[ back to publications page ]

This book, a revised version of a chapter from the author's dissertation (Shen, 1986), is an acoustic study of intonation in Mandarin Chinese. Shen's particular focus is on the pitch levels and pitch contours of the different intonation patterns, and on how intonation interacts with not only tone, but also stress. The interaction among the three components of tone, stress and intonation is made the more interesting by the fact that previous acoustic and perception studies of Mandarin have shown that their main acoustic correlate is fundamental frequency (Howie, 1976; Tseng, 1981; Coster and Kratochvil, 1984; and others cited in Shen, p.2) . Precisely for this reason, Fo measurements are the primary ones Shen made in her study. Shen isolates intonation to determine its main patterns, and then investigates its effects first on tone, and subsequently on tone and stress. This organization is reflected in Chapters 2, 3 and 4 respectively, which constitute the bulk of the book. The four other chapters include the introduction (Chapter 1), conclusion (Chapter 7), and two very short chapters discussing the phonological status of components of prosody (Chapter 5), and the main characteristics of intonation in a tone language such as Mandarin as opposed to those found in non-tone languages such as English or French (Chapter 6).

Shen lays out three problems with previous analyses and descriptions of Mandarin intonation in Chapter 1: (1) imperfect investigation methods, (2) confusion of multiple uses of pitch, and (3) inaccurate assumptions about intonation makeup. Concerning the first problem, Shen objects to the inaccuracies that can arise from impressionistic observations on the one hand, and to the recording of artificial, read speech instead of more natural free conversation. However, she also notes that studies that have employed natural running speech (e.g., Tseng, 1981) present yet another difficulty, namely the near impossibility of controlling and manipulating different factors. Shen's solution is to use a corpus of sentences, each of which was first orally presented to the subject for memorization before the relevant contexts were provided for the subject to say that sentence with a particular intonation pattern.

The second problem involves the fact that tone, stress and intonation are all primarily indicated auditorily by pitch, with pitch level in particular relevant to stress and intonation. Stress has hitherto been largely ignored in intonation studies on Mandarin. Shen's study thus serves to fill that gap by providing observations and some clarification of the interaction among these three prosodic phenomena.

Shen defines the third problem in terms of an opposition between two different viewpoints concerning the effect of intonation on tones in Mandarin: whether the imposition is successive, as proposed by Chao (1933:131ff), or simultaneous, as suggested by later scholars (e.g., Ho, 1977). By "successive", Chao means, for example, the adding of a rising pitch after an original falling pitch, to produce a falling-rising tone. "Simultaneous" addition is described by Shen (p.4) as involving the superimposition of an intonational trend over a whole on a sentence, so that incompatibility between intonation contour and lexical tone would result in tone neutralization, while congruency between these two prosodic phenomena would emphasize or enhance the contour features of the original tones. By "simultaneous", Shen is essentially combining two of Abe's (1980:9) three models of intonation makeup: namely, "conflictive" (when tone and intonation are incompatible), and "cumulative" (when they are compatible), both of which he opposed to "successive".

Shen's discussion of the problem would be clearer if she explained her terms, especially since Chao himself (1933) originally used the term "simultaneous" (in contrast with "successive") addition to mean widening and narrowing of pitch range, as well as to mean raising and lowering. He provided examples to illustrate both types of additions in Mandarin. Shen, however, disputes that successive addition occurs, on the basis of her results (p.38). Perhaps the dispute is spurious, and both are correct. The two types of additions may be occurring in two different types of utterances, only one of which was studied by Shen.

One of the important contributions of Shen's work is her exploration of the basic intonation patterns of Mandarin in Chapter 2. She selected six types of sentences, five of which represent different types of interrogative sentences. Only Type A sentences are statements, or declaratives. Types B and C are yes-no questions (B sentences are syntactically unmarked, or echo, questions, whereas C sentences are marked with a sentence-final ma particle). Types D and E sentences are disjunctive questions. (D represents a subset consisting of a verb and its negation, e.g., "Are you going or not?", whereas E represents disjunctive alternatives, e.g., "Did you go to school or stay home today?".) And finally, Type F sentences are WH-questions, which in Chinese do not trigger word-order change.

Shen recorded these six types of utterances using sentences in which every syllable bears Mandarin Tone 1, a high level tone. Shen obtained the mean Fo value for these sentence types at four selected points: starting point, highest peak, lowest trough, and ending point. (She plots the points for comparison in Figure 2.3 on page 20; however, since the lines in the legend do not match those on the graphs, it is necessary to turn to the mean Fo values in Table 2.8 on page 19 to determine which line represents which type of sentence.) Declarative sentences distinguish themselves in having a mid starting point, a mid-high value for the highest peak, and a low ending point. Yes-no questions, with or without the ma particle, differ from declaratives in having a mid-high starting point, a high value for the highest peak, and a high to mid-high for the ending point. In other words, yes-no questions exhibit an overall higher pitch register than declaratives. This last point is of theoretical interest, since Shen (p.27) uses it to argue against models with a movable topline but invariant baseline (Pierrehumbert, 1980; Gårding, 1983), on the grounds that they do not adequately capture the phonetic facts of Mandarin. Disjunctive questions share certain characteristics with both yes-no questions and declaratives. They have a mid-high starting point and a high value for the highest peak similar to yes-no questions, but they have a low ending point similar to declaratives. Differing only slightly from disjunctive questions are the last category, namely, WH-questions. Compared to disjunctive questions, WH-questions have a slightly higher starting point and a somewhat lower value for their highest peak. WH-questions are like disjunctive questions and declaratives in having a low ending point.

