Contrast or context, that is the question

Abstract

In the past decades, most research on question intonation has relied on controlled experimental tasks for the production and perception of syntactically well-formed, pragmatically or syntactically ambiguous, unbiased and genuine requests for new information.
The exploration of a real-life interaction in East Midlands British English shows that such questions are neither frequent nor typical. Prosodic information does undoubtedly contribute to the identification of questions, as shown by comparing questionhood ratings, based on either audio recordings or orthographic transcriptions only. However, compared to the role of syntactic, lexical, contextual, sequential, and interactional information, the role of speech prosody is shown to be marginal.
These findings suggest that prosodic information is difficult to isolate or to manipulate independently. Therefore, question intonation can only be adequately studied by taking into account naturalistic, context-rich, spontaneous behaviour.