The following is my contribution to a symposium on Richard Gaskin's The Unity of the Proposition. The symposium, together with Gaskin's replies, is scheduled to appear in the December 2009 issue of Dialectica.
GASKIN ON THE UNITY OF THE PROPOSITION
William F. Vallicella
While studying Richard Gaskin’s The Unity of the Proposition (Oxford 2008), the word ‘magisterial’came repeatedly to mind. Gaskin’s mastery of the history, literature, and dialectical intricacies of the problem of the unity of the proposition in all its ramifications is in evidence on every page. More than a treatment of a particular problem, Gaskin’s book is a systematic treatise in the philosophy of language organized around a particular but centrally important problem. To my knowledge, it is the most thorough and penetrating discussion of the unity of the proposition ever to appear. The fact that Gaskin’s solution to the unity problem is set within a systematic philosophy of language contributes to the book’s depth and richness, but also makes the task of the critic difficult. In a few pages, the critic cannot properly convey the systematic underpinnings of Gaskin’s formulation of the problem and his solution to it. And when the critic evaluates, he is forced to acknowledge that he is evaluating a solution embedded in a far-flung system whose ideas are mutually reinforcing. His critical points may then appear as ‘dialectical potshots’ if he cannot, as he cannot in a few pages, bring a competing system of mutually reinforcing ideas onto the field. These caveats having been registered, I proceed to sketch Gaskin’s project and raise some questions about his formulation of the unity problem. After conceding that Gaskin has solved the problem as he understands it, I will suggest that the problem lies deeper than he recognizes, and that the linguistic idealism in which he embeds his solution is problematic.
1. Gaskin’s Project
Gaskin’s main objective is to solve the problem of sentential/propositional unity. (vi) Since propositions "enjoy metaphysical priority in respect of unity" (viii) over declarative sentences, the central problem concerns the unity of the proposition, where a proposition is the meaning of a declarative sentence. Meaning embraces both sense and reference, and Gaskin posits propositions both at the level of sense and at the level of reference. The question of the unity of a proposition, whether at the level of sense or reference is: "what distinguishes propositions from mere aggregates of their components, and enables them to be true or false?" (18) What makes a proposition a unity capable of being either true or false, when neither the aggregate nor any member of the aggregate is capable of being true or false? The problem of unity is prior to the problem of truth since "false propositions are just as unified as true ones." (viii) This holds both for propositions at the level of sense (Fregean propositions) and at the level of reference (Russellian propositions). To save ink we can refer to these as F-propositions and R-propositions. Gaskin is committed to both types of proposition and maintains that one cannot countenance one type without countenancing the other. (115) There are true and false tokens of both types. It follows that there are false propositions at the level of reference. Whether this is a problem we will consider below in section 5.
A principle of compositionality holds for sentences, F-propositions, and R-propositions: they all have subsentential/subpropositional components. (66) If sentences and propositions were absolutely simple entities, the unity problem could not arise. But it arises for Gaskin in an acute form not only because sentences and propositions have constituents but also because he cannot accept a Frege-style solution in terms of the unsaturatedness of concept-expressions and concepts given his contention that all subsentential/subpropositional components are unsaturated. It is a central claim of Gaskin’s book that Frege’s context principle implies "that there is no warrant for Frege’s differential treatment of names and concept-expressions." (191) According to the context principle, words have meaning only in the context of a sentence: there is nothing more to the meaning of a word than its contribution to the meaning of a sentence in which it occurs. (188-189) Sentences are conceptually prior to words in that "words are a theoretical abstraction from sentences . . . ." (189) If this is right, then names and concept-expressions are on a par in respect of saturatedness/unsaturatedness. Either both are saturated or both are unsaturated. Gaskin maintains that both names and concept-expressions are unsaturated. Both are such by contrast with the saturatedness of the sentence. Sentences are saturated in that they, unlike their subsentential parts, can be used to say something either true or false. (192) Given the unsaturatedness of all subsentential and subpropositional components, Gaskin cannot avail himself of a Frege-style solution to the unity problem. He cannot say that names (objects) complete or saturate incomplete concept-expressions (concepts) and in so doing achieve the sort of unity that enables a sentence (proposition) to be either true or false. If Gaskin is right, Frege’s own context principle precludes Frege’s solution to the unity problem. Frege’s solution requires that names and their referents be complete while concept-words and their referents be incomplete; this asymmetry is supposed to cement together names (objects) and concept-words (concepts) directly without the services of a mediating link and thus without igniting an instantiation regress. Gaskin’s point, however, and a plausible one it is, is that Frege’s solution is ruled out by a rigorous adherence to Frege’s own context principle.
