Phrasal typology and the interaction of topicalization, wh-movement, and extraposition

(6) Whoi do you know [thati you can really trust]? (7) Which argumenti do you know [thati Sandy thought was unconvincing]? One could assume that the contrast illustrated in (1) to (5) can be accounted for by considering the grammaticalized discourse functions of the dislocated phrases. In the ungrammatical cases in (2) and (5), the dislocated phrase can be considered a grammaticalized topic – hence the term topicalization. In the grammatical cases in (6) and (7), the dislocated phrases can be analyzed as the focus of the sentence. In section 1 below, however, we will show that such an approach is seriously flawed. Instead, we will suggest that the contrast between (6) and (7) on the one hand, and (2) as well as (5) on the other hand, can be derived from the interaction of two assumptions. The first assumption concerns the typology of phrases in HPSG. Pollard/Sag (1994:391) assume that subjects are realized as subject daughters, and that topicalized phrases, wh-subjects, as well as whobjects are realized as filler daughters. Contrary to this assumption, we assume that wh-moved object daughters and ‘ordinary’ subject daughters in English are specifier daughters, and not filler daughters, or subject daughters, respectively. Topicalized phrases, however, are analyzed as filler daughters. An important consequence of this move is that subjects and wh-moved objects are locally related to a lexical head, viz. to the head whose specifier they are. Topicalized phrases are not locally related to a