LINGUIST List 4.458

Tue 15 Jun 1993

Disc: GB

Editor for this issue: <>

Directory

1. Connor Ferris, Re: 4.448 GB and non-GB
2. Swann Philip, 4.448 GB and non-GB
3. sowa, Re: Non-GB = non-person?
4. RichardHudson50, GB and not GB
5. , GB and non-GB
6. Anjum Saleemi, Re: 4.448 GB and non-GB
7. , GB and non-GB
8. Anjum Saleemi, Re: 4.448 GB and non-GB

Message 1: Re: 4.448 GB and non-GB

Barbara Need's recent posting touches on some of the mainsprings of
action and belief in any intellectual work. It doesn't take many years
of reasonably honest observation to realise that even highly intelligent
people (or perhaps that should be `even people who are highly competent
at passing examinations') even in their own specialist fields are
powerfully swayed by social and psychological motives and allegiances
(which `ought' not to come into things, and would usually be disowned)
and these largely govern how carefully and with what attitude they will
consider views and suggestions, and more radically, what views and
suggestions they will allow themselves to consider. Of course most
academic subjects are nowadays complex enough that a practitioner
can nearly always rest a rationalisation of her/his decision on the
back of some sort of argument based on some sort of data from the
field. Not that academic subjects contain more deviousness in this
respect than most other human activities. But it is still strange
that so many philosophers of science seem able to write at such length
about idealisation *within* science, and give no hint at all that the
exercise of science which they describe is itself an idealisation.
The practical results of the covert influences (good work neglected,
jobs not available, and so on) are tangible. Some may think it is
worth putting up a little resistance.

Reading Barbara Need's comments I was reminded of the role that
Freud played in the history of psychoanalysis. GB seems to me to
be a nice notation/model/whatever that formalizes some aspects of
grammar better than the Standard Theory, but GB qua GB is surely
what Chomsky himself would describe as a "notational variant" of
HPSG, CFG, LFG etc. Since we're dealing with pure competence at
an abstract level, the choice of formal apparatus is largely a
matter of taste constrained by what you are actually trying to
achieve. To utilize GB as a test of people's ideological commitment
to the possibility of a general theory of language is thus rather
like using Freud's theory of sexuality as a test of their belief
in a general theory of human development.

Philip Swann
University of Geneva

A friend of mine, who asked not to be named, made the following
observation:

 Any decent computer scientist can invent half a dozen new
 formalisms before breakfast. But in linguistics, only one
 person -- Noam Chomsky -- is allowed to invent formalisms.
 If anyone else dares to invent a new formalism, Chomsky promptly
 denounces it as a "notational variant" of one of his own.

Since my friend tries to maintain good contacts with both GB and
non-GB linguists, he didn't want to make this statement under his
own name. But since I have been working primarily in artificial
intelligence and computational linguistics, I am safely outside the
range of Chomsky's hit squad. In fact, it is considered a badge of
honor in AI to be denounced by Noam Chomsky.

John Sowa

It's been interesting to hear about the lack of unity in GB, but
as Barbara Need points out, the main point is in danger of
getting lost. In fact, if anything the disagreements among users
of GB underline the point. My original complaint was a personal
one: I don't like being ignored when syntacticians are being
counted; and Barbara adds another personal dimension, the
difficulty that some excellent non-GB syntacticians have in
getting jobs. Let me now introduce another complaint, which isn't
personal but concerns the health of our subject.

At a point where GB itself is divided on many fundamental issues,
how safe is it to assume that GB is the only available starting
point for the pursuit of truth? In a recent elementary survey
article I counted 10 theories of syntax which could reasonably be
described as "important" in some sense (and which didn't,
incidentally, include my own theory!). The aims of these theories
aren't in fact that different, so their current conclusions on
particular issues are comparable; nor is it that difficult to get
a rough understanding of them. So why not consider all these
theories as a pool of ideas, like a gene pool, from which one can
select? I know a theoretical pick-and-mix is potentially a
theoretical mess, but then how many existing theories could not
be described in some sense as a mess?

It all depends, I suppose, on where you think we are in the
history of linguistics. If you think the 36 years since 1957 are
a long time, then you may think we've reached the fine-tuning
stage where all the major issues have been settled. If on the
other hand you think that we're still working on questions that
the Greeks etc raised two thousand years ago, you may be more
aware of our continuing uncertainties over fundamentals and feel
it's premature to get too deeply into fine tuning. As I see them,
our present view of theory in terms of labelled packages (e.g.
GB) only make sense if we're into fine tuning.

