Quine's Critique of Analyticity

          The fourth of the five milestones that Quine finds in the history of empiricism is the abandonment of analyticity in the traditional analytic-synthetic dichotomy.  He calls his exclusive acceptance of synthetic statements "methodo­logical monism.”  The rejection of analyticity is one of the earliest theses in Quine's philosophy of lan­guage.  In his Dear Carnap, Dear Van Creath reports that when Quine had first met Carnap in March 1933, Quine was reading the manuscript for Carnap’s Logical Syntax as Carnap’s wife was typing it.  Creath notes that a brief shorthand note later found among Carnap’s archived papers reveals that Quine had asked whether or not the difference between the analytic axioms of arithmetic and the synthetic empirical claims about physical bodies is merely a difference of degree, which reflects our relative willingness to abandon the various beliefs under consideration.  Quine’s first published statement of the rejection of the traditional analytic-synthetic distinction is in his "Truth by Convention" (1936) originally in Philosophical Essays for A.N. Whitehead, and later reprinted in his Ways of Paradox.  Analytic statements are those that are true by linguistic convention, and they include the propositions of logic and mathematics.  Essentially his argument in this paper is based on the rejection of an infinite regress; he argues that some logic is needed and is presupposed to develop logic.   Thus he asks whether or not it makes any sense to say that the truths of logic and mathematics are destined to be maintained independently of our observation of the world, so that truth by convention may apply.
          Fifteen years later Quine’s critique of analyticity took a different tack in "Two Dogmas", where he formulated the Duhem-Quine thesis of semantical wholism, and attacked linguistic synonymy upon which analyticity is based.  The statement "No bachelor is married" is made analytic by substitution of synonyms "bachelor" and "unmarried man" in the statement "No unmarried man is married", because the latter statement is true in all interpretations of its nonlogical or descriptive terms.  Quine notes that Carnap explained analyticity by appeal to state descriptions; a statement is analytic if it is true in all state descriptions.  Quine says that appeal to state descriptions works only if the atomic statements of the language are mutually independent, i.e. if the language has no extralogical synonym pairs such as "bachelor" and "unmarried man.”  Thus on Quine's thesis, Carnap's criterion for analyticity in terms of state des­criptions is a reconstruction at best of logical truth, not of analyticity.  Quine argues that all instances of synonymy except those occurring in purely stipulative definitions introducing notational abbreviations are based on observed synonymy occurring in natural language.  These include synonymies occurring in reduction sentences, analytic sentences and Carnap's semantical rules; and they all depend on the thesis contrary to Duhem's thesis, that it is possible to determine the truth or falsehood of sentences in isolation from one another.  Invoking Duhem's thesis Quine rejects the distinc­tion between a factual component and a linguistic component in the truth of any individual statement, which is the basis for the analytic-synthetic distinction.
          Shortly after writing "Two Dogmas" Quine wrote "Carnap and Logical Truth" (1954) in Philosophy of Rudolf Carnap (1963).  This critical essay's most distinctive character­istic relative to Quine's prior essays is its treatment of the effects of linguistic and scientific change on analyti­city and logical truth.  Carnap's interest in philosophy was originally inspired by Einstein's use of non-Euclidian geo­metry and by Hilbert's formalistic approach to mathematics.  Quine says that the initial tendencies to treat geometries as true by convention together with the tendency toward formalization were extended to mathematical systems generally.  But Quine maintains that formalist mathematics has been "corrupted" by supposing that postulates are true by conven­tion, and he rejects the idea of semantically uninterpreted postulates.  Quine treats the subject of postulates in a manner similar to his earlier treatment of definitions in "Two Dogmas.”  He distinguishes two types of postulates: "legislative" and "discursive.”  The former type is a stipulative definition that merely introduces previously unused notation, and it initiates truth by convention.  Dis­cursive postulation on the other hand is a selection from a pre-existing body of truths, of certain ones for use as a basis from which to derive others initially either known or unknown.  Most notably what discursive postulation fixes is not truth, but only some particular ordering of the truth.  All postulation may be said to be conventional, but only legislative postulation admits to truth by convention.  The importance of the distinction, however, is that it refers to an act and not to any enduring consequences.  The conventionality in postulation is a passing trait, which is significant at the moving frontier of science, but which is useless in classifying the sentences behind the lines.  Conventionality is a trait of events and not of sentences.  And if legislative postulates are subsequently singled out in some later exposition, they have the status of discursive postulates in the subsequent exposition.  The artificiality of legislative truth does not linger as a localized quality, but suffuses with the corpus and becomes integral with it.   Quine does not explicitly reference Duhem in this context, but Duhem's wholism is clearly operative.  Quine says that legislative postulation occurs continually in the theoretical hypotheses of natural science.  The justification of any theoretical hypothesis can at the time of hypothesizing consist in no more than the elegance or convenience which the hypothesis brings to the containing body of laws and data.  There is indirect but eventual confrontation with empirical data, but this can be remote.  Furthermore, some such remote confirmation with experience may be claimed even for pure mathematics and logic.  A self-contained theory that can be checked with experience includes not only its various theoretical hypotheses of so-called natural science, but also such portions of logic and mathematics as it makes use of.  There is no line to be drawn between hypotheses that confer truth by convention and hypotheses that do not; logic and mathematics are not qualitatively different from the rest of science.
