Treffer: Individual Differences in Holistic Word Processing and the Role of Phonological Processing in Sentence Comprehension

Background: Recent research has documented individual differences in the reading profiles of skilled native English readers using a behavioural marker of holistic visual word processing (orientation sensitivity). A more holistic word reading profile is associated with a weaker correlation with phonological decoding for word identification. Interestingly, that is different from typical patterns in skilled reading in English. The current study aimed to further assess the underlying distribution of orientation sensitivity as a continuous measure in an unrestricted undergraduate population and to extend past findings to understand the role of phonological decoding in reading comprehension as a function of orientation sensitivity. Methods: A group of 137 SUNY New Paltz undergraduates completed an orientation sensitivity test to assess the overarching distribution. A subset of 55 SUNY New Paltz undergraduates completed additional reading-related tests (phonological decoding and reading comprehension) on an online platform (findingfive.com). The relationship between orientation sensitivity and reliance on phonological decoding on reading comprehension was assessed using regression. Results: Results found that the distribution of orientation sensitivity was not normal and skews to the right, but is not bimodal, thus warranting the use of regression instead of group statistics. Lastly, those with relatively greater orientation sensitivity had a weaker relationship between measures of phonological decoding and a reading comprehension measure, as predicted. Conclusions: The results extended the finding of differences in reliance on phonological decoding to reading comprehension as a function of orientation sensitivity. These results help refine our understanding of orientation sensitive readers and potential alternative route to successful reading found in the general population.

As Provided

AN0174912655;d8c01feb.24;2024Jan23.04:24;v2.2.500

Individual differences in holistic word processing and the role of phonological processing in sentence comprehension 

Background: Recent research has documented individual differences in the reading profiles of skilled native English readers using a behavioural marker of holistic visual word processing (orientation sensitivity). A more holistic word reading profile is associated with a weaker correlation with phonological decoding for word identification. Interestingly, that is different from typical patterns in skilled reading in English. The current study aimed to further assess the underlying distribution of orientation sensitivity as a continuous measure in an unrestricted undergraduate population and to extend past findings to understand the role of phonological decoding in reading comprehension as a function of orientation sensitivity. Methods: A group of 137 SUNY New Paltz undergraduates completed an orientation sensitivity test to assess the overarching distribution. A subset of 55 SUNY New Paltz undergraduates completed additional reading‐related tests (phonological decoding and reading comprehension) on an online platform (findingfive.com). The relationship between orientation sensitivity and reliance on phonological decoding on reading comprehension was assessed using regression. Results: Results found that the distribution of orientation sensitivity was not normal and skews to the right, but is not bimodal, thus warranting the use of regression instead of group statistics. Lastly, those with relatively greater orientation sensitivity had a weaker relationship between measures of phonological decoding and a reading comprehension measure, as predicted. Conclusions: The results extended the finding of differences in reliance on phonological decoding to reading comprehension as a function of orientation sensitivity. These results help refine our understanding of orientation sensitive readers and potential alternative route to successful reading found in the general population.

Highlights: What is already known about this topicOrientation sensitivity has been used as a measure of holistic orthographic coding.High orientation sensitivity was associated with a weaker correlation between sublexical phonological processing and word identification. What this paper addsRegression is a valid way to examine orientation sensitivity given the distribution of individual differences in the general population of skilled college readers.Greater orientation sensitivity was linked to a weaker correlation with phonological decoding at the level of comprehension, independent of reading skill. Implications for theory, policy or practiceSensitivity to orientation, and holistic orthographic coding, can be used to identify individual differences in readers who employ an atypical route to skilled reading that relies relatively less on phonological processing.This reading profile exists in the general population and is not necessarily linked to reading skill.Due to the decrease in reliance on phonological processing, understanding this naturally occurring reading profile could be relevant to understanding alternative routes to literacy for those with deficits in phonological processing.

