English Prosody Dataset for investigating English prosody acquisition

English prosody conveys grammatical structure, word meaning, and pragmatic information. However, research on second language (L2) acquisition has primarily focused on segmental properties of speech, with relatively less attention to prosody. We present the English Prosody Dataset to facilitate research on the acquisition and use of English prosody. The dataset includes contrastive examples of three prosodic features: phrase boundary, disambiguating syntactic dependencies; contrastive focus, disambiguating key information within a phrase; and lexical stress, disambiguating word sense through syllabic emphasis, all recorded by three female and three male voice actors speaking Standard Southern British English. Researchers can tailor experiments by selecting stimuli isolating individual acoustic dimensions (e.g., pitch and duration) and by adjusting perceptual difficulty using samples with reduced acoustic information. The dataset also provides acoustically unmanipulated recordings and sentence lists that can be used to develop naturalistic comprehension paradigms or to elicit speech.