home | philosophy index | notes | comment

Making Sense of Reference

by David Foss

I. Introduction

Consider the following sentences: “Hamlet slew Polonius.” “Thomas Mann vigorously denied reports that Adrian Leverkühn was in fact Arnold Schoenberg.” “Fidelio is Leonore.” By the mere form of such sentences it is impossible to distinguish them from such sentences as “Booth slew Lincoln,” “Bob Woodward vigorously denied reports that Deep Throat was in fact Alexander Haig,” and “Mark Twain is Samuel Clemens.” In our analysis of these statements, if we follow the approach suggested by Frege or Russell, those of the first set all come up either truth-valueless or false, while those of the latter set is come up true.[1] This is because, without regard to meaningful content, the referents of ‘Hamlet’, ‘Polonius’, ‘Adrian Leverkühn’, ‘Fidelio’ and ‘Leonore’ fail to properly exist (where proper existence is a feature of real things, like Thomas Mann, or the oak tree outside my window), and as a result any assertion containing these terms as primary designators will fail to refer, or pick out, its proper referent (i.e., a truth value). According to such an analysis, when I use a term such as ‘Hamlet’, I fail to designate anything, relying instead upon a wide set of conventional associations to comport my meaning. But it seems that I do speak about something; namely the character named Hamlet (and not merely the character of Hamlet). How are we to understand this aboutness if it is not reference? Reference would seem the natural place to locate such a use of the name ‘Hamlet’ (or any of the others mentioned). How is it that a term is said to successfully refer only when its nominatum is “real”? Unfortunately, the status of a sign’s nominatum as “real” is often presumed, or merely restated as “successful” designation, referring, or naming.

Still, why should our analysis of assertions made in the course of literary criticism (by which I mean to include any proposition which refers in part or whole to a fictional object or event) bother us, or be as pressing an issue for explication as our analysis of the assertions of scientific enquiry? After all, in the prior case we are dealing with the fuzziest regions of cognitive signification, and in the latter, the most clear and real. Assertions in science concern hard objective phenomenon, bodies, and relations. Assertions in literary criticism concern virtual objects, overflowing with interpretive ambiguities and poetic imagery, which at every turn evade dispassionate analysis. If we are to have a reasonably rigorous linguistic theory, it would seem that scientific language (which itself demands a particularly high degree of rigor) should be the model for a semantics of knowledge. A theory of knowledge is, of course, the driving force behind the search for clear rules of propositional truth-valuation. It is the truth-character of scientific speech, which strives to accurately model the world in language, that makes an analysis according to propositional truth-valuation so helpful.

Unfortunately, by exclusively attending to the truth-value of a particular knowledge claim (or more properly, of the propositional content of a particular knowledge claim) we loose our grasp upon the wider significations of the claim. It appears as a sort of Heisenberg-style uncertainty phenomenon, where by grasping hold of the truth of a proposition, we are no longer able to determine precisely, how or why its terms function the way they do. In particular, the insistence that a sentence possess a truth-value in virtue of the proposition it expresses, and this in virtue of the trans-linguistic reach of the proposition in comporting a real-world correspondence between its real-world subject and real-world predicate, obfuscates the semantic and pragmatic devices which map constituent signs, terms, names, expressions, etc., to their referentia, and furthermore, conceals the proprietary devices which distribute the success or failure of the particular claim across its embedded discourse.

This is, admittedly, a controversial claim. Why should a serious attention to detail in terms of the truth-valuation of propositions prevent a clear analysis of other aspects of knowledge claims? No one claims that an analysis of a knowledge claim is exhausted by an analysis of its truth-value. What I am suggesting, however, is that by introducing truth as a primitive characteristic of the propositions expressed by knowledge claims, we subvert our ability to adequately analyze the phenomenon of reference in itself, which we should realize forms the very foundation of the primitive truth definition (whether it be a sort of correspondence, coherence, or more exotic epistemological theory which defines propositional integrity in terms of the relationships of constituent referentia). The effort to analyze the truth conditions of a proposition by a reduction of propositional content into subject and predicate names, which are in turn analyzed in terms of their referentia, presupposes the occurrence of “reference.” Once the analysis is begun, and the study of a particular assertion has isolated the terms of predication, the actual mechanisms by which the terms successfully pick out their referentia have been whittled away. The mechanisms I have in mind are none other than the pragmatics of reference.

