Correlates of word likeness in the perception of pseudowords

Abstract

New words are often introduced into languages from borrowings or neologisms. Pseudowords (possible words of a language that have not yet been assigned meaning) are used in many experiments to test aspects of speech perception. However, what we know about the perception of pseudowords is relatively limited. In this study, we investigate word likeness ratings of pseudowords on a 1–7 scale. A total of 659 listeners responded to sets of 200 pseudowords from an item list of 9599 English-sounding pseudowords varying in number of syllables and morphological complexity. We analyze the listeners' responses with the following predictors: phonotactic predictability, duration, phonological neighborhood density, phonological uniqueness point, pseudo morphological complexity, and acoustic distinctiveness. We analyze the ordinal data to identify how these predictors influence listeners’ ratings. We also take average word likeness ratings of the pseudowords and use them to predict a different set of listeners’ reaction times in a lexical decision task. In this presentation, we discuss what properties of the pseudowords influence what makes them seem more word-like or not and how word-likeness affects auditory word recognition.