w/ Damian Blasi and Gary Lupyan


The recent discovery of a number of putatively sound-symbolic regularities across a large proportion of the world’s languages is incredibly interesting, promising that the application of Big Data techniques to corpus materials and word lists can inform us about universal principles of language, and thus, cognition. Including the findings of these types of techniques into our theories about the evolution of language, however, requires that we explore them carefully and experimentally in an attempt to delineate the source of those associations (which correlations are blind to).

This project seeks to explore the associations found in Blasi et al., for example that words for tongue often contain the consonant /l/. Following from previous research exploring cross-cultural sound-symbolic biases, we explore whether unfamiliar words for tongue from other languages are more likely to be matched with the correct meaning and more easily learned by English and Dutch speakers.

For example, are English speakers more likely to accept ‘lengua’ as a word for tongue because it contains two associated phonemes (/l/, /e/), compared to either ‘gjuhe’ (contains infrequent phoneme /j/) or ‘mova’ (no frequent or infrequent phonemes), despite all three being correct words for tongue. What about words like ‘elle’ that contain frequent phonemes but do not mean tongue? Will experimental participants accept those words more commonly as meaning ‘tongue’, or are other features related to the goodness of fit of entire words at play?


Using a series of experiments, we are attempting to explore the associations observed in the analysis of Blasi et al.’s data. Here we make use of a number of experiments, starting with the set of meanings for which there are both positive and negative phoneme associations (bone, breasts, dog, I, name, nose, tongue, we).

Additionally, we seek to describe methods for quantifying the overall adherence/goodness-of-fit of the languages surveyed to global sound-symbolic patterns, attempting to look at demographic factors that might be related to the presence and strength of sound-symbolism.