, Volume 22, Issue 4, pp 417–432

Idiolects and Language


    • Department of Social Sciences and CommunicationUniversity of Salento (Lecce)
Original Paper

DOI: 10.1007/s10516-011-9151-2

Cite this article as:
Chiffi, D. Axiomathes (2012) 22: 417. doi:10.1007/s10516-011-9151-2


The present paper is intended to analyse from a theoretical point of view the relationships between natural language and idiolects in the context of communication by means of the Davidson–Dummett controversy on the nature of language. I will explore from a pragmatic point of view the reliability of an alternative position inspired by the recent literalism/contextualism debate in philosophy of language in order to overcome some limitations of Dummett’s and Davidson’s perspectives on language, idiolects and communication.


Natural languageIdiolectsDummettDavidsonLiteralismContextualism

1 Dummett and Davidson

The philosophies of Davidson and Dummett do not share many ideas on the role of concepts like language, meaning, communication, etc. For instance, Dummett holds a verificationist and molecularistic theory of meaning (recently he prefers to talk of a theory of meaning based on justification), while Davidson holds a holistic and truth-conditional account of meaning. Again, Dummett maintains a full-blooded theory of meaning with a complex linguistic framework that could handle the effective communicative process, while Davidson (in his early works) is interested in a modest theory of meaning and in the condition occurring in principle in communication.1 A full-blooded theory of meaning is expected to explain our ability to apply a word correctly according to some public patterns of use, and without referring to any relationship between speaker and interpreter. Moreover, it must also explain how we grasp new concepts, i.e. a full-blooded theory of meaning is a theory of the actualunderstanding. On the other hand, a modest theory does not explain how we grasp new concepts and it does not refer to external links in the constitution of the meaning.2 Nevertheless, Davidson and Dummett share the idea that the theory of meaning must be based on something more primitive (respectively truth and verification). However, in the following pages more similarities will be found out.

The aforementioned differences between the two philosophers require a deeper analysis that falls partially outside the limits and aims of this paper. Yet, I intend to underline their views on the relationship between idiolects and common language. The main questions are the following:
  1. 1.

    Can communication occur without a shared language?

  2. 2.

    Which one is prior in communication? Language or idiolect?3

Before presenting Davidson’s and Dummett’s views to the questions I and II, I shall spend few words on Davidson’s picture of communication (see Davidson 1986), since it represents the starting point of the dispute. Davidson argues that a theory of communication needs a speaker and a hearer. They communicate with each other by means of two theories: the prior theory and the passing theory. From the hearer’s point of view the prior theory is the way in which he is prepared to interpret every utterance of the speaker, while the passing theory is the way he interprets the utterances of the speaker in a particular context. From the point of view of the speaker, the prior theory is the set of his beliefs concerning what counts as interpreter’s prior theory and his passing theory is the theory that he supposes the interpreter will apply. Davidson maintains that communication succeeds if the passing theory is shared. The prior theory at step n is the passing theory at the step n − 1. If communication does not stop at the step n, then the prior theory at the step n can be updated in a new passing theory at the step n + 1.
The prior theory supports the conditions of possibility of communication and it is not a “long range” theory as Dummett maintained.4 Dummett, in fact, prefers to talk of “long range” theory instead of prior theory and he argues that even Hacking (1986) was wrong in interpreting Davidson:

Hacking takes the prior theory of each participant to be that which he brings to the conversation at the outset, and his passing theory to be that into which it evolves in the course of the conversation (Dummett 1986, p. 459).

According to Dummett, the prior theory is not replaced by the passing theory, i.e. the prior theory is a theory about the way we understand the speaker in general, while the passing theory regards the way we understand some particular expressions in a specific context. The prior theory is not paramount for Davidson, because it is systematic and prepared, and yet it remains unshared, while the passing theory is systematic and shared, but not prepared. One cannot say that the prior theory is the long-range theory in Dummett’s sense, because the short-range theory at step n is not the long-range theory at the step n − 1.5
Only the passing theory offers the possibility that speaker and hearer can share some meanings. Indeed, we read in (Davidson 1986)

what interpreter and speaker share, to the extent that communication succeeds, is not learned and so is not a language governed by rules or conventions known to speaker and interpreter in advance; but what the speaker and interpreter know in advance is not (necessarily) shared, and so is not a language governed by shared rules or conventions. What is shared is, as before, the passing theory; what is given in advance is the prior theory, or anything on which it may in turn be based.

