Contrasting eye-movements in reading aloud and reading silently to localise a speech deficit.

Background and aims. We report the case of CS, an aphasic patient who made phonological errors in all spoken language tasks. We ask whether his deficit is at the level of abstract phonological encoding or at a post-encoding locus. We contrasted measures of eye-movements taken while CS was reading aloud and reading silently. Our hypothesis was that a central phonological deficit would affect both reading aloud and reading silently, but a post-encoding deficit would allow silent reading to proceed normally.
Method. CS and older adult controls (N=4) read sentences (N=96) aloud and silently while their eye movements were monitored (Eyelink 1000, SR Research) and their speech recorded. Comprehension questions ensured that all participants read for meaning in a normal way. First fixation times, gaze duration (sum of all fixations in a word from when it is first fixated until the eyes move to words further right in the sentence), and right-bounded duration (the total of all fixation times on a word from when a word was first fixated the eyes moved further right in the sentence) were analysed. First fixation is sensitive to dimensions associated with lexical access, like frequency. Gaze durations are affected by access to phonological information, and right-bounded duration indexes later comprehension processes (Calvo, 2001; Huestegge, 2010; Lee, Binder, Kim, Pollatsek, & Rayner, 1999).
Results. Linear mixed models were used to compare log fixation times from CS and controls during reading aloud and reading silently (see Figure 1). First fixation times did not differ between CS and controls in either silent reading (G2=.003, p=.96) or reading aloud (G2=0.067, p=0.80). CS’s gaze durations did not differ from controls in silent reading (G2=0.06 p=.80), but they were different in reading aloud (G2=4.53, p=0.03). Eye-movements were longer in reading aloud, consistent with articulatory difficulties forcing fixations to be longer so that the eyes do not get too far ahead when speech is affected by problems with articulatory planning. The shape of the distribution of fixation times was also consistent with this hypothesis, since the mode remained in the same position as the controls, but the tail lengthened. Right-bounded durations were not different in silent reading (G2=0.46, p=0.50), but were different in reading aloud (G2=9.41, p=0.002), where CS was slower than controls.
Conclusion. Eye-tracking during silent reading and reading aloud has the potential to reveal the locus of a speech deficit. CS was not different from controls on any measure in silent reading and not different on first fixations in reading aloud. A phonological encoding deficit should affect later measures in silent reading as well as reading aloud. We argue that CS’s eye-movement pattern is consistent with his speech error evidence, where there were a large number of false starts, strong frequency effects on fluency and length and position effects for non-words. Together, these suggested a limitation on resources at the interface between phonology and articulation.