Abstract
This essay examines the syntax of names. It argues that names are a syntactically and not just semantically distinctive class of expressions. Its central claim is that names are a distinguished type of anaphoric device—devices of explicit co-reference. Finally it argues that appreciating the true syntactic distinctiveness of names is the key to resolving certain long-standing philosophical puzzles that have long been thought to be of a semantic nature.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
In fairness to Fodor, though, he also says that syntax has an “external” face since “the syntax of a representation determines certain of its relations to other representations.” But I take syntax properly so-called to be constituted by the sort of representation-representation relations toward which Fodor seems here to gesture.
There are of course semantic relations among representations—relations of synonymy, for example. But such relations presuppose that expressions already stand in various representation-world relations.
This is not to say that narrowly Fodorian syntax is entirely irrelevant to syntax in the broader sense. If a competent speaker is to be able to recognize and competently deploy an expression in syntactically appropriate ways, then there must be something about the expression that enables such recognition and deployment. And intrinsic shape-like properties may serve as good cognitive handles, as it were, on syntax in the broader sense.
It is sometimes objected to my view that the twin properties of explicit co-referentiality of co-typical name tokens and referential independence of type-distinct name tokens are not, on their own, sufficient to distinguish names from certain other sorts of expressions. For example, all tokens of the type ‘tiger' refer (rigidly) to the species "tiger," it is sometimes observed. Similarly with the word ‘yellow'. And it is objected that one might therefore worry that my account fails to pick out any distinctive property of names. But I claim only that the twin properties of explicit co-referentiality and referential independence partially characterize the syntactic category NAME. Names are also expressions that, for example, may well-formedly flank the identity sign and well-formedly occupy the argument places of verbs. Some totality of such properties jointly constitute a broader, but still syntactically characterized class of expressions—the class of SINGULAR TERMS. Think of the category NAME as a distinguished subclass of the class of SINGULAR TERMS. Now included in the class of singular terms are also demonstratives and indexicals. I discuss the co-reference profiles of demonstratives in this essay and argue that demonstratives and names constitute something like a minimal pair within the category of singular terms. So my approach does require an antecedent analysis of singular termhood and I, admittedly, haven’t offered such an analysis here. But the point is that the twin properties of referential independence of type distinct tokens and explicit co-referentiality of co-typical tokens is meant to distinguish names from other elements of the class of singular terms. It is of no consequence for my view if other expressions in, say, the category PREDICATE or the category COMMON NOUN turn out to have “correlative” syntactic properties. In particular, that would not suffice to make predicates be names or names be predicates or to obliterate the important syntactic distinction between names and predicates.
Some have been tempted to assimilate names to variables (either bound or free) under assignment. But I do not think that this is quite right because I think that reference of every token of a given name is determined, as it were, in one fell swoop by facts about the name type of which the token is a token. What we might call the type-reflexivity of names is also reflected in what I take to be the fundamental semantic rule governing all names—if e ∈ NAME, and t is a token of e, then t rigidly designates the object o such that o BEARS e, where e is an expression type, rigid designation is a semantic/modal relation, and BEARS is a this worldly causal relation distinct from the semantic relation of rigid designation.
But I stress that to display that two names are co-incidentally co-referential is not to say that they are co-referential. The subject matter of a true identity statements is not metalinguistic but objective.
See also Récanati (1993).
For a further defense of this claim see Taylor (2010).
For a suggestive and helpful discussion of mental anaphora and its role in identity thinking and content-preservation, see Lawlor (2002).
Kant was perhaps the first to see clearly that an inner capacity for thinking with same-purport was essential to enable our thought to achieve what I have elsewhere called full objectivity (as opposed to mere objectuality). Consider, for example, the following passage from the A-Deduction
Without consciousness that that which we think is the very same as what we thought a moment before, all reproduction in the series of representations would be in vain. For it would be a new representation in our current state, which would not belong at all to the act through which it had been gradually generated, and its manifold would never constitute a whole, since it would lack the unity that only consciousness can obtain for it. If, in counting, I forget that the units that now hover before my senses were successively added to each other by me, then I would not cognize the generation of the multitude through this successive addition of one to the other, and consequently, I would not cognize the number
Kant takes the points made above about how numbers are cognized to be perfectly general and to apply to the cognition of any object as such.
