A New Look at Altaic.

• Author: Norman, Jerry

The goal of the enormous etymological compendium recently completed by Sergei Starostin and his collaborators is to convince its readers of the existence and validity of the Altaic language family. These three thick tomes represent by far the most ambitious work of this kind to date and are unlikely to be surpassed anytime soon, at least in terms of volume. In the conception of the authors, Altaic comprises not only the conventionally recognized Turkic, Mongolic, and Tungusic groups, but Korean and Japanese as well. The introduction contains five chapters which introduce the comparative phonology of Altaic, a section on comparative morphology, and the classification and dating of Proto-Altaic. Four further-sections provide articles on the structure and conventions of the dictionary, a bibliography, and abbreviations. The bulk of the work (pp. 267-1556) is a comparative dictionary arranged alphabetically according to the authors' Proto-Altaic reconstructions. The entire third volume is devoted to indices of individual languages as well as reconstructions.

Altaic comparative studies have a long and controversial history. The basic phonological correspondences among Turkic, Mongolian, and Tungusic were worked out by G. Ramstedt (1957) and Nicholas Poppe (1960). Subsequently their work was subjected to prolonged criticism from several scholars, among whom Gerhard Doerfer (2004) was one of the most insistent. The chief criticism has been that the majority of the cognates proposed by Ramstedt and Poppe can be considered loanwords, onomatopoeia, or chance resemblances. The conelusion of such criticism has been that Altaic is no more than an areal of Sprachbund phenomenon and cannot be considered a valid genetic grouping. Although this negative criticism has been very influential, leading almost to a consensus that no Altaic language family exists, supporters of the Ramstedt-Poppe theory have by no means disappeared. Starostin and his colleagues must be added to the list of those who defend the validity of the Altaic theory. Even among the supporters of Altaic as a valid linguistic family there is considerable disagreement about whether Japanese and Korean should be included in the family. The classic treatments of Ramstedt and Poppe cautiously included Korean in their foundational works but excluded Japanese. Starostin and his co-authors strongly support the inclusion of both Korean and Japanese in their genealogical scheme.

The authors of the Etymological Dictionary of the Altaic Languages (EDAL) consider Altaic a language family of great time-depth, extending back to the fifth millennium B.C., older than Indo-European by at least a thousand years. They view Altaic as consisting of three principal groups: Turko-Mongolian. Manchu-Tungusic, and Korean-Japanese. Apparently this view is based on lexicostatistical considerations (pp. 230-34). This conceptualization of Altaic subgrouping strongly affects the entire framework of this dictionary. Whereas most earlier Altaicists based their comparative work on the core groups of Turkic, Mongolic. and Tungusic, Starostin and his colleagues integrate Korean and Japanese into their reconstruction at the most basic level. This idea leads to a number of radical proposals concerning phonology.

The authors of EDAL, following a theory of the nostraticist, V. M. Illic-Svityc (1963, 1965), posit a three-way distinction for stops and affricates at the Proto-Altaic (PA) level. For example, the following correspondences are found for dental and velar stops.

PA Turkic Mongolic Tungusic Korean Japanese *t' *t *t *t *t *t *t *d *d *d *t *t *d *j *d *d *t *t *k' *k k *x *k *k *k *g *k *k *k *k *g *g *g *g *k *k Earlier Altaicists recognized only a two-way distinction for stops and affricates, *t and *d in the case of dentals and *k and *g in the case of velars. There can be little doubt that Illic-Svityc was influenced by his ideas concerning Nostratic, a super-family that includes Altaic, Uralic, and Indo-European among several other language groups--Afro-Asiatic, Kartvelian, and Dravidian. Note, first of all, however, that Korean and Japanese show no evidence for a three-way (or even a two-way) distinction. Moreover, the nature of the distinction is different in the two cases.

Without going into all the details, I personally find Illic-Svityc's three-way distinction for Altaic highly debatable. Although both V. I. Cincius (1949) and Johannes Benzing (1955) reconstructed a contrast of *k and *x for Proto-Tungusic, when we examine words beginning with *k in the Sravnitel'nyj Slovar' Tunguso-Man'czirskix Jazykov (TMS), we find that a disquieting number of these words have either a very restricted distribution, are loans from Mongolic, or are expressive or sound-symbolic in nature. It seems to me plausible that Proto-Altaic *k regularly became *x in Proto-Tungusic, leaving a distributional vacuum for a velar stop and that subsequently a *k was restored from various sources. Something similar happens in the history of Japanese: Proto-Japanese *p becomes h in Modern Japanese but initial p- still persists in many loanwords and onomatopoeia. Or we might took at Manchu where Proto-Tungusic *p becomes f; yet in Manchu there are still a significant number of words that have an initial p-; such words, it turns out, are almost all loanwords (chiefly from Chinese) or onomatopoeia. The possibility of isolated archaism can also not be...