 

-1.
.33..

at”

P.
..:..L..
I £53
a z.

 

:5.
u .

Iii .n‘l

..

. ...!|

mm

.5

naﬁamﬁ.&gﬁm§ _

1
L

9.1:

a.

H.533

7:1

mm! Aﬂwwwmﬁuﬂ

.9.

(.5

 

.0.
(17.73.64,...
@0an ..
y 33!: t

. :s
. 33v»
.4 {.1

CC"
'11:...8

10?) T

This is to certify that the
dissertation entitled

DEVELOPMENT OF MORPHOSYNTACTIC FEATURES
OF THE COMPLEMENTIZER PHRASE IN L2 RELATIVE
CLAUSES OF KOREAN LEARNERS

presented by

lL-JAE LEE

has been accepted towards fulfillment
of the requirements for the

PHD. degree in LINGUISTICS

 

 

W MM

Major Profé’ssor’ 3 Signature

/4 17.75.2199,;

 

Date

MSU is an Affirmative Action/Equal Opportunity Institution

LIBRARY
Michigan State
niversity

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/07 p:lClRC/DateDue.indd-p.1

DEVELOPMENT OF MORPHOSYNTACTIC FEATURES OF
THE COMPLEMENTIZER PHRASE IN L2 RELATIVE CLAUSES OF
KOREAN LEARNERS

By

Il-jae Lee

A DISSERTATION

Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Linguistics and Germanic,
Slavic, Asian and African Languages

2006

ABSTRACT

DEVELOPMENT OF MORPHOSYNTACTIC FEATURES OF
THE COMPLEMENTIZER PHRASE IN L2 RELATIVE CLAUSES OF
KOREAN LEARNERS

By

Il-jae Lee

This dissertation argues that the acquisition of phrase structure and the acquisition of
relevant morphosyntactic features are independent of each other, as discussed in Lardiere
(1998a, 1998b) and Haznedar (2003), and that the entire Ll phrase structure is available
- as the base of L2 acquisition. This claim supports the Full Transfer/Full Access (FT /FA)
hypothesis of Schwartz and Sprouse (1994, 1996). Cross-sectional elicited spoken and
written data from adult Korean learners of English show that they acquire that/a-relative
clauses retaining [—wh, +pred] prior to wh-relative clauses retaining [+wh, +pred] (cf.
Rizzi 1990). T hat/a-relative clauses are prevalent in the early stages of acquisition. Wh-
relative clauses, however, dominate as the complexity of relative CPs escalates, as the L2
proﬁciency increases, and as the monitoring time extends. Korean teachers, as well as
advanced L2 learners, typically opt for wh-relative clauses, whereas native speakers

generally dispense with wh-relative clauses and prefer that/a-relative clauses.

DEDICATION

To my parents and family.

Also to my teachers, colleagues, and friends.

iii

ACKNOWLEDGEMENTS

I thank all the people who helped me stay focused on completing this dissertation
which required much labor and many years of patience. All four members of my
committee inspired me and guided the development of my academic thinking.

Grover Hudson, my dissertation advisor, directed me from the beginning to the
completion of this dissertation with his exceptional insight and perspective on linguistic
and scientiﬁc thinking and writing. I want to express to him my deep gratitude and
profound appreciation.

Alan Munn, my syntax professor, taught me many technical details of how to
understand and compare critical research. His lectures contributed signiﬁcantly to the
theoretical basis for this dissertation.

Dennis Preston provided me with constant guidance, motivation, and assistance on
crucial points. The data collection practice ﬁom his lectures trained me to be more
rigorous in my own ﬁeldwork.

Cristina Schmitt taught me how to apply syntactic theories to the ﬁeld of language
learning, which gave me a practical foundation in the ﬁeld of second language acquisition.

Further, I also thank Yen-Hwei Lin for accepting me as a linguistics graduate student
at Michigan State University and for her valuable advice. Susan Gass, the original advisor,
introduced me to an interesting area in second language acquisition. Patricia Lunn, an

outsider reader, thoroughly read the entire dissertation and gave me insightful comments.

I thank the professors at Kangwon National University. I would like to extend my
special thanks to Nam-gook Kim, my M.A. advisor, for his dedicated advice and support.

I thank Professor Jong-mi Kim, who gave me endless guidance and my ﬁrst
academic appointment at the university.

I also especially thank the following professors for their support and encouragement:
Nak Seung Baek, Sae Youn Cho, Frederick 1. Kaplan, Ii Hwan Kim, Kunsoo Kim, Song
Kyo Kim, Bo Kyong Lee, Tae Hwa Park, Sung Kyun Shin, Jin Koo Yoon, and Yong-kyu
Yun.

I would like to thank Dami Lee at Hanyang University for broadening my
perspective on English academic societies in Korea and for her sincere advice.

I would also like to recognize the help of Rebecca Roeder, Peter Githinji, and
Christopher Wolfe. And, thank you, Koo Young Choi, John Grant, Jinwoo Hwang, Yeong
Hoon Kim, Borami Ku, Sehong Oh, Gab Dong Park, Kyeong Min Park, Moon Su Park,
Hyun Do Shin, Meen Um, Hyunsung Yoon, and Jee Young You.

For being with me through all this, I thank my family and my deceased father.

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................................ viii

LIST OF FIGURES ............................................................................................................. x

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

1.1 Signiﬁcance of the Study ......................................................................................... 7

1.1.1 Minimal Trees (MT) Hypothesis ................................................................... 9

1.1.2 Full Transfer/Full Access (FT/FA) Hypothesis ........................................... 15

1.1.3 Alternative Approach ................................................................................... 19

1.2 Syntactic Structures of CPs ................................................................................... 21

1.2.1 A Typology of CPs ....................................................................................... 25

1.2.2 Movement and Predication in English Relative CPs ................................... 28

1.2.3 Structure of the Korean Relative CP ............................................................ 35

1.3 Evidence from the Relative CP .............................................................................. 37

1.4 Chapter Summary .................................................................................................. 39
CHAPTER 2

2.1 Review of Literature .............................................................................................. 41

2.2 Chapter Summary .................................................................................................. 46
CHAPTER 3

3.1 Purpose of the Study .............................................................................................. 48

3.2 Research Questions ............................................................................................... 50

3.3 Research Hypotheses ............................................................................................. 51

3.4 Research Methodology .......................................................................................... 51

3.4.1 Participants and Classiﬁcation into Levels .................................................. 51

3.4.2 Data Collection Method ............................................................................... 54

3.4.3 Data Collection Instrument .......................................................................... 56

3.4.4 Research Administrators .............................................................................. 59

3.4.5 Data Treatment ............................................................................................ 60

3.4.6 Nonparametric Statistical Techniques ......................................................... 62

3.5 Chapter Summary .................................................................................................. 64
CHAPTER 4

4.1 Testing Hypothesis 1: SUB Relative CP ................................................................ 67

4.2 Testing Hypothesis 2: DO Relative CP .................................................................. 73

4.3 Testing Hypothesis 3: 10 Relative CP ................................................................... 76

4.4 Testing Hypothesis 4: SUB < DO ......................................................................... 81

4.5 Testing Hypothesis 5: DO < 10 ............................................................................. 84

4.6 Testing Hypothesis 6: SUB < 10 ........................................................................... 87

4.7 Chapter Summary .................................................................................................. 89

vi

CHAPTER 5

5.1 Discussion ............................................................................................................. 92

5.1.1 Korean Teachers and Native Speakers ...................................................... 103

5.1.2 Interpretation of the Findings .................................................................... 107

5.1.3 Other Findings ........................................................................................... 114

5.2 Signiﬁcance and Implications.. .......................................................................... 121

5.3 Chapter Summary ................................................................................................ 123
CHAPTER 6

6.1 Summary and Conclusions .................................................................................. 128

6.2 Shortcomings ....................................................................................................... 131

6.3 Future Research ................................................................................................... 134

APPENDIX A: Sentence Combination Test .................................................................... 136

APPENDIX B: Sample Pictures ...................................................................................... 137

APPENDIX C: Lead-In Statements for Spoken Data ...................................................... 143

APPENDIX D: Lead-In Statements for Written Data ..................................................... 146

APPENDIX E: Consent Forms ........................................................................................ 150

REFERENCES ................................................................................................................ 152

vii

LIST OF TABLES

Table l: Pre-test scores for LPL and HPL ......................................................................... 54
Table 2: Numbers of spoken and written relative CPS ....................................................... 61
Table 3: The Binomial test result of SUB relative CPS for LPL ........................................ 68
Table 4: The Binomial test result of SUB relative CPS for HPL ....................................... 69
Table 5: Frequencies and Crosstabulations for SUB relative CPS ..................................... 71
Table 6: The )(2 test for SUB relative CPS in the Spoken data ........................................... 72
Table 7: The x2 test for SUB relative CPS in the written data ........................................... 73
Table 8: The Binomial test result of DO relative CPS for LPL .......................................... 74
Table 9: The Binomial test result of DO relative CPS for HPL ......................................... 75
Table 10: The Binomial test result of IO relative CPS for LPL ......................................... 78
Table 11: The Binomial test result of 10 relative CPS for HPL ......................................... 79
Table 12: Numbers and percents of relative CPS for all L2 participants ........................... 80
Table 13: Numbers of tokens that produced both types of relative CPS ............................ 81
Table 14: Differences of the numbers of SUB and DO relative CPS ................................. 82
Table 15: The McNemar test result for SUB and DO relative CPS ................................... 83
Table 16: Differences of the numbers of DO and 10 relative CPS .................................... 85
Table 17: The McNemar test result for DO and 10 relative CPS ....................................... 86
Table 18: Differences of the numbers of SUB and 10 relative CPS .................................. 87
Table 19: The McNemar test result for SUB and 10 relative CPS ..................................... 88

viii

Table 20: Numbers of spoken and written relative CPS (repeated) .................................... 94

Table 21: Numbers of relative CPS of all learners ............................................................. 97
Table 22: Comparison of two inferences for SUB and DO relative CPS ........................... 98
Table 23: Numbers of relative CPS from Korean teachers and native speakers .............. 104

LIST OF FIGURES

Figure l: Phrasal structures of CP, IP, and VP ..................................................................... 7
Figure 2: L2 phrase structure under the MT hypothesis .................................................... 10
Figure 3: Development of CP projections under the MT hypothesis ................................. 11
Figure 4: Agreement between Spec-CP and Co ................................................................. 14
Figure 5: L2 phrase structure under the FT/FA hypothesis ................................................ 16
Figure 6(a): Morphosyntactic features in the relative CP .................................................. 24
Figure 6(b): Syntactic dependency relation in the relative CP ........................................... 24
Figure 6(c): Semantic predication in the relative CP ......................................................... 25
Figure 7: DO relative CP structure in Korean .................................................................... 36
Figure 8: Numbers of SUB and DO relative CPS for all learners .................................... 100
Figure 9: Numbers of SUB, DO, and 10 relative CPS of all learners .............................. 102
Figure 10: Numbers of relative CPS of Korean teachers ................................................. 105
Figure 11: Numbers of relative CPS of native Speakers ................................................... 106
Figure 12: Extracting positions in the SUB, DO, and 10 relative CPS ............................ 109

INTRODUCTION

The role of ﬁrst language (L1) grammar has been an extensive topic of research in
the acquisition of a second language (L2). The goal is to examine the extent to which
L2 acquisition is constrained by the linguistic principles and parameters of UG
instantiated in learners’ L1. In examining this problem, two models found in the
literature on this topic have been proposed. While both agree on the crucial role of L1-
instantiated UG in the development of L2 acquisition, they have somewhat different
perspectives in terms of the degree of availability of the L1 grammar.

The partial access model claims that only the principles and parameters of U0
instantiated in L1 are available to L2 learners. The full access model, on the other hand,
claims that the entire UG is available to L2 learners, just as with L1 acquisition.
Neither model is, however, able to account fully for the data which Show a dissociation
of phrase structure and its relevant morphosyntactic features, reported in Hawkins and
Chan (1997), Lardiere (1998a, 1998b), and Haznedar (2003).

Data from the complementizer phrase (CP) in the relative clause (relative CP)
present supplemental evidence that the acquisition of phrase structure and the
acquisition of morphosyntactic features are independent of each other. Such evidence
can also lend support to one model over the other, and may lead to a better

understanding of the acquisition of L2 grammar.

With respect to the two conﬂicting models, one issue that has received a
considerable amount of attention is the availability of L1 functional projections such as
the inﬂectional phrase (IP) and complementizer phrase (CP)l at the initial state of L2
acquisition. Of the two speciﬁc L2 models on that issue, one is the Minimal Trees
(MT) hypothesis (partial access model for Ll-instantiated UG) (Vainikka and Young-
Scholten 1994, 1996a, 1996b, 1998), which claims that “only lexical categories are
present in the earliest stages of both L1 and L2 acquisition, and during that acquisition,
functional projections develop in succession” (1996a: 7). The other is the Full
Transfer/Full Access (FT/FA) hypothesis (full access-to-UG model) (Schwartz and
Sprouse 1994, 1996), which claims that “the initial state of L2 acquisition is the ﬁnal
state of L1 acquisition” (1996: 40). That is, the entire L1 grammar including all
functional projections is readily available as the base of L2 acquisition.

In this dissertation, the data give evidence that morphosyntactic features in the
relative CP develop from [—wh, +pred] to [+wh, +pred]. This feature development is
independent of L1 phrase structure, which is available as a block. One may, therefore,
reach the generalization that the phrase structure of other types of CP that typically
appear at earlier stages of L2 acquisition is also fully available as a block. Necessary

morphosyntactic features, however, develop from [—] to [+].

 

' The determiner phrase (DP) as a functional category is not an issue in this study. Vainikka and Young-
Scholten (1996a: 31) claim that “since argument structure is not stated in terms of DPS, but rather in
terms of theta-roles, a DP need not be present from the earliest Stages of acquisition.”

The same data are evidence that the developmental pattern of the morphosyntactic
features in the relative CP observes the order anticipated by the Noun Phrase
Accessibility Hierarchy (NPAH, Keenan and Comrie 1977).

With respect to the investigation of the availability of functional categories,
previous research has focused on the types of CP structures that were typically elicited
from the data of natural speech in yes-no questions, wh-movement, and subordinate
clauses. These necessitate a CP, which can appear in earlier stages of L2 acquisition.
Among the structures that also contain a CP is the restrictive relative clause, although
the CP found in this type of clause is different from that found in questions or
declaratives2 because it involves different syntactic relationships and semantic co-
indexation within the phrase and between the matrix and relative clauses. Arguably, the
early acquisitional characteristics of the relative CP typically emerging at a more
advanced state of L2 acquisition elucidate another facet of the developmental trend of
CP acquisition. This fact is concerned with more deﬁnite and critical evidence for the
growth of L2 phrase structure in terms of the development of morphosyntactic features.

The study presented here is organized as follows. In Chapter 1, the major claims of
the two hypotheses, both of which are concerned with the issue of the early presence
of L1 ﬁanctional projections, are introduced in detail. This study opposes the MT

hypothesis and favors the FT/FA hypothesis. The Dissociation Hypothesis claims that

 

2 A typology of CPS and syntactic structures and morphosyntactic features of different CPS are
discussed in detail in Chapter 1.

the development of morphosyntactic features is dissociated from the acquisition of
phrase structure, which is already entirely available as a block at the onset of language
acquisition. In the next section dealing with syntactic structures of CPS, a typology of
CPS (cf. Rizzi 1990) is summarized. The syntactic structure of relative CPS in English
is discussed from the perspective of the interpretive rule [Make OP] proposed by
Munn (1994) and the rule of predication.

The next section introduces the structure of the relative CP in Korean. The cross-
linguistic comparison leads to a view that the relative CP in English is underlyingly
identical to that in English, but differs from English with regard to the existence of an
optional morphosyntactic features [+wh, +pred] in English. The examination of the
motivation for this study follows. One key motivation is that new data are needed to
better justify the earlier claims on the acquisition of L2 functional categories that were
based on data from natural speech in yes-no questions, wh-movement, and subordinate
clauses, all structures which contain sorts of CP. Previous studies, however, failed to
report on much data from relative CPS, possibly because complex structures like the
relative CP are only rarely produced in natural speech and appear at a later stage of
language acquisition.

Chapter 2 presents case studies based on the two contrasting models mentioned
above: Vainikka and Young-Scholten’s work on the MT hypothesis, and Schwartz and

Sprousc’s work on the FT /FA hypothesis. Subsequent controversial claims made by

each of these two hypotheses are pointed out. The dispute, however, seems still
unsettled. Studies on the Dissociation Hypothesis (Hawkins and Chan 1997; Lardiere
1998a, 1998b; Haznedar 2003), which substantially clarify the dispute between the two
hypotheses, are also discussed.

In Chapter 3, this study proposes that crucial data to support the Dissociation
Hypothesis can be found in English relative CPS produced by adult Korean learners.
The following sections contain relevant research questions, research hypotheses, and
research methodology. The data are collected by a cross-sectional elicited production
test using the picture-description instrument to obtain sufficient relative CPS within a
Single experiment. The production outputs are in the forms of Speaking and writing.

In Chapter 4, the statistical results for the research hypotheses are interpreted by
using various nonparametric statistical techniques.

In Chapter 5, the results are discussed in the context of the current research
questions in this study. The data of learners, teachers, and native Speakers, which are
examined in some detail here, provide support for the view that the Dissociation
Hypothesis iS an empirically explanatory model that accounts for the availability of L1
grammar for L2 acquisition. Other relevant ﬁndings from the data are summarized. A
variety of hybrid relative clauses that may reveal unknown properties of UG in human
language are investigated. The theoretical Signiﬁcance and pedagogical implications of

these investigations are pointed out.

Chapter 6 presents conclusions, mention the shortcomings in this experimental

study, and suggest needs for future research.

CHAPTER 1

1.1 Signiﬁcance of the Study

One issue that has received a great deal of attention in the ﬁeld of second language
(L2) research is the inﬂuence of ﬁrst language (L1) grammar on functional categories
such as the inﬂectional phrase (IP) and complementizer phrase (CP) in the early stages
of L2 acquisition (Lakshmanan and Selinker 1994; Schwartz and Sprouse 1994, 1996;
Vainikka and Young-Scholten 1994, 1996a, 1996b; Hawkins and Chan 1997; Lardiere
1998a, 1998b; Prévost and White 2000; Bhatt and Hancin-Bhatt 2002; Haznedar 2003).

Figure 1 presents a tree diagram illustrating the structural relationship of these phrases.

Figure 1: Phrasal structures of CP, IP, and VP

.-
.....
‘-
0‘.

An on—going dispute about the availability of CP at the onset of L2 acquisition has

been carried on, and two models have claimed rival but distinct viewpoints.

Theoretical and empirical shortcomings, however, exist in both models. A different
approach, such as investigation of the CP in the relative clause (hence, relative CP), iS,
therefore, needed. Relative clauses typically do not appear until later stages of L2
acquisition, but studying them may help us better understand the stages in the
development of previously acquired CPS and their morphosyntactic features such as
[:twh] for wh-movement and [ipred] for predication.

This dissertation argues against the Minimal Trees (MT) hypothesis (Vainikka and
Young-Scholten 1994, 1996a, 1996b), which claims that the functional categories IP
and CP gradually develop in L2 after the acquisition of the L1 lexical category VP.
That is, only the VP bearing the lexical items (therefore, Minimal Trees) is transferred
from L1 grammar, whereas functional categories like CP emerge as learners are
subsequently exposed to L2 input, which interacts with X’-Theory.

The opposing Full Transfer/Full Access (FT /FA) hypothesis (Schwartz and
Sprouse 1994, 1996) assumes that the entire Ll grammar, including the functional
categories IP and CP and their complement projections, is readily available as the base
of L2 acquisition.

One major inadequacy of both models is that they fail to distinguish between the
acquisition of phrase structure and the acquisition of morphosyntactic features. The
availability and the development of CP, including the relative CP, are still debatable

(Lardiere 1998a, 1998b, 2003; Schwartz 1998; Vainikka and Young-Scholten 1998;

Prévost and White 2000; Selinker and Lakshmanan 2001; Bhatt and Hancin-Bhatt
2002; Haznedar 2003). Overcoming that inadequacy can prove that, in the initial state
of L2 acquisition, morphosyntactic features may be absent, while the entire phrase
structure is available as a block. The morphosyntactic features gradually develop from
the absence of a feature to the presence of the feature.

In the following sections, both models are described in greater detail. The last
section discusses an alternative approach to the understanding of the acquisition of the
relative CP, an approach based on the assumption that syntactic and morphological

developments are independent of each other (Lardiere 1998a).

1.1.1 Minimal Trees (MT) Hypothesis

The MT hypothesis, based on the Weak Continuity Approach in L1 acquisition
(Radford 1988, 1990; Clahsen, Eisenbeiss, and Vainikka 1994), assumes that learners
make use of the fewest possible positions and projections needed to analyze the input
data, and no more (Vainikka and Young-Scholten 1994). The lowest possible position
and projection needed for learners, according to Vainikka and Young-Scholten, is the

VP as shown in Figure 2.

Figure 2: L2 phrase structure under the MT hypothesis

CP

/\

31"” A

C l/P\ ,,,,,,
L2 functional cate ones /\n/L\'\;" L1 lexical category

- L2 inﬂectional phrase

/'/V\- L] verb phrase

- L2 complementizer phrase

Vainikka and Young-Scholten claim that the functional projections of IP and CP
develop as learners identify a potential head of a phrase that is situated higher than the
VP. Once the head is identiﬁed from the input data, learners assume an argument
position provided by X’-Theory. They further claim that learners use only the
knowledge of L1 VP to establish the structural basis for L2 acquisition. After this point,
functional categories like IP and CP are situated higher in the syntactic tree projection
through the interaction of X’-Thcory with the input data. A variation of the target-like
grammar, an interlanguage, is due to the competition between the grammar based on
higher functional categories (IP or CF grammar) and that based on VP (Vainikka and
Young-Scholten 1996a).

The MT hypothesis, however, does not specify how relevant morphosyntactic
features in functional categories like CP mature. For example, what cannot be

determined under the hypothesis iS whether the morphosyntactic features [—wh] and

10

[+wh] mature in sequence or emerge together. One of the features can be a default and
the other an option, or both may be defaults. The MT hypothesis does not mention that
issue. Rather, the hypothesis implicitly assumes that the features develop when the
head of a phrase Situated lower in the syntactic tree develops prior to its Speciﬁer as

Shown in Figure 3.

Figure 3: Development of CP projections under the MT hypothesis

CF

I VP 1 VP

Moreover, the claim that an incomplete sentence produced from the VP grammar
competes with the grammar from a higher functional category, although plausible,
cannot be empirically proven. Such shortcomings will be resolved with the alternative
approach discussed in the last section of the chapter. Let us now examine in greater
detail the development of L2 functional categories assumed under the MT hypothesis.

The MT hypothesis (Vainikka and Young-Scholten 1996a) assumes the following
four sequential stages of syntactic development: (1) bare VP, (2) Finite Phrase (FF), (3)

AgrP, and (4) CP. At the initial VP stage, learners transfer their L1 VP, but not any of

11

the L1 ﬁmctional categories. Verb movement or overt production of auxiliaries and
modals are absent. Agreement and tense morphemes in [P are lacking. Therefore, yes-
no questions or wh-questions involving a CP are absolutely unimaginable. Learners
initially use the literal L1 word order, but switch to the target order when exposed to
negative evidence.

The F P stage is characterized by the presence of one functional category, a low IP-
level projection like aspect or tense, which provides a position for a raised verb as well
as a position for auxiliaries and medals. The production of yes-no questions or wh-
questions with a CP is still not possible, but optional verb raising is evident. The option
for apparent verb raising brings about two competing grammars: one with verb raising
and one with the verb still in VP. Agreement and tense morphemes in [P are still
underspeciﬁed so that learners still need to determine which Speciﬁc person, number,
and tense features must be assigned to which functional heads.

The AgrP stage involves the speciﬁcation of full agreement and tense features.
Verb raising is frequent, and auxiliaries and medals are common. Agreement and tense
features are acquired. Some embedded clauses with a complementizer are apparent,
and complex wh-questions can be produced. The status of CP at the AgrP stage is,
however, not clear, and learners seem to be in the process of acquiring a CP.

Finally, the CP projection gradually emerges independently of learners’ L1, and

learners start to produce subordinate clauses, yes-no questions, or wh-questions at the

12

CP stage. If learners, however, violate the rather strict implicational stages during the
acquisition of phrase structure as mentioned above, the MT hypothesis faces a critical
chaHenge.

In opposition to the MT hypothesis, Schwartz (1998) raises thorny questions
which could reveal a violation of strict implicational stages. For example, a
subordinated CP complement He thinks that Jane is pretty which appears in earlier
stages of L2 acquisition cannot be explained under the MT hypothesis because an L1
CP complement would not be permitted to transfer until the last stage. Lardiere
(1998a: 20) also argues against the MT hypothesis, contending that “there is no direct
correspondence between an afﬁx (morphological feature) and syntactic phrase
structure.” Accordingly, Haznedar (2003: 5) criticizes the strict criterion for
implicational stages assumed under the MT hypothesis and raises the question of
“whether overt production of morphemes associated with a functional category is
necessary for one to conclude that the category is acquired.”

On the other hand, Bhatt and Hancin-Bhatt (2002: 366) extend the MT hypothesis
and propose the Structural Minimality hypothesis, which claims that “at the initial state,
all the features (and their strengths) and categories relevant to the IP-system are also
borrowed from L1.” Learners then gradually re-analyze the input data to accommodate
the structure. That the 1P emerges much earlier in the developmental stage than the CP

is also claimed in Kaplan (1993) and Lakshmanan and Selinker (1994). Bhatt and

13

Hancin-Bhatt (2002) further assume that different aspects of the CP projections emerge
at different stages, and that learners opt for the most economical way to acquire a CP.
This choice projects Spec-CP only after acquisition of the head C°. Such a claim,
however, raises a serious theoretical objection: Spec-CP must be simultaneously
projected with C°, because C0 of the relative CP is always projected with an
unspeciﬁed feature which needs to attract a lexical item with relevant features to Spec-
CP, in order to check the unspeciﬁed feature of Co (Radford 1997) as illustrated in

Figure 4.

Figure 4: Agreement between Spec-CP and C°

CP

which
[ith L§C° /\

[iwh]

Hence, Bhatt and Hancin-Bhatt’s claim (ibid: 360) that “the Speciﬁc components
of CP emerge at different levels of language learning experience” seems theoretically
invalid for the case of relative CP. This dissertation argues against their claim, because
once the lexical items are selected to produce a relative CP, both Spec-CP and Co must
be readily available. What they Should have concluded, however, concerns the

emergence of morphosyntactic features of CP such as [-wh, +pred] and [+wh, +pred]

(Rizzi 1990), not projections of CP.

To summarize, according to the MT hypothesis and its later version, the Structural
Minimality hypothesis, CP is absent in the earliest stages of L2 acquisition, and
different types of CPS (e.g. yes-no question CP, wh-question CP, subordinate clause CP,
relative CP, etc.) emerge over time. The gradual emergence of CP projections and
relevant features under the MT hypothesis and the Structural Minimality hypothesis
seems to make speciﬁc and researchable predictions about developmental stages in the
acquisition of CP.

A serious theoretical modiﬁcation must, however, take place to correspond to
empirical data which show a full CP projection in the early stages of L2 acquisition
(Schwartz and Sprouse 1996; Sprouse and Schwartz 1998; Lardiere 1998a; Prévost
and White 2000; Haznedar 2003). The following hypothesis might offer a more

promising, though still insufﬁcient, explanation of the L2 acquisition of CP.

1.1.2 Full Transfer/Full Access (FT IFA) Hypothesis

The FT /FA hypothesis, based on the Strong Continuity Approach of full transfer
and full access to Universal Grammar (UG), claims that all the principles and
parameters instantiated in learners’ L1 grammar transfer as the initial state of an L2
(Lakshmanan 1993/1994, 1998; Lakshmanan and Selinker 1994; Schwartz and

Sprouse 1994, 1996; Epstein et a1. 1996, 1998; Hawkins and Chan 1997; Schwartz

15

1998; Haznedar 2001, 2003). Figure 5 illustrates the initial projections of phrase

structure in L2 acquisition under the FT/FA hypothesis.

Figure 5: L2 phrase structure under the FT /FA hypothesis

CP

/\

3"“ A

C lP

L1 functional categories o/l\\ Ll lexical category

 

- L1 inflectional phrase 1 ,'/VP\' L 1 verb phrase
- L1 complementizer phrase I,’ A
v

On ﬁrst exposure to L2 input data, the Ll-instantiated L2 grammar starts to
change when the current grammar cannot generate the intended target phrase structure.
The change results from a failure to assign a relevant structure for the data and forces
restructuring dependent on the options of UG. Therefore, the L2 developmental course
is partly determined by Ll grammar, L2 input, and UG options. That is, learners use
their entire Ll knowledge to establish the structural basis for L2 acquisition, even for
the production of wh-questions or relative CPS, which involve the highest functional
category CP. When the interlanguage grammar fails to provide the intended structure,

the options of UG guide learners, but individual differences determine choices among

16

the options and the ways to make the choice.

