Skip to main content
PLOS ONE logoLink to PLOS ONE
. 2021 Jan 22;16(1):e0243440. doi: 10.1371/journal.pone.0243440

Perceptual expertise with Chinese characters predicts Chinese reading performance among Hong Kong Chinese children with developmental dyslexia

Yetta Kwailing Wong 1,*, Christine Kong-Yan Tong 2, Ming Lui 3, Alan C-N Wong 2
Editor: Athanassios Protopapas4
PMCID: PMC7822259  PMID: 33481782

Abstract

This study explores the theoretical proposal that developmental dyslexia involves a failure to develop perceptual expertise with words despite adequate education. Among a group of Hong Kong Chinese children diagnosed with developmental dyslexia, we investigated the relationship between Chinese word reading and perceptual expertise with Chinese characters. In a perceptual fluency task, the time of visual exposure to Chinese characters was manipulated and limited such that the speed of discrimination of a short sequence of Chinese characters at an accuracy level of 80% was estimated. Pair-wise correlations showed that perceptual fluency for characters predicted speeded and non-speeded word reading performance. Exploratory hierarchical regressions showed that perceptual fluency for characters accounted for 5.3% and 9.6% variance in speeded and non-speeded reading respectively, in addition to age, non-verbal IQ, phonological awareness, morphological awareness, rapid automatized naming (RAN) and perceptual fluency for digits. The findings suggest that perceptual expertise with words plays an important role in Chinese reading performance in developmental dyslexia, and that perceptual training is a potential remediation direction.

Introduction

Reading is an essential skill to acquire in normal schooling experience. However, learning to read is difficult for some children affected by developmental dyslexia. Developmental dyslexia is characterized by difficulties in accurate or fluent word recognition, spelling, and decoding of words despite adequate instruction, intelligence and sensory abilities [1]. It is a prevalent neurodevelopmental disorder, affecting about 5–10% of the population depending on the definition adopted in various estimates [2].

The causes of developmental dyslexia are hotly debated. Theoretical proposals include impaired ability to perceive, access and manipulate sounds of spoken words in awareness (“phonological awareness”, [35]), deficits in rapid processing of auditory speech input [68], deficits in visuospatial processing along the magnocellular-dorsal pathway [9,10], deficits in processing of rapid stimulus sequences [11], general visual attentional deficit [12], deficits in processing crowded visual images [13], and excess neural noise in the brain regions important for reading [14]. The search for a single cause of dyslexia remains elusive, probably because it is a complex disorder caused by the interaction of multiple neural, cognitive, and genetic factors [1417].

This study investigates yet another factor that may help explain developmental dyslexia: perceptual expertise with words. Perceptual expertise refers to the excellent perceptual skills in individuals who can efficiently and effortlessly differentiate between different visual objects within their expertise domains [18]. While visual discrimination between different object categories (e.g., between a car and a bird) can be relatively easy, visual discrimination between different objects within an object category (e.g., between one car from another car) can be a lot more challenging. This ability is also referred to ‘subordinate-level categorization’–the ability to categorize visual images at more specific levels (e.g., ‘kitchen table’ instead of ‘table’) or even at individual level (e.g., “my kitchen table”, [19,20]). Previous research emphasizes how perceptual expertise is supported by excellent shape processing of expert objects [21,22]. In addition to shape processing, recent evidence shows that color and semantic information of the objects is also useful during expert object recognition [23,24].

In word reading, it goes without saying that fluent readers have excellent perceptual skills in discriminating between different words that are often visually similar, e.g., ‘car’ and ‘can’. Empirical evidence indicates that fluent readers can discriminate visually presented word sequences from highly similar alternatives at a glance [25]. Importantly, such perceptual expertise with words is learned, presumably through years of schooling in which they gain substantial experience in word recognition and reading. From this perspective, developmental dyslexia may involve a failure of developing adequate perceptual expertise with words to support fluent word reading despite frequent exposure to visual words in educational experience.

Converging evidence suggests that perceptual expertise with words may be an important predictor of word reading performance. First, it has a strong face validity to explain developmental dyslexia, because reading is impossible without first processing the visual appearance of the letters, characters, and words. Hence difficulties in discriminating between these visual codes should predict worse reading performance. Second, it is well established that children with developmental dyslexia often show under-activations in the occipitotemporal regions when processing word stimuli [2632]. Interestingly, visual expertise for object recognition is typically accompanied by an increased recruitment of the occipitotemporal regions for processing various domains of visual objects, e.g., faces [33], letters and words [3436], musical notation [37], birds and cars [38,39], chessboards [40], radiographs [41], and lab-trained expertise with computer-generated novel objects [4245]. It has been proposed that the engagement of the ventral visual cortex, in particular the visual word form area (VWFA) for words, is a result of perceptual expertise [46], suggesting that children with developmental dyslexia fail to develop perceptual expertise with word stimuli. Third, recent evidence identified impairment in recognizing faces and general objects in individuals with developmental dyslexia, suggesting that they may have deficits in visual discrimination in the ventral visual stream in general in addition to that for words [4749].

Perceptual expertise with words might be particularly important in Chinese reading because of three reasons. First, Chinese characters are visually complex. Evidence comes from the visual analyses of the 700 most frequently used Chinese characters and the lowercase Roman letters, which showed that Chinese characters were 3.7 times more complex (see Table A4 in [50]). Second, there are several thousand commonly used Chinese characters, and primary school students are expected to master more than 2,500 of them [51]. This poses a huge challenge to visual perceptual analyses during Chinese word reading. Third, the correspondence between the visual code (i.e., stroke patterns) and the phonemes in Chinese is relatively opaque and irregular [52]. According to the orthographic depth hypothesis, this might encourage readers to rely more on the visual-orthographic structure of the visual codes during reading [53]. These make Chinese reading a good candidate language system to explore the role of perceptual expertise with words in reading performance in developmental dyslexia.

Importantly, perceptual expertise with words is different from the several existing visual accounts of developmental dyslexia. First, it is different from the magnocellular-dorsal account which proposes that developmental dyslexia is related to the processing deficit in the dorsal ‘where’ pathway of the visual system for visuospatial processing [9,10]. Visual expertise for object recognition is typically found to engage the occipitotemporal cortex, which is in the ventral pathway of the visual system [33,34,3644,54,55]. The ventral pathway is traditionally regarded as the ‘what’ pathway for visual object identification and recognition, in contrast to the dorsal ‘where’ pathway for visuospatial processing [56,57].

Second, perceptual expertise with words does not simply reflect visual attention span or general visual attentional skills used to account for developmental dyslexia [12,58,59]. Measures of perceptual expertise and visual attention skills often appear similar because they both involve simultaneous recognition of multiple visual items presented horizontally and in brief durations. However, they have different assumptions about the underlying skills and the category specificity of these skills. In particular, the critical skills underlying perceptual expertise is shape processing, while that underlying attentional account is visual attention. Also, perceptual expertise is often highly specific to a certain object category but attentional skills are not. For example, a car expert can be outstanding in recognizing cars, but only has average performance with birds. This category-specific expertise in object recognition is consistent with the findings that behavioral effects indicating perceptual expertise are typically confined to one’s domain of expertise [21,22,6063]. The attentional account of developmental dyslexia, however, is often considered general and observable with different types of visual stimuli such as numbers, shapes or symbols [58,59,6466].

Third, perceptual expertise with words is different from orthographic processing of words. Orthography refers to how a script represents phonological, semantic and morphological information in a given writing system [67,68]. Orthographic processing often concerns how linguistic information can be abstracted from written scripts, e.g., the letter-to-phoneme correspondence [e.g., 53], the correspondence between letter order and semantics [e.g., 69], the abstract representation of letter and word identities despite their visual appearance in different cases and fonts [e.g., 70], and the frequency, validity and position of letter combinations [e.g., 71]. One important characteristic common to all these orthographic tasks is that one needs to be familiar with the language system to perform orthographic judgments. In contrast, perceptual expertise views words as ‘visual images’, which could be analogous to highly similar ‘line drawing patterns’ that need to be individuated and identified. Fluent discrimination of these images lies in the efficient extraction of visual diagnostic information [7275]. Importantly, the diagnostic parts of the words may or may not have linguistic values, e.g., contrast, edges, line junctions, terminations, intermediately complex units, etc. [7680]. Therefore it is possible to invite novices to perform visual discrimination judgments with unfamiliar objects [21,81,82], while it would be basically impossible for novices to perform orthographic judgments with an unfamiliar language system. While perceptual expertise with words and orthographic processing are different with different emphases and assumptions, they are not completely unrelated. In some cases, perceptual discrimination of words and orthographic processing might be partially supported by overlapping information, e.g., when some levels of visual diagnostic parts of the words happen to have nice phonemic correspondence. The extent to which they overlap likely depends on multiple factors, such as how the visual codes of a writing system differ from one another, the mapping between the visual code and various types of linguistic information, etc.

It is also important to note that perceptual expertise with words is different from word reading tasks that requires one to read words aloud because their critical task demands are different. For perceptual expertise with words, the critical task demand is to tell different words apart based on the visual input. To achieve this, the amount of visual exposure to the word stimuli is typically manipulated during measurement, such as to limit the presentation duration of the stimuli. Importantly, the discrimination task does not explicitly examine abilities in phonology, semantics and speech production. As a result, accurate discrimination of words can be accompanied by incorrect or missing phonological and phonemic representation of the words (e.g., a novice reader of Thai judging Thai words visually). In contrast, word reading tasks explicitly require one to read aloud the words, which involves accurate extraction of the visual information from the print, followed by extracting the grapheme-phoneme mapping, phonological or phonemic information, and eventually pronunciation of the word through the speech production system [83]. In other words, many abilities in addition to visual discrimination are explicitly assessed during word reading tasks, and correct performance in word reading tasks requires accurate visual discrimination, phonology and speech production. With these complex task demands, it makes sense that word reading tasks rarely manipulate the amount of visual exposure to the words, i.e., brief presentation of words is rarely seen during word reading tasks. In sum, even though both tasks use words as the stimuli, measures of perceptual expertise with words do not equal the measure of word reading because they focus on a subset of the task demands required in word reading.

Although words in measures of perceptual expertise are often located in different visuospatial locations horizontally (Fig 1), perceptual expertise with words does not simply refer to the type of visuospatial abilities with which children with dyslexia are often suggested to be superior [26]. In these visuospatial tasks, participants are often required to report complex visual patterns from memory, perform mental depth rotation of three-dimensional objects, or reproduce a map after exploring a three-dimensional virtual environment [26,84]. The task demands of these tasks heavily involve visual working memory and/or long-term memory, mental imagery of the structure of three-dimensional objects, and judgment of spatial distance and relationship between multiple objects. In contrast, perceptual expertise with words emphasizes on fine-level discrimination between two-dimensional word stimuli, which does not involve any explicit spatial judgment or spatial imagination of the stimuli (e.g., performing plane- or depth-rotation with words, or judging the distance and positions between the words). With a brief time gap between stimulus presentation and report, task performance is likely limited by visual perception instead of visual working memory. However, it is worth noting that words in perceptual expertise measures are often arranged in the two-dimensional space, in which the spacing between the words also influences the difficulty of the perceptual expertise measures, as in the case of visual crowding [13,63,85]. Overall, perceptual expertise with words taps on cognitive processes that are distinct from that captured by these visuospatial tasks.

Fig 1. A sample trial of the perceptual fluency task for characters.

