Is language processing organized in any way? As already discussed in the 2015-2016 lecture, the nucleus of language areas verifies several modularity criteria (Fodor, 1983) :
- Its neural architecture is fixed and reproducible from one individual to another, it has specific anatomical connections and even a distinct neurotransmitter receptor density profile.
- Its development is rapid, instinctive and specific to the human species (properties of the language acquisition device postulated by Chomsky).
- It appears to be largely specialized for language, as opposed to other brain systems organized in a more " horizontal " or " isotropic " way, which are involved in multiple cognitive tasks.
- Linguistic information seems to be encapsulated : the core network of language areas processes only a fraction of the available information, and responds identically, for example, to normal and " Jabberwocky " sentences (from which semantic information has been removed).
- It processes information automatically, inaccessible to consciousness. The lecture developed this last point by reviewing all the available data on unconscious language processing. Recent data, obtained as part of Lucie Berkovitch's thesis, suggest that access to a syntactic lexicon (which determines the category of words, e.g. noun or verb) occurs even when the word is rendered invisible by masking. Thus, the earliest and simplest stages of syntactic processing can be carried out non-consciously, whether it's grammatical categorization or the identification of syntactic features (e.g. singular or plural). In addition, the brain responds non-consciously to syntactic (ELAN wave) or semantic (N400 wave) violations, although perhaps only if the words are sufficiently close in time (there would thus be a dependency between working memory and awareness). Brain imaging shows that the temporal cortex continues to respond to unconscious sentences, even during sedation, while Broca's area appears to cease responding.
