id:contractio:20071115:p1
|
|
全文読めてしまいます....。
http://ru.philosophy.kiev.ua/library/goodman/07.htmlVII On Rightness of Rendering - 7. Rightness Reviewed
Briefly, then, truth of statements and rightness of descriptions, representations, exemplifications, expressions — of design, drawing, diction, rhythm — is primarily a matter of fit: fit to what is referred to in one way or another, or to other renderings, or to modes and manners of organization. The differences between fitting a version to a world, a world to a version, and a version together or to other versions fade when the role of versions in making the worlds they fit is recognized. And knowing or understanding is seen as ranging beyond the acquiring of true beliefs to the discovering and devising of fit of all sorts.
Procedures and tests used in the search for right versions range from deductive and inductive inference through fair sampling and accord among samples. Despite our faith in such tests, their claims as means for determining rightness may often seem obscure. Indeed, rather than being able to justify our confidence in inductive inference or in the procedures for taking fair samples, we look to the confidence itself for whatever justification there may be for these procedures. Choosing "green" rather than "grue" as projectible, or stirring and shaking a barrel of seed, may seem like rain-dancing — ritual with some celebrated successes and some dismissed failures that is cherished until too disastrous or disreputable. But so sour a view betrays a discredited demand for justification as convincing argument that a test or procedure will ensure, or at least improve our chances of, reaching right conclusions. We have seen, on the contrary, that rightness of categorization, which enters into most other varieties of rightness, is rather a matter of fit with practice; that
without the organization, the selection of relevant kinds, effected by evolving tradition, there is no rightness or wrongnessof categorization, no validity or invalidity of inductive inference, no fair or unfair sampling, and no uniformity or disparity among samples. Thus justifying such tests for rightness may consist primarily in showing not that they are reliable but that they are authoritative.
