Unfortunately, no studies of such a nature have been found. The closest approximation to our "ideal" study is the longitudinal research of Loban (1961, 1963, 1964, 1967). He presents data on reading achievement in grades 4, 5, 6, 7, and 8 for children who were evaluated for oral language competency by means of a vocabulary test administered orally, and teachers' ratings on (a) amount of language, (b) quality of vocabulary, (c) skill in communication, (d) organization, purpose and control of language, (e) wealth of ideas, and (f) quality of listening. The Stanford and California Reading Achievement Tests were used to assess reading achievement.

Figure 5 presents data from Loban's (1964) research. The figure has been constructed from data presented in Table 16 (p. 117) of Loban's report, which gives median years and months for students in high and low oral language groups who scored above or below the norm for their age group. High language ability students were those who scored two standard deviations above the mean of the oral language ratings obtained in kindergarten; low language ability students scored two standard deviations below the mean. In constructing Figure 5, we have assumed that the data for fourth grade students are deviations above or below average for 9-year-olds, that is, students beginning grade 4; for 10-year-olds at grade 5; 11-year-olds at grade 6, and so forth. Actual age figures were not given in Loban's report-just deviations from age norms. Thus the data points of Figure 5 are only close approximations. For our purposes they are accurate enough, however, for they clearly point to a strong relationship between oral language ability in kindergarten, and subsequent reading ability, at least for extreme groups.

While Loban nowhere presents correlation coefficients for oral language and reading achievement for the total sample of students studied, he does present scatter diagrams for grades 4, 6, and 8. From these diagrams we have computed contingency coefficients-.36, .49, and .52 for grades 4, 6, and 8 respectively.1 Thus, for the total group, as well as the extreme groups of Figure 5, there is a positive relationship between oral language and reading. Furthermore, the relationship grows with increases in grade level, which is consistent with the expectation that auding and reading test performance will be more highly correlated after the learning to read (decoding) phase.

Additional data regarding relationships between auding and reading are presented in Table 4. There we have what can only be regarded as a limited, unsystematic sample of the many studies that report correlations among all manner of measures of auding and reading tests. Literally dozens of unpublished master's theses and doctoral dissertations which report such correlations can be found cited in various sources (Duker's 1968 bibliography is a prime source). The auding and reading tests described in many of these unpublished reports, and also in many published reports, comprise a true potpourri of auding and reading tasks. In most cases auding and reading tests differ in content, response task, duration, and format.

1. The use of a 2 x 2 contingency table restricts the upper limit of the correlation coefficient to +.71 (Siegel, 1956). Thus the contingency coefficients tend to underestimate the relationship between oral language and reading.