Published in METAPHILOSOPHY, Vol. 15, No. 1, January 1984, pp. 26-34.



I. Introduction

In this paper I wish to raise a puzzle about the view I shall call Linguisij Idealism: the view, as Richard Rorty put it, that "what appears to us, or what we experience or what we are aware of, is a function of the language we use."1 I will first define more fully what Linguistic Idealism (LI) amounts to. Then I will raise and explore a puzzle which arises when we seriously think through the implications of LI. Finally, I will consider ways out for the Linguistic Idealist. My aim is not so much to resolve the puzzle as it is to provide a prima facie objection to LI.

II. Linguistic Idealism

Philosophical positions that imply that a person's cognitive states (e.g., his propositional attitudes) are a function of his language or conceptual structure are not uncommon in recent epistemology. When such a position goes to the extreme of making not just propositional attitudes but the very objects (or contents) of experience, a function of language I will speak of Linguistic Idealism.2 For our purposes, the notion of objects of experience includes at the least pain, sensations, emotions, etc. The claim that I am going to explore is that the kind of object (or content) one experiences is a function of one's language. Ordinary common sense supplies us with a picture of experience opposed to such an idea. It pictures experience (or its objects) as prelinguistic in the sense that it is identically available to non-speakers (children, animals) and to speakers, as well as to speakers of radically different languages. In short, what you experience has nothing to do with the language, if any, that you speak. Surely this common sense picture is too naive for many cases, but it seems inescapable for experiences as primitive or basic as pain. How could the experience of pain be any sort of function of language?

All this is denied by LI. For example, it is rejected by Richard Rorty who embraces a version of LI in order to defend his version of materialism: eliminative materialism (EM).3 He thinks other materialists are wrong to claim a strict identity between sensations and brain processes. Rather, he suggests that it is possible that talk about brain processes will simply replace talk about sensations in the future as our conceptualization of our experience shifts to a materialist interpretation. The future neurological talk will not be equivalent to the old way of talking -- for, he claims, one cannot attribute the properties of sensations such as sharp and throbbing to brain processes (225) -- but it may be a more adequate way, according to certain criteria, to describe our experience. This change in talk and conceptualization amounts to what Rorty has called the disappearance of pain. In response to the charge of his critics that whatever language we use, if it is to be adequate to describe our experiences it will have to include ways of talking which express what we express by our sensation talk, Rorty begins to elaborate LI by announcing a doctrine of a sort of plasticity of experience: "I would claim that if we got in the habit of using neurological terms in place of 'intense,' 'sharp,' and 'throbbing,' then our experience would be of things having neurological properties, and not of anything, e.g., intense . . . [my critics] are taking for granted that there is a sort of prelinguistic givenness about, e.g., pains which any language which is to be adequate must provide a means of expressing . . . If it were the case that we experienced the same thing when we used the new vocabularly as when we used the old, then their point would be sound. But there is nothing to be this 'same thing' " (228).4

What is at stake here is the denial of what Rorty and Sellers call the Myth of the Given, which Rorty characterizes as "the view that awareness comes first and language must follow along and be adequate to the initial awareness" (229).

Rorty even goes so far as to say that "the notion of a non-linguistic awareness is simply a version of the thing-in-itself -- an unknowable whose only function is paradoxically enough, to be that which all knowledge is about" (229). Rorty makes his point about the possible disappearance of pain by imagining two groups of people, one raised to speak conventional English and the other raised "to use only neurological predicates in the place of those conventionally used in introspective reports" (228). (I shall call this modified language, Neurolese.) He believes a paradox is involved in a demand for a common prelinguistic experience shared by the two groups. For, what is that common experience, "the intensity of the pain or the X-character of the brain process?" (225)5 Rather, he suggests we admit "there is nothing in common between the two experiences save that they are had under the same conditions -- viz. the manipulation of the body in certain specified ways. That the 'same descriptive role' is played is not a matter of the same feature or features being reported in either case, but simply a matter of the two sentences being used to answer the same question -- viz. what do you experience under the following conditions?" (228) This is a special case of Rorty's more general rationale for LI: "To say that 'X's appear to us as F' is merely to say that 'We customarily use "F" in making non-inferential reports about X's' " (228).

