Function without content

In this paper we provide new evidence for the assumption that the content of functional categories should be dissociated from its function (Ritter & Wiltschko 2009, 2011). In particular, we discuss evidence from lexicalization patterns. On standard assumptions, functional categories are intrinsically associated with content; consequently the exponents that spell out these functional categories would also be expected to be intrinsically associated with content. We argue that na spells out INFL before it is associated with any kind of content. This allows for a straightforward explanation of the seemingly erratic distribution of na. We further explore the consequences of this analysis and we provide evidence from the Down Syndrome grammar.