LINGUIST List 4.984

Tue 23 Nov 1993

FYI: Report on 2nd Comparative Workshop: Nostratic

Editor for this issue: <>


  1. , Report on Comp. Workshop

Message 1: Report on Comp. Workshop

Date: Tue, 9 Nov 93 20:10:38 ESTReport on Comp. Workshop
From: <>
Subject: Report on Comp. Workshop

Report on:
Second Workshop on Comparative Linguistics:
The Status of Nostratic: Evidence and Evaluation
Eastern Michigan University, Ypsilanti
October 21-22, 1993

 The workshop opened with welcoming remarks from
Marcia Dalbey, Head of the Department of English, EMU.
 The discussion of the Nostratic enterprize,
moderated by Joe Salmons (Univ. of Wisconsin/Purdue
U.), started with "The Insider's View of Nostratic"
presented by Mark Kaiser and it continued with Alexis
Manaster-Ramer's presentation of "The Outsider's View of

 Mark Kaiser (Souther Illinois U.) stated that no
substantive criticism of Nostratic has been put forward
yet and only the principle of Nostratic
has been rejected. He criticized Greenberg's mass
comparison and Bomhard's 1984 binary approach [though we
know that Bomhard's Nostratic is not binary any longer],
he also mentioned examples where Bomhard truncated
roots to make them match or where his phonological
reconstruction is not correct.
 Classical Nostratic undergoes changes: data
treatment is constantly refined, new (groups of)
languages are included, others are discarded; for the
time being the inclusion of further languages is a moot
question in his opinion. Afroasiatic may indeed turn out
to be a sister family (AA entities are to be revised).
Basic principles should be maintained like: working
strictly in accordance with established regular sound
correspondences, borrowings should be distinguished,
multilateral comparison is preferred to binary
comparison in order to avoid chance correspondences.

 Alexis Manaster-Ramer (Wayne State U.) sounded
evidently supportive of Nostratic although his was an
outsider's approach to the theory. He surveyed the
various attitudes toward the Nostratic hypothesis
ranging from deprecative allusions (Watkins'
Nostratosphere) to
constructive criticism; the latter, by the way, was well
demonstrated by his presentation. He cited examples
where the explanatory power of the Nostratic theory
becomes quite obvious (IE triune velars and vocalism,
IE st, sk clusters in initial position only), proposed a
feasible alternative for the reconstruction of Nostratic
affricates and, last but not least, pointed out
weaknesses like deglottalization of *q treated as a
normal process in the neighborhood of *p' in
Afroasiatic which is assumed on the basis of a handful
of examples and at the same time, non-deglottalized
variants are attested as well, plus there
are sporadic instances of deglottalization in
other environments. The case of Nostratic *sV-
'causative-desiderative' morpheme was also
questioned due to the inconsistency that it is supposed
to yield desiderative reflexes in IE and Altaic but
causative reflexes in Dravidian and AA. Not
to mention the unclear case of Dravidian voiced geminate
stops that have very few cognates outside Dravidian and
many examples show alternations with sonorant + stop
clusters, thus the proposed Nostratic origin (stop+H)
seems to be implausible.

 The above two papers were then discussed by
Brent Vine (Princeton U.), who considers himself an
interested onlooker. He raised questions like
"who is competent to do Nostratics?" since it is such a
vast field that it evidently has to be a collaborative
enterprise. He thinks that long-rangers often use
Pokorny's material uncritically ("fishing expeditions in
Pokorny's") although it is considered to be dated in
many respects by the modern state of Indo-European
studies. The reconstruction of Nostratic is lexically
based, the reconstruction of morphology has
its limitations since in the case of two-segment
morphemes the element of chance is high. [Let me
note here, that in the reconstruction of PIE two-segment
and single-segment morphemes are also established with
a high degree of certainty, furthermore
in the case of Nostratic there are several
cases where a two-segment morpheme is reconstructed on
the basis of evidence from 5 or 6 language families in
accordance with regular sound correspondences and
complete functional /semantic/ correspondence.]
 The Nostratic solution to the puzzle of IE
gutturals in Vine's opinion is considered
redundant since the gutturals no longer pose a problem
to Indo-Europeanists. [I wonder if there are Indo
Europeanists out there who are still bothered
about the status of gutturals and would be
pleased to see the Nostratic background if they knew
that there is such. Yes, I am fishing for comments from
experts who were not present at the workshop!]

