An analysis of reading skill development using E-Z Reader

Previously reported simulations using the E-Z Reader model of eye-movement control suggest that the patterns of eye movements observed with children versus adult readers reflect differences in lexical processing proficiency. However, these simulations fail to specify precisely what aspect(s) of lexical processing (e.g., orthographic processing) account for the concurrent changes in eye movements and reading skill. To examine this issue, the E-Z Reader model was first used to simulate the aggregate eye-movement data from 15 adults and 75 children to replicate the finding that gross differences in reading skill can be accounted for by differences in lexical processing proficiency. The model was then used to simulate the eye-movement data of individual children so that the best-fitting lexical processing parameters could be correlated to measures of orthographic knowledge, phonological processing skill, sentence comprehension, and general intelligence. These analyses suggest that orthographic knowledge accounts for variance in the eye-movement measures that is observed with between-individual differences in reading skill. The theoretical implications of this conclusion will be discussed in relation to computational models of reading and our understanding of reading skill development.     

Using Reinforcement Learning to Examine Dynamic Attention Allocation During Reading

A fundamental question in reading research concerns whether attention is allocated strictly serially, supporting lexical processing of one word at a time, or in parallel, supporting concurrent lexical processing of two or more words (Reichle, Liversedge, Pollatsek, & Rayner, 2009). The origins of this debate are reviewed. We then report three simulations to address this question using artificial reading agents (Liu & Reichle, 2010; Reichle & Laurent, 2006) that learn to dynamically allocate attention to 1–4 words to “read” as efficiently as possible. These simulation results indicate that the agents strongly preferred serial word processing, although they occasionally attended to more than one word concurrently. The reason for this preference is discussed, along with implications for the debate about how humans allocate attention during reading.     

Using reinforcement learning to understand the emergence of "intelligent" eye-movement behavior during reading

The eye movements of skilled readers are typically very regular (K. Rayner, 1998). This regularity may arise as a result of the perceptual, cognitive, and motor limitations of the reader (e.g., limited visual acuity) and the inherent constraints of the task (e.g., identifying the words in their correct order). To examine this hypothesis, reinforcement learning was used to allow an artificial "agent" to learn to move its eyes to read as efficiently as possible. The resulting patterns of simulated eye movements resembled those of skilled readers and suggest that important aspects of eye-movement behavior might emerge as a consequence of satisfying the constraints that are imposed on readers. These results also suggest novel interpretations of some contentious empirical results, such as the fixation duration costs associated with word skipping (R. Kliegl & R. Engbert, 2005), and theoretical assumptions, for example the familiarity check in the E-Z Reader model of eye-movement control (E. D. Reichle, A. Pollatsek, D. L. Fisher, & K. Rayner, 1998).     

Inhibitory neighbor priming effects in eye movements during reading

We report an eye movement experiment investigating whether prior processing of a word’s orthographic neighbor in a sentence influences subsequent word processing during reading. There was greater difficulty in early word processing when a target word’s neighbor rather than a control word appeared earlier in a sentence; this effect was uninfluenced by the relative lexical frequency of the word and its neighbor. We discuss this inhibitory neighbor priming effect in terms of competitive network models of word recognition and the process of lexical identification in the E-Z Reader model of oculomotor control (e.g., Reichle, Pollatsek, Fisher, & Rayner, 1998).     

词汇预测性对中文高、低阅读技能儿童眼动行为的影响

操纵目标词的预测性、词频和阅读技能水平,考察句子阅读中词汇预测性对高、低阅读技能儿童眼动行为的影响,揭示其在儿童阅读发展中的作用。结果显示：儿童对高预测词的跳读率更高、注视时间更短,且预测性与词频交互影响跳读率和注视时间;预测性对高阅读技能儿童早期的跳读率影响更大,而对低阅读技能儿童晚期的再阅读时间具有更大影响。结果表明：词汇预测性影响儿童阅读的眼动行为和词汇加工,且作用大小和发生时程受阅读技能调节。     

Unsegmented text delays word identification: Evidence from a survival analysis of fixation durations

The present study employed distributional analyses of fixation times to examine the impact of removing spaces between words during reading. Specifically, we presented high and low frequency target words in a normal text condition that contained spaces (e.g., “John decided to sell the table in the garage sale”) and in an unsegmented text condition that contained random numbers instead of spaces (e.g., “John4decided8to5sell9the7table2in3the9garage6sale”). The unsegmented text condition produced larger word frequency effects relative to the normal text condition for the gaze duration and total time measures (for similar findings, see Rayner, Fischer, & Pollatsek, 1998), which indicates that removing spaces can impact the word identification stage of reading. To further examine the effect of spacing on word identification, we used distributional analyses of first-fixation durations to contrast the time course of word frequency effects in the normal versus the unsegmented text conditions. In replication of prior findings (Reingold, Reichle, Glaholt, & Sheridan, 2012; Staub, White, Drieghe, Hollway, & Rayner, 2010), ex-Gaussian fitting revealed that the word frequency variable impacted both the shift and the skew of the distributions, and this pattern of results occurred for both the normal and unsegmented text conditions. In addition, a survival analysis technique revealed a later time course of word frequency effects in the unsegmented relative to the normal condition, such that the earliest discernible influence of word frequency was 112 ms from the start of fixation in the normal text condition, and 152 ms in the unsegmented text condition. This delay in the temporal onset of word frequency effects in the unsegmented text condition strongly suggests that removing spaces delays the word identification stage of reading. Possible underlying mechanisms are discussed, including lateral masking and word segmentation.     

语境预测性对中文高低阅读技能儿童预视加工的影响

采用呈现随注视变化的边界范式(boundary paradigm),操作目标词的语境预测性和预视类型,考察高、低阅读技能儿童在词汇预视加工中语境预测效应的差异。结果发现,高、低阅读技能儿童对语境预测性的使用在发生时程上存在差异,高阅读技能儿童从词汇副中央凹预视加工阶段开始利用语境预测性信息,而低阅读技能儿童则在加工晚期更加依赖语境预测性信息,符合预测编码框架理论对儿童语境预测性效应差异的解释。     

Using E-Z reader to model the effects of higher level language processing on eye movements during reading

Although computational models of eye-movement control during reading have been used to explain how saccadic programming, visual constraints, attention allocation, and lexical processing jointly affect eye movements during reading, these models have largely ignored the issue of how higher level, postlexical language processing affects eye movements. The present article shows how one of these models, E-Z Reader (Pollatsek, Reichle, & Rayner, 2006c), can be augmented to redress this limitation. Simulations show that with a few simple assumptions, the model can account for the fact that effects of higher level language processing are not observed on eye movements when such processing is occurring without difficulty, but can capture the patterns of eye movements that are observed when such processing is slowed or disrupted.     

Task effects reveal cognitive flexibility responding to frequency and predictability: Evidence from eye movements in reading and proofreading

It is well-known that word frequency and predictability affect processing time. These effects change magnitude across tasks, but studies testing this use tasks with different response types (e.g., lexical decision, naming, and fixation time during reading; Schilling, Rayner, & Chumbley, 1998), preventing direct comparison. Recently, Kaakinen and Hyönä (2010) overcame this problem, comparing fixation times in reading for comprehension and proofreading, showing that the frequency effect was larger in proofreading than in reading. This result could be explained by readers exhibiting substantial cognitive flexibility, and qualitatively changing how they process words in the proofreading task in a way that magnifies effects of word frequency. Alternatively, readers may not change word processing so dramatically, and instead may perform more careful identification generally, increasing the magnitude of many word processing effects (e.g., both frequency and predictability). We tested these possibilities with two experiments: subjects read for comprehension and then proofread for spelling errors (letter transpositions) that produce nonwords (e.g., trcak for track as in Kaakinen & Hyönä) or that produce real but unintended words (e.g., trial for trail) to compare how the task changes these effects. Replicating Kaakinen and Hyönä, frequency effects increased during proofreading. However, predictability effects only increased when integration with the sentence context was necessary to detect errors (i.e., when spelling errors produced words that were inappropriate in the sentence; trial for trail). The results suggest that readers adopt sophisticated word processing strategies to accommodate task demands.     

Rethinking parafoveal processing in reading: Serial-attention models can explain semantic preview benefit and N+2 preview effects

During reading, some information about the word to the right of fixation in the parafovea is typically acquired prior to that word being fixated. Although some degree parafoveal processing is uncontroversial, its precise nature and extent are unclear. For example, can it advance up to the level of semantic processing? Additionally, can it extend across more than two spatially adjacent words? Affirmative answers to either of these questions would seemingly be problematic for serial-attention models of eye-movement control in reading, which maintain that attention is allocated to only one word at a time (see Reichle, 2011). However, in this paper we report simulation results using one such model, E-Z Reader (Reichle, Pollatsek, Fisher, & Rayner, 1998), to examine the two preceding questions. These results suggest the existence of both semantic preview and N+2 preview effects, indicating that they are not incompatible with serial-attention models. We discuss the implications of these findings for models of eye-movement control in reading and provide a new theoretical framework for conceptualizing parafoveal processing during reading and its influence on eye movement behaviour.     

Investigating the causes of wrap-up effects: Evidence from eye movements and E-Z Reader

Wrap-up effects in reading have traditionally been thought to reflect increased processing associated with intra- and inter-clause integration (Just, M. A. & Carpenter, P. A. (1980). A theory of reading: From eye fixations to comprehension. Psychological Review,87(4), 329-354; Rayner, K., Kambe, G., & Duffy, S. A. (2000). The effect of clause wrap-up on eye movements during reading. The Quarterly Journal of Experimental Psychology,53A(4), 1061-1080; cf. Hirotani, M., Frazier, L., & Rayner, K. (2006). Punctuation and intonation effects on clause and sentence wrap-up: Evidence from eye movements. Journal of Memory and Language,54, 425-443). We report an eye-tracking experiment with a strong manipulation of integrative complexity at a critical word that was either sentence-final, ended a comma-marked clause, or was not comma-marked. Although both complexity and punctuation had reliable effects, they did not interact in any eye-movement measure. These results as well as simulations using the E-Z Reader model of eye-movement control (Reichle, E. D., Warren, T., & McConnell, K. (2009). Using E-Z Reader to model the effects of higher-level language processing on eye movements during reading. Psychonomic Bulletin & Review,16(1), 1-20) suggest that traditional accounts of clause wrap-up are incomplete.     

Eye movements and parafoveal preview of compound words: Does morpheme order matter?

Recently, there has been considerable debate about whether readers can identify multiple words in parallel or whether they are limited to a serial mode of word identification, processing one word at a time (see, e.g., Reichle, Liversedge, Pollatsek, & Rayner, 2009). Similar questions can be applied to bimorphemic compound words: Do readers identify all the constituents of a compound word in parallel, and does it matter which of the morphemes is identified first? We asked subjects to read compound words embedded in sentences while monitoring their eye movements. Using the boundary paradigm (Rayner, 1975), we manipulated the preview that subjects received of the compound word before they fixated it. In particular, the morpheme order of the preview was either normal (cowboy) or reversed (boycow). Additionally, we manipulated the preview availability for each of the morphemes separately. Preview was thus available for the first morpheme only (cowtxg), for the second morpheme only (enzboy), or for neither of the morphemes (enztxg). We report three major findings: First, there was an effect of morpheme order on gaze durations measured on the compound word, indicating that, as expected, readers obtained a greater preview benefit when the preview presented the morphemes in the correct order than when their order was reversed. Second, gaze durations on the compound word were influenced not only by preview availability for the first, but also by that for the second morpheme. Finally, and most importantly, the results show that readers are able to extract some morpheme information even from a reverse order preview. In summary, readers obtain preview benefit from both constituents of a short compound word, even when the preview does not reflect the correct morpheme order.     

Using eye movements to understand the leakage of information during Chinese reading

Psychonomic Bulletin & Review - How is attention allocated during reading? The present eye-movement experiment used a paradigm developed by Liu and Reichle (Psychological Science, 29,...     

Can We ‘Read’ the Eye-Movement Patterns of Readers? Unraveling the Relationship Between Reading Profiles and Processing Strategies

In an eye-tracking experiment we examined the risky reading hypothesis, in which long saccades and many regressions are considered to be indicative of a proactive reading style (Rayner et al. in Psychol Aging 21(3):448, 2006; Psychol Aging 24(3):755, 2009). We did so by presenting short texts—that confirmed or disconfirmed verb-based implicit causality expectations—to two types of readers: proactive readers (long saccades, many regressions) and conservative readers (short saccades, few regressions). Whereas proactive readers used implicit causality information to predict upcoming referents, and slowed down immediately when they encountered a pronoun that was inconsistent with these verb-based expectations, the conservative readers slowed down much later in the sentence. These findings were consistent with the predictions of the risky reading hypothesis and as such presented novel evidence for the general idea that the eye-movement profile of readers reveals valuable information about their processing strategy.     

The gaze-contingent moving window in reading: Development and review

The development of the gaze-contingent moving window paradigm (McConkie & Rayner, 1975, 1976) is discussed and the results of the earliest research are reviewed. The original work suggested that the region from which readers can obtain useful information during an eye fixation in reading, or the perceptual span, was asymmetric around the fixation point, and extended from 3–4 letter spaces to the left of fixation to about 14–15 letter spaces to the right of fixation. Subsequent research which substantiated these findings is discussed. Then more recent research using the moving window paradigm to investigate the following topics (1) effects of reading speed, (2) effects of reading skill, (3) effects of the writing system, (4) effects due to age, (5) effects related to deafness, and (5) effects related to schizophrenia is discussed. Finally, some extensions of gaze-contingent paradigms to areas other than reading are discussed.     

Lexical restructuring in the absence of literacy

Vocabulary growth was suggested to prompt the implementation of increasingly finer-grained lexical representations of spoken words in children (e.g., [Metsala, J. L., & Walley, A. C. (1998). Spoken vocabulary growth and the segmental restructuring of lexical representations: precursors to phonemic awareness and early reading ability. In J. L. Metsala & L. C. Ehri (Eds.), Word recognition in beginning literacy (pp. 89-120). Hillsdale, NJ: Erlbaum.]). Although literacy was not explicitly mentioned in this lexical restructuring hypothesis, the process of learning to read and spell might also have a significant impact on the specification of lexical representations (e.g., [Carroll, J. M., & Snowling, M. J. (2001). The effects of global similarity between stimuli on children’s judgments of rime and alliteration. Applied Psycholinguistics, 22, 327-342.]; [Goswami, U. (2000). Phonological representations, reading development and dyslexia: Towards a cross-linguistic theoretical framework. Dyslexia, 6, 133-151.]). This is what we checked in the present study. We manipulated word frequency and neighborhood density in a gating task (Experiment 1) and a word-identification-in-noise task (Experiment 2) presented to Portuguese literate and illiterate adults. Ex-illiterates were also tested in Experiment 2 in order to disentangle the effects of vocabulary size and literacy. There was an interaction between word frequency and neighborhood density, which was similar in the three groups. These did not differ even for the words that are supposed to undergo lexical restructuring the latest (low frequency words from sparse neighborhoods). Thus, segmental lexical representations seem to develop independently of literacy. While segmental restructuring is not affected by literacy, it constrains the development of phoneme awareness as shown by the fact that, in Experiment 3, neighborhood density modulated the phoneme deletion performance of both illiterates and ex-illiterates.     

Dyslexia: saccadic eye movements

This paper describes an extensive study of the parameters of saccadic eye movement in a group of 28 poor-reading children and a comparative normally reading group of 31 children. Ages ranged from 6.0 to 16.9 yr. Poor readers had normal intelligence but were lagging by at least two years in reading ability as compared to their peer age group. Parameters studied in refixation eye movements included saccadic latency, accuracy, velocity, and acceleration, as well as differences in latency for abduct and adduct movements. Shapes of saccades were also studied. It was shown that eye-movement performance in simple saccadic refixation was not distinguishably different for poor readers and normal readers, even though in reading, which requires higher processing, eye-movement characteristics are quite different. Maturational changes for various saccadic eye-movement parameters were also examined for the normal and poor reading groups.     

Reading sentences of uniform word length – II: Very rapid adaptation of the preferred saccade length

In the current study we investigated whether readers adjust their preferred saccade length (PSL) during reading on a trial-by-trial basis. The PSL refers to the distance between a saccade launch site and saccade target (i.e., the word center during reading) when participants neither undershoot nor overshoot this target (McConkie, Kerr, Reddix, & Zola in Vision Research, 28, 1107-1118, 1988). The tendency for saccades longer or shorter than the PSL to under or overshoot their target is referred to as the range error. Recent research by Cutter, Drieghe, and Liversedge (Journal of Experimental Psychology: Human Perception and Performance, 2017) has shown that the PSL changes to be shorter when readers are presented with 30 consecutive sentences exclusively made of three-letter words, and longer when presented with 30 consecutive sentences exclusively made of five-letter words. We replicated and extended this work by this time presenting participants with these uniform sentences in an unblocked design. We found that adaptation still occurred across different sentence types despite participants only having one trial to adapt. Our analyses suggested that this effect was driven by the length of the words readers were making saccades away from, rather than the length of the words in the rest of the sentence. We propose an account of the range error in which readers use parafoveal word length information to estimate the length of a saccade between the center of two parafoveal words (termed the Centre-Based Saccade Length) prior to landing on the first of these words.     

The role of form and meaning in the processing of written morphology: A priming study in French developing readers

Three visual priming experiments using three different prime durations (60 ms in Experiment 1, 250 ms in Experiment 2, and 800 ms in Experiment 3) were conducted to examine which properties of morphemes (form and/or meaning) drive developing readers’ processing of written morphology. French third, fifth, and seventh graders and adults (the latter as a control group) performed lexical decision tasks in which targets were preceded by morphological (e.g., tablette–TABLE, “little table–table”), pseudoderived (e.g., baguette–BAGUE, “little stick–ring”), orthographic control (e.g., abricot–ABRI, “apricot–shelter”), and semantic control (e.g., Tulipe–FLEUR, “tulip–flower”) primes. Across all groups, different patterns of priming were observed in both morphological and orthographic/semantic control conditions, suggesting that they all process morphemes as units when reading. In developing readers, the processing of written morphology is triggered by the form properties of morphemes, and their semantic properties are activated later in the time course of word recognition. In adults, patterns of priming were similar except that the activation of the form properties of morphemes decreased earlier in the time course of word recognition. Taken together, these findings indicate that French developing readers process both the form and meaning properties of morphemes when reading and support a progressive quantitative change in the development of morphological processing over the course of reading development.     

Tracking the acquisition of orthographic skills in developing readers: masked priming effects

A masked priming procedure was used to explore developmental changes in the tuning of lexical word recognition processes. Lexical tuning was assessed by examining the degree of masked form priming and used two different types of prime-target lexical similarity: one letter different (e.g., rlay-->PLAY) and transposed letters (e.g., lpay-->PLAY). The performance of skilled adult readers was compared with that of developing readers in Grade 3. The same children were then tested again two years later, when they were in Grade 5. The skilled adult readers showed no form priming, indicating that their recognition mechanisms for these items had become finely tuned. In contrast, the Grade 3 readers showed substantial form priming effects for both measures of lexical similarity. When retested in Grade 5, the developing readers no longer showed significant one letter different priming, but transposed letter priming remained. In general, these results provide evidence for a transition from more broadly tuned to more finely tuned lexical recognition mechanisms and are interpreted in the context of models of word recognition.