The FT/FA hypothesis (Schwartz and Sprouse 1994) suggests the following three
successive stages of a developmental pattern analyzed ﬁ'om the data by an adult native
Speaker of Turkish acquiring German as an L2: Stage I: XSV<+rm>, Stage 2:
XV<+ﬁn>S<+pmn>, and Stage 3: XV<+ﬁn>S<ipmn> At the ﬁrst stage, what follows
V<+rm> are unattested, and the most common structure is a non-subject initial structure
(X)SV<+rm>, in which X is usually adverbial. The verb undergoes head movement to
the head of AgrP (AgrPo) and then ultimately to C°. The subject originates inside the
VP, and raises to Spec-CP via Spec-AgrP. The motivation for movement to Spec-CP is
for nominative Case checking of the subject by way of a Spec-Head conﬁguration with
the verb settled in C°.3 The sentence- initial adverbial is base-generated in the extended
CP. Their analysis may partially explain the presence of functional projections,
although a paradigm of agreement and tense features is not fully speciﬁed in the early
stages of L2 acquisition.

The next stage is characterized by a different kind of non-subject initial structure
(X)V<+ﬁn>S<+pron>, which is the inversion of the previous stage, but only when the
subject is a pronoun. At this stage, X can be an adverbial phrase, a prepositional phrase,

or a direct object. The verb raises to Co via AgrPo, and the subject pronoun

 

3 According to Schwartz and Sprouse‘s data, it is evident that the subject moves to Spec-CP and the

verb raises to C°. lt, however, cannot be for nominative Case checking because nominative Case is
checked by Tense, independent of whether the subject is moves to C or not (Chomsky, 1995; Radford,
1997).

17

incorporates as a clitic to the ﬁnite verb in C° for nominative Case checking. Finally,
at the last stage XV<+ﬁn>S<ipmn>, learners are capable of inverting the verb with a non-
pronominal subject as well as with verb-initial declarative utterances.

Under this claim, all functional categories including CP are expected to be
available at the onset of L2 acquisition. If L2 data showed that the functional
projections gradually emerge in learners’ L2 knowledge, the FT /FA hypothesis would
lose its standing. If learners start to utilize the full-ﬂedged functional projections from
the earlier stages of L2 acquisition, the data of such projections can support the FT /FA
hypothesis, which this dissertation assumes.

Proponents of both models, with their supporting data and re-analyzing each
other’s data to support their own hypotheses, have continued to maintain their
respective opinions (Schwartz and Sprouse 1996; Vainikka and Young-Scholten 1996a).

Neither theory is, however, able to differentiate the acquisition of morphosyntactic
features from the acquisition of phrase structure. An alternative approach must be
sought. Thus the next section discusses an alternative to the two hypotheses. It
generally acknowledges the presence of the entire L1 grammar as well as the highest
functional category CP. This claim amounts to supporting the FT /FA hypothesis, but
further assumes that what gradually develops is not phrase structure per se, but the

morphosyntactic features associated with phrase structure.

1.1.3 Alternative Approach

In this section, an alternative approach to the two previously mentioned
hypotheses is discussed based on the pioneering works of Lardiere (1998a, 1998b).
Lardiere (1998a) argues for a distinction between abstract morphosyntactic features
and their syntactic representations, and she (1998b: 362) further proposes that there is
“a dissociation of syntax from morphology in L2 acquisition.” This approach supports
the FT /FA hypothesis by supposing that all the functional projections such as CP must
be available at the onset of L2 acquisition and that failure to produce the intended
structure may be attributed to a failure to map morphosyntactic features. Put another
way, the fully matured phrase structure for L2 acquisition exists ﬁ'om the beginning,
where the command of mapping the relevant features [—] and [+] is obscure.

In hypothesizing the developmental pattern for the relative CP consistent with this
alternative approach, it can be assumed that the relative CP structure always exists. In
addition, learners are assumed to have acquired other Simpler CP projections and their
morphosyntactic features containing Spec-CP and C0 from the earlier productions of,
for example, yes-no questions, wh-questions, subordinate clauses, etc. With the CP
projections already acquired, later in the developmental stage, depending on the
features acquired, learners may produce different types of relative CPS4 which literally

have the same meaning; e. g. the book which I have, the book t_h_qt I have, and the book

 

 

" A typology of CPS and their morphosyntactic features are discussed in the next section.

19

g I have. The lexical item which involves [+wh] and that and a need [-wh]. If, in
language development, a marked feature is acquired after an unmarked feature, high
proﬁciency L2 learners (HPL) may produce a greater proportion of [+wh] relative CPS
than low proﬁciency L2 learners (LPL), as HPL overcomes the difﬁculty of mapping
the feature [+wh].5

In such a case, the result neatly supports the emergence of CP projections before
the acquisition of speciﬁc features associated with them. The phrase structure must,
however, be available as a block for feature checking between Spec-CP and C°. This
assumption lends support to the FT/FA hypothesis. It ﬁtrther assumes that the
acquisition of speciﬁc features of a relative CP is independent of its phrase structure
projections. Lardiere (1998a, 1998b), therefore, supports the view that morphological
features are dissociated from syntactic development, and Haznedar (2003: 29) claims
that “the acquisition of L2 functional categories overcomes the mapping problems that
learners face in associating the surface form with underlying abstract features.”

Overall, I assume that the alternative approach sketched here better explains the
initial state and the development of relative CPS, which, in turn, may offer a better
understanding of previously acquired CPS. Moreover, I hypothesize that HPL produces

a greater proportion of [+wh] relative CPS than LPL.

 

5 Lakshmanan and Selinker (1994) also report the evidence from two adolescent L2 learners of English
that there is a developmental link between the realization of [—wh, —pred] and the realization of [-wh,
+pred].

20

Summing up from the alternative approach, I propose the Dissociation Hypothesis:
The development of morphosyntactic features is dissociated from the acquisition of
phrase structure, which is already entirely available as a block at the onset of the
acquisition of an L2.

The next section provides an illustration of different types of phrase structure that
contain a CP. Afterwards, a typology of CPS with distinct Speciﬁcations is discussed,
as proposed in Rizzi (1990). Movement and predication in English relative CPS are
then presented. A discussion on the Korean relative CP follows, since the intended data
are collected from the learners who Speak Korean as an L1 and acquire English as an

L2.

1.2 Syntactic Structures of CPS

CP is a functional category that contains a complementizer or an auxiliary in C°
and/or an operator in Spec-CP, which selects a sentence in IP (Radford 1997). Previous
research on the early state of L2 acquisition was typically based on CP-containing

structures such as the following:

(1) a. She thinks [cp [c that [IP she is pretty]]] (that-subordinate clause)
b. She thinks [cp [c e [[13 she is pretty]]] (a-subordinate clause)
c. [CP 0F [(3 Will; [11) you t,- work tomorrow]]] (yes-no question)
(1. [cp What,- [c will,- [11) you t,- buy tj tomorrow]]] (wh-question)

21

Each of the IP clauses in (1) is a complement clause of C°. An IP functions as the
complement of the Co immediately including it. In (la) a CP iS formed when the
based-generated complementizer that" in C0 merges with the IP, and Spec-CP in this
case is empty. In (lb) the null complementizer a subcategorizes a ﬁnite complement
clause (Lakshmanan and Selinker 1994)7 as does the overt complementizer that in
(la). The syntactic properties of a type of complement clause are encoded not in the
lexical item, but in the phrase structure. Spec-CP is also empty in (1b). CP can also be
ﬁlled by a question operator and an auxiliary or modal as in (lo) for yes-no questions
and as in (1d) for wh-questions. In (1c) the auxiliary will has adjoined to Co via head
movement from the head of IP,8 as motivated by unspeciﬁed features (Lasnik 1995) of
the based-generated null question operator GP in Spec-CP and C°. On the other hand,

in (1d) Spec-CP is occupied by a wh-phrase what acting as an overt question operator

 

" On a historical linguistic claim that that is a complementizer, refer to Van der Auwera (1984, 1985)
and Seppanen (1997). On a pragmatic perspective of that, refer to Cheshire (1996).

7 The null complementizer a can also select for a non-ﬁnite [—Tense] complement clause as well, as in
John preferred (for) Bill to leave for for deletion (Munn, personal communication).

8 Actually, when we say that an Aux is moved to C°, we do not exactly mean that C" contains the Aux.
It is assumed that there is a null C° to which the Aux adjoins. See the tree below.

A

Spec /C’\
/Co\ /P\\
I0 C0 /\

| F VP

22

raised from the verb complement Site, and the auxiliary will has adjoined as it does in
(1c).
Another type of CP is the relative CP as in (2) below, which has different

morphological and semantic structures from other CPS in (l).

(2) the cup [CP a/which/that [11) the girl is holding]] (relative CP)

The relative CP in (2) has never been a target structure for understanding the early
state of L2 functional category development (cf. Hawkins and Chan 1997) mainly
because earlier studies could not have acquired sufficient (especially production) data
containing relative CPS. Such complex syntactic structures as relative clauses occur
only infrequently in both L1 and L2 spontaneous Speech, especially from children and
beginning adult L2 learners (Crain and Thornton 1998). Nevertheless, the study of
relative CPS can provide valuable evidence for the development of morphosyntactic
features and for the syntactic dependency relation and semantic predication in CP, as

shown in Figures 6(a-c) for a direct objective relative CP.

23

Figure 6(a): Morphosyntactic features in the relative CP

CP

/\

9

Spec /C\

('3 IP
[—-wh, +pred]
[+wh, +pred] /l\

v NIP
l-wh, +pred]

[+wh, +pred]

Figure 6(b): Syntactic dependency relation in the relative CP9

 

Movement to Spec-CP

 

 

9 A detailed explanation about the structure of relative CPS is dealt in Section 1.2.2.

24

Figure 6(c): Semantic predication in the relative CP

 

 

/NP\
NP,‘ CP
. OP; C'
§/\ 0/\
f c ,~ 1P
..................... l’
Re-indexaticn (NP,- H C) V,
V/\NP
l
Co-indexation (NP,- H OF,- H NP,-) t,’

 

 

Findings from the current study support the Dissociation Hypothesis, which claims
that L2 phrase structure exists from the onset of L2 acquisition and that its
morphosyntactic features gradually mature. For this reason, the study on L2 relative
CP acquisition may promise to provide insight into a better understanding of how UG
determines syntactic structures. It is apparent that, without the data from relative CPs,
previous acquisition studies on CP are incomplete and unsatisfactory. The next section

discusses a typology of CPS as proposed in Rizzi (1990).

1.2.1 A Typology of CPS
Rizzi (1990) introduces a feature system which provides a partial speciﬁcation of

different types of C0 in yes-no questions, constituent questions, wh-questions,

25

subordinate clauses, relative clauses, etc. The feature system consists of [—] and [+] of

wh-operator and predication, as follows:

(3) [iwh, ipred]

These features are base-generated and speciﬁed on C0 and make possible four different
types of CP in terms of morphosyntactic features: [—wh, —pred], [+wh, —pred], [—wh,
+pred], and [+wh, +pred].

A C0 with [—wh] does not allow movement of a wh-phrase from the inside of IP to
Spec-CP. Therefore, a subordinate clause or a relative clause generated by movement
of a wh-phrase such as what, which, to whom, etc. cannot occur; that is, no overt
operator can be moved to Spec-CP. When [—wh] is speciﬁed with {—pred], C0 can only
head a non-predicated clause such as tensed subordinated clauses. In that case, no
element such as the trace of a moved constituent is available to be semantically co-
indexed with Co and the head noun in the matrix clause. Any sort of predication is
impossible. Therefore, in I think that she is pretty as in (la) and 1 think she is pretty as
in (1b), C0 has [-wh, -pred], which cannot assign a question operator to Spec-CP nor a
predicated clause as a complement, because the two negative features in Co Specify
declaratives. [+wh] is obligatory for questions. When [—pred] combines with [+wh], a

possible case of CP is a question which involves question operator movement from the

26

inside of a non-predicated IP to Spec-CP. Examples of this are found in sentences like I
know where you live and wh-questions as in (1d) and yes-no questions as in (lc),
which contains a base-generated null question operator in Spec-CP.

The feature [+pred] is distinctly exclusive for the relative CP. It allows Co to
license a relative operator'0 in Spec-CP. Hence, Co with [—wh, +pred] heads a relative
clause with a null operator moved from the inside of relativized IP to Spec-CP. This
results in the a-relative CP with a null operator in Spec-CP and a null relative
complementizer in Co or the that-relative CP with a null operator in Spec-CP and the
complementizer that in C°. Rizzi (1990: 57) claims, supposedly in UG, that “[—wh] Co
is obligatory, while [+wh] C0 is optional.” In that case, Co with [+pred] heading a
relative CP can optionally retain [+wh] to attract an overt relative operator from the
inside of relativized IP to Spec-CP. This possibility generates the wh-relative CP.
Summarized in (4) are the three possible cases of relative CPS, as in the typology of

CPS illustrated above:

(4) a. [-wh, +pred] a-relative CP: the cat [CP 0,- [c e [[p the girl is kissing t,-]]]
b. [—wh, +pred] that-relative CP: the cat [CP 9; [C that [11) the girl is kissing t;]]]
c. [+wh, +pred] wh-relative CP: the cat [CP which,- [c 0 [[p the girl is kissing t;]]]

 

'0 According to Rizzi (1990: 57ft), the relative operator in the relative CP is different from the null
operator which is base-generated in Spec-CP for yes-no questions as in (1c) or the overt wh-operator
raised from the position of the complement for wh-questions as in (1d). The relative operator is,
however, often simply referred to as ‘operator’ in this dissertation.

27

The next section discusses both the movement and the rule of predication in
English relative CPS in (4). The subsequent section reviews the structure of the Korean
relative CP. This dissertation examines research questions concerning whether learners
at different levels of proﬁciency produce a signiﬁcantly similar or different proportion

of [—wh] or [+wh] relative CPS.

1.2.2 Movement and Predication in English Relative CPS

The biclausal structure the cat which the girl is kissing stole a ﬁsh contains a
relative clause which the girl is kissing that is headed, in this example, by a relative CP
with an operator phrase containing which; ‘operator’ refers to the relative pronoun
which. Cinque (1982: 271) says that “the dual nature of the relative operator serves as
both an element binding the variable and an anaphor to the relative head.” Clause-
internally, the operator creates a relation by forming an operator-variable dependency
relation and interprets its syntactic and semantic information stranded in the variable.
Clause-extemally, the dependency relation yields morphosyntactic agreement with the
relative head. None of these relations are optional; they are all absolutely obligatory
(Chomsky 1982). Once there exists a relative clause as an open predicate, then the
variable must naturally seek an operator in the relative CP, or vice versa, to form a

dependency relation, which, in turn, must agree with the relative head in the main

28

clause.ll

The relative CP structure at issue ﬁindamentally concerns the relative operator
making its morphosyntactic agreement with C°. The following are the three types of
relative CPS being considered in this dissertation, in order to examine whether learners
acquire the default feature [—wh] and the optional feature [+wh] in the relative CP at

once or sequentially.

(5) a. Subject relative CP: the cat which _ is eating a ﬁsh
b. Direct object relative CP: the cat which the girl is kissing _
c. Indirect object relative CP: the cat which the girl is giving a ﬁsh to

(5a) contains a subject (SUB) relative clause with the gap (i.e., the position of the
variable) acting as the subject of the relative clause. (5b) and (5c) have an object
relative CP. (5b) contains a direct object (DO) relative clause with the variable in the
direct object position. The syntactic object is neither the relative head cat nor the wh-
operator, but the intangible variable which has the semantic role of the object. This
role preserves the local structural relations, while sensibly deﬁning the object as ‘Sister
to V’. (5c) contains an indirect object (10) relative clause with the variable acting as

the indirect object of the verb of the relative clause. According to previous accounts,

 

” Wiltschko (1998: 169) States this as the Extended Bijective Principle: “There is a bijective
correspondence between an operator-variable chain and a range.” That is, “each operator must A’-bind
exactly one variable, and each variable must be A’-bound by exactly one operator, and for each operator-
variable chain, there must be exactly one range” (cf. Chomsky 1982).

29

the wh-operator as an argument is inserted into the variable position in the relative
clause, and then, subsequently moves to Spec-CP headed by C°. In any case in (5a-c),
the wh-operator forms a dependency relation with the variable, and further makes
morphosyntactic agreement with the relative head and C° in the relative CP.

A question that arises immediately is the motivation for movement in the relative
clause when the wh-operator could, in fact, ﬁJIﬁII the variable’s functions in situ, and
not ﬁ'om the dislocated, more distant position Spec-CP. Under the feature checking
theory in the minimalist framework, the sole role of lexical items like the wh-operator
is to license and delete uninterpretable ﬁinctional features that attract compatible
lexical features for checking. Such uninterpretable functional features are the
triggering mechanisms for movement (Chomsky 1999). The uninterpretable feature
setting off wh-operator movement as in the relative CP is, following Rizzi (1990),
[+pred] C°. Rizzi claims that [+pred] C° licenses an uninterpretable lexical feature of
the wh-operator and attracts it to the checking domain in the relative CP; that is, Spec-
CP. The wh-operator movement is contingent on the principle of Full Interpretation,
which necessitates intact preservation of semantically-related features as well as the
complete deletion of the operator feature in the base (Chomsky 1994, 1995).

Adopting the interpretive rule [Make OP] proposed in Munn (1994),'2 the relative

 

'2 [Make OP] is an interpretive rule that applies to the LF representation to yield a structure that is
interpretable by the semantics. It also supervises the making of GP in Spec-CP and subsequent
operations (Munn, personal communication).

30

CP basically involves two sequential operations: movement and copy/deletion
processes. After wh-operator movement to Spec-CP (Move),l3 [Make OP] guides
operator dependency with the variable in the base position (Copy), deletion of non-
operator elements in Spec-CP (Delete 1),'4 and the operator copy in the base (Delete

2), as illustrated in (6a-d) for the SUB relative CP in (5a).

(6) a. the cat [cp which ] [ ] is eating a ﬁsh (Move)
b. the cat [cp which cat] [which cat] is eating a ﬁsh (Copy)
c. the cat [cp which ] [ cat] is eating a ﬁsh (Delete 1)
d. the cat [cp which ] [ ] is eating a ﬁsh (Delete 2)

In (6a) an uninterpretable feature in C° needs to attract the wh-operator to Spec-CP.
The variable of the operator is left behind in the base position. In (6b) [Make OP]
guides the wh-operator to bind the variable and to decode its syntactic and semantic
information and functions. Subsequently, the wh-operator in Spec-CP and the variable
in the base form a dependency relation. They are logically and fully interpreted as
[which cat], Showing a semantically identical dependency relation (Rizzi 1986;

Chomsky 1992; Munn 1994; Sauerland 1998; Bissell 1999; Fox 1999, 2002). The wh-

 

" Actually, it is not the operator per se that moves. Rather, it is the phrase that contains the operator,
called operator phrase (OP). '

a. the cat which the girl is giving a ﬁsh to

b. the cat to which the girl is giving a ﬁsh
In (a) the OP contains only the wh-operator which; whereas in (b) it contains the wh-operator which and
the pied-piped preposition to.
"’ [Make OP] considers unnecessary any element that comes along with the operator, and further
eliminates it. For example, in the roof of which Tom ﬁxed, [Make OP] only interprets which and deletes
the roof of in LF (Munn 1994).

31

operator in the relative CP is phonologically realized, while the variable in the relative
clause is logically and semantically interpreted. In (6c) [Make OP] deletes the
unnecessary non-operator element cat in Spec-CP. The which in the base is also
eliminated at this point, after its logical and semantic functions have been interpreted
by the which in Spec-CP. Lastly, in (6d) the undeleted variable cat in the base is
deleted because its identiﬁcation is recoverable from the relative head (Munn 1994). '5
Below in (7) and (8) are the derivations for the DO and 10 relative CPS in (5b) and
(5c). The processes are the same as the SUB relative CP in (6), but involve more

derivational constraints which may hinder successful production.

 

(7) a. the cat [CP which ] the girl is kissing [ ] (Move)
b. the cat [cp which cat] the girl is kissing [which cat] (Copy)
c. the cat [cp which ] the girl is kissing [ cat] (Delete 1)
d. the cat [cp Which ] the girl is kissing [ ] (Delete 2)
W \
(8) a. the cat [cp Which ] the girl is giving a ﬁsh to [ ] (Move)
b. the cat [cp which cat] the girl is giving a ﬁsh to [which cat] (Copy)
c. the cat [cp which ] the girl is giving a ﬁsh to [ cat] (Delete 1)
d. the cat [or Which ] the girl iS giving a ﬁsh to [ ] (Delete 2)

Williams (1980) proposes that wh-operator movement creates a copy in the

relative clause that can naturally turn into a variable, and the rule of predication

 

'5 The relative head has a certain potential range of c-commanding referents (Comrie 1989) where its
information can be identiﬁed.

32

assigns the same referential index16 from the operator to the relative head. Then, the
relative clause turns into a semantically open predicate clause with which the operator
needs to be co-referentially agreed (Chomsky 1977, 1982; Williams 1980; Fabb 1990;
Moon 1995). See (9).
(9) a. the cat,- which; whie-h; is eating a ﬁsh

b. the cat,- which; the girl is kissing whieh;

c. the cat; which,- the girl is giving a ﬁsh to which;
In (9a-c) after the wh-operator moves and binds the variable which which is left behind,
the rule of predication assigns them the same referential index from the operator. F abb
(1990), Seppéinen (1997), and Sharvit (1999a) claim that the relative head co-indexes
the variable by a mechanism of index percolation. On the contrary, Chomsky (1977),
Cinque (1982), Quirk et al. (1985), and Rizzi (1990) allege that it is not the relative
head that assigns the index to the variable, but the relative operator in the relative CP
bearing an anaphoric nature that binds the variable and Shares the same index.
Whichever the case may be, the relative head Shares a referential index with the
operator and the variable. The discussion so far touched brieﬂy on the explanation of
how the relative CP is traditionally assumed to be formed (cf. Chomsky 1977, 1992;
Cinque 1990; Lebeaux 1990, 1991; Borsley 1997; Alexiadou et a1. 2000).

Until now, I have focused only on the wh-relative CP; that is, [+wh] CP. 1n

 

'6 A referential index is not a syntactic element or a mental realization, but only a notation to indicate
the entities sharing the same properties.

33

contrast, the relative CPS in (10) do not contain an overt wh-operator; hence,

supposedly, they involve no visible'7 movement.

(10) a. the cat a the girl is kissing
b. the cat a that the girl is kissing

When relative CPS do not contain an overt wh-operator in Spec-CP, Co is assigned
[—wh], and a null operator is inserted in the variable position and subsequently moves
to CF (Kaplan and Whitman 1995).'8 That way, a dependency relation can be formed
and the variable’s syntactic and semantic information transfers to the null operator in
the relative CP. The null operator makes subsequent morphosyntactic agreement with
the relative head, and does the same for the overt wh-operator. Noticeably, the
complementizer that is present in C° in (10b). According to Rizzi’s (1990) analysis, the
semantically-related head C° undergoes an abstract agreement with the relative head.
In Saﬁr’s (1986) view, the second agreement between the complementizer that and the

relative head is a process of re-indexation, possibly, at post-LP, as in (11).

 

'7 Null operator movement is “overt” in the sense that it occurs before Spell-Out, although it is not
pronounced (Munn, personal communication).

'8 There is a different claim that the null operator and an arbitrary null variable pro are base-generated
(Chomsky 1977; Rizzi 1986; Cinque 1990; Wiltschko 1998; Bissell 1999; Komﬁlt 2000; Lee 2001).
They claim that Since Full Interpretation does not permit a free variable but a referential variable, the
null variable pro must be at once bound by a null operator (Saﬁr 1986; Chomsky 1989, 1992, 1995;
Epstein 1989; Chomsky and Lasnik 1993; Wiltschko 1998). In addition, unless there is an operator in
Spec-CP, relative clauses cannot be semantically interpreted. For this principled requirement, a null
operator is assumed to be directly base-generated in Spec-CP (Stahlke 1976; Chomsky 1977, 1982; Saﬁr
1986; Cinque 1990; McCloskey 1990; Shlonsky 1992; Moon 1995; Ariel 1999; Bissell 1999; Sharvit
1999b).

34

(11) a. the cat,- a; that the girl is kissing t,- (ﬁrst cooindexation)
b. the cat,- a that,- the girl is kissing t (re-indexation)
In (11a) the dependency relation [9,, t,~] is co-indexed with the relative head in the ﬁrst

round of co-indexation. In (1 lb) the relative head subsequently shares the index with

that in Co in the second round of co-indexation. As a result, the four entities (relative
head, Spec-CP, C°, and variable) bear the same referential index. The next section

discusses the relative CP structure of Korean.

1.2.3 Structure of the Korean Relative CP

In Korean, an SOV language, the relative CP precedes the head noun and has an

9

afﬁxal complementizer -nun,I which is affixed to the verb as the head of the relative

CP (Han 1992; Kaplan and Whitman 1995; Han and Kim 2004). -Nun has syntactic
and phonological preperties as shown below for SUB, DO, and 10 relative clauses in

(12).20

(12) a. SUB [cp 0P; [1p t,- pang-ul joaha] -nun] wonsungi
bread-ACC like C° monkey
‘the monkey that likes bread’
b. DO [cp 0P,- [1P wonsungi-ka t,- joaha] -nun] pang
monkey-TOP like c° bread
‘the bread that the monkey likes’

 

'9 The afﬁx -nun is used for the present, -n for the past, and -l for the future tense (Kim and Shin 1994).
2° For a detailed analysis of relative clauses in Korean, refer to Tagashira (1972). She argues that the
aﬂixal complementizer is a tense marker rather than a relative clause marker. Also cf. McCawley (1972),
Kuno (1974), Song (1991), Yang (1993), Lee, S.-B. (1994), among others.

35

c. 10 [cp OP,- [[9 namja-ka t,- pang-ul ju] -nun] wonsungi
man-TOP bread-ACC give C° monkey
‘the monkey that the man gives bread to’

As seen in Figure 7 for (12b), for example, the only available way to form the relative
CP in Korean is to raise a null operator generated in VP within IP into Spec-CP in the

sentence-initial position while realizing Co as man (Kang 1988).

Figure 7: DO relative CP structure in Korean

/\’

pang (bread)
1P °
. SUB I, “nun (that)

l
wonsun i-ka'
(mon ey) 1 VP [0
" ti V’ ha [present]

0

V

...-.'ks

.......................... joa- (like)

Therefore, such relative CPS as in English the cat which the girl is kissing with an
overt wh-operator in Spec-CP and no overt complementizer in C0 or the cat the girl is

kissing with a null operator in Spec-CP and no overt complementizer in Co are not
possible in Korean because lexical items corresponding to which, who, that, etc. are

missing. Keenan (1985) reports that relative pronouns are limited to languages that

36

allow postnominal relative CPS. There are no pronominal relatives in any language
which clearly present relative pronouns.21

The reason for disallowing relative pronouns can be that Korean does not allow Co
to carry the optional feature [+wh] in the relative CP, but only the default feature [—-wh],
which consequently prohibits overt wh-operator movement in that language. Further,
the afﬁxal complementizer -nun is obligatory. It cannot be absent like the
complementizer that in English DO and 10 relative CPS. Then, the most Similar type of
relative CP in surface structure between the two languages is the relative CP with the
default feature [—wh], not the optional feature [+wh], which Korean does not employ,
mentioned in Rizzi (I990). The next section presents the motivating grounds for the

current study.

1.3 Evidence from the Relative CP

One indicator of whether the morphosyntactic features in CP mature or whether
they are, instead, available as a set is the frequency of the use of relative CPS in normal
discourse. If HPL produces signiﬁcantly more [+wh] relative CPS than LPL, then we
can assume that the strength of [:twh] in the relative CP has been gradually developed

from [—wh] to [+wh]. In that case, we must accept the Dissociation Hypothesis, since

 

2' Keenan (1985) further explains that the use of a clause-ﬁnal complementizer in the prenominal
relative CPS is less common than the use of a clause-initial complementizer in the postnominal relative
CPS. A regular difference between prenominal and postnominal relative CPS concerns the form of the
main verb in the relative clause. In the former, the verb is almost always in some sort of non-ﬁnite form
as in Korean.

37

its ﬁ'amework postulates that the features gradually develop, while the entire syntactic
structure is available from the beginning. On the other hand, if both LPL and HPL
produce [—wh] and [+wh] in a similar proportion, then the result argues for the claim
that the entire syntactic structure is available horn the beginning along with the fully
matured morphosyntactic features. If this claim were true, we would have to
subsequently modify the generalization that L1 inﬂuence changes as L2 proﬁciency
increases (cf. Bardovi-Harlig 1987; Jarvis 2000), since all functional categories and
relevant features would be operative at the beginning of L2 acquisition.

