A free and publicly available stimuli set for statistical learning experiments in language and music

Abstract

The ability to encode regularities from the environment and abstract these patterns into structured cognitive representations is called statistical learning. Currently, one of the most studied paradigms of statistical learning is a word segmentation task originally introduced in a landmark paper by Saffran, Newport, and Aslin (1996). In this paradigm, participants hear a continuous stream of artificial words without acoustic cues to the boundaries separating the words. Then, they are asked to discriminate between words and non-words in the artificial language. One of the barriers that has impeded research progress using this paradigm is the lack of stimulus standardization. This lack of standardization is inefficient, causing researchers to create new stimuli for each new line of experiments. Furthermore, it makes the comparison of methods and results between different experiments difficult, if not impossible. The goal of the current report is to introduce a publically available stimuli set for language (syllable) and music (tone) segmentation experiments of statistical learning. All stimuli, data, and code associated with the current report are available for download on Open Science Framework, and we report here our validation experiments for both sets of stimuli.