[Lexicog] Lexical polysemy and prototype semantics

Thu Apr 15 20:13:29 UTC 2004

Dear Fritz,
There's lots of work on prototype analyses of polysemous words in the
linguistic (rather than lexicographical or metalexicographical) literature.
If you're interested in an attempt to systematically chart the relevance of
prototype theory for lexicography (in a somewhat more theoretical mode than
Patrick's reply), you might have a look at my paper "The definitional
practice of dictionaries and the Cognitive Semantic conception of
polysemy". Lexicographica 17: 6-21, 2001.
Best wishes,
Dirk Geeraerts

At 15-4-2004 13:38, you wrote:
>Patrick,
>
>I am glad to hear that kind of sceptical note from someone
>who seems to be a professional dictionary-maker. Most
>dictionaries try to be precise, as you say, or try "to
>nail down", as I say. That way of approaching meaning
>is, indeed, questionable and fraught with difficulties.
>It also does not take into account the creativity of
>language by its users.
>You are right about "spurious" precision. Often componential
>analysis which is a controversial heuristic device has been
>used to establish and number these senses. Meanings are
>fuzzier. I recommend Bart Kosko's "Fuzzy Thinking. The New
>Science of Fuzzy Logic" (New York: Hyperion).
>Your example of "to abandon" points to the need of prototype
>semantics to deal with word meaning(s). What is the proto-
>typical meaning of "to abandon?"
>Has anyone on the list applied prototype semantics to a
>polysemous term? (I have applied it to Greek CHARIS for those
>who are interested). There is some literature out there about
>"What is a prototypical 'lie'?"
>
>Fritz Goerling
>
>
>
>
>My view is that English monolingual dictionaries (of the kinds that I
>have been editing all my life) give a very distorted picture of polysemy.
>Part of the problem, it seems to me, is that word meaning is vague,
>while dictionaries try to be precise. Numbered senses improve clarity,
>accessibility, readability, etc., but they imply a kind of spurious
>precision, I think
>
>Consider a simple verb at the start of the alphabet -- "abandon".
>MWIII offers 7 senses and 3 subsenses.  CED also has 7 senses.
>NSOED has 6.  These appear to be mutually exclusive, but in fact
>they are not.  For example, it could be argued that only the context,
>not the sense of the verb, is different in "abandon a site", "abandon
>a person", and "abandon a vehicle".  The 6 or 7 senses could easily
>be reduced to 3 or 4 senses by rewriting some of the definitions at
>a more general level.   Alternatively, one could further split "abandon"
>into a dozen or more senses by treating, say, "abandon a refrigerator"
>as different from "abandon a car".
>
>NSOED lumps "Parisians abandoning their city to scalding sunshine"
>in with "a schoolgirl abandoning herself to grief" (because of the to-PP).
>But other dictionaries make the split differently, giving a higher
>priority to the reflexive pronoun and a lower priority to the PP.
>If you do this, "abandoning the city to something" ends up with
>"abandoning a site".
>
>So my first point is that there is no one "correct" way to split up the
>different uses of a word into meanings.  Definition writing is more a
>matter of market forces (how big do we want our dictionary to be?),
>and (dare I say it) of art, taste, and judgement, rather than the
>application of data-driven rules.
>
>My second point is that it's often better to read a group of different
>definitions as complementary rather than mutually exclusive. I don't
>know of any dictionary users who are taught to read definitions in
>this way, and even if they were, there is nothing in the dictionary text
>to tell them which definition groups are mutually exclusive and which
>are complementary.   (If I remember rightly, in the first edition of COD,
>1911, the Fowlers used numbers only for mutually exclusive sense
>groups.)
>
>My third point is that, even when splitting is well justified (i.e. when
>senses really are mutually exclusive), there is no indication of
>relative frequency.  For many polysemous words, one sense (or
>sometimes one group of complementary senses) accounts for 80%
>or 90% of the uses, while the remainder are quite rare.   So, for
>example, "abandoning oneself to something" accounts for only
>around 1% of all uses of  "abandon" in the British National
>Corpus -- a balanced and representative collection of texts.
>Some dictionaries record an even rarer use of "abandon", a
>domain-specific term in the insurance world, defined in NSOED
>as "relinquish a claim to (property insured) to underwriters."  This
>is the sort of sense that is supported by citations collected from
>domain-specific reading, rather than from corpus analysis of a
>general corpus. I think it's fair to say that this specialist sense
>accounts for much less than 0.1% of uses of "abandon" in general
>English, but of course it's just the sort of use that users of a large
>monolingual dictionary like to have explained.
>
>Patrick Hanks
>
>
>
>
>
>
>
>
>
>
>
>
>
>----- Original Message -----
>From: "Rusmadi Baharudin" <rusmadi at dbp.gov.my>
>To: <lexicographylist at yahoogroups.com>
>Sent: Thursday, April 15, 2004 1:22 AM
>Subject: RE: [Lexicog] Other topics?
>
>
> > What about the treatment of polysemous word in the dictionary? Polysemy
> > - a multiple but related meanings for a single form - poses a problem in
> > semantic theory and the semantic applications such as lexicography and
> > natural language processing system. It seem that in lexicographic
> > practice there is no objective criteria for the analysis and the
> > treatment of this polysemous word. Anyone out there to share a comment
> > on this matter?
> >
> > -----Original Message-----
> > From: lexicography2004 [mailto:lexicography2004 at yahoo.com]
> > Sent: Thursday, April 15, 2004 6:42 AM
> > To: lexicographylist at yahoogroups.com
> > Subject: [Lexicog] Other topics?
> >
> > What other lexicography topics would any of you on the list like to
> > discuss?
> >
> >
> > Wayne
> > -----
> > Wayne Leman
> > List Facilitator
> > Cheyenne website: http://www.geocities.com/cheyenne_language
> >
> >
> >
> >
> > Yahoo! Groups Links
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> > Yahoo! Groups Links
> >
> >
> >
> >
> >
>
>
>
>
>
>Yahoo! Groups Links
>
>
>
>
>
>
>
>
>
>Yahoo! Groups Links
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/lexicography/attachments/20040415/68eb24bc/attachment.htm>