Eye Movement Control In Reading Research Articles

Language models outperform cloze predictability in a cognitive model of reading.

Although word predictability is commonly considered an important factor in reading, sophisticated accounts of predictability in theories of reading are lacking. Computational models of reading traditionally use cloze norming as a proxy of word predictability, but what cloze norms precisely capture remains unclear. This study investigates whether large language models (LLMs) can fill this gap. Contextual predictions are implemented via a novel parallel-graded mechanism, where all predicted words at a given position are pre-activated as a function of contextual certainty, which varies dynamically as text processing unfolds. Through reading simulations with OB1-reader, a cognitive model of word recognition and eye-movement control in reading, we compare the model's fit to eye-movement data when using predictability values derived from a cloze task against those derived from LLMs (GPT-2 and LLaMA). Root Mean Square Error between simulated and human eye movements indicates that LLM predictability provides a better fit than cloze. This is the first study to use LLMs to augment a cognitive model of reading with higher-order language processing while proposing a mechanism on the interplay between word predictability and eye movements.

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.

Processing multiconstituent units: Preview effects during reading of Chinese words, idioms, and phrases.

Arguably, the most contentious debate in the field of eye movement control in reading has centered on whether words are lexically processed serially or in parallel during reading. Chinese is character-based and unspaced, meaning the issue of how lexical processing is operationalized across potentially ambiguous, multicharacter strings is not straightforward. We investigated Chinese readers' processing of frequently occurring multiconstituent units (MCUs), that is, linguistic units composed of more than a single word, that might be represented lexically as a single representation. In Experiment 1, we manipulated the linguistic category of a two-constituent Chinese string (word, MCU, or phrase) and the preview of its second constituent (identical or pseudocharacter) using the boundary paradigm with the boundary located before the two-constituent string. A robust preview effect was obtained when the second constituent, alongside the first, formed a word or MCU, but not a phrase, suggesting that frequently occurring MCUs are lexicalized and processed parafoveally as single units during reading. In Experiment 2, we further manipulated the phrase type of a two-constituent but three-character Chinese string (idiom with a one-character modifier and a two-character noun, or matched phrase) and the preview of the second constituent noun (identity or pseudocharacter). A greater preview effect was obtained for idioms than phrases, indicating that idioms are processed to a greater extent in the parafovea than matched phrases. Together, the results of these two experiments suggest that lexical identification processes in Chinese can be operationalized over linguistic units that are larger than an individual word. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Word length and frequency effects on text reading are highly similar in 12 alphabetic languages

Reading research robustly finds that shorter and more frequent words are recognized faster and skipped more often than longer and less frequent words. An empirical question that has not been tested yet is whether languages within the same writing system would produce similarly strong length and frequency effects or whether typological differences between written languages would cause those effects to vary systematically in their magnitude. We analyzed text reading eye-movement data in 12 alphabetic languages from the Multilingual Eye-Movement Corpus (MECO). The languages varied substantially in their word length and frequency distributions as a function of their orthographic depth and morpho-syntactic type. Yet, the effects of word length and frequency on fixation durations and skipping rate were highly similar in size between the languages. This finding suggests a high degree of cross-linguistic universality in the readers’ behavioral response to linguistic complexity (indexed by word length) and the amount of experience with the word (indexed by word frequency). These findings run counter to influential theories of single word recognition, which predict orthographic depth of a language to modulate the size of these benchmark effects. They also facilitate development of cross-linguistically generalizable computational models of eye-movement control in reading.

The function/content word distinction and eye movements in reading.

A substantial quantity of research has explored whether readers' eye movements are sensitive to the distinction between function and content words. No clear answer has emerged, in part due to the difficulty of accounting for differences in length, frequency, and predictability between the words in the two classes. Based on evidence that readers differentially overlook function word errors, we hypothesized that function words may be more frequently skipped or may receive shorter fixations. We present two very large-scale eyetracking experiments using selected sentences from a corpus of natural text, with each sentence containing a target function or content word. The target words in the two classes were carefully matched on length, frequency, and predictability, with the latter variable operationalized in terms of next-word probability obtained from the large language model GPT-2. While the experiments replicated a range of expected effects, word class did not have any clear influence on target word skipping probability, and there was some evidence for a content word advantage in fixation duration measures. These results indicate that readers' tendency to overlook function word errors is not due to reduced time spent encoding these words. The results also broadly support the implicit assumption in prominent models of eye movement control in reading that a word's syntactic category does not play an important role in decisions about when and where to move the eyes. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Examining semantic parafoveal-on-foveal effects using a Stroop boundary paradigm

The issue of whether lexical processing occurs serially or in parallel has been a central and contentious issue in respect of models of eye movement control in reading for well over a decade. A critical question in this regard concerns whether lexical parafoveal-on-foveal effects exist in reading. Because Chinese is an unspaced and densely packed language, readers may process parafoveal words to a greater extent than they do in spaced alphabetic languages. In two experiments using a novel Stroop boundary paradigm (Rayner, 1975), participants read sentences containing a single-character color-word whose preview was manipulated (identity or pseudocharacter, printed in black [no-color], or in a color congruent or incongruent with the character meaning). Two boundaries were used, one positioned two characters before the target and one immediately to the left of the target. The previews changed from black to color and then back to black as the eyes crossed the first and then the second boundary respectively. In Experiment 1 four color-words (red, green, yellow and blue) were used and in Experiment 2 only red and green color-words were used as targets. Both experiments showed very similar patterns such that reading times were increased for colored compared to no-color previews indicating a parafoveal visual interference effect. Most importantly, however, there were no robust interactive effects. Preview effects were comparable for congruent and incongruent color previews at the pretarget region when the data were combined from both experiments. These results favour serial processing accounts and indicate that even under very favourable experimental conditions, lexical semantic parafoveal-on-foveal effects are minimal.

Eye movement control in reading Chinese: A matter of strength of character?

This paper explores the processes underlying eye movement control in Chinese reading among a population of young 4th and 5th grade readers. Various proposals to explain the underlying mechanisms involved in eye movement control are examined and the paper concludes that the most likely account is of a two-factor process whereby the character is the main driver for longer saccades and that the word plays a role in shorter ones. A computational model is proposed to provide an integrated account of the dynamic interaction of these two factors.

Preview frequency effects in reading: evidence from Chinese.

Studies about sentence reading have shown that visual and lexical information beyond the currently fixated word can be integrated across fixations. The gaze-contingent boundary paradigm has been used widely to explore the extent to which parafoveal information can be processed before a word is fixated on. However, a critical review of the current literature suggests that unrelated mask previews are an unlikely baseline control with zero lexical activation, blurring the nature of experimental effects observed in the paradigm. The present study, therefore, aimed at shedding light on the effect of parafoveal mask properties through a manipulation of preview word frequency. Low-frequency preview words that are unrelated to target words elicited a larger interference than high-frequency preview words. We discuss implications of the preview frequency effect for computational models of eye-movement control in reading.

A dynamic adjustment model of saccade lengths in reading for word-spaced orthographies: evidence from simulations and invisible boundary experiments

ABSTRACT Contemporary models of eye movement control in reading assume a discrete target word selection process preceding saccade length computation, while the selection itself is assumed to be driven by word identification processes. However, a potentially more parsimonious, dynamic adjustment view allows both next word length and its content (e.g. orthographic) to modulate saccade length in a continuous manner. Based on a recently proposed center-based saccade length account a new regression model of forward saccade length is introduced and validated in a simulation study. Further, additional simulations and gaze-contingent invisible boundary experiments were used to study the cognitive mechanisms underlying skipping. Overall, the results support the plausibility of dynamic adjustment of saccade length in word-spaced orthographies. In the future, the present regression formula-based computational model will allow a straightforward implementation of influences of current and next word content (visual, orthographic, or contextual) on saccade length computation.

A Bayesian approach to dynamical modeling of eye-movement control in reading of normal, mirrored, and scrambled texts.

In eye-movement control during reading, advanced process-oriented models have been developed to reproduce behavioral data. So far, model complexity and large numbers of model parameters prevented rigorous statistical inference and modeling of interindividual differences. Here we propose a Bayesian approach to both problems for one representative computational model of sentence reading (SWIFT; Engbert et al., Psychological Review, 112, 2005, pp. 777-813). We used experimental data from 36 subjects who read the text in a normal and one of four manipulated text layouts (e.g., mirrored and scrambled letters). The SWIFT model was fitted to subjects and experimental conditions individually to investigate between-subject variability. Based on posterior distributions of model parameters, fixation probabilities and durations are reliably recovered from simulated data and reproduced for withheld empirical data, at both the experimental condition and subject levels. A subsequent statistical analysis of model parameters across reading conditions generates model-driven explanations for observable effects between conditions. (PsycInfo Database Record (c) 2021 APA, all rights reserved).

What Is the Difference? Rereading Shakespeare's Sonnets -An Eye Tracking Study.

Texts are often reread in everyday life, but most studies of rereading have been based on expository texts, not on literary ones such as poems, though literary texts may be reread more often than others. To correct this bias, the present study is based on two of Shakespeare’s sonnets. Eye movements were recorded, as participants read a sonnet then read it again after a few minutes. After each reading, comprehension and appreciation were measured with the help of a questionnaire. In general, compared to the first reading, rereading improved the fluency of reading (shorter total reading times, shorter regression times, and lower fixation probability) and the depth of comprehension. Contrary to the other rereading studies using literary texts, no increase in appreciation was apparent. Moreover, results from a predictive modeling analysis showed that readers’ eye movements were determined by the same critical psycholinguistic features throughout the two sessions. Apparently, even in the case of poetry, the eye movement control in reading is determined mainly by surface features of the text, unaffected by repetition.

The effect of contextual plausibility on word skipping during reading

Recent eye-movement evidence suggests readers are more likely to skip a high-frequency word than a low-frequency word independently of the semantic or syntactic acceptability of the word in the sentence. This has been interpreted as strong support for a serial processing mechanism in which the decision to skip a word is based on the completion of a preliminary stage of lexical processing prior to any assessment of contextual fit. The present large-scale study was designed to reconcile these findings with the plausibility preview effect: higher skipping and reduced first-pass reading times for words that are previewed by contextually plausible, compared to implausible, sentence continuations that are unrelated to the target word. Participants' eye movements were recorded as they read sentences containing a short (3–4 letters) or long (6 letters) target word. The boundary paradigm was used to present parafoveal previews which were either higher or lower frequency than the target, and either plausible or implausible in the sentence context. The results revealed strong, independent effects of all three factors on target skipping and early measures of target fixation duration, while frequency and plausibility interacted on later measures of target fixation duration. Simulations using the E-Z Reader model of eye-movement control in reading demonstrated that plausibility effects on skipping are potentially consistent with the assumption that higher-level contextual information only affects post-lexical integration processes. However, no current model of eye movements in reading provides an explicit account of the information or processes that allow readers to rapidly detect an integration failure.

The influence of parafoveal preview, character transposition, and word frequency on saccadic targeting in Chinese reading.

This article reports the results of an eye-movement experiment which manipulated the frequency and parafoveal preview (i.e., nonword, transposed-character, or identical) of 2-character Chinese target words using a gaze-contingent boundary paradigm (Rayner, 1975). The key findings were that progressive saccades were longer into high- than low-frequency target words, and that this word-frequency effect was more pronounced for identical than transposed previews. These findings suggest that Chinese readers adjust their saccade lengths in response to variables that influence the rate of parafoveal lexical processing. To examine the feasibility of this hypothesis, 2 computer simulations were completed that pitted this dynamic-adjustment account (Liu, Huang, Gao, & Reichle, 2017) against an account in which readers simply move their eyes to a small number of default saccade targets (e.g., the beginning or center of the upcoming word; Yan, Kliegl, Richter, Nuthmann, & Shu, 2010). The simulation results show that the dynamic-adjustment hypothesis more accurately describes our experimental findings using fewer parameters. The theoretical implications of the dynamic-adjustment account of saccadic targeting are discussed relevant to both models of eye-movement control in reading and modes of Chinese word identification. (PsycINFO Database Record (c) 2019 APA, all rights reserved).

Spill the load: Mixed evidence for a foveal load effect, reliable evidence for a spillover effect in eye-movement control during reading

It has been hypothesized that the processing difficulty of the fixated word (i.e., “foveal load”) modulates the amount of parafoveal preprocessing of the next word. Evidence for the hypothesis has been provided by the application of parafoveal masks within the boundary paradigm. Other studies that applied alternative means of manipulating the parafoveal preview (i.e., visual degradation) could not replicate the effect of foveal load. The present study examined the effect of foveal load by directly comparing the application of parafoveal masks (Exp. 1) with the alternative manipulation of visually degrading the parafoveal preview (Exp. 2) in adult readers. Contrary to expectation, we did not find the foveal-load interaction in the first experiment with traditional letter masks. We did, however, find the expected interaction in the second experiment with visually degraded previews. Both experiments revealed a spillover effect indicating that the processing of a word is not (always) fully completed when the reader already fixates the next word (i.e., processing “spills over” to the next word). The implications for models of eye-movement control in reading are discussed.

The dynamic adjustment of saccades during Chinese reading: Evidence from eye movements and simulations.

This article reports an eye-movement experiment in which participants scanned continuous sequences of Landolt-Cs for target circles to examine the visual and oculomotor constraints that might jointly determine where the eyes move in a task that engages many of the perceptual and motor processes involved in Chinese reading but without lexical or linguistic processing. The lengths of the saccades entering the Landolt-C clusters were modulated by the processing difficulty (i.e., gap sizes) of those clusters. Simulations using implemented versions of default-targeting (Yan, Kliegl, Richter, Nuthmann, & Shu, 2010) versus dynamic-adjustment (Liu, Reichle, & Li, 2016) models of saccadic targeting indicated that the latter provided a better account of our participants' eye movements, further supporting the hypothesis that Chinese readers "decide" where to move their eyes by adjusting saccade length in response to processing difficulty rather than by selecting default saccade targets. We discuss this hypothesis in relation to both what is known about saccadic targeting during the reading of English versus Chinese and current models of eye-movement control in reading. (PsycINFO Database Record (c) 2019 APA, all rights reserved).

Reading sentences of words with rotated letters: An eye movement study

Participants' eye movements were measured as they read sentences in which individual letters within words were rotated. Both the consistency of direction and the magnitude of rotation were manipulated (letters rotated all in the same direction, or alternately clockwise and anti-clockwise, by 30° or 60°). Each sentence included a target word that was manipulated for frequency of occurrence. Our objectives were threefold: To quantify how change in the visual presentation of individual letters disrupted word identification, and whether disruption was consistent with systematic change in visual presentation; to determine whether inconsistent letter transformation caused more disruption than consistent letter transformation; and to determine whether such effects were comparable for words that were high and low frequency to explore the extent to which they were visually or linguistically mediated. We found that disruption to reading was greater as the magnitude of letter rotation increased, although even small rotations affected processing. The data also showed that alternating letter rotations were significantly more disruptive than consistent rotations; this result is consistent with models of lexical identification in which encoding occurs over units of more than one adjacent letter. These rotation manipulations also showed significant interactions with word frequency on the target word: Gaze durations and total fixation duration times increased disproportionately for low-frequency words when they were presented at more extreme rotations. These data provide a first step towards quantifying the relative contribution of the spatial relationships between individual letters to word recognition and eye movement control in reading.

Visual letter similarity effects during sentence reading: Evidence from the boundary technique

The study of how the cognitive system encodes letter identities from the visual input has received much attention in models of visual word recognition but it has typically been overlooked in models of eye movement control in reading. Here we examined how visual letter similarity affects early word processing during reading using Rayner's (1975) boundary change technique in which the parafoveal preview of the target word was either identical (e.g., frito-frito [fried]) or a one-letter-different nonword (e.g., frjto-frito vs. frgto-frito). Critically, the substituted letter in the nonword was visually similar (based on letter confusability norms) or visually dissimilar. Results showed shorter viewing times on the target word when the parafoveal preview was visually similar than when it was visually dissimilar. Thus, visual letter similarity modulates the integration of parafoveal and foveal information during sentence reading. Future implementations of models of eye movement control in reading should incorporate a more developed orthographic-lexical module to capture these effects.

Influences on and consequences of parafoveal preview in reading.

During reading, information is extracted from upcoming words to the right of the currently fixated word, which facilitates recognition of those words when they are later fixated. According to the foveal load hypothesis (Henderson & Ferreira, 1990), this parafoveal preview benefit depends on how difficult the currently fixated word is to recognize. Furthermore, there is evidence that the influence of lexical variables (frequency and predictability) on word processing changes when no preview of that word is available. The present study reports two moving-window experiments in which the upcoming word to the right of fixation was either included in or excluded from the window. Through this manipulation, accurate parafoveal information was either available or not for each word in the paragraph. Two critical interactions between preview condition and lexical variables were observed. First, the word frequency at word N was found to be the primary influence on the amount of preview benefit obtained at word N+1, consistent with the foveal load hypothesis. Second, denial of preview eliminated the word predictability effect. These findings have implications for models of eye movement control in reading.

Eye movements when reading sentences with handwritten words.

The examination of how we read handwritten words (i.e., the original form of writing) has typically been disregarded in the literature on reading. Previous research using word recognition tasks has shown that lexical effects (e.g., the word-frequency effect) are magnified when reading difficult handwritten words. To examine this issue in a more ecological scenario, we registered the participants' eye movements when reading handwritten sentences that varied in the degree of legibility (i.e., sentences composed of words in easy vs. difficult handwritten style). For comparison purposes, we included a condition with printed sentences. Results showed a larger reading cost for sentences with difficult handwritten words than for sentences with easy handwritten words, which in turn showed a reading cost relative to the sentences with printed words. Critically, the effect of word frequency was greater for difficult handwritten words than for easy handwritten words or printed words in the total times on a target word, but not on first-fixation durations or gaze durations. We examine the implications of these findings for models of eye movement control in reading.

Beyond cloze probability: Parafoveal processing of semantic and syntactic information during reading

Abstract Theories of eye movement control in reading assume that early oculomotor decisions are determined by a word’s frequency and cloze probability. This assumption is challenged by evidence that readers are sensitive to the contextual plausibility of an upcoming word: First-pass fixation probability and duration are reduced when the parafoveal preview is a plausible, but unpredictable, word relative to an implausible word. The present study sought to establish whether the source of this effect is sensitivity to violations of syntactic acceptability. In two experiments, the gaze-contingent boundary paradigm was used to compare contextually plausible previews to semantically acceptable and anomalous previews that either matched or violated syntactic rules. Results showed that readers benefited from the convergence of semantic and syntactic acceptability early enough in the timecourse of reading to affect skipping. In addition, both semantic and syntactic plausibility yielded preview effects on target fixation duration measures, providing direct evidence of parafoveal syntactic processing in reading. These results highlight the limitations of relying solely on cloze probability to index contextual influences on early lexical processing. The implications of the data for models of eye movement control and language comprehension are discussed.