Behavior and Philosophy, Vol. 23 (1995), pp. 47-52{1}

Thinking Without English

Barbara Abbott


The argument cannot be about whether it is possible to think in a natural language. As Hauser points out, we have direct phenomenological evidence of that ability. Furthermore it would be almost incoherent to suppose that we are unable to entertain mental texts expressing thought content given that we can produce physical ones that do that. The argument probably cannot be about whether it is impossible to think without using a natural language either. Convincing thought exercises show that at least some thought is imagistic rather than verbal. (One requires you to imagine a square, and then imagine two lines being drawn diagonally across it from corner to corner, and then imagine horizontal lines across it about a third of the way down from the top and about a third of the way up from the bottom, and then count the partitions of the square.) Note also the shape-recognition experiments that suggest that subjects are rotating mental images (see Shepard and Cooper 1982).

There are at least several other issues to argue about. One is whether or not most of the thinking that we human beings do is in our natural language. So stated the issue is pretty vague, but it may in fact be what most people have in mind when they write on this topic. More explicit statements invoke the concept of productive thought. That concept is clearly invoked by both Lycan and Hauser. Possible points of contention include (a) whether a natural language is required for productive thought, in principle or only in practice, and (b) whether all (or most?) of our human productive thought is in fact carried out in our natural language. Hauser would definitely answer (b) in the affirmative, but he sometimes suggests that he also holds the stronger belief entailed by (a). One problem here is determining what is meant by "productive". As Hauser notes, in this kind of context it usually means 'unlimited novel', but since we cannot experience unlimited quantities of things we could not really tell whether that property was instantiated in, e.g., the thought of a languageless creature. In natural language the property of productivity is the result of recursion in the syntax plus compositionality in the semantics, and these properties are rightly stressed by both Hauser and Lycan, the latter of whom attributes them to mentalese (the proprietary language of thought). They will also play an important role in the arguments given below. The first seven are replies to Hauser, and they are followed by several more which provide support for Lycan's assumption that the language of thought is not a natural language. However, my purpose is even more negative than Hauser's. In arguing against mentalese he is arguing that the language of thought is a natural language. I am arguing against that, but I don't have any very clear idea of what kind of mentalese I could be arguing for.

Seven Replies to Hauser's Seven Arguments

(1) Counterarguments from infrahumans and feral humans. It is not clear that nonhuman animals, especially relatively smart ones such as chimpanzees, cannot think productively. Even pigeons can be taught new concepts (see (Herrnstein, Loveland, and Cable 1976), (Herrnstein 1979, 1982), cited in (Griffin 1984)). Chimpanzees can solve problems using implements in new ways (Kohler 1956; see also Chevalier-Skolnikoff 1989), so apparently they can have a novel thought -- one part of productivity.{2}

Is there any evidence of compositionality? Compositionality would seem to be a necessity for dealing with the world. What good would it be to have mental representations of kinds of things and not have those representations figure compositionally in thoughts about those things? It is hard to understand how a chimpanzee could use a stick to get termites to eat (Goodall 1971) without forming thoughts about where termites are, what termites are good for, and how to get termites, where the chimp's mental representation of 'termite' contributes compositionally to each of these thoughts.

What about evidence of recursion? The clearest kind of case might be evidence of thoughts about propositional attitudes of others, since sentences about propositional attitudes require recursion in natural language syntax. Premack and Woodruff (1978) argue that chimpanzees do infer beliefs and desires of others (although they are apparently aided by having been taught a languagelike code (cf. Premack 1983)). Another indication of thoughts about thoughts comes from evidence of intentional deception on the part of primates (cf. Whiten and Byrne 1988).{3}

With respect to feral human beings, Hauser says "On PLT [the proprietary language of thought hypothesis], it seems one should even expect the behavior of feral or language-deprived humans to be as productive as that of fully competent speakers." It does not seem that way to me. A human being who lacks a natural language thereby lacks the main vehicle of social and intellectual intercourse with other human beings. But surely much of the grist for our own intellectual mills comes from interactions with others. Isn't it generally true, for example, that the more education one has the more, and more complex, thoughts one can have? Given these considerations, it does not seem to me that we can draw any negative conclusions from the impoverished thoughts of the language-deprived. On the other hand we do have evidence that Genie (the feral child discovered in Los Angeles in 1970) had thoughts before she had language. Curtiss (1977, pp. 185-6) describes Genie's recounting of memories from the time before she was found; and Curtiss 1988 (p. 98) provides the examples below, among others. (The glosses are Curtiss's.)

a. Genie bad cold live father house. ('I had a bad cold when I lived in my father's house.')

b. Father hit Genie cry long time ago. ('When my father hit me, I cried, a long time ago.')

c. Genie have Mama have baby grow up. ('I have a Mama who has a baby who grew up.')

