[Lexicog] Draining corpora
Mike Maxwell
maxwell at LDC.UPENN.EDU
Wed Oct 20 01:38:23 UTC 2004
Ron Moe wrote:
> 1. The bathroom sink is draining slowly.
> 2. I think I've got the sink fixed. At least the water is draining now.
> 3. You forgot to drain the water in the sink.
> 4. This is the pipe that drains the sink.
<snip>
> I don't know if we need to develop a better theory of semantics or if we
> need to write definitions that can handle all the possible expressions. I
> suspect that there are general rules that would at least partially govern
> the types of expressions that are possible, such as what we see in (1-4) and
> in the 'safe' examples. But I don't have a theory, let alone rules, that
> enable me to avoid having to write at least four definitions for 'drain'. I
> don't think we need to, but I don't know where in the 'grammar' this sort of
> rule should be captured. Anybody have any ideas?
There is a literature on this. Unfortunately, I'm not in the office,
and won't be for at least another day. I believe one ref is
Pustejovsky, J., and Boguraev, Bran. 1996. Lexical semantics:
the problem of polysemy. Oxford; New York: Clarendon Press;
Oxford University Press.
(although I thought Pustejovsky was the sole author of the book I'm
thinking of, so maybe that's the wrong ref). I haven't read the ff.,
but it sounds like it should also be relevant:
Pustejovsky, James, and Boguraev, Branimir. 1994. A Richer
Characterization of Dictionary Entries: The Role of
Knowledge Representation. In Computational Approaches to
the Lexicon, eds. B.T.S. Atkins and Antonio Zampolli,
295-311. Oxford: Oxford U Press.
I think the quick summary is that in English (and presumably in some
other languages), intransitives like 1-2 are directly related to
transitives like 3-4. One question, of course, is how productive this
relation is: can we say it is rule governed? My hazy recollection of
Pustejovsky (and Boguraev?) is that it is quite productive in English
within a certain semantic domain, including verbs like 'spill', 'pour',
'evaporate', 'sublimate', 'melt', 'blow up', 'burn down/up', 'start' (as
in "They started the car/ The car started"), 'stop',... I'm sure you
can think of others.
There are of course some verbs which do not have this alternation, such
as 'dismantle' or 'construct' (although there is a famous literature on
sentences like "This book reads easily"--but that's quite another
issue). IIRC, the entailment is one way: if there is an intransitive/
inchoative, there will be a transitive/ causative, but not necessarily
the other way around. No, that's not right, there's "fall down". But I
think there is some generalization...
As for the relation between 1 and 2, this may be a matter of metonymy,
which again is quite common in English. I'm not sure what to say about
the relation between 3 and 4.
As to how all this ought to be represented in a dictionary, well that's
quite a different issue. I think the usual question of who the
dictionary is for is relevant. E.g. if you are a native English
speaker, but you don't happen to know the verb 'sublimate', then it may
be sufficient for the dictionary to give you the inchoative meaning
(which for this verb at least is much more common), and you'll
intuitively know that the causative meaning is possible. OTOH, if
you're a non-native speaker, it's probably necessary to give both (or
even all four) meanings.
> The variations in the examples above are by no means unusual. Fauconnier and
> Turner, in their book "The Way We Think," give the example of 'safe': "The
> child is safe," "The beach is safe," "The shovel is safe." The word 'safe'
> activates the scenario of 'danger' along with the features of 'savior'
> 'save' 'victim' 'dangerous person/instrument/time/place' 'safety'. It isn't
> that 'safe' has several meanings, or is modifying a person, place, or
> instrument respectively, but that it activates the scenario.
Reminds me of the scene in "Marathon Man" where Mengle asks, "Is it
safe?". If you haven't seen or read it, you should...
--
Mike Maxwell
Linguistic Data Consortium
maxwell at ldc.upenn.edu
------------------------ Yahoo! Groups Sponsor --------------------~-->
$9.95 domain names from Yahoo!. Register anything.
http://us.click.yahoo.com/J8kdrA/y20IAA/yQLSAA/HKE4lB/TM
--------------------------------------------------------------------~->
Yahoo! Groups Links
<*> To visit your group on the web, go to:
http://groups.yahoo.com/group/lexicographylist/
<*> To unsubscribe from this group, send an email to:
lexicographylist-unsubscribe at yahoogroups.com
<*> Your use of Yahoo! Groups is subject to:
http://docs.yahoo.com/info/terms/
More information about the Lexicography
mailing list