View in EDS HTML Full Text PDF Full Text

Understanding the Role of Eye Movement Pattern and Consistency in Isolated English Word Reading through Hidden Markov Modeling

Saved in:
Bibliographic Details
Title: Understanding the Role of Eye Movement Pattern and Consistency in Isolated English Word Reading through Hidden Markov Modeling
Language: English
Authors: Weiyan Liao, Janet Hui-wen Hsiao (ORCID 0000-0003-2271-8710)
Source: Cognitive Science. 2024 48(9).
Availability: Wiley. Available from: John Wiley & Sons, Inc. 111 River Street, Hoboken, NJ 07030. Tel: 800-835-6770; e-mail: cs-journals@wiley.com; Web site: https://www.wiley.com/en-us
Peer Reviewed: Y
Page Count: 29
Publication Date: 2024
Document Type: Journal Articles
Reports - Research
Descriptors: Foreign Countries, English (Second Language), Eye Movements, Markov Processes, Sight Method, Naming, Reaction Time, Reading Fluency, Visual Acuity, Individual Differences
Geographic Terms: Hong Kong
DOI: 10.1111/cogs.13489
ISSN: 0364-0213
1551-6709
Abstract: In isolated English word reading, readers have the optimal performance when their initial eye fixation is directed to the area between the beginning and word center, that is, the optimal viewing position (OVP). Thus, how well readers voluntarily direct eye gaze to this OVP during isolated word reading may be associated with reading performance. Using Eye Movement analysis with Hidden Markov Models, we discovered two representative eye movement patterns during lexical decisions through clustering, which focused at the OVP and the word center, respectively. Higher eye movement similarity to the OVP-focusing pattern predicted faster lexical decision time in addition to cognitive abilities and lexical knowledge. However, the OVP-focusing pattern was associated with longer isolated single letter naming time, suggesting conflicting visual abilities required for identifying isolated letters and multi-letter words. In contrast, in both word and pseudoword naming, although clustering did not reveal an OVP-focused pattern, higher consistency of the first fixation as measured in entropy predicted faster naming time in addition to cognitive abilities and lexical knowledge. Thus, developing a consistent eye movement pattern focusing on the OVP is essential for word orthographic processing and reading fluency. This finding has important implications for interventions for reading difficulties.
Abstractor: As Provided
Entry Date: 2024
Accession Number: EJ1441014
Database: ERIC
Full text is not displayed to guests.
FullText Links:
  – Type: pdflink
    Url: https://content.ebscohost.com/cds/retrieve?content=AQICAHj0k_4E0hTGH8RJwT4gCJyBsGNe_WN95AvKlDbXJGqwxwFhF_Xgw29CGemMTuR3x5h9AAAA4zCB4AYJKoZIhvcNAQcGoIHSMIHPAgEAMIHJBgkqhkiG9w0BBwEwHgYJYIZIAWUDBAEuMBEEDHBbo_l1fB7eEZpgUgIBEICBm8WBp82ZETFH90RUhPEYSKqmkIFNHffpqugAeKxQWXOYT0-WB82j6eyk-lTvkvtPUfMTDu_7ZGqRavcPGb_o1DUimz5ZM9ocpQnbhXlKkwU0tXqccarveV28w7_XBB-jwpZbkr1FIH7dw2EJAITlSdkgBFy4Y6hP7T37B4IY3YcT935X_V7MxLmAC2zLJ_-EqjYxXII7T6tlw_wG
Text:
  Availability: 1
  Value: <anid>AN0179878392;cgn01sep.24;2024Sep27.07:11;v2.2.500</anid> <title id="AN0179878392-1">Understanding the Role of Eye Movement Pattern and Consistency in Isolated English Word Reading Through Hidden Markov Modeling </title> <p>In isolated English word reading, readers have the optimal performance when their initial eye fixation is directed to the area between the beginning and word center, that is, the optimal viewing position (OVP). Thus, how well readers voluntarily direct eye gaze to this OVP during isolated word reading may be associated with reading performance. Using Eye Movement analysis with Hidden Markov Models, we discovered two representative eye movement patterns during lexical decisions through clustering, which focused at the OVP and the word center, respectively. Higher eye movement similarity to the OVP‐focusing pattern predicted faster lexical decision time in addition to cognitive abilities and lexical knowledge. However, the OVP‐focusing pattern was associated with longer isolated single letter naming time, suggesting conflicting visual abilities required for identifying isolated letters and multi‐letter words. In contrast, in both word and pseudoword naming, although clustering did not reveal an OVP‐focused pattern, higher consistency of the first fixation as measured in entropy predicted faster naming time in addition to cognitive abilities and lexical knowledge. Thus, developing a consistent eye movement pattern focusing on the OVP is essential for word orthographic processing and reading fluency. This finding has important implications for interventions for reading difficulties.</p> <p>Keywords: Visual word recognition; Eye movement; Optimal viewing position; EMHMM</p> <hd id="AN0179878392-2">Introduction</hd> <p>Individual differences in the quality of word representation, including orthography, phonology, and meaning, are shown to contribute to variation in reading performance (Perfetti, [<reflink idref="bib65" id="ref1">65</reflink>]). For example, better phonological awareness, letter identification, or semantic knowledge has been reported to be associated with better word reading performance in both first and second language learners including children (e.g., Hogan, Catts, & Little, [<reflink idref="bib32" id="ref2">32</reflink>]; Jasinska & Petitto, [<reflink idref="bib48" id="ref3">48</reflink>]), adolescents (Gottardo, Koh, Chen, & Jia, [<reflink idref="bib30" id="ref4">30</reflink>]), and adults (Williams & Morris, [<reflink idref="bib81" id="ref5">81</reflink>]). Cognitive abilities, in particular working memory (Knoop‐van Campen et al., 2018), selective attention, and executive functions (Haft et al., [<reflink idref="bib31" id="ref6">31</reflink>]), are also shown to be associated with word reading performance.</p> <p>Visual processing skills also play an important role. For example, the ability to rapidly identify an isolated familiar symbol as measured in the rapid automatized naming (RAN) task is associated with word reading fluency in children (Kim, Park, & Lombardino, [<reflink idref="bib50" id="ref7">50</reflink>]; Savage et al., [<reflink idref="bib74" id="ref8">74</reflink>]). Also, better selective visual attention was shown to be associated with better single word reading and nonword reading performance (Lancaster, Li, & Gray, [<reflink idref="bib54" id="ref9">54</reflink>]). Perceptual expertise literature has suggested that both global and local information are important for the recognition of visual stimuli. For example, both global holistic representation (i.e., integration of multiple features into a single perceptual representation, which can typically be assessed using the composite paradigm; Gauthier, Klaiman, & Schultz, [<reflink idref="bib28" id="ref10">28</reflink>]; Rossion, [<reflink idref="bib73" id="ref11">73</reflink>]) and local featural information are important for face recognition (Cabeza & Kato, [<reflink idref="bib9" id="ref12">9</reflink>]; Chuk, Chan, & Hsiao, [<reflink idref="bib16" id="ref13">16</reflink>]; Hsiao, An, Zheng, & Chan, [<reflink idref="bib38" id="ref14">38</reflink>]). Similarly, Chinese character recognition expertise involves an initial increase in holistic perception, as assessed using the composite paradigm (Tso, Chan, & Hsiao, [<reflink idref="bib79" id="ref15">79</reflink>]), followed by enhanced selective attention to local components through writing practice (Tso, Au, & Hsiao, [<reflink idref="bib77" id="ref16">77</reflink>]; [<reflink idref="bib78" id="ref17">78</reflink>]), and reading difficulties in Chinese are associated with impaired ability to selectively attend to local components (Tso et al., [<reflink idref="bib79" id="ref18">79</reflink>]). In lexical decisions of low‐frequency Portuguese words, faster response time (RT) was associated with more holistic word perception as assessed using the composite paradigm, suggesting the relationship between word processing efficiency and holistic perception (Ventura et al., [<reflink idref="bib80" id="ref19">80</reflink>]). In English word naming, global and local perceptual primes led to differential modulation effects on words with irregular and regular letter‐sound mappings (Franceschini et al., [<reflink idref="bib27" id="ref20">27</reflink>]). These findings suggest that global‐local perceptual processing abilities may also influence word reading performance.</p> <p>While the factors of lexical knowledge, cognitive abilities, and visual processing skills have been well examined, whether individual differences in eye fixation behavior during word reading can also account for variation in performance is less understood. In isolated word reading, the optimal viewing position (OVP) phenomenon has been consistently observed in skilled readers, where performance in word naming, lexical decision, or perceptual identification is the best when participants' initial eye fixation is being directed to the region between the beginning and the middle of the words (Brysbaert & Nazir, [<reflink idref="bib7" id="ref21">7</reflink>]). Factors accounting for this phenomenon include (<reflink idref="bib1" id="ref22">1</reflink>) visual acuity drop from center to periphery; (<reflink idref="bib2" id="ref23">2</reflink>) information structure of English words, where word beginnings are more informative for identification than word endings (e.g., Chan & Hsiao, [<reflink idref="bib13" id="ref24">13</reflink>]; cf. Hsiao & Cheng, [<reflink idref="bib41" id="ref25">41</reflink>]; Hsiao & Cheung, [<reflink idref="bib42" id="ref26">42</reflink>]); (<reflink idref="bib3" id="ref27">3</reflink>) perceptual learning and reading direction, that is, the enhancement of word discrimination sensitivity at the right visual field (VF) due to the left‐to‐right reading direction (Chung, Liu, & Hsiao, [<reflink idref="bib18" id="ref28">18</reflink>]; Hsiao, [<reflink idref="bib34" id="ref29">34</reflink>]; Nazir & O'Regan, [<reflink idref="bib62" id="ref30">62</reflink>]); and (<reflink idref="bib4" id="ref31">4</reflink>) a left VF/right hemisphere advantage in language processing (Hsiao & Lam, [<reflink idref="bib44" id="ref32">44</reflink>]; Lam & Hsiao, [<reflink idref="bib52" id="ref33">52</reflink>]; cf. Hsiao & Liu, [<reflink idref="bib47" id="ref34">47</reflink>]). Thus, the OVP phenomenon may reflect reading skill development through reading experience. Nevertheless, it remains unclear whether readers voluntarily direct their eye gaze to this OVP during isolated word reading and whether variation in the ability to consistently fixate the OVP is associated with variation in performance. It is possible that readers whose eye movement patterns involve a more focused eye fixation distribution at the OVP during isolated word reading have better word processing performance. Indeed, Linka, Borda, Alsheimer, de Haas, and Ramon ([<reflink idref="bib59" id="ref35">59</reflink>]) recently showed that in face identification, adult super‐recognizers fixated closer to the OVP for face recognition, which was at the bridge of the nose below the eyes (Hsiao & Liu, [<reflink idref="bib47" id="ref36">47</reflink>]), suggesting that consistent and focused eye fixation behavior toward the OVP is associated with better identification performance. In addition, consistent and focused eye fixation behavior toward the OVP during isolated word reading may be associated with cognitive or visual processing skills relevant to word reading performance. Consistent with this speculation, eye movements during face recognition (Chan, Chan, Lee, & Hsiao, [<reflink idref="bib12" id="ref37">12</reflink>]) and scene perception (Hsiao, Lan, Zheng, & Chan, [<reflink idref="bib45" id="ref38">45</reflink>]) are associated with executive function and selective attention abilities.</p> <p>Here, we examined whether eye movement pattern and consistency during isolated word reading predicts reading performance in addition to lexical knowledge and cognitive ability and visual skill factors. We used a machine learning‐based method, Eye Movement analysis with Hidden Markov Models (EMHMM; Chuk, Chan, & Hsiao, [<reflink idref="bib15" id="ref39">15</reflink>]), to obtain quantitative measures of individual eye movement pattern and consistency in isolated word reading. In EMHMM, each individual's eye movements in a task were summarized using a hidden Markov model (HMM), including person‐specific regions of interest (ROIs) and transition probabilities among these ROIs. Individual HMMs can be clustered to discover representative patterns, and similarities among individual patterns can be quantified using data log‐likelihoods given the HMMs. Thus, it can help us identify representative eye fixation patterns among readers using a data‐driven approach and quantify individual differences. In addition, consistency of eye movement pattern across trials can be quantified using entropy of the HMM (Hsiao, An, & Chan, [<reflink idref="bib36" id="ref40">36</reflink>]; Hsiao, Chan, An, Yeh, & Jingling, [<reflink idref="bib40" id="ref41">40</reflink>]; Hsiao, Liao, & Tso, [<reflink idref="bib46" id="ref42">46</reflink>]; Liao, Li, & Hsiao, [<reflink idref="bib58" id="ref43">58</reflink>]). Entropy is a measure of predictivity; higher entropy indicates higher randomness (Cover & Thomas, [<reflink idref="bib21" id="ref44">21</reflink>]) and thus less consistent eye movements.</p> <p>Here, we focused on lexical decision, word naming, and pseudoword naming tasks. We hypothesized that individual readers differ in eye fixation behavior during isolated word reading, and more focused and consistent eye fixation behavior toward the OVP is associated with better reading performance, particularly for tasks that rely more on orthographic knowledge such as lexical decision. This eye fixation behavior may be associated with cognitive or visual processing abilities important for reading. We recruited adult English as second language learners to ensure sufficient variance in word reading performance for the examination.</p> <hd id="AN0179878392-3">Methods</hd> <p></p> <hd id="AN0179878392-4">Participants</hd> <p>Participants were 128 (38 males) adult English as a second language learners in Hong Kong. One hundred and twenty‐three of them had Chinese, one had Tamil, one had Korean, one had Punjabi, one had Tagalog, and one had French as their first languages. They all indicated that English was their second language in the order of dominance and acquisition. The age of beginning acquiring English ranged from 1 to 12 (<emph>M</emph> = 4.09, <emph>SD</emph> = 2.17). Their self‐reported proficiency in English reading ranged from 5 (adequate) to 10 (perfect) (<emph>M</emph> = 7.77, <emph>SD</emph> = 1.08). Their age ranged from 18 to 30 (<emph>M</emph> = 21.7, <emph>SD</emph> = 3.03). They had similar college education backgrounds. All participants had normal vision or corrected to normal vision by contact lenses or glasses. According to a power analysis using G*Power, a sample size of 115 was needed to acquire a small to medium effect size (<emph>r</emph><sups>2</sups> = .09) in a linear multiple regression with three tested predictors and 19 predictors in total (<emph>β</emph> = .20, <emph>α</emph> = .05). The three tested predictors were the eye movement pattern as quantified using EMHMM, consistency of overall eye movement pattern as measured in overall entropy of the HMM, and consistency of the first fixation at the word as measured in the marginal entropy of the first fixation. We measured the overall entropy and marginal entropy of the first fixation separately since in the isolated word reading task, the first fixation involved locating the word stimulus for further processing, and thus its processing may be fundamentally different from the rest of the fixations during word reading. In addition, previous research has suggested that early eye fixations in a trial play a more important role in recognition performance than later fixations (albeit in face recognition; Chuk, Crookes, Hayward, Chan, & Hsiao, [<reflink idref="bib17" id="ref45">17</reflink>]; Hsiao & Cottrell, [<reflink idref="bib43" id="ref46">43</reflink>]). The other 16 predictors were language proficiency measures, accuracy and RT of verbal two‐back task, visuospatial two‐back task, RAN task, Navon task global and local processing trials, normalized accuracy and RT of Flanker task (see below for details), and horizontal and vertical VFs of the presented word (see Section 3.3 for details). All experiments were performed in accordance with the American Psychological Association ethical standards. Informed consent was obtained from all participants.