Lingering effects of disfluent material on comprehension of garden path sentences

Abstract

In two experiments, we tested for lingering effects of verb replacement disfluencies on the processing of garden path sentences that exhibit the main verb/reduced relative (MV/RR) ambiguity. Participants heard sentences with revisions like The little girl chosen, uh, selected for the role celebrated with her parents and friends. We found that the syntactic ambiguity associated with the reparandum verb involved in the disfluency (here chosen) had an influence on later parsing: Garden path sentences that included such revisions were more likely to be judged grammatical if the reparandum verb was structurally unambiguous. Conversely, ambiguous non-garden path sentences were more likely to be judged ungrammatical if the structurally unambiguous disfluency verb was inconsistent with the final reading. Results support a model of disfluency processing in which the syntactic frame associated with the replacement verb “overlays” the previous verb's structure rather than actively deleting the already-built tree.

This research was supported in part by grants from the National Science Foundation (BCS-9976584, DGE-0114378) and a Michigan State Department of Social Science Dean's Assistantship awarded to Ellen Lau. The authors thank Karl Bailey, Cristina Schmitt, and Jim Zacks for comments on the manuscript.

Notes

¹Note that the results of Bailey and Ferreira (2003) imply that such a parser's performance would nevertheless be improved further if it did make use of disfluency information.

²The reader may have noticed that the MV/RR garden path sentence presented here does not give rise to the same level of conscious difficulty as the famous example of the same type, The horse raced past the barn fell. It has been suggested that the relative ease of garden paths like The little girl selected for the role celebrated with her parents results at least partly from the inability of the PP for the role to appear after the verb in a MV structure, leading to earlier disambiguation, as in *The little girl selected for the role. vs. The horse raced past the barn (Gibson 1991; Pritchett 1988). But while the garden path is milder in these cases, such sentences have still been found to be measurably more difficult than their unambiguous counterparts; this will be demonstrated again in the following experiments as a baseline measure.

³By semantic here, we mean the “dictionary” meaning associated with the wordform and not the associated argument structure requirements. In the cases we will focus on, both the argument structure and the basic meaning of the substituting verbs are fairly similar; thus we make no claims about the possibility that this kind of semantic information from the disfluency lingers as well, although it certainly seems likely.

⁴In both experiments we inadvertently included a verb (raced) which can appear not only in the transitive MV and RR constructions discussed here, but can also appear intransitively. This may have resulted in a later disambiguation point than other items. Although it is unclear that our main predictions in this case would differ (disambiguation would presumably still happen at the verb), all reported effects remained significant when this item was excluded from analyses.

⁵In general, the PPs following the first verb would preclude an MV reading (*The little girl selected for the role Maria vs. The little girl selected Maria for the role). However, in English, it is possible for a PP to precede a direct object if the direct object is fairly long or “heavy” through a process known as “heavy NP-shift” (The little girl selected for the role her best friend Maria who had lived next door to her since kindergarten). Because such constructions are at least somewhat marked, we still believe that disambiguation generally happened at the PP in our material (see MacDonald, 1994). At the same time, we should note that our predictions do not at all depend on whether disambiguation takes place at the PP or at the second verb; in either case, the disfluency takes before the disambiguating material.

⁶A reviewer suggests that the main effect of disfluency may be due to the lack of obvious message-level motivation for the revisions in our disfluent materials (e.g., no striking difference in meaning between chosen and selected), leading them to the perceived as less natural.

⁷TAGs incorporate two fundamental operations, substitution and adjunction. At this point, we limit our hypothesised Overlay operation to substitution cases, although the implications of Overlay for adjunction is certainly an interesting issue which we plan to consider in the future.

⁸It is possible that, at least in some cases, the signal to the parser to try to initiate Overlay is elicited by the failure of the incoming word to match a structural prediction (as in prediction of a preposition in The little girl chosen …) rather than simple failure to find a substitution site.

Additional information

Notes on contributors

Ellen F. Lau

The first author is now at the Department of Linguistics, University of Maryland, College Park, MD. USA