          Quine elaborates by illustration: Suppose a scientist introduces a new term for a certain substance or force by an act of legislative definition or postulation.  Progressing, he then evolves hypotheses regarding further traits of the named substance or force.  And then further progressing he identifies this substance or force with one named by a complex term built up of other portions of his scientific voca­bulary.  This new identity will figure in the ensuing devel­opments quite on a par with the identity which first came by the act of legislative definition, or on a par with the law which first came by the act of legislative postulation.  And revision in the course of further progress can touch any of these affirmations equally.  Quine says that scientists proceeding in this way are not slurring over any meaningful distinction.  Legislative acts occur routinely. Carnap's dichotomy between analytic and synthetic, between truth by meaning postulate and truth by force of nature, has no clear meaning, even as a methodological ideal.  The fabric of our sentences, our web of beliefs as Quine calls them later, develops and changes through more or less arbitrary and deliberate revisions and additions of our own, more or less directly occasioned by the continuing stimulation of our sense organs.
          Carnap replies at the end of the volume in which Quine's critique was published.  He emphasizes that his explication of "analytic" has always been for a formalized language, one for which explicit semantical rules are specified and that lead to the concept of truth.  He rejects Quine's demand that semantical concepts such as analyticity and synonymy must also be explicated pragmatically by an empirical criterion in behavioristic terms applicable to natural language.  He therefore maintains that Quine's objections are not directed against his semantical expli­cata, and that A-truth is not objectionable.  Carnap then turns to Quine's critique of analyticity in situations where there is a change in artificial language, from L(n) to L(n+1).  Firstly Carnap agrees with much of what Quine says in "Two Dogmas", where Quine sets forth his neo-Duhemist wholistic thesis.  Carnap agrees that a scien­tist who discovers a conflict between his observations and his theory and who must therefore make a readjustment somewhere in the total system of science, has much latitude with respect to the places where a change is to be made.  Remarkably Carnap also agrees that in this procedure of readjustment, no statement is immune to revision, not even statements of logic or mathematics.  But Carnap rejects Quine's characterization of an analytic statement as one held true come what may.  And Carnap furthermore denies that a change in language invalidates the analytic-synthetic dis­tinction.  In defense of analyticity Carnap distinguishes two types of linguistic change.  The first type is a change of language from L(n) to L(n+1).  He says that this type constitutes a radical alteration and perhaps a revolution.  It occurs only at certain historically decisive points in the development of science.  The second type is a mere change in or an addition of a truth-value ascribed to an indeterminate statement.  An indeterminate statement is one having a truth-value that is not fixed by the rules of the language, i.e. by postulation of logic, mathematics, or perhaps physics.  This second type of change occurs "every minute" according to Carnap.  He says that his concept of analyticity has nothing to do with the first type of transition; his concept of analyticity refers only to some given language, L(n).  The truth of a sentence, S, in L(n) is based on meanings in L(n) of the terms occurring in S.  In L(n) analytic sentences cannot change their truth-value, and furthermore neither can the synthetic postulates of physics and their logical consequences.
