24/7 writing help on your phone
This essay will compare the conversational maxims of Grice with Sperber and Wilson’s Relevance theory, concluding that Grice has the more useful approach, and bodes well for the future of Natural Language Processing. The two theories attempt an analysis of language that goes beyond mere syntax or semantics – to discover the pragmatic meaning conveyed by a sentence, above and beyond the truth-conditional meaning of what is said. Crudely, pragmatics is the process of relating a sentence to the context in which it occurs, the context being either linguistic or non-linguistic.
In the former, the meaning of a sentence may depend upon surrounding sentences – both preceding it and those whose meaning may be influenced by the sentence in question. So we might determine the reference of pronouns: Anthony Burgess used to avoid Monday-morning Linguistic lectures.
His tutors scarcely saw him all term. Whereas non-linguistic contexts provide knowledge about the person producing the utterance, and what they wish to achieve from the communication:
Has anyone seen Anthony Burgess? (a) ie, Has anybody visually perceived him? (b) A sarcastic comment that Burgess is rarely in his lectures And so on.
Pragmatics, and semantics, need recourse to knowledge about the world or the domain modelled. There is, then, a distinction to be made between the literal content of a sentence and the context-dependent meaning of the utterance. With this background, we can examine Grice’s notion of conversational implicatures, which he first formulated in the 1960s but revised much later in Studies In the Ways of Words (1989).
The first general feature of human conversation is that rational agents engage in conversational implicature via cooperation, on the understanding that this form of co-operative exchange is to their mutual benefit; or at least that there is a mutually accepted direction to the conversation. Such an understanding is, according to Grice, achieved by the agents following a co-operative principle: “Make your conversational contribution such as is required, at the stage at which it occurs, by the accepted purpose or direction of the talk exchange in which you are engaged…” Bound within this are several maxims which combine to guide conversation along the most satisfactory lines.
Grice outlined four fundamental maxims: Maxim of Quality Try to make your contribution one that is true: (i) do not say what you believe to be false (i) do not say that for which you lack adequate evidence Maxim of Quantity (i) make your contribution as informative as is required for the current purposes of the exchange (ii) do not make your contribution more informative than is required Maxim of Relevance Make your contributions relevant Maxim of Manner Be perspicuous, specifically: (i) avoid obscurity of expression (ii) avoid ambiguity (iii) be brief (iv) be orderly These maxims are, it must be noted, an ideal of conversation. There may be a whole host of reasons why some are not always observed – environment, speaker, participants, motives, and so on.
Maxims can clash (eg, the maxim of Quantity versus the maxim of Quality) or openly disregarded. Yet they serve as strong, tangible guidelines for human discourse. This accounts for poetical metaphors, hyperbole and analogies which give language its richness and beauty, and also explains tautologies such as Moore’s Paradox. These maxims can be used to elaborate the conditions in which conversational implicature is evident. Here, an implicature is an inference of sorts, where the agents in a conversation make an inference which is invariably ambiguous in semantic terms. Consider again the example of Anthony Burgess attending lectures – if X utters -“Is Anthony in today?” Y may ask “What are you implying by that statement?”
Alternatively, the implication may be clear to X because of something mentioned in an earlier conversation – or because of the concept of mutual knowledge, where X and Y both know something about the situation. Grice is keen to note that the literal meaning of the above if very different to the implicatue which is so apparent – its contextual meaning. Implicature, then, appears to bridge the gap between what is said and what is actually meant. An implicature is worked out in the following way: “0 S has said that p (ii) there’s no reason to think S is not observing the maxims or at least the co-operative principle (iii) in order for S to say that p and be indeed observing the maxims or the co-operative principle, S must think that q (iv) S must know that it is mutual knowledge that q must be supposed if S is taken to be co-operating (v) S has done nothing to stop me , the addressee, thinking that q (vi) therefore S intends me to think that q, and in saying that p has implicated q.” (pp. 113/14 in Levinson, 1983)
For Grice, an implicature has several features: it is defeasible (can be cancelled, almost inductively), non-detachable, non-conventional, calculable and indeterminate. One can sarcastically say: “Anthony is never late for lectures” or “Anthony is always on time for his lectures” – and the implicature remains despite the re-wording. Conversational implicatures are also carried out by the utterances themselves, rather than by the sentences. Grice distinguishes between generalized implicature, and a second kind that arises when his maxims are intentionally violated. A generalised implicature might be: “Anthony likes lectures”
Which will imply that “I believe that Anthony likes lectures, it is true that he likes lectures, and I have sufficient evidence to show it to be the case.” But a maxim can be flouted, as in a response to the question – “Does anybody know where Anthony is?”: “Yes, he is on a bus which is currently travelling along Oxford Road, situated in Manchester, in the country of England, which forms part of the Western Hemisphere…” An answer which clearly flouts the maxim of Quantity. Sperber and Wilson (1986) posit a rival explanation known as Relevance Theory. They argue, from their psychological and linguistic perspectives of human cognition, that communication can be achieved using a code model and an inferential model. In the code model, a process of encoding and decoding takes place, where a signal is taken as an input, and a message is produced accordying to an underlying code.
