Attitudinal Intonation and the Inferential Process

The role of prosody in conveying affective meaning is complex. The complexity is reflected to some extent in the many labels used to describe ‘ways of speaking’ that could generally be described as affectively coloured. The search for prosodic correlates of emotional speech, however, is more successful for some labels than for others. I argue that some labels refer not to the affective prosody itself, but to the meanings implied by or inferred from utterances in a given interactional context. These meanings, particularly those suggestive of attitude or interpersonal stance, may, of course, arise in part from a perceived affective colouring of the voice such as sadness or anger. Some, on the other hand, may be generated by the strategic use of prosodic patterns that are not inherently ‘attitudinal’, but are in some way incongruent with the text or context, and set in train the process of interpretation of speaker meaning. The notion of incongruence, however, presupposes the notion of congruence, and I argue that if we are to fully understand the contribution of prosody to speaker meaning, the search for emotion in the voice should be complemented by the study of ‘normative’ use of prosody in interaction. 1. Attitudinal labels Scherer [16] regards emotion as a process, consisting of several components: physiological, cognitive, sociomotivational and ‘action tendency’. This complexity, he claims, is documented in the labels we use, which highlight different aspects of the ‘emotion process’. In addition to what we may learn about the nature of emotion from descriptions of ‘ways of being’, we can also glean useful information about utterance interpretation from labels used to describe more specifically ‘ways of saying’ and this information is crucial to our efforts to understand the role of affect in conversational interaction. Since many studies of the relationship between affect and speech make use of labels, it is useful to look at the labels themselves. The labels used to describe ‘ways of saying’ have been studied in a corpus of British English speech and writing, (ICE GB International Corpus of English, compiled at the Survey of English Usage at University College, London) [19]. Results suggest that participants in interaction make a distinction between acoustic features of the voice, the speaker states that provoke such features, and the utterance meanings that are the result of a process of interpretation using a wide range of cues including the assumed emotional state of the speaker. We find, for example, a category of labels that describe prosodic characteristics themselves: quiet, quick, wavering, high-pitched. These are lay terms for features that phoneticians have little difficulty in translating. If such terms are included in a study of prosodic correlates with verbal labels, e.g. criarde, rapide, dynamique (shrill, rapid, dynamic) [7], we would expect to find close acoustic correlates, precisely because these are lay attempts to describe acoustic features. Another category of labels describes the emotional state of the speaker which has been inferred from the acoustic signal As Scherer observed [16], such labels reflect different components of emotion. Some (e.g. excited, anxious, a bit down, placid) focus on the physiological aspects while others highlight cognitive aspects (e.g. incredulous, scornful, convinced, disappointed), i.e. the reason for the state rather than the state itself. The acoustic correlates of these are likely to be more closely related to the physiological response to an event than to the reasons for that response. A very different set of labels describes ways of saying that are actually ways of behaving, in other words, actions that require a receiver and a context. These are examples of interpersonal stance a potentially affectively coloured behaviour towards an interlocutor. In the literature on intonation, interpersonal stance is usually referred to as ‘attitude’, a use of the term which is different from the practice in social psychology. This difference can be the source of misunderstanding in the interdisciplinary study of emotion and speech, so it is worth explaining that ‘attitude’ in social psychology refers to beliefs and opinions which motivate or explain behaviour, while ‘attitude’ as traditionally used in the study of intonation refers to the behaviour itself. Labels referring to speaker behaviour do so primarily from the perspective of the hearer e.g. firm, sympathetic, arrogant, patronising, insistent. In other words they are more likely to be meanings that are inferred by the hearer than to be those necessarily intended by the speaker. Some of these perceptions may be reinforced by voice quality (e.g. a ‘warm’ voice may cue both ‘sympathetic’ and ‘patronising’), others are an even more indirect interpretation of the attitude of the speaker derived from a complex interaction of text, prosody and context. Most importantly, none of these are expressions of speaker state: it is possible to be sad or happy on your own, but it is not possible to be condescending on your own. How do such meanings arise? The kind of prosodic behaviour that might give rise to such interpretations is suggested by the final category of labels found in the study. These refer to discourse behaviour (abrupt, sudden, final), behaviour which may be intrinsic to the utterance itself (such as the extreme finality of a low terminal), but may also reflect how the utterance relates to other utterances, such as the timing of turns. Such effects may play an important part in the participants’ experience of the interaction. If a speaker sounds ‘final’ when the hearer wishes to continue, the experience may be a negative one for the hearer, and may generate the kind of impressions captured in the kind of labels already described above those which reflect the hearer’s perception of the speaker’s attitude or behaviour towards them, e.g. the impression that the speaker is being uncooperative, unfriendly or inattentive. Speech Prosody 2002 Aix-en-Provence, France April 11-13, 2002 ISCA Archive http://www.isca-speech.org/archive