linguistlist.org>
Editor for this issue: Martin Jacobsen <martylinguistlist.org>
worldnet.att.net>On March 5, "Henry M. Hoenigswald" <henryhMail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue
umich.edu>I dont know what exactly I did to offend Robert Ratcliffe, whom I do not know personally or professionally, but I apologize just in case. However, he does not cite any evidence in support of what his rather severe strictures. As for me, my impression about the state of Semitic studies (namely, that there are very few active Semitic LINGUISTS, that the field is dominated by philolo- gists, and that the philologists overwhelmingly doubt the reality of protolanguages (even Proto-Semitic) and increasingly voice doubts about the Afro-Asiatic language family) is based on extended discussions on the ANE list as well as discussions I have had with current or recent graduate students or faculty at some leading American departments, and the reading of some recent introductory texts. I dont of course mean to say that Semitic linguistics is dead, but I think the general linguistics audience does need to alerted to the fact that there are books entitled 'Introduzione alle lingue semitiche' or the like which preach that the Semitic languages are NOT related to Afro-Asiatic. Regarding Altaic, Dr. Ratcliffe, while admittedly not an expert, appears to have fallen in with experts on some of these languages who either do not accept the Altaic theory or else either do not have the skills or the will to evaluate the recent work pro or con this theory by, e.g., Doerfer, Janhunen, Vovin, Starostin, and others (incl. myself). Again, I do not mean to deny that the Altaic issue is still not entirely settled. All I want to do is alert general linguists to certain things of general importance. Although I support Altaic, I decry the lack of knowledge about just whatthe critics of Altaic havebeen saying just as much as I do the lack of knowledge of what my side has. As for Dr. Ratcliffe's attack on my characterization of Nichol's book, let me quote from this book (1992:4): "the evidence [sc. for Altaic] was reduced to the pronominal root resemblances and a set of putative cognates. When the cognates proved not to be valid, Altaic was abandoned and the received view now is that Turkic, Mongolian and Tungusic are unrelated" Now, this is precisely what I said she said, isn't it? And my point, which Dr. Ratcliffe does not bother to refute or challenge, is that (a) this is incorrect and (b) it is important for general linguists to know (a). However, books like this reach a much wider audience of linguists than do books and articles about the Altaic issues, whether written by pro- or anti-Altaicists, and so such misinformation is liable to be hard to live down. And THAT is my main concern, that general linguists do not get proper information about the state of comparative linguistics and perhaps for that very reason comparative linguistics is being gradually eliminated as a subject of study in our linguistics departments. Not being part of a such department myself (I teach computer science), there is little I can do "locally", but I am hoping that through LINGUIST and other forums (e.g., the forthcoming article on the state and history of Altaic in the Journal of Linguistics), I can help do something "globally". There is, of course, room for disagreements in comparative linguistics, as in any field. What I am opposed to is the wholesale elimination of teaching positions in the field and the wide dissemination of misinformation by people who are not themselves comparative linguists working on the languages at issue combined with the lack of information on what is really going on in comp. ling. that would be readily available and comprehensible to the general linguistic public. In saying this, I am not appointing myself a "spokesman" for comparative linguistics. I agree with professor Hoenigswald that comparative linguists have done themselves more harm than good in some ways, though we probably disagree about the specifics. I think the main problems are that (a) comp. linguists never articulated a coherent intellectual response to the view of many linguists who subscribe to Noam Chomsky's views that, since ling. is all about the learning of lg by a child, comp. ling. has no relevance to lx (a view I am not at all sure Noam actually shares), (b) comp. linguists have continued to publish their findings in (increasingly) specialized (and hence obscure) places w/o any attention to the need to keep the rest of us informed, (c) comp. linguists do not seem to have staged any kind of a fight when for example Language, which used to publish tons of comparative work, stopped doing so, (d) comp. linguists have not produced quality textbooks of historical linguistics, (e) comp. linguists have allowed internal debates (e.g., over Altaic or Amerind) take priority over the fate of the field, and have been more concered about trying to "shout down" others in the field than about having the field survive, and (f) comp linguists have been almost uniformly silent when the kinds of misinformation I alluded to above have been getting the attention of the broader linguistic public--probably feeling about these the way that all linguists feel about the misinformation regarding language in nonlinguistic publications (that it is too difficult and too hopeless a fight given the size and resources of the problem). AMRMail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue
fas.harvard.edu>
I want to commend Alexis for his remarks on the state of
comparative linguistics (9.313). He notes that it is (1) less taught
than it used to be, (2) an area where myth and facile misstatement
abounds, and (3) modern textbooks don't give comparative grammar the
attention they should.
