The E-Z reader model of eye-movement control in reading: comparisons to other models

A beginner's guide to eye tracking for psycholinguistic studies of reading

Eye tracking has been a popular methodology used to study the visual, cognitive, and linguistic processes underlying word recognition and sentence parsing during reading for several decades. However, the successful use of eye tracking requires researchers to make deliberate choices about how they apply this technique, and there is wide variability across labs and fields with respect to which choices are "standard." We aim to provide an easy-to-reference guideline that can help new researchers with their entrée into eye-tracking-while-reading research. Because the standards do - and should - vary from field to field or study to study as is appropriate for the research question, we do not set a rigid recipe for handling eye tracking data, but rather provide a conceptual framework within which researchers can make informed decisions about how to treat their data so that it is most informative for their research question. Therefore, this paper provides a description of eye movements in reading and an overview of psycholinguistic research on the topic, an overview of experiment design considerations, a description of the data processing pipeline and important choice points and implications, an overview of common dependent measures and their calculation, and a summary of resources for data analysis.

Visual attention matters during word recognition: A Bayesian modeling approach

It is striking that visual attention, the process by which attentional resources are allocated in the visual field so as to locally enhance visual perception, is a pervasive component of models of eye movements in reading, but is seldom considered in models of isolated word recognition. We describe BRAID, a new Bayesian word-Recognition model with Attention, Interference and Dynamics. As most of its predecessors, BRAID incorporates three sensory, perceptual, and orthographic knowledge layers together with a lexical membership submodel. Its originality resides in also including three mechanisms that modulate letter identification within strings: an acuity gradient, lateral interference, and visual attention. We calibrated the model such that its temporal scale was consistent with behavioral data, and then explored the model's capacity to generalize to other, independent effects. We evaluated the model's capacity to account for the word length effect in lexical decision, for the optimal viewing position effect, and for the interaction of crowding and frequency effects in word recognition. We further examined how these effects were modulated by variations in the visual attention distribution. We show that visual attention modulates all three effects and that a narrow distribution of visual attention results in performance patterns that mimic those reported in impaired readers. Overall, the BRAID model could be conceived as a core building block, towards the development of integrated models of reading aloud and eye movement control, or of visual recognition of impaired readers, or any context in which visual attention does matter.

Neural correlates for word-frequency effect in Chinese natural reading

Word frequency effect has always been of interest for reading research because of its critical role in exploring mental processing underlying reading behaviors. Access to word frequency information has long been considered an indicator of the beginning of lexical processing and the most sensitive marker for studying when the brain begins to extract semantic information Sereno & Rayner, Brain and Cognition, 42, 78-81, (2000), Trends in Cognitive Sciences, 7, 489-493, (2003). While the word frequency effect has been extensively studied in numerous eye-tracking and traditional EEG research using the RSVP paradigm, there is a lack of corresponding evidence in studies of natural reading. To find the neural correlates of the word frequency effect, we conducted a study of Chinese natural reading using EEG and eye-tracking coregistration to examine the time course of lexical processing. Our results reliably showed that the word frequency effect first appeared in the N200 time window and the bilateral occipitotemporal regions. Additionally, the word frequency effect was reflected in the N400 time window, spreading from the occipital region to the central parietal and frontal regions. Our current study provides the first neural correlates for word-frequency effect in natural Chinese reading so far, shedding new light on understanding lexical processing in natural reading and could serve as an important basis for further reading study when considering neural correlates in a realistic manner.

Aging and word predictability during reading: Evidence from eye movements and fixation-related potentials

The use of context to facilitate the processing of words is recognized as a hallmark of skilled reading. This capability is also hypothesized to change with older age because of cognitive changes across the lifespan. However, research investigating this issue using eye movements or event-related potentials (ERPs) has produced conflicting findings. Specifically, whereas eye-movement studies report larger context effects for older than younger adults, ERP findings suggest that context effects are diminished or delayed for older readers. Crucially, these contrary findings may reflect methodological differences, including use of unnatural sentence displays in ERP research. To address these limitations, we used a coregistration technique to record eye movements (EMs) and fixation-related potentials (FRPs) simultaneously while 44 young adults (18-30 years) and 30 older adults (65+ years) read sentences containing a target word that was strongly or weakly predicted by prior context. Eye-movement analyses were conducted over all data (full EM dataset) and only data matching FRPs. FRPs were analysed to capture early and later components 70-900 ms following fixation-onset on target words. Both eye-movement datasets and early FRPs showed main effects of age group and context, while the full EM dataset and later FRPs revealed larger context effects for older adults. We argue that, by using coregistration methods to address limitations of earlier ERP research, our experiment provides compelling complementary evidence from eye movements and FRPs that older adults rely more on context to integrate words during reading.

40 Years of Cracking the Orthographic Code: A Special Issue in Honour of Jonathan Grainger's Career

Anyone recounting the history of cognitive psychology will have to make early mention of the study of orthographic processing (starting in 1886 with the seminal work of Cattell, a doctoral student of Wilhelm Wundt); and anyone recounting the study of orthographic processing will have to make mention of Jonathan Grainger. An honorary member and former president of the European Society for Cognitive Psychology, Jonathan has dedicated nearly four decades of research to the mechanisms driving the recognition of letters, words and sentences during reading. In honour of Jonathan's career-which formally has come to a close in 2023-in this Special Issue several contemporaries and close collaborators highlight important advances that have been made in the past 40 years, and provide flavours of where the field stands today.

The fundamentals of eye tracking part 1: The link between theory and research question

Eye tracking technology has become increasingly prevalent in scientific research, offering unique insights into oculomotor and cognitive processes. The present article explores the relationship between scientific theory, the research question, and the use of eye-tracking technology. It aims to guide readers in determining if eye tracking is suitable for their studies and how to formulate relevant research questions. Examples from research on oculomotor control, reading, scene perception, task execution, visual expertise, and instructional design are used to illustrate the connection between theory and eye-tracking data. These examples may serve as inspiration to researchers new to eye tracking. In summarizing the examples, three important considerations emerge: (1) whether the study focuses on describing eye movements or uses them as a proxy for e.g., perceptual, or cognitive processes, (2) the logical chain from theory to predictions, and (3) whether the study is of an observational or idea-testing nature. We provide a generic scheme and a set of specific questions that may help researchers formulate and explicate their research question using eye tracking.

