Prelude

A number of papers on Quine's philosophy that have recently been published share one remarkable feature: they attempt to re-enact the Carnap-Quine controversy from the early 1950s that is usually supposed to have been decided in favour of a thorough-going holism, precluding special pleading for theoretical positions over and above the business of ongoing science. Those re-enactments do not, however, concur with this orthodox view. All of them find some of Carnap's points against Quine attractive and tend to find the latter's all-out pragmatic holism difficult to swallow and at times paradoxical. Here is a brief account of some of the complaints raised by four authors.²

Take ``analyticity''. Richard Creath considers this to be a legitimate theoretical term within Carnap's philosophy. It's purpose is to explicate a certain feature of scientific discourse, i.e. cognitive frameworks and their pragmatic status. Quine's objections, rather than addressing the intuitions involved, simply disregard the task Carnap set for himself. Such a move is, of course, open to Quine, but his alternative account is seriously defective, according to Creath. Forget about a-temporal analytic propositions exempt from correction, still: some epistemic principles we happen to hold must be granted a privileged position as our doctrine changes, as it is bound to do. If this were not the case we could never distinguish between more or less coherent, simple and conservative theories.³ Unquestioned belief in the sufficiency of such extremely general naturalistic criteria in explaining the dynamics of knowledge seems to be Quine's own dogma.

Another philosopher reconsidering the notion of analyticity is Paul Artin Boghossian. Given an linguistic framework, what could ``the meaning'' of an expression consist in? A plausible answer is to point to certain inference patterns: Some inferences would have to be constitutive of an expression's proper use. According to Quine such features are not to be sustained: There is no fact of the matter as to which inferences one is obliged to accept in any given situation. This follows from the thesis of radical indeterminacy of meaning, as P.A. Boghossian rightly points out. But - bracketing the doctrine⁴ - Quine's claim turns out to be extremely implausible. If there is any recognisable semantical role an expression can be seen as playing, it has to be singled out by distinguishing between its meaning-constitutive and its peripheral uses. Stripped from its characteristic (or, equivalently, equipped with arbitrary) inferential features expressions are just so many accumulated tokens, devoid of logical structure and discursive conviction.

Analyticity is linked to a particular view of linguistic frameworks and discursive practice. As long as there are problems internal to some particular system of concepts, solving them implies a commitment to use those concepts. Carnap's claim was that we have to distinguish practice and a sideways-on reflection on practice delineating external questions implied by the choice of the appropriate linguistic framework.⁵ Quine, again, seeks to undercut this duality in favour of a pervasive scientific ``internalism'' lacking clear-cut and permanent divisions. But, as Hugh Price has pointed out: this sounds like an utterly abstract ontological principle and cannot easily be reconciled with Quine's own insistence on starting from empirical data.⁶We are to be dissuaded from dividing our cognitive activities into matters of factual investigation and methodological reflection - why? Quine's tenet - insofar as it submits to naturalism - is itself provisional and none too plausible at that. People do, as a matter of fact, master a considerable number of highly complicated, partially overlapping, meaning systems. It seems a sectarian move to legislate them out of existence on ontological grounds. Speaking of ontology, here is a final worry. Quine seems to employ the ancient philosophical strategy of determining true reality:

The goal of philosophical ontology is to determine what really exists. Leave out the ``really'' and there's no philosophy; the ordinary judgement that there exists a city called Chicago stands unopposed.⁷

Naturalistic (like any other) ontology has to be able to identify the basic stuff the world is made from.

Stephen Yablo, in his draft, argues that Quine's ontological parsimony can only be effectively achieved by distinguishing the literal and the metaphorical. Quine owes us an account of what terms are allowed into the most general scientific theory that, by being true, determines what the world consists of by conferring ontological status to the values of its variables. These terms have to be separated out from expressions serving the purpose of elucidation and intermediate cognitive motivation. The literal/metaphorical distinction is, however, certainly more elusive than Carnap's split between synthetic and analytical sentences.

These short remarks are, admittedly, only gesturing towards the crucial issue: How seriously should Quine's holism be taken? There seem to be several reasons to defend Carnap's two-step approach and advocate a different treatment for matters of fact and questions of how to analyse various determinate, discursive formations respectively. As Richard Creath puts it

$\ldots$ we have two programs of epistemological research: Carnap's exploration of the role of conventions of language and Quine's investigation of global justification. As programs of research perhaps neither should be reviled as dogma.⁸

This paper will not present additional material to decide the Carnap-Quine controversy. At the present stage the discussion might profit from a more general point of view. I want to highlite an impressive theoretical construction embracing both sides of the alleged dichotomy. I am referring to a line of thought manifest in Wittgenstein's pre-Tractarian Notebooks and the remainder of this presentation will be an argument to the effect that the most prominent debate in post-war Analytic Philosophy had - in rough outline - already been resolved by the early Wittgenstein. His construction, it is true, is worked out within a different vocabulary. But I shall try to convince you that Wittgenstein offers an in principle account of how to reconcile reductionism and holism, logical analysis and open-ended explorative flexibility, in short: just the two counteracting tendencies at issue in the discussions briefly surveyed in this section.