The passing theory is where, accident aside, agreement is greatest. As speaker and interpreter talk, their prior theories becomes more alike, so do their passing theories (p. 442).

Davidson’s argument is that communication can occur even if there are no shared rules, regularities and conventions between the speaker and the hearer. Davidson argues that literal (first) meaning is characterized by the following properties:
  1. 1.

    first meaning is systematic, namely there are well-defined relations between the meanings of the utterances.

  2. 2.

    first meanings are shared, namely there is a common way to interpret some words.

  3. 3.

    first meanings are governed by learned conventions or regularities.

He maintains that there is no room for (3) in linguistic communication. Here is Davidson’s example: imagine a situation where a speaker uses the word “epitaph” in place of “epithet”. This linguistic phenomenon is called “malapropism”. When the interpreter understands that the speaker uses “epitaph” with the intention of referring to “epithet”, he is capable to interpret speaker’s words. In this case the concept of meaning involved in the communication is speaker’s meaning. According to Davidson, speaker’s meaning is systematic; in fact, if our theory of interpretation is compositional and with a finite base, we can learn in principle new meanings (and new modes of compositions of words).6 Moreover, speaker’s meaning is shared when both speaker and interpreter get a common understanding of a word. Hence, Davidson claims that speaker’s meaning possesses the properties (1) and (2), but he argues that it is not based on conventions and regularities. In fact, when the speaker utters “epitaph” with the intention of referring to “epithet”, the interpreter in his prior theory interprets “epitaph” according to the meaning of the English dictionary, while in his passing theory he can interpret “epitaph” referring to “epithet”. Note that this is a contextual dependency, since the interpreter uses his passing theory only when the speaker utters “epitaph”. In all the other contexts, the interpreter uses the word “epitaph” according to its literal meaning. Hence, Davidson concludes that communication does not need shared conventions, rules or regularities, but only the agreement on how to interpret the speaker. Davidson makes a leap forward asserting that:

I conclude that there is no such thing as a language, not if a language is anything like many philosophers and linguists have supposed. There is therefore no such thing to be learned, mastered, or born with. We must give up the idea of a clearly defined shared structure which language-users acquire and then apply to cases (Davidson 1986, p. 446).

If there is no room for common language, then communication can be based in principle solely on idiolects. Davidson’s challenge is to show that communication can occur just with the idiolects of the speaker and interpreter. Obviously, a shared language can facilitate communication, yet he argues that the concept of common language is not philosophically important. We just need converging passing theories. Dummett replies that we must distinguish three elements which are involved in communication: a language, a theory of meaning and a second order theory.7 He argues that language is first of all a phenomenon (not a theory) that cannot be eliminated; language has two main functions: it is an instrument of communication and a vehicle of thought.8 Hence, one cannot think to reduce language to a linguistic theory showing it is not philosophically important. Thus, we need a theory of meaning, which is able to explain how language works using some theoretical concepts like truth, reference, etc. According to Dummett, there is no place for the speaker’s intentions and beliefs within the theory of meaning. These are aspects playing a role in a second order theory, i.e. in speaker’s theory based on the theory of meaning. Dummett’s picture of communication has the merit of placing contextual dependence on a second-level theory of meaning, preserving the literal meaning of the first-level theory. In this model it might be hard to explain the introduction of new words and meanings, because it seems to be only a bottom-up process. Nevertheless, Dummett’s picture takes into account both the literal meaning and the speaker’s meaning, without denying the existence of a common language. Thus, Dummett considers idiolects like second-order theories which need to be based on a first level theory and on a language.9 Idiolects cannot be prior to common language, because they are based on a theory of meaning. Dummett firmly states the priority of common language on idiolects. He writes:

A language, in the everyday sense, is something essentially social, a practice in which many people engage; and it is this notion, rather than that of an idiolect, which ought to be taken as primary. We cannot, indeed, dispense with the notion of an idiolect, representing an individual’s always partial, and often in part incorrect, understanding of his language; but it needs to be explained in terms of the notion of a shared language, and not conversely (Dummett 1974, p. 135).