The gradual development of morphosyntactic features in CP can be tested with
SUB, DO, and IO relative CPS to understand whether the expected phenomenon
occurs more or less with a certain type of relative CP than others or whether there
exists a hierarchy among different relative CPS in terms of the acquisition of
morphosyntactic features. If a ranking is discovered, it may correspond - either
directly or inversely — to the Noun Phrase Accessibility Hierarchy (NPAH, Keenan and
Comrie 1977), which predicts the acquisition of SUB before DO and DO before 10
relative clauses. Along the same lines, if HPL produces more pied-piped constructions
such as the dog to which a girl is giving a cookie than LPL, then we can also assume
that the optional feature [+wh] has been further developed to attract more lexical items
to form a more complex type of relative CP. Furthermore, it will be interesting to see

how learners behave differently in the production of relative CPS in speaking and

38

writing Situations, where the degree of monitoring time differs.

1.4 Chapter Summary

Chapter I discussed the Signiﬁcance of the study and analyzed the syntactic
structure of relative CPS in English and Korean. It was proposed that, in the initial state -
of L2 acquisition, morphosyntactic features are absent, while the entire phrase
structure is available as a block. The morphosyntactic features gradually develop from
the absence of a feature to the presence of the feature.

The Minimal Trees (MT) hypothesis claims that only the L1 lexical category VP is
present at the earliest stage of L2 acquisition. Other ﬁrnctional categories like IP and
CP develop in succession. This dissertation partly adopts the Full Transfer/Full Access
(FT/FA) hypothesis. The FT /FA hypothesis claims that all L1 functional categories as
well as the L1 lexical category are present for L2 acquisition from the earliest stage.
That is, the entire Ll grammar — including the functional categories IP and CP, and
their complement projections — is readily available as the base of L2 acquisition.

An alternative approach, the Dissociation Hypothesis based on Lardiere (1998a,
1998b), proposed that there should be a distinction between abstract morphosyntactic
features and their syntactic representations such that the acquisition of syntax and its
features are independent of each other.

This chapter also discussed different types of CPS that previous research has

39

focused upon. Based on Rizzi’s (1990) analysis, two types of relative CPS with [—wh,
+pred] and [+wh, +pred] can give proof of a developmental pattern in the acquisition
of relative CPS. The former type renders the possibility of o-relative CPS or that-
relative CPS, neither of which allow for an overt relative operator in Spec-CP. The
latter type, however, allows an overt relative operator with a wh-phrase in Spec-CP.
Lastly, this dissertation hypothesizes that learners acquire the optional feature [+wh]
aﬁer acquiring the default feature [—wh] for each type of SUB, DO, and 10 relative
CPS.

Furthermore, an acquisitional hierarchy of morphosyntactic features in CF is
expected among SUB, DO, and 10 relative CPS, and the pied-piped constructions may
be derived more ﬁom HPL than from LPL. A comparison between spoken and written
outputs may reveal some crucial similarities and differences in the developmental
pattern of morphosyntactic features in CP based on different production modes. The
next chapter reviews previous studies supporting the MT hypothesis, the FT/FA

hypothesis, and the Dissociation Hypothesis.

40

CHAPTER 2

2.1 Review of Literature

Vainikka and Young-Scholten (1994), supporting the MT hypothesis, collected
cross-sectional data from 11 Turkish and 6 Korean adults acquiring German, and
proposed that learners acquire ﬁrnctional categories in a gradual developmental pattern
that is similar to the way in which children acquire Ll; that is, VP -—» IP -> CP. By the
CP stage, learners can produce subordinate clauses, wh-movement, and yes-no
questions. Vainikka and Young-Scholten (1996a) re-analyze previous studies by Zobl
and Liceras (1994), Epstein et a1. (1996), and Grondin and White (1996), and, in so
doing, further strengthen their own argument that only lexical projections constitute
learners’ initial state, and the functional categories IP and CP develop later.

Vainikka and Young-Scholten (1996b) provide additional evidence from Romance
Speakers, and conclude once again that L1 children and L2 adults acquire syntactic
structure in much the same manner. This model can be empirically tested with a
prediction that learners will acquire functional categories in a deﬁnite direction.
Nevertheless, their claim does not explain the developmental pattern for
morphosyntactic features associated with phrase structure. In other words, what needs
to be further investigated is whether Co is acquired before or after Spec-CP or
simultaneously, and whether morphosyntactic features of CP are available along with

phrase structure projections or independently, and also with respect to the complexity

41

of different relative CP structures; e.g., SUB, DO, and 10 relative CPS.

Schwartz and Sprouse (1994), supporting the FT/FA hypothesis, collected
extensive longitudinal interlanguage data from a Turkish speaker, Cevdet, learning
German. They conclude that Cevdet’s initial state of L2 acquisition of German was his
ﬁnal state of L1 acquisition of Turkish, and that each stage in his interlanguage
development conforms to what UG follows. Schwartz and Sprouse (1996) further
provide evidence for the inadequacy of Vainikka and Young-Scholten’s MT hypothesis
by re-analyzing previous studies of French speakers acquiring English. The French L2
Speakers had trouble with the sentence- internal placement of adverbs and produced the
S-V-Adv-O order in English. This suggests that the initial stage for learners already
includes a phrase marker with the adverb adjoined to VP.

A critical study by Bhatt and Hancin-Bhatt (2002) reports a comparison of the two
models in the L2 English grammar of Hindi Ll speakers, to Show a developmental
trend in the acquisition of CP. The result argues for the MT hypothesis and claims that
different CP projections develop in sequence. Their argument further supports the
gradual development of CP as reported in Japanese L2 learners of English in Kaplan
(1993). Clahsen, Penke, and Parodi (1993/1994) report as well that early L1 German
children also construct phrase-structure trees in a gradual fashion. Further, Bhatt and
Hancin-Bhatt (2002) make a theoretical proposal that the L2 initial State includes not

only VP, as hypothesized in the MT hypothesis, but also includes IP, and that only CP

42

develops later. However, they fail to further investigate whether morphosyntactic
features [—wh] and [+wh] of the relative CP develop along with their syntactic
projections or independently, or whether the two features emerge simultaneously or
whether one precedes the other.

In this dissertation, one crucial reason to reject the MT hypothesis and Bhatt and
Hancin-Bhatt’s (2002) claim is that C0 simultaneously necessitates the presence of
Spec-CP in the relative CP because Co needs to attract a lexical item with unspeciﬁed
features to Spec-CP in order to check the same unspeciﬁed features of C°. That is, the
components of CP structure cannot gradually emerge but must be present
Simultaneously with C0 in the relative CP.

Among earlier studies that support the FT /FA hypothesis, Lakshmanan and
Selinker (1994) and Selinker and Lakshmanan (2001) examine longitudinal data from
L1 Spanish and French children learning English as an L2, and conclude that Spec-CP
and C0 are operative from the very beginning of L2 acquisition. Grodin and White
(1996) argue against the MT hypothesis with data drawn from a longitudinal corpus of
spontaneous production data by two L1 English children who started attending French
kindergarten at age 5. Grodin and White suggest that functional categories and their
projections are present in the earliest utterances available. Epstein et a1. (1996) analyze
the error data from Japanese L2 learners of English who were tested on pre- and post-

posed sentence structures. They reject the no or partial transfer model such as the MT

43

hypothesis, and provide evidence that functional categories in L2 acquisition are fully
available to learners.

Parallel to this view, Lardiere (1998a) collected naturalistic longitudinal
production data from a Chinese woman, Patty, who had lived in the US for 18 years.
Lardiere reports that the relationship between morphosyntactic features and abstract
grammatical knowledge may be much less direct than other previous studies have
reported. Lardiere’s study rejects the MT hypothesis because it does not explain
irregular appearances of certain morphosyntactic features in Patty’s interlanguage. In
her study, however, Lardiere accepts the assumption made by the FT /FA hypothesis
about the existence of L1 phrase structure and the gradual emergence of its features.
These features may or may not match the native speaker’s knowledge.

Lardiere (1998b) re-analyzes the data from Patty and suggests that verb-raising is
optional for L2 learners, due either to failure to acquire verbal agreement morphology
or to an impairment of the morphosyntactic feature agreement system. She further
states that syntax and morphology are dissociated in L2 grammar. If so, it can be
assumed that the structure of relative CP and relevant morphosyntactic features for
agreement are also dissociated, and their acquisition is independent of each other.

Furthermore, Haznedar (2003) examines the status of the functional categories in
longitudinally collected data ﬁom a Turkish child, Erdem, acquiring English, and

presents counterevidence for the MT hypothesis or any partial transfer explanations.

She claims that the development of CP is not implicationally contingent on the prior
acquisition of IP. Haznedar (2003: 2) and a number of L2 researchers postulate the
Missing Surface Inﬂection hypothesis that “the lack of morphosyntactic forms in
interlanguage grammars reﬂects a problem with the realization of surface
morphology.”

Hawkins and Chan (1997) collected data from Chinese Speakers learning English,
based on the grammaticality judgment test, and make a similar proposition to the
Missing Surface Inﬂection hypothesis. Hawkins and Chan (1997: 187) argue that
“(the) divergence from native-Speaker representations is an effect of the inaccessibility
of features of functional categories in L2 acquisition.” They refer the claim as the
‘Failed Functional Features Hypothesis’.

Lakshmanan (2000) reports the ﬁndings of an experimental (cross—sectional) study
which investigated the L1 acquisition of the syntax and pragmatics of relative clauses
by Tamil-speaking children ranging in age from 2 years 11 months to 6 years 6 months.
She argues that the observed age-related differences stem primarily from performance
factors and that the younger children are probably not inferior to the older children in
regard to their knowledge of the syntax and pragmatics of Tamil relative clauses.
Lakshmanan’s ﬁnding assumes the availability of full-ﬂedged syntactic structure from
the beginning of language acquisition, but the feature speciﬁcation was different

among the younger and older children. That is, the acquisition level for

45

morphosyntactic features was different although, “from a developmental perspective,
children’s knowledge of the syntactic and the pragmatic module go hand in hand and
. neither syntax nor pragmatics may be developmentally prior to the other”

(Lakshmanan 2000: 615).

2.2 Chapter Summary

Chapter 2 reviewed previous literature on the availability of L1 grammar for L2
acquisition. Vainikka and Young-Scholten (1996a, 1996b) present data that support the
partial availability of L1 grammar and a gradual development of CP phrase structure in
L2 acquisition. They further re-analyze previous studies by Zobl and Liceras (1994),
Epstein et al. (1996), and Grondin and White (1996) to support their own argument.
Bhatt and Hancin-Bhatt (2002) put forth data to support Vainikka and Young-
Scholten’s claim, and they conclude that C0 emerges prior to Spec-CP in L2
acquisition.

On the other hand, Schwartz and Sprouse (1994, 1996) offer different data that
support the availability of full-ﬂedged L1 grammar for L2 acquisition, in line with
Lakshmanan and Selinker (1994), Epstein et a1. (1996), Grodin and White (1996), and
Selinker and Lakshmanan (2001).

Lardiere (1998a, 1998b) contends that the acquisition of syntax is independent of

the development of morphosyntactic features. While adopting Schwartz and Sprouse’s

46

claim, her claim presupposes a full availability of L1 CP from the beginning, but the
morphosyntactic features of L1 CP may be subject to gradual maturation. Hawkins and
Chan (1997) refer this divergence of morphosyntactic features from phrase structure as
the Failed Functional Features hypothesis. Haznedar (2003) ﬁrrther strengthens
Lardiere’s approach to the Dissociation Hypothesis, and proposes the Missing Surface
Inﬂectien hypothesis. Lakshmanan (2000) in her Ll study claims that syntax is not
available at a different stage, but that performance factors such as feature speciﬁcation
cause the children to produce different outputs.

None of the previous studies have, however, collected production data from
relative CPS, which may offer concrete insights into the development of functional

categories as well as into the presence of L1 grammar for L2 acquisition.

47

CHAPTER 3

3.1 Purpose of the Study

The purpose of this study is to provide empirical evidence that the fully developed
Ll functional categories are transferred and their morphosyntactic features develop in
succession. This evidence is in accord with the Dissociation Hypothesis, which seems,
therefore, to be superior to the MT hypothesis in its explanation of the early stages of
the acquisition of phrase structure and the acquisition of its morphosyntactic features.
This dissertation presents evidence to support the conclusion that one of the functional
categories, relative CP, is transferred from L1 and that [:twh], which is speciﬁc to the
relative CP, gradually develops from [—wh] to [+wh]. This ﬁnding can be generalized
for other CPS that emerge at earlier stages in L2 acquisition.

Cross-sectional data on SUB, DO, and 10 relative CPS are collected in an
experimentally controlled Situation; The three previously mentioned types of relative
CPS are examined in one study in order to test whether there is a relationship between
the syntactic and predicate complexity of a given CP structure and the development of
morphosyntactic features. A reason for employing the 10 relative CP is to collect pied-
piped constructions to investigate whether the [+wh] relative CP can be further

developed to attract a heavier, theoretically disfavored element to Spec-CP.22 Both

 

22 Under the minimalist framework, Chomsky (1992) proposes the Preference Principle that a
minimally restricted relative CP in Spec-CP is always preferred so long as the grammar converges in
this manner. In other words, according to this principle, only the relative operator itself should be

48

spoken and written data are collected from the same group of learners in order to
examine whether different production modes play a role in the development of
morphosyntactic features (cf. Ball 1996).

The intended linguistic data are different types of English relative CPS produced
by adult Korean learners. The availability of the afﬁxal complementizer marker with
[—wh, +pred] in Korean may encourage learners to dispense with the relative CP
having [+wh, +pred] in English. If so, learners might prefer to produce the relative CP
in English that is syntactically compatible with the relative CP in Korean, regardless of
the L2 proﬁciency level. If HPL produces a greater proportion of [+wh] relative CPS
than LPL, it appears that L2 morphosyntactic features gradually develop. The gradual
development may be affected by the complexity of relative CPS, leading to a sort of
acquisitional hierarchy that can be anticipated based on learners’ L2 knowledge of the
morphosyntactic features of relative CPS.

It is expected that although the statistical signiﬁcance across SUB, DO, and 10

 

logically contained in Spec-CP. Now look at the two samples below.

a. the cat which the girl is giving a ﬁsh to

b. the cat to which the girl is giving a ﬁsh
In (a) only the wh-operator moves alone, stranding the preposition to (i.e., preposition stranded
construction); while, in (b) to which, the whole prepositional phrase containing the wh-operator, moves
to Spec-CF (i.e., preposition pied-piped construction), which requires more difﬁculty, and thus should
be a disfavored derivation. The Preference Principle also favors (a) because it has a less restricted
element in Spec-CP, compared to (b) with a more restricted one. In addition, (b) requires more difficulty
than (a) does to copy and delete elements in the base (Munn 1994; Sauerland 1998; Saﬁr 1999). [Make
OP] copies and deletes to which; whereas in (a), only which is copied and deleted. Nevertheless, both
constructions in (a) and (b) are considered grammatical in English and, more or less, entail the same
interpretation; but they should not co-exist; (b) Should be an illegitimate output because different
derivations must compete, in the sense that the most economical one blocks other competing
alternatives arising out of the same set of lexical items.

49

relative CPS might vary, HPL would generally produce more [+wh] relative CPS than
LPL. Therefore, the expected conclusion would be that [:twh] of the relative CP
develops from [—wh] to [+wh]. This outcome would support the Dissociation
Hypothesis, which argues for the gradual development of L2 morphosyntactic features
and a full transfer of L1 CP as a block. Also, more [-wh] relative CPS will be produced
from DO than SUB relative CPS in speaking and writing situations. More DO than 10
relative CPS with the same feature will be produced as well.

This result would suggest that the development of relative CPS may vary
according to the syntactic and predicate complexities. The expected result can also be
related to the generalization by the Noun Phrase Accessibility Hierarchy (NPAH),
which predicts the acquisition of SUB before DO and DO before 10 relative clauses.23

To examine these hypotheses, two research questions are posed below.

3.2 Research Questions

Research Question 1: Do high proﬁciency L2 learners produce a greater proportion of
[+wh] SUB, DO, and 10 relative CPS in Speaking and writing situations than low
proﬁciency L2 learners?

Research Question 2: Do low proﬁciency L2 learners produce a greater proportion of
[—wh] relative CPS than high proﬁciency L2 learners as the complexity of
relative clause structures increases through SUB, DO, and 10 relative CPS and in
speaking and writing situations?

 

23 Studies on the acquisition of relative clauses in SOV languages report that the object relative CP
tends to be acquired prior to the subject CP by L1 children (N irmala 1980; Bai 1989) and L2 learners
(O’Grady et al. 2000; K0 2002; O’Grady 2003).

50

3.3 Research Hypotheses
To understand more deeply and in more detail the possible effects of the research

questions, six research hypotheses are proposed.

Hypothesis 1: High proﬁciency L2 learners produce a greater proportion of [+wh]
SUB relative CPS than low proﬁciency L2 learners in speaking and writing
situations.

Hypothesis 2: High proﬁciency L2 learners produce a greater proportion of [+wh] DO
relative CPS than low proﬁciency L2 learners in speaking and writing
Situations.

Hypothesis 3: High proﬁciency L2 learners produce a greater proportion of [+wh] 10
relative CPS than low proﬁciency L2 learners in speaking and writing
Situations.

Hypothesis 4: L2 learners produce a greater proportion of {—wh] DO relative CPS than
[-—wh] SUB relative CPS in speaking and writing situations.

Hypothesis 5: L2 learners produce a greater proportion of [—wh] 10 relative CPS than
[-wh] DO relative CPS in speaking and writing situations.

Hypothesis 6: L2 learners produce a greater proportion of [-wh] 10 relative CPS than
[—wh] SUB relative CPS in speaking and writing situations.

3.4 Research Methodology
This research was conducted with the MSU IRB approvals, IRB #04-897 and IRB

#X06-467.

3.4.1 Participants and Classification into Levels

The participants were Korean adults who were proﬁcient enough in English to be

51

capable of producing relative clauses in English in the experimental Situation.
Moreover, since English relative clauses are rather complicated structures for L2
learners, only Korean college students were interviewed. These students, 126 from
several English classes, were majoring or mirroring in English at Kangwon National
University in Chuncheon, Gangwon Province, Korea.

In order to classify participants according to level of English proﬁciency, a
preliminary sentence combination test was given24 in which two sentences containing
the same noun had to be properly joined by using a relative pronoun a, that, or
which.25 Twelve questions with fairly low level vocabulary were asked: four each
concerning the SUB, DO, and 10 relative CPS. Further, each relative CP consisted of a
matrix sentence containing either a subject or an object that was supposed to be the
relative head.

Furthermore, since Fox and Thompson (1990: 301) claim that “unlike nonhuman
subject heads, 85% of which were deﬁnite, only 28% of the non-human object heads
are deﬁnite,” humanness and deﬁniteness of the head noun were controlled to be non-

human and deﬁnite, for example, the apples, the dog, etc.26 Because the length of a

 

2" See Appendix A for the test sheet.

25 The test has been adapted from the one originally used in Gass (1979). Although such a paper-and-
pencil test for relative clause formation has provoked criticism in terms of pragmatic felicity (cf.
Hamburger and Crain 1982; Crain and Thornton 1998), it has frequently been used in L2 studies.

26 Fox and Thompson (1990: 310) further report that “the subject slot tends to be associated with
deﬁnite, identiﬁable, speciﬁc human referents, while the object Slot tends to be associated with non-
identiﬁable, non-human, non-Speciﬁc referents.” The sensitivity to humanness is also reported in Kim

and Shin (1994: 469) that “the Ononhuman'RC pattern in English and the RC'Ononhuman pattern in
Korean are relatively prevalent in comparison with other types.”

52

clause may affect relativization, as reported in Yamashita (1994), the sentences being
relativized were kept relatively Short.

Thus, a total of 12 relative CPS were collected from the participants: two each of
the types: SUB-SUB, DO-SUB SUB-DO, DO-DO, SUB-IO, and DO-IO relative CPS.

Each sample sentence is listed below.

SUB-SUB: The apples th_at_ grow in the mountain are expensive.

DO-SUB: Paul hates the_cg t_ha_t often breaks down.27

SUB-DO: The songs flit the old man wrote are very sad.

DO-DO: The man kicked meg 113311 his neighbor parked.

SUB-IO: The brother hates th_ec_at t_ha_t his sister gave a ﬁsh to.

DO-IO: The lady cleaned the room that the manager sent a bottle of wine to.

Participants were divided into two proﬁciency levels, LPL and HPL.28 Seventy-
nine participants were classiﬁed as LPL: Sixty-ﬁve of these participants had either
fewer than 7 correct answers out of the 12 questions, and 14 participants had more
than 6 correct answers without a single correct IO relative CP. The other 47
participants were classiﬁed as HPL. Table 1 shows the pre-test scores for LPL and

HPL.

 

27 What will be disregarded in this dissertation is whether the syntactic structure of the relative clause
branches into DP (the) and NP (dog that...) or DPfN P (the dog) and CP (that. . .).

2“ Two colleagues, a Korean graduate student, and a Korean professor in Applied Linguistics
acknowledged the level classiﬁcation and commented that it was reasonable, according to their
experience in language acquisition research.

53

Table l: Pre-test scores for LPL and HPL

 

 

 

 

SUB DO 10
LPL 176 correct 200 correct 25 correct
(79 learners, 316 tokens) (56%) (63%) (8%)
HPL 151 correct 159 correct 74 correct
(47 learners, 188 tokens) (80%) (85%) (39%)

LPL performed better on the DO than SUB relative CPS (176 and 200), while the
number of correct 10 relative CPS was just 25. HPL seemed to have done about the
same on the SUB and DO relative CPS (151 and 159), and roughly twice as well as the
10 relative CP (74).29 The proportions, in general, indicate that HPL was better on the
sentence combination test than LPL. One factor was more Signiﬁcant than others for
classifying participants into LPL and HPL: the outcome of 10 relative CPS.

Not all learners who took the initial test, however, participated in the actual

experiment. Forty-eight of 79 LPLS participated, and 40 of 47 HPLS.

3.4.2 Data Collection Method

A cross-sectional elicited production method was employed to acquire sufficient
data for the target structures within a single experimental session (cf. Hamburger 1980;
Hilderbrand 1987; Crain et a1. 1990; Labelle 1990, 1996; Pérez-Leroux 1995;

McDaniel et al. 1996; Thornton 1996; Crain and Thornton 1998). The task enabled the

 

29 Fox (1987) also reports that the ratio of subject and object relatives is 1:1 in her data of written
English.

experiment to evoke complex syntactic structures such as the relative CP, which occurs
quite infrequently in both L1 and L2 spontaneous speech. Speakers often avoided the
intended construction by choosing an alternative, Simpler means of expression.
Hilderbrand (1987) ﬁnds in her acquisition study with Ll children that this type of
elicitation task gives more direct access to the child’s underlying competence than the
spontaneous production or comprehension tasks. Crain and Thornton (1998) claim that
such a production task can elicit language learners’ grammar more directly than a
comprehension task because correct derivations from the lexicon do not take place by
accident.

Moreover, there are two other potential advantages of the elicited production task
over other tasks. One is that it can acquire readily replicable results and the other is
that the acquired data entail greater conﬁdence and precision with respect to what
learners meant in producing the relative clauses, as Hamburger and Crain (1982) state.
McKee et al. (1998) take the production data as the truer indicator of children’s early
syntactic competence. They assume that in both children and adults comprehension
data indicate aspects of language processing that may be inﬂuenced by factors other
than syntactic knowledge.

In the current study, the experimental setting and linguistic environments during
all sessions were managed in order to produce circumstances which were as similar as

possible for each participant. Irrelevant but immediate extraneous variables were

55

possibly the time of the day for data collection, and participants’ motivation and
sincerity in the experiment. Results substantiate that the elicited production task helps
uncover learners’ grammatical knowledge of relative CPS more directly than

comprehension tasks.

3.4.3 Data Collection Instrument

A picture-description instrument was employed with three critical aspects:
situation, lead-in statement, and participant’s production, as suggested in Crain and
Thornton (1998). Each Situation included two experimental pictures containing the
same object (e.g., book, teddy bear, etc.) but in different settings. A sample picture for
each relative CP is provided in Appendix B. For example, to elicit an 10 relative CP,
the participant was presented with two pictures of a teddy hear. The target object teddy
bear was marked with a different colored sticker in each of the two pictures. One
picture illustrated a teddy bear with a green sticker, and the teddy bear was lying on the
picnic table. The other picture had another kind of teddy bear marked with a yellow
sticker, and a girl was giving a ball to the teddy bear. In that Situation, the research
administrator asked the participant, “Which teddy bear has the yellow sticker?” An
appropriate answer would be, for example, “the teddy bear which a girl is giving a ball
to.” Pictures for eliciting the written output were different, but the contents were

Similar. For example, if an intended DO relative CP for a Spoken output was to utter

56

“The gloves that a lady is wearing,” the corresponding DO relative CP for a written
output was to write “The cup that a lady is holding.” The order of picture presentation
for different types of relative clauses and for the distractors was exactly the same in
each of the two output modes.

The sticker method was ﬁrst developed in Labelle (1990), and has since been
uniquely modiﬁed and applied in Lakshmanan (2000) and O’Grady et al. (2000),
among others.

This method of presenting the participant with the same target object in different
settings was intended to construct a pragmatically felicitous discourse context in which
the presuppositions of restrictive relative clauses could be satisﬁed. When just one
object is present in the context, the restrictive relative clause becomes unnecessary
because there is no potential ambiguity for referring to it (Hamburger and Crain 1982;
Crain and Thornton 1998).

For every new situation, the research administrator read a manipulative lead-in
statement to elicit a relative CP by using the set of pictures. The administrator, however,
did not see the pictures being presented, requiring the participant to give a detailed and
Speciﬁc answer.

Participants were told to consider the question-and-answer situation as a telephone
conversation in which two parties cannot see what they are describing. This method

was effective when a participant wrongly answered such as This one!, The first one!,

57

etc.

The lead-in statements did not contain any relative CPS, but only short declarative

and question sentences. A total of 14 situations were presented to each participant: 6

experimental situations and 8 distracting Situations. The 6 intended relative CPS

elicited from a participant consisted of two of each of the SUB, DO, and 10 relative

CPS.

The full list of the lead-in statements for the spoken and written Situations is in

Appendices C and D. Below is a version of a sample lead-in statement for eliciting an

10 relative CP.

Administrator:

Participant:

Administrator:

Administrator:

Participant:

Administrator:

Participant:

Administrator:

Participant:

In one picture, some children and a teddy bear are on the
picnic table. The teddy bear has a green star, right?

Yes.

In another picture, a girl is giving a candy to the teddy bear.
And it has a yellow star.

[A distracting question] Now tell me. Where is the candy in
the picture? Answer with a full sentence.

The candy is in the girl’s right hand.
Good. One more. Do you see the yellow and red stars?
Yes.

[The experimental question] Then, which teddy bear has the
yellow star?

[Response] The teddy bear that the girl is giving a candy to.

58

Before asking the experimental question, the administrator asked each participant
some unrelated questions to distract him and also to verify whether he or she was
following the lead-in statement. When he or She could not understand a question, it
was repeated. Furthermore, all participants were asked to answer in ﬁJll sentences and
to repeat any responses that were incomprehensible to the administrator.

The spoken output was recorded with Aiwa TP-VSS30 Cassette Recorder. It was
placed aside from the participants to minimize their fear of being recorded.30 The
average recording time per participant was about 9 minutes. After a short pause, the
written data elicitation procedure began and each participant was given a blank sheet
with numbers on it that corresponded to the experimental questions. The average
writing time was about 6 minutes. The reason for a Shorter response time for the
written output was partly that the participants were familiar by that time with how to
answer the questions. In addition, there were signiﬁcantly fewer questions and
repetition requests during the second experiment for the written output, and

participants were not allowed to make corrections to earlier responses.

3.4.4 Research Administrators
The research administrators for data collection were two adult native English

speakers working as English conversation teachers in Korea. They were informed

 

3° When the participants were discouraged with their spoken responses, the administrator encouraged
them by saying what they had said was actually very difﬁcult, but they did superbly.

59

about the purpose of the research, and administered the pilot study with some students
for practice. When gathering the actual data, they each called one participant at a time
into an empty classroom and introduced themselves and described the consent form.“
The initial contact person who notiﬁed the potential participants for the research was a

graduate student who had no personal relationship with them.

3.4.5 Data 'Ifeatment

The administrators audiotaped all spoken output and collected the written output,
and the researcher transcribed the tapes. When transcribing the output, the researcher
ﬁrst distinguished understandable responses from those that could not be understood
by the researcher and the administrators (e.g., The one who is sitting on a picnic table
whom the little girl giving). The latter were excluded. Also excluded were good
responses without a relative CP (e.g., The girl puts a book on the chair), or with a
relative CP but in an unintended structure like a passivized SUB relative CP (e.g., the
cup that is held by a woman) for the intended DO relative CP (e.g., the cup that a
woman is holding).32 When there was more than one response to a question (e.g., the
cup that a woman... Ah... the cup which a woman...), the ﬁrst grammatical relative
CP was selected.