 The afternoon session, moderated by Brian Joseph
(Ohio State U.), was devoted to historical and
methodological questions. Vitaly Shevoroshkin (Univ. of
Michigan) gave a survey of the history and evolution of
Nostratic ideas, noting that already Holger Pedersen,
who coined the term and first proposed the Nostratic
hypothesis, anticipated the danger of mass comparison.
And indeed in the adolescent period (Trombetti, Moeller,
Cuny) produced an enormous amount of mistakes. With the
publications of Collinder, Menges and Poppe, however, a
period of comparisons based on regular sound
correspondences set in. Nostratic studies became
really intensive with the research launched by Illich
Svitych, Dolgopol'skij and other scholars at the Moscow
Academy and it has been gaining supporters in European
countries (Yugoslavia, Czech Rep., Hungary, France) and
in the US too.
 He called attention to the rigorous nature of
the work carried out by Illich-Svitych and pointed out
that actually the first two volumes of his Nostratic
Dictionary should be treated separately from the first
fascicle of the 3rd volume, since the latter is a
collective work of the Moscow linguists who decided to
publish the dictionary after the author's premature
death. He also remarked that - despite several attempts
- the publication of the English translation of the
dictionary completed by Mark Kaiser is still in a limbo.

 Joseph Greenberg's paper on "The Convergence of
Nostratic and Eurasiatic" was read by Keith Denning.
The title comprises the basic tenet that the significant
changes in the views of Nostraticists in recent years as
to what language families should be classified as
Nostratic have reduced the difference between Nostratic
and Greenberg's Eurasiatic. [N.B. this is only regarding
the question of membership! I do not see much
convergence otherwise.] There is still some
incongruence even in respect of membership, because
Greenberg still does not include Dravidian and
Kartvelian. He considers AA to be a sister superstock to
Nostratic. At the same time it is intriguing that
Greenberg, referring to Blazek's investigations,
emphasizes that Kartvelian is connected to AA by a
significantly larger number of etymologies than to any
other Nostratic branch. [Actually Blazek mentions 65
etyma (max. 108 including less certain comparisons)
that connect Kartvelian with AA, but this is hardly
more than what he gives e.g. for Kartvelian-IE (62,
max. 100). Besides, Blazek also called attention to the
remarkable reserves of Dravidian-AA comparisons.] In
his forthcoming book Greenberg will list 63 grammatical
elements as well to support his hypothesis.

 Mark Hale (Harvard U.) in his discussion
outlined four criteria that govern a scholar
whether to pursue a hypothesis or not:
 1. significance to others or other topics,
 2. likelihood,
 3. feasibility of yielding results,
 4. independence of scholar (running risks).
[If a hypothesis catches on (as it is seen happening to
Nostratic) in any scientific field there is more to it
than just that more scholars will be positive at the end
of considering the four criteria outlined above. There
is a bulk of disquieting FACTS (!) that keep pushing a
hypothesis toward the center of attention in
professional circles.]
 Mark Hale's opinion supported Brent Vine's
concerning the reliability of Pokorny's dictionary and
also the status of the guttural problem in IE studies;
his idea that it is impossible and useless to write an
etymological dictionary raised the objection of the

 The second day started with papers investigating
the role of chance which is indeed a crucial point in
establishing true correspondences, the moderator was
Martha Ratliff (Wayne State U.).
 Robert Oswalt (California Indian Language
Center) presented a talk on "The Probabilistic
Evaluation of Similarities among Very Dissimilar
Languages". For over 25 years he has been
developing computer aided procedures to enable a
statistical determination of greater-than-chance
similarity between languages of the world. He carried
out intrastock and extrastock comparisons on the basis
of the 100-word list. He also examined the
effect of requiring 3 consonant matches and the effect
of having two alternative words in each semantic slot.
His calculations suggested that the branches of Altaic
are distant, furthermore, the Altaic-Uralic comparison
yields no significant resemblances, at least the basic
vocabulary does not reveal affinity between them. It is
still surprising though that a comparison between
Armenian and Hungarian yielded a relatively high index
in the range of 4 out of 5 or 5 out of 5 required
number of matches in the manner of articulation. Such a
relatively high index is easy to explain in the case of
German and Finnish by language contacts. The factor of
geographical position is decisive in many cases as
proximity can create affinity (IE neighboring branches
show a higher degree of correlation).

 Don Ringe (Univ. of Pennsylvania) provoked
probably the most fervent debate with his
paper entitled "A Probabilistic Evaluation of Indo
Uralic". His aim was to devise a test to discard
chance phenomena, to find a point beyond which it is not
likely that similarities between two (groups of)
languages are not random phenomena. Although
he opened by commenting that PIE and PU appear to be
the best candidates in the Nostratic group because the
probability of their relatedness is 1:47 or 1:48 which
looks better than random but is still not reliable.
Actually his final conclusion was that PIE-PU is only
one item above the threshold but doubted that a single
item could justify their relatedness. His further
assumptions were that chance similarity increases
with multilateral comparison and mathematically
demonstrated its disastrous failure.
 Most of the criticism of Ringe's investigation
was directed at the linguistic data that were
used as input for the calculations. This is indicative
of the fact that mathematical approaches to
languages are highly input sensitive, and if linguists
do not agree with the selection of the input
data there is no way to convince them that the result of
the investigation is correct. It was indeed
strange that the calculations were based on the
comparison of PIE and Proto-Fenno-Ugric rather
than PIE and PUralic, and available cognates were
missing from the basic list.