A common language can be related to an idiolect as the rules of a game can be related to what the player thinks they are. Therefore, idiolects are not the rules of the game but just what the speaker thinks a rule could be.10
In Wittgenstein’s terminology, we should be capable to distinguish between “one is following a rule from one is thinking to follow a rule”.11 Wittgenstein and Dummett maintain that the only way to understand if someone is following a rule is to share a practice. This implies that language is a social phenomenon. Davidson thus remarks:

a grasp of the concept of truth, of the distinction between thinking is so and its being so, depends on the norm that can be provided only by interpersonal communication; and of course interpersonal communication, and, indeed, the possession of any propositional attitude, depends on a grasp of the concept of objective truth.

Those who insist that shared practices are essential to meaning are half right: there must be an interacting group for meaning – even propositional thought, I would say – to emerge. Interaction of the needed sort demands that each individual perceives others as reacting to the shared environment much as he does; only then can teaching take place and appropriate expectations be aroused. It follows that meaning something requires that by and large one follows a practice of one’s own, a practice that can be understood by others. But there is no fundamental reason why practices must be shared (Davidson 1994, pp. 15–16).

According to Davidson, communication needs some practices that must be understandable for the interpreter. Interpreter and speaker can have different linguistic practices, yet they must be respectively understandable. A shared practice is not essential for communication; what is important for communication is the acknowledgment of the linguistic practices of the other and the possibility that others can grasp our linguistic practice. The ability to recognize a practice does not follow a set of static syntactic and semantic rules. Pragmatic phenomena and contextual dependence occurring in communication do not necessary belong to a precise set of syntactic and semantic rules. It seems that the “language of the linguists and philosophers” which Davidson says that does not exist is the one which accommodates only for syntactic and semantic rules. Instead, it is the pragmatic understanding that gives life to meaning. But if understanding is the primitive concept in communication, then the concept of meaning starts playing its role as a secondary concept. Dummett’s picture is the exact opposite. Literal meaning in a shared language is a primitive concept, while understanding is based on a theory of meaning. From Dummett’s account it follows that we usually use words with their literal meaning, viz., we just talk and we do not consider our interpreter as an alien in all the circumstances. We assume that our interpreter can be part of some shared practices. Usually, there is no room for a second-order theory; we speak according to our shared language, even if the interpreter and the speaker do not use the same set of words. And yet, speaker and interpreter share a fragment of language allowing them to communicate. This view leads to a molecularistic account of meaning, since the meaning of a word does not depend on the knowledge of the entire language, but on a relevant fragment of it. According to Dummett,12 if one assumes a holistic view of meaning (like Davidson), communication becomes impossible. Dummett’s argument is the following: consider that in the holistic account of meaning there is no distinction between analytical sentences and synthetic sentences and also that Frege’s principle of context holds. Then, assume that meaning is given by the inferential role. Moreover, the meaning of a word depends on the entire language (because there is no finite base of primitive meanings). But there are not two people with the same set of words and beliefs, hence it is never the case that two people can grasp a shared meaning, i.e., communication is impossible.13 Dummett does not provide a confutation of holism, but he shows that there can be awful consequences for communication if one holds a holistic view of meaning.14 A similar argument is presented in (Dummett 1977, p. 366):

no one sentence of language can be fully understood unless the entire language is understood. The understanding of a sentence comprises a readiness to recognize each possible means by which it might be deductively derived from true sentences; and, because we can place no restrictions upon which sentences might occur in the course of such a derivation, there is no proper fragment of the language of which we can say that, once it has been mastered, then a complete understanding of that sentence has been attained. Language, on such view, is a game with an immensely complicated system of rules, and, in order to grasp the significance of any move in the game, you must know all the rules.

Davidson criticises both Dummett’s view of language, according to which language is a game with a set of rules and conventions, and Dummett’s objection to semantic holism. The concept of truth cannot be considered equivalent to the act of winning a game, because the properties that can be ascribed to the concepts of truth and language are not equivalent with those necessary for playing a game. Namely, Davidson argues that:

Dummett drew an analogy between truth and the concept of winning a game. If we want to know what winning at a game is, we will not be satisfied by being told the definition of winning for each of several games; we want to know what makes the situation defined for each game a case of winning. Thinking of truth, the problem could be put this way: if we were exposed to speakers of a language we did not know, and were given a Tarski-style truth definition, how could we tell whether the definition applied to that language? A good question; but I do not believe it can be answered by attending to Dummett’s proposed convention. For it seems to me that nothing in language corresponds in relevant ways to winning in a game. The point is important because if Dummett is right, to understand what it is in language that is like winning in a game is to make the crucial connection between meaning as described in a theory of truth and the use of language in contexts of communication (Davidson 1984, pp. 4–5).

