Harmonic alignment and the hierarchy of pronouns in German

Abstract

The phonological feature [±stress] distinguishes between strong pronouns (which are capitalized here and henceforth) and others. The semantic feature [±animate] distinguishes between what I call unstressed pronouns and weak pronouns (cf. Cardinaletti & Starke (1996)). The phonological feature [±reduced] distinguishes between weak pronouns and reduced pronouns; the only reduced pronoun in Standard German is es. Finally, the phonological feature [±clitic] distinguishes between reduced pronouns and clitic pronouns. In contrast to what is the case with the other pronoun classes, the syntactic behaviour of clitic pronouns is mainly governed by phonological (rather than syntactic) constraints related to the distinction between proand enclisis, properties of potential clitic hosts, and so on (cf. Gartner & Steinbach (2000), Zifonun (2000)). In what follows, I abstract away from this last class. Table T0 shows how the remaining four pronoun classes behave with respect to the four syntactic operations mentioned above. Interestingly, the pattern is not arbitrary, but obeys strict implicational generalizations: If a given pronoun undergoes Wackernagel movement, all weaker pronouns also undergo this operation; if a given pronoun undergoes R-pronoun formation (optionally or obligatorily), then all weaker pronouns do so as well; and if a given pronoun resists coordination and topicalization, all weaker pronouns share this property. The main goal of this paper is to account for this pattern. The basic assumption is that for all four pronoun classes there is a constraint that forces a pronoun to show up in the Wackernagel position. All these Wackernagel constraints can in principle be