Ultimately, I would suggest, within a purely (classical) semantic analysis there is no difference between an assertion which names a fictional character, and one which names a physical or mathematical entity. The reason for this, as I will begin to argue here, is that the meaningful interplay of names can be accounted for without direct appeal to trans-linguistic attachments; and that the success (or failure) of reference can be best understood in terms of behavioral performances in concert with the utterance of a name. The approach will be one suggested by Joseph Camp, Jr., in “Why Attributions of Aboutness Report Soft Facts,”[2] where he works to combine a Sellarsian account of content with a possible worlds account of reference proper (which looks something like Kripke’s analysis of designators). Along the way I will consider whether the approach proposed by Camp is a threat at all to the sort of analysis offered by Frege and Russell. It might be objected, once the theory is roughly sketched, that Camp is simply offering an analysis of what occurs to the senses of names in the midst of a substantive identity claim; and that an analysis of reference can still proceed according to a model which treats a truth-value as propositional referent. Of course, this paper can only stand as an introduction to the threat posed by a Sellarsian linguistic theory to purely semantic theories of reference.

II. Talking About Characters

Camp’s use of the term “character” for the nominata of terms which refer (or report aboutness) makes sense given the context in which he introduces his analysis, where the principle objects are postulated entities, like “celestial spheres” and Maria (from Leonard Bernstein’s West Side Story). It is clear, however, that reference in general is being discussed, and we are being asked to give particular attention to the manner in which meaning traverses the language via names (as devices which denote). The language of “character” follows the central metaphor of a narrative theory of meaning.[3] “Character” also highlights the ways in which names, or their use, cognitively “pick out” individuals.

Names are more than simply continuity devices: more than merely vehicles for the preservation of topical regularity. Names are not mere placeholders, bridging disparate conversational (or semantic) narratives in order to conceptually (or even physically) locate the referent(s). For Camp, no names are merely proper or referential. Rather, any name, insofar as it meaningfully comports some object, imports a conceptual and conversational history of associations (never strictly semantic) which help to individuate and characterize the referent (or nominatum). Certainly, the class identified by ‘John’ shares nothing in virtue of the name. The name, term, or sign, might thereby be said to denote without connotation (in Mill’s language). But in order for the name to pick out a specific referent (eg., this John), a selection function must be provided gesturally: conversationally or contextually. This might be called the sense of John. And this conclusion, in itself, is no threat to Frege. Indeed, it often seems as though it is precisely the sense of a term or expression which fixes the referent. But Camp is reaching further than sense, to treat the narrative (even imaginative) context of local reference as relevant to referential content.

He suggests three aspects of “character” which distinguish this from a more classically understood nominatum.[4] First, a character-nominatum is said to possess “idiosyncratic traits” (individual propensities and capacities) which, as the name implies, are uniquely held of it, and are sufficient but not necessary for referential individuation. Second, as a speaker I must have “individual-specific knowledge” with respect to a character-nominatum in order that I be able to cognitively identify it. An understanding of character-object p involves certain p-specific facts which themselves are not readily inferred by a localization of known generalizations. (eg. I may know that p is brown, without knowing what physical laws make it so.) And third, “individual-specific attitudes”: “These are attitudes which cannot find full basis in the discernible properties of the individual in question.” Of course, this is not to say that there are no properties which cause us to have such attitudes; only that those properties need not be transparent to the one who possesses such attitudes.

There are clear cases where the treatment of a name’s referent as a character in an individual’s or group’s narrative is tremendously helpful. Recall one of the sentences which began this paper: “Thomas Mann vigorously denied reports that Adrian Leverkühn was in fact Arnold Schoenberg.” What is being asserted here? Thomas Mann is being credited with having denied the following assertion:

Adrian Leverkühn is Arnold Schoenberg.

