Volume 13 Number 1
October 1973

Maurine A. Fry and Carole Schulte Johnson

PROGRESS in learning to read is considered to be dependent upon the child’s general language development. Early intervention programs for culturally different children who do not achieve well in our public schools have tended to place primary emphasis on language development (Hodges, McCandless, and Spicker, 1971; Maccoby and Zellner, 1970). Programs and suggestions for language development among these children continue to increase, but only limited evidence exists to indicate what particular aspects of their oral language production are most closely associated with differences in reading performance.

In an earlier study that controlled for differences in chronological age, socio-economic status and intelligence among second-grade children, the authors found a relation between a number of measures of oral language production and reading achievement (Fry, Johnson, & Muehl, 1970). The present study, using a design that controlled for chronological age, sex, and intelligence, was conducted to further explore the relation between oral language and reading achievement among two groups of culturally different children (see Note 2).

Subjects. Forty-five second grade Pima-Maricopa Indian children attending public schools in Mesa, Arizona, and 24 attending a BIA Day School in nearby Scottsdale served as subjects (Ss) in the study.

All Ss were administered the Lorge-Thorndike Intelligence Test, Primary I or 11 Battery and the Metropolitan Achievement Test (MAT), Primary I or 11 Battery. Because a selection procedure apppeared to determine which children were sent to the public schools, it was predicted that the Day School Ss would perform at a chance level on the Primary II Batteries. Therefore, the Primary I Batteries were used with Day School Ss and, the Primary 11 Batteries with public school Ss. Due to prolonged absences, it was impossible to complete the data for seven Ss.

The remaining 62 Ss were divided into two reading groups based on the median grade equivalent obtained on the reading comprehensive subtest of the MAT. An above average (AA) group included those Ss who scored 2.0 or above on the reading comprehension subtest. A below average (BA) group included those Ss who received a grade-equivalent of 1.9 or less on the reading comprehension subtest (see Note 3).

Oral Language Sample. Each S’s oral language sample was obtained individually in response to 20 picture stimuli and recorded on magnetic tape by a Stenorette Recorder (see Note 4). Due to illness and transfer, the language sample was obtained from 60 Ss.

The oral language protocols were transcribed from the magnetic tapes, and each S was identified by number so that further analysis was done by the second author without knowledge of the S’s reading group assignment. During the process of transcribing the protocols, two Ss were eliminated. One S’s response was partially erased and the second Ss speech was impaired.

Following Strickland’s (1962) procedures, the language protocols were first divided into phonological units (PUs), a speech segment that occurs between pauses or voice stops and is represented by a rising or falling inflection. Each PU was further divided into communication units (CUs), a grammatically independent clause with any of its modifiers. Units within PUs which failed to meet the definition of a CU were classified as mazes (hesitations, false starts, and meaningless repetitions) or partials (incomplete CUs).

Previous work (Darley and Moll, 1960) has indicated that for adequate reliability 50 CUs or language response units per S are desirable. Twenty picture stimuli were ample to obtain 50 CUs for each S in the previous study (Fry, et al., 1970). However, only 35 of the remaining 58 Indian children responded with 50 CUs or more. To preserve as many protocols as possible, the decision was made to analyze all protocols that included at least 30 CUs. Six Ss were eliminated at this point due to less than 30 CUs produced, leaving a total of 52 Ss.

The language analysis involved counts of word production holding stimuli constant, e.g., total words and total different words used in responding to ten pictures. Holding production constant, the first 30 CUs of each S’s typed protocol were analyzed for number of pictures needed to produce 30 CUs and total number of words included in 30 CUs. Using the methods employed by Strickland (1962), Loban (1963) and Kean (1965), 25 variables involving more detailed analysis of the syntactical structure of CUs were also considered. Four other variables were included due to the results of the earlier work (Fry, et al., 1970). These four were: number of nominal compounds (Baby Bear), total contractions, use of "there is" and use of conjoining verbs.

In order to maintain proportionality or sex between reading groups, one female and nine BA males were randomly removed. Data obtained from 42 Ss (12 AA females, 12 AA males, 9 BA females and 9 BA males) were included in the statistical analysis. Table I includes the data on chronological age, and intelligence for each reading group and reading-sex sub-group. A 2x2 analysis of variance (ANOVA) on chronological age and on intelligence revealed no significant differences.

The absence of a significant difference in intelligence was surprising since IQ score range could not be controlled due to sample shrinkage. Correlations between IQ scores and reading comprehension grade equivalents (GE) for reading group and sex group were calculated and are presented in Table II. Since the two reading groups did not differ in intelligence, the absence of a relation between IQ score and GE would be expected for the BA group, and as Table II indicates, this was the case. However, neither the assignment of Ss to reading group nor the absence of a difference in intelligence would allow one to predict the different relation found between IQ score and GE among the males as opposed to the females. For these Ss, knowing the IQ score of a second-grade Indian girl would have some predictive value, but knowing the IQ score of an Indian boy would be of no help in estimating his reading achievement.

