THE EFFECTS OF CONTEXT ON UTTERANCE INTERPRETATION : SOME QUESTIONS AND SOME ANSWERS

In this paper I want to consider one aspect of the answer to this question. In considering this aspect of the interpretation question, I am going to foc~s on the interpretation of utterances of single sentences "single sentence utterances", for short. But the interpretation question (1) arises not only for single sentence utterances. It also arises in the case of "extended" utterances utterances which are the realization of a series of sentences. The term "discourse" is normally used to refer to such extended, or multi-sentence, utterances ,1 Given that there is an interpretation question for single sentence utterances as well as for multi-sentence utterances (=discourse), one should of course ask whether there is any connection between the answers to these

two questio~s.In the last part of the paper I briefly consider this matter.
In the study of utterance interpretation there is almost unanimous agreement on one point: it is not only an addressee's linguistic knowledge or his gramml:Lticalcompetence which enables him to successfully interpret utterances.There is, quite simply, a huge gap between, on the one hand, the information/meaning which an addressee can recover from an utterance on the basis of linguistic knowledge alone and, on the other hand, the information/meaning which the communicator intends to convey by means of this utterance.2Consider for instance the interpretation of the following, deceptively simple, utterance. (2) Peter's bat is too big.
The meaning which a hearer solely on the basis of his represented roughly as follows: can assign to this utterance ling'uistic knowledge can be (3) 'Some hitting instrument, or selme flying mammal, somehow associated with some entity bearing the name Peter has, at the time of utterance, some property of bigness to an extent which is excessive relative to sc)meunidentified referencepoint.'3 Let us call the meaning represented in (3) the "linguistic meaning" of (2).Now clearly this linguistic meaning of (2) is vague and underspecified in several respects.This can be highlighted by considering a number of questions which are left unanswered by (3), but which would normally have to be answered by a hearer to whom (2) is addressed.
* Should the ambiguous term bat be interpreted in its In the cricket situation outlined above, the speaker would intend to convey at least the inforlDation represented in (5); accordingly, a hearer will have t.o recover at least this information if he is to succeed in interpreting (2): (5) "At the time of utterance the following holds: the hitting instrument being used by the pers,onwho bears the name Peter.
and who is batting has the property of being too big for him to bat with successfully."tNote that in (5) all the questions left unanswered by (3) are answered.There is then a clear "e:ap"between the information which a hearer can recover for (2) lsolely on the basis of his linguistic knowledge, and the information which he has to recover in order to successfully interpret (2) in a specific situation.
The following question then arises in connection with the interpretation of utterances such as ( 2): (6) The "gap" question How does an addressee inanage to bridge the gap between (i) the information/meaning which he can recover for an utterance on the basis of his linguistic knowledge and (ii) the information/meaning which the communicator intends to convey with the utterance?
It is widely accepted that non-linguistic factors in the form of context play an important rOlE!in the bridging of the gap between the linguistic meaning of an.utterance and the meaning which a communicator intends to convey with this utterance.In addition to the term "context", there are several other terms used in the literature to refer to the non-linguistic factor(s) involved in determining utterance interpretation.These include, inter alia, "contextua.lknowledge" ,. "background .knowledge",and "world knowledge".5 The illustration given above of how the information conveyed by (2) can become more determinate by imagining (2) being uttered in the cricket situation, can be regarded as an illustration of how context co-determines utterance interpretation.To get an even clearer sense of the dramatic impact of context on utterance interpretation, imagine for yourself how differently the various questions left "unanswered" by the linguistic meaning of (2) will be answered if ( 2) is addressed to a hearer in the laboratory situation outlined below, instead of in the cricket situation outlined above.* the translator, who struggles with differences between the background knowledge of the rellders of a source text and that of the intended readers of the target language text,7 * the translation teacher, who has to prepare translators to deal successfully with this problem,s * the language teacher, who has to select texts for comprehension tests that are "f~dr", that is, that will not discriminate among students on the basis of non-linguistic background knowledge/experienc~,9 * the reading teacher, who, has to decide whether the comprehension problems of a poor reader is due to an inadequate language competence or to a lack of the requisite background knowledge (or maybe to a failure to integrate There is then a clear need on the part of several language professions for an understanding of the role of context in utterance interpretation.In line with the conference theme, one can also ask the converse question: Does linguistics, and pragmatics in particular, have anything to offer to all those language professionals who are involved with language in use, and who thus need to understand the role of context in utterance interpretation?
