|
Abstract:
In some structures in German involving a filler gap ambiguity
one can observe a contrast in the visibility of a reanalysis
effect that is related to the type of the disambiguating item, as
exemplified in (1).
(1) a. welche Frau sah der Mann which woman saw-sg the-nm
man
b. welche Frau sah den Mann which woman saw-sg the-acc man
c. welche Frau sahen die Manner which woman saw-pl the-nm
men-pl
The experimental data reported by Meng (1997) and Schlesewsky
et al. (1997) show an advantage for a SOV order if disambiguation
is effected by number agreement, as exemplified in (1c). But in
most constructions no effect is observed if Case is the
disambiguating factor, (1a/b). To further address this issue, we
present the results of an ERP-study and of a sentence repetition
task both of which confirm the absence of a Case dependent
reanalysis effect as in (1a/b). The absence of a reanalysis
effect in the case condition could be a consequence of the
costlessness of this type of ambiguity resolution or a result of
an erroneous misinterpretation of the second NP. We discuss
experiments which address the question and support the latter.
This view, the "attention-driven"strategy, is motivated by the
assumption that subjects decide about the validity of an initial
preference on the base of any information that they receive
during the incremental analysis regardless of whether or not this
information is relevant for the filler gap dependency. This
strategy is consistent with a human parser that tries to solve
ambiguities as soon as possible. For example, in (1a/b) there is
a subject preference in the parse. On encountering the verb there
is no information that contradicts the preference. Therefore the
number agreement on the verb is taken as evidence in support of
the subject preference analysis. Since there is apparent number
support for the preferred parse, the Case information on the
second NP is not attended to. In order to vary the visibility of
case information on the final phrase we contrasted sentences
where the second argument was realized as a pronoun instead of a
full DP. In this condition we found a higher reaction time on the
nominatively marked pronoun. This finding is consistent with the
so-called "attention-driven" strategy above.
Taken together, these findings shed a interesting light on the
discussion of the nature of reanalysis.
Meng, Michael (1998) Grammatik und Sprachverarbeitung:
Psycholinguistische Untersuchungen zur Berechnung syntaktischer
Strukturen. Unpublished manuscript, University of Jena.
Schlesewsky, M., Fanselow, G. Kliegl, R. and Krems, J. (to
appear). Preferences for grammatical functions in the processing
of locally ambiguous wh questions in German. In: B.Hemforth L.
Konieczny Cognitive Parsing of German. Dortrecht: Kluwer.
|
