Iris Alexia Hernández-González (1) y Jennifer Balade (2)
(1) Dept. de Didácticas Específicas, Universidad de La Laguna, Tenerife, España
(2) Dept. de Psicología Evolutiva y de la Educación, Universidad de La Laguna, Tenerife, España
(cc) Iris Hernández.
Human language is the result of the interaction between biological mechanisms, learning and brain organisation, without depending on isolated centres such as those postulated by Broca and Wernicke. Current evidence shows that a network distributed across frontal, temporal, and parietal areas, coordinated with perceptual, motor, and cognitive systems, allows internal representations to be transformed into words and generate meaning. This approach reveals that language is a dynamic process, where neural specialisation and plasticity combine for linguistic acquisition, processing, and production.
How do humans manage to transform thoughts into words? This mystery has fascinated the scientific community for centuries. Human language develops on a biological basis that enables its acquisition and use, as demonstrated by solid evidence such as early language acquisition even in contexts with limited stimuli (Macías et al., 2025), the existence of a critical period for its development (Lopera, 2007), and the identification of genes involved in general neurodevelopmental processes such as neuronal differentiation, migration, and connectivity, which indirectly affect the sensorimotor circuits necessary for speech (Fisher, 2017). However, not all aspects of language are innate. For example, writing is a cultural invention that requires systematic teaching and practice to master (Cuetos et al., 2020). In this context, Christiansen and Chater (2008) argue that what is inherited is not a set of specific grammatical rules, but rather general cognitive biases that guide learning. The authors argue that the hypothesis of an innate, biologically determined Universal Grammar lacks evolutionary viability, as language changes too rapidly for its supposed arbitrary regularities to have become genetically fixed. From this perspective, language has evolved culturally to fit the general limitations and predispositions of the human brain, and not the other way around, which would explain the remarkable efficiency with which children acquire their mother tongue. Large-scale language models, such as ChatGPT, provide empirical evidence that it is possible to generate complex grammatical language from massive exposure to linguistic data, without the need for a pre-specified innate grammar (Contreras-Kallens et al., 2023). These models, despite lacking semantic understanding and human social abilities, systematically produce grammatical sequences, which is proof of concept of the power of simple statistical learning to capture complex linguistic regularities.
To understand how language is organised in the brain, a classic model has historically been used that distinguished between a production centre (Broca’s area) and a comprehension centre (Wernicke’s area), connected by the arcuate fasciculus (Geschwind, 2010); a definition that is more anatomical than functional (Fedorenko et al., 2024). Although this model made a decisive contribution to neuropsychology, current data show that it is insufficient. Recent research (see review by Fedorenko et al., 2024) has revealed that language is not based on isolated modules, but on a widely distributed functional network that includes frontal and temporal areas of the left hemisphere (a hemispheric predominance that remains a subject of debate; Corballis, 2008; Sha et al., 2021). Additionally, Turker et al. (2023) showed results indicating that different components of language (e.g., semantics, syntax, or phonology) activate partially differentiated patterns within this distributed architecture, involving not only the frontotemporal core but also parietal, subcortical, and cerebellar regions.
This linguistic network can be easily identified through tasks that contrast linguistic stimuli with perceptually similar but meaningless stimuli, such as pseudowords (e.g., barifo or tomeru). Its topography is relatively stable in each person, although it varies anatomically between individuals, highlighting the importance of locating it functionally on a case-by-case basis. In their review, Fedorenko et al. (2024) concluded that this network functions similarly regardless of modality and operates in both comprehension and production; that is, the same frontotemporal areas are activated when listening to, reading, or producing spoken and written language, invalidating the classic separation between areas of comprehension and areas of production. Furthermore, this network is highly selective because it responds robustly to words and sentences but shows little activity in response to music, mathematics, logical reasoning, or demanding cognitive tasks.
Although coordinated with other brain systems, this network remains functionally distinct from them. The areas responsible for speech perception, for example, are located in the superior temporal cortex and process only the acoustic properties of the signal, regardless of whether or not it has meaning. Similarly, the areas responsible for articulation plan and execute movements, but do not represent meanings or linguistic structures. In addition, reading also depends on a specialised perceptual system called the word-shape visual area in the left occipitotemporal cortex, which responds to the visual shape of words but not to their meaning (McCandliss et al., 2003). All of these are perceptual and motor systems that feed into and receive information from the language network but are not part of it.
The question is: how does the human brain manage not only to process words, but also to understand narratives, construct mental models and interpret social context? Language processing does not end at the linguistic network but depends on its interaction with higher-order cognitive systems, such as the default network or the theory of mind network. When reading a sentence such as ‘Lucía arrived late at the airport and when she saw the screen she started running,’ the linguistic network processes the words, but the default network integrates the context, activates knowledge about how airports work, and allows us to infer that she is afraid of missing her flight. As can be seen, these networks integrate linguistic content with prior knowledge, social inferences, and the overall coherence of discourse.
Understanding how we speak, listen and think therefore requires abandoning the idea of isolated centres. The contemporary view suggests that language emerges from a neural symphony in which a highly specialised network decodes and encodes meanings, while other perceptual, motor and cognitive networks contribute to deploying language in the physical and social world.
References
Christiansen, M. H., y Chater, N. (2008). Language as shaped by the brain. Behavioral and Brain Sciences, 31, 489-509.
Contreras-Kallens, P., et al. (2023). Large language models demonstrate the potential of statistical learning in language. Cognitive Science, 47, 1-6.
Corballis, M. C. (2008). The gestural origins of language. En N. Masataka (Ed.), The Origins of Language: Unraveling Evolutionary Forces (pp. 11–23). Springer Science + Business Media.
Cuetos, F., et al. (2020). Psicología del Lenguaje. Editorial Médica Panamericana.
Fedorenko, E., et al. (2024). The language network as a natural kind within the broader landscape of the human brain. Nature Reviews Neuroscience, 25, 289-312.
Fisher, S. E. (2017). Evolution of language: Lessons from the genome. Psychonomic Bulletin & Review, 24, 34-40.
Geschwind, N. (2010). Disconnexion syndromes in animals and man: Part I. Neuropsychology Review, 20, 128-157.
Lopera, F. (2007). Procesamiento cerebral de las palabras y su impacto en los procesos de conocimiento. Páginas: Revista académica e institucional de la UCPR, 79, 5-30.
Macías, F. M., et al. (2025). Estimulación temprana y su influencia en el desarrollo del lenguaje oral. Pedagogical Constellations, 4, 239-261.
McCandliss, B. D., et al. (2003). The visual word form area: Expertise for reading in the fusiform gyrus. Trends In Cognitive Sciences, 7, 293–299.
Sha, Z., et al. (2021). The genetic architecture of structural left-right asymmetry of the human brain. Nature Human Behaviour, 5, 1226–1239.
Turker, S., et al. (2023). Cortical, subcortical, and cerebellar contributions to language processing: A meta-analytic review of 403 neuroimaging experiments. Psychological Bulletin, 149, 699-723.
Manuscript received on November, 13th, 2025.
Accepted on December 11th, 2025.
This the English version of
Hernández-González, I. A., y Balade, J. (2026). El cerebro que habla: La biología del lenguaje humano. Ciencia Cognitiva, 20:1, 1-3.
