Language universals in the brain: How linguistic are they?

Anybody’s search for language universals will depend on certain assumptions that are not themselves scientific in the strict sense of the empirical sciences, since they cannot be subjected to experimental testing. These basic assumptions are ontological, as they imply convictions of how those universals might exist, and they are epistemological because their mode of existence will determine how one can find out about them. Although I do not intend to digress into philosophical questions, it is nonetheless necessary at the outset to clarify certain preconceptions that will characterize this chapter. These are physicalist in nature and therefore the information I will provide in the discussions below will be most relevant to those who believe that minds are organized in certain ways because brains are. There are alternative positions one could take regarding universals. For example, to Saussure (1915/1972) universal principles of “langue” were communicative in nature, i.e., derived from social interaction, rather than individual minds or brains.1 In more recent cognitive science, minds have sometimes been likened to software or programs that can be implemented on just about any computational hardware (cf. Fodor, 1976; Gardner, 1987: p78f). The implication of this position would be that some universals of mind may exist without corresponding universals of brain. Conversely, Chomsky – arguably the most prominent thinker in modern linguistics – has insisted that “linguistic universals [are] principles that hold of language quite generally as a matter of biological (not logical) necessity”, which is ultimately based on a “genetically determined initial state” (Chomsky, 1980: p232). The physicalist approach taken in this chapter does not deny the possibility of universal principles of language that are not based on a unique neural architecture. In other words, different speakers (including speakers of different languages) may abide by abstract universal principles, but these principles – though universal descriptively – may be subserved by diverse types of neural processes. This universality could be called shallow because it is not found in neurocognitive architecture, but only in the description of linguistic behavior. Deep universality, on the contrary, would involve neurofunctional principles shared by all typically developing brains. I will argue below that from a developmental neuroscience point of view, it is not very likely that any deeply universal principles are specifically linguistic. Functional domains that begin to develop before the onset of language acquisition (such as visuomotor coordination, polymodal integration, joint attention, working memory) present us with much more probable roots for linguistic universals.