Character, proper names, and Frege's Puzzle

Filosofia Unisinos Unisinos Journal of Philosophy 17(1):75-80, jan/apr 2016 Unisinos – doi: 10.4013/fsu.2016.171.10 PHILOSOPHY SOUTH Character, proper names, and Frege’s Puzzle Filipe Martone1 ABSTRACT Kaplan’s (1989a) solution to the indexical version of Frege’s Puzzle in terms of the character of linguistic expressions has been greatly influential and much discussed. Many philosophers regard it as being correct, or at least as being on the right track. However, little has been said about how character is supposed to apply to proper names, and how it could account for the name version of the Puzzle. In this paper I want to fill this gap. I sketch some solutions to the name version of Frege’s Puzzle in terms of character, and argue that all of them are flawed in some way: they are either semantically implausible or fail to account for all relevant phenomena. Keywords: reference, Frege’s Puzzle, proper names, cognitive value, philosophy of language. Take a look at the following sets of sentences, the first involving only coreferential indexicals and the second involving only coreferential proper names: (1a) I am me [Bruce Wayne pointing at himself] (1b) I am him [Bruce Wayne pointing at some footage of Batman on TV] (1c) I fight criminals at night [said by Bruce Wayne in his Batman suit] (1d) He fights criminals at night [said by Alfred pointing at Bruce Wayne] (2a) Freddie Mercury is Freddie Mercury (2b) Freddie Mercury is Farrokh Bulsara (2c) Freddie Mercury is talented (2d) Farrokh Bulsara is talented It seems clear that the cognitive values of those sentences are different, even though the singular terms they contain refer precisely to the same thing. Sentences (a) are trivial, while sentences (b) appear informative. A speaker can sincerely accept (c) sentences and at the same time sincerely reject (d) sentences (and vice-versa) without being irrational. But how is this possible? How can the mere substitution of coreferential expressions affect the cognitive value of a sentence? What accounts for the difference in their epistemic profiles if they refer to the same object? This is what is traditionally called Frege’s Puzzle. Kaplan (1989a) managed to deal with the indexical version of the Puzzle quite well. In his theory, indexicals have two levels of meaning: character and content. The content of an indexical is the contribution it makes to the proposition expressed. Since Kaplan is a referentialist, the content of an indexical just is the object being referred to. Character, on the other hand, is the linguistic rule that is attached to the indexical that determines the content in a context. More precisely, the character is a function from contexts to contents, and this function is associated with the indexical by the rules of 1 Universidade Estadual de Campinas. Rua Cora Coralina 100, 13083896, Campinas, SP, Brasil. E-mail: [email protected] This is an open access article distributed under the terms of the Creative Commons Attribution 4.0 International (CC BY 4.0), which permits reproduction, adaptation, and distribution provided the original author and source are credited. Filipe Martone language. Because of this, it is plausible to think of it as the linguistic meaning of indexicals that is known by every competent speaker. Clearly, Kaplan cannot appeal to contents to account for the differences in cognitive value between sentences of the first set above, for they all have the same content. So, it must be the character, not the content, that explains cognitive value: since the characters of ‘I’ and ‘he’ are obviously different (they are different functions from contexts to contents: the character of ‘he’ is something like ‘the discriminated male’ and the character of ‘I’ is ‘the producer of this token’), and they are known by every competent speaker, those sentences may differ in cognitive value even if they refer to the same object. They refer to the same thing, but, since their descriptive rules are distinct, they do so in different cognitively significant manners.2 The name version of the Puzzle, however, is nowhere near as amenable as its indexical version to a solution in terms of character. The reason for this is simple: for most referentialists (including Kaplan), names are just labels for their referents. There is no other level of meaning to a proper name other than the object that it stands for. In short, the meaning of a proper name is exhausted by its referent. If this is right, then all coreferential names, no matter how syntactically distinct, have exactly the same meaning. Or, as Kaplan puts it, in proper names “all three kinds of meaning–referent, content, and character–collapse. […] Because of the collapse of character, content, and referent, it is not