Dick Hudson
Dept of Phonetics and Linguistics,
University College London,
Gower Street,
London WC1E 6BT
(071) 387 7050 ext 3152

Of course agreement within the GB school is not "universal" in an absolute
sense, but for syntacticians that find even the fundamental assumptions of
GB questionable, it is quite amazing to what extent GBers manage to agree
on highly specific assumptions and at the same time manage to virtually
monopolize the field (I think Barbara Need's experience is quite typical).
 Maybe some morphologists who feel close to GB (like Andrew Spencer)
have doubts about the role of "functional categories" in syntactic trees,
but my observations tell me that now probably the majority of GBers (and
especially students who are not burdened with years of their own research)
assume that there is an AgrSP, a TP, an AgrOP (and more), and that what
used to be "accusative Case assignment by the verb" is now "feature checking
in Spec,AgrO". And instead of "at s-structure", more and more people now
say "in overt syntax", following Chomsky 1992. (You can't notice this in
articles like Wexler & Poeppel 1993 yet, because articles in Language take
some time to appear and psycholinguists don't keep up with developments so
fast.)
 Evidently, many people find it normal to teach Chomsky's latest assumptions
to their students, and to change their own assumptions (or at least their
terminology and notation) when Chomsky changes them. I've heard a noted GB
syntactician say, "I just don't think of myself as important enough to have
my own theory". The logic seems to be this: One has to have SOME framework,
so in the default case one takes GB because of Chomsky's prestige, and
increasingly syntacticians who think they are "important enough" are edged
out of the field. (Note that GBers usually mean the Chomskyan school when they
say "the field", e.g. T. Ernst in Language 69.1 (1993):150. Chomsky himself
says in Knowledge of Language (p. 6-8) that generative grammar is not a
theory that could be refuted, but a field of study--so you can ignore others
who don't work in "the field".)
 I wonder if a similar situation exists in other disciplines. The only
analog I can think of is psychoanalysis, where Freudians accepted (or accept)
as given whatever Freud said.

Martin Haspelmath

Message 6: Re: 4.448 GB and non-GB

Some recent postings about GB/non-GB syntax reflect a controvery which is
probably going on continually: in various situations, settings, institutions,
etc. I strongly believe in the useful of debate, but it seems to me that the
controversy in question is often couched in terms which couldn't conceivably
be beneficial to anyone: for one thing, I'm often not sure if the points raised
are a comment on the sociology of the field, or it is indeed the case that
some empirical, methodological or theoretical issue is at stake. Take the
question of the post-Pollock proliferation of functional categories: I have my
doubts about some aspects of this line of research, but surely no one can claim
that the insights involved are trivial. I've often noticed that some people who
are very critical of GB exhibit uncritical acceptance of some other framework,
or some older (according to them worthier) formulation of the same set of
issues. It's probably useful to remember that today's dogma may be yesterday's
innovation, and that the process of scientific progress requires constant
exploration and experimentation with new ideas.

I don't really think there is actually such a thing as GB out there: the label
loosely refers to a set of researchers who share some assumptions, and who are
willing to try out some new possibilties collectively while remaining within
some constraints agreed upon by them. There are other groups of researchers
(for example those related to RG, various versions of PSG, etc.), who seem to
be engaged in the same sort of activity, and I think most people (including
the so-called GB linguists) would/should be happy that the field we belong to
is pluralistic: so anyone who normally prefers to stick to a certain framework
knows that not everything would be lost if the ideas forming the source of his
adopted framework were to dry out completely. It obviously would be nicer if
there was consensus on some of the basics, but that perhaps is going to happen
as a result of natural evolution rather than by design. So my point is: let's
carry on the work, but let's also keep talking to each other, that is if we
can do so without getting involved in acrimonious debate. Obviously people can
talk to each other only to the extent that they share a language, a set of
assumptions, etc., yet usually any interaction based on a certain amount of
objectivity and dispassionate thinking (to borrow one of Bertrand Russell's
favourite expressions) is very fruitful; for my part, I wouldn't mind using
some agreement features if doing so solved the problems I have to contend with;
in fact, my recent work on Hindi-Urdu syntax indicates that perhaps using a
Pollock-type framework for the analysis of Hindi-Urdu is a bit of an overkill.
A major problem which any universalist framework must come to terms with is
the following: to what extent is one justified in extrapolating from the
analysis of one language to another: related languages typically contain
subsets which are deceptively identical in some (many?) respects, and a closer
look is usually likely to reveal significant differences in the underlying
grammars. I think in some of the current linguistic theory the practice of
generalizing from one language to another is indulged in rather too quickly.
And mind you, problems of this sort occur in both GB and non-GB frameworks.

To conclude, the moral (if there can be one) is that no framework can by itself
guarantee that all of its practitioners will always produce work that is both
theoretically interesting and faithful to the data. One can always hope that
the two factors will converge, but in the absence of anything more definite
than hope, all we can do is to tolerate diversity, assuming that tolerance
will render it possible for the good ideas to surface, even if they are not
noticed immediately (or at all) for a variety of reasons. Eventually, time is
the best test of everything.

