 

THE-isle

mix r“ LIBRARY
' ' ‘ Michigan State
University

This is to certify that the
thesis entitled

SECOND LANGUAGE PROCESSING OF
WH-MOVEMENT IN ENGLISH:
THE EFFECTS OF FIRST LANGUAGE
AND LEARNING ENVIRONMENT

presented by

SE HOON JUNG

has been accepted towards fulfillment
of the requirements for the

Teaching English to Speakers

Master Of Arts degree 'n of Other Lan ua es

_______________._.__———-—-—

    

Major Professor’s Signature

._————’

$39.10

Date

__.__——-——-

MSU is an Afﬁrmative Action/Equal Opportunity Employer

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

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 KIProj/Aoc8Pres/ClRC/DaleDueJndd

SECOND LANGUAGE PROCESSING OF WH-MOVEMENT IN ENGLISH:
THE EFFECTS OF FIRST LANGUAGE AND LEARNING ENVIRONMENT

By

Se Hoon Jung

A THESIS
Submitted to
Michigan State University

in partial fulﬁllment of the requirements
for the degree of

MASTER OF ARTS
Teaching English to Speakers of Other Languages

2010

ABSTRACT

SECOND LANGUAGE PROCESSING OF WH-MOVEMENT IN ENGLISH:
THE EFFECTS OF FIRST LANGUAGE AND LEARNING ENVIRONMENT

By

Se Hoon Jung
The present study, through the replication and extension of J uffs’s (2005) research,
sought to investigate the effects of different L1 syntactic features (including two factors:
lack of wh-movement, and the head-ﬁnal VP feature) and different learning environment
on proﬁcient adult learners’ processing of wh-ﬁller/ gap constructions in English. The
three groups of English learners (Chinese ESL, Korean ESL, and Korean EFL) and a
group of native English speakers as control participated in a word-by-word self-paced
online reading and the consequent grammaticality judgment tasks that included three
types of wh-constructions that violate Subjacency and four types of grammatical long
distance (LD) wh-extraction. While all learner groups showed sensitivity to Subjacency
violation, the analysis of their accuracy scores and word-level reading proﬁles on
grammatical wh-structures showed a strong effect of L1 , in which, while lack of overt
wh-movement in L1 syntax was found to be disadvantageous for all learners’ processing
and comprehension of those structures, the head-ﬁnal characteristic of L1 Korean put
additional burden on L1 Korean learners’ wh-ﬁller/gap processing in English. The
analysis also found a signiﬁcant effect of learning environment, in which the Korean EFL
learners demonstrated greater processing difﬁculty as compared to the Korean ESL

learners especially in the extraction of wh-phrase from the ﬁnite clausal condition.

Table of Contents

List of Tables ....................................................................................... iii
List of Figures ...................................................................................... iii
Introduction .......................................................................................... 1
Review of Literature ............................................................................... 4
Research Questions ................................................................................. 20
Methods ............................................................................................. 21
Results ............................................................................................... 29
General Discussion ................................................................................. 47
Limitation and Future Research .................................................................. 52
Appendices .......................................................................................... 54
References ........................................................................................... 70

Table 1

Table 2

Table 3

Table 4

Table 5

Table 6

Table 7

List of Tables

Summary of Bio-data and Proﬁciency Test results ............................... 26
Mean accuracy scores on ungrammatical wh-extraction. . . . . . . .. 29
Mean accuracy scores on grammatical wh-extraction. . 31

Procedural description of parsing at the critical region:

extraction from ﬁnite clause ............................................................... 37
Means of RT at the critical region in the ﬁnite clausal condition... ............. 38

Procedural description of parsing at the critical region:

extraction from nonﬁnite clause 42

Means of RT at the critical region in the nonﬁnite

clausal condition ..................................................................... 45

Figure 1
Figure 2
Figure 3

Figure 4

Figure 5-

Figure 6

Figure 7

List of Figures

Comparison of mean accuracy scores by L1 .................................. 33
Comparison of mean accuracy scores by structure ........................... 34
Comparison of mean accuracy rates by learning environment .............. 35
Mean RTs on the subject wh-extraction from ﬁnite clauses ................. 39
Mean RTs on the object wh-extraction from ﬁnite clauses .................. 40
Reading proﬁles at the critical region of subject wh-extraction ............ 43
Mean RTs on the subject wh-extraction from nonﬁnite clauses ............ 44

Introduction

Accessibility to Universal Grammar (U G) in second language acquisition has long
been a primary research area speciﬁcally within the generative framework, in which
especially the question of whether or not (or more moderately, ‘to what extent’) adult
learners are able to acquire native-like second language (L2) grammatical knowledge has
been controversial among researchers (see, White, 2003). This question, however, has not
clearly been resolved, mainly due to inconsistent research results observed in a number of
UG-based L2 studies (Juffs, 1998b).

In light of this ongoing controversy, there also have been considerable efforts to
seek an alternative approach as a better model for explaining late learners’ variant and
relatively inferior L2 performance. One such effort is a syntax-based L2 research model
that takes psycholinguistic consideration of ﬁrst language (L1) sentence processing, such
as Active Filler Hypothesis (Frazier, 1987; Frazier & Clifton, 1989) or Principle-based
Parsing (Pritchett, 1992a), into account. The underlying assumption of this approach is
that adult learners may encounter processing problems in parsing certain target language
structures online (e.g., wh-constructions) regardless of their possession of relevant UG-
driven L2 knowledge (Juffs & Harrington, 1995; White & Juffs, 1998; Williams et. al.,
2001; Juffs, 2005). Given this assumption, some studies explored learners’ online
reading patterns to examine how they process the target structures differently from native
speakers, by implementing experimental psycholinguistic methods, such as a moving
window technique (e.g., Juffs & Harrington, 1995, 1996, Juffs, 2005) or an eye-tracking

method (e.g., Frenck-Mestre, C., & Pynte, J ., 1997).

If processability, in addition to competence deﬁcit, can be another problem of
adult learners, then the next question to consider is, “What factors make it more difﬁcult
for late L2 learners to process certain L2 structures?” One aspect that has been steadily
investigated as such a factor is the assumption that syntactic variation between and
among languages may be closely associated with different processing strategies, (e. g.,
Juffs & Harrington, 1995; White & Juffs, 1998). If this is the case, it will become evident
that learners must learn not only the required L2 grammatical system, but also relevant
L2 processing strategies to be fully proﬁcient in their L2 use (Marinis, 2003). In
formulating wh-questions, for example, the way a wh-element is placed in a sentence
varies from language to language. Some languages require a wh-word to be overtly
moved to the sentence-initial (or phrase-initial when it occurs in a relative clause) from
its original position (i.e., wh-movement; English or Spanish, for example). On the other
hand, some other languages require no such movement, at least on the surface level (i.e.,
wh-in-situ; Chinese, Korean or Japanese, for example). Based on such differences in
syntax, L2 learners whose native languages do not require wh-movement may have
difﬁculties in processing English wh-structures because the procedure of ﬁlling the wh-
gap (i.e., ﬁnding the pre-movement position of wh-words) may not be familiar to those
learners.

Taking this syntactic variation of wh-movement into consideration, Juffs and
Harrington (1995) investigated how adult Ll Chinese (i.e., wh-in-situ) ESL (English as a
Second Language) learners process English wh-constructions differently from native
English speakers. They measured individual participant’ word-level self-paced reading

times (RTs) and accuracy rates on ungrammatical and grammatical wh-constructions

using a moving window technique (Just et al., 1982). While their grammatical sensitivity
in rejecting ungrammatical wh-movement was comparable to the control (i.e., accuracy
rates on the ungrammatical wh—constructions), the study found that the Chinese group
showed signiﬁcantly slower reading patterns to those of the control group speciﬁcally at
the possible wh-gaps (i.e., subject or object candidate positions). Juffs and Harrington
claimed that late learners’ relatively poorer processing abilities (as they are compared to
those of the native speakers), caused by syntactic differences between L1 and L2, may be
the locus of adult learners’ inferior performance.

Juffs (2005), based on previous research (e.g., Fender, 2003, Juffs, 1998a), also
considered different headedness of the VP (i.e., head initial and head ﬁnal) as another
variable that may potentially inﬂuence adult L2 processing. With the test items developed
from Juffs and Harrington (1995), Juffs (2005) comparedLl Spanish ESL and L1
Chinese ESL learners’ (both head-initial: SVO) reading proﬁles and accuracy rates on
grammatical wh-extraction to those of Japanese ESL learners (head ﬁnal: SOV). In this
comparison, Japanese learners showed signiﬁcantly lower accuracy rates and slower RTs
than other experimental groups especially when they were to integrate wh-phrases to the
possible argument gaps, implying that the head-ﬁnal feature in their Ll made the target
structures more difﬁcult to process for Japanese speakers; however, as noted by Juffs, it
was not clear whether the Japanese speakers’ less accurate and slower processing patterns
was because of L1 effects or their initially lower L2 proﬁciency. Therefore, by matching
the proﬁciency level between groups, this study seeks to explore this aspect further (i.e.,

the effect of different headedness in L1) through the replication of Juffs (2005).

In addition to investigating the effect of ﬁrst language, the present study also
explores the effect of learning environment on late learners’ second language sentence
processing, namely ESL and EFL (English as a foreign language) contexts. There have
been ongoing debates in the generative SLA literature about the role of learning
environment with relation to the issue of UG availability, which remains controversial.
For example, those who argue against UG operation in L2A have argued that adult L2
learners in a foreign language learning context may not be able to acquire UG-driven L2
grammar because the types and amount of their L2 input is restricted to formal instruction
in a classroom setting, (e.g., Felix & ’Weigle, 1991, Johnson & Newport, 1991). On the
other hand, UG proponents argued that UG access by late learners can take place both in
the foreign and second language learning contexts (e.g., White & Juffs, 1998). With this
in mind, there has been little effort to explore how different learning environment and its
consequences (e. g., different types of input or different amount of input) may play a role
in late learners’ L2 processing of target language structures in real time. Therefore, by
comparing proﬁciency matched Korean ESL and Korean EFL learners’ reading proﬁles
and accuracy rates on the given structures, the present study seeks to reveal how different
learning environment and its consequences (e.g., different amount of exposure to L2 in

daily life) have an effect on late learners’ L2 processing.

Review of Literature
Parametric variation in wh-movement
The syntactic movement operation of wh-elements occurs if a language contains a

strong wh-feature (e.g., English) in the head of CP (i.e., C), in which wh-phrases are to be

raised to the Spec of CP for feature checking (Chomsky, 1995). To take an example from
English, in (2) who takes the movement operation and overtly moves from its original
argument position (i.e., the object argument of kiss, as in (1)) into the Spec of CPI,
During this procedure, movement is constrained by the Subjacency Principle in that a

constituent (i.e., a wh-word) may not cross more than one bounding node1 in a single

movement (Chomsky, 1977; 1981).

(1) John believed (that) Mary kissed who (m).

  
  

(2)
CPl
DP C'
W110i /\

C /TPI\

T!
T VP 1
<didj>
V CP2
believe /\

<Wh01> C'

  
 
 
 
 
 
 

DP T'
May

T VP2
PAST

V DP
kissed <whoz>

 

I In English, a bounding node generally refers to IP or TP, and DP (Hawkins, 2001; White, 2003).

5

Given this constraint on the movement operation, in (3) what,- cannot move to the Spec of
CP2 because that place is already occupied by another wh-element, who,- (i.e., wh-island

constraint). Another way of fronting who,- is moving it directly to the Spec of CPI;

however, this trial causes unacceptable movement (i.e., Subjacency violation) by crossing
two bounding nodes (i.e., TPI and TP2) in one single movement. A similar pattern of
Subjancency violation can also be found in (4):

(3) *[CPl Whati did [TPI Ryan think [CP2 WhOj [TP2 <wh0j> bought ti]]]]?

(4)

P2

DP

 

In (4), while the dashed line indicates the syntactically unacceptable movement, whoi
cannot move to the sentence—initial position because, from the Spec of CP2, it must cross
two bounding nodes (i.e., DP and TPI) in a single movement (i.e., complex NP island).
While languages whose wh-feature is strong must follow the Subjacency constraint

for making grammatically acceptable movements, languages such as Chinese, Korean or
Japanese do not need to obey this principle, at least not overtly in syntax, because wh-
feature of those languages is weak, hence having no reasons of operating movement for
feature checking, at least in syntax boundary. For those that fall into this category, a wh-
word stays in its original argument position overtly, which is shown in (5) through (8)
with examples from Korean and Japanese. In (6) and (8), the subjects of the embedded
clauses (i.e., nu-ka and dare0 ka, respectively) are the requested information in the form

of wh-phrases, and they are required to stay in the same position of Amy in (5) and (7).

9 m

(5) Ryanoun AME Mercedes-reul sa-tta go seng¢kak~he-tta.
Ryan-TOP Amy-NOM Mercedes-ACC buy-PAST-Comp think-past-Dec-Part
‘Ryan thought Amy bought a Mercedes.’

(6) Ryan°un nu-ka Mercedes-reul satta go seng-kak-hen0ni?
Ryan-TOP who-NOM Mercedes-ACC buy-PAST-Comp think—PAST Q-Part

‘Who did Ryan think bought a Mercedes?’

0 Japanese
(7) Ryanowa Amy-ga Mercedes'wo Ka°tta°to omoima-sita

Ryan-TOP Amy-NOM Mercedes-ACC Buy-Past-Comp think-PAST.

‘Ryan thought Amy bought a Mercedes.’

(8) Ryan-wa dare-ga Mercedes0wo ka-tta to omoimaositaoka?
Ryan-TOP who-NOM Mercedes-ACC buy-PAST-Comp think-PAST-Q-Part

‘Who did Ryan think bought a Mercedes?’

The eﬂects of L1 and learning environment on adult L2 acquisition: UG debates

While it is generally agreed that UG access is available in ﬁrst language
acquisition literature, it has been controversial in adult second language acquisition of
morpho-syntax especially in relation to the effect of L1 and learning environment (e.g.,
Bialystok, 1997; Bley-Vroman et al, 1988; Johnson & Newport, 1989, 1991; Li, 1998;
Patkowski, 1980; Schachter, 1989, 1990; Slabakova, 2003; White, 1992; White &
Genesee, 1996; White & Juffs, 1998). This section brieﬂy summarizes the previous UG-
based studies that investigated late learners’ acquisition of L2 Subjacency speciﬁcally in
relation to these two factors.