Keywords: orientation sensitivity; orthographic coding; individual differences; phonological processing; reading comprehension

Individual differences that support an atypical route to skilled reading in English have recently been identified through both behavioural assessments of reading (Hirshorn et al., [18]) and laterality of a neural signature of word processing (Carlos et al., [7]). The atypical route includes relatively weaker correlations between sublexical and phonological psycholinguistic factors and word identification (Hirshorn et al., [18]) and a more bilateral neural signature in the mid‐fusiform gyrus, often termed the Visual Word Form Area (Carlos et al., [7]). The behavioural findings of weaker correlations with phonological measures are especially intriguing since there is wide acceptance that phonological processing is a key foundational skill for becoming a proficient English reader (Ehri et al., [12]; Snowling, [39]; Wagner & Torgesen, [42]) and remains into adulthood (Milledge & Blythe, [27]; Pennington et al., [32]). Understanding an alternative route to reading with less reliance on such sub‐skills is relevant for models of reading and could be of interest to struggling readers with phonological processing impairments.

These atypical patterns in skilled English readers were discovered by exploring differences in a behavioural marker of holistic visual word processing: orientation sensitivity, an example of what has previously been referred to more broadly as distortion sensitivity (Hirshorn & Harris, [17]). The objective of using a measure like orientation sensitivity is to help identify individual differences in orthographic coding, such that an individual with greater sensitivity to an inverted word compared to an upright word, for example, utilises a more holistic orthographic coding approach. This reasoning draws from the face processing literature where face inversion has been used as a marker of holistic face processing (Farah et al., [14]). In the context of word processing, holistic coding is thought to capture the mapping between a specific visual form and its lexical equivalent, while analytic coding captures the mapping between sublexical orthographic units (e.g. graphemes, bigrams, etc.) and their phonological equivalents. Distorting or inverting the typical form of a word (or face) is thought to reveal these differences because it leads to the requirement for a more effortful sublexical approach in word identification (i.e. disrupts holistic processing). Thus, if distorting the typical form is relatively more disruptive to word identification, it suggests a more holistic coding approach is relied upon more. Indeed, past research found that word reading in individuals with greater orientation sensitivity was relatively more affected by whole‐word factors like length and to a lesser degree imageability, while individuals with less orientation sensitivity were relatively more affected by sublexical factors like biphone frequency (Hirshorn et al., [18]). Since English affords coding both holistically and analytically (i.e. whole word identification and sublexical‐based phonological decoding), a measure like this is necessary beyond simple measures of word identification of upright words to reveal differences in orthographic coding procedures.

More broadly, preferred orthographic coding strategy is thought to be a correlate of overall reading procedures. This has primarily been supported by cross‐linguistic work comparing Chinese–English and Korean–English bilinguals processing English (Akamatsu, [1], [2]; Ben‐Yehudah et al., [3]; Muljani et al., [28]; Pae et al., [30]; Pae & Lee, [31]; Wang et al., [43]), based on the idea that one's first language writing system influences how their second is processed (Kim et al., [20]; Koda, [21], [24]). Chinese and Korean are structured completely differently, with Chinese being a morphosyllabic/logographic system and Korean being an alphabetic syllabary, where letters are organised into syllable blocks. Several studies have shown group differences in that Chinese–English bilinguals are more affected than Korean–English bilinguals by distorting the typical visual presentation (e.g. orientation, case alternation, etc.) of English, their L2 (Akamatsu, [1], [2]; Ben‐Yehudah et al., [3]; Pae et al., [30]; Pae & Lee, [31]). This suggests that Chinese‐English bilinguals process English in a relatively more holistic manner than Korean–English bilinguals. Additionally, cross‐linguistic differences in orthographic coding strategies have also been linked with differences in lexical vs. sublexical biases in word identification (Ben‐Yehudah et al., [3]; Bhide, [4]). For example, Chinese–English bilinguals, who were more sensitive to distortion, were also biased more towards lexical‐level psycholinguistic measures, like frequency, and less towards sublexical‐level phonological measures, like consistency (Ben‐Yehudah et al., [3]). Compared to Korean–English bilinguals, Chinese–English bilinguals display relatively less reliance on phonological sublexical processing in the service of word identification (Koda, [21], [22], [23], [24]) and comprehension (Koda, [23]), despite having equivalent sub‐skills like nonword decoding.

