Pro-drop parameter, universal grammar and second language ...
-
Upload
khangminh22 -
Category
Documents
-
view
1 -
download
0
Transcript of Pro-drop parameter, universal grammar and second language ...
PLEASE TYPE THE UNIVERSITY OF NEW SOUTH WALES Thesis/Project Report Sheet
Surname or Family name: Chen Other name/s: N/A First name: Jin Song Abbreviation for degree as given in the University calendar: PhD School: Modem Language Studies Faculty: Arts and Social Sciences Title: Pro-drop Parameter, Universal Grammar 'llld Second Language Acquisition
of Chinese and English
Abstract 350 words maximum: (PLEASE TYPE)
This research is an attempt to consider the application of Chomsky's Principles and Parameters theory to the process of learning Chinese and English as the second language in terms of the prodrop parameter and other related features . It is concerned with the investigation of directional difficulty in second language acquisition by Chinese-speaking learners learning English as a foreign language (CFL) and by English-speaking learners learning Chinese as a foreign language (EFL). By choosing the syntactic area of the pro-drop parameter and other related linguistic features , such as null and non-null subjects/objects, base/non-base-generated topics, and locaVlong-distance reflexives for this study, it is expected that further insights will be obtained about the contrastive study of second language learning. An experiment was conducted on 135 Chinese-speaking learners of English and 70 English-speaking learners of Chinese to compare the learners ' L2 competence of syntax in Chinese and English respectively. The findings of this study have shown that Chinese and English are different languages that have different settings for the so-called pro-drop parameter and related language features . In terms of the linguistic features investigated in this study, it is suggested that there are no single directional difficulties in the second language acquisition of Chinese and English. While some language features are more difficult for EFLand CFL learners to acquire, others are more learnable in terms of the difficulty level. Some linguistic features are easier to "unlearn", while others are more difficult to "acquire". Factors such as the availability of positive evidence in the learners' input data as well as their ability to process the relevant information for the restructuring of their interlanguage of the target language, will have great impact on the directional difficulties in the second language acquisition of Chinese and English, respectively.
Declaration relating to disposition of project report/thesis
I am fully aware of the policy of the University relating to the retention and use of higher degree project reports and theses, namely that the University retains the copies submitted for examination and is free to allow them to be consulted or borrowed. Subject to the provisions of the Copyright Act 1968, the University may issue a project report or thesis in whole or in part, in photostat or microfilm or other copying medium.
I also authorize the publication by University Microfilms of a 350 word abstract in Dissertation Abstracts International (applicable to doctorates only) .
. . . 1 :J Signature
2:2-jo 7(;;oo J ... .... .... ...... ... ... .. .... .. Witness Date
The University recognises that there may be exceptional circumstances requiring restrictions on copying or conditions on use. Requests for restriction for a period of up to 2 years must be made in writing to the Registrar. Requests for a longer periods of restriction may be considered in exceptional circumstances if accompanied by a letter of support from the Supervisor or Head of School. Such requests must be submitted with the thesis/project report.
FOR OFFICE USE ONLY Date of completion of requirements
\ ~ . Cl · 0 \ ,.-----'---'--, ~~~al THIS SHEET IS TO BE GLUED TO THE INSIDE FRON
Pro-Drop Parameter, Universal Grammar and Second Language Acquisition of Chinese and English
by
JINSONGCHEN
DISSERTATION
Submitted to the Faculty of Arts and Social Sciences of
The University ofNew South Wales
In Partial Fulfillment
of the Requirements
for the Degree of
DOCTOR OF PHIT.,OSOPHY
The University ofNew South Wales
July, 2001
CERTIFICATE OF ORIGINALITY
I hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, nor material which to a substantial extent has been accepted for the award of any other degree or diploma at UNSW or any other educational institution, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis.
I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project's design and conception or in style, presentation and linguistic expression is acknowledged.
(Signod) ·I ~------------------------~
ACKNOWLEDGEMENTS
I am grateful to all members of my advisory committee members for their advice, encouragement and support. First and most importantly, I would like to thank those who have supervised my study for all their help and encouragement. I would like to extend my sincere thanks and appreciation to my thesis advisor, Associate Professor Peter Collins, for his kind assistance in every phase of the preparation of this dissertation. His insightful comments, valuable suggestions, patience and fine sense of humor have always provided me with stimulation for the development of my interest. I would like to thank my cosupervisor, Dr. Mengistu Amberber, for his enlightening advice, valuable comments, and for his support and encouragement in many ways. I would also like to thank Sue Hood for her help and encouragement in the early stages of this study. I am very grateful to Martin Cooper for his valuable advice on statistical analyses of this study.
I wish to thank the University ofNew South Wales (UNSW) for offering me an Overseas Postgraduate Research Scholarship; and I would like to thank Professor John lngleson for granting me a Faculty of Arts and Social Sciences Scholarship, both of which have made my Ph.D. studies possible. I would like to thank the School of Modem Languages, UNSW, for offering me an External Study Grant to travel to China conducting my fieldwork and collecting experimental data; for a postgraduate travel grant to attend the Australian Linguistic Institute held in Brisbane in June 1998.
My sincere appreciation also goes to all the students for their enthusiastic participation in the studies. I would like to thank the students in the following institutions who helped me to collect data: Sichuan University, Sichuan Educational College, Chengdu University of Science and Technology, Chengdu Yandaojie Senior Middle School, Beijing Language and Culture University, and the University ofNew South Wales.
I also want to express my sincere gratitude to my colleagues and friends: Henry Lee, Liping Xia, Xiang Chen, Lijiang Zhao, Vivian Y. Zhang, Ren Zhang, David Kilpatrick, X. L. Dai, Annie Y. Zhang, Shirley X.W. Shi, Angela Fan, Lily Shi, Kyung Seo Mo, L. K. Chiu, Lucia Deng, Yvonne Y. Ding, Shigenori Wakabayashi, Chiho Sunakawa, Jae Sung Lee. Finally and most of all, I thank my parents for their constant support and love.
ii
Abstract
This research is an attempt to consider the application of Chomsky's Principles and
Parameters theory to the process of learning Chinese and English as the second language
in terms of the pro-drop parameter and other related features. It is concerned with the
investigation of directional difficulty in second language acquisition by Chinese-speaking
learners learning English as a foreign language (CFL) and by English-speaking learners
learning Chinese as a foreign language (EFL). By choosing the syntactic area of the pro
drop parameter and other related linguistic features, such as null and non-null
subjects/objects, base/non-base-generated topics, and local/long-distance reflexives for
this study, it is expected that further insights will be obtained about the contrastive study
of second language learning. An experiment was conducted on 135 Chinese-spealdng
learners of English and 70 English-spealdng learners of Chinese to compare the leamers'
L2 competence of syntax in Chinese and English respectively. The findings of this study
have shown that Chinese and English are different languages that have different settings
for the so-called pro-drop parameter and related language features. In terms of the
linguistic features investigated in this study, it is suggested that there are no single
directional difficulties in the second language acquisition of Chinese and English. While
some language features are more difficult for EFL and CFL leamers to acquire, others are
more leamable in terms of the difficulty level. Some linguistic features are easier to
"unleam", while others are more difficult to "acquire". Factors such as the availability of
positive evidence in the learners' input data as well as their ability to process the relevant
information for the restructuring of their interlanguage of the target language, will have
great impact on the directional difficulties in the second language acquisition of Chinese
and English, respectively.
AGR AM ANOVA BC CA CAH CCH CFL CL EC ECP EFL EFC EPP EXP GB GC GF I Ll L2 LF NGT NL NP PFV SLA SPEC TC TL TNS UG
ABBREVIATIONS
agreement aspectualmarker analysis ofvariance blocking category Contrastive Analysis Contrastive Analysis Hypothesis Creative Construction Hypothesis learning Chinese as a foreign language classifier empty category the empty category principle learning English as a foreign language free empty category Extended Projection Principle experiential aspect marker government and binding governing category Grammatical Function INFL (inflection) first language second language logical form non-gap topic native language noun phrase perfective aspect marker second language acquisition specifier topic chain target language tense universal grammar
iii
Table of Contents
Acknowledgements i
Abstract ii
Abbreviations iii
Chapter One Introduction 1
1.1 Introduction 1 1.2 The Study ofDD and SLA 4 1.3 Purpose of this Research 8 1.4 Organization of this Thesis 11
Chapter Two Universal Grammar 13
2.1 Universal Grammar 13 2.2 UG as a theory of grammar 13 2.3 GB theory: levels of representation and movement 16 2.4 Subtheories of GB Theory 20
2.4.0 Introduction 20 2.4.1 X-bar Theory 21 2.4.2 C-command and M-movement 24 2.4.3 Government 26 2.4.4 Binding Theory 29 2.4.5 9-Theory (Theta- Theory) 32 2.4.6 Movement Theory 35 2.4.7 The Projection Principle (PP) and the
Extended Projection Principle (EPP) 42 2.4.8 Bounding Theory 43 2.4.9 Case Theory 44 2.4.1 0 Empty Categories and the Empty Category
Principle 45 2.4.11 UG and Language Acquisition 53
2.5 Conclusion 57
Chapter Three Universal Grammar and Second Language Acquisition 59
3.1 Introduction 3.2 Interlanguage 3.3 The Access ofUG in SLA
3.3.1 Direct Access to UG 3.3.1 Indirect Access to UG 3.3.2 No Access to UG
3.4 Evidence of Access to UG
59 60 63 66 66 67 68
iv
3.5 Evidence of Lack of Access to UG 3.6 The Role ofL1 in SLA
3.6.1 L1 and Language Transfer 3 .6.2 Parametric Variation
3.7 Learnability considerations and SLA 3.7.1 Direct Positive Evidence 3. 7.2 Indirect Positive Evidence 3. 7.3 Direct Negative Evidence 3. 7.4 Indirect Negative Evidence
3.8 Learnability problem and unlearning strategies 3. 8.1 Overgeneralizations 3.7.2 The Subset Principle
3.9 Conclusion
Chapter Four: Null Subject and Null Objects in Chinese
4.1 Introduction 4.2 Null Subjects
4.2.1. Null Subjects and Pro-drop Parameter 4.2.2. Two Principles to Identify a Pro-drop Language 4.2.3. C.-T. J. Huang's Modification Approach 4.2.4. Null Subjects and Morphological Uniformity 4.2.5. Topic Chain and Topic Identification 4.2.6. The Finite and Non-Finite Clause Distinctions in Chinese 4.2.7. Topic Structure and Base-Generated Topics in Chinese
4.3 Null Objects 4.3.1. C. -T. J. Huang's Variable Analysis 4.3.2. Xu & Langendoen's and Xu's Approach 4.3.3. Arguments by other linguists 4.3.4. Summary
Chapter Five: Chinese and English Reflexives: An Introduction
5.1 Chinese and English Reflexives 5.1.1. Local and Long-Distance Binding:
Theoretical Considerations 5.1.2. C. -T. J. Huang's Approach 5.1.3. Tang's Reindexing Approach 5.1.4. Battistella's Move-to-INFL Analysis 5 .1.5. Cole et al and Cole & Sung's Head Movement Approach 5.1.6. Relativized SUBJECT Approach
5.2. Long-distance Binding ofziji in L1 Acquisition 5.3. Long-distance Binding ofziji in SLA 5.4. Summary
v
71 74 75 77 82 83 85 86 88 90 90 95 100
102
102 102 102 105 109 112 115 117 122 125 125 132 134 135
137
137
137 142 143 145 147 151 153 154 157
VI
Chapter Six: The Experimental Study 158
6.0 Introduction 158 6.1. Research Questions and the Experimental Hypothesis 158 6.2. Informants in the Experimental Study 162
6.2.1. The EFL Informants 162 6.2.2. The CFL Informants 164 6.2.3. EFL informants and their background 166
6.3. The Instruments 170 6.3.1. The Acceptability Judgments Test 170 6.3.2. Arguments against and for the acceptability tests 170 6.3.3. Design considerations 174
6.4. The Administration and Test Procedure 176 6.5. The Procedures of the Data Analysis 179 6.6. SPSS and ANOV A Analysis 181
Chapter Seven: Analysis of the Experimental Data 185
7.0 Introduction 185 7.1. Test sentence structure for EFL informants 185
7.1.1. Test sentences with null subject subjects for the EFL informants 185
7.1.2. Test sentences with base-generated topics or non-gap topics for the EFL informants 187
7.1.3. Test sentences with null objects for the EFL informants 188 7.2. Test sentence structures for the CFL informants 190
7.2.1. Test sentences with null subject for the CFL informants 190 7.2.2. Test sentences with base-generated topics or
non-gap topics for the CFL informants 193 7.2.3. Test sentences with null object for the CFL informants 194
7.3. Analysis of the EFL and CFL informants' judgments 198 7.3.1. The EFL informants 198 7.3.2. The CFL informants 218
7.4. Analysis of Results 237 7.4.1. The acquisition of null subject by EFLICFL informants 237 7.4.2. The acquisition of nominal expletive it
by EFLICFL informants 249 7.4.3. The acquisition of BGT by EFLICFL informants 251 7.4.4. The acquisition of null object by EFLICFL informants 256
7.5. Conclusion 260
Chapter Eight: The Acquisition of Reflexives 263
8.0 Introduction 263 8.1. Reflexive Binding and Language Acquisition 264 8.2. Experiment and Test Sentences 270
vii
8. 3. The judgments on reflexive binding by EFLICFL ,informants 271 8.4. The Analysis of Results 273 8.5. Conclusion 276
Chapter Nine: Summary and Conclusion 278
9. 1. Summary and Conclusion 278 9.2. Limitations and suggestions for further studies 279
References 285
Appendix 310
1. Second Language Acquisition Research Survey Questionnaire. 311
2. Acceptability Judgment Test for Learners of English as a Foreign Language (EFL): English Version Instructions 312
3. Acceptability Judgment Test for Learners ofEnglish as a Foreign Language (EFL): Chinese Version Instructions. 313
4. Vocabulary List for the Acceptability Judgment Test ( CFL subjects) 314
5. Vocabulary List for the Acceptability Judgment Test (EFL subjects) 316
6. Acceptability Judgment Test for the Learners of English as a Foreign Language (EFL) · 317
7. Acceptability Judgment Test for the Learners of English as a Foreign Language(CFL) 321
8. Subject Information Statement for the Subjects 328
9. Consent Form for the Students Participating in Research Projects 329
10. Sentence Types in the Acceptability Judgment Test for the CFL Subjects 330
11. Sentence Types in the Acceptability Judgment Test for the EFL Subjects 337
12. SPSS EFL/CFL ANOVA Tables 1 & 2 341
13. SPSS Tukey HSD Tests Tables 7 (1-30) & 8 (1-2) 345
Chapter 1 Introduction
1.1 Introduction
In recent decades, there has been growing interest in applying the theory of Universal
Grammar (UG) to second language (L2) acquisition/learning (SLA) as well as first
language (L1) acquisition. Chomsky (1983: 118) defines UG as ''the inherited genetic
endowment that makes it possible for us to speak and learn human language". It is
claimed that children hmately posses some abstract linguistic structure, by which they
come to produce and comprehend their mother tongue as time passes by (Chomsky,
1965, 1981a, 1982, 1986a). The common human endowment, this innate abstract
structure is called "Universal Grammar", and, according to Chomsky, "consists of a
highly structured and restrictive system of principles with certain open parameters, to be
fixed by experience" (Chomsky, 1981b: 38). Further, claims Chomsky (Chomsky,
198lc), "by fixing each of the parameters, we determine a particular grmmnar which, let
us suppose, generates a specific language" (ibid., p.126). The idea is that a small number
of open parameters are related with some principles ofUG and that when specific
evidence of the language being learned has set the relevant parameter at a particular
value, a language is produced.
According to this formalized process ofL1 acquisition, any children will be ultimately
able to acquire a language, provided that they are living in a language community, and
that they ·are provided with evidence of that language, no matter whether they are born of
Chinese, English or Spanish parentage. What language children will acquire depends
1
wholly on what kind of evidence is involved in the process of their language acquisition.
The simple reason for this is that children are endowed with an omnipotent machine
known as UG, which can process any kind of evidence offered and therefore process a
specific language. Obviously if the two conditions, UG and evidence, are satisfied, all
children will undergo the process ofL1 acquisition smoothly in the same way, and they
will not encounter any particular difficulty in the course of language acquisition.
2
Adults however, unlike children who uniformly succeed in attaining native competence
in their L 1, generally fail to achieve native competence in their acquisition of an L2
(Bley-Vroment, 1989). This lack of success in SLA can be related to quite a few factors.
The L2 learners have already acquired knowledge of a native language (NL), i.e. their L1,
and will presumably bring their knowledge of the L1 to the acquisition of the L2. In other
words, since the L2 learners are faced with new evidence from the L2 different from the
old evidence from their NL, the two kinds of evidence from the two different languages
may interfere with each other. So, a few questions might be asked about the acquisition
of an L2: how do the knowledge of L 1 and the knowledge of L2 interact with each other,
how does the L 1 knowledge affect the L2 acquisition? Are there any aspects of L2
features, which are more or less difficult to acquire than others? For attempts to answer
such questions, we may find factors such as the following to be of concern in the study of
L2 acquisition: areas of difficulty, degree of difficulty and directionality of difficulty.
By area of difficulty we mean any linguistic features (either at the syntactic, semantic or
phonological level) of the L2, which present an intellectual challenge to the L2 learner as
a result of differences between the two languages, i.e. L1 and L2. For example, in
3
English, there is a rule adding "-s" to the verb in the present tense if the subject is a third
person singular. As this rule is unique to English, Chinese learners at the beginning stage
of English learning will fmd it very difficult to recognize this special phenomenon and
others. As a result, they will produce, from time to time, such ungrammatical sentences as
"* He like to play basketball", or"* She go to the movie very often", without having the
verbs inflected appropriately. Degree of difficulty refers to how much or to what extent
particular features of the L2 will be difficult for the L2 learner. With respect to the above
example, degree of difficulty would involve considering how much time it takes a
Chinese speaker of English to internalize the rule adding "s" to the verb when it is
needed.
Finally, directionality of difficulty (DD) refers to a situationjn which two languages
differ with respect to a certain language phenomenon (either at the syntactic, semantic or
phonological level), and in which this phenomenon is more difficult for learners of one of
the languages than of the other. For instance, the "-s" rule for verbs in English may be
more problematic for Chinese speakers learning English, or the non-occurrence of this
rule in Chinese may be more difficult for English speakers learning Chinese.
The three factors are closely interrelated. The directionality of difficulty entails area of
difficulty and degree of difficulty. The directionality of difficulty has constituted the
main obstacle for both language teachers and applied linguists. So, how do L 1 features
affect L2 acquisition? Are there L2 features that are more difficult or less difficult to
acquire than others? If there are, we have to provide an account for the directionality of
difficulty when it occurs in SLA. And we have to explain the phenomenon of some L2
features being more or less difficult than others.
1.2 The study of DD and L2 acquisition
4
From the beginnings ofthe study ofL2learning and teaching, the factor ofDD has been
theoretically discussed and experimentally investigated by applied linguists. Proponents
of Contrastive Analysis ('CA') (Fries, 1945; Lado, 1957), which was very influential in
SLA research from the 1940s to the 1960s, tried to solve the problem of predicting and
explaining difficulties which occur during the process oflearning an L2. They found that
L2learn~rs find easy those L2 structures, which are similar to those in their L1, and they
have difficulties where the two languages differ. The underpinnings of theCA approach
lie in the hypothesis that a detailed description of the similarities and differences between
the particular NL and the target language (TL) can show where the hidden difficulties are
for the L2 learners.
CA is based on a theory oflanguage that claims that language is a habit, and the major
source of errors in the production/reception of an L2 is the NL. Since one can account for
errors by considering differences between the L 1 and the L2, what one has to do in
learning an L2 is learn the differences, and the similarities can be safely ignored as no
new learning is involved. According to CA theorists, "difficulty and ease in learning are
determined respectively by differences and similarities between the two languages in
contrast" (Gass & Selii1ker, 1994: 60). However, the publication of Chomsky's classic
5
review of Skinner's Verbal behavior seriously challenged the behaviorist view of
language acquisition. Researchers such as Whitman and Jackson (Whitman & Jackson,
1972) concluded that, "contrastive analysis was inadequate to predict the interference
problems of a language learner". Long and Sato (Long & Sato, 1984) also pointed out
that the most damaging flaw of the CA was the dubious assumption that the learner could
depend solely upon an analysis of a linguistic product to yield meaningful insights into a
psychological process, i.e. L2 learning. In sum, the approach of CA failed to make the
prediction of learning difficulties accurate, and failed to make directional predictions
possible.
Given the inadequacy of CA to account for the facts of directional difference in difficulty
for the L2 learner, some researchers began to pursue other approaches. For instance,
Eckman (1977) incorporated the notion of typological markedness into the CA
hypothesis together with the concept ofimplicational universals to discuss the factor of
DD. In this way, he was able to predict that speakers ofPersian will have great difficulty
with English relatives; speakers of Arabic will have less trouble than Persian speakers,
but more than Chinese speakers, who, in tum, will have greater difficulty than Japanese
speakers. This, he argues, is due to the nature of human cognition, since ''humans learn to
do things which are less complex before they learn to do things which are more complex,
and further, no human being leams to do things which are more complex without also a
fortiori learning to do related things which are less complex" (1977: 315-330). English is
the most marked and Persian the least marked, with Japanese, Chinese and Arabic being
between the two extremes, respectively.
6
With the appearance of Chomsky's formal notion of parametric variation, many linguists
began to pay special attention to universal aspects of L2 learning. According to the theory
of parametric variation, languages are understood to vary from one another in the way in
which a particular parameter is set for some principles ofUG. White (White, 1985b: 47)
proposed that, "adults learning a second language have particular problems when their
mother tongue has activated a parameter of Universal Grammar which is not operative in
the second language. It is suggested that the learner will carry the parameter over fi:om Ll
to L2, causing transfer errors." Among the few researchers (Flynn, 1983; White, 1985b;
1986c; 1987b; Phinney, 1987; Liceras, 1988) who have considered markedness within
the parameterized model ofUG in L2learning, White has contributed the most
significant studies. In a series of experimental studies, White claims that, the 'pro-drop
parameter' in Spanish will pose a greater problem for Spanish speakers learning English
than the absence of a 'pro-drop parameter' in English will pose for English speakers
learning Spanish (White, 1985c). Concerning the principle of adjacency, it is more
difficult for French learners of English to master "strict" adjacency in English, while it is
easier for English learners of French to master the lack of strict adjacency in French
(White, 1987a). Since there are many models and definitions of markedness in the field
of linguistics and psycho linguistics, the terms concerning markedness are not used in this
research in order to avoid confusion.
However, from a learnability point of view, White (White, 1985b, c; 1986a, b; 1987b)
takes the position that DD in SLA depends on the availability of positive evidence to L2
7
learners. If a linguistic feature in L2 is more inclusive than that in the learners' L 1 (i.e. if
the linguistic feature in L2 includes a structure which is not instantiated in the L1
feature), the positive evidence in the L2 input can indicate to the L2 learners that their L 1
grammar is incompatible with the grammar of the L2 with respect to the language
features concerned and the change is motivated. However, in terms of a certain language
feature, when the L1 is more inclusive than the L2, the L2 learners will not have positive
evidence in the L2 input data to help them to unlearn the structures instantiated in the L1
and thus difficulties will occur. White (1986a) predicts that an L1 structure which can not
be disconfirmed by L2 data is likely to persist and remain as a candidate for what
Selinker (1972) terms "fossilization". In recent years, however, some doubts have been
expressed about the role of the availability of positive evidence in determining ease or
difficulty in SLA (cf. White, 1989; Sharwood-Smith, 1991; Sorace, 1991). Furthermore,
the availability of positive evidence alone is not a guarantee that L2 learners make use of
it for the restructuring of their interlanguage grammars.
Considering the learning of Chinese and English, it is certain that all the L2 learners will
experience some particular problems with the use of certain language points that are
unique to the target language (TL). So far, however, the exact nature of the problems is
far from having been identified. The limited research within the context of Chinese and
English on the factor ofDD for students learning Chinese and English as the L2, means
that there is a lot of work needing to be done in this area. Therefore, it is a worthwhile
endeavour to investigate both the learning of English by Chinese learners and the
learning of Chinese by English learners, and the theoretical study of L2 learning in a
wider sense.
8
In addition, some of the linguistic features under investigation, such as the features
concerning empty categories, are being discussed in depth both in Chinese linguistics in
particular and in linguistics in general. Moreover, an increasing number of speakers of
various language backgrounds, including native speakers of English, are learning Chinese
as an L2, and more than half of the students in China (i.e. middle school and university
students) have taken or will take English as their L2. On the one hand, the contrastive
study of Chinese and English learning by native speakers ofEnglish and Chinese is
needed for applied linguists, teachers and learners of the two languages. On the other
hand, although Chinese is spoken by over one billion people, research on the acquisition
of Chinese as an Ll and L2 is still in its infancy. This research is an attempt to contribute
to the unfortunate gap in the field of SLA in the context of L2 acquisition of Chinese and
English.
1.3 Purpose of this research
This research is an attempt to consider the application of Chomsky's Principles and
Parameters theory to the process oflearning Chinese and English as L2s in terms of the
pro-drop parameter and other related features. It is concerned with the investigation of
directional difficulty in SLAby Chinese-speaking learners of English as a foreign
language (EFL) and by English-speaking learners of Chinese as a foreign language
9
(CFL). More specifically, the syntactic area of the pro-drop parameter and other related
linguistic features, such as null and non-null subject/object, shmi/long-distance
reflexives, etc., will be investigated. It is expected that further insights will be obtained
about the contrastive study of L2 learning. Chinese and English are languages which have
different settings for the pro-drop parameter: when the parameter is set at the[+] value,
then it is Chinese with[+ pro-drop]; when the parameter is set at the[-] value, then it is
English with [-pro-drop].
In order to address the question of what causes difficulties for the learners of the two
languages, the pro-drop parameter and related language features will be discussed
throughout the research, and be explored respectively. The purpose is to examine:
(1) Whether there is a directional difficulty in the course of English learning by Chinese
speakers and in the course of Chinese learning by English speakers; Whether
Chinese-speaking learners would find it more difficult to learn English or would
English-speakers of Chinese find it more difficult to learn Chinese; why there is such
a directional difficulty, or why there is not one?
(2) Whether the non-availability of positive evidence in the L2 input causes learning
difficulty.
In this research, it is hypothesized that the availability of positive evidence in the L2
input leads to ease of acquiring the property and that the non-availability of positive input
leads to difficulty.
In order to address the above questions, the linguistic features presented (5)- (7) below
will be explored in subsequent chapters.
(3) Null/Non-null subject
e.g. a) zhe ge shiyan YIJmg kaishi, wo xiangxin this classifier experiment already start e hui chenggong.
I believe
will succeed
b) * The experiment has been started. I am sure e will be successful.
( 4) Null/Non-null object
a) ruguo if
ni xiechu you write
mai e. buy
yi ben hao a CL good
shu , wo book I
b) * If you write a good book, I'll definitely buy e.
(5) Short-/Long-distance reflexives
a) Zhangsani Johnson
renwei think
Lisij xiangxin Bob believe
b)* Jolmsoni thinks Bobj trusts himself+ ilf·
(6) Base-generated/non-base-generated topics
ziji;ij self
yiding definitely
a) zhe this
ben shu wo bu zhidao weishenmo ta bu CL book I not know why he not
xihuan. like
b) * This book I don't know why he doesn't like.
As can be seen from the above examples, Chinese allows both a null subject in finite
clauses and a null object, as is shown in (Sa) and (6a). However, the null subject in finite
clauses and the null object give rise to ungrammaticality in English, as in (5b) and (6b),
10
11
respectively. Chinese and English differ in that while English reflexives can only take the
local subject as its antecedent, as in (7b), the Chinese reflexive "ziji" can take the matrix
subject as well as the embedded subject as its antecedent, as in (7a). Unlike Chinese, in
which topics can be base-generated, such as "zhe ben shu" (=this book) in (8a), English
does not allow a base-generated topic, as is shown in (8b ), since the topic in English has
to be the result of movement and is subject to the Subjacency Principle 1•
1.4 Organization of the thesis
This thesis is divided into 9 chapters. Chapter 1 is the introduction, which deals with the
general ideas of why and how the present thesis is written. Chapter 2 is devoted to the
theoretical framework of Principles and Parameters Theory on which the present research
is based. It is also devoted to the discussion ofUG as a theory of grammar and UG as a
theory of L 1 acquisition, and introduces the linguistic facts concerning the linguistic
features which are under investigation in later chapters. In Chapter 3, I devote some space
to theoretical issues in SLA, such as the interlanguage, learnability theory, Ll transfer
and other related hypothesis. Chapter 4 deals with the empirical aspects of the research.
The theoretical hypothesis concerning the directionality of difficulty for the learners of
Chinese and English as the L2 is developed. The methodology of the investigation is
reported in detail. Chapters 5 to 8 are devoted to the study of the acquisition and non
acquisition of the null/non-null subject/object, short-/long-distance reflexives by Chinese
speaking learners of English as well as English-spealdng learners of Chinese. In Chapter
9, I draw some conclusions on the basis of the results obtained, some implications and
12
applications of the findings are proposed for the L2 acquisition of Chinese and English.
In addition, the contributions and limitations of the present research and some
suggestions for further studies are discussed.
1 Chomsky (1977) states that a trace may not be separated from its antecedent by more than one "bounding node". In other words, no element may be moved across more than one bounding node. The bounding node for English is considered to be IP and NP (cf. Chomsky, 1977; Lasnik & Uriagereka, 1988).
13
Chapter2 lJniversal(;raununar
2.1 lJniversal (;raununar
Universal Grammar (UG) is both a theory of properties of grammar as well as a theory of
the initial biological endowment for language (Flynn, 1987:32). UG is hypothesized to
consist of a finite set of abstract and linguistically significant principles underlying all
languages. It is argued that these principles define the "initial state" of the first language
(L1) learner's mind (Chomsky, 1980a: 69). The principles lay down the fundamental
features of natural language, whereas the parameters embody their variability, in that they
offer a limited range of choices as a consequence of which languages may differ by
taking one of the permitted values of each parameter (Saleemi, 1992: 11). UG, as a theory
of grammar, predicts that languages are constrained in form; and as a theory of a child's
"initial state", it predicts that language acquisition is constrained by the existence ofUG
(Lust, 1986).
In this chapter, we will briefly discuss UG as a theory of grammar, and its properties. We
will then introduce the relevant linguistic facts concerning the language features involved
in the study of the second language acquisition (SLA) of Chinese and English. And
finally, we will examine briefly the role ofUG as a theory ofL1 acquisition.
2.2 lJniversal (;raununar as a theory of graununar
As a theory of grammars, the goal ofUG is to provide "a system of principles,
conditions, and rules that are elements or properties of all human languages, not merely
14
by accident, but by necessity ... the essence ofhuman language" (Chomsky, 1975: 29). It
is ''taken to be a characterization of the child's prelinguistic state" (Chomsky, 1981a: 7).
Theories ofUG have changed considerably in the past decades. It seems that Chomsky's
earlier model of transformations "within the standard model of transformational grammar
was highly complex, with conditions and stipulations built in here and there in response
to particular problems that arose in the process of devising grammars that approached the
goal of descriptive adequacy" (Horrocks, 1987: 81 ). Later Chomsky shifted the emphasis
away by addressing the question "not of what the rule system for a grammar of English
must be like but rather of why the rule system has the properties that it does" (1987: 81).
That is to say, the emphasis has been shifted away from the development of the more or
less descriptively adequate rule systems to the development of more explanatorily
adequate theories of universal grammar.
Much of the work that will be discussed in this thesis has been conducted within the
framework of Government and Binding (GB) Theory or the Principles and Parameters (P
& P) approach within current linguistic theory (Chomsky, 1981a, 1986a, b). GB theory
leads to a complex overall theory involving abstract sub-theories but at the same time
creates a new simplicity; knowledge oflanguage comes down to variations in a small
number of properties. This is the major shift of the theory; rules and stipulations are to be
explained as the interactions of principles and lexical properties rather than existing in
their own right.
15
Within Chomskyan theory, learner behavior is assumed to belong to a theory of use
(White, 1989b: 18). UG is a theory of knowledge, not of behavior. It prefers general
statements that cover many instances rather than particular statements that cover only
one. UG specifies ''those aspects of these rules and principles that are uniformly attained
in language but underdetermined by evidence" (Chomsky, 1981a: 6). Furthermore,
"certain principles within this system are hypothesized to be associated with parameters
that specify dimensions of structural variation across all languages" (Flynn, 1987: 33).
The values of these parameters are assumed to be fixed by experience gained in the
language learning process. Chomsky believes that setting the values in one way or
another will have deductive consequences for the rest of the grammar. Thus "in a tightly
integrated theory with fairly rich internal structure, change in a single parameter may
have complex effects, with proliferating consequences in various parts of the grammar"
(Chomsky, 1981a: 6). In broad terms, UG is thought to involve a set of general principles
accounting for the universal aspects of human languages, as well as parameters, which
encode the possible variations of the basic universal features.
However, there is no consensus among researchers as to the nature of elements with
which parameters are associated. Chomsky describes these parameters as "genetically
permitted variations that exist as options in the universal grammar" (Gliedman, 1985:
372). The variations are encoded in terms of "values" of the parameters that must be
"set" by the learner based on experience with the target language. The values may be
binary, having a(+) or (-)value, or they may have multiple values. Once the learner has
set the parameters, the "core grammar" has been identified. As Chomsky (198lb:38)
points out, "UG consists of a highly structured and restricted system of principles with
certain open parameters, to be fixed by experience. As these parameters are fixed, a
grammar is determined, what we may call a 'core grammar'. "
16
However, a different view is put forward by a number of other researchers ( cf. Borer,
1983; Wexler & Manzini, 1987). Some argue that there is no parameterization in the
principles ofUG and parametric variation should be regarded as the result ofvariation in
the properties o:flexical items. Borer (Borer, 1983:27) holds the view that "languages will
differ in the availability of particular inflectional rules ... " Wexler & Manzini (Wexler &
Manzini, 1987) proposes that parameters are not properties of principles but of individual
lexical ~terns in the lexicon, and "parameterization is essentially lexical" (1987: 47). Even
Chomsky (Chomsky, 1991) shifts his position fmm the parameterization ofthe principles
ofUG to the parameterization of the properties of functional elements. Ouhalla (Ouhalla,
1991) suggests a further development, claiming that parametric variation involving
substantive elements is determined by variation in the properties of functional categories.
2.3 GB theory: levels of representation and movement
GB theory provides one of the most well developed approaches to the study ofUG (cf.
Chomsky 1981 a, 1982, 1986a, b). Within this framework, the "T -model" proposed by
Chomsky and Lasnik (Chomsky & Lasnik, 1977) is adopted to assume a basic tripartite
differentiation of grammar (syntax, phonology, and semantics). (See Figure below F1-1).
Phonetic Form (PF)
D- structure
l S - structure
Logical Form (LF) (Lasnik & Uniagereka, 1988: 1)
The T -model determines a model of core grammar. The levels provide the structural
foundation for the transformational and interpretive rules. Transformational movement
connects the various levels of representation, i.e. D-structure, S-structure, phonetic form
(PF), and logical form (LF) (see Figure F1-1). These D-structures express semantically
17
relevant grammatical functions and relationships. And they are generated by a set of rules
of the base component- phrase structure rules of a very special kind, and the lexicon. D-
structure is purely a structural manifestation of8-grids 1 (Chomsky, 1981a; Gass &
Selinker, 1994). S-structures are the products of the application oftransformational rules
to each D-structure through the application of 'move a' 2 or 'do anything anywhere'
(Lasnik & Saito, 1984). Syntactic movement takes place between D-structure and S-
structure, during which D-structure is uniquely mapped onto S-structure by means of the
general rule "Move ci', which says "move anything anywhere" (Saleemi, 1992).
1 9-criterion is a principle ofUG, which specifies that each argument bears one and only one theta-role, and each theta-role is assigned to one and only one argument.
2 UG makes available a transformational component consisting of the single rule schema Move a. (i.e. "move any category a. anywhere, i.e. move anything anywhere"). The child "learning" the grammar of some particular language then simply has to find out what the permissible values of a. are in that language; in English, for example, a. would include at least NP and Wh. Everything also should then follow from the interaction ofthis information and the child's knowledge of universal grammar (Horrocks, 1987:99).
18
S-structure is dually mapped onto PF and LF. PF is a level at which various phonological
and phonetic operations take place, together with certain stylistic rules (such as wanna
contraction). LF is a level of representation for those aspects of meaning that relate to
sentence structure, such as scope of quantifiers. The rule "move ci' also mediates between
S-structure and LF.
Chomsky (1981a: 17) points out that grammar must express the relationship between
what a phrase or sentence means and the form that phrase or sentence takes. The T -model
assumes that there is an indirect lin1c between meaning and surface form, which is
mediated by abstract levels of syntax. By proposing different levels of representation, the
model attempts to capture similarities in meaning between two phrases or sentences, even
though the form may change. Similarly, the model can capture the fact that two similar
forms may have different meanings (Juffs, 1996).
Chomsky (1981a: 135) proposed a theory oflinguistic competence, or grammar, which
could be divided into two types of subsystems: one based on "rules" as in (la), and one
based on principles in (1 b): (cf. Horrocks, 1987: 97-101)
(1a).i. lexicon
ii. syntax
a) categorial component
b) transformational component (Move a)
iii. PF component (PF = phonological form = sound)
iv. LF component (LF =logical form= meaning)
(1 b). i. bounding theory
ii. government theory
iii. 8-theory
iv. binding theory
v. Case theory
vi. control theory
The system in (1a) and (lb) is modular. This means that different parts ofthe "rule
system" are constrained by different parts of the subsystem of principles.
19
The Principles and Parameters (P&P) approach to language is designed to account for the
observation that while human languages obviously vary in many ways, they nevertheless
share certain fundamental properties (Chomsky, 1981a). The subcomponents in (1a) and
(1b) are assumed to constitute a theory ofthe principles which underlie all human
languages. Hence, the theory is a theory ofUG, which must also be able to account for
variation, without losing the insight that language is constrained by universal principles
Juffs, 1996: 6).
The representation at each level and the rule "Move a" are constrained by an additional
set of principles. For example, X-bar theory severely restricts the option for phrase
structure rules. The subjacency principle sets constraints on movement. It disallows
moving something ''too far" in one operation of movement. The projection principle
requires lexical properties to be "projected" to all levels of syntactic representation.
Binding theory concerns the relationships of anaphors, pronouns, names and other
20
variables to their antecedents. Other subsystems of principle include, for example, 8-
theory, Case theory, control theory, government theory, etc. (cf. Chomsky, 1981).
It is important to note that, though each of the grammar component functions fairly
autonomously, they are closely related in several ways. This is because Principles and
Parameters theory describes knowledge of language as an interlocking set of subtheories
consisting of principles and parameters. Though the rules and principles themselves are
simple, the complexity of linguistic phenomena, as well as variation in human language,
"is traced to the interaction of partially independent subtheories, each with its own
abstract structure" (Chomsky, 1981a: 135).
I Syntactic component
[
Base 1) lexicon
2) Categorial component liJa D - s ucture
Transformational component ~ucture
PF LF (Horrocks, 1987:98)
2.4 Subtheories of GB Theory
2.4.0 Introduction
The network of subtheories or modules ofUG, which interact among themselves to yield
the core grammar of a particular language, are briefly described one by one below in
order to pave the way for the subsequent chapters.
21
2.4.1 X-bar Theory
According to X-bar theory 3, all types of phrase need two internal levels of structure. It
proposes that all phrases in all languages share a simple cell-like structure with two levels
to each phrase: one (X", X-double bar) consists of the head and possible specifiers; the
other (X', X-bar 4) consists ofthe head (X) and possible complements. Note that specifier
and complement are not themselves syntactic categories but functional labels for
positions in the structure that may be filled by actual syntactic categories such as NPs and
VPs.
XP (X")
Sp~X' ~
X Head
(lexical category)
complement
X-bar syntax makes statements about phrase structure that are true for all phrases rather
than for one rule or one phrase type. It relies on two main structural relationships within
the phrase: one is the link between the head and the complements that are its sisters; the
3 The fundamentals ofX-bar theory are outlined in Chomksy (1973) and later refined in Emonds (1976), Jackendoff(1977), Chomsky (1981) and Stowell (1981). The latter and later Chomsky (1986b) were influential in extending the principles ofX-bar Theory to non-lexical categories, complementing earlier work by Bresnan (1970) and Fassi Fehri (1980) in relation to COMP/S' (Ouhalla, 1994:12).
4 Bars are currently symbolized as', i.e. N' and V'; a former convention depicts bars as lines above the letter, i.e. N, presumably abandoned because of its incompatibility with a conventional keyboard. The bar symbol then gives X-bar theory its name (cf. Cook, 1994).
22
other is between the specifier and the head. The order of these elements is parameterized
and hence may vary from one language to another.
Originally, the X-bar schema ranged mostly over selected lexical categories, such as
Noun ([+N, -V]), Adjective ([+N, +V]), and Preposition/Postposition ([-N, -V]). To
obtain conformity in the structural representation of all categories, Chomsky (Chomsky,
1986b) extends the X-bar schema in order to allow certain abstract and non-lexical
categories to function as head as well.
The X-bar framework is extended to the functional categories I(n:flection) and
C( omplementizer). The clausal categories previously labeled as S and S' are now labeled
IP5 and CP6 (I= INFL; C = Complementizer), as in (2) and (3):
(2) a. S = I"= [NP [I' [vP V ... ]]]
b. S'= C"= [ ... [C' C I"]]
(Chomsky, 198lb:3)
(3) (=S')
5 I represents the category of INFL/inflection. I' (1-bar) is an intermediate projection headed by I; and IP (inflection phrase) is a maximal projection headed by I.
6 CP stands for 'Complementizer phrase'. 'Complementizer' is used to denote the presubject position in clauses ('the complementizer position') which is typically occupied by a complementizer like that/if/for, but which can also be occupied by an inverted auxiliary in sentences such as Can you help?, where can is taken to occupy the complementizer position in the clause.
23
C IP (=S)
~ NP I'
~ I VP
/ Lexical phrases such as the VP, and NP are built around lexical heads; functional phrases
such as IP and CP are built around functional heads, which may contain lexical material
such as morphological endings but are not required to contain lexical material (Cook &
Newson, 1996: 150).
Under the extended X-bar theory, I is assumed to be the position where elements of AGR
(agreement), TNS (tense), ASP (pect) and NEG (ative) originate, as shown in (4):
(4) I
AGR TNS ASP NEG
However, as Ouhalla (1991).has pointed out, the analysis in (4) is problematic.
First of all, NEG in languages like English should not be treated as an inflectional
element; and the fact that I is multi-headed "violates the isomorphic constraint on
categories and their projections implicit in X-bar theory" (Ouhalla, 1991: 56). Besides,
the elements under the node of I can differ from one language group to another and if all
the elements are assumed to be under one single node (i.e. I node), it would be impossible
to predict variations in word order among languages (cf Pollock, 1989).
24
In order to address this question, Ouhalla (1991) suggests that all the elements previously
assumed under the node of I should be regarded as independent syntactic categories.
Thus, (4) can be treated as (5). Following the suggestions by Ouhalla, we will assume in
this research that AGR, TNS, NEG, and ASP are independent syntactic categories.
However, as I is used in most of the literature of Principles and Parameters approach, we
will continue treating the I node as the head of IP unless elements of AGR, TNS, NEG
and ASP are dealt with specifically.
(5) AGRP
~ SPEC AGR'
~ AGR' TNSP
~GP NE~P
ASP ( cf. Ouhalla, 1991)
2.4.2 C-command and M-command
C-conunand (constituent-command) and m-command are two very important concepts in
GB theory which are crucial in a number of the subsystems. C-command is a structural
relation between two consitituents. Chomsky (1986b: 8) describes c-command as
follows:
(6). a c-commands 13 iff a does not dominate 13 and every y [XP] that dominates a
dominates 13.
(Chomsky, 1986b: 8)
25
That is to say, a category a. c-commands another category j3 if and only if the first
branching node dominating a. also dominates j3.7
7 There are a number of definitions of C-command and Government in the literature, which share the same basic ideas, although differing as to details. To see how this definition works, consider the following tree:
s
I"' I
NP
I I
v I
"' VP
I "'
"' NP
I ~ y
The first branching node dominating the NP a is the S. This S also dominates the V ~and the NP y. Therefore, a c-commands both ~ and y. ~ and y, on the other hand, do not c-command a, since the first branching node dominating them is VP, which does not dominate a. They do, however, c-command each other. Consider another VP structure (1): VP
I V' I V'
l"'--1 "' "' VNP I I "' pp
I I I I
I P'
I I I I l"'-
p NP
I I I I Quit his job in the autumn
In this example, V c-commands the NP his job but not the PP in the autumn. On the other hand, V mcommands both the NP his job, the PP in the autumn and also the preposition in and the NP the autumn. P c-commands the NP the autumn. P c-commands the NP the autumn, and P also m-commands the NP the autumn. However, P does not c-command V: P', the first branching node dominating P, does not dominate V. P does not m-commands V either: there is a maximal projection PP which dominates P and does not dominate V.
In the following structure (as in example (2)), V c-commands PP (unlike in (1)), and it also m-commands the PP, and head P and the NP inside the PP. The difference between V and PP is identical to that in (1).
26
Following Aoun and Sportiche (1983), Chomsky proposes another notion called m-
commanding (1986b: 8). The definition ofm-command is that ify in (6) is restricted to
maximal projections, then a is said to m-command 13 under these circumstances. Them-
commanding element must be in a position which is higher than that or in the same
maximal projection as the elements it m-commands.
C-command is a structural configuration used in various parts of principles and
parameters theory to express a relationship between elements with one element being
"superior to" but not dominating other elements which are c-commanded by it. When it is
defined in terms of maximal projections, it may be termed "m-command".
2.4.3 Government
VP
I" I "'-V'
V'
I v I
I I I
Leave
"'-" " pp
I P'
I" I "-
P NP
I I I I
in the autumn
Government refers to a particular syntactic relationship of high abstraction
between a "governor" and an element that it governs. Government "plays
a central unifying role throughout the system" (Chomsky, 1982: 7). As we
will find out later, the Empty Category Principle crucially depends on
government; Case is assigned under the restrictions of government
(Chomsky, 1981a). Government, or more precisely the absence of
government, is crucially involved in the distribution of null pronouns.
The relations of government may be defined as follows (Chomsky, 1986b:
8; cf. Aoun & Sportiche, 1983):
(7) Government
a governs 13 iff a c-commands 13, and every barrier for 13
dominates a.
Where a = X0, in the sense of X -bar theory, and for simplicity barrier
may be taken to mean a maximal projection. As a rule, Case and 9-role
assignment occur under government. In other words, government can be
defined again as:
(8) a governs 13 if and only if
i) a is a governor (e.g. N, V, P, A, etc); and
ii) a m-commands 13; and
iii) no barrier intervenes between a and 13.
27
Maximal projections are barriers to government.
Governors are heads.
(Chomsky, 1986b: 8; Cf. Haegeman, 1994:137)
What, then, is a barrier? Chomsky believes that certain maximal
projections are barriers themselves, intrinsically, others become barriers
by inheritance. According to Chomsky, lexical categories are L-markers 8
and the L-marking determines what consititutes a barrier. However,
Chomsky makes a distinction between a blocking category (BC) and a
barrier. A maximal projection which is not L-marked is called aBC; and
not every BC is a barrier.
Chomsky defines "blocking category (BC)" as in (9) and "barrier" as in
(10):
(9). "y is aBC for 13 iffy is not L-marked andy dominates 13"
(10). y is a barrier for 13 iff(a) and (b):
a. y immediately dominates 8, 8 a 13 for 13;
b. y is a 13 for 13, y :;t: IP.
We noticed that in (lOa), immediately dominate is a relation between
maximal projections. Thus, non-L-marked maximal projections are
barriers (except for IP), but L-marked maximal projections are not. In fact,
28
a barrier is any maximal projection, which blocks a government relation
between two elements.
2.4.4 Binding Theory
Binding Theory specifies the structural relations that determine the
distribution and interpretation of three major nominal classes: anaphors
(reciprocals and reflexives), pronominals (pronouns) and referential
expressions (r-expressions, i.e. names or other referential lexical NPs) 9•
Binding Theory constrains the relationships between various kinds of
noun phrases. There is a locality restriction on the binding relationships
concerning anaphors and pronominals and the relevant local domain is
defined by the inclusion of the governor of pronouns. The local domain of
the pronoun is known as its governing category.
(11) Governing Category:
(i) a is a governing category for 13 if and only if a is the
minimal category containing 13, a governor of 13, and a
SUBJECT accessible to 13.
8 In order to refer to the special relation established between a lexical item and the complement which it governs and theta-marks, Chomsky (1986b: 14) introduces the term L-marking.
9 Non-argument NPs, such as expletive it in it seems that the end is nigh, or existential there in there is a fimdamentaljlaw in the government's economic policy, are outside the system.
29
(ii) A SUBJECT is AGR or the subject of an infinitive, a
gerund, an NP or a clause.
(iii) a is accessible to f3 if and only if f3 is in the c-commanding
domain of a and assignment to f3 of the index of a would
not violate * [y ... o], where y and o bear the same index.
The three principles of the binding theory are as follows:
(12) a. An anaphor is bound in its governing category.
b. A pronominal is free in its governing category.
c. An R-expression is free.
(Chomsky, 1982:188)
Central to all principles ofbindingtheory are c-command and governing
category. An element is said to bind another element if it c-commands it
and has the same index;
(13) "a binds f3 if a c-commands f3 and is co-indexed with f3"
(Chomsky, 1986a:164).
The importance of governing category lies in the fact that the lexical
domain needs to be defined in terms of governing category.
Take the following sentence as an example.
(14) a. Harry i shot himself i·
b.* Harry i shot him ilj·
30
Binding theory describes when different expressions may be co-indexed,
such as when him or himself may refer to the same person Harry. In the
above example, nouns such as Harry are classed as referring expressions
(r-expressions) in that their reference is necessarily to something in the
discourse outside the sentence rather than to some other element in the
sentence. The word himselfbelongs to the class of anaphors, made up of
subgroups such as reflexives -himself, herself, themselves, and so on, and
reciprocals such as each other. The word him belongs to the class of
pronominals among which are also numbered she, him, them, and so on.
The crucial difference between anaphors, pronominals, and referring
expressions is the area of the sentence within which they can be bound.
Anaphors are "bound" (which refers to the conjunction of c.:commanding
and co-indexing) within the clause or local domain. Where an anaphor is
unbound (i.e. has no suitable antecedent to bind it), the resulting sentence
is ungrammatical. Pronominals may be bound by NPs in other clauses or
be free to take their reference outside the sentence; referring expressions
are always free (the term free simply means not bound). Thus, binding can
be defined as:
(15). a binds r3 if and only ifthat a c-commands r3 and that a and r3 are
co-indexed.
However, the correct definition of Governing Category (GC) is somewhat
controversiaL A number of alternative ideas have been put forward.
Wexler and Manzini (1987) suggest that the definition ofGC may differ
31
from one lexical item to another not only across languages but also inside
the same language. According to Wexler and Manzini (1987:47),
"parameterization is essentially lexical; parameters are not properties of
principles but of individual lexical items in the lexicon". Huang (1982)
believes that the GC for anaphors and that for pronominals are different;
while the former requires a SUBJECT which is accessible, the latter does
not.
Huang (C. -T. J. Huang, 1982) modifies the notion ofthe GC as in (16):
(16). Governing category:
a is a governing category for 13 if and only if a is the minimal
category containing 13 and a SUBJECT which, if 13 is an anaphor, is
accessible to 13.
According to him, for pronouns, a SUBEJCT does not have to be
accessible and for anaphors, there is no modification: a SUBJECT has to
be accessible.
2.4.5 8-Theory (Theta-Theory)
8-Theory (Theta-Theory) deals with the assignment ofthematic roles (or
0-roles 10 for short, i.e. semantic roles, such as Patient (or theme), Agent,
32
33
Beneficiary, Goal and so on) to elements of the structure ofthe sentence.
9-roles express certain meaning relationships between elements, a type of
meaning directly relevant to the LF component and indirectly relevant to
the semantic component. The Principles and Parameters approach handles
9-roles in the 9-theory.
The lexical entry for any predicate 1 1 includes the 9-roles that its
arguments bear, such as Agent, Patient, etc. This is represented as a 9-grid.
For example, the lexical entry of the verb cook is in part:
(17). Cook <Agent, Patient>
The 9-grid <Agent, Patient> tells us that the verb cook takes two
arguments, an Agent and a Patient. The 9-grid of a predicate determines
which 9-roles its arguments bear.
The fundamental principle of 9-theory is the 9-Criterion, which requires
that "each argument bears one and only one 9-role" (Chomsky, 1981:36);
10 9-roles include Agent (the person or thing carrying out the action, as in "Mary bought a book"), Patient (the person or thing affected by the action, as in "Mary bought a book"), Goal (the recipient of the object of the action, as in "Mary gave Peter a book"), and the Theme (thing which is moved by the action, as in "Mary gave Peter a booR'), etc (See Horrocks, 1987: Chapter 2; Radford, 1988: Chapter 7).
1 1 Elements, which say something about entities or their relationships, are know as predicates and the entities concerned are known as arguments. A predicate then expresses the meaning relationship between arguments. Knowing the argument structure of a predicate means knowing not just how many arguments are involved but also what kind (Cook, 1996:160).
and each 6-role is assigned to one and only one argument. The 6-Criterion
ensures that the 6-roles listed in a predicate's lexical entry are assigned
each to a single argument and that no argument appears without bearing a
single 6-role.
6-Theory is responsible for determining the structural representation of
thematic relations between constituents. It is concerned with the
assignment of thematic roles to sentential consitituents, such as Agent,
Patient, Goal, etc. It is assumed that these are assigned to the complements
of lexical items as a lexical property. It is also assumed that the majority
of verbs "6-mark"12 the subject position of sentences containing them.
A lexical entry can assign a 6-role only to those positions that conform to
particular grammatical configurations of subject or object. An argument
position (A-position) is a Grammatical Function (GF) position such as
subjects or objects at D-structure. 6-roles are only assigned to A-positions.
It should be pointed out that 6-roles such as Patient should not be confused
with GFs; a patient may be assigned anywhere in the appropriate
projection, not just to the GF object position.
12 One part of0-theory concerns the process of how 0-roles get from the lexical entry of a predicate to the arguments that bear them. 9-roles are transferred from a predicate to its arguments by a process known as ().marking.
34
35
The subject GF is also different from the other A-positions in that a 8-role
may not necessarily be assigned to it. In English, we often use the nominal
expletive there and it in subject position which are not assigned any
external 8-roles. Verbs such as seem and appear do not assign external 8-
roles. Thus, there are A-positions 13, to which 8-roles have been assigned,
which are 8-marked and there are A-positions, to which 8-roles have not
been assigned, which are not 8-marked. All 8-roles are assigned to A-
positions but not all A-positions have 8-roles.
2.4.6 Movement Theory
Just as the gradual elimination ofthe peculiarities of individual rules has
resulted from the development of X -bar theory, so many separate rules
previously known as transformations have been subsumed into the general
principle of movement. UG is seen once again as limiting the ways in
which movement can take place, narrowing down the set of possible
human languages. Therefore, it is best to assume that there may be no
restrictions on movement at all. Any part of the sentence could move
anywhere.
13 A-position refers a position, which can be occupied by an argument, but not by a nonargument expression (e.g. not by an adjunct). These include Spec ofVP, Spec ofiP, and complement ofV and P. On the other hand, non-A-positions refer to the structural positions in which arguments are typically not found; these include Spec of CP and complement of I.
The general principle is called Move a, where alpha is a variable which
ranges over all categories. Move a can be simply defined as in (18):
(18). Move a
Move any category anywhere.
The theory of movement explores the restrictions that human languages
actually place on movement. It is a property ofUG that only ce1iain
elements may be moved, that they may only be moved to certain locations,
and that they may not move more than a certain distance. That is to say,
"Move a" is tightly constrained. Some of these restrictions apply
uniformly to all human languages, some are parameterized and vary
within limits from one language to another.
It should be noted that movement is a way of expressing a relationship
between one form and another "as if things moved". Another way of
expressing the relationship of D-structure to S-structure is via the concept
of a chain that records the links that make up this relationship. The
relationship between D-structure and S-structure of sentence (19) can be
seen as incorporating two chains: (where, t1), and (is, tz).
(19) Where1 isz the hospital t2 t1?
36
37
As Chomsky and Lasnik (1993: 522) have pointed out, "movement of an
element a always leaves a trace 14 and, in the simplest case, forms a chain
(a, t) where a, the head of the chain, is the moved element and t is its
trace."
There are three main types of movement:(!) Head-movement, which is the
movement of auxiliaries from I to C; (ii) Wh-movement 15, which is the
movement ofwh-constituents to the specifier ofCP; and (iii) NP-
movement, which is the movement associated with passive in which an
NP is moved to an empty subject position (the empty subject position is
often indicated by the symbol e for "empty").
There is a constraint specifically on head movement, called the Head
Movement Constraint (HMC). Following Chomsky (1986b), Ouhalla
(1991) defines the HMC as in (20):
(20). The Head Movement Constraint
A head category can only move to the head position immediately
preceding it.
14 An empty category, which encodes the base-position of a moved consitituent, is referred to as a trace and will be indicated by t (Haegeman, 1994:309).
15 Interrogative constituents such as whom are called wh-consitituents. Movement of question words is referred to as wh-movement.
38
Wh-movement concerns the movement ofwh-phrases. A wh-phrase is one
that contains a wh-word; these include the item how as well as words
starting with wh such as who and which.
Wh-movement can move the wh-phrase into a non-A-position (A'-
position). The A' -position to which the wh-phrase is moved should be an
empty position in the D-structure at the beginning of the sentence. The
specific position immediately dominated by CP now proves its usefulness.
The reason is that this position is empty except when a phrase such as a
wh-phrase moves into it. Wh-movement takes a wh-phrase from an A-
position and moves it to the specifier of CP, leaving a trace t behind it.
This type oftrace is consequently a wh-trace, also known as a variable.
As the specifier of CP is not an A-position, 9-roles cannot be assigned to it
at first; wh-movement goes from an A-position, where it gets its 9-role, to
a non-A-position, where it does not pick up an extra 9-role. This does not
violate the 9-criterion since the specifier of CP does not carry any
competing 9-roles.
Another movement rule generally known as ''topicalization"16 is a process
by which a consitituent is made into the topic of a sentence by being
16 Topicalization is a movement or other operation or structure that determines a Topic-Comment Structure. For example: in English, the movement that places the element beans in initial position in the sentence Beans, he hates. Topic-Comment Structure is a sentence structure in which the major syntactic division in
moved into a more prominent position at the front of the sentence. It
involves the movement of a phrase from an A-position into the specifier of
CP, an A' -position. The following sentences are examples of
topicalization (Emonds, 1976:31; Radford, 1997:273):
(21 ). a. [NP Such behavior we cannot tolerate _ in a civilized society.]·
b. [NP Each part John examined __ carefully.]
c. [NP Our daughters we are proud of __ .]
The bracketed NPs originate in the position marked __ , and are
preposed into the italicized position by the rule oftopicalization. Since the
topicalization involves movement of a phrase into the specifier of CP, it
excludes any preposed wh-phrase in the specifier of CP.
Wh-movement involves movement from an A-position to specifier of C, a
non-A-position. Questions such as "What did the cat kill?" involves wh-
movement of the wh-phrase what from the D-structure "the cat past kill
what" to the specifier position of CP, yielding the S-structure "what did
the cat kill t".
the sentence is between a topic and a comment on that topic rather than between a subject and a predicate. For example, in the sentence Beans, he hates, the element beans is the topic, and he hates is the comment.
39
40
Generally speaking, movement is defined as either substitution or
adjunction (Chomsky, 1986b). In transformational grammar, the additional
abstract structures are regarded as more basic and the ordinary structures
are seen as derived from them by various movement processes. For
example, (22) is derived from (23), and (24) from (25).
(22) s
~ (2~
NP s NP VP
~ ~ NP VP V NP
I v I I Stefan annoyed who
Who Stefan I
annoyed
(24). /---_ ~ NP VP NP VP
~ v VP
I ~ Stefan seemed to be a fool
I ~s e I ~
Seemed NP VP
stJfan~ To be a fool
(Examples from Borsley, 1991 :92-95)
There are two types of movement processes. In the derivation of (22) from (23), we have
an adjunction process. This makes the moved category the sister of some existing
category and the daughter of another instance of that category. It refers to the movement
of the xmax or X0 to another maximal or head category. The movement of the verb to I in
finite clauses is an instance of adjunction movement. The structures below (26a) and
(26b) illustrate the structures of adjunction.
41
(26) a. XP b. X
yp XP y X
In (26a) YP is the adjoined xmax, which is moved and adjoined to another xmas, that is,
XP. (26b) is the structure resulting from adjoining a moved X0 to another X0, that is X. It
is possibly subject to parametric variation whether the category is adjoined to the left or
the right of the host.
In the derivation of (24) from (25), we have a substitution process. This substitutes the
moved category for an existing empty category. The empty category is standardly
referred to as a "landing site" for movement.
Substitution processes are commonly known as structure-preserving (Emonds, 1970),
since unlike adjunction processes they do not change the structure of the sentence in any
way but just rearrange the lexical material.
42
2.4. 7 The Projection Principle (PP) and the Extended Projection Principle (EPP)
The principles and parameters approach emphasizes the words of the language, i.e. the
lexical items of the mental lexicon. The lexical entry is said to "project" onto the syntax,
or the structure of the sentence; the lexical specification of the word ensures that the
system has a particular form. This can be expressed in a central principle known as the
"Projection Principle".
Lexical information is syntactically represented. "Representations at each syntactic level
(i.e. LF, D- and S-structure) are projected from the lexicon, in that they observe the
subcategorizationprojection oflexical items" (Chomsky, 1981a:29).
Meanwhile, not only must lexical properties of words be projected in the system, but in
addition, regardless of their argument structure, sentences must have subjects. This
requirement is known as the Extended Projection Principle (EPP).
(27). Extended Projection Principle (EPP)
S~NP-AUX-VP
This requirement is not one that is specific to individual lexical items, but it is a general
grammatical property of all sentences.
43
2.4.8 Bounding Theory
"Bounding theory poses locality conditions on certain processes and related items"
(Chomsky, 1981 :5), so that movement is prohibited if too many "bounding nodes"
intervene between the starting point and the finishing point of each movement. Bounding
theory defines the boundaries for movement and thus determines how far an element can
be moved.
Chomsky (Chomsky, 1977) proposes a principle called the "Subjacency Principle",
which states that a trace may not be separated from its antecedent by more than one
"bounding node"; in other words, no element may be moved across more than one
bounding node.
(28). Subjacency Condition
Movement cannot cross more than one bounding node in a single step, where
bounding nodes are IP and NP (Haegeman, 1994: 402; cf. Chomsky, 1977;
Lasnik & Uriagereka, 1988).
Subjacency is a condition on movement (Move a) rather than a condition on
representations derived by movement. Subjacency becomes a locality condition on the
representation of traces in relation to their antecedent (the moved category), to the effect
that a given trace cannot be separated from its antecedent by more than one bounding
node. According to Rizzi (Rizzi, 1982), the theory assumes that no more than one
44
. bounding node can be crossed, but languages vary in their choice of what is to count as a
bounding node. English, for example, has S and NP as the bounding nodes, whereas
Italian chooses NP and S'. Generally speaking, movement across more than one
bounding node results in rapid deterioration of grammaticality.
2.4.9 Case Theory
Case theory deals with the principles of case assignment to constituents. Although case 17
is an overt property only of pronominal NPs in English, Chomsky assumes that all NPs
with lexical content are assigned (abstract) case 18• According to Chomsky (1986a: 74),
"in some languages, Case is morphologically realized, in others not, but we assume that it
is assigned in a uniform way whether morphologically realized or not."
The basic idea is that Case is assigned under government, and is assigned in the S-
structure. It is taken as structural Case because it is assigned according to the
Grammatical Function configurations of the sentence. In each ofthese Grammatical
Functions, a particular element acts as the "Case assigner". Generally, it is thought that
Verbs assign objective/accusative Case, Prepositions oblique Case and AGR or TNS
nominative/subjective Case. Sentences without AGR and TNS do not have subjects in the
nominative Case.
17 The different case forms of a pronoun are the different forms which the pronoun has in different sentence positions. Traditionally speaking, English has three cases- nominative, objective/accusative and genitive. Personal pronouns typically inflect overtly for all three cases, whereas noun expressions inflect only for genitive case.
45
As we will find out in later chapters, AGR and TNS are absent in Chinese. Therefore, the
subject position in Chinese is ungoverned and Case cannot be assigned structurally in this
position. Chomsky (Chomsky, 1981) suggests that Case can also be assigned inherently.
This is pruiicularly true in some non-configurational languages. It is assumed that in
Chinese, lexically realized NPs in subject position are inherently Case-marked.
One of the most important principles of Case theory is the Case filter, which states that
"every phonetically overt NP must be assigned abstract Case" (Chomsky, 1986a: 74).
That is to say, any S-structure that contains an NP with lexical content but no Case is
ungrrunmatical. The Case filter excludes overt NPs, which have either no Case or more
than one Case. This has significant consequences for the application of Move a, forcing
it to apply in some instances and making its application impossible in others.
2.4.10 Empty categories and the Empty Category Principle
In the Principles and Parameters approach, four types of empty categories have been
identified. As in the following examples, they are: NP-trace (the trace of a moved NP);
Wh-trace (the varia,ble/trace left by a moved Wh-phrase), PRO (the abstract subject of
infinitival clause) and pro (the phonologically null subject of finite sentences, which is
found in pro-drop languages).
18 Abstract Case is usually distinguished from case as an overt inflectional category by the use of an initial capital, and this convention is used henceforth.
Examples of empty categories:
a. NP-trace: John is likely t to succeed.
b. PRO: John tried PROto win.
c. Pro: pro habla. (Spanish)
d. Wh-trace/variable: Who did he see t?
PRO
PRO (read "big PRO") is an empty pronoun assumed within Principles and Parameters
Theory. PRO is the phonologically null NP occupying the subject position of an
infinitive clause, either controlled by a superordinate clause NP or assigned an arbitrary
reading. PRO is a base-generated element (meaning that it is present in D-structure
representations), and it is ungoverned and can never be Case-marked. This follows from
PRO's status as both anaphor and pronominal, which will be outlined below.
46
PRO acquires its referential content by means of a special indexing relation called
control, which is not subject to c-command. PRO is constrained by Control Theory,
which "determines the potential for the reference of the abstract pronominal element
PRO" (Chomsky, 1981 :6). PRO can take as its antecedent an NP that may be subject
(29a) or object (29b) in the matrix sentences. This is known as "obligatory control". With
obligatory control, PRO must be bound within its governing category, i.e. be an anaphor.
It is in this sense that PRO is held to be anaphoric.
47
PRO is also pronominal because it may have indefinite/independent arbitrary reference
(29c ). This is known as "arbitrary control". Thus, on the basis of binding theory, PRO has
the features ofboth [+anaphor] and [+pronominal].
(29). a. Johni wanted [PROi to sell the house].
b. John persuaded Billi [PROi to sell the house].
c. It is illegal [PROarb to travel without a valid ticket].
