LINGUIST List 8.1024

Mon Jul 7 1997

Review: Hung: Prosody and Acquisition

Editor for this issue: Andrew Carnie <>

What follows is another discussion note contributed to our Book Discussion Forum. We expect these discussions to be informal and interactive; and the author of the book discussed is cordially invited to join in. If you are interested in leading a book discussion, look for books announced on LINGUIST as "available for discussion." (This means that the publisher has sent us a review copy.) Then contact Andrew Carnie at

Message 1: Book Review : Feng-Sheng Hung, Prosody and the Acquisition ...

Date: Mon, 07 Jul 1997 17:17:12 +0800
From: David Henry Deterding <>
Subject: Book Review : Feng-Sheng Hung, Prosody and the Acquisition ...

Hung, Feng-Sheng, (1996). Prosody and the Acquisition of Grammatical Morphemes in Chinese Languages. Indiana University Linguistics Club: Bloomington, Indiana. 150 pages.

Reviewed by David Deterding <>


In this book, Hung studies the language development of 6 children, 3 of whom are acquiring the variety of Mandarin Chinese spoken in Taiwan (hereafter TMC) as their first language, and the other 3 of whom are acquiring Taiwanese, the variety of Hokkien that is a native language for most Taiwanese. In particular, he is concerned with the order and the ease with which these children learn to use grammatical morphemes in their respective language.

This research has considerable significance because of differences between the two languages. Though they are morphologically and syntactically similar, they have marked phonological differences, particularly as the most common grammatical morphemes in Mandarin have a schwa and also have a neutral tone, while those of Taiwanese have a full vowel and a full tone. Hung claims that the frequent use of weak syllables with a neutral tone makes Mandarin a stress-timed language, with a tendency for a trochaic foot consisting of a strong syllable followed by a weak syllable, while the absence of weak syllables makes Taiwanese a syllable-timed language; so his study provides an insight into the different acquisition paths for these two kinds of rhythm while keeping morphology and syntax constant.

Hung studies the children longitudinally, visiting each child 6 times between the ages of about 1.5 to 2.5 years. He obtains three kinds of data from them: spontaneous speech, imitation of selected phrases, and listening comprehension. Hung acknowledges serious problems with the listening comprehension exercises, so he relegates those data to the status of a pilot study; but he finds consistent and convincing evidence from the first two kinds of data to show that grammatical morphemes are more quickly and easily mastered in Mandarin than Taiwanese.

Hong concludes from this that, because Mandarin is stress-timed, there is a strong-weak template used by the children, and the grammatical morphemes are learned early because they constitute an essential part of this template. In contrast, because Taiwanese is syllable-timed, there is no metrical foot providing a rhythmic template, so all syllables are of equal weight, and infants concentrate on the semantically richer content morphemes. This explains why the Taiwanese-learning infants have a greater tendency to omit grammatical morphemes, even though one might assume that the grammatical morphemes of Taiwanese are phonetically more salient than their Mandarin equivalents.


This book is well-written and clearly presented, and it has some very interesting data that contribute valuable insights to our understanding of infants' acquisition of language.

I have two main reservations that I will discuss separately: the phonological structure of the grammatical morphemes; and the rhythmic nature of the two languages. Though these reservations constitute questions that should be considered further, they do not necessarily invalidate Hung's conclusions, and they certainly do not undermine the value of his fascinating and well-conducted study.

Phonological Structure of the Grammatical Morphemes

The morphemes on which Hung concentrates are as follows:

Mandarin Chinese

1. ge Classifier e.g. si-ge ren "four CL person" -- "4 people" 2. de Nominalizer e.g. da-de shu "big NOM book" -- "big book" 3. de Genitive e.g. wo-de shu "I GEN book" -- "my book" 4. zi Diminutive e.g. ya-zi "duck DIM" -- "duck" 5. zhe Durative e.g. kan-zhe wo "look DUR me" -- "looking at me" 6. le Completion e.g. ta chi-le "he eat COM" -- "he has eaten" 7. men Plural e.g. wo-men "I PLUR" -- "we"

All of these have a schwa, except for no. 4, which has a high vowel. And all have a neutral tone.


