Hacettepe University Graduate School of Social Sciences 

Department of Linguistics 

 

 

  

 

THE ANALYSIS OF THE L2 PERFORMANCES OF TURKISH 

SPEAKERS ON THE ISLAND CONSTRAINTS IN ENGLISH TO 

ASSESS THE VALIDITY OF THE INTERPRETABILITY 

HYPOTHESIS 

 

 

 

 

Sinan ÇAKIR 

 

 

 

 

A PhD Dissertation 

 

 

 

Ankara, 2014 



 
 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 



 
 

THE ANALYSIS OF THE L2 PERFORMANCES OF TURKISH SPEAKERS ON THE 

ISLAND CONSTRAINTS IN ENGLISH TO ASSESS THE VALIDITY OF THE 

INTERPRETABILITY HYPOTHESIS 

 

 

 

 

 

 

Sinan ÇAKIR 

 

 

 

 

Hacettepe University  Graduate School of Social Sciences 

Department of Linguistics 

 

 

 

 

 

A PhD Dissertation 

 

 

 

 

 

Ankara, 2014 



 
 



 
 



iii 
 

ACKNOWLEDGEMENTS 

 

 

My time at Hacettepe University has been influenced and guided by a number of people 

to whom I am deeply indebted. Without their help and support, this dissertation would 

never have been completed. Prof. Dr. IĢıl ÖZYILDIRIM, Prof. Dr. Deniz ZEYREK and 

Prof. Dr. Nalan BÜYÜKKANTARCIOĞLU have had the greatest impact on the 

preparation of this dissertation, so I thank them for their insights and guidance. I feel 

most fortunate to have worked with other committee members Prof. Dr. Özgür AYDIN 

and Assoc. Prof. Dr. Selçuk ĠġSEVER. 

 

I also would like to thank Assoc. Prof. Dr. Jason ROTHMAN, for his guidance and 

recommendations during the time I was in the University of Florida. Thanks are also 

due to all participants who contributed to my study willingly. I would also like to thank 

my true friend Arif BAKLA, who helped me constantly in the preparation of this 

dissertation. My thanks are also reserved for my colleagues in the Department of 

Linguistics in Hacettepe University. 

  

My deepest gratitude is reserved for my wife, Zeynep AYDIN ÇAKIR and my 

daughter, Elvin Cemre ÇAKIR. Without their constant love, support and 

encouragement, I would never have been able to produce this dissertation. 

 

 

 

 

 

 

 

 

 



iv 
 

ÖZET 

 

ÇAKIR, Sinan. Yorumlanabilirlilik Varsayımının Geçerliliğini Değerlendirmek Ġçin Ġngilizceyi 

Ġkinci Dil Olarak Edinen Türklerin Ġngilizcedeki Ada Kısıtlamaları Üzerindeki 

Performanslarının Ġncelenmesi, Doktora Tezi, Ankara, 2014. 

 

Evrensel Dilbilgisi ve ikinci dil edinimi arasındaki iliĢki üzerine geliĢtirilmiĢ son yıllarda öne çıkan 

varsayımlardan birisi Tsimpli ve diğerleri (2003), ve Tsimpli ve Dimitrakopoulou‟nun (2007) ortaya attığı 

Yorumlanabilirlilik Varsayımıdır.  Bu varsayıma göre sadece yorumlanabilir özellikler ve ana dilde var 

olan yorumlanamayan özellikler ikinci dil edinimi için eriĢilebilir durumdadır. Bu varsayımı destekleyen 

(Hawkins ve Hattori, 2006; Al-Thubaiti, 2011, gibi) ya da bu varsayıma karĢı duruĢ sergileyen (Rothman 

ve diğerleri, 2010; Bond ve diğerleri, 2011, gibi) çok sayıda çalıĢma mevcuttur. Bir baĢka deyiĢle, bu 

konudaki tartıĢmalar henüz son bulmamıĢtır. Yorumlanabilirlilik Varsayımı üzerine yapılmıĢ mevcut 

çalıĢmaların hiçbiri Türkçe-Ġngilizce dil çifti üzerine odaklanmamıĢtır. Ġngilizcede var olan 

yorumlanamayan ne-özelliğinin Türk ikinci dil öğrenicileri tarafından edinim sürecinin incelenmesi bu 

konudaki tartıĢma üzerine yararlı veriler ortaya sunacaktır.  

 

Bu çalıĢma, Ġngilizceyi ikinci dil olarak edinen Türklerin Ġngilizcedeki ne-taĢıma sırasında ortaya çıkan 

ada kısıtlamaları üzerindeki performanslarını inceleyerek Yorumlanabilirlilik Varsayımının geçerliliğini 

değerlendirmeyi amaçlamaktadır. ÇalıĢmanın verileri bir dilbilgisellik değerlendirme testi, ne-sorusu 

oluĢturma testi ve Türkçeden Ġngilizceye çeviri testi vasıtasıyla toplanmıĢtır. Ġngilizce anadil 

konuĢucularından oluĢan bir kontrol grupla (N:58) beraber, yaĢadıkları ülkeye (ABD ya da Türkiye) ve 

Ġngilizce yeterlilik düzeylerine (Ġleri düzey yada orta düzey) göre dört öğrenici grubu çalıĢmaya dahil 

olmuĢtur (sırasıyla N:46, N:38, N:20, N:30).  

Toplanılan veriler Kruskal Wallis H Testi ve Mann Whitney U Testi vasıtasıyla istatistiksel olarak 

incelenmiĢtir. Hedef dilde doğal veriye maruz kalmıĢ ve bu dilde üst seviye yeterliliğe sahip ikinci dil 

öğrenicilerinin ada yapılarını fark etmede ve dilbilgisel tümceler üretmede Ġngilizce anadil konuĢucuları 

kadar baĢarılı olduğu gözlemlenmiĢtir. Türkçede yorumlanabilir olmayan güçlü ne özelliğinin (uwh*) var 

olmadığı dikkate alındığında, bu katılımcıların performansı onların hedef dildeki parametrik değerleri 

edindikleri ve yorumlanabilir olmayan güçlü ne özelliğinin (uwh*) ikinci dil edinimi için de edinilebilir 

olduğu sonucunu ortaya koymaktadır.  Sonuç olarak çalıĢmanın bulguları Yorumlanabilirlilik 

Varsayımına karĢı çıkmakta ve Tam Transfer & Tam EriĢim Varsayımını desteklemektedir.  

ÇalıĢmanın bulguları ikinci dil edinimi sırasında doğal veriye maruz kalmanın önemini de 

vurgulamaktadır. Yapılan testlerin sonuçları, (uwh*) özelliğinin, sadece üst seviye Ġngilizce yeterliliğine 

sahip ve bu dilde doğal veriye maruz kalmıĢ ikinci dil öğrenicileri tarafından tam olarak edinilmiĢ 

olduğunu ortaya koymuĢtur. Bu açıdan, Evrensel Dilbilgisine eriĢim ve ikinci dil edinimi arasındaki 



v 
 

iliĢkiyi inceleyen çalıĢmaların doğal veriye maruz kalan bireyler üzerinde yürütülmesi gerekliliği ortaya 

çıkmaktadır.  

 

Anahtar Sözcükler  

Yorumlanabilirlilik Varsayımı, Ġkinci Dil Edinimi, Evrensel Dilbilgisi, Ada Kısıtlamaları  



vi 
 

ABSTRACT 

 

ÇAKIR, Sinan. The Analysis of the L2 Performances of Turkish Speakers on the Island Constraints in 

English to Assess the Validity of the Interpretability Hypothesis, A PhD Dissertation, Ankara, 2014. 

 

The Interpretability Hypothesis (Tsimpli et. al., 2003; Tsimpli and Dimitrakopoulou; 2007) claims that 

uninterpretable features which are not instantiated in L1 are unavailable for L2 acquisition. This 

hypothesis is supported (e.g. Hawkins and Hattori 2006; Al-Thubaiti, 2011) and opposed (e.g. Rothman 

et. al., 2010; Bond et. al. 2011) by many other studies, and this hypothesis is still hotly debated. In such 

studies, different language pairs should be extensively analyzed to reach at more concrete results. None of 

the recent studies that focus on Interpretability Hypothesis focus on Turkish-English language pairs. The 

acquisition process of the uninterpretable wh-feature in English by Turkish L2 learners of English might 

provide fruitful results for this debate.  

The present study aimed to analyze the validity of the Interpretability Hypothesis by analyzing the 

performances of the Turkish L2 leaners of English on island constraints on wh-movement in English. The 

data of the study were collected through a Grammaticality Judgment Test, Wh-Question Formation Test 

and Translation Test. Along with a native control group (N:58), four  learner groups were formed 

according to the place they live (USA or Turkey) and their level of proficiency in English (advance or 

intermediate) (N:46, N:38, N:20, N:30 respectively).   

The data gathered were statistically analyzed by using two non-parametric tests: Kruskal-Wallis H Test 

and Mann-Whitney U Test. It was observed that L2 learners of English who are exposed to positive 

evidence in a naturalistic learning environment in the target language can deal with these island 

constraints as well as the native speakers of this language. Since their mother tongue lacks the 

uninterpretable strong wh-feature (uwh*), this result suggests that they have already acquired the 

necessary L2 parameter values, and this uninterpretable feature is available in their L2 acquisition 

process.  Hence, the findings of the present study support the Full Transfer Full Access Hypothesis, 

standing against the Interpretability Hypothesis. 

The results of the study also emphasized the importance of positive evidence in L2 acquisition process. 

According to the results, the uninterpretable strong wh-feature (uwh*) appeared to be available only for 

the highly proficient L2 learner of English who are exposed to natural input in this language, yet not for 

the other group who acquired this language only in their home country.  As the results of the study 

suggest, to assess the availability of UG in SLA precisely,  such studies should be carried out on the 

participants who are exposed to natural input in the target language. 

