Version classiqueVersion mobile

Resemblance and Representation

 | 
Ben Blumson

5. Symbol Systems

Texte intégral

1It’s often argued that the theory of language ought to be subsumed by a general theory of representation, of which language is merely one kind. Ferdinand de Saussure, for example, writes “A language is a system of signs expressing ideas, and hence comparable to writing, the deaf-and-dumb alphabet, symbolic rites, forms of politeness, military signals, and so on. It is simply the most important of such systems. It is therefore possible to conceive of a science which ... would investigate the nature of signs and the laws governing them” (Saussure, 1972, 15). Language, according to Saussure, is merely a special case of representation in general.

2In a similar vein, Charles Peirce writes, “… there are three kinds of representations. 1st Those whose relations to their objects is mere community in some quality, and these representations may be termed Likenesses. 2nd Those whose relations to their objects consists in a correspondence in fact, and these may be termed Indices or Signs. 3rd Those the grounds of whose relation to their objects is an imputed character, which are the same as general signs, and these may be termed Symbols” (Peirce, 1868, 7). However, Peirce continues “… the rules of logic … have no immediate application to likenesses or indices, because no arguments can be constructed from these alone” (Peirce, 1868, 7).

3Lack of success has led to pessimism. Gilbert Harman, for example, writes “… there is no ordinary sense of the word ‘mean’ in which a picture of a man means a man or means that man. This suggests that Peirce’s theory of signs would comprise at least three rather different subjects: a theory of intended meaning, a theory of evidence, and a theory of pictorial depiction. There is no reason to suppose these three theories would have any principles in common” (Harman, 1977, 214). Language, according to Harman, has little in common with other kinds of representation, so the study of language cannot be subsumed under a study of representation in general.

4The last chapter argued for an analysis of depictive symbol systems analogous to the analysis of languages in use. But not all symbol systems are depictive or descriptive: as well as languages and pictures, there are, for example, traffic lights, musical scores, graphs and dewy decimal numbers. This suggests that the analyses of depictive and descriptive symbol systems are special cases of an analysis of symbol systems in use in general. This chapter argues that a symbol system is in use in a group if and only if the group conforms, has reason to conform and prefers others to conform if most do to a regularity of speaker meaning in the symbol system, and if they believe all this.

5The first section addresses some objections to the necessity and sufficiency of Lewis’ analysis of languages in use, and argues for replacing regularities of truthfulness with regularities of speaker meaning. The second section argues for a generalisation of this analysis and the analysis of depictive symbol systems which drops the clauses which are unique to each while retaining the clauses which are common to both. The third section uses this analysis to argue that not all depictions belong to symbol systems in use, and so depiction is not a kind of symbol system. The fourth section addresses an objection concerning non-literal language, and the fifth concludes.

5.1 Analysis of conventional language

6A language is used by a population, according to Lewis’ original analysis (1969, 177), if and only if there’s a convention of truthfulness in that language in that population or, in other words, if and only if there’s a convention of sometimes uttering sentences of the language in the indicative mood while trying to abstain from uttering indicative sentences which are false in the language, and trying to make true in the language sentences of the language addressed to one in the imperative mood, if those sentences are uttered by an appropriate agent – such as someone in authority, someone one can trust or someone whom one wants to please (Lewis, 1969, 184).

7But there are counterexamples to both the necessity and sufficiency of Lewis’ original analysis. The analysis is not necessary because a population of inveterate liars may have a language of use in which there is no regularity of truthfulness, since although they often utter sentences of their language, they never do so while trying to abstain from uttering sentences of their language which are false, but actively try to say sentences which are false, and although imperatives in their language are addressed to them by appropriate agents, they never try to make those imperatives true, but go out of their way to try to make them false (Lewis, 1969, 195; 1975, 182).

8The analysis is not sufficient because every conventional regularity of behaviour in a population can be redescribed as a conventional regularity of truthfulness in some language in the population. If garbage is conventionally collected on Tuesday, for example, there’s a conventional regularity of truthfulness in the language in which collecting garbage is an indicative sentence meaning that it’s Tuesday, since people sometimes collect garbage but abstain from collecting garbage when it’s not Tuesday. But the language in which collecting garbage means that it’s Tuesday is obviously not a language in use (Schiffer, 1993, 233).

