Intonation Patterns of Yes-No Questions for Chinese EFL learners

Report of Phonetic Research 9 Intonation Patterns of Yes-No Questions for Chinese EFL learners JI Xiaoli Zhejiang University Institute of Linguistics, CASS Jixiaoli@6.com WANG Xia Nokia Research Center, Beijing xia.s.wang@nokia.com LI Aijun Institute of Linguistics, CASS liaj@cass.org.cn Abstract The present study investigates Chinese EFL (English as a foreign language) learners intonation pattern of yes-no questions on the basis of AM theory. According to our study, American speakers adopt a low-level (L*) or low rising tone (L*H) on nuclear accents no matter the nuclear accent is on the medial or final part of a sentence. By contrast, Chinese EFL learners apply a high-level (H*) or falling (H*L) tone when a nuclear accent falls on the medial part of a sentence but a falling (H*L) or low rising tone (L*H) when it is on the final part. The final boundary tone of Chinese EFL learners can be either high (H%) or low (L%) while American speakers mainly apply the H% boundary tone. Besides, Chinese EFL learners pitch movements of nuclear accents in yes-no questions are similar to those of statements.. Introduction Backed by the AM theory, the paper intends to find out Chinese EFL learners intonation regularity, especially the pitch aspect of yes-no questions by a comparison with native American speakers, and to discuss if this is caused by negative transfer for EFL learners. We hope the results can benefit second language learning and Computer Aided Language Learning (CALL) systems. Pierrehumbert specified three types of tonal events for the tonal inventory of English intonation, [], namely, seven pitch accents (H*, L*, H*+L, H +L*, L*+H, L +H*, H*+H ), two phrase accents (H, L ) and two boundary tones (H%, L %). Ladd proposed an improved grammar which can be used to generate all the legal tunes of English. [] Based on AM theory, two intonation contours are described as follows []: Declarative contour contains one or more H* pitch accents and ends in sequence of L- phrase accents and L% boundary tone, i.e. (H*) H* L-L%. Yes-no questions contour, the second type, consists of one or more L* accents followed by H- phrase accent and H% boundary. Many researches have proved that falling patterns are generally used to imply certainty and confirmation in statements, while rising ones are used to indicate doubt and advice in questions. Xu & Liu [] studied the phonetic realization of statements and declarative questions in American English with respect to focus and word stress, and compared it with Chinese intonation. Figure The finite-state grammar of English intonation phrase in Ladd [] Different intonation models have been proposed in Chinese intonation research, [] such as the PENTA model, the Top-Bottom Line model, the STEM-ML model. In addition to the intonational model, based on the AM intonation theory, Lin Maocan [6] pointed out that there are two variables in Chinese intonation: the accents and the boundary tones. However, few phonetic studies have focused on EFL learners intonation patterns, such as yes-no questions. With only one Chinese dialect speaker as subject of his experiment, Shack [7] found that the Chinese speaker used the same general intonation patterns for statements, or not questions as well as for the did and unmarked questions. Chen [8] adopted Halliday s theory to analyze the intonation pattern distinction between American speakers and Chinese EFL learners in respect of tonicity, tonality and tone. However, there is only one yes-no question sentence in her material. The present paper studies the F contours of yes-no questions with various nuclear accent patterns and boundary tones. Comparisons will be made between Chinese EFL learners vs. American speakers, yes-no questions vs. statements, in order to find out Chinese EFL learner s intonation patterns and to explore their relationship with Chinese intonation patterns and negative transfer of L. Before further discussion, it should be pointed out that what this paper refers to as nuclear accent is the last accented word of a phrase [], and the nuclear word is the word bearing the nuclear accent. 7

Report of Phonetic Research 9. Materials The speech materials adopted for this paper consist of sentences [9]. Each sentence was uttered as two sentence types (statement vs. yes-no question) and two nuclear accent places (medial vs. final). Each sentence was read three times by each speaker, resulting in 8 utterances in total. Table List of test materials Sen. Y-N questions vs. statements Could you be mistaken? / You could be mistaken. Can I count on that? / I can count on that. Have you got enough money in hand? /You have got enough money in hand. Can you find out whether John will help? /You can find out whether John will help. Can I see him if I come back later? /I can see him if I come back later. Seven Americans and twelve Chinese served as subjects. The American speakers, six male plus one female, came from western America. The Chinese speakers, on the other hand, were Beijing residents whose native language was standard Chinese, seven of them being female, five male. All of them had no self-reported speech or hearing disorders. The speech data were phonetically and phonologically annotated by ToBI and IViE []. An American expert was invited to check the result of transcription, mainly pitch accents, nuclear accents and prosody organizations. The F for each target sentence was modified manually, extracted by Praat, transferred from Hz to semitone values and then normalized in tone letter space. F of each voiced phone was extracted in points except those creaky voices.. Result and analysis.. Pitch patterns of yes-no question with final nuclear accent Figures -6 display mean time-normalized F contours when the nuclear accent falls on the final word of the sentences. Vertical lines indicate phoneme boundaries. In the legend, A represents American speakers, and C for Chinese EFL learners; C, C, A, A stand for varieties of pitch movement patterns with the same nuclear accent. Nuclear words are all capitalized. Figure shows that when the nuclear accent of Sen. is on mistaken, American speakers apply a low-rising (L*H) tone on the nuclear word while Chinese EFL learners adopt a falling tone (H*L). Both of them present a high boundary tone (H%). American speakers use a low rising tone on later and hand in Sen. and Sen.. Chinese EFL learners apply a falling tone on hand. SEN Could you be MISTAKEN? could you be MISTAKEN k U d j u: b i: m I s t ei k s@ n Figure Time-normalized F contours of Sen. with mistaken as nuclear word T-Value SEN Can I see him if I come back LATER? can I see him if I come back LATER? k ae n ai s i: h (h-)i m s@ f ai k V(s@) m b ae k l ei dt s@r A LATER A C C LATER Figure Time-normalized F contours of Sen. with later as nuclear word SEN Have you got enough money IN HAND? have you got enough money IN HAND h ae v j u: g O: t s@ n V f m V n i: I n h ae n d A Figure Time-normalized F of Sen. with hand as nuclear word Besides high boundary tone (H%), Chinese EFL learners also use low boundary tone (L%).(C in Figure and C in Figure 6) From Figures and 6, we can see that when the nuclear accent is on the final syllable of the sentence, two patterns are observed for Chinese EFL learners-- 76.% apply a low rising tone(l*h) and.8% a falling one (H*L). American speakers provide a low rising tone (L*H) on the nuclear word that. (Note: The phonological annotation in this paper is based on Phonological Tier in IViE, thus phrase accent is not included.) C SEN Can I count on THAT? can I count on THAT k ae n ai k au n t O: n D ae t A C C Figure Time-normalized F of Sen. with that as nuclear word 7

Report of Phonetic Research 9 SEN Can you find out whether John will HELP? can you find out whether John will HELP k ae n j u: f ai n d au t w E D s@r dz A: n w I l h E l p A A C Figure 6 Time-normalized F of Sen. with help as nuclear word The data above illustrate that when the nuclear accent falls on the final word of a sentence, American speakers apply a low rising tone on nuclear words ending with a high boundary tone. But for Chinese learners, the type of nuclear accent is related to place of word stress. If the word stress is at the final syllable of a sentence, Chinese EFL learners adopt two types of nuclear accent patterns, low rising L*H or falling H*L, and the final boundary tones are high (H%) after L*H nuclear accent or low (L%) after H*L nuclear accent; if the nuclear accent is non-final, the nuclear pitch accent of Chinese EFL learners displays a falling H*L tone and the final boundary tones are high (H%) or low (L%) after it. Additionally, a pre-nuclear accent H*L (Note: It can also be transcribed as L*H in ToBI) can be observed for all Chinese EFL learners on the second word of the sentences, a phenomenon absent in the reading of native speakers... Pitch patterns of yes-no questions with medial nuclear accent In Figure 7, when nuclear word is at the medial position of a sentence, there are two types of nuclear accent patterns for Chinese EFL learners, high-level (H*) for C and C or falling (H*L) for C. The boundary tones also display two types, high (H%) or low (L%). In Figure 8, American speakers also apply two nuclear accent patterns, low rising tone (L*H) for A and falling tone (H*L) for A (only one speaker). Though nuclear contour types are different, the boundary tone is the same (H%). However, our study shows a difference for Chinese EFL learners, who apply a falling tone (H*L) on nuclear word with two different types of boundary tone, high (H%) or low(l%). SEN Can I COUNT on that? can I COUNT on that k ae n ai k au n t O: n D ae t SEN Have you got enough MONEY in hand? have you got enough MONEY in hand h ae v j u: g O: t s@ n V f m V n i: I n h ae n d A A C C Figure 8 Time-normalized F contours of Sen. with money as nuclear word SEN Can you find out whether JOHN will help? can you find out whether JOHN will help k ae n j u: f ai n d au t w E D s@r dz A: n w I l h E l p A A C Figure 9 Time-normalized F contours of Sen. with John as nuclear word In Figure 9, all Chinese EFL learners adopt a falling tone (H*L) on nuclear words. American speakers tend to use low rising tone (L*H) on the nuclear word John, yet there are two different boundaries, H% or % (no pitch movement at boundary in IViE). In conclusion, when nuclear accents are in the middle of sentences, Chinese EFL learners adopt two types of nuclear accent patterns, high-level(h*) or falling(h*l) while American speakers mainly apply the low-rising tone. The boundary tone of American speakers is high while the boundary of Chinese EFL learners may be low (L%) or high (H%). The other distinction is that there is always a pre-nuclear pitch accent H*L at the second word of the yes-no questions for Chinese EFL learners with a low initial boundary tone (%LH*L)... Comparison of pitch patterns between yes-no questions and statements This section shows mean time-normalized F contours of yes-no questions and statements with the same nuclear accent. The purpose is to find out the differences of pitch patterns of yes-no questions and statements between American speakers and Chinese EFL learners. In the following figures, Q represents yes-no question, S for statement, and the words capitalized are nuclear words. The first two words of questions and statements are different from each other. A C C C Figure 7 Time-normalized F contours of Sen. with count as nuclear word 7

Report of Phonetic Research 9 SEN Can I COUNT on that? Q can I COUNT on that S I COUNT on that can k ae n ai k au n t_h O: n D ae t Q COUNT A S COUNT A Figure Time-normalized F contours of Sen. for American speakers, nuclear accents are on the word count for both yes-no questions and statements. SEN Can I COUNT on that? Q can I COUNT on that S I COUNT on that can k ae n ai k au n t_h O: n D ae t Q COUNT C Q COUNT C Q COUNT C S COUNT C Figure Time-normalized F contours of Sen. for Chinese EFL learners, nuclear accents are on the word count for both yes-no questions and statements. Figure and Figure show that American speakers apply a low rising tone on nuclear accents with a high boundary tone (L*HH%) in yes-no questions but a falling tone on nuclear words with a low boundary tone(h*ll%) in statements. Chinese EFL learners also adopt a falling tone on nuclear words with a low boundary tone (H*LL%) in statements. As for yes-no questions, two nuclear accent patterns are observed from Chinese F contours, high-level tone (H*) (Q COUNT C and Q COUNT C in figure) or falling tone (H*L) (Q COUNT C in figure ). There are also two types of boundary tones, high or low. Though Chinese EFL learners apply a falling tone on nuclear words both in statements(s COUNT C in figure ) and in questions (Q COUNT C), the F contours after nuclear accent are different between statements and questions. The F contour of statements declines gradually to the end of the sentences. The F contours of questions also declines to the end of the sentence (Q COUNT C and C in figure ) or before the high boundary tone (Q COUNT C in figure ). But the F contour of questions continues in a higher pitch register than that of statements after nuclear accent. SEN Have you got enough MONEY in hand? got enough MONEY Q have you in hand S you have got enough MONEY in hand h ae v j u: g O: t s@(i)n V f m V n i: I n h ae n d S MONEY A Q MONEY A Figure Time-normalized F contours of Sen. for American speakers, nuclear accents are on the word money for both yes-no questions and statements. SEN Have you got enough MONEY in hand? Q have you got enough MONEY in hand S you have got enough MONEY in hand h ae v j u: g O: ts@(i)n V f m V n i: I n h ae n d S MONEY C Q MONEY C Q MONEY C Figure Time-normalized F contours of Sen. for Chinese EFL learners, nuclear accents are on the word money for both yes-no questions and statements. The same phenomenon can be found from Figure for American speakers. From Figure, we can see that Chinese EFL learners