Expressives, Communication and Representation of Consciousness

Diane Blakemore

Project funded by Leverhulme Major Research Fellowship (September 2011 – September 2013)

This project will investigate the meanings of a range of English expressions and constructions and non-linguistic symbols which are said to have an expressive function. This range of phenomena includes expletives (e.g. damn), NP epithets (e.g. the poppet, the bastard), diminutives (e.g. dearie, kitty), exclamations and syntactic exclamatives, stylistic devices (e.g. repetitions, hendiadys), interjections (e.g yuk, wow), and non-linguistic symbols such as punctuation (!!!!) and the emoticons (e.g. smiling/sad faces) used in electronic communication. While it will not focus on emotional intonation specifically, it will examine how lexical and syntactic expressives interact with prosody and non-verbal natural behaviour (e.g. gesture) for the communication of expressive effects in spoken communication. The fact that the term ‘expressive' has been applied to linguistic constructions, borderline linguistic phenomena, and  clearly non-linguistic phenomena raises the question of whether there can be a unitary account of expressive meaning. In this project I shall approach this question by attempting to explain how these different forms contribute to communication in everyday discourse and to the representation of consciousness in fiction.

Research Questions

Expressives have been treated as having marginal significance in linguistics despite their role in the communication of emotions, and until recently any extensive treatment has been restricted to their role as subjectivity markers in free indirect style (Banfield 1982; Fludernik 1993). However, Potts (2005, 2007) and Potts et al (2009a,b) have now demonstrated that they have important implications for semantics (see also Kaplan 1997). At the same time, research on the pragmatics of stylistic effects (Sperber & Wilson 1995; Pilkington 2000; Blakemore 2008), the pragmatics of interjections and non-verbal communication (Wharton 2003, 2009),  and emotional prosody (see http://emotion-research.net/ shows that an understanding of these phenomena is central to the explanation of how communication is achieved. This project builds on this research in the following ways:

1. Semantics (a): The existence of words whose meanings must be explained in terms of their contribution to ‘tone’ (i.e. their effects on the hearer’s feelings or imagination) rather than their contribution to the truth conditions of the sentences that contain them was noted by Frege (1892). However, studies of non-truth-conditional meaning have tended to focus on words whose meanings are not appropriately described in terms of their effect on the hearer’s feelings – namely, words such as still, too, already and connectives such as but, however, and moreover (Karttunen & Peters 1975, Grice 1989, Blakemore 1987, 2002). The exception to this trend is Kaplan (1997) who argues that in contrast with words which contribute to descriptive (truth conditional) content, expressions such as damn and the bastard contribute to expressive content and can only be accommodated in a semantics of use. Thus while an expression is descriptively correct if what it describes is the case, an expression such as the bastard is expressively correct only if the speaker is expressing the appropriate emotion or attitude. Recently, Potts (2005, 2007) and Potts et al (2009a,b) have developed Kaplan’s suggestions for a semantics of use arguing that it can be accommodated in a compositional semantics. Specifically, he has argued that such expressions alter ‘the expressive setting of the context of interpretation’.
   
   Kaplan’s and Potts’ approaches involve extending the semantics so that it specifies not only the truth conditions for expressions with descriptive meaning but also the conditions on the use of expressions with expressive meaning. In this project I argue that this type of approach cannot be maintained and that an account of the meanings of expressions such as damn and the bastard must be embedded in an approach in which linguistic meaning is subserviant to the cognitive processes involved in understanding utterances in context. In this approach the main distinction in semantics is not the distinction between meaning which contributes to truth conditions and meaning which does not (contra Kaplan and Potts), but a distinction between meaning which contributes to conceptual representations which undergo pragmatic processing and meaning which contributes to the procedures involved in pragmatic interpretation (cf Blakemore 1987, 2002, Wilson & Sperber 1993).
   