          Quine's critique of analyticity is principally directed against what Carnap called A-truth, which is truth based on the semantics of the descriptive vocabulary in the sentence, a lexical basis.  As a symbolic logician Quine continues to rely on logical truth, on the kind of sentence that Carnap calls L-truth, but his reasons are different than Carnap's.  In "The Ground of Logical Truth", the eighth chapter in his Philosophy of Logic, Quine admits to an acceptable sense of logical truth, the truth that is evident due to the grammatical structure of the logically true sentence.  But Quine rejects Carnap's doctrine of linguistic truth, the thesis that language alone can make logical truth indepen­dently of the nature of the world.  In view of Carnap's defense of analyticity, it is doubtful that Carnap continued to maintain such a view.  In any event, Quine maintains that the validity of logical truth depends on the relation of grammatical structure to the structure of the real world.  He argues that the distinction between the lexical and the grammatical is variable not only among different languages, but also within the same language.

Quine's Rejection of First Philosophy

          Quine’s taking Whitehead’s comment that logic shapes metaphysical thought beyond logic and making it his general theory of language, has another and even more important implication: Quine’s thesis of ontological relativity.  Thus the last of the five milestones in Quine's history of empiricism is what he calls the abandonment of the goal of a first philosophy.  By first philosophy he means any philo­sophy that is prior to natural science.  Traditionally meta­physics and epistemology are considered to be first philosophy.  Quine calls his position "naturalism.”  The term "naturalism" has meant many different things in the history of philosophy.  A term that Quine does not use is "scientism.”  In "Five Milestones" Quine defines his naturalism as the view that natural science is an inquiry into reality, a fallible and corrigible inquiry, but not answerable to any suprascientific tribunal, and not in need of any justification beyond observation and the hypothetico-deductive method.  This statement by Quine is not merely an affirmation of the autonomy of empirical sci­ence from metaphysics, as may be found in Duhem's philosophy of science.  Quine rejects the view that there is any philo­sophical tribunal for science, by which he means any know­ledge separate from empirical "common sense” that he views to be continuous with science in his wholistic philosophy of language.  Furthermore, Quine maintains that epistemology is an empirical discipline that he assimilates into empirical psychology, which for him is behavioristic psychology.  He describes the scientific epistemologist as asking how animals, presumably human, can have managed to have arrived at science from the limited information from surface stimula­tions, and as pursuing this inquiry to yield an account that pertains to the learning of language and the neurology of perception.
          Quine gives two reasons for his naturalism by which he rejects all first philosophy.  One reason is what he calls an "unregenerate" realism, the robust state of mind of the natural scientist who has never felt any qualms beyond the negotiable uncertainties internal to his science.  He expresses his realism even more emphatically in his "Scope and Language of Science" (1954) reprinted in Ways of Para­dox.  There he states that we cannot significantly question the reality of the external world or deny that there is evidence of external objects in the testimony of our senses.  For to do so is to dissociate the terms "reality" and "evi­dence" from the very application which originally did most to invest these terms with whatever intelligibility they may have for us.  He maintains that the notion of reality inde­pendent of language is derived from our earliest impres­sions, and then carried over into science as a matter of course.  The second reason for Quine's realism is what he calls the despair of being able to define theoretical terms gener­ally in terms of phenomena even by contextual definitions.  This is a rejection of the Logical Positivist problem for which reductionism of theoretical terms was thought to pro­vide an answer.  On the Positivist philosophy there is no justification for affirming the reality of theoretical enti­ties, unless these terms are firstly established as seman­tically meaningful.  The purported solution is the reduction of theories to observation sentences, which are the source for the semantics and ontology of theories.  Quine rejects the Positivists' problem, because it involves a prior onto­logy or first philosophy consisting in the Positivists' observation language.  In Quine's view Positivism is a kind of metaphysics, Positivists' antimetaphysical rhetoric notwithstanding.