 



vii 
 

Key Words 

Interpretability Hypothesis, Second Language Acquisition, Universal Grammar, Island Constraints 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 



viii 
 

TABLE OF CONTENTS 

 

KABUL VE ONAY ……………………......................................................................... i 

BİLDİRİM ………………………………………......................................................... ii 

ACKNOWLEDGEMENTS ………………………..................................................... iii 

ÖZET.............................................................................................................................  iv 

ABSTRACT ……….…..………………........................................…......…...……….. vi 

TABLE OF CONTENTS ..........................................................................................  viii 

LIST OF ABBREVIATIONS ....................................................................................  xv 

LIST OF TABLES ....................................................................................................  xvii 

LIST OF FIGURES…………………………………………………….…………. xviii 

CHAPTER 1: INTRODUCTION ................................................................................ 1 

1.1 . UG in L2 Acquisition ................................................................................. 1 

1.2 . Studies on the Interpretability Hypothesis& Full Transfer Full            

  Access Hypothesis Debate .............…….................................…………... 6 

1.3 . Statement of the Problem …………..…......…………………………… 14 

1.4 . The Purpose of the Study……………………………….…………….... 15 

1.5 . The Research Questions ……………………………………………….. 18 

1.6 . The Significance of the Study …………………………………….…… 18 

CHAPTER 2: THEORETICAL CONCEPTS .......................................................... 20 

          2.1. UG and L1 Acquisition ........................................………..………………. 20 

2.2. UG and L2 Acquisition ........................................….........…………………... 20 

                     2.2.1. Full Access Hypothesis ...................................................................21 

                     2.2.2. No Access Hypothesis ………….....................................................21 

                     2.2.3. Partial Access Hypothesis.....................................................................22 



ix 
 

                     2.1.4. Full Transfer / Partial Access Hypothesis ......................................  23 

                     2.2.5. No Transfer/ Full Access Hypothesis ............................................  23 

                     2.2.6. Full Transfer/ Full Access Hypothesis ............................................... 23 

                     2.2.7. Partial Transfer/ Full Access Hypothesis ........................................... 24 

                     2.2.8. Partial Transfer/ Partial Access Hypothesis ................................... 24 

 2.3. The Interpretability Hypothesis & Full Transfer Full Access 

                  Hypothesis ……………………………………………………………….. 25        

  2.3.1. Interface Hypothesis .................................................................... 26 

2.3.2. The Missing Surface Inflection Hypothesis ................................ 26 

2.3.3. Feature Assembly Hypothesis ..................................................... 27 

2.3.3. Morphological Underspecification Hypothesis ……................... 27 

2.4. Wh-phrases and Wh-movement ………………………………….......... 28 

 2.4.1. Wh-movement in English …………………………………........ 28 

 2.4.2. Wh-movement in Turkish …………………………………........ 29 

2.5. Island Constraints on Wh-movement ……………….……………………. 33 

 2.5.1. The Island Constraints proposed by Ross (1967)...…………….. 34 

  2.5.1.1. The Complex NP Constraint………….…………......... 34 

  2.5.1.2. Sentential Subject Constraint ……………....………… 35 

  2.5.1.3. Coordinate Structure Constraint ………...……………. 36 

 2.5.2. The Island Constraints proposed after Ross (1967)…….............. 37 

  2.5.2.1. Wh-island Constraint …………………………………. 37 

  2.5.2.2. Adjunct Island Constraint ……….………………......... 39 

  2.5.2.3. Left-Branch Constraint ……………..………………… 39 

  2.5.2.4. Right Roof Constraint ……………….………….......... 40 



x 
 

  2.5.2.5. Factive Island Constraint ……………………………... 40 

  2.5.2.6. Superiority Constraint ………………………………... 40 

  2.5.2.7. Negative Island Constraint ……………………............ 41 

 2.5.3. Subjacency …………………………………………………....... 41 

 2.5.4. The Condition on Extraction Domains and the Empty    

                     Category Principle ……..………………………….………......... 42 

 2.5.5. Barriers …………………………………………………………. 42 

 2.5.6. Island Constraints in the Minimalist Program….………………. 43 

 2.5.7. The Island Constraints in Turkish ……………………………… 45 

CHAPTER 3: METHODOLOGY ............................................................................. 54 

3.1. Participants ……………………………….………...…………………… 54 

3.1.1. The Control Group ……………………………………………... 54 

3.1.2. The Learner Groups ……………………..….………………….. 54 

3.1.2.1. Learner Groups who live in USA ……...……………... 55 

3.1.2.1. a. Learner Group 1 ……………………………. 55 

3.1.2.1. b. Learner Group 2……………………………. 56 

3.1.2.2. Learner Groups who live in Turkey ………………….. 56 

3.1.2.2. a. Learner Group 3 ……………………………. 56 

3.1.2.2. b. Learner Group 4 …………………………… 57 

3.2. Data Collection Tools …………………….……………………………... 57 

3.2.1. Michigan Placement Test …………………..…………………... 57 

3.2.2. The Grammaticality Judgment Test ……………………………. 58 

3.2.2.1. The Grammatical Test Items …………………………. 59 

3.2.2.2. The Test Items that Contain Wh-Island Violation……. 61 



xi 
 

    3.2.2.3. The Test Items that Contain Complex NP Island  

  Violation …………………………………………….. 62 

       3.2.2.4. The Test Items that Contain Sentential Subject     

                 Constraint Violation ……………………………….... 63 

3.2.2.5. The Test Items that Contain Adjunct Island   

              Constraint Violation ….…………….………………... 64 

3.2.3. The Wh- Question Formation Task ………………..…………... 65 

3.2.3.1. The Test Items on Wh- Island Constraint ….……….... 66 

3.2.3.2. The Test Items on Complex NP Island  

             Constraint …………………………………………….. 67 

3.2.3.3. The Test Items on Sentential Subject Constraint……....68 

3.2.3.4. The Test Items on Adjunct Island Constraint….............69 

3.2.3.5. The Test Items that do not Direct the Participants to  

             Violate Island Structures ………….……….................. 69 

3.2.4. The Translation Task …………………………………………... 70 

3.2.4.1. The Test Items on Wh- Island Constraint ……………. 72 

3.2.4.2. The Test Items on Complex NP Constraint ………….. 73 

3.2.4.3. The Test Items on Sentential Subject Constraint……... 74 

3.2.4.4. The Test Items on Adjunct Island Constraint ………… 75 

3.2.4.5. The Test Items that Do Not Focus on Any Island  

 Structure ……………………………………………..... 76 

3.3. Data Collection Process ……………........................................................ 78 

 3.3.1. Pilot Studies ……………………………………………………. 78 

3.3.2. Main Study ……………………………………………………... 81 



xii 
 

3.3.2.1. Data Collection Process for the Control Group ……… 81 

3.3.2.2. Data Collection Process for Learner Groups 1  

          and 2 ................................................................................. 82 

3.3.2.3. Data Collection Process for Learner Groups 3 

         and 4 ……….……………………………………...…….. 83 

3.4. Analysis of the Data …………………..………………………………… 84 

3.5. Limitations……………………………………………………………….. 85 

CHAPTER 4: FINDINGS OF THE STUDY ............................................................ 87 

4.1. Findings for the Grammaticality Judgment Test ………..…………… 87 

4.1.1. The Overall Results for the Grammaticality Judgment  

          Task …………………………………………………………….. 87 

4.1.2. The Grammaticality Judgment Task Results for Wh- Island  

          Constraint………………………………………………..……… 92 

4.1.3. The Grammaticality Judgment Task Results for Complex  

          NP Island Constraint………………………………….………… 94 

 4.1.4. The Grammaticality Judgment Task Results for Sentential 

   Subject Constraint …………………………………………….. 96 

 4.1.5. The Grammaticality Judgment Task Results for Adjunct Island 

                                Constraint ………………………………………………………... 97 

4.2. Findings for the Wh-Question Formation Task ……………...…….... 100 

4.2.1. The Overall Results for the Wh-Question Formation Task  ..… 100 

4.2.2. The Wh-Question Formation Task Results for Wh-Island 

          Constraint ……………………………………………….......... 102 



xiii 
 

4.2.3. The Wh-Question Formation Task Results for Complex NP 

           Island Constraint ……………………………………..………. 103 

4.2.4. The Wh-Question Formation Task Results for Sentential Subject  

          Constraint ………………………….…………….……………. 105 

4.2.5. The Wh-Question Formation Task Results for Adjunct Island 

          Constraint ……………………………………….…….………. 106 

4.2.6. The Avoidance Strategies Used by the Participants not to  

          Violate the Island Structures ………………..…..…………….. 108 

4.2.6.1. Providing a Short Response with a Simplex  

Clause …………………………………………...…………… 108 

4.2.6.2. Changing the Context ………………….....……...….. 111 

4.2.6.3. Parting the Sentence into two with the Use of  „and‟  

             Conjunction ……………………………………..…... 113 

4.2.6.4. Deleting the Item(s) that Causes Island  

             Violation…………………………………………...… 115 

4.2.7. Individual Analysis of Some Salient Items in the Wh-Question 

          Formation Task …………………….…………………..……... 118 

4.3. Findings for the Translation Task …………….……………………… 121 

4.3.1. The Overall Results for the Translation Task ………………… 122 

4.3.2. The Distribution of the Island Violations to the Groups ……… 123 

4.3.3. The Avoidance Strategies Used by the Participants not to Violate 

          the Island Structures …………………………………………... 125 

4.3.3.1. Pied-Piping ………………………………………….. 125 



xiv 
 

4.3.3.2. Changing the Sentences Structurally and  

             Semantically…………………………………………. 128 

4.3.3.3. Parting the Sentence into two with the Use of  „and‟ 

             Conjunction …………………………………………. 131 

4.3.3.4. Leaving in-situ ………………………………………. 134 

4.3.3.5. Using an Extra Wh- word in the Sentence ………….. 137 

4.3.4. Individual Analysis of Some Salient Items in the Translation  

          Task …………………………………………………………… 140 

4.3.4.1. The Analysis of the Test Items for Sentential Subject 

              Constraint…………………………………………… 140 

CHAPTER 5: DISCUSSION ...................................................................................  144 

5.1. The Analysis of the Findings in Accordance with the Research 

       Questions ……………………………………………………………….. 155 

CHAPTER 6: CONCLUSION ................................................................................. 160 

REFERENCES ……………………..……………………………………………… 164 

 

APPENDICES …………………..........……......……...…......................................... 174 

APPENDIX 1 ................................................................................................... 174 

APPENDIX 2 ................................................................................................... 180 

 

 

 

 

 



xv 
 

LIST OF ABBREVIATIONS 

 

AIC  : Adjunct Island Constraint 

ANOVA : Analysis of Variance 

BC  : Blocking Category 

CED  : Condition on Extraction Domain 

COMP  : Complementizer  

CNPC  : Complex Noun Phrase Constraint 

CP  : Complementizer Phrase 

CSC  : Coordinate Structure Constraint 

DP  : Determiner Phrase 

ECP  : Empty Category Principle 

FFFH  : Failed Functional Features Hypothesis 

FocP  : Focus Phrase 

GJT  : Grammaticality Judgment Test 

IP  : Inflectional Phrase 

LAD  : Language Acquisition Device 

LBC  : Left Branch Constraint 

LF  : Logical Form 

L1  : First Language 

L2  : Second Language 

NP  : Noun Phrase 



xvi 
 

NSP  : Null Subject Parameter 

OPC  : Overt Pronoun Constraint 

PF  : Phonetic Form 

PP  : Prepositional Phrase 

RRC  : Right Roof Constraint 

SLA  : Second Language Acquisition 

SPEC  : Specifier 

SPSS  : Statistical Package for Social Sciences  

SSC  : Sentential Subject Constraint 

TGG  : Transformational Generative Grammar 

TP  : Tense Phrase 

UG  : Universal Grammar 

uwh  : Uninterpretable wh-feature 

VP  : Verb Phrase 

vP  : Light Verb Phrase 

WIC  : Wh-Island Constraint 

 

 

 

 

 

 

 



xvii 
 

LIST OF TABLES 

 

Table 1. The Interpretation of Wh-expressions out of the Island Constraints in Turkish …..... 52 

Table 2. The Overall Results for the Wh-Question Formation Task ………………………... 100 

Table  3. The Wh-Question Formation Task Results for Wh-Island Constraint ………...……103  

Table 4. The Wh-Question Formation Task results for Complex NP Island Constraint ….….104 

Table 5. The Wh-Question Formation Task Results for Sentential Subject Constraint ….…. 105 

Table 6. The Wh-Question Formation Task results for Adjunct Island Constraint …...…..… 107 

Table 7. The Overall results for the Translation Task ………………………..……………… 122 

Table 8. The Distribution of the island violations to the groups ……...……………..………. 124 

 

 

 

 

 

 

 

 

 

 

 

 



xviii 
 

LIST OF FIGURES 

 

 

Figure 1. The overall success of the groups in the Grammaticality Judgment Task………… 88 

Figure 2. The overall success of the groups for the grammatical and ungrammatical test items in  

   the Grammaticality Judgment Task …………………………………………………. 90   

Figure 3. The Grammaticality Judgment Task results for Wh- Island Constraint ……………. 93 

Figure 4. The Grammaticality Judgment Task results for Complex NP Island Constraint .….   94 

Figure 5. The Grammaticality Judgment Task results for Sentential Subject Constraint .....…. 96 

Figure 6. The Grammaticality Judgment Task results for Adjunct Island Constraint ……...…. 98



1 
 

CHAPTER 1 

INTRODUCTION 

 

Generative Grammar emerged in 1950‟s as a result of work performed by Noam 

Chomsky and it was developed since then by his and his colleagues‟ studies. One of the 

outstanding arguments of this new framework was that people possess a kind of 

Language Faculty which is a part of human natural biological qualities. The innate 

linguistic knowledge that enables practically any child to learn any of about 6000 

existing languages at a given point in time is sometimes known as the „Universal 

Grammar‟. 

