Shanghai International Studies University
The speech sound, being the most important medium between our physical body and linguistic mind, is one of human beings’ oldest objects of study. In different cultures, it has been understood in different ways, depending on the role it plays in social life (e.g. in religious activities), the technologies available, the dominant philosophy, and various other factors. If we take the position that each writing system is a preliminary analysis of the speech sound, even more varieties may be considered.
The above having been said, one might assume that the Chinese, with a long history, an autonomous culture, and a unique writing system, must have a rather independent tradition in the study of the speech sound. Indeed, this is the belief of many Chinese scholars as well, to whom the fact that speech sound study (yinyunxue 音韻學, literally “pronunciation and rhyme study”) makes one of the three branches of the “basic learning” (xiaoxue 小學, somewhat similar to the Western trivium) in traditional Chinese scholarship and played a leading part in its culmination during the Qing Dynasty (1616-1911, the last Chinese dynasty) is enough to support the view.
In this post, I will show that the above view is at least partly wrong. At the heart of the speech-sound branch of the “basic learning,” a tension between the Chinese and the Indian approach has existed since the 2nd century A.D. Moreover, as the branch started to be modernised at the beginning of the 20th century, since which time it has been customarily called “historical Chinese phonology” (hanyu lishi yinyunxue 漢語歷史音韻學), the tension has been further complicated by the introduction of the Western approach. Throughout the process, the Chinese approach has never been really dominant, but it has never been dispensable, either. This poses a problem for those who argue for the independence of traditional Chinese scholarship, because “speech sound study” is regarded as a core component of the scholarship (in much the same way as philology is the basis for classical studies). On the other hand, this is also problematic for the modern linguists, who often feel that the modernisation of historical Chinese phonology is not enough.
1. The Chinese approach
The earliest systematic studies of the Chinese speech sound, according to historical records, are the Sound Categories (Sheng Lei 聲類) by Li Deng 李登 of the Three Kingdoms Period (220-280), and the Book of Rhymes (Yun Ji 韻集) by Lü Jing 呂靜 of the Western Jin Dynasty (266-316). However, these books were lost long ago, and all we know about them today is that they grouped Chinese characters into five sound categories, using the “five sounds” (wuyin 五音, i.e. gong 宮, shang 商, jue 角, zhi 徵, and yu 羽) from ancient music theory as their standard of reference. As an idea deeply rooted in the traditional Chinese philosophy about the “five elements” (wuxing 五行, i.e. “wood” mu 木, “fire” huo 火, “earth” tu 土, “metal” jin 金, and “water” shui 水), the “five sounds” is applicable to all the sounds in the world. Today, there are still stories of doctors who can judge the state of a patient by listening to him/her speaking so as to tell what element he/she lacks. But as far as the systematic application of this (and only this) philosophy to the Chinese speech sound is concerned, the above mentioned works seem to be the end. For at round their time, Indian linguistics started to be introduced into China with the entry of Buddhism. After that, although the influence of traditional Chinese philosophy was still there, and the terms it provided remained in use for a long time (some even till this day), the framework for the study of the Chinese speech sound became basically Indian.
Of course, the Chinese approach to the study of the speech sound did not really die. For the dominance of the Chinese writing system meant that a pre-theoretic Chinese linguistic thinking was always available – as a help as well as a constraint. At least on the surface, everything had to appear in the form of characters. Moreover, the study of the Chinese sound was always centred around character pronunciation: except for tones, features above the syllable received little attention.
2. The Indian approach
Indian linguistics was introduced into China around the 2nd century A.D., at least no later than the Eastern Han Dynasty (25-220). Its influence lasted till the end of the Song Dynasty (960-1279). In India, linguistics was itself a branch of a “basic learning” system called pañcavidyā (“the five branches of knowledge”, including śabdavidyā “linguistics”, hetuvidyā “logic”, cikitsāvidyā “medicine”, śilakarmasthānavidyā “arts and crafts”, and adhyātmavidyā “spiritual doctrines”), and was further divided into śikṣā “pronunciation”, chandas “meter”, vyākaraṇa “grammar”, and nirukta “etymology.” Curiously, it seems that only śikṣā had much influence on language study in China, though the influence surely went very deep.