Right on all three, and it is worse: in many areas, including
the study of native American languages which has been my own special
field, people believe that all there is to comparative grammar is the
assembly of lexical similarities and doing some statistics with
these. This is not how comparative grammar is done! My favorite
one-liner describing the field came from a course I took with the
distinguished Panini scholar Paul Thieme -- comparative grammar is the
process by which similarities between languages are converted into
correspondences. Somehow the notion that isolated vocabulary
comparisons are the basic tool of genetic classification crept in to
vitiate our methodology. Since I am quoting myself from a previous
publication I may as well continue for a bit: this doctrine was
explicitly stated by J.W. Powell, on page 11 of "Indian Linguistic
Families of America North of Mexico" BAE-R7 (21891): 7-148, but the
most influential source seems to have been A. L. Kroeber, "The
Determination of Linguistic Relationship" <italic>Anthropos </italic>3
(1913) 289-401. His contention here (Kroeber did not always hold this
view) survived, in America at least, the reasoned and urbane rebuttal
of Meillet in "Le probleme de la parente des langues"
<italic>Scientia</italic> 15 (1914), reprinted in <italic>Linguistique
historique et linguistique generale </italic> 1 (1948): 76-109 to
become a basic rule of thumb used by practitioners of language
classification and reaching its most ridiculous level in
Greenbergianism. {see IJAL 48.333 (1982)}.
As Alexis suggests, I believe we must get back to basics. I
would start by suggesting we go back to Meillet (see above) and, more
recently, Hoenigswald. To me, in fact, what we do when we do
comparative grammar is not very different in general outline from what
we do when we do descriptive grammar: in the latter we collect speech
fragments from native speakers as basic data, and set out to write
grammars which account for these data. In comparative grammar we
collect as data fragments from various putative daughter languages,
and set out to write a grammar of the protolanguage which will account
for them. To the extent we can do this, we can then claim that the
languages are genetically related. It seems to me that one of the
problems, as now emphasized again in the new book by R.M.W. Dixon
<italic>The rise and fall of languages</italic>, Cambridge University
Press 1997, is just that the current <italic>Zeitgeist</italic> does
not much consider the problems and techniques of working with real
data from real languages -- it is much more fun to theorize, and in
the case of comparative grammar it is much more fun to jump to
conclusions about Proto-World than it is to get down to the hard work
of comparing and analyzing data.
So, once again, I commend Alexis and support his emphasis that
we have to get back to basics with comparative grammar. Even though we
will no doubt never be able to write the grammar of Babelonian (excuse
me), there is much that needs to be done and we should get to it!