According to Davidson, when playing a game like chess, all players have to represent themselves as directed to win the game, they must follow the rules of the game in order to win, and winning a game can be an end in itself. These phenomena do not occur in the linguistic practice, and therefore truth cannot be compared to the idea of winning a game. He suggested that the concept of “assertion” might be equivalent to the concept of winning a game, because who performs an assertion wants the content of the assertion to be socially judged and recognized as true. Dummett’s equivalence “truth as winning a game” was criticised also by Jaakko Hintikka, who prefers the equivalence “truth as the existence of a winning strategy for the verifier” in the game semantics. According to Hintikka truth cannot be compared to the actual act of verification in game-semantics, but to the possibility in principle that the verifier could win the game, since there is a winning strategy for him.15

In case of a game with pre-existing rules the players can align their practices with them. Nonetheless, that does not imply that there is always one suitable practice for all rules.16 On the other hand, if the game lacks pre-existing rules, how can it be possible for any convention to be instituted? The question is equivalent to the following one: how can it be possible for two idiolects to produce a common language with conventions? I will answer all the aforementioned questions in Sects. 3 and 4. In the next section, I will place Dummett’s and Davidson’s view on language and communication within the dispute between literalism and contextualism, in order to explain the role that contexts and common language play in their views.

2 Literalism and Contextualism

The relationships between common language and idiolects can be considered on the light of the dispute between literalists and contextualists. In Recanati (2005) it is offered a good conceptual map among all the existing views which go from radical literalism (also known as protoliteralism) to radical contextualism. I will briefly describe all the positions in order to settle Dummett and Davidson in this conceptual map. Hence, I consider the following points of view:
  1. 1.

    Protoliteralism: this was the view of the Wiener Kreis. Natural languages are imperfect because they present contextual dependencies and indexical elements, differently than logical languages.

  2. 2.

    Eternalism: this is Quine’s view. Indexicality is a non-essential property of common language, because all contextual sentences can be reduced to eternal sentences.17

  3. 3.

    Conventionalism: this is the view for which the truth-conditions of a sentence are determined by the linguistic rules. There is no room for pragmatic and contextual dependencies. Indexicals are determined not by speaker’s meanings but they are determined by linguistic rules.

  4. 4.

    Minimalism: speaker meaning is based on the truth-conditional content, which is ruled by linguistic conventions. This is a bottom-up view of meaning. The literal meaning is the first meaning. The second-order meaning (which is context-dependent) is instituted on the first-order meaning.

The points of view 1–4 complete the set of all the literalist accounts of meaning. Then, there are two middle positions between literalism and contextualism that are 5 and 6:
  1. 5.

    Syncretic view: this is a top-down view about meaning. It does not suffice to know the literal meaning of a sentence, because every sentence needs a pragmatic modulation which determines the “actual” meaning; e.g., if one says: “the ham sandwich left without paying”, a process of “predicate transfer” occurs. The guy who has eaten the sandwich is replaced with the sandwich itself. This process is not linguistically controlled, because it involves only pragmatic aspects.

  2. 6.

    Quasi contextualism: literal meanings exist but they are useless theoretical entities. The only important part of the meaning is speaker’s meaning in a particular context.

For the contextualist accounts of meaning, no proposition can be expressed without a contextual modulation, i.e. literal meaning cannot be ascribed to sentences. The three contextualist points of view are the following:
  1. 7.

    Pragmatic composition: according to this view, words can have a literal meaning, but their modes of composition are pragmatically modulated. Hence, words can be interpreted with their linguistic meanings, while sentences need pragmatic factors in order to be interpreted.

  2. 8.

    Wrong format view: unlike the pragmatic composition, even the meaning of the words cannot go directly into the interpretation. Words receive their meaning from the interpretations based on speaker’s intentions.

  3. 9.