          Fundamental to Quine's second reason for rejecting first philosophy is his thesis of ontological relativity.  This thesis can be found in Quine's literary corpus even before he came to call it "ontological relativity" in the mid-1960's.  In "Two Dogmas" after rejecting the dogma of reductionism, he says that physical objects are conceptually imported into the linguistic system as convenient intermedi­aries, as irreducible posits comparable epistemologically to the gods of Homer.  What he calls the "myth" of physical objects is epistemologically superior to others including the gods of Homer, in that it has proved to be more effica­cious than other myths as a device for working a manageable structure into the flux of experience.  Microphysical enti­ties are posited to make the laws of macroscopic objects and ultimately to make the laws of experience more manageable.  Science is a continuation of common sense, and it continues the commonsense expedient of swelling ontology to simplify theory.  Shortly later in "Posits and Reality" (1955) Quine says that if we have evidence for the existence of bodies of common sense, we have it only in the way in which we may be said to have evidence for the existence of molecules.  All science is empirically underdetermined, and the only difference between positing microphysical and macrophysical enti­ties is that the theories describing the former are more underdetermined.  In this context Quine is using the term "underdetermined" in same sense as he used it in "Two Dog­mas" to express his neo-Duhemist wholistic view of language.
          The thesis of ontological relativity is also prefigured in Word and Object.  Just as Carnap recognized extensional vagueness, Quine recognized referential indeterminacy, which he calls referential "inscrutability.”  Inscrutability of reference is due to the semantic indeterminacy of direct ostension.  This indeterminacy is encountered when the field linguist attempts to translate a previously unknown language, but it also occurs more generally in all language, and is not distinctive of the translation situa­tion.  The context-dependence of semantics makes reference and ontology completely system-determined in the linguistic context that determines the semantics of a discourse including notably the context constituted by a scientific theory.  In chapter six of Word and Object Quine says that everything to which we concede existence is a posit from the standpoint of the theory-building process, and is simultaneously real from the standpoint of the theory that is built.  His phrase “ontological relativity” itself is set forth in "Ontological Relativity" (1968) in Ontological Relativity.  Quine uses the phrase explicitly on analogy with Einstein's relativity theory in physics.  He maintains that reference is nonsense except in relation to a coordinate system, where the coordinate system is some background language.  Asking for ontological reference in any more absolute way than by reference to a background language is like asking for absolute position or absolute velocity, rather than for position or velocity relative to a frame of reference.  The ultimate background language to which we take recourse in practice is our mother tongue, in which we take words at face value with their primitively adopted and ultimately inscrutable ontology.  Any subordinate theory must be interpreted by reference to this home language.  Quine opposes his thesis of ontological relativity to Carnap's thesis of the distinction between external ontolo­gical questions and internal factual questions set forth in "Empiricism, Semantics and Ontology.”  In Quine's view there can be nothing like Carnapian external questions which are external to the home language.  In "Carnap's Views on Onto­logy" (1951) reprinted in Ways of Paradox Quine maintains that ontological questions are on a par with questions in natural science.  Within science there is a continuum of gradations from the statements that report observations to those that reflect basic features of quantum theory and relativity theory.  Similarly statements of ontology and even of mathematics and logic form a continuation of this continuum, though these are more remote from observations than the central principles of quantum theory or relativity theory.  Quine says that the differences along this contin­uum are only differences of degree and not differences in kind.
          While the semantical wholism of the Duhem-Quine thesis has received much attention, it is seldom realized that Quine's rejection of all first philosophy is one of its most consequential implications for philosophy of science.  When the Duhem thesis of physical theory is extended to the whole of language, not only is all semantics made context-determined, but also all ontologies described by the semantics are made vulnerable to empirical criticism; there are no longer any privileged or protected ontologies.  Quine's thesis of ontological relativity has the historic and revolutionary effect of excluding all ontological considerations from the criteria for scien­tific criticism.  In his philosophy it is empirical adequacy of scientific theories that decides ontological questions, rather than prior ontological commitments that decide the acceptability of scien­tific theories.  Quine subordinates all questions of onto­logy to the empirical adequacy of the theory affirming the ontological claims in question.  He maintains that the human knower can never do better than to occupy the standpoint of one or another theory, whether the theory purports the existence of either macrophysical or microphysical entities.  All entities are "posits" affirmed by one or another theory, and all are worthy of our patronage just to the extent that the theory positing them is empirically adequate.  However detailed may be the relevant observation language, empirical underdetermination (in Quine's earlier sense) and its conse­quent semantical indeterminacy always admit alternative choices of theory.  And the consequent referential inscruta­bility may admit to as many correspondingly alternative choices of entities.