The data were coded on a spreadsheet of the SPSS for Windows 12.0.1, the

 

3 ' See Appendix E for the consent forms.
32 Passivization was also an avoidance strategy for the intended DO relative clause for the L1 children
and adults in Fragman’s (1997) study (also, cf. Keenan 1985; Pe’rez—Leroux I995; McDaniel et al. 1998).

60

‘I-
‘i
‘.
s.
l-
I.
.I
. lul-

participants on the rows and the variables on the columns. The variables on the
spreadsheet were the scores for the sentence combination test ‘score’, the proﬁciency
level ‘level’, the participant’s number ‘id’, Spoken SUB relative CPS ‘ssub’, Spoken
DO relative CPS ‘Sdo’, spoken 10 relative CPS ‘sio’, written SUB relative CPS ‘wsub’,
written DO relative CPS ‘wdo’, and written 10 relative CPS ‘wio’. The variable level
was nominally categorized as 1 for LPL and 2 for HPL. The features of the different
relative CPS were 1 for o-relative CPS, 2 for that-relative CPS, 3 for wh-relative CPS,
and 4 for pied-piped (ppd) relative CPS. Listed below in Table 2 is the numbers of
spoken and written relative CPS from 48 LPLS and 40 HPLS, which were used for the

statistical test.

Table 2: Numbers of spoken and written relative CPS

 

 

 

 

 

 

 

 

SUB DO 10
Spoken
[—wh] [+wh] [—wh] [+wh] [-wh] [+wh]
-ing33 that wk 5 that wk 9 that wh ppd
LPL (96)a 61 1 l6 4 8 13 8 4 10 0
HPL (80)b 3o 16 32 9 17 25 10 16 17 o
_ SUB DO 10
Written
I-Wh] [+wh] l-Whl [+wh] I-Wh] [+wh]
-ing that wh a that wh a that wh ppd
LPL (96) 42 9 l8 5 11 15 0 6 12 3
HPL (80) 15 19 43 13 12 28 10 l3 l8 5

 

 

 

a 48 LPLS x 2 tokens = 96 tokens b 40 HPLS x 2 tokens = 80 tokens

 

33 The -ing nonﬁnite clause is not an abbreviated progressive form of a relative clause (Quirk et a1.
1985), but I listed the ﬁgures to Show the frequency of this substitute for the SUB relative CP. Kim and
Shin (1994: 477) report that “English Speakers Show a strong preference for the use of prepositional
phrases or reduced passive modiﬁers over the use of full-ﬂedged relative-clause constructions.”

61

Two tokens from each relative CP were elicited from every participant; that is, two
from each SUB, DO, and 10 relative CP for both spoken and written outputs.
Therefore, the maximum number of tokens for LPL was 96 for each relative CP (48
LPLS x 2 tokens), and 80 for HPL (40 HPLS x 2 tokens). So, for example, a total of 25
spoken DO relative CPS for LPL (4 + 8 + 13) were successfully elicited out of the total
96 possible occasions; and 51 spoken DO relative CPS for HPL (9 + 17 + 25), out of

the total 80 tokens.

3.4.6 Nonparametric Statistical Techniques

The data were counted by the number of occurrence and further analyzed to make
inferences about the general tendency of the learners as a whole. The appropriate
statistical tests for such types of data are the Binomial test for calculating the choice
and the 12 test within the nonparametric statistical techniques (cf. Siegel and Castellan
1988; Franekel and Wallen 2003). The types of data and the expected sample size for
the present research ﬁt the general requirements for use of the techniques.

The procedure for hypothesis testing was followed in ﬁve steps as shown below,
listed by the order of performance.
1. Stating the null hypothesis (H0): A hypothesis of no differences, counter to the

research hypothesis (H 1), was formulated for the purpose of being rejected. Since

the testing wanted to reach a decision about differences, Hg was tested against H1.

62

H1 was accepted if Hg was rejected. Nevertheless, the degree of acceptance or
rejection could be deﬁnite in some cases, while there might have been other cases
in which it seemed questionable to accept the outcome of the statistical ﬁgures.

2. Choosing an appropriate statistical technique for testing Ho: An admissible
statistical test was selected for each hypothesis. The one-sample tests like the
Binomial test or the A} test used for the goodness-of-ﬁt type were carried out.
Comparison between independent samples was tested with the 12 test. A research
hypothesis, however, often required a few steps of testing, each of which involved
a different kind of statistical test depending on the nature of the variables being
compared.

3. Specifying a Signiﬁcance level (a) and sample Size (N): For all hypothesis testing, a
= .0534 was used for the probability that a statistical test would yield a feature
under which Hg was rejected. N was listed to draw for computing the statistical
test. N for the total was 88: 48 for LPL and 40 for HPL. Two tokens for each
relative CP (Spoken-SUB, Spoken-DO, Spoken-IO, Written-SUB, Written-DO,
and Written-IO) were elicited; that is, 12 possible tokens from each participant.

4. Deﬁning the region of rejection: The region of rejection, expressed by a, was a
region of the signiﬁcance level in which Ho could be rejected in favor of H1. The

region was affected by the nature of H1. If H | indicated the predicted direction of

 

’4 The level, a = .05, is commonly used in educational research (Fraeken and Wallen 2003) and
behavioral sciences (Siegel and Castellan 1988).

63

the difference, a one-tailed test was then called for. If H1 did not indicate the
direction of the predicted difference, then a two-tailed test was called for. The
degree of freedom was mentioned in this step.

5. Decision: Hg was rejected or accepted in favor of or in preference to H1 according to
the observed feature of signiﬁcance whose associated probability of occurrence

under Hg was equal to or less than a.

3.5 Chapter Summary

Chapter 3 introduced the purpose of the study — to provide empirical evidence
about the Dissociation Hypothesis, which fundamentally assumes that the fully
developed Ll functional categories are transferred and their morphosyntactic features
develop in succession from [—wh] to [+wh]. The results elicited from the examination
of the data on relative CPS can also be discussed in terms of the generalization made
by the NPAH. The research questions asked whether HPL produces a greater
proportion of [+wh] SUB, DO, and 10 relative CPS in speaking and writing situations
than LPL and also whether LPL produce a greater proportion of [—wh] relative CPS
than HPL, as the complexity of relative clause structures increases in speaking and
writing. The six research hypothesesspeciﬁcally proposed the possible outcomes out

of the research questions.

Participants in the experiment were Korean adults who study English as an L2 at a

 

Korean university and were considered capable of producing the intended relative CPS.
Initially, a sentence combination test was given to 126 potential participants for level
classiﬁcation into LPL and HPL. The resulting number of participants was 88: 48
LPLS and 40 HPLS.

A cross-sectional elicited production task was employed as the method for data
collection to acquire a sufﬁcient number of target structures within a Single
experimental session. Such a complex structure occurs quite infrequently in
Spontaneous speech and has to be experimentally evoked to maintain the consistent
data elicitation setting and linguistic environment for each and every participant.

The instrument for data collection was a picture-description task. It contained
three observable phenomena: situation, lead-in statement, and participant’s production.
Each situation contained two pictures with the same object but in a different setting.
The intention was to construct a pragmatically felicitous discourse context in which
the presuppositions of restrictive relative clauses could be satisﬁed.

Two adult native English speakers administered the data elicitation process at the
university that the participants were attending. The experiment administrators recorded
the spoken output on a cassette tape player and collected the written output. The
researcher transcribed the spoken output.

Any production of relative clauses not containing the intended relative CP was

discarded. When more than one response was collected for a single question, the ﬁrst

65

 

relative CP was selected for analysis.

A nonparametric statistical technique was selected for data analysis because the
data were counted by the number of occurrence. The appropriate statistical tests were
the Binomial test for calculating the choice and the 12 test within the nonparametric
statistical techniques. The procedure for hypothesis testing was followed in ﬁve steps:
stating the null hypothesis (Ho), choosing an appropriate statistical technique for
testing H0, specifying a signiﬁcance level (a) and sample Size (N), deﬁning the region

of rejection, and making the decision.

66

CHAPTER 4

4.1 Testing Hypothesis 1: SUB Relative CP

H1: High proﬁciency L2 learners produce a greater proportion of [+wh] SUB relative
CPS than low proﬁciency L2 learners in speaking and writing situations.

The ﬁrst part of Hypothesis 1 concerns whether LPL or HPL produced a
signiﬁcantly greater proportion of [+wh] SUB relative CPS than the one with [—wh], in
speaking and writing. Since the SUB relative CP could be produced only as a wh-
relative CP (the bear which is catching a ﬁsh) or a that-relative CP (the bear that is
catching a ﬁsh), the Binomial test was considered. The test is of the goodness-of-ﬁt
type, which indicates whether a signiﬁcant difference exists between the observed
relative frequency (converted into a proportion) of a variant and the expected number
of occurrence based on Ho. The subsequent statistical test was the )(2 test for two
independent samples, which determines the signiﬁcance of differences of frequency
between two independent groups.

1. H0: [—wh] = [+wh]. There is no difference between LPL and HPL in the probability
of occurrence of the [—wh] and [+wh] SUB relative CPS in speaking and writing. H I:
[-wh] < [+wh].

2. Statistical test: The Binomial test was chosen because the data were in two discrete
categories and the design was of the one-sample type. If the two possibilities [—wh]

and [+wh] are randomly selected, there is no reason to think that one feature is

67

preferred to the other feature under H0.

3. Signiﬁcance level: Let a = .05. N = 96 (48 LPLS x 2 tokens) and N = 80 (40 HPLS x
2 tokens).

4. Rejection region: Since H1 predicted the direction of deviation from randomness, a
one-tailed region of rejection was used. H0 would be rejected if the observed value of
the Binomial test was such that the probability associated with its occurrence under Hg

for the degree of freedom of l was equal to or less than a = .05.

Table 3: The Binomial test result of SUB relative CPS for LPL

 

 

Observed Test Asymp. Sig.

 

 

 

 

Features Categories N , .
Proportion Prop. (2-ta11ed)
[-wh, +pred] that 1 .06 .50 .000
Spoken-SUB
[+wh, +pred] which 16 .94
Total 17 1.00
. [—wh, +pred] that 5 .19 .50 .002a
Written-S UB
[+wh, +pred] which 22 .81
Total 27 1.00

a Z Approximation is used for the cells that are N > 25, which causes an asymptotic signiﬁcance

calculation.

5. Decision: We will discuss the data by LPL ﬁrst. In the experiment, LPL used a
greater proportion of [+wh] SUB relative CPS in both Speaking and writing. The data
are Shown in Table 3 for LPL. The N column lists the total numbers of relative CPS

produced.

In the speaking situation, the observed frequency of [—wh] SUB relative CPS for
LPL is just 1 out of 17, while the frequency is 16 out of 17 for [+wh] SUB relative CPS.
The probability associated with x S 1 for N = 17 is p = .000. Because this p is smaller
than a = .05, the decision is to reject H0 in favor of H 1 for the Spoken data.

In the writing Situation, the observed frequency of [—wh] SUB relative CPS for
LPL is 5 out of 27, while the frequency is 22 out of 27 for [+wh] SUB relative CPS.
The probability associated with x S 5 for N = 27 is p = .002. Because this p is smaller
than a = .05, the decision is also to reject H0 in favor of H; for the written mode.

We can conclude that [—wh] < [+wh] for LPL in both situations; that is, LPL
tended to prefer the [+wh] SUB relative CP to the one with [—wh] in both Speaking and
writing. On the other hand, HPL also used a greater proportion of [+wh] SUB relative

CPS. The data are shown in Table 4 for HPL.

Table 4: The Binomial test result of SUB relative CPS for HPL

 

 

Observed Test Asymp. Sig.

 

 

 

 

Features Categories N . .
Proportion Prop. (2-ta11ed)
—wh, + red that 16 .33 .50 .029“
Spoken-SUB [ p ]
[+wh, +pred] which 32 .67
Total 48 1.00
, {-wh, +pred] that 19 .31 .50 .003‘1
Wrrtten-SUB
[+wh, +pred] which 43 .69
Total 62 1.00

 

a Z Approximation is used for the cells that are N > 25, which causes an asymptotic Signiﬁcance

calculation.

In the speaking situation, the total number of [—wh] SUB relative CPS is 16 for
HPL, while the total number of [+wh] SUB relative CPS is 32. The probability
associated with x S 16 for N = 48 is p = .029. Because this p is smaller than a = .05,
the decision is to reject H0 in favor of H1. It can be concluded that [—wh] < [+wh] for
HPL; that is, HPL preferred the [+wh] SUB relative CP in the speaking situation.

In the writing Situation, the total number of [—wh] SUB relative CPS is 19 for HPL,
while the total number of [+wh] SUB relative CPS is 43. The probability associated
with x S 19 for N = 62 is p = .003. Because this p is smaller than a = .05, the decision
is to reject H0 in favor of H1. It can be conchrded that [—wh] < [+wh]; that is, HPL as
well as LPL tended to prefer the [+wh] SUB relative CP in both Speaking and writing.

The next part of Hypothesis 1 concerns whether LPL and HPL differ signiﬁcantly
with respect to the relative frequency of the [+wh] SUB relative CP in speaking and
writing. The )(2 test for two independent samples is employed.

1. Ho: LPL = HPL. There is no difference between LPL and HPL in the probability of
selecting the [+wh] SUB relative CP. H1: LPL < HPL.

2. Statistical test: The 12 test was used to determine the signiﬁcance of differences
between the two groups Since the data consisted of frequencies. For choosing the
[—wh] or [+wh] SUB relative CPS, there is no reason to think that one or the other is

preferred under Ho.

70

Table 5: Frequencies and Crosstabulations for SUB relative CPS

 

 

 

 

 

 

 

 

Frequencies
Valid Missing Total
Crosstabulation N Percent N Percent N Percent
Level*Spoken-S UB 65 36.9% 1 11 63.1% 176 100.0%
Level*Written-SUB 89 50.6% 87 49.4% 176 100.0%
Spoken-SUB Written-SUB
[-wh] [+wh] Total [-wh] [+wh] Total
Level LPL 1 16 17 Level LPL 5 22 27
HPL 16 32 48 HPL l 9 43 62
Total 17 48 65 Total 24 65 89

 

 

 

3. Signiﬁcance level: Let a = .05. N = 176. The total number of the participants was 88

(48 LPLS and 40 HPLS), and two tokens from each participant were collected. Table 5

Shows the frequency summary and crosstabulations for SUB relative CPS for LPL and

HPL.

4. Rejection region: Since H1 predicted the direction of deviation from randomness, a

one-tailed region of rejection was used. would be rejected if the observed value of )(2

was such that the probability associated with its occurrence under Hg for the degree of

freedom of 1 is equal to or leSS than a = .05.

5. Decision: In the experiment, there was a difference between LPL and HPL in the

probability of selecting the [+wh] SUB relative CP, especially in the Spoken data. The

two groups behaved differently when producing SUB relative CPS in that situation.

The value of )(2 for the Spoken data is Shown in Table 6.

71

Table 6: The )(2 test for SUB relative CPS in the Spoken data

 

Asymp. Sig. Exact Sig. Exact Sig.

 

Spoken-SUB Value df _ _ _
(2-S1ded) (2-Srded) (l-s1ded)

Pearson Chi-Square 4.898b l .027
Continuity Correctiona 3.580 1 .058
Likelihood Ratio 5.994 1 .014

Fisher’s Exact Test .029 .023
Linear-by-Linear Association 4.823 1 .028

N of Valid Cases 65

 

 

a Computed only for a 2x2 table

b 1 cell (25.0%) has expected count less than 5. The minimum expected count is 4.45.

In the speaking Situation, the probability of occurrence under Hg for )(2 _>_ 4.898
with the degree of freedom of l is p = .023. Because this p is less than a = .05, the
decision is to reject Ho. However, H1 is not supported either. The opposite of H1 is
supported instead, because the probability of occurrence of the [+wh] SUB relative CP
in Speaking is greater for LPL (p = .000 from Table 3) than HPL (p = .029 from Table
4). LPL produced a greater proportion of [+wh] SUB relative CPS in Speaking than
HPL.

In the writing Situation, there was, however, no difference between LPL and HPL

in the probability of selecting the [+wh] SUB relative CP. The value of 12 for these

data is shown in Table 7.

72

 

Table 7: The x2 test for SUB relative CPS in the written data

 

 

Asymp. Sig. Exact Sig. Exact Sig.

 

Written-SUB Value df
(2-sided) (2-Sided) (l—Sided)

Pearson Chi-Square 1.404b l .236

Continuity Correctiona .856 1 .355

Likelihood Ratio 1.473 1 .225

Fisher’s Exact Test .303 .178
Linear-by-Linear Association 1.389 1 .239

N of Valid Cases 89

21 Computed only for a 2x2 table

b 0 cells (0%) have expected count less than 5. The minimum expected count is 7.28.

The probability of frequency under Hg for 12 2 1.404 with the degree of freedom of 1
is p = .178. Because this p is greater than a = .05, the decision is to accept H0 in

preference to H1 in the writing situation.

4.2 Testing Hypothesis 2: DO Relative CP

H1: High proﬁciency L2 learners produce a greater proportion of [+wh] DO relative
CPS than low proﬁciency L2 learners in Speaking and writing situations.

The ﬁrst part of Hypothesis 2 concerns whether LPL or HPL produced a
Signiﬁcantly greater proportion of [+wh] DO relative CPS in speaking and writing.
Although the DO relative CP has three variants (a wh-relative CP with [+wh] and a a-
relative CP and a that-relative CP with [—wh]) the number of categories is two: [—wh]
and [+wh]. Therefore, the Binomial test was again used as it was used for Hypothesis 1.

1.110: [_th = [+wh]. H1: {—th < [+wh].

2. Statistical test: The Binomial test was again used as it was used for Hypothesis 1.

3. Signiﬁcance level: Let a = .05. N = 96 (48 LPLS x 2 tokens) and N = 80 (4O HPLS x
2 tokens).

4. Rejection region: Equal to or less than a = .05.

5. Decision: The Binomial test result is presented in Table 8 for LPL (See Table 2 for
all participants). We will discuss the data by LPL ﬁrst. The table shows that, out of a
total of 96 occurrences, LPL successfully produced 25 DO relative CPS in speaking: 4
o-relative CPS, 8 that-relative CPS, and 13 wh-relative CPS (12 [—wh] and 13 [+wh]).
Of the 31 DO relative CPS in writing, there were 5 o-relative CPS, ll that-relative CPS,

and 15 wh-relative CPS (l6 [—wh] and 15 [+wh]).

Table 8: The Binomial test result of DO relative CPS for LPL
Observed Test Asymp. Sig

 

 

 

 

LPL Features Categories N _ .
Proportion Prop. (2-ta11ed)
a 4
Spoken- [—wh, +pred] 12 .48 .50 1.000
that 8
DO
[+wh, +pred] which 13 .52
Total 25 1.00
. 9 . 5 a
Wrrtten- [—wh, +pred] 16 .52 .50 1,000
that 11
DO
[+wh, +pred] which 15 .48
Total 31 1.00

 

a Based on Z Approximation

In the speaking Situation, the probability associated with x S 12 for N = 25 is p =

74

1.000. Because this p is greater than a = .05, the decision is to accept H0 in preference
to H1. It can be concluded that LPL randomly selected a DO relative CP in Speaking.

In the writing situation, the probability associated with x S 16 for N = 31 is p =
1.000. Because this p is greater than a = .05, the decision is to accept H0 in preference
to H1. It can be concluded that LPL randomly selected a [—wh] or [+wh] DO relative
CP in both the speaking and writing situations.

The frequency data for HPL are presented in Table 9. The table shows that, of the
total of 80 occurrences in the spoken data, HPL produced 51 DO relative CPS: 9 a-
relative CPS, l7 that-relative CPS, and 25 wh-relative CPS (26 [—wh] and 25 [+wh]). Of
the 53 DO relative CPS in the written data, HPL produced 13 o-relative CPS, 12 that-

relative CPS, and 28 wh-relative CPS (25 [—-wh] and 28 [+wh]).

Table 9: The Binomial test result of DO relative CPS for HPL
Observed Test Asymp. Sig

 

 

 

 

HPL Features Categories _ _
Proportion Prop. (2-tarled)
a 9 a
Spoken— [—wh, +pred] 26 .51 .50 1.000
that 17
DO
[+wh, +pred] which 25 .49
Total 51 1.00
a 13 3
Written- [—wh, +pred] 25 .47 .50 .784
that 12
DO
[+wh, +pred] which 28 .53
Total 53 1.00

 

a Based on Z Approximation

75

In the Speaking Situation, the probability associated with x S 26 for N = 51 is p =
1.000. Because this p is greater than a = .05, the decision is to accept H0 in preference
to H1.

The writing situation was Similar. The probability associated with x S 25 for N
= 53 is p = .784. Because this p is greater than a = .05, the decision is to accept H0
in preference to H1. It can be concluded that HPL as well as LPL randomly selected
a [—wh] or [+wh] DO relative CP in both the speaking and writing Situations.

The x2 test for two independent samples to assess whether LPL and HPL would
differ with respect to the relative frequency of the [+wh] DO relative CP, was not
employed. The reason was that the both groups failed to produce a greater proportion
of [+wh] relative CPS in both speaking and writing. Therefore, the statistical procedure

for )(2 test became unnecessary unlike the procedure for the SUB relative CP.

4.3 Testing Hypothesis 3: 10 Relative CP

H1: High proﬁciency L2 learners produce a greater proportion of [+wh] 10 relative
CPS than low proﬁciency L2 learners in speaking and writing situations.

Hypothesis 3 concerns whether LPL or HPL produced a Signiﬁcantly greater
proportion of [+wh] IO relative CPS than [—wh] 10 relative CPS. Although an 10
relative CP can be produced in four ways: o-relative CP and that-relative CP with

[—wh] and wh-relative CP and pied-piped (ppd) relative CP with [+wh], the number of

76

categories under consideration is again two: [—wh] and [+wh]. The Binomial test was
appropriate for understanding whether a Signiﬁcant difference exists between the
observed relative ﬁequency (converted into a proportion) of a variant and the expected
number of occurrence based on Ho.

1. H0: [—wh] = [+wh]. H1: {—wh] < [+wh].

2. Statistical test: The Binomial test was again used as it was used for Hypotheses l
and 2.

3. Signiﬁcance level: Let a = .05. N = 96 (48 LPLS x 2 tokens) and N = 80 (40 HPLS x
2 tokens).

4. Rejection region: Equal to or less than a = .05.

5. Decision: The Binomial result is presented in Table 10 for LPL. The table Shows
that, out of a total of 96 occurrences in the spoken data, LPL produced 22 10 relative
CPS in speaking with 8 e-relative CPS, 4 that-relative CPS, 10 wh-relative CPS, and no
ppd-relative CPS (12 [—wh] and 10 [+wh]). Of the 21 10 relative CPS, the observed
frequencies in writing were no o-relative CPS, 6 that-relative CPS, 12 wh-relative CPS,

and 3 ppd-relative CPS (6 [—wh] and 15 [+wh]).

77

Table 10: The Binomial test result of 10 relative CPS for LPL

 

 

Observed Test Asymp. Sig

 

 

 

 

LPL Features Categories N .
PrOportron Prop. (2-tailed)
[ h, + ed] a 8 12 55 50 a
-w r . . .832
Spoken- p that 4
10 which 10
[+wh, +pred] 10 .45
ppd 0
Total 22 1.00
[ h ed] 0 O 6 29 50 a
-W a + r - - . 78
Written— p that 6 0
10 which 12
[+wh, +pred] 15 .71
ppd 3
Total 21 1 00

3 Based on Z Approximation

In the speaking situation, the probability associated with x S 12 for N =22 is p
= .832. Because this p is greater than a = .05, the decision is to accept H0 in preference
to H1. It can be concluded that LPL randomly selected an 10 relative CP in Speaking.

In the writing Situation, the probability associated with x S 6 for N = 21 is p = .078.
Because this p is greater than a = .05, the decision is to accept H0 in preference to H1.
It can be concluded that LPL randomly selected a [—wh] or [+wh] 10 relative CP in
both the speaking and writing situations.

The frequency data for HPL are presented in Table 11. The table shows that, of
the total of 80 occurrences, HPL produced 43 10 relative CPS in speaking with 10 a-

relative CPS, 16 that-relative CPS, 17 wh-relative CPS, and no ppd-relative CPS, (26

[-wh] and 17 [+wh]). Of the 46 10 relative CPS, the observed frequencies in writing
were 10 o-relative CPS, 13 that-relative CPS, l8 wh-relative CPS, and 5 ppd-relative

CP (23 [—wh] and 23 [+wh]).

Table 11: The Binomial test result of 10 relative CPS for HPL
Observed Test Asymp. Sig

 

 

 

HPL Features Categories _ _
Proportion Prop. (2-ta11ed)
a 10 a
[—wh, +pred] 26 .60 .50 .222
Spoken- that 16
10 which 17
[+wh, +pred] 17 .40
ppd 0
Total 43 1.00
[ h + ed] a 10 23 50 50 1000"
—-w , r . . .
Written- p that 13
10 which 18
[+wh, +pred] 23 .50
ppd 5

 

Total 46 1.00

 

a Based on Z Approximation

In the Speaking Situation, the probability associated with x _<_ 26 for N = 43 is p
= .222. Because this p is greater than a = .05, the decision is to accept H0 in preference
to H].

In the writing Situation, the probability associated with x S 23 for N = 46 is p =
1.000. Because this p is greater than a = .05, the decision is to accept H0 in preference

to H1. It can be concluded that HPL randomly selected a [~wh] or [+wh] 10 relative

CP in both the speaking and writing Situations.

79

The 12 test for two independent samples to assess whether LPL and HPL would
differ with respect to the relative frequency of the [+wh] 10 relative CP, was not
employed. The reason was that the both groups failed to produce a greater proportion
of [+wh] 10 relative CPS with in both speaking and writing.

In the ﬁrst step for testing Hypotheses 4, 5, and 6, the data of LPL and HPL were
combined Since the hypotheses were assumed for the learners as a whole, not based on
their proﬁciency levels. Table 12 shows the observed number of each relative CP in
speaking and writing from all 88 participants (48 LPLS and 40 HPLS). The expected

number of each relative CP is 176 (88 participants x 2 tokens).

Table 12: Numbers and percents of relative CPS for all L2 participants

 

 

 

 

 

SUB DO 10
I-Whl [+wh] I-Wh] [+wh] l-Whl [+th
17 48 38 38 38 27
Spoken
9.7% 27.3% 21.6% 21.6% 21.6% 15.3%
Total 65 (37%) 76 (43%) 65 (37%)
. 28 61 41 43 29 38
Written
15.9% 34.7% 23.3% 24.4% 16.5% 21.6%
Total 89 (51%) 84 (48%) 67 (38%)

A participant who produced a SUB relative CP, however, did not always produce a
DO relative CP, and vice versa, or a participant who produced a DO relative CP did not
always produce an 10 relative CP either, and vice versa. Therefore, those tokens which

were not successful in producing both types of relative CPS that each of Hypotheses 4,

80

5, and 6 intends to compare had to be eliminated. Table 13 shows the numbers of

research tokens after exclusion.

Table 13: Numbers of tokens that produced both types of relative CPS

 

 

 

 

Research Tokens
Hypothesis 4 Spoken-SUB (65) vs. Spoken-DO (76) 36
(SUB vs. DO) Written-SUB (89) vs. Written-DO (84) 58
Hypothesis 5 Spoken-DO (76) vs. Spoken-IO (65) 33
(DO vs. 10) Written-DO (84) vs. Written-IO (67) 54
Hypothesis 6 Spoken-SUB (65) vs. Spoken-IO (65) 34
(SUB vs. 10) Written-SUB (89) vs. Written-IO (67) 45

4.4 Testing Hypothesis 4: SUB < DO

H1: L2 learners produce a greater proportion of [—wh] DO relative CPS than [—wh]
SUB relative CPS in speaking and writing Situations.

The McNemar test for the signiﬁcance of differences with two related samples is
particularly applicable to this type of design in which measurement is in the strength of
a nominal scale. Thus, McNemar is apt for testing the effectiveness of a particular
difference on learners’ preference among variables.

1. Ho: SUB = DO. The probability that L2 learners would produce a greater proportion
of [—wh] SUB relative CPS was equal to the probability that they would produce a
greater proportion of [—wh] DO relative CPS. H1: SUB < DO.

2. Statistical test: The McNemar test for the Signiﬁcance of differences was chosen

because the study used two related samples, the SUB relative CP and the DO relative
CP, with nominal data. If [-—wh] and [+wh] are randomly selected for every relative CP,
there is no reason to think that one was preferred to the other.

3. Signiﬁcance level: Let a = .05. N = 176, the number of tokens from 88 participants
(48 LPLS and 40 HPLS) and two tokens from each participant.

4. Rejection region: Since H; Speciﬁed the direction of the predicted difference, the
region of rejection is one-tailed. The region of rejection consisted of all values of )(2
which were so large that they had a one-tailed probability associated with their

occurrence under Ho of .05 or less. The degree of freedom was 1.