The most important idea the Chinese learnt from Indian linguistics was that syllables could be further divided into an “initial” (shengmu 聲母, shengniu 聲紐, sheng 聲, niu 紐, zimu 字母, mu 母) and a “rhyme” (yunmu 韻母, yun 韻) (which includes the tone). Based on this idea, a technique called fanqie 反切 was invented for the analysis of character pronunciation, in which the pronunciation of one character was represented by two others that shared the same initial and rhyme with it respectively. For example, the pronunciation of 恪 (then pronounced [khɑk]) could be represented by 苦 (then pronounced [khuo]) and 各 (then pronounced [kɑk]); anyone who was not familiar with 恪 could get its pronunciation easily by putting the initial of 苦 and the rhyme of 各 together. This technique, however, had some obvious disadvantages: (1) One had to know the pronunciation of a number of basic characters to get the pronunciation of those more difficult ones. (2) Moreover, there was no agreement on which character to use for a particular initial or rhyme. Systematic presentations of the rhymes occurred later in the rhyme books; and an alphabet-like list of all the initials appeared even later during the Tang Dynasty (618-907). (3) With the passing of time, the same initial or rhyme may change into different sounds in different characters, turning the representation invalid. This, of course, was more a disadvantage for the later generations.
After the invention of fanque, the concept of the tone appeared, and a four-tone theory was proposed during the Southern and Northern Dynasties (420-589) to describe the tones of the Chinese syllables as ping 平 “flat”, shang 上 “rising”, qu 去 “falling”, or ru 入 “checked” (i.e. with a stop final) (we do not know the exact phonetic values of these tones). There is still some disagreement today on the actual origin of the theory, but it is generally accepted that the tone theory in India must have played a significant role in its appearance. To quote a well-known historian of Chinese linguistics on the issue: “the ‘high’, ‘flat’, ‘circumflex’ and ‘slanted’ tones used in the Buddhist monks’ recitation must have been related to the four tones in some way.” (He Jiuying 何九盈, A History of Ancient Chinese Linguistics 中國古代語言學史, Beijing 北京: Peking University Press 北京大學出版社, 2006. p. 115.)
The invention of fanqie and the discovery of the four tones enabled the compilation of the so-called “rhyme books” (yunshu 韻書). Generally, people were much more interested in rhymes than in initials because rhymes were important for verse composition, which was, for a long time, part of the national “imperial examinations” (keju 科舉) held to select government officials. As required by the rules of rhyming, any two rhyming syllables must have the same tone, the same main vowel, and the same final (if there is one), but not necessarily the same “medial” (jieyin 介音, i.e. anything between the initial and the main vowel, if there is one). For example, a character pronounced [tuan] (flat tone) would have the following phonological structure:
Thus, it could rhyme with [luan] (flat tone), [lan] (flat tone), and [lian] (flat tone), but not [luen] (flat, rising, or falling tone), [lua] (flat, rising or falling tone), [luak] (checked tone), or [luan] (rising or falling tone). Hence, a rhyme book must specify the tone and the “main vowel (+final)” part for the pronunciation of each character. Those characters which have the same tone and “main vowel (+final)” will form a “rhyming group” (yun 韻). Usually, a rhyme book would also divide each rhyming group into different “small rhyming groups” (xiaoyun 小韻), in which characters of exactly the same pronunciation are put together.
The most famous and authoritative of all rhyming books is the Qieyun 切韻 by Lu Fayan 陸法言 of the Sui Dynasty (581-619). The following is a page from this classical work:
We read from right to left, focusing on the highlighted parts. Thus, we have 十一○唐 first, which means that this is the 11th rhyming group of the present book (Qieyun 切韻 is divided into five books, of which this is the second), and that under this rhyming group, the first small rhyming group starts with the character 唐 (then pronounced [dɑŋ] in flat tone). At the end of the entry for唐 is its fanqie followed immediately by a note on the number of characters that belong to this particular small rhyming group (i.e. how many characters share the pronunciation of 唐). This same pattern of explanation is repeated for the second small rhyming group starting with 郎, and the third small rhyming group starting with 當. Surely, 唐, 郎, and 當 must rhyme with each other because they all belong to the same rhyming group. The interesting thing is: if we look at their fanqie in detail, we will see that the rhyme of 唐 is represented by 郎, the rhyme of 郎 by 當, and the rhyme of 當 by 郎. This is exactly what I meant when I said there was no agreement on which character to use for a particular initial or rhyme.
In the study of the rhymes and initials, there seems to have been an interesting division of labour between the Confucian scholars and the Buddhists: while the former were mostly interested in rhymes for such practical purposes as verse composition and preparation for examinations, the latter undertook the work on the initials. As previously mentioned, an alphabet-like list of initials was made in the Tang Dynasty, apparently by a monk called Shouwen 守溫. It is believed that this list was the forerunner of a better-known table of initials from the Song Dynasty, as shown below:
I have included a reconstructed Song Dynasty sound value for each of the initials represented, so that we can clearly see the connection of the table (especially in its layout) to the Sanskrit consonant alphabet:
By now, both the rhymes and the initials seem to have been sorted out in a framework established on a basically Indian foundation. But there were still some blurred areas. For example, the medials, since they are neither relevant to rhyming nor a part of the initials, had never been singled out for analysis; on the other hand, though the initials had been listed out, they were never specified in the rhyme books and differentiated from the medials as a distinct source of the difference between small rhyming groups. In the resolution of these issues, Indian linguistics again served as a guide. Imitating the spelling exercises in the teaching of the Sanskrit alphabet, some scholars started to make the so-called “rhyme tables”, of which the best-known (and also the oldest we now have) is the Yun Jing 韻鏡 (“Mirror of Rhymes”) by an anonymous author. The following is a page from this work:
One needs to know some technical details to fully understand the table, but a basic point is: this table brings the rhymes (in which the Confucian scholars were most interested) and the initials (on which the Buddhists worked most) together to show how they are combined to give syllables in Chinese. The rhymes are represented in the horizontal lines (the four large sections corresponding to the four tones), the initials in the vertical lines, and the small circles are possible pronunciations with no corresponding characters. In this way, the medials and the initials (which were more or less mixed up in the rhyme books) became immediately distinct.
There was still more work to do to fully clarify the phonological system of Chinese. For example, the rhymes in the above table were taken directly from the Qieyun 切韻 of the Sui Dynasty, but the initials listed may not be an actual representation of the initials of that same period. Many of these issues were resolved by scholars of the Qing Dynasty. I will not say more about their work since there was no more direct influence from Indian linguistics at their time. But undoubtedly, their achievements could not have been possible without the initial influence from India.
3. Tension between the Chinese and the Indian approach
Ever since the introduction of Indian linguistics into China, there has been a strong desire among Chinese scholars to make the Indian techniques, which they found so useful, more Chinese. Most often, they tried to translate Indian ideas by using traditional Chinese terms. For example, the “five sounds” were directly assigned to the five places of articulation (i.e. chun 唇 “lips”, she 舌 “tongue”, ya 牙 “back teeth”, chi 齒 “front teeth”, and hou 喉 “throat”), together with the “five elements”, the “five directions” (i.e. dong 東 “east”, nan 南 “south”, xi 西 “west”, bei 北 “north”, and zhong 中 “middle”), and so on. However, there was little consistency among scholars regarding the phonetic relevance of these terms; they only made the description of speech sounds more chaotic.
While direct translation caused serious confusion, some scholars went further to make the descriptive framework Chinese by giving it a Chinese philosophical basis. The most famous and influential of these scholars was Shao Yong 邵雍 (1011-1077) of the Song Dynasty. As a great master of traditional knowledge on yin, yang, the “five elements”, and the secret meanings of numbers, Shao tried to incorporate the basic notions in speech sound studies into his grand picture of the world in his masterpiece Huangji Jingshi 皇極經世, assigning traditional philosophical values to initials, finals, tones and their interactions. But in doing so, he was so obsessed with matching the phonetic notions to certain “perfect numbers” that he mystified and fossilised the notions instead of clarifying them with a plausible Chinese explanation. For example, he regarded the four tones and the four “divisions” (deng 等) in the rhyme table as determined by the four “images” (xiang 象) of the universe, but once these numbers were fixed, any “imperfect number” generated by the phonetic framework became inexplicable if not “altered” by force to suit this new philosophy. Under the influence of Shao, many scholars took to the philosophy of numbers as a key to understanding and improving the study of the speech sound, but none of them managed to build a philosophy that did not turn out to impose “perfect numbers” on their objects of study.
In a sense, what scholars like Shao did was perfectly understandable. After all, they had to depend on their traditional philosophy to comprehend the world, and if their body, life, and environment were all perfectly explicable in that philosophy and the many branches of knowledge it gave rise to (e.g. medicine, agriculture, architecture, politics, art, music, and literature), why should speech sounds be an exception? But the problem is: although this seemed to be a minor extension of a philosophy to cover a small part of the world whose majority it had already explained very well, under the surface, it was translation, in which no literal equivalence could be guaranteed, and definitely no transfer of grammar from one language to the other. In the subsuming of the Indian grammar under the Chinese, it turned out that what was finally subsumed was the Chinese. The various Chinese terms that remained in speech sound study, for example, all lost their original philosophical connotations: yin 陰 and yang 陽 came to mean “with no finals” and “with nasal finals” for the rhyme, or “in a syllable with a voiceless initial” and “in a syllable with a voiced initial” for the tone; qing 清 and zhuo 濁 came to mean simply “voiceless” and “voiced”.
Of course, the Indian approach was not unaltered, either. The most significant consequence of the approach assuming a Chinese appearance was probably the loss (or at least serious decrease) of its ability to represent phonetic values. During the 15 centuries after its entry into China, Chinese scholars never came up with a real phonetic alphabet for their language. Such a thing would have to wait till the coming of Western linguistics.
4. The Western approach
During the Qing Dynasty, studies of the Chinese speech sound reached a historical height, with the phonological systems of the language at different periods, especially Old Chinese, gradually disclosed. But precisely at this time, a bottleneck was encountered, and it seemed that no further progress was possible if there could not be a breakthrough. The situation was most clear in the case of the famous scholar Duan Yucai 段玉裁 (1735-1815):
Duan Maotang 段懋堂 (i.e. Duan Yucai 段玉裁) concluded from his rigorous study that in the Chinese of the time of the Book of Poems 詩經, a strict distinction existed between the zhi 支, the zhi 脂 and the zhi 之 rhyme. However, he was not able to tell the phonetic nature of this distinction. In a letter he wrote to Jiang Jinsan 江晉三 when he was old, he sighed: “Can you find out with certainty how the three rhymes were pronounced differently? I am old, and would die content if I could know the answer.” (古韻支、脂、之三部《三百篇》分用，段懋堂考之甚明，而不能讀為三種音，晚年以書問江晉三云：“足下能確知其所以分為三乎？僕老耄，倘得聞而死，豈非大幸？”)
Chen Li 陳澧 (1810-1882), A Study of Qieyun 切韻考, Beijing 北京: Zhongguo Shudian 中國書店, Vol. 6, 13-14, 1984.
Obviously, the bottleneck was the reconstruction of sound values. While the Indian techniques ensured the sorting out of the contrasts between different sounds, when it came to the phonetic nature of these contrasts, these techniques, assuming the form of Chinese characters, were no longer helpful. This shows that the tension between the Indian and the Chinese approach in traditional Chinese linguistics actually reached a climax in the Qing scholars.
It was at this climactic point that Western linguistics started to exert overwhelming influence. As far as the study of the speech sound is concerned, the Western approach brought in two important things: the phonetic alphabet, which was a powerful tool for the description of the speech sound, and the idea that ancient and contemporary pronunciations of a language are systematically related. The answer that Duan Yucai longed so much for, for instance, was finally given by the Swedish scholar Bernhard Karlgren (1889-1978), who reconstructed Middle and Old Chinese using the Western approach (though his reconstruction has been seriously criticised and extensively revised). This was to be the start of an inexorable process of modernisation for Chinese linguistics.
