试论等韵学之原理与内外转之含义

The Thcoretical Basis of the Rhyme-Tables and the Meaning of their Inner/Oufer Distinction

一、等韵学的历史地位二、等韵学的音韵基础三、内外转的含义四、结语一等韵学的历史地位在汉语音韵学的研究中,等韵学是最值得大书特书的辉煌成就。它的重要性不仅在于为中古汉语提供了许多极宝贵的资料,更重要的是它为汉语音韵学的研究提供了极巧妙的新方法,所以有人说等韵学就是中国的理论音韵学,这是极有见地的看法。事实上,有关汉语音韵的任何有意义的研究,不论是上古、中古或现代方言,也都离不开等韵学。在世界语言学史上,这也是我们可以自豪的了不起的成就,只可惜湮没太久,其重要性遂鲜为世人所知。 1.1 要正确地理解等韵学的性质与重要性,我们必须首先确认音韵学研究的真正目标。对一般人来说,研究某一语言(或方言)的语音。当然是说明该语言(或方言)中某些字如何读法;研究古音,也就是想知道某一时代的古人究竟是怎么说话的。

In traditional Chinese linguistic study, there are two types of phonological sources. The first type is rhyme dictionaries and the other type is dengyun(rhyme-tables). From a theoretical point of view, the highlight of Chinese hiStorical lin-guistics is clearly the dengyun tradition which first appeared in the Tang dynasty (618—907) and evolved into maturity in the Song dynasty. These rhyme tables in somewhat varied forms served as the technical basis for phonological study in China in the past, and have been used as the framework for historical reconstruc-tion of the Various periods of the Chinese language by modern linguists. Though they have been the object of numerous philological studies, only a few preliminary attempts have been made to uncover the phonological basis for the design and categorization of these tables. To this author, the highly ingenious design of the rhyme tables cannot possibly be a matter of pure invention. It must have been based upon the phonemic structure of an actual language—in this case, the sound system of one historical period of the Chinese language. Furthermore, the phonological con-trasts revealed by both the rhyme dictionaries and the rhyme-tables are by definition phonemic in nature, while the rhyme-tables, which are based upon such concepts as wuyin 'five types of sounds' , qing 'clear' , zhuo 'muddy' , deng 'division' , she 'rhyme-class' , etc., are really in agreement with the theory of 'distinctive features' in spirit, It follows that a proper understanding of these ancient materials can pro-bably be better acheived through a strictly phonemic approach. There are many puzzles in the dengyun study which can perhaps only be pro-perly explained on the basis of the underlying sound system. One of them is the nei-wai zhuan 'inner/outer turn' . All the rhyme-tables are marked as being either nei 'inner' or wai 'outer' . This seems to be a significant binary contrast, but its meaning has been a mystery for centuries. Many modern scholars have tried very hard to uncover the true meaning of this contrast. Most famous among them is Lo Changpei who, on the basis of Karl-gren' s 'phonetic' reconstruction of the Qieyun, claims that this means roughly the contrast between the higher vowels and the lower vowels. In so doing, he had to move the zhen-she from the outer to the inner, and move the guo-she and the dang-she from the inner to the outer, on the basis of some obscure and questionable versions of rhyme-table books. Lo' s proposal is so attractive that it has been accepted by many people, including myself. However, my more recent research has led me to believe that the high/low (or close/open) contrast of the vowels and the inner/outer contrast of the rhyme tables actually represent two completely different and unre-lated concepts. As a general rule, a word should appear in the same division of a rhyme table as the second word in its fanqie notation, but there are some exceptions, either be-cause of sound changes or arbitrary conventions of the tables. The so-called menfa represents a set of rules for properly locating words of different kinds of exceptions. One rule, called neiwai men 'the inner/outer door' , is concerned with the location of words with retroflex sibilant initials. Words of this kind were all entered in the second division, though the second word in their fanqie notation may appear in either the second or the third division. Reversely, words with an initial of a different kind may have a word of this kind as the second word in their fanqie notation, but these words may also appear either in the second or the third division. Obviously, in each of these two cases, a breach of the general rule mentioned above may occur. In order to help readers locate these words accurately, tables were thus labeled either 'inner' or 'outer' , to the effect that words of this kind in the inner tables must be recognized as third-division words, while those in the outer tables are truly second-division words. Thus, the traditional definition of the inner/outer contrast, though ad hoc, oversimplified, and misleading, is essentially correct, and the traditional classification of the rhyme tables with regard to this distinction should by no means be tampered with. In several previous publications, Ⅰ proposed a four-vowel system within the time-honored framework of Chinese syllable patterns, (C)(M)V(E), as the basis for the design of the rhyme tables. Furthermore, in an effort to buttress Lo' s proposal, and remove any vagueness from it, I claimed that the inner/outer contrast is a reflexion of the strict phonemic Contrast between the single high vowel and the three low vowels in this sound system. With the help of /y/ (palatalization) in the medial slot (M), the divisions of two types Of rhyme tables can be abstracted according to the nature of the syllable finals(see 2.4.2.). This interpretation of the inner/outer contrast, however, has the following difficulties: a) It forces us to reclassify on very shaky ground the rhyme-classes as Lo did. b) It destroys thereby the function of the inner/outer distinction for properly locating words, with retroflex sibilant initials. c) we are forced to combine guo and jia, jiang and dang, merely for the sake of the so-called 'complementary distribution', which actually is incomplete. d) Rhyme-table books 'of later dates, such as the sisheng Dengzi, indeed combine the above classes, as well as geng and zeng, and label them 'the inner/outer mixture' . Lo' s theory in effect destroys the distinction and, therefore, cannot provide a logical explanation for the mixture. e) It misleadsus in our effort to reconstruct an accurate sound system for Middle Chinese and trace its development to modern Mandarin dialects. If we reject Lo' s theory, we shall be under no pressure to modify the tradi-tional version of the inner/outer contrast, and we shall feel no prohibition in recon-structing an inner class with a low vowel, or an outer class with a high vowel. Indeed, r propose that, not only guo and dang, but also zeng, in the inner series should be reconstructed with the back low vowel, while zhen in the outer series should have the high vowel. My abstraction of the syllable finals of the rhyme tables implies one restric-tion: only central vowels can be palatalized. However, a sound system does not have to be so restricted. If we postulate that it is only the result of a sound change, we can then allow palatalized syllables with the back vowel as nucleus in our reconstruction for an earlier period, and explain the restriction on the back vowel by the following rule of assimilation. (?)→a/y____Sometime after the Qieyun, this sound change must have started. Like many other sound changes in Chinese, this one also picked up power gradually, that is, affecting one type of syllables at one time. By the time when rhyme tables were first conceived, it must have affected rhymes in all classes except those in guo, dang, and zeng. These rhymes could not join the rhyme with the same ending but different low vowels to form a single table, because (a) their third-division words had a different vowel, and (b) in the case of dang-jiang and zeng-geng, words with retroflex sibilant initials would clash for the same spot in the second division. Consequently, separate ad hoc tables must be drawn for them, and must be labeled 'inner' . When the sound change represented by the above rule eventually caught up vith them, the two reasons for their separation automatically disappeared Thus, 'inner/outer mixture' became inevitable. The above rule must be an accurate reflexion of the sound change that affected the two rhyme-classes, guo and zeng, which had zero and palatal endings respec-tively, but its application to the dang-she which had velar endings is rather doubtful, because nuclear vowels in Chinese syllables are always primarily conditioned by the ending and secondarily by the medial. Judging by their later development and reading in modern dialects, we can conclude that it was actually the jiang-she that joined the dang-she, when its nucleus changed from /a/to/(?)/ under the influence of the velar endings. Therefore, the above rule must include the restriction that it does not apply to syllables with velar ending, and another rule, (a→c/__η, k) which actually also affected the xiao-she(i. e., syllables with/w/as ending), must be added. The case with the zhen she is more thorny. By the traditional verbal defini-tion, it qualifies perfectly as an inner class, because no words appear in its second division except under the sibilant initials. It must be labeled wai, nevertheless, because these few words in the second division were deliberately set aside as two independent rhymes by the Qieyun compilers, and they rely on each other for fanqie notation. As we know, the Qieyun is very complex. Even after allowing a considerable amount of dialectal mix-up, probably we still have to reconstruct a more complex vowel system for it than the 4-rowel system proposed for the rhyme tables here. It seems that these two rhymes might each represent the residue of a mush larger rhyme of words with a lower high vowel in the pre-Qieyun time, but by the time of the Qieyun, all but the few syllables with retroflex initials changed and joined other rhymes. Before the rhyme-tables, the nucleus of these few remain-ing syllables must have also changed to the high vowel /i/, and must thus be entered into the same table with the other words of the zhen-she. Nevertheless, compilers of the rhyme-tables, who were extremely concerned with the Qieyun, madea special effort to indicate, by labeling the zhen-she 'outer' , that these few syllables form a separate group. This is obviously quite arbitrary, but it has a significant his torical implication.