LINGUIST List 5.322

Mon 21 Mar 1994

Disc: Mainstream Linguistics

Editor for this issue: <>

Directory

1. Paul Deane, 5.294 Mainstream Linguistics
2. Bruce Nevin, 5.310 Mainstream Linguistics
3. David Prager Branner, Chinese, mainstream
4. , Mainstream linguistics

I am delighted by Robin Schafer's observations about linguists' views
on "data" vs. "real language" and the job market for linguists.
I hope we can all take the very positive view expressed, namely that
all linguists are concerned with real language and that our differences
have to do with judgements about which data matter and which can safely
be left for later generations to address.
 I also think the jobs issue is critical. Incidentally, my estimate
of 10 syntax jobs in the last three years should be interpreted as ten
jobs specifically restricted to syntax or at least syntax/semantics where
syntax is the obvious priority. There are other jobs out there, but they
aren't in linguistics deparments. Roughly speaking, the job opportunities for
individuals with linguistic training may be classified as follows:

(i) Pure theoretical jobs in linguistics departments--a very rare commodity.

(ii) Language teaching jobs, either English Language in an English dept.,
 or a position in a foreign languages dept.

(iii) "Applied linguistics"--i.e., teaching English as a second language.

(iv) Computational linguistics.

Of these, (i) is awfully hard to get unless you graduate from one of the
top four programs in the country at a time when your research focus is hot;
(iii) is hard to get if your focus is theoretical linguistics; (iv) is growing
quickly but at present it is hard to get a job on the computer side unless your
degree is more computer science than linguistics. That leaves (ii), which is
where most THEORETICAL linguists end up if they stay in linguistics at all.
And that position has definite disadvantages, as the people making decisions
on one's tenure are people who often are quite skeptical about the value of
pure theoretical work rather than something "USEFUL" like TESOL or Composition
Theory. I think in the long run, linguists are going to prove indispensible in
the development of large language-based computer programs, and that our theor-
etical training will help us to separate approaches that seem plausbile but
are fundamentally unworkable from computational approaches that are sophisti-
cated enough to handle language at a realistic level. (I am fortunate enough
to have landed a private industry job in which my linguistic training is an
asset despite my lack of a comp. sci. degree.)

On to less happy matters. Steve Berman writes:

 Does someone hold a monopoly on "the truth"? Has Connoly's
 language teaching PER SE led him to linguistic generalizations
 such as those embodied in the Obligatory Contour Principle, the
 Head Movement Constraint, or the conservativity of quantifiers?

 ... if Connolly knows "the real explanation", and knows where and
 what the "real flora and fauna" [are], perhaps he would be so kind
 as to let us know ...

 Perhaps he would also tell us what our (let alone Chomsky's) good
 is?

And Samuel Wang replies:

 This kind of statement sounds like OCP and Head Movement Constraint
 are God-given truths. Any justification for that? If they are not
 God given truths, why are these `generalizations' important?

Please, gentlemen. If this is the best we linguists can do when airing our
differences, then maybe we deserve to have our departments under administrative
siege. It's very frustrating, I know, to have someone fail even to see the
relevance of work you have devoted your professional life to, but there are
real issues under the surface.
 For example, something like the "head movement constraint" requires
that one assume a formal theory in which movement of items through the
tree (leaving traces behind them at every step). Such a theory makes certain
phenomena seem related because formally we can relate them.(That's the job
of a formalism.) But the terms of a theory have to DENOTE something
or else the whole thing is just an exercise in puzzle-solving. And things
that seem obviously related on a formalism-internal view may not be so
obviously related to those on the outside. For example, GB theory uses
traces both to represent unfilled argument slots and to keep track of
extraction-paths across tree structure. The implication, in effect, is
that anaphors, unfilled argument slots, and extraction-paths (adjoined
traces in non-A positions) form a natural class. In HPSG type accounts,
extraction paths are marked by slash features, which claims in effect that
extraction paths are a natural class with phrase-level features. I don't
find the arguments particularly persuasive in either case--on theory
external grounds. So if you want to convince me that the way your theory
slices the data is the correct way, you have to do more than show me that
you can handle them with the same formalism. You need to give me evidence
that the similarities you choose to highlight are more significant than
the differences you choose to filter out. (But note that if one can only
view things in a theory-internal perspective, the phenomena that one
filters out are not merely set aside as less important; for oneself they
do not even exist. And that sets one up to pursue a theory past its
point of diminishing returns. We need people who disagree with us to
keep us honest.

