An emergentist view on functional classes

Building on Manzini 2014, Manzini & Savoia 2014, this paper aims to challenge the hypothesis that function words fall into classes. Furthermore, I show that the distribution and behaviour of Romance pronouns does not provide conclusive evidence to the claim that so-called classes result from the internal make-up of function elements. The distinction between classes of function words is often conceptualised in terms of inner syntax: strong elements are conceived as extended phrases, while clitics correspond – at least in the latter stage of their derivation – to a deprived structure, possibly to a single head exhibiting an affixlike behaviour. The correlation between the behaviour of function words and their syntactic makeup was advanced by Kayne 1975, who argued that clitics are heads inasmuch as they cannot be coordinated, focused, modified, used in isolation, etc. Kayne 1983 argued that certain clitics – noticeably, French subject clitics – are in fact phonological clitics as they show cues of phrasal behaviour. The status of phonological clitics has been revised in comparison to the status of Germanic weak pronouns, e.g. German es, which cannot be coordinated, modified, etc., although they are not bound to a specific host or to a dedicated syntactic position (see Holmberg 1986, 1991 a.o.). The Germanic data led Cardinaletti 1991, 1994, 1998; Cardinaletti & Starke 1996, 1999 to a more articulated typology of deficient elements by individuating a third class of pronouns, which Cardinaletti and Starke term weak. Interand intralinguistic variation follows from the distribution of pronominal forms across the three classes, as exemplified in the following table, which illustrates the status of certain Italian and German pronouns (from Cardinaletti & Starke 1996: 27, 29):