In another example (30), PRO is both anaphoric since it is coindexed with the higher
clause subject, and pronominal in that it acts as the subject of the lower clause.
(30) I like [PRO to play baseball].
In summary, PRO is characterized by the feature matrix [+anaphor, +pronominal], which
must not be governed. Its content is determined by control theory.
pro
pro (read "small/little pro", as opposed to the big PRO) is an empty pronoun assumed
within Principles and Parameters theory. It is the phonologically null subject of finite
sentences, which is found in pro-drop languages marked either by agreement as in
Spanish and Italian (31 ), or by some other means as in the case of Chinese and
Vietnamese (32) (cf. Taraldsen, 1978; Chomsky, 1982; Platt, 1993; Haegeman, 1994).
(31) pro vengo a las siete.
(I) am coming at seven.
(32) pro co the di hoc duoc.
(One) can go to school.
(Examples from Platt, 1993: 1 07)
48
Unlike PRO, pro is not controlled and has the normal pronominal capacity for
independent reference. In terms of binding, it is free in its governing category and subject
to Principle B of the binding theory. Hence, pro is the empty category counterpart of the
pronominal NP and is [+pronominal].
Languages which allow a pronominal subject to be left unexpressed are called "pro-drop
languages/null-subject languages", because they "drop" the subject pronoun. Spanish,
Italian and Chinese are pro-drop languages, whereas French and English are not. This
cross-linguistic variation is referred to as the "pro-drop parameter".
While the pattern of occurrence of PRO is more or less uniform across various languages,
whether a language can have a phonetically null subject in tensed clauses (i.e. pro) is
determined by the null-subject parameter.
49
NP-movement
NP-movement is involved in the derivation of passives and raising constructions, and has
the distinctive property of moving an NP into an empty subject position, which is Case
marked since the head of the chain is Case-marked.
NP-movement is triggered not only by passive verbs but also by so-called raising verbs
and by raising adjectives. Consider sentence (33a). (33b) is the S-structure representation
of (33a): Poirot is a derived subject.
(33). a. Poirot seems to have destroyed the evidence.
b. [rP Poiroti [r -s [ vP seem [rP ei to have destroyed the evidence]]]].
Because the subject of the lower clause is raised out of the clause and moved into a
higher clause, this movement is sometimes referred to as "NP-raising" or raising. Verbs
such as seem which induce raising are called "raising verbs".
NP-movement takes the NP from an A-position, leaving an NP trace behind. The moved
NP is co-indexed with the trace within the same governing category. Hence, Principle A
of binding theory suggests that the empty category NP-trace behaves like the overt NP
category ofanaphor; it has the feature of [+anaphor].
50
Wh-movement
Within Principles and Parameters theory, the ordinary obvious structure of a wh-question
will be derived from a more abstract structure by a movement process. It is normally
referred to as "wh-movement". In wh-movement, a wh-phrase is moved to the specifier
position of CP, a non-A- or A' -position, leaving a trace behind, which is co-indexed with
the wh-phrase. Consequently, the wh-trace must refer outside IP and be bound by the
specifier position of CP.
By Principle C of the binding theory, wh-trace is an R-expression like a full NP and has
the feature of[-anaphor] and [-pronominal]. Principle C ofthe binding theory requires
that no r-expression, including variables/wh-traces, should be bound anywhere by a c
commanding NP in an A-position. This does not preclude the possibility of A'-binding of
variables, i.e. the binding of variables by a c-commanding element in the specifier
position of CP.
NP-trace and wh-trace share the feature [-pronominal]. However, there are several
identified properties which distinguish NP-traces and wh-traces. These concern the
moved element, its landing site and its extraction site19•
19 NP-trace and wh-trace: survey of properties (See Haegeman, 1994: 440):
51
One point we should note is that all these empty NPs except PRO have overt
counterparts. The overt counterparts of NP-trace are reflexives, the overt counterparts of
pro are ordinary pronouns, and the overt counterparts of wh-trace are non-pronominal
NPs. The relationship between overt and empty categories ofNP can be distinguished by
features of [+/-anaphor] and [+/-pronominal] as follows (Cook & Newson, 1996: 256):
Relationship between overt and empty categories
Overt empty anaphor Pronominal
Reflexive pronouns NP-trace + Personal pronouns pro + r-expressions wh-trace
PRO + +
The idea that traces are licensed by being governed by lexical heads is known as the
Empty Category Principle (ECP). The ECP requires that an empty category must be
NP-trace Wh-trace
Moved category NP XP (NP, PP, etc)
Landing site A-position by A'-position by subsititution Substitution or adjunction NP-position [Spec, CP] or adjoined position
Properties of antecedent Case Yes No Chain A-chain A'-chain
Properties oftrace Features [+anaphor] [-anaphor]
[-pronominal] [-pronominal] Binding theory A c Theta theory Yes Yes Case No Yes (when target= NP) Governed Yes Yes
52
properly governed: a properly governs 13 iff (a) a governs 13 and (b) a is lexical or an
antecedent. Therefore, there are two ways in which a trace can be properly governed. The
first one can be called lexical/head government, and the second antecedent government. It
should be noted, however, that the Empty Category Principle is only about NP-traces and
wh-traces, rather than all empty categories. PRO and pro are not subject to the
requirement that they be properly governed.
The sentences in (34) are typical examples illustrating the effects ofECP, which is often
called "that-trace" phenomenon.
(34). a. Whoi do you think [cP ti [IP ti saw Bill]].
b. *Whoi do you think [cP thati [IP ti saw Bill]].
c. Whoi do you think [cP that [IP John saw ti]].
In (34a), the trace in the subject position of the embedded sentence is not lexically
governed. However, it is locally antecedent-governed by another trace left at the
landings-site by the wh-element, who. In (34b), where an overt complementizer, that, is
immediately followed by a trace, the trace is not locally antecedent-governed. Thus,
ungrammaticality results. However, an overt complementizer, that, may co-occur with a
trace in object position without ungrammaticality resulting, as in (34c). This is because
the trace in object position is lexically governed by the verb. Thus, the ECP is satisfied.
53
2.4.11 UG and Language Acquisition
Apart from a theory of grammar, UG is also a theory of acquisition. It is assumed that
languages are constrained in form by rules, principles and parameters. Researchers
working on UG in second language acquisition accept the premise that first language
acquisition is mediated by a system of innate principles and parameters, and that children
will apply "biologically determined principles and parameters to the structure-dependent
experience of primary language data" (Lust, 1986: 8). The motivation for postulating a
construct like UG is that adult linguistic competence is too complex to have been derived
solely from the linguistic input (Hornstein & Lightfoot, 1981; Eckman, 1995).
The claim that many properties of language are not available directly from input is known
variously as "the poverty ofthe stimulus problem", "Plato's problem", or "the logical
problem oflanguage acquisition" (Baker, 1979; Hornstein & Lightfoot, 1981; Chomsky,
1987; 1988). The general approach to acquisition from this perspective is lmown as the
"learnability approach" (Pinker, 1989).
The arguments for innateness in L1 acquisition are based on the following facts. First, the
evidence that the child has to rely on in acquiring his or her native language (i.e. the
speech of those around him and her) is degenerate (i.e. incomplete and imperfect). For
example, many of the sentences native speakers have intuitions about, both grammatical
and ungrammatical, have never been heard by them before. In consequence, the
extremely rich and complex linguistic rules which the child eventually acquires are
54
underdetermined by this evidence, that is, the evidence on its own is not sufficient for the
child to be able to work out the relevant rules. Native speakers are not limited to
producing only those strings, which have been heard in the input. The input is said to
underdetermine the knowledge oflanguage which adults attain. So, a question that should
be answered is: where does this knowledge about what constitutes a possible structure
come from, if it is not from input alone? It is the Principles and Parameters theory that
offers the only logical possibility that, the knowledge of grammar assumed to construct
the U G is innateness.
UG guides the way the child will interpret and organize the experience. As White has
pointed out, UG "consists of principles which constrain the form and functioning of
grammars. It gives the child advance knowledge of many abstract and complex properties
of language, so that these do not have to be learned solely on the basis oflinguistic input
or by means of general learning strategies. In language acquisition, then, there is an
interaction between the innate UG and the linguistic input from the language being
acquired." (White, 1989: 5) It is noted that all adult native speakers come up with
comparable Ll competence and all children acquire their native language in a remarkably
short period of time although they have been exposed to different kinds oflinguistic data.
Chomsky (1986a) reasons that whatever knowledge native speakers have about their
language, which they cannot have acquired through experience, must be attributable to
innate knowledge; and that whatever is innate must therefore be universal. The theories
of acquisition that presuppose UG claim that input alone is insufficient to account for
55
acquisition. First language (L 1) acquisition is assumed to proceed on the basis of
naturalistic positive evidence interacting with innate principles and parameters ofUG.
The positive evidences or input data "trigger" properties ofUG, causing UG parameters
to be set without the necessity for learning to take place. According to Chomsky
(Chomsky, 1988: 134), "Language learning, then, is the process of determining the values
of the parameters left unspecified by universal grammar."
So, the mastery of a language is not really the result of learning. Being equipped with UG
with its parameters and exposed to a particular language, such as Chinese or English, the
child will construct the grammar of the language he or she is exposed to. That is to say,
through an l-language20 theory, UG has a place for experience in language learning. If
this were not so, all children would end up speaking the same language. "The
environment determines the way the parameters of universal grammar are set, yielding
different languages" (Chomsky, 1988: 134). Thus, a child born and brought up in an
Arabic-speaking environment can later speak Arabic. For this reason, the term "learning"
is often replaced by the term "acquisition".
On the other side, however, although it is generally agreed that language is acquired with
the help of the biologically endowed knowledge, i.e. a set of abstract principles and
parameters, it is still quite controversial what elements parameters are associated with
and how they are "fixed" or "set". One traditional view is that parameters are associated
20 !-language is a term used by Chomsky (Chomsky, 1986a) to refer to language viewed as an element of the mind of a person who knows the language, acquired by the learner, and used by the speaker-hearer. It is used in contrast withE-language (Crystal, 1991: 170).
56
with principles, and the values of these parameters are fixed by experience attained in the
language learning process (Chomsky, 1981). However, Wexler & Manzini (Wexler &
Manzini, 1987: 47) claimed that "parameterization is essentially lexical"; parameters are
not properties of principles but of individual lexical items in the lexicon. This is the
called the "lexical parameterization hypothesis". And children set parameters by
acquiring lexical properties of a particular language (cf. Borer, 1983; Wexler & Manzini,
1987; Cook & Newson, 1996; Chomsky, 1991).
Recently, there has been another development. Ouhalla (1991) suggests that only
functional categories, such as determiners, complementizers and inflection, have
grammatical features such as number, person etc., and that only functional categories
have parameters. Parameters do not belong to principles but to functional categories. The
fixing of parameters means acquiring all the properties of functional categories (Radford,
1990; Tsimpli, 1991 ). This is called the "functional parameterization hypothesis", which
means that languages differ only in the properties they select for their functional
categories, and that lexical categories are universal and uniform across all languages.
Functional categories consititute a closed class and they c-select 21 the category of their
21 It is often assumed that lexical items have "c-selection" (category selection) properties. This is the ability to determine the type of complement that the lexical item may have. Traditional subcategories such as verbs can be seen as c-selecting either no complements, or a single NP complement, or two NP complements, respectively (Cook & Newson, 1996). C-selection is represented in the lexicon in terms of a subcategorization frame:
askL CP/NP] wonderLCP]
For example, verbs such as ask and wonder can both have a question clause as their complement: Mary asked [what the time was] Mary wondered [what the time was]
However, only ask can take a complement that is an NP: Mary asked [the time] *Mary wondered [the time]
57
complements. Unlike functional categories, lexical categories s-select 22 their
complements, i.e. they choose their complements according to the thematic roles they
assign (Ouhalla, 1991).
If it can be maintained that parametric variation among languages is restricted to
functional categories, we can say that children do not have to learn the properties of
lexical categories, which would be innate and universal. This would explain the fact that
children acquire lexical categories before functional categories.
2.4.12 Conclusion
SLA is a complex field whose focus is the attempt to understand the processes underlying
the learning of a L2. It involves a complex interaction between L1 and L2 and UG. And it
is their interaction that makes the study of SLA challenging. To meet this challenge, we
have to resort to various sophisticated tools. The subtheories and principles ofUG are a
useful tool for characterizing the properties of structures in the SLA learners'
interlanguage (I will discuss this notion in subsequent chapters) grammars. The view of
language acquisition in terms of parameter setting is the basis of current work in the
generative tradition. In this research, I take the "Principles and Parameters Theory" as the
22 The lexical entry for any predicate includes the 9-roles that its arguments bear. This is represented as a 9-grid, such that the Verb cook is in part:
cook <Agent, Patient> Thus the 9-grid of a predicate determines which 9-roles its arguments bear: only certain types of argument can bear certain types of9-role. And the predicate's ability to restrict the kind of arguments that accompany it is called "s-selection "(semantic selection).
theoretical linguistic framework for the study of the acquisition of Chinese and English
by Chinese-speaking learners and English-speaking learners.
58
59
Chapter 3 Universal Grammar and Second Language Acquisition
3. 1 Introduction
Recent developments in linguistic theory within the framework of Principles and
Parameters theory (Chomsky, 1981a, 1986a, 1986b, 1991) have exerted considerable
influence on the area of first language (L1) acquisition and second language acquisition
(SLA). Until the 1980s, most work in SLA was inextricably linked to language pedagogy
alone (Newmeyer & Weinberger, 1988). It was really not until the Principles and
Parameters approach, realized in Government and Binding Theory, began to be applied to
SLA that we see a truly theory-centered approach to the question ofSLA. One ofthe
impetuses for research within this Principles and Parameters framework has been the
well-known "logical problem" of language acquisition. The term, "the logical problem of
language acquisition", was coined by David Lightfoot (Horstein & Lightfoot, 1981). The
problem is to explain how children come to acquire the complex linguistic knowledge, or
properties of their native language in a relatively short period of time on the basis of input
that is assumed not to be sufficiently rich and precise and in the absence of crucial
negative evidence. This is a problem because the input vastly underdetermines the finally
achieved competence.
In order to deal with the "logical problem", a number of SLA researchers have been
trying to show that the principles and parameters ofUG constrain the range of the
hypotheses L2 learners apply to the learning of the L2 (Flynn, 1987, 1988, 1989, 1991,
1996; Liceras, 1988, 1989; White, 1988a, 1989, 1996). However, in spite of the great
60
effort made by L2 researchers to study SLA within aUG framework, there are still more
questions than answers in the field. A central question is: to what extent, if any, is adult
SLA constrained by the linguistic principles that determine L1 acquisition? What is the
role ofL1 in SLA? Is UG still available to adult L2 learners after they have mastered
their mother tongue? What is the relationship between language processing and language
acquisition? In the following sections, we will discuss the questions above and try to
arrive at a better understanding of these issues.
3.2 Interlanguage
The term 'interlanguage' (IL) was introduced by Selinker (1972) to refer to the structured
system which the L2learner constructs at any stage during the acquisition of the target
language (TL). TheIL hypothesis claims that learner languages are somehow different
from other language systems. The IL grammar is influenced by and independent of both
the L 1 and the L2 and has the features of each.
According to the Ontogeny Model ofSLA (O'Grady, Dobrovolsky, & Katamba, 1997),
there are two types of error in an IL grammar: transfer errors and developmental errors
(see Taylor, 1974; Zobl, 1980a, b). The former type of error reflects transfer from the Ll.
In contrast, developmental errors involve the same sort of mistakes that children make in
acquiring their L 1. The IL grammar is influenced by both the L 1 and L2, though the
proportion of influence is dependent on the learner's level of overall proficiency. In the
Ontogeny Model, advanced learners have low numbers of both transfer and
developmental errors. Not all learners, however, reach this advanced stage. It is common
61
in SLA for learners to reach a plateau in their development. When the interlanguage
grammar stops changing, it is said to have fossilized.
Adjemian (1976) argues that ILs are not only natural languages, but that their grammar is
peculiar in being permeable. He further develops the IL hypothesis and redefines the
central characteristics of ILs, such as systematicity, permeability1, and stability. The
notion of "systematicity" is based on the assumption that ILs are natural languages, and
that there exists an internal consistency in the rule and feature system which makes up the
IL. The internalized rules in the learners' IL grammars are systematic in ways that are
similar to grammars of natural human languages. By permeability, Adjemian (1976)
means that ILs should be viewed as linguistic systems, and as grammars: It is not the
"application" of transfer or overgeneralization that results in the speech forms produced
by the learner. Rules and forms that constitute the L2 learners' IL grammars at any stage
are susceptible to infiltration by both Ll and L2 rules and forms. This accounts for the
variability and instability of the leamers' ILs; the rules and forms in the TL are partially
acquired or improperly used and the rules and forms from the Lis permeate the emerging
IL grammars at various levels. In the following chapters, we will see some examples of
penneability in English speaking learners' IL grammar of Chinese and Chinese speaking
leamers' IL grammar of English.
Moreover, Adjemian claims that the notion of stability should be restricted to refer to
those parts of an IL system that have lost their permeability. Once permeability is lost,
1The notion of permeability refers to the degree to which rules from one system penetrate another rule system. In SLA, the IL can be permeated by rules from either the TL or the NL.
62
fossilization2 occurs, which prevents the L2 learners from achieving native-speaker
competence. The term 11fossilization11 has been used, if not invented, by Selinker (1972)
to refer to the process whereby certain 11linguistic items, rules, and sub-systems11 become
relatively permanently incorporated into the grammatical system of a L2 learner. That is
to say that, during the acquisition process, all learners tend to fossilize at least particular
forms or particular parts of their grammar; learners stop learning when their ILs still
contain some rules or forms different from those of the TL system. In this way, the
fossilized adult rule manifests itself as an error.
There is clear evidence that L2 learners generally fail to reach TL competence. However,
children do not fossilize before they reach the full L1 competence. This difference is one
of the most striking features that distinguish SLA from L 1 acquisition. One of the
research goals facing SLA researchers is to determine and 11discoverthe ways in which
fossilization may occur, what may be fossilized 11 (Adjemian, 1976: 315) and to explain
the reason why fossilization occurs in SLA. As we will find out in the following chapters,
some language features appears to manifest themselves as candidates for fossilization in
both the IL grammars of English speaking learners of Chinese and Chinese speaking
learners of English. We will try to offer some explanations as to why these language
features are susceptible to fossilization in the light oflearnability and parsability.
2 As defined by the Unabridged Random House Dictionary, "fossilization" refers to the process whereby "a linguistic form, feature, or rule, etc., becomes, permanently established in the IL ofL2 learner in a form that is deviant from the TL norm, and continue to appear in performance regardless of further exposure to the TL''.
63
3.3 The Access ofUG in SLA
Researchers working on UG in SLA now adopt the Principles and Parameters approach,
as realized in Government and Binding (GB) Theory (Chomsky, 198la, 1986a, 1986b),
"to characterize the native speakers' (NS) knowledge oflanguage, or linguistic
competence, and to explain how the acquisition of such competence is possible" (White,
1996: 85). It is argued that much of our linguistic competence stems from innate
knowledge, which takes the form of aUG, and that input alone is simply insufficient to
allow the child to attain full adult competence.
One ofthe major factors motivating aUG-based solution to the logical problem ofL1
acquisition is that, linguistic input to the child consists exclusively of so-called primary
linguistic data (PLD), or positive evidence (utterances in the input that children are
exposed to, or utterances in context) which assumedly, interact with innate principles and
parameters ofUG. The input data "trigger" properties ofUG and cause UG parameters to
be set without learning having to take place. Negative evidence, such as explicit
con·ection of deviant child utterances or information about ungrammaticality, is in
general neither available nor used, or plays a minimal role (White, 1996).
However, conditions for L1 acquisition differ from SLAin some respects, though the
research into the latter has been greatly influenced by the former. First of all, L2 learners
differ from L 1 learners in having the knowledge of another language available to them;
that is, the learners' L1 potentially plays a role in subsequent language learning (Flynn,
64
1987; Flynn & Martohardjono, 1993), although how much this affects them is much
debated.
The question of whether L2 learners, like L1learners, can have access to UG has been
the main topic of research among those interested in applying the Principles and
Parameters approach to SLA. It has been claimed that the L2 learner can get caught in
situations where negative evidence is necessary to avoid or correct a faulty understanding
ofthe L2 grammar (White, 1987b, 1996). In SLA situations, many L2learners, unlike L1
learners, may be exposed to negative evidence though whether this negative evidence
affects the SLA process or not is still controversial (White, 1991 a, 1992a, Schwartz &
Gubala Ryzak, 1992).
Adult L2learners also differ in that they have passed the so-called "critical period"3 for
language acquisition (Lenneberg, 1967). If there is a critical period, for instance, perhaps
that could explain failure to acquire uniformity in the level of competence attained by L2
learners. But, why is the failure not uniform? Once again, there are some researchers who
do not accept the critical period hypothesis at all (Birdsong, 1991; Flynn & Manuel,
1991; Scovel, 1988).
3 Lenneberg (1967) argues that the language acquisition device, like other biological functions, works successfully only when it is stimulated at the right time, a time which is referred to as the "critical period" or "sensitive period". The notion that there is a specific and limited time period of language acquisition is referred to as the critical period hypothesis (CPR). There are two versions of CPR. The strong version is that children must acquire their 11 by puberty or they will never be able to learn from subsequent exposure. The weak version is that language learning will be more difficult and incomplete after puberty (Lightbown & Spada, 1993).
65
According to White (1996: 90), an L2 acquisition theory can take as a working
hypothesis that L2learners do (or not) still have access to abstract principles in order to
establish the exact nature of L2 competence and to account for its acquisition. The L2
developmental question can also be addressed by looking at issues like parameter setting,
triggering data, learning principles, and so forth. Not surprisingly, work in L2 acquisition
done from the principles and parameters perspective has centered on the issue of the
availability ofUG.
There are various hypotheses regarding the availability of UG in adult SLA. White (1989:
48-49) has summarized the following possible positions for a role for UG in adult SLA:
i. UG is available and works exactly as it does in L1 acquisition u. UG is totally unavailable in L2 acquisition. iii. Access to UG is mediated via the Ll. There are actually two different
versions of this hypothesis: a. UG is inaccessible but any aspects of it available in the Ll can be used
in the L2. b. L2learners initially assume the Ll value ofUG parameters, but are
still able to tap UG. Hence, they can re-set to L2 parameter settings. iv. UG is available but does not work in identical fashion to L1 acquisition.
(White, 1989: 48-49)
White claims that the five positions can be reduced to two main ones: the hypothesis that
UG in some way or other plays a role in SLA (1, iiib and iv), versus the hypothesis that
UG is to all intents and purposes inaccessible (ii and iiia).
SLA researchers are still divided as to whether adult learners have access to the principles
and parameters ofUG in constructing their L2 grammars, and, if they do, what form that
access takes. Currently, no real consensus exists on the availability ofUG to L2 learners,
though generally speaking there are three different hypotheses regarding the availability
ofUG in adult SLA: "direct access to UG", "indirect access to UG" and "no access to
UG" (Cf. Cook, 1993; White, 1989, 1996).
3.3.1 Direct access to UG
66
According to the "direct access" (Cook, 1988, 1994; Cook & Newson, 1996) to UG
position, adult L2 grammars are constrained by UG principles in the same way as child
L1 grammars are. This position is exemplified in the work ofFlynn (1983, 1987, 1989,
1991, 1993), Flynn & Martohardjono (1991, 1992), Epstein et al (1996), Liceras (1989),
Finer & Broselow (1986), Lakshmannan (1992), Mazurkewich (1984), White (1988b,
1989b), and Schwartz (1993), among many others. This view states that all principles and
parameter values available to the child L1learner are also available to the adult L2
learner, although differences may certainly arise due to the different contexts (biological,
linguistic, social) within which child versus adult language learning takes place. It holds
the view that L2 speakers have parallel competence in L1 and L2, two instantiations of
UG. Learners should not commit "impossible" errors in L2. L2learners may employ the
principles ofUG and set the parameters without any reference to their L1 values, i.e. to
set parameters to UG-sanctioned values unlike those ofL1.
3.3.2 Indirect access to UG
In this position, UG is not available to SLA except through the L1 grammar (Cook,
1988: 182). "L2learning has access to L1 competence, which was ultimately based on
67
UG, and L2 competence will only reflect those parts ofUG that are made use of in the
L1" (Cook & Newson, 1996: 293). According to this hypothesis, properties shared by all
natural languages should emerge in the L2 grammars of adult learners, and learners
should be able to realize the parameter settings ofL1 in L2. But learners should be unable
to reset parameters to values not instantiated in Ll. White (1990a: 127) describes this as a
"compromise position" which assumes that "aspects ofUG encoded in L1 can still be
tapped", although it is "dead as an active force in SLA". In other words, the learner can
make use of principles already instantiated in the L1, but cannot use those that the L1 has
not taken up. All else must be learned in terms of general problem-solving strategies. UG
is available to adult L2 learners only in the form in which it is appears in L1.
3.3.3 No access to UG
The "no access" hypothesis claims that child L1 acquisition and adult SLA are
fundamentally different cognitive processes, the former deriving from the language
faculty, and the latter determined by nonlinguistic processes. In this position, UG is
unavailable to SLA and SLA proceeds through the use of general problem-solving
procedures This position is best exemplified in the work ofBley-Vroman (1989), Clahsen
(1988), Clahsen & Muysken (1986); Schachter (1990). The grammar ofthe L2 "does not
incorporate the principles and parameter settings ofUG, since these are not derivable
from input or from textbooks" (Cook, 1994:485). In other words, an L2 can be learnt to
some extent from a grammar book or from drills. L2 learning thus becomes an implicit
and explicit process in which "features of language are presented to learners and acquired
68
by them through conscious attention, or in which teachers correct particular mistakes in
the learners' speech, i.e. in which negative evidence has a role to play" (Cook, 1994:485).
Proponents of this position often appeal to Lenneberg's Critical Period Hypothesis (CPR)
(1967) for language acquisition. According to this view, the obvious difficulty that adult
learners have in acquiring new languages after puberty has been ascribed to the existence
of a "critical" or "sensitive" period. Within this period, the brain retains "plasticity" which
allows the UG to function, or alternatively allows cognitive-motor "programs" to become
instantiated neurally (Lieberman, 1991; Bates et al, 1992). People of"a certain age"
generally find it impossible to acquire a second language to "native" proficiency.
Interpreted in terms ofUG, the CPH claims is that the direct access position is
unavailable to older learners, for whom L2learning depends either on the Ll or on other
aspects of the mind.
3.4 Evidence of access to UG
In the 1970's the first studies emerged which looked at SLA through the assumptions of
the then-current version of generative grammar. Dulay and Burt (1974) studied the order
in which child L2learners acquire specific morphemes and classes ofwords (e.g. articles,
progressive -ing, third person singular-s). They found that children with different Lis
follow the same sequence of acquisition of English as a second language, a sequence
which coincides with that followed by child Lllearners ofEnglish. Dulay and Burt took
this as evidence that L 1 and SLA are essentially the same process, driven by the same
69
"language acquisition device". This challenged the prevailing view that learners'
knowledge ofLl is the chief organizing factor in SLA. In another study, Bailey, Madden
& Krashen (1974) reported that adult L2learners also follow the order of acquisition
which Dulay and Burt found among both child L 1 and child L2 learners, re-opening the
debate about whether the Critical Period Hypothesis holds for adult L2 learning.
However, these studies soon proved unconvincing. Various methodological problems,
reviewed by McLaughlin (1987: 31-34; 66-69), weakened the empirical basis ofthis
research.
In the late 1970's, Ritchie (1978) may have been first to design a study of SLA, which
investigated the presence of a proposed linguistic universal. He tested Japanese-speaking
learners of English for sensitivity to the "Right Roof Constraint", a forerunner of the
Subjacency principle. The subjects rejected extraposition of sentential subjects and
extraposition out ofNPs in L2 only when such movement violated the Right Roof
Constraint. Claiming that Japanese has no rightward movement rules, Ritchie argued that
these learners' success at distinguishing between legitimate and illegitimate movement in
L2 cannot result from their knowledge ofLl. In addition, input data does not inform
learners of the kinds of movement which are disallowed in L2, so Ritchie interpreted
these results as evidence that his subjects' hypotheses are innately constrained by the
Right Roof Constraint, presumed to be a language universal. And this is an argument for
direct access to UG, since the subjects appear to obey a constraint in L2, which Ritchie
claims has no analogue in L1.
In the last decade, evolution of a theory ofUG based on principles and parameters has
raised numerous new empirical issues, stimulating rewarding work in the field. Felix
70
(I 985, 1987, 1988) believes that UG is still available to L2learners, and he explains the
difference between L1 and SLAin terms of what he calls competing cognitive systems;
Ohler (1988) takes the same position as well. Felix (1988: 285) tested German-speaking
learners of English for their knowledge of a number of properties of L2 purportedly
controlled by UG but which are not present in Ll, including Superiority, parasitic gaps,
that-trace and Case Filter effects. He reports that "in most cases the majority oftest
subjects gave correct responses in judging structural contrasts that can neither be learned
from positive evidence nor can be transferred from German". Such results not only
support the availability ofUG to L2learners, but also in addition suggest that an L2
learners' observation of principles ofUG is not restricted to the instantiation ofthose
principles in Ll.
In his approach, it is assumed that principles ofUG are available to adults as well as to
children. The observed differences in acquisition are said to reside in the separate
faculties that the two groups bring to bear upon the learning experience. For children, it is
UG, for adults, it is UG in competition with late-acquired general principles of a
problem-solving nature. Felix was the first to recognize the need to specify what other
mental resources are uniquely at the disposal of adult learners. As we will see in later
chapters, Felix's approach, together with some parsability principle, provides an insight
for the explanation of fossilization in the English-speaking learners' acquisition of the
long-distance reflexive in Chinese.
71
3.5 Evidence of lack of access to UG
Several types of evidence are claimed to demonstrate no access to UG in L2 learning.
Clahsen & Muysken (1986), who claim that L2 learners do not have a parameter-based
grammar at all, argue for the strongest version for this position. Clahsen (19 8 8) claims
that UG disappears in the process of maturation, that the principles ofUG are only
available to L 1 acquisition, and that subsequent languages learned by adult learners are
acquired in a different fashion. Clahsen & Muysken (1986) found different sequences of
acquisition for L2 and L1learning of German. In their studies, L1learners started with
the correct underlying SOV order even if this yielded sentences that were wrong; L2
learners started with the correct surface SVO order even if they were wrong in the
underlying structure. The L2 sequence was claimed to demonstrate lack of access to UG.
Subjacency is the second area that has been widely claimed to provide evidence that there
is no access to UG in L2 grammar. Bley-Vroman, Felix & Ioup (1988) found that only
slightly over half of the L2 subjects they tested knew subjacency in sentences such as:
*What did Sam believe the claim that Mary had bought? Johnson & Newport (1991)
found that Chinese learners of English scored consistently below natives at subjacency.
Other arguments which support the no access hypothesis are: the knowledge of the L2 is
not so complete (Schachter, 1988; Bley-Vroman, 1989); some L2s are more difficult to
learn than others are (Schachter, 1988); the L2 gets fossilized (Schachter, 1988); and L2
learners vary in ways that L 1 learners do not.
72
Clahsen & Muysken (1989) later have made a slight concession. They claim that in SLA,
a set of stable principles in UG which hold for every language, such as structure
dependence, locality, etc. are available indirectly, i.e. through the mediation of their
mature Ll grammar. According to Clahsen and Muysken (1989), the crucial difference
between L 1 and SLA is that adults no longer have open parameters. They propose that
principles are available for SLA via the Ll, but the parameters cannot be reset; L2
learners carry over from the L1 knowledge of subjacency, the Case Filter, etc.; they
cannot reset parameters such as bounding nodes, governing category, etc. The evidence
Clahsen (1988), Clahsen & Muysken (1986, 1989) used to support their arguments
concerns different developmental sequences and rules that they have observed in Ll
acquisition and SLA of German word order. However, the validity of ruling out the
availability ofUG in SLAby matching the developmental patterns between L1
acquisition and SLA has been questioned by Flynn & O'Neil (1988a: 10), who suggest
that "in order to develop an explanatory theory of L2 acquisition, we must account for
both the learner's L1 knowledge and principles independent ofthis knowledge".
In a reply to Clahsen & Muysken, duPlessis et al. (1987) argue that because of
misleading properties in the L2 input or because ofL1 transfer, L2learners do not
necessarily set all of the proposed parameters correctly at the same time. There are
intermediate stages in SLA when the L2 learner shows evidence of having set only some
ofthese parameters appropriately. Analysis by Schwartz & Tomaselli (1990) shows that
the behavior of adult learners of German is indeed UG constrained, with a number of
interacting parameters being set such that these learners have an IL grammar that
resembles that of neither the L1 nor the L2.
73
A slightly different version from Clahsen and Muysken's is taken by Bley-Vroman
(1989) and Schachter (1988, 1989), who argue that what is at issue is not whether UG is
available to SLA, but rather the extent to which UG operates in SLA. According to them,
UG in its entirety is available only to L1learners; what are available to L2learners are
the principles ofUG instantiated in the L1. Uninstantiated principles are no longer
available to L2 learners, nor are the other possible parameter settings not instantiated in
the L1. These researchers believe that there are some strildng characteristics in adult
SLA, which include lack of uniform and complete mastery ofL2, and wide variation in
terms of ultimate achievement, fossilization, etc.
Following recent developments in the UG theory, Tsimply & Smith (1991), and Tsimply
& Roussou (1991) have taken a radical view and claimed that parameters are not
associated with UG principles but with lexical items and in particular, functional
categories. They argue that there can be no parameter resetting in SLA. Parametric
variation is exclusively determined by the different values associated with functional
categories. According to them, functional categories constitute an independent
component ofUG (Tsimpli & Ouhalla, 1990), the UG lexicon. In L1 acquisition, the "UG
lexicon" is assumed to become available to the L1 learner as result of maturation and is
associated with the effect of Critical Period. In SLA, on the other hand, the "UG lexicon"
is inaccessible to the adult L2 learner, because it is assumed that language learning at
74
stages after the Critical Period cannot make use of the mechanisms available to children.
In a word, UG principles are assumed to be still operative in SLA; however, parameter
resetting in SLA is excluded.
While there are indeed fundamental differences between L 1 acquisition and SLA, it
seems somewhat rash to claim that none of the parameters uninstantiated in the L1 can be
reset for the L2. Experimental evidence and intuitive observation seem to suggest that
although success in resetting all the parameters for the TL is extremely rare in SLA, L2
learners are able to reset some parameters for the TL. The problem facing SLA
researchers is to pinpoint the parameters that can be reset in SLA and those that cannot,
and to provide explanations for them.
3.6 The role of Ll in SLA
Unlike the child in L 1 acquisition, the adult L2 learner comes to the task of acquisition
with the knowledge 6f at least one other language. Most current theories of SLA do, in
fact, assume that the NL of the learner plays a role in acquisition; however, what role the
NL plays is less certain. The linguistic knowledge ofL1 represents the major source of
prior learning to affect the construction of the IL grammar. According to Selinlcer's
(1972) IL hypothesis, language transfer is one ofthe principal processes operating in the
ILs. Especially in the early stages of acquisition, there are strong tendencies for the
restructuring and constructing ofiLs to be affected by Ll features. Although not all the
features ofLl are equally susceptible to transfer or exert a visible influence on IL,
experimental evidence has shown that knowledge ofL1 plays an important role in the
construction of IL grammars (see Gass & Selinker, 1983).
75
Prior knowledge and experience of one language, such as L1, can shape our expectations
of the experience of learning a new language. SLA in the early stages is particularly
subject to such influences. Moreover, learner's knowledge of a prior acquired language
can inform the learner of a great deal about language in general, providing a rich
repertoire of hypotheses for the learner to draw on in the construction of a TL or L2
grammar. However, incorrect hypotheses based on L1 features will result in structures
that deviate :fi:om target forms while corresponding to structures in the L 1, and manifest
themselves as errors. Thus, a characterization of the role played by L1 experience and
knowledge in SLA is indispensable to our understanding of IL development.
3.6.1 Ll knowledge and language transfer
Transfer phenomena allow researchers to infer how prior knowledge of an acquired
language (L1) can shape the learners' construction of an IL grammar. The concept of
transfer has gone through significant changes in the development of SLA research. In the
early days, the study of transfer was dominated by two major theories: the Contrastive
Analysis Hypothesis (henceforth CAH) (Fries, 1945; Lado, 1957) and the Creative
Construction Hypothesis (henceforth CCH) (Dulay & Burt, 1974; Dulay, Burt &
Krashen, 1982). Each of these theories provides a distinct perspective on the role ofLl in
SLA. CAH emphasizes forms "that can be attributed to the L1, which are known as
transfer or interference errors". These errors mirror some L 1 structure and are not part of
76
L 1 acquisition data of the TL. CCH, on the other hand, concentrates on L 1
"developmental errors", which are common to learners of different mother tongue, but are
also found in the L1 acquisition data ofthe TL (see White, 1988b:36).
The CCH theory was originally proposed as a reaction to the shortcomings of CAH.
While the CCH was in vogue, the role ofL1 was substantially de-emphasized. The fact
that CAH was often associated with the behaviorist learning theory, meant that the
transfer errors seem to be considered an embarrassment to approaches that emphasize
universal aspects of SLA, to the extent that some deny their significance or play down
their importance (e.g. Dulay & Burt, 1974; Dulay, Burt & Krashen, 1982).
With the advent of cognitive psychology and generative grammar, there emerged a more
enlightened view of the nature ofL1 and language learning. Language is no longer
viewed as "stimulus-response" verbal behavior, but a cognitive system, which is creative
and productive; language learning is not regarded as habit formation, but to acquire the
knowledge of grammar of a language, which is represented in the human mind. The
nature of transfer is not seen in terms of structuralist and behaviorist view as L 1 habits
being carried over to the L2\ but as a cognitive view of "knowledge and representation"
(See Schachter, 1992). The counterproposal that language is a set of abstract rules, which
are acquired thanks to an innate endowment triggered by exposure to environmental
input, began to appeal to L2 researchers. Within the framework ofUG, language is
4 What was embodied in the CAH was the practice of linguistic analysis by item-by-item comparison of surface features. It was assumed that language acquisition is a matter of habit formation, ie.e. old habits impede the learning of new L2 habits, while those L1 patterns that conform to L2 patterns aid the learning of the L2 (Fries, 1945; Lado, 1957).
77
viewed as a system of interrelated abstract principles and constraints. According to this
assumption, any particular transfer phenomenon cannot be adequately viewed as a single,
independent feature that corresponds to a similar feature in L 1. It should be treated as a
certain manifestation or instantiation of an underlying principle.
3.6.2 Parametric variation
Research in the parameter-setting approach has provided another view ofthe nature ofLl
transfer in SLA (White, 1985b, 1986b, 1988b; Flynn, 1987). From this perspective, one
of the important points to consider in SLA is the fact that the L2learner has already
activated UG in learning his mother tongue. That is to say, the L2 learner has already had
the parameters ofUG set or fixed during the acquisition of their Ll. According to
Chomsky (1981 b, 1982), it may be assumed that linguistic universals are invariant across
languages (i.e. the same for all languages), but universal principles vary in subtle ways
from language to language, and that is how the concept of "parametric variation" is
introduced. Within this theory, it is suggested that a limited number of parameters are
associated with a particular principle and that specific evidence from the language being
learned will trigger the relevant parameter for that language. In SLA, however, many
learners are faced with situations where the values of the parameters in Ll differ from
those in L2, or some principles are inactive in Ll but active in L2. During the early stages
of SLA, the L2 learners' IL grammars are characterized by some settings which
correspond to the settings in their L1 but are not compatible with the L2 parameter
settings (White, 1986b ). Therefore, during the process of SLA, L2 learners face the task
78
of "unlearning" the L 1 values of the parameters, and resetting the parameter to the values
appropriate to the L2 settings. Whether the unlearning process and the resetting process
are successful or not depends on the L2 data the learners are exposed to. More
importantly, their success depends on whether the L2 learners are able to process the data
to figure out their linguistic significance. White's (1985b) research indicated that having
to change a parameter ofUG causes problems for language learners and that this is a
source of transfer errors, particularly at lower levels of proficiency.
It should be noticed that there are different positions on the role ofL1 among researchers
who argue for a UG perspective on transfer. How the prior knowledge of a mother tongue
may shape the acquisition of a L2 is far from uniformly solved. Some argue that L1
parameter settings are part of the IL grammar, either briefly or for a longer period of
time, and consequently L 1 parameter settings affect the way L2 learners attempt to
comprehend and produce the L2 (e.g. Phinney, 1987, Schwartz, 1987; White, 1985b,
1986b, 1988b). Some researchers make an implicit or explicit assumption that "UG can
start from scratch in SLA, without any effect from L1" (Mazurkewich, 1985, 1988).
Flynn (1985, 1987, 1988) holds a different view. Results of empirical studies
investigating the role of the head-direction parameter in adult SLAby Flynn (1983, 1984,
1987) have indicated that, in the earliest stages of acquisition, Japanese speakers learning
English as a second language, are able to acquire the English value of the head-direction
parameter. Her studies claim to have confirmed the hypothesis that L2learners are able to
assign new parametric values in the construction of the L2 grammar when there is a
79
mismatch between the L1 and the L2 (e.g. Flynn, 1983, 1984, 1987, 1991, 1993, Flynn &
Martohardjono, 1994).
In Flynn's studies of acquisition of the head-direction parameter by Spanish learners of
English, and Japanese and Chinese learners ofEnglish, two different patterns emerged.
The Japanese and Chinese subjects worked out the fundamental properties of the head
direction parameter for the L2 in much the same way that early L1learners do. The
Spanish subjects do not need to assign a new value to this parameter; they already have
the correct value set for their L 1. As a result, these speakers can construct the structure
established by this value in working out the properties of the parameter. Flynn argues that
both of these developmental patterns, in some sense, match L1 patterns of acquisition for
English; however, each corresponds to a distinct stage in L1 acquisition in this
developmental sequence.
According to Flynn, L1 does play a role in SLA, and all L2learners are sensitive to the
match or mismatch of the settings of parameters between the L 1 and the L2 :from the
early stages of acquisition. It is assumed that all L2 learners use the same principles of
syntactic organization isolated in Ll acquisition in the initial construction of the L2
grammar, regardless of a match or mismatch in parametric values between the L 1 and the
L2. In the case where values in the Ll do not match those ofthe L2, L2learners assign a
new value to match the L2. In the case where values do match, no such assignment is
necessary (Flynn, 1987). Flynn's arguments are mainly based on SLA data concerning the
setting of the head-direction parameter. However, data about the settings of other
80
parameters which are reported in the SLA literature, suggest that parameter setting or
resetting is not as straightforward as Flynn claims and the learners' Ll has a more
important role to play in the parameter resetting in SLA. Indeed, the differences between
Flynn's studies and others may reflect differences in the experimental methodologies
employed in obtaining data from various language learners tested5•
One of the convincing interpretations of the role ofL1 in SLA and overviews of the
transfer phenomenon is that given by White (1985a, 1985b, 1988b, 1992). White (1992a)
summarizes four areas that make current views of the phenomenon of transfer truly
different from earlier conceptualizations. These four areas are: levels of representation;
clustering of properties; interaction of parameters; and learnability. According to White,
when the L 1 and L2 have different parameter settings for a particular parameter, "L2
learners will be affected by the parameter already instantiated in their L1, so that, at least
initially, they carry over the setting already established for the mother tongue" (White,
1988b: 43). The fact that positive evidence in Ll has caused a parameter to be set in a
particular way may obscure the fact that positive evidence in L2 motivates a different
setting. In situations where it is really necessary to reset parameters, the learner may have
particular difficulties, reflected in the transfer of the L 1 value, and hence in the incidence
. of intmference errors. According to her view, L2 learners would assume at the initial
stage of the acquisition that L1 and L2 match. In order to "reset" or "reassign" a new
5 Eubank (1989) challenges Flynn's position and methodology. Her latest position argued in Epstein, Flynn & Martohardjono (1996) is questioned by Bley-Vroman (1996), Borer (1996), Carroll (1996), Eubank (1996), among others.
81
value to the L2 parameter where the L1 differs from the L2, L2 learners have to "unlearn"
the L 1 value of the parameter in SLA.
Transfer and developmental influences are also viewed as interacting. Zobl (1980)
suggests that, if there is a natural developmental stage in the L2 that corresponds to a
pattern in the learners' L1, then the learner will be more likely to persist in using that
pattern in IL than if the pattern did not exist in his Ll. Zobl cited the data of Spanish
speakers' simple preverbal negation in English as an example. Children acquiring English
as L 1 are also found to go through a stage where they produce utterances, such as I no go,
The glass no will break, etc. As the simple preverbal negation occurs in both the L2
learners' mother tongue and L2 as a developmental stage, the learners' IL grammars are
susceptible to the L1 influence. According to Zobl, "L2 structures that show influence
from the Ll may in certain cases be more recalcitrant to restructuring; in other words;,
they show a tendency toward fossilization" (Zobl, 1980: 477).
During the past decade, research results have provided ample evidence of transfer in the
construction of aspects of L2 core grammar: the pro-drop parameter (White, 1985b;
Rilles, 1986), the Subjacency parameter (White, 1985c), the adjacency parameter (White,
1986a), and the head-direction parameter (Flynn, 1987). Different positions all recognize
the crucial role of transfer, with the only difference being the status of those parts ofUG
not instantiated in L1. If the uninstantiated UG principles are still available, there should
be evidence of such principles at work in the L2 grammar; if such principles are no
82
longer available, this evidence should not be found. These are in principle testable
predictions generated by the different claims.
3. 7 Learnability Considerations and SLA
Learnability study is concerned with the "logical problem oflanguage acquisition"
(Hornstein & Lightfoot, 1981) and the nature ofthe data available to the learner. Issues of
learnability arise in both L1 and SLA regardless of the role attributed to UG. In recent
years, some SLA researchers have tried to adapt the learnability theories developed in L 1
acquisition to enlighten the view of leamability issues in SLA ( cf. Pankhurst, Sharwood
Smith &Van Buren, 1988; Yip, 1995; Gregg, 1996). White (1985a, 1989b) argues that L2
learners are faced with the same problem as that posed for Ll acquisition, i.e. the
linguistic data available to the child do not seem sufficiently informative to allow the
learner to work out the complex properties of the L2. Only in this sense, the goal of L2
learnability theory is the same as for Ll. However, the learnability problems SLA
researchers are trying to account for are not the same as those pursued in the field of L 1
acquisition. Given the characteristics of lack ofultimate attainment, general failure,
fossilization in SLA (Bley-Vroman, 1989), SLA researchers have to account for the
"unlearnability or 'partial outcome' (Davies, 1984) as much as learnability" that occurs in
the L2 context (Yip, 1995).
According to Gregg (1996: 57), the learnability problem in SLA involves four
components or parameters:
1. Initial state: The L2 learner has already internalized a grammar of a specific natural language; is this important? Has the adult learner, conversely, lost any
properties of ( 1 )? Does the learner begin the process of SLA with both U G and L1 grammar, or just the latter? Does it make a difference which Ll grammar is part of the initial state?
83
2. TL (=L2) grammar, or rather, the TL-related IL finally attained by the learner. 3. Input: Negative evidence is available in many cases; is it necessary? Useful?
Usable? 4. Learning mechanism: Does the parser vary from L1 to L2? That is, must one
develop a new parser as well as a new grammar to be parsed: Do the learning mechanisms survive into adulthood? Conversely, do adults use mechanisms that children do not or cannot?
One ofthe major factors motivating aUG-type solution to the logical problem ofL1
acquisition is that linguistic input to the child consists exclusively of positive evidence, or
primary linguistic data (PLD). Negative evidence, such as explicit correction of deviant
child utterances, is in general neither available nor used.
Learnability theory attempts to explain how a language learner progresses from one state
of knowledge to another based on language input. The input available to the learner is
normally divided into two main types: positive evidence, consisting of grammatical and
well-formed structures; and negative evidence, which provides the learner with the
information that a given structure is not well-formed. Moreover, the two types of
evidence can be further divided into direct and indirect forms, resulting in four logical
types, as can be shown in Figure 1.
Positive Negative Direct Triggering data Explicit correction Indirect lfP then~Q Non-occurrence
(Catapult Hypothesis) Figure 1 Four types of evidence m the mput
84
3.7.1 Direct Positive Evidence
Direct positive evidence, also known as primary linguistic data (henceforth, PLD), comes
from the adult language and assumed to consist exclusively of grammatical structures of
the TL. Positive evidence is generally assumed to be the basic type of evidence the
learner uses to construct the grammar and revise their hypothesis (Yip, 1995). If the TL
grammar has a language feature F, the examples ofF in the input data are the positive
evidence available to the learner.
The main problem with positive evidence is the explanatory evidence of how the input
data can "trigger" change in the learners' grammar. Here, "trigger" refers to an aspect of
the input data to the language learner that induces the setting of a parameter in one of its
values within the Principles and Parameters approach. The triggering issue is difficult to
deal with, for it requires an assessment of the role of the input. And that is a delicate
problem for a theory oflanguage acquisition which holds as one of its fundamental
hypotheses that much of the relevant information is not contained in the PLD. That is to
say, which are the crucial data required to set a parameter in a certain way? The standard
assumption is that triggering, "as opposed to learning, requires less robust data and less
exposure to the input", and this "raises the question of what exactly is needed to serve as
triggering information" (Meisel, 1995: 18). According to Carroll (1989), triggering
differs from learning in that it is predicted to happen faster, requiring less frequent and
less simple input data, and the developmental pattern is expected to be much more
uniform across individuals.
A related question about triggering data is how many tokens or what length of exposure
to the relevant data, are required for a change in the learners' grammar to take place.
Clahsen (1991b) observes that the child "seems to require a certain amount of positive
evidence to fix a parameter at a particular value". Randall (1987; 1992:100) goes on by
postulating that there exists a "triggering threshold", i.e. that a "sufficient number of
tokens" need to occur, but "we know surprisingly little about it". Lightfoot (1991) goes
further by claiming that data must be salient and frequent in order to be able to act as
trigger.
3. 7.2 Indirect Positive Evidence
85
Since grammar is an implicational network of systematically interrelated structures, not a
random collection of separate entities, positive evidence can serve as indirect evidence
for the occurrence or non-occurrence of another related structure, and causing change in
the grammar. In order to solve the problem of overgeneralization of dative structures in
L1 acquisition. Randall (1987, 1992) uses indirect positive evidence in her proposal. She
suggests that indirect positive evidence is based on a principle of the form [IF P THEN
NOT Q], which means that if construction P occurs in the language, construction Q
cannot occur. In other words, the learner can get the information ofnon-occurrence ofQ
indirectly through the presence ofP in the input, because positive evidence of a structure
can serve as indirect evidence for the occurrence or non-occurrence of a related structure.
In more recent work, Randall (1992) formalizes this idea as the "Catapult hypothesis".
Randall (1992) applies the Catapult Hypothesis to argue that lexical overgeneralizations
86
can be unlearned in a principled way by appealing to independently motivated
grammatical principles. During the grammar construction, it is assumed that exposure to
very little data can result in a great deal of the grammar of the language being acquired.
However, the question of how many tokens of a piece of evidence are required before
reorganization of the grammar can happen is still not well understood. And Randall
(1992) relates this to the threshold problem.
Randall's proposal depends crucially on establishing disjunctive principles of the form
[either A orB (exclusively)]. It is not without problems (cf. Gropen et al, 1989; Pinker,
1989l However, as we shall see in later chapters, the proposal Randall takes provides us
with a solution to the learnability problems in the acquisition and unlearning of the
null/non-null subject by Chinese second language learners of English and English second
language learners of Chinese.
3.7.3 Direct Negative Evidence
Explicit negative evidence, also named as direct negative evidence, is defined by Berwick
(1985: 41) as: "the (perhaps methodical) pairing of positive (syntactically well-formed)
and negative (syntactically ill-formed) sentences with the appropriate labels well-formed
and ill-formed. It could also include correction of ill-formed utterances (alternatively,
parses) via (a) explicit negative reinforcement (e.g. That's wrong) or (b) tacit negative
6 Pinker (1989) questions the validity of Randall's (1987) indirect evidence.
87
reinforcement (e.g. responding with the correct pattern, or not respondingt. A number of
empirical studies in L 1 acquisition have shown that children do not have systematic
direct negative evidence available to them (Pinker, 1984; Bowerman, 1983; Gropen et al,
1989). And some researchers claim that even if direct negative evidence is available, they
either would not or could not make use of it (Braine, 1971).
In SLA, the situation is different in that direct negative evidence is available to L2
learners. In L2 classroom situations, the learner is provided with systematic and
descriptive corrections with respect to some features ofTL grammar; in naturalistic
situations, L2 learners can also have access to certain forms of negative evidence, but in
the form of confirmation checks, requesting, clarification, etc. (Schachter, 1986b). On
negative evidence in SLA, see also Bley-Vroman, 1986; Rutherford, 1987; Rutherford &
Sharwood-Smith, 1988; Schachter, 1983, 1986b, 1991; Birdsong, 1989.
This raises the question of the role of negative evidence in SLA. In L1 acquisition, it is
assumed that negative evidence is not necessary; indeed, it is normally not even available
(Pinker, 1989). In order to solve the Baker's Paradox7 in L1 acquisition, Pinker (1989: 10)
proposes that it must be shown that negative evidence (1) exists; (2) exists in usable
form; (3) is used; (4) is necessary to avoid or recover from overgeneralization (cf. Beck
& Eubank, 1991, for SLA). However, whether these would work together is unclear. The
7 Baker's Paradox is proposed by Pinker (1989:7) to refer to the three aspects of problems facing the children: (1) the assumption of lack of negative evidence; (2) productivity of rules problem; (3) the arbitrariness of natural languages. According to Pinker, these three factors make acquisition of argument structure alternations in the verb lexicon impossible to explain.
88
role of direct negative evidence in adult SLA is still quite controversial. White (1991a,
1992a) has conducted a teaching experiment on the effectiveness of formal instruction in
parameter resetting in SLA. She carried out a study on adverb placement in English and
found the formal instruction on the "Verb Movement" Parameter had only short-term
effects on French learner's judgements. However, in response to her claim, Schwartz &
Gubala-Ryzak (1992) argue that the use of negative evidence (i.e. the formal instruction)
had not affected the underlying English competence of these French learners, because,
with regard to verb movement, the formal instruction did not result in reorganization of
the TL grammar.
From these considerations, some researchers' claim that negative evidence plays no
significant role (hence the "no negative evidence hypothesis") and it may be true, at least
for right now. As Williams (1987: ix) has pointed out, we agree that
"it may be well be that the [no negative evidence] hypothesis is a complicated half-truth, as much a part of the idealization of the problem as a part of the answer to the problem. At the present, when we have no idea where the truth lies, this hypothesis serves the very useful function of ... evaluating the consequences of proposals about the nature of grammar or the study of acquisition".
Yip (1995) argues that, the challenge for any theory ofLl that assumes this hypothesis, is
to account for the fact that learners are able to recover from overgeneralization even
without the input of negative evidence. In other words, given the unavailability of
negative evidence, how does the learner unlearn or "undo" the ungrammatical forms?
89
3. 7.4 Indirect Negative Evidence
It is suggested that, if the child notes certain linguistic structures or forms which are
expected to occur in the input but do not turn up, that fact could inform the learner about
the correctness of his hypothesis. And that fact can serve as a kind of evidence that such
forms are ungrammatical (e.g. Chomsky, 1981a). This is called the indirect negative
evidence, which Berwick (1985: 41, note 26) defines as follows:
"indirect negative evidence, the inference that if a linguistic construction P can appear in simple sentences and is not observed to appear, then P does not occur in any sentence, no matter how complex".
In order to use the indirect negative evidence, the learner has to actually notice the non-
occurrence of deviant structures which he expects to hear but does not appear in the
input, to help him decide whether his hypothesis about these linguistic forms is correct or
not.
Now let us take the pro-drop or null-subject parameter as an example. Assuming that
+pro-drop is the default value for the parameter (as is possible, cf. Rizzi, 1980, cited in
Berwick, 1985: 292), the learning of a non-pro-drop language may proceed in the
following manner. The learner will initially consider the +pro-drop value, which is the
more inclusive one, to be the correct value. But since examples corroborating this value
will not show up in the enumeration of a finite set of data of a certain complexity, the
acquisition procedure will instead select the more restrictive -pro-drop value8•
8 Saleemi (1992) proposed a slightly less conservative alternative, which he claims to be conservative enough to ensuring correct learning. The learning seems to encompass not just what he calls "positive identification", but also "exact identification", requiring some negative evidence. The logic is that sometimes, due to pragmatic/quasi-linguistic factors, the languages generated by parameter values may lack a (somewhat marginal) set of constructions which, had they existed, would be perfectly well formed. Therefore the learner has to notice their non-representation in the available data.
90
The effect of indirect negative evidence has been questioned by a number of language
acquisition researchers. Braine (1971) disagrees with the effect of indirect negative
evidence on the grounds of the overwhelming computational burden on the learner. He
believes that, given the irregularity in language, it is far too much for the learner to keep
track of a great range of construction types together with their ratios of occurrence and
non-occurrence. As Valian (1988, 1990a) points out, the problem with indirect negative
evidence is that, there is no principled way of knowing how long one should wait before
concluding that a specific structure will never emerge in the data. However, Saleemi
(1992: 135) argues that the use ofindirect negative evidence is constrained and well
defined, since it can only be used to "eliminate a subpart of a choice, but not to identify a
value, unless in collaboration with positive data". He believes that "positive data are still
considered to be the core of the primary data, with indirect negative evidence having an
essential but supplementary role, and direct negative evidence remaining irrelevant to the
process of acquisition".
3.8 Learning problems and unlearning strategies
3.8.1 Overgeneralizations
In order to learn a language, one has to make correct generalizations based on the finite
set of samples available in the input. In the course ofL1 acquisition and SLA, learners
inevitably overgenerate and undergenerate. One ofthe general issues in the theory of
learnability is the overgeneralization problem (Baker, 1979; Bowerman, 1987; Lasnik,
1981; Pinker, 1986, 1989) in the learners• grammars. As is well known, children almost
invariably succeed in developing correct grammars. But, why do learners generalize?
91
There are two reasons behind this. First, language requires generalizations. Since
language is infinite, learning a language means that the learner has to make correct
generalizations based on the finite available data they are exposed to. Since "a language
is an open-ended set, not a fixed list, so the child must generalize from these inputs to an
infinite set of sentences that includes the input sample but goes beyond it" (Pinker, 1989:
5). Moreover, it is a characteristic of human minds to generalize, and this is true for both
children and adults and is based on two assumptions. First is the least-effort principle, in
which the learner formulates the simplest rules that require minimum effort to process;
and the second is that learners "favor general rules because they are easier than restricted
rules" (Fodor & Crain, 1987: 35). General rules are inherently easier and simpler than
restrictive ones. As Baker (1979) has pointed out, rule generalizations can cause serious
learnability problems if the rule becomes overtly generalized and productive. Given that
learners do over generalize during their acquisition process, the question that arises then is
as follows: how do they manage to accomplish the enormously difficult unlearning task
required of them? Or, how they ever manage to retreat to the correct forms without the
help of direct negative evidence?
The traditional answer to this question is the assumption, that there is no negative
evidence available and that the child is conservative9 in nature and does not
9 However, Mazurkewich and White's (1984) study reveals that English-speaking children do operate with some kind of non-conservative lexical redundancy rule, which is formulated on the grounds of analogy and consequently inappropriate generalizations of dativization (cf. Pinker, 1989, Randall, 1985). Some of the examples are listed below. These sentences were incorrectly judged to be grammatical by their subjects (age approximately 9 to 15 years). The examples under (8a) are those of to-datives, those under (8b) of/ordatives.
8a. * David suggested Ruth the trip. *Susan explained Jane the problem. * Bob reported the police the incident.
92
overgeneralize at all ( cf. Baker's Paradox, note 7). In connection with the learning of
dative alternation in English, Baker (1979) distinguished two kinds of
overgeneralizations: 'benign errors' and 'embarrassing errors'. The former refers to the
overgeneralized inflections such as in goed,foots, etc. When the learner uses the incorrect
forms mentioned above, because of the consistent exposure to positive evidence (such as
the irregular forms ofwent,feet, etc), the correct forms will eventually replace the
incorrect ones. On the other hand, the embarrassing errors do not seem to be correctable
on the basis of positive evidence alone, and they are predicted to be difficult to "unlearn"
since the positive evidence in the input does not provide disconfirming evidence to reject
the incorrect hypothesis.
The following examples illustrate how some of the overgeneralizations may occur (from
Pinker, 1989: 7).
1). John gave a dish to Sam. John gave Sam a dish.
Give: NP1_NP2 to-NP3 Give: NP1_NP3 NP2
2). John passed the salami to Fred. John passed Fred the Salami.
Pass: NP1_NP2 to-NP3. Pass: NP1_NP3 NP2
3). John told a joke to Mary. John told Mary a joke.
Tell: NP1_NP2 to-NP3. Tell: NP1_NP3 NP2.
4). John donated a painting to the museum. *John donated the museum a painting.
5) John reported the accident to the police. *John reported the police the accident.
8b. * * Anne created Sarah a costume. * Tom captured Canada the prize. * Paul designed Claire a house.
93
According to the above examples, it would be a reasonable generalization that any verb
with the NP1_NPz to-NP3 argument structure (prepositional dative) could also have a
NPz_NP3 NPz argument structure (double-object dative). The problem is that not all the
verbs with the prepositional argument structure dativize, i.e. appear in both versions of
the alternation, as examples in (4) and (5) shows. Given the non-availability of negative
evidence, the learner has no way of 1m owing this.
According to Pinker (1989: 5), the solution to the learning problem works in this way:
"Constraints on the learner force him to entertain a restricted set of hypotheses that includes the correct hypothesis but excludes many others. The learner can then compare the predictions of a hypothesis (which sentences it generates) with the input data so that incorrect hypotheses can be rejected".
Pinker proposed there are four ways in which one of the child's hypotheses can be
incorrect ( cf. Figure 2).
H T H T T H
OOCID @@ (i) (ii) (iii) (iv)
"+" symbol = positive evidence "-"symbol= negative evidence
Figure 2: Learnability scenarios (Pinker, 1989: 6)
In Figure 2, "H' stands for the set of sentences generated by the child's incorrect
hypothesis and "T' refers to the set of sentences generated by the actual target language.
There are problems concerning disconfirming an overgeneralized hypothesis arising in
the absence of negative evidence. The child's hypotheses can be disjoint from the TL, as
94
in (i). In this case, any sentence in the input is sufficient to inform the child that the
hypothesis is not correct. When the child's hypothesis grammar and the TL are partially
intersected as in (ii); or when a hypothesized language is less inclusive than the TL (i.e.
the TL is the superset and the hypothesis language being the subset) as in (iii), the
positive evidence consisting of input sentences in the non-overlapping region of the TL
suffices to impel the child to reject the hypothesis. If, however, when the hypothesized
language is more inclusive than the TL, i.e. the hypothesis language forms a superset and
the TL the subset as in (iv), there is no positive evidence available to the learner to
correct their wrong hypothesis.
Now we will take a look at another example, the SLA problems of the English non-null
subject and non-null object by Chinese speakers. In Chinese, both the subject position in
infinite clauses and the object position can be empty (see later chapters 4 & 5), but they
can also be lexically filled, as in the following examples (3.1) and (3.2). The "null-
subject" or "pro-drop" parameter specifies the grammatical conditions, which must be
met for this empty category "pro" to be allowed to occur.
(3.1) a. women xiawu pengjian le yuehan de taitai, e zhangde hen piaoliang.
We afternoon meet PFV10 John DE11 wife, look very beautiful * A:ftemoon we met John's wife. e Looked very beautiful.
b women xiawu pengjian le yuehan de taitai, ta zhangde hen piaoliang. We a:ftemoon meet PFV John DE wife, she look very beautiful
(3.2)
10 PFV =perfective aspect marker.
11 DE= de, is a modifying marker in Chinese that occurs at the end of a prenominal modifier.
a. Ni duguo zhe ben shu le rna? e Duguo e le. You read this CL book PFV QM
*"Have you read this book? e Reade already." read this book PFV
b Ni duguo zhe ben shu le rna? wo/ta/tamen Duguo e le. You read this CL book PFV QM I/he/she/they read this book PFV "Have you read this book? I/He/Size/Tizey have read tlzis book already."
95
From the English translations of (3 .1 a, b) and (3 .2a, b), we can see that English does not
allow either the empty subject or the empty object. With respect to the pro-drop
parameters, English is the subset of Chinese while Chinese is the superset of English, i.e.
Chinese is more inclusive than English. If, as it is assumed, there is no negative evidence
available in SLA, Chinese learners ofEnglish would have no positive evidence in the
input to indicate that unlike their Ll Chinese, which is the TL, English does not allow the
empty subject and empty object. According to White (1986a), positive evidence in the Ll
has caused the parameter to be set in a particular way; this can obscure the fact that the
L2 requires a different setting. She also predicts that in parameter resetting in SLA, ifthe
setting for the parameter in the learners' Ll is more inclusive than that in the L2, it will be
difficult or impossible for the L2 learners to 'unset' the Ll setting in the absence of
necessary evidence. Following her prediction, we can infer that to unlearn the empty
subject and empty object during the acquisition of English can be a problem for the
Chinese learners, because the empty subject and empty object are likely to persist and
become potential candidates for fossilization. We will find out in the later chapters
whether this is the case.
96
3.8.2 The Subset Principle
In the 11 acquisition and leamability literature, it is usually accepted that leamability
constraints and learning principles prevent the L1leamers from producing
overgeneralized grammars. And one ofthe learning principles is "Subset Principle",
which was first proposed by Berwick (1985: 236) and further developed by Wexler &
Manzini (1987) and Manzini & Wexler (1987). The Subset Principle (henceforth SP) is a
condition on the sequence of grammars that characterize the process of language
acquisition such that early grammars in the process generate languages that are subsets of
those generated by later grammars in the process (Ritchie & Bhatia, 1996). Such
principles or constraints allow children to formulate only conservative hypotheses so that
language learning can proceed based on positive evidence alone; as it has been argued
that only positive evidence is available to L1leamers (Baker, 1979; Wexler &
Humburger, 1973). If the children overgeneralize by entertaining a grammar, which
generates a superset of the TL, positive evidence will not be enough to disconfirm the
wrong hypotheses. Instead, negative evidence will be necessary. Since negative evidence
is not available to L 1 learners, a learning principle of SP has been posited. Wexler &
Manzini (1987) developed the SP within a parametric theory of language acquisition
(Chomsky, 1981a) with a view that the SP is an independent learning principle which
interacts with the principles ofUG during the acquisition to guide the setting ofUG
parameters. In Wexler & Manzini (1987), the SP is formulated as follows: the learning
function maps the input data to that value of a parameter which generates a language; it is
not only compatible with the input data, but also the smallest among the languages
compatible with the input data.
97
According to the SP, when the learner is faced with input that could be accommodated by
more than one grammar, the learner must start off with the narrowest possible hypothesis
compatible with the available data. The more general superset options are chosen only
when there is specific positive evidence to support these options. The SP is motivated by
the observation that the evidence that is effective for the child in first language
acquisition is essentially limited to positive evidence (Ritchie & Bhatia, 1996: 703). "The
learner should hypothesize languages in such a way that positive evidence can refute an
incorrect guess" (Berwick, 1985: 37).