Fig 1

In this study, we tested whether perceptual expertise with words predict word reading in developmental dyslexia. Chinese children with developmental dyslexia participated in multiple tasks that measured their performance in Chinese word reading, perceptual expertise with words, non-verbal IQ, rapid automatized naming (RAN), phonological awareness, and morphological awareness (see Methods). Perceptual expertise with words was measured by a perceptual fluency task, which has been used to quantify visual expertise for different categories like words and musical notation [25,63,8688]. In this task, the time of visual exposure to sequences of four Chinese characters was manipulated and limited such that the speed of successful discrimination of sequences of Chinese characters at an accuracy level of 80% was estimated (see Methods). A separate perceptual fluency task for four-digit strings was also included. This task, together with RAN, ensured that any explanatory power of the perceptual fluency for characters on Chinese reading would not be explained by visuospatial abilities or discrimination abilities general to all kinds of visual stimuli and objects.

We addressed two questions. First, Pearson correlation analyses were performed to test whether perceptual expertise with words, as measured by the perceptual fluency task, is correlated with reading performance in developmental dyslexia. Second, using hierarchical regression analyses, we explored whether perceptual expertise with words is a unique predictor of variability in reading performance in developmental dyslexia, above and beyond other measures that are well-known to predict reading performance among children with developmental dyslexia, including age, RAN, phonological awareness and morphological awareness.

We were interested in investigating the variability of abilities among children with dyslexia; therefore, the difficulty level of the tasks was designed to be appropriate for this group of children. Based on our pilot testing, task difficulty levels that are appropriate for children with dyslexia were too easy for typically developing children, leading to ceiling effects, in particular for tasks that measured accuracy (see Methods). Since our study aimed to examine individual differences among those with dyslexia, it is important for all tasks to be off ceiling and floor, while it might not be as much of a concern for studies with other goals, such as that performed group comparison between children with dyslexia and typically developing children [26,49] or that performed logistic regression to categorize children as those with dyslexia or not [89]. Hence comparing the role of perceptual expertise with words among children with and without dyslexia was out of the scope of this study.

Method

Participants

Thirty-five (6 females, mean age = 11.2 years old, SD = 1.08 years old) Hong Kong Chinese students from Primary 3 to Primary 6 were recruited from their schools. All participants received clinical diagnosis of developmental dyslexia according to the Hong Kong Test of Specific Learning Difficulties in Reading (HKT-SpLD; [90]). This diagnosis ensured that they had specific learning difficulty in reading despite adequate instruction and normal intelligence; and that such difficulty was not caused by other neurological disorders. All participants reported to have normal or corrected-to-normal vision and hearing. Four additional participants withdrew during the session and their data were thus discarded. Some of the participants continued to engage in a subsequent training program, but the results of that are beyond the scope of the current paper. The project was approved by the Research Ethics Committee of Hong Kong Baptist University and the Survey and Behavioural Research Ethics Committee of the Chinese University of Hong Kong. Informed consents in written format were obtained from both the participants and their parents.

Material

One hundred and sixty Chinese characters were selected from a database of characters established by the Education Bureau for Primary 1 to Primary 5 students in Hong Kong [91]. In addition, the ten Chinese numerals (from 1 to 10) were also included for the speeded reading task. For the computerized tests in our study, stimuli were presented on a gaming monitor with high temporal sensitivity (BenQ XL2430T; with a refresh rate of 144Hz and GTG of 1ms), which was controlled by MATLAB (MathWorks, Natick, MA) and the Psychophysics Toolbox extension [92,93]. Digital numbers from ‘0’ to ‘9’ (except ‘1’) were used for the fluency task for digits because ‘1’ is simply a vertical stroke which easily stands out from other alternatives. The numbers in the font of Cambria were used for the rapid automatized naming (RAN) task.

Procedure

All participants engaged in a two-hour session with frequent breaks. They completed the tasks in the following order: perceptual fluency for characters and digits, RAN, speeded reading, non-speeded reading, non-verbal IQ, phonological awareness, morphological awareness. A 5-minute break was given after non-speeded reading and one after phonological awareness to keep up the motivation level of the participants and avoid fatigue. The participants’ parents filled out a questionnaire about participants’ education background, medium of instruction, general health, and types and intensity of therapy or training previously received.

Because the currently reported data were part of a training study, two sets of stimuli were prepared with matched difficulty for the pretest and posttest with counterbalanced order. Therefore, for the current data, half of the participants were tested with stimulus set 1 and the other half with stimulus set 2. Performances of the two groups of participants using two stimulus sets were comparable for all tasks (ts ≤ 1.42, ps ≥ .165), except for morphological awareness, in which the performances with the two stimulus sets were significantly different, t = -2.45, p = .02. This difference was driven by one of the subtasks, the word production task (see below; t = -3.21, p = .003) but not in the other subtask of concept production (p = .314).

Non-verbal IQ

The Raven’s Standard Progressive Matrices was used as a standardized measurement of nonverbal intelligence. During each trial, participants saw a visual geometric pattern with a missing part. Then, they verbally selected out of six options the one that would best complete the missing part of the geometric figure. The score was calculated by the number of correct trials out of all 36 trials.

Perceptual fluency

The task measured one’s perceptual expertise with Chinese characters and digits. The characters were commonly learned by Grade 1 students and therefore the participants were highly familiar with these stimuli. A similar task has been used for English words, digits, and musical notation [25,8688]. During each trial (Fig 1), a fixation cross first appeared at the centre of the screen for 200 milliseconds (ms), followed by a 500-ms pre-mask, the target image with a sequence of four characters or digits for a varied duration, and finally a 500-ms post-mask. After that, two images appeared side-by-side, with one identical to the target sequence, and the other as a distractor sequence generated by replacing one of the four characters or digits with a different one randomly drawn from the whole set of characters or the digits (‘0’ to ‘9’, except ‘1’ since it tends to stand out relatively to other alternatives). The position of the replaced character in the distractor was counterbalanced. The pre- and post-masks were grayscale images created by segments of prints (e.g., letters and digits; Fig 1). These target and distractor sequences were always arranged horizontally (Fig 1), and were checked by a native Chinese reader that none of them formed any semantically meaningful words or expressions. Participants pressed the ‘Z’ (left) or ‘M’ (right) key on the keyboard to indicate whether the left or right image was identical to the target, with no time limit for response. The duration of the target was changed trial-by-trial depending on the performance, in order to find the threshold duration required for a participant to achieve an 80% accuracy (QUEST; [94]). There were three blocks for Chinese characters and three for digits, and the order was the same for all participants (Chinese-digit-Chinese-digit-Chinese-digit). The first block for each type of stimulus contained 10 practice trials and 40 experimental trials, and the second and third blocks contained 3 practice trials and 40 experimental trials. The perceptual fluency scores for characters and digits were calculated by averaging the logarithm of the duration thresholds across blocks such that a lower value corresponds to higher fluency. Duration threshold is a type of response time measure, which tends to have a non-linear relationship with performance and makes the numerical differences difficult to interpret. For example, a 100ms response time improvement from 200ms to 100ms means a significant improvement (a 100% change), while that from 2100ms to 2000ms is relatively negligible (a 5% change). Transforming the response time measure with logarithm would linearize the relationship between duration threshold and performance, i.e., the same 100ms improvement between 200ms and 100ms would become 0.30 in the log scale, a much larger difference than that between 2100ms and 200ms (0.021). This is a commonly applied strategy to deal with non-linear relationship between variables [95], which makes the findings more interpretable.

Rapid Automatized Naming (RAN)

The RAN task with digits was adapted from the standard RAN task [96]. Ninety digits (each of the five digits 2, 4, 6, 7, and 9 repeated 18 times) were printed in random order on a sheet of A3 paper along multiple horizontal rows. Participants were asked to read out the digits printed on the paper from left to right, and from the top to the bottom rows, as quickly and as accurately as possible. A short practice list with the unused digits (0, 1, 3, 5, and 8) was given to make sure participants understood the instruction. They were then required to read the experimental list twice. The score was calculated by averaging across the two trials the number of digits correctly read minus the number of digits incorrectly read within 30 seconds.

Speeded chinese reading

This task measured the Chinese single character reading efficiency under time pressure, and was similar to the reading test commonly used in the reading research [e.g., 89,90,97]. The procedure was identical to the RAN task, except that Chinese characters were used. The 90 characters were composed of 80 Grade 1 level characters selected from an established database [91], plus ten Chinese numerals. Relatively easy characters were included in this task such that the reading accuracy would be high and performance would mainly vary on the speed of reading familiar characters.

Non-speeded chinese reading

This task measured the number of Chinese characters read correctly without any time pressure. Eighty-five Chinese characters commonly learned by Grade 1 to Grade 5 students [91] were included as testing stimuli such that the characters would cover a wide range of difficulty levels. Each of the characters was presented on a separate computer screen. Participants were instructed to read out loud each character or to say ‘pass’ if they did not know it. The score of this test was the number of characters read correctly. The 85 characters were ordered according to a descending order of frequency, calculated as a weighted average from various Chinese word frequency reports [91].

Phonological awareness

The task consisted of two parts. The first part adapted a version of the phoneme deletion task used in a previous study with Chinese students [89]. During each trial, the experimenter orally presented a real Chinese character, which corresponded to a syllable, and asked the participants to pronounce the sound when a given phoneme was deleted from the character. For example, /kei3/ without the /k/ sound should be pronounced /ei3/. Each trial involved deletion of either the initial or the final phoneme of a character.

The second part was an oddball task [98]. During each trial, participants heard three Chinese characters, i.e., three syllables, from a clip pre-recorded by a native speaker of Cantonese, a dialect of Chinese and the most common mother tongue in Hong Kong. Two of the three shared either the same onset phoneme, such as /k/ sound in kei3; or the ending rime, such as ei3 in kei3. The remaining character did not share the same onset or rime with the target pair, and thus became the oddball in the sequence. Cantonese is a tonal language, and the tone of a character is indicated by the number (e.g., ‘1’ denotes a high flat tone; ‘2’ denotes a rising tone; ‘3’ denotes a flat mid-pitch tone lower than ‘1’; etc.). Notably, identical onset phoneme and rime pairing with different tones could have distinct meanings (e.g., kei1, kei3, kei4 and kei5 may mean abnormal, to hope for, a flag, and to stand up respectively; and there are additional homophones with other meanings too). In the current manipulation, the tone could either be shared or all different among the 3 characters within each trial. Participants were required to indicate which sound was the most dissimilar to the others (i.e., the oddball) in the 3-character sequence. Each part was preceded by two practice trials with feedback, and would finish when all trials were finished or when a participant failed in four consecutive trials. There were 13 trials in part 1 and 12 trials in part 2. The task score was the sum of the numbers of correct trials in the first and the second part, with a maximum value of 25.

Morphological awareness

The task consisted of two parts. The first part adapted a version of a concept production task used in previous studies [89,99]. During each trial, participants listened to a scenario that described a novel object or concept, and were asked to come up with a word to represent a concept. An example scenario was this: “The scene on a hot day early in the morning is called a hot scene. What would we call a scene on a cold day early in the morning?” The correct answer would be a ‘cold scene’. There were 19 trials.

The second part adapted a version of the word production task used in previous studies [89,100]. During each trial, participants first listened to a two-character word (e.g., ‘紅色’, or ‘red color’), in which one of the morphemes was highlighted (e.g., ‘紅’, which means ‘red’). In the first 9 trials, participants were asked to use the same morpheme ‘紅’ to create a new word in which the morpheme would have the same meaning as that of the original two-character word (e.g., ‘紅蘋果’ or ‘red apple’, in which the morpheme ‘紅’ also means ‘red’). In the next 9 trials, participants were asked to use the same morpheme (e.g. ‘紅’ which means ‘red’) to create a new word in which the morpheme would have a different meaning (e.g., ‘花紅’ or ‘bonus’, in which the morpheme ‘紅’ means ‘extra’). Each part was preceded by two practice trials with feedback and would finish when all trials were finished or when a participant failed on four consecutive trials. The task score of this morphological task was the sum of the numbers of correct responses in the first and the second part, with a maximum value of 37.

