LINGUIST List 5.39

Tue 11 Jan 1994

Disc: The Nature Of Linguistics

Editor for this issue: <>


  1. Mark Seidenberg, secular religions
  2. Ecological Linguistics,Anderson,PRT, Linguistics as Battleground
  3. , RE: 5.31 The Nature Of Linguistics

Message 1: secular religions

Date: Fri, 7 Jan 94 17:22:32 PSTsecular religions
From: Mark Seidenberg <>
Subject: secular religions

Surely the best analogy is not between linguistics and science or
linguistics and religion but rather: linguistics and psychoanalysis.

Freud = Chomsky
Interpretation of Dreams = Syntactic Structures
libido = deep structure
free association = grammaticality judgments
Charcot = Zellig Harris
lots of defensiveness about whether really a science or not = LODAWRASON
numerous followers = numerous followers


One might also add:

eventually disintegrated into competing schools
main ideas absorbed by mainstream psychology but basic theory abandoned.

Who knows?

Mark Seidenberg

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

Message 2: Linguistics as Battleground

Date: 10 Jan 94 15:45 GMT
From: Ecological Linguistics,Anderson,PRT <>
Subject: Linguistics as Battleground

A topic of recent discussion has been Linguistics as Religion vs. as Science.

Sometimes what is happening can be understood via a related opposition:

Linguistics as battleground vs. Linguistics as cooperative endeavor

and this relates to the glee with which linguists attempt to debunk the
theories of others, beyond what is required by the explorations of science.
I will come back to this point at the end, after noting some characteristics of
the battle over Language in the Americas.

Despite all of these problems, of course, many Linguists do continue to make
discoveries and advance our knowledge in the field.

The quantity of noise and dissention is simply greater than need be the case,
and greater than I think would be the case in a field like engineering which
had a more solid grounding in use and which were undergoing similar ferment in
new discoveries, simply because everyone would agree that knowing how to
reliably get the best engineering results is the paramount common goal of all
participants, on pain of economic irrelevance.

The controversy over Greenberg's "Language in the Americas" seems to me to fit
both of the two frameworks mentioned above.

In the first framework, there are indeed representatives of standard orthodoxy
who apparently really do believe that Greenberg's work is one of the most
irresponsible things to have heppened to the field. This has a strong
religious side, though some of its proponents claim or believe they are acting
purely as scientists.

I have heard some of the same persons reflect what is ambiguously a religious
and a vested-interest point of view, that Greenberg's work will take students
away from standard reliable historical linguistics.[*endnote] This is
sometimes pure finance in a period of shrinking budgets.

There is no question that Greenberg has said some things that are wrong. So
what? So has every other scholar in any field **who has said anything
interesting at all**. There is also no question that Greenberg has made many
valuable contributions to linguistics.

But it is still possible to comment on how the linguistics profession is
dealing with this matters. The question revolves around the issue of
cooperation and the obligation to deal with facts.

One recent article in IJAL noted errors in Greenberg's work, and did have the
honesty to comment that those errors, if corrected, might not change
Greenberg's conclusions. But neither this article nor any other similar review
I have seen then had the methodological discipline to take the opposite point
of view, and test whether it would in fact change Greenberg's conclusions and
if not, what that might say about the robustness of the techniques he uses,
even given all the limitations of the techniques.

There have been a number of specialists making comments that a bunch of data in
Greenberg is wrong. A few years ago I wrote to three of these, asking to have
a copy of all of the corrections. Not a single one of them provided me with a
list of the corrections (probably because they actually had not written them
down, but had only noted them mentally when scanning Greenberg and put his book
away as bad scholarship). I do not doubt that the errors really are errors in
most of these cases. But what is noteworthy is the complete lack of a
cooperative attitude. The attitude has rather been that we must destroy the

This attitude is I guess what made the first reviewer in our journal Language
critize Greenberg methodologically, neglecting to discuss the very long
introductory chapter in Greenberg's book dealing precisely with most of the
same questions of techniques. And what made the second review in Language use
the term "megalolinguistics". From the point of view of science, the state of
Joseph Greenberg's ego is quite irrelevant. What may be called megalomania by
some may be called ambition and determination and stamina by others (and the
other views are just as irrelevant as the megalo view to the questions of facts
of language relationships).

