| |
Abstract:
We propose a novel explanation of preferences in attaching
relative clauses (RCs) within recursive nominals - i.e., noun
phrases with embedded NPs - which yield multiple attachment sites
for an RC (e.g., "the table by the chairs that was/were...").
Building on the ideas of feature percolation in recursive nominals
(e.g., Vigliocco & Nicol, 1998), we attribute crosslinguistic
variation in RC attachment preferences to differential processing
of agreement features.
In subject-verb agreement, "strong" features (Bock &
Eberhard, 1993) may percolate upwards from an embedded noun to the
head noun of a recursive nominal. We formulate this process within
the weighted feature processing mechanism of the competitive
attachment parser, which supports the communication of weighted
agreement features to an NP from its constituents. This feature
integration process increases the activation of the agreement
features of the highest NP of a recursive nominal. This factor
interacts with the decay of phrasal activation (Stevenson, 1994) to
yield the following account of observed modifier attachment
behaviors:
Our approach has several advantages. First, the proposal unifies
research on agreement processing and modifier attachment (cf. the
role of Case in Meng & Bader, 1997; Sauerland & Gibson,
1998). Heretofore the phenomena have largely been treated as
individual processes that both just happen to occur within
recursive nominals. Our approach shifts the focus to the recursive
nominals themselves, and an integrated view of the complex
interpretation process within them.
Second, by appealing to the processing of agreement features, we
ground the crosslinguistic differences in RC attachment to known
variation in syntactic features. Stronger features have been
proposed to influence the processing of agreement in a number of
languages (e.g., Vigliocco et al., 1996; Holscher & Hemforth,
1997). Under our account, it is not accidental that languages with
strong agreement also exhibit a greater tendency for high
attachment of an RC.
Third, the opposing high and low attachment tendencies for
nominal modifiers are integrated within a single parsing mechanism
(as in Gibson et al., 1996; but in contrast to Frazier &
Clifton, 1996; Hemforth et al., 1997). The result is a unified
account of the processing mechanisms at work in both agreement and
attachment, as well as a unified view of the influences on nominal
modifier interpretation.
References
Bock, K., & Eberhard, K. (1993). Meaning, sound and syntax
in English number agreement.
Language and Cognitive Processes,
8, 57-99.
Frazier, L., & Clifton, C. (1996).
Construal.
Cambridge MA: MIT Press.
Gibson, E., Schütze, C., Canseco-Gonzalez, E., & Hickok,
G. (1996). Recency preference in the HSPM.
Cognition,
59, 23-59.
Hemforth, B., Konieczny, L., & Scheepers, C. (1997). Modifier
attachment: Relative clauses and coordinations. Manuscript,
University of Freiburg.
Holscher, C., & Hemforth, B. (1997). Subject-verb agreement in
German: Evidence from production and comprehension. Manuscript,
University of Freiburg.
Meng, M., & Bader, M. (1997). Syntax and morphology in
sentence parsing. Manuscript, University of Jena.
Miyamoto, E. (1998). A low attachment preference in Brazilian
Portuguese relative clauses. Talk presented at AMLaP-98.
Sauerland, U., & Gibson, E. (1998). How to predict the
relative clause attachment preference. Talk presented at the
Eleventh Annual CUNY Conference on Human Sentence Processing, New
Brunswick NJ. BR> Stevenson, S. (1994). Competition and recency
in a hybrid network model of syntactic disambiguation.
Journal of Psycholinguistic Research,
23, 295-322.
Stevenson, S. (1995). Arguments and adjuncts: A surprising
asymmetry.
Proceedings of the Annual Conference of the Cognitive Science
Society,
748-753.
|