To get a firmer grasp of these abstract formulations it will be useful to consider the example of two musicians trained in radically different musical traditions listening to the same piece of music. Independently of LI, it is plausible to claim that the same sound events will create two totally different musical experiences for the two musicians; e.g., the experience of a symphony with its themes, developments, modulations, etc. might not be open to a Balinese musician, just as, conversely, the significance of the sounds in a Balinese religious ceremony will be unavailable to the western musician. We should probably not think of the contents of their musical experience as just sounds accompanied by differing thoughts and inferences -- rather, we should think of the musical objects as just as much direct objects of experience as is a face. Moreover, the divergence of the contents of their musical experience is an automatic consequence of the two different musical contexts that the two listeners place the music in. Put another way, the same sound events might have a totally different meaning or function if they occurred in a Balinese gamelan piece than if they occurred in a piece of western music; just as it is an automatic consequence that the identical gesture -- e.g., shaking the head may mean different things in different linguistic contexts. It is plausible next to note that languages are systems of conceptualization which pervasively affect our experience in much the same way as specialized systems of conceptualization (e.g., musical training) affect special areas of experience.

My example works to connect the two experiences by involving, or appearing to involve, a level of experience common to the two musician's experience, namely, sound. A problem arises however: How are the experiences of two linguistic groups related if the experiences have nothing in common? For example, suppose one goes a step further, as I'm sure Rorty would, and questions the idea that both musicians are hearing sounds which they incorporate into different musical experiences. One might argue that sound itself is a western concept, part scientific, part metaphysics, which a Balinese need not share, so that according to LI his experience may possibly be comprised of different components than the western composer's. If this is granted it is hard to see how to describe in any neutral way what is being experienced in two different ways. This conforms to Rorty's claim that there is nothing in common to the two experiences.

Can the two experiences still be connected? Yes, by following Rorty's suggestion of mentioning the circumstances under which they occur, related experiences occurring under the same circumstances. Unfortunately, pace Rorty, for many of our experiences (e. g., headache) there are no standard circumstances or external causes, so the foreign experience which is to be related to (or replace) our experience will have to be specified by specifying the circumstances in terms of our kind of experience. Thus LI is the claim that under circumstances, described in our language, in which we would have had certain experiences, speakers of a radically different language will not have our sort of experience, but rather, some other kind, a kind which we presumably cannot describe in our language because by hypothesis this aspect of their talk is untranslatable into our linguistic practices.6 They will have experience of a certain sort because of the language they speak. For instance, the speaker of Neurolese will experience certain brain processes occurring just in case he would have felt a headache were he a speaker of English.

III. Problems for LI

In this exposition I have had to ignore a number of difficulties with this sort of position. Serious questions can be raised, for example, concerning the delineation of the language that experience is supposed to be a function of: is experience a function of what is typically expressed in a language, or of what can be expressed, and if the latter, how do we determine its boundary so that we can confidently assert that what can be expressed in one language cannot be expressed in another? Questions can also arise concerning the connection -- or, as it seems in this position, lack of connection -- between language and behavior and social practices. If the meaning of an expression is largely determined by the social practices in which it occurs, it will not seem terribly relevant to determine that in a future language the theoretical connections (which are the focus for linguistic idealists) are different as long as the social practices in which the expressions are embedded remain roughly the same.7

These questions are too general in scope for this paper. Rather than explore external objections to LI, I wish to raise an internal problem, one to which LI has no immediate solution. What does LI tell us about the experience of a bilingual, for example, a speaker of both Neurolese and English? Intuitively it seems that LI has to hold that the bilingual will have two different experiences at the same time and if that charge can be sustained it provides reason to doubt the coherence of LI.

Consider a specific stimulus, for example, a pin prick applied to the finger of a speaker of both English and Neurolese. In this sort of case the argumei can be set out as:

  1. A bilingual in both English and Neurolese is possible.
  2. LI implies that the bilingual would have at the same time two different experiences of a particular stimulus (e.g., a pin prick).
  3. For a person to have at the same time two different experiences of a stimulus (e.g., a pin prick) is impossible.
If these premises are correct, it follows that LI is at fault and to be rejected. I will now turn to objections to these premises.

IV. Objections to (1)

I take (1) to be a simple instance of a more general truth that mult-lingualism is always possible between any two human languages. It cannot be argued that the bilingual is impossible on the grounds that otherwise LI would have to be rejected, unless there are independent grounds for holding LI with stronger intuitive plausibility than that of the bilingual. I think this is obviously not the case. There are plenty of actual cases of bilingualism between two conceptually very different languages such as English and Hopi. Presumably, every normal human has the potentiality for learning any human language, so what would there be about Neurolese and English which would make it impossible to speak both?

In the formulation of LI it is presumed that the speakers of the different languages are essentially the same sorts of beings, namely, ordinary persons. In particular, LI has to hold that it is possible for us, just as we are now, to learn to speak Neurolese. But if we can speak either language, why not both? To hold that there is an incompatibility between being able to speak English and being able to speak Neurolese suggests that speaking Neurolese alters its speakers in some way. But surely this is an unsubstantiated, presumably casual, thesis which, moreover, suggests that a speaker of English could not learn Neurolese and vice versa. And this would mean that LI and EM do not apply to us because Neurolese would, after all, not be open to us, and so the claim that our experience might be of brain processes would be unsubstantiated on grounds of Linguistic Idealism. People would have to be rather different from us in some unspecified way to speak Neurolese.

V. Objections to (2)

One might be inclined to reject premise (2) on the grounds that as we speak only one language at a time, LI need not imply that the bilingual has any more than one sort of experience at a time. This won't do because what is at issue here is not "speaking language L" in the performance sense of "speaking". LI rnust be the thesis that the language one knows ("speaks" in the competence sense) determines the nature of one's experience. For, one surely has experience when one is not speaking (or not speaking of one's experience) or perhaps even in a position to speak, e.g., during mental break­down. Thus the bilingual being in the situation of having an inflammation of his eye will feel a sharp and throbbing pain in his eye just because he speaks ordinary English even though he says nothing; simultaneously, he will have a totally different sort of experience -- that of certain neurological processes -- just because he speaks Neurolese.

VI. Objections to (3)

In answer to (3) it might be pointed out that we can certainly have two different experiences, or experiential contents, in connection with the same stimulus as when one hears a sound which simultaneously makes one feel bored. This rebuttal only works, however, by mentioning experiential contents which are conceptually unrelated to each other; whereas the relevant Neurolese and English experiences are so closely related to each other that one is meant to be a replacement for the other. In a sense, the linguistic expressions of the respective experiences are alternative but incommensurable descriptions of the same situation. For example, to the identical question in the same context, "What do you experience now?" we are supposed to have two significantly different answers: "Throbbing pain in my eye" or "Oscillating E-fibers". These answers refer to the same experiential situation and they describe that situation in two different and incompatible ways. Moreover, LI is a significant thesis only in the case of two different languages which constitute experience in ways which cannot be treated as mere stylistic variants of each other. This is not to say that LI implies that; any two particular languages, such as English and Neurolese, are as a matter of fact incommensurable -- though proponents of LI will hold that many of them are -- but only that if they are incommensurable, they will differently constitute experience. In the present case, the whole point of the strategy of eliminative materialism is founded on the assumption of incommensurability of English and Neurolese and the related incompatibility of sensations and brain processes as objects of experience.

In purely logical terms, it is difficult to demonstrate the impossibility of a bilingual in two incommensurable languages having both sorts of experience at once. In the case of sensations and brain processes, assuming there is one content of experience, then as we have seen, Rorty plausibly holds that content cannot be a brain process and also have phenomenal properties such as sharp and throbbing. Intuitively, the impossibility of having two different experiences of the same stimulus is analogous to the impossibility of a physical object being in two places at once, and like that case the impossibility is not in any simply way analytic.

