Understanding language processing & the brain

Language comprehension and production are feats of real-time neural computation: within 200 milliseconds of hearing a word, the brain has activated its sound, meaning, and grammatical properties simultaneously. This theme covers how the brain converts continuous acoustic streams into meaning — from phoneme perception and word recognition through syntactic parsing and pragmatic inference — and where these processes live in the cortex.

01

McGurk effect

Speech perception is audiovisual: /ba/ + /ga/ lip movements = /da/ — the brain integrates auditory and visual information automatically.

02

Garden-path sentences

"The horse raced past the barn fell" — the parser commits to the simplest reading, then must expensively reanalyse.

03

Broca's aphasia

Non-fluent, effortful speech with good comprehension — caused by damage to left inferior frontal gyrus (Broca's area, BA 44/45).

Paul Broca

1824–1880

Identified left inferior frontal gyrus as critical for speech production via the case of "Tan" (1861); founded cortical localisation of language.

Carl Wernicke

1848–1905

Described the posterior superior temporal gyrus as critical for language comprehension (1874); his model gave rise to the Wernicke-Geschwind framework.

Alan Baddeley

1934–

Proposed the phonological loop component of working memory (1974, 1986) — the subsystem that maintains phonological strings for language processing and word learning.

Harry McGurk

1936–1998

Demonstrated audiovisual speech integration with the McGurk effect (1976) — /ba/ + /ga/ lip movement = perceived /da/ — showing speech perception is multimodal.

Andrew Marslen-Wilson

1945–

Developed the Cohort Model of spoken word recognition (1987); demonstrated that listeners identify words before the final phoneme.

Gregory Hickok & David Poeppel

Developed the dual-stream model of auditory language processing (2000, 2007) — ventral (meaning) and dorsal (sensorimotor) streams — based on neuroimaging and aphasia data.

Categorical perception

Phonemes are perceived as discrete categories despite continuous acoustic variation — voice-onset time (VOT) is heard as either /b/ or /p/, never intermediate.

Cohort model

At each phoneme, all matching word-candidates are activated simultaneously; they are progressively eliminated until the uniqueness point — where one word remains.

Semantic priming

Prior exposure to a semantically related word (e.g., "doctor") speeds recognition of a target word (e.g., "nurse") — evidence for associative organisation of semantic memory.

Garden-path effect

Incremental syntactic parsing commits to the simplest reading; sentences that require reanalysis (e.g., "The horse raced past the barn fell") cause processing difficulty.

Phonological loop

A component of Baddeley's (1986) working memory model that temporarily stores phonological information — critical for sentence comprehension and new word learning.

Grice's cooperative principle

Speakers follow maxims of quantity, quality, relation, and manner; violations generate conversational implicatures — what is communicated beyond what is literally said.

Broca's area

Left inferior frontal gyrus (BA 44/45) — damage causes non-fluent speech with relatively preserved comprehension (Broca's aphasia); also involved in syntax, phonological WM, and hierarchical structure.

Wernicke's area

Left posterior superior temporal gyrus — damage causes fluent but meaningless speech (jargon, paraphasias) with severely impaired comprehension (Wernicke's aphasia).

Arcuate fasciculus

White-matter tract connecting Wernicke's and Broca's areas; damage causes conduction aphasia — fluent speech, intact comprehension, severely impaired repetition.

Dual-stream model

Hickok & Poeppel (2007): ventral stream (temporal lobe) maps sound onto meaning ("what"); dorsal stream (parietal-frontal) maps sound onto motor articulation ("how").

Conduction aphasia

Disconnection syndrome from arcuate fasciculus damage: fluent spontaneous speech, good comprehension, but severely impaired repetition.

Language lateralisation

Language is predominantly left-hemisphere dominant in ~95% of right-handers; right hemisphere contributes prosody, discourse coherence, and pragmatic inference.

What is the difference between Broca's and Wernicke's aphasia?+

Broca's aphasia results from damage to the left inferior frontal gyrus. Speech is non-fluent, effortful, and telegraphic ("want… coffee… now"), but comprehension is relatively preserved. Wernicke's aphasia results from damage to the left posterior superior temporal gyrus. Speech is fluent with normal prosody but contains paraphasias (word substitutions), neologisms, and jargon; comprehension is severely impaired. A useful distinction: Broca's patients know what they want to say but cannot produce it; Wernicke's patients produce words but do not know what they are saying or understand what is said to them.

What does the dual-stream model add beyond the Broca/Wernicke framework?+

The Wernicke-Geschwind framework predicted sharp dissociations corresponding to discrete lesion sites. The dual-stream model (Hickok & Poeppel, 2007) better accounts for neuroimaging data showing distributed, overlapping activation. The ventral stream (superior and middle temporal gyrus → ventral frontal lobe) handles sound-to-meaning mapping and is partially bilateral; the dorsal stream (posterior superior temporal cortex → parietal → premotor/IFG) handles sensorimotor mapping for articulation, phonological WM, and repetition. This explains why Broca's area activates during comprehension of complex syntax (it is part of the dorsal stream's hierarchical processing role, not just a speech motor region) and why lesion outcomes are more variable than the classic model predicted.

Why is the McGurk effect important for theories of speech perception?+

The McGurk effect (McGurk & MacDonald, 1976) demonstrates that speech perception is not a passive acoustic process — it is an active, multimodal computation. The brain integrates articulatory visual information (lip movements for /ga/) with auditory input (/ba/) and produces a percept (/da/) that matches neither source alone. The effect is automatic and survives knowing about it, demonstrating that audiovisual integration occurs early in perceptual processing, before conscious awareness. This undermines purely auditory accounts of speech perception and supports motor theories (Liberman's motor theory; Galantucci et al., 2006) in which perceived phonemes are abstractions over articulatory gestures.

Last reviewed July 2025
  1. 1.

    Friederici, A. D. (2011). The brain basis of language processing: From structure to function. Physiological Reviews, 91(4), 1357–1392.

    +About this source

    Comprehensive review of the neural infrastructure supporting language comprehension and production, integrating lesion and neuroimaging data.

  2. 2.

    Hickok, G., & Poeppel, D. (2007). The cortical organization of speech processing. Nature Reviews Neuroscience, 8(5), 393–402.

    +About this source

    Introduced the dual-stream model — ventral (meaning) and dorsal (sensorimotor) pathways for language processing.

  3. 3.

    McGurk, H., & MacDonald, J. (1976). Hearing lips and seeing voices. Nature, 264(5588), 746–748.

    +About this source

    Reported the McGurk effect — audiovisual integration in speech perception — demonstrating that speech perception is multimodal.