Extending the Private Language Argument

2.1 The problem of a private linguist’s stable memory

In a wide reading of the argument, the focus falls on the nature of the relation between internal mental experience and external linguistic behavior. Wittgenstein regards some external and shared symbols as indispensable to determining the content of private beliefs and, just as importantly, how identity can be maintained over time. He asks whether anyone who is deprived of an ability to externalize thoughts could ever determine whether the content of a thought at time (t1) was identical to the content of supposedly the same thought at a later time (t2)?

What is the criterion for the sameness of two [mental] images? [...] Before I judge that two images which I have are the same, I must recognize them as the same. And when that has happened, how am I to know that the word “same” describes what I recognize? Only if I can express my recognition in some other way, and if it is possible for someone else to teach me that “same” is the correct word here. (Wittgenstein 1953: 117, emphasis added).

Wittgenstein indicates that the mere externalization of thought is actually far from being sufficient to determine how, if at all, thoughts change over time. In fact, if one is to judge the identity of two mental images it is essential that they be externalized and shared by others. To highlight this point, Wittgenstein considers a case where a private linguist invents an arbitrary expression “X” that stands for a feeling S. Having done so, he keeps a diary of the feeling’s subsequent appearances. Every time he experiences S, therefore, he writes the phrase “X” in the diary. The problem is self-evident: in all but the first use of the invented phrase, he has no means of ascertaining that he correctly applies the expression “X” to the original sensation of S. The alleged private linguist would wonder: “I impress on myself the connexion between the sign and the sensation. – But ‘I impress it on myself’ can only mean: this process brings it about that I remember the connexion right in the future. But in the present case I have no criterion of correctness” (Wittgenstein 1953: 92, emphasis added).

It is impossible for the private language creator to be sure whether, at two points of time, his own experience Z is the same. This is because, in such a case, there is no external standard for correct recognition of the experiences’ identity. He may be convinced of having an accurate memory of the original feeling but, if this is so, “whatever is going to seem right to me is right” (Wittgenstein 1953: 92). No mere impression is sufficient for a relation of identity. Wittgenstein stresses that struggling to verify the identity of two mental images without reference to publicly available artifacts is akin to measuring a ruler with itself or, indeed, checking the credibility of a newspaper by comparing two copies of the same issue. If one had an excellent eidetic memory, one might preserve the original mental picture as the basis for making justified judgments about the identity of subsequent experiences. This would be like superimposing two photographic pictures in order to determine whether or not they differ. Since memory does not work that way, the original mental image cannot be accepted as a golden standard for recognizing identity. It is concluded, therefore that an external, publicly available criterion of identity is necessary to stabilize and objectify mental images. The moral of the story about the private linguist’s diary is that mental contents can only be co-constructed by acts amenable to public evaluation. On this view any appeal to mental content that is independent of social coordination is mere philosophical illusion (Candlish and Wrisley 2014).

2.2 Why a private linguist cannot follow a rule

The second, narrow interpretation of Wittgenstein’s argument pursues the issue of rule-following. While the wide interpretation challenges the intelligible use of the signs of a private language, the narrow interpretation maintains that there is no way in which private signs can be created. The argument boils down to answering the question of why it is impossible to follow a rule when a person is separated from other people and thus deprived of access to a public language (Kenny 2005). By definition, a private language consists of “sounds which no one else understands but which I 'appear to understand'” (Wittgenstein 1953: 94). For others, therefore, a private language is unintelligible. The key point here is that the language would also be unintelligible to the inventor. Why is that? In order to make sense of a sign – Wittgenstein claims –it needs to be used according to some kind of rule. The person who aims to invent a private language – the argument goes – must begin by defining the first word. Defining the first word in a private language would entail creating a stable conventional relation between an expression “X” and some mental event S. What would grant the stability of this conventional relation in the expression’s subsequent uses? For this to be possible, a private linguist would subsequently have to use the original conventional relation to link the expression “X” to the mental event Z by drawing on a rule allowing him or her to judge the correctness of each use of the expression. If the expression “X” once refers to S and on other occasions to mental states A, B or C..., “X” would be meaningless even for its alleged user. Therefore, if a private language user is to make sense of a sign, he or she must necessarily conform with a rule because only such conformity can ensure stability of the original relation between the expression and the mental state in the sign’s subsequent use (Candlish and Wrisley 2014; Glock 1996).