 

Universal Grammar (UG hereafter) is a theory of linguistics which claims that there are 

principles of grammar shared by all languages, and they are thought to be innate to 

human beings. It attempts to explain language acquisition in general rather than 

describing specific languages. It is part of an innate biologically endowed language 

faculty. It places limitations on grammars, constraining their form as well as how they 

operate. It includes invariant principles, as well as parameters. Thus, according to 

Chomsky, the child‟s language faculty incorporates a theory of Universal Grammar 

which includes a set of universal principles and a set of structural parameters. 

 

In this approach, it is claimed that human beings are pre-programmed for language 

learning. In other words, the acquisition of language is innate, and as soon as we are 

born, we start to acquire our native language by getting necessary input to our language 

acquisition faculty. According to Chomsky, there are some universal language principles 

that are shared by all languages, and language acquisition is an issue of parameter 

setting. 

 

1.1. UG IN L2 ACQUISITION 

 

When Generative Grammar was introduced by Noam Chomsky in 1950s, the 

accessibility of Universal Grammar on first language started to be discussed among 

linguistic environments. However, in the following years, another debate started to take 



2 
 

place among linguists: the accessibility of Universal Grammar in second language 

acquisition (SLA hereafter). Rod Ellis defines „Second Language Acquisition‟ as, “the 

subconscious or conscious processes by which a language other than the mother tongue 

is learnt in a natural or a tutored setting” (1985, p. 6).  This was a new area that had to 

be investigated by the generative grammarians. The accessibility of UG in this area had 

to be investigated as well. Hence, they started to have studies on this field as well as the 

ones on first language acquisition to assess the accessibility of UG. 

 

In the accessibility of UG in SLA, different possible scenarios claimed by the linguists 

are open to consideration. The linguists who investigate different aspects of SLA have 

different ideas on the accessibility of UG in SLA. The original hypotheses are 

summarized by Waber and Czendik (2002) as: 

 

1- No Access Hypothesis: UG is totally inaccessible to the adult L2 learner; learning 

takes place in terms of non-linguistic learning strategies. 

2- Partial Access Hypothesis: UG is partially available to the learner; only those 

parametric values characterizing the L1 grammar are available, the rest must be learnt in 

terms of non-linguistic learning strategies. 

3- Full Access Hypothesis: UG is fully available; differences in patterns of acquisition 

between L1 and L2 learners and the lack of completeness can be accounted for in other 

ways. It is possible to find supporters of all these hypotheses. As Herschensohn (1999) 

expresses;  

incompleteness of L2 parameter setting, the inability of L2ers to be 

complete in resetting parameter values, has been taken as evidence for No 

Access or Partial Access to UG, while L2 acquisition of parameter values 

not available in L1 is taken to support full access (p. 115). 

 

These three hypotheses are still alive today, yet with certain modifications and with 

different names. No Access Hypothesis which was originally developed by Clashen and 

Muysken (1986; 1989) and Bley-Vroman (1989) is also named as Fundamental 

Difference Hypothesis. In this hypothesis, L1 and L2 acquisition are claimed to be 

fundamentally different and L2 acquirers are believed to use other cognitive 

mechanisms in their acquisition process other than UG. Though UG is available for L1, 



3 
 

it is unavailable for L2 acquisition. Han (2004) and Long (2007) are some of the 

followers of this hypothesis. 

 

Full Access Hypothesis which was developed by Schwartz and Sprouse (1996) is 

recently uttered as Full Transfer & Full Access Hypothesis. It claims that L2 acquirers 

transfer their L1 internal grammars and restructure them via UG operations upon being 

confronted with L2 data that cannot be accounted for by the L1 configurations. Montrul 

et al. (2006), Tanner (2008), Rothman et al. (2010) and Bond et al. (2011) are some of 

the scholars who adopted this view.  

 

Partial Access Hypothesis which was developed in 1980‟s (White, 1986; Flynn, 1987) 

was modified in 1990‟s and named as Failed Functional Features Hypothesis (Tsimpli 

and Roussou, 1991; Smith and Tsimpli, 1995; Hawkins and Chan, 1997). This 

hypothesis proposes that grammars fossilize because functional features (like case or 

agreement) are only accessible during first language acquisition. Such features are not 

available for L2 acquirers. In the last decade, this hypothesis was modified again. Its 

newest version, the Interpretability Hypothesis, proposes that only uninterpretable 

features that are not instantiated in L1 are unavailable for L2 acquisition (Tsimpli et al., 

2003; Hawkins, 2005; Hawkins and Hattori, 2006; Tsimpli and Dimitrakopoulou, 2007; 

Tsimpli and Mastropavlou 2007). This hypothesis maintains that uninterpretable 

features are subject to critical period constraints and they are inaccessible to L2 learners. 

L1 parametric values associated with these features resist re-setting in L2 acquisition; 

on the other hand, interpretable features are accessible to the L2 learner, even if L2 

differs from the native language. 

 

The interpretable and uninterpretable features were introduced by Chomsky in his 

checking theory in the minimalist program. According to this theory, the interpretable 

features (animate- inanimate; human-non-human; singular-plural etc.) have an effect on 

the semantic interpretation of the sentence while the uninterpretable features (uwh, uD,  

uclausetype etc.) exist only for syntactic operations. They can be strong or weak 

according to the parametric differences among languages. These uninterpretable 

features and their interpretable counterparts must be checked against one another. The 



4 
 

checking procedure can be done in two ways. If the uninterpretable feature is strong, it 

forces its interpretable counterpart to move overtly to the specifier position of the 

projection where it exists. On the other hand, if it is weak, the checking procedure can 

be done via AGREE. That is, the uninterpretable feature searches for its interpretable 

counterpart by looking at its c-commanding domain. If it finds such a feature, it checks 

its features with this interpretable feature that stays in-situ, without forcing it to move to 

the specifier position of the projection it exists.  

 

Once the uninterpretable and the related interpretable features match, the 

uninterpretable features are deleted from the derivation. If there are any uninterpretable 

features that are not deleted from the derivation before the final spell-out, the derivation 

crashes, and becomes grammatically unacceptable. Therefore, all uninterpretable 

features should be deleted after having been checked against their interpretable 

counterparts.  

 

For instance, in the case of wh-movement in English, the uninterpretable, strong wh-

feature [uwh*] that exist in the matrix CP must be checked against its interpretable 

counterpart. The wh-word that carries the interpretable (wh) feature has to move to the 

matrix spec CP position since the uninterpretable (uwh) feature is strong in English. 

After the checking procedure, the [uwh*] feature is deleted from the derivation. In the 

following derivation, this checking process is presented: 

 

 

 

 

 

 

 

 

 

 

 



5 
 

(1) What did Sue claim that she saw? 

 

In this derivation, matrix C‟ contains a strong uninterpretable wh-feature, [uwh*], which 

must be checked against an interpretable wh-feature. This interpretable wh-feature 

exists in the wh-word, „what‟, which originates within the lower CP. Since the 

uninterpretable wh-feature is strong in English, this wh-word has to move to the spec 

CP position of the matrix clause. Once this movement is done cyclically, the 

interpretable and uninterpretable features are checked with each other and the 

uninterpretable wh-feature is deleted from the derivation.  

 



6 
 

According to Hawkins and Hattori (2006), the interpretable features should be available 

lifelong for all learners since they carry some semantic load. If they become unavailable 

after a critical period, some semantic information cannot be acquired by the second 

language acquirers, and this is implausible. Yet, the uninterpretable features do not have 

any effect on the semantic claims that only interpretable features are available for 

language learners lifelong.   

 

The debate among these three hypotheses is very much alive today. The most 

remarkable debate is between the followers of the Full Transfer & Full Access 

Hypothesis (Montrul et al., 2006; Tanner, 2008; Rothman et al., 2009; Rothman et al., 

2010, Bond et al., 2011) and the Interpretability Hypothesis (Tsimpli et al., 2003; 

Hawkins, 2005; Hawkins and Hattori, 2006; Tsimpli and Dimitrakopoulou, 2007; 

Tsimpli and Mastropavlou 2007).  

 

The scholars who favor either of the hypotheses carry out research on language pairs in 

which an uninterpretable feature exists in the target language but lacks in the source 

language. According to the findings of the scholars who favor the Interpretability 

Hypothesis, L2 acquirers cannot acquire the target uninterpretable feature (e.g. Tsimpli 

and Mastropavlou, 2007, p. 178; Hawkins and Hattori, 2006, p. 269). They perform 

significantly worse than native speakers on the tests which focus on the acquisition of 

this uninterpretable feature. On the other hand, the scholars who favor the Full Transfer 

/ Full Access Hypothesis present the cases in which such uninterpretable features are 

successfully acquired by L2 learners. Leaning on such findings, they claim that all 

features, interpretable or uninterpretable, are available for L2 acquisition.   

 

1.2. STUDIES ON THE INTERPRETABILITY HYPOTHESIS & FULL 

TRANSFER / FULL ACCESS HYPOTHESIS DEBATE 

 

Tsimpli et al. (2003) examined the L2 acquisition of Greek clitics and determiners by 

native speakers of Russian and Serbian. As they expressed, these languages lack a 

determiner system that distinguishes between a definite and indefinite article. 20 adult 

subjects took part in the study. The persistent problems in the use of definite articles and 



7 
 

clitics were claimed to show a long-lasting (possibly fossilized) effect of L1 properties 

due to the misanalysed features borne by these elements. They conclude that the definite 

article and the 3rd person clitic are D elements which are not intrinsically specified for 

an uninterpretable feature. This, by assumption, causes learnability problems to L2 

learners. 

 

In a similar study, Tsimpli and Mastropavlou (2007) investigated the acquisition of 

pronominal clitics and determiners in Greek by adult and child native speakers of 

Turkish and Russian. They emphasize the fact that  the definite article and the third-

person clitic in Greek possess uninterpretable features only, whereas the indefinite 

article and first-/second-person clitics carry an interpretable feature of [- definiteness] 

and [person] respectively. In the study, it is argued that the inaccessibility of 

uninterpretable features leads to a misanalysis of the third-person clitic and the definite 

article in advanced L2 grammars. The results of their study supported this hypothesis. 

