If you find a broken link, or have any questions about the Interlocutor, please feel free to contact

Prof. Jim Peterman

*David Wingate is currently finishing a degree in philosophy at Grinnell College. His focuses are linguistic and existential philosophy.

  In the Philosophical Investigations, Wittgenstein introduces the term "private language" to discuss the possibility of a private language of sensationoa language whose terms are ostensively defined to refer to private sensations and are therefore unintelligible to other speakers. The phrase "private language" and the argument of the Investigations to which it has come to refer are the subject of much scholarly debate and confusion. I wish, here, to focus on one such case.

In the second chapter of The Language of Thought, Jerry Fodor characterizes Wittgenstein's private language argument as inapplicable to Fodor's claims that we must have a language of thought by which we learn a first language. Fodor's argument has two forks that need explication. The first insists that we must have a private language to learn the meanings of predicates in a first-learned language and the second dismisses Wittgenstein's private language argument as inapplicable to Fodor's first, positive claim. The two arguments are related in that the second defends the possibility of the first. By Fodor's account, he and Wittgenstein disagree with respect to their accounts of how we might learn language. However, the nature of the dispute is more trenchant than a simple dispute about the manner in which we may learn a first language. This supposed disagreement is symptomatic of a larger disagreement as to what counts as "meaning" and as language. As I will show below, Fodor posits propositional attitudes as appropriate for explaining the conventions that govern the use of language and such a move (implicitly) challenges Wittgenstein's private language argument.

Fodor's essay rests on three main tenets which merit brief enumeration here for the sake of grounding the discussion. The first two claims are that "learning a first language is a matter of hypothesis formation and confirmation . . ." and "that learning a first language involves at least learning the semantic properties of its predicates" (59). In defense of his first assumption, Fodor argues that learning the predicates of a language is a matter of concept formation and that the formation of such concepts has been shown to be a matter of hypothesis formation and testing. In support of this last step, he cites sources such as Burner, Goodnow, and Austin (1956). Fodor does not defend his second assumption. He proposes that it "will be granted by anyone who is willing to suppose that there is anything at all to the notion of semantic properties as psychologically real" (59). For the purposes of this paper, I do not take issue with either of these assumptions, which will be fundamental to Fodor's assertion that a language of thought must be necessary for learning a first language. However, his third assumption will prove to be problematic. It demands "that S learns the semantic properties of P only if S learns some generalization which determines the extension of P" (59). Here "S" denotes a speaker who learns an arbitrary predicate "P" of his first learned language. This assumption is also crucial to Fodor's subsequent positing of a "private language" of thought and, as I will show below, places him at odds with Wittgenstein on the question of the relation of learning to understanding. 

For Fodor, these three assumptions form a linear progression from which he concludes that "[a speaker] S learns [a predicate] P only if S learns a truth rule for P" (59).¹  He contends that if language learning is a matter of hypothesis formation and testing, "then among the generalizations about a language that the learner must hypothesize and confirm are some which determine the extensions of predicates of that language" (59). If we learn a language by hypothesis formation and testing, then we must learn a generalization for determining a term's extension because that generalization (by assumption 3) is necessary in learning the semantic properties of a language (2^nd assumption). And by Fodor's account, such a generalization for determining the extensions of predicates is a truth rule. Thus, Fodor arrives at the claim that "S learns P only if S learns a truth rule for P" (59).  This posited necessity of a truth rule for determining the extension of predicates of a learned language brings Fodor to posit the necessity of "private languages." He asserts that one cannot learn that "[a predicate] P falls under [a truth rule] R unless one has a language in which P and R can be represented" (64). From this assertion Fodor concludes that one cannot learn a first language unless one already has a language with which to represent P and R. Since P, by this account, is a predicate of a first learned language; R may not be described in terms of a learned language for fear of infinite regress. If R were a predicate of a learned language, then we could no longer express a truth rule for learning a first language by relating P and R. Thus, it seems we must have some representational system, which is not learned, with which we may characterize R.

