Syntactic prediction in language comprehension: Evidence from either...or.

Abstract

Readers' eye movements were monitored as they read sentences in which two noun phrases or two independent clauses were connected by the word or (NP-coordination and S-coordination, respectively). The word either could be present or absent earlier in the sentence. When either was present, the material immediately following or was read more quickly, across both sentence types. In addition, there was evidence that readers misanalyzed the S-coordination structure as an NPcoordination structure only when either was absent. The authors interpret the results as indicating that the word either enabled readers to predict the arrival of a coordination structure; this predictive activation facilitated processing of this structure when it ultimately arrived, and in the case of Scoordination sentences, enabled readers to avoid the incorrect NP-coordination analysis. The authors argue that these results support parsing theories according to which the parser can build predictable syntactic structure before encountering the corresponding lexical input.

The return of the repressed: Abandoned parses facilitate syntactic reanalysis☆

“…This account would emphasize the indirect nature of the evidence in favor of the claim that the ultimately correct analysis was, in fact, adopted prior to the main point of ambiguity in the critical conditions. This evidence comes from previous studies (Staub & Clifton, 2006;Frazier et al, 2000) that were not explicitly designed to investigate the parser's analysis at the relevant points (i.e., at the point of reading or in Experiment 1, and at the point of reading the subordinate clause verb in Experiment 2), and from sentence completion studies carried out in association with the present eyetracking experiments which found, first, that when sentence-initial either was present, or is about twoand-a-half times as likely to be treated as a clausal coordinator as when either was absent, and second, that a completion with a subordinate clause-final verb phrase was significantly more likely to consist of a simple intransitive verb when this verb phrase was preceded by verb and.…”

“…This finding could be interpreted as reflecting the mild reanalysis that is predicted to take place on the ambiguous noun phrase in the either no comma condition. As noted above, Staub and Clifton (2006) found that in general an or noun phrase string is read more quickly when either is present than when either is absent. In the present experiment, then, it is possible that the one-versus-three pattern resulted from a combination of this facilitation from either and a reanalysis effect, working in the opposite direction, in the either no comma condition.…”

“…However, there were significant differences in the probability of regressing from this region in different conditions, as discussed below, and as a result first pass time may have been differentially influenced by trials in which the reader regressed after making only one or two fixations (see Altmann, Garnham, & Dennis, 1992;Rayner & Sereno, 1994;Staub & Clifton, 2006). Therefore, first pass time for those trials on which readers did not make a regressive eye movement was also computed.…”

“…In fact, because much of the variability in go-past time is due to first pass variability, effects of experimental manipulations on reanalysis difficulty may be missed even in the go-past measure. In order to focus more closely on the process of reanalysis itself, the mean time spent to the left of the final region after first fixating it, and before leaving it to the right, was computed (see Staub & Clifton, 2006). For ease of discussion, this measure will be referred to as regression path left.…”

“…The first comes from Staub and Clifton (2006), who monitored participants' eye movements as they read sentences like (3a-d): One finding from this experiment was that reading time beginning on the word or was significantly reduced by the presence of the word either before the coordinate structure. This reduction in reading time continued for several words, and did not interact with sentence type; the size of the effect was very similar for sentences in which two clauses were coordinated (3a-b) and in sentences in which two noun phrases were coordinated (3c-d).…”