Distinct names are ipso facto referentially independent. Moreover, unless it is made explicit either by context or by the addition of an identity statement explicitly linking the two, the use of referentially independent names generates a defeasible imputation of distinctness. Because of this fact, substitution of names that are merely co-incidentally, as opposed to explicitly co-referential is not, in general, guaranteed to preserve what I have elsewhere called dialectical significance. It is unfortunately, quite easy to confuse failure to preserve dialectical significance with failure to preserve truth value. A clear failure to appreciate the significance of the difference between failure to preserve truth value and failure to preserve dialectical significance is exemplified in Saul (1997). Though Saul was one of the earliest philosophers to take explicit notice of the failure of co-referring names to be freely substitutable even in simple sentences containing no embedding constructions a lack of a full understanding of the distinctive co-reference profiles of names led her to draw a number of erroneous conclusions. See also Saul (2007).
My claim is not that co-referring and co-typical deictic tokens can never be interpreted as co-referential. There are in fact sentences in which it seems all but mandatory that two co-typical deictic tokens be interpreted as co-referential. Consider the following:
-
(a)
Ted saw that man and Bill saw that man too
-
(a′)
Ted saw (that man)i and Bill saw himi too
-
(b)
John hates that man because that man is a cad
-
(b′)
John hates (that man)i because hei is a cad
On the default reading, an utterance of (a) would seem to be roughly equivalent to an utterance of (a′). Similarly, an utterance of (b) is roughly equivalent to an utterance of (b′). It may be tempting to conclude that there can indeed obtain a relation of anaphoric dependence between subsequent and antecedent deictic tokens of the same type. But what we really have here is co-reference through what I call demonstration sharing. Co-reference through demonstration sharing occurs when a speaker uses the reference fixing demonstration associated with an “antecedent” deictic to also fix the reference for a subsequent deictic. When two token deictics share a demonstration, they will indeed co-refer, but co-reference through demonstration sharing is a purely pragmatic phenomenon. It is, moreover, a species of coincidental co-reference rather than a form of linguistically controlled explicit co-reference.
-
(a)
Brandom (1994) makes a similar point. Where I distinguish between devices of explicit co-reference and devices of de novo references, he distinguishes between repeatable and unrepeatable tokenings. Names and anaphoric pronouns are repeatable, while demonstratives generally are not. Names, he says, must be “understood as anaphoric dependents—elements in an anaphoric chain that is anchored in some name-introducing tokening.” Moreover, Brandom argues—quite correctly, in my view—that anaphoric reference of the sort that requires repeatability is more fundamental than what I have called de novo reference. As he puts it:
Deictic uses presuppose anaphoric ones. One cannot coherently describe a language in which expressions have demonstrative uses but no pronominal uses (although the converse is entirely possible). For indexical uses generally, like deictic ones, are essentially unrepeatable according to types. Different tokenings of ‘this’ or ‘here’ or ‘now’ are not in general recurrences of each other, or even co-identifiable. (464-65)
Though the precise letter of Brandom’s views are rather different from my own, there is significant overlap between the spirit of his view and the spirit of my own, at least on this score.
See Jeshion (forthcoming) for a far more developed argument against the names as predicates view. In rejecting the view that we must give a unified treatment of genuine names and name-like predicates, Jeshion argues that the predicativist fails even to give a unified treatment of predicative uses of name-like expressions. For an earlier criticism along similar lines to Jeshion’s see Boër (1975).
And in this I take myself to be in disagreement with Kit Fine’s (2007) relational semantics. Moreover, my approach helps explains why we have different reactions to sentences like (a) and (b) below:
-
(a)
Clark Kent went into the phone booth and Clark Kent came out.
-
(b)
Clark Kent went into the phone booth and Superman came out.
