Difficulty in identifying cognates and loanwords

pattern is clearly iambic, and although epenthetic schwas often occur within or after the minor syllable, there is no contrastive vowel nucleus. In Northern Kammu, the minor syllable can carry contrastive tone Svantesson and Karlsson 2004. The existence of sesquisyllables in some South-East Asian languages, coupled with some well- documented examples of non-tonal disyllabic proto forms which have acquired tone and reduced to monosyllabic forms for example Old Cham Tsat, described by Sagart 1993, have both been cited as evidence for the existence of sesquisyllabic forms in proto languages in the Tai-Kadai family Ostapirat 2000, Pittayaporn 2009. Matisoff 1990:547 argued that southeast Asian languages undergo a “millenial dance” between four different types of syllable: 1 consonantally complex monosyllables; 2 consonantally simple monosyllables; 3 disyllabic compounds or tight collocations; and 4 sesquisyllables. The stage in which disyllabic compounds reduce to sesquisyllables is one which we have amply demonstrated is taking place in Sui. Although all of the Sui dialects which we surveyed exhibited some sesquisyllabic forms, none of these forms were consistently elicited across the Sui region. For most examples of a sesquisyllabic word that we found in one dialect, we found the same word in fully disyllabic bimorphemic form in another. Thus we argue that Sui is still essentially a monosyllabic language. Whether Proto-Kam-Sui was monosyllabic or sesquisyllabic is a matter for debate and will probably not be decided until the historical processes of monosyllabification and tonogenesis have been more adequately explained. As we examine the historical development of the Sui dialects in more detail, we do, however, find that Proto-Kam-Sui sesquisyllabic forms are a neat and useful way of explaining some of the sound changes which have taken place.

3.3.4 Difficulty in identifying cognates and loanwords

One of the greatest challenges in southeast Asian comparative linguistics is the correct identification of cognates and loanwords. Contact between the Sinitic, Hmong-Mien and Tai-Kadai languages goes back over two thousand years, and most of the modern languages in these branches are typologically similar—largely isolating and monosyllabic, with rich contour tone systems, a limited number of rime codas usually restricted to -p, -t, -k and -m, -n, - ŋ at the most and relatively limited options for onset consonant clusters. The chances of two etymologically unrelated words looking phonetically similar are therefore higher than for many other language families. Campbell 2004:355 says that as a rule of thumb, “5–6 per cent of the vocabulary of any two compared languages may be accidentally similar.” 5 And in writing this, he did not specifically have languages in the Sinosphere in mind. The temptation for historical linguists in the Sinosphere, then, is to formulate historical hypotheses based on etymologically dubious correspondence sets. It is particularly easy to find apparent cognates in Chinese because of the huge number of Chinese characters recorded in historical documents. For example, the Qie Yun 《切韵》 rhyme dictionary, published in 601AD mid Sui dynasty, contains over 11,500 characters, with a total of 193 individual rhymes spread across four tone categories, i.e., a mean of almost 60 different Chinese characters for each rhyme although of course some rhymes are far more common than others. The Guang Yun 《广韵》 , published in 1008AD early Song dynasty but representing an earlier pronunciaton contains over 26,000 characters. The Qie Yun and Guang Yun are basic reference works for all reconstructions of Early Middle Chinese pronunciation. The Kangxi Character Dictionary 《康熙字典》 1716AD, early Qing dynasty contains a colossal 47,025 characters. For virtually any monosyllabic word in modern Hmong-Mien and Tai-Kadai languages, the diligent researcher is almost bound to find a Middle Chinese or Early Mandarin form which is both phonetically and semantically similar. Just to cite one example, at first glance the Sui for ‘ox horn’, qaːu¹ or paːu¹, appears to be cognate with the Chinese 角 t ɕjau³ Mandarin, kjaw³ Early Mandarin. Further investigation reveals that this is probably not the case. Tone category and onset correspondences indicate that other Chinese loanwords relating to livestock and agriculture, such as ‘yoke’ ʔiːt⁷, ‘field’ tjen² and ‘dog’ qau³ the latter two 5 See Ringe 1992 for an extensive discussion of the factor of chance in language comparison. preserved as reading forms by experts in traditional Sui script, were borrowed at an earlier stage, well before Early Mandarin. Pulleyblank 1991 reconstructs the Late Middle Chinese pronunciation of ‘horn’ as kjaːwk D . Other Chinese loans in Sui show that modern Sui consistently retains -p, -t and -k finals from Middle or Old Chinese loans. Added to this, it seems unlikely that the Sui, who have presumably been raising livestock for many centuries, should borrow the word for ‘horn’ at such a late stage. We discuss Chinese and other Tai-Kadai loanwords in Sui in a little more depth in chapter 6. Suffice to say that throughout our analysis we adopt a cautious approach to identifying loanwords from neighbouring languages. At the same time, we do make use of the clues that probable older Chinese loanwords give us about subsequent sound changes which have occurred in the Sui language.

3.4 Data and conventions