The Two-Component Theory of Proper Names and Kripke’s Puzzle






JeeLoo Liu




[Online Version Only]




The Two-Component Theory of Proper Names and Kripke’s Puzzle


Kripke's puzzle is an old and familiar story.  It was put forward in Kripke's 'A puzzle about Belief.'[1979]  But even today it still has such a charm that people are drawn to it time and time again.  In this paper I shall use his puzzle as the stepping stone for developing a new description theory of proper names.   

      Kripke tries to defend his direct reference theory against the charge that it cannot explain the role of proper names in an epistemic context (such as belief, thought, etc.).  There are many famous puzzles involving substitution salva veritate for different names of the same referent, and the description theory can easily dissolve them by suggesting that different names have different senses.  These puzzles were considered to be defeating the direct reference theory of proper names.  Kripke thus tries to demonstrate a similar puzzle that does not involve different names, and thus does not involve different senses.  Using his principle of disquotation and principle of translation,[1] Kripke presents a puzzle which involves a Frenchman Pierre who is attributed the following set of beliefs:

            (1)       Pierre believes that London is pretty.

            (2)       Pierre believes that London is not pretty.

According to Kripke, the two belief reports attribute a contradiction to Pierre, even though Pierre himself cannot be interpreted as being inconsistent.[2]     

      Kripke also discusses another puzzle which invokes only the principle of disquotation and no translation is involved.  This is the example of Peter’s two beliefs concerning the politician/musician Paderewski.  In this case, we get a similar set of contradictory belief reports:

            (3)       Peter believes that Paderewski has musical talent.

            (4)       Peter believes that Paderewski has no musical talent.[3]

Kripke thinks that these puzzles generate the same difficulty for both the direct reference theory and the description theory.  The conclusion he draws from these puzzles is that they reveal a general feature of belief contexts that such contexts resist substitution, and the failure of substitution has no bearing on whether one adopts a direct reference theory or a description theory. 

      There are numerous approaches in dealing with Kripke's puzzle:[4]

1. Stopping the generation of the puzzle: One could reject one or both of Kripke's principle of disquotation and principle of translation, so as to terminate the generation of these puzzling cases.[5]

2. Biting the bullet: One could simply accept the verdict that Pierre and Peter have inconsistent beliefs and argue that we all do, thereby showing that the puzzle is no puzzle at all.[6]

3. Dissolving the puzzle: One could give proper names a different analysis so that the puzzle gets dissolved under this new analysis.

      My approach is of the third kind.  Following Marcus and Katz, I argue that Kripke's puzzle applies only to a direct reference theory such as his own.[7]  There are, of course, other versions of the direct reference theory that may avoid generating this kind of puzzle.  The new direct reference theorists (such as Nathan Salmon, Mark Richard and Gareth Evans) incorporate some elements of the description theory into their direct reference theories.  What I am developing in this paper, on the other hand, is a  new description theory of proper names which incorporates some elements of the direct reference theory into the description theory.  I shall also explain why we should have a description theory rather than a direct reference theory, even though the two sides are meeting in the middle ground.  Since the decline of the description theory of proper names follows from Kripke's attack, my paper will treat Kripke's numerous criticisms of the description theory as the main challenge for my new description theory.  In what follows I will first briefly explain why Kripke's theory of proper names does not give us the whole story.  I will then introduce my theory which I call the two-component description theory of proper names.  My proposal will be based on the rejection of the commonly assumed sharp separation between semantics and pragmatics.  Using some of the familiar cases Kripke sets up against the traditional description theories, I will explain how my theory gives a different story.  Finally, I will go back to Kripke’s puzzle and show how my theory can avoid attributing a contradictory set of beliefs either to Pierre or to Peter, and thereby dissolve the puzzle that Kripke poses for the description theory.  


§ I. The Insufficiency in Kripke's Theory of Proper Names

For Kripke, proper names are 'rigid designators' in the sense that they designate the same individuals across possible worlds.  However, what a theory of names should explain, first and foremost, is not how reference gets fixed across possible worlds, but how reference gets fixed in our actual world.  I think Kripke gives us too simplified a story in the latter respect.  For example, Kripke talks about 'Nixon' as being fixed in our world.  Kripke seems to assume that we all know which Nixon it is, to whom we then assign all the possible situations.  However, it is not the case that if one simply mentions the name 'Nixon,' the name itself will do the job of getting the correct person being discussed.  Suppose someone names his dog after the former president Nixon.  One day the dog owner is on his way home, and his neighbor, who takes interest in politics, informs him: 'Nixon is dead.'  In this case the dog owner would most likely take it to mean that his dog Nixon had died.  Some explanation is required in this context to reveal the fact that it was the former president Nixon, not the dog Nixon, who died that day.  The reason why proper names alone are insufficient is that there might be, and in fact usually is, more than one individual who is called by that name.  If a proper name used in our world can have more than one bearer, which is the one we single out in our discourse?   

      Kripke appeals to a causal chain picture as a support for his direct reference theory.  He says, 'Someone, let's say, a baby, is born; his parents call him by a certain name.  They talk about him to their friends.  Other people meet him.  Through various sorts of talk the name is spread from link to link as if by a chain.' [1972, p. 91]  According to such a picture, names are supposed to designate the one originally dubbed by that name, and as the name passed on to us through a chain of communication, we intend to use the name to refer to the same person referred to by the previous user of the name.  Thus, 'a speaker who is on the far end of this chain' is able to, simply by virtue of using the name, refer to the individual initially dubbed that name.  The problem with such a picture is that there might be multiple causal chains from one name leading back to different objects named.[8]  In most cases of proper names, especially when only the first or the last name is used in the discourse, there are parallel causal chains linking different people or objects to their identical name.  From one name mentioned, it might not be so clear which one of the objects named was at the beginning of this causal chain of communication.  In Kripke's causal theory, communication depends on the audience's intention to use the same reference as the speaker does.  But such an intention does not guarantee success. 

       Suppose someone visits a museum and the tour guide introduced a painting as one done by Giacometti.  Since the visitor only knows of one Giacometti, the one who made those slender-shaped figures (i.e. Alberto Giacometti), the visitor thought that this painting was done by him.  However, the painting was actually done by Gustav Giacometti, the sculptor’s father.  When the tour guide announced:

            (5)       This was painted by Giacometti,

her utterance was associated with Gustav Giacometti.  On the other hand, if the visitor tries to report this information to a friend by uttering (5), the utterance would be associated with Alberto Giacometti.  Communication via causal chain goes astray in this case.

      What the above example shows, is that the occurrence of a proper name in daily discourse does not automatically reveal the chain(s) behind the usage of that name.  Whether it does or not would depend on how many possible candidates there are.  When there is more than one possible referent of the name, change of speaker associations may take place from one speaker to the next.  In order for the audience to fix the right one that bears the name, some other mechanism is required.  What we need, is a description of the speaker's associations of the name, while the causal chain itself is insufficient to secure successful communication.  This is where the call for a description theory comes in.


