tone perception and production by cantonese-speaking and english- speaking l2 learners of mandarin...

36
Tone perception and Tone perception and production by Cantonese- production by Cantonese- speaking and English- speaking and English- speaking L2 learners of speaking L2 learners of Mandarin Chinese Mandarin Chinese Yen-Chen Hao Yen-Chen Hao Indiana University Indiana University

Upload: taya-brand

Post on 14-Dec-2015

226 views

Category:

Documents


1 download

TRANSCRIPT

Tone perception and production by Tone perception and production by Cantonese-speaking and English-Cantonese-speaking and English-speaking L2 learners of Mandarin speaking L2 learners of Mandarin

ChineseChinese

Yen-Chen HaoYen-Chen Hao

Indiana UniversityIndiana University

Purpose of this study:Purpose of this study:

Examine the perception and production of Examine the perception and production of Mandarin Chinese tones by second language Mandarin Chinese tones by second language learners whose native languages differ.learners whose native languages differ.

We expect that the native languages of second We expect that the native languages of second language learners will influence their language learners will influence their acquisition of the second language. In acquisition of the second language. In addition, there may be discrepancy between addition, there may be discrepancy between their perception and production of L2 their perception and production of L2 elements.elements.

Target: Acquisition of Mandarin Target: Acquisition of Mandarin Chinese tonesChinese tones

Tone 1: high-level tone (55) Tone 1: high-level tone (55) Tone 2: high-rising tone (35) Tone 2: high-rising tone (35) Tone 3: low-dipping tone (214)Tone 3: low-dipping tone (214) Tone 4: high-falling tone (51)Tone 4: high-falling tone (51)

Cross-language tone perceptionCross-language tone perception Gandour & Harshman (1978) compared the Gandour & Harshman (1978) compared the

discrimination of paired synthetic tones by discrimination of paired synthetic tones by speakers of Thai, Yoruba and English. Some speakers of Thai, Yoruba and English. Some auditory dimensions were important for all three auditory dimensions were important for all three groups of subjects, like pitch height and length of groups of subjects, like pitch height and length of the stimuli. But English speakers did not attach the stimuli. But English speakers did not attach much importance to the contour, while speakers much importance to the contour, while speakers of tone languages did.of tone languages did.

Tone and non-tone language speakers may attach Tone and non-tone language speakers may attach importance to different perceptual cues.importance to different perceptual cues.

Cross-language tone perceptionCross-language tone perception Stagray & Downs (1993) found that speakers of tone Stagray & Downs (1993) found that speakers of tone

languages are not sensitive to slight frequency changes languages are not sensitive to slight frequency changes because they make more categorical judgments of pitch.because they make more categorical judgments of pitch.

Tone language speakers tend to have more categorical Tone language speakers tend to have more categorical perception of pitch patterns than non-tone language perception of pitch patterns than non-tone language speakers.speakers.

Francis et al. (ms) assessed Chinese and English Francis et al. (ms) assessed Chinese and English speakersspeakers’’ identification of Cantonese tones before and identification of Cantonese tones before and after training. There is no significant difference between after training. There is no significant difference between these two group before training. I.e. they are good at the these two group before training. I.e. they are good at the same tones and poor at same ones as well. same tones and poor at same ones as well.

L1 influence may not manifest substantially all the time.L1 influence may not manifest substantially all the time.

Second language acquisition of Second language acquisition of Mandarin tonesMandarin tones

Perception: Perception: Kiriloff (1969) and Chen (1997) found that English-Kiriloff (1969) and Chen (1997) found that English-

speaking learners often confused tone 2 (35) &3 speaking learners often confused tone 2 (35) &3 (214) in perception.(214) in perception.

Production: Production: Shen (1989)found that American learners made more Shen (1989)found that American learners made more

register errors with Tone 1 (55) and 4 (51) and fewer register errors with Tone 1 (55) and 4 (51) and fewer errors with Tone 2 & 3. It was attributed to that Tone errors with Tone 2 & 3. It was attributed to that Tone 1 & 4 are most similar to the pitch patterns in English 1 & 4 are most similar to the pitch patterns in English and thus more susceptible to L1 interference.and thus more susceptible to L1 interference.