Attention to food stimuli in binge eating disorder: Electrophysiological evidence

Attentional biases towards food play an important role in the pathology of binge eating disorder (BED). Later stage electrophysiological potentials (P300, late positive potential) present promising markers of motivated attention with high temporal, albeit low spatial resolution. Complementing this, the N2pc is an earlier-latency component providing the possibility of more directly analyzing visuospatial attention. Therefore, we tested a group with BED (N = 60), as well as an overweight (OW; N = 28) and normal weight (NW; N = 30) group without BED in a Go/No-Go paradigm using food and nonfood distractor images. Only the OW group in exclusively the Go trials displayed a stronger spatial attention allocation towards nonfood distractors as evidenced by an increased N2pc amplitude. In the P300's time window, the OW group displayed no attentional bias towards food and the NW group only did so in the absence of a target. Solely the BED group allocated more motivated attention towards food distractors both in Go and No-Go trials. In the following late positive potential (LPP), the OW group exhibited a general attentional bias towards food distractors, while the BED group only did so in the absence of a target. These results are discussed in light of the incentive sensitization theory and a potential early attentional suppression of potent distractors.

Information entropy facilitates (not impedes) lexical processing during language comprehension

It is well known that contextual predictability facilitates word identification, but it is less clear whether the uncertainty associated with the current context (i.e., its lexical entropy) influences sentence processing. On the one hand, high entropy contexts may lead to interference due to greater number of lexical competitors. On the other hand, predicting multiple lexical competitors may facilitate processing through the preactivation of shared semantic features. In this study, we examined whether entropy measured at the trial level (i.e., for each participant, for each item) corresponds to facilitatory or inhibitory effects. Trial-level entropy captures each individual's knowledge about specific contexts and is therefore a more valid and sensitive measure of entropy (relative to the commonly employed item-level entropy). Participants (N = 112) completed two experimental sessions (with counterbalanced orders) that were separated by a 3- to 14-day interval. In one session, they produced up to 10 completions for sentence fragments (N = 647). In another session, they read the same sentences including a target word (whose entropy value was calculated based on the produced completions) while reading times were measured. We observed a facilitatory (not inhibitory) effect of trial-level entropy on lexical processing over and above item-level measures of lexical predictability (including cloze probability, surprisal, and semantic constraint). Extra analyses revealed that greater semantic overlap between the target and the produced responses facilitated target processing. Thus, the results lend support to theories of lexical prediction maintaining that prediction involves broad activation of semantic features rather than activation of full lexical forms.

Direct lexical control of eye movements in Chinese reading: Evidence from the co-registration of EEG and eye tracking

The direct-lexical-control hypothesis stipulates that some aspect of a word's processing determines the duration of the fixation on that word and/or the next. Although the direct lexical control is incorporated into most current models of eye-movement control in reading, the precise implementation varies and the assumptions of the hypothesis may not be feasible given that lexical processing must occur rapidly enough to influence fixation durations. Conclusive empirical evidence supporting this hypothesis is therefore lacking. In this article, we report the results of an eye-tracking experiment using the boundary paradigm in which native speakers of Chinese read sentences in which target words were either high- or low-frequency and preceded by a valid or invalid preview. Eye movements were co-registered with electroencephalography, allowing standard analyses of eye-movement measures, divergence point analyses of fixation-duration distributions, and fixated-related potentials on the target words. These analyses collectively provide strong behavioral and neural evidence of early lexical processing and thus strong support for the direct-lexical-control hypothesis. We discuss the implications of the findings for our understanding of how the hypothesis might be implemented, the neural systems that support skilled reading, and the nature of eye-movement control in the reading of Chinese versus alphabetic scripts.

The Role of Occipitotemporal Network for Speed-Reading: An fMRI Study

The activity of occipitotemporal regions involved in linguistic reading processes, such as the ventral occipitotemporal cortex (vOT), is believed to exhibit strong interactions during higher-order language processing, specifically in the connectivity between the occipital gyrus and the temporal gyrus. In this study, we utilized functional magnetic resonance imaging (fMRI) with psychophysiological interaction (PPI) and dynamic causal modeling (DCM) to investigate the functional and effective connectivity in the occipitotemporal network during speed reading. We conducted the experiment with native Japanese speakers who underwent and without speed-reading training and subsequently performed established reading tasks at different speeds (slow, medium, and fast) while undergoing 3-Tesla Siemens fMRI. Our activation analyses revealed significant changes in occipital and temporal regions as reading speed increased, indicating functional connectivity within the occipitotemporal network. DCM results further demonstrated more intricate effective connections and high involvement within the occipitotemporal pathway: (1) reading signals originated from the inferior occipital gyrus (iO), distributed to the vOT and the posterior superior temporal sulcus (pSTS), and then gathered in the anterior superior temporal sulcus (aSTS); (2) reading speed loads had modulation effects on the pathways from the aSTS to vOT and from the iO to vOT. These findings highlight the complex connectivity and dynamic interactions within the occipitotemporal network during speed-reading processes.

Processing causal structure sentences in Mandarin Chinese: an eye movements study

It remains uncertain whether causal structure prediction can improve comprehension in Chinese sentences and whether the position of the headword mediates the prediction effect. We conducted an experiment to explore the effect of causal prediction and headword position in Chinese sentence reading. Participants were asked to read sentences containing causal connectives with their eye movements recorded. In the experiment, we manipulated the causal structure of the sentence and the position of the headword. We found a promoting effect of causal structure on first-pass reading time and a hindering impact on total reading time. However, the effect was not mediated by the headword position. The results show that causal syntactic prediction facilitated early-stage processing and increased the integration cost in the late stage of Chinese sentence processing. These findings also support the constraint-based approach, which suggests an isolation between semantic and syntactic processing.

An information-theoretic analysis of targeted regressions during reading

Regressions, or backward saccades, are common during reading, accounting for between 5% and 20% of all saccades. And yet, relatively little is known about what causes them. We provide an information-theoretic operationalization for two previous qualitative hypotheses about regressions, which we dub reactivation and reanalysis. We argue that these hypotheses make different predictions about the pointwise mutual information or pmi between a regression's source and target. Intuitively, the pmi between two words measures how much more (or less) likely one word is to be present given the other. On one hand, the reactivation hypothesis predicts that regressions occur between words that are associated, implying high positive values of pmi. On the other hand, the reanalysis hypothesis predicts that regressions should occur between words that are not associated with each other, implying negative, low values of pmi. As a second theoretical contribution, we expand on previous theories by considering not only pmi but also expected values of pmi, E[pmi], where the expectation is taken over all possible realizations of the regression's target. The rationale for this is that language processing involves making inferences under uncertainty, and readers may be uncertain about what they have read, especially if a previous word was skipped. To test both theories, we use contemporary language models to estimate pmi-based statistics over word pairs in three corpora of eye tracking data in English, as well as in six languages across three language families (Indo-European, Uralic, and Turkic). Our results are consistent across languages and models tested: Positive values of pmi and E[pmi] consistently help to predict the patterns of regressions during reading, whereas negative values of pmi and E[pmi] do not. Our information-theoretic interpretation increases the predictive scope of both theories and our studies present the first systematic crosslinguistic analysis of regressions in the literature. Our results support the reactivation hypothesis and, more broadly, they expand the number of language processing behaviors that can be linked to information-theoretic principles.

Cloze probability, predictability ratings, and computational estimates for 205 English sentences, aligned with existing EEG and reading time data

We release a database of cloze probability values, predictability ratings, and computational estimates for a sample of 205 English sentences (1726 words), aligned with previously released word-by-word reading time data (both self-paced reading and eye-movement records; Frank et al., Behavior Research Methods, 45(4), 1182-1190. 2013) and EEG responses (Frank et al., Brain and Language, 140, 1-11. 2015). Our analyses show that predictability ratings are the best predictors of the EEG signal (N400, P600, LAN) self-paced reading times, and eye movement patterns, when spillover effects are taken into account. The computational estimates are particularly effective at explaining variance in the eye-tracking data without spillover. Cloze probability estimates have decent overall psychometric accuracy and are the best predictors of early fixation patterns (first fixation duration). Our results indicate that the choice of the best measurement of word predictability in context critically depends on the processing index being considered.

Early parafoveal semantic integration in natural reading

Humans can read and comprehend text rapidly, implying that readers might process multiple words per fixation. However, the extent to which parafoveal words are previewed and integrated into the evolving sentence context remains disputed. We investigated parafoveal processing during natural reading by recording brain activity and eye movements using MEG and an eye tracker while participants silently read one-line sentences. The sentences contained an unpredictable target word that was either congruent or incongruent with the sentence context. To measure parafoveal processing, we flickered the target words at 60 Hz and measured the resulting brain responses (i.e. Rapid Invisible Frequency Tagging, RIFT) during fixations on the pre-target words. Our results revealed a significantly weaker tagging response for target words that were incongruent with the previous context compared to congruent ones, even within 100ms of fixating the word immediately preceding the target. This reduction in the RIFT response was also found to be predictive of individual reading speed. We conclude that semantic information is not only extracted from the parafovea but can also be integrated with the previous context before the word is fixated. This early and extensive parafoveal processing supports the rapid word processing required for natural reading. Our study suggests that theoretical frameworks of natural reading should incorporate the concept of deep parafoveal processing.

Lower-level oculomotor deficits in schizophrenia during multi-line reading: Evidence from return-sweeps

Reading fluency deficits in schizophrenia (Sz) have been attributed to dysfunction in both lower-level, oculomotor processing and higher-level, lexical processing, according to the two-hit deficit model. Given that prior work examining reading deficits in individuals with Sz has primarily focused on single-line and single-word reading tasks, eye movements that are unique to passage reading, such as return-sweep saccades, have not yet been examined in Sz. Return-sweep saccades are large eye movements that are made when readers move from the end of one line to the beginning of the next line during natural passage reading. Examining return-sweeps provides an opportunity to examine lower-level, oculomotor deficits during reading under circumstances when upcoming higher-level, lexical information is not available for visual processing because visual acuity constraints do not permit detailed lexical processing of line-initial words when return-sweeps are programmed. To examine the source of reading deficits in Sz, we analysed an existing data set in which participants read multi-line passages with manipulations to line spacing. Readers with Sz made significantly more return-sweep targeting errors followed by corrective saccades compared with healthy controls. Both groups showed similar effects of line spacing on return-sweep targeting accuracy, suggesting similar sensitivities to visual crowding during reading. Furthermore, the patterns of fixation durations in readers with Sz corroborate prior work indicating reduced parafoveal processing of upcoming words. Together, these findings suggest that lower-level visual and oculomotor dysfunction contribute to reading deficits in Sz, providing support for the two-hit deficit model.

Visuospatial, oculomotor, and executive reading skills evolve in elementary school, and errors are significant: a topological RAN study

We investigate the development of visuospatial and oculomotor reading skills in a cohort of elementary school children. Employing a longitudinal methodology, the study applies the Topological serial digit Rapid Automated Naming (Top-RAN) battery, which evaluates visuospatial reading skills leveraging metrics addressing crowding, distractors, and voluntary attention orientation. The participant pool comprises 142 students (66 males, 76 females), including 46 non-native speakers (21 males, 25 females), representing a diverse range of ethnic backgrounds. The Top-RAN dataset encompasses performance, error, and self-correction metrics for each subtest and student, underscoring the significance of these factors in the process of reading acquisition. Analytical methods include dimensionality reduction, clustering, and classification algorithms, consolidated into a Python package to facilitate reproducible results. Our results indicate that visuospatial reading abilities vary according to the task and demonstrate a marked evolution over time, as seen in the progressive decrease in execution times, errors, and self-corrections. This pattern supports the hypothesis that the growth of oculomotor, attentional, and executive skills is primarily fostered by educational experiences and maturation. This investigation provides valuable insights into the dynamic nature of these skills during pivotal educational stages.

Column setting and text justification influence return-sweep eye movement behavior during Chinese multi-line reading

People regularly read multi-line texts in different formats and publishers, internationally, must decide how to present text to make reading most effective and efficient. Relatively few studies have examined multi-line reading, and fewer still Chinese multi-line reading. Here, we examined whether texts presented in single or double columns, and either left-justified or fully-justified affect Chinese reading. Text format had minimal influence on overall reading time; however, it significantly impacted return-sweeps (large saccades moving the eyes from the end of one line of text to the beginning of the next). Return-sweeps were launched and landed further away from margins and involved more corrective saccades in single- than double-column format. For left- compared to fully-justified format, return-sweeps were launched and landed closer to margins. More corrective saccades also occurred. Our results showed more efficient return-sweep behavior for fully- than left-justified text. Moreover, there were clear trade-off effects such that formats requiring increased numbers of shorter return-sweeps produced more accurate targeting and reduced numbers of corrective fixations, whereas formats requiring reduced numbers of longer return-sweeps caused less accurate targeting and an increased rate of corrective fixations. Overall, our results demonstrate that text formats substantially affect return-sweep eye movement behavior during Chinese reading without affecting efficiency and effectiveness, that is, the overall time it takes to read and understand the text.

