BEEP BOOP
24 views
Skip to first unread message
Asad Sayeed
Apr 27, 2014, 12:55:29 PM4/27/14
to linguistic...@googlegroups.com
Am I like, number three on this list?
Anyway, anyone who wants to know who I am can just Google my name and ignore any links that refer to someone who appears to be a Pakistani economist with the same name as me.
My current interest in semantics and semantic formalisms is in formalisms that can be deployed in modeling incremental processing. I've also gotten interested in the incremental interpretation of QR-type structures.
Yours,
--Asad.
Anyway, anyone who wants to know who I am can just Google my name and ignore any links that refer to someone who appears to be a Pakistani economist with the same name as me.
My current interest in semantics and semantic formalisms is in formalisms that can be deployed in modeling incremental processing. I've also gotten interested in the incremental interpretation of QR-type structures.
Yours,
--Asad.
Darryl McAdams
Apr 30, 2014, 4:01:53 AM4/30/14
to linguistic...@googlegroups.com
What sorts of challenges does incremental semantics face, and what's the benefit of having an incremental system? aside from the obvious ability to capture actual facts about humans!
- darryl
Asad Sayeed
May 1, 2014, 12:47:08 PM5/1/14
to linguistic...@googlegroups.com
The motivations for representing semantic processing incrementally
include, of course, economic motivations (building dialog systems that
can comprehend and react in "real time"), and they also include the kind
of "basement-level" concerns of psycholinguistics and cognitive science
(ie, the actual facts about humans!). Do semantic expectations have an
effect on the parser that are separate from syntactic expectations, how
do we represent this, and so on?
From a "linguist" point of view (as opposed to what I guess Norbert
Hornstein might call a "languist" POV), it has always seemed to me
rather unlikely that the "interface" requirements of LF are entirely
independent of the "real-world" temporal ordering of the
incoming/outgoing semantic representations, so for me, exploring
incremental semantic representations also presents an opportunity to
understand whether and how much "time" matters to competence.
The challenges of doing incremental semantics are the very classic
challenges, I suppose, of revising expectations when you've already
committed to a structure, which isn't too dissimilar to the problems of
doing syntax incrementally. One thing we haven't done, from a
psycholing/engineering perspective, is coming up with a realistic way to
connect a developing semantic structure to its larger context (usu. via
probability), something that is much better developed for syntax. From
the perspective of the formal representation, there is the question of
the extent to which an incremental semantics needs to respect axioms of
formal logic. I've learned that some people say "of course!" whereas for
me, a representation that isn't strictly logical and allows some
"incorrect" inferences shouldn't be ruled out if it makes the formalism
more convenient and economical.
Yours,
--Asad.
include, of course, economic motivations (building dialog systems that
can comprehend and react in "real time"), and they also include the kind
of "basement-level" concerns of psycholinguistics and cognitive science
(ie, the actual facts about humans!). Do semantic expectations have an
effect on the parser that are separate from syntactic expectations, how
do we represent this, and so on?
From a "linguist" point of view (as opposed to what I guess Norbert
Hornstein might call a "languist" POV), it has always seemed to me
rather unlikely that the "interface" requirements of LF are entirely
independent of the "real-world" temporal ordering of the
incoming/outgoing semantic representations, so for me, exploring
incremental semantic representations also presents an opportunity to
understand whether and how much "time" matters to competence.
The challenges of doing incremental semantics are the very classic
challenges, I suppose, of revising expectations when you've already
committed to a structure, which isn't too dissimilar to the problems of
doing syntax incrementally. One thing we haven't done, from a
psycholing/engineering perspective, is coming up with a realistic way to
connect a developing semantic structure to its larger context (usu. via
probability), something that is much better developed for syntax. From
the perspective of the formal representation, there is the question of
the extent to which an incremental semantics needs to respect axioms of
formal logic. I've learned that some people say "of course!" whereas for
me, a representation that isn't strictly logical and allows some
"incorrect" inferences shouldn't be ruled out if it makes the formalism
more convenient and economical.
Yours,
--Asad.
Darryl McAdams
May 2, 2014, 9:24:29 PM5/2/14
to linguistic...@googlegroups.com
what would you recommend as basic reading on the problem?
Reply all
Reply to author
Forward
0 new messages