Darwin-L Message Log 11: 121–149 — July 1994
Academic Discussion on the History and Theory of the Historical Sciences
Darwin-L was an international discussion group on the history and theory of the historical sciences, active from 1993–1997. Darwin-L was established to promote the reintegration of a range of fields all of which are concerned with reconstructing the past from evidence in the present, and to encourage communication among scholars, scientists, and researchers in these fields. The group had more than 600 members from 35 countries, and produced a consistently high level of discussion over its several years of operation. Darwin-L was not restricted to evolutionary biology nor to the work of Charles Darwin, but instead addressed the entire range of historical sciences from an explicitly comparative perspective, including evolutionary biology, historical linguistics, textual transmission and stemmatics, historical geology, systematics and phylogeny, archeology, paleontology, cosmology, historical geography, historical anthropology, and related “palaetiological” fields.
This log contains public messages posted to the Darwin-L discussion group during July 1994. It has been lightly edited for format: message numbers have been added for ease of reference, message headers have been trimmed, some irregular lines have been reformatted, and error messages and personal messages accidentally posted to the group as a whole have been deleted. No genuine editorial changes have been made to the content of any of the posts. This log is provided for personal reference and research purposes only, and none of the material contained herein should be published or quoted without the permission of the original poster.
The master copy of this log is maintained in the Darwin-L Archives (rjohara.net/darwin) by Dr. Robert J. O’Hara. The Darwin-L Archives also contain additional information about the Darwin-L discussion group, the complete Today in the Historical Sciences calendar for every month of the year, a collection of recommended readings on the historical sciences, and an account of William Whewell’s concept of “palaetiology.”
--------------------------------------------- DARWIN-L MESSAGE LOG 11: 121-149 -- JULY 1994 --------------------------------------------- _______________________________________________________________________________ <11:121>From email@example.com Tue Jul 19 16:12:15 1994 Date: Tue, 19 Jul 94 14:11:53 PDT From: firstname.lastname@example.org (Peter H. Salus) To: email@example.com Subject: Re: historical lx Well, as Peano wrote a book (in Italian) on Grassmann (in the 1880s), I'd look there, first. I've never read it. Peter _______________________________________________________________________________ <11:122>From firstname.lastname@example.org.Hawaii.Edu Tue Jul 19 19:47:29 1994 Date: Tue, 19 Jul 1994 14:46:58 -1000 (HST) From: Joel Bradshaw <email@example.com.Hawaii.Edu> Subject: Uniformitarian Twain To: firstname.lastname@example.org Here's a wonderful Mark Twain parody of uniformitarianism dating from his 1875 Life on the Mississippi. He cites the following figures, which I have tabulated, and then draws scientific conclusions in the accompanying paragraphs, whose spelled-out numbers I have converted to digits. Otherwise, the text is a direct quote. Length of Mississippi, Cairo to New Orleans Year Length (miles) 1699 1215 1722 1180 1875 973 Now if I wanted to be one of those ponderous scientific people, and "let on" to prove what had occurred in the remote past by what had occurred in a given time in the recent past, or what will occur in the far future by what has occurred in late years, what an opportunity is here! Geology never had such a chance, nor such exact data to argue from! Nor "development of species," either! Glacial epochs are great things, but they are vague--vague. Please observe: In the space of 176 years the Lower Mississippi has shortened itself 242 miles. That is an average of a trifle over one mile and a third per year. Therefore, any calm person, who is not blind or idiotic, can see that in the Old Oolitic Silurian Period, just a million years ago next November, the Lower Mississippi River was upwards of 1,300,000 miles long and stuck out over the Gulf of Mexico like a fishing rod. And by the same token any person can see that 742 years from now the Lower Mississippi will be only a mile and three quarters long, and Cairo and New Orleans will have joined their streets together, and be plodding comfortably along under a single mayor and a mutual board of aldermen. There is something fascinating about science. One gets such wholesale returns of conjecture out of such a trifling investment of fact. Mark Twain, Life on the Mississippi (New York: Signet Classics, 1961), p. 120. Joel Bradshaw email@example.com _______________________________________________________________________________ <11:123>From DARWIN@iris.uncg.edu Tue Jul 19 23:15:34 1994 Date: Wed, 20 Jul 1994 00:14:50 -0500 (EST) From: DARWIN@iris.uncg.edu Subject: Sarich, Campbell, and deep roots To: firstname.lastname@example.org Organization: University of NC at Greensboro The different comfort levels that systematists and historical linguists experience when exploring the bases of their respective trees is something that came up here once before. It stuck me as interesting then and it still does, I suppose because I haven't quite sorted it out in my mind yet. Kent Holsinger proposed one useful phylogenetic analogy. Let me try another. The linguists, when they get down near the base of the tree (that is, when they examine very old events), have precious few characters to work with, and these characters tend to be rather simple word matchings, rather than conjugation systems and more complex things (pardon me for not knowing the proper terms). They also have some things that a systematist might call "process assumptions", that is, ideas about how linguistic change takes place, generally speaking. These processes would include word formation by onomatopoeia, etc. The attitude the linguists seem to be taking is that of a systematist faced with, say, a short chunk of molecular sequence data that may have experienced multiple hits. This means that there are relatively few characters to work with, and from what we know or think we know of the processes of change, the similarities that can be seen may not mean much. "Multiple hits" refers to the problem one can face with molecular data given that the only nucleotides are A T G C; if you have enough time, any particular site will mutate (be hit) more than once, and the history will be unrecoverable because the events are all over-written. Vince accuses the linguists of ignoring parsimony. The linguists may not be conversant with the particular role this idea has played in arguments about phylogenetic inference. Anyone wanting an extensive survey can consult: Sober, Elliott. 1988. _Reconstructing the Past: Parsimony, Evolution, and Inference_. Cambridge: MIT Press. But the basic point Vince is making is: isn't it much simpler to assume that these various widespread similarities are due to inheritance than to all sorts of special processes? But I think Vince would agree that if one had a short chunk of sequence data that had a reasonable probability of having been subject to multiple hits, then even though one _could_ do a standard cladistic parsimony analysis on it and get a tree, the tree might not mean very much. That, as I understand it, is rather like what the linguists are saying. But as a systematist, I'm predisposed to being more comfortable down near the bottom of the tree than the linguists seem to be, so I'm still not completely satisfied with their discomfort. I think there is a very important subtext here, however: one that emerges in Vince's criticism of Campbell (forcefully expressed, as is his wont). The arguments that Vince quotes from Campbell are almost exact copies of the kinds of arguments made by the early critics of cladistic analysis: "systematists shouldn't bother with phylogeny because there is lots of convergence, and in plants lots of hybridization, and all of these things so obscure the history that you're really wasting your time trying to reconstruct phylogeny; just make your practical groupings and don't try to do more." The last twenty or thirty years in systematics have shown that criticism to be completely hollow: the problem with earlier treatments of phylogeny was not that they were obscured by convergence or hybridization, but rather that the principles of phylogenetic inference were poorly understood. Thus a systematist, hearing the sort of argument made by Campbell in Vince's quotation, almost certainly will react somewhat dismissively, since many hard battles have been fought (and won) over that sort of issue in systematics in the last thirty years. This is not to say that that is necessarily the correct reaction; I'm simply trying to tease out the sources of the disagreement, intellectual, historical, and disciplinary. Bob O'Hara, Darwin-L list owner Robert J. O'Hara (email@example.com) Center for Critical Inquiry and Department of Biology 100 Foust Building, University of North Carolina at Greensboro Greensboro, North Carolina 27412 U.S.A. _______________________________________________________________________________ <11:124>From DARWIN@iris.uncg.edu Wed Jul 20 11:04:11 1994 Date: Wed, 20 Jul 1994 12:04:01 -0500 (EST) From: DARWIN@iris.uncg.edu Subject: Comet info and astronomical palaetiology To: firstname.lastname@example.org Organization: University of NC at Greensboro Astronomy, as William Whewell recognized, has an important palaetiological component: As we may look back towards the first condition of our planet, we may in like manner turn our thoughts towards the first condition of the solar system, and try whether we can discern any traces of an order of things antecedent to that which is now established; and if we find, as some great mathematicians have conceived, indications of an earlier state in which the planets were not yet gathered into their present forms, we have, in pursuit of this train of research, a palaetiological portion of Astronomy. In recognition of astronomical palaetiology, in recognition of this the 25th anniversary of the first moon landing, and also in recognition of the impact of comet Shoemaker-Levy 9 on Jupiter, I have set up a special link on the Darwin-L gopher that can provide you with the latest information on the comet impact. Connect to the Darwin-L gopher (rjohara.uncg.edu) and look in the directory Other Network Resources. Those who are using gopher clients that are able to display images (such as the TurboGopher on the Macintosh) will be able to view some of the latest pictures of the impact from an assortment of telescopes around the world. More on astronomical palaetiology shortly, namely, a discussion of the phylogeny of asteriods. ;-) Bob O'Hara, Darwin-L gopher Robert J. O'Hara (email@example.com) Center for Critical Inquiry and Department of Biology 100 Foust Building, University of North Carolina at Greensboro Greensboro, North Carolina 27412 U.S.A. _______________________________________________________________________________ <11:125>From firstname.lastname@example.org Wed Jul 20 11:26:09 1994 Date: Wed, 20 Jul 94 11:18:42 EDT From: email@example.com (Peter D. Junger) To: firstname.lastname@example.org Subject: Re: Re: On the shoulders of giants I was not able to respond to this thread earlier from my home machine, though I don't