The decoded content of an utterance combines with what is mutually manifest to the agents, ie, their context, which inputs to the inferential model. This conveys the speaker’s intentions. In the latter, a set of premises are taken as input, and a set of conclusion are derived from them as output. Both types of processing are utilised in verbal communication. The relevance is attained because humans fix their attention towards information that appears to be relevant. Sperber and Wilson insist that the act of communicating is to grab this attention of the other agents. Relevance here is based around human cognitive processes, which, they insist, try to achieve the greatest cognitive effect using the least cognitive effort. In a discourse, participants select the most relevant interpretation according to this definition – andeach utterance possesses just one such interpretation.
Contextual effect is vital here – if an assumption has a contextual effect in a context, then it is relevant in that context, so there is an established link between the assumption and the context. One must also note the processing effort which brings about the contextual effect – they argue that “the mind assesses its own efforts and their effects by monitoring physio-chemical changes in the brain”, which has an obvious parallel in connectionism. They also talk of the cognitive environment which affects the analysis of information. They define it as “a set of facts that are manifest to…” an agent. Here a fact is manifest to an individual at a certain time, but only if the individual can represent it mentally and can accept this representation as being true or probably true.
Sperber and Wilson argue that a communicator intends to make certain assumptions manifest to his audience, by modifying his cognitive environment – and not by directly modifying their thoughts. How different are the two approaches? Grice approaches language from philosophy; Sperber and Wilson from psycholinguistics, so there is an obvious contrast from the beginning. Their talk of cognition indicates a move towards a pragmatics with a solid psychological base; whereas Grice takes up the philosophical challenge to analyse ‘ordinary’ language. Grice is also subject to criticism from the two psychologists – for example, his idea of mutual knowledge is implausible in psychological terms – for to be manifest is weaker than to know or assume something, so is less implausible.
Instead Sperber and Wilson posit a theory of mutual manifestness, which does indeed sidestep the infinite regress of Grice, and appeal to ostensive acts which alter the cognitive environment of both speaker and audience. Grice’s notion of mutually accepted assumptions can also be criticised – in order for both agents in a discourse to know they share mutual knowledge, these assumptions need to be drawn by them both, and they both need to know this fact as well, and so on ad infinitum. Since a cognitive environment does not get caught in an infinite number of assumptions, Sperber and Wilson state that it can explain situations where information is exploited in a theory, unlike the Gricean perspective.
They also claim that communication in general has the aim of increasing “the mutuality of the cognitive environment…rather than guarantee… strict duplication of thoughts…”, which of course is Grice’s assertion of what implicatures achieve. By claiming that the speaker only wants to alter his own cognitive environment, not the thoughts of the hearer, Sperber and Wilson escape this. Sperber and Wilson, remember, posit two models – code and inferential, both of which are essential to explain human discourse. Grice’s maxims seem to use just the inferential model, by inferring a set of conclusions from a set of premises.
Moreover, their principle of Relevance suggest the existence of heuristics, some innate, some acquired via experience, and it implies that there is a degree of relevance in all communication. This can be worked out from contextual effect and processing effort. Grice merely appeals to us to be relevant’ in his maxim of relation; further, that norms are acquired and need to be known for proper communication. In contrast, Sperber and Wilson argue that communicators do not follow their principle of relevance: indeed, they could not violate it even if they so desired.
The principle of relevance is always communicated, even if the communicator fails to be relevant. Grice argues that ironies can be successfully recognised and understood, eg:. ‘Life is Life. As a tautology, this breaks the maxim of quantity – it communicates no information and is therefore not being informative. The relevance theory of Sperber and Wilson lack the language games permitted by Grice. This is because the communicator does not adhere to an explicit principle of relevance, since ‘communicating’ implies that the information is relevant. Thus a speaker cannot deliberately flout the principle of relevance. It is however possible that the communicator does not manage to make his utterance seem relevant to the addressee.
