ABSTRACT
In this chapter some objections to linguistic idealism are tabled, and it is conceded that the doctrine has to undergo a certain restriction if it is to retain plausibility, though not a restriction that undermines its spirit. There are several candidates for kinds of object that cannot be referred to that have been raised in the literature. These are divided into indistinguishable entities and undefinable entities. The first group comprises objects existing in certain physical, mathematical, or linguistic symmetries; the second comprises indefinable mathematical objects and perhaps some cases of physical or modal underdetermination. It is contended that physical and some mathematical symmetries can be broken by embedding relevant structures in a wider context. Recalcitrant mathematical symmetries, as well as linguistic symmetries, generate referential indeterminacies; but these arise anyway at the metalinguistic level (as the permutation argument shows) and are not a problem for linguistic idealism. That is because these indeterminacies reside in the nature of things; the world is in se indeterminate in the relevant respects. It is not as though there is something that language should do but cannot. By contrast, the existence of mathematical, and perhaps physical and modal, indefinables, forces a split-level approach on our ontology. At the basic level we can say that linguistic idealism holds: every object can be differentially named and described. But at a secondary or derived level unnameable and indescribable objects may arise. These latter objects are, however, constructed from materials available at the basic level: they are in a clear sense dependent on that basic level. So although the letter of linguistic idealism is not preserved in full generality, the fact that derived objects, some of which can escape its net, are metaphysically dependent on basic objects to which the doctrine applies without reservation, ensures that the thesis of linguistic idealism remains substantially and fundamentally true. There are in fact two main dimensions to the split-level approach adopted in this book: (i) we have a distinction between a primary, non-theoretical level of communication using true and false sentences, on the one hand, and a secondary level at which theoretical and semantical questions arise, and at which entities are posited, including objects (broadly construed), propositions, and the world itself; (ii) we have a primary level at which linguistic idealism applies sans phrase—everything in our ontology can be named and described—and also a secondary level at which new objects can be generated by the resources available at the primary level, and these new objects include ones that escape the net of language.