know why. But in any case, as has been mentioned, the current efforts to discover the sources of what is traditionally termed OTSOG is a feeble recapitulation of the seminal and definitive work done by Thomas Merton in his _Standing on the Shoulders of Giants, A Shandean Postscript_. It is shocking that only one member of this list so far has expressed familiarity with this work, which is perhaps _the_ paradigmatic example of the application of the historical sciences to a trope. Peter D. Junger Case Western Reserve University Law School, Cleveland, OH Internet: JUNGER@SAMSARA.LAW.CWRU.Edu -- Bitnet: JUNGER@CWRU _______________________________________________________________________________ <11:126>From email@example.com Wed Jul 20 18:42:29 1994 Date: Wed, 20 Jul 94 16:42:18 PDT From: firstname.lastname@example.org (Peter H. Salus) To: email@example.com Subject: Re: Sarich, Campbell, and deep roots In 1970 Howard Aronson (U of Chicago) and I were both visiting professors at the Linguistics Summer Institute at Ohio State. Inter alia, I was teaching Hittite and Howie was offering Kartvelian. We decided to hold a common last meeting, open to all and sundry, on Hittite-Kartvelian Linguistic Relationships. I led off, explaining that in our Chomskian paradigm, we no longer sought either phonological nor morphological parallels, for these were mere surface phenomena. We now looked for deeper, underlying relationships. Here, Howie took over and explained how Georgian had lexical items for hand, mother, five, etc. He wrote them on the board. I then wrote the corresponding items in cuneiform Hittite. Howie then explained the concept of lexical gaps and I solemnly pointed out that *neither* language had a word for "left-handed violinist." At this point, much to my delight, the late Ken Naylor let out a shriek, laughing hysterically, and pulled out a handkerchief to dab his eyes. We went on for about half an hour more, but couldn't go further. We were asked several times to re-hold the seminar, but felt it was a one-timer, a hapax legomenon, so to speak. There is a light side to historical linguistics, too. Some of you might want to glance at _Studies Out in Left Field_, which has been reprinted by Benjamins. I admit to my complicity in that, too. Peter _______________________________________________________________________________ <11:127>From firstname.lastname@example.org Thu Jul 21 23:58:00 1994 To: email@example.com Subject: Re: Sarich, Campbell, and deep roots Date: Fri, 22 Jul 94 00:57:57 -0400 From: Sally Thomason <firstname.lastname@example.org> Bob O'Hara asks just the right questions in considering differences between Vince Sarich's view of the Greenberg controversy and the view that has been expressed by linguists on DARWIN. The short answers to the issues he raises are, yes, historical linguists know about parsimony; no, convergence and hybridization are not the main reasons for historical linguists' rejection of Greenberg's methods and results; and no, the inheritance-vs.-diffusion dichotomy that Sarich sets up does not correspond to any dichotomy anyone will find in historical linguistics, because the dichotomy is simplistic. The crucial general point (responding to Bob O'Hara now) is that most historical linguists believe that the linguistic evidence on which hypotheses of language relationships are based degenerates over time, to such an extent that the kinds of systematic correspondences that are needed to establish relationships disappear after some thousands of years -- more than six thousand, probably not much more than ten thousand. Our belief is not an article of faith, but is rather a conclusion based on the results of studies of language change over many decades. That is, we have solid empirical evidence that change obscures similarities over time, and that systematic correspondences also disappear over (more) time; and from that evidence we extrapolate back several thousand years from the most distant relationships we have established, to conclude that by then it is not at all likely that enough systematic correspondences will survive in related languages to permit us to prove that the languages are related. Those are the short(ish) answers. If you're interested in details, read on. It may indeed occur to other historical scientists to wonder if historical linguists have heard of Occam's Razor, given Sarich's comments; in fact, however, parsimony is, and has long been, a cornerstone of our methodology. We use it constantly, we teach it to our students, it appears in historical linguistics textbooks. But surely, in other historical sciences too, the question of just how it is to be applied in any specific instance is not always trivially obvious: first you have to agree on the premises. Sarich lays out the premises of his argument in a way that historical linguists would not accept. Since Bob has raised the issue, I'll comment on the three major disagreements here. First, Bob says (citing Sarich), `isn't it much simpler to assume that these various widespread similarities are due to inheritance than to all sorts of special processes?' The answer would of course be yes if the premise were solid, but the crucial phrase here is `these various widespread similarities': that is precisely what Greenberg's critics deny that Greenberg has demonstrated. And we have reasons. (1) Data errors in Greenberg's book LANGUAGE IN THE AMERICAS (LIA) are so numerous that it's hard to take seriously groupings based on them. A number of specialists in various well-established language families in the Americas have published analyses of the data in LIA with estimates of error rates ranging from 60% to 90% (well, there's one 100% error case, in which not one of the words Greenberg cites from a particular language actually comes from that language -- rather, the words come from several other languages). Greenberg does not deny the existence of the errors; his response to this criticism is that his method is so powerful that it transcends data errors. His critics are dubious about a method that is so independent of the evidence cited in support of its results. And, in particular, most of us would like to see results based on accurate data before deciding whether there is enough matching to require a historical explanation. (2) Linguists also point to the large number of words one can find, in inspecting any group of languages, that are similar in form and meaning but are demonstrably unconnected etymologically (e.g. English bad and the Persian word that sounds the same and means the same, and English much and Spanish mucho, which are not etymologically related -- though English and Persian and Spanish are in fact related languages). This is the issue of chance, which Don Ringe addressed in a message to DARWIN recently. What Don didn't mention (I think) was that he published a monograph analyzing the statistics in LIA and arguing that test cases show Greenberg's results to be well within the range of chance similarity -- and that's accepting Greenberg's data as is, without eliminating erroneous data. Greenberg published a reply to Ringe's monograph, but his sole response to Ringe's actual statistical analysis was that the statistician father of a student of his had written him, Greenberg, a letter praising his statistics. No one has impeached Ringe's statistics, as far as I know; and if Ringe's results are accurate, then there is nothing in the LIA data that rises above the level of chance. (3) And finally, there's the issue of borrowing, which, though nontrivial, is to my mind the least significant of the three types of objection I list here. Greenberg's method relies absolutely on the absence of much borrowing (in the kinds of forms he examines, primarily basic vocabulary but also grammatical morphemes) among languages. Greenberg claims that borrowing is a minimal factor in language history, except in nonbasic vocabulary. I don't think any historical linguists nowadays would agree with him (but I admit that my own reluctance to agree with him is based on the fact that I co-wrote a book a few years ago arguing, with a lot of supporting evidence that has generally been accepted since then, that borrowing is a pervasive phenomenon in language history). Greenberg's response to such counterevidence is to repeat his claim; he has not provided any argumentation or evidence to support his view on this point. My overall point here is that Greenberg's evidence looks solid as long as you don't examine the data critically, or consider the probability of chance resemblance, or consider the probability of borrowing (though no one would claim that close matches in words covering a whole continent or two could be due to "massive borrowing" over a large area -- a charge that has been leveled repeatedly against Greenberg's critics, but without any reference to a published or unpublished statement of such a claim). But note that this doesn't mean that Greenberg is wrong; for all I know, he may be right. The argument, rather, is that he has not made his case on the basis of any linguistic evidence that he has presented. Second, and potentially more interestingly for DARWINers, similarity itself is dubious as a criterion for establishing distant linguistic relationship. With closely-related languages, there's no problem: cognate forms, i.e. forms inherited from the common parent, in (say) German and English do indeed look very similar, and they also tend to have very similar meanings (mother/Mutter, father/Vater, brother/Bruder, etc.). But the further back you go in time, and therefore in distance from the common parent language, the less similar cognate forms will be in form and meaning. Not always, of course: some sounds are more stable in certain positions in a word than others, and the accidents of history leave other things unchanged too: the words for `mother' in Latin and Sanskrit look quite similar to each other and to English, too. But many other cognates aren't so obvious, phonetically and/or semantically: if it weren't for the archaic English spelling, would one easily connect English knight (phonetically [nayt]) and German Knecht (which means `low fellow' and is phonetically [knext])? Here's the point: linguistic evidence for historical relationships (as I said above) degenerates over time in a way, or at least to an extent, that the evidence in some (all?) other historical sciences doesn't. That doesn't make it impossible to establish relationships, but it is why linguists require systematic correspondences rather than mere unsystematic similarities: if there are systematic correspondences (like p : ch in some Salishan languages, or the famous dw : rk correspondence between ancient Greek and Armenian), then chance can be quickly ruled out as the source of the matchings, regardless of whether the forms being compared are phonetically similar or not. In closely-related languages you typically have *both* similarity *and* systematic correspondences -- e.g. English/German pairs like knee/Knie, knuckle/Knochen, knot/Knoten, where English has lost the phonetic k and German has kept it -- but similarity is neither necessary nor, in the absence of systematic correspondences, sufficient to prove linguistic relationship. One often finds very little overall phonetic similarity in genuine cognates in the deepest relationships that have been well established in the linguistic literature; that's why we are skeptical of methods that rely on linguistic similarity alone to support claims of