Prevailing approaches do not have a computational lexicon

[Mark Johnson]

Hi Carl,

I think I understand the second issue here, and in fact I think that LFG
and R-LFG don't suffer from it precisely because c(onstituent)-structure
isn't integrated into the logic of features (i.e., there isn't a single
"logic" of all of LFG, but instead it consists of a heterogenous
collection of different but coupled "logics").

Is the problem as follows?  Suppose a substring is ambiguous; it has one
analysis as an NP ACC, and another completely different one as an NP
DAT.  In a type-logical system or similar in which the feature structure
logic (specifically, disjunction and conjunction) is tightly integrated
with the c-structure (so to speak) so that there's only really one
logic, one would then be able to prove that the string also derives the
"over-specified" category NP DAT /\ ACC.  But unfortunately ambiguous
strings don't behave like the lexically neutral forms "Frauen", showing
that the "/\" cannot be regular logical conjunction.  (But it still
could be a different binary operator in a multi-modal logic, couldn't it?)

The reason why I think that LFG and R-LFG don't suffer from this is that
each distinct c-structure is associated with its own f-structure; an
f-structure is formed from the constraints from exactly one c-structure,
so this merging of features from different c-structures simply cannot occur.

Have I got it right (more or less)?

Thanks,

Mark

Carl Pollard wrote:

>Hi Mark,
>
>>Are you referring here to examples of syncretism (or neutrality, or
>>underspecification, as opposed to ambiguity) like the celebrated
>>
>>   Er findet und hilft Frauen
>>   he finds  and helps women-ACC/DAT
>>
>>If so, what would it mean (in your terms) to analyze this in terms
>>of what you called feature cancellation?
>>
>Yes -- you know about the 1995 ACL paper that Sam Bayer and I wrote (you
>can get it from my Web page), and Sam's Language paper a year or so
>later.
>
>
>There are (at least) a couple of serious problems with analyzing
>syncretism in terms of Lambek's meet ($\sqcap$) operation. The first
>one you're probably already aware of: that there is no way to
>neutralize [F a, G b] and [F c, G d] without also getting
>[F a, G d] and [F c, G b] in the mix.
>
>The other is a foundational problem with the "frame semantics" (= the
>prosodic interpretation of types as strings), pointed out in Neal
>Whitman's (2002) dissertation: there is no way to distinguish
>neutrality from ambiguity, under the standard assumption that
>phonological identity entails prosodic identity. For example,
>accusative and genitive neutralize in Polish (Dyla 1984) but
>accusative and nominative don't, even when the forms are homophonous.
>In order to retain the ambiguity/neutrality distinction and at
>the same time the standard interpretation of meet as intersection
>in the frame semantics, one must say that homophonous forms
>that don't neutralize have distinct prosodies, e.g. the Polish
>nominative and accusative interrogative pronouns have to have
>distinct prosodies <co, 1> and <co, 2>. (This consequence was
>arrived at independently by Whitman, and by Glyn Morrill (p.c.
>to Whitman).)
>
>Examples like this suggest three analytic options:
>
>1. Continue to maintain that the ambiguity/neutrality distinction is
>   syntactically based, but accept that prosodies can include a
>   phonologically vacuous component (diacritic, uniquifier, index,
>   subscript, whatever you want to call it). This is what Morrill
>   (p.c.) opts for.
>
>2. Say that the neutrality/ambiguity distinction has no syntactic
>   basis, but rather, whether ambiguous forms can neutralize is
>   governed by pragmatic factors. This is the tentative conclusion
>   drawn by Whitman)
>
>3. Continue to maintain BOTH that the ambiguity/neutrality distinction
>   is syntactically based, and that prosody has no phonologically
>   vacuous component, but that the type logic is something other than
>   that of Lambek 1958/1961 (or elaborations thereof). This is the
>   option I am exploring (but the type logic that I have found to
>   work is not resource-sensitive).
>
>Carl
>
>That's a very interesting idea.  Given this set-up, can you predict what
>aspects of grammar will show resource sensitivity and which ones won't?
>
>
>Metaphorically, you can use syntactic/semantic resources freely, but
>you have to write IOU's every time. Prosody is the IOU's.
>
>Carl
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/lfg/attachments/20021009/6b413751/attachment.htm>