上古音芻議

Studies on Archaic Chinese Phonology

本文論漢語上古音聲韻系統，大體針對李方桂先生〈上古音研究〉有所商略。壹節論單一聲母及介音，主張：（1）取消圓唇聲母，仍以開合兩分。（2）據中古四個等，區分上古為甲、乙、丙、丁四韻類，甲類無介音，其餘分別具r、j、i介音，取消rj複合介者的構擬。（3）聲母計有幫p、滂ph、並bh、明m、端t、透th、定dh、泥n、來l、見k、溪kh、群gh、疑ŋ、精ts、清tsh、從dzh、心s、邪z、影ʔ、曉h、匣ɦ二十一個，並四類韻俱全。照、穿、床之二等出於精、清、從，三等絕大部分出於帶s或z複母（或詞頭）的端、透、定，小部分亦出於精、清、從，審禪則無論為二等為三等，均分別出於心、邪。日母出於帶s複母（或詞頭）的泥，喻四則本讀zɦ複母。貳節論上古韻部及擬音，主張：（1）古韻以分二十二部最為合適，陰與入之分不若其合。（2）受方言影響，音有正有變，不必凡字於周代只有隸屬一個韻部的讀音，亦不必任何異音都要求其音變條件。（3）陰聲字不具塞音尾，歌部則獨具韻尾r，其情形如同國語之兒化。（4）宵部陰聲原是談部的陰聲，其後始脫離了陰陽關係，其入聲亦自葉部分出；侵緝原亦有陰聲，今則混入了幽。（5）舉出四十餘組自幽部轉讀入微文部字例，證實上古陰聲本不具g、d韻尾。

This paper is concerned with the reconstruction of Archaic Chinese phonology, mainly with that proposed by late professor Fang-kuei Li in the 1970s. Part one discusses initials and medials, and suggests that: (1) there is no need for labial-velars, instead the kai-he dichotomy should be maintained as a part of rime; (2) there are four classes of rimes: A, B, C, and D; while A rimes lack any medial, B, C, D ones get -*r-, -*j-, -*i- as their medial element respectively; and it turns out that the combination of -*rj- could be removed; (3) there are 21 simple initials, namely: *p-, *ph-, *bh-, *m-, *t-, *th-, *dh-, *n-, *l-, *k-, *kh-, *gh-, *ŋ-, *ts-, *tsh-, *dzh-, *s-, *z-, *ʔ-, *h-, *ɦ-, and they can appear with all classes of rimes. The alveolo-palatal affricates (Zhao, Chuan, Chuang) with II-grade in Ancient Chinese came from *ts-, *tsh-, *s-; while the same ones with III-grade derived from *t-, *th-, *dh- with prefixes *s- or *z- for the most part and from simple *ts-, *tsh-, *dzh- also in a few cases. The alveolo-palatal fricatives (Shen, Chan) came from *s-, *z- respectively; and the corresponding nasal (Zhi) derived again from prefix *s- plus dental nasal *n. The zero initial (Yu-si) of Ancient Chinese originated from consonant cluster *zɦ-. Part two deals with the rime categories and concludes that: (1) the 22-category system is preferable; (2) by considering dialect interference, the requirement that each single word should be classified into only one rime-category is abandoned; (3) words belonging to Yin-sheng rime-categories bear no consonant endings with exceptions of those from Ge category, which has -*r as its ending, a case quite similar to the -r suffixation of Mandarin Chinese; (4) during the pre-Archaic period, words of Xiao category, both Yin-sheng and Ru-sheng, are parts of Tan category; similarly, the original corresponding Yin-sheng words of Qin and Qi categories also merged with You category in Archaic Chinese; (5) the claim that endings -*g or -·*d do not exist in any Yin-sheng rimes can be justified again by more than forty pieces of evidence exemplified in this paper.