Contrastive Study on Tonal Patterns Between Accented and Standard Chinese Aijun Li1, Ziyu Xiong1, Xia Wang2 1Institute of Linguistics, Chinese Academy of Social Sciences 5, Jianguomennei Dajie,Beijing,100732 {liaj,xiongzy}@cass.org.cn 2Nokia Research Center, Beijing [email protected] Abstract. This paper depicts a contrastive study on tonal patterns of monosyllables and disyllabic words between Standard Chinese (SC) and three regional accented Chinese(RAC): Shanghai, Xiamen and Taiwan. It is found that for mid level accent speakers, tonal contours of the third tone syllable uttered in isolation or at the final position of a prosodic constituent are always different for RAC from those of SC, displaying a less concave contour or even no concave at all. Moreover, even if the third tone contours are similar in isolation, they still may be different in final positions of bi-syllabic words between RAC and SC. The tonal register and range are compared among SC, RAC and dialects, showing that the RAC tends to keep the same tonal register and range as in dialects, and tonal range is harder to change. A perceptual experiment was carried out to find the third tone perceptual patterns for 3 accented Chinese. All the results imply that third tonal production is a difficult task for some dialectal speakers. Tonal acquisition of the citation form can not completely solve all tonal acquisition problems; tonal combination pattern still needs acquisition. The tonal register and range may be harder to be acquired than tonal contours. Keywords: tonal patterns, contrastive study, accented Chinese, tone production, tone acquisition 1. Introduction There are 10 major dialect groups in China, namely the groups of Mandarin, Jin, Wu, Hui, Xiang (or Hunanese), Gan, Yue (or Cantonese), Min, Ping and Hakka. People from different dialectal areas might not be able to communicate with each other simply because the differences among the dialects are so significant. So, nowadays Standard Chinese (also called Putonghua or Mandarin) is widely used all over China and is popularized by Education Ministry of China. The accented Chinese refers to the Mandarin spoken by dialectal Chinese speakers. The accented Chinese can be regarded as the inter-language in L2 acquisition theory. In the area of accented Chinese study, most researchers have focused on qualitative and phonological description on dialects in popularizing the Standard Chinese. Although some contrastive studies between two dialects or between a dialect and the SC have been published, few investigations have been carried out from the perspective of phonetics. In recent years, we have been doing phonetic analysis on accented Chinese from the view of language teaching and objective evaluation on accent levels. In paper [1], we investigated the difference between accented Shanghai Chinese and SC, finding that tonal patterns in isolation and combination are quite different between SC and accented Shanghai Chinese especially for third tone. This reminds us that there is a tendency for speakers to have tone production problem if those dialects lack the similar tonal pattern as in SC. There are many contributions on Mandarin tone acquisition. Aiming to assess whether typed natural language interaction with a computer using tone-marked Mandarin pinyin can lead to improved tone production in American L2 learners of Chinese, Mitchell Peabody etc.[2] analyzed the tonal patterns of L2 learners and native Chinese teachers in continuous utterances. They found that the tone contours for the native teachers were highly consistent and predictable, but the students' tone profiles varied widely and typically bore little resemblance to the teacher targets. Wang Yue etc. made acoustic and perceptual evaluation of Mandarin tone productions before and after perceptual training for American learners. She found that post-training tone contours approximate native norms to a greater degree than pre-training tone contours. Furthermore, pitch height and pitch contour are not mastered in parallel, with the former being more resistant to improvement than the latter.[3] Studies of hemispheric processing have consistently demonstrated that Mandarin tones are processed differently by native and nonnative speakers. While for native speakers, the neural substrate underlying the ability to identify Mandarin tone is predominantly lateralized in the left hemisphere, this hemispheric specialization for lexical tone is not characteristic of nonnative speakers of Mandarin. [4] In most contributions as mentioned above, Mandarin learner’s native language (L1) is a stress or an accent language. But what we focus here is a special situation, i.e. both source language (dialect) and target language (SC) belong to Chinese tone language family even if some of them may have huge discrepancy like two languages. Tseng made a contrastive investigation on the prosodic properties of intonation between Putonghua (Mandarin spoken in China) and Guoyu (Mandarin spoken in Taiwan),she found that Putonghua demonstrated a general higher register than Guoyu.