Cogsci 2013
31st July – 3rd Aug, 2013
Berlin, Germany
The visual processing deficits of Chinese dyslexics
Orieta H.Y.Wong & Michael C.W.Yip
Department of Psychological Studies
The Hong Kong Institute of Education
() ()
Abstract
The present research aimed at identifying cognitive processing that works at the initial stage of letter identification. First, we used the word superiority task to set the baseline of visual detection rate and then we used the letter matching task to evaluate the phonemic transference rate of the participants. It was hypothesized that physically-different letters required longer processing time as the judgment depended not only on visual detection, but also on a process of phonemic transference. Our results found that the control group had a higher letter identification rate in study one and a faster reaction time in study 2 when compared with the dyslexic group, showing that visual detection rate was a good predictor of reading disability. However, phonemic transference rate was not a good predictor of dyslexia. One of the reasons may be that the process of letter identification involved cognitive skills other than that of phonemic transference alone.
Keywords: visual processing deficit; Chinese dyslexic; superiority effect; letter matching task; visual word recognition
Some additional findings:
(1) Deficits in dyslexia is modality specific across different cultures
We found that out of the 16 words given, about half (8 words) of them were non-frequent words to junior students (S1 & S2) in a Band 2 Chinese medium school. These non-frequent words included: “calf, seal, pods, pots, scam, sham, slot and spot.” (Coglab version of WSE). This implied that unless knowledge had been gained through learning, an automatic grapheme-phoneme conversion would not have taken place. For this reason, we proposed a hypothesis of word familiarity effect on word identification in Chinese L2 learners of English. The same case may not happen in L1 learner of English as they can pick up the phonemic rules by simple matching and extend the same knowledge to self-learning of new vocabularies.
(2) Deficits in dyslexia is modality specific across different subjects
When we looked at the results of each participant individually, we found that in some cases, the reaction time of specific dyslexic students in the physical identical letter case (e.g. ~student 019: 671.82 msec; ~student 020: 657.49 msec; student 025: 618.4 msec; student 029: 607.52 msec) was actually shorter than the mean reaction time of that of the control group (RT11: 759.054), showing that these students could actually detect letters/ words visually faster than some normal reading ability participants.
(3) Dyslexia is modality specific
To locate how sensory modalities work in specific students, we analyze specific students in detail. Take a dyslexic participant 006 as an example. 006 had an accuracy rate of 54.17% in the letter-in-word case in the Word Superiority Effect task (WSE) in Study One, a bit lower than the mean rate of dyslexic students (~56.675%). However, she had an accuracy rate up to 70.83% when identifying single letters in WSE, much higher than that of the mean rate of dyslexic students in the single letter case (~65%). Judged in this way, the student might not risk a deficit in visual detection and visual letter identification, but rather, deficits in transferring the visual code to a phonemic or [lexical] representation (Posner, 1986, p.80).
Such a speculation was proved by the fact that the student had a slower phonemic transference rate (~139.07msec) as compared with the mean phonemic transference rate of the dyslexic group (119.312msec). It was judged that the student was slow in phonemic interpretation, which in turn slower lexical decision of the visually exposed words on the screen and hence a lower accuracy rate in the letter-in-word case in the Word Superiority Task in Study One.
(4) As for another dyslexic participant 029, aside from detecting visual stimulus faster than the normal group, his phonemic transference rate as calculated (RT2-RT1) was also a bit faster (~24.08 msec) than the mean reaction phonemic transference rate of the normal group (~29.86 msec). The same student 029 was found to be restless during the test though and gained exceptionally low marks in the First Term English examination (~5%). We suspected that he risks problems either in motivation or attention. We suggest therefore a modality specific hypothesis for the etiology of dyslexia.
(5) To further prove our observation, a correlation test between letter/word identification rate in study one and mean reaction time in study two was conducted. Result showed that the reaction time of the dyslexic group in the letter matching task was negatively correlated with word identification rate in WSE.
Correlations
WORD / RT2WORD / Pearson Correlation / 1 / -.293
Sig. (2-tailed) / . / .289
N / 15 / 15
RT2 / Pearson Correlation / -.293 / 1
Sig. (2-tailed) / .289 / .