What is Gaskin’s solution to the unity problem as he understands it? The problem arises because a declarative sentence is not a mere list of its constituent words, and a proposition is not a mere aggregate of its components. A solution to the unity problem must take the form of a specification of what distinguishes a mere list from a sentence, and a mere aggregate from a proposition. One cannot look to syntactics (19) nor to pragmatics (390) for an answer. "What makes a bunch of words a sentence rather than a mere list . . . is that its semantic analysis generates the instantiation regress." (391) That is Gaskin’s solution in a nutshell. In the case of the proposition, Gaskin’s proposal is similar: what makes a number of word-meanings a proposition rather than a mere aggregate is that its semantic analysis generates the non-vicious instantiation regress. Gaskin’s central thesis is that Bradley’s regress is "the metaphysical ground of the unity of the proposition." (345) Bradley’s regress, far from being vicious, is precisely that which "guarantees our ability to say anything at all." (345, emphasis in original) The regress is indeed "a necessary condition of the constitution of language." (419)
2. Why is There a Unity Problem Given Rigorous Adherence to the Context Principle?
For Gaskin, then, all semantically significant subsentential lexical components are on a par as unsaturated. But they are also all on a par as names referring to (unsaturated) objects. For Gaskin, "the whole point of the context principle is that it gives us reason to say that all words are names . . . ." (197) Sentences are therefore composed of names. The unity problem Gaskin thinks he faces is thus of the "starkest form": what is the difference between a declarative sentence and a mere list of names? (280). There is, however, something puzzling about Gaskin’s formulation of the unity problem. For Gaskin takes it to be a consequence of the context principle that "words are a theoretical abstraction from sentences . . . ." (189) But if so, why should there be any problem about how words form sentences? If sentences are primary, and taken as given, then unity is not something to be achieved or effected by a combining of pre-given subsentential parts, but something presupposed. Why then is there a problem about unity? If the context principle is interpreted as Gaskin interprets it, in a manner to show that all subsentential components are on a par as unsaturated, then it would seem that the context principle so interpreted would also show that there is no problem about unity. For if words and their referents are abstractions from sentences and propositions respectively, then there is no problem about how words and their referents combine to form sentences and propositions respectively, for the simple reason that, had no such unproblematic combinations been available, then there would have been no sentential/propositional unities from which to abstract words and their meanings in the first place.
Nevertheless, Gaskin, well aware of the objection just raised, persists in thinking that there is a unity problem: "what distinguishes a proposition – an entity composed in the right way of conceptualized objects – from a mere aggregate of such objects? The sheer fact that all objects are unsaturated, and that they are, in appropriate combinations, of the right shapes to form propositions is not of itself enough to effect their being fitted together into propositions." (314) But why not, given that words are abstractions from sentences and their meanings abstractions from propositions? Why is there need for an effecting of unity when sentential and propositional components are abstractions from pre-given propositional unities? Talk of effecting unity presupposes that words and their referents are antecedently available as semantic building-blocks; but then words and their referents cannot be viewed as "theoretical abstractions."(189) Could Gaskin be waffling between two different models of the relation between a proposition and its components?
