[lingtalks] CRL Talks Tomorrow (Tues, Mar 17 at **2:30pm**) - Anna Holt / Lisa Rosenfelt: CUNY LAND!!

[Jamie Alexandre]

Hope to see you tomorrow!  Once again, please note the time change -- the
talks will start at 2pm.

----------------------------------------

Last week we heard how "two is not better than one", but this week we've set
out to show that, sometimes, two *can* be even better than one, with a
two-for-one presentation of CUNY-bound talks by Anna Holt and Lisa
Rosenfelt!  Come early for the happy half-hour (with snacks and drinks) in
CSB 215 at *2pm*.  All are welcome, so we hope to see you there!  Please
note that *we will be starting earlier* so as not to overlap with the
memorial service for Kuroda (http://ling.ucsd.edu/kuroda/) that is starting
at 4:30.

*2:00-2:30 (CSB 215)**:* Happy half-hour
*2:30-3:15 (CSB 280): *Anna Holt: "Resolving Conflicting Information from
First-Mention Biases and Discourse Event Structure in Ambiguous Pronoun
Interpretation in a Short Story Paradigm"
*3:15**-4:00 (CSB 280): *Lisa Rosenfelt: "No ERP evidence for automatic
first-pass parsing: Pure word category violations do not elicit early
negativity"

 **********************************************
CRL Happy Half Hour @ 2:00 in CSB 215
CRL Talks @ 2:30 in CSB 280
**********************************************
*Resolving Conflicting Information from First-Mention Biases and Discourse
Event Structure in Ambiguous Pronoun Interpretation in a Short Story
Paradigm
(Anna Holt and Gedeon Deak)*

Making anaphoric judgments in a discourse context holds several novel
challenges when compared with simple, intra-sentential anaphoric resolution.
Adults use the lexical features of a pronoun (e.g. gender, animacy, and
number) as the most reliable source of information for disambiguation.
However, when lexical features of a pronoun are underspecified, adults use
conflicting strategies with which to determine the referent of a pronoun.
Adults have a well-known preference for considering the first out of two or
more entities in a sentence—often the grammatical subject and the continuing
discourse topic-- as the most salient one (preferred pronoun referent)
(Arnold, Eisenband, Brown-Schmidt, & Trueswell, 2000). However, recent work
(Rohde, Kehler and Elman, 2007) suggest adults also use strategies which
take into account event structure and discourse cohesion when determining
the referent of a pronoun in an inter-sentential story completion paradigm.
For instance, participants prefer interpretations consistent with ongoing
action (e.g. adults spontaneously produce more goal continuations for
pronouns following sentences with a perfective verb than an imperfective
verb.) We tested how adults resolve conflicting cues to inter-sentential
pronoun interpretation, including the first mentioned entity, the most
frequently named entity, and the entity predicted by verb aspect and verb
semantics. We created a set of five sentence short stories which involve two
actors. The two actors participate in a short exchange using a transfer of
motion verb. An ambiguous pronoun undergoes an intransitive action, and
participants are asked to choose which actor is the referent of the pronoun.
Throughout these stories, we vary 1) whether or not the current topic is
also the initial subject of the first sentence (presence or absence of a
topic switch) 2) whether or not the last-mentioned actor is also the initial
subject of the first sentence and finally 3) whether the event structure
predicted by the intransitive verb suggests a goal continuation from the
actor in the transfer-of-motion sentence or a source continuation. We
collected responses as the reaction time to choose the appropriate actor
following story presentation and the percentage of choices of initial story
topic. Future work will additionally collect eye-tracking data, as pronoun
resolution in unambiguous situations is typically resolved with 200ms
(Arnold, Eisenband, Brown-Schmidt, & Trueswell, 2000).

----------------------------------------------------------------------------------------------

*No ERP evidence for automatic first-pass parsing: Pure word category
violations do not elicit early negativity
(Lisa Rosenfelt, Christopher Barkley, Kimberly K. Belvin, Chia-lin Lee, Kara
Federmeier, Robert Kluender, and Marta Kutas)
*
Certain neurocognitive processing models [1,2] map early left anterior
negativity (eLAN) onto automatic first-stage parsing—because it is elicited
by purported grammatical category violations, by hypothesis interfering with
initial syntactic assignments—and late positivity (P600) onto processes of
reanalysis occurring later in time. Crucially, however, eLAN (followed by a
P600) has been reliably elicited only by words following missing nominal
heads, as in Max's __ OF [3] and im __ BESUCHT (“visited in the __”) [4]. In
the latter, most common paradigm, the violation occurs when a verb replaces
the expected noun. Thus noun/verb violations that do not elicit early
negativity [5,6] and grammatical verb gapping that does [7,8] become
relevant to the discussion.

We compared ERP responses to word category violations with (a,b: “ungapped”)
and without (c: “gapped”) phrasal heads in stories that required reading for
comprehension rather than monitoring for syntactic well-formedness.

[...]

In sum, violation of the expected grammatical category of an incoming word
is not a sufficient (i) [5,6] condition for eliciting early negativity; it
seems to be reliably elicited only in paradigms that gap phrasal heads (ii)
[3,4,7,8,9]. If early negativity is sensitive to gapping rather than to
grammatical category per se, it cannot be the index of an automatic
first-pass parse assigning preliminary syntactic structure. Without a
reliable ERP index of modular first-pass parsing, a crucial piece of
neurocognitive evidence in support of serial parsing models is called into
question.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://pidgin.ucsd.edu/pipermail/lingtalks/attachments/20090316/53a11b1b/attachment.htm>