Is theology more of a field than a father is a king? Modelling semantic relatedness in processing literal and metaphorical statements.

One major question in the study of metaphors historically is: Are different mechanisms involved in the comprehension of figurative statements versus literal statements? Many studies have addressed this question from a variety of perspectives, with mixed results. Following Harati, Westbury, and Kiaee (Behavior Research Methods, 53, 2214-2225, 2021), we use a computational (word embedding) model of semantics to approach the question in a way that allows for the quantification of the semantic relationship between the two keywords in literal and metaphorical "x is a y" statements. We first demonstrate that almost all literal statements (95.2% of 582 statements we considered) have very high relatedness values. We then show that literality decisions are slower for literal statements with low relatedness and metaphorical statements with high relatedness. We find a similar but smaller effect attributable to the cosine of the vectors representing the two keywords. The fact that the same measurable characteristics allow us to predict which metaphors or literal sentences will have the slowest literality decision times suggests that the same processes underlie the comprehension of both literal and metaphorical statements.

Evaluating the predication model of metaphor comprehension: Using word2vec to model best/worst quality judgments of 622 novel metaphors.

In this paper our goal is to undertake a systematic assessment of the first, most widely known, and simplest computational model of metaphor comprehension, the predication model developed by Kintsch (Cognitive Science, 25(2), 173-202, 2000). 622 metaphors of the form "x is a y" were selected from a much larger set generated randomly. The metaphors were judged for quality using best/worst judgments, which asks judges to pick the best and worst metaphor from among four presented metaphors. The metaphors and their judgments have been publicly released. We modeled the judgments by extending Kintsch's predication model (2000) by systematically walking through the parameter space of that model. Our model successfully differentiated metaphors rated as good (> 1.5z) from metaphors rated as bad (< -1.5z; Cohen's d = 0.72) and was able to successfully classify good metaphors with an accuracy of 82.9%. However, it achieved a true negative rate below chance at 36.3% and had a resultantly low kappa of 0.037. The model could not distinguish unselected random metaphors from those selected by humans as having metaphorical potential. In a follow-up study we showed that the model's quality estimates reliably predict metaphor decision times, with better metaphors being judged more quickly than worse metaphors.