Imageability: now you see it again (albeit in a different form)

Imageability is a complex and controversial measure as was already noted by Paivio et al. (1968) when they first introduced the term. According to these authors imageability should describe the ease/difficulty with which “words arouse a sensory experience” (p. 4) and therefore should closely correlate with concreteness. Since concrete objects are experienced through the senses, the words that denote them should easily re-evoke that sensory experience: words high in concreteness should also be high in imageability and vice versa. However, Paivio et al. also noted an interesting asymmetry: some terms with low concreteness ratings (i.e., abstract terms) had high imageability ratings. The peculiarity of these words is that they have strong emotional connotations (Paivio et al., 1968, p. 7). (They also noted a few terms with high concreteness and low imageability ratings such as e.g., astrolab, but these are easy to explain: people know that they denote concrete objects, but they do not know what they look like). Westbury et al. (2013) aim is mainly pragmatic, and consists in “estimating imageability using algorithmic methods grounded in empirical measures,” and we are quite sympathetic to this endeavor. Nevertheless, the measures these authors rely on clearly imply a theoretical account able to deal with the anomalies of this construct. Their strategy is to distinguish two aspects of imageability: (a) the relationship between imageability and context, and (b) the relationship between imageability and emotionality. The line of reasoning they adopt to develop their method is the following.