Philosophy of Language:
The Early Modern Background

Curtis Brown

This handout provides a brief summary of the historical background to verificationism. Ian Hacking, in Why Does Language Matter to Philosophy?, defends the claim that the seventeenth- and eighteenth-century philosophers he discusses did not have a "theory of meaning" at all, but for our purposes it doesn't make a lot of difference whether they had no theory of meaning or a bad theory of meaning. So I will cheerfully describe their views about language as involving such a theory.

Descartes, Locke, Berkeley, Hume, and others had a rather well-worked out theory of ideas, a theory modeled on the recent success of the physical sciences. Science had made great progress by theorizing that macroscopic objects were composed of basic particles (or "corpuscles") that were very tiny; the properties of macroscopic objects were to be explained in terms of the interactions of the properties of these basic particles. Similarly, Locke and company thought progress in psychology would come from identifying the basic building blocks of the mind ("simple ideas") and explaining how more complicated mental equipment was the result of combinations of these building blocks. While these philosophers had a detailed picture of ideas, however, their account of language was essentially just grafted onto this theory.

The main thesis of the philosophers just mentioned was that words signify ideas. But a number of questions need to be answered before it is clear what this means.

1. What does signify mean? Hacking emphasizes that in his view it does not mean "mean." He thinks this because (a) it is not clear whether to describe their theory as an ideational, referential, or behavioral theory (see e.g. pp. 19-22); (b) signification seems to be a relation of precedence or consequence rather than one of meaning (20); and (c) the philosophers who held this view did not worry about the inverted-spectrum problem (how do I know you've got the same ideas I do under the same circumstances?), but should have if they were defending a theory of meaning (44-49).

2. What are ideas? According to Hacking, there are three main features of the seventeenth-century view of ideas.

  1. Ideas are nonlinguistic. We could perfectly well have the same ideas we in fact have even if we had never developed a language to stand for them. Language is necessary for communication, but is incidental to thought; indeed we think most clearly when we concentrate on the ideas our words signify rather than on the words themselves. As Hacking writes, "for real thinking . . . try to get as far as you can from words" (17).
  2. Ideas mediate between us and the outside world. For example, when we see a tree, the light waves which bounce off the tree and strike us in the retina produce in our minds an idea of a tree; we see the real tree indirectly, by virtue of directly perceiving this idea of a tree. When we think about the tree, again we think about it indirectly by directly manipulating our idea of the tree. (The idea that all our cognitive dealings with the world are mediated by ideas is closely related to what is called the representative theory of perception.)
  3. Ideas are like images. Images are ideas, for Locke and company, and in many ways they are the paradigm examples of ideas. Sometimes it seems as though the empiricists think all our ideas just are images; Hacking is careful not to go this far (31), but stresses that "we are aware of ideas through a faculty akin to sight" (33). Thus, for example, we can tell whether we have a certain idea or not by introspecting. We survey the contents of our minds by means of the "mind's eye," which is thought of very much on analogy with the real eye.

3. Where do ideas come from? The rationalists and the empiricists disagreed sharply over this question. The rationalists held that at least many of our ideas are innate. The empiricists, on the other hand, held that all our ideas ultimately come from experience There are two sorts of experience: perception, or experience of the external world, gives us ideas of external objects; reflection, or experience of our own minds, gives us ideas of mental contents and operations. For the empiricists, the recommendation that we not be bewitched by words but focus on the ideas they signify amounts to a recommendation to focus on the experiences relevant to the words in question. This is the fundamental idea the verificationists developed and modified.

4. Are there abstract ideas? The difficulties of the eighteenth-century picture of language begin to emerge when we consider whether there are any abstract ideas or not. Locke blithely assumed that there were; Berkeley and Hume argued that there were not, and Berkeley in particular found the whole idea of abstract ideas outrageous.