linguistic relationships at even greater time depths. Experience has taught us that obviously similar forms at great time depths are more likely to be due to chance or to fairly recent borrowing than to inheritance. This skepticism is not intended by anyone, however, as a blanket rejection of all efforts to establish distant genetic relationships: if someone can show widespread systematic correspondences in similar or dissimilar forms, historical linguists will accept that evidence gladly. Third, Sarich's logic in his `Occam's Razor' excerpt applies to the question of diffusion vs. inheritance as a source of similarities. But in setting up his simple dichotomy -- which is more parsimonious as an explanation, inheritance or borrowing? -- Sarich relies on the apparent assumption that there is no way to distinguish shared forms that result from inheritance from those that result from borrowing. Since historical linguists do have ways of making this distinction, Sarich's premise is flawed, and his argument is therefore aimed at a straw man that no linguist would believe in. In the Na-Dene example he cites, for instance, the problem is not as he states it. It's true that Haida shares numerous words with the other languages in the proposed Na-Dene family, but, crucially, most of those words are not in the basic vocabulary. Instead, they cluster in particular nonbasic semantic domains, such as animal names -- types of words that are very often borrowed among languages. But in every well-established language family in the world, genuine cognates are at least as common, and indeed always or almost always more common, in the basic vocabulary than in the rest of the lexicon; the reason, to oversimplify somewhat, is that you don't usually need to borrow a word for something you already have a word for. (`Basic vocabulary', I hasten to add, is not a precise theoretical concept; it's a rough-and-ready category, used by all comparativists. The lists include words like `mother', `sun', `walk', ....) So if common ancestry is the source of the shared vocabulary in this case, the uniformitarian criterion is violated, because that would make Na-Dene (with Haida included) unique among language families in having a sharp discrepancy between basic vocabulary (few shared words) and nonbasic vocabulary (numerous shared words). Aside from the issue of basic vs. nonbasic vocabulary, there are other ways of distinguishing borrowed from inherited material, having to do with the nature of the similarities: close matchings in certain words but not elsewhere in the vocabulary, for instance, which tends to be a red flag even if some of the close matchings are in the basic vocabulary. (This, for instance, helps pick out the French borrowings in English, which are much more similar and which have different sound correspondences than the actual cognates inherited by French and English from Proto-Indo-European.) But the methods for distinguishing borrowed from inherited material require preservation of enough of the original sound structure (of the inherited material and/or of the borrowed words) to apply the methods to; that's why, with claims of very distant relationships, it can be impossible to distinguish borrowed from inherited material, even in cases where one can be reasonably sure that there is some kind of historical connection between two languages. Sally Thomason email@example.com _______________________________________________________________________________ <11:128>From firstname.lastname@example.org Fri Jul 22 08:40:24 1994 Date: Fri, 22 Jul 94 09:03:22 EDT From: email@example.com (Kent Holsinger) To: firstname.lastname@example.org Subject: Re: Sarich, Campbell, and deep roots One minor comment on Sally Thomason's very interesting post about the reasons historical linguists are skeptical of Greenberg's work. In the course of that comment she writes: > linguistic evidence for historical relationships (as I said above) > degenerates over time in a way, or at least to an extent, that the > evidence in some (all?) other historical sciences doesn't. I don't know enough about other historical sciences to comment intelligently, but this doesn't seem that different to me, in principle, from the problems biological systematistists face. Sally attributes this loss of evidence to the increasing difficulty of recognizing cognates and grammatical structures as languages become more distantly related. That seems entirely parallel to the difficulties that systematists face in recognizing homologous structures in distantly related groups. It's pretty easy to see that the bones in the middle ear of humans and chimps are homologous. It's far more difficult to see that the bones in the middle ear of humans are homologous to certain jaw bones in fish. Recognizing the "systematic correspondences" is one of the fundamental problems biological systematists face. In some ways, however, the problem in linguistics may be more similar to the problems in molecular systematics than to those in morphological systematics, as Bob O'Hara suggested. One of the reasons these systematic correspondences can be recognized with morphological data is that many structures have an enormous number of characters associated with them. When it is possible to show a detailed correspondence in many characters and, possibly, a transformation series linking two seemingly disparate structures, then we have reasonable confidence in pronouncing the structures homologous. Molecular systematists dealing with nucleotide sequence data, on the other hand, have only four possible character states at every position in the sequence: A, T, C, & G. Two *random* sequences of 1000 bases in length are expected to be identical at about 25% of the positions. (Actually, they'll be identical at a bit more than that, once nucleotide compositional biases are taken into account and once you take into account the fact that the sequences will be aligned in such a way as to maximize positional identity.) As a result, when nucleotide sequences are very divergent from one another, there is no way to tell whether they are descended from a common ancestor. Does this bear any resemblance to the problem posed by having a limited number of phonemes available? Is it possible that the difference in attitudes between historical linguists and biological systematists reflects biologists confidence that life is monophyletic, that there is, therefore, a single history of life to be discovered, and a willingness to express hypotheses about that relationship, even if they are later rejected? The most recent issue of _Nature_ that I've had the chance to look at had a paper suggesting rodents, rabbits, and hares form a monophyletic group, based primarily on evidence from a single fossil. Would historical linguists publish a paper based on similar amounts of evidence postulating a relationship among a group of languages? I'm not suggesting that biological systematists are right in what they do and historical linguists are not. I'm just wondering if the differences reflect not fundamental differences in the problems we face, but differences in how willing we are to postulate relationships. -- Kent +------------------------------------------------------------------------+ | Kent E. Holsinger Internet: Kent@Darwin.EEB.UConn.Edu | | Department of Ecology & Holsinge@UConnVM.UConn.Edu | | Evolutionary Biology BITNET: Holsinge@UConnVM | | University of Connecticut, U-43 | | Storrs, CT 06269-3043 | +------------------------------------------------------------------------+ _______________________________________________________________________________ <11:129>From email@example.com Fri Jul 22 11:05:22 1994 To: firstname.lastname@example.org Subject: historical lx again Date: Fri, 22 Jul 1994 12:05:01 EDT From: Don Ringe <email@example.com> Dear Colleagues-- Many thanks to Sally Thomason for her clear and sensible explanation of the problems in historical/comparative linguistics. What I have to add concerns the monograph of mine to which she refers. Though I think my conclusions hold up, readers should be aware that the piece has plenty of shortcomings, beginning with the extreme primitiveness of the math. Some of these were acknowledged in my exchange with Prof. Greenberg (especially fn. 57a, which is just plain wrong--fortunately it's just an aside, not part of the argument); others were pointed out in Bill Poser's judicious book notice in a recent issue of *Language*. So far as I can see, the main contribution of the monograph is in the way it fits the probabilistic model to the structure of the data--a point to which far too little attention has been paid in the past. What we need next is further work by a linguist and a mathematician in tandem, so as to make the method more sophisticated and powerful without losing the solid grounding in linguistic structure. Cheers! --Don Ringe _______________________________________________________________________________ <11:130>From GROBE@INS.INFONET.NET Fri Jul 22 12:28:55 1994 Date: Fri, 22 Jul 1994 12:26:58 -0500 (CDT) From: GROBE@INS.INFONET.NET To: firstname.lastname@example.org, email@example.com, firstname.lastname@example.org, Galileo@muwayb.ucs.unimelb.edu.au, email@example.com, firstname.lastname@example.org, email@example.com, firstname.lastname@example.org Subject: soc.history.science gatewayed mailing list To all interested historians of science: Cold Spring Harbor Laboratory is developing a gateway for the proposed history of science newsgroup, so that interested people who do not read Usenet groups may participate in the group via e-mail. Nathaniel Comfort (email@example.com) and Corp Reed (firstname.lastname@example.org) will be the CSHL contacts for the mailing list. It is planned that the list will be called hist-sci and that its address will be email@example.com. Information about subscribing to the list will be posted on soc.history.science and related newsgroups and sent to appropriate mailing lists, once the listserv is set up and is running. For users of Mosaic, it will also be possible to subscribe to the list through a page located at Cold Spring Harbor Laboratory's World Wide Web site (http://www.cshl.org/). Nathaniel C. Comfort _______________________________________________________________________________ <11:131>From DARWIN@iris.uncg.edu Sat Jul 23 21:22:54 1994 Date: Sat, 23 Jul 1994 13:14:32 -0500 (EST) From: DARWIN@iris.uncg.edu Subject: Reviews of Classical archaeology, history, etc. (fwd from BMR-L) To: firstname.lastname@example.org Organization: University of NC at Greensboro --begin forwarded message-------------- Date: Fri, 22 Jul 1994 03:05:04 -0400 From: Bryn Mawr Reviews <email@example.com> Subject: BMR: Scholia Reviews Sender: firstname.lastname@example.org SCHOLIA REVIEWS Scholia aims to provide critical reviews of publications in the field of ancient Greek and Roman art, archaeology, history, literature and philosophy as soon as possible after they appear. The editors also believe that reviews should be as detailed, informative and comprehensive as possible. In order to make it possible for the journal to provide reviews of this kind, given the constraints under which it is produced, reviews will be published over the international electronic network to registered subscribers. Subscription to the electronic reviews is free and without restriction. Once published, the reviews will be archived at the University of Natal, Durban and the University of Pennsylvania, USA, from which they can be retrieved