9Liars have a language of use in which they are not truthful and garbage collectors are truthful in a language which is not their language of use, because garbage collectors do not mean anything by collecting garbage whereas liars mean something by lying. This suggests that languages in use should be defined, not in terms of conventional regularities of truthfulness in a language, but in terms of conventional regularities of speaker meaning in a language (Schiffer, 1993, 233). In other words, this suggests the original plan of analysing speaker meaning directly in terms of intention, and sentence meaning jointly in terms of convention and speaker meaning.

10A regularity of speaker meaning in a language, is a regularity of uttering sentences in the language only if they mean what the speaker means by uttering them (Schiffer, 1987, 252). In other words, speaker meaning in a language is uttering sentences in the language approximately only if one intends an audience to believe, make it the case, imagine or have some other attitude towards the truth-condition of the sentence in the language, by means of recognition of one’s intention. A regularity of speaker meaning in English, for example, is a regularity of uttering ‘the door is shut’ only if one means the door is shut, ‘shut the door’ only if one means shut the door ... and so on.

11So a language is used by a group, according to this suggestion, if and only if there’s a conventional regularity of speaker meaning in the language, or a convention of uttering sentences in the language only if they mean what their speaker means by uttering them. English, for example, is supposed to be our language of use because we have a convention of speaker meaning in English or, in other words, a convention of uttering English sentences only if their meaning in English is what we mean by uttering them: we utter ‘the door is shut’ only if we mean that the door is shut, ‘shut the door’ only when we mean the door to be shut, ... and so on.

12One clarification. The conditional in this analysis shouldn’t be interpreted as material, since if it were, the conditions of the analysis would be trivially met if the antecedent of the conditional is false. The convention of silence during the eleventh minute of the eleventh hour of the eleventh month, for example, would count as a convention of speaking every language at that time, since by not uttering sentences of any language, speakers utter those sentences only if they speaker mean what the sentences mean in the language. But the convention of silence during the eleventh minute of the eleventh hour of the eleventh month is obviously not a convention of speaking any language.

5.2 Analysis of symbol systems in use

13The natural generalisation of this analysis of language in use to symbol systems is: a symbol system is in use if and only if there’s a convention of speaker meaning in that symbol system, where speaker meaning in a symbol system is perpetrating characters of the symbol system only if they represent a state of affairs their perpetrator speaker means by perpetrating them. Traffic lights, for example, are in use because there’s a convention of speaker meaning in the symbol system of traffic lights: the lights are red only if we are meant to stop, green only if we are meant to go and orange only if we are meant to be careful.

14This analysis accommodates many depictive symbol systems without amendment (Bennett, 1974; Novitz, 1977). Maps, for example, are a symbol system in use since there’s a regularity of speaker meaning in the symbol system of maps, there’s reason to conform to this regularity if others do, there’s an alternative regularity – speaker meaning in the symbol system in which the bottom instead of the top of a map represents north – which there would be reason to conform to if others conformed to it, others are preferred to conform to this regularity if most do, and it’s believed that all these conditions obtain. In other words, maps are used in virtue of a convention of speaker meaning.

15But depiction is still a counterexample to the necessity of this analysis, because some depictive symbol systems are not mediated by convention, even in part. The symbol system of colour charts, for example, in which colour samples are used to represent colour shades, is a symbol system in use. But the regularity of speaker meaning in the symbol system of colour charts is not conventional, because there’s not an alternative regularity which we would have reason to conform to if others did: the symbol system in which samples are the same colour as the shades they represent is the only symbol system which allows the represented shade to be seen by looking at the sample.

16Globes, for example, also seem to belong to a regularity of meaning in a symbol system which lacks any equally good alternatives. Mercator projections, for instance, are not as good as globes because they inevitably distort the Earth’s shape by making the poles seem larger than they actually are; other systems of projection produce other distortions. Any globe, of course, could be replaced by a larger or smaller globe, but that replacement would not amount to a change in the symbol system being used but just to another representation in the same system. So depictions and depictive symbol systems need not conform to the analysis of convention (Lopes, 1996, 132-135).

17Depiction is not the only illustration of the point that not all symbol systems in use are mediated by convention. Imagine, for example, a species incapable of learning, which knows the grammar and vocabulary of its language innately. For this species, there’s no alternative language in which a regularity of speaker meaning would be preferred if it were preferred by others, because everyone in the population prefers to speak the only language they know. In this case, the species’ language of use is mediated not by convention, but by psychological necessity (Peacocke, 1976, 169). Human language is mediated by convention, but this is contingent, and not a matter of necessity.

18The fact not all symbol systems are mediated by convention might be taken to suggest that symbol systems should be analysed in terms of any, instead of only conventional, regularities of speaker meaning, as follows: a symbol system is in use if and only if there is a regularity of speaker meaning in that symbol system. So the symbol system of traffic lights, for example, is supposed to be in use just because there’s a regularity of speaker meaning in the symbol system of traffic lights or, in other words, merely because the lights are red only if we’re meant to stop, green only if we’re meant to go and orange only if we’re meant to be careful.

19If this analysis were sufficient, then every depiction would belong to a symbol system in use, because there is a regularity of speaker meaning in the symbol system defined by the function which takes every depiction to the state of affairs its audience is intended to believe, imagine, bring about ... or so on, by means of recognition of this intention. Because Leo intended the Mona Lisa, for example, to induce in us the belief that Lisa smiles by means of recognition of his intention, Leo conformed to a regularity of speaker meaning in a symbol system which takes the Mona Lisa to the state of affairs of Lisa’s smiling.

20But this analysis entails that every depiction belongs to a symbol system in use only because it entails every arbitrary function from utterances to what is meant by them is a symbol system in use. There’s a regularity of speaker meaning, for example, in the language defined by the function taking English sentences to what English speakers mean by them and German sentences to what German speakers mean by them, since English speakers utter ‘it’s raining’ and ‘es regnet’ only if it’s raining, ‘snow is white’ and ‘schnee ist weiß’ only if snow is white, ... and so on for all English and German sentences.

21But the function taking English sentences to what English speakers mean by them and German sentences to what German speakers mean by them obviously isn’t a language in use, because the fact that there’s a regularity of speaker meaning in this language is simply a by-product of the fact that there’s a regularity of speaker meaning in English and also a regularity of speaker meaning in German. In general, it’s not sufficient for a symbol system in the abstract to be a symbol system in use that there’s a regularity of speaker meaning in that symbol system, because that regularity might simply be a by-product of the use of other symbol systems or disparate acts of communication.

22This suggests dropping from the analysis of symbol systems of use not the analysis of convention as a whole, but just those parts of the analysis of convention which entail that choice of symbol system is arbitrary, leading to the following analysis:

(20) A symbol system is in use in a group if and only if everyone in the group:
a. perpetrates characters of the symbol system only if they speaker mean what the characters represent in the symbol system
b. has reason to perpetrate characters of the symbol system only if they speaker mean what the characters represent in the symbol system
c. prefers other members of the group to perpetrate characters of the symbol system only if they speaker mean what the characters represent in the symbol system, on condition that most do
d. and everyone in the group believes (a)-(d).

23Traffic lights, for example, are in use because the lights are red only if we’re meant to stop, green only if we’re meant to go and orange only if we’re meant to be careful, there’s reason to conform to this regularity, conformity is preferred if most conform, and all this is believed.

24This analysis generalises both the analysis of conventional language and the analysis of depictive symbol systems by retaining the clauses common to both while dropping the clauses which are unique to each. In particular, while it’s still required that there be a reason for conforming to a regularity of speaker meaning in a symbol system, it’s not required that this reason be precedent or the conformity of others as is required by the analysis of conventional language or that the reason be resemblance of characters to their extensions, as in the analysis of depictive symbol systems. An innately known language, for example, may be spoken simply because of psychological necessity.

25Requiring that everyone in the group has reason to conform to the regularity, without specifying what kind of reason, may seem redundant, since if the group conforms to a regularity, it seems they must have reason to conform to that regularity. But this is not the case, since the explanation of a group’s conformity to a regularity may be brutely causal, rather than rational. If the explanation of a group’s conformity to a regularity of speaker meaning in a symbol system is brutely causal, rather than rational, then even if everyone in the group prefers others to conform on condition that most do, and everyone believes all this, the symbol system is not used by the group.

