Large-scale research carried out both in the United States and in Canada in the past 30 years has provided compelling evidence that the critical variable in second language development in children is not the amount of exposure, but the timing and the manner of exposure [to the language of instruction].
For example, in 1997, researchers Thomas and Collier tracked 42,000 children in the United States, who entered school. These children had been placed under one of several programs that varied extensively in how much children were exposed to English language instruction and when they were exposed.
In one of these programs, non-English speaking children were placed directly into English-medium classrooms thereby receiving all of their education in English. In other programs, children received three years of initial instruction via their first language (with preparatory English language instruction) before moving on to an English-medium classroom.
In the most extreme (and innovative) type of program, children received up to six years of instruction in their first language while receiving, at the same time, six years of preparatory instruction in English before they were inserted into English-only classrooms for middle school.
To the surprise and consternation of many educators (and parents), the children who received all of their education in English learned the least amount of English and scored the most poorly on nationally normed and standardized tests of academic achievement, finishing, as a group, at the bottom 10 percent.
They were also the most likely to drop out of school before finishing. Conversely, children participating in one of the six-year programs actually completed their secondary education scoring well above the national norm for their native English-speaking peers.
Because of the sharp contrast between “popular belief” and the findings or predictions of educational researchers, we tested Grade 3 children participating in the Lubuagan First Language Experiment to measure their English language development.
Three components of English language learning were tested -- English listening, English grammar, and English reading.
In the Lubuagan experiment, there are three schools in each program with approximately 120 children per program.
The test results for English language development show a small but consistent advantage for those children in the First Language Component (FLC) program (in which most initial instruction in reading is in the mother tongue). In technical terms, none of the differences is statistically significant.
The test results clearly run contrary to the fear and concern of many parents (and educators) that children beginning in the first language are going to suffer educationally as a result. Not only are these children learning to read in their own language, they are actually learning more English than their peers who are receiving all of their instruction in English.
A Math test was also administered to both students receiving instruction in their first language, as well as those receiving instruction primarily in English and Filipino (i.e., the Control Group).
Steve Walter notes that Math is of theoretical interest to researchers on the Language of Instruction question. Thomas and Collier have suggested that one of the fundamental virtues of first language instruction lies precisely in its superiority as a vehicle for teaching and learning more complex and subtle concepts. Math is one such example. The logic is simple. Grappling with subtle, nuanced, or complex material requires well-developed language skills for communicating shades of meaning as well as new concepts and models that often have to be introduced by means of analogy.
Unlike the English test above, the results of the Math test are statistically significant. Walter notes, “The content of the Math test was taken directly from the Grade 3 curriculum. No effort was made to test the reasoning skills about Math concepts beyond the knowledge and skills taught in the national curriculum. The Math test was not subdivided by domain and so it is presented as a single score. The mean score for the control group was 8.7 (SD = 3.67) while that of the control children was 11.0 (SD = 6.98). The difference is statistically significant (T = 2.26; p = .026) though the level of significance is relatively modest.
This test result for Math coincides nicely with the predictions of research on mother tongue education. Children cope better with the introduction of a cognitively complex material when it is presented in their first language. Therefore, they are likely to do better on a test of Math skills, as Math gets more and more cognitively demanding as a child progresses in school.
SIL researchers will continue with the Lubuagan study for seven more years to show the long-term advantage of students educated in their mother tongue first. SIL invites educators and interested parties to interact with them on these education questions as we all seek together to extend education opportunities among ethnolinguistically diverse societies.
Greg and Diane Dekker are literacy specials for SIL International (http://www.sil.org and www.sil.org/asia/Philippines), a volunteer, non-profit organization that has worked in the Philippines since 1953. In cooperation with the Department of Education, SIL members carry out linguistic research and documentation of the languages and cultures of Philippine ethnolinguistic communities. For more information about multilingual education in the Philippines write to literacy_philippines@sil.org.
