Property analysis:

--Derivative analysis: analyze a capacity by analyzing steps in a procedure.

--Specification of a capacity by the end it brings about; this leaves open many different sets of steps that can produce that end.

--Ordinary definitions are literal because each component property listed in the definition really is possessed by any object meeting that definition.  But a recursive definition... that doesn't list any properties of the objects at all. (The only way we get a property is by reading the recursive clause as a constructive process, amd stating that the object has the property of possibly-being-generated-by-that  -process

*****

Dilemma: on the one hand, clearly, recursive definition is not intended as an account of the actual, historical construction of, e.g., WFSs; instead, it is intended as an account of what "qualifies as" a WFS.

But: in what sense, then, can we say that '(p&q)' is a genuine part of '((p&q)&r)', but that '(p' isn't?
Answer: the recursive rules specify, in effect, not just whether a string is a WFS, but also what its parts are.  That's part of the information specified in the rules (rather trivially, since all the parts will also be  WFSs.)

--

*****

cartoon: *Hippocrates* April 96, p. 11:

