"Any such definition would depart from the lexical meaning (and the current use of the term in AI) unless it allows that there can be pro and con arguments": That is unquestionably so, but I do not understand why the possibility of counterarguments has to be part of the definition itself. I have seen definitions that make it so. For example, Sinnott-Armstrong and Fogelin in the 9th edition of Understanding Arguments define an argument as "giving reasons for or against some claim" (my emphasis). In the current edition they have a different definition that contains nothing corresponding to the "or against" part, and quite rightly so in my view, because it is redundant. If I argue against a claim or position or argument, I must be trying to establish some claim about that claim or position or argument--that it is false, confused, misleading, unwarranted, or whatever.

So to define an argument as, e.g., "an attempt to prove or establish a conclusion" (Robert Ennis's definition in Daryl Close's collection), does not in any way exclude the possibility of an argument being directed against a claim or a position or an argument, as far as I can see.

Miles Rind

Douglas Walton wrote: