Toward an appropriate baseline for measures of eye movement behavior during reading

In empirical studies of human eye movement behavior during reading, it is common to compute various summary measures from the data, but these measures are typically not evaluated with respect to corresponding measures of baseline performance. The authors present a method for deriving an appropriate baseline by mapping the actual behavior to a random perturbation of the text being read, and they find surprising similarities between the baseline and the empirical data. The practical message from these findings is that the importance of a particular factor in explaining either empirical or simulated eye movement patterns should be evaluated using an appropriate baseline. In addition to this methodological point, the authors suggest that eye movement behavior in reading may be constrained by the properties of an eye-guidance system that has adapted to the coarse-grained statistical properties of written language.     

A contingent speech technique in eye movement research on reading

A novel eye-movement-contingent method is presented. It builds on and extends established eye-movement-contingent visual display change methods in that it uses movements of the eyes to control the presentation of acoustic information during sentence reading. In one implementation, an irrelevant spoken word is presented when the eyes cross a predetermined spatial boundary before they move on to a selected visual target word. The relationship between the spoken word and the visual target is manipulated, and the pattern of interference, caused by the presentation of the spoken word, is used to determine the nature and time course of activated representations. Results from three recently completed experiments in which the technique was used show that a word’s phonological code remains active after it has been read and that the activated code has speech-like properties.     

The processing of homophonic homographs during reading: Evidence from eye movement studies

Research on the processing of homophonic homographs during reading is reviewed. The primary dependent variable considered was fixation time on target homographs. Both the characteristics of the homograph (whether there are two equally likely meanings or one dominant meaning) and the characteristics of the preceding context (whether it is neutral or contains disambiguating information) were varied. When the preceding context was neutral, readers fixated longer on balanced homographs (homographs having two equally likely meanings) than on control words matched on frequency and length, but didn't look any longer at biased homographs (homographs having a highly dominant meaning) than matched control words. However, when the preceding context disambiguated toward the subordinate meaning, readers fixated longer on a biased homograph than a matched control word (the subordinate bias effect). Attempts to eliminate the subordinate bias effect are described and the implications of our research for models of lexical ambiguity resolution are discussed.Preparation of this paper was supported by grant DBS-9121375 from the National Science Foundation. We wish to thank our colleagues Sara Sereno and Chuck Clifton for their helpful comments on earlier drafts of this paper.     

Covert attention and eye movements during reading

Eye movements were monitored during the reading of spatially transformed text in order to examine covert attentional processes in reading. In some conditions, the sequence of letters within a word was congruent with (i.e. in the same direction as) the sequence of words in the sentence; in other conditions the direction of letters within words and the direction of words in the sentence were incongruent. In addition, the window of visible text was varied so that in some conditions only the fixated word (and all preceding words) were visible, whereas in other conditions the fixated word and the succeeding word were both visible. Readers were able to extract more parafoveal information from text when the words themselves were normal than when the letters within the words were transformed. However, with practice, readers were able to use some parafoveal information even when the words were transformed. The most important finding was that the congruity of the word and letter order had no reliable effect on the ability to extract parafoveal information and influenced reading performance only when the words themselves were normal. We conclude that covert attention in reading is not a letter-by-letter scan that sweeps across the page, but either an asymmetric spotlight held constant on each fixation or a shifting of an attentional spotlight extending across multiletter units (possibly words) with the direction of shifts of attention closely coupled to the direction of eye movements.     

The time-course of recovery from interruption during reading: eye movement evidence for the role of interruption lag and spatial memory

Two experiments examined how interruptions impact reading and how interruption lags and the reader's spatial memory affect the recovery from such interruptions. Participants read paragraphs of text and were interrupted unpredictably by a spoken news story while their eye movements were monitored. Time made available for consolidation prior to responding to the interruption did not aid reading resumption. However, providing readers with a visual cue that indicated the interruption location did aid task resumption substantially in Experiment 2. Taken together, the findings show that the recovery from interruptions during reading draws on spatial memory resources and can be aided by processes that support spatial memory. Practical implications are discussed.     

Eye movement control during reading: Evidence for direct control

In three experiments, subjects read text as their eye movements were monitored and display changes in the text were made contingent upon the eye movements. In one experiment, a window of text moved in synchrony with the eyes. In one condition, the size of the window was constant from fixation to fixation, while in the other condition the size of the window varied from fixation to fixation. In the other experiments, a visual mask was presented at the end of each saccade which delayed the onset of the text, and the length of the delay was varied. The pattern of eye movements was influenced by both the size of the window and the delay of the onset of the text, even when the window size or text delay was varying from fixation to fixation. However, there was also evidence that saccade length was affected by the size of the window on the prior fixation and that certain decisions to move the eye are programmed either before the fixation begins or are programmed during the fixation but without regard to the text fixated on. The results thus provide support for a mixed control model of eye movements in reading, in which decisions about when and where to move the eyes are based on information from the current fixation, the prior fixations, and possibly, other sources as well.     

An anatomically constrained, stochastic model of eye movement control in reading

This article presents SERIF, a new model of eye movement control in reading that integrates an established stochastic model of saccade latencies (LATER; R. H. S. Carpenter, 1981) with a fundamental anatomical constraint on reading: the vertically split fovea and the initial projection of information in either visual field to the contralateral hemisphere. The novel features of the model are its simulation of saccade latencies as a race between two stochastic rise-to-threshold LATER units and its probabilistic selection of the target for the next saccade. The model generates simulated eye movement behavior that exhibits important characteristics of actual eye movements made during reading; specifically, simulations produce realistic saccade target distributions and replicate a number of critical reading phenomena, including the effects of word frequency on fixation durations, the inverted optimal viewing position effect, the trade-off between first and second fixation durations of refixated words, and the dependence of parafoveal preview benefit on eccentricity.     

自然阅读的脑成像研究及其与眼动技术结合

以往的研究者主要采用单词系列呈现的实验范式对词、句子和篇章等材料进行阅读脑成像研究。针对近年来使用自然阅读范式的功能磁共振(fMRI)研究进行分析,建议在该范式下使用脑-眼动同步记录的方式,并介绍了基于注视编码的fMRI数据分析原理。新的范式和技术提高了fMRI研究的生态效度,有利于对阅读中眼动加工的脑机制进行探索,是相关研究的新趋势。未来需要在自然阅读fMRI实验中结合实验设计及其他技术以克服时间分辨率低的问题,同时要注意探讨阅读和眼动脑区的关系。     

The orthographic uniqueness point and eye movements during reading

Recent research found that naming and lexical decision times for words with an early orthographic uniqueness point (OUP) were faster than for words with a late OUP ( Kwantes & Mewhort, 1999a ; Lindell, Nicholls, & Castles, 2003 ). A word's OUP corresponds to the letter position in the word where that word is differentiated from other words. These results have been presented as evidence for sequential letter processing in visual word recognition ( Kwantes & Mewhort, 1999a ). In two experiments, we attempted to extend these results to a more natural reading situation by recording participants’ eye movements. Readers read sentences with early or late OUP words embedded in them. In both experiments, we manipulated the amount of parafoveal information available during reading. Readers did not show any consistent benefit for reading words with an early OUP regardless of the amount of preview available. Our results are at odds with the naming and lexical decision data and prove problematic for models that predict OUP effects.     

The word grouping hypothesis and eye movements during reading

The distribution of landing positions and durations of first fixations in a region containing a noun preceded by either an article (e.g., the soldiers) or a high-frequency 3-letter word (e.g., all soldiers) were compared. Although there were fewer first fixations on the blank space between the high-frequency 3-letter word and the noun than on the surrounding letters (and the fixations on the blank space were shorter), this pattern did not occur when the noun was preceded by an article. R. Radach (1996) inferred from a similar experiment that did not manipulate the type of short word that 2 words could be processed as a perceptual unit during reading when the first word is a short word. As this different pattern of fixations is restricted to article-noun pairs, it indicates that word grouping does not occur purely on the basis of word length during reading; moreover, as the authors demonstrate, one can explain the observed patterns in both conditions more parsimoniously without adopting a word-grouping mechanism in eye movement control during reading.     

A head-mounted display holder for eye movement recording

A lightweight head-mounted display holder is described which eliminates the need to hold the head fixed during eye movement recording. A comparison of this device with a standard fixed-head condition indicates that performance is comparable under the two conditions but that subjects report less discomfort with the head-mounted display holder.     

A head-mounted scene holder for eye movement research

An inexpensive head-mounted scene holder is described. In contrast to the usual biteboard/fixed-head approach, the present apparatus allows head movements without introducing substantial artifact. This enables eye movement investigations of brain-damaged or psychiatric subjects who often cannot maintain prolonged head immobility; also, the unit makes possible subjects’ verbal reports coincident with visual scanning.     

Perceived movement of the afterimage during eye movements

The question of whether an afterimage viewed in a dark field appears to move during eye movement was studied by comparing recordings of eye movements with recordings of reports of perceived movement. The correlation was found to be quite good even under conditions where the eye movements were spontaneous rather than specifically directed. The results were taken to support the hypothesis that the behavior of the retinal image is “interpreted” by taking into account information concerning what the eyes are doing.     

Effects of syntactic prominence on eye movements during reading

Three experiments explored whether patterns of eye movements during reading might help explain syntactic prominence effects that are typically observed using reaction time tasks. Participants read sentences in which target words were in syntactically prominent or syntactically less prominent positions. Across all three experiments, using three types of syntactic prominence manipulations, there were fewer fixations and shorter reading times for words in more prominent positions, indicating that enhanced accessibility of syntactically prominent words is not caused by increased processing time. Rather, syntactic prominence appears to facilitate early encoding/lexical access and sentence integration processes while also, as shown previously, increasing activation of concepts in a comprehender’s sentence or discourse representation. We propose that enhanced encoding and sentence integration processes can be attributed to an increase in attentional resources for more prominent concepts, and that this increase derives from readers’ immediate sensitivity to informational prominence contours that are signaled by syntax.     

iMap: a novel method for statistical fixation mapping of eye movement data

Eye movement data analyses are commonly based on the probability of occurrence of saccades and fixations (and their characteristics) in given regions of interest (ROIs). In this article, we introduce an alternative method for computing statistical fixation maps of eye movements—iMap—based on an approach inspired by methods used in functional magnetic resonance imaging. Importantly, iMap does not require the a priori segmentation of the experimental images into ROIs. With iMap, fixation data are first smoothed by convolving Gaussian kernels to generate three-dimensional fixation maps. This procedure embodies eyetracker accuracy, but the Gaussian kernel can also be flexibly set to represent acuity or attentional constraints. In addition, the smoothed fixation data generated by iMap conform to the assumptions of the robust statistical random field theory (RFT) approach, which is applied thereafter to assess significant fixation spots and differences across the three-dimensional fixation maps. The RFT corrects for the multiple statistical comparisons generated by the numerous pixels constituting the digital images. To illustrate the processing steps of iMap, we provide sample analyses of real eye movement data from face, visual scene, and memory processing. The iMap MATLAB toolbox is editable and freely available for download online ().     

Masking of foveal and parafoveal vision during eye fixations in reading

A window or visual mask as moved across text in synchrony with the reader's eye movements. The size of the window or mask was varied so that either information in foveal or parafoveal vision was masked on each fixation. In another experiment, the onset of the mask was delayed for a certain amount of time following the end of the saccade. The results of the experiments point out the relative importance of foveal and parafoveal vision for reading and further indicate that most of the visual information necessary for reading can be acquired during the first 50 msec that information is available during an eye fixation.     

Mislocated fixations can account for parafoveal-on-foveal effects in eye movements during reading

Contrasting predictions of serial and parallel views on the processing of foveal and parafoveal information during reading were tested. A high-frequency adjective (young) was followed by either a high-frequency word(n) (child) or a low-frequency word(n) (tenor), which in turn was followed by either a correct (performing) or an orthographic illegal word(n + 1) (pxvforming) as a parafoveal preview. A limited parafoveal-on-foveal effect was observed: There were inflated fixation times on word(n) when the preview of word(n + 1) was orthographically illegal. However, this parafoveal-on-foveal effect was (a) independent of the frequency of word(n), (b) restricted to those instances when the eyes were very close to word(n + 1), and (c) associated with relatively long prior saccades. These observations are all compatible with a mislocated fixation account in which parafoveal-on-foveal effects result from saccadic undershoots of word(n + 1) and with a serial model of eye movement control during reading.     

Direction-specific motion thresholds for abnormal image shifts during saccadic eye movement

Eye movements were monitored and a target circle subtending an angle of 7^o was made to move during and dependent on the eye movements. Thresholds of detection of the resulting abnormal image displacements were obtained. Thresholds were low when both the eyes and the target moved either horizontally or vertically. They were higher by a factor of two or more when the eye movements and the target motions were not in the same plane. In the latter conditions, two processes account for the detection of target motion. One is a compensation process where the extent of that component of the motion of the retinal image of the target which is parallel to the eye movement is compared with the extent of the eye movement. The other process detects an angle between the plane of the target image motion and the plane of the eye movement. Our results indicate that the higher thresholds occurred when detection of this angle was required.     

Mislocated fixations can account for parafoveal-on-foveal effects in eye movements during reading

Contrasting predictions of serial and parallel views on the processing of foveal and parafoveal information during reading were tested. A high-frequency adjective (young) was followed by either a high-frequency word_n (child) or a low-frequency word_n (tenor), which in turn was followed by either a correct (performing) or an orthographic illegal word_{n + 1} (pxvforming) as a parafoveal preview. A limited parafoveal-on-foveal effect was observed: There were inflated fixation times on word_n when the preview of word_{n + 1} was orthographically illegal. However, this parafoveal-on-foveal effect was (a) independent of the frequency of word_n, (b) restricted to those instances when the eyes were very close to word_{n + 1}, and (c) associated with relatively long prior saccades. These observations are all compatible with a mislocated fixation account in which parafoveal-on-foveal effects result from saccadic undershoots of word_{n + 1} and with a serial model of eye movement control during reading.