Table 14: Differences of the numbers of SUB and DO relative CPS

 

 

 

 

Spoken-SUB & Spoken-DO Written-S UB & Written-DO
Spoken-SUB [—wh] [+wh] Written-SUB [—wh] [+wh]
[—wh] 4 4 [—wh] 13 5

[+wh] 13 15 [+wh] 15 25

5. Decision: According to Table 14, the Spoken SUB and DO relative CP data with 36
tokens Show that the participants who preferred [—wh] for both the spoken SUB and
DO relative CPS are 4 (the left ‘4’ in the table) and those who preferred [+wh] for the
same relative CPS are 15. The participants who Showed a preference for different

features for different relative CPS are 13 and 4 (the second ‘4’ in the table). These data

82

are the ones that we are interested in, and the statistics calculates the differences of
preference based on these two values.

Thirteen is the number of the participants who selected [+wh] SUB relative CPS,
but preferred [—wh] DO relative CPS; That is, they selected [+wh] for the SUB relative
CP, but selected [-wh] for the DO relative CP. Four is the number of participants who
selected [—wh] for the SUB relative CP, but preferred [+wh] for the DP relative CP.

For the written SUB and DO relative CP data in Table 14, the participants who
chose [+wh] for SUB relative CPS and [—wh] for DO relative CPS are 15. The
participants who chose [-wh] for SUB relative CPS and [+wh] for DO relative CPS are
5. The statistics uses these two values which Show a difference of occurrence. Those
who preferred [—wh] SUB and [—wh] DO relative CPS are 13, and those who preferred
[+wh] SUB and [+wh] DO relative CPS are 25. Statistics, however, leaves out these
two values which Show no diﬁ’erence of preference.

Table 15 reveals that the probabilities of differences of preference under Hg for the

spoken and written SUB and DO relative CPS.

Table 15: The McNemar test result for SUB and DO relative CPS

Spoken-SUB & Spoken-DO Written-S UB & Written-DO
N 36 58
Exact Sig. (2-tailed) .049“ .041

 

 

 

 

 

a . . . . .
Binomial distribution used.

83

In the Speaking situation, the probabilities for the Spoken SUB and DO relative
CPS Show that p = .049 for 36 research tokens. The value, however, has to be halved
because a one-tailed test is called for and the table gives two-tailed values. Therefore,
the actual value when the degree of freedom is l is .0245. Because the probability
under Ho associated with the occurrence we observed is p = .0245 and is less than a
= .05, the observed value of 12 is in the region of rejection, and thus the decision is to
reject H0 in favor of H1.

In writing, the probability of occurrence under Hg for 58 tokens is half of p = .041;
that is .205 with the degree of ﬁeedom of 1. Because the probability under Ho
associated with the occurrence we observed is p = .0201 and is less than a = .05, the
observed value of )(2 is in the region of rejection, and thus the decision is also to reject
H0 in favor ofH1.

According to the data, learners as a whole produced a greater proportion of [-wh]
DO relative CPS than [—wh] SUB relative CPS in both the Speaking and writing

situations.

4.5 Testing Hypothesis 5: DO < [0

H1: L2 learners produce a greater proportion of [—wh] 10 relative CPS than [—wh] DO
relative CPS in speaking and writing Situations.

1. Ho: DO=10. H1: D0<10.

84

2. Statistical test: The McNemar test was chosen again as it was used for Hypothesis 4.
3. Signiﬁcance level: Let a = .05. N= 176.

4. Rejection region: a = .05 or less.

Table 16: Differences of the numbers of DO and 10 relative CPS

 

 

 

 

Spoken-DO & Spoken-IO Written-DO & Written-IO
(33 tokens) Spoken-IO (54 tokens) Written-IO
Spoken-DO [—wh] [+wh] Written—DO [-wh] [+wh]
[—wh] 16 3 [—wh] 14 10

[+wh] 7 7 [+wh] 10 20

5. Decision: According to Table 16, the tokens that Show a difference of preference for
[—wh] DO and 10 relative CPS in the speaking situation are 7 (the left ‘7’ in the table)
and 3. Seven indicates the number of the participants who selected [+wh] for the DO
relative CP, but chose [—wh] for the 10 relative CP. Three is the number of the
participants who selected [—wh] for the DO relative CP, but preferred [+wh] for the 10
relative CP. The probability of difference of preference is calculated, based on these
numbers.

Other two tokens l6 and 7 are excluded from the statistical analysis. Sixteen is the
number of the participants who preferred [—wh] for both the DO and 10 relative CPs;
seven is the number of those who preferred [+wh] for the same relative CPS.

In the writing situation, the number of the participants that Showed a difference of

85

preference for [+wh] for the DO relative CP and for [—-wh] for the 10 relative CP is 10
(the left ‘10’ in the table). The right ‘10’ in the table is a difference of preference for
those who selected [—wh] for the DO relative CP and for [+wh] for the 10 relative CP.
From the same table, the statistics disregards the numbers 14 and 20. Fourteen is the
number for those who preferred [—wh] and 20 is the number for those who preferred
[+wh] for both DO and 10 relative CPS. Table 17 below presents the statistical result

for the differences of preference.

Table 17: The McNemar test result for DO and 10 relative CPS

Spoken-DO & Spoken-IO Written-DO & Written-IO
N 33 54
Exact Sig. (2-tailed) .344a 1.000a

 

a . . . . .
Binomial distribution used.

In the speaking Situation for 33 tokens, the probability of occurrence for
preference shows that p = .344. The value again needs to be halved because a one-
tailed test was called for, and the table gives two-tailed values. Therefore, the actual
value when the degree of freedom is 1 is .172. Because the probability under Ho
associated with the occurrence we observed is p = .172 and is greater than a = .05, the
observed value of )(2 was not in the region of rejection, and thus the decision was to
accept H0 in preference to H1.

In the writing situation for 54 tokens, the probability of occurrence under H0 is

86

half of p = 1.000; that is .500 with the degree of freedom of 1. Because the probability
under H0 associated with the occurrence we observed is p = .500 and is greater than a
= .05, the observed value of 12 is not in the region of rejection, and thus the decision is
also to accept H0 in preference to H1.

These data lead to the conclusion that the learners as a whole did not seem to
produce a greater proportion of [—wh] 10 relative CPS than [-—-wh] DO relative CPS in

either the Speaking or writing Situations.

4.6 Testing Hypothesis 6: SUB <10

H1: L2 learners produce a greater proportion of [—wh] 10 relative CPS than [—wh] SUB
relative CPS in Speaking and writing Situations.

1. Ho: SUB = 10. H1: SUB < [0.
2. Statistical test: The McNemar test was chosen as it was used for Hypotheses 5 and 6.
3. Signiﬁcance level: Let a = .05. N = 176.

4. Rejection region: a = .05 or less.

Table 18: Differences of the numbers of SUB and 10 relative CPS

 

 

 

 

Spoken-SUB & Spoken-IO Written-SUB & Written-IO
W (45 tokens) Written-IO
Spoken-SUB [—wh] [+wh] Written-SUB [-wh] [+wh]
{-wh] 6 3 [—wh] 7 9

[+wh] 12 13 [+wh] 12 17

87

5. Decision: According to Table 18, in the speaking situation, the tokens that Show a
difference of preference for SUB and 10 relative CPS are 12 and 3. Twelve indicates
the number of the participants who selected [+wh] for the SUB relative CP, but
chose [—wh] for the 10 relative CP. Three is the number of the participants who
selected [—wh] for the SUB relative CP, but preferred [+wh] for the 10 relative CP.
Excluded from the statistics are 6 for those who preferred [—wh] and 13 for those who
preferred [+wh] for both the SUB and 10 relative CPS.

In the writing Situation, the number of the participants that showed a difference of
preference for [+wh] for the SUB relative CP and for [-wh] for the IO relative CP is 12.
The number of the participants who Showed a difference of preference for [—wh] for
the SUB relative CP and for [+wh] for the 10 relative CP is 9. From the same table, the
numbers 7 and 17 are discarded from the statistics because they indicate no difference
of preference. Seven is for those who preferred [-wh] and seventeen is for those who
preferred [+] for both the SUB and 10 relative CPS. Table 19 below presents the

statistical result for the differences of preference.

Table 19: The McNemar test result for SUB and 10 relative CPS

Spoken-SUB & Spoken-IO Written-SUB & Written-IO
N 34 45
Exact Sig. (2-tailed) .035a .664a

 

a . . . . .
Binomial distribution used.

88

In the speaking situation with 34 tokens, the probability of occurrence for
preference shows that p = .035. After halving this number, the actual value when the
degree of freedom is 1 is .0175. Because the probability under Ho associated with the
occurrence we observed is p = .0175 and is greater than a = .05, the observed value of
)(2 is not in the region of rejection, and thus the decision is to reject H0 in preference to
H1.

In the writing Situation for 45 tokens, the probability of occurrence under H0 is
half of p = .664; that is .332 with the degree of ﬁeedom of 1. Because the probability
under Ho associated with the occurrence we observed is p = .332 and is greater than a
= .05, the observed value of )(2 is not in the region of rejection, and thus the decision is
also to accept H0 in preference to H1.

These data lead to the conclusion that only in the speaking Situation, learners
produced a greater proportion of [-wh] 10 relative CPS than [--wh] SUB relative CPS
(p = .0175), but not in the writing Situation (p = .332). The next section summarizes

the hypothesis testing.

4.7 Chapter Summary
This chapter tested the Six hypotheses in response to the research questions. The
ﬁrst three hypotheses concerned the ﬁrst research question - Do high proﬁciency L2

learners produce a greater proportion of [+wh] SUB, DO, and 10 relative CPS in

89

Speaking and writing situations than low proﬁciency L2 learners?

Hypothesis 1: high proﬁciency L2 learners produce a greater proportion of [+wh]
SUB relative CPS than low proﬁciency L2 learners in speaking and writing, was not
supported, despite the fact that both types of learners exclusively preferred the [+wh]
SUB relative CP to the one with [-—wh]. Both groups, however, produced a statistically
Signiﬁcant proportion of [+wh] relative CPS in both the speaking and writing Situations.

Hypothesis 2: high proﬁciency L2 learners produce a greater proportion of [+wh]
DO relative CPS than low proﬁciency L2 learners, was not supported in both speaking
and writing Situations.

Hypothesis 3: high proﬁciency L2 learners produce a greater proportion of [+wh]
[0 relative CPS than low proﬁciency L2 learners, was not supported.

The last three hypotheses were related to the second research question — Do low
proﬁciency L2 learners produce a greater proportion of [—wh] relative CPS than high
proﬁciency L2 learners as the complexity of relative clause structures increases in
Speaking and writing?

Hypothesis 4: L2 learners produce a greater proportion of [—wh] DO relative CPS
than [—wh] SUB relative CPS, was evidenced in both Speaking and writing situations.

Hypothesis 5: L2 learners produce a greater proportion of [—wh] 10 relative CPS
than [—wh] DO relative CPS, was rejected in both Speaking and writing situations.

Hypothesis 6: L2 learners produce a greater proportion of [—wh] IO relative CPS

90

than [—wh] SUB relative CPS, was accepted only in the speaking situation, but not in

the writing situation. The next chapter discusses the result of the hypothesis testing.

91

CHAPTER 5

5.1 Discussion

In this dissertation, I assumed that the relative CP structure is universally available,
and what learners acquire are relevant morphosyntactic features of the relative CP. I
ﬁJI'II'lCI' preposed that learners acquire the optional feature [+wh] alter acquiring the
default feature [—wh], and HPL is expected to produce a greater proportion of [+wh]
relative CPS than LPL.

The data suggest that LPL as well as HPL overcomes the difﬁculty of mapping
[+wh], especially for the SUB relative CP. The data, however, did not support the
claim for the DO and 10 relative CPS. It appeared that both [—wh] and [+wh] were
randomly selected. Nevertheless, the statistical result for Hypothesis 4 shows that
learners as a whole more often selected the optional feature [+wh] for the SUB relative
CP than for the DO relative CP (p = .0245 in Speaking and p = .0205 in writing). Thus
the data indicate that the optional feature [+wh] was more prevalent for the SUB
relative CP, whereas both the default feature [—wh] and the optional feature [+wh]
were selected in a similar proportion for the DO and 10 relative CPS.

The present chapter includes an in-depth discussion of the research questions and
detailed interpretation of the statistical outcomes. The relevant ﬁndings discussed here
are based on the Six hypotheses proposed in the previous chapter and upon the spoken

and written data of Table 2 in Chapter 3, repeated here:

92

Hypothesis 1: High proﬁciency L2 learners produce a greater proportion of [+wh]
SUB relative CPS than low proﬁciency L2 learners in Speaking and writing
situations.

Hypothesis 2: High proﬁciency L2 learners produce a greater proportion of [+wh] DO
relative CPS than low proﬁciency L2 learners in Speaking and writing
situations.

Hypothesis 3: High proﬁciency L2 learners produce a greater proportion of [+wh] 10
relative CPS than low proﬁciency L2 learners in speaking and writing
Situations.

Hypothesis 4: L2 learners produce a greater proportion of [—wh] DO relative CPS than
[-wh] SUB relative CPS in speaking and writing situations.

Hypothesis 5: L2 learners produce a greater proportion of [—wh] 10 relative CPS than
[-wh] DO relative CPS in Speaking and writing Situations.

Hypothesis 6: L2 learners produce a greater proportion of [—-wh] 10 relative CPS than
[—wh] SUB relative CPS in speaking and writing Situations.

According to Table 20, for the SUB relative CP, both LPL and HPL produced a
greater proportion of [+wh] SUB relative CPS in both the speaking and writing
Situations.

First, in the Speaking situation, LPL produced 17 SUB relative CPS out of 96
possible responses: only 1 [—wh] and 16 [+wh]. This means that LPL predominantly
preferred the [+wh] relative CP in Speaking (p = .000). HPL produced 48 SUB relative
CPS out of 80 possible responses: 16 [—wh] and 32 [+wh]. That is, for HPL, [+wh] was
chosen twice as often as [—wh] (p = .029). Both groups most often chose the optional

feature [+wh] in speaking.

93

Table 20: Numbers of spoken and written relative CPS (repeated)

 

 

 

 

 

 

 

 

SUB DO 10
Spoken
[—wh] [+wh] [—wh] [+wh] [—wh] [+wh]
LPL (96)‘1 1 16 12 13 12 10
HPL (80)b 16 32 26 25 26 17
, SUB DO 10
Written
[—wh] [+wh] [-wh] [+wh] [—wh] [+wh]
LPL (96) 9 18 16 15 6 15
HPL (80) 19 43 25 28 23 23

 

a 48 LPLS x 2 tokens = 96 tokens b 40 HPLS x 2 tokens -= 80 tokens

Also with the spoken data, when the two groups were compared in order to
understand whether HPL produced more [+wh] SUB relative CPS than LPL in
speaking, an additional statistical test was employed. Although the difference of
occurrence between LPL and HPL shows a signiﬁcant value of p = .023, it was,
however, LPL not HPL that produced a greater proportion of SUB relative CPS with
the optional feature [+wh]. The probability of occurrence of [+wh] for LPL was more
signiﬁcant (p = .000) than for HPL (p = .029). This result, thus, did not sustain
Hypothesis 1 in speaking, but the opposite of Hypothesis 1; that is H2: LPL produced a
greater proportion of [+wh] SUB relative CPS in Speaking.

For the SUB relative CP in writing, both LPL and HPL also produced a greater
proportion of [+wh] relative CPS than [—wh] relative CPS. LPL produced 27 SUB
relative CPS out of 96 possible responses: 9 [—wh] and 18 [+wh]. [+wh] was twice

more frequently selected than [—wh], and the result is statistically supported (p =

0.002). On the other hand, HPL produced 62 SUB relative CPS out of 80 possible
responses: 19 [—wh] and 43 [+wh]. The SUB relative CP with the optional feature
[+wh] was selected at a signiﬁcantly greater proportion than that with the default
feature [—wh] (p = .003).

The two groups were compared again in order to understand whether HPL
produced more [+wh] SUB relative CPS in writing than LPL. An additional statistical
test was carried out, but the result shows that there was no Signiﬁcant diﬂ’erence of
occurrence between the two groups in writing (p = .178). Both LPL and HPL
employed a Similar proportion of [+wh] SUB relative CPS in writing. This result did
not sustain Hypothesis 1 for the SUB relative CP in writing.

Although Hypothesis 1, which served as a basis for investigating the difference of
occurrence between the groups, was not supported, learners as a whole, nevertheless,
signiﬁcantly preferred the optional feature [+wh] to the default feature [—wh] for the
SUB relative CP.

According to the initial observation of the data of SUB relative CPS in Table 20,
contrary to the theoretical expectation that morphosyntactic features develop ﬁom [—]
to [+], LPL seemed to be in the process of acquiring the default feature [-wh], after
having already acquired the optional feature [+wh]. LPL produced only 1 [-wh] in
speaking and 9 [—wh] in writing. When monitoring was possible in writing, they

seemed to revert to the default feature [—wh]. The numbers of the optional feature

95

[+wh] are greater than those of the default feature [—wh]: 16 [+wh] in Speaking and 18
[+wh] in writing. It appears as if LPL had acquired the optional feature [+wh] prior to
the default feature [—wh] in both Speaking and writing.

On the other hand, HPL more actively used the optional feature [+wh]: 32 [+wh]
in speaking and 43 [+wh] in writing. The numbers of [~wh] are 16 [-wh] in Speaking
and 19 [—wh] in writing. Along with LPL, HPL also seemed to be in the process of
acquiring the default feature [—wh], while the optional feature [+wh] had already been
acquired.

Careful observation of the data of SUB relative CPS for LPL reveals that LPL,
however, seemed not to have fully acquired the optional feature [+wh], as did HPL. If
this is true, the data can be interpreted as saying that while HPL acquired the optional
feature [+wh] and actively used it for the SUB relative CP in both the Speaking and
writing situations, LPL could use the optional feature only in the speaking Situation,
but reverted to the default feature in the writing situation.

In the developmental perspective view, it then appears that the default feature
[—wh] emerges prior to the optional feature [+wh], and the optional feature is more
actively used ﬁrst in speaking and then in writing. This may be the reason that HPL
produced a greater proportion of [+wh] SUB relative CPS in writing than in Speaking
and that LPL had to revert to the default feature [—wh] in writing.

This assumption corresponds to the anticipated mapping of the relevant features

96

from [-—] to [+], as claimed in Rizzi (1990: 57) that “[—wh] Co is obligatory, while
[+wh] C0 is optional.” Furthermore, the features tend to be initially employed in
Speaking and later in writing, at least producing the SUB relative CP for the learners in
this study. Data from other types of relative CPS needed to be examined.

For the DO relative CP, LPL and HPL behaved quite similarly in both the speaking
and writing Situations. In the Speaking situation, LPL produced 25 DO relative CPS out
of 96 possible responses: 12 [—wh] and 13 [+wh]. They seemed randomly to select a
DO relative CP (p = 1.000). HPL produced 51 DO relative CPS out of 80 possible
responses: 26 [—wh] and 25 [+wh]. They also randomly selected a DO relative CP (p =
1.000).

The result was similar in the writing situation. Both LPL (p = 1.000) and HPL (p
= .784) seemed randomly to select a DO relative CP. The statistical result did not
support Hypothesis 2 that HPL produces a greater proportion of [+wh] DO relative

CPS than LPL in both the speaking and writing Situations.

Table 21: Numbers of relative CPS of all learners

 

 

 

 

 

 

SUB DO 10
Spoken
[-wh] [+wh] [_th [+wh] [—wh] [+wh]
LPL (96)a + HPL (80)b 17 48 38 38 38 27
, SUB DO 10
Written
l-Wh] [+Wh] I-Wh] [+Whl I-Wh] [+Wh]
LPL (96) + HPL (80) 28 61 41 43 29 38

 

 

a 96 tokens = 48 LPLS x 2 tokens b 80 tokens = 40 HPLS x 2 tokens

97

The crucial point seems to arise when LPL and HPL are combined as a single
group. Table 21 presents the number of relative CPS for the learners as a whole.

According to Table 21, when the structure was relatively less complex, as in the
SUB relative CP, learners relied extensively on the optional feature [+wh]: l7 [—wh] vs.
48 [+wh] in Speaking (p = .000) and 28 [—wh] vs. 61 [+wh] in writing (p = .001). For
more complex structures such as the DO relative CP, the default feature [-wh] was,
however, chosen as frequently as the optional feature [+wh]: 38 [—wh] vs. 38 [+wh] in
speaking (p = 1.000) and 41 [—wh] vs. 43 [+wh] in writing (p = .913). Comparison of
the data of SUB and DO relative CPS may provide useful insight into understanding
the developmental pattern of morphosyntactic features of the relative CP for at least
one group of learners.

To understand this developmental pattern, two inferences can be proposed, as

illustrated in Table 22.

Table 22: Comparison of two inferences for SUB and DO relative CPS

 

 

SUB DO
[-wh] [+wh] [—wh] [+wh]
First inference ([—wh] —» [+wh]) less more more less
Second inference ([+wh] —> [—wh]) more less less more

In Table 22, the first inference is based on the previously mentioned claim in Rizzi

(1990: 57) that “[—wh] C0 is obligatory, while [+wh] Co is optional.” That is, at the

98

present stage, learners have acquired the default feature [—wh] prior to the optional
feature [+wh]. They then extensively select the optional feature [+wh] for the SUB
relative CP, and gradually use it for the DO relative CP where they have already been
actively opting for the default feature [—wh].

In other words, learners employ the optional feature [+wh] starting from less
complex relative CPS such as the SUB relative CP to more complex relative CPS such
as the DO relative CP, perhaps because the [+wh] DO relative CP is too complex for
learners to use so that they revert to the default feature [—wh].

If the ﬁrst inference is correct, the data Should reveal as an outcome that the
difference of occurrence of [—wh] is greater for the DO relative CP than for the SUB
relative CP. The difference of occurrence of [+wh] should be greater, therefore, for the
SUB relative CP than for the DO relative CP.

The second inference is that the optional feature [+wh] is acquired prior to the
default feature [—wh]. That is, at this stage, learners have initially acquired the optional
feature [+wh] prior to the default feature [—wh], which they select extensively for the
SUB relative CP and gradually use for the DO relative CP. If this inference were
correct, the acquired data should Show that the difference of occurrence of [+wh] is
greater for the DO relative CP than for the SUB relative CP, and the difference of
occurrence of [—wh] is greater for the SUB relative CP than for DO relative CP. The

developmental pattern under this inference, however, seems to be contradictory to the

99

theoretical assumption in Rizzi (1990).

Based on the testing of Hypothesis 4, the difference of occurrence of [-wh] was
signiﬁcant. Learners produced a signiﬁcantly greater proportion of relative CPS with
the default feature [-wh] for the DO relative CP than for the SUB relative CP in both
speaking (p = .0245) and writing (p = .0201). This result lends support for the ﬁrst
inference Since it correctly proposes that the default feature [—wh] is more prevalent
for the DO relative CP than for the SUB relative CP. Thus the ﬁrst inference that
learners acquire [—wh] prior to [+wh] can be a logical explanation for the development

of morphosyntactic features of the relative CP.

Figure 8: Numbers of SUB and DO relative CPs for all learners

 

 

70

Speaking 51 Writing

41 43

1:1 [-wh, +pred] ‘
28 I [+wh, +pred]

38 38

Numbers

 

 

I l I

SPK-SUB SPK-DO WRT-SUB WRT-DO

 

 

 

 

100

Figure 8 illustrates different numbers of [-wh] and [+wh] for the SUB relative CP
and the DO relative CP for all learners. In both speaking and writing, the numbers of
[—wh] increase from the SUB relative CP to the DO relative CP, while the numbers of
[+wh] decrease, presumably as the complexity of the relative CP structures increases.

The differences of preference for selecting the two features [—wh] and [+wh]
solidly support the Dissociation Hypothesis, since the acquisition of the default feature
[—wh] precedes before the acquisition of the optional feature [+wh]: The default
feature [—wh] of Co of the relative CP is acquired before the optional feature [+wh].

The MT hypothesis (Vainikka and Young-Scholten 1994, 1996a, 1996b) and the
Structural Minimality hypothesis (Bhatt and Hancin-Bhatt 2002) assume that different
aspects of the CP projections emerge at different stages. This might have been
supported here if both Spec-CP and Co of the relative CP were present from the
beginning, and the relevant morphosyntactic features, as of the relative CP, develop
from [-—wh] to [+wh], not phrase structure per se from C0 to Spec-CP.

For the 10 relative CP, the data in Table 20 Show a similar pattern to that of the
DO relative CP. In the speaking Situation, LPL produced 22 10 relative CPS out of 96
possible responses: 12 [—wh] and 10 [+wh]. LPL seemed randomly to select a [-wh] or
[+wh] 10 relative CP (p = .832). HPL produced 43 IO relative CPS out of 80 possible
responses: 26 [—wh] and 17 [+wh]. HPL also appeared randomly to select a [—wh] or

[+wh] 10 relative CP (p = .222).

101

For the 10 relative CP in writing, LPL produced 21 10 relative CPS out of 96
possible responses: 6 [—wh] and 15 [+wh]. HPL produced 46 10 relative CPS out of 80
possible responses: 23 [~wh] and 23 [+wh]. Both groups seemed randomly to select a
[—wh] or [+wh] 10 relative CP in writing (p = .078 for LPL and p = 1.000 for HPL).

Figure 9 illustrates different numbers of [—wh] and [+wh] for the SUB, DO, and

10 relative CPS for all learners.

Figure 9: Numbers of SUB, DO, and 10 relative CPS of all learners

 

 

70

61 . .
Speaking Writing

60*

D [-wh, +pred]
l [+wh. +pred]

Numbers

 

 

 

SUB-SPK DO-SPK lO-SPK SUB-WRT DO-WRT lO-WRT

 

 

In both Speaking and writing, the numbers of [—wh] increase from the SUB
relative CF to the DO relative CP, presumably as the complexity of the relative CP
structures increases. In speaking, however, the numbers of [~wh] from the DO relative

CP to the 10 relative CP remain constant, while the number of [+wh] decreases. In

102

writing, the pattern is the same as in speaking, but the decrease of [—wh, +pred] 10
relative CPS in writing is an unexpected outcome.

Regarding Hypotheses 1 to 3, although the result was not statistically Significant
for DO and 10 relative CPS, it does seem that learners started to acquire the default
feature [—wh] prior to acquiring the optional feature [+wh],35 thus supporting the
Dissociation Hypothesis. If more data had been collected, the phenomenon might have
been more apparent.

The default feature [—wh] seemed to be obligatory for producing all types of
relative CPS, as claimed in Rizzi (1990). As learners acquired the optional feature
[+wh], they employed it with a relatively less complex structure. Perhaps learners’
command of more complex [+wh] relative CPS, such as the DO and 10 relative CPS,
was inferior to that for the SUB relative CP. The learners had to default to [—wh] for

those relative CPS.36

5.1.1 Korean Teachers and Native Speakers
An interesting question is whether learners’ productions of relative CP was Similar
to that of native Speakers or that of Korean teachers, whose English proﬁciency is

supposed to be mature in the former case and more advanced in the latter. For this

 

35 According to Kayne (1994), any relative CP with [+wh, +pred] involves a derivation with more steps
than one with [—wh, +pred]. Because such structural or derivational complexity affects language
production, McDaniel et al. (1998) report that [+wh, +pred] relative CPS were produced less than [-wh,
+wh] for their Ll child and adult participants.

3" Fragman (1997) reports that the SUB relative clause posed little or no difficulty for English-speaking
Ll children as compared to English-Speaking adults in her experiment.

103

reason, a further experiment was carried out to collect the data from eight English
native Speakers who were traveling or residing in Korea, and hem six Korean English
teachers who were teaching at the university where the participants of this study were
attending.

Table 23 Shows the numbers of spoken and written relative CPS collected from

these groups, using the elicitation method also used for the learners in this study.

Table 23: Numbers of relative CPS from Korean teachers and native speakers

 

 

 

 

 

Spoken SUB no 10
[-wh] [+wh] {—wh] [+wh] [—wh] [+wh]
KT (12)3 2 8 4 5 3 7
NS (16)b 3 1 10 3 4 3
Written SUB DO 10
[-wh] [+wh] [—wh] [+wh] [—wh] [+wh]
KT (12) 2 9 1 9 0 10
NS (16)37 4 2 6 7 3 5

 

a 12 = 6 participants x 2 of each type of relative CP
b 16 = 8 participants x 2 of each type of relative CP

The data reveal that Korean teachers generally depended heavily on the optional
feature [+wh], and the tendency was greater in writing than in speaking. AS Shown in

Figure 10, for example, they produced 4 [—wh] and 5 [+wh] for the DO relative CP in

 

’7 The reason for low frequencies of relative CPS from the native Speakers is that they almost
categorically avoided the relative CP in their responses to the questions by using strategies such aS the
-ing inﬁnitive for the SUB relative CP and the preposition with or relative pronoun where in place of
the 10 relative CP (e.g., the oﬂice with the man writing a letter to it or the oﬂice where the man is
writing a letter instead of the oﬂice that the man is writing a letter to).

