A Statistical Account on Word Order Variation in German

Author: Daniela Kurz

Editor:

In this paper we present a corpus-based study involving the linear order of subject, indirect object and direct object in German. The aim was to examine several hypotheses derived from Hawkins' (1994) performance theory. In this context it was crucial to examine whether and to which extend length influences the order of subject and objects. The analysis was based on data extracted from the annotated NEGRA corpus (Skut et al., 1998) and the untagged Frankfurter Rundschau corpus. We developed an analysis system operating on the untagged corpus that facilitates the acquisition of data and subsequent statistical analysis. We describe this system and discuss the results drawn from the analysis of the data. These results do not support the theoretical assumptions made by Hawkins. Furthermore, they suggest the investigation of other factors than length.