Reading and Language Processing

John M. Henderson, Murray Singer, Fernanda Ferreira, John M. Henderson, Murray Singer, Fernanda Ferreira

• Publisher: Psychology Press
• Year: 2013
• Print ISBN: 9780805819038
• eBook ISBN: 9781317759553
• Topic: Personal Development
• Subtopic: History & Theory in Psychology

This volume was designed to identify the current limits of progress in the psychology of reading and language processing in an information processing framework. Leaders in their fields of interest, the chapter authors couple current theoretical analyses with new, formally presented experiments. The research -- cutting-edge and sometimes controversial -- reflects the prevailing analysis that language comprehension results in numerous levels of representation, including surface features, lexical properties, linguistic structures, and idea networks underlying a message as well as the situations to which a message refers. As a group, the chapters highlight the impact that input modality -- auditory or written -- has on comprehension. Finally, the studies also capture the evolution of new topic matter and ongoing debates concerning the competing paradigms, global proposals, and methods that form the foundation of the enterprise.

The book presents current accounts of research on word-, sentence-, and text-processing. It will prove informative for experimental psychologists as well as investigators in cognitive science disciplines such as computer science, linguistics, and educational psychology. The book will also be very helpful to graduate students who wish to develop expertise in the psychology of language processes. For them, it collects, in a single volume, readings that are representative of progress concerning many central problems in the field. As such, it is distinct from the numerous collected volumes that concentrate on a single issue. Complete author and subject indexes facilitate effective use of the volume.

1

Reading and Language Processing: Paradigms, Proposals, and Procedures

Murray Singer   University of Manitoba

John M. Henderson   Michigan State University

Fernanda Ferreira   Michigan State University

This volume is devoted to reading and language processing, an area that has been central to the study of human cognition since the inception of modern cognitive psychology. For example, in his classic book, Cognitive Psychology, Neisser (1967) devoted 4 of 11 chapters to topics in reading and language. These chapters were Words as Visual Patterns, Speech Perception, Active Verbal Memory, and Sentences. In addition, most of the other chapters in Neisser’s book included discussions of related topics and presented experiments in which reading and/or language played a major role.

In the 1990s, many of the topics discussed by Neisser continue to draw the attention of researchers, and new ones have been assigned high research priority. The early maturation of this field is characterized both by the evolution of new topic matter and by serious debates concerning the competing paradigms, global proposals, and methods that form the foundation of the enterprise. Our attempt to identify a representative set of contributors who are conducting research on problems that are currently cutting-edge or controversial in reading and language processing has dovetailed with our desire to highlight the latter debates. We will briefly identify these features, and link them to some of the present contributions.

Competition among paradigms in the study of language processes has resulted from the emergence of connectionist modeling (Rumelhart & McClelland, 1986) as a serious competitor to the antecedent symbol processing systems in cognition (Newell & Simon, 1972). The tension between these paradigms is evident in the field of language processing, but the advantages of the two approaches may also be fruitfully merged. In the present volume, for example, Just and Carpenter apply their Capacity Constrained READER model to their pupillometric data reflecting the fluctuation of effort during reading. CC READER is a hybrid model consisting of a symbolic production system and an activation-based connectionist model (see Spivey-Knowlton, Trueswell, & Tanenhaus for a purer connectionist approach). Likewise, the construction-integration model of Kintsch (1988; see Moravcsik & Kintsch, this volume) blends the symbolic construction of propositional networks with the settling, according to connectionist principles, of activation in those networks. The interplay between the two paradigms will likely characterize research in this field in the 1990s and perhaps beyond.

These dominant paradigms form a backdrop for the evaluation of global proposals which, although general to the study of cognition, also take forms specific to the problems of language processing. Theorists must ask whether language processes are modular or interactionist, serial or parallel, and bottom-up or top-down; and whether or not these processes are executed to completion immediately upon the encoding of a spoken or written phrase. The modularity hypothesis of Fodor (1983), for example, suggests that a syntactic module might form an important basis of parsing processes. Clifton concludes that his data favor the existence of a dedicated and informationally-encapsulated syntactic module, whereas Spivey-Knowlton et al. interpret their own data to favor the free interaction of information from syntactic, semantic, and other levels of analysis.

Global proposals such as the modular versus interactionist competitors may be associated with the paradigms discussed earlier. For example, the alternative of unrestricted interaction of information is correlated with the connectionist framework, but this correlation is by no means perfect. Many connectionist models posit different cognitive systems, such as those devoted to lexical, syntactic, and semantic processing. These systems represent a degree of segregation of function, although they are not modules according to Fodor’s (1983) definition. As a result, a connectionist model may exhibit elements of both of the ostensively antagonist modular and interactionist processing analyses.

Because the modern study of language processes is barely a quarter of a century old, the logic and value of relevant research procedures are still being worked out. This emerges as a prominent theme of this volume. In the use of eye fixations to gain insight about language processing, researchers are still considering subtle but important differences in the value of first pass, second pass, and cumulative fixations; as well as of the significance of fixations upon different text regions (current region, previous region, next region) relative to a critical word or phrase. Eye fixation measures are reported by Daneman and Reingold, Ferreira and Henderson, Henderson and Ferreira, Pollatsek, Raney, Lagasse, and Rayner, and Clifton. New implications of these data were highlighted in the two chapters by Ferreira and Henderson. Another incisive technique, the use of pupillometry to monitor on-line fluctuations of resource demands in reading, was explored by Just and Carpenter. Caveats concerning the use of the “moving window” display technique, in which successive sections of text are revealed in response to a subject’s button presses, were offered by Spivey-Knowlton et al.

It would be pointless for language processing researchers to be embarrassed about controversies of method, and unwise for them not to address the controversies. The debates indicate that investigators are conscientiously scrutinizing the alternative methods that are available, and the alternate interpretations of each one.

Paradigms, theories, and methods serve, of course, to expose and organize the subject matter and content of a field. In reading and language processing, a major advance and prevailing analysis recognizes that language comprehension results in numerous levels of representation, including the levels of surface features, lexical properties, linguistic structures, and idea networks (or textbases) underlying a message, as well as the situations to which a message refers. The interplay among these levels is emphasized in the study of Moravcsik and Kintsch. Surface features are particularly highlighted in Levy, Barnes, and Martin’s study of the impact of the repetition of words and syntactic structures on reading fluency. Lexical access was scrutinized by Buchanan and Besner and by Daneman and Reingold. Linguistic and parsing processes constituted a primary focus in the investigations of Clifton, of Ferreira and Henderson, Just and Carpenter, and Spivey-Knowlton et al. The construction of a propositional textbase is addressed in the chapters of Masson and of Singer, and both of those studies also bear on the extraction of a causal situation model from the textbase. The situational level was also emphasized in Dixon, Harrison, and Taylor’s study of the derivation of action plans from procedural texts.

Many other trends and issues can be discerned in the present chapters. One that promises to be of considerable importance in the near future is the growing evidence that individual differences among readers are associated with qualitative differences in their processing patterns. Reader differences in cognitive resources were examined by Just and Carpenter, and differences in readers’ background knowledge were addressed by Moravcsik and Kintsch.

The present chapters may be of most direct concern to experimental psychologists, but we hope that the findings will be of interest to investigators in several of psychology’s companion disciplines in cognitive science. Those chapters that inspect parsing processes may bear on linguistic theories of language structure. The computational models explicit or implicit to these investigations are pertinent to studies of natural language processing in computer science. Reading practitioners may detect important clues about basic reading processes in several of the chapters.

The collection of the chapters in this volume was undertaken in order to produce a special issue of the Canadian Journal of Experimental Psychology. We thank Gordon Winocur, editor of the Journal at the initiation of the project, for inviting us to edit the special issue. With his consultation, we invited the contributors to submit empirical reports of new research on topics central to their overall research programs in reading and language processing. We were very gratified that all of the researchers whom we invited to contribute agreed to do so. The result is 13 articles covering what we think are some of the most important and interesting areas of contemporary cognitive research.

We would like to thank the authors for giving us the honor of presenting their work. Most of the authors also served as reviewers, and we thank them for their time and effort in that regard as well. We would also like to thank Michael Anes, Tom Carr, Vic Ferreira, Albrecht Inhoff, Karen McClure, Paul van den Broek, and one anonymous reviewer for providing insightful comments on the articles. We are grateful to Colin Macleod, current editor of the Canadian Journal of Experimental Psychology, who provided us with guidance during the later stages of the journal phase of the project. Finally, thanks are due to Judi Amsel, who helped us transform the special issue into book form.

References

Fodor, J. A. (1983). The modularity of mind. Cambridge, MA: The M.I.T. Press.

Kintsch, W. (1988). The role of knowledge in discourse comprehension: A construction-integration model. Psychological Review, 95, 163–182.

Neisser, U. (1967). Cognitive psychology. New York: Appleton Century Crofts.

Newell, A., & Simon, H. A. (1972). Human problem solving. Englewood Cliffs, NJ: Prentice Hall.

Rumelhart, D. E., McClelland, J. L., and the PDP Research Group. (1986). Parallel distributed processing: Explorations in the microstructure of cognition (Vol. I). Cambridge, MA: Bradford Books.

2

Reading Aloud: Evidence for the Use of a Whole Word Nonsemantic Pathway

Lori Buchanan and Derek Besner University of Waterloo

Abstract It is widely assumed that the presence of an associative priming effect during the oral reading of orthographies with consistent spelling-sound correspondences signals the use of an orthographic code for lexical access (the addressed routine). Relatedly, the failure to observe such a priming effect has been taken to indicate the use of a routine that relies on subword spelling-sound correspondence knowledge (the assembled routine). This logic depends on the assumptions that (a) only the addressed routine (whole word orthographic knowledge) can produce priming, and (b) that it necessarily does so (i.e., is automatic). The present experiments show that, taken alone, neither the presence nor absence of priming effects in oral reading permit an inference as to whether the addressed or assembled routine is used. Converging operations which do permit such an inference are reported. The data support the view that (i) components of the word recognition system operate interactively such that use of the assembled routine yields priming under certain conditions, and (ii) normal readers of a shallow orthography use a nonsemantic, whole word pathway to name words.

Résumé II est généralement admis que, dans le cas d’orthographes peu profondes (où il y a correspondance systématique entre la graphie et la prononciation), la présence d’un effet d’amorçage associatif durant la lecture orale marque l’utilisation d’un code orthographique pour accéder au lexique (programme adressé). Par ailleurs, on considère que l’absence d’un effet d’amorçage indique l’utilisation d’un programme qui repose sur la connaissance des correspondances entre la graphie et la prononciation à un niveau inférieur à celui du mot (programme assemblé). Cette logique dépend des hypothèses suivantes: a) seul le programme adressé (connaissance orthographique des mots complets) peut produire l’amorçage et b) il le produit nécessairement (c’est-à-dire automatiquement). Les expériences que nous avons menées montrent qu’on ne peut inférer, uniquement de la présence d’effets d’amorçage ou de leur absence durant la lecture orale, que le programme adressé ou bien le programme assemblé est utilisé. Les opérations convergentes qui permettent une telle inférence sont exposées dans le rapport. Les données recueillies montrent que i) les composantes du système de reconnaissance des mots fonctionnent en interaction de sorte que l’utilisation du programme assemblé produit un effet d’amorçage dans certaines conditions et ii) les lecteurs normaux qui utilisent une orthographe peu profonde empruntent une voie d’accès non sémantique aux mots complets pour nommer les mots.

Until recently, research on visual word recognition has focussed on questions concerning how words printed in English are read. This analysis has yielded a remarkable consensus concerning some of the underlying processes (e.g., see reviews by Carr & Pollatsek, 1985; Patterson & Coltheart, 1987; see also Paap, Noel & Johansen, 1992) and has resulted in the dual route model¹. An extension of this work to cross-orthography investigations has resulted in a large body of research. One position was articulated in the strong version of the Orthographic Depth Hypothesis, which held that the consistency of spelling-sound correspondences within an orthography dictates which of the routines of the dual route model are used during reading. Shallow orthographies (i.e., those with consistent spelling-sound correspondences) were argued to be always read aloud prelexically (e.g., Allport, 1979; Bridgeman, 1987; Hung & Tzeng, 1981; Morton & Sasanuma, 1984; Turvey, Feldman, & Lukatela, 1984). This position has since been tempered by the linguistic observation that syllabic stress can not always be derived on the basis of prelexical phonology (e.g., Katz & Frost, 1992). Consequently, some researchers assume that prelexical phonology serves to activate a phonological lexicon which then serves to mediate naming or semantic access (e.g., Carello, Lukatela & Turvey, 1988; Besner & Smith, 1992). Another solution is to assume that, additionally, a whole word orthographic routine which addresses lexical phonology is also functional in shallow orthographies (e.g., Besner, 1987; Besner & Hildebrandt, 1987; Besner & Smith, 1992; Besner, Patterson, Lee & Hildebrandt, 1993; Frost, Katz & Bentin, 1987; Katz & Feldman, 1983; Katz & Frost, 1992; Patterson, 1990; Sebastian-Galles, 1991; Seidenberg, 1985a,b).

While the strong version of the Orthographic Depth Hypothesis can thus be rejected on logical grounds, the methodology that has been used to investigate reading in various orthographies nevertheless bears on some issues which are important in their own right. Several lines of evidence relevant to how different orthographies are read are based on critical but un...

Table of contents

• Cover
• Title Page
• Copyright Page
• Table of Contents
• 1. Reading and Language Processing: Paradigms, Proposals, and Procedures
• 2. Reading Aloud: Evidence for the Use of a Whole Word Nonsemantic Pathway
• 3. What Eye Fixations Tell Us About Phonological Recoding During Reading
• 4. The Use of Information Below Fixation in Reading and in Visual Search
• 5. Eye Movement Control During Reading: Fixation Measures Reflect Foveal but Not Parafoveal Processing Difficulty
• 6. Thematic Roles in Sentence Parsing
• 7. Reading Processes During Syntactic Analysis and Reanalysis
• 8. Context Effects in Syntactic Ambiguity Resolution: Discourse and Semantic Influences in Parsing Reduced Relative Clauses
• 9. The Intensity Dimension of Thought: Pupillometric Indices of Sentence Processing
• 10. Causal Bridging Inferences: Validating Consistent and Inconsistent Sequences
• 11. Writing Quality, Reading Skills, and Domain Knowledge as Factors in Text Comprehension
• 12. Effects of Sentence Form on the Construction of Mental Plans from Procedural Discourse
• 13. Transfer of Fluency Across Repetitions and Across Texts
• 14. Episodically Enhanced Comprehension Fluency
• 15. Reading and Language Processing: Similarities and Differences
• Author Index
• Subject Index