Re: Chinese tone and pitch ("Alexander L. Francis" )

Subject: Re: Chinese tone and pitch
From:    "Alexander L. Francis"  <afrancis(at)HKUSUA.HKU.HK>
Date:    Thu, 10 May 2001 12:02:18 +0800

At 10:10 AM +0800 5/10/01, lonce wyse wrote: >Hi - > > Chinese is a tonal language, but it is *possible* to understand >Chinese without the pitch information: > >1) Chinese speakers whisper, too! (However, there are formant changes >correlated with pitch in speech) Also, at least for Mandarin Chinese each tone is associated with a distinct amplitude contour. I don't speak Mandarin, but I have been analyzing some Mandarin stimuli and can usually tell the tone of a token just from the amplitude waveform (without listening). Such non-F0 cues might help with understanding whispered speech. See Whalen & Xu (1992) [Information for Mandarin tones in the amplitude contour and in brief segments, Phonetica, 49, 25-47] Note, however, that Whalen & Xu cite a paper by Arthur Abramson that apparently suggests that "tones tend not to be recovered well in whispered speech." But I know Cantonese speakers can understand whispered Cantonese, because my students can whisper to one another quite fluently during class! Furthermore, in my experience, there seems to be much less correlation between amplitude envelopes and tones in Cantonese (which also has more distinct tones than Mandarin). But I don't know of any published research on this topic. Perhaps semantic/pragmatic/contextual considerations play a stronger role than non-F0 acoustic cues in understanding whispered Cantonese. >2) Chinese songs are almost completely devoid of the lexical tonal >information. (However, songs are special in their use of language - >especially regarding redundancy) This is not entirely true, at least for Cantonese. See Wong & Diehl's 1999 abstract in JASA: Melody-tone relation in Cantonese songs. Journal of the Acoustical Society of America, 106, 2286(A). >3) Tones are often sacrificed in natural speech (albeit using pragmatic >and contextual "rules") I'm not sure what you mean by "sacrificed" -- there is certainly a great deal of contextually determined variation (some of it phonological, a great deal more of it phonetic/physiological). But that doesn't mean that the "correct" tone isn't recoverable by listeners. And the amount of phonologically conditioned tone change depends in part on the dialect. Mandarin has a lot more phonological tone sandhi than Cantonese. >4) Monotone computer speech synthesis is also understandable - by both >humans and machines - due to joint word frequency statistics as well as >semantic context. In normal (not-monotone) Cantonese this kind of information is also probably quite important because there seem to be quite a few homophones (including same tone). But I would guess there's at least as much homophony in Mandarin, given fewer tones and a more phonotactically limited syllable structure. >Aren't there any native Chinese speakers who want to "pitch in" here? I >would be interested to know if deaf-from-birth people learn to speak >with tones at all. > >Regards, > - lonce > >p.s. Related topic: I believe Bruno Repp (among others) found that pitch >in Chinese is processed in areas of the brain associated with language >rather than with music. There's an early paper by Van Lancker & Fromkin that, if I remember correctly, shows a right-ear advantage (left-hemisphere advantage) for right-handed native speakers of a tone language (Thai?) discriminating syllables with different pitch patterns, while right-handed native speakers of English show no ear advantage. This is interpreted as suggesting that Thai speakers process spoken pitch preferentially in the hemisphere dominant for language, while English speakers do not (and may process pitch preferentially in the right hemisphere, with music and emotional aspects of speech like "tone of voice," etc.). More recent imaging work by Gandour and his colleagues seems to support this hypothesis. And I'm sure there's been other imaging work on this topic (by Zatorre?) -alex -- Alexander L. Francis Post-doctoral Fellow afrancis(at) Dept. of Speech and Hearing Sciences Tel. (852) 2859-0561 The University of Hong Kong Fax. (852) 2559-0060 Prince Philip Dental Hospital 34 Hospital Road Hong Kong

This message came from the mail archive
maintained by:
DAn Ellis <>
Electrical Engineering Dept., Columbia University