|
Abstract:
Spontaneous speech disfluencies occur frequently, but little
is known about how they are processed by hearers. In false starts
like (FS), an utterance is aborted then restarted:
(FS) ... There was once this king .. and [he was chasing this
animal .. very ..] he was on this hunt ...
False starts are important because, as in garden path
sentences, an initial determination of sentence structure is
revised on line. Using speech stimuli and measuring word
monitoring latencies, Fox Tree (1995) found evidence of an
increased processing load during some FS repairs.
The present study measures event-related brain potentials
(ERPs) time-locked to the onset of the repair. Recent ERP studies
have found increased positive deflections about 600 ms after the
disambiguating word in garden path sentences. If this P600
reflects processes involved in the revision of an initial
syntactic analysis, then the repair of a false start should
elicit a similar effect.
For stimuli, the speech of people retelling fairytales from
memory was digitally recorded and 40 false start utterances
identified. Fluent controls were constructed by digitally
excising the false start,
(CTRL) There was once this king .. and he was on this hunt
...
The target beginning with "he" occurs as fluent continuation
in (CTRL) and as a repair in (FS). Thirty additional pairs of
fillers, e.g., "Mary got up and walked her jet/dog" were read and
recorded. The materials were split into two balanced lists,
pseudo-randomized, and presented over headphones, with each
sentence followed by a forced-choice "makes sense?" question.
Preliminary analysis (12 subjects) found the expected sentence
final auditory N400 for the fillers. Analysis of the (FS) repair
found that the target elicited a positive deflection relative to
the fluent control starting about 400 ms poststimulus and most
pronounced at the midline sites.
These results suggest that false start repairs are processed
differently from fluent utterances within 400 ms of the onset of
the repair. Although neither parietally maximal nor bilateral,
the false start ERP effect appears to have the polarity and
latency of a typical P600, which would be expected if the
comprehension processes involved in repairing a false start are
similar to those involved in recovering from the misanalysis of a
local ambiguity.
Fox Tree, J. E. (1995). The effects of false starts and
repetitions on the processing of subsequent words in spontaneous
speech.
Journal of Memory and Language,
34, 709--738.
|
