How to Measure English Learners' Development More Accurately

Blog Post
March 27, 2015

The ancient Greek philosopher Heraclitus famously posited that it is impossible to step into the same river twice. That is, by the time you go back for the second dip, the water you touched the first time is long downstream. This makes it challenging to get clear understanding of the river: is it full of fish? What’s its temperature? Etc. Each splash into the water is simply one slice of time. To get more complete knowledge, you’d need to measure it over a longer period of time.

This is as good an analogy as any for illustrating the challenges of responsibly assessing language learners. Each assessment is a limited time slice, and since these students’ abilities will show up differently on assessments—whether they are measuring English language proficiency or content knowledge—as their English language skills develop. This is why we usually measure these students’ linguistic and academic growth, rather than focusing only on their proficiency. We want to monitor how they’re performing over time, even when they fall short of mastery. But there are other problems with how we use assessments to define these students’ success.

Like any good diagnosis of a problem, we need to start by outlining the status quo. In our current system, we check the English language abilities of all students and then classify those with “limited English proficiency” as “English Learners” (though different states use different terms for this classification). Students who fall into this group must, pursuant to No Child Left Behind, receive research-based language support services. These continue until their English skills develop to a point when they’re able to access instructional content in English, participate in mainstream academic classes (taught in English), and demonstrate their academic knowledge on the state’s math and literacy assessments. When students reach that point, they are reclassified as “former English Learners” and cease to be language learners as far as the state is concerned. At that point, No Child Left Behind requires that states monitor the academic performance of former ELs for two years after they are reclassified. Here’s a diagram of the process from my 2014 paper, Chaos for Dual Language Learners:

There’s a basic level of systemic coherence to the approach: screen students to assess their linguistic strengths and needs, offer supports, measure their academic progress, check their English development, and end language services when they are no longer necessary.

But it has a data problem related to the river analogy I used above. We often hear about achievement gaps between students classified as English Learners (ELs) and students who are not classified. But there’s a problem with that framing. As Working Group for ELL Policy researchers Megan Hopkins, Karen D. Thompson, Robert Linquanti, Kenji Hakuta, and Diane August put it in an article several years ago, current policies create “a ‘revolving door’ effect, as more [English] proficient students exit and less [English] proficient students enter the EL subgroup.” That is,

Under current policy, the more successful schools are in reclassifying their ELs, the more poorly their EL subgroup performance looks...This poses a problem for accountability because it provides faulty information about the performance of the EL subgroup on long-term outcomes.
In other words, once an EL student develops his or her English language skills to a point where he or she begins to perform well on math and literacy assessments, he or she leaves the EL group. As a result, the EL group isn’t a static pool of students. Each time educators dip into that “river” with an assessment, they’re surveying a meaningfully different group of students.

Not only does this contribute to an unfair and inaccurate narrative about language learners—that, as a group, they are supposedly a drag on schools’ academic performance—but it makes accountability systems problematic. As Hopkins and her co-authors put it, “as former ELs are systematically removed from the subgroup, it becomes impossible to determine which schools and practices are successful for these students.”

Congress could simply require that districts monitor former ELs beyond the current two years—all the way until graduation.

And without information about how various instructional practices support English acquisition, it’s hard to set appropriately rigorous expectations for students, teachers, schools, or districts or to build comprehensive policy systems that support ELs. As I chronicled in Chaos for Dual Language Learners, states define ELs in a wide variety of ways—but most of federal law treats these students as a constant, commonly-defined subgroup. But if one state uses a low English proficiency bar and reclassifies many ELs after just a year or two of language services, it’s hard to compare their approach (let alone their results) with a state that uses a different approach and generally takes longer to reclassify ELs.

Those differences in policies mean that each state’s EL subgroup can vary in important ways from other states’ EL populations. It’s impossible to set the right reclassification policies without considering states’ approaches to ELs’ language supports, assessment, and more. To do that, we’d need better data on how ELs—and former ELs—develop and perform over a much longer time frame.

Fortunately, there’s a straightforward fix to this problem. Congress could simply require that districts monitor former ELs beyond the current two years—all the way until graduation. Hopkins and her co-authors suggest doing this by creating a “Total English Learners” (TELs) subgroup. This new grouping would make it possible for states to disaggregate the achievement data of all current and former ELs, which would make it possible to see how different states’ approaches to educating ELs work across a longer time horizon.

In addition, this change would dramatically change how the education system views—and treats—language learning students. When I discussed the TEL proposal with a colleague who works with DLLs on a daily basis, she was excited enough to reanalyze her students' math achievement data in that way. The results were amazing: she found that the TEL group started slightly behind native English speakers at the beginning of this year, but outgrew—and outperformed—the native English speakers by the mid-year interim assessment. She agreed that this provided a more accurate reflection of these students’ knowledge than the current system.

So: whenever it gets serious about rewriting No Child Left Behind, Congress really should consider making this (relatively easy) fix. Let’s hope that day comes soon (though I wouldn’t hold my breath).

(For more reform ideas from the Working Group on ELL Policy, click here.)

--Note: This post is part of New America’s Dual Language Learners National Work Group. Click here for more information on this team’s work."