Second language acquisition of Second language acquisition of Mandarin tonesMandarin tones

Production (cont.)Production (cont.) MiracleMiracle’’s research (1989) showed that American s research (1989) showed that American

learners made roughly the same amount of errors learners made roughly the same amount of errors across the four tones. across the four tones.

Chen (1997) found alien level tones like /22/ and /33/ Chen (1997) found alien level tones like /22/ and /33/ often substituted the target tones. often substituted the target tones.

No agreement on which tones are most difficult for No agreement on which tones are most difficult for second language learners.second language learners.

There may be discrepancy between perception and There may be discrepancy between perception and production.production.

Tone acquisition by people with Tone acquisition by people with different linguistic backgroundsdifferent linguistic backgrounds

Sun (1998): learners with Sun (1998): learners with ““tone language tone language experienceexperience”” did not make fewer errors than did not make fewer errors than those who without. But their errors were more those who without. But their errors were more consistent. Their perception and production consistent. Their perception and production did not change as the target appeared in a did not change as the target appeared in a different position.different position.

Specific research questionsSpecific research questions

1.1. Which tones are more difficult for second Which tones are more difficult for second language learners? language learners?

2.2. Do second language learners have the same Do second language learners have the same error patterns in tone perception and error patterns in tone perception and production? production?

3.3. Do learners whose L1 is tonal (e.g. Do learners whose L1 is tonal (e.g. Cantonese) have different error patterns from Cantonese) have different error patterns from learners whose L1 is not (e.g. English)?learners whose L1 is not (e.g. English)?

SubjectsSubjectssubjectsubject Native Native

languagelanguagebeginbegin LOLLOL Daily useDaily use

C1C1 CantoneseCantonese 9 9 6 yrs6 yrs 50%50%

C2C2 CantoneseCantonese 1919 < 1 yr< 1 yr 10%10%

C3C3 CantoneseCantonese 88 3 yrs3 yrs 85%85%

A1A1 EnglishEnglish 3333 1.5~2 yrs1.5~2 yrs 0%0%

A2A2 EnglishEnglish 2020 3 yrs3 yrs 2%2%

A3A3 EnglishEnglish 1717 7 yrs7 yrs 80%80%

ExperimentExperiment

Stimuli: Stimuli:

- 4 syllables: /wa- 4 syllables: /waŋŋ/, /ji/, /jo/, /ma//, /ji/, /jo/, /ma/- MonosyllabicMonosyllabic

4 target tones × 4 syllables × 2 repetitions = 324 target tones × 4 syllables × 2 repetitions = 32- DisyllabicDisyllabic

4 target tones × 4 syllables × 2 positions = 324 target tones × 4 syllables × 2 positions = 32

TasksTasks

IdentificationIdentification: perception + linguistic categorization: perception + linguistic categorization MimicryMimicry: perception + production: perception + production ReadingReading: linguistic categorization + production: linguistic categorization + production

TasksTasks1.1. IdentificationIdentification: Subjects listened to the 32 : Subjects listened to the 32

monosyllabic and 32 disyllabic nonsense words monosyllabic and 32 disyllabic nonsense words presented in random order, and marked the tone of presented in random order, and marked the tone of each syllable. each syllable.

2.2. MimicryMimicry: Subjects listened to the same 64 stimuli : Subjects listened to the same 64 stimuli but in a different order and repeated each word but in a different order and repeated each word immediately after hearing it.immediately after hearing it.

3.3. ReadingReading: Subjects were provided with a list of the : Subjects were provided with a list of the same 64 stimuli yet in another order. Each syllable same 64 stimuli yet in another order. Each syllable was spelt in was spelt in pinyinpinyin with tonal diacritics. Subjects with tonal diacritics. Subjects read each word. read each word.