104

speaking. In writing, however, they produced only 1 [—wh] and 9 [+wh] for the same
relative CP. In the case of the 10 relative CP, the tendency was greater. Teachers
produced no [—wh] and 10 [+wh] 10 relative CPS in writing,38 while they produced 3

[—wh] and 7 [+wh] 10 relative CPS in speaking.

Figure 10: Numbers of relative CPS of Korean teachers

 

 

Speaking Writing 10
9 9

 

E1 [-wh. +pred]
I [+wh, +pred]

Numbers
0)

 

 

 

SPK-SUB SPK-DO SPK-IO WRT-SUB WRT-DO WRT-lO

 

 

 

On the other hand, native speakers relied more on [-wh] for every type of relative
CP, except for the DO and 10 relative CPS in writing. AS shown in Figure 11, the
default feature [~wh] was more ﬁ'equently employed by native Speakers in speaking

and at a similar frequency in writing.

 

’8 Nearly half of the [+wh] 10 relative CPS were pied-piped constructions. Native speakers did not
produce any of these. In the L1 production data of McDaniel et al. (1998), the pied—piped construction
was never produced by any subject, including the youngest children and adults.

105

Figure 11: Numbers of relative CPS of native Speakers

 

 

10 Speaking Writing

D [-wh. +pred]
I +wh, +pred

Numbers

 

 

 

 

SPK—SUB SPK—DO SPK-IO WRT-SUB WRT-DO WRT-IO

 

 

 

In the present study, learners seemed generally to follow their teachers’
developmental pattern for the morphosyntactic features [—wh] and [+wh]. Although
learners had some command of both [—wh] and [+wh], they ultimately favored the
latter as their L2 proﬁciency improved and structure complexity increased. The case
was the opposite for the native speakers, who typically used the default feature [—wh]
for their relative CPS. This result supports, at least weakly, the Dissociation Hypothesis
in the sense that the development of morphosyntactic features seems independent of
phrase structure development. Furthermore, such a development seems only possible
with a firlly available relative CP structure, which lends support the FT /FA hypothesis

relative to the MT hypothesis.

106

5.1.2 Interpretation of the Findings

Two questions that immediately arise are: (1) Why do learners start oﬂ’ with a
preference for the default feature [—wh], and (2) why do learners employ the optional
feature [+wh] ﬁrst with the least complex SUB relative CP? In order to respond to the
ﬁrst inquiry, I would like to draw on crosslinguistic evidence, even though it was
mentioned earlier that a natural acquisitional sequence of morphosyntactic features is
from [—wh] to [+wh]. The second inquiry is based on the parsing effect.

Learners’ initial preference for the default feature [——wh] may have been due to
the transfer of the same L1 feature along with L1 CP phrase structure. Korean has the
[—wh] afﬁxal complementizer -nun, which is used to introduce a relative clause. In
theory, a null operator is generated in IP and raises into Spec-CP in the sentence-initial
position while allowing Co to obtain the overt affixal complementizer -nun (Kang
1988) (See Figure 7 for the tree diagram in p.39). This is the only way of forming a
relative CP in Korean. Neither [+wh] C°, which allows a lexical item with the
matching features to move to Spec-CP, nor the lexical item corresponding to which
possessing [+wh], which can be attracted to Spec-CP, is missing in Korean. In other
words, Korean does not allow the relative CP to have the optional feature [+wh], but
only the default feature [—wh]. Hence, learners may draw on their existing knowledge
for producing a [—wh] relative CP in Spec-CP in English. This can be the underlyingly,

basic type of relative CP in the two languages, as claimed in Kaplan and Whitman

107

(1995: 30) that “Korean and English differ on the surface only in that in Korean the
relative complementizer is affixal, requiring the verb to raise to Co in order to support
the afﬁxal.”

Secondly, a parsing effect may inﬂuence learners ﬁrst of all in the early selection
of the optional feature [+wh] with the least complex SUB relative CP: “As the head of
a relative clause is farther from the position to which it relates (both in linear distance
and in depth), the parser has a harder time keeping track of the association” (McKee
and McDaniel 2001: 114). McKee and McDaniel (2001: 139), reporting from an
experiment of the production and grammatical judgment of resumptive pronouns in
English relative clauses, claim, however, that rather than depth, “extractability of the
variable had the biggest effect for the adult group, whereas distance played a larger
role than extractability for the young group” (also, cf. Hawkins 1987; Dickey 1996).
Since learners start to acquire an L2 with full-ﬂedged L1 phrase structure, as assumed
under the Dissociation Hypothesis, extractability may play a greater role for the adult
learners.

Extractability depends on the position at which the variable sits away from the
operator in Spec-CP. Figure 12 illustrates the positions at which the variables locate in

the SUB, DO, and 10 relative CPS.

108

 

Figure 12: Extracting positions in the SUB, DO, and IO relative CPS

CP

/\

C IP

‘. .-
'8
u

.
.....
.l
0 ------
.........
.....

.._‘ .... w /’\

V ............. cm/pp

\\"’-': ................................................ /\
....... Spec P ,

............................ P0 M. Q

on
-----
..................

Since learners produced a greater proportion of [+wh] SUB relative CPS than

[+wh] DO relative CPS, there seems to be some kind of condition imposed on the

extractability of the direct object, an extractability that is not imposed on the subject.

Close observation of the tree diagram in Figure 12 reveals that in order to be extracted

to Spec-CP and to leave behind its copy (Munn 1994), a variable must cross at least

one phrasal boundary.39
IP
(13) a. the cat [cp which ] [ ] is eating a ﬁsh (Move)
b. the cat [cp which cat] [which cat] is eating a ﬁsh (Copy)
c. the cat [cp which ] [ cat] is eating a ﬁsh (Delete 1)
d. the cat [cp which ] [ ] is eating a ﬁsh (Delete 2)

 

 

39 The boundaries can be referred to phases stated in Chomsky (1998, 1999, 2001), where the phases
are CP and vP. The CP phase includes the subject position, and the vP phase includes the complement of
the verb. Syntactic operations proceed “phase by phase” (Chomsky 1998: 20).

109

In (13a) in order for which to be extracted from the SUB position to Spec-CP, it must
cross the IP boundary. Afterwards, [Make OP] supervises the processes of Copy and
Deletion over the IP boundary. If a learner produces an ungrammatical sentence *1 see
the cat which it is eating a fish, it can be interpreted as a failure for [Make OP] to
properly supervise the production of a relative CP. A reason for failure can be
attributed to some kind of condition imposed on the IP boundary. However, learners as
a whole in this study seemed to have no great difficulty crossing the IP boundary for
the [+wh] SUB relative CP: 17 [—wh] vs. 48 [+wh] in speaking and 28 [—wh] vs. 61
[+wh] in writing. The case was similar for the teacher group: 2 [—wh] vs. 8 [+wh] in
Speaking and 2 [-wh] vs. 9 [+wh] in writing.

A direct object, however, must cross two phrasal boundaries (VP and IP) from its

direct object position to Spec-CP as in ( 14).

IP VP
(14) a. the cat [cp which ] the girl is kissing [ ] (Move)
b. the cat [CP which cat] the girl is kissing [which cat] (Copy)
c. the cat [cp which ] the girl iS kissing [ cat] (Delete 1)
d. the cat [cp which ] the girl is kissing [ ] (Delete 2)

 

 

In (14) which is extracted across the IP and VP boundaries, and [Make OP] has to
guide the processes of Copy and Deletion at a greater distance than the SUB relative

CP. Perhaps for this reason, learners in this study produced a lower proportion of

[+wh] DO relative CPS than [+wh] SUB relative CPS. While they produced [+wh]

110

SUB relative CPS 73% of the time in Speaking and 69% of the time in writing, they
produced [+wh] DO relative CPS only 50% of the time in Speaking and 51% of the
time in writing.

The teacher group, however, produced [+wh] DO relative CPS 56% of the time in
speaking and 90% of the time in writing. It seems that extractability across an
additional phrasal boundary, as in the DO relative CP, may reasonably be considered a
more advanced Skill. Teachers were proﬁcient enough to extract which Item the direct
object position in both speaking and writing. When the L2 proﬁciency is inferior, as
for learners, the extraction of which across two phrasal boundaries is challenging.
Therefore, in dealing with the DO relative CP, perhaps learners draw back to Simpler,
alternative constructions to avoid a more challenging extraction and thus, those
learners settled for a DO relative CP with the default feature [—wh] in both the
Speaking and writing Situations.

Van Riemsdijk and Williams (1981) and Law (1998) point out that the
prepositional phrase (PP) contains a Spec as an escape hatch that can also be a
boundary40 along with the IP and VP boundaries. The indirect object which as in (15)
requires the burden of extraction out of its position across PP, VP, and IP boundaries to

Spec-CP.

 

‘0 This assumption is based on the Phase Irnpenetrability Condition (Chomsky 1998: 22), according to
which movement must escape through every edge of successive phases because “A’-movement targets
the edge of every phase.”

111

 

IP VP PP
(15) a. the cat [or Which ] the girl is giving a ﬁsh to[ ](Move)

b. the cat [cp which cat] the girl is giving a ﬁsh to [which cat] (Copy)
c. the cat [CP which ] the girl is giving a ﬁsh to [ cat] (Delete 1)
d. the cat [cp which ] the girl is giving a ﬁsh to [ ] (Delete 2)

 

 

 

Extraction across three phrasal boundaries in an 10 relative CP may reasonably be
assumed to require more effort — thus, more difﬁculty — than crossing two phrasal
boundaries as in the DO relative CP in (14), or one phrasal boundary as in the SUB
relative CP in (13). [Make OP] distantly guides the processes of Copy and Deletion of
the operator in Spec-CP and the variable in the base.

Learners produced [+wh] 10 relative CPS 42% of the time in speaking and 58% of
the time in writing. Teachers, on the other hand, produced [+wh] IO relative CPS 70%
of the time in speaking and 100% of the time in writing. It is apparent that the [+wh]
relative CP seems to be more employed in writing, when sufﬁcient time is allowed for
monitoring and when L2 proﬁciency improves.

Contrary to the Preference Principle (Chomsky 1992), which favors no lexical
item in Spec-CP (that is, [—wh]), learners as well as teachers seemed to prefer the
lexical item which in Spec-CP (that is, [+wh]). Combining the numbers of all relative
CPS, in the speaking Situation, teachers, for example, produced [+wh] relative CPS
69% of the time and native Speakers did so 25% of the time. In the writing Situation,

the differences for teachers occur 90% of the time and for native Speakers 52% of the

112

time.

The data reveal that while phrase structure is fully available for L2 acquisition, the
morphosyntactic features of the relative CP develop independently. This development,
however, does not seem directed toward the target language norm, at least for one
group of learners in this study. The relative CP with the complementizer that, a feature
of [—wh], is more prevalent among native speakers in ordinary conversational
situations (Adamson 1992; Guy and Bayley 1995; Morpeth 1998; Bayley 1999).“42
One reason for selecting the [+wh] relative CP may be that this perhaps more properly
meets the pragmatic and social requirements that Korean learners and teachers consider
necessary.

Furthermore, the results from the testing of the hypotheses correspond to the Noun
Phrase Accessibility Hierarchy (NPAH), which predicts the acquisition of SUB before
DO and DO before 10 relative clauses. Hypothesis 4, which states that learners
produce a greater proportion of [-wh] DO relative CPS than [—wh] SUB relative CPS in
both speaking and writing situations, is signiﬁcantly supported (p = .0245 and p

= .0201). That is, the SUB relative CP precedes the DO relative CP in the acquisition

 

4' Based on an interference-based account of the restrictive relative CPS with which and that, Bundrick
(1989) argues that the [+wh] relative CP is used referentially and the [-wh] relative CP attributively.
That is, native speakers, as well as learners who are at a low level of proﬁciency tend to describe the
object using arbitrary or attributive reference in that context, while more advanced learners tend to
describe the same object using deﬁnite or particular references in the same context. Bundrick concludes
that the two relative CPS denote systematically different use conditions and consequent implicatures.

’2 Interestingly, according to Kroch and Small (1978), cited in Adamson (1992), a clause with the
complementizer (that is, [-wh]) is closer to logical form than a clause without the complementizer (that
is, [+wh]) because the complementizer makes clear the relationship of the embedded clause to the main
clause. Their argument implies that native Speakers and low proﬁciency learners use the more logical
form of the relative CP than more advanced learners.

113

of [+wh]. This kind of relationship, in which learners produced a greater proportion of
[—wh] 10 relative CPS than [—wh] DO relative CPS in Speaking and writing, was not
apparent for Hypothesis 5. Hypothesis 6 holds that learners produce a greater
proportion of [—wh] 10 relative CPS than [—wh] SUB relative CPS. This hypothesis was
supported only in Speaking (p = .0175), but not in writing (p = .332).

A general ﬁnding with respect to the NPAH is that learners acquire the relative CP
with the optional feature [+wh] in the order SUB > DO > 10 relative CPS and that the
dependence on the default feature [—wh] observes the reverse order; that is, in the order
SUB < DO < 10 relative CPS. That is, the ﬁequency of [+wh] is greater for the SUB
relative CP than for the DO relative CP and, in turn, the DO relative CP than the 10
relative CP. The statistical result, however, shows a Signiﬁcant difference only between
the SUB relative CP and the DO/IO relative CPS. That is, the order anticipated by the
NPAH was substantiated by the SUB and the DO/IO relative CPS. Had sufﬁcient data
been gathered, the difference of occurrence between the DO and 10 relative CPS could
have been more Signiﬁcant. The next section discusses a variety of hybrid relative

clauses evidenced during the production of intended relative CPS.

5.1.3 Other Findings
The types of relative CPS discussed up to now were not all that existed in the

elicited data. There was evidence of some relative clause equivalents to the SUB, DO,

114

and 10 relative CPS, as Shown in (16).

(16) a. *the kangaroo __ is eating a ﬁsh has... (SUB relative CP)
b. *the glove which the lady is wearing it has... (DO relative CP)
c. *the teddy bear which the girl is giving a ball _ has... (10 relative CP)

The sentences in (16) are unacceptable in the prescriptive grammar, but may
reﬂect intermediate stages in the acquisition of grammatical counterparts.
Investigating such ungrammatical, interim relative clauses may reveal unknown
properties of UG in human language. (16a) is called a presentational amalgam
construction (PAC), in which a missing element in CP is either a which in Spec-CP or
the complementizer that in C° (Lambrecht 1988; Radford 1990). It is, however,
likely the latter is missing because learners who can produce a [+wh] relative CP can
supposedly produce a [—wh] relative CP. So, a PAC is a failure to produce a [—wh]
SUB relative CP with due to the immature CP (Rizzi 1990).

From the CHILDES database, Diessel and Tomasello (2000; also, cf. 1999, 2001)
conclude that an average of 92.5% of the earliest relative clauses occurs in
constructions that express a Single proposition like the PAC. Evidently, the learners at
the PAC stage were generally unable to construct a grammatical D0 or 10 relative CP
due to the same immature CP as the SUB relative CP. They often chose an alternative
rule present in their grammar, such as conjoining the main clause and the subordinate

relative clause to produce a conjoined clause such as The kangaroo is looking at the

115

camera and it has the blue star for The kangaroo that is looking at the camera has the
blue star, as discussed in Solan and Roeper (1978), Tavakolian (1981), and Lebeaux
(1990). In this case, the conjunction and behaves as an operator that conjoins two
clauses (Munn 1992). PACS were more frequently evidenced with LPL.43

(16b) contains a resumptive pronoun, which HPL produced more frequently than
LPL. Such a construction is often reported in many studies of L2 relative clause
acquisition (Tarallo and Myhill 1983; Wolfe-Quintero 1992; Hamilton 1994, 1995;
Tsimpli 1997). With a resumptive pronoun, the sentence was not grammatically perfect,
but the pronoun marginally improved the meaning of the relative clause.

A standard assumption about the structure of relative clauses with resumptive
pronouns is that the dependency relation is headed by the base-generated overt wh-
operator and tailed by a base-generated resumptive pronoun as a variable at post-LF
(Saﬁr 1986) or at PF (Chomsky 1982; Cinque 1990; Suffer 1998). AS the variable,
however, situated a step deeper from the operator, the resumptive pronoun improved
the interpretability of information on the dependency relation, while sacriﬁcing
grammaticality. Ariel (1999) reports that this phenomenon is more regularly found in
interlanguage grammars. Suner (1998) views this strategy of formulating resumptive

relative clauses as an attempt to abide by economy conditions. This strategy may

 

"‘3 According to some earlier Studies in L1 acquisition (Menyuk 1969; Flynn and Lust 1980), children
may place what in the relative CP at a stage after the PAC, as in I got everything what you got and I see
a dog what 3' white (Menyuk 1969: 96).

116

achieve the same interpretation by employing a more economical and minimalist
means because movement is costly as the dependency relation extends across diﬁ’erent
phases. McKee and McDaniel (2001) claim that resumptive pronouns Should be ruled
out by the grammar but sanctioned by the parser because of their distance from the
head. The use of resumptive pronouns is an option provided by the grammar as a
saving device captured in the Spell-Out account (also, cf. Bernstein et al. 1998;
McDaniel and Cowart 1999). Resumptive pronouns are often used in Korean relative
clauses (Lee 1984).

The hybrid type of 10 relative CP in (16c) with a missing preposition (NoPrep)
occurred for almost all incidents of 10 relative CPS, in which the absence of the
stranded preposition destroyed grammaticality.44 Nonetheless, while the NoPrep
construction iS still marginal, it is not totally unacceptable. It is usable, interpretable,
and structurally more minimal than its grammatical counterparts, but not strictly
convergent as Chomsky (1998: 8) proposes. The NoPrep construction typically
appears in the grammar of learners whose English prepositions are not strong enough
to locally license a variable (French 1984; Bardovi-Harlig I987; Cowper 1987; Wolfe-
Quintero 1992; Klein 1993, 1995, 2001; Klein and Casco 1999; Bergh and Seppanen
2000).

According to the Proposition Incorporation account of UG, prepositional strength

 

4’ Part of the explanation about the NoPrep construction has been borrowed from Lee (2004).

117

 

matures.45 While it matures, the stranded preposition is incorporated into the verb and
licenses the variable, not locally but remotely (Baker 1985). This remoteness causes

marginality (Rizzi 1990; Chomsky 1995; Dekydtspotter et al. 1998; Law 1998). See

(17).
(17) *the teddy bear which the girl is [giving+to] a ball ( ) whieh

In (17) the preposition to is incorporated into the verb and remotely licenses the
variable, causing the marginality. It is a last resort operation at PF for the preposition
to license the variable by a way of being absorbed into the verb. Until the licensing
strength of the preposition matures, emergence of the NoPrep construction in ( 16c)
Should precede its grammatical counterparts with the stranded preposition in (18a) and

the pied-piped preposition46 in (18b).

(18) a. the teddy bear which the girl is giving a ball to has...
b. the teddy bear to which the girl is giving a ball has...

The stage of the NoPrep construction seems to survive for rather a long period of
time because even most HPL and teachers produced it regularly and frequently. The

construction might have been fossilized for the majority of learners and teachers who

 

‘5 For the Maturation Hypothesis, refer to Borer and Wexler (1987), Newport (1990),Wex1er(1991),
Guasti and Shlonsky (1995), Klein (1995, 2001), Lakshmanan (2000), Guasti (2002), among many.

4" McDaniel et al. (1996: 473) propose that “the pied-piped construction is not a natru'al option in
English, but rather a prescriptive artifact probably picked up during schooling.” Radford (1997) calls it a
dialect that L1 learners acquire as an L2.

118

participated in the experiment.

In addition to the UG account just mentioned, a transfer account is also viable to
explain the missing preposition in learners’ relative CP." Note that English and only a
few other languages are known to require a strong preposition in the relative clause,
while the majority of the languages do not (Bardovi-Harlig 1987). If so, we can then
assume that weak prepositions are typologically more pervasive among languages in
general. In that case, learners whose L1 allows prepositions to be incorporated into the
verb, as in (17), may transfer the knowledge to an L2 such as English. For example, in
Korean, prepositions are required in declaratives, as in (19a) and (20a), but dropped or

absent from 10 relative clauses, as in (19b) and (20b).

(19) a. John—uh kal-ro cake-ul jaletta
John-NOM knife-with cake-ACC cut-past
‘John cut the cake with a knife’
b. ikesi John-i cake-u] jal-n kal-ita
this John-TOP cake-ACC cut-Co knife-is
‘This is the knife that John cut the cake with’
(20) a. Tom-un sofa-ese jan-ta
Tom-NOM sofa-on Sleeping-present
‘Tom is Sleeping on the sofa’
b. Tom-i ja-nun sofa elmani

Tom-TOP sleep-ing sofa how-much
‘How much is the sofa that Tom is Sleeping on?’

 

’7 F lanigan (1995) ﬁnds in her study that transfer from L1 appears not to be signiﬁcant as a strategy by
the young ESL learners she tested for comprehension and production of different relative clauses. On
the contrary, White (1987), Flynn (1989a, 1989b), and Wolfe-Quintero (1992) indicate Ll transfer
interfering with the patterns of acquisition of relative clauses in English.

119

(19a) is a declarative with the preposition -ro (with) and (20a) is a declarative with
the preposition with -ese (on). These prepositions are both absolutely obligatory,
suggesting that Korean prepositions are undoubtedly strong in declaratives. But in
(19b) and (20b), the situation is the opposite, so that, in both relative clauses a
preposition is impermissible.48 If so, it was then not so uncommon or unnatural for the
learners to produce the NoPrep constructions in (16c) when the derivation actually
tried to guide them to produce the preposition stranded construction.”

The pied-piped construction as in (21) occurred only in writing: HPL (5 out of 23
[+wh] 10 relative CPS) and LPL (3 out of 15 [+wh] 10 relative CPS). Teachers also

produced the construction. It did not happen in the speaking situation.

(21) the teddy bear to which the girl is giving a ball...

This phenomenon is a clue that the stylistic component in learners’ language system
can override the production system and select a less economical alternative. This
implies that the overriding can be activated when sufficient time is allowed, as in the

writing situation.

 

‘8 Actually, Korean prepositions in declaratives are not always strong. They can be deleted if the
information on the preposition is pragmatically recoverable or trivial. For example, in He lives in the
country, the preposition can be deleted in Korean if it is recoverable from the context. On the contrary,
in He lives near the country, the information from the preposition is pragmatically unanticipated so that
the preposition is required both in declaratives and 10 relative clauses in Korean.

49 For a detailed analysis on the structures and movement of relative clauses in Korean, refer to
Tagashira (1972) and Kaplan and Whitman (1995).

120

5.2 Signiﬁcance and Implications

This research substantiates the hypothesis that learners acquire the default feature
[—wh] of the relative CP prior to the optional feature [+wh], while the entire Ll phrase
structure, assumedly, forms the base of L2 acquisition. This ﬁnding offers a different
perspective to early L2 studies that have failed to view L2 syntax acquisition in terms
of dissociating the acquisition of relative clause structure and the acquisition of its
morphosyntactic features.

The dissociation of the relative CP structure and its relevant features further
resolves the controversial issue of the availability of L1 grammar for L2 acquisition.
This supports the FT/FA hypothesis (Schwartz and Sprouse 1994, 1996) that the
entirety of L1 grammar is available for L2 acquisition from the early stages and rejects
the MT hypothesis (Vainikka and Young-Scholten 1994, 1996a) that L2 functional
categories gradually mature. The morphosyntactic features mature independently, and
they may be fossilized at a steady state (Hawkins and Chan 1997; Lardiere 1998a,
1998b; Haznedar 2003). Ultimately, the studies on L2 syntax acquisition should
distinguish the acquisition of phrase structure from the acquisition of its
morphosyntactic features.

The acquisition of the optional feature [+wh] observes the acquisitional order
anticipated by the Noun Phrase Accessibility Hierarchy. This feature was most

apparently employed for the SUB relative CP and then the DO/IO relative CPS. On the

121

 

same ground, the dependence on the default feature [—wh] was more noticeable in the
order of the 10 relative CP, DO relative CP, and SUB relative CP. It ﬁts the reverse
order anticipated by the NPAH. If a sufficient amount of data had been collected, the
order between the DO and 10 relative CPS could have been statistically signiﬁcant in
this experiment.

Judging from the results of the data from teachers, it appears that the optional
feature [+wh] dominates in all types of relative CPS as L2 proﬁciency increases. This
seems to be true for both speaking and writing Situations. Native Speakers, however,
predominantly favor the default feature [—wh] for all relative CPS.

This research provides pedagogic implications. It shows evidence that the way in
which the Korean curriculum is used to teach L2 learners of English is, in some aspects,
different from the way in which native speakers use English. Learners as well as
teachers tend to employ a more complex type of structure as proﬁciency or monitoring
time increases, which was not evident from the data derived from native Speakers.

Nakamori (2002) Shows that the hierarchical teaching method that respects
developmental sequences was a great help to Japanese learners learning English
relative clauses in high school, although new kinds of developmental errors occurred
such as the resumptive relative clause. A Similar ﬁnding is also reported in Kupferberg
and Olshtain (1996).

The relative clause structures in English and Korean may be syntactically Similar,

122

 

but they are quite distinct in terms of word order. Unlike the relative clause in English,
the relative clause in Korean occurs frequently in ordinary speaking and writing (Crain
and Thornton 1998). Korean learners of English need to master the usage of English
relative clause structure in order to precisely produce intended meaning in English (cf.
Bundrick 1989). The Korean curriculum, therefore, should dedicate a substantial
portion of grammar lessons to this goal. Without learning the relative clause structure,
learners can only produce a series of declaratives using conjunctions like and.
Typically, English teachers in Korea do not seem to distinguish between which and
that. All teachers and learners whom I asked about the difference between the two
lexical items responded that the two can be interchangeably used or that that is Simpler
because it works for both humans and non-humans, while which works only for non-
humans. None of those questioned knew that learners tend to produce a that-relative
CP at a low proﬁciency level and gradually shift to produce a which-relative CP as

their proﬁciency increases.

5.3 Chapter Summary

Chapter 5 discussed the results of hypothesis testing, reported hybrid relative
clauses, and pointed out the signiﬁcance and implications of this dissertation. An in-
depth discussion was presented interpreting the statistical outcomes of the hypotheses.

The results lend support to the view that the emergence of relative CP projections

123

precedes the acquisition of speciﬁc features [-wh] and [+wh] associated with them.
This, in turn, seems to support the view that syntactic phrase structure is available as a
block for feature checking between Spec-CP and C°, which further supports the
Dissociation Hypothesis and the FT /FA hypothesis.

It seems apparent that learners ﬁrst acquired the default feature [—wh], and then
the optional feature [+wh] for all types of relative CPS. As learners, however, acquired
the optional feature [+wh], they started to employ that feature with relatively less
complex structures such as the SUB relative CP. Furthermore, the same feature was
more actively selected when learners had sufficient monitoring time to produce the
intended relative CP. Such a phenomenon was conspicuous in more proﬁcient learners
such as teachers. That is, as L2 proﬁciency increased, the use of the optional feature
[+wh] was more recurrent. Also, as the monitoring time increased, the use of the
feature was also more noticeable. The optional feature [+wh], however, decreased as
the complexity of the relative CP increased, and learners depended more on the default
feature [—wh]. It seems that learners’ command of more complex relative CPS had
probably not been developed enough, and they had to rely on the default feature [—wh],
in order to produce these complex relative CP structures.

In addition, the outcomes were compared to data collected from Six Korean
English teachers who taught at the same university that the learners attended, and data

from eight native speakers of English who were traveling or residing in Korea. The

124

purpose of the comparison was to further understand whether the general
developmental pattern of morphosyntactic features of the relative CP is on a par with
that of the English produced by teachers or whether it was on a par with that of the
English produced by native speakers. The result showed that learners were on a
developmental pattern similar to that of the teachers, who heavily depended on
employing the optional feature [+wh]. Native speakers made use of the default feature
[—wh] in most cases.

Two inquiries were raised: (1) Why did the L2 learners start off with [—wh] and
tend to use it as a default over [+wh] and (2) why did the L2 learners employ [+wh]
first with the least complex SUB relative CP? For the ﬁrst inquiry, a cross-linguistic
explanation was discussed. Learners might have transferred the previously existing
[—wh] of the relative CP in L1 Korean and used it for early L2 production of English
relative CPS.

For the second inquiry, a parsing effect was discussed to explain the early selection
of the optional feature [+wh] with the least complex SUB relative CP. Extractability of
which to cross the IP, VP, or PP boundaries from its base position to Spec-CP may play
a great role in the production of successful relative CPS. Extraction across all of the
three phrasal boundaries in an 10 relative CP may reasonably be assumed to be more
difficult than crossing IP and VP boundaries as in the DO relative CP or just IP

boundary as in the SUB relative CP in (13). When extraction fails, learners seem to

125

revert to the default feature [-wh].

Unlike native Speakers preferring the default feature [—wh], learners and teachers
preferred a more restricted OP with the optional feature [+wh], possibly due to the
pragmatic and social requirements that they felt were necessary.