This necessary representational system is what Fodor describes as the language of thought. Since such a representational system is not a learned language, its use could not be mediated by public conventions. If it were informed by such conventions, then it would be learned. Note that, for Fodor, this language of thought would be the basis for learning the truth rules for the use of our predicates. It must be assumed that his proposed language of thought is also the basis for learning the predicates of sensation speech. If there were some other faculty by which we represented the extensions of predicates of sensation speech, Fodor would have no reason to even approach Wittgenstein's private language argument. Since Wittgenstein's private language argument argues that the validity of our sensation speech is, in fact, mediated by public conventions, Fodor's concern for the first part of his argument is to show that the private language argument does not challenge his proposed language of thought. While Fodor's general concern is respectable, his arrival at the claim that we must have a language of thought is problematic with respect to maintaining his claim that Wittgenstein's private language argument does not challenge his future characterization of the possibility of such a language of thought. The third assumption that Fodor makes in arriving at the necessity of a language of thought was that a speaker must learn a generalization by which to determine the extensions of predicates in order to learn the semantic properties of that predicate. While this assumption does not directly conflict with the private language argument, it does lead to implications that challenge Wittgenstein's commitment to a use theory of meaning, which is a central basis for Wittgenstein's argument.

Fodor's critique of Wittgenstein

To begin, I wish to briefly review Fodor's characterization of the private language argument. The impetus for Fodor's characterization is to defend his claims to the necessity of a language of thought from Wittgenstein's private language argument. Although Fodor initially claims not to "propose to enter the miasma of exegetical dispute that surrounds the private language argument," he in fact offers two possible readings of Wittgenstein's argument (68). Fodor allows Wittgenstein "two ways of characterizing a private language: either as one whose terms refer to things that only its speaker can experience or as a language for the applicability of whose terms there exists no public criteria" (69). Fodor takes Wittgenstein's argument to discuss either sensation or linguistic practice. By the first count, the argument would be taken to reject the notion that we may have private terms of sensation, i.e., sensations of which others cannot have knowledge. By the second count, the argument is taken to reject the notion of a language whose terms acquire meaning by something other than publicly accessible rules.  Wittgenstein introduces the idea of private language by asking us to imagine a language in which "individual words . . . refer to what can only be known to the person speaking; to his immediate private sensations" (PI, 243).

This introductory note seems to introduce the first fork of Fodor's proposed dichotomy: a language whose terms refer to things that only the speaker can experience. Wittgenstein's initial question, although split by a section break, does not end there, however. He goes on to consider the question "How do words refer to sensations?" It is this crucial connection of the private language argument, between language and sensation, that Anthony Kenny notes in his discussion of private languages. He observes that the argument attacks both the mistake about experience that "experience is private" as well as the mistake about language that "words can acquire meaning by bare ostensive definition," i.e. without employing public criteria (180). The crux of the private language argument is that sensations may not be named privately because language simply does not permit this; meaningful terms depend upon public corroboration. Thus, claims like "I know that I am in pain" seem awkward to us. This intimate connection between the nature of sensation and the manner in which we speak of it (Fodor's two forks) prompts Wittgenstein to note that "It can't be said of me at all (except perhaps as a joke) that I know I am in pain" (246).

Fodor's dichotomy of aspects of Wittgenstein's argument is inappropriate; Wittgenstein does speak of both private sensation and public criteria, but would not allow that they be separated in such a manner. There is, in Wittgenstein's private language argument, a strict relationship between the issue of private sensation and public criteria. Wittgenstein prefaces his argument by inquiring "How do words refer to sensations?" and notes that this question is tantamount to asking, "how does a human being learn the meanings of the names of sensations?" (244). Here Wittgenstein prefigures his intention to explore the relation of learning and reference in sensation speech. The special thing about sensation speech for Wittgenstein is that its words refer to sensations only insofar as they are legitimated by the public conventions in which we learn them; the forks of Fodor's proposed dichotomy are therefore inseparable. Fodor's error has been to separate the argument's corrections of the mistakes about experience and the mistakes about linguistic meaning.  This connection between correcting mistakes in talk of sensation and in the nature of language appears throughout the private language argument. Wittgenstein claims that our first person sensation speech replaces primitive responses to those sensations. For example, "I am in pain" might replace crying (PI, 244). As a response to the interlocutor's objection that it would be impossible to teach a child the use of sensation speech if we showed no outward signs of those sensations, Wittgenstein offers the following example:

