Cognates are words that share both meaning and spelling across languages, such as the Dutch-English pair “school,” while non-identical cognates like Dutch “rivier” and English “river” share meaning but differ slightly in form. For decades, psycholinguists have observed that bilinguals recognize cognates faster than other words, a phenomenon known as cognate facilitation. The prevailing explanation held that this speed advantage arises because both languages activate their versions of the word simultaneously, sending converging signals to a single shared meaning. But whether this facilitation reflects a temporary surge of activation or a deeper structural change in the bilingual lexicon remained an open question.

Woumans and Duyck designed an elegant experiment to settle the matter. Sixty-nine Dutch-English bilinguals completed a visual lexical decision task, in which they judged whether letter strings on a screen were real words or not, across five consecutive days. On the first and fifth days, participants were tested in one language, either their native Dutch or their second language English. On the intervening days, they practiced the same task in the other language. Crucially, the word lists included identical cognates, non-identical cognates, non-cognate translation equivalents like “dog” and “hond,” matched control words, and non-words, allowing the researchers to tease apart the contributions of meaning and form.

The logic behind the design was straightforward. If cognate facilitation is driven purely by shared meaning, then translation equivalents, which share meaning but not form, should benefit from cross-language practice just as much as cognates do. If instead the effect depends on the similarity of word forms, then cognates should show a larger transfer of practice benefits across languages than translation equivalents. And if identical cognates share a single mental representation across languages, as some models propose, they should show a disproportionately large jump in performance that non-identical cognates cannot match.

The results told a clear story. Across all word types, participants got faster with practice, reflecting a general procedural learning effect. But when the researchers compared performance before and after practice in the other language, both identical and non-identical cognates showed significantly larger cross-linguistic transfer than translation equivalents or control words. Translation equivalents, despite sharing meaning with their counterparts in the practiced language, showed no additional benefit beyond the general practice effect. This pattern held for both reaction times and accuracy scores.

Perhaps most strikingly, the transfer effects for identical and non-identical cognates did not statistically differ from each other. If identical cognates truly shared a single structural representation in the bilingual lexicon, as the shared orthographic representation account predicts, they should have shown a much larger transfer than their non-identical counterparts. The fact that they did not suggests that the bilingual lexicon does not simply merge identical cognates into one entry but instead relies on graded cross-linguistic similarity effects that operate along tightly coupled orthographic and phonological processing pathways.

These findings challenge the idea that shared meaning alone can explain enduring cognate facilitation. While short-term priming studies have shown that translation equivalents can facilitate each other briefly within a single experimental session, the multi-day paradigm used here reveals that such semantic effects are transient and do not accumulate into lasting cross-linguistic benefits. Instead, the enduring transfer observed for cognates points to structural modifications within the bilingual lexicon that depend on the degree of form overlap between languages.

The authors acknowledge an important limitation: because Dutch and English share the same writing system, orthographic and phonological similarity are correlated in their stimuli, making it impossible to disentangle the two. Future research using cross-script language pairs or systematically varying orthographic and phonological consistency will be needed to determine whether the transfer effects observed here are driven by spelling, sound, or an interaction between the two.

Nevertheless, the study provides compelling evidence that the bilingual mind is not organized into separate language compartments but into a richly interconnected network where form similarity shapes how experience in one language reshapes processing in the other. For the millions of people who navigate multiple languages daily, the message is clear: every word you read in one language leaves a mark on how you will read its relatives in another.

Subject of Research: Cross-linguistic transfer of cognate facilitation in bilingual visual word recognition

Article Title: Enduring cognate facilitation: evidence for form-based transfer in bilingual word recognition

Article References: Woumans, E., & Duyck, W. (2026). Enduring cognate facilitation: evidence for form-based transfer in bilingual word recognition. Frontiers in Language Sciences, 5, Article 1792224. https://doi.org/10.3389/flang.2026.1792224

Image Credits: AI Generated

DOI: 10.3389/flang.2026.1792224

Keywords: bilingualism, cognate facilitation, lexical decision, orthographic representation, cross-language activation, psycholinguistics, word recognition, translation equivalents, Dutch-English bilinguals, form-based transfer, shared representations, language processing