3. Two Construals of the Unity Problem
Let us now delve a bit deeper into the tension just uncovered. Everyone will agree that there cannot be a problem about the unity of a proposition if propositions are simple: they must be composite entities composed of components. In the simple monadic case, the proposition that a is F has at least two components, a and F-ness. But which is ontologically prior, the proposition or its components? And what is the nature of the composition: is it ‘real’ or ‘conceptual’? Borrowing some terminology from Gaskin, should we think of the components as "antecedently available" or as "ex post facto parts abstracted from wholes"? (408) Depending on how we answer this question, we get two different views of how a proposition is related to its components and two different construals of the unity problem. To say that a and F-ness are antecedently available is to say that they can exist in reality without forming the proposition that a is F. But if a and F-ness are ex post facto parts abstracted from the whole that is the proposition, then a and F-ness cannot exist in reality without forming the proposition that a is F. The exact formulation of the unity problem depends on whether one thinks it possible for there to be such mere aggregates as a + F-ness in reality as opposed to merely in our thought. I take the ‘ontological’ view that such aggregates can and do exist in reality apart from our thought, while Gaskin takes the ‘conceptual’ view that they exist only as abstracta in our thinking. So, for Gaskin, the unity problem is "the distinctively theoretical question what differentiates a proposition from a mere aggregate – what constitutes the moment of propositional unity when we move in thought from a mere aggregate to a proposition." (386, emphasis added) Gaskin goes on to say that "from a conceptual point of view," the propositional unity is not simply given. (386) This is because one can make a conceptual distinction between propositional constituents and the "moment" of their unity. To which my response will be: indeed, but only because the propositional unity is taken to be ontologically given.
For Gaskin, however, there is no distinction between propositional components and their moment of unity in reality: propositional unity is given ontologically as a sort of brute fact. For me, however, it is not: propositional unity at the level of reference requires an explanation, and it had better not be circular. So for me, the unity problem is the problem of locating the ontological, not merely conceptual, ground of the difference between the mere aggregate a + F-ness and the corresponding Russellian proposition. This ontological ground is an entity that explains the peculiar togetherness of a and F-ness, the togetherness that enables the resulting proposition to be either true or false.
I am not assuming for the purposes of this criticism that there is what Gaskin says there isn’t, namely, "pre-propositional reality" (386) if this is taken to mean that there are objects bare of properties, or first-level properties that are not instantiated. But I am assuming that there is "pre-propositional reality" (386) if this is taken to mean that a and F-ness can exist in reality at the level of reference without a instantiating F-ness. (If an object cannot exist without instantiating some property, it doesn’t follow that it must instantiate F-ness; and if a first-level property cannot exist without being instantiated by some object, it does not follow that it must be instantiated by a.) Gaskin, however, by conceiving of the parts of a proposition as "ex post facto parts abstracted from wholes" (408) is in effect assuming that a and F-ness cannot exist in reality without a instantiating F-ness. But then the unity problem disappears. The problem is to specify what more there is to a’s being F than a + F-ness. What makes the problem difficult is that the ‘more’ cannot be a special unifying ingredient internal to a’s being F. The problem as I formulate it disappears on Gaskin’s assumption that a and F-ness cannot exist except as tied together in a’s being F.
From my point of view, then, Gaskin’s account of the unity of the proposition is no account at all inasmuch as he evades the real problem by simply helping himself to the proposition and its unity. To put it another way, his semantic explanation of propositional unity is circular: he takes unified propositions as ontologically given, as ontological brute facts, and then uses them to explain the unity of propositions via the benign external instantiation regress. The problem, once again, is to specify what makes a proposition more than the mere aggregate of its constituents, what makes it be a unity capable of being either true or false, when neither the aggregate nor any member of the aggregate is capable of being true or false. Using some terminology that Gaskin himself employs, we want to know what is the "metaphysical ground of the unity of the proposition." (345) Gaskin’s answer is that the benign instantiation regress is the metaphysical ground we seek. (345) Let our example be the proposition expressed by ‘Socrates is wise,’ symbolized by ‘Ws.’ Let ‘E2,’ ‘E3,’ ‘En’ symbolize dyadic, triadic, and n-adic exemplification. Now consider the (admittedly benign) infinite series of propositions . . . En+1Ws, EnWs, . . . , E3Ws, E2Ws, Ws. If a metaphysical ground is an ontological ground, then the metaphysical ground of the unity of the base proposition Ws cannot be E2Ws; nor can E3Ws be the ground of E2Ws, and so on. This is because the infinite series cannot exist unless the base proposition is already constituted as a unitary structure capable of being true or false. Gaskin's explanation, in other words, is circular: his explanation of the unity of Ws presupposes the unity of Ws. And the same goes for every proposition 'to the left' of Ws in the regress as above depicted: in every case, the explanation of EnWs in terms of En+1Ws is circular in that it presupposes that EnWs is on hand as a full-fledged unity.
