Frontiers in Psychology (Apr 2014)
Reliability of Visual-World Eyetracking for Lexical and Sentence Comprehension Tasks
Abstract
Introduction Relatively little is known about the processing changes that support treatment-induced recovery in aphasia. Visual-world eyetracking has been used to detail language processing deficits in aphasia (Meyer, Mack, & Thompson, 2012; Mirman, Yee, Blumstein, & Magnuson, 2011), and may be useful for detecting processing changes that result from treatment. However, the test-retest reliability of eyetracking has not been fully established. One previous study suggested that eyetracking is a reliable measure of lexical access (Farris-Trimble & McMurray, 2013). However, that study used a large number of trials (n=640), which may be impractical in treatment research. Additionally, no previous studies have examined the reliability of eyetracking for sentence comprehension tasks. The present study tested the reliability of eyetracking for lexical and sentence comprehension tasks, using a practical number of trials. Method Unimpaired college-aged adults (n=21) participated in eyetracking experiments testing lexical access and sentence comprehension, in two sessions one week apart. In the lexical access experiment (n=30 trials), participants viewed arrays of four objects and heard a noun matching one object. In the sentence comprehension task (n=48 trials), participants viewed two drawings with reversed participant roles and listened to an active or passive sentence (e.g. The dog is chasing the cat/The cat is being chased by the dog). Participants were instructed to click on the picture that matched the auditory stimuli. Eye data were analyzed in time windows in which a rise in fixations to the target picture was observed: 300-800 ms from word onset in the lexical access experiment, and 0-500 ms from verb offset in the sentence comprehension experiment. Eye fixation patterns were modeled using mixed effects empirical logit regression, with maximal fixed and random effects. These models tested for effects of test session at the group level. To test for reliability of individual eye fixation patterns, participant-specific parameters from each test session were entered into intraclass correlations (ICCs). Results See Figure. In both experiments, no significant effects of session (main effects or interactions with other variables) were observed at the group level (p’s > 0.05). For both experiments, moderately strong correlations were observed in the rise of fixations to the target picture within individuals across sessions (lexical access: ICC(2,1) = 0.531; p = 0.007; sentence comprehension: ICC(2,1) = 0.582, p = 0.002). Discussion The results indicate that eyetracking is a reliable measure of the latency of lexical access and sentence comprehension, even with a limited number of trials. Therefore, this method may prove useful for investigating lexical access and sentence comprehension changes as a result of treatment for aphasia. Figure 1. Proportion of fixations to the target picture across sessions in (a) lexical access and (b) sentence comprehension trials. Vertical lines indicate windows for statistical analyses.
Keywords
