Open the app

The bilingual brain: Arturo Hernandez and why native language is different from foreign language (2013)

The bilingual brain: Arturo Hernandez and why native language is different from foreign language (2013)

Arturo Hernandez is a neuroscientist and cognitive psychologist at the University of Houston who has spent more than two decades studying how two or more languages are stored and used in a single brain. His 2013 book The Bilingual Brain (Oxford University Press) gathers that work around one deceptively simple question: for the brain, is a second language the same kind of thing as the first, or something different? The short answer — and the reason the book matters for anyone learning a language as an adult — is that a native language and a foreign language are partly different objects in the brain, and the difference is not a fixed fact about the languages themselves but a consequence of when and how each was learned.

This entry is about that neurobiology specifically: where a second language sits, why it can sit somewhere slightly different from the first, and what changes as you get better. It is the biological companion to our entry on bilingualism as a phenomenon (its definitions, myths and everyday reality); here we stay under the skull.

Hernandez and the one-or-two-systems question

The oldest debate about the bilingual brain is whether the two languages share a single system or occupy two. Hernandez's answer is neither a flat "one" nor a flat "two" — it is emergentist. He argues that the way a language is represented is not decided in advance by a language "module" but emerges from the interaction of the brain's general learning machinery with the input a person actually gets, at the age they get it. A language learned early, soaked up through the senses and the body, becomes wired into the brain in one way; a language learned later, often through explicit study, gets wired in another. There is no separate "second-language organ" waiting to be filled — both languages recruit the same general-purpose left-hemisphere network that any language uses. What differs is how deeply and how automatically each is embedded in it.

This is why the honest picture is "partly shared, partly separate". The two languages draw on one network and can overlap almost completely; but the degree of overlap is not guaranteed. It is earned, and it depends on two factors Hernandez keeps returning to: the age at which the second language was acquired, and the proficiency the learner reaches. Those two factors pull in interesting directions, and separating them is the heart of the modern story.

Age of acquisition versus proficiency

The single most influential piece of evidence here is an fMRI study by Kim, Relkin, Lee and Hirsch, published in Nature in 1997 under the title "Distinct cortical areas associated with native and second languages". They scanned bilinguals while they silently generated sentences in each of their languages and found a striking split that depended on when the second language had been learned:

  • In Broca's area (the frontal region tied to producing speech), late bilinguals — people who learned their second language in adulthood — showed the two languages sitting in spatially separated, though neighbouring, patches of cortex. Early bilinguals, who acquired both languages in childhood, used essentially the same frontal tissue for both.
  • In Wernicke's area (the temporal region tied to comprehension), there was little or no such separation: both languages drew on overlapping tissue regardless of when the second was learned.

The result was a landmark because it put a physical location on an intuition. For the part of the brain that produces language, a language learned late really does get its own slightly separate real estate — the brain does not simply slot it into the machinery already built for the mother tongue. The comprehension side, meanwhile, is more forgiving and tends to share.

But age is not the whole story, and this is the crucial correction. A decade of later work — much of it by Daniela Perani and Jubin Abutalebi — showed that proficiency is often the stronger factor. In their review "The neural basis of first and second language processing" (2005) and related studies, the pattern is a convergence hypothesis: as a learner's competence in the second language rises, the brain activity for that language converges on the same network used for the first, and the extra regions a beginner leans on quietly drop away. A low-proficiency late learner recruits more brain — wider frontal activation, more effortful control regions — to hit roughly native performance. A high-proficiency learner, even one who started late, increasingly looks native under the scanner. Attained proficiency, in other words, can override much of the age effect on where a second language lives.

So the two findings are not in conflict; they describe different moments. Age of acquisition sets the starting configuration and leaves a durable mark, especially on production and on the sound system (accent is the most age-sensitive layer of all). Proficiency then reshapes that configuration over time, pulling the second language toward the first's network the better you get. The encouraging reading for an adult learner is that the age you started is a starting condition, not a ceiling — the brain keeps rewiring toward native-like organization as competence grows.

Control and switching: the competition between languages

Hernandez uses a memorable image: in a bilingual brain, two languages coexist "almost like two species live in an ecosystem" — active at the same time, sharing resources, but also competing, especially under stress or fatigue. This is not a metaphor for confusion; it reflects a well-documented fact. When a bilingual intends to speak one language, the other does not politely switch off. Both are partly activated in parallel, and the words of the unwanted language have to be actively held back. Managing that competition is a job in its own right, and the brain has dedicated machinery for it.

That machinery is a language-control network, distinct from the language network itself and drawn mostly from the brain's general executive-control system. Work by Abutalebi, Green and colleagues implicates a consistent set of regions: the anterior cingulate cortex (conflict monitoring — noticing when the wrong language is intruding), the dorsolateral prefrontal cortex (suppressing the non-target language and correcting errors), the basal ganglia / caudate (selecting the intended language and driving lexical access), and the inferior parietal lobule (holding the goal — "we are speaking English now" — in mind). David Green's earlier inhibitory control model framed the core mechanism cleanly: to speak language A, the system inhibits language B in proportion to how strongly B is competing.