by Gopher or FTP. Instructions on how to retrieve reviews electronically will be published in the journal itself along with a list of books received. The editor reserves the right to publish the full text of a review in the journal itself. Contributors of reviews are therefore requested to submit an abstract (300-500 words) together with the full text of their review. Contributions should preferably be sent by e-mail or on disk followed by one clearly printed copy by air mail. HOW TO SUBSCRIBE TO SCHOLIA REVIEWS In order to receive electronic reviews from Scholia simply send a request to Scholia@owl.und.ac.za. Your e-mail address will be added to the distribution list of Scholia Reviews. HOW TO OBTAIN SCHOLIA REVIEWS BY GOPHER Gopher to OWL.UND.AC.ZA and follow the path Campus Information System Faculty Information Classics Scholia Reviews The reviews are classified by the year in which they appeared e.g 1 (1992) and are listed by number, author, title and reviewer e.g. (1) Perkell, Vergil's Georgics (Davis). HOW TO OBTAIN SCHOLIA REVIEWS BY FTP (File Transfer Protocol) FTP to OWL.UND.AC.ZA. When you are asked for your name type: ANONYMOUS When asked for a password type in your email address and press ENTER. You do not have to use upper case letters. Then type: CD PUB/UND/CLASSICS/REVIEWS You can then list the contents of the directory by typing: LS To read a file type MORE followed by the filename (these are UNIX commands). Files are listed by year, number and author e.g. 92-1-Perkell = Review number 1, 1992, review of Perkell, Vergil's Georgics. SCHOLIA REVIEWS AT PENNSYLVANIA Scholia is pleased to announce that the reviews of the journal are now available on the ccat gopher at the University of Pennsylvania. We hope that access to the reviews will be more convenient at this location. We are grateful to Professor James O'Donnell and the University of Pennsylvania for making this possible. GOPHER ACCESS Gopher to ccat.sas.upenn.edu and look under menu item 8 (Electronic Publications and Resources). Scholia Reviews appear as item 19. GOPHER BOOKMARK The gopher bookmark that will let you or anybybody else add this to their own gopher menu is: Type=1 Name=Scholia Reviews (Classical Studies) Path=1/scholia Host=ccat.sas.upenn.edu Port=5070 URL: gopher://ccat.sas.upenn.edu:5070/11/scholia FTP ACCESS The ftp address is also ccat.sas.upenn.edu, login as anonymous, then cd pub cd scholia ls J.L. Hilton Reviews Editor: Scholia 20 July 1994 --end forwarded message---------------- _______________________________________________________________________________ <11:132>From GGALE@VAX1.UMKC.EDU Sun Jul 24 00:15:24 1994 Date: Sat, 23 Jul 1994 13:00:10 -0600 (CST) From: GGALE@VAX1.UMKC.EDU Subject: This should be useful/interesting to many members. --g To: email@example.com, firstname.lastname@example.org, email@example.com, HPSST-L%QUCDN.firstname.lastname@example.org ----------- Begin Forwarded Message ------------ Date: Sat, 23 Jul 94 11:58 EDT From: Ronald_E_DOEL@umail.umd.edu (rd87) Subject: ORAL HISTORY ARCHIVES: A Request To: email@example.com SURVEY OF ORAL HISTORY COLLECTIONS IN HISTORY OF SCIENCE HELP REQUESTED For an appendix to an article (to appear in a book on oral history in the history of science), I would appreciate learning about archival interview collections, in any country, which meet all of the following criteria: | 8 or more interviews in collection (roughly!) | interviews and transcripts are available to qualified scholars at a library or archive (or will soon be deposited in such a facility) | address the physical and biological sciences, including the social, cultural, and political relations of these scientists (but not medicine or the social sciences) Unfortunately space will only permit a terse *listing* of individual collections. It would be most helpful if information is provided in the following format: 1. TITLE OF COLLECTION (including year begun) 2. ISSUES COVERED (i.e., did interviews primarily address internal science themes, institutional developments, biography, gender or race issues?) Please include total number of interviews and (or) total hours of interviews in collection 3. LOCATION OF TRANSCRIPTS (ie, at what library/archive? Please note any significant restrictions on their use by scholars, e.g., closed until 1996) For university archives with significant interviews with science faculty, a single notice for these holdings is preferred *unless* distinct interview projects exist with different aims and motivations. I would also be glad to hear of any articles describing collections of oral history interviews in these fields, such as that published by Joan Warnow-Blewett in _Osiris_ 7 (1992). A list of collections already called to my attention appears below: * Archives for History of Quantum Physics (1961), American Philosophical Society * Recombinant DNA History (1975), MIT * Sources for History of Modern Astrophysics (1976), American Institute of Physics * [Interviews with scientists] ( ? ), Dwight D. Eisenhower Presidential Library * International Project in History of Solid State Physics (1979), American Institute of Physics * Faculty Oral History Project ( ? ), California Institute of Technology, Archives. * Space Astronomy Oral History Project (1981), Smithsonian Institution (National Air and Space Museum) * Women in Science and Engineering ( ? ), MIT * Hazel de Berg recordings ( ? ), Australian scientists, National Library of Australia * Space Telescope History Project (1983), Smithsonian Institution (National Air and Space Museum) * Laser History Project (1983), American Institute of Physics * History of Theories of Mass Extinction by Meteorite Impact and Alternative Causes (1984) [privately held at present] * Princeton Mathematical Community in 1930s (1985), Mudd Library, Princeton University * [Interviews in Physics and Allied Sciences], (1986) American Institute of Physics. * Smithsonian Videohistory Program (1986), Smithsonian Institution. * Glennan-Webb-Seamans Project for Research in Space History (1987). Smithsonian Institution, National Air and Space Museum. * Rand History Project (1987), Smithsonian Institution (National Air and Space Museum) * Biomolecular Science Initiative (1988), Oral History Program, UCLA. * MIT Radiation Laboratory (1991) IEEE History Center, Rutgers University, NJ * [Interviews with scientists] ( ? ), Columbia University Oral History Research Office. Many thanks. Apologies to those who received this note from several lists. Ronald E. Doel Postdoctoral Fellow MRC 311, DSH/NASM Smithsonian Institution Washington, DC 20560 tel. 202/ 357-2828 fax: 301/ 209-0882 email: firstname.lastname@example.org (Niels Bohr Library at AIP) please cc email to email@example.com ------------ End Forwarded Message ------------- _______________________________________________________________________________ <11:133>From CRAVENS@macc.wisc.edu Sun Jul 24 10:35:49 1994 Date: Sun, 24 Jul 94 10:35 CDT From: Tom Cravens <CRAVENS@macc.wisc.edu> Subject: Re: Sarich, Campbell, and deep roots To: firstname.lastname@example.org Kent Holsinger raises good questions regarding the differences in approach of historical biological sciences and historical linguistics. I'm much too ignorant of the former to draw any analogies, but perhaps I can help a little bit by illustrating some of the difficulties in establishing genetic linguistic relationships at great time depth. A number of factors serve to obscure relationships. I'll cite just four which pop to mind immediately, illustrating with Romance languages, which have diverged in a relatively short time span. 1. Selection of variants. Caps represent Latin. French tete `head' TESTA 'earthenware pot' > 'head' Spanish cabeza `head' CAPU(T) 'head' 2. Extreme phonological erosion. [ ] enclose phonetic transcription. French [u(t)] 'August' AUGUSTU Spanish [agosto] 'August' " French [o] 'water' AQUA Spanish [a(g)wa] 'water' " 3. Borrowing. French olive OLIVA Spanish aceituna Arabic zeituna 4. Semantic divergence. French chef 'boss' CAPU(T) 'head' Spanish cabo 'end' " These are extreme examples, in the sense that, given the short time depth, much else has been affected little enough that the genetic relationship can be established to everyone's satisfaction. And, by having an approximation of the protolanguage at hand (Latin), as well as all but innumerable cognate Romance dialects for further comparison, and external historical knowledge available (long contact with Arabic in Spain), everything can be sorted out nicely. But the examples are also real, and illustrative of what happens quite commonly in language over time. If we factor in continued developments of these sorts over several millenia, we find relationships getting ever more obscure as we go back in time. At this point, presuppositions about what constitutes acceptable method, that is, about what produces plausible results, become crucial. One basic criterion is that phonological correspondences be demonstrable (and semantically coherent). For chef-cabo, for example, we can find a large set of examples in which French has a "sh" sound and Spanish has [k], preceding different vowels in French, but always corresponding to the position before [a] in Spanish (champ - campo, chien - O.Sp. can, cher - caro, etc.) So we posit that French once had an [a] in that position, and that before the French vowels changed, [k] became "sh" when preceding [a]. We continue the process for the rest of the phonological structure, and we can posit a cogent proto-form. We can dig in the semantics, comparing the relationships of 'end' and 'boss' with those known to exist in established language families, and, if either or both is available, search texts or query native speakers for intermediate and/or other meanings of chef and cabo, to see if initial congruence in meaning makes sense. (In this example, everything will fall into place very neatly to show that the two are reflexes of the same etymon. Similar examples are found and we posit genetic affiliation of the two languages.) Importantly, although with a few other examples to test it by, the chef-cabo relationship is dead obvious to an historical linguist with a little experience, the transparency of their equivalence is actually the result of painstaking application and reapplication of detailed, informed analysis, checked and rechecked against general principles. It's a rigorous process (ask any beginning student who has been required to show every step!). I don't think that most historical linguists would object a priori to the idea that widely differing language families might be historically related. It's just that--as Sally explained--as time depth increases, the accumulation of changes of the sort outlined at the outset here makes establishing relationships increasingly difficult, to the point of hopeless once we go back more than a few thousand years, IF the analyst is held to generally recognized standards of rigor in reconstruction. The less the analyst holds him/herself to these standards, the less likely it is that the results will be viewed as credible, since without the standards there's no way to distinguish chance similarity from cognate relationship. Tom Cravens email@example.com firstname.lastname@example.org P.S. Apologies if all this is enthusiastically incoherent. Having a six-year-old clamoring for attention is not conducive to cogent thought. Off to play dollies! _______________________________________________________________________________ <11:134>From RMBURIAN@VTVM1.CC.VT.EDU