§ II. The Two-Component Description Theory of Proper Names

What I propose here is a two-component description theory of proper names (or, TCD).  I think two questions should be separated: 'What does the name mean?' and 'How does the name refer?'.  According to TCD, descriptions are associated with the use of proper names in two ways: One is the description that gives the meaning of the name, such as 'an individual called such and such (by a certain linguistic community).'  The other is the set of descriptions that the speaker would use, if asked, to specify the intended referent.  These two descriptions compose the sense, or the semantic value, of a proper name.  We shall treat the two kinds of descriptions as an ordered pair:


[P]       <'an individual called "F" (by a certain linguistic community),' F = a set of descriptions>

The first description determines the denotation of the name.  The set of descriptions F, on the other hand, fixes the reference of the use of a name by that speaker.  I distinguish 'denoting' and 'referring' in roughly the following way:[9] denoting is a semantic relation; it is something that a name does, referring is a pragmatic relation; it is something that the speaker does.[10]  The denotation of a proper name is a set, which consists of members whose qualifying property is that they are all called by that name (by a certain community).[11]  The denotation sets the range of possible referents, and the speaker's associated descriptions get us to the particular referent within this range.  Now I shall discuss these two components separately.

The First Component   The first component of the descriptive sense of a proper name is its meaning, which is different from the sense.  In contrast to Bach's analysis, I argue that the meaning of a proper name is not analyzed as a definite singular term 'the one called such-and-such' but as an indefinite singular term 'an individual called such-and-such (by a certain community).'[12]  For instance, the meaning of the name 'Nixon' is 'an individual called "Nixon",' and the denotation of the name is the set of all members called 'Nixon.'  Since a name is generally used to designate one member of the denoted set, not the whole set, we should further distinguish the denotation of a name itself and the denotation of the name in use in the following way: the denotation of the name 'F' in the language is a set, call it l, and l = {all individuals called 'F'}, while the denotation in each particular use of 'F' is one member of l.[13]

      In the above analysis of the meaning of proper names, 'the calling relation' and the parentheses '(by a certain community)' both need some explication.  The calling relation is not merely established through one's being addressed to in a certain way.  'Being called as' is also one's being mentioned as, one's being referred to as, one is being introduced as, one's being spoken of as, etc.[14]  Someone's shouting 'Hey, you!' at someone else certainly does not establish 'Hey, you!' as the name for that other person.  Furthermore, such a calling relation is established through a communal act; it is what the linguistic community jointly has done that establishes the use of a proper name.  For the purpose of communication, the use of the name cannot be restricted to only one person (even though it could be stipulated by one person initially).  For instance, one may be directly addressed to by the title 'Sir' by some people, but one would not normally be called by and mentioned as 'Sir' by all members in the community.[15]  This added condition ('by a certain community') rules out such calling relations as a naming relation.  There are, however, no strict linguistic rules, other than conventional usage, governing the use of names.  'Seven' could be a name; 'Moon Unit' could also be a name.  Even a title could become a name when it is used by a community as a name.  'Jack the Ripper' is an example at hand. 

      Furthermore, my theory proposes an indexical treatment of the usage of proper names; that is, it treats names as indexed to a certain linguistic community.  Kripke's causal chain picture gives us a model explaining how an individual falls into the denoted set of a proper name, but one amendment needs to be made to that picture: The community involved does not have to be the original community to which the object named belongs.  When we are dealing with proper names in English, the English speaking community is the linguistic community relevant to the case.  A proper name, take 'Socrates' for example, should be given the following analysis:

            [S]       'Socrates' = 'an individual called "Socrates" by our linguistic community'

This treatment does not rely on the fact that Socrates was called 'Socrates' in his times since it is pretty clear that he was not.  'Socrates' is a name we use for him in English.  Names used in English, however, are not necessarily English names.  'Mitterand' and 'Chirac' are good examples.  Proper names have an interesting status in that they are sometimes not translated and are directly used in a different language.  But there are no hard and fast rules about translation for proper names.  Many names are not translated (such as different family names in Latinate languages) while many are (such as names of places or first names like 'Peter' and 'Pierre,' 'John' and 'Jon,' etc.).  Instead of concluding, as some do, that names are thus not part of a language, I argue that names are indexed to the language and it is the language users who determine whether those names are to be translated, or to be incorporated into the language as such.  The key point of this analysis is the indexical term 'our.'  'Our' is indexical to the user of the name, not fixed to the present writer and readers of this paper (that is, not fixed to English).[16]  Our analysis preserves this 'semi-independent' status of proper names: they can either be translated or directly used in a different language.  When they are translated, the translated names become the names for the individuals in the new language.  And when they are directly used, those original names are the names for those individuals in the new language as well.  Even if the present linguistic community is not the one that initiated the calling relation, as long as the name in the initial language is preserved in the present language, we can say that the object named is called such-and-such by our present linguistic community.[17] 

      Here I should explain why I think that the calling relation is a better analysis than similar meta-linguistic analyses such as 'being named such-and-such' or 'the bearer of a name.'  The analysis 'being called such-and-such' is better than the analysis 'being named such-and-such,' because the naming relation is usually based on the original naming ceremony, whereas the calling relation has a broader application.  First of all, the calling relation is defined in terms of the present linguistic community, while the naming relation is defined in terms of the original community.  Naming presupposes an initial dubbing in the presence of the object named.  The naming relation would imply that we can only talk about objects that were named in our linguistic community.  Kripke assumes that the usage of a name is passed on through a causal chain to refer back to the individual originally dubbed.  But it could very well happen that somewhere in the causal chain an error occurred, and a historical person (or a remote object) that is called by our name was not originally so called.  With a historical figure, for example, it might well have happened that somewhere in the historical chain there is a translation or even a mis-translation of the original name, such that the name which certain later communities (such as ours) come to use is no longer the same name given to the individual by the original community.[18]  The same could happen to the name-passing chain of any remote object.  Searle [1983] gave us an example presented initially by Gareth Evans.  Searle writes, '"Madagascar" was originally the name of a part of Africa.  Marco Polo, though he presumably satisfied Kripke's condition of intending to use the name with the same reference as "the man from whom he heard it," nonetheless referred to an island off the coast of Africa, and this island is now what we mean by "Madagascar".'[p. 237]  Furthermore, there are also cases of broken causal chains of names, such that we would have to conjure up a name to accomplish referring.  For instance, we call the prehistoric woman whose remains were recently found 'Lucy' even though it is certain that she was not named so initially.  When anthropologists talk about Lucy, they are talking about that woman.  With all these cases, the calling relation allows the names to be legitimate names.  As long as our linguistic community intentionally use the name to refer to someone or something, the name is established for the object named.

      Secondly, as should be obvious now, the calling relation is broader than the use of genuine proper names (i.e. the name with which one is dubbed at birth), since the calling relation would include things like nicknames, pen-names, assumed names, aliases, etc.  The individuals may get certain names even though they were not originally named such and such; one can come to have a name if one is being called by that name by a certain community.  For instance, Marilyn Monroe was not baptized 'Marilyn Monroe.'  Each calling relation is relative to a certain community.  For the community prior to the time when she adopted her alias, 'Marilyn Monroe' was not her name.  But in our linguistic community nowadays, 'Marilyn Monroe' was her name.  The use of names is thus relative to the community in question.  By talking about the calling relation instead of the naming relation, I intend to capture this wider range of names.[19] 

      The bearer analysis has the same advantage over the naming analysis as the calling analysis does.  I don't have any strong objection to that analysis.  My only complaint for the bearer analysis is that it still does not tell us how a person becomes the bearer of a name. 

The Second Component   Taken by itself, a name (in each use) only signifies an indeterminate object that is a member of the denoted set.  But in our daily discourse, a proper name is always associated with one particular object.  I argue that proper names cannot single out particular objects without the speaker's intention, which is specifiable by a set of descriptions.  The second component of the sense of a proper name is this set of descriptions in virtue of which the speaker accomplishes reference to a particular agent.  The importance of the speaker's intention in aiding a proper name to fix the reference is especially obvious in cases of first name reference.  At a party where there were four people by the first name 'Mike,' a confusion can easily arise.  But when the hostess called out from the kitchen: 'Mike, do you want to go potty now?', the situation is not confusing even though it may be amusing in some way.  Everyone present should understand that she was addressing her three-year-old son Mike. 