*p < .05

**p < .01

To assess sex x reading group interactions in oral language usage, 24 variables with a high frequency of occurrence (used by > 22 Ss) were tested by 2 x 2 ANOVAs. In all 24 variables, there was no indication of a sex x reading group interaction or a main effect for sex at the .05 level. In three instances, a main effect for sex reached the .10 level with the males needing fewer pictures to emit 30 CUs, using more words and using more different words in describing ten pictures. One syntactic variable revealed a significant effect for reading group. Table III contains the analysis of variance summary for the 12-sentence pattern. The 12-sentence pattern is a simple structure composed of only a subject and verb slot, e.g., "He can play." BA group Ss tended to use this pattern approximately twice as much as AA Ss (BA X = 4.39, AA X = 2.00).

Nine variables that 20 Ss or more did not use at all were dichotomized and tested by chi square with Yates’ correction. All usage was tested against sex first and then reading group. The use of conjoining verbs was not independent of reading group classification at the .05 level. More Ss in the AA group made use of conjoining verbs than did the Ss in the BA group. A significant association with sex was obtained for the 1 2b 5 M variable. In this pattern, 1 represents subject and its modifiers, M represents linking verb, 5 represents the linking verb complement, and M represents the movable expression of place. A larger proportion of the males made use of this construction than did the females. When each reading-sex group was tested separately, use of the construction was independent of reading group classification.

*p X .01

While interesting, little confidence can be placed in the absence of a relation between intelligence and reading achievement among second-grade Indian boys. Abundant data should be available to support or refute this result, but little data seems to have been published. In a recent study of seventh-grade Ss (McCandless, Roberts, and Starnes, 1972), the girls’ standardized achievement test results were better predicted by a conventional group intelligence test than were those of boys. The practical difference may be slight, but the correlation for disadvantaged girls (white and black) was .84 as compared to .56 for disadvantaged boys (black and white).

As with other recent investigations (Bougere, 1969; Fry, et al., 1970; Menyuk, 1969), essentially no significant sex differences in oral language production were found.

These data lent minimal support to a relation between oral language production and reading achievement among second-grade Indian children. The more advanced readers did use more conjoining verbs and used the simple noun-verb sentence considerably less often than the poorer readers. However, due to the disproportionate representation of public school Ss in the AA group, the differences in oral language usage may associate with SES background, as well as reading performance.

It is also possible that a stronger relation between oral language production and reading achievement would have appeared if more than 30 responses per subject could have been included in the oral language sample. Another possibility that must be considered is that a relation between oral language and reading achievement only obtains when extreme groups are involved and these two groups of Indian children were not that different in reading achievement (AA Group Mdn. = 2.1, Range = 2.0-4.7) (BA Group Mdn. = 1.8, Range = 1.5-1.9).

Bougere, M. B. "Selected Factors in Oral Language Related to First-Grade Reading Achievement." Reading Research Quarterly, 1969, 4, 31-58.

Darley, F. L. and Moll, K.L. "Reliability of Language Measures and Size of Language Samples." Journal of Speech and Hearing Research, 1960, 3, 166-173.

Fry, M. A., Johnson, C. S., and Muehl, S. "Oral Language Production in Relation to Reading Achievement Among Select Second Graders." In D. J. Bakker and P. Satz (Eds.), Specific Reading Disability, Rotterdam: Rotterdam University Press, 1970, pp. 123-146.

Kean, J. M. "An Exploration of the Linguistic Structure of Second and Fifth Grade Teachers’ Oral Classroom Language." Unpublished doctoral dissertation, Kent State University, 1965.

Loban, W. D. The Language of Elementary School Children. Champaign: National Council of Teachers of English, 1963.

Maccoby, E. E. and Zellner, M. Experiment in Primary Education. New York: Harcourt Brace Jovanovich, 1970.

McCandless, B. R., Roberts, A., and Starnes, T. "Teachers’ Marks, Achievement Test Scores, and Aptitude Relations with Respect to Social Class, Race, and Sex." Journal of Educational Psychology, 1972, 63, 153-159.

Menyuk, P. Sentences Children Use. Cambridge, Mass.: MIT Press, 1969.

Strickland, R. C. "The Language of Elementary School Children: Its Relationship to the Language of Reading Textbooks and the Quality of Reading of Selected Children." Bulletin of the School of Education, Indiana University, 1962, 38, No. 4.

1 This study was supported, in part, by a faculty grant to the first author from Arizona State University. The authors gratefully acknowledge the cooperation of the participating schools and the assistance of Tom Wilkins and Don Milam in data collection and analysis.

2 The data necessary to control for socio-econornic status (SES) could not be obtained.

3 Public school Ss were disproportionately represented in the AA Group (3:1).

4. The procedures and pictures are described in more detail in Fry et al., 1970.