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 peter@usenix.org  Tue Jul 19 16:12:15 1994

Date: Tue, 19 Jul 94 14:11:53 PDT
From: peter@usenix.org (Peter H. Salus)
To: darwin-l@ukanaix.cc.ukans.edu
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 bradshaw@uhunix.uhcc.Hawaii.Edu  Tue Jul 19 19:47:29 1994

Date: Tue, 19 Jul 1994 14:46:58 -1000 (HST)
From: Joel Bradshaw <bradshaw@uhunix.uhcc.Hawaii.Edu>
Subject: Uniformitarian Twain
To: darwin-l@ukanaix.cc.ukans.edu

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
bradshaw@uhunix.uhcc.hawaii.edu

_______________________________________________________________________________

<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: darwin-l@ukanaix.cc.ukans.edu
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 (darwin@iris.uncg.edu)
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: darwin-l@ukanaix.cc.ukans.edu
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 (darwin@iris.uncg.edu)
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 junger@samsara.law.cwru.edu  Wed Jul 20 11:26:09 1994

Date: Wed, 20 Jul 94 11:18:42 EDT
From: junger@samsara.law.cwru.edu (Peter D. Junger)
To: darwin-l@ukanaix.cc.ukans.edu
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 peter@usenix.org  Wed Jul 20 18:42:29 1994

Date: Wed, 20 Jul 94 16:42:18 PDT
From: peter@usenix.org (Peter H. Salus)
To: darwin-l@ukanaix.cc.ukans.edu
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 sally@isp.pitt.edu  Thu Jul 21 23:58:00 1994

To: darwin-l@ukanaix.cc.ukans.edu
Subject: Re: Sarich, Campbell, and deep roots
Date: Fri, 22 Jul 94 00:57:57 -0400
From: Sally Thomason <sally@isp.pitt.edu>

   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
   sally@isp.pitt.edu

_______________________________________________________________________________

<11:128>From kent@darwin.eeb.uconn.edu  Fri Jul 22 08:40:24 1994

Date: Fri, 22 Jul 94 09:03:22 EDT
From: kent@darwin.eeb.uconn.edu (Kent Holsinger)
To: darwin-l@ukanaix.cc.ukans.edu
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 dringe@unagi.cis.upenn.edu  Fri Jul 22 11:05:22 1994

To: darwin-l@ukanaix.cc.ukans.edu
Subject: historical lx again
Date: Fri, 22 Jul 1994 12:05:01 EDT
From: Don Ringe <dringe@unagi.cis.upenn.edu>

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: caduceus-l@beach.utmb.edu, cocta-l@nosferatu.cas.usf.edu,
        darwin-l@ukanaix.cc.ukans.edu, Galileo@muwayb.ucs.unimelb.edu.au,
        hastro-l@wvnvm.wvnet.edu, hopos-l@ukcc.uky.edu,
        hpsst-l@qucdn.queensu.ca, htech-l@sivm.si.edu
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
(comfort@cshl.org) and Corp Reed (corp@cshl.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 hist-sci@cshl.org.

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: darwin-l@ukanaix.cc.ukans.edu
Organization: University of NC at Greensboro

--begin forwarded message--------------

Date: Fri, 22 Jul 1994 03:05:04 -0400
From: Bryn Mawr Reviews <bmr@ccat.sas.upenn.edu>
Subject: BMR:  Scholia Reviews
Sender: bmr-l@cc.brynmawr.edu

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: hopos-l@ukcc.uky.edu, darwin-l@ukanaix.cc.ukans.edu,
        sci-tech-studies@ucsd.edu, HPSST-L%QUCDN.bitnet@vm42.cso.uiuc.edu

----------- 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: oha-l@ukcc.uky.edu

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: nbl@aip.org     (Niels Bohr Library at AIP)
please cc email to rd87@umail.umd.edu

------------ 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: darwin-l@ukanaix.cc.ukans.edu

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
cravens@macc.wisc.edu
cravens@wiscmacc.bitnet

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: darwin-l@ukanaix.cc.ukans.edu

   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: darwin-l@ukanaix.cc.ukans.edu
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          rmburian@vtvm1.cc.vt.edu   voice: 703 231-6760
Science Studies                   or
Virginia Tech           rmburian@vtvm1.bitnet        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: darwin-l@ukanaix.cc.ukans.edu
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
listserv@ukanaix.cc.ukans.edu, or gopher to rjohara.uncg.edu (152.13.44.19).

