CSCAN ROUNDS
Thursday September 11th 2014, 1:00pm
PROCESSING NOVEL DEFINITES: EVIDENCE AGAINST ACCOMMODATION
It is a common assumption that the prototypical use of a definite noun phrase is the anaphoric one. A referent is introduced by an indefinite DP in the discourse and is then returned to by the definite (see example in (1)). That means that definite NPs commonly refer to discourse-given information. In other words, the definite determiner presupposes familiarity.

(1) Yesterday evening I saw an artiste at the Barelli Circus. The artist was rope walking.
(2) Yesterday evening I went to the Barelli Circus. The artist was rope walking.

Therefore, it is a striking phenomenon that definite noun phrases are able to introduce new discourse referents as well (see example in (2)).
The underlying constraints for the use of novel definites are not sufficiently understood so far. One prominent account is based on the assumption that the familiarity presupposition can be accommodated if a bridging inference can be drawn to a given discourse referent (Barelli Circus in (2); see e.g. Clark, 1975; Burkhardt, 2006; Singh et al., submitted). An alternative proposal assumes that a novel definite is licensed if the entity it refers to is part of a (small enough) local frame introduced by the discourse (Evans, 2005; Frazier, 2006). The prediction of the Small Frame Account is that whenever a local frame is available in the context, a definite should be preferred to an indefinite. This prediction differs from the prediction of the Familiarity Account which assumes an additional process, namely accommodation, which might not lead to measureable processing costs if the novel definite is highly predictable, but does not predict a preference for a novel definite compared to an indefinite since accommodation of the familiarity presupposition has to take place in any case.

In my talk, I will present two experiments (AJT & SPR) comparing the processing of novel definites compared to indefinites in German. The results revealed evidence for the Small Frame Account, i.e. a preference for novel definites if a local frame is present.
PRESENTED BY
Britta Stolterfoht
from the University of Tübingen
INVITED BY
Christoph Scheepers