Now let us take a look at the phenomenon on the conditions of reflexives, as listed in
examples (1) and (2).
(1) Johni blames himselfi.
(2) Peten thinl(S [IP Johnj blames himselfiuj].
As the above examples shown, an English reflexive such as himself must have an
appropriate antecedent within the clause in which it occurs, such as John in (1). In (2), it
refers to an antecedent, which is outside that clause, such as Peter, and this is disallowed.
That is, the antecedent must be John, the subject of the embedded clause and cannot be
Peter, the subject of the main clause.
On the other hand, Chinese reflexive ziji may have a long-distance antecedent outside the
clause in which it occurs, as in (3). Either Lulu or Liping or Xiaotong may be the
antecedent of ziji.
98
(3) Lului shuo [ IP Lipingj yiwei [IPXiaotongk zeguai ziji ilj!I,]
Lulu say Liping think Xiaotong blame self
Lulu says Liping thinks Xiaotong blames her/herself.
Obviously, as we can see from the above examples, the domain in which the antecedent
of a reflexive may occur in English is a subset of that in Chinese. The SP argues that
learners start off with the narrowest possible hypothesis compatible with the available
data, i.e., they must select the subset grammar since they proceed from positive evidence
only. Children would initially set the most restrictive value for reflexives (as we can see
from the English examples); only by being given the positive evidence of examples such
as (3) can the children change the setting and revise the grammar to allow long-distance
reflexives. In this way, positive evidence in the input can motivate the learners to expand
their IL grammar to a more generalized hypothesis, which is compatible with the TL
grammar. The problem arises only when the learners' Ll makes the learners' initial
hypothesis for the TL too general. And the recovery from a more general hypothesis to a
less general one is difficult or even impossible in the absence of negative evidence.
If the learners' hypothesis follows the ordering proposed by the SP, there is no problem of
overgeneralization. However, learners do not simply start with the narrowest hypothesis
in SLA, since the ILs frequently represent supersets with respect to the TL grammar. For
one reason or not, L2learners may not necessarily notice the linguistic significance ofthe
positive evidence in the input for the construction of the TL grammar even if the positive
evidence is available to them.
99
So far, a number of studies have been conducted to examine the claim that SP not only
can apply toLl acquisition but also to SLA. White (1989d) reviewed several studies
investigating different parameters ofUG to which the SP has been said to apply. She
concluded that the SP does not guide adult SLAin the acquisition of the Governing
Category Parameter by adult Korean learners (Finer, 1989; Finer & Broselow, 1986); by
Japanese learners (Finer, 1989; Hirakawa, 1989), and by Chinese and Spanish learners
(Thomas, 1989) of English. In an attempt to evaluate White's hypothesis, MacLaughlin
(1992; 1994) considered the Case-Adjacency Parameter (Chomsky, 1981a, 1986a;
Stowell, 1981), the Pro-drop Parameter (Hyams, 1986), and the Bounding-Node
Parameter (Rizzi, 1982) for subjacency and reflexive binding. MacLaughlin concluded
that in fact none ofthese parameters provides support (or the lack thereof) for the SP
because it does not apply to them. He even argued that the preconditions for the principle
to apply are not met in any of these studies and concluded that the principle is irrelevant
in SLA.
On the other hand, however, Berent (1994) reviewed several studies and suggested an
alternative analysis or interpretation of the data, which enabled him to argue that the SP is
indeed relevant to adjacency phenomena (White, 1989c), the Configurationality
Parameter (Zobl, 1988), and the Governing Category Parameter (White, 1989b).
According to his interpretation of the results of studies on SLA of English relative
clauses, Berent argues that the SP does apply to SLA. Both Berent (1994) and
MacLaughlin (1994) argued that, in all the cases where L2learners do seem to be guided
100
by the SP, the parameters investigated do not meet the Subset Condition12 (i.e., they are
not in a subset/superset relation). Berent suggested that the SP does apply if parameters
do meet the Subset Condition.
The evidence presented in the literature remains inconclusive and limited to a small
number of parameters. Nevertheless, the SP can make some predictions for the L2
context. Without available negative evidence, when the IL generates a superset in relation
to the TL with respect to a hypothesis (or parameter setting), the SP predicts the situation
to be unlearnable. L2 learners often adopt a hypothesis that is based on Ll or a
hypothesis between the TL and the L 1. And these are the situations where learning
difficulties occur.
As we will find out from the later chapters, in acquiring the long-distance antecedents in
Chinese, the English learners of Chinese may fail to extend a less inclusive grammar to a
more inclusive grammar in spite of the presence of positive evidence in the input. To
expand a subset grammar to a superset grammar may not be as straightforward as the SP
predicts. Though there are arguments that many areas of grammar that are not
parameterized (cf. Rutherford, 1989a) and the SP may not available to L2learners, I
12 Wexler & Manzini (1987) proposed the Subset Condition that requires a proper subset-superset relation to obtain between languages defined by any two values of parameter. Wexler & Manzini argue that the Subset Condition is the precondition required to be met before the SP can apply. "It is necessary that two values of a parameter in fact yield languages which are in a subset relation to each other (i.e., one is a subset of the other). This requirement we call the Subset Condition. It is necessary for the Subset Condition to hold in order for the Subset Principle to apply" (Wexler & Manzini, 1987: 45). Later, they proposed another Independence Principle to solve the "many-parameter problem", because there is more than one parameter in a language, setting just one parameter does not suffice to allow the calculation of subsetsuperset relations within a language apart from the effects of other parameters. Wexler & Manzini (1987: 46) argued that the Subset Condition and Independence Principle are assumed to be "necessary and sufficient for the Subset Principle to apply in all cases".
101
believe the SP can still be applied to characterize some potential learning difficulties in
SLA.
3.9 Conclusion
For the availability ofUG problem in SLA, I have introduced direct access to UG,
indirect access to UG and no access to UG position. I take the view that L2 learners do
have access to UG principles and parameters. However, adult L2 learner may not have as
direct access to UG as a child does, since adult L2learners can not obtain ultimate Ll
competence and can only succeed partially and their competence varies from individual
to individual. In our view, L2learners still have access to UG in varying degrees and they
can re-set parameters for the TL during the SLA process. However, the extent to which
they can have access to UG and what parameters are prone to be/not to be resettable in
SLA remains a further issue.
I have also discussed the IL grammar and the transfer problem in SLA. I believe that the
role of previous knowledge and transfer is seen as a pervasive force in shaping IL
grammars. Fossilization occurs only when the permeability ofiL grammar is lost. SLA
leamability theory has to address both "unleamability" as well as "learnability" problems.
Whether a TL feature is learnable or partially learnable depends on the various factors,
such as the types of evidence available to the learner and the processing factors, etc.
102
Chapter 4 Null Subjects and Null Objects in Chinese
4.1 Introduction
In the following chapter, using a Government and Binding Framework, we will make a
contrastive survey of aspects of Mandarin and English in terms of the language features
involved in the investigation forthcoming in later chapters. First, I will discuss aspects of
the Null Subject or Pro-Drop Parameter and topic structures in Mandarin and English. I
will then look at the null object phenomenon in the two languages and finally, discuss
Chinese and English reflexives.
4.2 Null Subject
4.2.1 Null Subject and Pro-Drop Parameter
The generative linguistics literature on the Null Subject or Pro-Drop Parameter (Rilles,
1986; Phinney, 1987; White, 1985, 1986) is vast. It has been observed (Perlmutter, 1971)
that in languages such as Italian, Spanish and Chinese, it is possible to omit the subject in
a finite declarative sentence or clause as shown in the following examples:
(1 ). Spanish
(a) Juan come
Juan eats
(b) e 1come
He eats
(2). Italian
1 e = Empty Pronoun.
103
(a) lui parti
he left
(b) e parti
he left
(3). Chinese 2(C. -T. J. Huang, 1982)
(a) Zhangsan shuo [ta xiawu hui lai] Zhangsan say he afternoon will come "Zhangsan said that he will come this afternoon"
(b) Zhangsan shuo [e xiawu hui lai] Zhangsan say afternoon will come "Zhangsan said that he will come this afternoon"
In each ofthe (b) examples in (1) and (3), the pronominal subject is "missing" but
speakers understand it to be there covertly.
Languages that allow for the missing subject have been described variously as pro-drop
languages, subject-drop languages or null subject languages. The cross-linguistic
variation which concerns whether a language can have null subjects; or, whether the
subject of a clause can be suppressed or not (Newmeyer, 1988) is known as the Pro-Drop
or Null Subject Parameter.
Chomsky (1981a: 240) noted that there is a "clustering of properties related to the pro-
drop parameter" and listed the following as the most common properties:
1). the omission of subject pronouns; 2). the free inversion of subject and verbs in declarative sentences; 3). subject extraction from a wh-island; 4). apparent violation of the that-Trace Filter;
2 Throughout the thesis, tones are omitted from Chinese examples for typographical convenience.
104
5). empty resumptive pronouns in embedded clauses.
According to this parameter, a distinction can be made between pro-drop languages and
non-pro-drop languages. Hence the following contrast of grammatical Italian sentences
with the ungrammatical English equivalent of Italian sentences (Riemsdijk & Williams,
1986: 300-301):
( 4). Missing subject pronoun:
Ec! parlano di linguistics They-speak of linguistics
"They talk about linguistics"
(5). Subject freely postposed from subject position:
1. Le brigate rosse hanno telefonato The brigades red have called
"The red brigades have called"
ii. EC1 hanno telefonato le brigate rosse
"The red brigades have called"
(6). Apparent violation of the that-trace filter:
Chi1 credi che EC verra
"Who do you believe will come"
(7). Subject extraction out of a wh-island:
La donna che1 non so quando EC1 abbia telefonato The woman who not I-know when has called
"the woman about whom I don't know when she has called"
(8). Null resumptive pronoun:
il professor che1 non so chi abbia detto che EC1 sia arrivato
3 EC refers to empty category.
105
The professor who not I-lmow who has said that has arrivato "the professor about whom I don't lmow who said that he has arrived"
It is clear that languages like Italian and Spanish observe the pro-drop parameter, being
able to optionally delete the pronominal subject from a tensed clause, to freely invert the
order of the subject and verb in a declarative sentence, etc. Languages like English and
French, on the other hand, do not comply with the pro-drop parameter, thus permitting
only very restricted omission of the subject pronoun, and no free inversion of the subject
and verb, etc.
4.2.2 Two Principles to Identify a Pro-Drop Language
The early formulations of the Null Subject Parameter focused on richly inflected
languages such as Italian and Spanish and attributed the possibility of a null subject to
rich verbal inflection. Based on his observation and analysis of mainly Romance
languages, Taraldsen (1978) argues that there are some languages which may have null
pronominal subjects in tensed clauses, because they are quite rich in morphological
inflection in their verbs, which contains all features ofthe missing subjects. That is to
say, ifthere is a "rich .enough" system of verb-subject agreement in a language, then its
verb is marked with the grammatical features of the subject, including its person, number
and its gender.
Following Taraldsen (1978), Chomsky (1981a, 1982) claimed that the distribution of the
Pro-Drop Parameter is determined by the principle ofrecoverability, or what Jaeggli
(1982) terms the "identification hypothesis". It is widely accepted that the null subject in
finite sentences occurs in Romance languages because they have a rich enough subject-
106
verb agreement system to determine the content of the null subject (cf. Chomsky, 1982;
Taraldsen, 1978).
By recoverability, is meant that the morphological richness of the verb is sufficient to
restore both the semantic reference and syntactic contents of the unexpressed subject. In
other words, if the dropped pronominal subject can be recovered from the verb itself, then
this subject is unnecessary, and would sound or look somewhat redundant. As C. -T. J.
Huang (1989: 185) has pointed out, the recoverability requirement indicates that "a
pronoun may drop from a given sentence only if certain important aspects of its reference
can be recovered from other parts of the sentence".
In a language like Italian or Spanish, the subject of a finite clause may drop, because the
agreement marking on a finite verb is sufficiently rich to recover or identify its reference.
But for languages such as English and French, which have relatively "impoverished"
agreement morphology, pro-drop is prohibited from the subject position of a finite clause.
However, the identification hypothesis (Jaeggli, 1982) predicts that the deletion of
pronominal objects will not be permissible either in a pro-drop language or in a non-pro
drop language, because neither of the two exhibits any property of verb-object agreement.
Consequently, when the null object pronouns occur, their grammatical features cannot be
discerned from the verbal inflection. The result will be an ungrammatical sentence.
The agreement-based features of the Pro-Drop Parameter, which seems to be quite
plausible in the case of certain well-studied European languages, present an obvious
challenge when Chinese data is examined. It is generally agreed that Chinese has no such
107
inflectional morphology on its verbs, i.e. Chinese has no subject-verb AGR (agreement)
features at all. Based on the observations above, it seems analogously reasonable to
predict that Chinese will not allow any null elements either in the subject position or in
the object position, since the nature of the subject and object could not be recognized if
omitted. However, what occurs in Chinese is just the opposite: the null elements can,
more often than not, occur rather freely in both the subject and object position of a tensed
clause. For example, in the Chinese discourse presented in (9) below, all of speaker B's
answers are acceptable discourse whereas some of the answers in the English examples'
in (10) are not acceptable.
(9). Speaker A:
SpeakerB:
Zhangsan kanjian Lisi le rna? Zhangsan see Lisi ASP QM "Did Zhangsan see Lisi?"
(a). Ta kanjian ta le. He see he ASP "He saw him."
(b). e kanjian tale. "[He] saw him."
(c). Ta kanjian e le. "He saw [him]."
(d). e kanjian e le. "[He] saw [him]."
(e). wo cai [e kanjian e le] I guess see ASP
"I guess [he] saw [him]."
(f). Zhangsan shuo [e kanjian e le]. Zhangsan say see ASP
* "Zhangsan said that [he] saw [him]." (adapted from C. -T. J. Huang, 1984a:533)
(10). Speaker A: Did John see Bill yesterday? Speaker B: a). Yes, he saw him.
b).* Yes, e saw him. c).* Yes, he saw e. d).* Yes, e saw e. e).* Yes, I guess e saw e. f).* Yes, John said e saw e. (Adapted from C. -T. J. Huang, 1984a:532)
108
As can be seen in the above examples, unlike English, which allows neither null subjects
in finite clauses nor null objects, Chinese grammar allows both overt (i.e. phonetically
realized) subjects and objects, and null subjects and objects.
As is well known, Chinese has no subject-verb agreement at all. The early formulations
of the Null Subject Parameter theory in the early 1980s (Jaeggli, 1982; Rizzi, 1982;
Hyams, 1986) ignored and failed to explain the fact that "inflection-poor" languages,
such as Chinese, also allow null subjects and null objects in finite sentences. As the
following Chinese examples from C. -T.J.Huang (1989:187) illustrate, both subjects and
objects may drop from finite sentences in Chinese.
(11). Zhangsan kanjian Lisi le rna? Zhangsan see Lisi ASP Q "Did Zhangsan see Lisi?"
(12). A. [ta] kanjian [ta] le. He see he PFV4
"[He] saw [him]".
B. wo xiang [ta] kanjian [ta] le. I think he see he PFV "I think [he] saw [him]."
(13). A. Zhangsan shuo [e hen xihuan Lisi] Zhangsan say very like Lisi
"Zhangsan said that he liked Lisi."
4 PFV refers to Perfective Aspect Marker.
B. Zhangsan shuo [Lisi hen xihuan e] Zhangsan say Lisi very like Zhangsan said that Lisi liked [him].
109
(Adapted from C. -T. J. Huang, 1989:187, his examples (1), (2) and (4a)).
In this example, the null subjects are not identified by a rich agreement system as in the
Romance-type languages, but identified by the discourse topic (13b) and a noun phrase in
the matrix sentence (13a) respectively. As can be seen above, Chinese is regarded as a
pro-drop language, because it allows null subjects; in contrast, English is a non-pro-drop
language. Null subjects can occur quite freely in Chinese, but null subjects are not
allowed in finite sentences in English.
4.2.3 C. -T. J. Huang's Modification Approach
The types of sentences illustrated in (9) and (1 0) have prompted C. -T. C. Huang (1982,
1984a, b, 1987, 1989) to revise Chomksy's agreement-based theory of the Pro-drop
Parameter as well as Taraldsen's identification generalization so as to account for the
facts of Chinese.
In the work of Chomsky (1981a, 1982, 1986a, 1989), the study of two null elements such
as control (the occurrence of a null pronominal in the subject position of a non-finite
clause) and the study of pro-drop (a null pronominal in other positions) are generally
considered to be distinct. In Chomsky (1981a), these two elements are called ["big'1 PRO
,and ["small"] pro. PRO is assumed to be a pronominal anaphor, which under standard
assumptions must be ungoverned. Therefore, within each language, it is limited or occurs
in positions such as the subject of a non-finite clause or infinitives. On the other hand, pro
110
is a pure pronominal and unlike that of PRO, the occurrence of pro is not universal across
languages (C. -T. J. Huang, 1989: 185). Pro occurs only where there is a rich enough
agreement to identify it. Thus, it occurs in subject position in languages with rich subject
verb agreement such as Spanish and Italian, and in object position in languages with rich
verb-object agreement such as Pashto (cf. C. -T. J. Huang, 1989:186). Within each
language, PRO is limited to the subject position of a non-finite clause. Unlike the
occurrence of PRO, that of pro is not universal across languages (C. -T. J. Huang, 1989:
185).
C. -T. J. Huang (1982, 1984a, b, 1989) suggested that, in Chinese, the null subject in
finite sentences can be pro or a variable. He claims that, taking Chinese as a pro-drop
language, one has to face the question of why Chinese allows pro subjects in the absence
of AGR. As is pointed out above, since Chinese has no AGR features, the proposal that
pro occurs in languages with sufficiently rich agreement, will not work for Chinese. C. -
T. J. Huang (1982) argued that "the pro-drop phenomenon may happen in one of two
types of languages, either a language with a very rich agreement system, or in a language
without agreement at all. On the other hand, a language with a rather meager system of
agreement, such as English, does not allow pro-drop" (pp. 366-367). C. -T. J. Huang
argues that pro can occur in Chinese if it is locally identified by a nominal element
(rather than by AGR, as Chomsky and others propose).
C. -T. J. Huang (1989:193) suggested a modification of Chomsky's (1981a) agreement
based proposal, called the "Generalized Control Rule" (GCR), under which PRO and pro
may both occur.
111
(14). Generalized Control Rule (GCR)
(C. -T. J. Huang, 1984a:552) Coindex an empty pronominal with the closest nominal element.
(C. -T. J. Huang, 1989:193) An empty pronominal [pro or PRO] is controlled in its control domain (if it has one).
(15). a is the control domain for f3 iff it is the minimal category that satisfies both (a) and (b): a. a is the lowest S or NP that contains (i) (3, or (ii) the minimal maximal
category containing (3, and b. a contains a SUBJECT accessible to (3.
C. -T. J. Huang (1989) proposed the abolishment of the distinction between PRO and pro
originated in Chomsky (1982). According to C. -T. J. Huang, PRO and pro are in fact the
same entity, and the identification requirement for pro and the control requirement
holding of PRO reduce to a single principle, i.e. the GCR (Herbert, 1995:227). According
to his GCR, an empty pronominal takes the closest nominal element as its antecedent. A
nominal element, according to C. -T. J. Huang, refers to either NP or A GR. In this way,
pro should be identified by its "closest nominal element". In the case of a pro-drop
language, it is AGR; but in the case of the subject of a subordinate clause in Chinese (16),
it is the subject of the next higher clause.
(16). Zhangsan shuo [ei jian guo Lisi]. Zhangsan say meet EXP Lisi * "Zhangsan says that e has met Lisi".
According to C. -T. J. Huang, the null subject of the embedded clause in (16) is pro,
coindexed with the subject of the matrix sentence, Zhangsan, because there is no AGR in
the embedded sentence and the closest nominal element is the matrix subject.
112
4.2.4 Null subjects and Morphological Uniformity
Hyams (1987a) made a distinction between licensing conditions for empty categories and
identification conditions for these empty categories (discussed in Jaeggli & Safir, 1989).
The idea is that null subject languages such as Chinese and Italian share the property that
their morphological systems are uniform. That is, whereas Chinese has only underived
inflectional forms, Italian possesses only derived ones. This is exemplified in the
paradigm in Atkinson (1992: 117), and repeated in (17) below:
(17) Chinese (18) Italian wo gongzuo io lavoro "I work" m gongzuo tu lavori "You work" ta gongzuo lui/lei lavora "He/she/it works" women gongzuo noi lavoriamo "We work" mmen gongzuo VOl lavorate "You (pl.) work" tam en gongzuo loro lavorano "They work"
The Chinese examples in (17) are well-formed despite the absence of agreement with the
range of interpretations indicated:
gongzuo ["work"] "I work/you work/he, she, it works/we work/you (plural) work/they work"
As we can see from examples in (17) and (18), Chinese and Italian share the property of
being morphologically uniform in their verbal paradigms, but this uniformity takes quite
different forms in the two languages. "In Italian, a verbal paradigm is uniform by virtue
of all its forms being inflected; in Chinese, uniformity is achieved by none of the forms in
the paradigm being inflected" (Atkinson, 1992: 118).
English, on the other hand, exhibits a mixed paradigm by virtue of the fact that it has only
a two-way inflectional contrast.
113
(19) I sleep you sleep he/she/it sleeps we sleep you sleep they sleep
Jaeggli & Safir (1989) developed Hyams's (1987a) linguistic account of null subjects and
proposed a sharp distinction between licensing and identification. By licensing, is meant
the "grammatical properties that a language must contain in order for null subjects of
tensed clauses to be admissible" (Davies, 1996: 475). Identification, on the other hand, is
the grammatical mechanism for assigning reference to a null element. fu the case of null
subjects in tensed clauses, identification has been attributed either to the morphological
infmmation encoded in the verb (as in inflection-rich languages) or to discourse
information (as in inflection-poor languages).
Jaeggli & Safir (1989) proposed a principle responsible for licensing null subjects based
on the promise that it is not the richness of inflectional paradigms, but rather their
morphological uniformity, which plays a role in the licensing. They offer the following
formulation of the null subject parameter, which determines when null subjects in tensed
sentences are licensed:
Null Subject Parameter (Jaeggli & Safir, 1989: 29)
(20) Null subjects are permitted in all and only languages with morphologically uniform paradigms.
Morphological uniformity is defined as follows:
(21) Morphological Uniformity: An inflectional paradigm P in a language L is morphologically uniform ifP has either only underived inflectional forms or only derived inflectional forms.
114
Under this definition, Spanish and Italian, and Japanese and Chinese, are morphologically
uniform; the former two have inflectional endings throughout the paradigm since all
verbal paradigm have derived forms, and the latter two have no inflectional endings at all
since all verbal paradigms have only underived forms. English, on the other hand, is not,
and is morphologically mixed, since some forms have endings (e.g. he run-s) while
others do not (e.g. run). According to Jaeggli & Safir (1989),pra is licensed only in those
languages with a morphologically uniform paradigm.
According to the formulation in (19), morphological uniformity is a necessary but not a
sufficient condition for null subjects. Not all languages that are morphologically uniform
allow null subjects, as is the case in German5 (21):
(22) Ich arbeit-e 1SG "I work" Du arbeit-est 2SG "You (singular familiar) work" Er arbeit-et 3SG "He/she/it works" Wir arbeit-en 1PL "We work" lhr arbeit-et 2PL "You (plural) work" Sie arbeit-en 3PL "They work"
For referential null subjects to appear in a language, Jaeggli & Safir propose that null
subjects must also be identified. Now let us return to the Italian paradigm in (18). In
Italian, null subjects are licensed by virtue of its having a morphologically uniform
inflectional paradigm. It was also noted that Italian is a language with rich inflectional
morphology. According to Hyams (1986), it is the features in INFL (or AG 6) that
5 It is widely conceded that Gennan is not a null subject language. However, it is reported that at least one dialect allows null expletives in certain environments in the indefinite extraposition construction (Davies, 1996: 488; Atkinson, 1992: 119).
6 For agreement, 'AGR' is more conventional, but I will follow Hyams' (1986:27) usage for reference to her work. AGR contains features of person, number and gender (9-features), and of course languages differ in which, if any, of these features they encode.
115
eventually get "spelled out" as verbal inflections which permit identification. In a word,
null subjects occur in Italian because they are licensed via morphological uniformity and
they are identifiable by means of rich agreement. This is what she calls agreement-
identification.
4.2.5 Topic Chains and Topic Identification
Following Tsao (1977), C. -T. J. Huang (1982, 1984a: 549) maintains that languages can
be broadly characterized either as "discourse-oriented!' or "sentence-oriented".
Chinese and Japanese are notable members of the former; English, French and Italian
belong to the latter.
According to C. -T. J. Huang, an empty category (EC) in zero-topic languages7 can (and
sometimes must) be bound by an NP in the discourse rather than in the matrix sentence.
C. -T. J. Huang argues that "its reference must, in other words, be the discourse topic,
someone or something that a given discourse is about" (1984a: 541). C. -T. J. Huang
(1984a) argues that discourse-oriented languages such as Chinese, Japanese, and Korean
have a rule of "Topic NP Deletion":
[Topic NP Deletion] ... operates across discourse to delete the topic of a sentence under identity with a topic in a preceding sentence. The result of such a deleting process is formally a topic chain (p.549).
The examples in (23), based on Huang's (55), show Tsao's (1977) Topic NP Deletion.
7 C.-T. J. Huang (1984a) distinguishes zero-topic languages like Chinese from non-zero-topic languages like English and Italian. Based on this, he suggested that there are four different types oflanguages. (1) zero-topic, pro-drop languages, such as Chinese and Portuguese, (ii) non-zero-topic, non-pro-drop languages, such as English and French, (iii) zero-topic, non-pro-drop languages, such as Gennan, (iv) nonzero-topic, pro-drop languages, such as Italian and Spanish.
(23). [Zhangguo, difang hen da,] [ e, renko hen duo,] China place very big population very many
[ e, tudi hen feiwo,] [ e, qihou ye hen hao,] land very fertile climate too very good
[ e, women dou hen xihuan.] we all very like
"(As for) China, (its) land area is very large. (Its) population is very big. (Its) land is very fertile. (Its) climate is also very good. We all like (it)."
In (23), the ECs are identified by a sometimes null sentence topic or topic chain. In the
116
discourse-oriented languages, the notion of "discourse-topic" seems to be interacting with
purely syntactic processes to identify empty subject positions 8• If this is correct, there is
another mechanism available for the identification of empty subjects in Chinese, namely,
topic-identification. This is schematized in (24) below:
(24) [Top ei] [s ei gongzuo]
According to C. -T. J. Huang, "Top" is an operator in an A-bar position which binds, and
therefore identifies, eh the variable but not pro in the subject position. So, there are two
mechanisms9 whereby the identification requirement on empty subjects can be satisfied:
agreement-identification in sentence-oriented languages where agreement features must
be present in a case-governing category (e.g. a category containing both AGR and TNS
features) and topic-identification in discourse-oriented languages (Atkinson, 1992:120).
8 See Xu (1986) for more information on the role of discourse notions in establishing an appropriate taxonomy of empty categories for Chinese.
9 Following C. -T. J. Huang (1984a), Hyams & Jaeggli (1988) discuss a third kind of null subject identification, namely, through a c-commanding nominal. They suggest that null subjects in Chinese can be identified by the c-commanding nominal in the case of embedded sentence, or by the topic, which may be dropped. This type of identification is not of concern in our discussion.
117
The morphological uniform approach to the presence or absence of null subjects is a
distinct improvement over previous analyses, in that it is based on data from a fairly wide
range of languages, including Japanese, Chinese, Spanish, Italian, German, Irish, Danish,
Swedish, French, Arabic and American Sign Language. However, this approach does
lack intuitive appeal in that there seems to be no logical connection between the presence
or absence of overt subject pronouns. The morphological uniform approach does not state
the precise nature of the relationship between null subjects and morphological uniformity.
4.2.6 The Finite and Non-Finite Clause Distinctions in Chinese
During the past years, many linguists (C. -T. J. Huang; 1982; Li, 1985, 1990; Tang, 1990;
Chiu, 1995; Yuan, 1997) suggested that, in spite of the lack of morphological differences,
there exist distinctions in Chinese clauses parallel to those observed between finite and
non-finite clauses cross-linguistically 10•
C. -T. J. Huang (1982: 367) assumes that the INFL of a Chinese finite clause is an
obligatory governor or proper governor because it is truly lexical. What is dominated by
INFL in Chinese is often a true lexical category. These include: you 'have' (the perfective
aspect marker), zai 'at' (the progressive aspect marker), hui 'will' (the future aspect
marker), le (pronounced liao) 'to finish' (the perfective aspect marker), guo 'to have done'
(the experiential aspect marker), or zhe 'attained' (the progressive aspect marker), etc.
10 There are linguists who believe that there is no systematic way to distinguish finiteness from nonfiniteness either syntactically or morphologically in Chinese, such as Y. Huang (1987, 1989 & 1994) or Hou (1988). Hou suggests that it is not necessary to distinguish finite verbs from non-finite verbs in Chinese as it is in English. Instead, all Chinese verbs are generally considered as finite. The reason is that in Chinese the syntactic function of a tensed clause is just the same as that of a non-finite one. Thus, since Chinese does not have a necessary distinction between finite and non-finite verbs, it will not have any distinction between tensed clauses and non-tensed ones.
118
According to C. -T. J. Huang, all these elements occur preverbally as independent lexical
categories, not as affixes 11 (such as English suffixes -s, -ed, -ing, etc., which cannot be
used as independent lexical items). The aspect markers in Chinese make it possible for
the subject variable and the empty topic to be properly governed by INFL, thus satisfying
theECP.
C. -T. J. Huang suggests that in Chinese the aspect markers can be used to identify finite
and non-finite clauses, and only verbs in finite clauses can be mounted with aspect
markers. For example, C. -T. J. Huang identifies the embedded sentences in the following
sentences as nori-finite and the null subject in them as PRO (C. -T. J. Huang, 1982:
11 Some examples of le and you provided by C.-T. J. Huang (1982: 165): (i). Zhangsan piping-le ta. (ii) Zhangsan mei-you piping ta.
Piping-ASP he not-have criticize he "Zhangsan criticized him." "Zhangsan did not criticize him."
(iii) * Zhangsan you piping ta. (iv) * Zhangsan mei piping-leta. Have criticize he not criticize-ASP he
(v) * Zhangsan mei-you piping-le ta. (vi) Not-have criticize-ASP he
(i)-(ii) show that le is correctly suffixed to a verb in an affirmative sentence but appears preverbally in a negative sentence. (iii)-(v) show that the reverse situation is unacceptable, and (v) that the two items you and le are mutually exclusive.
12 However, Hou (1988) argued that aspect markers cannot be used to distinguish the finite embedded clauses from the non-finite, because the embedded clauses are non-finite does not prevent the use of aspect markers in them. He listed the following examples to illustrate his point (examples A, Band C = Hou's 14, 15 and 17).
A). ta bi Lisi lai-guo, (couqiao ni bu zai jia). He force Lisi come EXP fortunately you not at home *"He forced Lisi to have come; (fortunately you were not at home)".
B). (xingkui) wo quan Zhangsan mai-Ze zhe ben shu fortunately I persuade Zhangsan buy PFV this CL book
(xianzai mai bu dao le). now buy not get PART "(Fortunately), I persuaded Zhangsan to have bought this book; (now you won't be able to buy it)".
C). wo zhunbei lai -zhe bu zou, (kan ni na wo zenmo ban).
119
(25). wo bi Lisi [PRO lai] I force Lisi come "I force Lisi to come." (C. -T. J. Huang, 1982: 169, example (108))
(26). wo zhuenbei [PRO mingtian lai] I prepare tomorrow come "I prepare to come tomorrow." C. -T. J. Huang (1982: 354, example (87))
(27). wo quan Zhangsan [PRO bu mai zhe ben shu] I persuade Zhangsan not buy this CL book "I persuaded Zhangsan not to buy this book." C. -T. J. Huang (1982: 354, example (88))
Following Li's (1990) evidence suggested in her examples13, Yuan (1997:470) agrees that
"finite clauses, but not non-finite clauses, are barriers to the syntactic relation between the
negative polarity item and the negation, and that the finite and non-finite distinction
exists in Chinese". According to Li (1990), a negation marker in Chinese can license a
I prepare hang-on PRG not leave see you take me how deal-with "I prepare to be hanging on and refuse to leave (and see how you'll deal with me)"
Since the aspect markers do not refer directly to time, they can be used in non-finite clauses, as is the case in English (see examples in D and E) and Chinese (see examples in A, B and C).
D) ....... to have done .... .. E) ....... to be doing .... ..
We believe that there is neither tense nor agreement in Chinese verbs, and the categories of TNS and AGR are absent in Chinese syntax (or are morphologically empty in Chinese).
13 Li (1990) suggests that a negation marker in Chinese can license a negative polarity such as "renhe" in an embedded non-finite clause, but not in an embedded finite clauses, illustrated in the following examples (I) and (ii) ((9) and (10) in Li, 1990: 2):
(i)a. * wo meiyou gaosu guo ta ni zuo renhe shiqing I not-have tell EXP him you do any thing
*"I have not told him you did anything."
b. wo meiyou quan guo ta qu zuo renhe shiqing I not-have persuade EXP him go do any thing "I have not persuaded him to do anything."
(ii) wo mei jiao ta pai ni qu quan tamen zuo renhe shiqing. I not-have ask he send you go persuade them do any thing "I have not asked him to send you to persuade them to do anything."
120
negative polarity item such as renhe "any" in an embedded non-finite clause but not in an
embedded finite clause. There is a contrast between sentences containing persuade-type
verbs and those containing tell-type verbs. It is claimed that persuade-type verbs
typically take non-finite clauses as their complements, whereas tell-type verbs typically
take finite clauses as their complements.
Following Wexler (1994), Yuan (1997) suggests that "it is the underspecification of1NS
and AGR features that licenses null subjects in Chinese" (p. 469). Moreover, from
Wexler's (1994) proposal ofunderspecification ofTNS in some child language14, he
assumes that AGR is available in Chinese though it is underspecified, because it is
believed that when TNS exists, AGR is present (Yuan, 1997). The presence of AGR in
Chinese is evidenced by the fact that the interpretation of the long-distance binding of the
Chinese reflexives ziji, is believed to depend on its coindexation with a higher AGR via a
local AGR (cf. Progovac, 1992, 1993), or to depend on its movement to AGR at LF (cf.
Cole & Wang, 1995). According to Yuan, what TNS specifies, for example, is finiteness
alone, and not the values of [+/-past].
I agree with the suggestion by Yuan (1997) that it is the "underspecification of1NS
(tense) and AGR (agreement) features that licenses null subjects in Chinese" (p.469).
Since Chinese lacks TNS and AGR features (cf. C. -T. J. Huang, 1982; Li, 1990), INFL
in Chinese lacks contrastive features. And this is why TNS and AGR are underspecified
in Chinese (Yuan, 1997). According to Chomsky (1993, 1995), inflectional features of a
14 Wexler (1994) discussed a number of possible explanations to account for the phenomenon of root infinitives in child Ll acquisition of English, French, German, Dutch, etc. He suggested that TNS in these child languages was underspecified in the sense that it lacked contrastive values, such as [+past] or [-past].
121
verb are the intrinsic property of the verb, and they project together with the verb,
forming a complex with the verb root.
"Functional heads like TNS and AGR do not have morphological elements in
them; their role is to serve as checking domains for relevant V-features.
Inflectional features can be strong or weak. Strong features are visible at PF and
they must raise to INFL (i.e. TNS and AGR) at S-structure so that features can be
checked at PF. If features are weak, then they are not visible at PF, and they will
not raise to INFL for checking until LF. This is to comply with the general
principle of economy (Procrastinate). That is, do nothing until it is necessary"
(Yuan: 1997:470).
4.2. 7 Topic Structure and Base-Generated Topics in Chinese
Some linguists have studied the characteristics of topic in Chinese. In Chinese linguistics,
Y. R. Chao (1968) was one of the first linguists to propose that Chinese has a topic-
comment structure. He suggests that the grammatical meaning of subject and predicate in
a Chinese sentence is topic and comment. Some examples of topic-comment
constructions are:
(Topic) (Comment) (28). Zhege ren erduo ruan.
This man ear soft "(As for) this man, the ear is soft." (He is gullible)
(Topic) (Comment)
(29). Shige li wuge Ian le. Ten pear five spoil ASP "(Of) ten pears, five have spoiled."
(Topic) (Comment) (30). Pengyoujiu de hao; yishang xin de hao.
Friend old DE good clothes new DE good "(Of friends), old ones are best; (of) clothes, new ones are best." (Xie Tianwei, 1992: 17)
122
From a syntactic point of view, topic is the constituent in the sentence-initial position. It
is important to lmow that topic position in Chinese is always different from the normal
(S-structural) subject position, i.e. [Spec, IP].
Li & Thompson (1976a) suggested that there is a distinction between "topic-prominent"
and "subject-prominent" languages, the former being the discourse-oriented languages
(like Chinese) and the latter sentence-oriented ones. In subject-prominent languages like
English, all sentences must have a subject, a property that Chomsky (1982:10) described
as part of the Extended Projection Principle. This accounts for the use of pleonastic
element or expletive such as "it" and "there" in weather-predicate sentences and raising-
predicate sentences in English15• ill so-called topic-prominent languages such as Chinese,
Japanese or Korean, structural subjects are not an obligatory element, or a basic
requirement of the sentence. Chinese does not have an expletive comparable to English
"it". This is illustrated in example sentences (31) and (32): (examples 8 & 9, Yuan, 1997:
473)
15 In English, these "dummy" subjects it or there may be needed whether or not it plays a semantic role, such as the following examples (Li & Thompson, 1976: 467):
(i) It is raining. (ii) It is hot in here. (iii) It is possible that the war will end. (iv) There is a cat in the garden.
(31). Kuai yao xiayu le. Soon will rain PART "It is going to rain soon. 11
(32). Kanshangqu Zhangsan hen lei. Seem Zhangsan very tired "It seems that Zhangsan is very tired."
123
Li & Thompson (1976: 459) suggested that "the notion of topic may be as basic as that of
subject in grammatical descriptions", and that "languages may differ in strategies in
construction sentences according to the prominence of the notions of topic and subject"16•
According to them, the topic in Chinese cannot be viewed as derived by movement from
some argument position in the sentence. It is claimed that topic is not determined by the
verb and that topic selection is independent of the verb. The following sentences (33)-
(36) provide some examples (adopted from LI & Thompson, 1976: 479, examples (60) to
(63)).
(33). Huang-se de tu-di dafen zui heshi Yellow-colour DE soil manure most suitable *"The yellow soil (topic), manure is most suitable."
(34). Nei-zuo fangzi xingkui qu-nian mei xia-xue that-classifier house fortunate last-year not snow *"That house (topic), fortunately it didn't snow last year."
(35). Dongwu wo zuzhang bao-shou zhengce animal I advocate conservation policy *"Animals (topic), I advocate a conservation policy."
(36). Zei-jian shiqing ni bu neng guang mafan yi-ge ren this-classifier matter you not can only bother one person *"This matter (topic), you can't just bother one person. 11
16 According to Li & Thompson (1976: 459), there are four types oflanguages: (i) subject-prominent languages; (ii) topic-prominent languages; (iii) languages that are both subject-prominent and topicprominent; (iv) languages that are neither subject-prominent nor topic-prominent.
124
As can be seen from examples (33) to (36), the topics the yellow soil, that house, animals,
and this matter are not determined by the verbs. Li & Thompson argue that these
examples provide clear evidence that there is no process oftopicalization or movement.
Chinese topics leave no "gaps", and they are "base-generated" rather than derived by
movement from some other position in the sentence. The "topic prominence" of Chinese,
Li & Thompson claimed, is illustrated by the extensive use of "base-generated" or "non-
gap" topics, i.e. topics that bear no anaphoric relationship to a constituent in the comment
sentence.
One of the classic examples cited by Li & Thompson (1976: 462) is the following
sentence (3 7):
(37). Nei-chang huo, xingkui xiaofangdui lai de zao. That-classifier fire fortunately fire-brigade come COMP early "That fire (topic), fortunately the fire brigade came early."
As C. -T. J. Huang (1984: 550) points out, sentences like (37) "must count as basic forms
in that they cannot be plausibly derived from other, more 'basic' forms". C. -T. J. Huang
(1984: 561) believes that the topic position of a sentence must be available at D-structure.
"Therefore, for each variable bound to a topic, there are two possible ways to
derive it: it may be created by movement as a wh-trace, or it may start out as an
EC at D-structure and later be coindexed with the topic (and become a variable)"
(C.-T. J. Huang, 1984: 561).
Cole (1987) further proposes that the topic in Chinese is a basic unit of a sentence which
is an adjunct being adjoined to I", as in (38)17•
17 According to C. -T. J. Huang and Cole's approach, the topic Zhongguo de minshan dachuan in sentences ( a)-(b) below is not derived from some non-topic constituent of the sentence and it must be "basegenerated", or what we call "non-gap topic" (C. -T. J. Huang, 1984a, 1989; Cole, 1987: 604).
125
(38) C"
~ SPEC C'
c I"
I VP
4.3. Null Objects
4.3.1 C. -T. J. Huang's Variable Analysis
As can be seen in the following example (38), Chinese allows null objects as well as null
subjects in finite sentences. English, on the other hand, allows neither null subjects nor
null objects in finite clauses.
(3 8) Zhangsan shuo [ e bu renshi e] Zhangsan say not know *"Zhangsan said e didn't know e ." "Zhangsan said (he) didn't know (him)."
C. -T. J. Huang's (1982, 1984a) approach is perhaps the best-lrnown analysis ofECs
(Empty Categories) in Chinese. In his analysis, he argues that the ECs in subject position
can be PRO, pro or a variable, but that the ECs in object position are always variables. C.
(a) Zhongguo de minshan dachuan, wo zhi qu-guo Huangshan. China DE famous scenic places I only go EXP Yellow Mountains *"Famous scenic places in China, I have only been to the Yellow Mountains".
(b) Zhe wei nus hi wo bu jide yiqian wo zai nar jian- guo. This CL lady I not recall before I in where meet EXP *"This lady, I cannot recall where I have mete before."
The same analysis is available for (b), where the topic zhe wei nus hi is not created through movement like a wh-trace, otherwise the "subjacency principle" would be violated. The topic in (b) is base-generated and the empty object in the embedded wh-clause is also base-generated and becomes a variable after it is coindexed
with the topic.
126
-T. J. Huang suggests that an EC in object position is a variable, the result of the
movement (in the syntactic component) to an A'-position of a base-generated empty
object. According to him, there is an asymmetry18 of the ECs in subject position and
object position in Chinese. As can be seen in the example sentences (39) and (40), C. -T.
J. Huang suggests that the null subjects in sentence (39) can be coreferential with the
subject of the matrix clause Zhangsan or with someone whose reference is fixed outside
of the matrix clause. The EC is pro in the former case, and in the latter, the EC is a
variable.
(39) Zhangsan; shuo eu1 bu renshi Lisi. Zhangsan say not know Lisi "Zhangsani says (hei) does not know Lisi".
(40) Zhangsani shuo Lisi bu renshi e*i!J· Zhangsan say Lisi not know Zhangsani says Lisi does not know (himi).
(examples from sentences (19c) and (19d) in C. -T. J. Huang (1984a: 537))
The following English sentences illustrate the similarities and differences in distribution
and reference between English and Chinese.
(41) John said that he knew Bill.
(42) John said that Bill knew him.
Like the Chinese example sentences in (39), the embedded subject pronouns in sentence
( 41) may refer either to John or to someone whose reference is different from John.
However, C. -T. J. Huang argues that, in the Chinese example sentence (40), the
embedded object EC can refer only to someone whose reference is fixed outside of the
18 Xu & Langendoen (1985) and Xu (1986) argue that Mandarin does not display the asymmetry claimed by C. -T. J. Huang. They propose a different approach to the analysis. The argument hinges in part on the status of certain disputed judgements and on the validity of tests that C. -T. J. Huang applied to apparent counterexamples to his analysis adduced by Xu and by Xu & Langendoen.
127
entire sentence. Since it cannot be coindexed with the matrix subject Zhangsan, the null
object in the embedded clause can neither be a PRO nor a pro. C. -T. J. Huang argues that
"while null subjects may be regarded as genuine null pronominals (pro's), a null object is
best analyzed as a variable A'-bound by an operator that is itself null" (1989: 187).
According to him, the null object can only be a variable because it is in accordance with
Principle C of Chomsky's binding theory (Chomsky, 1981) that a variable, like other
referring expressions, cannot be coreferential with a c-commanding nominal occurring in
an argument position, i.e. it cannot be A-bound19• According to him, not only are null
objects in Mandarin Chinese and Portuguese languages variables, but in fact null
pronominals (pro) can not occur in object position in any language (in the absence of an
identifying clitic or object agreement marker).
C. -T. J. Huang (1984b & others) has claimed that the possibility of null variable objects
in such languages as Mandarin is a side effect of the existence ofnull objects. C. -T. J.
Huang (1984a: 541) argues that "when a topic NP appears in addition to a matrix subject,
an embedded object EC (null object) is mostly naturally interpreted as bound by that
topic". Example sentence (43) below is identical in surface form to (40), except it
contains an overtly present topic "nage ren" (=that man), whose reference is different
from the matrix subject Zhangsan. In both sentences of ( 40) and ( 43), the null objects
19 Within Chomsky's (1981, 1982) Government and Binding framework, the identity of an EC is functionally determined according to the following principle (Chomsky, 1981: 330):
a. An EC is a pronominal if and only if it is free or locally bound by an element with an independent thematic role, and a non-pronominal otherwise. b. A non-pronominal EC is an anaphor if and only if it is locally A-bound, and a variable if locally
A'-bound.
128
refer to a topic, and the only difference between them is that in the former the topic is not
covertly present in the sentence, but can be probably inferred from the discourse. Thus,
the analysis ofthe sentence in (40) can be assimilated to (43).
(43) neige ren; Zhangsan shuo Lisi bu renshi e;. That man Zhangsan say Lisi not know "That man, Zhangsan said that Lisi didn't know e".
Furthermore, sentence ( 40) can be represented as ( 44), with an empty topic binding the
embedded null object, corresponding to (43) with a lexical topic (cf. C. -T. J. Huang,
1984a: 542).
( 44). (= 40) [TOP ei]
"*[Himi],
[Zhangsan shuo [Lisi bu renshi ei]] Zhangsan say Lisi not know
Zhangsan said that Lisi didn't know ei".
According to C. -T. J. Huang, we assume that the embedded object has been topicalized
and appears in the sentence-initial position, and there is nothing missing in sentence ( 40).
Similarly, within the traditional analysis of topicalization, there is nothing missing in
object position in sentence (44) (= 40). C. -T. J. Huang claims that what is really missing
in ( 44) is the topic, that is, "an object is topicalized first before it is deleted from the topic
position" (C. -T. J. Huang, 1984a: 542).
From the example ( 44), it seems that movement is involved in topicalization in Chinese.
However, the movement does not seem to be subject to the Subjacency Principle. C. -T.
J. Huang argues that there is a difference between Chinese and English with respect to
Subjacency Principle. In the example sentences below, the movement out of a sentential
subject (as in (45)) is in violation ofthe Subjacency Principle in English sentences and
the movement out of a complex NP clause (as in ( 46)) also gives rise to an
129
ungrammatical string in English. However, sentences ( 45) and ( 46), which constitute a
violation of the Subjacency Principle under the assumption that the topics are derived by
movement, are grammatical sentences in Chinese.
( 45) ta daman keyi jie zhe ben shu, danshi he certainly can borrow this CL book but
(TOP ei) [cP ta xianzaiJle ei] shi bu keneng de. He now borrow be not possible PART
*"He can certainly borrow this book, but that he wants to borrow e now is not possible."
"He can certainly borrow this book, but it is unlikely that he is able to borrow (it) now."
( 46) Zhe ge dianying hen jingcai, this CL movie very interesting
[NP [cP kan guo ei de] ren] bu duo. Look EXP20 DE people not many
*"This movie is very interesting, but people who have seen e are not many." "This movie is very interesting, but there are not many people who have seen (it)."
According to C. -T. J. Huang, the Subjacency Principle does not apply to Chinese. The
topic position of a sentence must be available at D-structure since Chinese is a topic-
prominent language (cf. 4.2.7.).
"Therefore, for each variable bound to a topic, there are two possible ways to
derive it: it may be created by movement as a wh-trace, or it may start out as an
EC at D-structure and later be coindexed with the topic (and become a variable)"
(C. -T. J. Huang, 1984a: 561).
20 EXP refers to experiential aspect marker.
130
Thus, the null objects ( 45) and ( 46) are "not created by movement but must have been
generated in the base (and identified as a pro at D-structure, that is, an empty resumptive
pronoun) and become a variable by definition at the point where it is coindexed with its
local A'-binder" (C. -T. J. Huang, 1987: 331).
C. -T. J. Huang provides us with example sentences (47) and (48) (=Huang's (1984a:
560) (78) and (79)) to illustrate the point that the Subjacency Principle is available in
Chinese and there is subject-object asymmetry in Chinese.
( 4 7) a. Zhangsanb ei changge de shengyin hen haoting. Zhangsan sing DE voice very good-to-hear "Zhangsanb the voice with which [hei]sings is good."
b. *Zhangsani wo hen xihuan ei changge de shengyin. Zhangsan I very like sing DE voice "Zhangsani, I like the voice with which [hei] sings."
( 48) a. Zhangsanb ei xie de shu bu shao. Zhangsan write DE book not few "Zhangsanb the books that [hei] wrote are not few."
b. *Zhangsanb wo nian-le bu shao ei xie de shu. Zhangsan I read-LE not few write DE book "Zhangsanb I have read quite a few books are [hei] wrote."
In sentence (47a) and (48a), "extraction" is allowed from a complex NP in subject
position; but such an extraction is impossible from a complex object NP, as in (47b) and
(48b). When we take a look at sentences (47a) and (48a), it appears that the Subjacency
Principle does not apply to Chinese, because these grammatical sentences "constitute a
violation of Subjacency under the assumption that the sentence is derived by movement"
(C. -T. J. Huang, 1984a: 561). If this is the case, then (47b) and (48b) should be well-
formed in Chinese, but they are not. In (47a) and (48a), movement from the position of
131
the EC to that of John is excluded by Subjacency. In order to be admitted, e then would
have to be base-generated, and it is by the functional definition (see footnote 16) a
pronominal before coindexing takes place. C. -T. J. Huang argues that movement does
not derive ECs in sentences (47a) and (48a), but that they are base-generated
pronominals.
According to his GCR (see 4.2.3), in (47a), the EC as a pronominal must be identified by
the closest potential antecedent. There is no nominal closest within the relative clause
controlling e in either (47a) or (48a). The heads of the relative clauses, shengyin
(meaning, "voice") (in (47a)) and shu (meaning, "book") (in (48a)), are the closest
nominal elements, but they cannot serve as the controllers or antecedents for the ECs,
because they are already coindexed with something else.
C. -T. J. Huang (1984a: 561) argues that "the heads of a relativized contribution that is
already coindexed with something does not count as a potential antecedent of another
EC". If this stipulation is right, the next closest nominal element in ( 4 7 a) and ( 48a) is the
topic Zhangsan, which may be coindexed with the ECs. In this way, (47a) and (48a) are
thus well-formed because the ECs may start out as pronominals and become variables
when they are coindexed with the topics Zhangsan in accordance with the GCR ( cf.
Huang, 1984a: 561).
When we take a look at sentences ( 4 7b) and ( 48b ), however, the ECs cannot be derived in
the same manner, because the closest potential antecedents for the ECs are the matrix
132
subjects wo (meaning, "I"), not the topics Zhangsan, which leads to ungrammaticality of
the sentences. In a word, there are two distinct syntactic processes for the topics in C. -T.
J. Huang's approach: movement and base-generation. Movement is subject to the
Subjacency Principle, and the base-generation has something to do with the GCR.
4.3.2 Xu & Langendoen's and Xu's Approach
C. -T. J. Huang's analysis is challenged by some linguists, notably by Xu & Langendoen
(1985) and Xu (1986). In Xu's (1986) study ofECs in Chinese, he claims to have found a
totally new class ofECs fundamentally different from those defined in Chomsky (1982)
and he proposed the "free empty category" (FEC). Xu (1986) argues against C. -T. J.
Huang's approach by claiming that all embedded null objects are coreferential with topics
and not with matrix subjects, making them A'-bound almost unexceptionally.
Sentence (49) is one ofthe counterexamples given by Xu (1986) to illustrate his point
((49 =Xu's (9)).
( 49) Xiaotou yiwei mei ren kanjian e. thief think no man see The thief thought nobody saw (him).
Xu (1986) points out that the verb kanjian in (49) allows an empty embedded object to be
coreferential with the matrix subject. He claims that in sentences such as (50), hearers
would "naturally interpret the matrix subject as the antecedent of the embedded object
unless context compelled them to do otherwise" (Xu, 1986: 78).
133
(50) haizi yiwei mama yao zeguai e le child think mother will reprimand SFM21
"The child thinks his mother is going to reprimand (him)."
According to Xu, this is different from the particular sentence such as (51) cited by C. -T.
J. Huang (1982: 349) ((51)= (7) in C. -T. J. Huang) where hearers would be disinclined
to take Zhangsan as the controller of the embedded object without additional context.
(51) *Zhangsan shuo ni kanjian le PR022
Zhangsan say you see AM23
*"Zhangsan said that you saw."
C. -T. J. Huang (1982) argues that the empty object in the embedded clause cannot refer
to the matrix subject. He claims that only pragmatic and contextual factors may allow an
interpretation according to which coreference becomes possible. That is to say, sentences
like ( 49) are not pragmatically neutral. However, as X. G. Li (1988) points out, if the
matrix subjectXiaotou (meaning, "thief') in (49) is replaced with a semantically neutral
word, Zhangsan, the embedded null object can still be bound by the matrix subject, as in
(52). This example shows that matrix subjects can bind embedded null objects. Xu's
arguments against C. -T. J. Huang appear to be reasonable. It seems that C. -T. J. Huang's
analysis cannot handle language speakers who permit free co-indexing of empty objects
with matrix arguments.
(52) Zhangsani yiwei mei ren kanjian ei. Zhangsan think no man see "Zhangsan thought nobody saw (him)."
21 According to Xu (1986: 75), SFM refers to Sentence Final Marker. 22 C. -T. J. Huang (1982: 349) uses PRO to refer to e. 23 AM refers to Aspectual Marker.
134
4.3.3 Arguments by other linguists
Other linguists, such as X. G. LI (1988), Henry (1988, 1989) and Yuan (1997) also
challenge C. -T. J. Huang's position. However, these linguists do not take null objects as
belong to the FEC categories (see 4.3.2.) invented by Xu (1986), but regard it as pro
instead. C. -T. J. Huang (1984a) claims that null pronominal (pro) cannot occur in object
position in Chinese, it cannot occur in object position in any languages in the absence of
an identifying clitic or object-verb agreement (cf. Cole, 1987: 598).
However, as Henry (1988: 83) points out, C. -T. J. Huang's claims about the impossibility
of empty objects being a universal characteristic oflanguage is shown to be incorrect,
because it entails a claim that empty objects in all languages must be variables coindexed
with an empty topic. As Cole (1987: 602-603) points out, some languages such as Korean
and Thai do appear to have null pronominal objects because it "allows both null
pronominal and null variable objects". We agree with this view and regard Chinese as
belong to the same group as Korean and Thai explained in Cole's typology24•
4.3.4 Summary
In this section, we have reviewed and discussed the previous versions of analysis of the
null subjects as well as the base-generated topic in Chinese. We believe that the null
subject in finite clauses in Chinese is PRO and the lexically realized NP in the subject
24 Following C. -T. J. Huang (1984a, b: 1985), Cole (1987: 608) proposed that there are four types of null objects languages. They are: "(1) those not permitting either null pronominal or null variable objects; (2) those permitting null variable objects but not null pronominal objects; (3) those permitting null pronominal objects but not null variable objects; and ( 4) those permitting both null pronominal and null variable objects.
135
position has inherent Case rather than structural Case. Following C. -T. J. Huang (1984a),
we accept the view that there are two ways to generate a topic in Chinese: it can be base
generated, as is the case in Chinese, or it can be the result of movement. The topic in
English, however, cannot be base-generated; it can only be created by movement as a wh
trace.
We have seen in this section that C. -T. J. Huang's argument that the null object in
Chinese cannot be bound by an element in an argument position (A-bound) and that it
can only be analyzed as a variable is problematic. To say that the empty topic in Chinese
is properly governed by INFL is also an invalid view (C. -T. J. Huang, 1984a, 1989). It is
AGR and TNS that govern the subject position in finite clauses. However, since in
Chinese AGR and TNS are absent, it is a clear indication that the subject or topic in
Chinese is not governed, not to mention it being properly governed. A number of
linguists (such as Si (1988), X. G. Li (1988) and Henry (1989)) have challenged C. -T. J.
Huang's view of INFL in Chinese being a proper governor.
However, C. -T. J. Huang is correct in claiming that there are two ways of deriving a null
object in Chinese: It may be created through topicalization like a wh-trace; or it can be
base-generated. The null object is a variable in the former case, while it is a pro in the
latter. The null object created by movement and bound by an overt topic is a variable. As
there is strong evidence that INFL in Chinese is not a governor, C. -T. J. Huang's analysis
of the null object as a variable no longer holds; thus pro is the only interpretation left to
us. There is nothing to prevent pro in the object position from referring to an entity
outside the matrix sentence.
136
137
Chapter 5 Chinese and English Reflexives: An Introduction
5.1. Chinese and English Reflexives
5.1.1. Local and Long-Distance Binding: Theoretical Considerations
Most linguists agree that English reflexives must find their antecedents in the local
clause, whereas the Chinese reflexive ziji (meaning 'self) can take any subject Noun
Phrase (NP) to be its antecedent (Cole, Hermon & Sung, 1990; Cole & Sung, 1994;
Progovac, 1993; Wexler & Manzini, 1987). The so-called long-distance (henceforth, LD)
reflexives are attested in East Asian languages such as Chinese, Japanese' and Korean
(Cole, Hermon & Sung, 1990), but do not occur in all languages. English, on the other
hand, has only local reflexives, i.e. reflexives that must find their antecedent in the local
clause. Thus, in the following example (1 ), only Harry can serve as an antecedent for the
reflexive himself, but not Bill or Tom.
(1) Tom1 thinks BillJ knows Harryk likes himself *i!*Jtk·2
In addition to the reflexives in object position as shown in (1), the reflexive ziji can also
appear in preverbal positions, as in (2).
(2) Zhangsan1 renwei ziji1 neng huode aoyunhui menpiao.
Zhangsan think self can get Olympic ticket
* Zhangsan thinks that himself can get an Olympic ticket.
1 Japanese has a long-distance reflexive zibun (meaning 'self') as well as two types of local reflexives, karezisin, kanojo-zisin, etc. ('himself', 'herself') and zibun-zisin ('self-self'). The Japanese local reflexives are considered to be formal and are less frequent than zibun (White, Hirakawa & Kawasaki, 1995: 130).
2 These coreference possibilities are indicated with subscripts, the asterisk denoting impossibility of coreference. Examples (1) and (2) given here are quoted from Cole & Sung (1994: 355), i.e. their (1) and (2).
138
Embodied in Chomksy's (1981a) Government and Binding Theory, the basic explanation
for this locality condition (see 4.2) is that the first potential antecedent for the reflexive,
termed SUBJECT (read 'big subject'), closes off the domain for the reflexive. That is to
say, a reflexive cannot be bound across a SUBJECT, i.e. construed with an antecedent
across a SUBJECT. However, this locality condition proves too strong in the case of
morphologically simple reflexives, which consist of only the self morpheme (Yang, 1983;
Nakamura, 1987). In fact, the monomorphemic reflexives in some languages can "take
antecedents indefinitely far from the anaphor" (Cole & Sung, 1994: 355), i.e. be bound
across instances ofpotential antecedent SUBJECTs, as illustrated in the following
examples (3) from Chinese:
(3) Zhangsani renwei Lisij zhidao Wangwuk xihuan ziji i!j/k·
Zhangsan think Lisi know Wangwu like self 'Zhangsan thinks Lisi knows Wangwu likes him/himself.'
In example (3), ziji can refer to any of the three NPs, and the sentence is thus three-ways
ambiguous. While himself can only refer to Harry in example (1), ziji can refer to either
Wangwu, Lisi or Zhangsan in example (3). Apart from having monomorphemic reflexive
or bare reflexive ziji, Chinese also has what Cole et al (1990) call the phrasal reflexives
or morphologically complex reflexives3 in the form of "pronoun+ ziji" (such as ta-ziji 4
3 Depending on the preference of the individual, the two types of Chinese reflexive anaphors have been variously termed as foiiows (Wu, 1992:52):
(1) ziji (self): long-distance reflexive, bare reflexive, monomorphemic reflexive, nonphrasal reflexive;
(2) pronoun + ziji (himself, herself, yourself): short-distance reflexive, compound reflexive, polymorphemic reflexive, phrasal reflexive.
4 There are three different Chinese characters for ta depending on whether it is male, female, or inanimate third person. However, the three pronouns are not distinguished phonologicaiiy. For a detailed discussion of ziji and pronoun+ ziji, see Huang (1994). For the acquisition of ta ziji by Chinese children, see Chien, Wexler & Chang (1993).
'himself/herself or ni-ziji 'yourself), which is comparable to the English him-self I her-
self or your-self. Battistella & Xu (1990) suggest that the phrasal reflexive has the
structure in (4):
(4) NP
~ DET N
I I Pronoun ziji
Li and Thompson (1981: 137) treat the phrasal reflexive as a free variant of the bare
reflexive ziji, arguing that ziji "may optionally be preceded by a pronoun that is
coreferential with the subject of the sentence". Thus, we have the following phrasal
reflexives:
wo zijl myself
ni ziji -- yourself
ta ziji ~ himself
herself
itself
women -- ourselves
nimen --- yourselves
tamen --- themselves
139
As noted by Wang & Stillings (1984), the bare reflexive ziji is different from the phrasal
reflexive 'pronoun+ ziji' since the binding distance of phrasal reflexive is shortened. For
the phrasal reflexive 'pronoun+ ziji', only the closest subject is a possible antecedent. As
can be seen in the example sentence (4), like the case with himselfin English, ta-ziji
could only select the closest antecedent, Wangwu. LD reflexive binding, then, "cannot be
a language-specific, but rather a reflexive-specific property" (Bennett & Progovac, 1995:
79).
(5) Zhangsani renwei [Lisij zhidao Wangwuk xihuan ta ziji *il*jtk].
Zhangsan think Lisi lrnow Wangwu like he self
'Zhangsan thinks that Lisi lrnows that Wangwu likes himself.
140
In recent analyses of reflexive binding, it is assumed that there are two types of anaphors.
The LD/local properties are linked to these types, either through movement at logical
form (LF) (Cole et al, 1990; Katada, 1991; Pica, 1987; Reinhart & Reuland, 1993) or via
coindexation with agreement (AGR) (Aikawa, 1993; Progovac, 1992, 1993). Each type
has a cluster of properties associated with it, as shown in (5)
(5) a. xmax (phrasal) reflexives: morphologically complex, require local binding,
allow binding to subject and non-subject, e.g. himself, herself in English.
b. X0 (head) reflexives: monomorphemic, allow LD binding, require binding
to subject, e.g. zibun, in Japanese.
As we have seen in Chapter 2, Chomsky (1981a) proposes Binding Condition A (B, i, ii,
iii), given in (6), to account for the distribution of reflexives in English.
(6). A. An anaphor must be bound in its governing category.
B. A pronominal is free in its governing category.
i). a is a governing category for 13 if and only if a is the minimal category
containing 13, a governor of 13, and a SUBJECT accessible to f3.
ii). A SUBJECT for an anaphor is ~P, IP], [N:P, NP], or AGR.
iii). a is accessible to 13 if and only if 13 is in the c-commanding domain of a
and assignment to 13 of the index of a would not violate *[y .. . 8 ... ], where
y and 8 bear the same index.
The crucial property of (6) is that reflexives must have a syntactically prominent
antecedent (e.g. c-commanding) in a local domain 5( e.g. governing category or GC). In
other words, reflexives must obey a locality condition (GC) and syntactic prominence
141
conditions ( c-commanding). Thus, according to this theory, reflexive binding is syntactic
and observes structural constraints like c-command. However, the bare reflexive ziji in
Chinese does not obey the conditions of A in binding theory; ziji can have antecedents
outside its GC, behaving like a pronominal, and can be bound to a non-e-commanding
NP. For instance, the reflexive ziji in (2) can be replaced with a pronominal ta (meaning
'he') as in (7), and the sentence is still well-formed.
(7) Zhangsani renwei tai neng huode aoyunhui menpiao.
Zhangsan think he can get Olympic ticket
Zhangsan thinks that he can get an Olympic ticket.
5 Wexler & Manzini (1987) suggest that the local domain for the binding relation is defined as the Governing Category and it is parameterized among lexical items (i.e. the anaphors themselves) in different ways for different languages:
(1) Governing Category Parameter (henceforth GCP) y is a governing category for a if and only if y is the minimal category that contains a and: A. has a subject, or B. has an INFL; or C. has a TNS; or D. has an indicative TNS; or E. has a root TNS. (Wexler & Manzini, 1987: 53)
The relationships among the parameter values of GCP are inclusive, and exhibit subset relationships. For instance, when an NP is included in an NP with a subject NP, this NP with a subject NP is always included in a minimal category with an INFL; this minimal category with an INFL is always included in a minimal category with a TNS; and this minimal category with a TNS is always included in a minimal category with arootTNS. Therefore, only NP 1 can be the antecedent of the reflexive in (2), but no other NPs can: (2) [ROOTS TNS NP1 [TNs NP1 [INFL NPk [NPwitlt a subject NPr reflexive *V*J!*M]]]] Thus, in sentence (3), an anaphor has to refer to a NP within its local domain defined by its syntactic construction, an anaphor refers to NP2, but not to NP1 in (4): (3) Lewinsky 1 thinks that [Bi11 2 dislikes himself •;q]. (4) NP J .. • [urelocaldomaln NP z ... anaphor] The most relevant choice for Chinese ziji is (IE). But since there is no Tense in Chinese, the appropriate choice would be something like "the root sentence is the governing category" as suggested in Wang and Stilling (1984)(cf. Wakabayashi, 1996: 268).
142
5.1.2. C. -T. J. Huang's Approach
In order to account for the difference in behavior between simple and complex reflexives,
C. -T. J. Huang (1982) proposes a modified version ofthe Governing Category
(henceforth GC) Principle, as in 2.4.4 in Chapter 2 and repeated here in (8):
(8) a is a governing category for 13 iff a is the minimal category containing 13 and
a SUBJECT which, if 13 is an anaphor, is accessible to 13.
According to C. -T. J. Huang, the GC for ziji in (2) is the matrix sentence since the only
accessible subject it has is the matrix subject. In example sentence (7), the GC for ta is
the embedded sentence since it contains ta and a SUBJECT (namely ta), which does not
have to be accessible to ta. Thus, ziji in (2) is a reflexive and it is subject to Principle A of
the binding principle and is bound in its GC; and ta in (7) is a pronominal and it is subject
to Principle B of the binding theory and is free in its GC. However, C. -T. J. Huang's
modification fails to account for sentences such as (9), where the reflexive ziji is in object
position in the embedded sentence.
(9) Zhangsani renwei Lisij xiangxin zijiifj· Zhangsan believe Lisi trust himself
Zhangsan believes Lisi trusts himself.
In sentence (9), according to the GC principle, Lisi is an accessible SUBJECT to ziji; thus
the embedded sentence Lisi xiangxin ziji must be the GC for ziji. That is to say, ziji can
only take Lisi as its antecedent. Nevertheless, Chinese grammar allows ziji to have the
matrix subject Zhangsan as its antecedent as well, which indicates that the redefinition of
143
the GC principle is unable to provide an explanation for the possible behaviors of the
bare reflexive ziji.
5.1.3. Tang's Reindexing Approach
Tang (1989) has adopted a different approach to the LD-bound reflexives. She proposes
two language-specific rules for LD-bound ziji: a "feature copying rule" and a "reindexing
rule". She argues that these two language-specific rules are necessary to account for the
LD binding of ziji and the blocking effect. She claims that the LD binding of the bare
reflexive ziji to the matrix subject is blocked unless the subjects of all clauses between
ziji and the matrix subject happen to agree in person features. It is suggested that
"Binding Condition A needs to be modified to include two additional parameters: subject
orientation and sub-command6 conditions. The parameters account for the subject
orientation property of ziji and the fact that a sub-commanding NP can be the antecedent
ofziji" (Pan, 1995: 97).
According to the reindexing approach, ziji will get its features from the 'pro + ziji'
structure, in which the empty pronoun pro, after the application of the Binding Theory,
6 According to Tang (1989), a sub-commander is defined as an NP contained in a subject NP, which ccommands the reflexive. This notion can be schematically represented below:
s I \
NP VP I \ I \
NP NV NP
I I I I My pride hurt myself
Where the bold NP is a sub-commander
I44
may transfer optionally its features (such as person, number and gender) to -ziji, thus
turning -ziji into a LD reflexive. In other words, if the intervening subjects have
compatible features (or feature matching), ziji can be LD-bound. But if any one of them
has different features, then the blocking effect will be observed. After that, ziji undergoes
an obligatory and iterative reindexing rule, which reindexes ziji with the potential NP or
subject of the next higher GC. Although the reindexing approach works better than the
original Binding Condition A (Chomsky, 1981a) for ziji, it is an ad hoc solution. As noted
by Battistella (1989) and Huang & Tang (1991), there is no independent evidence
showing why a local reflexive can be turned into a LD one. Let us look at some example
sentences7 (10), (11) and (12) from Tang (1989).
(10) Woduede nii dui ziji*ilj mei xinxin.
I think you to self no confidence
"I think that you had no confidence in yourself."
(11) Zhangsanduede wo/nii dui ziji*ilj mei xinxin.
Zhangsan think I!you to self no confidence
"Zhangsan thought that Ilyou had no confidence in myself/yourself."
(12) Zhangsani zhidao wo/nii juede Lisii dui ziji*il*jlk mei xinxin.
Zhangsan know Ilyou think Lisi to self no confidence
"Zhangsan knew that I!you thought that Lisi had no confidence in himself."
Under the reindexing approach, the reindexing rule is applied iteratively and obligatorily,
LD binding is impossible if the feature-matching requirement between the LD antecedent
and the reindexed reflexive is not met. This is why the LD binding in sentences (1 0), (11)
7 The example sentences (IO), (II) and (I2) are from Tang's (I989) sentence (48), (50) and (51).
145
and (12) are disallowed. However, if we replace wo ('I') and ni ('you') in the above
example sentences (10), (11) and (12) with the third person pronoun ta or with another
name, then the matrix subject is a possible antecedent.
5.1.4. Battistella's Movement-to-INFL analysis
Battistella (1989) proposes to analyze the LD binding property of ziji in the 'movement-
to-INFL' framework first developed in Lebeaux (1983) and later adopted in modified
form in Chomsky (1986a) and Pica (1987). According to Battistella, reflexives in Chinese
move in LF :fi.·om argument position into INFL position (i.e. the head ofiP). The subject-
orientation8 of ziji is immediately accounted for through subject-head (AGR) agreement;
ziji gets its person, number and gender features through spec-head agreement from the
subject it is coindexed with. The LD binding of ziji is the result ofiNFL-to-INFL
movement of ziji at LF, as illustrated in (13).
8 Subject orientation refers to the fact that LD reflexives can be bound only to subjects but not objects, as illustrated in the following Chinese (a) and Russian (b) examples (Battistella & Xu, 1990; Rappaport, 1986). Cole & Sung (1994) consider subject orientation as one of the typological properties ofLD reflexives.
(a) Zhangsani songgei LisiJ zijiil"i de shu. Zhangsan give Lisi self De book "Zhangsan gave Lisi a book about him."
(b) Milicioneri rassprasival arestovannogoi o sebeil*i policeman questioned suspect about self "The policeman questioned the suspect about himself."
According to Battistella, INFL-to-INFL movement must be cyclic. By requiring
successive cyclic head movement to INFL, the blocking effect9 is explained, for the
features of the intervening first or second person pronouns will not be compatible with
those of the local third person subject from which ziji acquires its features. Battistella's
146
(1989:997) approach suggests that ziji moves to INFL, and that "INFL and the subject are
the locus of grammatical-agreement rules in UG to obtain the result that subjects of
clauses between ziji and its long-distance antecedent agree in person features". Once it
moved to INFL position, ziji or its trace behaves like the node AGR. "When ziji occurs in
the matrix clause at LF, the subject of each clause that ziji has passed through will have a
trace of ziji in INFL." As a result, the subjects of each of these clauses must agree with
one another since agreement will obtain between each subject and the AGR-like features
of the trace of ziji in INFL.
(13) [Zhangsan ziji-INFL renwei [Lisi 1,-INFL zhidao [Wangwu 1~-INFL xihuan t]]]
Zhangsan ziji thinks Lisi knows Wangwu likes
fu example sentence (13), ziji has moved from object position in the embedded clause to
the INFL of its own clause. From there, it further moves to the INFL of the intermediate
clause, and finally to the INFL of the matrix clause, thus explaining the possibility that
ziji might refer to Zhangsan or Lisi in addition to Wangwu. However, while the
Movement-to-INFL approach provides an insightful account for the LD binding in
9 In Chinese and other East Asian languages such as Japanese and Korean, when an immediately higher subject differs in person and number features from a lower subject, LD reflexives are blocked. This is the so-called "blocking effect" (cf. Cole & Sung, 1994: 363; Progovac, 1993:760).
147
Chinese, the reason why English does not have LD reflexives remains unclear, at least in
Battistella's attempt (Cole et al, 1990).
5.1.5. Cole et al and Cole & Sung's Head Movement Approach
Following Battistella, Sung (1990), Cole et al (1990) and Cole & Sung (1994) also take
the INFL-to-INFL head movement approach for Chinese reflexive ziji. In this theory, the
reflexive ziji may be raised in LF into INFL position (i.e. the head of IP), and may then
move from INFL to INFL (therefore, head to head), in consequence of the movement, the
reflexives' LD binding effects are explained. This is the reason why the head movement
theory is also called "INFL-to-INFL theory". Under this approach, the local reflexive is
treated as a full NP (i.e. a maximal projection) xmax, whereas the reflexive ziji is a zero
projection (i.e. a head element) X0• They assume that head-movement is a process of
adjunction rather than substitution. While an X0 element can adjoin to an X0 position,
xmax elements must adjoin to xmax positions (Sung, 1990).
The two different versions differ in their account of the reason why LD reflexive is
excluded in English. In Cole et al (1990), both X0 and XP reflexives undergo LF head-to-
head (INFL) movement. The reason that LD reflexive is excluded in English is that
English INFL is functional but not lexical. So it will fail to L-mark10 the higher VP; thus,
the higher VP is a barrier for the movement of the reflexive from lower C to the higher
10 A head a L-marks ~ if a is lexical and 9-govems ~· According to Chomsky (1986b ), lexical categories are L-markers and the L-marking determines what constitutes a barrier. Not every blocking category is a barrier. Non-L-marked maximal projections are barriers (except for IP), but L-marked maximal projections are not. See Chomsky (1986b), Cullicover (1997) for more concept ofL-marking.
148
INFL, causing an Empty Category Principle violation. Chinese has LD reflexive binding
because INFL in Chinese is lexical according to C. -T. J. Huang (1982) who assumes that
Chinese INFL is lexical to account for the non-existence of subject-object asymmetry11 in
NP extraction. However, the hypothesis that INFL is lexical does not hold water. As is
well known, Chinese verbs have no inflection except for the aspect marker; whereas
English has richer inflection than Chinese does. So, it is hard to justify the claim that
Chinese INFL but not English INFL is lexical.
Cole & Sung (1994) made a clear distinction between X0 and XP reflexives (i.e.
monomorphemic and polymorphemic reflexives) and claimed that only the
monomorphemic reflexive ziji can be LD bound. They argued that the polymorphemic
reflexives in English and Chinese (actually in all languages) can only have local
antecedents. Cole & Sung (1994) argued that there is a correlation between the emptiness
ofiNFL and the existence of the blocking effect in a language, because the blocking
effect only happens in Chinese, and it crucially depends on the emptiness of INFL.
As Pan (1995: 94) pointed out, the LF movement analysis of ziji is based on successive
cyclic movement and the assumption that the local subject will block the LD binding of
11 C.-T. J. Huang (1982) argued that Chinese lacks AGR in INFL. His evidence consisted of sentences like:
Zhangsan1 shuo s• [ziji1 hui lai]. Zhangsan say self can come 'Zhangsan said that himself will come'.
Such sentences indicate that the binding domain for the anaphor ziji is the matrix S' (sentence), because it is only in the matrix S' that ziji has a binder. This entails that there is no accessible SUBJECT (i.e. no AGR) in the embedded S' containing ziji. Huang claims that although certain subject-object asymmetries are widespread in the languages of the world, they do not occur in Chinese (Huang, 1982). Saying that Chinese lacks AGR provides an explanation.
149
ziji. However, not all the local animate subjects block LD binding-- only first person and
second person NPs do. But the LF movement analyses of reflexives fail to explain why
only first and second person NPs can block the LD binding of ziji. The LF movement
analysis of reflexives does not have an advantage with regard to the explanation ofthe
blocking effect. It makes wrong predictions regarding the LD binding in sentences like
(14).
(14) Zhangsan shuo wode/nide baogao ba [ta zijl]iJ*j haiku-le.
Zhangsan say my/your report BA he self hurt-hard-PERF
"Zhangsan said that my/your report hurt him very much."
Pan (1995: 94) argues that, the LD binding of the compound form ta ziji is unexpected in
Battistella's (1989) and Cole & Sung's (1994) approach. And for Cole et al (1990), it is
difficult to explain why the blocking effect is not observed in (14) where their percolation
process will predict that LD binding is impossible.
CP1
A Spec C'
~ c1 ~ ~ ~I'
Zhanlsan not a barrier
ziji r p2
~Lr~ (_J,:. 1\p
,.J (k,.ow)!\
not a barrier
Spec A C3~
J::."" t" I VP 3~
UJ~(UM) r Figure 1: Chinese (ziji in object position) t' N
L--t (Cole at al, 1990:6)
150
151
5.1.6. Relativized SUBJECT approach
Progovac (1992, 1993) proposes a relativized SUBJECT analysis for LD binding of
reflexives cross-linguistically. Observing the mixture of specifiers ([NP, IP], [NP, NP])
and heads (AGR) in the SUBJECT definition in (6ii), Progovac (1992, 1993) proposes to
relativize SUBJECT according to the X-bar compatibility dimensions of reflexives along
the line of Rizzi (1990), in the GB framework. Following Yang (1983) and Pica (1987),
Progovac claims that Chinese ziji is morphologically simple (i.e. X0, zero projection)
reflexive and therefore can only take X0 category to be its SUBJECT which is Agreement
(AGR), as in (12) and (13).
(12) An X0 reflexive must be bound to AGR, as the only salient ( c-commanding)
X0 category.
(13) AGR is the only SUBJECT for X0 reflexives. (Progovac, 1993: 672)
The fact that Chinese lacks morphological AGR is uncontroversial. However, this does
not mean that Chinese has no AGR feature12• Instead, Chinese has a syntactically null
AGR which is "anaphoric", or dependent on coindexation with higher AGR, as argued by
Borer (1989). Therefore, the LD binding of ziji is accounted for by the assumption that
there is an AGR in Chinese and following Borer (1989), this AGR is anaphoric. That is to
say, AGR is anN-type element which is referentially dependent on the subject13•
12 Aoun (1986) argues that Chinese lacks AGR in INFL since its verbs do not vary morphologically depending on the choice of subject. Lightfoot (1991:32) agrees that by saying that Chinese lacks AGR accounts for some puzzling similarities between English and Chinese.
13 The predictions are: (i) the local subject will always be the antecedent of ziji; (ii) LD binding of ziji is possible only if the lower AGR is compatible with, thus anaphoric to, the AGR in question; otherwise LD
152
Let us take a look at the following example (14).
Zhangsan think Lisi know Wangwu criticize ASP self
According to Progovac (1993), if AGR is bound to a higher AGR, the SUBJECT is now
the whole AGR chain, and the domain extends up to the head of the chain. In the example
sentence (14), ziji has to be bound to AGR3 in the local clause, which is anaphorically
linked to AGR2, which is further anaphorically linked to AGRr. By transitivity, ziji is
automatically bound to AGRr and AGR2. An "AGR chain" is formed when AGRs
(AGRr, AGR2 and AGR3) are matched with each other in pronominal features, allowing
ziji to be bound by the intermediate subject Lisi, the matrix subject Zhang san, as well as
the local subject Wangwu. However, when the features of all AGRs are not compatible
with each other, the AGR chain is broken. Consequently, only the local binding of ziji is
allowed. Hence the existence ofthe so-called "blocking effect" 14• That is to say, zijibeing
a bare reflexive, cannot take the matrix subject as its antecedent unless the subjects of the
clauses between ziji and the matrix subject all agree in person features. As example (15)
shows, the long-distance binding is blocked in Chinese when the different-person AGR
intervenes.
binding will be blocked-- the so-called "blocking effect"; (iii) subject-orientation is achieved by subject and AGR coindexation (Spec-head agreement) (cf. Pan, 1995: 70).
14 Tang (1985, 1989), quoting Y.-H.Huang (1984), claims that LD binding of ziji is possible only if all the subjects of the clauses intervening between the potential antecedent and ziji agree in person features; othetwise, LD binding is blocked. This is the so-called "blocking effect". However, the blocking facts are more complicated than presented in Tang (1989). Huang and Tang (1991) point out that an intervening subcommanding NP with different 9-features (which include person, number and gender) can also induce the blocking effect (cf. Pan, 1995: 27).
(15) Zhangsan AGR1 shuo [ WOj AGR2 zhidao [Lisik AGR3 chang piping ziji */*/k]]
Zhangsan say I know Lisi often criticize self
In this example, 1st person AGR2 intervenes between two 3rd person AGRs, AGR1 and
AGR2, preventing AGR-chain formation. As a result, the domain for ziji remains local.
5.2. Long-Distance Binding of ziji in Ll Acquisition
153
It has been reported that children acquiring different languages, whether LD or local
(including English), tend to bind their reflexives long-distance in the early stages (Wexler
& Chien, 1985; Hyams & Sigurjonsdottir, 1990). Wexler & Chien (1985) showed that
only twenty per cent of English speaking children at the age of 2 years 6 months have
local (adult) binding. In another study, Chien & Wexler (1987) compared Chinese and
English-speaking children's knowledge of reflexives and pronouns. They found out in an
act-out task (namely the 'party game'), that although it is perfectly grammatical in
Chinese for ziji to have a LD antecedent, Chinese-speaking children through age 6
showed a strong preference for the local binding of the reflexive ziji (Wexler & Chien,
1987; cf. Chien, Wexler & Chang, 1993). In the act-out task, the subject was encouraged
to choose between the LD antecedent and the local antecedent for the reflexive ziji, but
he/she was allowed to select only one antecedent. Thus, the result seemed to reveal the
subject's preference for a particular reading.
C. -T. J. Huang & Tang (1991) have argued that LD binding is caused by the movement
of ziji in LF to a non-argument (A') position, which is an adjoined position, not a subject
154
or object-like position. Using a yes/no picture-judgement task (i.e. whether a given
picture matches a given sentence), Chien, Wexler & Chang (1993) tried to suggest that
something is missing in the child's grammar which delays the LD binding and found that
most children preferred the local antecedent than the LD antecedent. As compared to the
act-out task, they claim~d that this yes/no picture-judgement task seemed to tap more
closely in to the subject's grammatical knowledge. Their explanations for the results are
that: either children did not move ziji at LF (in a sense of C. -T. J. Huang & Tang
(1991)); or children moved ziji at LF, but they did not have any problem in terms of
transferring referential features from a higher matrix subject NP to an NP in a non-subject
like or non-object-like position. They argued that children's preference indicated that they
did not interpret ziji locally simply because of a pragmatic or "minimal distance"
tendency. However, it seems that the work in the L1 acquisition of Chinese LD/local
binding has raised more questions than it has been able to answer (Chien, 1992; Chen,
1995; Chien, Wexler & Chang, 1993).
5.3. Long-Distance Binding of ziji in SLA
Reflexive binding is one of the major areas that have attracted great interests among the
L2 researchers. One of the controversial questions under discussion addresses L2
learners' access to UG as well as the resetting of parameters in SLA. As to the questions
of whether adult L2 learners can reset parameters and whether the Subset Principle is
operative in LSA, studies on UG access in SLA have inconclusive results. Some studies
claim that the Subset Principle is not operative in SLA (e.g. Finer & Broselow, 1986;
155
Thomas, 1989, 1991, 1993; Cook, 1990; Hirakawa, 1990; Lakshmanan & Teranishi,
1994), or that experimental data do no necessarily support this argument (e.g. Yuan,
1994). However, most researchers claim that UG is available and parameters are reset in
SLA (e.g. Finer & Broselow, 1986; Thomas, 1989, 1990, 1993; Hirakawa, 1990; Bennett,
1994; Eckman, 1994; Matsumura, 1994; Wakabayashi, 1996; Bennett & Progovac,
1998).
fu an experimental study of the cross-linguistic properties of reflexives in English,
Chinese and Spanish, Christie (1992) examined the issue of whether L2learners would
recognize the relationship between the binding domain and the proper antecedent in the
Head movement of anaphors at LF (Cole, Hermon & Sung, 1990; Sung & Cole, 1991).
The subjects include Spanish- and Chinese-speaking learners of English, English-
speaking learners of Spanish and English-speaking learners of Chinese. As regards the
Chinese, five out of seven advanced learners (i.e. English-speaking learners of Chinese)
allowed LD binding and three of those five allowed binding of ziji to either subject or
object. Christie concludes that the data from learners of Chinese is inconsistent with the
movement of ziji at LF, providing little evidence that LD binding is correlated with
subject orientation15 in L2learners' grammar. Thus, UG is not available to L2learners.
15 Subject orientation refers to the fact that LD reflexives can be bound only to subjects but not objects. Cole & Sung (1994) consider subject orientation as one of the typological properties ofLD reflexives. Please refer to this Chinese example sentence.
Zhangsan songgei Lisii zijiil*i de shu. Zhangsan give Lisi self DE book
'156
Another related study was conducted by D. -D.Chen (1995). Using a written stimulus-
question-answer task, D. -D.Chen investigated 8 English-speaking and 9 French-speaking
adults learning Chinese as a second language, as well as a control group of 28 educated
Chinese native speakers. Her findings revealed that L2 learners consistently chose local
binding. She argues that it may be the case that ziji does not present the nature of a LD
anaphor as much as is assumed in the literature. In other words, ziji might actually require
a local antecedent for a default interpretation, and this forces learners to favor the local
binding, even though ziji can be grammatically bound either LD or locally. She further
argues that unless a given context forces a learner to choose an antecedent beyond its
local domain, the default interpretation of ziji is a local anaphor. She also points out that
pragmatic factors should not be ignored when Chinese reflexivization is examined16•
Chen argues that learners' preference for local binding could be an indication that ziji
actually requires a local antecedent rather than a LD antecedent for a default
interpretation.
5.4. Summary
Languages differ as to whether they require reflexives to be local binding or long-
distance binding. Languages like English only permit the local binding of reflexives,
whereas languages like Chinese allow the antecedent of the reflexive to be local or long
16• D.D.Chen (1995: 11) revealed that there might exist a scenario: "ziji behaves more like a local anaphor
when it is presented in an isolated sentence, but a LD anaphor when it appears in a discourse context". Other linguists like Xu (1993, 1994) claim that semantic and discourse factors as well as syntactic factors are involved in the choice of an antecedent for ziji; Y. Huang (1994) argues that only pragmatic principles are able to characterize various properties of Chinese binding.
157
distance bound. English has only one type of reflexive, which is the phrasal reflexive.
Chinese, on the other hand, has the bare reflexive "ziji" as well as the phrasal reflexive
"pronoun+ ziji" (e.g. "ta-ziji"). The phrasal reflexives are maximal projections (Mmax).
Because of the so-called "blocking effect", i.e., barriers blocking their way up the tree,
the phrasal reflexives are unable to move out of their own clauses and thus can only take
a local antecedent. Since the phrasal reflexive can only take a local antecedent, it is also
called a short-distance reflexive (SDR). The bare reflexive, however, being able to move
out of its own clauses and undergo a successive cyclic head movement up the tree, it can
take long-distance antecedent as well as the local antecedent. Nevertheless, unless the
subjects of the clauses between ziji and the matrix subject all agree in person features, the
bare reflexive ziji cannot take the matrix subjects as its antecedent.
Chapter 6 Experimental Study
6 Introduction
In this chapter, I am going to look at the empirical study of this research by presenting
my empirical hypotheses and describing the research process involved. I will first
introduce the EFL (learners ofEnglish as a foreign language) and CFL (learners of
Chinese as a foreign language) informants involved in this study. Then, I will describe
the research procedures, as well as the administration and conducting of acceptability
judgment tests (AJT) with the EFL and CFL test groups.
6.1 Research Questions and the Experimental Hypothesis
158
Bloomfield (1933: 29) once argued that acquiring language is "doubtless the greatest
intellectual feat any one of us is ever required to perform". During the process of
learning, Chinese does present a lot of problems for the CFL learners. When talking to
some English speakers learning Chinese, one might certainly hear them complaining
about their pains in recognizing, memorizing and writing Chinese characters, about their
problems in pronouncing the same character with different yet correct tones, etc. During
the process oflearning, Chinese does present a lot of problems for the CFL learners.
Similarly, English learning is far from easy for some Chinese speakers, because even the
most advanced learners are found to commit serious grammatical errors in spoken and
written English. In fact, as two languages (Ll and L2) vary in lexis, syntax and
phonology, it would be expected that any L2 would cause difficulties in the learning
159
process. Only the degree of difficulty for the L2 learner is different. Depending on the
learner of the two different languages, both English and Chinese may serve as either the
mother tongue (Ll) or the target language (L2).
With respect to the language features involved in this study, such as the base/non-base
generated topics, null/non-null subjects, null/non-null objects as well as the long/short
distance reflexives, Chinese is more inclusive since it allows all the language features that
are allowed by the English grammar. At the same time, it allows language structures that
are absent in English. English is therefore less inclusive. For instance, Chinese allows not
only sentences having subject or object preserved, but also sentences having subject or
object dropped. English only allows sentences where pronominal subjects or objects are
obligatorily used. Therefore, the grammar of English is less wide than that of Chinese and
is properly contained within the Chinese grammar. The attempt of this study is to find out
the directional difficulties, if any, in the process of! earning Chinese and English for the
native speakers of English and for the native speakers of Chinese respectively. It is
hypothesized that directional difficulties are manifested in the different direct positive
evidences available to the L2 learner and these differences will have some impact on the
SLA of English and Chinese.
In terms of the language structures in SLA, if the Ll is more inclusive than the L2, the L2
learners have no direct positive evidence in the L2 input to help them undo the language
structures absent in the Ll. In this case, directional difficulties will arise. On the other
side, in terms of the language structures in SLA, if the L2 is more inclusive than the Ll,
the L2 learner will acquire the direct positive evidence, which indicates the difference
between L 1 and L2 with respect to the language features concerned and the change is
motivated.
160
With respect to the grammatical phenomena discussed here, Chinese is more inclusive
than English and consequently represents a superset of English grammar in this sense. As
far as the SLA of Chinese and English is concerned, it is hypothesized that Chinese
would be more difficult for English speakers to learn, while English would be less
difficult for Chinese speakers to learn. In terms of the language structures tested in this
study, it is hypothesized that:
(1). Concerning the null/non-null subject feature in finite sentences, it is predicated that it
would be much easier for the CFL informants to learn subject PRO in Chinese in contrast
with EFL informants learning English. Because CFL informants are exposed to a large
number of null subjects in the L2 input, they can acquire the null subject in Chinese quite
easily. However, the "undoing" process for the EFL informants learning English is much
longer since there is no direct positive evidence available to inform them that null subject
is disallowed in English.
(2). Concerning the null/non-null object feature, the acquisition of pro would be easier
for CFL informants than the undoing process for the EFL informants. Only CFL
informants have direct positive evidence available in the input.
161
(3). Concerning the base/non-base-generated topics, it would be more difficult for EFL
informants to learn English than for CFL informants to learn Chinese since direct positive
evidence is not available to EFL learner in the input. While CFL informants have direct
positive evidence in the input indicating the possible appearance of base-generated topic
in the TL Chinese, EFL informants have no way of !mowing that because English does
not allow the base-generated topic.
(4). Concerning the long/short-distance reflexives, it would be easier for CFL informants
to learn the Chinese long-distance reflexive than for the EFL informants to undo their
long-distance reflexives in their acquisition of English. EFL informants have no direct
positive evidence in their input showing that long-distance binding of the reflexive is
disallowed in the TL English.
In all the hypotheses above, the directional difficulty would be from Chinese to English.
We are going to test the above hypotheses in the acceptability judgment tests for both
EFLand CFL informants. This prediction of the directional difficulty in learning Chinese
and English as L2 maybe diagrammed as follows:
Chinese English
+null subject (PRO) -null subject (PRO) +non-null subject easier +non-null subject +null object (pro) ... -null subject (pro) ~
+non-null object +non-null object + base-generated topic .. - base-generated topic +non-base-generated topic +non-base-generated topic + long-distance reflexive harder - long-distance reflexive + short-distance reflexive + short-distance reflexive
162
6.2 Informants in the Experimental Study
More than 200 people served as informants in the experimental study. There were two
language groups, the EFL group and the CFL group. I will first look at the EFL group,
and then the CFL group. To prepare for the analysis of the informants' performances
according to their levels of command of the target languages, informants participating in
this study were divided into different proficiency levels/groups. Three sources of
information were used to group the informants: (1 ). The length oftime they had been
studying English at the time of the experiments; (2). The informants' total scores for the
distracter items of the experiment in the present study; (3). The informants' scores on the
cloze test and multiple choice test which they took at the onset of the present experiment.
6.2.1 The EFL informants
All together, 13 5 Chinese-speaking learners of English as a foreign language took part in
this study. There were 13 native speakers of English who served as controls. The EFL
informants include: 50 students studying English at Yandaojie Senior Middle School,
Chengdu, China; 67 students learning English in the Intensive Language Training Center
at Chengdu University of Science and Technology and Sichuan University, Chengdu,
China; 9 teachers of English (Chinese background) from Sichuan University; and 9
Chinese-background teachers who are visiting scholars in Sydney, Australia. Some
background information about the EFL informants can be found in Table 6-1: The EFL
Informants Background Information.
In the EFL Group, cross-sectional data were collected during the spring and summer
sessions of 1999. All subjects were native speakers ofMandarin Chinese. According to
the information gathered from the questionnaire the subjects filled out, 120 of the
subjects speak no other languages except Chinese and English; 18 of them speak some
Japanese; 6 of them speak some German or French, as a second foreign language.
Most of the subjects were learning English in classroom settings at the time of the
experiment. But some of them, especially those in EFL Group 5, who were Chinese
research students in Australia, have been exposed to a natural language learning
environment. According to their responses in the questionnaire, the length of time they
had been learning English ranged from 3 years to 12 years.
163
In the EFL group, informants were divided into 7 proficiency levels or groups. They are
lower and upper beginners, lower and upper intermediate, advanced and native groups.
As a result, the 148 informants were divided into seven groups, with 25 in EFL Group 1,
25 in Group 2, 23 in Group 3, 22 in Group 4, 22 in Group 5, 18 in Group 6, and 13 in
Group?. The designations of lower and upper elementary, lower intermediate,
intermediate, upper intermediate and advanced groups are, of course, relative. However,
they do approximate the respective grammar scores that are typically obtained by learners
who are at comparable levels of proficiency.
An attempt was made to collect additional data from learners of other English programs
at different overseas institutions. But due to restrictions imposed by the human subjects
consent office at the particular university, and the unavailability of the students of a
164
particular proficiency group, it was not possible to administer some experiments in a
classroom setting. However, it is assumed that the data reported on here were at least
moderately representative of the overall population oflearners of English from these six
groups oflearner background. It is assumed that the responses obtained generally
represent the actual state of their underlying linguistic knowledge. There were 13 native
speakers of English serving as controls in this research. They were native speakers of
English who are students or teachers around the Sydney area.
Table 6-1: The EFL Informants Background Information
Age No of Time living Native (Mean) years in English speaker
studying speaking of English country English
Group 1 16.44 4.28 None No
Group 2 16.61 4.64 None No
Group 3 29.74 8.83 None No
Group 4 30.77 10.18 None No
Group 5 31.73 11.86 0.52 No
Group 6 31.89 13.44 4.05 No
Group 7 29.54 N/A N/A Yes
Total
G1 =Senior 1, G2 =Senior 2, G3 = ILTC 1, G4 = ILTC 2, G5 =Visiting Scholar, G 6 =Teachers of English, G 7 = Controls (Native speakers of English).
6.2.2 The CFL informants
No of Subjects
25
25
23
22
22
18
13
148
The CFL group consisted of70 learners of Chinese as a foreign language and 16 native
speakers of Mandarin Chinese serving as controls. All these CFL learners were native
165
speakers ofEnglish, except for a few learners of German, Russian and Swiss background.
However, most native speakers of English would regard their English proficiency as
native-like, since they have lived in English-speaking countries for quite a long time. The
CFL informants were divided into 5 proficiency levels or groups. They are beginners,
lower intermediate, intermediate, advanced and native groups.
The 70 CFL informants included students studying Chinese at the University ofNew
South Wales, University of Sydney, Sichuan University and Beijing Language and
Culture University in China. More background information can be found in Table 6-2:
The CFL Informants Background Information. The 16 native speakers of Chinese serving
as controls in the CFL groups were in-service teacher training students majoring in
educational studies and Chinese studies at Sichuan Education College, Chengdu, China.
They had little knowledge of or exposure to English.
Table 6-2: The CFL Informants Background Information
Age No of Time living Native (Mean) years in Mandarin speaker
studying Chinese of Mandarin speaking Mandarin
country Group 1 23.28 0.89 0.66 No
Group 2 26.17 2.05 1.22 No
Group 3 27.88 3.29 2.11 No
Group 4 31.35 11.76 3.05 No
Group 5 26.87 N/A N/A Yes
Total
G 1 = 1st year & beginners, G2= 2nd year, G3= 3 rd year, G4= teachers, G5= control
No of Subjects
18
18
17
17
16
86
166
Most CFL informants are English-speaking students learning Chinese either in China or
Australia. Some CFL students have already spent some time in China. Different from the
teaching programs in China, most of the beginning CFL learners in Australia are more
exposed to the Chinese phonemic system pinyin, whereas the learners in China would
have been equally exposed to pinyin as well as the Chinese characters. In this study, most
of the informants in CFL Group 1 are 1st year beginner students from the University of
New South Wales and the University of Sydney. Informants in CFL Group 2 are 2nd year
students mainly from Sichuan University, China. Group 3 are 3rd year students from
Beijing Language and Culture University. Groups 4 mainly consists of teachers of
Chinese as well as some advanced students from Australia and China in the above
mentioned universities.
6.3.3 EFL informants and their background
Education in China is "divided into four categories, namely, basic education, secondary
vocational and technical education, regular higher education and adult education" (Li,
1997: 2). Basic education encompasses pre-school education, primary and general
secondary education. Primary schooling usually lasts six years; secondary education is
divided into two stages: junior secondary and senior secondary (i.e. junior middle school
and senior middle school), with each usually lasting three years.
In China, primary school normally begins at the age of 6, and lasts 6 years. At the age
around 12, children progress to a three-year junior secondary school. After another three
167
years in junior middle school, they complete their nine-year compulsory education. After
junior secondary 3, some would then go to the secondary vocational and technical
schools; and a few would leave formal education at this point to get jobs. Meanwhile,
most stay on for another three years studying in senior middle schools, and then take the
National University Matriculation Examination in order to go to various kinds of
universities. About twenty percent of those who take the matriculation exams will
succeed in gaining university admission.
In both junior and senior middle schools in China, English is taught as a compulsory
course, with two or three hours a week in junior middle schools and three to four hours in
senior ones. There are two sessions each year, with about twenty weeks in each term. In
the senior middle school where part of the experiments took place, there were native
English teachers teaching oral English for around one hour each week. Some students
also have private tuition in English outside classroom hours. The students have direct
contact with the native speakers of English, though the exposure time and the access to
English audio and video learning materials is quite limited.
Most normal undergraduate university courses in China last four years, with the
exception of medical courses lasting five years. For Chinese students of English at
various universities, they seldom have the opportunity to study, travel or work in English
speaking countries before their graduation. This is different from the foreign language
students here in Australia, where they can normally spend a year or so as an exchange
student in the target language speaking countries during their university study.
168
While conducting the experiment in Yandaojie Senior Middle School, Chengdu, China,
the researcher was able to sit in and observe some classroom teaching of English given
by some Chinese teachers ofEnglish. Most of the classroom teaching was predominantly
teacher-centered, based on traditional teaching methodology. Very few role-plays or
communicative methodology were used, except the "English comer" extra-curriculum
activity where students gathered together around a comer in the middle school to practise
their oral English once a week, thus improving their communicative abilities.
The learners from Chengdu University of Science and Technology and Sichuan
University had much greater access to English, such as easy access to language
laboratories and up-to-date audio and video language learning materials, as well as a lot
of contact with native speakers of English teaching in the Intensive Language Training
Center. English was used as the medium of instruction in the classroom setting.
The experimental tests were administered to the students in the Four-Skill Program and
TOEFL Program. The Four-Skill Program refers to the intensive training of students'
English in terms of the four core areas of Listening, Speaking, Reading and Writing.
Participants in the English training courses were highly motivated in learning English.
Most of them were fee-paying students who wanted to brush up and improve their
English skills in order to go abroad or in preparation for some English proficiency
examinations. Only a few of the students emolled in the Four-Skill Program and TOEFL
Program were high school graduates.
Many of them were university graduates who have already passed the National
University English Proficiency Test (Level Four) in China. That test, whose format is
very similar to that of TOEFL test, is designed to assess the graduates' English
proficiency levels before graduation. Those who failed the test at Level Four, which is
required for any would-be graduates, would not be awarded a bachelor's degree unless
they pass the test at a later date. Compared with the middle school students, these
students have a higher proficiency level in terms of their time of exposure to English
learning as well as their proficiency levels.
169
The next group of informants consisted of Chinese scientists or visiting scholars currently
living in Australia. They were advanced students, because not only have they studied
English for a relatively long time, but they had also passed various English tests, such as
TOEFL, IELTS, etc., and undergone intensive English training before their arrival in
Australia. They were upper intermediate learners of English in this EFL Group. The last
learner group consisted of English teachers of Mandarin background. They were the
advanced learners in this research. Finally, there were 14 native speakers of English
serving as controls in the EFL group and they were residents working or studying in
Sydney. Most of them hardly had any knowledge of Mandarin Chinese.
170
6.3. The Instruments
6.3.1. The Acceptability Judgement Test
In investigating principles and parameters, the researcher needs to find out whether the
language learners have knowledge of the ungrammaticality or acceptability of certain
sentence structures. The theoretical assumption is that the judgements of the subject as to
the (un) acceptability of a sentence to some extent reflect their underlying linguistic
competence. That is, respondents are asked to make acceptability judgements on a set of
sentences which included randomly mixed acceptable and unacceptable examples of the
linguistic structure under consideration, in order to access their knowledge of
unacceptability, and thus infer actual states of the learners' interlanguage grammars.
6.3.2. Arguments against and for the acceptability judgement tests
Acceptability Judgement Tasks (AJTs) or Grammaticality Judgement Tasks have come to
be used with some frequency in SLA research since the mid 1970s, to investigate
knowledge of universal principles and to attempt to access the linguistic competence of
L2learners in their L2 (White, 1988; Schachter, 1989). AJTs are ones in which sentences
are presented to subjects who are then asked to judge whether or not the sentences are
acceptable, correct, grammatical, well-formed, good and so on. There are both
advantages and disadvantages to using a grammaticality judgement task.
171
Some researchers argue against the use of grammaticality judgements in L2 research, and
argue for the exercise of great caution in their use. These researchers include Birdsong
(1989), Ellis (1990, 1991), Kellerman (1985), Christie & Lantolf (1992), Schachter &
Yip (1990) and Goss et al (1994). They have discussed a number of potential problems
with grammaticality judgement tasks, including issues of response bias and the use of
semantic (rather than syntactic) criteria in determining grammaticality, which might skew
the test results. Birdsong (1989) argues that grammaticality judgment tasks "can be
somewhat limited as to their informativeness because subjects could be exhibiting a
response bias in judging the sentences". According to him, performance variables such as
the mode and order of presentation, provision of context, balancing of test items, types of
response required, complexity of the test stimuli, the types of forms the subjects are
instructed to identify and so on, can have an effect on the learners' metalinguistic
performance. i.e. language activities that aim to discover the internal linguistic
lmowledge of the performers. The subject may simply not be capable of the
metalinguistic skills necessary to perform on the grammaticality judgment successfully.
Ellis (1991), however, cautions that performance constraints may actually influence
acceptability or grammaticality judgements and urges researchers to carefully validate
these instruments and how they are to be used to make claims about grammatical
competence.
However, many researchers, including Bley-Vroman et al (1988), Chaudron (1983), Gass
(1994), Munnich et al (1994), White (1989), Gowan & Hatasa (1994), Murphy (1997),
Cowart (1997), and Davies and Kaplan (1998) support the use of grammaticality
172
judgement tasks. Such task allows experimenters to construct the sentences in such a way
that the specific (or linguistic phenomenon) under investigation will be included (Cowart,
1997). These researchers argue that L2 acceptability/grammaticality judgements do
provide valid data for L2 research. Experimenters can have some control over the
experiment by "manipulating certain categorical variables so as to induce effects on a
continuous variable (acceptability)" (Cowart, 1997: 44). Subjects are forced to consider
the specific structure under investigation and the experimenter can include violations of
that structure (White, 1989). According to Cowan & Hatasa (1994), because they
eliminate the need for the learners to access performance-related processing strategies,
the use of grammaticality judgment tasks has the potential to provide a fairly direct
window into grammatical competence.
Although the experimental practice of eliciting grammatical judgements has been
criticized by some researchers, it is argued that the methodological advantages outweigh
any disadvantages that may be involved. In the absence of other practical elicitation
techniques, acceptability/grammaticality judgement tasks remain the most convenient
approach to collecting data related to a language learner's intuitive sense of syntactic
structure. As Gass (1994) pointed out, "by comparing judgment data with data based
more directly on linguistic principles, it has been shown that judgment data can, when
used properly and appropriately, be used in second-language acquisition research" and
that grammaticality judgments "are indeed reflective of patterns of second-language use"
(p.320). As a whole, I believe the acceptability/grammaticality judgement tasks are
suitable and valid means for the present study.
173
The Acceptability Judgements Tests in the present experiment consisted of72 items (see
Appendix I for a complete version of the English Acceptability Judgement Tests). Since
judgements on the designed sentences might be subtle, test participants were given a 7-
point scale on which to judge a sentence1• The respondents were asked to concentrate on
how they felt about the sentence, and using the scale to indicate how acceptable or
unacceptable they considered them to be. Training in judging sentences using the scale
was given with Chinese and English sentences shortly before the test delivery (see
Questionnaire Instructions in Appendix III: English Instructions and Chinese Instructions
in Appendix IV).
While doing the test, the informants were asked to judge the acceptability of the
experimental sentences. For each question, informants were asked to assess the degree of
confidence they had in their judgment by ticking the Likert-like2 (Likert, 1932) 7-point
score answers like the following:
-3 = completely unacceptable -2 =fairly unacceptable
-1 = slightly unacceptable 0 = unable to decide
+ 1 = slightly acceptable +2 = fairly acceptable
+3 =completely acceptable
1 See Juffs (1996: 183) for an example of such a scale in L2 acquisition experiment.
2 The Likert score (Likert, 1932) asks individuals respond to a series of statements by indicating whether they "strongly agree", "agree", "are undecided", "disagree" and "strongly disagree" with each statement. "Strongly agree" may be assigned a weight of 5 points, while "strongly disagree" may get a score of 1.
174
A value of"-3" meant that informants were 100% certain that the sentence was
unacceptable, and "+3" meant that they were 100% certain that the sentence was
completely acceptable. Other values indicated different degrees of certainty, with "0"
signifying Unable to Decide or Not Sure.
Finally, to control for response bias, there were 8 distracter sentences (5 grammatical and
3 ungrammatical) which had nothing to do with the target sentences in the experiment.
There were two versions of the same questionnaire to control for order of presentation.
6.3.3. Design Considerations
Due to the extensive investigation and discussion on the use of the acceptability and
grammaticality judgement data in empirical research (see 6.4.3), it has been realized that
the control of secondary variables is a very important issue in SLA research. In order to
produce clear results for the present study, careful consideration was taken to improve the
experimental design, the administration of the questionnaire and the test procedures.3
Several secondary variables were carefully controlled before the cross-sectional
investigation was carried out. One of the most common problems found in the acquisition
of syntactic structures is that the observation of data may be contaminated by
3 According to Schutze (1996:169), many stimulus and procedure variables can have effects on grammaticality judgments. "Serial order, repeated presentation, deliberate judgment strategies, modality, register, preparation, and judgment speed are all features of the elicitation task that might contribute systematically to variation in judgments. So might stimulus features, including the various types of contextual material, the meaningfulness of the sentence, the perceived frequency of the sentence structure, and idiosyncratic properties of its lexical item." He pointed out that "perhaps the biggest lesson is the importance of the instruction we give to subjects" (Schutze, 1996: 169).
175
performance factors or secondary factors, such as the use of vocabulary unfamiliar to the
subjects. Effort has been made to limit the number of new words that might cause
understanding problems for the subjects. Most of the vocabulary used in the experiment
sentence was limited to basic words and familiar use as much as possible. To guarantee
that subjects knew all the words used in the test sentences, a copy of the Chinese-English
and English-Chinese vocabulary list (see Appendix 4 and Appendix 5) were provided.
Sentence length and complexity are directly related to difficulty (see Cowart, 1997:46).
The experimental sentences were carefully controlled in length and they were designed to
be as simple as possible. There were two token sets of sentence structures of concern in
each category. The range for the number of words in the target sentences was 3 to 17.
One set of sentences corresponded to a particular sentence structure, with each set of
sentences consisting of two control sentences and two experimental sentences in most
cases (see Appendix 6 & 7). It should be noted that the test sentences were designed in a
way that pairs of the control and experimental sentences were identical except for the fact
that they are unique in terms of the aspect of concern. It is hoped that the subjects'
judgments of control and experimental sentences were based on their differences in the
aspect being tested.
Wherever necessary, the general situation for each test item, indicating the content of the
target sentence and its immediate discourse context, was provided. The purpose was for
the informants to recognize the discourse situation while avoiding the difficulty of
understanding a single sentence without context. The informants were told to make their
176
judgments according to this context. In order to make sure these experimental sentences
were appropriate, native speakers of Mandarin and English verified all the sentences,
during the experimental design.
Instead of asking informants to make absolute judgments of right or wrong, the
researchers decided to ask the subjects to make judgment on the appropriateness or
acceptability of the seven scales from "completely unacceptable" to "completely
acceptable". In the CFL group experiment, both Chinese characters and the pinyin system
were used in order to avoid the difficulty of reading and writing Chinese characters in the
questionnaire.
6.4. The Administration and Test Procedure
The researcher administered the acceptability judgement task to two types of informants.
They are the Chinese learners ofEnglish (i.e. EFL students) and the learners of Chinese
(CFL students).
Permission was obtained from individual instructors and appointments were made to
guarantee that enough class time was available for completion of the test. Before
delivering the experiment sentences, the experimenter explained to the subjects the
purpose of the testing session and informed the subjects that this was an experiment. No
personal results were to be recorded officially or disclosed to other parties and whatever
response they produce would be useful and interesting to the researcher. Most of the
177
explanations and instructions during the tests were given in the subjects' native language,
or second language on a few occasions.
An informant's Background Information sheet was provided to the test participants prior
to the testing session. In this way, individual differences on potentially relevant factors
such as age, sex, educational background and second language background were noted on
the questionnaire so that variability attributable to them can be examined later in the
analysis.
The instructions for the tests were delivered to each of the subjects on a separate piece of
paper before the testing session started. The instructions were read aloud by a native
speaker in the subjects' native language. Prior to completing the grammaticality
judgement task, the subjects were required to complete the five practice test items at the
beginning of the test pamphlet to familiarize themselves with the test format. This was to
give the subjects some experience of the conditions under which the test was to be
administered. The practice test and the acceptability judgement task, both of which were
pen-and-paper tests (as opposed to computer-administered tests), are provided in their
entirety in the Appendix 6 & 7. After the completion of the practice examples, a short
discussion time was allocated to subjects, if they have any questions regarding the
response they had made to the five practice examples.
After that, each subject was given a copy of the Chinese-English or English-Chinese
vocabulary list (See Appendix 4 & 5) and was required to memorize the unfamiliar words
178
in advance. During the experiment design process, effort was made to limit the number of
new words that might cause understanding problems for the subjects. The subjects were
told that they would be allowed to consult the vocabulary lists while doing the tests.
Every effort was made to ensure that subjects understood the meaning of the testing
session and how to do the test. The subjects were told to provide "gut reactions"
regarding the acceptability of the sentences, and were instructed to refrain from going
back to a previous item once a judgement had been made. When subjects felt comfortable
with the test format, and all questions had been answered, the acceptability judgement
task was distributed. The 72 task items appeared on 5 single sheet of A4 size paper.
Although all subjects were given ample opportunity to complete the test, they were told
that it should be completed within a 35-minute time limit. The experimenter carefully
watched the subjects during the testing session to make sure that subjects did not go back
and double-check their answers when they had completed the test. When all subjects had
finished and all task forms had been collected, the experimenter thanked the class and
answered any final questions about the purpose of the research.
The students studying in the same class (i.e. Senior Class 1, Senior Class 2, Four-Sldll
Program class, and TOEFL class) were taken as a testing group. There were two separate
testing sessions for each testing group. The randomized sets of the acceptability
judgement tests were administered in the first testing session, followed by the Chinese
and English cloze tests and multiple choice tests to all of the EFLand CFL subjects
respectively, including the native controls in the two language groups.
179
On some occasions, the informants completed testing sessions individually; especially for
those additional subjects for whom the researcher had no control over the completion of
the questionnaire and the feedback was sent back from overseas. With the assistance of a
computer specialist, some feedback was collected via Internet by the survey questionnaire
published online as a homepage. In this way, the researcher could obtain extra subjects, if
necessary, and the result is quite efficient.
6.5. Procedures of the Data Analysis
After the data collection, the collected data of questionnaires was transferred to coding
sheets by hand, and then transferred to the SPSS (Nie et al, 197 5) computer package
database. SPSS refers to the Statistical Package for Social Sciences and its updated
version SPSS 8.0 (SPSS, 1999). It is the most commonly used statistical software
worldwide.
On the coding sheets, each item is represented by a column. Thus, column 1 represents
the id (identity) of the informant whose data is being coded. Column 2 and 3 are the
informants' sex (1 for male, 2 for female) and age. Column 5 represents the informants'
years of exposure to English learning. Column 6 presents the informants' proficiency
level, or proficiency group, from Group 1 "elementary" to Group 6 "advanced" learners,
and Group 7 being the native speakers' group. Column 8 and plus are the answers to the
acceptability judgment sentences. After the data were entered onto the computer
180
database4, a printout of the data was obtained and this was compared with the original
coding sheets to check that the information was entered accurately, thus avoiding missing
values.
The computer is essential for this analysis of quantitative research data, since it can
handle complex analyses oflarge amounts of data in a very short time and at a very
reasonable cost. SPSS version 8.0 for Windows5 (SPSS,, 1999) is used in this research to
perform the analyses needed. After the data was entered into the computer in a sensible
and organized way, I designed a table displaying the research hypothesis and statistical
procedure.
Research Hypothesis: There is no effect of different language proficiency group on informant's AJT judgment scores (i.e. the means of the EFL/CFL Group will not differ).
Significance level? .05 1- or 2-tailed? 2-tailed
Design: Dependent variable(s)? Measurement? Independent variable( s )? Measurement? Independent or repeated measure? Other features? Statistical procedure? ( Cf. Hatch & Lazaraton, 1991: 317)
AJT Judgment scores Scores (interval) EFL/CFL Groups Nominal (7/5levels) Independent groups Fixed effect design One way ANOVA
4 SPSS has been modified for use with desktop computers and its input format is similar to that used by software such as Excel 97. Each variable is assigned a distinct column into which the data for that variable are entered.
5 SPSS 8.0 for Windows (SPSS, 1999) is a comprehensive system for analyzing data. SPSS can take data from almost any type of file and use them to generate tabulated reports, charts, and plots of distributions and trends, descriptive statistics, and complex statistical analyses. The URL for the SPSS home page is http://www.spss.com/.
181
As the above table shows, the Hypothesis for the study is that there is no difference in
AJT scores for different EFLICFL Groups. The dependent variable is AJT scores. The
independent variable is EFLICFL Groups. This is a one-way design because there is one
dependent variable and only one independent variable (with 7/5 levels). I assume that the
AJT test scores are interval data and this is a fixed-effect design.
6.6. SPSS and one way ANOV A analysis
According to Cowart (1997: 134), analysis of variance (or commonly, ANOV A) "makes
it possible for a researcher to apply overall control to a set of comparisons so that the
threshold value the researcher selects at the outset is the one that actually obtains in the
analysis of results". AN OVA procedures allow the researcher to make many comparisons
within a single data set simultaneously.
The goal of ANOV A is to explain the variant in the dependent variable in terms of
variance in the independent variables. "In a one way design, there is only one dependent
variable and only one independent variable with three or more levels" (Hatch &
Lazaraton, 1991: 312). The comparisons ofthe means on the dependent variable are
made across the levels.
With the data entered onto the SPSS database, we can compute the within-group and
between-groups variances fairly easily and then find the F ratio. To conduct a one way
analysis of variance (ANOVA) on SPSS, we select the Statistics menu, click on Compare
182
Means and One WayANOVA ... to open the One WayANOVA dialogue box, and follow
the rest of the procedures. We can obtain the ANOVA table and the Test of Homogeneity
of Variances.
The F ratio is the ratio of between-groups variance to within-groups variance, as
indicated below:
F = Between group variance/Within group variance
One way analysis of variance is "performed on the variance of the groups, focusing on
whether the variability between the different groups is greater than the variability within
each of the groups" (Seliger & Shohamy, 1989: 232). Thus, a significant F will occur
when the variability among the groups is greater than the variability within each group.
When the reverse is true, the F is likely to be insignificant because that means that
differences among the groups are not large enough to be significant6• A significant F
value tells us that the population means are probably not all equal.
A one way analysis of variance will result in an F value. That F value is entered into the
F table, which can be found in most statistics textbooks, to find out whether, given the
size of the sample, F is statistically significant. If it is significant, the researcher can
6 Statistical significance "related to the confidence" which the researchers can have in the correlations obtained. "The significance of the correlation depends on the size of the correlation and the size of the sample from which it has been obtained" (Seliger & Shohamy, 1989: 220). When a correlation is found to be significant statistically, the researcher can reject the null hypothesis that states that there are no significant correlations between the variables. The p-value on which the rejection is based is referred to as the level of significance. The conventional minimum p-value needed to reject the null hypothesis is p s;;; .5 orp s;;; .1.
183
reject the null hypothesis7• "However, the researcher does not lrnow where, in all these
comparisons, the significant differences are" (Seliger & Shohamy, 1989: 235). So, when
researcher obtained a significant F, they need to perform subsequent analysis to find out
where in the data the significance occurs.
This requires post hoc analysis. "Methods that allow for the testing of all differences are
lrnown as post hoc (after the event) procedures" (Lockhart, 1998: 328). Post hoc
procedures allow the researcher to explore the data and evaluate whatever differences
appear interesting. Post hoc analysis is when we hunt through the data for any
significance. To conduct a one way ANOV A with post hoc analysis, click on the Post
Hoc .•. command push bottom to open the One Way AN OVA: Post Hoc Multiple
Comparisons sub-dialogue box. Then, click on the check box for Tukey in the number of
multiple comparison options available. Finally, click on Continue and then Ok. A table
of Tukey HSD Multiple Comparisons is generated by SPSS.
In interpreting this output, we must ensure that the homogeneity assumption has not been
violated. If Levene's test for homogeneity of variances is not significant (P ~ 0.5), we can
be confident that the population variances for each group are approximately equal.
To determine whether we have a significant F ratio, we use the df (degree of freedom),
the F-value and F-probability. The correct way to determine significance is to use the
critical F tables. Significance can also be determined by looking at the F-probability
7 The usual symbol for the null hypothesis is Ho; the "H stands for hypothesis and the subscripted zero indicates that something in that hypothesis is zero, or null" (Cooper, 1999: 2-40).
184
value. lfp c .05, we can reject the null hypothesis and accept the alternative hypothesis
which states that AJT scores are different across EFL/CFL groups.
Having obtained a significant result, we can go further and determine, using the Tukey
HSD test (SPSS, 1999: 76; Cooper, 1999: 7-15), where the significance lies. The result of
the Tukey HSD test identifies where differences in EFL/CFL Groups lie. Moreover, if the
homogeneity of variance assumption has been violated (p c .05), thus any interpretation
of the main effect must be undertaken with caution.
185
Chapter? Analysis of Experimental Data
7 Introduction
In this chapter, I am going to look at the acquisition and unlearning process of the null
subject and null object by the EFL and CFL learners respectively. I will also examine the
directionality of difficulty in the acquisition and unlearning of the base-generated topics
by EFL and CFL informants.
7.1. Test sentence structure for EFL informants
7.1.1 Test sentences with null subjects for the EFL informants
Five sentence structures, illustrated in (1-5), were designed to test whether the EFL
learners were aware that, unlike their Ll Chinese, English does not allow null subject
(PRO) to occur in the subject position of the finite sentences. These sentences involve the
null and non-null subjects in the acceptability judgment for the EFL informants. The
format of the acceptability judgment sentence made it possible to test a number of
variables relevant to this study.
For each sentence structure, there were two control sentences and two experimental
sentences. Each control sentence and its matched experimental sentence were deliberately
designed as identical to the other in every aspect except for the linguistic feature being
investigated. In this way, it was hoped that any difference that emerged in the judgments
between the control sentence and the experimental sentence by the informant would
reveal findings relevant to the linguistic feature under investigation. Sentences la-5a
186
were examples of experimental sentences, and sentences 1 b-Sb were examples of control
sentences.
(1) Sentences with null subject in matrix sentences (SEFL/NS-MS)
(1a). Experimental
*Johnson has bought a new refrigerator. Is made in China.
(1 b). Control
Johnson has bought a new refrigerator. It is made in China.
(2) Sentences with null subject in embedded sentences (SEFL/NS-ES)
(2a). Experimental
*I asked David whether our coach would come today, David said would certainly come.
(2b). Control
I asked David whether our coach would come today, David said he would certainly come.
(3) Sentences with both null subject and null object (SEFL/NS-NO)
(Sa). Experimental
*He has promised to buy his daughter a new toy, but so far hasn't bought yet.
(5b). Control
He has promised to buy his daughter a new toy, but so far he hasn't bought her one yet.
(4) Sentences with absence ofweather-predicate expletive (SEFL/Exp-WhP)
(3a). Experimental
*Suddenly, snowed heavily.
(3b ). Control
Suddenly, it snowed heavily.
(5) Sentences with absence of raising-predicate expletive (SEFL/Exp-RP)
(4a). Experimental
*Appears that they don't know this matter.
( 4b ). Control
It appears that they don't know this matter.
7.1.2 Test sentences with base-generated topics or non-gap topics for the EFL informants
187
Three sentence structures with base-generated topics and non-gap topics, illustrated in (6-
8), were designed to find out whether EFL learners were comfortable with the fact that, in
English, topics cannot be based-generated with sentential subjects.
(6) Sentences ofbase-generated topics with sentential subject (SEFL/BGT-SS)
(6a). Experimental
*That photocopier it is very likely that you won't be able to use now.
(6b). Control
It is very likely that you won't be able to use that photocopier now.
(7) Sentences ofbase-generated topics with Wh-island (SEFL/BGT-Wh)
(7a). Experimental
*This movie I don't know why he doesn't like.
(7b). Control
I don't know why he doesn't like this movie.
188
(8) Sentences with non-gap topics (SEFL/NGT)
(Sa). Experimental
*Her best friends, I have only met Lily.
(8b ). Control As regards her best friends, I have only met Lily.
7.1.3 Test sentences with null objects for the EFL informants
Seven sentence structures, illustrated in (9-14), were designed to find out whether the
EFL informants were comfortable with the fact that, unlike their L1 (Chinese), English
sentences with object pro (or null object) are unacceptable in the target language
(English). As we have indicated before (see 4.2), in Chinese, the third person pronoun in
object position referring to an inanimate entity is obligatorily absent. An attempt was
made in the experimental design to distinguish between an inanimate object pro and an
animate object pro, to find out whether such a distinction would be reflected in the
informants' judgments.
(9) Sentences with null object in sentential subject (SEFL/NO-SS)
(9a). Experimental
He is using that pen right now. *It is likely that you won't be able to borrow right now.
(9b). Control
He is using that pen right now. It is likely that you won't be able to borrow it right now.
(10) Sentences with null object in Wh-island (SEFL/NO-Wh)
(lOa). Experimental
*I like this novel very much, but I don't know why she doesn't like.
(lOb). Control
I don't know why she doesn't like this novel.
(11). Sentences with inanimate null object in main sentence (SEFLIINO-MS)
(lla). Experimental
Johnson's bicycle has gone wrong. I am going to repair for him.
(11b). Control
Johnson's bicycle has gone wrong. I am going to repair it for him.
(12). Sentences with animate null object in main sentence (SEFL/ANO-MS)
(12a). Experimental
*David says John has got a new girlfriend, but I haven't met yet.
(12b ). Control
David says John has got a new girlfriend, but I haven't met her yet.
(13). Sentences with inanimate null object in subordinate sentence (SEFLIINO-SS)
(13a). Experimental
*Lily lost her car last week, but David says the police have found for her.
(13b). Control
Lily lost her car last week, but David says the police have found it for her.
(14). Sentences with animate null object in subordinate sentence (SEFL/ANO-SS)
( 14a). Experimental
189
*John said those clients were in the hotel lobby, but I told him I did not find there.
(14b). Control
John said those clients were in the hotel lobby, but I told him I did not find them there.
(15). Sentences with inanimate null object coindexed with an argument in an adjunct (SEFL/INO-Adjunct)
(15a). Experimental
*Ifyou write a good book, I will defmitelybuy.
(15b). Control
If you write a good book, I will definitely buy it.
7.2. Test sentence structures for the CFL informants
7.2.1. Test sentence with null subject for the CFL informants
Five sentence structures for the CFL informants were designed to find out whether
English-speaking learners of Mandarin Chinese accept null subject in Chinese finite
clauses, and whether they accept sentences with both null subject and null object
structures.
190
Chinese does not allow the insertion of nominal expletive "it" in the language. However,
expletive "it" often appears in English in terms of weather verbs (e.g. "It is raining") and
in terms of raising constructions (e.g. "It is difficult"). The division of the test sentences
according to the presence or absence of the nominal expletives made it possible to find
out whether the subjects judged the sentences differently.
In English, topics are normally created through movement (see 4.2.7), whereas in
Chinese, those topics are base-generated, i.e. not through any kind of movement. The
base-generated topics in Chinese are quite acceptable and grammatical in nature. As in
sentences (16-22), Chinese sentences with base-generated topics are included to find out
whether the CFL informants would regard them as (un) grammatical or accept them.
191
In the test sentences for the CFL informants, each control sentence corresponds to an
experimental sentence, as in (16a), (17a), (18a), (19a) and (20a). It should be noted that,
however, some control sentences in the test sentences for CFL informants were not
grammatical and not acceptable in Chinese. Because of the insertion of the nominal
expletives, test sentences like (16b) and (17b ), for instance, are not grammatical.
(16). Sentences with null subject in matrix sentence (SCFL/NS-MS)
(16a). Experimental
Lao Li mai le yi tai xin de binxiang, shi zhongguo zao de. Lao Li buy PFV1 one CL2 new DE3 refrigerator is China made DE
*"Lao Li has bought a new refrigerator. Is made in China".
(16b). Control
Lao Li mai le yi tai xin de binxiang, ta shi zhongguo zao de. Lao Li buy PFV one CL new DE refrigerator it is China made DE
"Lao Li has bought a new refrigerator. It is made in China."
(17). Sentence with null subject in embedded sentence (SCFL/NS-ES)
(17a). Experimental
wo wen Xiao Wang jiaolian jintian lai bu lai, I ask Xiao Wang coach today come not come
Xiao Wang shuo yiding hui lai. Xiao Wang say certainly will come
*"I asked Xiao Wang whether our coach would come today. Xiao Wang said would certainly come".
(17b ). Control
wo wen Xiao Wang womende jiaolian jintian lai bu lai,
1 PFV refers to perfective aspect marker. 2 CL refers to the classifier. 3 DE refers to the modifying marker in Chinese that occurs at the end of a prenominal modifier.
I ask Xiao Wang our coach today come not come
Xiao Wang shuo ta yiding hui lai. Xiao Wang say he certainly will come
"I asked Xiao Wang whether our coach would come today. Xiao Wang said he would certainly come."
(18). Sentences with both Null Subject and Null Object (SCFL/NS-NO)
(18a). Experimental
shang ge yue ta daying gei nu'er mai yi ge xinde last CL month he promise give daughter buy one CL new
danshi zhidao xianzai ye mei mai. but until now still not buy
wanju, toy
*"Last month he promised to buy his daughter a new toy, but so far hasn't bought yet. "
(18b). Control
shang last
ge yue ta daying gei nu'er mai yi ge xinde wanju, CL month he promise give daughter buy one CL new toy danshi zhidao xianzai ta ye mei mai. but until now he still not buy
"Last month he promised to buy his daughter a new toy, but so far he hasn't bought her one yet."
(19). Sentences with absence ofweather-predicate expletives (SCFL/Exp-WhP)
(19a). Experimental
*turan, ta xia qi le da xue. Suddenly, it fall start CL big snow
"Suddenly, it snowed heavily."
(19b). Control
turan, xia Suddenly, fall
qi le da xue. start CL big snow
"Suddenly, snowed heavily." (ungrammatical in English)
192
(20). Sentences without raising-predicate expletive (SLCFL/Exp-RP)
(20a). Experimental
haoxiang Xiao Wang sheng bingle. seem Xiao Wang get sick PART
"Seems that Xiao Wang is ill." (ungrammatical in English)
(20b ). Control
*ta haoxiang Xiao Wang sheng bing le. It seem Xiao Wang get sick PART
"It seems that Xiao Wang is ill."
7.2.2. Test sentence with base-generated topics or non-gap topics for the CFL informants
Three sentence structures with base-generated topics and non-gap topics, illustrated in
(21-23), were designed to find out whether CFL informants were comfortable with
Chinese sentences with base-generated topics and would regard them as acceptable. In
193
the CFL informant's Ll, i.e. English, topics are normally created through movement, the
grammatical Chinese sentences like (21 b), (22b) and (23b) are designed to find out
whether CFL informants judge them differently.
(21). Sentences with base-generated topics and sentential subjects (SCFL!BGT-SS)
(21a). Experimental
na tai chuanzhenji ni xiang xianzai yong shi bu keneng de. that CL fax you want now use be not possible PART
*"That fax machine that you want to use now is impossible".
(21b). Control ni xiang xianzai yong na tai chuanzhenji shi bu keneng de. you want now use that CL fax be not possible PART
?"That you want to use that computer now is not possible".
(22). Sentences with base-generated topics and Wh-islands (SCFL/BGT-Wh)
(22a). Experimental
zhe ge dianying wo bu zhidao weishenme ta bu xihuan. "this CL movie I not know why he not like"
*"This movie I don't know why he doesn't like."
(22b ). Control
wo bu zhidao weishenme ta bu xihuan zhe ge dianying. I not know why he not like this CL movie
"I don't know why he doesn't like this movie."
(23). Sentences with non-gap topics (SCFL/NGT)
(23a). Experimental
tade pengyou, wo zhi jian- guo Xiao Wang. his friends I only meet EXP Xiao Wang
*"His friends, I have only met Xiao Wang."
(23b ). Control
zhiyu tade pengyou, wo zhi jian- guo Xiao Wang. As for his friends I only meet EXP4 Xiao Wang
"As for his friends, I have only met Xiao Wang."
7.2.3. Test sentence with null object for the CFL informants
Six sentence structures for the CFL informants were designed to find out whether
English-speaking learners of Chinese would initially reject a large proportion of correct
194
195
Chinese sentences with null objects under the influence of their Ll (i.e. English) settings
in the learners' IL grammar. Through the test sentences, we would like to find out
whether the CFL informants were comfortable with the fact that, unlike their Ll, English,
Chinese sentences with object pro (null object) are quite acceptable in the target language
(Chinese).
In the following example sentences, each experimental sentence (24a), (25a), (26a),
(27a), (28a) and (29a) has a corresponding control sentence, illustrated in (24b), (25b),
(26b), (27b), (28b) and (29b), respectively. In Chinese, the third person pronoun in object
position referring to an inanimate entity is obligatorily absent (see 4.3). We made a
distinction in the experimental design between an inanimate object pro and an animate
object pro, in order to find out whether such a distinction would be reflected in the
subjects' judgments.
(24). Sentences with null object (object pro) in sentential subject (SCFL/NO-SS)
(24a). Experimental
ta zhengzai yong nar zhi gangbi. ni xiang xianzai yong shi bu keneng de. He PRG5 use that CL pen you want now use be not possible PART
*"He is using that pen. That you want to use now is impossible."
(24b ). Control
ni xiang xianzai yong nar zhi gangbi shi bu keneng de. you want now use that CL pen be not possible PART
?"That you want to use that pen is not possible."
(25). Sentences with null object (object pro) and Wh-island (SCFL/NO-Wh)
(25a). Experimental
4 EXP refers to experiential aspect marker. 5 PRG refers to progressive aspect marker.
(26).
