Questions like "what is the meaning of a word?" paralyze us because we feel there must be a thing called "meaning" that we ought to be able to point to. We should ask a different question: "what is an explanation of the meaning of a word?" We should learn the grammar of the word "meaning" from the form these explanations take.
Ostensive definition—by which I explain a word by pointing to the thing it denotes—involves an act of interpretation. If say "pencil" to a non-English speaker while pointing to a pencil, that person might interpret "pencil" to mean "wood" or "one" or "hard." So perhaps it is the act of interpretation that gives a word meaning.
If we define understanding as interpretation, we conceive language as signs for words that are dead until a mental process, understanding, gives them life. There is the mechanical process of manipulating these signs, and the organic process of giving them meaning. But if the mental process of understanding "red" is simply a matter of picturing the color in the mind, we could replace understanding the word with the physical process of looking at a chart with colors in one column and the words for these colors in the other column. So we are wrong to think of understanding words as a mental process, Wittgenstein argues. Words are brought to life not by mental processes, but by their grammatical use.
We are often puzzled by how the mind and mental processes work, not because they are complicated, but because we are confused. We think of the mind as the substance through which mental phenomena pass. We think of the mind this way not because of any experimental evidence, but from a feeling that something of the sort must exist.
If thinking is a matter of operating with signs, then we can think with the hand when writing or with the larynx when speaking. We speak literally when we say thinking takes place on paper or in the mouth, but we speak metaphorically when we say thinking takes place in the head. Thinking with the hand and mouth is not analogous to thinking with the head. If the cases were analogous, someone could locate where in the head a particular thought takes place. We could say, for example, "the visual image is located two inches behind the bridge of my nose." Perhaps someone could make such a claim, but we would not understand it, for we have not learned that way of using those words. Similarly, we do not understand the diviner's claim that he feels in his hand that there is water three feet under the earth. It is not simply that we do not know the diviner's art. We also do not know what it would mean to have a feeling in one's hand that corresponds with an underground current. If we are to understand him, he must explain how he came to learn how to estimate the depth of underground currents.
There are two ways of explaining how we come to learn things, two ways of answering the question "why do you proceed as you do?" The first is an explanation of a cause, in which we hypothesize about how our environment has shaped us. The second explanation is a justification our behavior based on a set of rules or norms. Both identifying causes and following rules answer the question "why do you proceed as you do?", so we think of the two explanations as analogous.