I grant that Gaskin's Bradley-type regress arises and must arise, that it is actually (not potentially) infinite, and that it is benign. I also admit that the regress furnishes us with a criterion, in the epistemological sense of a test, for distinguishing between a proposition and its corresponding mere aggregate and a declarative sentence and its corresponding mere list. The criterion or test is that, if the semantic analysis of the proposition/sentence gives rise to Bradley's regress, then it is a proposition/sentence, while if it does not, then it is an aggregate/list. What I deny is that the regress has any explanatory value as regards the metaphysical (ontological) question of the unity of the proposition. Thus, although the regress is benign, it does not do an ontologically explanatory job. It does not do such an explanatory job because, as an external regress between propositions, as opposed to an internal regress within a proposition, it presupposes the very datum that needs explaining, namely, the unity of the base proposition, Ws in our example.
My point, then, is that Gaskin has not furnished us with the metaphysical (ontological) ground of the unity of the proposition. He has instead furnished us with a criterion or test for distinguishing sentences from lists and propositions from aggregates. But this criterion or test cannot be what makes or constitutes a proposition a unity, because, in order for the test to be applicable, the proposition must already be on hand as a full-fledged unity.
Gaskin and I are disagreeing, not about the solution to the unity problem, but about its exact nature. If the problem were merely to furnish a criterion or test for telling apart propositions/sentences and their corresponding aggregates/lists, then I would say that Gaskin has solved the problem. After all, as Gaskin says, the elements of a mere aggregate are "radically unrelated to one another." (415) This being the case, no mere aggregate can give rise to a Bradley-type regress. But the unity problem as I see it lies deeper. There are two quite different unity questions here, the first properly metaphysical (ontological), the second epistemological-semantic:
Q1. What makes the antecedently available sub-propositional constituents of a proposition into a unity capable of being true or false? What unites object and concept in a monadic proposition? A proposition is more than its constituents as their unity: what is the ontological ground of this unity?
Q2. What distinguishes pre-given propositions from mere aggregates of their components when the latter are taken to be ex post facto abstractions from pre-given propositional wholes?
I believe that Gaskin has provided an answer to (Q2). The answer is that propositions but not aggregates give rise to an instantiation regress upon semantic analysis. But in answering (Q2) he has not thereby answered (Q1). Now it would be churlish and absurd for me to criticize Gaskin for failing to answer a question that he does not attempt to answer or one he considers insusceptible of an answer. But I suggest that the interest and importance of (Q2) derives from its conflation with (Q1). For, as argued in the preceding section, a rigorous adherence to the context principle would seem to issue in the dissolution, rather than solution, of the (Q2) unity problem. Gaskin’s claim that "The sheer fact that all objects are unsaturated, and that they are, in appropriate combinations, of the right shapes to form propositions is not of itself enough to effect their being fitted together into propositions" (314) suggests that he remains under the spell of (Q1).
4. Truth-Bearers but No Truth-Makers at the Level of Reference?
Gaskin and I conceive the unity problem differently. Let us now try to understand what is really at the bottom of our disagreement. As I see it, the fundamental unity problem concerns the unity of truth-making facts, where a fact is not, as Gaskin maintains, a true proposition, but a truth-maker of a true proposition. If we think of propositions as linguistic entities, as Gaskin does, then truth-makers are extralinguistic as the ontological grounds of (some) truths. Propositions on any scheme are truth-bearers and thus either truth or false. Truth-makers, by contrast, are all ‘true,’ or to put the point precisely, they are all such as to obtain. A truth-maker that does not obtain does not exist. So the question arises as to what constitutes the difference between a truth-maker and its constituents. What enables a truth-maker to make true a true proposition when the mere aggregate of its constituents lacks this power? In the case of a truth-bearer, one can say that its unity is just its truth-or-falsity; but one cannot say that the unity of a truth-maker is just its obtaining-or-nonobtaining, and this for the reason that nonobtaining truth-makers do not exist. Since for Gaskin the unity of a proposition is just its truth-or-falsity (18), and since there is no "pre-propositional reality" (386), and hence no extralinguistic truth-makers (116), there simply is no fundamental unity problem as I have described it deeper than the problem of the unity of the proposition.
Truth-maker theory has its difficulties, and so does the notion that there is an ontology of truth deeper than the semantics of ‘true,’ but Gaskin’s linguistic idealism, according to which, and in supplementation of a famous Wittgensteinian dictum, "The whole truth is that the world comprises both everything that is the case and everything that is not the case," (118) also has its difficulties. One of them is the idea that there are false Russellian propositions.
