Refine
Year of publication
Document Type
- Article (24)
- Postprint (9)
- Doctoral Thesis (8)
- Monograph/Edited Volume (1)
Keywords
- reading (42) (remove)
Institute
- Department Psychologie (26)
- Department Linguistik (4)
- Department Erziehungswissenschaft (2)
- Humanwissenschaftliche Fakultät (2)
- Potsdam Research Institute for Multilingualism (PRIM) (2)
- Strukturbereich Kognitionswissenschaften (2)
- Department Grundschulpädagogik (1)
- Department Sport- und Gesundheitswissenschaften (1)
- Institut für Mathematik (1)
- Mathematisch-Naturwissenschaftliche Fakultät (1)
The evaluation of process-oriented cognitive theories through time-ordered observations is crucial for the advancement of cognitive science. The findings presented herein integrate insights from research on eye-movement control and sentence comprehension during reading, addressing challenges in modeling time-ordered data, statistical inference, and interindividual variability. Using kernel density estimation and a pseudo-marginal likelihood for fixation durations and locations, a likelihood implementation of the SWIFT model of eye-movement control during reading (Engbert et al., Psychological Review, 112, 2005, pp. 777–813) is proposed. Within the broader framework of data assimilation, Bayesian parameter inference with adaptive Markov Chain Monte Carlo techniques is facilitated for reliable model fitting. Across the different studies, this framework has shown to enable reliable parameter recovery from simulated data and prediction of experimental summary statistics. Despite its complexity, SWIFT can be fitted within a principled Bayesian workflow, capturing interindividual differences and modeling experimental effects on reading across different geometrical alterations of text. Based on these advancements, the integrated dynamical model SEAM is proposed, which combines eye-movement control, a traditionally psychological research area, and post-lexical language processing in the form of cue-based memory retrieval (Lewis & Vasishth, Cognitive Science, 29, 2005, pp. 375–419), typically the purview of psycholinguistics. This proof-of-concept integration marks a significant step forward in natural language comprehension during reading and suggests that the presented methodology can be useful to develop complex cognitive dynamical models that integrate processes at levels of perception, higher cognition, and (oculo-)motor control. These findings collectively advance process-oriented cognitive modeling and highlight the importance of Bayesian inference, individual differences, and interdisciplinary integration for a holistic understanding of reading processes. Implications for theory and methodology, including proposals for model comparison and hierarchical parameter inference, are briefly discussed.
The perceptual span describes the size of the visual field from which information is obtained during a fixation in reading. Its size depends on characteristics of writing system and reader, but-according to the foveal load hypothesis-it is also adjusted dynamically as a function of lexical processing difficulty. Using the moving window paradigm to manipulate the amount of preview, here we directly test whether the perceptual span shrinks as foveal word difficulty increases. We computed the momentary size of the span from word-based eye-movement measures as a function of foveal word frequency, allowing us to separately describe the perceptual span for information affecting spatial saccade targeting and temporal saccade execution. First fixation duration and gaze duration on the upcoming (parafoveal) word N + 1 were significantly shorter when the current (foveal) word N was more frequent. We show that the word frequency effect is modulated by window size. Fixation durations on word N + 1 decreased with high-frequency words N, but only for large windows, that is, when sufficient parafoveal preview was available. This provides strong support for the foveal load hypothesis. To investigate the development of the foveal load effect, we analyzed data from three waves of a longitudinal study on the perceptual span with German children in Grades 1 to 6. Perceptual span adjustment emerged early in development at around second grade and remained stable in later grades. We conclude that the local modulation of the perceptual span indicates a general cognitive process, perhaps an attentional gradient with rapid readjustment.
During sentence reading the eyes quickly jump from word to word to sample visual information with the high acuity of the fovea. Lexical properties of the currently fixated word are known to affect the duration of the fixation, reflecting an interaction of word processing with oculomotor planning. While low level properties of words in the parafovea can likewise affect the current fixation duration, results concerning the influence of lexical properties have been ambiguous (Drieghe, Rayner, & Pollatsek, 2008; Kliegl, Nuthmann, & Engbert, 2006). Experimental investigations of such lexical parafoveal-on-foveal effects using the boundary paradigm have instead shown, that lexical properties of parafoveal previews affect fixation durations on the upcoming target words (Risse & Kliegl, 2014). However, the results were potentially confounded with effects of preview validity.
The notion of parafoveal processing of lexical information challenges extant models of eye movements during reading. Models containing serial word processing assumptions have trouble explaining such effects, as they usually couple successful word processing to saccade planning, resulting in skipping of the parafoveal word. Although models with parallel word processing are less restricted, in the SWIFT model (Engbert, Longtin, & Kliegl, 2002) only processing of the foveal word can directly influence the saccade latency.
Here we combine the results of a boundary experiment (Chapter 2) with a predictive modeling approach using the SWIFT model, where we explore mechanisms of parafoveal inhibition in a simulation study (Chapter 4). We construct a likelihood function for the SWIFT model (Chapter 3) and utilize the experimental data in a Bayesian approach to parameter estimation (Chapter 3 & 4).
The experimental results show a substantial effect of parafoveal preview frequency on fixation durations on the target word, which can be clearly distinguished from the effect of preview validity. Using the eye movement data from the participants, we demonstrate the feasibility of the Bayesian approach even for a small set of estimated parameters, by comparing summary statistics of experimental and simulated data. Finally, we can show that the SWIFT model can account for the lexical preview effects, when a mechanism for parafoveal inhibition is added. The effects of preview validity were modeled best, when processing dependent saccade cancellation was added for invalid trials. In the simulation study only the control condition of the experiment was used for parameter estimation, allowing for cross validation. Simultaneously the number of free parameters was increased. High correlations of summary statistics demonstrate the capabilities of the parameter estimation approach. Taken together, the results advocate for a better integration of experimental data into computational modeling via parameter estimation.
It is well-documented that academic achievement is associated with students' self-perceptions of their academic abilities, that is, their academic self-concepts. However, low-achieving students may apply self-protective strategies to maintain a favorable academic self-concept when evaluating their academic abilities. Consequently, the relation between achievement and academic self-concept might not be linear across the entire achievement continuum. Capitalizing on representative data from three large-scale assessments (i.e., TIMSS, PIRLS, PISA; N = 470,804), we conducted an integrative data analysis to address nonlinear trends in the relations between achievement and the corresponding self-concepts in mathematics and the verbal domain across 13 countries and 2 age groups (i.e., elementary and secondary school students). Polynomial and interrupted regression analyses showed nonlinear relations in secondary school students, demonstrating that the relations between achievement and the corresponding self-concepts were weaker for lower achieving students than for higher achieving students. Nonlinear effects were also present in younger students, but the pattern of results was rather heterogeneous. We discuss implications for theory as well as for the assessment and interpretation of self-concept.
German orthography systematically marks all nouns (even other nominalized word classes) by capitalizing their first letter. It is often claimed that readers benefit from the uppercase-letter syntactic and semantic information, which makes the processing of sentences easier (e.g., Bock et al., 1985, 1989). In order to test this hypothesis, we asked 54 German readers to read single sentences systematically manipulated by a target word (N). In the experimental condition (EXP), we used semantic priming (in the following example: sick → cold) in order to build up a strong expectation of a noun, which was actually an attribute for the following noun (N+1) (translated to English e.g., “The sick writer had a cold (N) nose (N+1) …”). The sentences in the control condition were built analogously, but word N was purposefully altered (keeping word length and frequency constant) to make its interpretation as a noun extremely unlikely (e.g., “The sick writer had a blue (N) nose (N+1) …”). In both conditions, the sentences were presented either following German standard orthography (Cap) or in lowercase spelling (NoCap). The capitalized nouns in the EXP/Cap condition should then prevent garden-path parsing, as capital letters can be recognized parafoveally. However, in the EXP/NoCap condition, we expected a garden-path effect on word N+1 affecting first-pass fixations and the number of regressions, as the reader realizes that word N is instead an adjective. As the control condition does not include a garden-path, we expected to find (small) effects of the violation of the orthographic rule in the CON/NoCap condition, but no garden-path effect. As a global result, it can be stated that reading sentences in which nouns are not marked by a majuscule slows a native German reader down significantly, but from an absolute point of view, the effect is small. Compared with other manipulations (e.g., transpositions or substitutions), a lowercase letter still represents the correct allograph in the correct position without affecting phonology. Furthermore, most German readers do have experience with other alphabetic writing systems that lack consistent noun capitalization, and in (private) digital communication lowercase nouns are quite common. Although our garden-path sentences did not show the desired effect, we found an indication of grammatical pre-processing enabled by the majuscule in the regularly spelled sentences: In the case of high noun frequency, we post hoc located parafovea-on-fovea effects, i.e., longer fixation durations, on the attributive adjective (word N). These benefits of capitalization could only be detected under specific circumstances. In other cases, we conclude that longer reading durations are mainly the result of disturbance in readers' habituation when the expected capitalization is missing.
German orthography systematically marks all nouns (even other nominalized word classes) by capitalizing their first letter. It is often claimed that readers benefit from the uppercase-letter syntactic and semantic information, which makes the processing of sentences easier (e.g., Bock et al., 1985, 1989). In order to test this hypothesis, we asked 54 German readers to read single sentences systematically manipulated by a target word (N). In the experimental condition (EXP), we used semantic priming (in the following example: sick -> cold) in order to build up a strong expectation of a noun, which was actually an attribute for the following noun (N+1) (translated to English e.g., "The sick writer had a cold (N) nose (N+1) ..."). The sentences in the control condition were built analogously, but word N was purposefully altered (keeping word length and frequency constant) to make its interpretation as a noun extremely unlikely (e.g., "The sick writer had a blue (N) nose (N+1) ..."). In both conditions, the sentences were presented either following German standard orthography (Cap) or in lowercase spelling (NoCap). The capitalized nouns in the EXP/Cap condition should then prevent garden-path parsing, as capital letters can be recognized parafoveally. However, in the EXP/NoCap condition, we expected a garden-path effect on word N+1 affecting first-pass fixations and the number of regressions, as the reader realizes that word N is instead an adjective. As the control condition does not include a garden-path, we expected to find (small) effects of the violation of the orthographic rule in the CON/NoCap condition, but no garden-path effect. As a global result, it can be stated that reading sentences in which nouns are not marked by a majuscule slows a native German reader down significantly, but from an absolute point of view, the effect is small. Compared with other manipulations (e.g., transpositions or substitutions), a lowercase letter still represents the correct allograph in the correct position without affecting phonology. Furthermore, most German readers do have experience with other alphabetic writing systems that lack consistent noun capitalization, and in (private) digital communication lowercase nouns are quite common. Although our garden-path sentences did not show the desired effect, we found an indication of grammatical pre-processing enabled by the majuscule in the regularly spelled sentences: In the case of high noun frequency, we post hoc located parafovea-on-fovea effects, i.e., longer fixation durations, on the attributive adjective (word N). These benefits of capitalization could only be detected under specific circumstances. In other cases, we conclude that longer reading durations are mainly the result of disturbance in readers' habituation when the expected capitalization is missing.
German orthography systematically marks all nouns (even other nominalized word classes) by capitalizing their first letter. It is often claimed that readers benefit from the uppercase-letter syntactic and semantic information, which makes the processing of sentences easier (e.g., Bock et al., 1985, 1989). In order to test this hypothesis, we asked 54 German readers to read single sentences systematically manipulated by a target word (N). In the experimental condition (EXP), we used semantic priming (in the following example: sick → cold) in order to build up a strong expectation of a noun, which was actually an attribute for the following noun (N+1) (translated to English e.g., “The sick writer had a cold (N) nose (N+1) …”). The sentences in the control condition were built analogously, but word N was purposefully altered (keeping word length and frequency constant) to make its interpretation as a noun extremely unlikely (e.g., “The sick writer had a blue (N) nose (N+1) …”). In both conditions, the sentences were presented either following German standard orthography (Cap) or in lowercase spelling (NoCap). The capitalized nouns in the EXP/Cap condition should then prevent garden-path parsing, as capital letters can be recognized parafoveally. However, in the EXP/NoCap condition, we expected a garden-path effect on word N+1 affecting first-pass fixations and the number of regressions, as the reader realizes that word N is instead an adjective. As the control condition does not include a garden-path, we expected to find (small) effects of the violation of the orthographic rule in the CON/NoCap condition, but no garden-path effect. As a global result, it can be stated that reading sentences in which nouns are not marked by a majuscule slows a native German reader down significantly, but from an absolute point of view, the effect is small. Compared with other manipulations (e.g., transpositions or substitutions), a lowercase letter still represents the correct allograph in the correct position without affecting phonology. Furthermore, most German readers do have experience with other alphabetic writing systems that lack consistent noun capitalization, and in (private) digital communication lowercase nouns are quite common. Although our garden-path sentences did not show the desired effect, we found an indication of grammatical pre-processing enabled by the majuscule in the regularly spelled sentences: In the case of high noun frequency, we post hoc located parafovea-on-fovea effects, i.e., longer fixation durations, on the attributive adjective (word N). These benefits of capitalization could only be detected under specific circumstances. In other cases, we conclude that longer reading durations are mainly the result of disturbance in readers' habituation when the expected capitalization is missing.