unnatural to say of proper names that they have no meaning other than their referent” (Kaplan, 1989a, p. 562). Therefore, there is no difference in meaning between ‘Farrokh Bulsara’ and ‘Freddie Mercury’ that can be exploited by the referentialist in order to account for the difference in cognitive value of the sentences in the second set above. All there is to the semantics of coreferential names is simply identical. How can one account for the cognitive value of proper names in terms of character, then3? Kaplan himself acknowledges the difficulty (e.g. Kaplan, 1989a, p. 562). His theory simply does not have enough resources to solve Frege’s Puzzle as arising for names. In fact, this is a reason to suspect in principle his attempt to explain cognitive value via character, even in the case of indexicals. If we have a phenomenon and a theoretical entity that purports to explain it, then the fact that this entity does not even begin to explain a recognized subclass of the same phenomenon gives us enough 2 reason to suspect that it was not the entity that we were looking for in the first place. Therefore, if character really has this supposed epistemic dimension, it should have this dimension for all singular terms, not only for indexicals. In other words: if cognitive value could plausibly be explained by character, then all phenomena of cognitive value should at least be initially treatable via character. But, in the case of proper names, they clearly are not. Characters seem ill suited to explain the cognitive value of names right from the outset. If this is correct, then it looks like that the apparent relation between cognitive value and character in the case of indexicals that Kaplan was so enthusiastic about was merely incidental. But let us not be so pessimistic. Perhaps Kaplan is wrong about proper names. Perhaps they do have more than one level of meaning besides the referent. If character and content in proper names do not coincide in the way that Kaplan believes they do, then maybe character is able solve the name version of the Puzzle after all. This, of course, is obviously a major deviation from Kaplan’s original theory. But, as we have seen, if the referentialist does not part ways with Kaplan regarding proper names, then she has no hope when it comes to solving the name version of the puzzle in terms of character. There are several candidates for the characters of proper names, but it seems clear that all of them will be descriptions of some sort. Since characters are rules that determine the content, they must be given descriptively, just like the characters of indexicals. It is important to stress that this is in no way incompatible with direct reference. The biggest lesson we take from direct reference is that names are not equivalent to descriptions at the level of content (i.e., they do not contribute descriptive material to propositions). It is perfectly compatible with the general referentialist framework, then, for proper names to have a level of descriptive meaning just as indexicals do. The descriptions that state the characters of proper names have to function like descriptions coupled with Kaplan’s dthat operator:4 they will express conditions that must be satisfied by an object in order for it to be their extension, but contribute only their extension to propositional content. In short, the dthat operator turns descriptions into directly referential terms. The character of a proper name, then, could very well function as a description of its bearer combined with a dthat operator.5 This is of course an oversimplified explanation. However, for the purposes of this paper, we do not need to go into further details. 3 One might wonder why Kaplan should care if his solution does not apply to proper names. The reason is the following. If one believes that cognitive value is an aspect of meaning, as Kaplan does, it is an extremely ad hoc move to claim that the same phenomenon is accounted for by something semantic in the indexical case (i.e., by characters) and by something non-semantic in the case of proper names (e.g. non-semantic guises, as Salmon (1986) proposes, or anything pragmatic for that matter). Either cognitive value is an aspect of meaning or not. This is why proper names pose a difficult problem to referentialists who are sympathetic to Frege’s criterion of adequacy for semantics, i.e., the thesis that semantics must account for Frege’s Puzzle. And this is why Kaplan is so frustrated that his solution via character does not extend to proper names. In sum, if one shares the Fregean intuition that cognitive value is a feature of meaning, then one must show how something semantically relevant tracks differences in cognitive value adequately for all singular terms, not just for indexicals. Thanks to an anonymous referee for making me flesh this out more precisely. 