27.06.2013 Views

Gibson Ferguson Language Planning and Education Edinburgh ...

Gibson Ferguson Language Planning and Education Edinburgh ...

Gibson Ferguson Language Planning and Education Edinburgh ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

50 <strong>Language</strong> <strong>Planning</strong> <strong>and</strong> <strong>Education</strong><br />

3.2.1 Research on the effectiveness of bilingual education programs<br />

One of the most influential of the early program evaluations was the Baker <strong>and</strong><br />

De Kanter (1981) study. Federally funded to inform policy decision-making, this,<br />

like other early studies, was narrowly focused on the comparative effectiveness of<br />

Transitional Bilingual <strong>Education</strong> (TBE) <strong>and</strong> English-only program models, <strong>and</strong><br />

on a limited range of outcomes – specifically English language attainment <strong>and</strong><br />

performance on certain non-language subjects tested in English.<br />

Working within a positivist research paradigm, Baker <strong>and</strong> De Kanter (1981)<br />

reviewed well over 100 primary studies, identifying only 28 as methodologically<br />

sound on the basis of such criteria as the presence of a comparison group, r<strong>and</strong>om<br />

assignment between treatment <strong>and</strong> comparison groups <strong>and</strong> adjustment for preexisting<br />

differences between groups. Having surveyed these 28, they concluded that<br />

there was no consistent evidence for the effectiveness of bilingual education <strong>and</strong><br />

advised that ‘exclusive reliance on this instructional method is clearly not justified’<br />

(Baker <strong>and</strong> De Kanter 1981: 1). They did, however, support experimentation with<br />

‘structured immersion’ approaches.<br />

Shortly after its publication, 23 of the 28 studies in the Baker <strong>and</strong> de Kanter<br />

(1981) review were reanalysed by Willig (1985), 15 employing the statistical technique<br />

of meta-analysis to measure <strong>and</strong> combine the size of program effects, even when<br />

these were not statistically significant, <strong>and</strong> to adjust for methodological flaws in some<br />

of the studies reviewed. She concluded, in sharp contrast to Baker <strong>and</strong> de Kanter,<br />

that there were ‘positive effects for bilingual programs … for all academic areas’<br />

(Willig 1985: 287), <strong>and</strong> that bilingual education using the native language was better<br />

than the alternative of no special treatment (that is, ‘sink or swim’).<br />

Unsurprisingly, given the politicised climate of program evaluation, these<br />

contrasting findings have been strongly contested. Supporters of BE note the<br />

questionable ‘vote counting’ technique of the Baker <strong>and</strong> De Kanter study, the narrow<br />

range of outcomes it considers <strong>and</strong> the decisions to exclude certain studies. Opponents,<br />

meanwhile, point to the small number of studies reviewed by Willig (1985)<br />

<strong>and</strong> the mechanistic nature of the meta-analysis procedures.<br />

The details of these criticisms need not detain us, however, as it may be more<br />

useful here to consider more recent studies in the same tradition, namely those by<br />

Ramirez et al. (1991), Rossell <strong>and</strong> Baker (1996) <strong>and</strong> Greene (1997).<br />

3.2.1.1 The Ramirez et al. study (1991)<br />

The Ramirez study is significant both for its quasi-experimental longitudinal design<br />

<strong>and</strong> for including – in a three-way comparison – the less commonly evaluated late<br />

exit bilingual program (bilingual instruction up to sixth grade). 16 A total of 2,300<br />

Spanish-speaking pupils enrolled in English-only ‘structured immersion’, early exit<br />

TBE (65–75 per cent use of English) as well as late exit bilingual programs were<br />

followed at nine sites over four years. Efforts were made to match comparison groups<br />

for background characteristics, to establish fidelity of program implementation <strong>and</strong>

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!