«DOCUMENT RESUME ED 363 732 CE 064 960 AUTHOR Fowler, Anne E.; Scarborough, Hollis S. TITLE Should Reading-Disabled Adults Be Distinguished from Other ...»
Two kinds of diagnostic tools are currently in wide use in adult literacy programs, though neither was designed to detect reading disability, per se. (For reviews, see Kirsch & Guthrie, 1977-78; Sticht, 1988; Venezky, 1992; Venezky, Bristow, & Sabatini, in press). One kind are tests of basic skills, such as the comprehension section of the Test of Adult Basic Skills or the Stanford Diagnostic Reading Tests. These measures provide some index of overall function, potentially useful for initial placement of individuals and for monitoring the effectiveness of programs. They are easy to administer and, because they can be administered in groups (except for non-readers), they are efficient. They have face validity inasmuch as the actual materials being read are of interest/relevance to the reader. Finally, because they do not require reading aloud, dignity is preserved.
TECHNICAL REPORT T193-7 The other measure currently in wide use are tests of functional abilities, including those used in the National Adult Literacy Survey (NALS) and the Young Adult Literacy Survey (YALS). These tests require adults to give short answers in regard to everyday reading materials such as bus schedules or newspaper advertisements and assess numerical abilities as well as reading skill. This type of measure has two main virtues: efficiency and face validity. It has been used in population testing to provide new estimates of adult literacy levels.
Although both kinds of measures provide some initial assessment of overall reading status, and hence the severity of the problem, neither is sufficient to plan for treatment. As discussed earlier, performance on standardized comprehension measures depends upon two kinds of abilities: word recognition and listening comprehension. A single score, such as the eighth-grade level, could be achieved via the combination of high intelligence and very limited word recognition skills, via average intelligence combined with well-developed decoding skills that have not yet become automatic and so overly stress verbal memory, or via well developed word recognition skills combined with limited understanding of the content of the passage. The problem may be further exacerbated by the equation of an adult comprehension score with norms developed on children. As Venezky (1992) suggests, "a sixth grader who reads at a sixth-grade level and an adult who reads at a sixth-grade level usually have widely different reading abilities and require different forms of instruction, yet by grade level measures they are classed as identical" (p. 3). It should also be noted that of all standardized reading measures, those assessing reading comprehension have proven to be most problematic, in large part because of the tremendous influence of background knowledge on reading comprehension. On the other hand, tests of functional abilities are even more subject to confounding problems. In their present form, these measures incorporate word recognition, general comprehension, experience with the particular materials, and even mathematical and problemsolving ability. Venezky notes that the major functional reading test correlates less with standardized measures of reading comprehension (.70) than it does with standardized measures of mathematics. In short, while these tests are very useful for providing a population index of overall literacy, performance on them reflects so many skills that it is impossible to plan treatment wisely on the basis of such a score.
Researchers have generally relied on more analytic diagnostic instruments than those currently used by clinicians. They typically
NATIONAL CENTER ON ADULT LITERACY 69focus less on establishing a person's overall reading level and more on establishing a profile of strengths and weaknesses in component skills to aid in determining why an individual is at a particular level. In assessing reading, therefore, researchers typically distinguish between measures of word recognition, decoding, comprehension of connected text, oral reading fluency and oral language facility (e.g., listening comprehension). When no standardized instruments are available, furthermore, the researcher can develop experimental measures (e.g., for assessing decoding automaticity, or for examining the effects of topic familiarity on comprehension). While such a complicated fractionation of skill assessment is both inefficient and unwarranted for achieving the clinician's goals, the careful delineation of component skill profiles has been useful in gaining an understanding of reading problems from the research, and some of these contributions can be carried over into the practical sphere.
An ideal diagnostic measure battery for adults seeking literacy instruction should provide information about five aspects of functioning, although some of these skills can be inferred from other measures. First, as in current practice, a test of reading comprebension should continue to be used to determine the overall reading level. Overall proficiency is most related to adaptive function and incorporates performance on all other skills.
Many tests suitable for this purpose are available. As noted earlier, however, comprehension scores can be strongly influenced by a person's familiarity with the topic discussed in the text, so the appropriateness of a test's content should be taken into account in selecting a measure of reading comprehension.
Second, a diagnostic measure of reading should include a measure of listening comprebension (or verbal aptitude) to identify any problems with spoken language that may be impeding the adult's current reading performance and that may have to be worked on to facilitate improvement in reading. Verbal facility has several components, but these tend to be highly correlated with each other such that assessing any single aspect can provide a rough index of overall proficiency in spoken language. For example, listening comprehension can be measured by obtaining spoken responses to orally presented materials that are similar to the written passages on corresponding reading comprehension tests (e.g., as on the Durrell scales or the Woodcock-Johnson Psychoeducational Battery). Alternatively, measures of single word vocabulary knowledge (e.g., Peabody Picture Vocabulary Test) or selected verbal subtests from verbal intelligence tests (e.g, Wechsler Adult Intelligence Scales) could be used for this purpose.
TECHNICAL REPORT TR93-7 As noted above, regarding reading comprehension measures, it is important to select language tests that have content appropriate to clients' backgrounds.
Third, a word recognition measure in which clients must read aloud words presented in isolation is essential to any sensitive diagnosis of reading problems in adults. Research on diverse samples of adults who have varied widely in overall reading levels indicates very clearly that it cannot be assumed that word recognition skills are fully developed in any reader whose reading comprehension is deficient. Used in comparison to reading comprehension scores, a word recognition measure provides an important clue as to how much emphasis should be placed on lower-level and higher-level processing in instruction. This recommendation to grant a more prominent role in the diagnostic process to assessing word recognition gains further support from a study by Venezky, Bristow, and Sabatini (cited in Venezky, 1992).
They found that a locator vocabulary screen functioned just as well as, if not better than, a full three-hour test battery in making placement decisions for adult students in Adult Basic Education or GED classes. Ideally, a measure of pseudoword decoding would accompany the word recognition measure, since this provides the clearest evidence regarding the reader's grasp of the correspondences between letters and sounds, a skill that importantly underlies the entire reading process. Although there are presumably several tests that meet these needs, one available pair of measures that is commonly used in research on both children and adults are the Word Identification (real words) and Word Attack (pseudowords) subtests of the Woodcock-Johnson Psychoeducational Battery (mentioned above). These measures are well normed for adults and are quick to administer. In addition, for adults whose decoding skills are neither fully mastered nor wholly lacking, a supplementary diagnostic assessment (e.g., using the stimuli from Part II of the Decoding Skills Test) can provide more detailed information as to which sound-letter regularities have and have not been mastered.
Fourth, when good word recognition/decoding and good listening comprehension are seen in conjunction with poor reading comprehension,2 the examiner should pursue the possibility that decoding.automaticity is lacking, creating a bottleneck that impedes comprehension. To this end, it is important for the examiner to note, during the administration of the real word and pseudoword recognition tests, how slowly or haltingly the adult makes his or her responses. If an automaticity problem is indicated, then a normed test of reading
NATIONAL CENTER ON ADULT LITERACY 71
speed/efficiency, such as the Gray Oral Reading Test, can be given.
Because this test requires extensive reading aloud by the client, which may be unsettling, its use is only recommended when there is reason to suspect an automaticity problem.
Fifth, an interview with the client is obviously an essential component of treatment planning. The interview should provide information regarding the person's educational history, perceptions of which aspects of reading are causing difficulty, and broader vocational or personal goals for which help is being sought. The interview should also seek to determine whether the person's reading problems might be associated with visual or hearing difficulties or with a prior incidence of head injury or disease. Last, if mathematics proficiency is also to be included in the treatment objectives, then it, too, should be assessed directly, using a test that focuses on numerical calculation rather than on solving verbally presented problems (in which poor reading can interfere with performance).
This set of measures would provide sufficient information to guide decisions regarding initial placement and instructional needs and could also be used to evaluate the progress of an individual or program over time. The proposed measures are moderately dignity-preserving inasmuch as a minimum of oral reading is required and most recommended tests are designed to be discontinued once the examinee has made a certain number of errors; further preservation of dignity would be at the expense of sacrificing important information. The set of measures is also reasonably efficient, insofar as few of the tests are overly long, and scoring is generally quite straightforward. In contrast, the widely used TABE battery requires nearly three hours of testing (Venezky, 1992), which is far more than is needed for the assessment procedure that has been recommended. The criterion of face validity, however, is not as clearly met by the proposed battery.
Fortunately, it is clear from research, and from personal experience, that adults are willing to suspend face validity when working with a sympathetic tester. In administering pseudoword measures, for example, often one can stress just how strange the task is, but that it nevertheless helps to understand how a person figures out words. It is encouraging to read, in a review of selfperceived needs of learning-disabled adults, that 62% expressed a willingness to be tested, even though they felt that they already understood their problems.
In sum, a picture of an adult's overall level of reading proficiency, a profile of his or her strengths and weaknesses in different component processes (understanding spoken and written TECHNICAL REPORT T193-7 material, recognizing printed words, decoding based on lettersound regularities, and processing of text in a fast and efficient manner), and information about the client's educational history and current objectives can be derived from the set of measures proposed. Novice readers would be expected to show poor word recognition accuracy and low overall achievement levels; if their listening comprehension skills are relatively strong, a primary instructional emphasis on decoding and practice in word recognition would be called for, whereas if listening comprehension is also weak, then instruction aimed at revealing regularities in both spoken and written language would be needed.
Nonautomatic readers would be expected to show moderate accuracy but slow speed of word recognition, indicating a need for practice (especially speeded practice) in the reading of both isolated words and connected text. Weak comprebenders would have difficulties in fully extracting meaning from both written and oral language, regardless of the speed and accuracy of their lowerlevel reading skills. For these individuals, instruction could also focus on strategies and methods for improving comprehension.
Finally, independent of an individual's overall reading level and socioeconomic or educational background, pure cases of specific reading disability could be identified on the basis of far better proficiency in listening comprehension than in reading skills, but a distinction between specific reading-disabled adults and other poor adult readers cannot readily be made for those who do not show this clear-cut profile.