|
Abstract:
Mauner and Koenig (1998) argue that readers rapidly access
agent information when it is lexically encoded in a passive verb,
as in (1a), but not when it is merely logically entailed of a
situation described by an intransitive verb, as in (1b). The only
significant differences in their eye-monitoring study were in total
reading times: intransitive sentences took longer to read than
short passives at both verb and post-verb regions.
Selective rereading of intransitive but not passive verb regions
(40% vs. 4% regressions back to verb regions, respectively)
suggests that argument information was processed during first pass
reading. But, longer intransitive reading times could have been due
to the markedness of either the intransitive "middle" or auxiliary
verbs, or to difficulty in accessing the argument structures of
rarer "middle" verbs relative to passive verbs. Subsidiary analyses
of these data and a new eye-monitoring study were conducted to
address these concerns.
Reanalysis.
The null difference in First Pass (FP) reading in the post-verb
region is surprising. We examined whether spillover effects from
the verb region were depressed in the intransitive condition
because intransitives elicited more regressive eye-movements.
Participants who produce regressive eye-movements out of the region
have shorter FP times than participants who do not. Thus the
greater proportion of regressions in intransitives could have
masked spillover effects. When we excluded FP times that led to a
regression, intransitives (M = 805 ms, SE = 58) elicited longer FP
times than passives (M = 700 ms, SE = 36). This suggests that
argument structure information was processed in FP reading.
Study.
We compared first pass and total reading times to short passive
and intransitive sentences following rationale clauses as in (1a)
and (1b) and subordinate control clauses which do not require an
agent for interpretation, as shown in (2a) and (2b). Intransitives
following rationale clauses elicited longer first pass and total
reading times than passives in the post-verb region and longer
total times in the verb region. No differences in first pass or
total reading times emerged for control sentences.
Our reanalysis provides further support for argument structure
information becoming available for interpretation as soon as a verb
is recognized. The replication of longer times for intransitive
sentences only when preceded by expressions whose interpretation
requires an agent demonstrates that longer reading times to
intransitive sentences in our first study cannot be attributed to
oddities associated with intransitive materials and their argument
structure information.
References
Mauner, G., & Koenig, J.-P. (1998). Lexical-semantic vs.
conceptual sources of implicit argument information. Talk presented
at the 11th Annual CUNY Conference on Human Sentence Processing,
March 19-21, New Brunswick, NJ. BR> Matin, E., Shao, K. C.,
Boff, K. R. (1993). Saccadic overhead: Information-processing time
with and without saccades.
Perception & Psychophysics,
53(4), 372-380.
|
