contrast is not that clear for BIR and bir.
Moreover, if BIR is an adverb outside of the pseudo-incorporated unit, what does it scope
over? If it is an adverbial modifier, its scope should be the whole VP. The fact that it
modifies only the nominal book rather than the noun+verb complex as a whole sheds
doubt on its adverbial property. The structures that have BIR has the reading ‘Ali read
ONE book’ rather than ‘Ali book-read ONCE.’ This scope-taking property shows that the
nominal and the verb are two distinct units and BIR scopes over the nominal only. It is
possible to have BIR with intransitive verbs. In such structures it has the meaning
‘ONLY’ or ‘ONCE’ but appearantly in the structures described above, it does not have
such an interpretation.
59
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In addition to these, the account that treats bare nouns and bir N nouns alike under a
pseudo-incorporation analysis, cannot explain why the structures that have bir have an
indefinite reading, while the bare objects do not. Another problem that has not been
addressed is the difference between denoting ‘singularity’ (in the case of unstressed bir)
and ‘quantity’ (without singularity, in the case of BIR).
2.3.3 Bir N objects, referentiality and DP projection
As discussed in Section 1, in the present study, I analyze bare objects as non-referential
Bir N objects contrast with them as it is possible to have referential nominals with bir N
structure1 9 . As discussed above bir N objects can be pronominalized and thus they have a
DP projection. As also discussed above, they can be modified by subject relative clauses.
In such relative clause examples, there is a difference between the interpretations of the
nominals. See the examples below. The difference between (102) and (103), where we
have the bare nominals is that with bir the speaker is looking for a single secretary,
whereas in (103) the number of secretaries is not specified.
1 9 In Schroeder (2000), too, bir is analyzed as a referentiality marker similar to the Bulgarian indefinite
article. According to Nilsson (1985), too, it is a marker that distinguishes referential and non-referential
nominals. Her analysis o f the indefinite article is similar to the accusative marker. She states that the
indefinite quantifiers establish a greater independence of the noun vis-a-vis the predicate (Nilsson,
1985:47). Bir N nominals and accusative marked objets contrast with bare N objects, which undergo
incorporation in her analysis. In a more recent study Aydemir (2004), proposes a similar account. She states
that bir N nominals are arguments, whereas bare Ns are not, they are incorporated to verbs. Her argument is
based on the properties that were discussed above, namely modification, adjacency, pronominalization and
aspectual properties o f the object+verb complexes
60
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(102) Fransizca bil-en bir sekreter ari-yor-um [a single secretary]
French know-REL bir secretary look+for-prog-lS
‘I am looking for a secretary who can speak French.’
(103) Fransizca bilen sekreter an-yor-um. [any number of secretaries]
French know-REL secretary look+for-prog-lS
‘I am looking for a secretary/ secretaries who can speak French.’
In the next chapter, where I discuss the differences between overtly case marked and non
case marked objects, I present a detailed account of the structure and interpretation of the
bir N objects. I argue that bir N objects are assigned structural case and they move to the
AgrOP which is immediately above VP. They do not bear an accusative case morpheme,
because their case is deleted under adjacency to the verbs.
2.3.4 A summary
Bir N nominals and bare N nominals are different because bir N nominals have DP and
NumP projections and due to these projections they have definiteness and number
specifications, whereas bare Ns lack these projections. Bir N nominals can be referential,
while bare N nominals are always non-referential. They both appear adjacent to verbs but
this does not necessarily mean that they have the same structure. They both lack overt
case morphology. In the case of the bare objects case is not assigned. In contrast, the bir
N objects are assigned case, but their case is deleted under adjacency. Both object types
61
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
have narrow scope with respect to negation and adverbs due to their position which is
below negation and adverbs.
2.4 Scope of non-case marked objects
In this section I show that both bir N objects and bare objects take narrow scope with
respect to negation and adverbs.
2.4.1 Negation
In (104), I present an example of bir N object and a bare object that interact with
negation. In both sentences the object has narrow scope with respect to negation and the
sentences read as ‘Ay§e did not eat any apples. ’
(104) a. Ay§e bir elma ye-me-di. neg > bir elma
Ay§e an apple eat-past
‘Ay§e did not eat an apple.’
b. Ay§e elma ye-me-di. neg > elma
Ay§e apple eat-past
‘ Ay§e did not eat an apple.’
In the discussion above, I proposed that the scope taking properties of these objects are
related to their structures and positions with respect to negation. In the case of the bare
62
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
objects, the object remains within VP and the negation, which is above VP c-commands
and scopes over the object (105). The object cannot move to any position above negation
or outside of VP due to the lack of NumP and DP. In the case of the bir N objects, bir N,
which has a DP projection and which is assigned structural case moves to AGRoP, which
is immediately above VP (105). It does not move any higher, it remains there and
undergoes case deletion. Note that bir N object is still below negation. Negation c-
commands and scopes over bir N object.
(105) a. [ [ D P subject...[Negp[AgrOpbir N [V p...bm NV-neg]...] V-neg ]]]]
b. [ [ D P subject • • • [NegP [AgrOP [vp. .. N V neg] ...] V-neg ]]]]
In summary, although the bir N objects and bare objects do not have the same structures
and occupy the same positions in the clause structure, they have the same scope-taking
property, which is dependent upon their position with respect to negation.
2.4.2 Adverbs
The same kind of relationship is seen in the interaction of objects with the adverb twice.
In (106) a. and b. the objects interact with the adverb twice and have narrow scope with
respect to the adverb.
63
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(106) a.Ay§e iki kere bir sandalye it-ti twice > indef
Ay§e twice a chair push-past
‘Ay§e pushed a chair twice.’ (any chair)
b.Ay§e iki kere sandalye it-ti twice > indef
Ay§e twice chair push-past
‘Ay§e pushed a chair twice.’ (any chair)
I attribute their scopal behavior to the same kind of positional property. Both object types
are lower than the adverb twice and have narrow scope due to their positions, which are
shown in (107).
( 1 0 7 ) a. [ [ D P subjec t ...[ b W c e ... [AgrOpbir N [Vp . . . ¥]...] ]]]]
b. [ [ D P s u b je c t • • • [ tw ice... [A g rop [v p ... N ¥ ] . . . ] ] ] ] ]
2.5 Chapter summary
In this chapter, I discussed the properties of the non-case marked objects. I argued that
bare objects and bir N objects are structurally different. Bare objects lack NumP and DP
projections and they remain in VP projection. They are not assigned case. The bir N
objects have both projections and move out of VP and are assigned case. Their case is
deleted under adjaceny. I also argued that they have narrow scope with respect to
negation and adverbs because they are both in positions that are lower than the negation
64
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
and adverb positions. In the next chapter, I will discuss the properties of overtly
marked objects and contrast them with non-case marked objects.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3 Accusative-marked indefinite objects
In Chapter 2 ,1 presented non-case marked objects and discussed their syntactic
properties. In this chapter, I present accusative marked indefinite objects with a particular
focus on their scope properties and contrast them with their non-case marked
counterparts.
Examples of accusative marked objects are repeated here in (108). In each example the
object occurs in a different surface position with respect to the verb (preverbal adjacent to
the verb, preverbal non-adjacent to the verb, and postverbal position) but has the same
interpretation as far as the scope relationship is concerned i.e., it has wide scope with
20
respect to negation .
(108) a. Ay§e bir e lm a-y i ye-me-di. [D P subj bir N-acc V-neg]
Ay§e an apple-acc eat-past
‘Ay§e did not eat a particular apple.’
2 0 Sentential stress falls on the verb in each case, i.e., neither the object, nor the numeral bir have focal or
contrastive stress. The role of stress placement on interpretation will be discussed later.
66
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
b. Bir elma-yi Ay§e ye-me-di. [bir N-acc D P subj V-neg]
an apple-acc Ay§e eat-past
‘Ay§e did not eat a particular apple.’
c. Ay§e ye-me-di bir elma-yi [D P subj V-neg bir N-acc]
Ay§e eat-past an apple-acc
‘ Ay§e did not eat a particular apple.’
Accusative marked objects exemplified above contrast with non-case marked objects
both in terms of their syntactic properties and their interpretation. They can scramble
freely as seen in the examples above, while the non-case marked objects that have a non
specific interpretation and narrow scope with respect to negation must be adjacent to
verbs. Accusative marked indefinites can be modified by relative clauses (109) and
adjectives (110) without any restriction in contrast with both types of non-case marked
objects. Similar to the bir N objects, but unlike the bare objects, they can be
pronominalized (111), and the verb can undergo ellipsis (112).
(109) a. Ay§e market-ten al-dig-i bir elma-yi Hasan-a ver-di.
Ay§e market-abl buy-REL an apple-acc Hasan-dat give-past
‘Ay§e gave Hasan.an apple-acc that she bought at the market.’
67
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
b. Ay§e yer-e dii§-en bir elma-yi Hasan-a ver-di.
Ay§e floor-dat fall-REL apple-acc Hasan-dat give-past
‘Ay§e gave Hasan an apple-acc that fell on the floor.’
(110) Ay§e ye§il bir elma-yi Hasan-a ver-di.
Ay§e green one apple-acc Hasan-dat give-past
‘Ay§e gave Hasan a green apple-acc.’
(111) Ay§e o-nu Hasan-a ver-di.
Ay§e it-acc Hasan-dat give-past
‘Ay§e gave it to Hasan.’
(112) Ay§e Hasan-a bir elma-yi ver-di, Ahmet-e bir armut-u.
Ay§e Hasan-dat one apple-acc give-past Ahmet-dat one pear-acc
‘Ay§e gave Hasan an apple-acc, Ahmet a pear-acc.’
In the literature, the accusative marked indefinites are analyzed as specific (Erguvanli,
1984, En?, 1991, Zidani-Eroglu, 1997, 2001) partitive (En<;, 1991), referential (Zidani-
Eroglu, 1997, Oztiirk, 2004) or presuppositional (Kelepir, 2001, 2003; Nakipoglu-
Demiralp, 2004) as opposed to non-specific, non-referential, non-partitive and non-
presuppositional indefinite objects that do not bear overt case morphology. While each
analysis acknowledges that the case marked and non-case marked objects are different
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
terms of their interpretation, their approaches to the actual interpretation of the accusative
case differ.
In this study, I propose that the accusative case is assigned formally and it is not
semantic, contrary to most of the literature. I argue that the accusative case is assigned
regardless of interpretation of the object and it can be deleted when the object is adjacent
to verbs, except for objects that bear possessive morphology. The adjacent position is the
presentational focus position and presupposed objects move to escape from the focus
position to one of the three positions: postverbal, preverbal, preceding the subject and
higher than AgrOP but lower than subject. The interpretation of the object does not come
from the case morpheme. Rather, the accusative case cannot get deleted when the object
has a particular interpretation and when it moves outside of the AgrOP.
In section 3.1,1 present a description and an overview of the accusative marked
indefinites together with the other kind of accusative-marked objects and discuss
examples. I show that accusative marked indefinites have wide scope with respect to
other constituents in the examples that I discuss. In Section 3 .2 ,1 present structures in
which accusative marked objects can have narrow scope with respect to other
constituents. I attribute their exceptional behavior to the occurrence of the accusative
morpheme due to formal requirements, and the inability of case to delete under particular
conditions, namely when the object has possessive morphology. I argue that in such
examples the morpho-phonological realization of the morpheme and its interpretation are
independent. Accusative case can be overt due to formal reasons and in such structures
69
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
does not necessarily have an impact on the interpretation or the scope of the object. In
such instances, it has to occur adjacent to the verb and bear the focal stress as it is not
different from the non-case marked objects in terms of its interpretation and restrictions
on scrambling. When it has a presuppositional interpretation it is displaced from a verb-
adjacent position. Through this discussion, I conclude that presuppositional objects
(rather than any case-marked object) can scramble in Turkish and thus scrambling is
dependent on presuppositionality, not the morpho-phonological realization of the
accusative case. Moreover, presuppositional objects are required to scramble so that they
escape from the focus position, where only non-presuppositional constituents can be
hosted. In Section 3 .3 ,1 present the structures that I adopt. Section 3.4 is devoted to the
postverbal objects. Section 3.5 discusses an alternative account and argues that the
analyses presented in the earlier sections can capture the behavior of the objects in a
better way. Section 3.6 summarizes the discussion and concludes the chapter.
3.1 Interpretation of the accusative marked objects
In Chapter 2 ,1 proposed that Turkish has a DP projection and the accusative case
morpheme is realized on an abstract D head. It is a phonological realization of the
feature but it can be deleted under adjacency. In the sections that follow, I discuss
interpretational properties of the accusative marked indefinite objects.
70
case
the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.1.1 Accusative-marked versus non-case-marked objects
According to the context of use, accusative marked objects have various interpretations.
In the examples above, for example, the apple that has not been eaten is ‘a particular’
apple or any member of an apple set that has been introduced in the previous discourse,
or ‘an apple’ that the speaker has in mind or s/he believes or presupposes that it exists.
Note that these interpretations can be different in terms of their truth conditions
depending on the context. The important property of the accusative marked objects for
the purpose of this study is that they contrast with the non-case marked objects
exemplified in (113) and (114) which do not need to be introduced prior to the utterance,
do not need to be a member of a previously mentioned set or the speakers do not need to
have a presupposition regarding their existence. The sentence in (113) means that Ayse
did not eat any apples and in the one in (114) it is possible that there are no apples
around.
(113) Ay§e bir elma / elma ye-me-di.
Ay§e an apple / apple eat-past
‘Ay§e did not eat (an) apple.’
(114) Ay§e bir elma / elma ye-mek isti-yor.
Ayse an apple / apple eat-inf want-prog
‘ Ay§e wants to eat (an) apple.'
71
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In addition to the indefinites, definites, proper names and pronouns bear an overt
accusative case marker in the language when they appear in a direct object position.
Examples are seen in (115)-(122). In all these examples the accusative marker is
obligatory and the object has a presupposed interpretation.
(115) Proper Noun
Ay§e Hasan-i/ *-0 gor-du
Ay§e Hasan-acc / *- 0acc see-past
‘Ay§e saw Hasan.’
(116) Definite noun
Ay§e elma-yi/*-0 ye-di
Ay§e apple-acc/ *0acc eat-past
‘Ay§e ate the apple.’
(117) Pronoun
Ay§e o-nu / *-0 oku-yor.
Ay§e it-acc/* 0acc read-prog
‘Ay§e is reading it.’
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(118) Demonstrative + Noun
Ay§e bu kitabi /*-0 oku-yor.
Ay§e this book-acc/*-0acc read-prog
‘Ay§e is reading this book.’
(119) Noun + Possessive
Ay§e Mehmet-in kitab-i-m / *-0 oku-yor.
Ay§e Mehmet-gen book-poss&3S-acc/*-0 read-prog
‘Ay§e is reading Mehmet’s book.’
(120) Generic
Kedi-ler-i /*-0 sev-er-im.
Cat-plu-acc / *-0acc like-aor-lS
‘I like cats.’
(121) Quantifiers
a. Ay§e biitun kitap-lar-i/*-0 oku-du.
Ay§e all book-plu-acc/-*0acc read-past
‘Ay§e is reading all the books.’
b. Ay§e her kitab-i / *-0 oku-du.
Ay§e every book-acc/-*0acc read-past
‘Ay§e is reading every book.’
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(122) Which N, Who
a. Ay§e hangi kitab-i / *-0 oku-yor?
Ay§e which book-acc /*-0acc read-prog
‘Which book is Ay§e reading?’
b. Ay§e kim-i / *-0 gor-du?
Ay§e who-acc/*-0acc see-past
‘Whom did Ayse see?’
Although proper names, as definite expressions, are assumed to require an accusative
case in the literature, there are exceptions. In those examples, where the accusative case
is not obligatory, the interpretation that it adds to the structure becomes more salient. As
seen in the example in (123), the accusative case is deleted when the object does not refer
to the actual object or the person, but rather it refers to any piece of music composed by
Mozart. When the same object bears the accusative case, it either refers to the person, in
this case Mozart himself, or it implies that the object or the person is a member of a set of
composers or CDs. Note that the deletion of the accusative case is not possible with other
proper names that cannot have a non-presupposed interpretation. In (124), for example, it
is obligatory.
(123) Ay§e Mozart(-i) dinl-iyor.
Ay§e Mozart(-acc) listening
‘ Ay§e is listening to Mozart’
74
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(124) Ay§e Mehmet-i / *Mehmet dinl-iyor.
Ay§e Mehmet-acc / Mehmet listening
‘Ay§e is listening to Mehmet.’
This interpretation o f the accusative case with proper names thus provides further
evidence that the accusative case appears on a presuppositional object and is not realized
on any proper name. A similar contrast is seen in the wh-word ‘what’ and the pronoun
birsey ‘something’ which can appear with or without an accusative marker. When they
bear the accusative case they refer to an object that the speaker has in mind. The ‘what’
question with the case marker can be uttered by a speaker who has a set of objects or a
particular object in his/her mind and who is asking to which one of these objects Ayse is
listening. When it is not case-marked it refers to any object, not necessarily a member of
a set. When Ayse is looking for something (without a case marker) to eat, it does not
matter what she eats, when the object is case-marked she has a particular thing in mind21.
(125) Ay§e ne(-yi) dinli-yor?
Ay§e what(-acc) listen-prog
‘What is Ay§e listening?’
2 1 Nilsson (1985), too, observes that what+acc is closer to the wh-word which in English. See her Chapter 4
for more examples.
75
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(126) Ay§e bir§ey(-i) yemek isti-yor?
Ay§e something(-acc) eat-inf want-prog
‘Ay§e wants to eat something.’
The interpretational differences between the case marked and non-case marked objects
are attributed to the ‘specificity’ in the sense of partitivity (e.g., En9, 1991) or
referentiality (e.g., Zidani-Eroglu, 1997), or ‘presuppositionality’ (e.g., Kelepir, 2001)
which is defined as ‘what the speaker has in mind.’ The following sections present a brief
discussion of each account. In the discussion below I show that this particular
interpretation is possible when the object can alternate with the non-case marked
counterpart and the nonpresuppositional interpretation is achieved through the absence of
the object case marking.
3.1.2 Partitivity and D-linking
In En? (1991) the notion of specificity is based on that of partitivity and d-linking in the
sense of Pesetsky (1987). In other words, specific objects are linked to a previous
discourse and the object is a member of a previously mentioned set. In (127) for
example, where the sentences in b and c follow the utterance in a, the girls in b sentence
are the members of the set of the children entering the room. The object with the
accusative case marking has this set membership interpretation. The object (the girls) in
c, without a case marking, refers to any girls, not necessarily the ones mentioned in a.
Because the sentence cannot be related to the previous discourse, it sounds odd.
76
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(127) a. Oda-m-a birka? ?ocuk gir-di
my-room-dat several children entered
'Several children entered my room.'
b. Iki kiz-i tam-yor-du-m
two girl-acc know-prog-past-IS
‘I knew two girls (from the set of children that entered my room)’
c. Iki kiz tam-yor-du-m.
Two girl know-prog-past-IS
'I knew two girls'.
Partitivity is only one of the interpretations that can be attributed to the accusative
marked indefinites. As the later work reveals accusative marked indefinites can be used
to refer to objects in non-partitive contexts. In the following sections I discuss such non
partitive examples.
77
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.1.3 Referentiality and out-of-the-blue indefinites
Zidani-Eroglu (1997) proposes that indefinites can have three different interpretations in
Turkish22. These types of indefinites are: 1- concealed partitives, 2- relational specifics,
3-out-of-the-blue indefinites. The concealed partitives are the partitives that Eng (1991)
introduces. The relational specifics are the ones that can be used with ‘a certain’ and the
out-of the-blue indefinites refer to referential objects that the speaker has in mind as
exemplified in (128), (129) and (130). In these examples, the objects are not members of
a previously mentioned set, so they are not partitive.
(128) Ali birkac belge-yi inceli-yor.
Ali several document-acc study-prog
‘Ali is studying several documents.’
(129) Ali dun bir kitab-i oku-du.
Ali yesterday one book-acc read-past
‘Ali read a book yesterday.’
(130) Ali uzun zamandir bir soru-yu dti§iin-uyor.
Ali long time one question-acc think-prog
‘Ali has been thinking about a question for a long tim e.’
2 2 Zidani-Eroglu (1997) does not propose three different syntactic structures for these indefinites and does
not state whether it is necessary to have three different structures.
78
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
She proposes that the indefinites that can be mentioned out-of the-blue are similar to the
referential indefinites that Fodor and Sag (1982) discuss. She suggests that all three types
of indefinites are accommodated in discourse; therefore they are specific just as the
partitive indefinites. In Zidani-Eroglu’s (1997) analysis, the interpretation of the object is
closely related to the case morphology.
3.1.4 Presuppositionality
Kelepir (2001:80-81) also observes that accusative marked indefinites can refer to ‘what
the speaker has in mind’ without assuming a set membership. In the accounts that were
mentioned, the accusative marked objects were analyzed as specific and specificity is
defined either as d-linking or referentiality in the sense of Fodor and Sag (1982). In
Kelepir (2001), it is shown that pre-suppositionality and referentiality are different:
accusative marked objects are pre-suppositional, but not necessarily referential or
partitive. In example (131), the translator that Ahmet is looking for is not necessarily a
member of a set previously mentioned in the discourse. Therefore, the accusative
indefinite in the example are not partitive, but still specific if specificity is defined as
‘what the speaker has in mind.’ Although the interpretation that Kelepir discusses is
‘what the speaker has in mind,’ there is one more reading of the sentence and that is
‘what the agent has in mind.’ Therefore, the sentence in the answer in (131) can be
followed by the statement ‘... but I don ’ t know who ’ (131) indicating that the speaker
does not have a particular person in mind, but s/he knows that the agent of the action is
79
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
looking for a particular person that he has in mind. In both cases, there is a person either
the speaker or the agent has in mind, therefore the accusative case marked object still
refers to a certain person and is presuppositional.
(131) Q: Ahmet bugiinlerde ne yap-iyor?
Ahmet nowadays what doing
‘What is Ahmet doing these days?’
A: Bir pevirmen-i an-yor.
An interpreter look.for-prog
‘He is looking for a translator’
A ’: ... ama kim-i bil-mi-yor-um.
But who-acc know-prog-IS
‘.. .but I don’t know whom’
To test presuppositionality, Kelepir (2001) uses conditional clauses. The contrast between
case-marked and non-case marked indefinites can be seen clearly in the case of if-clause
islands. Compare the examples in (132) and (133). In these examples, when the object is
case marked the speaker presupposes that there is a mistake in the document, and when it
is not case marked, there is no such presupposition, that is, it is possible that there are no
mistakes in the document.
80
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(132) Bir hata-yi bul-ur-sa-n, git-me-n-e izin ver-eceg-im.
one mistake-acc find-aor-cond-2S go-inf-poss-datpermit-fut-lS
‘If you find a mistake, I will let you go.’
(133) Bir hata bul-ur-sa-n, git-men-e izin ver-eceg-im.
one mistake find-aor-cond-2S go-inf-poss-dat permit-fut-IS
‘If you find a mistake, I will let you go.’
Kelepir (2001) also argues that such indefinites cannot be considered referential in the
sense of Fodor and Sag (1982) (contrary to what Zidani-Eroglu (1997) proposes)
because, they do not always get the widest scope, which was the prediction of Fodor and
Sag's (1982) referentiality account. I will discuss such 'narrow scope' cases in the sections
below and argue that they are not necessarily contrary examples.
3.1.5 Are accusative marked indefinites quantifiers?
Diesing, (1992) based on the data reported in En? (1991), associates accusative case
marking with wide scope and proposes that accusative marked indefinites are quantifiers
that obligatorily undergo quantifier raising at LF (c.f. May, 1977, May 1985). In her
account, accusative objects are external to VP (134), and they have a presuppositional or
specific reading and wide scope with respect to other quantifiers and negation due to their
syntactic position. A similar account is found in Kennedy (1994). She proposes that the
81
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
caseless objects remain in VP and are licensed through strict sisterhood relationship.
Case-marked objects, on the other hand, move outside the VP.
(134) [birN -acc ... [ v p ...]]
As also discussed by Zidani-Eroglu (1997) in detail, the accusative marked indefinites
are different from genuine quantifiers as they can escape scope islands. In the example in
(135), the accusative marked object in the if-clause can have a wide scope reading which
would not be possible if it were a quantifier.
(135) Eger Ali okulla ilgili iki rapor-u dii§unup duruyorsa i§i zor.
‘If Ali school-with related two report-acc keeps thinking job-3poss difficult.’
If Ali has been thinking about two reports / two of the reports related to
school, he faces a difficult task.
((48) in Zidani-Eroglu (1997: 116))
3.1.6 Summary and a generalization
We have seen that the accusative marked indefinites have various interpretations which
are not limited to partitivity or D-linking. Based on what we have seen, we can
summarize the realization conditions of the accusative case as in (136).
82
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(136) Accusative Case Realization
Objects are overtly accusative-marked when (i) they are presuppositional, (ii)
they are members of a presupposed set.
The first part of the condition takes care of the objects discussed in Kelepir (2001). The
second part of the condition accounts for the types of indefinite objects that are discussed
by Eng (1991). In both conditions, discourse/pragmatics is involved and it interacts with
the interpretation of the case morpheme. It is also important to note that this does not
mean that the presuppositional interpretation comes from the accusative case. As I will
discuss below, I propose that the objects bear an accusative case morpheme when they
are already presuppositional.
So far we have seen how the accusative marked indefinites and non-case marked
indefinites are different in terms of their interpretations. In the section below, I discuss
the scope properties o f the two object types.
3.2 Accusative case and scope
3.2.1 Wide Scope
We saw in the discussion above that the interpretation of the accusative-marked objects
closely reflected on the presence or absence of the case morpheme. In this section I will
argue that regardless of their position, case marked objects have wide scope with respect
to adverbs and negation. This is what we see in the case of the negation-indefinite
83
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
interaction examples repeated here in (137). In all three positions, the object has wide
scope with respect to negation.
(137) a. Ay§e bir elma-yi ye-me-di. [D P sub j bir N-acc V-neg]
Ay§e an apple-acc eat-past
‘Ay§e did not eat a particular apple.’
b. Birelma-yi Ay§e ye-me-di. [bir N-acc D P subj V-neg]
an apple-acc Ay§e eat-past
‘Ay§e did not eat a particular apple.’
c. Ay§e ye-me-di bir elma-yi. [D P subj V-neg bir N-acc ]
Ayse eat-past an apple-acc
‘Ay§e did not eat a particular apple.’
In (138), we see the interaction of the indefinite object with the adverb twice. As
predicted, the accusative marked objects have wide scope with respect to the adverb
twice regardless of their relative position to the adverb (in each case, one object is acted
upon twice)23. When the object is not case marked, it has narrow scope (any number of
objects is involved in the action).
2 3 Accusative marked objects in the examples can have various interpretations such as ‘a particular chair’
rather than ‘any one of the chairs.’ The discussion of the indefinites in this section and throughout the
84
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(138) a. Ayse iki kere bir sandalye / sandalye it-ti twice > bir N / N
Ayse twice a chair /chair push-past
‘Ayse pushed a chair twice.’ (any chair)
b. Ayse iki kere bir sandalye-yi it-ti bir N-acc > twice
Ayse twice a chair-acc push-past
‘Ayse pushed a particular chair twice.’ (one of the chairs)
c. Ay§e bir sandalye-yi iki kere it-ti bir N-acc > twice
Ayse a chair-acc twice push-past
‘Ayse pushed a particular chair twice.’ (one of the chairs)
Therefore it is the case morpheme that signals the presuppositionality of the object, rather
than the linear order of the adverb or the negation morpheme. Since the accusative
marked objects can have the same interpretation and have the same scope assignment in
both scrambled and non-scrambled positions2 4 , we can conclude that the objects do not
scramble to check a particular feature (e.g., specificity). Rather, as I will argue below, by
virtue of being outside of the narrow scope domain (which is restricted by the maximal
dissertation is restricted to the experimental conditions/contexts that I used.
241 refer to the linear ordering of constituents here. I argue that accusative marked objects that have a
presuppositional interpretation are scrambled regardless of their surface position. In the statement here,
however, by scrambling, I just refer to the surface displacement of the object.
85
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
projection that is above VP where both bir N and bir N-acc objects move), they have a
presuppositional interpretation. Under the present analysis, scrambling is treated as a
discourse/focus related re-shuffling of the constituents, as the interpretation of the objects
do not come from various maximal projections where the objects land to check their
features (see Schaeffer, 1997, and later work for such an account for Dutch). No matter
where objects appear in the structure, they have a presuppositional interpretation as the
interpretation is realized within the DP domain. The scope assignment condition is
summarized in (139).
(139) Scope Assignment2 5
Overtly accusative-marked objects have wide scope w.r.t. negation and
adverbs.
The condition above states that the scope assignment to objects is independent of other
factors such as word order and that the accusative marked objects will have wide scope
with respect to negation and adverbs. In the following section, I discuss some examples
that appear to be counter examples to this generalization and revise the scope assignment
condition stated above.
2 5 This condition will be revised in the following sections.
86
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.2.2 Can accusative marked indefinites take narrow scope?
All the accounts mentioned above acknowledge that the accusative marked indefinites are
different from the non-case marked indefinites and that case morphology is closely
related to the interpretation of the objects. In contrast with the accounts discussed above,
Komfilt (2001) argues that accusative marked objects can be non-specific. Komfilt
(2001) makes a distinction between formal and semantic reasons behind the occurrence
of the accusative case. She proposes that it is possible for an accusative marked object to
have a non-specific interpretation if its realization is driven by a formal requirement. Her
examples come from the use of partitives in imperatives exemplified in (140) and (141).
She argues that the occurrence o f the accusative case in such examples is due to a formal
reason (as opposed to a semantic reason), i.e., the possessive marker, which always
requires the accusative case. See Nilsson (1985) for a similar observation.
(140) Kitaplar-dan/-in ikisi-ni al, geri-si-ni kutu-da birak.
Book-plu-abl/-gen two-poss-acc take remainder-poss&3S-acc box-loc leave
‘Take (any) two of the books and leave the remainder (of the books) in the
box.’
(141) *Kitaplar-dan/-in ikisi al, geri-si-ni kutu-da birak.
Book-plu-abl/-gen two-poss take remainder-poss&3S-acc box-loc leave
‘Take (any) two of the books and leave the remainder (of the books) in the
box.’
87
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
According to Komfilt (2001), because o f the nature of the imperatives, the objects in the
examples cannot refer to ‘what the speaker has in mind,’ because it would be odd for the
addressee to guess which books she has in mind. Therefore Komfilt (2001) argues that
the objects cannot be ‘specific’ and the realization of the accusative case is independent
of interpretation. It is important to note, however, that the objects in the examples are not
specific only if the notion of specificity is defined in relation to what the speaker has in
mind. If we adopt a definition which is based on partitivity and set membership (c.f. Eng,
1991), the objects in Komfilt’s examples can be analyzed as specific, because they are
members of a previously mentioned or visually available set of books. In other words, it
is not true that the objects in the examples are non-specific.
Although these examples do not necessarily lead us to the conclusion that the accusative
case requirement is due to formal reasons and is independent of a particular
interpretation, it is still worth pursuing the idea that case marked objects are not always
specific. In the following discussion I will propose that the possessive morpheme in the
examples is playing a role in the realization of the accusative case despite the
interpretation. Now refer to the contrast in (142) and (143) once more (Example (98) in
Kelepir, 2001:78). As discussed by Kelepir (2001) when the object bears an accusative
case and occurs in an if-clause, it has a pre-suppositional interpretation. The speaker
knows or presupposes that there is at least one mistake in the document. In the example
with the non-case marked object, however, such a presuppositional reading is not
accessible.
88
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(142) Bir hata-yi bul-ur-sa-n, git-me-n-e izin ver-eceg-im.
one mistake-acc find-aor-cond-2S go-inf-poss-dat permit-fut-lS
‘If you find a mistake, I will let you go.’
(143) Bir hata bul-ur-sa-n, git-me-n-e izin ver-eceg-im.
one mistake find-aor-cond-2S go-inf-poss-dat permit-fut-lS
‘If you find a mistake, I will let you go.’
Now see the example in (144), where the object has a possessive marker together with an
accusative case morpheme but still does not have to have a presuppositional reading. In
other words, the structure is ambiguous between the two readings, while in the accusative
example above, only the presuppositional reading is available26.
(144) Bir hata-m-i bul-ur-sa-n git-me-n-e izin ver-eceg-im.
one mistake-poss&lS-acc find-aor-cond-2S go-inf-poss-dat permit-fut-IS
‘If you find a mistake of mine, I will let you go.’
The example shows that whenever there is possessive morphology, the accusative case
cannot be deleted although the object is not presuppositional. The interpretation of (144)
is similar to the one that we have in the case of the non-case marked object above. The
2 6 It can be disambiguated with stress shift. The role of stress shifts/ prosody on interpretations will be
discussed later.
89
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
accusative case is obligatory on objects that have a possessive morpheme, and its
behavior and the interpretation show that it is required by a formal reason that Komfilt
(2001) proposes. In such cases, the object has narrow scope with respect to negation and
27
adverbs despite the overt case morpheme .
Another objection to the wide scope reading of the accusative marked objects in all
positions comes from Kelepir (2001, 2003) who observes that when the verb is negated,
and the subject is an NPI, the wide scope reading of the accusative marked indefinite is
blocked.
(145) Kimse parti-ye bir arkada§-im-i davet et-me-mi§.
Nobody party-loc a friend-poss&lS-acc invite-neg-perf
‘Nobody invited any friend of mine’
* A friend of mine such that nobody invited him/her
Note that the object in this example bears a possessive morpheme too. When the object in
the example does not bear the possessive morphology, that is, when the object is “a
student” or “a friend”, rather than “a friend of mine” wide scope reading is salient.
Therefore, it is likely that the reading results from the possessive morpheme just as it is
2 7 Kural (1997) observes that scope relations can change when there is an interaction with binding theory,
i.e. objects have narrow scope although they are case marked when the binding theory demands it. He does
not provide an account. Whether or not binding theory can account for the realization and the interpretation
o f the accusative case needs to be investigated.
90
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the case in the examples above28. The example further supports Komfilt's proposal that
the formal realization of the accusative morpheme is independent of its interpretation.
(146) Kimse parti-ye bir arkada§-i davet et-me-mi§.
Nobody party-loc a friend-acc invite-neg-perf
‘Nobody invited a friend’ (A friend such that nobody invited him/her)
The dual behavior of the accusative case is not something peculiar to the accusative case
in Turkish. A similar occurrence property is reported for Kannada by Lidz (2000). Lidz
observes that there is a correlation between the optionality of the accusative case and the
possible interpretations that an object can have in Kannada. In that language, too, the
object can either appears in the bare form, or it can bear an accusative case morpheme
when it is singular or inanimate. When the object is plural and animate, the accusative
case is obligatory and the object can have a narrow scope reading despite the case
morpheme. If the accusative case is optional, the narrow scope reading is not possible
with the accusative case and is restricted to the non-case marked object. In other words,
when a singular object is case marked, narrow scope is not possible, as the language
provides the bare object option for narrow scope reading.
2 8 Kelepir (2001, 2003) does not state the reason why she was using objects that have a possessive
morphology or whether it makes a difference in the interpretation when she does not have a possessive
marked object in this particular example. In order to account for the narrow scope reading in this example
she proposes that indefinites are interpreted as choice functions (Reinhart, 1997, Winter, 1997, Kratzer,
1998, Matthewson, 1999, among others). The choice function proposal cannot account for the contrast
between objects that have a possessive morphology and those that do not. Kelepir (2001) proposes that the
accusative marked indefinites are pre-suppositional but they do not necessarily take the widest scope.
91
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Moreover, if the subject is quantificational, obligatorily case marked objects have only
narrow or intermediate scope, not the widest scope. In our example in (145), too,
accusative case is obligatory,29 there is a quantificational subject, and the object behaves
like the objects in Kannada, i.e., it cannot have the widest scope reading. In structures
that do not have a possessive morpheme, however, we can have access to a wide scope
interpretation (146) and the narrow scope interpretation is realized by the non-case-
marked objects (148).
2 9 Kelepir (2001:94) states that accusative case can be omitted after the possessive marker and gives the
following ‘possessive compound’ example. In Turkish in so-called ‘possessive compounds’ the compound
marker on the head noun is homophonous with the third person singular possessive marker — si. However,
in such examples, the status of — s i as a possessive marker is questionable. It is typically analyzed as
compound marker, rather than an actual possessive marker in the literature (van Schaaik, 2000 among
others).
(i) Bugtin balik 9orba-si iq-ti-k
Today fish soup drink-past-lP
‘We drank fish-soup today.’
The strongest argument for the analysis of - s i as possessive is that - s i is interchangeable with the other
possessive markers (other persons) (e.g., balik gorba-m versus *balik gorba-si-m ‘my fish soup’). Note
however that when we have — m instead o f - s i, the accusative case is obligatory. That is, in such examples,
when we have an actual possessive marker, the accusative case is still obligatory. Similarly, when the
possessive interpretation of -si is emphasized by the addition of the possessor onun ‘his’ the case marker
cannot be dropped. Therefore when - s i is a possessive marker, it behaves like the examples I discussed
above. Kelepir (2001) acknowledges that the status of s i is controversial
(ii) Bugiin ben-im balik gorba-m-i / *9orba-m-0 i9 -ti.
Today my fish soup-poss&lS-acc / soup-poss& lS-0acc drink-past-lP
‘S/he drank my fish soup today’
(iii) Bugiin o-nun balik porba-si-m/ *9orba-si-0 i9 -ti-k.
Today his fish soup-poss&3S-acc / soup-poss&3S-0acc drink-past-lP
‘We drank his fish soup today’
92
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(147) *Kimse parti-ye bir arkada§-im-0 davet et-me-mi§.
Nobody party-loc a friend-poss& lS-0acc invite-neg-perf
‘Nobody invited any friend of mine.’
(148) Kimse parti-ye bir arkada§-0 davet et-me-mi§.
Nobody party-loc a friend-0acc invite-neg-perf
‘Nobody invited a friend to the party.’
Based on the data he provides, Lidz (2000: 335) concludes that the semantic composition
is independent o f morphological form and “the range of morphological options interacts
with the range of semantic options to determine the meaning of a given morpheme in a
given context.” In the Turkish examples, too, it seems that the realization of the
accusative morpheme is not necessarily determined by semantic factors and can be
related to idiosyncratic formal requirements. The generalization about accusative case
deletion can be stated as in (149):
(149) Accusative Case Deletion
(i) Accusative case gets deleted if the object is non-presuppositional and
appears adjacent to the verb.
(ii) Accusative case cannot delete when the object carries a possessive
morpheme.
93
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Based on what we have seen in this section, we can re-formulate the scope assignment
condition as in (150). Note that this revised condition excludes the objects that have a
possessive morpheme and that can have narrow scope with respect to negation and
adverbs.
(150) Scope Assignment (revised)
Overtly accusative-marked objects have wide scope w.r.t. negation and
adverbs when they are presuppositional.
In this section, I discussed the scope taking properties of the objects and saw that the
unexpected behavior of some accusative marked indefinites can be attributed to the
idiosyncratic morpho-phonological structure of the language which restricts the case
deletion when the object carries a possessive morpheme. In the next section, I show that
the narrow scope and non-presuppositional interpretation of objects are restricted to the
immediately preverbal position. I present more data and discuss the role of adjacency in
the interpretation of objects’.
3.2.3 Adjacency
Let us first go back to our original example and see whether there can be another factor
why the narrow scope reading is the only available reading when the object is adjacent to
the verb and co-occurs with a quantifier subject. First, let us look at the example in (151).
94
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In this example, the object which bears the accusative case and the possessive morpheme
is scrambled to the sentence initial position, where it precedes the quantifier subject.
(151) Bir arkada§-im-i kimse parti-ye davet et-me-mi§.
a friend-poss&lS-acc nobody party-loc invite-neg-perf
‘Nobody invited a friend of mine to the party.’
The crucial point is that the object has the widest scope when it is in a scrambled
position, therefore the narrow scope reading is restricted to the immediately preverbal
position where the object is adjacent to the verb30. In the discussion that follows, I will
3 0 Kelepir (2001), observes that when the object is accusative marked and adjacent to the verb, the structure
is ambiguous as follows: (1) There is a book that all the students read (there is one common book for all the
students), (2) For every student there is a specific book (every student has a different book, but it is a
specific book for each). When the accusative-marked object is not in the immediately preverbal position as
in (ii), the wide scope interpretation is the only interpretation possible. Therefore, adjacency plays a role in
scope assignment.
(i) Her ogrenci bir kitab-i oku-du every student > a book-acc
Every student a book-acc read-past a book-acc> every student
‘Every student read a book.’
(ii) Bir kitab-i her ogrenci oku-du *every student > a book-acc
a book-acc every student read-past a book-acc > every student
‘A book such that every student read it.’
The examples that she discusses are different from the ones I discuss here. In her examples, the subjects are
quantifiers and she is discussing the scope of objects with respect to the quantifier subjects. Besides the
object has an intermediate scope, rather than a true narrow scope. The object still refers to a particular book
for each individual subject, which can be chosen from sets that were defined for each individual subject.
Therefore, they do not take narrow scope in the sense that the object refers to any book (as it is the case
with non-case marked objects). Because there is a set for each individual subject and the book is drawn
from that set, the kind of scope that these objects have is not the same as the scope that we have in bare
objects or non-case marked objects and a contrast between case marked and non-case marked objects is still
maintained. Moreover, we do not have this reading when the subject is a single person i.e., the distributive
reading is not accessible in examples such as (iii), therefore, what causes the intermediate or distributive
reading is the quantifier in the subject position. More importantly, when the sentence is negated (iv), the
95
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
propose that the adjacency requirement for narrow scope reading can be attributed to the
restrictions on the information/ focus structure of the language.
Although the canonical word-order is SOV in Turkish, it is generally agreed that simple
Turkish sentences have the following (152) pragmatically determined word order (Dede,
1986, Erguvanli, 1984, Issever, 2003, among others).
(152) Topic ...Focus.. .V / Predicate.. .Background
The immediately preverbal position is the neutral focus position31 where the new
information is located. Accusative marked indefinites bear the given information or old
information in sentential discourse and this property conflicts with their position unless
the interpretation is controlled with a particular stress shift to another constituent. The
object has wide scope with respect to negation, although it cannot scope over the subject.
(iii) Ogrenci bir kitab-i oku-du.
student a book-acc read-past
‘The student read a book.’
(iv) Her ogrenci bir kitab-i oku-ma-di.
Every student a book-acc read-past
‘Every student read a book.’
Therefore, although I have the same proposal with respect to the adjacency to verbs, the data I am dealing
with are different. Whether or not such subject-object scope relations can be accounted for under my
analysis needs further investigation.
3 1 Goksel & Ozsoy (2000a, 2000b) propose that the focus position is not necessarily the immediately
preverbal position but rather focused constituents can appear in a larger preverbal domain. In the statement
here I refer to a neuter intonation focus position which is immediately preverbal.
96
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
non-presuppositional objects that carry new information occur in the immediately
preverbal position and bear the focal stress.
Kural (1992:15) notes that accusative marked indefinite objects are somewhat “awkward”
in immediately preverbal position (153)a. They sound better when they are non-adjacent
(153)b.
(153) a. ?Ahmet yol-da [U 9 ki§i]-yi gor-mii§.
Ahmet on the way three person-acc saw
‘Ahmet saw three (specific) people on the way.’
b. Ahmet [ii? ki§i]-yi yolda gdr-mu§.
Ahmet three person-acc on the way saw
‘Ahmet saw three (specific) people on the way.’
The incompatibility of the accusative marked objects and the ‘new information’ or ‘non-
presuppositional constituent’ position may be the reason why accusative marked
indefinites do not sound good to Kural (1992). Once the stress is shifted to the verb or
any other constituent, the structure sounds better, as the object is no longer interpreted in
the immediately preverbal position.
Let us go back to the narrow scope accusative marked object example in Kelepir (2003),
which, potentially, has a similar problem. In the example, the object ‘a friend of mine’ is
97
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
located in the immediately preverbal position, which is the new information position.
When the object is non-adjacent to the verb, it is more likely to be interpreted as the old
information and this focus effect results in a tendency to attribute a specific reading to it.
When the object is adjacent to the verb, it is interpreted in a new information slot and
thus a non-specific reading is forced.
To gain a wide-scope reading, then, accusative marked indefinites should appear in a
given information position in addition to bearing case morphology as the case
morphology by itself, cannot be sufficient to provide a particular interpretation. Shifting
the stress to the verb, is a way to establish a wide scope interpretation for the accusative
marked objects that appear in the preverbal position. This account is compatible with
Nakipoglu-Demiralp (2004) that argues that scrambling is driven by focus/prosody
requirements rather than syntactic/semantic constraints and the interpretation of the
object is based on this focus driven reshuffling.
In this section we saw that the accusative marked indefinites have to be in a position that
is compatible for their presuppositional interpretation. In the following section, I propose
a clause structure for Turkish that can derive the readings that we have been discussing so
far.
98
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.3 Putting everything into structure
3.3.1 Presuppositionality within DP
In the discussion above, I showed that the accusative marked indefinite objects are
different from the bare and bir N nominals in terms of their interpretation. I also showed
that the interpretation of the accusative case morphology is dependent on other factors
such as the presence of the possessive morphology and being non-adjacent to verbs. I
propose that the presuppositionality reading of the object is realized within DP. When the
DP has a presuppositional reading, the accusative case is realized overtly (154) when the
DP has a weak interpretation, the accusative case is deleted (155). In these two cases, the
objects occur in different structural position, although they both appear adjacent to the
verb on the surface.
(154) ACC on strong DP
D
strong ^ + D-linked / Presuppositional
+ACC
99
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(155) ACC (deletion) on weak DP
D ’
Nui
- D-linked / Non-presuppositional
+ACC (deleted under adjacency)
In the case of the objects that bear possessive morphology, the case morpheme cannot be
deleted (156), but the object does not necessarily have a presuppositional interpretation.
The occurrence of the possessive morpheme in such examples is more like the occurrence
of other (lexical) case morphemes such as the dative and locative that cannot be deleted
even though the object is non-referential (157).
(156) Possessive objects+ACC
D ’
Dweak/strong -------- ► - /+ D-linked / (non)presuppositional
bir arkadas-poss +ACC (cannot delete)
(157) Telefon-a/*-0 bak-acak bir sekreter ari-yor-um
Telephone-dat/-0dat answer-fut a secretary looking-for
‘I am looking for a secretary to answer phones.’
100
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In the following section, I discuss how the presuppositionality of the object interacts with
the structure and discourse organization.
3.3.2 Scrambling as a way to escape from non-presuppositionality
In order to get a presuppositional interpretation, an object has to be outside the non-
presuppositional interpretation domain, which is the VP adjacent position. This position
can also be treated as the focus position and objects that bear the old information and
those that have a presuppositional interpretation need to escape from this position to be
able to maintain their presuppositional reading. Note that these objects do not
move/scramble to a position to get an interpretation. They move to escape from a
position that is incompatible with the interpretation that they have (158)a. Those objects
that are not presuppositional remain in the VP adjacent position and their case is deleted
under adjacency to verb (158)b. These objects that have a weak D (c.f. Milsark, 1974, de
Hoop, 1992, for example) and that remain in focus domain are not incompatible with
their interpretation.
(158) Narrow-scope (non-presuppositional) domain of objects
a. [ [ DP subject • • • [NegP
b. [ [ DP subject • • • [NegP
[AgrOP bir N-acc [v p ...] ]
U g r o p b ir N-aee [Vp...] ]
•••]]]
•••]]]
101
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
This account is compatible with the focus driven scrambling account proposed by
Zubizarreta (1998), where the defocalized constituents move to a position out side of the
focus area in order to ensure that the focused constituent can occur in a position where it
can get main focal stress. In Zubizarreta’s (1998) account, presuppositional constituents
scramble to a position where they can have defocalized interpretation. In the present
account, too, the accusative marked object that has a presuppositional interpretation move
to a position outside of the focalized domain.
The structures of the scrambled objects are shown in (159) and (160). In the first one the
object scrambles to a position to the left of the subject, escaping from the non-
presuppositional domain. In the second structure, it scrambles to a lower position, below
the object, but still outside the presuppositional domain. Note that in the second case the
accusative marked objects appear adjacent to the verb in the linear order, although it is in
a non-adjacent position above the NegP and AgrO. In such a position it no longer bears
the focal stress. This is the structure where either the verb is stressed or a pause is
inserted after the object. Thus the prosody indicates that the object is no longer in the
focus position.
(159) hir N-acc D P subject V-neg
[bir N-acc... [ DPsu b ject • • • [N eg p Ugropbir N aee [vp...]]]]]]
t I
102
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(160) DPS u b ject bir N-acc V-neg