But maybe you feel that there is no real need for language professionals concerned with the role of context in utterance interpretation to turn to linguistics and pragmatics in particular -for enlightenment.You could argue, for instance, that there are several conceptions of context available which could serve these language professionals well, and that there is therefore no need for them put in any effort to become acquainted with pragmatic theories which deal with the matter.13To counter this line of argument, I am going to illustrate the inadequacy of three possible answers to the "what is context?" question.For lack of a better term, I will refer to these answers as "popular" answers.By illustrating the inadequacy of these answers, I will hopefully convince those language professionals concerned with utterance interpretation that, in order to gain real insight into the role of context in utterance interpretation, they have to go beyond these "popular" conceptions of context.In the very last part of the paper I will then briefly introduce you to some recent developments in pragmatics which do seem to provide us with new, and worthwhile, insights into the role of context in utterance interpretation.

2.
Three inadequate answers to the "what is context?" question 2.1 A first answer: "Context is the situation of the utterance." One popular answer to the "what is context?" question goes as follows: the context for the interpretation of an utterance is the situation in which it is uttered, that is, its situational context.lf There can be no doubt that the situation in which an utterance is uttered, can somehow affect its interpretation.The two different interpretations of (2) in the cricket situation and the laboratory situation sketched above seem to bear this out.
If, however, we want to understand how a hearer manages to interpret an utterance, 'context' cannot be defined with reference to aspects of the situation or "external environment" in which the utterance is uttered.The reason for this is really quite.simple.The external environment does not affect language directly, but only via the speaker's and hearer's knowledg~of this environment.As Blass (1990:31) explains, "not everything that could potentially be perceived attracts attention.Moreover, people perceiving the same physi.calenvironment do not necessarily represent it to themselves in the same way." The general point is then this: if we want to understand what the context is that affects a hearer's interpretation of an utterance, we have to focus on thl~"environment inside the hearer's head", and not on the external environment.
There is a second reason why an adequate general notion of ,context' cannot be defined with reference to the situational context of an utterance: many of the contextual assumptions which In sum, then: the claim that context is the situational context of an utterance does not represent an adequate general answer to our "what is context?" question.
2.2 A second answer: "Context is the co-text of the utterance." A second possible answer to the "what is context?" question, is that context is the co-text -or linguistic context of an utterance.More specifically, the claim is that the context for the interpretation of an utterance is the set of assumptions explicitly expressed by the utterances in the preceding part of the discourse.lS It is undoubtedly the case that' assumptions explicitly expressed in a discourse can affect the interpretation of an utterance preceded by this discourse.Context cannot, however, be equated with co-text.Let me briefly mention two reasons why the notion that context is co-text cannot form the basis for an adequate general answer to the "what is context?" question.
A first reason is that the interpretation of single, isolated utterances is affected by context, even though they do not have a co-text. 16Consider for instance the interpretation of the following utterance, addressed by a child -rushing into the house -to her mother: (9) The dog is chewing my toys!
In order to successfully interpret this utterance, the hearer, amongst other things, has to identify what dog the child is referring to, and what she (the hearer) is supposed to do in response to the utterance.Contextual, or background, knowledge clearly plays an important role in this interpretation.In this case context co-determines the interpretation of an utterance even though the utterance does not have a co-text (or linguistic context).In the case of such isolated utterances, then, we have context without any co-text.The saDIe is true for the first sentence in an extended discourse.17 A second reason why context cannot be equated with co-text relates to the interpretation of utterances which are embedded in a larger discourse.It is not the case that all the contextual assumptions required for the interpretation of such an utterance are explicitly expressed by earlier parts of the discourse.
Consider for instance the interpretation of Mary's response to Peter's question in (10)18: (10) a. Peter: Do you want some coffee:' b.Mary: Coffee will keep me aWakE!.