      More should be said about how this set of descriptions aids in securing reference.  A proper way to understand how the associated descriptions fix the reference, is to take them to express the speaker's beliefs of the object.  The speaker may believe them to be true, but some of the beliefs may turn out false.  Even if some or all of the beliefs are false of the referent, they are nonetheless beliefs about that referent.  This is the speaker's intentionality at work.  In general cases of name-using, the speaker has something/someone in mind, and intends to refer to this particular object by the use of the name.[20]  The speaker does not need to have direct acquaintance with the object in order to have some beliefs about it.  Naturally, these beliefs may be perceptual beliefs, and thus what the speaker would describe are based either on his/her mental images (Cindy - 'the woman with red hair') or on the conditions under which he/she established physical contact with the object (George - 'the man whom I met yesterday').  However, the descriptions could also be based merely on knowledge by description.  In those cases the descriptions would describe a piece of information the speaker previously acquired of the object (Plato - 'the one who wrote The Republic'); or the circumstances under which the speaker acquired that information (Farrakhan - 'the one I read about in the newspaper').  Therefore, speaker-associated descriptions do not merely describe properties the object presumably has; they describe also the conditions under which the speaker came to know about that object.  Even if the descriptions of the speaker's beliefs may be false of the object, the descriptions of the epistemic conditions would establish some "parasitic links" between the speaker and the referent.  This is how F fixes the reference of the utterance. 

      The direct reference theorists often employ the same criticism Kripke has against the description theories:  It may happen that Moses never went into politics; it may happen that Socrates was not snub-nosed; it may happen that Plato was not Aristotle's teacher, it may happen that Gödel did not prove the incompleteness of arithmetic, etc..  In other words, it may happen that all of the beliefs that we as a linguistic community have about the referent are false, and thus even parasitically the speaker cannot refer to the right object since no one would satisfy those descriptions.  However, I think such a criticism commits a fallacy of conflating actuality with possibility, and conflating knowledge with beliefs.  What we say about Socrates, Plato, etc. is based on our knowledge of these people in the actual world.  Of course it could have happened that our knowledge is just a bunch of false beliefs, and that those possible scenarios could have happened in our actual world.  But these are "counterfactual" considerations.  To criticize the description theories in this way, is to demand an absolute warranty of our knowledge of the actual world.  I can also turn the table around and demand such a warranty in the causal chain to which the direct reference theorists appeal.  In general cases of knowledgeable name using (barring those counterfactual scenarios), some of the speaker's associated descriptions should be satisfied by the intended referent.  According to TCD, we refer to one particular member of the denoted set of the name, because that one satisfies the other descriptions we have in mind about him/her.  But in cases when no one in the denoted set satisfies any of the descriptions, the speaker simply fails to refer by using the name. 

      If the set of descriptions (the F) includes both descriptions of the speaker's beliefs and descriptions of the epistemic conditions of the speaker's coming to have those beliefs, then it is a rather large set.  Only in rare cases would the F be an empty set.  Suppose that the speaker picks up the name from a party conversation without knowing anything about the referent, he/she would at least associate the description such as 'the person whom they were talking about,' 'the person who did such and such according to them,' etc. with the use of the name.  As long as the context is “informative,” one can always acquire new information concerning the referent.  Even when one forgets the complete context in which one acquires the use of the name and fails to recall any information concerning the referent, one would have a minimal description such as 'is a person,' 'is a city,' etc.  If the speaker in using a name fails even in providing those minimal descriptions, then the description he/she associates would be <'an individual called "F",' F = 0>.  In this case I would say that the speaker is using the proper name attributively in Donnellan's terminology and he or she thereby fails to refer.[21]               

      One might object that if the F is based on individuals' beliefs about the object, then there is no guarantee that we can ever talk about the same thing since we might all have very different beliefs.  When the associated descriptions include the epistemic conditions of the speakers, the problem is even more pronounced: almost no one shares the same epistemic conditions.  How does this affect the analysis of proper names given by TCD?  My answer is that the meaning of proper names in different people's utterance is the same: it is 'an individual called such-and-such.'  The semantic value of a proper name in different utterances, on the other hand, would be different from person to person.  Semantic value corresponds to cognitive value.  This difference in semantic, and thus cognitive, values explains why one would take 'Cicero is Tully' to be a trivial statement, while another would take it to be informative.

      Another problem for TCD is this:  Even though there are many 'Nixons' in our linguistic community, there is only one Nixon to whom we refer by that name when we talk about Nixon in the present English.  How does this happen?  Here I wish to introduce a pragmatic notion 'the realm of discourse.'  A realm of discourse is defined as the set of things discussed by (thought by) a group of people that engage in the discourse.  If an individual enters a public realm of discourse, then this group of people would usually employ the name in the same way.  For instance, even if it is quite likely that there were many people called 'Socrates' (in Greek) around the time that our Socrates was living, he was the only one who was significant enough to enter the realm of our public discourse.  Thus if you look up the word 'Socrates' in an encyclopedia, the descriptions you get would be of this particular Socrates.  This does not mean that these descriptions are synonymous with the name 'Socrates;' it only means that the reference of 'Socrates' is generally fixed in our discourse.  But the fixing is not done by the name itself, it is done by the general intention of the participants of the discourse.  The use of a public name is intended to refer to that particular individual, and the Fs different people associate with the use of the name will be largely the same.  There is no mystery in how different people come to share largely the same F: we learn about the world through interaction with others.  As a result, in cases of public discourse, proper names are used as if  they were singularly denotative.  It is not because these names really denote a singular object, but because the participants share the intention to refer to the same individual.  This is why we can talk about Socrates, about Nixon, or about Moses without any other specification.  It is not, however, a linguistic rule that we have to talk about this Socrates, this Nixon or this Moses.  It is rather a rule of pragmatics.  A. P. Martinich [1997] defines 'pragmatics' in this way: 'Pragmatics is the study of how language is used.... Pragmatics focuses on the interaction between speakers and hearers.  The major idea that guides research in this area is that speaking is intentional behaviour and governed by rules.'[p. 12]  Rules of pragmatics depend on the context and the intention of speakers and hearers.  Under the present theory, it is not a wrong use if one uses a certain name, which normally picks up one particular member in the name-set, to pick out another member.  It would be a wrong use, on the other hand, if one uses a certain name to pick out someone by a different name. 

      To recap, according to TCD, the sense (or the semantic value) of a proper name should include two components, one determines the name's denotation and the other determines the speaker's reference by using the name.  I call the first component the meaning of the proper name; the second component, the associated descriptions of the proper name.  Anyone who satisfies the same calling relation (namely, being-called-by-the-same-name) can be qualified as a member of the same set.  The denotation of a proper name is thus a set, which could have multiple, single, or even no members (if the name is an empty name).  The referent of a proper name, on the other hand, is the particular member of that set, the member which is being picked out by the intention of the speaker.[22]  The speaker’s intention is expressible by descriptions of her beliefs, her mental images, her epistemic relationship with the object, etc.  The intention with the associated descriptions “fix” the reference for the speaker’s utterance, and thus there is no ambiguity of a proper name in an utterance (as long as the speaker knows what she has in mind). 