Contra Saul (1997) and Saul (2007) (a) and (b) do not differ in truth value. (Here I bracket the fact that since there is no Clark Kent and no Superman either, each of (a) and (b) are both, on my view, truth valueless. See Taylor (2014) for a more realistic treatment of such fictional names.) Nonetheless (a) and (b) can, in the right setting, differ in dialectical significance, since (a) does and (b) does not generate any imputation of distinctness. The author of the Superman comic books cleverly exploits the referential independence and co-incidental co-reference of ‘Clark Kent’ and ‘Superman’ to very good pragmatic effect. In particular, it is part of the background story, within which these names are at home, that although Clark Kent and Superman are one and the same person within the story, this fact is unknown to most characters in the story—though it is known to the readers of the story. It would be an interesting exercise to trace all the pragmatic effects of this setup in greater detail. I lack the space to do so here. I will just say that it seems to me that whenever the two names are used together in the same sentence or context, the reader is supposed to attribute a sort of in the story distinctness to Clark and Superman at least for the other characters, while recognize that the characters are, in fact, confused and that in the story Clark Kent just is Superman. Moreover, the use of either name in isolation from the other evokes in the story continuity of reference and identity. But the deeper point is that this subtle exploitation of the facts about the co-reference profiles of names depends entirely on the broadly syntactic character of names and does not require us to any way bring to bear any sort of semantic analysis of names and naming.
-
(a)
For more on the distinction between narrowly or directly asserted content and total utterance content see Taylor (2007).
It is worth briefly contrasting my own view with a view that is similar in spirit but differ in detail from my own. I have in mind the views of Fiengo and May (2006). Though they seem to agree with much of the general spirit of much of what I say in this essay, they introduce a notion of de dicto content apparently to preserve certain Fregean intuitions. But given the arguments of this essay it is hard to see why any view roughly in the same spirit as mine should be moved by Fregean considerations. To make clear the points of contention between their view and my own, I need to introduce three bits of their own technical apparatus: (a) notion of a translation statement; (b) the notion of an assignment; and (c) the distinction between a de dicto and a non-de dicto logical form.
A translation statement is a statement of the form:
-
(25)
⌈“X” translates “Y”⌉
where X and Y are expressions that may be either co-indexed or not co-indexed. This notion of “translation” seems intended, at least in part, to capture speakers’ beliefs about coreference relations—both coreference relations of the linguistically marked variety and coreference relations of the linguistically unmarked variety. Co-indexed expressions are guaranteed to be translations of one another. Speakers have purely linguistic or grammatical grounds for their knowledge of such translation statements. But even expressions which are not co-indexed can stand in the translation relations to each other. There are, however, no purely linguistic grounds, according to Fiengo and May, for knowledge of such translation statements. As they put it, “Translation statements may also hold between noncoindexed expressions, and these are learned case by case.”
Consider next the notion of an assignment. An assignment, they say, is a “semantic belief” about the reference of an expressions. Such beliefs may be characterized by sentences of the form:
-
(26)
⌈“[NPi X]” has the value NPi ⌉
where X is a schematic letter covering the syntactic contents of NPi. Hence:
-
(27)
“[NPi Cicero]” has the value Ciceroi
and:
-
(28)
“[NPi Tully]” has the value Tullyi
represent beliefs about the semantic values of ‘Cicero’ and ‘Tully’ respectively. Now Fiengo and May use a system of indices to mark what they call expression identity. In their system, if NP’s are occurrences of the “same expression” then they are co-indexed. And if NP’s are not occurrences of the same expression, they are not co-indexed. Co-indexed expressions are coreferential. Expression that are not co-indexed may also co-refer, but their coreference will not be linguistically marked by the system of co-indexing.
Finally, consider the distinction between what they call de dicto and what they call non-de dicto logical forms. A dicto logical form, roughly, is a logical form that has an assignment as an explicit constituent, while a non de dicto logical form is a logical form that does not have an assignment as a constituent. According to Fiengo and May:
When speakers make assertions the Assignments believed by the speakers need not be part of the logical form of the sentences that speakers utter. They may be, rather, elements of the context that the speaker assumes. Assignments, however, are not always just background to assertions; they may stand as part of the truth conditions of the assertion itself, in which case they give rise to apparent exceptions to the Assignment principle.