Bley-Vroman et a1. (1988) sought to ﬁnd evidence of maturational effects on L2
acquisition and examined 92 adult Korean speaking learners of English by testing their
grammatical sensitivity to Subjacency and ECP (Empty Category Principle) violation via
a grammaticality judgment task (GJ T). The focus of the research was whether or not
those adult learners acquired subtle knowledge of English subjacency that is not operative
in their L1. In their experiment, participants showed approximately 75 percent accuracy
rates in rejecting ungrammatical wh-movement and ECP, implying that those participants
correctly understood the rules of wh-movement in English, at least, at above the chance
level. However, the authors suggested two other possible interpretations for such high
accuracy results; ﬁrst, UG could be operative in adult L2 acquisition, but its strength may

be weaker than in child L2 acquisition. Second, a fundamentally different learning

“cognitive system, a general problem solving system”, rather than UG operation, may be
applied in adult L2 learning (p. 27).

Similarly, Schachter (1989; 1990) also investigated L1 Korean late ESL learners’
acquisition of Subjacency via a GJ T. The result showed that the participants did not show
the relevant knowledge of Subj acency by accepting the ungrammatical wh-movement as
correct in many cases; the accuracy rates were even below the chance level (i.e., less than
50 percent in the Yes/No grammaticality judgments). Schachter concluded that late L2
learners cannot access UG, and therefore they cannot acquire UG-driven L2 knowledge
(i.e., they cannot acquire knowledge of the Subjacency Principle in L2 as long as it is not
operative in their L1 5).

Johnson and Newport (1991) tested Chinese-speaking learners of English to
examine the effects of age on L2 acquisition. They divided participants into four groups
based on learners’ age of arrival in the English speaking environment. The test results
from the GJT on Subjacency revealed that participants who were exposed to ESL
contexts after puberty performed signiﬁcantly more poorly on the test as opposed to the
other comparison groups, whereas the three other experimental groups, divided by age of
arrival from the age of 4 to 16 (hence, roughly before puberty), showed results relatively
comparable with the control group.

In addition, Johnson and Newport found that there was a continuous decline in
participants’ performance along with the age of arrival. They concluded that there is a
maturational effect in the process of second language acquisition, and consequently L2
learners would not be able to acquire subtle syntactic features of L2 that are not operative

in their L1 3 if their learning started after puberty (i.e., approximately 15-16 years old,

according to Long, 1990), or if their L2 input before puberty was limited to formal
instruction (e.g., a foreign language environment).

In relation to those studies, Felix and Weigle (1991) investigated more
speciﬁcally the effect of learning environment in SLA. They included English wh-
extraction as a part of the test items in a GJT and tested 77 German high school EFL
students whose exposure to English was limited to formal classroom instruction. For their
English learning, participants usually received rather non-communicative English lessons
(e. g., traditional audio-lingual and grammar-translation methods) and their L2 proﬁciency
was represented roughly as “beginning, intermediate, and advanced learners” by the
researchers (p. 166). In their conclusion, Felix and Weigle claimed that EFL learners
whose L2 input is restricted to formal classroom settings cannot access UG.

On the other hand, White and Juffs (1998), based on their ﬁndings, argued that
late learners can access to UG regardless of their learning environments. They examined
ESL/EFL learners’ grammaticality judgments on Subjacency violation in English and
their sentence-level reading proﬁles on grammatical wh-constructions. The results
showed that both ESL and EFL learners performed well, showing almost equivalent
accuracy rates in rejecting ungrammatical wh-extraction as compared to the control.
From this result, the authors claimed that those adult learners were able to correctly
distinguish possible and impossible wh-movement in English although Subjacency is not
attainable in their L1 (i.e., Chinese). The authors concluded that UG is available
regardless of the place (i.e., types or amounts of L2 input) in which L2 learning occurs.

In addition to investigating the availability of UG, White and Juffs (1998) also

attempted to explore the effect of learner processability in the same study by measuring

10

participants’ sentence-level RTs and comparing them to those of the control. This
comparison showed that although learners’ average RTs were signiﬁcantly slower than
native speakers, their accuracy rates and reading patterns in terms of predictable
difﬁculty were comparable to the control group; all groups performed better on object
extraction than subject extraction. While this ﬁnding was correspondent to that of Jorden
(1991), Schacter (1990) and Schacter and Yip (1991), White and Juffs suggested that
processing difficulties could be a better explanation for learners’ relatively poorer
performance on grammatical subject wh-extraction (i.e., less accurate for their
grammatical judgments and slower in their readings). Although their aim was in

supporting UG availability in the L2 domain, this implication they suggested became a

crucial starting point for later studies (Juffs, 2005; Juffs & Harrington, 19952).

The ejﬁzcts of L1 and learning environment: L2 processing of wh—ﬁller/gap constructions
While UG-based research has not provided reliable accounts for inconsistent L2
performance of late learners, research interest in how late learners actually process L2
structures online has steadily grown. Followed by White and Juffs (1998), Juffs and
Harrington (1995) focused more on aspect of L2 sentence processing. They brought the
Principle-based Parsing framework that has been widely accepted in the L1 processing
literature (Fender, 2008; Gibson and Hickok, 1993; Stowe, 1986; Williams, 2006) in
which Juffs and Harrington adopted two primary principles proposed by Pritchett (1992a),

as shown in (9) and (10).

 

2 It is necessary to note for clariﬁcation that the experiment in White and Juffs (1998) preceded the one
reported in Juffs and Harrington (1995) (Juffs, 2005, p.123).

11

(9) Theta Reanalysis Constraint (TRC):
“Syntactic reanalysis which re-interprets a theta-marked constituent as outside of a
current theta domain is costly.” (Pritchett, 1992a, p.15)
(10) Generalized Theta Attachment (GTA):
“Every principle of the syntax attempts to be maximally satisﬁed at every point

during processing.” (Pritchett, 1992a, p.138)

The intrinsic basis for these two principles is that the parser processes the sentence
in an incremental manner when it attempts to ﬁnd the place for wh~ﬁllers. In other words,
native speakers try to ﬁll the wh-gap as soon as possible as they read the sentence
incrementally word-by-word (i.e., GTA). The incremental characteristic of sentence
processing is also assumed as a native speaker’s basic processing strategy in Frazier’s
(1987) Active Filler Strategy (AF S) (Dallas & Kaan, 2008). AFS assumes that the ﬁller’ s'
syntactic category is hypothesized at each grammatically possible position and each gap
is tested as a candidate for a ﬁller position until the parser successfully ﬁnd the right
place (Frazier and Clifton, 1989).

The evidence of incremental processing manner can be found in processing Garden

Path (GP) sentences, an example of which is shown in (l 1):
(11) After Steve ate the soup proved to be poisoned (Pritchett, 1992b, p. 326).

In processing this sentence, the soup would be initially analyzed as a direct object of ate,
but then this analysis must be revised when the parser runs into the incoming word
proved; the soup must be reanalyzed as a subject of proved in the main clause. During

this procedure, additional processing burden may be required for readers under the

12

prediction by the TRC because the syntactic reanalysis has to be made from the adverbial
to the main clause (i.e., reinterpretation of the argument, the soup across the different
theta domain; ate 9 proved domain), possibly causing garden-path effects (Juffs and
Harrington, 1996).

While assuming that Principle-based Parsing accounts might work in describing
late learners’ inferior L2 performance, Juffs and Harrington (1995) hypothesized that late
English learners whose Lls lack Subjacency Principle (i.e., wh-in-situ) would encounter
processing difficulty to some extent when they are involved in a wh-ﬁller/ gap procedure.
This is because the incremental wh-ﬁller/ gap procedure—which has been generally
accepted as a general parsing mechanism by native English speakers—is not required in
their L1 processing of wh-constructions. The test items they analyzed include three types
of ungrammatical wh-extraction and four types of grammatical wh-extraction. The
grammatical wh-extraction types are shown in (12) through (15) with the sub-details of
incremental ﬁller-gap procedures in each structure; the examples and its procedural

description of each were described based on Juffs and Harrington (1995) and Juffs (2005):

(12) Who does Andy expect to meet Laura? (subLect extractiog, nonﬁnite clauses)

a) Initial posit: wh-word as an object of the main clause
' Who,- does Andy expect t,- ?

b) Reanalysis: object 9 subject of the embedded nonﬁnite clause
0 Who,- does Andy expect [1p t,- to]?

c) Reanalysis: subject 9 object of the embedded clause

0 Who,- does Andyk expect [Cp [1p PR0,c to meet t,- ?]]

13

d) Reanalysis and parsing termination: object 9 subject of the embedded clause

0 Who,- does Andyk expect [(7) [11) t; to meet Laura? ] ]

(13) Who did the manager expect to hire? (Object extraction, nonﬁnite clauses)

a) Initial Posit: wh-word as an object in the main clause
0 Who did the manager expect t,- ?
b) Reanalysis: object of the main clause 9 subject of the embedded clause
° Who did the manager,( expect [1p t,- to]?
c) Reanalysis and parsing termination: subject 9 object of the embedded clause
' Who,- did the managerk expect [Cp {1}) PRO;( to hire tJ] ?
(14) Who; did Soojin think t; loves Jenny? (subject extraction, ﬁnite clauses)
a) Initial posit: wh-word as an object of the main clause
' Who,- did Soojin think t,-?
b) Reanalysis: object 9 subject of the embedded ﬁnite clause
0 Who,- did Soojin think [[p t,- loves? ]
c) Parsing termination
0 Who,- did Soojin think [11) t,- loves Jenny]

(15) Who does John think Ryan met yesterday? (object extraction, ﬁnite clauses)

a) Initial Posit: wh-word as an object of the main verb believe
0 Who,- does John think t,-?
b) Reanalysis and parsing termination: object (main clause 9 embedded clause)

0 Who,- does John think Ryan met t,~ yesterday?

14

As shown above, the examples in (12) ~ (15), through its procedural description,
show that there can be different levels of parsing complexity depending on the place of
the gap position and the ﬁniteness of the embedded clause, in which extraction of subject
are expected to be relatively more difficult to process than extraction of object (i.e.,
subject-object asymmetry) within the identical clausal condition, and subject wh-
extraction ﬁ'om nonﬁnite clause requires the most complicated parsing process (i.e., more
frequent experience of TRC). To take an example from (12), the extraction of wh-phrases
from a subject gap in a nonﬁnite clause is considered to be the most complicated
structure in terms of its processing complexity. In (12), the wh-word is initially tested in
the object position of the matrix verb expect (12a), then the parser must change this
hypothesis as it encounters the next word to; the initially posited object is to be analyzed
as a subject in the nonﬁnite embedded clause (12b). The second reanalysis occurs in (12c)
after the parser processes the next unit, to in which the previously posited subject must be
reanalyzed as an object of the embedded clause, as in (12c). Lastly, the parser must go
back to its second analysis (12b) when the object trace is ﬁlled with Mary. On the other
hand, object extraction from ﬁnite clauses seems to require relatively less complicated
action for parsing, in which the parser takes only one time of theta reanalysis, as in (15).

As noted earlier, White and Juffs (1998) speculated that there may be
processability problems in late learners’ L2 performance. However, because they simply
measured sentence-level RTs, it was not clear at which points learners had more
processing problems thereby spending more time. To clarify this issue, Juffs and
Harrington (1995) conducted a test of Subjacency with late Ll Chinese ESL learners by

implementing an on—line moving window procedure and the consequent GJ T. In this

15

procedure, each sentence was given one word at a time, and each individual participant
was allowed to control the pace of word presentation by pushing the designated button at
his/her own pace. In this way, Juffs and Harrington measured each individual’s word-
level RTs. The results showed that Chinese participants’ performance on the
ungrammatical wh-extraction was comparable to that of the control group, but that they
were less accurate and slower on grammatical subject wh—extraction, which, according to
the Principle-based Parsing account, requires heavier parsing procedure (e.g., case
reassignment, reanalysis of structure position). In the conclusion, the authors claimed that
adult learners’ processing deﬁcit, rather than deﬁcient L2 competence, may be the better
explanation for adult L2 learners’ inferior performance on the given stimuli.

In a follow-up study, Juffs (2005) replicated Juffs and Harrington (1995) with
three highly proﬁcient groups of ESL groups: L1 Spanish, Ll Chinese, and LI Japanese
speakers. The L1 Spanish group (wh-movement & SVO in L1) was added to conﬁrm the
L1 effects (absence/presence of wh-movement in L1) found in the previous research. By
comparing accuracy scores and reading patterns of the L1 Spanish (wh-movement) ESL
to those of the L1 Chinese Japanese (wh-in-situ) ESL groups, the author hypothesized
that the Spanish group would outperform the Chinese and Japanese groups because,
Spanish speakers might be more familiar with the incremental wh-ﬁller/gap activities
from their L1 processing experiences (i.e., positive transfer).

In addition, J uffs (2005) also considered different headedness of VP (i.e., word
order) as another independent variable in this study; thus the Japanese group (SOV, head-
ﬁnal) was also included. The prediction of this factor was based on Fender (2003) and

Juffs (1998a; 1998b). In those studies, L2 learners whose Lls are the head-ﬁnal (i.e.,

16

SOV) showed more difﬁculties “when incrementally integrating English words into
phrase and clause structures” (Fender, 2003, p.296). In SOV languages such as Japanese
and Korean, arguments of a verb (e. g., subject or object) are processed before the verb.
Given that a verb generally constrains its arguments, the parser in those head-fmal
languages may delay its decision of case and theta assignment until it processes the verb
(Pritchett, 1992b). Taking such differences of parsing strategies between Ll (i.e.,
Japanese, head-ﬁnal: delay of argument integrations into VP) and L2 (i.e., English, head-
initial: argument integrations in an incremental fashion), J uffs postulated that L1
Japanese ESL learners may have additional processing loads as compared to the other
ESL groups when parsing the head initial English wh-constructions.

To brieﬂy summarize the results of Juffs (2005), the effect of the L1 in terms of
wh-movement was conﬁrmed to some extent; while all participants showed subject-
object asymmetry, the Spanish group showed better accuracy rates and faster RTs than
the Chinese and Japanese groups in processing grammatical wh-extraction. In terms of
different headedness of VP, the Japanese group showed more difﬁculties not only in
ﬁlling subject gaps, but also in “integrating objects into the parse” (p.143). However,
Juffs could not elicit a strong claim that different headedness of VP in L1 may play a role
in L2 sentence processing because, as previously mentioned, the L1 Japanese group’s
English proﬁciency level was initially lower than the other groups. Thus it was not clear
whether their poor performance was inﬂuenced by their L1 or by their low L2 proﬁciency.