Eye movement analysis for real-world settings using segmented linear regression

Eye movement analysis is critical to studying human brain phenomena such as perception, cognition, and behavior. However, under uncontrolled real-world settings, the recorded gaze coordinates (commonly used to track eye movements) are typically noisy and make it difficult to track change in the state of each phenomenon precisely, primarily because the expected change is usually a slower transient process. This paper proposes an approach, Improved Naive Segmented linear regression (INSLR), which approximates the gaze coordinates with a piecewise linear function (PLF) referred to as a hypothesis. INSLR improves the existing NSLR approach by employing a hypotheses clustering algorithm, which redefines the final hypothesis estimation in two steps: (1) At each time-stamp, measure the likelihood of each hypothesis in the candidate list of hypotheses by using the least square fit score and its distance from the k-means of the hypotheses in the list. (2) Filter hypothesis based on a pre-defined threshold. We demonstrate the significance of the INSLR method in addressing the challenges of uncontrolled real-world settings such as gaze denoising and minimizing gaze prediction errors from cost-effective devices like webcams. Experiment results show INSLR consistently outperforms the baseline NSLR in denoising noisy signals from three eye movement datasets and minimizes the error in gaze prediction from a low precision device for 71.1% samples. Furthermore, this improvement in denoising quality is further validated by the improved accuracy of the oculomotor event classifier called NSLR-HMM and enhanced sensitivity in detecting variations in attention induced by distractor during online lecture.

Processing difficulty while reading words with neighbors is not due to increased foveal load: Evidence from eye movements

Words with high orthographic relatedness are termed "word neighbors" (angle/angel; birch/birth). Activation-based models of word recognition assume that lateral inhibition occurs between words and their activated neighbors. However, studies of eye movements during reading have not found inhibitory effects in early measures assumed to reflect lexical access (e.g., gaze duration). Instead, inhibition in eye-movement studies has been found in later measures of processing (e.g., total time, regressions in). We conducted an eye-movement boundary change study (Rayner, Cognitive Psychology, 7(1), 65-81, 1975) that manipulated the parafoveal preview of the word following the neighbor word (word N+1). In this way, we explored whether the late inhibitory effects seen with transposed letter words and words with higher-frequency neighbors result from reduced parafoveal preview due to increased foveal load and/or interference during late stages of lexical processing (the L2 stage within the E-Z Reader framework). For word N+1, while there were clear preview effects, there was not an effect of the neighborhood status of word N, nor a significant interaction. This suggests that the late inhibitory effects of earlier eye-movement studies are driven by misidentification of neighbor words rather than being due to increased foveal load.

Unique Patterns of Eye Movements Characterizing Inattentive Reading in ADHD

OBJECTIVE

We aimed to identify unique patterns of eye-movements measures reflecting inattentive reading among adults with and without ADHD.

METHOD & RESULTS

We recorded eye-movements during uninterrupted text reading of typically developed (TD) and ADHD adults. First, we found significantly longer reading time for the ADHD group than the TD group. Further, we detected cases in which words were reread more than twice and found that such occasions were much more frequent in participants with ADHD than in TD participants. Moreover, we discovered that the first reading pass of these words was less sensitive to the length of the word than the first pass of words read only once, indicating a less meaningful reading.

CONCLUSION

We propose that high rate of words that were reread is a correlate of inattentive reading which is more pronounced among ADHD readers. Implications of the findings in the context of reading comprehension are discussed.

Beta- and gamma-band cortico-cortical interactions support naturalistic reading of continuous text

Large-scale integration of information across cortical structures, building on neural connectivity, has been proposed to be a key element in supporting human cognitive processing. In electrophysiological neuroimaging studies of reading, quantification of neural interactions has been limited to the level of isolated words or sentences due to artefacts induced by eye movements. Here, we combined magnetoencephalography recording with advanced artefact rejection tools to investigate both cortico-cortical coherence and directed neural interactions during naturalistic reading of full-page texts. Our results show that reading versus visual scanning of text was associated with wide-spread increases of cortico-cortical coherence in the beta and gamma bands. We further show that the reading task was linked to increased directed neural interactions compared to the scanning task across a sparse set of connections within a wide range of frequencies. Together, the results demonstrate that neural connectivity flexibly builds on different frequency bands to support continuous natural reading.

Neglect dyslexia: whole-word and within-word errors with parafoveal and foveal viewing

Patients with left-sided neglect dyslexia often omit whole words positioned on the left, termed whole-word errors, or commit errors on the left-sided letters of words, termed unilateral paralexias. In addition, the errors have been shown to be exacerbated by simultaneously presented distractors, which has been interpreted as a failure of selective attention. In two experiments, we examined the dependency of these error types on parafoveal versus foveal viewing. The first experiment used a paradigm with parafoveal targets and distractors; the second a paradigm with foveal targets and parafoveal distractors. This enabled a separate evaluation of the influences of stimulus position within an egocentric frame, a two-word allocentric frame, and a within-word allocentric frame. First, regarding whole-word errors, we found the expected spatial and distractor effects with parafoveal targets and distractors. With foveal targets and parafoveal distractors, however, the spatial effect was effectively eliminated. Surprisingly, intrusions from the distractor word were common in distractor conditions. This is consistent with an egocentric account and not a two-word allocentric account. Second, we found that unilateral paralexias remained largely consistent regardless of spatial position or the presence of a distractor. Thus, there is a contrast in spatial and distractor effects between whole-word errors and unilateral paralexias. These results are consistent with three distinct deficits: an egocentric deficit across space resulting in whole-word errors, a failure of selective attention that results in whole-word intrusion errors, and a within-word allocentric deficit resulting in unilateral paralexias.

Human attention during goal-directed reading comprehension relies on task optimization

The computational principles underlying attention allocation in complex goal-directed tasks remain elusive. Goal-directed reading, that is, reading a passage to answer a question in mind, is a common real-world task that strongly engages attention. Here, we investigate what computational models can explain attention distribution in this complex task. We show that the reading time on each word is predicted by the attention weights in transformer-based deep neural networks (DNNs) optimized to perform the same reading task. Eye tracking further reveals that readers separately attend to basic text features and question-relevant information during first-pass reading and rereading, respectively. Similarly, text features and question relevance separately modulate attention weights in shallow and deep DNN layers. Furthermore, when readers scan a passage without a question in mind, their reading time is predicted by DNNs optimized for a word prediction task. Therefore, we offer a computational account of how task optimization modulates attention distribution during real-world reading.

The transposed-word effect provides no unequivocal evidence for parallel processing

Studies using a grammaticality decision task have revealed surprising flexibility in the processing of word order during sentence reading in both alphabetic and non-alphabetic scripts. Participants in these studies typically exhibit a transposed-word effect, in which they make more errors and slower correct responses for stimuli that contain a word transposition and are derived from grammatical as compared to ungrammatical base sentences. Some researchers have used this finding to argue that words are encoded in parallel during reading, such that multiple words can be processed simultaneously and might be recognised out of order. This contrasts with an alternative account of the reading process, which argues that words must be encoded serially, one at a time. We examined, in English, whether the transposed-word effect provides evidence for a parallel-processing account, employing the same grammaticality decision task used in previous research and display procedures that either allowed for parallel word encoding or permitted only the serial encoding of words. Our results replicate and extend recent findings by showing that relative word order can be processed flexibly even when parallel processing is not possible (i.e., within displays requiring serial word encoding). Accordingly, while the present findings provide further evidence for flexibility in the processing of relative word order during reading, they add to converging evidence that the transposed-word effect does not provide unequivocal evidence for a parallel-processing account of reading. We consider how the present findings may be accounted for by both serial and parallel accounts of word recognition in reading.

The role of visual crowding in eye movements during reading: Effects of text spacing

Visual crowding, generally defined as the deleterious influence of clutter on visual discrimination, is a form of inhibitory interaction between nearby objects. While the role of crowding in reading has been established in psychophysics research using rapid serial visual presentation (RSVP) paradigms, how crowding affects additional processes involved in natural reading, including parafoveal processing and saccade targeting, remains unclear. The current study investigated crowding effects on reading via two eye-tracking experiments. Experiment 1 was a sentence-reading experiment incorporating an eye-contingent boundary change in which reader's parafoveal processing was quantified through comparing reading times after valid or invalid information was presented in the parafovea. Letter spacing was jointly manipulated to compare how crowding affects parafoveal processing. Experiment 2 was a passage-reading experiment with a line spacing manipulation. In addition to replicating previously observed letter spacing effects on global reading parameters (i.e., more but shorter fixations with wider spacing), Experiment 1 found an interaction between preview validity and letter spacing indicating that the efficiency of parafoveal processing was constrained by crowding and visual acuity. Experiment 2 found reliable but subtle influences of line spacing. Participants had shorter fixation durations, higher skipping probabilities, and less accurate return sweeps when line spacing was increased. In addition to extending the literature on the role of crowding to reading in ecologically valid scenarios, the current results inform future research on characterizing the influence of crowding in natural reading and comparing effects of crowding across reader populations.

Parallel phonological processing of Chinese characters revealed by flankers tasks

An important and extensively researched question in the field of reading is whether readers can process multiple words in parallel. An unresolved issue regarding this question is whether the phonological information from foveal and parafoveal words can be processed in parallel, i.e., parallel phonological processing. The present study aims to investigate whether there is parallel phonological processing of Chinese characters. The original and the revised flankers tasks were applied. In both tasks, a foveal target character was presented in isolation in the no-flanker condition, flanked on both sides by a parafoveal homophone in the homophone-flanker condition, and by a non-homophonic character in the unrelated-flanker condition. Participants were instructed to fixate on the target characters and press two keys to indicate whether they knew the target characters (lexical vs. non-lexical). In the original flankers task, the stimuli were presented for 150 ms without a post-mask. In the revised flankers task, we set the stimulus exposure time (duration of the stimuli plus the blank interval between the stimuli and the post-mask) to each participant's lexical decision threshold to prevent participants from processing the target and flanker characters serially. In both tasks, reaction times to the lexical targets were significantly shorter in the homophone-flanker condition than in the unrelated-flanker condition, suggesting parallel phonological processing of Chinese characters. In the revised flankers task, accuracy rates to the lexical targets were significantly lower in the unrelated-flanker condition compared to the homophone-flanker condition, further supporting parallel phonological processing of Chinese characters. Moreover, reaction times to the lexical targets were the shortest in the no-flanker condition in both tasks, reflecting the attention distribution over both the target and flanker characters. The findings of this study provide valuable insights into the parallel processing mechanisms involved in reading.

Altered Eye Movements During Reading With Simulated Central and Peripheral Visual Field Defects

Purpose

Although foveal vision provides fine spatial information, parafoveal and peripheral vision are also known to be important for efficient reading behaviors. Here we systematically investigate how different types and sizes of visual field defects affect the way visual information is acquired via eye movements during reading.

Methods

Using gaze-contingent displays, simulated scotomas were induced in 24 adults with normal or corrected-to-normal vision during a reading task. The study design included peripheral and central scotomas of varying sizes (aperture or scotoma size of 2°, 4°, 6°, 8°, and 10°) and no-scotoma conditions. Eye movements (e.g., forward/backward saccades, fixations, microsaccades) were plotted as a function of either the aperture or scotoma size, and their relationships were characterized by the best fitting model.

Results

When the aperture size of the peripheral scotoma decreased below 6° (11 visible letters), there were significant decreases in saccade amplitude and velocity, as well as substantial increases in fixation duration and the number of fixations. Its dependency on the aperture size is best characterized by an exponential decay or growth function in log-linear coordinates. However, saccade amplitude and velocity, fixation duration, and forward/regressive saccades increased more or less linearly with increasing central scotoma size in log-linear coordinates.

Conclusions

Our results showed differential impacts of central and peripheral vision loss on reading behaviors while lending further support for the importance of foveal and parafoveal vision in reading. These apparently deviated oculomotor behaviors may in part reflect optimal reading strategies to compensate for the loss of visual information.

Lexical Processing Strongly Affects Reading Times But Not Skipping During Natural Reading

In a typical text, readers look much longer at some words than at others, even skipping many altogether. Historically, researchers explained this variation via low-level visual or oculomotor factors, but today it is primarily explained via factors determining a word's lexical processing ease, such as how well word identity can be predicted from context or discerned from parafoveal preview. While the existence of these effects is well established in controlled experiments, the relative importance of prediction, preview and low-level factors in natural reading remains unclear. Here, we address this question in three large naturalistic reading corpora (n = 104, 1.5 million words), using deep neural networks and Bayesian ideal observers to model linguistic prediction and parafoveal preview from moment to moment in natural reading. Strikingly, neither prediction nor preview was important for explaining word skipping-the vast majority of explained variation was explained by a simple oculomotor model, using just fixation position and word length. For reading times, by contrast, we found strong but independent contributions of prediction and preview, with effect sizes matching those from controlled experiments. Together, these results challenge dominant models of eye movements in reading, and instead support alternative models that describe skipping (but not reading times) as largely autonomous from word identification, and mostly determined by low-level oculomotor information.

