NNEP (University of New Hampshire)

Steve Bayne

I would like to thank the people at the NNEP as well as the hosts at the University of New Hampshire. Interest in Sellars is high at New Hampshire and since this paper was inspired by Sellars this seems a fitting place to offer these remarks. In the late 1960's I had occasion to hear Wilfred Sellars lecture from time to time. On one of those occasions while discussing verbal dispositions and the nature of thought, he asked his audience to consider what is going on when in the middle of a sentence a speaker pauses and then, following some reflection, continues. My recollection of the substantive content of his remarks faded over the years, but the issues they raised continued to exercise considerable influence over how I would come to view the nature of thought. In what follows I would like to discuss what it is to think in words, based on Sellars' interesting thought experiment.

What actually led Sellars to entertain this particular thought experiment I don't think we will ever know for sure, but I do have my suspicions, and here is one of them. Chomsky in 1959 attacked behaviorism famously, using arguments that even today are quite persuasive. One argument was that language is generative, that is, from a small number of elements an infinite number of novel sentences is available to any competent language user. It is unrealistic to believe that the occasion of use for each sentence is the exercise of an acquired disposition to behave. The problem is that behaviorism makes language unlearnable. If, as behaviorists had suggested, language is learned in chunks as big as sentences, what would explain suppressing mid-sentence the disposition to utter a sentence?

In his 1938 lectures at the University of Chicago, Russell averred that a sentence could be regarded as a series of words (IMT p. 38). About the same time Carnap in the Logical Syntax of Language generalized the notion from sentences to any linguistic elements, saying,

The syntax of language is concerned with the structures of possible serial orders of any elements whatsoever. (LSL. p.6).

However, Chomsky in Syntactic Structures would later devastate this simple picture of sentences as word strings in his critique of Markov chains as providing a Turing computable model of natural language. Still, it was the idea of a sentence as a series of words that actuated Wittgenstein in his discussion of the relation between sentences and thoughts. In a rather remarkable passage, Wittgenstein expresses the view that sentences regarded as series of words relate to thoughts the way series of numbers may be viewed as related to algebraic formulas:

The lightening like thought may be connected with the spoken thought as an algebraic formula is with the sequence of numbers which I work out from it. (PI 320)

Earlier in his Philosophical Investigations, Wittgenstein invited his readers to consider a language game in which one person orders another, following presentation of a series of numbers, to continue the series in the same way. The important thing here is the idea of continuing the series, as in Sellars' case where the speaker pauses and then continues the series of words. The question that leads us in the direction of thought and not just sentences is this: Why would a speaker stop? And what would cause such a speaker to begin again, that is, continue the series? Wittgenstein considers an analogous case in the language game we just described, the person is given the order to continue the series

...but when he wants to go on he hesitates and can't do it. Are we to say that he was wrong when he said he could go on? (Wittgenstein PI 181)

In the case of Sellars' speaker, he hesitates. But the question is why? He may hesitate because he realizes he is about to use an offensive word; but he may also hesitate because he doesn't really know what he wants to say; his thought is not clear. As Wittgenstein remarks:

How do I choose the 'right' word? ...Without doubt it is sometimes as if I were comparing them by differences in smell...(PI 218).

Our speaker cannot find the right word, and this explains his hesitation. Does he find the right word - only if there is such a thing and not just a better word - that is in the same way he might find the right continuation of a mathematical series?

How do I know when I arrive at the right word and can now continue my sentence? Indeed, someone may help me out and I say "Yes! That is the very word I am looking for!" Does such a person know my thoughts better than I do, or is it rather than I am receiving assistance on what to think? I believe the latter is the case: we arrive at the right word not by applying a strategy in lieu of comparing all the possibilities but by building on what has come before the word we seek. It is the preceding word or structure that acts as pointer toward the active site of sentence generation, and it is thinking that does the generating. Such processes result in language in use as what contains the thought in some yet unspecified sense of 'contain'. Wittgenstein is right, then, when he remarks

When I think in language...the language is itself the vehicle of thought. (PI 329)

But this brings us to another issue, that of whether language is indeed primarily for packaging thoughts or alternatively, perhaps, primarily for communication. So let us return to the matter of what thinking in words is.

A mocking bird searches for the "right" chirp; an artist searches for the right combination of oils from the palette. The artist may be duplicating another painting but is he "thinking in oil"? Is the mocking bird thinking "in" chirps? Are we working with the right idea of what it is to think "in" anything?

Gilbert Ryle offers a compelling argument that we never think "in" words:

Perhaps while composing his speech the word 'apolaustic' occurs to him, so he considers it and rejects it as too pompous, or because he is not sure what it means. He thought it up, he thought about it and he scrapped it for a reason. But surely we cannot say that his thinking was "in" this rejected word? What of the phrases and sentences that finally constitute his speech? Had he thought in them while composing? Obviously not. He had then been *searching* for them, and his searching could no more be described as in its objectives than prospecting could be described as being done in diamonds or in nuggets of gold. ("Thought and Soliloquy" in On Thinking by Gilbert Ryle. ed. by Konstantin Kolenda).

When we are said to think "in" words what is meant is that what we do consists in doing things with words as writing "in" sand is doing something with sand. Similarly, when we multiply "in" Roman numerals it is just that we are doing something with Roman numerals. We need to distinguish the activity of thinking from the tokens that make up the sentence expressing what we think. Ryle's objection, therefore, is not a good reason for rejecting the idea that we think "in" words. More accurately thinking is what goes on "between" words. Earlier Ryle had argued that the only difference between thinking to ourselves and speaking our thoughts is that in the former case we are "keeping our thoughts to ourselves" (Concept of Mind p.27). But is an inner whistling the withholding of an outer whistling in the "same" way that imagining the song being played on the piano is withholding actually playing it on the piano? Inner speech is not just withholding outer speech. It is one thing to pause in search of the right word. It is another thing to interrupt a sentence in order to be discrete. I, therefore, reject the idea of thinking in words as simply withholding spoken words.

