Abstract
A long-standing puzzle in the analysis of coordination is the so-called SGF coordination (Subject Gap in Finite/Fronted constructions) in German, first discussed by Höhle (1983). The syntactic analysis of SGF constructions is challenging for any type of syntactic framework, as they seem to violate basic assumptions of accessibility or distribution in coordination constructions.
SGF constructions have been analysed in terms of asymmetrically embedded constituents (Wunderlich 1988, Höhle 1990, Heycock and Kroch 1993, Büring and Hartmann 1998) or symmetric conjuncts (Steedman 1990, Kathol 1995, 1999). Asymmetric embedding is problematic as it involves extraction asymmetries, or an analysis of coordination as adjunction. Symmetric analyses need to assume special licensing conditions that are not independently motivated. In particular, we argue that the symmetric analysis of Kathol (1999) is lacking independent syntactic motivation, and fails to account for related asymmetric coordinations of verb-last and verb-fronted (VL/VF) sentences.
We present a multi-factorial LFG analysis of asymmetric coordination, building on independently motivated principles of correspondence between c--structure, f--structure, and i(nformation)--structure. SGF coordination is analysed as symmetric coordination in c--structure. Binding of the (prima facie) inaccessible subject of the first conjunct is enabled, at the level of f--structure, by asymmetric projection of a "grammaticalised discourse function (GDF)", a TOPIC, FOCUS or SUBJ function (Bresnan 2001). Asymmetric GDF projection is motivated by relating the semantic and discourse-functional properties of asymmetric coordination to well-known discourse subordination effects of modal subordination (Frank 1997, Frank and Kamp 1997). In conjunction with word order constraints in the optimality model of Choi (2001), our analysis explains the mysterious word order constraints of asymmetric coordination, and some puzzling scoping properties.