   Semantics (b): Sentence adverbials such as regrettably or fortunately are amongst the expressions which Frege believed to contribute to tone rather than truth conditions (also see Ifantidou (1993)). However, although such expressions communicate the speaker’s feelings and clearly do not contribute to what Potts calls ‘at issue’ content, they are not included in either his or Kaplan’s treatment of expressive meaning. This is possibly because they do not share all of the features of the expressives discussed by Potts: they are not descriptively ineffable (see below); in contrast with expressives such as damn their repetition leads to redundancy rather than strengthening of the emotion communicated; and in contrast with expressives such as damn they must be understood as communicating the speaker's attitude towards the proposition expressed by the sentence they introduce (as noted by Potts, damn may be understood to be directed at something situational outside the sentence which contains it). This raises the question of whether there is a criterion for distinguishing the expressives discussed by Kaplan and Potts from other cases of 'non-at-issue' content which communicates information about the speaker’s feelings. (See also 5. below)
2. Syntactic form and expressive meaning: Kaplan’s notion of expressive correctness also underlies attempts to analyse the expressive functions of exclamations (You’re all here!) and syntactic exclamatives (What a day, How tall you are) in semantics. Thus Rett (2008) argues that while both exclamations and exclamatives result in the expression of surprise, they are expressively correct under different conditions: an exclamation is expressively correct only if its context is salient, while an exclamative is expressively correct only if its content is additionally about a degree which must exceed a contextually relevant standard. According to Rett, this difference is reflected in the fact that the illocutionary force operators included in the semantic representations of exclamations and exclamatives have different domains, while the similarity between their interpretations is reflected in the fact that these operators have the same value  - the expression of surprise. In this project, I ask to what extent the interpretation of exclamations and exclamatives can or should be explained in the semantics, and explore the role of pragmatics in the interpretation of such utterances.
3. Expressives and non-verbal communication: Expressive meaning is not only communicated by linguistic expressions and constructions, but is also communicated by expressions on the borderline of linguistics, e.g.interjections (cf Wilkins 1992, Scherer 1994, Wharton 2003, 2009, Schroeder); non-linguistic symbols (e.g !!! and emoticons); intonation (see Banse & Scherer 1996, Wichmann 2000, Wilson & Wharton 2006, Schroeder (http://www.dfki.de/~schroed ); natural non-linguistic behaviour (e.g. gestures, facial expressions (cf Ekman 1982, Wharton 2009); or a combination of a number of these. (As Wichmann (2000) points out, it is unlikely that there are emotions communicated uniquely by intonation.) Cruse (1986) has noted that linguistic expressives have properties shared by non-verbal communicative behaviour. This raises the question of whether we should maintain a distinction between linguistically encoded expressive meaning and natural expressive behaviours, or whether expressive phenomena are distributed along a continuum ranging from natural signals to non-natural or properly linguistic signals (cf Wharton 2003, 2009, Schroeder 2000).
4. Descriptive ineffability: Potts (2007) argues that expressives are descriptively ineffable, arguing that an expression such as the bastard may be used to communicate a whole range of emotions from extreme anger to affection, and that its  meaning cannot be pinned down in conceptual terms out of context. Potts cites Blakemore’s (2002) work on discourse markers in support of his claim about the descriptive ineffability of expressives. In fact, Blakemore’s argument is that while an expression such as however cannot be said to correspond to a particular concept, it encodes a procedure for the derivation of a specific set of conceptual representations in a given context. This sort of ‘descriptive ineffability’ must be contrasted with that associated with such stylistic devices as repetition (cf Sperber & Wilson 1995) and hendiadys (cf Blakemore 2008) which have vague effects. As Sperber & Wilson (1995:224) have shown, these devices achieve relevance through a wide array of weak implicatures and in this way contribute to the affective mutuality of speaker and hearer (see also Pilkington 2000). This raises the question of whether the descriptive ineffability of Potts’ expressives is more like that of expressive stylistic devices rather than the descriptive ineffablity of Blakemore’s procedural discourse markers.
5. Emotions and attitudes: As Wichmann (2000) points out, research in affective prosody has often conflated the expression of 'attitude' and the expression of 'emotion' (cf Crystal 1995). Wichmann rejects Couper-Kuhlen’s (1986) suggestion that while emotion is a speaker state, attitude should be viewed as speaker behaviour, and proposes instead that while expressive intonation reflects both (pure) emotional states (e.g. being sad) and emotions arising from propositional states (e.g. impressed that p), attitudinal intonation reflects interactional behaviour (e.g. being rude to someone). However, seems possible that what Wichmann calls attitudinal intonation can be explained in terms of its role in indicating emotions in particular kinds of communicative contexts. This project assesses this distinction in the light of Wharton’s (2009) research on the pragmatics of non-verbal behaviour and asks whether it can be maintained for either prosody or linguistic expressives.
6. Expressives and point of view: It has been argued (e.g. by Cruse 1986, Potts 2005) that expressives are speaker-oriented in that where they do occur in indirect discourse, they must be interpreted as communicating the point of view of the speaker reporting a subject’s thoughts or speech rather than the emotions or attitudes of that subject. However, Harris & Potts (2009) argue that while speaker-oriented interpretations are the ‘pragmatic default’, there are certain types of contexts which are compatible with non-speaker-oriented interpretations. However, their account does not explain why speaker-oriented interpretations should be the ‘pragmatic default’, why the contexts identified by Potts are compatible with non-speaker-oriented interpretations, and how the hearer decides on a particular interpretation in a specific case.
7. Expressives and the representation of consciousness in fiction: The idea that  speaker-oriented interpretations are the ‘pragmatic default’ raises the question of how we should accommodate the use of expressives in free indirect representations of thought in fiction where there is no narrator ‘speaking’ in the text and expressives are characteristically attributed to a character whose thoughts are being represented (cf Banfield 1982, Fludernik 1993). More generally, the project asks whether or to what extent the use of expressives in the representation of consciousness in fiction differ from its use in direct speech.

Workshops and Conferences

Workshop on Expressives and Affective Prosody, part of the 2011 conference 'The Prosody-Discourse Interface'

University of Salford, Greater Manchester  12 September 2011 –  14 September (incl)

Workshop:

• Chris Potts, University of Stanford
• Marc Schroeder, German Research Centre for Artificial Intelligence
• Diane Blakemore

Conference:

• Nicole Dehe, University of Konstanz
• John Local, University of York

Visit the Conference page