These examples bring out two further, related points. The first is that Genie cannot really be said to have acquired English. The second is that the recursiveness in evidence here must have been driven by her thought rather than her (incomplete at best, and possibly nonexistent) knowledge of English syntax. Thus the balance of the feral human evidence would seem to favor the claim that a natural language is not necessary for productive thought.

(2) The calculator argument. The significance of this argument is not clear. We all seem to agree that a recursive compositional system is necessary for productive thought -- the issue is whether or not monolingual humans have one or two of these at their command.

(3) Another phenomenological argument. I have already allowed for verbal qualia. Problematic for Hauser would be the existence of (productive) nonverbal mental experiences. Many times I have had the experience of having a dream, major parts of which were inexpressible in English. The sense of these dreams is one of an alien intellect representing utterly strange relationships and goals. When we recount our dreams to others we automatically mold them to the constraints of our expressive powers, but typically this is at the expense of much of the original. Dreams seem certainly to be the result of a productive mental process, and their medium does not seem to be a natural language.

(4) The moron's objection. Whenever I try to write a paper I feel like a moron. Why is it so difficult to put my thoughts into English prose if they already are in English prose? Why is it so hard to make my pencil smart? What am I doing spending all these hours in front of my MAC if not translating my thought into English? And why is there such a thing as a "tip of the tongue" phenomenon if my natural language is my language of thought?

(5) The argument from empirical inaccessibility. How accurate is the knowledge of the thoughts of others that we get by talking with them? On the NLT hypothesis that our languages of thought are natural and other public languages, the task of communicating our thought to others who speak the same language should be trivial. All we have to do is open our mouths and speak the thought, or write it down. No encoding or decoding is necessary since, on this hypothesis, the thought already is in sentences of our natural language. But we all know that communication is not at all trivial, and misunderstanding of each others' mental states is the order of the day.{4}

(6) The argument against testability. It's not clear that the World War II decryption analogy works here. The Japanese did not know what the messages said. That surely makes the decryption task much more difficult. In our case presumably we will be able to arrange situations in which the semantics is given, so it should not be impossible to figure out what the brain forms are. (This kind of decryption is similar (although not identical) to the morphological analysis problems that are standard in introductory linguistics classes, in which students are given the unparsed target text plus an interpretation for the whole and required to identify the meaningful parts.)

(7) The argument against simplicity. My first answer to this was: What we're after is the truth, and Occam's razor can be damned; besides, since when did Mother Nature take the simplest route to anything? But Goodman tells me that this view is perverse (1972, 337).{5} Goodman also tells me that determining in actual cases which of two hypotheses is simpler is not always simple. I think that applies to this case. It is true that if our mental language is our outer language we have one fewer language to postulate. However then we postulate a discontinuity between ourselves and the rest of the animal kingdom. And this discontinuity may create further complexity in the explanation of our own evolution.

Some Additional Arguments

Only some of Hauser's arguments directly concern how humans think. The rest are indirect, but several more direct arguments cast gloom on his research program.

(1) The argument from the arts. When I play my fiddle I play completely by ear. I am not all that proficient, it is true, but I do make up embroidery to go with pieces that my band plays when I am taking a break (a solo), and I do not see how I could be using English to do it. Presumably jazz musicians do not use their natural language when they make music.{6} An artist friend tells me that when he used to make sculptures he was conscious of lapsing from an explicit English monologue with himself into languageless reflection on how to proceed.

