The present project aims to bridge the gap between different research traditions, bringing together experts on visual word recognition, reading development, oculomotor control, spatial attention, and computational modeling, with the overarching goal of developing a new model of basic processes in reading that builds on knowledge gained from prior research on single word recognition and prior research on eye guidance during reading.
The main empirical objective of this project was to develop and test a new paradigm for studying the spatial integration of orthographic information during reading - the Flanking Letters Lexical Decision task. The theoretical objective was to develop and test a new computational model of orthographic processing and multi-word reading.
We have used the Flanking Letters Lexical Decision (FLLD) task in several series of experiments.
We have performed sentence reading experiments using the «boundary technique« to modify the stimulus at position N+1 when the reader's eyes are looking at the word at position N, in order to investigate orthographic parafoveal-on-foveal influences during reading.
We have also developed and tested a novel Rapid Parallel Visual Presentation (RPVP) paradigm, whereby sequences of words are presented simultaneously for a very brief duration (150 ms) and participants are asked to report the identity of one word at a post-cued location.
With J. Snell (PhD student working in collaboration with M. Meeter, Amsterdam) we have completed several experiments using the FLLD paradigm, as well as sentence reading experiments investigation orthographic parafoveal-on-foveal effects, that have revealed that the integration of orthographic information is limited to the sublexical level (Snell, Vitu, & Grainger, 2017), and that parallel processing across several words continues beyond the lexical level (syntactic and semantic effects) but without being integrated in to a single channel (Snell, Meeter, & Grainger, 2017). We have also run experiments with the novel RPVP paradigm and have shown that the grammatical structure of word sequences influences single word identification in this paradigm - a «sentence superiority effect«.
With J. Mirault (PhD student funded by the ANR) we have run several experiments to test specific predictions of our computational model, and specifically with respect to the role of inter-word spacing during sentence reading. This work will be shortly submitted for publication.
A first version of the computational model has been implemented with J. Snell and M. Meeter, and an article describing the model is about to be submitted.
One outstanding result is the novel «sentence superiority effect« observed with the RPVP paradigm (work submitted for publication).
The developmental research in collaboration with B. Lété (Lyon 2) will begin this June, and we will pursue the experimentation with adult participants and the computational modelling.
Grainger, J. (2017). Orthographic processing: A “mid-level” vision of reading. Quarterly Journal of Experimental Psychology, in press.
Snell, J., Meeter, M., & Grainger, J. (2017). Evidence for simultaneous syntactic processing of multiple words during reading. PLoS ONE.
Snell, J., Vitu, F., & Grainger, J. (2017). Integration of parafoveal information during foveal word reading: Beyond the sublexical level? Quarterly Journal of Experimental Psychology, in press.
Years of research has led to the development of sophisticated accounts of the basic processes involved in reading comprehension, and particularly with respect to orthographic processing of single words. Equally intensive research, performed largely in parallel with the aforementioned efforts, has given rise to detailed proposals describing the mechanisms involved in controlling eye movements during reading. The present project aims to bridge the gap between these two research traditions, bringing together experts on visual word recognition, reading development, oculomotor control, spatial attention, and computational modeling, with the overarching goal of developing a new model of basic processes in reading that builds on knowledge gained from prior research on single word recognition and prior research on eye guidance during reading.
The specific aims of the project are threefold. First we aim to develop and test a recently introduced paradigm for reading research, the flanking letters lexical decision task, a paradigm that holds much promise as a tool for investigating the spatial integration of orthographic information across words during reading. This novel technique represents a significant paradigm shift for reading research, which has been dominated until now by research investigating the temporal integration of information during single word recognition, or research investigating higher-level integration processes operating at the sentence-level. The new paradigm will be used to increase our understanding of how orthographic information can be processed in parallel across several words, and therefore how adjacent words can influence processing of the word that is fixated. The results obtained with this task will be compared with experiments investigating the same issues, but using eye movement recordings during sentence reading.
Second, we will apply the new paradigm in a large-scale investigation of how spatial integration of orthographic information develops during learning to read. Here we propose to apply the FLLD paradigm to investigate the development of parallel orthographic processing and the spatial integration of orthographic information thought to accompany such parallel processing. This will provide the first investigation of the transition from a strictly sequential “one word at a time” approach to reading, thought to be typical of beginning readers, to a more efficient reading strategy that involves extracting orthographic information from several words at the same time. By testing children at each grade of primary education, we can plot the developmental trajectory of spatial integration processes.
Third, we aim to develop and test a new model of multi-word reading, which we call “O-code Reader”. This model describes how orthographic information, extracted from several words in parallel, is used to correctly identify the word that is fixated, while at the same time enabling cross-talk between neighboring words. In a first step, a static version of this model, the “Bag-of-Bigrams” (BoB) model, will be tested using the results obtained with the flanking letters lexical decision task. In a second step, we will develop a dynamic version of the model that incorporates mechanisms for eye-movement control during reading.
We expect that the results of the proposed research will establish a new theoretical foundation for reading research, solidly anchored in a detailed description of the cooperative interactions between two of the most elementary aspects of reading behavior, that is, orthographic processing and eye-movement control.
Monsieur Jonathan Grainger (Centre National de la Recherche Scientifique Délégation Provence et Corse Laboratoire de Psychologie Cognitive)
The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.
EMC Laboratoire d'Etudes des Mécanismes Cognitifs
CNRS DR12 UMR7290 Centre National de la Recherche Scientifique Délégation Provence et Corse Laboratoire de Psychologie Cognitive
Help of the ANR 260,000 euros
Beginning and duration of the scientific project: September 2015 - 36 Months