Wittgenstein argues that the idea of the private rule does not make sense. For the rule to be private, he argues, no publicly shared criterion can be used in judging whether or not the rule is being followed. However, without external criteria for following a rule, the only possible standard is a person’s private conviction. On such a view, a person’s internal conviction about rule-following becomes indistinguishable from actual rule-following. This conclusion leads to entrapment in a category mistake because, by definition, following a rule belongs to a different epistemic and ethical category than thinking that a rule is being obeyed. In other words, Wittgenstein’s argument suggests that when a private language user thinks that he follows the rule, he would follow the rule and that this would hold regardless of whether or not he actually follows the rule. The result is plainly absurd. If, by definition, the private linguist cannot draw on shared rules, social coordination, and external artifacts, nobody – not even a private linguist – can know whether a rule is actually being followed or whether the meaning applied to an expression in a given situation is correct (or not). In this case, private language is unintelligible for its user. There is no way to ascertain that the meaning he applies to the expression “X” in a particular situation is the same as the original meaning of this expression. In consequence, a private language user cannot possibly know what an expression he utters actually means. For this reason, not only the use but also the invention of a private language expression appears to be impossible (Glock 1996; Prechtl 2004).

Wittgenstein reflects upon two aspects of meaning-making: he both challenges the internal mental picture and appeals to a semiotic relation between a mental picture and its expression (Candlish and Wrisley 2014; Glock 1996). The third aspect of meaning-making – the materiality of signification is left unexplored. Apparently, any difference between a stable record and talk in interaction is assumed inessential in meaning-making. The difference is neglected because it is assumed that our engagement with materiality can be left out. For instance, while discussing possible uses of private expressions, Wittgenstein’s analysis depends almost exclusively on literacy practices such as keeping a diary, checking a timetable or calendar, or looking at a watch. This applies in spite of the fact that he attempts to define a private language in terms of speech sounds (Wittgenstein 1953: 92–94). Accordingly, three interconnected considerations of private language come to the fore: memorizing mental images, creating a stable practice of using a sign, and the need for an external, independent standard for individual impressions. Given that problems of memorization, stability, externality, and independency drive discussion about distinct semiotic and cognitive functions of speech and writing (Kravchenko 2009; Olson 1996; Trybulec 2013, 2017), one can ask whether considering different modes of presenting linguistic patterns (or even “kinds of linguistic activity”) brings new insight on the private language argument. Wittgenstein seems to take a graphic mode of signification as a core of linguistic practice. In so doing, he leaves the distinction between talk in interaction and writing aside. This is because the materiality of signification is considered insignificant. If one acknowledges fundamental differences between graphic and vocal meaning-making, a new question emerges. How, if at all, is the private language argument affected by taking the materiality of linguistic activity seriously?

3.1 Can a private linguist take a language stance?

Those who take a distributed language perspective concur that culture distinguishes human behavior from that of other animals. While animal behavior can be described as biomechanical process, human coordination requires additional explanations that depend on skills with cultural practices and artifacts (Cowley 2007, 2019). Although this may seem quite obvious, the statement’s consequences challenge commonsense views of language and cognition. They fit Wittgenstein’s critique of private language in tracing meaning-making to real-time social coordination that creates shared meaning spaces. This view rules out any possibility that the meaning of linguistic expressions derives from stable or private mental images. Indeed, if mental images or sensations are not self-intelligible, how can we seem to make sense of them? Both Wittgenstein and those taking a distributed language perspective trace mental phenomena and their meaning to sharable practices of coordination that arise from continuous coupling between producing expressions and generating reactions. Shareability is thus essential to meaning-making. On such a view, it is inconceivable that an individual could ever identify mental states without reference to the cultural and material resources that become familiar during socialization.