Results

Descriptive statistics and reliability of the measures

Descriptive statistics (means and SD) of all measures were reported in Table 1. To check if there was sufficient precision and variability in the reading and component skill measures, the reliabilities of all measures were computed (Table 1). The high reliability values (all above .83) indicate high internal consistency of the measures in general for finding correlations between the reading and the component measures.

Table 1. Descriptive statistics and reliabilities of different reading and component skill measures.

For perceptual fluency, the corresponding threshold values are directly transformed from log duration values in the table for easier interpretations of the findings.

Task Mean (SD) Range Possible Range Type of Reliability Reliability
1. Speeded Reading 42.33(11.77) 11.5–67 0–90 Test-retest reliability 0.899
2. Non-speeded Reading 59.83(16.89) 2–81 0–85 Cronbach’s alpha 0.961
3. Morphological Awareness 24.69 (5.67) 7–34 0–37 Split-half reliability 0.837
4. Phonological Awareness 9.69 (4.28) 0–19 0–25 Split-half reliability 0.83
5. RAN 57.77(13.21) 34–88 --- Test-retest reliability 0.929
6. Perceptual fluency for Digits
            log duration 2.58(.41) 1.91–3.61 --- Cronbach’s alpha 0.889
            corresponding threshold (ms) 381.4 82.2–4109.7
7. Perceptual fluency for Characters
            log duration 2.87(.42) 2.17–3.72 --- Cronbach’s alpha 0.849
            corresponding threshold (ms) 747.7 146.2–5218.2
8. Raven’s Progressive Matrices 30.91(3.89) 19–36 0–36 --- ---

The standard deviations (SD) are omitted because they are computationally different before and after applying the log transformation and thus any direct transformation of these values would be misleading.

Relationship between perceptual fluency for characters and reading

Table 2 shows the pairwise Pearson-Product correlations among the reading and component skill measures. Several aspects of the pattern of correlations are noteworthy. First, both speeded and non-speeded reading scores were highly correlated with perceptual fluency for characters, r(33) = -.665 & -.610 respectively, ps < .001 (Fig 2). Second, the speeded and non-speeded reading scores were correlated with perceptual fluency for digits, r(33) = -.477, p = .004, and r(33) = -.358, p = .035 respectively. Third, the speeded and non-speeded reading scores were also correlated with RAN, r(33) = .649, p < .001, and r(33) = .398, p = .018 respectively. Fourth, the reading measures were not correlated with morphological or phonological awareness (rs < .323, ps > .058). Overall, there were high correlations between reading and component skill measures such as perceptual fluency for characters, perceptual fluency for digits, and RAN; these component skill measures were also correlated significantly.

Table 2. Pearson correlation matrix for the speeded and non-speeded reading as well as other component skill measures.

Speeded Reading Non-speeded Reading Morphological Awareness Phonological Awareness RAN Perceptual fluency for Digits Perceptual fluency for Characters
Speeded Reading 1
Non-speeded Reading .790 1
Morphological Awareness .283 .323 1
Phonological Awareness .188 .214 .220 1
RAN .649 .398 -.045 -.077 1
Perceptual fluency for Digits -.477 -.358 -.341 -.209 -.495 1
Perceptual fluency for Characters -.665 -.610 -.433 -.236 -.477 .727 1

Note: Correlations significant at p < .01 were in black and bold, while those significant at p < .05 were in black. Non-significant correlations are in grey.

Fig 2.

Fig 2

Scatterplots of the scores of (A) speeded reading and (B) non-speeded reading against perceptual fluency for characters in log duration.

Perceptual fluency for characters as a unique predictor of reading

To examine if perceptual fluency specific to characters can further explain reading performance beyond the predictors commonly found in past studies of developmental dyslexia, an exploratory hierarchical regression was conducted with age, non-verbal IQ, phonological awareness, morphological awareness, RAN, and perceptual fluency for digits entered simultaneously in the first block, followed by perceptual fluency for characters in the second block.

The results are summarized in Tables 3 and 4. When speeded reading was the dependent variable, perceptual fluency for characters accounted for an additional 5.3% of variance in addition to other predictors (Block 1: R2 = .654, F(6,28) = 8.83, p = .000019; Block 2: R2 = .707, F(7,27) = 9.29, p = .000008; ΔR2 = .053, ΔF(1,27) = 4.82, p = .037; Table 3). With non-speeded reading as the dependent variable, perceptual fluency for characters accounted for an additional 9.6% of variance in addition to other predictors (Block 1: R2 = .494, F(6,28) = 4.56, p = .002; Block 2: R2 = .590, F(7,27) = 5.56, p = .000482; ΔR2 = .096, ΔF(1,27) = 6.35, p = .018; Table 4).

Table 3. Results of the hierarchical regression analysis for variables predicting speeded reading.

  Block 1 / Step 1 Block 2 / Step 2
  B SE β t p B SE β t P
(Intercept) -54.23 26.21 -2.07 0.05 -19.86 29.15 -0.68 0.50
Age 3.66 1.29 0.34 2.84 0.01 3.22 1.23 0.30 2.63 0.01
Non-verbal IQ 0.48 0.45 0.16 1.06 0.30 0.35 0.43 0.11 0.81 0.42
Morphological Awareness 0.23 0.31 0.11 0.75 0.46 0.10 0.29 0.05 0.33 0.75
Phonological Awareness 0.24 0.35 0.09 0.68 0.50 0.19 0.33 0.07 0.58 0.57
RAN 0.58 0.12 0.65 4.74 < .001 0.49 0.12 0.55 4.09 < .001
Perceptual fluency for Digits -0.27 4.11 -0.01 -0.07 0.95 4.93 4.53 0.17 1.09 0.29
Perceptual fluency for Characters -10.49 4.78 -0.38 -2.20 0.04
R2 0.654 0.707
ΔR2 0.654         0.053        

B and β stand for unstandardized and standardized beta respectively.

Table 4. Results of the hierarchical regression analysis for variables predicting non-speeded reading.

  Block 1 / Step 1 Block 2 / Step 2
  B SE β t p B SE β t P
(Intercept) -70.64 45.5 -1.55 0.13 -3.77 49.43 -0.08 0.94
Age 7.22 2.24 0.46 3.22 0.003 6.37 2.08 0.41 3.06 0.01
Non-verbal IQ 0.24 0.78 0.05 0.30 0.77 -0.02 0.72 -0.004 -0.03 0.98
Morphological Awareness 0.47 0.53 0.16 0.89 0.38 0.21 0.50 0.07 0.43 0.67
Phonological Awareness 0.40 0.60 0.10 0.66 0.51 0.30 0.56 0.08 0.55 0.59
RAN 0.48 0.21 0.37 2.26 0.03 0.31 0.21 0.24 1.53 0.14
Perceptual fluency for Digits -0.42 7.14 -0.01 -0.06 0.95 9.71 7.68 0.24 1.27 0.22
Perceptual fluency for Characters -20.41 8.10 -0.51 -2.52 0.02
R2 0.494 0.590
ΔR2 0.494         0.096        

B and β stand for unstandardized and standardized beta respectively.

Given the relatively small sample size, we examined whether the above models involved data overfitting by examining the predicted R-squared of the models. This method removes a data point from the dataset, generates the regression model and evaluates how well the model predicts the missing observation. Large discrepancy between the original R-squared value and the predictive R-squared value indicates that the model does not predict new observations as well as it fits the original dataset, and therefore suggests that overfitting might have occurred in the regression model. For example, an original R-squared of 0.5 and a predictive R-squared of 0.05 would mean that the main contribution to the original R-squared involves overfitting. We found that for speeded reading, the original R-squared was 0.707 and the predictive R-squared was 0.426. For non-speeded reading, the original R-squared was 0.590 and the predictive R-squared was 0.319. There was a considerable difference, yet the results were still largely generalizable.

Discussion

In this paper, we examined whether perceptual expertise with words predicts reading performance in developmental dyslexia. We observed that performances in both speeded and non-speeded reading tasks were significantly predicted by perceptual fluency for characters. Moreover, hierarchical regression analyses showed that perceptual fluency for characters uniquely explained an additional 5.3% and 9.6% variance in speeded and non-speeded reading task performance respectively, after controlling for the contributions of age, non-verbal IQ, phonological awareness, morphological awareness, RAN and perceptual fluency for digits. These findings established that reading performance can be predicted by ability to discriminate between visually similar word sequences. Such visual perceptual ability is an important predictor of reading performance in developmental dyslexia, above and beyond the contributions of the major factors that were well-evidenced to account for reading difficulties in developmental dyslexia. These findings have important theoretical implications to the understanding of and the interventions for developmental dyslexia in Chinese.

Perceptual expertise and word recognition

In contrast to some researchers’ rejection of visual perception as a possible cause of developmental dyslexia [101,102], the current findings highlight the important role of perceptual expertise with words in Chinese children with developmental dyslexia. It echoes with the recently renewed interest in understanding the visual factors in developmental dyslexia [103], adding perceptual expertise with words to the considerations of building a multifactorial model of developmental dyslexia [1417].

The lack of perceptual fluency in discrimination may impede the development of reading skills because of several reasons. First, the decreased perceptual fluency for characters likely indicates the failure to develop sufficient sensitivity to the diagnostic information of the words and characters during development, which leads to confusion between words with similar visual features or shapes. This perceptual deficit might have caused reading difficulty among children with developmental dyslexia. This perceptual bottleneck might lead to an even more significant problem in more advanced reading materials in which the number of similar visual alternatives tends to increase. For example, among the common Chinese characters for primary school students [51], the number of Chinese characters with the radical of ‘口’ (means ‘mouth’), e.g., 吵, 叫, 吃, 呢, 吸, etc., increases from 29 for Grade 1 students to 60 for Grade 2 students. The increasing number of similar visual alternatives makes visual word discrimination and identification more and more challenging.

Second, the lack of perceptual fluency in discrimination may pose limitations to the development of efficient mappings between units of the visual codes and their corresponding linguistic units, such as phonological or semantic units. Intuitively, when one cannot discriminate between similar visual codes effectively, attempts to associate the visual codes to linguistic units would be error-prone, e.g., linking the pronunciation of a Chinese character to two visual codes that are similar looking and hence confusable, or simply linking the pronunciation of a character to a wrong visual code. This may lead to difficulty in tasks that require one to retrieve these linguistic associations, including reading and RAN. This is consistent with the previous findings that individuals with developmental dyslexia have difficulty in learning associations between orthography and sound patterns [104].

Third, individuals with dyslexia have less reading experience needed to develop expertise compared with typically developing individuals given the same instruction or revision time. As a result, the accumulated amount of perceptual experience would lag behind typically developing children, resulting in a further enlarged gap in the perceptual fluency for words. In other words, a decreased perceptual expertise with words might be both a cause and an effect of reading impairment. Future studies can clarify the causal role of visual perceptual ability in developmental dyslexia by intervention studies [105]. Despite the fact that visual perceptual ability cannot fully account for developmental dyslexia, as shown by the well-documented phonological deficits in developmental dyslexia observed in pre-reading babies and toddlers [106108], our findings showed that taking visual perceptual ability into account would help capture and explain the individual variabilities in developmental dyslexia, which are often huge [109].

Given the current findings, the next important question is to clarify how perceptual expertise with words compares between children with dyslexia and typically developing children, and how this factor explains reading in these two groups. In the current study, the tasks were designed to cater for the range of abilities of children with dyslexia, some of which would have been too easy and resulted in ceiling effects with typical readers. Similarly, some of the tasks with appropriate difficulty levels for typical readers could often be too difficult for children with dyslexia, leaving their performance levels at the floor.