Sun Jul 24 15:32:22 1994 Date: Sun, 24 Jul 94 16:18:52 EDT From: "Richard M. Burian" <RMBURIAN@VTVM1.CC.VT.EDU> Subject: What does 'historiography' mean? To: email@example.com Nearly a week ago, Bonnie Blackwell responded to my request for suggestions for a course on the history and historiography of genetics with the following question. I have been off the list for a few days and apologize for not responding more promptly. ----------Message from Bonnie Blackwell------------------------- Date: Mon, 18 Jul 1994 12:43:59 -0500 (EST) From: "Bonnie Blackwell, (519)253-4232x2502" <BONN@nickel.laurentian.ca> To: firstname.lastname@example.org Subject: Re: Materials for a course in History of Genetics may i ask how you define the word historigraphy? how does that differ from history? a nonhistorian -------------End Quotation-------------------------------------- I, too, am not an historian and have not got an official account of the meaning of historiography. I was startled to find that neither the Oxford English Dictionary nor some of the older dictionaries I have at home list the meaning I had in mind. I know, however, that it is current among historians; cf. Helge Kragh, _An Introduction to the His- toriography of Science_ (Cambridge, 1987). I have only glanced at this book and do not have my copy at hand at the moment. If I get a chance, I will dig it up and see whether it sheds more light on the term than what follows. The meanings of historiography in the OED are derivative from those for historiographer. The meanings listed for that term are "1. A writer or compiler of a history ..." [first use cited, 1494]; '2. An official historian, appointed in connexion with a court, or some public institution' [first use cited, 1555], and 3. One who describes or gives an account of some natural object or objects ...; a writer of natural history [first cited use, 1579. The only cognate term that takes a step in the direction of the usage I intended is historiographic, which is defined as Pertaining to history, or the delineation of the historical sciences [first use cited, 1807]. I learned to use 'historiography' informally from a number of his torian friends. As I understand it, it refers to the underlying approach or theory taken by an historian (or group of historians) to history in general or an area of history, or a specific range of topics. In particular, what is at stake in the choice of historiography is deciding what boundaries to use in delimiting the history of concern, determining what factors to take into account in trying to explain the principal events or relationships in question, and deciding (or establishing criteria to decide) when a particular (historical) explanation of those events and relationships is adequate. Since I will be dealing with history of genetics, let me use examples concerned with that topic. For shorthand purposes, I will employ a bit of hyperbole and caricature. A number of the older histories written by scientists are 'great man' histories [they mention VERY few women!]. They have the general form here is a problem or group of problems that was misunderstood by the early workers in the field, here are the insights or set of experiments brought to bear by the great man or men at the center of the history, and here is how the field progressed thanks to the brilliance of that work. Other histories, e.g., Kohler's new _Lords of the Fly_ [about the use of Drosophila by the Morgan group] view genetics as a system for the production of knowledge. Some of these center on the role of bench workers, experimental systems, production techniques, etc., and treat these as basic to the understanding of what was and could be accomplished -- far more basic than the contribution of specific individuals. At another extreme are a number of articles written from a social constructivist perspective, according to which certain lines of work are (wholly?!) explained by the demands of the funding agencies, social needs, and so on. Thus there have been attempts to explain R.A. Fisher's entry into, and career in, population genetics in terms of his preoccupation, (reinforced by many of those in his social network) with eugenic questions and the socially motivated dissatisfaction with the style of answers proposed to eugenic problems by the work of such biometricians as Karl Pearson and his allies. Thus, Leonard Darwin, in part in his capacity as head of a eugenics society, not only encouraged Fisher, but arranged publication of his first key paper in the _Proc. Royal Soc., Edinburgh_ after the R. S. _London_ refused to publish that work. But more; the actual content of Fisher's theory is (according to some -- not me) to be understood in terms of the specific eugenic concerns he gleaned and gained from his social context. A fourth historiographical tradition views the delimitation of 'solvable' and even 'acceptable' problems within genetics to depend on the disciplinary background of the protagonists entering, or gaining control of, the discipline of genetics and on the developments in those neighboring disciplines as well as within genetics. Historians persuaded of this approach sometimes explain the differences in the development of genetics in different countries by the pre-existing differences in the development of neighboring disciplines in those countries. Thus, exemplary problems of the historiography of X (e.g., genetics) include (1) how to delimit X (e.g., should non-Mendelian eugenic work be included? should the work of Mendel himself, now recognized by many _not_ to be a Mendelian in the sense established by around 1910, be included?; (2) what sorts of explanatory factors to take into account (the concern with eugenics or with practical breeding; the different disciplinary orientations, traditions, and tools of those who came to found genetics; the differences in national scientific establishments and traditions; specific experimental outcomes; the influence of pragmatism and philosophical reductionism; .... ????); and (3) when to count a proffered historical explanation or account as adequate to the (main lines of the) task at hand. Hope this helps. I continue to be interested in receiving suggestions about the construction of a course on the history and historiography of genetics. (To repeat, I will have a small graduate audience of students from science studies and, I hope, biology.) Please reply to me off list unless your suggestions have general interest for the members of Darwin-L. Richard Burian email@example.com voice: 703 231-6760 Science Studies or Virginia Tech firstname.lastname@example.org fax: 703 231-7013 Blacksburg, VA 24061-0247 _______________________________________________________________________________ <11:135>From DARWIN@iris.uncg.edu Sun Jul 24 18:01:54 1994 Date: Sun, 24 Jul 1994 19:01:46 -0500 (EST) From: DARWIN@iris.uncg.edu Subject: July 24 -- Today in the Historical Sciences To: email@example.com Organization: University of NC at Greensboro JULY 24 -- TODAY IN THE HISTORICAL SCIENCES 1794 (200 years ago today): CHRISTIAN HEINRICH PANDER is born at Riga, Latvia. Pander will enter the University of Dorpat in 1812, where will study natural history and medicine. He will continue his studies at the Universities of Berlin, Gottingen, and Wurzburg, and will receive his medical degree from Wurzburg in 1817. A student of the great embryologists Karl Ernst von Baer and Ignaz Dollinger, Pander will be best remembered for his research on the development of the chick. He will spend the greater part of his career, however, pursuing investigations in geology and paleontology, and among his more important works will be ge zur Geognosie des russischen Reichs_ (St. Petersburg, 1830), and _Monographie der Fossilen Fische des silurischen Systems der Russisch-Baltischen Gouvernements_ (St. Petersburg, 1856). Today in the Historical Sciences is a feature of Darwin-L, an international network discussion group for professionals in the historical sciences. For more information about Darwin-L send the two-word message INFO DARWIN-L to firstname.lastname@example.org, or gopher to rjohara.uncg.edu (184.108.40.206). _______________________________________________________________________________ <11:136>From email@example.com Mon Jul 25 07:13:36 1994 Date: Mon, 25 Jul 1994 08:13:45 -0400 From: ad201@FreeNet.Carleton.CA (Donald Phillipson) To: firstname.lastname@example.org Subject: Historiography and social science Cf. Richard Burian's excellent long post July 24. RB's "social constructivist" usage appears much the same as what French-speaking scholars in "sciences humaines" call "problematique." In a case too close to home, a thesis topic of tackling the unwritten history of government science in Canada 1914-64, designed to continue and complete an excellent Toronto dissertation on 1867-1914, was vetoed at Montreal Univ. for lack of "problematique." This type of conflict seems common in Canada, not only between English- and French-language communities but between at least 2 schools in French Canada (Quebec and Montreal) and Eastern and Western scholars working in English. An example is the judgement whether the Canadian Pacific Railway was the great enabler of settlement on the prairies or a racket run by Eastern Interests to fleece gullible migrants; another is Creighton's "Forked Road," the question whether the Canadian polity and culture ought to follow a British model or an American one or a distinct and unique Canadian one. We therefore talk about "prairie radical" or "continentalist" schools of historiography -- which remains cognate with the etymology i.e. history defined by print (rather than by objects, law, current ideas in people's heads etc.) Examples can be found in Carl Berger's _The Writing of Canadian History_ (a good book) and Serge Gagnon's _Quebec and its Historians (not.) -- | Donald Phillipson, 4180 Boundary Rd., Carlsbad | | Springs, Ont., Canada K0A 1K0; tel: (613) 822-0734 | | "What I've always liked about science is its independence from | | authority"--Ontario Science Centre (name on file) 10 July 1981 | _______________________________________________________________________________ <11:137>From email@example.com Mon Jul 25 08:21:15 1994 Date: Mon, 25 Jul 1994 09:22:18 -0500 To: firstname.lastname@example.org From: email@example.com (J. A. Witkowski - Banbury Center, CSHL) Subject: Re: Historiography and social science >An example is the judgement whether the Canadian Pacific Railway >was the great enabler of settlement on the prairies or a racket run by >Eastern Interests to fleece gullible migrants I am interested that this sort of question has to be treated as one *or* the other. Why not both? _______________________________________________________________________________ <11:138>From firstname.lastname@example.org Mon Jul 25 13:01:55 1994 Date: Mon, 25 Jul 1994 11:01:48 -0700 (PDT) From: Scott C DeLancey <email@example.com> Subject: Re: Sarich, Campbell, and deep roots To: firstname.lastname@example.org On Sun, 24 Jul 1994, Kent Holsinger wrote: > In some ways, however, the problem in linguistics may be more similar to > the problems in molecular systematics than to those in morphological > systematics, as Bob O'Hara suggested. I think so. Morphological comparisons in biology seem more analogous to typological comparisons in linguistics (e.g. the observation that two languages both have postpositions rather than prepositions, have a special case marking for the subject rather than the object of a transitive