Deux exposés d'Asaf BACHRACH au DEC-ENS

[Lea Nash]

Dans le cadre du programme d'échange MIT-France entre le DEC-ENS et le 
Département de Linguistique du MIT, Asaf BACHRACH (MIT, Linguistics 
Department) donnera deux exposés à l'ENS les 23 et 26 mars prochains, 
aux horaires et salles indiqués ci-après:

1) "Antecedent Contained Deletion as VP sharing"

Vendredi 23 mars, 15h-17h
Salle de séminaire du DEC - ENS, 29, rue d'Ulm - 75005 -  RdC.


2) "Parametric study of visual word processing combining MEG and fMRI"

Lundi 26 mars, de 15h30 à 17h30
Salle Beckett - ENS, 45, rue d'Ulm - 75005 Paris


___________________________________
Antecedent Contained Deletion as VP sharing
-----------------------------------------------

Vendredi 23 mars, 15h-17h - DEC

Asaf BACHRACH (MIT, Linguistics)
(Joint work with Roni Katzir, MIT)


Résumé: Antecedent Contained Deletion (ACD) is generally assumed to 
involve VP ellipsis (VPE). Under this view, either QR or extraposition 
of the relative clause applies in the matrix VP to produce an 
appropriate antecedent VP.

i) John read every book Mary did
ii) John [read x] [every book Mary did [read x]]

In this talk we discuss a number of puzzles for the VPE account. We show 
that under this account ACD requires DP movement which violates known 
locality conditions on both QR and extraposition. We will present a new 
account of ACD, where VPE is replaced with VP sharing. The VP moves out 
of the relative clause and re-merges as  the matrix VP. This account 
captures the correct locality conditions on ACD as well as a number of 
other features of this construction.


______________________________________________________
Parametric study of visual word processing combining MEG and fMRI
------------------------------------------------------------------------

Lundi 26 mars, 15h30-17h30 - ENS, Salle Beckett

Asaf BACHRACH (MIT, Linguistics)
(Joint work with John Gabrieli, Susan Gabrieli, Alec Marantz and Dafna 
Palti)

Résumé: I will report preliminary  findings of a multi-modal imaging 
study of visual word processing designed to test the possibility of 
analyzing single trials of imaging experiments via correlations with 
continuous stimulus variables (Hauk et al 2006). Visual word processing 
has been one of the central  domains  of research  in the cognitive, and 
neurocognitive,  study of language. While behavioral paradigms have 
provided much insight into to the role of a number of lexical dimensions 
in word processing, reaction time, being a delayed, dense, measure shed 
only little light on the fine temporal structure of lexical access. 
Hemodynamic imaging technics on their own allow for the localization of 
different aspects of visual word processing but lack the temporal 
resolution that is required for the interpretation of the localization 
results. Electrophysiological methods provide high temporal resolution 
but  lack in spatial resolution. Our multi-modal study benefits from the 
high spatial resolution of fMRI and the temporal resolution of MEG. 
Another problem in the study of lexical access (and of language more 
generally) has been the fact that many different  lexical dimensions 
(e.g. frequency and length) covary  to a very high degree.  This 
property makes the widely used subtraction design  problematic. 
Following the approach of Hauk et al, we opted for a fully parametric 
design. Subjects performance on a lexical decision task was monitored 
using Magnetoencephalography (MEG) and trial related Functional Magnetic 
Resonance Imaging (fMRI) (two separate sessions). In the parametric 
design, each item was coded for 'string properties' (length, average 
bigram count and size of orthographic neighborhood) and 'lemma 
properties' (token frequency and imageability). Parameters were drawn 
from The English Lexicon Project (Balota et al 2002) and MRC 
Psycholinguistic Database (Wilson 1988). Words were chosen to vary 
continuously along all stimulus dimensions and such that the stimulus 
variables themselves were maximally decorrelated. The fMRI data provided 
spatial localization of parameter-related activation while the MEG data 
was used for temporal localization, with individual trial data 
correlated with stimulus variables ms by ms. Furthermore, individual 
subject  structural MRI and fMRI data were used to constrain the MEG 
forward model (MNE software,  Martinos Center MGH). In this talk I will 
discuss in particular spatio-temporal activation patterns in and around 
the Word Form Area (Cohen et al 2000) in the context of the debate 
regarding the putative role of this area in visual word processing.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/parislinguists/attachments/20070312/41eb8a2e/attachment.htm>