Despite this challenge in measurement, it is important to understand the role of perceptual expertise with words in explaining reading performance in different groups of readers, which helps clarify the contribution of perceptual expertise with words to reading in general. For example, it is possible that perceptual expertise with words is important for all readers such that it predicts reading for all readers including children with and without dyslexia and mature adult readers. This is consistent with the observation that children tend to shift away from phonological strategies to orthographic-based strategies when their reading skills advance [110], which suggests that perceptual fluency may become even more important when one learns to read more fluently. Alternatively, it could be the case that perceptual expertise with words is important at early phases of literacy development, but is not a good predictor of reading fluency among typically developing readers (as in the case of sensitivity to configuration of characters, [20]). Future studies should also extend and test whether perceptual expertise with words may explain the development of reading skills in different groups of readers.

In this study, we have demonstrated the importance of perceptual expertise with words in Chinese reading, while its importance for predicting reading performance in developmental dyslexia in other languages remains unknown. Further studies should clarify whether this factor is particularly important in Chinese reading, potentially because of its visual complexity and opaque nature, or that this factor is also important for other languages. For example, Indian languages involve hundreds of characters that are fewer than that in Chinese, but many more than that in alphabetic languages in general. It is possible that Indians may rely more heavily on perceptual expertise with words in reading compared to other alphabetic language users. In languages that involve the use of diacritics (e.g, Hebrew), fine-grained visual analyses are important such that perceptual expertise with words may also play an important role in reading. Hence it is possible that perceptual expertise with words may be a marker of reading development common to different languages.

It is recently clarified that object recognition ability can be explained by both domain-general and domain-specific abilities, both being independent from general intelligence [111].

In our study, both perceptual fluency for characters and for digits involve domain-general visual perceptual abilities, which could explain why they were both significantly correlated with reading skills (though with different strengths of the correlation coefficients; Table 2). However, it is important to note that the perceptual fluency for digits served as a control measure to capture domain-general object recognition ability such that, after partialling out its contribution, the perceptual fluency for characters represents ability specific to character recognition (Tables 3 and 4). Also, although the performance for digits was better than that for characters in general (Table 1), the QUEST estimation was performed by a monitor with very high refresh rate (1 ms per frame) which resulted in 1 ms per step of estimation. Given that both duration thresholds were very far from ceiling performance, and that both tasks showed very high discriminatory power, as indicated by the high reliability of both tasks (Table 1), it is unlikely that the discriminatory power of either task was constrained. In sum, our findings demonstrate that word reading skills in developmental dyslexia can be predicted by perceptual expertise with words, as a type of domain-specific abilities in object recognition.

Nonetheless, it is unclear to what extent domain-general abilities in object recognition explain word reading in developmental dyslexia. Recent studies have reported impaired performance in object recognition in adults with developmental dyslexia, suggesting deficit in object recognition is not limited to words but may be generalized to other object categories [4749,112; see also 111]. However, these findings are inconsistent in how ‘general’ the higher-level visual impairment is, i.e., whether one should expect it to be observed in all object categories in general [48], or only in some [49]. Also, it is unclear how these visual deficits in other object categories are related to their word reading abilities, e.g., whether these general high-level visual deficits cause word reading deficits or whether these general deficits are independently caused by other conditions of the participants. This question would be directly relevant to the formulation of intervention strategy (see below), e.g., whether the intervention should be specific to words or general to multiple object categories. In the current study, the hierarchical regression results showed that perceptual fluency for digits did not predict word reading performance, while perceptual fluency for characters did (Tables 3 & 4). Since domain-general object recognition skills should be captured by both perceptual fluency measures, we did not observe any evidence for the role of domain-general skills in uniquely predicting word reading in multiple regression analyses. However, it is important to note that this study was not designed to look at domain-general object recognition skill in the sense that only one control object category (digit) was included. Ideally, one should include more categories of objects to better capture domain-general object recognition skills, and therefore this should be clarified by future studies.

Task demands of the perceptual fluency task and reading

Although the perceptual fluency task focused on the ability to visually discriminate between words, it is worth noting that this task is unlikely a purely visual task. For example, since the stimuli were real-world Chinese characters, existing associations between the visual form of the characters and other linguistic representations (e.g., phonological, morphological or semantic units) might have contributed to their performance. Also, the stimuli were Grade 1-level characters that participants should have learned previously, and therefore visual and verbal memory might have contributed to their judgment. Do these make the visual fluency task essentially a ‘reading’ task such that the findings reported here simply showed ‘reading predicts reading’?.

Several considerations would help inform this discussion. First, one should clarify the exact definition of ‘reading’ that one adopts. If ‘reading’ refers to recognition of the visual form of the word only and does not consider any subsequent linguistic processes, then this would be highly similar to the visual discrimination ability measured by the perceptual fluency task. Alternatively, if ‘reading’ requires one to correctly extract the pronunciation and/or the semantics of the words based on the visual code, then regardless of whether the ‘reading’ is done aloud or silently, this is critically different from the perceptual fluency task which did not explicitly examine the accuracy of responses in these linguistic domains.

Second, with the use of real-world and learned Chinese characters, one might have activated existing associations between the visual form of the characters and other linguistic units during the perceptual fluency task. In this sense, the perceptual fluency task appears to be highly similar to ‘reading’ since linguistic information can potentially play a role. In response to this, one should consider the fact that word recognition involves many complicated cognitive processes that tend to interleave with each other. Readers likely rely on multiple processes to solve most tasks and therefore a single process cannot be truly isolated. To investigate a specific cognitive process, a more effective way is to use specific task designs combined with control measures to emphasize the interested process and minimize the contribution of others.

For the perceptual fluency measure, visual discrimination ability was emphasized using a sequential matching paradigm with speeded presentation. To perform well, one’s perceptual analyses of the character sequences within the brief presentation time must be sufficiently adequate to differentiate it from the distractor, otherwise any subsequent linguistic processes could be error-prone and thus not helping. It is also worth noting that our use of four-character sequences imposed a much higher visual perceptual demand compared with other word recognition tasks that also employed speeded presentation with one single word [e.g., 113]. When the perceptual analyses are adequately performed, there is no way to stop one from activating the subsequent linguistic processes associated with the characters. However, the contribution of these linguistic processes to one’s performance has been minimized in the current design in several ways. One is that the task did not explicitly examine these linguistic responses, as discussed above, and therefore their existence and accuracy were not considered. This is in contrast to any ‘reading’ task that requires correct pronunciation or semantic retrieval of the words. The other is that, under brief presentation of the characters, the room for activating the linguistic processes is largely constrained. For example, with a presentation time of 300ms or below, it is not straightforward even for adult native readers to retrieve the verbal labels of all characters, at least within one’s awareness, before the characters disappear. In other words, with a faster presentation duration of the characters, the role of visual discrimination becomes more important while that of subsequent linguistic units becomes more limited. The last one is that the use of random character sequences removed any semantic contexts such that characters of any pronunciation or meaning could fit in well, and therefore the usefulness of linguistic information to inform correct responses was largely reduced.

Finally, given the use of learned characters, participants might have retrieved their verbal labels, that might have helped the retention of the stimuli during the time gap between the study and test images. In this case, one might suspect that the current findings simply reflected verbal short-term memory, as a common product of ‘reading’, instead of perceptual fluency with characters. However, this is unlikely because of several reasons. First, the time gap between the study and test images was merely 500ms. This largely limited the amount of memory decay during the retention period [114], and therefore one’s working memory ability should have minimal influence on his or her performance. Second, while we acknowledge the potential contribution of verbal labels to one’s performance, this was shared by the perceptual fluency task with digits and should therefore be partialled out in the multiple regression analyses.

In sum, the perceptual fluency task is not simply another ‘reading’ task that requires correct access to linguistic information. While the perceptual fluency task may involve many cognitive processes, the critical task demand that determines one’s performance is visual discrimination, especially when the stimulus is presented with brief durations. The contribution of other cognitive abilities, including that of short-term memory, is further tempered with the control measures, in particular the perceptual fluency with digits, which should have partialled out the contribution of other major cognitive abilities known to contribute to reading performance with dyslexia.

Comparing perceptual expertise and other cognitive skills

It is important to note that perceptual expertise with words does not simply refer to general visual ability as in other visual accounts of developmental dyslexia [9,10,12,13]. In our study, the perceptual fluency task for digits served as a control measure, which required an identical task and visual attentional span as the perceptual fluency task for characters. Hence, the unique contribution by perceptual fluency for characters is not simply domain-general visual abilities [115], sensorimotor abilities [101], visual distortions, illusions or fatigues that stemmed from visual stress in general [116]. Instead, it involves high-level visual processes in differentiating between similar visual objects within a category, in this case, Chinese characters. This subordinate-level visual categorization typically engages the ventral visual pathway [33,34,3744,54,55], in contrast to the magnocellular-dorsal theory of developmental dyslexia [9,10].

Perceptual expertise with words does not simply reflect the degree of visual crowding experienced by readers [13,117]. It has been demonstrated that perceptual expertise typically leads to alleviation of visual crowding that is specific to the expert object category, but not with other untrained shapes [63,86]. Importantly, the perceptual fluency task with digits were identical to that with characters in terms of how the stimuli were spatially presented, and therefore general visual crowding effects that is constrained by eccentricity should have been partialed out in our regression model by the perceptual fluency task with digits [118]. Furthermore, perceptual fluency with characters is critically different from visual crowding. In a visual crowding task, participants are typically told to report the target and ignore the nearby distractors [85,118,119], while the perceptual fluency tasks in our study required participants to attend to all items. This task difference is critical, because training focused on visual crowding did not lead to improved reading in a past study [120], while trainings focused on recognition of all of the presented letters did [83,93]. These suggest that the development of perceptual expertise involves skills to address the visual crowding problem, but visual crowding per se cannot fully explain the visual skills involved in perceptual expertise. Note that the visual crowding discussed here largely concerns the between-object visual crowding, that is largely constrained by eccentricity [85,118,119,121]. Additional factors may affect the exact degree of visual crowding experienced with a specific stimulus category, such as target complexity, flanker complexity, the similarity and complexity differences between target and flankers, and self-crowding [122124]. How these factors affect the category-specific visual crowding experienced during the perceptual fluency measure with characters and digits is unclear, and further studies should examine this and clarify its relationship with perceptual expertise.

Although perceptual expertise with words is conceptually different from visual attention span (see Introduction), it is interesting to further evaluate the details of the tasks used to measure visual attention span and perceptual expertise with word, because these specifics would determine how much the actual measurements overlap. For example, measures of visual attention span often apply very brief presentation duration (e.g., 200ms) to measure what participants can perceive without making additional saccades [58,59,66]. This would make sure that participants are simultaneously processing several visual elements (or ‘multi-element processing’) [125]. However, in perceptual expertise measures, individuals are highly varied in terms of how fast they can recognize the stimuli [25,37]. Hence, only the top experts can demonstrate excellent perception with single visual fixation (and hence meeting the requirement of tasks of visual attention span), while it is possible for other participants to rely on multiple saccades for recognition. Moreover, visual span measures that involve discrimination of highly similar shapes or objects (e.g., using Chinese characters for studying Chinese reading or letters for studying French reading) [126,127], or that involve visual crowding [65] would involve visual discrimination skills central to perceptual expertise, while perceptual expertise measures that present stimuli over a large visual span would reflect skills critical to visual span measures. In the current study, the perceptual fluency task for digits served as a control measure, which required an identical visual attention span as that for characters, and therefore the contribution of visual attention span [12,65] should have been partialed out by the perceptual fluency task for digits in our regression model.