What I would like to do is see a republication of Greenberg's basic data sets,
with all the possible factual corrections made, and with careful evaluation of
the degree to which such corrections strengthened apparent historical bonds
between particular languages or families, using his techniqes (bond A weaker by
3%, bond B stronger by 17%, or whatever in a more networking group of links).
In a paper to the Berkeley Linguistics Society a few years ago I attempted
this. In that paper I also attempted to consider whether a language family in
which Greenberg had specialized (Andean-Equatorial) had in effect generated
more supposed correspondences with other language families by his techniques,
since they could be influenced by his degree of familiarity with the
vocabulary. I did not publish the paper at that time.

Does anyone else have parts of Greenberg's data sets on computer? Does anyone
have sets of corrections by specialists which have appeared in print, or a
bibliography of such sets of corrections organized in some way? Is there a
possibility of starting on a joint publication of this later this year?


1. My own view is that it is the obligation of the purveyors of historical
linguistics to be both interesting and open to new ideas, such as what to do
when our existing technical methods cannot find answers to a given historical
question, yet we are sure the question is answerable. This was made explicit
to me years ago in the case of Algonquian-Ritwan, where a noted scholar was
convinced of the relationship but at that time said that standard historical
methods could not establish it.

2. It is also an obligation of modesty on the part of all scientists to accept
that the techniques currently known are not necessarily the only ones possible,
that our inability to analyze a question now does not mean it will be forever
unanalyzable. This is to me relevant in the case of scholars who maintain that
one "cannot" penetrate beyond a certain historical depth because change will
obliterate all traces.

3. A third case is the dogma that one can only compare at a minimum CVC roots,
nothing shorter. As a statistical preference, this is of course valid. As an
absolute, it is not, since it would *by definition rather than by empirical
fact* completely exclude for example Siouan from historical comparisons. We
simply have to develop new techniques to deal with such cases, and then refine
our techniques in the other cases to also make use of less full material, under
whatever safeguards we can devise.

4. In theoretical linguistics, careers are made by the theorietical virtuosity
of the propounders, rather more than we should wish if we are thinking of our
obligations to society at large to increase reliable knowledge. Too rarely is
there concern for whether a theory proposed is actually true, more concern for
whether we *can* recode the data in such and such a way. This has, among other
effects, prevented the easy comparison of matches between data patterns and
theories proposed, because the theories are presented in ways that are more
incommensurable that need be the case.

5. Much of this reminds me of a nice phrase used by Jim McCawley years ago:
"Madison Ave. Si, Pennsylvania Ave. No", meaning that advertising is legitimate
in scientific persuasion, but not political manipulation. (I think I would go
further than my paraphrase here, adding a positive obligation to recognize and
deal forthrightly with alternative views, not to suppress them by omission.
The motivation to attack is not legitimate, the motivation to question
cooperatively is.)

End of Endnotes)

Sincerely, Lloyd Anderson
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue

Message 3: RE: 5.31 The Nature Of Linguistics

Date: Mon, 10 Jan 94 19:26 GMT
Subject: RE: 5.31 The Nature Of Linguistics

In Linguist 5-31, Joe Stemberger counters Neil Smith's refutation of the notion
that linguistics is a religion on the grounds that religions do go through
periods of massive upheaval and acrimonious debate from time to time.

It is perhaps salutory to think that this kind of stirring within Religions is
not confined to historical antiquity, and here in the UK the Bishop of Durham
is doing precisely that, to the consternation of many, in questioning the
veracity of the Crucifixion, and, this last Xmas, of the Virgin Birth, and the
trappings of the Christmas story, suggesting that they are merely myths,
metaphors or useful fictions of that sort.

So perhaps we can look forward to a continuation of the battles over the true
nature of linguistics in the same way.

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