On Pure Syntax (Uncontaminated by Information Structure)

Abstract

Word order correlates with distinctions of information structure, and Gunther Grewendorf’s work has contributed much to our understanding of what the pertinent regularities are in German and other languages such as Italian or Japanese, and it has also shaped our understanding of how these regularities are linked to grammar. Grewendorf (1980) constitutes one of the first concrete proposals of capturing the impact of informational distinctions on German word order. Grewendorf and Sabel (1994, 1999) developed one of the most detailed models of word order variation in the middle field. Grewendorf has also contributed substantially to recent developments concerning the left periphery of clauses, based on the view that categories like focus and topic are directly represented in the syntax. Rizzi (1997) proposed that there are Topic and Focus heads situated in the higher functional layers of the clauses, and that the specifiers of these heads host topic and focus phrases, respectively if not already in the surface representation, then at least at LF. Rizzi’s view has been elaborated for German by Frey (2004), Haftka (1995), Pili (2000), and, of course, Grewendorf (2005a,b). Such approaches in which information structure is directly coded in the syntax constitute one of two extreme ends of a continuum of models of the syntax-information structure interaction. Correlations between positions and informational functions do not necessarily imply that information structure and word order are directly related by syntactic laws. Rather, informational concepts might be correlated with other properties of phrases (say, their length), which could then be the ones really affected by the forces determining word order, such as processability, see, e.g., Hawkins (1994) for such a view.1 In this paper, I will argue that information structure concepts indeed do not play an immediate role in syntax. While word order can reflect information structure categories, it does so because these categories are encoded phonologically in a certain form (or because of the semantic consequences of information structure distinctions) to which