Considering perceptual expertise with words may help inform seemingly contradictory findings in studies of visual attention span. Children with dyslexia showed deficits in visual attention span for letters and digit strings but not for symbol strings, and such category selectivity of the deficit has been interpreted as the deficit in symbol-sound mapping that children had acquired for letters and digits, but not about other visual problems such as dysfunctions of the visual word form system [59]. In another study, however, children with dyslexia showed deficit in visual attention span that was similar regardless of whether the stimulus was nameable or non-nameable, which suggested that the deficit is visual but not verbal in nature [58]. Considering the factor of perceptual expertise provides a novel angle to these seemingly contradictory findings. Since children typically have much more experience with recognizing letters and digits than with symbols, a possible alternative explanation of the findings in Ziegler et al. [59] was that their participants had higher perceptual expertise with letters and digits than with symbols. Further studies may directly test this alternative explanation to clarify this issue.

Phonological awareness and word recognition

It is interesting to observe that phonological awareness did not predict either speeded reading or non-speeded reading (Table 2). Phonological awareness refers to the awareness of and access to the sounds of one’s language [128,129]. While the deficit in phonological awareness is regarded as a major cause of developmental dyslexia for alphabetic languages [3,4], its role in developmental dyslexia in logographic languages such as Chinese is less clear. Earlier studies reported that phonological skills predicted reading performance in typically developing Chinese children [129,130]. However, similar findings were not observed in a subsequent large-scale study [131].

Findings regarding its contribution to developmental dyslexia were also mixed. For example, while about 30% of Hong Kong Chinese children with dyslexia showed deficits in phonological skills, the unique contribution of phonological awareness did not reach significance when other factors were included in hierarchical regression models [109]. Phonological awareness also failed to distinguish children with developmental dyslexia and typically developing children using logistic regression [89]. However, in a longitudinal study using logistic regression, phonological processing during the 3rd year in kindergarten predicted dyslexia outcome a year later in Grade 1 [132]. While phonological awareness remains to be important for Chinese language learning for some researchers [89], other researchers considered phonological awareness as less important and not one of the ‘core problems’ in Chinese developmental dyslexia [109,126].

Our results showed that phonological awareness did not correlate with either speeded or non-speeded reading performance, consistent with the previous findings [109]. Given the good reliability of and the absence of ceiling or floor effects in our phonological awareness measure, our findings support the idea that phonological awareness skills may not be an important unique predictor of reading performance among Chinese children with developmental dyslexia.

Morphological awareness and word recognition

It is also interesting to observe that morphological awareness did not correlate with speeded reading and non-speeded reading performances (Table 2). Morphological awareness refers to the awareness of, the ability to reflect on and the ability to manipulate the structure of the smallest meaningful units, i.e., morphemes, in words [133]. In recent years, morphological awareness has been proposed to be a core theoretical construct for explaining Chinese reading abilities [89]. Supporting evidence comes from its ability to predict Chinese character recognition in typically developing children [134], to longitudinally predict Chinese character recognition in typically developing children [131], and to distinguish between children with dyslexia from age-matched controls [88; but see 135].

To understand our seemingly inconsistent finding with these evidences, it might be useful to consider the task demands of the morphological awareness measures. In the current study, two subtasks were used. One was the concept production task, which tapped onto how well children understand the morphological structure of the multi-character words. This morphological structure provides very useful hints for the gist of the meaning of multi-character words (e.g., whether it is a type of flower, a type of fish, or a type of machine), and therefore enhancing the semantic transparency of the words [131]. This is a relatively easy task (mean = 16.0 out of 19 points, SD = 3.47, range = 2–19) and is commonly used with younger children. We adopted this easy task to pick up the variance in morphological awareness in relatively weaker readers among children with dyslexia.

The other task was a word production task, which was relatively more difficult (mean = 8.71 out of 15, SD = 3.31, range = 1–15) and was often adopted for older children. To perform well, one needs to fulfill two task demands: to differentiate whether the homophones in different two-character words were the same or different characters, and whether these homophones carried the same meaning or not. To achieve these, it is helpful to retrieve and discriminate between the visual codes of the target character based on the pronunciation. Given the large number of homophones in Chinese language, the ability to discriminate between the retrieved visual codes becomes even more helpful. This hypothesis is supported by the significant correlation between morphological awareness and perceptual fluency for characters (Table 2).

In other words, the morphological awareness measures in the current study included two tasks, each tapped onto different aspects of morphological skills with different difficulty levels. Our participants were heterogeneous in terms of their abilities in word reading and in different aspects of morphological skills, as demonstrated by the huge range of performance in each task. It is possible that the previously observed relationship between morphological awareness and reading could be observed more easily in a relatively more homogeneous sample (e.g., among typically developing children; e.g., [138,141]), or in categorizing participants into the dyslexic and control groups which could be more robust given the heterogeneity of the data [89]. This is a possible account of the inconsistent findings and should be examined in further studies.

Visual training as a potential intervention strategy

Demonstrating the role of perceptual expertise with words in developmental dyslexia sheds light onto a possible direction for intervention of developmental dyslexia. It is well-established that visual training in the laboratory can effectively and efficiently improve high-level visual processing [21,22,43,74,86,136138]. These training paradigms are typically computerized, involving various visual judgments such as naming, discrimination or visual search. The course of the training is typically carefully and gradually tuned such that the difficulty level of visual judgments becomes more challenging with time, e.g., by introducing more visually challenging stimuli, or by requiring faster responses within a shorter time window.

Importantly, the human visual system is very sensitive and responsive to visual training, which has been shown to work well in various populations including typical adults [21,22,43,74,86,136138], individuals with visual impairment [e.g., 139], and the elderly [e.g., 140]. It also works well with different kinds of objects including faces [141], words [142], musical notation [86,88], and various computer-generated novel objects [21,22,43,143]. Visual training works well even when the to-be-learned perceptual signal is task-irrelevant and/or unconscious [74,144,145]. Lab visual training often leads to significantly improved visual skills within 8–10 hours of training, and is accompanied with large-scale neural changes in the occipitotemporal cortex and other brain regions [4245,142,143]. Therefore, it is reasonable to expect that visual training may also help children with developmental dyslexia improve visual discrimination of words and therefore develop their perceptual expertise with words. As discussed above, this may help these children improve their reading performance by enhancing their efficiency in discriminating between visually similar words, strengthening the association of words with their linguistic units based on more accurate representation of the visual codes, and alleviating the vicious cycle between reduced perceptual fluency and reduced reading experience.

The task demand of visual perceptual training is shared with some existing intervention strategies. For example, COREVA, a visual attention span intervention for children with dyslexia which has gained empirical evidence for its effectiveness, involves training the fine-level visual discrimination skills, which is central to perceptual expertise development [127,146]. COREVA included three tasks, visual search and discrimination, visual matching and visual parsing. For visual search and discrimination, participants were required to identify targets among distractors in which “their visual similarity (between targets and distractors) was typically high” [p. 130, 127]. For visual matching, participants were required to perform a simultaneous matching task–to identify whether two strings of letters, drawings or symbols were identical or not as accurately and as fast as possible. This was highly similar to the perceptual fluency task except for the sequential versus simultaneous presentation of the stimuli, and similar perceptual training protocols have been shown to enhance perceptual expertise [74,86]. For visual parsing, participants were required to search for bigrams or trigrams in a long string of letters as fast as possible. This required participants to recognize a specific combination of letters among other highly similar letters, and again essentially training up one’s ability to discriminate between highly similar visual objects. In sum, COREVA presents stimuli of letters and highly similar symbols rapidly with the requirement of speeded responses, which essentially improves users’ perceptual expertise in addition to other skills.

It is also important to note that perceptual expertise training is different from that focused on low-level visual perceptual training [147]. In this study, training on visual texture discrimination improved reading performance in logographic language users with developmental dyslexia [147]. Texture discrimination typically involves judging basic visual features as line orientation over a large visual field covering the visual periphery, and are often referred to as early visual processes engaging the primary visual cortex [148,149]. In contrast, perceptual expertise with words typically focuses on a few characters presented at the fovea, and these processes are referred to as higher or late visual processes engaging the more downstream visual areas corresponding to shape and object recognition [33,34,3644,54,55]. While it was difficult to pinpoint what visual skill(s) involved in texture discrimination caused the reading improvement, this type of training likely addresses different types of visual bottleneck of the reading deficit in developmental dyslexia, in contrast to the high-level and domain-specific perceptual training discussed here.

The perceptual training discussed here is also different from general cognitive and perceptual training. It has been proposed that action video game (AVG) training can improve reading performance in developmental dyslexia [150,151]. While this type of training may improve general visual attention and cognitive functions, it does not involve the use of words or characters in the training. Therefore, it is unlikely that AVG training can help fine-tune the high-level perceptual representations of words in the visual system, or lead to improved visual judgments among similar visual instances of words and characters. In contrast, the perceptual training proposed here directly involves words and characters, and participants are required to discriminate among visually similar alternatives. Therefore, the action video game training and the perceptual fluency training may be complementary to each other to target on different types of deficits observed in children with developmental dyslexia.

The above discussion highlights how visual perceptual training is similar or different from other intervention strategies. Indeed there are many more intervention strategies that have been proposed, and some were evaluated systematically [152]. An implication of the current paper concerns perceptual fluency training as a possible intervention strategy that might have unique contribution to improving reading in children with dyslexia. However, it is not our goal to propose that perceptual fluency training is superior to other types of training. Instead, we believe that developmental dyslexia has multiple potential causes, and hence effective intervention likely involves multiple strategies. Discriminating between visual codes effectively is one of the fundamental skills in reading, which supports the development of other multimodal skills underlying effective reading. Hence perceptual fluency training could potentially be combined with other trainings to provide more comprehensive intervention for children with dyslexia.

Acknowledgments

This work should be corresponded to Y.W. at yetta.wong@gmail.com or at RM 308, Ho Tim Building, The Chinese University of Hong Kong, Hong Kong, or A. W. at alanwong@cuhk.edu.hk or at 334, Sino Building, The Chinese University of Hong Kong, Hong Kong.

Data Availability

All data are available in OSF DOI 10.17605/OSF.IO/DCTH6.

Funding Statement

This work was supported by the Language Fund under Research and Development Projects 2018-19 of the Standing Committee on Language Education and Research (SCOLAR), Hong Kong SAR (to Y.W.), and the Direct Grant for Research at the Chinese University of Hong Kong (to Y.W.). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

References

Decision Letter 0

Athanassios Protopapas

23 Mar 2020

PONE-D-20-07055

Visual discrimination skills predict Chinese reading performance among Hong Kong Chinese children with developmental dyslexia

PLOS ONE

Dear Dr. Wong,

Thank you for submitting your manuscript to PLOS ONE. To save everyone's time, I screen incoming manuscripts before sending them out for review. There are two issues that prevent me from considering this submission further as it stands:

First, your submission does not adhere to the data sharing guidelines of PLOS ONE. You only indicate that data will be posted after acceptance and that there will be restrictions to access. However, minimal data must be included with the manuscript, provided with submission, unless extraordinary (fully justified) circumstances prevent this, and any foreseen restrictions to accessing all the data upon publication must be fully documented and justified. Please consult the PLOS ONE data policy (https://journals.plos.org/plosone/s/data-availability) to ensure full compliance before resubmission.