While previous cross‐linguistic studies of orientation sensitivity have been helpful in identifying an alternative route to skilled reading in English as a second language, there is still much unknown about how such an alternative route manifests in a typical skilled native English reading population. However, the concept of holistic word processing in a single writing system has been explored. For example, individuals with dyslexia who displayed slower word identification also showed less of a holistic orthographic coding profile using a variation on word inversion and word jumbling (Conway et al., [10]). Hence, a greater holistic profile was associated with greater reading skill. While not in English, Ventura and colleagues found a similar pattern in Portuguese, similarly an alphabet, using the composite task, another measure of holistic word processing, and found that holistic processing of visual words is related to higher efficiency in visual word recognition by skilled readers (Ventura et al., [41]). This is in contrast to the previous work that found no connection between reading skill and orientation sensitivity, although that was in a very small sample (Hirshorn et al., [18]). Nevertheless, these studies have identified differences of holistic word processing as a correlate of reading skill. In contrast, the current paper aims to understand more what individual differences in holistic processing could mean about reading procedures within a single writing system, independent of overall skill.

Previous research found a connection between reliance on phonological processes and word identification as a function of orientation sensitivity, such that those with higher orientation sensitivity had a weaker connection between phonological decoding and single overt word identification (Hirshorn et al., [18]). One natural extension would be that this differential reliance on phonological processes would also transfer to the level of reading comprehension. Indeed, phonological decoding has repeatedly been shown to predict reading comprehension (Engen & Høien, [13]; Shankweiler et al., [37]), and although the correlation may become weaker with age (García & Cain, [15]), it still remains into adulthood. However, reading comprehension is much more complex than single word identification and involves additional syntactic, semantic/vocabulary, memory processes, and inferences making (Cain, [5]; Cain & Oakhill, [6]; Nouwens et al., [29]; Van Dyke et al., [40]). For that reason, it is not necessarily true that the pattern of less reliance on phonological decoding in orientation sensitive readers would remain at the level of reading comprehension. Nevertheless, based on ideas such as the lexical quality hypothesis (Perfetti, [33]), which suggests that lower level representations at the word level have consequences for higher level skill like comprehension, we hypothesised that the same hallmark pattern of a weaker correlation between phonological processing and reading skill in higher orientation sensitive readers would remain at the level of comprehension.

Lastly, in pursuit of understanding the consequences of individual differences in holistic processing, the overall distribution of orientation is important to know. Since past research looked a very restricted sample, the nature of the distribution of orientation sensitivity in a larger population is unknown. Because individuals with high orientation sensitivity have not been identified as different, due to high skill, they likely never reached the level of additional investigation. Therefore, the current study aims to identify how common higher orientation sensitivity is, if there are distinct groups, or whether the measure exists along a normal continuum.

To summarise, the current study aims to (1) understand the distribution of orientation sensitivity in a larger sample, as preliminary studies were conducted on small and highly constrained group analyses, and (2) expand the scope of understanding how orientation sensitivity may relate to differences in reliance on phonological processing at the comprehension level.

Methods

Participants

Participants were 137 right‐handed (105 female, 24 male, 8 preferred not to say; mean age = 21.3 years, SD = 3.8), native monolingual English‐speaking undergraduates at SUNY New Paltz with no reported history of hearing or vision issues, learning or reading difficulties, mental illness or neurological problems. A subset of 55 participants (33 females, 21 male, 1 preferred not to say; mean age = 21.9 years, SD = 3.3) was part of a separate study that collected additional psycholinguistic measures. The study was approved by the Human Research Ethics Board at SUNY New Paltz, and participants were given class credit.

Orientation Sensitivity

Sensitivity to atypical orientation was assessed using a lexical decision task in which the stimuli were presented in upright and inverted (rotated 180°) orientations (see Figure 1 for example). Words were blocked such that upright stimuli were presented first, followed by inverted stimuli, with words and non‐words randomised within a block. Words were chosen to neither have extremely high nor low lexical frequency (min log HAL frequency = 6.16, max = 12.85, mean = 8.76) and to have one syllable. There were 40 words and 40 non‐words, with half of each presented in each orientation. The upright and inverted conditions were matched for frequency, length, bigram frequency and orthographic neighbours.

Participants made a two alternative forced decision about each stimulus (real word or nonword), presented in blocks (upright and inverted). The task was self‐paced, with the next word appearing after each decision had been preceded by a fixation cross for 500 ms. Participants were instructed to answer as quickly and as accurately as possible. Orientation sensitivity was calculated as the ratio of median reaction time for correct inverted trials divided by median reaction time for correct upright trials, such that a larger number indicates greater orientation sensitivity.

Reading Skills

The reading skills of participants were assessed using tasks that were easily adaptable to an online platform (i.e. did not require overt responses).

Reading Comprehension

Reading comprehension was assessed using the Passage Comprehension Subtest of Woodcock Reading Mastery Tests (WRMT‐Revised, Form H; Woodcock, [44]) using the standardised W‐scores for statistical analyses as they are the preferred measure for most statistical comparisons and provide an equal‐interval measure of test performance. The test requires participants to supply missing words that best fit the context of short passages. It was adapted for online responses, such that participants read sentences with a blank and typed the single word that best completed the sentence.

Phonological Decoding Test

Phonological decoding was tested using the lexical knowledge battery (Perfetti & Hart, [34]), which consisted of a list of non‐words, some of which when pronounced sounded like real words (e.g. kote sounds like a real word (coat), whereas lote does not). During the phonological decoding test, participants saw one word at a time and were asked to identify if they sounded like a real word or not. Participant performance for the phonological decoding test was measured by d′ (i.e. sensitivity to pronunciation of words) factoring in both hit rate and false alarms.

Overall Procedure

Data was collected remotely using the online platform FindingFive.com. Participants were asked to find time in a quiet room without distraction to complete the experiment. For the participants that completed several tasks, they were given a break after each task. The remainder of this session was devoted to tasks for an ongoing database data collection, and it will not be discussed further.

Results

Orientation sensitivity was calculated as in previous studies (Hirshorn et al., [18]) as the median reaction time for correct trials of inverted stimuli divided by upright stimuli. Accuracy was high (mean = 91.16, standard deviation = 0.052; see Table 1), confirming engagement in the task. The mean W‐score in reading comprehension was 528 (SD = 9.97), and the mean d′ for phonological decoding was 1.72 (SD = 0.57).

1 Table Descriptive statistics of orientation sensitivity measure (N  = 137)

<table><thead valign="bottom"><tr><th /><th>Reaction time</th><th>Accuracy</th></tr><tr><th /><th>Mean</th><th>Standard deviation</th><th>Mean</th><th>Standard deviation</th></tr></thead><tbody valign="top"><tr><td>Orientation sensitivity ratio</td><td align="char" char=".">1.75</td><td align="char" char=".">0.74</td><td>0.96</td><td>0.1</td></tr><tr><td>Upright real word</td><td align="char" char=".">903.88</td><td align="char" char=".">316.84</td><td>0.93</td><td>0.03</td></tr><tr><td>Upright non&#8208;word</td><td align="char" char=".">1116.34</td><td align="char" char=".">501.34</td><td>0.91</td><td>0.12</td></tr><tr><td>Inverted real word</td><td align="char" char=".">1348.87</td><td align="char" char=".">604.57</td><td>0.9</td><td>0.12</td></tr><tr><td>Inverted non&#8208;word</td><td align="char" char=".">1939.22</td><td align="char" char=".">916.09</td><td>0.87</td><td>0.12</td></tr></tbody></table>

The first goal of this study was to assess the distribution of orientation sensitivity in a larger and less restricted sample. It was assumed that there would be a range of orientation sensitivity in the broader population, but the type of distribution was unknown. The range of orientation scores was graphed (range: 0.82–4.83; Figure 2a), where larger numbers indicate greater sensitivity to orientation (inverted/upright). Although most orientation sensitivity ratios were near 1, there was significant variability (Table 1).

