The Effect of Salience on Chinese Pun Comprehension: An Eye-tracking - - PowerPoint PPT Presentation

The Effect of Salience on Chinese Pun Comprehension: An Eye-tracking Study

Wang Xiaolu Zhejiang University

Contents

• 1. Introduction
• 2. Research Question
• 3. Eye-tracking Experiment
• 4. Results and Discussion
• 5. Conclusion

• 1. Introduction
• Puns are prevalent in daily language use.

– During branding, cowboys have sore calves. (Coulson 2007) – 东边日出西边雨，道是无晴却有晴/情。

• 1. Introduction
• Pun is a special case of ambiguity, where two

meanings are accessed and maintained at the same time. – Accessed in what way?

• 晴-情
• 情-晴
• 晴

情

• 1. Introduction
• The Visual World Paradigm (VWP)

Target Phonological competitor Distractor Distractor

• 1. Introduction
• Findings indicate that the fixation proportion on the

picture referents is modulated by: – phonological information – semantic information – shape information

(Allopenna, Magnuson, & T anenhaus, 1998, Huettig & McQueen, 2007）

• Weakness of the VWP

– pictures have to be used

• 1. Introduction
• The Visual World Paradigm (printed-word version)

car

• 1. Introduction
• Studies from English and Dutch show that fixation

proportion on the printed-words is still found modulated by – phonological /orthographic information – maybe not by semantic information

• 1. Introduction
• Shen, W., Qu, Q. Q., & Li, X. S. (2016): short-

preview condition

Distractor (school) Semantic competitor (nurse) Phonological competitor (hanger) Distractor (insect)

• 1. Introduction
• Shen, W., Qu, Q. Q., & Li, X. S. (2016).

Phonological competitor Semantic competitor Distractor

• 2. Research Question
• Is the salient meaning accessed earlier in the

context of Chinese puns as predicted by the graded salience hypothesis?

Material & Design

• Auditory Stimuli

– replication condition: 40 unambiguous sentences – pun condition: 40 pun sentences – 80 filler sentences; 4 blocks – mostly from newspaper headlines – recorded by a female native Chinese speaker sampled at 44.1 kHz

• 3. Experiment

• 3. Experiment

Material & Design

• Visual Stimuli

– 80 printed-word quadraplet

– Two-character Chinese words matched in word frequency, and the total number of word strokes. – Semantic competitors (Mean >3); phonological competitors or the distractors (Mean <2).

• 3. Experiment

Material & Design

• Visual Stimuli

– Meaning salience: cross-modal lexical decision task.

Semantic Competitor 1 = salient meaning (shorter RT) Semantic Competitor 2 = less salient meaning (longer RT)

– Screen location counterbalanced

• 3. Experiment
• Participants

– 27 college students (mean age= 21.2, SD= 2.0) – normal or corrected to normal vision

• Apparatus

– Eyelink 1000 plus sampled at 1000 Hz – 19-in. monitor seated 60cm from participants, 75 Hz refresh rate, 1024*768 pixels resolution – chin rest with forehead support

• 3. Experiment
• Experiment structure:

• An example trial of the pun condition
• 3. Experiment

Phonological competitor Distractor Semantic competitor 1 Semantic competitor 2

• Playback of an example trial in aggregate mode:

Semantic competitor 2 Distractor Phonological competitor Semantic competitor 1

• 4. Results

中国大学,请不要被排行绑了。

• Fixation proportion to different competitors:

(pun condition)

• 4. Results

Salience effect

• Statistical analysis (pun condition)

– Logit mixed models – Global analysis: 200-800ms; 800-1400ms – Local analysis: 800-1000ms; 1000-1200ms; 1000-1200ms – p value: Bonferroni correction

• 4. Results

• Statistical results (pun condition)

– Salient meaning (Semantic competitor 1 ) received significantly more fixations than the less salient meaning (Semantic competitor 2) during the 200- 800ms time window (p = 0.013*)

distractor

Predictor Estimate SE Wald-Z Adjusted p

(Intercept) Semantic competitor 1 0.26 0.09 2.79 <0.01** Semantic competitor 2

• 0.22

0.09

• 2.48

0.013* Phonological competitor

• 0.22

0.09

• 2.49

0.012*

• 4. Results

• Statistical results (pun condition)

– However, global analysis in the 800-1400ms time window shows no such difference (p = 0.45)

Predictor Estimate SE Wald-Z Adjusted p (Intercept) Semantic competitor 1 0.28 0.07 4.06 <0.001*** Semantic competitor 2

• 0.07

0.09

• 0.76

0.45 Phonological competitor

• 0.45

0.09

• 4.87

<0.001***

• 4. Results

• Statistical results (pun condition)

– Still, the local analysis in the 800-1000ms time window (approximately 200ms after the pun word offset) reveals that semantic competitor 1 receives more fixation than the semantic competitor 2. (p = 0.04*)

Predictor Estimate SE Wald-Z Adjusted p (Intercept) Semantic competitor 1

• 0.45

0.06

• 7.22

<0.001*** Semantic competitor 2

• 0.18

0.09

• 2.05

0.04* Phonological competitor

• 0.58

0.09

• 6.17

<0.001***

• 4. Results

• 4. Discussion
• Chinese par ticipants fixated more on the

semantically-related words compared to the distractors.

– Different from research on alphabetic languages (e.g. English and Dutch), where no such semantic effect was reliably found. – Alphabetic languages have a stronger orthographic form-sound connection; logographic languages, such as Chinese, have stronger orthographic form-meaning relation. – The printed version of the VWP is applicable to semantic research on Chinese.

• 4. Discussion
• Salient meanings of the pun word are accessed

first when listening to Chinese puns. – Cognitive advantage of salient meanings – Lend support to the graded salience hypothesis proposed by Giora (2003).

• 5. Conclusion
• Pun comprehension is guided by meaning

salience associated with the pun word.

• Pun experience may vary with the different

linguistic experience of the pun receiver.

• The printed word VWP is a valid and promising

tool for semantic research on Chinese.

感 谢 聆 听