[PS-1.53] Number interference as predicted by cue-based retrieval

Nicenboim, B. 1 , Engelmann, F. 2 , Suckow, K. 3 & Vasishth, S. 1, 4

1 University of Potsdam, Germany
2 University of Manchester, UK
3 University of Goettingen, Germany
4 CEREMADE, Université Paris Dauphine, France

On a cue-based retrieval account of sentence processing (Van Dyke & Lewis,
2003), heads such as verbs provide cues that are used to distinguish between a
dependent and other competitor items in memory. Similarity-based interference
occurs when several items share a feature associated with a cue, which makes
it harder to identify the target of the retrieval. There is wealth of evidence
showing similarity-based interference effects caused by semantic and syntactic
features in verb-dependency resolution. However, the number feature has only been indirectly studied in number attraction experiments, and has shown results
inconsistent with cue-based retrieval. We show that a high-powered self-paced
reading study in German (N=83 and N=100) with two singular distractors
(assistant and director) can produce a small slowdown at the retrieval site
(had.sg) of the singular subject (philanthropist), when compared with a
condition with only plural distractors (9 ms, CI=[0,18] between conditions);
this is as predicted by cue-based retrieval:

(1) The philanthropist, who the assistant(s) the director(s) greeted had.sg, sat.sg in the committee.

Our findings provide evidence in favor of general dependency creation via
retrieval and can be helpful in constraining theories that explain number attraction.