Based on a test of normality, the distribution in this sample is not normal (Figure 3), but appears to be right skewed (skewness score = 1.93), and possibly bimodal, with peaks in this sample at an orientation sensitivity ratio of 1.3/1.4 and smaller one around a ratio of 2. Interestingly, the apparent dip between two distributions around a ratio of 1.7 is similar to where groups were defined in a previous study with a much smaller sample (i.e. orientation ratio ~1.5; Hirshorn et al., [18]). However, a diptest was performed using R diptest package (R. D. C. Team, [35]) to assess unimodality versus multimodality (Hartigan & Hartigan, [16]), and it was determined that the distribution was not multimodal D = 0.027, p = 0.76. This validated the use of orientation sensitivity as a predictor in regression analyses instead of group design analyses.

However, of main interest was to test whether there was a difference in the relationship between phonological decoding and reading comprehension as a factor of orientation sensitivity. This was done in a smaller sample (N = 55) from whom additional reading measures were collected, but more than twice as big as in previous research (Hirshorn et al., [18]). The distribution in the subset was similar to the full dataset (Figure 2b, skewness score = 1.72). A hierarchical regression was computed with reading comprehension as the dependent measure, first adding orientation sensitivity and then phonological decoding. Lastly, the interaction term between orientation sensitivity and phonological decoding was added (e.g. orientation sensitivity * phonological decoding) (Table 2). Results showed that orientation sensitivity was a modestly significant predictor (p = 0.033), such that the greater orientation sensitivity ratio the higher the comprehension score. Phonological decoding was a robust significant predictor of reading comprehension (p < 0.001), such that the better the phonological decoding score, the higher the reading comprehension score, which is consistent with past research, even in adults (García & Cain, [15]; Milledge & Blythe, [27]). However, when phonological decoding was added first, the correlation between orientation sensitivity and reading comprehension was no longer significant (p = 0.42) once phonological decoding was controlled for (Table 3). Importantly, in both models the interaction term between orientation sensitivity and phonological decoding still explained a significant amount of variance in reading comprehension measures (R<sups>2</sups> = 0.05, p = 0.031), even after accounting for phonological decoding, such that for readers with relatively higher orientation sensitivity ratio (i.e. holistic word processing), the relationship between phonological decoding and reading comprehension was weaker than in those with lower orientation sensitivity ratios (see Figure 4). For the sake of visualisation, even though the inversion sensitivity measure is treated as a continuous variable in the model, the orientation sensitivity measure was coded as lower sensitivity for those below the median split of 1.5 and those above as higher. With these group definitions, there were no differences in reading comprehension (M<subs>lower sensitivity</subs> = 525.77, SD<subs>lower sensitivity</subs> = 10.89; M<subs>higher sensitivity</subs> = 529.5, SD<subs>higher sensitivity</subs> = 8.94; t(52) = 1.38, d = 0.37, p = 0. 17). Interestingly, there was a small group difference in phonological decoding, but it was the higher sensitivity group that on average scored higher (M<subs>lower sensitivity</subs> = 1.59, SD<subs>lower sensitivity</subs> = 0.62; M<subs>higher sensitivity</subs> = 1.90, SD<subs>higher sensitivity</subs> = 0.42; t(52) = 2.16, d = 0.59, p = 0.035).