1. e Classifier e.g. si-e lang "four CL person" -- "4 people" 2. e Nominalizer e.g. toa-e chheh "big NOM book" -- "big book" 3. e Genitive e.g. goa-e chheh "I GEN book" -- "my book" 4. a Diminutive e.g. kau-a "dog DIM" -- "dog" 5. leh Durative e.g. che-leh chiah "sit DUR eat"--"eat while sitting" 6. teh Progressive e.g. goa teh khun "I PROG sleep"--"I am sleeping" 7. ti Locative e.g. goa di chia "I LOC here" -- "I am here"

All of these have a full vowel, and all have a full tone.

If we consider these morphemes from the two languages, only the first 5 are directly comparable. (The direct equivalent of both 6 and 7 from the Taiwanese list is 'zai' in Mandarin, but as this has a full tone and a full vowel, it was not included in the imitation tasks.) Of the first five, the durative marker was never produced spontaneously by the infants of either language group, so the comparison rests mostly on the first four.

We should note that in all these first four, the Taiwanese morpheme consists just of a vowel, but the Mandarin consists of a consonant followed by a vowel. One wonders to what extent this CV syllable shape contributes to the Mandarin grammatical morphemes being easier to produce and more resistant to omission. Given that so many syllables in both languages end in a vowel, one wonders to what extent these Taiwanese grammatical morphemes remain perceptually salient for an infant: even though the Taiwanese grammatical morphemes do indeed have full vowels and full tones, one wonders whether they may not have a greater tendency to merge into the previous syllable in the perception of infants.

Rhythmic Differences between Mandarin and Taiwanese

Abercrombie (1967:97) insisted that all languages can be classified as either stress-timed or syllable-timed. However, more recently, many linguists have suggested that there is a continuum between stress-timing and syllable-timing (e.g. Dauer, 1983). In this respect, I wonder how the Mandarin Chinese spoken in Taiwan (TMC) should be described.

It is true that the Mandarin Chinese spoken in Beijing (BMC) is often described as stress-timed. One of the contributory factors to this rhythmic classification is the wide use of the neutral tone, not just in the grammatical morphemes discussed above, but also in the second syllable of compounds such as 'zhi-dao' ("to know") and 'yi-sheng' ("doctor"), with the result that there tends to be an alternation of strong and weak syllables. As Hung points out (p.19), one of the characteristics of TMC is that the neutral tone is avoided in compounds, so that both 'zhi-dao' and 'yi-sheng' would usually have a full tone on the second syllable. This being the case, one might ask how stress-timed TMC is -- surely it is somewhere between BMC and Taiwanese on the continuum, and one should be careful about classifying it as stress-timed.


In addition to the two main reservations that I have discussed above, there are a few less crucial issues that I raise here. They do not interfere much with Hung's main conclusion, that differences in the rhythm of Mandarin and Taiwanese offer different acquisition paths for infants, but they do raise some interesting questions that might be worth discussing.

The issues I will discuss are: the nature of tone sandhi; the semantic matching exercise; and the contribution of tone sandhi to syntactic parsing.

Tone Sandhi in Taiwanese

Hung (p.14) lists the seven citation tones of Taiwanese, giving them their traditional numbers. Then (p.16) he lists the rules of tone sandhi, showing how each citation tone gets changed when it is in non-final position. (The third tone, the Low tone, is omitted from this list of tone-sandhi rules on p.16. I assume this is an oversight, as it is usually claimed that the low tone becomes a falling tone after sandhi (Peng, 1994).)