Following Camp, we should translate this assertion roughly as:

(Ex) (Ey) (In our inner narrative of things x is characterized as Arnold Schoenberg, and Thomas Mann’s thought-story characterized y as Adrian Leverkühn, and x reappears as y.)

Thomas Mann can then be taken as having disputed the third conjunct of the analysis. And this seems plausible. Both Leverkühn and Schoenberg are particularly susceptible to treatment as “characters,” the one fictional and the other quite real. The appropriateness of treating them in terms of character is further strengthened by the particular relevance of personality in this case. But what about an example in which neither character-nominatum is a fiction or a person? Consider the following:

Hesperus is Phosphorus.

Here we can view Camp’s analysis as an interpretation of nominata as characters in a shared narrative. We translate this assertion as:

(Ex) (Ey) (Our archaic thought-story characterized x as ‘Hesperus’, and our archaic thought-story characterized y as ‘Phosphorus’, and x reappears as y.)

This seems plausible enough. The cognitive relevance of such a claim does concern the wider significations of each object, as it is treated in our language. But Camp adds an important wrinkle:

The claim that character y is character x, or that x reappears as y, is correctly assertible on the basis of, and in the context of, a good interpretation of the narratives in question, which interpretation requires seeing y as x. Human intelligence being what it is, there will always be other interpretations equally “good,” but which have different goals and interests in view, in the context of which these claims are not correctly assertible. [...] The reappearance of a character from one narrative in another is a soft fact, a matter of interpretation in precisely the sense I have described.[5]

As it is expressed here, the matter of whether to affirm the third conjunct (i.e. whether it is sensible to view the appearance of two signs as naming the same character) is one of interpretation on the (partial or full) characterization of each term’s nominatum. The interpretation not only asserts a shared nominatum, but by the phenomenon of “reappearance” an intersection of complimentary or contradictory narratives. Where the narratives are complimentary (eg. Phosphorus reappears as Hesperus) an equivocation of reference expands the characterization of the mutual nominatum. Where they are contradictory or incongruous (eg. Classical momentum reappears as Relativistic momentum)[6] an equivocation of reference may express a genuine disagreement concerning the appropriate characterization of the nominatum. In both cases, however, there may be circumstances and goals under which an equivocation of referent will not be “correctly assertible.” If we seek to make clear the nuanced peculiarity of Hesperus (at least as it was understood archaically), it will be unhelpful (and possibly “wrong”) to point out that it is Phosphorus.

Asserting a convergence of nominatum (a reappearance of character) involves the claim that it is helpful to view one thought-story’s character as possessing many or all of the attributes of another thought-story’s character, even if such characteristics are never explicitly ascribed to such a character by one of the thought-stories. Following Camp’s example, an historian who favors continuity will assert that it is helpful to view the Islamic sphere-cluster character as playing an inferential role just like the sphere-cluster character for the ancients, even if much of those roles were never explicitly ascribed to this character by the Islamic astronomers. And in my example, this same historian will likely assert that is helpful to view Einstein’s momentum character as playing an inferential role “just like” (or analogous to) the momentum character for Newton, even though the substantive formulation and behavior of Relativistic momentum diverges from the one proposed by Newton (recall, Newton offered his definition of momentum stipulatively as the product of mass and velocity, where each of these were taken as absolute values; and relativistic momentum is not only calculated differently, taking mass to be variable according to velocity, where velocity is anything but an absolute measure of the rate of motion, but such momentum is no longer conserved across collisions[7]).

Agreement as to the characterization of a sphere-cluster model or momentum, or any object of inter-subjective signification, is treated as a complex ascription of general translatability (“reappearance”) across the inner thought-stories of individual members to the agreement.

Nuanced ascriptions of collective thought often introduce their nuances by implying the “presence in thought” of a shared intentional object, shared by all the members of the collective. The best picture of what this really amounts to is this: The members of the collective have their inner narratives. An ascriber of collective thought to these people will want to portray certain characters in each person’s inner narrative as reappearances of characters in other people’s inner narratives, to get the effect of ascribing a truly joint intellectual process while at the same time highlighting certain individuals’ contributions to the collectively endorsed story — the nuancing. ... But it will always be optional not to do this, just as reidentifications of characters across ordinary literary works are optional.[8]

Such a collective might be as broad as all English speaking persons, or as selective as the editors of the Oxford English Dictionary. Is it optional to view ‘Hesperus’ as designating the same character-nominatum as ‘Phosphorus’? This still seems a rather strange thing to say of the assertion S2. Part of such a claim is that it not be merely a matter of textual ambiguity whether the nominatum of ‘Hesperus’ is distinct from the nominatum of ‘Phosphorus’. The optional nature of identity claims, as they have been described thus far, occurs where we are ascribing an identity of reference across thought-story boundaries. It is optional whether we endorse the claim that ‘Bohr and Thompson corrected Democritus on the nature of the atom,’[9] precisely because it is not at all obvious (and indeed cannot be established) that Democritus meant by ‘atom’ a structure susceptible to the analysis provided by Thompson, and then Bohr. Indeed, it could well be argued that the discernment of complexity within atomic structure (protons, electrons, orbits, etc.) already violates the metaphysically fundamental character atoms possessed for Democritus. What did Democritus have in mind then? We might endorse the claim above in order to emphasize the constitutive role the nominata of ‘atom’ continue to play in our theory of physical chemistry, and to suggest that it is helpful to view this (the conception of the actual structure of that constitutive building block) as a point of authentic disagreement between Bohr and Democritus.

But what about intra-thought-story identity claims? Might they also be open to a similar ambiguity? ‘Hesperus is Phosphorus’, ‘Mark Twain is Samuel Clemens’, and assertions of this sort are not so plainly attributions of referential agreement between distinct character-worlds (though they are ascriptions of referential agreement between distinct terms).

One suggestion is that we view such assertions according to what Camp calls “branching intersubjectivity.”[10] A branching is where one narrative takes the nominata of several terms to be distinct, while another takes the terms to name a common referent. Such a branching is how we should view the claim S2. After all, this is an assertion about archaic characterizations of ‘Hesperus’ and ‘Phosphorus’, and a claim that the extant character of Venus (which is the extant, or proprietary, re-characterization of ‘Hesperus’ and ‘Phosphorus’) is the intentional target of both (or is an extant reappearance of these two characters in one). What is being asserted by S2 is the usefulness of preserving much of the characterizations we had previously held of ‘Hesperus’ and ‘Phosphorus’, now in concert of a single character (also named ‘Venus’). This helps explain how S2 might not be correctly assertible in all contexts. We might just as well assert that neither ‘Hesperus’ nor ‘Phosphorus’ succeeded in picking out the same object, if we were more interested either in preserving archaic use or in dismissing exotic elements of the archaic characterizations (this is something I might want to do by denying that Bohr’s atom was the self same atom of Democritus).

The inverse of “branching” is ontological confusion. If the interpretive indeterminacy of “correctness” introduced by “branching” seemed to threaten our ability to make scientific judgments with respect to the preservation of referential agreement across thought-story boundaries, “confusion” now threatens its viability within a given thought-story. “In general,” Camp says, “I suggest that aboutness is the degenerate case of ontological confusion.”[11] An ontological confusion is irreducible, as a phenomenon of belief about character-nominata. Thinking that A and B (where we take A and B to be distinct entities) are characterized univocally by ‘P’ (believing that A is B) cannot be identified with merely possessing a group of false beliefs. One could attribute singularly to A (or B) all of the known properties and activities of both A and B, without ever believing that A is B. Nor can it be identified with the possession of a “de re confusion” (i.e., believing of A that it is B). One can have no beliefs simply of A (or strictly of B) so long as she believes that A is B.

