THE MISREADING OF HIGH-FREQUENCY WORDS'

THE MISREADING OF HIGH-FREQUENCY WORDS' Richard L. Allington, Ph.D. James T. Fleming, Ed.D. State University of N e w York at Albany While the misread...
1 downloads 0 Views 434KB Size
THE MISREADING OF HIGH-FREQUENCY WORDS' Richard L. Allington, Ph.D. James T. Fleming, Ed.D. State University of N e w York at Albany While the misreading of visually similar words has been thought to lie in a visual-perceptual deficit, recent research has seriously undermined such hypotheses. This study suggests

that such errors are more typically a function of a lack of effective and efficient integration of semantic and syntactic cues with visual information.

T h e etiology of the misreading of high-frequency words (e.g., was-saw, went-want) has long been thought to lie in a visual-perceptual deficit (cf. Orton, 1925, 1937). However, a variety of recent research studies has demonstrated that even though poor readers often misread these words when presented in isolation, they sustain no visual- perceptual deficit when compared to normal and good readers on tasks requiring visual discrimination and visual memory (Allington, Gormley, & Truex, 1976; Vellutino, Steger, 8c Kandel, 1972; Vellutino, Steger, Desetto, & Phillips, 1975; Vellutino, Steger, Moyer, Harding, & Niles, 1977). SimilarIy, others have demonstrated that optical similarity is not a sufficient condition to account for these errors (Shankweiler & Liberman, 1972). One as yet unexplored facet of these misreadings is the relative availability and use of the semantic and syntactic cue systems by readers. That is, good readers may misread less often in context as a result of effective utilization of these cues; while poor readers utilize these cues less and therefore misread high-frequency words more often (Smith, 1973). This study was designed to test the effects of context on the identification of high-frequency words. Since most previous research has presented these words in isolation, o r in no-context situations, we have no firm basis for discussing misreadings which occur when reading in context. Neither do we have clear evidence that poor readers differ from good readers in their ability to employ the semantic and syntactic cues when reading in context.

METHOD Materials

A passage of approximately 250 words was selected from a second-grade basal reader. T h e passage was rated at a 2.8 readability level (Spache, 1953). 'This research was supported by a grant from the Research Foundation o f the State University of New York. 0022-4669178l1201-004lSOl.0010

__

418

THE JOURNAL OF SPECIAL EDUCATION VOL. 12/NO. 4/1978

0 RW

There once was a girl who was good and kind, but very poor. Many nights she and her mother had to go to bed hungry. Finally, there was no food anywhere i n the house, and they were very hungry. put once kind forest how nuts she want a forest cooking mother eat woman go this the very was ripe to for many to was words until the these poor went knew food your pot i t time the say began a black said kind handing spoke to they

Figure 1. Illustrative samples of the text in each condition: original (0)and random words (RW). This passage was then altered with the words randomly ordered. T h e experimental materials were the passage in the two conditions - original (0)and random word (RIV). For both conditions the stimuli were presented in horizontal lines retaining the left-right and top-to-bottom reading patterns (see Figure 1).

Subjects All fourth-graders in a rural elementary school were individually screened for reading ability on the Word Identification Test of the Woodcock Reading hlastery Tests. Students with raw scores between 50 and 96 (2.0 and 3.6 grade equivalents) were classified as poor readers, since their achievement lagged behind grade placement by 1 year. Twelve subjects were randomly selected from this group for the experimental tasks. Students with raw scores between 108 and 127 (grade equivalents of 4.5 and 7.0) were classified as good readers, and 12 subjects were also randomly selected. T h e mean raw score for the poor reader group was 76.25 (2.8 grade equivalent), while the mean raw score for the good reader group was 113.25 (4.9 grade equivalent). T h e mean chronological ages for the poor and good reader groups were 9.5 and 9.6 years, respectively. Sex distribution for the poor reader group was 9:3 males to females, while for the good reader group the ratio was 6:6. Procedure

All subjects were tested individually in a small room adjacent to the classrooms. Subjects were told they were going to read a story two times but that one version of the story had been altered to make it more difficult to read. They were then directed to simply read each passage aloud as best they could. Subjects were told at the beginning of each experimental session that this was not a test but rather an experiment to see how persons learn to read. Each subject was presented with the passage in both conditions, though the order of presentation was randomly assigned to control for effects of familiarization with the experimental materials. Each subject’s performance was tape recorded for later analysis.

,

THE jOURNAL OF SPECIAL EDUCATION VOL. 12INO. 4/1978

TABLE 1

THE HIGH-FREQUENCY WORDS Began black but came ***her *how knew

look no

on put some tell *then *there these they took

**very want **was well went *were when who

41 9

TABLE 2 MEANS A N D STANDARD DEVIATIONS FOR BOTH GROUPS ON EACH TASK Context

X

5D

Good readers

35.5

Poor readers

33.7

Random

X

5D

1.17

35

1.21

2.53

28

4.51

From Allington, 1977.

. ‘Each asterisk marks additional occurrences in materials.

Scoring T h e data of particular interest were the number of high-frequency, loiv discriminability words misread in each condition. These words were those which had had a greater than 5% error rate in an earlier study of the “confusability,” o r discriminability, of 120 high-frequency words (Allington, 1977b). Table 1 lists these words. T h e data on misreading of these words were gathered from listening to the audiotaped recordings of the experimental sessions. Both experimenters scored the recordings independently. Discrepancies between scorings were minimal, and these were resolved by repeated listenings to the recordings.

