The Handbook of Discourse Analysis
The Handbook of Discourse Analysis The Handbook of Discourse Analysis
112 Diane Blakemore captures the attention of the hearer and, second, contextual effects which would not have been achieved by the second segment alone. These analyses have described (28b) and (29b) as reformulations. However, I have argued that this description must itself be analyzed in terms of the notion of interpretive representation. The question of whether an utterance is relevant as an interpretation (rather than a description) is not a question about how it is connected to the preceding text, but a question about the relationship between the proposition it expresses and the thought it represents. As Sperber and Wilson (1986, 1985/6) and Wilson and Sperber (1992) have shown, the notion of interpretive representation is involved in the analysis of a range of phenomena; for example, reported speech, free indirect speech, interrogatives, irony, and metaphor. In some cases, an utterance may be relevant as an interpretation of a thought that has been communicated by an utterance that is not part of a continuous text, and in other cases it may be relevant as an interpretation of a thought that has not been communicated at all. Indeed, according to relevance theory, the identification of an utterance as a reformulation follows from an aspect of interpretation which is fundamental to the way in which the relevance of all utterances is established, and will not itself contribute to the identification of contextual effects. This is not to say that a hearer, or, indeed, an analyst, will not describe the utterance as a reformulation. The point is that such a description is a consequence of the recognition that the utterance is an instance of interpretive rather than descriptive language use. 4 Implications for Discourse Understanding In this chapter I have focused on an approach to discourse which assumes that discourse coherence provides the key to a theory of discourse comprehension, and have shown how in a relevance theoretic framework hearers’ intuitions about coherence can be explained as a consequence of the hearer’s search for an interpretation that is consistent with the Principle of Relevance. However, work in relevance theory is not just concerned with the reassessment of coherence relations. It has also shown how the notion of optimal relevance can be used to explain those aspects of comprehension which are claimed to be a consequence of the search for discourse coherence. For example, recently Wilson and Matsui (1998) have compared the predictions made by Asher and Lascarides’s (1995) coherence-based heuristics for disambiguation in discourse with those made by relevance theory. Whereas relevance theory claims that the same criterion of consistency with the Principle of Relevance explains disambiguation in both isolated utterances and extended texts, Asher and Lascarides’s heuristics are designed to supplement the word-association heuristics given in the artificial intelligence literature for disambiguating isolated utterances. Wilson shows that neither the heuristics for isolated utterances nor the heuristics for discourse make the correct predictions and argues that disambiguation phenomena are more satisfactorily explained in terms of the notion of optimal relevance. The criterion of consistency with the Principle of Relevance also provides a unitary explanation for the assignment of reference in isolated utterances and discourse sequences such as (30) (from Wilson 1992):
Discourse and Relevance Theory 113 (30) Sean Penn attacked a photographer. The man was quite badly hurt. While it is often claimed that reference resolution is affected by the relative accessibility of the candidate referents, it is also agreed that an account based on accessibility alone would make the wrong predictions. For example, Herb Clark (1977) proposes that reference assignment in examples like (30) is affected by the number and plausibility of the assumptions needed to introduce the intended referent; but as Wilson (1992) and Matsui (1993, 1995) show, this proposal does not deal with all examples. Candidate referents must also be evaluated in terms of a pragmatic criterion that the overall interpretation is supposed to meet. However, Wilson and Matsui (1998) have shown that neither the attempts to define such a criterion in terms of truth (cf. Lewis 1979; Sidner 1983) nor the attempts to develop a coherence-based criterion (cf. Hobbs 1979; Fox 1987) explain reference resolution in all cases. Moreover, a criterion which is powerful to choose among the various interpretations of an utterance on either of these grounds could do so only by considering them all. As Wilson (1995) says, this “would create a combinatorial explosion of gigantic proportions, and be quite unlike what hearers actually do.” Her relevance theoretic analyses of examples that are problematic for both truth- and coherence-based accounts show that what hearers actually do is to accept the first interpretation that is consistent with the Principle of Relevance and that the speaker could have manifestly foreseen. Within coherence-based approaches to discourse, expressions like utterance-initial so, well, still, after all are classified as discourse markers, a term which is intended to reflect the role that these expressions play in marking, signaling, or indicating how one unit of discourse is connected to another (cf. Levinson 1983: 87–8; Fraser 1990; Mann and Thompson 1987; Sanders et al. 1993; Knott and Dale 1994). 10 Since relevancebased approaches are concerned with processes of utterance understanding rather than the structure of discourse, and appeal to contextual effects rather than coherence relations, it is not surprising that relevance theoretic analyses of these expressions are significantly different from coherence-based ones. For example, whereas Sanders et al. analyze but as an explicit guide to a range of coherence relations (namely, Contrast, Antithesis, Contrastive Cause–Consequence), my 1987 analysis treats but as an expression which constrains the interpretation process by narrowing down the search for the intended contextual effects. Thus while this analysis, like Sanders et al.’s, treats but as expressing either contrast or denial of expectation (cf. Lakoff 1971), it does this not by analyzing it in terms of a marker of coherence relations, but by analyzing it as an instruction for the recovery of contextual effects. 11 The analysis of a discourse marker as an expression which links units of discourse would seem to imply that it cannot be used discourse initially. However, as the examples in (31–2) show, this is clearly not the case: (31) (speaker sees hearer come in laden with shopping) So you’ve spent all your money. (32) (speaker takes an enormous slice of cake) After all, it is my birthday.