Another interesting ﬁnding from J uffs’s (2005) study was that the learner group
showed the greatest difﬁculties in processing subject gaps in the ﬁnite clausal condition

while the control group, consistent with the prediction of Principle-based Parsing, had

17

relatively more problems in ﬁlling subject gaps in the nonﬁnite clausal conditions. Such
discrepancy between natives and nonnatives was also found in Juffs and Harrington
(1995), in which L1 Chinese ESL learners also showed the lowest accuracy scores and
the longest RTs on the subject wh-extraction from ﬁnite clause. J uffs claimed that this
result might be due to “the severe garden path effects” that are caused by two adjacent
ﬁnite verbs (Juffs, 2005, p.144), through which L2 learners might experience
comparatively more parsing difﬁculty than native speakers. Jelliffe and Juffs (2004) also
found similar patterns when they investigated the two types of English relative clauses
with L1 Chinese and L1 Spanish ESL learners, in which English learners spent
signiﬁcantly longer RTs when they processed object relative clauses where two adjacent
verbs were presented (e. g., ‘the tiger that the lion chased climbed the tree ’ as compared

to “the tiger that chased the lion climbed the tree’, as cited in Juffs, 2005, p. 144).

 

 

Based on the review of literature, the main purpose for this replication study can
be summarized as follows: Firstly, Juffs (2005) was not able to make a strong prediction
about the effect of the different headedness of VP on late learners’ performance on L2
English wh-ﬁller/gap parsing mainly due to initially different English proﬁciency
between the Japanese and the other experimental groups. Therefore, it is necessary to
scrutinize this aspect again by matching group proﬁciency.

Secondly, as noted earlier, there has been little attempt to examine the role of L2
learning environment on late learners’ L2 processing of wh-ﬁller/ gap constructions
speciﬁcally when there is no such processing strategy in their L1. Although not directly
related to the current research topic, Dussias and Sagarra (2007) investigated how

differential amount of exposure to L2 (i.e., immersion experience in a ESL environment

18

vs. limited L2 English experience in Spain) might have an effect on proﬁcient Spanish-
English bilinguals’ processing of L1 Spanish relative clauses, speciﬁcally in relation to
their NP (i.e., an antecedent) attachment preferences, in which they revealed that those
learners in the ESL context tend to employ L2 processing strategies in their L1
processing. They argued that daily exposure to L2 English inﬂuenced their NP
attachment preferences in L1. Taking this into consideration, it seems to be reasonable to
hypothesize that different amount of exposure to the target languages may play a role in
the development or implementation of learners’ target language processing strategies.

As mentioned previously, White and Juffs (1998) addressed the issue of
processability in relation to the learning environment factor; however, they could not
observe what types of processing strategy were actually implemented by learners because
they simply measured sentence-level RTs. Furthermore, those two L2 groups (Chinese
ESL and EFL) signiﬁcantly differed in terms of their mean age at the time of the test and
of their ﬁrst exposure to communicative language use; the mean age difference between
the two was almost 10 years3, and the mean ages of ﬁrst exposure to “nonformal L2 input”
were signiﬁcantly different ﬁ'om one another, according to the authors (p. 125).

Given that the age-related loss in language ability may be cumulative even after
puberty (e. g., Long, 1990), this casts doubt whether those comparisons between ESL and
EFL truly reﬂect the effect of learning environment. Therefore, age factors (both in terms
of mean age at the test and the ﬁrst exposure to the communicative L2 uses) need to be

controlled for ESL and EFL groups to investigate the effects of environment

 

3 Two groups’ mean ages at the time of the test were 32.25 years old for the ESL and 22.8 for the EFL.

19

independently. With these matters in mind, the present study seeks to ﬁnd the answers

from the following research questions.

Research Questions

1.

Do proﬁcient late English learners (including both ESL and EFL) show
grammatical sensitivity to structures that violate Subjacency? That is, are their
accuracy rates on ungrammatical wh-extraction comparable to those of the control?
Do late English learners show subject-object asymmetry when processing types of
grammatical wh-extraction? That is, do they have more difﬁculty in processing
subject extraction than object extraction as predicted by Principle-based Parsing? If
so, between subject extractions from ﬁnite and non-ﬁnite clauses, which
construction is more difﬁcult for learners to parse?

Are Chinese ESL and Korean ESL learners less accurate in their grammaticality
judgments and slower at the critical regions than native speakers when they process
types of grammatical long distance wh-extraction? If so, are Chinese ESL learners
more accurate and faster than Korean ESL learners? That is, does the head-ﬁnal
characteristic in Korean make wh-ﬁller/ gap procedures more difﬁcult for Korean
English learners to perform as compared to L1 Chinese ESL learners whose L1
headedness is head-initial like English?

In their performance on grammatical LD wh-extraction, are Korean ESL learners
more accurate (in their grammaticality judgments) and faster ((in ﬁlling wh-gaps)
than Korean EFL learners? If so, does it indicate differences in learning
environments can play a role in late learners’ processing of English wh-

constrictions?

20

Methods

With these research questions in mind, this study implemented a moving window
procedure, which includes both self-paced sentence reading and the consequent GJ T on
each stimulus. Participants’ accuracy rates on each structure and word-level RTs on each
stimulus sentence were measured ﬁ'om this procedure.

In order to investigate the effect of L1 (i.e., absence of Subjacency in L1 and the
different headedness of VP), Chinese and Korean ESL learners were included in the
research as well as the control group. While these two languages are characterized as wh-
in-situ languages, they are different in terms of the headedness of VP; Chinese is a head-
initial language (i.e., SVO), while Korean is a head ﬁnal language (i.e., SOV). Korean,
in many respects including word order (i.e., headedness of VP) or case marking systems,
is similar to Japanese (Fender, 2008; Pritchett, 1992b). Given that, replicating Juffs (2005)
with Korean English learners is relevant as a way of conﬁrming L1 effects of the
headedness of VP on L2 English wh-ﬁller/ gap processing. In addition, Korean learners of
English in a foreign language (i.e., EFL) context were also included to investigate the
effect of learning environment.

The experiment took place in two places: 1) the SL3 (Second Language Studies)
testing room in Wells Hall (BIOOD) at Michigan State University for the ESL and control
group experiments, and 2) CSL (Center for the Study of Language) Laboratory at Kyung
Hee University in Seoul, Korea, for the EFL experiment. Both places were equipped with
desktop personal computers and 15 or 17 inch LCD monitors, and the experiment was

carried out individually using those computers.

21

Individual participants attended a single 40~90 minute session by appointment
with the researcher. During this time, all participants except native English speakers
completed: 1) Consent form (see Appendix A), 2) proﬁciency test, 3) background
questionnaire (see Appendix B), and 4) moving window task. Participants who
participated as control only completed 1) and 4). After completing the moving window
task, all participants were given brief information about the present study through the
Debrieﬁng form (see, Appendix E), and then, as advertised, were paid either 10 or 15

dollars (U SD) cash or the equivalent 15,000 Won (KW) for compensation“.

Participants

Because the goal of this study was to investigate the role of learner processability
on second language acquisition by highly proﬁcient late learners, L2 English proﬁciency
was monitored and controlled through the administration of MTELP (Michigan Test of
English Language Proﬁciency). If one’s L2 proﬁciency level is not high enough, and if
one’s performance in terms of accuracy and reading proﬁles is poor, then it may become
unclear whether those poor outcomes were caused by their initially low proﬁciency level
or their processing problems. As a result, recruiting highly proﬁcient L2 English learners
was necessary to avoid this conﬂict. Participants’ proﬁciency level was assessed through
the grammar (section 1: 40 test items) and vocabulary (section 2: 40 test items) section of
MTELP, following Juffs and Harrington (1995), White and Juffs (1998), and Juffs (2005).

Through the administration of MTELP, 2 participants in the Korean EFL group

were excluded because of their low scores. For this exclusion, proﬁciency test result of

 

4 During the preliminary test, ESL learners spent approximately 80 minutes while native English speakers
spent about 35 minutes for the experiment. It was because native Speakers did not take the proﬁciency test
and background questionnaire. Given the differences of time spent, subjects who participated as a control
were paid 10 dollars and subjects who are nonnative English speakers were paid 15 dollars.

22

Chinese ESL group in Juffs (2005) was used to determine the outliers for this study. In
Juffs (2005), the Chinese group showed the highest mean scores on the proﬁciency test
among all experimental groups (p. 131). Taking this into consideration, their total mean
scores (58.03) minus one standard deviation (12.59) was used as the cut-off score (i.e.,
45.44). In addition, if one did not achieve 60 percent accuracy on the three types of ﬁller
(declaratives including complex NP or relative clause, and short distance wh-movement)
or 50 percent accuracy on their overall grammaticality judgments, then his/her data were
excluded in the analysis. The ﬁrst two ﬁllers were the ones Juffs (2005) used for data
exclusion. The reason was that if participants cannot make correct judgment on sentences
that are structurally simpler than the targeted structure, it would not be likely that they
can judge long-distance wh-extraction correctly (p. 130). With these two types, this study
added a structure that includes a short distance wh-movement for the same reason. One
participant in the Korean EFL group who achieved 46 percent accuracy overall on the
GJT was also taken out from the analysis.

In addition to proﬁciency, participants’ ages at ﬁrst exposure to communicative
English use were controlled during the recruitment process. In the present study, the
deﬁnition of late learners followed Johnson & Newport (1991) and White and Juffs
(1998), in which late learners were considered as those who were not exposed to
communicative L2 use before age of 16. This condition was informed to the participants
as they read the eligibility of participation in the consent form, and the speciﬁc time at

ﬁrst exposure to communicative English was asked through the background questionnaire.

23

All except one Korean ESL participant satisﬁed this conditions.

Both ESL groups and the control group consisted mostly of graduate students who
were enrolled either in the MA TESOL (Teaching English to Speakers of Other
Languages) program or SLS (Second Language Studies) Ph.D. program at Michigan
State University (MSU). These students are generally considered to be highly proﬁcient
in English6. One Chinese participant was a graduate student enrolled in the College of
Education at MSU and two Korean participants were also graduate students enrolled in
the School of Journalism at MSU. For the Korean EFL group, graduate students either in
the English Language and Literature department or the English Education department at
Kyung Hee University participated in the experiment.

Participants’ bio-data and MTELP results were compared using one-way
AN OVA with alpha set at .05, a summary of which is shown in Table 1. One noticeable
improvement from the original study (i.e., Juffs, 2005) was that all groups showed
statistically comparable proﬁciency level from the test. Length of residence was
signiﬁcantly different because of the Korean EFL group, but the Chinese and Korean
groups were not different from each other.

Participants’ amount of English use in their daily life was asked through the
background questionnaire as a supplementary tool to get more insight on ESL and EFL

learners’ English usage habits (see, Table 1). The statistical analysis showed that both

 

5 This participant reported that she had a 1 hour English conversation class (3 times a week, for 2 months)
in Korea at the age of 13; however, that class was taught by a Korean English teacher, and it was not an
entire immersion class, according to the participant. Given that, the data ﬁ'om this participant was kept for
further analysis

6 The minimum TOEFL (Test of English as a Foreign Language) score for admissions to the MA TESOL
and SLS program is 600 on the PET, 250 on the CBT, or 100 on the IBT format. The minimum score for
the IELTS (International English Language Testing System) is 7.0 for MA TESOL and 7.5 for SLS
program, respectively. (This information can be found at: 1) http://www.msu.edu /~lIngIang/tesol
/admission.html, and 2) http://sls.msu.edu/index.php/admissions, respectively.

24

Chinese and Korean ESL learners spent signiﬁcantly more time in using English than
Korean EFL learners. Korean ESL learners used English slightly less time than Chinese

ESL learners, but they were not signiﬁcantly different from one another.

Stimuli

The tested sentences used in the present study were developed based on the items
used in Juffs (2005)7 (see Appendix C). The items in the self-paced reading task and the
grammaticality judgment task were made up of 27 structure types (6 sentences for each
structure, totaling 162 sentences), including 4 types of ungrammatical wh-extraction, 5
types of grammatical LD wh-extraction, 2 types of simple wh-extraction (i.e., short-
distance), 3 types of garden path structures, 6 types of declaratives containing reduced
relative clauses, 1 type of ungrammatical sentences with overpassivization, 3 types of
transitivity related structures, 2 types of declaratives containing complex NP and relative
clause, and 1 type of nonsensical sentences . Among these structures, 3 types of
ungrammatical wh-extraction8 and 4 types of grammatical wh-extraction (examples in
(12) - (15)) were used for the analysis of the current research. Examples of

ungrammatical wh-extraction that violate Subjacency are shown in (16) through (18).

(16) * What,- did Tom hear the story that Ann cooked t,- for Bill? (*Complex NP island)
(17) * Who,- did John call his mom after he kissed t,- last week? (*Adjunct clause island)

(18) * What, did Sam see the man who wanted t,~ from Jane? (*Relative clause island)

 

7 The stimuli were provided by Dr. Alan Juffs at University of Pittsburgh and used in the present research
under his permission.
The construction of ungrammatical subject wh-extraction from ﬁnite clauses with the complementizer

that (Structure Type #4) was initially included for analysis. However, it had to be excluded due to low
mean accuracy of the control group on sentences of this type, which was below the chance level (41.02%).

25

426. mo. v R as “coach? buqaowmawmm 053 8.38 95.0 ...
.352 S. v A 8 about? bugommawmm 29$ 38:. @595 I.
._o>o_ So. v a. 8 “cocotmu begowﬁwmm 203 838 9.80 1....

.8:on 50253 bouncotmu «505:me 2: 88%.: 93885-8 98 85 £802 .862

 

 

 

8... a. .N as 28.: 2.9. .23... 3.2 .38 r... .855 was 8.38. .8 .825 .59 583
.3. a. .N 3.8 .88 8... m... 3.8. .88. .. .285 8288 ~8.8m. =. Bass 533
a... a. .N NE. ”a... 8.... 2.... 8.2 ”8.8 r. .28... 8.58... 8.8m a. .825 583
:3. a. .m 8.... .323 88 .88 o: 08.... t... .855 8.83 8.8m a. .825 583
88 a. .N 8... 3:; 8.. . s3... :8 08.8. r. .285 8.83. F8.8m a. .528 583

m... a. .m 2.8 8.8 8.8 8.8 8... $8 .8 u 88: .soe 38.2
mm. a. .N 55 R. . m 2... 8.8 8... 2.; 8.. u :8: 5.883, $th
8. a. .N 22 8. . m PM 8.; 8... ES. 8.. n 88.. asseo .33.).

8.3 Q» .m mod aenmd m3; nnmdm “2.: «no. _ m I... 85:25 85283 mo 3:04
a... Q .N 42 2.8 83. 2.... com 8.... a... 3 888.8858 88 a 8....
8. a. .N 8.. 8d Rm 8.: 5A 8.2 S e 8.8%.”. .2... a a...

.w. a. .N 8.8 3.8 8... 8. . m 8.8 Sam 8...
a. E 3. x. a. 2
a. 8 Anne GT5 GT5
4...”. 523. 4mm 523. 4mm 3.85

 

83.8.. has Ozmmoﬁoi 38 23365.8 gash

_ 2an

26

Some sentences were revised from the original test items to obtain more precise
reading proﬁles. For example, (19) was one of the original sentences used in Juffs (2005)
and it was revised in this study by changing the object DP her sister to his family, and its
antecedent the girl to the boy respectively: the revised sentence is shown in (20).

(19) Who did the girl expect to invite her sister last night? (Original one in Juffs, 2005)

(20) Who did the boy expect to invite his famin last night? (Revised version)

This revision was necessary because of the ambiguity of her in terms of its case
representation; it can be either genitive or accusative case. Given that the parser processes
a sentence incrementally, and that the moving window displays only one word at a time,
her in (19) will be initially interpreted as an object of the verb invite with accusative case
assignment; however, this analysis must be revised to genitive when the parser encounter
the next word sister. Considering this processing aspect, sentences that include genitive
her within the object DP may require additional processing efforts as compared to other
DP types. This may cause the discrepancies between sentences in terms of processing
burden and time. To resolve this problem, six sentences that include genitive her in the

object DPs were revised using another genitive pronoun (e.g., his, your, their).

Moving window procedure

A moving window task was designed using the software package Linger (version
2.94, programmed by Doug Rohde at the Massachusetts Institute of Technology), in
which individual words in a sentence were displayed in a sequential order from left to the
right as participants pressed the space key to display the next word. In this way, their RTs

at each word were recorded. After they read the last word of each sentence, subjects were

27

asked to make a grammaticality judgment on the sentence by pressing the designated key
(green key for ‘YES’ and the red key for ‘NO’), through which individual subjects’
accuracy scores were recorded (1 point for correct judgment, and 0 point for incorrect
judgment). Tested sentences were programmed to appear on a screen in a random order
to avoid potential task effects.

Before the main task, instruction and a practice session were given to all
participants in order for them to have better an understanding about the task procedure
and to be familiar with the task (see, Appendix D). During this session, participants were
instructed l) to read sentences as naturally as possible, and 2) to focus on the structure of
the sentence rather than meaning for their grammaticality decisions. Some participants in
the preliminary test reported that they judged some sentences as incorrect because they

thought the meaning was strange. Two such examples are given in (21) and (22):

(21) The large birds killed in the garden could not see the cat.

(22) The large birds killed almost every week could not see the cat.

Although these sentences are syntactically well-formed, there was a possibility that
readers can encounter a problem in the semantic domain depending on how they interpret
the subject of each sentence. That is, if the subject, the large birds is interpreted as a
direct entity that performs the action of the main verb, see (i.e., agent), the meaning of
each sentence can be perceived as odd because the modifying clause indicates that the
subject is already dead and consequently cannot take any further actions. Taking this into
consideration, it was necessary to make readers focus more on processing of structure

than appropriateness of meaning.

28

Results

Accuracy scores on ungrammatical wh-extraction

The inclusion of ungrammatical wh—extraction to the moving window procedure
and a GJ T was to investigate to what extent late learners whose L1 lacks wh-movement
are sensitive and accurate in detecting types of Subjacency violation in English as
compared to the control group. In order to scrutinize this aspect, one-way AN OVA
(Analysis of Variance) was carried out for all group comparisons at 95 percent
conﬁdence level with the Tukey procedure for post hoc analysis. Table 2 shows a
summary of the mean accuracy scores on ungrammatical wh-extraction (i.e., Subjacency
the score of 1.00 is equivalent to 100 percent accuracy. For example, the overall accuracy
violation), in which all numeric values are rounded off to the second decimal place.
Given the coding methods (1 point for correct answers and ’0 points for incorrect answers),
score of Chinese ESL group on ungrammatical wh-extraction was 0.80, which indicates

that the group made average 14.4 correct judgments from 18 given sentences.

 

 

 

Table 2
Mean accuracy scores on ungrammatical wh-extraction
Chinese ESL Korean Korean EFL English
(n=15) ESL (n=15) (n=15) (n=13)
M (SD) M (SD) M (SD) M (SD)
*Complex NP 0.83a 0.76 0.59a,b 0.94b
islands (n=6) (0.19) (0.23) (0.23) (0.15)
*Adjunct clause 0.74 0.62c 0.67d 0.89c,d
islands (n=6) (0.21) (0.12) (0.30) (0.16)
*Relative clause 0.81 0.80 0.72 0.82
islands (n=6) (0. 15) (0.14) (0.19) (0.08)
Total 0.80e 0.73f 0.66e,g 0.89f,g
Ungrammatical (n=18) (0.11) (0.13) (0.16) (0.09)

 

Note. Means that are co-indexed represent the signiﬁcant difference between groups.

29

There was a signiﬁcant difference in the overall mean accuracy scores among
groups, F (3, 54) = 8.354, p < .001, in which the Chinese ESL group showed the highest
accuracy rates among the learner groups and their mean accuracy was comparable to the
control in all cases. On the other hand, the Korean EFL group displayed the poorest
performance among groups; the mean accuracy scores of the Korean EFL learners were
signiﬁcantly lower than those of the Chinese ESL (p <05) and the control group (p
< .001), but they were not different from the Korean ESL group (p = .467). The Korean
ESL group was not different from the Chinese ESL group (p = .420) in any cases, but
they were reliably different from the control (p < .01)9.

However, these signiﬁcant differences between the Korean ESL/BF L learners and
the control groups must be interpreted with caution. That is, those learners’ relatively
lower accuracy rates might be due to the word-by-word sentence reading task, which may
require readers (including the native speakers) to experience more processing burden for
their comprehensions and grammatical judgments (J uffs, 2005). Furthermore, the means
of the control group (i.e., 89 percent) are much higher than the control group in Juffs
(2005), who scored approximately 74 percent, which made the level of signiﬁcance
between the L1 Korean groups and the control to be more considerable. Taking these into
consideration, all learners groups demonstrated their grammatical sensitivity to the
Subjacency violations, at least to some degree, by correctly rejecting these violations at

above the chance level.

 

9 The source of this difference was their lower accuracy results on the adjunct clause island violation. If
compared after excluding this structure from the analysis, then Korean ESL group also showed comparable
accuracy to the control.

30

Accuracy scores on grammatical wh-extraction

The inclusion of four types of grammatical LD wh-extraction (i.e., wh-ﬁller/gap
procedure, which is expected to experience TRC) was to investigate how syntactically
distinct L1 characteristics (i.e., wh-in-situ and the different headedness of VP) and
different learning environment (i.e., ESL and EFL) affect late learners’ wh-gap ﬁlling
processing. Table 3 provides a summary of each group’s mean accuracy scores on these
structures. All groups except Chinese ESL showed the greatest difﬁculty in subject wh-
extraction from ﬁnite clauses, and object wh-extraction from nonﬁnite clauses were the
easiest for all groups. The Chinese ESL group had the lowest scores on object wh-

extraction from ﬁnite clauses, but they also showed comparably low accuracy scores on

Table 3
Mean accuracy scores on grammatical wh-extraction

 

Chinese ESL Korean ESL Korean EFL English Control

 

 

(n=15) (n=15) (n=15) (n=13)
M (SD) M (SD) M (SD) M (SD)
Subject extraction 0.66 0.59 0.32 0.73
(ﬁnite), (n=6) (0. 1 7) (0.25) (0.28) (0.20)
Object extraction 0.64 0.68 0.52 0.83
(ﬁnite), (n=6) (0.23) (0.16) (0.33) (0.20)
Subject extraction 0.72 0.68 0.61 0.79
(nonﬁnite), (n=6) (0.19) (0.16) (0.23) (0.26)
Object extraction 0.76 0.78 0.74 0.86
(nonﬁnite), (n=6) (0.11) (0.12) (0.18) (0.15)
0.69 0.68 0.55 0.80
Total (n=24)
(0.11) (0.11) (0.15) (0.14)

 

31

subject extraction from ﬁnite clauses"). While these results are consistent with the results
of Juffs (2005), they were different from the prediction made by Principle-based Parsing
that assumes subject wh-extraction from nonﬁnite clauses to be the most difﬁcult
structure in the incremental processing, as in (12). Furthermore, all learner groups were
slightly less accurate in ﬁlling object gaps in ﬁnite clauses than in ﬁlling subject gaps in
nonﬁnite clauses. This result was not only inconsistent with the Principle-based Parsing
account, but it was also different from that of J uffs (2005), in which all groups, including
the control, performed slightly better on obj ect-extraction from ﬁnite clauses than subject
extraction from nonﬁnite clauses. In the present study, although learners showed subject-
object asymmetry within the same clausal condition, the processing problems of learners
seem to be closely associated with the ﬁniteness.

In order to further explore the effect of ﬁrst language and learning environment, a
3 x 4 GLM (General Linear Model) repeated measures AN OVA was implemented at 95
percent conﬁdence level. If any signiﬁcance was found between groups, then the Tukey
procedure was used for all post hoc comparisons. In this analysis procedure, the two
between-subject independent variables (i.e., different L1 and different learning
environment) were separated into two sets of statistical analysis (i.e., 2 sets of AN OVAs),
The separation of those two variables was necessary because the current research did not
include a Chinese EFL group (i.e., no interactions between L1 and learning environment).
If the L1 factor (2 values: Chinese Korean, and the control) and learning environment
factor (3 values: Korean ESL, Korean EFL, and the control) were included as between-

subject variables in a single statistical analysis, then the outcomes would be conceptually

 

10 The low mean score of Chinese group on this structure was due to extremely low score of one Chinese
participant who received a zero score. Mean scores after excluding this participant were higher than those
from subject wh-extraction.

32

problematic due to inappropriate group combinations during the statistical analysis (i.e.,
Chinese ESL and Korean ESL group combination as opposed to the Korean EFL group
for learning environment effect analysis, or Korean ESL and Korean EFL group

combination as opposed to the Chinese ESL group for L1 effect analysis).

a) The eﬂect of L1 : Accuracy comparison of Chinese ESL and Korean ESL
The Chinese ESL, Korean ESL, and the control group were included to explore
the effect of L1. In this analysis, L1 was the between-subject independent variable and
four types of grammatical wh-extraction were within-subj ect independent variables. The
results showed a main effect of L1 , F (2, 40) = 4.529, p < .05, in which the mean
accuracy of the Chinese ESL and Korean ESL group were signiﬁcantly lower than the
control, but not different from each other (p = .944); the mean accuracy of the Chinese

ESL group were slightly higher than those of the Korean ESL group (see Figure 1).

 

 

 

1001
5
:5 9o«
5‘
g 30‘ /
0
<33
5 70* 7% ﬂ
0
2
60

Chinese ESL KoreanESL English Control
L1

Figure 1. Comparison of mean accuracy scores by L1

33

There was also a main effect of structure, F (3, 120) = 4.856, p < .05, indicating
that accuracy scores between structures were signiﬁcantly different from one another;
subject wh-extraction from ﬁnite clauses were the least accurate while object wh-
extraction from nonﬁnite clauses were the most accurate (see Figure 2). There was no
signiﬁcant interaction of L1 and structure, F (6, 120) = .539, p = .778, in which all

participants show similar accuracy patterns to those of four structures.

 

 

 

100 1
g
3 9o «
r32
5‘
g 80 -
8 /
<2 A -
8 70 ~ /
d)
2
60
Subject Object Subject Object
(Finite) (Finite) (Nonﬁnite) (Nonﬁnite)
Structure

Figure 2. Comparison of mean accuracy scores by structure

b) The eﬂect of learning Environment: Comparison of Korean ESL and EFL
The Korean ESL/BF L groups and the control group were compared to examine the
effects of learning environment. The analysis showed a main effect of learning
environment, F (2, 40) = 13.017, p < .001; while both groups were signiﬁcantly inferior
to the control group, Korean EFL group also performed signiﬁcantly less accurately than
the Korean ESL group (see Figure 3). There was also a main effect of structure, F (3,

120)= 10.918, p < .001 , in which accuracy patterns on the given structures were identical

34

to those of L1 effect analysis (see Figure 2). There was no signiﬁcant interaction between
learning environment and structure, F (6, 120) = 1.738, p = .118, implying that both

groups had similar patterns on their grammaticality judgments to the given structures.

 

 

100 -
§
V 90 -
8
53, 80~
g 70 ~
<1
5 60 ~
2

50

Korean ESL Korean EFL English Control
L1 '

Figure 3. Comparison of mean accuracy rates by learning environment

Reading proﬁles on grammatical wh-extraction

Individual participants’ word-level RT data on the four types of grammatical wh-
extractions were analyzed to gain more insight on late learners’ processing patterns. The
data from the control group were not included in the RT analyses. As noted by Juffs
(2005), the inclusion of native speakers in RT analyses always tend to make the
comparisons highly signiﬁcant because their reading patterns are generally much more
stable and faster than those of L2 learners. .

In addition, several treatments were made following general RT analysis
paradigms. First, only RTs that were correctly judged were included for the statistical

analyses. Thus those that were answered incorrectly were coded as missing values and

35

eventually excluded during the individual mean RT calculation process (e.g., Fender,
2008; Juffs, 1998a, 2005; Juffs & Harrington, 1995; 1996). From this exclusion process,
approximately 36 percent of error data from the learner groups had to be eliminated in the
analyses. In addition, RTs that went beyond more than three times standard deviation
from the group mean of that region were also excluded, following the method
implemented in Juffs (2005).