</p> <hd id="AN0179878392-5">Materials</hd> <p>Participants performed a list of isolated word reading tasks, followed by a list of cognitive ability tests that were found to be related to reading performance (e.g., Chang, Wang, Cai, & Wang, [<reflink idref="bib14" id="ref47">14</reflink>]; Commodari, [<reflink idref="bib19" id="ref48">19</reflink>]; Franceschini, Bertoni, Gianesini, Gori, & Facoetti, [<reflink idref="bib26" id="ref49">26</reflink>]; Georgiou, Parrila, & Liao, [<reflink idref="bib29" id="ref50">29</reflink>]). The cognitive ability tests were used to control for individual differences in cognitive abilities in our data analysis.</p> <hd id="AN0179878392-6">Isolated word reading tasks</hd> <p>Participants performed lexical decision, word reading, and pseudoword reading tasks in separate blocks. In all reading tasks, the stimuli were chosen from standard validated tests for second language learners to ensure that the stimuli were at an appropriate difficulty level to assess their performance. The stimuli consisted of words of different lengths to form a representative set of stimuli rather than specific to a particular word length. During the tasks, we presented the word stimuli at the center of one of the four quadrants on the screen (randomly determined) to reflect that in real life readers may locate an isolated word from different directions. Although when words are viewed in the context of a sentence, they are read from left to right, we adopted this design to be consistent with previous studies in face or visual object recognition where initial scanning direction was typically counterbalanced (An & Hsiao, [<reflink idref="bib1" id="ref51">1</reflink>]; Chuk et al., [<reflink idref="bib17" id="ref52">17</reflink>]; Hsiao & Cottrell, [<reflink idref="bib43" id="ref53">43</reflink>]). In all reading tasks, stimuli were presented in black with a white background with a resolution of 1024×768, under a viewing distance of 51 cm. Each five letters occupied 1.5° of visual angle (Brysbaert, Vitu, & Schroyens, [<reflink idref="bib8" id="ref54">8</reflink>]). An EyeLink 1000 eye tracker (SR Research Ltd., Canada) was used. A forehead‐rest was used to reduce head movement. Recalibration was performed when the gaze position error was larger than 1° of visual angle (e.g., Feng, [<reflink idref="bib25" id="ref55">25</reflink>]; Que, Zheng, Hsiao, & Hu, [<reflink idref="bib68" id="ref56">68</reflink>]). A more stringent criterion such as 0.5° of visual angle may be used to enhance eye tracking accuracy in future work. EyeLink default settings for cognitive research were used in data acquisition. EMHMM was used to analyze eye movement data.</p> <hd id="AN0179878392-7">Lexical decision</hd> <p>Stimuli were from the Lexical Test for Advanced Learners of English (LexTALE; Lemhöfer & Broersma, [<reflink idref="bib56" id="ref57">56</reflink>]), designed to assess English proficiency as a second language. The lexical decision task has a moderate to good internal reliability: The split‐half reliabilities of average percentage of correct responses, Spearman−Brown corrected, was .814 in Dutch participants and .684 in Korean participants (Lemhöfer & Broersma, [<reflink idref="bib56" id="ref58">56</reflink>]). Each string of letters had 4−12 letters. There were 60 trials in total. The task was designed for English as second language learners with a wide variety of language backgrounds. Participants started with a central solid circle for drift correction, and then were presented with a string of letters at the center of one of the four quadrants (randomly determined) on the screen. They were asked to judge whether the presented stimulus was a real English word or not by pressing the corresponding keys, that is, to use the left hand to press the "F" key if the stimulus was a real English word and use the right hand to press the "J" key if the stimulus was a pseudoword. The stimulus was presented until response. Accuracy, RT of correct trials, and eye movements were recorded. Following the LexTALE scoring, accuracy was calculated by averaging the percentages correct for the word and nonword item types (Lemhöfer & Broersma, [<reflink idref="bib56" id="ref59">56</reflink>]), and similarly for RT of correct trials.</p> <hd id="AN0179878392-8">Word naming</hd> <p>Stimuli were selected from Letter‐Word Identification of Woodcock‐Johnson III Tests of Achievement (Woodcock, McGrew, & Mather, [<reflink idref="bib82" id="ref60">82</reflink>]). Each word had 5−14 letters. In this test, the stimuli were divided into seven groups for participants in different age groups: (<reflink idref="bib1" id="ref61">1</reflink>) preschool to kindergarten children; (<reflink idref="bib2" id="ref62">2</reflink>) Grade 1 children; (<reflink idref="bib3" id="ref63">3</reflink>) Grade 2 children; (<reflink idref="bib4" id="ref64">4</reflink>) Grade 3 to Grade 4 children; (<reflink idref="bib5" id="ref65">5</reflink>) Grade 5 to Grade 6 children; (<reflink idref="bib6" id="ref66">6</reflink>) Grade 7 to average adults; and (<reflink idref="bib7" id="ref67">7</reflink>) college and above average adults. We chose stimuli suitable for participants from college and above average adult level based on the age of our participants. There were 28 trials in total. Participants started with a central solid circle for drift correction, and then were presented with a word at the center of one of the four quadrants (randomly determined) on the screen. Participants were asked to read out the presented word to a microphone as quickly as possible. The stimulus disappeared when the microphone detected the onset of the pronunciation and the next trial started after the accuracy of the pronunciation was recorded by the experimenter. Accuracy (judged by the experimenter), RT of correct trials, and eye movements were recorded.</p> <hd id="AN0179878392-9">Pseudoword naming</hd> <p>Stimuli were selected from Word Attack of the Woodcock‐Johnson III Tests of Achievement (Woodcock et al., [<reflink idref="bib82" id="ref68">82</reflink>]). Each pseudoword had 2−11 letters. In this test, the stimuli were divided into two groups for participants in different age groups: (<reflink idref="bib1" id="ref69">1</reflink>) preschool to Grade 2 children; and (<reflink idref="bib2" id="ref70">2</reflink>) Grade 3 to adults. We chose stimuli suitable for participants from Grade 3 to adult level based on the age of our participants. There were 29 trials in total. Participants started with a central solid circle for drift correction, and then were presented with a pseudoword at the center of one of the four quadrants (randomly determined) on the screen. Participants were asked to pronounce the presented pseudoword to a microphone. The stimulus disappeared when the onset of the pronunciation was detected by the microphone and the next trial started after the accuracy of the pronunciation was recorded by the experimenter. Accuracy (judged by the experimenter), RT of correct trials, and eye movements were recorded.</p> <hd id="AN0179878392-10">Cognitive ability tests</hd> <p></p> <hd id="AN0179878392-11">Navon task for global/local attention</hd> <p>The Navon task measures participants' global and local attention (Navon, [<reflink idref="bib63" id="ref71">63</reflink>]). There were 128 trials in total. Each trial started with a drift correction at the center of the screen, and then participants were presented with a hierarchical letter pattern: a large letter consisted of small letters. Under a viewing distance of 51 cm, each large letter subtended a horizontal and vertical visual angle of 5.25° x 6.68°, with each small letter subtended a horizontal and vertical visual angle of .47° x .47°. Participants judged whether a target letter (either letter D or E) was presented regardless of whether it was at the global (large letter) or local (small letter) level. We measured the accuracy and RT of the trials where there was a target letter at the global and the local level separately as the measure of global and local information processing ability, respectively.</p> <hd id="AN0179878392-12">RAN for isolated familiar symbol identification</hd> <p>The RAN task measures the ability of identifying an isolated symbol (Denckla & Rudel, [<reflink idref="bib24" id="ref72">24</reflink>]; Siddaiah, Saldanha, Venkatesh, Ramachandra, & Padakannaya, [<reflink idref="bib75" id="ref73">75</reflink>]). Letter stimuli were used, and the size of the letter was consistent with that in the reading tasks. Each trial started with a fixation cross presented at the center of the screen. Participants then were presented with a single letter at the center of the screen and were asked to name it as fast as possible (Callens, Tops, & Brysbaert, [<reflink idref="bib10" id="ref74">10</reflink>]). The stimuli were presented until the response was detected. Their accuracy and RT were recorded.</p> <hd id="AN0179878392-13">Verbal and visuospatial two‐back for working memory</hd> <p>These tasks measure the ability of working memory (Lau, Eskes, Morrison, Rajda, & Spurr, [<reflink idref="bib55" id="ref75">55</reflink>]). There were 24 trials in each block, and three blocks in each task. In the verbal two‐back task, each trial started with a fixation cross at the center of the screen. Then, participants were presented with a single number at the center of the screen for 1000 ms, followed by a blank screen for 2500 ms. They were asked to judge whether the presented number was the same as the number presented two trials back. In the visuospatial two‐back task, each trial started with a fixation cross at the center of the screen. Then, participants were presented with a single symbol appearing at one out of 12 possible locations on the screen for 1000 ms, followed by a blank screen for 2500 ms. They were asked to judge whether the stimulus was at the same location as the one presented two trials back. Under a viewing distance of 30 cm, the stimulus presented in each trial subtended a vertical visual angle of 1.72°. Their accuracy and RT were measured.</p> <hd id="AN0179878392-14">Eriksen flanker task for selective attention</hd> <p>The flanker task measures the ability to attend to a target while suppressing surrounding incongruent information (Ridderinkhof, Band, & Logan, [<reflink idref="bib71" id="ref76">71</reflink>]). There were 60 trials in each block and two blocks in total. Each trial started with a fixation cross at the center of the screen. Participants then were presented with five arrows arranged horizontally at the center of the screen and were asked to judge the direction of the arrow at the center, which was flanked by two arrows each on the left and right. Under a viewing distance of 40 cm, the flanking arrows subtended a horizontal and vertical visual angle of 1.86° x 1.86°, and the target arrow subtended a horizontal and vertical visual angle of 1.29° x 1.29°. There were three conditions in the task: congruent, incongruent, and neutral. In the congruent condition, the flanking arrow direction was congruent with the target arrow. In the incongruent condition, the flanking arrow direction was incongruent with the target arrow. In the neutral condition, the flankers were diamond shapes instead of arrows and thus did not indicate any direction. Flanker effect on accuracy and RT was measured as (performance in incongruent trials – performance in congruent)/(performance in incongruent + performance in congruent).</p> <hd id="AN0179878392-15">Design</hd> <p>For each word reading task, we first examined their general eye movement behavior across different quadrants of the screen. To examine whether participants' initial fixation locations differed when the word stimulus was presented in different quadrants of the screen, we used ANOVA on the fixation locations relative to the word center in the x‐ and y‐coordinates separately with fixation number (first vs. second), vertical VF (upper vs. lower), and horizontal VF (left vs. right) as the independent variables. We also used ANOVA on the average number of fixations per trial with vertical VF (upper vs. lower), and horizontal VF (left vs. right) as the independent variables.</p> <p>We then used EMHMM to quantify participants' overall eye movement pattern and consistency across all trials. We used stepwise multiple regression to examine whether eye movement pattern and consistency measures were among the best predictors for accuracy and RT of the correct trials in the lexical decision task, word naming task, and pseudoword naming task. In addition to eye movement pattern and consistency measures, the potential predictors included accuracy in the word reading tasks as language proficiency measures (i.e., accuracies of the lexical decision, word naming, and pseudoword naming tasks), verbal two‐back accuracy and RT for verbal working memory, visuospatial two‐back accuracy and RT for visuospatial working memory, flanker effect on accuracy and RT for selective attention, RAN accuracy and RT for isolated familiar symbol identification, global and local information processing abilities as measured in the Navon task accuracy and RT, and horizontal and vertical VFs of the presented word. More specifically, lexical decision accuracy was used as a language proficiency predictor for the regression models of the word naming and pseudoword naming performance (including accuracy and RT), whereas word naming and pseudoword naming accuracy were used as language proficiency predictors for the regression models of the lexical decision performance (including accuracy and RT). In addition to these language proficiency and cognitive ability measures, since presenting VF may affect visual word recognition performance, we included horizontal VF and vertical VF as two dichotomous categorical variables in the stepwise regression model as potential predictors.</p> <p>The EMHMM toolbox (Chuk et al., [<reflink idref="bib15" id="ref77">15</reflink>]) was used to provide a quantitative measure of eye movement pattern and consistency that takes both spatial and temporal dimensions of eye movements into account (see below).