The attribution of an ontological confusion can be analyzed in a manner similar to S1 or S2, where we take the variables to pick out distinct characters in our thought-story, but pick out a single character (to reappear as one character), in the thought-story of a confused individual. The implication of such an analysis, reaching beyond a mere depiction of the nature of our disagreement with a confused believer, casts doubt on the supervalence of our own “unconfused” standpoint. We are offering the confused believer another interpretation of the characters in our common language. The rightness of such a move will depend, not (simply) upon appeals to “facts,” but upon the usefulness (according to some set criteria) of our proposed equivocation. Not only is the semantically determinate facticity of nominata undermined, but the ontological acuity of our own thought, even where it is most “rational,” is no longer assured (though it still possesses a prima facia legitimacy in cases of proprietary compliance).

The clarity and distinctness of thought we attribute to one another, the lack of ontological confusion, is a matter of regimentation by those who attribute it — no doubt justified by our explanatory interests as they stand, but optional nonetheless.[12]

Of course, Camp realizes this comes terribly close to a full endorsement of a central notion for classical Idealism. Reports of aboutness are viewed as reports which ostensibly concern the content, not of the world, but of our (inner) narratives (be they of the ‘World’, or not). While idealistic in its portent on this level, the view, if grounded in a wider pragmatics, promises not to be viciously idealistic.

III. Of Sense and Character

Has Camp merely conflated sense and reference? Certainly Frege would contest that we have simply been speaking of the sense of denoting terms by speaking of “characters.” If Camp’s picture is correct with respect to the meaningful content of embedded identity assertions (of the sort, ‘Hesperus is Phosphorus’ or ‘Bohr corrected Democritus on the nature of the atom’), then it appears that such assertions fail to actually refer at all, or do so only indirectly, if we mean to maintain the model of propositional-referent as truth-value. It would seem that an assertion of the sort ‘Hesperus is Phosphorus’ is, according to this analysis, really about the “manner and context of presentation”[13] with respect to these terms.

Frege can be seen as dividing the characterization of a nominatum by a name into two categories: the sense and the image. The sense is the characterization brought about by the proprietary behavior of the name in general usage: it is, he says, “the totality of designations of which the proper name is a part.”[14] Knowledge of a nominatum involves the capacity to judge membership in the set of a nominatum’s proper senses (that is, involves the articulation of “individual-specific knowledge”). And this itemization, Frege suggests, is one which can never be completed (human intelligence being what it is), further strengthening the similarity of sense with the content of “character.”

The speaker-specific analysis of character, however, suggests a possible closer allegiance with Frege’s “image,” which is interpreted as the cognitive experience of using or hearing a name, as it appears to the single person.[15]

The nominatum of a proper name is the object itself which is designated thereby; the image which we may have along with it is quite subjective; the sense lies in between, not subjective as is the image, but not the object either.[16]

Fortunately, the privacy of content imparted by image is not congenial to “character,” whose primary strength is the development of an inter-subjective account of the objects of thought.

Of course, the reason for this exercise in re-imagining character as sense (or as a cluster of senses), is that interpretive choice can coexist with truth-as-referent, if that choice is taken to involve the content of terms not intended to actually pick out an existent object.

The sense of a subordinate clause is usually not a proposition but only part of one. Its nominatum is therefore not a truth-value. The reason for this is either: that the words in the subordinate clause have only indirect nominata, so that the nominatum, not the sense, of the clause is a proposition, or, that the clause, because of a contained indeterminately indicating constituent, is incomplete, such that only together with the principal clause does it express a proposition.[17]

Similarly, we can view Russell’s distinction between primary and secondary occurrences of denoting phrases in propositions, as an effort to partially address gaffs in intended reference,

A secondary occurrence of a denoting phrase may be defined as one in which the phrase occurs in a proposition p which is a mere constituent of the proposition we are considering, and the substitution for the denoting phrase is to be effected in p, not in the whole proposition concerned.[18]

How might Camp’s analysis be seen as an analysis of secondary occurrences, or of denoting terms embedded in subordinate clauses? Could it be a cloaked restatement of the trouble “speaker reference” seems to make for traditional theories of reference? By “speaker reference,” I have in mind the distinction drawn by Saul Kripke between it and “semantic reference,” as part of a general response to Keith Donnellan’s proposed objection to Strawson and Russell in “Reference and Definite Descriptions.”

