|
Abstract:
It is generally believed that during the processing of
filler-gap constructions, the filler is kept activated in working
memory (WM) until it can be assigned to its gap (Frazier &
Flores D'Arcais, 1989). Unlike in English, German WH-questions with
masculine noun phrases are immediately disambiguated by the case
marking of the WH-filler. Nevertheless, reading time studies
demonstrate increased reading times in German object WH-questions
(Schlesewsky et al., submitted), suggesting that despite its
disambiguation, the object WH-filler has to be maintained in WM
until it can be linked to its gap.
Experiment 1: Evidence from Event-Related Potentials
An ERP study was conducted to investigate the temporal course of
maintaining WH-fillers active in WM between filler and gap.
Indirect subject and object WH-questions (see examples [1] and [2])
were presented visually to 21 participants. Filler-gap distance was
varied by inserting either one or three prepositional phrases (PPs)
between the question word and the second noun phrase (NP) of
sentences such as [1] and [2].
[1] [S-WH] Karl fragt sich, wer ___ [1 PP vs. 3 PPs] den Doktor
gerufen hat.
Karl asks himself, who (NOM) [1PP/3PPs] the doctor (ACC) has
called.
[2] [O-WH] Karl fragt sich, wen [1 PP vs. 3 PPs] der Doktor ___
gerufen hat.
Karl asks himself, who (ACC) [1PP/3PPs] the doctor (NOM) has
called.
Multiword ERPs revealed a significant sustained negativity with
left-anterior focus for object as compared to subject WH-questions
between the filler and the second NP. This effect was observed in
WH-questions with long filler-gap distance only and is interpreted
as indicating increased WM demand (cf., King & Kutas, 1995;
Kluender & Münte, 1998; Ruchkin et al., 1997). We conclude
that in object WH-questions, the filler is maintained active in WM
until it can be integrated with its gap.
Experiment 2: Evidence from functional Magnetic Resonance
Imaging
In order to specify the brain regions subserving this syntactic WM
process, an fMRI study at 3 Tesla (EPI, 8 axial slices, TR=1s,
TE=30ms) has been conducted with the sentences that elicited the WM
effect in Experiment 1, and two additional sentence conditions of
equal length (see examples [3] and [4]). These two conditions
operationalize the contrast between subject and object WH-questions
but, as the PPs are positioned after the gap, they do not contain a
long-distance movement like conditions [1] and [2].
[3] [Subject-WH] Karl fragt sich, wer ___ den Doktor [3 PPs]
gerufen hat.
[4] [Object-WH] Karl fragt sich, wen der Doktor ___ [3 PPs]
gerufen hat.
Results from 14 participants suggest that a bihemispheric
network supports the maintenance of the WH-filler in working
memory. Regions of the inferior frontal gyrus (left inferior and
bilateral superior portions of pars opercularis/BA 44) and several
foci along left and right superior temporal sulcus show increased
activation for object WH-questions with a long distance movement
(condition [2] with 3 PPs) ascompared to object WH-questions with a
short filler-gap distance [4].
|