apply a falling tone (H*L) on nuclear words in both statements and yes-no questions. But the pitch register of a yes-no question is higher than that of a statement from the beginning of the sentence. The pitch accent of Sen. and Sen. falls on the last word for both yes-no questions and statements. American speakers apply a low rising tone on nuclear words with a high boundary tone in yes-no questions and a falling tone on nuclear words with a low boundary tone in statements. Chinese EFL learners adopt a falling tone on nuclear words in both questions and statements. The boundary tone of questions is high while that of statements is low (figure ). Figure shows us that American speakers apply a rising tone on nuclear words in questions (L*H) but a falling tone in statements (H*L). Figure 6 shows that Chinese EFL learners also apply a falling (H*L) tone on nuclear words in statements. But two types of nuclear accent patterns are found in yes-no questions, falling (H*L) or low rising (L*H). SEN Could you be MISTAKEN? Q could you be MISTAKEN S you could be MISTAKEN k U d j u: b i: m I s t ei k s@ n S MISTAKEN A S MISTAKEN C Q MISTAKEN A Q MISTAKEN C Figure Time-normalized F contours of Sen. for American speakers and Chinese EFL learners: both Q and S having the nuclear accents at sentence finals. 7

Report of Phonetic Research 9 SEN Can you find out whether John will HELP? Q can you find out whether John will HELP S you can find out whether John will HELP k ae n j u: f ai n d au t w E D s@r dz O: n w I l h E l p S HELP A Q HELP A Figure Time-normalized F contours of Sen. for American speakers SEN Can you find out whether John will HELP? Q can you find out whether John will HELP S you can find out whether John will HELP k ae n j u: f ai n d au t w E D s@r dz O: n w I l h E l p S HELP C Q HELP C Q HELP C Figure 6 Time-normalized F contours of Sen. for Chinese EFL learners In brief, the differences between yes-no questions and statements for American speakers are nuclear contour patterns and boundary tones. F contours of yes-no questions consist of a L* nuclear pitch accent and end in a H% boundary tone while F contours of statements contain a H* nuclear pitch accent followed by a L% boundary tone no matter the nuclear accent is at the medial or final of the sentence. When a nuclear accent is not on the final syllable of the sentence, Chinese EFL learners apply the same tone on nuclear words in questions and statements, that is, the H*L tone. The boundary for statements is low (L%) while that for questions is low (L%) or high (H%). When a nuclear accent falls on the final syllable of the word, Chinese learners adopt a falling tone on the nuclear accent with a low boundary tone (H*LL%) in statements while a falling tone on the nuclear word with a low boundary tone (H*LL%) or a low rising tone on the nuclear word with a high boundary tone in questions (L*HH%). A pre-nuclear H*L pitch accent is always observed at the second word of the yes-no questions for Chinese EFL learners with a low initial boundary (%L).. Comparison of pitch patterns of yes-no questions with different nuclear words for Chinese EFL learners In order to find out the effects of nuclear accents on F contours of yes-no questions, this section compares F contours of the same sentence with different nuclear words. In the legend of the following figures, the words capitalized stand for nuclear words, while C, C, and C means varieties of intonation patterns with the same nuclear accent. SEN Can I count on that? Can I count on that k ae n ai k au n t O: n D ae t COUNT C COUNT C COUNT C THAT C THAT C Figure 7 Time-normalized F contours of Sen. with different nuclear words for Chinese EFL learners We can see from Figure 7 that the pitch values of count in COUNT C and COUNT C are higher than those of count in THAT C and THAT C. In Figure 8 the pitch values of money in MONEY C and MONEY C are higher than those of money in in hand. In fig. 9 the pitch value of John in JOHN C is higher than in HELP C and HELP C. These imply that the pitch values of nuclear words are higher than those of nonnuclear words for Chinese EFL learners. SEN Have you got enough money in hand? have you got enough money in hand h ae v j u: g O: t s@ n V f m V n i: I n h ae n d MONEY C MONEY C IN HAND C Figure 8 Time-normalized F contours of Sen. with different nuclear words for Chinese EFL learners SEN Can you find out whether John will help? Can you find out whether John will help k ae n j u: f ai n d au t w E D s@r dz A: n w I l h E l p JOHN C HELP C HELP C Figure 9 Time-normalized F contours of Sen. with different nuclear words for Chinese EFL learners. Discussion and Conclusion Results show Chinese EFL learners nuclear pitch patterns of yes-no questions are quite different from those of American speakers. American speakers apply a low rising tone (L*H) on nuclear words in yes-no questions no matter the nuclear accent is on the medial or final part of a sentence. When the nuclear accent is not on the final syllable of the sentences, Chinese EFL learners always apply a high-level tone (H*) or a falling tone (H*L) on the nuclear accent. A low rising tone (L*H) is one variety that can be found when 76