4 Cf. Kaplan (1978, 1989a, p. 521-22, 1989b, p. 578-582). Kaplan says that dthat should be understood as a true demonstrative, nor as a real operator. For our purposes here, however, this does not make much difference. 5 Note that the dthat operator is not the same as the actually operator, which is usually appealed to by descriptivists about proper names to deal with Kripke’s objections. The former stays at the level of character, so to speak, while the latter carries over to content. In other words, the actually operator just rigidifies a description; it does not turn it into a directly referential term like dthat. Filosofia Unisinos – Unisinos Journal of Philosophy – 17(1):75-80, jan/apr 2016 76 Character, proper names, and Frege’s Puzzle However, contrary to indexicals, which have more or less easily statable descriptive characters, finding which descriptions are good candidates for the characters of proper names is an enormously difficult task. Since these descriptions function as characters, they have to somehow be cognitively accessible to all speakers who are competent regarding the name in question, and they must be responsible for determining its referent. The options seem to be the following: (1) (2) (3) (4) Causal-historical chain description Specific-name metalinguistic description Generic-name metalinguistic description Context-sensitive description Options (1) and (2) are more conservative, for they treat names as non-indexicals, i.e., as expressions whose character is constant and whose content does not depend on certain parameters of the context, exactly like Kaplan does. Options (3) and (4), on the other hand, treat proper names as indexicals, i.e., as expressions whose character is context-sensitive, and thus are a far cry from Kaplan’s theory. Let us begin by examining option (1). Many people, including Kaplan, were convinced by Kripke’s causal-historical picture of how names refer. According to Kripke (1980), a given tokening of a name refers to the object it does because it is part of chain that goes back to an initial baptism, when the name was first introduced as the name of the baptized object. This name is passed on from speaker to speaker, and it is in virtue of being causally connected to the object itself that my utterance of that name refers to the precise object it does. In short, the reference is fixed externally, by the name’s ancestry, not by some fact internal to my cognition. It is also important to stress that, in this picture, names are individuated in terms of baptism ceremonies. So two coreferential names are the same name if and only if they were introduced in the same ceremony. Conversely, two coreferential names are different if and only if they were introduced to the linguistic community in two different baptisms. Hence, to each name corresponds one and only one causal-historical chain, which originates in a baptism ceremony. A consequence of this view is that, for example, a noun like ‘Ludovic’ will be systematically ambiguous: there is no single name ‘Ludovic’, but as many different names–spelled identically–as there have been baptism ceremonies. In fact, if the same person is named ‘Ludovic’ twice in two distinct ceremonies, there will be two distinct names spelt ‘Ludovic’, not a single one.6 Kaplan believes that the role of this causal chain is pre-semantic or, as he puts it, metasemantic (Cf. Kaplan, 1989b, p. 573). This means that the causal chain is not somehow built into the meaning of a given proper name; it functions only to determine which name is being used, and hence which thing is being referred to. So, on a given occasion of discourse in which 6 a name is used, contextual cues determine which causal chain is being exploited, and thus which name is being tokened, in the same way that contextual cues determine the meaning of an ambiguous expression such as ‘bank’. This is completely different from the way in which context determines the content of an indexical expression: for indexicals, context-sensitivity is built into their characters, and thus into their meanings, whereas names have a context-insensitive meaning. In other words, the referent of a name is not determined in virtue of an aspect of its meaning, but pre-semantically, by the chain that brought the name to the speaker. Context is relevant only in determining which chain is being exploited. The context does not include a parameter to which the meaning of the name is sensitive. In short, for Kaplan, causal chains merely fix the referent in Kripke’s sense; they are not encoded in the name’s meaning. Option (1) is the view that, contrary to Kaplan and Kripke, causal chains are in fact encoded in the name’s meaning. More