The most serious problem with this study, however, was that it does not really qualify as a two-way bilingual program. Both English and Spanish-dominant children were in the same classroom, but the program had much more emphasis on English than Spanish. It appeared to be an English immersion program with some support in the primary language: it was supposed to be 50/50, but “it was clear from classroom observations and interviews that the 50/50 split was not implemented ... teachers ... utilized more Spanish than English and lacked Spanish resources in all content areas” (p. 16). Even though the schools were on the border (Brownsville), “most students had a strong preference for English” and “teachers stressed the English TAAS [Texas Assessment of Academic Skills] for their students” (p. 16). Alanis also notes that there was no emphasis placed on Spanish literacy: “the original goal of the program was only Spanish oral proficiency as opposed to equal levels of bilingualism and biliteracy” (p. 16).²

Some of the studies discussed previously also compared two-way students to English dominant and native speakers of English. Recall that Castillo (2001) found that two-ways did better than a group of native English speakers (n = 30), but only 11 two-way students were tested and there was no information on their initial level of English. Ajuria (1994)’s two-way students also did better than native speakers of English in the mainstream (n = 20), but recall that the mainstream classroom was severely lacking in many essentials and a modified version of a standardized test was used. Clayton (1993) reported that two-way students did better than a group of Native American children with similar backgrounds, but it is not clear whether the “two-way” children were in two-way or whether it was a combined sample of children in two-way and TBE: The sample size for the comparison with Native American children suggests it was a combined group: there were 83 children in the two-way group and 59 in the TBE group in the two-way vs. TBE comparison, but 150 in the two-way vs. Native American comparison.

Thomas and Collier (2002) also provide a comparison with native speakers of English in their Houston sample. All native speakers performed around the 50th percentile in English reading; two-way students scored above this level at all grade levels tested, an impressive result even though there was no control for variables such as neighborhood or poverty.

Discussion

With respect to question (1), “Do they acquire significant amounts of English?, several studies reassure us and show that children in two-way programs improve in English. Table 7 presents grade five results in NCE’s and percentiles for the three studies using standardized tests, including all groups tested. Even discounting the lowest score as an outlier (9th percentile in Lindholm), the results are variable, and not always high enough for reclassification, and the Thomas and Collier score is much higher than the others.

10