André-Georges Haudricourt (1911-1996)

Summary: Haudricourt’s main legacies to the field of Chinese historical phonology are his systematic account of tonogenesis, and his reconstruction of final *-s and of labiovelars. The method that he proposed under the term of Panchronic Phonology holds special promise for the study of the languages of China in their universal context.

He investigated a variety of language groups -European, as well as Asian and Oceanic. As his fragile health did not allow him to conduct much fieldwork himself, he mostly worked on the basis of available reports, which at the time were relatively deficient. He excelled at identifying the shortcomings of these sources by taking into account the language background of the persons who collected the data: reconstructing, as it were, the investigators' analytic process. He also applied this approach to the analysis of the origin of the peculiarities of the Vietnamese alphabet (1949, translated 2010).
He is remembered as an impressive mentor, at his best in informal exchanges, going straight to the heart of scientific issues. His unpolished style often makes his articles hard to read, but reflects the precision and vividness of his insights. A translation of his most seminal papers is currently in preparation (ed. by Martine Mazaudon and Alexis Michaud, Berlin: De Gruyter Mouton, forthcoming); his more anecdotal pieces, including interviews (Haudricourt and Dibie 1987;Haudricourt and Jacquesson 1989;Haudricourt, Dessein, and Swiggers 1997), are also extremely instructive, reflecting his profound erudition and commitment to down-to-earth scientific method, as well as his nonconformism and acute sense of irony.

The relationship between Vietnamese and Chinese
Haudricourt demonstrated (pace Maspero 1912) that Vietnamese and Chinese are not genetically related, teasing out layers of borrowings, and emphasizing that typological similarities, such as the presence of lexical tone, do not constitute evidence for phylogenetic groupings (1953). This represents an important contribution towards defining the boundaries of the Sino-Tibetan family.

Systematic account of tonogenesis in Chinese
Haudricourt's account of the origin of tones in Vietnamese (1954) addresses issues of tonogenesis in a range of Asian languages, including Chinese. The discussion takes one step further the insights of previous scholars (most prominently Maspero 1912), concluding: "Old Chinese only had three tones up until the 9 th century. At that time, the voiced stops /g/, /j/ [IPA: ɟ], /d/ and /b/ devoiced to /k/, /c/, /t/, /p/, aspirated or not depending on dialects and on tones. From that point on, the musical height [as opposed to the contour] of the tone became a contrastive phonological property used to distinguish words. Two words which used to be distinct because one had an onset /k/ and the other had an onset /g/ came to be distinguished solely by the higher tone of the former. The three-tone system thus became a six-tone system" (translation by Marc Brunelle).
A systematic account of the two stages of tonogenesis in Chinese was provided by Haudricourt as part of his general model of tonogenesis in Asia (1961), including important observations on the behavior of the mid series of initials in tonogenesis: "We have seen that Ancient Chinese [=Middle Chinese] had only one series of sonorants. These sonorants were not involved when the voiced and voiceless initial stops merged, and so they form a "mid" series of initials. For example, in Sino-Vietnamese, [=Sino-Annamese], a Chinese dialect of the 10 th century preserved in Vietnam as a language of high culture (Maspero 1912:91-93): We see the paradox that B and C have been confused in the low series, while they remained distinct in the "mid" series. The paradox is only apparent: the tonemes of the "mid" series had no reason to be modified, whereas those of the "low" series had to be distinguished from whose of the "high" series.
Modern Chinese has an analogous treatment (Egerod 1956:24): Haudricourt 1968 and further studied by Rivierre 1993Rivierre , 2001. This demonstrates the usefulness of a cross-language approach to address problems found in the study of one specific language. Haudricourt referred to this approach as "panchronic" (Haudricourt 1940(Haudricourt , 1978Hagège and Haudricourt 1978). The aim is to formulate generalizations about sound change that are independent of any particular language or language group. Panchronic laws are obtained by induction from a typological survey of precise diachronic events whose analysis brings out their common conditions of appearance. In turn, these laws can be used

Other contributions to the reconstruction of Old Chinese
Haudricourt's contribution to Old Chinese reconstruction is not limited to his account of tones. He clarified several rhyming patterns found in the Book of Odes. Words with final stops /-p -t -k/ rhyme with words in departing tone (去聲 qùsheng) according to their MC pronunciation: for instance, words in the zhà 乍and zuó 昨series (MC: *dzraeH and *dzak, respectively) rhyme, as do words in the bì 敝 and piē 瞥 series (MC : *bjiejH and *phiet).
"Already the fact that Ancient Chinese [=Middle Chinese] possessed -p, -t, -k, -m, -n, -ng but not -b, -d, -g suggests these latter; And this seems so much the more natural as the experience from other languages shows that the mediae [i.e. voiced stops] more easily fall than the tenues. To cite my own language again, there are many Swedish dialects where bed > bē but bet > bet. It is therefore rather likely that it is a final g and a final d we have to expect in words like 乍, 敝" (Karlgren 1923:27-28, cited in Baxter 1992.
Some years later, Karlgren realized that the words that he reconstructed with final *-g sometimes rhyme with words whose phonetic series in Old Chinese has no trace of a final stop. At that point, he concluded that final *-g must be generalized to the entire set of syllables that rhymed with his *-g-final syllables. This resulted in a reconstructed language that had almost no open syllables. Karlgren's reconstruction was adopted by many scholars.
Haudricourt's theory (1954b) that the departing tone comes from *-s explains this phenomenon: the words with departing tone rhyming with words in final stop should be reconstructed with final clusters *-ks, *-ts or *-ps. Moreover, from the point of view of historical morphology, Haudricourt's theory of tonogenesis leads to the reconstruction of several *-s suffixes (in particular a nominalizing suffix) which can be shown to be cognate with those found in conservative languages of the Sino-Tibetan family, such as Tibetan.
This theory has been accepted by nearly all specialists in Chinese historical phonology, with the exception of some scholars in China. Li Fang-Kuei (1971:33-34) mentions Haudricourt's hypothesis but considers that the internal evidence from Chinese does not allow for any hard-and-fast conclusion, and provisionally chooses to retain Karlgren's reconstructions with final *-g. This cautious choice tends to be reinterpreted by some of Li Fang-Kuei's epigones as an outright rejection of Haudricourt's analysis. As for Wáng Lì (1989: vol. XVII, pp. 213-247), he points out that the reconstructions of final voiced stops by Bernhard Karlgren andWalter Simon (1928-1929) are problematic, but he is apparently unaware of Haudricourt's proposal.
A second major finding, also reported in the 1954b article, consists in the hypothesis that there were labiovelars in Old Chinese: This idea was used later by Yakhontov 1960a andPulleyblank 1962 to revise the reconstruction of the Old Chinese vowel system, and is the basis for the six-vowel system that is common to the recent systems of Starostin 1989, Baxter 1992 and Zhèngzhāng Shàngfāng 郑張尚芳 2003. On the other hand, two of the ideas proposed by Haudricourt have been disproved since the time of publication.
First, in his discussion of the MC initials *dzy-(常 cháng) and *zy-(船 chuán), he argues that the contrast is only due to dialect mixture. However, we now know that these two initials have entirely distinct origins in Old Chinese, and reflect a genuine contrast, though this contrast is poorly preserved in modern dialects. The phonetic reconstruction of these initials in MC by Karlgren was erroneous: it was based on misleading evidence found in the 切韻 Qièyùn. The rhyme tables in the Qièyùn clearly suggest an interpretation as voiced palatal affricate and voiced palatal fricative for these two initials, but this agrees neither with earlier texts, nor with transcriptions of Sanskrit in Chinese characters (see Lù Zhìwéi 陸志韋 1947 andPulleyblank 1962:67-68).
Second, his hypothesis concerning the existence of spirants, which he set up to solve the issues raised by Karlgren's reconstruction of voiced stops in Archaic Chinese, is now known to be mistaken. Yakhontov 1960b andPulleyblank 1962 put forward another interpretation, which is now accepted by most scholars: that Karlgren's *d-should instead be reconstructed as a lateral. A clear account of the necessity of reconstructing laterals in Old Chinese is provided by Baxter (1992:196-199).
This overview has attained its goal if it has shown (i) that Haudricourt's legacy to the field of Chinese historical phonology is not limited to advances in the study of tonogenesis, and (ii) that the method proposed by Haudricourt under the term of Panchronic Phonology holds special promise for the study of the languages of China in their universal context.