</p> <hd id="AN0179878392-16">Procedure</hd> <p>Participants first completed a demographic background questionnaire, Language Experience and Proficiency Questionnaire (LEAP‐Q; Marian, Blumenfeld, & Kaushanskaya, [<reflink idref="bib61" id="ref78">61</reflink>]). A "hole‐in‐the‐card" dominant eye test was carried out to assess eye dominance for monocular eye tracking during the reading tasks (Barbeito, [<reflink idref="bib4" id="ref79">4</reflink>]; Rodriguez‐Lopez et al., [<reflink idref="bib72" id="ref80">72</reflink>]). More specifically, participants used both hands to form a triangle and were asked to look through the triangle at a sticker on the wall using one eye. If they reported that they could not see the sticker when the right eye was closed, their dominant eye was the right eye, and vice versa.</p> <p>Participants were then asked to perform the reading tasks first. In each reading task, they started with a 9‐point calibration and validation procedure. During the validation, the experimenter ensured that each gaze position error was smaller than 1° visual angle. Afterward, they performed the cognitive tests in the following order: verbal and visuospatial two‐back, flanker task, RAN task, and Navon task.</p> <hd id="AN0179878392-17">Eye movement data analysis using EMHMM</hd> <p>Here, we used a data‐driven machine learning model‐based method, EMHMM (Chuk et al., [<reflink idref="bib15" id="ref81">15</reflink>]), to obtain quantitative measures of individual eye movement patterns and consistency in word reading. This method takes individual differences in both temporal and spatial dimensions of eye movements into account. More specifically, in EMHMM, each participant' eye movements in a visual task were summarized using an HMM (a type of time‐series statistical model in machine learning), including person‐specific ROIs and transition probabilities among these ROIs. Parameters of the HMM were estimated from the participant's eye movement data using the variational Bayesian expectation maximization algorithm (Bishop, [<reflink idref="bib6" id="ref82">6</reflink>]). Accordingly, in the current study, each participant's eye movements in a word reading task were summarized using an HMM. Then, following previous studies using EMHMM (e.g., An & Hsiao, [<reflink idref="bib1" id="ref83">1</reflink>]; Chan et al., [<reflink idref="bib12" id="ref84">12</reflink>]; Chan, Barry, Chan, & Hsiao, [<reflink idref="bib11" id="ref85">11</reflink>]; Chuk et al., [<reflink idref="bib15" id="ref86">15</reflink>]; Chuk et al., [<reflink idref="bib16" id="ref87">16</reflink>]; Chuk et al., [<reflink idref="bib17" id="ref88">17</reflink>]; Coutrot, Hsiao, & Chan, [<reflink idref="bib20" id="ref89">20</reflink>]; Hsiao et al., [<reflink idref="bib38" id="ref90">38</reflink>]; Liao et al., [<reflink idref="bib58" id="ref91">58</reflink>]; Zhang, Chan, Lau, & Hsiao, [<reflink idref="bib83" id="ref92">83</reflink>]), for each reading task, we clustered all individual HMMs into two groups according to their similarities through the variational hierarchical expectation maximization algorithm (Coviello, Lanckriet, & Chan, [<reflink idref="bib23" id="ref93">23</reflink>]; [<reflink idref="bib22" id="ref94">22</reflink>]) to reveal two representative eye movement patterns. The similarity between an individual's eye movement pattern and a representative eye movement pattern then could be assessed using the log‐likelihood of the individual's eye movement data being generated by the representative pattern HMM (for more details about the EMHMM methodology, please refer to Chuk et al., [<reflink idref="bib15" id="ref95">15</reflink>]; Chuk et al., [<reflink idref="bib17" id="ref96">17</reflink>]). This quantitative measure of eye movement pattern allowed us to examine whether eye movement pattern could account for an individual's reading performance. In addition, consistency of an eye movement pattern across trials in the task can be quantified by the entropy of the corresponding HMM (Hsiao et al., [<reflink idref="bib36" id="ref97">36</reflink>]; Hsiao et al., [<reflink idref="bib40" id="ref98">40</reflink>]; Hsiao et al., [<reflink idref="bib46" id="ref99">46</reflink>]). Entropy is a measure of predictivity; higher entropy indicates higher randomness (Cover & Thomas, [<reflink idref="bib21" id="ref100">21</reflink>]) and thus less consistent eye movements across trials. Here, we used the overall entropy of the HMM to measure the overall consistency of the participant's eye movements across trials. In addition, we calculated the marginal entropy of the first fixation to measure the consistency of the participant's first fixation in a trial in the given reading task. We measured the marginal entropy of the first fixation, in addition to the overall entropy, since previous studies have suggested that early fixations in a trial play a more important role in the recognition of visual stimuli (Chuk et al., [<reflink idref="bib16" id="ref101">16</reflink>]; Hsiao & Cottrell, [<reflink idref="bib43" id="ref102">43</reflink>]). In addition, the first fixation in a trial involved locating the word stimulus from somewhere outside the word (here the center of the screen), and thus may be functionally different from other fixations in the trial.</p> <p>When training individual HMMs, we set the range of possible number of ROIs to be 1−10. EMHMM uses a variational Bayesian approach to determine the optimal number of ROIs from the preset range for each model. Each individual model with a different preset number of ROIs was trained for 300 times, and the model with the highest data log‐likelihood was used. When we clustered individual HMMs to discover two representative, contrasting eye movement patterns in each reading task, the optimal number of ROIs used for creating the two representative HMMs was decided by the Bayesian method (Lan, Liu, Hsiao, Yu, & Chan, [<reflink idref="bib53" id="ref103">53</reflink>]). The clustering algorithm was run for 300 times, and the result with the highest data log‐likelihood was used. To quantify an individual's eye movement pattern along the dimension of the two representative, contrasting eye movement patterns, pattern A and pattern B, following previous studies (e.g., An & Hsiao, [<reflink idref="bib1" id="ref104">1</reflink>]; Chan et al., [<reflink idref="bib12" id="ref105">12</reflink>]; Hsiao et al., [<reflink idref="bib38" id="ref106">38</reflink>]; Hsiao et al., [<reflink idref="bib46" id="ref107">46</reflink>]), we defined A‐B scale as (A – B)/(|A| + |B|), where A refers to the individual's data log‐likelihood given pattern A HMM, and B for that given pattern B HMM. A more positive value indicated higher similarity to pattern A.</p> <hd id="AN0179878392-18">Results</hd> <p>Table 1 shows participants' performance in different word reading tasks.</p> <p>1 Table Descriptive statistics of accuracy and RT in isolated word reading tasks</p> <p> <ephtml> <table><thead><tr><th /><th /><th align="center">Mean</th><th /></tr><tr><th /><th /><th>95% CI</th><th /></tr><tr><th /><th align="center">Mean</th><th>[LL, UL]</th><th align="center">SD</th></tr></thead><tbody><tr><td>Lexical decision accuracy</td><td>.749</td><td>[.731, .767]</td><td>.104</td></tr><tr><td>Lexical decision RT</td><td>1431.177 ms</td><td>[1356.870 ms, 1505.485 ms]</td><td>424.847 ms</td></tr><tr><td>Word naming accuracy</td><td>.647</td><td>[.621, .673]</td><td>.149</td></tr><tr><td>Word naming RT</td><td>1956.868 ms</td><td>[1811.087 ms, 2102.650 ms]</td><td>833.490 ms</td></tr><tr><td>Pseudoword naming accuracy</td><td>.748</td><td>[.723, .773]</td><td>.144</td></tr><tr><td>Pseudoword naming RT</td><td>2070.189 ms</td><td>[1935.732 ms, 2204.646 ms]</td><td>768.745 ms</td></tr></tbody></table> </ephtml> </p> <hd id="AN0179878392-19">Lexical decision</hd> <p>We first examined whether participants' initial fixation locations differed when the word stimulus was presented in different quadrants of the screen (Fig. 1). The results showed that in the x‐coordinate, a main effect of horizontal VF was found, <emph>F</emph>(<reflink idref="bib1" id="ref108">1</reflink>, 126) = 32.233, <emph>p</emph> < .001, <emph>η</emph><subs>p</subs><sups>2</sups> = .085: participants' first two fixations were to the left of the word center when the word was presented in the left VF, and were to the right of the word center when it was presented in the right VF. This result suggested that participants generally overshoot the word horizontally if we assumed they targeted the word center. This effect interacted with fixation number, <emph>F</emph>(<reflink idref="bib1" id="ref109">1</reflink>, 126) = 9.467, <emph>p</emph> = .003, <emph>η</emph><subs>p</subs><sups>2</sups> = .001: participants' first fixations were more to the right and closer to the word center than their second fixations in left VF, <emph>t</emph>(<reflink idref="bib126" id="ref110">126</reflink>) = 2.992, <emph>p</emph> = .017, <emph>d = </emph>.223, whereas their first and second fixations did not differ in the right VF, <emph>t</emph>(<reflink idref="bib126" id="ref111">126</reflink>) = −1.435, <emph>p</emph> = .480. In the y‐coordinate, a main effect of fixation number was found, <emph>F</emph>(<reflink idref="bib1" id="ref112">1</reflink>, 126) = 23.443, <emph>p</emph> < .001, <emph>η</emph><subs>p</subs><sups>2</sups> = .002: participants' second fixation was higher and closer to the word center than the first fixation. Assuming participants aimed to fixate the word center, this result suggested that they used the second fixation to correct the first fixation location. A marginal main effect of vertical VF was also found, <emph>F</emph>(<reflink idref="bib1" id="ref113">1</reflink>, 126) = 3.004, <emph>p</emph> = .085, <emph>η</emph><subs>p</subs><sups>2</sups> = .007: participants' first two fixations were lower in the lower VF than in the upper VF. We then examined the average number of fixations per trial when the word stimulus was presented in different quadrants of the screen, we found no effect of vertical VF, <emph>F</emph>(<reflink idref="bib1" id="ref114">1</reflink>, 126) = .036, <emph>p</emph> = .849, horizontal VF, <emph>F</emph>(<reflink idref="bib1" id="ref115">1</reflink>, 126) = .007, <emph>p</emph> = .933, or their interaction, <emph>F</emph>(<reflink idref="bib1" id="ref116">1</reflink>, 126) = .512, <emph>p</emph> = .476, suggesting that participants made a similar number of fixations per trial across the quadrants.</p> <p> <img src="https://imageserver.ebscohost.com/img/embimages/rdk/CGN/01sep24/cogs13489-fig-0001.jpg?ephost1=dGJyMNXb4kSepq84yOvqOLCmsE6epq5Srqa4SK6WxWXS" alt="cogs13489-fig-0001.jpg" title="1 Participants' first and second fixation locations in (a) x‐coordinate and (b) y‐coordinate in lexical decision when the word stimulus was presented in different quadrants of the screen (*p < .05; **p < .01; ***p < .001). The error bars represent the standard deviations." /> </p> <p></p> <p>Fig. 2 shows the two representative eye movement patterns. In pattern A, participants most likely started with a fixation in the Red ROI around the word center and stayed looking at the same ROI afterward. Sometimes they looked below the word center (Cyan) or slightly to the word beginning (Blue) and stayed there. In pattern B, in contrast, participants always looked around the region between the word beginning and center, the typical OVP for skilled readers. Seventy‐six participants were classified as using pattern A, whereas 52 participants were classified as using pattern B. We used Kullback‐Leibler divergence estimation to examine whether the two representative patterns significantly differed from each other (e.g., Hsiao et al., [<reflink idref="bib45" id="ref117">45</reflink>]; Hsiao et al., [<reflink idref="bib46" id="ref118">46</reflink>]). More specifically, we calculated the log‐likelihood of participants' data being generated by pattern A and pattern B HMMs and used ANOVA to test whether data from participants adopting pattern A were more likely to be generated by pattern A HMM than pattern B HMM, and vice versa for data from participants adopting pattern B. The results showed that the two patterns significantly differed: data from participants adopting pattern A were more likely to be generated by pattern A HMM than pattern B HMM, and vice versa for data from participants adopting pattern B, <emph>F</emph>(<reflink idref="bib1" id="ref119">1</reflink>, 126) = 162.4, <emph>p</emph> < .001, <emph>η</emph><subs>p</subs><sups>2</sups> = .563, 90% CI [.4682, .6322]. Participants using the two patterns did not differ in average number of fixations per trial, <emph>t</emph>(<reflink idref="bib126" id="ref120">126</reflink>) = 1.276, <emph>p</emph> = .204, <emph>d</emph> = .230, fixation duration, <emph>t</emph>(<reflink idref="bib126" id="ref121">126</reflink>) = 1.938, <emph>p</emph> = .055, <emph>d</emph> = .349, or saccade length, <emph>t</emph>(<reflink idref="bib126" id="ref122">126</reflink>) = −.374, <emph>p</emph> = .709, <emph>d</emph> = −.067. The two representative patterns did not differ in overall entropy, <emph>t</emph>(<reflink idref="bib126" id="ref123">126</reflink>) = 1.022, <emph>p</emph> = .309, <emph>d</emph> = .184. However, they differed in marginal entropy of the first fixation: pattern B's first fixation, which was at the OVP, was more consistent than that in pattern A, <emph>t</emph>(<reflink idref="bib126" id="ref124">126</reflink>) = 2.175, <emph>p</emph> = .032, <emph>d</emph> = .391. Both eye movement pattern (A‐B scale) and marginal entropy of first fixation were correlated with lexical decision RT (Table 2).