|
Abstract:
Previous accounts of preposing (a.k.a. 'topicalization') have
argued that its felicitous use requires that the preposed element
constitute a LINK between the current utterance and the prior
discourse (Vallduvi 1992, Birner and Ward forthcoming; see also
Prince 1981). One means of satisfying this requirement is for the
link to stand in a partially-ordered set, or POSET, relation
within the current discourse segment (Ward 1988), as illustrated
in 1:
1a. Baseball's ok. Basketball I like a lot better. b.
Baseball's ok. I like basketball a lot better.
Here, the link (basketball) and trigger (baseball) in (1a) are
related to the inferred poset (sports) via a type/subtype
relation. Preposing, unlike canonical word order (CWO),
explicitly and conventionally marks the relevant constituent
(basketball in (1)) as a poset-related link, thus serving as a
pragmatic cue to sentence processing.
From Birner and Ward's account, it follows that even poset
members implicitly related to the link (e.g. sports in 1a) should
become more activated in memory with preposed links than with CWO
links. To test this hypothesis, a recognition experiment was
performed in which participants were shown sets of dialogs, as in
2:
2. A: I don't mind all the ants in the basement. B: Yeah, but
the cockroaches I could definitely do without.
Dialogs were presented on a computer screen, with half of B's
utterances in CWO and half in preposed word order. After a block
of dialogs, participants were presented with either the link
(cockroaches in (2)), the trigger (ants), the poset (insects), or
some unrelated word (chair). Participants were asked if the
presented word was one used in the dialogs, or not. The critical
trials were those in which the word represented the poset, which
participants should have identified as 'new'. The response time
to such items is a function of how active they were in memory, as
words that are highly activated should be slower to be identified
as new (McKoon and Ratcliff 1986). Our results support Birner and
Ward's account: Response times to posets inferred from preposed
links were significantly longer (1330 ms) than response times to
posets inferred from CWO (1221 ms); t(10) = 3.2158, p < .01.
These results provide evidence for the crucial role of the link
in the processing of preposing constructions. We are currently
running a series of follow-up experiments designed to show that
this effect is specific to the link of the preposing, and not a
sentence-level effect of preposing in general.
Birner, B., and Ward, G. (1998).
Information Status and Noncanonical Word Order in
English.
Amsterdam/Philadelphia: John Benjamins. In press.
McKoon, G., and Ratcliff, R. (1986). "Inferences about
predictable events."
Journal of Experimental Psychology: Learning, Memory, and
Cognition
12, 82-91.
Prince, E. F. (1981). "Topicalization, Focus-Movement, and
Yiddish-Movement: A pragmatic differentiation."
Berkeley Linguistics Society
7, 249-264.
Vallduvi, E. (1992).
The Informational Component.
New York: Garland.
Ward, G. (1988).
The Semantics and Pragmatics of Preposing.
New York: Garland.
|
