Natural language research announcements

[Margaret Fleck <mfleck AT cs.uiuc.edu>]

-------- Original Message --------
Subject: LPBB
Date: Mon, 31 Oct 2005 09:36:18 -0600
From: Sarah Brown-Schmidt 
<brownsch AT uiuc.edu>
 (by way of Judy Allen)
To: Recipient List Suppressed:;


Language processing brownbaggers:

I hope to see you at this weeks' brown-bag, Thursday at 12:30 in Beckman 4269.

We will be hearing practice talks and posters for the upcoming
Psychonomics meeting. Below are 2 of the abstracts that will be
presented.  If you would like to practice your talk or poster but
haven't emailed me the abstract yet, that's no problem, just bring it
along-- See you Thursday!   -Sarah Brown-Schmidt

#1: Relative Clause Comprehension in Mandarin
Yowyu Lin & Susan M. Garnsey
Abstract: Speakers of both English and Japanese find object relative
clauses harder to understand than subject relatives. Although these
languages differ in many ways (default word order, whether relative
clauses follow or precede the head nouns they modify, etc.), they have
in common that subject relatives follow the default word order pattern,
while object relatives do not. In contrast, in Mandarin it is object
relatives that have the default word order, leading Hsiao and Gibson
(2003) to predict that they should be easier in Mandarin, and they
collected reading time data that provided weak support for their
prediction. In two Mandarin reading time studies, we found and
replicated robust support for this prediction, but we also found that
differences between subject and object relatives disappeared when head
nouns were less confusable with other nouns in the sentences,
consistent with Lewis' (1999) claims about the role of similarity-based
interference during sentence comprehension.


#2: Speech errors reflect newly learned phonotactic constraints

Jill A. Warker and Gary S. Dell

If speakers repeatedly produce a set of syllables in which all
occurrences of, say, /f/ are syllable onsets and all /s/'s
are codas, their speech errors will rapidly come to reflect
these constraints. For example, when /f/'s slips, they will slip
to other onset positions, not to coda positions. We attribute
this effect to the implicit learning of phonotactic constraints
within the experiment. In four experiments, we show that more complex
constraints, such as /f/ appearing in an onset only if the vowel is
/ae/, can also be acquired and can influence speech errors. These
constraints are learned much more slowly, however. We present
a model of the data to illustrate our view that the language
production system adapts to recent experience while also
continuing to reflect the accumulated experience of a lifetime
of speaking and listening.