Again, Wittgenstein points to the awkwardness produced by our misconceptions about sensation and linguistic meaning/use when they are juxtaposed. If we attempt to ostensively define a private term of sensation, then the use theory of meaning demands that there exist some rule by which we employ the word in our speech if that term is to be meaningful. If such a rule does not exist, then our word lacks meaning. If we invent the technique of using the word, then there need be no reason to speak of the word as necessarily private; we might easily teach our rule of employment for the word to another person. Similarly, if we find that the rule for the word's use already exists in the public sphere, then we need merely reference this rule like a signpost of our language to explain the use of the word to another. 

The end focus of these claims is that the use theory of meaning does not permit us to privately define words of sensation speech. In other words, saying a name while inwardly focusing our attention on a sensation is not sufficient to give the term meaning. In an oft-cited metaphor, Wittgenstein compares the giving of a private ostensive definition to the metaphor of my left hand giving something to my right hand and considering such an exchange to be a gift. The possibility of this scenario is to be taken as obviously inane. The reader is expected to ask what the significance of such an exchange might be. Wittgenstein suggests that "the same [question] could be asked if a person had given himself a private definition of a word; I mean, if he has said the word to himself and at the same time has directed his attention to a sensation" (268). Here the possibility of definition through ostensive definition becomes crucial. The terms of a language of private sensation are limited to reference via ostensive definition; as noted above, they cannot depend on public criteria for their meaning since they would then no longer be private. Ostensive definitions are neither impossible nor even uncommon.

Wittgenstein deals with the notion of ostensive definition early in the Investigations. He notes in section 28 that "an ostensive definition can be variously interpreted in every case" and, in section 30, that it "explains the use-the meaning-of the word when the overall role of the word in language is clear." In a related analogy, Wittgenstein speaks of the function of holding up a king from a chessboard and saying, "This is the king." He emphasizes that such an ostensive definition "does not tell [us] the use of this piece . . . unless [we] already [know] the rules of the game up to the last point" (31). Thus, although ostensive definition may have its function in learning a language, it is insufficient to determine a predicate's meaning. Words of sensation speech that are defined ostensively acquire meaning only through public linguistic convention. In retrospect, we can see that the impetus for Wittgenstein's private language argument is his commitment to a use theory of meaning; it is this commitment that makes the possibility of a private language of sensation non-sensical. The language-game of sensation speech simply does not permit us to speak of sensations based on private criteria. Our language of sensations requires that it be couched in social linguistic practices to prevent it from becoming as non-sensical as the left-to-right-hand gift. 

Wittgenstein does not restrict this commitment, to limiting the ability of ostensive definition to fix a term's meaning only in relation to a broader language-game, to the private language argument. It is an implication of Wittgenstein's use theory of meaning. In the Blue Book, he discusses the way we might understand a diviner who says that he "feels that the water is five feet under the ground" (9). Although we understand the meaning of "I feel" and "five feet under the ground" independently, the grammar of the whole phrase has not yet been explained. Wittgenstein claims not to understand the phrase because he is limited to knowing "what a word means in a certain context" (9). Ostensive definition of the phrase by relating its parts is simply insufficient by Wittgenstein's account. Wittgenstein's proposed resolution to the stalemate is that the diviner offers the following explanation: "when I have a certain feeling of tension in my hands, the words ¬three feet' spring up in my mind" (10). This "certain feeling of tension" seems suspiciously similar to the possibility of a private sensation. But Wittgenstein claims that such an explanation does not tell us anything about the sensation. Such an explanation does not, in fact, fix a reference to something private. He remarks that the diviner's explanation is permissible, but that the statement that he feels the depth of the water "will have neither more, nor less, meaning than [his] explanation has given it" (10). Furthermore, Wittgenstein elaborates by suggesting that "if experience shows that the actual depth of the water always agrees with the words ¬n feet' which come into your mind, your experiences will be very useful for determining the depth of water" (10). This scenario of the Blue Book is similar to Wittgenstein's example from the private language argument of a term "S," which a speaker invents to refer to a sensation that is supposedly private. For Wittgenstein, this sort of pointing does not take on meaning until we have an external (public) corroboration for its use. The use of the hypothetical symbol "S," which we initiated to "refer" to a sensation by focusing our attention upon it, is (potentially) inconsistent and meaningless until we find external corroboration for its useoan increase in blood pressure in the case of Wittgenstein's example. However, at this point (270), "S" names a sensation because of the way in which it is employed within a language-game. Moreover, in the same way that the diviner's statement was (potentially) useful for determining the depth of water rather than a referenced sensation, the use of "S" refers to an increase in heart rate rather than referring explicitly to a private sensation.

