Acquisition of Reading Skills (1986)

Cultural Constraints and Cognitive Universals

Barbara R. Foorman, Alexander W. Siegel

• Publisher: Routledge
• Year: 2017
• ISBN:
  9781351385763
• Edition: 1
• Topic: Education
• Subtopic: Education General

Originally published in 1986.

In literate societies around the world, children begin instruction in reading somewhere between the ages of five and seven years. On one level their tasks are very similar – learn the sound-symbol relationship of their languages and apply their linguistic and cognitive skills to gain meaning from print. On another level their tasks seem to vary – orthographies and sound-symbol relationships differ, and cultures' attitude towards reading and children's motivation to achieve range widely.

This book considers both universal and culturally constrained aspects of the process of learning to read, with the first four chapters exemplifying cognitive universal approaches to reading, and the last four highlighting cultural constraints. It will be of use to researchers and students, as well as teachers requiring an insight into how reading skills are acquired.

1 Cognitive and Linguistic Components of Reading Ability

Charles A. Perfetti
University of Pittsburgh

Whether there are cognitive universals in learning to read will doubtless prove a difficult question. However, a search for generalizations about reading processes may suggest some constraints on the form of such universals or at least may suggest what kinds of processes are candidates for universals.

There are two general questions to pose: (1) What is the nature of ability differences in reading? (2) How does a child become a skilled reader? Within each of these questions, specific possibilities concerning the role of linguistic and cognitive components will be raised. Along the way, I will draw not only on my own research, but on the advances provided in general by cognitive science. Although each conclusion will be based predominately on work with English-speaking populations, I offer all conclusions in the belief that they may, with appropriate qualification, be general.

INDIVIDUAL DIFFERENCES IN READING ABILITY

Ability differences in reading are pervasive in some nonobvious ways. We have recently begun studying two types of college-age individuals who have problems reading. One could be called a compensated adult dyslexic (CAD). He and she (we have one of each so far) have managed to hide their difficulties in reading so successfully that not only are they getting by in college, they are getting by rather well. One of the CADs is an engineer, and the other is a student in liberal arts. The first fits the high-spatial, “right-brain” syndrome fairly well—good space, good math, bad verbal. The second is more dramatic. She reads fluently under some conditions and has a large vocabulary. We think the vocabulary, along with the heavy use of context and visual pattern recognition, is part of how she gets by. Both CADs seem to be either average or well above average in everything except reading. Eventually, we may know something about the cognitive and linguistic components of their reading ability and how they compensate for their “dyslexia.” However, one thing is clear even without detailed analysis. Such individuals are scarce. I doubt that among a college population they are as frequent as one in a thousand.

Far more common is the second type individual we are studying. This is the college student low in reading skill (LRS). Unlike the uncommon CAD, LRS individuals have more pervasive problems. They are not only low in general intelligence because their reading skill lags behind their nonverbal skills, but they also are more likely than the CADs to have smaller vocabulary and to have trouble in comprehension even of spoken language. They show little evidence of compensation. Rather, they struggle with texts, as evidenced by slow reading and marginal comprehension.

One of the conclusions to come out of this work, and from much other research in reading and dyslexia, is that it is more common to find a strong association among verbal abilities than it is to find a disassociation. Verbal things, including reading, go together. The exception, such as the CAD, is interesting because it is an exception. This is the first generalization to propose: Reading ability depends to a considerable extent on general verbal abilities.

Cognitive and Linguistic Factors in LRS

What are the factors that contribute to a reading problem serious enough to be apparent but not serious enough to keep a person from going to college? The answer to this question is not completely clear. For one thing, we define the problem by reference to percentiles or other norm-referenced markers. This ensures there will be some LRS individuals by definition. For another thing, the measurements we (and cognitive studies generally) have taken have been a restricted set of tasks. In our case, we examined simple tasks of letter matching, vocalization latency, memory span, auditory same-different judgments, and a few others. These were similar to the tasks used by Jackson and McClelland (1979), Hunt (1978), and others. It is possible to discover that a certain factor is important in a skill only if experimental tasks are designed to tap it. In the case of adult reading ability, cognitive studies have been good at providing detailed analysis of reading tasks and related appropriate tasks to test possible low-level components of the task. At the same time, we have been remiss at providing analysis of language tasks, including spoken language comprehension, and there has been little opportunity to observe general language processing components that may affect both reading and other language processes. This is one of the needs of future research. Nevertheless, with proper caveats, there are some conclusions to draw.

The Jackson and McClelland Studies. Jackson and McClelland (1979) used a measure of effective reading speed to reflect both comprehension accuracy and speed of reading. As defined by effective reading speed, high ability readers were in the top quartile, and low ability readers were in the bottom quartile of the population of UCSD college freshmen and sophomores. These students were given several information processing tasks designed to provide information about which basic processes were contributing to ability differences. Several conclusions from their experiments stand out: (1) ability differences were not associated with recognition of single letters (as measured by recognition thresholds); (2) ability differences were associated with performance on matching tasks, including the time to make a matching decision on pairs of letters (A a), pairs of synonyms, and pairs of nonlinguistic patterns; (3) ability differences were associated with listening comprehension performance, measured on texts comparable to those for the text reading task that defined the ability groups.

Results of the various tasks, including some not mentioned here, were analyzed in regression models. Jackson and McClelland concluded that there are two primary ability factors, one a general language comprehension factor that is modality-free. The other is a name retrieval factor that operates when a symbol name is retrieved from memory in response to a visual input, regardless of whether the input is linguistic. The last qualification accommodates a subsequent finding by Jackson (1980) that speed of matching, even of nonsense drawings arbitrarily associated with names, predicted ability. The cognitive processes implicated here are (1) general language comprehension ability and (2) simple symbol activation from visual input.

The Hammond Study. A recent study by Hammond (1984) seems to modify the conclusion by Jackson and McClelland. It also demonstrates how the range of tasks used necessarily limits the range of processing generalizations that can be made. Unlike Jackson and McClelland, Hammond included a task of vocalization latency and an auditory same-different task. The vocalization latency task was used because, in our earlier studies of reading ability, it had consistently proved to be the best predictor of children’s reading comprehension ability. It is simply a measure of the elapsed time between the presentation of a visual stimulus and the onset of the vocalization that produces its name. A particularly important theoretical fact in explaining children’s reading ability is a finding of Perfetti, Finger, and Hogaboam (1978). They found that the reading ability of third grade subjects was predicted by their vocalization latencies to printed words but not to colors or pictures. (Vocalization latencies to digits were only weakly associated with reading skill.) Clearly, the conclusion is that linguistic inputs were critical to children’s reading ability. Naming time for nonalphabetic visual stimuli was unrelated to reading ability.

This conclusion has had two points of contrast. One is with the Jackson and McClelland result previously described. Children’s reading ability is associated with linguistic processes but adults’ reading ability is not. Second is its contrast with the “naming deficit” observed for dyslexic children (Denckla & Rudel, 1976). This deficit includes slower than normal naming time for colors, numbers, and pictures, as well as words. This is in clear contrast to what we found in the Perfetti et al. (1978) study.

Hammond’s (1984) experiment may shed some new light on these contrasts. She found that college subjects’ word vocalization latencies were associated with their reading ability. She also found that digit naming latency was connected with reading ability. Thus, like young dyslexics, the college subjects of Hammond’s study gave evidence of a generalized naming “deficit.” This finding is compatible with the suggestion that the speed of word identification increases with age or reading experience, but within limits set by general nonlinguistic memory retrieval. The latter process is assumed to be the process by which the memory representation of any nameable (symbolic) input is retrieved. Thus, for low ability children without neurological problems, word retrieval speed has not reached the potential set by general symbol retrieval speed. For dyslexics, who may have a severe problem with symbol retrieval, and for normal adults, whose word retrieval speed has reached the potential set by symbol retrieval speed, it is the general symbol retrieval process that limits the rate of word identification. (This hypothesis is further illustrated in Perfetti [1983].) In addition, Hammond found evidence for a true decoding factor in the ability differences of this population. This was seen in latencies to pseudowords, which in a multiple regression analysis predicted reading ability independently of the other general name retrieval variables (e.g., digit naming and word reading). Thus, there may be both a general symbol retrieval factor and a true decoding factor among this population.

A second addition by Hammond (1984) to previously used tasks is also interesting. Subjects were asked to classify as “same” or “different” digitized recordings of spoken syllables, e.g., /di/ - /di/ (”same") and /di/ - /ti/ (”different”). On some trials the two syllables were spoken by the same voice. On other trials, one syllable was spoken by a female voice and one syllable was spoken by a male voice. The task was always to respond “same” or “different” on the basis of the syllable identity, not the voice. Thus, this task was a kind of acoustic analog to the visual letter match task. Hammond found that the voice made a difference, just as in the visual task letters in the same case (AA) were matched more quickly than letters in different cases (Aa). It took more time to recognize two syllables as the same when they were spoken by different voices, one male and one female, than when they were spoken by the same voice. Further, Hammond found that high ability readers were faster than low ability readers on this acoustic task. It is possible that skilled readers show some generalized linguistic abilities, beyond visual word identification, that are for the less skilled reader.

These data call into question the conclusion that low skill among college readers has two independent sources, one in general comprehension and the other in nonlinguistic symbol retrieval. Instead, they suggest a single modality-independent factor, associated with abstract symbol retrieval. And, because of the pseudoword differences, they also suggest a factor associated with true decoding.

Finally, a recent study by Palmer, MacLeod, and Hunt (1985) may further modify the earlier interpretation of Jackson and McClelland concerning ability differences. The pattern of results Palmer and his colleagues obtained, using the kinds of tasks and regression procedures common to the McClelland and Jackson and Hammond studies, suggest separate reading abilities in speed and comprehension. Rapid name retrieval, they conclude, is related to speed of reading but not to comprehension. The evidence again suggests that the comprehension factor is a general one, subsuming both listening comprehension and reading comprehension.

At this point, conclusions concerning comprehension and speed factors in adult reading ability can be only tentative. Quite a bit hinges crucially on which tasks are selected and how comprehension and reading rate are measured, as well as the comparability of populations. Nonetheless, there are generalizations that have survived all three studies previously discussed and additional studies that are currently underway in Pittsburgh. The main conclusion is that two general cognitive factors account for differences in reading ability among college students: Less able readers are slower to access a symbol representation in memory, and they are less able to understand spoken language. Both of these conclusions are in much need of further analysis. For example, it is not clear whether the symbol activation factor is general or specific to print, although, on balance, the general factor hypothesis may be more nearly correct. And we know very little about how to decompose the general comprehension factor that includes both reading and listening. Nevertheless, the generalizations are secure and provide part of the basis for a general theory of ability differences in adult reading.

Reading Ability or Verbal Intelligence?

One question that arises is whether the kind of ability under discussion should be considered reading ability or verbal intelligence. This question is especially important for identifying a distinct category of dyslexia in which reading disability might be an isolated problem, unconnected to other functioning. My earlier reference to a project on college students with reading problems was intended, in part, to emphasize the infrequency of problems unique to reading among college populations. The typical college student who does not read well is likely to have a full complement of weaknesses in verbal knowledge, including syntax, vocabulary, spelling, and composition. Accordingly, it probably does not matter too much whether we refer to differences in reading ability or to differences in verbal intelligence. It is reasonable to assume that, at this general level of skill, the verbal abilities that matter for are the same ones that matter for verbal intelligence.

This conclusion is consistent with a re-analysis of the Jackson and McClelland data carried out by Carroll (1980). He showed that an information processing measure of symbol memory retrieval (the letter comparison task of Posner and Mitchell [1967]) correlated almost as highly with subjects’ verbal SAT scores as with their effective reading speed scores. Hunt and his colleagues have shown that performance on this letter matching task is related to measures of verbal ability, measured by SAT-like tests (Hunt, 1978). Indeed, they have emphasized the same kind of symbol retrieval process that studies of reading have implicated. Thus, at the college level reading ability and verbal ability may be approximately the same thing.

There is an important caveat to making a generalization based on these findings: There is no reason to think that they apply to noncollege adults. The reason that verbal ability and reading ability are the same thing for college students is that their academic experience allows their reading to function up to the general limits that their language ability permits. This may not hold for noncollege adults. For example, it seems not to hold for less well educated persons for whom reading lags behind spoken language comprehension (Sticht, 1979). On the other hand, it is likely that many people who show marginal literacy following normal opportunities for learning will also show marginal general verbal abilities. An interesting question, not being addressed to my knowledge, is whether such adults can continue to improve their verbal skills in the absence of reading. It is certainly possible that exposure to literate speech and demands for ve...

Table of contents

• Cover
• Half Title
• Title Page
• Copyright Page
• Table of Contents
• Preface
• Introduction
• 1 Cognitive and Linguistic Components of Reading Ability
• 2 Some Reflections on Verbal Efficiency Theory
• 3 Phonological Skills and Learning to Read and Write
• 4 Phonological Skills and Learning to Read and Write: Reactions and Implications
• 5 How do Japanese Children Learn to Read?: Orthographic and Eco-cultural Variables
• 6 Non-Alphabetic Codes in Learning to Read: The Case of the Japanese
• 7 Beginning Reading in Chinese and English
• 8 The Reading Achievement Game: Cognitive Universals and Cultural Constraints
• Author Index
• Subject Index