My general point can be put this way: The expression by itself is not secondary; it is the USE that might be secondary—if only one could use it this way.
I think many people are like me in this regard and unlike Wittgenstein. I don’t know how to work with the fat Wednesday lean Tuesday example; Wittgenstein was apparently able to. And I think that this is partly why most people think that secondary uses are linguistic dead-ends. Wittgenstein’s choice of fat Wednesday as his primary example in the discussion about secondary sense does disservice to his discussion.
If I’m right, when people read Wittgenstein’s discussion of secondary sense, they tend to miss its point. And that’s because what grabs their attention in this discussion is the dead-end (which they can see) and not the use (which they can’t). They have the expression—“fat Wednesday”—they know it must be a secondary use (because Wittgenstein said so), they realize that they have no idea what to do with it, and they accept that to some people the expression is natural. They think that those people too have no use for the expression—that for them too it is a linguistic dead end—and they think that the only difference between them and those other people is that the latter find the expression natural and they don’t. They end up identifying the secondary-senseness of the example with its dead-endness. They come to think that this is the defining thing about secondary senses in general: that they are linguistic dead ends.
But this is not right at all. People just THINK they have an example of secondary use when they mention ‘Wednesday is fat’; but most people actually don’t. An expression is not an example of secondary sense for you if you don’t know how to make a secondary use of the expression.
If that’s right, we need another example to work with. One suggestion (also taken from Wittgenstein, so it has that seal of approval) is ‘e is lighter than o’ (BB 139). Here, many people do have the inclination to say that. They will even be able to sort the other vowels from light to dark. That is, they do know how to make a secondary use of that expression, and in this sense it is not a linguistic dead-end for them. (Another example that seems to work for most is the fitting of names to faces. Many people know how to argue about this.)
Instead of linguistic dead endness, therefore, I take the defining feature of secondary uses to be that they are detached from the regular technique of using the relevant words, and from the regular array of interests that we express using those words. The use of ‘lighter’ and ‘darker’ in ‘e is lighter than o’ is in a way extracted from the normal pattern of applications, cut off from the typical pool of reasons we have for calling upon these words. Secondary uses are not linguistic dead-ends, they are more like alleyways that run parallel to the main road, shadow it, draw on it, nourish from it, but never quite run into it.
I’m grateful to Ed Dain for the discussion