The argument from the arts raises the issue of modality, about which I am prepared to say almost nothing. In a modular mental architecture of the kind described by Fodor (1983), inputs from different modalities are transformed by the input modules into some form that the central processing unit can deal with. The fact that we are conscious of different modalities suggests that this central processing unit can deal with different forms. If this is right, then the medium that the central processing unit is using must be one that is 'multi-modal', but I find my own central processing unit strained by trying to imagine what it is like.

(2) The arguments from other people.

(2a) Sachs' argument. Sachs 1967 reports that subjects in recall experiments were unable to distinguish syntactic variants of the sentences they had heard earlier, although they could distinguish semantic variants. This suggests that there is some kind of nonlinguistic, or at any rate non-natural-linguistic, storage vehicle. There is also more recent evidence that even short term storage is abstract and 'propositional' and does not retain the surface syntactic form of sentences (see Potter and Lombardi 1990, Reder and Kusbit 1991).{7}

(2b) Vendler's, Stalnaker's, and Hall's arguments. Vendler (1972) and Stalnaker (1976) point out a number of ways in which language differs from thought. We can make mistakes in what we say -- slips of the tongue, or faulty presentations of our thought; but it's hard to imagine making the corresponding types of mental mistakes -- intending to think one thought and thinking another instead, or thinking ambiguously. We can want to say something and not know how, but we cannot want to think something and not know how. We also talk very differently about beliefs than we do about sentences. We don't talk about the number of words in a belief, or about how long it is, or about what the constituent structure is, or about whether the subject contains a relative clause; and Hall (1985) points out that the verb believe cannot occur with a quoted sentence as object.{8}

(2c) Fodor's argument. The main argument of (Fodor 1975) needs to be mentioned here although I do so with some trepidation. The argument is that you need to know a language in order to learn a language. In more detail: to learn the meanings of words you need to be able to form and test hypotheses about what those meanings are, but to form such hypotheses you must already be able to express the meanings in question. The problem is that this argument, in conjunction with the currently common assumption that word meanings are not componential (i.e. they are not composed out of a relatively small number of semantic components which together give necessary and sufficient conditions for denotation) leads to the hard-to-accept conclusion that we are innately endowed with an extremely large mental lexicon.{9} Nevertheless the argument seems to be sound, and my own inclination would be to hesitate before throwing out a componential approach completely.

Back to: Abstract; Doing Without Mentalese by L. Hauser.

Go to: B. Abbott's Homepage; L. Hauser's Homepage.

Comments to: B. Abbott; L. Hauser.


Chevalier-Skolnikoff, S. (1989). Spontaneous tool use and sensorimotor intelligence in Cebus compared with other monkeys and apes. Behavioral and Brain Sciences, 12, 561-627.

Curtiss, S. (1977). Genie: A Psycholinguistic Study of a Modern-Day "Wild Child". New York: Academic Press.

Curtiss, S. (1988). Abnormal language acquisition and the modularity of language. In Newmeyer, F. J. (Ed.), Linguistics: The Cambridge Survey. Volume II Linguistic Theory: Extensions and Implications (pp. 96-116). Cambridge: Cambridge University Press.

Fodor, J. A. (1975). The Language of Thought. Cambridge, Massachusetts: Harvard University Press.

Fodor, J. A. (1981). The present status of the innateness controversy. In Representations (pp. 257-316). Cambridge, Massachusetts: MIT Press.

Fodor, J. A. (1983). The Modularity of Mind. Cambridge, Massachusetts: MIT Press.

Fodor, J. A. (1987). Why there still has to be a language of thought. In Psychosemantics (135-154). Cambridge, Massachusetts: MIT Press.

Goodall, J. (1971). In The Shadow of Man. Boston: Houghton Mifflin.

Goodman, N. (1972). Problems and Projects. Indianapolis: Bobbs-Merrill.

Griffin, D. R. (1984). Animal Thinking. Cambridge, Massachusetts: Harvard University Press.

Hall, R. J. (1985). An argument that the language of belief is not English. Philosophical Studies 48, 235-240.

Herrnstein, R.J., Loveland, D.H., & Cable, C. (1976). Natural concepts in pigeons. Journal of Experimental Psychology: Animal Behavior Processes, 2, 285-302.

Herrnstein, R.J. (1979). Acquisition, generalization, and discrimination reversal of a natural concept. Journal of Experimental Psychology: Animal Behavior Processes, 5, 118-129.

