Theoretical Issues in Reading Comprehension

Perspectives from Cognitive Psychology, Linguistics, Artificial Intelligence and Education

Rand J. Spiro, Bertram C. Bruce, William F. Brewer, Rand J. Spiro, Bertram C. Bruce, William F. Brewer

• Publisher: Routledge
• Year: 2017
• ISBN:
  9781351607230
• Edition: 1
• Topic: Psychology
• Subtopic: Cognitive Psychology & Cognition
• Index: Psychology

Research in cognitive psychology, linguistics, and artificial intelligence – the three disciplines that have the most direct application to an understanding of the mental processes in reading – is presented in this multilevel work, originally published in 1980, that attempts to provide a systematic and scientific basis for understanding and building a comprehensive theory of reading comprehension. The major focus is on understanding the processes involved in the comprehension of written text. Underlying most of the contributions is the assumption that skilled reading comprehension requires a coordination of text with context in a way that goes far beyond simply chaining together the meanings of a string of decoded words. The topics discussed are divided into five general areas: Global Issues; Text Structure; Language, Knowledge of the World, and Inference; Effects of Prior Language Experience; and Comprehension Strategies and Facilitators, and represent a broad base of methodology and data that should be of interest not only to those concerned with the reading process, but also to basic science researchers in psychology, linguistics, artificial intelligence, and related disciplines.

I   GLOBAL ISSUES

The chapters in this book cover a wide range of topics from diverse perspectives. Nevertheless, several themes run throughout, indicating some consensus about the major questions. This section contains three chapters that each focus on different aspects of the broad themes. Later chapters will return to these themes as they pertain to specific areas of inquiry.

The authors in this section argue that any adequate model of reading comprehension will possess three essential characteristics. It will be multilevel, interactive, and hypothesis-based. Multilevel implies that knowledge structures at several different levels are actively used in the reading process. Traditionally proposed levels include orthographic, phonological, lexical, syntactiç, and semantic. Clearly, however, higher level knowledge sources such as inference rules and expectations about story structure are also crucial components of the skilled reading process.

To say that the model needs to be interactive means that, although the knowledge sources or levels seem to form a “natural” hierarchy running from orthographic knowledge to expectations about discourse structure, communication between these levels is not limited to adjacent members of the hierarchy. Thus, the knowledge sources interact in a heterarchical fashion. The general model proposed by some psychologists (e.g., Gough, 1972; LaBerge & Samuels, 1974), which involves a visual input progressing linearly through the various knowledge levels to arrive finally at a “meaning,” is not supported here. Instead, it appears that each knowledge source can contribute input at various points in the complex process of comprehending text. Comprehension proceeds from the top down as well as from the bottom up. Comprehension is “driven” by preexisting concepts as well as by the “data” from the text (Bobrow & Norman, 1975).

The coordination of these multiple contributions requires a mechanism for collecting evidence for various interpretations of the text. This is the hypothesis-based aspect of reading. Three characteristics of these hypotheses are relevant here. First, these hypotheses may be tacit. Second, a hypothesis represents a possible interpretation that may later either be continued or rejected. Third, part of the structure of a hypothesis is the specification of those pieces of evidence that would support or contradict it.

Several existing reading theories share significant properties with the general form described here. For example, Goodman (1973) describes receptive language processes in general as hypothesis-based, defining them as “cycles of sampling, predicting, testing, and confirming.” Productive reading is seen as requiring strategies that facilitate the selection of the most useful cues. Smith (1971) also emphasizes the contribution of what he terms “nonvisual” information to reading. This nonvisual knowledge includes what people already know about reading, language, and the world in general. He argues particularly that reading is not decoding to sound, but rather that semantic and other nonvisual processes intercede between visual processes and reading aloud. Perfetti (1975) proposes at least three levels of sentence processing that obviously require corresponding levels of knowledge. He also focuses more explicitly on how the various component processes might interact, basing his overall conclusions on the fact that all the processes that occur during reading comprehension must share a “limited capacity processor.”

Though the approach here shares much with that of these and other investigators, there are also some differences in emphasis. The authors are more explicit in the designation of different levels of knowledge sources, particularly in the areas Goodman terms “semantic.” Examples of types of knowledge recognized are word semantics, logical inference rules, social action patterns, story schemata, and strategic knowledge about how to use the various knowledge sources. In addition, the explicit definition of the interaction between knowledge components is considered to be of the utmost importance. This approach suggests the possibility that some unskilled reading may be the result of not knowing how to use and interweave knowledge, rather than of a lack of knowledge itself.

Another emphasis in these chapters is on the dynamic nature of the reading process. These authors propose that a reader’s working hypothesis may be wrong and that at various points during the reading process it may be in a state of limbo, only partially specified, needing more evidence. They consider as well the possibility that as a consequence of some of the intermediate stages, the reader must “back up” and rehypothesize about the meaning of a text.

A final theme in these chapters is the emphasis on structure-building. These structures or schemata are important for both the final representation of the meaning of the text and the intermediate hypotheses that are so crucial to attaining the final goal. Three classes of knowledge are necessary for building such structures. First of all, a reader must have sufficient information about the types of schemata that are possible at each level, how to recognize them, and what implications they have for further processing. Second, the reader must have strategic knowledge—information on how to use the structural knowledge, what priorities to use in evaluating hypotheses, and what form the final “understood” structure should take. Third, there is knowledge about the way the purpose of reading a particular text relates to the structural and strategic knowledge used.

The global features we have outlined are addressed with differing emphasis in each of the three chapters in this section. Adams focuses on the levels of analysis involved in reading comprehension and their interaction. Rumelhart discusses characteristics of schemata and their role in hypothesis formation and testing. Woods also concerns himself with hypothesis-related processes, as well as drawing implications for reading comprehension from research on computer simulation of natural language processing.

ACKNOWLEDGMENT

Andee Rubin contributed to the writing of this section’s introduction.

REFERENCES

Bobrow, D. G., & Norman, D. A. Some principles of memory schemata. In D. G. Bobrow& A. M. Collins (Eds.), Representation and understanding: Studies in cognitive science. New York: Academic Press, 1975.

Goodman, K. S. Psycholinguistic universals in the reading process. In F. Smith (Ed.), Psycholinguistics and reading. New York: Holt, Rinehart & Winston, 1973.

Gough, P. B. One second of reading. In J. F. Kavanagh & I. G. Mattingly (Eds.), Language by ear and by eye. Cambridge, Mass.: MIT Press, 1972.

LaBerge, D., & Samuels, S. J. Toward a theory of automatic information processing in reading. Cognitive Psychology, 1974, 6, 293–323.

Perfetti, C. A. Language comprehension and fast decoding: Some psycholinguistic prerequisites for skilled reading comprehension. Paper presented to The Development of Reading Comprehension Seminar of the International Reading Association. Newark, Del., July 1975.

Smith, F. Understanding reading. New York: Holt, Rinehart & Winston, 1971.

1 Failures to Comprehend and Levels of Processing in Reading

Marilyn Jager Adams

Center for the Study of Reading,
Bolt Beranek and Newman Inc.

Reading deficiency is one of the most significant problems facing educators today. By recent estimates, as many as 40% of the school-age children in the U.S. may be handicapped by reading difficulties (Goldberg & Schiffman, 1972). The significance of the problem, however, is only partially reflected by such statistics, as reading difficulties may result in poor performance in other educational activities. Reading is one of the basic ways of acquiring information in our society and in academic settings in particular. The individual who cannot read well is at a serious disadvantage with respect to educational and, consequently, vocational opportunities.

Why so many children have trouble learning to read is not well understood. In some cases, mental or physical disabilities can be cited as the underlying cause. But more often, reading problems have not been clearly associated with diagnosable mental or physical deficits. This has led to the definition of clinical syndromes such as dyslexia and minimal brain dysfunction that acknowledge and label the problem but do not explain it.

A basic assumption of this chapter is that skilled reading depends on a multiplicity of perceptual, linguistic, and cognitive processes and that, for many children, reading difficulties reflect the inadequate development of one or more of these processes. The purpose of this chapter is to consider some of the processes that may be especially problematic for the young reader. The chapter begins with an overview in which skilled reading is described as the product of both analytic and synthetic, or bottom-up and top-down activities. Following this overview, potential sources of difficulties are discussed under three general topics: word recognition, syntactic processing, and semantic processing.

OVERVIEW

For the skilled reader, the processes involved in reading are so well learned and integrated that written information can flow almost automatically from sensation to meaning. As the letters of the text are identified, they simultaneously prime or set up expectations about the identities of the words to which they belong. As the words are identified, they prime the most probable syntactic and semantic structures. More generally, since the end products of each level of analysis are the elements for some other level, the information is naturally propagated upward through the system, through increasingly comprehensive levels of analysis. This is known as bottom-up processing. While all of this is happening, the partially activated candidates at each level are competing for completion; as they do so, they reciprocally prime or facilitate the processing of their missing elements. This is known as top-down processing. For the skilled reader, top-down and bottom-up processing are occurring at all levels of analysis simultaneously as she or he proceeds through the text. The reader is therefore able to make optimal use of the information on the page, the redundancy of the language, and the contextual environment with minimal effort. The top-down processes ensure that lower order information that is consistent with the reader’s expectations will be easily assimilated, as it will already have been partially processed. Meanwhile, the bottom-up processes ensure that the reader will be alerted to any information that is novel or that does not fit her or his ongoing hypotheses about the content of the text. (For a more thorough description of these processes, see Rumelhart’s Chapter 2 in this book, or Adams & Collins, 1979.)

The efficient operation of such a system depends as much on the information in the reader’s mind as on the information in the written text. If the reader is lacking any critical skill or piece of knowledge, the flow of information through the system will be obstructed. In these cases, the reader must find a way to compensate. One option is to direct extra processing energy to the difficulty until it is resolved; for example, the reader may pause and articulate a difficult word. Alternatively, she or he may rely on top-down processes to evade the problem; for example, she or he may use contextual information to infer the meaning of an unfamiliar word. Both of these solutions are normal and adaptive and are regularly used by skilled readers. Thus, one kind of difficulty that we might expect of beginning readers is they might fail to adopt either of these strategies. However, equally serious problems might arise if they adopt either of these strategies to the extreme.

The danger of relying too heavily on top-down processing is obvious. The proper balance between the information that the reader should bring to the text and that which the text should bring to the reader will be lost. To the extent that guesses are based on prior guesses, the individual is not really reading in any useful way. Yet, as is discussed in the sections to follow, some of the most basic aspects of reading may also be the most foreign for the beginner. By contrast, the beginning reader already has a wealth of real-world knowledge, and in terms of content, her or his required reading materials are probably quite simple. Thus, she or he may find, for example, that the identity of an unfamiliar word can often be guessed as accurately and more easily than it can be sounded out or that the conceptual relationships implied by a sentence can be inferred without attending to its syntactic structure. It should not be surprising to find young readers who have learned to depend on such top-down strategies.

In the long run, the alternative strategy of focusing attention on the difficulty may be more adaptive. At least it provides an opportunity for learning. The danger in using this strategy is that comprehension may consequently suffer. The problem is that the human mind is a limited-capacity processor. As LaBerge and Samuels (1974) have pointed out, the reader can selectively direct attention to any particular subprocess but only by taking it away from deeper levels of analysis. In G. Stanley Hall’s words (1911), true reading only occurs “when the art has become so secondarily automatic that it can be forgotten and attention be given solely to the subject matter. Its assimilation is true reading and all else is only the whir of the machinery and not the work it does [p. 134].”

The problem of limited processing capacity is especially critical for the young reader. First, many of the necessary subskills are not well learned and, therefore, demand considerable attention. Second, the functional memory capacity of the young child tends to be less than that of the adult. It is not entirely clear why this is so: Some have argued that the span itself increases with age (e.g., Farnham-Diggory, 1972); some have attributed it to young children’s failure to “chunk” or organize the material for efficient storage (e.g., Flavell, 1970; Olson, 1973; Simon, 1974); still others have argued that it only reflects the differential effort that children must invest in the encoding of to-be-remembered items (e.g., Huttenlocher & Burke, 1976). Regardless of which explanation is correct, the important implication for the pre...