In a sense, the influence of Western linguistics was facilitated by the tension between the Indian and the Chinese approach. For having long suffered from the final denial of detailed precision in a framework that seemed to promise phonetic exactness, many scholars saw a long-awaited solution in the Western approach. The following is a famous example of some Qing scholars being “enlightened” by Westerners’ alphabetic representations of their own language:
Hardly had I finished reading the book A Help to Western Scholars (A textbook written by the French Jesuit N. Trigault (1577-1628) for Western learners of Chinese) when I suddenly realised the rationale and rules underlying the qiezi 切字 (i.e. characters used in fanqie 反切). (予閱卷未終，頓悟切字有一定之理，因可為一定之法。)
Yang Xuanqi 楊選杞, Writings on the Combinations of Initials and Rhymes 聲韻同然集 (1659)
So, at least as far as its phonetic part was concerned, the Western approach was both new and familiar. It is therefore easily understandable why, of all the branches of traditional Chinese learning, the study of the speech sound was among the first to connect to Western science (Ge Zhaoguang 葛兆光, An Intellectual History of China 中國思想史, Vol. 2 Shanghai 上海: Fudan University Press 復旦大學出版社, 2001). But the curious thing is: knowledge of the speech sound was at the same time regarded as the a centrepiece of the “basic learning”, which was itself the foundation of all traditional Chinese learning. As Duan Yucai 段玉裁 says, “what is the most important in the study of the classics is to find out the correct meaning, and the most proper way to do so is by means of phonetic analysis” (治經莫重於得義，得義莫切於得音). As a result, the tension I have been discussing was still there – complicated, though, by the addition of a Western element, for the key issues in the “centre of the centre” of the Qing scholars’ intellectual world, namely the phonetic techniques that had become so important to the understanding of their ancestors’ words, were only vaguely understood in the Chinese philosophical tradition. Even today, anyone seriously studying ancient Chinese works has to have a solid basis in the so-called “historical Chinese phonology,” but to do so, one must be a student of Western phonetics first.
On the other hand, linguists are not all satisfied with the way historical Chinese phonology is done, because Chinese characters have never ceased to be a significant part in it: all the initials, rhymes, and tone categories are still named today by characters according to convention, and have to be learnt by heart, and, of course, the documents to be dealt with are all written in characters, too. Notably, some so-called “Princeton School” scholars, including J. Norman and A.-Y. Hashimoto, tried to break away from the character-based tradition altogether and to rely completely on historical reconstructions from modern Chinese dialects. (See, e.g. J. Norman and W. Coblin, “A new approach to Chinese historical linguistics”, Journal of the American Oriental Society, Vol. 115, No. 4: 576-584, 1995.) However, their approach has been strongly opposed, because if characters were completely dispensed with, the evidence that could be used to reach any safe conclusion would be reduced to an insufficient minimum (E. Pulleyblank, “Qieyun and Yunjing: the essential foundation for Chinese historical linguistics”, Journal of the American Oriental Society, Vol. 118, No. 2: 200-216, 1998).
A more recent outburst of the tension passed down from ancient times can be seen in a heated discussion among Chinese linguists on the methodology of historical Chinese phonology since 2001 (for background information see Collected Papers on the Methodology of Chinese Historical Phonology 音韻學方法論討論集, Beijing 北京: The Commercial Press 商務印書館, 2009). The focus of the discussion seems to be the reconstruction of Old Chinese phonology, but the central problem in this reconstruction is the way the Chinese data in the form of characters should be treated. As a result of a divergence in how to interpret ancient Chinese sounds represented by characters, some scholars found a lack of scientific strictness in the works of others, while their own reconstructions have been criticised for being overly complicated by unnecessary rigidity with characters and thus too different from Middle Chinese or any other language to be possible.
Some of the key ideas in this post can be found in my paper “‘Inner Logic’ in the modernisation of Chinese linguistics”, Journal of Oriental Studies, 44(1 & 2): 25-44, 2012.
English introductions to historical Chinese phonology can be found in the following works:
S. Auroux, et al. (ed.), History of the language sciences, Vol. 1. Berlin: Walter de Gruyter, 2000.
M. A. K. Halliday, “The origin and early development of Chinese phonological theory”, in R. E. Asher and E. J. A. Henderson (ed.), Towards a history of phonetics. Edinburgh: Edinburgh University Press, 123-140, 1981.
G. Lepschy (ed.), History of linguistics, Vol. 1. London: Longman, 1994.
R. A. Miller, “The Far East”, in T. A. Sebeok, et al. (eds.) Current trends in linguistics, Vol. 13, Historiography of linguistics. The Hague: Mouton, 1213-1264, 1975.
J. Norman, 1988. Chinese. Cambridge: Cambridge University Press.
How to cite this post
Zhu, Lei. 2013. ‘Historical Chinese phonology as a meeting ground for the Indian, the Chinese, and the Western linguistic tradition’. History and Philosophy of the Language Sciences. https://hiphilangsci.net/2013/09/04/historical-chinese-phonology-as-a-meeting-ground-for-the-chinese-the-indian-and-the-western-linguistic-tradition