说长重短轻原则:以汉语和英语为例

不少前人观察到,长音节倾向于有重音(长重),短音节倾向于无重音(短轻),不过反例似乎也很多。比如,根据电子词典CELEX的标注,英语有7401个单语素词,共含11947个音节,其中长重的反例(长却轻)占18%(如routine[ru:][’ti:n]①“惯例”的第一音节、cactus[’kæk][təs]“仙人掌”的第二音节),短轻的反例(短却重)占12%(如happy[’hæ][pi]“幸福”和city[’sɪ][ti]“城市”的第一音节)。本文以汉语和英语为对象,全面考察各种反例。汉语的事实相对清楚,长重短轻主要反映在普通音节和轻声音节的区别上(Chao,1968:38;林茂灿、颜景助,1980,1990;罗常培、王均,1981:135)。在英语中,反例主要源于对英语音节划分及重音标注的片面理解。关于音节划分,前人有三个错误假设(Kahn,1976;Jensen,2000):第一,音节划分先于重音指派;第二,音节划分基于声母优先;第三,所有辅音都在音节之内。本文认为:音节划分和重音指派是同步完成的,无先后之分(Duanmu,2019);而且,词尾的辅音不一定都在音节之内,有的可以处于音节之外(extrametrical,Hayes,1980;Halle&Vergnaud,1987;Duanmu,2008)。关于重音标注,本文同意Chomsky&Halle(1968)的观点,即非弱化元音皆有重音(主重或次重)。根据正确的音节划分和重音标注方法,反例最多只占1%。因此,长重短轻在汉语和英语中都有充分的证据,不只是一种倾向,而是音系理论中的一条重要规律。

It has often been observed(such as Jespersen,1909;Prokosch,1939;Fudge,1969;Hoard,1971;Bailey,1978;Murray&Vennemann,1983;Kager,1989;Prince,1992;Wells,1990;Hammond,1999;Duanmu,2008)that(a)long syllables tend to attract stress and(b)short syllables do not,which I shall call the LengthStress Correspondence.However,counter-examples seem easy to find.For example,according to the CELEX lexicon(Baayen et al.,1995),English has 7401 monomorphemic words,with a total of 11947 syllables,of which 18%are long yet unstressed(such as the first syllable in routine[ru:][’ti:n]and the second syllable in cactus[’kæk][təs]),and 12%are short yet stressed(such as the first syllable in happy[’hæ][pi]and city[’sɪ][ti]).This study offers a close examination of counter-examples to the Length-Stress Correspondence in Chinese and English.The case in Chinese is over straightforward:as previous scholars have observed,syllables with a lexical tone are long and stressed,and syllables without a lexical tone are short and unstressed(such as Chao,1968:38;Lin&Yan,1980;1990;Luo&Wang,1981:135).In the case of English,most prominent examples are the result of unsupported assumptions concerning syllabification and stress annotation.With regard to syllabification,I argue against three common assumptions(such as Kahn,1976;Jensen,2000):(i)syllabification precedes stress assignment,(ii)syllabification is based on Maximal Onset,and(iii)all consonants are inside a syllable.Instead,I argue that syllabification and stress assignment is achieved simultaneously(Duanmu,2019),where syllabification is based on word-edge phonotactics(the Law of Initials and the Law of Finals,Vennemann,1988).In addition,not all word-final consonants need to be inside a syllable;instead,some can be extrametrical(Hayes,1980;Halle&Vergnaud,1987;Duanmu,2008).In the new perspective,the analysis of happy is[’hæp][i],and that of city is[’sɪt][i],where all stressed syllables are long.In addition,the analysis of cactus is[’kæk][tə],where is extrametrical and the unstressed second syllable is short.With regard to stress annotation,I follow Chomsky&Halle(1968)that all syllables with an unreduced vowel have stress(primary or secondary).Thus,the analysis of routine is not[ru:][’ti:n]but[,ru:][’ti:n],where the first syllable has secondary stress.An exhaustive examination of the 11947 syllables in the 7401 monomorphemic English words show that counterexamples constitute no more than 1%.Therefore,there is strong evidence in both English and Chinese for the Length-Stress Correspondence.It is not just a tendency but an important generalization in phonology.