In a discussion we were having about the Tibetan evidential particle yin, Nathan Hill sent me an article by Nicholas Tournadre entitled "Arguments against the Concept of 'Conjunct' / 'Disjunct' in Tibetan" from Chomolangma, Demawend und Kasbek, Festschrift für Roland Bielmeier (2008), 281-308. As I started reading through the article with the hope of finding how yin functions as a sort of equational verb or copula, I was caught up short by some preliminary remarks about the classification of Tibetan that Tournadre makes at the beginning of his paper.
Based on his 20 years of field work throughout the Tibetan language area and on the existing literature, Tournadre estimates that there are 220 "Tibetan dialects" derived from Old Tibetan and currently distributed across five countries: China, India, Bhutan, Nepal, and Pakistan. In a forthcoming work, Tournadre states that these "dialects" may be classed within 25 "dialect groups," i.e., groups that do not permit mutual intelligibility. According to Tournadre, the notion of "dialect group" is equivalent to the notion of "language," but does not entail standardization. Consequently, says Tournadre, if the concept of standardization is set aside, it would be more appropriate to speak of 25 languages derived from Old Tibetan rather than 25 "dialect groups."
Tournadre maintains that this is "not only a terminological issue but it gives an entirely different perception of the range of variation. When we refer to 25 languages, we make clear that we are dealing with a family comparable in size to the Romance family which has 19 groups of dialects." (These Romance "groups" are named in footnote 2: Portuguese, Spanish, Asturian-Leonese, Aragonese, Catalan, Gascon, Provençal, Gallo-wallon, French, Nones-Cadorino, Friulian, Venetan, Lombardo, Corsican, Italian, Napolitan-Sicilian, Sardinian, Aromanian, and Daco-Romance.) I would prefer to call Romance a "branch," reserving "family" for Indo-European. The main thing to keep in mind, though, is that, when we refer to 25 "languages," we make clear that we are dealing with a collective entity ("branch" in my terminology) that is comparable in size to Romance which has 19 "groups."
To quote Tournadre directly, "This perspective is quite different from dealing with several dialects of a single language. So I propose to adapt the terminology to reflect the linguistic diversity of the area and speak of Tibetic languages (or groups of dialects) derived from Old Tibetan."
Tournadre goes on to list the twelve major Tibetic languages and the thirteen minor Tibetic languages. Some of the latter consist of only a single dialect and between a few hundred and a few thousand speakers.
These views are by no means idiosyncratic with Tournadre, but reflect the thinking of a major research team headed by Roland Bielmeier at the University of Bern that works on Tibetan dialects. The position of the dozen or so scholars who have been working at The Tibetan Dialects Project of the Institute of Linguistics at Bern is that "Tibetic is a very useful term to designate a very precise group of languages all directly derived from Old Tibetan."
I believe that it is time for Sinologists to take a cue from the Tibetologists. Out of a total population of more than 1.3 billion in the People's Republic of China, there are supposedly almost 1.2 billion speakers of "Chinese." This is held to be a single "language," not a "family," "branch," or "group" of languages, but a monumental, monolithic LANGUAGE consisting of multitudinous "dialects." Since many of these so-called "dialects" are mutually unintelligible, language taxonomists fudge a bit by calling some of them "major dialects," "sub-dialects," and so forth. However, several of the "major dialects" — all of which are mutually unintelligible — have upwards of 20 million speakers: Mandarin (c. 850 million), Wu (nearly 100 million), Cantonese or Yue (around 90 million), Min (over 50 million), Xiang and Hakka (approximately 35 million each), Gan (roughly 20 million), and Jin ([a disputed subdivision of Mandarin] about 45 million). Moreover, many of these huge "major dialects" comprise scores of "dialects" that have a very low degree of mutual intelligibility (or none at all) and a highly complex set of internal relationships. For example, Min is divided into Eastern Min, Southern Min, Central Min, and so forth, and these branches are further subdivided into varieties that are quite different among themselves. For instance, Southern Min is made up of Quanzhou, Zhangzhou, Amoy (Xiamen), Teochew (Chaozhou), Leizhou, and Hainanese, all of which are significantly dissimilar.
Whether or not Sinitic and Tibetic are genetically related, how can it be that there is only a single Sinitic "language" with 1.2 billion speakers of innumerable "dialects," while Tibetic — with somewhere around two million speakers worldwide — is divided into 25 "languages"?
All the usual arguments in favor of Sinitic or "Chinese" being a single language (common culture, common script, common history, common ethnicity, common polity, and so forth) do not hold water. For instance, Chinese characters have been used to write Japanese, Korean, and Vietnamese, but that certainly does not make Japanese, Korean, and Vietnamese Sinitic languages. Conversely, there are Sinitic languages written in other scripts (e.g., Dungan, which is written in Cyrillic), yet they are Sinitic nonetheless. Comparable arguments may be brought forward against all of the other "common" features that are frequently adduced in favor of there being only a single "Chinese" language consisting of a myriad "dialects." In any event, when making a case against the indivisibility of Sinitic, it is not necessary to rebut each of these "common" features individually, since they are largely or wholly extralinguistic.
If we were to apply the same principles of classification to Sinitic that have been applied to Indo-European, Semitic, Austronesian, Austroasiatic, and so forth, we would soon enough find that — like them — Sinitic is a family of languages that may be organized into branches and groups. Adopting such an approach to Sinitic would help not only to illuminate many problems in the historical development of Sinitic itself, but would contribute to enhanced comparison with other language families.