(27).
wo hen xihuan zhe ge dianying keshi wo bu zhidao weishenme ta bu xihuan I very like this CL movie but I not know why she not like
*"I like this movie very much, but I don't know why she doesn't like."
(25b ). Control
wo bu zhidao weishenme ta bu xihuan zhe ge dianying I not know why she not like this CL movie
"I don't know why she doesn't like this movie."
Sentences with inanimate null object (object pro) in main sentence (SCFL!INO-MS)
(26a). Experimental
tade danche hui le, wo dasuan mingtian bang ta xiuli his bicycle bad PEF I plan tomorrow help him fix
*"His bicycle has gone wrong. I am going to repair for him tomorrow."
(26b ). Control
*tade danche hui le, wo dasuan mingtian bang ta xiuli ta. his bicycle bad PEF I plan tomorrow help him fix it
"His bicycle has gone wrong. I am going to repair it for him tomorrow."
Sentences with animate null object (object pro) in main sentence (SCFL/ANO-MS)
(27a). Experimental
wo like jiu renchulai le zhexie nu yanyuan, I at once then recognise PFV female actor
houlai Xiao Wang ye renchulai le later Xiao Wang also recognise PFV
*"I immediately recognised these actresses, and later Xiao Wang also recognised."
196
(28).
(27b ). Control
wo like jiu renchulai le zhexie nu yanyuan, I at once then recognise PFV female actor
houlai Xiao Wang ye renchulai le tamen. later Xiao Wang also recognise PFV them
"I immediately recognised these actresses, and later Xiao Wang also recognised them."
Sentences with inanimate null object (object pro) in subordinate sentences (SCFL/INO-SS)
(28a). Experimental
shang zhou Xiao Wang de qiche bei dao le. last weekXiao Wang DE car be stolen PFV
danshi Lao Lee shuo jingcha yijing bang ta zhaodao le. but Lao Lee say police already help him found PFV
*"Xiao Wang's car was stolen last week, but Lao Lee says the police have found for him."
(28b ). Control
*shang zhou Xiao Wang de qiche bei dao le. last weekXiao Wang DE car be stolen PFV
danshi Lao Lee shuo jingcha yijing bang ta zhaodao le ta. but Lao Lee say police already help him found PFV it
"Xiao Wang's car was stolen last week, but Lao Lee says the police have found it for him."
(29). Sentences with animate null object in subordinate sentences (SCFL/ANO-SS)
(29a). Experimental
Lao ban shuo naxie kehu zat jiudian datang nei, Boss say those clients at hotel lobby in
keshi wo gaosu ta zai na'er mei zhaodao. but I tell him at there not find
*. "The boss said those clients were in the hotel lobby, but I told him I didn't find there."
197
(30).
(29b ). Control
Laoban shuo naxie kehu zai jiudian datang nei, Boss say those clients at hotel lobby in
keshi wo gaosu ta zai na'er mei zhaodao tamen. but I tell him at there not find them
"The boss said those clients were in the hotel lobby, but I told him I didn't find them there."
Sentences with inanimate null object coindexed with an argument in an adjunct (SCFL/INO-Adjunct)
(30a). Experimental
ruguo m xiechu yiben hao shu, wo yiding mai. If you write a good book I definitely buy
*"Ifyou write a good book, I will definitely buy."
(30b). Control
Ruguo ni xiechu yiben hao shu, wo yiding mai ta. If you write a good book I definitely buy it
"If you write a good book, I will definitely buy it."
7.3. Analysis of the EFL and CFL informants' judgments
7.3.3. The EFL Informants
All the sentences with the null and non-null subjects and objects, base-generated/non-
base-generated topics in the acceptability judgment tests for the EFL informants can be
found in Appendix 10.
198
199
7.3.3.1. Judgments on the SEFL/NS-MS and SEFL/NS-ES
In Charts 7-1 and 7-2, the lines of dashes with circles represent the mean scores of AJT
judgments on the control sentences, i.e. sentences with subject position lexically filled. In
judging the EFL/NS-MS sentences, except those in Group 5, most informants have
Chart 7-1:
Chart 7-2:
Mean Scores of AJT on the S-EFL/NS-MS sentences by the EFL informants
Chart 7-1: S-EFLINS-MS 4.0,------------,
3.0 ---------------------------------- ·;."
"' 2,{)( .·::::::_~-~.:::. 'f):. __ '"f:J-_____ ___ ,...!!! ____ --.... _e .....
1.0 -- ------------------------------------
1-~ 0.0 --------------------------------------
0 ~ -1.0 ----- ---~-------------------------
~ -2.0 -------------------------~ • ::~::~: <:: ro
;:E -3.0 0 Control 5-EFUNS-MS 1 2 3 4 5 6 7
· .EFL Groups
·Mean Scores of AJT on the S-EFL/NS-ES sentences By the EFL informants
Chart 7-2: S-EFL/NS-ES 3.0,-------------,
/
2.0 ------------------------------"if-<'---,/"
,...0.... _.0' 1.o -~-----" .... e-·: . .-.::e""- ----------------
~ 0.0 --------------------- ----------------
0 ~ ~ -2.0 -------------------------------------<::
• Expennental
S-EFUNS-ES
lll :;; -3.0 0 Control S-EFUNS-ES
2 3 4 5 6 7
· .EFL Groups
200
learned to recognize the sentence structures designed to test the null subject in finite
clauses. The solid lines with black squares represent the mean scores of judgments on the
experimental sentences, i.e. sentences which are ungrammatical because of the null
subject in the finite clauses.
Group 1 in Chart 7-1, Group 1-4 in Chart 7-2 seem not be able to reject the
ungrammatical experimental sentences, since they are elementary and intermediate EFL
learners. These informants' judgments were quite influenced by their L1, Chinese, in
which a null subject may occur in the subject position of a finite clause. From the
multiple comparisons in Tukey HSD test conducted subsequently (see Table 7-1A in
Appendix 13), we could see that except for Group 1, all the EFL groups (Groups 2-6)
show no significant difference from the native group (Group 7), in rejecting the
experimental sentences ofEFL/NS-MS. However, in terms of the judgment ofEFL/NS
ES sentence, there seems to be some restructuring going on at this stage in the
informants' IL grammars ofEnglish.
In judging the control sentence ofEFL/NS-MS, there are significant differences between
Group 4, Group 5 and Group 7, in rejecting the experimental sentences. In judging the
control sentence ofEFL/NS-ES, except for Group 5, there are no significant differences
between other groups (see Table 7-1B in Appendix 13).
7.3.3.2. Judgments on the gEFL/NS-NO sentences
Chart 7-3: Mean Scores of AJT on the S-EFL/NS-NO sentences by the EFL informants
Chart 7-3: S-EFUNS-NO 3.0..-------------,
2.5 --------------------------------------
2.0 ------- --------------------- ----------
1.5 -------------------- :.:,(2)-:.-:.-f!:.-: __ _ h--e
1.0 -:: ::e;-""'-- ------.-------------------.5 --- ------ ------------------------
~ 0,0 ---------------- ---------------------
~ -1 0 ----------------------------------- -·
g _1:5 -------------------·------------·---·· • Experimental (f) S·EFUNS·NO m -2.0 ---------------·-·-------------------· --. :E -2.5 0 Control S·EFUNS·NO
1 2 3 4 5 6 7
.EFL Groups
201
As we can see from Chart 7-3, in judging the ungrammatical experimental sentences of
EFL/NS-NO (i.e. the solid line with black squares), the informants in Groups 1-3 are
quite indeterminate. This is assumed to be due to the fact that, at early stages of
acquisition, the learners do not have much L2 knowledge and their mastery of the
sentence structures tested is relatively poor. As we can see from the control sentence (i.e.
the dashed line with circles), their AJT scores are among the lowest in different EFL
groups. However, when the informants' English proficiency level improves, they become
more determinate in rejecting the experimental sentences. Nevertheless, it is noticed that
Group 4, Group 5 and Group 6 are almost parallel in rejecting the experimental
sentences. Though they are able to reject the experimental sentences when they become
more proficient, it seems the restructuring stage is having a lasting effect on the learners.
202
Generally speaking, Chinese allows both null subject in the finite clause and null object,
while English allows neither of them. It might be suggested that the informant's rejection
ofEFL/NS-NO sentence is the result of their awareness of the ungrammaticality of the
null subject in English only, but not the null object. Notice that the pattern of the line
representing the EFL informant's judgment on the experimental sentences of EFL/NS-NO
is similar to that ofEFL/NS-MS than that ofEFL/NS-ES.
7.3.3.3. Judgments on the SEFL/Exp-WhP and SEFL/Exp-RP sentences
From Chart 7-4, we can see that informants generally agree to accept the control
sentence, i.e. sentence with the nominal expletive "it" lexically filled. Informants' rejected
the experimental sentences, which do not use the expletive "it". As can be seen from the
Chart 7-4, with the improvement of proficiency levels, EFL learners reflect gradual
overall rejection for the missing expletive sentences.
In Chart 7-5, again, we noticed an overall preference for the control sentences (see Table
7-3 in Appendix 13). EFL learners appear to experience a rather dramatic jump in
improvement on the rejection of experimental sentences. However, the rejection process
stops at Group 5, because when proficiency level further improves, Gioup 6 as well as
the native speaker's judgments (Group 7) of EFL!Exp-WhP are more accepting than those
of the previous levels, or non-native groups. This is believed to be related to the fact that
some of the experimental sentences, such as Seems he likes this movie are quite
acceptable in colloquial English even though they are unacceptable from a grammatical
203
point of view. Some advanced learners may have accepted this kind of structure because
of their intensive exposure to colloquial English usage.
Chart 7-4:
Chart 7-5:
Mean Scores of AJT on the S-EFL/Exp-WhP sentences by the EFL informants
Chart 7-4: 8-EFUExp-WhP 3.5,...----'------------,
2.5 -------------------- --------------rr _...:_ /
/ 1.5 ---------- -· ••••••• ·-. --- --· __ , __ ·---·.- •• /
<1~·---e--- --..o---0 .5 ••••••••••••• '9.. ........... ----------------
~ -.5 ~------------------------------ -
0 ~ -1.5 --------------------·--·--·······-~ _•_;::~::::hP (/) -2 5 --------------------------·-··------------@ · 0 Control
:* -3.5 S-EFLJExp-WhP 12 3 4 56 7
.EFL Groups
Mean Scores of AJT on the S-EFL/Exp-RP sentences by the EFL informants
~ 0
Chart 7-5: 8-EFUExp-RP 3.0..-------------.,
2.5 ---··-··-···-·--······---------------
2.0 ---------------------------._,.fr::::-. /
1.5 ·-:_-:_;;:; :..:B<-·:...·::e--·,.-0.- ·--·--- ---
1.0 -------------------- ·- --------------
.5 • ··-···-- --------- --·-···-- ---- ·-·- --
~ -1.0 ·---------------------------------·-·
~ -1.5 ··--------------···------------------ • :~:::::~~p @ -2.0 ····------------··-··---------------- ---:* -2.5 0 Control 5-EFL/Exp-RP
2 3 4 5 6 7
.EFL Groups
7.3.3.4. Judgments on the SEFL/BGT-SS and SEFL/BGT-Wh sentences
Chart 7-6:
Chart 7-7:
Mean Scores of AJT on the S-EFL/BGT-SS sentences by the EFL informants
Chart 7-6: S-EFUBGT -SS 3.5~---------------,
2.5 ----------------------------------------_, -/
/,(:[ 1.5 ------------;B-.-::..-------------;,.. ___ --------
- ,...- --o----e -"'0'
.5 -------------------------------------------
• Experimental ~ -1.5 ------------------------------- ----------8 S·EFLIBGT -ss ~ -2 5 -------------------------------------------·
0 Control
m :iE -3.5 S-EFLIBGT-SS
2 3 4 5 6 7
.. EFL Groups
Mean Scores of AJT on the S-EFL/BGT-Wh sentences by the EFL informants
Chart 7-7: S-EFUBGT-Wh 4.0~------------,
3.0 ----------- .. -------- .. - ........................... -------e---
---0---e--4--:::.::~-~-~---------2.
1-~ 0.0
~ 8 ~ -2.0 ------------------------------------------ro Q)
:iE -3.0J...-------------' 1 2 3 4 5 6 7
.EFL Groups
• Experimental
S-EFUBGT-Wh
0 Control
S-EFUBGT-Wh
204
205
From the Charts 7-6 and 7-7, I noticed that the judgments by Groups 1-5 on the
EFLIBGT -SS control sentences and judgments by Groups 1-4 on the EFL/BGT-Wh
control sentences are both quite indeterminate. This suggests that informants from
beginner's to intermediate proficiency level may not have mastered sentences with wh
island and extraposed clause. From the multiple comparisons in Tukey HSD test (see
Table 7-4 in Appendix 13), we find that significant differences are found between Groups
1-5 on the one hand, and Group 7 on the other, in judging the control sentences of
EFL/BGT -SS. However, there is no significant difference found in Tukey HSD test
between Groups in judging the control sentences ofEFLIBGT-Wh. Even though the
judgment of control sentences in both cases improves gradually while proficiency level
improves, as we can see from Chart 7-7, the judgments ofEFL/BGT-Wh control
sentences are quite indeterminate, and this suggests that informants might have more
difficulties with sentences with wh-island structure.
In judging the EFLIBGT -SS experimental sentences, there appears to be a gradual
improvement in accuracy in judging the ungrammatical sentences, except Group 2.
Generally speaking, when the informants become more proficient, they are more
determinate in rejecting the incorrect experimental sentences.
In judging the EFL/BGT-Wh experimental sentences, there is also a general tendency of
improvement in judging the incorrect sentences, with the only exception of Group 6. This
may indicate the indeterminate restructuring which might even happen to advanced
learners. It should be noticed that, the result of multiple comparisons in Tukey HSD (see
206
Table 7-7 in Appendix 13) show that, except for Group 1, all the learner groups (Groups
2-7) show no significant difference from the native group (Group 7) in rejecting the
experimental sentences.
On the whole, from the judgments on the EFL/BGT -SS and EFL/BGT-Wh sentences, we
can conclude that EFL learners do not have much difficulty in rejecting the
ungrammatical experimental sentences of the base-generated topic in English.
7.3.3.5. Judgments on the SEFL/NGT sentences
In judging the EFL/NGT control sentences (see Chart 7-8), no significant difference is
found between any of the learner groups (Groups 1-6) and the native speaker group
(Group 7) (see Table 7-8 in Appendix 13).
Chart 7-8: Mean Scores of AJT on the S-EFL/NGT sentences by the EFL informants
~ 0
Chart 7-8: S-EFL/NGT 3,0.,-----------,
2.5 ----------------------------------------
2.0 --------------- -------------------------
1.5 ------------------- --------------------~e--
1.o -::;;..-.e--.:.-~-.:.-.:.-0=--=--.:o-<------ ----.5 - --------------------------------------
~ 8 • Experimenlal
(J) S-EFUNGT ~ -2.0 ---------------------------------------- ---:2 -2.5 0 Control E-EFUNGT
2 3 4 5 6 7
.EFL Groups
207
In judging the EFL/NGT experimental sentences, significant differences are found
between Groups 1-3 on the one hand and Group 7 on the other (see Table 7-8 in
Appendix 13). As we can see from the solid lines with black squares in Chart 7-8, there is
a steady improvement of accuracy in rejecting the experimental sentences by the EFL
learner groups. That is to say, the informants become more accurate in rejecting the
experimental sentences when their proficiency in English improves gradually.
This fmding shows that, EFL learners would accept English sentences with topics
preceded by prepositions, such as "as regards", "out of', etc., as native speakers of
English would do in many circumstances. Meanwhile, when their proficiency in English
improves, EFL learners will be aware of the fact that base-generated topics are not
allowed in English even though they are perfectly okay in their L1, Chinese. In this way,
their improved proficiency will help them to eliminate the incorrect English sentence
structures with non-gap topics.
7.3.3.6 Judgments on the SEFL/NO-SS and SEFL/NO-Wh sentences
The line of dashes represents the EFL informants' judgments on the control sentence of
EFL/NO-SS with circles in Chart 7-9. As we can see in Chart 7-9, the EFL learners'
judgments on the control sentences ofEFL/NO-SS by Groups 1-5 are relatively
indeterminate in that the structure of sentences with an extraposed clause by Groups 1-5
are at a marginal level. The result of multiple comparisons in Tukey HSD tests indicates
that there is significant difference between Groups 1-5 on the one hand, and Group 7 on
Chart 7-9:
Chart 7-10:
Mean Scores of AJT on the S-EFL/NO-SS sentences by the EFL informants
Chart 7-9: S-EFUNO-SS 4.0,....--------------,
3.0 ------------------------------ ;;:;:.::
/ 2.0 --------------------------- / __ ----- ---
~-0---G---e/ --e-
0 0.0
"' !!! 8 rJ) -1.0 c ro Q)
:z -2.0.1-----------' 1 2 3 4 5 6 7
· .EFL Groups
• Experimental
S-EFUNO-SS
0 Control S-EFUNO-SS
Mean Scores of AJT on the S-EFL/NO-Wh sentences by the EFL informants
Chart 9-10: S-EFUNO-Wh 4.0,..------------,
3.0 .......... -------------------------------xr-
1.0
I-< 0.0
~ -1.0
8 ~ -2.0 ro
.-" -~-e. ... =--e~=-e---e---- --------
• Experimental
S-EFL/NO-Wh
~ -3.0.1------------ 0 Control S-EFUNO-Wh 1 2 3 4 5 6 7
EFL Groups
208
209
the other (see Table 7-9 in Appendix). No significant difference is found between Group
6 and Group 7. This suggests that, like the native speakers of Group 7, informants in this
learner group had mastered the sentence structures with an extraposed clause.
The solid lines with black squares represent the EFL learners' judgments on the
experimental sentences ofEFL/NO-Wh. As we can see from Chart 7-9, significant
differences are found between the judgments by Groups 1-4 and by the native speakers in
EFL learner Group 7 (see Table 7-10 in Appendix 13). This suggests that informants in
Groups 1-4 are reluctant to or having difficulty in rejecting sentences with null object in
an extraposed clause.
It is interesting to note that, while judging the experimental and control sentences of
EFL/NO-SS, informants in Groups 1-4 experienced a stage of indeterminate status. After
that period, learners began to master the sentence structures with an extraposed clause.
When looking at the judgments on EFL/NO-Wh sentences, similar results emerged, as is
shown in Chart 7-10. In judging the control sentences, significant differences are found
between Groups 1-2 and Groups 4-5 on the one hand, and Group 7 on the other (see
Table 7-10 in Appendix 13). No significant difference is found between Group 3 and
Group 7 in the judgment ofEFL/NO-Wh control sentences. However, since Group 3's
mean score in judging the control sentence of EFL/NO-Wh (mean difference = .8696) is
quite close to that of Group 2's (mean difference = .9800), we assume that the sentence
210
structure with a wh-island also had an indeterminate status in the learners' IL grammar of
English, like that of Groups 1-2 and Groups 3-4,
No significant difference is found between Group 6 and the native speaker Group 7 in
judging the control sentences ofEFL/NO-Wh. This suggests that the learners in Group 6
had already mastered the sentence structure ofWh-island in English. In judging the
ungrammatical experimental sentences ofEFL/NO-Wh, the informants in Groups 1-5, as
is shown in Chart 7-10, are quite indeterminate in rejecting the incorrect experimental
sentences of EFL/NO-Wh. Significant differences are found between Groups 1-6 on the
one hand, and Group 7 on the other, injudging the experimental EFL/NO-Wh sentences
(see Table 7-10 in Appendix 13). Like the case of judging the EFL/NO-SS sentences,
EFL learners are having difficulty in rejecting the ungrammatical English sentence with
null object in a wh-island.
7.3.3.7. Judgments on the SEFL/INO-MS and SEFL/INO-SS sentences
In Charts 7-11 and 7-12, the solid lines with black squares represent the EFL informants'
judgments on the experimental sentences ofEFLIINO-MS and EFLIINO-SS. The lines of
dashes with circles represent their judgments on the control sentences.
As Chart 7-11 shows, the informants in Groups 1-6, together with the native speakers in
Group 7, generally accepted the control sentences, which are correct in English. There is
no significant difference found between Groups 1-6 and Group 7 injudging the control
Chart 7-11: Mean scores of AJT on the S-EFL/INO-MS sentences by the EFL informants
Chart?-11: S-EFUINO-MS 3.0,------------,
,... .0--.,.,..xr---G----- .,.. .....
2. .:-. =@--- --- ___________ (;2: -----------
~ '0 -1.0 ---------------------------------- --
~ 8 • Experimental en -2.0 - • •• • ---··· • •• -------------------- •• - S-EFUINO-MS c Cll
~ -3.0.1-----------1 0 ControiS-EFUINO-MS 1 2 3 4 5 6 7
.EFL Groups
Chart 7-12: Mean scores of AJT on the S-EFLIINO-SS Sentences by the EFL informants
Chart 7-12: S-EFUINO-SS 3.0r-----------,
2.0 ---------------------------- --ft--:.-:. I
....G---8-- / 1.o -:- -:--e"":.:-------------::-.0.:-----------
f-
< '0 -1.0 rn
~ -2.0 -------------------------------------
• Experimental
S-EFUINO-SS c Cll Q) ::2: -3.0J-_________ _J 0 Control S-EFLIINO-SS
7 2 3 4 5 6
. EFL Groups
211
212
sentences ofEFL/INO-MS (see Table 7-11 in Appendix 13). Again, no significant
difference is found between Groups 1-6 and Group 7 in judging the control sentences of
EFL/INO-SS (see Table 7-12 in Appendix 13).
However, when we look at the judgments of the experimental sentences ofEFL/INO-MS
and EFL/NO-SS, we can see that informants in all learner groups were unable to reject
the ungrammatical experimental sentences with inanimate null object either in the main
sentences or in the subordinate sentences. They all failed in this respect without any
exception. As we can see from the Tukey HSD multiple comparisons tests, in judging the
experimental sentences ofboth EFL/INO-MS and EFL/INO-SS (see Table 7-11 and 7-12
in Appendix 13), there are significant differences between the native speaker Group 7 and
all the other EFL learner groups. From this result, we can conclude that the EFL
informants have failed to reject the ungrammatical English sentences with null object.
7.3.3.8. Judgments on the SEFL/ANO-MS and SEFL/ANO-SS sentences
In Charts 7-13 and 7-14, the solid lines with black squares represent the EFL informants'
judgments on the experimental sentences ofEFL/ANO-MS and EFL/ANO-SS, i.e.,
sentences with animate null object in object positions. The lines of dashes with circles
represent their judgments on the control sentences, i.e., sentences with the animate object
pronouns.
Chart 7-13: Mean scores of AJT on the S-EFL/ANO-MS Sentences by the EFL informants
Chart 7-13: S-EFUANO-MS 4.0..------------,
3.0 ------------------- ----------;B"= /
/ 2.0 ------------ -;.-.-.-0--"'f:f ----------
e --e --1.0 ..................................................................... ..
~ 0.0
'0 "' ~ 8 • Experimental
(f) -2.0 ------------------------------------ S-EFUANO-MS
m ::!: -3.0 0 Control S-EFUANO·MS
1 2 3 4 5 6 7
EFL Groups
Chart 7-14: Mean scores of AJT on the S-EFLIANO-SS Sentences by the EFL informants
Chart 7-14: S-EFUANO-SS 3,0,------------,
2.o -.-------------------.--- -----6---·-.0---E)---0"-- --
1. - -e--::: .. "': .. --------- ................... -------
'0 ·1.0 --------------------------- --------
"' ~ • Experimental ~ -2.0 --------------------·------------- - S-EFUANO-SS c <U
~ -3.0.1----------__J 0 Control S-EFUAN0-88 1 2 3 4 5 6 7
EFL Groups
213
214
According to the multiple comparisons Tukey HSD tests results, the EFL learners'
judgments on the experimental and control sentences ofEFL/ANO-MS and EFL/ANO
SS have not been affected by the distinction between the animate object pronoun and the
inanimate object pronoun in Chinese. There is no difference between the judgments on
the control sentences with the animate object pronoun and those with third person
singular object pronoun referring to inanimate entities, which were reported in the
. previous section.
The EFL informants' judgments are quite similar in that there is no significant difference
found between Groups 3-6 and Group 7 in judging the control sentences ofEFL/ANO
MS (see Table 7-13 in Appendix 13). Also, there is no significant difference found
between Groups 1-6 and Group 7 in judging the control sentences ofEFL/ANO-SS (see
Table 7-14 in Appendix 13). The EFL learners normally accepted the control sentences
that are grammatical in English.
On the other hand, as is indicated by the solid lines with black squares in Charts 7-13 and
7-14, learner Groups 1-3 failed to reject the ungrammatical experimental sentences of
both EFL/ANO-MS and EFL/ANO-SS. There are significant differences between EFL
learner Groups 1-6 and Group 7 in rejecting the ungrammatical experimental sentences of
EFL/ANO-MS (see Table 7-13 in Appendix 13). There are also significant differences
between Groups 1-5 (except Group 4) and Group 7 in rejecting the ungrammatical
experimental sentences ofEFL/ANO-SS (see Table 7-14 in Appendix 13). From these
215
results, it is suggested that most of the early and intermediate EFL learners are unable to
reject the ungrammatical English sentences with null object.
7.3.3.9. Judgments on the SEFL/INO-Adjunct sentence
Chart 7-15: Mean scores of AJT on the S-EFLIINO-Adjunct Sentences by the EFL informants
Chart 7-15: S-EFUINO-Adjunct 3.0..--------------,
2.0 •••••••••• •••••··•·•••••• •••••• ·::o-·,;.·-" ft--B---o---e--,.,
1.0 • .: ••••••••••••••••••••••••••••••••••••••
~ 0 "' ·1.0 • ••••••• •• ••••• •• •••• •• •••• ••••• •• • •• •• • Experimental ~ 8 S-EFUINO-Adjuncl
(/) ·2.0 •••••••••••••••.•••.•••••••••••••••••••• ---
0 Control c: m :2 -3.0.1------------' S-EFUINO-Adjuncl
1 2 3 4 5 6 7
EFL Groups
In Chart 7-15, the line of dashes with circles represents the EFL groups' judgments on the
control sentences ofEFLIINO-Adjunct, i.e., sentences with an inanimate pronoun it
coindexed with an argument in an adjunct (see sentence (15b)). The solid lines with black
squares represents the EFL groups' judgment on the experimental sentences, i.e.,
sentences with an inanimate null object coindexed with an argument in an adjunct (see
(15a)).
As we can see from the Chart, the EFL informants' judgments on the control sentences
are quite indeterminate. There is no significant difference between Groups 2-6 and Group
216
7 (except Group 1 and Group 7) (see Table 7-15 in Appendix 13). When we look at the
learner groups' judgments on the experimental sentences ofEFLIINO-Adjunct, we can
also see the indeterminate development of informants' IL grammars. As is confirmed by
the Tukey HSD tests, only Groups 1-3 are significantly different from the native speaker
Group 7 in judging the experimental sentences, with the rest ofthe Groups 4-6 having
marginal differences from one another. This has provided us evidence that EFL
informants are experiencing difficulty in rejecting the ungrammatical English sentence
with null object.
7.3.3.10. Summary ofEFL learners' judgments
As we can see from the above analysis, for most of the EFL informants, especially
intermediate and high intermediate learners, there is not much difficulty in rejecting the
ungrammatical experimental English sentences with null subject in the finite clause.
When encountering with both null subject in finite clause and null object, most EFL
informants are able to reject the incorrect EFL/NO-NO sentences, though it is suggested
that is a result of their awareness of the ungrammaticality of the null subject in English,
not the null object.
In judging the use of the nominal expletive it, the EFL informants expressed an overall
preference for the control sentences, i.e., sentences with the nominal expletive it lexically
filled, rejecting the sentences without the expletive it either as "less acceptable" or
"completely unacceptable".
217
From the judgments on the sentences of base-generated topics, EFL informants from
beginners' to intermediate proficiency levels may not have mastered sentences with wh
island or extraposition. To some extent, the EFL learners' judgments are quite
indeterminate. Nevertheless, when the EFL informants become more proficient, they are
more aware of the ungrammaticality of the base-generated topic in English and are more
determinate in rejecting those incorrect experimental sentences.
As the judgments on the non-gap topics has shown, when their proficiency in English
improves, EFL informants will be more aware of the fact that base-generated topics are
not allowed in English even though they are perfectly acceptable in their Ll, Chinese. In
this way, their improved proficiency will help them to eliminate the incorrect English
sentence structures with non-gap topics.
However, from the judgments on the experimental sentences with null object, most of the
EFL informants have failed to reject the ungrammatical English sentences with null
object. All learner groups are unable to reject the ungrammatical experimental sentences
with inanimate null object either in the main sentences or in the subordinate sentences. It
is an indication that EFL informants are experiencing difficulty in rejecting the
ungrammatical English sentences with null object, their improvement in proficiency has
little effect on their indeterminacy in their IL grammars of English.
218
7.3.4. The CFL Informants
7.3.4.1. Judgments on the SCFL/NS-MS and SCFL/NS-ES sentences
All the sentences with null and non-null subjects and objects, base-generated/non-base
generated topics in the acceptability judgment tests for the CFL informants can be found
in Appendix 10.
In Charts 7-16 and 7-17, the lines of dashes with circles represent the mean scores of AJT
judgments on the control sentences, i.e. sentences with subject position lexically filled in.
As the charts show, though the line of dashes with circles in Chart 7-16 move in a zigzag
way, it is within a limited range and quite flat in some way, like the other line of dashes
with circles in Chart 7-17. This indicates that the CFL informants have more or less learnt
the sentence structures provided in the test like the native speakers do.
The solid lines with black squares in Charts 7-16 and 7-17 represent the CFL informants'
judgments on the experimental sentences. From the two charts, we found that the
learners' judgments on the experimental sentences were quite similar to one another. The
solid lines of black squares present themselves in a rather flat way. From the multiple
comparisons Tukey HSD tests conducted subsequently, we could see that there is no
significant difference found between any of the learner groups and the native speaker
group, Group 5 (see Tables 7-16 and 7-17 in Appendix 13). This is an indication that
Chart 7-16: Mean scores ofAJT on the S-CFL/NS-MS Sentences by the CFL informants
Chart 7-16: S-CFUNS-MS 3.0..------------,
2.0 --------------------------------------
...... -=V -¥ -:B---1.0 - -- ----------------------- ------------
0.0 --------------------------------------
-2.0 --------------------------------------
2 3 4
CFL Groups
• Experimental
5-CFUNS-MS
Chart 7-17: Mean scores of AJT on the S-CFL/NS-ES Sentences by the CFL informants
Chart 7-17: S-CFUNS-ES 3.0.....-----------,
2.0 - ----------------------- --------------
1.0 -:.:----..A __ ,.,...__
1-
~ 0.0 --------------------------------------0 "' e! 8 .1•0 ---------------·---------------------- • Experimental rn S-CFUNS-ES
m :E -2.0J------------' 0 Control S-CFUNS-ES
2 3 4 5
CFL Groups
219
220
CFL informants, when they come across null subjects in the subject position of the finite
clause, do not experience much difficulty in learning the Chinese sentence structures.
7.3.4.2. Judgments on the SCFL/NS-NO sentences
Chart 7-18: Mean scores of AJT on the S-CFL/NS-NO Sentences by the CFL informants
Chart 7-18: S-CFUNS-NO 3.0..-----------,
2.5 ----------------- --------------- -----
2.0 -------- ------- ----------------------
1.5 -------- ------------------------------ 'Q' --
1.0 ~-------~~------------------~ '0 .5 -------------------------------------(/)
~ 0.0 ------------------------------------- • Experimental
~ -.5 ------------------------------------- S-CFUNS-NO m :2 -1.0 0 Conlrol S-CFUN8-NO
1 2 3 4 5
CFL Groups
As Chart 7-18 shows, the solid line with black squares represents the informants'
judgments on the control sentences where both the subject and the objects are lexically
filled. The line of dashes with circles represents the judgments on the experimental
sentences, i.e. sentences with both a null subject and a null object. From the previous
chapters (see Chapter 4), we know that Chinese allows both null object and null subject
in finite clauses. From Chart 7-18, it is noticed that there seems a general preference of
the experimental sentences over the control ones, except Group 1. Notice that the
judgments of the experimental and control sentences are very close to one another.
221
Nevertheless, it is difficult for us to find out whether the preference is genuine or not,
because there is no significant difference between the judgments on the experimental and
control sentences by the CFL learner groups and the native speaker group, Group 5 (see
Table 7-18 in Appendix 13). This has served as an indication that the informants at the
intermediate learner groups may have accepted the null subject and null object as
grammatical in their acquisition of Chinese.
7.3.4.3. Judgments on the SCFL/Exp-WhP and SCFL/Exp-RP sentences
In Chinese, the use of a nominal expletive in subject position leads to ungrammatical
sentences. Native speakers normally would not use this kind of expletive in subject
position unless it is done so on purpose to create a special effect in discourse. Even so, it
would be hard to find such examples. The line of dashes with circles represents the
control sentences, i.e. sentences that are acceptable, without the use of a nominal
expletive in subject position. The solid line with black squares represents the
ungrammatical and unacceptable sentences, i.e. sentences using the nominal expletive in
subject position. From the multiple comparisons HSD Tukey tests conducted, we can see
that there is no significant difference found between Groups 1-4 on the one hand and
Group 5 on the other, in rejecting the control sentences ofExp-WhP (see Table 7-19 in
Appendix 13). There is only significant difference found between Group 1 on the one
hand and the native speaker group, Group 5, on the other in rejecting the control
sentences ofExp-RP (see Table 7-20 in Appendix 13). In this case, the informants of
Group 1 judge the control sentences indeterminately. As we can see, informants in Group
Chart 7-19: Mean scores of AJT on the S-CFL/Exp-WhP Sentences by the CFL informants
Chart 7-19: S-CFUExp-WhP 3.5,--------------,
.5 ------------------------------------------
1-~ -.5 ------------------------------------------
0 <~---1.5 _____ :-_:::~::---..,-------------------------- • Experimental
~ ""0-- --e---- ~ S-CFUExpoWhP ~ 8
(J) -2 5 ------------------------------------------<: •
0 Control
m :2 -3.5 S-CFUExpoWhP
2 3 4 5
CFL Groups
Chart 7-20: Mean scores ofAJT on the S-CFL/Exp-RP Sentences by the CFL informants
Chart 7-20: S-CFUExp-RP 3.5,-----------,
2.5 -------------------------------------
1.5 ~----------=----~ .5 -------------------------------------
1-~ -.5 -------------------------------------0 ......... ~ -1.5 _____ -::,.._& ___________________________ -8 ... ... ... • Experimental
rn ·2.5 ---------------~:0:-.-:-.-::-.-=-.e:-.:-.:-.:-. S-CFUExpoRP <: ro ~ -3.5 0 Control S-CFL/ExpoRP
1 2 3 4 5
CFL Groups
222
1 may also have difficulty accepting the experimental sentences of Exp-WhP since there
is significant difference between Group 1 and the rest of the groups, Groups 2-5 (see
Table 7-19 in Appendix 13). However, the same thing is not found in the judgment of
223
experimental sentences ofExp-RP (see Table 7-20 in Appendix 13). Generally speaking,
this may indicate the fact that beginning learners are experiencing some trouble accepting
sentences without the use of a nominal expletive in subject position in their IL grammar
of Chinese.
7.3.4.4. Judgments on the SCFL/BGT-SS and SCFLIBGT-Wh sentences
From Charts 7-21 and 7-22, we can see that the lines of dashes with circles represent the
judgments on the control sentences of BGT -SS and BGT-Wh. The solid lines with black
squares represent the judgments on the experimental sentences. In the control sentences,
there is no base-generated topic and every argument position is lexically filled (see (21 b)
and (22b) in 7.2.2.). In the experimental sentences, however, there is a base-generated
topic coindexed with a variable in a sentential subject or in a wh-island (see (21a) and
(22a) in 7.2.2.).
From the two charts, we can find some similar graph patterns. It seems that, as the
informants' proficiency in Mandarin Chinese increases (see Charts 7-21 and 7-22), no big
improvement in accuracy is found in the judgments of the experimental sentences by the
informants in CFL learner Groups 1-4. It seems that there is no sharp increase in the
acceptability of the sentences with the base-generated topic in the informants' IL
Chart 7-21: Mean scores of AJT on the S-CFL/BGT-SS Sentences by the CFL informants
Chart 7-21: S-EFUBGT-SS 3.0,--------------,
2.0 ----------------- -------------------
-1.0 -- ----------------------------------(/)
~
~ -2.o ------------------------------------ • :.::::;.~s c: m :2 -3.0 .1------~-------' 0 Control S-CFL/BGT-SS
2 3 4 5
CFL Groups
Chart 7-22: Mean scores of AJT on the S-CFL/BGT-Wh Sentences by the CFL informants
Chart 7-22: S-CFUBGT-Wh 3.0,...-----------,
2.0 ------------------------------------
e----1.0 ·:.::::.=;(U--~-=--e..:-.:-.:-.:-___ -----
1-
< 0 -1.0 ------------------------------------
~ 0 • Experimental
~ -2.0 ------------------------------------ S-CFUBGT-Wh c: <11
~ -3.0.__ ________ __, 0 Control S-CFUBGT-Wh
2 3 4 5
CFL Groups
224
grammar of Chinese until they have reached the stage beyond Group 4, i.e., Group 5 (the
native speaker's group). In both cases, the mean scores of AJT of informants' in CFL
learner Group 4 were around 0.5, just above the zero level, meaning they were just
beginning to adopting the acceptability of the sentences with the base-generated topic.
225
From the multiple comparisons in Tukey HSD test conducted subsequently (see Table 7-
21 and 7-22 in Appendix 13), we could see that there is significant difference found
between Group 5 (the native speaker's group) and the rest of the groups (Groups 1-4) in
judging the experimental sentences. The results indicate that the informants in CFL
learner Groups 1-4 judged the experimental sentences ofBGT-SS and BGT-Wh quite
differently from the native speakers in CFL learner Group 5. This may indicate the fact
that there is difficulty for the beginners, intermediate as well as higher intermediate
informants to accept the sentences with a base-generated topic. One has to remember that
both the control sentences and experimental sentences represented in Charts 7-21 and 7-
22 were grammatical in Chinese and they are, to some extent, acceptable to the native
speakers of Chinese. As we can see from the charts, almost all informants judged the
control sentences as acceptable, while their support for the experimental sentences were
increasing very slowly in accordance with their proficiency levels in Chinese.
7.3.4.5. Judgments on the sCFL/NGT sentences
As Chart 7-23 shows, the solid lines with black squares represent the experimental
sentences, with "non-gap topics" (see 4.2.7 in Chapter 4) in the initial position of the
sentence which were not preceded by anything, as in sentence (23a) in 7.2.2. The control
sentences, represented by the line of dashes with circles, all have non-gap topics preceded
by a preposition zhiyu (=as for) in the initial position of the sentence, as in sentence
(23b) in 7 .2.2.
Chart 7-23: Mean scores of AJT on the S-CFL/NGT Sentences by the CFL informants
1--~ 0
Chart 7-23: S-CFL/NGT 3,0,-----------,
2.5 ----------------------- -----------------
::: ~~~~~~~i~~~=~=~ 1.0 i'::::' .. -- --------------------------------
.5 ----------------------------------------
~ 0.0 ---------------------------------------- • Experimental
~ -.5 ---------------------------------------- S-CFUNGT ro ~ -1.0 0 Control S-CFUNGT
2 3 4 5
CFL Groups
In Chinese, although the use of zhiyu (= as for) may sound a little bit redundant to the
native speakers of Mandarin, both the control sentences and experimental sentences of
226
NGT (non-gap topics) are quite grammatical and are acceptable to native speakers. From
the multiple comparisons Tukey HSD tests, we could see that there is no significant
differences found between any of the learner groups and the native speaker group, Group
5 (see Table 7-23 in Appendix 13). The only difference is that it might have different
preference levels to suit a spoken context. As we can see from Chart 7-23, all the
informants give the control sentences a "positive 1" level judgment, i.e., "slightly
acceptable". We can see from the chart that all the CFL learner groups judge control
sentences as acceptable, like the native speaker group (Group 5) does. It should be noted,
227
however, that there is less support for the acceptability of the control sentences in the
judgments by Group 5 (the native speaker group). It might be a reflection of the fact that
informants in CFL learner Group 5, the native speaker group, are aware of the clumsiness
and redundant use of zhiyu (= as for). This was later confirmed by de-briefing
conversations carried out with some of the informants after the questionnaire session.
Some of the informants did mention the stylistic heaviness of the control sentences.
7.3.4.6. Judgments on the SCFL/NO-SS and SCFL/NO-Wh sentences
As Charts 7-24 and 7-25 show, the solid lines with black squares represent the judgments
on the experimental sentences of CFL/NO-SS and CFL/NO-Wh respectively. The lines of
dashes with circles represent the judgments on the control sentences of CFL/NO-SS and
CFL/NO-Wh. The CFL/NO-SS sentences are those with null object in a sentential subject
(see sentences in (24) in 7.2.3.), and the CFL/NO-Wh sentences are those with null object
in a wh-island (see sentences in (25) in 7.2.3.).
From the multiple comparisons Tukey HSD tests, we find that there are significant
differences found between the CFL learner Group 1, and the native speaker group (Group
5) in their judgments of both the control sentences and experimental sentences of
CFL/NO-Wh (see Table 7-25 in Appendix 13). There is a significant difference found
Chart 7-24:
Chart 7-25:
Mean scores of AJT on the S-CFL/NO-SS Sentences by the CFL informants
Chart 7-24: S-CFUNO-SS 3.0.,--------------,
2.5 ...................................................................... ..
. 5 -------------------------------------
~ 0.0 -------·----------------------------- • Experimental
~ -.5 ------------------------------------- S-CFUN0-5S co ~ -1.0 0 Control S-CFUNO-SS
2 3 4 5
CFL Groups
Mean scores of AJT on the S-CFL/NO-Wh Sentences by the CFL informants
1-:t 0 U)
Chart 7-25: S-CFUNO-Wh 3.0r------------.
2.5 ----------- ... -------------------------
2.0 -------------------- ---,::_-_:o---~::.
1.5
" ,. 1.0 - Y----------------------------------
.5 -------------------------------------
~ 0.0 ------------------------------------- • Experimental
~ -.5 -------------------------------······ S-CFUNO-Wh co Q)
::2: -1.0J....------------' 2 3 4 5
CFL Groups
0 Control S-CFLINO-Wh
228
229
between the CFL learner Group 1 and the native speaker group (Group 5) in their
judgments of the experimental sentences ofCFL/NO-SS (see Table 7-24 in Appendix
13). As we can see from the charts, the lines of dashes and solid lines are quite close to
each other, sometimes crossing one another. Generally speaking, it is an indication that
CFL learners do not have much difficulty in accepting the grammatical Chinese sentences
with null object in a sentential subject or in a wh-island. It seems that their mastery of the
sentence structures tested is directly related to the improvement of their Chinese language
proficiency. When the CFL learners become more comfortable with the sentence
structures provided in the tests, i.e., when their proficiency in Chinese improves, the
acceptance of the Chinese sentences with null object in a sentential subject or in a wh
island is more accurate. It increases with the change of proficiency levels.
7.3.4.7.Judgmentson the SCFL/INO-MS and SCFL/INO-SS sentences
In Charts 7-26 and 7-27, the solid lines with black squares represent the CFL informants'
judgments on the experimental sentences ofCFL/INO-MS and CFL/INO-SS, i.e.,
sentences with null object or object pronoun in object position. The line of dashes with
circles represent their judgments on the control sentences of CFL/INO-MS and
CFL/INO-SS, i.e., sentences with an overt third person singular pronoun ta (=it) in
object position referring to an inanimate entity.
As we lmow, Chinese does not allow an overt third person singular pronoun ta (=it) to
occur in object position if it refers to an inanimate entity. Normally, it is obligatory to
Chart 7-26: Mean scores of AJT on the S-CFLIINO-MS Sentences by the CFL informants
Chart 7-26: S-CFUINO-MS 3.5,------------,
3.0 -------------------------------------
2.5 --------------------------- -
2.0 r-:::::-___ ----------------------------1.5 ------------------- ------------------
1.o -=~--0-"":.::.::~·:.:::.-·:::e---::::::_ ~ .5 -------------------------------------
0 0.0 -------------------------------------
"' e! •·5 ------------------------------------- • Experimental
~ ·1.0 ------------------------------------- S.CFUINO-MS
ffi -1.5 ------------------------------------- ---::E -2.0 0 Control 5-CFL/INO-MS
1 2 3 4 5
CFL Groups
Chart 7-27: Mean scores of AJT on the S-CFLIINO-SS Sentences by the CFL informants
Chart 7-27: S-CFUINO-SS 3.0..----------,
2.5 -------------------------------------
2.0 ------ --------------------- ----- -----
1.5~---
~ 1.0< ·:..::..::.:=o::..::.::.::e:::.::.::.:e:.::.::.:::< 'to- .5 ----------·--------------------------0
"' § 0.0 ------------------------------------- • Experimental
rn •. 5 ----------------------··············· S-CFL/INO-SS ffi ~ -1.0 0 Control S·CFUINO-SS
1 2 3 4 5
CFL Groups
230
231
have null object (pro) in object position. Because of this characteristic in Chinese, unlike
the control sentences of other sentence structures, the control sentences of CFL/INO-MS
and CFL/INO-SS are ungrammatical in Chinese. Only the experimental sentences of
CFLIINO-MS and CFLIINO-SS would be acceptable, i.e., sentences with null object
referring to an inanimate entity would be acceptable to the native speaker.
As Charts 7-26 and 7-27 indicate, we can see that the CFL informants in Groups 1-5 do
not have much difficulty in accepting the grammatical experimental sentences with
inanimate null object in either the main sentence (CFL/INO-MS) or the subordinate
sentence (CFLIINO-SS). According to the multiple comparisons Tukey HSD tests, in
judging the experimental sentences with inanimate null object, there is no significant
difference found between any of the CFL learner groups. That is to say, in judging the
grammatical experimental sentences of CFL/INO-MS and CFL/INO-SS, no significant
difference is found between Group 5 (the native speaker group) and the rest of the learner
groups (Group 1-4) (see Tables 7-26 and 7-27 in Appendix 13).
In judging the control sentences ofCFL/INO-MS and CFLIINO-SS, i.e., the
ungrammatical sentences with the overt inanimate object pronoun, again, no significant
difference is found between the native speaker Group 5 and any of the CFL learner
groups (see Tables 7-26 and 7-27 in Appendix 13). It seems that the native speakers in
the Group 5 as well as learners in CFL Groups 1-4 were aware of the fact that there is a
distinction between the more acceptable sentence with inanimate null object (pro) and the
less acceptable sentence with the overt inanimate object pronoun. On the other hand,
232
compared with the experimental sentences used in the subordinate sentence, it seems that
the informants would more readily accept the grammatical experimental sentence with
inanimate null object in the main sentences. As we have already found out, there is no
significant difference found in the judgments between the CFL learner groups and the
native speaker Group 5 on the control sentences of CFL/INO-MS and CFLIINO-SS. This
may partly be due to the reason that overt inanimate object pronoun can be used for
strong emphasis in Chinese in certain context and thus the judgments are at a marginal
level.
7.3.4.8. Judgments on the SCFL/ANO-MS and SCFL/ANO-SS sentences
From Charts 7-28 and 7-29, we can find out the CFL learner groups' judgments on
sentences with animate null object (pro) in main sentences (CFL/ANO-MS) and in
subordinate sentences (CFL/ANO-SS). In Charts 7-28 and 7-29, the solid lines with black
squares represent the CFL informants' judgments on the experimental sentences of
CFL/ANO-MS and CFL/ANO-SS, i.e., sentences with animate null object (pro) in object
position. The lines of dashes with circles represent their judgments on the control
sentences, i.e., sentences with animate pronouns in object position.
As we lmow, both animate null object (pro) and object animate pronouns are acceptable
in Chinese. If the overt animate null object or object animate pronoun refers to an
animate entity, although it may make the sentence sound stylistically heavy or "clumsy",
233
it is also normally acceptable to the native speaker for the purpose of emphasis in speech.
As we can find out from Charts 7-28 and 7-29, all groups ofCFL learners
Chart 7-28: Mean scores of AJT on the S-CFL/ANO-MS Sentences by the CFL informants
Chart 7-28: S-CFUANO-MS 3.0,-------------,
2.5 -------------------- ----------------
2.0 ------------------------------------
1.5 ------------------------------------
1.0 --------- ---------------------------
---~---e---~---,~ .5 ------------------------------------
t/)
~ 0.0 ------------------------------------ • Experimental
~ -.5 ------------------------------------ S-CFUANO-MS ctl
~ -1.0 0 ControtS-CFUANO-MS
2 3 4 5
CFL Groups
Chart 7-29: Mean scores of AJT on the S-CFLI ANO-SS Sentences by the CFL informants
Chart 7-29: S-CFUANO-SS 3.0,-------------,
2.5 ------------------------------------
2.0 ------------------------------.-.---
1.5 --------.- --------------------------
1.0 -------:.a:·----------- ------------1- ~· ----0--_
.5<~:: :-::-::E):::::-.::-------------- .:::-.-:(D
~ 0.0 ------------------------------------
~ -.5 ------------------------------------ -
8 .1.0 ------------------------------------ • Experimental Cl) S-CFUANO-SS ~ -1.5 ------------------------------------ --·
:::!: -2.0 0 Control S-CFUANO-SS
1 2 3 4 5
CFL Groups
234
accepted the experimental sentences ofboth CFLIANO-MS and CFL/ANO-SS, i.e.,
sentences with animate null object in object position. There is no significant difference
found between the native speaker Group 5 and the rest of the CFL learner groups, Groups
1-4 (see Tables 7-28 and 7-29 in Appendix 13). In the CFL learners' judgments of the
control sentences ofCFLIANO-MS and CFLIANO-SS (i.e., sentences with overt animate
pronouns in object position), there is an indication that CFL learners' acceptance of the
use of overt animate pronouns in object position is gradually increasing in accordance
with their improvement in Chinese proficiency. In other words, the control sentences
become more acceptable to them once they are more proficient in Mandarin Chinese.
It should be noted that there is slightly less acceptance of the use of overt animate
pronouns in object position among the native speakers in the Group 5 because they are
somewhat aware of the stylistic heaviness of that use. However, for the informants in the
CFL learner groups, they are not clear about the stylistic difference between the use of
overt animate object pronouns and the use of animate null object (pro). The more
advanced learners in CFL Groups 3-4 treat both the experimental and control sentences
as equally acceptable, whereas the native speakers made differentiated judgments.
It is noted that, the informants in CFL Groups 1-4, like the native speakers in Group 5,
showed a preference for the grammatical experimental sentences over the control
sentences in their judgments of the CFLIANO-MS, i.e., sentences with animate null
object (pro) in the main sentences.
235
7.3.4.9. Judgments on the SCFLIINO-Adjunct sentences
Chart 7-30: Mean scores of AJT on the S-CFL/INO-Adjunct Sentences by the CFL informants
Chart 7-30: S-CFL!INO-Adjunct 3.0..--------------.,
2.5 ----------------------------------------
2.0 ----------------------------------------
1.5 ------------------:;,:------~ 1.0 ------ -------------------------------0----" .5 ~:::: ====et == ::::.:::: e;:.; ..,.~~- -----------
f- 0.0 ----------------------------------------
~ 0
-.5 -----------------------------------------
-1.0 ---------------------------------------- • Experimental
~ -1.5 ···---·--·----·--··············--------- S·CFUINO·Adjunct
~ -2.0 ---------------------------------------- --·
ffi .2.5 ---------------------------------------- 0 Control ~ -3.0 S·CFUINO-Adjunct
1 2 3 4 5
CFL Groups
Chart 7-30 indicates the CFL groups' judgments on the sentences ofCFL/INO-Adjunct.
The solid line with black squares represents the experimental sentences, and the line of
dashes with circles represents the control sentences, i.e., ungrammatical sentences. The
control sentences are unacceptable to the native speakers because they contain inanimate
pronoun ta (=it) in object position co-indexed with an argument in an adjunct.
Like the native speakers in the Group 5, it seems that the CFL learners in Groups 1-4
showed a general preference for the grammatical experimental sentences over the control
sentences of CFLIINO-Adjunct. However, no significant difference is found between the
native speaker Group 5 and the rest of the CFL Groups 1-4 in judging the control
sentences ofCFLIINO-Adjunct (see Table 7-30 in Appendix 13). Similarly, there is no
significant difference found between the CFL groups 1-4 and the native speakers in
judging the experimental sentences.
7.3.4.10. Summary ofCFL learners' judgment
236
Most of the CFL informants do not experience much difficulty in the acquisition of null
subject in the subject position of the finite clauses. Except for the beginners, CFL
informants at intermediate or above learner groups judged the null subject in finite
clauses as acceptable. They also accepted the null subject and null object as grammatical
in their acquisition of CFL/NS-NO sentences. Regarding the judgments of the CFL/Exp
WhP and CFL/Exp-RP sentences, the CFL informants at elementary level rejected
sentences without the use of a nominal expletive in subject position in their IL grammar
of Chinese. However, when their proficiency in Chinese gradually improves, they would
become aware of the fact that insertion of the nominal expletive is disallowed in Chinese.
On the judgments of CFL/BGT -SS and CFL/BGT-Wh sentences, almost all the CFL
informants judged the control sentences of non-base-generated topic (with every
argument position lexically filled) as acceptable. The CFL informants' support for the
experimental sentences ofbase-generated topic coindexed with a variable in a sentential
subject or in a wh-island, increases very slowly in accordance with their proficiency
levels in Chinese. It seems that it may take quite some time before the CFL informants
can acquire this feature in Chinese. This is also the case with the judgments of CFL/NGT
(non-gap topics) sentences.
237
Generally speaking, the CFL informants do not have much difficulty in accepting the
grammatical Chinese sentences with null object, which is contrary to the findings of their
EFL counterparts, i.e., the EFL informants are having trouble in rejecting the
ungrammatical English sentences with null object (7.3.3). The CFL informants are aware
of the fact that there is a distinction between the more acceptable sentence with inanimate
null object and the less acceptable sentence with the overt inanimate object pronoun ta
(=it) in object position. However, the CFL learners, especially those at a higher
proficiency level, treated sentences with animate null object (pro) and sentences with
overt animate object pronouns as equally acceptable, since they are not sensitive to the
stylistic clumsiness of the use of overt animate pronouns in object position. The native
speakers notice the difference in use.
7.4. Analysis and discussion of results
7.4.1. The acquisition and unlearning of null subject by CFL/EFL informants
As we can see from the analysis (7.3.3.10), most of the EFL informants, especially
intermediate and post-intermediate learners, did not have much difficulty in rejecting the
ungrammatical experimental English sentences with null subject in the finite clauses.
Most of the CFL informants (see 7.3.4.10) did not experience much difficulty in the
acquisition of null subject in the subject position of the finite clauses. However, this is
contrary to our assumption concerning the directional difficulty in the acquisition and
undoing process of the null subject by EFL/CFL learners respectively. Based on the
238
direct positive evidence available to the learners, we had assumed that it would be more
difficult for EFL informants to undo the null subject in their acquisition of English than
for CFL informants to learn the null subject in their acquisition of Chinese. This is
because EFL learners would presumably not have any direct positive evidence in their
input to demonstrate the fact that English disallows null subject in the finite clauses,
whereas CFL informants would find themselves being exposed to high doses of direct
positive evidence, i.e., null subject in their input. As is indicated above, this assumption is
not borne out at all. There seems to be no directional difficulty in the acquisition and
unlearning of null subject by the CFL/EFL informants respectively. I would like to
explore the possible reasons in the following sections.
7.4.1.2. CFL informants and the acquisition of null subject
We assumed that language acquisition relies on direct positive evidence available to the
learners in the input and that Ll transfer plays an important part in L2 acquisition. Direct
positive evidence is assumed to be the basic type of evidence learners use to construct the
grammar and revise their hypotheses. In the case of CFL informants learning Chinese,
since Chinese allows null subject in the finite clauses, the high frequency of null subjects
in the input data provides direct positive evidence for CFL learners. However, it remains
unexplained how the obviously abundant input data can trigger the change in the learners'
IL grammar of Chinese. In the acquisition of null subject by CFL informants, learners are
exposed to input data from the TL (Chinese), which is inconsistent with the values of
language properties in the Ll grammar (English). It might be the fact that the
239
contradictory input data from Chinese triggers the restructuring in the learners' IL
grammars of Chinese. While being exposed to high frequency input data of null subjects,
the restructuring process has reached a "triggering threshold" (Randall, 1987; 1992)
which would enable the null subjects acquire a positive value in the learners' IL
grammars of Chinese. However, this explanation is not without problems.
If the direct positive evidence were the only source of input in their acquisition, it would
be difficult for CFL informants to differentiate null subject (Subject PRO) from pro in
UG, because they function differently. The direct positive evidence in the input can only
inform CFL informants that kind of "gap" is allowed in subject position in a finite clause
in Chinese. It says nothing about the type of gap, and the status of the lexically filled
subject NP. It is concluded that, relying on direct positive evidence alone, there would be
insufficient information to trigger the necessary change in CFL informants' IL grammars
of Chinese.
Randall (1987, 1992) suggested that indirect positive evidence could trigger a substantial
change in the learners' IL grammars. His suggestion has shed some light on the
explanation of substantial change in CFL informants' IL grammars of Chinese. According
to him, the indirect positive evidence is based on a principle of the form [IF P THEN
NOT Q], which means that if construction P occurs in the language, construction Q
cannot occur (see 3.7.2). In other words, the learner can get information about the non
occurrence of Q indirectly through the presence of P in the input, because positive
evidence of a structure can serve as indirect evidence for the occurrence or non
occurrence of a related structure.
240
In the acquisition of the null subject in Chinese by CFL informants, I believe that the
non-occurrence of the features of AGR and TNS in Chinese triggers the acquisition of the
null subject by CFL informants. In terms of Randall's suggestion, the evidence ofNOT P
is [-AGR] and [-TNS], and Q represents the realization of[+ the null subject]. That is to
say, CFL informants can easily notice the absence of agreement markers and tense
markers in the input data during their acquisition of Chinese. Whereas English makes use
of do-support, be-insertion and copula to "bear" AGR and TNS features ( cf. Ouhalla,
1991; Yuan, 1997), Chinese does not use them because there is "no AGR or TNS" for
them to bear (Yuan, 1997). The indirect positive evidence would indicate to CFL
informants that the subject position in the finite clause in Chinese is ungoverned and null
subject is allowed. Meanwhile, CFL informants would get to know that the lexically
filled NPs or pronominals in subject position in Chinese have inherent case, not structural
case. I believe that the indirect positive evidence is more useful to CFL informants in
their acquisition of null subject than direct positive evidence, but the latter can be used as
a reinforcement feature in the acquisition process. Once the informants find out that AGR
and TNS are absent in Chinese, they are able to predict that null subjects are allowed in
the finite clause in Chinese, and the direct positive evidence reinforces this assumption in
their acquisition of null subject in finite clauses.
241
7.4.1.3. EFL informants and the undo process of null subject
What is the acquisition mechanism for EFL informants? I had predicted that there is no
direct positive evidence that could indicate to EFL informants that, null subject is
disallowed in the TL, English. It would be more difficult for EFL informants to undo the
null subject setting in the finite clause during their acquisition of English. However, our
results have indicated that this is not the case. EFL informants did not experience much
difficulty in rejecting the incorrect English sentences with null subjects. How could this
happen? Why are EFL informants able to reject the ungrammatical English sentences
with null subjects in finite clauses?
As I have shown above, there is no direct positive evidence available to EFL informants,
informing them that null subjects are disallowed in finite clauses in English.
Nevertheless, one could argue that indirect negative evidence is still available to them,
which may help them to trigger the change in their IL grammars of English. Recall that
indirect negative evidence refers to the non-occurrence of some expected language
features that can cause change in the IL grammar (Yip, 1995: 59; see also 3.7.4.). EFL
informants would normally expect the appearance of null subjects in English as in their
Ll, Chinese. However, the expectation is not realized since the null subjects never show
up in the input, except in some child English sentences (see discussion later). EFL
informants might infer from this observation, that null subjects are simply disallowed in
adult English. Having noted this, it would be naive to conclude that the non-occurrence of
a linguistic feature in the finite input data can be used as a solid evidence for the
ungrammaticality of the structure, since language is infinite. What language data the EFL
242
informants are exposed to are also infinite. Thus, non-occurrence is not reliable evidence
for this acquisition process of change of grammar. Moreover, even if non-occurrence is
said to be useful evidence for language learners (Bowerman, 1983, 1985), it may not be
available, or ineffective and inefficient in the EFL informants' input data. The EFL
informants' exposure data could be quite misleading insofar as they might think that
subjects in finite clauses in English are "optional" (Hyams, 1996: 92), like those in child
language. This is misleading since their input data contain both grammatical and
ungrammatical sentences.
I have mentioned that null subjects can be found in some child English sentences
(Hyams, 1983, 1986, 1996) or in English and French diaries6 (Haegeman, 1990; Pierce,
1992), Some early null subject examples can be found in the following sentences:
(31). English: A. Want look a man
(from Hyams (1983))
B. a). Hug Mommy.
b). Play bed.
c). Writing book.
6 Haegeman (1990) observed that early null subjects are found in certain abbreviated varieties of English and French languages in the register of diaries, such as the following:
(1). A very sensible day yesterday._ saw noone. _took the bus to Southwark Bridge. _ walked along Thames Street ....
(2). _ m'accompagne au Mercure, puis ala fare ... '(he) takes me to Mercure, then to the station .. .'
s'est donne sou vent l'illusion de l'amour .. . '(he) often gave himself the illusion oflove ... ' _me demande si ... je lui eus montre les notes ... '(I) ask myself if ... I would have shown him the notes' (data from Haegeman (1990); cited by Rizzi, 1994: 254)
243
d). See running.
(from Bowerman (1973), citing Bloom (1990: 491))
(32). French: A. a). est pas gros
(is not big)
b). va checher l'auto
(goes to look for the car)
c). est casse
(is broken)
d). boit
(drinks)
(from Pierce, 1992: 109)
Around the age of two and a half years, children typically produce sentences with null
subjects at frequencies between 30% (cf. Valian, 1989) and 65% (cf. Jaeggli & Hyams,
(1988), based on the Brown corpus for the child Adam). Many researchers, under the
heading of the Null Subject Parameter7 (Rizzi, 1982, 1986; Jaeggli & Safir, 1989), have
tried to explore whether the child null subject phenomenon was related to the null subject
option that many adult languages possess. However, research during the past decade has
provided a strongly negative answer, i.e. "early subject omission is not like the option
found in adult Null Subject Languages" (Rizzi, 1998: 15).
7 Children drop subjects quite freely around the age of two (cf. Rizzi, 1982, 1994), regardless of whether or not the target language is a null subject language. However, learners drop subjects freely but not obligatory objects (Valian, 1991; Wang, Lillo-Martin, Best & Levitt, 1992; Hyams & Wexler, 1993;). Hyams (1986) described this phenomenon in terms of the Null Subject parameter (Jaeggli & Safir, 1989; Rizzi, 1982, 1986, 1994) and proposed that the initial setting is the null subject value, or [+]null subject. Learners of languages such as English and French need to reset the parameter on the basis of their experience, and "this is normally done a few months after the second birthday" (Rizzi, 1994: 250).
244
According to Valian (1990a, 1990b), neither a[-] null subject nor a[+] null subject value
can adequately characterize children's initial assumption about the setting of the null
subject parameter8• Her arguments rest on the notions of parser limitations and
misleading input. In her view, positive evidence is not sufficient to trigger a parametric
change since children's [-null subject] grammar would not allow them to appreciate the
significance of the input received. Remember that the learners need to be exposed to
certain types of triggering evidence to set parameters correctly. So, a Chinese learner
starting off with a[+ null subject] grammar would take sentences with the non-
appearance of null referential subjects as evidence for the need to reset the parameter to
[-null subject].
There are cases, however, in which the type of input the EFL learner receives mimics the
trigger in some respect, as exemplified below in (33) and (34):
(33). a. Seems like the class always wakes up five minutes before the bell
rings.
b. Feels like real silk.
8 There are two possible initial settings proposed for the null subject parameter: one in which the initial setting is[+] null subject (or[+] morphologically uniform) (Hyams, 1987a, 1991) and one in which the initial setting is[-] null subject (Bloom, 1990, 1993). However, Valian (1990a, 1990b) argues that, because of inherent limitations in the learner's parser as well as the presence of misleading input, the single-value theory, i.e., the learner begins acquisition with the null subject parameter set for either the English-like value or the Italian-like value, cannot work, no matter which value is chosen as the initial one. Valian (1990a, 1990b) proposed a radically alternative dual-value view, arguing that the learner embarks on the acquisition process with both values of the parameter available, and uses theory-confirmation procedures to decide which value is best supported by the available data.
c. Guess I should be going.
c. Ever been to Chicago?
(from Schmerling, 1973: 580-582)
(34). a. Wanna cookies?
b. Bet you can't stack those blocks.
c. Gonna get your lunch, and I'll be right back.
(from Schulte-Nafeh & Dussias, 1993: 103)
245
These cases are labeled "misleading input" because they can potentially cause the EFL
learner hearing English to arrive at the wrong conclusion about their TL, English. As we
can see, the sentences in (33) and (34) are not totally grammatical, because they do not
have overt subjects, which is a requirement of the English grammar. Nevertheless, these
sentences are acceptable in certain contexts, especially in spoken discourse. EFL learners
are exposed to fully grammatical utterances without subjects, in the form of imperatives.
Examples of imperative sentences, or sentences such as Wanna go?, Want lunch now? or
Don't know (cf. Examples in (34)) would normally be acceptable in colloquial English.
In fact, for native speakers of English, an acceptable sentence does not have to be fully
grammatical (Valian, 1991: 33); but "it does have to be comprehensible or appropriate"
(Valian, 1990a: 121). Among native speakers of English, the sentences in (33) and (34)
are not uncommon at all. Hearing repeated instances of sentences as in (33) and (34),
without overt subjects, EFL learners could easily be misled into thinking that English,
like their L1 Chinese, allows null subjects in finite clauses. These instances could cause
EFL learners to incorrectly conclude that they are learning a [+null subject] language. By
246
contrast, native English speakers, it is argued, are always aware that sentences like those
in (33) and (34) are colloquial. That is to say, native English speakers regard those
sentences as not 'standard'. However, it is more difficult for EFL learners to identify
these sentences as 'non-standard' in TL English, taking into consideration that these
sentences in (33) and (34) are standard in EFL learners' L1, Chinese.
As Haegeman (1990) has observed in her findings, whereas the main subjects can be
dropped, embedded subjects cannot, for example:
(35). A. can't find the letter that I need.
B. *I can't find the letter that need.
(from Haegeman, 1990; citing Rizzi, 1994: 255)
EFL learners are aware that if the null subject occurs in an embedded clause, it would
result in an ungrammatical and unacceptable sentence. They could deduce that null
subject is disallowed in the English grammar. As Valian (1990a: 132) points out, "that
subject omission is ungrammatical in English can be seen from the fact that it is both
ungrammatical and unacceptable if the subject follows a subordinating (but not
coordinating) conjunction, or the subject is in an embedded clause. One can acceptably
say, 'think I saw him', but not 'I think saw him'."