    Meaning eliminativism: we do not need literal meanings at all. All meanings are context-dependent. From the past uses of a word or a sentence, we can obtain the contextual meaning of them.

I claim that Dummett’s position conveys some aspects of minimalism, while Davidson’s point of view (especially for the later Davidson) is wrong format view. First of all, I will consider Dummett’s view. According to minimalism, speaker’s meaning is based on the literal meaning of a sentence and it represents exactly the same difference between a first order theory and a second order theory in Dummett’s terminology.18 Moreover, consider this passage of Dummett about indexicals in a paper that is devoted to the ontology of time:

Every place can be called both ‘here’ and ‘there’, both ‘near’ and ‘far’, and every person can be called both ‘I’ and ‘you’: yet ‘here’ and ‘there’, ‘near’ and ‘far’, and ‘I’ and ‘you’ are incompatible. It would be no use for an objector to say that London is nearby far away, but far away nearby, or that it is ‘here’ there but ‘there’ here, since it can also be called ‘nearby nearby’ and ‘“here” here’, and so on. Similarly, it would be no use an objector saying ‘You are “you” to me, but “I” to you’, because everyone can be called both ‘“you” to me” and ‘“I” to me’ (Dummett 1978, p. 353).

In this difficult passage, Dummett claims that indexicals need to follow some linguistic rules, because the meaning of indexical expressions cannot be just “in the head” of the speaker. Hence, Dummett’s view of meaning and communication can be placed upon the literalist corner of the dispute between contextualists and literalists. On the other hand, Davidson’s view falls in the contextualist corner of the dispute, notably in the position called wrong format view. Namely, both sentences and words need to be interpreted on the grounds of conveying passing theories. Although Davidson argues that a common language is not important for communication, I do not consider Davidson within the most radical contextualist position, namely meaning eliminativism, which denies that communication needs linguistic meanings and common language. In fact, writes Davidson:

what is essential (for communication) is a basic framework of categories and rules, a sense of the way English (or any) grammars may be constructed, plus a skeleton list of interpreted words for fitting into the basic framework. […] The general framework or theory, whatever it is, may be a key ingredient in what is needed for interpretation, but it can’t be all that is needed since it fails to provide the interpretation of particular words and sentences as uttered by a particular speaker (Davidson 1986, p. 444).

Hence, even the later Davidson should recognize the role to common language; but this role is philosophically unimportant since it comes to be useless for contextual dependency. Davidson maintains that communication in a particular context can occur without a shared language and without conventions. Moreover, one can define an idiolect without a common language. Namely, language occurs when idiolects overlap, therefore idiolects are conceptually prior to common language. By contrast, Dummett argues that contextual dependency cannot be understood without a shared language. Common language is conceptually prior to idiolects and communication cannot occur without a common language.

In the next section I will analyse the main questions and I will argue that from two different practices (or idiolects) it is possible to constitute a social convention (or a common language).

3 Idiolects, Language and Conventions

The concept of idiolect can be interpreted in two different ways, which I call idiolect1 and idiolect2. Idiolect1 is the strong metaphysical account of the concept, i.e. two people cannot have the same idiolect and two idiolects cannot even present a non-trivial intersection. Moreover, the speakers of the idiolects do not share any practice. Hence, if one assumes this concept of idiolect, communication is impossible without a common language. There is no way to grasp the meaning of speaker’s words for the interpreter. Therefore, idiolect1 is not the concept involved in the Dummett-Davidson dispute. If we assume that two idiolects can have a non-trivial intersection or that two people can master the same idiolect and share some practices (of course, there is still the open possibility that the two idiolects are different), then the concept of idiolect which corresponds to these properties is what I call idiolect2. This is the concept of idiolect involved in the dispute. That’s why when I use the word idiolect in this paper I refer to idiolect2.

We know that if there is a common language two people can communicate without paying attention to their idiolects. The two idiolects can overlap, they can have a non-trivial intersection, they can be different and it is also possible that an idiolect is contained into another idiolect.19 If speaker and interpreter share the same idiolect they can easily communicate. If the idiolects have a non-trivial intersection or an idiolect is contained into another, then communication is partial, but communication can come to be better and better if it keeps on going. If the two idiolects are completely different and the speaker and the interpreter do not share any practice and there is no metalanguage speaking of the two idiolects, then there is no direct communication.20 In any case, if the speaker and the interpreter start sharing a practice, then there is a theoretical way for which they can institute a convention, namely a conventional language in this particular case.