In Volume 5 issue 310 of today (Mon 21 Mar 1994), Vicki Fromkin,
IYO1VAFMVS.OAC.UCLA.EDU, refers to "myths about Chomsky's virulent
attacks on the opposition." She says:

> One would like to see 'real' evidence behind any such claims. None of
> which has been (or can be) forthcoming since they are false claims.

For a presentation at ICHoLS VI this past August, I revisited Chomsky and
Halle's arguments against the phoneme, especially as expressed in
_Current Issues in Linguistic Theory_ (Chomsky 1964). I remembered being
outraged by this when I read it in the mid 1960s. At the time, my main
interests were in syntax and semantics, and not in phonology, and I did
not pursue the matter. I suppose that's the main reason I didn't
remember (last spring) why I had found the arguments there so outrageous.

Revisiting Chomsky's account brought it all back, and this time I delved
into a detailed review of what was actually said by one of those whom he
attacked in that essay (chapter 4 of _Current Issues_), namely, his
teacher Zellig Harris. I was astonished at the extent to which Harris's
views are distorted, his claims misrepresented, and his statements
isolated from context and misleadingly recontextualized. Harris was
saying things very different from what Bloch and the others were saying,
but they were all lumped into "taxonomic phonemics." I can think of only
two interpretations: that Chomsky really did not understand what Harris
was saying--he has said as much in an interview someplace, I think, but
probably with reference to syntax--or that he was deliberately
misrepresenting Harris. Either is difficult to excuse, and the latter
would be reprehensible.

The ICHoLS paper is on the linguistics archive server,
linguistics.archive.umich.edu, in linguistics/papers, in the file
paper.ps. This is a severely pruned version of the conference pre-paper,
which is in the same location in the file contrast.ps. These are both
PostScript files produced by MicroSoft Word. Also in that directory is
the file postscrp.ini, the MicroSoft initialization file containing
definitions of the PostScript commands used in the *.ps files. To print
a file, queue the init file ahead of it; or you can concatenate the two
files first, creating a combined file to queue to your printer. Some
special characters, such as quote marks, em-dashes, and ellipses, may not
print correctly, depending on the font loaded in your printer. There are
also flat ASCII versions of these files (paper.pln and contrast.pln),
lacking any font shifts, etc., and also the MS Word files for DOS
(paper.doc and contrast.doc). There have been minor revisions of the
short paper for the proceedings; the longer paper is a preliminary draft
in need of review and revision.

I particularly asked Fritz Newmeyer to look at the longer paper when I
saw him at the conference (he did not stay long enough to hear my talk),
and he did, and said that he thought it was well argued, but that he
didn't know enough about it, and he would like to see the explanation
from the other side. So would I. I have hoped to submit the longer
paper for publication when I have time to deal with it. (I have to
support my family by means completely unrelated to linguistics, and I
have to budget my time and other resources carefully. I suspect that the
ramifications of this are not obvious to those of you who work in an
academic environment. With very rare exceptions, I cannot go to
conferences or attend lectures, for example. And there is the small
matter of writing a dissertation, resumed after 20 years, also in my
"spare time".) So, OK, perhaps this form of "publication" will have to
do after all, at least this year. I would welcome comments and
suggestions.

Although I do this somewhat against my better judgment, I am moved to do
so because I feel that Vicki's Panglossian dismissal of perceptions
shared by many of us should not go entirely unanswered.