Yours, kvt
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue
HUMnet.UCLA.EDU>I have a few comments on the subject, mainly concerning Robert R. Ratcliffe's message. I have elsewhere discussed Alexis's views and concerns, including in p.c. Therefore, I will set apart from Robert's message those points he made which might be construed as ad hominem to Alexis. With that understood, I not only agreed with most of what Robert said, but admired the way he articulated it. Again, I want to set aside the ad hominem stuff, since I think his criticism applies more generally to a point of view which I agree with Robert is (or should be?) "anachronistic", though perhaps not as "bizarre" in historical context as I wish it were for the common concerns of current historical linguistics. Starting with that point, I felt it worthwhile to comment on some things that I thought need amplification, and some things that I thought might be overstated and misleading to those who have not yet formed opinions on the matter. Robert writes: > I feel that his apparent obsession with GENETIC classification is >both bizarre and anachronistic. This could be said about a lot of scholars. Honeigswald, in an acompanying message, undoubtedly had Greenberg and Merritt in mind (Alexis dissociates himself from them, but this message is not about Alexis). In any case, the word "anachronistic" struck me as particularly appropriate to "genetic" classification, not only because of the large role played by language contact in the evolution of languages, but because of the original motivation for genetic classification in *romantic nationalism* from the beginning of the 19th c. (until who-knows-when). The tree model of linguistic evolution, compellingly simple and simply a part of the entire story, served the political identity purposes of romantic nationalism in providing the basis for origin myths for sociopolitical entities seeking recognition of some sort. Origin myths, providing a common origin for a current or desired sociopolitical entity, are quite widespread among cultures and were wholeheartedly adopted by romantic nationalism for reasons that we need not go into, cf. the Trojans for the Romans, according to the Aeneid, Indo-European (called "Indo-Germanische") and the much admired connection with Hindu civilisation for the 19th c German intellectuals, struggling for a common-origin myth. In accordance with this myth, genetic classification was just a means to an end, under the assumption that a sociopolitcal entity needs an origin myth to motivate its members and establish its "legitimacy", particularly in the face of more powerful and apparently more unified entities, e.g., France for the early 19th c German romantic nationalists. That, of course, has become anachronistic (at least in academic and scientific discourse). Germany etc became somewhat unified, and since then the rules of the political game have been changed both politically and scientifically in power states, often to the displeasure of remaining romantic nationalists in various "minority" cultures. Yet, the preoccupation with singling out some UNITARY origin for a language, as for a culture, persists, now as an "obsession" (or acquired predisposition) without a clear purpose. A vaguer purpose of chronologically layering the contacts which have led to current cultures remains the underlying purpose. Somehow, for many historical linguists, the notion of a "single" origin still has primacy, and the problems involved in coordinating into some single ancient period of time the various things that can be reconstructed is short-changed by scholars who impatiently dismiss contact phenomena as "noise", but argue incessantly about which phenomena are contact and which are "original" (indicating no consensus on the principles for establishing such things, a glaring weakness in historical linguistic methodology at present). Robert continues: > The classic genetic tree model of the nineteenth century >reflects only ONE type of historical situation-- that where a >population speaking a more or less homogeneous language splits into >two or more groups which then separate to the point that they >completely and permanently lose contact with each other. We know that it is not a necessary condition that groups lose contact with each other for their languages (more accurately, speech varieties) to diverge. Loss of contact is tangential to the point of the basic tree model, and ultimately clouds the issue of what is wrong with it. >But based on >more modern research on sociolinguistic variation, dialectology, >creoles and pidgins, and language contact, we have to conclude that >the situation envisaged by the classic genetic model is by no means >the most common or typical situation. I think the modern research that R alludes to here has not yet resolved the problem of what form of linguistic change is (statistically?) most common or typical (esp internally spontaneous vs. contact-induced vs. multiply-caused), and even raises not yet resolved questions about the extent to which the classic genetic model can be maintained at all, or at least how to (re)interpret it. This is not about opposing one set of research tools with another, but about how to coordinate them, and in so doing discover how they have to be modified, and how the understandings underlying and motivating them have to be modified. In the final analysis, it is about how to recognize and then explain similarities and differences among languages. This spans the range from "universal grammar"to the most minute and seemingly "trivial" detail in any speech variety. > Although Semitic was the >first language family recognized (pre-dating Jones's discovery of IE, >by about 900 years), Neo-grammarian principles were never fully >embraced by Semitists. The important, triumphant and troublesome point is whether or not the notion of a common ancestor which was not the same (or even essentially the same) as any of the attested languages was part of that recognition. If so, it had no influence on IE studies, since at first IEists thought that Sanskrit was