Second, I am worried about the terminology used in your submission, and in particular with respect to the critical task, which may turn out to be misleading for readers in the field, as it is used in the title as well as the body of the manuscript. Specifically, you talk about "visual discrimination" where in fact you task is neither "visual" in the commonly encountered sense nor "discrimination". Of course it is visual in the sense that one must see the details of the characters correctly in order to perform the task. But the term "visual" is typically reserved for skills that are not specific to certain kinds of stimuli, as this is commonly understood to be a generic designation. Instead, your findings concern the visual processing of characters, and indeed beyond performance with digits, therefore squarely excluding any generic "visual" component and instead concerning the processing of orthographic material. As expected based on the effects of experience and as your previous published work demonstrates, perceptual expertise with characters covaries with reading skill, so there is nothing surprising or problematic about that. Character recognition expertise is a marker of reading proficiency, and it would be very surprising if dyslexia was not associated with more difficulty in recognizing, retaining, and matching characters.

in your study, as I understand it, the task concerns the identification of characters and indeed includes a memory component, further distancing from what might be considered a strictly "visual" task, as participants must hold the character sequence in memory in order to respond in the two-alternative-forced-choice setup of the task. Which brings me to the second term, namely "discrimination". In a discrimination task one typically must distinguish between two or more stimuli (or aspects thereof), usually (but not always) responding whether they are the same or different. Of course more complex discrimination arrangements do exist, but I find it difficult to classify your task as discrimination when the participant must remember a character (or digit) sequence and subsequently match it to a displayed array of characters. This sounds like an identification task to me. If you disagree, please justify your choice of terms in the revision, so that reviewers will be clear about how the terms are used.

If the above comments indicate that I have misunderstood your task or some critical aspect of it, please clarify your description to avoid similar misunderstanding by others. Finally, please note that "novelty" or perceived theoretical importance are not considered as publication criteria for PLOS ONE, so you do not need to feel any pressure to establish any of these; please set up the rationale and terminology of the study to be more precisely in line with the tasks used.

We would appreciate receiving your revised manuscript by May 07 2020 11:59PM. When you are ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter.

To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

Please include the following items when submitting your revised manuscript:

  • A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). This letter should be uploaded as separate file and labeled 'Response to Reviewers'.

  • A marked-up copy of your manuscript that highlights changes made to the original version. This file should be uploaded as separate file and labeled 'Revised Manuscript with Track Changes'.

  • An unmarked version of your revised paper without tracked changes. This file should be uploaded as separate file and labeled 'Manuscript'.

Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

We look forward to receiving your revised manuscript.

Kind regards,

Athanassios Protopapas

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements:

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at http://www.plosone.org/attachments/PLOSOne_formatting_sample_main_body.pdf and http://www.plosone.org/attachments/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Please provide additional details regarding participant consent, both for the children and their parents or guardians. In the ethics statement in the Methods and online submission information, please ensure that you have specified what type you obtained (for instance, written or verbal, and if verbal, how it was documented and witnessed).

3. We note that you have stated that you will provide repository information for your data at acceptance. Should your manuscript be accepted for publication, we will hold it until you provide the relevant accession numbers or DOIs necessary to access your data. If you wish to make changes to your Data Availability statement, please describe these changes in your cover letter and we will update your Data Availability statement to reflect the information you provide.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files to be viewed.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org. Please note that Supporting Information files do not need this step.

PLoS One. 2021 Jan 22;16(1):e0243440. doi: 10.1371/journal.pone.0243440.r002

Author response to Decision Letter 0


12 May 2020

[The following content has also been uploaded as the attachment of 'Response To Reviewer']

Dear Prof. Protopapas,

Thank you for your constructive comments. We have now clarified our wordings hoping to minimize confusion among researchers from different backgrounds.

We sincerely hope that now you would find the manuscript suitable for the review process.

Thank you very much for your time.

Yetta Wong & Alan Wong

*********************

PONE-D-20-07055

Visual discrimination skills predict Chinese reading performance among Hong Kong Chinese children with developmental dyslexia

PLOS ONE

Dear Dr. Wong,

Thank you for submitting your manuscript to PLOS ONE. To save everyone's time, I screen incoming manuscripts before sending them out for review. There are two issues that prevent me from considering this submission further as it stands:

First, your submission does not adhere to the data sharing guidelines of PLOS ONE. You only indicate that data will be posted after acceptance and that there will be restrictions to access. However, minimal data must be included with the manuscript, provided with submission, unless extraordinary (fully justified) circumstances prevent this, and any foreseen restrictions to accessing all the data upon publication must be fully documented and justified. Please consult the PLOS ONE data policy (https://journals.plos.org/plosone/s/data-availability) to ensure full compliance before resubmission.

RESPONSE: Thank you for this suggestion. During the preparation of the dataset, we identified errors in a small subset of manually input data, which led to minor adjustments of the reported numbers (mostly in the tenths and hundredths decimal places). These did not affect the major pattern of the results. We have now corrected these errors with tracked changes. We have also included the dataset ready to be shared upon acceptance of the manuscript, which is available here:

https://osf.io/dcth6/?view_only=5ab8b6d793b54f24b5fcbf5471bed83c

*****

Second, I am worried about the terminology used in your submission, and in particular with respect to the critical task, which may turn out to be misleading for readers in the field, as it is used in the title as well as the body of the manuscript. Specifically, you talk about "visual discrimination" where in fact you task is neither "visual" in the commonly encountered sense nor "discrimination". Of course it is visual in the sense that one must see the details of the characters correctly in order to perform the task. But the term "visual" is typically reserved for skills that are not specific to certain kinds of stimuli, as this is commonly understood to be a generic designation. Instead, your findings concern the visual processing of characters, and indeed beyond performance with digits, therefore squarely excluding any generic "visual" component and instead concerning the processing of orthographic material. As expected based on the effects of experience and as your previous published work demonstrates, perceptual expertise with characters covaries with reading skill, so there is nothing surprising or problematic about that. Character recognition expertise is a marker of reading proficiency, and it would be very surprising if dyslexia was not associated with more difficulty in recognizing, retaining, and matching characters.

RESPONSE: In the initial draft of the manuscript, we thought some of the reading researchers may not care about the ‘perceptual nature’ of this task and therefore we described the task as ‘visual’ for simplification.

Based on your comments, we realized that this simplification could actually cause confusion for other researchers who care about these differences.

Now, we called this factor ‘perceptual expertise with words’ (or ‘perceptual expertise with characters’ when referring to Chinese characters) throughout the manuscript, and the task ‘perceptual fluency task’.

*****

in your study, as I understand it, the task concerns the identification of characters and indeed includes a memory component, further distancing from what might be considered a strictly "visual" task, as participants must hold the character sequence in memory in order to respond in the two-alternative-forced-choice setup of the task. Which brings me to the second term, namely "discrimination". In a discrimination task one typically must distinguish between two or more stimuli (or aspects thereof), usually (but not always) responding whether they are the same or different. Of course more complex discrimination arrangements do exist, but I find it difficult to classify your task as discrimination when the participant must remember a character (or digit) sequence and subsequently match it to a displayed array of characters. This sounds like an identification task to me. If you disagree, please justify your choice of terms in the revision, so that reviewers will be clear about how the terms are used.

If the above comments indicate that I have misunderstood your task or some critical aspect of it, please clarify your description to avoid similar misunderstanding by others. Finally, please note that "novelty" or perceived theoretical importance are not considered as publication criteria for PLOS ONE, so you do not need to feel any pressure to establish any of these; please set up the rationale and terminology of the study to be more precisely in line with the tasks used.

RESPONSE: As described above, we now avoid describing the underlying factor as ‘discrimination’ by referring it to ‘perceptual expertise with characters’. Hopefully this would be more accurate and minimize confusion for researchers with different backgrounds.

Attachment

Submitted filename: ResponseLetter_v1.docx

Decision Letter 1

Athanassios Protopapas

6 Jul 2020

PONE-D-20-07055R1

Perceptual expertise with Chinese characters predicts Chinese reading performance among Hong Kong Chinese children with developmental dyslexia

PLOS ONE

Dear Dr. Wong,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

In particular, the reviewers are unanimous in their overall positive evaluation, as well as in their evaluation that your manuscript only partially fulfills the essential criterion of being technically sound with conclusions fully supported by the data, and they provide several constructive suggestions for improvement on this front. Although there are several points raised, and many additional sources of information recommended to be taken into account, it seems possible to me that a revised manuscript may be able to address these criticisms, and I would therefore like to give you an opportunity to do that in a revision. 

Please submit your revised manuscript by Aug 20 2020 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

  • A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

  • A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

  • An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Athanassios Protopapas

Academic Editor

PLOS ONE

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: (No Response)

Reviewer #2: All comments have been addressed

Reviewer #3: (No Response)

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Partly

Reviewer #2: Partly

Reviewer #3: Partly

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: I appreciate the opportunity given or reviewing this paper which can potentially contribute to this society. However, there are a few concerns I have as below.

1. Literature review: The authors mentioned "children with developmental dyslexia fail to develop perceptual expertise with word stimuli." on p.5. However, this is quite an under debate. This issue is about the argument whether dyslexic children have talents in their visuospatial abilities which have been discussed for a long time in either alphabetic languages (e.g., Brunswick, Martin, & Marzano, 2010 ) or Chinese (e.g., Wang & Yang, 2013). It's more about whether it focuses on detailed or gross information of the visual stimuli. So, the authors are suggested to make a more solid argument here with appropriate citations.

Brunswick, N., Martin, G. N., & Marzano, L. (2010). Visuospatial superiority in developmental dyslexia: Myth or reality?. Learning and Individual Differences, 20(5), 421-426.

Wang, L. C., & Yang, H. M. (2011). The comparison of the visuo-spatial abilities of dyslexic and normal students in Taiwan and Hong Kong. Research in Developmental Disabilities, 32(3), 1052-1057.

2. Research questions: The two research questions are quite the same.

3. Research questions: The authors mentioned "Ceiling effects could easily result if the tasks were performed by typically developing children. Hence comparing the role of perceptual expertise with words in normal readers and readers with dyslexia was out of the scope of this study." on p.9. However, the claimed ceiling effect should have proof, otherwise, it is not convincible. For the research in this field, it's not normal to find the study without a reference group, especially the targeted issue isn't a very popular and well-accpted one, as deficient orthographic knowledge, in Chinese contexts.

4. Methodology: Another key flaw in the design of this study is the lack of taking visual perception into consideration, which is considered to be crucial to Chinese reading (e.g., Meng et al., 2011) as well as one of the core deficits of Chinese dyslexia (e.g., Ho et al., 2004).

Meng, X., Cheng-Lai, A., Zeng, B., Stein, J. F., & Zhou, X. (2011). Dynamic visual perception and reading development in Chinese school children. Annals of Dyslexia, 61(2), 161-176.

Ho, C. S. H., Chan, D. W. O., Lee, S. H., Tsang, S. M., & Luan, V. H. (2004). Cognitive profiling and preliminary subtyping in Chinese developmental dyslexia. Cognition, 91(1), 43-75.

5. Methodology (Perceptual fluency): Although the meanings between original stimulus and replacing one were checked, the visual similarities are also matter. Considering the importance of visual modality in Chinese character reading, I expect a prior examination of the stimuli used like Liu, Chen, and Chung (2015) did.

Liu, D., Chen, X., & Chung, K. K. (2015). Performance in a visual search task uniquely predicts reading abilities in third-grade Hong Kong Chinese children. Scientific Studies of Reading, 19(4), 307-324.

Reviewer #2: The manuscript presents a study investigating the relation between perceptual expertise with Chinese characters and Chinese word reading in Hong Kong Chinese children with dyslexia. A task measuring perceptual expertise in Chinese character processing, with an adapted visual presentation duration is used to test the individual threshold at which a string of Chinese characters can be discriminated. Individual performance on this task correlated both with speeded and non-speeded reading of Chinese words presented in lists. Hierarchical regression analyses showed that performance on the perceptual expertise task also explained variance in speeded and non-speeded reading after taking into account age, non-verbal IQ, phonological awareness, morphological awareness, rapid automatized naming and performance on the perceptual expertise task but using strings of digits rather than Chinese characters. The authors conclude that perceptual expertise with words plays an important role in Chinese reading and that perceptual training is a potential route to remediation.