      Under TCD, reference is not direct; rather, reference is mediated through the speaker’s associated descriptions.  This feature separates TCD from any form of direct reference theory.  Let us now turn to the issue of indirect reference.  The ordered pair [P] should determine the actual referent, the a, of each utterance.  The first component of [P] describes any indeterminate member of the denoted set l = {all individuals called 'F'}.  The second component of the ordered pair picks out that particular member, the a, and a belongs to l.  Thus a should be a member of the denoted set that is being singled out by the speaker's associated descriptions.  Reference cannot be successful without using the right name.  As Burge [1973] remarks, 'A proper name occurring in a sentence used by a person at a time designates an object if and only if the person refers to that object at that time with that proper name, and the proper name is true of that object.'(my italics) [p. 435]  This is simply the linguistic rule of name-using and the social habit of following that rule.[23]  For instance, if I intend to refer to Plato by using the name 'Aristotle,' even if I associate all the right descriptions (such as the one who wrote the Republic, etc.), I do not refer to Plato by that name.  One may argue that just as Donnellan can use descriptions that don't match to “refer to” a particular individual (what he calls the referential use of descriptions), we can also use names that are not true of the object to refer to a particular individual.  What matters, the line of argument goes, is what the speaker intends to refer to.  Kripke seems to have taken this line of argument.  He gives the example in which two people see Smith in the distance and mistake him for Jones.  One person asks: 'What is Jones doing?,' while the second person answers: '(He is) raking the leaves.'  According to Kripke, even though the name 'Jones' 'never names Smith,' 'in some sense, on this occasion, clearly both participants in the dialogue have referred to Smith.'[24]  My reply is that there is a sense in which we say that the first speaker succeeds in referring, but the speaker does not use the name to refer in this case.  What the speaker does instead, is to refer by means of other contextual expediency (such as pointing, gazing, etc.).  The use of the name 'Jones' plays no significant role in the referring act.  We should thus analyze the Jones-Smith scenario in the following way:


            (6) a = <--- , F = {the person whom we are looking at now}>, and a is raking the leaves.

In other words, the only means the two speakers actually use to accomplish their referring to Smith is their contextual relationship to Smith.  The name 'Jones' used in this context is simply a "misnomer."

      Truth values are assigned to the utterance of a sentence, or we can say, to the proposition expressed by an utterance.  According to Strawson [1956], sentences themselves have meaning but no truth value, and yet we can make use of a sentence to 'express a true or false proposition.'[p. 223]  As he says, 'Meaning is a function of the sentence or expression; mentioning and referring and truth or falsity, are functions of the use of the sentence or expression.'[p. 224]  A sentence of the type 'Nixon is dead' in itself cannot be assigned a definite truth value, since it means something such as 'Someone called "Nixon" is dead.'  In our normal usage where we talk about Richard Nixon, the utterance is true after April of 1994.  But the utterance 'Nixon is dead' by our dog-owner's neighbor would be true only if the intended referent is the former president Nixon, and would be false if the intended referent is the dog Nixon.  The need to assign different truth values to different utterances of the same sentence shows that after the ordered-pair of descriptions fixes an a for an utterance, it is this a that we evaluate when we assign a truth value to that utterance.  The semantic content of the proper name is incorporated into the semantic content of the sentence in which the name appears, through the identification (via descriptions) of the a in question.  Take 'Socrates is wise' for example.  Our analysis of its truth value will be rendered as such: 'Socrates is wise' is true iff an individual called 'Socrates' (by our linguistic community) is singled out in virtue of a set of descriptions in the mind of the speaker, and this individual is wise.[25]

      It is not the case that all utterances can be successfully assigned a truth value.  There are cases when the set of descriptions cannot generate any real a, such as in the case of empty names.  If the fictional name 'Worf,' for example, has never been used by any actual person, then l = {all individuals called 'Worf'} is an empty set.  Most cases of fictional names, however, do not belong to this category.  Santa Claus does not exist, but there is a town called 'Santa Claus.'  There is no Pegasus, but there might be companies named 'Pegasus.'  I argue that the a doesn't get fixed in these cases not because l is empty, but because F fails to pick out any member in l.  In cases where there are entities called by that name, but the speaker's descriptions fail to pick out any member of the set, the reference is vacuous and the name is empty.  Under TCD, a sentence containing an empty name is not meaningless.  In Kent Bach's words, 'reference failure does not lead to loss of meaning.'[1984, p. 174]  Depending on the scope reading, in some cases such a sentence would be false while in others it would have no truth value.  

        Previously I have distinguished meaning and sense.  Now based on what I have said about the assignment of truth values, I wish to introduce a third notion: content.  The content of a proper name is the referent mediated by the ordered pair (the two components) of descriptions:

[Q]      a (<'an individual called "F",' F = a set of descriptions>)

where a is the object referred to, 'F' is the name the speaker uses to refer to a, and F is the set of descriptions the speaker would use to specify a.[26] 

            Using [Q], the above example of 'Nixon is dead' is thus analyzed as (7) and (8):


(7)    Nixon (<'an individual called "Nixon",' F = {is a man, is a former U.S. President, has a large nose and sad-looking eyes,...}>) is dead.


(8)    Nixon (<'an individual called "Nixon",' Y = {is a dog, has a large nose and sad-looking eyes,...}>) is dead.

      With the Giacometti case mentioned earlier, TCD would fare much better than Kripke's theory.  Our analysis would render the tour guide's remark as (5'):


            (5')      This was painted by Giacometti (<'an individuals called "Giacometti",' F = {a Swiss painter working in the late 19th Century, etc.}>).

On the other hand, when the museum visitor also utters (5), his utterance should be analyzed as (5"):


(5")     This was painted by Giacometti (<'an individuals called "Giacometti",' F = {a Swiss sculptor who made slender-shaped figures, etc.}>).

I think in this case the tour guide and the visitor use the same sentence-type, which can be analyzed as meaning that the painting was done by someone called 'Giacometti.'  However, their utterances are not the same in the way that the tour guide refers to Gustav Giacometti and she is right, while the visitor refers to Alberto Giacometti and he is wrong. 

      This shift of intention also poses a problem for Kripke's theory concerning speech reports, while TCD can handle this sort of problems easily.  Suppose the visitor says

            (9)       The tour guide said this was painted by Giacometti,

he report a different content of her utterance even though he uses the same words she used.  Under Kripke’s theory, the visitor would be referring to Gustav Giacometti since he did intend to use the name the tour guide did.  Therefore what he utters would be true as well.  Consider the fact that the visitor knows nothing about Gustav Giacometti and his associated descriptions actually pick out the-Giacometti-who-made-those-slender-shaped-figures, it does not seem correct to say that he is referring to Gustav Giacometti simply because he has heard the name 'Giacometti' from someone who did refer to Gustav.  On the other hand, TCD can allow us to assign different propositional contents as well as different truth-values to the utterances made by different speakers.  There are conceivably many other cases where intentional fixing changes from one speaker to the next.  Kripke’s causal theory of reference fails to explain these cases.

      I now want to show how TCD deals with some of the problems Kripke presents as a refutation of the description theory.  One of Kripke's attacks focuses on William Kneale's description theory of names.  Kneale's theory is that the meaning of a name is simply 'the individual called by that name.'  Kneale argues that statement (10) is trifling or non-informative:

      (10)           Socrates was called 'Socrates.'

If (10) is trifling, then it must be because the name 'Socrates' itself has already included the information given by the predicate.  Therefore, Kneale concludes, 'Socrates' means 'the individual called "Socrates".'

