LINGUIST List 3.176

Tue 25 Feb 1992

Disc: Semantics, Selectional Restrictions

Editor for this issue: <>


Directory

  1. "R.Hudson", semantic subcategorisation
  2. Johanna Rubba, Selectional restrictions
  3. Robert Goldman, Selection restrictions

Message 1: semantic subcategorisation

Date: Thu, 20 Feb 92 08:46:00 +0semantic subcategorisation
From: "R.Hudson" <uclyrahucl.ac.uk>
Subject: semantic subcategorisation

Eric Schiller says that every theory he knows can handle semantic
restrictions on arguments, such as the restriction that the phrase to
which -ISH is attached must be a time of day, a colour, etc. I take his
word for Autolexical Syntax, but my impression is that at least some
modern theories would have difficulty in imposing semantic restrictions
on arguments EXCEPT for restrictions which can be imposed in terms of
semantic (theta) roles. To take a simple example, how can we require the
complement of FLOCK to name a group of sheep or goats, but not of cows?
Or again, how do we require the subject of German FRESSEN to be an animal,
in contrast with ESSEN (`eat'), which takes human subjects but is otherwise
synonymous? This kind of thing used to be handled in TG by selectional
features but they don't seem to be much in evidence in GB; and they
at least don't get much discussion in other theories like LFG and G, GPSG,
HPSG, RG, CG.

Dick Hudson
Dept of Phonetics and Linguistics,
University College London,
Gower Street,
London WC1E 6BT
(071) 387 7050 ext 3152
home: (081) 340 1253
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue

Message 2: Selectional restrictions

Date: Fri, 21 Feb 92 11:50:38 PSSelectional restrictions
From: Johanna Rubba <rubbabend.UCSD.EDU>
Subject: Selectional restrictions


I wish Arran lots of luck with a formalist account of these phenomena.
If he's interested in a cognitive semantic account, I can refer him
to a few works.

For the "Rolls Royce eats gasoline" type example, he can look at
conventional metaphor, especially perhaps the Great Chain of Being
model; see George Lakoff and Mark Johnson, Metaphors we live by,
U Chicago, 1980; but especially George Lakoff and Mark Turner,
"More than cool reason: A field guide to poetic metaphor" (year??
Publlisher??), and Mark Turner "Death is the mother of beauty" (1987, U Chic).

For the "Road climbs up the hill" type of example, Ron Langacker
has developed an account within Cognitive Grammar under the topics
'abstract motion' and 'subjective motion.' See his recent book,
"Concept, image, and symbol: The cognitive basis of grammar",
1991, Mouton de Gruyter, pp. 149-164, and references cited therein
(his 1987 and 1991 monographs set out the theory in all its glory).

He'll find examples in all these works, and lots more if he looks
around through texts or corpora or whatnot. There are, I'm sure,
tons of them.

Jo Rubba, UC San Diego
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue

Message 3: Selection restrictions

Date: Fri, 21 Feb 92 17:18:01 CSSelection restrictions
From: Robert Goldman <rpgcs.tulane.edu>
Subject: Selection restrictions


Along the lines of your
5) I picture something like:
 a) check to see if normal selectional restrictions are broken
 b) check the knowlege about the objects to find an alternative
 reading which does not break the restrictions.
 c) if none can be found, reject the sentence as not sensible.
 (or in a limited implementation as not understood).

I would suggest reading Hobbs, et. al's excellent work on the TACITUS
project at SRI. May I suggest the following?
inproceedings(Hobbs:87,
	author =	{Jerry R. Hobbs and Paul Martin},
	year =		{1987},
	booktitle =	IJCAI-87,
	title =		{Local Pragmatics},
	pages =		{520-523}

where IJCAI-87 is the proceedings of the International Joint
Conference on Artificial Intelligence; and

inproceedings(Hobbs:88,
	author =	{Jerry R. Hobbs and Mark Stickel and
			 Paul Martin and Douglas Edwards},
	year =		{1988},
	booktitle =	ACL-88,
	title =		{Interpretation as Abduction},
	pages =		{95-103}

ACL being the Association for Computational Linguistics' yearly
conference proceedings.

For the gory details, write to Jerry Hobbs and/or Mark Stickel for SRI
tech reports.

I would just like to warn about one danger of this scheme: that is
the assumption that a literal reading is always going to be preferable
to a metaphorical/metonymic reading. This seems to me unlikely.
(Hobbs' scheme does not make this assumption.) cf. also work done by
Eugene Charniak on myself, which is very similar to the work done
here, but whose foundations are in probability theory rather than in
theorem-proving.

R
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue