Intersecting constraints: Why certain constraint-types overlap while others don’t
Marc van Oostendorp, Michael T. Putnam, Laura Catharine Smith · Oxford University Press eBooks · 2016
This chapter presents further arguments in favor of the claim that all computation happens at the same time, demonstrating that such a parallel architecture of human grammar is optimal in design when compared with modular alternatives. A logical consequence is that models that advance a parallel architecture of grammar seem inherently easier to implement in the analysis of syntax–phonology interface phenomena than analyses couched within some version of the derivational inverted T-model. Some level of modularity, which is not explicitly required in such a model of the grammar, is also derived: syntactic and phonological constraints are at least potentially interspersed in the same grammar. It is argued that both consequences are desirable and allow us to derive certain crucial properties of the phonology–syntax “interface”: in particular the fact that both “modules” can see the constituent structure of the other, but for the most part, not its (featural) content.