RESULTS T h e 2 (groups) x 2 (treatments) ANOVA with repeated measures on the last factor demonstrated that the recognition of the high-frequency, low discriminability words in context did not differ between groups ( F = 5.18; df = 1,22; p >.01). Since the experimental material was selected to approximate the instructional level of the poor readers, this result should not be too surprising. Both groups were able to read this material and recognize nearly all of the target words (see Table 2 for group means and standard deviations on the tasks). Differences between groups did exist on the recognition of the highfrequency, loiv discriminability words in the no-context, o r random order, condition (F= 26.95; df = 1,22; p < .01). That is, the good readers were significantly more accurate on recognition without context. T h e good readers’ performances in the random order condition was similar to their performances in context, while the poor readers’ accuracy dropped substantially.

DISCUSSION Neither the hypotheses of Orton (1925, 1937) nor Smith (1973) fully explain the performance of the poor reader group. T h e difference in their performances in the two conditions suggests that these poor readers were able to effectively employ the semantic and syntactic cues available in the context condition. This supports Goodman’s (1965) contention that context facilitates word

420

THE JOURNALOF SPECIAL EDUCATION VOL. 12/NO. 4/1978

recognition. Smith (1973) holds a similar point of view but, as noted, has argued that misreadings which violate semantic and syntactic constraints indicate a lack of attention to these cues. T h e lack of differentiation in the performance of good and poor readers ignore semantic and syntactic cues when available. Similarly, the poor readers’s performances in context undermines Orton’s hypothesis that these errors are attributable to visual similarity, or the lack of visual differentiation. IVhile at first glance the poor readers’ performances in the no-context condition may seem to suggest support for a visual similarity hypothesis, the recent research of Vellutino et al. (1972, 1975) and Allington et al. (1976) undermines this perceptual deficit hypothesis. Rather, an equally plausible hypothesis is the associational deficit suggested by Vellutino et al. (1975). That is, the poor readers may not have reached the word identification levels in no-context conditions. However, poor readers approximated the performance of good readers when reading in context, a significant improvement over their no-context performance. This suggests that poor readers can employ available language cues to reduce inaccuracy when reading. This conclusion raises two important points for those working with disabled readers. First, assessments of word identification which provide no contextual information (e.g., flashcards, word lists) seem generally to provide less than adequate information about a learner’s reading ability and thus should be de-emphasized. This seems particularly true for disabled readers, as demonstrated in this study and an earlier effort (Allington, in press). Second, since disabled readers were more accurate in context, it would seem most beneficial to structure such situations as often as is possible. Since the ultimate goal of remedial instruction in reading is the development of reading ability, not accuracy on word lists, instruction should focus on reading in context. This would take advantage of the strength of the learner rather than focusing on a weakness which is tangential to the ultimate goal. Of final note are the data on rate of reading, since the poor readers took over twice as long on the average to read the context passage as the good readers (2.75 minutes versus 1.01 minutes). Thus, while accuracy did not differ in this condition, the good readers were more efficient. These data suggest that good readers employ either word recognition strategies or contextual information more efficiently. T h e poor readers’ performances are marked with disfluency. T h e word-by-word reading suggests that the poor readers employed semantic and syntactic cues differently than did the good readers. That is, while the good readers seemed to predict and read in phrases, the poor readers seemed to use semantic and syntactic cues after the fact, to confirm or reject responses. Instruction for these disabled readers might profitably focus on the development of fluency, or increased efficiency, of reading. This increased efficiency would allow the disabled reader to read more material; and increased reading typically improves the quality of reading (Allington, 1977a).

THE JOURNAL OF SPECIAL EDUCATION VOL. 121NO. 4/1978

421

References reading: A search for causes. I n J. Allington, R. L. If they don't read much Kavanagh S: I. hlattingly (Eds.), Lnnhow they ever gonna get good? Journal guage bj ear and by q e : The relatioxships ofReading, 1977.21, 57-61. (a) betureen speech and reading. Cambridge, Allington, R. L. T h e visual confusability of hlass.: h l I T Press, 1972. high frequency words. Journal of Learnirtg Disabilities, 1977, 10, 444-449. (b) Smith, F. Psycholinguistics and reading. New York: Halt, Rhinehart 8- IVinston, 1973. Allington, R. L. IVord identification abilities of severely disabled readers: A Spathe, G. A new readability formula for comparison in isolation and context. primary grade reading materials. Jouriral of Reading Behavior, in press. Elernentarj School Journal, 1953, 53, Allington, R. L., Gormley, K. G., 8.Truex, 410-413. S. T. Poor a n d normal readers' Vellutino, F. R.. Steger, J. A., S: Kandel, achievement on visual tasks involving G. Reading disability: An investigation high frequency , low disc rim inability of the perceptual deficit hypothesis. Corwords. Journal of Learning Disabilities, tex, 1972,8, 106118. 1976,9, 292-297. Goodman, I(. A linguistic study of cues Vellutino, F. R., Stcger, J- A-, Desetto, L.2 & Phillips, F- Immediate and delayed and miscues in reading. Elemenkary Eftgrecognition of visual stimuli in poor and lkh, 1965,42, 639-643. normal readers. Jouriial of Expen1nel1fal Orton, S. T. Ij'ord-blindness in school Child Psycholog), 1975,19, 223-232. children. Archives of h'eurology and PSJVellutino, F. R., Steger, B. hi., hloyer, S. chiatr).. 1925,14, 581-615. C., Harding, C. J., S: Niles, J. A. Has the Orton, S. T. Reading, writing, and speech perceptual deficit hypothesis led us problems in children. London: Chapman astray? Journal of Learning Disabilities, and Hall, 1937. Shankweiler, D., 8- Liberman, I. Y. hfis1977,10, 375-385.