Morgan, A. M. 1 , von der Malsburg, T. 2 , Ferreira, V. S. 1 & Wittenberg, E. 2
1 UC San Diego, Psychology Department
2 UC San Diego, Linguistics Department
English resumptive pronouns (RPs; (1)) are enigmatic due to their low acceptability but regular production. Hofmeister & Norcliffe (2014) argue that their function may be to facilitate comprehension. This conclusion, however, was based on decreased reading times, not comprehension.
1. It was Mr.Dino that Mr.Rabbit wondered whether Ms.Frog tickled ___GAP/him_RP with a feather.
We conducted a self-paced reading experiment (N=96) similar to H&N?s, but also measuring comprehension, and eliminating pragmatic information that can support comprehension even without full syntactic parsing, as in (1).
We too found faster reading times following RPs (vs. gaps). More importantly, comprehension measures show that with RPs, sentences were comprehended *worse* than with gaps, even in sentence types where RPs are commonly produced.
The comprehension results suggest that RPs led comprehenders to build locally coherent but globally infelicitous dependencies. This is supported by a second one-item sentence-picture matching experiment (N=300) replicating the same effect.
These results suggest that speakers probably do not produce RPs to benefit comprehension. Instead, RPs are likely produced to satisfy local argument structure requirements when structural constraints force speakers to abandon forming long-distance dependencies (authors, in prep).