REGRESSION-CONTINGENT ANALYSES OF EYE-MOVEMENTS DURING SENTENCE PROCESSING - REPLY TO RAYNER AND SERENO

Altmann, Garnham, and Dennis (1992) explored contextual influences on syntactic ambiguity resolution by monitoring eye movements during reading. In order to resolve a conflict of interpretation given that the different eye movement measures yielded different patterns, we introduced a regression-contingent analysis of reading times, separating trials according to whether the eyes departed from the region of interest with a leftward (regressive) or rightward movement. Rayner and Sereno (1994) argue that various assumptions which they claim underlie the motivation for introducing the regression-contingent measure are in fact flawed. In this paper I demonstrate that these assumptions are incorrectly ascribed to us (while agreeing that they are incorrect), and that Rayner and Sereno's re-analysis of an earlier study by Rayner, Garrod, and Perfetti (1992) neither questions nor threatens the generalizability of the regression-contingent measure. Finally, I discuss some of the uncertainties surrounding the interpretation of first-pass reading times which further motivate the measure we adopted in Altmann et al.'s (1992) study.