      Kripke [1972] rebukes this argument by pointing out that (10) 'isn't trifling on any view,' because it could happen that the Greeks didn't call Socrates 'Socrates.'[p. 69]  I agree with Kripke on this point.  But an important feature in (10) is the past tense verb ('was') used by Kneale.  I think if (10) is stated as

      (11) 'Socrates is called "Socrates",'

then that statement is trifling or non-informative.  Under (11), Kneale's argument could support the meta-linguistic analysis of the meaning of 'Socrates.' 

      How do we explain the difference between the triviality of statement (11) Socrates is called 'Socrates' and the non-triviality of statement (10) Socrates was called 'Socrates.'?  Under my theory, the calling relation is fixed to the present language used, thus (10) as analyzed in the following way is not trifling:


(10')    An individual called 'Socrates' by the present English-speaking community was also called 'Socrates' by the ancient Greeks.

And (11) is analyzed in this way which clearly shows how it is a non-informative statement:


(11')    An individual called 'Socrates' by the present English-speaking community is called 'Socrates' by the present English-speaking community.

This shows that Kneale is partially correct: the meaning of the name 'Socrates' does include a piece of meta-linguistic information.

      Kripke's second criticism of Kneale's description theory is that it violates what he calls 'the non-circularity condition.'  The noncircularity condition stated by Kripke is this:


(C) For any successful theory, the account must not be circular.  The properties which are used in the vote must not themselves involve the notion of reference in a way that it is ultimately impossible to eliminate.[1972, p. 68]

What is barred by (C) is a circular theory of reference, which uses the notion of reference itself in defining the way to fix the reference.  Kneale’s theory of proper names, at least as Kripke interprets it, uses 'the individual called such and such' both as the meaning of 'Socrates' and as a way of referring to Socrates.  Kripke [1972] argues: 'Obviously if the only descriptive senses of names we can think of are of the form "the man called such and such,"...then whatever this relation of calling is is really what determines the reference and not any description like "the man called 'Socrates'".'[p. 70]  That is to say,  if the description one uses to refer to Socrates is simply 'the individual called "Socrates",' then anyone can call x by such and such a name and then refer to x by using that name.  In this criticism, Kripke seems to take the word 'calling' used in this context to be expressing the same notion as 'referring,' and he thus charges Kneale's theory with the violation of the non-circularity condition.  However, under my theory, 'being called such and such' only gives us a descriptive property of members of the name-set (the denotation); it does not determine the exact referent.  TCD does not violate the non-circularity condition in that as a theory of reference, what determines the reference is the speaker’s intention which is expressed by the associated descriptions.  Reference is not accomplished by using this description 'being called such and such.'  In contrast to Kneale's theory that uses the same description to give the meaning and to fix the reference of a proper name, TCD separates the functions of the two components of descriptions.  The associated descriptions in the speaker’s mind do not “give the meaning" of proper names.  What those descriptions do, is to help identify the referent fixed by the speaker’s intention.  What determines the meaning, on the other hand, is the denotation of the name.  We need to have the two kinds of descriptions to complete both denoting and referring.  Once the two components are assigned separate roles, there is no circularity in the definition of reference.

      Kripke presents another case that is supposed to be a problem for the description theory.  In the example (borrowed from Wittgenstein), Kripke discusses the following statement:

            (12)     Moses does not exist. 

The reason why (12) presents a problem for the description theory, according to Kripke, is that the Biblical descriptions should not be used to "fix" the reference of the name 'Moses.'  For one thing, it might happen that Moses did exist but he did not do the things the Bible attributed to him.  '[I]n that case maybe no one would have done any of the things that the Bible relates of Moses.  That doesn't in itself mean that in such a possible world Moses wouldn't have existed.'[Kripke 1972, pp. 66-67]  This argument does not strike me as a convincing one, because (12) does mean that the Moses described in the Biblical story does not exist.  Even if there were a Moses who had never gone into politics or religion, (12) is still true.  Under my theory, if there were people called 'Moses' so the name does not denote an empty set, but somehow all the descriptions we associate with this name does not pick out any in the set, then there is no individual who is picked out by the name 'Moses' in our usage.  The analysis of statement (12) is thus this:


(12')    It is not the case that there is an a such that a belongs to l = {all individuals called 'Moses'} and Fa.

      Another problem that Kripke [1972] attributes to the description theory involves Gödel.  This is a case where 'the person named by that name did not satisfy the descriptions usually associated with it, and someone else did.'[p. 254]  If the only description we have about Gödel is 'the man who proved the incompleteness of arithmetic,' and it could turn out that Gödel didn’t really prove it, but someone else called 'Schmidt' did, then the description we give would fix Schmidt for the name 'Gödel.'  With my theory, however, such a situation would not occur.  The descriptions we give to the name 'Gödel' in this case would be an ordered pair:   


(G)      <'an individual called "Gödel"'; F = {is the man who proved the incompleteness of arithmetic.}>

      If Gödel did not actually prove the incompleteness of arithmetic, no one else would satisfy this ordered pair of descriptions.  However, such an analysis is still insufficient.  (G) gives a more limited sense to the name 'Gödel' than what the name usually has in our discourse.  The F that a speaker associates with the name 'Gödel' is generally something like (G'):


(G')     <'an individual called "Gödel"'; F = {is the man who proved the incompleteness of arithmetic, is the person whose name I have read in many logic books,...}>

In other words, the F the speaker associates with the name is generally not a single description, but a set that includes descriptions of the speaker's epistemic conditions of the name.  Thus, the speaker can still use the name to parasitically refer to Gödel even if she is misinformed about who actually proved the incompleteness of arithmetic.  But now what Kripke wants us to consider is not just the misinformation of one speaker, but a counterfactual situation in which our Gödel did not prove the incompleteness of arithmetic.  This demand is committing what I called earlier 'the conflation between actuality and possibility.'  How we refer to an individual in a possible world is simply not the same issue as how we refer in our actual world.  The way we succeed in referring to Gödel by using the limited information that we have of him, is based on the fact that we learned about him in this way.  It would not be the case that anyone who happens to satisfy the sole description 'the man who proved the incompleteness of arithmetic' would thus become the referent of the name 'Gödel.'  It would rather be the case that the man Gödel would satisfy the set of descriptions, because part of the descriptions describe the epistemic conditions under which we learned about this name. 

      Finally, I shall address the Modal Argument Kripke sets against the description theory.  Kripke thinks that the descriptional approach gets the counterfactual wrong, because 'although the man (Nixon) might not have been the President, it is not the case that he might not have been Nixon (though he might not have been called "Nixon").'[1972, p. 49]  By this argument, both the qualitative description ('was a President of the United States') and the meta-linguistic description ('was called "Nixon"') fail to "fix" the reference across possible worlds.  No doubt this is a good argument, but I don't think the description theorist necessarily insists on fixing reference across possible worlds purely by the descriptions one uses to fix the reference in our world.  Let us consider these two statements:

            (13)     Nixon might not have been Nixon.

            (14)     Nixon might not have been called 'Nixon.'                 

My theory gives them the following analyses:


            (13’)    There is an a, a belongs to l = {all individuals called 'Nixon'} & Fa, and it is possible that (aa).

            (14')    There is an a, a belongs to l = {all individuals called 'Nixon'} & Fa,  and it is possible that (a is not called 'Nixon').

