A schema for illocutionary act identification with prosodic feature

We propose a new discrimination schema for illocutionary acts using prosodic features based on experimental results.We performed a series of experiments in which subjects were asked to identify the sentence type and intonation contour of given stimuli. Given the transcribed sentence with contextual information, the subjects were able to identify correctly the sentence type of 85% of 290 sentences. With information about the intonation contour types, they could correctly identify 90% of speech acts. We find evidence that illocutionary acts can be signaled by specific contour types. These typical contours are realized in the sentence final boundary tone; a neutral or falling tone for assertion and request, a rising tone for question. An intonation contour is then identified using an algorithm that calculates the range and slope of the upper and lower bounds of unwarped segmental contour, and matches these against predefined contour templates. This algorithm could correctly recognize 78% of the pitch contour types in the utterances. Furthermore, this automated intonation contour classification, nearly 90% of speech acts could be correctly identified.