Journal of Chinese Linguistics Vol.12 – 1984

Abstract 摘要
It is generally believed that every stressed syllable in Mandarin has a tone and that when a syllable is weak-stressed, the tone becomes neutralized. However, there are stressed neutral-toned syllables as well as weak-stressed toned syllables. Furthermore, while the neutral tone in certain forms (e.g. jiějie ‘elder sister’, yǐzi ‘chair’, etc.) does not trigger a sandhi change, the neutral tone in other forms (e.g. xiáojie ‘unmarried woman’, zóuzou ‘take a walk’, etc.) does. Grammatical approaches to this problem in previous works are found to be inadequate. This paper proposes the following explanation: Neutral tone and weak stress are two separate entities, hence do not interact in terms of cause and effect. In Mandarin there are certain grammatical morphemes which lack tone at both the underlying and the surface level. These are usually weak-stressed. But even when they are normal-stressed, they do not trigger sandhi changes. This category includes suffixes, particles and reduplicated kinship reference terms. In Mandarin, tones are phonemic; stresses are not. Toned syllables may fluctuate between normal and weak stress in accordance with various factors (e.g. sentence intonation, familiarity of the term, emotion, etc.). When a toned syllable is weak-stressed, its duration becomes shorter, and its tonal characteristics less distinct. This is how it is confused with truly toneless syllables. But a weak-stressed toned syllable is still viable in triggering sandhi changes. Indeed, sandhi changes indicate the speaker’s anticipation of a toned syllable, regardless of the degree of tonal retention for that syllable after it survives weak stress. Dictionary listings of the lexical neutral tone items are simply chaotic. The present study examined the tonal readings of 760 disyllabic lexical items as recorded in three sources. The rate of inconsistency or optionality was 72.2%. In the case where a shift between a neutral tone reading and a full tone reading was supposed to signal a change in meaning, 48 out of 55 items (i.e. 87.3%) nullified the claimed distinction. Such inconsistencies not only exist in dictionaries, but also in the intuitions of native speakers. The extensive inconsistencies indeed pertain to fluctuations between normal stress and weak stress for toned syllables rather than between toned syllables and toneless ones. The implication of the above findings is that existing dictionaries need revision with respect to the tonal annotations: All the lexical items marked with the neutral tone should be given their full tones. As to the stress patterns for such toned syllables, they are non-phonemic and variable, hence need not and cannot be prescribed.

国语或普通话之轻声 (neutral tone) 与轻音 (weak stress) 一向被视为互为因果且不可分割。然轻声与重音有时并存，全声 (full tone) 与轻音亦可同在。此外同是所谓轻声词而变调现象却不一致。而类如‘词界’之区分不足以解释其间之差异。本文提出如下解释：国语中某些词素本无声调；此类词素多作轻读，然亦可作重读。而不论轻重读皆不致引起前字之变调。此类词素包括虚词的，了，着，们，子，吗，吧等及亲属称谓之重复词（如姐姐）。国语之轻重音无辩义作用。而带有声调（全声）之词可能因句子之语调，该词之熟悉性，情感，等等因素而于轻重读之间有所变化。带有声调之词轻读时音长变短，调值亦有不同程度之淡化。然不论淡化至何种程度，仍足以引起前字之变调。动词之重复词（如走走），方位词里，作宾语之人称代词，以及语法上无规可循之所谓“轻声词”皆为常作轻读之有声（全声）音节。诸词典内对轻声词之注音往往有所不同。作者以五年之间出版之三本词典内所记之760条双音节词作为比较。其中72.2％之词条各词典之注音不一致或同一词典内有同义之轻重二读。至于所谓轻重音二读作不同解释之论，于55条双音词中之48条（即87.3％）内，此一区别不能成立。声调有辨义作用；轻重读则无。诸词典内双音节词所谓轻声与全声之混乱实为全声音节轻重读之改变。