|
Abstract:
A recent ERP study by Coulson, King & Kutas (1998) on the
processing of double nominative ungrammaticalities in English
showed that, in comparison to correct sentences, sentences such as
(1) elicit a biphasic LAN-P600 pattern on the pronominal NP
inducing the ungrammaticality.
(1) The plane-nom took we-nom to paradise
In our talk, we will present the results of an ERP study
examining the ERP patterns elicited by double nominative (2a) and
double accusative (2b) ungrammaticalities in German.
(2a) Welcher Dichter besuchte der Gärtner
which-nom poet visited the-nom gardener
(2b) Welchen Dichter besuchte den Gärtner
which-acc poet visited the-acc gardener
The language-specific differences between English and German led
us to predict that the ERP-pattern for German should differ from
the one found by Coulson et al..
English has a rigid word order in which linear positions
determine the grammatical function of the NP arguments, thus making
interpretation possible despite incorrect case marking on an NP, as
in (1). By contrast, in a language with a free word order such as
German, grammatical functions are determined by overt
(morphologically realized) case information on the NPs rather than
by their linear positions. As a consequence, the assignment of
grammatical functions and therefore thematic interpretation should
be impossible in a sentence containing two identically marked
NPs.
Given that the N400 is a marker of uninterpretability while the
LAN is not, the biphasic N400-P600 pattern that we found in
ungrammatical sentences such as (2) reflects this inability of the
parser to assign a specific thematic role to the sentence
arguments. In addition, we found that the N400 but not the P600 in
the double nominative condition (2a) was delayed and shorter in
duration in comparison to the double accusative ERP-pattern. This
may be due to the fact that double nominatives are initially easier
to interpret.
The difference in interpretability between double nominative and
double accusative ungrammaticalities may be taken in support of the
view that the nominative has the status of a default case (see e.g.
Bittner & Hale 1996). Thus, the second argument position made
available by a transitive verb may license an attachment site for
an unmarked (nominative) object. Such an interpretation of our
results is confirmed by a second experiment using NP-NP-V
structures in which no differences were found between the
ungrammatical conditions.
The results will be discussed with respect to the role of case
information in reanalysis processes (Fodor & Inoue 1999).
|