[ [ DP su b ject [bir N -acc... [N egp [A grop bir N acc [v p ...]]]]]]
t__________ 1
In the case of the possessive objects, if the object is presuppositional, it scrambles in the
same fashion described above to a position that is outside of the non-presuppositionality
domain (161). If it has a non-presuppositional interpretation it remains within the lower
domain, where the non-case marked bir N objects occur (162). Its case cannot be deleted,
but it can still have non-presuppositional interpretation due to its position.
(161) DP su b ject bir N-poss-acc V-neg / bir N-poss-acc DPS U bjec t V-neg
a. [ [ DPsu b ject [bir N-poss-acc.. .[N eg p [A grop bir N poss - ac-e [v p ...]]]]]]
t__________ |
b. [bir N-poss-acc [ DPsu b ject • • -[N eg p [ A grop bir N -poss-acc [vp...]]]]]]
t________________I
(162) DPS u b ject Zu'r N-(poss-)acc V-neg
a. [ [DPsubject •••[N egp U grO P bir N-poss-acc [vp...]]]]]]
b. [ [ DPsubject • • - [N egp [A grop bir N-acc [v p ...]]]]]]
The scope of the accusative marked objects, then, depends on the presuppositionality of
the objects. If they are d-linked or presuppositional, they have wide scope with respect to
the other constituents and as a requirement of being presuppositional, they move to a
position above NegP so that they can escape from the non-presuppositional domain. Only
103
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
those objects that are outside of this domain can have wide scope with respect to negation
or adverbs.
3.4 Postverbal Objects
Postverbal objects are important for the present study because accusative marked objects
obligatorily have specific interpretation although they follow negation in linear order.
(163) Ay§e ye-me-di bir elma-yi. bir elma-yi >neg; *neg> bir elma-yi
Ay§e eat-neg-past an apple-acc
‘Ay§e didn’t eat an apple’.
Their specific reading comes from (at least) two properties. First, the postverbal position
is a position for backgrounded information (e.g., Erguvanh, 1984, Zimmer, 1986), and
the reference of the object in the backgrounded position is discourse-predictable,
presupposed or is recoverable from previous discourse. New referents in discourse are
never postposed, as predicted, non-specific indefinites never occur in the postverbal
position (Erguvanh, 1984). The fact that post-verbal constituents cannot have focal stress,
and question words cannot appear postverbally are also in line with the backgrounding
property of the postverbal position.
Secondly, in Turkish, the postverbal constituents are obligatorily outside the linear
domains of the verb, which appears in the terminal position in Turkish due to the head-
104
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
final property of the language (Goksel, 1997, Goksel & Stavroula, 1995). They are
different from the postverbal objects in English where the objects are still in the (c-
command) domain of the verb. According to Kural’s (1997) proposal, in which they are
analyzed as adjuncts to CP, the post-verbal objects are outside the hierarchical domain of
the verb as well. In his account, the postverbal constituents are base-generated in a
preverbal position, and are moved to the postverbal position. His proposal is based on the
following facts.
Postverbal constituents cannot appear in embedded clauses (164) and Kural attributes this
property to their adjunct status i.e., their position, which is, as he shows, high enough to
interact with the main verb.
(164) a. Ahmet-in Bema-ya [bu kitab-i] ver-me-si-ni isti-yor-um.
Ahmet-gen Bema-dat this book-acc give-neg-poss&3S-acc want-prog-IS
‘I want Ahmet to give this book to Bema.’
b. * Ahmet-in Bema-ya ver-me-si-ni [bu kitab-i] isti-yor-um.
c. Ahmet-in Bema-ya ver-me-si-ni isti-yor-um [bu kitab-i].
Post-verbal quantifiers have wider scope with respect to the preverbal quantifiers as seen
in the examples in (165), (166), (167) and (168) below ((18) and (19) of Kural (1997)).
105
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(165) Herkes dun ara-mi§ U 9 ki§i-yi.
everyone yesterday call-perf three person-acc
‘Everyone called three people yesterday.’
(166) U 9 ki§i-yi dim ara-mi§ herkes.
three person-acc yesterday everyone call-perf
‘Everyone called three people yesterday.’
(167) U 9 ki§i dim ara-mi§ herkes-i.
three person yesterday call-perf everyone-acc
‘Three people called everybody yesterday.’
(168) Herkes-i dun ara-mi§ U 9 ki§i.
Everyone yesterday call-perf three person
‘Three people called everybody yesterday.’
According to Kural (1997), postverbal constituents c-command preverbal subject and
object QNPs at the level of representation where scope is assigned and thus they always
have wide scope with respect to the preverbal QNPs. Although Kural (1992, 1997)
shows clearly that the postverbal objects are in a position where they c-command the verb
and subjects, their scope property with respect to other constituents do not necessarily
result from their position. As stated earlier, accusative marked indefinite objects do not
have to be in a c-commanding position to have wide scope. Rather, they have wide scope
106
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
in any position due to the properties of their DP as long as they are outside of a particular
domain. Their case morphology cannot get deleted.
(169) [c p [ D P s u b j e c t • • ■ [ N e g p . . . | L \ g r O P bir N ace [v p ...] • [ • •]]]] bir N-acc]
What is interesting about the postverbal objects is that their scope property is not marked
with case morphology only. Their linear (and hierarchical) position signals that they have
a particular DP that is associated with wide scope interpretation.
3.5 An alternative account for the accusative case
According to the recent account of Oztiirk (2004) case morphology is a referentiality
marker and a type shifter that turns predicates into arguments. It is equivalent to the
determiners in determiner languages such as English. In that account, Turkish does not
have a DP projection and what substitutes the DP projection is the case morphology. In
this account, non-case marked bir N objects and bare objects both pseudo-incorporate to
verbs (a la Massam, 2001) and they do not qualify as arguments of the verbs. Only those
case marked objects are assigned referentiality and are arguments.
The type-shifting account assumes that non-case marked objects, because they do not
bear a case morpheme, cannot be arguments. As I discussed in the previous chapter, this
is problematic for bir N objects especially, which can occupy argument positions. Bir N
objects are referential in the sense of Givon (1978), which is adopted by Massam (2001).
107
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Oztiirk (2004) proposes a pseudo-incorporation account a la Massam, but adopts the
referentiality definition of Fodor and Sag (1982) instead of the one proposed by Givon
(1978), which Massam (2001) adopts. The difference between the two referentiality
definitions is important because according to Givon's referentiality bir N objects are
referential (and non-specific) in Turkish, whereas in Fodor and Sag's (1982) definition
they are not referential (in the sense that they do not refer to particular, specific objects).
In other words, according to the original account of Massam (2001) there is no contrast
between case marked versus non-case marked objects and thus, case cannot be treated as
a type shifter or an indicator of pseudo-incorporation. Note also that, in determiner
languages (e.g., English) the indefinite article is analyzed as the head of a DP and 'an N'
objects can be arguments. Analyzing case as the only type-shifter requires that only the
specific / presuppositional objects can be arguments in Turkish. In the type-shifting
account no justification is presented for the reason why this should be the case.
In the present analysis I propose that the occurrence of the morpheme cannot be restricted
with a syntactic type-shifting role. The realization and the interpretation of the accusative
case is driven by much more complex mechanisms which involve an interaction of
multiple components of grammar, as discussed above.
3.6 Summary and Conclusion
In the previous chapter, I showed that the bare nominals are generated within VP and
remain there due to their structural properties (that they lack NumP and DP) and they
108
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
form some kind of a loose complex predicate within VP. I also showed that bir N
nominals, although they can move out of VP, still have to be adjacent to verbs, because
they are non-presuppositional. Their case is deleted under adjacency. Here are these
structures, once more, together with the accusative marked indefinite structures.
(170) a.NV [...[ [v p [np book] read]]]
b. Bir N [... [bir book-aee [ v p [ d p bir book acc] read ]]]
c. Bir N-acc [ bir book-acc... [bir book acc[V p [D P bir book acc] read ]]]
Those object structures that have bir, that is, those that have NumP and DP projections,
move out of VP so that they can escape from the verbal domain and have an independent
interpretation. Both the accusative marked indefinites and the non-case marked
indefinites move to the [spec, AGRoP] and bear an abstract case feature. I assume that
once the DP is projected and the nominal is referential and have definiteness and
specificity features (+or -) they move out. The motivation for this movement does not
come from any specificity, discourse related reasons, but it is simply a structural property
of nominals that have DP projection. Bir N-acc and bir N nominals move to the same
position, and bir N nominals cannot scramble any further, due to their non-
presuppositionality. Bir N-acc nominals cannot stay in this position because they are
presuppositional. They move to a higher position and can have wide scope. The
realization of the case morpheme is dependent upon the type of the DP (weak or strong)
and presuppositionality of the object. The interpretation of the overtly realized case
morpheme then further depends on the position where this object occurs. If it is d-linked
109
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
or presuppositional, then the object has wide scope over negation and the adverb and
obligatorily moves to a position above NegP. If it is realized only because it cannot delete
due to some idiosyncratic morpho-phonological reason, it appears adjacent to the verb
and can have narrow scope.
In Turkish objects do not move for checking a feature such as specificity, rather they
scramble for discourse reasons and information re-shuffling and to escape from certain
positions that are incompatible with their interpretation. Once they escape the narrow
scope domain, scrambling does not change the scope relations of the constituents as the
objects' scope is determined on the basis of the properties of the DP projections rather
than their surface ordering.
In the present study, to investigate the scope assignment to the indefinite objects I study
children’s and adults’ comprehension of accusative marked and non-case-marked objects
that interact with the adverb twice and negation. When the object is case-marked, it can
scope over the adverb twice, i.e., one particular chair is pushed in two pushing actions.
When we have the object in a non-case-marked form, it has narrow scope with respect to
the adverb (i.e., either two or one chair may be involved in the event). In the experiments
I conduct, I have two linear versions of the accusative marked object, one adjacent to the
verb, following the adverb, and the other preceding the adverb, non-adjacent to the verb.
The position of the accusative marked object does not result in a different scope
interpretation with respect to the adverb.
110
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In the case of the negated sentences we have a similar contrast between case marked and
non-case-marked objects. While the accusative marked objects scope over negation and
have a wide scope reading (i.e., there is an apple such that Ay§e didn’t eat), when the
object is not case-marked, it has narrow scope (i.e., it is not the case that Ay§e ate an
apple). Therefore the sentences that have scope ambiguity in languages, such as English,
have morphologically distinct counterparts in Turkish.
I l l
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4 Early indefinites, hypotheses and predictions
Various studies have been conducted on the acquisition of indefinites, specificity and
scope in typologically different languages. In this chapter, I present their accounts and
discuss their findings. Earlier studies can be categorized into three groups based on their
major claims: (1) Those that argue that children’s scope interpretation is related to (i) the
relative linear position of the indefinites and other quantifiers (i.e., the isomorphism
account) or (ii) the hierarchical position of the indefinites (i.e., the c-command account),
(2 ) those that argue that children’s and adults’ wide scope indefinites are different and the
difference is due to the relatively late acquisition of “discourse rules” rather than different
syntax-semantics mapping and (3) those that stress the role of the indefinite articles and
their semantic features that interfere with scope relations. In the following sections, I will
present a summary of all these accounts and finally close this chapter with a discussion of
their relevance and predictions that they possibly have for Turkish.
4.1 Isomorphism
According to the isomorphism account, children’s interpretation of indefinites is heavily
based on the linear ordering of indefinites in surface syntax. More specifically, in this
account, those quantified noun phrases (QNPs) that precede in linear order, have scope
over the others that follow them. The evidence for the isomorphic account comes from
children’s acquisition of any and some in English, and a contrast observed in the
acquisition of every and some indefinites in subject and object positions. Although the
112
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
present study does not focus on the acquisition of QNPs, a discussion of the work done
on QNPs is included in this chapter as the isomorphism account has implications for the
acquisition of scope in Turkish. Similarly, although the present study is based on an
analysis of the object nominals, a discussion of earlier comparative work on subjects and
objects are presented, because the contrast is important for the isomorphism account.
4.1.1 Any and some puzzle
In English, the relative interpretation of quantified NPs does not mirror the position that
these elements occupy in overt syntax. In both (171) and (172) the QNPs with any and
some occupy the object position, which is to the right of the negation in English, but any
has narrow scope while some has wide scope with respect to negation. In other words,
the sentence in (172) implies that no apples were eaten, while (173) implies that there
were some apples that were not eaten. The sentence in (171) has an isomorphic
interpretation, because the order of negation and the QNP mirrors the scope relation (_,3).
The sentence in (171), however is non-isomorphic because although the linear order of
negation and the QNP is the same as the one in (171), the semantic relationship is reverse
0 - ) .
(171) John didn’t eat any apples. > 3 ; * 3 >
(172) John didn’t eat some apples.
113
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Any is a negative polarity item that needs to be licensed and must be interpreted in the
scope of the element that licenses it (Baker, 1970, Ladusaw, 1979, 1980, Linebarger
1980, Progovac, 1994, among others). By contrast, some is not a negative polarity item
and cannot be interpreted under the scope of the negation that is in the same clause.
Therefore the sentences with some and any do not have scope ambiguity. A scope
ambiguity can be seen in a sentence such as the one in (173), where the QNP an apple
has scope ambiguity with respect to negation. In one reading, it implies that no apples
were eaten, in the other one, it reads as there is an apple such that it was not eaten. So
both isomorphic and non-isomorphic readings are possible with one syntactic form.
(173) John didn’t eat an apple. -,> 3 ; 3 > - '
To summarize so far, the isomorphic interpretation is obligatory in (171), it is not
possible in (171) and it is optional in (173) depending on the lexical nature of the
quantifier element. This variable behavior of quantifiers presents a potential challenge for
a young learner as one can predict. To confirm the difficulty, Thornton (1995) and
O ’Leary and Crain (1994) test children’s knowledge of some and any and find that
children have an adult-like interpretation of any but they produce some in contexts in
which any would be appropriate. The utterance in (174), from O ’Leary and Crain (1994)
is such an example.
114
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(174) So he didn’t get some money (E.G. 4; 10)
Following these studies, Musolino (1998) and Musolino, Crain and Thomtom (2000)
study children’s QNPs and their interaction with negation and conclude that a sentence
can have only an isomorphic interpretation in child grammar, i.e., children’s scope
assignment reflect the surface ordering of the quantifiers and negation. With such a
proposal they can explain why children master any earlier and have difficulty with some.
In the interpretation of any, children need to map the surface ordering to the semantic
structure. But in some there is not one-to-one correspondence. Below is a more detailed
description of the study that leads to the isomorphism proposal.
4.1.2 Every orange, some guys and two slices of pizza
Musolino (1998) investigated the scope interpretation of English speaking children
between the ages 3;10-7;3. In test sentences, he had indefinite objects or subjects and
negation as exemplified in (175) and (176). The indefinite objects every N and some N
had narrow scope with respect to negation. In the case of every N, the interpretation was
non-adult-like, while in the case of some N, which was the control structure, the
interpretation was adult-like. The examples with two N represented a case where the
object has scope ambiguity. They compared the scope assignments o f these structures
with those of the subject indefinites that are shown in (176). In those structures every N
had scope ambiguity, while some N represented a case where the subject had wide scope.
115
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(175) [neg... QNP]
a. Smurf didn’t buy every orange.
b. The detective didn’t find some guys / someone. *“■ > 3; 3 > -■
c. Cookie Monster didn’t eat two slices of pizza. -■ > 3; 3 >
(176) [QNP... neg]
a. Every horse didn’t jump over the fence. - > V; V > -
b. Some girls won’t ride on the merry-go-round. > 3; *3 >
The experiments showed that, when objects that have wide scope with respect to negation
appear in subject positions, children have an adult-like interpretation of the indefinites.
They have difficulty with the same indefinites when they appear in object position,
following the negation in English. Musolino (1998) attributed this non-adult-like
behavior to the isomorphic property of child grammar.
Here are the details of the experiments. I conducted a truth value judgment task on the
comprehension of the five grammatical structures. In the experiment that tested the first
structure that is exemplified in (175)a, 20 children between the ages 3;11-6;0 (mean age:
4; 10) witnessed an act out-story where Smurf bought one o f the oranges, but did not buy
some others at a grocery store. Children accepted the statement of the puppet (175)a
(narrow scope reading of the indefinites) 85% of the time, therefore children’s
interpretation of the sentence was adult-like. In the other structure, the sentence in (175)b,
116
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
was accepted by all the adult subjects for a description of a scene where the detective did
not find some of the guys (wide-scope reading), but the children accepted the sentence as
a remark of this scene in 50% of the cases. %65 of the younger children (mean age: 4;7)
rejected this reading. So children could assign narrow scope to objects in an adult-like
fashion, but they failed to have a wide scope interpretation o f the indefinites in the object
position.
The experiments conducted on the comprehension of subject indefinites revealed that the
problem children have is not in the assignment of wide scope reading, but rather the
assignment of wide scope reading to the indefinites in the object position. In the
experiment that tested children’s comprehension of every N in the subject position, with a
context where two horses jumped over a fence and one out of three horses did not, more
than 90% of the 20 children subjects (mean age 5;11, age range: 4;0-7;3) behaved like
adults and accepted the sentence in (176)a as the correct description of the scene (wide
scope reading). Children did even better in the experiment that tested the scope
assignment to some N in the subject position. All the children (mean age 4; 11, age range:
4;10-6;2) accepted the wide scope reading, just like adults.
The experiments conducted on two N indefinites confirmed the results reported above. In
this experiment, Cookie Monster ate two slices of pizza, and did not eat two other slices.
A group of children with mean age of 4;7 and adults made judgments about the puppets
utterances such as (175)c. While all adults accepted the test sentences as the correct
commentary on the scene, revealing that they could assign wide scope reading to two N
117
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
object, 50% of the children (3;11-6;1, mean age:4;10) rejected the wide scope reading. In
the younger age group (mean age 4;3), children accepted the wide scope reading in only
27% of the instances.
The findings of Musolino (1998) support earlier findings of Bergsma-Klein (1996),
namely that children do not assign wide scope reading to objects that obligatorily have
wide scope reading for adults. In order to account for this contrast, Musolino (1998) and
later Musolino, Crain and Thornton (2000) proposed that children’s responses resulted
from an isomorphism, that is, children go with the syntactic ordering (linear ordering) of
the negation and the nominal, rather than their semantic scope. In (177), their observation
is stated ((12) in Musolino, Crain & Thornton, 2000).
(177) The observation o f isomorphism
Unlike adults, young children systematically interpret negation and QNPs
on the basis of their positions in overt syntax.
According to Musolino, Crain and Thornton (2000) the reason that lies behind children’s
isomorphism is the inability o f their grammar to generate non-isomorphic structures.
They observe that there is a parametric variation across languages that allow non
isomorphic interpretations. Some languages such as Chinese, do not allow non
isomorphic scope assignment. Therefore the Chinese value of the parameter allows a
subset of the scope assignment options possible for languages such as English. Thus, the
learners adopt the subset values of the parameter and then reset the parameter at later
118
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
phases o f language development. Another parametric variation is observed in the types
of QNPs. Following Homstein (1984), Musolino, Crain and Thornton (2000) propose
that QNPs are of two types. Type-1 QNPs are subject to no-movement and have their
interpretation through choice functions (cf. Reinhart 1997). Type-2 QNPs are derived
through movement, which is a grammatical mechanism. The interpretation of Type-2
QNPs is fixed by overt position with respect to other quantifiers and negation. Type-1
QNPs are ambiguous. Since Type-1 quantifiers have more interpretative option than the
Type-2 QNPs, children initially treat the QNPs as being Type 2, which is the most
restrictive option.
In their experiments Musolino (1998) and Musolino, Crain and Thornton (2000) do not
control for stress (except in the case of some indefinites). In the case of the some
indefinites, the word some is stressed fully to maintain the quantifier interpretation.
4.1.3 Challenges
M usolino’s (1998 and subsequent work) analysis can account for the English data and it
explans the acquisition path. However, it is challenged in cross-linguistics grounds. As
will be seen in Section 4.2 below, children unexpectedly behave in a similar way i.e.,
they have narrow scope of objects with respect to negation, when acquiring an SOV
language, where negation follows the object as well (Lidz & Musolino, 2002). In Dutch,
as will be discussed in the following sections, high objects can have narrow scope
reading although they precede negation and adverbs (Kramer, 2000).
119
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Moreover, Su (2001) observes that English speaking children do not treat double-object
sentences such as The Mermaid gave a lady every flower isomorphically. Under the
isomorphism hypothesis they should have a wide scope of a lady over every flower due to
the linear ordering. This is not borne out in the experiments. Su (2001) reports that only
44% of the children had the wide scope reading of a N (adult preference for wide scope
was 94%). Su’s findings can be attributed to the difference between QNP-QNP
interaction versus QNP-negation interaction, or the property of the indefinite article, as
she argues. No matter what the reason behind the behavior of children is, the findings are
problematic for a strict isomorphism account.
In the following chapters, I will argue that the reason why children have a non-adult-like
reading of the sentences does not have anything to do with isomorphism. Rather, it is just
as a matter o f coincidence that the reading that children have corresponds to the linear
ordering of constituents in English.
4.2 Linear versus hierarchical ordering
One of the predictions of the isomorphism hypothesis was that when negation and a
quantified NP are reversed, the scope assignment of children should be reversed too. So
it is predicted that in those languages where the quantified objects precede negation,
children should be able to assign wide scope to objects. To test this prediction, Lidz and
Musolino (2002) conducted experiments on four-year-old English and Kannada speaking
120
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
children (24 children for each language). Kannada is an SOV language, where the
objects precede the negation, which is embedded in the verb complex. Therefore in terms
of its word order it provides a contrast with English as seen in (178) ((25) in Lidz &
Musolino, 2002).
(178) Naanu eraDu pustaka ood-al-illa two books... neg (Kannada)
I-nom two book read-inf-neg
‘I didn’t read two books’ neg... two books (English)
Lidz and Musolino (2002) found that the prediction of the isomorphism hypothesis is not
borne out. In contrary, children speaking Kannada behaved like the English speaking
children in their scope interpretation. Both English and Kannada speakers accepted the
wide scope reading of negation in 81% and 75% of the cases, and rejected a wide scope
reading of the indefinite NP in 65% and 80% of the cases (English and Kannada,
respectively). Based on the findings, the authors concluded that isomorphism, which is
based on the linear ordering of constituents, was not the right way to go, rather the
findings suggest that it is the hierarchical ordering of constituents that determined
children’s interpretation. In English, negation precedes and also c-commands the object,
and in Kannada, verb+negation complex follows the object, but still c-commands it. The
results of the study indicated that children relied on the c-command relationship between
the indefinite object and negation, rather than the linear precedence for scope
interpretation. In other words, children cannot do quantifier raising and thus the objects
cannot scope out of negation.
121
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Lidz and Musolino (2002) discuss various reasons that may be behind children’s non
adult-like behavior. One of them is the lack of grammatical knowledge of one of the
operations that is responsible for the interpretation of the wide-scope indefinites (c.f.
Kramer, 2000). The other one is the restrictions on children’s language processing. They
do not propose any analysis or a conclusive reason that could capture children’s
behavior. Nor do they present any evidence to show that it is the c-command relationship
that determines children’s behavior. Rather, their conclusion is based on the parallelism
o f the c-command relationship and the results, and the lack of such a correlation between
the linearity and children’s scope interpretations. It is possible that children’s behavior is
determined by an independent factor, which, as a matter of coincidence, parallels the
hierarchical similarity and the linear contrast.
In later work, Lidz et al. (2003) show that children do not have a problem with quantifier
raising. They test other structures which involve QR, but does not require an interaction
with negation and see that children behave adult-like in such structures, displaying a
competence for QR in their grammar. Lidz et al. (2003) conclude that the difficulty
children have in QNP-negation structures then should be attributed to some other kind of
difficulty. Later Viau, Lidz and Musolino (2005) argue that children’s comprehension is
effected by both discourse disadvantages and processing factors. Their proposal of
difficulty with processing is supported by Trueswell et al. (1999), who show that five-
year-old children do not take into account relevant discourse/pragmatic principles when
resolving temporary ambiguity in sentences such as ‘put the frog on the napkin into the
122
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
box’ (versus the unambiguous form ‘put the frog that’s on the napkin into the box’) and
they showed little or no ability to revise their initial parsing commitments. In contrast,
adults pay attention to the discourse constraints and revise their incorrect parsing
commitments. Trueswell’s et al. (1999) study does not deal with QNPs and whether or
not the QNP-negation interaction will have a similar conclusion has not yet been
investigated.
4.3 Discourse Integration
Another series of experiments was conducted on the acquisition of specificity (and scope)
in the Dutch language (Kramer, 2000, Schaeffer, 2000). Although these studies have
different analyses of the target adult language, and the paths children take to reach adult
grammar, their main conclusion is the same: it is ‘discourse integration’ what presents a
challenge to children. In this section, I will present what ‘discourse integration’ means in
each account, why it is difficult for children and how Dutch children master it in the end.
In Dutch, specificity or scope marking is achieved through word order variations .
Specific objects obligatorily scramble to the left of adverbs and negation (179) while
non-specific objects remain to the right, adjacent to the verb (180). Studies done on
Dutch have shown that children have a narrow scope reading of high indefinites (those
3 2 Specificity marking through word oder variations is challenged by Ruys (2001). I will discuss this issue
later.
123
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
that obligatorily have wide scope in adult speech) at a particular time in language
development. Children’s interpretation of low indefinites, however, is observed to be
adult-like at an early age. The so-called high objects are higher than the adverbs and the
negation in the hierarchical structure. They also precede the adverbs and negation in the
linear ordering. It is interesting to note that they still have narrow scope in children’s
33
speech, which is not predicted by the isomorphism and c-command accounts .
(179) Ik wil een paard even vasthouden. [high/specific]
I want a horse PART hold
‘I want to hold a (particular) horse for a minute.’
(180) Ik wil even een paard vasthouden. [low/non-specific]
I want PART a horse hold
‘I want to hold a horse for a minute.’
While the accounts of Dutch data vary with respect to the syntactic structure that they
adopt for Dutch and the details of the learning pattern, they all attribute the findings to
the late acquisition of the “discourse integration” rule which requires children to integrate
what they experience here and now to the events in the past (any object versus, one of the
3 3 Lidz (p.c.) notes that indefinites that Kramer (2000) tested are not quantifiers, thus they are not relevant
to the c-command account. According to Gualmini (2005), Kramer (2000) has a methodological flaw and
her results do not necessarily represent Dutch child grammar.
124
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
objects that were introduced earlier in the discourse). Thus, it is maintained that
children’s behavior in scope assignment is determined by the phases they go through in
pragmatic development, or pragmatics-syntax/semantics interface. In this section,
representing the discourse integration hypothesis, Schaeffer’s (1997; 2000) and Kramer’s
(2000) analyses of Dutch child data and the adult Dutch grammar that they each adopt are
presented and discussed. Schaeffer’s results are mostly based on the interpretation of
definite objects, but they have implications for the interpretation of object referentiality in
general. Kramer’s results are based on indefinites only.
4.3.1 The concept of non-shared knowledge
4.3.1.1 Schaeffer’s Dutch grammar
Following Szabolcsi (1983) and Abney (1987), Schaeffer (1997, 2000) assumes that
referential expressions are DPs and takes the overt determiner to be the morpho-syntactic
expression of referentiality. In her analysis, the morpho-syntactic marking of
referentiality and the [referential] feature coincides in adult speech, but they can be
disassociated in child language. In other words, the presence of a morpho-syntactic
marker for referentiality does not necessarily imply that a nominal bears the [referential]
feature. Proper names, that are refential, start at N and move to D, and in that way the
referentiality can be expressed. Following Koopman (1998) she assumes that pronouns
are inherently referential and they have person and number features. The indefinites with
referential interpretation are DPs, just like definites. Parallel to the definite determiner,
the indefinite determiner occupies the D position and it is the realization of referentiality.
125
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The non-referential indefinites are NumPs and in such nominals the indefinite determiner
occupies the Num position rather than the D position. Thus when DP is projected the
nominal is referential, when it is not projected it is non-referential. She further proposes
two different referential nominals: Discourse-related and non-discourse-related. An
example for a non-discourse related nominal is “the sun,” which does not need to be
mentioned in the previous discourse to be identified by the hearer. The other referential
nominals (e.g., “the tree”) need to be related to the previous or past discourse to be
identified, so they are discourse related. Following Sportiche (1992), Schaeffer proposes
two different syntactic positions for discourse related and non-discourse related objects,
thus she proposes that discourse relatedness is syntactically marked. The projection that
hosts the discourse-related nominals is called DiscP (Discourse Phrase) and the one that
is occupied by the non-discourse-related objects is called RefP. DiscP is higher than RefP
in the hierarchical structure and they are both above negation (181).
In summary, in Schaeffer’s analysis, the non-referential objects are NumP and they
appear below negation. If the objects are referential, they have the DP projection and
their landing site above NegP is determined by their discourse relatedness. If they are
understood to be referential because they have an antecedent in the previous discourse,
they are represented as referential and discourse-related and move via SpecRefP to
SpecDiscP. In SpecRefP their referentiality is licensed and in SpecDiscP, their discourse
relatedness is licensed. Therefore movement to a particular position is obligatory to have
a particular interpretation. Under this analysis, the adverbs can appear either immediately
126
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
above VP (low adverbs) or immediately above TP (high adverbs), which is a position
below DiscP and above RefP (181).
(181) .. .[AgrSP [DiscP (high adverb) [TP [RefP [NegP [AgrOP (low adverb) [VP
[DP V ...]]]]]]]]
In this analysis, the NegP projects above AgrOP and its position is fixed while the objects
remain either below it or moves above. Similarly, the adverbs appear in fixed positions.
Low adverbs appear to the right of negation, while high adverbs are higher, to the left. In
her work, she studies the high adverbs and negation, as the object scrambling that is
driven by the [referential] feature is over negation and high adverbs, rather than the low
adverbs. The movement over a low adverb is done for case reasons.
4.3.1.2 Non-discourse related interpretation of objects
Schaeffer (1997; 2000) looks at the acquisition of objects in low and high positions. She
includes definite DPs, proper names, personal and demonstrative pronouns and indefinite
objects in her study. She conducts truth value judgment tasks and elicited production
tasks on 49 children between the ages 2;4 and 6;10.
She observes that Dutch 2-year-olds (mean age: 2;8) move specific preferential) objects
(definite DPs and proper names) over negation 30% of the time (this movement is
obligatory for adults). They move personal pronouns and demonstrative pronouns over
negation in 33% and 17% of the cases respectively. With indefinite objects, they behave
127
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the same, move the indefinite objects over negation 33% of the cases. Their task with
indefinites is more complicated though, as they needed to make a distinction between
referential and non-referential indefinites. The interesting finding is that even the adults
moved the indefinite object in 66% of the cases, so adults had a difficulty with
identifying the referentiality of the indefinite objects as well. Based on this “unexpected”
result, Schaeffer (2000) bases her analysis of referentiality on the results of the definite
objects. It should be noted though, that the difference between adults and children are
reported to be significant in indefinites although the adults have an unexpectedly low
score (Schaeffer, 2000: 72). According to this account, a nominal expression is marked
for specificity only if the distinction between discourse and non-discourse relatedness is
made. If the object is not marked for specificity (as it is the case for 2 year olds), it does
not rise. In the case of an overt object, this results in non-scrambled object, i.e. the object
remains below NegP.
Schaeffer (2000) suggests that the reason why 2-year-olds do not always mark a noun for
specificity is the fact that they have not acquired the pragmatic discourse rule which
requires the child to take into account the interlocutor. She refers to this rule as ‘concept
of non-shared knowledge,’ stated in (182) ((18) in Schaeffer, 2000), which she proposes
to be a pragmatic property.
(182) Concept of non-shared knowledge
Speaker and hearer knowledge are always independent.
128
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
She proposes that 2-year-olds do not have the concept of non-shared knowledge,
therefore they do not mark referentiality. Therefore, in children’s grammar, referential
objects do not move up to DiscP. According to her findings children at the age of 3;0
perform adult-like with respect to direct-object scrambling.
With respect to the adverbs, children’s behavior is assumed to be the same. The youngest
children in the 2-year old age group do not have any adverbs in their speech. The three
year old subjects however, are reported to be adult-like. In this part of the analysis, too,
Schaeffer’s results are mostly based on definite objects, rather than the indefinites.
Schaeffer states that in child speech, the presence of the referentiality marker does not
imply referentiality. Therefore an object with a definite determiner can be non-referential.
She does not discuss whether the absence of the marker obligatorily means non-
referentiality. In her analysis, those that bear referentiality marker can be referential or
non-referential which means that the marker is not a referentiality marker for children. If
there is no referentiality marker, then those that do not bear the marker (bare nouns for
example, if they exist in early grammar) should be able to have referential interpretation
in child speech. This issue is left undiscussed.
The discussion of referentiality is then based on the nominals that bear a referentiality
marker in adult speech, namely definite or indefinite nominals. It is interesting to see that
children’s treatment of pronouns is the same as the treatment of definite or indefinite
objects, namely they are not moved above negation in 2-year old children’s speech either
129
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(Schaeffer 2000: 82). According to Koopman’s analysis of pronouns, however, they
should inherently have referentiality. If they are inherently referential, they should always
appear to the left of negation. This does not happen. So the questions of what kind of
interpretation the pronouns have in child speech and whether inherent referentiality holds
for children or not remain to be issues that are not addressed. According to Schaeffer,
they are referential but they are not “marked” for referentiality and this is what prevents
movement. If they are referential and despite that they do not move, the link between
referentiality and movement requirement is weakened.
It is actually possible that referentiality (or specificity) and scrambling are not
interdependent, just as I argued for Turkish case marking. Objects scramble for
independent reasons as also proposed by Ruys (2001). Schaeffer (p.c.), too, states that the
‘a certain’ indefinites do not have to scramble in Dutch. If this is actually the case, what
children are doing is not that non-adult-like. In child grammar, just as it is the case in
adult-grammar, specific objects do not have to move.
Another important comment to be made about Schaeffer’s study is that it is a study on the
acquisition of definiteness rather than referentiality or specificity. As reported in the
study, even the performance of the adults were about 40% correct (or expected) in the
experiments on referential versus non-referential readings and this is attributed to a
possible methodological flaw. Therefore it is possible that what was mastered by the age
of 3:0 was definiteness, rather than referentiality or specificity. The results of the study
by Kramer (2000), which will be discussed in the next section, confirms this argument by
130
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
showing that children do not have an adult-like comprehension of referential indefinite
objects up to the age of 6;0.
Schaeffer’s (2000) study is not a comprehension study, so the results cannot be directly
compared to the results of the English studies discussed earlier although it has similar
results. It is important to note that the conclusion that it has, namely that the children’s
difficulty can be attributed to a pragmatic difficulty, is consistent with the later
comprehension study on Dutch (Kramer, 2000). It is also very much in line with the
position I will be defending in the following chapters.
4.3.2 Non-integration hypothesis
4.3.2.1 Kramer’s Dutch grammar
Kramer (2000) looks at the comprehension of high (183) and low indefinites (184) in
Dutch at experimental settings. She reports that high indefinites are interpreted like low
indefinites by her subjects at an early age of development and confirms the earlier
findings of Schaeffer (2000) that children do not mark specificity by movement. The two
proposals are very different though. Kramer (2000) proposes that the high indefinites take
narrow scope interpretation just like the low indefinite objects. In Schaeffer’s account,
however, movement is not obligatory, but objects can have specific/wide scope
interpretation regardless of movement.
131
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(183) Ik wil een paard even vasthouden. [high/specific]
I want a horse fo r a minute hold
‘I want to hold a (particular) horse for a minute.’
(184) Ik wil even een paard vasthouden. [low/non-specific]
I want fo r a minute a horse hold
‘I want to hold a horse for a minute.’
Kramer adopts van Geenhoven’s (1998) semantic incorporation theory as the target adult
grammar. In this account indefinites are categorized as either predicative objects or free
variables. The predicative objects occur in low object positions and are semantically
incorporated to verbs. By such an incorporation operation, they are absorbed by the verb.
Thus, verbs combining with low indefinite objects have an intransitive argument
structure, and bear intransitive morphology in languages such as West Greenlandic where
intransitivity is marked overtly, whereas those verbs that combine with high indefinite
objects remain transitive. Predicative objects always have narrow scope with respect to
negation and adverbs. In this account most verbs have two entries in the lexicon. The
indefinites that appear in higher positions are interpreted as free variable indefinites.
Unlike the predicative nominals they have a variable that must be bound through an
accommodation mechanism, which locates the variable of the indefinite in a context. The
free variable indefinites are insensitive to the operators that affect the verb and always
have wide scope with respect to negation and adverbs. In this account, although the
lexical content appears in a low position in syntax, it may be bound at a high level. They
132
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
have an interpretation as long as they are accommodated in discourse, that is, when they
have a pre-established reference in discourse.
To be able to interpret the indefinites, a hearer applies a bridging mechanism (based on
Clark, 1977), through which s/he relates the utterances to previous discourse and
identifies the accommodation site of an indefinite. The speaker has to provide sufficient
information to the hearer so that the hearer can construct a serious o f inferences based on
what the discourse provides. The inferences guide the hearer to the accommodation site
of the indefinite, that the speaker intends.
4.3.2.2 The non-integration hypothesis
According to the non-integration hypothesis of Kramer (2000) stated in (185), children
acquire the predicative interpretation in indefinites early. The free variable interpretation
is acquired later because it requires discourse integration. Because the interpretation of
the high indefinites is not consistent with the surface syntactic position of the NPs, she
concludes that her findings support van Geenhoven’s (1998) semantic incorporation
account of indefinites.
(185) Non-integration hypothesis
Children acquire the predicative interpretation of indefinites early. The free
variable interpretation is acquired later because it requires discourse
integration.
133
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In her interpretation of indefinite objects in child grammar, she focuses on the semantic-
type error in which high indefinites are interpreted as low indefinites (predicates) instead
of free variables, i.e. A particular horse is interpreted as any horse. In van Geenhoven’s
(1998) mechanism, there is no direct association between syntax and the semantic
interpretation of objects. In this analysis a high position does not necessarily imply a high
reading. Objects in the high positions can be semantically incorporated and have
predicate interpretation. According to Kramer, this is what is happening in child speech.
In the experiments on direct objects she looked at (i) direct objects and negation and (ii)
direct objects and the use of the adverb twice. High versus low indefinite objects were
tested with children in the age groups 4;0-8;0 through a comprehension experiment.
Subjects were provided with scenes where one out of three objects were not acted upon.
For example, an agent catches two out of three fish. The example sentences are presented
in (186) and (187).
(186) De jongen heeft g[een vis] gevangen.
‘The boy did not catch any fish’
(187) De jongen heeft [een vis] niet gevangen.
‘The boy did not catch a particular fish’
In negation experiments, in the low condition 100% of the child responses were adult
like. In high condition only 16% of the child responses were adult-like. That is, while
134
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
children were treating low indefinites in an adult-like fashion, their interpretation of high
indefinites signaled a semantic type error. There is observed to be improvement by age.
In the the twice experiments, in the low condition, 93% of the responses were adult like.
In the high condition, 39% were adult like. In summary, the low indefinites took narrow
scope for all children, as they did for adults. The high indefinites also took narrow scope
for many children but never for adults.
The results of these experiments suggest that high indefinites can be interpreted as
predicates, which do not need to be accommodated and do not need discourse integration.
As opposed to the adult grammar adopted by Schaeffer (1997, 2000), Kramer’s (2000)
adult grammar does not require a close association between syntactic position and
interpretation. Kramer’s (2000) results are not expected under Schaeffer’s proposal,
because in her system once the object moves to a higher position, it is already marked for
specificity and objects can have a specific interpretation without movement34. The results
in Kramer (2000) are not predicted by the isomorphism account either because high
objects, although they precede negation and adverbs have narrow scope with respect to
them. Similarly, a hierarchy/c-command-based account cannot capture the results
because high objects are apparently higher in the hierarchy as well.
3 4 Schaeffer (2000) does not talk about indefinites but does not state that her account excludes indefinites,
or indefinites should be analyzed in a different way.
135
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
According to Kramer (2000), the finding that the interpretation o f high indefinites by
children does not correspond to their syntactic positions is accounted for under van
Geenhoven’s (1998) semantic incorporation analysis as in this analysis the syntactic
position and the interpretation are not directly relevant. Therefore, according to Kramer’s
analysis of low and high objects, the interpretation of objects is blind to their syntactic
positions. Then the questions of why the children do not ever interpret low objects as
specific and why they have to incorporate remains undiscussed. If there is no relationship
between the syntactic position and interpretation, this is a logical possibility.
The original semantic incorporation analysis of van Geenhoven (1998) is based on the
weak-strong case distinction. That is, those objects that have weak case can be absorbed
by the verb. Schaeffer (1997, 2000), however, shows that pronouns can appear in the low
position too, therefore an analysis of the low objects should include pronouns as well.
Incorporation of pronouns is not possible under the definition of semantic incorporation,
as they are referential by definition and they do not have weak case. Therefore the
behavior of children needs a more complicated account than a simple semantic
incorporation account. Moreover, Kramer (2000) observes that the same kind of error is
seen in nominals in the subject position, too. Although van Geenhoven (1998:178) states
that predicate subjects can incorporate, none of the examples she gives are unergative
verbs. Therefore it is not clear whether her analysis allows agent incorporation.
No matter what the target structure is and how children acquire it, based on what we
know about Dutch, (Schaeffer, 1997, 2000, Kramer, 2000), English and Kannada (Lidz &
136
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Musolino 2002), assuming that language acquisition follows a somewhat universal path,
we can conclude that children are late in the comprehension of wide-scope objects and
the difficulty they have can be attributed to a non-adult-like ‘pragmatic’ component and
discourse difficulty. This seems to account for the behavior of the children acquiring the
languages discussed so far. If the results are attributed to a pragmatic difficulty we
should expect it to be a universal difficulty as pragmatics is not language-specific.
However, cross-linguistic studies that focus on similar structures showed that children’s
treatment of scope interaction in other languages is not the same as those in Dutch. As
will be discussed in the next sections, Su (2001) and Miller and Schmitt (2003) among
others show that children acquiring languages such as Chinese and Spanish do not have
difficulty with wide scope reading of the indefinites.
4.4 Cardinal reading of indefinite articles
In both English and Dutch, both the specific and non-specific interpretations of indefinite
objects are achieved by nominals that have an indefinite article. In Dutch the difference
between the indefinite article a and the cardinal number one is marked with a stress
variation, in English the difference is marked with two different lexical items. In this
section, I will report experiments that are conducted on Mandarin Chinese and Spanish
spoken in Chile, where the narrow scope is expressed with bare objects and those
nominals that have a cardinal number/ indefinite article is ambiguous with respect to
scope.
137
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.4.1 Chinese yi-ge and Lexical Factor Analysis
4.4.1.1 Su’s Chinese Grammar
Chinese does not have overt definite or indefinite articles that correspond to English the
or a. According to Su (2001), bare noun phrases can be definite, indefinite or generic.
Indefmiteness is expressed by numeral phrases. As seen in (188), yi-ge can refer to a
single entity or it can have an existential interpretation {at least one) (189), or can have a
free choice meaning (190). In other words, yi-ge can have exactly one interpretation, but
in contrast to the English one it does not have to mean exactly one (Su, 2001:111-112).
(188) wo jintian mai-le yi-ge huaping,keshi houlai meimei ba huaping dapuole.
I today buy-ASP a-CL vase but later sister BA vase break-CL
‘I bought a vase today, but later (my) sister broke the vase’
(189) he yi-bei niunai zai qu xuexiao
drink a-cup milk then go school
‘Drink a cup of milk before going to school.’
(190) yi-ge hao xuesheng hui zhunshi jiao zuoye
a-CL good student will on-time turn-in homework
‘A good student will turn in homework on time’
138
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.4.1.2 The Lexical Factor Analysis
Su (2001) investigated the way in which Chinese and English speaking children
interpreted the indefinite yi-ge and a respectively. A series of experiments are conducted
using the Truth Value Judgment task (Crain & Thornton 1998). In one o f the
experiments, which is the most relevant to the present study, she tested the hypothesis
that at an early stage of development, English speaking children interpret a as a free
choice any, whereas Chinese speaking children consider yi-ge to mean exactly one. The
difference between any interpretation and exactly one interpretation is important because
it has an impact on the truth value judgments. For example, they presented sentences
such as the one in (191) to children in two different contexts. In one, Mickey Mouse did
not ride any dogs (narrow scope context), and in the other he did not ride one o f the dogs
(wide scope context). The experiment involved three dogs in total.
(191) Mickey Mouse didn’t ride a dog.
a. There is a dog such that MM didn’t ride (‘one of the dogs’)
b. It is not the case that MM rode a dog (‘any dogs’)
The wide scope and narrow scope readings are given in the following the example in
(191). In the context mentioned above the wide scope reading would be true. The narrow
scope reading in (191)a is equal to the interpretation ‘he didn’t ride any dogs,’ which
would be false in a situation where two dogs (out of three) were ridden. When the
indefinite article reads as ‘he didn’t ride one (=exactly one) dog’, the sentence would be
true again in the same situation, namely when two dogs were ridden, because the number
139
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
of dogs that were ridden is not one. According to Su’s (2001) analysis of English and
Chinese, in English the sentences with a reads as ‘any N ’ while the objects in Chinese
sentences are equal to ‘exactly one N ’ which results in a difference in truth value
judgements.3 5
Su (2001) tests 24 English speaking children between the ages 4;2 and 6;4 (mean age:
5;3) and 27 Chinese speaking children between the ages 4;9 and 6; 10 (mean age: 5;9)36.
The results of the experiment showed that Chinese-speaking children predominantly
accepted the “one of the dogs” (wide-scope) reading (77%), while the English-speaking