In particular, Fiengo and May claim that both in the context of identity statements and in the context of (de dicto) ascriptions of propositional attitude ascriptions, assignments make their way into asserted truth conditional contents. To be sure, they acknowledge that although we may sometimes directly ascribe belief in an assignment to an agent, typically such ascriptions are merely tacit. But they claim that assignments need not be explicitly represented in order to make their ways into truth conditional contents. Speakers sometimes:
…commit themselves to a claim about the terms under which the agent holds a belief, and provide information as to what the agent believes over and above that given by the overt primary attribution.
I reject this sort of approach. To appreciate why, begin by considering the following two statements:
-
(29)
John believes that Cicero was a Roman.
-
(30)
John believes that Tully was a Roman.
Fiengo and May claim that (29) and (30) each have two logical forms—a de dicto logical form and a non-de dicto one. The two logical forms of (28) are represented by (31) and (32), while the two logical forms of (30) are represented by (33) and (34) below:
-
(31)
John believes that [Cicero1 was a Roman]
-
(32)
John belives that [[Cicero1 was a Roman] and [“Cicero1” has the value Cicero1]]]
-
(33)
John believes that [Tully2 was a Roman]
-
(34)
John believes that [[Tully2 was a Roman] and [“Tully2 “has the value Tully2]]
The idea is that a speaker has the option of ascribing an attitude in two different ways, using two different “logical forms.” She will use a non-de dicto logical form when her ascription is not intended to reflect the names that her ascribee herself would use to refer to the objects that the ascribed belief is about. She will use a de dicto logical form when her ascription is intended to reflect the ascribee’s own use.
But I suspect that Fiengo and May have, like Frege, committed a subject matter fallacy. It may well be correct to say that in certain contexts it is possible to “convey” information about “assignments” by an utterance of an attitude ascription. But even if one can show that such information can be conveyed in a given conversational setting, by a given utterance, it wouldn’t follow from that fact alone that claims about assignments enter into the strict literal truth conditions of the relevant utterance. In fact, it is not obvious how and why such information could manage to become part of the strict literal truth conditions of attitude ascriptions. To be sure, in claiming that de dicto logical forms are just that –logical forms, F &M may mean to suggest that at some level of representation a sentence like (29) above has a structure like that of (32) above. But F&M adduce no evidence at all that (29) is structurally ambiguous. Alternatively, they might think that (29) has some hidden or suppressed constituent that possibly calls for a contextually supplied assignment as its value. But I know of no reason to think this. One could also think that assignments get into truth conditions as either Perryesque unarticulated constituents or as the consequence of something like Recanati’s free enrichment. Now I have argued against all such approaches elsewhere and won’t repeat these arguments here. My point here is not that those approaches have been thoroughly refuted. It’s just that F&M don’t tell us anything about how a sentence like (29)—which makes no explicit mention of any metalinguistic information about assignments, which does not obviously contain some hidden slot or parameter that might take on an assignments as its value, which seems in no way structurally ambiguous or semantically incomplete on its face—would manage to have a logical form and truth conditions like that represented in (32). F&M owe us some story, but their book contains no such story.
Another view that shares much of the spirit of the views presented here is due to Fine’s (2007). But Fine wishes to complicate the semantics in ways that I find unnecessary. In particular, he thinks that we have to build representations of what we might call referential coordination directly into the propositional contents of our sentences. Thus on his view, “Cicero = Cicero” and “Cicero = Tully” will, in one sense, express the same singular proposition, but the proposition expressed by the first will contain a representation to the effect that ‘Cicero’ contributes Cicero not once but twice, while the proposition expressed by the second will contain two uncoordinated occurrences of Cicero. According to Fine, moreover, we must distinguish representations which represent the same object from representations which represent objects as being the same. He argues, for example, “Cicero = Cicero” represents Cicero “as the same,” while “Cicero = Tully” represents the object as being the same w/o representing the object as the same. Though I think Fine is on to an important distinction, what he sees as part of the semantics, I see as part of syntax. But part of the difference between Fine and me, however, may be that he sees syntax as a matter of mere typography. On my view, of course, syntax concerns considerably more. It has to do with expression classes, their formation rules, transformations that can be made on them, governance relations, anaphoric dependency relations, co-reference profiles and the like. Syntax so understood not only helps to explain various distributional facts, but also to “constrain” without fully determining semantic interpretation and relations. Perhaps if Fine were to adopt this broader notion of syntax his view and mine would be considerably closer to converging.
