Abstract
Wordplay and verbal irony are not as different as they might seem at first glance. In fact, they share the property of being uneigentlich (‘non-actual’), i.e., the meaning of an utterance is somehow dissociated from its wording. Thus the question arises whether this kind of speech is signaled in the production process, and whether the cues thus created can be detected by naive listeners. This contribution presents an overview of the various means by which verbal irony may be conveyed to the listener. This is followed by an empirical study which demonstrates that verbal irony is indeed inherently ambiguous and that the underlying message is phonetically coded. The question addressed is how disambiguation takes place on the phonetic level. In other words - how does a speaker signal the intended meaning and how are listeners able to get the underlying message. The study deals with single-word utterances in sincere and ironic settings. Parameters considered include average voice fundamental frequency (F0) and related measures, intensity of the voice signal, and duration. The results of the acoustic measurements show that production results vary with type of irony (sarcasm vs. kind irony). It is argued that the underlying emotional states have to be taken into account for the interpretation of the measurement results. The perception study yields overall recognition rates of about 70 %, the sincere utterances being identified significantly better than the sarcastic ones in the positive stimulus set (sarcasm) and the ironic ones better than the sincere ones in the negative stimulus set (kind irony).
