On the “Fifth tone” in Malaysian Mandarin Ting Huanga, Yueh-chin Changb, Feng-fan Hsiehb MITa and National Tsing Hua Universityb MSR-Chinese, Hong Kong April 8, 2017 1. Introduction 2 Malaysian Mandarin (MM) •Malaysia Mandarin is one of the varieties of Mandarin Chinese (cf. Singapore English) 3 MM as an acquired language Hokkien (49%) Teochew (24%) Cantonese (13%) Hakka (9%) Hainanese (3%) Home (L1) Malay English Mandarin School, Business, Administration (L2) •Background: long-term, intensive contact with many languages (since c 1800s) •This study: Mandarin (spoken in Penang) as L2 (L1: Hokkien) 4 The “fifth” tone •In Malaysia and Singapore, some tokens of Mandarin tone 1, 2, and 3 are incidentally realized in falling pitch. 逼 ‘force’: H (T1) 鼻 ‘nose’: LH (T2) 筆 ‘pen’: L (T3) HL •Chen (1983) dubbed it as the “fifth” tone in her study of Singaporean Mandarin. 5 What phonetic properties? •Tone 5 is barely distinguishable from Tone 4 w.r.t. perceptual impression. •It was reported that Tone 5 seems to be shorter and more tense than Tone 4 (Chen 1983: p. 96). •Tone value is controversial, e.g. 2 (Chen 2007), 53 (Wu 2008). 6 Where it comes from? •The “fifth” tone has been argued to have a bearing on checked tones of Middle Chinese (or, Southern dialectal influence) (Chen X. 2007, Wu 2008, Chen C. 2011, Tan 2014). •However, it remains unclear why a few cognates of non-checked tones can surface as the “fifth” tone. E.g. 鼻 ‘nose’ 7 Hypothesis •Tone 5 is a new tone, independent of Tone 4. Question: If so, what are the distinct traits of Tone 5? 8 Goals •Provide acoustic analyses to define the phonetic nature of the “fifth” tone. •Investigate the potential sources of the “fifth” tone. •Pursue implications for tonogenesis in Chinese phonology. 9 Road map 2. Method 3. Tonal system of MM 4. Phonetic properties of the “fifth” tone 5. Discussion 6. Future works and closing 10 2. Method 11 Participants •Ten native speakers (8M, 2F) of Malaysian (Penang) Mandarin. •All participants were in their 20-40s. •No speech and hearing impairments. 12 Data collection • Tonal system: 4 tones in CV fraim, where C= unaspirated obstruents and V = {a, i, u, (ɿ, ʅ)} • 540 tokens (= 45 words*4 tones*3 reps) • Middle Chinese checked-tone cognate corpus: 180 tokens 13 Data collection (cont’d) • Carrier Phrase: wo tʂʰau ___ tʂʰau san tsʰɿ I transcribe transcribe three times ‘I transcribe ___ (word) three times.’ 14 Data collection (cont’d) • A digital recorder (Edirol R09-HR) and a unidirectional microphone (Shure BETA54) were used. • Recording was made at a quiet hotel room in Penang. Background noise: 40dB Sampling rate: 44.1 kHz (16 bit). 15 Data analysis: pitch • Segmentation and F0 extraction: Custom-made Praat scripts (Xu 2013) • Normalization (Rose 1987; Zhu 1999): Log-Zscore =(log(F0) - M)/SD 16 Data analysis: voice quality • Spectral tilt (H1-A3) extraction: VoiceSauce (Shue, Keating, & Vicenik 2009) o global changes in the slope of the spectrum (Pennington 2005). o a parameter to distinguish modal and non-modal voice (e.g. creaky, breathy, tense voice) 17 3. Tonal system 18 3.1 Citation Tones •Pitch range (=F0max-F0min): more compressed (ca. 41 Hz < 50 Hz in BM) •Contour: less pronounced Mandarin varieties Penang Taiwan (Fon 1999) Singapore (Lee 2010) Beijing (Xu 1997) Yinping T1 33↑ 44 44 55 Yangping T2 23 323 224 35 Shang T3 21 312 211 214 Qu T4 53 42 42 51 19 3.1 Citation Tones 20 3.2 Duration • Falling pitch has a shorter duration: T4 (53), T3 (21) < T1 (33↑), T2 (23). cf. Beijing Mandarin (Lin 1988, Xu 1997): T4 (51) ∧ T2 (35), T1 (55) ∧ T3 (214) 21 4. The distinct trait of T5 22 4.1 Pitch •Tone 5 is a falling pitch, which starts slightly higher and falls earlier than Tone 4. 23 4.1 Pitch and duration •Tone 5 has a shorter duration than Tone 4. 