Herrnstein, R.J. (1982). Stimuli and the texture of experience. Neuroscience and Biobehavior Review, 6, 105-117.

Kohler, Wolfgang (1956). The Mentality of Apes. New York: Humanities Press.

Lycan, W. G. 1993: A deductive argument for the representational theory of thinking. Mind and Language, 8, 404-422. An earlier version of this paper was presented by Lycan at the University of Rochester Conference on Belief and Belief Attribution: Rochester, NY, May 1991.

Pinker, S. (1989). Language acquisition. In Posner, M. I. (Ed.), Foundations of Cognitive Science (pp. 359-399). Cambridge, Massachusetts: MIT Press.

Potter, M. C. & Lombardi, L. (1990). Regeneration in the short-term recall of sentences. Journal of Memory and Language, 29, 633-654.

Premack, D. (1983). The codes of man and beasts. The Behavioral and Brain Sciences, 6, 125-167.

Premack, D. & Woodruff, G. (1978). Does the chimpanzee have a theory of mind? The Behavioral and Brain Sciences, 4, 515-526.

Reder, L. M. & Kusbit, G. W. (1991). Locus of the Moses illusion: Imperfect encoding, retrieval, or match? Journal of Memory and Language, 30, 385-406.

Sachs, J.S. 1967: Recognition memory for syntactic and semantic aspects of connected discourse. Perception and Psychophysics 2, 437-442.

Shepard, R. N. & Cooper, L. A. (Eds.). (1982) Mental images and their transformations. Cambridge, Massachusetts: MIT Press.

Stalnaker, R. C. (1976). Propositions. In MacKay, A. & Merrill, D. (Eds.), Issues in the Philosophy of Language (79-91). New Haven, Connecticut: Yale University Press.

Vendler, Z. (1972). Res Cogitans: An Essay in Rational Psychology. Ithaca, New York: Cornell University Press.

Whiten, A. & Byrne, R.W. (1988). Tactical deception in primates. Behavioral and Brain Sciences, 11, 233-273.


1. The mental lexicon was estimated at roughly the size of the OED by Fodor in a talk at the University of Michigan in the mid 1980's. In general discussion at the SUNY Buffalo Cognition and Representation Conference Fodor expressed the view that the innate mental lexicon consists of roughly the monomorphemic core of English and other languages -- hypothesized to overlap extensively. See also (Fodor 981).^

2. In discussion at the Buffalo conference following this paper Jerry Fodor suggested a better property to use here would have been "systematicity", which has been characterized by Fodor and Lepore as roughly, the fact that any language (/mind) that can express (/entertain) the proposition P will also be able to express (/entertain) many propositions that are conceptually close to P. (If a mind can entertain the thought that aRb, then it can entertain the thought that bRa; if it can entertain the thought that P -> Q then it can entertain the thought that Q -> P; and so forth. (Fodor & Lepore 1992: pp. 175-6, footnote omitted.)^

3. The last three paragraphs are essentially a nod in the direction of a very large literature. Griffin (1984) is a readable discussion with many references.^

4. I wish to acknowledge the fact that this pat answer ignores or overlooks complex issues in appearing to suggest that sentences contain content.^

5. I must also thank Bill Lycan for alerting me to the insufficiency of my first answer and Rich Hall for helping me locate Goodman.^

6. Hauser has pointed out to me that this ability of musicians may well be considered to be grounded in an external recursive system and so does not argue against his main claim. However the argument from visual art immediately following does not seem as vulnerable to such a reply from Hauser.^

7. In fact it appears that non-outer-language mental storage is the default assumption of most cognitive psychologists and other cognitive scientists. See the remarks in (Pinker 1989, 360), which I thank Fernanda Ferreira for pointing out to me.^

8. Interestingly, Hall is careful to distinguish believe from think in this respect. See (Hall 1985) for discussion.^

9. The mental lexicon was estimated at roughly the size of the OED by Fodor in a talk at the University of Michigan in the mid 1980's. In general discussion at the SUNY Buffalo Cognition and Representation Conference Fodor expressed the view that the innate mental lexicon consists of roughly the monomorphemic core of English and other languages -- hypothesized to overlap extensively. See also (Fodor 1981).^