The pied-piped construction was evidenced only in the writing situation. As would
be expected, the teacher group used this construction the most out of the three groups:
learners, teachers, and native speakers.

The result also corresponded to the Noun Phrase Accessibility Hierarchy to a
certain extent. Learners seem to acquire the [+wh] relative CP in the order of SUB >
DO > 10, and the use of the [—wh] relative CP seems to be apparent in the order of
SUB < DO < 10. The statistical results, however, Show only the signiﬁcant difference
between the SUB relative CP and the DO/IO relative CPS. More data may have to be
collected to improve the statistical signiﬁcance.

Next, ungrammatical but acceptable relative clauses that learners constructed in
the process of producing the intended relative CPS were reported: the presentational
amalgam construction (Lambrecht 1988) with a missing default feature [—wh] for the
SUB relative CP, the relative clause with a resumptive pronoun in the base for the DO
relative CP, and the IO relative clauses with a missing preposition (NoPrep). The
NoPrep construction was prevalent because, as is common in Korean, the prepositional

strength is not mature enough to license the variable or 10 relative clauses with a

126

missing preposition (Bardovi-Harlig 1987; Wexler 1991).

In conclusion, it is proposed here in agreement with the Dissociation Hypothesis
(Lardiere 1998a, 1998b; Haznedar 2003) that ﬁrture studies on the acquisition of L2
syntax Should distinguish the acquisition of phrase structure from the acquisition of
relevant morphosyntactic features. One pedagogic ﬁnding of the current study iS that
learners acquire the English produced by their teachers, although the quality of the

teachers’ English was very different ﬁ'om that of the English of native Speakers.

127

CHAPTER 6

6.1 Summary and Conclusions

Although previous studies have dealt with the CPS of questions and declaratives,
none have examined the productions of relative CPS, despite the fact that studying the
relative CP could potentially contribute crucial empirical information about the
development of morphosyntactic features [+wh] and [-wh]. This study, therefore,
investigated this issue from the theoretical perspective of both the Full Transfer/Full
Access (FT/FA) hypothesis and the Dissociation Hypothesis, looking Speciﬁcally at
whether both of the morphosyntactic features mentioned above are available at the
same time or whether they develop in succession, one after the other, and whether the
result would abide by the Noun Phrase Accessibility Hierarchy (NPAH) to a certain
degree.

Two extensive research questions were pr0posed: (1) Do high proﬁciency L2
learners produce a greater proportion of [+wh] SUB, DO, and 10 relative CPS in
speaking and writing Situations than low proﬁciency L2 learners? (2) Do low
proﬁciency L2 learners, in Speaking and writing situations, produce a greater
proportion of [—wh] relative CPS than high proﬁciency L2 learners as the complexity
of relative clause structures increases? By way of a cross-sectional elicited production
method with a picture-description instrument, the data were collected from Korean

university students majoring or minoring in English. A nonparametric statistical

128

technique was selected for data analysis because the data were counted by the number
of occurrence.

Although some of the speciﬁc hypotheses have not been Signiﬁcantly supported,
the FT/FA hypothesis and the Dissociation Hypothesis seemed, in general, to be better
models to account for the early state of L2 CP acquisition and the development of
relevant morphosyntactic features. Learners did not seem to acquire the two features
Simultaneously, but the default feature [—wh] was initially prevalent and the optional
feature [+wh] emerged afterwards, starting from a less complex structure such as the
SUB relative CP. The optional feature [+wh], which was acquired later, was more
often selected when the structure was not complex and also when learners had
sufficient time for monitoring, as in the writing Situation.

Additional data were collected from Korean teachers of English, and the data Show
that these teachers predominantly preferred the optional feature [+wh] across-the-
board. This tendency was even greater in the writing situation.

AS the Dissociation Hypothesis leads us to expect, the development of
morphosyntactic features was independent of syntax. Korean learners and teachers
initially preferred a less restricted OP (that is, [-wh]) at earlier stages, but preferred a
more restricted OP (that is, [+wh]) at a later stage of L2 acquisition. This ﬁnding is
contrary to the Preference Principle (Chomsky 1992), preferring no lexical item in

Spec-CP. The development of these features was from [—wh] at earlier stages to [—wh]

129

and [+wh] mixed at the intermediate stage, and ﬁnally, to [+wh] at the end Stage.

The data collected from native Speakers, however, showed that, in general, they
seemed to dispense with the optional feature [+wh]. They seemed to be satisﬁed with
the default feature [—wh] across-the-board.

A cross-linguistic reason was given for learners starting off with the default feature
[—wh]. This study suggests that learners might have relied on their existing knowledge
of [—wh] in their L1 Korean, which does not have [+wh].

Sociolinguistically, learners might have felt a necessity to produce stylistic
productions for pragmatic and social requirements.

Extractability based on the parsing explanation was also provided to explain why
learners ﬁrst used the optional feature [+wh] with the less complex SUB relative CP.
The contention is that such usage was because the SUB relative CP involves the least
number of phrasal boundaries for a variable to be extracted to Spec-CP from its base
position within 1P.

It was found that the results also corresponded to the NPAH to a certain extent.
Learners employed the optional feature [+wh] in the order of SUB > DO > 10 relative

CPS.50

 

5“ The same order as anticipated by the NPAH is also reported in Gas (1979, 1980), Chiang (1980),
Adjémian and Liceras (1984), Hyltenstam (1984), Hansen (1986), Prideaux and Baker (1986), Eckman
et al. (1988), Eckman et a1. (1989), Doughty (1991), Shin and Lee (1992), Hamilton (1994, 1995), Kim
and Shin (1994), Lee, Y.-K. (1994), Sadighi (1994), Aarts and Schils (1995), Tanaka (1996), Aviar
(1997), [to (1997, 2001), Tezel (1998), Park (2000), O’Grady (2001), Kim (2001), Nakamori (2002),
among others.

130

This study cited some hybrid relative clauses that learners produced in the course
of constructing the intended relative CPS. These hybrid relative clauses contained: (1)
no element in CP for the SUB relative CP, (2) resumptive DO relative CP, and (3) the
NoPrep 10 relative CP.

Lastly, the dissociation between phrase structure and morphosyntactic features in
L2 acquisition was pointed out. Pedagogically, it was emphasized that the quality of
the teachers’ English was very different from that of the English of native speakers.
The next two sections discuss the Shortcomings of the research and suggest the need

for future research.

6.2 Shortcomings

There are theoretical, experimental, and statistical shortcomings which were not so
crucial to the overall result, but might have inﬂuenced it in a particular sense.

Theoretically, this research has focused only on the relative CP, in order to prove
whether morphosyntactic features in the functional category are available as a whole
or whether they develop in succession. The result revealed that morphosyntactic
features gradually mature. The ﬁndings, however, are too premature to be generalized
for other functional categories such as IP or DP (determiner phrase) or the lexical
category VP. In other words, it is not apparent whether the viability of the FT/FA

hypothesis and the Dissociation Hypothesis resulted from the relative CP provides a

131

broad view in respect to other syntactic categories.

A conceptual shortcoming existed in the determination of how early was early in
the acquisition of relative CPS. The view was adopted that the early stage was not
necessarily found at the very onset of language acquisition, but at the beginning stage
of a certain phrase structure like the relative CP or the subjunctive mood, which can
occur only in more proﬁcient learners.

Experimentally, some crucial evidence occurred from the data by teachers and
native speakers, but they were not included as research groups for testing the research
hypotheses. If the data by HPL are combined with the data by teachers, the diﬂ‘erence
of occurrence between LPL and HPL may be more statistically Signiﬁcant for all
relative CPS.

The participants were limited to Korean university adults who had learned English
in the academic setting. If the research, however, had instead limited the participants to
Korean adults who had learned English in a natural setting or had resided for a long
period of time in an English speaking country, the result could have been different
from the one reported in this dissertation and would probably have been similar to that
of native Speakers

Another experimental shortcoming lies in the classiﬁcation of participants, who
were categorized either into LPL or HPL. Such groupings lacked a clearer picture of

the developmental pattern than the classiﬁcation into, for example, LPL, MID, and

132

HPL, where HPL and LPL may unveil a distinctive contrast. Such a study would have
required a much larger amount of data.

Furthermore, the data collection time for each participant was, on average, no
more than 15 minutes, though some participants sometimes felt that the time was too
long. The less proﬁcient participants, especially, often seemed to be exhausted at the
end of the experiment, perhaps due to nervousness. This nervousness may, to some
degree, have introduced variability in the experiment, with respect to the validity of the
data. The time constraint was one of the main reasons that the researcher decided not
to study a more complex type of relative CP, such as the genitive relative CP.S '

The ideal ratio for distractors and target tokens is 3:1 in cognitive sciences
(Schmitt, personal communication) but, after a pilot study, it was adjusted to 2:1 due to
the time constraint and learners’ exhaustion.

Two research administrators collected the data. Therefore, there might have been
some unknown individual differences in the way of presenting the lead-in statements,
although the administrators attempted to treat each participant in exactly the same way.

It would have been nearly impossible to collect all the data if only one research

 

5 ' In essence, the genitive relative CP with whose possessing [+wh] may be developed out of the [-wh]
relative CP. McDaniel et al. (2002: 55) argue with data from their English-Speaking children that “thatis
is either an alternate form of whose or a contraction of the complementizer that with the genitive marker
3.” McDaniel et al. further claim that “thats was not a Speech error” (ibid: 57) and that “American
English is changing in the direction of allowing the that Is constructions That Is is just an alternative
form of whose for adults.” (ibid: 68). Seppanen and Kjellmer (1995: 397) also argue for early
appearances of the [-wh] genitive relative CP with that is, as indicated in their journal title called ‘The
dog thatls leg was run over’. Seppanen and Kjellmer say that “the genitive thatis is a neologism which
seems to be comparatively recent in all varieties of the language but which shows Signs of spreading in
present-day usage.”

133

administrator had had to deal with about 100 participants.

In the statistical analysis, the researcher decided to accept the NoPrep construction
as a successful 10 relative CP because most 10 relative CPS were produced with a
missing preposition. The data, otherwise, would not have contained a sufficient

number of 10 relative CPS.

6.3 Future Research

In order to overcome the shortcomings mentioned in the previous section, a
number of needs for future research are suggested to further strengthen the ﬁndings
ﬁom this dissertation. Studies on the morphosyntactic features of other syntactic
categories like IP, DP, VP, etc. must be carried out to generalize the claim of the
Dissociation Hypothesis. The same experiment conducted with Korean adults who
have immigrated or been residing in an English Speaking country may reveal a
comparative acquisitional view of the development of morphosyntactic features
mentioned in this dissertation.

Another interesting study would be to compare how learners behave in terms of
the production of morphosyntactic features of the relative CP when they converse with
native vs. non-native speakers, with the more educated vs. the less educated, in more
or less formal circumstances (cf. Pavesi 1986), etc. If data are collected from a

sufﬁciently large number of participants at a variety of proﬁciency levels, a

134

comprehensive layout of the developmental pattern might be captured. It may be,
however, possible only with a paper-and-pencil test, computerized test, survey, etc. to
collect a massive amount of data. Interviews or longitudinal studies would be
impractical for such a study.

Japanese learners of English would also be a good target group to focus on in
investigating the order of acquisition between [—wh] and [+wh], because, unlike the
Korean language, Japanese does not have an overt complementizer. If the ﬁndings in
this dissertation are, in fact, universally applicable, it can be assumed that Japanese
learners would also start off with the default feature [—wh] before the optional feature
[+wh] in the production of relative CPS.

Yet another interesting study may be to properly document the types of alternative
constructions that native Speakers employ instead of producing the complex DO, 10,
or genitive relative CPS.52 Those alternative constructions may be the structural norms
that learners should have targeted, instead of relative CPS.

Findings here can perhaps be pedagogically implemented to help learners better

comprehend prescriptive grammar and practical grammar.

 

52 McDaniel et al. (1998) report that their L1 subjects who avoided the relativized genitives often used
grammatical alternatives instead of being reduced to blubbering ungrammatical strings.

135

 

APPENDIX A: Sentence Combination Test

o E’a‘b‘lHlB. EJEAI that, which, oPJ 10881.11, QC" ”1.1
Cl?- FJOlE Eillb’lll Ulkilﬁ. E’JEAI E751 Ag- Ell £3. 52%
’é‘AIBIMIR. @1211 and $8 but-5.3. Algﬁlll lellB.

 

1. (A) The apples are expensive. (B) The apples grow in the mountain.

 

2. (A) The computer is made in Korea. (B) The computer works very well.

 

 

 

 

 

 

 

 

3. (A) Everyone likes the dog. (B) The dog can swim very well.

4. (A) Paul hates the car. (B) The car often breaks down.

5. (A) The cat chases the bird. (B) Torn owns the cat.

6. (A) The songs are very sad. (B) The old man wrote the songs.

7. (A) Paul kissed the ring. (B) His wife bought the ring.

8. (A) The man kicked the car. (B) His neighbor parked the car.

9. (A) The ofﬁce opens at 11:00. (B) Jack wrote the office an urgent e-mail.
10. (A) The cat hates dogs. (B) His sister gave the cat a ﬁsh.

 

11. (A) The lady cleaned the room. (B) The boy sent the room a bottle of wine.

 

12. (A) Mr. Lee owns the company. (B) SONY offers the company a new deal.

 

136

APPENDIX B: Sample Pictures

A Sample Picture for SUB relative CP in the Speaking Situation (SUB l)

 

137

A Sample Picture for DO relative CP in the Speaking Situation (DO 1)

 

138

A Sample Picture for 10 relative CP in the Speaking Situation (IO 1)

 

139

A Sample Picture for SUB relative CP in the Writing Situation (SUB 1)

 

140

A Sample Picture for DO relative CP in the Writing Situation (DO 1)

 

141

A Sample Picture for 10 relative CP in the Writing Situation (IO 1)

I

\V.
‘1}?!

eat we.
\. 8‘
ﬁlm?

 

142

APPENDIX C: Lead-In Statements for Spoken Data

Lead-In Statement

. Listen carefully to the words and sentences as I describe the pictures.
. When you answer, use my words. Don’t use different words.
. I have never seen the pictures, so when you say like “right one, left one, red one, top

picture, or bottom picture,” I will not understand.

. You will see a set of 14 pictures.
. When I ask you a question, please be sure you understand the question exactly!
. These are very easy. So don’t worry!

 

QWN—r

Page 1

. There are 2 pictures - each with a clock, right?

. In one picture, a clock is on the sofa. And the sofa has a yellow star.

. In another picture, a clock is on the table. And the table has a green star.
. Now tell me. Where is the yellow (or green) star in the picture?

 

_

Page 2

. There are two persons in the picture.
. What would you call the person with the blue star?

Is she a lady, a young lady, a young woman, or a woman?

. How about the person with the yellow star?

Lady, young lady, young woman, or woman?

 

Page 3 (SUB 1)

. Two pictures of kangaroos.
. In one picture, a kangaroo is drinking water with his ﬁ'iends. A kangaroo is drinking

water with his friends. And it has a green star.

. In another picture, a kangaroo iS looking at the camera man. A kangaroo is looking

at the camera man. And the kangaroo has a blue star, right?

. Then, look at the blue and green stars. Which kangaroo has the blue star?

 

Ut-hbéNt—

Page 4

. In one picture, a girl has a butterﬂy in her hands. And the butterﬂy has a yellow star.
. In another picture, a teacher is talking with a student, right?

. Then, tell me in a full sentence. What is the girl doing in the ﬁrst picture?

. Again, the question is: What is the girl doing in the first picture?

. One more. What is the teacher doing in the second picture?

I43

g

Page 5 (DO 1)

. I believe there are two pictures - each with a lady, right?
. Now look at the gloves. They both have gloves, right?
. In one picture, a lady is carrying gloves and a handbag. A lady is holding gloves and

a handbag, and there’s a blue star on the gloves.

. In another picture, a lady is wearing gloves. A lady is wearing gloves. And her

gloves have a red star, right?

. Look at the blue and red stars. Tell me. Which gloves have the red star?

 

AWN.—

Page 6

. Two men, right? A Korean man and a western man.

. Now listen. They both speak very good English.

. If you choose, who do you want to be your English teacher?
. The Korean man or the western man?

 

Page 7

. OK. In one picture, a Korean man is smiling and reading a book. And the book has a

yellow star.

. In another picture, a black man is looking for a book in the library. And the book

has a blue star.

. Now listen. What is the Korean man doing in the picture?
. One more. See the black man? Which star does his book have?

 

Page 8 (IO 1)

. Two pictures of teddy bears.
. In the ﬁrst picture, a teddy bear is Sitting on the picnic table. A teddy bear is sitting

on the picnic table. And it has a purple star.

. In the next picture, a girl is giving a ball to the teddy bear. A girl is giving a ball to

the teddy bear. And it has a blue star, correct?

. Look at the blue and purple stars. Which teddy bear has the blue star?

 

ﬂ

Page 9

. A picture of a lady.
. OK. Tell me. Where is the red star in the picture?
. One more. Where is the yellow star?

 

144

Page 10
This time, two ladies.

. The same question as before. They both speak very good English.
. If you choose, who do you want to be your English teacher?
. The Korean person or the western person?

 

Page 11 (SUB 2)

. Now monkeys!
. In one picture, a mother monkey is carrying a baby monkey on her back. A mother

monkey is carrying a baby monkey on her back. And the baby monkey has a
yellow star.

. In another picture, a baby monkey is holding on to a tree. A baby monkey iS holding

on to a tree. And it has a white star, right?

. Look at the yellow and white stars. Which monkey has the white star?

 

N...

Page 12 (DO 2)

. Now potatoes!
. In one picture, a lady is happily thinking of a potato. A lady is happily thinking of a

potato. And that potato has a green star, right?

. In another picture, a lady is holding a potato. A lady is holding a potato. And it has a

red star, right?

. Look at the green and red stars. Which potato has the red star?

 

Page 13

. There is a telephone in one picture. And there is a swing set in another picture,

right?

. Say in full sentence. Where is the yellow star exactly? Where is the red star?

 

Page 14 (IO 2)

. These are very interesting pictures.
. In one picture, a man is writing an e-mail to his company. A man is writing an e-

mail to his company. And a blue star is on the company, right?

. In another picture, a man is writing a letter to the company. A man is writing a letter

to the company. And the company has a red Star, right?

. Did you understand!!
. Look at the red and blue stars. Which company has the red star?

145

APPENDIX D: Lead-In Statements for Written Data

Lead-In Statement

Page 1

1. There are 2 pictures, right?

. In one picture, there are a teddy bear and a chair. And the teddy bear is sitting on the
chair. And the chair has a yellow star.

. In another picture, a dog is Sitting on a picnic table. And the dog has a green star.

4. Question 1: Where is the green star in the picture?

 

Page 2

1. There are two persons in the picture.

. Question 2: The person with the red star? Do you think she is a lady, a young lady, a
young woman, or a woman? Please write in a full sentence.

 

Page 3 (SUB 1)

1. Two pictures with bears, right?

. In one picture, three bears are looking at the camera. Three bears are looking at the
camera. And the bear in the middle has a green star.

. In another picture, a bear iS catching a ﬁsh. A bear is catching a ﬁsh. And that bear
has a red star, right?

. Look at the green and red stars. Question 3: Which bear has the red star?

 

Page 4

1. In one picture, a girl has long green hair and there is a red star.

2. In another picture, a woman has long gray hair and there is a blue star, right?
. Question 4: Which star is on the green hair?

Please write your answer in a full sentence!

 

146

 

 

Page 5 (D01)

1. Two pictureS--each with a lady, right? Look at the cups, please.

. In one picture, a lady is kissing a cup. A lady is kissing a cup. And the cup has a

yellow star.

. In another picture, a lady is holding a cup. A lady is holding a cup. And it has a

white star, right?

. Question 5: Which cup has the yellow star?

 

Page 6

. Two men, right? A western man and a Korean man.

2. Question 6: Who do you want to be your English teacher?

. The Korean man or the western man? Write your answer in a sentence.

 

#MNV—

Page 7

. There are 7 lipsticks in the picture, right? Listen, remember, and write.
. The blue star is on the lipstick on the very left.

. The red star is on the third lipstick from the right. Understand?

. Question 7: Where is the red star?

 

Page 8 (IO 1)

. In one picture, some children and a teddy bear are on the picnic table. Some children

and a teddy bear are on the picnic table. The teddy bear has a green star, right?

. In another picture, a girl is offering a candy to the teddy bear. A girl is oﬂering a

candy to the teddy bear. And it has a yellow star. Is that correct?

. Question 8: Which teddy bear has the yellow star? Write it in sentence.

 

_

Page 9

. A picture of a lady.
. OK. Write in full sentence. Question 9: Where is the blue star?

 

147

Page 10

1. Two ladies, right? A Korean and a Canadian.

2. Question 10: Who do you want to be your English teacher?
. The Korean or the Canadian? Write it in sentence.

 

Page 11 (SUB 2)

1. Glasses!

. In one picture, an eye doctor is adjusting her patient’s glasses. An eye doctor is
adjusting her customer’s glasses. So She is helping her. And the glasses have a
red star, right?

. In another picture, a woman is wearing glasses. A woman is wearing glasses. And
the glasses have a green star, right?

. Look at the red and green stars. Question 11: Which glasses have the green star?

 

Page 12 (DO 2)

1. Pictures of two women.

. In one picture, a woman is taking a bath. And she is tilting a wine glass and pouring
red wine into the water. The wine glass has a red star.

. In another picture, a woman is lifting a wine glass. A woman is lifting a wine glass.
And the glass has a blue star, right?

. Look at the red and blue stars. Question 12: Which glass has the blue star?

 

Page 13
l. A young man.
. Question 13: Where is the yellow star in the picture?

 

Page 14 (IO 2)

1. Two men!

. In one picture, a man is working in his office. A man is working in his office. And
the office has a red star.
. In another picture, a man is writing a report to his office. A man is writing a report to

I48

his office. And the office has a blue star, right? The small picture is the ofﬁce.

4. Look at the red and blue stars. Question 14: Which office has the blue star?

149

APPENDIX E: Consent Forms

Dissertgtion Title: DEVELOPMENT OF MORPHOSYNTACTIC FEATURES OF
THE COMPLEMENTIZER PHRASE IN L2 RELATIVE CLAUSES OF KOREAN

LEARNERS
Responsible Project Investigator Secondary Investigator
Dr. Susan Gass (gass@msu.edp) Il-jae Lee (ijlee@m§u.edu)
University Distinguished Professor Doctoral Candidate
English Language Center Department of Linguistics
A-714 Wells Hall Michigan State University
Michigan State University East Lansing, MI 48824

East Lansing, MI 48824
Phone: (517) 353-0800
Fax: (517) 432-1149

By signing this consent form, you are agreeing to participate in a study. The study
is for the above secondary investigator’s doctoral dissertation from the Department of
Linguistics and Languages at Michigan State University in the US. A purpose of the
study will be to collect and analyze English spoken and written production data by
Korean college learners who are currently taking a major course in the English
department at Kangwon National University in Korea.

You will be asked to describe and write about some pictures. Your voice has to be
recorded into a cassette tape for later transcription. Signing this consent form also
allows the investigator to record your voice. The experiment will take no more than 20
minutes.

Notice that participation is voluntary; you may choose NOT to participate at all, or
you may refuse to answer certain questions or may discontinue the experiment at any
time without penalty or loss of beneﬁts to which you are otherwise entitled. We assure
that your privacy and conﬁdentiality will be protected to the maximum extent
allowable by law.

Your signature above indicates your voluntary agreement to participate in this
study and to record your voice.

Signature

 

If you have any questions about this study, please contact the investigator: If you have questions or
concerns regarding your rights as a study participant, or are dissatisﬁed at any time with any aspect of
this study, you may contact - anonymously, if you wish - Peter Vasilenko, MD, Chair of the University
Committee on Research Involving Human Subjects (UCRIHS) by phone: 001-1-517-355-2180, fax: 00]-
1-517—432-4503, e-mail: pprihs@msp.§dp, or regular mail: 202 Olds Hall, East Lansing, MI 48824.

150

Dissertation Title: DEVELOPMENT OF MORPHOSYNTACTIC FEATURES OF
THE COMPLEMENTIZER PHRASE IN L2 RELATIVE CLAUSES OF KOREAN

 

 

LEARNERS
Responsible Project Investigator Secondary Investigator
Dr. Grover Hudson (hudspn@msu.edu) Il-jae Lee (ijlee@msp.edp)
Dept. of Linguistics and Languages Bongeu-dong #43
Michigan State University Chuncheon, Gangwon-do
East Lansing, MI 48824 Korea 200-020
Phone: (517) 355-8471 Phone: 011-82-33-256-9443

Fax: (517) 432-2736

By signing this consent form, you are agreeing to participate in a study. The
study is for the above secondary investigator’s doctoral dissertation ﬁ'om the
Department of Linguistics and Languages at Michigan State University in the US. It
will be to collect and analyze English Spoken and written production data by Korean
college learners who are currently taking a major course in the English department at
Kangwon National University in Korea, English teachers, and native English Speakers.

You will be asked to describe and write about some pictures. Your voice has to be
recorded into a cassette tape for later transcription. Signing this consent form also
allows the investigator to record your voice. The experiment will take no more than 20
minutes.

Notice that the participation is voluntary; you may choose NOT to participate at
all, or you may refuse to answer certain questions or may discontinue the experiment
at any time without penalty or loss of beneﬁts to which you are otherwise entitled. We
assure that your privacy and conﬁdentiality will be protected to the maximum extent
allowable by law.

 

If you have any questions about this study, please contact the investigator
(Bongeu-dong #43, Chuncheon, Gangwon-do, 200-020 Korea. Tel: (033) 256-9443 or
Email: iilee@msu.edu2. If you have questions or concerns regarding your rights as a
study participant, or are dissatisﬁed at any time with any aspect of this study, you may
contact - anonymously, if you wish - Peter Vasilenko, MD, Chair of the University
Committee on Research Involving Human Subjects (UCRIHS) by phone: 001-1-517-
355-2180, fax: 001-1-517-432-4503, e-mail: ucrihs@,msu.edy, or regular mail: 202
Olds Hall, East Lansing, MI 48824.

Your signature above indicates your voluntary agreement to participate in this
study and to record your voice.

Signature: Date:

 

151

REFERENCES

Aarts, F. and Schils, E. (1995). Relative clauses, the accessibility hierarchy and the
contrastive analysis hypothesis. IRAL 33, 47-63.

Adamson, H. D. (1992). Social and processing constraints in relative clauses.
American Speech 67, 123-133.

Adjémian, C. and Liceras, J. (1984). Accounting for adult acquisition of relative
clauses: Universal Grammar, L1, and structuring the intake. Universals pf
Second Langpage Acquisition, ed. by F. R. Eckman, L. H. Bell, and D. Nelson,
101-118. Rowley, MA: Newbury House Publishers.

Alexiadou, A, Law, P, Meinunger, A. and Wilder, C. (2000). Introduction. The Syptax
of Relative Clauses, ed. by A. Alexiadou, P. Law, A. Meinunger, and C. Wilder,
1-51. Philadelphia, PA: John Benjamins Publishing Company.

 

Ariel, M. (1999). Cognitive universals and linguistic conventions: the case of
resumptive pronouns. Studies in Langpage 23, 217-269.

Aivar, E. S. (1997). Levels of complexity in relative clauses. Studia Anglica
Posnaniensia XXXII, 115-126.

Baker, M. C. (1985). Incorporation: A Theory of Grammatical Changing. Doctoral
dissertation. MIT.

Bai, B. L. (1989). Relativization in child language and history. Langpage Variation and
Language Change, ed. by A. Mukherjee, 141-152. Hyderabad, India: Osmania
University.

Ball, C. N. (1996). A diachronic study of relative markers in Spoken and written
English. Language Variption and Change 8, 227-258.

Bardovi-Harlig, K. (1987). Markedness and salience in second-language acquisition.
Langpage Learning 37, 385-407.

Bayley, R. (1999). Relativization strategies in Mexican-American English. American
Speech 74, 115-139.

Bernstein, J. B, McDaniel, D. and McKee, C. (1998). Resumptive pronoun strategies
in English speaking children. Proceedings of the 22nd Annual Boston

University Conference on Langpage Development, ed. by Annabel Greenhill,
Mary Hughes, Heather Littleﬁeld, and Hugh Walsh, 58-68. Somerville, MA:

Cascadilla Press.

152

Bergh, G and Seppanen, A. (2000). Preposition stranding with wh-relatives: a
historical survey. English Language and Lingpistics 4, 295-316.

Bhatt, R. M. and Hancin-Bhatt, B. (2002). Structural minimality, CP and the initial
state in second language acquisition. Second Language Research 18, 348-392.

Bissell, T. (1999). Further Evidence for Null Pronominal Variables. Master’s thesis,

University of California, Santa Cruz. (www.mit.edu/~bissell/Nulngquf)

Borer, H. and Wexler, K. (1987). The maturation of syntax. Parameter Setting, ed. by T.
Roeper and E. William, 123-172. Reidel, Dordrecht: Kluwer Academic
Publishers.