(13') and (14') seem to preserve the intuitive distinction that Kripke makes with regard to (13) and (14).  In other words, we fix the referent of 'Nixon' in our world in the same way TCD describes, and then we assign possible counterfactuals to this fixed referent.  This fixed-in-the-actual-world brings in indexicality to the present language and the realm of discourse.  A name used in our present discourse will always pick out the same individual across possible worlds, because it is this person whose counterfactuals we are considering.  Joseph Almog in his ‘Naming Without Necessity' suggests that naming is naming, and necessity is necessity, and that the connection Kripke tries to draw between the two is unfounded.  Almog argues that we should distinguish two stages in our semantic theory: the generation stage and the evaluation stage.  In the generation stage, we generate the propositional constituent (such as an individual person) corresponding to a name.  And the question for this stage is whether the name refers to the individual via some descriptive content.  In the evaluation stage, we evaluate the truth of the proposition in a possible world.  And the question for this stage is whether the individual (not the name) bears modal attributes.  Almog writes, 'The two questions are definitely different.  One concerns language.  The other is metaphysical, having nothing to do with names.  The two questions are not only different; they are independent of each other.  First, one could hold the semantical view that names refer by means of descriptive concepts, and yet couple this stand with the metaphysical view that objects...bear modal attributes....  Conversely, ... we could have naming without necessity.  One could believe that names do not refer by means of descriptive concepts, and couple this semantic view with a skeptical metaphysical attitude toward modal individualism.'[p. 229]  Almog himself holds the second view.  It should clear from what has been argued in this paper that I hold the first view, which Almog calls 'necessity without naming.'

      In summary, TCD is a more complete theory than Kripke's direct reference theory, and it does not have the same problems that older description theories do.  As Martinich [1997] remarked, 'Perhaps behind Kripke's puzzle is an even more general misconception about language: the belief that language is self-contained and that purely linguistic knowledge is sufficient for using language.'[p. 31]  I think what is being left out in this self-contained view on language is the speaker.  In contrast to this picture of a two-term relation between language and the world, the two-component description theory is based on the picture of a three-term triadic relation amongst language, speaker and the world.[27]  In such a picture, we can speak about the world both because we, as speakers, intend to refer to things or events in the world, and because the language we use gives descriptions of the world.  The speaker apprehends the meaning of terms in a language prior to choosing the terms for the intended reference.  On the other hand, the speaker’s intention picks out the one called such and such, and it is the speaker’s intention that determines the referent within the denoted set. 

      Even though referring seems to be a subjective, or intersubjective speech act, denoting is objective in the sense that it is governed by the fact that someone was indeed called such-and-such.  What Kripke’s causal chain gives us is a theory about denotation; about how an individual gets to be a member of the set denoted by the name.  But it does not give us a theory about reference; about how a speaker speaks about a particular individual called by that name.  Kripke is right in emphasizing the importance of the relation of causal chain, but this relation simply does not give us the whole story. 

      There are people who take speaker reference as pragmatics, not semantics, of names.[28]  I argue, however, that semantics cannot be separated from pragmatics, and speaker reference should be considered as part of the semantics of names in the language.  Language in itself sometimes gives only a partial proposition, and we have to consider the speaker's intention to complete the content of the proposition.  At the same time, language sometimes gives multiple propositions, and we also need to consider the speaker's intention to pin down the particular proposition expressed.  Without the speaker aspect, the semantics of language cannot be either complete or accurate.  The two components cannot be separated if our usage of a name is to bring us to the right individual.[29] 

      What prompted the whole debate between the direct reference theory and the description theory is exactly the issue of speaker reference.  Or, as Searle puts it, 'Both theories are attempts to answer the question, "How in the utterance of a name does the speaker succeed in referring to an object?".'[Searle 1983, p. 234]  According to the direct reference theory, names refer directly; that is to say, nothing that mediates between the name and the referent is semantically significant.  On the other hand, according to the description theory, names refer in virtue of the descriptions associated with the use of the name.  So the main issue being debated on is this: Is the reference of proper names mediated by any description?  The debate between the direct reference theory and the description theory is thus a debate concerning the issue of reference.  The debate can also be construed as a debate concerning the meaning of proper names.  The direct reference theorists argue that the semantic value of a proper name is simply its referent.  The description theorists, on the other hand, argue that the semantic value of a  proper name is the set of descriptions associated with the name.  This characterization of the core issue demonstrates further that the issue of meaning (or the semantic value) of names is closely related to the issue of reference, or how the reference is determined.  Thirdly, Searle [1983] presents the debate as a debate between internalists and externalists.  He says, 'The issue is simply this: Do proper names refer by setting internal conditions of satisfaction..., or do proper names refer in virtue of some external causal relations?.' [p. 233]  Lastly, Kripke seems to think that this is a debate between subjectivism and objectivism when he says, 'It is not how the speaker thinks he got the reference, but the actual chain of communication, which is relevant.' [1972, p. 93]  All these interpretations of the key issue demonstrate the fact that the issue of reference, albeit a pragmatic issue; albeit an issue of subjective speech act, is not separable from the issue of semantics; the issue of the objective semantic value of proper names.

      While putting the same emphasis on the meta-linguistic analysis of the meaning of proper names, my TCD is distinguished from Bach's NDT and Katz's PMT is exactly the incorporation of pragmatics into the semantics of proper names.  Both NDT and PMT are taken to be merely a semantic theory; a theory of sense, not of reference.[30]  Both Bach and Katz argue that their theories are therefore not responsible for answering Searle's criticism that the meta-linguistic sense of the name is insufficient in terms of fixing the reference in contexts.  They also both think that their theories are thus immune to Kripke's circularity argument.  I have argued, however, that the issue of reference is an essential part of a semantic theory of proper names.  Take out an answer to the question of reference, Bach's and Katz's theories of proper names, though correct to some extent, become uninteresting.

      Kripke says that any theory of beliefs and names must deal with the puzzles about Pierre’s and Peter’s belief, so now I will go back to Kripke’s puzzle.



§ III. An Application of TCD: Back to Kripke's Puzzle and Others

Kripke thinks that the case of Pierre’s belief resembles the case of Jones’ belief about Cicero and Tully.  So we shall begin with the latter case.  Why is it that Jones may believe that Cicero is bald while denying that Tully is?  With the analysis of the two-component description theory, we can explain that it is because 'a person called "Cicero"' and 'a person called "Tully"' are different descriptions.  Even when the only description Jones associates with both the name 'Cicero' and the name 'Tully' is 'is a Roman orator,' we would still have the difference in names in Jones’ mind.  So we can have:


(15)     Jones believes: Cicero (<'an individual called "Cicero",' F = {is a Roman orator}>) is bald, and

            Jones believes: Cicero (<'an individual called "Tully",' F = {is a Roman orator}>) is not bald.

This kind of analysis does not require that all propositions involving different names express distinct beliefs of the subject.  If, for instance, Sally knows that Cicero is Tully, then Sally's belief would be expressed as:


            (16)     Sally believes: Cicero (<'a man called "Cicero",' F = {a Roman orator, is also called 'Tully,'...}>) is bald.

In this way we would not have too many beliefs individuated by the different names the subject chose to express her belief. 

      With the case of Paderewski, Peter uses the same name but he associates different descriptions with the name.  Our ascription should be like (17):


(17)     Peter believes: Paderewski (<'an individual called "Paderewski",' F = {is a musician,...}>) has musical talent,


            Peter believes: Paderewski (<'an individual called "Paderewski",' Y = {is a politician,...}>) has no musical talent.