clause, etc.) Similarities of this kind are of no use in establishing genetic relationship, primarily because there is generally a limited set of possibilities--for languages to choose from. Nearly all languages have adpositions, which can come either before or after their noun phrase (i.e. they must be either prepositions or postpositions), so taking any two languages at random there is a .5 chance that they will match in this feature. > Molecular systematists dealing with nucleotide sequence data, on the other > hand, have only four possible character states at every position in the > sequence: A, T, C, & G. Two *random* sequences of 1000 bases in length > are expected to be identical at about 25% of the positions. (Actually, > they'll be identical at a bit more than that, once nucleotide compositional > biases are taken into account and once you take into account the fact that > the sequences will be aligned in such a way as to maximize positional > identity.) As a result, when nucleotide sequences are very divergent from > one another, there is no way to tell whether they are descended from a > common ancestor. Does this bear any resemblance to the problem posed by > having a limited number of phonemes available? Yes, but the molecular problem is simpler in that it is chemical. That is, there are four and only four nucleotides, and each is invariable, so there's no question whether a particular sequence matches another or not. But languages don't all use the same inventory of units. Thus when you're dealing with linguistic forms you can't know a priori whether similar strings in two languages should count as a match or not. If one language has a voiced stop /b/, a voiceless /p/, and a voiceless aspirated /ph/, while another has a voiceless /p/ and an ejective /p'/, and we have a form /ban/ in the first and /p'an/ in the second with similar meanings, is that a match or not? Certainly if we are surveying the languages with a view to trying to decide whether or how they are related, we will note that pair as a likely-looking match. But it can't count as proof of anything until we have evidence that /b/ in the first language is the historical equivalent of /p'/ in the second. Scott DeLancey email@example.com Department of Linguistics University of Oregon Eugene, OR 97403, USA _______________________________________________________________________________ <11:139>From firstname.lastname@example.org Thu Jul 28 13:37:29 1994 From: Prof Vince Sarich <sarich@qal.Berkeley.EDU> Date: Thu, 28 Jul 1994 11:36:35 -0700 To: email@example.com Subject: Sarich, Campbell, and deep roots This is just a warning that the next piece is a long one. I'll be sending along some more specific thoughts on Thomason, Cravens, and Ringe in a later posting. Vincent Sarich firstname.lastname@example.org _______________________________________________________________________________ <11:140>From email@example.com Thu Jul 28 20:17:35 1994 From: Prof Vince Sarich <sarich@qal.Berkeley.EDU> Date: Thu, 28 Jul 1994 18:16:42 -0700 To: firstname.lastname@example.org Subject: Deep roots and measuring diffusion Some of you may be tempted to take various of the recent "Sarich, Campbell, and deep roots" messages seriously with respect to the difficulties inherent in discovering deep roots, and separating diffusional effects from common ancestry. Let me try to disabuse you of that temptation by simply recounting my own efforts of a few years ago along those lines. I wanted to get some sense of just how rapidly, on the average, ancestral terms would accumulate sufficient semantic and/or phonetic change so as to render them unrecognizable in a descendant language. That is, I wanted to see how seriously I should take the arguments of various linguists to the effect that one just couldn't hope to trace relationships at a time depth of 11,000 years; that is, Amerind. I provide a recent quote to that effect by a major anti-Greenbergian, Terence Kaufman (of Campbell and Kaufman fame in a previous posting of mine): (from Amazonian Linguistics: Studies in Lowland South American Languages, edited by Doris L. Payne, 1990: 23) ".... a temporal ceiling of 7,000 to 8,000 years is inherent in the methods of comparative linguistic reconstruction. We can recover genetic relationships that are that old, but probably no older than that. The methods possibly will be expanded, but for the moment we have to operate within that limit in drawing inferences." Kaufman then argues that, since the Americas are KNOWN to have been inhabited longer than the alleged limits of the comparative method, "the proof of a common origin for the indigenous languages of this hemisphere is not accessible to the comparative method as we know it." The quotes and the comment between them are taken from Merritt Ruhlen's On the Origin of Languages, pg. 183. Ringe, Cravens, and Thomason have provided a similar climate in these "pages". Do we have to take Kaufman's word for this? Do we need to go to an "expert" to check out the situation? I think not. What I did was to sit down with list of basic words (having added a few terms having to do with agriculture), Buck's A Dictionary of Selected Synonyms in the Principal Indo-European Languages, and Watkins' The American Heritage Dictionary of Indo-European Roots. I then asked what proportion of the reconstructed proto-Indo-European terms had survived in easily recognizable form (obviously allowing for some semantic and phonetic change) in modern Croatian (the only language other than English for which I could do this at all). For example, Watkins gives "many" as deriving from the PIE "menegh-"; the Croatian word is "mnogo". PIE for "snow" is "sneigwh" (the "w" here is given as a small superscript); the Croatian is "snjeg". The Croatian "zemlja" (Earth, ground, soil) clearly derives from the PIE "ghem". And so on. My minimum estimate was that 60% of the PIE forms have readily recognizable descendants in modern Croatian. A linguist colleague of mine, Bill Wang, and I then tried this with native speakers of Spanish and Bengali and got similar results. Dictionary work in other languages also suggests that the Croatian, Spanish, and Bengali results were representative. In other words, even after the 7,000 or so years separating PIE from its extant descendants, it is still more likely than not that a given PIE term will have a recognizable descendant in a given IE language. What about over 11,000 years? If it's 60% over 7,000 years, then it's (assuming each term has an equal chance of changing) 45% over 11,000. If 50% over 7,000, then 34% over 11,000. I think most linguists would not allow the equal probability of change assumption (and neither would I), but this would only strengthen my argument, as then those terms surviving longest would be less likely to change in the future. An even more telling way of looking at this is to note that we have PIE "sneigwh", English "snow", Croatian "snjeg, French "niege", Irish, "snechte", Lithuanian "sniegas". First, do we really need anything beyond a modicum of common sense to see the relationship among these? Do we need to "prove" the relationship by the use of the "comparative method"? Then, consider that lineages here are deep ones (in the context of IE). The English (Germanic), French (Romance), Croatian (Slavic), and Irish (Celtic) lineages have probably been separated from one another by 5,000 years or so. Lithuanian (Baltic) and Croatian are somewhat more closely related -- say 3,500-4,000 years of separation. I then count a total of about 2K+5K+5K+5K+1K+4K+4K, or 26,000 years of lineages connecting the 5 extant IE languages with one another and PIE over which the PIE term for that fluffy white stuff that falls from the sky in winter has maintained sufficient semantic and phonetic integrity for us to see the relationships without "expert" guidance. So what kind of sense do Kaufman's comments make now? I trust -- NONE! Now let me extend this argument to the Amerind situation -- which is where I started with it. Bob O'Hara noted: "I think Vince would agree that if one had a short chunk of sequence data that had a reasonable probability of having been subject to multiple hits, then even though one could do a standard cladidtic parsimony analysis on it and get a tree, the tree might not mean very much. That, as I understand it, is rather like what the linguists are saying." Well, yes I would agree that under those circumstances the tree might not mean very much. But that is not at all like what the linguists are saying. The reason is that in the Amerind situation we are not dealing with the linguistic equivalent of a short chunk of sequence data. As I noted in my earlier postings, the fact that there are many, many equivalents of the "snjeg" situation among Amerind languages is not a secret; nor has it been for many years; nor did most linguists begin treating it as a secret until 1987 -- the year Greenberg's Language in the Americas appeared. So let me remind them and you that they had been perfectly happy to admit that there were "many, many equivalents of the 'snjeg' situation among Amerind languages". For example (and these are taken from a previous posting), we have Campbell and Kaufman: "We do not take at all kindly to WB's (1981:908) caricature of our reservations concerning widespread forms, called Pan-Americanisms by some, for such reservation is a standard criterion of distant genetic research in the Americas (Campbell 1973). We in no way appealed to or necessarily believe in the hypothesis attributed to us of "a gigantic Proto-Amerind phylum" (WB 1981:908), rather we made reference to the legitimate practice in the investigation of remote relationships in the Americas of avoiding widespread forms. It is generally recognized that certain forms recur with similar sound and meaning in very many American Indian languages (cf. Swadesh 1954). Acknowledgement of the widespread forms presupposes no particular explanation; while some may feel that these support some far-flung genetic connection (cf. Swadesh 1954; 1967; Greenberg, 1960; etc.), it is possible that some widely shared similarities may be due to onomatopoeia, sound symbolism, perhaps diffusion, accident, or other undetermined factors." William Bright then repeats this almost cavalier granting of the existence of large numbers of similarities among Amerind languages in his book American Indian Linguistics and Literature (1984:25): "I would not be opposed to a hypothesis that the majority of recognized genetic families of American Indian languages must have had relationships of multilingualism and intense linguistic diffusion during a remote period of time, perhaps in the age when they were crossing the Bering Strait from Siberia to Alaska. We can imagine that the so-called pan-Americanisms in American Indian languages, which have attracted so much attention from 'super-groupers' like Greenberg, may have originated in that period." Here Bright echoes Levine on Haida. Both seemingly go out of their way to highlight the inherent flaw in their arguments. Bright writes of "multilingualism and INTENSE linguistic diffusion", just as Levine wrote of "EXTREMELY PROLONGED contact". In other words, it isn't a small number of similarities that link Haida with the other Na-dene languages, and the many Amerindian languages with one another -- otherwise why the use of "intense" and "extremely prolonged" -- those similarities must be many and obvious, as Greenberg and Ruhlen keep emphasizing, seemingly to no avail. And