[Lexicog] Percentage of idioms vs single words

[Richard Rhodes]

Mike Maxwell wrote:

>I'll try very hard to make this my last posting on this topic, lest I wear
>out your monitors :-).

No, you don't get out of this so easily.

>
>>  It seems to me that there are two sets of assumptions in play in this
>>  discussion. First, what does it mean to be lexicalized? And second,
>>  what is the role of pragmatics?
>
>And third, what needs to be included in a dictionary? (Or as I might put it,
>what goes in a dictionary and what goes in an encyclopedia?)

You're right, that is a third assumption.

	I'd like to distinguish between what theoretically belongs in
a lexicon and what goes into the book we call a dictionary. The
constraints on the latter include many imposed by editors and
publishers, as well as those imposed by limits of manpower. The
constraints on the former depend on positions one takes theoretically.
	We're clearly in different camps. I do not believe that there
is any principled distinction among what are traditionally called
syntax, lexicon, and encyclopedia. There are certainly kinds of
knowledge crucial to language use that are more clearly grammar-like
and less lexical, and there are kinds of linguistically relevant
knowledge that are more clearly like knowledge of the world, and less
clearly lexical, but my position is that you can't draw a principled
line dividing the lexicon from either the "grammar" or the
"encyclopedia".

	Since you'll balk at both ends of this assertion, let me work
on the harder sell, that there is no principled distinction between
grammar and lexicon.
	We already cheat a lot to maintain the fiction that there is
a distinction. On the lexical side we do this in four ways. 1) We
acknowledge that part of speech is relevant to a lexical entry. 2) We
include, either in theory or practice, various kinds of collocational
information, e.g. selectional restrictions on verbs. 3) We are
trained to ignore enormous amounts of collocational information as if
it were "automatic", e.g. that hear refers to a particular relation
between a sentient being and a sound source (modulo metonymies). And
4) we just tacitly omit various kinds of lexically relevant
information, like the fact that put is an obligatory three place verb
- agent theme location.

	That the lexicon and syntax can't be divided is shown in more
dramatic ways by recent work on the constructions central to examples
like:

	The more, the merrier.

and

	What's that fly doing in my soup?

These show that some syntactic constructions crucially include
lexical items. Thus it is in principle impossible to modularize
lexical items off from syntax. It turns out that there are so many of
these kinds of constructions that the folks working on the syntactic
side of this (most notably Chuck Fillmore and Adele Goldberg) are now
talking about a "constructicon" which includes both lexical items
with their associated syntax, constructions with crucial lexical
parts, and constructions with more general syntactic properties.

	Unfortunately, I have to run now. But I'll have more to say
about lexicalization and encyclopedic knowledge in a future
contribution.

Rich Rhodes

BTW Levi's work is a book: The Syntax and Semantics of Complex
Nominals. New York: Academic Press, 1978. Dated, yes, but full of
important insights, required reading for anyone serious about English
noun compounds.

--
******************************************************************

  Richard A. Rhodes
  Department of Linguistics
  University of California
  Berkeley, CA 94720-2650
  Voice (510) 643-7325
  FAX (510) 643-5688

******************************************************************
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/lexicography/attachments/20040206/41441553/attachment.htm>