The apparent need for abstract ideas is clearest in the case of reasoning. In geometry, for instance, we are interested in reasoning about the properties of all triangles (e.g. we might prove that the angles of a triangle must sum to 180 degrees). It seems natural to regard such reasoning as a process of considering the concept of a triangle in the abstract and seeing what follows from it. For such reasoning, it seems, we need an idea of triangle that is not an idea of any particular kind of triangle: we seem to need, in short, an abstract idea of triangle.

Berkeley opposed the doctrine that there were abstract ideas with two main considerations. (1) He purported not to be able to find any abstract ideas in his own mind, and suggested that others who examined the contents of their own minds also would find no such thing. (This should already set us wondering about the capacities of the mental analogue of sight. Locke thought he discerned abstract ideas in his own mind. How could he have been so mistaken about the contents of his own thought?) (2) He argued that in geometry, for example, one need not have an abstract idea of a triangle: one can instead take an idea of a particular kind of triangle, then simply disregard the specific aspects of the triangle that are irrelevant to the proof one is engaged in. But hints at a problem with the view that meanings are like images, since in this case it seems that the idea of a triangle must be accompanied by a sort of commentary that explains which features are relevant and which are not. If to give the meaning of "triangle" we need not just an idea but also a commentary, we may wonder whether we can't just skip the image and make use of the commentary.

Criticisms of the 17th-Century View of Language

Disregarding Hacking's warnings not to regard the 17th-century view of language as a theory of meaning, let us anachronistically read it as though it were. Then the idea is that (1) words signify or mean ideas, and (2) ideas are like images. What is wrong with this sort of account? Here are some criticisms that have been made of it. The first two are Wittgensteinian in spirit; my formulations are indebted to Simon Blackburn, Spreading the Word (Oxford: Oxford University Press, 1984). The general idea behind them is that in a theory that explains the meaning of words in terms of ideas or images, the problem of how words apply to things in the world is liable to simply get relocated as the problem of how ideas apply to things in the world.

1. Images are not sufficient for meaning. Assuming we want to talk about anything other than particular objects, we will need ideas that apply to more than one thing. But how do we determine which things an image refers to? The image I usually associate with "dog" is an image of my old dog, Pom-Pom. (My sister named him.) But what determines whether the image refers just to Pom-Pom, or to all Pomeranians, or to all small dogs, or to all dogs, or to all animals, or whatever? As with the abstract idea of triangle, it seems that the image needs an accompanying commentary to explain which things it does and does not refer to. But if this commentary is in words, then the project of explaining the meanings of words in terms of ideas is a failure. (Cf. Blackburn, pp. 45-46.)

2. Images are not necessary for words to have meaning. At some level, the abilities which give content to our words cannot require images. Perhaps we could identify a red flower by comparing flowers with a mental image of red (though surely we often don't do it this way). But how do we identify which mental images are mental images of red? Not by comparing them to a mental image of red to see which ones match. But if we can just directly identify which mental images are red, it seems that we might as easily just directly identify which flowers are red. (Cf. Blackburn, p. 49.)

3. We associate no images with many classes of words. Or at least, if we do, they have nothing to do with the meanings of the words. I have no images for prepositions (e.g. "to," "for"), or for logical connectives ("and," "or," "if . . . then," and so on). And even if I did, it's hard to see how they could have anything to do with the meanings of the words. (Compare: Vladimir Nabokov reports in Speak, Memory that as a child he associated a different color with each letter of the alphabet. This must have been lovely, but it had nothing to do with his ability to spell.)

4. No account of sentence meaning. A more general difficulty with this sort of theory is due to its emphasis on the meanings of individual words. It's hard to get a coherent account of the meaning of a sentence by thinking of it as a collection of images, one for each word. For that matter, it seems that sentence meanings must be more than just a collection of individual word meanings, no matter what account we give of the meanings of words. A sentence is more than a list. This is one reason the positivists, who in many ways were the heirs of the empiricists, focussed on sentences rather than individual words; Quine e.g. has emphasized the extent to which taking the unit of meaning to be the sentence rather than the word was a significant advance. (See e.g. W. V. Quine, "Five Milestones of Empiricism," in his book Theories and Things (Cambridge: Harvard University Press, 1981.)