Brain basis of meaning, words, constructions, and grammar

Discussing aspects of the putative brain basis of meaning-bearing units, ranging from phonemes, morphemes and words over word sequences and schematic syntactic structures to "rules of grammar", we here address the neurobiological foundation of linguistics and the putative beneficial impact of language and neuroscience research. Lexical entries are instantiated in the brain as distributed neuronal assemblies spread out over fronto-temporal language areas further extending into motor and sensory regions. This is not just the case for words referring to physical objects, concrete actions and bodily sensations but also for those used to communicate about more abstract ideas. Recurrent word sequences and more abstract constructions generalizing over such specific sequences are also stored in the brain, possibly by processes distinct from word storage as we argued. In a brain-inspired model of grammar, their stored representations take the form of sequence detectors and sequence detector aggregates, DCNAs, respectively. Rules of grammar can then be viewed as emergent properties of multiple stored sequences which are bound together in assemblies as a consequence of string segments being able to substitute for one another in the same structural slot on the basis of their semantic commonality. Such semantically-coloured combinatorial assemblies can further merge together to form brain correlates of highly abstract constructions, such as those licensing the mere sequence of (any) noun and (any) verb.
Apart from discussing the possible emergence of grammar rules (or 'constructions') in the brain, we argue that positing them in the first place is necessary to account for the experimental observation that the brain responds differently to grammatically acceptable phrases than to ungrammatical ones; crucially, unlike what many neural network modellers suggest, the brain's (relatively reduced) immediate response signalling acceptance is not dependent on string probability, since this response to a grammatical and common sequence is not significantly different in strength from the response to an equally grammatical but very uncommon sequence. Rule-conformity rather than string probability is what explains the observed brain responses.
We then provide an answer to the question whether Cognitive and Construction Grammar frameworks are neuroscientifically plausible. Our answer can be summed up as "Yes, with a but". More specifically, it seems correct to treat human language as an extension of other (and older) neurocognitive features, such as the brain's general ability to build and establish links between object and action representations and register sequential events. It also seems correct to treat syntax and semantics as interwoven, as syntactic circuits in the brain emerge from sequences whose segments belong together semantically. It may also be that there are brain correlates of schematic but still inherently meaningful constructional schemata with semantic features bound to them, so that a semantic contribution of a constructional template could be explained. Finally, although the brain correlates of syntactic constructions can range from medium-level to highly abstract (thereby supporting in part the lexicon-syntax continuum tenet), and in spite of the existence of multi-word lexical items, we would warn against a total abolition of a lexicon-syntax distinction, as words and (even common) sequences of words trigger different, in fact opposite, brain responses relative to their erroneous counterparts (pseudowords and illegitimate word combinations, respectively). It may therefore be advantageous to keep observing the traditional distinction between, on the one hand, word-level ('lexical') items and, on the other hand, instantiations of above-word ('syntactic' or 'phrasal') templates in the grammar. Interestingly, the neurophysiological manifestations of the lexicon/syntax distinction may help address specifically linguistic questions with neuroscience tools, as we here exemplify by highlighting the case of verb-particle combinations. Needless to say that this latter strategy requires a firm link between the language of neurons and that used to speak about language, that is, a neurobiologicallinguistic theory that provides the translation and thereby connection of the two worlds.

Publication year:2011

Keywords:grammar, constructions, semantics, phrasal verbs (verb-particle combinations), neuroscience, Mismatch Negativity (MMN)