 William Baxter in his discussion of the session
expressed his approval of mathematical methods since
they increase objectivity but he added that it is better
to use them for testing rather than for solving
problems. On Oswald's paper he commented that it
presented not the traditional way a linguist would find
matches but the method is sound as far as looking for
similarities in the world's languages. Concerning
Ringe's paper he asserted that establishing a criterion
derived from the distribution of consonants within the
100-word list may induce circularity of reasoning. To
this Ringe added that circularity can be avoided since
any size of word list can be used as input.
Baxter then concluded that it is not a mathematical
truth that multilateral comparison fails in all
cases, especially if you find always 3 matches from 4
 Mark Hale found the construction of the initial
list a problem because there is no formal way of
resolving the semantic issues and added that the
linguistic approach has always been more productive in
finding real equations.

 The afternoon session concentrated on family
specific connections, Anthony Aristar (Texas A&M U.)
acted as moderator.
 Carlton Hodge surveyed "Implications of Lislakh
for Nostratic". Lislakh unites Lisramic (Afroasiatic)
and Indo-European but Hodge's comparisons seem to
significantly differ from the concept of Nostratic. He
analyzed non-conformist consonant sets in Semitic, then
compared core vocabulary sets from the Afroasiatic
branches and Indo European to arrive at a table of sound
correspondences between AA branches and IE. His
resulting system of stops is strangely uniform
unlike the sophisticated set of correspondences
postulated by the Nostratic hypothesis. Hodge
also surveyed cases of Lislakh consonants occurring with
accompanying features like aspiration and nasalization.
Consonant ablaut is of special significance in Hodge's
reconstructions and he finds that prothetic alif can
account for the presence of both CVC(C) and CCV(C)
patterns of the same root in Lislakh.

 Alexander Vovin (Univ. of Michigan) represented
another outsider's opinion with his paper "Nostratic and
Altaic: The Level of Relationship". He discussed some
of the inadequacies of Nostratic from the Altaic point
of view. Altaic personal pronouns do not show regular
correspondence with Nostratic. The Altaic forms with
initial m- ( are obviously secondary, the PA
form can be reconstructed as *bV(-n-) 'I'. The with initial t- is attested only in Mongolian
and Proto-Mong. *t- does not correspond to Proto
Manchu-Tung., PJapanese and PTurkic *s-, the initial
consonant of the pronoun. The sound
correspondence is also broken in the case of the direct
object suffix, which can be reconstructed for PA as *
ba/*-bd and thus cannot be derived from PN *-mA.
 Then Vovin examined PA, PU and PIE lexical
correspondences [Dravidian, Kartvelian and Afroasiatic
were not his concern here, though he does not discard
the possibility of their affiliation with Nostratic]. He
thinks that Altaic is still a stronger case for
Nostratic than Dravidian or AA. He treated the lexical
correspondences with a deliberate ultraconservative
approach [with an eye to achieving greater reliability
of results], i.e. accepted only the comparisons that had
straightforward semantics and that were supported by
attested forms from more than one branch of the same
family. Thus he found the following distribution of
convincing correspondences: PA- PU 47, PA-PIE 44, PIE-PU
47. To account for these as sheer chance resemblances is
impossible, at the same time, borrowing (from PIE via PU
to PA) is unlikely because then we would be faced
with a phonetic development rather difficult to explain:
 PIE C[+voice] > PU C[-voice] > PA C[+voice] or
 PIE C[+stop] > PU C[-stop] > PA C[+stop].
 His final conclusion was that Altaic is related
both to PIE and to PU but for lack of the common
personal pronouns it is better not to include it
in Nostratic but to consider it a separate related

 In the discussion William Rozycki (Indiana U.)
pointed out that Illich-Svitych was correct
in positing PA initial *p- and *k'- and the postulation
that "r" is primary and "z" is a secondary
development in Altaic can also be supported, although
specialists in the wake of Clauson would reject these
ideas. Evidently Illich-Svitych had a thorough
comparativist's approach and his Nostratic hypothesis
managed to come to correct conclusions in several
 Numerous comments were made by the above
mentioned participants, Eric Hamp and the
group of graduate students present at the workshop which
was concluded with a panel discussion followed by a
pleasant party at Helen Arsitar-Dry's house.

 To sum up, I would say that it is very promising
that the number of interested onlookers is
increasing. It is even more promising that the
interested onlookers take the effort and point out
fallacies of Nostratic reconstructions. And it is more
than promising that interested onlookers take
the courage and point out the merits of the Nostratic
 And did we enjoy the intellectual wrestling we
all were part of during those two days? For
that thanks are due to the organizing committee: Helen
Dry, Keith Denning, Brian Joseph,
Alexis Manaster-Ramer, Martha Ratliff, Joe Salmons.

Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue