| |
Abstract:
In subordinate/main clause ambiguities like (1), late closure
(LC) sentences (1a) are preferred over early closure (EC) sentences
(1b). This preference has been the major source of evidence for the
Late Closure principle (Frazier & Rayner, 1982). This ambiguity
has also been used to investigate other issues, such as sensitivity
to verb transitivity and factors that influence reanalysis.
1a) After Janet left the party she ran into some old friends.
1b) After Janet left the party began.
In the studies using this ambiguity, the sentences were
presented without commas between the subordinate and main clauses.
Commas eliminate the garden path effect in these sentences (Hill
& Murray, 1997) and are often inserted to produce unambiguous
controls. The practice of comma omission as a means for testing
parsing strategies rests on two assumptions (Frazier & Rayner,
1982): First, commas at these clause boundaries are optional
stylistic features; second, parsing strategies develop and function
independently of punctuation, thus the parser is in its normal mode
of operation when reading comma-less EC sentences. Juliano and
Tanenhaus (1994) previously examined this structure and found a 90%
rate of comma use in edited text. We explore the possibility that
commas are not optional in this structure in natural written
production and, within a constraint-based framework, they may be an
important source of information that constrains the developing
interpretation. To examine the rate of comma use in EC/LC
sentences, 140 participants completed 12 sentence-initial
fragments, consisting of a subordinating conjunction followed by a
proper name (e.g., "While Lorna..."). The 645 sentences produced
with the relevant structure ("While Lorna VP NP...") were coded for
closure, comma usage, embedded verb transitivity, and case marking
of the NP immediately following the verb.
Late closure sentences were produced more often (60%) than early
closure sentences. Commas were used in 78% of the cases, and in 65%
of the no-comma cases at least one other cue was present. Thus,
participants produced only 50 temporarily ambiguous sentences (8%).
Comma usage did not differ as a function of closure or as a
function of the presence of other disambiguating cues. This
suggests that comma usage is much more consistent than previously
assumed and that comprehenders do not deal with closure ambiguities
on a regular basis (cf. Frazier & Rayner, 1982).
To examine the nature of the comma as a constraint on
processing, we further considered comma use as a function of the
particular subordinating conjunction. Different subordinating
conjunctions yielded different rates of comma usage (range = 67% -
91%). If the parser is sensitive to these frequencies, then they
may predict the amount of difficulty encountered in EC sentences
when the expected comma is absent. We are currently conducting a
self-paced reading study to investigate this possibility, and will
discuss the results with regard to current ambiguity resolution
theories.
References
Frazier, L., & Rayner, K. (1982). Making and correcting errors
during sentence comprehension: Eye movements in the analysis of
structurally ambiguous sentences. Cognitive Psychology, 14(2),
178-210.
Hill, R. L., & Murray, W. S. (1997). Punctuated parsing:
Signposts along the garden-path. Paper presented at the 10th Annual
CUNY Conference on Human Sentence Processing, Santa Monica, CA.
Juliano, C., & Tanenhaus, M. K. (1994). A constraint-based
lexicalist account of the subject/object attachment preference.
Journal of Psycholinguistic Research, 23(6), 459-471.
|