N / 15 / 15
If the cognitive process of visual detection and phonemic transference work automatically in a parallel way in the case of a normal student in study one, then it is certainly not so in study 2 when the task was pressured by accuracy. The comparatively much longer reaction time in study two signified that other cognitive processes, other than that of letter detection alone, had taken place in study two.
In fact, when the researcher was carrying out the experiment, some students claimed that they can actually count how many mistakes they have made. This showed that the students could have made automatic choices in the matching task before they struck the key and in this sense, a faster reaction time could point to the fact that orthographic images had left memory traces for a delayed cognitive process of phonemic transference and this further supported the hypothesis of the dual-route model, suggesting that visual word recognition could work in a modality specific way.
(6) To support this hypothesis, regression tests on the identification rate of letter-in-word case in study one and reaction time of study two were carried out. The findings of particular interest was that for the control group, the accuracy rate of the matching task showed a positive correlation with the accuracy rate of letter-in-word case on word superiority effect (WSE) while the result of that of the dyslexic group showed a negative one. To show such pattern more clearly, quadratic regression graphs on both the control group (fig. 5-7) and the dyslexic group (fig. 8-10) were plotted.
(7) The graph for the control group shows that the accuracy rate for letter-in-word case in study one is positively correlated with the reaction time for physical identical letter match, name code match and phonemic transference in study two in the sense that a longer reaction time in study two was positively correlated with a more accurate letter recognition rate in study one. It seemed that for the letter-in-word case in study one, there was more or less a tendency for students in the normal group to identify a letter first by identifying a word and in this sense, a cognitive process of phonemic transference should have taken place automatically by transforming the orthographic letter to its phonemic and auditory codes before a judgment, enlightened by such a contextual effect, was made and for this reason, a longer reaction time was needed.
The same strategy seemed to have been employed in study two in that longer reaction time was positively correlated with a higher letter identification rate. From this perspective, it seems that the process of phonemic transference is, to a larger or lesser extent, a natural automatic process for students with normal reading ability.
As for the dyslexic group, the graph shows a negative trend of correlation. It seems that for some students on the faster spectrum, a cognitive process of automatic orthographic comparison have taken place, the cognitive skill of which allows faster detection also in the letter-in-word case in WSE. However, for other students who are much slower in visual physical match of identical letter, it seems that either the visual detection rate is too slow for visual detection in 40 milliseconds or that phonemic transference does not take place fast enough for letter in word identification. In fact, those students on the faster spectrum can perform even better than some students in the normal group. Their difficulties in reading, involves not so much deficit in visual word recognition than deficit in other cognitive areas.
(8) What then finally affect the development of reading disabilities?
Against our prediction, the phonemic transference rate of the dyslexic group does not show a positive or a negative correlation with letter-in-word identification rate (Refer to Figure 10). We suspect therefore other psychological factors like motivation, attention and word familiarity effect and/or those areas involved in the end stage of word processing might affect more.
(9) As does the normal group, therefore, the dyslexic group shows modality specific cognitive processing concerning letter identification albeit at a globally slower rate. This implies that somewhat different modalities have involved in the cognitive process of word processing when reaction time is a variable. In contrast, both modalities of visual detection and phonemic transference have positive effects on the latency of word identification in both control and dyslexic groups, with a different degree of sophistication, at a different stage of cognitive process of word identification. In fact, such variation can be seen, not only between two different groups, but also among the same group of dyslexic students and this signifies that further tests needed to be done to correctly identify the affecting variables for a particular dyslexic student, though group performance on the whole showed a globally slower level of letter/word identification rate. Data reviews therefore that for some students in the dyslexic group, particular strong inclination on one modality, namely the visual-detection modality and the phonemic transference modality may benefit for higher accuracy rate in the word superiority task. However, a deficit in any one of the modalities may also slower the cognitive processes and lowers the accuracy rate in visual word decoding in general.
(10) The data above shows that identification of modality specific dyslexia can be done through monitoring Reicher & Wheeler’s (1969, 1970) word superiority effect task (WSE) and Posner’s visual letter matching task.