8 February: Two linguistics talks Feb 15: E. Engdahl + R. Cooper
Dublin Computational Linguistics Research Seminar: Index of February 2013 | Dublin Computational Linguistics Research Seminar - Index of year: 2013 | Full index
University of Gothenburg) speak on Feb 15 (Friday) at TCD.
(1) Engdahl on Unbounded dependencies: grammar and processing
2pm, Large Conference Room, Computer Science Dept (O'Reilly)
(sponsored by CCLS: Centre for Computing and Language Studies)
(2) Cooper on Judgement, truth and probability: taste...
4pm, Room 3074, Arts Building (near Nassau St gate)
(DCLRS: Dublin Computational Linguistics Research Seminar)
ABSTRACTS
(1) Unbounded dependencies =96 on the interaction between grammar
and processing
Understanding and producing language is extremely fast and usually
effortless. We are not aware of doing anything special when we
talk to each other. But when linguists look at the structures
that are being built during the processing, a rather complex picture
emerges. This picture is corroborated by what psycholinguists find
when they look at the activity in the brain during listening or
reading, In this talk I will discuss what is involved in the
production and comprehension of so-called unbounded dependencies.
These are structures where the interpretation of an initial phrase
can not be fully determined until much later in the sentence, as
in the following example.
What was it you said I shouldn't forget to tell them to bring tomorrow?
In order to understand this question, the addressee has to keep
the initial question word what accessible until s/he has identified
the verb bring in the most embedded clause.
(2) Judgement, truth and probability: taste, (dis)agreement and compromise
There is a considerable literature on exchanges such as the following
A: This soup tastes great
B: No, it's horrible
What are A and B disagreeing about? What, if anything, gets entered onto
A's and B's dialogue gameboards as a commitment resulting from this
exchange? A standard approach to these cases is to start from a notion of
proposition defined in terms of truth in possible worlds and relativize
this notion in some way to context possibly involving A's and B's
beliefs.
We will try to argue instead that we should turn this around: we start =
from a notion of judgement, taking our inspiration from type theory. An
assertion represents our judgement about a situation being of a certain
type. In a large number of cases there is, in addition to this judgement
by an agent, also a putative fact of the matter: the situation actually =
is, or is not, of the type. When there is a fact of the matter a view of
this can be entered as a commitment on the gameboard. Otherwise, we can
only enter information about the judgement.
Some support for this view perhaps comes from hybrid cases where the =
issue of whether there is a fact of the matter is unclear.
A: This milk is sour
B: I'm not so sure
There is arguably a fact of the matter as to whether the milk is sour - =
on the other hand people's tolerance for sour milk varies and milk is =
sour to a varying degree. This suggests a probabilistic or
gradient view of judgements of whether the milk is sour with different
thresholds of sourness being used by different agents.
Consider dialogues involving epistemic modals:
A: Sam might have arrived in Paris by now
B: She must have
A: Sam must have arrived in Paris by now
B: She might have
Are A and B agreeing or disagreeing and what should be entered as a
commitment on the gameboard? It seems that they are in partial
agreement, a compromise. If we treat epistemic modals in terms of
probabilities as has recently been suggested by Lassiter and Goodman then
it seems natural to say that the shared commitment is to the
lower probability of Sam having arrived in Paris. It is perhaps not so
obvious what the shared commitment should be if we take them to be sets
of possible worlds where A and B have different accessibility relations.
Our proposal is therefore that we should analyze these examples using a
probabilistic type theory which builds on a notion of individual
judgements.
_______________________________________________
cogsci mailing list
cogsci@scss.tcd.ie
https://lists.scss.tcd.ie/mailman/listinfo/cogsci
Dublin Computational Linguistics Research Seminar - Index of February 2013 | Index of year: 2013 | Full index