Processing Japanese subject and object relative clauses by advanced learners: learners: Comparison with native speakers by a wholewhole-sentence reading experiment Barış KAHRAMAN, Atsushi SATO, Mariko KOIDE, Mariko UNO, Miwa TAKEMURA and Hiromu SAKAI Graduate School of Education, Hiroshima University [email protected]

CBL

Center for Brain Science of Language Acquisition and Language Learning

INTRODUCTION INTRODUCTION  Subject relative clauses (SRCs) are easier to process than object relative PREVIOUS (Kanno, 2007; O’Grady et al. 2003; Aydın, 2007) PREVIOUS STUDIES STUDIES in in L2 L2 clauses (ORCs) in many languages (both in L1 and L2)  Participants were either elementary or intermediate. SRC : [The teacher who saw the student] was smoking a cigarette. ⇒ How about the advanced learners? : [The teacher who the student saw] was smoking a cigarette. ORC  Only accuracy rates of SRCs and ORCs were compared.  Recent studies from Asian languages have argued that Structural Distance ⇒ How about the time spent on processing of SRCs and ORCs? Hypothesis accounts for the processing asymmetry between SRCs and ORCs AIM AIM of of the the PRESENT PRESENT STUDY STUDY L1 ⇒Japanese: Ueno & Garnsey, 2008; Korean: Kwon et al., 2006; L2 ⇒Japanese: Kanno, 2007; Korean: O’Grady et al. 2003; Turkish: Aydın, 2007



Structural Distance Hypothesis (SDH) 

We examined whether SDH can accurately predict the processing difficulty of SRCs and ORCs for advanced learners We also compared advanced learners’ performance with those of native speakers

Participants: Task:

EXP 1: Chinese-speaking learners of Japanese (CLJ: N=21) EXP 2: Native speakers of Japanese (NSJ: N=24) Whole-sentence reading experiment

MAJOR MAJOR FINDINGS FINDINGS The number of syntactic nodes between a filler and a gap determines the processing difficulty of RCs (O’Grady, 1997; Hawkins, 1999).

 EXP 1. CLJ read SRCs faster than ORCs

⇒ SDH accounts for this result

 EXP 2. NSJ read ORCs faster than SRCs

⇒ SDH cannot explain this result ⇒ Task might have also an impact

EXPERIMENT EXPERIMENT 11 Aim:

To examine whether advanced learners of Japanese process SRCs faster than ORCs in Japanese, as predicted by SDH.

Participants:

21 advanced Chinese-speaking learners of Japanese (CLJ) from Hiroshima University 19 of them have 1st level of JLPT -Learning Japanese for 3 to 13 years (avr = 7 years; SD = 2.19) -Living in Japan for 6 months to 7 years (avr = 3ys; SD = 2.61 ) 2 × 20 targets (plausibility was tested by a norming study) + 48 fillers (passives, causatives, etc.); Latin Square Design Sensei-o mita gakusei-wa tabako-o sutte-ita.

Materials : SRC :

Teacher-ACC saw student-TOP cigarette-ACC smoke-Past PROG The student who saw the teacher was smoking a cigarette.

ORC :

Sensei-ga

mita

gakusei-wa

tabako-o

RESULTS RESULTS Data trimming:

Accuracy → 2 items & 2 participants < 65%; RT → ±2.5 SD

Overall accuracy:

85%; SRC = 83% & ORC = 88% [F1 (1,18) = 3.01, p < .09; F2 (1,17) = 1.03, p < .32]

ORCs were slightly higher than SRCs Reading times:

SRC = 8871 ms; ORC = 9961 ms [F1 (1,18) = 8.5, p < .01]

9500 RT (millisecond) 9000

[F2 (1,17) = 4.54, p < .05] This tendency does not change even if LOG transformation is performed

8500

 CLJ read SRCs faster than ORCs

8000

sutte-ita.

ORC

Teacher-NOMsaw student-TOP cigarette-ACC smoke-Past PROG The student who the teacher saw was smoking a cigarette.

Procedure :

**

10000

SRC

DISCUSSION DISCUSSION Participants were instructed to answer YES-NO questions after they have completely understood the sentences;  The results are consistent with previous studies in L2 & L1 - CLEFT questions: Gakusei-ga/wo mita-nowa sensei desu ka? L1 ⇒Japanese: Ishizuka, 2005; Miyamoto & Nakamura; Ueno & Garnsey, 2008; Korean: Kwon et al., 2006; Is it a teacher that / the student saw/ saw the student? L2 ⇒Japanese: Kanno, 2007; Korean: O’Grady et al. 2003; Turkish: Aydın, 2007 - Answer of target sentences was YES;  SDH accurately predicts the processing difficulty of RCs even for advanced learners - Sentences were presented by Linger 2.94 (by Doug Rohde); - Familiarity of lexical items was checked soon after the EXP. EXPERIMENT EXPERIMENT 22

Aim:

To examine how NSJ process sentences used in EXP 1, and to compare CLJ with NSJ.

Participants:

24 university student from Hiroshima University

Materials:

Identical to Experiment 1

Procedure: RESULTS RESULTS

Identical to Experiment 1

Data trimming: Overall accuracy:

Reading times: 4000

**

3800 RT (millisecond)

SRC = 3503 ms; ORC = 3891 ms [F1 (1,20) = 6.44.01, p < .01]

3600

[F2 (1,18) = 3.8, p < .07]

3400

 NSJ read ORCs faster than SRCs

3200

Accuracy → 1 item & 3 participants < 65%; RT → < 8000ms & ±2.5 SD 89%; SRC = 82% & ORC = 96% [F1 (1,20) = 14.04, p < .001; F2 (1,18) = 11.53, p < .001]

ORCs were higher than SRCs EXPERIMENT 1

→

CLJ read SRCs faster than ORCs

EXPERIMENT 2

→

NSJ read ORCs faster than SRCs

Why the results of EXPERIMENT 2 differed from previous studies

ORC

SRC

DISCUSSION DISCUSSION  This result is contrary to previous studies & EXP1  This result cannot be explained by SDH GENERAL DISCUSSION GENERAL DISCUSSION PRELIMINARY PRELIMINARY IMPLICATIONS IMPLICATIONS  L2 learners cannot process sentences as native speakers do (Clashen & Felser, 2006)  Not only the processing difficulty of RCs, but also processing time differs between learners and native speakers →Lack of automaticity (Segalowitz, 2003)  Chinese (L1) did not influence CLJ’s performance in Japanese (L2) (Felser et al.

ALTERNATIVE → TASK? ALTERNATIVE EXPLANATIONS EXPLANATIONS 2003; Kanno, 2007, Papadopulou & Clashen, 2003) 1. [Instruction?] There was no time limitation & Participants were asked to answer questions after understanding sentences completely → ORCs are easier to process than SRCs in Chinese (Hsiao & Gibson, 2003) → Participants might have read a sentence many times  Structural distance between the filler and the gap might not be the sole factor that determines the processing difficulty of RCs 2. [Question type?] Right answer was YES & questions were only CLEFT → Task might have also an impact on the processing of RCs → Participants might have used an unusual strategy ACKNOWLEDGEMENT: This research was supported by Japan Society for the Promotion of Science Grant-in-Aid for 3. [Lexical items?] Words were adjusted to learners (JLPT 3rd level) & we could Scientific Research (#20320060), Japanese Ministry of Education, Culture, Sports Science & Technology Grant-in-Aid for Scientific Research on Priority Areas -System Study on Higher-Order Brain Functions-(#20020020), Japan Promotion for not unify tense-aspect of RC verbs. Science & Technology (RISTEX Brain Science and Education Project). TL -MAPLL 2009 → Mixed use of tense-aspect of might have had an impact on RT

KYUSHU UNIVERSITY / FUKUOKA