-
(25)
References
Boër, S. (1975). Proper names as predicates. Philosophical Studies, 27, 389–400.
Brandom, R. (1994). Making it explicit: Reasoning, representing, and discursive commitment. Cambridge, MA: Harvard University Press.
Burge, Tyler. (1973). Reference and proper names. Journal of Philosophy, 70, 425–439.
Chomsky, N. (1981). Lectures on government and binding. Foris Publications.
Chomsky, N. (1995). Minimalism. Cambridge, MA: MIT Press.
Fara, D. G. (2015). ‘Literal’ uses of proper names. In A. Bianchi (Ed.), On reference. Oxford: Oxford University Press.
Fara, D. G. (unpublished ms). Names as predicates.
Fiengo, R., & May, R. (2006). Belief De Lingua. Cambridge, MA: MIT Press.
Fine, K. (2007). Semantic relationalism. Malden, MA: Blackwell.
Fodor, J. (2002). The mind doesn’t work that way: The scope and limits of computational psychology. Cambridge, MA: MIT Press.
Frege, G. (1977). In P. T. Geach & M. Black (Eds.), Translations of the philosophical writings of Gottlob Frege. Oxford: Basil Blackwell.
Jeshion, R. (forthcoming). “Names Not Predicates” in On Reference. Andrea Bianchi (ed). Oxford: Oxford University Press.
Lawlor, K. (2002). Memory, anaphora and content preservation. Philosophical Studies, 109, 97–119.
Matushansky, O. (2006). Why rose is the rose: On the use of definite articles in Proper Names” in O. Bonami and P. Cabredo Hofherr, eds. Empirical Issues in Syntax and Semantics, 6, 285–307.
Perry, J. (2001). Reference and reflexivity. Stanford: CSLI Publications.
Perry, L., & Korta, K. (2011). Critical pragmatics: An inquiry into reference and communication. Cambridge: Cambridge University Press.
Quine, W. V. O. (1960). Word and object. Cambridge, MA: Harvard University Press.
Récanati, F. (1993). Direct reference: From language to thought. Oxford: Blackwell Publishers.
Saul, J. (1997). Substitution and simple sentences. Analysis, 57(2), 102–108.
Saul, J. (2007). Simple sentences, substitution, and intuitions. Oxford: Oxford University Press.
Taylor, K. (2003). Reference and the rational mind. Palo Alto, CA: CSLI Publications.
Taylor, K. (2004). The syntax and pragmatics of the naming relation. In C. Bianchi (Ed.), The semantics/pragmatics distinction. Palo Alto, CA: CSLI Publications.
Taylor, K. (2007). A little sensitivity goes a long way. In Gerhard Preyer & Georg Peter (Eds.), Context-sensitivity and semantic minimalism: new essays on semantics and pragmatics. Oxford: Oxford University Press.
Taylor, K. (2010). On Singularity. In Robin Jeshion (Ed.), New essays on singular thought. Oxford: Oxford University Press.
Taylor, K. (2014). The things we do with empty names: objectual representations, non-veridical language games, and truth similitude. In Manuel García-Carpintero & Genoveva Martí (Eds.), Thinking and talking about nothing. Oxford: Oxford University Press.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Taylor, K.A. Names as Devices of Explicit Co-reference. Erkenn 80 (Suppl 2), 235–262 (2015). https://doi.org/10.1007/s10670-014-9706-x
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10670-014-9706-x