Uralic & Altaic Language and Archaeology

There are at least two issues raised by this wide-ranging paper that seem to invite comment. The first concerns what I take to be critical hyperbole regarding the identification of either the steppe cultures or the Bactrian Margiana complex with Indo-Iranian. While we may agree that there is no one-to-one relationship between material remains and language, there are still degrees of geo-linguistic plausibility. For example, that Altaic, Ugric (I think Lamberg-Karlovsky must mean Finno- Ugric here), and Elamo-Dravidian have equal claims on the areas involved with Indo-Iranian is not impossible but surely not very probable. From the earliest written testimony of the 1st millennium b.c. it appears that almost the entire region discussed was occupied, according to tribal and personal names and the occasional item of vocabulary, by the eastern branch of Iranian We can chart the later evidence for Turkish (Altaic) presence in this area, so it is at least very unlikely that Andronovo was Altaic. I know of no one who assigns the Finno-Ugric languages to the steppe lands (the reconstructed vocabulary of all the branches is adamantly arboreal and reflects largely a hunting-gathering- fishing economy, with later Indo-Iranian loanwords providing our earliest evidence of stock breeding and agriculture [see Napolskikh 1997]. That the Bactrian Margiana complex was Elamo-Dravidian is possible (see below), but making the steppe cultures Elamo-Dravidian (up to the end of Andronovo at ca. 900 b.c.?) would require an archaeological performance of Von Da¨niken like proportions to get the Indo-Iranians from wherever one wants to stash them for several millennia to their historical locations. The same goes for making the Andronovans a language family that became extinct before we have any written evidence. The only way out of this corner is either to accept Renfrew’s (1987:189–97) “Plan A,” which sees linguistic continuity in India and Iran from the spread of the Neolithic (a position that he himself has rightly abandoned [e.g., Renfrew 1999:280–81], or to assume that the Indo-European homeland itself was in South (Misra 1992) or Central Asia (Nichols 1997), models which throw up other problems so enormous that they make the Indo-Iranian issue look like child’s play. Andronovo and the Bactrian Margiana complex really do appear to be the only game(s) in town, and Lamberg-Karlovsky has indicated why they appear to be mutually exclusive solutions, neither capable of resolving the problem by itself. There is no single culture (using the word in the widest—even most unjustified—sense) that can link the Indus, Iran, Central Asia, and the steppe lands together. As Lamberg-Karlovsky indicates, a solution that rests on some “form of symbiosis” between the steppe tribes and the Bactrian Margiana complex remains “utterly elusive.” For some time I have tried to inch forward on this front because I too have felt the inadequacy of employing the spread of material culture as proxy evidence for linguistic movement. The problem here is not just the longrehearsed criticism against assuming that potsppeople/ language but that there are clear instances, the Indo-Iranians being a case in point, in which there is no hint of the distribution of any archaeological assemblage that might correlate with the distribution of the target language group. The situation is so dire that we can’t even make the type of mistake Lamberg-Karlovsky warns us about! Working from first principles, it has seemed to me that archaeologists engaged in tracing linguistic entities need to concentrate on the archaeological manifestation of language shift, and this may be independent of the trajectories of material culture (Mallory 1992, 1998b). Language shift may occur in some obvious instances of subsistence differences, for example, the Neolithic models posed by Renfrew (1987) and Bellwood (2000), but in many instances I suspect that we are dealing with language shift due to social differences that are not obvious in the archaeological record. Impressed by Ronald Atkinson’s (1994) treatment of the spread of the Acholi in Uganda, I think that we should be looking for evidence for competing social organizations and attempting to predict which would most likely bring about linguistic shift and expansion. Lamberg- Karlovsky (1994) has already reviewed the ethnohistorical evidence for suggesting that the Bactrian Margiana complex might have been organized as a khanate. The four-tiered political system evident in Central Asia in historical times bears a close resemblance to the fourtiered political structure reconstructed for Proto-Indo- Iranian by Emile Benveniste (1973). What is of considerable interest is the highest tier, the one which would incorporate the greatest number of people—the ∗da´syu (Old Indic dasyu, Avestan dahyu). In a recent study Lubotsky (n.d.), this term has been regarded as a non-Indo- European substrate term that was borrowed into Proto- Indo-Iranian along with a series of other words associated with religion (words for “priest,” “magic,” deities, and even “soma”). Lubotsky has suggested that these words may have had a Central Asian source, and it seems to me that the Bactrian Margiana complex, with its elaborate (to a steppe pastoralist) ritual architecture, would make a plausible candidate (and if Elamo-Dravidian, we might at least hope for some lexical correlations between the putative loanwords and their proposed sources; otherwise, these people may have spoken a language that has not survived).My current model, admittedly no more supported by archaeological evidence than the previous discussion alluded to by Lamberg–Karlovsky, is to assign some form of Indo-Iranian identity to the Andronovo but see their expansion southward in terms of their adoption of both political and religious concepts (including material manifestations of these concepts) from the Bactrian Margiana complex. The spread of Indo-Iranian languages then would be through the vector of the Andronovo culture on the steppe but by way of the Bactrian Margiana complex to its south. Steppe tribes that came into contact with the Bactrian Margiana complex would be required to retain their language (Indo-Iranian) but would gain a more incorporative social organization from their neighbours as well as a series of religious concepts and practices, perhaps in the same way that the Acholi of Uganda retained their Luo language but gained from their Bantu neighbours the more incorporative chiefdom system which permitted them to carry both their new social organization and their own language to the north (Atkinson 1994). Both the social and the religious organization (see Erdosy 1995) of Bactrian Margiana–complexinspired Indo-Iranians would then become the vector for language spread southward. Obviously, all of this would require far more intimate relationships between the Andronovo and the Bactrian Margiana complex than the existing distribution of “mutually exclusive” material culture would permit, and what is clearly at stake, as I suspect Lamberg-Karlovsky would agree, is our confidence in our ability to read the record of social processes from the archaeological record.
J.P. Mallory School of Archaeology and Palaeoecology, Queen’s University, Belfast, Northern Ireland