Mouse Tracking for Reading (MoTR): A New Naturalistic Incremental Processing Measurement Tool, Journal of Memory and Language
Ethan G. Wilcox, Cui Ding, Mrinmaya Sachan, and Lena A. Jäger
[http]
Abstract
We introduce Mouse Tracking for Reading (MoTR) a new incremental processing measurement tool that can be used to collect word-by-word reading times. In a MoTR trial, participants are presented with text, which is blurred, except for a small region around the tip of the mouse. Participants must move the mouse to reveal and read the text. Mouse movement is recorded, and, using a postprocessing pipeline we present, can be analyzed to produce scanpaths as well as word-by-word reading times. We validate MoTR in two suites of experiments. In the first experiment, we collect data for the English-language Provo Corpus (Luke and Christianson, 2018). We analyze scanpaths and show that participants interpolate between two types of strategies for reading during a MoTR trial – sometimes they fixate on individual words, somewhat akin to eye-tracking, while other times they produce a more constant pass over the text, slowing down in response to processing difficulties. Taking these strategies into account, we show that the word-by-word reading times produced by our data analysis pipeline correlate well with previously collected eye-tracking data for this corpus, and that these correlations are higher than those produced by SPR data, which we also collect for the corpus. Furthermore, we demonstrate that there is a linear relationship between by-word MoTR values and word-level surprisal values, as has been previously shown for eye-tracking data (Smith and Levy, 2013). In the second experiment, we assess whether MoTR can be used to study sentence processing phenomena in targeted psycholinguistics experiments. Using materials from Witzel et al. (2012), we show that MoTR can reveal English speakers’ preferences for low attachment during online sentence comprehension. We argue that MoTR presents a compelling tradeoff between multiple experimental considerations: It is cheap to run and can be presented in a browser enabling the collection of data over the internet. It is more naturalistic than some alternative processing measures, allowing participants to skip words and regress to previous sentence regions. Finally, it has good sensitivity, detecting signatures of psycholinguistic processing behaviors from a relatively small number of participants.