2 Table Hierarchical regression examining predictors of reading comprehension, order 1

<table><thead valign="bottom"><tr><th>Order</th><th>Added predictor</th><th>R² Change</th><th>F change</th><th>df1</th><th>df2</th><th>p</th></tr></thead><tbody valign="top"><tr><td align="center">1</td><td>Orientation sensitivity ratio</td><td>0.089</td><td align="char" char=".">4.8</td><td>1</td><td>53</td><td align="char" char=".">0.033</td></tr><tr><td align="center">2</td><td>Phonological decoding</td><td>0.382</td><td align="char" char=".">34.68</td><td>1</td><td>52</td><td align="char" char=".">&#60;0.001</td></tr><tr><td align="center">3</td><td>Orientation sensitivity ratio&#160;*&#160;phonological decoding</td><td>0.05</td><td align="char" char=".">4.92</td><td>1</td><td>51</td><td align="char" char=".">0.031</td></tr></tbody></table>

3 Table Hierarchical regression examining predictors of reading comprehension, order 2

<table><thead valign="bottom"><tr><th>Order</th><th>Added predictor</th><th>R² change</th><th>F change</th><th>df1</th><th>df2</th><th>p</th></tr></thead><tbody valign="top"><tr><td align="center">1</td><td>Phonological decoding</td><td>0.464</td><td align="char" char=".">42.42</td><td>1</td><td>53</td><td align="char" char=".">&#60;0.001</td></tr><tr><td align="center">2</td><td>Orientation sensitivity ratio</td><td>0.007</td><td align="char" char=".">0.65</td><td>1</td><td>52</td><td align="char" char=".">0.42</td></tr><tr><td align="center">3</td><td>Orientation sensitivity ratio&#160;*&#160;phonological decoding</td><td>0.05</td><td align="char" char=".">4.92</td><td>1</td><td>51</td><td align="char" char=".">0.031</td></tr></tbody></table>

Discussion

The current study expanded on past research exploring a newly identified sub‐group of skilled English readers, those who are relatively more sensitive to orientation of word presentation. Past research had identified that orientation sensitivity was linked to less reliance on sublexical factors in word identification and that phonological decoding was less closely tied to word identification (Hirshorn et al., [18]). The current study expanded with a larger sample and additionally looked at the role of phonological decoding in reading comprehension, as a factor of orientation sensitivity. The results are consistent with past findings that a range of orientation sensitivity is present in a typical skilled reading population and that the distribution is not bimodal. Furthermore, the finding that greater orientation sensitivity predicts a weaker correlation between phonological decoding and reading comprehension is a natural extension from previous results that similarly found a disconnect between phonological decoding, which relies on sublexical analysis and single‐word reading.

Even with past studies that used a group design, it was assumed that there would be a range of orientation sensitivity scores, as it is a continuous measure, but the nature of the distribution was unknown. While still a relatively smaller sample, the current study suggests that the natural distribution of orientation sensitivity is not bimodal but may be skewed to the right. What contributes to these variations is still an open question, and may be a combination of method of instruction, which has been shown to influence reading procedures (Connelly et al., [9]) and inherent individual differences in information processing, which merits further work.

The finding that orientation sensitivity predicted some variance in reading comprehension (Table 2), such that those with higher orientation sensitivity had slightly higher comprehension scores, was unexpected, as there was a trending finding in the opposite direction in previous work, that non‐orientation sensitive readers had better reading comprehension (Hirshorn et al., [18]). However, the second analysis of the data where phonological decoding was added first eliminated this correlation (Table 3), suggesting that it does not explain any additional variance after controlling for phonological skill. Further, when comparing readers above and below the median split of orientation sensitivity, there were no significant differences in reading comprehension. These small discrepancies across data sets suggest that there may not be a strong and consistent relationship between orientation sensitivity and overall reading skill, and any apparent small differences may be a function of the specific sample collected, but future replications will be needed to clarify this relationship.

