Words and Sentences

   Date: Wed, 11 May 94 13:35:30 PDT
   From: "KENNETH D. LITWAK" <klitwak@vnet.ibm.com>

	While in general meaning may be conveyed by sentences or larger units,
   that seems to create a practical problem in at least two cases.  If it
   is the case that the context solves most ambiguities, and individual
   words don't especially matter,

Arggh!  That is not my point!  (and probably not Keith Massey's,
either.)  My point is NOT that individual words do not contribute to
the meaning of the sentence, nor that word choice is not significant.
My point is that an approach to lexical disambiguation that tries to
do disambiguation at the level of the word is not even descriptively
adequate (i.e. it just won't work).  The proper level to deal with
lexical dismbiguation is at the level of the utterance (and of the
utterance, at the level of text/dialogue).

				  why do writers and speakers (all of
   us, I'd argue) often pause, trying to mentally find just the "right" word?
   Sounds like a waste of time if words alone don't convey meaning.

As I said, that isn't what either Keith or I said, and, frankly, I'm
completely stymied as to how anybody could get such an idea out of
what either of us said.  Perhaps you are confusing the level of
representation at which an interpretation is determined, from the
levels which contribute to that interpretation?  Clearly, words have
meanings; but these meanings get focused by the contexts in which the
word appears.

Note also that Firth's dictum ``By a word's context shall ye know it''
is actually totally in sync with this search for le mot juste: we have
to find the word that conveys the right meaning and is also most
compatible with its context.  See the discussion of interpretation as
simultaneous equation solving below.

   Second, this does not address how to understand the meaning of a sentence
   in which one or more key words admit of multiple meanings.  This is not a
   case of ambiguity.

Ahem, how can a word ``admit of multiple meanings'' if it is not
lexically ambiguous?

		       This is a case of understanding which senses are
   possible and from those, which are meant.

This was the point of my message which produced ``une silence
profonde'', as my old French teacher used to put it: lexical
disambiguation is like solving simultaneous equations.  That is,
typically, we are not confronted with an utterance N words long, of
which N-1 are unambiguous, and 1 alone is ambiguous.  Rather, we are
confronted with utterances, many of whose words are ambiguous.
Understanding the utterance is a process of finding a global
interpretation whose individual components mutually support each
other's separate interpretations.

					      When someone using Californian
   slang says "That's really bad", the meaning of bad is not ambiguous,
   but the word "bad" does have multiple meanings and the intent of the
   sentence hinges on which of those meanings is intended.  If we are
   only going to look for meaning at the sentence level,

Well, I'd already talked about the text/dialogue level playing a role.
But the entire context, including extra-textual issues, such as the
speaker, circumstances of the text/dialogue, etc. play a role.  I
ignored mentioning them in my previous message since they seemed
beyond the scope of the discussion.

							 how do you
   propose to come to terms with Paul saying Jesus is a hilastErion?
   The meaning of this word is crucial for what the sentence means,
   and this word's "meaning" determines what the sentence means, not
   the other way around, and so in all cases (except perhaps for
   political speeches that drone on saying nothng -- intentionally),
   words may not stand on their own, but the postings I've seen today
   seem to be saying we can dismiss looking at the usages of a given
   word in determining its use and meaning in a given sentence.

Please cite me anywhere where either Keith or I ever said anything so
block-headed.  Look, I'll try to make this as easy as possible: all
I've been trying to say is that any approach to lexical disambiguation
that asks: ``How do we disambiguate this word?'' will not work.  The
more adequate approach to lexical disambiguation is one that asks:
``How do we disambiguate this word in its given context of use?''  In
my previous message, the aspects of context of use that I talked about
were the utterance it appeared in, and, further, the text/dialogue
context in which that utterance appeared.  But clearly factors
external to the current textual context are salient as well.  Among
these are the way a given speaker or writer uses a particular word.

Now in the case of a language which is no longer spoken (Modern Greek
is not Koine, even though it is descended from it), these problems are
multiplied.  Certainly, studying all the uses of a particularly
problematic word, either within a particular genre, period, or
author's corpus to determine what its range of meanings is is totally
reasonable.  (Nor did anything I say rule this out.)

   How would those who have expressed these opionions suggest
   understanding things like, John's use of alEtheia, or Paul's
   use of charis?  I'd usggest a much more symbiotic relationship
   between words and sentences than seems to have been expressed
   so far.

Ahem, the message I posted a while ago, which first used the metaphor
of lexical dismabiguation as being akin to solving simultaneous
equations was trying to make just this point.  Perhaps I should
suggest that you try reading more carefully?