</p> <p> <img src="https://imageserver.ebscohost.com/img/embimages/rdk/CGN/01sep24/cogs13489-fig-0002.jpg?ephost1=dGJyMNXb4kSepq84yOvqOLCmsE6epq5Srqa4SK6WxWXS" alt="cogs13489-fig-0002.jpg" title="2 Representative eye movement patterns in lexical decisions. Ellipses show ROIs as 2‐D Gaussian emissions. Table on the right shows transition probabilities among the ROIs. Priors show the probabilities that a fixation sequence starts from the ellipse. Small image shows ROI assignments of fixations and the corresponding heatmap. ROI 4 (pink) in pattern A and ROI 5 (Cyan) in pattern B captured outlier fixations. In pattern B, ROIs 1−4 were overlapped since the same number of ROIs were used to generate the two representative patterns to facilitate meaningful comparisons during clustering." /> </p> <p></p> <p>2 Table Intercorrelations of the potential predictors in the stepwise multiple regression predicting lexical decision accuracy and RT</p> <p> <ephtml> <table><thead><tr valign="bottom"><th align="left" /><th align="center">Lexical decision A‐B scale</th><th align="center">Lexical decision OE</th><th align="center">Lexical decision ME</th><th align="center">Word naming accuracy</th><th align="center">Pseudoword naming accuracy</th><th align="center">Verbal two‐back accuracy</th><th align="center">Verbal two‐back RT</th><th align="center">Visuospatial two‐back accuracy</th><th align="center">Visuospatial two‐back RT</th><th align="center">RAN accuracy</th><th align="center">RAN RT</th><th align="center">Navon task global trial accuracy</th><th align="center">Navon task global trial RT</th><th align="center">Navon task local trial accuracy</th><th align="center">Navon task local trial RT</th><th align="center">Flanker effect in accuracy</th><th align="center">Flanker effect in RT</th></tr></thead><tbody><tr><td>Lexical decision A‐B scale</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td></tr><tr><td>Lexical decision OE</td><td>.2480002</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td></tr><tr><td>Lexical decision ME</td><td>.3360002</td><td>.6150002</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td></tr><tr><td>Word naming accuracy</td><td>−0.065</td><td>0.077</td><td>0</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td></tr><tr><td>Pseudoword naming accuracy</td><td>−0.021</td><td>0.101</td><td>0.009</td><td>.7430002</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td></tr><tr><td>Verbal two‐back accuracy</td><td>−0.067</td><td>−0.166</td><td>−0.131</td><td>−0.015</td><td>0.086</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td></tr><tr><td>Verbal two‐back RT</td><td>0.037</td><td>0.053</td><td>−0.011</td><td>0.143</td><td>0.052</td><td>−0.157</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td></tr><tr><td>Visuospatial two‐back accuracy</td><td>−0.012</td><td>−.2170002</td><td>−0.171</td><td>−0.106</td><td>−0.011</td><td>.6910002</td><td>−0.133</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td></tr><tr><td>Visuospatial two‐back RT</td><td>0.01</td><td>0.065</td><td>0.041</td><td>0.158</td><td>0.112</td><td>−0.009</td><td>.7550002</td><td>−0.157</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td></tr><tr><td>RAN accuracy</td><td>−0.077</td><td>−0.081</td><td>−0.132</td><td>−0.061</td><td>−0.07</td><td>−0.02</td><td>0.021</td><td>−0.029</td><td>0.052</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td></tr><tr><td>RAN RT</td><td>−.2590002</td><td>−0.049</td><td>0.015</td><td>−0.09</td><td>−0.018</td><td>−0.108</td><td>−0.056</td><td>−0.1</td><td>−0.013</td><td>−0.037</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td></tr><tr><td>Navon task global trial accuracy</td><td>−0.045</td><td>−0.097</td><td>0.015</td><td>−0.005</td><td>−0.025</td><td>0.172</td><td>0.003</td><td>0.171</td><td>−0.053</td><td>0.11</td><td>−0.106</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td></tr><tr><td>Navon task global trial RT</td><td>.1850002</td><td>−0.087</td><td>−0.142</td><td>0.069</td><td>0.036</td><td>−0.013</td><td>0.139</td><td>−0.003</td><td>0.105</td><td>0.009</td><td>−0.131</td><td>−.4100002</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td></tr><tr><td>Navon task local trial accuracy</td><td>0.079</td><td>−0.089</td><td>0.046</td><td>−0.077</td><td>−0.078</td><td>0.139</td><td>−0.073</td><td>0.121</td><td>−0.146</td><td>0.055</td><td>−0.089</td><td>.8190002</td><td>−.3310002</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td></tr><tr><td>Navon task local trial RT</td><td>0.172</td><td>0.053</td><td>0.138</td><td>0.054</td><td>−0.106</td><td>−.1870002</td><td>.4260002</td><td>−.3400002</td><td>.3190002</td><td>−0.028</td><td>−0.171</td><td>0.022</td><td>.3710002</td><td>0.082</td><td align="center">–</td><td align="center">–</td><td align="center">–</td></tr><tr><td>Flanker effect in accuracy</td><td>.1760002</td><td>−0.006</td><td>0.125</td><td>−0.096</td><td>−0.132</td><td>−0.047</td><td>.2910002</td><td>−0.007</td><td>.2920002</td><td>−0.036</td><td>−0.068</td><td>0.096</td><td>0.073</td><td>0.106</td><td>.2620002</td><td align="center">–</td><td align="center">–</td></tr><tr><td>Flanker effect in RT</td><td>−0.025</td><td>−0.101</td><td>−0.114</td><td>−0.034</td><td>−0.035</td><td>0.039</td><td>−0.126</td><td>0.062</td><td>−0.139</td><td>−0.024</td><td>0.099</td><td>0.011</td><td>−0.098</td><td>−0.027</td><td>−0.13</td><td>−0.165</td><td align="center">–</td></tr></tbody></table> </ephtml> </p> <p>1 Abbreviations: ME, marginal entropy of the first fixation; OE, overall entropy.</p> <p>2 <sups>*</sups><emph>p</emph> < .05; **<emph>p</emph> < .01; ***<emph>p</emph> < .001.</p> <p>In a separate analysis, we examined whether participants' eye movement pattern when the stimulus was presented in different quadrants of the screen was generally classified as the same pattern as their overall eye movement pattern derived by summarizing trials across all quadrants. Specifically, for each participant, we classified the eye movement data in each quadrant into pattern A or pattern B according to the two representative pattern models, and examined whether the resulting pattern is the same as the participant's overall pattern. For each quadrant separately, we then calculated the number of participants using the same pattern as their overall pattern and the number of participants using a different pattern from their overall pattern, and used a chi‐square test to examine whether this distribution was significantly different from chance (i.e., an equal distribution between the two cases. See e.g., Chuk et al., [<reflink idref="bib16" id="ref125">16</reflink>]). The results showed that there were more participants adopting the same pattern as their overall pattern than chance in the upper‐left quadrant (99/128), <emph>χ</emph><sups>2</sups>(<reflink idref="bib1" id="ref126">1</reflink>) = 38.281, <emph>p</emph> < .001, in the upper‐right quadrant (105/128), <emph>χ</emph><sups>2</sups>(<reflink idref="bib1" id="ref127">1</reflink>) = 52.531, <emph>p</emph> < .001, and in the lower‐right quadrant (96/128), <emph>χ</emph><sups>2</sups>(<reflink idref="bib1" id="ref128">1</reflink>) = 32.000, <emph>p</emph> < .001, but not in the lower‐left quadrant (64/128), <emph>χ</emph><sups>2</sups>(<reflink idref="bib1" id="ref129">1</reflink>) = 0.000, <emph>p</emph> = 1.000.</p> <p>Stepwise multiple regression was used to examine which variables significantly predicted lexical decision accuracy and RT. The intercorrelations among the potential predictors are summarized in Table 2. The results showed that lexical decision accuracy was best predicted by visuospatial working memory, local visual processing efficiency, and word phonological knowledge, but not by any eye movement measures. In contrast, lexical decision RT was best predicted by eye movement pattern in addition to local visual processing efficiency, word phonological knowledge, selective attention, and visuospatial working memory: higher similarity to the pattern that focused at the OVP (pattern B) was associated with shorter lexical decision RT. All variables had strong split‐half reliability (Table 3).</p> <p>3 Table Correlations and stepwise regressions of lexical decision task</p> <p> <ephtml> <table><thead><tr><th>Variable</th><th align="center"><italic>r</italic></th><th align="center"><italic>β</italic></th><th align="center"><italic>R</italic><sup>2</sup></th><th align="center">Δ<italic>R</italic><sup>2</sup></th><th align="center"><italic>r</italic><sub>SB</sub></th><th align="center"><italic>VIF</italic></th></tr></thead><tbody><tr><td>Lexical decision accuracy</td><td align="left" /><td align="left" /><td>.152</td><td align="left" /><td>.72</td><td align="left" /></tr><tr><td align="left">Word naming accuracy</td><td>.3640001</td><td>.3760001</td><td align="left" /><td>.1320001</td><td>.99</td><td>1.009</td></tr><tr><td align="left">Navon task local trial RT</td><td>.076</td><td>.1260001</td><td align="left" /><td>.0070001</td><td>.85</td><td>1.134</td></tr><tr><td align="left">Visuospatial two‐back accuracy</td><td>.048</td><td>.1200001</td><td align="left" /><td>.0130001</td><td>.90</td><td>1.140</td></tr><tr><td>Lexical decision RT</td><td align="left" /><td align="left" /><td>.270</td><td align="left" /><td>.94</td><td align="left" /></tr><tr><td align="left">Navon task local trial RT</td><td>.4070001</td><td>.4640001</td><td align="left" /><td>.1660001</td><td>.85</td><td>1.187</td></tr><tr><td align="left">Visuospatial two‐back accuracy</td><td>.086</td><td>.2130001</td><td align="left" /><td>.0570001</td><td>.90</td><td>1.174</td></tr><tr><td align="left">Lexical decision A‐B scale</td><td>.1930001</td><td>.1300001</td><td align="left" /><td>.0170001</td><td>.99</td><td>1.019</td></tr><tr><td align="left">Word naming accuracy</td><td>−.1410001</td><td>−.1070001</td><td align="left" /><td>.0120001</td><td>.91</td><td>1.011</td></tr><tr><td align="left">Navon task local trial accuracy</td><td>.1800001</td><td>.1090001</td><td align="left" /><td>.0110001</td><td>.93</td><td>1.034</td></tr><tr><td align="left">Flanker effect in RT</td><td>.032</td><td>.0810001</td><td align="left" /><td>.0060001</td><td>.95</td><td>1.020</td></tr><tr><td>Lexical decision A‐B scale</td><td align="left" /><td align="left" /><td>.081</td><td align="left" /><td>.91</td><td align="left" /></tr><tr><td align="left">RAN RT</td><td>−.1880001</td><td>−.1810001</td><td align="left" /><td>.0350001</td><td>.96</td><td>1.005</td></tr><tr><td align="left">Horizontal VF</td><td>.1830001</td><td>.1830001</td><td align="left" /><td>.0330001</td><td align="center">–</td><td>1.000</td></tr><tr><td align="left">Flanker effect in accuracy</td><td>.1240001</td><td>.1110001</td><td align="left" /><td>.0120001</td><td>.70</td><td>1.005</td></tr></tbody></table> </ephtml> </p> <p>3 <sups>*</sups><emph>p</emph> < .05; **<emph>p</emph> < .01; ***<emph>p</emph> < .001.</p> <p>Since eye movement pattern predicted lexical decision RT, a follow‐up analysis was conducted to examine what factors, including cognitive abilities and presented VF, best predicted eye movement pattern in A‐B scale. The results showed that longer RAN RT, better selective attention, and left VF presentation was associated with higher similarity to the pattern that focused at the OVP (pattern B). All variables had strong split‐half reliability (Table 3).</p> <hd id="AN0179878392-22">Word naming</hd> <p>We first examined whether participants' initial fixation locations differed when the word stimulus was presented in different quadrants of the screen (Fig. 3). The results showed that in the x‐coordinate, there was a main effect of horizontal VF, <emph>F</emph>(<reflink idref="bib1" id="ref130">1</reflink>, 127) = 10.897, <emph>p</emph> = .001, <emph>η</emph><subs>p</subs><sups>2</sups> = .024: participants' first two fixations were to the left of the word center in the left VF, and were at the word center in the right VF. This effect interacted with fixation number, <emph>F</emph>(<reflink idref="bib1" id="ref131">1</reflink>, 127) = 4.312, <emph>p</emph> = .040, <emph>η</emph><subs>p</subs><sups>2</sups> = .001: the horizontal VF effect was stronger in the second fixation, <emph>t</emph>(<reflink idref="bib127" id="ref132">127</reflink>) = −3.764, <emph>p</emph> = .001, <emph>d</emph> = .465, as compared with the first fixation, <emph>t</emph>(<reflink idref="bib127" id="ref133">127</reflink>) = −2.486, <emph>p</emph> = .067, <emph>d</emph> = .310. In the y‐coordinate, a main effect of fixation number was found, <emph>F</emph>(<reflink idref="bib1" id="ref134">1</reflink>, 127) = 25.291, <emph>p</emph> < .001, <emph>η</emph><subs>p</subs><sups>2</sups> = .004: participants' second fixation was higher and closer to the word center than the first fixations. This effect interacted with vertical VF, <emph>F</emph>(<reflink idref="bib1" id="ref135">1</reflink>, 127) = 6.745, <emph>p</emph> = .011, <emph>η</emph><subs>p</subs><sups>2</sups> = .001: the fixation number effect was observed only in the upper VF, <emph>t</emph>(<reflink idref="bib127" id="ref136">127</reflink>) = 6.594, <emph>p</emph> < .001, <emph>d = </emph>.456, but not in the lower VF, <emph>t</emph>(<reflink idref="bib127" id="ref137">127</reflink>) = 1.541, <emph>p</emph> = .416 (Fig. 3). As for the average number of fixations per trial, there was no main effect of vertical VF, <emph>F</emph>(<reflink idref="bib1" id="ref138">1</reflink>, 127) = .380, <emph>p</emph> = .539, horizontal VF, <emph>F</emph>(<reflink idref="bib1" id="ref139">1</reflink>, 127) = 2.421, <emph>p</emph> = .122, or their interaction, <emph>F</emph>(<reflink idref="bib1" id="ref140">1</reflink>, 127) = .056, <emph>p</emph> = .813, suggesting that participants made a similar number of fixations per trial across the quadrants.</p> <p> <img src="https://imageserver.ebscohost.com/img/embimages/rdk/CGN/01sep24/cogs13489-fig-0003.jpg?ephost1=dGJyMNXb4kSepq84yOvqOLCmsE6epq5Srqa4SK6WxWXS" alt="cogs13489-fig-0003.jpg" title="3 Participants' first and second fixation locations in (a) x‐coordinate and (b) y‐coordinate in word naming when the word stimulus was presented in different quadrants of the screen: (*p < .05; **p < .01; ***p < .001). The error bars represent the standard deviations." /> </p> <p></p> <p>Fig. 4 shows the two representative eye movement patterns. In pattern A, participants most frequently looked at the ROIs centered at either the word beginning (Red), to the right of the word center (Green), or below the word center (Blue), and stayed in the same ROI afterward. In contrast, in pattern B, participants mainly looked at the ROIs centered at either the word center (Pink) or the area below the word center (Red/Green/Blue). Sixty‐one participants were classified as using pattern A, whereas 67 participants were classified as using pattern B. The two patterns significantly differed: data from participants adopting pattern A were more likely to be generated by pattern A HMM than pattern B HMM, and vice versa for data from participants adopting pattern B, <emph>F</emph>(<reflink idref="bib1" id="ref141">1</reflink>, 126) = 103.16, <emph>p</emph> < .001, <emph>η</emph><subs>p</subs><sups>2</sups> = .450, 90% CI [.3437, .5332]. Participants adopting the two patterns did not differ in average number of fixations per trial, <emph>t</emph>(<reflink idref="bib126" id="ref142">126</reflink>) = .565, <emph>p</emph> = .573, <emph>d</emph> = .100, or saccade length, <emph>t</emph>(<reflink idref="bib126" id="ref143">126</reflink>) = 1.890, <emph>p</emph> = .061, <emph>d</emph> = .337. However, participants adopting pattern A had shorter fixation duration than participants adopting pattern B, <emph>t</emph>(<reflink idref="bib126" id="ref144">126</reflink>) = −2.371, <emph>p</emph> = .019, <emph>d</emph> = −.416. Pattern A had higher overall entropy, <emph>t</emph>(<reflink idref="bib126" id="ref145">126</reflink>) = 1.987, <emph>p</emph> = .049, <emph>d</emph> = .354, and marginal entropy of the first fixation than the pattern B, <emph>t</emph>(<reflink idref="bib126" id="ref146">126</reflink>) = 2.713, <emph>p</emph> = .008, <emph>d</emph> = .480.</p> <p> <img src="https://imageserver.ebscohost.com/img/embimages/rdk/CGN/01sep24/cogs13489-fig-0004.jpg?ephost1=dGJyMNXb4kSepq84yOvqOLCmsE6epq5Srqa4SK6WxWXS" alt="cogs13489-fig-0004.jpg" title="4 Representative eye movement patterns in word naming. Ellipses show ROIs as 2‐D Gaussian emissions. Table on the right shows transition probabilities among the ROIs. Priors show the probabilities that a fixation sequence starts from the ellipse. Small image shows the ROI assignment of fixations and the corresponding heatmap." /> </p> <p></p> <p>Similar to the lexical decision task, in a separate analysis, we examined whether participants' eye movement pattern when the stimulus was presented in different quadrants of the screen was generally the same as their overall eye movement pattern derived by summarizing trials across all quadrants, by using a chi‐square test to compare the distribution of participants using same versus different patterns against the chance distribution in the four quadrants separately. The results showed that there were more participants adopting the same pattern as their overall pattern than chance only in the upper‐left quadrant (92/128), <emph>χ</emph><sups>2</sups>(<reflink idref="bib1" id="ref147">1</reflink>) = 24.500, <emph>p</emph> < .001, but not in the upper‐right quadrant (61/128), <emph>χ</emph><sups>2</sups>(<reflink idref="bib1" id="ref148">1</reflink>) = .281, <emph>p</emph> = .596, in the lower‐left quadrant (61/128), <emph>χ</emph><sups>2</sups>(<reflink idref="bib1" id="ref149">1</reflink>) = .281, <emph>p</emph> = .596, or in the lower‐right quadrant (61/128), <emph>χ</emph><sups>2</sups>(<reflink idref="bib1" id="ref150">1</reflink>) = .281, <emph>p</emph> = .596.</p> <p>Stepwise multiple regression was used to examine which variables significantly predicted word naming accuracy and RT. The intercorrelations among the potential predictors are summarized in Table 4. The results showed that word naming accuracy was best predicted by lexical decision accuracy, working memory ability, and local visual processing efficiency, but not by any eye movement measure. In contrast, word naming RT was best predicted by consistency of the first fixation in a naming trial in addition to lexical decision accuracy, selective attention, and visuospatial working memory: a more consistent first fixation location was associated with shorter word naming RT. All variables had strong split‐half reliability (Table 5).</p> <p>4 Table Intercorrelations of the potential predictors in the stepwise multiple regression predicting word naming accuracy and RT</p> <p> <ephtml> <table><thead><tr valign="bottom"><th align="left" /><th align="center">Word naming A‐B scale</th><th align="center">Word naming OE</th><th align="center">Word naming ME</th><th align="center">Lexical decision accuracy</th><th align="center">Verbal two‐back accuracy</th><th align="center">Verbal two‐back RT</th><th align="center">Visuospatial two‐back accuracy</th><th align="center">Visuospatial two‐back RT</th><th align="center">RAN accuracy</th><th align="center">RAN RT</th><th align="center">Navon task global trial accuracy</th><th align="center">Navon task global trial RT</th><th align="center">Navon task local trial accuracy</th><th align="center">Navon task local trial RT</th><th align="center">Flanker effect in accuracy</th><th align="center">Flanker effect in RT</th></tr></thead><tbody><tr><td>Word naming A‐B scale</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td>−0.024</td><td>−0.016</td><td>−0.092</td><td>0.091</td><td>0.014</td><td>−0.108</td><td>−0.005</td><td>0.146</td><td>−0.012</td><td>−0.032</td><td>0.074</td><td>−0.096</td></tr><tr><td>Word naming OE</td><td>.2230002</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td>−.2530002</td><td>0.085</td><td>−.2040002</td><td>0.149</td><td>0.007</td><td>−0.137</td><td>−0.087</td><td>−0.014</td><td>−0.09</td><td>−0.002</td><td>0.07</td><td>−0.062</td></tr><tr><td>Word naming ME</td><td>.2130002</td><td>.5660002</td><td align="center">–</td><td align="center">–</td><td>−0.008</td><td>−0.152</td><td>0.028</td><td>−0.057</td><td>−0.025</td><td>−0.124</td><td>0.039</td><td>−0.11</td><td>0.09</td><td>0.006</td><td>0.158</td><td>−.2080002</td></tr><tr><td>Lexical decision accuracy</td><td>−0.133</td><td>0.044</td><td>−0.083</td><td align="center">–</td><td>0.07</td><td>.1990002</td><td>0.057</td><td>.1790002</td><td>−0.015</td><td>−0.12</td><td>0.051</td><td>0.044</td><td>0.003</td><td>0.108</td><td>0.046</td><td>−0.093</td></tr></tbody></table> </ephtml> </p> <ulist> <item>4 Abbreviations: ME, marginal entropy of the first fixation; OE, overall entropy.</item> <item>5 <sups>*</sups><emph>p</emph> < .05; **<emph>p</emph> < .01; ***<emph>p</emph> < .001. The rest of the intercorrelations were presented in Table 2.</item> <item>5 Table Correlations and stepwise regressions of word naming task</item> </ulist> <p> <ephtml> <table><thead><tr><th>Variable</th><th align="center"><italic>r</italic></th><th align="center"><italic>β</italic></th><th align="center"><italic>R</italic><sup>2</sup></th><th align="center">Δ<italic>R</italic><sup>2</sup></th><th align="center"><italic>r</italic><sub>SB</sub></th><th align="center"><italic>VIF</italic></th></tr></thead><tbody><tr><td>Word naming accuracy</td><td align="left" /><td align="left" /><td>.150</td><td align="left" /><td>.99</td><td align="left" /></tr><tr><td align="left">Lexical decision accuracy</td><td>.3640001</td><td>.3770001</td><td align="left" /><td>.1320001</td><td>.72</td><td>1.012</td></tr><tr><td align="left">Visuospatial two‐back accuracy</td><td>−.078</td><td>−.1290001</td><td align="left" /><td>.0090001</td><td>.90</td><td>1.137</td></tr><tr><td align="left">Navon task local trial RT</td><td>−.025</td><td>−.0970001</td><td align="left" /><td>.0080001</td><td>.85</td><td>1.141</td></tr><tr><td>Word naming RT</td><td align="left" /><td align="left" /><td>.055</td><td align="left" /><td>.80</td><td align="left" /></tr><tr><td align="left">Flanker effect in accuracy</td><td>.1340001</td><td>.1320001</td><td align="left" /><td>.0180001</td><td>.70</td><td>1.006</td></tr><tr><td align="left">Lexical decision accuracy</td><td>−.1310001</td><td>−.1420001</td><td align="left" /><td>.0180001</td><td>.72</td><td>1.004</td></tr><tr><td align="left">Visuospatial two‐back accuracy</td><td>.0920001</td><td>.1040001</td><td align="left" /><td>.0100001</td><td>.90</td><td>1.187</td></tr><tr><td align="left">Word naming marginal entropy of the first fixation</td><td>.0980001</td><td>.0970001</td><td align="left" /><td>.0090001</td><td align="left" /><td>1.187</td></tr><tr><td>Word naming marginal entropy of the first fixation</td><td align="left" /><td align="left" /><td>.034</td><td align="left" /><td align="left" /><td align="left" /></tr><tr><td align="left">Horizontal VF</td><td>−.1220001</td><td>−.1220001</td><td align="left" /><td>.0150001</td><td align="center">–</td><td>1.000</td></tr><tr><td align="left">Vertical VF</td><td>.0990001</td><td>.0990001</td><td align="left" /><td>.0100001</td><td align="center">–</td><td>1.000</td></tr><tr><td align="left">RAN RT</td><td>−.0970001</td><td>−.0970001</td><td align="left" /><td>.0090001</td><td>.96</td><td>1.000</td></tr></tbody></table> </ephtml> </p> <p>6 <sups>*</sups><emph>p</emph> < .05; **<emph>p</emph> < .01; ***<emph>p</emph> < .001.</p> <p>Since the marginal entropy of the first fixation was one of the best predictors for word naming RT, a follow‐up analysis was conducted to examine what factors, including cognitive abilities and presented VF, best predicted marginal entropy of the first fixation. The results showed that longer RAN RT, right VF presentation, and upper VF presentation were associated with a more consistent first fixation during word naming (Table 5).</p> <hd id="AN0179878392-25">Pseudoword naming</hd> <p>We first examined whether participants' first and second fixation locations differed when the word stimulus was presented in different quadrants of the screen (Fig. 5). The results showed that in the x‐coordinate, a main effect of horizontal VF was found, <emph>F</emph>(<reflink idref="bib1" id="ref151">1</reflink>, 126) = 31.767, <emph>p</emph> < .001, <emph>η</emph><subs>p</subs><sups>2</sups> = .058: participants' first two fixations were to the left of the word center where the word was presented in the left VF, and were to the right of the word center when it was presented in the right VF, suggesting that they generally overshoot the word horizontally. This effect interacted with fixation number, <emph>F</emph>(<reflink idref="bib1" id="ref152">1</reflink>, 126) = 10.830, <emph>p</emph> = .001, <emph>η</emph><subs>p</subs><sups>2</sups> = .003: participants' first fixation was more to the right and closer to the word center than their second fixations in left VF, <emph>t</emph>(<reflink idref="bib126" id="ref153">126</reflink>) = −3.363, <emph>p</emph> = .006, <emph>d</emph> = .251, whereas their first and second fixations did not differ in the right VF, <emph>t</emph>(<reflink idref="bib126" id="ref154">126</reflink>) = 1.479, <emph>p</emph> = .453. These results were consistent with those in the decision‐making task. In the y‐coordinate, a main effect of fixation number was found, <emph>F</emph>(<reflink idref="bib1" id="ref155">1</reflink>, 126) = 10.762, <emph>p</emph> = .001, <emph>η</emph><subs>p</subs><sups>2</sups> = .001: participants' second fixation was higher and closer to the word center than the first fixation. This effect interacted with the effect of vertical VF, <emph>F</emph>(<reflink idref="bib1" id="ref156">1</reflink>, 126) = 32.364, <emph>p</emph> < .001, <emph>η</emph><subs>p</subs><sups>2</sups> = .004: the fixation number effect was significant in the upper VF, <emph>t</emph>(<reflink idref="bib126" id="ref157">126</reflink>) = 6.628, <emph>p</emph> < .001, <emph>d</emph> = .443, but not in the lower VF, <emph>t</emph>(<reflink idref="bib126" id="ref158">126</reflink>) = −1.659, <emph>p</emph> = .350 (Fig. 5). As for the average number of fixations per trial, we found no effect of vertical VF, <emph>F</emph>(<reflink idref="bib1" id="ref159">1</reflink>, 127) = .881, <emph>p</emph> = .350, horizontal VF, <emph>F</emph>(<reflink idref="bib1" id="ref160">1</reflink>, 127) = .297, <emph>p</emph> = .587, or their interaction, <emph>F</emph>(<reflink idref="bib1" id="ref161">1</reflink>, 127) = 2.292, <emph>p</emph> = .133, suggesting that participants made a similar number of fixations per trial across the quadrants.</p> <p> <img src="https://imageserver.ebscohost.com/img/embimages/rdk/CGN/01sep24/cogs13489-fig-0005.jpg?ephost1=dGJyMNXb4kSepq84yOvqOLCmsE6epq5Srqa4SK6WxWXS" alt="cogs13489-fig-0005.jpg" title="5 Participants' first and second fixation locations in (a) x‐coordinate and (b) y‐coordinates in pseudoword naming when the word stimulus was presented in different quadrants of the screen: (*p < .05; **p < .01; ***p < .001). The error bars represent the standard deviations." /> </p> <p></p> <p>Similar to word naming, as shown in Fig. 6, in pattern A, participants looked at either the regions below the word beginning (Red and Yellow) or word end (Blue), or at these two regions (Green and Pink). In contrast, in pattern B, participants mainly focused at the word center (Red and Brown). Seventy‐five participants were classified as using pattern A, whereas 53 participants were classified as using pattern B. The two patterns significantly differed: data from participants adopting pattern A were more likely to be generated by pattern A HMM than pattern B HMM, and vice versa for those adopting pattern B, <emph>F</emph>(<reflink idref="bib1" id="ref162">1</reflink>, 126) = 153.7, <emph>p</emph> < .001, <ephtml> <math display="inline" altimg="urn:x-wiley:03640213:media:cogs13489:cogs13489-math-0001" xmlns="http://www.w3.org/1998/Math/MathML"><semantics><msubsup><mi>η</mi><mi>p</mi><mn>2</mn></msubsup><annotation encoding="application/x-tex">$\eta _p^2$</annotation></semantics></math> </ephtml> = .549, 90% CI [.4528, .6204]. Participants using the two patterns did not differ in average number of fixations per trial, <emph>t</emph>(<reflink idref="bib126" id="ref163">126</reflink>) = −.521, <emph>p</emph> = .603, <emph>d</emph> = −.093, fixation duration, <emph>t</emph>(<reflink idref="bib126" id="ref164">126</reflink>) = −.252, <emph>p</emph> = .802, <emph>d</emph> = −.045, saccade length, <emph>t</emph>(<reflink idref="bib126" id="ref165">126</reflink>) = 1.274, <emph>p</emph> = .205, <emph>d</emph> = .229, overall entropy, <emph>t</emph>(<reflink idref="bib126" id="ref166">126</reflink>) = .828, <emph>p</emph> = .394, <emph>d</emph> = .154, or marginal entropy of the first fixation, <emph>t</emph>(<reflink idref="bib126" id="ref167">126</reflink>) = 1.237, <emph>p</emph> = .218, <emph>d</emph> = .022.</p> <p> <img src="https://imageserver.ebscohost.com/img/embimages/rdk/CGN/01sep24/cogs13489-fig-0006.jpg?ephost1=dGJyMNXb4kSepq84yOvqOLCmsE6epq5Srqa4SK6WxWXS" alt="cogs13489-fig-0006.jpg" title="6 Representative eye movement patterns in pseudoword naming. Ellipses show ROIs as 2‐D Gaussian emissions. Table on the right shows transition probabilities among the ROIs. Priors show the probabilities that a fixation sequence starts from the ellipse. Small image shows ROI assignment of fixations and the corresponding heatmap. Note that in pattern B, the ROI 2 (Green) and 3 (Blue) were overlapped with ROI 4 (Pink) since the same number of ROIs were used to generate the two representative patterns to facilitate meaningful comparisons during clustering." /> </p> <p></p> <p>Similarly, in a separate analysis, we examined whether participants' eye movement pattern when the stimulus was presented in different quadrants of the screen was generally the same as their overall eye movement pattern derived by summarizing trials across all quadrants. The results showed that there were more participants adopting the same pattern as their overall pattern than chance in all quadrants[<reflink idref="bib1" id="ref168">1</reflink>]: the upper‐left quadrant (104/128), <emph>χ</emph><sups>2</sups>(<reflink idref="bib1" id="ref169">1</reflink>) = 50.000, <emph>p</emph> < .001, in the upper‐right quadrant (101/128), <emph>χ</emph><sups>2</sups>(<reflink idref="bib1" id="ref170">1</reflink>) = 42.781, <emph>p</emph> < .001, in the lower‐left quadrant (92/128), <emph>χ</emph><sups>2</sups>(<reflink idref="bib1" id="ref171">1</reflink>) = 24.500, <emph>p</emph> < .001, and in the lower‐right quadrant (99/128), <emph>χ</emph><sups>2</sups>(<reflink idref="bib1" id="ref172">1</reflink>) = 38.281, <emph>p</emph> < .001.</p> <p>Stepwise multiple regression was used to examine which variables significantly predicted pseudoword naming accuracy and RT. The intercorrelations among the potential predictors are summarized in Table 6. The results showed that pseudoword naming accuracy was best predicted by RAN RT, but not by any eye movement measures. In contrast, pseudoword naming RT was best predicted by consistency of the first fixation in a naming trial in addition to lexical decision, selective attention, and visuospatial working memory: a more consistent first fixation location was associated with shorter pseudoword naming RT. A follow‐up analysis showed that no cognitive ability measures or presented VF was associated with consistency of the first fixation during pseudoword naming. All variables had strong split‐half reliability (Table 7).