Roeper and Weissenborn (1990: 155) suggest that subordinate clauses can provide a
locus for "unique triggers" for setting the pro-drop parameter. Valian (1990a, 1990b)
found that none of the English speaking children in her study omitted subjects in tensed
embedded clauses, supporting Roeper and Weissenborn's (1990) proposal for a unique
247
trigger for the subordinate clause. But, is the lmowledge of such a unique trigger
available to the adult L2 learner as well? Some evidence indicates that this is not the
One of the questions that still needs to be answered is: what triggers the change in EFL
learners' IL grammar of English? Why, contrary to our expectations, are EFL learners
able to reject ungrammatical English sentences with null subject? I have suggested that,
during the acquisition of English, EFL learners have no direct positive evidence in the
input data that can enable them unlearn the null subject. The argument of indirect
negative evidence also has some problems. As pointed out by Saleemi (1992: 135),
"positive data are still considered to be the core of the primary data, with indirect
negative evidence having an essential but supplementary role, and direct negative
evidence remaining irrelevant to the process of acquisition"10• As I have indicated above,
direct positive evidence and indirect negative evidence are not proper candidates for this
undo process of null subject acquisition. However, if we adopt Randall's (1987)
postulation [IF P THEN NOT Q], it is argued that indirect positive evidence can play an
active role in this undo or unlearning acquisition process. 'EFL learners would become
9 Though an interesting theory by itself, it is not obvious whether Roeper and Weissenborn's (1990) proposal can be applied in the context of adult SLA. An examination of the results of White's (1985b) study of pro-drop parameter in adult SLA indicates that the Spanish speaking subjects accepted English sentences with missing subjects in tensed embedded clauses. So it is doubtful whether the knowledge of the subordinate clause, which functions as a unique trigger, can be available to the adult L2 learner as well ( cf. Gass & Lakshmanan, 1991:200).
10 In the case of the Null Subject Parameter, learners will expect to hear null subject sentences as well as sentences with subjects phonetically filled. When the input data provides no sentences with subjects, that constitutes indirect negative evidence. On the other hand, direct negative evidence would only occur if the environment directly informs learners that there were no sentences with phonetically realized subjects. Generally speaking, indirect negative evidence could occur in the form of the absence of sentences; direct negative evidence, however, occurs rarely, if at all (Valian, 1990b).
248
aware of the fact that null subjects are disallowed in English and the subject position in
the English finite clause must be lexically filled ( cf. 7 .4.1.2.). Through the presence of P,
EFL learners can receive information about the non-occurrence of Q, i.e., the null subject
in English. Here, the evidence of P refers to the triggering data, which indicate the
presence of AGR and TNS in English, including tense and agreement markers, such as
the copula, do-support, the progressive auxiliary be (cf. Yuan, 1997). Yuan (1997)
investigated the unlearning of null subjects and null objects by 159 Chinese learners in
their SLA of English. He proposed that EFL learners reject the null subject "as a result of
the specification of the contrastive values ofTNS and AGR in the learners' L2 grammar
of English" (ibid., p.489). He used the verbal expletives do in do-support and copula be
as examples to illustrate how EFL learners can be made aware of the specification of
AGR and TNS in English. According to him, EFL learners would not be able to use the
features of AGR and TNS proposals at the beginning stages of acquisition. However, the
encounter with these informative data in the input is sufficient to make them aware of the
presence of AGR and TNS features in English. I agree with this point that EFL learners
will eventually acquire that "AGR and TNS are specified in English" (ibid., p.491) when
their proficiency level improves gradually. Recall that in Chinese, there is no subject-verb
agreement at all. Because of this distinction, EFL learners can infer that null subject is
disallowed in English finite clauses and the subject position in English must be lexically
filled to receive the nominative case assigned by AGR.
7.4.2. The acquisition of nominal expletive it by EFL/CFL informants
Expletive subjects are also called "dummy" subjects or pleonastic subjects. They are
referentially empty NPs whose only function is to fill an unoccupied subject position
(also see 4.2.7.). In English, the expletive subjects it and there can occur in raising
predicate sentences, as in the following examples:
(36). It is possible that I might quit smoking.
(3 7). There is a cat in the garden.
Expletive it also occurs in weather predicate sentences, such as:
(38). It is raining.
(39). It is snowing.
(40). It's cold outside.
249
However, Chinese sentences do not have such expletive subjects. Compare the following
examples:
(41). Zai xia yu le.
PROG11 fall rain PRT
(It's) raining.
11 PROG refers to progressive aspect.
250
(42). Keneng12 wo hui zaodian wancheng.
Possible I will earlier finish
(It's) possible (that) I will finish earlier.
(examples from Yip, 1995: 83)
In Sections 7.3.3.3 and 7.3.4.8, we have indicated that the insertion of expletives in
subject position results in ungrammatical sentences in Chinese. We found that CFL
informants noticed the non-occurrence of AGR elements in Chinese, leading them to the
conclusion that Chinese does not require nominal expletives. In our experiments (see
7.3.4.8.), most CFL informants accepted the null subject in sentences with weather
predicates, with the exception of some informants in low proficient elementary level who
were not confident enough to make the correct judgements. Since weather predicates in
Chinese can sometimes have thematic subjects13, it may be that some elementary level
beginning learners of CFL informants take the nominal expletive as a referential pronoun.
12 Linguist such as Li (1990) argues that, on theoretical grounds, Chinese lacks the expletive null subject proexp ( exp refers to experiential aspect). Li argues that the example of ( 42) may involve use of ken eng as an adverb. As pointed out by Yip (1995), it seems necessary to recognize in Chinese a non-referential null subject in a number of constructions, such as the example given by Gao et al (1994), arguing that proexp occurs as the subject of impersonal predicates such as lun in the following example:
prOexp lun ni xi wan tum you wash dish
'It's your tum to wash the dishes.' Yip (1995) suggests that Chinese is a type (d) language with respect to null subjects (see Saleemi (1992: 112) for a classification of four different types oflanguage with respect to null subjects), lacking overt expletives but allowing null expletive subjects. 13 CFL informants might have encountered sentences below because weather predicates in Chinese can optionally take thematic subjects.
(1). shangzhou (tianqi) hen re. Last week, weather very hot *"Last week the weather was very hot".
(2) wo xiake de shihou, (tian) zheng xiazhe da xue. I finish class DE time sky PRG fall big snow *"When I finished the class, the sky was snowing heavily."
DE refers to definiteness effect, a modifying marker that occurs at the end of a prenominal modifier. PRG refers to progressive aspect.
251
In Hyams' (1986) early proposal for the pro-drop parameter theory, the parameter has a
default value setting of [ + pro-drop] in child L 1 acquisition of English. When English
speaking children notice the presence of expletives, such as the pleonastic it or existential
there, they trigger the resetting of the parameter [ + pro-drop] to [- pro-drop]. However,
can the use of expletive it in English be informative to EFL learners and have any
triggering effect on their resetting from null subject to non-null subject? The answer is
not so obvious. The obligatory use of the expletive in weather predicates and raising
predicates sentences "has, if any, very limited triggering effects on the unlearning of the
null subject" by EFL learners (Yuan, 1997: 489). For the EFL informants, they
sometimes encounter examples of ungrammatical but acceptable sentences in colloquial
English (see examples in (33) and (34)), which do not have the required expletive in the
subject position. From our experiment (see 7.3.3.3, Chart 7-5), it can be seen that the
most advanced group of EFL learners experience a rather dramatic improvement in the
rejection of experimental sentences, i.e. the raising-predicate sentences without the
expletive it. However, the rejection process stops at advanced levels, with native
speakers' judgment at a marginal level. This is due to the fact that some advanced
learners as well as the native speakers have accepted these kind of structures because of
their intensive exposure to colloquial English usage. It is suggested that, with the
presence of AGR and TNS in English, this indirect positive evidence could trigger the
acceptance of expletive it and the rejection of null subject in EFL learners' IL grammar of
English. Since AGR and TNS are specified in English, Chinese learners of English are
forced to find something to fill in the subject position with weather predicate or raising
predicate sentences.
252
7.4.3. The acquisition ofBGT by EFL/CFL informants
Both Chinese and English grammars generate sentences with topics. However, according
to Li & Thompson (1976a), the distinction between "topic-prominent" and "subject-
prominent" languages can be related to the distinction between discourse-oriented
languages (like Chinese) and sentence-oriented ones. In subject-prominent languages like
English, all sentences must have subjects, which accounts for the presence of pleonastic
expletives such as it and there in English. Chinese, on the other hand, a topic-prominent
language, does not have such pleonastic elements since structural subjects are not
obligatory elements. Moreover, in Chinese language, "sentences of the form topic-
comment abound and must count as basic forms in that they cannot be plausibly derived
from other, 'more basic' forms" (C-T. J. Huang, 1984: 550). The derived topic/non-base-
generated topic structure and base-generated topic structures are exemplified in ( 43) and
( 44), respectively.
(43). A. Pijiui wo bu he Beer I not drink 'Beer I don't drink.'
B. Wine I drink, but beer I don't.
(44). A. Pijiu wo zui xihuan Qingdao. Beer I most like Tsingtao 'As for beer, I like Tsingtao [a brand of beer] the best.'
B. Yosemite pubu hen piaoliang Yosemite waterfall very beautiful 'As for Yosemite, the waterfalls are very beautiful.' (examples from Yip, 1995: 78)
In example (43), as in English, the topic construction in Chinese is derived by movement
of a non-subject to the topic position. However, the base-generated topic in examples
14 PRT refers to sentence final particle.
253
(44a) and (44b) can not be derived by any movement. Chinese topics leave no "gaps",
and they are "base-generated" rather than derived by movement from some other position
in the sentence (also see Section 4.2.7. in Chapter 4.).
In 7.3.3.4 and 7.3.3.5, we found that in judging the BGT (base-generated topic)
experimental sentences, there was a gradual improvement pattern in judgment accuracy.
Generally speaking, when EFL informants become more proficient, they are more
determinate in rejecting the incorrect experimental sentences. That is to say, beginning
EFL learners and some intermediate EFL learners are unable to reject the base-generated
topics because their proficiency level has not reached a relatively competent stage to be
able to deal with these sort of English sentences in the experiment. However, the
improvement in proficiency has resulted in more rejection of the BGT in English.
As we all know, the EFL informants' Ll (Chinese) allows base-generated topic in
sentence construction. But how does this change of grammar happen, i.e., what has
triggered the change in the informants' IL grammars which rejected the BGT?
During the acquisition of English, the presence of AGR and TNS in English can serve as
indirect positive evidence, revealing EFL informants that BGT is not possible in English.
EFL learners will initially assume that, like their Ll Chinese counterparts, both the topic
and the object variable in English can be base-generated, thus allowing sentences with the
non-gap topic and with the topic co-indexed with a variable in a wh-island as possible
sentences in English. Many beginning learners and some early intermediate learners
accepted those sentences (see 7.3.3.4 and 7.3.3.5.). However, since the input of AGR in
254
English is frequent and sufficient, EFL learners will learn that the subject position in
English is assigned nominative case, rather than inherent case. This information will tell
the learners indirectly that the topic in English has to be case-marked structurally as well.
fu order to have structural case, the topic must have a preceding preposition to assign
oblique case to it. Otherwise, movement has to be involved. In this way, as their
proficiency improved, EFL learners started to accept the non-gap topic sentence with a
preceding preposition, rejecting the sentences with the object variable co-indexed with a
topic in a wh-island or in an extraposed clause.
For the CFL informants (see 7.3.4.4. and 7.3.4.5.), elementary, intermediate and high
intermediate learners were all reluctant to accept the experimental sentences ofBGT and
the non-gap topic. Almost all informants judged the control sentences as acceptable,
while their support for the experimental sentences was constrained by their proficiency
levels in Chinese. Even though the CFL learners had available positive evidence, there
was little corresponding increase in the acceptability ofBGT and non-gap topics in the
CFL learner's IL grammars of Chinese.
In my experiment, CFL learner Groups 1-4 judged the experimental sentences ofBGT
quite differently from the native speakers in CFL learner Group 5. Though CFL Group 4
were high intermediate learners who had been learning Chinese for around ten years,
their acceptance of the BGT sentences was far from straightforward yet increasing
steadily in accordance with their proficiency levels.
255
In Yuan's (1995) study ofEnglish-speaking learners' IL grammars of Chinese, he
suggested that the processing factors, grammatical factors and misleading evidence were
to blame for the difficulty of acceptance of the base-generated topic. Following Hawkins
(1990), he adopted the principle of Minimal Attachment ("postulate the fewest nodes
consistent with the grammar"), proposed by Frazier (1978, 1985) and Frazier and Rayner
(1988), to account for the parsing of Chinese sentences with base-generated topics. He
suggests that the "English-speaking learner is likely to process the base-generated topic
as a subject NP and require the backtracking strategy to reanalyze the topic after
subsequently encountering the subject NP" (Yuan, 1995: 593). The base-generated topic
can cause even more difficulty for the learner's parser when it is long and complex. Yuan
further suggests that because of the differences between topic-prominent and subject-
prominent languages, native speakers of Chinese and English "should accordingly
demonstrate certain differences in their organization of parsing procedures" (ibid.,
p.593)15• However, whether these differences between topic-prominent and subject-
prominent languages cause parsing difficulties still needs to be tested empirically. In fact,
I believe that the CFL learner's late acquisition of base-generated topics is more related
to grammatical factors in the development of their IL grammars of Chinese.
On the issue of initial state ofL2 acquisition, according to Vainikka and Young-Scholten
(1994), the functional projections such as CP and IP initially are missing in L2 grammar
and are acquired only later on. Eubank (1994) believes that they may initially be
underspecified and only develop fully later on. Following Yuan (1995), I assume that the
15 Since English is recognized as a subject-prominent language, topicalization is also possible. However, according to Li and Thompson (1976), use of the topic in English is much lower in frequency and less productive than in topic-prominent languages such as Chinese. A subject, a wh-word, or a complementizer normally occupies the initial position of a sentence in English.
256
syntax of Chinese comprises functional projections of IP and CP, and that base-generated
topics in Chinese are generated in SPEC ofCP. Beginning learners of Chinese would
have substantial difficulty in processing Chinese sentences with a topic since CP is absent
from initial L2 development, and there is no space for the topic in their L2 syntax. In my
experiment, beginning learners had no problem accepting control sentences, i.e.,
sentences without BGT and were reluctant to accept experimental ones, i.e., sentences
with BGT. However, the situation changes when the informants' proficiency level
increases. More advanced learners accept the control sentences as well as the
experimental ones because those sentences were all acceptable to the native speakers of
Chinese to some extent. My result has supported Yuan's (1995) suggestion that, CP
projection is missing from initial L2 acquisition of Chinese and it is not posited in the L2
Chinese grammar "until the learner has reached an advanced level" (p.597). Beginning
and intermediate learners can not make use of the positive evidence which can trigger CP
projection in L2 Chinese grammars, since the input data which can trigger the projection
ofCP in L2 Chinese grammars are sentences with topics. Without CP projection, CFL
learners do not judge the experimental sentences of BGT as acceptable.
7.4.4. The acquisition of null objects by EFL/CFL informants
As noted in Sections 7.3.3.6 to 7.3.3.9, in judging experimental sentences with null
object, most of the EFL informants accepted the ungrammatical English sentences with
null object. Most learner groups were unable to reject the ungrammatical experimental
sentences with inanimate null object either in the main sentences or in the subordinate
sentences. This is an indication that EFL learners are reluctant to reject the
257
ungrammatical English sentence with null object, their improvement in proficiency has
little effect on their IL grammars of English.
The CFL informants (see results from 7.3.4.6 to 7.3.4.9), however, do not have much
difficulty in accepting the grammatical Chinese sentences with null object, which is
contrary to the findings of their EFL counter-parts. The CFL informants are intuitively
aware of the fact that there is a distinction between the more acceptable sentences with
inanimate null objects and the less acceptable sentences with the overt inanimate object
pronoun ta in object position. However, the CFL informants, especially those at a higher
proficiency level, treat sentences with animate null object and sentences with overt
animate object pronouns as equally acceptable, since they are not sensitive to the stylistic
clumsiness of overt animate pronouns in object position. Native speakers, however, may
notice the difference in use.
Why do EFL learners accept the experimental sentences with null object, and why are
they reluctant to unlearn null object in their acquisition ofiL grammars of English? Why,
on the other hand, do CFL learners not have much difficulty in acquiring the language
feature of null object during their acquisition of Chinese? As we already mentioned,
Chinese is a discourse-oriented language (Li & Thompson, 1976a; C.-T.J. Huang, 1984),
which has topic-drop as a common language feature. In Chinese, the null object is a result
of moving the object to the topic position and then having it deleted by the Topic NP
Deletion Rule (C.-T. J. Huang, 1984a), if it is identifiable with a topic in a topic chain. In
English, topicalization is allowed whereas topic-drop is not permitted. With respect to the
258
language feature of topic-drop, Chinese has a setting of [ + topic-drop] and English [-
topic-drop]. In this case, Chinese is more inclusive than English in that it allows sentence
structures that are impossible in English.
According to Gass (1997), if CFL learners select the more restricted grammar, i.e.
English, then the evidence from the input alone (which includes the full range of possible
Chinese sentences) immediately allows these learners to modify their hypothesis because
sentences in the superset are heard. Since CFL learners have sufficient direct evidence16
informing them that null objects are possible in their target language (Chinese), most
CFL learners do not have much difficulty in acquiring null object in Chinese. During
their acquisition of IL grammars of Chinese, CFL informants come across Chinese
sentences with phonetically null objects at a very high frequency. This information could
be sufficient for CFL learners to realize that null objects are allowed in Chinese and this
input could serve as a trigger for the change or restructuring in the CFL learners' IL
grammars of Chinese.
Shi (1989, 1992) believes that, like the CP in English, the topic chain is the largest
syntactic unit in Chinese. Following Shi (1989, 1992), Yuan (1997) suggests that one of
the syntactic differences between English and Chinese is that English has CP as its largest
syntactic unit, whereas Chinese has topic chain as its largest syntactic unit. Moreover, the
topic chain may consist of more than one CP and has all the functions normally assumed
16 Parameters are clusters of properties. According to Gass (1997), if one is learned, they are all learned. fu consideration of a parameter that assumes a clustering of grammatical structures, the input serves only as a catalyst to trigger certain changes in the learner's grammar.
259
for CP. During their acquisition of English, EFL learners "may incorrectly take some
semantically closely-related sentences as a TC and co-index the null object with some NP
within this assumed topic chain" (ibid., p.491). However, since nothing in the input data
can inform the EFL learner that this is not the case, they cannot "unlearn" or "undo" the
more inclusive [ + topic-drop] setting without the necessary positive evidence in the target
language English.
From a learnability point of view, the presence of the more inclusive setting [ + topic-
drop] in the learners' L1 may become a persistent source of interference in the acquisition
of English. According to White (1985, 1987), if a certain more inclusive parameter has
already been set in a particular way because of positive evidence in the L1, it would be
difficult for the L2 learner to undo this setting without some necessary positive evidence.
Thus, EFL learners find it very hard or impossible to unlearn or "undo" this more
inclusive setting in the absence of necessary positive evidence.
However, do EFL learners have indirect negative evidence 17 (Chomsky, 1981) in
unsetting the null object in their acquisition of English? The answer is obviously no,
since the indirect negative input is not reliable in language acquisition. As Pinker (1996)
pointed out, "since language is infinite and creative, any finite sample of input is bound
to lack many (indeed, an infmite number of) perfectly grammatical sentences" (p.231 ).
Meanwhile, it would be unimaginable for the learner to keep track of a great range of
17 Chomsky (1981) suggests that a learner might resort to the use of"indirect negative evidence". If a certain form never appears in the input, the learner might assume that "it is ungrammatical, without requiring overt corrections, disapproval, evidence of comprehension failure, and so on" (Pinker, 1996: 321).
260
sentence types together with their ratios of occurrence and non-occurrence (Braine,
1971). Moreover, as pointed out by Valian (1988, 1990a), there is no principled way of
knowing how long one should wait before concluding that a specific structure will never
emerge in the data. Therefore, the argument of indirect negative evidence has several
problems. It is suggested that the indirect negative evidence is not a reliable source of
input in language acquisition.
In a word, there is no positive evidence in English, whether direct or indirect, which can
demonstrate that null objects are impossible. The idea of indirect negative evidence
during the acquisition of English by EFL learners is also refuted. Since EFL learners do
not have evidence for the impossibility of the null object in English, EFL learners would
have more difficulty rejecting the ungrammatical sentences with null object than the CFL
learners do. Because of the lack of informative evidence, EFL learners do not know that
the null objects are disallowed in English. Meanwhile, the setting of[+ topic-drop] could
become a persistent source of interference in the acquisition ofEnglish due to the fact
that EFL learners have already had the setting of [ + topic-drop] in their L 1. It would be
difficult for them to "undo" this more inclusive setting in the absence of any positive
evidence in the target language. It is suggested that interaction between the Ll setting and
the IL grammar of the target language can probably delay this unlearning process further.
7.5. Conclusion
The results presented in this chapter indicate that there is no directional difficulty in the
acquisition and undo process of the null subject by EFL/CFL informants, which is
261
contrary to our assumption. For EFL learners, there is no direct positive evidence
available in the input that can indicate to EFL learners that English does not permit null
subject in finite sentences. However, sufficient indirect positive evidence, including do
support, copula and the auxiliary, etc., can inform the EFL learners about this difference
and thus trigger the necessary change in their IL grammars towards the TL. During the
learners' learning process, i.e., the acquisition and undo process of the null subject, the
indirect positive evidence reflecting the AGR and TNS features (i.e. the presence or
absence of AGR and TNS in Chinese) can help EFLand CFL learners acquire and
unlearn the null subject. For the CFL learners, once the learners found out that AGR and
TNS are absent in Chinese, they would be able to predict that null subjects are allowed in
the finite clause in Chinese. As for EFL learners, once their proficiency in English
improves, they would eventually be aware of the fact that AGR and TNS are specified in
English, enabling them to acquire and unlearn the null subject during the acquisition
process.
The acquisition and rejection of nominal expletives by EFLand CFL learners are also
related to the influence of the indirect positive evidence, such as the occurrence or non
occurrence of AGR features in the TL. We found that CFL learners have noticed the non
occurrence of AGR features in Chinese, indicating to them that Chinese does not require
nominal expletives. In our study, most informants accepted the null subject in the
experimental sentences.
262
This study has found that the acquisition and rejection of base-generated topics by CFL
learners are not as easy as for EFL learners, though we predicted that might be the case.
For the EFL learners, once again, the presence of AGR and TNS during the acquisition of
English can serve as indirect positive evidence, indicating to EFL learners that BGT is
impossible in English. Thus many EFL informants accepted the experimental sentences.
Most elementary and intermediate CFL learners were reluctant to accept the experimental
sentences ofBGT and the non-gap topic, unless they had reached a relatively advanced
level.
During the acquisition and unlearning of the null object by EFLand CFL learners, the
direction of difficulty moves from Chinese to English. That is to say, CFL learners
normally accept grammatical Chinese sentences with null object, whereas most of the
EFL learners reject the ungrammatical English sentences with null object. The lack of
informative evidence is blamed for the difficult unlearning process the EFL learners have
experienced, whereas sufficient direct positive evidence has made it relatively easy for
CFL learners to acquire the null object during their acquisition of Chinese. It is suggested
that it would be more difficult for the learners to change from a more inclusive grammar
to a less inclusive one, if insufficient evidence is available, resulting in the occurrence of
the triggering.
263
Chapter 8 The acquisition of reflexives
8.0 Introduction
As is shown in 5.1.1, English reflexives must have their antecedent in the local clause,
whereas the Chinese reflexive ziji can take any subject NP to be its antecedent ( cf. Cole
& Sung, 1994; Wexler & Manzini, 1987). fu the following examples, only Harry can
serve as an antecedent for the reflexive himself, but not Bill or Tom in (1 ), while ziji can
refer to any of the three NPs, namely, Lulu, Liping or Xiaotong in (2).
(1 ). Tomi thinks Billj knows Harryk likes himself *i/*jtk·
(2). Lulu i shuo Lipingj yiwei Xiaotongk zeguai ziji iljtk Lulu say Liping think Xiaotong blame self
"Lulu says Liping thinks Xiaotong blames her/herself'.
Whereas English has only phrasal reflexive, Chinese has two types, the bare reflexive ziji
and the phrasal reflexive in the form of "pronoun+ ziji'' (such as ta-ziji 'himself/herself
or ni-ziji 'yourself). Both Chinese and English phrasal reflexives are short-distance
reflexives (i.e. they have to take a strictly local antecedent). Being blocked by some
barriers from moving up the tree, they cannot move out of their own clauses to undergo
successive movements. When the features of the antecedent in different clauses are not
matched with one another, the long-distance binding of ziji is blocked. As examples (3) to
(5) indicate, ziji can only refer to the subjects in the embedded clauses Zhangsan, ni 'you'
and Lisi respectively.
(3). Woi zhidao Zhangsanj piping le Zl]l *ilj· I think Zhangsan criticize ASP self
"I know that Zhangsan criticized self'.
( 4). Zhangsani zhidao niJ p1pmg le Zl]l *i!J Zhangsan know you criticize ASP self "Zhangsan knows that you criticized self'.
( 5). Zhangsani juede WOJ zhidao Lisik piping le Zl]l *it*Jtk Zhangsan think I know Lisi criticize ASP self "Zhangsan thinks that I kilow that Lisi criticized self'.
264
In the following sections, I shall examine the acquisition and undo process of the Chinese
long-distance reflexive (LDR) ziji by CFL and EFL learners respectively. It is
hypothesized that it will be easier for the CFL informants to acquire the Chinese LDR in
their acquisition of Chinese than for the EFL informants to undo the LDR in their
acquisition of English. I have argued in previous sections (e.g. Section 7.4.) that language
acquisition relies on the positive evidence available to learners in the input, whether
direct or indirect. Positive evidence can trigger a substantial change in CFL/EFL learners'
IL grammars, causing the acquisition or the undoing process to take place towards the
norm of the target language. However, based on positive evidence in the language input
alone, can we explain how CFL learners acquire the knowledge that Chinese ziji is long-
distance bound? Are they able to acquire the knowledge that ziji is not long-distance
bound when the potential antecedents do not agree with each other in person or number?
I will answer these questions in the following sections.
8. 1. Reflexive binding and language acquisition
A number of theories have been proposed to account for cross-linguistic variations in the
properties of binding in general and the Chinese reflexives in particular (see Chapter 5).
Wexler and Manzini (1987) developed a modular theory of parameter setting which
assumes Chomsky's (1986a: 166) Binding Principle (i.e., an anaphor is bound in its local
265
domain) to be a principle ofUG. Wexler and Manzini (1987) suggest that the local
domain for the binding relation is defined as the governing category, as in (6). The
relationships among the parameter values of the governing category parameter are
inclusive, and exhibit subset relationships.
(6). Governing Category Parameter y is a governing category for a iff y is the minimal category that contains a and A. has a subject, or B. has a INFL; or C. has a TNS; or D. has an indicative TNS; or E. has a root TNS. (Wexler and Manzini, 1987: 53)
Wexler and Manzini' s theory parameterizes the notions of governing category and proper
antecedent. An English reflexive has the value A of the governing category parameter,
since it refers to an NP within a minimal category which contains the anaphor and has a
subject. For example, in sentence (7), himself refers to Leo but not to Michael.
(7). Michaeli thinks that [Leoj dislikes himself *i!j].
However, Chinese reflexive ziji has the value E of the governing category parameter,
since it refers to an NP only if the NP is included in a minimal category with a root TNS.
As can be seen in (8), a Chinese translation version of (7), ziji in Chinese can have the
matrix subject Zhangsan or the embedded subject Lisi as its antecedent, indicating that
the root clause is a governing category for ziji in Chinese.
(8). Zhangsani renwei Lisij buxihuan ziji ifj· Zhangsan think Lisi dislike self "Zhangsan thinks that Lisi dislikes himself'.
266
With the five different language settings or values of the governing category parameter
forming an inclusion hierarchy1, English forms the least inclusive language, with Chinese
the most inclusive one. Wexler and Manzini (1987) later proposed the Subset Principle in
order to solve the learning problems encountered by the learners, given that the evidence
available to the learner at a certain stage of development would not be able to determine
the correct parameter setting. According to the Subset Principle, the learner initially starts
with the setting for the parameter that yields the smallest language. That is to say, the
learner starts by choosing the unmarked setting (6A), which defines the most restrictive
domain within which reflexives find their antecedents. A child learning English, for
example, would start out assuming that a reflexive must be bound within the domain of a
subject. A child learning Chinese starts out assuming the same, but revises this when he
comes across sentences where ziji is bound within a larger domain. The process of
revision is repeated until the child arrives at a grammar, which is minimally compatible
with the input data. According to Wexler and Manzini (ibid.), since L1learners always
make conservative hypotheses about what can serve as the antecedent of a reflexive,
overgeneralization does not arise. Children always adopt a more inclusive setting when
they are confronted with reflexives bound within larger domains.
Although the Subset Principle in setting governing category and proper antecedent
parameters has enlightened L1 acquisition research, it is controversial whether a Subset
Principle indeed operates in SLA (Zobl, 1988; White, 1989a, 1989b; Thomas, 1993,
1 For the other marked settings, Manzini and Wexler (1987) indicate that Italian se is subject to parameter setting (6B), and Icelandic sig to (6D). Finer and Broselow (1986: 156) cite an unspecified Russian anaphor as in (6C) (cf. Manzini and Wexler, 1987; Thomas, 1995).
267
1995; Yuan, 1994; Wakabayashi, 1996l If the governing category of reflexives in the
L1 is more inclusive than that in the L2, which is the case with Chinese or Japanese
learners of English, then how are reflexives acquired? Some studies have been concerned
with whether adult learners can reset the governing category and proper antecedent
parameters when reflexives in the L1 differ in their settings from reflexives in L2.
Finer and Broselow (1986) investigated the interpretation of English pronouns and
reflexives by six adult native speakers of Korean. Korean reflexive casin can be bound in
the root sentence as well as in the embedded sentence, like that of Japanese zibun or
Chinese ziji. Their results suggest that Korean-speaking learners of English employ an
intermediate governing category setting, which is neither the correct unmarked L2 one
nor the maximally marked L1 setting. Their informants bound 91.7% of reflexives to
local antecedents while the reflexive was inside a tensed embedded sentence. However,
the informants chose 37.5% of the reflexives bound long-distance and 58.3% of the
reflexives locally bound when confronted with infinitive sentences. In a later study, Finer
(1991) increased the participant pool by 20 Japanese and Korean-speaking learners and
claimed that L2 learners can successfully reset the proper antecedent parameter. The
results suggest that, in tensed embedded sentences, only 2% of the reflexives in the
judgments by the Japanese speakers and 5% by the Korean speakers were bound long-
distance.
2 For a critical summary ofManzini and Wexler, please see Thomas (1993; 1998) and Ying (1999), among others.
268
Hirakawa's (1990) data on the acquisition of English reflexives by 65 speakers of
Japanese are more or less consistent with those of Finer and Broselow. The informants
indicated an overall preference for local antecedents for English reflexives in tensed
embedded clauses. Thomas (1989, 1991) investigated the acquisition of English
reflexives by native speakers of Chinese and Spanish, as well as native speakers of
Japanese and Spanish. She claims that transfer ofLl parameter values cannot account for
all learners' errors in L2, since some Spanish as well as Japanese and Chinese-speaking
learners bind English reflexives long distance. Based on her experimental results,
Thomas suggests that L2 learners are constrained by UG in their construction of IL
grammars of reflexives, arguing that L2 learners can reset parameters for the target
language in spite of the lack of positive evidence in the input data.
In a later study, Thomas (1995) investigated whether L2learners of Japanese who have
acquired the long distance property of Japanese reflexive zibun also know that it must be
subject-oriented, i.e. zibun must take a subject as its antecedent. Her experiments include
58 learners of Japanese as a foreign language, 34 in a low proficiency group and 24 in a
high proficiency group. Subjects were tested with a truth-value judgment task involving
stories and pictures. The experimental sentences were both monoclausal (with binding to
subjects or objects) and biclausal (with binding to long distance subjects or long distance
objects). In Japanese, binding to subject is grammatical in both monoclausal and biclausal
sentences, whereas binding to object is ungrammatical. The results show that most of the
informants at a high proficiency level who bound zibun long-distance rejected object
antecedents, whereas the lower proficiency group failed to bind reflexives long distance.
269
Since a number of informants (35 out of 58) failed to bind the reflexive long distance, the
issue of subject-orientation could not be investigated for these learners. Thomas's (1995)
results suggest that, "at least at a high proficiency level, most L2learners of Japanese
who bind zibun long distance also reject non-subject antecedents, consistent with a key
prediction of movement in LF" (ibid., p.232). Her results indicate that the lack of
correlation between long-distance binding and subject orientation cannot serve as
evidence for learners' UG-inconsistent behavior.
In his latest article, Wakabayashi (1996) examines experimental data of SLA of English
reflexives by 40 Japanese-speaking learners. His evidence suggest that SLA is
systematic, that L2learners' interpretations of English reflexives reflect their linguistic
knowledge, and that UG is available in SLA (ibid., p.265). Wakabayashi's experimental
data suggest that SLA of the referential properties of English reflexives is achieved by
resetting parameter values gradually. More recently, Christie and Lantolf (1998) have
attempted to determine whether L2learners' grammars exhibit a set of properties that are
integral to the assumption that anaphors move in Logical Form. However, their guarded
conclusion indicates that their finding neither supports nor strongly disconfirms the claim
that UG constrains learners' knowledge of anaphors.
SLA studies so far have yielded inconclusive evidence with respect to L2 learners' access
to UG and L1 effects in L2 acquisition. In terms of the acquisition of Chinese reflexives
by non-native speakers of Chinese, few studies have been carried out. As will be seen in
the next section, the results suggested in this section are parallel to the results reported in
270
Chen's (1995) study (see section 5.3.). That is, English-speaking learners consistently
chose a local antecedent for ziji, having difficulty in acquiring the long-distance reflexive
(cf. Thomas, 1991; Chen, 1995).
8.2. Experiment and Test Sentences
Different types of sentences were designed to find out whether the CFL informants would
allow ziji to have long-distance binding, i.e. to have an antecedent outside the embedded
clause in which it occurs, like example (8).
(8). Zhangsani zhidao Lisij dui zijii!j mei you xmxm.
Zhangsan know Lisi to self no have confidence
Meanwhile, the EFL informants were tested to find out whether they were aware that
reflexives in object position have to be bound locally, i.e. they can only have short-
distance binding, as in (9), which is the English version of (8).
(9). Luciai thinks Tinaj doesn't have confidence in herself *ifj·
The complete set of test sentences concerning short-distance and long-distance reflexives
for both the EFL/CFL learners can be found in Appendix 10 and 11. In the questionnaire,
EFL/CFL informants were asked to choose one answer as the correct interpretation from
the multiple choices provided, such as example (1 0).
(1 0). Shirley thinks Tina hates herself. Here: herself= 1) Shirley
2) Tina 3) either Shirley or Tina 4) don't know
Each type has two tokens, and all sentences were presented with Chinese characters or
English for the CFL and EFL informants respectively. The informants were asked to
271
make intuitive judgments for the test sentences. Although "don't know" was provided as
a possible answer, informants were instructed to avoid choosing it unless necessary. The
informants used in the reflexive experiment were the same as the ones used for the rest of
the study.
8.3. The Judgments on reflexive binding by the EFL/CFL learners
Chart 8-1: Mean scores of AJT on reflexive binding by EFL groups
2.1 2.1 r 0
~ I
2.0 2.0 r , 0 , :::1 , , (C
, 6. , iii' 1.9 , , 1.9 Iii ,
If :::1
, , £ , , OJ
1.8 , , 1.8 5' 1- , , c.
~ ·-----· 5' , (C
..... , , 0 1.7 , 1.7
~ , , ,
8 , , ..
en 1.6 ,' 1.6 ..... ~--
, 0 ---·-c: ro Q)
:E 1.5 1.5
2 3 4 5 6 7
EFL Groups
The EFL informants' judgements on object reflexive are represented by the dotted line in
Chart 8-1. As we can see from the chart, informants in Groups 1-3 did not show steady
behaviors in rejecting the ungrammatical long-distance binding for reflexives in English,
since their judgments are quite indeterminate. As is shown clearly, more proficient
informants in Groups 4-6 were starting to reject the long-distance binding of reflexives in
English. According to the Tukey HSD Tests conducted, there is no significant difference
between Group 7, the native group, and the non-native Groups 4-6 in judging the long-
272
distance binding of reflexives in English (see Table 8-1 in Appendix 13). However, there
are significant differences found between the native group and the elementary and
intermediate Groups 1-3. Generally speaking, the EFL informants' preferences are
moving towards local binding for reflexives in English, because more than 60% of the
EFL informants preferred local binding over long-distance binding for reflexives in
English. Only a little over 30% of the EFL informants chose long-distance binding for
reflexives in English in the experiments, many of them being elementary EFL learners.
Chart 8-2: Mean scores of AJT on reflexive binding by CFL groups
1.9,-----------------,
1.8
1.7
1.6
~ 1.5
0 ~ 8 1.4 "' c: m
·----.---------~-------' ' I
I
' I ' ' I ' I ' I
' I ' I
I
:2 1.3 oi.-------.----.-----.-----J. 1 2 3 4 5
CFL Groups
r 0
~ I
1.8 r 0 ::l ur c.
1.7 iii' or ::l
!il 0"
1.6 5' c. :;· cc
1.5
1.4
In Chart 8-2, the dotted line illustrates CFL informants' judgments on object reflexive
binding. Recall that for the Chinese reflexive ziji in object position, both local and long-
distance binding are possible, whereas for English reflexives, only local binding is
possible (see Section 8.0). As we can see from the chart, CFL informants in Groups 1-4
showed a general preference for local binding of reflexives ziji, consistently rejecting the
correct long-distance binding of reflexive ziji. Even when the informants' proficiency in
273
Chinese improved, as in the case of Groups 3 and 4, there was little improvement in
judging the long-distance binding of reflexive ziji as acceptable. Generally speaking, all
the CFL groups, except the native Group 5, rejected the long-distance binding of ziji
incorrectly. For the native Group 5, 6 out of 16 informants (37.5%) rejected the long
distance binding of reflexive ziji. According to the Tukey HSD Tests conducted, there is
no significant difference found between native Group 5 and the rest of the non-native
groups in judging the reflexive binding of ziji (see Tables 8-2 in Appendix 13).
8.4. The Analysis of Results
As we can see from the previous section (8.3), EFL informants showed a general
preference for local binding over long-distance binding for reflexives in English.
Previous experiments reported in the SLA literature (cf. Finer & Broselow, 1986; Finer,
1991; Hirakawa, 1990; Thomas, 1990, 1991, 1995) indicate that learners of English as a
foreign or second language consistently identify reflexives in English finite sentences
with local antecedents. L2 learners prefer local binding regardless of whether only local
binding or both local and long-distance bindings are demonstrated in the learners' L1s.
The present study is consistent with the results reported in these researches. According to
the governing category parameter (see 8.0 (6A-E)), proposed by Wexler and Manzini
(1987), English reflexives are subject to setting (5E) of the parameter, the most restrictive
domain in which they find their antecedents. L2learners ofEnglish, as in the case of
Chinese, Japanese and Korean, do not have positive evidence in the input to indicate that
long-distance binding for reflexives is not allowed in English, because their L1s have a
setting of ( 6E) and allow long-distance binding for the reflexives. However, there are two
274
types of reflexives in Chinese, one is the bare reflexive ziji and the other the phrasal
reflexives like pronoun + ziji (see also 5.1.1.). According to Sung and Cole's (1991)
theory of successive cyclic head movement (see also 5 .1.5. ), the binding of phrasal
reflexives has to be strictly local and bare reflexives can be bound both locally and long
distance. As we have seen in previous chapters (see 5.0), reflexives in English are
phrasal, and cannot undergo the successive cyclic head movement, which is possible for
the bare reflexive. When Chinese learners of English are exposed to English reflexives in
the input data of English, they recognize them as phrasal reflexives. As a consequence,
they regard them as having the same properties as phrasal reflexives in their L1. Based on
this correct assumption, Chinese learners bind English reflexives to local antecedents
only, rejecting sentences that allow long-distance binding for English reflexives.
Moreover, since the bare reflexives and phrasal reflexives contrast in surface forms by
demonstrating a distinct mental representation in the learners' L1s, L2 learners recognize
English reflexives as phrasal reflexives by the surface form. Since no bare reflexive
occurs as an independent word in the EFL learners' input data, the successive cyclic head
movement for bare reflexives is inactive in the learners' IL grammars of English. The
local binding of reflexives becomes a default for reflexives in English. This is what this
study has found and it explains why EFL learners generally rejected the long-distance
binding for English reflexives in our study.
For the CFL learners, however, the story is different. As we can see from Chart 8-2, CFL
informants in Groups 1-4 generally preferred the local binding of ziji, consistently
rejecting the correct long-distance binding of reflexive ziji. Even when the learners'
proficiency in Chinese improved, there was little improvement in accuracy in their
275
judgments. The findings reported here reveal that CFL learners accepted only local
binding, and rejected long-distance binding for ziji in their IL grammars of Chinese. That
is to say, they bound ziji locally for the test sentences that were supposed to allow long
distance antecedents as well as the local ones. Questions that arise are whether the lack of
long-distance binding for ziji really means that the CFL learners had not yet acquired the
knowledge in question. Since CFL informants had difficulty accepting the long-distance
domain for ziji, it seems that they were unaware that Chinese allows ziji to be long
distance bound. However, since there is positive evidence in the target language which
should remind the learners of the fact that ziji is a bare reflexive, this should in tum
inform them that ziji can be long-distance bound in Chinese. Given that CFL informants
have positive evidence in their input data that the local and long-distance binding are
possible in Chinese, the question arises as to why they only accept local binding and
reject the long-distance domain for ziji. One possible explanation may be that CFL
informants failed to notice that positive evidence existed in the input data, which means
they failed to realize the grammatical significance of the relevant linguistic data for
restructuring the learners' IL grammars. Instead of interpreting the data by problem
solving abilities, CFL informants just use what is known to them, i.e. their L1 knowledge,
to analyze the data. Thus, choosing the local antecedent for ziji as the obvious taken-for
granted answer.
However, there might be another answer to this question. As D-D.Chen (1995: 11)
suggested, it may not be the case that CFL informants had difficulty acquiring knowledge
oflong-distance binding, but that "ziji might actually require a local antecedent for a
276
default interpretation and this forced the learners to favor local binding, even though ziji
can be grammatically bound long-distance". In my study, all the non-native speaker
groups consistently rejected the long-distance binding, though the native control group is
more determinate in accepting long-distance binding as correct. It should be noticed that
there is no significant difference found between the native speaker group and the rest of
the CFL learner groups in judging the reflexive binding of ziji. CFL informants, it is
suggested, would choose the local binding of ziji as a default interpretation unless a given
context forces them to choose an antecedent beyond its local domain. The native control
group, however, does not need this given context in order to reach the conclusion. Since
this study did not take pragmatic factors into consideration, it is suggested that further
studies should not ignore the pragmatic implications of this explanation3•
8.5. Conclusion
Our findings indicate that EFL informants showed a general preference for local binding
over long-distance binding for reflexives in English. In the acquisition of reflexives in
English, EFL informants were not influenced by the long-distance binding in their Ll,
Chinese. Since there are two types of reflexives (i.e. bare and phrasal) in the learners' Ll
Chinese, EFL learners were able to recognize English reflexives as phrasal by the surface
form and rejected long-distance binding of reflexives in English as unacceptable. The
local binding of reflexives thus becomes default for reflexives in English. All the CFL
informants, however, bound ziji locally for the test sentences that were supposed to allow
3 Some linguists, such as Y.Huang (1994), claim that only pragmatic principles should be involved to characterize the various properties of Chinese binding; Xu (1993, 1994) suggests that semantic, discourse factors and syntactic factors are all involved in the choice of an antecedent ziji.
277
long-distance binding as well as the local ones, even though there is relevant positive
evidence in the input. It is suggested that CFL informants would choose the local binding
of ziji as a default interpretation unless a given context forces them to choose an
antecedent beyond its local domain. Meanwhile, it is also suggested that the difficulty
CFL learners have in acquiring knowledge oflong-distance binding of the Chinese
reflexive ziji, is related to their interpretation of the relevant language data they had
exposed for the restructuring of their IL grammars for the target language.
278
Chapter 9 Summary and Conclusion
9.1 Summary and conclusion
This study has shown that English and Chinese are different languages that have different
settings for the so-called pro-drop parameter. In terms of the linguistic features
investigated in this study, including the null/non-null subject, null/non-null object,
base/non-base-generated topics, local/long-distance reflexives, it is suggested that there
are no single directional difficulties in the second language acquisition of Chinese and
English respectively. The findings of this study suggest that some language features are
more difficult for EFL and CFL informants to acquire, whereas other features are more
learnable in terms of the difficulty level. Some linguistic features are easier to "unlearn",
while others are more difficult to "acquire". Many other factors also play a part in
forming the directional difficulties in the second language acquisition of Chinese and
English. These crucial factors are: the availability of positive or negative evidence in the
learners' input data, the learners' ability to process the relevant information in the
available input for the restructuring of their IL grammar of the TL, and finally the
acquisition process between the learners' Ll and the developmental process of the TL,
i.e. Chinese or English.
In terms of the languages features investigated in this study, English speakers learning
Chinese (CFL learners) and Chinese speakers learning English (EFL learners) at various
proficiency levels showed different degrees of difficulty in acquiring or unlearning those
features. In terms of null and non-null subjects, neither EFL nor CFL informants
279
experience much difficulty in unlearning or acquiring null subjects in English or Chinese,
respectively. Comparing both types of informants, there is no directional difficulty
involved in this particular language feature in the second language acquisition of Chinese
and English. In terms of null and non-null objects, however, EFL informants generally
fail to reject the ungrammatical experimental sentences of null-objects during their
acquisition of English, whereas CFL informants have little difficulty in acquiring null
objects in Chinese. In this case, the directional difficulty is from Chinese to English,
since Chinese is more inclusive than English in this aspect of grammar.
In terms of the acquisition and undo process ofbase-generated topics by EFLand CFL
informants, only intermediate and advanced EFL informants are more determinate in
rejecting the incorrect experimental sentences of base-generated topics in English,
whereas CFL informants generally accept non-base-generated topics in their acquisition
of Chinese. Moreover, CFL and EFL informants do not experience much difficulty
during their process of acquisition and rejection of nominal expletives in Chinese and
English respectively. For this aspect of grammar, there is no evidence that directional
difficulties play a part in second language acquisition of Chinese and English.
9.2. Limitations and suggestions for further studies
Like many other experimental studies, the present research cannot escape from the
following limitations. First, this study was a cross-sectional, rather than a longitudinal
study, so it is not possible to analyze the linguistic features used by the same learners at
280
different developmental stages. However, the use of cross-sectional grammaticality
judgment tasks offers the advantage of requiring less time and being cost-effective, and
thus was chosen as the most suitable way to L2 data. Secondly, it should be pointed out
that some informants in this study had acquired English or Chinese in the classroom
situations only, while others had acquired it both in classrooms and in natural situations.
Though it is arguable whether the use of the classroom setting as the primary source of
SLA data collection should be viewed as a limitation, efforts should be made to obtain
data in more natural, non-classroom settings in the future ( cf. Dittmar, 1992; Perdue,
1993; Klein & Perdue, 1992). Most of the CFL informants had studied Chinese in
different institutions, and had worked or lived in China or Taiwan for some time.
However, apart from our advanced learners, most of the EFL informants had not studied
English in English-speaking countries. This may have had some impact on the
informants' performance of syntactic competence. Thirdly, because of the availability of
resources, I only found and recruited 70 CFL informants to carry out the grammaticality
judgment tasks. It is suggested that the number of English speaking learners of Chinese
should be expanded in order to provide a more balanced comparison between the EFL
and CFL learners' performance in regard to the linguistic features under investigation.
It is suggested that further studies should be conducted with more test data, and designed
with different types of tasks. For instance, some production task might be included, in
which informants would be required to write a free composition or a passage on a given
topic. Alternatively, CFL or EFL learners' free conversation on a given topic might be
designed. Data collected in this way would be a reliable reflection of the ability ofL2
learners, and as a result, findings would be more accurate and useful.
281
A further limitation was related to the theoretical model this study adopted. As discussed
in the introductory chapter, the theoretical orientation of this study is the widely
discussed principles and parameters theory of Universal Grammar as detailed in
Chomsky's (1981a, 1982) Government and Binding Theory. Many applied linguists have
undertaken empirical research within the Principles and Parameters framework with
particular attention paid to issues relating to the acquisition and unlearning of the Null
Subject Parameter in adult second language acquisition. This study is an attempt to
investigate the directional difficulties concerning the linguistic features of null/non-null
subjects, null/non-null objects, base/non-base-generated topics, local/long-distance
reflexives, etc, during the second language acquisition of Chinese and English,
respectively. The results in this study indicate that there is no single directional difficulty
involved during the acquisition process for the CFL or EFL learners respectively. There
are other factors that play a part in forming the difficulties. Except for the factors
suggested in this study, further studies are needed to find out the crucial factors at work.
There is a need for more SLA research relating to directional difficulty as well as the
phenomena of null subject parameter to be carried out within the principles and
parameters approach. Ever since the introduction of Government and Binding (GB)
theory, many research reports have appeared related to the phenomena of null subject
languages in the past two decades. Nevertheless, the relevant L2 research has only
scratched the surface of the many acquisition issues raised in this area.
Moreover, with the emergence of the Minimalist Program (Chomsky, 1993, 1995a,
1995b ), recent developments within the theory of principles and parameters have changed
282
the way the grammar is viewed. Lexicon, LF (Logical Form) and PF (Phonological
Form) have replaced the previous three levels of syntactic representation (D-structure, S
structure and LF (Logical Form)). There has been a shift in linguistic theory from the
principles and parameters approach of GB Theory toward the current formulations of the
Minimalist Program. Although this new development may cause a significant paradigm
shift within the field of formal linguistics, it remains to be seen if this new research
orientation gains widespread acceptance among those interested in data-driven research
ofLl and L2 acquisition. The present study is generally based on the earlier dominant
paradigm of GB models of Principles and Parameters approach and this may be a
shortcoming in this regard. However, research paradigm shifts are inevitable and should
always be anticipated. It is far from right to assume that each new paradigm represents
the final state of the theory, and that all previous theories must therefore be rejected. It is
argued that the theoretical framework of Principles and Parameters model, along with its
various sub-theories, is suitable as a foundation for inquiry in the SLA context and in this
study. It is argued that the syntactic parameterization of parameter (re-)setting provides
an accessible and testable method for the collection and analysis of SLA data. Further
research is needed to test the applicability of the Minimalist Program in relation to the
linguistic features under investigation in this study.
It is a fact that Chomsky did not think much about language teaching when he began to
establish the framework of his generative grammar, and there has been a belief in the
field of language learning that the Chomsky's linguistic theory has no relevance to the
teaching ofL2. However, any linguistic theory including generative linguistics will,
directly or indirectly, shed light on problems in Ll and L2 learning. But how to recognize
283
the applicability of the Principles and Parameters theory is still a problem which requires
further understanding. If one agrees with the view that adult L2 learning resembles
processes of L2 learning in some way, it might be suggested that L2 learners should
merely be exposed to large samples of the target language so that the target language can
be acquired unconsciously. With substantial and robust L2 linguistic input (i.e. the direct
and indirect positive evidence), it seems that the innate language faculty rather than
explicit grammar teaching, should be able to do most of the acquisitional work (see
Schwatz, 1993; Trahey, 1996; for details). However, to what extent can current
grammatical theory be brought into L2 classroom settings if the decision has been made
to consciously teach grammatical properties and rules for L2 learners? It seems that this
applicability issue has been debated over a number of years and still remains, for the most
part, unresolved. White (1987a, 1989a, 1990b, 1991a) is one of the researchers who has
written extensively on the topic of the relationship ofUG theory and SLA. White (1989b:
182) agrees that the overall goal ofUG-based SLA research is to come to an
understanding ofhow second languages are learned, but she added "even when such
understanding is attained, this does not necessarily offer clear insights into the best way
to teach languages". But White (ibid.: 182) also admits that "it is possible that specific
grammar teaching and correction in the language classroom can sometimes fill a gap, not
covered by positive evidence form the L2". Nevertheless, whether highlighting
potentially problematic L2 surface-structure properties can result in lasting acquisitional
effects remains open to question, especially if it is argued that the linguistic properties
involved are consequences of deep-structure syntactic principles and parameters. It is
suggested that further research is needed to address the theory-to-practice issue, if the
284
recent formulations of linguistic theory are to exert profound implications on L2
classrooms. However, as White (1985b: 58) has indicated in her influential UG-oriented
study, "it is worth investigating whether UG plays a role in L2 acquisition and, if so,
whether the Ll affects the way that UG operates in L2". In a word, the factor of
directional difficulty for Chinese learners of English and English learners of Chinese is a
very interesting and important area in the learning and teaching of Chinese and English as
the L2. The findings of this study indicate that the relationship between UG and L2 study
will continue to be worthy of investigation.
REFERENCES
Adjemian, C. (1976). On the nature ofinterlanguage systems. Language Learning, 26: 297-320.
285
Aikawa, T. (1993). Reflexivity in Japanese and LF Analysis of Zibun Binding. Doctoral Dissertation. Ohio State University.
Aoun, J. & Sportiche, D. (1983). "On the formal theory of government", The Linguistic Review, 3, 211-235.
Atkinson, M. (1992).Children 's Syntax: An Introduction to Principles and Parameters Theory. London: Basil Blackwell.
Bailey, N., Madden, C. & K.rashen, S. (1974). Is there a 11natural sequence11 in adult second language learning? Language Learning, 24: 235-243.
Baker, C. L. (1979). "Syntactic theory and the projection problem". Linguistic Inquiry, 10, 533-581.
Battistella, E. (1989). Chinese reflexivization: a movement to INFL approach. Linguistics 27: 987-1012.
Battistella, E. & Xu, Y. (1990). Remarks on the reflexives in Chinese. Linguistics 28: 205-240.
Bates, E., Thal, D. & Janowsky, J. (1992). Early language development and its neural correlates. In Handbook ofNeuropyschology, Vol.7: Child Neuropsychology, ( eds.). Rapin, I. & Segalowitz, S. Elsevier.
Beck, M.-L. & Eubank, L. (1991). Acquisition theory and experimental design: A critique ofTomasello and Herron. Studies in Second Language Acquisition, 13(1): 73-76.
Bennett, S. (1994). Interpretation of English reflexives by adolescent speakers of SerboCroatian. Second Language Research, 10: 125-56.
Bennett, S. & Progovac, L. (1995). Evidence of transfer and Universal Grammar in second language acquisition of reflexive binding. McGill Working Papers in Linguistics, 9: 77-101.
Bennett, S. & Progovac, L. (1998). Morphological status of reflexives in second language acquisition. In Flynn, S., Martohardjono, G. & O'Neil, W. (eds.). The generative Study of Second Language Acquisition. Hillsdale, NJ: Lawrence Erlbaum, 187-214.
Berent, G. (1994). The subset principle in first and second language acquisition. fu E. Tarone et al ( eds.)., Research Methodology in Second Language Acquisition, 17-39, Hillsdale, NJ: Erlbaum.
Berwick, R. (1985). The Acquisit~on of Syntactic Knowledge. Cambridge: Mass: MIT Press.
286
Birdsong, D. (1989). Metalinguistic Performance and Interlanguistic Competence. New York: Springer.
Birdsong, D. (1991). On the notion of"critical period" in UG/L2 theory: A response to Flynn and Manuel. fu L. Eubank ( ed.), Point Counterpoint: Universal Grammar in the Second Language. Amsterdam: J. Benjamins.
Birdsong, D. (1992). Ultimate attainment in second language acquisition, Language, 68: 706-755.
Bley-Vroman, R. (1986). Hypothesis testing in second-language acquisition theory. Language Learning, 36: 353-376.
Bley-Vroman, R. (1989). ''What is the logical problem of foreign language learning?" fu Gass, M.S. & Schachter, J. (eds.). Linguistic Perspectives on Second Language Acquisition. Cambridge: Cambridge University Press.
Bley-Vroman, R. (1996). What we have to explain in foreign language learning. fu Commentary, Behavioral and Brain Sciences, 19: 716-717.
Bley-Vroman, R., Felix, S. & Ioup, G. (1988). The accessibility ofUniversal Grammar in adult language learning. Second Language Research, 4: 1-32.
Bloom,P .(1990).Subjcetless sentences in child language. Linguistic Inquiry,21 :491-504.
Bloomfield, L. (1933). Language. New York: Holt, Rinehart & Winston.
Boe, D. R. (1996). Parameter Resetting in Adult Second Language Acquisition. Doctoral Dissertation. fudiana University.
Borer, H. (1983). Parametric Syntax: Case Studies in Semitic and Romance Languages. Dordrecht: Foris.
Borer, H. (1989). Anaphoric AGR. fu Jaeggli, 0. & Safir, K. (eds.). The Null Subject Parameter. Dordrecht: Reidel.
Borer, H. (1996). Access to Universal Grammar: The real issue. fu Commentary, Behavioral and Brain Sciences, 19: 718-720.
287
Borsley, R. D. (1991). Syntactic Theory: A Unified Approach. London: Edward Arnold.
Bowerman, M. (1973). Early Syntactic Development: A Cross-Linguistic Study, with Special Reference to Finnish. Cambridge: Cambridge University Press.
Bowerman, M. (1983). How do children avoid constructing an overly general grammar in the absence of feedback about what is not a sentence? Papers and Reports on Child Language Development, 22: 23-35. Stanford University.
Bowerman, M. (1985). What shapes children's grammars? In Slobin, D. (ed.). The CrossLinguistic Study of Language Acquisition (Vol.2).Erlbaum, Hillsdale, New Jersey.
Braine, M. (1971). On two types of models of the internalization of grammars. In Slobin, D. I. (ed.), The Ontogenesis of Grammar. New York: Academic Press.
Carroll, S. (1989). Language acquisition studies and a feasible theory of grammar. Canadian Journal of Linguistics, 34: 399-418.
Carroll, S. (1996). Parameter-setting in second language acquisition- explains and explanandum. In Commentary, Behavioral and Brain Sciences, 19: 720-721.
Chao, Y. R. (1968). A Grammar of Spoken Chinese. Berkeley & Los Angeles: University of California Press.
Chaudron, C. (1983). Research on metalinguistic judgments: a review of theory, methods and results. Language Learning 33, 343-77.
Chen, D. -D. (1995). The L2 acquisition of the Chinese reflexive ziji. Penn Worldng Papers in Linguistics, 2: 37-52.
Chien, Y. -C. (1992). Theoretical implications of the principles and parameters model for language acquisition in Chinese. In Chen, H. -C. & Tzeng, O.J.L. (eds.). Language Processing in Chinese. Amsterdam: North-Holland.
Chien, Y-C. & Wexler, K. (1987). A comparison between Chinese-speaking and Englishspeaking children's acquisition of reflexives and pronouns. Paper presented at the 12th Annual Boston University Conference on Language Development, Boston, Mass.
Chien, Y. -C. Wexler, K. & Chang, H. -W. (1993). Children's development oflongdistance binding in Chinese. Journal of East Asian Linguistics, 2: 229-259.
Chiu, B. (1995). An object clitic projection in mandarin Chinese. Journal of East Asian Linguistics 4: 77-117.
Chomsky, N. (1965). Aspects ofthe Theory of Syntax. Cambridge, Mass.: MIT Press.
Chomsky, N. (1975). Reflections on Language. New York: Random House.
Chomsky, N. (1977). On wh-movement. In P.Culicover, T.Wasow, & A. Akmajian (eds.). Formal Syntax, 71-132. New York: Academic Press.
Chomsky, N. (1980a). "On binding". In Linguistic Inquiry, 11, 1-46.
Chomsky, N. (1980b). Rules and Representations. New York: Columbia University Press.
Chomsky, N. (1981a). Lectures on Government and Binding. Dordrecht: Foris.
Chomsky, N. (1981b). Principles and parameters in syntactic theory. In Explanation in Linguistics: The Logical Problem of Language Acquisitions. ( eds.). by N. Hornstein & D. Lightfoot. London: Longman.
288
Chomsky, N. (1981c). Markedness and core grammar. In A. Belletti et al (eds.) Theory of Markedness in Generative Grammar: Proceedings ofthe 1979 Glow Conference. Pisa, Scuola Normale Superiore.
Chomsky, N. (1982). Some Concepts and Consequences ofthe Theory of Government and Binding. Cambridge, Mass.: MIT Press.
Chomsky, N. (1983). Language Universals and Linguistic Typology. Oxford: Basil Blackwell.
Chomsky, N. (1986a). Knowledge of Language: Its Nature, Origin and Use. New York: Praeger Publishers.
Chomsky, N. (1986b). Barriers. Cambridge, Mass.: MIT Press.
Chomsky, N. (1988). Language and Problems of Knowledge. The Mangua Lectures. Cambridge, Mass. :MIT Press.
Chomsky, N. (1991). "Some notes on economy of derivation and representation". In Freidin, R. ( ed.). Principles and Parameters in Comparative Grammar. Cambridge, Mass.: MIT Press.
Chomsky, N. (1992). A minimalist program for linguistic theory. MIT Working Papers in Linguistics, Cambridge, Mass.
Chomsky, N. (1993). A minimalist program for linguistic theory. InK. Hale & S. J. Keyser (eds.), The View from Building 20: Essays in Linguistics in Honor of Sylvain Bromberger, 1-52, Cambridge, Mass.: MIT Press.
289
Chomsky, N. (1995a). Bare phrase structure. In G. Webelhuth (Ed.), Government and Binding Theory and Minimalist Program: Principles and Parameters in Syntactic Theory, 383-440, Oxford: Blackwell.
Chomsky, N. (1995b ). The Minimalist Program. Cambridge, Mass.: MIT Press.
Chomsky, N. & H. Lasnik. (1977). Filters and controls. Linguistic Inquiry, 8, 425-508.
Christie, K. (1992). Universal Grammar in the second language: an experimental study of the cross-linguistic properties of reflexives in English, Chinese and Spanish. Doctoral Dissertation, University of Delaware, Newark.
Christie, K., & J.P. Lantolf. (1992). The ontological status oflearner grammaticality judgments in UG approaches to L2 acquisition. Rassegna Italiana de Linguistica Applicata 24: 31-52.
Christie, K. & J.P. Lantolf. (1998). Bind me up bind me down: reflexives in L2. In Flynn, S., Martohardjono, G., & O'Neil, W. (eds.). The Generative Study of Second Language Acquisition. Mahwah, NJ: Lawrence Erlbaum.
Clahsen, H. (1988). Parameterized grammatical theory and language acquisition: a study of acquisition of verb placement and inflection by children and adults. In Flynn, S. & O'Neil, W. (eds.), Linguistic Theory in Second Language Acquisition. K.luwer: Dordrecht.
Clahsen, H. (1991a). Child Language and Developmental Dysphasia: linguistic studies ofthe acquisition of German. Amsterdam: J. Benjamins.
Clahsen, H. (1991b). Constraints on parameter setting: A grammatical analysis of some acquisition stages in German child language. Language Acquisition, 1: 361-391.
Clahsen, H, & Muysken, P. (1986). The Availability of Universal Grammar to adult and children learners: a study ofthe acquisition of German word order, Second Language Research, 2/2: 93-119.
Clahsen, H. & Muysken, P. (1989). The UG paradox in L2 acquisition. Second Language Acquisition, 5: 1-29.
Cole, P. (1987). Null object in Universal Grammar. Linguistic Inquiry, 18/4: 597-612.
Cole, P. Hermon, G. & Sung, L.-M. (1990). Principles and parameters oflong-distance reflexives. Linguistic Inquiry, 21: 1-22.
Cole, P. & Sung, L. -M. (1994). Head movement and long-distance reflexives. Linguistic Inquiry, 25: 355-406.
Cook, V. J. (1988). Chomsky's Universal Grammar: An Introduction. Oxford: Basil Blackwell.
290
Cook, V. J. (1991). The poverty-of-the-stimulus argument and multi-competence. Second Language Research, 7 (2), 103-117.
Cook,V.J. (1992). Evidence for multi-competence. Language Learning, 42(4), 557-591.
Cook,V.J.(1993).Linguistics and Second Language Acquisition. Basingstoke: Macmillan.
Cook, V. J. (1994). The metaphor of access to Universal Grammar in L2 learning. In Ellis, N.C. (ed.), Implicit and Explicit Learning of Language. Academic Press.
Cook, V. & Newson, M.(1996). Chomsky's Universal Grammar. Oxford: Blackwell.
Cooper, M. (1999). Introduction to Experimental Design and Analysis. Lecture notes. Sydney: The University ofN ew South Wales.
Cowan, R. & Y. Hatasa. (1994). Investigating the validity and reliability of native speaker and second language learner judgments about sentences. In E. Tarone et al ( ed.): Research Methodology in Second Language Acquisition. Hillsdale, NJ: Lawrence Erlbaum Associates, 287-302.
Cowart, W. (1997). Experimental Syntax: Applying Objective Methods to Sentence Judgments. Thousand Oaks, California: Sage Publications.
Cowper, E. A. (1992). A Concise Introduction to Syntactic Theory: The GovernmentBinding Approach. Chicago: University of Chicago Press.
Culicover, P. W. (1997). Principles and Parameters: An introduction to syntactic theory. Oxford University Press.
Dalrymple, M. (1993). The Syntax of Anaphoric Binding. CSLI Lecture Notes Number 36. Center for the Study of Language and Information.
Davies, A. (1984). "Introduction". Interlanguage, (eds.)., Davies, A., Criper, C. & Howatt, A. P. R., ix-xv. Edinburgh: Edinburgh University Press.
Davies, W. (1996). Morphological uniformity and the null subject parameter in adult SLA. Studies in Second Language Acquisition, 18: 475-493.
Davies, W. D. & T.I. Kaplan. (1998). Native speaker vs. L2 learner grammaticality judgments. Applied Linguistics, 19/2: 183-203.
Dittmar, N. (1992). Grammaticalization in second language acquisition: Introduction.
Studies in Second Language Acquisition, 14: 249-257.
Downey-Vanover, J. (1994). Adult Access to Universal Grammar in Second Language Acquisition. Doctoral Dissertation, University of Maryland, College Park, MD.
291
Dulay, H., & Burt, M. K. (1974). A new perspective on the creative construction process in child second language acquisition, Language Learning, 24: 253-278.
Dulay, H., Burt, M.K., & Krashen, (1982). Language Two. Oxford: Oxford University Press.
DuPlessis, J., Solin, D., Travis, L., & White, L. (1987). UG or not UG, that is the question: A reply to Clahsen and Muysken. Second Language Research, 3: 56-75.
Eckman, F. R. (1977). Markedness and the contrastive analysis hypothesis. Language Learning. 27: 315-330.
Eckman, F. R., & Highland, D., & Lee, P. W., & Mileham, J., & Weber, R.R. (eds.). (1995). Second Language Acquisition: Theory and Pedagogy. Mahwah, N.J: Lawrence Erlbaum Associates.
Eckman, F. R. (1994). Local and long-distance anaphora in second-language acquisition. In E. Tarone et al ( eds.). Research Methodology in Second Language Acquisition. Hillsdale, NJ: Lawrence Erlbaum, 207-25.
Ellis, R. (1990). Grammaticality judgments and learner variability in H. Burmeister and P.L. Rounds (ed.): Variability in Second Language Acquisition: Proceedings of the Tenth Meetings of the Second Language Research Forum. Uegene, OR: University of Oregon, Department of Linguistics, 25-60.
Ellis, R. (1991). Grammaticality judgments and second language acquisition. Studies in Second Language Acquisition, 13: 161-186.
Emonds, J. (1970). Root and Structure-Preserving Transformations. Ph.D. Thesis. MIT.
Emonds, J. (1976). A Transformational Approach to English Syntax. New York: Academic Press.
Epstein, S. D., Flynn, S, & Martohardjono, G. (1996). Second Language Acquisition: Theoretical and Experimental Issues in Contemporary Research, Behavioral and
Brain Sciences, 19: 677-758.
Eubank, L. (1989). Parameters in L2 Learning: Flynn revisited. Second Language Research, 5(1): 43-73.
Eubank, L. (1994). Optionality and the initial state in L2 acquisition. InT. Hoekstra & B.
292
D. Schwartz (eds.), Language Acquisition Studies in Generative Grammar: Papers in hour of Kenneth Wexler from the 1991 GLOW workshops. Amsterdam: J. Benjamins.
Eubank, L. (1996). Methodological problems with Epstein, Flynn, and Martohardjono's research. In Commentary, Behavioral and Brain Sciences, 19: 724-725.
Felix, S. W. (1985). More evidence on competing cognitive systems. Second Language Research, 1: 47-72.
Felix, S. W. (1987). Cognition and Language Growth. Dordrecht: Foris.
Felix, S. W. (1988). UG-generated lrnowledge in adult second language acquisition. In Flynn, S. & O'Neil, W. (eds.), Linguistic Theory in Second Language Acquisition. Dordrecht: Kluwer.
Finer, D. (1989). Binding parameters in second language acquisition. Paper presented at the Second Language Research Forum, University of California at Los Angeles.
Finer, D. (1991). Binding parameters in second language acquisition. In Eubank, L. (ed.), Point-Counterpoint: Universal Grammar in the Second Language. Amsterdam: J. Benjamins.
Finer, D., & Broselow, E. (1986). Second language acquisition of reflexive-binding. In Proceedings ofNELS 16, 154-168, Amherst: University ofMassachusetts, Graduate Linguistics Students Association.
Flynn, S. (1983). Differences between first and second language acquisition: setting the parameters of universal grammar. In D. Rogers & et al (eds.). Acquisition of Symbolic Skills. New York: Plenum Press.
Flynn, S. (1984). A universal in L2 acquisition based on a PBD typology. In Eckman, E., Bell, L. & Nelson, D. (eds.)., Universals of second language acquisition, 75-97, Rowley, Mass.: Newbury House.
Flynn, S. (1987). A Parameter-Setting model of L2 Acquisition: Experimental Studies in Anaphora. Dordrecht: Reidel.
Flynn, S. (1988). Nature of development in L2 and implications for theories of language acquisition in general. In Flynn, S. & O'Neil, W. (eds.). Linguistic Theory in Second Language Acquisition. Dordrecht: Kluwer.
Flynn, S. (1989). The Role ofhead-initial/head-final parameter in the acquisition of English relative clauses by adult Spanish and Japanese speakers. In Gass, S.M. & Schachter, J. (eds.). Linguistic Perspectives on Second Language Acquisition. Cambridge: Cambridge University Press.
Flynn, S. (1991). Government-binding: parameter setting in second language acquisition. In Crosscurrents in Second Language Acquisition and Linguistic Theories. (eds.) Ferguson, C. & Huebner, T. Amsterdam: J. Benjamins.
Flynn, S. (1993). Interactions between L2 acquisition and linguistic theory, In Confluence: Linguistics, L2 Acquisition and Speech Pathology, ( ed.), Eckman, F. Amsterdam: J. Benjamins.
Flynn, S. (1996). A parameter-setting approach to second language acquisition. In Ritchie, W. & Bhatia, T. (eds.). Handbook of Second Language Acquisition. New York: Academic Press.
Flynn, S. & Manuel, S. (1991).Age-dependent effects in language acquisition: an evaluation of "critical period" hypotheses. In L. Eubank ( ed.), Point Counterpoint: Universal Grammar in the Second Language, 117-145, Amsterdam: J. Benjamins.
Flynn, S. & Martohardjono, G. (1991). How critical is age in adult second language acquisition? Paper presented at the annual meeting of the American Association for the Advancement of Science, Washington, D. C.
Flynn, S. & Martohardjono, G. (1992). Mapping from the initial state to the final state: The separation of universal principles and language-specific properties. Paper
Presented at the Conference on Syntactic Theory and First Language Acquisition: Crosslinguistic Perspectives, Cornell University.
293
Flynn, S. & Martohardjono, G. (1993). Towards theory-driven language pedagogy. Paper presented at the 22nd annual Linguistics Symposium of the University of Wisconsin-Milwaukee.
Flynn, S. & Martohardjono, G. (1994). Mapping from the initial state to the final state. The separation of universal principles and language specific properties. In Lust, B., Whitman, J. & Komflit, J. ( eds.)., Syntactic theory and first language acquisition: cross linguistic perspectives: Vol.1 Phrase structure. Hillsdale, NJ: Erlbaum.
Flynn, S. & O'Neil, W. (1988). (ed.) Linguistic Theory in Second Language Acquisition. Dordrecht: Kluwer.
Fodor, J. & Crain, S. (1987). Simplicity and generality of rules in language acquisition. In Mac Whinney, B. ( ed.)., Mechanisms of Language Acquisition. Hillsdale, New Jersey: Erlbaum.
Frazier, L. (1985). Syntactic complexity. In D. R. Dowty, L. Karttunen, & A.M. Zwicky (eds.), Natural Language Parsing: Psychological, Computational, and Theoretical
Perspectives. Cambridge: Cambridge University Press.
Frazier, L. & Rayner, K. (1988). Parameterizing the language processing system: leftversus right-branching within and across languages. fu J. A. Hawkins (ed.), Explaining Language Universals. Oxford: Basil Blackwell.
Fries, C. (1945). Teaching and Learning English as a Foreign Language. Ann Arbor: University of Michigan Press.
294
Gao, Z.-M., Huang, C.-R. & Wei, W.-J. (1994). Argument structure and empty expletive: a study of event-evaluative predicates in Chinese. Paper presented at ICCL 3, HK..
Gass, S.M. (1994). The reliability of second language grammaticality judgments. fu E.Tarone et al. (eds.): Research Methodology in Second Language Acquisition, 303-322, Hillsdale, NJ: Lawrence Erlbaum Associates.
Gass, S.M. & Lakshmanan, U. (1991). Accounting for interlanguage subject pronouns. Second Language Research, 7: 181-203.
Gass, S.M. & Selinker, L. (1983). Language Transfer in Language Learning. Rowley, Mass.: Newbury House.
Gass, S.M. & Selinker, L. (1994). Second Language Acquisition: An Introductory Course. Hillsdale, N.J.: Lawrence Erlbaum Associates.
Gliedman, J. (1985). An interview with Noam Chomsky. fu V. Clark, P.A. Eschholz & A. F. Rosa (eds.). Language: Introductory Readings. NewYork: StMartin's Press.
Goss, N.Z. Ying-Hua, & J.P. Lantolf. (1994). Two heads may be better than one: Mental activity in second language grammaticality judgments. fu E. Tarone et al ( eds.): Research Methodology in Second Language Acquisition, 263-286, Hillsdale, NJ: Lawrence Erlbaum Associates.
Green, C. F. (1996). The origins and effects of topic-prominence in Chinese and English interlanguage. IRAL, XXXN/2, May 1996, 119-133.
Gregg, K. R. (1996). The logical and developmental problems of second language acquisition. fu Ritchie, W. C. & Bhatia, T. K. ( eds.)., Handbook of Second Language Acquisition. San Diego: Academic Press.
Gropen, J., Pinker, S., Hollander, M., Goldberg, R. & Wilson, R. (1989). The learnability and acquisition of the dative alternation. Language, 65: 203-257.
Haegeman, L. (1990). Understood subjects in English diaries. Multilingua, 9:157-199.
Haegeman, L.(1994).lntroduction to Government and Binding Theory. Oxford:Blackwell.
295
Hatch, E. M. & Lazaraton, A.(1991). The Research Manual. NY: Newbury House.
Hawkins, J. A. (1990). A parsing theory of word order universals. Linguistic Inquiry, 21: 223-261.
Henry, A.M. (1988). Empty Categories in Chinese. Doctoral Dissertation. University of Ulster.
Hermon, G. (1994). Long-distance reflexives in UG: theoretical approaches and predictions for acquisition. In Lust, B., Hermon, G. & Kornfilt, J. (eds.). Syntactic Theory and First Language Acquisition Cross-Linguistic Perspectives. Volume 2: binding, dependencies, and learnability. Hillsdale: Lawrence Erlbaum Associates.
Rilles, S. (1986). Interlanguage and the pro-drop parameter. Second Language Research, 2: 33-52.
Hirakawa, M. (1989). L2 Acquisition of English Reflexives by Speakers of Japanese. Paper presented at the Conference on the Interaction of linguistics, Second Language Acquisition and Speech Pathology, University of WisconsinMilwaukee.
Hirakawa, M. (1990). A study of the L2 acquisition of English reflexives. Second Language Research, 6: 60-85.
Hornstein, N. & Lightfoot, D. (eds.). (1981). Explanation in linguistics. The Logical Problem of Language Acquisition. London: Longman.
Horrocks, G. (1987). Generative Grammar. London: Longman.
Huang, C-T. J. (1982). Logical Relations in Chinese and the Theory of Grammar. Doctoral Dissertation. MIT.
Huang, C-T. J. (1983). A note on binding theory. Linguistic Inquiry, 14: 554-561.
Huang, C-T. J. (1984a). On the distribution and reference of empty pronouns. Linguistic Inquiry, 15, 531-574.
Huang,C-T.J.(1984b).On the typology of zero anaphora. Language Research,20:85-105.
Huang, C-T. J. (1989). Pro drop in Chinese: a generalized control theory. In Jaeggli, 0. & Safir, K. (eds.), The Null Subject Parameter, 185-214. Dordrecht: Kluwer.
Huang, C-T. J. & Li, Y-H. A. (1996). New Horizons in Chinese Linguistics. Dordrecht: Kluwer.
296
Huang, C. -C. J. & Tang, C. -C. J. (1991). The local nature oflong-distance reflexive in Chinese. In Koster, J. Reuland, E. (eds.). Long-distance Anaphor. Cambridge: Cambridge University Press.
Huang, Y. (1994). The Syntax and Pragmatics of Anaphora: A Study with Special Reference to Chinese. Cambridge: Cambridge University Press.
Huang, Y. -H. (1984). Reflexives in Chinese. Studies in English Literature and Linguistics, 10.
Hyams, N. (1983). Acquisition of Parameterized Grammars. Doctoral Dissertation. CUNY, New York.
Hyams, N. (1986). Language acquisition and the theory of parameters. Dordrecht: Reidel.
Hyams, N. (1987a). The setting of the null subject parameter: a reanalysis. Paper Presented to the Boston University Conference on Child Language Development.
Hyams, N. (1987b). The theory of parameters and syntactic development. In Roeper, T. & Williams, E. (eds.). Parameter Setting, Dordrecht: Reidel.
Hyams, N. (1989). The null subject parameter in language acquisition. In J. Weissenborn, H. Goodluck, & T. Roeper (eds.). Theoretical Issues in Language Acquisition, Hillsdale, NJ: Lawrence Erlbaum.
Hyams, N. (1991). The genesis of clausal structure. In Meisel, J. (ed.). The Acquisition of Verb Placement, 371-400. Dordrecht: Kluwer.
Hyams, N. (1996). The underspecification of functional categories in early grammar. In Clahsen, H. (ed.). Generative Perspectives on Language Acquisition, 91-127. Amsterdam: J. Benjamins.
Hyams, N. & Wexler, (1993). On the grammatical basis of null subjects in child language. Linguistic Inquiry, 24: 421-459.
Jaeggli, 0. (1982). Topics in Roman Syntax. Foris: Dordrecht.
Jaeggli, 0. & Hyams, N. (1988). Morphological uniformity and the setting of the null subject parameter. In Blevins, J. & Carter, J. ( eds.). Proceedings of the 181
h
Meeting of the North East Linguistics Society, Vol.l. 238-252.
Jaeggli, 0. & Safir, K. (1989). The null subject parameter and parametric theory. In Jaeggli, 0. & Safir, K. (eds.). The Null Subject Parameter. Dordrecht: Kluwer.
297
Jackendoff, R. S. (1977). X-Bar Syntax: A Study of Phrase Structure. Cambridge: MIT.
Johnson, J, & Newport, E. (1991). Critical period effects on universal properties of language: The status of subjacency in the acquisition of a second language. Cognition, 39: 215-258.
Juffs, A. (1996). Learnability and the Lexicon: Theories and Second Language Acquisition Research. Amsterdam: J. Benjamins.
Katada, F. (1991). The LF representation of anaphors. Linguistic Inquiry 22: 287-313.
Kellerman, E. (1985). Dative alternation and the analysis of data: A reply to Mazurkewich. Language Learning, 35: 91-106.
Kellerman, E., Sharwood Smith, M. (1986). Cross-Linguistic Influence in Second Language Acquisition. Oxford: Pergamon Press.
Klein, W. & Perdue, C. (1992). Utterance Structure: Developing Grammars Again. Amsterdam: J. Benjamins.
Lado, R. (1957). Linguistics Across Cultures. Ann Arbor: University of Michigan Press.
Lakshmanan, U. (1994). Universal Grammar in Child Second Language Acquisition. Amsterdam: J. Benjamins.
Lakshmanan, U. & Teranishi, K. (1994). Preferences versus grammaticality judgments: some methodological issues concerning the governing category parameter in second-language acquisition. In E. Tarone et al ( eds.). Research Methodology in Second Language Acquisition. Hillsdale, NJ: Lawrence Erlbaum, 185-206.
Lasnik, H. & M. Saito. (1984). On the nature of proper government. Linguistic Inquiry. 15, 2: 235-290.
Lasnik, H. & Uriagereka, J. (1988). A Course in GB Syntax: Lectures on Binding and Empty Categories. Cambridge, Mass.: MIT Press.
Lebeaux, D. (1983). A distributional difference between reciprocals and reflexives. Linguistic Inquiry 14: 723-730.
Lenneberg, E. (1967). Biological Foundations of Language. New York: John Wiley.
Li, C. N. & Thompson, S. A. (1976a). Subject and topic: a new typology. In Li, C. ( ed.). Subject and Topic, New York: Academic Press.
Li, C. N. & Thompson, S. A. (1976b). On the issue of word order in a synchronic grammar: a case against 'movement transformations', Lingua 39: 169-181.
Li, C. N. & Thompson, S. A. (1981). Mandarin Chinese: Afunctional reference grammar. Berkeley & Los Angeles: University of California Press.
298
Li, D.X., & Li, H. (1997). General Survey of Education in China. Department ofForeign Affairs ofthe State Education Commission, People's Republic of China. Beijing: Beijing Normal University Press.
Li, W. D. (1996). Second Language Acquisition of Topic-Comment Structures in Mandarin Chinese. Doctoral Dissertation, University of Alberta.
Li, X. (1998). Adult L2 accessibility to UG: an issue revisited. In Flynn, S. Martohardjono, G. & O'Neil, W. (eds.). The Generative Study of Second Language Acquisition. Hillsdale, NJ: Lawrence Erlbaum.
Li, X. G. (1988). Pronominal Null Objects in Chinese. In Yu, Y. F. (ed.). 1988 Proceedings of the Second International Conference on Generative Grammar. Harbin: Heilongjiang University Press.
Li, Y. -H. (1985). Abstract Case in Chinese. Doctoral Dissertation. University of Southern California, Los Angeles.
Li, Y. -H. (1990). Order and Constituency in Mandarin Chinese. Dordrecht: K.luwer.
Liceras, J. (1988). Syntax and Stylistics: more on the pro-drop parameter. In Pankhurst, J., Sharwood Smith, M. & Van Buren, P. (eds.). Learnability and Second Languages. Dordrecht: Foris.
Liceras, J. (1989). On some properties of the pro-drop parameter: looking for missing subjects in non-native Spanish. In Gass, S. S. & Schachter, J. ( eds.). Linguistic Perspectives on Second Language Acquisition. Cambridge: Cambridge University Press.
Lieberman, P. (1991). Uniquely Human: The evolution of speech, thought, and selfless behavior. Cambridge, Mass.: Harvard University Press.
Lightbown, P.M. & Spada, N. (1993). How Languages Are Learned. Oxford: Oxford University Press.
Lightfoot, (1991). How to Set Parameters. Cambridge, Mass.:MIT Press.
Likert, R.A. (1932). A technique for the measurement of attitudes. Archives of Psychology, No: 40.
Lillo-Martin, D. (1986). Two kinds of null arguments in American Sign Language. Natural Language and Linguistic Theory 4: 415-444.
Long, M. & Sato, C. (1984). Methodological issues in interlanguage studies: an interactionist perspective. In Davies, A. Criper, C. & Howatt, A. ( eds.). Interlanguage, 253-79. Edinburgh: Edinburgh University Press.
299
Lust, B. (ed.). (1986). Studies in the Acquisition of Anaphora: Defining the Constraints. Vol.1. Dordrecht: Reidel.
Manzini, R. & Wexler, K. (1987). Parameters, binding theory and learnability. Linguistic Inquiry, 18: 413-444.
Mazurkewich, C. (1984). The acquisition of the dative alternation by second language learners and linguistic theory, Language Learning, 34: 91-109.
Mazurkewich,C.(1985).Syntactic markedness and language acquisition, SSLA,7:15-36.
Mazurkewich, C. (1988). The acquisition of infinitive and gerund complements by second language learners. In Flynn, S. & O'Neil, W. (eds.). Linguistic Theory in Second Language Acquisition. Dordrecht: Kluwer.
McLaughlin, B. (1987). Theories ofSecondLanguageLearning. London: Edward Arnold.
McLaughlin, D. (1995). Language acquisition and the subset principle, The Linguistic Review, 12: 143-191.
Meisel, J. M. (1995). Parameters in acquisition. In Fletcher, P. & MacWhinney, B. ( eds.)., The Handbook of Child Language. Oxford: Blackwell.
Munnich, E., S. Flynn, & G. Martohardjono. (1994). Elicited imitation and grammaticality judgments tasks: What they measure and how they relate to each other. In E. Tarone et al. (ed.): Research Methodology in Second Language Acquisition, 227-243, Hillsdale, NJ: Lawrence Erlbaum Associates.
Nakamura, M. (1987). Parameterized extension of binding theory. MIT Worldng Papers in Linguistics 9: 193-223.
Napoli, D. J. (1993). Syntax: theory and problems. New York: Oxford University Press.
Newmeyer, F. & Weinberger, S. (1988). The ontogenesis ofthe field of second language learning research. In: Flynn, F. & O'Neil, W. ( eds.) Linguistic Theory in Second Language Acquisition. Dordrecht: Kluwer.
Nie, N.H., Hull, C. H. Jenkins, J. K. Steinbrenner, K., & Bent, D. H. (1975). SPSS: Statistics Package for the Social Sciences. New York: McGraw Hill.
300
Ohler, L. (1988). Commentary: neurolinguistics and parameter setting. In Flynn, S. & O'Neil, W. (eds.), Linguistic Theory in Second Language Acquisition. Dordrecht: Kluwer.
O'Grady, W., Dobrovolsky, W., & Katamba, F. (1997). Contemporary Linguistics: an introduction. New York: Longman.
Ouhalla, J. (1991). Functional Categories and Parametric Variation. London: Routledge.
Ouhalla, J. (1994). Introducing Transformational Grammar: from rules to principles and parameters. London: Edward Arnold.
Pan, H. -H. (1995). Locality, Self-Ascription, Discourse Prominence and Mandarin Reflexives. Doctoral Dissertation. University of Texas at Austin.
Pankhurst, J. , Sharwood-Smith, M. & Van Buren, P. (1988). ( eds.). Learnability and Second Languages: a book of readings. Dordrecht: Foris.
Perdue, C. (1993). Adult Language Acquisition: Cross-Linguistic Perspectives. Cambridge: Cambridge University Press.
Perlmutter, D. (1971). Deep and Surface Constraints in Syntax. New York: Holt, Rinehart and Winston.
Phinney, M. (1987). The Pro-drop parameter in second language acquisition. In Roeper, T. & Williams, E. (eds.)., Parameter Setting. Dordrecht: Reidel.
Pica, P. (1987). On the nature of the reflexivization cycle. In McDonough, J. & Plunkett, B. (eds.). Proceedings ofNELS 17:483-500. UniversityofMassachusetts at Amhest: Graduate Linguistics Students Association.
Pierce, A. E. (1992). Language Acquisition and Syntactic Theory: A Comparative Analysis of French and English Child Grammars. Dordrecht: Kluwer.
Pinker, S. (1984). Language Learnability and Language Development. Harvard, Mass: Harvard University Press.
Pinker, S. (1986). Productivity and conversatism in language acquisition. In Demopoulos, W. & Marras, A. ( eds.)., Language Learning and Concept Acquisition. Norwood, NJ: Ablex.
Pinker, S. (1989). Learnability and Cognition: the Acquisition of Argument Structure. Cambridge: MIT Press.
Pinker, S. (1996). Language Learnability and Language Development. Cambridge,
Massachusetts: Harvard University Press.
Pollock, J. -Y. (1989). Verb movement, UG and the structure ofiP. Linguistic Inquiry, 20: 365-424.
301
Progovac, L. (1992). Relativized SUBJECT: long-distance reflexives without movement. Linguistic Inquiry, 23: 671-680.
Progovac, L. (1993). Long-distance reflexives: movement-to-fufl versus relativized SUBJECT. Linguistic Inquiry, 24: 755-772.
Radford, A.(1988).Transformational Grammar. Cambridge: Cambridge University Press.
Radford, A. (1990). Syntactic Theory and the Acquisition of English Syntax. Cambridge, Mass.: Blackwell.
Radford, A. (1997). Syntax: A Minimalist Introduction. Cambridge: Cambridge University Press.
Randall, J. (1987). fudirect Positive Evidence: Overgeneralizations in Language Acquisition. Bloomington, fud.: fudiana University Linguistics Club.
Randall, J. (1992). The catapult hypothesis: an approach to unlearning. fu Weissenborn, J., Goodluck, H. & Roeper, T. (eds.)., Theoretical Issues in Language Acquisition: Continuity and change in development, 93-138, Hillsdale, NJ: Lawrence Erlbaum.
Reinhart, T. & Reuland, E. (1990). Reflexivity. Linguistic Inquiry 24: 657-720.
Riemsdijk, H. van & Williams, E. (1986). Introduction to the Theory of Grammar. ·cambridge, MA: MIT Press.
Rizzi, L. (1980). Comments on Chomsky. fu Mehler, J. (ed.)., Proceedings of the June 1980 Conference on the Cognitive Sciences. Paris: CNRS.
Rizzi, L. (1982). Issues in Italian Syntax. Dordrecht: Foris.
Rizzi, L. (1986). "Null objects in Italian and the theory of pro". Linguistic Inquiry, 17: 501-557.
Rizzi, L. (1990). Relativized Minimality. Cambridge: Mass.: MIT Press.
Rizzi, L. (1994). Early null subjects and root null subjects. fu Lust, B., Hermon, G., & Kornfilt, J. (eds.). Syntactic Theory and First Language Acquisition: CrossLinguistic Perspectives. Vol.2. Hillsdale: Lawrence Erlbaum Associates.
302
Rizzi, L. (1998). Remarks on early null subjects. Proceedings of the 22"d Annul Boston University Conference on Language Development. Vol.1. (eds.). Greenhill, A. & Hughes, M. et al. Somerville, Mass.: Cascadilla Press.
Ritchie, W. (1978). The right roof constraint in adult-acquired language. In Ritchie, W. (ed.), Second language Acquisition Research: Issues and Implications. New York: Academic Press.
Ritchie, W. (1983). Universal grammar and second language acquisition. In Rogers, D. & Sloboda, J.( eds.), The Acquisition of Symbolic Skills. New York: Plenum Press.
Roeper, T. & Weissenborn, J. (1990). How to make parameters work. In Frazier, L. & de Villiers, J. (eds.). Language Processing and Language Acquisition, 147-162. Dordrecht: Kluwer.
Rutherford, W. (1987). Second language grammar: Learning and teaching. New York: Longman.
Rutherford, W. & Sharwood-Smith, M. (1988). Grammar and second language teaching: A book of readings. New York: Newbury House.
Saleemi, A. P. (1992). Universal Grammar and Language Learnability. Cambridge: Cambridge University Press.
Schachter, J. (1983). Nutritional needs oflanguage learners. In Clarke, M.A. & Handscombe, J. (eds.)., On TESOL '82: Pacific perspectives on language learning and teaching, 175-189, Washington, DC: TESOL.
Schachter, J. (1986a). In search of systematicity in interlanguage production. Studies in Second Language Acquisition, 8: 119-134.
Schachter, J. (1986b). Three approaches to the study of input. Language Learning, 36: 211-225.
Schachter, J. (1988). Second language acquisition and its relationship to Universal Grammar. Applied Linguistics, 9: 219-235.
Schachter, J. (1989). Testing a proposed universal. In Gass, S. & Schachter, J. (eds.), Linguistic Perspectives on Second Language Acquisition, 73-88, Cambridge: Cambridge University Press.
Schachter, J. (1990). On the issue of completeness in second language acquisition. Second language Research, 6: 93-124.
Schachter, J. (1991).Corrective feedback in historical perspective. Second Language Research, 7(2): 89-102.
Schachter, J. (1992). A new account oflanguage transfer. In Gass, S. & Selinker, L. (eds.)., Language Transfer. Amsterdam: J. Benjamins.
Schmerling, S. F. (1973). Subjectless sentences and the notion of surface structure. In Papers from the Ninth Regional Meeting ofthe Chicago Linguistic Society. Chicago Linguistic Society, Chicago, Tilinois.
Schulte-Nafeh, M. & Dussias, P. (1993). Parameters, learnability and the null subject parameter. El Two Talk: Arizona Working Papers in Second Language Acquisition and Teaching (SLAT), Vol.l, 1:85-107.
Schutze, C. (1996). The Empirical Base of Linguistics: Grammaticality Judgments and Linguistic Methodology. Chicago: University of Chicago Press.
Schwartz, B. (1987). The Modular Basis of Second Language Acquisition. Doctoral Dissertation: University of Southern California, Los Angeles.
303
Schwartz, B. (1993). On explicit and negative evidence effecting and affecting competence and "Linguistic Behavior", Studies in Second Language Acquisition, 15: 147-163.
Schwartz, B. & Gubala-Ryzak, M. (1992). Learnability and grammar reorganization in L2A: Against negative evidence causing the unlearning of verb movement. Second Language Research, 8, 1-38.
Schwartz, B. & Tomaselli, A. (1990). Some implications from an analysis of German word order. In Abraham, W., Kosmeijer, W., & Reuland, E. (eds.)., Issues in Germanic syntax, 251-274, Berlin: de Gruyter.
Scovel, T. ( 1988). A time to speak: A psycholinguistic inquiry into the critical period for human speech. New York: Newbury House.
Seliger, H. W. & Shohamy, E. (1989). Second Language Research Methods. Oxford: Oxford University Press.
Selinker,L.(1972).Interlanguage. International Review of Applied Linguistics .1 0:209-231.
Selinker, L., Swain, M. & Dumas, G. (1975). The interlanguage hypothesis extended to children. Language Learning, 25: 139-152.
Selink:er, L. & Lakshmanan, U. (1992). Language transfer and fossilization: the multiple effects principle. Gass, S. & Selinker, L. Language Transfer in Language Learning, 197-216.
Sharwood Smith, M. (1991). Speaking to many minds: on the relevance of different types
304
oflanguage information of the L2 learner.Second Language Research, 7/2,118-32.
Shi, D. (1989). Topic chain as a syntactic category in Chinese. Journal of Chinese Linguistics, 17: 223-259.
Shi, D. (1992). The Nature of Topic Comment Constructions and Topic Chains. Doctoral Dissertation, University of Southern California, Los Angeles.
Si, H. (1988). Empty categories and subject government in Chinese. In Y. E., Yu (ed.) 1988 Proceedings of the Second futernational Conference on Generative Grammar, Harbin.
Stowell,T.(1981).Elements ofPhrase Structure. Doctoral Dissertation. Cambridge, MIT.
Sun, L.-M. (1990). Universals of Reflexives. Doctoral Dissertation. University oflllinois.
Sung, L.-M. & Cole, M. (1991). The effects of morphology on long-distance reflexives. Journal of Chinese Linguistics. 19: 42-61.
Tang, C. -C. J. (1985). A study of reflexives in Chinese. MA Thesis, National Taiwan Normal University.
Tang, C. -C. J. (1989). Chinese reflexives. Natural Language and Linguistic Theory, 7.
Tang, C.-C. J. (1990). Chinese Phrase Structure and the Extended X'-Theory. Doctoral Dissertation. Cornell University.
Tang, C. -C. J. (1994). A note on relativized SUBJECT for reflexives in Chinese. fu Lust, B., Hermon, G. & Kornfilt, J. (eds.). Syntactic Theory and First Language Acquisition: Cross-Linguistic Perspectives. Hillsdale: Lawrence Erlbaum Associates.
Taraldsen, T. (1978). On the NIC, Vacuous Application, and the That-trace Filter. Bloomington: fudiana University Linguistic Club.
Taylor, B.(1974).Toward a theory oflanguage acquisition. Language Learning 24:23-35.
Thomas, M. (1989). The interpretation of English reflexive pronoun by non-native speakers. Studies in Second Language Acquisition, 11: 281-303.
Thomas, M. (1991). Universal Grammar and the interpretation of reflexives in a second language. Language, 67: 211-39.
Thomas, M. (1993). Knowledge of Reflexives in a Second Language. Amsterdam: J. Benjamins.
Thomas, M. (1995). Acquisition of the Japanese reflexive zibun and movement of anaphors in Logical Form. Second Language Research, 11: 206-234.
305
Thomas, M. (1998). Binding and related issues in L2 acquisition: commentary on Part III. In Flynn, S., Martohardjono, G., & O'Neil, W. (eds.). The Generative Study of Second Language Acquisition. 261-276, Hillsdale, NJ: Lawrence Erlbaum.
Trahey, M. (1996). Positive evidence in second language acquisition: some long term effects. Second Language Research, 12: 111-139.
Tsao, F. (1977). A Functional Study of Topic in Chinese: The First Step toward Discourse Analysis, Doctoral dissertation, University of Southern California, Los Angeles, California.
Tsimpli, I. M. (1991). "On the maturation of functional categories: early child speech." University College London Worldng Papers in Linguistics, 3, 123-148.
Tsimpli, I. M. & Ouhalla, J. (1990). Functional Categories, UG and Modularity. Ms. . London: UCL & QMW.
Tsimpli, I.-M. & Roussou, A. (1991). Parameter-resetting in L2? University College London Working Papers in Linguistics, 3: 149-170.
Tsimpli, I.-M. & Smith, N. (1991). Second Language Learning: Evidence from a Polyglot Savant. University College London Working Papers in Linguistics, 3: 171-183.
Vainikka, A., & Young-Scholten, M. (1994). Direct access to X'-theory. InT. Hoestra & B. D. Schwartz (Eds.), Language Acquisition Studies in Generative Grammar: Papers in honor of Kenneth Wexler from the 1991 GLOW workshops. Amsterdam: J. Benjamins.
Valian, (1988). Positive evidence, indirect negative evidence, parameter setting, and language learning. Ms., Hunter College.
Valian, V. (1989). Children's production of subjects: competence, performance, and the null subject parameter. Papers and Reports on Child Language Development, 28: 156-163.
Valian, V. (1990a). Logical and psychological constraints on the acquisition of syntax. In Frazier, L. & de Villiers, J. ( ed)., Language Processing and Language Acquisition. Kluwer: Academic Press.
Valian, V. (1990b ). Null subjects: a problem for parameter-setting models oflanguage acquisition, Cognition, 35: 105-122.
Valian, V. (1991). Syntactic subjects in the early speech of American and Italian children.
Cognition, 40: 21-81.
Van Riemsdijk, H. Williams, E. (1986). Introduction to the Theory of Grammar. Cambridge: MIT Press.
Wakabayashi, S. (1996). The nature ofinterlanguage: SLA of English reflexives. Second Language Research, 12: 265-303.
Wang, Q., Lillo-Martin, D., Best, C. T., & Levitt, A. (1992). Null subjects versus null objects: some evidence from the acquisition of Chinese and English. Language Acquisition, 2: 221-254.
Wang, J.L. & Stillings, J.T. (1984). Chinese reflexives. In Li, X.Y. et al (eds.). Proceedings of the First Harbin Conference on Generative Grammar. Harbin, China: Heilongjiang University Press.
306
Wexler, K. (1994). Finiteness and head movement in early child grammars. In Lightfoot, D. & Hornstein, N. (eds.). Verb Movement, 305-350. New York: Cambridge University Press.
Wexler, K. & Chien, Y-C. (1985). The development oflexical anaphors and pronouns. Papers and Reports on Child Language Development. 24: 138-149.
Wexler, K. & Hamburger, H. (1973). On the insufficiency of surface data for the learning oftransformationallanguages. In Hintikka, K. J., Moravcsik, J. M., E. & Suppes, P. (eds.)., Approaches to Natural Language, 166-179, Dordrecht: Reidel.
Wexler, K. & Manzini, M. R. (1987). "Parameters and learnability in binding". In Roeper, T. & Williams. E. ( eds.), Parameter Setting, 41-76, Dordrecht: Reidel.
White, L. (1985a). Is there a logical problem of second language acquisition? TESL Canada, 2(2): 29-41.
White, L. (1985b ). The pro-drop parameter in adult second language acquisition. Language Learning, 35: 47-62.
White, L. (1985c ). The acquisition of parameterized grammars: subjacency in second language acquisition. Second Language Research, 1: 1-17.
White, L. (1986a). The principle of adjacency in second language acquisition. InS. Gass ( ed.). Second Language Acquisition: a Linguistic Perspective. Cambridge: Cambridge University Press.
White, L. (1986b). Implications ofparametric variation for adult second language acquisition ofthepro-drop parameter. In V. Cook (ed.). Experimental
Approaches to Second Language Learning. Oxford: Pergamon.
White, L. (1987a). Against comprehensible input: the input hypothesis and the development of L2 competence. Applied Linguistics, 8, 96-110.
307
White, L. (1987b). Markedness and second language acquisition: the question oftransfer. Studies in Second Language Acquisition, 9, 261-286.
White, L (1988a). Island effects in second language acquisition. In Flynn, S. & O'Neil, W. ( ed.). Linguistic Theory in Second Language Acquisition. Dordrecht. Kluwer.
White, L, (1988b). Universal Grammar and language transfer. In Pankhurst, J., Sharwood-Smith, M. & Van Buren, P. (eds.). Learnability and Second Languages. Dordrecht: Foris.
White, L. (1989a). The principle of adjacency in second language acquisition: Do L2 learners observe the subset principle? In Gass, S, & Schachter, J. (eds.), Linguistic Perspective on Second Language Acquisition, 134-158, Cambridge: Cambridge University Press.
White, L. (1989b ). Universal Grammar and Second Language Acquisition. Amsterdam: J. Benjamins.
White, L. (1989c). The adjacency condition on case assignment: do L2learners observe the Subset Principle? In Gass, S. & Schachter, J. ( eds.), Linguistic Perspectives on Second Language Acquisition, 134-158, Cambridge: Cambridge University Press.
White, L. (1989d). Linguistic universals, markedness and learnability: comparing two different approaches. Second Language Research, 5: 127-140.
White, L. (1990a). Second language acquisition and universal grammar. Studies in Second Language Acquisition, 12, 121-134.
White, L. (1990b). Another look at the logical problem of foreign language learning: a reply to Bley-Vroman. Linguistic Analysis, 20, 50-63.
White, L. (1991a). Adverb placement in second language acquisition: some effects of positive and negative evidence in the classroom. Second Language Research, 7, 133-161.
White, L. (1991 b). The verb-movement parameter in second language acquisition. Language Acquisition, 1, 337-360.
White, L. (1992a). Universal Grammar: Is it just a new name for old problems? In S. Gass & L. Selinker (eds.), Language Transfer in Language Learning,
219-234, Amsterdam: John Benjamins.
White, L. (1992b ). On triggering data in L2 acquisition: a reply to Schwartz and Gubala-Ryzak. Second Language Research, 8, 120-137.
308
White, L. (1995). Input, triggers, and second language acquisition: can binding be taught? In Eclanan, F., Highland, D., Lee, P., Mileham, J. & Weber, R. (eds.). Second Language Acquisition Theory and Pedagogy, 63-78. Hillsdale, NJ: Lawrence Erlbaum.
White, L. (1996). Universal grammar and second language acquisition: current trends and new directions. In Ritchie, W. & Bhatia, T. (eds.). Handbook of Second Language Acquisition, 85-120. New York: Acedemic Press.
White, L., Hirakawa, M. & Kawasaki, T. (1995). Second language acquisition oflong distance reflexives: effects and non-effects of input manipulation. McGill Working Papers in Linguistics, 12: 129-151.
Whitman, R. & Jackson, K. (1972). The unpredictability of contrastive analysis. Language Learning, 22: 29-41.
Williams, E. (1987). Introduction. In Roeper, T. & Williams, E. (eds.)., Parameter Setting, Dordrecht: Reidel.
Wu, H.M. (1992). Aspects of Chinese Binding. Doctoral Dissertation. University of Washington.
Xie, Tianwei. (1992). An Examination ofthe Topic-Prominence of Chinese Language Learner's Interlanguage. Doctoral Dissertation, University of Pittsburgh.
Xu, L.-J. (1986). Free empty category. Linguistic Inquiry, 17/1: 75-93.
Xu, L. -J. (1993)The long-distance binding of Ziji. Journal of Chinese Linguistics, 21: 123-141.
Xu, L. -J. (1994). The Antecedent ofZiji. Journal of Chinese Linguistics. 22: 115-136.
Xu, L. -J. & Langendoen, D. T. (1985). Topic structures in Chinese. Language, 61: 1-27.
Yang, D.W. (1983). The extended binding theory ofanaphor. Linguistic Research, 19: 169-192.
Ying, H. G. (1999). Access to UG and language transfer: a study ofL2 learners' interpretation of reconstruction in Chinese. Second Language Research, 15:
41-72.
Yip, V. (1989). Aspects of Chinese/English Interlanguage: syntax, semantics and learnability. Doctoral Dissertation, University of Southern California.
309
Yip, V. (1995). Interlanguage and Learnability: From Chinese to English. Amsterdam: John Benjamins.
Yuan, B. (1994). Second language acquisition of reflexives revisited. Language, 3: 539-545.
Yuan, B. (1995). Acquisition of base-generated topics by English-speaking learners of Chinese. Language Learning, 45: 567-603.
Yuan, B. (1997). Asymmetry ofNull Subjects and Null Objects in Chinese Speakers' L2 English. SSLA, 19: 467-497.
Zobl, H. (1980a). The Formal and Developmental Selectivity ofL1 Influence on L2 Acquisition. Language Learning, 30: 43-57.
Zobl, H. (1980b ). Developmental and transfer errors: their common bases and (possibly) differential effects on subsequent learning.TESOL Quarterly,Vol.XN,4:469-479.
Zobl, H. (1988). Configurationality and the Subset Principle: the acquisition ofV1 by Japanese learners of English. In Pankhurst, J., Sharwood-Smith, M. & Van Buren, P. ( eds. )., Learnability and Second Languages, 116-131, Dordrecht: Foris.
311
Appendix 1
Second Language Acquisition Research Survey Questionnaire
This questionnaire is designed as part of a research work on the second language acquisition of Chinese and English. As a research student at the Linguistics Unit, School of Modern Language Studies, UNSW, I would like to invite you as my informants. I am much obliged to your cooperation. Please fill in the form below and answer the questions on the questionnaire if you are interested in the research. Thank you very much.
Background Information
Nationality:
Sex: (Male) (female)
Age: please tick your age category 16-19 (_) 20-24 {_) 25-29 _) 30-34 ( 35-39 (_) 40-44 (_) 45-49 ) 50-59 (_
Place of Birth: _________ (Country)
Are you a) a native speaker of English ____ _ b) a native speaker of Mandarin Chinese
Time learning Mandarin Chinee/English: ___ (years) ____ (months)
Time living in English/Mandarin speaking country: ____ (years)
Current education status: ___ (high school) __ (TAFE) ___ (University) __ (Graduate degree)
Language(s) used at home (if bilingual, please state):
Language(s) of instruction in secondary education: English I Mandarin Chinese I Cantonese I Korean I Japanese Others (please specify) :
Self-perceived language First language: Second language: Third language: Others (if any):
Language tests taken in TOEFL ( __ ) IELTS ( ) Chinese HSK ( __ ) Others ( ______ _
proficiency level: High ( --High ( --)
High ( --)
High ( --)
the past: Score ( ____ _ Overall Band ( __ ) score ( _____ ) Score ( _____ )
Average ( ) Low ( --Average ( - ) Low ( --Average ( - ) Low ( --Average ) Low
(months)
) ) ) )
Appendix2
Acceptability Judgement Test for Learners of English as a Foreign Language (EFL)
The English Version
Instructions:
312
The purpose of this questionnaire is to find out how English speakers of Chinese judge the acceptability of some Chinese sentence structures, and particularly how they rank order sentences with respect to one another.
In these tests, you will be asked to indicate whether certain sentences "sound" more or less acceptable than others. You will soon realise that some sentences are completely acceptable or unacceptable (and you may have an. immediate reaction to them), whereas other sentences may be acceptable, or unacceptable in varying degrees.
Although some of the sentences look similar, there are differences between them and you should treat each one as a separate sentence. The best approach is to judge each sentence fairly quickly. That is, do not try to count up the number of times you felt a particular way, but rather indicate the alternative that seems a reasonable estimate. Please trust your first impression, without trying to recall grammatical rules.
Read the sentences in the questionnaire, preferably in less than 8 to 10 seconds. Please judge the acceptability of these sentences and try to distinguish the degrees of acceptability as you move on. Please circle the numbers [-3 -2 -1 0 +1 +2 +3 ] after reading each sentence. For each question, choose from the following alternatives:
-3 = completely unacceptable -2 = fairly unacceptable
-1 = slightly unacceptable 0 = unable to decide
+1 slightly acceptable +2 = fairly acceptable
+3 = completely acceptable
On the following scale after each sentence, circle the response that best describe your judgement on the acceptability of the Chinese sentence, structures. Please remember that you will have to answer each sentence fairly quickly, preferably in less than 8 seconds. Before we start the experiment, you will be given four practice sentences.
Please note that the sentence structures in the practice sentences are not the sentence structures to be tested in the experiment; they are just example sentences to show how the sentences should be dealt with. Thank you for your cooperation.
Appendix3
Acceptability Judgment Test for Learners of Chinese as a Foreign Language (CFL)
THE CHINESE VERSION
~~~"ill.~
*~~~~~~~~~~q:.~~~~~*~~~®~*~T~~~~~~~M~ili~ 5f!IJllJTo :EE~J£:®~;1J!MJ:JI~r:p, ~ff~~~~-1-~T, IJ..t5f!IJll}f~ffJMJ~~~a :ff®~T ~%:g·J:if1W~51t~~~~~ft~~51t~, Tiif :!i3 5'!--®~T 9ld:EE~f!ll~JJE..t~ 1JJ~51t~~~ t~~51i':~o ~-1'-~TlErifl~~~ilJ~51i':~~~~IWJo ~~_t, ::(f®~-=f"&::(fF:f(fMJX>.f'.m ~*,R~ilJ~51t~~MJ~~o*~~~~~m~~~~®~~~~o
1f®~T;r3'7l§*~~ll.t, ~~~...t#-~#fji§ja :EE~~1l!~~X>.t~-1-~T:ffiSizHfiM!ltMJ5f!IJllJTa :f:!f:-1-~TIWf.WJ~5f!IJllJTII>J'I'~~~::.t~, :S::!!f~~Mi!:ttA.?e+~#o :!i35'r, ~W~~i...t=Utl!.JH M~~*~*~JllJT, iffi'~#l{~~ B;§'~J~TFcM~~ili~iiUJJ$~" l!M~~iffi'.:fl=~iito M~f.WJ~~~..t~~-1-~T,:EEf.WJ~~1'~T~IWJ~X>.i'~T*fi~llJT"*•:f:!f:-t-~TIW ilJ~~~~~~.~ill~:f:!f:-1-~T*~~-1-~~:
[-3 -2 -1 0 +1 +2. +3]
-3. ~X>.t~t~~j!l: -2. ;ft:J~;UE~j!l:
...
-1. ~~~~ o. 3G~ll}f~
+1. t~~j!l: +2. ~~t~~j!l:
+3. %~t~~j!l:
:EEfro ~~-1-1l:J T J§: , :(:EJ§"'iti3'd tl:l MJ~ 1-5f!~Jll}f !fr :j:!G ili :S:~~ ~ B:S::W $ ~~ ~J llJT o i!tiE>fi ; ~~:(£~1'1:iJT..t~M*~IWII>J'~,-:!$~1l!:fffS:(:E+~~~tl:l~llJT"
:EE7f~fC.I~~~illi, ~ffJ~fittl1!1-1-f91J~, {EI.J!®{91J~If!ll9~?l!!~~~:zfqt~~~IW~~ 3G~. 'l?:ifJR~FH*fH!!bf$~~~1l!1JJ:tfi9{91J'il:Jiffi'Ba
313
314
Appendix4
Vocabulary list for the Acceptability Judgment Tests (CFL subjects)
The Chinese-English Vocabulary List
]£~ chenggong to: succeed :t ilJ:kJII mingshan dachuan famous scenic places 1'T~ dasuan intend It!~ dianying movie ~ diu lose ~w gaosu tell :X:T' guanyu about ~-=[- haizi child :m hua flower ~T huai le go wrong 1m hui. return 3< huo fire ~ Jiao teach
•* jingcha police '" ~.ill kanjian see i*~ kecheng course/subject ~ keng cold w~ mianshu look familiar &p niag year :tcJl nu'er daughter ~ qian money JFf qizi wife =m bang to help -~ yitao (classifier) ~?& dan gran of course fg.:t!ll. dianshi television ~tl::t gangcai just now :9:IDli.t nupengyou girl friend H haox:iang seem/appear fgJm diannao computer ic:m jide remember PI~ ken eng possible JE!Jl:. lis hi history ~fiffj meiyouyong of no use/useless
~~- zongdu Governor ~&:f qunian last year
315
s* rib en Japan :!m* ruguo if -h\'-1 renwe1 think ~jig shengdan Christmas ft.Z,. shenme what ·~ }• taolun discuss .91--W waimian outside '-r1t.Z.. weishenme why fii.l!m wenti question :mm xiangxin believe
'"' ~ xiang think mu~. xiaoxi news ~~ xihuan like -~ yiding certainly ~i* yinggai should IE:pjjj zhengque correct ~:lit zhidao know z.m zhihou after
ll* zhunbei prepare ~ dane he bike lf1=~ zuotian yesterday }!JfQ,{ suoyi so/therefore !l~m tushuguan library :k± nushi lady IZSI1!7 yin wei because m,.z,. zenme how ~JJJ zhaodao find iiJJJ. zhidao till :1.:~ zhongyao important ~B ziji self ~ xueqi term }f1 yong to use
316
Appendix5
Vocabulary list for the Acceptability Judgment Tests (EFL subjects)
The English-Chinese Vocabulaty List
address n. :!mi.lf: appear v :[lf~, {JJI.-¥ as regards ~.::f be able to ~~'~ believe v. ;f§f~ borrow v. 1!¥ buy v. ~ computer n. E§.H~ confidence n. f~. ~J[j~
course n. 1*~ definitely adv w~im,-~:!tl! experiment n. ~~ :,"'1"
familiar a. ~~ go wrong ~7 Governor n. ,ffi,1f library n. m=tH~ likely a. PI~ look familiar ~_t~ lose v. ~ money n. ~ once adv. tifj~
police n .. ~~ problem n. fii.Im promise V. ~@: recently adv. IB1I repair v. ~~ seem v. !if~, fW.~ sell v. ~ so far JIJ §iW:fgJl: successful a. Jit:;fj term n. ~m
Appendix 6 Acceptability Judgement Test for the
Learners of English as a Foreign Language (LEFL)
Practice 1: He is an Australian, he can't speak Chinese. [ -3 -2 -l 0 +l +2 +3 J
317
Practice 2: He sleeps like a pig. -3 -2 -1 0 +l +2 +3 J
Practice 3: He likes Beijing very much. [ -3 -2 -l 0 +1 +2 +3 ]
Practice 4: John seems that he is late. [ -3 -2 -l 0 +l +2 +3]
1 Appears that they don't know this matter. [ -3 -2 -l 0 +l +2 +3 ]
2 Allan has bought a new computer, but he doesn't know how to use it.
[ -3 -2 -1 0 +1 +2 +3
3 The next morning, it rained very hard. [ -3 -2 -l 0 +l +2 +3
4 We once met David's daughter. Was very cute. [ -3 -2 -1 0 +1 +2 +3 ]
5 It is very likely that you won't be able to use that fax machine now.
[ -3 -2 -1 0 +l +2 +3 ]
has been cold recently. [ ' It very -3 -2 -1 0 +1 +2 6
7 John said that those students were in the library, but I told him I didn't find them there. [ -3 -2 -l 0 +1 +2 +3 ]
+3
8 This investigation has been started. I hope will be successful. [ -3 -2 -l 0 +1 +2 +3 ]
9 If you write a good book, I will definitely buy it. [ -3 -2 -1 0 +l +2 +3 ]
10 I asked David whether our coach would come next week.
11 This
12 It is
David said he would certainly come. [ -3 -2 -l 0 +l +2 +3 J
news it is possible that she won't believe at [ -3 -2 -l 0 +1 +2 +3 ]
this book that the teacher asked you to read. [ -3 -2 -1 0 +1 +2 +3
13 The little boy feared that his father would beat. [ -3 -2 -1 0 +l +2. +3 ]
14 Suddenly, it snowed heavily. [ -3 -2 -1 0 +l
first.
+2 +3 ]
]
16 I ask John whether he has ever seen the Governor. He says he has seen him on TV. [ -3 -2 -1 0 +1 +2 +3
17 Famous scenic places in China, I have only been to Yellow Mountain.
[ -3 -2 -1 0 +1 +2 +3 ]
18 He has promised to buy his daughter a new computer, but so far hasn't bought yet. [ -3 -2 -1 0 +1 +2 +3
19 It seems that Johnson is ill. [ -3 -2 -1 0 +1
318
+2 +3 )
20 David says John has got a new girlfriend, but I haven't met yet. [ -3 -2 -1 0 +1 +2 +3 ]
21 I asked David whether our coach would come next week, David said he would certainly come. [ -3 -2 -1 0 +1 +2 +3 ]
22 This problem it is quite possible that he will be able to solve. [ -3 -2 -1 0 +~ +2 +3 ]
23 He teaches us Mandarin Chinese last term, but will no longer teach us this term. [ -3 -2 -1 0 +1 +2 +3 ]
24 He is using that pen right now. It is likely that you won't be able to borrow now. [ -3 -2 -1 0 +1 +2 +3 ]
25 Mary lost her bike last week,
26
27
but John says the police have found it for her. [ -3 -2 -1 0 +1 +2 +3 ]
Suddenly, snowed heavily. -3 -2 -1 0 +1
Seems that Johnson is ill. -3 -2 -1 0 +1
+2
+2
28 That fax machine it is very likely that you won't be able to use now.
[ -3 -2 -1 0 +1 +2 +3
29 He teaches us Mandarin Chinese last term,
30
but he will no longer teach us Chinese this term. [ -3 -2 -1 0 +1 +2 +3 ]
Has been very cold recently. -3 -2 -1 0 +1 +2
+3
+3
+3 ]
31 It is quite possible that he will be able to solve this problem. [ -3 -2 -1 0 +1 +2 +3 ]
32 Johnson has bought a new refrigerator. Is made in China. [ -3 -2 -1 0 +1 +2 +3 ]
33 This movie I don't know why he doesn't like. -3 -2 -1 0 +1 +2 +3 )
34 I can't remember where I have met this lady before. -3 -2 -1 0 +1 +2 +3 ]
35 When they heard the news, they won't believe it at all. [ -3 -2 -1 0 +1 +2 +3 ]
36 It appears that they don't know this matter. [ -3 -2 -1 0 +1 +2 +3 ]
37 The members of her family, I have only met her younger sister. [ -3 -2 -1 0 +1 +2 +3 ]
38 Mary's bike has gone wrong. I am going to repair for her tomorrow.
[ -3 -2 -1 0 +1 +2 +3 ]
39 David says John has got a new girlfriend, but I haven't met her yet.
[ -3 -2 -1 0 +1 +2 +3
40 That child didn't let his parents know, himself went to London yesterday.
[ -3 -2 -1 0 +1 +2 +3
41 I can't remember where I have met this gentleman before. [ -3 -2 -1 0 +1 +2 +3 ]
42 He has promised to buy his daughter a new computer, but so far he hasn't bought her one yet. [ -3 -2 -1 0 +1 +2 +3 ] .
43 We once met David's daughter. She is very pretty. [ -3 -2 -1 0 +1 +2 +3 ]
44 I don't know why he doesn't like this movie. [ -3 -2 -1 0 +1 +2 +3 ]
45 As regards famous scenic places in China, I have only been to Yellow Mountain. [ -3 -2 -1 0 +1 +2 +3 ]
. 45 It is likely that you won't be able to borrow the pen right now. [ -3 -2 -1 0 +1 +2 +3 ]
47 Johnson has bought a new hi-fi. It is made in Japan. [ -3 -2 -1 0 +1 +2 +3
48 This gentlemen I can't remember where I have met before. [ -3 -2 -1 0 +1 +2 +3 ]
49 His son is only eight years old, but himself is able to ride a bicycle now. [ -3 -2 -1 0 +1 +2 +3 ]
so I don't know why he doesn't like this article. [ -3 -2 -1 0 +1 +2 +3 ]
51 If you write a good book, I will definitely buy. [ -3 -2 -1 0 +1 +2 +3 ]
319
53 I like this movie very much, but I don't know why she doesn't like. [ -3 -2 -1 o +1 +2 +3 ]
54 This gentleman looks very familiar, but I can't remember where I have met before. [ -3 -2 -1 0 +1 +2 +3 1
55 Alan has bought a new computer, but he doesn't know how to use. [ -3 -2 -1 0 +1 +2 +3 1
56 Mary's bike has gone wrong. I am going to repair it for her tomorrow. [ -3 -2 -1 0 +1 +2 +3
57 I have got a problem with my work.
57
58
As
Mary said that she would help me to solve it. [ -3 -2 -1 0 +1 +2 +3 1
regards the members of her family, I have only met younger sister. [ -3 -2 -1 0 +1 +2
her +3
I immediately recognised these students, and later Peter also recognised. [ -3 -2 -1 0 +1 +2 +3
59 Lily lost her bike last week, but John say's the police have found for her. [ -3 -2 -1 0 +1 +2 +3 1
60 This investigation has been started. I hope it will be successful. [ -3 -2 -1 0 +1 +2 +3 1
61 I have got a problem with my work. Mary said she would help me to solve. [ -3 -2 -1 0 +1 +2 +3 1
63 John said those students were in the library, but I told him I didn't find there. [ -3 -2 -1 0 +1 +2 +3 ]
64 I ask John whether he has ever seen the Governor. He says he has seen on TV. [ -3 -2 -1 0 +1 +2 +3
65 It is quite possible that he will be able to solve this problem. [ -3 -2 -1 0 +1 +2 +3 ]
66 I immediately recognised the students, and later Mary also recognised them. [ -3 -2 -1 0 +1 +2 +3 ]
67 When they heard the news, they won't believe at all. [ -3 -2 -1 0 +1 +2 +3 1
68 There is a park near our school. [ -3 -2
69 David knew Johnson trusts himself. Here: himself = A) David
B) Johnson
-1 0 +1
70 Steve thinks Michael doesn't have confidence in himself. Here: himself = A) Steve [ ]
B) Michael [ ]
+2 +3 1
320
Appendix 7
Acceptability Judgment Test for the Learners of Chinese as a Foreign Language (CFL)
Practice 1 : ·fl!tdi!:m*~J11E.Aa ta shi aadaliya ren.
[-3 -2 -1 0 +1 +2 @
Practice 2: ~.OO..n~-1'-:9;-~.!&a :ftlg;jj:1ft~3ta wo jiangua zhege nu yanyuan. ta feichang piaaliang.
[-3 -2 -1 0 +1 +2 @
Practice 3: 'E~:J~d!:2fs:~;i§J)ijill~~~a ta jiushi zheben shu laashi jiaa du de .
@) -2 -1 0 +1 +2 +3]
Practice 4: it{i!!~)C~:ftt"WT? shui ta de wenzhang piping le?
@ -2 -1 0 +1 +2 +3]
1 :!Ef~1tMfJ~Tflff~-t-$·1fa haa xiang tamen bu liaajie zhege shiqing.
[-3 -2 -1 0 +1 +2 +3]
2 'J'?$~T --e-mr~lt!..W, @;l!{i!!~~:lJt~~ffl'Ea Xiaa Li maile yitai xin de diannaa, danshi ta bu zhidaa zenme yang ta.
[-3 -2 -1 0 +1 +2 +3]
3 jf{=x!ft~, 'Er1ElT*1ma di er tian zaachen, ta xia qi le dayu.
[-3 -2 -1 0 +1 +2 +3]
4 1tMfJiW.7C;j-.00.. T 'J'~~:tl:JL, -l£~q&ilT~a tamen qiantian kan jian le Xiaa "Zhang de nu'er, zhang de hen ke'ai.
[-3 -2 -1 0 +1 +2 +3]
5 18\~Jm.tfffl~~ -e-~jt;fJL;!!~il]"~~~ a ni xian zai yang natai chuanzhenji shi bu keneng de.
[-3 -2 -1 0 +1 +2 +3]
6 'E;~nn~&~a ta wai mian hen leng.
[-3 -2 -1 0 +1 +2 +3]
321
7 *!lili~~.®~~t~:m~-till, 1iJ ~~~Wft!!~tE1l~JLm:~JJJftkfll. Laoshi shuo naxie xuesheng zai tushuguan li,
keshi wo gaosu ta wo zai na'er mei zhaodao tamen. [-3 -2 -1 0 +1 +2 +3)
8 ~~~~B~*~.~~~~~~. zhe ge diaocha yijing kaishi, wo xiangxin hui chenggong.
[-3 -2 -1 0 +1 +2 +3)
9 :!m*f$~1fl-lfs.::flf~' ~-71:~-E. Ruguo ni xiechu yiben hac shu, wo yiding mai ta.
[-3 -2 -1 0 +1 +2 +3]
10 :fl(;jiij;j\.:E~~A,.7C*~*' ;J,.:EM.-71:~*· wo wen Xiao Wang jiaolian jintian lai bu lai,
Xiao Wang shuo yiding hui lai. [-3 -2 -1 0 +1 +2 +3)
11 ~~¥ML~f$311ltl:~iJF1lk~.~;{c!Ji.
12
13
14
zhege xiaoxi ni xianzai gaosu ta you dian taizao. [-3 -2 -1 0 +1 +2 +3]
::J5 J)ifinq f$;g ff.J :!H~:;zJs.:~. Laoshi jiao ni kan de shi zheben shu. [-3 -2 -1
~~~Jit'la1tlr5t~Hr -ftk. zhe nanhai jui pa ta fuqin da ta. [-3 -2 -1
~?&, 'E151JII§ T:k~.
0 +1 +2 +3)
0 +1 +2 +3)
tu ran, ta gua qi le da feng. [-3 -2 -1 0 +1 +2 +3]
15 1$~&:,~.*1:1", =:+~~LJ§", § B~f¥tm. Ni bubi qu guandeng, sanshi miao zhang zhihou, ziji hui ting de.
[-3 -2 -1 0 +1 +2 +3]
16 ;j\:$fiij:fl(;.JX!.m:.JX!.j:J:,m,tf' ~i#.=il(;;£E~;tm.J:.JX!.:(:t1!ko Xiao Li wen wo jianmei jianguo zongdu,
wo shuo wo zai dianshi shang jianguo ta. [-3 -2 -1 0 +1 +2 +3]
17 c:f:!OOtm~ilJ:kJI(, ~.R~l:t:J!iti.LJ. zhongguo de mingshan dachuan, wo zhi qu guo Huang Shan.
[-3 -2 -1 0 +1 +2 +3]
,,
322
18 ..t1-J.l11!!~Jf>Z~:tr.-JL~-f:l'jflt!.JW, @~JU3JII.:r±i:!!.~~o shangge yue ta daying gei nu'er mai yitai xin diannao,
dashi zhidao xian zai ye mei mai. [-3 -2 -1 0 +1 +2 +3]
19 ~:llf~;J'.:E~T o
ta hao ·xiang Xiao Wang bing le. [-3 -2 -1 0 +1 +2 +3]
20 *.:EiJl;J,~~T 1-~Ht93r.-IDJ:tt, @.~~3a!~~ lh!.:M:o Lao Wang shuo Xiao Zhang zhao le ge xin de nupengyou,
danshi wo huai meiyou jianguo. [-3 -2 -1 0 +1 +2 +3]
21 '.f2fij]Jj\.:E.~~~~*~*• 1]\.:Ei3t11!!-~~*o "WO wen Xiao Wang jiaolian jintian lai bu lai,
Xiao Wang shuo ta yiding hui lal. [-3 -2 -1. 0 +1 +2 +3]
22 :1! -t-•mf8\lfll11klti1:;?9:ffl o
zhe ge shiqing ni he ta taolun mei yong. [-3 -2 -1 0 +1 +2 +3]
23 ..t~M11!!W.:'i21fJJ:f:t, @.~:i!~M~~T o
shang xueqi ta jiao women zhongwen, danshi zhe xueqi bu jiao le. (-3 -2 -1 0 +1 +2 +3]
24 11klE:r±Jf3~j(m~, f8\~Uii:r±fl.ffl:;!i!;:~fmft1Io Ta zheng zai yong nar zhi gangbi, ni xiang xianzai jieyong shi bu keneng de.
[-3 -2 -1 0 +1 +2 +3]
25 lt'F~;J'.:E.* T ~ ff$, ff!.:.ll!:;J,~i3tfi?~E.!&:fW{fk~~JT~o Zuotian Xiao Wang diu le zixingche,
26
27
28
danshi Xiao Zhang shuo jingcha yijing bang ta zhaodao le ta. [-3 -2 -1 0 +1 +2 +3]
~~, iSIJ®T::kJXt,. tu ran, gua qi le da feng. [-3 -2 -1 0 +1 +2 +3]
:llf~;J,.:E.7WT o hao xiang Xiao Wang bing le. [-3 -2 -1 0 +1 +2 +3]
~~ il' ftf1J;:f1Lf8\~~:r±m ~~m~~ft11. natai chuanzhenji ni xian zai yong shi bu keneng de.
[-3 -2 -1 0 +1 +2 +3]
323
29 J:~WJf&~~fiJ9='Jt, 1§.:H&~~WJf&;;:r;;~~fiJ9='JtT o
shang xueqi ta jiao women zhongwen, danshi zhe xueqi ta bu jiao women zhongwen le. (-3 -2 -1 0 +1 +2 +3]
30 7-Hiif~l!*o wai mian hen /eng. (-3 -2 -1 0 +1 +2 +3]
31 f$5f'll{tl!i;tJ~~-1-fil.l»Ji'5tJt.l o ni· he ta taolun zhage wenti mei yong. [-3 -2 -1 0 +1 +2 +3]
32 ~$~T -~(19?1j(~' ~J:f:r(@:Jfr(i9o Lao Li mai le yi tai xin de binxiang, shi Zhong Guo zao de.
(·3 -2 -1 0 +1 +2 +3]
33 ~1-Jt!.~~~1'Jlilt~1t.Z.{II!;;:r;;-l{::gX. zhe ge dianying wo buzhidao wei shenme ta bu xihuan.
[-3 -2 -1 0 +1 +2 +3)
34 ~~:lit~4JJU mr=Jt:tEI!llJ!JL.ro.:et~-fE:k±. wo bu jue de yiqian wa zai na'er jian guo zhe wei nushi.
[-3 -2 -1 0 +1 +2 +3]
35 [email protected], f&fiJ1!:zJ>:~;kfWi'E. Tamen ting le :zhege xiaoxi zhihou, taman genben bu xiangxin ta.
[-3 -2 -1 0 +1 +2 +3]
36 'E:iff~:ftkffJ;;:r;;T•:l!-t-.·hf. ta hao xiang taman bu lfaojie zhege shiqing. f-3 -2 -1 0 +1 +2 +3]
37 :fd!!*.m.ll9A. ~..R.m.:M:M!~~. ta jia li de ran, wo jian guo ta mama. [-3 -2 -1 0 +1 +2 +3]
~ f&il9~~T.~tr~~~~~~~. Ta de danche huai le, wo dasuan mingtian bang ta xiuli.
[-3 -2 -1 0 +1 +2 +3]
39 ;lt.:E1JI\INIE1~T ~(rl}3J:Jll.l:ilt:, 1§.:i:~l!'5t;ff .m.:et:fd!!. Lao Wang shuo Xiao Zhang zho fe ge xin de nupengyou,
danshi wo huai meiyou jianguo ta. [-3 -2 -1 0 +1 +2 +3]
40 ~1-~-=ft3t;fi 1!r-Wf&'l5:§-5f'll:4!!:4!!, ~ B--1-AII1=~J!J:It~ ~ T. Ne ga haizi meiyou gaosu baba he mama,
ziji yige ren zuotian dao beijing qu fe. [-3 -2 -1 0 +1 +2 +3]
324
41 ~=1'1EH~QA ftr~:tE!J$JL.OO.~:i! fil~A. Wo bu jide yiqian wo zai na'er jianguo zhe wei faoren.
[-3 -2 -1 0 +1 +2 +3]
42 .J:-1'-JJ fti!~Elz~:fcJL~-~tfi~JW, ffi.R;:KJ!JlJR;{Eftl!tl:!.&f€1-:fcJL~tfiMJ~JJW. shangge yue ta daying gei nu'er mai yiliang xin diannao,
danshi zhidao xian zai ta ye mei gei nu'er mai xin.de diannao. [-3 -2 -1 0 +1 +2 +3]
43 1tl!ffJiifT:JC;'j-Ji!. 7,J,*~WidL, :RtE*~~~~wr~. taman qiantian kan jian le Xiao Zhang de nu'er, ta zhang de hen ke'ai.
[-3 -2 -1 0 +1 +2 +3}
M ~=1'~m~*~~=1'~~+~~. Wo bu zhidao weisheme ta bu xihuan zhe ge dianying.
{-3 -2 -1 0 +1 +2 +3]
45 .¥:-TJ:f:I!E(Ig.:1!ii.l.r:kJII, ~R~~~w. zhiyu zhongguo de mingshan dachuan, wo zhi qu guo Huang Shan.