5. The Incoherence of False Russellian Propositions
As noted above, Gaskin maintains that there are true and false propositions at the level of reference. They serve as the referents of true and false declarative sentences respectively. These Russellian propositions are abstract objects that "contain as literal constituents the worldly entities . . . introduced by the semantically significant parts of those sentences" that refer to them. (57) Thus ‘Venus is uninhabited’ (my example) refers to a true R-proposition that contains Venus itself, that massive chunk of physical reality, along with the property of being uninhabited. "Venus is less massive than Mercury’ (my example) refers to the false R-proposition having as literal constituents Venus, Mercury, and the relation less massive than. One might wonder how abstracta could have concreta among their constituents. Gaskin’s counters this worry with the observation that most philosophers find it unproblematic that sets, which are abstract objects, often include concrete items. So why balk at propositional abstracta harboring concreta? (59) This is a plausible, if not absolutely compelling, response to the worry.
More troublesome than the abstractness of Gaskin’s propositions in re (as he also calls them) is the claim that some of them are false. It is easy to see how there can be false propositions, false Fregean Gedanken, at the level of sense. A false Fregean proposition is just the sense of a false declarative sentence. But it is difficult to see how there could be false propositions at the level of reference. Sentences and the Fregean Thoughts they express are representational in a straightforward sense: they represent the disposition of the world of concreta. They either represent it as it is or as it is not. Thus the Fregean proposition that Socrates is wise represents Socrates as wise. It is a unity of sense that says that Socrates is wise whether or not he actually is wise. A proposition at the level of reference, however, is arguably not representative of anything: such propositions ‘on the ground’ are what get represented without representing anything in their turn. They just are dispositions or arrangements of concreta. From this it would seem to follow that all Russellian propositions are true, albeit in a sense of ‘true’ different from though analogous to the sense of ‘true’ as a predicate of propositions. On one picture, the correspondence picture that Gaskin rejects, (115) true Fregean propositions are ‘verified’ (made true) by truth-making facts all of which obtain. On this sort of scheme, a Russellian proposition is identified with a truth-making fact or state of affairs in the realm of reference, an entity that does not in its turn represent anything but serves instead as a terminus of representation. On this way of thinking, a true R-proposition is just an actual state of affairs, and a false R-proposition is no proposition or state of affairs at all: it is just an aggregate of would-be state of affairs constituents.
Gaskin parries this critical thrust as follows. "The sense in which false [Russellian] propositions lack the actuality of actual states of affairs is just this: while the components of false propositions are real enough, those components are not, as it happens, related in the way in which, in the proposition in question, they are said to be related." (111) This is puzzling since it is not clear in what sense an R-proposition, whether true or false, says anything. The puzzle is exacerbated when the R-proposition is false. About what does a false R-proposition say something false?
Gaskin tells us that that "False propositions at the level of reference differ from true ones at that level neither in point of unity nor in their ontological status: the only difference is that their constituents are being said (by the proposition) to be related in a way in which they are not related." (112) So the R-proposition denoted by ‘Venus is inhabited’ says something false about its own constituents, namely, Venus and being inhabited. It says that they are related in a way in which they are not related. And which way is that? By way of instantiation. Thus the false R-proposition says about its subject-constituent, Venus, that it instantiates its property-constituent, being uninhabited, when it does not. But in that case the R-proposition lacks unity, contrary to what Gaskin has claimed about false R-propositions. We just heard him say that false propositions at the level of reference do not differ from true ones at that level in point of unity. But without the subject-constituent instantiating the property-constituent, there is no unity, and thus no proposition at all, but only a mere aggregate.
Or consider the false Russellian proposition expressed by ‘Russell = Whitehead.’ It says something false about its constituents, Russell and Whitehead. It says that they are related in a way in which they are not related. And which way is that? By way of identity. Thus the R-proposition says of Russell that he is identical to Whitehead when he is not. But in that case the R-proposition lacks unity.
