1September 2002

2Whatever area of research you look at, the problems that need addressing at the beginning of the 21^st century are becoming increasingly complex. They require active interdisciplinarity, which has become the fundamental basis of research policies, regardless of what their field and origin may be. Interdisciplinarity does not hide the need to promote a comparison between different, close or remote cultural areas. This observation paved the way for the creation of the “Asia Network” (Réseau Asie) so that it could coordinate research on different Asian countries better.

3To illustrate the relevance of such a comparative approach, I would like to discuss the issues raised by the classification of east and southeast Asian languages.

4Linguists traditionally divide the world’s 5,000 to 6,000 languages – nearly half of which will disappear by the end of the century – into 400 to 500 families of various sizes. Some of them, such as the Austronesian family, include more than 1,200 languages, while others, only one; Basque is one well-known example of such a language isolate.

5During the 20^th century, few linguists set out to compare different families with each other and suggested that families be grouped into “superfamilies”. This methodological point of view has developed since the 1980s at the request of the impetus of linguists such as Greenberg1 or Ruhlen2, who had made earlier suggestions (i. e. Austric, Nostratic and renamed “Eurasiatic”, etc.) and put forward new superfamilies such as Dene-Caucasian or Amerind. These are still bitterly argued hypotheses within the linguist community.

6A study of the situation in East and South-East Asia unveils the various possibilities of how languages could be classified as well as how difficult it is to have reached a definitive decision in favour of any of them.

7This vast geographic area is generally said to have five major linguistic families: Sino-Tibetan, Austronesian, Austro-Asiatic, Tai-Kadai and Miao-Yao (Hmong-Mien).

Big families

8Sino-Tibetan (ST) is generally divided into two branches that are believed to have been separated 6,000 years ago, at a time when archaeologists identified the creation of what they call “the start of China” (Chang) 3: Sinitic languages on the one hand and Tibeto-Burman languages (TB) on the other. Some 250 languages belong to TB, including Tibetan, Lolo, Jingpo, etc. As for Sinitic languages, they are made up of seven groups: northern Chinese (Mandarin), Xiang, Gan, Wu, Min, Kejia (Hakka) and Yue (Cantonese). To this day, the theory of the ST family remains disputed. Some linguists4 believe Sinitic languages to be nothing more than a branch of TB, or even a sub-sub-sub branch. If this were the case, ST would not even have existed.

9The Austronesian (AN) family dates back to approximately 5,000 BC. It is one of the oldest families to ever be identified, even older than the Indo-European family, including over 1,200 languages. According to Blust5, the AN family has ten branches, nine of which represent only 26 languages and which are native to Taiwan. The tenth branch includes over 1,100 languages from Madagascar to the east of Polynesia (cf. Diamond, 20006).

10Sagart7 on the other hand, considers AN to be made up of only six branches, all present in Taiwan, with only one represented elsewhere. In this case, Malayo-Polynesian languages would only become a sub-sub-sub-sub branch of a Paiwan-Polynesian group.

11The Austro-Asiatic family (AA), which was established around 1880, accounts for about 150 languages. It is sub-divided into two branches: the Munda languages from northern India, and the other Mon-Khmer languages, mostly spoken in Vietnam and Cambodia. Similar to AN, it is possible that AA may date from 4,000 to 5,000 BC.

12Tai-Kadai (Daic) included the Thai languages and the Kadai languages. This family obviously includes Thai, but also Lao, Zhuang, Nuyi as well as the Kam-Sui languages (Kam, Sui, Mulao, Maonan, etc.) from southern China.

13Lastly, the Miao-Yao (Hmong-Mjen) family includes thirty languages that can be divided into two groups: the Miao languages group (Hmong languages spoken in Guizhou, Yunan and Indochina, Hmn spoken in western Guizhou and Qo Xiong spoken in western Hunan) and the Yao languages group (Mjen, Mun and Tsao Min.)

Superfamilies

14Various groupings between the aforementioned five families were put forward in the last century. Almost all combinations were tried and tested.

15Leaving out the Sino-Indo-European hypothesis, which is still very weak (Pulleyblank8), four hypothetical scenarios of four superfamilies were up for discussion: Austro-Tai, Austric, Sino-Caucasian, Sino-Austronesian (recently renamed Sino-Tibetan-Austronesian).

16Regarding the Austro-Tai family, Schlegel9 had already noticed a link between Tai and Austronesian languages. It was Benedict10, however, who confirmed, as early as 1942, that there really was an Austro-Tai superfamily, which included two branches, Tai-Kadai and Austronesian, to which Miao-Yao was sometimes added. This hypothesis is still defended by Ostapirat11 and challenged by Thurgood12, among others.

17The hypothetical grouping of the AA and AN families into the Austric family was also previously suggested by Schmidt13. Recently, Blust14 and Reid15 took this up. Other linguists even included Miao-Yao and Tai-Kadai to the group. However, Diffloth16 considers these links to be far from convincing. Moreover, he believes that proto-AA originated far south of the southeast sub-continent, or even in India itself, which makes the Austric hypothesis quite far-fetched.

18After identifying lexical aspects common to ST, northern Caucasian and Yeniseian (Ket language), Starostin17 suggested, in the early 1980s, that there be a superfamily: Sino-Caucasian. Later, Bengston18 included Basque, Burushaski and Na-Dene to this group to create the superfamily that Ruhlen called “Dene-Caucasian”.

19Finally, at the start of the 1990s, Sagart19 argued that the Chinese and Austronesian languages might be connected and concluded that there was thus a superfamily – “Sino-Austronesian” – that was divided into two branches, Sinitic and Austronesian languages. He had initially left aside the Tibeto-Burman languages, but he has now reconsidered this hypothesis and acknowledged Sino-Tibetan as part of the construct (with its two branches, Sinitic languages and Austronesian languages) and has put forward the idea of a new superfamily, Sino-Tibetan-Austronesian (STAN) 20.

20Tai-Kadai languages were also included in STAN, supposedly dating back to 6,500 BC, at the time when cereals were starting to be cultivated in the Yellow River valley.

A super-superfamily: PEA (Proto-East Asian)

21The supposed connections between the various aforementioned linguistic families naturally led researchers to imagine an ancestral language common to all these languages. Starosta21 has recently suggested a super-super family, which he calls “Proto-East Asian” (PEA). According to this theory, PEA is supposed to have been spoken in Central China, along the Han and Yellow Rivers, around 10,000 to 8,000 BC. This consists of only two branches: STAN (taking into account Sagart’s suggestion22 to include Tai-Kadai as well) and Austro-Asiatic. Thus, there would be no need for an Austric superfamily.

22This hypothesis of a single super-superfamily in East and South-East Asia is merely a theory, but it could be supported by findings in typological studies, or even research into areal linguistics, which are more likely to use syntactic rather than phonetic or lexical criteria. A number of syntactic properties common to most languages of this vast area have been identified.

Languages and genes

23Population genetics, which aims to study the genetic variability of our species in order to reconstruct the history of human beings from their origins until this day, is helpful for linguists in their classification of languages.

24From the moment geneticists started analysing DNA in depth in the 1980s, a significant amount of data was gathered. While bio-chemical and molecular techniques developed, theoretical evolutionary models were required to interpret genetic material in a historical context at the same time.

25Attempts to correlate genetic distance – a key concept in population genetics – and linguistic distance were made, and led to the discovery of a few similarities between the genetic classification of populations and the classification of language superfamilies. Other papers later refuted the existence of incontestable correlations between the genetic classification of communities and language classification.

26An important piece of research, conducted by Chinese geneticists studying the distribution of Gm and Km immunoglobulin allotypes in 74 populations of the People’s Republic of China, concluded that there were genetic differences between northern Han and southern Han populations. The division follows a line situated at latitude 30° north. (Zhao and Lee23, Zhao et al.24) The comparison with 33 other populations outside China also revealed that northern Hans are likely to belong to a group that also includes the Athabaskans (speaking north American Na-Dene languages, i. e. belonging to the Dene-Caucasian superfamily), the Eskimos, the Japanese, the Koreans and the Mongolians (all of whom speak Eskimo-Aleut or Altaic languages that belong to the Eurasiatic superfamily). Southern Hans, on the other hand, would be closer to the Thais (Tai-Kadai languages), the Indonesians, the Filipinos (Austronesian languages) and also the Vietnamese (an Austro-Asiatic language); in short, populations whose languages are related to the Austric superfamily.

27After studying the genetic profile of 28 Chinese populations, Chu et al.25 strongly supported this north-south opposition. Similarly, Su et al.26 whose study focused on the Y chromosome, noticed that genetic polymorphisms were much more prevalent in southern than northern populations, which only shared a very limited subset of haplotypes with the former group.

28The Han ethnic unity would therefore be cultural, not biological. It could also be non-linguistic since Sinitic languages are either linked to Dene-Caucasian regarding the languages spoken in northern China, or to the Austric superfamily in the case of southern Chinese languages.

29Ding et al.27, after analysing three genetic markers and a human virus contend that such a north-south distinction is groundless. They point out that the Zhuang people from southern China, who speak Zhuang (a Tai-Kadai language) are genetically very close to the Chinese people of the north, much closer than they are to the Vietnamese (who speak Austro-Asiatic languages) or the Malay (whose languages belong to the Austronesian superfamily). They conclude that the differences observed between the northern and southern Chinese languages are to be ascribed to simple cultural phenomena, but have no genetic origin whatsoever.

30Yao et al.28 also concur that a simple division between the southern and northern Chinese languages fails to account for the genetic structure of the Hans, which is far more complex.

31This debate on Asian populations, such as the discussion on the existence of superfamilies is not about to be solved any time soon. The most recent and promising studies, regardless of the contradictory hypotheses and uncertainty that they have led to so far, are unusual in that they are truly interdisciplinary and intercultural. Linguists have gone beyond their own capacity and felt the need to work not only with geneticists, but also with archaeologists, paleoanthropologists and paleodemographers, in trying to achieve the “new synthesis” that Renfrew has been hoping for since the early 1990s. At the same time, specialists of specific cultural areas (China, Japan, India and South-East Asia) have also extended their research beyond their preferred cultural areas.

32It was time that France, like other European countries such as the Netherlands, or the United States, set up its own “Asia Network” to coordinate these initiatives.