[5] Richard and Torgerson investigated pitch and tone register differences between native speakers of Taiwan and Beijing Mandarin by means of an acoustic analysis of three speech styles. Results suggest that tones produced in Taiwan Mandarin are in a slightly lower register than those produced in Beijing Mandarin. Speech style was not a significant predictor of pitch register in long segments of recorded speech. [6] Duanmu has provided several examples of tonal differences between Taiwan and Beijing Mandarin, including the opinion that Taiwan Mandarin is characterized by the frequent use of low tones while Beijing Mandarin uses rising or high tones, a difference of tone register.[7] By comparing the tonal contours between SC and Min accented Chinese, we found mid-accent Min speakers have difficulty to produce the third tone contour, because Min dialect lacks a similar low rising contour tone as in SC.[8] In this paper we try to look more dialectal regions to contrast their tonal pattern including tonal contour, tonal range and tonal register. Three regions are selected as Shanghai, Xianmen and Taipei, among which MinNan dialect is spoken in southeast area of Fujian Province and also part of Taiwai. Xianmen dialect is the standard MinNan Dialect. Taipeiness speak Min or Haka, but we only focus on Min dialect speakers in this paper. Wu dialect is a group of dialects spoken in Shanghai, Zhejiang, southern Jiangsu, and part of Fujian and Anhui. Wu dialect is the second biggest dialect running after Mandarin. Shanghai dialect is regarded as the standard Wu dialect. Nowadays, most of Xianmenese, Shanghainese and Taipeinese also speak RAC (GUYU called in Taiwan) while communicating with non-native speakers or dealing with public affairs. RAC is an inter-language which may be put somewhere between dialect and SC. According to the outline of SC spoken skill test constituted by Chinese Ministry of Education, accent is categorized from light to heavy into 3 levels (1-3) and two sub-levels from light to heavy as A and B within each level. In this paper, we only focus on mid-accent level speakers. We hope the results will be helpful for SC learning and automatic accent-level evaluation. 2. Materials There are three kinds of corpus concerned: SC, RAC and dialect corpus (MDC). We selected some speakers with medial accent from each sub-corpus as listed in table 1. The recorded speech includes monosyllables with 4 citation tones and disyllabic words with 16 tonal combinations. The speakers recruited from Taipei can speak both GuoYu and Taiwan Minnan dialect. Table 1. Speaker distribution for each sub-corpus Corpora RAC/MDC SC Shanghai(SH) 30 20 Xiamen(XM) 6 10 Taiwai (TW) 4 4 3. Lexical tone There are four lexical tones in SC, in 5-letter tone scale, they are 55, 35, 214 and 51 respectively. The third tone is a low rising tone (concave tone) in isolation. There are 7 lexical tones in Xiamen dialect (DXM) and Taiwan Minnan dialect (DTW) with 2 checked tones (T6,T7); 5 lexical tones in Shanghai dialect (DSH) with 2 checked tones(T4, T5) as shown in table 2. So we may find that DTW, DXM and DSH don’t have a contour tone like T3 of SC. Table 2. Phonological tones of SC, Xianmen, Shanghai and Taiwan Min dialects (in 5-letter tone scale) languages T1 T2 T3 T4 T5 T6 T7 SC 55 35 214 51 XM/ TW 55 24 53 21 22 5 2 dialect SH dialect 51 13 23 5 2 4. Tonal patterns of monosyllables Five-letter tone scale is used to analyze the tonal features i.e. X=5*(F0-F0min)/ (F0max-F0min). Where X refers to the pitch value under analysis; F0max and F0min refer to the maximum and minimum of the pitch value in Semitone (100Hz as reference frequency). Tonal patterns of 4 citation tones of SC are given in Figures 1 and 2. Four citation tones of accented Shanghai Mandarin (ASH) are shown in figures 3 and 4. Four citation tones of accented Xiamen Mandarin (AXM) are shown in Figures 5 and 6. Figures 7-9 are citation tones for DXM, DTW and DSH.. Comparing these figures, we find there is no phonological difference between lexical tones of SC and ASH (Figures 3 and 4). But a significant different can be found for tone T3 of AXM and ATW (Taiwan Guoyu) speakers, whose tonal features are LLM(214) for SC (figures 1 and 2), ML(31) for AXM and ATW respectively (Figures 5,6,10,11). In this case, T3 of AXM and ATW is prone to be perceived as falling or a neutral tone in SC. By observing the speakers’ mother tones, we may explore that the mis-production of T3 is mainly caused by the lack of such kind of tone in DXM and DTW. According to the Language learning theory, the speakers will produce the new elements by the most similar elements of their own languages.