Re: THEORY: two questions

And Rosta wrote:

>> >In my view, typological universals are due to constraints imposed by the
>> >hardware - the implementation of language in the brain - and to functional,
>> >darwinian selectional pressures (languages have to be fit to do their job).
>> >UG does not exist as anything beyond a mere language faculty that
>>imposes no
>> >constraints upon language except those that follow from general properties
>> >of the mind/brain. I should come clean and declare myself to be a rabid
>> >antiempiricist platonist.
>>
>> So I guess you believe that those aspects of language design which are
>> decidedly non-functional are remnants of earlier developmental stages
>> in language evolution--the cognitive equivalent of an appendix, or a whale's
>> finger bones.  I suppose that's a valid point of view.
>
>I'm not sure what you have in mind by "decidedly non-functional".
>Disfunctional
>elements would not be remnants of earlier stages in lg evolution; just
>remnants of an earlier stage in the history of lg. I think that in general lg
>is disfunctional only in 2 ways. First, temporarily, in the history of a
>particular lg (e.g. if sound changes create too many homophones). Second,
>like biological evolution it operates on the principle "just good enough is
>good enough".
My favourite example of a "decidedly non-functional" feature of grammar
is the constraint which says that a wh-phrase may not move out of
one half of a coordinate structure--i.e. the ungrammaticality of sentences
such as "Who did you see and Bill?".  There doesn't seem to be any
obvious communicative reason for the coordinate structure constraint.
Sentences like "Who did you see and Bill?" are not especially difficult
to process (no more so than many other constructions which are judged
grammatical).  And yet, in every language that I'm aware of where this
phenomenon has been tested, such sentences are judged ungrammatical,
and are rarely if ever produced spontaneously.  I find it hard to believe
that the coordinate structure constraint is the product of functional
parameters or darwinian selectional pressures.  It just appears to be an
arbitrary side-effect of language design.

>> >Of course, as a believer in UG, you have a harder time, and indeed, coping
>> >with differences between languages has been and continues to be the
>>Achilles
>> >heel in Generative Grammar - an ineradiable stain of arbitrariness.
>>
>> Actually, UG is the easy way out.  Superficially, language appears to be
>> autonomous from other cognitive systems, with its own unique properties,
>> and so generativists like me attribute the language faculty to some
>>mysterious
>> black box in the brain.  I prefer to think of UG as a working model
>> rather than
>> a belief.  Belief implies faith, and I have no faith in UG.  It's
>> just that the
>> UG explanation is, IMHO, the best one we've been able to come up with so
>> far.
>>
>> As for the ineradiable stain you mention:  There's nothing wrong with
>> arbitrariness, as long as it can be persuasively explained away as
>> being superficial.  :-)
>
>I was thinking of the conspicuous lack of practical success of the highly
>appealing notion of parameters, and the apparent fundamental conceptual
>conflict between Minimalism's working assumption that UG is 'perfect' and the
>failure (possibly remedied since early Minimalism?) to explain syntactic
>variation across lgs in terms of interface/output conditions.
Well, to ask whether Minimalism has "failed" or not is jumping the gun,
considering the theory has only been around for the last seven years or so.
And anyhow, Minimalism does *not* attempt to explain cross-linguistic
variation in terms of interface/output conditions at all, at least not as
far as I understand it.  UG is taken to be 'perfect' only in the sense that it
is devoid of redundancies and underspecified structural features.  All
cross-linguistic variation resides in the lexicon--viz. that aspect of the
mental grammar which is not innate, but acquired.  In other words, there
are no parameters in Minimalism other than lexically/morphologically-
driven parameters, so that once the lexicon has been mastered the mental
grammar is fully formed.

As for the notion of parameters, this is like the notion of UG:  It is
not an explanation, but a way of expressing what needs to be explained.
To say that the grammar has parameters which need to be set is just a
fancy way of saying that there is cross-linguistic variation, an empirical
fact.  So parameters definitely exist.  The work lies in trying to figure out
what the parameters are and how they work.  If generative grammar
has failed to answer that question satisfactorily, that's only because
the data is so complex.  From what little I know--and I freely admit that
my knowledge of such things is limited--non-UG-based theories have
had just as little success in this regard.

Matt.