_______________________________________________________________________________

<11:136>From ad201@freenet.carleton.ca  Mon Jul 25 07:13:36 1994

Date: Mon, 25 Jul 1994 08:13:45 -0400
From: ad201@FreeNet.Carleton.CA (Donald Phillipson)
To: darwin-l@ukanaix.cc.ukans.edu
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 witkowsk@cshl.org  Mon Jul 25 08:21:15 1994

Date: Mon, 25 Jul 1994 09:22:18 -0500
To: darwin-l@ukanaix.cc.ukans.edu
From: witkowsk@cshl.org (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 delancey@darkwing.uoregon.edu  Mon Jul 25 13:01:55 1994

Date: Mon, 25 Jul 1994 11:01:48 -0700 (PDT)
From: Scott C DeLancey <delancey@darkwing.uoregon.edu>
Subject: Re: Sarich, Campbell, and deep roots
To: darwin-l@ukanaix.cc.ukans.edu

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			delancey@darkwing.uoregon.edu
Department of Linguistics
University of Oregon
Eugene, OR 97403, USA

_______________________________________________________________________________

<11:139>From sarich@qal.berkeley.edu  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: darwin-l@ukanaix.cc.ukans.edu
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
sarich@qal.berkeley.edu

_______________________________________________________________________________

<11:140>From sarich@qal.berkeley.edu  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: darwin-l@ukanaix.cc.ukans.edu
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
sarich@qal.berkeley.edu

_______________________________________________________________________________

<11:141>From kent@darwin.eeb.uconn.edu  Fri Jul 29 07:53:36 1994

Date: Fri, 29 Jul 94 08:53:59 EDT
From: kent@darwin.eeb.uconn.edu (Kent Holsinger)
To: darwin-l@ukanaix.cc.ukans.edu
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 streelma@chuma.cas.usf.edu  Fri Jul 29 08:55:45 1994

Date: Fri, 29 Jul 1994 09:50:10 -0400 (EDT)
From: "Jeffrey Streelman (BIO)" <streelma@chuma.cas.usf.edu>
Subject: Re: Deep roots and measuring diffusion
To: darwin-l@ukanaix.cc.ukans.edu

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
streelma@chuma.cas.usf.edu

_______________________________________________________________________________

<11:143>From sally@isp.pitt.edu  Sat Jul 30 10:51:20 1994

To: darwin-l@ukanaix.cc.ukans.edu
Subject: Re: Deep roots and measuring diffusion
Date: Sat, 30 Jul 94 11:51:14 -0400
From: Sally Thomason <sally@isp.pitt.edu>

   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
   sally@isp.pitt.edu

_______________________________________________________________________________

<11:144>From wright@clark.net  Sat Jul 30 14:19:21 1994

Date: Sat, 30 Jul 1994 15:19:19 -0400 (EDT)
From: Bob Wright <wright@clark.net>
Subject: Re: Deep roots and measuring diffusion
To: darwin-l@ukanaix.cc.ukans.edu

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 dasher@netcom.com  Sat Jul 30 14:47:38 1994

Date: Sat, 30 Jul 1994 12:47:56 -0700
From: dasher@netcom.com (Anton Sherwood)
To: darwin-l@ukanaix.cc.ukans.edu
Subject: Re: Deep roots and measuring diffusion

Wright's language article is quite readable and entertaining.
	*\\* Anton

_______________________________________________________________________________

<11:146>From peter@usenix.org  Sat Jul 30 14:51:46 1994

Date: Sat, 30 Jul 94 12:51:36 PDT
From: peter@usenix.org (Peter H. Salus)
To: darwin-l@ukanaix.cc.ukans.edu
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 staddon@psych.duke.edu  Sat Jul 30 16:16:01 1994

Date: Sat, 30 Jul 94 17:15:57 EDT
From: staddon@psych.duke.edu (John Staddon)
To: darwin-l@ukanaix.cc.ukans.edu
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, iap8ewh@mvs.oac.ucla.edu

_______________________________________________________________________________

<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: darwin-l@ukanaix.cc.ukans.edu
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