The Claim That Launched a Thousand Curricula
Ask any educator why bilingual education is valuable and you will hear, alongside the obvious linguistic benefits, a cognitive claim: learning a second language makes you smarter, not just more linguistically capable but genuinely better at focusing, planning, switching between tasks, and ignoring irrelevant distractions. The claim rests on two decades of research, much of it from Ellen Bialystok's influential lab at York University in Toronto, and it has shaped education policy from Toronto to Singapore to the bilingual immersion programs spreading across American suburbs.
The idea is elegant. Bilinguals constantly manage two active language systems, suppressing the one they are not using while activating the one they are, and this continuous juggling exercises executive control, the cognitive machinery responsible for focus, task-switching, and working memory.
The evidence, it turns out, was being filtered before anyone could see the full picture.
What 152 Studies Actually Show
In 2018, Minna Lehtonen and colleagues at Åbo Akademi University published the most comprehensive meta-analysis of the bilingual advantage to date, collecting 891 effect sizes from 152 studies on adults across six domains of executive function: inhibition, shifting, working memory, monitoring, attention, and verbal fluency, including unpublished data that had never made it through the journal pipeline.
Before correcting for publication bias, the raw numbers showed what the published literature had long suggested: a very small bilingual advantage for inhibition, shifting, and working memory, though the effect sizes hovered near the floor of statistical detection.
Then Lehtonen's team applied standard corrections for publication bias, statistical methods that adjust for the well-documented tendency of journals to publish positive findings and reject null results. The advantage vanished completely. Across all six domains, the corrected effect size dropped to approximately zero, and in verbal fluency, bilinguals actually performed slightly worse than monolinguals, likely because splitting linguistic exposure between two languages reduces vocabulary depth in each one. The bilingual advantage was not small; once the full evidence was visible, it did not exist at all.
The File Drawer Proved It
Three years earlier, Angela de Bruin, Barbara Treccani, and Sergio Della Sala had published a smoking gun in Psychological Science, comparing studies presented at scientific conferences, where results are accepted before they are known, to those that eventually appeared in peer-reviewed journals. Of 104 conference abstracts on bilingualism and executive control, roughly half supported an advantage and half did not.
But 63% of the pro-advantage studies made it into print, compared with only 36% of the null results. The publication pipeline was filtering out negative findings at nearly double the rate, constructing a literature that looked far more supportive than the raw evidence warranted.
Scale Did Not Help
If the advantage existed but was just small, larger and more representative samples should have detected it.
In 2019, Anthony Dick and colleagues analyzed 4,524 children aged 9 to 10 from the Adolescent Brain Cognitive Development study, the largest long-term brain development study in the United States, testing for bilingual advantages across multiple executive function tasks. Their result, published in Nature Human Behaviour, was unambiguous: no advantage in any measure. A year later, Nichols and colleagues tested 11,000 adults in a cognitive battery published in Psychological Science and reached the same conclusion, with a sample nearly 75 times larger than the typical study in Bialystok's original research program.
The Strongest Counterargument
Bialystok and other proponents have not conceded, and their most compelling response, articulated in a 2019 PNAS paper by DeLuca, Rothman, Bialystok, and Pliatsikas, is that bilingualism is not a binary category but a spectrum: a heritage speaker who uses their second language only at home is not engaging the same cognitive demands as a simultaneous interpreter switching languages hundreds of times per hour.
If the cognitive advantage depends on the intensity, duration, and context of bilingual experience, then studies grouping all bilinguals together may dilute a real effect that exists only in specific subpopulations. But this reasoning also makes the hypothesis nearly unfalsifiable: if no amount of negative evidence can close the case because the right subgroup has not yet been tested, the hypothesis becomes a moving target rather than a testable scientific claim. In a systematic review, Gunnerud and colleagues found that better-designed, more recent studies consistently produce smaller effects, with the proportion supporting an advantage shrinking as methodological controls improved.
Put It in Numbers
The uncorrected effect size for the bilingual advantage in inhibition, the strongest domain in the meta-analysis, was approximately d = 0.06, which on an IQ-like scale translates to an advantage of 0.9 points. The cognitive boost from a single cup of coffee is larger, and the effect of one good night of sleep is roughly five times greater. An advantage of 0.9 points falls below what any person could detect in daily life, below the measurement error of the instruments used to assess it, and below the threshold that any educational intervention could meaningfully exploit.
That is the number before correcting for bias. After correction, it is zero.
What We Didn't Prove
This evidence applies specifically to executive function, the set of cognitive abilities that the bilingual advantage hypothesis has always targeted. Speaking two languages opens access to two literatures, two cultures, two professional networks, and the approximately 3.3 billion people whose primary language is not English, benefits that stand entirely on their own.
The meta-analysis focused on adults and the ABCD study tested children aged 9 to 10, leaving open the question of whether very early bilingual exposure produces advantages in toddlers that dissipate by school age. The evidence on bilingualism and dementia delay is a separate and contested literature that has not undergone the same quality of bias-corrected meta-analytic review.
The Bottom Line
The bilingual advantage in executive function was one of the most widely cited findings in cognitive psychology for two decades. It influenced school curricula, parenting advice, and government education policy. When researchers corrected for the systematic filtering of negative results, the advantage disappeared entirely across 152 studies and 891 effect sizes. Two subsequent studies with samples in the thousands confirmed the null. Learning a second language remains one of the most valuable things a person can do. It just does not make you measurably smarter.
What You Can Do
- Learn languages for communication, culture, and career. Better cultural competence, broader job markets, deeper literary and intellectual access. These are the real reasons, and they are more than enough.
- Be skeptical of "brain training" claims attached to any activity. Music, chess, language, apps: when researchers correct for bias, most transfer effects shrink to near zero.
- Treat "bilingualism prevents dementia" claims with caution. That is a separate body of evidence, not supported by the same quality of bias-corrected review, and currently contested.
- For parents: pursue bilingual education for real reasons. "It will make your child smarter" is not one the evidence supports. Communication, cultural identity, and economic opportunity are sufficient justification.
- Watch for the file-drawer effect in other fields. The bilingual advantage is a textbook case of how publication bias can sustain a false consensus for decades. If a finding sounds too clean across dozens of studies, ask what happened to the studies that found nothing.