Adult L2 learners showed poor performance in the use of the definite article compared 

to child L2 learners, which was argued to stem from the inaccessibility of 

uninterpretable features. As for the use of third-person clitics, both child and adult L2 

learners showed poor performance, which was interpreted by them as L1 influence, 

especially for the child L2 learners.  

 

In another study, Tsimpli and Dimitrakopoulou (2007) investigated the use of subject 

and object resumptive pronouns in L2 wh-interrogatives. Resumptive pronouns are 

linked to wh-movement, regulated by uninterpretable functional features. While 

resumptives are not allowed in single-clause sentences in Greek and English, Greek 

optionally allows resumptives in two- and three-clause wh-questions. Therefore, these 

scholars predicted that the learners would have problems in abandoning the resumptive 

strategy in L2 wh-interrogatives. They got the following results:  first, the abstract 

properties of subject–verb agreement in Greek were transferred to the learners‟ L2 

grammar, more strongly by the intermediate learners than by the advanced ones. While 

the advanced group showed an object advantage in rejecting resumptives, the 

intermediate learners were hovering at about 40% incorrect acceptance with both 

subjects and objects. Secondly and thirdly, animacy and d-linking may have aided the 



8 
 

learners in making some progress in rejecting the L1 resumptive strategy. However, 

learners never got close to native speaker performance. Therefore, Tsimpli and 

Dimitrakopoulou (2007) concluded that uninterpretable formal features, such as 

(subject, object) agreement, cause learnability problems even at advanced stages of 

acquisition.  

 

Similarly, Hawkins and Hattori (2006) had a study on the interpretation of English 

multiple wh-questions by Japanese speakers. They focused on the acquisition of the 

uninterpretable feature that forces wh-movement in interrogatives in English. Nineteen 

L1 speakers of Japanese (a wh-in-situ language that lacks the movement-forcing 

feature) who are highly proficient speakers of English were asked to interpret bi-clausal 

multiple wh-questions in English (like Where did the professor say the students studied 

when?). Their responses were compared with those of a native speaker control group. It 

is argued that the results are consistent with the unavailability of the uninterpretable 

feature. They concluded that there is a critical period for the selection of uninterpretable 

syntactic features for the construction of mental grammars. 

 

Kong (2005) investigated the acquisition of obligatory overt arguments in L2 English by 

adult L1 Chinese speakers. The results supported the view that older learners (who 

passed the critical period) cannot reset their parameter values. He states that the learners 

were more successful in disallowing null matrix subjects than null arguments in other 

positions because they made a small adjustment to the use of topic chains while the 

parameter settings of Chinese are maintained. The results of his study suggest that 

parameter values associated with functional categories are inaccessible to L2 learners 

after the critical period, which opposes to Full Access Hypothesis.  

 

Al-Thubaiti (2007) tested the “Interpretability Hypothesis” by examining the effect of 

age on the knowledge shown by proficient Saudi Arabic speakers of L2 English of two 

subtle linguistic properties associated with uninterpretable features: (i) the Gap Strategy 

in wh-interrogatives, and (ii) Reflexive Binding. While the former is differently 

instantiated in Arabic and English, the latter is similarly present in both languages. In 

the study, the advanced adult starters showed a persistent L1 effect (delearning problem) 



9 
 

in the acquisition of the gap strategy in wh-interrogatives, but hardly had any problems 

with reflexive binding. It was concluded that, fossilization is selective, and is a reflex of 

L1-L2 grammatical differences. For L2 adult starters, a persistent L1 effect is inevitable 

when the L1 property (resumptives) is associated with uninterpretable features encoded 

in the L1 lexicon, but not in the L2. On the other hand, where an L2 property (reflexive 

binding) is associated with uninterpretable features encoded similarly in L1 and L2, it 

can be successfully acquired by all age groups. Based on this evidence, he stated that a 

maturational account can be extracted from testing properties contrastively instantiated 

in L1 and L2. 

 

Beside these studies that support the existence of a critical age for the parameter 

resetting in L2, there are also some studies which claim the opposite. For instance, 

Montrul et al. (2006) examined the influence of maturation in adult bilingualism in the 

acquisition of Spanish accusative clitics and word order. They analyzed whether early 

bilinguals (heritage speakers) have more native like knowledge than late (post-puberty) 

bilinguals (L2 learners) of Spanish clitics and alternative word order due to the fact that 

they were exposed to naturalistic input early in childhood. The overall results of the 

study showed that early and late bilinguals performed alike. Even though the late 

bilinguals were more inaccurate than the early bilinguals at rejecting sentences in some 

conditions, the two bilingual groups showed the same pattern of responses as the 

monolingual control group; that is, they correctly accepted grammatical sentences and 

correctly rejected ungrammatical sentences with ungrammatical word orders. In short, 

all groups knew that clitics precede finite verbs and follow nonfinite verbs in Spanish. 

The results of their study support the view that maturation does not play a significant 

role in UG-SLA relationship. L2 learners whose first language lacks clitics (i.e., 

English speakers) are able to acquire new functional projections not instantiated in their 

language, which supports the full access account.  

 

In a case study, Tanner (2008) examined the role of critical age hypothesis in second 

language acquisition. He showed that DP concord in German is acquirable by English 

speakers who are acquiring German as a second language, although their native 

language lacks DP concord. The participants in the study showed success in matching 



10 
 

agreement features for elements within a DP, though he seems to have trouble acquiring 

inherent features of agreement-controlling nouns and properly spelling out case 

morphology, particularly dative. These facts are against the claim that uninterpretable 

features are unacquirable post-puberty. The results of this case study supported models 

of L2 acquisition which hold that uninterpretable features from the UG lexicon are 

available to post-puberty learners. 

 

In another study, Rothman et al. (2009) investigated adult acquisition of the Overt 

Pronoun Constraint (OPC) and inflected infinitives by L1 English learners of L2 

Portuguese. Their study challenges the Interpretability Hypothesis claimed by Tsimpli 

and colleagues. The data they obtained demonstrated that advanced learners of L2 

Portuguese acquired the OPC and the syntax and semantics of inflected infinitives with 

native-like accuracy. Since inflected infinitives require the acquisition of new 

uninterpretable φ-features, the data were claimed to provide evidence against Tsimpli 

and colleagues‟ Interpretability Hypothesis, supporting Full Transfer / Full Access 

Hypothesis.  

 

Judy et al. (2008) focused on the narrow syntax of the Spanish DP, and examined the 

acquisition of gender and number features of nouns, adjectives and determiners by L1 

English learners of adult L2 Spanish. These features determine DP internal word order 

(the adjective with respect to the noun), determiner-noun-adjective overt morphological 

accord and are both interpretable (on the head noun) and uninterpretable (on adjectives 

and determiners). After testing an intermediate and advanced group of adult English 

learners of L2 Spanish, they demonstrated across two linguistic tasks that L2 learners 

could acquire new features as evidenced by their knowledge of grammatical gender and 

the semantic construals of adjective placement in L2 Spanish. The results were against 

the approaches which claim that some features which are not instantiated in the L1 are 

no longer available to adult learners. 

 

Bond et al. (2011) used event-related potentials to investigate two factors in the second 

language (L2) processing of agreement: the role of number and gender features in the 

native language (L1), and the impact of individual differences between learners. The 



11 
 

study contributed to the growing body of literature investigating the role of the L1 in L2 

morpho-syntactic processing, providing evidence that adult learners even at low 

proficiency can exhibit development of native-like processing of (a) uninterpretable 

features that are not present in their L1, as well as (b) novel instantiations of features 

that are shared between the L1 and L2, supporting claims of full-access theories (e.g., 

Schwartz and Sprouse, 1994, 1996), and opposing to the Interpretability Hypothesis.  

 

In another study, Rothman et al. (2010) explored the adult acquisition of L2 nominal 

phi-features. Specifically, they analyzed the acquisition of the uninterpretable N feature 

that triggers noun raising in Spanish. This feature exists in Spanish whereas lacks in 

English, and it causes different adjective-noun orders for set-denoting and kind-

denoting interpretations for Spanish DP‟s. In the study, they focused on the syntactic 

and semantic reflexes in the related domain of adjective placement in two experimental 

groups: English-speaking intermediate (n = 21) and advanced (n = 24) learners of 

Spanish, as compared to a native-speaker control group  n = 15). Results of their study 

showed that, on some of the tasks, the intermediate L2 learners appear to have acquired 

the syntactic properties of the Spanish determiner phrase but, on other tasks, to show 

some delay with the semantic reflexes of pre-nominal and post-nominal adjectives. 

Crucially, however, their data demonstrated full convergence by all advanced learners 

and thus provided evidence opposite to  the predictions of representational deficit 

accounts (e.g., Hawkins and Chan, 1997 ; Hawkins and Franceschina, 2004 ; Hawkins 

& Hattori, 2006) Hence, the results of the study support the full access account.  

 

Mendez and Slabakova (2012) carried out a study as a response for Tsimpli and 

Dimitrakopoulou (2007) in which Greek native speakers‟ knowledge of gaps versus 

resumptive pronouns in English wh-movement was examined. Mendez and Slabakova 

(2012) focused on the resumptive pronouns in Spanish, which have similar 

characteristics to that of Greek. They divided Spanish native speakers into those who 

accept resumptives and those who do not; then they checked their acceptance of gaps 

and resumptives in English. Their results indicated that both groups of advanced 

learners, those that do and those that don‟t have resumptives in their individual 

grammars, have acquired the ungrammaticality of resumptives in English, opposing the 



12 
 

findings of Tsimpli and Dimitrakopoulou (2007). According to these results, they 

concluded that the uninterpretable features that are necessary for the use of the 

resumptive pronouns are available for L2 acquisition as well.  

  

As it is seen, the Interpretability Hypothesis and Full Transfer Full Access Hypothesis 

are hotly debated in the recent studies that examine parameter resetting in L2 

acquisition. 

 

One of the grammatical structures that attract the attention of the scholars who carry out 

researches on the availability of Universal Grammar in second language acquisition 

process is the island constraints on movements  (e.g. Johnson and Newport, 1991; White 

and Juffs, 1998; Li, 1998). Island structures are the ones that prohibit the movement of 

the elements to higher nodes in the derivation. The extraction of the items out of these 

structures causes ungrammaticality in the sentence. For instance, in the sentence below, 

the wh-island constraint is violated: 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 



13 
 

(2) Who did George claim when he saw? 

 

The movement of the wh-word „who‟ to matrix spec CP position is prohibited by the 

other wh-word „when‟ which occupies the spec position of the lower CP.  Such 

constraints were first introduced by Ross (1967) and extended since then. (See the 

Theoretical Concepts part for detailed information on wh-movement and the island 

constraint phenomena). 

 

Before the introduction of the Minimalist Program, the reason for the interest of UG-

SLA scholars on these structures was that the acquisition process of such constraints is 

the poverty of stimulus issue. If such constraints were acquirable for L2 learners, it 

meant that UG was accessible for them, if not, UG was inaccessible for L2 acquisition.  