Adler criticised Sperber and Wilson for merely simplifying Grice’s first 3 maxims in their postulation of contextual effects, which seems a fair argument. He also insisted that their ‘processing effort’ was effectively Grice’s maxim of manner and co-operative principle. Morgan and Green similarly argue that they merely subsume Grice’s maxims. However, they cannot capture truthfulness since it has been subsumed into one large principle; whereas Grice can give properties to certain maxims, by giving the maxim of quality the highest priority. Where they fundamentally break from Grice is over his metaphors and irony.
They say there is no discontinuity between literal uses and metaphors – it is simply a question of interpretation, and there are no intermediate steps since they are processed directly depending on the relevant meaning. Grice retorts that irony and metaphors do indeed break his maxim of quality – yet communicators get around this, because people understand metaphors apart from their literal useage, so that the statement – “He eats like a horse” does not literally mean the perceived human is eating straw from a nosebag, and it is not misinterpeted along these lines either. Again, this seems dependent upon the context to at least some extent. However, in conversation we must assume that the maxims are not being broken (or at least not all of the maxims are being broken).
To say that “Anthony Burgess is never late for lectures.” could imply that the speaker is defending his friend against accusations of his being late, or that the speaker is in fact a lecturer who is employing sarcasm. It is the context which determines the interpretation of such statements, and this indicates that the utterance is first interpreted literally before it is interpreted metaphorically. This contrasts with the theory of Sperber and Wilson, who claim that a listener will select a context which will maximizes the relevance of an utterance. This reverses the notion that relevance is a ‘variable’ which needs to be calculated – in fact it is the context which is the variable, and stems from their reliance upon relevance being, at least in part, innate.
They make an interesting point that Grice may be employing an “ex post facto” argument, where the chosen interpretation of an utterance is justified by an appeal to context, the utterance itself and general expectations of the behaviour of the conversing parties. However, as Sperber and Wilson note, in some cases, “…an equally convincing justification could…have been given for some other interpretation that in fact was not chosen…” We may also consider the differing conceptions of inference involved here. Although both rely on it, Grice appeals to a defeasible inductive implicature. Sperber and Wilson, in contrast, argue that inference is nondemonstrative, ie, the truth of the premises only make the truth of the conclusion probably true. Further, they wish to say that deduction is the key to analysing non-demonstrative inference, which includes implicature.
However, they admit to relying upon an unorthodox system of logic to arrive at this deductive model, and this is always a precarious move, which is invariably found lacking. An interesting attempt to model language is worthy of mention. It tackles the maxims of Grice – not least because Sperber & Wilson themselves confess that the vagueness of their theory makes computer implementation difficult. The model is known as the ‘Genial Understanding System’ (GUS), and was modelled on pragmatics by Bobrow et.al (1977).
Developed in the mid-1970s, the primary interface for GUS is natural language. GUS carries out a task-oriented dialogue with a user who is planning a round trip to a city in California. It contains strong expectations as to the response most users would give to its questions, and can understand vague, and even incomplete, answers. It seems to follow the Gricean maxims which is largely achieved by task-specific knowledge, for example what trip specifications are like. The computer simulates a co-operative and sympathetic human operator such as: “Hello, my name is GUS. I can help you plan a simple trip by air. Where do you want to go ?” The human would give relevant details of the planned trip in response to questions phrased in natural language: “Would you like PSA flight 102 that arrives at 9.15 am.?” until eventually we reach the following position: “I have confirmed the following flight : PSA flight 102 on Wednesday May 28 from San Jose to San Diego that leaves at 7.30am. and arrives at 9.15 am.”
Gricean maxims are clearly put to good use here, with the computer being relevant, truthful, perspicuous, informative, and so on – for example, it does not list all the flights entering San Diego before 10am. The success of such an implementation may answer some of the criticisms from Sperber and Wilson, that Grice is too abstract. It can also be seen, that, despite their protestations to the contrary, Sperber and Wilson offer no real advancement over Grice’s maxims, and in fact suffer in comparison to him. Relevance theory is much more vague than its rival, and ultimately fails in theoretical and practical terms; whereas Grice has produced a series of conversational guidelines which illuminate the complexity and mystery of language.
👋 Hi! I’m your smart assistant Amy!
Don’t know where to start? Type your requirements and I’ll connect you to an academic expert within 3 minutes.get help with your assignment