A Multiple‐Channel Model of Task‐Dependent Ambiguity Resolution in Sentence Comprehension

Abstract

Traxler, Pickering, and Clifton (1998) found that ambiguous sentences are read faster than their unambiguous counterparts. This so-called ambiguity advantage has presented a major challenge to classical theories of human sentence comprehension (parsing) because its most prominent explanation, in the form of the unrestricted race model (URM), assumes that parsing is non-deterministic. Recently, Swets, Desmet, Clifton, and Ferreira (2008) have challenged the URM. They argue that readers strategically underspecify the representation of ambiguous sentences to save time, unless disambiguation is required by task demands. When disambiguation is required, however, readers assign sentences full structure--and Swets et al. provide experimental evidence to this end. On the basis of their findings, they argue against the URM and in favor of a model of task-dependent sentence comprehension. We show through simulations that the Swets et al. data do not constitute evidence for task-dependent parsing because they can be explained by the URM. However, we provide decisive evidence from a German self-paced reading study consistent with Swets et al.'s general claim about task-dependent parsing. Specifically, we show that under certain conditions, ambiguous sentences can be read more slowly than their unambiguous counterparts, suggesting that the parser may create several parses, when required. Finally, we present the first quantitative model of task-driven disambiguation that subsumes the URM, and we show that it can explain both Swets et al.'s results and our findings.

A Computational Evaluation of Sentence Processing Deficits in Aphasia

“…For example, in recent work Caplan, Michaud, and Hufford (2013) have investigated 61 persons with aphasia on different syntactic structures and different tasks (sentence-picture matching, sentence-picture matching with auditory moving window presentation, and object manipulation). Based on this large dataset, they concluded that the underlying deficit may be a reduction in processing resources; they also found an interaction between task demands and parsing difficulty (a topic that is of great interest in unimpaired sentence processing; see Loga cev & Vasishth, 2015;P. Loga cev and S. Vasishth (2014), unpublished data.…”

Models of retrieval in sentence comprehension: A computational evaluation using Bayesian hierarchical modeling

“…Furthermore, the activation-based model may be able to account for Logačev and Vasishth's (2016) observation that the parser seems to behave in a way that resembles a race between interpretations (low attachment vs. high attachment) but it is also taskdependent (as assumed by Swets et al, 2008). This could be achieved by setting the timeout (the parameters of the accumulator associated with the retrieval failure) to be task-dependent: longer timeouts when instructions or context encourage attentive reading and shorter timeouts when a full interpretation is not needed for successfully completing the experimental task.…”

“…In contrast, for the activation based-model with different variances, the relationship between reading times at the different conditions would depend on comprehension accuracy. Future work that includes measures of reading times and queries for the comprehension of the relative clause, as well as manipulates task demands could compare the activation-based model with different variances, the direct-access model, and the model presented in Logačev and Vasishth (2016), which subsumes the unrestricted race model and allows it to be task-dependent.…”

How do violations of Gricean maxims affect reading?

“…Similarly, research has shown that syntactic ambiguity does not necessarily slow down comprehension (e.g., Logačev & Vasishth, 2016;Swets, Desmet, Clifton, & Ferreira, 2008;Traxler, Pickering, & Clifton, 1998;Van Gompel, Pickering & Traxler, 2000).…”