There is a way of defusing the problem about having incompatible experiences by taking seriously Rorty's claim that to have an experience of| certain sort is just to habitually talk a certain way, i.e., to use words which express or describe that sort. So to say that the bilingual both has pain and experiences C-fibers firing is just to say that he is both disposed to describe his experience with words like "pain", "sharp", "throbbing" as an English speaker, and disposed to describe his experience in neurological terms as a speaker of Neurolese. In short, he merely has two compatible verbal dispositions. This saves the position at the price of embracing an extreme behaviorism that collapses having experience into having verbal dispositions. As such it makes LI trivially true; the disappearance of sensations, for example, becomes no more than the claim that English and Neurolese are two different languages, a point which is not in dispute.

VII. Other Ways Out for LI

I shall now consider a way of avoiding the bilingual objection by weakening the relationship claimed between language and experience. In effect, the position I have been objecting to treats language as sufficient to determine a kind of experience if the subject of experience satisfies all the other bodily conditions such as being impinged upon by an appropriate stimulus, having| properly functioning sense organs, etc. Instead of this it might be proposed that language is merely a necessary condition for experience of a certain sort to have a certain sort of experiential content one has to speak a language which expresses that sort. Thus one could be in perfect somatic order, attentively listening to music, a speaker of English, but yet not necessarily hear sounds. In this way premise (2) is undermined: the bilingual will not necessarily have both sorts of experiences; being pricked by a pin he won't necessarily feel pain, nor C-fibers firing.

One might wonder what the missing factor is that is needed to make experience fall into a certain order. Perhaps we could think of it as a belief in, or commitment to, the language and its conceptual system. This is to say that it is possible to know and speak a language without being committed to its way of looking at the world. To be sure, we almost always are so committed, but this modified Linguistic Idealism points out that it is logically possible not to be committed; and it proposes to understand the bilingual as being committed to only one of the two languages because, as we have seen, it is impossible to be committed to both languages at once. His experience is that dictated by the language he believes in or is committed to.

Because of the conflicting tendencies of LI and EM I have kept EM in the background throughout this paper. Yet it is worth noticing that this version of LI seems fatal to the aims of EM as they were first delineated. For example, EM makes the claim that it may turn out that we are 'really' experiencing C-fibers firing when we think we are experiencing pain. Yet, if it is a necessary condition of experiencing C-fibers firing -- or in general, having neurological processes as the content of one's experience -- that one speak a language which describes such events, namely Neurolese, how can it be true that non-speakers of Neurolese are 'really' experiencing these events?

What causes this modified LI to have as many problems as the unmodified version is the extreme generality of its claimed application: simply to all experience. Treating language as a necessary condition of experience has some plausibility for sophisticated sorts of experience -- such as would be involved in attributing any rather complex predicate to an object of experience -- but the view under examination alleges to apply to all strata of experience, even at such a primitive level as to apply to pain. So a number of unlikely consequences follow. The most obvious, that no one can feel pain or have sensations unless they speak a language of a certain sort, may be accepted by proponents of LI. But it will be harder to accept the consequences that a speaker of ordinary English may not be experiencing pain -- even though he appears to be -- because he does not meet some other unspecified condition such as commitment to the language. Worse, lacking this condition (though there is no way we could tell this) he might not be having any sort of experience.8 Certainly the unfortunate who speaks no language has no sort of experience, not even of the primitive sort seemingly exemplified by sensations. I should think these consequences are absurd. For sophisticated sorts of experience we can apply this modified LI by holding that if the subject of experience does not speak the appropriate language, while not having the sophisticated content, he will have experience of some more primitive sort, which lacking the necessary language cum conceptualization he cannot construe in sophisticated ways. Thus he has some kind of experience, and one would have thought that pain would be a prime candidate for such basic (i.e., nonlinguistic) experience. Rorty's point, of course, is that one cannot draw any strict or absolute line between primitive and sophisticated experience, and speaking for EM he adds that it is possible that the line should not be drawn at pain. Yet to make this point he adopts a much stronger thesis, namely, that there is no line at all, i.e., there is no primitive or non-linguistic experience. We now have sufficient evidence that this leads to paradoxical consequences. If language is a sufficient condition, then he has the bilingual-puzzle. If language is a necessary condition, then he has the possibility of speakers who have no kind of experience at all