</p> <p>6 Table Intercorrelations of the potential predictors in the stepwise multiple regression predicting pseudoword naming accuracy and RT</p> <p> <ephtml> <table><thead><tr valign="bottom"><th align="left" /><th align="center">Pseudoword naming A‐B scale</th><th align="center">Pseudoword naming OE</th><th align="center">Pseudoword naming ME</th><th align="center">Lexical decision accuracy</th><th align="center">Verbal two‐back accuracy</th><th align="center">Verbal two‐back RT</th><th align="center">Visuospatial two‐back accuracy</th><th align="center">Visuospatial two‐back RT</th><th align="center">RAN accuracy</th><th align="center">RAN RT</th><th align="center">Navon task global trial accuracy</th><th align="center">Navon task global trial RT</th><th align="center">Navon task local trial accuracy</th><th align="center">Navon task local trial RT</th><th align="center">Flanker effect in accuracy</th><th align="center">Flanker effect in RT</th></tr></thead><tbody><tr><td>Pseudoword naming A‐B scale</td><td align="center">–</td><td align="center">–</td><td align="center">–</td><td>0.116</td><td>0.075</td><td>−0.055</td><td>0.069</td><td>−0.077</td><td>0.048</td><td>−0.148</td><td>0.142</td><td>−0.069</td><td>.2180002</td><td>0.134</td><td>0.128</td><td>0.057</td></tr><tr><td>Pseudoword naming OE</td><td>0.152</td><td align="center">–</td><td align="center">–</td><td>0.026</td><td>−.2510002</td><td>0.029</td><td>−.2160002</td><td>0.111</td><td>−0.069</td><td>−.2340002</td><td>−0.147</td><td>0.158</td><td>−0.105</td><td>.2130002</td><td>0.135</td><td>−0.13</td></tr><tr><td>Pseudoword naming ME</td><td>.2400002</td><td>.6330002</td><td align="center">–</td><td>−0.078</td><td>−0.167</td><td>−0.039</td><td>−.2080002</td><td>0.049</td><td>−0.13</td><td>−0.13</td><td>−0.052</td><td>−0.034</td><td>−0.035</td><td>0.136</td><td>0.173</td><td>−0.049</td></tr><tr><td>Lexical decision accuracy</td><td>.116</td><td>−.078</td><td align="center">–</td><td align="center">–</td><td>0.07</td><td>.1990002</td><td>0.057</td><td>.1790002</td><td>−0.015</td><td>−0.12</td><td>0.051</td><td>0.044</td><td>0.003</td><td>0.108</td><td>0.046</td><td>−0.093</td></tr></tbody></table> </ephtml> </p> <ulist> <item>7 Abbreviations: ME, marginal entropy of the first fixation; OE, overall entropy.</item> <item>8 <sups>*</sups><emph>p</emph> < .05; **<emph>p</emph> < .01; ***<emph>p</emph> < .001. The rest of the intercorrelations were presented in Table 2.</item> <item>7 Table Correlations and stepwise regressions of pseudoword naming task</item> </ulist> <p> <ephtml> <table><thead><tr><th>Variable</th><th align="center"><italic>r</italic></th><th align="center"><italic>β</italic></th><th align="center"><italic>R</italic><sup>2</sup></th><th align="center">Δ<italic>R</italic><sup>2</sup></th><th align="center"><italic>r</italic><sub>SB</sub></th><th align="center"><italic>VIF</italic></th></tr></thead><tbody><tr><td>Pseudoword naming accuracy</td><td align="left" /><td align="left" /><td>.012</td><td align="left" /><td>.71</td><td align="left" /></tr><tr><td align="left">RAN accuracy</td><td>.1110001</td><td>.1110001</td><td align="left" /><td>.0120001</td><td>1.00</td><td>1.000</td></tr><tr><td>Pseudoword naming RT</td><td align="left" /><td align="left" /><td>.093</td><td align="left" /><td>.82</td><td align="left" /></tr><tr><td align="left">Lexical decision accuracy</td><td>−.1710001</td><td>−.1950001</td><td align="left" /><td>.0290001</td><td>.72</td><td>1.012</td></tr><tr><td align="left">Flanker effect in accuracy</td><td>.1680001</td><td>.1530001</td><td align="left" /><td>.0300001</td><td>.70</td><td>1.029</td></tr><tr><td align="left">Pseudoword naming marginal entropy of the first fixation</td><td>.1030001</td><td>.1070001</td><td align="left" /><td>.0110001</td><td>.71</td><td>1.005</td></tr><tr><td align="left">Visuospatial two‐back accuracy</td><td>.0910001</td><td>.1110001</td><td align="left" /><td>.0110001</td><td>.90</td><td>1.008</td></tr><tr><td align="left">Flanker effect in RT</td><td>−.1140001</td><td>−.1080001</td><td align="left" /><td>.0110001</td><td>.95</td><td>1.038</td></tr></tbody></table> </ephtml> </p> <p>9 <sups>*</sups><emph>p</emph> < .05; ***<emph>p</emph> < .001.</p> <hd id="AN0179878392-28">Discussion</hd> <p>In the current study, we tested the hypothesis that in isolated word reading, individual readers differ in eye fixation behavior, and how well readers voluntarily direct their eye gaze to the OVP for word recognition (Brysbaert & Nazir, [<reflink idref="bib7" id="ref173">7</reflink>]) is associated with reading performance in addition to lexical knowledge and cognitive abilities. We found that across different reading tasks, different representative eye movement patterns were discovered through clustering, suggesting the role of task demands on eye movement planning (Hsiao & Chan, [<reflink idref="bib39" id="ref174">39</reflink>]; Hsiao et al., [<reflink idref="bib38" id="ref175">38</reflink>]). In particular, in lexical decision, individual readers differed in whether they voluntarily fixate their eye gaze at the word center or the OVP (i.e., the area between word beginning and the word center), and higher eye movement similarity to the representative pattern focusing at the OVP as opposed to the other pattern focusing at the word center was associated with shorter lexical decision RT. Indeed, word recognition time is shown to increase when participants' initial fixation is being directed to a location away from the OVP (Liu & Li, [<reflink idref="bib60" id="ref176">60</reflink>]). We further showed that in lexical decision, individuals can differ in the ability to voluntarily fixate at the OVP, which accounts for individual differences in RT.</p> <p>In addition to eye fixation behavior, better local but not global visual processing ability was associated with faster lexical decision RT and higher pseudoword naming accuracy. Previous research has shown that among skilled English readers, those with faster reading RT showed higher sensitivity to constituent letters as demonstrated in masked form priming paradigms (Andrews & Hersch, [<reflink idref="bib2" id="ref177">2</reflink>]; Andrews & Lo, [<reflink idref="bib3" id="ref178">3</reflink>]; Perry, Lupker, & Davis, [<reflink idref="bib66" id="ref179">66</reflink>]). Consistent with these findings, Sigurdardottir, Arnardottir, and Halldorsdottir ([<reflink idref="bib76" id="ref180">76</reflink>]) reported that in adult participants feature‐based face matching accuracy explained additional variance in accuracy and speed of naming words and pseudowords, whereas global‐form face matching performance did not. These results suggested that performance in naming words and pseudowords was associated with local but not global processing ability. Similarly, Chinese readers with limited writing experience perceived characters more holistically with longer recognition RT, suggesting the importance of sensitivity to local components for expert reading performance (Tso et al., [<reflink idref="bib77" id="ref181">77</reflink>]; [<reflink idref="bib78" id="ref182">78</reflink>]; Tso et al., [<reflink idref="bib79" id="ref183">79</reflink>]).</p> <p>Interestingly, higher similarity to the eye movement pattern focusing at the OVP in lexical decision was associated with longer RAN RT, suggesting that the ability to accurately fixate at the OVP may conflict with the ability to rapidly identify an isolated familiar symbol. Similarly, a more consistent first fixation in word naming was associated with longer RAN RT. Identifying a multi‐letter word may require simultaneously attending to multiple features that distinguish it from its orthographic neighbors, in contrast to identifying an isolated symbol. Consistent with this speculation, local processing ability, but not RAN, was associated with lexical decision RT and word naming accuracy. Indeed, Georgiou et al. ([<reflink idref="bib29" id="ref184">29</reflink>]) reported that RAN RT was negatively correlated with isolated English word naming time.</p> <p>Eye movement pattern was not associated with word naming or pseudoword naming RT or accuracy. The discovered representative eye movement patterns were both dispersed and did not show concentration at the OVP. In contrast, word and pseudoword naming RT were associated with marginal entropy of the first fixation, with higher entropy associated with longer RT. A consistent first fixation location may result from extensive reading experience (Joseph, Liversedge, Blythe, White, & Rayner, [<reflink idref="bib49" id="ref185">49</reflink>]), which leads to more efficient word recognition through perceptual learning (Nazir & O'Regan, [<reflink idref="bib62" id="ref186">62</reflink>]). It could also reflect expertise in extracting important information through visual routine learning (Hsiao, [<reflink idref="bib35" id="ref187">35</reflink>]; Hsiao et al., [<reflink idref="bib36" id="ref188">36</reflink>]; Hsiao et al., [<reflink idref="bib46" id="ref189">46</reflink>]; Liao, Chong, & Hsiao, [<reflink idref="bib57" id="ref190">57</reflink>]). Indeed, lower marginal entropy of the first fixation (i.e., more consistent first fixation location) during word naming was found to be associated with right VF word presentation, suggesting its relevance to the typical left‐to‐right reading direction where the next word for reading appears in the right VF. In contrast, this association was not found in pseudoword naming.</p> <p>Among the reading tasks, eye movement pattern and consistency were associated with lexical decision, word naming, and pseudoword naming RT in addition to lexical knowledge/phonological knowledge and cognitive abilities. Thus, eye movement behavior may play an important role in visual word form identification, or more specifically orthographic processing/word reading fluency. Previous research has reported that the differences between children's and adults' eye movement behavior during reading could be better accounted for by their difference in linguistic proficiency rather than oculomotor control ability (Reichle et al., [<reflink idref="bib70" id="ref191">70</reflink>]). Also, adult readers had more consistent first fixation locations during reading than children (Joseph et al., [<reflink idref="bib49" id="ref192">49</reflink>]), suggesting that consistent eye fixation behavior during reading may result from extensive reading experience. Consistent with our speculation, a similar phenomenon has recently been reported in face identification, where children gradually develop more consistent eye movement behavior during learning to recognize faces with higher consistency associated with better recognition performance (Hsiao, An, Hui, Zheng, & Chan, [<reflink idref="bib37" id="ref193">37</reflink>]), and super‐recognizers fixated closer to the theoretical optimal fixation point for face identification when viewing faces (Linka et al., [<reflink idref="bib59" id="ref194">59</reflink>]). As word decoding skills are essential for reading comprehension (Hoover & Gough, [<reflink idref="bib33" id="ref195">33</reflink>]), developing a consistent visual routine to fixate at the OVP may further contribute to sentence reading performance. Thus, deficits in eye movement control or visual routine development, which is related to executive function and selective attention abilities (Hsiao et al., [<reflink idref="bib36" id="ref196">36</reflink>]; Hsiao et al., [<reflink idref="bib46" id="ref197">46</reflink>]; Liao et al., [<reflink idref="bib57" id="ref198">57</reflink>]), or simply insufficient experience in developing a consistent visual routine (cf. Hsiao et al., [<reflink idref="bib37" id="ref199">37</reflink>]), could be potential factors underlying reading fluency disorders. Indeed, abnormality in eye movements during reading has been commonly found in individuals with dyslexia (Prado, Dubois, & Valdois, [<reflink idref="bib67" id="ref200">67</reflink>]; Razuk, Barela, Peyre, Gerard, & Bucci, [<reflink idref="bib69" id="ref201">69</reflink>]). The quantitative measures of eye movement pattern and consistency using EMHMM can potentially be used for early identification of individuals at risk of reading difficulties (Benfatto et al., [<reflink idref="bib5" id="ref202">5</reflink>]; Nerušil, Polec, Škunda, & Kačur, [<reflink idref="bib64" id="ref203">64</reflink>]). Future work may examine whether consistent and focused eye fixation behavior at the OVP during isolated word reading contributes to reading fluency in sentence reading and whether it can be used for earlier identification of reading fluency disorders.</p> <p>Note that in the current study, we used stimuli from the Woodcock‐Johnson III Tests of Achievement (Woodcock et al., [<reflink idref="bib82" id="ref204">82</reflink>]), which is a norm‐referenced instrument for monitoring progress in achievement in different age groups. We only used the stimuli from the level of college and above average adults in order to match the age of our participants, and this limited the number of trials available in the word naming and the pseudoword naming tasks. Future work may try to replicate the current findings with a larger number of trials. Another limitation of the study was that we directly used lexical decision, word naming, and pseudoword naming accuracy as measured in the LexTALE test and the Woodcock‐Johnson III Tests of Achievement as the English proficiency measures in the regression models, instead of using a separate English proficiency measure. In future studies, we may use a separate English proficiency measure for the regression models, and also consider including a measure of L1 proficiency as a potential predictor.</p> <p>Note also that the current study focused on adult English as second language readers and most of the participants' first language was Chinese. Future work may examine whether similar phenomena can be observed in native English readers, whether readers' first language experience modulates the effects, and whether similar effects can also be observed in isolated word reading in other languages.</p> <hd id="AN0179878392-29">Conclusion</hd> <p>In conclusion, through a machine‐learning‐model‐based data‐driven approach EMHMM, we discovered two representative eye movement patterns when readers were making lexical decisions through clustering, with one focusing on the OVP and the other focusing on the word center. Through quantifying readers' eye movement pattern along the contrast between the two representative patterns, we showed that how well readers voluntarily directed their eye gaze to the OVP was associated with faster lexical decision RT. In addition, a more OVP‐focused eye fixation behavior was associated with longer RAN RT, suggesting conflicting visual processing abilities required for identifying isolated letters and letter strings. Also, a more consistent first fixation was associated with faster word naming and pseudoword naming RT, which may reflect more extensive reading experience or a better developed visual routine. Thus, the ability to develop a consistent eye movement pattern focusing on the OVP plays an important role in visual word form processing and reading fluency. EMHMM allowed us to discover and quantify individual differences in eye movement behavior in a data‐driven fashion, leading to this discovery. This finding has important implications for ways to facilitate reading fluency development and early detection of reading fluency disorders.</p> <hd id="AN0179878392-30">Data availability statement</hd> <p>The data and materials for all experiments and scripts for data analysis are available in the OSF repository, https://osf.io/8rw6x/?view%5fonly=73690157a3ae49bfbb2050c771da70bb. The EMHMM code is available on the website <ulink href="http://visal.cs.cityu.edu.hk/research/emhmm/">http://visal.cs.cityu.edu.hk/research/emhmm/</ulink>.</p> <hd id="AN0179878392-31">Open Research Badges</hd> <p>This article has earned Open Data and Open Materials badges. Data and materials are available at https://osf.io/8rw6x/?view_only=73690157a3ae49bfbb2050c771da70bb.