Heaviness and Constituent ordering : a Corpus-based study in Persian

International audienceThis paper presents a corpus-based study of the relative order between the direct object (DO) and the indirect object (IO) in the preverbal domain in Persian, a SOV language. The relative weight is one of the main factors stipulated in explaining constituent order preferences, namely several studies have pointed out the tendency to postpone heavy constituents (heavy NP shift). This weight effect is either accounted for in terms of parsing (Hawkins, 1994) or in terms of planning and production (Wasow 1997, Arnold et al. 2000, Stallings et al.1998, among others), planning or comprehending long and/or complex constituents requires more memory or resource than short and simple ones. Since this explanation lies on general principles of cognition, it has been sometimes assumed that “short-before-long” is universal. However, building on extensive data from typologically different languages, Hawkins have highlighted an asymmetry between VO and OV languages. The latter display the mirror image preference, i.e. “long-before-short”. Empirical corpus-based and experimental studies on Japanese, a consistent head-final language, have indeed confirmed this generalization (Yamashita & Chang, 2001). Our study shows a similar effect in Persian; however, relative length plays a secondary role, since the position of the DO mainly depends on its degree of determination