| |
Abstract:
English simple past tense markers are semantically ambiguous
(Kratzer, 1998). They can be used to mark multiple meanings, which
must be marked by distinct tense morphology in other languages.
(1) "Max stood up. John greeted him."
a. M. stond op. J. groette hem. (simple past, greeting >
standing)
b. " J. had hem gegroet. (past perfect, greeting <
standing)
The English discourse in (1) can be understood as having the
event of the second sentence precede or follow the event of the
first. These two meanings require different tense forms in Dutch,
as shown in the translations in (1a-b). This paper presents two
self-paced reading studies looking at how the processor resolves
this semantic ambiguity. Results indicate that the processor
proceeds in two steps in interpreting a tense morpheme, making use
first of anaphora resolution processes (Partee, 1984) and then
later inferential reasoning regarding discourse coherence
(Lascarides & Asher, 1993).
Study 1 examined the role that discourse coherence relations
play in resolving the tense ambiguity. Subjects read brief
past-tense paragraphs, whose final sentence contained an adverbial
disambiguating the sentence's coherence relation to preceding
context. (See sample item (2).) The adverb either indicated that
the sentence's event *followed* the event of the preceding one
(2a-c) or indicated that it *preceded* it (2b-d). The adverbial
appeared in either preposed (2c-d) or sentence-final (2a-b)
position. Reading times for the segments containing the adverbial
were higher for the Preceding adverbial (2b-d) than for the
Following one (2a-c), F1(1,59) = 6.29,p<.02; F2(1,23) = 7.01,
p<.02. However, there was no interaction between the content and
the position of the adverb (F1,F2<2), indicating that subjects
were not committing early to a final interpretation of the tense
marker and then later being forced to revise it in the
sentence-final conditions. Instead, they waited for discourse
relation information before committing to a full
interpretation.
Study 2 examined the role of anaphora resolution in interpreting
tense by looking at when the processor consults preceding context
to locate a past Reference Time (Partee, 1984) for the current
sentence's tense. Subjects read an unambiguous simple past-tense
sentence preceded by either a past-tense context or an incompatible
future-tense one. The critical sentence contained either a
transitive or an obligatorily distransitive verb, and its coherence
relation to the preceding context was ambiguous until the final
segment. In both the transitive and ditransitive conditions,
subjects read the critical sentence more slowly in the future
context than the past context starting at segment 3, after the
finite verb plus the direct object, F1(1,43)=8.52, p<.01,
F2(1,23)=4.30, p<.05. In line with previous results (Trueswell
& Tanenhaus, 1991), subjects appeared to be consulting context
to find a RefT fairly soon after the tense marker. They did not
appear to delay until the coherence relation involved became clear
as in Study 1. Subjects thus appeared to assign a partial
interpretation to the tense marker early, based on anaphora
resolution and independent of discourse coherence information.
Together, these results argue in favor of views in which both
anaphoric and discourse-coherence processes play a role temporal
relations (Kehler, 1999). The two types of information contribute
to separate semantic processing decisions.
SAMPLE ITEMS
(2) A senior aide in the Senator's press office was accused of
harassment. He denied the claims but resigned anyway.
a. He took | advantage of | someone else | at his next job| on
Capitol Hill.
b. He took | advantage of | someone else | at his previous job| on
C.Hill.
c. At his next job | on Capitol Hill | he took | advantage of |
someone else.
d. At his previous job | on C.Hill | he took | advantage of |
someone else.
|
