Abstract
A recent dramatic increase in the number and scope of chronometric and norming lexical megastudies offers the ability to conduct virtual experiments—that is, to draw samples of items with properties that vary in critical linguistic dimensions. This paper introduces a bootstrapping approach, which enables testing of research hypotheses against a range of samples selected in a uniform, principled manner and evaluates how likely a theoretically motivated pattern is in a broad distribution of possible outcome patterns. We apply this approach to conflicting theoretical and empirical accounts of the relationship between the psychological valence (positivity) of a word and its speed of recognition. To this end, we conduct three sets of multiple virtual experiments with a factorial and a regression design, drawing data from two lexical decision megastudies. We discuss the influence that criteria for stimuli selection, statistical power, collinearity, and the choice of dataset have on the efficacy and outcomes of the bootstrapping procedure.
Thanks are due to two anonymous reviewers, as well as Emma Bridgwater, Constance Imbault, Daniel Schmidtke, and Amy Beth Warriner for their valuable comments on earlier drafts, and to the audience of the Tucson Mental Lexicon Workshop, London, ON, Canada where this work was presented in November, 2013.
Notes
1 Since the critical difference between theoretical accounts is in the pairwise comparisons between valence levels (negative vs. positive and negative vs. neutral) and not in whether there is an overall difference between the three levels, in what follows we concentrate on statistical power of t tests, and not ANOVA.
2 A model type where frequency was allowed to interact with both the linear and quadratic terms of valence did not yield significant results and is not reported further. Also, for simplicity we only report models with valence and frequency as predictors. Models with a larger set of predictors including age of acquisition and word length (see Kuperman et al., Citation2014, for the list) showed very similar results and are not reported here.