David Lewis, in his doctoral dissertation, has proposed a rigorous treatment of the concept of convention.21 He has also offered many examples explaining how a convention can be instituted. Unlike a social norm which entails a sanction when not followed, a convention is a social behaviour that does not entail a sanction. A sanction is externally defined and imposed, while the breakdown of communication is something internal. There is no external authority in communication. Communication occurs in principle without a privileged point of view between the speaker and the interpreter, while people follow a convention because it is considered as a common interest to behave according to it in a society, where there must exist some authorities. A convention is usually determined by coordination games or by phenomena that need coordination problems.

Imagine two guys rowing a canoe. The best way to row the canoe is synchronizing their movements, which is a social convention, because everybody prefers to follow that practice if the others follow it, too. The conformity to a practice is an equilibrium of coordination in this context.

According to Lewis, a rigorous definition of convention is the following:

“a regularity R in the behaviour of members of a population P when they are agents in a recurrent situation S is a convention if and only if it is true that, and it is common knowledge in P that, in almost any instance of S among members of P,

(1) almost everyone conforms to R;

(2) almost everyone expects almost everyone else to conform to R;

(3) almost everyone has approximately the same preferences regarding all possible combinations of actions;

(4) almost everyone prefers that any one more conform to R, on condition that almost everyone conforms to R;

(5) almost everyone would prefer that any one more conform to R’, on condition that almost everyone conform to R’, where R’ is some possible regularity in the behaviour of members of P in S, such that almost no one in almost any instance of S among members of P could conform both to R’ and to R” (Lewis 1969, p. 78).

Once a convention starts running within a population the members of the population start solving the problems of coordination and they keep following the convention. If a shock occurs in the common states of affairs to the members of a population, then the convention can be changed.22 If one wants to understand how a conventional language can come to be instituted, we can imagine the following condition as an adaptation of Lewis’ example. Consider two individuals A and B. They do not share a language. A utters “XYZ” when he meets B. If they want to meet again after 1 week the easiest thing to do is to go to the same place, because the best place to go for A is the place where B will go and, conversely, the best place to go for B is the place where A will go. If A succeeds to meet B, also B succeeds to meet A. This is the best strategy both for A and B. If other members of the population start following this convention, probably, in this way “XYZ” will end to mean something like “see you here again”. In this way from an individual expression can be abstracted a linguistic convention. Obviously, this is a probabilistic process and some empirical difficulties can occur; e.g., the name “kangaroo” seems to mean in an Australian dialect just a species of kangaroo. Probably other cases similar to the kangaroo case can occur. Nevertheless, I am just interested in the fact that a conventional language can occur in principle from different idiolects, even if there is a switch from the original meaning to the new one.
Davidson considers Lewis’ view about conventions. He writes:

What exactly is the necessary convention? It cannot be that speaker and hearer mean the same thing by uttering the same sentences. For such conformity, while perhaps fairly common, is not necessary to communication. Each speaker may speak his different language, and this will not hinder communication as long as each hearer understands the one who speaks. It could even happen that every speaker from the start had his own quite unique way of speaking. Something approaching this is in fact the case, of course. Different speakers have different stocks of proper names, different vocabularies, and attach somewhat different meanings to words. In some cases this reduces the level of mutual understanding; but not necessarily, for as interpreters we are very good at arriving at a correct interpretation of words we have not heard before, or of words we have not heard before with the meanings a speaker is giving them. Communication does not demand, then, that speaker and hearer mean the same thing by the same words; yet convention requires conformity on the part of at least two people (Davidson 1984, pp. 13–14).