The co-text of Mary's utterance alone will not enable Peter to decide whether Mary is (indirectly) sayin~that she does want coffee, or whether she is saying that she does not want coffee.
In order to decide what she is "really trying to say", Peter has to bring to the interpretation process either one of the following two assumptions: (11) a. Mary wants to stay awake.First, even if this position can be upheld for utterances forming part of some larger discourse, it will still leave us without an account of context for the interpretation of an isolated utterance or for the interpretation of the first utterance in a text or discourse.
Second, such a characterization of context will have the effect that the context for the interpretation of an utterance consists of a very large set of assumptions -and this, as I will try to show, creates a serious problem.Consider in this connection the following discourse. 22  (12) (i) Oh, fish for breakfast.(ii) If I had known it was going to be fish, I would have put my contact lenses in.
One possible interpretation of (12ii) is that the speaker would have liked to eat fish, but that she will not be able to since she cannot properly see the fishbones without her contact lenses.How can this interpretation come about?
Let us first focus on the encyclopaedic knowledge about fish accessed by the first utterance.This knowledge includes the knowledge that fish have bones which can be difficult to see, that they live in water, that they swim, that they have gills, that they have fins, that they provide food, that there is a whole industry dedicated to the harvesting of fish, and so on.In terms of the conception of 'context' with which we are now dealing, all this information will form part of the conversational background, and hence the context, of the utterance (12ii).The first utterance also pefers to breakfasts.On the conception of 'context' outlined above, this first utterance will also contribute a huge amount of encyclopaedic information about breakfasts to the context for the interpretation of the second utterance.Clearly, the sumtotal of encyclopaedic information associated with fish and breakfasts is very large.(And, of course, the more extended the discourse, the larger the body of encyclopaedic information associated with all the concepts which feature in the assumptions explicitly and implicitly communicated by the discourse preceding an utterance.)Let us, .for the sake of the illustration, focus only on the information provided via the concept 'fish' to the context for the interpretation of (12ii), leaving the contribution of the concept 'breakfast' aside.Now note that out of the vast amount of information which is to be regarded as the context for the interpretation of an utterance under the conception of context that we are now considering, only the assumption that fish have bones which can be difficult to see, are required to help the reader interpret (12ii).Much of the information which belongs to the so-called conversational common ground has nothing to do with the interpretation of an utterance which belongs to a discourse.
By equating context with the conversational common ground, then, we seem to end up with a very unenlightening notion of 'context': it may include what we want, but it includes a whole lot more besides this.
Our fish example can be used to illustrate a third shortcoming of the conception of context as the conversational background.Contrary to what is assumed in this definition of 'context', context is in any event not fully determined by the preceding discourse.To put it differently: context is not fixed in advance to the interpretation of an utterance.Consider again the interpretation of (12ii) along the lines set out earlier.
This interpretation also requires the following background assumption: that contact lenses improve the vision of a person who does not otherwise have normal vision.But this assumption clearly did not form part of the context which the hearer had available before he started the interpretation of (12ii).Rather, this assumption (along with other assumptions deriving from the hearer's encyclopaedic information about contact lenses) is added to the context, or conversational background, only as part of the process of interpreting (12ii) itself.This shows that context is not something "given" and "fixed" in advance, available at the start of the interpretation of an utterance.Instead, it suggests that conte~t is constructed as part of the interpretation pro.cess.
If we now add this third shortcoming of the "context is the conversational background" conception to the second shortcoming identified above, it should be clear that this is a very unsatisfactory conception of context indeed.On the one hand, this conception of context leads to the inclusion of a great deal of irrelevant information in the co'ntext, while, on the other hand, it still fails to include all the relevant information.It seems then that this conception of context, rather paradoxically, includes both too much and too little.
We can use the fish example to highlight yet another shortcoming of the "context is the conversational background" answer to the "what is context?" question.In the discussion above of the assumptions made available to the context via the concept 'fish' the focus fell on s1;ereotypical knowledge about fish.But an individual could also have non-stereotypical knowledge about fish which may affect the interpretation of an utterance such as (12ii).Suppose" for instance, that the addressee of Mary's utterance knows that fish is a food loathed by Mary.Then this assumption will also be added to the set of assumptions forming the context, or conversational background, of Mary's utterance.Also, in addition to the assumption about contact lenses improving vision, the information about contact lenses "activated" by the reference to the contact lenses in the (12ii) will add to the "conversationELIbackground" the assumption that it is time-consuming to insert contact lenses.Now, had the hearer used the assumption that JI1aryloathed fish and the assumption that inserting contact lemses is time-consuming when interpreting (12ii), he would obviously have come up with an entirely different interpretation, namely that Mary would not have bothered to come for breakfast if she had known that there that there was fish for breakfast.This illustrates that context cannot be equated with a fixed body of uniquely determined information available at the time of utterance.To put it simply: it shows that context is not given.Rather, it is chosen or selected by the hearer as part of the interpretation process.This then constitutes another reason why the claim that context is the conversational background does not provide an adequate general answer to the "what is context?" question.
To summarize the findings of this discussion of three question that were highlighted by the preceding discussion.
First: an adequate notion 'context' must be defined with reference not to the external environment, but with reference to the environment "inside the individual's head".That is, an adequate notion 'context' should be a psychological notion.
Second: an adequate notion 'context' must permit a wide variety of assumptions to function as context for the interpretation of utterances.These include: assumptions derived through perceptual processing of the external environment, all kinds of encyclopaedic information stored in memory, including social and cultural knowledge, as well as assumptions added to the memory through the processing of preceding discourse.
Third: an adequate notion 'context' should provide a sufficiently restricted characterization of the context for the interpretation of a specific utterance, so that it excludes information which does not really have anything specific to do with the interpretation of an utterance.
Fourth: an adequate notion 'context' must be reconcilable with the fact that the context for the interpretation of an utterance is not fixed in advance of the interpretation of the utterance.
Fifth: an adequate notion 'context' must be reconcilable with the fact that the context is not "uniquely determined", but that it is subject to choices on the part of the addressee throughout the interpretation process.Does linguistics, and pragmatics in particular, then offer us a more adequate answer to the "what is context?" question than the three answers reviewed above?Specifically, does it offer an answer that meets at least the five requirements identified above?I would like to introduce you very briefly to a recently developed theory of pragmatics of which it is claimed that it has indeed led to greater insight into the role of context in utterance interpretation.
The theory in question is called "relevance theory".
Relevance theory was developed by Dan Sperber and Deirdre Wilson in a series of publications since the early eighties.A third point: Relevance theory claims to be able to account not only for the successful communication of determinate aspects of communication, but also for the less determinate, vaguer aspects of communication so prevalent in literary interpretation, for example.In this account, too, the notion 'context' plays the central role.31A fourth point: Although I have not considered the second "context"-question -"how does context work?" -at all in this paper, it is worth noting that relevance theory also provides a detailed answer to this important question.Here too the theory claims that general principles of cognition determine how context interacts with the linguistic meaning of an utterance to determine its interpretation.This Dleans, amongst other things, that the theory claims that context determines all the various aspects of utterance interpretation listed above in the same way.32A fifth point: Relevance theory is claimed to be a generat heory, applicable to all forms of discourse.This means, amongst other things, that relevance theory is a theory of the interpretation of single sentence utterances, as well as of multi-sentence utterances, that is, discourse.Moreover, the theory is applicable to all the various forms of discourse that are distinguished: formal and informal, spoken and written, planned and unplanned, and so on.Blass (1990:41-42)  In section 1 the issue was raised of whether there is any connection between the answer to the interpretation question (1) for single sentence utterances and the answer to this question for multi-sentence utterances, that is, discourse.These comments by Blass imply that, for relevance theory, these two answers are indeed closely connected.In fact, one could say that for relevance theory the problem of accounting for the interpretation of single sentence utterances is simply a special case of a more general problem, namely that of accounting for the interpretation of all the various forms of discourse.I have no hesitation in stating that, for anyone interested in the role of context in utterance interpretation both the single sentence and the multi-sentence variety it will be worthwhile to become acquainted with relevance theory, as well as with some of the descriptive/analytical work done within the framework of this theory.And I specifically include here the various language professionals who have to deal in some way or another with the effects of context on utterance interpretation.
But note that by claiming that recent developments in pragmatics can provide these language professionals with a better understanding of the role of context in utterance interpretation, I am not claiming that pragmatics will also provide the professionals with ready-made solutions to their practical problems.The connection between, on the one hand, theories about language and, on the other hand, solutions to practical problems faced by language professionals, is virtually never simple or straightforward.A good theory could provide the professional with greater insights into the parti.cularaspect of language in which he/she is interested.But to "translate" these insights into workable solutions to practicl!llproblems will require a great deal of hard work on the part of the language professional.There is, moreover, no guarantee that any specific theoretical insight will be translatable into a practical solution for some problem.
Language professionals who turn to pragmatics for greater understanding of the role of context in utterance interpretation also have to accept that in pragmati<:s,like most other subfields of linguistics, there is little consensus.Like other subfields of linguistics, pragmatics is characterized by theoretical diversity.Moreover, there is a great deal of debate about the merits and shortcomings of the various proposals on offer.In the case of relevance th~ory, in particular, a rather fierce debate has developed about the merits of the theory, with participants beirig widely divided in their opinion on the merits of the theory.39All of this means that the language professional who turns to pragmatics, and to relevance theory in particular, in the hope of obtaining final, definitive answers to questions about the role of context in utterance interpretation, will be disappointed.But this most certainly does not warrant the conclusion that the language professional can safely ignore recent work in pragmatics on the role of context in utterance interpretation.At the very least, an acquaintance with this work will help the language professional to understand how extensive and complex the effect of context on utterance interpretation is.
Consequently, the language professional will be in a better position to recognize the shortcolnings of.various simplistic answers to questions 'about the ro,le of context in utterance interpretation.Moreover, a look at some of the recent descriptive/analytical work done within the framework of relevance theory will make it clear that vague, general statements about the role of context in utterance interpretation are unlikely to lead to an increased understanding of this important matter.Such understanding will come about only as a result of rigorous, detailed analyses of the role of context in the interpretation of utterances, analyses that are based on clear theoretical definitions of key notions such as 'context'.
To conclude, then.Pragmatics cannot provide the language professional with final, definitive answers to questions about the role of context in utterance interpretation, nor with readymade solutions to the practical problems faced by the language professional in this regard.What pragmatics can do, though, is to provide the language professional with valuable insight into the role of context in utterance interpretation.
12 In this connection, note that if context plays a significant role in the comprehension of discourse, this will affect the measurement of the readability of texts.Specifically, it will mean that the readability of texts will not be determined in full by measures defined in terms o~purely structural elements of a text.Rather, an important aspect of readability will be how well the text takes into account the "context" which the reader brings to the text.See Huckin 1983 for a discussion of a cognitivist approach to readability that takes into account, for instance, the reader's prior familiarity with the subject matter.
13 This line of argument of course begs the question of whether these conceptions of 'context' do not have their origins in linguistics anyway.But I will not pursue this matter any further here.
14 Blass (1990:29-30) mentions Malinowski, Halliday, Firth, Labov, and Hymes as examples of linguists who adopt such a conception of context.The discussion that follows on the shortcomings of the idea that context should be characterized as situational context is based on the discussion by Blass (1990:30-31).
15 See Sperber and Wilson 1986:133 for a discussion of this hypothesis.
16 See Blass 1990:8, 73 for a similar argument about the effect of context on the interpretation of isolated utterances.17 See Blass 1990:73-4 for an illus~ration of how context also determines the interpretation of the first sentence of a discourse.
18 For a more detailed discussion of this particular example, see Sinclair and Winckler 1991:55-60.
See Sperber and Wilson 1986:133-134 for another illustration of the fact that the context for the interpretation of an utterance contains more than the assumptions explicitly expressed by the preceding discourse.
19 See Sperber and Wilson 1986:133ff.for an extensive discussion of this view.The discussion that follows is a simplified version of their argument against this position.I make use of different examples, however.20 See Levinson 1989:464 for this term.