children preferred that reading 33% of the time. Su (2001) attributed this result to the
lexical properties of English a, and Chinese yi-ge. Chinese-speaking children at an early
age interpret yi-ge as “exactly one”. In English, however, because there are two lexical
items, a and one, respectively, children distinguish the two readings and understand a N
as a bare N37. Because English speaking children do not attribute any cardinal number
reading to a due to another lexical item that corresponds to the same reading, they do not
3 5 It is important to stress that this is just a shortcoming of the experimental methodology. In the Chinese
case, the value judgement is the same as the value judgement of wide scope reading, whereas in English it
is the same as narrow scope.
3 6 Note that Chinese children are 6 months older than the English children. Six months is a big difference in
the period of early development, so the results may be influenced by the age difference.
3 7 Note that this interpretation is in line with Slobin’s (1973) one form, one function hypothesis which
states that in early child grammar, each form has only one function. Because there is another form ‘one’ in
the language children do not attribute ‘one’ reading to ‘a’ in English. The same logic works in the opposite
direction in Chinese: Because there is an option of bare objects, children do not attribute the narrow reading
to objects with yi-ge.
140
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
understand the sentence as ‘one o f the dogs was not ridden.’ Su (2001) formulates her
hypothesis as in (192).
(192) Lexical Factor Analysis
Children’s non-adult interpretation on quantifier scope assignment results
from a lexical idiosyncracy between English a, and Chinese yi-ge. At an early
stage of development, English speaking children analyze the indefinite a as
meaning any (i.e., non-specific), whereas Chinese-speaking children consider
yi-ge to mean exactly one (i.e., specific).
4.4.2 Spanish una and the scope of bare singulars
The lexical factor analysis of Su (2001) finds support in cross-linguistic works. Miller &
Schmitt (2003) look at children’s interpretation of indefinites in Spanish spoken in Chile
from this perspective. Following a similar methodology they test children’s and adults’
interpretation of indefinites such as (193) and (194). In the first one the object has the
indefinite article/numeral una, while the second one is a bare object without una. In
adult speech, (193) is ambiguous between the wide and narrow scope readings, while
(194), which has a bare nominal, allows only a narrow scope interpretation of the object.
What is interesting for Spanish is that, the indefinite article una is homophonous between
one and a, just as it is the case in Chinese.
(193) El ninito no trajo una pelota. neg > una pelota; unapelota >neg
The boy neg bought a ball
'The boy didn’t bring any balls.' / 'The boy didn’t bring one of the balls'
141
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(194) El ninito no trajo pelota. neg > pelota; *pelota > neg
The boy neg brought ball
'The boy didn’t bring any balls.' / *'The boy didn’t bring one of the balls'
According to the analysis of Miller and Schmitt (2003), the objects that have una are
marked for number while the bare objects are ‘featureless. ’ In terms of their scope
taking properties they are similar to the incorporated objects of West Greenlandic as
analyzed in van Geenhoven (1998), polarity indefinites in St’at’imcets reported by
Matthewson (1999), bare nominals in Hungarian (Farkas & Swart, 2003), and genitive
objects in Russian (Neidle, 1988). According to the analysis of Masulla (1992), the bare
singulars in Spanish do not have the full DP projection, rather they are NPs and they
incorporate to verbs to satisfy the Visibility Condition. They are restricted to the object
position, which is considered to be an evidence for incorporation, and they always
receive the narrowest scope.
Miller and Schmitt (2003) hypothesize that if children cannot assign wide scope to
indefinites (cf. c-command and non-discourse-integration accounts) then, they should
have the same interpretation of bare objects and objects with the indefinite
article/numeral. If, however, the proposal of Su (2001) holds cross-linguistically, then
considering the cross-linguistic similarities between Spanish and Chinese bare objects,
they predict that children will have a narrow scope interpretation of the bare objects, but
not the other object type.
142
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
A Truth Value Judgment task was conducted on 24 Spanish speaking children between
the ages 4;5 and 5;11 (mean 5;0) and 30 adults. The testing stories involved four objects,
three of them were acted upon and one was left behind. A boy decided to take three out
of four balls, for example. The test sentences (stated in (193) and (194) above) were
presented to subjects. A rejection indicated a narrow scope reading of the sentences.
Adults and children rejected the bare object sentences in 99% and 76% of the cases,
respectively. In the case of the una N objects, the rejection response was recorded in 43%
and 36% of the cases for adults and children respectively. Thus, the results of the study
showed that Spanish-speaking children, just like the Spanish-speaking adults, can
distinguish between the obligatory non-specific reading of bare nominals and the
optional specific reading of the una N nominals. Miller and Schmitt (2003) argue that
children are accessing a number reading of the nominal with una due to its homophony
with the numeral one; thus they conclude that their results support Su’s (2001) Lexical
Factor Hypothesis.
4.4.3 Response ambiguity problem
As Miller and Schmitt (2003) also observe, the cardinal reading and the wide scope
(specific) reading result in the same right or wrong response in the experiments. When
there are more than two objects, and one of them is not acted upon, the readings ‘it is not
the case that one object was acted upon’ (exactly one reading) and ‘there is an object such
that it was not acted upon’ (wide scope reading) are both true. The narrow scope reading
‘it is not the case that an object was acted upon’ is false. Therefore, when a child accepts
143
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
a sentence, it is not clear whether s/he is accepting the number reading or the wide scope
reading, both, as a matter of experimental design, correspond to the ‘yes’ response. Miller
& Schmitt (2003) and Su (2001) both attribute the adult-like wide scope reading to the
number interpretation of the structures. However, the two readings overlap only because
of the shortcomings of the methodology and the results do not necessarily say anything
about children’s wide scope interpretation. The ambiguity that results from the
interference of the number reading only blurs the results. Although it may be true that the
number reading may have an effect on the interpretation of wide scope indefinites, the
results of the experiments reported by Su (2001) and Miller and Schmitt (2003) do not
necessarily show that children have an early access to the wide scope reading.
Moreover, Su (2003) further tested four-year-old children and adults to see whether
Chinese-speaking children rely on the c-command relationship to interpret the
ambiguous sentences such as the one in (195). She used objects that have two, three etc.
instead of yi-ge to eliminate the exactly one reading.
(195) Xiaogou meiyou zhuadao liang-zhi tuzi
dog didn’t catch two-CL rabbit
‘The dog didn’t catch two rabbits’
For a negation wide-scope reading, children accepted the sentences 35% of the time and
adults 72% of the time (difference: p<.05). For a QNP wide scope reading, children
accepted the sentences 63% of the time, and adults 35% of the time (again different at
144
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
p<.05). The results showed that unlike English and Kannada adults who accepted the
wide scope readings over 85% o f the time, Chinese-speaking adults did not prefer a wide
scope reading of indefinites. In contrast, Chinese-speaking children interpreted the QNPs
as having wide scope over negation, revealing that they do not follow a c-command
principle. Su (2003) attributed the behavior of the children to a misinterpretation of the
structure, more specifically, she argued that children interpret the nominal two rabbits as
THE two rabbits, assigning a null definite article to the structures. The results of this
experiment also showed that the conclusion reached for the earlier Su (2001) study
cannot hold. If what was guiding children’s interpretation were the number reading o f yi-
ge, and its contrast with the English a, we would not have the same result with structures
that do not contain exactly one reading. The results, in contrast, show that children do
have wide scope reading independent of the cardinal number reading.
In summary, we have Dutch, English, and Kannada on the one hand, as having narrow-
scope reading of the wide-scope indefinite objects in children due to discourse
integration and c-command reasons. On the other hand, we have languages such as
Chinese and Spanish, which, allow an early adult-like wide-scope interpretation of the
indefinite objects. In the next section, I will present the predictions for Turkish based on
the earlier findings on scope and the structure of Turkish.
145
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.5 Predictions for Turkish
Considering the adult structure that I adopted, here are the hypotheses regarding the adult
grammar: Adults’ accusative marked indefinites will have wide scope with respect to
negation and the adverb twice when they are not adjacent to verbs. (2) Adjacency to the
verb may have an impact as the accusative marked indefinites do not necessarily have
wide scope with respect to other constituents in this position (c.f. Kelepir 2001 and the
analysis I adopted). According to the present account, bir N objects should have narrow
scope with respect to negation and the adverb twice since the deletion of the case
morphology indicates that they are in AgrOP where they cannot scope over negation and
adverbs. In all the accounts I have cited so far and the present account I adopted, adults’
bare N objects have narrow scope. They should never have wide scope as they cannot
interact with other constituents due to the complex predicate formation.
I proposed that the accusative case morpheme is the realization of the formal case feature
on the object. When an object is displaced, the case morpheme cannot get deleted and
displacement takes place when the object has some kind of a preusppositional or partitive
interpretation. On the one hand, it is a formal case marker. On the other hand, its full,
adult-like interpretation requires a complex pragmatic knowledge and an interface of
various components of grammar so that the presuppositionality can be understood. In the
present study, I propose that children have access to the formal properties of the
accusative case but do not have an adult-like interpretation it.
146
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
What we already know about the early child grammar is that children recognize the
accusative case as a formal case morpheme and they know that they cannot delete it when
the object is displaced. Earlier studies show that children never delete the accusative case
in scrambled positions (Ekmekgi, 1986, Komfilt, 1994). Moreover, Slobin and Bever
(1982) observe that children use the accusative case as a clue to detect the object (as
opposed to subject) in a sentence. Therefore, the earlier literature shows that the children
have access to the formal properties of the accusative case, i.e., it is a case marker that
marks the object and that it cannot be deleted when the object is displaced. The
interpretation of the object and thus the case morpheme is something else.
Children do not delete case on the scrambled objects, but do they actually know that only
presuppositional objects can scramble? Do they use the accusative case when they are
referring to presuppositional objects only? Ozcan (1993) reports that children overuse the
accusative case when they are referring to new referents in narrations. Kuntay (1995,
2002) shows that children do not use the non-case marked indefinite forms to refer to
new referents until the age of seven. Ketrez (2004a) shows that when imitating bir N
structures, children add an accusative marker, so overuse the case marker. It is possible
that the overuse of the accusative case is something similar to the overuse of the article
the in English child language, in other words, children do not know that their knowledge
is not always shared by the other interlocuters i.e., they mark the object that are
presuppositional for themselves (c.f. Karmiloff-Smith 1979, Matthewson & Schaeffer,
2000, among others). In the present study, I argue that children’s interpretation of the
147
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
accusative marked objects is not adult-like. Moreover, they do not have access to the
relationship between scrambling, overt case morpheme and the interpretation.
I propose that children have the formal morpho-syntactic properties of the accusative
case. However, they do not have an adult-like interpretation of the accusative marked
indefinite objects up to a later developmental stage. They overproduce the accusative
case because they do not know when they can delete it. I hypothesize that neither the
linear order of constituents, nor the hierarchical position of the objects with respect to
other constituents will have a significant effect on children’s interpretation of the objects.
What is missing is a mechanism to interpret the accusative case. I predict that the
accusative marked indefinites will have a non-adult-like interpretation in all the structures
I study.
Let us now look at the predictions based on the other theories. Table 1 below presents a
summary of the predictions for Turkish. According to the isomorphism account, those
objects that precede negation in linear order will have wide scope with respect to
negation38. If this account works, we should have an asymmetry in the treatment of the
postverbal and preverbal case-marked objects. Moreover, bir N objects that are in the
preverbal position should have wide scope interpretation with respect to negation and
post-verbal objects.
3 8 The discussion o f the predictions is based on the negation-object interaction. I assume that the same
discussion can hold for the adverbs.
148
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 1: Hypotheses and predictions for Turkish
Hypotheses Predictions for Turkish (Scope at the youngest age)
Isomorphism
C-command
Lexical Factors
D-related
Bir N-acc (preverbal) > NEG
NEG> bir N-acc (postverbal)
Bir N-acc (preverbal) > bir N-acc (postverbal)
Bir N, bare N > NEG > bir N-acc (postverbal)
Bir N-acc > NEG > bir N=bare N
Bir N(acc) > NEG (> bare N)
NEG > Bir N-acc=bir N=bare
For Turkish, the c-command hypothesis predicts that the accusative marked indefinites
that are not adjacent to verbs (and high indefinites in Dutch) will have wide-scope
reading with respect to adverbs and negation for children even at the earliest age because
they are in a c-commanding position. The postverbal object especially, which are in a
higher position than the preverbal objects should have wide scope in child speech during
the early ages. The post verbal objects’ position is more salient than the other accusative
marked objects as they are outside the linear verbal domain as well. If the Lexical Factor
hypothesis (Su, 2001) is universal, considering the properties of bir, we would expect
children to have a wide scope reading of the accusative marked indefinites as well as the
bir N indefinites from the early ages onwards.
It is important to note that Turkish has more properties that possibly interact with each
other when compared to other languages. It has both case marking and scrambling, and as
the discussion of the adult grammar suggested, it is not clear which one has more impact
on the interpretation of scope. Moreover, it has both bare and non-case marked
indefinites with bir, so case-marked versus non-case marked distinction would not be
149
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
able to capture the scope differences that are represented by high and low indefinites in
Dutch. In the following chapters, I will show that the accusative marked indefinites do
not necessarily have wide scope with respect to other constituents. Children at the
youngest age group treat all the object types alike regardless of their positions or
morphology. In that sense, the results I will report will be very much similar to the results
of Dutch reported by Kramer (2000). I will, however, have a different account for the
developmental pattern of the indefinite objects.
150
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5 Indefinite objects and the adverb twice
This chapter presents two experiments that test the scope of indefinite objects with
respect to the adverb twice. The first experiment, which is an act-out task, shows that
children’s treatment of accusative marked indefinite objects is not adult-like, although
they have a high preference for a 1-object (wide scope) reading when the object carries a
case morpheme. The second experiment, which is a follow-up to the first one, shows that
the high wide scope reading observed in the first experiment is misleading and the
children allow a narrow scope reading of the accusative marked objects, which is not
possible in adult speech.
5.1 Experiment 1: An act-out task
In this experiment, participants use small objects to act out the sentences such as let the
doll push a chair twice or let the doll push a chair-acc twice. In the former, the child is
expected to push any two chairs, or one chair twice, while in the latter, where the object
bears an accusative case, only one chair can be acted upon. This experiment is used to
test whether or not children have an adult-like comprehension of the sentences and
whether the subjects assign different scope to objects that have a case morpheme. The
sentences that are under investigation have four different forms and two different scope
relations exemplified in (196).
151
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(196) a. Ay§e iki kere bir sandalye / sandalye it-sin. neg > bir N / N
Ayse twice a chair /chair push-PAST
‘Let Ayse push a chair twice.’ (any chair)
b. Ay§e iki kere bir sandalye-yi it-sin. bir N-acc > neg
Ayse twice a chair-acc push-PAST
‘Let Ayse push a particular chair twice.’ (one of the chairs)
c. Ay§e bir sandalye-yi iki kere it-sin. bir N-acc > neg
Ayse a chair-acc twice push-PAST
‘Let Ayse push a particular chair twice.’ (one o f the chairs)
The technique has been used since Chomsky (1969) and is observed to be very effective
in testing children’s comprehension. It is also the method used in Kramer (2000) for
testing the scope of high and low indefinite objects with respect to the adverb twice.
5.1.1 Participants
18 four-year-old (mean age:4;2, age range: 3;0-4;6,) and 18 five-year-old (mean age: 5;2,
age range: 4;10-5;9) children were randomly drawn from three different kindergartens in
Istanbul. 12 university students voluntarily participated in the study as the adult control
group. All the participants belonged to middle class or upper middle class families and
152
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
were native speakers of the standard (Istanbul) dialect of Turkish. Every subject was
tested on all four structures exemplified in (196) above.
5.1.2 Procedure
The participants were provided with small toys and were asked to act out the sentences
that were produced by an experimenter. The toys included kitchen objects that were
placed on a table. What children saw was a complete kitchen with various utensils,
furniture and four dolls standing in the kitchen. On the left hand-side of the kitchen, a
backyard was created with a bunch of flowers, three horses and three goats. Prior to the
experiment the kitchen and backyard objects were introduced to the participants by
saying their names so that the children get familiarized with the objects. If the children
wanted to play with the objects they were allowed to play with them and discover the
toys. Then the experimenter introduced the main character of the experiment, a doll
named Ayse, and the participants were asked to act-out the twice sentences with this main
character.
Each testing session was preceded by a training session, through which the participating
child was tested on counting. The rules of the games were taught during the training
session. Those children who could not count and those who gave irrelevant responses, or
did not want to play the game as instructed were immediately discarded. All the subjects
acted out 6 test sentences of each type (24 in total) and 14 filler sentences. A complete
list o f all the sentences is given in Appendix-1. There were three different versions that
153
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
varied in the sequence of the object types. In one version, for example, children heard the
accusative marked objects first, whereas in the other version, they started with a non-case
marked type. Children and adults were equally distributed to the versions.
When the object in the test sentence was accusative marked, participants were expected
to act upon one single object (one of the objects, acted out twice). When the accusative
marking was absent, their action might involve multiple objects (any object(s)). Those
responses, in which one object is involved, were coded as a 1-object reading. The other
one was coded as a 2-object reading. The results were evaluated according to (i) the
mean scores of each object type in each age group, (ii) the response pattern. In the latter
one, participants were divided into 3 groups according to their response tendency (1-
object, 2-object, mixed pattern). In each phase o f the analysis, (i) children were compared
to adults, (ii) participants’ comprehension of 4 object types were compared, to see
whether case and/or adjacency play a role on their scope interpretation.
5.1.3 Results
5.1.3.1 Mean scores
Figure 1 below represents the age groups and their scope assignment to four object types.
The results based on mean score results are shown according to the age groups in Table
2. What is important in the discussion in this section is the scores of the accusative
marked indefinites as the non-case marked objects are ambiguous. They are grammatical
in both 1-object and 2-object readings. As seen in Figure 1, children’s interpretation of
154
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the four object types is alike. Although they have high wide scope scores, their scores are
still lower than adults’.
Figure 1: 1-object reading in Experiment 1
w 100
0 5
c
' " U
~o
age 4 age 5 adults
■
- bir N-acc twice V
-twice bir N-acc V
-twice bir N V
- twice N V
Age groups
InTable 2, the percentages show the percentage of 1-object reading, that is, the times
when participants included one single object and acted upon this particular object twice.
Table 2: 1-object reading in Experiment 1 (mean scores)
Age 4 (mean: 4;2) Age 5 (mean: 5;2) Adults
bir N-acc twice Verb 3.94 (65.66)* 5.11 (85.16)* 6.00 (100)
(SE=.460) (SE=.241) (SE=.00)
twice bir N-acc Verb 4.33 (72.1) 5.16(86) 5.40 (90)
(SE=.45) (SE=.30) (SE=.41)
twice bir N Verb 3.94 (65.66) 4.38 (73) 4.00 (66.6)
(SE=.46) (SE=.49) (SE=.66)
twice N Verb 3.66 (61) 5.00 (83.33) 4.58 (76.33)
(SE=.62) (SE=.30) (SE=.66)
(*) p< 05 (difference from adults)
The raw scores are based on the mean of 1-object reading responses calculated out of 6.
The percentages and standard errors o f mean are given between parentheses. The first
155
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
two rows of the data show the accusative-marked sentence type results. Then follow the
non-case marked and bare objects. The columns represent three age groups.
A 3 (age groups) X 2 (case) X 2 (bir) X 2 (adjacency) ANOVA was run on 1-object
reading (indefinite > twice) scores to determine which factor effects the scope
assignment. The 3 age groups were 4 year olds, 5 year olds and adults. The two case
conditions were the accusative marked versus non-case marked objects. The bir factor
refers to the two versions that have bir (Bir N-acc and bir N) versus the bare N objects.
Two adjacency versions were the ones in which the object followed or preceded the
adverb twice.
The ANOVA results showed that accusative case (F(l,180)=5.271, j><.05) and age
(F(2,180)=7.783, p<.01) have main effect on the 1-object preference. The presence of bir
(F(l,180)=.790,p=.375), and the object position with respect to twice (F(l,180)=.015,
p=.902), do not have a main effect on the interpretation of sentences. No interaction
effect is observed between the factors. In the following discussion, I look at the scores
across different age groups in order to see how children’s scope interpretation of the
indefinite objects is different from that of adults. Then, within each age group, I look at
subjects’ interpretation of the objects and see how they treat the accusative marked and
non-case marked objects.
Children and adults are different with respect to their interpretation of the scope of
indefinites, only when the indefinite object is non-adjacent to the verb. In all the other
object types, their behavior is the same as adults. When the object is not adjacent to the
156
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
verb, the accusative marked object has a 1-object reading for all the adults. Five-year-old
subjects have this reading 85.16% of the time, while the children in the youngest age
group have the 1-object reading of the indefinite in 65.66% of the cases. ANOVA results
and the Games-Howell post-hoc test results suggest that both children groups’
performances are significantly worse than the adults (F(2,45)=8.638, p<.01 and
F(2,45)=8.638,/K.01, four and five year olds), and the increase observed between the
four-year-olds and five-year-olds is not significant (F(2,45)=8.638, n.s.). When the
adverb is placed above both the object and the verb (hence the twice bir N-acc Verb,
structure) and the object appears adjacent to the verb, children’s responses are not
significantly different from that of adults (F(2,45)=2.033 , n.s.). Adults involve only one
object in their actions in 90% percent of the cases. The child participants have 1-object
reading in %86 and 72.1% of the cases (five and four year olds respectively). Although
the ANOVA results show that adjacency is not one of the factors that effect scope
assignment, we see that children are different from adults only in the case of the non-
adjacent objects. It is very likely that this is due to the decrease of 1-object reading in
adult’ s speech (100% versus 90%) rather than children’s treatment of the adjacent versus
non-adjacent object types differently. A paired samples t-test conducted on adults’
adjacent versus non adjacent objects reveal that adults treat these types differently
(f(ll)=2.679,/?<.05)39.
3 9 When the two accusative marked structures (bir N-acc) are merged, there is a significant difference
between the three age groups (F(2, 93)=8.991,p<.01). In this case, the four year olds are significantly
different from the five year olds and adults (F(2, 93)=8.991, p<.05 andp <.01 respectively) and there is no
significant difference between the five year olds and adults (F(2, 93)=8.991, n.s.).
157
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
As seen in the second half of the table, children are adult-like in their interpretation of
non-case marked bir N objects (F(2 ,45)=.216 , n.s.) and bare objects (F(2, 45)=2.141,
n.s.). Now let us have a look at the accusative marked bir N-acc objects and bir N objects
that both appear adjacent to the verb and that differ only in their case morphology. A
paired samples t-test is run to test whether the accusative case makes a difference in the
interpretation of bir N-acc and bir N structures and it is observed that neither 4-year-olds
nor 5-year-olds have significantly different scores (f(17)=.789, n.s. and ?(17)=1.613, n.s.)
in two sentence types. In other words, the accusative case does not have an impact on the
interpretation of the objects, i.e., accusative marked objects do not necessarily have wider
scope in child speech in adjacent position. In adult’s speech, in contrast, these case-
marked and non-case marked objects are treated differently (7(1l)=2.679,/?<.05).
In summary, adults have a wide scope reading of the accusative-marked indefinites and
case appears to be a factor that affects scope assignment in the experiments. While all the
adults have only a wide scope reading of the object when the object is scrambled position,
this interpretation falls down to 90% when the accusative marked object is adjacent to the
verb and the adverb occurs above the object. Children (both age groups) allow the 2-
object reading in accusative marked indefinites in a non-adult-like fashion. Overt case
morphology does not have an impact on their scope assignment. There has been no effect
of bir “a/one” on the interpretation of the scope of the indefinite as the participants do not
treat bare objects and the bir N(acc) objects differently.
158
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5.1.3.2 Response patterns
In section 5.1.3.1 above, I reported the results according to the mean scores of each age
group. When we have a closer look at the individual children’s results, we see that there
are different populations of children in terms of their response patterns. In other words, in
each age group, some children have a strong tendency for 1-object reading, while others
prefer 2-object reading in the majority of cases. In this part of the analysis, I look at the
results according to children’s response pattern. I group those children who have a
preference for a particular reading in 5 or 6 of the cases together. Those that have an
equal number of cases (3-4 1-object and 3-4 2-object reading) are marked as mixed. Then
the 1-object reading scores are compared with a combination of 2-object and mixed
pattern subjects. Table 3 and Table 4 below presents the number of subjects in each
sentence type and age group. The numbers between parentheses are the percentages.
Table 3: Response pattern in Experiment 1 (adults)
type 1-object 2-object & mixed 2-object mixed
bir N-acc twice V 12 (100%) 0 0 0
twice bir N-acc V 11 (91%) 1 (8.3%) 1 (8.3%) 0
twice bir N V 6 (50%) 6 (50%) 2(17% ) 4 (33%)
twice N V 7 (58%) 5 (42%) 0 5 (42%)
As seen in the table, 100% of the adult subjects (n=12) have 1 object reading when the
object is accusative marked and not adjacent to the verb. When the object is adjacent, one
out of 12 subjects has 2-object reading in a majority of the instances. Therefore, adults
show a preference for the 1-object reading and their response pattern is not different with
two different adjacency conditions (Fisher Exact Test, two tailed, p=\.Q). In their
159
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
interpretation of non-case marked objects 50% (n=6) of the adults prefer 1-object
reading, the rest have either a mixed (n=4) or a 2-object pattern (n=2). When the object is
a bare noun, 58% (n=7) o f the adults have 1-object pattern, the rest have a mixed pattern.
Interestingly none of the adults have a 2-object pattern with bare nouns. While there is no
difference between the adults’ response patterns in two non-case marked objects (twice
bir N Verb versus twice N Verb) (Fisher Exact Test, two tailed, j?=1.0), there is a
difference trend in their treatment of the case marked (twice bir N-acc verb) versus non
case marked objects (twice bir N Verb) (Fisher’s Exact Test, two tailed, p=.06). In other
words, bir does not affect the interpretation, but case morphology has an effect trend.
Note that both readings (1-object and 2-object) are grammatical when the object is not
case marked, whereas only the 1-object reading is expected when the accusative marking
is present. The results of adults confirm the results of the score analysis.
Next I turn to children’ response pattern. Table 4 below shows the results of the children
subjects in two age groups. The response pattern analysis reveals that children do not
prefer a 2-object (narrow scope) reading, but rather the difference between adults and
children results from a mixed response pattern that they mostly have. In the younger age
group, only two subjects had a 2-object pattern in the sentence type bir N-acc twice V. In
the other accusative sentence type, only one subject has a 2-object pattern. A similar
picture is seen in the other age group. Only a small number of children prefer 2-object
reading, which is ungrammatical in adult grammar. The majority of the subjects either
give mixed responses or they have the 1-object response. When we compare their 1-
object versus 2-object responses, leaving out the children that have a mixed pattern, we
160
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
see a clear preference of the 1-object reading over the 2-object reading (Fisher Exact
Test, two tailed, p<. 01). Although children’s response pattern is similar to that of adults,
based on this picture, we still cannot conclude that the children are adult-like because
only 44 % of the children in the youngest age group (n=8) and 67% o f the five-year old
children (n=12) have only the 1-object reading when the accusative marked object is not
adjacent to the verb. Note that this is the only reading available to the adults. That is,
children are still non-adult-like when they are compared to adult groups’ 1-object reading
scores.
Table 4: Response pattern in Experiment 1 (children)
type age 1-obj. 2-obj & mixed 2-obj. mixed
bir N-acc twice V 4 8 (44%)* 10 (55%) 2(11% ) 8(44%)
5 12 (67%)* 6 (33%) 0 6 (33%)
twice bir N-acc V 4 10 (55%)* 8 (44%) 1(5%) 7 (39%)
5 14 (78%) 4 (22%) 1 (5%) 3 (17%)
twice bir N V 4 9 (50%) 9 (50%) 3 (16%) 6 (33%)
5 12 (67%) 6 (33%) 2(11% ) 4 (22%)
twice N V 4 10 (55%) 8 (45%) 7 (39%) 1 (6%)
5 13 (72%) 5 (27%) 1 (5%) 4 (22%)
(*) p <.05 (difference from adults)
In the structure twice bir N-acc V, when the object is adjacent to the verb, 55% (n=10) of
the younger children and 78% of the older children (n=14) have the 1-object reading,
while 91% (n=l 1) of the adults have this reading only. The difference between adults and
four-year-olds is significant (Fisher’s Exact Test, two sided, p<.05) but there is no
difference between five-year-olds and adults (Fisher’s Exact Test, two sided, p=.622).
The four year olds are non-adult like, but they have a clear preference of the 1-object
161
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
reading over the 2-object reading (Fisher’s Exact Test, two sided,p< .01). It is important
to note that, this preference for 1-object reading may be a general tendency to act upon
one object, rather than a tendency to assign wide scope exclusively to the accusative
marked objects only. Compare the percentages of all object types: 44%, 55%, 50% and
50%. The 1-object reading across all the object types is similar. See Figure 2, Figure 3,
Figure 4 and Figure 5 for a comparison of children’s and adults’ treatment of four object
types.
Figure 2: Response pattern in “bir N-acc twice V'
□ mixed
□ 2-object
■ 1-object
adults
Figure 3: Response pattern in “twice bir N-acc V'
□ mixed
□ 2-object
B 1 -object
adults
162
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 4: Response pattern in “twice bir N V’’
ag e 4 age 5 adults
□ mixed
□ 2-object
■ 1 -object
Figure 5: Response pattern in bare objects
ag e 4 age 5 adults
□ mixed
□ 2-object
■ 1 -object
When the object is not case-marked, children’s response pattern is quite similar to what
we see in adults. Note that the high scores of 1-object reading o f non-case marked objects
do not imply anything because these structures are ambiguous. What is important is that
the number of children who have the 1-object reading with accusative marked indefinite
objects is not any higher than the number of children who have this reading with non
case marked objects. These results contrast with what we had in adults, who treat case-
marked versus non-case marked objects differently. These results that the children’s
treatment of case marked objects is not adult-like confirm the results we had in the scores
analysis.
163
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5.1.4 Discussion and Conclusion
The overall results of the experiment show that the children in the younger age group do
not necessarily have 1-object reading of the accusative marked indefinites, which is the
only grammatical reading for adults. Although the number of subjects who prefer the 2-
object reading is quite low, children are still not adult-like in terms of their distribution to
mixed and 1-object patterns.
The results of the experiment are similar to the act-out task results in Kramer (2000) in
the sense that wide scope (specific, high, 1-object) indefinites can have a narrow scope
(non-specific, low, 2-object) reading in young children. However, we cannot conclude
that they are exactly like the Dutch results. Despite the difference between adults and
children, which shows that children allow the 2-object reading that is not possible in adult
grammar, the scores of 1-object reading are very high, when compared to the 1-object
scores in Dutch (as reported in Kramer, 2000). Therefore, based on the results of the
present analysis, it is difficult to conclude that children behave like children acquiring
Dutch, although they allow narrow scope reading o f accusative marked indefinites in a
non-adult-like way.
The high wide scope results, however, can be attributed to the shortcomings of the act-out
test, which can only show that there are children who prefer 2-object reading or 1-object
reading. It does not show clearly whether the structures are ambiguous for children (and
they pick up one of the readings) or they have only one reading and that is what they act
164
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
out. In other words, those children who act upon one object in the accusative marked
structures may allow the 2-object reading (which is non-adult-like) as well.
In the next experiment, I address this question and test whether children who allow thel-
object reading allow only the 1- object reading with the accusative marked objects.
Interestingly, this does not seem to be a problem in the Dutch study as the children's 1-
object scores are already pretty low. In the present study, however, before we conclude
that children's accusative marked indefinite objects can take scope over the adverb in an
almost adult-like way, we need to make sure that the 1-object reading they have is the
only reading they have access to. The second experiment that is discussed in the next
section is designed to address this question. I will evaluate the results of both twice
experiments together in section 5.2.4 below.
5.2 Experiment 2: Follow-up to Experiment 1
5.2.1 Participants
20 four-year-old (mean age: 4;2, age range: 3;6-4;6) and 20 five-year-old (mean age: 5;1,
age range 4;7-5-10) children participated in the experiment. 12 university students
participated in the study as the adult control group.
5.2.2 Procedure
The same setting as the one in Experiment 1 was used, but instead of asking the
participants to act-out the sentences, an experimenter acted out the sentences and the
165
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
children were asked whether or not they were acted out correctly. In this experiment,
there were three child characters that were playing a game with Ayse, the main character
of Experiment-I. Now, Ayse wanted them to act out some sentences that she produced.
The participants were told that sometimes the children in the game could not do the
action well, because they did not understand the orders. When they could not perform the
action correctly, the participant was asked to help them (i.e., to show them how they were
supposed to do it). For example, Ayse wanted one of the children in the game to push one
of the chairs (with an accusative marked object) twice, but the character pushed two
different chairs. The subject was asked whether the child could do it correctly. If the
subject said “yes,” her response was evaluated as a possible reading for that scene, that is,
2-object reading for an accusative-marked object. If the subject said “no,” then he or she
was told to show how the child in the game was supposed to do the action or tell us why
it was wrong so that we could confirm that the response was given for a relevant reason.
The experiment involved six sentences for each accusative marked-type and 12 filler
sentences that were followed by either correct or incorrect performances. Before the test
sentences, children went through five training sentences so that they could learn the rules
of the game and also see how the children failed to perform some sentences. The
complete list of test sentences and warm-up items are presented in Appendix-1. This
experiment involved only the accusative marked objects, as the non-case marked objects
allow both readings, that is, they are ambiguous for adults in terms of scope. Therefore,
the expected response for all the test sentences was “no.” The reason why we did not
have the “yes” option in half of the test sentences is that the experiment was conducted as
166
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
a follow-up to Experiment 1, where we already saw that children have a 1-object reading
of the sentences with case marking. Experiment 2 was conducted to check whether they
allow the 2-object reading as well, so it is a continuation of the earlier experiment.' In
order to take care of the 'yes' response bias, they were presented with training sentences
with easier scenes. In that way, they saw that some scenes should be rejected and some
should be accepted. Only those children who passed the training sentences (who are able
to give both no and yes responses appropriately) were included in the test. So, the
children who were tested knew that it was possible to reject a sentence. More
importantly, they had filler sentences that were randomly placed in among the actual test
sentences and half of these filler sentences required yes responses and the other half
required no responses. Again, the children were included in the analysis if they gave the
expected responses in the fillers. This required them to give “no” responses. In other
words, children did have both “yes” responses and “no” responses in the experiment
although the correct response in the test sentences were “yes” only.
5.2.3 Results
5.2.3.1 Mean scores
Table 5 below shows the results of the experiment according to mean scores of each age
group. The scores are the scores of 2-object reading, that is, the cases when children
accepted the 2-object acting with the accusative marked objects. In 67% of the responses
both child groups allowed the 2-object reading, which is not a possible reading for adults.
167
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 5: 2-object reading scores in Experiment 2
Aye 4 (range: 3; 6-4; 6) Age 5 (range: 4; 7-5; 10) Adults
bir N-acc twice V
twice bir N-acc V
3.75* (62%, SE=.56)
4.30* (72%, SE=.43)
3.85* (64%, SE=.54)
4.20* (70%, SE=.49)
0
1 (.01%)
(*) /K.001 (difference from adults)
Experiment 2, then, revealed that children allowed the 2-object reading of the accusative
marked indefinites and the high 1-object responses in Experiment-1 reflect children’s
preference only rather than the only reading that they have access to.
In (197) and (198) below are some examples of what children say when they reject the
wrong actions. It is interesting to note that when they were describing the way they
understand the sentence with the accusative-marked indefinite, they use words such as
the same chair, or only this chair. 70% of children failed to give such responses and
accepted the sentences.
(197) Iki kere it-ecek-ti aym sandalye-yi (Kerem, 4;5,23)
Two time push-fut-past same chair-acc
‘She would push the same chair-acc twice.’
(198) Sadece bu-nu boyle it-me-si gerek (Pelin, 5;7,9)
Only this-acc like-this push-inf-poss&3S necessary
‘She should push only this-acc like this.’ [=pushing one chair twice]
Figure 6 shows the accept responses of the three age groups. As it is seen in this figure,
children accept a reading which is not acceptable for adults. In this figure, we also see
168
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
that the children’s interpretation of the two accusative marked indefinites goes parallel to
each other, that is, they do not treat the two object types (adjacent and non-adjacent)
differently. It is important to note that in adult speech, accusative marked indefinites can
have narrow scope with respect to other constituents when they are adjacent to verbs. I
attributed this to the non-deletion of the accusative case although the object stays in
AgrOP. The fact that children have a 2-object reading both in the scrambled and non
scrambled positions show that the behavior of the accusative marked indefinites cannot
be attribute to a restriction on movement in child speech. So, Schaeffer’s (1997, 2000)
account cannot apply to Turkish, because in her account, objects do not necessarily
move, but can still be referential reading regardless of their position, therefore what is
missing is scrambling only. In Turkish, objects do not have 1-object reading, although
they are in scrambled positions.
Figure 6: 2-object reading in Experiment 2.
— 100
O )
c
I
age 4 age 5 adults
- bir N-acc twice V [2-object]
-twice bir N-acc V [2-object]
Age groups
The results are confirmed by the ANOVA results. The 2 X 3 ANOVA conducted on the
placement of the adverb and three age groups, shows that the position of the object with
169
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
respect to the adverbial twice does not have a main effect on the readings
(F(l,98)=2.117, n.s.), but the main effect of age is significant (F(2,98)=24.666,/?<.001).
In other words, while there is a significant difference between the age groups, neither the
children, nor the adults have a difference in their responses in two different object
positions. The results of the experiment confirm that children allow the 2-object reading
of the accusative marked objects in a non-adult-like fashion. The high percentage of the
1-object reading observed in the first twice experiment (65% and 85% for four and five
year old children respectively) is misleading as it shows a preference for one reading
only, rather than displaying the only reading children can get. In the next section, we
look at children’s distribution according to the response patterns.
5.2.3.2 Response patterns
Table 6, Table 7 and Figure 7 display the distribution of the readings within each
individual age group. As seen in Table 6, all the adults have only 1-object reading when
the object is not adjacent to the verb. When the object is adjacent, one adult has a 2-
object pattern. There is an obvious preference for 1-object reading with both accusative
marked object types.
Table 6: Response Pattern in Experiment 2 (adults)
type 1-obj. 2-obj. & mixed 2-obj. mixed
bir N-acc twice V 12 (100%) 0 0 0
twice bir N-acc V 11 (91%) 1(8.3%) 1(8.3%) 0
170
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 7 below shows the children’s response patterns. A majority of four and five year
old groups have a preference for a 2-object reading (65% and 70% respectively) when the
object is not adjacent to the verb (Fisher’s Exact Test, two sided,/?<.01). Note that no
adult has a 2-object preference with this structure. In the other object type, too, the
majority of the children accepts the 2-object reading. There is no significant difference
between the two child age groups while both age groups are significantly different from
adults in their accept scores (Fisher’s Exact Test, two sided, p<.01) in both object types.
Table 7: Response pattern in Experiment 2 (children)
type age 1-obj. 2-obj.
& mixed
2-obj. mixed
bir N-acc twice V 4 5 (25%)* 15 (75%) 13 (65%) 2 (10%)
5 6(30%)* 14 (70%) 13 (65%) 1 (5%)
twice bir N-acc V 4 3 (15%)* 17 (85%) 14 (70%) 3 (15%)
5 8 (40%)* 12 (60%) 12 (60%) 0
(*) p <.05 (difference from adults)
Figure 7 below represents the age group’s scope assignment to the accusative marked
objects (two object types merged). The difference between the adults and the children is
obviously bigger than what we observed in Experiment 1 (See Figure 2 and Figure 3
above for a comparison).
171
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 7: Response pattern in Experiment 2 (two object types merged)
□ mixed
□ 2-object
■ 1 -object
age 4 ag e 5 adults
Figure 8 below represents the scores of the two twice experiments. The lines show the 1-
object and 2-object reading scores of two object types (adjacent and non-adjacent) at two
experiments.
As seen in the figure, adults have 1-object reading of the accusative marked indefinites
and it is the only reading they have. Four-year-old children allow both 1-object and 2-
object reading of the accusative marked indefinites. The five-year-old group starts
distinguishing the scope of these objects.
Figure 8: 1- and 2-object readings in Experiments 1 and 2.
o -
100 - 1
O )
C
“D
C
C O
I
adults age 4 age 5
Age groups
- bir N-acc twice V [1 -object]
-twice bir N-acc V [1-object]
- bir N-acc twice V [2-object]
-twice bir N-acc V [2-object]
172
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5.2.4 Discussion and Conclusion
The two twice experiments reported in this chapter confirmed that, in adult speech, as
long as the object is outside the AgrOP domain, i.e, in a position where its case is not
deleted, it has wide scope with respect to other constituents, in this particular case, the
adverb twice. The surface position of the accusative marked indefinite with respect to the
adverb (See (199) and (200)) did not have a significant effect on adults’ interpretation of
the objects. They had wide scope with respect to the adverb in both positions. Only one
adult had a narrow scope reading of the accusative marked indefinite in the adjacent
position. It is likely that this happened because he was not paying attention to the
prosodic structure, which marks the position of the accusative marked indefinite and
interpreting the object as if it is still in AgrOP (201). Note that when the object is
adjacent to the verb in the surface structure, its position i.e. whether or not it is in AgrOP,
is not clear unless the prosodic cues are followed. When it is in AgrOP it bears the focal
stress, when it is higher, it is not stressed. In the experiments, the presence or absence of
bir did not have an impact on adults’ interpretation of the objects.
(199) [A grsp- • • [bir N-acc[ twice.. .[[A g rO P bir N aee [V p bir N acc V] ] ] ]
(200) [Agrsp• • • [[ twice... [bir N-acc ... U grO P bir N acc [vp bir N acc V] ] ] ]
(201) [A grS P . • • [[ twice... [ [A g rO P bir N-acc [V p bir N acc V] ] ] ]
Children are different in their interpretation, namely, their accusative marked indefinites
can have narrow scope with respect to the adverb, and thus they allow a 2-object reading
173
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
in both scrambled and non-scrambled positions. Both the score analysis and the response
pattern analysis yield to the same conclusions. Unlike the adult subject who had narrow
scope interpretation in the adjacent position only, children can assign narrow scope to
objects both in the scrambled and non-scrambled positions. As seen in Figure 8 above
they do not differentiate the two object types according to their linear position. Therefore,
we cannot attribute the children’s non-adult like behavior to prosody. If that were the
reason, just like adults they would have an asymmetry between the two positions.
The results show that neither the overt movement nor the case morphology has an impact
on children’s scope assignment to objects. Their treatment of objects rather suggests that
they reconstruct the object to their AgrOP position. I propose that this happens when they
are unable to interpret the object due to an interpretational difficulty with the information
structure. In Chapter 8 ,1 will discuss the properties of this operation in detail.
The results, then, showed that the acquisition pattern of Turkish is similar to Dutch,
namely that the types of objects that have only the wide scope reading in adult speech can
have narrow scope reading in child grammar and children are not sensitive to the surface
position of the objects. In Table 8, we see a comparison with the scores of the twice test
in the Kramer (2000) study. The scores are the accept scores of the subjects in narrow
scope condition. Dutch-speaking children accept a narrow scope reading of the
accusative marked indefinites in narrow scope condition in 51% of the cases. Turkish
speaking children allow this reading in a similar fashion in accusative marked indefinites.
174
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 8: Dutch and Turkish twice-test results compared
Dutch Turkish
twice > indefinite (Kramer, 2000) Bir N-acc (merged)
Children 51% 63%
Adults 8% 0%
Despite the similarity of the results, I do not assume a semantic incorporation analysis of
the objects. Objects do not need to get incorporated to verbs and undergo an argument
reduction operation to have a narrow scope interpretation. On the contrary, I propose that
they are reconstructed to their AgrOP positions and this is possible when they maintain
their argument status.
No effect of bir is observed in the experiments; therefore the results are not as predicted
by the Lexical Factor hypothesis of Su (2001). Similarly, as also discussed above,
because the linear position of the adverb and the object does not have an impact on
children’s interpretation of the objects, the isomorphism account cannot capture the facts
in Turkish. We can have the same argument for the c-command account. The children do
not seem to be sensitive to hierarchical ordering either as the accusative marked
indefinites do not necessarily scope over twice.
This chapter presented children’s scope of indefinite objects with respect to the adverb
twice. In the next chapters, three series of experiments conducted on the scope interaction
of objects with negation will be reported and discussed.
175
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6 Indefinite objects and negation
Experiment 3, presented in this chapter, tests whether children’s accusative marked
indefinites can have narrow scope with respect to negation. The experiment shows that
both the accusative case morphology and the numeral bir have impacts on the
interpretation of the indefinite objects and the children are close to adults in their scope
assignment to accusative marked indefinites. There is thus a contrast between the results
of the twzce-experiment, where children have lower wide scope scores.
In this experiment, children and adults were presented with accusative marked and non
case marked sentences as exemplified in (202) and asked to make judgments regarding
the scope of the indefinite object with respect to negation.
(202) a. Ay§e bir elma / elma ye-me-di. neg > bir N / N
A. an apple /apple eat-NEG-PAST
‘Ay§e did not eat an apple.’ (no apples were eaten)
b. Ay§e bir elma-yi ye-me-di. bir N-acc > neg
Ayse an apple-ACC eat-NEG-PAST
‘ Ay§e did not eat a particular apple/ an apple’
176
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
c. Bir elma-yi Ay§e ye-me-di. bir N-acc > Neg
an apple-ACC Ayse eat-NEG-PAST
‘Ay§e did not eat a particular apple.’
When the object is displaced from the AgrOP, either overtly or covertly, it is expected to
scope over negation. In the case of the accusative marked objects above, the objects are
assumed to be in displaced positions, even when they occur adjacent to verbs in the actual
sentences.
6.1 Experiment 3
6.1.1 Participants
38 four-year-old, 32 five-year-old children and 14 adults participated in the experiment.
Children were randomly drawn from three different kindergartens in Istanbul. Adults
were college students in Istanbul and Los Angeles and participated in the experiments
voluntarily. They were all native speakers of the Istanbul dialect of Turkish and belonged
to the middle/upper middle socio-economic class. The four sentence types were tested on
four different groups of children.
Table 9 below shows the number of subjects and their distribution to each object type as
well as the age range of subjects in each group.
177
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 9: Participants in Experiment 3 (children).
Age No= Age range Mean age
bir N-acc XP V-neg 4 9 3;7-4;5 4;1
5 9 4;7-5;9 5;0
bir N-acc V-neg 4 9 3;6-4;5
4;1
5 7 4;7-5;0 4;7
bir N V-neg 4 8 3;6-4;4 4;0
5 9 4;6-5;7 5;0
N V-neg 4 12 3;7-4;5 4;0
5 6 4;6-5;8 5;0
6.1.2 Procedure
A version of a truth-value judgment task (Crain & Thornton, 1998) was used. Two
experimenters conducted the experiment in a quiet room at kindergartens. One of the
experimenters told short stories to the children while showing pictures on a computer
screen. The other experimenter played two hand-puppets, who were introduced to
children prior to the game. Children were tested individually and were told that they
would play a game in which two puppets compete with each other. The puppets listened
to a story told by the experimenter and then they were asked what happened at the end of
the story. Puppets took turns when they were speaking, that is, each puppet talked about a
different story. If the puppet understood what happened at the end of the story, that is, if
s/he stated what happened at the end correctly, the child rewarded him/her with a candy.
The task of the child in this game was to determine whether or not the puppet gave the
correct answer and also to determine the winner at the end of the game. With adult
participants, the same setting was used, but instead of the puppets, the experimenter told
the test sentences as a conclusion or summary of the stories.
178
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Children heard three test sentences/ scenarios (all the same object type) and five filler
stories (see Appendix 2 for the complete list of the sentences). These sentences were
accompanied by a one-object (wide scope) context in which there were three objects and
only one of these objects were not acted upon. An example of a wide-scope condition and
the test sentences are presented in (203) below40.
(203) Hide and Seek Game
Story:
Burada gocuklar saklanbag oynuyorlarmiy Ahm et’i ebe yapmiglar. Obiir
gocuklar da ko§mu§ gitmig saklanmiy Ahm et’in biitiin gocuklari bulmasi
gerekiyormuy Ahmet gocuklari aramig aramiy bir ttirlti bulamamiy Tam
vaz gegiyormuy agacin arkasina saklanan gocugu gdrmii§! Onu hemen
sobelemiy sonra sepetlerin arkasina saklanan gocugu da bulmuy Geriye
bir gocuk kalmiy Ama obiir gocugu arami§ aramig bir tiirlii bulamamiy
En sonunda vazgegmiy ben bunlari bulamayacagim demiy Yuriimiig
gitmiy
'The children are playing hide-and-seek here. Ahmet becomes the seeker.
The other children hide. Ahmet starts looking for them. At first he cannot
find anybody.. .He decides to give up. He thinks he is not good at this
game. Then he sees the child who was hiding behind the tree! Right after
that, he finds the girl who was behind the baskets, too. But no matter how
hard he tries, he cannot find the other kid. In the end he decides to give
up.'
4 0 In addition to the three test sentences and five fillers, children were presented with the same type of
objects with an alternative type of scenarios in which 'no objects' were acted upon. The results of these test
items are presented and discussed in Appendix 3. So the children heard six test sentences (with two
conditions, and five filler sentences).
179
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(204) Test sentences
a. PUPPET: Ahmet bir 9ocuk bul-a-ma-di.
Ahmet a/one child find-neg-past
‘Ahmet could not find a child’
b. PUPPET: Ahmet 9 0 0 1 k bul-a-ma-di.
Ahmet child fmd-neg-past
‘Ahmet couldn’t find a child.’
c. PUPPET: Ahmet bir 9 0cug-u bul-a-ma-di.
Ahmet a/one child-acc find-abil-neg-past
‘Ahmet could not find a child-acc’
d. PUPPET: Bir 9 0 cug-u Ahmet bul-a-ma-di
a/one child-acc Ahmet find-abil-neg-past
‘Ahmet couldn’t find a child.’
e. CHILD: Accept (Wide scope) / Reject (Narrow scope)
Prior to the test, children also heard five training sentences. The aim of the training
stories was to teach the child the rules of the game; therefore they were simpler than the
actual test sentences. The fillers had similar stories, but the sentences that the puppets
uttered contained intransitive verbs. Those children who failed in two or more training
180
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
sentences were not allowed to continue the test. Similarly, I eliminated the children who
failed in more than one filler sentence and those who gave irrelevant responses to the
‘why’ questions.
Just as I did in the twice experiment, I had four types of objects but I presented them to
four different groups of children not only to prevent a possible influence of one reading
on the other, but also to limit the length of the testing session. So, each child had only one
type of object in his/her test sentences. I had three object types that were adjacent to the
verb-negation complex. For the other type, the accusative marked object that is non-
adjacent, I either inserted a locative phrase between the object and the verb ((205)-(206),
or I switched the subject and the object controlling the intonation ((207)-(208)) by
dropping the sentential stress on the verb.
(205) Ash [dolab-a] bir §i§e-yi koy-ma-di. DPiocation bir N-acc V-neg
Ash [fridge-dat] bir bottle-acc put-neg-past
‘Asli didn’t put a bottle-acc in the fridge.’
(206) Ash bir §i§e-yi [dolab-a] koy-ma-di. bir N-acc D P iocation V-neg
Ash a bottle-acc [fridge-dat] put-neg-past
‘Asli didn’t put a bottle-acc in the fridge’
181
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(207) [Ahmet] bir ?ocug-u bul-a-ma-di. DPsu b j bir N-acc V-neg
Ahmet bir child-acc find-abil-neg-pass
‘Ahmet could not find a child.’
(208) Bir focug-u [Ahmet] bul-a-ma-di. bir N-acc D P subj V-neg
A/one child-acc [Ahmet] find-abl-past
‘Ahmet couldn’t find a child-acc.’
6.1.3 Results
6.1.3.1 Mean scores
The scores of adults and children are seen in Table 10 and Table 11 respectively. The
scores are the scores of the cases when children and adults accepted the sentences as a
commentary on the scenarios where one out of three objects was not acted-upon. The
expected reading for the accusative indefinites is accept, namely indefinite having wider
scope over negation. It is rejected for the non-case marked indefinites.
Table 10: Results of Experiment 3 (adults, mean accept scores)
object type adults
bir N-acc XP Vneg 3.0 (100%)
XP bir N-acc Vneg 2.71 (90.3%, SE=.163)
bir N Vneg 1.42 (47%, SE=.358)
N Vneg .000
As seen in Table 10 adults accepted the wide scope reading of the accusative marked
indefinites in non-adjacent and adjacent conditions in 100% and 90% of the cases. In the
case of the bir N objects, they still have some wide scope reading in around half of the
182
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
cases although it is not case marked. They never assign wide scope to bare objects.
Considering the analysis of the bare objects, this result is expected. Table 11 below
displays the children’s scores.
Table 11: Results of ex periment 3 (children, mean accept scores)
Object type Age 4 Age 5
bir N-acc XP V-neg
XP bir N-acc V-neg
bir N V-neg
N V-neg
2.44 (81%, SE=.37)
2.55 (85%, SE=.33)
2.12 (70%, SE=.398)
1.66 (55%, SE=.35)
2.66 (89%, SE=.21)
2.42 (81%, SE=.61)
1.66 (55.3%, SE=.408)
1.33 (44%, SE=.49)
(*) p <.05 (difference from adults)
The children’s wide scope reading of the accusative marked indefinites is around 80%.
They assign wide scope to bir N objects in 70% of the cases although they are not case
marked. Interestingly their bare objects are assigned wide scope with respect to negation
in 55% of the cases, which is around chance. Therefore, the non-case marked objects and
the bare objects do not have narrow scope in child speech.
Here are the ANOVA results and an overview. A 3 (age) X 2 (case) X 2 (bir) X 2
(adjacency) ANOVA was run on the 1-object reading (indefinite > negation) scores.
Three age groups (4 year olds, 5 year olds and adults), two case conditions (case marked
and non-case marked objects), two bir conditions (case marked and non case marked
objects that have bir and the bare objects), and two adjacency conditions (those that are
adjacent to verbs and those that are scrambled to a non-adjacent position) were included
in the analysis.
183
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The results showed that accusative case (F (l, 113)=11.039, /><01) and the presence of
bir (F(l,113)=8.917,/K.01) have main effects on scope assignment. Age has an effect
trend (F(2,l 13)=2.717,p=.070) on the 1-object preference. Adjacency does not have an
effect (F (l,l 13)= .313, n.s.). In other words, both the case morphology and the numeral
bir affect the interpretation of sentences. The subjects interpret the case marked and non
case marked objects differently and just like adults, they consider the overt case
morpheme as a signal for object displacement. Moreover, bir N objects pattern more like
the accusative marked objects than the bare objects. No interaction effects are observed
between the factors. Figure 9 displays the age groups and accept scores.
Figure 9: Results of Experiment 3 (Wide scope reading over negation)
100
in
a >
i _
o
o
in
Q .
C D
O
O
<
age 4 age 5 adults
— ■— bir N-acc XP Vneg
— * — bir N-acc Vneg
— x — bir N Vneg
- - o - • N Vneg
Age groups
As seen in Figure 9, adults treat the four object types differently (F(3, 52)=48.498,
/?< 01). The difference is observed between the accusative-marked objects and bir N
objects (F(3, 52)=48.498, p<.05, F(3, 52)=48.498 p<.01, adjacent and non-adjacent
objects respectively). A similar difference is recorded between the accusative marked
184
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
objects and bare N objects (F(3, 52)=48.498,p< 01, both adjacent and non-adjacent
objects). There is no difference between the two accusative marked object types (adjacent
and non-adjacent) (F(3, 52)=48.498, ns). As the ANOVA results suggest, adults treat the
two non-case marked objects {bir N and bare N) differently as well (F(3, 52)=48.498,
p<.05). Therefore bir has an impact on the interpretation of non-case-marked objects, i.e.,
bare objects and bir N objects are not treated alike. Because the bir N objects and bir N-
acc objects are treated differently in terms of their scope assignment, we can suggest that,
it is not only bir that triggers wide scope assignment to bir N-acc objects. Both case and
bir have impact on the interpretation; i.e., they are treated as indicators of object
displacement.
ANOVA results run on the scores of the four- and five-year-old children reveal that
children do not treat structures differently (F(3,34)=1.305, n.s , F(3,27)=2.936, n.s., four-
and five-year-olds respectively). Neither the case morphology, nor the adjacency to
verbs, nor the presence or absence of bir has an effect on four-year old children’s scope
assignment. In other words, although the children have very high wide scope scores, their
treatment of all object types alike suggest that they are not at the adult-level in their
comprehension.
In summary, adults’ scope interpretation of the case marked and non-case-marked object
types was different. The two accusative marked indefinites are treated alike, i.e.,
adjacency did not play a role. Their accusative marked indefinite objects had wide scope
and the bare objects had narrow scope. The scores of bir N objects, which are not case
185
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
marked and which have bir ‘a/one,’ are around chance. The case morphology, as well as
bir ‘a/one’ had a role on adult’s scope assignment. Children’s scope interpretation is not
different across object types. Their bare objects and non-case marked objects do not
receive narrow scope either, therefore they are scopeless. All object types are treated very
similar to each other, thus children do not seem to differentiate the case marked versus
non-case marked objects despite the high wide scope scores.
6.1.3.2 Response patterns
Table 12 and Table 13, as well as Figure 10, Figure 11, Figure 12 and Figure 13 show the
results of the response pattern analysis. The expected response was “accept” for the wide
scope indefinites. All the adults (n=14) accepted this reading o f the accusative marked
object when the object is not adjacent to the verb. When the object and the verb were
adjacent, 78% (n=l 1) of the adults accepted the wide scope reading, while 21% of them
(n=3) had a mixed pattern. There was no adult who had a rejection pattern. Despite the
three adult subjects who have a mixed pattern, the difference between the accept scores
of adjacent versus non-adjacent objects is not significant (Fisher’s Exact test, two sided,
p=1.0). We can conclude, then, that adults’ accusative marked objects, regardless of
adjacency, had wide scope with respect to negation.
When the object was not case marked (bir N), 36% of adults (n=5) rejected the wide
scope reading of objects, 36% of them (n=5) accepted this reading and the rest, 28%
(n=4), had a mixed pattern. In other words, adults did not have a preference for one of the
readings when the object was not case marked, but had bir. In the case of the bare objects
186
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
all the adult participants had only narrow scope. The results were similar to what we
observed in the scores analysis.
Table 12: Response pattern in Experiment 3 (adults).
type
Accept Reject & mixed Reject Mixed
bir N-acc XP Vneg
bir N-acc Vneg
bir N Vneg
N Vneg__________
14 (100%)
11 (78%)
5 (36%)
0
0
3 (21%)
9 (64%)
14 (100%)
0
0
5 (36%)
14 (100%)
0
3 (21%)
4 (28%)
0
Table 13 shows the children’s response pattern. There is no difference observed in
children in their response pattern to the two accusative marked object types (adjacent
versus not adjacent). In the younger child group, in both sentence types, there is one child
who consistently rejected the wide scope reading of accusative marked objects. In the
older group, no child rejected the wide scope reading. This finding is quite similar to
what we observe in adults. When the objects are case-marked, the majority of the subjects
in all three age groups tend towards wide scope.
When the object is not accusative-marked, 50% of four-year-old (n=4) and 33% of five-
year old children (n=3) accept the wide scope reading of the object. Four-year-olds treat
the accusative marked versus non-case marked objects differently (Fisher’s Exact test,
two sided, p<. 05), while five-year-old children's scores are not significantly different.
There is no difference between the two child age groups or between children and adults
in accept responses.
187
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 13: Response pattern in Experiment 3 (children).
type age accept reject
& mixed
reject mixed
bir N-acc XP V-neg 4 7 (77%) 2 (23%) 1 (11%) 1 (12%)
5 7 (77%) 2 (22%) 0 2 (22%)
bir N-acc V-neg 4 7 (78%) 2 (22%) 1 (11%) 1 (11%)
5 4 (57%) 3 (43%) 0 3 (43%)
bir N V-neg 4 4 (50%) 4 (49%) 1 (12%)* 3 (37%)
5 3 (33%) 6 (66%) 2 (22%) 4 (44%)
N V-neg 4 4 (33%) 8 (67%) 3 (25%)* 5 (42%)
5 1 (17%) 5 (83%) 2 (33%)* 3 (50%)
(*) p<.0005 (different from adults)
Figure 10: Response pattern in bir N-acc XP V-neg
□ mixed
□ narrow
■ wide
adults
Figure 11: Response pattern in bir N-acc V-neg
□ mixed
□ narrow
■ wide
adults
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In summary, accusative-marked indefinites have wide scope with respect to negation in
the majority of the children’s grammar. The proportion o f children who accept the wide
scope reading in accusative-marked indefinites is the same as the proportion of adults
who allow wide scope in the non-case-marked objects. In the case of the non-adjacent
objects, no adults have narrow or mixed pattern while two children (in each age group)
have either mixed or narrow scope pattern. The results are similar to the analysis of the
scores which detected only a difference trend in age groups.
Figure 12: Response pattern in bir N V-neg
100 --------------- --------------- ---------------
80
□ mixed
□ narrow
■ wide
age 4 age 5 adults
In bare objects, 33% (n=4) of the four-year-old and 17% o f five-year-old children (n=l)
accept the wide scope reading of objects. In the five-year-old group less children seem to
accept the wide scope reading of bare objects, which is not accepted by any adults, but
the difference between the child groups is not significant (Fisher’s Exact test, two sided,
p=.6). Children do not treat the two non-case marked object types (bir N and bare
objects) differently.
189
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Adjacency to the verb does not have a significant effect on scope. Children’s response
pattern is similar to adults. When the object is not accusative-marked, children are less
likely to accept the wide scope reading of the object, which suggest that case marking
make a difference for children just as it does for adults. When the objects are in bare
form, we see a difference between adults and children. While no adult accepts the wide
scope reading for a bare object, it is possible for some four-year-olds.
Figure 13: Response pattern in bare N V-neg
100 -I --------------- --------------- ---------------
80
□ mixed
□ narrow
■ wide
age 4 age 5 adults
The most important observation is that children do not seem to be very different from
adults despite a small number of children who allow wide scope reading in structures that
never have wide scope in adult speech or vice versa.
6.1.4 Discussion and Conclusion
In the negation test, adults’ accusative-marked indefinites had wide scope with respect to
negation. There was no difference in the interpretation of the adjacent versus non-
adjacent objects. Therefore, as was the case in the twice test, the overt case morpheme on
the objects, or its inability to delete, rather than word order or adjacency, was recognized
190
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
as the signal of the objects’ structural position. As was the case with the twice test,
children allowed the narrow scope reading of the accusative marked indefinite objects in
the negation experiment. However, the scores were much higher and closer to the adults'
as compared to the twice test results. Children rejected the wide-scope reading in 15% of
the cases in wide-scope context when the object was non-adjacent to the verb. It is quite
low as compared to the scores of the Dutch children reported in Kramer (2000), who
allow the ungrammatical reading in 84% of the cases. In the following sections, I will be
discussing the reasons behind these differences between the two languages and also
between the results in the two experiments I presented.
6.2 Further Discussion of the Results
In addition to case, bir had an impact on the interpretation in adult speech, and bir N
objects and bare objects were not treated alike. This raises the question of whether or not
Turkish is one of those languages in which the numeral has a significant effect on the
scope assignment (c.f. Su, 2001).
Let us have a look at the languages and wide scope readings. Table 14 below presents the
wide-scope reading of the indefinite objects in Dutch, English, Spanish, Chinese and
Turkish. In the Turkish columns, I show all three types of objects that have bir, so that I
can compare these with the other structures in other languages.
191
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 14: Languages compared (wide scope readings, percentages).
O'
0
1
i
o
o
Children *16% *33% 84% 78% *85% 81% 70%
Adults 100% 50% 57% 64% 100% 90% 47%
(*) p <.05 (difference from adults)
In terms of the adults’ treatment of the wide scope indefinites, Turkish is similar to Dutch
in the scrambled accusative marked objects, i.e., adults do not allow a narrow scope
reading of the object in a scrambled position. In those nominals, that are adjacent to the
verbs, adults allow narrow scope reading in 10% of the cases. In those nominals where
we do not have case, in 47% of the cases adults have wide scope reading, which is only
around chance. This final result is similar to the languages, such as English, where the
same form of the object has two different scope interpretations (compare 47% to 50% of
English). In children’ s interpretation of the accusative marked indefinites, we see that
Turkish and Dutch are different and Turkish is closer to the Chinese and Spanish side of
the scale by restrictive narrow scope interpretation of the case marked object. In other
words, this picture suggests that Turkish is one of those languages where the numeral one
had an impact on the scope interpretation of the indefinite objects.
192
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Just as predicted, while bare nominals had narrow scope for adults, the interpretation of
the bir N nominals was ambiguous (their wide scope assignment was around chance). Let
us consider attributing this finding to the fact that bir has a singularity interpretation
similar to yi-ge in Chinese and it is this property that triggers wide scope reading. This
would explain why we have a high wide scope rate in bir N objects that does not have
case as well. However, although bir N has unexpectedly high rate of wide scope, it is
important to note that its wide scope score is not as high as the accusative-marked ones.
This may be attributed to the contrast created by the case morphology. As far as the
typological properties of Turkish are concerned, such an analysis seems to be valid. It is
interesting to note, however, that we did not have an effect of bir on the interpretation of
scope with respect to the adverb twice (See the Experiment 2 results). The results in the
negation experiment and the second twice experiment were surprisingly different. While
children failed to assign wide scope to accusative marked objects in twice sentences in
the majority of the cases, they did much better in the negation experiments. The Lexical
Factor account, which attributes the high wide scope rates to the numeral reading of bir
cannot account for this difference unless we can find a reason why only the scope of the
objects with respect to negation are affected by the presence of bir.
Let us consider the difference between bare objects and bir N objects, which are both
non-case marked and which are expected to have narrow scope with respect to negation
and the adverb twice. A reason for the scope difference between the bare objects and bir
N objects may be that it is not natural to have bir N objects in negated sentences. Bare
objects sound more natural for an intended meaning of'no objects.' For example, if a
193
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
speaker would like to say that ‘Mickey Mouse did not eat any apples (no apples are
eaten), the most appropriate object type to use would be the bare object. Because there is
such an unambiguous alternative in the grammar of Turkish, the use o f the ambiguous
one is not expected. The participants are, therefore, more inclined to attribute a different
reading, in this case, a cardinal number reading to bir N objects rather than the ‘no
object’ reading. Thus what the participants were accepting was the cardinal reading rather
than the wide scope reading, which both correspond to the ‘accept’ responses. In contrast,
the bir N type of objects sound natural in sentences with twice and the subjects can assign
narrow scope to them without confusing it with the cardinal number one. Therefore, the
results of the negation experiments might be related to the expectations of the speakers,
which is a pragmatic reason rather than a semantic or syntactic one. The difference
between the results of the twice experiment and the negation experiment support an
analysis that attributes the participants’ behavior to a pragmatic difficulty, while it shows
that the Lexical Factor hypothesis cannot hold at least for Turkish.
In addition to the pragmatic disadvantage that makes the bir N objects infelicitous in the
negation context, there is another contextual disadvantage that most probably biases the
subjects towards a number reading. As will be discussed below, the number reading is
related to the wide scope reading because of the particular method used in the
experiment, not necessarily because there is a semantic relationship between the two.
Let’s have a closer look at the contexts of the experiments now and see how the number
reading and the wide scope reading overlap due to the scenarios used in Experiment 3.
194
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In the scenarios of Experiment 3, there were three objects, two o f them were acted upon,
one is left behind. In such a situation, as shown in (209), the number interpretation and
the wide scope interpretation had the same correct response (accept) with the wide scope
indefinites. That is, the children will accept the statement if they have access to the wide
scope reading (i.e., there is an apple such that MM didn’t eat). But they will also accept
the statement if they have the cardinal reading (i.e., it is not the case that MM ate ONE
apple). In the case of the wide scope reading, the statement is accepted because there is
one apple that was not eaten. In the case of the cardinal reading, it will again be accepted
because MM is eating two apples, not one. So the expected response of the two readings
is the same but due to independent reasons. Because bir N objects are not natural in a
negated sentence, whose intended meaning is ‘no object,’ the subject is more inclined to
consider the number reading.
(209) Situation: Two out of three objects were acted upon.
Mickey Mouse didn’t eat an apple.
(i) Narrow scope reading: It is not the case that MM ate an apple/ any apples.
R EJEC T (because he ate some apples).
(ii) Wide scope reading: There is an apple such that MM didn’t eat it.
ACCEPT (because there is an apple that is not eaten).
(iii) Cardinal reading: It is not the case that MM ate ONE apple.
ACCEPT (because there is exactly one apple that MM didn’t eat).
195
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In an experimental design with a scenario with three objects, it is not possible to see
which reading the children are accepting41. Therefore, the results of Experiment 3 are
hard to interpret. The high wide scope reading scores may be the cardinal reading scores,
and they cannot be interpreted as children’s almost adult-like mastery of scope
assignment in the case of the accusative marked indefinites. If this is true, the results of
the negation test is not comparable to the results of the twice test, and thus, the difference
between them is not problematic.
6.3 A redesign proposal
This problem of the scenarios calls for a follow up test that will clarify the results. To this
end, Experiment 4 was conducted. In Experiment 4, which will be presented in Chapter 7,
the scenarios are modified to reduce the number reading bias. Now compare the situation
in (210) with the one in (209). In the latter scenario, there are two objects in total, instead
of three. One of the objects is acted upon and the other one is not acted upon. The narrow
scope reading (i.e., It is not the case that MM ate an apple) is false. The wide scope
reading (i.e., there is an apple that MM didn’t eat) is true. The cardinal reading (i.e., it is
not the case that MM didn’t eat ONE apple) is false, as he ate exactly one apple. The
crucial difference between the two scenarios is that in the latter scenario the accept
4 1 One way to learn why children accept/reject the sentences would be to ask them why they accepted or
rejected the sentences. However, it is difficult to ask children ‘why’ the sentences were ‘correct’ when they
accept it. In the case o f the ‘no’ responses, we could ask them what the sentence was supposed to be, or
what really happened.
196
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
responses unambiguously correspond to the wide scope reading, whereas in the first one
they correspond to the number reading as well.
(210) Situation: One out of three objects were acted upon.
Mickey Mouse didn’t eat an apple.
(i) Narrow scope reading: It is not the case that MM ate an apple/ any apples.
R EJEC T (because he ate some apples).
(ii) Wide scope reading: There is an apple such that MM didn’t eat.
ACCEPT (because there is an apple that is not eaten).
(iii) Cardinal reading: It is not the case that MM ate ONE apple.
R EJEC T (because he ate exactly one apple).
The response ambiguity problem still exists in the second scenarios with respect to the
reject responses, which are ambiguous between the number reading and the narrow scope
reading, as one of the responses (accept or reject) has to be the same as the response for
the number reading. Because the major research question I am investigating is whether or
not children can assign wide scope to indefinites, it is more important for us to have an
unambiguous response pattern for the wide scope. Therefore the second scenarios are
better for the purpose of the study, although they are not perfectly unambiguous. The
question of why children reject the test sentences, is also important, but of secondary
importance. Besides, the rejection responses can be disambiguated by asking the children
why they rejected the sentences. Asking the why question and getting an explanatory
response is much easier in the case of rejection responses.
197
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The bir N object sentences will still not be as natural as the bare object sentences in a
negation context but disambiguation of the number versus wide scope reading will
improve the result in an expected way. As will be discussed in the next section, such a
modification changes the results of the negation tests in a predicted way and it shows that
the response pattern of the children is effected by the number interpretation and what
children were accepting in Experiment 3 was the cardinal one reading, rather than the
wide scope reading. With the modification, the results of the twice test and the negation
tests will show greater similarity.
Before I go on to the next Experiment, here is a summary of the issues discussed
regarding the results of Experiment 3:
Because there were three objects that were acted upon in the Experiment 3 scenarios, we
cannot know whether the children were accepting the cardinal reading or the wide scope
reading as both readings correspond to the accept responses. Because there is a bare
object possibility for the narrow scope (or ‘no object’) reading in Turkish, which makes
the bir N objects infelicitous in negation contexts, the use of bir N(acc) object may bias
the subjects towards a cardinal interpretation (ONE object is not acted upon, not two or
more), and therefore they accept the test sentences. The cardinal number reading and the
wide scope reading are not necessarily related. The overlap is due to a methodological
shortcoming; the ambiguity of the accept responses.
198
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Experiment 4, presented in the next chapter, will take care of the problem in such a way
that the accept scores will correspond to the wide scope reading only. Therefore, we will
know that when children accept a sentence they do not accept it because of the number
interpretation. I will discuss the implications of the negation experiment in the next
chapter after I present the results of the follow-up experiments on negation.
199
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
7 Another look at the indefinite objects and negation
This chapter presents two follow up experiments to Experiment 3, discussed in the
previous chapter. Experiment 4 tests children's and adults' comprehension of indefinite
objects with a modification in the scenarios. It also includes post-verbal objects that were
not included in any of the earlier experiments. In addition, there is one more age group, 6
year olds, in addition to the four and five year old age groups. The experiment shows that
both the accusative case and the numeral bir have an influence on the comprehension of
indefinite objects by adults. Experiment-5, which is a follow-up to Experiment 4, shows
that the influence of bir can be eliminated if the sentences accompany more felicitous
scenarios. Thus, the results of the two experiments together suggest that the accusative
case is the only factor that affects adults’ scope assignment in the experiments. In other
words the overt case morpheme signals that the object is in a scrambled position where
case omission is not possible. With the modification of the scenarios, children’s wide
scope assignment drops to 20% and becomes similar to what we report in the twice tests.
The new results show that children do not differentiate case marked versus non-case
marked objects at 4;0 years of age, assigning narrow scope to both. They start
distinguishing objects at the age of 5;0 but their comprehension does not reach an adult
like level even by the age 6;0. Implications of these findings are discussed in the final
section of this chapter.
200
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
7.1 Experiment 4: Follow-up to Experiment 3
Experiment-4 tests adults and children’s comprehension of indefinite objects repeated
here in (211) to (215). In addition to the structures tested in the earlier experiments, a new
structure, the post verbal objects were included in the test (213).
(211) [Bir fi9egi] kegi ye-me-di. [Bir N-acc] D P S U bj V-neg
a flower-acc goat eat-neg-past
‘The goat did not eat a flower-acc’
(212) Ke?i [bir 9i?egi] ye-me-di. DPsu b j [bir N-acc] V-neg
goat a flower-acc eat-neg-past
‘The goat did not eat a flower’
(213) Ke9i ye-me-di [bir 9i9eg-i] D P subj V-neg [bir N-acc]
goat eat-neg-past a flower-acc
'The goat did not eat a flower'
(214) Ke9i bir 9i9ek ye-me-di. DPsu b j [bir N] V-neg
Goat a flower eat-neg-past
'The goat did not eat a flower'
201
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(215) Ke?i 9i?ek ye-me-di. D P S U bj [N] V-neg
Goat flower eat-neg-past
'The goat did not eat a flower'
7.1.1 Participants
148 children and 139 adults participated in the experiments. They were all native
speakers of Turkish, belonged to middle or upper middle class families and were living in
the two big metropolitan cities (Istanbul and Izmit) in the North-west of Turkey. Children
were drawn from four different kindergartens. They were all monolingual children who
had a normal language development pattern as reported by their teachers. Upon their
completion of the test, they were awarded with a color sticker. Adult participants were
undergraduate students at Bogazi9i University in Istanbul. Thirty of them participated
voluntarily upon an announcement that they received in a class they were attending, and
they did not receive anything for their participation. The others were recruited through
the subject pool of the Psychology Department at Bogaziqu University and received extra
credit in a psychology class that they were taking as a requirement for that semester.
They came from various majors.
Each sentence structure was tested on a different group of participants to avoid confusion.
There were five object types and five groups of people per age.
Table 15 presents the number, age range, and the mean age of participants that were
distributed to the sentence structures. In this experiment, I included one more age group
(six-year-olds) to the child group. Children whose ages range between 3;3 and 4;5 were
202
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
included in the youngest age group. Children who were between the ages 4;6 and 5;6
were included in the 5-year age group. Children who were 5;7 and older represented the
older age group. Adult participants’ age ranged between the ages 17-23.
Table 15: Participants in Experiment 4 (accusative-marked objects)
Age Group Mean Age Age range No=
[Bir N-acc] D P sub j V-neg 3 4;2 3;9-4;5 10
5 5;0 4;6-5;5 10
6 6 ;2 5;10-6;6 9
Adult 24
DPsubj [bir N-acc] V-neg 4 4;0 3;3-4;4 10
5 5;0 4;7-5;4 9
6 6 ;l 5;7-6;5 11
Adult 32
DPsubj V-neg [bir N-acc] 4 3;9 3;5-4;4 10
5
5;1 4;6-5;6 11
6 6 ;0 5;10-6;6 10
Adults 25
Table 16: Participants in Experiment 4 (non-case-marked objects)
Age Group Mean Age Age range V o -
DPsubj [bir N] V-neg 4 4;0 3;0-4;4 10
5 5;5 4;8-5;6 11
6
6;1 5;8-6;6 8
Adult 33
DPsubj [N] V-neg 4 3;8 3;5-4;3 9
5 5;0 4;8-5;4 10
6 6 ;0 5;7-6;4 10
Adult 25
7.1.2 Procedure
A truth-value judgment task (Crain & Thornton, 1998) was used. Two experimenters
conducted the experiment at kindergartens in a quiet room that was reserved for the
203
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
experiment. One of the experimenters acted out scenes in front o f the child using small
toys. The other experimenter used a hand-puppet, Kermit the Frog, who was introduced
to children prior to the game. Children were tested individually and were told that they
would play a game in which they would hear a story and then a puppet would tell what
happened at the end of the story. If the puppet was right about his comment, he would be
awarded with two candies. If the puppet failed to make the correct comment, he would
receive only one candy. In the latter case, the child was asked why the puppet was wrong
and was expected to give a relevant explanation for his or her judgment. Children were
told that their task in this game was to determine whether or not the puppet gave the
correct answer and to determine how many candies he would receive each time. The
puppet was rewarded even when he was judged to be wrong because children liked
giving him candies and they could have had the tendency to say that he was right all the
time if they were not giving him anything when he was wrong. Two candies, rather than
two different kinds of rewards for true and false responses were used to shorten the
decision time of the child and to avoid personal preferences/biases o f children towards
different rewards.
The scenarios involved four major characters: Seda, her brother Ahmet, their
friends/neighbors Osman and Ayse. The venue for the events was Seda’s kitchen and the
backyard of her house. Seda’s kitchen included a dinner table, a stove, a sink, two bottles
of soda, two plates and other kitchen objects. She also had a cat and a mouse in the
kitchen. Next to the kitchen, there was a garden with flowers, two horses, two goats and a
tree. In the scenarios, the characters played various games and performed everyday
204
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
activities such as eating, washing the dishes, riding horses. The activities were presented
in a logical order, as if they were following one another, rather than being presented as
different, independent scenes. All the activities and filler/control scenarios were
randomized across different versions. In one version, for example, they moved the chairs
to the garden, played a jumping game, and then ate their dinner. In the other version, they
ate their meal, rode horses, and then played the jumping game in the backyard and moved
the chairs back to the kitchen.
All the characters, objects and locations were introduced to the children prior to the game
and the children were allowed to discover them by playing with them for a while if they
wanted to. While the experimenter was acting out the actual scenes, children were asked
to help the experimenter so that they would become involved and pay more attention to
the scenes. They were asked to set the table before the dinner scene, for example, or they
were asked to choose where the characters hid in the hide and seek game.
Every scenario included a main agent, two objects to act upon, alternative objects and
alternative agents. A sample scenario and the test sentences were presented in (216). In
this scenario, the goal of the agent, the goat, was to eat both flowers. Due to an
unexpected problem (he was not tall enough to reach one of the flowers), he can eat only
one flower. The other one is eaten by another agent, who is tall enough to reach the other
flower. In this scenario, the narrow scope reading (i.e., it is not the case that the goat ate a
flower) is not possible because a flower is eaten. The statements must be rejected if the
subject had only the narrow scope interpretation. The number interpretation (i.e., it is not
205
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the case that the goat ate one flower) must be rejected as well, since he ate exactly one
flower. The wide scope reading {i.e., there is a flower such that the goat didn’t eat it) is
the only reading that is acceptable.
(216) A sample scenario: The goat eating flowers
Karm ag olan kegi bahgede dolagip yiyecek ariyormug. Bakmig otlarin
arasinda iki gigek gormiig. Ama kegi gigekyemeyi hig sevmiyormug.
Qigeklerin yamndan gegmig, etrafa bakmig, yiyecek bagka bir gey bulmaya
caligmig ama bagka higbirgey yokmug. O da ag kalmaktansa en iyisi
gigekleri yiyeyim demig. Gidip yakindaki gigegi afiyetle yemig. Karm hala
gok agmig. Biraz yukarida biryerde bir gigek daha varmig. Ama o gigege
bir tiirlii boyu yetigmemig. O da onu birakmig eve donmiig. O gittikten
sonra bir at gelmig, kalan gigegi de o at yemig. Atm boyu uzun oldugu igin
yukaridaki gigeklere bile yetigebiliyormug.
'The goat, which is still hungry, is searching the garden to find something
to eat. He sees two flowers among the grass. But he didn’t like flowers. So
he keeps looking for some other things to eat. But he could not find
anything. So he had to eat the flowers. He ate one of the flowers. He was
still so hungry. So he wants to eat the other one too. But the other flower
was somewhere high. He tried to reach it, but no matter how hard he tried
he could not get it. In the end he gave up and returned home. A horse that
was passing by finds the flower after the goat left and ate it. He could
reach the flower because he was a very tall horse.'
(217) Test sentences
a. KERMIT: Ke?i bir 9i?eg-i ye-me-di. DPS U b j [bir N-acc] V-neg
goat a flower-acc eat-neg-past
'The goat did not eat a flower'
206
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
b. KERMIT: Bir gigeg-i kegi ye-me-di. [bir N-acc] D P sub j V-neg
a flower-acc goat eat-neg-past
The goat did not eat a flower'
c. KERMIT: Kegi ye-me-di bir gigeg-i. D P sub j V-neg [bir N-acc]
goat eat-neg-past a flower-acc
'The goat did not eat a flower'
d. KERMIT: Kegi bir gigek ye-me-di. D P S U b j [bir N] V-neg
goat a flower eat-neg-past
'The goat did not eat a flower'
e. KERMIT: Kegi gigek ye-me-di. D P sub j [N] V-neg
goat flower eat-neg-past
'The goat did not flower-eat'
f. CHILD: Accept (wide reading) / Reject (narrow reading)
The sentences that the puppet would produce were recorded prior to the game and played
from a small laptop computer so that all the subjects could be provided with the same
intonation. The puppet was played with the second experimenter, whose task was to
animate the puppet and also to play the test sentences recorded in the computer. The
puppet was located behind the laptop monitor, as if he was hiding. Children were told
207
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
that the puppet was very shy and was not very comfortable in talking to children, so what
he would say was recorded prior to the game, as it was too obvious that the sound was
coming from the computer, not from the puppet. Children did not have any problems with
the setting and interacted with the puppet.
With adult participants, the same setting and the recordings were used but the puppet part
was excluded. They were tested in a class-room in groups of up to 15 students circled
around a table where the toys were placed. The same scenes were acted out in front of
them, and the test sentences were played at a computer. They were given a sheet, where
the scenes were labeled (e.g., “hide-and-seek”) and they circled the words “right” or
“wrong” across each scene label right after they heard the test sentence. They were not
provided with the sentences on the sheet so that they can make judgments only on what
they “hear,” rather than reading and marking the answer without listening to the
sentences. In that way, they were tested under the same conditions that were used for
children.
In all the scenarios, there were two objects (i.e., two plates to wash, two goats to feed,
two flowers to eat etc.), one of them was acted upon, the other one was not acted upon by
the major agent. Participants heard six test sentences (all the same object type) and five
filler / control stories (See Appendix 4 for the complete list of the sentences). The filler
stories had sentences with intransitive verbs (i.e., dii§- ‘fall,’ otur- ‘sit’) but included a
location or source (i.e., attan ‘from a horse,’ sandalyeye ‘on a chair’) so that the number
of words per sentence could be the same as the test sentences.
208
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Prior to the test, children heard three training/warm-up sentences, which were uttered
accompanying a hide-and-seek game. The aim of the training stories was to teach the
child the rules of the game and also to show that the puppet can be wrong; therefore they
were simpler than the actual test sentences. If a child failed in more than one trial
sentence, s/he was not allowed to continue the test. Similarly, children who failed in more
than one filler sentence and who gave irrelevant responses to the “why” questions, were
discarded.
Each sentence was played twice and the sentential stress fell on the verb. The indefinite
article/numeral bir was not stressed and the final-r was dropped so that the numeral
interpretation of the word is avoided at least to some extent.
Each yes response (or acceptance) corresponded to a wide scope reading. The no
responses (or rejections) were interpreted as a narrow scope reading of the indefinite
objects. With the accusative marked indefinites, then, which had wide scope, all the right
responses were yes responses. A possible yes bias was controlled by the fillers. With the
non-case marked objects, the expected responses were “rejection,” and a possible
response bias is controlled by the fillers again.
209
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Two types of analyses were conducted on the data. First, the mean scores of each object
type were compared within and across age groups. Each subject had a score of wide-
score reading calculated out of 6 . In the second analysis, the subjects were divided into
three groups according to their response tendency. Subjects whose score is 5-6 was
categorized as wide pattern. Those subjects that have scores of 0-1 were included in the
narrow pattern. The other scores (2, 3, 4) were labeled as mixed pattern subjects.
7.1.3 Results
7.1.3.1 Mean scores
The results of adults are presented in Table 17. The first part of the table displays the
results of the case marked objects. As seen in the table, adults do not treat the two
accusative marked objects differently, i.e., they all have a wide scope reading with
respect to negation in the majority of the cases: For adults overtly scrambled and adjacent
objects have wide scope reading over negation in 97% of the cases. Similarly, postverbal
objects are assigned wide scope reading in 93% of the instances. This picture suggests
that the accusative case is considered a clue for the position of the object, and thus the
interpretation, by adults.
210
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(215) Kegi 9iGoat flower eat-neg-past
'The goat did not eat a flower'
7.1.1 Participants
148 children and 139 adults participated in the experiments. They were all native
speakers of Turkish, belonged to middle or upper middle class families and were living in
the two big metropolitan cities (Istanbul and Izmit) in the North-west of Turkey. Children
were drawn from four different kindergartens. They were all monolingual children who
had a normal language development pattern as reported by their teachers. Upon their
completion of the test, they were awarded with a color sticker. Adult participants were
undergraduate students at Bogazi9i University in Istanbul. Thirty of them participated
voluntarily upon an announcement that they received in a class they were attending, and
they did not receive anything for their participation. The others were recruited through
the subject pool of the Psychology Department at Bogazifi University and received extra
credit in a psychology class that they were taking as a requirement for that semester.
They came from various majors.
Each sentence structure was tested on a different group of participants to avoid confusion.
There were five object types and five groups of people per age. Table 15 presents the
number, age range, and the mean age of participants that were distributed to the sentence
structures. In this experiment, I included one more age group (six-year-olds) to the child
202
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
group. Children whose ages range between 3;3 and 4;5 were included in the youngest age
group. Children who were between the ages 4;6 and 5;6 were included in the 5-year age
group. Children who were 5;7 and older represented the older age group. Adult
participants’ age ranged between the ages 17-23.
Table 15: Participants in Experiment 4 (accusative-marked objects)
Age Group Mean Age Age range No=
[Bir N-acc] D P sub j V-neg 3 4;2 3;9-4;5 10
5 5;0 4;6-5;5 10
6 6 ;2 5;10-6;6 9
Adult 24
DPsubj [bir N-acc] V-neg 4 4;0 3;3-4;4 10
5 5;0 4;7-5;4 9
6
6 ;l
5;7-6;5 11
Adult 32
DPsubj V-neg [bir N-acc] 4 3;9 3;5-4;4 10
5 5;1 4;6-5;6 11
6 6 ;0 5;10-6;6 10
Adults 25
Table 16: Participants in Experiment 4 (non-case-marked objects)
Age Group Mean Age Age range No=
DPsubj [bir N] V-neg 4 4;0 3;0-4;4 10
5 5;5 4;8-5;6 11
6 6;1 5;8-6;6 8
Adult 33
DPsubj [N] V-neg 4 3;8 3;5-4;3 9
5 5;0 4;8-5;4 10
6 6 ;0 5;7-6;4 10
Adult 25
203
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
7.1.2 Procedure
A truth-value judgment task (Crain & Thornton, 1998) was used. Two experimenters
conducted the experiment at kindergartens in a quiet room that was reserved for the
experiment. One of the experimenters acted out scenes in front of the child using small
toys. The other experimenter used a hand-puppet, Kermit the Frog, who was introduced
to children prior to the game. Children were tested individually and were told that they
would play a game in which they would hear a story and then a puppet would tell what
happened at the end of the story. If the puppet was right about his comment, he would be
awarded with two candies. If the puppet failed to make the correct comment, he would
receive only one candy. In the latter case, the child was asked why the puppet was wrong
and was expected to give a relevant explanation for his or her judgment. Children were
told that their task in this game was to determine whether or not the puppet gave the
correct answer and to determine how many candies he would receive each time. The
puppet was rewarded even when he was judged to be wrong because children liked
giving him candies and they could have had the tendency to say that he was right all the
time if they were not giving him anything when he was wrong. Two candies, rather than
two different kinds of rewards for true and false responses were used to shorten the
decision time of the child and to avoid personal preferences/biases of children towards
different rewards.
204
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The scenarios involved four major characters: Seda, her brother Ahmet, their
friends/neighbors Osman and Ayse. The venue for the events was Seda’s kitchen and the
backyard of her house. Seda’s kitchen included a dinner table, a stove, a sink, two bottles
of soda, two plates and other kitchen objects. She also had a cat and a mouse in the
kitchen. Next to the kitchen, there was a garden with flowers, two horses, two goats and a
tree. In the scenarios, the characters played various games and performed everyday
activities such as eating, washing the dishes, riding horses. The activities were presented
in a logical order, as if they were following one another, rather than being presented as
different, independent scenes. All the activities and filler/control scenarios were
randomized across different versions. In one version, for example, they moved the chairs
to the garden, played a jumping game, and then ate their dinner. In the other version, they
ate their meal, rode horses, and then played the jumping game in the backyard and moved
the chairs back to the kitchen.
All the characters, objects and locations were introduced to the children prior to the game
and the children were allowed to discover them by playing with them for a while if they
wanted to. While the experimenter was acting out the actual scenes, children were asked
to help the experimenter so that they would become involved and pay more attention to
the scenes. They were asked to set the table before the dinner scene, for example, or they
were asked to choose where the characters hid in the hide and seek game.
Every scenario included a main agent, two objects to act upon, alternative objects and
alternative agents. A sample scenario and the test sentences were presented in (216). In
205
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
this scenario, the goal of the agent, the goat, was to eat both flowers. Due to an
unexpected problem (he was not tall enough to reach one of the flowers), he can eat only
one flower. The other one is eaten by another agent, who is tall enough to reach the other
flower. In this scenario, the narrow scope reading (i.e., it is not the case that the goat ate a
flower) is not possible because a flower is eaten. The statements must be rejected if the
subject had only the narrow scope interpretation. The number interpretation (i.e., it is not
the case that the goat ate one flower) must be rejected as well, since he ate exactly one
flower. The wide scope reading (i.e., there is a flower such that the goat didn’t eat it) is
the only reading that is acceptable.
(216) A sample scenario: The goat eating flowers
Karm ag olan kegi bahgede dolagip yiyecek any or mag. Bakmig otlarin
arasinda iki gigek gormtig. Ama kegi gigekyemeyi hig sevmiyormug.
Qigeklerin yamndan gegmig, etrafa bakmig, yiyecek bagka birgey bulmaya
caligmig ama bagka higbirgey yokmug. O da ag kalmaktansa en iyisi
gigekleri yiyeyim demig. Gidip yakindaki gigegi afiyetle yemig. Karm hala
gok agmig. Biraz yukarida bir yerde bir gigek daha varmig. Ama o gigege
bir tiirlii boyu yetigmemig. O da onu birakmig eve donmug. O gittikten
sonra bir at gelmig, kalan gigegi de o at yemig. Atm boyu uzun oldugu igin
yukaridaki gigeklere bile yetigebiliyormug.
'The goat, which is still hungry, is searching the garden to find something
to eat. He sees two flowers among the grass. But he didn’t like flowers. So
he keeps looking for some other things to eat. But he could not find
anything. So he had to eat the flowers. He ate one of the flowers. He was
still so hungry. So he wants to eat the other one too. But the other flower
was somewhere high. He tried to reach it, but no matter how hard he tried
he could not get it. In the end he gave up and returned home. A horse that
was passing by finds the flower after the goat left and ate it. He could
reach the flower because he was a very tall horse.'
206
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(217) Test sentences
a. KERMIT: Kecpi bir 9i?eg-i ye-me-di. DPS U b j [bir N-acc] V-neg
goat a flower-acc eat-neg-past
'The goat did not eat a flower'
b. KERMIT: Bir 9i?eg-i ke9i ye-me-di. [bir N-acc] DPS U b j V-neg
a flower-acc goat eat-neg-past
'The goat did not eat a flower'
c. KERMIT: Ke9i ye-me-di bir 9i9eg-i. DPsu b j V-neg [bir N-acc]
goat eat-neg-past a flower-acc
'The goat did not eat a flower'
d. KERMIT: Ke9i bir 9i9ek ye-me-di. D P sub , [bir N] V-neg
goat a flower eat-neg-past
'The goat did not eat a flower'
e. KERMIT: Ke9i 9i9ek ye-me-di. D P su b j [N] V-neg
goat flower eat-neg-past
'The goat did not flower-eat'
f. CHILD: Accept (wide reading) / Reject (narrow reading)
207
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The sentences that the puppet would produce were recorded prior to the game and played
from a small laptop computer so that all the subjects could be provided with the same
intonation. The puppet was played with the second experimenter, whose task was to
animate the puppet and also to play the test sentences recorded in the computer. The
puppet was located behind the laptop monitor, as if he was hiding. Children were told
that the puppet was very shy and was not very comfortable in talking to children, so what
he would say was recorded prior to the game, as it was too obvious that the sound was
coming from the computer, not from the puppet. Children did not have any problems with
the setting and interacted with the puppet.
With adult participants, the same setting and the recordings were used but the puppet part
was excluded. They were tested in a class-room in groups of up to 15 students circled
around a table where the toys were placed. The same scenes were acted out in front of
them, and the test sentences were played at a computer. They were given a sheet, where
the scenes were labeled (e.g., “hide-and-seek”) and they circled the words “right” or
“wrong” across each scene label right after they heard the test sentence. They were not
provided with the sentences on the sheet so that they can make judgments only on what
they “hear,” rather than reading and marking the answer without listening to the
sentences. In that way, they were tested under the same conditions that were used for
children.
In all the scenarios, there were two objects (i.e., two plates to wash, two goats to feed,
two flowers to eat etc.), one of them was acted upon, the other one was not acted upon by
208
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the major agent. Participants heard six test sentences (all the same object type) and five
filler / control stories (See Appendix 4 for the complete list of the sentences). The filler
stories had sentences with intransitive verbs (i.e., dii§- ‘fall,’ otur- ‘sit’) but included a
location or source (i.e., attan ‘from a horse,’ sandalyeye ‘on a chair’) so that the number
of words per sentence could be the same as the test sentences.
Prior to the test, children heard three training/warm-up sentences, which were uttered
accompanying a hide-and-seek game. The aim of the training stories was to teach the
child the rules of the game and also to show that the puppet can be wrong; therefore they
were simpler than the actual test sentences. If a child failed in more than one trial
sentence, s/he was not allowed to continue the test. Similarly, children who failed in more
than one filler sentence and who gave irrelevant responses to the “why” questions, were
discarded.
Each sentence was played twice and the sentential stress fell on the verb. The indefinite
article/numeral bir was not stressed and the final-r was dropped so that the numeral
interpretation of the word is avoided at least to some extent.
Each yes response (or acceptance) corresponded to a wide scope reading. The no
responses (or rejections) were interpreted as a narrow scope reading of the indefinite
objects. With the accusative marked indefinites, then, which had wide scope, all the right
209
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
responses were yes responses. A possible yes bias was controlled by the fillers. With the
non-case marked objects, the expected responses were “rejection,” and a possible
response bias is controlled by the fillers again.
Two types of analyses were conducted on the data. First, the mean scores of each object
type were compared within and across age groups. Each subject had a score of wide-
score reading calculated out of 6 . In the second analysis, the subjects were divided into
three groups according to their response tendency. Subjects whose score is 5-6 was
categorized as wide pattern. Those subjects that have scores of 0-1 were included in the
narrow pattern. The other scores (2, 3,4) were labeled as mixed pattern subjects.
7.1.3 Results
7.1.3.1 Mean scores
The results of adults are presented in Table 17. The first part of the table displays the
results of the case marked objects. As seen in the table, adults do not treat the two
accusative marked objects differently, i.e., they all have a wide scope reading with
respect to negation in the majority of the cases: For adults overtly scrambled and adjacent
objects have wide scope reading over negation in 97% of the cases. Similarly, postverbal
objects are assigned wide scope reading in 93% of the instances. This picture suggests
that the accusative case is considered a clue for the position of the object, and thus the
interpretation, by adults.
210
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 17: Results of Experiment 4 (adults, mean scores)
Object Type No= Indefinite > Negation
[Bir N-acc] D P subj V-neg
DPsubj [bir N-acc] V-neg
DPsubj V-neg [bir N-acc]
DPsubj [bir N] V-neg
24 5.83 (97%, SE=.078)
33 5.82 (97%, SE=.092)
25 5.60 (93%, SE=.208)
28 2.04 (34 %, SE=.413)
25 .24 (4 %, SE=.119) DPsubj [N1 V-neg
In contrast, when the object does not bear overt case morphology, it has narrow scope
reading with respect to negation. Adults assign wide scope reading to bir N objects in
34% of the cases, which is significantly lower than chance (t(27)=-2.335,/><.05) and thus
shows that they have a preference for narrow scope, rather than assigning no scope, or
arbitrary scope to them. Note that this is different from the results of Experiment 3 where
the narrow scope assignment was recorded around chance (47%). In bare objects, the
wide scope reading drops to only 4% of the cases.
A Factorial ANOVA was conducted on the comprehension scores of adults to see what
determines the scope assignment in objects. Case, bir, word order (preverbal vs.
postverbal) and adjacency to verb were entered into the analysis as possible factors. The
results indicated that case had a main effect (F(l,131)=200.8,/><.001) on the
interpretation of indefinite objects, i.e., adults treated the case-marked and non-case
marked objects differently. The numeral bir still had a main effect (F(l,131)=31.741,
/?<.001), ie., bir N(-acc) objects and bare objects were also treated differently by adults.
Word order (preverbal and postverbal), and adjacency to verb, however, did not have a
main effect (F(l,131)=.653, n.s., F(l,131)=.002, n.s.).
211
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In summary, the accusative case and bir determine the scope assignment in adults’
indefinite objects. The linear position of the object with respect to negation does not
have any influence on the scope assignment. Therefore no isomorphism is observed. The
response pattern results, which will be reported in the next section, together with the
children’s response pattern results present a similar picture.
The results of children are presented in Table 19 and Figure 14. As seen in the tables,
four-year-old children treat all the object types alike (F(4, 44)=.745, p=.567). They have
the wide scope reading of case marked objects in 25%, 18% and 33% of the cases in
preverbal non-adjacent, preverbal adjacent and postverbal object types respectively. First,
the scores of both preverbal objects are below chance (i(9)=2.355,/?<.05, t(9)=3.243,
p< 05 two object types respectively), so children’s accusative marked indefinites in the
preverbal position have narrow scope. The wide scope assignment to the postverbal
object, however, is around chance (i(9)=1.732,^=.117). Second, the scores are not
different from each other so the children treat the accusative object types alike regardless
of their adjacency and word order properties. Note that the highest raw score is that of the
postverbal objects. The lowest wide scope assignment is to the adjacent accusative
marked objects, though the difference between the object types is not significant
(F(2,27)=.564, p=.516), and the response pattern results suggest that children’s
preference for wide scope reading of postverbal objects is not any higher than that of
others (See Table 20 below). Four and five year old children’s treatment of bir N objects
is quite similar to their treatment of accusative marked objects. Namely, the youngest
group has wide scope reading of this object type in 20% of the cases. Bare scores, which
212
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
are shown in the same table, indicate that children have a wide scope reading of bare
objects in 10% or fewer cases. Therefore, children’s comprehension of bare objects is
adult-like.
Table 18: Results of Experiment 4 (accusative-marked, children, mean scores)
Object Type N= Mean Age Indefinite > Negation
[Bir N-acc] D P sub j V-neg 10 4;2 1.50 (25%, SE=.637)
10 5;0 3.10(52% , SE=.862)
9 6 ;2 4.80 (80%, SE=.291)
DPsubj [bir N-acc] V-neg 10 4;0 1.10(18%, SE=.586)
9 5;0 3.11 (52%, SE=.754)
11
6;1
3.45 (57%, SE=.6 6 6 ))
DPsubj V-neg [bir N-acc] 10 3;9 2.00 (33%, SE=.577)
11 5;l
2.73 (41%, SE=.752)
10 6 ;0 3.90 (65%, SE=.640)
Table 19: Results of Experiment 4 (non-case-marked, children, mean scores)
Object Type N= Mean Age Indefinite > Negation
DPsubj [bir N] V-neg 10 4;0 1.20 (20%, SE=.80)
11 5;5 2.09 (34%, SE=.625)
8 6 ;l 2 (33%, SE=.906)
DPsubj [N] V-neg 9 3;8 .56 (9%, SE=.242)
10 5;0 .30 (5%, SE=.213)
10 6 ;0 .60 (10%, SE=.40)
It would be interesting to have the wide scope reading in the postverbal objects higher
because, as discussed in the previous chapter, post verbal objects are clearly outside the
domain of VP, in contrast to the preverbal objects. Earlier wide scope assignment to post
verbal objects would be in line with the c-command account of Lidz and Musolino
(2002). Similarly, the low scores in the adjacent accusative marked indefinites would
support an account that proposes that adjacent accusative indefinites can have narrow
scope (c.f. Kelepir, 2000 and the account adopted here). The results, however, do not
213
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
show any statistical difference. Moreover, the response patterns do not display any
difference in scope assignment.
The five-year-olds have wide scope of the preverbal and adjacent type accusative objects
in 52% of the cases. Postverbal objects are assigned wide scope in 41% of the instances.
For five-year-olds bir N objects have wide scope in 34% of the cases. With bare objects
the wide scope assignment drops to 5%. At 5 years of age, children start distinguishing
the bare objects and non-bare objects as also seen in Figure 14.. A factorial ANOVA
conducted on five-year-olds’ scores suggest that case morphology does not have an effect
(F( 1,47)=.297, n.s.), but bir has an effect trend F(l,47)=3.627, p=.06).
Figure 14: Results of Experiment 4 (Object types and age groups)
100 -i
C O
C D
1 _
o
o
C O
-4 -»
Q .
C D
O
O
<
age 4 age 5 age 6 adults
— ■— bir N-acc XP Vneg
— A— bir N-acc Vneg
— • — Vneg bir N-acc
— x— bir N Vneg
- - O - ■ N Vneg
Age groups
As seen in Figure 14, bir N objects pattern similarly to the bir N-acc objects although
they are not case marked. This trend shows that bir has some impact on scope assignment
(c.f. Su, 2001). It is important to note, however, that although bir N-acc and bir N objects
214
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
without case pattern similarly, neither of them have wide scope above chance level.
Therefore, the objects that have bir and bare objects are differentiated but bir does not
result in an early wide scope assignment. Rather, bare objects have narrow scope and the
other object types are scopeless. In the earlier age group, however, they were clearly
assigned narrow scope. Children, then, around the age of 5 stop assigning default scope
to bir N objects, but they do not assign wide scope either. Another important point is that
children’s scope assignment to bir N objects are not different from adults’ scope of bir N
objects (around 30%). Therefore the picture shows that bir N-acc objects are behaving
like bir N objects (which is adult like). The cardinal number theories predict it to be the
other way around, namely, bir N objects are expected to behave like bir N-acc objects
having wide scope. Therefore the trend observed in bir N objects in the five-year-old
group cannot be interpreted as a result supporting the Lexical Factor hypothesis (Su,
2001).
When the children are 6 years old, preposed objects have 80% wide scope. No increase is
observed in adjacent objects, which are recorded to have around 57% wide scope.
Postverbal objects have wide scope around 65% of the cases. At this age, the bir N
objects do not pattern like the bir N-acc objects anymore. A high increase is observed in
the preverbal objects in the 6 year olds’ speech. It is interesting to note that no such
increase is recorded for other object types. 6 year olds have neither a main effect of case
(F(l,45)=9.797, n.s.) nor bir (F(l,45)=2.46, n.s.). So, even at the age of 6;0, children’s
treatment of case is not adult-like despite the increase in wide scope reading.
215
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
7.1.3.2 Response patterns
The results of the response patterns that are presented in Table 20 and Figure 15 to The
response patterns are not very different from what we saw in the score analysis. The only
important difference between the two approaches to data is observed in the 6 year olds’
wide scope reading of the preposed objects. Although the scores were calculated as high
as 80%, the response pattern analyses showed that only 50 % of the subjects accepted the
wide scope reading consistently. The others had a mixed pattern. Therefore the children’s
treatment of the case-marked objects is different from the adults who all have wide scope
reading even at the age of six.
Figure 19 below further suggest that children do not have a wide scope interpretation in
the accusative marked indefinites at the age of four. 80-90 % of the four-year-old
children have either narrow scope or a mixed response pattern. Note that in preposed and
adjacent object types the narrow scope children are much more than the mixed pattern
children (See Figure 15 and Figure 16).
With postverbal objects however, half of the children show a mixed pattern which
suggest that they apply a guessing strategy rather than having a consistent preference for
one reading (Figure 17). Remember that their wide scope reading was around chance in
the analysis of the scores as well.
216
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 15 : Response pattern in bir N-acc DP V-neg structure.
□ mixed
□ narrow
■ wide
age 4 age 5 age 6 adults
Figure 16: Response pattern in DP bir N-acc V-neg structure.
age 4
□ mixed
□ narrow
■ wide
age 5 age 6 adults
Figure 17: Response pattern in DP V-neg bir N-acc structure
age 4
□ mixed
□ narrow
I wide
age 5 age 6 adults
It is difficult to evaluate these results. One reason may be that the postverbal objects are
higher than negation in the hierarchy and in contrast to the preverbal objects, their
position is much more salient to children. If this is the case, then the hierarchical ordering
217
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
does play a role in children’s scope assignment (c.f. Lidz & Musolino, 2002). Another
reason may be the non-canonical word order. It is possible that children cannot interpret
the SVO word order. Because they do not know what kind of interpretation they should
assign, they assign scope arbitrarily. The reason is more likely to be the latter one, as we
would expect a higher wide scope assignment rate if the reason were their hierarchical
position.
A support for children’s difficulty with non-canonical word order comes from Batman-
Ratrosyan (2003), who observes that children cannot understand the sentences when
objects are in non-canonical positions. She does not deal with scope of indefinite objects
though. Rather she looks at the subject-object distinction and the observation that
children have difficulty with non-canonical word orders is based on the non-case marked
objects. In her account, once the object is case-marked, children do not have any
difficulty in recognizing it as an object (this is originally proposed by Slobin and Bever
(1982)) and interpreting the SVO order. The findings I report here show that although the
case marked objects can be recognized as objects in non-canonical orders (c.f. Slobin &
Bever, 1982, Batman-Ratrosyan, 2003), they cannot have adult-like interpretations in
terms of scope o f the object with respect to other constituents. These results show that,
confirming my hypothesis, children treat the accusative case as a formal case marker
(thus recognize the accusative marked objects as ‘objects’ as opposed to subjects), but
they do not yet know that the overt case morphology shows that these objects are
displaced and thus they should have a particular interpretation. Their behavior with the
218
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
postverbal objects show that neither overt scrambling nor the case morpheme as an
indicator of the scrambled position can result in a wide scope reading in child grammar.
At 5 years, more children with an adult-like wide scope reading of the accusative-marked
objects, but their number is still not more than the subjects who have mixed or narrow
reading patterns. Even at the end of the period analyzed, children do not reach the adult
level in their scope assignments, but a decrease is observed in the number of children
who assign narrow scope to the accusative-marked objects (See Figure 15, Figure 16 and
Figure 17 above).
Table 20: Response Pattern in Experiment 4 (accusative-marked objects)
Object Type Age Accept
(wide)
Reject
& Mixed
Reject
(narrow)
Mixed
[Bir N-acc] D P su b j V-neg 4 2 (2 0 %) 8 (80%) 6 (60%) 2 (2 0 %)
5 5 (50%) 5 (50%) 3 (30%) 2 (2 0 %)
6 5 (50%) 5 (50%) 0 5 (50%)
adult 24 (100%) 0 0 0
DPsubj [bir N-acc] V-neg 4 1 ( 1 0%) 9 (90%) 8 (80%) 1 ( 10%)
5 3 (33%) 6 (6 6 %) 3 (33%) 3 (33%)
6 4 (36%) 7 (64%) 4 (36%) 3 (28%)
adult 31 (94%) 2 (6 %) 0 2 (6 %)
DPsubj V-neg [bir N-acc] 3 1 ( 1 0%) 9 (90%) 4 (40%) 5 (50%)
5 4 (36%) 7 (64%) 5 (45%) 3 (18%)
6 6 (60%) 4 (40%) 1 ( 1 0%) 3 (30%)
adult 24 (96%) 1 (4%) 1 (4%) 0
Table 21 below shows the response pattern of children in non-case marked objects. As
seen in the table, a majority of the children have narrow reading of the non-case marked
objects. With bare objects, their narrow scope preference is obvious, no child in the 4 or 5
year group have wide scope reading, only one six year old subject has a wide scope
219
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
tendency with bare objects. With birN objects, their behavior is similar to the case-
marked objects, they have narrow scope tendency but some children still have wide scope
interpretation of objects in the majority of their responses.
Table 21: Response Patt em in Experiment 4 (non-case-marked objects)
Object Type Age Accept Reject Reject Mixed
(Wide) & Mixed (Narrow)
DPsubj [bir N] V-neg 4 2 (2 0 %) 8 (80%) 8 (80%) 0
5 2(18% ) 9 (82%) 5 (45%) 4 (36%)
6 2 (25%) 6 (75%) 5 (62%) 1 (13%)
adult 6 (2 1 %) 22 (79%) 23 (58%) 6 (2 1 %)
DPsubj [N] V-neg 4 0 9 (100%) 8 (89%) 1( 1 1%)
5 0 10 ( 1 0 0%) 9 (90%) 1 ( 1 0%)
6 1 ( 1 0 %) 9 (90%) 8 (80%) 1( 1 0%)
adult 0 25 (100%) 23 (92%) 2 (8%)
Figure 18: Response pattern in DP bir N V-neg
100 -|
80 -
60 -
40 ■
20 -
o -
age 4 age 5 age 6 adults
The response patterns are not very different from what we saw in the score analysis. The
only important difference between the two approaches to data is observed in the 6 year
olds’ wide scope reading of the preposed objects. Although the scores were calculated as
high as 80%, the response pattern analyses showed that only 50 % of the subjects
accepted the wide scope reading consistently. The others had a mixed pattern. Therefore
220
□ mixed
□ narrow
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the children’s treatment of the case-marked objects is different from the adults who all
have wide scope reading even at the age of six.
Figure 19: Response pattern in bare objects
100 -i
80 -
60 •
40 ■
20 ■
0 1 1 ^ 1 1
age 4 age 5 age 6 adults
7.1.4 Discussion and Conclusion
In Experiment 4 we saw that the case morphology played a major role in scope
assignment in adults’ speech and children are in the process of acquiring the relationship
between overt case morphology, position of the objects and their interpretation. We also
found that bare objects predominantly had narrow scope reading for all subjects. The
results are not surprising, as the theories that I discussed earlier and the account that I
adopted predict them. The interesting result was that some adults (in 34% of the cases)
still had wide scope interpretation of BirN objects that were not case marked despite the
modification in the scenarios. Although there was a big enough difference in the
proportion of wide scope reading in case-marked and non-case marked objects this time
(34% versus 96% (average of three accusative marked object scores)), and the wide scope
reading is lower than chance, the factorial ANOVA test still detected bir as a factor
effecting wide scope. This is not expected in the model I adopted and it is problematic for
221
□ mixed
□ narrow
■ wide
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
those theories that assign the same structure and scope taking property to bir N and bare
N objects as well. Note that the results cannot be attributed to the cardinal number
reading ambiguity anymore because this was fixed in Experiment 4. How could bir N
objects still have wide scope? One reason may be another kind o f bias in the scenarios,
which possibly results in a wide scope reading and which is independent of the number
reading. Let us first have a look at the scenarios that are used in Expeiment 4 and discuss
what may be wrong with them for “narrow scope” interpretation.
In Experiment 4, because the goal of the experiment was to test whether the subjects have
wide scope reading or not, the scenarios were built in such a way that they force a wide
scope reading of the object: A well-defined set of objects were introduced and one o f the
members of this set was not acted upon. We had two objects (e.g., two flowers) and the
goal o f the major agent was to act upon both objects (e.g., eat both flowers). Due to an
unexpected problem (e.g., he was not tall enough to reach one of the flowers), he cannot
achieve his goal. One object is acted upon, but the other one is not (e.g., one out o f two
flowers is eaten). The expected outcome in this scenario was acting upon two objects,
while the actual outcome is that only one object was acted upon. Thus, what is
unexpected is leaving one object behind. Because it is the unexpected (or surprise) part,
the hearer expects the speaker to talk about the object that was not acted upon. The
scenario, drawing the attention to that object, biases the subject towards a wide-scope
222
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
reading42. The problem with this bias is that the use of a non-case marked object to talk
about such a situation becomes infelicitous. By changing the expected outcome, however,
it is possible to have a situation which is more felicitous for non-case-marked objects,
without changing the scope-assignment of the object. One way to change the expected
outcome is to change verbs used in the experiment. Verbs that denote negative ending
such as break, spill, lose can express an opposite expected outcome in a way that will be
discussed below.
Let us look at the scenario in (218) where the goal of the agent is not to act upon (or
affect) any of the objects (i.e., jumping over the chairs without hitting/turning over43 any
of them). Due to an unexpected situation (i.e., she decides to jump backwards and crashes
into the last chair), she cannot achieve her goal.
(218) The jumping game
Ahmet and Seda bir atlama oyunu oynuyorlarmiy Bir sepet ve iki sandalyeyi
yan yana dizmi§ler. Onlari devirmeden iistlerinden atlamaya cahpyorlarmiy
Ay§e sepetin ustiinden atlamiy Birinci sandalyenin ustunden de atlami§. Bu
atlama i§i cok kolaymiy ters donerek bile sandalyeyi devirmeden atlayabilirim"
demiy Arkasim ddnmii§, atlamaya cali§mi§, ama bu sefer becerememi§,
sadalyeye takdmi§, sandalye tie birlikte du§mu§.
Ahmet and Seda are playing a jumping game. The one who can jump over two
chairs without touching them (turning over them) will be the winner. Seda starts
4 2 See Gualmini (2005) for a similar discussion of the two different kinds o f scenarios used in Muslino
(1998).
4 3 The verb devir that is used in this experiment has a negative involuntary causative meaning, i.e., ‘turn
over something unintentionally, accidentally.
223
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the game. She jumps over one of the chairs without a problem. Then she says
‘this is so easy! I can jump even backwards!’ She turns back, and tries to jump.
But this time she hits the chair and the chair is turned over.
(219) Test sentences
a. K ERM IT: Seda bir sandalye devir-me-di. D P su bject [bir N] V-neg
Seda a chair tum-over-neg-past
'Seda did not turn over a chair'
b. K ERM IT: Seda sandalye devir-me-di D P subject [N] V-neg
Seda a chair tum-over-neg-past
'Seda did not chair-tum-over'
c. CHILD: Accept (wide reading) / Reject (narrow reading)
In both scenarios (the ones in Experiment-4 and those in Experiment-5) there are two
objects involved and one of them is acted upon/ affected and the other one is not affected.
The difference between the scenarios is in the expected outcome: In the former one, the
goal of the agent was to affect both objects (i.e. eat both flowers) whereas in the latter one
the goal is to affect none (i.e., not to turn over any chairs). What is important in this
difference is the following: In the former scenarios, the surprise object is the one that was
not acted upon (the hearer expects the speaker to talk about it, not the other object) and
this forces to a wide scope interpretation of the object. In the latter one, it is the object
224
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
that was acted upon (the speaker is expected to make a statement about the object that
was affected).Table 2 2 below presents a summary of the two experiment conditions.
Table 22: Summary o f the conditions in Experiment 4 and 5
Experiment 4
‘the flower story’
Experiment 5
‘the chair story’
Expected Acting out both objects Acting out no objects
out-come: (2 objects out of 2 ) (0 object out of 2 )
Actual out-come: One object is acted upon. One object is acted upon.
Surprise object: The one that was not acted upon The one that was acted upon
Hearer expectation Comment on the object Comment on the object
(attention): that was not acted upon that was acted upon
In the next experiment, children’s and adult’s comprehension of non-case-marked objects
were tested again to see whether the modification of the scenarios discussed in this
section will make any difference in their comprehension.
7.2 Experiment 5: Follow up to Experiment 4
7.2.1 Participants
The participants in this experiment were the same children and adults who participated in
Experiment. See Table 16 in Section 7.1.1 for their details.
225
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
7.2.2 Procedure
The follow-up experiment was conducted with the same methodology (See Section 7.1.2
of Experiment 4) with a modification in the scenarios as discussed in Section 7.1.4.
7.2.3 Results
7.2.3.1 Mean scores
The results of Experiment 5 are given together with Experiment 4 at Table 23 below and
in Figure 20 following it. As seen in the table, the modification of the scenarios has an
effect in the adults’ interpretation. The wide-scope reading of bir N indefinites dropped to
9% from 34% in adults.
Table 23: Results of Experiment 4 and 5 compared (mean scores)
Object Type N= Age Experiment 4 Experiment 5
indefinite > negation
DPS u b j [bir N] V-neg 10 4;0 1 .2 0 (2 0 %) 0.3 (5%)
11 5;5 2.09 (34%) 0.18(3% )
8
6 ;1
2 (33%) 0.5 (8 %)
33 Adult 2.1 (34%) 0.54 (9.1%)
DPsubj [N] V-neg 9 4;0 .56 (9%) 0
10 5;0 .30 (5%) 0
10 6 ;0 .60 ( 10%) 0.2 (3.3%)
25 Adult 2.4 (4%) 0
In children, too, there is a decrease in a similar fashion. As clearly seen in the figure, bir
N objects pattern more like the bare objects as predicted when the felicity conditions are
met. Such a big change is not observed in bare objects, as their wide scope reading was
already low (around 1 0%) in all age groups.
226
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 20: Results of Experiment 4 and 5 compared (mean scores)
100
t/j
a )
i _
o
o
w
- 4 - »
Q .
< D
O
O
<
age 4 age 5 age 6
Age groups
adults
-X — birN (Experiment 4)
O - - bare N (Experiment 4)
- ■ — birN (Experiment 5)
- • — bare N (Experiment 5)
With the Experiment 4, our finding was that both bir and case have impacts on the scope
assignment of indefinite objects. Experiment 5, where the scenarios were changed in such
a way that the wide-scope bias was reduced, showed that the wide scope reading
tendency is not any more than the wide-scope reading assigned to bare nouns. Thus, the
results of Experiment 4 and 5 together suggest that only the case morphology affects
scope assignment in adults’ indefinites. In children, a similar change pattern is observed.
When the felicity conditions are satisfied bir N objects and bare objects are not different
in terms of their scope.
7.2.3.2 Response patterns
The response pattern analysis confirms the results. No subject has a wide scope reading
of non-case-marked objects. As seen in Table 24, a majority of the subjects either reject
227
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the wide scope reading or they have a mixed pattern. No subjects assigns wide scope to
the bir N objects, and in this sense, they are not different from the bare objects.
Table 24: Response Pattern in Experiment 5 (non-case-marked objects)
Object Type n= Age Accept Reject
& Mixed
Reject Mixed
DPsubj [bir N] V-neg 10 4;0 0 10 ( 1 0 0 %) 9 (90%) 1 ( 1 0%)
11 5;5 0 11 ( 1 0 0 %) 11 ( 1 0 0 %) 0
8
6;1
0 8 ( 1 0 0 %) 7 (8 8 %) 1 ( 12%)
28 adults 0 28 ( 1 0 0%) 25 (90%) 3 (10%)
DPsubj [N] V-neg 9 3;8 0 9 (100%) 9 (100%) 0
10 5;0 0 10 ( 1 0 0%) 10 ( 1 0 0 %) 0
10 6 ;0 0 10 ( 1 0 0%) 10 ( 1 0 0 %) 0
25 adults 0 25 (100%) 25 (100%) 0
7.2.4 Discussion and Conclusion
In this chapter I presented two experiments that tested children’s comprehension of
indefinite objects that interact with negation. They revealed that adults recognize the
overt case morphology as an indicator of the position, thus the interpretation of the
object. Children were still in the process of acquiring the relationship between case
morphology and interpretation. In addition to the case morphology, bir had an effect on
the interpretation of the objects. In Experiment 4 and the follow-up Experiment 5, the
degree of bir influence on the scope assignment was reduced and than eliminated.
The results of the experiments that are reported in this chapter are different from that of
the earlier negation experiments reported in Chapter 6 in a predicted way. In the earlier
negation experiments, the children in the youngest age groups had a wide scope
228
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
interpretation of the accusative marked indefinite objects in about 80% of the cases and
the results were similar to the ones reported for Chinese and Spanish. The scores were
quite high when compared to the scores of Dutch and English. Moreover, the results in
the negation experiments and the twice experiments were surprisingly different. While
children failed to assign wide scope to accusative marked objects in twice sentences, they
did much better with negation experiments. In the discussion o f the earlier experiments, I
argued that the results of the negation experiments might be related to a contextual bias.
When two objects were acted upon and one is left behind, the number interpretation and
the wide scope interpretation had the same right response (accept) with the wide scope
indefinites. That is, it is not possible to know whether children are accepting the test
sentences because they attribute them a wide scope reading or a cardinal number (exactly
one) reading. In the new experiments, this problem was eliminated with using two
objects, and as predicted, the results were different. Children had wide scope reading in
around 20% of the cases, which was similar to the Dutch results. Therefore, the earlier
results suggested that Turkish was similar to Chinese-like languages, whose scope
interpretation is claimed to be dependent on the semantic properties of the indefinite
article. The later experiments, however, showed that the earlier findings failed to
represent the real picture and once the test conditions were fixed, the results of Turkish
children were much similar to Dutch-type languages. I argued that the high scores could
be attributed to the number reading of bir, which was not necessarily a structural or
semantic effect. Rather, it was a shortcoming of the experimental design and the
expectations of the participants, which is a pragmatic artifact. The new findings are also
compatible with the results of the twice experiments, and they altogether suggest that
229
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
children’s scope assignment to accusative-marked indefinite objects in both conditions
(adverbs and negation) is governed by the same constraint or developmental property.
Another difference between the earlier findings and the findings in the later experiments
is the scores of the non-case marked objects. In the earlier experiments, children and
adults assigned wide scope reading to the objects although they were not case marked.
The wide scope assignment was as high as 70% in children. The present results suggested
that this finding was also due to the methodological (or pragmatic) reasons and cannot
imply that non-case marked objects can have wide scope reading.
However, it is important to note that bare objects and bir N objects are not treated alike
by the participants. This may be more likely to be a consequence of the structural
differences between bir N objects and bare objects and their interaction with the verb,
rather than the semantic properties of bir. The reason (no matter what it is) that was
affecting the scope taking property of bir did not affect the bare objects. Bir N objects are
DPs, which can have [±specific] feature, while bare objects are bare NPs and form
complex predicates with verbs. Therefore they do not interact with other constituents in
terms of scope.
We can go even further and propose that in other languages too, the difference in results
is not a consequence of the semantic properties of the indefinite articles/ numeral one, but
rather it is a consequence of a more general typological difference between languages,
namely, presence or absence of bare nominals as unambiguous narrow scope objects.
230
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
More specifically, we can argue that in those languages which allow bare objects in
argument positions, indefinite article/numeral + N object structure will be more likely to
be assigned wide scope reading. Such an analysis would predict that in bare object
languages, wide scope reading (which is dependent on the so-called exactly one reading
of indefinite articles) emerges early. Dutch and English do not have a bare object option
for narrow scope, so their indefinite article+N objects can have narrow scope reading.
Spanish and Chinese however have bare objects and their objects with indefinite article
or numeral are more likely to get wide scope assignment. Therefore, a typology based on
the possibility of bare objects in a language rather than the semantic properties of
indefinite articles seem to be easier to explain the results.
In this chapter, I presented two follow-up experiments to the negation experiment that I
reported in Chapter 6 and discussed the results. In the following chapter, I will evaluate
the findings and propose an analysis.
231
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
8 Evaluation of the results and the proposal
8.1 Summary of major results
In the earlier chapters, I presented and discussed the results of the experiments. I showed
that the adults assigned wide scope interpretation to the accusative marked objects in the
experiments. In other words, they knew the relationship between case morphology and
the position of the objects. The children in the youngest age group could not have an
adult-like interpretation of the objects. An improvement is observed, but even at 6;0, they
could not reach an adult-like level. In (220) I present the clause structure that I adopted in
this study once more. The indefinite objects can occur in all four positions numbered in
this structure.44 As discussed earlier, non-presuppositional objects occur in (1) and their
case gets deleted. The other positions can be occupied by the presuppositional objects
which bear overt case morphology.
(220) [CP ...CD [DPsu b je c t © ...[NegPtAgrOP CD [VP...] ] . . . ] ] . . . © ]
The positions (2), (3) and (4) are the positions where objects take wide scope with respect
to negation. In (1) the object is interpreted under the scope of negation. In child grammar
that I presented in the earlier chapters, children’s indefinite objects in the positions in (2),
(3) and (4), had an interpretation as i f they are in (1). In other words, they had narrow
441 am excluding the bare objects in this discussion because their position is not relevant in this discussion.
232
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
scope with respect to negation. A similar behavior is observed in the case of the
interaction of objects with adverbs.
This is a description of children’s behavior based on the experimental findings. In this
chapter I evaluate the findings, discuss the implications of the findings for the hypotheses
that I discussed in Chapter 4 and propose an analysis for children’s behavior. In the
discussion of the results in earlier chapters, I concluded that children have a difficulty
with the interpretation of objects. Before I discuss the implications of the present
findings, in the next section, I discuss a possible alternative conclusion.
8.2 Scope of negation?
Can the results be attributed to the acquisition of negation, rather than the indefinite
scope? In other words, could it be possible that in child grammar, negation is very high in
the hierarchical structure and thus no other constituent can scope over it? In early child
grammars, negation appears either at the beginning or the final end of the utterances,
suggesting that it may be in a c-commanding position. In the examples in (221), (222)
and (223), the negation appears at the beginning of the utterances.
(221) No the sun shining (Bellugi 1967)
(222) Pas la poupee dormir (Pierce 1992)
NEG the doll to. sleep
‘The doll is not sleeping.’
233
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(223) Nein diese Messer Auau (Park 1981)
NEG this knife hurt
‘This knife won’t hurt (her).’
In Turkish, too, similar examples are recorded. See for example the utterances in (224)
and (225) (from Aksu-Kog & Slobin, 1985 and Koskinen, 2000). In Turkish, the negation
appears at the end of children’s utterances, which can be related to head directionality.
What is interesting about these examples is that the form of the negation marker is not the
negative marker that an adult would use in these structures. Target adult structures are
presented between the brackets. Adult grammar would mark negation with the negative
suffix -m a attached to the stem of the verb. Instead of embedding the negation into the
verb complex, the child is expressing it at the end of the clause, i.e., outside of the verbal
complex (which includes tense and agreement).
(224) yap-acag-im, lh. [Adult form: yap-ma-yacag-im]
Do- fut-lS, NEG
‘I will not do it.’
(225) Ay, koy-du-m yok. [Adult form: koy-ma-di-m]
Oh put-past-IS NEG
‘Oh I didn’t put.’
234
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
These examples show that negation is in a high position in child grammar and thus, it is
marked with an external negation item that is placed above the clause. These examples
raise the question of whether or not children’s wide scope assignment to negation in the
present study can be attributed to the position of the negation in children’s grammar
rather than the interpretation of the objects. In other words, could we propose that
indefinites cannot take wide scope with respect to negation because no constituent can
scope over negation?
The strongest evidence against such a proposal is the findings based on the experiments
with the adverb twice. If children’s difficulty had to do with negation only, we would not
have the same results in the twice experiments. The similarity of the findings in both
experiments suggests that the reason is related to the interpretation of the objects, rather
than the scope o f negation. Moreover, the negation errors that are exemplified above are
recorded before the age 2;0, therefore they are examples from a much earlier
developmental period. The pattern that I report in the present study continues until a
much later age.
Therefore, the results and the patterns of the errors support my earlier conclusion that
children’s behavior is due to a non-adult-like interpretation of the objects. In the next
section, I discuss the implications of the Turkish data for hypotheses based on other
languages.
235
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
8.3 Implications for language acquisition hypotheses
The summary of the predictions of each hypothesis and whether or not Turkish findings
confirm the predictions are given in Table 25 below.
Table 25: Hypotheses and predictions for Turkish revisited
Hypothesis Predictions fo r Turkish
(Scope at the youngest age)
Confirmed?
Isomorphims Bir N-acc (preverbal) > NEG
NEG> bir N-acc (postV)
Bir N-acc (preverbal) > bir N-acc (postV)
Bir N, bare N > NEG > bir N-acc (postV)
No
C-command Bir N-acc > NEG > bir N=bare N No*
Lexical Factors Bir N(acc) > NEG (> bare N) No
D-related NEG > Bir N-acc=bir N=bare Yes*
(*) Modified versions of these accounts can capture the Turkish facts.
The isomorphism account of Musolino (1998) and subsequent work predicts that the
constituents that precede in linear order, have wider scope with respect to those that
follow. It is predicted that in Turkish, the objects that precede negation (both adjacent and
non-adjacent ones) will have wide scope with respect to negation, whereas the postverbal
objects will take narrow scope because they follow negation in the linear order.
Isomorphism account also predicts an asymmetry between post verbal and preverbal
accusative marked indefinites as one of them precedes and the other follows negation.
The experiment results showed that this is not what was home out in Turkish. Neither the
children nor the adults treated the objects according to their linear ordering. At their
youngest age, children treated all the object types alike. In the later ages, they started
236
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
differentiating case-marked versus non-case marked objects. No affect o f linear ordering
is observed in this differentiation.
Secondly, the c-command account of Lidz and Musolino (2002) predicts that the
accusative marked objects, which are VP-extemal and which occur in a c-commanding
position, will have an early wide scope with respect to negation in Turkish. In the case of
the post-verbal objects, especially, a wide scope interpretation effect is expected, if what
determines the scope assignment is the hierarchical ordering. The results of the
experiment showed that the c-command relationship of the objects did not have an impact
on scope assignment. Children in the youngest age groups treated all the objects alike
regardless of the hierarchical ordering. Even post-verbal objects, which are analyzed as
adjuncts to CP (a la Kural 1992, 1997) could take narrow scope for children. Therefore,
the present results cannot be explained by a c-command relationship of the objects with
negation45.
Thirdly, according to the proposal of Su (2001) and the subsequent work, the numeral bir
should have an impact on the interpretation of the scope assignment o f indefinite objects.
This prediction could explain the wide scope bias in objects that have the numeral bir in
the first negation experiment. However, by a modification of the scenarios, I showed that
the impact of bir can be reduced. Once the scenarios are fixed, the cardinal reading
4 5 But see the discussion in the next section for the similarities between the analysis adopted here and the c-
command account.
237
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
hypothesis does not hold for Turkish children's treatment of indefinite objects. It is
possible to conclude that it has some impact, since bir N objects are not treated like the
bare objects in terms of scope assignment, but this impact is not as big as it was proposed
by Su (2001).
Finally, the theories that are based on children's difficulty with discourse integration
predict that accusative marked indefinites, regardless of their linear position, will be a
late acquisition in Turkish. As discussed in the chapter on the adult grammar. Its
realization of the accusative case (i.e., inability to delete) and interpretation is heavily
based on its discourse-relatedness. The results of the experiments, i.e., children's
difficulty with the wide scope interpretation of the accusative marked objects, and their
narrow scope assignment to all object types are correctly predicted by the discourse
integration accounts proposed in Kramer (2000) and Schaeffer (1997, 2000). In the
earlier sections, I showed that the accusative case is a formal realization of the case
feature in Turkish. Its discourse-related interpretation comes about when it scrambles to a
field which is associated with a presuppositional interpretation. What is missing in child
grammar, then, is the ‘presuppositionality’ association with the case morpheme and this
is in line with the discourse-based accounts of indefinite acquisition. However, the
account that is adopted here is crucially different from the accounts defended by Kramer
(2000) and Schaeffer (2000).
According to Kramer (2000), children’s high objects which are analized as free variable
indefinites (a la van Geenhoven, 1998) have a predicate interpretation in child grammar.
238
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
This is because children’s grammar lacks a ‘discourse integration’ mechanism. The
predicate interpretation of both object types is due to their ‘semantic incorporation’ to
verbs. In this study, I do not attribute children’s treatment of objects to semantic
incorporation through which the thematic roles of objects are absorbed by the verb. In the
present analysis, it is crucial that the objects occupy an A-position and maintain an
argument status. With the model that I will present in the next section, objects do not
need to incorporate.
According to Schaeffer (2000), objects need to scramble to particular positions higher in
the clause structure to have particular interpretations. In my account, the objects do not
have particular destinations to go to. They gain their interpretations through escaping
from the focus or non-presuppositonality area. They have their interpretations because
they are not in a particular position (i.e., AgrOP). In contrast, in Schaeffer’s model,
objects move to RefP or DiscP to have a discourse related interpretation. More
importantly, in my account objects scramble for prosodic reasons, which is not taken into
consideration in Schaeffer’s account although it is likely that focus/prosody plays a
crucial role in the interpretation of objects in Dutch, too (Ruys, 2001).
8.4 Getting back in focus
As stated earlier, I attribute the findings to a discourse-related difficulty. In that sense, the
way I interpret the results is similar to the conclusion of the studies on Dutch. However,
in contrast to the accounts defended in Schaeffer (2000) and Kramer (2000), I propose
239
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
that when children cannot interpret an indefinite object, they assign a default
interpretation to it. The default interpretation for an indefinite object is its interpretation
in the highest A-position, which is AgrOP. In the clause structure that I presented above
(repeated here in (226)), the default position is (1).
(226) [C P ...(D [ D P subjeC t . . . © . - W U g r o r © [vr...] ]
The non-case marked objects are already in AgrOP, so they have an adult-like
interpretation. The accusative marked objects, no matter where their position with respect
to negation or adverbs is, have an interpretation as if they are in a lower position, where
they have narrow scope reading during the earliest periods of acquisition. Note that the
positions other than (1) are derived/scrambled positions and only the presuppositional
objects can scramble to those positions. More importantly, in those positions the objects
are defocused obligatorily. As we discussed earlier, accusative marked objects cannot
bear the focal stress. The contrast is clear especially when the accusative marked objects
appear adjacent to the verb but have a wide scope reading. In such cases, the focal stress
falls on a constituent other than the object and such a shift in the prosody provides
evidence that the object is in a scrambled position although it appears adjacent to the
verb.
I propose that children have a difficulty with the relationship between presuppositionality
and scrambling. I also propose that they cannot interpret the focus-related prosodic
240
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
variations, which is related to presuppositionality and scrambling. In child grammar, just
as it is the case in adult grammar, the accusative case is a formal realization of the case
feature. It is not associated with a particular interpretation in child grammar, as children
do not have access to the relationship between scrambling and case deletion. The mastery
of this relationship is crucial for an adult-like interpretation of presuppositionality. The
case morphology does not have an interpretational property for the child, neither does the
focal stress shifts. Therefore all objects are treated alike in terms of their interpretation
and the interpretation that they have is their interpretation in AgrOP, which is the neutral
focus position.
According to Cinque’s (1993) Nuclear Stress Rule (NSR), which is proposed as a
universal rule, the constituent in the mostly embedded position gets the primary stress.
This position is the neutral focus position across the languages (Cinque 1993, Zubizarreta
1998). This is also the position for non-presuppositional constituents by default.
Santos (2005, in progress), studies children’s interpretation o f the focus operator in
Portuguese and proposes that NSR is innate. Her proposal is based on the observation
that children have a tendency to focus the constituent that is in the most embedded
position by applying NSR. The constituent that is in this position is the direct object in a
regular transitive sentence. Any kind of defocalizations of the objects is difficult for
children as the resulting prosodic structure is a derived structure. In other words,
computationally, NSR is easier than any kind of focus shift for children and thus those
structures where the focused constituent is the most embedded one are easier for them.
241
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
According to Santos’s (2005) Default Focus Hypothesis (DFH), children start with a
default focus assignment which corresponds to the application of the universal NSR of
Cinque (1993). According to DFH, children assume that only the mostly embedded
constituents are focused.
In Portuguese, for example, the focus element ‘only’ can precede the subject and when it
does, it scopes over the subject only. Santos (2005, in progress) tests children’s (mean
age: 3;8) comprehension of yes-no questions with the focus element ‘only.’ She observes
that children understand the sentences as if the objects (or VPs) are focused, although for
adults only subject focus reading is possible. She argues that the prevalence of an object
focus reading is due to the universal NSR of Cinque (1993), which seems to be in
operation in child speech.
Portuguese child language is not the only example for an early NSR generalization. The
same kind of conservative focus assignment is observed across various typologically
different languages. For example, Gualmini, Maciukaite and Crain (2002) test English-
speaking children’s comprehension of sentences such as (227) and (228). In the former
one only is associated with the most embedded constituent and the focus is marked with
neutral stress. In the latter, the focal stress is shifted to another constituent.
(227) Barney only sold a cake to SNOW WHITE
(228) The farmer only sold a BANANA to Snow White
242
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The children were adult-like in their interpretation of the first sentence where the focal
stress is its neutral position. They interpreted the second sentence in (228) as if the focal
stress is on Snow White. In other words, children prefered to interpret only as if it is
associated with the most embedded constituent, regardless of the stress shift.
Endo (2004) reports a similar observation regarding the comprehension of only by
Japanese children (age range: 3;6-6;3, mean age 4;7). When the object is focused with
dake/sika ‘only,’ children’s comprehension was adult-like in 80% of the cases. When the
focal stress is shifted to the subject, in only 34% of the cases children behave like adults.
Szendroi (2003) reports similar findings for Dutch. Philip (1999) shows that the pattern
observed in other languages can be seen in Norwegian as well. Similarly Paterson et al.
(2003) show that English speaking children avoid the focus shifts achieved through the
occurrence of only in various positions in a sentence in a similar manner. All these
crosslinguistic data can be accounted for by the DFH of Santos (2005). Children across
different languages seem to apply the NSR of Cinque (2003) as Santos (2005) suggests.
In the present study, I attribute children’s difficulty with the scrambled objects to a
similar kind of developmental property. Remember that the most embedded position in
the Turkish clause structure is the position where the non-case marked objects occur. This
is where the focal stress falls according to NSR. In the case of the non-case marked
objects, the focus is in its neutral position. Thus it is not surprising that the children can
interpret the sentences correctly in the majority of the cases.
243
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The accusative marked objects, however, are defocalized. Although they are defocalized
in child grammar, they have an interpretation as if they are in the focus position. In other
words, children seem to keep the objects in their neutral focus positions in terms of their
interpretation. Note that DFH (Santos, 2005, in progress) does not say that the objects
must be in the focus position. It only says that the constituent that is in the mostly
embedded position must bear the focus interpretation. The results of the present study,
however, show that, at least in the case of Turkish, children have a tendency to keep the
objects in the most embedded position46. Because this most embedded position is lower
than the negation and adverbs, objects cannot scope over them.
In summary, because children have not yet mastered the interpretation of focus related
prosodic variations, they cannot interpret the objects in an adult-like manner. They seem
to put the objects back in the most embedded positions, due to a tendency to focus them.
Once they are in this position, just like the non-case marked objects, they cannot take
wide scope with respect to negation and adverbs. Thus, in the present account, I attribute
children’s behavior to a late development of focus or information structure which effects
their scope-taking property. Therefore in this account, children’s non-adult-like behavior
does not result from a deficient scope-taking property of objects. Rather, the scope
properties of the objects are adult-like but due to a conservative focus assignment
mechanism, objects cannot move out of the focus or non-presuppositionality field, where
they can only take narrow scope with respect to negation and adverbs.
4 6 Whether or not this observation holds crosslinguistically needs further investigation.
244
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In the discussion above I stated that the results are problematic for the c-command
account of Lidz & Musolino (2003) because the accusative marked indefinite objects
occur in scrambled positions where they c-command negation. The c-command model
predicts a wide scope reading of the accusative marked objects with respect to negation
due to their position. However, a modified version of the Lidz & Musolino (2003) model
can still capture the results reported here. Note that in the analysis I proposed above, the
objects are placed back in AgrOP, which I analyze as the default or neutral focus
position, due to the restrictions on focus assignment in child speech. Once they get in this
low position, they are c-commanded by negation and adverbs, and what determines scope
is the c-command relationship. Therefore, the analysis that I propose is compatible with
the c-command account if it is modified in such a way that it allows a focus related
reconstruction47. It further explains why objects cannot scope over negation in English as
well. In addition, such a focus related analysis can account for the results of the Dutch
data reported by Kramer (2000) in a much simpler way. If we attribute children’s
behavior to the difficulty with focus shifts and information structure, we do not need an
incorporation analysis, which is a less economic operation as it requires absorption of
theta-roles, changes in the argument structures of verbs etc. Another disadvantage of the
incorporation account is that it requires us to assume completely different adult and child
471 am using the word ‘reconstruction’ in its literal sense. I do not imply any quantificational operation
underlying this operation.
245
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
grammars, while the present account proposes two grammars that are different only in the
way they organize information structure.
8.5 Implications for Turkish clause structure
In addition to the implications for the language acquisition theories, the present findings
have implications for the current discussion of indefinites and clause structure of adult
grammar. First, the results of the adults confirm that accusative case is a factor, actually
the only factor that effects subjects’ scope assignment to objects in Turkish (c.f. En?
1991, and the subsequent work). Note that this does not mean that it is the accusative case
that assigns scope. Rather, it is treated as a cue, together with the prosodic structure of the
language for scope assignment.
Because the accusative marked indefinites can have narrow scope reading in child
speech, we can argue that the case morphology and the scope interpretation do not always
go together and they are not acquired simultaneously. Therefore the results of the present
study are problematic for a theory that predicts a simultaneous acquisition of case
morphology and its interpretation as they are assigned under the same functional
projection (c.f. Oztiirk 2004). The present study shows that the adult-like interpretation of
the case morphology is delayed and emerges much later than the production of case
morphology, thus, they are not necessarily interdependent for overt realization. In the
present study, I did not only propose that they are independent, I also suggested that the
interpretation and the formal assignment of case are taken care of at different components
246
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
of grammar. Such a separation could take care of the developmental pattern of the
accusative case.
The interpretation of the bare objects and the bir N objects is not predicted by an account
that treats the two object types alike. The finding that bir N objects can be assigned wide
scope in particular contexts, especially in the first negation experiment, poses a problem
for an account that treats bir N objects as a part of the verb complex (c.f. Oztiirk, 2004).
247
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
9 Accusative case omissions in a Turkish child’s speech
9.1 Introduction
In the earlier chapters I showed that children do not have an adult-like interpretation of
the accusative marked indefinite objects. In other words, the accusative marked
indefinites could have narrow scope with respect to negation and the adverb twice in
early child speech. I also argued that children do not have a syntactic deficiency. What
they are missing is that the connection between the interpretation of the objects and
scrambling. The goal of this chapter is to present an example of a child grammar which
lacks a similar kind of association between case marking, interpretation and scrambling.
Earlier work on scrambling and case marking (e.g., Ekmek9i, 1986) focused on the
question whether or not children’s objects in scrambled positions, and they showed that
they did even before the age 2;0. No study has concentrated in detail on the case omission
errors, which were considered to be rare and highly exceptional. In what follows, I
present a case-study conducted on a monolingual Turkish child’s accusative case
omission errors recorded before the age 2;0. Although they are few in number, they
display a pattern that is worth attention: In the samples analyzed, (1) specific objects are
not necessarily case marked, (2) case omission errors always occur in the preverbal
position48.1 argue, once more, that the absence of obligatory object case morphology in
4 8 Interestingly, in other child languages, too, the specific objects occur in “non-specific” syntactic
structures in less than 10% of the cases. In Swahili for example, the verb carries an object agreement
marker when the object is specific. Deen (2004) reports that children (age range: 1;8-3;1) omit the object
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the subject’s speech does not result from a syntactic deficiency. In contrast, the omission
pattern shows that the child’s grammar is governed by syntactic constraints that are
similar to the ones in operation in adult speech; namely the omission of the accusative
case occurs under adjacency to verbs. The omission of case regardless of the
interpretation of the object shows that where child and adult grammars differ is the
connection between scrambling, case and interpretation.
9.2 Method
Spontaneous speech samples of a monolingual Turkish child, Deniz, was recorded
longitudinally between the ages l;3-2;0 with approximately 20-day intervals49. All multi
word utterances that contained a verb and a direct object were targeted for analysis.
Object+Verb combinations were studied with a particular focus on the accusative case
omission errors and the position of the object with respect to the verb. The three positions
that were studied were 1- preverbal adjacent to verb (PreV-Adj), 2- preverbal non-
adjacent to verb (PreV-NonAdj) and 3-postverbal (PostV). Proper names, pronouns and
agreement in 7% of the obligatory contexts. So specific objects occur without object agreement. In
Russian,specific objects occur postverbally. According to Avrutin & Bran (2001) Russian children (age
range l;7-2;3) incorrectly produce the specific objects in preverbal position in 9.7% of the cases. The
Turkish child I report in this chapter omits the accusative case in 8% o f the clearly obligatory contexts.
There seems to be consistency across various languages in terms of specificity marking in child speech.
Whether or not children’s speech in other languages display a pattern such as the one I discuss here needs
further investigation.
4 9 The data analyzed in this chapter were collected for the project “The Longitudinal Study of the
Acquisition of Turkish,” which was supported by a grant from the Bogazici University Research Fund.
249
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
contextually specific common noun objects were included in the analysis and coded
according to their reference and context.
9.3 Results
The first accusative case is recorded at the age of 1;3, which is the first session recorded,
but the productive use of the accusative case and other nominal morphology takes place
only around the age of 1;6. (Ketrez, 1999, and subsequent work). The first object+verb
utterance is recorded at 1;650. During the period analyzed, 528 object+verb utterances
were recorded. 465 of these objects (88.1%) occurred in the PreV-Adj position, 13 of
them (2.5%) were in PreV-NonAdj positions and 50 of them (9.4%) occurred
postverbally. 283 of these preverbal objects (55%) were not case-marked. In other words,
majority of the objects occurred in the immediately preverbal position, which is the
canonical object position in Turkish5 1 . Scrambling to postverbal position is more common
than scrambling to a preverbal non-adjacent position. Table 26 presents a summary of the
distribution. As seen in Table 26, 22 (8%) of the case omissions were recorded in
contexts where case morphology was obligatory. In other words, they were clearly
ungrammatical.
5 0 First two-word utterances contained a subject or a vocative and a verb and are recorded before the age
1;6.
5 1 Slobin & Bever (1982) report that the SOV order is the most common word order in both adult and early
child speech and in the absence of any morphological evidence children treat the second nominal as the
object. Batman-Ratyosyan & Stromswold (2000) and Batman-Ratyosyan (2003) conduct similar
experiments and replicate the results of Slobin & Bever (1982).
250
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 26: Objects in three different positions.
PreV-Adj PreV-NonAdj PostV
0 *0 acc 0 *0 acc 0 *0 acc
260 22 183 0 0 13 2 0 48
Total 465 (88.1%) 13 (2.5%) 50 (9.4%)
The other case omissions were analyzed as grammatical omissions because the intended
meaning could be interpreted as a non-specific reference, as in the case o f the example in
(229). In this example, Deniz covers a doll she is holding with a blanket and the sentence
without a case marker on the object reads as 'I did doll covering.' It is grammatical with
this intended meaning. However, contextually it would be more appropriate to use an
accusative marked version of the object because there was only one definite doll she had
in her hand. This utterance is categorized as a grammatical case omission example but it
is also possible that the child was intending a specific reference, the particular doll that
she was playing with rather than a general doll-covering event and omitting the case.
Because such examples are hard to evaluate, they were not included in the ungrammatical
category. The discussion of the case omissions in the rest of this chapter focuses on the
clear cases where the omission is clearly ungrammatical.
(229) A grammatical case omission (common noun):
CHI: bebek o(r)t-tu-m. [age: 1;7,3]
doll-0acc cover-PAST-lS
‘I covered (a/the?) doll’
251
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
MOT: aa bebek 6rt-mii§ kiz-im.
oh do 11-0ACC cover-PERF daughter-poss&lS
‘oh my daughter covered (a/the?) doll’
A similar kind of ambiguity is seen in the use of proper names. In (230), Deniz wants her
mother to draw a picture of Biidii, a Sesame Street character. The sentence is grammatical
without a case marker with the intended meaning 'draw a (picture of) Budii' or 'any
picture of Budii.' Such examples, too, are analyzed as grammatical case omission
instances although they could be ungrammatical omissions if the child’s intention were to
refer to the particular person. Only 13% of the proper names were marked with the
accusative case in the preverbal position and thus had a specific interpretation. The
majority of the proper names was produced without a case marker, and had a non-specific
interpretation. These examples, too, as it is the case with the common nouns are hard to
evaluate in terms of the child's real intention. The possibility of examples such as (232),
where the case omission is clearly ungrammatical on proper names, suggest that she
might be omitting the case although her intention was a specific reading .
(230) A grammatical case omission (proper noun) [age: 1;9]
Budii yap.
Biidii make (=draw)
'Make/draw Budii'
252
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The examples for the utterances that were clearly ungrammatical are given in (231),
(232), (233) and (234). The example in (231) is especially interesting because she has a
grammatical production o f the case morpheme together with the omission of the case in
the same utterance. What is important to notice is that she produces the accusative case in
the post-verbal position (the first part of the utteance), and omits it when the object
occurs pre-verbally (the second part of the utterance).
(231) Itti-yoy-um bu-nnay-i, bu-nnay-0 vey [age: 1; 10,3]
want-prog-IS this-plu-acc this-plu-*0acc give.
‘I want these, give these (to me)’
In the example in (232), she omits case in another obligatory position: a proper noun.
Similarly in (233), a demonstrative pronoun lacks case in the preverbal position and in
(233) case is omitted in another obligatory context where the object obviously has a
definite reference.
(232) Lale-0 anlat. [age: 1; 10,3]
Lale-*0acc tell/narrate
‘Talk about Lale (literally: Tell Lale)’
(233) Ayi-cik bu-0 deyred-iyo [age: 1;11,23]
bear-dim this-*0acc watch-prog
‘The little bear is watching this’
253
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(234) Du kiymiti-0 vey-iy mi-tin? [age: 1;10,3]
That red-*0acc give-aorist que-2sg
‘Could you give (me) that red (one)?’
Thus the specific or definite objects were not necessarily case marked in the child's
speech. As discussed above, it is possible that those grammatical utterances that have a
non-referential interpretation may be case omission instances as well. When objects
(specific or non-specific) were not case marked, they occur in the preverbal position, i.e.,
no case omission is observed in scrambled positions (post-verbal or preverbal)5 2 . Table
27 summarizes the use and omission of the accusative case in three positions.).
Table 27: Grammatical production and omission of the accusative case
PreV-Adj PreV-NonAdj PostV
0 acc *0acc acc 0 acc *0acc acc 0 acc *0acc acc
CN 192 7 39 0 0 3 2 0 13
PN 17 3 2 0 0 2 0 0 11
PRO 0 12 78 0 0 5 0 0 23
WH 46 0 49 0 0 0 0 0 0
QNP 5 0 0 0 0 0 0 0 0
N-poss 0 0 15 0 0 3 0 0 1
Total 260 22 183 0 0 13 2 0 48
5 2 There are two exceptions, both are generic uses of bare objects. The following example is uttered in the
context of candy distribution, where ‘candy’ was already introduced into the discourse. See Goksel &
Kerslike (2005) for a discussion of such examples.
(i) CHI: al-di-m dekey.
take-past-IS candy
'I took candy'
254
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The first column shows the type of the objects. CM and PM stand for Common Noun and
Proper Noun respectively. They are followed by pronouns (PRO) and wh-words (WH All
the grammatical case omissions are recorded on the wh-word what. No ungrammatical
case omission is observed in other wh-words. 5 QNPs are all the occurences of the word
something, which can be used without an overt case marker. It is interesting to note that
she never omits the accusative case in objects that bear possessive morphology. As we
discussed in Chapter 3, case omission is not allowed in such objects in the language. I
attributed the inability to omit case in this kind of objects to the morpho-syntactic
requirements of the language. It is interesting to note that she is not violating this
constraint. In summary, what we see is that she is omitting case in those types of objects
which allow case omission and she grammatically produces the case morpheme in those
classes of objects which never allow case omission in adult speech.
9.4 Analysis
In adult speech, the objects that are not overtly case-marked can appear in the
immediately preverbal position. In other words, preverbal position is a legitimate position
for case omissions. However, in adult speech, only the non-specific/non-presuppositional
objects are in this position. As seen in the ungrammatical examples, in this child’s
speech, pronouns, proper names, and other specific objects can appear without case as
well.
255
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
I propose that children and adults have the same syntactic structures and their production
is restricted by the same syntactic principles. In adult speech, however, specific objects
obligatorily scramble and thus their case cannot get deleted. In child speech, the
requirements are more flexible: Specific objects can remain adjacent to verbs and lack
overt case. What is deficient, then, is the constraint that the specific objects have to
scramble and thus bear case-marking overtly. In many languages such as English,
specificity is not marked overtly and the specificity interpretation is heavily based on the
contextual cues. In this child's speech too, the interpretation of the object, scrambling and
overt case morphology do not necessarily go together.
Such a proposal that the child does not lack any syntactic knowledge correctly predicts
that (i) the omission errors are seen at a later developmental phase, when the child’s
speech already provides evidence for adult-like syntactic operations, (ii) she can produce
case morphology grammatically as well (iii) omission errors observed in other types of
nominal morphology that are not necessarily associated with specificity are rare when
compared to the accusative omission errors. In the following sections, I discuss these
predictions and show that the accusative case omission errors cannot be attributed to a
developmental phase of her syntactic development.
9.4.1 Grammatical production of nominal morphology
Around the time the omission errors occur, Deniz can produce other nominal and verbal
morphology grammatically. The evidence for morpho-syntactic knowledge comes from a
256
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
study that analyzes her nominal morphology development. Ketrez & Aksu-Kog (2004),
shows that Deniz’s morphological development reaches a stability period around the age
1;95 3 . Figure 21 shows the child's different lemma versus type scores. What is important
to notice in this figure is the distance between the two lines, which represent the morpho-
syntactic variation in Deniz’s speech. As seen in the figure, her morphological
development can be divided into three periods. Between the ages 1;3 and 1;6, no
morphological variation is observed (hence the lines overlap). Between 1;6 and 1;8 her
morphological variation increases (see the increasing distance between two lines) and the
increase ceases around 1;9, that is the difference between the number of types and
lemmas becomes stable after 1;9 (the lines go parallel to eachother)54. The errors
discussed in this chapter are observed after 1;8. As predicted by the proposal, they are
observed at a later developmental stage when the morpho-syntactic variety in her speech
already reaches a stable stage.
5 3 The figure is based on her nominal morphology only. Her verbal morphology follows a similar
developmental pattern (Aksu-Ko? & Ketrez, 2003).
5 4 The decrease in both lemmas and types in the session recorded at 1; 11 is due to the number of words
recorded in this session, which is shorter than the other sessions. Therefore it is not important for the
purpose o f the discussion. What is important is the distance between the two lines, which represents the
morphological variation.
257
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 21: Morphological variety in Deniz’s speech
— & —Lemma B Type
200 n
100
1;3 1;4 1;5 1;6 1;7 1;8 1;9 1;10 1;11 2;0
age
9.4.2 Asymmetry between accusative case and other nominal morphology
Further evidence for the proposal is that she can produce case morphology correctly, as
well, in preverbal and non-preverbal positions. Accusative case omission errors are
observed in 8% of the obligatory contexts, as discussed above. So she can produce the
accusative case morphology grammatically.
Table 28: Grammatical use and omission of nominal morphology
ACC5 5 DAT LOC ABL INS/
COM
GEN POSS PLU
Gram, use 00 112 50 83 1 3 47 37 10
*0 4 8 2 0 2 13 11 0
*0%
m
• r- .
3.4% 14% 2.3% 0 % 40% 13% 22% 0%
Gram, use
o
202 190 153 45 26 114 197 6
*0 21 8 0 0 0 3 2 0
*0% 9.4%* 4% 0% 0% 0% 2.5% %1 0%
(*) p <.05 (increase)
5 5 The accusative tokens in this table include one-word utterances or utterances that do not contain a verb so
that a more efficient comparison can be made with the other cases which do not necessarily occur in
utterances that contained a verb. Therefore the number o f tokens is not the same as the numbers discussed
258
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
As also predicted by the proposal, omission of other nominal morphology, which is not
necessarily associated with specificity is less common during the period analyzed
especially after 1;95 6 ’ 57,. As predicted, Dative, Locative, Instrumental and Commutative
case omissions are less frequent, Ablative and plural morpheme omissions are never
recorded.
Moreover, as seen in Table 28, the number of omissions in other nominal morphology
decreases in the period between l:9-2:0, while an increase is observed in the accusative
omissions (chi-square=3.98, p<.05). This asymmetry between the accusative case and the
other nominal morphology is clearly seen in Figure 22 below.
above or reported in Table-1, which included the utterances that contained both a verb and an object only.
5 6 Genitive case is associated with specificity as well (See Komfilt, 2003, among others) therefore a similar
pattern is expected, but not observed in the production of the genitive case. However, the discussion of
specificity and genitive case in the literature is based on the specific subjects that carry genitive case in
embedded clauses. Majority of the examples in Deniz’s data are nominals that appear in predicate positions
(e,g, bu teyp dede-nin (this cassette player grandpa-gen) ‘this cassette player is grandpa’s) and around the
time I analyze she doesn’t have any genitive marked embedded subjects. Whether or not all the uses of the
genitive case in all types of nominals and contexts are the same needs further investigation.
5 7 According to Ay gen (1999, and subsequent work), all case marked objects are specific. Examples such as
‘telefona bak (telephone-dat answer) ‘answer phone’ in examples such as ‘I am looking for someone to
answer phones,’ where the telephone is obviously non-referential and non-specific, is problematic for such
an analysis. In the present study, I treat the accusative case differently than other case markers, which could
appear due to lexical reasons (See Komfilt 2003, among many others for a similar approach).
259
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 22: Omission of the accusative case and other nominal morphology
40 - 1
- O - D A T
—o — LOC
— X — COM
—A— GEN
—O — PLU
—X— POSS
1 ;9-2;0
age
9.4.3 Evidence from the verbal domain
It has been observed that the syntactic development in nominal and verbal domains is
simultaneous (See Hyams 1996, for example). Children’s speech provide evidence that
when they are going through root infinites (or ‘aspect only’) stage throughout which their
verbs have aspectual/modal properties but do not necessarily refer to events that are
located in specific points in time (i.e., tense), their nominals lack specificity properties as
well58. In this study, I propose that Deniz already has specificity in her speech but does
not necessarily know the relationship between scrambling, case morphology and the
interpretation of the object. Such a proposal predicts that she has already gone through
the aspect-only stage in her grammar.
5 8 It is not clear whether there is any parallelism between the ‘aspect only’ stage and ‘the roor infinive’
stage.
260
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
This is what is reported in Aksu-Koq (1998), who analyzes the use o f tense/aspect
markers in Deniz’s speech. Deniz’s tense/aspect development is divided into two stages
for each tense/aspect morpheme. During the first stage, she uses each tense/aspect marker
with one class of verbs. For example, the past tense marker -DI is produced only on
achievement verbs and refers to a final state of a completed action, while the progressive
marker -Iyor exclusively goes with activity verbs. According to this study, by the age of
1 ;9, Deniz is already in a stage where she can use each tense/aspect marker with different
types of verbs and thus provides evidence that she can have tense reference.
In summary, I showed that the omission pattern Deniz follows suggests that she is
omitting case in positions that are possible case omission positions in the target language.
The properties of the case omissions and her morpho-syntax in general during the period
analyzed suggest that her non-adultlike behavior with respect to the accusative case
morphology is not due to her syntactic development but rather, they can be attributed to
her late acquisition of the relationship between scrambling and case morphology.
9.5 Discussion and Conclusion
In this chapter, I argued that the accusative case omission errors observed in a Turkish
child's speech do not result from a syntactic deficiency; in contrast, I proposed that the
pattern of the errors shows that the child's grammar is very similar to the adult grammar.
The evidence was based on the observation that the omission errors are observed in a
position where objects without case morphology can occur in adult speech. Based on the
261
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
grammatical use of case morphology in scrambled positions in early child speech,
Komfilt (1994) already argued for an early configurational grammar for Turkish. In the
present study, I suggested that the absence of case in this particular position, just as its
grammatical production in scrambled positions, provides evidence for an early
configurational grammar that has already been defended by Komfilt (1994). In this way,
this study presents a case where a child’s omission errors, just like the correct and
systematic use of morphemes, provide evidence for their existence in the syntactic
representation (c.f. Borer & Rohrbacher 2003).
The present findings are similar to the arguments in Giirel (2000) and Haznedar (2003)
who study adult and child second language acquisition of case morphology in Turkish.
They argue that the absence of case does not result from a syntactic deficiency in their
subjects’ speech. The findings reported here are similar to the findings reported in other
scrambling languages such as Japanese as well. Japanese children, too, drop case
(accusative and nominative) only in positions that are legitimate for case omission in
adult speech. The only difference between child and adult Japanese is that children omit
case more frequently than adults (Otsu 1994; Miyamoto et. al. 1999, among others).
I proposed that what is non-adult-like in child grammar is the relationship between case
marking, scrambling and interpretation. The strongest evidence for this was the omission
of the accusative case in the obligatory contexts, which shows that these objects are in
non-scrambled positions. Such a proposal is in line with the arguments for the acquisition
of the specific objects and direct object scrambling in Dutch. According to Schaeffer
262
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(1997, 2000), for example, specificity marking is optional in young children acquiring
Dutch. Children, around the age 2;0 years do not scramble specific/definite objects,
including pronouns, although scrambling of specific objects is obligatory in adult
grammar. Schaeffer attributes the children's behavior to a late acquisition of the 'concept
of non-shared knowledge' that requires the child to mark referentiality or specificity. In
the present study, too, I argued that the Turkish child does not know that specific or
presuppositional objects must scramble to a position where their case cannot get
deleted59.
Finally and more importantly, the findings in this chapter are in line with the findings
which are based on experiments that tested children's comprehension o f the accusative
marked versus non-case marked objects reported in this dissertation. In the other chapters
I argued that children do not differentiate case marked from non-case marked indefinite
objects. The omission pattern of the accusative case in Deniz’s speech complements the
experimental findings because they show that children do not know that specific objects
5 9 Gordishevsky & Schaeffer (2004) studies the relationship between case and number in early Russian and
propose that children use the nominative case in stead o f the accusative case on plural objects during a
period when their NumP is underspecified. Underspecified Number blocks the movement o f the plural
objects to the D head and prevents the case feature from getting to the DP level where it can be valued and
checked. The result is the nominative case realization in stead of the accusative case, which they argue to
be the default case form in the language. Turkish does not have a phonological realization of the
nominative case. If there is a default form of the nominals it is very likely that it is the bare form o f the
objects. In this sense, if the bare forms are the default forms in the language, the findings discussed in this
chapter might be similar to the findings reported in Gordishevsky & Schaeffer (2004), as well. Whether of
not Gordishevsky & Schaeffer’s (2004) default case analysis applies to Turkish needs further investigation.
In the present analysis, I am assuming that the non-case marked objects are assigned accusative case and
their case is deleted under adjacency.
263
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
should move, that is, they do not know that the overt occurrence of the accusative case
signals a scrambled position where presuppositional objects can occur.
I have to acknowledge that the findings discussed in this chapter are based on one single
child’s corpora and cannot be attributed to a general trend in the acquisition of Turkish
without further investigation on more children’s speech. It is still important to see that at
least in one child’s grammar, the overt realization (or the omission) of the case
morphology does not necessarily coincide with a particular kind of interpretation of the
objects and the case omission does not violate any syntactic constraint in the target
language.
264
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
10 Conclusion
10.1 Summary of findings and implications
This dissertation presented an analysis of indefinite objects in Turkish alongside the
results of experimental studies conducted on their comprehension by Turkish children. In
chapters 2 and 3, discussing the adult grammar, I proposed that accusative case is only a
formal overt realization of the case feature of objects. It cannot get deleted when the
object is in a scrambled position and thus it signals a scrambled position, a particular
interpretation and scope-taking properties. The results of the experiments confirmed that
in adult speech the overt accusative case was the only factor that signaled scope
assignment to objects. Children did not have an adult-like interpretation of the overtly
accusative marked objects during the developmental period I studied. Their case marked
objects could have narrow scope with respect to negation and adverbs regardless of their
linear position in child speech although they could take only wide scope in adult speech.
Children’s scope assignment improved with age, but did not reach an adult-like level
even at the age of 6;0
Both in adult speech and child speech bir ‘one/a’ played a role in affecting scope
assignment in the first set of experiments, biasing the subjects towards a wide scope
reading. In the last two sets of experiments, the wide scope reading was shown to be an
artifact of the experimental design, and disappeared once the impact of bir was
eliminated and the experimental scenarios modified. Other possible factors, such as word
265
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
order variations and surface adjacency to verbs did not affect children's or adults'
comprehension o f indefinite scope.
I concluded that the formal properties of accusative case assignment were accessible to
children, but they had difficulty linking scrambling with presuppositionality and focus-
structure with prosodic variations. These results thus provide evidence for a modularity
modal which allows the development of morpho-syntax to follow a different time line
from the development of other components, specifically, information structure. I further
showed that the present results could not be explained by current theories of scope
acquisition but they are compatible with a modified version of accounts that attribute
children’s non-adult-like behavior to a late acquisition of discourse pragmatics and the
account that is based on a c-command relationship of constituents. In contrast to the
accounts defended in Schaeffer (2000) and Kramer (2000), I proposed that when children
could not interpret an indefinite object, they assigned a default interpretation to it. In
other words, when the late development of the pragmatics component or the information
structure makes it difficult to interpret scrambled objects, the child assigns a default
interpretation to the scrambled indefinite object in its highest A-position, which is also
the neutral focus position. This analysis is compatible with the results of the studies done
on the acquisition of focus which universally report that children have a tendency to keep
the most embedded constituent in the neutral focus position. Accusative marked objects,
then, regardless of their surface position with respect to negation or adverbs are
interpreted as if they are in the highest A-position, be it AGRoP or internal to the VP,
which is lower than the lowest possible position of adverbs and negation. Because it is
266
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
lower, they are construed as having narrow scope with respect to adverbs and negation
during the earliest periods of acquisition. This account does not only explain why
accusative marked indefinites have a narrow scope with respect to negation and the
adverb twice in Turkish, it also accounts for the children's treatment of objects in other
languages.
Acquisition of an adult-like interpretation of accusative case requires the mastery of the
interaction between various components of grammar, namely morpho-syntax, semantics
and pragmatics. As is well known, a delayed development of one of the components may
have an impact on the production and the comprehension of the structures that require
multiple interactions. Although there is no delay in the syntactic structures, the delay in
the interpretation suggests that the interaction between different components cannot be
mastered by children. I proposed that in the case of accusative case in Turkish, the
delayed acquisition of the discourse pragmatics or information structure component
results in a non-adult-like comprehension of accusative marked indefinites. I showed that
the children have access to the morpho-syntax of their grammar and are capable of
producing accusative case in natural speech. Previous research and the spontaneous
production data I reported in Chapter 9, both support the proposal that children do have
an adult-like syntactic structure at the age I study in this dissertation. What is missing and
what gives rise to a non-adult-like behavior in indefinites, then, is the relatively slow
emergence of adult-like interpretation of scrambling and presuppositionality.
267
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In turn, the analysis crucially requires that the adult grammar allow the disassociation of
the formal properties of positions and their interpretation. In the case of the accusative
morpheme in Turkish, formal case assignment and presuppositionality interpretation must
be assigned at two different phases in the derivation. I argued that the formal properties
precede the discourse properties, i.e., children go through a period where they have
access to the formal properties of the accusative case, but lack the adult-like discourse
related ones as they require a series of more complicated interfaces. The child has to
know that stress shift to a constituent other than the object means that the object is
defocalized and it is in a scrambled position where the presuppositional objects can
occur. In such instances, the accusative case morpheme is not a formal case marker
anymore.
10.2 Future research
This study focused on the comprehension of the objects only, as the major concern of the
study was the accusative case morpheme and its interaction with the interpretation of
objects. It would be interesting to look at the comprehension of the indefinite subjects
especially because they seem to present different results crosslinguistically. Musolino
(1998) for example reports that children do not have an isomorphism effect on subjects
which can be assigned adult-like wide scope with respect to negation. In contrast Kramer
(2000) finds that the children participating in her experiments have difficulty with
subjects as well. Investigation of subjects would also be interesting because nominative
case does not have a phonological realization in Turkish and the definiteness/specificity
268
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
is marked exclusively through prosodic features and scrambling. Therefore it would be a
good testing ground for children’s interpretation of prosody and its association with scope
variation.
Another interesting follow-up topic would be children’s interpretation of focus particles.
There has been no study conducted on children’s comprehension of stress shifts or the
scope of focus particles in Turkish. In the earlier discussion in Chapter 8 ,1 assumed that
the studies conducted on other languages can be extended to Turkish and Turkish
children will assign default scope, which is their scope in AgrOP, to focus particles
regardless of their linear position. In other words, the results of the present study and my
analysis of these results predict that children younger than 4;0 years of age will assign a
default focus interpretation to objects, as the most deeply embedded constituents.
Another interesting research topic would involve subjects with specific language
impairment (SLI). Recently Schaeffer (2005) tested children with SLI with an application
of the tests used in Schaeffer (2000) with a prediction that SLI children and normal
children will behave the same. This prediction is due to her conclusion in her original
study that what children lack is a discourse rule, concept of non-shared knowledge, which
is a property of the pragmatic component. According to her analysis, children do not
know that they need to ‘mark specificity’ by scrambling specific objects to higher
positions. In the discussion in the preceeding chapters, I argued that her analysis does not
capture all the facts about indefinitenes reported in Kramer (2000). Although the details
of the mechanics of scrambling are different in her approach and in mine, we have a
269
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
common conclusion and that is children’s non-adult-like behavior is related to a
discourse-related deficifiency. In SLI children pragmatic component is not impaired.
Because their syntactic component is impaired and because normal children’s non-adult-
like behavior has nothing to do with a syntactic development, SLI children and normal
children are expected to be the same in terms of their interpretation o f pragmatic
conditions. This is exactly what was borne out in her study. Because I have the same
proposal that the children’s non-adult-like comprehension of the objects is not due to a
syntactic deficiency, I predict that Turkish children with SLI will not behave differently
than the subjects in the present study.
270
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
References
Abney, S. P. (1987). The English Noun Phrase in its Sentential Aspect. PhD
Dissertation, MIT.Cambridge,Mass.
Aksu-Ko?, A. & Slobin, D. I. (1985). The acquisition of Turkish, In Slobin, D. I. (ed)
Cross-linguistic Study Language Acquisition. Vol:l. New Jersey: Lawrence
Earlbaum Associates.
Aksu-Ko?, A. (1988). Input and tense-aspect in Turkish. First language 18:3. 255-280.
Aksu-Ko?, A. & Ketrez, F. N. (2003). Early verbal morphology in Turkish: Emergence
of inflections. In D. Bittner, W.U. Dressier & M. Kilani-Schoch (eds.) Mini
paradigms and the emergence o f verb morphology. Berlin: M. De Gruyter 27-52.
Avrutin, S. & Brun, D. (2001). The expression of specificity in a language without
determiners: Evidence from child Russian. In A. H-J. Do, L. Dominguez, & A.
Johansen (eds.) The BUCLD 25 Proceedings. Somerville, MA: Cascadilla Press.
70-81.
Avrutin, S. & K. Wexler (1992). Development of principle B in Russian: coindexation at
LF ans coreference. Language Acquisition 4, 259-306.
Aydemir, Y. (2004). Are Turkish preverbal bare nouns syntactic arguments? Linguistic
Inquiry, 35(3) 465-474.
Aygen, G. (1999). Specificity and Subject-Object positions/Scope interactions in
Turkish.Paper presented at the Conference on Turkic linguistics at Manchester
Univ. 8-10 April. 1999.
Baker, C. L. (1970). Double negation. Linguistic Inquiry 1. 169-196.
Baker, M. (1985) The Mirror Principle and morphosyntactic explanation, Linguistic
Inquiry 16. 373-415.
Baker, M. (1988). Incorporation: A theory o f grammatical function changing. Chicago,
Chicago University Press.
Batman-Ratrosyan, N. (2003). Acquisition o f Word Order and Morphology in Turkish.
Unpublished PhD Dissertation. Rutgers University. New York.
Batman-Ratyosyan, N. & Stromswold, K. (1999). What Turkish acquisition tells us about
underlying word order and scrambling. Proceedings o f the 23rd Annual University
o f Pennsylvania Linguistics Conference, 6, 37-52.
271
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Batman-Ratyosyan, N. & Stromswold, K. (2000). Early bare stems in an agglutinative
language. In Proceedings o f the 25th Boston University Conference on Language
Development. 102-113.
Batman-Ratyosyan, N. & Stromswold, K. (2001). Morphosyntax is Easy, Discourse/
Pragmatics is Hard. In Proceedings o f the 26th Boston University Conference on
Language Development. 793-804
Besler, D. (1999). Thequestion particle -m l and Movement in Turkish. Unpublished MA
Thesis. Bogazici University. Istanbul.
Borer, H. & Rohrbacher, B. (2003) Minding the absent. Language Acquisition. 10 (2).
Borer, H. & Wexler, K. (1987) The maturation of syntax. In T. Roeper, E. Williams,
(eds.). Parameter setting. 123-172. Dordrecht: D. Reidel Publishing co.
Brown, R. (1973). A first language: The early stages, Cambridge: Harvard University
Press.
Chomsky, N. (1975). The logical structure o f linguistic theory. New York: Plenum.
Chomsky, N. (1981). Lectures on government and binding. Dordrecht: Foris.
Chomsky, N. (1986). Knowledge of language: Its nature, origin and use. New York:
Praeger.
Cinque, G. (1993). A null theory of phrase and compound stress. Linguistic Theory. 24:2.
239-297.
Clark, H. H. (1977). Bridging. In Johnson-Laird, P.N. & P.C. Wason (Eds), Thinking:
Readings in cognitive science. Cambridge: Cambridge University Press. 411-420.
Clahsen, H. & Penke, (1992). ‘The acquisition of agreement morphology and its syntactic
consequences: New evidence on German child language from the Sunore
Corpus.’ In Meisel, J. (ed.) The Acquisition o f verb placement: Functional
categories and V2 phenomena in language development. Kluwer Academic
Publishers, Dordrecht, The Netherlands.
Crain, S. & Thornton, R. (1999) Investigations in Universal Grammar: A Guide to
Experiments on the Acquisition o f Syntax and Semantics. Cambridge, Mass. MIT
Press.
272
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Dede, M. (1986) Definiteness and Referentiality in Turkish nonverbal sentences. In D.
Slobin & K. Zimmer (Eds.) Studies in Turkish Linguistics. Amsterdam: John
Benjamins. 147-163.
Deen, K. U. (2004). Object agreement and specificity in Swahili. In Brugos, A.
Micciulla, L. & Smith, C.E. (eds.) BUCLD 28 Proceedings. Somerville, MA:
Cascadilla Press. 129-140.
Diesing, M. (1996) Indefinites. Linguistic Inquiry Monograph 20. Cambridge: the MIT
Press.
EkmekQi, O. (1979). Acquisition o f Turkish: A longitudinal study o f the early language
development o f a Turkish child. Unpublished Ph.D Dissertation. University of
Texas. Austin.
Ekmekci, O. (1982). Language development of a Turkish child: A speech analysis in
terms of length and complexity. Journal of Human Sciences 1(1). 103-112.
Ekmekgi, O. (1986). Significance of word order in the acquisition of Turkish. In D.
Slobin & K. Zimmer (Eds.) Studies in Turkish Linguistics. Amsterdam: John
Benjamins. 253-264.
En9, M. (1991). The Semantics of Specificity. Linguistic Inquiry 22, 1-25.
Endo, M. (2004). Developmental issues on the interpretation of focus particles by
Japanese children. In Brugos, A. Micciulla, L. & Smith, C.E. (eds.) BUCLD 28
Proceedings. Somerville, MA: Cascadilla Press. 141-152.
Erdal, M. (2000). Clitics in Turkish, In Goksel, A & Kerslake, C. (eds.) Studies on
Turkish and Turkic Languages. Turcologica 46. Wiesbaden: Harrasowitz Verlag.
41-48.
Erguvanli, E. E. (1984) The Function o f Word Order in Turkish Grammar. Berkeley:
University of California Press.
Farkas, D. and Swart.H. 2003. The semantics o f incorporation. CSLI Publications.
Fodor, J. A. (1983). Modularity o f mind. Cambridge, Mass.: MIT Press.
Fodor, J. & Sag, I. (1982). Referential and Quantificational Indefinites. Linguistics and
Philosophy. 5. 355-398.
Fukui, N. (1995). Theory o f projection in syntax. Studies in Japanese Linguistics, Vol. 4.
Center for the Study of Language and Information & Kurosio Publishers: Tokyo.
273
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Guilfoyle, E. & Noonan, M. (1988). ‘Functional categories and language acquisition.’
Paper presented at the 13th annual Boston University Conference on Language
Development.
Gordisevsky, G. & J. Schaeffer (2004). On the relation between case and number in early
child language. In Falk, Y. N. (ed.). Proceedings oflsrail Association o f
Theoretical Linguistics 20. 23-24 June, 2004.
van Geenhoven, V. (1998) Semantic Incorporation and Indefinite Descriptions:
Semanticand Syntactic aspects o f noun comprehension in West Greenlandic.
Stanford: CSLI Publications.
Givon, T. (1978). Definiteness & Referentiality. In Greeberg, J. (ed.). Universals o f
Human Language. MIT Press. Cambridge. 291-330.
Gomeshi, J. (2003). Plural marking, indefinites and the noun phrase. Studies in
Linguistics 57(2). 47-74.
Goksel, A. (1988). Bracketing paradoxes in Turkish nominal compounds. In Koc, S.
(Ed.) Studies on Turkish Linguistics. Ankara: Middle East Technical University.
Press. 287-298.
Goksel, A. (1993). What’s the mirror principle? In Cobb, M. and Y. Jiang (Eds). SO AS
Working Papers in Linguistics and Phonetics 3. 168-205.
Goksel, A. (1997). Linearity, focus and the postverbal position in Turkish. In Johanson,
L. & E.A. Csato, Locke, V. Menz, A. & Winterling, D. (eds.) Proceedings o f the
Seventh International Conference on Turkish Linguistics. Wiesbaden:
Harrossowitz. 86-106.
Goksel, A. & C. Kerslake (2005). Turkish: A comprehensive grammar. London:
Routledge.
Goksel, A. & A. S. Ozsoy (2000) Is there a focus position in Turkish? In Goksel,A. & C.
Kerslake (eds.) Studies on Turkish and Turkic Languages, Turcologica 46.
Harrassowitz Verlag-Wisbaden. 219-228.
Goksel, A. & A. S. Ozsoy (2000) dA: a focus/topic associated clitic in Turkish. Lingua
113:205-232.
Goksel, A. & Stavroula, T. (1995). Linear Domains: Variations between Greek and
Turkish. Studies in Greek Linguistics. Proceedings o f the 16th Annual Meeting o f
the Department o f Linguistics Faculty o f Philosophy. Aristotle University of
Thessaloniki. 4-6 May 1995. 193-203.
274
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Gualmini, A. (2005). The ups and downs o f child language: Experimental studies on
children’ s knowledge o f entailment relationship and polarity phenomena. New
York & London: Routledge.
Gualmini, A. & Musolino, J. (2003). The scope of partitivity in child language. Paper
presented at the CUNY conference, MIT, Cambridge. March 27-29, 2003.
Gualmini, A., Maciukaite, S. & Crain, S. (2002). Children's insensitivity to contrastive
stress in sentences with ONLY. PWPL 9.1. Proceedings of the 26th Annual PLC.
University of Pennsylvania.
Gurel, A. (2000). Missing case inflection: Implications for second language acquisition.
In C. Howell, S. A. Fish & T. Keith-Lucas (eds.). Proceedings o f the 24th Annual
Boston University Conference on Language Development, Somerville, MA:
Cascadilla Press, 379-390.
Haegeman, L. (1996). Verb second, the split CP and early null subjects in Dutch finite
clauses. GenGenP 4, Geneva: Department of General Linguistics, University of
Geneva. 133-175
Harley, H. (1995). Subjects, events and licensing. PhD Dissertation. MIT.
Haznedar, B. (2003). Missing surface inflection in adult and child L2 acquisition. In
Juana M. Liceras et al. Proceedings o f the 6th Generative Approaches to Second
Language Acquisition Conference (GASLA 2002). Somerville, MA: Cascadilla
Proceedings Project, 140-149.
Homstein, N. (1984). Logic and Grammar. Cambridge, MA: MIT Press.
de Hoop, H. (1992) Case Configuration and Noun phrase interpretation. NewYork:
Garland.
Hyams, N. (1983). The acquisition o f parameterized grammars. Unpublished PhD
Dissertation. City University of New York.
Hyams, N. (1986). Language acquisition and the theory o f parameters. Dordrecht:
Kluwer.
Hyams, N. (1992). The genesis of functional categories. In Meisel, J. (ed). The
acquisition o f verb placement: Functional categories and V2 phenomena in
language development. Dordrecht: Kluwer. 371-400.
Hyams, N. (1996) Underspecification of functional categories in early grammar. In.
Clahesen, H. (ed.) Generative perspectives on language acquisition. Amsterdam:
John Benjamins. 91-128.
275
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Issever, S. (2003). Information structure in Turkish: the word order prosody interface,
Lingua 113, 1025-1053.
Karmiloff-Smith, A. (1979). A functional approach to language: A study o f determiners
and reference. Cambridge: Cambridge University Press
Kelepir, M. (1997). The implications o f Antisymmetry in Theory o f Syntax in Turkish
word order. Unpublished MA Thesis. Bogazici University. Istanbul.
Kelepir, M. (2001). Topics in Turkish Syntax: Clausal Structure and Scope.
Ph.D.Dissertation. MIT.
Kelepir, M. (2004). Intervention effects in the interpretation of Turkish and Japanese
indefinites. In Csirmaz, A. Youngjoo Lee & M. A. Walter (Eds.). Proceedings of
the Workshop on Altaic in Formal Linguistics 1. Cambridge, MA: MIT Working
Papers in Linguistics 46. 49-63.
Kennelly, S. D. (1994). Caseless Existential DPs in Turkish. Paris 8 Working Papers in
Linguistics.
Ketrez, F. N. (1999). Early Verbs and the Acquisition o f Turkish Argument Structure.
MA Thesis, Bogazisi University, Istanbul.
Ketrez, F. N. (2004a). Children’s accusative case and indefinite objects. Dilbilim
Arastirmalari 2004. 63-74.
Ketrez, F. N. (2004b). -lAr-marked nominals and three types of plurality in Turkish.
Proceedings o f CLS 39.
Ketrez, F. N. & Aksu-Koc, A. (2004). Early nominal morphology in Turkish: Emergence
of case and number. Ms. Under review.
Knecht, L. E. (1986) Subject and object in Turkish. MIT: Ph.D. Dissertation.
Koopman, H. (1998). ‘Internal and external distribution of pronominal DPs’ In Johnson,
K. & Roberts, I. (eds.). Papers in memory of Osvaldo Jaeggli. Reidel, Dordrecht,
The Netherlands.
Korkmaz, Z. (2003). Turkiye Turkcesi Grameri (Sekil Bilgisi). Ataturk Kultur Dil ve
Tarih Yuksek Kurumu. No: 827. Ankara: Turk Dil Krumu Yayinlari
276
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Komfilt, J. (1994). Some remarks on the interaction of case and word order in Turkish:
Implications for Acquisition. In Lust, B., Suner, M. & Whitman, J. (eds.)
Syntactic Theory and First language Acquisition: Cross-linguistic Perspectives.
Vol.l: Heads, Projections and Learnability. New Jersey: Lawrence Earlbaum
Associates. 171-199.
Komfilt, J. (1997). Turkish. London: Routledge.
Komfilt, J. (2001). Non-specific partitives and the unreliability of specificity markings.
Paper presented at LSA Annual Meeting, Washington, DC.
Komfilt, J. (2003). Scrambling, Subscrambling and Case in Turkish. In Karimi, S. (ed.)
Word order and scrambling. Oxford: Blackwell Publishing. 125-155.
Komfilt, J. (2004). DOM and two types of DSM in Turkish. Paper presented at the
workshop on DSM. Nijmegen. July 2004.
Koskinen, P. (2000). Children’s acquisition of negation: Turkish evidence. In Goksel, A.
& Kerslake, C. (eds.) Studies on Turkish and Turkic Languages. Turcologica 46.
Wiasbaden: Harrassowitz Verlag.299-306.
Kramer, I. (2000). Interpreting Indefinites: An experimental Study o f Children’ s
language comprehension. MPI Series in Psycholinguistics.
Kratzer, A. (1998). Scope or pseudoscope? Are there wide-scope indefinites? In S.
Rothsteined., Events and Grammar, 163-196. Dordrecht: Kluwer.
Kural, M. (1992). Properties of Scrambling in Turkish. Ms., UCLA.
Kural, M. (1993). V-to-(I-to)-C in Turkish. In Beghelli, F. & Kural, M. (eds.) Recent
papers in syntax, semantics and computational linguistics. UCLA Occasional
Papers in Linguistics, Volume 11.
Kural, M. (1997). Postverbal constituents in Turkish and linear correspondence.
Linguistic Inquiry 28: 498-519.
Kuroda, S. Y. (1988). Whether we agree or not. A comparative syntax of English and
Japanese. Linguistic Investigations 12: 1-47.
Kiintay, A. (1995). Developing referential cohesion in elicited Turkish narratives.
Unpublished M. A. Thesis. University of California. Berkeley.
Kiintay, A. (2002). Development of the expression of indefinites: Presenting new
referents in Turkish Picture-series stories, Discourse Processes 33(1). 77-101.
277
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Ladusaw, W. (1979). Negative polarity items as inherent scope relations. PhD
Dissertation. University of Texas at Austin.
Ladusaw, W. (1980). Negative polarity items as inherent scope relations. New York:
Garland.
Lebeaux, D. (1988). Language acquisition and the form of grammar. PhD Dissertation.
University of Massachusets, Amherst.
Linebarger, M. (1980). The grammar of negative polarity. PhD Dissertation. MIT.
Cambridge, MA.
Lidz, J. (2000a). The morpho-semantics of object case in Kannada. In Bird, S. Camie, A.
Haugen, J. D. andNorquest, P. (eds.). Proceedings ofWCCFL 18, 325-336.
Lidz, J. (2000b). The grammar of Accusative case in Kannada, Ms. Northwestern
University.
Lidz, J. & Musolino, J. (2002). Children’s command of quantification. Cognition 84
113-154.
Lidz, J., McMahon,E. Syrett, K., Viau, J., Anggoro, F. Peterson, J., Sneed, E., Bunger, A.
Flevaris, T., Graham, A. Grohne, K., Lee, Y. & Strid, E. (2003). Quantifier
Raising in 4-year-olds. Proceedings o f the 28th Boston University Conference on
Language Development.
Marantz, A. (1991). Case and Licencing. Proceedings of ESCOL 8 . University of
Maryland, Baltimore. 234-253.
Massam, D. (2001). Pseudo noun incorporation in Niuean. Natural Language and
Linguistic Theory, 19: 153-97.
Masulla, P. (1992). Incorporation and Case Theory in Spanish: A crosslinguistic
perspective. University of Washington.
Matthewson, L. (1999). On the interpretation of wide scope indefinites. Natural
Language Semantics 79-134.
Matthewson, L. & J. Schaeffer (2000). Grammar and Pragmatics in the acquisition of
article systems. In Gilkerson, J., Becker, M. & Hyams, N. (eds). UCLA Working
Papers in Linguistics. No. 5. Language Development and Breakdown 1. 1-39.
May, R. (1977). The Grammar o f Quantification. Ph.D. Dissertation. MIT.
278
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
May, R. (1985). Logical Form: Its structure and derivation. Cambridge, Mass. MIT
Press.
Meisel, J. & Mueller, N. (1992). ‘Finiteness and verb placement in early child grammars:
Evidence from simultaneous acquisition of two first languages: French and
German.’ In Meisel, J. (ed.) The Acquisition of verb placement: Functional
categories and V2 phenomena in language development. Kluwer Academic
Publishers, Dordrecht, The Netherlands.
Miller, K. & Shmitt, C. (2003).The interpretation of Indefinites and Bare singulars in
Spanish Child Language. The 6 th Conference on the Acquisition of Spanish and
Portuguese and First and Second Languages. University of New Mexico, October
16, 2003.
Milsark, G. (1974). Existential Sentences in English. Ph.D. thesis, MIT, Cambridge, MA.
Miyagawa, S. (1991). Case realization and scrambling. Ms.
Miyamoto, E.T., K. Wexler, T. Aikawa & S. Miyagava (1999). Case dropping and
Unaccusatives in Japanese Acquisition. Proceedings o f the 23rd Annual Boston
University Conference on Language Development. Cascadilla Press 443-452.
Mithun, M. (1984). The evalution of noun incorporation. Language 60. 847-893.
Miner, K. L. (1986). Noun stripping and loose incorporation in Zuni. International
Journal o f American Linguistics 52(3). 242-254.
Miner, K. L. (1989) A note on noun stripping. International Journal o f American
Linguistics 55. 476-77.
Musolino, J. (1998) Universal Grammar and the Acquisition o f Semantic Knowledge: An
experimental Investigation into the Acquisition o f Quantifier Negation Interaction
in English. Unpublished PhD Dissertation. University of Maryland, College Park.
Musolino, J., Crain, S. & Thornton, R. (2000) Navigating negative quantificational
Space. Linguistics 38(1). 1-32.
Musolino, J. & Lidz, J. (2003). The scope of isomorphism: Turning adults into children.
Language Acquisition, 11(4). 277-291.
Nakipoglu-Demiralp, M. (2004). The Turkish accusative and the relation between focus
structure and object displacement. Ms. Bogazici University. Istanbul.
Neidle, C. (1988). The role o f case in Russian Syntax. Kluwer Academic Publishers.
279
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Nilsson, B. (1985). Case marking semantics in Turkish. Ph.D Dissertation. University of
Stockholm.
Nilsson, B. (1986). Object incorporation in Turkish. Aksu-Kog A. & E. Erguvanli-Taylan
(eds). Proceedings of Turkish Linguistics Conference. August 9-10,1984.
Istanbul. Bogazici University Publications. 113-128.
O ’Lary & Crain, S. (1994). Negative polarity items (a positive result) positive polarity
items (a negative result). Paper presented at the 1994 Boston University
Conference on Language Development.
Orgun, O. C. & Inkelas, S. (2004). Turkish direct objects: Incorporation of word
order? Proceedings of the 40th Chicago Linguistics Society Meeting. University of
Chicago.
Otsu, Y. (1994). Case-marking Particles and Phrase structure in Early Japanese
Acquisition. In Lust, B., Suner, M. & Whitman, J. (eds.) Syntactic Theory and
First language Acquisition: Cross-linguistic Perspectives. New Jersey: Lawrence
Earlbaum Associates. 159-169.
Ozcan, F. H. (1993). Coherence in Narratives o f Turkish-speaking children: The role o f
noun phrases. Unpublished PhD Dissertation. University of Reading. U.K.
Ozcan, F. H. (1997). Definite and indefinite nouns in the discourse of Turkish-speaking
children. In Johanson, L., E.A. Csato, Locke, V. Menz, A. & Winterling, D. (eds.)
Proceedings o f the Seventh International Conference on Turkish
Linguistics.Wiesbaden: Harrossowitz. 553-567.
Oztiirk, B. (2004). Case, referentiality and phrase structure. Unpublished Ph.D
Dissertation. Harvard University. Cambridge, MA.
Park, T-Z. (1981). The development of syntax in the child with special reference to
German. Innsbrucker Beitrage zur Kulturwissenschaft Sonderheft 45. Institut fur
Spachwissenschaft der Universitat Innsbruck.
Paterson et al. (2003). Children’s comprehension of sentences with focus particles.
Cognition, 89(3). 263-294.
Philip, W. (1999). Children want only a right-conservative determiner. University o f
Massachusetts Occasional Papers in Linguistics 24. Amherst. University of
Massachusets. 221-243.
Platzack, C. (1990). ‘A grammar without functional categories: A syntactic study of early
Swidish child language’ Nordic Journal o f Linguistics 13. 107-126.
280
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Platzack, C. (1992). Functional categories in early Swedish. In Meisel, J. (ed.) The
Acquisition o f verb placement: Functional categories and V2 phenomena in
language development. Kluwer Academic Publishers, Dordrecht, The
Netherlands.
Poeppel, D. & Wexler, K. (1993). ‘The full competence hypothesis of clause structure in
early German.’ Language 69. 1-33.
Prince, E. F. (1981). Towards a taxonomy of given-new information. In P. Cole (Ed.),
Radical pragmatics. New York: Academic Press, 223-255.
Progovag, L. (1994). Negative and Positive polarity: A Binding approach. Cambridge:
Cambridge University Press.
Radford, A. (1988). Small children’s small clauses. Transactions o f the Philological
Society 86. 1-46.
