![Sample our Behavioral Sciences journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/110801/TOC-Subject-Sample-2021-Behavioral-Sciences.jpg"})
Abstract
Many deaf individuals do not develop the high-level reading skills that will allow them to fully take part into society. To attempt to explain this widespread difficulty in the deaf population, much research has honed in on the use of phonological codes during reading. The hypothesis that the use of phonological codes is associated with good reading skills in deaf readers, though not well supported, still lingers in the literature. We investigated skilled and less-skilled adult deaf readers' processing of orthographic and phonological codes in parafoveal vision during reading by monitoring their eye movements and using the boundary paradigm. Orthographic preview benefits were found in early measures of reading for skilled hearing, skilled deaf, and less-skilled deaf readers, but only skilled hearing readers processed phonological codes in parafoveal vision. Crucially, skilled and less-skilled deaf readers showed a very similar pattern of preview benefits during reading. These results support the notion that reading difficulties in deaf adults are not linked to their failure to activate phonological codes during reading.
This research was supported by a National Science Foundation Science of Learning Center Program under cooperative agreement SBE 1041725, an NIH (National Institutes of Health) Grant HD26765 and an American Recovery and Reinvestment Act supplement to that grant, and by a Postdoctoral Fellowship (Fonds de recherche sur la société et la culture 125964) to N. N. Bélanger. We thank the participants, Jesse “Rupert” Dubler, Jullian Zlatarev, Bernhard Angele, Timothy Slattery, and Klinton Bicknell for help with data collection or analysis, and Reinhold Kliegl, Matt Traxler, and Sarah White for their comments on an earlier version of the paper.
Notes
1 The groups were not quite matched on the number of participants as the important factor was to closely match the skilled deaf readers to the skilled hearing readers on their reading skill. Additionally, note that there is no control group for the less-skilled deaf readers as it would be highly unlikely to find a group of nondyslexic hearing readers matched on age, reading level, and nonverbal IQ.
2 The successive difference contrasts for type of preview were set up in the linear mixed-effects model (LMM), such that the measures for the identity condition were subtracted from the unrelated condition (to provide a measure of overall preview benefit effects), the measures from the homophone condition were subtracted from the identity condition (for the orthographic preview), and the measures from the orthographically similar condition were subtracted from the homophone condition (for the phonological preview effects).
3 The SKD readers appear to show an effect of phonological preview benefit for the HFLF preview–target pairs, but not for the LFHF preview–target pairs, whereas the SKH readers show the opposite pattern. However, this is true for first fixations only (not for gaze durations). In the separate analyses performed for each group, the interaction Frequency × Homophone–Orthographically. Similar (orthographically similar) contrast (phonological preview) is significant for the SKH readers (t = –2.05) but does not even come close to significant for the SKD readers (t = 0.99). Interestingly, the pattern of means for the four conditions for the SKH readers is pretty stable across first fixations and gaze durations, but the interaction between interaction Frequency × Homophone-Ortho. Similar contrast only reached significance in the first-fixation measures. For the SKD readers, however, the “reverse” pattern of phonological preview in first fixations is not present any more in gaze durations. Examination of the pattern of refixations across frequency and preview conditions, which may explain the difference in effects of phonological preview across first fixations and gaze durations for the SKD readers, we find that between the homophone and orthographically similar conditions, refixations patterns are similar across both frequency conditions with 3% more refixations in the homophone condition than in the orthographically similar condition for the HFLF condition and also a 3% difference between those two preview conditions in the LFHF condition (see ). Thus the different pattern in gaze durations for the SKD readers, relative to the pattern of result in first fixations, cannot be explained by different refixation rates across the different frequency conditions. It is hard to justify the presence of a phonological preview benefit in the LFHF condition for the skilled deaf readers in first fixations based on the present data.
4 The overall preview benefit found in the present experiment is of smaller magnitude than what has been found in prior experiments (for example, Pollatsek et al., Citation1992). We explain this difference by the fact that our hearing participants were drawn from the general population to be matched on age to our deaf participants; thus they may read at a slightly lower level than the undergraduates generally tested in reading experiments investigating skilled readers.
5 The probability of refixating a word is a function of the landing position of a saccade within a word, but also of the launch site of the saccade from the previous word (see Rayner, Citation2009). Rayner (Citation1979) found that readers' eyes generally land in the middle of the first half of a word (preferred viewing location, PVL). It has been shown that the proportion of refixations increases when the eyes do not land in an optimal position in a word (O'Regan, Citation1990; Rayner, Sereno, & Raney, Citation1996). To that effect, we conducted an analysis to investigate the launch sites and landing positions for the three groups of readers. SKD readers launched their saccade prior to the target word farther away (M = 5.8 characters away from the landing position in the target word) than did SKH (M = 5.1) or LSKD readers (M = 4.9; p < .03). A farther launch site for the SKD readers, relative to the SKH readers, is consistent with a landing site closer to the beginning of the target word and farther away from the PVL, thus increasing the probability of refixations. However, surprisingly, the SKH and SKD readers did not differ in the landing position within targets (M = 2.3 vs. 2.2 characters into the word: p = .16). That the SKH and SKD readers did not differ in landing positions cannot explain the lower probability of refixations found for the SKD readers. Rather we suggest (see also Bélanger, Slattery, et al., 2012) that SKD readers are extremely efficient when reading and processing words/text. This is also consistent with the higher skipping rate (see Footnote 6) and lower refixations rate found for the SKD readers.
6 Across conditions, the target word was skipped 24% of the time, with the identical condition skipped 27% of the time, and the other three conditions skipped roughly 23% of the time (with no significant differences). Interestingly, the SKD skipped the target words significantly more often (29% of the time) than did the SKH (24%: p = .001) and the LSKD readers (20%: p < .0001). Similarly, the probability of regressing back to the target in the identical condition was 14%, whereas the other conditions were regressed to about 17% of the time (no significant differences were found between conditions). The SKD readers regressed back to the targets significantly less often (14%) than did the SKH readers (17%: p = .05) and the LSKD readers (19%: p = .03).