And perhaps others will be encouraged to brave the very real political
risks and document the basis for those perceptions. I know something
about the political risks. A good part of why I left the field in 1974
was because "everyone knew" that Harris' theories were obsolete, passe'
"taxonomic linguistics," unworthy of intellectual or academic respect,
and I could not see how to continue the PhD program where I was without
abandoning intellectual integrity. I could not pursue research that I
believed had lasting value and would lead somewhere and also get a
degree, and I could not get a degree without pursuing research that I
believed was a passing fad and would not lead anywhere, except for
possible historical interest in the sociology of the field. (Hindsight
bears out that judgment.) More recently, and at a different institution,
I backed into phonology for my dissertation topic because of intolerance
for non-standard views in syntax and semantics. But I do not deceive
myself that I have a potential career in linguistics, as the field is
constituted today, and so perhaps I am less encombered by risk than many.
I wish you courage. But I also wish you well.

At the least, I urge you to please consider the evidence before
concluding that there are no instances of Chomsky's virulent attacks on
the opposition--and in this case, it wasn't even the opposition! Harris
supported Jakobsonian feature analysis (though not as excluding all other
tools of analysis), and indeed it was surely he who first told Chomsky
about it.

 Bruce Nevin
 bnlightstream.com

On the subject of GB analyses of "exotic" languages, mentioned by Leo
Connolly:

There is a dearth of material not only of the GB-type of but of all
formal linguistics and of language generally in Chinese. At best one sees
formal studies of the dialects of a few big cities, but rural dialects,
comprising at least 99% of all Chinese, are simply undescribed, let alone
unanalyzed. They are rapidly deteriorating - in 20 years it will be very
hard to find good informants for conservative varieties. Rural dialects
are vastly more diverse than urban dialects, but very few people are
doing fieldwork on them, even in China. Few fieldworkers at all get
closer to the countryside than the county seats, yet the dialects of county
seats as a rule have much more in common with each other than with the
dialects of the true countryside.

I don't dispute that formal studies have their value. But how can anyone
"generalize" without having seen even 1% of the data? It doesn't make
sense to me. There is no possible substitute for rural fieldwork.


 David Prager Branner
 Asian Languages and Literature
 University of Washington, DO-21
 Seattle, WA 98195

 <charmiiu.washington.edu>

 Leo Connolly (CONNOLLYmemstvx1.memst.edu) wrote:
 > So while it's fine to "hone in" on one area of language, the result
 > can be, and too often is, a reluctance to deal with the difficult
 > data. Don't confuse me with facts; I've got to theorize! And so the
 > exotic data get lost.

 > What I'm saying is this: GB linguists try very hard to explain what
 > they can (surprisingly often, some fine point of English) but really
 > don't deal with the interesting, "exotic" data. That's too bad.

 There are (at least) two kinds of "difficult" or "exotic" data: data
 from languages that we classify as exotic (because they are languages
 we don't know and which aren't taught in our schools), and in-depth
 data from a particular language. An example of the latter would be
 the COMP-trace phenomena discovered in English in the '70s by Joan
 Bresnan, if I remember. So counterintuitive were these data that had
 they been from a South American Indian language, I suspect many
 linguists would have rejected them.

 So far as I can tell, even back in the '60s generative syntacticians
 were using both kinds of data (Japanese is one language that comes to
 mind). I doubt whether it was ever true that English was the sole
 source of data, but today it seems doubtful to me, at least, that
 English is even the main source of data.

 As for the "real" reason for inversion in Italian psych-verb
 constructions (another issue Connoly and his interlocuters refer to),
 in many such issues there are really two questions: why inversion (or
 whatever construction) occurs (a pragmatic issue, perhaps, having to
 do with the speaker's intent), and the exact constraints on the
 construction--why it sometimes doesn't occur, or why it affects the
 particular part of the sentence it does (probably a syntactic issue).
 An example is wh-fronting: it is fairly obvious why wh-phrases in many
 languages move to the front of the sentence (the pragmatic issue), but
 one may understand that without understanding the syntactic limits
 (why you can say "Who did John claim that Kim kissed?" but not "Who
 did John claim that kissed Kim?"). The answer to the pragmatic
 question (which is typically the one typologists ask) usually provides
 no clue to the syntactic question (the one the GBers are asking), nor
 vice versa. But then nobody ever said language was simple!