McEvoy was quoting a passage from Witters's "PI", which ends -- the passage,
not the whole "PI":
"Well, I assume that he acts as I have described. Explanations come to an end
somewhere.—But what is the meaning of the word 'five'? No such thing was in
question here, only how the word 'five' is used."
This reminds me of ... Grice.
In "Prolegomena to Logic and Conversation," Grice does quote from Witters's
"PI," but does not elaborate. Grice is concerned with Witters on
i. A horse can look like a horse.
Grice's point is that, for Witters, regardless of whether (i) is true or false,
or gappy, 'look like' is, to echo McEvoy's passage above, "is NOT used" like
that.
For Grice, truth is more important than use (or usage). And surely (i) is
_true_.
Why is it _odd_, or _misleading_ (and thus not 'used' much)? "Well," to echo
Witters (note that he starts that part of the quote with the very English
sentential opener, "Well, ..." which triggers its own implicatures), (i) is not
used much, unless you are a philosopher (recall Witters/Moore conversations, "I
know that's a tree") because it flouts a couple of what Grice conversational
maxims or in "Prolegomena" (and earlier in "Causal Theory of Perception" --
excursus --) "general features of discourse", of cooperative discourse, as it
were.
In this way, it may be argued that Witters is too hasty to conclude that
because philosophical explanations have to stop somewhere, meaning = usage. He
should have considered general features of discourse, but then he would have
been Grice, and we don't like that!
(i) compares with
ii. A: That horse looks like a horse.
B: It _is_ a horse.
A: I never said it wasn't.
For Witters, it seems, "looks like" is used when we have "A" and "B", so we can
say:
iii. The Mona Lisa looks like a woman I once met in Rome.
It would be uninformative for a Louvre visitor guide to tell the spectator that
the Mona Lisa looks like the Mona Lisa. But museum guides ain't philosophers,
not even Griceian ones!
And so on.
Cheers,
Speranza