Wittgenstein’s private language argument leaves aside how children are socialized into hearing sound as meaningful patterns. If the socialization of hearing is addressed, one is bound to ask how a private linguist could grant identity to his own vocal expressions? In approaching this issue, I treat the situation of the private linguist as analogous to that of a baby learning to speak. A private linguist, like an infant early in semiotic development, cannot secure the meaning of expressions without shared practices that provide external criteria for correct vocalizations. Just like an infant deprived of participation in social routines, a private linguist could not know what s/he is referring to in naming ambiguous and fluid impressions or feelings. Moreover, without following a social rule, he or she could not use an ostensive definition to create a meaningful relation between expression and mental content. So, what is learned from comparing the situation of an infant to that of a private linguist? In early development, infants lack the ability to recognize a fluent stream of speech as consisting of meaningful and stable patterns. Learning to hear continuous vocalizations in terms of discrete patterns is necessary for granting identity to any vocalization. Cowley (2011) denies that physical resemblance is sufficient for recognizing the identity of what adults can hear as two vocalizations of the “same” pattern. This is because utterances are ambiguous, rather as are pictures in a Rorschach test. For example, in the first card of the Rorschach test the observer can recognize a moth or a cat or some other living or imaginary creature. However, this does not mean that the picture really contains a figure of a moth or a cat or indeed of any other shape. While there are more and less common associations, there is no one correct interpretation. The inkblot itself contains no figure and, for this reason, perceiving an inkblot derives from personal and cultural associations that allow one to project distinctive features of some entities on the inkblot (Cowley 2011). Further evidence for the claim that pictures do not contain ready-made objects can be drawn from the case of a Necker cube. The observer of this wireframe drawing, which lacks any visual cues that determine whether its front consists of a lower-left or upper-right square, constantly switches between different views and, by so doing, produces the illusion of a moving image. The point is that the physical structure of the Necker cube is not, of itself, sufficient to determine the correct interpretation of what is seen.

Cowley (2011: 11–15) argues that, like picture perception, speech perception draws on ambiguous stimuli. In both cases the process of construction is needed to perceive genuinely ambiguous stimuli in relation to relevant features. Figures and shapes perceived in pictures – like a moth or a cat in a Rorschach test – are not objects in any standard sense because they lack existence independently of the perceiver. Cowley (2011) uses Gibson’s term “a virtual object” to distinguish between regular objects and patterns constructed in the flow of perception. A moth in the Rorschach test or a particular view of the Necker cube are thus virtual objects.

The same applies to hearing a continuous and ambiguous stream of linguistic vocalization as a verbal pattern. On this view, the virtual objects are irreducible to the physical structure of sounds. Depending on how a person’s attending and hearing have been socialized, various objects will be identified. For Cowley, this is central to learning to hear virtual objects as part of vocalizing. The process of training a special attitude toward talk in interaction aimed at mastering a kind of listening that enables infants to discover virtual objects in the flow of vocalizations is called taking a language stance. Just as in the case of interpreting ambiguous figures, no pre-linguistic child could know in advance how to correctly map patterns onto features of vocalization. In fact, an infant cannot even realize that there are relevant features, since one and the same vocalization can be heard around different acoustic patterns. To discover regularities and stable patterns in speech, babies have to develop a special way of attending to speech sounds. Learning to hear vocalizations in terms of the language stance arises from attending to virtual objects.

Progress in recognizing virtual objects depends on participation in a rich environment of social routines. Since this presupposes a history of enacting social interactions, taking a language stance is a genuinely social competence (Cowley 2011: 16–18). Language stance can be at times conceived as similar to Dennett’s intentional stance (Dennett 1987). They both are conceptualized in terms of attitudes directed to identify virtual aspects of observed phenomena. Ascribing behavior with beliefs and desires is in a sense analogous to distinguishing virtual patterns in the flow of vocalizations. However, if we agree with Wittgenstein’s private language argument, an intentional stance seems to be groundless without the language stance. As Wittgenstein claims, there are no stable and therefore predictable mental phenomena without social interactions that serve as scaffoldings for the standardization of mental states. A private linguist cannot take an intentional stance just because he would not know to what s/he is referring to in naming ambiguous desires and beliefs. In order to secure stability and predictability of mental states, an intentional stance must be grounded in social coordination that ensures normative standards for taking a language stance.

Why does the language stance matter for the private language argument? In brief, infants have to hear speech sounds in a special way if they are to attribute identity to a caregiver’s vocalizations. This is a necessary condition if children are to gain the skills that are needed for linguistic and many other kinds of social competency. If taking a language stance must derive from engaging in social routines, a private linguist who had no opportunity to enact such a routine would not hear his own vocalizations as stable verbal units. As Cowley (2011) argues, taking a language stance serves as a kind of attention scaffolding that enables an infant to discover the verbal listening needed to recognize virtual objects in the vocalizations of their caregivers and themselves. Something analogous applies to the private linguist who has to distinguish virtual patterns in his/her own vocalizations. Wittgenstein forbids his private linguist basic conventions such as ostensive definition in that this would contaminate the sterile conditions of his thought experiment with remnants of public language. By the same token, the private linguist should also be deprived of the possibility of applying a language stance to his/her own vocalizations. With such a move, however, it would become impossible to produce stable expressions at all.

