Effects of Semantics and Syntax on Chinese and Spanish Native - - PowerPoint PPT Presentation

Effects of Semantics and Syntax

• n Chinese and Spanish

Native Speakers‘ Chinese Sentence Comprehension

Yu-Tzu Chang Georgetown University

syntax

Ln speakers

Introduction

2

syntax

L1 speakers

• How do people process sentences?
• L1 = L2 = Ln?
• Different L1 Ln language combinations?

(L1 CH L3 JP = L1 CH L2 EN )

Purpose

• Processing strategy in Chinese:

Spanish Vs. Chinese native speakers

• Which language processing theory?

– Shallow Structure Hypothesis (SSH; Clahsen & Felser, 2006) – Competition Model (MacWhinney 1997) – Good Enough Representation (Ferreira, 2003)

LITERATURE REVIEW

3 processing models Mandarin Differences between sentence structure

Shallow Structure Hypothesis ( Clahsen and Felser, 2006) Methods Complex sentences e.g. ambiguous & syntactic dependencies Results 1. Native speakers can process sentences with deep and complete sentence representations 2. Foreign language learners usually ignore syntactic cues and use shallow structure when comprehending sentences Problems & Questions 1. L1ers process with shallow and incomplete representation? 2. Don’t foreign language learners have opportunities to use deep and complete representation? 3. Will experiments using simple sentences as material instead of complex sentences produce different results?

5

Good Enough Hypothesis (Ferreira, 2003)

Findings

adults can process language both algorithmically and heuristically

Response to C&F

Adult native speakers do not always perform complete representation when processing language. They often miscomprehend many sentences because of the default NVN heuristics

6

Competition Model (MacWhinney，Bates，and Kligel， 1984) Purpose To investigate language users’ reliance on different linguistic cues in development , such as word order and semantic features (animacy, plausibility) Results Learners are more likely to use both semantic and syntactic cues as their proficiency increases

• Beginner: use L1 cues to process L2
• Advanced: use L2 cues to achieve more native-like

proficiency Response to C&F The processing of foreign languages can achieve native- like level with the use of L2 cues

Semantics > Syntax Miao (1981); Miao, Chen, & Ying (1986); Li et al. (1993) Syntax > Semantics Chen & Hsu(2017); Hsu (to appear) Purpose explored the role animacy cue plays when two L1 groups

• f speakers comprehend Mandarin.

Results native Mandarin speakers relied more on semantic cues than syntactic cues. NVN heuristics: participants with different proficiency Examples Ball kicks away the sister 球踢走了妹妹 Ball kicks away the sister 球踢走了妹妹

8

Mandarin sentence processing

Differences & Similarities between Mandarin & Spanish

Mandarin Spanish similarities

• 1. Function of Chinese ‘ba’ (patient marker) & Spanish ‘a’ (to):

N1 agent、N2 patient N ‘Ba’ NV N V ‘a’ N

• 2. Passive structures: N1 patient、N2 agent

N ‘Bei’ (agent marker) NV N ‘Ser’ (be verb) VN Differences- 1.meaning 2.Word order 1. ‘Ba’= an object being affected, dealt with 2. Patient+ ‘bei’ +V (NV) 1. ‘a’= pointing out the direction of the object, which does not necessarily have the meaning of affectedness 2. Se (syntactic marker)+ V + patient (VN)

• Syntactic rules with high frequency, credibility and validity in Spanish and Mandarin

Ølinguistic cues for Spanish and Chinese: Syntax > Semantics

9

Research Questions

• For the L1 Spanish L3 Chinese learners, what similarities or

differences exist between low-proficiency and high-proficiency, in terms of accuracy and response times, during Chinese sentence comprehension?

• What cues do L1 Spanish and L1 Mandarin speakers rely on when

processing sentences in L3/L1 Mandarin?

• Which language processing theory?