14 
 

After the introduction of the Minimalist Program, such constraints maintain their 

importance on UG-SLA studies. In fact, the acquisition process of such locality 

constraints are still at the heart of such studies (e.g. Hawkins and Hattori, 2006). The 

reason for this interest is that such constraints are directly related with the existence or 

non-existence of an uninterpretable feature: [uwh*] in minimalist accounts, as well as 

being the poverty of stimulus issue. The languages that have overt wh-movement are 

claimed to have this uninterpretable feature, whereas the wh-in-situ languages are 

believed not to have this feature.  Hence, having a study on the L2 acquisition of a 

language that has this [uwh*] feature by speakers whose mother tongue lacks it, is a 

valid research area for the scholars who carry out researches on UG-SLA relationship, 

and who particularly favor the Interpretability Hypothesis or Full Transfer& Full Access 

Hypothesis.  

 

1.3. STATEMENT OF THE PROBLEM 

 

In the present study, it is hypothesized that uninterpretable features are available for L2 

acquisition along with the interpretable features. Against the claims of the 

Interpretability Hypothesis, the source for the incompleteness in L2 acquisition cannot 

be rooted from inaccessibility to certain uninterpretable features. As Rothman (2010) 

also argues, empirical studies have demonstrated that L2 grammars provide robust 

evidence of incidental acquisition of target L2 properties, even despite a poverty of the 

stimulus for their instantiation. Support for this claim comes from Schwartz and 

Sprouse (2000) who argued that L2 competence for such properties constitutes robust 

evidence that adult SLA is guided by the same inborn linguistic properties as child L1 

acquisition. 

 

On the other hand, the results for the studies which claim that certain uninterpretable 

features are unavailable for L2 acquisition cannot be ignored altogether. Leaning on the 

performance of L2 acquirers, such studies claim that the target uninterpretable features 

are unavailable for second language learners. What seems to be problematic in such 

studies is that the reasons for the incompleteness in L2 acquisition process should be 

sought for in other sources. As Rothman (2010) further claims, inaccessibility to UG 



15 
 

cannot be the source of these problems. He argues that contemporary full access 

approaches maintain that L2 incompleteness does not necessarily result from deficits 

within the syntax but instead may emerge from external learnability constraints and 

interface vulnerabilities (See the Theoretical Concepts part for detailed information on 

the hypotheses which focus on the incompleteness in L2 acquisition process). Leaning 

on the findings of Sorace (2005), Rothman (2010) exemplifies that the syntax-

morphology-phonology interface can be shown to be especially vulnerable in adult 

SLA. According to him, adult L2 acquirers, irrespective of their L1 and of the target L2, 

typically demonstrate target-deviant use of L2 functional morphology, differing from 

native adults and child L1 learners in their use of correct functional morphology (e.g., 

nominal agreement, verbal agreement, verbal tense, aspect, modal morphology) in 

discourse performance. Although incorrect L2 morphological production often 

decreases over time with an increase in proficiency levels, L2 morphological use rarely 

indistinguishably matches that of native speakers, even at the highest of L2 proficiency 

levels. Hence, he favors the Full Transfer / Full Access Hypothesis, which claims that 

interpretable or uninterpretable, all features are available for L2 acquisition. The present 

study supports this view, as well.   

 

All in all, it is apparent that there is a need for further studies in this debate. In such 

studies, different language pairs should be extensively analyzed to reach at more 

concrete results. None of the recent studies that focus on Interpretability Hypothesis- 

Full Transfer, Full Access Hypothesis debate focus on Turkish-English language pairs. 

The acquisition process of the uninterpretable wh-feature in English by Turkish L2 

learners of English might provide fruitful results for this debate.  

 

1.4. THE PURPOSE OF THE STUDY  

 

The main aim of the study is to analyze the island constraints that put restrictions on 

wh-movements in languages to examine the „Interpretability Hypothesis‟ which was put 

forward by Tsimpli & Dimitrakopoulou (2007), and supported or opposed by many 

other linguists. Turkish is a wh-in situ language and such languages are claimed to lack 

the uninterpretable [uwh*] feature (e.g. Hawkins, 2005, p.129). Therefore, the island 



16 
 

effects show unstable characteristics in Turkish. On the other hand, this uninterpretable 

feature exists in English and this language strictly obeys the island effects. Hence, 

having a study on the acquisition of the island constraints in English by Turkish learners 

may provide valuable data to support or refute “the Interpretability Hypothesis” 

mentioned above. The main purpose of this study is to provide data from Turkish L2 

acquirers of English to examine the validity of the “Interpretability Hypothesis” in UG-

SLA relationship. If the adult Turkish L2 acquirers can be as successful as the 

monolingual English speakers, the data obtained can be viewed as evidence that refutes 

the “Interpretability Hypothesis”. On the other hand, if these participants become 

significantly less successful compared to the other group, the results can be considered 

as a support for this hypothesis.  

 

The study also aims to analyze the possible role of the positive evidence in a naturalistic 

learning environment. More specifically, the study aims to investigate if there are any 

differences between the performances of Turkish speakers who are exposed to positive 

evidence by living in a country where English is spoken as a mother tongue, with the 

ones who acquired this language completely in their home country. It is highly possible 

that being exposed to positive evidence in a naturalistic learning environment in L2 

acquisition may influence the success of the language users. Some of the studies that 

favor the Interpretability Hypothesis and focus on UG-SLA relationship are carried out 

on the L2 learners who are exposed to positive evidence in the target language by living 

in an environment where this language is spoken as a mother tongue for a long time 

(e.g. Hawkins and Hattori, 2006; Tsimpli and Mastropavlou 2007); yet some others (e.g. 

Tsimpli and Dimitrakopoulou, 2007; Kong, 2005; Al-Thubaiti, 2007) were carried out 

just on the ones who acquired it in their home country. In this respect, being (not) 

exposed to positive evidence in a naturalistic learning environment in L2 acquisition 

might have played some role in the results obtained in such studies. The participants of 

Tsimpli and Dimitrakopoulou (2007) were native speakers of Greek acquiring English 

as a foreign language. They were students at Aristotle University in Thessaloniki.  The 

participants of Kong (2005) were 75 Chinese speakers learning L2 English in China. Al-

Thubaiti had a similar case as well; only one of his adult participants had stayed in an 

English-speaking country. Carrying out studies on the participants who had never stayed 



17 
 

in an English-speaking country might have influenced the results they obtained in their 

studies.  

 

Another aim of the study is to detect the possible role of the mother tongue influence on 

L2 acquirers. As it has already been stated, English obeys the island constraints on wh-

movement but these constraints show unstable characteristics in Turkish. While it is 

possible to interpret wh-arguments out of the islands in Turkish, it is not possible to do 

the same for wh-adjuncts. Hence, it is possible to conclude that while Turkish and 

English behave similarly on some structures, they differ in some others. However, such 

similarities and differences originate from different sources. While the requirement for 

the overt wh-movement to Spec CP to satisfy the [uwh*] feature and the violation of 

island structures in this overt movement cause the ungrammatical structures in English, 

the government and binding relationships between the [Qu]-operator and wh-words in-

situ cause the (un)grammaticality in Turkish. (See the Theoretical Concepts part for 

detailed information on the wh-movement phenomena in English and Turkish). Such 

patterns may play a role in the acquisition of the island constraints.  Hence, the study 

aims to analyze the effects of the mother tongue parameter values on L2 acquirers.   

 

The study also aims to investigate the possible avoiding strategies which can be applied 

by both native and L2 learners of English in order not to violate island structures. After 

the analyses of the results of the production tasks like wh-question formation task and 

translation task, it might be possible to determine the strategies that the participants 

make use of in order to escape island violations. In these tests, the participants are 

directed to violate target island structures by means of the context provided for them. 

The participants have to develop some strategies to avoid violating the island structures, 

and one of the aims of the present study is to detect these strategies (See the 

methodology part of the study for detailed information on the tests of the study).  

 

The final aim of the study is to find out which island structures appear to be more 

problematic for L2 acquirers; that is to say, the study aims to find an answer for the 

question that among the four island constraints (Wh-island Constraint, Complex NP 

Constraint, Sentential Subject Constraint and Adjunct Island Constraint) which one of 



18 
 

them appear to be more difficult in L2 acquisition process. (See the methodology part of 

the study for detailed information on the island constraints that are focused on the study 

 

1.5. THE RESEARCH QUESTIONS  

 

The study aims to find answers for the following research questions: 

 

1- Does the performance of late L2 acquirers of English (whose L1 is Turkish) on the 

island constraints in wh-movement support “the Interpretability Hypothesis” put 

forward by Tsimpli and Dimitrakopoulou (2007)? In other words, when the performance 

of these L2 acquirers is compared with that of native speakers of English, is there a 

significant difference in the results as asserted by “the Interpretability Hypothesis”? 

 

2- Does being exposed to positive evidence in a naturalistic learning environment in L2 

acquisition process influence the success of the L2 acquirers in their performance on the 

island constraints on wh-movement? 

 

 3- When the structural differences between L1 and L2 on the target island constraints 

are taken into account, do such patterns play a role on the performance of L2 acquirers? 

In other words, do the mother tongue parameter values have an influence on the L2 

acquirers in their second language acquisition process? 

 

4- What avoidance strategies, if any, do the participants make use of in order not to 

violate island structures? 

 

5- Among the 4 island structures investigated in the study, on which structures the L2 

acquirers perform better than the others? 

 

1.6. SIGNIFICANCE OF THE STUDY 

 

Some of the studies that favor the Interpretability Hypothesis  which focus on UG-SLA 

relationship are carried out on the L2 learners who are exposed to natural input in the 

target language by living in an environment for a long time where this language is 



19 
 

spoken as a mother tongue (e.g. Hawkins and Hattori, 2006; Tsimpli and Mastropavlou, 

2007). Yet, some others (e.g. Tsimpli and Dimitrakopoulou, 2007; Kong, 2005; Al-

Thubaiti, 2007) were carried out just on the ones who acquired it in their home country.  

In this respect, being (not) exposed to natural input in L2 acquisition might have played 

some role in the results obtained in such studies.  That is to say, it is highly possible that 

exposure to positive evidence in a naturalistic learning environment might be playing 

some important role in determining the availability of universal grammar in second 

language acquisition process. 

 

In the review of the literature, it was observed that none of the UG-SLA studies, which 

stand for or against the Interpretability Hypothesis, take into account the possible role of 

the positive evidence that is received in a naturalistic learning environment. That is to 

say, none of such studies contained two L2 learner groups: one acquiring it in home 

country alone, the other acquiring it in an environment where this language is spoken as 

a mother tongue. If the data of these studies had been collected from these two L2 

learner groups, it would have been possible for us to determine the possible role of the 

exposure to positive evidence in a naturalistic learning environment. 

 

In this respect, the present study is the first to take into account the possible role of 

positive evidence that is received in a naturalistic learning environment in second 

language acquisition process. It is hypothesized that the exposure to natural input in an 

environment where the target language is spoken as a mother tongue has an important 

role in determining the availability of UG in L2. To that end, the data of the present 

study have been collected both from Turkish L2 learners of English who live in their 

home country alone, and who live in a country where this language is spoken as a 

mother tongue. The comparison between the performances of these learner groups is 

presumed to display the role of the exposure to positive evidence in a naturalistic 

learning environment during L2 acquisition process.  