EvaluationEvaluation

2 Chinese native speakers listened to the 2 Chinese native speakers listened to the recording of Mimicry and Reading and judged recording of Mimicry and Reading and judged the tone of each syllable. They did the work the tone of each syllable. They did the work independently and did not know the target tone independently and did not know the target tone when they were judging.when they were judging.

English speakers: Identification English speakers: Identification (3 positions × 8 times × 3 subjects = 72 tokens)(3 positions × 8 times × 3 subjects = 72 tokens)

responseresponseTargetTarget

11 22 33 44

1 (level 55)1 (level 55) 99%99% 1%1%

2 (rising 35)2 (rising 35) 4%4% 83%83% 11%11% 1%1%

3 (dipping 3 (dipping 214)214)

1%1% 15%15% 81%81% 3%3%

4 (falling 51)4 (falling 51) 3%3% 1%1% 96%96%

English speakers: Mimicry English speakers: Mimicry (3 positions × 8 times × 3 subjects × 2 judgments (3 positions × 8 times × 3 subjects × 2 judgments

= 144 tokens)= 144 tokens)

responseresponsetargettarget

11 22 33 44

1 (level 55)1 (level 55) 100%100%

2 (rising 35)2 (rising 35) 4%4% 82%82% 12%12% 1%1%

3 (dipping 214)3 (dipping 214) 7%7% 90%90% 3%3%

4 (falling 51)4 (falling 51) 3%3% 1%1% 96%96%

English speakers: Reading English speakers: Reading (3 positions × 8 times × 3 subjects × 2 judgments (3 positions × 8 times × 3 subjects × 2 judgments

= 144 tokens)= 144 tokens)

responseresponsetargettarget

11 22 33 44

1 (level 55)1 (level 55) 94%94% 1%1% 6%6% 6%6%

2 (rising 35)2 (rising 35) 4%4% 74%74% 24%24%

3 (dipping 3 (dipping 214)214)

3%3% 20%20% 77%77%

4 (falling 51)4 (falling 51) 6%6% 1%1% 1%1% 92%92%

Specific research questionsSpecific research questions

1.1. Which tones are more difficult for second Which tones are more difficult for second language learners? language learners?

2.2. Do second language learners have the same Do second language learners have the same error patterns in tone perception and error patterns in tone perception and production? production?

3.3. Do learners whose L1 is tonal (e.g. Do learners whose L1 is tonal (e.g. Cantonese) have different error patterns from Cantonese) have different error patterns from learners whose L1 is not (e.g. English)?learners whose L1 is not (e.g. English)?

DiscussionDiscussion

Tone 2 (35) and 3 (214) are most problematic. Tone 2 (35) and 3 (214) are most problematic. The confusion is bidirectional.The confusion is bidirectional.

Error rate: Reading > ID > MimicryError rate: Reading > ID > Mimicry They have problems associating the pitch They have problems associating the pitch

contour with the corresponding category contour with the corresponding category (component shared by Reading & ID). (component shared by Reading & ID).

Cantonese speakers: IdentificationCantonese speakers: Identification

responseresponsetargettarget

11 22 33 44

1 (level 55)1 (level 55) 82%82% 1%1% 17%17%

2 (rising 35)2 (rising 35) 14%14% 76%76% 8%8% 1%1%

3 (dipping 3 (dipping 214)214)

3%3% 21%21% 76%76%

4 (falling 51)4 (falling 51) 3%3% 1%1% 96%96%

Cantonese speakers: MimicryCantonese speakers: Mimicry

responseresponsetargettarget

11 22 33 44

1 (level 55)1 (level 55) 90%90% 10%10%

2 (rising 35)2 (rising 35) 1%1% 98%98% 1%1%

3 (dipping 3 (dipping 214)214)

21%21% 79%79%

4 (falling 51)4 (falling 51) 99%99%

Cantonese speakers: ReadingCantonese speakers: Reading

responseresponsetargettarget

11 22 33 44

1 (level 55)1 (level 55) 88%88% 4%4% 8%8%

2 (rising 35)2 (rising 35) 7%7% 73%73% 20%20%

3 (dipping 3 (dipping 214)214)

28%28% 72%72%

4 (falling 51)4 (falling 51) 12%12% 1%1% 87%87%

DiscussionDiscussion

If Tone 3 is misidentified, it is almost always If Tone 3 is misidentified, it is almost always as Tone 2, but not vice versa. as Tone 2, but not vice versa.