Anjum Saleemi
Linguistics Programme
National University of Singapore

Of course agreement within the GB school is not "universal" in an absolute
sense, but for syntacticians that find even the fundamental assumptions of
GB questionable, it is quite amazing to what extent GBers manage to agree
on highly specific assumptions and at the same time manage to virtually
monopolize the field (I think Barbara Need's experience is quite typical).
 Maybe some morphologists who feel close to GB (like Andrew Spencer)
have doubts about the role of "functional categories" in syntactic trees,
but my observations tell me that now probably the majority of GBers (and
especially students who are not burdened with years of their own research)
assume that there is an AgrSP, a TP, an AgrOP (and more), and that what
used to be "accusative Case assignment by the verb" is now "feature checking
in Spec,AgrO". And instead of "at s-strcuture", more and more people now
say "in overt syntax", following Chomsky 1992. (You can't notice this in
articles like Wexler & Poeppel 1993 yet, because articles in Language take
some time to appear and psycholinguists don't keep up with developments so
fast.)
 Evidently, many people find it normal to teach Chomsky's latest assumptions
to their students, and to change their own assumptions (or at least their
terminology and notation) when Chomsky changes them. I've heard a noted GB
syntactician say, "I just don't think of myself as important enough to have
my own theory". The logic seems to be this: One has to have SOME framework,
so in the default case one takes GB because of Chomsky's prestige, and
increasingly syntacticians who think they are "important enough" are edged
out of the field. (Note that GBers usually mean the Chomskyan school when they
say "the field", e.g. T. Ernst in Language 69.1 (1993):150. Chomsky himself
says in Knowledge of Language (p. 6-8) that generative grammar is not a
theory that could be refuted, but a field of study--so you can ignore others
who don't work in "the field".)
 I wonder if a similar situation exists in other disciplines. The only
analog I can think of is psychoanalysis, where Freudians accepted (or accept)
as given whatever Freud said.

Martin Haspelmath

Message 8: Re: 4.448 GB and non-GB

Some recent postings about GB/non-GB syntax reflect a controvery which is
probably going on continually: in various situations, settings, institutions,
etc. I strongly believe in the useful of debate, but it seems to me that the
controversy in question is often couched in terms which couldn't conceivably
be beneficial to anyone: for one thing, I'm often not sure if the points raised
are a comment on the sociology of the field, or it is indeed the case that
some empirical, methodological or theoretical issue is at stake. Take the
question of the post-Pollock proliferation of functional categories: I have my
doubts about some aspects of this line of research, but surely no one can claim
that the insights involved are trivial. I've often noticed that some people who
are very critical of GB exhibit uncritical acceptance of some other framework,
or some older (according to them worthier) formulation of the same set of
issues. It's probably useful to remember that today's dogma may be yesterday's
innovation, and that the process of scientific progress requires constant
exploration and experimentation with new ideas.

I don't really think there is actually such a thing as GB out there: the label
loosely refers to a set of researchers who share some assumptions, and who are
willing to try out some new possibilties collectively while remaining within
some constraints agreed upon by them. There are other groups of researchers
(for example those related to RG, various versions of PSG, etc.), who seem to
be engaged in the same sort of activity, and I think most people (including
the so-called GB linguists) would/should be happy that the field we belong to
is pluralistic: so anyone who normally prefers to stick to a certain framework
knows that not everything would be lost if the ideas forming the source of his
adopted framework were to dry out completely. It obviously would be nicer if
there was consensus on some of the basics, but that perhaps is going to happen
as a result of natural evolution rather than by design. So my point is: let's
carry on the work, but let's also keep talking to each other, that is if we
can do so without getting involved in acrimonious debate. Obviously people can
talk to each other only to the extent that they share a language, a set of
assumptions, etc., yet usually any interaction based on a certain amount of
objectivity and dispassionate thinking (to borrow one of Bertrand Russell's
favourite expressions) is very fruitful; for my part, I wouldn't mind using
some agreement features if doing so solved the problems I have to contend with;
in fact, my recent work on Hindi-Urdu syntax indicates that perhaps using a
Pollock-type framework for the analysis of Hindi-Urdu is a bit of an overkill.
A major problem which any universalist framework must come to terms with is
the following: to what extent is one justified in extrapolating from the
analysis of one language to another: related languages typically contain
subsets which are deceptively identical in some (many?) respects, and a closer
look is usually likely to reveal significant differences in the underlying
grammars. I think in some of the current linguistic theory the practice of
generalizing from one language to another is indulged in rather too quickly.
And mind you, problems of this sort occur in both GB and non-GB frameworks.

To conclude, the moral (if there can be one) is that no framework can by itself
guarantee that all of its practitioners will always produce work that is both
theoretically interesting and faithful to the data. One can always hope that
the two factors will converge, but in the absence of anything more definite
than hope, all we can do is to tolerate diversity, assuming that tolerance
will render it possible for the good ideas to surface, even if they are not
noticed immediately (or at all) for a variety of reasons. Eventually, time is
the best test of everything.

Anjum Saleemi
Linguistics Programme
National University of Singapore