Camp’s style of analysis is directly addressed to situations of attributed reference, and finds in such attributions a deep semantic indeterminacy with respect to the assertibility of identity claims. This is not an analysis of “attributive reference,” in Donnellan’s language. A certain degree of confusion is possible on this point, and is therefore worth clarifying. Treating the semantic phenomenon of reference in terms of “attributions of reference” seeks to locate the act of referring in a pragmatic context, where “to refer” entails a cluster of linguistic and action-guiding commitments. Donnellan’s “attributive reference” involves the intention of a speaker to explicitly refer through the mediation of (or directly to) a designator’s connotation (or character). But why should Kripke’s thoughts concerning semantic and speaker reference be relevant here?

In part, it is because we might try to re-articulate Camp’s analysis of identity claims in terms of an analysis of claims about the presence (and nature) of the denoting terms in the idiolect of particular individuals (or the dialect of particular communities), and not (or only indirectly) about the nominata of these terms, where a semantic ambiguity exists concerning the use of names reporting the denoting terms (eg. I can report the presence of a particular name, ‘John’, which denotes an individual person John; or I can report the presence of a name, whatever it may be, which denotes John). And this begins to sound very much like Kripke’s analysis of speaker reference. The objects of specific intention, and even general intention, become the referring terms themselves, and more generally, the senses of these terms: it is optional whether to see Bohr and Democritus as attempting to describe the same object by ‘atom’, just because neither Bohr nor Democritus were in a position to articulate a mutual “general” intention with respect to the characterization, or sense, of the nominatum of ‘atom’. Character reappears as sense. So an interpretive choice exists, in part, due to a semantic ambiguity brought about by taking this sense as the nominatum in cases of “complex” reference.[19]

If this is right, without too great a harm to the spirit of Camp’s analysis, we could translate S2' as something like the following:

(Ex) (Ey) (In our archaic thought-story x is (tenselessly) characterized by the proprietary use of ‘Hesperus’, and in our archaic thought-story y is characterized by the proprietary use of ‘Phosphorus’, and x reappears as y.)

where x and y are the senses of ‘Hesperus’ and ‘Phosphorus’ respectively. It is a complex case of intended reference because I intend by ‘Hesperus’ (and ‘Phosphorus’) to refer to a general class of proprietary uses, but more directly refer to a specific class of uses I take as proprietary (which I believe coincides with the general class, but which may not). This would certainly explain why such an assertion has cognitive significance, given the fact that both terms select the same nominatum, namely Venus. But there are two problems with such a reading of Camp’s analysis.

The first problem is that it has said nothing of the referent itself. The analysis attends so closely to the mechanism of reference (the senses of the terms), that the object of mutual designation slips its grasp. Within such an analytic frame, it hardly matters whether the terms in fact pick out the same object. But this is the point of asserting such a claim. ‘Hesperus is Phosphorus’ is not an assertion whose correctness can be assured (or even understood) without some appeal to the actual thing named by each term. In fact, by making such an assertion, I am hardly even thinking about the respective senses of the two terms (even though it is they which make such a claim non-trivial).

This brings us to the second problem. The final conjunct of this translation (“x reappears as y”) interprets the identity assertion to concern the senses of ‘Hesperus’ and ‘Phosphorus’. But this would be a rather ridiculous claim; and would in fact be false whenever our identity assertion succeeded in being non-trivial (i.e., was an identity assertion across terms with differing senses). By claiming ‘Hesperus is Phosphorus’, I may be claiming that they can be used interchangeably throughout the language, but I am certainly not claiming that their inferential roles are identical in every case. I am asserting something about a specific part of each term’s proprietary use; but it is the part which is most sensitive to the status and nature of the nominatum itself. In a manner of speaking, it is the thing itself, not the mechanism of designation.

Characters are not senses, though they are objects for which “characterization” (their sense or senses) is of paramount importance. Is this not true of Frege’s nominata as well? An explanation of the optional correctness of character based identity claims must lie elsewhere.