          Quine's rejection of prior ontological criteria in scientific criticism is also consistent with scientific realism, which gives the tested and nonfalsified explanation the role of defining ontology.  Realism is not established by science; it is a prior prejudice.  But science lets empirical justify the ontological claim that the explanation describes the real world.  This thesis is not only characteristic of the contemporary Pragmatist philosophy, but was also the practice of Galileo, Einstein and Heisenberg.  In developing his theory of relativity Einstein posited relativistic time as real instead of Newton's absolute time, and he rejected Lorentz's relegation of relativistic time to the status of apparent time and Lorentz's retention of New­ton's absolute time as real.  A central thesis of the Copenhagen interpretation, or at least Heisenberg's noninstrumen­talist version, is its realistic claims about the wave-particle dualism and the indeterminacy principle, and Heisenberg referenced Einstein's realism in relativity theory as a pre­cedent.  However, the Copenhagen wave-or-particle dualism thesis cannot be affirmed on the basis of the mathematical equations of the quantum theory, since the mathematical expression has no syntactical categories for referencing entities.  And Heisenberg’s potentia ontology for the indeterminacy relations is also an added ontological claim about entities no less so than deBroglie-Bohm deterministic pilot-wave-and-particle ontology making the indeterminacy relations due to errors of measurement that are in principle correctable.  The ontological claim justified by the empirical adequacy of the tested and nonfalsified mathematically expressed theory is limited to what the theory actually says, and the explanation is otherwise silent about ontology, and awaits further experimental findings.  The prac­tice of letting the empirical adequacy of a theory operate as the criterion for the acceptability of its ontology did not begin with Einstein or Heisenberg.  A historic and well known example is Galileo's realistic interpretation of the Copernican theory, which placed him in conflict with the Aristotelian ontology enforced by the Roman Catholic Papacy.
          This is a distinctively and thoroughly Pragmatist view that separates Quine from both his Positivist and Romanticist predecessors.  Ironically it also separates him from certain other aspects of his own philosophy.  One such aspect is his behavioristic epistemology.  The Romanticists insist upon and the Positivists in­sist against the introduction of "mentalism" in explanations in the social and behavioral sciences.   But on the contemporary Pragmatist philosophy of science, this ontological issue is decided by the empirical adequacy of the behavioral and social science theories.  Different theories in different sciences at different times or even at the same time will admit different ontologies.  Quine's behavior­istic "naturalized" epistemology is actually an exception to his thesis of ontological relativity.
          Another such inconsistent aspect is Quine’s ontological reductionism and his consequent de facto nominalism.  In his “Introduction” to his Dear Carnap, Dear Van Richard Creath states that Quine’s ontological reductionist agenda was due to Quine’s interpreting Carnap’s Logical Syntax in a manner that was nearly wholly unintended by Carnap.  Carnap argued in Logical Syntax that talk which appears to be about possibilities, properties, relations, numbers, etc. can be reconstrued to be talk about sentences, predicates, etc.  Creath says that in Quine’s “Lectures on Carnap”, a prepublication report on the theses of Logical Syntax given to the Society of Fellows at Harvard in 1934, Quine had interpreted Carnap to mean that there are no such metaphysical entities, and that philosophy therefore is syntax as a program of ontological reduction.  Creath states that in fact Carnap actually rejected both the affirmation and the denial of the existence of such metaphysical entities as properties, because Carnap believed at the time that such discourse is metaphysical nonsense.  Later Carnap took a more pragmatic view of such entities as intensions and properties.  But for the duration of his career Quine continued in his ontological reductionist agenda, which apparently resulted from his early misinterpretation of Carnap, notwithstanding Quine’s later formulation of his ontological relativity thesis.  This persistence is inconsistent; ontological relativity renders logical elimination for the purpose of ontological reduction a philosophically pointless exercise, because its acceptance implies the rejection of any and all prior ontological commitments that would motivate the ontological reductionism.  Ontological relativity makes all ontological commitments a posteriori to empirical criticism, and together with the empirical underdetermination of all theories results in ontological pluralism, not reductionism.  But Quine is neither the first nor the last philosopher-king to exercise a sovereign's right of eminent domain in his own philosophy, and exempt his preferred convictions from his own laws.