Based on Lai Yi-shiu (2009). Cognitive

linguistics.

2

• Motivation: Much research has been dedicated to the tense-lax vowel contrast production, but few studies focus on L2 learners’ discriminatory and assimilatory patterns of complete English vowels inventories.

• Experiment1: investigated Chinese-speaking learners’ English vowel discrimination. Experiment2: how these learners classified English vowels and assimilated them to Mandarin categories.

• Finding: Assimilatory patterns between L1-L2 segments in Perceptual Assimilation Model may not fully account for the perception saliency hierarchy and suggested one Tri-dimensional Model for interpreting L2 vowel perception.

Abstract

3Introduction

• In the field of second language acquisition, English vowels pronunciation plays an essential role.

• Especially, contrasts of tense and lax vowels, which are present in English but not in Mandarin Chinese, have been extensively studies in the interlanguage phonology for Chinese-speaking EFL learners.

• In these studies, Chinese speakers often confuse and mispronounce English tense/lax vowel pairs, and such a failure may result in misunderstandings when they converse with native English speakers.

4Introduction

• Speech Learning Model(SLM): L1-L2 segment inventories and adopts the terminology of “similar/old sounds” and “new sounds” in interpreting speech learning. E.g.

• Similarity Effect: Most L2 learners fail to articulate or discriminate some non-native contrasts because they interpret them as being equivalent. E.g.

• Perceptual Assimilation Model(PAM): L2 learners perceptually assimilate non-native contrasts to L1 phonetic categories, and that non-native perception is often filtered by linguistic experience.

5Introduction

• Purpose1: Chinese-speaking EFL learners in Taiwan differentially discriminate which English vowels Experiment1: How Taiwanese EFL learners discriminated English vowel pairs and which pairs posed the greatest challenges for them were addressed.

• Perpose2: Chinese-speaking learners classify which English vowels as similar or new tokens.

Experiment2: Taiwanese EFL learners were asked to classify English vowels as similar or new tokens and transcribe these vowels with Mandarin Chinese phonetic symbols or with IPA notations.

6Literature ReviewPhonological Comparison among English and Mandarin Vowels

• English vowel system is composed of 11 vowels Tongue articulation: high-front high-back low-front low-back Tenseness: tense lax‧Chinese vowel system 7 simple vowels 4 diphthongs

7Literature ReviewPhonological Comparison among English and Mandarin Vowels• Vowels shared in English and Chinese

familiar or similar sounds for Mandarin speakers

• Vowels not shared in English and Chinese

unfamiliar or new sounds• Tenseness dose not play a significant role in

distinguishing Chinese phonemes, native speakers may encounter difficulty in producing and perceiving lax vowels, which contain marked features.

8Literature ReviewSecond/Foreign Language Phonology of Vowel Perception• Universal factors: The template of Mandarin

tense-only vowels has been internalized and further become filters when native speakers of Mandarin begin to acquire English as a foreign language.

• Perceptions of English vowels( front vowels): Mandarin speakers would be assumed to perform better in their identification of English vowels , than . However, Mandarin speakers have better perception than .

9Literature ReviewSecond/Foreign Language Phonology of Vowel Perception• New vowels: Mandarin speakers mainly

contrasted new pairs of vowels by length, while English speakers distinguished these pairs by more parameters (duration and the first two formats).

• High vowels: Mandarin speakers showed excellent perception of English high vowels in listening task.

• Tense-lax vowels contrast??

10Literature ReviewSecond/Foreign Language Phonology of Vowel Perception• SLM: The labels “new” and “similar” can

be assigned only after evaluating learners’ judgments of similarity.

• Similarity Effect: On the contrary, the greater dissimilarity between L1 and L2 sounds is perceived, the easier L2 learners might acquire L2 sounds.

• PAM:L2 learners may perceptually assimilate non-native contrasts to their L1 phonemic categories.

11Literature ReviewSecond/Foreign Language Phonology of Vowel Perception• PAM:

non-native language native language

12MethodResearch Questions

• How did Taiwanese EFL learners discriminate English vowels? To what extent did learners of high English proficiency differ from those of low English proficiency?

• How did Taiwanese EFL learners assimilate English vowels to their L1 Mandarin phonetic categories? To what extent did learners of high English proficiency differ from of low English proficiency.

13MethodParticipants• 90 speakers of Mandarin Chinese in

Taiwan who learn English as a foreign language.

• College students, ages ranged from 19 to 22

• They had been learning English for at least 6 years.

• Non of them lived in an English- speaking country.

• Pre-test is one simulated TOEIC test, and divided the participants on the basis of the test scores.

14MethodParticipants

Group 1 Group245 participants 45 participants

10males 35femles 20males 25females

mean score of 530 mean score of 352

English majors non-English-majors

HEFL LEFL

15MethodExperiment1: English vowel discrimination

• Goal: Examine Taiwanese EFL learners’ English vowel discrimination.

• Procedure: Participants listened to the pre-recorded sounds to decided whether two segments in each word were the same or different. If the same, listeners had to circle SAME and circle the word matched the segment they heard. If different, listeners had to identify which words were the first and second sounds they heard.

16MethodExperiment1: English vowel discrimination• Materials: 7 English minimal pairs in 3 trial-types,

repeated twice by 2 English teachers, and added 8 distracters.

