I've been following this interesting exchange between And and Joe, and
learning a lot. I sympathise strongly with both sides: like Joe, I want to
avoid null elements unless I'm really forced to accept them, and I don't
think the evidence for traces is at all convincing. On the contrary. But
like And I think there sometimes is pretty compelling evidence that they
exist *as syntax* - i.e. they're not there just for the sake of the
semantics, as Joe implies.
        For example, take those null subjects of Joe's (and Paul Kay's):
(1) a. Scared you, didn't I?
      b. Blew it, didn't you?
I think we *must* reconstruct the 'missing words' *as words* in order to
explain the tags. For example, how else can you explain the "haven't" in
(2), contrasting with "didn't" in (1b)?
(2) Blown it, haven't you?
Worse still, how do you explain "there" in (3)?
(3)  No point, is there?
It's true that we may well store many or all of these examples as
ready-made utterances, but that's a side issue; and we may even have a
specific construction for the pattern (though I can't imagine what its
properties would be). The question is whether we try to understand them in
terms of more general structures and the answer must, surely, be yes. In
other words, I suspect strongly that we recognise (at some level of
awareness) the similarities between (1-3) and the full sentences in (4).
(4) a I scared you, didn't I?
      b You blew it, didn't you?
      c You've blown it, haven't you?
      d There's no point, is there?
To me this is a question of psychological fact not of elegance or anything
like that. I'm not sure how you would test it, but one way would be to
present subjects with a range of reduced sentences of this type, some with
sensible tags and others with silly ones (e.g. No point, has there?) and
ask them to pass judgement (e.g. which were said by a foreigner?). If you
included a handful that were neologisms (e.g. No xerox ink, is there?) you
could check for generalisability.
        This is why I'm so impressed by the evidence from case agreement, which
seems to me to be rock-hard evidence for syntactic (rather than semantic)
structure. If the predicative adjective agrees in case with the subject,
and you can't see a subject, there must nevertheless be one for the
predicative to agree with. But this isn't the only syntactic evidence that
I find convincing - And's extraction out of 'VP deletion' sites is just as
solid.
Dick


At 09:02 21/05/2002 +0200, you wrote:
>And Rosta wrote:
>
> >Before replying to the specifics of Joe's last two messages, and
> >arguments about whereabouts in which constructions there is or
> >isn't some null or empty or covert element, I think we need to
> >establish what a null or empty or covert element would be.
> >
> >Joe says:
> >#Personally, I think that anything that is related to movement and
> >#deletion is suspect and needs a thorough examination before being
> >#accepted.
> >#
> >#As for psycholinguistic experiments, as far as I know, there isn't any
> >#knockdown psychological evidence in favor of empty categories. This is
> >#something I'd like to look into further later on, however. Some authors
> >#claim to have evidence for empty categories, but the facts aren't so
> >#clearcut, in my opinion.  Nonetheless, neurologists and psychologists
> >#will be key in determining who's right.
> >
> >We are probably too far apart on fundamentals to debate this
> >fruitfully, seeing as you see syntax as an extension of phonology,
> >and psychological evidence as being paramount.
> >
> >I find it hard to guess what *you* mean by "empty category".
> >
>I take empty categories to be a sort of place holder:
>
>-- they have no phonology so as to make them readily observable;
>-- Chomsky's notion of "full interpretation" requires that they take
>      their semantics from "somewhere else" (though, as far as I can
>      tell, nobody pays much attention to this).
>
>In any event, the function of empty categories is merely to maintain
>the self-containedness of grammar--without them, you'd sometimes have
>to reach outside the so-called computational system to do syntax.
>
> >As for what I mean, there are different sorts, but what they
> >generally have in common is that they are elements of grammatical
> >structure that are interpreted but not pronounced.
> >
>Where does their interpretation come from?
>
> >I would have thought that the psycholinguistic evidence for such
> >things was not relatively sparse.
> >
>With traces, the problem has been that the experimental predictions
>are the same with or without traces. Big PRO has the same problem.
>On the other hand, I know of no corroboration of little pro--in
>fact, I'm unaware of experiments done on little pro.
>
>Joe
>
>

Richard (= Dick) Hudson

Phonetics and Linguistics, University College London,
Gower Street, London WC1E  6BT.
+44(0)20 7679 3152; fax +44(0)20 7383 4108;
http://www.phon.ucl.ac.uk/home/dick/home.htm