2. Is the manuscript technically sound, and do the data support the conclusions?

Based on my reading of the manuscript the experiments seem to have been conducted rigorously and including appropriate sample sizes. Compared to other studies on dyslexia I expected to also see data from a control group, yet conclusions can be drawn based on this dataset alone. I responded that the conclusions are partly supported by the data mainly because there are some aspects of the manuscript which in my opinion could be clarified and further discussed considering the existing literature. There is a large literature on dyslexia and though it is impossible to cover all these in a manuscript it seemed at times that there were some missing links between the perceptual expertise literature (in music etc.) and the dyslexia literature. In the following paragraphs I outline three main points I consider could be improved and some additional minor comments or questions. I enjoyed reading the manuscript and I hope the reviews will be of use to the authors.

1. I am not completely sure that based on the provided description and arguments I have a clear understanding of the (a) concept of perceptual expertise and its limits, (b) to what point perceptual expertise can be considered a potential cause of dyslexia rather than a consequence of less reading experience, and (c) the distinction between perceptual expertise and orthographic processing.

(a) Overall I understand that the authors consider perceptual expertise a domain-specific ability, distinguishing their proposal from other visual theories of dyslexia and supporting this with the reported differences in the contribution of the digit as compared to the character perceptual fluency tasks. On page 26 when the authors discuss differences in the digit and character perceptual fluency tasks they consider that the unique contribution of the character task to word reading after accounting for performance on the digit task indicates that the character perceptual fluency task reflects a domain-specific skill (suggesting again it is a consequence of reading experience?). While I understand the logic and appreciate the inclusion of the digit version of the task, I think there are potential limitations to this reasoning. Digits are fewer and less visually complex than the Chinese characters, thus it could be expected that acquiring perceptual expertise is less challenging and that the task might be inherently easier. Since similar patterns of correlations were found between these two tasks and reading skills (albeit stronger for the character than the digit span) could it be that the difficulty in the task with characters doesn’t tap into a different domain but has better discriminatory power? A secondary note is that if this skill is considered domain specific then I am not sure it can be reconciled with the results of the visual texture training study that led to improvements in reading (page 5).

(b) I was not sure whether the authors consider perceptual expertise only a consequence or also a potential cause of dyslexia. The example of the car expert on page 6 would suggest that reading experience alone might result in better perceptual fluency (as a car expert becomes particularly good at discriminating cars because they spend a lot of time looking at cars). In the discussion the authors do suggest that it could be both a cause and consequence of dyslexia, but I am not sure it is clear how it could be a cause.

(c) The authors suggest that perceptual expertise is not related to orthographic processing and I believe they consider that it does not rely on knowing the mappings between characters and linguistic units. On the other hand, the authors acknowledge (page 23) that reading experience can improve perceptual expertise and that “perceptual fluency may become more important when one learns to read more fluently” (page 25). This is also the case for orthographic processing which becomes more important after readers of alphabetic orthographies have moved beyond decoding and start processing multiple letters and larger orthographic units. This can also affect letter processing in tasks that are not reading. Indeed, reading experience allows readers of alphabetic orthographies to also become better at identifying letters in words (word superiority effect) in the Reicher-Wheeler paradigm. I was wondering why this is not considered to be the case in these perceptual fluency tasks. On a related note, on page 5 the authors mention that differences in orthographic depth (additionally to those of character visual complexity) could also lead readers to rely more on the visuo-orthographic structure of the visual codes. Would this also support that this perceptual expertise is not just visual but is related to the mappings between characters and linguistic units and is more like orthographic processing than suggested?

2. The authors link the literature on perceptual expertise and training perceptual expertise. I am not very familiar with this literature so when reading the manuscript I found myself thinking about the multi-element processing aspect of these tasks (also found in RAN tasks) and visual attention span studies that I am more familiar with. Indeed the paradigm used in this study, that uses different presentation durations depending on performance, clearly differentiates it from visual attention span tasks (in which it is set at around 200 ms to allow a single fixation on the string). Nevertheless, it seems that the results of some visual attention span studies could inform the interpretation of those presented in this manuscript and strengthen the discussion. I mention some of those with similar paradigms and others in children with dyslexia learning to read in Chinese (as far as I know the latter use a visual 1-back paradigm) in case they are of interest. In case the authors disagree with this this view perhaps the studies would still allow them to explain more specifically what their own assumptions are and how they differ from other visual theories of dyslexia. Each of the visual theories of dyslexia mentioned in the manuscript differ greatly (some are visual only, other auditory and visual, other related to magnocellular processing), so I believe it is difficult to set a new theory apart from all of the previous theories without considering the other theories in more depth. They could also discuss whether they consider the aspect of multi-element processing plays a role in their paradigm.

Lobier, M., Zoubrinetzky, R., & Valdois, S. (2012). The visual attention span deficit in dyslexia is visual and not verbal. Cortex, 48(6), 768-773.

Ziegler, J. C., Pech‐Georgel, C., Dufau, S., & Grainger, J. (2010). Rapid processing of letters, digits and symbols: what purely visual‐attentional deficit in developmental dyslexia?. Developmental Science, 13(4), F8-F14.

Valdois, S., Peyrin, C., Lassus-Sangosse, D., Lallier, M., Demonet, J. F., & Kandel, S. (2014). Dyslexia in a French–Spanish bilingual girl: behavioural and neural modulations following a visual attention span intervention. Cortex, 53, 120-145.

Zhao, J., Liu, M., Liu, H., & Huang, C. (2018). Increased deficit of visual attention span with development in Chinese children with developmental dyslexia. Scientific reports, 8(1), 1-13.

Chen, N. T., Zheng, M., & Ho, C. S. H. (2019). Examining the visual attention span deficit hypothesis in Chinese developmental dyslexia. Reading and Writing, 32(3), 639-662.

Regarding training, there is a visual attention span training program that might also be of interest (COREVA® training program: Valdois et al., 2014) because to my knowledge it includes tasks similar to those suggested by the authors: visual discrimination, string matching. As far as I know a version in Chinese does not exist.

3. On page 26 the authors consider the possibility of perceptual fluency training and discuss studies focusing on improving visual processing skill and perceptual expertise in other domains. I think that it might also be helpful to explain how a potential improvement in perceptual fluency for character processing (without any training of mappings with linguistic units) could transfer to reading skills and whether/why this training could be superior to a phonological training or a training of character-sound associations.

The above were the major points related to the manuscript. I also have some more minor comments or questions that I mention below:

-In the final paragraph on page 7, the authors focus on the differences in processing in reading aloud vs perceptual expertise tasks and suggest that the latter does not involve mapping between characters and linguistic units. Is this really the case? I would assume that the depth of processing is likely to depend on the task (reading aloud, lexical decision, perceptual expertise) but not necessarily that the perceptual expertise task is only visual.

-On page 9 the authors mention that ceiling effects would result if the tasks were performed by typically developing readers. I was wondering why this is so since in the speeded naming there could still be variability in fluent readers and the non-speeded task items were chosen so that they would be appropriate for P1 to P5.

-On page 10, if no group differences were found between participants presented with List 1 and those presented with List 2 this could be reported.

-Do the authors consider that performance on the perceptual fluency task would be related to visual or verbal memory?

-On page 12, is RAN typically presented as a list rather than a matrix when testing in Chinese or was it presented like this to be more like the word reading?

-On page 14, I was wondering whether the measures from the two morphological awareness tasks were correlated. The second task seems quite complex to me and I was wondering how participants performed.

-Neither morphological nor phonological awareness correlated with reading. Is this surprising or is it a common finding in reading in Chinese? Were the scores perhaps at floor/ceiling?

-Morphological awareness correlated with performance on the perceptual fluency tasks. Would this also indicate that performance on the perceptual fluency tasks is something more than visual processing or is the common variance related to something else?

-Table 1. It was not clear to me what each measure reflects (accuracy, speed sec-ms) and whether in these tasks there are minimum/maximum minimum possible scores. If there aren´t actual minimum maximum scores, then perhaps providing the range of scores could be helpful for the reader. I also consider that presenting the raw threshold values from the perceptual fluency tasks (additionally to the log transformed values) could be useful so the reader can more easily interpret the numbers.

-Tables 3 and 4. I am not sure I understand what the capital B stands for.

3. Has the statistical analysis been performed appropriately and rigorously?

It seems that the analyses have been conducted appropriately although some additional information on the tests used and distribution of the data could be presented. I am not sure it is mentioned but I assume that data for each variable were normally distributed since parametric tests have been used. I am not an expert on regression models but based on my experience I was a bit surprised that it was possible to fit a model including so many variables with only 35 participants without overfitting. Is it the case that there is more than one observation per participant for the reading scores? Is there a way the authors could check for overfitting? I was also wondering whether, when adding perceptual fluency for characters in the speeded and non-speeded reading models it was possible to check if adding this variable significantly improved the model overall (additionally to checking the variance it explained after it was added).

4. Have the authors made all data underlying the findings in their manuscript fully available?

Yes, the data is available.

5. Is the manuscript presented in an intelligible fashion and written in standard English?

Yes.

Reviewer #3: The introduction is clear and easy to follow. The analysis of results appears to be appropriate, and it was good to see reports on reliability of all tasks. My comments are mostly about discussion points, interpretations, and some lack of details.

Alternative interpretations of your results:

You should absolutely discuss alternative interpretations of your data. For example, in your perceptual fluency task, you manipulate the duration of the target (characters or digits) depending on performance and find that poorer readers need more time to process characters. While you interpret this as a perceptual expertise deficit, would this ever have been unexpected even from the standpoint of other theories of dyslexia, as slow reading is one of the characteristics of the disorder in the first place? If they read poorly, the characters will have disappeared before they can read them all successfully and hence, they cannot match them well unless they are shown for a longer time. What if this is e.g. because poorer readers take longer converting each character to a phonological code? Or they have poorer verbal working memory, which could play a part in this task? Or they have problem with crowding or object recognition regardless of experience, as characters are likely more self-crowding (crowding occurs between the parts of an object, see Martelli et al. 2005) and more visually complex than numbers? Etc.

Martelli, M., Majaj, N. J., & Pelli, D. G. (2005). Are faces processed like words? A diagnostic test for recognition by parts. Journal of Vision, 5(1), 6-6.

Category specificity:

As you pointed out in the beginning of the article (p. 6), perceptual expertise is often highly specific to a certain object category. Would you expect a perceptual expertise deficit in dyslexia to be specific to words/characters? If so, why? If not, how would you expect this problem to manifest for other visual objects? The current evidence for this is mixed. E.g. Gabay et al. found problems in dyslexic readers for faces, an expertise category, but not for cars, leading these authors to suggest that: “…DDs’ impaired performance on face and word stimuli can be accounted for by difficulties in learning or gaining perceptual expertise (and the ability to make finegrained discrimination among a group of homogeneous exemplars).” Sigurdardottir et al. (2018) found a problem with faces but not novel objects, again in accordance with a visual expertise account, where they: “…speculate that reading difficulties in dyslexia are partially caused by specific deficits in high-level visual processing, in particular for visual object categories such as faces and words with which people have extensive experience.” Sigurdardottir et al. (2019) again found problems with faces, but they found this regardless of experience with faces (own vs. other-race faces), leading them to say that: “Visual problems in dyslexia are not demonstrably dependent on visual experience.”

