A focused constituent contains the most prominent stress of the clause (Selkirk 1984, Reinhart 1995). Reinhart accounts for this by a PF/LF mapping rule. I extend this view to Hungarian, a language with contrastive focus movement, and show that a range of data, some of which pose a problem for a featuredriven approach, can be accounted for straightforwardly. Among these are the uniqueness of focus movement and the fact that verb-focusing does not strand the particle of particle-verbs (verb-movement generally strands it). The analysis extends to blocking effects between focusing and a phenomenon called particle climbing. It is concluded that the [+Focus]-feature is unnecessary to account for the data. Finally, it is shown that the alleged identificational focus vs. new information focus distinction of É.Kiss (1998b) is not supported by Hungarian data.
Download Full PDF Version (Non-Commercial Use)