precisely, they are encoded at the level of character. So, on this view, the character of a name ‘N’ would be given by a description such as ‘[dthat] the individual who lies at the other end of the historical chain that brought this token of “N” to me’. As Kaplan puts it, such a theory will “regard the historical chain theory as a part of semantics, as giving the meaning rather than as telling us how to discover it” (Kaplan, 1989b, p. 574). If this is plausible, then names would have two layers of meaning after all: the character, which is given by a description of the causal chain that introduced the name to the community, and the content, which is exhausted by its referent. This is why sentences like (2a), (2b), (2c), and (2d) have different epistemic profiles. Since the names ‘Freddie Mercury’ and ‘Farrokh Bulsara’ have clearly distinct causal histories, they have distinct characters. If they have distinct characters, and character is tied to cognitive value, then they have different cognitive values. Voilà: sentences (2a), (2b), (2c), and (2d) differ in cognitive value even though they have the same content. There are two main problems with this solution. The first has nothing to do with Frege’s Puzzle, however. It is about the cognitive role that causal chain descriptions supposedly play. The characters of indexicals are more intuitively conceived as linguistic meanings because they are, in some sense, grasped by every speaker of the language. And the descriptions that state these characters are fairly simple. Causal descriptions, on the other hand, are much more complex and seem much more cognitively demanding. It is not very intuitive to say that causal descriptions are rules that have to be mastered by competent speakers for the correct use of names: they require substantive beliefs about baptisms, causal connections, linguistic communities, etc. These sorts of beliefs may or may not be required for the linguistic practice in general. But so are beliefs about sounds, symbols, behaviors and This is essentially Kaplan’s example of the mischievous Babylonian. Cf. Kaplan (1990). Filosofia Unisinos – Unisinos Journal of Philosophy – 17(1):75-80, jan/apr 2016 77 Filipe Martone all sorts of beliefs about the world which are not built into meanings. The idea that the character of a name is a description of its causal chain, then, is somewhat implausible. Second, and more seriously, this proposal does not solve Frege’s Puzzle for names even if it is semantically plausible. Consider the following situation described by Kaplan: I may introduce a new proper name word and send it on its journey. When it returns to me–perhaps slightly distorted phonologically by its trip through other dialects–I can competently take it into my vocabulary without recognizing it as the very same word! Shocking! (Kaplan, 1989a, p. 563). If option (1) is right, there is just one baptism ceremony in this case, so there is just one name with a single character, call it ‘N’. When I encounter this name again, it may nevertheless be informative for me to be told that ‘N is N’, even though I was the one who introduced it in the first place. The name does not even have to be phonologically distorted; it can be spelt and pronounced in the exact same way as when I introduced it. This also occurs in Paderewski-like scenarios.7 It may be informative for me to be told that ‘Paderewski is Paderewski’ even though both occurrences of the name exploit the same causal chain and thus have the same character. How can this be possible? Option (1), then, seems barely tenable as a solution to Frege’s Puzzle. Option (2) seems a little more plausible, but it suffers from the same problem when it comes to explaining informativeness. This option presupposes much of the causal-historical picture of how names are individuated, but the descriptions that state the characters of names are much less cognitively demanding than descriptions of causal chains. They are metalinguistic descriptions like ‘[dthat] the bearer of ‘N’’. As seems clear, this description is a piece of knowledge that everyone acquires upon learning a new name, so they can more plausibly function as cognitively accessible characters. Note that in option (2) names are specific, i.e., they are individuated in terms of the baptisms by which they were introduced. As explained earlier, on this kind of view there is no single name ‘Ludovic’ with a single meaning, but many different names spelled identically, each with one single meaning. We can express this fact by subscripting the names: ‘Ludovic1,’ ‘Ludovic2,’ ‘Ludovic3,’ etc. So, for example, the character of ‘Ludovic1’ would be ‘[dthat] the bearer of ‘Ludovic1’’, which would be different from ‘the bearer of ‘Ludovic2’’, and so on. This guarantees that the description picks out the right individual, because the causal chain determines which specific name is loaded into the character, and the character then 7 determines the content. This explains why it can be informative to be told that ‘Ludovic1 is Ludovic2’: their characters are given by different descriptions, since different names occur in them. However, Paderewski cases are also left unaccounted for by this proposal. Since the causal chain pre-semantically individuates one single name, ‘Paderewski1’ (because he was baptized only once, presumably), the same characters would occur in both sides of ‘Paderewski1 is Paderewski1.’ This sentence should be trivial, but it is not. Let us now look at option (3). As mentioned above, this option treats names as indexicals. This is already very suspicious. As Kaplan says, “those who suggest that proper names are merely one species of indexical depreciate the power and the mystery of the causal chain theory” (Kaplan, 1989a, p. 563). But let us give it a shot. In this view, names are not individuated by baptisms. They are conceived as generic nouns. Thus, a generic name like ‘Paderewski’ (or ‘Ludovic’) will have a single, context-sensitive character–‘[dthat] the bearer of ‘Paderewski’’–which will refer to whoever is called ‘Paderewski’ in the context of its use. In other words, this option attributes to the generic name ‘Paderewski’ one univocal metalinguistic and context-sensitive meaning, which determines its content in a context. Generic names, then, function exactly like indexicals such as ‘here’ and ‘I’. Needless to say, this option is not very persuasive in its own right. Generic names do not appear to have meanings by themselves without being associated with a particular object; generic names are precisely that: generic8. Moreover, if there is just one single character for every possible occurrence of ‘Paderewski’, we will run into serious semantic problems. Suppose that there are two different men called ‘Paderewski’ in a given context. If I say ‘Paderewski is taller than Paderewski’, both occurrences of the generic name ‘Paderewski’ in that sentence will have the same character in the same context, and hence should determine the same content. In short, this sentence would refer to the same man twice, and thus would express a contradiction. This is obviously absurd, because what I said is perfectly reasonable, and may even be evidently true. To avoid this problem, we will have to say either that (a) the context shifts midsentence, guaranteeing that each occurrence of ‘Paderewski’ refers to a different individual, or (b) that the characters of each occurrence of ‘Paderewski’ are somehow different, so that different objects are determined by them; otherwise, this sentence will always come out false. However, if we maintain that (a) the context shifts midsentence while the character of ‘Paderewski’ remains the same, then there should never be any informative occurrence of ‘Paderewski is Paderewski’. In other terms, every possible occurrence of ‘Paderewski is Paderewski’ will trigger a shift in Cf. Kripke (2011) for his exposition of the famous Paderewski case. 8 If one defends the predicate view of names, as Burge (1973) and Fara (2015) do, then generic names do have meanings; they function like predicates that are true of objects that have those names (thanks to an anonymous referee for pointing this out). I find this position extremely counter-intuitive, but I will not argue against it here. In fact, I suspect that, even if such views are plausible, they would still suffer from the same kind of problem in accounting for Frege’s Puzzle as the views I am discussing. Filosofia Unisinos – Unisinos Journal of Philosophy – 17(1):75-80, jan/apr 2016 78 Character, proper names, and Frege’s Puzzle context; but, since the characters remain identical, every occurrence of this sentence should turn out trivial. But clearly there are non-trivial occurrences of ‘Paderewski is Paderewski’, as Kripke argued. In order to explain cognitive value in terms of character, there cannot be a difference in cognitive value without a difference in character. So, saying that the context shifts midsentence solves a semantic problem (it guarantees that the right individuals are picked out), but this view fails to account for Frege’s Puzzle. Alternatively, we can say that (b) the character of ‘Paderewski’ changes in each occurrence of the name, while the context remains the same. This would force us, however, to hold that characters are occurrence individuated, i.e., that each new occurrence of the name will produce a different character. Remember, this is required in order for the theory to deliver the correct result: if the characters in ‘Paderewski is taller than Paderewski’ are not individuated in terms of their