Radford, A. (1990). Syntactic theory and and the acquisition o f Syntax: The nature o f
early child grammar o f English. Basil Blackwell: Oxford, England.
Radford, A. (1994). The syntax of questions in child English Journal o f child language
21,211-236.
Reinhart, T. (1997). Quantifier Scope: How Labor is divided between QR and choice
functions. Linguistics & Philosophy 20. 335-397.
Reinhart, T. (1999). The processing cost of reference set computation: guess pattern in
acquisition. Lecture notes given in Rutgers cognitive colloquium.
Ritter, E. (1991). Two functional categories in noun phrases. Evidence from Modem
Hebrew. In Rothstein, S. (ed.). Syntax and Semantics 25: Perspectives on Phrase
Structure. New York: Academic Press. 37-62.
Rizzi, L. (2000). Remarks on early null subjects. In M.-A. Friedemann and L. Rizzi
(eds.). The acquisition of syntax. Harlow, England: Longaman.
Roeper, T. & J. Weissenbom (1990). How to make parameters work: Comments on
Valian. In Frazier, L. and J. de Villiers (eds.) Language Processing and Language
Acquisition. Dordrecht: Kluwer. 147-162.
Runner, J. T. (1994). A specific role for AGR. In Benedicto & Runner, J. T. (eds.)
UMASS Occasional Papers 17: Functional Projections.
281
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Ruys, E.G. (2001). Dutch scrambling and the strong-weak distinction. Journal o f
Comparative Germanic Linguistics 4\ 39-67.
Santos, A. L. (2005). Getting in focus: Acquisition evidence for the nuclear stress rule.
Ms. Universidade de Lisboa and UCLA.
Santos, A. L.(in progress) The acquisition o f minimal answers. Doctoral Dissertation.
Universidade de Lisboa
Schaeffer, J. (1997). Direct object scrambling in Dutch and Italian child language.
Unpublished PhD. Thesis. UCLA.
Schaeffer, J. (2000). The acquisition o f direct object scrambling and clitic placement
syntax and pragmatics. Language Acquisition and Language Disorders. Volume
22. Amsterdam/Philadelphia: John Benjamins Publishing Company.
van Schaaik, G. (1996) Studies in Turkish grammar. Turcologia. Wiesbaden:
Harrossowitz.
Schroeder, C.(1999). Turkish nominal phrase in spoken discourse. Wiesbaden:
Harrassowitz.
Slobin, D. I. (1973). Cognitive prerequisites for the development of grammar. In C.A.
Ferguson and D.I. Slobin (eds.) Studies of child language development: New
York: Holt, Rinehart and Winston.
Slobin, D. I. (1985) Cross-linguistic Study Language Acquisition. New Jersey: Lawrence
Earlbaum Associates.
Slobin, D. I. & Bever, T. (1982). Children use canonical sentence schemas: A cross-
linguistic study of word order and inflections. Cognition, 12(3), 229-265.
Su, Yi-ching (2001). Scope and Specificity: A cross-linguistic Study on English and
Chinese. Unpublished PhD. Dissertation. University of Maryland, College Park.
Su, Yi-ching (2003). Children don’t always follow c-command as a scope principle.
Paper presented in GLOW 2003 Workshop on Generative Approaches to
Language Development. Lund, Sweden.
Swift, L. B. (1963). A Reference Grammar o f Modern Turkish. Indiana University
Uralic and Altaic Series 19. Bloomington.
Szabolcsi, A. (1983). Focussing properties, or the trap of first order. Theoretical
Linguistics 10: 125-145.
282
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Szendroi, K. (2003). Acquisition evidence for an interface theory of Focus. In J. van
Kampen & S. Bauw (eds.) Proceedings o f GALA 2003. Vol.2. Utrecht: LOT.
Tateishi, K. 1983. Subjects, SPEC and DP in Japanese, Proceedings ofNELS 25.
405-418.
Thornton, R. (1995). Children’s negative questions: a production/comprehension
asymmetry. In J. Fuller, H. Han, and D. Parkinson (Eds) Proceedings o f ESCOL,
Ithaca, NY: Cornell University.
Trueswell, J. C., I. Sekerina, N. M. Hilland, M/ L. Logrip (1999). The kindergarten-path
effect: Studying on-line sentence processing in young children. Cognition 73. 89-
134.
Tura, S. (1973). A Study o f the articles in English and their counterparts in Turkish.
PhD. Dissertation. The University of Michigan. Ann Arbor.
Tura, S. (1986). Definiteness and referentiality in Turkish nonverbal sentences. In Slobin,
D.I. & K.Zimmer (eds.) Typological Studies in Language: Studies in Turkish
Linguistics 8 . Benjamins, Amsterdam.
Tunc (Karabay), F. (1997). On some problems of Turkish nominal compounds.
Unpublished BA Thesis, Bogazici University.
Valian, V. (1990). Null subjects: A problem for parameter setting models of language
acquisition. Cognition, 35, 105-122.
Valian, V. (1991). Syntactic subjects in the early speech of American and Italian
children. Cognition, 40, 21-81.
Valian, V. (1994). Children's postulation of null subjects: Parameter setting and language
acquisition. In B. Lust, G. Hermon, & J. Komfilt (Eds.), Syntactic theory and first
language acquisition: Cross-linguistic perspectives. Vol. 2: Binding,
dependencies, and learnability (pp. 273-286). Hillsdale, NJ: Erlbaum.
Valian V. & Eisenberg, Z. (1996). The development of syntactic subjects in Portuguese
speaking children. Journal o f Child Language, 23, 103-128.
Verrips, M. & Weissenbom, J. (1992). ‘Routes to verb placement in early German and
French: The independence of fmiteness and agreement.’ In Meisel, J. (ed.) The
Acquisition o f verb placement: Functional categories and V2 phenomena in
language development. Kluwer Academic Publishers, Dordrecht, The
Netherlands. 283-331.
283
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Viau, J., Lidz & M., J. (2005) Pragmatics & the role of experience in overcoming
isomorphism. Paper presented at the Annual LSA Meeting at Oakland, CA.
Winter, Y. (1997) Choice Functions and the scopal semantics of indefinites. Linguistics
and Philosophy 20, 399-467.
Zidani-Eroglu , L. (1997). Indefinite Noun Phrases in Turkish, Unpublished Ph.D.
Dissertation. University of Wisconsin Madison.
Zimmer, K. (1986). On the function of post-predicate subject pronouns in Turkish. In
Aksu-Ko?, A. and Erguvanli Taylan, E. (eds.). Proceedings of the Turkish
Linguistics Conference. 195-206.
Zubizarreta, M.L. (1998). Prosody, Focus and Word Order. Cambridge, MA: The MIT
Press.
284
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Appendix 1: Test items in Experiments 1 and 2
Experiment-1
The sentences are from version-1. The other versions have the same sentences, but the
order of sentence/ object type varies. The sentence types are not mixed. The fillers are
inserted after every two-three test sentences. No more than three test sentences and no
more than two fillers followed one another.
Test Sentences
bir N-acc twice Verb
1. Cocuk bir sandalyeyi iki kere itsin “get the child push a particular chair twice”
2. Cocuk bir cocugu iki kere opsun. “get the child kiss a particular child twice”
3. Cocuk bir tabagi iki kere dondursun“get the child turn over a particular plate twice”
4. Cocuk bir siseyi iki kere yuvarlasin“get the child roll a particular bottle twice”
5. Cocuk bir cocugu iki kere ziplatsin“get the child make a particular child jump twice’'
6. Cocuk bir keciyi iki kere itsin. “get the child push a particular goat twice”
twice bir N-acc Verb
1. Cocuk iki kere bir sandalyeyi ceksin“get the child pull a particular chair twice”
2. Cocuk iki kere bir tabagi yikasin “get the child wash a particular plate twice”
3. Cocuk iki kere bir tabagi ellesin “get the child touch a particular plate twice”
4. Cocuk iki kere bir cocugu gidiklasin“get the child tickle a particular child twice”
5. Cocuk iki kere bir yemegi karistirsin“get the child mix a particular pot twice”
6 . Cocuk iki kere bir kovayi kaldirsin“get the child raise a bucket twice”
twice N Verb
1. Cocuk iki kere cocuk dondursun “get the child make a child turn around twice”
2. Cocuk iki kere at kostursun “get the child make a horse run twice”
3. Cocuk iki kere sandalye cevirsin “get the child turn over a chair twice”
4. Cocuk iki kere kapi acsin “get the child open a door twice”
5. Cocuk iki kere cicek koklasin “get he child smell a flower twice”
6 . Cocuk iki kere kavanoz yuvarlasin “get the child roll a jar twice”
twice bir N Verb
1. Cocuk iki kere bir sise atsin “get the child throw a bottle twice”
2. Cocuk iki kere bir cocuk dusursun “get the child make a child fall twice”
3. Cocuk iki kere bir sandalye tekmelesin “get the child kick a chair twice”
4. Cocuk iki kere bir tabak isirsin “get the child bite a plate twice”
285
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5. Cocuk iki kere bir kova bosaltsin “get the child empty a bucket twice”
6 . Cocuk iki kere bir sandalye sallasin “get the child shake a chair twice”
Filler sentences
1. Cocuk masanin ustune ciksin
2. Cocuk bahceye ciksin
3. Cocuk yere otursun
4. Cocuk ata binsin
5. Cocuk ciceklerin ustunden atlasin
6 . Cocuk masanin etrafmda uc kere donsun
times”
7. Cocuk yere yatsin
8 . Cocuk arkadaslarina el sallasin
9. Cocuk beyaz kecinin ustunden atlasin
10. Cocuk dolabin yaninda dursun
11. Cocuk siyah ati sevsin.
12. Cocuk siyah ata binsin.
13. Cocuk ocagin arkasina saklansin
List of objects
Kitchen: A fridge with two doors and a drawer, a stove, two pots, one big spoon, a dinner
table, four chairs, a basket full of bottles, three jars, a coffee cup, two buckets full of
flowers, three children, a dishwasher, sink with a fountain, four plates.
Backyard: Three horses, three goats, a bunch of flowers on the ground and in a basket.
Experiment-2
Training sentences
1. Sentence:Cocuk masanin altina girsin “go under the table”
Scene: climbs on top of the table [wrong action]
2. Sentence: Cocuk sandalyeye otursun “sit on the chair”
Scene: sits on the floor [wrong action]
3. Sentence: Cocuk iki kere ziplasin “jump twice”
Scene: the child jumps twice [correct action]
4. Sentence:Cocuk bir ata binsin“ride a horse”
Scene: the child rides a horse [correct action]
5. Sentence Cocuk uc tabagi bir sepete koysun. “put three plates into the basket”
Scene: the child puts a plate into the basket [wrong action]
Test sentences
All statements imply one object. Two objects are involved in the action performed in
front of the child. There are three agent children named Seda, Ahmet and Osman. The
286
“get the child climb up the table
“get the child go out to the backyard”
“get the child sit down”
“get the child ride on a horse”
“get the child jump over the flowers”
“get the child turn around the table three
“get the child lie down”
“get the child waive-hand to her friends”
“get the child jump over the goat”
“get the child stand next to the fridge”
“get the child caress the black horse”
“get the child ride the black horse”
“get the child hide behind the stove”
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
instructions are given by another agent, Ayse. Objects and the setting are the same as in
Experiment-1.
Bir N-acc twice Verb
1. Seda bir sandalyeyi iki kere itsin “Seda, push a particular chair twice”
2. Seda bir tabagi iki kere dondursun “Seda, touch a particular plate twice”
3. Ahmet bir siseyi iki kere yuvarlasin.“Ahmet, roll a particular bottle twice”
4. Ahmet bir sandalyeyi iki kere tekmelesin. “Ahmet, kick a particular chair twice”
5. Osman bir tabagi iki kere sallasin. “Osman shake a particular plate twice”
6 . Seda bir keciyi iki kere ziplatsin “Seda make a particular goat jump twice”
twice bir N-acc Verb
1. Seda iki kere bir sandalyeyi ceksin. “Seda pull a particular chair twice”
2. Osman iki kere bir tabagi yikasin. “Osman, wash a particular plate twice”
3. Osman iki kere bir tabagi ellesin. “Osman, touch a particular plate twice”
4. Ahmet iki kere bir cicegi koklasin. “Ahmet, smell a particular flower twice”
5. Ahmet iki kere bir ati kostursun. “Ahmet, make a particular horse run twice”
6 . Osman iki kere bir tabagi isirsin. “Osman, bite a particular plate twice”
Filler sentences
1. Cocuk ocagin arkasina saklansin
2. Seda bir tabagi sepete koysun.
3. Cocuk masayi sallasin
4. Cocuk masanin ustune ciksin
5. Cocuk iki sandalyeyi dondursun
6 . Cocuk yere otursun
7. Cocuk ata binsin
8 . Cocuk ciceklerin ustunden atlasin
9. Cocuk arkadaslarina el sallasin
10. Cocuk yere yatsin
'‘hide behind the stove” [correct action]
'‘put a plate into the basket” [wrong action]
“shake the table” [correct action]
“climb up the table” [wrong action]
“turn over two chairs” [correct action]
“sit down” [wrong action]
“ride on a horse” [correct action]
“jump over the flowers” [wrong action]
“waive your hand to your friends” [correct]
“lie down on the floor” [correct action]
11. Cocuk beyaz kecinin ustunden atlasin “jump over the white goat” [wrong action]
12. Cocuk masanin etrafinda uc kere donsun “turn around the table three times” [wrong
action]
287
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Appendix 2: Test items in Experiment 3
Training Sentences
W arm -up-01.Bu kizin adi Asli ’ ymis. Bunlar da Seda ve arkadaslariymis. Burda
gelmisler A sli’ dan izin istiyorlar bahcede oynamah icin. Asli da onlar “tabii gelin burada
bahcede oynayabilirsiniz ” diyor. / The name of this girl is Asli. And these are Seda and
her friends. They came her to ask Asli’s permission to play in her backyard. Asli says “of
course you can play in my backyard”
PUPPET: Bu resimde Asli babasiyla konusuyor
“In this picture, Asli is talking to his father”
Expected response: NO (she is talking to the kids)
Warm-up-02.Ahcz7erm oglenyemegi vaktiymis. Cok acikmislar. Burda otlari
yiyorlarmis. Yemekte biraz ot bir kac tane de cicek varmis. / Look it is the lunch time for
the goats. They are very hungry. They are eating grass. The have some grass and some
flowers for lunch.
PUPPET: Keciler sut iciyor
“The goats are drinking milk”
Expected response: NO (they are eating grass)
W arm -up-03.5a^ bu Seda. Burada bahcede oynuyor, Burada yaninda bir tane sandalye
var ama o sandalyeye oturmak istemiyor. Cunku daha yorulmamis ve diger cocuklarla
oynuyormus. / This is Seda. She is playing the backyard. There is a chair right next to her.
She does not want to sit yet because she is not tired and she would like to play with the
other kids.
PUPPET: Seda sandalyede oturmuyor
“Seda is not sitting on the chair”
Expected response: YES
W arm -up-04. Burada masanin ustunde uc tane bardak varmis, Asli onlari mutfaga
goturmek istiyormus. Cunku burada dururlarsa dusup kirilabilirlermis. Ama sadece iki
tane eli oldugu icin iki bardak tasiyabiliyormus sadece. O yuzden bardaklardan biri
simdilik masanin ustunde kalmis. / Asli has three cups on the table. She would like to take
them to the kitchen because they may fall down and break if she leaves them there. But
because she has only two hands, she can carry only two cups at a time. So, this one is left
on the table for the time being.
PUPPET: Asli bir bardagi masada birakiyor
“Asli is leaving a cup on the table”
Expected response: YES
288
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
W arm-up-05. Ayse ata binmek istiyormus. Atina girmis, ben ata binmek istiyorum demis.
Ama ato o sirada yemekyiyormus ve oyuzden atina binememis. En sonda gidip biraz
sonra tekrar gelmeke karar vermis. / Ayse would like to ride her horse. She goes to her
horse, but the horse says, he is having his lunch now, so she cannot ride him now. So she
decides to go away and come back later.
PUPPET: Ayse ata biniyor
“Ayse is riding a horse”
Expected response: NO (she is not riding her horse, she goes away)
Test Sentences
Version-1: Accusative indefinites: Bir N-acc Verb / Bir N-acc XP Verb
Test-01. Bu kizin adi Asli’ymis. Marketten bir suru gazoz almis, boyle sepetin icine
doldurmus eve getirmis. Simdi hepsini dolaba koymasi gerekiyormus. Burada bir de
tencere varmis siselerin yaninda, onu da dolaba koyacakmis. Acmis dolabi, tencereyi
yerlestirmis. Sonra sira siselere gelmis. Ama bir bakmis, dolapta siseleri koyacak yer
yokmus. O zaman ben de onlari sonra dolaba koyarim demis, siseleri sepetin icinde
birakmis gitmis. / The name o f this girl is Asli. She buys lots o f soft drinks from the
supermarket and puts them in basket. Now she needs to put these into the fridge. She
also has a pot next to them. She would like to put that into the fridge too. She opens the
fridge, places the pot in there but then she sees that there is no room fo r the bottles. So
she decides to put the bottles later, she leaves them in the basket and goes away.
PUPPET: Asli dolaba bir siseyi koymadi / Asli bir siseyi dolaba koymadi
“Asli did not put a particular bottle in the fridge”
Expected response: NO (Asli did not put any bottles)
Test-02. Burada da Asli bulasik yikiyormus. Kirli tabaklari biz guzel bulasik makinasina
dizmis. Hepsini yikamak istiyormus. Ama bir tanesini masanin uzerinde unutmus. Onu
sonradan gormus. Bu tabagi da elimde yikayayim demis, lavabonun icinde yikamis. / Asli
is washing the dishes here. She puts all the plates into the dishwasher. She wants to wash
all o f them. But she forgets one o f them on the table. She decides to wash that one in the
sink.
PU PPET: Asli makinada bir tabagiyikamadi /A sli bir tabagi makinada yikamadi
“Asli did not wash a particular plate in the dishwasher”
Expected response: YES
Test-03. Burada cocuklar saklanbac oynuyorlarmis. Ahmet’i ebe yapmislar. Obur
cocuklar da kosmus gitmis saklanmis. Ahmet’in butun cocuklari bulmasi gerekiyormus.
Ama aramis aramis bir turlu bulamamis. En sonunda vazgecmis, ben bunlari bulamicam,
demis. Yurumus gitmis. / Children are playing hide-and-seek here. Ahmet becomes the
289
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
seeker. The other children hide. Ahmet starts looking fo r them, but no matter how hard he
tries, he cannot fin d the kids. In the end he decides to give up and go away.
PUPPET: Ahmet bir cocugu bulamadi /B ir cocugu Ahmet bulamadi
“Ahmet could not find a particular child”
Expected response: NO (Ahmet could not find any kids)
Test-04. Asli cok susamis. Burada masanin ustunde uc sise gazoz bulmus. Hepsini icmek
istiyormus. Almis bir siseyi, icmis bitirmis. Sonra obur siseyi almis, onu da icmis. Tam
en sonuncuyu icecekken arkadasi gelmis. O da son siseyi arkadasina vermis. / Asli is
very thirsty. She finds three bottles of soft drink on the table. She would like to drink all
of them. She takes and drinks one, and then she drinks the other one. When she is about
the drink the last one, her friend arrives. She gives the last bottle of drink to her friend.
PUPPET: Asli bir gazozu icmedi / Bir gazozu Asli icmedi
“Asli did not drink a soda pop“
Expected response: YES
Test-05. Cicekyemeyi cok seven bir keci varmis. Burada icinde uc tane cicek olan bir
sepet bulmus. Bu ciceklerden biriniyemis bitirmis. Obur cicegi deyemis. En sonucuyu da
yemek istiyormus ama uzanmis uzanmis, ona bir turlu boyu yetismemis. En sonunda
vazgecip yuruyup gitmis. O gittikten sonra bir at gelmis, o cicegi de o at yemis. / This
goat likes eating flowers. He finds a basket with three flowers. He eats one of the
floewer, the he eats the other one. He would like to eat the last one too, but he cannot
reach that one because he is not tall enough. He finally gives up and goes away. After he
leaves a big horse comes and eats that flower.
PU PPET: Keci bir cicegi yemedi / Bir cicegi keci yemedi
“The goat did not eat a particular flower”
Expected response: YES
Test-06. Ayse kecilerini beslemek istiyormus. Bir sepetin icine bir suru ot doldurmus
getirmis. Ama keciler, biz cicek yemek istiyoruz, ot sevmiyoruz demisler. Bahcede bir
yerde cicek bulmuslar, onlari yemeye baslamislar. Ayse de elindeki otlari atina vermis. /
Ayse would like to feed her goats. She brings a basket full of grass and puts in front of
them. But the goats would like to eat flowers, so they do not like the grass. They go away
and find some flowers there and start eating them, and Ayse gives the grass in the basket
to her horse.
PUPPET: Ayse bir keciyi beslemedi / Bir keciyi Ayse beslemedi
“Ayse did not feed a particular goat”
Expected response: NO (Ayse did not feed any goats)
Version-2: Non-case marked indefinites: Bir N Verb / N Verb
290
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Test-01. Bu kizin adi Asli’ymis. Marketten bir suru gazoz almis, boyle sepetin icine
doldurmus eve getirmis. Simdi hepsini dolaba koymasi gerekiyormus. Acmis dolabi, bir
kac tane sise koymus, sonra birkac sise daha koymus. Elinde bir tane sise kalmis. Ama
bir bakmis, dolapta sise koyacak yer yokmus. O zaman ben de bunu sonra dolaba
koyarim demis, siseyi sepetin icinde birakmis gitmis. / The name o f this girl is Asli. She
buys lots o f soft drinks from the supermarket and puts them in a basket. Now she needs to
put these into the fridge. She opens the fridge, places some bottles, then she places some
more. Then she sees that there is no room fo r the last bottle. So she decides to put that
one later, she puts it back in the basket and goes away.
PUPPET: Asli dolaba bir sise koymadi / Asli dolaba sise koymadi
“Asli did not put a bottle into the fridge”
Expected response: NO (She doesn’t put one of the bottles only, she put the others)
Test-02. Burada da Asli bulasik yikiyormus. Bir suru kirli tabak bir de tencere varmis.
Elepsini yikamak istiyormus. Once tencereyi bulasik makinasinin icine koymus. Sonra bir
bakmis, tabaklari koyacak yer kalmamis. O da tabaklari sonra yikamaya karar vermis. /
Asli is washing the dishes here. She has lots o f dirty dishes and a pot. She puts the pot
into the dishwasher first. The she sees that she does not have any room fo r the dishes. So
she decides to wash the dishes later.
PUPPET: Asli makinada bir tabak yikamadi /A sli makinada tabak yikamadi
“Asli did not wash a plate in the dishwasher”
Expected response: YES
Test-03. Burada cocuklar saklanbac oynuyorlarmis. Ahmet’i ebe yapmislar. Obur
cocuklar da kosmus gitmis saklanmis. Ahmet’in butun cocuklari bulmasi gerekiyormus.
Agacin arkasina saklanan cocugu hemen bulmus. Sepetlerin arkasina saklanan cocugu da
bulmus. Ama obur cocugu aramis aramis bir turlu bulamamis. En sonunda vazgecmis,
ben bunlari bulamicam, demis. Yurumus gitmis. / Children are playing hide-and-seek
here. Ahmet becomes the seeker. The other children hide. Ahmet starts looking fo r them,
he finds the child who was hiding behind the tree. He finds the girl who was behind the
baskets, too. But no matter how hard he tries, he cannot find the other kid. In the end he
decides to give up and go away.
PU PPET: Ahmet bir cocuk bulamadi / Ahmet cocuk bulamadi
“Ahmet could not find a child”
Expected response: NO (He couldn’t find one of the kids, he found the others)
Test-04. Asli cok susamis. Burada masanin ustunde uc sise gazoz ve bir caydanlik dolusu
cay bulmus. Acaba ne icsem diye dusunmus. En sonunda cay icmeye karar vermis. Bir
bardaga cay doldurmus ice ice gitmis. Gazozlar da masanin ustunde kalmis. / Asli is very
thirsty. She finds three bottles o f soft drink and a kettle full o f tea on the table. She cannot
decide what to drink. In the end she decides to pick up a cup o f tea. She fills the cup and
leaves. The bottles remain on the table.
291
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
PUPPET: Asli bir gazoz icmedi / Asli gazoz icmedi
“Asli did not drink a soft drink”
Expected response: YES
Test-05. Cicek yemeyi pek sevmeyen bir keci varmis. Burada icinde cicek olan bir sepet
bulmus.Cicek yemeyi sevmedigi icin, sadece sepetteki otlari yemis, cicekleri sepette
birakmis, yurumus gitmis. / This goat does not likes eating flowers. He finds a basket
with flowers. Because he does not like flowers, he eats up the grass and leaves the
flowers in the basket. Then, he goes away.
PUPPET: Keci bir cicek yemedi / Keci cicek yemedi
“The goat did not eat a flower”
Expected response: YES
Test-06. Ayse kecilerini beslemek istiyormus. Bir sepetin icine bir suru ot doldurmus
getirmis. Keciler cok sevimisler. Ama bu keci, ben cicek yemek istiyorum, ot
sevmiyorum demis. Bahcede bir yerde cicek bulmus, onlari yemeye baslamis. Obur
keciler de sepetteki otlari yemisler. / Ayse would like to feed her goats. She brings a
basket fu ll o f grass and puts in front o f them. But this goat says he likes eating flowers so
he goes away and fin d some flowers there and starts eating them, and Ayse gives the
grass in the basket to the other goats.
PUPPET: Ayse bir keci beslemedi / Ayse keci beslemedi
“Ayse did not feed a goat”
Expected response: NO (she did not feed one the goats, she fed the other two)
Filler Sentences
Filler-01. Burda bahcede cocuklar oynuyorlarmis. Ayse de onlari seyrediyormus. Acaba
ben de cocuklarin oyununa katilsam mi diye dusunuyormus. Sonra orada dolasan atini
gormus. Ben en iyisi atimla oyanayim demis. / The kids are playing in the garden and
Ayse is watching them. She is thinking whether she should join the kids’ game. Then she
sees her horse that is wandering around and decides to ride her horse.
PUPPET: Ayse cocuklarla oynamadi
“Ayse did not play with the kids”
Expected / correct response: YES
Filler-02. Burada cocuklar oyun oynuyorlarmis. Sonra bu ikisi cok yorulmuslar. Biz
biraz yere oturup dinlenelim demisler. Seda da onlarla oynuyormus ama pek yorgun
degilmis. Onun icin o boyle ayakta kalmis. / These children were playing here. The these
two gets very tired and decide to sit down and take a rest fo r a while. Seda was playing
with them but she did not feel tired. So she stays standing.
292
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
PUPPET: Seda yere oturmadi
“Seda did not sit down”
Expected / correct response: YES
Filler-03. Burada cocuklar gene saklanbac oynuyorlarmis. Osman da nereye
saklanacagina bir turlu karar veremiyormus. Acaba ciceklerin arkasina mi saklansam,
yoksa masanin altina mi girsem diyormus. En sonunda, ciceklerin arkasina saklanmaya
karar vermis. / The kids are playing hide-and-seek. Osman cannot decide where to hide.
He thinks, “ Should I hide behind the flowers? Or perhaps I should go under the table... ”
In the end, he decides to go behind the flowers again.
PU PPET: Osman ciceklerin arkasina saklanmadi
“Osman did not hide behind the flowers”
Expected / correct response: NO
Filler-04. Ayse’nin cani sikilmis. Asli’ya hadi gel alisverise gidelim demis. Ama Asli
cok yorgunmus o yuzden disari cikmak istemiyormus. En sonunda evde oturup sohbet
etmeye karar vermisler. / Ayse is very bored. She says “le t’ s go fo r shopping!” to Asli.
But Asli is very tired and does not want to go out. In the end they decide to sit at home
and chat.
PU PPET: Asli ve Ayse alisverise gitmediler
“Asli and Ayse did not go shopping”
Expected / correct response: YES
Filler-05. Aysenin iki tane ati varmis. Almis atlarini Asliya gelmis. Hadi gel ata binelim
demis. Asli ata binmekten korkiyormus. Sen bin istersen ama ben binmek istemiyorum.
Ben yurumek istiyorum demis. Ayse atina binmis, Asli da diger atin yaninda yurumus,
gitmisler. / Ayse has two horses. She comes to Asli with her horses and says “let’ s ride
horses!” but Asli is afraid o f horses and does not want to ride a horse. In the end, Ayse
rides on a horse and Aslo walks along side the horse. They all together walk away.
PUPPET: Asli ata binmedi
“Asli did not ride a horse”
Expected / correct response: YES
293
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Appendix 3: ‘No object’ condition in Experiment 3
In Experiment 3 that I reported in Chapter 6 children were provided with scenarios where
one out of three objects was not acted upon. In the hide-and-seek game, for example, the
seeker can find two out of three children and fails to find one child. I call this condition
‘one-object condition.’ It is a typical wide scope context for objects. If children had a
wide scope reading, they accept the accusative marked test sentence in one object
condition, which, in Turkish, reads as 'one of the objects was not acted upon.' If they had
a narrow scope reading they comprehend the sentences as ‘no object was acted upon’ and
thus rejected it. During the actual experiments, in addition to the one object condition, I
presented the test sentences with a contex where no objects were acted upon to see
whether the children who reject the sentences in one object condition, accept them in 'no
object condition.' In the 'no object' version of the hide-and-seek game, for example, the
seeker cannot find any children no matter how hard he tries and in the end gives up. Thus
in Experiment 3 children made judgments about six sentences in total (3 with no object
condition and 3 with one object condition), but in Chapter 6 I reported the experiments
with one object condition (wide scope condition) only. In this Appendix, I present those
results that are excluded in the discussion in Chapter 6 and explain the reason why they
were excluded.
I did not include the results of the no object condition in the discussion of the results in
Chapter 6 , because the no object scenarios are ambiguous due to the truth conditions of
the sentences. One object scenarios present 'wide scope' contexts, but no object scenarios
are not true narrow scope counterparts and the accept results are do not necessarily show
that children have narrow scope reading of the objects.
In the no object scenarios, where no objects are acted upon, we present children a
particular set of objects and even though none of these objects are acted upon, each of
these objects is a specific object as they are members of a definite set in this particular
experimental setting. In other words, when no objects are acted upon, it is also the case
that one of the objects is not acted upon. Interestingly, however, although the no object
contexts are ambiguous, the subjects presented a differentiated response pattern in two
conditions. Adults accepted the accusative-marked objects in one object condition and
rejected them in no object conditions. In the Tables below, I present the scores of the no
object condition, together with the results of the one object condition. The first column
shows the responses in one object condition while the second column is for the scores of
the no object condition. The scores in the tables are the accept scores. The expected
reading for the accusative indefinites is accept in the former condition, namely indefinite
having wider scope over negation, and reject in the latter condition (though accept score
is not ungrammatical due to the reason mentioned above).
294
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table A: Results of experiment 3 (adults).
Object type One object No objects
bir N-acc XP
Vneg
3.0 (100, SE=.00) .21 (7, SE=.154)
XP bir N-acc
Vneg
2.71 (90.3,
SE=.163)
.64 (21, SE=.289)
bir N Vneg 1.42(47,
SE=.358)
1.78 (59, SE=.394)
N Vneg .0 0 0 2.92 (97,
SE=7143E.02)
Table B: Results of experiment 3 (chi
Object type One object No object
bir N-acc XP
Vneg
2.44 (81,
SE=.37)
1.66 (55,
SE=.40)*
XP bir N-acc
Vneg
2.55 (85,
SE=.33)
1.88 (63,
SE=.35)*
bir N Vneg 2.12(70,
SE=.398)
2.25 (75,
SE=2.25)
N Vneg 1.66 (55,
SE=.35)
2.41 (80,
SE=.28)
dren, mean age: 4;2)
(*) p<.05 (difference from adults)
Table C: Results of experiment 3 (children, mean age: 5;2'
Object type One object No objects
bir N-acc XP
Vneg
2.66 (89,
SE=.21)
1.55 (52,
SE=.33)*
XP bir N-acc
Vneg
2.42 (81,
SE=.61)
1.71 (57,
SE=.420)
bir N Vneg 1.66 (55,
SE=.408)
1.88 (63,
SE=.42)
N Vneg 1.33 (44,
SE=.49)
2.66 (87,
SE=.23)
(*) p<.05 (difference from adults)
A 3 (age groups) X 2 (case) X 2 (bir) X 2 (adjacency) ANOVA was run on no object
condition scores. The results showed that accusative case (F( 1, 113)=4.191, p=.043), age
(F(2,113)=.014,p=.070) and bir (F(l,113)=6.521,/?=.012) have a main effect, and
adjacency does not (F (l,l 13)=.997,p=.320), that is, we have the same/ similar result in
both conditions (the no object condition reported here and the one object condition
discussed in Chapter 6 . The only difference between the two conditions is observed on
the age factor. In other words, in the one object condition that I reported in Chapter 6 ,
295
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
children were adult like. In the one object condition, however, age groups are different.
Below are the details of the test results and the implications.
A paired samples t-test conducted on adults responses in two scope conditions showed
that when the object is accusative marked, no matter what the adjacency condition is,
adults have preference for wide scope reading of the object (f(13)=18.004,p<.01(13)=-
5.836,p<.0\ non-adjacent, and adjacent objects respectively). When the object does not
have a case marker adults did not have a scope preference (f(13)=.673, n.s.) for bir N
objects. When the object is in the bare form, however, adults never accept a wide scope
reading of the object (t(13)=41.00, p<.01) and has a clear preference of narrow scope of
the object. Therefore, although the no object condition is ambiguous in a way described
above, the adult subjects have a differentiated behavior in two conditions and treat the no
object context as a true narrow scope context.
Paired samples t-test run on the two scope conditions of four-year-old subjects reveal
that, the non-adjacent objects are not treated differently in two different scope conditions
(f(8)=1.673, n.s.), while with adjacent objects children have a preference for wide scope
(f(8)=2.828, p< 05). With non-case marked and bare N objects they accept a wide scope
reading in 70% and 55% o f the cases respectively. They have narrow scope reading with
these object types in 55% and 80% of the instances (non-case marked and bare objects).
The paired-samples t-test analysis reveal that with neither non-case marked objects, nor
the bare ones, children have a preference for one of the scope readings (f(7)=.357, n.s,
11)=1.750, n.s.). Five year old children have a wide-scope preference for the non-
adjacent object (1(8)=4.264,/?<.01). 5-year-old children’s non-adjacent accusative
indefinite objects have wide scope with respect to negation. No significant scope
preference is observed for the adjacent object (f(6)=1.369, n.s.). A paired samples t-test
conducted on bir N scores does not show a significant difference (1(8)=.610, n.s.) in two
condition. In bare nouns, however, they have a narrow scope interpretation in 87% of the
cases in no object context. Their scores reveal a preference for a narrow-scope reading
(t(5)=2.697, p<.05).
Despite the ambiguity o f the no object condition, the two conditions present an
asymmetry in responses. Adults accept the accusative marked indefinite objects in one
object condition and reject them in no object condition. In the one object condition
children and adults are observed to be the same (no effect of age is recorded), whereas in
the no object condition, the age factor plays a role in the subjects' behavior.
296
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Appendix 4: Test items in Experiment 4 and 5
Test items include 3 warm-up scenarios, 6 test scenarios and 5 fillers/control items. Test
sentences have 5 different versions that differ in word order and object morphology.
Each child hears only one version of the test sentences.
Warm-up / Training Items
Warm-up-01: Seda ve Ahmet saklanbac oynayacaklarmis. Seda “ama iki kisi saklanbac
oynamak hie zevkli degil, keske bir kisi daha olsa,” demis. Tam o sirada oradan Osman
geciyormus, onlar da Osman'I oyuna davet etmisler. Osman “ama benim en sevdigim
cizgi film baslayacak simdi, ben onu seyretmek istiyorum,” demis. Seda “ama o filmin
baslamasina daha cok var, hadi gel biraz oynayalim, sonra gidip seyredersin” demis.
Osman da “tamam o zaman, hadi biraz oynayalim demis. Uc cocuk oyuna baslamislar.
Ahmet'i ebe yapmislar, Seda ve Osman da saklanacak yer aramaya baslamis. / Seda and
Ahmet are playing hide and seek. Seda says 'it is no fun to play this game with two
people. I wish there were one more person'. Then they see Osman passing by and invite
him to join them, but Osman says 'I am going home to watch my favorite cartoon, I
cannot play with you now'. Seda says 'you still have some time before it starts. Let's play
together for a while, and then you can go home and watch it'. Osman accepts this, and
the three children start playing hide-and-seek. Ahmet becomes the seeker. Seda and
Osman looks fo r a place to hide.
KERMIT: Osman cocuk-lar-la oyna-ma-di
Osman-child-plu-com play-neg-past
'Osman did not play with the children'
Expected response: Reject
Warm-up-02: Seda hemen agacin arkasina saklanmis. Osman once masanin altina
saklanmak istemis. Ama Ahmet beni burada hemen bulur demis, baska bir yer aramis.
Firinin arkasina mi saklansam acaba demis, ama orayi da begenmemis. Ama en sonunda
gene masanin altina saklanmaya karar vermis. / Seda hides behind the tree. Osman
decides to go under the table at first, then he changes his mind. He looks fo r another
place, goes behind the stove, but then thinks that the table was a safer place to hide. Goes
back to the table and hides under it.
KERMIT: Osman masa-nin alt-i-na gir-me-di
Osman table-gen&3S under-poss&3S-dat enter-neg-past
'Osman did not go under the table'
Expected response: Reject
297
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Warm-up-03: Ahmet Sedayi hemen bulmus. Osmani aramis aramis bir turlu bulamamis.
Sonunda “Osman hadi cik artik, bulamiyoum seni” demis. Osman cok sevinmis, hemen
gelmis. Ikinci defa oynamaya baslamislar. Bu sefer de Seda ebe olmus. Ahmet hemen
mutfakta bir yere saklanmis. Osman saklanacak yer aramis. Su agaca ciksam, Seda beni
orada bulamaz demis. Ama agaca cikmak hie de kolay degilmis, hem dusup bir yerini
incitebilirmis. Sonunda vazgecip, yine eski yerine, masanin altina saklanmaya karar
vermis. Hem nasil olsa onu orada bulamiyorlarmis. I Ahmet finds Seda immediately. He
cannot find Osman, and then in the end he says 'Okay, I cannot fin d you!'. Osman comes
out o f the table and they decide to play again. This time Seda becomes the seeker. Ahmet
hides somewhere in the kitchen. Osman looks fo r somewhere in the garden. He says that
the tree is the safest place the hide. He wants to climb up the tree. But it was not easy to
climb up the tree. He doesn’ t want to fa ll down and injure himself. He thinks that hiding
under the table again is a better idea. They could not fin d him there last time! So he goes
under the table again.
KERMIT: Osman agac-a cik-ma-di
Osman tree-dat climb-neg-past
'Osman did not climb-up the tree'
Expected response: Accept
Test Sentences/Scenarios
Test-01: Seda cok susamis. Hemen mutfaga gitmis birseyler icmek icin. Bir bakmis
masanin uzerinde iki tane gazoz var. Ama cani gazoz icmek istemiyormus. Portakal suyu
var mi acaba diye dolaba bakmis, ama dolapta portakal suyu yokmus... O da mecburen
gazozlara donmus. Hemen almis bir tanesini likir likir icmis. Tam otekini de iciyormus
arkadasi Ayse gelmis. “aaa Seda, gazoz mu iciyorsun? Bana da versene” demis. Seda da,
ben bir tane icmistim zaten. A1 bunu da sen ic demis, obur gazozu arkadasina vermis./
Seda is very thirsty. She goes to the kitchen and sees two bottles o f soda on the table. But
she doesn’ t want to drink soda. She wants orange juice. She goes and checks the fridge...
but cannot find anything else to drink... So she goes back to the sodas. She drinks one o f
them, when is about to drink the other one, her friend Ayse arrives, 'oh are you drinking a
soda?', she says. 'Give me one too!' Seda gives the soda to her friend and says 'I already
drank one, take this one'. Ayse takes the soda and drinks it.
KERMIT: Seda bir gazoz-u ic-me-di (version-1)
Seda a soda-acc drink-neg-past
'Seda did not drink a soda'
KERM IT: Bir gazoz-u Seda ic-me-di (version-2)
a soda-acc Seda drink-neg-past
298
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
'Seda did not drink a soda'
KERM IT: Seda icmedi bir gazozu (version-3)
Seda drink-neg-pasta a soda-acc
'Seda did not drink a soda'
K ERM IT: Seda bir gazoz icmedi (version-4)
Seda a soda drink-neg-past
'Seda did not drink a soda'
K ER M IT: Seda gazoz icmedi (version-5)
Seda soda drink-neg-past
'Seda did not soda-drink'
CHILD: Accept (wide reading) / Reject (narrow reading)
Test-02: Ayse'nin iki kecisi varmis. Kecilerin yemek zamani gelmis. Ayse onlari
beslemek icin kocaman iki sepet ot doldurmus... Ama yolda gelirken sepetlerden
birindeki ot dokulmus... Ayse otun dokuldugunu farketmemis. Sepetleri getirip kecilerin
onune koymus. Kecilerden biri otunu afiyetle yemis. Ama obur keci ac kalmis... O da
bahceye yiyecek birsey bulmaya gitmis.. / Ayse has two goats and it is the lunchtime fo r
the goats. She fills two baskets with grass and takes them to the goats. But on her way,
she drops the grass in one o f the baskets. She does not notice that until she arrives where
the goats are. She puts the baskets in front o f the goats. One o f them eats up whatever it
has in his basket. The other one does not have anything.
K ER M IT: Ayse bir keci-yi besle-me-di (version-1)
Ayse a goat-acc feed-neg-past
'Ayse did not feed a goat'
KERM IT: Bir keci-yi Ayse besle-me-di (version-2)
Ayse a goat-acc feed-neg-past
'Ayse did not feed a goat'
KERM IT: Ayse besle-me-di bir keci-yi (version-3)
Ayse feed-neg-past a goat-acc
'Ayse did not feed a goat'
KERM IT: Ayse bir keci besle-me-di (version-4)
Ayse a goat feed-neg-past
'Ayse did not feed a goat'
KERM IT: Ayse keci beslemedi (version-5)
299
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Ayse goat feed-neg-past
'Ayse did not goat-feed'
CHILD: Accept (wide reading) / Reject (narrow reading)
Test-03: Kami ac olan keci bahcede dolasip yiyecek ariyormus. Bakmis otlarin arasinda
iki cicek gormus. Ama cani cicek yemek istemiyormus ama baska birsey bulamamis.
Gidip yakindaki cicegi afiyetle yemis. Biraz yukarida bir yerde bir cicek daha varmis.
Ama o cicege bir turlu boyu yetismemis. O da onu birakmis... O gittikten sonra bir at
gelmis, kalan cicegi de o at yemis. Atin boyu uzun oldugu icin yukaridaki ciceklere bile
yetisebiliyormus. / The goat who was still hungry was searching the garden to find
something to eat. He sees two flowers among the grass. He doesn’ t like flowers. He
doesn’t want to eat them.. He searches fo r something else, but no w ay’ .There is nothing to
eat around... Finally goes to one o f the flowers and eats it up. He wants the other one
too. But the other flower is somewhere high,. He tries to reach it, but no matter how hard
he tries he cannot get it. In the end he gives up and turns back home. A horse that was
passing by finds the flower after the goat leaves and eats it. He can reach the flower
because he is a very tall horse.
KERMIT: Keci bir ciceg-i ye-me-di (version-1)
goat a flower-acc eat-neg-past
'The goat did not eat a flower'
KERMIT: Bir ciceg-i keci ye-me-di (version-2)
a flower-acc goat eat-neg-past
'The goat did not eat a flower'
KERMIT: Keci ye-me-di bir ciceg-i (version-3)
goat eat-neg-past a flower-acc
'The goat did not eat a flower'
KERMIT: Keci bir cicek ye-me-di (version-4)
goat a flower eat-neg-past
'The goat did not eat a flower'
KERMIT: Keci cicek ye-me-di (version-5)
goat flower eat-neg-past
'The goat did not flower-eat'
CHILD: Accept (wide reading) / Reject (narrow reading)
Test-04: Ayse ve Seda yemeklerini yemisler. Seda hemen yemekten kalkar kalmaz
disariya cikmis... Ayse arkasindan seslenmis... Oyuna baslamadan once bulasiklari
yikalamalari gerekiyormus. Seda mutfaga geri donmus ve Ayse’ye “ben tabaklari
yikayayim, sen de tencereyi yika’ demis. Ayse de Kabul etmis. Seda tabaklardan bir
300
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
tanesini yikamis. Ama obur tabagi masada unutmus, Ahmet ile konusmaya dalmis. Ayse
de “seda tabaklarda birini yikamayi unuttu galiba, olsun, o tabagi da ben yikarim demis.
O kalan tabagi da Ayse yikamis. / Ayse and Seda ate their meal. Right after the meal,
Seda runs to the garden to play. Ayse wants her to help her with the dishes... She comes
back and tells Ayse that she would like to wash the plates and asks Ayse whether she
wants to do the pot. She goes to the sink washes one o f the plates, and then she starts
talking to Ahmet who was passing by. She forgets the other plate on the table. Ayse
notices that the other plate is not washed. She goes to the sink and washes it.
K E R M IT: Seda bir tabag-i yika-ma-di (version-1)
Seda a plate-acc wash-neg-past
'Seda did not wash a plate'
K E R M IT: Bir tabagi Seda yikamadi (version-2)
a plate-acc Seda wash-neg-past
'Seda did not wash a plate'
K E R M IT : Seda yikamadi bir tabagi (version-3)
Seda wash-neg-past a plate-acc
'Seda did not wash a plate'
K E R M IT: Seda bir tabak yikamadi (version-4)
Seda a plate wash-neg-past
'Seda did not wash a plate'
K E R M IT: Seda tabak yikamadi (version-5)
Seda plate wash-neg-past
'Seda did not plate-wash'
CHILD: Accept (wide reading) / Reject (narrow reading)
Test-05: Ayse ve Seda cay iceceklermis. Seda servis yapmak icin Ayse’nin gelmesini
bekliyormus ama Ayse cok gee kalmis. Seda da ben en iyisi caylari bardaklara koymaya
baslayayimdemis. Caydanligi almis, kendi bardagini doldurmus. Ama o sirada
caydanliktan eli yanmis. Cani cok yandigi icin caydanligi Ayse'ye vermis. "A1 Ayse sen
de kendi bardagini doldur demis." Ayse de kendi bardagini doldurmus. / Ayse and Seda
would like to dink tea. Seda doesn’ t want to serve the tea and waits fo r Ayse to come join
her. But Ayse is late, so she decides to start serving... She pours it in one o f the cups and
fills it. When she is done, she spills some on her hand and burns herself. In the
meanwhile, Ayse arrives. Because Seda’ s hand hurts a little, she asks Ayse to fill the
other cup.
301
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
KERMIT: Seda bir bardag-i doldur-ma-di (version-1)
Seda a cup-acc fill-neg-past
'Seda did not fill a cup'
KERMIT: Bir bardag-i Seda doldur-ma-di (version-2)
a cup-acc Seda fill-neg-past
'Seda did not fill a cup'
KERMIT: Seda doldur-ma-di bir bardag-i (version-3)
Seda fill-neg-past a cup-acc
'Seda did not fill a cup'
KERMIT: Seda bir bardak doldur-ma-di (version-4)
Seda a cup fill-neg-past
'Seda did not fill a cup'
KERMIT: Seda bardak doldurmadi (version-5)
Seda cup fill-neg-past
'Seda did not cup-fill'
CHILD: Accept (wide reading) / Reject (narrow reading)
Test-06: Seda ve Ayse masayi ve sandalyeleri bahceye cikarip orada bir atlama oyunu
oynayacaklarmis. Once acaba sepetlerin ustunden mi atlasak diye dusunmusle... Ama
sepetler cok kolaymis. O yuzden sandalyelerle oynamaya karar vermisler. Ayse ben iki
sandalyeyi de gotururum, sen de masayi getir demis. Almis iki sandalyeyi ust uste
koymus. Ama bu sefer sandalyeler cok agir olmus. Teker teker goturmeye karar vermis.
Bir sandalyeyi bahceye kadar tasimis. Seda’ya “hadi sen de kendi sandalyeni getir. Sonra
masayi da beraber tasiyalim,” demis. Kalan sandalyeyi de bahceye Seda cikarmis. / Seda
and Ayse decide to take the chairs to the garden and play a jumping game there. First
they cannot decide what they can jum p over. They leave the chairs in the kitchen, and try
the baskets... But baskets are too easy. So they end up taking the chairs again. Ayse tells
Seda that she can take them both. She puts one on top o f the other and try to carry them
together. Because they are very heavy, she cannot carry them. She decides to take them
one by one and leaves one behind. She takes one to the garden. Seda takes the other to
help her.
302
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
KERMIT: Ayse bir sandalye-yi gotur-me-di (version-1)
Ayse a chair-acc take-neg-past
'Ayse did not take a chair'
KERMIT: Bir sandalye-yi Ayse gotur-me-di (version-2)
a chair-acc Ayse take-neg-past
'Ayse did not take a chair'
KERMIT: Ayse gotur-me-di bir sandalye-yi (version-3)
Ayse take-neg-past a chair-acc
'Ayse did not take a chair'
KERMIT: Ayse bir sandalye gotur-medi (version-4)
Ayse a chair take-neg-past
'Ayse did not take a chair'
KERMIT: Ayse sandalye gotur-me-di (version-5)
Ayse chair take-neg-past
'Ayse did not chair-take'
CHILD: Accept (wide reading) / Reject (narrow reading)
Additional test sentences/scenarios for version 4 & 5 (Experiment 5)
Test-07: Seda ve Ayse sise tasima oyunu oynuyorlarmis. Iki siseyi ve bir tabagi
dusurmeden bir tepsi uzerine koyarak ve sandalyelerin uzerinden atlayarak gecmeleri
gerekiyormus. Seda siselerden birini dusurmeden gecirmis. Sira obur siseye gelmis.
Boyle sise tasimak cok kolay, ben siseyle beraber tabagi da goturebilirim demis. Ikisini
de tepsiye koymus. Ama busefer de tepsi cok agir olmus. Yine de ilk sandalyenin ve
sepetin uzerinden gecmis. Tam son sandalyenin uzerinden atlayacakken elindeki tepsi
devrilmis, sise dusmus. / Seda and Ayse are playing a bottle-carrying game. They will try
to carry two bottles and a plate on a plate jumping over two chairs. Seda takes one o f the
bottles and can take them over to the other side. Then she goes back, takes the other
bottle. She takes the plate too, thinking that she can carry two objects. She jumps over the
first chair without a problem, but then when she is jumping over the last chair, the tray
turns over and she drops the bottle.
KERMIT: Seda bir sise dusur-me-di (version-4)
Seda a bottle drop-neg-past
'Seda did not drop a bottle'
KERMIT: Seda sise dusur-me-di (version-5)
Seda bottle drop-neg-past
'Seda did not bottle-drop'
303
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHILD: Accept (wide reading) / Reject (narrow reading)
Test-08: Ahmet bir parkta dolasiyormus. Cok guzel iki cicek gormus. Bunlari Ayse’ye ve
Sedaya gotursem ne kadar cok sevinirler demis ama o parkta cicek koparmak yasakmis.
Hem de park bekcisi ciceklerin yaninda duruyormus. Ahmet de cicek koparadan oradan
uzaklasmis. Ama sonradan bekci ciceklerin yanindan ayrilinca, hemen geri donmus,
ciceklerden birini koparmis, kacmis... /A hm et is walking around at a park. He sees two
very beautiful flowers. He wants to pick them and take them to Ayse and Seda. But picking-
up flowers is prohibited in this park and the park guard was there. He goes away without
picking up the flowers and then he comes back after the park-guard leaves. He picks up a
flower and runs away.
KERMIT: Ahmet bir cicek kopar-ma-di (version-4)
Ahmet a flower pick-neg-past
'Ahmet did not pick up a flower'
K ER M IT: Ahmet cicek kopar-ma-di (version-5)
Ahmet flower pick-neg-past
'Ahmet did not flower-pick'
CHILD: Accept (wide reading) / Reject (narrow reading)
Test-09: Seda arkadaslarina kahve ikram edecekmis. Iki fincana kahve doldurmus. Acaba
iki fmcani birden kahveyi dokmeden goturebilir miyim diye dusunmus. Yavas yavas
gidersem kahve dokulmez demis. Ama tam guzel guzel yoluna devam ederken kedisi
ayaklarina dolanmis, kahve fincanlarindan biri devrilivermis...Neyseki ikisi birden
dokulmemis. / Seda is serving coffee. She fills two cups with coffee but she doesn’ t know
whether she can take both o f the cups at the same time. They may be too heavy and it is a
long way to. She says she can do it if she walks really slowly. She walks to the garden,
when she was about the deliver the coffee, one o f the coffee cups fa ll from the tray.
KERMIT: Seda bir kahve dok-me-di (version-4)
Seda a coffee spill-neg-past
'Seda did not spill a coffee'
KERMIT: Seda kahve dok-me-di (version-5)
Seda a coffee spill-neg-past
'Seda did not spill a coffee'
CHILD: Accept (wide reading) / Reject (narrow reading)
Test-10: Ayse ve Ahmet bir atlama oyunu oynuyorlarmis. Bir sepet ve iki sandalyeyi yan
yana dizmisler. Onlari devirmeden ustlerinden atlamaya calisiyorlarmis. Ayse sepetin
ustunden atlamis. Birinci sandalyenin ustunden de atlamis. Bu atlama isi cok kolaymis,
ters donerek bile bu sandalyeyi devirmeden atlabilirim" demis. Arkasini donmus,
304
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
atlamaya calismis, ama bu sefer becerememis, sadalyeye takilmis, sandalye ile birlikte
dusmus. Ahmet and Seda are playing a jumping game. The one who can jum p over two
chairs without touching them (turning over them) will be the winner. Seda starts the
game. She jumps over one o f the chairs without a problem. Then she says ‘ this is so easy!
I can jum p even backwards! ’ She turns back, and tries to jump. But this time she hits the
chair and the chair is turned over.
KERMIT: Seda bir sandalye devir-me-di (version-4)
Seda a chair tum-over-neg-past
'Seda did not turn over a chair'
KERMIT: Seda sandalye devir-me-di (version-5)
Seda a chair tum-over-neg-past
'Seda did not chair-tum-over'
CHILD: Accept (wide reading) / Reject (narrow reading)
Test-11: Ayse kecileriyle bahcede oynuyormus. Sonra keciler yaramazlik olsun diye
kacip gitmisler. Ayse de arkalarindan kosmus. Ne kadar hizli kacarlarsa kacsinlar, onlari
nasil olsa yakalarim. Cunku ben cok hizli kosabiliyorum demis... Kecilerden birini hemen
yakalamis. Ama obur keci cok uzaga kacmis, gitmis, kaybolmus. Onu her yerde aramis,
bulmaya calismis ama bir turlu bulamamis./ Ayse is playing with her goats in the garden.
And then all o f a sudden, the goats run away. Ayse tries to chase them. She thinks that no
matter how fast they go, she can get them back... She jumps on one o f the goats... It was
easy to catch... But while she was trying to catch this one, the other one goes farther
away... and Ayse loses it.
KERMIT: Ayse bir keci kaybet-me-di (version-4)
Ayse a goat lose-neg-past
'Ayse did not lose a goat'
KERMIT: Ayse keci kaybet-me-di (version-5)
Ayse a goat lose-neg-past
'Ayse did not goat-lose'
CHILD: Accept (wide reading) / Reject (narrow reading)
Test-12: Ayse iki tane balik yakalamis, getirmis, aksama pisirecekmis. Kedisi hemen
baliklarin kokusunu almis. Ama Ayse kediye sakin baliklara dokunma, onlari biz
yiyecegiz, sen kendi yemegini ye demis. Kedi de tamam deyip kendi yiyecegini yemeye
baslamis... Ama baliklar cok guzel kokuyormus, kedinin de kami cok acmis,
dayanamayip, baliklardan bir tanesini alip yiyivermis. Neredeyse obur baligi da
yiyecekmis ama Ayse'nin geldigini duyunca kacmis saklanmis. / Ayse catches two fish.
She takes them to the kitchen. She tells her cat not to touch the fish... She gives the cat
some cat food to eat... But the fish smells so good. The cat eats his cat food and then she
305
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
cannot help going to the fish... She grabs one o f the fish and starts eating it... Then Ayse
comes back and sees the scene. She is so angry with the cat. The cat runs away...
KERMIT: Kedi bir balik ye-me-di (version-4)
Cat a fish eat-neg-past
'The cat did not eat a fish'
KERMIT: Kedi balik ye-me-di (version-5)
Cat fish eat-neg-past
'The cat did not fish-eat'
CHILD: Accept (wide reading) / Reject (narrow reading)
Fillers / Control items
Filler-01: Ayse’nin iki ati varmis. Seda’ya hadi gel bereber ata binelim demis. Ama Seda
ata binmekten korkuyormus. "Ben ata binmek istemiyorum, cok korkuyorum," demis.
Ayse sonunda Seda’yi ikna etmis, ve beraber atlara binip dolasmaya baslamislar. / Ayse
has two horses. She invites Seda to horseback riding. Seda says she was afraid o f horses
and doesn’ t want to ride on a horse... Ayse convinces her to give it a try...
KERMIT: Seda at-a bin-me-di.
Seda horse-dat ride-neg-past
'Seda did not ride on a horse'
Expected response: Reject
Filler-02: Seda atla dolasirken atin onune bir keci cikmis. At keciden cok korktugu icin
ziplamaya baslamis. Seda da cok korkmus tabii. Eyvah dusuyorum galiba demis. At
ziplamis ziplamis. Seda’yi ustunden atmis (version 1) / Seda ata sikica tutunmus,
dusmeden durmayi basarmis (version-2). / While they were riding their horses around the
woods, a goat gets into their way. The horse is startled. He starts jumping... Seda is so
afraid that she will fall... Ayse tells her to hold on the horse tight... But it doesn’ t help
much. She falls from the horse (version 1) / ... Son the horse calms down and Seda remains
on the horse... (version 2)
KERMIT: Seda at-tan dus-me-di.
Seda horse-abl fall-neg-past
'Seda did not fall from the horse.'
Expected response: Reject (version 1) / Accept (version 2)
Filler-03: Seda ve Ayse mutfaga donmusler. Ayse Seda’ya “hadi sen su sandalyeye otur,
ben de yemek hazirlayayim,” demis. Seda da oturup, Ayse’nin yemek hazirlamasini
beklemis (version-1). / Seda, “olmaz ben de sana yardim edeyim demis, beraber yemek
306
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
hazirlamislar (version-2) Sonra beraber yemek yemisler. / .Seda and Ayse go back to the
kitchen. Ayse tells Seda to have a seat. Seda takes a seat and waits fo r Ayse to prepare
the meal (version 1) / Seda doesn’ t want to sit and wait Ayse. So she offers help and they
prepare the meal together, (version 2)
KERMIT: Seda sandalye-ye otur-ma-di.
Seda chair-dat sit-neg-past
'Seda did not sit on a chair'
Expected response: Reject (version 1) / Accept (version 2)
Filler-04: Seda ve Ayse oturlarken, birden bir fare cikip Seda’nin sandalyesinin altindan
gecmis gitmis. Seda fareden cok korkarmis. Hemen sandalyesinin ustune cikmis, “fare,
fare, fare var burada” diye bagirmaya baslamis. Fare sozunu duyar duymaz, kedisi
kosmus gelmis. Fareyi kovalamaya baslamis. Fare kacmis, kedi kovalamis, fare kacmis
kedi kovalamis... Sonra fare birden gozden kaybolmus. Kedi de cok yoruldugu icin
vazgecip, geri donmus./ While Ayse and Seda are sitting in the garden, a mouse appears.
Seda is so scared. She climbed up her chair and started screaming. Her cat hears her
voice and comes immediately to catch the mouse. She looks around but cannot find the
mouse.
KERMIT: Kedi fare-yi yakala-ma-di
Cat mouse-acc catch-neg-past
'The cat did not catch the fish'
Expected response: Accept (version 1&2)
Filler-05: Ahmet alisverise gidiyormus. Seda'ya gelip, "hadi sen de benimle gel," demis.
Seda o sirada Ayse ile oynuyormus. Ahmet'e "biz simdi oynuyoruz ben alisverise gitmek
istemiyorum demis. Ahmet de alisverise tek basina gitmek zorunda kalmis. / Ahmet was
going fo r shopping. He invited Seda too. But Seda was playing with Ayse in the garden
and she refused the invitation. Ahmet went to shopping alone.
KERMIT: Seda alisveris-e git-me-di
Seda shopping-dat go-neg-past
'Seda did not go shopping'
Epected response: Accept (version 1&2)
307
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Grammaticalization and the development of functional categories in Chinese

