| |
Abstract:
A recurrent theme within the parsing literature is whether
prosody influences initial parsing decisions, as consistent with
Constraint-based models, or is limited to syntactic revisions, as
in some versions of the Garden Path model. Empirical results have
been mixed (Beach, 1991; Marslen-Wilson et. al., 1992; Nagel et.
al., 1994; Watt & Murray, 1996). In answering this question,
however, three issues must be addressed. What is the nature of the
prosodic information that might differentiate a temporary syntactic
ambiguity? How consistently does this information occur in
production? Are listeners sensitive to that information? These
issues were investigated using the temporary syntactic ambiguity
inherent at the conjunction in the following types of sentences:
(1a) Mary both sold vegetables and bought seeds.
(1b) Mary both sold vegetables and seeds.
Earlier experiments showed that both can take scope over
conjoined VPs, as well as over conjoined phrases within the VP, and
recent norming data demonstrate that this is true not only at The
Ohio State University, but also at Rutgers. This is reminiscent of
the global ambiguity of the scope of only in (2), that Rooth (1992)
and others claim to be regularly disambiguated prosodically, by
pitch accent placement.
(2) Karl only eats herring for breakfast
Five native speakers of English, each aware of the structural
contrast between (1a) and (1b), and aware that the experimenter was
interested in the way speakers produce them, were recorded
producing target utterances in response to brief contexts followed
by the question, "Did you know that?" The utterances were ToBI
(Tone and Break Indices) transcribed to analyze the location and
type of pitch accents, phrase accents, and boundary tones. Although
one speaker consistently produced two distinct tunes prior to the
disambiguating region by manipulating accentuation and phrasing,
the other four speakers produced a variety. The results suggest
that even when speakers are aware that the prosody of coordinate
structures is at issue, most fail to provide consistent
disambiguating cues.
Naïve listeners were then asked to identify whether
fragments, ending at the offset of I>and, started a conjoined NP
or conjoined VP sentence. Listeners were able to use the cues
provided by the one speaker who consistently manipulated
accentuation and phrasing: 62% of the conjoined NP fragments and
60% of the conjoined VP fragments were identified correctly. For
three of the other four speakers, listeners showed an overall bias
to identify fragments as conjoined NPs, a result compatible with
sentence completion data collected earlier.
The combined results of these experiments suggest that although
speakers often fail to provide reliable cues, listeners'
interpretations are constrained by reliable information, when
available.
References
Beach, C. M. (1991). The interpretation of prosodic patterns at
points of syntactic structure ambiguity: Evidence for cue trading
relations.
Journal of Memory and Language,
30, 644-663.
Marslen-Wilson, W. D., Tyler, L. K., Warren, P., Grenier, P.,
& Lee, C. S. (1992). Prosodic effects in minimal attachment.
The Quarterly Journal of Experimental Psychology,
45A, 73-87.
Nagel, H. N., Shapiro, L. P., & Nawy, R. (1994). Prosody and
the processing of filler-gap sentences. Journal of Psycholinguistic
Research, 23 (6), 473-485.
Rooth, M. (1992). A theory of focus interpretation.
Natural Language Semantics,
1, 75-116.
Watt, S. M., & Murray, W. S. (1996). Prosodic form and parsing
commitments.
Journal of Psycholinguistic Research,
25 (2).
|