Thus, throughout the private language argument, Wittgenstein insists on a connection between the rejection of the privacy of sensation and the commitment to a use theory of meaning by which our terms acquire meaning through the context of public criteria. Without such a connection our sensation speech would be as meaningless as a left-to-right-hand "exchange." The import of Fodor's mischaracterization of the private language argument is that it seems to form the basis for future mistakes. More specifically, Fodor focuses on what he takes to be the second possible reading of the private language argument in constructing his positive account of how we might characterize a language of thought. He foreshadows that his argument will be, at the very least, subject to the criticism that public conventions determine meaning, what Fodor took to be the second of the possible readings of the private language argument. I will contend below that such a limited application of Wittgenstein's argument lays the groundwork by which Fodor's positive claims, as to how we might represent the language of thought, do in fact put him at odds with Wittgenstein's private language argument. 

Fodor's positive account

Despite misunderstanding the private language argument, Fodor's general concern that learning a language seems to require a system of representation by which we characterize the extensions of its predicates is still valid. If the private language argument insists upon public criteria as the basis of (at least, sensation) language, the possibility of learning a first language does seem problematic. Though Fodor's third assumption, as we saw above, conflicts with Wittgenstein's use theory of meaning, it may turn out that Wittgenstein is simply wrong on this point. In other words, were Fodor's argument for the necessity of a language of thought upheld, Fodor's characterization of a language of thought would still be incompatible with Wittgenstein's private language argument; Wittgenstein's argument would still apply to Fodor's positive account of private language. Bearing in mind the questionable basis of Fodor's argument, I would like to characterize the positive claims that lead him to a characterization of a language of thought.²

Recall from above that Fodor had two ways of characterizing Wittgenstein's private language argumentoas discussing either private terms of sensation or the necessity of public criteria in fixing linguistic practice. Fodor suggests that what he is going to propose in the way of a language of thought will be subject to, at least, the second fork of this dichotomy and focuses on this potential problem. Fodor seems to mark the phrase "public criteria" as a term in need of review. He proposes that the consistent employment that the public criteria refer to should be the product of a "stable relationship between the way the terms are used and the way the speaker believes the world to be"opropositional attitudes (71).³ With this new focus, he remarks that the new "challenge that the private language argument poses to the notion of a language of thought is, therefore this: Show how such a relation [between linguistic forms and propositional attitudes] could be mediated by something other than public conventions" (73).

The importance of such a shift should not be overlooked. Fodor's characterization of the private language argument shifts the emphasis that the argument places on public convention to the realm of propositional attitudes. Fodor believes that such a move will earn him the right to do away with the necessity of public conventions for a language of thoughtoI will discuss the acceptability of this move below. Fodor needs to make such a move to demonstrate that the "conventions" of his proposed language of thought need not be public.  This shift of focus allows Fodor to make the following argument. If the comprehensibility of a public language requires that there be a stable correlation between the propositional attitudes and the linguistic practices of the speaker/hearer, then we may characterize a rule governing linguistic conventions as follows:

Such a rule insists that the meaning of our predicates is fixed by the conventionality of our assent to their pertaining only in case we believe that they are the case. Fodor believes that such a characterization of public language opens a viable option for characterization of a language of thought. Fodor claims that if we think of [a is F] as a proposition of an internal code, then we need only make two minor changes to the above characterization to allow it to hold for a language of thought as well. First, "assent to" is replaced by a sequence of one or more of the basic relations from which computational relations to internal formulae are constructed" and, second, "is conventional" is replaced by 'is nomologically necessary "(78). Fodor holds this sort of characterization of the representation relation for formulae of an internal representational system as sufficient to answer the challenge posed by the private language argument. Although this characterization is persuasiveothe modifications allotted to the initial framework do not depend on public conventionothe positing of (C) as a way of representing the extensions of our predicates does in fact challenge the private language argument.  Fodor assumes that the coherence of a public language requires a stable relationship between the word's use and the speaker's beliefs and the conclusion that Fodor draws from this assumption is problematic. Fodor's shift to propositional attitudes as the basis for coherence between the word and its use is inappropriate as a point to shift between public language and the language of thought. As seen above, Fodor believes that propositional attitudes are the fulcrum about which we may turn from a rule in the public language to a rule for determining extensions in the language of thought. I will show below that, although Wittgenstein's private language argument is not devastating to Fodor's overall concern, it does challenge his use of propositional attitudes in characterizing our speech of sensations. And, if Fodor cannot account for sensation speech, his rule for determining the extensions of termsoa primitive necessity for learning a first language, by his accountois incomplete at the very least. 

Fodor's use of propositional attitudes, as they relate to linguistic forms, is simply ineffective as a way to describe how language is used or learned if Fodor hopes additionally to avoid a confrontation with Wittgenstein's argument. More specifically, Fodor's use of propositional attitudes leads him to formulate a rule for learning the extensions of predicates in the language of thought which is directly at odds with Wittgenstein's private language argument. Let's recall the original (i.e. pre-substitution for the language of thought) version of Fodor's rule for determining the extension of a term.

Fodor notes that the status of his augmentation of (C) is based on an analogy to (C) itself. Although he does not pursue the exact nature of that analogy, he does commit to using his characterization of language at (C) as the validation for his augmentation of (C). However, we saw above that (C) is, at the very least, incomplete in that it cannot account for first person sensation speech with out directly challenging the private language argument. The augmented version of (C) will, therefore, suffer from the same problem.

We saw above that Fodor mischaracterizes Wittgenstein's argument. He inappropriately separates Wittgenstein's rejection of private sensation speech and his insistence on public criteria in fixing the meaning of a term. I also noted that Fodor's argument for (C) as a possible rule for determining the extensions of the predicates of a language of thought was derivative of Fodor's focus on the second of these two proposed interpretations of Wittgenstein's argument. However, it appears now that such a separation and subsequent focus was unfair to Wittgenstein's argument. Fodor's view of the private language argument and subsequent proposal of a rule by which to determine the extensions of a language's predicates does not, in fact, concur with Wittgenstein's treatment of first person sensation speech in the private language argument.

A few remarks need to be made on the propriety of my application of the private language argument to Fodor's proposed generalization, (C). First, it might be offered on Fodor's behalf that such application is unfair in that it denies Fodor the option of appealing to the distinction between superficial and deep grammar, a distinction that Wittgenstein also employs. While, in general, distinguishing between superficial and deep grammar is often important, it does not seem relevant in this case. To employ such a distinction on behalf of Fodor would note that the phrase "I believe I am in pain" might instead simply be read as "I am in pain." Such translation would remove the problem that Fodor's argument encounters, Wittgenstein's rejection of belief as inapplicable to first person sensation speech. However, such a move would remove from (C) its efficacy in characterizing our conventions of speech. By such a scenario, we would no longer have the left side of (C)'s second bi-conditional. We would be left with:

By such a suggested defense of Fodor, the second bi-conditional seriously challenges any conventional concept of the criteria for attributing "pain" to S. By the proposed defense, the convention upon which S's appropriate use of the phrase depends is that S is in pain only when he assents to "I am in pain." And such a claim violates Fodor's own admonition that behavior is insufficient to demonstrate meaning (63). Fodor himself notes that our speech acts are not necessarily dependent upon our mental/physical state. Such a separation of Fodor from the private language argument would simply not be effective. Assenting to being a state of pain does not require that our being in pain is actually the case.  Second, I do not mean to suggest that we must champion Wittgenstein's private language argument over Fodor's characterization of the possibility of a language of thought. The impetus of my argument has been to show that, despite Fodor' claims that Wittgenstein's argument is irrelevant, the argument does in fact conflict with Fodor's characterization of a rule by which we may determine the extensions of predicates in a language of thought. 