</p> <ref id="AN0179878392-32"> <title> Footnotes </title> <blist> <bibl id="bib1" idref="ref22" type="bt">1</bibl> <bibtext> Consistent with this finding, in an exploratory analysis, we found that the overall entropy of participants' eye movement pattern during the pseudoword naming task was lower than that in the word naming task, <emph>t</emph>(127) = 3.600, <emph>p</emph> = .001, and marginally lower than that in the lexical decision task, <emph>t</emph>(127) = 2.081, <emph>p</emph> = .098. Those in the lexical decision and word naming tasks did not differ significantly, <emph>t</emph>(127) = −1.125, <emph>p</emph> = .501.</bibtext> </blist> </ref> <ref id="AN0179878392-33"> <title> References </title> <blist> <bibtext> An, J., & Hsiao, J. H. (2021). Modulation of mood on eye movement pattern and performance in face recognition. Emotion, 21 (3), 617 – 630.</bibtext> </blist> <blist> <bibl id="bib2" idref="ref23" type="bt">2</bibl> <bibtext> Andrews, S., & Hersch, J. (2010). Lexical precision in skilled readers: Individual differences in masked neighbor priming. Journal of Experimental Psychology: General, 139, 299 – 318.</bibtext> </blist> <blist> <bibl id="bib3" idref="ref27" type="bt">3</bibl> <bibtext> Andrews, S., & Lo, S. (2012). Not all skilled readers have cracked the code: Individual differences in masked form priming. Journal of Experimental Psychology: Learning, Memory, and Cognition, 38, (1), 152 – 163.</bibtext> </blist> <blist> <bibl id="bib4" idref="ref31" type="bt">4</bibl> <bibtext> Barbeito, R. (1981). Sighting dominance: An explanation based on the processing of visual direction in tests of sighting dominance. Vision Research, 21, 855 – 860.</bibtext> </blist> <blist> <bibl id="bib5" idref="ref65" type="bt">5</bibl> <bibtext> Benfatto, M. N., Seimyr, G. Ö., Ygge, J., Pansell, T., Rydberg, A., & Jacobson, C. (2016). Screening for dyslexia using eye tracking during reading. PLoS ONE, 11, (12), e0165508.</bibtext> </blist> <blist> <bibl id="bib6" idref="ref66" type="bt">6</bibl> <bibtext> Bishop, C. M. (2006). Pattern recognition and Machine Learning. NY : Springer New York.</bibtext> </blist> <blist> <bibl id="bib7" idref="ref21" type="bt">7</bibl> <bibtext> Brysbaert, M., & Nazir, T. (2005). Visual constraints in written word recognition: Evidence from the optimal viewing‐position effect. Journal of Research in Reading, 28, (3), 216 – 228.</bibtext> </blist> <blist> <bibl id="bib8" idref="ref54" type="bt">8</bibl> <bibtext> Brysbaert, M., Vitu, F., & Schroyens, W. (1996). The right visual field advantage and the optimal viewing position effect: On the relation between foveal and parafoveal word recognition. Neuropsychology, 10, (3), 385 – 395.</bibtext> </blist> <blist> <bibl id="bib9" idref="ref12" type="bt">9</bibl> <bibtext> Cabeza, R., & Kato, T. (2000). Features are also important: Contributions of featural and configural processing to face recognition. Psychological Science, 11, (5), 429 – 433.</bibtext> </blist> <blist> <bibtext> Callens, M., Tops, W., & Brysbaert, M. (2012). Cognitive profile of students who enter higher education with an indication of dyslexia. PLoS One, 7, (6), e38081.</bibtext> </blist> <blist> <bibtext> Chan, F. H. F., Barry, T. J., Chan, A. B., & Hsiao, J. H. (2020). Understanding visual attention to face emotions in social anxiety using hidden Markov models. Cognition and Emotion, 34, (8), 1704 – 1710.</bibtext> </blist> <blist> <bibtext> Chan, C. Y. H., Chan, A. B., Lee, T. M. C., & Hsiao, J. H. (2018). Eye movement patterns in face recognition are associated with cognitive decline in older adults. Psychonomic Bulletin & Review, 25, (6), 2200 – 2207.</bibtext> </blist> <blist> <bibtext> Chan, A. B., & Hsiao, J. H. (2016). Information distribution within musical segments. Music Perception, 34, (2), 218 – 242.</bibtext> </blist> <blist> <bibtext> Chang, X., Wang, P., Cai, M. M., & Wang, M. Z. (2019). The predictive power of working memory on Chinese middle school students' English reading comprehension. Reading & Writing Quarterly, 35, (5), 458 – 472.</bibtext> </blist> <blist> <bibtext> Chuk, T., Chan, A. B., & Hsiao, J. H. (2014). Understanding eye movements in face recognition using hidden Markov models. Journal of Visualization, 14, (11), 1 – 14.</bibtext> </blist> <blist> <bibtext> Chuk, T., Chan, A. B., & Hsiao, J. H. (2017). Is having similar eye movement patterns during face learning and recognition beneficial for recognition performance? Evidence from hidden Markov modeling. Vision Research, 141, 204 – 216.</bibtext> </blist> <blist> <bibtext> Chuk, T., Crookes, K., Hayward, W. G., Chan, A. B., & Hsiao, J. H. (2017). Hidden Markov model analysis reveals the advantage of analytic eye movement patterns in face recognition across cultures. Cognition, 169, 102 – 117.</bibtext> </blist> <blist> <bibtext> Chung, H. K. S., Liu, J. Y. W., & Hsiao, J. H. (2017). How does reading direction modulate perceptual asymmetry effects? Quarterly Journal of Experimental Psychology, 70, (8), 1559 – 1574.</bibtext> </blist> <blist> <bibtext> Commodari, E. (2017). Novice readers: The role of focused, selective, distributed and alternating attention at the first year of academic curriculum. I‐Perception, 8 (4), 204166951771855. https://doi.org/10.1177/2041669517718557</bibtext> </blist> <blist> <bibtext> Coutrot, A., Hsiao, J. H., & Chan, A. B. (2018). Scanpath modeling and classification with hidden Markov models. Behavior Research Methods, 50, (1), 362 – 379.</bibtext> </blist> <blist> <bibtext> Cover, T. M., & Thomas, J. A. (2006). Elements of information theory. John Wiley & Sons.</bibtext> </blist> <blist> <bibtext> Coviello, E., Chan, A. B., & Lanckriet, G. R. G. (2014). Clustering hidden Markov models with variational HEM. Journal of Machine Learning Research, 15, 697 – 747.</bibtext> </blist> <blist> <bibtext> Coviello, E., Lanckriet, G. R., & Chan, A. B. (2012). The variational hierarchical EM algorithm for clustering hidden Markov models. In P. Barlett (Ed.), Advances in Neural Information Processing Systems 25 (pp. 404 – 412). New York : Curran Associates, Inc.</bibtext> </blist> <blist> <bibtext> Denckla, M. B., & Rudel, R. (1974). Rapid "automatized" naming of pictured objects, colors, letters and numbers by normal children. Cortex, 10, (2), 186 – 202.</bibtext> </blist> <blist> <bibtext> Feng, G. (2006). Eye movements as time‐series random variables: A stochastic model of eye movement control in reading. Cognitive Systems Research, 7 (1), 70 – 95. https://doi.org/10.1016/j.cogsys.2005.07.004</bibtext> </blist> <blist> <bibtext> Franceschini, S., Bertoni, S., Gianesini, T., Gori, S., & Facoetti, A. (2017). A different vision of dyslexia: Local precedence on global perception. Scientific Reports, 7, 17462.</bibtext> </blist> <blist> <bibtext> Franceschini, S., Bertoni, S., Puccio, G., Mancarella, M., Gori, S., & Facoetti, A. (2020). Local perception impairs the lexical reading route. Psychological Research, 85, 1748 – 1756.</bibtext> </blist> <blist> <bibtext> Gauthier, I., Klaiman, C., & Schultz, R. T. (2009). Face composite effects reveal abnormal face processing in autism spectrum disorders. Vision Research, 49, (4), 470 – 478.</bibtext> </blist> <blist> <bibtext> Georgiou, G. K., Parrila, R., & Liao, C. H. (2008). Rapid naming speed and reading across languages that vary in orthographic consistency. Reading and Writing, 21, (9), 885 – 903.</bibtext> </blist> <blist> <bibtext> Gottardo, A., Koh, P. W., Chen, X., & Jia, F. L. (2017). Models of English and Chinese word reading for adolescent Chinese‐English bilinguals. Reading and Writing, 30, (7), 1377 – 1406.</bibtext> </blist> <blist> <bibtext> Haft, S. L., Caballero, J. N., Tanaka, H., Zekelman, L., Cutting, L. E., Uchikoshi, Y., & Hoeft, F. (2019). Direct and indirect contributions of executive function to word decoding and reading comprehension in kindergarten. Learning and Individual Differences, 76, 101783.</bibtext> </blist> <blist> <bibtext> Hogan, T. P., Catts, H. W., & Little, T. D. (2005). The relationship between phonological awareness and reading: Implications for the assessment of phonological awareness. Language Speech and Hearing Services in Schools, 36, (4), 285 – 293.</bibtext> </blist> <blist> <bibtext> Hoover, W. A., & Gough, P. B. (1990). The simple view of reading. Reading and Writing, 2, (2), 127 – 160.</bibtext> </blist> <blist> <bibtext> Hsiao, J. H. (2011). Visual field differences can emerge purely from perceptual learning: Evidence from modelling Chinese character pronunciation. Brain and Language, 119, (2), 89 – 98.</bibtext> </blist> <blist> <bibtext> Hsiao, J. H. (2024). Understanding human cognition through computational modeling. Topics in Cognitive Science, 16 (3), 349 – 376. Portico. https://doi.org/10.1111/tops.12737</bibtext> </blist> <blist> <bibtext> Hsiao, J. H., An, J. H., & Chan, A. B. (2020). The role of eye movement consistency in learning to recognize faces: Computational and experimental examinations. In Proceedings of the 42nd Annual Conference of the Cognitive Science Society (pp. 1072 – 1078). Cognitive Science Society.</bibtext> </blist> <blist> <bibtext> Hsiao, J. H., An, J., Hui, V. K. S., Zheng, Y., & Chan, A. B. (2022). Understanding the role of eye movement consistency in face recognition and autism through integrating deep neural networks and hidden Markov models. npj Science of Learning, 7, 28.</bibtext> </blist> <blist> <bibtext> Hsiao, J. H., An, J., Zheng, Y., & Chan, A. B. (2021). Do portrait artists have enhanced face processing abilities? Evidence from hidden Markov modelling of eye movements. Cognition, 211, 104616.</bibtext> </blist> <blist> <bibtext> Hsiao, J. H., & Chan, A. B. (2023). Visual attention to own‐ vs. other‐race faces: Perspectives from learning mechanisms and task demands. British Journal of Psychology, 114, (S1), 17 – 20.</bibtext> </blist> <blist> <bibtext> Hsiao, J. H., Chan, A. B., An, J., Yeh, S.‐L., & Jingling, L. (2021). Understanding the collinear masking effect in visual search through eye tracking. Psychonomic Bulletin & Review, 28, 1933 – 1943.</bibtext> </blist> <blist> <bibtext> Hsiao, J. H., & Cheng, L. (2013). The modulation of stimulus structure on visual field asymmetry effects: The case of Chinese character recognition. Quarterly Journal of Experimental Psychology, 66, (9), 1739 – 1755.</bibtext> </blist> <blist> <bibtext> Hsiao, J. H., & Cheung, K. (2016). Visual similarity of words alone can modulate hemispheric lateralization in visual word recognition: Evidence from modelling Chinese character recognition. Cognitive Science, 40, (2), 351 – 372.</bibtext> </blist> <blist> <bibtext> Hsiao, J. H., & Cottrell, G. W. (2008). Two fixations suffice in face recognition. Psychological Science, 9, (10), 998 – 1006.</bibtext> </blist> <blist> <bibtext> Hsiao, J. H., & Lam, S. M. (2013). The modulation of visual and task characteristics of a writing system on hemispheric lateralization in visual word recognition – A computational exploration. Cognitive Science, 37, 861 – 890.</bibtext> </blist> <blist> <bibtext> Hsiao, J. H., Lan, H., Zheng, Y., & Chan, A. B. (2021). Eye movement analysis with hidden Markov models (EMHMM) with co‐clustering. Behavior Research Methods, 53, 2473 – 2486.</bibtext> </blist> <blist> <bibtext> Hsiao, J. H., Liao, W., & Tso, R. V. Y. (2022). Impact of mask use on face recognition: An eye tracking study. Cognitive Research: Principles and Implications, 7, 32.</bibtext> </blist> <blist> <bibtext> Hsiao, J. H., & Liu, T. T. (2012). The optimal viewing position in face recognition. Journal of Vision, 12, (2), 1 – 9.</bibtext> </blist> <blist> <bibtext> Jasinska, K. K., & Petitto, L.‐A. (2018). Age of bilingual exposure is related to the contribution of phonological and semantic knowledge to successful reading development. Child Development, 89, (1), 310 – 331.</bibtext> </blist> <blist> <bibtext> Joseph, H. S. S. L., Liversedge, S. P., Blythe, H. I., White, S. J., & Rayner, K. (2009). Word length and landing position effects during reading in children and adults. Vision Research, 49, 2078 – 2086.</bibtext> </blist> <blist> <bibtext> Kim, D., Park, Y., & Lombardino, L. J. (2015). Rapid automatized naming, word‐level reading, and oral reading fluency in first‐grade Korean readers at risk for reading difficulties. Asia Pacific Education Review, 16, (3), 447 – 459.</bibtext> </blist> <blist> <bibtext> Knoop‐van Campen, C. A. N., Segers, E., & Verhoeven, L. (2018). How phonological awareness mediates the relation between working memory and word reading efficiency in children with dyslexia. Dyslexia, 24, (2), 156 – 169.</bibtext> </blist> <blist> <bibtext> Lam, S. M., & Hsiao, J. H. (2014). Bilingual experience modulates hemispheric lateralization in visual word processing. Bilingualism: Language and Cognition, 17, (30), 589 – 609.</bibtext> </blist> <blist> <bibtext> Lan, H., Liu, Z., Hsiao, J. H., Yu, D., & Chan, A. B. (2023). Clustering hidden Markov models with variational Bayesian hierarchical EM. IEEE Transactions on Neural Networks and Learning Systems, 34 (3), 1537 – 1551. https://doi.org/10.1109/tnnls.2021.3105570</bibtext> </blist> <blist> <bibtext> Lancaster, H. S., Li, J., & Gray, S. (2021). Selective visual attention skills differentially predict decoding and reading comprehension performance across reading ability profiles. Journal of Research in Reading, 44, (3), 715 – 734.</bibtext> </blist> <blist> <bibtext> Lau, E. Y. Y., Eskes, G. A., Morrison, D. L., Rajda, M., & Spurr, K. F. (2010). Executive function in patients with obstructive sleep apnea treated with continuous positive airway pressure. Journal of the International Neuropsychological Society, 16, (6), 1077 – 1088.</bibtext> </blist> <blist> <bibtext> Lemhöfer, K., & Broersma, M. (2012). Introducing LexTALE: A quick and valid lexical test for advanced learners of English. Behavior Research Methods, 44, 325 – 343.</bibtext> </blist> <blist> <bibtext> Liao, W., Chong, W. C., & Hsiao, J. H. (2022). Does word boundary information facilitate Chinese sentence reading among beginning readers? In Proceedings of the 44th Annual Conference of the Cognitive Science Society. Cognitive Science Society.</bibtext> </blist> <blist> <bibtext> Liao, W., Li, S. T. K., & Hsiao, J. H. (2022). Music reading experience modulates eye movement pattern in English reading but not in Chinese reading. Scientific Reports, 12, 9144.</bibtext> </blist> <blist> <bibtext> Linka, M., Borda, M. D., Alsheimer, T., de Haas, B., & Ramon, M. (2022). Characteristic fixation biases in super‐recognizers. Journal of Visualization, 22, (8), 1 – 15.</bibtext> </blist> <blist> <bibtext> Liu, P., & Li, X. (2013). Optimal viewing position effects in the processing of isolated Chinese words. Vision Research, 81, 45 – 57.</bibtext> </blist> <blist> <bibtext> Marian, V., Blumenfeld, H. K., & Kaushanskaya, M. (2007). The language experience and proficiency questionnaire (LEAP‐Q): Assessing language profiles in bilinguals and multilinguals. Journal of Speech, Language, and Hearing Research, 50, (4), 940 – 967.</bibtext> </blist> <blist> <bibtext> Nazir, T. A., & O'Regan, J. K. (1990). Some results on translation invariance in the human visual system. Spatial Vision, 5, (2), 81 – 100.</bibtext> </blist> <blist> <bibtext> Navon, D. (1977). Forest before trees: The precedence of global features in visual perception. Cognitive Psychology, 9, 353 – 383.