Secondly, the remaining RT data in the analyses had to be log transformed to
compensate for the positively skewed characteristic of RT data (Juffs, 1998a, 2004, 2005,
2006). The values for skewness with the raw data at the critical region (Word 5 & 6 in the
ﬁnite clausal condition, Word 7 & 8 in the nonﬁnite clausal condition) ranged from 1.908
to 9.376. Those highly skewed values were adjusted through the data log transformation,
after which all the skewness values used in the analyses fell into the moderately

acceptable range of -/+2.

RT analyses on grammatical wh-extraction: ﬁnite clausal condition

As was expected, reading proﬁles of the control group were much faster and
ﬂatter than the learner groups across the structures (see Figure 4, 5, 7, and 8) In addition,
both the Korean ESL and EFL learners represented much sharper increased RTs than the
Chinese ESL learners whenever they processed verbs. Figures 4 and 5 show each group’s
word-level reading proﬁles on wh-extraction ﬁ'om ﬁnite clauses; subject and object gap,
respectively. In this clausal condition, the two places were selected as critical regions for
RT analyses following Juffs (2005). These are the ﬁﬂh and sixth word positions where

the parser makes initial posit and reanalysis, as shown in Table 4.

36

Table 4 -
Procedural description of parsing at the critical region: extraction ﬁom ﬁnite clause

 

 

W5 W6
LD Subject mama Reanalysis
extraction Object of the matrix clauses -) Subject of the embedded clause
LD Object I_mt_ial_msl_t: Reanalysis
extraction Object of the matrix clause -) Object of the embedded clause

 

In the case of subject wh-extraction, all learner groups apparently spent longer
time at the subject wh—gap (i.e., pass, the verb of the embedded clause in Figure 4) where
the parser is supposed to experience the TRC. One remarkable result at this region was
that the mean RT of the Korean EFL group was not only faster than that of the Korean
ESL group, but also slightly faster than that of the Chinese ESL group. Furthermore, this
RT pattern appeared similarly at the critical region in the object extraction (i.e., his,
determiner of the subject NP of the embedded clause in Figure 5) in which, while all
learner groups spent much less time as compared to the subject extraction, the Korean
EFL group was slightly faster than the other learner groups. Looking at other regions, all
learner groups displayed increased RTs in the object gap in the embedded clause in
Figure 5 (i.e., passed, the verb of the embedded clause), where the parser conﬁrms the
previously made reanalysis at the Word 6 position in which the two Korean groups were
slightly slower than the Chinese group, but not statistically different from one another
(from one-way ANOVA statistics: F 2, 39, = .818, p = .449).

In order to further examine these different reading proﬁles, two sets of 2 x 2 x 2
GLM repeated measures AN OVAs were carried out at 95percent conﬁdence level with

one between-subject variable (L1 in one set, and learning environment in another,

37

independently) and two within-subj ect variables (2 structure types; subject & object gap,
and 2 places as a critical region: word 5 & 6). The RTS used in this statistical procedure

are shown in Table 5.

 

 

Table 5

Means of RTs at the critical region in the ﬁnite clausal condition
Group Raw Mean (SD) Log-transformed Mean (SD) N”
1 (Chinese ESL) 525.27 (146.09) 6.22 (.28) 14
3‘33?“ 2 (Korean ESL) 732.51 (338.32) 6.51 (.42) 15
3 (Korean EFL) 707.13 (315.05) 6.59 (.41) 10
. 1 503.92 (135.29) 6.19 (.27) 14
03$? 2 727.58 (380.82) 6.46 (.52) 15
3 837.24 (437.51) 6.61 (.50) 10
1 1344.05 (873.83) 6.99 (.59) 14
suggct 2 1903.67 (974.77) 7.37 (.60) 15
3 1230.63 (732.03) 7.01 (.61) 10
1 652.68 (235.80) 6.42 (.37) 14
03:“ 2 696.61 (269.86) 6.49 (.33) 15
3 584.39 (148.69) 6.36 (.30) 10

 

a) The eﬂ'ect of L1 : Chinese and Korean ESL RT 3 (Finite)
The analysis demonstrated no main effect for L1 in the ﬁnite clausal condition,
but it was very close to the signiﬁcance level, F (1, 27) = 4.138, p = .052. There was a
signiﬁcant effect of structure, F (l, 27) = 36.292, p < .001, effect of region, F (1, 27) =
57.867, p < .001, and reliable interaction between structure and region, F (1 , 27) = 33.475,

p<.001.

 

“ There were a few participants who did not make any correct answers on one of the given structures:
Subject extraction: 1 Chinese ESL and 2 Korean ESL participants; Object extraction: 2 Korean EFL
participants. There was 1 Korean EFL participant who had 1 correct answer on the subject extraction, but
the RT at the critical region of the structure was replaced with a missing value because it surpassed the
cutoff line (i.e., 3 times the SD from the mean).

38

woman—o 8am 89a :ouombxoﬁB «8 Sam 05 no mhm 502 .V 95ka

2888 ﬁaemlxl .5 526ml 48 58621-1 .28 685.519.

«$95. 2% S 3983 hi 33.an .28.? 568 of Eu 3%:

 

 

com

.\I .. .. w . . . 1111.-....11 o8

 

coo

 

com _
8.2

K . / \ coc—
/<\ cow ﬂ

Q / /.// \\\
\ ///V \

 

oocN

39

«.865.

2%

$930 BEG Boa nomﬁgoﬁa 8030 05 :o mhm :82 .m. 3wa

2880233511 .5 aooeml 4% 568M181 3mm 9..qu01...

5

nuance.

«28

E

tom?

L323

3»

Sn

BS:

 

 

 

 

 

 

 

 

 

 

 

 

com

cow

coo

cow

coo_

ocm_

cov—

coc_

40

These results indicate that while the Chinese ESL group was slightly faster than the
Korean ESL group, both groups showed very similar reading proﬁles in the given
structures, speciﬁcally within the critical region; both groups spent signiﬁcantly longer
time for parsing reanalysis in the subject extraction than in the object extraction. Because
of such similarities in reading patterns between groups, there was no reliable L1 by
structure interaction, F (1, 27) = 1.675, p = .207, no signiﬁcant L1 by region interaction,
F (1, 27) = .197, p = .661 , and no signiﬁcant L] by structure by region interaction, F (1,

27) = 1.528,p = .227.

b) The eﬂect of learning environment: Korean ESL and EFL RTs (Finite)

Reading proﬁles of Korean ESL and EFL learners were analyzed to see if there
was any effect for learning environment. Similarly to the previous analysis of the L1
effect on RT, there was a main effect for structure, F (l, 23 = 24.042, p < .001), an effect
for region, F (1, 23) = 16.284, p <. 001, and interaction of structure by region, F (1, 23) =
33.475, p <.001, exhibiting subject-object asymmetry in the ﬁnite clausal circumstance.
However, the result revealed no main effect for learning environment F (1, 23) = .211, p
= .651; in both structure types, the Korean EFL group was slightly faster than the Korean
ESL group at the subject gap (i.e., word 6), but they were slightly slower than the Korean
ESL group in the initial gap (i.e., word 5), which brought a signiﬁcant learning
environment by region interaction (F, 1, 23 = 7.535, p < .05). It seems that such different
RT patterns by region might decrease the group differences in RT to some degree, thus
making no signiﬁcance.

Although the two groups were not different each other in their reading times at the

critical regions, the reading proﬁles of the Korean EFL group showed Spillover effects

41

after the critical region. In order to explore if these spillover effects are unique to the
Korean EFL group, one-way AN OVA was conducted with the three regions within the
structure: Word 7, Word 8, and the sum of Word 7 & Word 8 in the subject wh-extraction.
The results showed a signiﬁcant effect of learning environment on Word 8 (F 1, 23 =

14.838,p < .001) and on the sum ofWord 7 & Word8 (F 1, 23 =13.304,p = .001).

RT analyses on grammatical wh-extraction: nonfinite clausal condition

While Principle-based Parsing analysis predicts subject wh-extraction from
nonﬁnite clauses to be the most difﬁcult structure for parsing, the subject-object
asymmetry in the nonﬁnite clausal condition may occur aﬁer the parser encounters the
object gap and the consequent word in the lower clause in each case (i.e., invite in Figure
6, hire in Figure 7), from which the parser in both cases must revise its previous analysis
(subject 9object trace, as in 12c and 13c), as shown in Table 6. Then, in processing the
following word in subject extraction (i.e., his, the determiner of the object DP), the parser
must rapidly go back to its previous analysis because the object gap is already ﬁlled with

another DP (as in 12d), which eventually adds additional burden to the parser. On the

 

 

Table 6
Procedural description of parsing at the critical region: extraction ﬁ'om nonﬁnite clause
W7 W8
Reanalysis from the previous trial .
Reanalys1s

Subject gap (Subject of the embedded clause) .
9 Object of the embedded clause 9 Subject Of the embedded clause

Reanal s1s from the rev1ous trial Pars1ng Termm_at10n

Object gap (Subject of the embedded clause)

9 Object of the embedded clause -) Conﬁrm the prevrous analysis

42

woman? Ema—mac: 88m 838.55-“? you .33. 05 do £3 :82 6 SEER

28:58 ﬁaemnxl Em 5262...... 4mm goof-1 nmm 88301.1

 

 

 

 

 

 

 

 

«ENE 25 ANNEQ ME 8.55 8 8638 Ace 2: En Sb:
o
com
\. 1111111111 111... . , o9V
E cow
1‘ 4 com

 

\ coo—

 

\ oom~

 

Go:

43

woman—o BEES: 88m. nouombxoéz, Soﬁa 05 no ﬂux :82 N 6.33%

26558 Ewemti HE agent... qmm goof-1 mm 68:30.61

38.: BS cox 6S S\ 22‘ 2 88.68 Swarm“: of 36 2b:

 

 

oc—

 

com

 

cam

 

 

.817. 1/// ;. > - _ I- M...
«Q / /\ /\ \\
\ // \/ \\

 

\ 2 U x can
R Q can

com

 

coo ~

Table 7

Means of R Ts at the critical region in the nonﬁnite clausal condition

 

 

Group Raw Mean (SD) Log-transformed Mean (SD) N

1 (Chinese ESL) 601.68 (160.46) 6.36 (.28) 15

51337861 2 (Korean ESL) 777.32 (291.58) 6.59 (.39) 15
3 (Korean EFL) 823.62 (367.20) 6.62 (.47) 15

1 559.86 (204.37) 6.27 (.27) 15

033,670! 2 876.79 (467.07) 6.65 (.51) 15
3 774.77 (444.14) 6.52 (.53) 15

1 564.46 (170.24) 6.30 (.27) 15

Stiybjgct 2 612.11 (221.07) 6.36 (.35) 15
3 662.47 (222.89) 6.44 (.35) 15

1 475.50 (96.84) 6.15 (.19) 15

035;“ 2 640.26 (265.79) 6.40 (.36) 15
3 799.14 (275.87) 6.54 (.40) 15

 

other hand, in case object extraction, the parser can conﬁrm its analysis after the next

word (i.e., for in Figure 7) and terminates its parsing. For this reason, Word7 and Word 8
were placed as a critical region for further analyses, following Juffs (2005). The RTs for
these regions are shown in Table 7. The statistical procedures were the same as those

used in the analysis of wh-extraction from ﬁnite clauses.

a) The effect of L1 .' Chinese and Korean ESL RT s (Nonﬁnite)
While the Chinese ESL group showed faster RTs in almost all regions both in subject and
object extraction, there was a main effect of L1 in the critical region, F (1, 28) = 5.322, p
< .05. There was also a main effect of region, F (1, 28) = 10.1 12, p < .01, in which

Chinese ESL and Korean ESL learners spent longer time at the object gap in both

45

structures; differently from the prediction by Principle-based Parsing, learners in both
groups did not Show much difﬁculty when they encountered the object DP in ﬁgure 6,
and thus their reading proﬁles at this region became similar to those in the object wh-
extraction. Inﬂuenced by this, there was no main effect of structure F (l , 28) = .574, p
= .455, no signiﬁcant structure by region interaction, F (1, 28) = .013, p = .574, no
signiﬁcant L1 by structure interaction, F (1, 28) = 3.334, p = .079, and no signiﬁcant L1
by region interaction, F (1, 28) = 1.984, p = .170.

In addition to the L1 effect in the critical region, the Korean ESL learners also
spent much longer time than the Chinese ESL learners at the object gap of the matrix
clause (i.e., expect in Figure 6 & 7) and at the subject trace (i.e., to in Figure 6 &7) in the
embedded clause in both extraction cases, which were signiﬁcantly different from one

another, F(1, 28) =11.722,p < .01.

b) The eﬂect of learning environment: Korean ESL and EFL R Ts (Nonﬁnite)
The analysis of reading proﬁles of the two Korean groups (ESL and EFL) in the nonﬁnite
clausal conditions showed no signiﬁcant effect of learning environment F (1', 28) = .058,
p = .811., no signiﬁcant learning environment by structure interaction, F (l, 28) = .189, p
= .667, and no signiﬁcant interaction between learning environment and region, F (1 , 28)
= .206, p = .206. These results demonstrate that the reading proﬁles of the two L1 Korean
groups were fairly comparable to one another in both structures and both regions.

There was also no main effect of structure, F (1, 28) = .165, p = 688, and no

signiﬁcant interaction between structure and region. Although the parsing prediction was
that learners would spend more time on subject wh-extraction, especially at the place

where the parser must perform its third reanalysis (i.e., Word 8 in Figure 6, a constituent

46

of an object DP in the embedded clause; see Table 6 for procedural description), the
appearance of an object DP did not make learners Spend longer time at all.

Furthermore, a similar phenomenon was found at the subject trace in the nonﬁnite
clauses (i.e., to, Word 6, in Figures 6 and 7). While it was assumed that learners might
experience processing difficulties at this point due to TRC, it was much faster than even
the object gap in the main clause (i.e., expect, the main verb where the initial posit

occurs).

General Discussion

The main goal of this replication study was to explore the role of syntactically
distinct Ll features and L2 learning environment on highly-proﬁcient adult leamers’
processing of wh-constructions in English. Given differentdegrees of parsing complexity
during the incremental wh-ﬁller/gap procedure, the underlying hypothesis of the present
study was that late learners would experience relatively greater parsing difﬁculty as
compared to native English speakers when they perform wh-ﬁller/gap analysis through
incremental processing.

