Semantic transparency affects the phonetic signal

Abstract

Previous research has highlighted the impact of phonological factors and morphological structure on compound pronunciation, but the role of semantics in this interplay remains largely unexplored. Building on psycholinguistic findings that demonstrate semantic effects in compound processing, this study examines whether these effects extend to the production of English nominal triconstituent compounds. Using a state-of-the-art computational model to derive semantic transparency measures, we assessed the predictability of compound meanings from their constituents. Our regression model of experimental speech data revealed a notable duration difference between the first and second constituent and the third constituent for opaque compounds. For semantically transparent compounds, we find a shortening of the third constituent, diminishing the larger duration difference observed for less transparent compounds. These findings underscore the importance of semantic considerations in phonetic analyses of compound constituents, complementing prior research on morphological and phonological correlates of phonetic variation.