To anyone concerned with the question of Bangani,

The issue of Bangani has recently been revived through discussion on the Vyakaran list (as well as elsewhere). The following observations were sent earlier to the Vyakaran (S. Asia) list and may be of some interest to subscribers to the Indoeuropean list, too.

The controversy over Bangani and the authenticity of its apparent evidence for a centum language in northern South Asia does not seem to be coming to an end. In the opinion of some scholars, the claims by Dr. George van Driem and Dr. Suhnu Ram Sharma that their own fieldwork shows Dr. Claus-Peter Zoller's centum forms in Bangani to be spurious has in effect laid the claim -- and the controversy -- to rest. Recent fieldwork by Professor Anvita Abbi (Linguistics and English, Jawaharlal Nehru University, New Delhi) supports Zoller's evidence and in so doing casts doubt on the fieldwork and/or claims of van Driem and Sharma. As a consequence, Zoller's evidence must be taken seriously and its implications for Indo-European comparative linguistics and for South Asian linguistic prehistory must be carefully considered.

As is well known, in the course of fieldwork on Bangani, a language of northern South Asia, Zoller unearthed lexical items that appear to show centum developments of PIE palatalo-velars, instead of the satem outcomes expected in an Indo-Aryan language. At the same time, the language also contains lexical layers that are clearly Indo-Aryan and therefore satem; some of these result from recent influence of languages such as Hindi, others exhibit features typical of the northern languages belonging to the Indo-Aryan/Indo-Iranian family.

Among the forms with centum features are various words derivable from PIE *GenH-, such as OgnOM 'unborn' and gOnNO 'give birth', as well as kOtrO 'fight' (cf. Skt. Zatru-, Gaul. catu- 'battle') and dOkru 'tear' (*(d)aKru). [O = open _o_; G(h), M = nasalization, N = retroflex nasal, z = s with hacek, Z = palatal voiceless sibilant, S = retroflex voiceless sibilant, K = PIE palatalo-velars, uu = long [u:], I = Slav. front jer.]

While some of the forms are marked as doubtful, either by Zoller or by Abbi, and some other forms involve etymologies from Pokorny that many Indo-Europeanists would consider uncertain, there remains an impressive residue. What is especially interesting is that dOkru 'tear', with its initial d-, suggests affiliation with a western Indo-European language (cf. Gk. dakru, Lat. dacruma > lacrima, Germ. Zaehre, Engl. tear), while more eastern members show forms without d-: Skt. aZru, Av. asru, Lith. azara, Toch. B akruuna. More western affiliation is also suggested by lOktO 'milk' and gOsti 'guest (of honor)', which have good correspondences in Gk. galakt-, Lat. lact- and Lat. hostis, Gmc. *gasti-, OCS gostI, but not in more eastern Indo-European languages. Note that these forms do not necessarily contain original palatalo-velars (the fact that OCS has _gostI_ may be attributable to the transition-area status of Slavic and Baltic between satem and western centum languages); but they are nevertheless important, since they suggest western IE (rather, than, say Tocharian or even Indo-Iranian) origin.

Van Driem and Sharma claim that their fieldwork suggests that Zoller's forms are spurious, that some are based on misidentification and others are simply non-existent. In a recent summary of arguments pro and con, Dr. Kevin Tuitte further suggests that Zoller may have fallen victim to fieldwork consultants' tendency to provide evidence that they think may please the investigator. Even a priori, however, the latter suggestion is dubious, since it would be hard to imagine how illiterate villagers would be able to know that words like _dOkrO, lOktO, gOsti_ would please an investigator (to have that knowledge would require more than a superficial understanding of comparative Indo-European linguistics).

In January 1997 I had the opportunity to meet with Abbi and to go over some of her Bangani notes from fieldwork that she recently conducted in situ. She will provide a fuller report on her work in due course, but has asked me to provide a preliminary report, so as to set the record straight. While van Driem and Sharma appear not to have actually entered Bangani-speaking territory but limited themselves to interviewing Bangani speakers on the fringes of the territory, Abbi went into the territory and interviewed, among others, at least one monolingual speaker of Bangani. According to her fieldwork, most of Zoller's forms are genuine.

Her fieldwork also confirms that the lexicon of Bangani contains at least three layers: Words of the type _dOkrO, lOktO, gOsti_, words that exhibit "northern" Indo-Aryan features, and words that seem to be borrowed from more southern Indo-Aryan languages, such as Hindi.

Given these circumstances, Bangani poses several challenges to linguistics.

First, there is the question of what appears to be western centum influence. At this point, the evidence for this influence is highly suggestive; but a larger amount of words of the same type would certainly be helpful to allay worries that we might be dealing with chance similarities. (Zoller's data also contain a number of words in which RUKI apparently fails to apply. But RUKI-_S_ merges with dental _s_ (and with _Z_) in most of Indo-Aryan, and there are well-known problems with RUKI in Nuristani; as a consequence, words of this type do not provide unambiguous evidence -- unless we were dealing with words of the type _dOkrO, lOktO, gOsti_ which, qua words, seem to indicate western IE origin.)

A related question is the nature of the western centum influence. Words like _gOsti_ seem to rule out Greek influence (and thus the possibility that we are dealing with linguistic echoes of Alexander's army); _lOktO_ would eliminate Germanic and Celtic; and _kOtrO_ would eliminate Greek and Latin. That is, no known western centum language could be the source for all of the relevant words. At the same time, the fact that *a and *o exhibit the same outcome (O, no doubt via *a, see below) suggests possible affiliation with the Balto-Slavo-Germanic group (or possibly with Antalolian?).

The fact that *a and *o are reflected as O further suggests that, whatever the source of the words, they participated in the Bangani change of earlier *a to O and that therefore they must have entered (the ancestor of modern) Bangani prior to that change. But that change may be a very recent one. The question of what time these words entered Bangani therefore cannot be satisfactorily answered at this point.

Moreover, it is not at all clear whether the words in question actually entered Bangani, or whether they are part of the original lexicon of the language, and the northern Indo-Aryan lexical layer is a later accretion, comparable to the clearly secondary layer of southern Indo-Aryan words.

It is to be hoped that more extensive field work on Bangani will unearth evidence that will make it possible to answer some of these questions, or at least to make it possible to more clearly establish the nature of the different lexical layers of Bangani and their relationship to each other. Moreover, as noted earlier, the evidence for western IE influence or origin at this point is still rather limited; if more evidence could be found this would definitely strengthen the claim that Bangani contains a significant layer of centum vocabulary.

Hans Henrich Hock
Professor of Linguistics and Sanskrit
Linguistics, 4088 FLB, University of Illinois
707 S. Mathews, Urbana, IL 61801
telephone: (217) 333-0357 or 333-3563 (messages)
e-mail: hhhock@staff.uiuc.edu
fax: (217) 333-3466

Acting Director, Program in South Asian and Middle Eastern Studies 211 International Studies Building, 910 S. Fifth, Champaign, IL 61820 telephone: (217) 0796, fax: (217) 333-6270