From the intonation patterns of these various sentence types, Shen extracts three basic tunes: Tune I consists solely of declarative sentences, Tune II consists of both marked and unmarked yes-no questions, and Tune III consists of the remaining sentence types, namely, disjunctives (A-not-A and alternative interrogatives) and WH-questions.

However, despite their overall similarity, finer distinctions within Tune III might separate disjunctives from WH-questions, in that disjunctives not only have a steeper slope to the highest peak, but the highest peak also has a mean Fo value that is higher than the corresponding point for WH-questions across all six subjects. The two types of yes-no questions under Tune II also differ slightly from each other. For four of the subjects, at least, echo questions consistently have a higher mean Fo value than ma-particle questions at the ending point. The mean Fo difference is 17.3 Hz. The net result is that these unmarked questions have the highest ending point, contrasting with declaratives having the lowest ending point, thus maximizing their functional differences using intonation. For only two of the subjects is the difference in ending point negligible (4.6 Hz). (This may merely be a matter of individual differences, since one of these subjects has been in the U.S. for five years, the longest period of time, while the other is Beijing-born but later resided in Xi'an.)

Chapter 3 contains useful and interesting acoustic description of the effect of intonation on the four lexical tones in Mandarin, as well as on the neutral tone and the tonal variants that surface in third tone sandhi. The main data are presented in Figure 3.3 (pages 42 through 47), a set of examples of Fo contours illustrating the different sentence types. This figure would have been easier to interpret if explicit labelling of sentence types had been provided, and the three basic tunes from Chapter 2 were also specified. Minor typographical errors in the figures include tone diacritics that are not consistently omitted to mark neutral-tone syllables. There are also a few small errors in phonetic transcription of the sentences. For the general readership, it would probably be helpful to have included English glosses, or at least the Pinyin romanization from Appendix I corresponding to the phonetic transcription. Similar remarks hold for Chapter 4, which is a shorter chapter and highlights some of the effects of intonation on both tone and stress. Overall, these two chapters are useful in furthering our understanding of the nature of tone, stress, and intonation in the language.

While the basic description in these two chapters is fairly straightforward, Shen's analysis of how intonation interacts with tone and stress is sometimes difficult to follow. This partly stems from Shen's outline of the second problem in her introductory chapter. She takes her background information on Mandarin stress essentially from Chao (1948, 1968), as if to adopt his approach, but makes some changes to Chao's transcription, which effectively change the analysis. Chao (1968:35, 38) identifies three levels of stress: weak, normal, and contrasting (or contrastive). Weak stress is synonymous with neutral tone: a syllable with neutral tone has tone range flattened to practically zero, relatively short duration, and a pitch value that is parasitic on that of the preceding syllable. Most cases of weak stress are enclitics, involving a small number of morphemes, such as suffixes and particles, which always have weak stress and thus always occur in the neutral tone. Contrasting stress, serving similar functions as in English, has a wide pitch range, long duration, and usually increased loudness. All other syllables have normal stress. In the case of polysyllabic words and phrases with normal stress containing full-tone on each syllable, degree of phonetic stress differs according to position: stress is strongest on the last syllable, next on the first, and least on the medial syllable(s). Chao (1948:26) does not distinguish among these levels of normal stress phonemically. They contrast as a group to atonic syllables.

Shen's analysis of stress effectively differs from Chao's in at least two respects. The first pertains to the location of weak stress. Shen uses the term 'weak' stress to refer to medial syllables that Chao treated as full-tone syllables with tertiary stress, a phonetic variant of "normal" stress. A second related difference arises with respect to where neutral tone surfaces. Shen transcribes full tones for some syllables that Chao transcribed as having neutral tone, i.e., weak stress. In particular, Shen (p. 2) transcribes trochees with lexical tone on both syllables, although such words are transcribed by Chao with lexical tone on the first syllable only (e.g., zhidao 'know') and neutral tone on the second. Given these two differences, the overall result is that Chao ties weak stress to neutral tone, so that neither occurs without the other. Shen, on the other hand, permits underlying full tone on weakly-stressed syllables. Stress and intonation then determine whether a given syllable is toneless on the surface or not. What is then not made explicit is how stress is assigned.

