| |
Abstract:
Aspectual coercion has been proposed in the linguistic
literature (e.g. Moens and Steedman (1988)) as a semantic operation
which resolves a mismatch between the aspectual properties of the
verbal predicate, on one hand, and the aspectual requirements of a
(temporal) verbal modifier, on the other. In general, aspectual
readings are either open-ended (atelic) or bounded (telic). For
example, the only possible reading of "John threw a stone at the
scarecrow for three hours" is as an (atelic) repetitive event that
lasted 3 hours; this interpretation is coerced because the
predicate 'throw a stone' is naturally telic, but durative
modifiers of the form 'for X time' coerce atelic event readings.
Coercion is not necessary in the similarly repetitive counterpart
"John threw stones at the scarecrow for three hours," because the
predicate 'throw stones,' containing a bare plural object, is
already atelic.
Pinango, Zurif, and Jackendoff (1999) report an effect of
coercion, conceptualized as increase in processing load, in
sentence pairs contrasting temporally modified telic and atelic
predicates. However, Pinango et al.'s test sentences simultaneously
differ in whether they instantiate a single or a repetitive event
reading. Their results can alternately be interpreted to indicate
that repetitive readings impose a higher processing burden than
non-repetitive readings, leaving open the issue of whether the
operation of coercion has any observable processing
consequences.
Our study investigates this question for one aspectual class of
verbs, traditionally termed 'achievements'. This class of verbs was
chosen because the lexical aspect of its members can be
unambiguously telic. The verbs were combined with either an
indefinite singular or a bare plural direct object (resulting in a
telic or repetitive atelic reading of the predicate, respectively),
and further modified with a durative or an aspectually neutral
adverbial. This resulted in a 2 X 2 factorial design with both
cardinality of object and property of modifier as variables:
1. Even though Howard sent a large check to his daughter for
years, she refused to accept his money.
2. Even though Howard sent large checks to his daughter for years,
she refused to accept his money.
3. Even though Howard sent a large check to his daughter last
year, she refused to accept his money.
4. Even though Howard sent large checks to his daughter last year,
she refused to accept his money.
Sentences (1) and (2) both result in a repetitive
interpretation; however, in (1) this interpretation is due to
coercion forced by the modifier, whereas no coercion is needed in
(2). If coercion increases processing load, the reading time for
the critical regions - the region introducing the modifier and
subsequent material - should increase in the coercion condition
(singular object) relative to the non-coerced condition (plural
object). Sentences (3) and (4), on the other hand, contain a
neutral modifier in the critical region. Therefore, we expect no
difference between these two conditions in reading time for the
critical region.
We use a phrase-by-phrase self-paced stop-making sense reading
task to measure both reading time and number of rejections for the
critical adverbial phrase region . As expected, subjects are
significantly delayed when reading the critical region and
subsequent material in the coerced repetitive condition (1)
relative to the non-coerced repetitive condition (2). In addition,
subjects reject the critical region as nonsensical more frequently
in condition (1) than in condition (2); however, the majority of
subjects who did so judged the overall sentence as sensible,
suggesting that they resolved the aspectual ambiguity later on. No
such differences were observed for the critical regions in
conditions (3) vs. (4). We conclude that the observed delay is a
processing correlate to aspectual coercion.
|