Borsley, R. D. (1992). More on the difference between English restrictive and non-
restrictive relative clauses. Journal of Linguistics 28, 139-148.

Borsley, R. D. (1997). Relative clauses and the theory of phrase structure. Lingpistic
lnqug’ 28, 629-647.

Bundrick, C. (1989). An interference-based account of restrictive relative which and
that. Studies in the Lingpistic Sciences 19, 19-31.

Cheshire, J. (1996). That jacksprat: an interactional perspective on English that.
Journal of Pragmatics 25, 369-393.

Chiang, D. L. (1980). Predictors of relative clause production. Research in Second

Langpage Acquisition: Selected Papers of the Los Angeles Second Langpage
Acquisition Research Forum, ed. by R. C. Scarcella and S. D. Krashen, 142-
145. Rowley, MA: Newbury House Publishers.

Chomsky, N. (1977). On wh-movement. Formal Smtax, ed. by P. W. Culicover, T.
Wasow, and A. Akmajian, 71-132. New York: Academic Press.

Chomsky, N. (1982). Some Concepts and Consequences of the Theory of Government
and Binding. Cambridge, MA: The MIT Press.

Chomsky, N. (1989). Some notes on economy of derivation and representation. MIT
Working Papers in Lingpistics 10, 43-74.

Chomsky, N. (1992). A minimalist program for linguistic theory. MIT Occasior,1_al_
Papers in Lingpistics l.

Chomsky, N. (1994). Bare phrase structure. MIT Occasional Papers in Lingpistics 5.

Chomsky, N. (1995). The Minimalist Proggam. Cambridge, MA: The MIT Press.

153

Chomsky, N. (1998). Minimalist inquiries: The ﬁ'amework. MIT Occasional Papers in
Lingpistics 15.

Chomsky, N. (1999). Derivation by phrase. MIT Occasional Papers in Linguistics 18.

Chomsky, N. (2001). Beyond explanatory adequacy. MIT Occasional Papers in
Lingpistics 20.

Chomsky, N. and Lasnik, H. (1993). The theory of principles and parameters. Syntax:
An International Handbook of Contemporam Research, ed, by J. Jacobs, A.
von Stechow, W. Stemeﬂed, and T. Vennemann, 506-569. Berlin: Walter de
Gruyter.

Cinque, G (1982). On the theory of relative clauses and markedness. The Lingpistic
Review 1, 247-294.

 

Cinque, G (1990). Types of A-Dependencies. Cambridge, MA: The MIT Press.

Clahsen, H., Penke, M. and Parodi, T. (1993/1994). Functional categories in early child
German. Langpage Acquisition 3, 395-429.

Comrie, B. (1989). Language Universals and Linguistic Typology: Sﬂtax and
Mogphology. (2nd edition). Chicago: The University of Chicago Press.

Cowper, E. A. (1987). Pied piping, feature percolation and the structure of the noun
phrase. Canadian Journal of Linguisgtic_s 32, 321-338.

Crain, S, McKee, C. and Emiliani, M. (1990). Visiting relatives in Italy. Langpage
Processing and Langpage Acquisition, ed. by L. Frazier and J. de Villiers, 335-
356. Dordrecht: Kluwer Academic Publishers.

Crain, S. and Thornton, R. (1998). Investigations in Universal Grammar: A Guide to

Experiments on the Acquisition of Smtax and Semantics. Cambridge, MA:
The MIT Press.

Dekydtspotter, L, Sprouse, R. A. and Anderson, B. (1998). Interlanguage A-bar
dependencies: binding construals null prepositions and Universal Grammar.

Second Language Research 14, 341-358.

Dickey, M. W. (1996). Constraints on the sentence processor and the distribution of
resumptive pronouns. University of Massachusetts Occasional Papers in
Lingpistics 19, 157-192.

Diessel, H. and Tomasello, M. (1999). Why complement clauses do not include a that-
complementizer in early child language. Berkeley Lingpistics Society 25, 86-
97.

154

Diessel, H. and Tomasello, M. (2000). The development of relative clauses in
spontaneous child Speech. Cogpitive Lingpistica 11, 131-151.

Diessel, H. and Tomasello, M. (2001). The acquisition of ﬁnite complement clauses in
English: a corpus-based analysis. Cogpitive Linguistics 12, 97-141.

Doughty, C. (1991). Second language instruction does make a difference: evidence

from an empirical study of SL relativization. Studies in Second Langpage
Acquisition 13, 431-469.

Eckman, F. R., Bell. L. and Nelson, D. (1988). On the generalization of relative clause
instruction in the acquisition of English as a second language. Applied
Lingpistics 9, 1-20.

Eckman, F. R., Moravcsik, E. A. and Wirth, J. R. (1989). Irnplicational universals and
interrogative structures in the interlanguage of ESL learners. Language
Learning 39, 173-205.

Epstein, S. D. (1989). Quantiﬁcation in null operator constructions. Lingpistic Inqug'
20, 647-658.

Epstein, S., Flynn, S. and Martohardjono, M. (1996). Second language acquisition:
theoretical and experimental issues in contemporary research. Behavioral and

Brain Sciences 19, 677-758.

F abb, N. (1990). The difference between restrictive and nonrestrictive relative clauses.
Journal of Lingpistics 26, 57-78.

F lanigan, B. O. (1995). Anaphora and relativization in child second language
acquisition. Studies in Second Language Acquisition 17, 331-351.

Flynn, S. (1989a). Spanish, Japanese and Chinese speakers’ acquisition of English
relative clauses: new evidence for the head-direction parameter. Bilingpalism
Across the Life Span: Aspects of Acquisition. Maturity, and Loss, ed. by K.
Hyltenstam and L. K. Obler, 116-131. Cambridge: Cambridge University Press.

Flynn, S. (1989b). The role of the head-initial/head-ﬁnal parameter in the acquisition
of English relative clauses by adult Spanish and Japanese speakers. Linguistig
Perspectives on Second Lapgu_age Acquisition. ed. by S. M. Gass and J.
Schachter, 89-108. Cambridge: Cambridge University Press.

Flynn, S. and Lust, B. (1980). Acquisition of relative clauses: developmental changes
in their heads. Cornell Working Papers in Lingpistics 1, 33-45.

Fox, B. A. (1987). The noun phrase accessibility hierarchy reinterpreted: subject
primacy or the absolute hypothesis. Langpage 63, 856-870.

155

Fox, B. A. and Thompson, S. A. (1990). A discourse explanation of the grammar of
relative clauses in English conversations. Langpage 66, 297-316.

Fox, D. (1999). Reconstruction, binding theory, and the interpretation of chains.
Lingpistic Inqup'y' 30, 157-196.

Fox, D. (2002). Antecedent—contained deletion and the cepy theory of movement.
Lingpistic Inqug' 33, 63-96.

Fraenkel, J. R. and Wallen, N. E. (2003). How to Design and Evaluate Research in
Education. (5th edition). Boston: McGraw-Hill Higher Education.

F ragman, C. 1997. On assessing the distinctive properties of child grammar: the case
of relative clause production in French. Proceedings of the let Annual Boston
University Conference on Langpage Development, ed. by Elizabeth Hughes,
Mary Hughes, and Annabel Greenhill, 170-181. Somerville, MA: Cascadilla
Press.

French, M. (1984). Markedness and the acquisition of pied-piping and preposition
stranding. McGill Working Papers in Lingpistics 2, 131-144.

Gass, S. M. (1979). Language transfer and universal grammatical relations. Langpage
Learning 29, 327-344.

Gass, S. M. (1980). An investigation of syntactic transfer in adult second language
learners. Research in Second Langpage Acquisition: Selected Papers of the Los
Angeles Second Langpage Acquisition Research Forum, ed. by R. C. Scarcella
and S. D. Krashen, 132-141. Rowley, MA: Newbury House Publishers.

Grondin, N. and White, L. (1996). Functional categories in child L2 acquisition of
French. Langpage Acquisition 5, 1-34

Guasti, M. T. (2002). Language Acquisition: The Growth of Grammar. Cambridge,
MA: The MIT Press.

 

Guasti, M. T. and Shlonsky, U. (1995). The acquisition of French relative clauses
reconsidered. Langpage Acquisition 4, 257-276.

Guy, (1 R. and Bayley, R. (1995). On the choice of relative pronouns in English.
American Speech 70, 148-162.

Hamburger, H. (1980). A deletion ahead of its time. Cogpition 8, 389-416.

Hamburger, H. and Crain, S. (1982). Relative acquisition. Smtax and Semantics:
Language Development, Volume 1, ed. by S. A. Kuczaj 11, 245-74. Hillsdale,
NJ: Lawrence Erlbaum Associates.

156

Hamilton, R. L. (1994). Is implicational generalization unidirectional and maximal?
evidence ﬁ'om relativization instruction in a second language. Language
Learning 44, 123-157.

Hamilton, R. L. (1995). The noun phrase accessibility hierarchy in SLA: determining
the basis for its developmental effects. Second Language Acquisitiqn Theog
and Pedagogy, ed. by F. R. Eckman, D. Highland, P. W. Lee, J. Mileharn, and R.
R. Weber, 101-114. Mahwah, NJ: Lawrence Erlbaum Associates.

Han, C.-H. and Kim, J.-B. (2004). Are there “double relative clauses” in Korean?
Linguistic Inquiry 35, 315-37.

Han, J .-I. (1992). Syntactic movement analysis of Korean relativization. Language
Research 28, 335-357.

Hansen, L. (1986). Universals in relative clause acquisition: evidence from child and
adult L1 and L2 learners of Hindi-Urdu. Language Learning 36, 143-157.

Hawkins, R. (1989). Do second language learners acquire restrictive relative clauses
on the basis of relational or conﬁgurational information? the acquisition of
French subject, direct object and genitive restrictive relative clauses by second
language learners. Second Language Research 5, 156-188.

Hawkins, R. and Chan, C. Y.-H. (1997). The partial availability of Universal Grammar
in second language acquisition: the ‘failed functional features hypothesis.’

Second Language Research 13, 187-226.

Haznedar, B. (2001). The acquisition of the IP system in child English. Studies in
Second Language Acquisition 23, 1-39.

Haznedar, B. (2003). The status of ﬁrnctional categories in child second language
acquisition: evidence from the acquisition of CP. Second Language Research
19, 1-41.

Hildebrand, J. (1987). The acquisition of prepositional stranding. Canadian Journal of
Linguistics 32, 65-85.

Hyltenstam, K. (1984). The use of typological markedness conditions as predictors in
second language acquisition: the case of pronominal copies in relative clauses.
Second Languages: A Cross-Linguistic Perspective. ed. by R. W. Anderson, 39-
58. Rowley, MA: Newbury House Publishers.

Ito, A. (1997). Japanese EFL learners’ test-type related interlanguage variability. JALT
Journal 19, 89-105.

 

Ito, A. (2001). Japanese EFL learners’ processing in English relativization. ITL:
Review of Applied Linguistics 133-134, 325-345

157

Jang, Y.-J. and Han, H. (1999). Relative clauses in child language: semantic parsing
and the continuity hypothesis. Journal of Pan-Paciﬁc Association of Applied
Linguistics 3, 53-70.

Jarvis, S. (2000). Methodological rigor in the study of transfer: identifying Ll
inﬂuence in the interlanguage lexicon. Language Learning 50, 245-309.

Kang, M.-Y. (1988). Syntactic movement in Korean relativization. Linguistics in the

Morning Calm 2, ed. by the Linguistic Society of Korea, 347-362. Seoul:
Hanshin Publishing Co.

Kaplan, T. I. and Whitman, J. B. (1995). The category of relative clauses in Japanese,

with reference to Korean. Journal of East Asian Linguistics 4, 29-58.

Kaplan. T. I. (1993). The Second Language Acquisition pf Functional Categories:
Complementizer Phrases in English and Japanese. Doctoral dissertation.

Cornell University.
Keenan, E. L. (1985). Relative clauses. Language Typology and Syutaqtic Description,
Volume II: Complex Constructions, ed. by Timothy Shopen, 141-170.

Cambridge: Cambridge University Press.

Keenan, E. L. and Comrie, B. (1977). Noun phrase accessibility and Universal

Grammar. Linguistic Inqug’ 8, 63-99.

Kim, A. H.-O. and Shin, H.-S. (1994). Information ﬂow and relative constructions in

Korean discourse. Theoretical Issues in Korean Linguistics, ed. by Young-Key

Kim-Renaud, 463-493. Stanford, CA: CSLI.

Kim, S.-K. (2001). Structured Input and Production Practice in Foreign/Second

Language Acquisition. Doctoral dissertation. Temple University.

Klein, E. C. (1993). Toward Second Language Acquisition: A Study of Null-Prep.
Dordrecht: Kluwer Academic Publishers.

Klein, E. C. (1995). Evidence for a ‘wild’ L2 grammar: when PPS rear their empty

heads. Applied Linguistics 16, 87-117.
Klein, E. C. (2001). (Mis)construing null prepositions in L2 intergrammars: a

commentary and proposal. Second Language Research 17, 37-70.

Klein, E. C. and Casco, M. (1999). Optionality in English non-native grammars.
Proceedings of the 23rd Annual Boston University Conference on Language
Development, ed. by A. G, M. Hughes, H. Littleﬁeld, and H. Walsh, 349-360.

Somerville, MA: Cascadilla Press.

158

Ko, Y.-H. (2002). Towards processing easiness: examining oral production of L2
Korean RC. Handout from Second Language Forum in Toronto.

Kornﬁlt, J. (2000). Some syntactic and morphological properties of relative clauses in
Turkish. The Smtax of Relative Clauses. ed. by A. Alexiadou, P. Law, A.
Meinunger, and C. Wilder, 121-159. Philadelphia, PA: John Benjamins

Publishing Company.

Kuno, S. (1974). The position of relative clauses and conjunctions. Linguistic Inquiry
5, 117-136.

Kupferberg, I. and Olshtain, E. (1996). Explicit contrastive instruction facilitates the
acquisition of difficult L2 forms. Language Awagness 5, 149-165.

Labelle, M. (1990). Predication, wh-movement, and the development of relative
clauses. Language Acquisition 1, 95-119.

Labelle, M. (1996). The acquisition of relative clauses: movement or no movement?
Language Acquisition 5, 65-82.

Lakshmanan, U. (1993/1994). The boy for the cookie: some evidence for the
nonviolation of the case ﬁlter in child second language acquisition. Language
Acquisition 3, 51-97.

Lakshmanan, U. (1998). Functional categories and related mechanisms in child second

language acquisition. The Generative Study of Second Language Acquisition,
ed. by S. Flynn, G Martohardjono, and W. O’Neil, 3-16. Mahwah, NJ:

Lawrence Erlbaum Associates.

Lakshmanan, U. (2000). The acquisition of relative clauses by Tamil children. Journal
of Child Language 27, 587-617.

 

Lakshmanan, U. and Selinker, L. (1994). The status of CP and the tensed
complementizer that in the developing L2 grammar of English. Segond

Language Research 10, 25-48.

Lambrecht, K. (1988). There was a farmer had a dog: syntactic amalgams revisited.
Berkeley Linguistics Society 14, 319-339.

Lardiere, D. (l998a). Case and tense in the ‘fossilized’ steady state. Second Language
Research 14, 1-26.

Lardiere, D. (l998b). Dissociating syntax ﬁ'om morphology in a divergent L2 end-state
grammar. Second Language Research 14, 359-375.

Lardiere, D. (2003). Revisiting the comparative fallacy: a reply to Lakshmanan and
Selinker, 2001. Second Language Research 19, 129-143.

159

Lasnik, H. (1995). Case and expletives revisited: on greed and other human failings.
Linguistic Inquuy' 26, 615-633.

Law, P. ( 1998). A uniﬁed analysis of P-stranding in Romance and Germanic.
Proceedings of NELS 28, 219-234.

Lebeaux, D. (1990). The grammatical nature of the acquisition sequence: adjoin-a and
the formation of relative clauses. Language Processing and Language
Acquisition, ed. by L. Frazier and J. de Villiers, 13-82. Dordrecht: Kluwer
Academic Publishers.

Lebeaux, D. (1991). Relative clauses, licensing, and the nature of the derivation.

Syutax and Semantics, Volume 25: Perspectives on Phrase Structure: Heads
and Licensig. ed. by S. Rothstein, 209-239. San Diego: Academic Press.

Lee, F. (2001). Relative clauses without wh-movement. Proceedings of NELS 31, 321-
331.

Lee, l.-J. (2004). L2 indirect objective relative clause productions and minimalism.
Korean Journal of Applied Linguistics 20, 55-84.

Lee, S.-B. (1994). A semantic interpretation of generic NPS in relative clauses.
Language Research 30, 25-40.

Lee, S.-W. (1984). On the resumptive pronouns in Korean relative clauses. Language
Research 20, 51-59.

Lee, Y.-K. (1994). The Effect of Readiug on the Acquisition of English Restrictive
Relative Clauses. Doctoral dissertation. University of Southern California.

Matsumoto, Y. (1996). Interaction of factors in construal: Japanese relative clauses.
Grammatical Constructions: Their Form and Meaning, ed. by Masayoshi
Shibatini and Sandra A. Thompson, 103-124. Oxford, England: Clarendon

Press.

McCawley, J. D. (1972). Japanese relative clauses. The Chicago Which Hunt: Papers
from the Relative Clause Festival. ed. by P. M. Peranteau, J. N. Levi, and G C.
PhareS, 205-214. Chicago: Chicago Linguistic Society.

McCloskey, J. (1990). Resumptive pronouns, A’-binding, and levels of representation

in Irish. Smtax and Semantics, Volume 23: The Smtax of the Modern Celtic
Languages, ed. by R. Hendrick, 199-248. San Diego: Academic Press.

McDaniel, D. and McKee, C. (1996). Children’s oblique relatives. Proceedings of the

20th Annual Boston University Conference on Language Development, ed. by
A. Stringfellow, D. Cahana-Amitay, E. Hughes, and A. Zukowski, 472-482.

Somerville, MA: Cascadilla Press.

160

McDaniel, D. and Cowart, W. (1999). Experimental evidence for a minimalist account
of English resumptive pronouns. Coguition 70, BIS-324.

McDaniel, D, McKee, C. and Bernstein, J. B. (1998). How children’s relatives solve a
problem for minimalism. Language 74, 308-334.

McDaniel, D, McKee, C. and Cairns, H. S. (1996). Methods for Assessing Children’s
Smtax. Cambridge, MA: The MIT Press.

McDaniel, D, Stickney, H, E, Fowler, S. and McKee, C. (2002). What’s thats? English
Studies 83, 53-69.

McKee, C. and McDaniel, D. (2001). Resumptive pronouns in English relative clauses.
Language Acquisition 9, 113-156.

McKee, C, McDaniel, D. and Snedeker, J. (1998). Relatives children say. Journal of
Psycholinguistic Research 27, 573-596.

Menyuk, P. (1969). Sentences Children Use. Cambridge, MA: MIT Press.

Moon, G-S. (1995). A First Course in Formal Semantics. Seoul: Hanshin Publishing
Co.

Morpeth, N. van den E. (1998). WH- vs. T H-relativization as a stylistic diagnostic:
reporting on a real-time study of language change. Leuvensche Bijdragen 87,
47-57.

Munn, A. (1992). A null operator analysis of ATB gaps. The Linguistic Review 9, 1-26.

Munn, A. (1994). A minimalist account of reconstruction asymmetries. Proceedings of
NELS 24, 397-410.

Nakamori, T. (2002). Teaching relative clauses: how to handle a bitter lemon for
Japanese learners and English teachers. ELT Journal 56, 29-40.

NeWport, E. L. (1990). Maturation constraints on language learning. Cognitive Science
14, 11-28.

Nirmala, C. (1980). Relativization in children’s speech. Osmania University Papers in
Linguistics 6, 53-70.

O’Grady, W. (2001). A linguistic approach to the study of language acquisition.
Jourer Pan-Paciﬁc Association of Applied Linguistics 5, 57-71.

O’Grady, W. (2003). The radical middle: nativism without Universal Grammar. The

Handbook of Second Language Acquisition, ed. by C. Doughty and M. Long,
43-62. Malden, MA: Blackwell Publishing.

161

O’Grady, W, Yamashita, Y, Lee, M.-S., Choo, M.-H. and Cho, S.-E. (2000).
Computational factors in the acquisition of relative clauses. Manuscripts.
University of Hawaii at Manoa. -

Park, H.-S. (2000). Markedness and learning principles in SLA: centering on
acquisition of relative clauses. Journal of Pan-Paciﬁc Association of Applied
Linguistics 4, 87-114.

Pavesi, M. (1986). Markedness, discoursal modes, and relative clause formation in a
formal and an informal context. Studies in Second Language Acquisitiqn 8, 38-
55.

Pérez-Leroux, A. T. (1995). Resumptives in the acquisition of relative clauses.
Language Acquisition 4, 105-138.

Prévost, P. and White, L. (2000). Missing surface inﬂection or impairment in second
language acquisition? evidence ﬁ'om tense and agreement. Second Language
Research 16, 103-133.

Prideaux, G D. and Baker, M. (1986). Strategies and Structures: The Processing of
Relative Clauses. Philadelphia, PA: John Benjamins Publishing Company.

Quirk, R, Greenbaurn, S, Leech, G and Svartvik, J. (1985). A Comprehensive
Grammar pf the English Language. London: Longman.

Radford, A. (1988). Small children’s small clauses. Transactions of the Philologipal
Society 86, 1-43.

Radford, A. (1990). Smtactic Theory and the Acquisition of English Smtax.
Cambridge, MA: Blackwell Publishing.

Radford, A. (1997). Syutactic Theog and the Structure of English: A Minimalist
Approach. New York: Cambridge University Press.

Rizzi, L. (1986). Null objects in Italian and the theory of pro. Linguistic Inquuy' 17,
501-557.

Rizzi, L. (1990). Relativized Minimality. Cambridge, MA: The MIT Press.

Sadighi, F. (1994). The acquisition of English restrictive relative clauses by Chinese,
Japanese, and Korean adult native Speakers. IRAL 32, 141-153.

 

Saﬁr, K. (1986). Relative clauses in a theory of binding and levels. Linguistic Inquuy'
17, 663-689.

Saﬁr, K. (1999). Vehicle change and reconstruction in A-chains. Linguistic Inquu'y 30,
587-620.

162

Sauerland, U. (1998). On the Making of Meaning of Chains. Doctoral dissertation.
MIT.

Schwartz, B. and Sprouse, R. (1994). Word order and nominative case in nonnative
language acquisition: a longitudinal study of (L1) Turkish German

interlanguage. Language Acquisition Studies in Generative Grammar, ed. by T.
Hoekstra and B. Schwartz, 317-368. Amsterdam: John Benjamins Publishing

Company.

Schwartz, B. and Sprouse, R. (1996). L2 cognitive states and the full transfer/full
access model. Second Language Research 12, 40-72.

Schwartz, B. D. (1998). On two hypothesis of “transfer” in L2A: minimal trees and
absolute Ll inﬂuence. The Generative Study of Second Language Acquisition.
ed. by S. Flynn, G Martohardjono, and W. O’Neil, 35-60. Mahwah, NJ:
Lawrence Erlbaum Associates.

Selinker, L. and Lakshmanan, U. (2001). How do we know what we know?; why do
we believe what we believe? Second Language Research 17, 323-328.

Seppanen, A. (1997). Relative that and prepositional complementation. English
Language and Linguistics 1, 111-133.

Seppanen, A. and Kjellmer, G (1995). The dog that’s leg was run over: on the genitive
of the relative pronoun. English Studies 76, 389-400.

Sharvit, Y. (1999a). Functional relative clauses. Linguistics and Philosophy 22, 447-
478.

Sharvit, Y. (1999b). Resumptive pronouns in relative clauses. Natural Language and
Linguistic Theogy 17, 587-612.

Shin, S.-S. and Lee, H.-Y. (1992). Relativization in Korean. Language Research 28,
383-398.

Shlonsky, U. (1992). Resumptive pronouns as a last resort. Linguistic Inquuy' 23, 443-
468.

Siegel, S. and Castellan, Jr. H. J. (1988). Nonparametric Statistics for the Behavioral
Sciences. (2nd edition). New York: McGraw-Hill Humanities.

Solan, L. and Roeper, T. (1978). Children’s use of syntactic structure in interpreting

relative clauses. University of Massachusetts Occasional Papers in Linguistics
4, 106-126.

Song, J .-J . (1991). Korean relative clause constructions: conspiracy and pragmatics.

Australian Journal of Linguistics 11, 195-220.

163

Sprouse, R. A. and Schwartz, B. D. (1998). In defense of Full Transfer in Gerrnan-
English and French-English interlanguage: comparative L2 acquisition research.
Proceedings of the 22nd Annual Boston University Conference on Languaga
Development, ed. by A. Greenhill, M. Hughes, H. Littleﬁeld, and H. Walsh,
726-736. Somerville, MA: Cascadilla Press.

Stahlke, H. F. W. (1976). Which that, Language 52, 584-610.

Suﬁer, M. (1998). Resumptive restrictive relatives: a crosslinguistic perspective.
Language 74, 335-364.

Tagashira, Y. (1972). Relative clauses in Korean. The Chicago Which Hunt: Papers
from the Relative Clause Festixal, ed. by P. M. Peranteau, J. N. Levi, and G C.
Phares, 215-229. Chicago: Chicago Linguistic Society.

Tanaka, Y. (1996). The Comprehension and Acquisition of Relative Clauses by

Japanese High School Students through Formal Instruction. Doctoral
dissertation. Temple University.

 

Tezel, Z. (1998). The Acquisition of Restrictive Relative Clause Conﬁgurations in
English as a Second Language. Doctoral dissertation. Indiana University

Tarallo, F. and Myhill, J. (1983). Interlanguage and natural language processing in
second language acquisition. Language Learning 33, 55-76.

Tavakolian, S. L. (1981). The conjoined-clause analysis of relative clauses. Language
Acquisition and Linguistics Theogy, ed. by S. L. Tavakolian, 167-187.
Cambridge, MA: MIT Press.

Thorton, R. (1996). Elicited production. Methods for Assessing Children’s Smtax, ed.
by D. McDaniel, C. McKee, and H. S. Cairns, 77-102. Cambridge, MA: The

MIT Press.

Tsimpli, I. M. (1997). Resumptive strategies and L2A: a minimalist account.

Proceedings of the let Annual Boston University Conference on Language
Development. ed. by Elizabeth Hughes, Mary Hughes, and Annabel Greenhill,

639-655. Somerville, MA: Cascadilla Press.

Vainikka, A. and Young-Scholten, M. (1994). Direct access to X’-theory: evidence
from Korean and Turkish adults learning German. Language Acquisition
Studies in Generative Grammar, ed. by T. Hoekstra and B. Schwartz, 256-316.
Amsterdam: John Benjamins Publishing Company.

Vainikka, A. and Young-Scholten, M. (1996a). Gradual development in L2 phrase
structure. Second Language Research 12, 7-39.

164

Vainikka, A. and Young-Scholten, M. (1996b). The early stages in adult L2 syntax:
additional evidence from Romance Speakers. Second Language Research 12,
140-176.

Vainikka, A. and Young-Scholten, M. (1998). The initial state in the L2 acquisition of

phrase structure. The Generative Study of Second Language Acquisition, ed. by
S. Flynn, G Martohardjono, and W. O’Neil, 17-34. Mahwah, NJ: Lawrence

Erlbaum Associates.

Van der Auwera, J. (1984). Subject vs. non-subject asymmetries in the relativization of
embedded NP’S. Sentential Complementation, ed. by W. de Geest and Y.
Putseys, 257-269. Dordrecht: Foris Publications.

Van der Auwera, J. (1985). Relative that — A centennial dispute. Journal of Linguistics
21, 149-179.

Van Riemsdijk, H. and Williams, E. (1981). NP-structure. The Linguistic Review 1,
171-217.

Wexler, K. (1991). Some issues in the growth of control. Department of Brain &
Cognitive Sciences. MIT Occasional Paper 44, 141.

White, L. (1987). Markedness and second language acquisition. Studies in Second
Language Acquisition 9, 261-286.

Williams, E. (1980). Predication. Linguistic Inquiry 11, 203-238.

Wiltschko, M. (1998). On the syntax and semantics of (relative) pronouns and
determiners. Journal of Comparative Germanic Linguistics 2, 143-181.

Wolfe-Quintero, K. (1992). Learnability and the acquisition of extraction in relative
clauses and wh-questions. Studies in Second Language Acquisition 14, 39-70.

Yamashita, J. (1994). An analysis of relative clauses in the Lancaster/IBM Spoken
English corpus. English Studies 75, 73-84.

Yang, B.-S. (1993). Clause and information structure of Korean relative clauses in role
and reference grammar. ESCOL ’92, 282-293

Zobl, H. and Liceras, J. (1994). Functional categories and acquisition orders. Language
Learning 44, 159-180.

165

IIIIIIIIIIIIIIIIIIIIIIIIIIIIII

llllllllllilllllllllllllllllllllllilllllllllllll