 

 

 

 



20 
 

CHAPTER 2 

THEORETICAL CONCEPTS 

 

In this chapter, elaborate information is presented on universal grammar-second 

language acquisition studies. In addition, wh-movement and the island constraint 

phenomena, which constitute the target structures of the study, are explained thoroughly.  

 

2.1. UG AND L1 ACQUISITION  

 

The logical problem of language acquisition has led to proposals that certain aspects of 

knowledge must be innately present in the first language learner in the form of 

Universal Grammar. As White (2003) points out, certain properties of language are too 

abstract, subtle and complex to be acquired without assuming some innate and 

specifically linguistic constraints on grammars and grammar acquisition (p. 3). This 

innateness hypothesis constitutes one of the fundamental aspects of Generative 

Grammar. In this framework, Universal Grammar is seen as a part of an innate, 

biologically endowed language faculty. It places limitations on grammars, constraining 

their form as well as how they operate. It includes invariant principles, as well as 

parameters. As Radford argues, “Since universal principles of grammatical structure do 

not have to be learnt, the child‟s structural learning task is limited to that of parameter 

setting” (1997, p. 21). 

 

2.2. UG AND L2 ACQUISITION  

 

After the introduction of the innateness hypothesis in first language acquisition, another 

debate emerged: If language acquisition is innate and controlled by Universal Grammar, 

then what is the function of UG in second language acquisition? As Flynn (1989) 

suggests in her article;  

as is well known UG as a theory of acquisition characterizes L1 learning but 

does not make explicit predictions about L2 acquisition. However, if 

principles of UG do not in fact characterize a language faculty that is 

biologically determined and that is necessary for the acquisition of an L1, 



21 
 

then it seems quite reasonable to assume that principles of UG also play a 

role in L2 acquisition (p. 92). 

 

Naturally, while investigating the accessibility of UG in second language acquisition, all 

aspects of L2 acquisition cannot be dealt with at once. A specific aspect of it should be 

investigated in order to be able to get valuable data. Linguists usually prefer to 

investigate the acquisition of a parameter or a principle to assess the accessibility of UG 

in L2 acquisition. Initially, three different hypotheses were developed by the linguists 

who examined the role of UG in Second Language Acquisition in the first years of such 

studies: 

 

2.2.1. Full Access Hypothesis 

 

UG is fully available; differences in patterns of acquisition between L1 and L2 learners 

and the lack of completeness can be accounted for in other ways. Schwartz and Sprouse 

(1996) are pioneers of this hypothesis. This theory maintains that the grammatical 

properties of a speaker‟s L1 constitute the initial state in L2 acquisition. Thereafter L2 

speakers restructure their grammars on the basis of evidence from the target language 

which is not „„parsable‟‟ by the initial-state grammar.  

 

2.2.2. No Access Hypothesis 

 

Proponents of this position (e.g. Clashen and Muysken, 1986; 1989; Clashen, 1988; 

Bley-Vroman, 1989) assume that L1 and L2 acquisition proceed along the lines of 

fundamentally different cognitive processes. UG is totally inaccessible to the adult L2 

learner; learning takes place in terms of non-linguistic learning strategies like 

distributional analysis, analogy, and hypothesis formation and testing. Freeman and 

Long, who are also proponents of this hypothesis, put forward that “Certain researchers 

argue against access because of incompleteness, because the innate capacity for 

language learning declines with age” (1987, p. 116). 

 

 

 



22 
 

2.2.3. Partial Access Hypothesis 

 

UG is partially available to the learner (Tsimpli and Roussou, 1991; Smith and Tsimpli, 

1995). As Waber and Czendik (2002) state, in this hypothesis, only those parametric 

values characterizing the L1 grammar are available, the rest must be learnt in terms of 

non-linguistic learning strategies (p. 1). As explained in Kong (2005, p. 230), there are 

three versions of this view: (a) Principles of UG are still available and constrain 

grammar building, but learners cannot reset parameters: that is to say, UG is only 

accessible through L1; (b) Adult L2 learners can in principle access principles and 

parameters of UG but are unable to access some of the features of functional categories; 

(c) A specific class of functional features, in particular, strength features, become 

inaccessible.  

 

These hypotheses were in clear opposition to one another, incorporating sharp edges and 

making straightforward and precise predictions about the effects to follow. This 

rendered them easily testable and made them perfect subject for empirical research. 

However, the studies carried out on this field showed that the matters were more 

complex than it had seemed at first. As a consequence of that, all of the initial, 

straightforward hypotheses had to be redefined.  

 

As Kaltenbacker (2001) points out, the basic problem of the original hypotheses was 

that they addressed the role of UG in L2 acquisition as one single topic, although the 

issue consists of two clearly distinct questions, which are: 

 

1- What is the quality of the initial state? Is it constrained by UG, does it violate UG, or 

is it simply the L1 final state? 

 

2- Which role does UG play in the process from the L2 initial state to the L2 final state? 

Is UG operative and are the learning strategies supplied by the LAD still active? Is UG 

non-operative, or does UG just provide passive knowledge that is accessible via the 

learners‟ L1?    

 



23 
 

White (2000) drafts five different approaches that address both issues independently. 

These five approaches can be described briefly as follows (pp. 134-139): 

 

2.2.4. Full Transfer / Partial Access 

 

Under this view, the L2 initial state consists of the L1 final state. Universal properties 

that are not instantiated in the L1, are lost in L2 acquisition. Parameter resetting is not 

possible under this approach. Knowledge of UG is only retrievable as far as it is 

embodied in properties of the L1. 

 

2.2.5. No Transfer/ Full Access 

 

This hypothesis assumes that the L2 grammar is acquired on the basis of UG principles 

and parameters interacting freely with L2 input. The L2 learner‟s initial state parallels 

that of the L1 acquirer, and L1 and L2 developing grammars are also expected to be 

similar. All universal principles and parameter settings as well as functional categories 

and feature values are available to learner. Features of UG that are not incorporated in 

the L1 are still fully accessible to the L2 learner at any age.  

 

2.2.6. Full Transfer / Full Access 

 

This hypothesis was developed by Schwartz and Sprouse (1996).  It states that L1 and 

L2 differ with respect to their starting point, but are similar with respect to involvement 

of UG. Learners are assumed to transfer their L1 internal grammars and restructure them 

via UG operations upon being confronted with L2 data that cannot be accounted for by 

the L1 configurations. Again all UG options that are available to an L1 learner are also 

accessible by an L2 learner. This approach predicts that parameter resetting will take 

place once the input yields data that cannot be accounted for by the L1 internal 

grammar.  

 

 

 



24 
 

2.2.7. Partial Transfer / Full Access 

 

Proponents of this approach hold the view that the L2 initial state draws on properties of 

both the L1 and UG concurrently. There are however, different claims about the actual 

proportion of the L1 that is transferred. „Minimal Trees Hypothesis‟ which was 

proposed by Vainikka and Young-Scholten (1994; 1998) suggests that only lexical 

categories are transferred into the L2, while the functional categories are not. In the 

initial state, L2 learners are assumed to project only NP and VP, but not DP, IP or CP. 

According to this hypothesis, L2 learners gradually project functional categories just as 

L1 acquirers are assumed to do. In other words, in this view it is believed that the 

emergence of functional categories is not dependent on the L1, and hence there is no 

transfer; rather they emerge in response to L2 input. An alternative suggestion (Eubank 

1994; 1996) claims that both lexical and functional categories established in the L1 are 

transferred into the L2, but feature values, e.g. (+/- strong Agr), are not. This view is 

named as “Valueless Features Hypothesis”. Unlike the Minimal Trees Hypothesis, in 

this hypothesis it is asserted that both functional and lexical categories are available 

from the L1, but the strength of these features is not available. Acquisition involves 

acquiring the appropriate feature strength of the L2. 

 

2.2.8. Partial Transfer / Partial Access 

 

This approach suggests that only some universal properties of the L1 are transferred, 

and some of the UG components that are not exhibited in early L2 interlanguage do not 

emerge. As stated above, partial Transfer/Full Access hypothesis claims that appropriate 

feature strengths (strong or weak) cannot be transferred from L1, but they can be 

acquired during the process of L2 acquisition. On the other hand, Partial Transfer/Partial 

Access Hypothesis asserts that certain functional features are never specified for 

strong/weak values during the course of L2 acquisition.   

 

Along with these five hypothesis that claim that UG is accessible in L2 acquisition 

(partially or fully), the Fundamental Difference Hypothesis, which puts forward that 

adult second language learners do not have access to UG at all (Bley-Vroman, 1989; 



25 
 

1990; Clahsen and Muysken, 1986; 1989; Clahsen and Hong, 1995), is also alive today. 

Han (2004) and Long (2007) are some of the proponents of this view.  

 

It is still possible to find supporters of all these hypotheses. However, in recent studies, 

two hypotheses came to fore and they are hotly debated. The first one is the 

Interpretability Hypothesis which is a revised version of the partial access account. This 

hypothesis puts forward that uninterpretable features are subject to critical age and they 

are inaccessible for L2 learners (Tsimpli and Dimitrakopoulou, 2007).  The other one is 

a relatively older one: Full Transfer, Full Access Hypothesis (e.g. Schwartz and 

Sprouse, 1996; White, 2003) which claims that all features (both interpretable and 

uninterpretable) are available for L2 learners and the cause for the incompleteness in L2 

acquisition should be explained in other ways.   

 

2.3. THE INTERPRETABILITY HYPOTHESIS & FULL TRANSFER FULL 

ACCESS HYPOTHESIS 

 

Tsimpli et al. (2003), Hawkins (2005), Hawkins and Hattori (2006), Tsimpli and 

Dimitrakopoulou (2007) and Tsimpli and Mastropavlou (2007) put forward “The 

Interpretability Hypothesis”, which adopts assumptions regarding the critical period 

hypothesis for language acquisition. In particular, this hypothesis maintains that 

uninterpretable features are subject to critical period constraints and, as such, they are 

inaccessible to L2 learners. That is to say, L1 parametric values associated with these 

features resist re-setting in L2 acquisition; on the other hand, interpretable features are 

accessible to the L2 learner, even if L2 differs from the native language. In other words, 

the claim is that interpretable features are accessible to the L2 learner whereas 

uninterpretable features are difficult to identify and analyze in the L2 input due to 

persistent, maturationally-based, L1 effects on adult L2 grammars.   

 

In fact, this hypothesis is not new. It is the updated version of the Failed Functional 

Features Hypothesis (FFFH hereafter) which was favored by Tsimpli and Roussou 

(1991), Smith and Tsimpli (1995), Hawkins and Chan, (1997) and Franceschina (2001; 

2005). FFFH proposes that grammars fossilize because functional features are only 

accessible during first language acquisition. Interpretability Hypothesis is the re-



26 
 

formulation of this hypothesis; yet it makes claims only on uninterpretable features. As 

a matter of fact, both of them are the revised versions of the partial access hypothesis, 

which claims that UG is only partially available for L2 learners. In the literature, these 

hypotheses are also named as the Representational Deficit Approach, which attributes 

L2 inflectional variability or error to a failure in the selection of parameterized formal 

features.  