Conceming the ﬁrst research question that speciﬁcally asked to what extent late
learners were sensitive to Subjacency, conﬁrming learners’ having relevant L2 syntactic
knowledge of wh—movement was necessary in this study to exclusively observe the
effects of learner processability on L2 performance by late learners. With this in mind, all
L2 learner groups were fairly sensitive in detecting Subjacency violations in English, in
which especially the L1 Chinese ESL learners were not statistically different from the

native English Speakers, who achieved 89 percent accuracy. The L1 Korean ESL and

47

. EFL learners were reliably different from the control; however, they rejected
ungrammatical wh-constructions at a rate much above the chance level (73 percent and .
66 percent, respectively). Consequently, although wh-movement and the consequent
Subjacency constraints are not operative in their Lls, late learners both in the L1 Chinese
and Korean groups demonstrated that they understood syntactic rules of wh—movement in
English to a great extent. These results provide an important implication when
interpreting individual learners’ performance (both accuracy and RT data) on
grammatical wh-extraction. That is, even if they incorrectly judge the grammatical wh-
constructions as ungrammatical, their wrong judgments might be due to their processing
problems, rather than their lack of L2 competence on English wh-structures.

Despite this evidence, however, the accuracy scores and reading proﬁles of the
experimental groups on the four types of grammatical wh-extraction revealed that, as
compared to the control group, learners had considerably heavier processing difﬁculty in
reading and comprehending these ﬁller-gap related constructions (i.e., signiﬁcantly lower
accuracy scores and longer RTs). Moreover, they represented different levels of
processing difﬁculty depending on the type of gap argument (i.e., more difﬁculty in
subject gap than in object gap), as hypothesized through the second research question that
was asked to investigate whether late learners showed predictable processing difficulty
based on Principle based Parsing.

However, such different degrees of parsing complication indicated by learners
seem to only partially correspond with the predictions made by Principle-based Parsing
analysis. That is, although all groups showed subject-object asymmetry within the same

clausal condition (i.e., more parsing problems in subject wh-extraction than in object wh-

48

extraction, following Principle-based Parsing), the overall results did not obey Principle-
based Parsing accounts. In particular, all participants, including the control, showed the
greatest parsing problems in subject wh-extraction from ﬁnite clauses. While this result is
consistent with the results shown by Juffs (2005), it is different from the Principle-based
Parsing account that assumes subject wh-extraction from nonfmite clauses to be the most
difficult structure to process. For such discrepancies between learner performance and the
prediction of Principle based Parsing account, Juffs’s (2005) explanation that participants
might have experienced garden path effects when parsing two adjacent ﬁnite verbs in a
word-by-word moving window task may be a better explanation.

Furthermore, learners’ accuracy scores on object wh—extraction ﬁom ﬁnite clauses
were comparable (Korean ESL) or slightly lower (Chinese ESL, Korean EFL) than those
on subject wh-extraction from nonﬁnite clauses, and their reading proﬁles were not
signiﬁcantly different from one another. Given these results, as Juffs (2005) also
mentioned, English learners’ processing difﬁculty might be more closely related to the
ﬁniteness of the verbs (or tense morphology) over the parsing complexities made from
incremental processing.

In responding to the third research question about the effect of ﬁrst language, the
lack of wh-movement and consequent wh-ﬁller/ gap procedure in L1 processing was
apparently disadvantageous for both the Chinese ESL and the Korean ESL group as
compared to the control group, in that all learner groups were signiﬁcantly less accurate
on grammatical wh-extraction and their reading proﬁles were signiﬁcantly slower than

the control.

49

In addition, the study conﬁrmed that the different headedness of VP between L1
and L2 put additional burden on Korean ESL learners’ processing and understanding of
wh-constructions in English. Although the proﬁciency-matched Chinese ESL and Korean
ESL learners were not statistically different from one another in their grammaticality
judgments on the given structures, Korean EFL learners’ reading patterns were
Signiﬁcantly slower than Chinese ESL learners not only at the critical regions, but also at
every verb in the sentence where gap reanalysis is required. Such reading proﬁles as
those of the Korean group are consistent with those of the Japanese ESL learners in J uffs
(2005) and Fender (2003, 2008).

The study also found an effect of learning environment (i.e., research question 4).
While their proﬁciency and age factors (both mean age at the time of the test and at ﬁrst
exposure to nonformal L2 learning environment) were matched with the Korean ESL
learners, Korean EFL learners displayed much heavier processing difficulty compared to
Korean ESL learners, especially when they processed subject and object wh-extraction in
the ﬁnite clausal condition; the Korean EFL group’s mean accuracy scores were even
below the chance level on subject extraction (32 percent), and close to the chance level
on object extraction (52 percent), which were considerably poorer than those of the
Korean ESL learners (59 percent and 68 percent, respectively). As hypothesized, this
might be due to Korean EFL learners’ limited amount of exposure to L2 in their daily life.
As shown in Table 1, the weekly amount of English use for Korean EFL learners (total
average 17.30 hours per week) was signiﬁcantly less than that of Korean ESL learners

(total average 44.87 hours per week).

50

The reading proﬁles of the two groups (Korean ESL and EFL) were not much
different ﬁom one another. However, when considering the extremely low accuracy
scores of the Korean EFL group in the ﬁnite clausal condition and the consequent
elimination of such a large amount of error data in the RT data, it is questionable if the
represented RTs of the Korean EFL learners in these structures can purely reﬂect their
general processing patterns or ability.

In addition, the Korean EFL learners showed some evidence that they did not
consistently employ incremental processing strategies for ﬁlling wh-gaps as soon as
~ possible (i.e., GTA). In examining their reading patterns, Korean EFL learners showed
much ﬂatter or even faster reading proﬁles than expected, speciﬁcally at the regions
where the parser is assumed to meet heavier processing load from wh-gap reanalysis (i.e.,
Word 6 position in Figures 5, 6, and 7). These ﬁndings are Similar to the claim made by
Marinis et al (2005) that L2 learners may not fully bring syntactic information into play
in real-time processing. If this were the case for the EFL learners, it might be because of
their signiﬁcantly less L2 processing experience speciﬁcally on English wh-ﬁller/gap
structures, as compared to the ESL learners. Moreover, the targeted test items in the
current study were all interrogative sentences containing long-distance wh-movement,
which might not be likely to be used often especially in the EFL setting. Consequently,
those learners in the EFL condition would have had more processing problems, compared
to learners in the ESL condition, from their inconsistent (or unfamiliar) implementation
of relevant L2 processing strategies and/or their irrelevant use of L1 processing strategies
(i.e., delay of gap-ﬁlling until the verb is processed at the end of the sentence; SOV) in

real time processing.

51

Limitations and Future Research

There are a few limitations in the present study. First, the present study only
investigated English learners whose L1 8 are wh-in-situ. Consequently, the analysis was
not fully able to respond to the question of to what extent the wh-in-situ characteristic of
Lls can be more disadvantageous for those L2 learners as compared to other L2 learners
having wh-movement operation in their L1 3 (e. g., the L1 Spanish ESL group in Juffs
(2005)). Secondly, the current study was not able to examine the interaction of L1 and
learning environment because Chinese EFL learners were not included in the research
design. Therefore, in future research, it would be beneﬁcial to include ESL and EFL
groups from various Ll backgrounds (including both wh-movement and wh-in-situ L1 5)
in order to have more insight into and reliable generalization about the role of ﬁrst
language and learning environment on L2 processing.

In addition, the question of why the different headedness of VP between L1 and
L2 makes integrating arguments into phrases more difficult for L2 learners needs to be
further investigated. Both in Juffs (2005) and the present study, the Li Japanese and L1
Korean English learners, respectively, showed, through their reading proﬁles, greater
difficulty whenever they processed verbs. In English, overt case marking only appears in
several pronouns (e.g., his, him, or her) and wh-words are generally ambiguous in terms
of their case representation until the parser ﬁnd the gap. In the case of Korean, on the
other hand, although case marking can be optional in colloquial speech, wh-phrases
generally take overt case morphology in both in writing and speaking, as in (6). One
speculation is that such cross-linguistic differences in overtness of case morphology can

make integration of arguments into verb phrases more difﬁcult for Korean English

52

learners, because they must put additional effort to assign unmarked cases to each
argument. With this in mind, in future research, it would also bea useful to further study
the role of cross-linguistically distinct features of cases and case morphology in L2

sentence processing.

53

APPENDIX A

Participant Consent Form A
(For Native English Speakers)

You are being asked to participate in a research study about second language sentence processing.
This consent form is provided to give you information about the study and to explain risks and
beneﬁts of participation. Also it should be noted that participation is voluntary, and you can
withdraw or refuse to answer any particular questions in any tasks. Please, feel ﬂee to ask the
researchers any questions at any time. You may use the contact information below.

Research study title: “Experimental second language processing of wh-constructions in English:
The effects of ﬁrst language and learning environment”

Department and Institution: Department of Linguistics and Germanic, Slavic, Asian and
African Languages, Michigan State University.

Contact Information: If you have any questions or concerns, you may contact with Dr. Spinner
via email at spinneMuedu or by phone at 517-353-0754. You can also make
contact with Se Hoon Jung via email at iungsMuedu or by phone at 517-575-
M. Contact via mail is also available and you can mail at A-714 Wells Hall,
Michigan State University, East Lansing, MI 48824

Pumse of Research: The study aims at investigating to what extent English learners whose
native languages are syntactically different from English process English sentences
similarly or differently from native speakers.

Elig‘bilig: You are able to participate in this study if you speak English as your ﬁrst language.

What you will do: The study will involve 1 session, which will take approximately 40-50
minutes. During the session, you will be asked to read sentences (one sentence at a
time) on a computer screen and to judge if each one of those are grammatically
acceptable.

Commnsation: By participating in this study, you will receive $10 (USD) cash payment.

Potential beneﬁts: You will not directly beneﬁt from your participation in this study. However,
your responses will be used to investigate to what extent foreign and second language
learners process the target language differently from the native speakers of that
language. The information gained from the study could have implications for second
language acquisition theories.

Potential risks: There are no risks or discomforts expected as a result of your participation. Your
participation in the study will not affect your enrollment in the university, and your
performance during the tasks will not affect your grades in your classes. The data will
be used only for research purposes.

54

Confidentiality: You will not be required to provide your name or any other personally
identiﬁable information. Your performance in this study will be kept conﬁdential and
your information will be safely secured by the researchers. All the data will be
accessible only to the assigned investigators and will be kept securely. No personal
information such as your name will be given in public. Instead, you will be identiﬁed
only by a number throughout the data collection process as well as any further
publication.

If you have any questions that are not answered on this consent form, please feel free to ask at
any time. If you have any questions or concems about your role and rights as a research
participant, would like to obtain information or offer input, or would like to register a complaint
about this research study, you may contact, anonymously if you wish, Michigan State University
Human Research Protection Program at 517-355-2180, FAX 517-432-4503, or e-mail
irb@msu.edu, or regular mail at: 202 Olds Hall, MSU, East Lansing, MI 48824.

Thank you for your participation.

Your signature below indicates that you have read this consent form, and that you volunteer to
participate in this study.

 

 

 

Participant’s signature Printed name liafeTnm/dd/yy)

55

P_articipant Consent Form B
(For Learners of English)

You are being asked to participate in a research study about second language sentence processing.
This consent form is provided to give you information about the study and to explain risks and
beneﬁts of participation. Also it should be noted that participation is voluntary, and you can
withdraw or refuse to answer any particular questions in any tasks. Please, feel free to ask the
researchers any questions at any time. You may use the contact information below.

Research study title: “Experimental second language processing of wh-constructions in English:
The effects of ﬁrst language and learning environment”

Department and Institution: Department of Linguistics and Germanic, Slavic, Asian and
African Languages, Michigan State University.

Contact Information: If you have any questions or concerns, you may contact with Dr. Spinner
via email at spinneﬂmuedu or by phone at 51 7-353-0754. You can also make
contact with Se Hoon Jung via email at (10¢st edu or by phone at 51 7-575-
5885 (Lansing) and _t_1_t 01 7-555-6838(Sem_a_l). Contact via mail is also available and
you can mail at A-714 Wells Hall, Michigan State University, East Lansing, MI
48824. It is recommended that contact via email for Dr. Spinner or via email/phone
will be the fastest way to talk to the investigators.

Pugppse of Research: The study aims at investigating to what extent English learners whose
native languages are syntactically different from English process English sentences
similarly or differently from native Speakers.

Elig’biligy: You are eligible to participate in this study if
1) you speak Korean or Chinese as your ﬁrst language and English as your second
language, AND
2) you did not learn English in communicative contexts before 16 years old.

What you will do: The study will involve 1 session, which will take approximately 90 minutes.
During the session, you will be asked
1) to take the grammar and vocabulary sections of the Michigan Test,
2) ﬁll out a short questionnaire about your language learning history, and
3) to read sentences (one sentence at a time) on a computer screen and to judge
if each one of those are grammatically acceptable.

Commnsation: By participating in this study, you will receive 15 dollars (US) cash payment or
the equivalent in Korean Won (KWC).

Potential beneﬁts: You will not directly beneﬁt ﬁom your participation in this study. However,
for non-native English speakers, you will be able to practice reading in English. In
addition, your responses will be used to investigate how second language learners
process the target language online. The information gained from the study could have
implications for second language acquisition theories.

Potential risks: There are no risks or discomforts expected as a result of your participation. Your
participation in the study will not affect your enrollment in the university, and your

56

performance during the tasks will not affect your grades in your classes. The data will
be used only for research purposes.

Conﬁdentiality: You will not be required to provide your name or any other personally
identiﬁable information. Your performance in this study will be kept conﬁdential and
your information will be safely secured by the researchers. All the data will be
accessible only to the assigned investigators and will be kept securely. No personal
information such as your name will be given in public. Instead, you will be identiﬁed
only by a number throughout the data collection process as well as any further
publication.

If you have any questions that are not answered on this consent form, please feel free to ask at
any time. If you have any questions or concerns about your role and rights as a research
participant, would like to obtain information or offer input, or would like to register a complaint
about this research study, you may contact, anonymously if you wish, Michigan State University
Human Research Protection Program at 517-355-2180, FAX 517—432-4503, or e-mail
irb@msu.edu, or regular mail at: 202 Olds Hall, MSU, East Lansing, MI 48824.

Thank you for your participation.