METHODS

Participants

50 L1 Mandarin speakers

48 L1 Spanish L3 Mandarin speakers

14 beginners- currently learning text book ‘Contemporary Chinese’ Book 3 Lesson 10 to Book 4 34 intermediate-high : passed HSK 3.4/TOCFL A2.B1

12

Material

Sentence structures 13 Animacy

Material presentation

14

f j

Space

Who is the doer Coffee kicked away the sister

Implausible IA SVO

Order of Material

• Order of animacy, sentence structures, and plausibility:

random 'Ba' AI P filler 'Ba' IA IP filler SVO AI IP filler 48 SVO 48 'Ba' (agent marker)

• r

48 'Bei' (patient marker) 96 fillers total: 192

15

Procedure

• L1 Mandarin
• L1 Spanish

Language Background Practice

• 10 trials

formal experiment

16

Language Background Vocabulary Review

Practice

• 10 trials

formal experime nt

RESULTS & DISCUSSION

Statistics

L1 processing & L3 processing L1 vs. L3 Accuracy (R)

• logistic mixed model
• fixed effects: animacy, sentence

structure, plausibility

• random effects: participants, items
• fixed effects:

language proficiency Response time (SPSS、 Excel)

• Generalized Linear Models

(three-way ANOVA with mixed effect modeling)

• fixed effects: animacy, sentence

structure, plausibility

• random effects: participants, items
• fixed effects:

language proficiency

18

L1 Mandarin speakers

• Accuracy:

implausible reverse NVN = Plausible reverse NVN My sister was kicked by the coffee 我妹妹被咖啡踢了

• RT:

passive > S ba OV > SVO Mary was hit by Jane > Jane Ba Mary hit > Jane hit Mary 小明被小王打傷了> 小王把小明打傷了 > 小王打傷了小王 ∵ 1) NVN heuristics (SVO & Ba) faster than reverse NVN heuristics (pasive) 2) Animacy is not significant 3) Plausibility do not influence accuracy Ø Preference in NVN heuristics Ø Syntactic cues are dominant ∵reverse NVN cost more time

L3 Mandarin L1 Spanish speakers

Accuracy: high-level learners = L1 Mandarin speakers low-level learners: animacy AI> IA>AA Mary drinks the coffee > The alarm clock wakes me up > Mary hit Jane RT: low- level learners: animacy IA > AA The alarm clock wakes me up > Mary hit Jane high- level learners: plausibility & sentence structure (1) passive structure < ‘ba’ construction < SVO Mary was hit by Jane < Jane Ba Mary hit < Jane hit Mary 小明被小王打傷了< 小王把小明打傷了 <小王打傷了小明 ∵

• a. similar passive structure in Spanish & Mandarin
• b. semantic differences between ‘Ba’ and “a”

preference in syntactic markers over NVN heuristics (2) implausible SVO < plausible SVO The coffee drinks Mary < Mary drinks the coffee 咖啡喝了小明 < 小明喝了咖啡 if syntactic markers are present, implausibility disappeared ‘ba’ & ‘bei’ were not significant under implausible conditions Ø Preference in syntactic markers over plausibility Accuracy & reaction time: Plausible AA passive (only syntactic cues) = syntactic + semantics cues (imPlausible IA passive) Mary was hit by Jane = Coffee was cooked by the cat 小明被小王打傷了 = 咖啡被小貓煮了

Conclusion

1) Not support C&F’s Shallow Structure hypothesis：

• Mandarin language learners are still capable of comprehending the sentences with
• nly syntactic cues

2) Not fully consistent with the Good Enough Representation model：

• Mandarin native speakers’ reaction time: passive structure (violates NVN) > ba’ &

SVO (doer-receiver NVN)

• > NVN heuristics as a language processing strategy
• High-level learners: plausible sentences- passive structure (reverse doer-receiver

pattern ) < ‘ba’ < SVO

• > relied more on syntactic markers than NVN heuristics.

3) Supports competition model, but the dominant cue of Mandarin is syntax not semantics：

• Mandarin native speakers tended to rely on syntactic cues, while native Spanish

speakers relied more on animacy cues in the beginning of the learning process, but as their proficiency level increases, their reliance on syntactic cues enhanced

21

Thank You!