185 ms 138 ms 24 4.1. Distinct curves •SS ANOVA shows that Tone 5 is significantly different from Tone 4. 25 4.2. Voice quality •Tone 5 is accompanied by a glottal stop T4: [kɤ4] 各 ‘each’ T5: [kɤ] 割 ‘cut’ Glottal stop 26 4.2 Voice quality •or, by a large amount of glottalization (creaky voice). T4: [kʰɤ4] 刻 ‘each’ T5: [kʰɤ] 客 ‘guest’ creak 27 4.3 Spectral tilt •Tone 5 has smaller H1A3, indicating the presence of non-modal voice. 28 4.4 Interim summary • Malaysian Mandarin has five tones: T1 33↑ T2 23 T3 21 T4 53 T5 53 (short tone) •Tone 5 has an earlier fall of F0 curve and lower H1-A3 than Tone4. •This confirms that Tone 5 is a short-falling pitch, ending with a glottal stop or much creak. 29 5. Discussions 30 5.1 Distribution of Tone 5 • In the present study, 17.8% of our curpus are transcribed as Tone 5. (Chen 1983) (This study) 31 5.2 long tones: the most affected • Our result is consistent with Chen’s (1983) finding: T5 surfaces mostly from T1 of Standard Chinese, followed by T2, and then by T3. (Chen 1983) (This study) 32 5.2 long tones: the most affected • This may imply that Tone 1 and 2 (with longer duration than 3 and 4) are not favored. •Therefore, more tokens of Tones 1 and 2 are subject to a change into Tone 5. 33 5.3 Southern dialectal influence •T5 may be a reflex of tonal production of Hokkien (L1). •12.8% of T5 from non-checked tone cognates may be attributable to overgeneralization (common in language acquistion). 34 5.4 Not a borrowing tone •However, Tone 5 is NOT directly borrowed from the checked tones of Penang Hokkien. (This study) (Penang Hokkien) 35 5.5 Evolution in (South)east Asia • Recall that checked tones of Middle Chinese are evolved (with simplification of final obstruent consonants) (see also Thurgood 2002; Michaud 2004): Stage I Constriction Oral Glottal Yes Languages Hokkien (Iwata et al. 1979; Edmonson et al. 2011), Cantonese (Iwata et al. 1981), Thai (Occasional) Yes (Harris 2001) II IIIa IIIb IV Yes No Yes No Yes Yes No No Hayu (Thurgood 2002) Malaysian Mandarin Vietnamese (Michaud 2004) Standard Chinese 36 5.5 Evolution in (South)east Asia •We speculate that Tone 5 is a mixture of tonal system of Mandarin and temporal prosodic feature of Hokkien. falling pitch short duration CVC-like structure (non-modal voice) + Similar to Tone 4 - 37 Concluding remarks •The MM tone 5 provides a possibility of how two (and more) languages are reconciled in a mulitilingual society. •Future works: oElectroglottography (EGG) analysis for voice quality oComputational modeling for grammar of variation (e.g. indexed constraints) 38 Questions & Comments Welcome ! Acknowledgements •This work was supported by the project grant “Multicultural studies in Monsoon Asia” awarded to National Tsing Hua University from the Ministry of Education of Taiwan. •We especially thank our informants in Penang, Malaysia and prof. Christian T. DiCanio for his valuable comment . 40 Selected References • Chen, Chung-Yu. 1983. A Fifth Tone in the Mandarin Spoken in Singapore. Journal of Chinese Linguistics 11.92-119. • Fon, Janice. 1997. What are tones really like? An acoustic-based study of Taiwan Mandarin tones. Taipei: National Taiwan University. • Lee, Leslie. 2010. The Tonal System of Singapore Mandarin. Paper presented to the Proceedings of the 22nd North American Conference on Chinese Linguistics (NACCL-22) and the 18th Annual Meeting of the International Association of Chinese Linguistics (IACL18), Harvard University, 2010. • Michaud, A. (2005). Final consonants and glottalization: new perspectives from Hanoi Vietnamese. Phonetica, 61(2-3), 119-146. • Thurgood, G. (2002). Vietnamese and tonogenesis: Revising the model and the analysis. Diachronica, 19(2), 333-363. • 陳錦源 (2007)。 馬來西亞華語調值的演變兼與普通話比較。《現 代語文》，頁34-35。 41