| |
Abstract:
Although recent psycholinguistic work has found effects of
prosodic phrasing on the resolution of syntactic ambiguities, it
has generally been thought that prosody does not influence
lexical disambiguation, particularly in early stages of
processing. Three experiments explore the effects of two levels
of prosodic phrasing, the intonational phrase (IPh), and
phonological phrase (PPh), on the resolution of within-category
lexical ambiguity, and demonstrate that prosodic phrasing has
early influences on semantic interpretation.
In Experiment 1, strongly-biased polysemous words (glasses)
were presented in semantically neutral sentence-initial clauses
(Although the glasses were ugly...), followed by either an IPh or
PPh boundary. A second clause resolved the ambiguity (...they
held a lot of juice). End-of-sentence "makes-sense" judgment
times did not differ significantly between the two prosodies for
continuations instantiating the dominant meaning of the
polysemous word. However, reanalysis to the subordinate meaning
took longer following IPh than PPh boundaries, suggesting that
more extensive interpretive processing had taken place following
the higher-level boundary.
In Experiment 2, strongly-biased polysemous words (calf) were
presented in clauses weakly biased toward their subordinate
interpretation (Because his calf is injured badly...), followed
by an IPh or PPh boundary. Semantic associates of either the
dominant or subordinate meaning of the polysemous word (COW/LEG)
were visually presented immediately after the prosodic boundary.
Following IPhs, but not PPhs, semantic associates of the
subordinate meaning of the polysemous word were named faster than
dominant-associated targets, demonstrating an immediate effect of
the level of prosodic boundary on semantic interpretation.
IPh boundaries have longer durations than their PPh
counterparts, so the priming found in Experiment 2 may have been
due in part to the relatively longer processing time available
for contexts followed by IPh boundaries. To test this
possibility, Experiment 3 employed materials identical to those
in Experiment 2, except that the naturally-produced IPh/PPh
fragment pairs were digitally altered to match in duration.
Preliminary results replicate those for Experiment 2.
Phonological, phonetic, and RT analyses for each experiment
are presented to argue that the processing effects cannot be
explained solely on the basis of durational differences, but
instead depend on the phonological contrast between IPh and PPh
boundaries. The results demonstrate that semantic processing, and
the use and availability of contextual information, is influenced
by prosodic structure. We claim that the human sentence processor
builds a prosodic representation in early stages of comprehension
and is guided by it at phonological, syntactic, and semantic
levels of interpretation.
|