26Suppose, for example, there’s a group of people who sneeze only if they pronounce a certain sentence which tickles their noses and which they use to mean that it’s raining. Then there’s a regularity in the group of uttering sentences of a language in which sneezing means that it’s raining only if they speaker mean what the sentence means in that language. If everyone in the group is proud of any eccentricity they all happen to have, then they prefer the others to conform to this regularity if most do. And they might all believe all this. Even so, a language in which sneezing means it’s raining is not their language of use, since they don’t conform to the regularity for a reason, but merely because of a physiological accident (Peacocke, 1976, 176).

27Since the requirement that everyone in the group prefer others to conform if most do is merely inherited from the analysis of convention, it may not seem necessary. But this is not the case, since the use of a symbol system in a group, like the adoption of a convention, requires coinciding interest: just as a group adopts a convention because they have a common interest in some end which is furthered by the adoption of a common pattern of behaviour, a group adopts a symbol system for use because they have a common interest in communication about a certain subject matter, which is furthered by the adoption of a common symbol system.

28Suppose, for example, that the members of an automobile club conform to a regularity of driving vintage cars only if they mean that they are wealthy, that they have reason to conform to this regularity because they’re prepared to risk the expense of denting their vintage cars only if they mean that they are wealthy, and that everyone in the club believes all this, but they don’t prefer other members of the club to drive vintage cars only if they mean that they’re rich, because they don’t mind other members’ cars being dinted. Since the regularity is not adopted in response to a common interest, the club doesn’t use a symbol system in which driving vintage cars represents wealth.

5.3 Depiction outside of symbol systems

29Whether depiction should be defined as a kind of symbol system is hard to adjudicate: since there is a function which takes every depiction to what it represents, it’s trivially true that every depiction belongs to a symbol system in the abstract. The distinction between symbol systems in the abstract and in use makes adjudicating this question easier, because whereas it’s trivial that every depiction belongs to a symbol system in the abstract, it’s non-trivial whether every depiction belongs to a symbol system in use. The analysis defended in the last section shows not all depictions belong to symbol systems in use, and so depiction should not be analysed as a kind of symbol system.

30The function taking English sentences to what English speakers mean by them and German sentences to what German speakers mean by them isn’t a symbol system in use, according to this analysis, since though English speakers conform to a regularity of uttering sentences in this language only if they mean what the sentences mean in the language, not all English speakers believe this. Although it’s true that people utter ‘es regnet’ only if it’s raining and ‘schnee ist weiß’ only if snow is white, for example, most English speakers don’t believe this – as far as they know, ‘es regnet’ is uttered only if it’s sunny and ‘schnee ist weiß’ only if snow is black.

31For the same reason, this analysis does not entail that every depiction belongs to a symbol system in use. Although there’s a regularity of perpetrating depictions only if the perpetrator speaker means what the depiction represents, this is not believed by all perpetrators of depictions, so condition (d) isn’t met by the function which takes each depiction to what it represents. Not everyone believes Nude Descending a Staircase, for example, is perpetrated only if it’s meant that a nude is descending a staircase – as far as people unfamiliar with cubism believe, Nude Descending a Staircase may only be perpetrated if several knights in armour are filing down a staircase.

32So the symbol system defined by the function which takes each depiction to the state of affairs it represents is not a symbol system in use. But this doesn’t show that not all depictions belong to symbol systems in use, because though there’s no single symbol system in use to which every depiction belongs, it may be true that all depictions belong to some symbol system in use or another, but that different depictions belong to different symbol systems in use. Whereas the Mona Lisa, for example, may belong to a symbol system in use of renaissance painting, Nude Descending a Staircase may belong to a symbol system in use of cubist painting.

33But even if everyone in a group believes there’s a regularity of speaker meaning in a symbol system, the symbol system may not be in use if they don’t prefer others to conform to a regularity of speaker meaning in that symbol system if most do. Cubist painting, for example, may not be a symbol system in use because cubists may not prefer other cubists to conform to a regularity of speaker meaning in that system on condition that most do. If Marcel, for example, prefers his paintings to be original, then he might not prefer that others perpetrate Nude Descending a Staircase only if they mean that a nude is descending a staircase but may prefer them, less originally, to perpetrate it only if they mean that knights in armour are filing down a staircase.

