¹ Linguistic survey of India, ed. Sir George A. Grierson: Vol. 3, Tibeto-Burman family; Part 2, Specimens of the Bodo, Naga, and Kachin groups (Calcutta, 1903).

² Robert Shafer, Classification of the northernmost Naga languages, Journal of the Bihar Research Society 39.225-64 (1953). Here Shafer considers not only the languages usually regarded as Bodo but also a number of the northern and eastern ‘Naga’ languages, and maintains that these show a particularly close relation to Bodo. It is true that these ‘Naga’ languages show enough specific lexical correspondences with the Bodo group to make this an attractive conjecture; but even if it should prove to be justified, it would not disturb the coherence of the older Bodo group. The languages of this group are surely more intimately related to each other than to any of the ‘Naga’ languages.

The data presented in this paper are less extensive than Shafer's, in that I am concerned with only four languages while he assembled material from no less than twenty-two. On the other hand, my phonetic data are better than his, and allow me to set up correspondences and reconstructions with a degree of confidence that would be unwarranted on the basis of the older vocabularies that Shafer was obliged to rely on.

Most of the correspondences in this paper were developed before I learned of Shafer's article; it is therefore gratifying to find that many of his conclusions agree with mine. On some points, however, I believe that my data force a revision, or perhaps rather allow a simplification, of the relations that he assumes. For instance, Shafer reconstructs three sets of stops, which he writes *p *t *k, *p' *t' *k', *p'² *t'² *k'². No more than two series are found in any of the languages that I deal with, and for the most part there is a simple correspondence between the series of the several languages. It is true that one set is uniformly aspirated except in Kachari, where its members have unaspirated allophones, and that the other set is voiced in some of the languages and voiceless in others; but in any single language the distinction is always clear. Since workers accustomed to recording the languages of India are attuned to hearing the distinction between aspirate and unaspirate, they sometimes note it even for languages where it is nondistinctive. Such records may have led Shafer to believe that there were three contrasting sets of stops in the languages of this group as well.

Similar inconsistencies in the original recordings, which are extremely hard to interpret, may have caused Shafer to set up his two types of *n and *w, and his astonishing array of four different *l's.

³ The data used in this paper were collected while I held a Ford Foundation fellowship, for which I here express my profound thanks. I am greatly indebted to Henry M. Hoenigswald and Nicholas C. Bodman, who both read an earlier version of the paper and offered helpful suggestions.

⁴ J. Burton-Page, An analysis of the syllable in Boro, Indian linguistics 16.334-44 (1955); P. C. Bhattacharya, Glimpses from Boro folksongs, ibid. 17.240-4 (1957); Sidney Endle, The Kacharis (London, 1911); id., Outline grammar of the Kachari (Bårå) language, as spoken in District Darrang, Assam (Shillong, 1884).

⁵ William Cornyn, Outline of Burmese grammar (Language dissertation No. 38, 1944).

⁶ Paul Benedict, Studies in Indo-Chinese phonology, 2: Tibeto-Burman final -r and -1, HJAS 5.114-27 (1940).

⁷ In addition to sources mentioned earlier, note the following: Major A. Playfair, The Garos (London, 1909); J. F. Needham, Outline grammar of the Singhpo language (Shillong, 1889); Robert Shafer, The vocalism of Sino-Tibetan, JAOS 60.302-37 (1940), 61.18-31 (1941). Shafer's transcriptions unfortunately are sometimes rather bizarre; but he has labored so prodigiously in searching through existing dictionaries and word lists, and in assembling apparent cognates, that I can only be thankful for his industry.