Bilingual Research Bolsters Both Sides

Results Differ on Same Test Scores

Though they agree on almost nothing, people on both sides of the recent battle to reshape bilingual education in Massachusetts voice a common refrain: Look at the research.

Unfortunately, that’s easier said than done.

The data that opponents and proponents cite often have more to do with political leanings or opinions of bilingual education, while studies promoting the other side are dismissed or ignored.

That came into clear view last week during a legislative hearing on the future of bilingual education. Those backing efforts by Silicon Valley millionaire Ron Unz to eliminate bilingual programs in Massachusetts trumpeted their reports. Those seeking to preserve it pushed studies that support their point of view.

To muddle the situation further, both sides used the same test-score statistics to back their claims, leaving members of the Legislature’s education committee scratching their heads.

“One of the key problems is we don’t have good definitions of what it means to be a bilingual-education program,” said Paul Abraham, chair of Simmons College’s education department and director of its master’s program for teaching English as a Second Language.

“It goes everywhere from some minor support by an untrained aide to a full-fledged, two-way bilingual program such as the Amigos program in Cambridge – and everywhere in between.”

In Massachusetts, which has the nation’s oldest bilingual-education law, students who don’t know English can spend up to three years taking most of their classes in their native tongues, with the level of English rising as they move through school.

But critics say bilingual students often spend more than three years in bilingual classes and don’t learn English quickly enough.

Unz’s ballot initiative, one of four proposals debated last week, would place students in regular classes after just one year of intensive English immersion.

Bilingual-education proponents point to studies by a pair of George Mason University professors, Virgina P. Collier and Wayne P. Thomas, that analyzed 700,000 student records in five school districts around the country from 1982-1996.

They argue for teaching limited-English students in their native tongues for several years. Students who received native-language instruction and English instruction “at least through Grade 5 or 6″ are “doing well in school as they reach the last of the high school years,” their study says.

Collier and Thomas point out that two-way bilingual programs – in which English- and non-English-speaking children learn both languages together – have high success rates. But they also note that some poorly run bilingual programs “are no more successful than the best English-only programs in the long term.”

WestEd, an education research and development agency, also outlined three reviews of bilingual-education research in 1992, 1997, and 1998 – including one by Jay P. Greene, a scholar now at the conservative Manhattan Institute – that suggest there is some benefit in teaching students in their native language.

On the other side, Boston University professor Christine H. Rossell has published a paper critiquing the methods of the Collier-Thomas study. She also recently completed a study finding that the Unz initiative in California – known as Proposition 227 – has been implemented haphazardly.

Districts are letting students stay beyond English immersion classes for more than one year, her report found, for example.

Rossell’s report makes no judgment about Proposition 227 because the test-score data isn’t reliable. But she supports immersion as a way to teach limited-English students.

In a 1996 study, Rossell reviewed about 300 evaluations of bilingual programs, but found just 72 to be acceptable statistically. And of those, only one-fifth showed that traditional bilingual-education produced higher test scores in reading, her report found.

“The research evidence does not support transitional bilingual education as a superior form of instruction for limited-English-proficient children,” stated the study, which itself has been attacked for its methods.

Test scores don’t help.

Bilingual opponents say rising Stanford 9 scores of limited-English students in California show that one-year immersion is working.

Proponents draw the opposite conclusion: that the achievement gap between limited-English students and their peers is widening, showing immersion is no fix.

But statisticians say it’s wrong to draw conclusions from existing scores.

The Stanford 9 is not specifically a proficiency test for speaking, reading, or writing English. And no one has controlled for other factors that could affect test scores such as lower class sizes, more money spent in classrooms, and the type of immersion or bilingual programs the students were in.

Researchers noted that it’s not unusual for studies to reach opposite conclusions.

Said John B. Willett, Harvard Graduate School of Education dean, “In that context, it’s possible for the opponents to say one thing, and for the other side to say something else.”

Anand Vaishnav can be reached at

Comments are closed.