PDF

Connectionist phonology

PDF

Beyond words and phrases: A unified theory of predicate composition

PDF

Asymmetry of scope taking in wh -questions

PDF

A descriptive grammar of San Bartolome Zoogocho Zapotec

PDF

Factors influencing the interpretation of novel words as adjectives in 4-year-old Spanish -speaking children

PDF

Differences in the development of citizenship: Residents' motivations and capacities for participating in neighborhood associations in Los Angeles

PDF

Distributional cues and subject identification in the production of subject -verb agreement

PDF

Children in transition: Popular children's magazines in late imperial and early Soviet Russia

PDF

Issues in the syntax of resumption: Restrictive relatives in Lebanese Arabic

PDF

"Master of many tongues": The Russian Academy Dictionary (1789--1794) as a socio -historical document

PDF

Gossip, letters, phones: The scandal of female networks in film and literature

PDF

A comparative study of focus constructions

PDF

Number marking and definiteness in Bangla

PDF

Form and meaning: Negation and question in Chinese

PDF

Access factors and homerun experiences: Making readers of our children

PDF

Imagining China: "Niehai Hua" as a national narrative

PDF

A longitudinal study into the learning style preferences of university ESL students

PDF

A cut above: Fashion as meta -culture in early -twentieth -century Russia

PDF

Direct objects in Persian

Asset Metadata

Creator Ketrez, Fatma Nihan (author)

Core Title Children's scope of indefinite objects

School Graduate School

Degree Doctor of Philosophy

Degree Program Linguistics

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag language, linguistics,OAI-PMH Harvest

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Borer, Hagit (committee member), Hyams, Nina (committee member), Kural, Murat (committee member), Li, Audrey (committee member), Mintz, Toben H, (committee member), Pancheva, Roumyana (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c16-453888

Unique identifier UC11340987

Identifier 3196829.pdf (filename),usctheses-c16-453888 (legacy record id)

Legacy Identifier 3196829.pdf

Dmrecord 453888

Document Type Dissertation

Rights Ketrez, Fatma Nihan

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA