The Classical Tibetan Language.

• Author: Miller, Roy Andrew

The publication of a new and impressive attempt at a d hardly proves that h was zero: the question that must first be asked is, how else could the Tibetans have rendered foreign diphthongs except by using h? All these problems about h have already been more than adequately discussed and explained in the existing literature. B should have used his energy and the space available in his book for more of his own original contributions, in which he often shows himself a master, rather than, as in this matter of h, unsuccessfully attempting to reopen old, actually well-settled, controversies.

Any grammar is a linguistic document; and confronted with a linguistic document one will, these days at least, always have to ask, what kind of linguistics is it? Fortunately, B's enthusiasm remains firmly rooted in Tibetan itself, not in displaying his familiarity with the latest fads in linguistics, so that we are spared transformations, generations, deep structures, and stratifications as well. And it would probably be reading more into B's text than he intended if one were to suggest that his frequent (and mostly quite general) allusions to "semantic" this-or-that betray any serious inclination toward the doctrines of the latter-day neo-Chomskyites. Actually, the linguistics one finds in these pages is by-and-large good old-fashioned American descriptivism of the 1950s, generally the gospel a la Bloomfield, though he is never mentioned, and (as already noted, supra) new terms that help obscure many links with the past have been coined whenever possible (e.g., "equative" [pp. 253, 255-58] for the copula).

Even "Immediate Constituent" syntactic analysis, the Bloomfieldian "Great Satan" that was the first of the fallen angels to bite the dirt at M.I.T., is readily and illuminatingly invoked in these pages (e.g., pp. 191 n.; 193, 200 n.; 303 [a particularly spectacular example of I-C bracketing!!, and passim; it might also have been invoked and would have facilitated the description at pp. 235 n. and 241, as well as at p. 272, where far more than "change in word order" is at issue). In a word, B's linguistics is old-fashioned, solid, reliable, and useful; only the names have been changed "to protect the innocent."

Especially in view of B's almost entirely sound and traditional approach to questions of general linguistics and linguistic theory, it is a particular disappointment to find his analysis frequently running aground on a major methodological block-of-stumbling that often threatens to vitiate his entire voyage.

The language that B is describing is a written language. It is a dead language. No one now alive has ever heard it. Its written records give us data only about its segmental phonology (vowels, consonants), but tell us nothing about its stress, tone, pitch, or other presumably related suprasegmentals.(12) Linguistic analysis - "grammars" - can operate only with data that exist, not with entities that are missing from our texts, whether the texts are older written records or contemporary field-notes. Nevertheless, B early (pp. 5, 74 n., 90-92) and throughout his grammar attempts to anchor vital segments of his description and analysis of Written Tibetan onto a chimera, onto something that we do not know anything about because it is not in our texts, i.e., "stress." We are asked to believe in an analysis keyed to differences between "a single Stress Group" and "a Disyllabic Stress Group"; this latter "sometimes creates a Compound, which is a new stress group with a new meaning" (p. 91). We are told of "the basic disyllabic rhythm of the language" (p. 190 n.); but what we are never told is how B knows all these details about the suprasegmental features of what he at the same time (and quite correctly) insists is "above all a written rather than a spoken language ... freed from the constraints imposed by the transience and noise of spoken messages" (p. 194) - freed also, one may only add, from all the information and any of the linguistic-analytic clews that may well have been present in the now-lost anht have added that the term was once commonly used by many in Tibetan studies as well (including the reviewer), who may or may not have been "in conscious imitation" of anything. At any rate, my present considered view (and I think that of many others) is that this Sinological usage is itself highly suspect, and the less it is imitated the better. At least we know texts called (probably too loosely!) the "Chinese Classics"; but what are the "Tibetan Classics"? The language that B writes about so vividly and whose grammar he explicates so interestingly is "the language of written Tibetan texts" (p. 36); why not simply and unambiguously call it "Written Tibetan"? Nor is the problem ameliorated by random references to a lower-case "classical Tibetan" (e.g., pp. 37, 70 n.). On the diachronic scale we find Old Tibetan (p. 10 n. and passim) and Middle Tibetan (pp. 132 n., 139 n., 146 and passim). The ostensibly clear-cut parameters of the former are blurred by frequent and overlapping references to "archaic manuscripts from Central Asia," which never make clear how (or, if) their language differs from Old Tibetan; and so also for what is dubbed "an archaic chronicle found near Tun-huang" (p. 202). Middle Tibetan becomes an enormous catchall term covering a full nine centuries (p. 19), and Proto-Tibetan (in the usually accepted sense of the term) is confused with Pre-Tibetan (p. 10 n.).(3) The best, because it is the most straightforward, categorization of Old Tibetan (as distinct from Written Tibetan) defines it as that early variety of the language that had six or seven vowels (a, i, i, u, e, o, and possibly u); an automatic yod-glide between labials and i, i;, and the so-called da-drag. But the "extra" vowels that figure in every Old Tibetan manuscript are never mentioned ("There are five vowels in Old Tibetan - u, o, a, i, and e," p. 55); the yod-glide is treated only superficially and in passing; and the references to the da-drag (pp. 168 n., 175-76, 187, inter alia) are scattered and perplexing, never clearly making the essential point that this phenomenon too represents one of the critical distinguishing features of Old Tibetan phonology.

Related problems involving terminological precision similarly complicate B's discussion of the word, word-classes, and word-categories. Here the potential for serious misunderstanding is great indeed, particularly when B begins to base his grammatical analysis on meaning (which is generally a perilous pursuit) and at the same time substitutes translation into English for linguistic meaning (which is always fatal). His [section 7], "Words" (pp. 97-159) is an important discussion, and central to his grammar, but it is shot through with internal contradictions arising from an essential confusion between `meaning' and `[English] translation', which of course are not and never can be one and the same. No amount of trendy "new-speak" terminology, whether "the reading of texts" (p. 1), or "lexicalization" (p. 4), or "processing" (pp. 197, 199 n.), or "semantic nucleus" (pp. 137-38), can disguise the rents and fissures that arise in B's argumentation when he attempts to pour the new wine of his often genuinely original and useful analysis into the old bottles of translation-based grammar: immediately they leak badly. Nor does the invocation of the mental processes of "the Tibetan reader" (p. 304) do much except remind one of Hermann Paul's quest after this same will-of-the-linguistic wisp.

References to "semantically exocentric units" (pp. 92 n., 101 n.) and the like show that B has read (or heard) something of Bloomfieldian linguistics, but then end up doubly disappointing when the promise they hold out of a rigorously descriptive form- and syntactical-based system is betrayed by the trivializations inherent in "meaning = translation" (e.g., "the head member may not substitute for the whole collocation without changing the meaning of what is asserted," p. 102 n., which really only means, "without changing the usual Eibetan is more distantly related to Burmese; even more distantly to languages spoken by Naked Nagas and other hill tribes of Assam; and more distantly still to Chinese" (p. 7). But his examples of what he calls "Systematic Correspondences" between Chinese and Tibetan are the usual ones from the older literature, and mostly self-destructing: "the word[s] for 'three' (Old Tibetan gsum Old Chinese *s??m) ... both begin with a dental fricative" (p. 9). But of course they do not, and no amount of historical-phonological Turnubung can ever get the g- out of the Tibetan or add it to the Chinese.

The striking historical evidence that doublets such as Written Tibetan dgu ~ rgu '9' (noted p. 75 n., but there ascribed to "phonetic weakening"!) provide for the possibility that all the Tibetan numerals are nothing more than the result of loans from late Old Chinese into pre-Tibeto-Burman at a period just prior to a later and often diverse prefixation within some of the borrowing languages (a process that early proliferated in the Tibetan domain of Tibeto-Burman but one that at the same time atrophied in the Burmese sector) is passed over in silence. Thus, Chinese *sam '3', *nguo '5', and *kieu '9' would, for example, have originally been borrowed as *sum, *nga, and *ku (these forms are directly and quite simply reflected in Written Burmese sum, nga), and kui); but pre-Tibetan began to add various prefixes, and to yield g-sum '3', l-nga '5', and variously *d- + *ku [greater than! dgu ~ *r- + ku [greater than] rgu '9', etc. What begins (p. 8) as a promisingly succinct account for students of the comparative method of the Neogrammarians too soon thereafter crashes in flames ("the only way actually to demonstrate that two or more languages are cognate descendants of a common ancestral language is to reconstruct the common language from which they descended," p. 9). If this were...