As I see it, the idea of a false R-proposition issues in inconsistency. As a proposition, an R-proposition must say something about something. But as a Russellian proposition, it cannot say something about something wholly distinct from itself. So an R-proposition must say something about its own constituents. But if what it says is false, then the constituents are not related in the way they would have to be related to form a proposition. Now the only way in which Socrates and foolishness can be related to form the false R-proposition that Socrates is foolish is by Socrates’ instantiating foolishness. Without instantiation, the two components form a mere aggregate. But if Socrates instantiates foolishness at the non-representational level of reference, then the R-proposition that Socrates is foolish is true. This implies that this particular R-proposition must be both true and false. It must be true to be a Russellian proposition, but it must be false to be a false Russellian proposition. Since this is a contradiction, we ought to conclude that there are no false R-propositions.
Lest the reader think that I am begging the question against Gaskin by assuming what he denies, namely, that there are truth-making facts, let me make it as clear as I can that Gaskin’s position is internally contradictory. He is committed to the following:
1. R-propositions are propositions, truth-bearers, linguistic entities; they are not extralinguistic truth-makers: they say something about something and can be either true or false.
2. Since R-propositions are ground-level propositions in re, an R-proposition cannot say anything about something wholly distinct from itself.
3. The false R-proposition that Socrates is foolish, being both Russellian and false, must say about Socrates himself and foolishness itself that they are related in a way in which they are not related in reality. (From 2, 3)
4. The constituents of our sample R-proposition are not related in reality by instantiation: Socrates does not instantiate foolishness. If they were so related, the proposition would be true.
5 For a proposition to be a proposition, to say something and be capable of being true or false, it must be a unity and not a mere aggregate of constituents. Hence in our sample false R-proposition Socrates must instantiate foolishness.
6. (4) and (5) cannot both be true. Socrates must instantiate foolishness for the R-proposition to be a proposition (as opposed to a mere aggregate), but it cannot instantiate foolishness given that it is a false R-proposition.
My point is that Gaskin’s theory of propositions is logically inconsistent. This is an ‘immanent’ criticism: I am not coming at his theory with the ‘transcendent’ assumption that there are truth-makers. To avoid his contradiction while retaining the rest of his premises, one must reject (1). One could do that by holding that so-called Russellian propositions are in reality truth-making facts or states of affairs, and that all of them obtain. But in doing so one would abandon linguistic idealism.
6. Concluding Questions About Linguistic Idealism
As I understand it, Gaskin’s linguistic idealism is the thesis that the world is "the totality of (true and false) propositions at the level of reference" (420) in conjunction with the thesis that there is no "pre-propositional reality" (386) in the sense that there are no subpropositional ontological building blocks out of which propositions are constructed. So at bottom everything at the level of reference is a true or false Russellian proposition. Since Bradley’s regress is the "metaphysical ground" of the unity of every proposition true or false (345), the famous regress "sustains the very existence of the world . . . ."(420) So the regress is not merely benign, it is positively virtuous, so much so that one is tempted to speak of the apotheosis of Bradley’s regress in Gaskin’s system.
Gaskin’s linguistic idealism implies that at the level of reference every form of unity is propositional unity: "if the elements of an aggregate at the level of reference are configured with (related to) one another in any way at all . . . then the elements of such an aggregate are indeed propositionally unified." (415, emphasis added) Thus there are no forms of unity below the level of propositions nor above their level: there are no extralinguistic truth-making facts and no irreducible unities of consciousness. This makes for an interesting contrast of Gaskin’s linguistic idealism with my version of onto-theological idealism as presented in A Paradigm Theory of Existence: Onto-Theology Vindicated (Kluwer 2002). We both introduce an item that "sustains the very existence of the world" (420) and we both deem this item to be infinitistic. Call it the Unifier. But whereas for me the Unifier is an infinite mind, for Gaskin the Unifier is an infinite regress. And what the Unifier unifies is different on the two schemes. Gaskin’s Unifier unifies propositions while mine unifies truth-making facts or states of affairs. And while his world is a world of true and false Russellian propositions, mine is a world of truth-making facts.
To pursue this comparison further would require a separate article. Gaskin generously devotes five pages of his book to "Vallicella’s Onto-theology," (370-374) to which the reader may refer for Gaskin’s side of the comparison. Let me just sketch in conclusion three reasons why I do not accept Gaskin’s position. The first I have sufficiently elaborated in section 5 supra: Gaskin’s system requires both true and false R-propositions, but the very idea of a false R-proposition is incoherent. The second is that the existence of the world cannot be sustained by any such frail reed as language. The world for Gaskin is just the totality of true and false propositions, and propositions are linguistic entities. (115) This would seem to make the existence of the world dependent on the existence of language. But in what sense?