In analyzing the interaction of intonation with other prosodic phenomena, Shen distinguishes a phonetic level from a phonological one (p.3). A more detailed account is needed of the distribution of the tasks performed by these two different levels. It is not clear how Shen assigns underlying representations in the phonology. On page 39, Shen refers to syllables losing their "distinct intrinsic fundamental frequency pattern." Whether that Fo pattern refers to the citation tone on such syllables or an underlying tone that may or may not correspond to the citation tone is not made explicit. Even equating underlying tone with the tone contour assigned by the dictionary entry, as Shen does on page 64, does not fully clarify whether the tone contour is the isolation tone contour or the tonal value that surfaces in the given context of particular lexical entries.

Understanding which tonal variant is regarded as the underlying tone is particularly crucial in the case of Tone 3. This tone has three variants: it is low falling-rising [214] prepausally (including in citation context), mid-rising [35] before another Tone 3, and low(-falling) [21] elsewhere. While the citation variant [214] is often treated as the basic form, especially by traditional scholars in China, Tone 3 has also been variously analyzed in North America as a low tone, taking the low [21] variant as the basic, or underlying, form (Hartman, 1944; Hockett, 1947; Kratochvil, 1968). This is because the low tone variant occurs in the greatest number of environments, namely, before all tones except Tone 3, while the citation form only occurs prepausally.

If Shen had treated the low variant as the underlying form of Tone 3, her analysis in Chapter 3 could have been simpler. First of all, Shen (p.52-53) observes that Tone 3 changes to the rising [35] variant even though it precedes the falling [21] variant. Shen regards this as revealing "a contradiction between phonological prediction and actual phonetic output [since] according to phonological prediction, a 3rd tone would change to a rising tone only when preceding another 3rd tone." This problem disappears if one treats Tone 3 as underlyingly /21/.

Shen makes no overt statement as to rule-ordering and which derived forms might be fed into a later rule. A statement as to which tones, underlying and/or derived, are indicated by the terms 'R' (Rising) and 'F' (Falling) in the rules, and how they differ from 'T3' (Tone 3) would have been desirable, as would sample derivations for determining what the underlying representations are, when stress is assigned, and how one proceeds, methodologically, from phonological representation to phonetic output.

Overall, the main contributions of the monograph are in Chapters 2, 3 and 4. With respect to typographical errors, they are relatively few in number. One error that recurs in the book is the spelling of 'Hua' for 'Hoa' in Monique Hoa's name, found both in the text (e.g., page 60), and in the bibliography in reference to that author's 1980 (not 1981) dissertation. Other minor typos found in the bibliography include 'Jong' instead of 'Jiong' for Jiong Shen's name, '1967' instead of '1970' for Yao Shen's reference, and incomplete reference on Coster and Kratochvil (1984).

Given the paucity of acoustic literature on intonation in Mandarin in general, and even less dealing specifically with the interaction among tone, stress and intonation in their use of fundamental frequency, Shen's monograph will hold an important place in the literature on this topic for some years to come.

References

Chan, M.K.M. (1989) On the status of 'basic' tones, Acta Linguistica Hafniensia, 21, 5-34.

Chao, Y.R. (1933) Tone and intonation in Chinese, Bulletin of the Institute of History and Philology, 4, 121-134.

Chao, Y.R. (1948) Mandarin Primer. Cambridge: Harvard University Press.

Chao, Y.R. (1959) The morphemic status of certain Chinese tones, Transactions of the International Congress of Orientalists in Japan, 4, 44-48.

Chao, Y.R. (1968) A Grammar of Spoken Chinese. Berkeley: University of California Press.

Coster, D. & Kratochvil, P. (1984) Tone and stress discrimination in normal Beijing speech. In New Papers on Chinese Language Use (B. Hong, editor), 119-132. Canberra: Contemporary China Centre Research School of Pacific Studies, Australia National University.

Gårding, E. (1983) A generative model of intonation. In Prosody: Models and Measurements (A. Cutler & D.R. Ladd, editors), 11-26. Berlin: Springer.

Hartman, L. (1944) The segmental phonemes of the Peiping dialect, Language, 20, 28-42.

Ho, A. (1977) Intonation variations in a Mandarin sentence for three expressions: interrogative, exclamatory, and declarative, Phonetica, 34, 446-456.

Hockett, C. (1947) Peiping phonology, Journal of the American Oriental Society, 67, 253-267.

Howie, J. (1976) Acoustical Studies of Mandarin Vowels and Tones. Cambridge: Cambridge University Press.

Kratochvil, P. (1968) The Chinese Language Today: Features of an Emerging Standard. London: Hutchinson and Company.

Pierrehumbert, J. (1980) The Phonology and Phonetics of English Intonation. Ph.D. dissertation, M.I.T.

Shen, S. (1986) Contrastive Study of Mandarin Chinese and French Interrogative Intonologies. Ph.D. dissertation, University of California, Berkeley.

Tseng, C. (1981) An Acoustic Phonetic Study on Tones in Mandarin Chinese. Ph.D. dissertation, Brown University.

Return to Publications Page.

Copyright (c) 1996 Marjorie K.M. Chan. All rights reserved.
Created: 11 May 1996.
URL: http://people.cohums.ohio-state.edu/chan9/pubn/shen-rev.htm