|
Abstract:
First, a series of speeded grammaticality judgement
experiments examining double case ungrammaticalities has shown that
wh-clauses containing two nominative arguments (1) give rise to a
so-called "illusion of grammaticality_, i.e. are erraneously judged
to be grammatical. This is not the case for the analogous
structures containing double accusatives (2), which are generally
correctly judged to be ungrammatical. There is good evidence that
this difference results from the markedness contrast between
nominative (the default case; Bittner & Hale 1996) and
accusative.
(1) *Welcher Botschafter aus der Vorstadt besuchte der Minister?
which-nom ambassador from the suburbs visited the-nom minister (2)
*Welchen Botschafter aus der Vorstadt besuchte den Minister?
which-acc ambassador from the suburbs visited the-acc minister
When the judgement accuracies for the declarative counterparts
of the sentences in (1) and (2) are considered (see (3) and (4)),
an intriguing pattern emerges. While the double accusatives show no
accuracy differences between wh and declarative clauses, double
nominative declarative clauses are systematically judged with a
higher accuracy than double-nominative wh-constructions.
(3) *Der Botschafter aus der Vorstadt besuchte der Minister.
the-nom ambassador from the suburbs visited the-nom minister
(4) *Den Botschafter aus der Vorstadt besuchte den Minister.
the-acc ambassador from the suburbs visited the-acc minister
Thus, the data show a clear contrast between nominative-initial
declaratives and the other clause types examined. Such a finding is
fully consistent with a prominent theory of German clause structure
(initially proposed by Travis (1984)), which considers
subject-initial declarative clauses to be IPs and all other main
clauses to be CPs. Therefore, an initial [-wh] nominative DP will
be integrated into the grammatical structure as the specifier of
IP, where it is then immediately identifiable as a subject by
virtue of its position alone. By contrast, an initial [+wh]
nominative DP will be integrated into [Spec,CP], a position that
may also be occupied by accusative wh-elements and topicalised
constituents. Thus, there is no clear structural information
available to the parser which could identify the initial DP as a
subject. When the parser encounters the second nominative marked
DP, a conflict arises between the information carried by the
incoming element and the structure built up so far. Only in the
case of a declarative clause does the parser have structural
information available to it to confirm its initial
subject-interpretation of the first DP. This leads us to propose
that structural information is more prominent than morphological
information during parsing and thus weightier during diagnosis
processes.
Further evidence that the parser differentiates between
morphological and structural information is provided by a number of
reading-time experiments. The fact that structurally ambiguous
constructions do not elicit higher processing times than their
unambiguous counterparts is well-established within the
psycholinguistic literature. However, this appears not to hold for
morphological ambiguities: German sentences in which the initial
wh-phrase is ambiguous (5a), elicit higher reading times on this
initial phrase than their unambiguous counterparts (5b). We will
argue that this morphological ambiguity with respect to the
grammatical function of the feminine wh-phrase gives rise to
processing difficulties that are independent of any structural
property.
(5) Ich frage mich _
I ask myself ...
a. welcher Mann wahrscheinlich den Botschafter ausraubte
which-nom man probably the-acc ambassador robbed
b. welche Frau wahrscheinlich den Botschafter ausraubte
which-nom/acc woman probabably the-acc ambassador robbed
|