21 Sperber and Wilson (1986:134ff) set out a detailed argument against this position.
22 This example is a slightly amended and expanded version of an example first given by Blakemore (1988:235).

( 7 )
The laboratory situation Imagine that at the time of utterance the speaker and hearer are discussing a biology experiment for which a small mammal is required.Several specimens have been suggested by people involved in the ~xperiment, including a bat proposed by an individual called Peter.Acknowledging that context is an important factor in utterance interpretation is of course only a first step towards a general theory of discourse interpretation.Such a theory will have to provide detailed answers to at least the following two questionstry (inter alia) to develop answers to the "what is context?" question and the "how does context work?" question.6But the concern with the role of context in utterance interpretation, and discourse interpretation more generally, extends much beyond the domain of the theoretical .linguist.In fact, every language professional 'who is concerned with the effective use of language in communication is somehow or another concerned with these issues.The list of language professionals who belong to this class is quite long.The list includes, amongst others, the following:

Stellenbosch*
Papers in Linguistics, Vol. 25, 1992, 103-132 doi: 10.5774/25-0the literature teacher, who has to deal with interpretation problems faced by students who come backgrounds which differ a great deal from that of writer of a literary text,11 and, of course, the from the * all kinds of "professional" communicators -speakers and writers.12 play a role in utterance interpretation are not derived from the Stellenbosch Papers in Linguistics, Vol. 25, 1992, 103-132 doi: 10.5774/25-0-78 external environment by means of visual or auditory perception.The context for the interpretation of utterances include assumptions derived from an individual's memory, including social and cultural assumptions, and various assumptions derived from encyclopaedic information contained in the memory.Consider again the interpretation of (2) Peter's bat is too big in the cricket situation sketched above.Information derived through visual processing of the physical environment clearly plays a role in the interpretation of this utterance.Bu.t encyclopaedic knowledge about cricket retrieved from the hearer's memory also plays a significant role in the interpretation process.To "know" that Peter is playing cricket at the time of utterance, an individual both has to see what is going onperceptual processing of the physical environment -and he has to interpret what he sees in the light of certain encyclopaedic knowledge about cricket -information recovered from his memory.
Linguistics, Vol. 25, 1992, 103-132 doi: 10.5774/25-0-78 'context'.As regards the second requirement: it restricts the context to the actual assumptions used by the hearer in the interpretation process, thus avoiding the problem of a context that includes too much.As regards the third requirement: given the heterogeneous nature of an individual's cognitive environment from which the context for the interpretation of an utterance is selected, it automatically makes provision for a large variety of assumptions to function as contextual assumptions.A closer study of the theory will show that it also meets the fourth and fifth requirements outlined above.Relevance theory's answer to the "what is context?" question thus represents an improvement on the three popular conceptions of context I have reviewed earlier.For this reason the theory deserves the attention of all those interested in the role of context in discourse interpretation.Let me briefly mention a few other interesting points in connection with relevance theory's account of the role of context in utterance interpretation.utterances is context bound.In terms of relevance theory, there is no utterance intEtl:Qretationwithout context.29A second point: Relevance theory claims that context is involved in many aspects of the interpretation of an utterance.The list of aspects of utterance interpretation which relevance theory claims are context bound is quite long, and it includes30: * recovering any possible stylistic effects, including poetic effects.
provides a very clear statement on the status of relevance theory as a general theory of discourse interpretation: "Relevance theory when construed as a pragmatic theory, takes the whole of communicative discourse, planned and unplanned, formal and informal, connected and unconnected, as its domain, and shows that the Stellenbosch Papers in Linguistics, Vol. 25, 1992, 103-132 doi: 10.5774/25-0-78 principles involved in understanding it are essentially the same."33

A
sixth point: Relevance theory has stimulated some detailed analyses of the role of context in various specific aspects of discourse interpretation including translation,34 the interpretation of literary texts,35 the role and function of discourse connectives,36 the role and function of discourse particles,37 and the interpretation of conditional sentences.38 b. Mary does not want to stay awake.
Stellenbosch Papers in Linguistics, Vol. 25, 1992, 103-132 doi: 10.5774/25-0-78 Let me briefly give a few arguments why such a conception of context also fails to provide us with an adequate general answer to the "what is context?" question. 21