The effect of decay and lexical uncertainty on processing long-distance dependencies in reading
(2020)
To make sense of a sentence, a reader must keep track of dependent relationships between words, such as between a verb and its particle (e.g. turn the music down). In languages such as German, verb-particle dependencies often span long distances, with the particle only appearing at the end of the clause. This means that it may be necessary to process a large amount of intervening sentence material before the full verb of the sentence is known. To facilitate processing, previous studies have shown that readers can preactivate the lexical information of neighbouring upcoming words, but less is known about whether such preactivation can be sustained over longer distances. We asked the question, do readers preactivate lexical information about long-distance verb particles? In one self-paced reading and one eye tracking experiment, we delayed the appearance of an obligatory verb particle that varied only in the predictability of its lexical identity. We additionally manipulated the length of the delay in order to test two contrasting accounts of dependency processing: that increased distance between dependent elements may sharpen expectation of the distant word and facilitate its processing (an antilocality effect), or that it may slow processing via temporal activation decay (a locality effect). We isolated decay by delaying the particle with a neutral noun modifier containing no information about the identity of the upcoming particle, and no known sources of interference or working memory load. Under the assumption that readers would preactivate the lexical representations of plausible verb particles, we hypothesised that a smaller number of plausible particles would lead to stronger preactivation of each particle, and thus higher predictability of the target. This in turn should have made predictable target particles more resistant to the effects of decay than less predictable target particles. The eye tracking experiment provided evidence that higher predictability did facilitate reading times, but found evidence against any effect of decay or its interaction with predictability. The self-paced reading study provided evidence against any effect of predictability or temporal decay, or their interaction. In sum, we provide evidence from eye movements that readers preactivate long-distance lexical content and that adding neutral sentence information does not induce detectable decay of this activation. The findings are consistent with accounts suggesting that delaying dependency resolution may only affect processing if the intervening information either confirms expectations or adds to working memory load, and that temporal activation decay alone may not be a major predictor of processing time.
Process-oriented theories of cognition must be evaluated against time-ordered observations. Here we present a representative example for data assimilation of the SWIFT model, a dynamical model of the control of fixation positions and fixation durations during natural reading of single sentences. First, we develop and test an approximate likelihood function of the model, which is a combination of a spatial, pseudo-marginal likelihood and a temporal likelihood obtained by probability density approximation Second, we implement a Bayesian approach to parameter inference using an adaptive Markov chain Monte Carlo procedure. Our results indicate that model parameters can be estimated reliably for individual subjects. We conclude that approximative Bayesian inference represents a considerable step forward for computational models of eye-movement control, where modeling of individual data on the basis of process-based dynamic models has not been possible so far.
Reading requires the assembly of cognitive processes across a wide spectrum from low-level visual perception to high-level discourse comprehension. One approach of unravelling the dynamics associated with these processes is to determine how eye movements are influenced by the characteristics of the text, in particular which features of the words within the perceptual span maximise the information intake due to foveal, spillover, parafoveal, and predictive processing. One way to test the generalisability of current proposals of such distributed processing is to examine them across different languages. For Turkish, an agglutinative language with a shallow orthography-phonology mapping, we replicate the well-known canonical main effects of frequency and predictability of the fixated word as well as effects of incoming saccade amplitude and fixation location within the word on single-fixation durations with data from 35 adults reading 120 nine-word sentences. Evidence for previously reported effects of the characteristics of neighbouring words and interactions was mixed. There was no evidence for the expected Turkish-specific morphological effect of the number of inflectional suffixes on single-fixation durations. To control for word-selection bias associated with single-fixation durations, we also tested effects on word skipping, single-fixation, and multiple-fixation cases with a base-line category logit model, assuming an increase of difficulty for an increase in the number of fixations. With this model, significant effects of word characteristics and number of inflectional suffixes of foveal word on probabilities of the number of fixations were observed, while the effects of the characteristics of neighbouring words and interactions were mixed.