Two things follow. First, switching languages has a measurable cost — a small time penalty and extra activation in that control network — because the previously suppressed language has to be released and the other reined in. Second, this constant management is the leading candidate explanation for the "bilingual advantage" sometimes reported in executive function: a brain that spends its life resolving language competition may get better at conflict monitoring in general. (That advantage is genuinely debated and smaller and less reliable than early enthusiasm suggested — but the control network it rests on is not in doubt.)

L1 vs L2: what really differs

Pulling the threads together, here is what actually separates a native language from a foreign one in the brain — and, just as importantly, what does not.

  • Automaticity, not location. The biggest difference is not where the languages are but how effortlessly they run. The native language is deeply automatic: during early development the brain specializes — dedicating circuitry to the sounds, grammar and meanings it hears constantly — until using it costs almost no conscious effort. A later language, until heavily practised, leans on more effortful, more consciously controlled processing.
  • The sound system is the most age-bound layer. Phonology — hearing and producing the exact contrasts of a language — is laid down very early and is the hardest thing to fully rebuild later. This is why a late learner can reach flawless grammar and vocabulary and still carry an accent: the perceptual and motor templates for speech sounds are the most age-sensitive part of the whole system.
  • Grammar and vocabulary are more reshapeable. As proficiency climbs, the syntax and lexicon of a second language converge on the native network, becoming more native-like in both behaviour and brain activation. These layers are far more open to adult reorganization than the sound system.
  • The control demand is extra. A monolingual never has to inhibit a competing language; a bilingual always does. That added layer of management is a permanent structural difference between one language and two, regardless of proficiency.

"For the brain, the native language is very different from the second language" is therefore true — but the difference is a gradient set by age and closed by proficiency, not an unbridgeable wall. This mirrors what the classic aphasia studies of Broca and Wernicke first revealed about language having a physical seat in the cortex: both of your languages live in the same neighbourhood, but a late-learned one has to work its way in.

What this means for learning a language

The neurobiology is not just interesting — it prescribes how an adult should study. The recurring theme is that a late-learned language starts in effortful, consciously-controlled memory and has to be driven toward the automatic, native-like organization the brain reaches through practice.

  • Expect to learn through explicit memory first — then automate it. Michael Ullman's declarative/procedural account maps neatly onto all of this: an adult tends to route a new language, grammar included, through declarative memory (the same system that stores facts) before the grammatical machinery migrates into fast, unconscious procedural memory. Knowing a rule is only the first stage; the goal is to make it run without thinking. That shift is exactly the move from declarative to procedural memory, and it happens through repeated meaningful use, not more explanation.
  • Chase proficiency, not youth. Since proficiency is what pulls a second language onto the native network, the useful lever is not the age you started (you cannot change it) but the depth of practice you put in. High usage and high competence are what make the brain treat the language as native-like.
  • Train production directly. The age effect bites hardest on the output side — the frontal, speech-producing machinery Kim's study found partly separated for late learners. That circuitry is built by actively generating language, not by recognizing it, which is why the Taalhammer method has you produce whole sentences from meaning rather than merely review them.
  • Attack the sound system early and deliberately. Because phonology is the most age-sensitive layer, accent and listening deserve focused, early attention — lots of listening to build the perceptual templates, and lots of speaking to build the motor ones. Waiting until grammar is "finished" leaves the hardest layer for last.
  • Don't fear the two-language competition — it's normal. Interference and the odd mid-sentence intrusion are the control network doing its job, not a sign of failing. The more you use the target language in real contexts, the better that network gets at keeping it in front. On the critical-period question specifically — how much age actually constrains a late start — see the critical period.

The bilingual brain does not build a second, separate language organ. It takes the one language network it already has and teaches it to run a second system — a system that starts out effortful and slowly, with proficiency, becomes native-like. The age you began sets where that journey starts; the practice you put in decides how far it goes.

Frequently asked questions

Are two languages stored in the same place in the brain?

Largely yes — both a first and a second language recruit the same distributed left-hemisphere network, so there is no separate "second-language organ". But the overlap is not automatic. Kim et al. (1997) found that in Broca's area, a language learned in adulthood occupies a slightly separate patch from the native language, while a language learned in childhood shares the same tissue; comprehension regions overlapped regardless of age. As proficiency rises, the second language increasingly converges onto the native network. So the two languages share one system, with the degree of overlap set by age of acquisition and proficiency.

Does it matter more what age I started or how good I get?

Both matter, but for adult learners the more encouraging factor is proficiency. Age of acquisition leaves a durable mark — especially on pronunciation, the most age-sensitive layer. But research by Perani, Abutalebi and others shows that as competence grows, brain activity for the second language converges on the native-language network, and much of the early age-related difference fades. Attained proficiency is often a stronger predictor than starting age of how native-like a second language becomes in the brain. The age you started is a starting condition, not a ceiling.

Why does my native language feel so much more automatic than my foreign one?

Because your brain specialized for it during early development, dedicating circuitry to its sounds, grammar and meanings through years of constant use until the language runs with almost no conscious effort. A later language, until heavily practised, leans more on effortful, consciously-controlled processing and on declarative (fact-like) memory rather than fast procedural memory. The gap closes with use: the more you actively produce the language, the more its grammar migrates into automatic processing and the more native-like it feels.

Sources

Related