It has repeatedly been noted in the literature (Crysmann, 2000a; Spencer, 1991, among others) that weak pronominals in European Portuguese present some diverging evidence as to their status as lexical affixes or postlexical clitics: while their placement properties w.r.t. their host are certainly governed by syntactic and semantic properties (e.g. precedence of a filler or a downward monotone quantifier Crysmann, 1999), they also exhibit a number of properties characteristic of lexically attached formatives: they display arbitrary gaps, both in the set of clitic combinations and in the combinations between clitic and host, they sometimes give rise to affix fusion, they are subject to rigid and idiosyncratic ordering, they involve semantic idiosyncrasies, they can be infixed between inflectional affixes and the stem, and they also display a number of morphophonological idiosyncrasies, including haplology.
Within recent work in HPSG (Crysmann, 2000a,b), it has been suggested that this tension can be resolved by appealing to the distinction between tectogrammatical signs and phenogrammatical domain objects (dom-obj) (Kathol, 1995): while clitic-verb complexes are represented as a single lexical sign, the phonological contributions of the host and the clitic cluster are represented on two separate domain objects, making the cluster as a whole available to syntactic ordering. Morphological and semantic idiosyncrasies, however, are effectively encapsulated in the lexical component.
Concentrating on the phonology of function words in EP, Vigario (1999a,b) has argued that despite the morphophonological idiosyncrasies displayed by pronominal affixes, their behaviour w.r.t. lexical and phrasal phonological rules sets them apart from other inflectional affixes. She therefore assumes that EP weak pronominals attach to their hosts postlexically, adjoining to (proclisis) or incorporating into (enclisis) the prosodic word of the host. In this paper, I will show that the distinction between lexical sign's and word-level domain-objects already provides a comfortable basis for an account of these phenomena as well. The analysis will be couched in terms of Declarative Phonology, drawing on insights gained in Bird (1995); Bird and Klein (1994), and Walther (1999): in particular, I will assume a surface-oriented, partially lexicalised approach to prosodisation.
[Back to home page]
[email: crysmann@coli.uni-sb.de]
[File formats]