At a Standoff (conclusion)

Before considering how to resolve this impasse between Fodor and Wittgenstein, I want to remind us where we've been with respect to the relation of Fodor's positing of a language of thought to Wittgenstein's private language argument. I began by demonstrating Fodor's mischaracterization of the Wittgensteinian argument. Fodor separates Wittgenstein's rejection of languages whose terms refer to sensations that only the speaker can experience and languages that have no public criteria for their use into distinct lines of argument. Such a separation was unacceptable because it neglects the fact that these two corrections that Wittgenstein offers are not independent; they relate our sensation speech to the public criteria by which our words acquire meaning. Such a mischaracterization brought Fodor to propose propositional attitudes as a means to escape what he took to be the challenge of the private language argumento"show how such a relation [between linguistic forms and propositional attitudes] could be mediated by something other than public conventions" (73). Fodor's use of proposition attitudes, as he employs them with respect to the idea that learning a language requires at least learning a rule for the extensions of its predicates, does in fact conflict with Wittgenstein's private language argument.  Such an impasse is not easily negotiated, but there is hope for both sides. One way to resolve the conflict would be for Fodor to take a second crack at the private language argument, discrediting it and thus allowing his particular use of propositional attitudes in characterizing a rule for determining the extensions of predicates in a language of thought. However, I am partial to the private language argument and would like to propose an alternative. Fodor's main concern is providing a basis for defending the work of speculative psychology. To that aim, he wishes to show the possibility of a representational system by which we may characterize the language of thought. In it self, this aim does not necessarily conflict with either Wittgenstein's use theory of meaning or his private language argument. It seems that, baring a coherent rejection of the private language argument, Fodor would be better served to keep the impetus of his research program while removing the aforementioned problematic tenets. Enumerating a representational system for the language of thought need not suppose that learning a first language demands that we have a rule by which to determine the extensions of its predicates. As suggested above, such a rule seems impossible for every predicate. Additionally, by not explaining the use of public language through propositional attitudes we do not therefore reject the possibility of a "private language" of thought. As a further suggestion, I offer that Fodor has no need to call his system of representation a private language. On the first count, it does not seem that such a representational system, even by Wittgenstein terms, would be necessarily private. Fodor's posited private language is private only insofar as it is not used for communication between people. The fact that we may characterize such a representational system at all suggests that it follows consistent rules to which we have access. If Fodor's private language were truly private in the Wittgensteinian sense, we could not speak of it at all. On the second count, Fodor admits that his proposed "private language" may not be a language at all in any traditional sense; "although computation presupposes a representational language, it does not presuppose that . . . it must be a natural one" (56). Given all this, it seems that the only use Fodor has for referring to his proposed representational system as a "private language" is to position his thoughts within the paradigm of Wittgenstein's argument. Such a positioning might easily be maintained without calling his system a "private language" and thereby confusing the issue. In short, although Fodor's claims do conflict with Wittgenstein's argument, we need not necessarily reject the possibility of representing the language of thought by upholding the private language argument. 

Works Cited

1. I place this term in quotes because I will contend below that Fodor's language of thought need not be considered to be a private language.
2. Fodor himself claims that the same argument might be constructed from the point of view of an intentionalist rather than that of an extentionalist (60).
3. Note that fodor's connection of use to propositional attitudes does not constitute a use theory of meaning as I use the term in this paper. Rather, I use "use theory of meaning" to denote what Fodor rejects by positing the importance of propositional attitudes for fixing criteria of usage; namely, a connection between the way the world is and our linguistic practices.
4. The ramifications of this example for the substituted rule of a language of thought are considered below.
5. Note that I have used caret and number symbols to separate the subject and predicate clauses of the original statement "I am in pain." I do so to keep these two parts, which Fodor separates, distinct due to the difficulty of inserting a phrase with a first person subject into (C). I do not hold such fluxes in the grammar to be problematic since they reflect the various uses to which the speaker (S) will supposedly employ them. The form of Fodor's rule requires such changes.