• Skeleton: [h_ t]• 7English minimal pairs:

11 11

• 3 trial-types: A-A,A-B, and B-A• 2 teachers’ voices• 8 distracters: e.g. hit-hid, heed-heat, hate-hayed 7×3×2 ＋ 8=50

17MethodExperiment2: English vowel assimilation

• Materials: the same as experiment 1• Participants’ tasks:

(1) To label each of 11 English vowels as “Similar” or “New” vowels (2) If the vowels are categorized as “Similar” sounds, participants had to transcribe these vowels with Zhuyin Fuhao or with IPA transcriptions.

18MethodData Analysis

• Pair 1: SAME 0score Distinguished 1 score• 3 trails: A-A

• Focus: (a) perceptual discrimination of English vowels (b) perceptual assimilation of English vowels to Mandarin Chinese vowel categories

A-B

B-A

the highest score is 3

19Result and DiscussionEnglish Vowel Discrimination

• The language experience between two different groups acted as a significant factor in distinguish English vowel contrasts.

• Both groups tended to misperceive English vowel pairs to some extent.

20Result and DiscussionEnglish Vowel Discrimination

The hierarchy reflected some similar patterns shared by Taiwanese EFL learners of different proficiency levels.

Finding in pair was in disagreement with Luo(2002), in which speakers of Mandarin Chinese had excellent perception of English high vowels. The reason might be in Luo’s study, only high vowels were examined and excluded mid and low vowels.

21Result and DiscussionEnglish Vowel Discrimination

Misperceiving directions

Misperceived pairs as tense vowels:

Pair 1, 4, 5

Different groups respond differently to one tense-lax vowel pair:

Pair 2

No dominant misperceiving directions:

Pair 3, 6, 7

22Result and DiscussionEnglish Vowel Assimilation

tokens

groups

similar vowels

newvowels

HEFL

LEFL

23Result and DiscussionEnglish Vowel Assimilation

24Result and DiscussionEnglish Vowel Assimilation

25Result and DiscussionDiscussion

• HEFL group performed significantly better than the LEFL group in discriminating most English vowel pairs.

• Most English tense-lax vowel contrasts displayed a greater tendency to be perceived as tense.

• Different English proficiency levels displayed different perceptual assimilation patterns in Experiment 2. (HEFL--- little overlap between English tense vowels and their lax counterparts. LEFL---high degree of overlap in classification between English tense and lax vowels)

26Result and DiscussionDiscussion• Support to the Speech Learning Model,

most English lax vowels were all rated as “new” phones, and tense vowels were “similar” phones for HEFL learners.

• LEFL learners consider as the similar, and less discriminating this pair, which supports Similarity Effect, the more and L2 segment was perceptually to similar to that of an L1 segment.

• Agree with the importance of L2 learning experience. (LEFL learners tend to assimilate English tense-lax vowels as L1 tense vowels, HEFL learners become aware of discrimination of English vowel pairs.)

27Result and DiscussionDiscussion

Perceptual Assimilation ModelThree perceptual assimilation patterns of non-

native segments (a) being assimilated to a native category (b) being assimilated as uncategorizable

speech sound (c) not being assimilated to speechDiscrimination is expected to be excellent when

two non-native categories are assimilated to two different native categories,( UC or CU) and poor discrimination occurs when two non-native categories are assimilated to a single native category. (CC or UU)

?

x

28Result and DiscussionDiscussion

PAM is helpful in two ways (1) It is advantageous in accounting for the

UC or CU types were better discriminated than the CC types in LEFL group.

e.g. (2) Provide plausible account why the English

pairs poses fewer challenges for HEFL than for LEFL. (HEFL--- CU, LEFL--- CC)

29Result and DiscussionDiscussion

PAM fail to address two questions(1) Why did the HEFL group perform the best

in the UU pair than the CU or UC pairs ?

(2) Under which effect did these learners experience different degree of difficulties in L2 vowel discrimination?

(UU) ＞ (CU or UC)

30Result and DiscussionDiscussion

• To account for these unexpected findings , the researcher drew on the facts from markedness effects in the world languages.

(1) Sonority scale: Ranking in perception saliency in English

vowel discrimination might be a major result of the sonority scale.

Low vowels are claimed to be most sonorous , while high vowels least sonorous, [i] context might be comparatively less salient to ear and be more challenging for listeners.

The finding confirmed the universal sonority

scale

＞ ＞

＞ ＞

31Result and DiscussionDiscussion

(2) Minimal Sonority Distance (MSD) Model

e.g. high vowels are 1, mid vowels are 2, low vowels are 3

Sonority distance in is 3-1=2 Sonority distance in is 2-2=0 Vowel pairs with higher MSD settings

were easier to discriminate than those with lower MSD settings.＞

32Result and DiscussionDiscussion

Tri-dimensional model for L2 vowel perception

33Conclusion

• Learners of different English proficiency levels demonstrated similar patterns, but with different degrees of sensitiveness in perceiving English vowel pairs and slightly different misperceiving directions were also identified.

• PAM alone might not fully account for the current finding. One tri-dimensional model for L2 vowel perception, in which language factors and markedness effects were actively involved, was thus suggested.

34Conclusion

• Tense/ lax distinctions in English should thus be made explicit to EFL learners by phonological structures and tri-dimensional model in achieving competence at segmental levels.

• L2/FL phonemic categories can be enhanced with auditory training and appropriate methods, like minimal pairs, in which learners attention could be drawn systematically to the confusing segments in perception.

• The further testing with tri-dimensional model must be carried out to substantiate its feasibility and accountability in the future L2 perceptual studies.

35