 

 In contrast with the Interpretability Hypothesis and its older versions, Full Transfer Full 

Access Hypothesis claims that UG is fully accessible for the L2 learners. The 

researchers who favor this hypothesis (e.g. Montrul et al., 2006; Tanner, 2008; Rothman 

et al., 2009; Rothman et al., 2010; Bond et al., 2011) examine the acquisition of 

uninterpretable features that exist in the target L2 but lacks in L1 in order to get 

empirical evidence to refute the Interpretability Hypothesis. In fact, this approach does 

not ignore the fact that L2 acquisition cannot reach at ultimate attainment in general. It 

simply claims that this incompleteness does not stem from inaccessibility to UG. The 

reason for this incompleteness is usually sought for in different sources. Some of such 

hypotheses are summarized as follows:  

 

2.3.1. Interface Hypothesis  

 

 According to this hypothesis, the inability to reach at ultimate attainment in L2, that is 

to say, inability to become as proficient as a native language user in the second 

language, stems from the interaction between syntax and other grammar internal 

modules (syntax-semantics, syntax-phonology interfaces) or the grammar external 

modules (syntax-pragmatics, syntax-discourse). That is to say, the integration of 

syntactic knowledge with other types of information, especially with grammar external 

ones, is believed to be more problematic for L2 learners than properties that require 

only syntactic knowledge (Sorace, 2000; 2003; 2004; 2005; Valenzuela, 2006).  

 

2.3.2. The Missing Surface Inflection Hypothesis 

 

 According to this hypothesis, the lack of (or the variable use of) morphological forms 

in interlanguage grammars reflects a problem with the realization of surface 



27 
 

morphology, rather than an impairment in the domain of functional projections or 

feature strength. Learners‟ imperfect mapping of specific morphological forms to 

abstract categories may be the source of inability for ultimate attainment in L2. In cases 

like processing difficulties or communication pressures, learners may resort to defaults, 

forms that are underspecified in some features. However, it is argued that L2 learners 

have acquired the relevant features of the terminal nodes in the syntax and it is only the 

imperfect lexical access to the whole set of morpho-syntactic features that brings 

forward L2 learners‟ variability in production (Haznedar and Swartz, 1997; Prevost and 

White, 2000; Haznedar, 2003). 

 

2.3.3. Feature Assembly Hypothesis  

 

This hypothesis views incompleteness in L2 acquisition as a mapping problem as well. 

Lardiere (2008) looks for the sources of morphological variability in the complexity of 

the morpho-syntactic knowledge that has to be acquired by L2 learners. Her hypothesis 

maintains that the ways in which grammatical features are morphologically combined 

and conditioned present formidable learning problems in L2 acquisition. The prediction 

of this account is that the more re-assembly of features the L2 learner must do, the more 

difficulty she will face, and such morphemes (and meanings) will take longer to acquire 

(Lardiere, 2008).  

 

2.3.4. Morphological Underspecification Hypothesis 

 

According to this hypothesis incompleteness in L2 acquisition stems from 

representational deficits; not from deficits in syntactic but in morphological 

representation. Therefore, L2 morphology errors in this account involve the systematic 

substitution of underspecified, representationally-simpler forms across comprehension 

and production (McCarthy, 2007; 2008). 

 

Although these four hypotheses (and many others) do not openly express their support 

for the Full Access Full Transfer Hypothesis, they serve such a function since they all 



28 
 

put forward that there is no representational deficit for syntax in L2 acquisition, and UG 

is fully accessible for L2 acquirers.  

 

2.5. WH- PHRASES AND WH-MOVEMENT 

 

Since the present study focuses on island constraints on wh-movement to check the 

validity of the Interpretability Hypothesis, it is necessary to demonstrate the wh-phrase 

structures in the languages that the study focuses on. In general, languages differ with 

respect to the strategies they employ for wh-question formation. While in some 

languages wh-expressions are forced to move to matrix spec CP positions to form wh-

questions, in some others, such obligatory movements to sentence initial positions are 

not observed. The wh-expressions in these languages can stay in their base positions in 

interrogative sentences. In that respect, languages have been mainly classified into wh-

movement languages and wh-in-situ languages.  

 

2.5.1. Wh-Movement in English 

 

English has overt syntactic wh-movement (Watanabe, 2003, p. 203).  As Adger (2003) 

expresses, languages like English that have overt wh-movement possess a [uwh*] 

feature in their CP‟s, which forces the wh-expressions to move to the specifier position 

of this phrase. This uninterpretable feature makes it impossible for the wh-expression to 

stay in-situ if the [uclausetype] feature in C is [Q], that is to say, if it is an interrogative 

sentence. If there are more than one wh-element in the clause, the [uwh*] feature 

attracts only one of them, and it is the closest wh-item according to the “Attract Closest 

Principle”. He further explains how echo questions (e.g. Medea saw who?) are excluded 

from this rule by stating that they are not questions in the usual sense of the word, and 

their C‟s do not bear a [Q] feature. Since their CP‟s do not bear a [Q] feature, there is no 

[uwh*] feature in this phrase, either. Hence, wh-expressions in echo questions stay in-

situ and do not move to the specifier of the CP. As he further points out, in complex 

sentences, where the matrix clause is interrogative and the wh-elements are located in 

the embedded clause, the wh-words must move to the specifier position of the matrix 

CP cyclically, moving through the CP of the embedded clause (341-370).   



29 
 

Hawkins and Hattori (2006) make a similar comment on this issue. They state that wh-

word or phrase must be fronted in non-echo interrogative clauses in English (p. 274). 

Following Adger‟s statements, they summarize the wh-movement phenomenon in 

languages as; 

in the case of a wh-question, there are two ways for feature checking. In 

wh-in-situ languages, valuing occurs directly between [C, Q] and the target 

wh-word/phrase in-situ, without any movement. In languages where wh-

words/phrases are fronted, there is an additional requirement that valuing 

of [uwh:] occur within the immediate projection of interrogative C. This 

requirement is represented as an asterisked [uwh*:] feature. This forces a 

wh-word/ phrase to move to the specifier of [C, Q, uwh*:] to value 

[uwh*:] (pp. 275-276). 

 

In a similar fashion, Haegeman (1991) argues that wh-movement in English is not done 

by adjunction, but by substitution; wh-elements must move to spec CP (p. 355). She 

further claims that the landing site for wh-movement in English must be a position 

which is not specified for the phrasal category. Spec CP is a suitable position for this 

respect. A non-filled spec CP can receive phrasal constituents of any syntactic category 

(1991, p. 348). Hence, unfilled specifier positions of CP‟s are suitable landing sites for 

wh-movement in languages like English. 

 

2.5.2. Wh-movement in Turkish  

 

In Turkish, wh-phrases are said to remain in-situ both in main and embedded clauses 

(Akar, 1990; Özsoy, 1996; Uzun, 2000; Kornfilt, 2003; 2008; ĠĢsever, 2009; Çele and 

Gürel, 2011). Wh-in-situ languages lack the uninterpretable [uwh*] feature, and do not 

have overt wh-movement (Adger, 2003). That is to say, the wh-words do not have to 

move to sentence initial position to form wh-questions. Although Özsoy (2009) claims 

that some specific scrambling structures should be regarded as overt wh-movement in 

Turkish, this claim does not change the main characteristics of Turkish language: it is a 

wh-in-situ language. As Özsoy (1996) states, Turkish does not possess a syntactic rule 

of wh-movement, i.e. the wh-phrase appears in situ in the surface structure in a Turkish 

wh-question. Wh-phrases like Ne (What), Kime (Whom), and Ne zaman (When) 

respectively occur in the positions in which their NP-counterparts would be found in a 



30 
 

regular Turkish sentence. They do not have to move to S-initial position as their English 

counterparts have (p. 3).  

 

Huang (1982) developed a theory to explain the wh-formation strategy in wh-in-situ 

languages. According to this theory, the wh-phrases that do not move in S-structure 

move at Logical Form (LF hereafter). Thus, at LF, the wh-phrases in situ should have 

the same configurations as the syntactically moved wh-phrases (as cited in Tanaka, 

1999, pp. 371-372). She further states that the difference between the wh-movement 

found in Turkish and a language like English is that while the rule applies at the 

syntactic level in the latter, it applies at LF in the former (p. 4). This LF movement 

hypothesis is supported and used by several scholars and it is still frequently 

encountered in the literature. 

 

In fact, there are alternative views to the LF-movement approach in wh-in-situ 

languages. For instance, Aoun and Li (1993) put forward that wh-movement does not 

take place as a covert movement in LF in wh-in-situ languages like Chinese. The wh-

words are claimed to be co-indexed and interpreted by means of a Question Operator 

(Qu-Operator) in overt syntax. Their claims are supported and adopted to Turkish case 

by Arslan (1999) and Görgülü (2006). According to these scholars, wh-movement in 

Turkish does not take place in LF, but in overt syntax through a binding relationship 

between the [Qu]-operator and the wh-words that remain in-situ. Following, Aoun and 

Li (1993), they propose that although the wh-word does not move, the Null operator, 

with which it is co-indexed, may move from its base position in the clause containing 

the wh-word to the matrix spec CP position. The movement of the Qu-operator is 

subject to ECP, which accounts for the asymmetry in the extraction of wh-arguments 

and wh-adjuncts (Arslan, 1999, p. 23). That is to say, since wh-arguments are lexically 

governed, the Qu-operator can be base generated within the matrix spec CP position. On 

the other hand, since wh-adjuncts are not lexically governed, they are in need of 

antecedent government. Therefore, the Qu-operator is generated in the same position 

where the wh-word is generated, and then, it moves to the matrix spec CP position. This 

movement is subject to ECP effects, which accounts for the adjunct argument 

asymmetry in Turkish.   



31 
 

 

On the other hand, ĠĢsever (2009) claims that wh-movement takes place in overt-syntax 

in Turkish through feature movement. He takes into account the close interaction 

between wh-in-situ and focus in Turkish, and argues that the syntax of Turkish wh-in-

situ includes a focus-driven movement of a focus-accented wh-phrase operator to the 

lower Spec FocP, whereby it is attracted to Spec CP to satisfy the [uwh*] feature of C. 

In other words, he suggests that Turkish wh-phrases in-situ have a null operator in their 

spec positions, which undergoes movement in overt syntax. When the wh-words are 

focused in Turkish, the operator first moves to the Spec position of the FocP, then to the 

spec position of the CP to satisfy the [uwh*] feature of C.  