Lexical competition in the flankers task revisited

We investigated the impact of flanking stimuli that are orthographic neighbors of central target words in the reading version of the flankers task. Experiment 1 provided a replication of the finding that flanking words that are orthographic neighbors of central target words (e.g., BLUE BLUR BLUE) facilitate lexical decisions relative to unrelated word flankers (e.g., STEP BLUR STEP). Experiment 2 tested the hypothesis that this facilitatory effect might be due to the task that was used in Experiment 1 and in prior research-the lexical decision task. In Experiment 2 the task was perceptual identification, and here we observed that orthographic neighbor flankers interfered with target word identification. Experiment 2 also included a bigram flanker condition (e.g., BL BLUR UE), and here the related bigram flankers facilitated target word identification. We conclude that when the task requires identification of a specific word, effects of lexical competition emerge over and above the facilitatory effects driven by the sublexical spatial pooling of orthographic information across target and flankers, and that the inhibitory influence of lexical competition has an even stronger impact when flankers are whole words.

GECO-CN: Ghent Eye-tracking COrpus of sentence reading for Chinese-English bilinguals

The current work presents the very first eye-tracking corpus of natural reading by Chinese-English bilinguals, whose two languages entail different writing systems and orthographies. Participants read an entire novel in these two languages, presented in paragraphs on screen. Half of the participants first read half of the novel in their native language (Simplified Chinese) and then the rest of the novel in their second language (English), while the other half read in the reverse language order. This article presents some important basic descriptive statistics of reading times and compares the difference between reading in the two languages. However, this unique eye-tracking corpus also allows the exploration of theories of language processing and bilingualism. Importantly, it provides a solid and reliable ground for studying the difference between Eastern and Western languages, understanding the impact and consequences of having a completely different first language on bilingual processing. The materials are freely available for use by researchers interested in (bilingual) reading.

Is there a processing advantage for verb-noun collocations in Chinese reading? Evidence from eye movements during reading

A growing number of studies show a processing advantage for collocations, which are commonly-used juxtapositions of words, such as "joint effort" or "shake hands," suggesting that skilled readers are keenly perceptive to the occurrence of two words in phrases. With the current research, we report two experiments that used eye movement measures during sentence reading to explore the processing of four-character verb-noun collocations in Chinese, such as ("revise the article"). Experiment 1 compared the processing of these collocations relative to similar four-character expressions that are not collocations (e.g., , "revise the ending") in neutral contexts and contexts in which the collocation was predictable from the preceding sentence context. Experiment 2 further examined the processing of these four-character collocations, by comparing eye movements for commonly-used "strong" collocations, such as ("protect the environment"), as compared to less commonly-used "weak" collocations, such as ("protect nature"), again in neutral contexts and contexts in which the collocations were highly predictable. The results reveal a processing advantage for both collocations relative to novel expressions, and for "strong" collocations relative to "weak" collocations, which was independent of effects of contextual predictability. We interpret these findings as providing further evidence that readers are highly sensitive to the frequency that words co-occur as a phrase in written language, and that a processing advantage for collocations occurs independently of contextual expectations.

Not all grammar errors are equally noticed: error detection of naturally occurring errors and implications for eye-tracking models of everyday texts

Grammar errors are a natural part of everyday written communication. They are not a uniform group, but vary from morphological errors to ungrammatical word order and involve different types of word classes. In this study, we examine whether some types of naturally occurring errors attract more attention than others during reading, measured by detection rates. Data from 211 Danish high school students were included in the analysis. They each read texts containing different types of errors: syntactic errors (verb-third word order), morphological agreement errors (verb conjugations; gender mismatches in NPs) and orthographic errors. Participants were asked to underline all errors they detected while reading for comprehension. We examined whether there was a link between the type of errors that participants did not detect, the type of errors which they produce themselves (as measured in a subsequent grammar quiz), and the type of errors that are typical of high school students in general (based on error rates in a corpus). If an error is infrequent in production, it may cause a larger surprisal effect and be more attended to. For the three subtypes of grammar errors (V3 word order, verb errors, NP errors), corpus error rates predicted detection rates for most conditions. Yet, frequency was not the only possible explanation, as phonological similarity to the correct form is entangled with error frequency. Explicit grammatical awareness also played a role. The more correct answers participants had in the grammar tasks in the quiz, the more errors they detected. Finally, we found that the more annoyed with language errors participants reported to be, the more errors they detected. Our study did not measure eye movements, but the differences in error detection patterns point to shortcomings of existing eye-tracking models. Understanding the factors that govern attention and reaction to everyday grammar errors is crucial to developing robust eye-tracking processing models which can accommodate non-standard variation. Based on our results, we give our recommendations for current and future processing models.

The graded predictive pre-activation in Chinese sentence reading: evidence from eye movements

Previous research has revealed that graded pre-activation rather than specific lexical prediction is more likely to be the mechanism for the word predictability effect in English. However, whether graded pre-activation underlies the predictability effect in Chinese reading is unknown. Accordingly, the present study tested the generality of the graded pre-activation account in Chinese reading. We manipulated the contextual constraint of sentences and the predictability of target words as independent variables. Readers' eye movement behaviors were recorded via an eye tracker. We examined whether processing an unpredictable word in a solid constraining context incurs a prediction error cost when this unpredictable word has a predictable alternative. The results showed no cues of prediction error cost on the early eye movement measures, supported by the Bayes Factor analyses. The current research indicates that graded predictive pre-activation underlies the predictability effect in Chinese reading.

Eye Movement Traces of Linguistic Knowledge in Native and Non-Native Reading

The detailed study of eye movements in reading has shed considerable light into how language processing unfolds in real time. Yet eye movements in reading remain inadequately studied in non-native (L2) readers, even though much of the world's population is multilingual. Here we present a detailed analysis of the quantitative functional influences of word length, frequency, and predictability on eye movement measures in reading in a large, linguistically diverse sample of non-native English readers. We find many similar qualitative effects as in L1 readers, but crucially also a proficiency-sensitive "lexicon-context tradeoff". The most proficient L2 readers' eye movements approach an L1 pattern, but as L2 proficiency diminishes, readers' eye movements become less sensitive to a word's predictability in context and more sensitive to word frequency, which is context-invariant. This tradeoff supports a rational, experience-dependent account of how context-driven expectations are deployed in L2 language processing.

Visual attention modulates the transition from fine-grained, serial processing to coarser-grained, more parallel processing: A computational modeling study

During reading acquisition, beginning readers transition from serial to more parallel processing. The acquisition of word specific knowledge through orthographic learning is critical for this transition. However, the processes by which orthographic representations are acquired and fine-tuned as learning progresses are not well understood. Our aim was to explore the role of visual attention in this transition through computational modeling. We used the BRAID-Learn model, a Bayesian model of visual word recognition, to simulate the orthographic learning of 700 4-to 10-letter English known words and novel words, presented 5 times each to the model. The visual attention quantity available for letter identification was manipulated in the simulations to assess its influence on the learning process. We measured the overall processing time and number of attentional fixations simulated by the model across exposures and their impact on two markers of serial processing, the lexicality and length effects, depending on visual attention quantity. Results showed that the two lexicality and length effects were modulated by visual attention quantity. The quantity of visual attention available for processing further modulated novel word orthographic learning and the evolution of the length effect on processing time and number of attentional fixations across repeated exposures to novel words. The simulated patterns are consistent with behavioral data and the developmental trajectories reported during reading acquisition. Overall, the model predicts that the efficacy of orthographic learning depends on visual attention quantity and that visual attention may be critical to explain the transition from serial to more parallel processing.