We are told that the "whole point of the context principle [is that] in the beginning was the sentence . . ." (242) and that words, as "theoretical abstractions" from sentences (189) together with their senses and referents, as abstractions from Fregean and Russellian propositions respectively, are theoretical posits introduced to explain sentences and how they mean. We are also told that it is a consequence of the context principle that all words are names (197). But the objects denoted by these names are theoretical posits: "The existence of objects is thus fallout from the existence of language. . . ." (242) It is not because of objects in the world that our sentences are true or false; it is because our (declarative) sentences are true or false that the expressions they contain have worldly referents. (241-242) The ultimate point of language is not to say things about the world; "the ultimate point of the world is to model the meaningfulness of language." (242)
Gaskin is not saying that we make the world by our talking about it. The language-world link is forged at the level of whole sentences, not at the subsentential level, and "the truth-values of sentences are not in general up to us . . . ." (44) Although the truth-values of sentences do not in general depend on us, they don’t depend on the world either: the being-true of true sentences lacks a metaphysical ground. (116) So while we should not impute to Gaskin what I will call extreme linguistic idealism, his idealism is not free of taint inasmuch as it makes the existence of worldly objects dependent on, "fallout from," (242) the existence of language. Gaskin maintains not only that that there is "no pre-linguistic access to reality," but also that "reality is itself constituted by our linguistic access to it. . . ." (260) This certainly sounds like an objectionably strong form of idealism.
There are passages, however, which convey the impression of innocuousness and which suggest what could be called anemic linguistic idealism. This is the view that the world, as the totality of true and false Russellian propositions, is dependent, not on any particular language or family of languages, but merely on the possibility of there being some language or other: "an object’s existence is dependent only on the the possibility of its being referred to in some language or other." (44) How anemic this is depends on the weakness of the sense of ‘possibility’ in play. Suppose we agree that a transcendental realist is one who "asserts the independence of objects and properties from language as such." (117) Such a realist could, consistently with his realism, maintain the logical possibility of an object’s being referred to in some language or other. What the transcendental realist maintains is that objects exist and have (some) of the properties they have whether or not any language exists. This does not require him to say that it is not logically possible that they be objects of discourse within some language. A better indication of Gaskin’s actual view is given by his claim that "worldly entities . . . are essentially objects of discourse." (44) But if so, then worldly entities are essentially objects of discourse from within some language or other. But if worldly entities are essentially such objects, then it is not possible that they exist unless some language exists. And that is just preposterous. Before there were any languages, there was the Earth and its moon and the solar system, etc. It is surely not essential to the existence or nature of these entities that they be objects of discourse. If the world needs sustaining, and it does, and if the sustaining must take an idealistic form, which is plausible, then it cannot be language which does the sustaining especially if one considers that language is one of the things in the world that needs sustaining. Something more robust than language is needed to sustain the world. In sum, it is not entirely clear what Gaskin’s linguistic idealism comes to, and to the extent that it is clear, it strikes this critic as untenable.
My third reason for not accepting Gaskin’s system is that it implies that everything at the level of reference which is composite as opposed to simple exhibits propositional unity, which is to say that every composite entity at the level of reference is a Russellian proposition. Now consider my present unity of consciousness. I see a bird on a branch and I hear its chirping. The contents of my consciousness at a time form a synchronic unity and therefore on Gaskin’s scheme constitute a Russellian proposition. But surely a unity of consciousness is not an R-proposition. For one thing, such propositions are abstract (57), and it is difficult to see how my present buzzing, blooming unity of consciousness is anything abstract . It is a living, particular, concrete unity of experience. Nor is it easy to see how my present unity of consciousness could be the meaning of a declarative sentence which is what it would have to be to be an R-proposition. Meaning is meaning for a mind, which implies that (i) minds are logically and ontologically prior to propositions, and (ii) the unity of consciousness is logically and ontologically prior to the unity of propositions. If so, not all unity is propositional unity. Perhaps on another occasion I will fill in these sketchy remarks.