TITLE: Common Tai and Archaic Chinese. AUTHOR(S): Nishida, Tatsuo CITATION: Nishida, Tatsuo. Common Tai and Archaic Chinese.. 音声科学研究 1975, 9: 1-12 ISSUE DATE: 1975 URL: http://hdl.handle.net/2433/52582 RIGHT: STUDIA PHONOLOGICA IX (1975) Common Tai and Archaic Chinese Tatsuo NISHIDA 1. A number of scholars have discussed the problem of whether there is any re lationship between the Thai and Chinese languagesl. Among them, K. Wulff has carried out the most comprehensive research. In his Chinesisch und Tai2, Wulff first made detailed comparative studies ofTai dialects, and then comparing them with Karlgren's reconstructed form of Ancient Chinese. This work, while an excellent and enduring one, fails to establish regularities of phonemic correspond ence between the two languages. The Thai and Chinese languages are very close to each other not only in phonemic inventory and syllabic pattern, but also in syntax, and show one-to-one correlations. Nevertheless it is not easy to find sys tematic correspondences between individual phonemes and morphemes of the two languages. Therefore, it is quite natural to pose the question ofwhich family the Thai language should be said to belong t03• 2. It was previously proposed that the Sino-Tibetan language family, including Thai, should be brought into the Indonesian language family, and the whole in corporated into a larger group oflanguages.4 In his Uber das Verhaltnis des Malayo Polynesischen zum Indochinesischen, Wulff again took up the question of the relation ship between these two families. But this attempt, which proposed the existence offormal similarity between Thai and Indonesian, and the probability of certain co-relationships between them, can be endorsed only when further concrete com parisons have been made. In the same year, Paul Benedict suggested a new lan guage family, "Kadai", which would consist of the following languages: Laqua and Lati in the Tonkin plateau, Kelao in southern China and Li in Hainan, whose proper placement in language families was so far unknown.5 By using Kadai, he tried to relate Thai to Proto-Indonesian.6 Benedict has summarized his as sertion under the following headings: Tatsuo NISHIDA (j~:!HEft.m):Professor of Linguistics, Faculty of Letters, Kyoto University. 1. R. Shafer 1957: 163-166, 190-193. 2. K. Wulff 1934. 3. Amongthearticles treatingthissubjectare: Hsing Kung-wan 1955a 1955b, ChangKung-chin 1958. 4. Conrady, A. Bine merkwurdige Beziechung zwischen den austrischen und den indochine sischen Sprachen, Kuhn Festschrift (Munich, 1916:475-504) Neue austrisch-indochinesische Parallelen, Hirth Anniversary Vol. (26-66). 5. P.K. Benedict 1942. 6. Haudricourt,independentlyofBenedict'sresearch,comparedthefourteenwordsofLati,Laqua, Kelao and Dai (=Li) with the Common Tai which he himself had deduced, concluding that these languages had primitive forms before the Thai language suffered changes under the influence of Chinese and Khmer. Haudricourt 1948. 2 Tatsuo NISHIDA 1. The true Indonesian substratum on the Asiatic mainland is represented by four scattered languages in southern China, northern Tonkin, and Hainan, all of which constitute a single linguistic stock (Kadai). 2. The recognition of the Kadai stock, which shows numerous points of contact with Thai, opens the way to a new interpretation of the latter as a more distant member of an archaic Thai-Kadai-Indonesian linguistic complex. Despite his valuable research and its originality, which has drawn many an"" thropologists and linguists to accept his theory, Benedict fails to prove adequately the existence of his Kadai linguistic family.7 Besides larger linguistic groups with individual characteristics, such as Thai, Chinese, Burmese and Tibetan, there exist in Asia many languages which exhibit intermediary traits of several language groups but which cannot clearly be as signed to one. Most of these languages have not been investigated in detail. Benedict, using the restricted documents so far published, gave the name "Kadai" collectively to the four languages sharing Thai and Indonesian characteristics. Before recognizing these languages as one language family, however, further in vestigation should be made of languages which have similar traits. At the same time, their relationships to the Mak-Sui-Kam and Chuang language groups should be borne in mind. Also calling for further thorough study is the basis of Wulff and Benedict's argument that an Indonesian lexicallayer should exist in the Thai language, whether their common elements be genealogical or borrowed. In any case, this possibility cannot be discarded prior to investigation. 3. On the relationship between the Thai and Chinese languages, A.-G. Haudri court states as follows:8 On voit, d'apres cette reconstitution des phonemes de la langue commune, combien Ie thai est eIoigne du chinois. Seul coincide Ie systeme des consonnes finales et des tons; les voyelles et les initiales sont profondement differentes. Les mots de la langue commune incontestablement proches de mots chinois sont les noms de nombres, des techniques militaires (cheval, selle, elephant, jouguet) et des techniques artisanales (metier a tisser, ouvrier, papier), bref un vocabulaire de civilisation susceptible d'emprunt. Au contraire Ie nom des parties du corps etIe vocabulaire agricole ont peu d'affinite avec Ie vocabulaire chinois correspondant. Now, is it proper to infer from this that these two languages belong to com pletely different families, or that, though belonging to the same family, they are both in rather remote kinship and that words showing structural similarity are Chinese loan words in Thai? Though I cannot present a definite answer to such a broad problem, I should like to discuss an aspect of this problem. 4. To begin with, the question to be proposed is whether it can be concluded, as 7. For example, J.H. Greenberg, Historical Linguistics and unwritten Languages (Kroeber, Anthro,bology Todqy, Chicago 1953). 8. Haudricourt 1948: 235-6. Common Tai and Archaic Chinese 3 Haudricourt does, that Thai vocabulary indicating parts of the body is derived from stems or roots other than Chinese. Choosing thirty words at random, let us examine the word forms of the two languages.9 Common Tai Archaic Chinese Ancient Chinese 'head' hr;;m HI siog > -'gt > 'hair' phrom HI piwat piwvt .~. > 'ear' hruu HI ilidg l} > 'nose' dang HI bhidd bhji > 'eye' taa Ml miok miuk § > 'mouth' paak, suup khu o > 'tooth' qhi;m H3, van Ll ihidg tshi -~ > 'tongue' lin L3 dhiat dihiat E > 'face' hnaa H3 ngan ngan &.~ > 'neck' G::>::> Ll kieng kiang > 'chin' gaang Ll gidg > 'cheek' keem M3 kiap kiep > 'hand' mi 1;1 siog si~u 'finger' niu L3 iidr > tsi 'elbow' s::>::>k tiog > iidU > 'leg' qha HI kiak kiak > 'knee' kheeng H3 ghieng rieng > 'body' ton Ml sien sien > 'waist' 'ew Ml 'iog 'iau > 'shoulder' Baa M2 kian kien > 'belly' puum Ml piok piuk > 'liver' tap k~m kan > 'heart' cai Ml sidm sidm > 'lungs' p::>::>t phiwad phiwvi fjrfJ > 'stomach' d::>::>ng L3 giwdd jw~i ~ > 'flesh' nid L3 iliok lliiuk ~ > 'blood' lidt xiwet xiwet 1frJ. > 'muscle' Ten MI ki;m ki;;m JFj > 'skin' phiu HI pliwo piu JI > 'navel' sai-Brii Ml dzhidr dzhiei ~ Clearly, the words 'eye', 'nose', 'mouth', 'shoulder', and 'liver' are very dif ferent in form, while some of the other words show striking formal similarities, among which are 'waist', 'knee', 'lungs', 'muscle', 'skin' and 'navel'. These words 9. Below I will use Karlgren's reconstructed forms for Archaic and Ancient Chinese as in his Grammata Serica Recensa, Stockholm 1957, with the exception of the aspiration which will be denoted by 'h' instead of'. The Common Tai forms cited here are quoted from my own personal notes and other unpublished papers. Cf. T. Nishida 1954, 1955. 4 Tatsuo NISHIDA may possibly be loan words as well as cognate words derived from the same stem. We must inquire, therefore, whether there are other words which show formal similarity of the type illustrated by the above examples. What concerns us in particularis the correspondence betweenvowels and finals, such as: the correspond ence of 'waist'-Common Tai -ew: Chinese -iog> -iau (hence abbreviated C.T. and Ch. respectively) is paralleled by 'cat'-C.T. mew 1.1: Ch. ~i miog>miau, and 'to finish'-C. T. lew1.3: Ch. T liog (?»liau. Similarly, 'knee'-C.T. -eeng: Ch. -ieng>-iang is paralleled by 'hard'-C. T. kheeng HI: Ch. !fJ kieng>kiang, mr and 'powder'-C. T. peeng Ml: Ch. pieng(?»piang: 'navel'-C.T. -ai: Ch. -ieg>-iei, -idr>-iei by 'fowl'-C.T. kai M2: Ch. ~t kieg>kiei, 'ladder'-C.T. ?dai Ml: Ch. ~ thidr>thiei, and 'to till'-C.T. thai HI: Ch. ~ lidr>liei. On the contrary we cannot find examples which parallel 'lungs'-C. T. -;:)Jt: Ch. -iwad>-iwvi1o, 'muscle'-C.T. -en: Ch. -idn>-idn, and 'skin' C. T. -iu: Ch. -iwo>-iu.ll As a result of this we cannot but hesitate to treat these three words as loan words from Chinese. Besides the above-cited seven words, the five words: 'hair,' 'cheek,' 'belly,' 'leg' and 'flesh' bear formal resemblances between the two languages and, therefore, may possibly be cognate words. Out of these five ex amples, in the first three and the last two respectively there are similar correspond ences. In 'hair,' 'cheek' and 'belly', the Tai final nasal consonants correlate with Chinese stops, i.e. -m: -t, -m: -p, -m: -k. Other examples of this correspondence can be found: -n: -t 'to eat'-C. T. kin Ml (T.M.C. *krin): Ch. nt kidt>kidt; -n: -k 'stone'-C.T. hrin HI: Ch. :E diak>iiak; and -ng: -k 'force'-C.T. reeng 1.1: Ch. JJ lidk>lidk. On the contrary, a Tai final stop sometimes corresponds to a Chinese final nasal; i. e. 'scale (fish)'-C.T. klet: Ch.• lien>lien, and 'whip' -C.T. vaat: Ch.• pian>pian. These final consonants, whose function cannot be clarified, may be regarded as a sort ofsuffix. We can explain that 'hair' has *phrwa""""prwa(?) as its common form, suffix -m being suffixed in Tai and suffix -t in Chinese. Similarly it can be inferred that 'cheek' consists of the common root *kee- and suffix -m (Tai) or -t (Chinese); 'belly' of root *pu- or *piu- and suffix -m (Tai) or -k (Chinese); 'to eat' of root *kia- and suffix -n (Tai) or -t (Chinese); 'scale' of root *kle- and suffix -t (Tai) or -n (Chinese) and 'whip' of root baa- and suffix -t (Tai) or -n (Chinese). On the other hand, the last two examples 'leg' and 'flesh' have this kind of suffix only in Chinese, and not in Tai. As for 'leg,' the Tai form qha HI is iden tical with the common root, the Chinese form kiak being suffixed by -k. Like wise, 'flesh', whose root is almost equal to Tai nid 1.3, is suffixed by -k, forming Chinese niok. On the contrary, in the following examples, only the Tai words retain suffixes -m, -ng and the Chinese forms have no suffix. 10. 'To take off' Ch. ~ thwat-thuat: C.T. th:>:>t is close to this. 11. For the otherTaiforms corresponding to Ancient Chinese !Ji't{fJ'tyu rhyme, see p. 8 and foot note 18 of this paper. Common Tai and Archaic Chinese 5 'ditch' C.T. khum HI Ch. M ku > k~u *khu- 'bitter' C.T. khom HI Ch. ~ kho > khuo *kho- 'drum' C.T. kbJng MI Ch. fit ko > kuo *kb::>- 'belly' C.T. dJ::>ng L3 Ch. Jl± dho (?) > dhuo *dJJ- In addition to the examples cited thus far, we can show more examples in which Tai words denoting parts ofthe body, in spite ofthe apparent lack offormal resemblance to the Chinese equivalents, may probably have common stems or roots. Let us take 'blood' as an example. The Tai lidt and the Chinese xiwet have apparently no formal relation. But here we cannot completely deny the probability that these two forms are derived from the same *qwhlidt.12 Just as we cannot deny that Tai khem HI ('needle') and Chinese # iidm>tsidm have a common stem, we cannot assert that qhidu H3 ('tooth') has no relation at all with Chinese it ihidg>tshi. Again, Tai mi LI (,hand') is formally very remote from Chinese .:f siog>sidU, but we may safely assume a correlation between m (Tai) and z- (Chinese) in these two words. 'time' C. T. mid L2 Ch. Iftf didg > zi md- (?) 'tree' C. T. mai L3 Ch. itt diu > ziwo md- (?)13 These paralleled correspondences can not be entirely accidental. We must try to detect regular or systematic correspondences underlying the apparent ir regularities between the two languages. In my opinion, out of the above-cited thirty words for body parts, some sixty percent may share common stems or roots. 5. In some cases the Tai form abc corresponds to the Chinese form a'b'c' when aligned with the Ancient Chinese rhyme divisions. For instance, Common Tai aa corresponds to the Ancient Chinese ::~U~ kuo group Div. II Jffffl rna yun. 'horse' IE rna > rna maa L3 1\\9 'tusk' 7f nga > nga ngaa LI14 'value' ~ ka ' > ka gaa L2 'tea' 1k dha > dha }aa LI -ien and -un correspond to rbt~ shan group Div. III 1wffl (sien rhyme) and Div. IV 9dfJ! (hien rhyme) respectively: Div. III 'to be connected' ~ lian > lian lien LI 'to change' ~ plian > pian plien M2 'to cut' !lY tsian > tsian cien MI 12. To this correspond Mak phjaat and Sui phjat, so I infer Labio-Velaire *qWh_ as a common initial phoneme. See T. Nishida 1955: 34. 13. Apart from this mth- can be supposed for 'hand.' From the common forms *md-, *mth-, the shift into Tai and Chinese resembles that of *mth-, *md- in Tibeto-Burmese. *mdugs 'time', *mdong 'tree.' 14. To 'horse' correlate 'to ride' Ch. ,~1a- ghia>ghjie C.T. khii HI, khwii HI and 'saddle' Ch. tfJ( 'an>'an: C.T. 'aan, and to 'tusk' correlates 'elephant' Ch. ~dziang>ziang: C.T. jaang L3. 6 Tatsuo NISHIDA > Div. IV 'hard' ~ kien kien keen M2 > 'flat' h phian phien pheen H2 In comparison with the vocabulary for parts of the body these examples have de finite similarity and most of the Tai words coincide with the Ancient Chinese forms. Hence the strong possibility that these words were borrowed from Chinese into Tai, especially about the time when Ancient Chinese was current. Never theless there are no firm grounds to maintain this. 6. I have so far cited several examples which share the same roots or stems, but I do not think that examples of this kind can be increased so as to enable us to prove the relationship between the two languages. Intervention of various factors makes it hard to prove their kinship. At the same time, however, it should be emphasized that this research contains another significance. That is to say, first ly, as a result ofcomparing the two languages, the Tai form, be it a loan word or not, gives grounds to revise and modify the reconstructed form ofArchaic Chinese, and then, Archaic Chinese forms present a foundation on which to construct the earlier stage of Common Tai. 7. To begin with, let us discuss the first matter. Under Div. I -EIX~ ko rhyme corresponding to the above-mentioned *1~ kuo group Div. II ~~ rna rhyme come the following examples 'left' Ch. II tsa > tsa zaai L3 < draai 'song' Ch. iW\ ka > ka gaai Ll 'that' Ch. t.l~ nar > na naai L3 'fire' Ch. :k xwar > xwa15 vai Ll < vaai? The vowels of the four words, all belonging to ko rhyme in the Ancient Chi a, nese, can be supposed to be but in the Archaic Chinese 'left' and 'song' belong to Class 35 (according to Karlgren's classification), while 'fire' and 'that' to Class 8. We take theformer to be aand thelatterar. The assumption ofarin thelatter is based upon the fact that in Shi king rhyme and Hie sheng characters the words ofthis group come into contactwith the words of-n class and -rclass. xwar rhymes with '!dr (Shiking 154) and dh!dr (212), and 'that' rhymes with gan, x!~m and ulan (215). As opposed to this, Class 35 has no relation with the class which has a final consonant. The assumed -ar in 'fire' and 'that' can be well supported by the fact that Tai aai corresponds to it. But also for the words in Class 35, at least, for 'left' and 'song', we may properly assume *-ar as a result of the com parison with the Tai equivalents, though we have no direct support for this in Archaic Chinese documents. From the viewpoint of comparative linguistics, we would like to make up for the defect that the reconstruction of Archaic Chinese is only based upon Shi king and Hie sheng characters. 