No default syntactic scope for advance planning in sentence production: Evidence from finite mixture models

Mean response-onset latencies for sentences that start with a conjoined noun phraseare known to be consistently longer than for sentences starting with a simple noun phrase.This is seen as evidence for the hypothesis that grammatical encoding operates onphrase-size units. We used Bayesian mixture models to contrast this hypothesis with thepossibility that pre-planning is not determined by syntax, but optional. We reanalysed 11published datasets to directly compare these hypotheses. In all cases data were fromimage-description experiments in which participants produced sentences with conjoined(e.g., The dog and the hat) and simple initial noun phrases (The dog). We found consistentevidence for the optional pre-planning hypothesis. We report two follow-up studies:Experiment 1 (N=78) eliminated possible confounds and replicated the advantage for theoptional pre-planning hypothesis. Experiment 2 (N=57) eliminated grammatical encodingby asking participants to only name the images (dog, hat) and showed similar results toExperiment 1. This research provides, for the first time, direct evidence that pre-planningfor conjoined noun phrases is not by default determined by the syntactic processing system.