In an interview with LePore, Davidson clarified his view saying that he used to understand questions in French, and then he could reply in English. He said to Lepore

I sometimes carry on conversations with people who are speaking French while I speak English. When I read papers in Europe, I’m constantly asked questions in German, French and Spanish, and I answer them in English. They understand me and I understand them. Of course, you might say that we have a background that allows us to translate back and forth; that’s true. But, we could do it from scratch; it just would be very hard.23

In Davidson (1994) there is the same kind of example, but he wrote at that time that he could reply in English to some letters he had received in French and Spanish. Michael Dummett had offered a reply to Davidson’s claim that communication is possible if speaker and interpreter use different languages, making a distinction between the active use of a language and a passive understanding of a language.24 Imagine a community where twenty languages are spoken, where everyone can speak only one of these languages but everyone was taught to understand all languages. Probably, we should prefer to consider the linguistic phenomenon of that community as one language with “a remarkable variety of different forms”, and this language is the basis for communication. For instance, it is very easy for Italian people to understand Spanish and vice versa, but Spanish people cannot expect from an Italian speaker to master Spanish idiomatic expressions, if the Italian speaker does not know Spanish. Nevertheless, these kinds of mixed languages do not seem to be “the language of philosophers and linguistics” considered by Davidson.

4 Concluding Remarks

I want to come back now to the main questions exposed in the Sect. 1. My thesis is that from sharing a practice a (linguistic) convention can be instituted. I agree with Dummett’s reply to the main question I, namely that communication cannot occur without a shared language. If there is no common language between speaker and interpreter, a conventional language (in Lewis’ sense) can come to exist. Anyway, in some circumstances, idiolects can be prior to a common language, i.e. if there is at least a shared practice due to the strategies of coordination among the members of a population. Therefore, Davidson’s reply to the main question II can be acceptable in some contexts.25 So far so good. But still, a shared language does not imply that it is a structured language. A shared language lacking the structure (especially the syntax) must be something similar to what Strawson called the “feature placing” expressions such as: “rain”, “fire” etc., which respectively refer to “it is raining” and “something is burning”.26 These expressions make explicit the presence of a feature (such as: general kinds, very simple concepts with very evident observative counterparts, etc.) in an environment.27 This ‘small language’ can suffice to show the possibility—envisaged by Davidson—that any language can be instituted by some specific utterances or idiolects. The semantic and syntactical structures can help communication, but even in this case it is required the pragmatic dimension of language, due to the presence of some phenomena like malapropisms, etc. Hence, pragmatics seems to be one of the most important fields of communication. In this respect, I endorse Lycan’s views on the nature of pragmatics for which it is possible to make a distinction in the language between the “semantic pragmatic” and the “pragmatic pragmatics”.28 Semantic pragmatics concerns the determination of the content of an utterance by means of a context. Indexicals, particularized conversational implicatures, speaker meanings, etc. and all the context-dependent features of the language belong to the “semantic pragmatics”. The main idea behind “semantic pragmatics” is that to handle all the contextual features of an utterance with a semantic theory, e.g. Davidson’s treatment of indexicals can be viewed as belonging to the “semantic pragmatics”. Instead, the “pragmatic pragmatics” concerns those features of the language that do not strongly depend on the context, such as: speech acts, generalized conversational implicatures and some systems of formal pragmatics.29 Thus, it is also of paramount importance the acknowledgment of the pragmatic aspects of a structured language not depending on any particular context of utterance.30 Thus, the theoretical relationships between language and idiolects may be clarified with the help of new pragmatic theories and with different ways of modelling natural language at different level of abstraction rather than with simple semantic and theories of meaning.


For the distinction between modest theories and full-blooded theories (see: Dummett 1974, p. 102).


Note that “modest theory” does not stand for a theory with limited goals, i.e. a modest theory can achieve ambitious results, too. What is “modest” is the methodology, not the aims. A vivid example of a modest theory is the one presented in Davidson (1967), while the most recent works of Davidson do not fit anymore in a modest program, because of the role that charity and triangulation play in Davidson’s theory of meaning and communication. See Gaifmann (1996).


Note that not all possible combinations of the questions I and II make sense and therefore the two main questions are not completely separated. For instance, if one assumes that there is no shared language in communication, then one cannot maintain that the shared language is prior to the idiolects.


Dummett (1986).


Karen Green also manifests some perplexities for the identification of Dummett’s pair “short range theory” and “long-range theory” with Davidson’s pair “prior theory” and “passing theory”. Cf. Green (2001).


The problem of the linguistic creativity can be addressed with different approaches, e.g., the lexical productivity, the recursiveness, and the assignment of new uses of previous words (such as metaphors and some kinds of idiolects). As noted in Groenendijk and Stokhof (2005), these three approaches to linguistic creativity do not stand only for three different views about the same phenomenon, viz. every theory supports a different ontology. This implies that we need to use our conceptual tools “as if” the given underlying our linguistic theory should be exactly the one that fits better in our pre-established approach.


Cf. Dummett (1986).


See Dummett (1989). Usually, one cannot easily make a distinction between the aspects regarding our language and the ones regarding our thoughts. But when something goes wrong one can become aware of the interplay between language and thought. See Wittgenstein (1953, §330).


We know that the conversational implicatures explain the pragmatic and intuitive part of meaning that cannot be handled by the standard truth-conditional semantics. Dummett’s distinction between a first-order theory and a second-order theory can be handled by Grice’s theory of conversational implicatures. But according to Grice, the speaker’s meaning is based on the communicative intentions of the speaker, while, for Dummett, intentionality does play a secondary role in the theory of meaning. Nevertheless, also Dummett argues that part of our understanding is implicit, and a theory of meaning should make it explicit. In any case, he does not maintain that some unintentional linguistic phenomena (like malapropisms) can play any role in the theory of meaning.

Two different types of conversational implicatures are usually isolated: the generalized conversational implicatures (which are inferable without a strong appeal to a particular context) and the particularized conversationalimplicatures (which are mainly contextual). Namely, the first order theory is what Grice calls “what is said” (the descriptive meaning or the truth-conditional one), while the second-order theory is a generalized conversational implicature. Therefore, the generalized conversational implicatures are an important interface between purely semantic and pragmatic aspects. See Grice (1975) and Levinson (2000).

Another important class of non-truth-conditional aspects of the natural language is determined by means of the conventionalimplicatures, which are based on the conventional meaning of the words. Karttunen and Peters (1979) suggested to formalize the conventional implicatures by means of two formulae. The first formula expresses the literal (truth-conditional) meaning of a sentence, while the second its conventional implicature. In this way it is even possible to formalize in a recursive manner the non-truth-conditional aspects of language. See Gamut (1991, chapter VI).


Cf. Dummett (1986).


See Wittgenstein (1953, §202).


Dummett (1973).


This is Dummett’s point of view. See Penco (2007). Davidson could solve this problem with his theory of radical interpretation, which starts “from scratch”, without any appeal to any common knowledge between speaker and interpreter.


Cf. Penco (2007).


For further details, see Hintikka (1987).


See the example in Wittgenstein (1953, §200).


Cf. Quine (1960).


An indexical expression can be classically viewed as an open sentence, namely the interpretation of a sentence containing indexical is a function from contexts to propositions, while a proposition is a function from possible worlds to truth values. See Stalnaker (1970).


If one holds a strong view about the concept of “idiolect”, then there is no possibility that two different idiolects can belong to the same shared language. That is not the case if the concept of idiolect involved is understood as a partial grasp of a language, or an implicature.


Nevertheless, a third idiolect overlapping the other two idiolects can work like a partial interpreter allowing a process of indirect communication.


See Lewis (1969).


Note that not all the problems of coordination can be solved referring to conventions. Imagine a situation where there are two individuals. We promise them that they can get a great sum of money if they choose the same number, but they cannot communicate with each other. In the 40% of the cases, the two individuals choose the number 1 (see Schelling 1960). Of course, this choice does not depend on the rationality of the individuals, but it depends on aesthetic, accidental, or social issues (known as focal points) which can offer reasons for simple collective choices. If a focal point becomes common knowledge, namely the probability of choosing the number 1 becomes close to 1, this fact can suffice to convert a focal point into a convention.


Davidson (2004).


Dummett (1994).


Note that the use of conventions is rejected by Davidson. Hence, in my approach to the problem of the priority between idiolects and language the target is Davidsonian, not the methodology.


See Strawson (1959), from p. 202.


In Searle (2006) it is argued that the “feature placing” expressions are a sort of very primitive language common both to humans and to animals.


See Lycan (2000). Of course, there can exist some linguistic systems which provide both aspects of pragmatics, but this does not suffice to eliminate the dichotomy.


Dalla Pozza and Garola (1995) is a vivid example of the “pragmatic-pragmatics” tendency.


See Levinson (2000).



I am grateful to Martin Stokhof, Carlo Dalla Pozza, Carlo Penco and Davide Sergio for critical comments on an earlier version of this paper.

Copyright information

© Springer Science+Business Media B.V. 2011