Eye movements during reading in beginning and skilled readers: Impact of reading level or physiological maturation?

We begin by presenting and examining relevant data in the literature on eye movements in reading, from childhood to adulthood. In particular, we discuss the differences found in eye movements during reading between children in different age groups and with different reading levels and skilled adult readers in terms of word recognition and sentence processing. We then critically discuss two hypotheses that account for the differences between children and adults' eye movement during reading: one being reading age itself - the changes in eye movement patterns in reading are regulated by the level of reading proficiency and its automatization - and the other being the role of maturation of oculomotor control and, consequently, its possible changes in eye movement patterns during reading. Finally, we list gaps in the research field and suggest that future research will benefit from investigating eye movements during reading in ex-illiterate adults who are in the process of learning to read in order to isolate both reading and maturational factors.

The determinants of saccade targeting strategy in neurodevelopmental disorders: The influence of suboptimal reading experience

Whether eye-movements deficits are causal in reading disorders (RD) or rather a consequence of linguistic processing difficulty experienced by disabled readers has been extensively debated.Since RD are frequently comorbid with the Neurofibromatosis type1 (NF1), children with NF1 were used as a comparison group for children with dyslexia in this study.Eye movements were recorded while 21 dyslexic, 20 NF1, and 20 typically developing children performed an oculomotor lateralized bisection task. In this experiment, we manipulated the type of stimulus - discrete (words and strings of hashes) versus continuous (solid lines) - and the visual field where the stimulus was displayed (left vs right). The results showed that (1) only proficient readers (TD and NF1 without RD) showed fully developed oculomotor mechanisms for efficient reading, with a clear preferred viewing location located to the left of the word's centre in both visual fields, and fine-tuned saccade targeting guided by the between-character space information and (2) NF1 poor readers mirrored the dyslexic eye movement behaviour, with less accuracy and more variability in saccadic programming, no sensitivity to the discreteness of the stimuli, particularly in the left visual field. We concluded that disruption to oculomotor behaviour reflectsthe fact that many of the processes involved in reading are not yet automatized for children with RD, independently of NF1. This suggests that the differences in saccade targeting strategy between children with and without RD would be secondary consequences of their reduced reading experience.

Dynamic attentional bias for pictorial and textual food cues in the visual search paradigm

Previous studies have found that individuals have an attentional bias for food cues, which may be related to the energy level or the type of stimulus (e.g., pictorial or textual food cues) of the food cues. However, the available evidence is inconsistent, and there is no consensus about how the type of stimulus and food energy modulate food-related attentional bias. Searching for food is one of the most important daily behaviors. In this study, a modified visual search paradigm was used to explore the attentional bias for food cues, and eye movements were recorded. Food cues consisted of both food words and food pictures with different energy levels (i.e., high- and low-calorie foods). The results showed that there was an attentional avoidance in the early stage but a later-stage attentional approach for all food cues in the pictorial condition. This was especially true for high-calorie food pictures. Participants showed a later-stage conflicting attentional bias for foods with different energy levels in the textual condition. They showed an attentional approach to high-calorie food words but an attentional avoidance of low-calorie food words. These data show that food-related attentional bias varied along with different time courses, which was also modulated by the type of stimulus and food energy. These findings regarding dynamic attentional bias could be explained using the Goal Conflict Model of eating behavior.

Modeling task effects in human reading with neural network-based attention

Research on human reading has long documented that reading behavior shows task-specific effects, but it has been challenging to build general models predicting what reading behavior humans will show in a given task. We introduce NEAT, a computational model of the allocation of attention in human reading, based on the hypothesis that human reading optimizes a tradeoff between economy of attention and success at a task. Our model is implemented using contemporary neural network modeling techniques, and makes explicit and testable predictions about how the allocation of attention varies across different tasks. We test this in an eyetracking study comparing two versions of a reading comprehension task, finding that our model successfully accounts for reading behavior across the tasks. Our work thus provides evidence that task effects can be modeled as optimal adaptation to task demands.

Prediction at the intersection of sentence context and word form: Evidence from eye-movements and self-paced reading

A key issue in language processing is how we recognize and understand words in sentences. Research on sentence reading indicates that the time we need to read a word depends on how (un)expected it is. Research on single word recognition shows that each word also has its own recognition dynamics based on the relation between its orthographic form and its meaning. It is not clear, however, how these sentence-level and word-level dynamics interact. In the present study, we examine the joint impact of these sources of information during sentence reading. We analyze existing eye-tracking and self-paced reading data (Frank et al., 2013, Behavior Research Methods, 45[4], 1182-1190) to investigate the interplay of sentence-level prediction (operationalized as Surprisal) and word Orthography-Semantics Consistency in activating word meaning in sentence processing. Results indicate that both Surprisal and Orthography-Semantics Consistency exert an influence on several reading measures. The shape of the observed interaction differs, but the results give compelling indication for a general trade-off between expectations based on sentence context and cues to meaning from word orthography.

Developmental trajectories of eye movements in oral and silent reading for beginning readers: a longitudinal investigation

Eye movements provide a sensitive window into cognitive processing during reading. In the present study, we investigated beginning readers' longitudinal changes in temporal and spatial measures of eye movements during oral versus silent reading, the extent to which variation in eye movements is attributable to individual differences and text differences, and the functional form of growth trajectories of eye-movement variables. Data were from 363 English-speaking children (52% male; 59.8% White) in the US who were followed longitudinally from Grade 1 to Grade 3. Results showed a rapid decrease in temporal eye-movement measures (e.g., first fixation) and an increase in spatial eye-movement measures (initial landing position) in both oral and silent reading. The majority of variance in eye movements was attributable to individual differences whereas some variance in initial landing position was due to text differences. Most eye-movement measures had nonlinear growth trajectories where fast development tapered off near the end of Grade 3 while initial fixation count and total gaze count in silent reading had a linear growth trajectory. The findings provide a first large-scale look into the developmental progression of eye movements during oral and silent reading during a critical period when reading skills rapidly develop.

