Dear SftP people,
The disagreement between Ivan Handler & Michael Goldhaber
masks a profound point. In old-fashioned game theory (which is
sufficiently up-to-date to accommodate the point), a play of the
game would correspond to what handler wants to call a behavior.
But the player doesn't just have one play of the game, because
there are inputs from the other player --or also, of course, in
more realistic models, inputs from the environment. Accordingly,
we speak of a player having a strategy: that's a rule leading to
a play in each of the possible circumstances. Though it's not
emphasized in a first exposition of the idea, we note that no
real player lugs around a handbook of what-to-do-in-all-
circumstances; but there is some reaction pattern; maybe we ought
to say "reaction pattern" where we say "strategy".
In the same way, a language is not a bunch of utterances,
but a bunch of potential utterances. And, though it's not
emphasized in the textbook I last taught this from, a reaction
pattern: each speaker has a way of producing, in a given case, an
utterance --a different rule for a different speaker. I don't
agree with Goldhaber that proto-language of apes or dogs offers
no material for potential analysis in linguistics. The essential
point is that, even if our analysis does not introduce an elusive
concept of "meaning" (probably it should, but some might try to
duck it), it must recognize that the space of speakers'
strategies (in the game-theoretic sense) is enormously bigger and
differently structured than the space of utterances.