- Page 84 and 85: 62 Deborah Schiffrin and semantical
- Page 86 and 87: 64 Deborah Schiffrin shows that tea
- Page 88 and 89: 66 Deborah Schiffrin parallel claus
- Page 90 and 91: 68 Deborah Schiffrin see E. Abraham
- Page 92 and 93: 70 Deborah Schiffrin Bolinger, D. 1
- Page 94 and 95: 72 Deborah Schiffrin Hoyle, S. 1994
- Page 96 and 97: 74 Deborah Schiffrin Schiffrin, D.
- Page 98 and 99: 76 Neal R. Norrick 4 Discourse and
- Page 100 and 101: 78 Neal R. Norrick In the following
- Page 102 and 103: 80 Neal R. Norrick that between ref
- Page 104 and 105: 82 Neal R. Norrick Prince (1981), C
- Page 106 and 107: 84 Neal R. Norrick But the semantic
- Page 108 and 109: 86 Neal R. Norrick entailments invo
- Page 110 and 111: 88 Neal R. Norrick interpretation,
- Page 112 and 113: 90 Neal R. Norrick In a second exam
- Page 114 and 115: 92 Neal R. Norrick REFERENCES Abrah
- Page 116 and 117: 94 Neal R. Norrick Goodwin, Majorie
- Page 118 and 119: 96 Neal R. Norrick Kuno, Susumu (19
- Page 120 and 121: 98 Neal R. Norrick Prince, Ellen (1
- Page 122 and 123: 100 Diane Blakemore 5 Discourse and
- Page 124 and 125: 102 Diane Blakemore theory of disco
- Page 126 and 127: 104 Diane Blakemore it is not clear
- Page 128 and 129: 106 Diane Blakemore example, the in
- Page 130 and 131: 108 Diane Blakemore (20) a. John br
- Page 132 and 133: 110 Diane Blakemore However, it doe
- Page 136 and 137: 114 Diane Blakemore If these expres
- Page 138 and 139: 116 Diane Blakemore REFERENCES Ashe
- Page 140 and 141: 118 Diane Blakemore Unger, C. (1996
- Page 142 and 143: 120 Gregory Ward and Betty J. Birne
- Page 144 and 145: 122 Gregory Ward and Betty J. Birne
- Page 146 and 147: 124 Gregory Ward and Betty J. Birne
- Page 148 and 149: 126 Gregory Ward and Betty J. Birne
- Page 150 and 151: 128 Gregory Ward and Betty J. Birne
- Page 152 and 153: 130 Gregory Ward and Betty J. Birne
- Page 154 and 155: 132 Gregory Ward and Betty J. Birne
- Page 156 and 157: 134 Gregory Ward and Betty J. Birne
- Page 158 and 159: 136 Gregory Ward and Betty J. Birne
- Page 160 and 161: 138 Laurel J. Brinton 7 Historical
- Page 162 and 163: 140 Laurel J. Brinton of particular
- Page 164 and 165: 142 Laurel J. Brinton Shakespeare,
- Page 166 and 167: 144 Laurel J. Brinton argues that i
- Page 168 and 169: 146 Laurel J. Brinton typical of th
- Page 170 and 171: 148 Laurel J. Brinton 1 What is the
- Page 172 and 173: 150 Laurel J. Brinton politeness. 1
- Page 174 and 175: 152 Laurel J. Brinton that this are
- Page 176 and 177: 154 Laurel J. Brinton functions; -s
- Page 178 and 179: 156 Laurel J. Brinton Linguistics,
- Page 180 and 181: 158 Laurel J. Brinton Kastovsky, Di
- Page 182 and 183: 160 Laurel J. Brinton semantic-prag
112 Diane Blakemore<br />
captures the attention <strong>of</strong> the hearer and, second, contextual effects which would not<br />
have been achieved by the second segment alone.<br />
<strong>The</strong>se analyses have described (28b) and (29b) as reformulations. However, I have<br />
argued that this description must itself be analyzed in terms <strong>of</strong> the notion <strong>of</strong> interpretive<br />
representation. <strong>The</strong> question <strong>of</strong> whether an utterance is relevant as an interpretation<br />
(rather than a description) is not a question about how it is connected to<br />
the preceding text, but a question about the relationship between the proposition it<br />
expresses and the thought it represents. As Sperber and Wilson (1986, 1985/6) and<br />