If, instead of all this, Rorty means to adopt the weaker claim that while there is a primitive level of experience it is possible that pain does not lie on that level, then it would have to be accepted as well that it is possible that pain does lie on that level. I would argue that this is what the argument about Eliminative Materialism ought to be about, and so it is illegitimate to make a blanket appeal to LI, as Rorty does. It must be shown that LI sensibly applies, for example, to pain. The conclusion we can draw concerning EM is that it is possible that pain lies above the line of basic experience. I suppose in a sense this shows that it is possible that pain disappear, but it shows it in a way consistent with Rorty's opponents being right as well. For, this means no more than that sensations are candidates for the application of LI. This establishes the weak conclusion that it is possible that EM is not meaningless. No one need deny that. The intuitions of common sense do not need to be characterized as necessarily true; pain is not necessarily basic. Common sense need only assert that as a matter of act it is basic.

VIII. Conclusion

What can be concluded concerning LI? Two crucial aspects emerge:
(a) What logical relation is LI asserting between language and experience? (b) How extreme is LI in application, does it apply to all experience even to basic experience? Together the answers to these questions characterize the varients of LI that we have considered. The idea motivating LI has merit, but I think we have seen that most of these varients are untenable. I believe the bilingual objection is a good reason to reject the versions which hold that (given the background conditions) language is a sufficient condition of experience of a certain kind. I think the puzzle that results when we take all levels of experience as being subject to language as a necessary condition shows that the only tenable position is to take language as a necessary condition only of sophisticated experience. Modified in this way, which no longer gives any particular support to EM, Linguistic Idealism is a position worth further exploration.


1 Richard Rorty, "In Defense of Eliminative Materialism", The Review of Metaphysics XXIV (September, 1970): 112-121; rpt. in Materialism and the Mind-Body Problem, ed. D. M. Rosenthal (Englewood Cliffs, N.J.: Prentice-Hall, 1971), pp. 227f. Parenthetical page references in the text refer to this paper.

2 This view is especially popular in recent philosophy of science; c.f., Thomas Kuhn, The Structure of Scientific Revolutions, 2nd ed. (Chicago: University of Chicago Press, 1970), ch. X. The Whorf-Sapir hypothesis in linguistics provides another example of such a view.

3 Rorty's original exposition of EM is given in "Mind-Body Identity, Privacy, and Categories", The Review of Metaphysics, XIX (September 1965), pp. 24-54; rpt. in Rosenthal, op cit. pp. 174-199. I think there is reason to doubt that his "In Defense of Eliminative Materialism", in which he expounds LI is entirely consistent with the position he is defending (EM). In the remainder of the paper my focus of attention will be on LI rather than EM.

4 The underlying assumption here is that the neurological terms are in no way equivalent to our ordinary sensation descriptions.

5 Again, the assumption is clearly that the two languages cannot be adequately translated into each other.

6 If it is translatable, there is no basis for a claim that their experience will be different. Nonetheless, there seems a difficulty in ever establishing that they are untranslatable. It could be argued that in our case of English and Neurolese we can express in English the neurological descriptions, indeed the neurological descriptions are already in a sense, a part of English. However, it isn't just a question of pairing expressions, or pairs of sentences, of English and Neurolese, but of pairing self-ascriptions, and here our current linguistic practice mitigates against translating ordinary sensation self-ascriptions into neurological ones.

7 It should be noted, however, that even on this view we could generate a version of LI just in case linguistic social practice should change. However, what would count as the disappearance of pain on this view seems more plausible: requiring the disappearance of pain-behavior and related social practices, this would simply be the literal disappearnce of pain.

8 Perhaps even while he is describing his experience.