At first, it seems to make sense to side with Prior (Objects of Thought. Oxford. Oxford. 1971. p. 14), who maintained that while we may think "in" sentences we do not think sentences. But we need to separate 'thought' as a propositional attitude meaning roughly 'believe' from thinking as an activity, more particularly as something conative. So I may say that I was thinking while washing the dishes I do not say I was believing while washing the dishes. If we suppress our inclination to think of thoughts as taking objects and instead view thoughts as events or activities, it is easier to think of sentences as things that can be thought. By treating thought as an "attitude" towards propositions we have been held to a different conception, one that places thought alongside acts like beliefs and recollections, rather than alongside processes such as computation and deliberation.

Let us say, then, that a willful act may require an act of will but that an act undertaken thoughtfully may not require an act of thought. Washing dishes is not an act of thought, although it may be undertaken thoughtfully. Think-ing in words may eventuate in a thought without entailing the existence of a thought. Now we ask: Is searching for the right word part of thoughtfully building a sentence the way carefully trying to get the right amount of sugar is part of thoughtfully preparing a cup of tea? (Compare preparing YOU a cup of tea.) Sentences thoughtfully constructed express thoughts. Cups of tea prepared thoughtfully do not express thoughts. A cup of tea that is not thoughtfully prepared is still a cup of tea. A sentence if it is to express a thought is something that must be thoughtfully prepared in order to be.

Suppose I slap my knee and you come up with the right word. Then I do it again with the same result. What is strange here? To think a thought is to "have" a thought; but, to picture a picture is not to "have" a picture. Don't we want to say that there must be the "right" connection between thinker and word for the word to be the right word? Otherwise you might think that I am merely controlling your thoughts or claims of rightness. One objection to the idea of thinking in words is its inwardness, its privacy. Imagine I am able to effect a neurological hook up from my brain so that my discursive thought processes are expressed as the configuration of beads arranged on an abacus. Here I might come to believe that there was some causal correlation between my verbal images and the arrangement of beads. If I can do this, then I can also imagine all my verbal imagery completely disappearing, although I could know what I was thinking by looking at the abacus. Here there is no question of privileged access. I can even imagine taking the abacus into a room where I could not see it, perhaps even having you tell me from the other room what I am thinking. Conversely, I can imagine having the abacus removed and knowing my thoughts by restored verbal imagery.

Suppose I had to look at the beads on the abacus to know what to say next. This sharply contrasts with using inner speech as a "model" of what I am trying to say. Suppose others could see the abacus; what would be the point of the move from abacus to speech; indeed, what would be the point of the move from covert to overt speech? I will now consider the difference between a mathematical series and a series of words which occur while thinking in words.

Continue a series one way and you have one series; continue it another way and you have another series. But continue a sentence another way and yet you may retain the same thought.

When a continuation of a series affects what the "right" next element is we must ask "Is this really a series?" But any word we choose may affect the next choice of words, as we think aloud.

If the right word were given by a rule then using that word would be obligatory in the way that a rule for the series makes the following entry obligatory. In other words, if only one word "fits" then that word must be used or the sentence is never competed. But this is hardly ever the case, and so it appears that more than one word is bound to "fit" and that the right one is not the one the semantics deems obligatory. Then why is it used? There is in the case of assertion no intended word. The occurrence of a word does not allow us to infer what sentence was intended. None was.

Phenomenologically there are similarities between our inner speech and reading; but there are differences. When I read there is no question similar to the one we have been raising about the right word, although it would appear that I seldom if ever pause mid sentence in inner speech to search for the right word before continuing my inner sentence. But does this mean that in the case of inner speech we are simply reading inner words? If I pause, while reading, over some particular word, there may be a question whether I can read the language of the script, whereas if I pause while thinking in words aloud there is seldom the question whether I know how to think.

Pausing during "spoken thought" differs from the student pausing when asked to continue a series the teacher has written on the board. Wittgenstein, was discussing a two person type "game" where we have a student continue a series the teacher began, or an observer watching a teacher and a student continuing a series (PI 157). It is only taken in the reflexive case that the problem deepens, an issue of understanding becomes an issue concerning the nature of thought.

When Wittgenstein says:

We take a sentence and tell someone the meaning of each of its words; this tells him how to apply them and so how to apply the sequence too. (PI 175e)

This can't be right. What Wittgenstein overlooks is that knowledge of how to apply a sentence correctly includes knowing how to apply a sentence in dialogue. Compare Wittgenstein's arithmetic game of continuing a sequence in the "right" way with another game. Only in this one a complete sequence is given and the game is to answer with another sequence. In other words the game is played with a sequence of sequences. If I am to reply to a sequence with the correct sequence I cannot rely on my knowledge of elements of a sequence in continuing a sequence of sequences. Similarly, I cannot rely on my knowledge of words to sustain a meaningful continuation of a dialogue. Yet another reason for separating two cases: first, continuing a mathematical series, and second, coming up with the right word in continuing a series of words in thought (always compare to finding the "right" word in Quine type radical translation).

Summarizing, I have concluded that there really is such a thing as thinking in words. Following twists in the thread Sellars was attempting to unravel, I have argued both that continuing a sentence is not so much like continuing a mathematical series, and that thinking in words is not in any compelling way yet to be made explicit the exercise of verbal dispositions.

Bleached of soul, finite, transitory, and bound to an earthly body, speechless thought moves over the fibers of our brain like water over the face of the deep. That we think, we know because we think. That we are, we know because we think; but what is it to think? And what is it to know that we think?