In listing the rules of tone sandhi, Hung shows the pitch contour after sandhi, but he does not state whether the tone "becomes" another tone or not. In most cases, one can interpret the modified pitch contour as one of the original tones; but in one case, that for tone 2, the falling tone, the new pitch contour is HM which is not one of the original tones. Other linguists have assumed that tone 2 becomes tone 1, which is the high level tone. So I am not sure of Hung's position: do tones adopt their own, possibly unique, modified contour, or do they change into a different tone?

The issue is further unclear because later on, when discussing children's errors, Hung does suggest that tone 2 should become tone 1 after tone sandhi (p.94) and that tone 7 should become tone 2 after tone sandhi (p.95), (even though his own rules (p.16) suggest that tone 7 should adopt a low pitch contour, which represents tone 3).

So, do tones "become" other tones after sandhi? Or do they adopt their own modified pitch contours?

The Semantic Matching Exercise

To determine whether his subjects could understand the semantic role of the grammatical morphemes, Hung presented them with two sentences that differed only in the presence or absence of one of the morphemes and asked them to match them with pictures. This is problematic, as in almost every case, one of the sentences is rather strange, and it is hardly surprising that the infants were unable to do the task successfully. As an example, consider the following Mandarin pair:

(a) zhe-li you liang-ge dan here have two CL egg -- "there are two eggs here"

(b) zhe-li you liang dan here have two load -- "there are two loads here"

The problem here is that, while sentence (a) is a perfectly common-place sentence, in (b) 'liang dan' is rather more advanced vocabulary which two-year-old infants might well not be familiar with.

Another example is:

(a) zhe shi mi-feng de fang-zi this is bee GEN house -- "this is a house for bees"

(b) zhe shi mi-feng ... fang-zi this is bee house -- "these are bees and this is a house"

The trouble here is that, for it to be natural, sentence (b) would usually have a word meaning "and", such as 'gen' or 'he' between the two elements in the list. I tried this one on my wife, who is a native speaker of both TMC and Taiwanese: she only reluctantly accepted sentence (b) as grammatical, and she found it difficult to match it against the picture of bees and a house. No wonder Hung found his subjects uncooperative at this task!

To his credit, Hung accepts the problems with these semantic exercises, and places little emphasis on the results. One wonders if it is really possible to devise an exercise for Chinese where two sentences differ only in the presence or absence of a grammatical morpheme and both are equally meaningful for two-year-old infants.

Tone Sandhi and Syntactic Parsing

The sentences just discussed do suggest one thing further that might be worth considering. If we look at the Taiwanese equivalent for the second pair of sentences, we have:

(a) chia u phang e chhu here have bee GEN house -- "There is a house for bees here"

(b) chia u phang ... chhu here have bee house -- "There are bees and there is a house."

There is an interesting difference between (a) and (b): 'phang' has a high tone, tone 1, which in (a), as a result of tone sandhi, would adopt a mid contour (or become tone 7); but in (b), because 'phang' and 'chhu' are two separate noun phrases, 'phang' would retain its citation tone and not undergo tone sandhi. This means that, in some situations in Taiwanese, the grammatical morphemes may be redundant, as the following would already be distinct (though not necessarily fully grammatical):

(a) chia u phang7 chhu (b) chia u phang1 chhu

Hung says (p.86) that both groups of infants learn the correct tone system early. One wonders what help the extensive tone sandhi rules of Taiwanese might provide infants in parsing a sentence syntactically.


Abercrombie, D. (1967). Elements of General Phonetics. Edinburgh University Press: Edinburgh.

Dauer, R. M. (1983). 'Stress timing and syllable-timing reanalyzed', Journal of Phonetics, 11, 51-62.

Peng, Shu-hui (1994). 'Effects of prosodic position and tonal context on Taiwanese Tones'. Ohio State University Working Papers in Linguistics, 44, 166-190.

Reviewer: David Deterding National Institute of Education, Nanyang Technological University 469 Bukit Timah Road Singapore 259756