Gabay, Y., Dundas, E., Plaut, D., & Behrmann, M. (2017). Atypical perceptual processing of faces in developmental dyslexia. Brain and language, 173, 41-51.

Sigurdardottir, H. M., Fridriksdottir, L. E., Gudjonsdottir, S., & Kristjánsson, Á. (2018). Specific problems in visual cognition of dyslexic readers: Face discrimination deficits predict dyslexia over and above discrimination of scrambled faces and novel objects. Cognition, 175, 157-168.

Sigurdardottir, H. M., Hjartarson, K. H., Gudmundsson, G. L., & Kristjánsson, Á. (2019). Own-race and other-race face recognition problems without visual expertise problems in dyslexic readers. Vision research, 158, 146-156.

No control group:

On page 9, you say: “Note that we were interested in investigating the variability within children with dyslexia such that the difficulty level of the tasks was designed to be appropriate for their ability. Ceiling effects could easily result if the tasks were performed by typically developing children. Hence comparing the role of perceptual expertise with words in normal readers and readers with dyslexia was out of the scope of this study.” You have Raven’s, which should cover a wide range of abilities, perceptual fluency, which by definition covers a wide range of abilities as it uses a staircase procedure, and RAN and speeded reading, both of which measure time which again should cover a wide range of abilities. I can see how non-speeded reading, phonological awareness, and possibly morphological awareness might have ceiling effects in a typical sample, but you have lots of tasks that would be fine for a control group.

Adding details:

Some details in the procedure are missing. E.g. on page 11, you say that you used a premask in the perceptual fluency test. What kind of mask, what were its properties? You say that you showed four characters. Were they always aligned vertically or did that vary? Was the choosing of the to-be-replaced character location (first, second, third, fourth) random, counterbalanced, other? Why did you use characters 0-9 except 1 in the perceptual fluency task, but digits 2, 4, 6, 7, and 9 in RAN? Was the choice of logarithm of the duration thresholds based on how these perceptual fluency tasks have been previously run with e.g. English words, musical notation etc., or was it idiosyncratic for this study, and then why? In phonological awareness, you talk about tone, can you briefly explain to non-Chinese speakers what you mean? Or was it just three different people with three different voices that read the characters?

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Li-Chih WANG

Reviewer #2: No

Reviewer #3: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

PLoS One. 2021 Jan 22;16(1):e0243440. doi: 10.1371/journal.pone.0243440.r004

Author response to Decision Letter 1


16 Sep 2020

Please refer to the attachment called 'Response to reviewers' as we included some figures in the responses, which cannot be shown here.

Attachment

Submitted filename: ResponseLetterToReviewers_v4.pdf

Decision Letter 2

Athanassios Protopapas

12 Oct 2020

PONE-D-20-07055R2

Perceptual expertise with Chinese characters predicts Chinese reading performance among Hong Kong Chinese children with developmental dyslexia

PLOS ONE

Dear Dr. Wong,

Thank you for submitting your manuscript to PLOS ONE. The reviewers note that you have made an admirable effort to address the comments of the first round of reviews, resulting in an improved manuscript, and I agree. Both reviewers are positively inclined to the eventual acceptance of your manuscript for publication at PLOS ONE, as am I. However, Reviewer #3 makes a number of critical observations, with which I agree, and which I believe you should fully address in the manuscript before it can be accepted. I am therefore inviting you to submit a revision in which you address these issues.

In particular, it seems to me that you greatly overplay a supposed theoretical difference between "perceptual fluency" and reading, while you downplay the role of short-term verbal memory in your "perceptual fluency" task.  I agree with Reviewer #3 that your study essentially shows that "reading predicts reading" (although measured in different ways and perhaps stressing different aspects). This does not mean that the study is not useful or cannot be accepted for publication, however as there are at least two of your colleagues (Reviewer #3 and myself) who are of this opinion, it is reasonable to imagine that there are probably more out there, so it seems wise to take this point more seriously in the manuscript in order to maximize your impact. Your perceptual fluency task is a masked word identification task; and for all of the reasons you mention that concern word specificity as an aspect of expertise, for some of us this means that this is basically a word reading task.  Note that tasks manipulating word presentation duration are not unheard of in the word recognition literature; indeed there are recent attempts to use such tasks to assess "word reading automaticity" (Roembke et al., 2019; https://doi.org/10.1037/edu0000279). I believe that there are some problems with that approach as well, but the point is that more readers might share the view of Reviewer #3 regarding what it is you are measuring.

Furthermore, the need to retain a set of 4 words so that it can be matched after 500 ms suggests that verbal short-term memory may carry most of your effect, as noted by Reviewer #3. It seems reasonable to imagine that your participants performed the task by (verbally) remembering the words, or at least by partial support from verbal memory, as you also acknowledge in the manuscript. The only way to ensure this is not the case is to use unfamiliar visual stimuli that have no verbal association, but this is obviously impossible by definition when you are specifically interested in perceptual expertise with words. (This again goes back to the idea that your perceptual expertise is a kind of a reading task.) The potential involvement of verbal memory is also an issue with much VAS research, especially the commonly used full-report version of VAS, and there are indications that when this aspect of the task is removed then the association with reading may diminish or disappear (Banfi et al., 2018; https://doi.org/10.1371/journal.pone.0198903).  Again, this does not invalidate your study or your manuscript, but it does suggest that some of your claims need to be greatly tempered, with these alternative views taken into serious consideration in the manuscript. 

Let me point out that PLOS ONE does not require that your study leads to a definitive novel theoretical contribution; it only requires that your conclusions are supported by the data. Therefore, tempering or adjusting your conclusions to allow for different conceptions of the tasks does not diminish the potential of your manuscript for acceptance; instead, it increases it. In this sense I want to stress that these comments are purely constructive.

Finally, I would also like to point out a couple of additional minor comments to take into account in your revision:

On p. 3, phonological awareness has nothing to do with perception of the sounds, it is a meta-linguistic skill that concerns conscious awareness and manipulation.

On p. 9 you argue against a "spatial" interpretation partly on the basis that no 3rd dimension is involved. I think this is a misunderstanding. Two-dimensional space still concerns spatial relations, therefore, two-dimensional visuospatial skills are conceivable and measurable. There is no requirement of depth for something to be considered "spatial".

On p. 15 you describe the PA task as involving characters but PA is a purely oral skill that has nothing to do with characters or any other aspect of the written language. I imagine that here you mean "syllables" or "words" rather than characters (after all, characters are orthographic entities, they cannot be spoken, only visually presented)

Please submit your revised manuscript by Nov 26 2020 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.  

Please include the following items when submitting your revised manuscript:

  • A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

  • A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

  • An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Athanassios Protopapas

Academic Editor

PLOS ONE

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #2: All comments have been addressed

Reviewer #3: (No Response)

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #2: Yes

Reviewer #3: Partly

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #2: Yes

Reviewer #3: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #2: Yes

Reviewer #3: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #2: Yes

Reviewer #3: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #2: Thank you for addressing all my previous comments. I really enjoyed reading the revised version and also the sections added based on other reviewers' comments.

Some last suggestions (that you may or may not want to take into account):

1. Discussing the absence of correlations with morphological and phonological awareness at the very end of the discussion might not be the best approach since it is nicer if you conclude with the important findings of your paper.

2. I think the tests you did to check overfitting are informative so you might want to provide this information at some point.

3. I believe you mention COREVA but then change to COVERA. This is not very important. I felt that this section wasn't very clearly linked to your work. This may be just my feeling but I mentioned the battery because it might be of interest in relation to your work rather than because it had to be mentioned in detail.

Reviewer #3: I want to reiterate my comments from the previous review that I think that the paper is quite well written and that the analyses appear to be appropriate. I do not have particular problems with this paper being accepted, but I add some comments below for further guidance. I stand by my previous comment that alternative explanations should be discussed further, although the authors have added some discussion on this which is good. It is in the end up to the editor to decide whether further action should be taken to address my comments below, and I do not believe that I need to re-review the paper unless the editor thinks that this is necessary of course.

“…perceptual fluency for characters predicted speeded and non-speeded word reading performance.” I know that the authors do not agree but many would simply say that this means that reading predicts reading.

p. 8: “…word reading explicitly requires one to read aloud the words” – most often, people read silently so I do not think you can make such a claim.

p. 9: “With a brief time gap between stimulus presentation and report, the requirement on visual working memory is relatively minimal.” I disagree. In your task, the delay period is 500 ms and masked. This task therefore is in essence a visual working memory task almost by definition. Iconic memory shouldn’t last this long, especially with a mask, see e.g. https://www.frontiersin.org/articles/10.3389/fpsyg.2014.00971/full

p. 9: “A separate perceptual fluency task for four-digit strings was also included. This task, together with RAN, ensured that any explanatory power of the perceptual fluency for characters on Chinese reading would not be explained by visuospatial abilities or discrimination abilities general to all kinds of visual stimuli and objects.” Here, again, I disagree. You yourselves say that “Chinese characters are visually complex” and state that this is one reason why perceptual expertise with words might be particularly important in Chinese reading. Digits are much less visually complex than Chinese characters so the explanatory power of the perceptual fluency for characters for Chinese reading could possibly be explained by e.g. discrimination abilities general to all kinds of visual stimuli and objects.

P. 26: „In the current study, the tasks... would have been too easy and resulted in ceiling effects with typical readers.“ You cannot say this as some of the tasks were staircase tasks and therefore have no ceiling/floor effects by definition.

P. 29: „Since domain-general object recognition skills should be captured by both perceptual fluency measures, we did not observe any evidence for the role of domain-general skills in predicting word reading.“ This cannot be be claimed, as perceptual fluency for digits (tapping into domain-general skills, among other things) DID predict word reading, i.e. it was correlated with it. However, the character fluency task (presumably tapping into domain-specific skills) explained further variance not explained by the digit task.

P. 30: „...perceptual fluency task with digits were identical to that with characters in terms of how the stimuli were spatially presented, and therefore general visual crowding effects should have been partialed out in our regression model by the perceptual fluency task with digits.“ I would be careful in making such as claim, as characters are likely more self-crowding and more visually complex than numbers, as pointed out in my previous review.

Minor details: „Perceptual expertise with words might be particularly important in Chinese reading because of two reasons“ (p. 5) should read: „Perceptual expertise with words might be particularly important in Chinese reading because of three reasons“. „Recent studies have reported impaired performance in object recognition in children with developmental dyslexia“ (p. 28) should read: „Recent studies have reported impaired performance in object recognition in adults with developmental dyslexia“.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #2: No

Reviewer #3: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

Decision Letter 3

Athanassios Protopapas

23 Nov 2020

Perceptual expertise with Chinese characters predicts Chinese reading performance among Hong Kong Chinese children with developmental dyslexia

PONE-D-20-07055R3

Dear Dr. Wong,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Athanassios Protopapas

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

Acceptance letter

Athanassios Protopapas

5 Jan 2021

PONE-D-20-07055R3

Perceptual expertise with Chinese characters predicts Chinese reading performance among Hong Kong Chinese children with developmental dyslexia

Dear Dr. Wong:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Athanassios Protopapas

Academic Editor

PLOS ONE

Associated Data

    This section collects any data citations, data availability statements, or supplementary materials included in this article.

    Supplementary Materials

    Attachment

    Submitted filename: ResponseLetter_v1.docx

    Attachment

    Submitted filename: ResponseLetterToReviewers_v4.pdf

    Attachment

    Submitted filename: ResponseLetter_R3_v2.docx

    Data Availability Statement

    All data are available in OSF DOI 10.17605/OSF.IO/DCTH6.


    Articles from PLoS ONE are provided here courtesy of PLOS

    RESOURCES