Reading spaced and unspaced Korean text: Evidence from eye-tracking during reading

In written Korean, spaces appear between phrasal units ("eojeols"). In Experiment 1, participants read sentences in which space information had been manipulated. Results indicated that removing spaces or replacing them with a symbol hindered reading, but this effect was not as disruptive as previously found in English. Experiment 2 presented sentences varying in the proportion of eojeols that ended with postpositional particles as well as the presence/absence of spaces. Results showed that space removal interfered with reading, but its effects were weaker when the sentence contained more postpositional particles. This suggests that postpositional particles provide an extra cue to word segmentation in Korean texts. These findings are discussed in relation to the unique characteristics of the Korean writing system and to the models of eye-movement control during reading in different languages.

Enhancing the Sense of Attention from an Assistance Mobile Robot by Improving Eye-Gaze Contact from Its Iconic Face Displayed on a Flat Screen

One direct way to express the sense of attention in a human interaction is through the gaze. This paper presents the enhancement of the sense of attention from the face of a human-sized mobile robot during an interaction. This mobile robot was designed as an assistance mobile robot and uses a flat screen at the top of the robot to display an iconic (simplified) face with big round eyes and a single line as a mouth. The implementation of eye-gaze contact from this iconic face is a problem because of the difficulty of simulating real 3D spherical eyes in a 2D image considering the perspective of the person interacting with the mobile robot. The perception of eye-gaze contact has been improved by manually calibrating the gaze of the robot relative to the location of the face of the person interacting with the robot. The sense of attention has been further enhanced by implementing cyclic face explorations with saccades in the gaze and by performing blinking and small movements of the mouth.

Prediction as a basis for skilled reading: insights from modern language models

Reading is not an inborn human capability, and yet, English-speaking adults read with impressive speed. This study considered how predictions of upcoming words impact on this skilled behaviour. We used a powerful language model (GPT-2) to derive predictions of upcoming words in text passages. These predictions were highly accurate and showed a tight relationship to fine-grained aspects of eye-movement behaviour when adults read those same passages, including whether to skip the next word and how long to spend on it. Strong predictions that were incorrect resulted in a prediction error cost on fixation durations. Our findings suggest that predictions for upcoming words can be made based on the analysis of text statistics and that these predictions guide how our eyes interrogate text at very short timescales. These findings open new perspectives on reading and language comprehension and illustrate the capability of modern language models to inform understanding of human language processing.

Capacity limits in sentence comprehension: Evidence from dual-task judgements and event-related potentials

There is an ongoing controversy over whether readers can access the meaning of multiple words, simultaneously. To date, different experimental methods have generated seemingly contradictory evidence in support of serial or parallel processing accounts. For example, dual-task studies suggest that readers can process a maximum of one word at a time (White, Palmer & Boynton, 2018), while ERP studies have demonstrated neural priming effects that are more consistent with parallel activation (Wen, Snell & Grainger, 2019). To help reconcile these views, I measured neural responses and behavioral accuracy in a dual-task sentence comprehension paradigm. Participants saw masked sentences and two-word phrases and had to judge whether or not they were grammatical. Grammatically correct sentences (This girl is neat) produced smaller N400 responses compared to scrambled sentences (Those girl is fled): an N400 sentence superiority effect. Critically, participants' grammaticality judgements on the same trials showed striking capacity limitations, with dual-task deficits closely matching the predictions of a serial, all-or-none processing account. Together, these findings suggest that the N400 sentence superiority effect is fully compatible with serial word recognition, and that readers are unable to process multiple sentence positions simultaneously.

Context-based facilitation of semantic access follows both logarithmic and linear functions of stimulus probability

Stimuli are easier to process when context makes them predictable, but does context-based facilitation arise from preactivation of a limited set of relatively probable upcoming stimuli (with facilitation then linearly related to probability) or, instead, because the system maintains and updates a probability distribution across all items (with facilitation logarithmically related to probability)? We measured the N400, an index of semantic access, to words of varying probability, including unpredictable words. Word predictability was measured using both cloze probabilities and a state-of-the-art machine learning language model (GPT-2). We reanalyzed five datasets (n = 138) to demonstrate and then replicate that context-based facilitation on the N400 is graded, even among unpredictable words. Furthermore, we established that the relationship between word predictability and context-based facilitation combines linear and logarithmic functions. We argue that this composite function reveals properties of the mapping between words and semantic features and how feature- and word-related information is activated on-line.

Insights Into the Processing of Collocations During L2 English Reading: Evidence From Eye Movements

We report an eye movement experiment that investigates the effects of collocation strength and contextual predictability on the reading of collocative phrases by L2 English readers. Thirty-eight Chinese English as foreign language learners (EFL) read 40 sentences, each including a specific two-word phrase that was either a strong (e.g., black coffee) or weak (e.g., bitter coffee) adjective-noun collocation and was either highly predictable or unpredictable from the previous sentence context. Eye movement measures showed that L2 reading times for the collocative phrases were sensitive to both collocation strength and contextual predictability. However, an interaction effect between these factors, which appeared relatively late in the eye movement record, additionally revealed that contextual predictability more strongly influenced time spent reading weak compared with strong collocations. This was most likely because the greater familiarity of strong collocations facilitated their integration, even in the absence of strong contextual constraint. We discuss the findings in terms of the value of collocations in second language learning.

Probabilistic modeling of orthographic learning based on visuo-attentional dynamics

How is orthographic knowledge acquired? In line with the self-teaching hypothesis, most computational models assume that phonological recoding has a pivotal role in orthographic learning. However, these models make simplifying assumptions on the mechanisms involved in visuo-orthographic processing. Against evidence from eye movement data during orthographic learning, they assume that orthographic information on novel words is immediately available and accurately encoded after a single exposure. In this paper, we describe BRAID-Learn, a new computational model of orthographic learning. BRAID-Learn is a probabilistic and hierarchical model that incorporates the mechanisms of visual acuity, lateral interference, and visual attention involved in word recognition. Orthographic learning in the model rests on three main mechanisms: first, visual attention moves over the input string to optimize the gain of information on letter identity at each fixation; second, top-down lexical influence is modulated as a function of stimulus familiarity; third, after exploration, perceived information is used to create a new orthographic representation or stabilize a better-specified representation of the input word. BRAID-Learn was challenged on its capacity to simulate the eye movement patterns reported in humans during incidental orthographic learning. In line with the behavioral data, the model predicts a larger decline with exposures in number of fixations and processing time for novel words than for known words. For novel words, most changes occur between the first and second exposure, that is to say, after creation in memory of a new orthographic representation. Beyond phonological recoding, our results suggest that visuo-attentional exploration is an intrinsic portion of orthographic learning seldom taken into consideration by models or theoretical accounts.