15. This Archaic Chinese form canbetracedbackto *xvaror *xbar<sbar(?). AndKachinwan 'fire' corresponds to this *xvar or *xbar. Common Tai and Archaic Chinese 7 Now, to discuss the second problem, let us begin with the following three words 1. 'nine' Ch. 1L kiug > ki;;m T.C. kau M3 2. 'old' Ch. 1j ghiug > ghi;:m T.C. kau M2 3. 'to scratch' Ch. 11 sog > sau T.C. kau Ml These three basic words are, almost without doubt, derived from the same stems. We can give several examples which fall under the correspondences of Ancient Chinese k- (JJ!-ffJ: kien): Common Tai k-; An. Ch. gh- (;flf.-ffJ: kun): C. T. k-; and An. Ch. s- ()~\f~ sin): C. T. k-. They are 'wide'-Ch. fit kwang: C.T. kwaang M3, 'to grow old'-Ch. ghji: C. T. kee M2, 'before'-Ch. :5t sien> si;:m: C. T. kJJn M2, etc. The vowel correspondence of Ancient Chinese idU (jt Myou rhyme): C. T. au can be perceived not only in 'nine' and 'old', but also in 'pigeon'-Ch. 1t~ khidu: C. T. khau HI and 'mountain'-Ch. E: khidu: C. T. khau HI. Besides 'to scratch', there are 'early'-Ch. 1f!. tsau: C. T. Jau 1.3 and 'liquor'-Ch. ~ Iau: C.T. hlau H3>*khlau, which are in the correspondence au of Ancient Chinese (~M hao rhyme): Common Tai -au. We may safely suppose that the above-mentioned correspondences of the three words show firstly that Common Tai k- has three origins, and secondly that Common Tai -au has two origins. In my paper treating the comparative study of the Common Tai and the Mak and Sui languages, I assumed that Common Tai k- is derived from *k- and *kr-l6. The *k- and *kr- show the following correspondences: 1. *k- (a) C. T. k- : Mak k- : Sui q- e.g. 'fowl'-*kai, 'old'-*kau, 'before'-*kJJn (b) C. T. k- : Mak ~- Sui ts- e.g. 'nine'-*kau, 'to grow old'-*kee, 'to collect'-*kep 2. *kr- C.T. k- Mak s- : Sui ts- e.g. 'to eat'-*krin, 'straw sandals'-*kridk Further comparison of these words with Chinese permit us to infer *k- in 'nine' and 'fowl', *g- in 'old' and 'to grow old', *kj- in 'before' and *kr- in 'to eat'. As for the correspondence Ch. s-: C. T. k-, we infer the transitions 'be fore' *kjon>kjdn>sidn 7t, and 'each other' *kjang>siang t§ (C. T. kan). Here we cannot explain under what conditions the *g- of'old' and 'to grow old', which was deduced from the comparison of Common Tai *g- (a phoneme presumed to exist in the Mak and Sui languages) with Chinese, later split up into g- and k-. In the paper already referred to, I inferred three kinds of *-au from Common Tai -au and the corresponding Mak and Sui forms, representing them by-aul -auz and -aus. 1. -aUl C. T. -au: Mak -au Sui -au 16. T. Nishida 1955: 36-37. 8 Tatsuo NISHIDA e.g. 'to howl'-*hau, 'light (not heavy)'-*?bau, 'master'-*cau 2. -auz C. T. -au: Mak -aau Sui -au e.g. 'to take'-*'au, 'horns'-*qhau, 'pillar'-*sau, 'old', 'liquor' 3. -aUg C. T. -au: Mak -u Sui -u e.g. 'nine' When these words are compared with Archaic Chinese three kinds of cor respondence in Common Tai -au can be found, regardless of the related Mak -au, -aau, -;;)u.17 1. a) C. T. -au Archaic Chinese -uk, -uk C. T. khau 'rice' Ch. ~ kuk> kuk khau 'horns' Ch.}fj kuk> kak ·au 'to hold': Ch. W 'uk> 'ak b) C. T. -au Archaic Chinese -iug above-cited 'nine', 'old', 'pigeon', 'mountain' 2. C. T. -au Archaic Chinese -og above-cited 'early', 'to scratch', 'liquor', < 'to howl' C. T. hau khrau (?) : Ch. rut kiog> kieu 3. C.T. -au Archaic Chinese -iU18 C.T. sau 'pillar' Ch. tt dhiu>d.hiu C.T. cau master' Ch.'±' tiu> tsiu The final stem form of the words belonging to the correspondence 1. a) was -uk of -uk in Archaic Chinese, while in Proto-Tai it was -ug, e.g. Proto-Tai 'rice' *kug, 'horns' *khug, 'to hold' *·ug. From this we can infer that there was no difference between the final form of Proto-Tai in 1. a) and that in 1. b). Ancient Chinese m~ yu rhyme (-iu) is divided into two Archaic Chinese forms -iug (Class 31) and -iu (Class 34) according to whether it comes into con tact with -uk, -ok or not in Shi King rhymes and Hie Sheng characters. The 'master' and 'pillar' which fit under 3 above have no contact with the rhymes of-uk or -og class; hence we can infer that they are Archaic Chinese tiu and dhiu. Taking into consideration the result of the comparison with Common Tai, how ever, it seems more suitable to put the two words under -iug class (Class 31), in parallel with the above-stated La.) and l.b.). At least *ciug and *siug must be accepted as their Proto-Tai equivalents. In other words, as in 1.b), opposed to La), these forms are the stem belongs to Div. III rhyme. What is ofgreater interest is the vocabulary with correspondence in 2) above. 17. On the development offinal stops, the forms of Tushan and Sek dialects are very suggestive. Li 1971, Gedney 1970, Haudricourt 1963. 18. Besides this, for ~ yu rhyme, thereare exampleswhichcorrespondto CommonTai-JJ. 'Fa x: ther'-Ch. bhiwo>bhiu: C.T. b::>J L2; 'magician'-Ch. ,m; miwo>miu: C.T. hm::>J HI. HenceitcanbesaidthatthisdichotomyinArchaicChineseiscorrectinviewofthecorrespond ing ofThai. These two examples agree with the Archaic Chinese forms, while the above ex ample 'skin' coincides with Ancient Chinese form. Common Tai and Archaic Chinese 9 These words appear to possess the stern form ofDiv. I rhyme, but in fact, in paral lel with 'to cry' kiog, they were the forms which should be classified under Div. III rhyme with medials -1-, -r- of -log, -rog. 'Liquor' was originally *khlog, as compared with its Tai equivalent *khlau. 'to cry' kiog was derived from *krog, and 'to scratch' sog from *khjog. 'Early' tsog is assumed to have been originally *drog or *trog by the analogy of 'left' tsa: Tai zaai<dzaai<draai. Karlgren invested the main vowel 0 ofog in those words with the force ofclosed 0 near U19• Therefore it can be assumed that the main vowel of the Ancient Chinese forms corresponding to the Tai -au was u. But this main vowel -u- does not correspond to -u ofthe Tai -au. The latter -u can be regarded as the form derived from final -g. It can be assumed that the main vowel u was changed through d into a. This shift is paralleled with the transition from Archaic to Ancient Chinese. 'rice' *khug > khdu > khau 'master' *ciug > cidu > cau 'to scratch' *khjug > khjdu > kau 'liquor' *khlug > khldu > khlau 8. These words correspond to the Archaic Chinese forms rather than the Ancient Chinese ones. Hence there is less probability that these words were borrowed into Tai from Chinese. Nevertheless, even words with these conditions may have been borrowed Chinese words. It can thus be inferred from the following fact. It is almost certain that the series of words indicating terms of the twelve-year cycle were borrowed from Chinese, among which are:20 Archaic Chinese Ancient Chinese Pa-po Ahom Tai-lii > 'ox' 3± thniog thidu pao plao pau3<plau<*thlau (?) 'hen' Wziog > idU rao rao hrau4<hrau 'hare' gp mlQg > mau mao mao mau3<mau We cannot determine when these Chinese words were first introduced into the Tai language, butsince their process ofchange is very close to that ofthe words treated under l.b) and 2, there is the possibility that, if these are clearly Chinese loan words, the latter were also borrowed from Chinese. But I should like to refute this hypothesis. I think it more appropriate to suppose that a series of cyclic terms was borrowed and then incorporated into Tai, and suffered the same change, because they were equal in their syllabic patterns to words with original Tai sterns. Tai and Chinese are quite alike in their syllabic patterns, so that words borrowed from one language into the other followed the same process of shift as the native words did. This fact complicates the work of comparison of these two languages. Concerning the history of the Tai language, there existno documents which 19. Karlgren 1954: 246. 20. T. Nishida 1961, F.K.Li 1945.