occurrences, this sentence will always express a contradiction. But we cannot consistently and in a non-ad-hoc way maintain that characters are occurrence individuated just for these problematic cases. They are not semantically special in any sense. Hence, characters must be occurrence individuated for all cases. Note that this view delivers the correct result even for cases where two distinct occurrences refer to the same object. Suppose I do want to express a contradiction with ‘Paderewski is taller than Paderewski’. There is nothing semantically implausible about two distinct characters determining the same content in the same context, thus delivering the intended contradiction – just as I can say ‘I am taller than him’ while pointing to a mirror. This is perfectly in line with Kaplan’s framework: two different characters can determine the same object in the same context, but not necessarily so. If this is right, however, there should never be an uninformative case of ‘Paderewski is Paderewski’. Since every numerically distinct occurrence of ‘Paderewski’ would produce a different character, then they should always have different epistemic profiles. Yet this is clearly false, for there are trivial cases of ‘Paderewski is Paderewski’ (when Paderewski himself utters this sentence, for example). We also have no compelling semantic reasons to say that the characters of ‘Paderewski’ are identical in trivial cases and different in informative ones. Aside from considerations about cognitive value, those cases have no significant semantic differences that justify special semantic treatment. This leaves option (4). On this view, names function just like indexicals, but their characters vary from speaker to speaker or even for the same speaker in different occasions. Each speaker, then, attributes her own character to a given name in a given context, and this character determines the content. In a sense, this view is very similar to Fregean and Russellian descriptivism: each speaker associates one definite description with a name in a context of use, and this description determines the referent. Just like traditional descriptivism, in option (4) there is no single privileged description community-wise or even speaker-wise; as long as the referent remains the same, there is (apparently) no problem. The main difference is that, contrary to traditional descriptivism, these descriptions are not encoded in the propositional content of sentences containing names; they are turned into directly referential terms and are confined to the level of character. So, for instance, if I say ‘Socrates is a great person’ and associate the description ‘the greatest soccer player from Ribeirão Preto’ with ‘Socrates’, its character will be something like ‘[dthat] the greatest soccer player from Ribeirão Preto’ for me. The content, of course, would be just Socrates himself. Similarly, my audience may associate different descriptions, and thus different characters, with the same name. In other words, this view claims that characters vary contextually, based on the sort of information the speaker has about the referent and what sort of information is relevant in the context of communication. This is why ‘Paderewski is Paderewski’ can be informative to me: I associate different descriptions with each occurrence of the name, and thus the character of this sentence would be something like ‘[dthat] the pianist called ‘Paderewski’ is [dthat] the statesman called ‘Paderewski’’. The propositional content is just a self-identity, but the characters of the two occurrences of the name ‘Paderewski’ are different, so it is an informative self-identity. And this also explains why the same sentence can be trivial: if I associate the same character with both occurrences of ‘Paderewski’, then ‘Paderewski is Paderewski’ will turn out to be uninformative to me. In sum, then, this view claims that the characters of names are only determined in the speaker’s idiolect by the information she has, not by general rules of language. Moreover, proper names turn out to be mere placeholders for definite descriptions, for they do not have constant meanings at the level of character, much like the dthat operator itself. Because this view is so similar to Fregeanism and Russellianism about proper names, they solve Frege’s Puzzle more or less in the same manner. Additionally, this view manages to avoid Kripke’s modal arguments, because the description associated with a name is not expressed at the level of content. These descriptions are also not merely rigidified: they are turned into genuine directly referential expressions by something like the dthat operator. Finally, contrary to Kaplan’s theory, characters of names will no longer be constant functions, but merely functions from