Pan-Mandarin ToBI System



A Comparison of Tiers in (English) ToBI and Those Posited for Mandarin and Cantonese:*

Pan-Mandarin ToBI - Seven Tiers

1. Intone Tier (was: Pitch Range Tier)
   Marks both global and local characteristics of intonation
   %reset -- marks the begining of a new pitch downtrend or pitch reset (See Figure 11 [wave file].)
   %q-raised -- marks the raised pitch range of echo questions (See Figure 4 [no wave file] and Figure 10 [wave file].)
   %e-prom -- marks local expansion of pitch range due to emphatic prominence
   (See Figure 5 [wave file] (narrow focus) vs. Figure 6 [wave file] (broad focus).)
   %compressed -- marks reduction of pitch range of syllables that follow the expansion of pitch range under %e-prom
   (See Figure 5 [wave file].)

   Marks boundary tones
   H% -- high boundary tone at the end of an utterance
   (See Figure 11 [wave file] (H%) vs. Figure 12 [wave file] (L%).
   Also see Figure 1 [wave file] (H%).)
   L% -- low boundary tone at the end of an utterance
   (See Figure 2 [wave file] and Figure 12 [wave file].)

2. Break Indices Tier
   Prosodic phrasing of utterances are represented by break indices.
   Adopting for the current system the set developed by Tseng and Chou 1999a,b):
   0 -- marks reduced syllable boundary
   1 -- marks normal syllable boundary
   2 -- marks minor-phrase boundary
   3 -- marks major-phrase boundary
   4 -- marks breath group boundary
   5 -- marks prosodic group boundary
   (See Figure 2 [wave file] -- men in ni21 men 'you (pl)' is reduced to just the initial bilabial nasal and re-syllabified with ni; the boundary between ni and the bilabial nasal is labelled with break index 0.)

3. Style Tier
   Relative degrees of stress
   S3 -- for syllables with fully-realized lexical tone
   S2 -- for syllables with substantial tone reduction (e.g. undershooting of tonal target with duration reduction)
   S1 -- for syllables that have lost their lexical tonal specification in weakly-stressed positions
   S0 -- for toneless syllables such as SFP's
   (See Figure 2 [wave file] and Figure 8 [wave file].)

4. Romanization Tier (was: Romazi Tier)
   Pinyin romanization with tone numbers, based on dictionary entries. (The lexical tones used are: '55' for Tone 1, '35' for Tone 2, '21' (instead of the citation form, '214') for Tone 3, and '51' for Tone 4.)

5. Code Tier
   Different speech styles/varieties of Mandarin are marked with e.g., <GY, <PTH, and <RGH representing Guoyu, Putonghua, and Rugaohua.
   (See Figure 7 [no wave file] -- The <PTH-word label marks the beginning of the substitution of the word 'children,' and <RGH marks the point where the speaker resumes a pure Rugaohua code.)

6. Word Tier (was: Hanzi Tier)
   Transcription of Chinese characters (cannot yet label using xwaves+).

7. Phonetic Transcription Tier
   Transcription in IPA symbols (cannot yet label using xwaves+).

[ MC's Home Page | MC's ChinaLinks ]

[ DEALL Home Page | The Ohio State University ]

NOTE: This is a draft prepared for my Autumn Quarter's Chinese 889 seminar. Please do not cite! Labelling conventions, names and number of tiers, etc., are subject to change without notice -- for both the Mandarin and Cantonese ToBI systems given here. This page represents work-in-process by the Pan-Mandarin ToBI group consisting of Shu-hui Peng (National Chi Nan University, Taiwan), Mary Beckman, Marjorie Chan, Ok Joo Lee, and Tsan Huang in Summer 1999. Collaboration with Chiu-yu Tseng (Academia Sinica, Taiwan) began with a joint presentation at the ToBI workshop in San Francisco in August 1999. (ToBI is an acronym standing for Tone and Break Indices, and is a standard originally developed by an international group of researchers for transcribing English intonation and has since been applied to other languages, including our current project on Chinese (i.e., Mandarin, Cantonese, and Taiwanese (the last not yet underway)). For transcribing English intonation, see the (original English) ToBI Homepage for information and guidelines for ToBI labelling that were developed at OSU's Department of Linguistics under Prof. Mary Beckman.

Research on the Chinese ToBI systems at The Ohio State University is part of the project on "Establishing a Repository of Linguistically Varied, Prosodically Transcribed Spoken Language Data" (nicknamed "Speech Warehouse"), under Principal Investigators Mary Beckman (Linguistics) and Marjorie Chan (DEALL), Robert Kasper (Linguistics), Terrell Morgan (Spanish & Portuguese), Craige Roberts (Linguistics) and Don Winford (Linguistics). The project is supported by the interdisciplinary seed grant on Spoken Language Understanding and Generation (SLUG), funded by The Ohio State University Office of Research.

Official web pages for Pan-Mandarin ToBI and Cantonese ToBI will be online later. This web page, together with GIFs, was prepared by Marjorie Chan for her Autumn 1999 offering of the Chinese 889 seminar, the topic of which was "Intonation and Sentence-Final Particles." The figures and labelling conventions presented here are based on materials prepared by Shu-hui Peng, who served as the post-doc on the Mandarin project in Summer Quarter 1999, and continues to be associated with the project. Thanks go to Peggy Wong for converting the audio files from xwaves format to .wav format on the Unix late one evening this fall.

Note: A pre-publication copy of the M_ToBI manuscript is available online here.

URL:     http://people.cohums.ohio-state.edu/chan9/MToBI.htm