[-3 -2 -1 0 +1 +2 +3]
46 18\:m31Ut£fff m~~.xm~~~iift~~W. ni xiang xianzai jieyong na zhi gangbi shi bu keneng de.
f-3 -2 -1 0 +1 +2 +31
47 ~'$~7 -ftMJ?lj(~ ~ 'E~!f!@ll![t!9. Lao Li mai fe yi tai xin de binxiang, ta shi Zhong Guo zao de.
[-3 -2 -1 0 +1 +2 +3J
48 J! 1tl:±~~:!\t:1l~ ftr~tEIJ)JjiJL.OO.~. zhe wei nushi wo bu jue de yiqian zai na'er jian guo.
(-3 -2 -1 0 +1 +2 +3]
49 1tl!tf9JL-T;;J-J\.;J;l, @.:liHI BB~~~llt$7. Ta de erzi cai basui, danshi ziji yijing hui qiche fe.
[-3 -2 -1 0 +1 +2 +3] .
50 ~=1'9;Ull~*~ .fl!!~:J.~:i!·Jt-. wo bu zhidao wei shenme ta bu xihuan zha pian wenzhang.
[-3 -2 -1 0 +1 +2 +3]
51 3ln*~lWf.f:l-.2f>:~~. ~-m~. Ruguo ni xiechu yiben hac hu, wo yiding mai.
(-3 -2 -1 0 +1 +2 +3]
325
s2 flk/FT~:a1-mm.. ·{8\;fl:lfll!i~&~m. Ta bu liaojie zhe ge qingkuang, ni he ta taolun meiyou yang.
[-3 -2 -1 0 +1 +2 +3]
s3 ~~iW~:l!1-E!!m, -ar :lfk~~9if:l:itt~tr-z.:l!th/Fif~.
326
Wo hen xihuan zhege dianying, keshi wo bu zhidao weisher:~me ta bu xihuan. [-3 -2 -1 0 +1 +2 +3]
54 :1! 1ft~ A qi ~, {£!.~~~ ieq-i P..( 1lrf=J.21'£!!lJIOLJl!.lt. Zhe wei laoren hen mianshou, danshi wo bu jide yiqian wo zai na'er jianguo.
[-3 -2 -1 0 +1 +2 +3]
55 lj'*~T-f:l'~(!gE!!Mi, {E!.;Ji!:flk~~nt~~.fflo Xiao Li maile yitai xin de diannao, danshi ta bu zhidao zenme yang.
[-3 -2 -1 0 +1 +2 +3]
56 fi!!IW~~T, ~:t.r~IYJ:JC='I'Hfi!!M'E:-a Ta de danche huai le, wo dasuan mingtian bang ta xiuli ta.
[-3 -2 -1 0 +1 +2 +3]
57 ~~1-fiil/Hi, 1]\*l#.~.:E.f!~=W~~tk:'E. Wo you ge wenti, Xiao Li shuo Lao Wang neng bang wo jiejue ta.
[-3 -2 -1 0 +1 +2 +3] "'
58 .3[ T.:ht!H~~.m.tW A, ~ 71- .!X!3i:U!k~~. zhiyu ta jia li de ren, wo jian guo ta mama.
[-3 -2 -1 0 +1 +2 +3]
59 :JJa.ll:;j!JitV..lflT~.®!lf:~. ffl'-*ll,*'IJLV..lfl-*Y. Wo like jiu renchu le zhexie xuesheng, houlai Xiao Li ye renchu lai le.
[-3 -2 -1 0 +1 +2 +3]
so 1$.*11\:E*T m .ff:$, {E!.;!i!:JH~15lW~E.~='mfi!!~UlJT. Zuotian Xiao Wang diu le zixingche,
danshi Xiao Zhang shuo jingcha yijing bang ta zhaodao le. [-3 -2 -1 0 +1 +2 +3]
61 :l!1-~:titB~7f~, 'it:#f{~'E~.JiX;:W. zhe ge diaocha yijing kaishi, wo xiangxin ta hui chenggong.
[-3 -2 -1 0 +1 +2 +3]
62 ~~1-fiilmf, Jj\~~.:E.~fH;p(;f$tk:. Wo you ge wenti, Xiao Li shuo Lao Wang nang bang wo jiejue.
[-3 -2 -1 0 +1 +2 +3]
63 :EJ/ililiDJf!.®ey:~:t£00 ~iff ll!, iii :IM~i!riiFfl!!:reSI! JL &~JIJ. Laoshi shuo naxie xuesheng zai tushuguan fi,
keshi wo gaosu ta zai na'er mei zhaodao [-3 -2 -1 0 +1 +2 +3]
64 1NI!: raJ:re.OO. &.00.~.~ 'If, ~ill.~:tEJt~J:.!i!.~. Xiao Li Wen wo jianmei jianguo zongdu, we shuo wo zai dianshi shang jianguo.
[-3 -2 -1 0 +1 +2 +3]
65 1$;f1lft!!t-;Jit:lZ.:-t-fi:il.rm~~ m. Ni he ta taolun zhe ge wenti meiyou yong.
[-3 -2 -1 0 +1 +2 +3]
66 :reJZ:~Jiti.\lliT ~ .®ey:~, E*'J'*-& i.\lli * T fl!!ffJ. Wo like jiu renchu Je zhexie xuesheng, houlai Xiao Li ye renchu lai le tamen.
[-3 -2 -1 0 +1 +2 +3]
67 fl!!ifJI!RT~-t-m.m.zm, fl!!1fJ~*=1'-m1l!f. Tamen ting Je zhege xiaoxi zhihou, tamen genben bu xiangxin.
[-3 -2 -1 0 +1 +2 +3]
68 ~{f]ey::f3Cllft]lf~1-1HE. women xuexiao fujin you ge gongyuan. [-3 -2 -1 0 +1 +2 +3]
69 ;J,*IJ...:;J;];EM;;m{jlf~ B. Xiao Li renwei Lao Chen xiangxin ziji. ~ll! : ~ B = 1 H'* r 1
2):EMi [ ]
70 ;Mff9Jl:lJ!;J,.:E».." ~ B &~ {i!f't'. Xiao Zhang zhidao Xiao Wang dui ziji meiyou xingxin. ~ll!: ~ B = 1) ;MfE [ 1
2) lj\3:. [ ]
71 :EYi!i~T.:E,Mii~~*-'f~ B~Mc$. Lao Shi du le Wang Xiaojie xie de guyu ziji de gushi. ~11!: ~B=1):EYili [ l
2) .:Eij\MI. [ ]
72 :if;~:;J;];J,!hi&~ p. Lao Ma renwei Xiao Ma hen ziji. )!:£ : ~ B = 1) ;J'l::b [ l
2):E*
327
328
Appendix 8, Subject Information Statement for the Subjects
Dear friend:
I am a PhD student in the Linguistics Unit, School of Modern Language Studies, Faculty of Arts and Social Sciences, the University ofNew South Wales. I am undertaking this research under the supervision of Associate Professor Peter Collins (Head of Linguistics Unit) and Dr Mengistu Amberber. The topic of the research is: Pro-drop Parameter, Universal Grammar and Second Language Acquisition of Chinese and English.
I am asking your cooperation and participation in my research, which aims to investigate the directionality of difficulty in second language acquisition by Chinese-speaking learners learning English as a foreign language and by English-speaking learners learning Chinese a·s a foreign language.
During the research, you will be given a questionnaire to fill out. Please fill in the Second Language Acquisition Survey Form and answer the questions on the questionnaire. The questionnaire will take 30-35 minutes to complete. After the completion of the ,. questionnaire, please follow the instructions and finish the Close Test at the back of the questionnaire.
No materials that may identifY the individuals will be published. Participation in this research is voluntary. You can withdraw from this research whenever you wish without the penalty or prejudice, and all data associated with you will be destroyed if you request this. Every effort will be made to keep the confidentiality of all research data, subject to any possible legal requirements for disclosure. Though the research will contribute to the general knowledge in this field, no guarantee can be given that a benefit will occur to the participants.
Yours sincerely
Jinsong Chen
Linguistics Unit School ofModem Language Studies Faculty of Arts and Social Sciences The University ofNew South Wales Sydney, NSW 2052 Australia
Appendix9
THE UNIVERSITY OF NEW SOUTH WALES
SCHOOL OF MODERN LANGUAGE STUDIES
The University of New South Wales Linguistics Unit
School of Modem Language Studies
Consent Form for the Students Participating in Research Projects
Name ofparticipants: __________ _
Project Title: Second Language Acquisition Survey
Name of Supervisors: NProf. P. Collins; Dr. M. Amberber
1. I consent to participate in the above project, the particulars of which have been explained to me and are appended hereto.
2. I authorize the investigator or his assistant to conduct the questionnaire survey.
3. I acknowledge that:
(a) the possible effects of project have been explained to me to my satisfaction;
329
(b) I have been inf01med that I am free to withdraw from the project at any time and to withdraw any unprocessed data previously supplied, without penalty or prejudice;
(c) The project is for the purpose of research and/or teaching and not for treatment; there is no guarantee that a benefit will occur to the participants;
(d) I have been informed that the confidentiality of the information I provide will be safeguarded subject to any legal requirements.
Signature: _________ _ Date: -------(Participant) (day/month/year)
330
Appendix 10
Sentence Types in the Acceptability Judgment Test for the CFL Subjects
Sentences concerning the null subjects and the base-generated topics in the acceptability judgment test for the CFL learners.
A. Null subject in matrix sentences
1 a) 1iH{11fti7C~ .!l6 T ;j\*SS:kJL, 1£~-{H!fP.f~a (experimental) tamen qiantian kan jian le Xiao Zhang de nu'er, zhang de hen ke'ai.
b) 1t!!111fti7C~ .!l6 T Jj\*llS:kJL, :Q&-f£q~1l!PT~a (control) tamen qiantian kanjian le Xiao Zhang de nu'er, ta zhang de hen ke'ai.
2 a) ;l5*~T -:--ilmff!MJ17J<~, ~t:f:t~~SSa (experimental) Lao Li mai le yi tai xin de binxiang, shi Zhong Guo zao de.
b) ;/5*~T-~mff!MJ17.]<$~, ~~t:f:t~~IMJa (control) Lao Li mai le yi tai xin de binxiang, ta shi Zhong Guo zao de.
B. Null subject in embeded ~entences
3 a) ~-1'-i»H~B~:TfMI', ~if§1~~..1Vt~a (experimental) zhe ge diaocha yijing kaishi, wo xiangxin hui chenggong.
b) ~-i'-i»U~B~7f:Mt, ~if§1~'E~..IVt~a (control) zhe ge diaocha yijing kaishi, wo xiangxin ta hui chenggong.
4. a) ~~~;j\.:EW:~4-U~*' 1]\.:E-m-~~*a (experimental) wo wen Xiao Wang jiaolian jintian lai bu lai, Xiao Wang shuo yiding hui lai.
b) -~fiij;j\.:EW:~4-U~*' 1]\.:E-TMAm-~~*a (control) wo wen Xiao Wang jiaolian jintian lai bu lai, Xiao Wang shuo ta yiding hui lai.
331
C. Sentences with both null subject and null object
5 a) _t~ J.I 1f!!~E)Z~:kJL~-~1WitHw, {§.:;li&1i¥1l:OO:r±i:IL"N:~o (experimental) shangge yue ta daying gei nu'er mai yitai xin _diannao, dashi zhidao xian zai ye mei mai.
b) _t~ J.I 1fl!~E)Z~:kJL~-~~JtHw, 1.§:;li&1i¥1J:OO:r±1il!i:IL~~:kJL~~!¥.JitHruo (control)
shangge yue ta daying gei nu'er mai yitai xin diannao, danshi zhidao xian zai ta ye mei gei nu'er mai xin de diannao.
6 a) _t~:JW1t!!~~{fJJ:f:t)C; {§.:;li&~~~~~To (experimental) shang xueqi ta jiao women zhongwen, danshi zhe xueqi bu jiao le.
b) _t~~·fiMt~{fJJ:f:t)C, 1§.£~~1m~~~{f]lfr)CTo (control) shang xueqi ta jiao women zhongwen, danshi zhe xueqi tabu jiao women zhongwen le.
D. Sentences with absence of weather-predicate
7 a) ~~1!ff1J!~o (experimental) wai mian hen leng. ..
b) 'E'~~ W1!~ o (control) ta wai mian hen leng.
8 a) ~~' !Stl®T::k~a (experimental) tu ran, gua qi le da feng.
b) ~~' 'E!Stl®T::k~a (control) tu ran, ta gua qi le da feng.
E. Sentences without raising-predicate expletive
9 a) :f$~;J,±frJ5IT o (Experimental) hao xiang Xiao Wang bingle.
b) 'E':f$~;J,±frJ5IT o (control) ta hao xiang Xiao Wang bing le.
10 a) 1$~1fl!{f]~TfM.~ ~$·!! o (experimental) hao xiang tamen bu liaojie zhege shiqing.
b) 'E':kf~111!{f]~TfM.~ ~~PI! o (control) ta hao xiang tamen bu liaojie zhege shiqing.
F. Sentences with base-generated topics and sentential subject
11 a) ;§~ i:f{~~1JI.i$~~1±J§ ~/f'l:i.T~~I¥.1 o (experimental) natai chuanzhenji ni xiang xianzai yong shi bu keneng de.
b) {$~~{±}§;§~ i:f1~~1JI.~/FI:i.T~~a11 o (control) ni xiang xianzai yong natai chuanzhenji shi bu keneng de.
18 a) ~ 1'-~'tl-f$5fl:l.flf!i-'.fi.t~.m o (experimental) zhe ge shiqing ni he ta taolun mei yong.
b) f$5fl:l1!f!.i-'.fi.~~ -t-•·tt~J§ o (control) ni he ta taolun zhege shiqing mei yong.
G. Sentences with base-generated topics with Wh-island
13 a) ~1'-EE!.~:lJ<;/F~~;/gft~{t!!/f'.:g.~o (experimental) zhe ge dianying wo buzhidao wei shenme tabu xihuan.
b) :1J<;/F9;Q~~ft~11!!/F:g.:xj(~1'-EE!.~o (control) wo bu zhidao wei shenme tabu xihuan zhe ge dianying.
14 a) ~ 1}ti{±:lJ<;/F 1Jit~~ ~ IDi:lJ<;:fr 00~) LJh!aii:o (experimental) zhe wei nushi wo bu jue de yiqian wo zai na'er jian guo.
b) :lJ<;/F:JAM~~IDi~tEI!lJ~JLJR.~m:k±o (control) wo bu jue de yiqian wo zai na'er jian guo zhe wei nushi.
H. Sentences with non·gap topics
15 a) :l«!!~!l!IWA, ~~..!l6:if:l«!!:!r*Wieo (experimental) ta jia li de ren, wo jian guo ta meimei.
b) ~7:kfk~!l!!¥.J.A, ~~Jh!.:if:f«!!Wief*o (control) zhiyu ta jia 1i de ren, wo jian guo ta meimei.
16 a) J:f:t®IW.«tlr:XJJI, :lJ<;~~ufttrl.ro (experimental) zhongguo de mingshan dachuan, wo zhi qu guo Huangshan.
b) ~-TJ:f:t®IW.«I.IJ:XJ!I, :lJ<;~~NJttlro (control) zhiyu zhongguo de min~shan dachuan, wo zhi qu guo Huangshan.
332
Sentences concerning the null object in the test for the CFL learners
L Sentences with null object in sentential subject
17 a) 1miE:tE.ffl11~3tfffi~, 1$m~:tE{tf.ffl~7f'1lJij~I¥.Jo (experimental) ta zheng zai yong nar zhi gangbi, ni xiang xianzai jieyong shi bu.keneng de.
b) 1$~33ll:tEfff .ffl;lJ~3tfffi~~7f'1lJij~I¥.J o (control) ni xiang xianzai jieyong na zhi gangbi shi bu keneng de.
18 a) {tg7f'T -~ 1-'r!Wt£ , ~5FIHmi;~JitN7ff .ffl o (experimental) ta bu liaojie zhe ge qingkuang, ni he ta taolun meiyou yong.
b) 1$5fl:r1mi;;Ji-t~ 1-!'iiJMN;ff .ffl o (control) ni he ta taolun zhe ge wenti meiyou yong.
J. Sentences with null object in Wh-island
19 a) ~~~~X.X~-1'-E§~, 1lJ~~71'9alnl~fr~:hii!71'.X.Xo (experimental) wo hen xihuan zhege dianying, keshi wo bu zhidao weishenme ta bu xihuan.
b) ~71'9all!t~fr.Z.:hll!~:g:.X.X~1-EE!.~a (control) wo bu zhidao weisheme tabu xihuan zhe ge dianying.
20 a) ~1JI*.A~!W~, {E~~71'1.c~iQ.{IDJ=it:tE!!JJ~JL~l:to (experimental) zhe wei laoren hen inianshou, danshi wo bu jide yiqian wo zai na'er jianguo.
b) ~~-J.c~ill.{llli~:r±!!JJ~Jh~~Uz*Aa (control) wo bu jide yiqian wo zai na'er jianguo zhe wei laoren.
K. Sentences with inanimate ~ull object in main sentence
21 a) 1]\*~T ~it~JWEE!.JI%i, {EI.~{fg7f'~l!t~~Jtlo (experimental) Xiao Li mai le yitai xin de di~ao, danshi tabu zhidao zenme yong.
b) Jj\*~T~f:t~!Wit!.H~, {EI.~{fffi::t:~m~~.fflBa (control) Xiao Li mai le yitai xin de diannao, danshi ta bu Zhidao zenme yong ta.
22 a) figJWlji$:1/f\7, =!tt.T-1!137C:fW1fM~3M!.o (experimental) ta de danche huai le, wo dasuan mingt~an bang ta xiuli.
b) 1miW1!i$$T, =!t:fJ.J!ij7C:fi!J1iM~3M!.Ba (control) ta de danche huai le, wo dasuan mirtgtian bang ta xiuli ta.
333
L. Sentences with animate null object in main sentence
23 a) ~.:EiMA,*3tT -1'MIW:"9::&, ill~~:BS&;ff ..lh!.:t:i:o (experimental) Lao Wang shuo Xiao Zhang jiao le yige xin de nuyou,
danshi wo hai meiyou jianguo.
b) ~ .:E -ml]\$3'tT -1-tslf IW:-9::& , {E ~~:BS&;ff ..lh!.:t:i:il!!o (control) Lao Wang shuo Xiao Zhang jiao le yige xin de nuyou,
danshi wo hai meiyou jianguo ta.
24 a) ~:fl.~Jml;-b\lli T~.@~~' )§}jEJj\$iQ.iJ.-lli-*T o (experimental) wo like jiu renchu le zhexie xuesheng, houlai Xiao Li ye renchu lai le.
b) ~:fl.~tlml;-b\lliT~.@~~' Fo*l]\$iQ.iJ.-lli*T1iMf1o (control) wo like jiu renchu le zhexie xuesheng, houlai Xiao Li ye renchu lai le tamen.
M. Sentences with inaimate null object in subordinate sentence
25 a) J$5'Cij\.:E-* T §.fT.$, ill~Jj\*-mW*Bf&=!W{iM:~JIJT o (experimental) zuotian Xiao Wang diu le zixingche,
. danshi Xiao Zhang shuo jingcha yijing bang ta zhaodao le.
b) J$:X;j\.:E-* T § .fi$, {E~!]\*-m~~Bf&~{tM~JIJT~o (control) zuotian Xiao Wang diu le zixingche,
danshi Xiao Zhang shuo ji:hgcha yijing bang ta zhaodao le ta.
26 a) ~;ff-1'-l'i:iJm!, 1]"$-m~.:E~MW~f!i¥tk:o (experimental) wo you ge wenti, Xiao Li shuo'Lao Wang neng bang wo jiejue.
b) ~;ff-1'-l'i:iJm!, 1]\$-m::?5.:E~~=IW~~f:k:Bo (control) wo yo1:1 ge wenti, Xiao Li shuo Lao Wang neng bang wo jiejue ta.
334
335
N. Sentences with animate -null object in subordinate sentence
27 a) ~lfrliiJWJ~.®~~:tEm =f:Hij"£, PI ;%~11fifF1il!~:tE~~JLt!StJtJJJ o (experimental) laoshi shuo naxie xuesheng zai tushuguan li, keshi wo gaosu ta zai na'er mei zhaodao.
b) *lfrliiJWJ~.®~~:tEm "*111r £ , PI ;%~11rifF1iE~:t£~~ JLN~JJJ1&1fJ o (control) laoshi shuo naxie xuesheng zai tushuguan li,
keshi wo gaosu ta wo zai na'er mei zhaodao tamen.
28 a) ;j\*fn.I~~&~u,~J.I, ~iM.~Eit:f~Lt.~:tlo (experimental) Xiao Li wen wo jianmei jianguo zongli, wo shuo wo zai dianshi shang jianguo.
b) ;J\*fn.J~]!.t§t~u,~J.I, ~iM.~:tJ:Jt;'f11Lt~:Ut1fl!o (control) Xiao Li wen wo jianmei jianguo zongli, wo shuo wo zai dianshi shang jianguo ta.
0. Sentences with inanimate null object coindexed with an argument in an adjunct
29 a) :!tll=*~~lli-*-1«-"*, ~-~~o (experimental) ruguo ni xiechu yiben hao hu, wo yiding mai.
b) :!tll=':~~tl:\-*-1«-~, ~-~~'Eo (control) ruguo ni xiechu yiben hao shu, wo yiding mai ta.
30 a) 1&1fJI!ffT:i!-1-¥i~U~tZJff, 1il!1f1ifl!*~.#r1~o (experimental) tamen ting le zhege xiaoxi zhihou, tamen genben bu xiangxin.
b) 1il!1fJI!ffT:1!1'-1i!t~.z]ff, 1il!1fJifi!*~#H~'Eo (control) tamen ting le zhege xiaoxi zhihou, tamen genben bu xiangxin ta.
336
Sentences concerning short and long-distance reflexives in the test for the CFL learners
P. Object reflexives
34 1]\*}A~~.±;f§1~§ Sa Xiao Li renwei Lao Wang xiangxin ziji. ~£: § B = 1) 1j\*
=2)~.±
35 ~=1m:li*lm.XU"§ B&1f1~JL.'a
[ ] [ ]
Xiao Zhang zhidao Xiao Wang dui ziji meiyou xingxin.
~!!!: § B = 1) ~==2) *lm
[ ] [ ]
= 3) !]\*~~.± =4)~%1~
=3)*=~*1m =4)~%1~
[ J [ ]
[ ] [ ]
Appendix ll
Sentence Types in the Acceptability Judgement Test in the EFL Subjects
Sentences concerning the null subjects and the base-generated topics in the acceptability judgement test for the EFL subjects
A. Sentences with null subject in matrix sentence
1. a). We once met David's daughter. Is'very cute. (experimental)
b). We once met David's daughter. She is very cute. (control)
2. a). Johnson has bought a new refrigerator. Is made in Japan. (experimental)
b). Johnson has bought a new refrigerator. It is made in Japan. (control)
B. Sentence with null subject in embedded sentence
3. a). This investigation has been started. I hope will be successful. (experimental)
b). This investigation has been started. I hope it will be success~ul. (control)
4. a) • I 'asked David whether our coach would come next week. David said would certainly come. (experimental)
b). I asked David whether our coach would come next week. David said he would certainly come. (control)
c. Sentences with both null subject and null object
5. a). He has promised to buy his daughter a new computer, but so far hasn't bought yet. (experimental)
b) . He has promised to buy his daughter a new computer, but so far he hasn't bought her one yet. (control)
6.a). He teaches us Mandarin Chinese last term, but will no longer teach us this term .. (experimental)
b). He teaches us Chinese last term, but he will no longer teach us Chinese this term. (control)
D. Sentence with absence of weather-predicate expletives
7. a). Has been very cold recently. (experimental)
b). It has been very cold recently. (control)
8. a). Suddenly, snowed heavily. (experimental)
b). Suddenly, it snowed heavily. (control)
337
E. Sentence without raising-predicate expletive
9. a). Seems that Johnson is ill. (experimental)
b). It seems that Johnson is ill. (control)
lO.a). Appears that they don't know this matter. (experimental)
b). It appears that they don't know this matter. (control)
F. Sentence with base-generated topics and sentential subjects
ll.a). That fax machine it is very likely that you won't be able to use now. (experimental)
b). It is very likely that ybu won't be able to use that fax machine now. (control)
18.a). This problem it is quite possible that he will be able to solve. (experimental)
b). It is quite possible that he will be able to solve this problem. (control)
G. Sentence with base-generated topics and Wh-islands
13.a). This movie I don't know why he doesn't like. (experimental)
b). I don't know why he doesn't like this movie. (control)
14.a). This lady I can't remember where I have met before. (experimental)
b). I can't remember where I have met this lady before. (control)
H. Sentence with non-gap topics
15.a). The members of her family, I have only met her younger sister. (experimental)
b). As regards the members of her family,
I have only met her younger sister. {control)
16.a). Famous scenic places in China, I have only been to Yellow Mounta~n. (experimental)
b). As regards famous scenic places in China, I have only been to Yellow Mountain. (control)
338
Sentences concerning the null object in the test for the EFL subjects
I. Sentence with null object (object pro) and sentential subject
17.a). He is using that pen right now. It is likely that you won't be able to borrow now.
(experimental)
b). It is likely that you won't be able to borrow the pen right now. (control)
18 a) I think this problem won't be very difficult for him.
It is quite possible that he will be able to solve. (experimental)
b). It is quite possible that he will be able to solve this problem . . (control)
J. Sentence with null object and Wh-island
19.a). I like this movie very much, but I don't know why she doesn't like. (experimental)
b). I don't know why he doesn't like this movie. (control)
20.a). This old man looks very familiar, but I can't remember where I have met before. (experimental)
b). I can't remember where I have met this old man before. (control)
K. Sentence with inanimate null object in main sentence
2l.a). Alan has bought a new computer, but he doesn't know how to use. (experimental)
b). Alan has bought a new computer, but he doesn't know how to use it. (control)
22.a). Mary's bike has gone wrong. Tomorrow I am going to repair for her. (experimental)
b). Mary's bike has gone wrong. Tomorrow I am going to repair it fir (control)
L. Sentence with animate null object in main sentence
23 a) David says John has got a new girlfriend, but I haven't. met yet. (experimental)
her.
b). David says John has got a new girlfriend, but I haven't met her yet. (control)
339
24.a). I immediately recognized these students, and later Mary also recognized. (experimental)
340
b). I immediately recognized the students, and later Mary also recognized them. (control)
M. Sentence with inanimate null object in subordinate sentence
25 a) Mary lost her bike last week, but John says the police have found for her. (experimental)
b). Mary lost her bike last week, but John says the police have found it for her. (control)
26.a). I have got a problem with my work. Mary said she would help me to solve. (experimental)
b). I have got a problem with my work. Mary said she would help me to solve it. (control)
N. Sentence with animate null object in subordinate sentence
27 a) John said those students were in the library, but I told him I didn't find there. (experimental)
b). John said those students were in he library, but I told him I didn't find them there. (control)
28.a). I ask John whether h has ever seen the Governor. He says he has seen on TV. (experimental)
b). I ask John whether he has ever seen the Governor. He says he has seen him on TV. (control)
o. Sentence with inimima te null object co-indexed with an axgument in an adjunct
29. a). If you write a good book, I will definitely buy. (expe.r:imental)
b). If you write a good book, X will definitely buy it. (control)
30.a). When they heard the news, they won't believe at all. (experimental)
b). When they heard the news, they won't believe it at all. (control)
Sentences concerning local and long-distance reflexives in the t~st for the EFL subjects
P. object reflexives
34 John knows Bill trusts himself. Here: himself = 1) Bill [
= 2) John [
35 Peter thinks John doesn't have confidence in himself. Here: himself = 1) Peter [ ]
= 2) John [ ]
341
Appendix 12
SPSS EFL/CFL ANOV A Tables 1 & 2
ANOVA Table 1
Sum of Mean I
Squares df Square F Slg. ~~!mental Between Groups 28.1-40 6 4.690 2.470 .027 S-EFIJNS-MS Within Groups 267.689 141 1.899
Total 295.829 147 Control S-EFIJNS-MS Between Groups 17.991 6 2.998 2.913 .GlO
Within Groups 146.136 141 1.029 Total 163.127 147
Experrnental S-EFIJNS..ES Between Groups 109.553 6 18.259 8.782 ,000 Within Croups 293.170 141 2.079 Total 402.723 147
Control 8-EFIJNS-ES Between Groups 34.211 6 5.702 3.165 .006 Within Croups 2&1.038 141 1.802 Total 288.248 147
Experimental Between Groups 58.335 6 9.723 .• 5.009 .000 S-EFUNS-NO • Within Groups 273.665 141 1.941
Total 332.000 147 . Control S-EFUNS-NO Between Groups 16.605 6 2.767 1.-421 .211
Within Croups 274.630 141 1.948 Total 291.235 147
Experimental Between Groups 34.522 6 5.754 2.551 .022 ~'>·EFL/E)(p-WhP Within Croups 317.996 1-41 2.255
Total 352.519 147 Control S..EFUExp-WhP Between Groups 68.436 IS 11.406 7.993 .00()
Within Croups 201.204 141 1.427 Total 269.640 1-47 .
Experimental Between Groups 18.906 6 3.151 1.508 .180 S-EFUExp-RP Within Croups 29-4.720 141 2.090
Total 313.627 147. Control S-EFUExp-RP Between Groups 18.426 6 3.071 2.004 .069
Within Groups 216.039 141 1.532 Total 23-4 . .c65 147 <
Experimental Between Groups 78.945 6 13.158 5.792 .000 S-EFIJBGT-SS Within Croups 320.298 141 2.272
Total 399.243 147 Control S-EFUBGT -SS Between Groups 40.215 6 6.702 4.473 .000
Within Groups 211.265 1-41 1.498 Total 251.480 147
Experimental Between Groups 82.363 6 13.727 4.784 ,000 8-EFIJBGT-Wh Within Croups 404.576 141 2.869
Total 486.939 147 Control S-EFUBGT-Wh Between Groups 12.336 6 2.056 2.276 .040
Within Groups 127.367 141 .903 Total 139.703 147
Experimental E-EFUNGT BetWeen· Groups 66.803 6 11 . .f67 5.259 ,000 Within Groups 307.«0 141 2.180 Total 376.243 147
342
ANOVA Table 1
Sum of Mean ~§quarei df Square F Slg.
Control E-EFUNGT Betweer. Groups 5.757 6 .959 .580 .746 Within Groups 233.183 141 1.654 Total 238.939 147
Expermental Between Groups 46.879 8 7.813 3 ... 94 .003 5-EFUNO·SS WHhln Groups 315.328 141 2.236
Total 362.208 147 Control 8-EFIJNO-SS Between Groups 45.172 e 7.529 5.415 .000
Within Groups 196.018 141 1.390 Total 241.1fJI) 147
Experimental Between Groups 169.224 e 28.204 13.708 .000 S.EFUNO-Wh WHhln Groups 290.105 141 2.0~7
Total 459.329 147 Control 8-EFUNO-Wh Between Groups 17.776 $ 2.963 3.535 .003 -• Within Groups 1113.157 141 .838
Total 13!5.932 147 . Experimental Between Groups 143.126 6 23.854 10.972. .000 5-EFUJNO-MS Within Groups 306.5.49 141 2.17<1
Total 449.676 147 Control 8-EFl/INQ-MS Between Groups 5.7<U $ .951 1.112 .358
Within Groups 120.531 141 .855 Total 126.235 147
Experimental Between Groups 93.375 e 15.562 9.623 .Coo 5-EFUANO-MS Within Groups 228.036 14~ 1.817
Total 321.410 147 Control 5-EFl.IANo-MS Between Groupe 42.899 e 7.150 <1.36<4 .000
Within Groups 230.993 1-41 1.638 Total • 273.892 147
EXp6ffmentat Between Groups 109.623 8 18.270 9.475 .000 5-EFUJNO..SS Within Groups 271.890 141 1.928
Total 381.519 147 Control 8-EFLIINO-SS Between Groups 23.781 6 3.964 2.526 .024
Within Groups 221.224 141 1.569 Total 245.005 147
Experimental Between Groups 1-48.432 8 24.739 10.739 .000 s-EFUANO..SS Within Groups 324.818 1<t1 2.304
Total -473.250 147 Control s-EFUANO-SS Between Groups 14.816 8 2.~9 1.7<t1 .118
Within Groups 199,981 141 1.418 Total 214.797 147
Experimental Between Groups 56.797 e 9.466 5.193 .000 5-EFUINO-Adjunct Within Groups 257.034 141 1.823
Total 313.831 147 Control S.EFUJNO·Adjunct BetweM Groups 16.090 6 2.682 2.314 .037
Within Groups 163.403 141 1.159
.. T~tal 179.493 147
343
ANOVA Table 2
Sum of Mean Squares df Square F 8~
Experimental Between Groups .925 o4 :231 .282: .889 S-CFUNS..MS WHhfn Groups 66.485 81 .821
total 67.410 85 Control S..CFUNS..MS B~tween Gro\lpS .303 .--;f" 7.563E-{)2 .064 .987
WHhln Groups 73.255 81 .904 Total 73.558 85
Experimental Between Groups .258 4 6.448E-02 .043 .996 S-CFUNS..ES WHhln Croups 121.-451 81 1 . .C99
Total 121.709 85 Control S-CFUNS..ES Between Groups .568 4 .1<42 .122 .974
WHhfn Croups 94.039 81 1.161 Total 94.608 85
Experimental Between Groups 3.388 --:-. .IW.t .572 .684 o4 -S-CFUNS-NO , WHhln Croups 119.987 81 1.-481 Total 123.375 85
.. Control S-CFUNS-NO Between Groups 2.082 -:c .52(} .369 .816
WHhln Croups 108.2-47 81 1.336 Total 110.328 85
Experimental Between Groups 13.705 4 3.426 -4.478 .003 S.CFUExp-WhP Within Groups 61.98-4 81 .765
Total 75.689 85 Control S-CFUExp-WhP Between Groups 12.032 .. 3.008 1.238 .302
Within Croups 196.663 81 2.430 Total 208.895 85
experimental Between Gro\lps 13.020 " 3.255 1.866 .120 S-CFLIFX'fl"RP Within Croups 139.619 81 1.724
Total 152.640 85 Control 8-CFLJExp-RP Between Groups 32.819 4 8.205 4.<431 .003
Within Croups 1-49.966 81 1.852 Total 182.805 85
Experimental Between Groups 26.(108 4 6.502 3.346- ·.014 S-CFUBGT-SS Within Groups 157.402 81 1.943
Total 183 • .1110 85 Control S-CFUBGT .SS Between Groups 10.805 <4- 2.701 1.523 .203
Within Groups 1<43.709 81 1.77<4 Total 154.515 85
Experimental BetWeen Groups 17.249 .c <4.312 2.333 .063 S-CFUBGT-Wh WHhln Groups 149.6115 81 1.848
Total 166.933 IS5 Control S-CFUBGT-Wh Betweon Groups 6.264 .. 1.566 .964 .421
Within Groups 128.037 81 1.592 Total 135.201 as
Experimental S-CFUNGT Between Grdups 5.22<4 .. 1.306 .820 .516 Within Groups 128.997 81 1.593 Total 134.221 a a
344
/\NOVA Table 2
sum of Mean Squares df Sguare F Sig.
Control S-CFUNGT Between Groups 9.432 4 2.358 1.619 .177 Within Groups 117.940 81 1.-456 Total 127.372 e~
Experimental Between Groups 20.605 "' 5.151 3.578 .010 S-CFUNO-SS Within Groups 116.616 81 1.-440
Total 137.~1 85 Control S-CFUNO·SS Between Groups 9.606 4 2.402 1.609 .18(}
Within Groups 12o.gso 81 1.493 Total 130.558 85
Experimental Between Groups 19.701 4 4.925 4.432 ,003 S-CFUNO-Wh Within Groups 00.020 81 1.111
Total 109.721 85 Control S-CFLINO-Wh Between Groups 17.864 4 4.-46$ 3.540 .010
. Within Groups 102.194 81 1.262 Total 120.058 85
Experimental Between Groups 2.625 4 .656 .853 .496 S-CFUINO-MS Within Groups 62.331 81 .no
Total 64.956 85 Control S-CFUINO-MS Between Groups 2.661 "' .665 .407 .803
Within Gloups 132.563 81 1.637 Total 135.224 85
Experimental Between Groups .30-4 4 7.600E-02 .077 .989 S-CFUANO-MS Within Groups 80.045 61 .988
Total 80.349 85 Control S-CFUANO-MS Between Groups .675 4 • 169 .111 . .918
Within Groups 123.595 81 1.526 Total 124.270 85
Experimental Between Groups 1.508
"' .377 .345 .847
S-CFUINO-SS Within Groups 88.550 81 1.093 Total 90.058 85
Control S-CFUINO-SS Between Groups 8.472E-02 .. 2, 118E-02 .• 013 1.000 Within Groups 136.450 81 1.685 Total 136.535 85
Experimental Between Groups 2.566 "'
.642 .430 .787 S-CFUANO-SS Within Groups 120.946 81 1.493
Total 123.515 85 Control S-CFUANO-SS Between Groups 1.233 -4 .308 .159 .956
WHhfn Groups 156.898 81 1.937 Total 158.131 85
Experimental Between Groups 6.758 4 1.609 1.255 .295 S-CFUINO-Adjune1 Within Groups 109.0-47 81 1.346
Total 115.805 8~ Control S-CFUINO-Adjunct Between Groups 3.549 -4 .887 .493 .741
Within Groups 145.696 81 1.799 Total 149.247 85
345
Appendix 13
SPSS Turkey HSD Tests Tables 7 (1-30) and 8 (1-2)
Table7-1
TukeyHSD
95% Confidence Mean Interval
(J) EFL Difference Std. Lower Upper Dependent Variable (I) EFL Groupe_ Group$ (1-J) Error Slg. Bound Bound ~~!mental 7 1 -1.6000" .4711 .012 -2.9891 -.2109 S-EFUNS-MS 2 -.9800 .-4711 .365 -2.3691 .4091
3 -.8913 .-4781 .504 -2.3009 .5183 4 -.7955 .4820 .649 -2.2166 .6257 5 -.7273 .4820 .740 -2.1-48-4 .6939 6 -.3889 .5015 .987 -1.8675 1.0897
Con1rol S-EFUNS-MS 7 1 .9800 .3489 .071 -4.28E.Q2 2.0028 2 .6800 .3469 .+40 -.3428 1.7028 3 .8478 .3520 .195 -.1901 1.8858 .. 1.068.2* .3549 .042' 2.1ne-02 2.1146 . 5 1.36:W .3549 .002 .3172 2.-410l 6 .6667 .3693 .544 ·.4221 1.7554
•. The mean difference is significant at the .05 level.
Table7-2
TukeyHSD
95% Confidence Mean lnteJVal
(J) EFL Difference Std. Lower Upper DeQendent Variable (I) EFL Groups Groups _(1-J) Ermr Slg. Bound Bound ElCpermental S-EFUNS-ES 7 1 -2.4215" ,-4931 ,000 ·3.8752 -.9678
2 -2.2215. .4931 .ooo -3.6752 ·.7678 3 ·2.1137• .5003 .000 -3.5889 •.6385 4 -2.5979• .5044 .000 -4.0851 -1.1107 5 ·.9615 .5044 .476 ·2.4488 .5257 6 -.6838 .5248 ,851 -2.2311 .8636
Control S·EFLJNS-ES 7 1 1.51331 .. .4590 .010 .2299 2.9363 I 2 1.1631 .4590 .147 -.1901 2.5163 3 1.57~3· .4658 .013 .2021 2.9484 4 1.4913* .-4696 .025 .1068 2.8757 5 1.1276 .-4696 .196 ·.2568 2.5120 6 .5897 .-4886 .892 •. 8507 2.0302
•• The mean difference is slgmflcant at the .05 level.
346
Table7-3
TukeyHSO
95% Confidence Mean Interval
(J) EFL Difference Std. Lower Upper Dependent Variable (I) EFL Groups Gro~ _(1-J) Error Sig. Bound Bound
~~rlmentaf 7 1 -2.1846* .47&4 .000 -3.5891 -.7801 S-EFUNS-NO 2 -1.7246* .4764 .005 -3.1291 -.3201
3 -1.9064* .4834 .002 -3.3316 -.4811 4 -1.0664 .4674 .302 ·2.5033 .3705 5 -1.0210 .4874 ,356 -2.4579 .4159 6 -.9679 .5071 .474 -2.4630 .5271
Control 5-EFUNS-NO 7 1 .8306 .4772 .588 -.5762 2.2378 2 .9308 .4772 .447 •.4762 2.3376 3 .4699 .'1843 .960 -.9579 1.8977 4 .4580 .4882 .966' - -.9614 1.8975 ,, 5 .1653 .4882 1.000 -1.2541 1.62~7 • 6 3.632E-02 .5080 1.000 -1.4613 1.5340
•. The mean difference is significant at the .05 level.
Table 7-4
TukeyHSD
95% Confidence Mean Interval
{J) EFL Difference std. Lower Upper Dependent Variable (f) EFL Groups Groups (1-J) Error Slg. Bound Bound Experimental 7 1 -1.9138* .5135 .004 -3.4i278 -.3998 S-EFUExp-WhP 2 -1.5338* .5135 .045 -:.i.047B -2.0E-02
3 -1.2408 .5211 .206 -2.777'/Z .2956 4 -1.1538 .5254 .297 -2.7028 .3951 5 -1.1538 .5254 .297 -2.7028 .3951 6 -1.0427 .5466 .475 -2.6543 .5688
Control 5-EFLIEXp-WhP 7 1 1.7923* .4085 .000 .5880 2.9900 2 1.9123. .4085 .000 .7080 3.1166 3 2.0184* .4145 .000 .7003 3.2.f05 4 1.8287* .4179 .000 .5966 3.0607 5 1.6923. .4179 .001 .4602 2.9244 6 .369~0 .4348 .982 •.9229 1.6409
•. The mean difference Is significant at the .05 level.
347
Table7·5
TukeyHSD
95% Confidence Mean Interval
(J) EFL Difference Std. Lower Upper Dependent Variable (I) EFL Groups Groups (I.J) Error s~ Bound Bound Experimental 7 1 -.4277 .49-« .978 ·1.8852 1.0298 S-EFLJExp-RP 2 -.3277 .49« .995 -1.7852 1.1298
3 f-4.6823E-02 .5017 1.000 -1.5259 1.4322
• .4196 .5058 .982 -1.0716 1.9107 5 .5105 .5058 .952 -.9807 2.0016" 6 ,3590 .5262 .994 -1.1925 1.9104
Control S-EFUExp-RP 7 1 .9523 .4233 .269 -.2956 2,2002 2 .9323 .4233 .294 -.3156 2.1802 3 .71.oCO .4295 .641 -.5!523 1.9804 4 .851-f .o4330 ·""36' - -.4253 2.121!1 5 .6241 .4330 .779 -.6526 1.90.08 6 2.1368E-03 .4505 1.000 -1.3305 U262
Table7-6
Tukey HSD
95% Confidence Mean Interval
(J}EFL Dlrferencl!l Std. Lower Upper Dependent Variable (I) EFL Groups Groups (I.Jl Error ~ Sound Bound Experimental 7 1 -:2.:2862* .5154 .000 -3.8056 -.7667 S.EFUBGT-SS 2 -1.1462 .5154 .282 -2.6656 .3733
3 -1.-4983 .5230 .063 -3.0402 4.358E-02 4 -1.2325 .5273 .226 -2.7870 .3220 5 -.5052 .5273 .963 -2.0598 1.0493 6 -.1517 .~86 1.000 -1.7691 1.46_57
Con1rol S·EFUBGT -SS 7 ·1·· 1.5108. .4186 .000 .2767 2.1448 2 1.8708* .4186 .000 .6361 3.1048 3 1.2525. .4247 .050 2.472E-04 2.5048 4 1.59•W .4282 .004 .3319 2.8569 5 1.5944* .4282 .004 .3319- 2.8569 8 .7585 .4455 .614 -.555().· 2.0721
•. Th!! rn11an dlffamnm~ Is _slgnllleal'lt at the .o~ level.
348
Table 7-7
TukeyHSD
95% Confidence Mean Interval
(J) EFL Difference Std. Lower Upper Dependent Variable (I) EFL Groups Groups (1-J) Error Slg. Bound Bound Experimental 7 1 -2.3154* .5792 .001 -4.0231 -.6077 S-EFUBGT-Wh 2 -1.4554 .5792 .155 -3.1631 .2523
3 -1.4632 .5678 .163 -3.1961 .2697 4 -.6199 .5926 .811 -2.5670 .9272 5 -.1836 .5926 1.000 -1.9307 1.5635 6 -.6709 .6165 .932 -2.4887 1.1468
Control S-EFUBGT-Wh 7 1 .8862 .3250 .092 -7.202E-02 1.8443 2 .8062 .3250 .166 -.1520 1.7643 3 .9114 .3298 .083 -6.095E-02 1.8837 4 .7780 .3325 .225 -.2023 1.7582
' 5 ,5507 .3325 .645 -.4296 q310 6 .2628 .3459 .989 -.7571 1.2827
•. The mean difference Is significant at the ,05 level.
Table 7-8
TukeyHSD
95% Confidence Mean Interval
(J) EFL Difference Std. Lower Upper Dependent Variable (I) EFL Groups Groups J!-J) Error Slg. Bound Bound ExperlmentalS.-EFUNGT 7 1 -2.1431* .5049 .000 -3.6317 -.6544
2 -1.5431* .5049 .036 -3.0317 -5.44E-02 3 -1.5535* .5124 .039 -3.0642 -4.29E-02 4 -1.2185 .5166 .216 -2.7415 .3045 5 -.7185 .5166 - .807 -2.2415 .8045 6 -.2286 .5375 1.000 -1.8132 1.3560
Control S-EFUNGT 7 1 .6831 .4397 .712 -.6134 1.9796 2 .4231 .4397 .962 -.8734 1.7196 3 .5535 .4462 .878 -.7621 1.8691 4 .5140 .4499 .915 -.8124 1.8404 5 .4913 .4499 .931 -.8351 1.8176 6 .1731 .4681 1.000 -1.2069 1.5531
•. The mean dlrference Is significant at the .05 level.
349
Table7-9
TukeyHSD
95% Confidence Mean Interval
(J) EFL Difference Std. Lower Upper Dependent Variable (I) EFL Groups Groujlll_ JI-Jl Error SJg. Bound Bound Experimental 7 1 -1.9538* .511-4 .003 -3.-4615 -.4462 S-EFLJNO-SS 2 -1.7138* ,5114 .014 -3,2215 -.2062
3 -1.8278 .. .5189 .008 -3.3!517 -.2979 4 ·1.5a57 .. .5231 .039 -3.1281 -4.33E-02 5 -1.1993 .5231 .247 -2.7-417 .:3431 6 -.8483 .5443 .709 -2.4531 .7565
Control S-EFUNO-SS 7 1 1.8000*" .403;2 .000 .6113 2.9887 2 1.6200* .4032 .001 .4313 2.8087 3 1.3261* .4091 .020 .1199 2.5323 .. 1.2955* .4125 .028' 7.937E-02 2.5115 • 5 1.38tW .4125 .014 .1703 _2..6024 6 .3689 .4292 .972 -.876-4 1.6542
•, The mean dltferenc:A ll; ~lgnlflcant at the .05 level.
Table7-10
TukeyHSD
95% Confldence Mean Interval
(J) EFL Difference Std. Lower Upper Dependent Variable (I) EFL Groups Groups (I.J) Error Sig, Bound Sound
1 ~~rimental 7 1 -3.7831* .4905 .000 -a.2202 -2.3370 S-EFUNO-Wh 2 -3.4231* .4905 .000 -4.86112 -1.sno
3 -3.0753* .49n .000 -4.54;27 -1.6078 4 -3.1731* .5018 .000 -4.6525 -1.6936 5 -3.1276* .5018 .000 -4.6071 -1.6482 6 -1.5897• .5221 .038 -3.1290 -5.0'5E-02
Control 5-EFUNO-Wh T 1 1.1000* .3130 .008 .1771 2.0229 2 .9800* .3130 .029 5.712E-02 1:9029 3 .8696 .3176 .089 -{I.BiilE-02 1.8061 -4 1.oooo• .3202 .030 5.56-4E-02 1.9442 5 1.0227* .3202 .024 7.857E-02 1.9669 6 .3056 .3332 .970 -.6768 1.28£'.~.
•. The mean difference Is significant at the .05 level.
350
Table7-11
TukeyHSD
95% Confidence Mean Interval
(J} EFL Difference Std. Lower Upper Dependent Variable . (I) EFL Groups Groups (1-J) Error - SJg. Bound Bound Experimental 7 1 -3.7323" .50-42 ·:ooa -5.2188 -2.2458 S-EFLJINO-MS 2 .-2.6323* .5042 .000 -4.1188 -1.1458
3 -2.6010* .5116 .000 -4.3095 -1.2926 4 -1.8287* .~158 .007 -3.3495 -.3079 5 -2.0787• .5158 .001 -3.5995 -.5579 6 -1.6090 .. .5367 .043 -3.1913 -2.7E-02
Control S-EFLJINO-MS 7 1 .4200 .3161 .839 -.5121 1.3521 2 .4600 .3161 .771 -.4721 1.3921 3 4.348E-02 .3208 1.000 -.902-4 .9893 4 .2273 .3234 .992- -.7263 1.1809 5 .3864 .3234 .89£ -.5672 1.~400 6 5.5556E.02 .3365 1.000 -1.0477 .9366
•. Thernuan ·difference Is significant aUhe .. 0!5 level.
Table7-12
TukeyHSD
95% Confidence Mean Interval
(J) EFL Difference · Std. Lower Upper Dependent Variable (I) EFL Groups Groups (1-J) Error SIQ, Bound Bound
• Experimental 7 1 ·2.9169• .4748 .000 -4.3169 ·1.5170 S-EFUINO-SS 2 -2.1009* .4748 .000 -3.5969 -.7970
3 -2.12{)4* .4818 .000 ·3.5410 -.6998 4 -1.3042 .4858 .102 ·2.7364 .1281 5 -1.3269 .4858 .091 -2.7592 .1053 6 -.6~0 .505-4 .822 -2.1782 .8022
Control S-EFLJJNO-SS 7 1 .8923 .4283 .363 -.3705 2.1551 2 1.2323 .4283 .061 -3.05E-02 2.-4951 3 .8010 .4346 .519 -.4804 2.0824 ... .7605 ..4382 .592 -.5314 2.0524 5 1.0559 .4382 .194 -.2360 2.3-479 6 8.120E-02 .4559 1.000 -1.2630 1.425<4
•. The mean difference Is significant at the .05 !&Vel.
351
rable7-13
TukeyHSO
95% Confidence Mean Interval
(J) EFL Dirrerence Std. Lower Upper Dependent Variable (I) EFL Groups Groups (1-J) Error Sig. Bound Bound Experimental 7 , -3.1277* .-4349 .000 -4.-4098 -1.8456 S-EFIJANO-MS 2 -2.2277* .434g .000 -3.5098 -.9456
3 -2.0903* .«13 .000 -3.3913 -.7893 ... -1.7395* .4449 .002 -3.0512 -.4279 5 -1.7395° .4449 .002 -3.0512 •.4279 6 -1.4186* .4629 .035 -2.7835 -5.41E-02
Control S-EFIJANO-MS 7 1 1.6600* .4377 .003 .3696 2.9504 2 1.4000* .4377 .023 .1096 2.6904 3 1.2609 .4441 .068 -4.00E-02 2.5703 .. .9318 .4478 .36-4 -.3863 2.2519 . 5 1.0662 ..4478 .205 -.2519 2.3883 6 .1369 .-4859 1.000 -1.2346 1.5124
•. The mean difference Is slgnlllcant at the .O!l level.
Table7-14
TukeyHSD
95% Confldenoe Mean Interval
(J) EFL Difference Std. Lower Upper Dependent Variable (I) EFL Groups Groups (1-J) Error ~- Bound Bound Experimental 7 1 -3.3692* .5190 .000 -4.9194 -1.8!591 S-EFIJANO-SS 2 -2.5692• .5190 .000 -4.1194 ·1.0591
3 -2.6171* .5267 .000 -4.1698 -1.0643 4 -1.5647 .5310 .050 -3.1301 7.571E-04 5 ·1.6374* .5310 .010 -M029 -.2720 6 -.7970 .5524 .779 -2.4258 .8318
r-Control 8-EFIJANO-SS 7 1 .02-46 .4072 .399 -.3760 2.0252 2 .9246 .-1072 .258 -.2760 2.12!52 3 .5151 .4132 .876 -.7033 1.7334 4 .5664 .4166 .823 -.6619 1.7948 5 .3046 .4166 .969 -.8437 1.6129 ~ 2.350E-~2 .4335 1.000 -1.2545 1.3015
•. The mean dlfferllm:e i11 signlfir.ant at the .051~vel
352
Tabfe7-15
TukeyHSD
Mean 95% Confidence Interval (J) EFL Difference Std. Lower Upper
Dependent Variable (I) EFL Groups Groups (1-J) Error Slg. Bound Bound Experimental 7 1 •2.1o462° .4617 .000 -3.5073 -.7850 S-EFl.JINO-Adjunct 2 -1.6062. .4617 .009 -2.9673 -.2450
a -1.9331" .-4685 .001 -3.3144 -.5518 4 -1.0280 .4723 .308 -2.4205 .3646 5 -1.323.ol .4723 .075 -2.7160 6.913E..02 6 -.7906 .4914 .677 -2.2395 .6583
Control S-EFUINO-Adjunct 7 t 1.1.o~oo• .3681 .032 5.471E-02 2.2253 2 .5400 .3681 .765 -.5-453 1.6253 3 .3913 .3735 .943 -.7100 1.4926
" .3864 .3766 .9-l6. -.7240 Vl967 5 .4773 .3766 .867 -.6330 pa1s 6 .1667 .3918 1.000 -.9886 1.3219
•. The mean difference I$ significant at the .05 level.
353
Table7-16
TukeyHSD
95% Confidence Mean lnteJVal
(J) CFL Difference S1d. Lower Upper Dependent Variable (I) CFL Groups Groups (I.J) Error S_lg, Bound Bound ~rimental 5 1 .1285 .3113 .994 -.7401 .9970 5-CFUNS.MS 2: 9.3750E-02 .3113 .998 -.962~ .n4s
3 .1121 .3156 .997 -.7664 .9926 <4. ~.1232 .3156 .995 ·1.0037 .7573
Control S-CFUNS-MS 5 1 .1076 .3268 .997 -.8041 1.0193 2 .1354 .3268 .994 -.nea 1.0471 3 1.8382E-03 .3312 1.000 -.9261 .9224 .. .1158 3312 .997 •.808-4 1.0400
Table7-17
TukeyHSD
95% Confidence Mean lnteJVal
(J) CFL Difference Std. Lower Upper Dependent Variable (I) CFL Groups Gro~ (I.J)_ Error s~ Bound Bound Experimental 5 1 3.4722E-02 .4207 1.000 -1.2086 1.1392 S-CFUNS-ES 2 4.861E-02 .4207 1.000 -1.1253 1.2225
a 5.515E-02 .4265 1.000 -1.1349 1.2452
" 9.1912E-02 .4265 1.000 -1.2820 1.0981 Control S-CFUNS-ES 5 1 .2257 .3702 .973 -.8073 1.2587
2 3.125E-02 .3702 1.000 -1.0017 1.0042 3 6.066E-02 .3753 1.000 -.9865 1.1078 4 3.125E-02 .3753 1.000 ·1.0159 1.07'84
Table7-18
TukeyHSD
95% Confidence Mean Interval
(J) CFL Difference Std. Lower Upper Dependent Variable . (I) CFL Groups Groups ji.J) . Error Slg. Bound Bound Experimental . 5 1 .5-486 .4182 .685' -.6182 1.7154 S-CFUNS-NO 2 .1875 .4182 .991 -.9793 1.3543
3 .1434 .4239 .997 -1.0395 1.3262 .. 2.574E-02 .-4239 1.000 -1.1571 1.2086 Control s-cFUNS-NO 5 1 .4306 .3972 .814 -.a1n 1.5388
2 .1528 .3972 ,995 -.9555 1.2610 3 .1103 .-4027 .999 -1.0132 1.2338 .. 2.206E-02 .4027 1.000 -1.1014 1.1456
354
Table7-19
TukeyHSD
95% Confidence Mean Interval
(J) CFL Difference Std. Lower Upper Dependent Vllrl!lble (I) CFL Groupa Groupe (1-J) Error Slg. Bound Bound Experimental 5 1 .9861* .3006 .013 .1475 1.8248 5-CFUExp-WhP 2 .5417 .3006 .379 -.297(} 1.3803
3 9.659E-02 .3047 .998 •.7546 .9458 4 5.1471E-02 .3047 1.000 -.9016 .7987
Control S-CFLJExp-WhP 5 1 -.9931 .5357 .350 -2.4876 .5015 2 -.6319 .5357 .763 -2.1265 .8626 3 8.4559E-02 .5430 1.000 -1.5997 1.4300
" -.2316 .5430 .993 -1.7467 1.2835 •. Th11 miiM difference Is significant at the .05 h;rvvl.
Tllblt:t7·20
TukeyHSD
95% Confidence Mean Interval
(J) CFL Diffluence Sid . Lower Upper Dc:pcndent Vnn11hle . _Q).CFL Grcu~ Groups (1-J) Error S)Q. Bound Bound Expenmental 5 1 •.5278 .4511 .768 ·1.7864 .7309 S-CFUExp-RP 2 -1.0278 .4511 .163 ·2.2864 .2309
3 ·1.0682 .4573 .131 ·2.3642 .1877 4 -.7941 .4573 .418 ·2.0701 .4818
Control S-CFUExp-RP 5 1 -1.3472* .-4675 .039 -2.6518 -4.3E-02 2 ·.5417 .4675 .775 -1.8Ml2 .7629 3 .2279 .4740 .989 -1.0945 1.550<4 -4 .2868 .4740 .974 -1.0357 1.6093
•, Th11 meM difference Is siQnmcant lilllh11 .05 lllVIll
355
Table7-21
TukeyHSD
95% Confidence Mean Interval
(J)CFL Difference Std. Lower Upper Dependent Variable . (I) CFL Grotlp.s Groups (I..J) Error Slg~ Bound Bound ~?Perlmental 5 1 1.60-42" .4790 .011 .2678 2.9<C06 S..CFUBGT -SS 2 1.1697 .-4790 .120 -.1767 2.4961
3 .3493 .-4656 .-410 -.5055 2.2041 ... .4669 .4856 .871 -.8879 1.8217
Control S-CFIIRGT·SS 5 1 1.0000 ,4577 .196 -.2770 2.2770 2 .5556 .-4577 .743 -.7214 1.6325 3 .GOOO .4640 .817 -.7945 1.79<t5 4 .1176 .4640 .999 -1.1769 U122
•. The mean difference Is significant at the .OS level
Table 7-22
TukeyHSD
95% Confidence Mean Interval
(J) CFL Difference Std. Lower Upper Dependent Variable (I) CFL Groups Grol!!!_s (1-J) Error ~ Bound Bound EXperimental 5 1 1.3715• .4671 .034 6.828E-02 2.6148 5-CFUBGT -Wh 2 .8715 .4671 .344 •.4317 2.1748
3 .6673 .4735 .624 •.6539 1.9884
" .4908 .4735 .838 -.8303 1.8120 Control S·CfUOGJ •Wh 5 1 .7465 .4335 .426 -.4630 1.9561
2 .5243 .4335 .7-46 -.6853 1.7339 3 .3511 .4395 .930 -.8751 1.5773 -4. .1158 .4395 .999 -1.1104 1.3420
•. The mean difference IS significant at the .05 level.
Table 7-23
TUk&yHSD
95% Confidence Mean Interval
(J) CFL Difference Std. Lower Upper Dependent Variable (I) CFL Groups GroupG (1-.D_ Error s~. Bound Bound ~~!mental :J 1 .7292 .4336 .451 -.4807 1.9390 5-CFUNGT 2 .5903 .4336 .654 ·.6196 1.8001
3 .5772 .-4396 .634 -.6493 1.8037
" .4301 .4396 .864 -.7963 1.6566 Control &CFI /NGT 5 1 .7361 .-4146 .395 -.oi207 1.8929
2; .1528 .-4146 .996 w1,Q0.40 1.3096 3 9.5588E-02 .-4203 .999 -1.2683 1.0771 .. w,1838 .4203 .992 -1.3566 .9889
356
Table7-24
TukeyHSO
95% Confidence Mean Interval -
(J) CFL Difference Std. Lower Upper DePEndent Variable (I) CFL Groups Groups ji..Jt Error Slg. Bound Bound Experimental 5 1 1.3611. .4123 .Q12 .2106 2.5114 5-CFUNO-SS 2 .9167 .4123 .182 -.2336 2.0670
3 .7353 .o4179 .404 -.4308 1.9014 4 .2059 .4179 .988 -.9602 1.3720
Conlrol S-CFLINO-SS 5 1 .9861 .4199 .140 -.1654 2.1576 2 ,5694 .o4199 .657 -.6021 1.i.of09 3 .5515 .4256 .695 -.6361 t.7391 4 .2279 .4256 .983 -.9591. U155
•. The mean difference is significant allhe .05 level.
Table7-25
TukeyHSO
95% Cantlclence Mean Interval
(J)CFL Difference Std. Lower Upper Dependent Variable . (I) CFL Grou~ Gra~ jbl)_ Error ~ Bound Bound EXperimental 5 1 1.326<4* .3622 .004 .3157 2.3371 S-CFLJNO-Wh 2 .6319 ·~ • .J13 •.3787 1.6426
3 .5110 .3672 .635 -.5135 1.5356 .. 6.985E-02 .3672 1.000 -.9547 1.01).44 Control S-CFLJNO-Wh 5 1 1.291r .385Q .011 .2148 2.3685
2 .6806 .3859 .402 -.3963 1.757.of 3 .3603 .3912 .888 -.7313 1.4519
" .1838 .3912 .990 -.9078 1.2755 •. The mean difference Is significant at the .05 level
Table 7-26
TukeyHSD
95% Confidence Mean Interval
(J) CFL Difference Std. Lower Upper Dependent Variable (I) CFL GrOUJlS Gro~ ji..J) Error ~lg. Bound Bound Experimental ::; 1 .5347 .3014 .396 -.3063 1.3757 S-CFLJINO-MS 2 .2569 .301-4 .913 -.5840 1.0979 . 3 .2390 .3056 .935 -.6136 1.0915 .. .1507 .3056 .988 -.7018 1.0033 Control S-CFLJINO-MS 5 1 .3958 .4396 .896 -.8306 1.6223
2 .3125 .<4396 .953 -.9139 1.5389 3 .5478 .«56 .734 •,6955 1.7911 -4 .3713 .«56 .919 ~.8720 1.61.46
357
Table 7-27
TukeyHSD
95% Confidence Mean Interval
(J} CFL Difference Std. Lower Upper Dependent Variable (I} CFL Groups Groups {1-J) Error SJ2: Bound Bound Experimental 5 1 .3368 .3592 .881 -.6656 1.3392 S-CFUINO-SS z .2257 .3592 .970 -.ns1 1.2281
3 9.007E-02 .3642 .999 -.9261 1.1062 .. 1.838E-03 .3642 1.000 -1.01-43 1.0180 Con1rol S-CFUINO-SS l5 1 1.3889E-02 .44$0 1.000 -1.2582 1.2304
2 6.944E-02 .4460 1.000 -1.1749 1.3137 3 2.206E-02 .4521 1.000 -1.2393 1.2835
,4 5.1-47E-02 .4521 1.000 -1.2099 1.3129
Table 7-28
TukeyHSD
95% Confidence Mean Interval
(J) CFL Difference Std. Lower Upper Dependent Variable {I) CFL GroUJXS Gro~ (1-J}_ Error SJg. Bound Bound Experimental 5 1 .1424 .3<416 .994 -.8107 1.0954 S-CFUANO-MS 2 8.681E-02 .3<416 .999' -.8662 1.0398
3 1.65-fE-02 .3<463 1.001> -.9496 .9827 4 1.2868E-02 .3463 1.000 -.9790 .9533
Control S-CFUANO-MS 5 1 7.292E-02 .-4244 1.000 -1.1113 1.2572 2 j-6.5972E-02 .4244 1.000 -1.2502 1.1183 3 -.1379 ,4303 .998 -1.3384 1.0626 4 -.1673 .4303 .995 -1.3678 1.0332
Table 7-29
TukeyHSD
95% Confidence ' Mean Interval
(J) CFL Difference std. Lower Upper Dependent Variable (I) CFL Group$ Groups (1-J) Error S_lg. Bound Bound El<perlmental :5 1 .4340 .4199 .839 -.7375 1.6055 S-CFUAN0-5S 2 .2396 .4199 .979 -.9319 1.4111
3 .3327 .4256 .935 -.8549 1.5203 .. 9.19tE-03 .4256 1.000 -1.1784 1.1968 Control S-CFUANO-SS 5 1 7.9861E-02 ... 782 1.000 -1.4141 1.2544
2 2.4306E-02 .4782 1.000 -1.3566 1.3100 3 -.2040 .4848 .993 ~1.5567 1.1486 .. -.3217 . ..C848 .964 -1.6743 1.0309
Table7-30
TukeyHSD
Mean (J) CFL Difference
Dependent Variable (I) CFL Groups Groups (1-J) Experimental 5 1 .7535 8-CFUINO-Adjunct 2 .7257
3 .4283 4 . 2813
Control 8-CFUINO-Adjunct 5 1 .4479 2 .3646 3 .5165 .. 7.537E-02
Table 8-1
Dependent Variable: refteXive binding TukayHSD
Mean (J) EFL Difference
(I) EFL Grouos Groups {1-J) 7 1 .4000•
2 .4200. 3 .3478* 4 .2500 5 .2500 6 .1389
Std. Error .100& .1006 .1021 .1029 ,1()2g
.1071 •. The mean difference is significant at the .05 level.
T•blt8-2
Dependent Variable: reflexive binding of zl.il TukayHSO
95% Confidence Interval
Std. Lower Upper Error Slg. Bound Bound
.3987 .331 -.3569 1.8656
.3987 .369 -.3867 1.6381
.4041 .826 -.6993 1.5560
.4041 .957 -.8464 . 1.4089
.4608 .667 -.8379 1.7337
.<1608 .932 -.9212 1.6504
.4671 .803 -.7869 1.8200
.4671 1.000 -1.2281 1.3788
95% Confidence Interval
Lower Upper Sia. Boood Bound .001 .1033 .6967 .001 .1233 .7167 .012 4.sne..()2 .6489 .187 ~.3513E-02 .5535 .187 j.5.3513E-02 .5535 .854 -.1769 .4547
M!Uin ~g.nlkience Interval (J)CFL Difference Std. Lower Upper
Jl)_ CFL Groups Groups (J.J) Error Sia. Bound Bound 5 1 -.4028 .1577 .089 -.8-428 3.720E..Q2
2 -.3472 .1577 .189 -.7872 9.276E..02 3 -.3897 .1599 .116 -.8357 5.632E-02 4 -.3309 1599 .243 -.7769 .1151
358