If the constraint appears too radical, one can consider a private linguist who can take a language stance. However, even under this concession, he would still be unable to use his/her original vocalizations consistently. This is because, in creating an expression to name a mental image, the private linguist would need to produce some kind of simple vocal gesture. At the same time, using a language stance, he will attend to one of many virtual objects that vocal gestures can evoke. How will he attend to this virtual object? He will concentrate his attention on some set of features of the sound experience. As Wittgenstein puts it “I concentrate my attention on the sensation – and so, as it were, point to it inwardly” (Wittgenstein 1953: 92, §258). This is roughly a description of taking a language stance. In the context, however, it serves as the standard criticism of the possibility of private ostensive definition (Wittgenstein 1953: 92).

The necessary condition for future correct use of the vocal gesture is that the private linguist will correctly remember the original virtual object attended to during the first use of the vocal gesture. Without external criteria for correct recall, the private linguist cannot ensure that in subsequent uses of a vocal gesture he will attend to the same virtual object as used in the original act of producing the vocalization. If it is possible for the private linguist to misremember the virtual object that he attended to in the original vocalization, he might not really know what exactly he is saying In the absence of external standards for correct repetition, he could not know what aspects of a vocalization are significant by virtue of referring to his own impression. It follows, therefore, that the private linguist is not only unable to secure the identity and stability of his own impressions and definitions, but is also powerless to ensure the identity of his own expressions.

3.2 Why a private linguist has limited metalinguistic awareness

David Olson’s developmental research provides independent support for Cowley’s (2011) abstract view of the language stance. Olson shows that, during individual development, the emergence of the idea of discrete linguistic units and stable meanings depends on acquiring skills in literacy. How does this bear on the abstract considerations regarding private language? Empirical data provided by a metacognitive theory of writing support the wide interpretation of the private language argument in clarifying the relation between an expression and its internal mental contents. Olson shows that the ability to read and write grants children both a new awareness of language and new ways of conceptualizing how thought relates to expression. If Wittgenstein’s private linguist is assumed to be preliterate, his task of securing stable meaning for private expression will be quite different from that of a literate person. Taking a glance at how children gain new linguistic awareness makes it possible to imagine the linguistic awareness of a preliterate private linguist and, thus, how his/her task would differ from that described by Wittgenstein.

Olson allies with the distributed language perspective in rejecting code models of writing and reading (Cowley 2019; Jarvilehto et al. 2009; Kravchenko 2009). He shows that, when children learn to read, they gain a new competence not only to decode graphic symbols, but also to perceive spoken language in the terms presupposed by use of the writing system. The thesis that writing consists of external objects that are easy to refer to – entities known as words, letters, and sentences – may seem trivial. What is less obvious, however, is Olson’s claim that people without written language have limited metalinguistic awareness. This observation can be illustrated by two interconnected claims of special importance for private language argument. First, the systematic practices of embedding spoken expressions into a stable written record grants preliterate children extended phonological and lexical awareness. Second, literacy practices establish new criteria for the identity of linguistic repetition (Olson 1994, 2013, 2016).

Studies on the development of phonetic awareness supports the thesis that alphabetical writing does not reflect preexisting metalinguistic knowledge but, rather, offers a model that seems to enhance recognition of speech-patterns. For example, to recognize the component parts of syllables and perceive words as composed of distinct sounds that can be represented by letters of the alphabet, elementary knowledge of writing is needed. Thus, in the classic task of manipulating phonemes, where participants are asked to remove the initial consonant “w” from the word “water,” for example, preliterate subjects fail (Morais et al. 1986; Read et al. 1986). Simply, literate and preliterate people segment vocalizations differently. This claim confirms and extends Cowley’s (2011) view of the language stance, since it suggests that recognition of virtual objects in the flow of vocalizations depends not only on language acquisition but also on experience with how materiality can be integrated with linguistic practice. Research on lexical awareness provides more details of this process.

