[Lexicog] Lexical polysemy

[Patrick Hanks]

RE: [Lexicog] Lexical polysemyAmsler, Robert wrote:

>>From the perspective of a computational lexicologist what I'd like to see is subsenses conveyed within a decision tree structure, with a series of tests at each branch point to determine when a sense qualifies for each major and minor specialization. Tests can be described in terms of constaints on the arguments the verb takes or in terms of usage characteristics such as the constructions in which the verb appears, but the need here is to form a higher level of testing for each sense distinction than just giving examples. 

>If the usage were "abandon a(n) <X>" then how would you describe the constraints on <X>? I >would guess that if <X> is a human being for whom one is a legal guardian, then there would be a different sense than if <X> is an inanimate entity which one owns. Does it matter whether one >abandon's <X> by taking it someplace and leaving it there vs. <X> being real estate property which one abandons by failing to pay mortgages or property taxes and leaving no forwarding address. Once again, the law might define a difference. This isn't to say that the legal precedents should be the arbiter of meanings, since there are many distinctions that are not based on legal differences (though, differences which ARE treated differently under the law would almost always seem to correspond to different senses in ordinary language), but that it would be useful to offer testing criteria for taking the branches of meaning rather than just descriptions and examples. Descriptions suffer from ambiguity which can be very hard to interpret (especially if one is consulting the dictionary because one doesn't know the meaning of every word in the description). Tests are somewhat better because they presuppose that the test can be asked and answered. I find test criteria, especially those with YES/NO answers, to be more precise than descriptions.

REPLY

Hello Bob - 

Long time no see!  Hope you are well. 

> how would you describe the constraints on <X>

Well, I would deny there are constraints. In my corpus analysis work I see plenty of preferences but no constraints. 

> I would guess that if <X> is a human being for whom one is a legal guardian, then there would be a different sense than if <X> is an inanimate entity which one owns. 

.. and somewhere in between is an animate entity which one owns.  If you abandon your dog, is it more like abandoning your wife or more like abandoning your car? It may be politically incorrect (but still true) to point out that "legal guardian" and "owner" are semantically quite close, especially insofar as the meaning of "abandon" is concerned. Anyway, is a husband the legal guardian of his wife?  Haven't we moved on from there?

Most lexical sets that I've looked at are like this: some clear prototypical members but no clear boundaries.  Where does this leave your decision tree structures?

Your discussion of sense distinctions based on test criteria and your comments that "descriptions suffer from ambiguity which can be very hard to interpret" suggest that, for you, sense distinctions must be stipulative idealizations rather than classifications based on observed usage.  I.e. a given society or speech community can STIPULATE (legally or otherwise) that a pet belongs in the set of animate entities with rights to legal guardianship or whatever. Right? 

But then you have to have had some sort of ownership relationship with something before you can abandon it, don't you? If this is right, then cars, refrigerators, wives, children, dogs and (pet) cockroaches would all belong in the same lexical set in relation to "abandon" -- though perhaps not in relation to any other verb.  The contrast would be with "abandoning oneself to grief", assuming that ownership has nothing to do with the interpretation of the reflexive.

Friendly greetings,


Patrick




  ----- Original Message ----- 
  From: Amsler, Robert 
  To: 'lexicographylist at yahoogroups.com' 
  Sent: Friday, April 16, 2004 7:24 PM
  Subject: RE: [Lexicog] Lexical polysemy




  From the perspective of a computational lexicologist what I'd like to see is subsenses conveyed within a decision tree structure, with a series of tests at each branch point to determine when a sense qualifies for each major and minor specialization. Tests can be described in terms of constaints on the arguments the verb takes or in terms of usage characteristics such as the constructions in which the verb appears, but the need here is to form a higher level of testing for each sense distinction than just giving examples. 

  If the usage were "abandon a(n) <X>" then how would you describe the constraints on <X>? I would guess that if <X> is a human being for whom one is a legal guardian, then there would be a different sense than if <X> is an inanimate entity which one owns. Does it matter whether one abandon's <X> by taking it someplace and leaving it there vs. <X> being real estate property which one abandons by failing to pay mortgages or property taxes and leaving no forwarding address. Once again, the law might define a difference. This isn't to say that the legal precedents should be the arbiter of meanings, since there are many distinctions that are not based on legal differences (though, differences which ARE treated differently under the law would almost always seem to correspond to different senses in ordinary language), but that it would be useful to offer testing criteria for taking the branches of meaning rather than just! descriptions and examples. Descriptions suffer from ambiguity which can be very hard to interpret (especially if one is consulting the dictionary because one doesn't know the meaning of every word in the description). Tests are somewhat better because they presuppose that the test can be asked and answered. I find test criteria, especially those with YES/NO answers, to be more precise than descriptions.



------------------------------------------------------------------------------
  Yahoo! Groups Links

    a.. To visit your group on the web, go to:
    http://groups.yahoo.com/group/lexicographylist/
      
    b.. To unsubscribe from this group, send an email to:
    lexicographylist-unsubscribe at yahoogroups.com
      
    c.. Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service. 


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/lexicography/attachments/20040417/35e6f2b9/attachment.htm>