Wilson and Sperber (1992) have shown, the notion <strong>of</strong> interpretive representation is<br />
involved in the analysis <strong>of</strong> a range <strong>of</strong> phenomena; for example, reported speech, free<br />
indirect speech, interrogatives, irony, and metaphor. In some cases, an utterance may<br />
be relevant as an interpretation <strong>of</strong> a thought that has been communicated by an<br />
utterance that is not part <strong>of</strong> a continuous text, and in other cases it may be relevant as<br />
an interpretation <strong>of</strong> a thought that has not been communicated at all. Indeed, according<br />
to relevance theory, the identification <strong>of</strong> an utterance as a reformulation follows<br />
from an aspect <strong>of</strong> interpretation which is fundamental to the way in which the relevance<br />
<strong>of</strong> all utterances is established, and will not itself contribute to the identification<br />
<strong>of</strong> contextual effects. This is not to say that a hearer, or, indeed, an analyst, will<br />
not describe the utterance as a reformulation. <strong>The</strong> point is that such a description is a<br />
consequence <strong>of</strong> the recognition that the utterance is an instance <strong>of</strong> interpretive rather<br />
than descriptive language use.<br />
4 Implications for <strong>Discourse</strong> Understanding<br />
In this chapter I have focused on an approach to discourse which assumes that<br />
discourse coherence provides the key to a theory <strong>of</strong> discourse comprehension, and<br />
have shown how in a relevance theoretic framework hearers’ intuitions about coherence<br />
can be explained as a consequence <strong>of</strong> the hearer’s search for an interpretation<br />
that is consistent with the Principle <strong>of</strong> Relevance. However, work in relevance theory<br />
is not just concerned with the reassessment <strong>of</strong> coherence relations. It has also shown<br />
how the notion <strong>of</strong> optimal relevance can be used to explain those aspects <strong>of</strong> comprehension<br />
which are claimed to be a consequence <strong>of</strong> the search for discourse coherence.<br />
For example, recently Wilson and Matsui (1998) have compared the predictions<br />
made by Asher and Lascarides’s (1995) coherence-based heuristics for disambiguation<br />
in discourse with those made by relevance theory. Whereas relevance theory claims<br />
that the same criterion <strong>of</strong> consistency with the Principle <strong>of</strong> Relevance explains<br />
disambiguation in both isolated utterances and extended texts, Asher and Lascarides’s<br />
heuristics are designed to supplement the word-association heuristics given in the<br />
artificial intelligence literature for disambiguating isolated utterances. Wilson shows<br />
that neither the heuristics for isolated utterances nor the heuristics for discourse<br />
make the correct predictions and argues that disambiguation phenomena are more<br />
satisfactorily explained in terms <strong>of</strong> the notion <strong>of</strong> optimal relevance.<br />
<strong>The</strong> criterion <strong>of</strong> consistency with the Principle <strong>of</strong> Relevance also provides a unitary<br />
explanation for the assignment <strong>of</strong> reference in isolated utterances and discourse<br />
sequences such as (30) (from Wilson 1992):