This gives us no problem since it is reasonable for anyone to think that there are two 'Paderewski' being referred to in the two utterances.  Similarly, Peter can assent to a sentence such as 'Paderewski is not Paderewski' by taking it to mean 'this Paderewski is not that Paderewski' without violating the law of contradiction. 

      Kripke asks us to decide whether the sentence 'Pierre believes that London is pretty' is true or false.  But as I argued earlier, sentences themselves do not have truth value.  What we should do in this case is to find the proper proposition expressed by, or the semantic value of, the utterance.  With Pierre’s belief, my proposed ascription is this: 


(18)     Pierre believes: London (<'an individual called "Londres" (by the French-speaking community),' F = {is a city in England, is the city of which I have seen a postcard,...}>) is pretty,


            Pierre believes: London (<'an individual called "London" (by the English-speaking community),' Y = {is a city in England, is where I reside at the moment,...}>) is not pretty.

The name within the quotation marks will not be translated.  Thus, even when Pierre associates the same set of descriptions with 'Londres' and 'London,' his beliefs do not express the same propositions.  I thus think that Kripke’s puzzle could not be generated under a properly laid out description theory such as my TCD.

      Kripke [1979] asks: 'What is it about sentences containing names that makes them -- a substantial class -- intrinsically untranslatable, express beliefs that cannot be reported in any other language?'[p. 129]  I think the reason is that proper names are really dependent on the communal usage of a linguistic community.  Statements such as 'Londres is London,' 'Eiffel Tower is la Tour Eiffel,' 'Köln is Cologne' are by no means trivial.  They convey important information in language acquisition.  The way a name is given and used is very much dependent on the conventions of a linguistic community and the sub-groups within.  By giving a standard translation of names, we are also changing the context and the epistemic condition of the subject. 

      Finally, I want to explain why I think the theory of proper names should not be any form of direct reference theory. The main difference between the direct reference theory and the description theory lies in the assertion concerning whether reference is direct or mediated.  In this paper I have argued how reference has to be mediated through the two sets of descriptions, and thus the direct reference theory simply takes the wrong approach.  The first component of my TCD, the meta-linguistic description of the meaning of the name, is necessary in the mediation of reference.  And as I have argued, the meaning is simply 'an individual called such-and-such (by a certain community).'  That is to say, the reference of a proper name has to be mediated through social, conventional usage of the name.  Secondly, when we talk about an object, the object being discussed always comes into our discourse via one perspective (mode of presentation) or another.  The second component of TCD captures how speaker reference is mediated through descriptions of the way (the mode) in which the object is presented to the speaker.  Kripke's causal chain or Donnellan's historical explanation take the perspectives out of the speaker's mind and put it in the mind of an 'omniscient observer of history.'[31]  But our language is used by people like us and we are not omniscient.  This fact explains why substitution salva veritate, which poses no problem for an omniscient observer, always poses a problem in an epistemic context involving ordinary speakers.  This also illustrates the deficiency of direct reference theory in general. 

      Kripke’s conclusion concerning the puzzles involved in the epistemic context seems pessimistic.  He says, 'When we enter into the area exemplified by Jones and Pierre, we enter into an area where our normal practices of interpretation and attribution of belief are subjected to the greatest possible strain, perhaps to the point of breakdown.  So is the notion of the content  of someone’s assertion, the proposition it expresses.'[1979, p. 135]  However, I think the problem of substitution salva veritate is a serious problem for the direct reference theory.  What it pushes for, is not to abandon the hope of finding an acceptable belief ascription, but always to consider the subject's meta-linguistic beliefs as well as her other relevant beliefs about the object.  With a description theory properly laid out that captures those other beliefs, there is no puzzle about beliefs.







Almog, Joseph 1986.  'Naming Without Necessity', Journal of Philosophy 87, 210-242.

Bach, Ken 1984.  Thought and Reference, Oxford: Clarendon Press.

Burge, Tyler 1973.  ‘Reference and Proper Names’, Journal of Philosophy 70, 425-439.

Burgess, John 1996.  'Marcus, Kripke, and Names’, Philosophical Studies 84, 1-47.

Corlett, J. A. 1989.  'Is Kripke's Puzzle Really A Puzzle?', Theoria 55, 95-113.

Donnellan, Keith 1966.  'Reference and Definite Description’, in Martinich (ed.) The Philosophy of Language, first edition, 1985, 236-248.

Donnellan, Keith 1977.  ‘Speaking of Nothing’, in Schwartz (1977), 3-31.

Fodor, Jerry 1994.  The Elm and the Expert: Mentalese and Its Semantics, Cambridge: The MIT Press, Appendix A: Names.

Kaplan, David 1975.  ‘Dthat’, in Martinich (ed.) The Philosophy of Language, first edition, 1985, 315-328.

Katz, Jerold 1977.  'A Proper Theory of Names’, Philosophical Studies 31, 1-80.

Katz, Jerold 1990.  'Has the Description Theory of Names Been Refuted?' in Meaning and Method: Essays in  Honor of Hilary Putnam, New York: Cambridge University Press.

Katz, Jerold 1994.  ‘Names Without Bearers’, Philosophical Review 103, 1-39.  

Kripke, Saul 1972.  Naming and Necessity, Cambridge: Harvard University Press.

Kripke, Saul 1977.  'Speaker Reference and Semantic Reference’, in Martinich (ed.) The Philosophy of Language, first edition, 1985, 249-268.

Kripke, Saul 1979.  'A Puzzle about Belief’, reprinted in Propositions and Attitudes, 102-48, (eds.) Salmon & Soames, Oxford: Oxford University Press, 1988.

Kvart, I. 1981.  'Kripke's Belief Puzzle’, in Midwest Studies in Philosophy, (eds.) French et al,  287-325.

Loar, Brian 1987.  'Names in Thought’, Philosophical Studies 51, 169-185.

Marcus, R. B. 1981.  'A Proposed Solution to a Puzzle about Belief’, in Midwest Studies in Philosophy, vol. 6, 501-510, (ed.) Peter French, Minneapolis: University of Minnesota Press.

Martinich, A. P. (ed.) 1996.  The Philosophy of Language, Oxford: Oxford University Press, third edition.

Martinich, A. P. 1997.  'Philosophy of Language’, Chapter 1 of Routledge History of Philosophy, Volume X: Philosophy of Meaning, Knowledge and Value in the Twentieth Century, (ed.) John Canfield, London: Routledge.

McMichael, A. 1987.  'Kripke's Puzzle and Belief "Under" A Name’, Canadian Journal of Philosophy 17, 105-126.

Over, D. E. 1983.  'On Kripke's Puzzle’, Mind 92, 253-256.

Pettit, Philip 1984.  'Dissolving Kripke's Puzzle about Belief’, Ratio 26, 181-194

Salmon, Nathan 1986.  Frege's Puzzle, Cambridge: MIT Press, Appendix.

Searle, John 1958.  'Proper Names’, in Martinich (ed.) The Philosophy of Language, first edition, 1985, 270-274.

Searle, John 1983.  'Proper Names and Intentionality’, in Searle: Intentionality, Cambridge: Cambridge University Press.

Schwartz, Stephen (ed.) 1977.  Naming, Necessity, and Natural Kinds, Ithaca: Cornell University Press.

Strawson, P. F. 1956.  ‘On Referring’, in Martinich (ed.) The Philosophy of Language, first edition, 1985, 220-235.

Wettstein, Howard 1988.  'Cognitive Significance Without Cognitive Content’, in Themes from Kaplan, (eds.) Almog, Perry and Wettstein, New York: Oxford University Press, 1989.


