How One Brain Speaks Two Languages: The Discovery of a Single Grammatical Engine

For decades, anyone who speaks more than one language has experienced the occasional cognitive collision: accidentally applying the grammatical rules of one tongue while speaking another. A Spanish-English bilingual might say "I have 20 years" instead of "I am 20," directly translating the structural logic of Spanish into English vocabulary.

These everyday language mashups have long fueled a central debate in cognitive science. Does a bilingual person possess separate, language-specific "grammatical engines" in their brain—one rulebook for English and a completely distinct one for Spanish—that sometimes crash into each other? Or does the brain rely on a single, unified system to process the structural rules of every language a person knows?[1][3]

A landmark study published this week in the Journal of Neuroscience has fundamentally redrawn our understanding of multilingual cognition, providing definitive evidence that the brain operates on a single, shared grammatical engine. Led by researchers at New York University, the study dismantles the "dual-engine myth," revealing that human language is built from abstract neural computations that transcend any specific tongue. The findings suggest that the brain does not waste resources building separate anatomical storage rooms for different languages; instead, it implements grammar as a highly reusable, universal computational loop.[2][4]

"Our research suggests that brains have a single grammatical engine that fuels all of the languages we speak—rather than separate engines for each one," explained Esti Blanco-Elorrieta, an assistant professor of psychology and neural science at NYU and the senior author of the study.

By demonstrating that the exact same brain patterns support grammatical transformations in both English and Spanish, the research team has provided some of the clearest neural evidence to date that grammatical computations are shared across languages. The findings indicate that human language is not a collection of isolated silos, but rather a unified cognitive architecture built from neural computations that transcend any one specific vocabulary.[1][3]

To capture the lightning-fast speed of human speech processing, the researchers utilized high-resolution magnetoencephalography (MEG). This advanced neuroimaging technology maps the exact magnetic fields generated by electrical activity in the brain, allowing scientists to watch grammatical computations unfold millisecond by millisecond. Because language processing happens almost instantaneously, traditional functional MRI scans—which measure blood flow over seconds—are often too slow to capture the precise moment a grammatical rule is applied. MEG provided the temporal resolution necessary to track the brain's structural assembly line in real time.[2][4]

During the MEG scans, the researchers subjected fluent Spanish-English bilingual participants to a morphological stress test. Participants were tasked with instantly transforming singular words into their correct grammatical plural forms across both languages. For example, they would hear the English word "boat" and have to produce "boats," or hear the Spanish word "barco" and produce "barcos." As the participants performed these rapid-fire transformations, the MEG sensors tracked which neural pathways were firing to apply the pluralization rules.[1][3]

The empirical tracking data unmasked an identical, language-transcendent neural template firing across both tongues. The researchers observed that Spanish-English bilinguals engage a shared left frontal-temporal network when producing grammatically appropriate forms, regardless of which language they are speaking. This common neural signature emerges incredibly early during the speech planning process. The physical overlap in brain activity was so precise that researchers could not distinguish whether the brain was processing English or Spanish based solely on the grammatical computation pattern.[2][6]

To ensure that the brain was actually applying a universal grammatical rule rather than simply retrieving memorized plural forms from a mental dictionary, the NYU team introduced a critical control variable: pseudowords. Participants were asked to apply grammatical rules to completely fabricated, made-up words, such as "paple."

Because these words do not exist in any language, the brain could not rely on rote memorization or past experience; it had to actively compute the grammatical structure on the fly. This morphological stress test was designed to isolate the pure computational mechanics of the brain's language center, stripping away the influence of vocabulary familiarity.[3][4]

The results of the pseudoword test were unambiguous. The exact same shared neural mechanism fired when participants pluralized the made-up words, proving that the brain implements grammar as an abstract, reusable template. Furthermore, the researchers tested "cognates"—words that share similar meanings, spellings, and pronunciations across languages due to common roots. Across real words, cognates, and pseudowords, the single grammatical engine remained the consistent driver of structural language processing.[2][3]

These findings represent a massive shift from how neuroscience historically viewed the bilingual brain. Early research often framed bilingualism as a "disruption" or an "add-on" to the processing of a person's native language, operating under the assumption that the brain had to work overtime to manage competing linguistic systems.

Judith Kroll, a psycholinguist at the University of California, Irvine, noted that subsequent studies began to show bilingual brains displaying physical advantages, such as more efficient white matter. This new study takes that understanding further, showing that the brain achieves this efficiency by making its core neural networks do double duty.[1][6]

The discovery of a shared grammatical engine also reframes how we understand the occasional errors bilingual speakers make. When individuals slip and mix up grammatical rules across tongues, it is not because two separate language engines are colliding or misfiring. Instead, it is a byproduct of a highly efficient, unified system handling multiple vocabularies through a single computational pipeline. The brain is simply applying its universal structural template to the active lexicon, occasionally mapping the specific parameters of one language onto the words of another.[3][4]

Independent experts have praised the study's methodological rigor and its broader implications for cognitive science. Mirjana Bozic, a cognitive neuroscientist at the University of Cambridge who was not involved in the research, called the findings "highly informative, providing elegant and convincing evidence that bilingual speakers rely on shared neural mechanisms."

She noted that the results perfectly align with previous initial evidence suggesting that the front left side of the brain is typically involved in processing the grammatical structure of sentences across different languages. The confirmation of a single grammatical engine solidifies theories that have been debated in psycholinguistics for decades.[1][6]

The research, supported by grants from the National Science Foundation and the National Institutes of Health, also holds profound implications for the fields of artificial intelligence and computational linguistics. As engineers build increasingly complex Large Language Models (LLMs), there is an ongoing debate about whether artificial neural networks should compartmentalize different languages or process them through shared parameters. The human brain's reliance on a single, highly reusable computational loop suggests that true multilingual efficiency lies in shared structural representations rather than siloed language-specific modules.[4][5]

Despite the clarity of the MEG data, transparent uncertainty remains regarding how far these findings generalize across vastly different language pairs. The NYU study focused specifically on English and Spanish—two Indo-European languages that, despite their differences, share significant structural and historical overlap. A critical open question is whether the single grammatical engine operates identically when a bilingual speaker navigates languages with fundamentally different architectures, such as English and Mandarin, or languages that rely heavily on tone and complex morphological casing.[1][6]

Additionally, researchers have yet to determine how the age of language acquisition affects the development of this shared neural template. It remains unknown whether individuals who learn a second language late in adulthood integrate it into the exact same grammatical engine as seamlessly as those who are raised bilingual from birth.

The brain's plasticity changes over time, and future studies utilizing the same millisecond-by-millisecond MEG tracking will need to explore these developmental variables to map the full boundaries of the brain's linguistic architecture and understand how age impacts structural language processing.[2][3]

Ultimately, the discovery that one brain uses a single engine to speak two languages offers a profound testament to human cognitive efficiency. By building abstract, reusable neural mechanisms, the brain avoids the metabolic cost of maintaining redundant systems. As researchers continue to decode these shared pathways, the findings promise to offer new insights into how we communicate, how we might better teach new languages, and how we can protect linguistic function in the face of neurological decline.[1][4]