Comparison in terms of reference. A difference in sense does not necessarily imply a difference in reference. Overlap of reference is all that is needed for objective comparison in the sense that one sentence can preclude another from being true in virtue of referential relations without the sentences sharing any words or being of the same language/theory. This potentially enables objective comparison even in the absence of any deductive relation between the sentences themselves. But is there such referential overlap? Surely the answer is generally no if one thinks in terms of a description theory of reference (as Kuhn and Feyerabend seem to do e.g. in their discussions of Newtonian versus Einsteinian mass: there is no referential overlap because nothing can satisfy both descriptions). But this proves nothing because the description theory of reference has been discredited independently of the issue of incommensurability. Its problems include the facts that it makes most scientific terms not refer at all (e.e. Bohr's electron) and that descriptions are generally incomplete and fallible (e.g., descriptions notwithstanding, we may still recognise a three-legged albino tiger as a tiger). An alternative is the causal theory of reference (Kripke, Putnam), where reference is established as terms are introduced and then remain fixed. Entities may be defined ostensively (in simple cases) or in terms of the observable effects it causes (for theoretic terms such as electricity). Such definitions are not formal; rather the speakers will infer their"causal" essence for themselves. Thus there may be referential continuity without descriptive continuity. But the causal theory goes too far: it precludes reference change altogether (and, even worse, it seems to equate phlogiston and oxygen). To avoid this we must allow re-naming ceremonies as well as the initial naming-ceremony. It may seem that this leaves us where we started: no referential continuity. But Sankey thinks he has a way out in insisting that the only defining property of a theoretical term is its causal role. This (supposedly) separates phlogiston and oxygen and leaves the door open for (non-causal) descriptive change without change in reference. This is all extremely sketchy and it is quite impossible to see how it is supposed to apply for example in the case of Newtonian versus Einsteinian mass.
Untranslatability. This is a detour from Sankey's main argument. A perfect translation preserves extension in all possible worlds (cf. "unicorn"), i.e., identity of reference determinations. Translation may fail in two ways, as may be illustrated by the medieval concept of impetus. Its reference determination as "a force causing uniform motion in a straight line" is not permissible in Newtonian physics, and so translation fails. But impetus was also defined by ostension. From the point of view of Newtonian physics this was a separate "token" of impetus, and one which would translate as momentum. But faithful translation is still impossible since it would have to convey the presumed identity of these two reference determinations. Nevertheless, this does not preclude communication, as some have argued, because "we can learn a language ... from scratch, as a child learns them, without the detour through our native tongue" (Feyerabend, p. 113). Nor is the intranslatability thesis incoherent because Kuhn's and Feyerabend's proofs of it involves doing what they say is impossible, namely expressing ancient scientific terms in modern language. This objection misses the point sice scientific language is a sublanguage of the total language; expressability in the total language English proves nothing, for this is not the language into which translation fails.
Kuhn on reference. Kuhn said in Structure that Newtonian mass and Einstenian mass were "by no means identical." Sankey claims that this "seems to rule out common reference altogether" (p. 155; I disagree). He then claims that this is absurd because "neither term can refer successfully if it in fact fails to refer to stereotypical masses" (p. 155; it is of course naive to think that there is such a thing as "stereotypical mass"). To remove this alleged absurdity it is suggested that Kuhn's "referent" be read as "putative referent" (p. 159). Taken in conjunction with the description theory of reference to which Kuhn seems committed (in his proof that the two masses differ), this has, according to Sankey, "a number of objectionable features" because it "appears to yield a mistaken analysis of the relation between the two concepts of mass" (p. 160). E.g.: "it implies that Newton's theory cannot be contradicted by denying that mass is a conserved quantity. [Proof:] Given that the Newtonian concept of mass is a concept which is conserved, to deny that mass is conserved is ipso facto not to speak of the same kind of mass." And I suppose x=0 does not contradict x=1 because we are "ipso facto" talking about two different x's? Second example: "if Newtonian 'mass' only refers to mass if it is conserved, then certain sorts of experimental results are ruled out altogether. For example, it would be impossible to discover empirically that the mass referred to in Newtonian explanations of physical motion is convertible with energy." There is nothing "objectionable" or "mistaken" about this. On the contrary, it would be very objectionable if Newtonian physics did not rule out any experimental results. More importantly: it is precisely this possibility of an experimental decision between Newtonian and Einsteinian mechanics that Sankey wants to claim that his reference-based approach enables. Sankey is thus condemning Kuhn's approach as "objectionable" and "mistaken" precisely because it precludes the conclusion that he wants to draw.
In his conclusion, Sankey claims to have "deflated" incommensurability: "We have found no reason to take incomparability of content as the inevitable result of conceptual disparity between theories. ... The various referential overlap relations which may obtain between theories ensure that appropriately related statements from such theories may be compared with respect to agreement; so that empirical evidence may support one while disconfirming the other" (p. 221). But of course no one has ever claimed that conceptual disparity *must* lead to incomparability of content, only that it in fact does so in a number of actual cases. It is utterly ridiculous to pretend to have "deflated" this claim by asserting that it *may* be false. Vague allusions to "various referential overlap relations" prove nothing but Sankey's dilettantism. Sankey never clarifies what these referential overlap relations are supposed to be, why we should believe that they generally obtain in scientific revolution, or how they are instantiated in specific examples. On the contrary, as we saw above, Sankey seems to be operating with the extremely simplistic notion that either "mass" applies to a rock or it does not. The actual state of affairs is surely much more complex: the concepts of Newtonian and Einsteinian mass both try to pick out different but related properties of a rock. Whether this relation translates into a referential overlap which allows objective test is far from clear.