contexts to contents just like any other indexical. However, though it avoids Kripke’s modal argument, option (4) does not avoid his semantic argument. If it is my job to associate a description, and thus a character, with a name on an occasion of use, then my utterance will determine the wrong object if I associate the wrong information with the name. Think of the name ‘Einstein’. If I associate the description ‘the father of the atomic bomb’ with it, then its character will be ‘[dthat] the father of the atomic bomb’ and will pick out whoever satisfies this description. This character, of course, does not determine Einstein, but Oppenheimer. Therefore, the proposition I express will be a proposition about Oppen- Filosofia Unisinos – Unisinos Journal of Philosophy – 17(1):75-80, jan/apr 2016 79 Filipe Martone heimer, and not about Einstein. This, of course, runs counter to well-established arguments about reference-determination for proper names. Moreover, we intuitively say that I have false beliefs about Einstein, not that I am thinking and saying true things about Oppenheimer. If characters of proper names are determined internalistically, in my idiolect, then it is always possible for my utterances to determine the wrong objects if I have mistaken beliefs about them. Additionally, this picture makes communication very mysterious. If the characters of names are potentially multiplied by all the speakers in a given situation, how do we explain their grasp of what is said? If there is no significant overlap between the information speakers associate with the name uttered in a given occasion, then the route to reference will be so different for each of them that it seems very hard to explain how they arrive at the same content and know that they do so. In fact, it seems rather miraculous. As with indexicals, we intuitively say that it is the job of the speaker to exploit a single character that must be grasped and interpreted by the audience. In short, reference seems to be a two-place relation between speaker (or expression-in-context) and reference, not an n-place relation between every single person in a communication exchange and the referent. Of course, we may associate a huge body of information with a given name, but to claim that we also express or display this information through an utterance of this name, even if this is done via character, is hard to swallow. Worse yet, if I have conflicting individuating information about a given object, I refer successfully to it only if I happen to associate the right description when I use it; otherwise, I will refer to something else. Not even my own uses of the name will be consistent, on this view. It seems, then, that all options for explaining the name version of Frege’ Puzzle in terms of character are flawed in some sense. They either fail to account for cognitive value or are semantically implausible. This brings us back to the point I raised at the beginning of the paper: if character is unable to account for the cognitive value of proper names successfully, then we have serious reasons to suspect that it also fails to account for the cognitive value of indexicals. Kaplan’s solution to the indexical version of Frege’s Puzzle certainly seems plausible, but close scrutiny may reveal that the apparent relation between character and cognitive value is only accidental. But this is a matter for another investigation. References BURGE, T. 1973. Reference and Proper Names. Journal of Philosophy, 70(14):425-439. http://dx.doi.org/10.2307/2025107 FARA, D.G. 2015. Names are Predicates. Philosophical Review, 124(1):59-117. http://dx.doi.org/10.1215/00318108-2812660 FREGE, G. 1960. Translations from the Philosophical Writings of Gottlob Frege. Oxford, Blackwell, 244 p. KAPLAN, D. 1978. Dthat. In: P. COLE (ed.), Syntax and Semantics. Cambridge, Academic Press, p. 221-243. KAPLAN, D. 1989a. Demonstratives. In: J. ALMOG; J. PERRY; H.K. WETTSTEIN (eds.), Themes from Kaplan. Oxford, Oxford University Press, p. 481-564. KAPLAN, D. 1989b. Afterthoughts. In: J. ALMOG; J. PERRY; H.K. WETTSTEIN (eds.), Themes from Kaplan. Oxford, Oxford University Press, p. 565-614. KAPLAN, D. 1990. Words. Proceedings of the Aristotelian Society, 64(suppl.):93-119. http://dx.doi.org/10.1093/aristoteliansupp/64.1.93 KRIPKE, S. 1980. Naming and Necessity. Cambridge, Harvard University Press, 172 p. KRIPKE, S. 2011. A Puzzle about Belief. In: S. KRIPKE, Philosophical Troubles. Collected Papers Vol. 1. New York, Oxford University Press, p. 125-161. http://dx.doi.org/10.1093/ acprof:oso/9780199730155.003.0006 SALMON, N. 1986. Frege’s Puzzle. Cambridge, MIT Press, 194 p. Filosofia Unisinos – Unisinos Journal of Philosophy – 17(1):75-80, jan/apr 2016 Submitted on October 3, 2015 Accepted on March 13, 2016 80