Lastly, the current study does replicate the hallmark difference in the relationship between phonological decoding and reading skill as a function of orientation sensitivity, this time at the level of reading comprehension, which is consistent with research in Chinese–English bilinguals (Koda, [23]). This result was not guaranteed based on the past findings of this pattern at the word level, due to additional factors that contribute to comprehension. Importantly, these results hold while controlling for phonological decoding skill. While the amount of additional variance (~5%) explained by the interaction between orientation sensitivity and phonological decoding is modest, it still suggests that there are individual differences in the extent to which phonological decoding is a predictor of reading comprehension in the more general population. While it is true that the relationship between phonological decoding and comprehension weakens over time as a function of age and experience (García & Cain, [15]), it is also accepted that the relationship remains significant into adulthood with typical skilled readers showing strong decoding skill (García & Cain, [15]; Milledge & Blythe, [27]). Thus, it is important to reiterate that this pattern of less reliance on phonological processing is not only atypical in English readers, but is often considered to be not optimal, as it is often associated with a profile of an individual with a reading disorder who is forced to compensate for impaired phonological processing (Jackson & Doellinger, [19]; Savill et al., [36]; Shaywitz et al., [38]). In contrast, in the current study, the readers who had relatively greater orientation sensitivity, who had a weaker correlation between phonological decoding and reading comprehension, scored significantly higher on the phonological decoding measure, highlighting that they are likely not compensating for impaired phonological processing. This is what makes the pattern observed in individuals with higher orientation sensitivity and no history of a reading disorder noteworthy.

Nevertheless, this finding further begs the question of what may be relied upon relatively more to offset a weaker connection with phonological decoding and comprehension in those with greater orientation sensitivity. To have a more thorough understanding of differences in reading processing in higher orientation sensitive readers, future research would benefit from a deeper exploration of word versus non‐word processing as a function of orientation sensitivity. As holistic word processing is more relevant in real word compared to non‐words, it would be predicted that a weaker correlation with phonological decoding, as observed in this study, would only be relevant in examining real words. In general, more temporally sensitive measures could also be helpful in determining any differences in lexical representation in higher orientation sensitive readers that may provide insight into why phonological decoding is a weaker predictor of different measures of reading skill. Interestingly, there are other subpopulations of skilled English readers who also display this pattern of less reliance on phonological measures (Hirshorn & Harris, [17]). They include deaf native signers, compensated dyslexic readers, and the aforementioned Chinese–English bilinguals. A common pattern in these populations is that morphological knowledge appears to play a relatively larger role in overall reading skill (Cavalli et al., [8]; Deacon et al., [11]; Lam et al., [25]; Law et al., [26]; Zhang & Koda, [45]). Exploring the role of morphological processing, among other potential protective factors such as visual memory, may be a fruitful direction for future research in understanding how or why this pattern emerges in those with greater orientation sensitivity and no evidence of reading disorder. Lastly, having a more thorough understanding of this seemingly naturally occurring alternative route to skilled reading could have implications for instructional approaches for those with impaired phonological processing, or simply individuals who show early evidence of a preferred approach to orthographic coding.

Acknowledgements

The authors would like to thank the Summer Undergraduate Research Experience Award at SUNY New Paltz (to CW) for supporting student involvement in research and Julie Fiez for reading an initial draft of the manuscript.

Conflict of Interest Disclosure

There are no conflicts of interest to report.

Ethics Approval Statement

The study was approved by the Human Research Ethics Board at SUNY New Paltz.

Data Availability Statement

The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.

By Elizabeth A. Hirshorn; Emma Reilly; Alison Louche‐Robert and Cody Wojszysnki

Reported by Author; Author; Author; Author

Elizabeth Hirshorn received her training at the University of Rochester in the Brain and Cognitive Sciences Department and University of Pittsburgh at the Learning Research and Development Center. Dr Hirshorn is currently an Associate Professor of Psychology at SUNY New Paltz where she is the head of the Diversity in Language Lab that utilises both behavioural and electrophysiological methods to investigate individual differences in reading. She is the Director of the Masters Program in Psychological Science. Her research currently focuses on atypical routes to skilled reading and how individual differences in reading may transfer to other areas of cognition.