| |
Abstract:
This poster provides evidence for the Structural Agreement
Check (SAC), a detailed statement of the agreement checking process
in comprehension. The SAC presupposes that features are checked on
structure which has been built using structural parsing principles
and majorcategory information (the "Structure Evaluation"
hypothesis) (Deevy, 1999). Previous work on agreement comprehension
has focused on the "local plural" effect in which the presence of a
plural in a complex subject DP disrupts comprehension of
(grammatical) singular agreement (in, e.g., the key to the cabinets
was ...). Explanations assume that the singular nominal feature is
underspecified (e.g., Pearlmutter, Garnsey, & Bock, 1995) and
that the marked plural feature can migrate within the subject DP
(Nicol, Forster, & Veres, 1997). Thus, when agreement is
checked, it may appear that the plural feature is associated with
the head noun.
The SAC builds on this work in adopting the assumptions of feature
underspecification and movement. It goes further by stating that
checking must be satisfied immediately once the verb is identified
as bearing agreement features, and that agreement is enforced
through checking the specification (or underspecification) of the
links of the tree between subject and agreeing verb (Clifton,
Frazier, & Deevy, 1999). The current study shows the need for
both additional assumptions. A word-by-word self-paced reading
study tested the processing of subject Relative Clauses (RCs) in
which the verb must agree with one of two NPs in a complex DP. The
auxiliary agreed exclusively with NP1 or with NP2 (1a-d), or it
agreed with either NP (1e,f). It was assumed that the structural
relation between RC and head may be underspecified (Frazier &
Clifton, 1996); thus, at the point when agreement must be
checked,the structure underlying the check may not be built.
1. John was excited to meet...
a. the nieceNP1 of the actorsNP2 [who was recently starring in a
very successful play].
b. the niecesNP1 of the actorNP2 [who was recently starring in...
c. the niecesNP1 of the actorNP2 [who were recently starring in...
d. the nieceNP1 of the actorsNP2 [who were recently starring in...
e. the nieceNP1 of the actorNP2 [who was recently starring in...
f. the niecesNP1 of the actorsNP2 [who were recently starring in
...
Results show that reading slowed significantly in the plural
NP2-singular verb condition (1a vs. 1b). This difficulty cannot
receive the sameexplanation as the local plural effect; this would
require NP1 attachment with accidental migration upward of the
plural feature. However, systematic high attachment predicts
checking difficulty in (1b) and (1d) relative to (1c), which was
not observed. An account in terms of a systematic NP2 attachment
and check is also ruled out because the plural conditions (1c, 1d)
did not differ. It will be shown that agreement information may
guide RC attachment and that the pattern is best accounted for in
terms of feature markedness, immediacy of checking and enforcement
of the check through feature passing. In addition, it will be
suggested that apparent accidental feature migration can be
explained in terms of the immediacy requirement and systematic
feature passing.
References
Clifton, Jr., C., Frazier, L., & Deevy, P. (1999). Feature
manipulation in sentence comprehension. In E. DiDomenico & M.
DeVicenzi (Eds.),
Rivisti di Linguistica.
Deevy, P. (1999). I>The Comprehension of English Subject-Verb
Agreement. Unpublished doctoral dissertation, University of
Massachusetts, Amherst.
Frazier, L., & Clifton, Jr., C. (1996).
Construal.
Cambridge, MA: MIT Press.
Nicol, J. K., Forster, K. I., & Veres, C. (1997). Subject-verb
agreement processes in sentence comprehension.
Journal of Memory and Language,
36, 56-587.
|