Report of Phonetic Research 9 a nuclear accent falls on the final syllable of a sentence. Comparing to the research on intonation patterns of Mandarin, it s easy to find the effects of negative transfer of L. Jia pointed out that [], for Mandarin statements, the phonetic realization of the intonational stress is to enlarge the pitch range of the under-focus position, and compress the pitch range of the post-focus syllables. The enlargement is realized through raising the H target tone, while leaving L tone basically unchanged. Ling [] proved that the phonetic realization of the sentence stress for Tone (LL) in Mandarin is mainly to raise the H tone around instead of lowering the L tone. Besides, Chen [] and Wang [] found that Tone is the most difficult one to bear the sentential stress among four tones in Mandarin. From the above researches, it can be seen that the intonation stress in Mandarin is realized mainly through H tone. Thus, it can be inferred from the above that Chinese EFL learners may unconsciously borrow high-level or high falling tones from their native language to nuclear accents in their English. Table YN intonation patterns for Chinese ELF learners I Initial boundary Pre-nu clear %L H*L Initial boundary Pre-nu clear Nuclear (media word) final boundary H* L% H*L L% / H% Nuclear final (final word) boundary H* L L% II %L H*L L*H (final H% syllable) In conclusion, American speakers pitch patterns of yes-no questions consist of a low-rise nuclear pitch accent and a high boundary tone (L*HH%). A falling tone has been adopted by one native speaker on a nuclear word for Sen.. Yet this is demonstrated by Nancy []: 6.8% of American speakers adopt low-level tone (L*) on nuclear words,.88% use a falling tone (H*+L) and.89% use a high-level tone (H*) on nuclear words. Besides the low rising tone (L*H), Chinese EFL learners apply two other pitch pattern for yes-no questions: high-level tone (H*) or low rising tone (L*H). The pitch patterns are closed related to the places of nuclear accent and of word stress (In Table ). References [] Pierrehumbert, J. B., The Phonology and Phonetics of English Intonation, Ph. D dissertation, MIT, 98. [] Ladd, D. R., Intonational phonology, Cambridge University Press, Cambridge.996. [] Beckman, M. and G. Ayers, Guidelines for ToBI labeling, Ohio State University. 997. [] Liu, F. and Xu, Y., Question Intonation as Affected by Word Stress and Focus in English, 6th International Congress of Phonetic Sciences, Saarbrucken,Germany, 7. [] Li,A. and Lin, M. Studies on Chinese Prosody-Professor WU Zongji s thoughts on prosody and their far-reaching implications, Proc. of PCC 8, Beijing, China, 8. [6] Lin, M., Interrogative Mood and Boundary Tone in Chinese, Report of Phonetic Research, Beijing, 6, pp. -. [7] Shack, K., Comparison of Intonation Patterns in Mandarin and English for a Particular speech, In Katherine M. Crosswhite & Joyce McDonough (Eds.). University of Rochester Working Papers in the Language Sciences. Vol. Spring, No.. [8] Chen, H., On Chinese EFL learner s English Intonation Patterns, Shanghai Foreign Language Education Press, Shanghai, 6. [9] Wang, X., LI, A., Xiong, Z., Yin, Z., Multi-accent and Multi-lingual Speech Corpus, Proc. of O-COCOSDA8, Japan. [] Jia, Y. and L,i A., An Introduction to IViE English Labeling System, Report of Phonetic Research, Beijing,, pp.76-8. [] Jia, Y., Xiong, Z. and Li, A. The Effect of Focal Accents upon Sentential Pitch in Standard Chinese, Report of Phonetic Research, Beijing, 6, pp.6-68. [] Ling, F., A Study on the acoustic correlates of emphasized Tone in Mandarin Chinese, master dissertation, Peking University,. [] Chen, J., Contrastive study on prosodic aspects for Standard and Shanghai-Accented Chinese, master dissertation, Zhejiang University,. [] Wang Y., Chu, M. and He, L., An experimental study on the distribution of the focus-related and semantic accent in Chinese, The Proceeding of the 6th International Conference on Spoken Language Processing, Tianjin,. [] Hedberg, N., Juan M. ect. Tonal Constituents and Meanings of Yes-No Questions in American English, Proceedings of Speech Prosody, Dresden, Germany, 6. 77