Braze, D., Mencl, W.E., Shankweiler, D.P., Tabor, W. & Schultz, A. (2006, July 6). Skill-Related Differences in the Online Reading Behavior of Young Adults: Evidence from Eye-Movements. Talk given at the Thirteenth Annual Meeting of the Society for the Scientific Study of Reading. Vancouver, Canada. [abstract] [PDF]

Abstract: This study focuses on the apprehension of combinatorial meaning of sentences. We recorded eye-movements (EM) during silent reading of temporarily ambiguous sentences to examine individual differences in gaze patterns as a function of those capacities that support reading comprehension. We collected measures of reading and listening comprehension, decoding skill, verbal memory, vocabulary, and experience with print. Participants were 46 young-adults, 16 to 24 y.o., primarily adult-school and community college students representing a wide range of reading skill. Results indicate that reader skill is positively correlated with sensitivity to the local, moment-to-moment, demands of text comprehension, as evidenced through EM measures.

Participants silently read individual sentences from a computer screen while their eye-movements were recorded. Experimental materials included temporarily ambiguous sentences where the true meaning could only be determined subsequent to the point of ambiguity. A portion of these were 'garden-path' sentences, where the preferred interpretation at the point of ambiguity turns out to be incorrect. For example, in sentence (a) the phrase "on the ground" is ambiguous with respect to whether it modifies the verb "set," denoting the place onto which the lid was set, or modifies the noun "lid," denoting the location of the lid at the present time (from the writer's frame of reference). A substantial body of research in the sentence processing literature indicates that the verb modifying interpretation is preferred. Thus, when the disambiguating phrase, bracketed in (a), "on the trashcan" is reached, indicating the noun-modifying interpretation is incorrect, a reanalysis is necessary. Typically, this reanalysis effect is evidenced through longer gaze durations on the disambiguating phrase, relative to similar phrases that do not serve to obviate an established meaning, as in the bracketed phrases in control sentences (b) and (c). Our primary interest lies in individual variation in the magnitude of the reanalysis effect as a function of reading skill and supporting capacities. Results indicate (1) longer reading times in bracketed regions of 1a relative to 1b and 1c, replicating the well-established garden path effect; (2) skill differences account for a large portion of variance in overall reading times; (3) The magnitude of the garden-path effect decreases as a function of increasing skill, primarily Vocabulary knowledge as contrasted with decoding skill or listening comprehension.

a. Bruno will set the lid on the ground [on the trashcan] and then he'll sort the recycling.
b. Bruno will set the lid [on the trashcan] and then he'll sort the recycling.
c. Bruno will set the lid on the ground [beside the trashcan] and then he'll sort the recycling.