IV. Propositional Aboutness

We might suppose that the interpretive ambiguity is due to a failure to see which proposition is expressed by a sentence containing a name (eg. ‘John’). Propositionally, it is thought, the referent is clear. The semantic referent would be the “normal” (“general”) or “proper” propositional content of a term as it appears in our sentence. How is this proper mapping determined?

Somehow, in passing from the sentence to a proposition, ‘John’ becomes a trans-linguistic pointer, which uniquely and unambiguously picks out a real object named John (or are we to say, “which is uniquely John himself”?). The nature of propositions becomes terribly tenuous here. Are propositions, like mathematical objects and assertions, part of a virtual reality, self-sufficient and distinct from physicality? Surely reference would not be clarified by such a model. Are they real-world objects, somehow floating about in the ether of social space, “seen” only by direct cognition? This seems even more exotic and strange a suggestion, pushing propositions into a trans-sensory realm of artifactual “reality,” but again does not help explain reference. Are they, instead, nothing other than the real-world objects, and the relations they illuminate; the objects which are intentionally appealed to by rational locution? This would mean the proposition expressed by “Socrates is mortal” is the membership relation of the man Socrates to the class of all mortal things (now, before, and ever after). This appears to have some merit. Scientific assertions now bear a wonderful degree of reputability: propositions are not merely about the real world, they are the real world.

Unfortunately, assertions about fictions, or about language itself, or even mathematics, become strangely contorted (at best). It would seem that a sentence like “Hamlet killed Polonius” not only fails to refer to real-world persons, but fails to have propositional content! Likewise, though not as difficult, “Bank is an ambiguous sign” must be explained in terms of all actual utterances of “Bank,” and the class of ambiguous signs (where a sign is a movement from speech acts to ostensible objects, which themselves could be other speech acts). Of course, cognitions could not be propositions (unless we are keen to a materialist doctrine of mental states, which rather directly model propositional content in themselves), though they could be (and are) of propositions. Similarly, a speech act or sentence could not be a proposition, but could easily refer to one. This should suggest the more profound difficulty. In our effort to explain reference in terms of propositional content, we have simply passed propositions over to the other side. The divide still exists between signs and nominatum, but now lies between expressions (names, words, speech acts, etc.) and propositions, where it had once been between propositions and nominata.

The problem of reference now curiously resembles the mind/body problem in metaphysics. No matter how we frame the story, we always have trouble moving across the categorical divide between the cognitive (linguistic) and the objective (that about which we speak). But it also does an injustice the traditional role of propositions. Frege viewed propositions as playing very much the same role in guiding sentences to truth values as sense in guiding designators to their nominata.

Why then is reference generally framed as a semantic (or propositional) phenomenon? A purely semantic analysis of reference is guided by the insistence that “truth-values” are the proper measure of propositional integrity — or that a binary valuation of propositions, as “true” of “false,” as a primitive function on propositional content — depends upon the real-world existence of the referentia (nominata) of constituent terms. Still, I would maintain that ‘Hamlet slew Polonius’ is not false. Nor is it truth-valueless. ‘Hamlet slew Polonius’ is true: which is to say, as speaker, I endorse inferences about how the act of killing reflects upon Hamlet’s character, and so on. The proviso that Hamlet and Polonius are fictional does not seem to change this. What is changed is the treatment of the battery of inferences surrounding issues of each character’s physicality, to the extent that such issues are left textually vacuous (eg., ‘Hamlet lived in 1492’ is not false, but absurd).

Shifting analysis from a question of P’s truth, to an analysis of the claim ‘P is true’, avoids many of the paradoxes of “empty reference.” The prior commits us to an analysis of P which aims for propositional objectivity; that is, it is an analysis which treats the proposition expressed by ‘P’ (and not an utterance of ‘P’) as an object susceptible to empirical verification. The latter abandons, or leaves untouched, any claim concerning propositional objectivity, and merely concerns an analysis of inferential commitments entailed by particular truth claims.

