The Uralic Language Family: Facts, Myths, and Statistics

Though titled The Uralic Language Family, this book's main thesis is that no real evidence supports the commonly held view that such a family actually exists. According to M, the various sub-groups of what is traditionally regarded as "Uralic" cannot in fact be demonstrated to form a genetic unit. M discusses what she believes are three fundamental problems with the Uralic Hypothesis. First, Hungarian is not demonstrably related to the two Ob-Ugric language clusters in Western Siberia (Khanty and Mansi), which means there is no "Ugric" node in the Finno-Ugric side of Uralic. Second, the glaring inability of linguists to reconstruct Proto-Uralic convincingly at any level of structure indicates that Samoyedic, Finnic, Saamic, Ob-Ugric, and the Hungarian isolate should be regarded as separate families (it is unclear to me whether M supports the genetic unity of Permian and Volgaic with the rest of Finnic). Third, because each of these groups, in turn, displays well-known, though usually ignored, individual affinities with other Eurasian families—notably Yukagir and certain Turkic, Mongolic, Tungusic languages—some of the Uralic sub-groups may eventually prove to have closer areal or even genetic ties with non-Uralic languages. Amassing evidence from a variety of disciplines and, more importantly, employing the impact of what M sees as the absence of convincing linguistic proof despite more than a century of attempts to describe Uralic in terms of the family tree model, the author calls for a "paradigm shift" (p. 278) to demolish the "myth" of Uralic genetic unity.

My review will argue that in making her arguments, M tends to minimize the best evidence—primarily lexical—that supports Uralic as a valid genetic node, though one whose constituent branches have undergone extensive areal contact mutually as well as with non-Uralic languages. Still, even if one accepts Uralic as a family on the basis of shared basic vocabulary, then M is undoubtedly correct in emphasizing that it is a family quite unlike Indo-European, for which much of the morphosyntax as well as core vocabulary can be systematically reconstructed with some confidence. This fact starkly contradicts the received opinion commonly held even by non-Uralicists, who cite Uralic as one of the few language families for which the family tree model is clearly applicable (cf., for instance, Dixon 1997:28). Mis also correct in urging Uralicists to work at untangling Eurasia's many contact-based relationships instead of attempting to reconstruct an elaborate Uralic proto-language, a task that appears to be impossible whether one accepts Uralic or not. Despite being fully justified both in casting a wider net of historical inquiry at Eurasian languages in general, and in paying greater attention to language mixing in particular,M fails to make a convincing case that any subset of Uralic is actually more closely related genetically to languages outside Uralic itself. Despite M's numerous valid criticisms of individual facets of the Uralic Hypothesis, most of which I can accept, I am unable to reject Uralic as a genetic grouping due to the existence of a core of largely Uralic-specific basic vocabulary items. These include such words as for 'two', 'eye', 'tongue', 'nest', and several dozen others, which—despite the impossibility of reconstructing their precise proto-forms—are nevertheless unlikely to have acquired such a far-flung distribution through language contact involving originally unrelated languages.

It is probably for this reason, too, that Uralic is accepted as a "proven" language family even by linguists such as R.M.W. Dixon (1997) who tends to be extremely skeptical of any newly proposed genetic groupings and rejects all deep genetic families out of hand. Yet in recent years there has risen a growing chorus of voices among Uralicists in opposition to the traditional hypothesis. These include T.-R. Viitso (1996), A. Künnap (2000), and R. Taagepera (2000), all of whom argue that areal rather than genetic factors are the prime historical motivations behind the observed similarities among the Uralic sub-branches. These and other dissenting Uralicists seem to be arguing that Uralic itself should at best be regarded as a sort of "language mesh" in the sense defined by Fortescue (1998) when entertaining the possibility of some sort of ancient affinity involving Uralic, Eskimo-Aleut, and several Northeast Asian groups. It is possible that linguists such as Taagepera (2000) are basically correct in describing Uralic as a "lingua franca with roots", and yet the main ingredients of the Sprachbund to which they allude are still genetically related and thus ultimately "Uralic" in a meaningful sense. This is the view I am inclined to support.

M surveys the earliest historiography to demonstrate that serious doubts about Uralic genetic unity date back to the very beginnings of Uralic studies, but have largely been brushed aside by mainstream linguistics. She revisits the data used by J. Budenz, a key original proponent of Uralic over a century ago in his successful polemic with Á. Vámbéry and other skeptics, and finds his argumentation completely inadequate by modern standards. M then argues that the Uralic Hypothesis was subsequently perpetuated more through popular belief rather than on the basis of hard evidence that has subsequently stood the test of time.

At the core of this book is a discussion of the uses (and abuses) of the Comparative Method in light of the Uralic proto-language problem. Here, one can only applaud M's insistence on employing real language data rather than abstractions such as proto-forms, which are often skewed by the reconstructor in ways designed to facilitate proof of the very hypothesis being argued for. M uses a battery of statistics to demonstrate—insufficiently in my opinion—that pan-Uralic cognates are too few in number and too frequently shared by languages outside Uralic to be diagnostic of a unique genetic grouping. The core lexical evidence alluded to can be found in two appendixes (pp. 280–283). These lists contain several dozen basic vocabulary items, most of which are still, in my opinion, most likely to stem from a common source because they are found in all or nearly all Uralic subgroups; also, the presence of random look-alikes in other language families is usually too unsystematic to cast doubt on the specifically Uralic provenance of these words (items such as 'fish' and 'name' being the exception, since apparent cognates occur widely both in and beyond Uralic).

M discusses at length the greater difficulty of finding pan-Uralic cognates in morphosyntactic systems. This fact offers a more serious vantage from which to attack the Uralic Hypothesis. Nevertheless, even here there are sufficient grammatical commonalities within Uralic if not across all of Uralic at least to suggest a genetic explanation. The subjective vs. objective verb conjugations are one typological trait that comes to mind. The systematization of much of this evidence can be found elsewhere (cf., for example, Abondolo 1998), and I will not repeat it here except to say that M does not discount all of it. The suggestion that language contact may have altered much of the original grammatical structure inherited from proto-Uralic strikes me as more plausible, though M is fully justified in arguing against positing the bygone existence of traits that cannot be demonstrated anywhere in the primary data just because believing in them is convenient for the particular hypothesis at hand. This issue notwithstanding, it is still the lexical rather than morphosyntactic evidence that lends the strongest support to a unique genetic origin for Uralic languages.

M makes her most convincing proposals with regard to the problems attendant in trying to group Hungarian with Ob-Ugrian. She demonstrates the spuriousness of the belief that the proper nouns Mansi and Magyar, and Hungary and Yugria¹ are cognate, and alleges that this false assumption seems to have provided the original impetus for grouping Hungarian, Khanty and Mansi into a unified "Ugric" sub-node of Uralic. Although there are similarities in basic vocabulary among the three "Ugric" languages of the type found elsewhere in Uralic, there is scant indication these languages share any special affinity that would warrant their unique sub-grouping within the family. Perhaps the best piece of lexical evidence can be found in basic vocabulary such as the shared terms for 'three', 'twenty' and several other simple numbers alongside pan-Uralic 'two'. Here, interestingly, Mansi basic vocabulary patterns consistently closer to Hungarian than either language does to Khanty (again in sharp contradiction to the traditional view that Khanty-Mansi separated only long after the separation of Hungarian). The shallow time depth proposed by Honti (1998:327) for the break-off of Hungarian (2,500 years ago) and the division of Khanty and Mansi (less than 2,000 years ago) seems unlikely. More importantly, the traditional tree-branching internal structure of "Ugric" seems overly simplistic. Khanty and Mansi, like Hungarian, definitely deserve more analysis from the perspective of language contact. Ethnographic evidence suggests that the Khanty and Mansi people are cultural and physical amalgams of forest-dwelling tribes and steppe horse-breeders. It is plausible that the latter ethnic component involved speakers of a Hungarian-type language and that significant language mixing occurred in the history of these two peoples. The received notion that Uralic contains a discrete Ugric node consisting of Hungarian plus the more closely related pair of Ob-Ugrian languages should probably be discarded in favour of a more complex model involving major contact events, possibly along the lines of the multi-trunked Sprachbund tree proposed by Taagepera (2000). Based on lexical evidence, however, I would still place all three "Ugric" languages clearly inside Uralic, even if not in a unified node of their own.

In her discussion of Ugric, M further shows that Hungarian shares significant commonalities in proper names with the Turkic language Bashkir, an ethnonym that offers a more plausible candidate as cognate to Magyar than does Mansi. The Hungarian-Bashkir connection is perhaps best explained, however, as resulting from Hungarian influence on Bashkir, rather than as evidence for deep Hungarian-Turkic affinities of the type M suggests. However, as is the case with Ob-Ugrian, M's treatment of these aspects of Hungarian linguistic history is plausible enough to raise doubts about traditional interpretations; but her evidence is insufficient to support any concrete alternative proposal.

Another noteworthy aspect of this book is M's many allusions to other possible genetic groupings involving some or all of the Uralic languages (though little hard evidence for any of these linkages is actually spelled out). Similar ideas have been pursued with more vigor by scholars like Joseph Greenberg, as well as by the Nostraticists, but these linguists have worked entirely within the family tree model, taking little or no stock of areal contact as a major historical factor. M is right that pan-Eurasian lexical and grammatical similarities must be re-examined seriously from both a genetic and areal perspective without being circumscribed by the preconceived boundaries of assumed family trees, which tend to raise some of the commonalities to the level of evidence while blatantly ignoring others.

Finally, M brings to the discussion interesting parallel information from physical anthropology, archeology, Eurasian history, and several other disciplines to show that there is no convincing evidence to support the existence of a unified Uralic people, culture, or gene pool anywhere in the historical record. Nor is there any evidence to support the spread of Uralic-speaking peoples northwestward from South Siberia, as is traditionally assumed; in fact, all of the known data about pre-pastoral northern Eurasia indicates the movement of peoples, cultures, and technologies in the opposite direction, from Europe to Asia. This means that Uralic is either very old (Early Holocene would be my assumption, in conjunction with the first peopling of much of this area) or it never existed at all, which is M's contention.

Despite my rejection of M's central thesis that Uralic is not a genetic family, I still accept much of the skepticism contained in her ground-breaking study. I also urge both Uralicists and non-Uralicists alike to take this book seriously. M succeeds admirably in shedding doubt on many widespread, yet apparently indefensible assumptions about Uralic languages. However, she does not replace them with any new ones of her own. It remains to be seen whether this thought-provoking reassessment of facts, myths and statistics in Uralic linguistics actually produces the kind of paradigm shift for which its author is arguing. Nevertheless, it is now obvious that no advance in Uralic studies can occur unless Uralicists tackle head-on the many unresolved issues that M brings so eloquently to the fore.