 

 As a matter of fact, the Feature Movement Hypothesis proposed by ĠĢsever (2009) 

appears to cause a great problem for the basic assumptions of the present study. As he 

asserts, Turkish has a [uwh*] feature in its matrix CP‟s just like the overt wh-movement 

languages like English.  If such a feature exists in Turkish, then it becomes impossible 

to assess the validity of the Interpretability Hypothesis in English-Turkish pair, since 

both languages contain the same uninterpretable feature. However, having such a 

conclusion out of this proposal might be misleading. It is certain that there is a 

difference between these two languages: one of them has compulsory wh-movement, 

but the other one does not. Hence, if syntax will continue to explain the motivation of 

the movements through existence or non-existence of interpretable-uninterpretable 

features, then it is for sure that English has an uninterpretable feature which Turkish 

lacks. It might not be the [uwh*] feature which has been uttered so far, but an additional 

one, like [uwhx*] feature which can only be satisfied through the movement of lexical 

items, rather than just the features. That is to say, what ĠĢsever (2009) proposes is quite 

reasonable and offers new insights into wh-in-situ phenomena in Turkish, and it is not 

incompatible with the basic assumption of the present study: while there is a strong 

uninterpretable feature in English which causes the compulsory movement of the wh-

words to the spec CP positions, such a feature does not exist in Turkish and wh-words 

can stay in-situ in this language. Besides, it should also be taken into account that there 

is a close relationship between (not) having compulsory wh-movement and island 

effects. Island constraints are strictly obeyed in wh-movement languages, which is not 



32 
 

the case for the wh-in-situ languages. Therefore, it can be deduced that the movement of 

the lexical items which carry some phonological load with themselves out of island 

structures cause problems, and this is not the case for the feature movements. In the 

present study, then, what is being examined is the relationship between the 

(non)existence of [uwhx*] feature and island constraints, and this small change does not 

cause any problem for the general assumptions of the study. For the unity of the study, 

though, this feature will continue to be uttered as [uwh*] feature in the following 

chapters.  

 

In turn to other aspects of the wh-in-situ phenomena in Turkish, there are novel 

proposals produced by Görgülü (2006) and Arslan (1999), as well. As Görgülü (2006) 

asserts, wh-words in Turkish do not have inherent interrogative force by themselves but 

their interpretation is dependent on other elements in the structure. For instance, when 

“ki” particle exists in the CP, the wh-word receives only the non-interrogative reading. 

According to Görgülü, what this indicates is that the “ki” particle binds the wh-word 

yielding only the non-interrogative reading. He provides following example for this 

case: 

 

(3) Cem kim-i sev-di ki! 

     Cem-NOM who-ACC love-PAST particle 

     (i) “Cem loved no one”. 

     (ii) * “Who did Cem love?” 

 

In the absence of such a particle in the CP, the sentence above receives only the 

interrogative reading, which indicates that the wh-word in-situ is bound by a null [Qu]-

operator in CP. 

 

(4) Cem kim-i sevdi? 

     Cem-NOM who-ACC love-PAST? 

     (i) *“Cem loved no one” 

     (ii) “Who did Cem love?” 

 



33 
 

According to Görgülü (2006), in the absence of such CP level operators, it is claimed 

that the (non)existence of TP level operators like “Gen-operator” and “negation” 

determine the interpretation of the wh-words.  

 

The [Qu]-operator hypothesis also provides a plausible explanation for the argument-

adjunct asymmetry that exists on the island constraints on wh-movement, which form 

the focus of this study. As Arslan (1999) puts forward, although Özsoy (1996) 

successfully explains the argument-adjunct asymmetry in the Complex NP Constraint 

and Adjunct Island Constraint via LF movement analysis, this solution does not provide 

sufficient explanation for all structures. She further points out that the structures which 

fail to receive an explanation include examples that involve adjunct wh-phrases within a 

sentential subject; adjunct wh-phrases within postpositional phrases; and structures in 

which the wh-element co-occurs with the operator yalnızca 'only' (p .53).  

 

2.6. THE ISLAND CONSTRAINTS ON WH-MOVEMENT 

 

Munn (2007) defines island constraints as;  

there are a number of constraints on movement. These constraints have been 

traditionally called island constraints based on the metaphor that a syntactic 

island is a phrase of which elements cannot get off, just as a person cannot 

get off of an island without extra help of a bridge or a boat (p. 2). 

 

 Although there are also some island constraints which put restrictions on NP 

movements, majority of such island constraints are viewed on wh-movements, and 

especially on languages like English which have overt syntactic wh-movement.  

 

The earliest attempt to lay out a general principle to explain the constraints on 

movement comes from Chomsky (1964) where the A-over-A Condition is introduced 

(as cited in Lasnik, 2010, p.2; Hoffmeister and Sag, 2009, p.3). In this constraint, an 

element of category A cannot be extracted out of a phrase of category A. As Hoffmeister 

and Sag (2009, p.04) exemplify, this principle rules out all sentences where an NP has 

been extracted from an NP, as in the sentences below: 

 

(5) *What did he know [NP someone who has ___ ]? 



34 
 

(6) *What did you see the man read [NP the book that was on ]?   

 

However, Ross (1967) claimed that A-over-A constraint fails to provide correct 

explanations for all structures in which an element A is extracted from a category A. 

Besides, there are other structures in which movement of the elements are prohibited. 

Hoffmeister and Sag (2009) provide the following examples: 

 

(7) *Who would you approve of [NP my seeing ]? 

(8) *Which astronaut did you read [NP a book about ]? 

(9) *Which dignitaries do you think [[Sandy photographed the castle] and [Chris 

visited]? 

 

Counter examples like these led Ross to introduce a number of distinct island 

constraints that are still part of the descriptive vocabulary of the modern syntactic 

literature. 

 

2.6.1. The Island Constraints Proposed by Ross (1967) 

 

Ross (1967) proposed that there are many types of syntactic islands, that is, 

constructions out of which it is not possible to move. Some of these constraints are 

given below: 

 

2.6.1.1. Complex NP Constraint (CNPC) 

 

Movement is prohibited out of a noun phrase which includes a clause, either a nominal 

complement clause, or a relative clause. For instance; 

 

 

 

 

 

 



35 
 

(10) * When do you remember the young girl who you met?  

 

In this derivation, the wh-expression „when‟ originates within the lower CP and moves 

to the specifier position of the matrix CP. However, this movement is prohibited by the 

DP that c-commands the lower CP. The derivation crashes and results in 

ungrammaticality. This constraint is called as the Complex NP Constraint.  

 

2.6.1.2. Sentential Subject Constraint (SSC) 

 

 No element can be moved out of the subject position of a clause. For instance; 

 

 

 

 

 

 

 

 



36 
 

 

(11) * What is that George has already bought known by everybody? 

 

In this derivation, the wh-word „what‟ originates within the lower CP which originates 

within the matrix spec vP. First, the whole lower CP moves to the matrix spec TP 

position for case checking purposes. Then, only the wh-word, „what‟ moves to the 

matrix spec CP position to check its interpretable wh features. Yet, Sentential Subject 

Constraint claims that no elements can be moved out of sentential subject positions, in 

other words, from the matrix spec TP positions. Hence, this derivation is ungrammatical 

according to this constraint since the wh-word moves out of the sentential subject 

position.   

 

2.6.1.3. Coordinate Structure Constraint (CSC)  

 

In a coordinate structure, no coordinate may be moved, nor may any element contained 

in a coordinate be moved out of that coordinate unless it moves from all coordinates. 

For instance;  

 

 

 



37 
 

 

 

(12) * Who does Mary love Jason and?  

 

In this derivation, the wh-word „who‟ originates within a coordinate structure and its 

movement to a higher node is prohibited by the Coordinate Structure Constraint.  

     

2.6.2. The Island Constraints Proposed After Ross (1967) 

 

 After Ross, (1967), the number of island constraints was extended with the works of the 

scholars like Kiparsky and Kiparsky (1970), Chomsky (1973), Ross (1984), Schafer 

(1995) etc.      

 

2.6.2.1. Wh-Island Constraint (WIC) 

 

Wh-extraction is prohibited out of another wh-clause. For instance; 

 

 

 



38 
 

(13) * Who did Mary tell you when she visited?  

                                                                                      

In this derivation, there are two wh-expressions both of which originate within the 

lower CP. One of these wh-words, „when‟, moves to the spec position of the lower CP 

and fills this node. The other wh-word, „who‟, has to move directly to the spec CP 

position of the matrix clause because the lower spec CP position has already been filled. 

Since this movement is not done cyclically, it violates the Wh-island Constraint. 

  

 

 

 

 



39 
 

2.6.2.2. Adjunct Island Constraint (AIC) 

 

Movement is prohibited out of an adjunct/adverbial. For instance; 

 

(14) * For whom did Peter stay at home as he wanted to wait? 

 

The wh-expression „for whom‟ originates within the adjunct clause. When it moves to 

the matrix spec CP position, it violates the Adjunct Island Constraint.  

 

 Along with these island constraints, there are also others which are mentioned in the 

literature like Left Branch Constraint, Right Roof Constraint, Factive Island Constraint, 

Negative Island Constraint and Superiority Constraint. 

 

2.6.2.3. Left Branch Constraint (LBC) 

 

An island constraint which states that an NP which is the left branch of a larger NP is an 

island. For instance; 

 



40 
 

(15) * [Which senator‟s] did you meet [e wife]?  

 

2.6.2.4. Right Roof Constraint (RRC) 

 

The constraint, in English and some other languages, by which some element of a 

sentence which occurs further to the right than its expected „logical‟ position cannot 

occur more than one clause boundary to the right of that position. For instance; 

 

(16) *That [it surprises you] [is amusing] [that Lisa smokes]  

 

2.6.2.5. Factive Island Constraint 

 

Kiparsky & Kiparsky (1970) proposed Factive Island Constraint. They note that there 

are two classes of predicates, those that presuppose the truth of their sentential 

complements (factives) and those that do not (non-factives) (In Cuba 2006, p.123). 

They exemplify the factive verbs as: regret, resent, hate, comprehend, grasp, make clear, 

like; and nonfactives as: believe, claim, say, assert, is possible, conjecture. For instance; 

 

(17)- *How do you regret that you behaved t ?   

(18)- How do you think that you behaved t ?  

 

2.6.2.6. Superiority Constraint 

 

Chomsky (1973) stated that extractions cannot apply to an element if it could have 

applied to a superior element (p. 276). The sentences below exemplify this constraint: 

 

(19) Who ___ saw what? 

(20)* What did who see ___? 

 

 

 

 



41 
 

2.6.2.7. Negative Island Constraint 

 

Ross (1984) proposed Negative Island Constraint. Sentences like (20) in which an 

adjunct wh-phrase has been extracted over negation are judged less acceptable than 

sentences like (21) in which an object wh-phrase has been extracted (as cited in 

Gieselman et al, 2010, p.1). 

 

(21) ??How precisely didn‟t the student report her results __? 

(22) ?Which results didn‟t the student report __ very precisely? 

 

2.6.3. Subjacency 

 

As the number of the island constraints were extended, the scholars started search for 

principles or conditions to unify such constraints. The first attempt came from Chomsky 

(1973). He proposed the Subjacency condition to gather some of these island constraints 

under one title. Chomsky (1973) defines subjacency as: 

 

1- if X is superior to Y in a phrase marker P [roughly, if X asymmetrically c-commands 

Y], then Y is „subjacent‟ to X if there is at most one cyclic category C … Y such that C 

contains Y and C does not contain X. Thus, if Y is subjacent to X, either X and Y are 

contained in all the same cyclic categories or they are in adjacent cycles. 

 

2- No (movement) rule can involve X,Y; X superior to Y if Y is not subjacent to X (as 

cited in Lasnik, 2010, p.3) 

 

In this condition, it is claimed that A-bar movement may not cross more than one 

bounding node.  In English, bounding nodes were claimed to be the NP‟s and IP‟s.  

Hence, the grammar excludes examples like the following: 

 

(23) * It was a new company that Simon spread [NP the rumor that [IP they started]]. 

(24)* What did [IP [NP the attempt to find] end in failure]? (pp. 4-5). 

  



42 
 

Yet