Your signature below indicates that you have read this consent form, and that you volunteer to
participate in this study.

 

 

 

Participant’s signature Printed name Date (mm/dd/yy)

57

Appendix B
Background Questionnaire
X All information you provide will be kept conﬁdential and will be reported anonymously.

Section 1. Basic Information

 

1. Subject No: (Assigned by the researcher)
2. Gender (Check one): 1:1 Male 1:1 Female
3. Age: years old.

 

4. What is your native language?

 

5. (For MSU students)
How long have you been studying in the United States? 1:1 Yes 1:1 No
For years months.

6. Academic Status
(Check one, or two if needed; for example, if you are currently taking classes at ELC
while you are a senior at college in your home country check both at ELC and Senior)

[:1 ESL student El Freshman El Sophomore 1:1 Junior
1:1 Senior 1:1 Graduate 1:1 Other (please
specify)

 

7. Have you learned any foreign languages other than English?
924 If yes, please specify all information 1) including which language, 2) length of study

 

 

 

 

 

 

 

 

I have learned for month(s)
I have learned for month(s)
8. Have you ever studied or lived ABROAD? 1:1 Yes 1:1 No
it If yes,
1) Where? , 2) How long? , 3) Main Purpose:
1) Where? , 2) How long? , 3) Main Purpose:

 

 

*** [Example] 1) Where? Australia 2) How long? 6 mth 3) Purpose: Univ. exchange program
1) Where? Canada 2) How long? 2 years 3) Purpose: Degree Program

 

 

58

Section 2. English learning experience

9. When did you start learning English?
I was (about) years old when I started learning English.

10. How many years have you studied English?
C] Less than 1 year 1:1 1 year 1:1 2 years 13 3 years 1:1 4 years
1:1 5 years E] 6 years 1:1 7 years 1:1 8 years D 9 years
1:1 More than 9 years (for about years)

11. How old were you when you were ﬁrst exposed to English communication
environments? What was the context of that situation? (For example, taking
conversation classes, study abroad program, or ESL classes would be such cases).

I was years old. The context was

 

12. How much time do you usually spend using English from outside of the classroom?

 

 

 

 

 

 

 

 

Lan ua e . . . Use of time
Skgillsg Description (Weekly)
Newspaper (including online or intemet news),
Reading magazine, book (including textbooks), academic (about
journal, and any other types of reading activities in hours) per week
English
Homework assignment, essay, diary, English (about
Writing chatting, and any other types of writing activities in hours) TvTe—ek
English pe
Listening Pop music (in English), lecture, TV, radio, and any (about
other types of listening activities in English hours) per week
Conversing with native speakers of English, in- (about
Speaking class discussion in English, and any other types of hours) pe-r—weEk
speaking activities

 

Again your privacy will be secured. Thank you for your participation.

59

 

Appendix C
Experimental sentences for moving window procedures

X These test items were obtained from Dr. Alan Juffs at University of Pittsburgh,
and revised under his permission.
X Total 27 Structures, 162 Sentences

# N-WHM-CompNP l 1 What did Tom hear the story that Ann cooked for Bill?

# N-WHM-CompNP 2 1 Who did Bill believe the story that Ann hated at work?

# N-WHM-CompNP 3 1 Who did Tom believe the claim that Ann painted at home?

# N-WHM-CompNP 4 1 What did Bill believe the story that Ann studied at school?

# N-WI-IM-CompNP 5 1 What did Ann hear the news that Jane received from Tom?

# N-WHM-CompNP 6 1 Who did Sam hear the news that Jane called last night?

# N-WI-IM-AdjunctC 7 2 Who did John call his mom after he kissed last week?

# N-WHM-AdjunctC 8 2 What did Ann leave the kitchen after she cooked last night?

# N-WHM-AdjunctC 9 2 What did John begin his homework after he studied last night?
# N-WHM-AdjunctC 10 2 Who did Tom confuse his wife when he invited last night?

# N—WHM-AdjunctC ll 2 What did Tom break his glasses when he played last night?

# N-WHM-AdjunctC l2 2 What did John sell his bicycle when he needed last week?

# N-WHM-RC 13 3 What did Sam see the man who wanted from Jane?

# N-Wl-IM-RC l4 3 What did Tom meet the woman who rented last week?

# N-WHM-RC 15 3 What did Bill buy the bicycle which needed last year?

# N-WHM-RC l6 3 What did Ann lend the student who studied at school?

# N-WHM-RC l7 3 Who did Tom see the woman who collected in hospital?

# N-WHM-RC 18 3 Who did Jane like the man who carried for Tom?

# N-WHM-SubjFinTHAT 19 4 Who did Ann suggest that liked the manager at the office?

# N-WHM-SubjFinTHAT 20 4 What did Sam report that passed his bicycle in the street?

60

# N-WI-[M-SubjFinTHAT 21 4 Who did Ann know that studied the textbook in the library?

# N-WHM-SubjFinTHAT 22 4 Who did John know that watched the patient in the hospital?

# N-WHM-SubjFinTHAT 23 4 Who did Jane report that saved the driver from the crash?

# N-WHM-SubjFinTHAT 24 4 Who did Bill know that visited the woman at the hotel?

# Y-WHM-ShortObjExt 25 5 What did Mary need when she planned a party last week?

# Y-WI-IM-ShortObjExt 26 5 What did Tom lock after he packed his bag last night?

# Y-WHM-ShortObjExt 27 5 What did John earn when he painted the house for Mary?

# Y-WHM-ShortObj Ext 28 5 Who did Bill meet after he parked his car at school?

# Y-WHM-ShortObjExt 29 5 Who did Ann call when she stayed with Tom last week?

# Y-WHM-ShortObjExt 30 5 Who did Tom ﬁght when he played soccer last night?

# Y-WHM-FinSubj 31 6 Who did the woman suggest liked the manager at the ofﬁce?

# Y-WHM-FinSubj 32 6 What did the man report passed his bicycle in the street?

# Y-WI-IM-FinSubj 33 6 Who did the teacher know studied the textbook in the library?

# Y-WHM-FinSubj 34 6 Who did the nurSe know watched the patient in the hospital?

# Y-WHM-FinSubj 35 6 Who did the sailor report saved the woman from the ship?

# Y-WHM-FinSubj 36 6 Who did the police know visited the woman at the hotel?

# Y—WHM-FinObjWithoutThat 37 7 Who did the woman suggest the manager liked at the ofﬁce?
# Y-WHM-FinObjWithoutThat 38 7 What did the driver report his truck passed in the street?

# Y-WHM-FinObjWithoutThat 39 7 What did the teacher know the student studied in the library?
# Y-WHM-FinObj WithoutThat 40 7 Who did the nurse know the police arrested at the hospital?
# Y-WHM-FinObjWithoutThat 41 7 Who did the woman report the sailors saved from the ship?
# Y-WHM-FinObjWithoutThat 42 7 Who did the police know the thief visited at the hotel?

# Y-WI-IM-FinObjWithThat 43 8 What did Tom know that his friend studied in the library?

# Y-WHM-FinObjWithThat 44 8 Who did John think that his sister invited to the party?

# Y-WHM-FinObjWithThat 45 8 Who did Bill know that the police arrested at the airport?

# Y-WHM-FinObjWithThat 46 8 What did Jane suggest that the student studied in the library?

61

# Y—WHM-FinObj WithThat 47 8 What did John report that the teacher corrected in the exam?
# Y-WHM-FinObjWithThat 48 8 What did Tom know that his mother cooked for the party?
# Y-WHM-NonFSubj 49 9 Who did the students want to teach the course last year?

# Y-WHM-NonFSubj 50 9 Who did the teacher want to learn the rules last week?

# Y-WHM-NonFSubj 51 9 Who did the boy expect to invite his family last night?

# Y-WHM-NonFSubj 52 9 Who did the manager expect to meet the customer last month?
# Y-WHM-NonFSubj 53 9 Who did the woman want to arrest the thief last week?

# Y-WHM-NonFSubj 54 9 Who did the actor report to like his movie very much?

# Y-WHM-NonF Obj 55 10 What did the students want to learn in the course this term?

# Y-WI-IM-NonF Obj 56 10 Who did the police want to arrest at the airport last night?

# Y-WHM-NonFObj 57 10 Who did the manager expect to hire for the job last week?

# Y-WHM-NonFObj 58 10 Who did the teacher expect to meet at the school last week?

# Y-WHM-NonF Obj 59 10 What did the police report to know» about the crime last night?
# Y-WHM-NonF Obj 60 10 Who did the student report to like in the movie last week?

# Y-GP 61 11 When the manager served the wine splashed his jacket and tie.

# Y-GP 62 11 Before the student guessed the answer appeared on the next page.

# Y-GP 63 llAﬁer the students played the game stayed very popular at school.

# Y-GP 64 ll Aﬁer the children cleaned the house looked very neat and tidy.

# Y-GP 65 l 1 Before her sister parked the car needed some gas and oil.

# Y-GP 66 l 1 When the teacher helped the learned the rules last week.

# Y-NGP 67 12 After the president talked the guests offered him some red wine.

# Y-NGP 68 12 Before the man shouted the thief climbed through the small window.

# Y-NGP 69 12 After the woman died her husband married a woman from England.

# Y-NGP 70 12 When the student arrived the professor asked her about her trip.

# Y-NGP 71 12 When the actor appeared the crowd pushed towards his big car.

# Y-NGP 72 12 Before the teacher agreed the students played soccer in the yard.

62

# Y-GP 73 13 The doctor knew the nurse liked the man from England.

# Y—GP 74 13 The student heard his teacher hated the book about cars.

# Y-GP 75 13 The teacher knew the student studied the lesson on birds.

# Y-GP 76 13 The manager reported the woman earned the money last year.

# Y—GP 77 13 The sailor heard his friend the ship last year.

# Y-GP 78 13 The police reported the thief visited the house last week.

# Y-RRC 79 14 The famous man known by every student was a very nice person.

# Y-RRC 80 14 The large birds eaten in the garden could not see the cat.

# Y-RRC 81 14 The black horse chosen at the stadium won ﬁve prizes this year.

# Y-RRC 82 14 The bad boys seen in the aﬁemoon were playing in the park.

# Y-RRC 83 14 The small child ﬂown from his home was very sick last night.

# Y-RRC 84 14 The young teacher driven to the school started to teach this week.

# Y-RRC 85 15 The famous man known almost every place was a very nice person.
# Y-RRC 86 15 The large birds eaten almost every week could not see the cat.

# Y-RRC 87 15 The black horse chosen nearly every week won ﬁve prizes this year.
# Y-RRC 88 15 The bad boys seen almost every day were playing in the park.

# Y-RRC 89 15 The small child ﬂown almost every day was very sick last night.

# Y-RRC 90 15 The young teacher driven nearly every day started to work this week.
# Y-RRC 91 16 The famous man invited by every student was a very nice person.

# Y-RRC 92 16 The large birds killed in the garden could not see the cat.

# Y-RRC 93 16 The black horse noticed at the stadium won ﬁve prizes this year.

# Y-RRC 94 16 The bad boys criticized during the moming were playing in the park.
# Y—RRC 95 16 The small child moved from his home was very sick last night.

# Y-RRC 96 16 The young teacher hired at the school started to teach this week.

# Y-RRC 97 17 The famous man invited almost every time was a very nice person.
# Y-RRC 98 17 The large birds killed almost every week could not see the cat.

63

# Y-RRC 99 17 The black horse noticed nearly every time won ﬁve prizes this year.

# Y-RRC 100 17 The bad boys criticized almost every day were playing in the park.

# Y-RRC 101 17 The small child moved almost every day was very sick last night.

# Y-RRC 102 17 The young teacher hired nearly last year started to teach this week.

# Y-RRC 103 18 The famous man painted by every student was a very nice person.

# Y-RRC 104 18 The large birds attacked in the garden could not see the cat.

# Y-RRC 105 18 The black horse raced at the stadium won ﬁve prizes this year.

# Y-RRC 106 18 The bad boys watched during the aftemoon were playing in the park.

# Y-RRC 107 18 The small child visited at his home was very sick last night.

# Y-RRC 108 18 The young teacher helped at the school started to teach this week.

# Y-RRC 109 19 The famous man painted nearly every month was a very nice person.

# Y-RRC 110 19 The large birds attacked almost every week could not hear the cat.

# Y-RRC 111 19 The black horse raced nearly every month won ﬁve prizes this year.

# Y-RRC 112 19 The bad boys watched almost every day were playing in the park.

# Y-RRC 113 19 The small child visited almost every day was very sick last night.

# Y-RRC 114 19 The young teacher helped almost every day started to work this week.

# N-Overpass 1 15 20 Mary knew that the book was disappeared from the shelf yesterday.

# N-Overpass 1 16 20 Tom said that an accident was happened in the street yesterday.

# N-Overpass 117 20 Mary knew that the guests were arrived late at the party.

# N-Overpass l 18 20 John said that the apples were fallen because of the wind.

# N-Overpass 1 19 20 Tom knew that the dog was died because it was old.

# N-Overpass 120 20 Jane Said that the problem was existed before the bad storm.

# Y-UNACCUSATIVE 121 21 Tom knew that the book disappeared from the shelf last week.
# Y-UNACCUSATIVE 122 21 Jane said that the accident happened suddenly in the dark street.
# Y-UNACCUSATIVE 123 21 Mary knew that the guests arrived late at the party yesterday.
# Y-UNACCUSATIVE 124 21 John said that the apples fell because of the strong wind.

64

# Y-UNACCUSATIVE 125 21 Ann knew that the dog died because it was very old.

# Y-UNACCUSATIVE 126 21 Sam said that the problem existed because of the bad storm.

# Y-TRANSITIVE 127 22 In the summer the student changed his work for three months.

# Y-TRANSITIVE 128 22 Three months ago the owners closed the factory near the town.

# Y—TRANSITIVE 129 22 In the morning the policeman started the search for the thief.

# Y-TRANSITIVE 130 22 In the morning the teacher stopped the computer for a break.

# Y-TRANSITIVE 131 22 In the winter the manager reduced the pollution from the factory.

# Y-TRANSITIVE 132 22 In the summer the farmers grew the tomatoes in the ﬁelds.

# Y-OPTRAN-INTRAN 133 23 After several weeks his work changed at the new store.

# Y-OPTRAN-INTRAN 134 23 After several months the factory closed for the summer vacation.
# Y-OPTRAN-[NTRAN 135 23 After several hours the search started in a different place.

# Y-OPTRAN-INTRAN 136 23 After several minutes the computer stopped due to a problem.
# N-OPTRAN-TNTRAN 137 23 In some cities the pollution reduced due to new laws.

# Y-OPTRAN-INTRAN 138 23 In the north the tomatoes grow best in the sun.

# Y-FILL-DechompNP 139 24 Ann heard the news that Tom needed a car last week.

# Y-FILL-DechompNP 140 24 Jane believed the story that Bill cooked a chicken last night.

# Y-FILL-DechompNP 141 24 Tom believed the story that Ann borrowed a bicycle last week.
# Y-FILL-DechompNP 142 24 Ann heard the news that Jane wanted a present from Tom.

# Y-FILL-DechompNP 143 24 John believed the story that Ann painted the kitchen last week.
# Y-FILL-DechompNP 144 24 Tom heard the news that Jane called the manager at home.

# Y-FlLL-DeclRC 145 25 Sam married the woman who solved the problem for him.

# Y-FILL-DeclRC 146 25 Tom called the woman who rented the house last week.

# Y-FILL-DeclRC 147 25 Bill repaired the bicycle which needed new tires last night.

# Y-FILL-DeclRC 148 25 Ann helped the student who studied the Indians last year.

# Y-FILL-DeclRC 149 25 Tom invited the woman who visited the patient in hospital.

# Y-FILL-DeclRC 150 25 Jane liked the man who carried the bag for Ann.

65

# Y-FILL-ShortWH 151 26 When did the teacher leave?

# Y-FILL-ShortWH 152 26 What did the manager earn?

# Y-FILL-ShortWH 153 26 Who did the police arrest?

# Y-FILL-SimpDecl 154 26 The student studied the chapter on animals.

# Y-FILL-SimpDecl 155 26 The manager called the customers.

# Y-FILL-SimpDecl 156 26 The teacher helped the students.

# N-FILL-Nonse 157 27 Who did at the kitchen Mary see in the kitchen last week?
# N-FlLL-Nonse 158 27 The parents of the brought a cake in morning the crown.
# N-FILL-Nonse 159 27 Who policeman the did see sailor the called last week?
# N-FILL-Nonse 160 27 Helped teacher last week the should man in the summer.
# N-FILL-Nonse 161 27 Into the studied went the policeman woman night the in.

# N-FlLL-Nonse 162 27 Before week in Mary for Ann changed rules the into.

66

Appendix D

Instruction and practice script in the Moving Window procedure

1. Instruction

Welcome! You are invited to an experiment in which you will be reading several types of
sentences on the computer screen. Before you encounter each sentence, you will ﬁrst see
a row of dashes like this: --—- -- --- ---- -- --- ---- --—-- ----- ---- ---- -- - ---------

These dashes are covering the words in the sentence. When you press the SPACE BAR,
the ﬁrst word will appear. Then with every press of the SPACE BAR, the next word will
appear and the previous word will be changed to dashes again. In other words, you will
see only one word of the sentence at one time.

When you read each sentence, you can control your reading at your own pace. You
should read as quickly as you can, but BE SURE that you UNDERSTAND the meaning.
If you think you did not understand the sentence because you pushed the space bar TOO
FAST, then you may SLOW DOWN for the next items.

When you ﬁnish reading the last word of each sentence, press the SPACE BAR again.
The dashes will go away and you will see a question about the sentence you just read.
The question will ask you whether or not the sentence is a grammatically acceptable
sentence of English.

For your answer, press the GREEN key if you think the sentence was grammatically
CORRECT/ACCEPTABLE in English, or press the RED key if you think the sentence
was grammatically INCORRECT/UNACCETABLE in English.

When you make a judgment for each sentence, please ingnore the PUNCTUATION. For
example, for some sentences you may not see a comma where you expect it. Your
decision about the grammaticality must not be inﬂuenced by PUNCTUATION.

Also, please focus more on the structure of the sentence than the meaning. Some
sentences might be somewhat illogical in meaning. It is expected that you concentrate on
the STRUCTURE.

Try to answer as quickly and accurately as possible. If you are NOT sure of the answer,
try to pick the better answer.

You can take a rest after every 30 sentences you read. You can also have a short break
before you start each sentence. When the experiment is over, a screen will appear telling
you to stop. At that point, let the experimenter know you are ﬁnished.

If you have any questions about the procedure, ask the experimenter now.

67

2. Practice
# practice 1 -
This sentence is a practice sentence to help you get used to a moving window display.

# practice 2 -
This sentence is also for practice, and is followed by a question. // Are you following this
so far?

The next few practice items will be more like real experimental sentences.
# practice 3 — Where did Ann go for breakfast last Sunday? // Is this correct?

The sentence you just have read is, "Where did Ann go for breakfast last Sunday?"
This is a correct and grammatical sentence, asking where Ann went for breakfast last
Sunday, so you would press the GREEN KEY for YES.

# practice 4 - The man will bought a new car yesterday evening. // Is this correct?

The sentence you just have read is ..."The man will bought a new car yesterday evening."
This is not grammatical English, so you might press the RED KEY for NO.

Here are some more practice sentences. For those examples, you would not receive any
feedback. Remember that in the main experiment you will not receive any feedback about
the answer}

# practice 5 - What does Sam do on his day off every Tuesday? // Is this correct?

# practice 6 - The dog jumped over the fence into the river. // Is this correct?

# practice 7 - Who did Tom see on the bus? // Is this correct?

# practice 8 - Where did about Ann tell the story the in classroom? // Is this correct?
# practice 9 - What is the best saw Tom on his vacation in Chicago? // Is this correct?
# practice 10 -Who was horse at the news that bird met? // IS this correct?

That's all for the practice. If you have any question, ask the experimenter now.
Otherwise, you may begin the experiment.

68

Appendix E

Debrieﬁng form

Debrieﬁng
The study you have participated in is interested in how second language learners process second
language sentences, speciﬁcally sentences which contain wh—words (in English). While some
languages such as Chinese or Korean do not make any movement of wh-word in wh-questions,
some other languages such as English or Spanish must move the wh-word from its original place
to the beginning of sentence. Generally, it is well known that native English speakers try to ﬁnd
the original place of moved wh-word as soon as possible as they process the sentence
incrementally. Taking this into account, this study seeks to ﬁnd answer to the question, “to what
extent do Korean and Chinese speaking English learners process English wh-structures similarly
to or differently from native English speakers?”
You were asked to read a number of sentences that appeared on the computer screen, and you
could read only one word at a time. From this experiment, your word-level reading time was
measured. The word level reading time indicates the time you spent before you read the next
word. At the end of every sentence, you were asked to make a judgment about the correctness of
the sentence. This was to see how sensitive you are to grammatical and ungrammatical English
wh-constructions.

As compensation for my participation in the research study,

I acknowledge receipt of (10USD, 15USD, 15,000 Won) cash ﬁ'om Se Hoon Jung.

 

Your Signature

69

References

Bialystok, E. (1997). The structure of age: in search of barriers to second language
acquisition. Second Language Research, 13, 116-137.

Bley-Vroman, R., Felix. S., & G. Ioup. (1988). The accessibility of Universal Grammar
in adult language learning. Second Language Research 4(1), 1-32.

Chomsky, N. (1977). On Wh-movement. In P. Culicover, T. Wasow, & A. Akrnajiam
(Eds), Formal syntax mp. 71-132). New York: Academic Press.

Chomsky, N. (1981). Lectures on government and binding. Dordrecht: Foris.
Chomsky, N. (1995). The Minimalist Program. Cambridge, MA: MIT Press.

Dallas, A., & Kaan E. (2008). Second language processing of ﬁller-gap dependencies by
late learners. Language and Linguistics Compass, 2(3), 372-388.

Dussias, P. E., & Sagarra, N. (2007). The effect of exposure on syntactic parsing in
Spanish-English bilinguals. Bilingualism: Language and Cognition, 10(1), 101-
1 16.

Felix, S., & Weigl, W. (1991). Universal grammar in the classsroom: The effect of formal
instruction on second language acquisition. Second Language Research, 7, 162-
180.

Fender, M. (2003). English word recognition and word integration skills of native
Arabic- and J apanese-speaking learners of English as a second language. Applied
Psycholinguistics, 24, 289-315.

Fender, M. (2008). L1 effects on the emergence of ESL sentence processing skills of
Chinese and Korean ESL learners: A preliminary study. Languages in Contrast,
8(1), 47-73.

Frazier, L. (1987). Processing syntactic structures: Evidence ﬁ'om Dutch. Natural
Language and Linguistic Theory, 5, 519-559.

Frazier, L., & Clifton, C. (1989). Successive cyclicity in the grammar and the parser.
Language and Cognitive Processes, 4, 93-126.

70

Frenck-Mestre, C., & Pynte, J. (1997). Syntactic ambiguity resolution while reading in

second and native languages. Quarterly Journal of Experimental Psychology, 50A,
119—148.

Gibson, E., & Hickok, G. (1993). Sentence processing with empty categories. Language
and Cognitive Processes. 8, 147-171.

Hawkins, R. (2001). Second language syntax: A generative introduction. Oxford:
Blackwell Publishers.

Jelliffe, J. D. and Juffs, A. (2004). Processing of subject and object relative clauses by
Chinese-speaking and Spanish-speaking learners of English. Poster presented at

GASLA VII, Bloomington, Indiana, April 2004.

Johnson, J ., & Newport, E. (1989). The null subject parameter and parametric theory. In
0. Jaeggli and K. Saﬁr (eds.), The null subject parameter (pp. 1-44).
DordrechtzKluwer.

Johnson, J ., & Newport, E. (1991). Critical period effects on universal properties of
language: The status of subjacency in the acquisition of second language.
Cognition, 39, 215-258.

Jorden, P. (1991). Linguistic knowledge in second language acquisition. In L. Eubank
(Ed.), Point counterpoint (pp. 1990218). Amsterdam: John Benjamins.

Juffs, A. (1998a). Main verb vs. reduced relative clauses ambiguity resolution in second
language sentence processing. Language Learning, 48, 107-47.

Juffs, A. (1998b). Some effects of ﬁrst language argument structure and morphosyntax
on second language sentence processing. Second Language Research, 14(4), 406-
242. ‘

Juffs, A. (2005). The inﬂuence of ﬁrst language on the processing of wh-movement in
English as a second language. Second Language Research, 21(2), 121-151.

J uffs , A . (2006). Processing reduced relative versus main verb ambiguity in English as a
second language: A replication study with working memory . In R. Slabakova , S.
Montrul , & P. Prévost (Eds.), Inquiries in linguistic development in honor of
Lydia White (pp. 213 — 234). Amsterdam : Benjamins .

71

Juffs, A., & Harrington, M. (1995). Parsing effects in second language sentence
processing: subject and object asymmetries in wh-extraction. Studies in Second
Language Acquisition, 1 7, 483-516.

Juffs, A., & Harrington, M. (1996). Garden path sentences and error data in second
language sentence processing. Language Learning, 46(2), 283-326.

Just, M. A., Carpenter, P., & Woolley, J .D. (1982). Paradigms and processes and in
reading comprehension. Journal of Experimental Psychology: General, 3, 228-
238.

Li, X. (1998). Adult L2 accessibility to UG: An issue Revisited. In Flynn, S.,
Martohardjono, G. and O’Neill, W., editors, The generative stuay of second
language acquisition (pp. 89-110). Hillsdale, NJ: Erdbaum.

Long, M. (1990). Maturational constraints on language development. Studies in Second
Language Acquisition, 12, 251-285.

Marinis, T. (2003). Psycholinguistic techniques in second language acquisition research.
Second Language Research, 19(2), 144—161. ‘

Marinis, T., Roberts, L., Felser, C., & Clashen, H. (2005). Gaps in second language
processing. Studies in Second Language Acquisition, 27(1), 53-78.

Pritchett, B. (1992a) Grammatical competence and parsing performance. Chicago:
University of Chicago Press.

Pritchett, B. (1992b). Parsing with grammar: Islands, heads, and garden paths. In Goodluck,
l-l., & Rochemont, M., (Eds), Island constraints: theory, acquisition, and processing
(pp. 321-350). Dordrecht: Kluwer.

Patkowski, M. (1980). The sensitive period for the acquisition of syntax in a second
language. Language Learning, 30, 440-472.

Rohde, D. (2001). Linger (version. 2.94) [software]. Available from
http://tedlab.mit.Edu/~dr /Linger/

Schacter, J. (1989). Testing a proposed universal. In S. Gass & J. Schachter (Eds),

Linguistic perspectives on second language acquisition (pp. 73-88). Cambridge,
UK: Cambridge University Press.

72

Schachter, J. (1990). On the issue of completeness in second language acquisition.
Second Language Research, 6, 93-124.

Schachter, J ., & Yip, V. (1990). Grarnmaticality Judgments: Why does anyone object to
subject extraction? Studies in Second Language Acquisition, 12, 373-3 79.

Slabakova, R. (2003). Semantic evidence for functional categories in interlanguage
grammars. Second Language Research, 19(1), 42-75.

Stowe, L. (1986). Parsing wh-constructions: evidence for on-line gap location. Language
and Cognitive Processes, 2, 227-246.

White, L. (1992). Subjacency violations and empty categories in L2 acquisition. In H.
Goodluck & M. Rochemont (Eds), Island constraints (pp. 445-464). Dordrecht:
Kluwer.

White, L. (2003). Second language acquisition and Universal Grammar. Cambridge,
England: Cambridge University Press.

White, L., & Genesee, F. (1996). How native is near-native? The issue of ultimate
attainment in adult second language acquisition. Second Language Research, 12,
233-265.

White, L., & Juffs, A. (1998). Constraints on wh-movement in two different contexts of
normative language acquisition: Competence and processing. In Flynn, S.,
Martohardjono, G. and O’Neill, W. (Eds), The generative study of second language
acquisition (pp. 111-130). Hillsdale, NJ: Erdbaum.

Williams, J. N. (2006). Incremental interpretation in second language sentence processing.
Bilingualism: Language and Cognition, 9(1), 71.88.

Williams, J ., Moebius, P., & Kim, C. (2001). Native and non-native processing of

English wh-questions: Parsing strategies and plausibility Constraints. Applied
psycholinguistics, 22(4), 509-540.

73

MICHI AN STATE UNIVERSITY LIBRARIES

1111111111 11111111

3 1293 3063 L070