if they are there, then specific evidence of their having resulted from diffusion has to be presented (which, of course, no one has attempted to do); otherwise retention from common ancestry is the only acceptable explanation. I hope the "snjeg" example (and I could have used any number of others) has convinced you that words can maintain readily recognizable semantic and phonetic integrity over a period of time greatly exceeding the 11,000 years that speakers of Amerind languages have been here. And recovering those words in their ancestral condition should be a rather straightforward enterprise. The reason is that there are so many Amerind language lineages on which we have data -- probably at least 100 clades (families), each with an internal diversity comparable to that seen among IE languages, and more different from one another than, say, Greek and English. This means that if there was a proto-Amerind many known Amerind languages would retain the proto-Amerind terms in semantically and phonetically recognizable form. I would suggest to any doubters on the claim that in fact they do so retain hundreds of proto-Amerind terms to take a look at Merritt Ruhlen's On the Origin of Languages (Stanford University Press, 1994) -- especially Ch. 9: Amerind T'A?NA 'child, sibling', where we have the "snjeg" situation replicated in an Amerind context. Finally, let me close this by showing that the effects of diffusion can in fact often be isolated, identified, and measured objectively in the absence of any historical data. I apologize for the fact that I don't know how to paint pictures with words in our current e-mail technology, so please bear with me. Take 4 Micronesian languages: Sonsoroi, Ulithi, Nama, and Ifaluk. The distances between them -- given by the equation D = 100(-lnC), where C is the level of cognacy on a scale of 0 to 1 -- are: S/U, 45; S/N, 80; S/I, 54; U/N, 69; U/I, 35; N/I, 64. We can then draw a small tree with Sonsoroi and Ulithi sharing a period of common ancestry after the separation of the Nama lineage. The amounts of change from the Sonsoroi/Ulithi node are: to Sonsoroi, 28; to Ulithi, 18; to Nama, 52. Without an outgroup we can't know how much of that 52 accumulated along the common S/U lineage after the separation of Nama. We now attempt to place Ifaluk. From the S/U/N node, the Ifaluk distances (overages) are: with reference to Sonsoroi, 26 (that is, 54 minus 28); with reference to Ulithi, 17 (35 minus 18); with reference to Nama, 12 (64 minus 52). The idea then is to place the Ifaluk lineage so as to minimize the discrepancies among the 3 overages. One can immediately see that there is going to be a problem, as you can't at the same time put Ifaluk closer to both Nama and Ulithi (which would increase the overages with respect to those two, and decrease it with respect to Sonsoroi). But you can argue that if one places the origin of the Ifaluk lineage 7 units ((26-12)/2) along the Nama lineage with 19 units of change from there to modern Ifaluk, then the fact that the resulting Ifaluk/Ulithi distance is 44 (that is, 19+7+18), as compared to the actual 35, can be explained by positing a diffusional event of 9 units between Ifaluk and Ulithi after the separation of Sonsoroi from the latter. I might note here that these cognacy data are based on a 500-word list, so 9 units equals a not insignificant 45 words. In other words, diffusion is indicated when the tree distance comes out greater than the actual distance. A second possible solution would place the origin of Ifaluk 5 units along the Ulithi lineage with a subsequent diffusional event of 15 units between Ifaluk and Nama. This seems less likely on several grounds, but explaining would require me to do the whole tree involving 17 languages, and this forum isn't the place for such an effort. This is not to argue that you can always see diffusion through this procedure. Obviously any diffusion between 2 terminal lineages after their initial separation will simply have the effect of indicating a time of separation later than the "actual" one. But it is to say that the idea that, without a historical record, we can't tell whether similarities are due to common ancestry or diffusion (and sound symbolism, onomatopoeia, areal effects, etc.) is nonsense. A phylogenetic analysis such as that outlined above will isolate the latter from the former just as effectively for languages as it does for molecules or morphology. ------------------------------------------------------------------------------- A while ago Kent Holsinger pointed out that Joe Felsenstein had shown that one could not, even in principle, using allele frequency data ALONE, distinguish among common ancestry, gene flow, or some combination of the two, as explanations for observed patterns of genetic distances among populations. My thought here is to ask why anyone would have ever thought that one could so distinguish -- on the basis of the allele frequency data ALONE -- and why Felsenstein felt that it was necessary to "prove" that one could not? The same argument applies, as I have already noted, to historical linguistics. The question that Kaufman, Campbell, Thomason, Cravens, Ringe, and all the other anti-Greenbergians out there need to answer is why they do not reject out of hand a comment such as Bright's (repeated here): "I would not be opposed to a hypothesis that the majority of recognized genetic families of American Indian languages must have had relationships of multilingualism and intense linguistic diffusion during a remote period of time, perhaps in the age when they were crossing the Bering Strait from Siberia to Alaska. We can imagine that the so-called pan-Americanisms in American Indian languages, which have attracted so much attention from 'super-groupers' like Greenberg, may have originated in that period." as simply nonsense? How can one possibly consider that a hypothesis requiring who knows how many events ("the majority of recognized genetic families of American Indian languages must have had relationships of multilingualism and intense linguistic diffusion during a remote period of time ........") is somehow to be preferred to one of common ancestry -- which is one event? The question here -- and Bright's statement brings into sharp focus -- is why common ancestry is seen as wanting as an explanation? That is the point of Occam's Razor, and I don't think it's even addressed in the various comments made in response to my posting. I think it obvious that I am not setting up the simplistic dichotomy Thomason accuses me of. I am not being so silly as to deny that diffusion occurs. Nor -- see above -- do I think that there is no way, absent a historical record, to separate out the effects of diffusion and common ancestry. But I do maintain that is necessarily a directionality in any productive historical linguistics effort. You cannot start with diffusion as the possible explanation of similarity or identity, because there will then be nothing left to explain. Nor can you (if you want to get anywhere) consider diffusion and common ancestry as equally likely explanations. What you do is to first ask what can be explained by common ancestry -- THE POINT HERE BEING THAT IT WILL NEVER BE EVERYTHING -- and then go to diffusion, sound symbolism, areal effects, onomatopoeia, etc., for those identities or similarities for which common ancestry is found wanting as an explanation (as with Ifaluk and Ulithi). That's more than enough for this posting! I'll take on the specifics of Thomason, Craven, and Ringe in a separate piece. Vincent Sarich email@example.com _______________________________________________________________________________ <11:141>From firstname.lastname@example.org Fri Jul 29 07:53:36 1994 Date: Fri, 29 Jul 94 08:53:59 EDT From: email@example.com (Kent Holsinger) To: firstname.lastname@example.org Subject: Re: Deep roots and measuring diffusion As Vince Sarich pointed out in his long (and interesting) post, I mentioned a few days ago that > Joe Felsenstein had shown that one could not, even in principle, using > allele frequency data ALONE, distinguish among, common ancestry, gene flow, > or some combination of the two, as explanations for observed patterns of > genetic distances among populations. He then goes on to inquire > My thought here is to ask why anyone would have ever thought that one could > so distinguish -- on the basis of allele frequency data ALONE -- and why > Felsenstein felt that it was necessary to "prove" that one could not? The reason? Vince provided a good example of why in his disucssion of Sonsori, Ulithi, Nama, and Ifaluk. (I'll ignore Ifaluk, because it *does* suggest substantial diffusion as the only explanation.) If I've understood Vince correctly (and please remember that I'm treading far from my expertise hear, so correct me gently), the relationships among Sonsori, Nama, and Ifaluk, based on a distance measure derived from shared cognacy is N S U \ \/ \ / / Felsenstein's point (adapted to this situation) is this: This tree reflects shared similarity in cognates. From that similarity ALONE there is no way to tell whether S & U share more cognates because of more extensive borrowing between them, because of a shared history independent of N, or some combination of the two. -- Kent +------------------------------------------------------------------------+ | Kent E. Holsinger Internet: Kent@Darwin.EEB.UConn.Edu | | Department of Ecology & Holsinge@UConnVM.UConn.Edu | | Evolutionary Biology BITNET: Holsinge@UConnVM | | University of Connecticut, U-43 | | Storrs, CT 06269-3043 | +------------------------------------------------------------------------+ _______________________________________________________________________________ <11:142>From email@example.com Fri Jul 29 08:55:45 1994 Date: Fri, 29 Jul 1994 09:50:10 -0400 (EDT) From: "Jeffrey Streelman (BIO)" <firstname.lastname@example.org> Subject: Re: Deep roots and measuring diffusion To: email@example.com The Felsenstein paper of interest to this debate on linguistic reconstruction is Felsenstein, J. 1982. How can we infer history and geography from gene frequencies? J. Theor. Biol. 96: 9-20. todd streelman usf firstname.lastname@example.org _______________________________________________________________________________ <11:143>From email@example.com Sat Jul 30 10:51:20 1994 To: firstname.lastname@example.org Subject: Re: Deep roots and measuring diffusion Date: Sat, 30 Jul 94 11:51:14 -0400 From: Sally Thomason <email@example.com> There are two problems with Vince Sarich's claim that his Croatian example proves the people he contemptuously calls "experts" (the shudder quotes are his) to be wrong in dismissing Greenberg's methodology and evidence. He feels that, in sitting down with a list of basic words in Croatian and two authoritative sources on Indo-European etymology, and then comparing the lists, he has proved that Greenberg's methodology works. But in using *reconstructed* Proto-Indo-European words, he is not in fact using anything like Greenberg's methodology---first, because he has loaded the dice by making use of the results of over a century of "expert" research, and second, because in using the reconstructed words he has halved the distance between any two modern Indo-European languages that belong to separate branches of the family. The first point means that he has the benefit of the experts' knowledge (a) that the languages are related in the first place, (b) that certain words can be usefully compared while