Where to place a phrase? An informational and generative approach to phrasal extraposition

Abstract

In the following paper, we aim to cast light on the placement of prepositional phrases (PPs) in the so-called postfield, the position behind the right sentence bracket. Our focus is on the period of early New High German from 1650 to 1900. In a first step, extraposition will be correlated with Information Density (’ID’, Shannon 1948). ID is defined as “amount of information per unit comprising the utterance” (Levy & Jaeger 2007: 1). It can be calculated as surprisal. The higher the surprisal values the higher the impact on working memory and the more likely perceiving difficulties become (e.g. Hale 2001). We expect PP with such high surprisal values to be more likely to be placed in the postfield where more memory capacities are available than in the middle field. We test this hypothesis on a corpus of scientific articles and monographs dealing with medicine and theology and taken from the Deutsches Textarchiv (DTA, BBAW 2019). We only find evidence for the hypothesis in the timespan from 1650 to 1700 and for the rare case that attributive PPs are placed in the postfield. Since this has already been shown for attributive relative clauses (Voigtmann & Speyer 2021), we want to take this up and argue for a similar generative analysis for attributive PP and relative clauses in a second step.