essentially the same as Proto-IE. Once it was recognized that no current language is the ancestor of other current languages, serious reconstruction began, and the problems we have discussed above flooded in, at first unnoticed. Hence one does see, particularly in the older >literature, or particularly in the Italian school, references to >"Common Semitic" rather than "Proto-semitic," indicating some sort of >agnosticism about the notion of proto-languages. The distinction is worthwhile if it is made more explicit. "Common Semitic" refers to what attested Semitic languages have in common, with the inference that they inherited it from an earlier time when whatever they had in common existed as a SINGLE entity. The "Proto-Semitic" notion usually goes further to boldly and less justifiably claim that they descended directly from a SINGLE "language" (except for the "noise" mentioned above). But a "language" is more than a set of words. We are still learning how much more it is. (Undeniably, Semitic languages also share some very striking grammatical processes, distinct from most of their historically attested neighbours, but it is a very tricky proposition to assign them to the same time period as the common lexicon, not to mention the problem of establishing what is indeed in the *earliest* reconstructable lexical stratum.) >on the other hand, the standard models of language relationship and >language change are much too IE-centric. Some aspects of these >models-- including models of subclassification and models of >morphological change-- really don't work well for Semitic. Various things are implied here, and I don't understand them all. If standard models are IE-centric, then they bring the same problems inherent in the IE model to other language families -- and, indeed, they do. As for models of subclassification and morphological change, I'm not sure what is meant, but I would suppose that general problems in linguistic analysis, particularly grammatical analysis, are involved, no less for IE languages than for Semitic. Just to be more concrete about the first point for a moment, the implications of Hittite for Indo-European reconstruction are, I think, extremely severe. One view has Hittite (and its Anatolian relatives) as the most conservative IE language, esp with respect to grammar. Another view has Hittite grammar as formed largely by contact with other attested languages of similar type, and thus not particularly conservative in terms of *IE* (nor is its lexicon particularly conservative of IE, as a whole). This is quite apart from the so far plausible notion that "active" or "agentive" languages (like Hittite) can typologically evolve into "nominative" languages like the other IE "branches". This means that as a speculative model, the connection between Hittite grammar and more general IE grammar can be explored and be insightful for understanding the processes involved in the evolution of IE grammar, WITHOUT it necessarily being the case that Hittite conserves that "Pre-IE" grammar. Similarly, the study of other Afro-Asiatic languages can turn out to be insightful for studying the prior evolution of Semitic grammar, without the inevitable inference that Semitic grammar "descends" from Proto-AA grammar in some simple-minded way. This realization should allow the different brands of Semiticists to profit from each other's work and avoid pointless arguments about things they should realize they do not yet know. All they need to be committed to is something more specific than that Semitic is or is not "simply a direct descendent" of something called "Proto-AA". (All scholars are always doomed to be the blind men arguing about the elephant to future scholars, who will be arguing about a different elephant. It's blind men and elephants all the way down -- and up.) >So while Semiticists do have a responsibility to try to apply these >models, general historical linguists do have a responsibility to take >into account data from Semitic which doesn't fit these models and to >try to modify the models accordingly. Unfortunately, general historical linguists are not in a position to recognize the depths of the issues which specialists encounter. They seem to be at the mercy of specialists who may recycle ill-conceived ideas from the generalists. To avoid this vicious cycle, every historical linguist needs to be a specialist in some area and also a generalist. I see no other way out, and, if there's something wrong with the "health" of historical linguistics as a discipline, I suspect that's where the major problem lies. >language contact and typological universals are interesting areas of >research right now, and until we know more about these things arguing >for or against an Altaic family is utterly pointless. I agree entirely. But I also reiterate my last point, lest generalist typologists simplify and reductivize their second-hand information into fantasies with regard to restrictions on paths of linguistic change. >I don't see how we can do it if we commit ourselves monomaniacally to >an outdated pseudo-Darwinian model of language relationships, and >ignore the work of sociolinguists, dialectologists, pidgin/creole >specialists, typologists, and theorists of all stripes interested in >language universals. Let's just note for the record that the linguistic tree notion, which I mentioned above, preceded the Darwinian tree-radiation of life forms concept, and rests more directly on the personal family tree with its sexually motivated binary splits. After Darwin, the pseudo-scientific "genetic" terminology was grafted onto the linguistic tree notion, and, of course, all kinds of racist implications were then attached to it, some of which still remain as an "empty" racism without the added notion of "natural superiority". I sometimes wonder how current US (and wider) white supremacists like the "Aryan Brothers" would take it if they realized they were calling themselves "Iranians". Would they give up swastikas (< Sanskrit) and start wearing black turbans like Khomeini et al?Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue