Stefan,
You are writing a dissertation on distributional methods?
Perhaps you can tell me what the current orthodoxy is towards what I
understand was Chomsky's earliest controversial result: that distributional
methods applied to language result in a loss of generality of representation.
i.e. In "Logical basis of linguistic theory"(at least as reported by Fred
Newmeyer in his "Generative Linguistics -- A Historical Perspective"):
"Halle has pointed out that it is generally impossible to provide a level of
representation meeting the biuniqueness condition without destroying the
generality of rules..." (Chomsky, LBLT)
I understand this was Chomksy's earliest controversial result, that it created
quite a stir, and essentially killed the distributional analysis which had
been the dominant linguistic orthodoxy of the day.
Am I right that his objection was never really successfully addressed, only
forgotten? I can't find too much about it. Perhaps you know.
Thanks,
Rob Freeman
On Wednesday 28 September 2005 01:57, Stefan Bordag wrote:
> Dear Dana,
>
> > I am looking for the list of linguistic principles used to extract
> > semantic information from corpora. I mean principles like, for example,
> > the Distributional Hypothesis introduced by Hearst.
>
> Are you looking for principles, methods, algorithms or implementations?
> principles: syntagmatic and paradigmatic relations
> methods: statistical, pattern based, syntactic annotation based
> algorithms: ah, many different ;-)
> implementations: unfortunately not too many freely available, depends on
> what you really need.
>
> As far as I know, Harris introduced the distributional hypothesis. Hearst
> was one of those who implemented it.
>
> If you could provide me with more detailed information about your goals,
> perhaps I could help you since I am writing a dissertation on that topic
> currently.
>
> Best regards,
> Stefan Bordag
This archive was generated by hypermail 2b29 : Fri Oct 14 2005 - 09:19:55 MET DST