Summary: New research reshapes our understanding of the bilingual brain, showing that multilingual speakers do not store separate grammatical rulebooks for each language. Instead, a single shared neural mechanism—observable with high-resolution brain imaging—supports grammatical computation across languages.
Using magnetoencephalography (MEG) to record brain activity on a millisecond timescale, researchers at New York University tracked Spanish–English bilingual participants as they performed immediate grammatical transformations on real words, cognates and invented pseudowords. The fine-grained neural recordings revealed the same underlying computational pattern across both languages, indicating that grammar is handled by a reusable, language-general neural process rather than by multiple language-specific engines.
- Dismantling the dual-engine theory: Although bilingual speakers sometimes inadvertently apply rules from one language while speaking another, the neural evidence shows these slips arise from a shared grammatical system rather than from separate, competing systems.
- High-temporal-resolution measurement: The team used MEG to capture the brain’s magnetic fields with millisecond precision, allowing them to observe the exact timing of grammatical computations as participants generated plural forms and integrated them into speech planning.
- Behavioral task: pluralization: Participants heard singular nouns in English or Spanish and were asked to produce the correct plural forms (for example, turning “boat” into “boats” or “barco” into “barcos”), a task that forces real-time application of morphological rules.
- Pseudoword controls: To rule out reliance on memorized word forms, the study included made-up words such as “paple.” The same neural pattern emerged when participants pluralized pseudowords, demonstrating that the brain applies an abstract grammatical formula to novel items.
- A reusable computational template: Across different sounds, cognates and fabricated words, the left-lateralized fronto-temporal network activated in a consistent way, supporting the idea of a language-general computation for grammatical transformations.
- Implications for learning and cognition: If grammar is supported by a common neural mechanism, acquiring additional languages may rely more on supplying new vocabulary to an existing computation than on building entirely new grammatical systems.
- Funding: This work received federal support from the National Science Foundation and the National Institutes of Health, emphasizing its relevance to linguistics and cognitive health research.
Source: NYU
Everyday bilingual slips
It is common for bilingual speakers to momentarily borrow syntax from one language when speaking another (for example, saying “I have 20 years” under influence of a second language). Such occurrences prompted the question: do bilinguals maintain distinct neural “grammatical engines” for each language? The NYU study shows this is not the case. Instead, a single neural computation supports grammatical operations across both English and Spanish.
Led by Esti Blanco-Elorrieta (assistant professor of psychology and neural science at NYU) with first author Xuanyi Jessica Chen, the study appears in Journal of Neuroscience. The researchers designed a task that required rapid grammatical computation: listeners heard singular nouns and had to produce the plural, while MEG recorded the brain’s millisecond dynamics. The experiment included true words, cognates and pseudowords to test how the brain handles familiar and novel forms.
Results showed a left-dominant fronto-temporal network engaged roughly 100 ms after the cue to generate the correct inflected form. Multivariate analyses demonstrated that the neural signature of this computation generalized across English and Spanish, across different pluralization patterns, and even to pseudowords—supporting the conclusion that the brain implements grammatical transformations as abstract, generative operations rather than as language-specific rule sets.
“Our data indicate that a single grammatical engine fuels the languages we speak,” says Esti Blanco-Elorrieta. “The same brain patterns support grammar in English and Spanish, suggesting that core grammatical computations can transcend any single language’s surface forms.”
Key Questions Answered:
A: The study suggests these slips occur because the brain runs all vocabulary through the same computational loop. Because a single neural mechanism serves multiple languages, features of one language can briefly influence production in another when processing demands are high or when contextual cues favor one form over another.
A: Magnetoencephalography (MEG) records magnetic fields generated by neural activity with millisecond precision. Language computations occur extremely rapidly, so MEG was necessary to observe the exact timing and pattern of neural activity involved in forming grammatical word forms and to show that the same pattern appears across languages.
A: Pseudowords rule out the possibility that neural responses simply reflect retrieval of memorized whole words. When participants applied grammatical rules to invented forms, the same neural pattern was observed, indicating an abstract grammatical computation at work rather than lookup of stored items.
Editorial Notes:
- This article was edited by an editor at Neuroscience News.
- The journal paper was reviewed in full by the reporting team.
- Additional editorial context was provided by staff.
About this language and neuroscience research news
Author: James Devitt
Source: New York University (NYU)
Contact: James Devitt
Image: Image credited to Neuroscience News
Original Research: Open access. “A Shared Neural Mechanism for Abstract Grammatical Computations Across Languages in Bilinguals” by Xuanyi Jessica Chen and Esti Blanco-Elorrieta. Journal of Neuroscience. DOI: 10.1016/j.scib.2026.02.053
Abstract
A Shared Neural Mechanism for Abstract Grammatical Computations Across Languages in Bilinguals
A central question in cognitive neuroscience concerns how the brain carries out abstract computations that must generalize across superficially different inputs. Language offers a strong test: the same grammatical operation—such as pluralization—can take different surface forms across languages. The critical issue is whether such transformations are implemented by language-specific neural systems or by abstract mechanisms that generalize across linguistic contexts.
Using MEG, the researchers tracked millisecond-by-millisecond neural dynamics as highly proficient Spanish–English bilinguals produced singular and plural noun forms in both languages. The experimental design separated semantic number, phonological change, grammatical inflection and produced language. Adjusting words to their grammatical context engaged a left-lateralized fronto-temporal network beginning around 100 ms after the cue. Multivariate decoding showed that the neural patterns supporting this computation generalized across languages, across different plural forms, and to pseudowords, demonstrating that equivalent grammatical operations are instantiated in the same neural substrates despite surface differences. These findings provide time-resolved evidence that grammatical transformations rely on a language-general computational mechanism and illustrate how bilingualism can illuminate general principles of neural organization.
Funding: Supported by grants from the National Science Foundation (BCS-Grant 2446452) and the National Institutes of Health (R00 DC019973-01).