| |
Abstract:
Previous research on ambiguous whNP1-NP2-V clauses in Dutch
has shown a subject-object preference (Frazier, 1987; Frazier &
Flores d'Arcais, 1989; among others), which is reduced or even
reversed if the second NP is a pronoun (Kaan, 1997). One
explanation for this modulation by the type of NP2 is frequency:
pronouns are more often used as subject than as object. An
alternative explanation can be given in terms of syntactic
integration difficulty, along the lines of Gibson (1998). According
to Gibson, integration of a trace and a
wh-
phrase is easier when this dependency is intervened by a 1st or 2nd
person pronoun, compared to, e.g., a full NP. Assuming that in
Dutch the object, but not the subject, postulates a trace at the
verb, the dependency between whNP1 and the object trace crosses NP2
in object-subject-V clauses. Processing object-subject clauses
should hence be easier when this NP2 is a 1st or 2nd person pronoun
than when it is a full NP.
The present study was aimed at teasing apart these two
explanations, using the paradigm in (1):
The second NP was either a full NP or
jullie
(you-PL). The word order was disambiguated by the finite auxiliary
(singular: subject-object, plural: object-subject); the lexical
verb appeared only two positions later. Materials were controlled
for semantic reversibility.
Under the integration account, one should expect an effect of
NP2 on the order preference only at the lexical verb, i.e., when
the object trace is integrated with whNP1. Under a frequency-based
account, however, one expects a modulation already at the point of
disambiguation.
Self-paced reading times showed no differences among the
conditions at the disambiguating auxiliary. At the following word,
reading times were longer for object-subject than subject-object
clauses in the NP condition, but showed the reversed pattern for
jullie,
which is in accordance with a frequency-based account. This
interpretation is confirmed by additional data showing that
jullie
is more difficult as an object than as a subject even in
unambiguous clauses.
At positions following the lexical verb, however, reading times
were longer for the object-subject than the subject-object
conditions, but more so in the NP than in the
jullie
conditions, confirming an integration-based account. We will
present additional data showing that the type of NP also affects
the processing of unambiguous dependencies, as is expected in an
integration model.
The present data thus suggest that order preferences are
affected by both frequency and syntactic integration
difficulty.
References
Frazier, L. (1987). Processing syntactic structures: Evidence from
Dutch.
Natural Language and Linguistic Theory,
5, 519-559.
Frazier, L., & Flores d' Arcais, G. (1989). Filler driven
parsing: A study of gap filling in Dutch.
Journal of Memory and Language,
28, 331-344.
Gibson, E. (1998). Linguistic complexity: Locality of syntactic
dependencies.
Cognition,
68, 1-76.
Kaan, E. (1997).
Processing Subject-Object Ambiguities in Dutch.
Ph.D. thesis, University of Groningen.
|
