|
Abstract:
The recent discussion about the nature of reanalysis
processes raised questions about the influence of syntactic
features. The resolution of ambiguous wh-sentences in German is an
especially illuminating example. In this language, we are able to
solve an ambiguity induced by a Case ambiguous
wh-
phrase either via number agreement with the sentence internal verb
(1c) or via Case marking of a second DP (1a/b).
In a number of experiments one can find a reliable reanalysis
effect, if the initial assignment of a case feature is rejected on
the basis of number information. On the other hand, all studies
show a non-homogeneous pattern with respect to reanalysis if the
ambiguity is solved via the second DP (Meng, 1998; Schlesewsky et
al., 1998).
There are different explanations for the non-existence of
visible reanalysis effects on the second DP (diagnosis model by
Fodor & Inoue, mismatch principle by Bader & Meng). All of
those theories have one thing in common: the assumption that there
is a reanalysis effect at the second DP but that this effect is
weak, and therefore hard to detect. In this paper we will challenge
this assumption.
In the first part of this paper we will present a number of
studies that show a clearly detectable reanalysis effect induced by
Case mismatch on the second DP. On the basis of these data we argue
that the undetectability of a reanalysis effect in sentences like
(1a) is a consequence of the linear order of the verb and its
arguments. In the second part we will present some grammaticality
judgment data. In contrast to the sentences in (1), we used
ungrammatical sentences and extended the design by replacing the
full second DP by a personal pronoun (2).
The data show that subjects judge double nominative
ungrammaticalities at chance level if the second argument is
realized as a full DP. In double accusative constructions or in
double nominative ones that include a pronoun in the second
argument position the accuracy of correct responses is higher at 80
%.
To explain this pattern, we concluded that the undetectability
of reanalysis in sentences like (1b) is a consequence of the formal
feature of the second DP and of the linear order (distance) between
both arguments and the verb. We assumed that the data support the
idea of a so-called "expectation-driven" non-syntactic influence in
human parsing.
|
