Collier and Thomas (2002) provide some additional data. Two-way children in Houston appeared to be well ahead of children acquiring English whose families refused special help for their children. These “refusers” scored at the 20th percentile level at grade five (NCE = 32). Children in ESL-only programs, however, reached the 66th percentile in grade 5 (NCE = 59), doing better than two-way children. Two-way children in Houston scored at the 52nd percentile in grade 5 (NCE = 541.3). It is possible, however, that those assigned to ESL had higher levels of English when entering school, and the data are not controlled for poverty and neighborhood, as are students in Collier and Thomas’ other Houston comparisons (see below).

Group Three: Comparison with Other Bilingual Education Programs

Comparison to Transitional Bilingual Education

In Clayton (1993), “Spanish-speaking” students who had participated in transitional bilingual education outperformed “Spanish-speaking” students who had been in a two-way program on English reading. The two-way students had been in the program for at least four years. The results were similar when parental education was controlled, but the difference was greater when parents in both groups had no education. Two-way students outperformed Native American students from similar SES status. Students were in grades 3 through 8. We do not know if “Spanish-speaking” meant limited in English, however. Clayton suggests that the two-way program had an inferior reading program (p. 155) and was a newer program (p. 157). Table 2 presents data from two successive years of testing.

Table 2:
Comparison of Two-way and TBE Graduates, Grades 3-8

From: Clayton(1993)

Cazabon, Lambert and Hall (1993) report on the English proficiency of Spanish- speaking children in a two way program in Massachusetts, as compared to children “in a standard bilingual program” (p. 5). Two-way and comparison children were similar in social class and intellectual ability, as measured by Raven Matrices test.

It is not clear that all children classified as Spanish speakers (“Spanish amigos”) were Spanish dominant. In grade 1, for two cohorts combined, seven out of 47 were considered English-dominant (Lau classification score of 4 or 5 out of 5), and none were classified as completely Spanish-dominant. This measure was not applied at the beginning of the program, but seven months into the first grade; it is quite possible that the children had acquired considerable English conversational competence by then.

5