many, many others can't be (since in using the etymologies as his master list he automatically excludes *all* the non-cognates, i.e. all the random noise), and (c) that certain sounds that aren't similar at all can nevertheless be connected historically through regular sound change. In addition, he makes use of his own knowledge of Croatian: he knows which bits of Croatian words are prefixes and which are suffixes, so he can find a word's root without even having to think about it. Consider, for instance, his claim that `the Croatian "zemlja" (Earth, ground, soil) clearly derives from the PIE "ghem".' Oh? Are "z" and "gh" similar phonetically? No, they aren't: they have completely different articulations. Might they nevertheless be connected historically? Yes, sure. How do I know that? Because Indo-Europeanists have shown that a change from (something like) gh to z occurred regularly in Slavic. And what about the -lja in zemlja? Oh, that can safely be ignored, because it's a suffix. There's more: it would be hard to compare the Croatian word with a word in a modern IE language that looks like ghem, because, as far as I can recall (I'm in Montana for the summer, 2000 miles from my reference books), there aren't any. I'm not sure there are any even in ancient IE languages: Sanskrit has a phoneme gh, but, unless my memory is faulty, it would have h instead of gh in the descendant of a root like PIE ghem, and the root's vowel would certainly be a, not e. The second point I mentioned above has to do with the question of time depth. Proto-Indo-European was last spoken between 5000 and 6000 years ago (this is an estimate, but it's generally accepted by specialists), so that's the time depth for PIE and Croatian. But it's not the time depth for (say) Croatian and English, because their nearest common ancestor is PIE; they have each undergone five or six thousand years of change, independent of each other. So comparing a modern IE language with reconstructed PIE forms, or for that matter with an ancient language like Sanskrit, is certain to give give much better results than comparing two modern languages from separate branches. This doesn't, of course, mean that Sarich is wrong in believing that one can find lots of cognates in two modern IE languages, like Croatian and English (though I doubt his 60% figure for cognates). But to do such an experiment without loading the dice, he would have to choose two languages that are known to be related but whose common parent language has not been reconstructed at all, so that the sound correspondences and lexicon of the common ancestor are unknown; and the ancestor would also have to be at least 5000 years back in time. Since he wouldn't have the reconstructed parent-language words, he also wouldn't be able to ignore from the beginning all the words that are *not* inherited from a common ancestor; and he wouldn't come up with anything like 60% cognates between the two modern languages, using any list of basic vocabulary. Even that, however, would not approximate Greenberg's methodology. To do that, Sarich would also have to choose languages that he didn't know at all, so that he wouldn't be able to find a word's root from just eyeballing the word. (A large percentage of the errors in Greenberg's book stem from wrong assumptions about where a word's root is: for instance, he would take a word like Salish "xmench", which means "like, love", and compare it to other languages' words that mean roughly the same thing and have the form "men" -- assuming that x- is a prefix and -ch is a suffix. The problem is that the actual root of the Salish word is xm; -ench is a suffix meaning "belly".) But suppose he did an experiment with Indo-European languages using Greenberg's exact methodology and still found it easy to show that the languages are related. (This would be possible as long as he avoided the truly aberrant languages, which of course would have to be a matter of luck in his choice of languages to compare -- he couldn't exlude them deliberately without again making improper use of prior knowledge about the language family.) Would doing this successfully show that Greenberg's critics are all wet? No, because the time depths Greenberg is making claims about are, by Greenberg's own estimate, twice as great (12,000 years) as the time depth for IE; and by some other estimates the time depths are greater still. Sarich is confident that doubling the time would not decrease the number of cognates to the point where the relationship would be undetectable, using Greenberg's methodology. Maybe he's right, *if* one could also assume that the shared words would hold steady and not undergo enough sound change to make them unrecognizable. One must also assume that they would not undergo enough semantic change to make them unrecognizable. That's what those of us who have criticized Greenberg's methods doubt. It's not what our knowledge of sound change and semantic change make us expect. Sarich's claim that there are, `in the Amerind situation', `many, many equivalents of the "snjeg" situation' is simply false. There are none -- no forms at all in separate branches of Greenberg's proposed super-family whose etymology can be traced through regular sound changes back to a common ancestor. There are a few striking lookalikes that are tolerably widespread in the Americas, but that's quite a different matter: there are no regular correspondences, and in fact there aren't even any recurring correspondences (a weaker criterion) between sounds in words of same or similar meanings. I've read Merritt Ruhlen's T'A?NA paper, and it is nothing at all like a replication of the "snjeg" situation. No historical linguist would accept it as one; the words Ruhlen compares do not all carry the same or even obviously similar meanings, some don't have very similar sounds, some of them are erroneous, there are no regular or recurring correspondences to support his equations of particular sounds, etc. Finally, I found Sarich's claim that `the effects of diffusion can in fact often be isolated' interesting, since it provides support for historical linguists' claim that one has to be careful not to let borrowing skew the results of comparing languages in an effort to determine whether they're related or not: all you have to do is perform a simple thought experiment in which, of a small closely-related group, all the languages have disappeared (a common phenomenon in human cultural history) except the two that are linked by extensive borrowing. Will you then be able to distinguish borrowed from inherited vocabulary? No: no evidence. Sally Thomason firstname.lastname@example.org _______________________________________________________________________________ <11:144>From email@example.com Sat Jul 30 14:19:21 1994 Date: Sat, 30 Jul 1994 15:19:19 -0400 (EDT) From: Bob Wright <firstname.lastname@example.org> Subject: Re: Deep roots and measuring diffusion To: email@example.com Just in case any of you are baffled by this ongoing debate about alternative approaches to historical linguistics (Greenbergian vs. traditional), I might humbly direct you to a primer I wrote on the subject, published in the April 1991 issue of the Atlantic Monthly, called "Quest for the Mother Tongue." It is suitable for beginners. I can't claim complete impartiality (I have vaguely Greenbergian sympathies), but I tried to present both sides of the argument accurately. I believe you can order back issues by calling the Atlantic at (617) 536-9500. --Robert Wright Washington, DC _______________________________________________________________________________ <11:145>From firstname.lastname@example.org Sat Jul 30 14:47:38 1994 Date: Sat, 30 Jul 1994 12:47:56 -0700 From: email@example.com (Anton Sherwood) To: firstname.lastname@example.org Subject: Re: Deep roots and measuring diffusion Wright's language article is quite readable and entertaining. *\\* Anton _______________________________________________________________________________ <11:146>From email@example.com Sat Jul 30 14:51:46 1994 Date: Sat, 30 Jul 94 12:51:36 PDT From: firstname.lastname@example.org (Peter H. Salus) To: email@example.com Subject: Re: Deep roots and measuring diffusion I feel I should refrain from detailed dissection of Sally Thomason's posting regarding Sarich and Croatian. However, I found one passage irritating enough to try a brief response. ST states: >>Consider, for instance, his claim that `the Croatian "zemlja" (Earth, >>ground, soil) clearly derives from the PIE "ghem".' Oh? Are "z" and >>"gh" similar phonetically? No, they aren't: they have completely >>different articulations. Resisting an exhaustive list of distinctive features or acoustic data, I'd merely like to point out, lest ST has forgotten, that both [gh] and [z] are voiced and that neither is (say) bilabial nor nasal. Similarity is something that Troubetzkoy and Kurylowicz found difficult on a profound level. If I can set up a voiced/voiceless distinction, then no two voiced sounds can possibly have "completely different articulations." As someone who was critical of Joe Greenberg's African lumping over 30 years ago, and who had to accept the work 25 years ago, I find it difficult to throw out the American baby ... Peter ________________________________________________________________ Peter H. Salus #3303 4 Longfellow Place Boston, MA 02114 +1 617 723-3092 _______________________________________________________________________________ <11:147>From firstname.lastname@example.org Sat Jul 30 16:16:01 1994 Date: Sat, 30 Jul 94 17:15:57 EDT From: email@example.com (John Staddon) To: firstname.lastname@example.org Subject: Re: Deep roots and measuring diffusion Re: The piece on community rating: Michael Kinsley had a nice account of these "unintended consequences" in a recent New Yorker (2-5 weeks ago, as I recall). John Staddon _______________________________________________________________________________ <11:148>From IAP8EWH@MVS.OAC.UCLA.EDU Sat Jul 30 20:22:44 1994 Date: Sat, 30 Jul 94 18:22 PDT To: DARWIN-L@ukanaix.cc.ukans.edu From: IAP8EWH@MVS.OAC.UCLA.EDU Subject: Deep roots and measuring diffusion Here is a naive question prompted by the controversy over Amerind. Do the Pan-Americanisms mentioned by Campbell and Kauffman and by Bright apply also to the Na-Dene languages, which according to Greenberg are phylogenetically separate from Amerind although their speakers occupy adjacent territories? An affirmative answer would support diffusion, chance, etc., while a negative answer would support a phylogenetic unity for Amerind. Has this been checked? Any references would be appreciated. Thanks. Eric Holman, email@example.com _______________________________________________________________________________ <11:149>From GA3704@SIUCVMB.SIU.EDU Sun Jul 31 21:51:53 1994 Date: Sun, 31 Jul 94 21:49:49 CST From: "Margaret E. Winters <GA3704@SIUCVMB.SIU.EDU>" <GA3704@SIUCVMB.SIU.EDU> To: firstname.lastname@example.org Subject: deep roots and measuring diffusion I would like to second Bob Wright on his recommendation of his own "Atlantic Monthly" article from 1991 on the controversy in historical linguistics between Greenberg and others. It makes good reading and, I have found, presents things in a way that can be used in a second level historical class to the satisfaction of the students and the teacher. Cheers, Margaret Winters _______________________________________________________________________________ Darwin-L Message Log 11: 121-149 -- July 1994 End
© RJO 1995–2016