34Moreover, because some depicters don’t believe that there are regularities of speaker meaning in any symbol system, their depictions don’t conform to any regularity of speaker meaning which meets condition (d). It is probable, for example, that the first person to depict something did not believe in any other instance of speaker meaning, in which case the person did not believe in a regularity of speaker meaning in any symbol system. So even though the first depicter conformed to regularities of speaker meaning in many symbol systems, not everyone believed in any of these regularities, because the first depicter believed in none of them.

35It might be objected that single occurrences are degenerate regularities, so that even if only one depiction were ever perpetrated, there’d be a regularity of speaker meaning in a symbol system defined by the function taking that depiction to the state of affairs its audiences is intended to believe, imagine, bring about ... or so on by means of recognition of intention, constituted by the single act of speaker meaning that state of affairs by perpetrating that depiction. In this case, then even the first depicter would have believed he or she conformed to a regularity of speaker meaning in a symbol system, simply in virtue of believing that he or she meant something by the depiction.

36But if single occurrences were degenerate regularities, then even a single behavioural occurrence could count as conventional. If, for example, both of us go to a party in the hope that the other will, there’s another party on the same night we would have been equally happy to go to, both of us prefer to go to the same party as the other, and both of us are aware of all this, then our going to this party would count as a convention, even if we never go to another party our whole lives long. Since single occurrences of this kind are intuitively not conventions, single occurrences should not be regarded as degenerate regularities in the analyses of convention or depictive symbol systems.

5.4 Meaning outside conventional language

37The argument just given that not all depictions belong to symbol systems in use might be thought to prove too much, on the grounds that a similar argument undermines the analysis of languages in use. The problem arises because there’s in fact no regularity of uttering English sentences only if one means what the sentence means in English, because English sentences are often uttered loosely, when their meaning in English merely approximates what one means by them, and non-literally, when their meaning in English is not what one means by them. If I utter ‘John is a fine friend’ ironically, for example, then I mean not that John is a fine friend, but that he’s a false friend.

38One response to this problem is to agree that there is no regularity of speaker meaning in literal English, but argue instead that there is a regularity of speaker meaning in a more complex language, which consists of a function that takes English sentences and contexts to their non-literal rather than their literal truth-conditions or meanings. The function which defines this language would take the sentence ‘John is a fine friend’ in contexts in which it’s uttered ironically to the truth-condition of John’s being a false friend, ‘Juliet is the sun’ in contexts in which it’s uttered metaphorically to the truth-condition of Juliet’s being pretty, ... and so on (Lewis, 1975, 183).

39But this response to the problem is incorrect, because although there is a regularity of speaker meaning in the language defined by the function taking English sentences and contexts to their non-literal truth-conditions or meanings which English speakers have reason to conform to, this function may not meet the third and fourth conditions of the analysis of being a symbol system in use: it may be that not everyone prefers others to conform to a regularity of speaker meaning in this language if most do and it may be that not everyone believes everyone conforms to a regularity of speaker meaning in this language, even though everyone does conform to it.

40If, for example, we all spoke non-literally in order to show off our cleverness by the aptness of our metaphors, then although we would all conform to a regularity of speaker meaning in the language defined by the function taking English sentences in contexts to their non-literal truth-conditions, we would not prefer others to conform to a regularity of speaker meaning in that language: we would prefer others to conform to a regularity of speaker meaning in literal English, in order that the cleverness of our own speech be more impressive in comparison. In this case, the regularity of speaker meaning in non-literal English wouldn’t meet clause (c) of the analysis.

41Moreover, it’s possible we conform to a regularity of speaker meaning in non-literal English, without believing that we conform to a regularity of speaker meaning in non-literal English. It might be that we utter ‘Juliet is the sun’, for example, only if Juliet is pretty, but that each time we utter ‘Juliet is the sun’, we believe that we don’t conform to a regularity of speaker meaning in non-literal English because we believe that that ‘Juliet is the sun’ is uttered on all other occasions only if Juliet is the sun. In this case, non-literal English is not our language in use, because it does not meet clause (d) of the analysis of symbol systems in use.

42If the correct response to this were to weaken clauses (c) and (d) of the analysis, that would undermine the argument that not all depictions belong to symbol systems in use, which relied on those clauses. But the correct response is to weaken the analysis to reflect that speakers may utter a sentence of their language not only if they mean by it what it means in their language, but also if they mean by it something non-literal, which the audience is intended to infer from its literal meaning. Although ‘Juliet is the sun’, for example, is not uttered only if it’s meant that Juliet is the sun, what is meant by ‘Juliet is the sun’ is inferred from the fact that it means Juliet is the sun in English.

