Semantic and auditory factors in processing syntactic ambiguity

1
Semantic and auditory factors in processing
syntactic ambiguity Filip Loncke, Beverly
Colwell Adams, Kimberly Edwards, Laura Leviski,
Janet Stack, Ji eun Kim, and Jennifer Starkey
Communication Disorders Program - University of
Virginia
What factors influence our interpretation of
syntactically ambiguous sentences?
Syntactic Ambiguity
Results
Sentence Type There was a main effect of
sentence type F (2, 1403) plt .01 Consistent
with Loncke et al. (2007), the present results
show a preference for the low attachment strategy
for unambiguous and abstract sentences. The
high attachment strategy and the low attachment
strategy were split fairly evenly (92/110)
between the syntactically ambiguous/semantically
ambiguous sentences, such as (1). (See Figure
1) Pauses The pattern of results for pauses, P0,
P1, and P2, for both unambiguous sentences and
abstract ambiguous sentences, showed a commanding
preference for the low attachment interpretation
(See Figures 2 and 3). Examining only the
syntactically ambiguous sentences, in the P0 and
P2 conditions, there was a clear preference for
high attachment. Most interestingly, however,
for the P1 pause position condition, there was a
dramatic increase in the number of sentences that
were interpreted with the low attachment
strategy almost equaling the high attachment
preference t(38) 5.085, plt.001 (see Figure 4).
Consider the syntactically ambiguous sentence
(1). (1) The apple is in the bowl next to the
banana. The sentence processing strategy, high
attachment, guides sentence processing to the
interpretation that the apple and the banana are
in the bowl. In contrast, the low attachment
strategy steers toward the interpretation that
the banana is located next to the bowl. This
type of syntactically ambiguous sentences such as
(1) is also semantically ambiguous. Now,
consider syntactically ambiguous sentences that
are semantically unambiguous like (2). (2) The
apple is in the bowl next to the sink. Only one
semantically plausible interpretation is produced
by the low attachment strategy the apple is
located in the bowl and the bowl is next to the
sink.
Which interpretation will the listener give to a
syntactically ambiguous sentence? If the
syntactic structure is the sole or dominant force
in determining the interpretation, a preference
for a high syntactic resolution can be
expected. If other factors interfere (such as
acoustics, semantics, and/or pragmatics), other
responses can be expected along with high
syntactic resolution.
Hypotheses

• We examined the following hypotheses
• Interpretation of syntactically ambiguous
  sentences will be made using the high attachment
  strategy.
• Strategically placed pauses will shift the
  preferred high attachment strategy to low.

Figure 2
Figure 1
Previous studies
Although not without controversy (Loncke, Adams,
Stack, et al., 20071), previous research has
consistently reported that the high attachment
processing strategy is made initially and is
indeed the preferred strategy for interpretation.
Loncke, et. al.s results suggested that the
final interpretation of even syntactically
ambiguous sentences may be influenced by
semantics.
Method
Figure 4
Figure 3
Participants Thirty-nine participants, all
undergraduate students of the University of
Virginia, participated in this study.
Materials Thirty-six target sentences were
developed with consideration to the independent
variable, sentence type. One third (12) of the
sentences were syntactically ambiguous and
semantically ambiguous such as (1). Twelve
sentences were syntactically ambiguous, but
semantically unambiguous such as (2). Twelve
sentences were syntactically and semantically
ambiguous using the names of geometric shapes in
the noun positions such as (3). (3) The circle
is in the rectangle next to the triangle. The
second independent variable was pauses. In the
P0 condition, no pauses were added to the
auditorally-presented computer-generated
sentences. In P1, an exaggerated pause was
inserted after the first preposition The apple
is in // the bowl next to the banana (sink). In
P2, the pause was inserted after the first
prepositional phrase The apple is in the bowl //
next to the banana (sink). Procedure Each
participant, tested individually, listened to
each target sentence, one-at-a-time. After
listening, participants were asked to make a
sketch of their interpretation of what they had
heard.
The results of the 2007 study indicate that
responses differ , depending on the conditions.
The use of abstract sentences such as the
square is on top of the rectangle next to the
triangle helps us to understand the relation
between semantics and syntax.
Discussion
In the psycholinguistic literature, syntactically
ambiguous sentences are examined to investigate
the initial parsing strategy that guides the
attachment of constituent phrases. The
assumption has been that the most efficient
strategy (high attachment) guides the initial
attachment and unless semantics or pragmatics
suggests a different interpretation, the initial
strategy was the final strategy. In the present
study, we suggest that the final interpretation
of pre-tested syntactically ambiguous/
semantically ambiguous sentences, maintain some
level of subtle semantic/pragmatic influence.
With sentences such as (3) that incorporated only
geometric shapes in the noun positions, we report
that participants made their final interpretation
using the low attachment strategynot high. The
results suggest, also, that the
non-syntactic/non-semantic factor, time,
strategically placed, can alter interpretation.
Our focus in the current study was to determine
if pauses in spoken text can change the preferred
interpretation of syntactically ambiguous
sentences.
1Loncke, F., Adams, B., Dodson, V., Cao, L.,
Stack, J., Kim, J., Craig, J. (2007, November
17). Clinical relevance of auditory perception of
syntactic ambiguity. Poster presented at the
Annual Convention of the American Speech-Language
and Hearing Association (ASHA). Boston.
Acknowledgment
Conduction of these studies were made possible
through the participation of several groups of
students from the University of Virginia
throughout the years 2005-2008. Special thanks to
Valerie Dodson, Amanda Spear and Alyssa Weltman
who were instrumental in a pre-pilot study and a
pilot study.
Contact Filip Loncke, ftl4n_at_virginia.edu or
Beverly C. Adams, bca5y_at_virginia.edu