But what does this do to truth? Can we still suppose that the nominatum of a sentence (or alternatively, the nominatum of a proposition) is its truth-value? What sort of claim is involved here? Are we still bound to assert that all true sentences name the same “character” (i.e., the true); and likewise all false sentences? What kind of “character” is truth? Immediately, we should recall that our analysis will concern the truth-claim and not the truth of the claim. A question of truth (eg., ‘is it true that P?’) will be a question of practical endorsement and commitment to proprietary entailments, and not merely a question of strict propositional integrity, though indirectly, it will involve justificatory strategies which may find offense in existential gaffs (of the sort involved in literary assertions).

... one interpretation might be better worked out than the other, or more illuminating for some other reason, but in principle each could be as “right” as the other, because each serves its own set of interpretive goals.[20]

Scientific truth, or existential accuracy, should not be viewed as an interpretive goal itself, but should itself be seen in terms of other goals, whether merely instrumental, aesthetic, political, or some more complex (though no less humanistic) value cluster.

What then would it mean to talk about the propositional content of a sentence? I suggest that it is no other than the proprietary content of that sentence indexed according to the author’s idiolect. This might at first seem a trivial point. There are, however, two ways in which propositional objectivity is undermined here (and this would seem to be far from trivial). First, propositional content is a matter of proprietary use; and second, the assertion of proprietary validity is projected from within the author’s idiolect onto the public language. If I claim ‘P is the propositional content of ‘Q’’, I am asserting that we ought to take a speaker as meaning P when she asserts ‘Q’, where P and ‘Q’ are also taken as conforming with proprietary use (which in turn, commits me to the claim that I mean by P and ‘Q’ what any responsible speaker, in my language, should mean). The semantic reference of a term, or even a sentence, would then be a variant of speaker’s reference, where a further claim is embedded, to the effect that this object is the appropriate one to take as referenced (named).

V. The Act of Referring

When I refer to an object, I talk about it. Can a term or sign “refer”? Or is reference a thing only a speaker can do? I refer to the Sun by the use of ‘sun’, appropriately contextualized in conversation or action. Reference is an act. It cannot be understood semantically. It is not a semantic artifact. The semantic content, the propositional nominatum, of a sign cluster (a sentence presented in some manner or context) is dependent upon, or fixed by, the interplay of its conversational occurrence and the proprietary nominata of its embedded terms.

What else could do the work of selecting, out of among the nominal class ‘John’, the individual object relevant to semantic analysis? A semantic analysis needs a unique John, unless we are willing to give up the chase and accept a fundamental cognitive ambiguity. Pragmatics can resolve the ambiguity, but at the cost of maintaining a strictly semantic analysis of reference along the lines of Frege or Russell. Our analysis here uproots the semantic self-sufficiency of signs in terms of the determination of their referentia. But by relocating the occurrence of reference in action (perhaps as part of a wider theory of speech acts) we avoid the threat of idealism and the challenge that Camp’s analysis can say nothing about the actual referentia of terms.

The interests and goals which determine the “correctness” or “confusedness” of aboutness are only available contextually: in the pragmatic locatedness of an assertion of referential unanimity or distinctness. Referential success or failure might be a part of semantic determination. But the precise existential status of the referent can not be obtained by a purely semantic analysis. When I speak of ‘Hamlet’, I do indeed speak about something. That about which I speak is not merely the set of characteristics attributed to the character Hamlet, but the “character” himself. I can say, of Hamlet, that he was not confused or hesitant out of a weakness of will, but was caught by conflicting notions of justice (avenge thy father’s murder; and obey and respect thy mother’s husband and King). An evaluation of such an assertion need never look to the existential status of the nominatum of ‘Hamlet’, but also never depends merely upon the conventional associations of the name itself. I am not asserting that the icon ‘Hamlet’ ought to be viewed as comporting a conflict of moral imperatives; it is not the name I am speaking about, but he who I name by it.

Philosophy of Language: Reference and Singular Terms
PHIL-725-01, Georgetown University
Fall 1991
(© David Foss, January 10, 1994)

Send comments to hok007@shlobin-foss.net
Last modified September 1, 1998

home | philosophy index | notes | comment