There is confusion between Tone 1 (55) and 4 There is confusion between Tone 1 (55) and 4 (51).(51).

Error rate: Reading > ID > MimicryError rate: Reading > ID > Mimicry Problem comes from linguistic categorization.Problem comes from linguistic categorization. Tone 3 biased toward 2 is persistent across Tone 3 biased toward 2 is persistent across

tasks.tasks.

Comparison of English and Comparison of English and Cantonese-speaking subjectsCantonese-speaking subjects

Similarity:Similarity: Both groups make errors probably because they have not Both groups make errors probably because they have not

yet formed a robust association of the linguistic category yet formed a robust association of the linguistic category and the pitch pattern. and the pitch pattern.

Differences:Differences: English-speaking subjects have bidirectional confusion English-speaking subjects have bidirectional confusion

between Tone 2 and 3. Almost no problem with Tone 1 and between Tone 2 and 3. Almost no problem with Tone 1 and 4.4.

Cantonese-speaking subjects misidentified Tone 3 as 2, but Cantonese-speaking subjects misidentified Tone 3 as 2, but not so much vice versa. They confuse Tone 1 and 4 not so much vice versa. They confuse Tone 1 and 4 sometimes.sometimes.

Explanations for the differencesExplanations for the differences English-speaking learners: English-speaking learners: - - Tone 2 and Tone 3 have similar F0 onset and contour Tone 2 and Tone 3 have similar F0 onset and contour

(both have rising)(both have rising)- Li & Thompson (1977) found that L1 children learning - Li & Thompson (1977) found that L1 children learning

Chinese have more problems with Tone 2 and 3.Chinese have more problems with Tone 2 and 3.- Hume and Johnson (2003) measured the perceptual - Hume and Johnson (2003) measured the perceptual

space between the four Mandarin tones and found that space between the four Mandarin tones and found that Tone 2 and 3 are closer to each other for both Chinese Tone 2 and 3 are closer to each other for both Chinese and English speakers.and English speakers.

The distinction between Tone 2 and 3 is hard.The distinction between Tone 2 and 3 is hard.

Cantonese-speaking learners:Cantonese-speaking learners:

L1 interferenceL1 interference

Cantonese tone system (Bauer and Benedict, 1997) Cantonese tone system (Bauer and Benedict, 1997) and corresponding Mandarin tonesand corresponding Mandarin tones

High High levellevel

High High risingrising

Mid Mid levellevel

Mid-Mid-low low levellevel

Mid-Mid-low low fallingfalling

Mid-Mid-low low risingrising

CantoneseCantonese 55/5355/53 2525 3333 2222 2121 2323

MandarinMandarin 55/5155/51 3535

Mandarin Chinese Cantonese Tone 1 (55) → variant 55 of high level tone

↑ free variation

↓ Tone 4 (51) → variant 53 of high level tone

Mandarin Chinese CantoneseTone 2 (35) ↘

high rising tone (25 or 35)Tone 3 (214) ↗

Hypothesized L1-L2 tone mapping:Hypothesized L1-L2 tone mapping:

(1) Tone 1&4 confusion(1) Tone 1&4 confusion

(2) Tone 3 biased as 2(2) Tone 3 biased as 2

For the futureFor the future……

Add a control group of Chinese native speakersAdd a control group of Chinese native speakers Increase subjects to minimize individual Increase subjects to minimize individual

variationvariation Conduct a Mandarin-Cantonese tone mapping Conduct a Mandarin-Cantonese tone mapping

experimentexperiment

ReferencesBauer and Benedict. (1997). Modern Cantonese Phonology. Berlin: Mouton de Gruyter.Chen, Q. (1997). Toward a sequential approach for tonal error analysis. Journal of the

Chinese Language Teachers Association 32: 21-39.Gandour, J. T. and Harshman, R. A. (1978). Crosslanguage differences in tone perception: a

multidimensional scaling investigation. Language and Speech 21: 1-33. Hume, E. and Johnson, K. (2003). The impact of partial phonological contrast on speech

perception. Proceedings of the 15th International Congress of Phonetic Sciences: 2385-88.

Kiriloff, C. (1969). On the auditory perception of tones in Mandarin. Phonetica 20: 2-4.Li, C. N. and Thompson, S. A. (1977). The acquisition of tone in Mandarin-speaking

children. Journal of Child Language 4: 185-99.Miracle, W. C. (1989). Tone production of American students of Chinese: A preliminary

acoustic study. Journal of Chinese Language Teachers Association 24: 49-65.Shen, S. X. N. (1989). Toward a register approach in teaching Mandarin tones. Journal of

Chinese Language Teachers Association 24: 27-47.Stagray, J. R. and Downs, D. (1993). Differential sensitivity for frequency among speakers

of a tone and a nontone language. Journal of Chinese Linguistics 21: 143-63.Sun, S. H. (1998). The development of a lexical tone phonology in American adult learners

of standard Mandarin Chinese. University of Hawai‘i Press.

Thank you!Thank you!

Comments? Questions?Comments? Questions?

English (mono)English (mono)

responseresponsetargettarget

  1 2  3  4

1 (level 55)1 (level 55) 0.99 0.01 0 0

2 (rising 35)2 (rising 35) 0.05 0.78 0.17 0

3 (dipping 3 (dipping 214)214)

0.01 0.08 0.91 0

4 (falling 51)4 (falling 51) 0.02 0 0.02 0.96

English (initial)English (initial)

responseresponsetargettarget

  1 2  3  4 

1 (level 55)1 (level 55) 0.92 0.01 0.03 0.04

2 (rising 35)2 (rising 35) 0.025 0.75 0.2 0.025

3 (dipping 3 (dipping 214)214)

0.01 0.1 0.84 0.05

4 (falling 51)4 (falling 51) 0.058 0.017 0.017 0.91

English (final)English (final)

responseresponsetargettarget

  1 2  3  4 

1 (level 55)1 (level 55) 1 0 0 0

2 (rising 35)2 (rising 35) 0.025 0.83 0.14 0

3 (dipping 214)3 (dipping 214) 0.01 0.23 0.76 0

4 (falling 51)4 (falling 51) 0 0.01 0.19 0.8

Cantonese (mono)Cantonese (mono)

responseresponsetargettarget

  1 2  3  4 

1 (level 55)1 (level 55) 0.98 0.017 0 0

2 (rising 35)2 (rising 35) 0.05 0.85 0.1 0

3 (dipping 3 (dipping 214)214)

0 0.18 0.82 0

4 (falling 51)4 (falling 51) 0.12 0 0 0.88

Cantonese (initial)Cantonese (initial)

responseresponsetargettarget

  1 2  3  4 

1 (level 55)1 (level 55) 0.77 0.0177 0 0.22

2 (rising 35)2 (rising 35) 0.07 0.82 0.11 0.01

3 (dipping 3 (dipping 214)214)

0.017 0.21 0.775 0

4 (falling 51)4 (falling 51) 0.04 0.017 0 0.94

Cantonese (final)Cantonese (final)

responseresponsetargettarget

  1 2  3  4 

1 (level 55)1 (level 55) 0.875 0.025 0 0.1

2 (rising 35)2 (rising 35) 0.067 0.84 0.09 0

3 (dipping 3 (dipping 214)214)

0 0.33 0.67 0

4 (falling 51)4 (falling 51) 0.06 0.013 0 0.92