Olson shows that literacy practices are also central to perceiving words as autonomous and objectified individual entities that are susceptible to systematic reflection and definition (Olson 2016: 98–114). More specifically, preliterate children confuse a word with a thing denoted or, in other terms, treat them as tokens that are fully integrated with the world. For example, when asked to say a long word, children often identify a long object by saying, for example, “train.” In studies by Homer and Olson (1999) children from 4–6 years of age were presented with the written expression: "Three little pigs." The experimenter read it aloud, and the child was asked to repeat the sentence in order to ensure that it was understood. Then, for example, one of the words was covered and the child was asked: “What do you think the sentence means right now?” Typically, the answer was: “Two little pigs.” Similar studies were conducted in a range of configurations. When the child was asked to pretend to write “Three little pigs,” the child produced three indefinite marks. When asked to write “One little pig,” one mark was made. The results suggest that, for preliterate children, word-tokens are intertwined with reality and can be exchanged for a real thing in a world. The idea of a word as a conventional symbol that can appear in the flow of vocalizations is to be traced to skills derived from learning to write (Homer 2009; Homer and Olson 1999).

Olson suggests that preliterate children lack a clear concept of "word" as an autonomous lexical unit. According to him, there is no evidence that the concept of “word” is universal (Olson 1994). Rather, it is a construct of literate cultures. As indicated by Chafe (1985) words are not basic units of talk in interaction. Rather, the building blocks of utterance last around two seconds, possess a specific intonation contour, and consist of several – five to seven – items to which a speaker can consciously attend. The acquisition of the term “word” arises with writing and reading skills. What is more, the first concept of a word that children acquire is associated with its graphic form.

Social anthropology further supports the link between literate skills and metalinguistic awareness. In investigation of how memory co-evolved with literate culture, Carruthers (1990) stresses that writing should not be conceived as an external storage system for information that would otherwise be stored in individual memory. The real impact of writing consists in creating new standards for memorizing. For Carruthers, whereas preliterate cultures apply loose standards for memorizing, literate societies treat written record or quotation as a model of correct recall. Jack Goody shows something similar: in preliterate cultures, adults fail to distinguish between literal and paraphrased repetitions of a story. This is not because they fail to remember but, rather, because they do not distinguish between a paraphrase and a verbatim repetition (Goody 1977). Developmental psychology delivers even more detailed evidence for tracing the distinction between literal repetition and paraphrase to literacy. Olson (1994: 126–127) and colleagues designed an experiment to show that preliterate children under the age of 6 find it hard to distinguish paraphrase and verbatim repetition. Child participants viewed a puppet theater scene featuring two characters: Big Bird and Teddy Bear. The scene exhibits a situation where Big Bird says a sentence that Teddy Bear repeats either literally or by using paraphrase. Although children between 4 and 5 years of age correctly identify Big Bird’s intentions and thus correctly and incorrectly paraphrase, only those above the age of 6 correctly perform both the paraphrase task and one using verbatim repetition. Anthropological and psychological data indicates that preliterate participants – adults and children – treat different expressions as identical when they have a similar meaning or express the same intentions. Even basic literacy grants participants new criteria for utterance identity: it separates literal meaning from intended meaning or a paraphrase. Olson shows how new criteria for the identity of expressions emerge from practices with external and stable artifacts.

Why does a metacognitive theory of writing clarify the situation of a preliterate private linguist? First, such a linguist lacks awareness of autonomous and stable linguistic units and he cannot use this knowledge in the process of creating his own private expressions. Without picturing a stable word-object, he could never know what aspects of an expression were relevant for a meaning. Indeed, were he outside a social context, he could not know that stable vocal patterns could be heard at all. These conclusions refine arguments based on considerations about the language stance (Cowley 2011). They are further complemented by the second conclusion from the metacognitive theory of writing where preliterates do not differentiate an expression from its referent. If a preliterate private linguist treated expressions as emblems – like children in Olson’s research – he would systematically mix vocal gesture with the impression it was intended to denote. In the extreme case, the vocalization would merge with the impression by creating an amalgamated semiotic structure whose performative functions create a language of magic (Malinowski 1923). Finally, as Wittgenstein stresses, in the process of meaning-making, to stabilize the identity of meaning and mental content one needs outer criteria. If a metacognitive theory of writing is correct, the same argument applies to identity of expression. Indeed, after gaining some experience with written artifacts, as Olson shows, people begin to apply new criteria for utterance identity. Without these, they are unable to distinguish exact and approximate repetition. A private linguist without literacy skills would be unable to ascertain if the subsequent applications of an expression were literal or approximate repetitions. Therefore, s/he would be unable to assure the stability of his/her own vocalizations.