The Acquisition of Academic English by Children in Two-Way Programs: What does the Research Say?

Table 8
Summary of Studies

study	results	commentary
NO COMP. GROUP
Sugarman & Howard	gains in English proficiency	tests unstandardized, no comparison group
		no measure of initial competence
de Jong (2002)	2-way achieve 29 to 37 percentile by grade 5	small n, no comparison group
		no measure of initial competence
Lindholm	2 -way achieve 23 to 30 percentile by grade 6	no comparsion group
	one group only at 9 percentile in grade 5
Thomas/Collier (Oregon)	2-way better than state, district norms	small n (12), initial competence not clear
COMP. W. MAINSTREAM
Ajuria	2-way better than comparisons in mainstream	small n (11,7) ,only 1 year, no gains
		second semester, comparisons in print-
		deprived environment, modified test
		no measure of initial competence
Castillo	2-way much better than comparisons in	small n (4 comparisons!), up to grade 2
	mainstream	only, 2-way scores very high after grade
		one, then decline, no measure of initial
		competence
Thomas/Collier (Houston)	ESL only > TBE > submersion	uncontrolled
COMP. W. TBE
Clayton	TBE graduates do better than 2-way graduates	no measure of initial competence
		2-way had "inferior" reading program
Cazabon et. al.	2-way do better than TBE	small n, up to grade 2 only, initial
		English competence unclear
Thomas/Collier (Houston)	2-way better than TBE	grade 1 scores very high, then decline
		no measure of initial competence
		2-way advantage present very early
COMP W. DEVELOPMENTAL
de Jung (2004)	2-way better than developmental	small n, ceiling on reading test
		Dev. have better oral compr. at K
COMP. W. Nspeakers of Eng
Alanis	2-way = English dominant	small n, 2-way did not begin with zero
		English; 2-way not really 2-way
Ajuria	2-way better than English native speakers	test modified, mainstream class print-deprived
		one year only
Castillo	2-way better than English native speakers	up to grade 2 only; no measure of intial
		competence; small n (11 2-way children)
Clayton	2-way better than Native American children	not clear if 2-way group alone was
		included or entire 2-way + TBE group
Thomas/Collier (Houston)	2 way better than English native speakers	Uncontrolled (see also above)

The results are thus encouraging but they are not the overwhelming and massive support we sometimes read about in the popular press.

It should be pointed out that supporters of bilingual education have criticized studies claiming to support immersion for similar flaws. In Krashen (1996) I faulted Gersten (1985) for a small sample size and a short duration (up to the end of grade 2), I also noted that Rossell’s (1990) analysis was suspect because scores were very high long before reclassification.

Rossell and Baker (1966) criticize many studies for not controlling for individual differences at the outset of the program, which is also a problem in the studies discussed here. I argued (Krashen, 1996) that with large numbers of studies, this is not a problem if there is no reason to suspect differences; large numbers of studies provide quasi-randomization. This is clearly what is called for in the case of two-way bilingual education.