43A language in the abstract is a language in use, according to this suggestion, if and only if there’s a convention of taking utterances of sentences in the language as prima facie evidence that the utterer means what they mean in the language. English is our language in use, for example, because there’s a convention of taking utterances of ‘it’s raining’ as prima facie evidence the utterer means it’s raining and utterances of ‘Juliet is the sun’ as prima facie evidence that the utterer means that Juliet is the sun. In the latter case, this prima facie evidence is outweighed, and allows the audience instead to infer that the speaker means that Juliet is pretty (Peacocke, 1976, 172).

44In general, this leads to the following analysis of symbol systems in use:

(21) A symbol system is in use in a group if and only if everyone in the group:
a. takes perpetrations of characters of the symbol system as prima facie evidence that the perpetrator means what the characters represent in the symbol system
b. has reason to take perpetrations of characters of the symbol system as prima facie evidence that the perpetrator means what the characters represent in the symbol system
c. prefers other members of the group to take perpetrations of characters of the symbol system as prima facie evidence that the perpetrator means what the characters represent in the symbol system, on condition that most do
d. and everyone in the group believes (a)-(d).

45Traffic lights, for example, are in use because red lights are prima facie evidence that we’re meant to stop, green lights are prima facie evidence that we’re meant to go and orange lights are prima facie evidence that we’re meant to be careful.

5.5 Conclusion

46So a symbol system is in use in a group if and only if there’s a regularity of taking the perpetration of a character in that symbol system to be prima facie evidence that the perpetrator means what it represents, everyone in the group has reason to conform to that regularity, everyone prefers everyone to conform to that regularity on condition that most do and everyone believes all this. Traffic lights, for example, are in use because there’s a conventional regularity of taking red lights as prima facie evidence that we’re meant to stop, green lights as prima facie evidence that we’re meant to go and orange lights as prima facie evidence that we’re meant to be careful.

47It follows from this analysis that since not all depictions belong to symbol systems in use, depiction should not be defined as a kind of symbol system. Though perpetrating a picture is prima facie evidence that the perpetrator means what it represents in some symbol system, this is not sufficient for the picture to belong to a symbol system in use because it may not be the case that others are preferred to conform to that regularity on condition that most do and it may not be the case that it’s believed that those regularities obtain. Depictive symbol systems are an important, but not the only, kind of depiction.

48It might be objected that this conclusion – that depiction is not a kind of symbol system – undermines the analogy between depiction and language, since the analysis of languages in use encompasses all of the phenomena it’s expected to, whereas the analysis of depictive symbol systems does not. While languages, for example, have their depictive analogues in maps, chess diagrams and other symbol systems, it may seem that depictions which don’t belong to symbol systems lack linguistic cousins, so that the analogy between depiction and description is not as close as the analogy between languages and depictive symbol systems originally seemed to suggest.

49But this objection would be incorrect. Just as some depictions fall under the analysis of depictive symbol systems and others do not, some spoken and written utterances belong to languages whereas others fall outside it. If somebody calls out loudly in alarm, for example, then, whilst their call trivially belongs to some language in the abstract or another, whether it falls under the analysis of languages in use will depend on the preferences, reasons and beliefs of the speaker and their population. Spoken and written utterances of this kind – instances of speaker meaning without sentence meaning – are analogous to depictions not belonging to depictive symbol systems.

50Just as there is a distinction between the meaning of sentences in a language of use and what speakers mean by their utterances, there’s a distinction between the extensions of depictions in a depictive symbol system and what perpetrators represent by perpetrating depictions. The analysis of languages in use is an analysis of the meaning of sentences in a population’s language of use, which is why it excludes examples of speaker meaning in the absence of a language, such as shouts of alarms or ad hoc gestures. By paralleling this analysis, the definition of depictive symbol systems captures the aspect of depiction which is analogous to sentence meaning.

CC-BY-4.0

Le texte seul est utilisable sous licence CC BY 4.0. Les autres éléments (illustrations, fichiers annexes importés) sont « Tous droits réservés », sauf mention contraire.

Acheter

Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search