</bibtext> </blist> <blist> <bibtext> Nerušil, B., Polec, J., Škunda, J., & Kačur, J. (2021). Eye tracking based dyslexia detection using a holistic approach. Scientific Reports, 11, 15687.</bibtext> </blist> <blist> <bibtext> Perfetti, C. (2007). Reading ability: Lexical quality to comprehension. Scientific Studies of Reading, 11, (4), 357 – 383.</bibtext> </blist> <blist> <bibtext> Perry, J. R., Lupker, S. J., & Davis, C. J. (2008). An evaluation of the interactive‐activation model using masked partial‐word priming. Language, Cognition and Neuroscience, 23, 36 – 68.</bibtext> </blist> <blist> <bibtext> Prado, C., Dubois, M., & Valdois, S. (2007). The eye movements of dyslexic children during reading and visual search: Impact of the visual attention span. Vision Research, 47, (19), 2521 – 2530.</bibtext> </blist> <blist> <bibtext> Que, Y., Zheng, Y., Hsiao, J. H., & Hu, X. (2023). Studying the effect of self‐selected background music on reading task with eye movements. Scientific Reports, 13, 1704.</bibtext> </blist> <blist> <bibtext> Razuk, M., Barela, J. A., Peyre, H., Gerard, C. L., & Bucci, M. P. (2018). Eye movements and postural control in dyslexic children performing different visual tasks. PLoS ONE, 13, (5), e0198001.</bibtext> </blist> <blist> <bibtext> Reichle, E. D., Liversedge, S. P., Drieghe, D., Blythe, H. I., Joseph, H. S. S. L., White, S. J., & Rayner, K. (2013). Using E‐Z reader to examine the concurrent development of eye‐movement control and reading skill. Developmental Review, 33, (2), 110 – 149.</bibtext> </blist> <blist> <bibtext> Ridderinkhof, K. R., Band, G., & Logan, G. (1999). A study of adaptive behavior: Effects of age and irrelevant information on the ability to inhibit one's actions. Acta Psychologica, 101, (1999), 315 – 337.</bibtext> </blist> <blist> <bibtext> Rodriguez‐Lopez, V., Barcala, X., Zaytouny, A., Dorronsoro, C., Peli, E., & Marcos, S. (2023). Monovision correction preference and eye dominance measurements. Translational Vision Science & Technology, 12, 18.</bibtext> </blist> <blist> <bibtext> Rossion, B. (2008). Picture‐plane inversion leads to qualitative changes of ace perception. Acta Psychologica, 128, (2), 274 – 289.</bibtext> </blist> <blist> <bibtext> Savage, R., McBreen, M., Genesee, F., Erdos, C., Haigh, C., & Nair, A. (2018). Rapid automatic naming predicts more than sublexical fluency: Evidence from English‐French bilinguals. Learning and Individual Differences, 62, 153 – 163.</bibtext> </blist> <blist> <bibtext> Siddaiah, A., Saldanha, M., Venkatesh, S. K., Ramachandra, N. B., & Padakannaya, P. (2014). Development of rapid automatized naming (RAN) in simultaneous Kannada‐English biliterate children. Journal of Psycholinguistic Research, 45, 177 – 187.</bibtext> </blist> <blist> <bibtext> Sigurdardottir, H. M., Arnardottir, A., & Halldorsdottir, E. (2021). Faces and words are both associated and dissociated as evidenced by visual problems in dyslexia. Scientific Reports, 11, 23000.</bibtext> </blist> <blist> <bibtext> Tso, R. V. Y., Au, T. K., & Hsiao, J. H. (2014). Perceptual expertise: Can sensorimotor experience change holistic processing and left side bias? Psychological Science, 25, (9), 1757 – 1767.</bibtext> </blist> <blist> <bibtext> Tso, R. V. Y., Au, T. K., & Hsiao, J. H. (2022). Non‐monotonic developmental trend of holistic processing in visual expertise: The case of Chinese character recognition. Cognitive Research: Principles and Implications, 7, 39.</bibtext> </blist> <blist> <bibtext> Tso, R. V. Y., Chan, R. T. C., & Hsiao, J. H. (2020). Holistic but with reduced right‐hemisphere involvement: The case of dyslexia in Chinese character recognition. Psychonomic Bulletin & Review, 27, (3), 553 – 562.</bibtext> </blist> <blist> <bibtext> Ventura, P., Fernandes, T., Pereira, A., Guerreiro, J. C., Farinha‐Fernandes, A., Delgado, J., Ferreira, M. F., Faustino, B., Raposo, I., & Wong, A. C.‐N. (2020). Holistic word processing is correlated with efficiency in visual word recognition. Attention, Perception, & Psychophysics, 82, 2739 – 2750.</bibtext> </blist> <blist> <bibtext> Williams, R. S., & Morris, R. K. (2004). Eye movements, word familiarity, and vocabulary acquisition. European Journal of Cognitive Psychology, 16, (1–2), 312 – 339.</bibtext> </blist> <blist> <bibtext> Woodcock, R. W., McGrew, K. S., & Mather, N. (2001). Woodcock‐Johnson III Tests of Achievement. Itasca, IL : Riverside.</bibtext> </blist> <blist> <bibtext> Zhang, J., Chan, A. B., Lau, E. Y. Y., & Hsiao, J. H. (2019). Individuals with insomnia misrecognize angry faces as fearful faces while missing the eyes: An eye‐tracking study. Sleep, 42, (2), zsy220.</bibtext> </blist> </ref> <aug> <p>By Weiyan Liao and Janet Hui‐wen Hsiao</p> <p>Reported by Author; Author</p> </aug> <nolink nlid="nl1" bibid="bib65" firstref="ref1"></nolink> <nolink nlid="nl2" bibid="bib32" firstref="ref2"></nolink> <nolink nlid="nl3" bibid="bib48" firstref="ref3"></nolink> <nolink nlid="nl4" bibid="bib30" firstref="ref4"></nolink> <nolink nlid="nl5" bibid="bib81" firstref="ref5"></nolink> <nolink nlid="nl6" bibid="bib31" firstref="ref6"></nolink> <nolink nlid="nl7" bibid="bib50" firstref="ref7"></nolink> <nolink nlid="nl8" bibid="bib74" firstref="ref8"></nolink> <nolink nlid="nl9" bibid="bib54" firstref="ref9"></nolink> <nolink nlid="nl10" bibid="bib28" firstref="ref10"></nolink> <nolink nlid="nl11" bibid="bib73" firstref="ref11"></nolink> <nolink nlid="nl12" bibid="bib16" firstref="ref13"></nolink> <nolink nlid="nl13" bibid="bib38" firstref="ref14"></nolink> <nolink nlid="nl14" bibid="bib79" firstref="ref15"></nolink> <nolink nlid="nl15" bibid="bib77" firstref="ref16"></nolink> <nolink nlid="nl16" bibid="bib78" firstref="ref17"></nolink> <nolink nlid="nl17" bibid="bib80" firstref="ref19"></nolink> <nolink nlid="nl18" bibid="bib27" firstref="ref20"></nolink> <nolink nlid="nl19" bibid="bib13" firstref="ref24"></nolink> <nolink nlid="nl20" bibid="bib41" firstref="ref25"></nolink> <nolink nlid="nl21" bibid="bib42" firstref="ref26"></nolink> <nolink nlid="nl22" bibid="bib18" firstref="ref28"></nolink> <nolink nlid="nl23" bibid="bib34" firstref="ref29"></nolink> <nolink nlid="nl24" bibid="bib62" firstref="ref30"></nolink> <nolink nlid="nl25" bibid="bib44" firstref="ref32"></nolink> <nolink nlid="nl26" bibid="bib52" firstref="ref33"></nolink> <nolink nlid="nl27" bibid="bib47" firstref="ref34"></nolink> <nolink nlid="nl28" bibid="bib59" firstref="ref35"></nolink> <nolink nlid="nl29" bibid="bib12" firstref="ref37"></nolink> <nolink nlid="nl30" bibid="bib45" firstref="ref38"></nolink> <nolink nlid="nl31" bibid="bib15" firstref="ref39"></nolink> <nolink nlid="nl32" bibid="bib36" firstref="ref40"></nolink> <nolink nlid="nl33" bibid="bib40" firstref="ref41"></nolink> <nolink nlid="nl34" bibid="bib46" firstref="ref42"></nolink> <nolink nlid="nl35" bibid="bib58" firstref="ref43"></nolink> <nolink nlid="nl36" bibid="bib21" firstref="ref44"></nolink> <nolink nlid="nl37" bibid="bib17" firstref="ref45"></nolink> <nolink nlid="nl38" bibid="bib43" firstref="ref46"></nolink> <nolink nlid="nl39" bibid="bib14" firstref="ref47"></nolink> <nolink nlid="nl40" bibid="bib19" firstref="ref48"></nolink> <nolink nlid="nl41" bibid="bib26" firstref="ref49"></nolink> <nolink nlid="nl42" bibid="bib29" firstref="ref50"></nolink> <nolink nlid="nl43" bibid="bib25" firstref="ref55"></nolink> <nolink nlid="nl44" bibid="bib68" firstref="ref56"></nolink> <nolink nlid="nl45" bibid="bib56" firstref="ref57"></nolink> <nolink nlid="nl46" bibid="bib82" firstref="ref60"></nolink> <nolink nlid="nl47" bibid="bib63" firstref="ref71"></nolink> <nolink nlid="nl48" bibid="bib24" firstref="ref72"></nolink> <nolink nlid="nl49" bibid="bib75" firstref="ref73"></nolink> <nolink nlid="nl50" bibid="bib10" firstref="ref74"></nolink> <nolink nlid="nl51" bibid="bib55" firstref="ref75"></nolink> <nolink nlid="nl52" bibid="bib71" firstref="ref76"></nolink> <nolink nlid="nl53" bibid="bib61" firstref="ref78"></nolink> <nolink nlid="nl54" bibid="bib72" firstref="ref80"></nolink> <nolink nlid="nl55" bibid="bib11" firstref="ref85"></nolink> <nolink nlid="nl56" bibid="bib20" firstref="ref89"></nolink> <nolink nlid="nl57" bibid="bib83" firstref="ref92"></nolink> <nolink nlid="nl58" bibid="bib23" firstref="ref93"></nolink> <nolink nlid="nl59" bibid="bib22" firstref="ref94"></nolink> <nolink nlid="nl60" bibid="bib53" firstref="ref103"></nolink> <nolink nlid="nl61" bibid="bib126" firstref="ref110"></nolink> <nolink nlid="nl62" bibid="bib127" firstref="ref132"></nolink> <nolink nlid="nl63" bibid="bib39" firstref="ref174"></nolink> <nolink nlid="nl64" bibid="bib60" firstref="ref176"></nolink> <nolink nlid="nl65" bibid="bib66" firstref="ref179"></nolink> <nolink nlid="nl66" bibid="bib76" firstref="ref180"></nolink> <nolink nlid="nl67" bibid="bib49" firstref="ref185"></nolink> <nolink nlid="nl68" bibid="bib35" firstref="ref187"></nolink> <nolink nlid="nl69" bibid="bib57" firstref="ref190"></nolink> <nolink nlid="nl70" bibid="bib70" firstref="ref191"></nolink> <nolink nlid="nl71" bibid="bib37" firstref="ref193"></nolink> <nolink nlid="nl72" bibid="bib33" firstref="ref195"></nolink> <nolink nlid="nl73" bibid="bib67" firstref="ref200"></nolink> <nolink nlid="nl74" bibid="bib69" firstref="ref201"></nolink> <nolink nlid="nl75" bibid="bib64" firstref="ref203"></nolink>
Header DbId: eric
DbLabel: ERIC
An: EJ1441014
AccessLevel: 3
PubType: Academic Journal
PubTypeId: academicJournal
PreciseRelevancyScore: 0
IllustrationInfo
Items – Name: Title
  Label: Title
  Group: Ti
  Data: Understanding the Role of Eye Movement Pattern and Consistency in Isolated English Word Reading through Hidden Markov Modeling
– Name: Language
  Label: Language
  Group: Lang
  Data: English
– Name: Author
  Label: Authors
  Group: Au
  Data: <searchLink fieldCode="AR" term="%22Weiyan+Liao%22">Weiyan Liao</searchLink><br /><searchLink fieldCode="AR" term="%22Janet+Hui-wen+Hsiao%22">Janet Hui-wen Hsiao</searchLink> (ORCID <externalLink term="https://orcid.org/0000-0003-2271-8710">0000-0003-2271-8710</externalLink>)
– Name: TitleSource
  Label: Source
  Group: Src
  Data: <searchLink fieldCode="SO" term="%22Cognitive+Science%22"><i>Cognitive Science</i></searchLink>. 2024 48(9).
– Name: Avail
  Label: Availability
  Group: Avail
  Data: Wiley. Available from: John Wiley & Sons, Inc. 111 River Street, Hoboken, NJ 07030. Tel: 800-835-6770; e-mail: cs-journals@wiley.com; Web site: https://www.wiley.com/en-us
– Name: PeerReviewed
  Label: Peer Reviewed
  Group: SrcInfo
  Data: Y
– Name: Pages
  Label: Page Count
  Group: Src
  Data: 29
– Name: DatePubCY
  Label: Publication Date
  Group: Date
  Data: 2024
– Name: TypeDocument
  Label: Document Type
  Group: TypDoc
  Data: Journal Articles<br />Reports - Research
– Name: Subject
  Label: Descriptors
  Group: Su
  Data: <searchLink fieldCode="DE" term="%22Foreign+Countries%22">Foreign Countries</searchLink><br /><searchLink fieldCode="DE" term="%22English+%28Second+Language%29%22">English (Second Language)</searchLink><br /><searchLink fieldCode="DE" term="%22Eye+Movements%22">Eye Movements</searchLink><br /><searchLink fieldCode="DE" term="%22Markov+Processes%22">Markov Processes</searchLink><br /><searchLink fieldCode="DE" term="%22Sight+Method%22">Sight Method</searchLink><br /><searchLink fieldCode="DE" term="%22Naming%22">Naming</searchLink><br /><searchLink fieldCode="DE" term="%22Reaction+Time%22">Reaction Time</searchLink><br /><searchLink fieldCode="DE" term="%22Reading+Fluency%22">Reading Fluency</searchLink><br /><searchLink fieldCode="DE" term="%22Visual+Acuity%22">Visual Acuity</searchLink><br /><searchLink fieldCode="DE" term="%22Individual+Differences%22">Individual Differences</searchLink>
– Name: Subject
  Label: Geographic Terms
  Group: Su
  Data: <searchLink fieldCode="DE" term="%22Hong+Kong%22">Hong Kong</searchLink>
– Name: DOI
  Label: DOI
  Group: ID
  Data: 10.1111/cogs.13489
– Name: ISSN
  Label: ISSN
  Group: ISSN
  Data: 0364-0213<br />1551-6709
– Name: Abstract
  Label: Abstract
  Group: Ab
  Data: In isolated English word reading, readers have the optimal performance when their initial eye fixation is directed to the area between the beginning and word center, that is, the optimal viewing position (OVP). Thus, how well readers voluntarily direct eye gaze to this OVP during isolated word reading may be associated with reading performance. Using Eye Movement analysis with Hidden Markov Models, we discovered two representative eye movement patterns during lexical decisions through clustering, which focused at the OVP and the word center, respectively. Higher eye movement similarity to the OVP-focusing pattern predicted faster lexical decision time in addition to cognitive abilities and lexical knowledge. However, the OVP-focusing pattern was associated with longer isolated single letter naming time, suggesting conflicting visual abilities required for identifying isolated letters and multi-letter words. In contrast, in both word and pseudoword naming, although clustering did not reveal an OVP-focused pattern, higher consistency of the first fixation as measured in entropy predicted faster naming time in addition to cognitive abilities and lexical knowledge. Thus, developing a consistent eye movement pattern focusing on the OVP is essential for word orthographic processing and reading fluency. This finding has important implications for interventions for reading difficulties.
– Name: AbstractInfo
  Label: Abstractor
  Group: Ab
  Data: As Provided
– Name: DateEntry
  Label: Entry Date
  Group: Date
  Data: 2024
– Name: AN
  Label: Accession Number
  Group: ID
  Data: EJ1441014
PLink https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=eric&AN=EJ1441014
RecordInfo BibRecord:
  BibEntity:
    Identifiers:
      – Type: doi
        Value: 10.1111/cogs.13489
    Languages:
      – Text: English
    PhysicalDescription:
      Pagination:
        PageCount: 29
    Subjects:
      – SubjectFull: Foreign Countries
        Type: general
      – SubjectFull: English (Second Language)
        Type: general
      – SubjectFull: Eye Movements
        Type: general
      – SubjectFull: Markov Processes
        Type: general
      – SubjectFull: Sight Method
        Type: general
      – SubjectFull: Naming
        Type: general
      – SubjectFull: Reaction Time
        Type: general
      – SubjectFull: Reading Fluency
        Type: general
      – SubjectFull: Visual Acuity
        Type: general
      – SubjectFull: Individual Differences
        Type: general
      – SubjectFull: Hong Kong
        Type: general
    Titles:
      – TitleFull: Understanding the Role of Eye Movement Pattern and Consistency in Isolated English Word Reading through Hidden Markov Modeling
        Type: main
  BibRelationships:
    HasContributorRelationships:
      – PersonEntity:
          Name:
            NameFull: Weiyan Liao
      – PersonEntity:
          Name:
            NameFull: Janet Hui-wen Hsiao
    IsPartOfRelationships:
      – BibEntity:
          Dates:
            – D: 01
              M: 09
              Type: published
              Y: 2024
          Identifiers:
            – Type: issn-print
              Value: 0364-0213
            – Type: issn-electronic
              Value: 1551-6709
          Numbering:
            – Type: volume
              Value: 48
            – Type: issue
              Value: 9
          Titles:
            – TitleFull: Cognitive Science
              Type: main
ResultId 1