[1]The two principles can be stated as follows:

(PD)      If a normal English speaker, on reflection, sincerely assents to ‘p’, then he believes that p.

(PT)      If a sentence of one language expresses a truth in that language, then any translation of it into any other language also expresses a truth (in that language).

[2]Kripke says that we must say that Pierre has contradictory beliefs, that he believes that London is pretty and he believes that London is not pretty, even though Pierre himself 'cannot be convicted of inconsistency.' [1979, p. 122]

[3]For details of these puzzles, see Kripke [1979]. 

[4]For discussions on these puzzles, see Marcus [1981]; Pettit [1984]; Kvart [1981]; Over [1983]; Corlett [1989]; Salmon [1986]; McMichael [1987]; and Loar [1987].

[5]For example, see Marcus [1981].

[6]For example, see Martinich [1997].

[7]Marcus [p. 501] argues that the puzzle is 'a predicament that is generated by the theory of direct reference of names taken in conjunction with a plausible disquotation principle relating belief to assent.'  Katz [1990, p. 32] also argues that Kripke's puzzle 'isn't a puzzle for description theorists, since they reject Mill's view of proper names.' 

[8]Katz [1994, p. 18] makes the same criticism of Kripke’s theory.

[9]Even though the distinction made here is not commonly adopted, it has been adopted by others using different terminology.  Kripke distinguishes 'semantic reference' and 'speaker reference' in the case of descriptions.  According to him, 'If the speaker has a designator in his idiolect, certain conventions of his idiolect determine the referent in the idiolect: that I call the semantic referent of the designator...The speaker's referent is the thing the speaker referred to by the designator, though it may not be the referent of the designator, in his idiolect.' [1977, pp. 256-57]  Bach also seems to make a similar distinction when he talks about 'speaker reference' and 'linguistic reference.'[1984, 141].  This is the kind of distinction I intend to capture, but I prefer to separate the two terms.  Following Donnellan, I shall say that 'referring is not the same as denoting,' [Donnellan 1966, p. 236] and I use 'denotation' in roughly the same way that Kripke uses 'semantic reference.'  But with regard to 'reference', or 'speaker's reference' in Kripke's terminology, I differ from Kripke as well as Donnellan in one major respect: they both think that the speaker can refer to something outside the realm of semantic reference, while I take the speaker's reference to be restrained by the semantic reference.  (More on this later.) 

[10]As Strawson [1956, pp. 223-224] says, '"Mentioning" or "referring", is not something an expression does; it is something that someone can use an expression to do.' 

[11]The condition in the parentheses is used to rule out any arbitrary stipulation by one individual.  Furthermore, the condition brings in the linguistic community in which the calling relation gets established.  This is a three-term relation between the name, the object named, and the linguistic community, and is thus different from the two-term relation between a name and its bearer as championed by Katz.  For the contrast of the two views, see Katz [1990, pp. 38-9]. 

[12]This is basically in agreement with Burge’s predicate treatment of proper names.  Burge argues that even though proper names are usually used in singular and unmodified form, 'they play the role of predicates, usually true of numerous objects--on all occurrences.' '[Proper names] play instead the roles of a demonstrative and a predicate.  Roughly, singular unmodified proper names, functioning as singular terms, have the same semantical structure as the phrase "that book."  Unlike other predicates, proper names are usually used with the help of speaker-reference and context, to pick out a particular.' [Burge 1973, pp. 431-32] 

[13]Katz seems to have made a similar distinction between the set and the individual.  He calls the former 'type reference' and the latter 'token reference.'  Katz [1977, pp. 35-36] says, 'Let us call the referent of a word or expression its "type reference," and let us call something referred to in the use of a word or expression its "token reference."'  In Katz [1990] he also states, 'The type-reference of a proper noun is the collection of its bearers.'  And that 'the criterion for a literal application of a token of the type is that the referent of the token belongs to the type-reference.' [1990, p. 48]

[14]These descriptions are all included in the definition of the word 'call.' 

[15]If, however, everyone in the community calls a person 'Sir' and mentions him as 'Sir' to others, then I would think that the person has adopted 'Sir' as his name.  In a similar way, nicknames become proper names. 

[16]In the present discussion on proper names, we are dealing only with English and the English speaking community.  Thus, the parentheses (by our linguistic community) will sometimes be omitted for simplicity.

[17]How the name gets passed on from other languages to the present one could be explained by homophonic translation, semi-homophonic translation, or replacement of the original name with a new name, etc.  It is generally assumed, though not guaranteed, that the names in the present language have legitimate sources in other languages.

[18]For example, Moses might not have been called 'Moses,' but it is the name we come to use to refer to the one that did all those things the Bible attributed to him.  More on Moses later.

[19]Naturally, if we generalize too much, then the theory would no longer be a theory of "proper names".  But in view of the fact that so many widely accepted "proper names" of people are often the result of a stipulation or a change of names (e.g. in immigrating to a new country, one may create a new translation of one's native name and henceforth be known by that newly adopted name; one may go through a self-stipulated name-changing process and demand to be called by the new name from that point on, etc.), I feel that we should not include only names that were dubbed to the individuals at birth. 

[20]Wettstein [1988] calls the speaker’s 'having something in mind' a 'cognitive fix' of the speaker, and he argues that such a cognitive fix is not required for a successful referring act.  However, by 'cognitive fix,' Wettstein means that the speaker can correctly distinguish the referent from everything else in the universe, while I claim that correctness is not required here.

[21]According to Donnellan [1966], a description is used referentially if the speaker has the object in mind, and it is used attributively if the speaker simply uses a description to pick out whoever satisfies the description.

[22]Even though pragmatics relies on the interaction between speakers and hearers, it is mainly the speaker's intention that fixes the reference of an utterance.  The hearer may very well have a different set of associated descriptions upon hearing the name mentioned.  If the hearer's descriptions would pick out the same referent as the speaker-intended referent, then communication is successful.  If not, misunderstanding gets generated. 

[23]Katz makes a similar observation.  He says, 'What makes a word the right name for a thing is that the thing fits or conforms to the meaning of the word in the language.' [1990, p. 47]

[24]Kripke [1977, p. 257].  Of course this is an example of 'speaker's reference.' 

[25]I wish to thank A. P. Martinich for suggesting this analysis to me.

[26]Having the object itself in the proposition does not make the theory 'directly referential', since the object is mediated through the descriptions. 

[27]The philosophers who hold such a two-term relation would be Tyler Burge, Howard Wettstein, etc.  The three-term relation, on the other hand, seems to be explicit or implicit in the philosophy of Donald Davidson, Gareth Evans, etc.  Another kind of picture neglected here is a two-term relation between the speaker and the world, which seems to be implicit in the different theories of speaker’s meaning.

[28]Examples are Katz [1990] and Fodor [1994, pp. 111-112]

[29]This does not mean, however, that an individual necessarily fails to refer without using language.  One could, for instance, point to a particular object without saying anything and yet refer successfully to that object.  Kripke’s example of two people asking 'What is Jones doing?' when it was in fact Smith being referred to, seems to fall into this sort of cases.  In this case, I argue, the two speakers refer demonstratively, and not via the name  'Jones.' 

[30]Bach says, 'NDT... does not even purport to be a theory of reference.  It is nothing more than a modest theory of the modest meaning of names.'  [Bach 1984, p. 161]  Katz also says that his PMT 'is (part of) a theory of sense, not a theory of reference. [Katz 1990, p. 40]

[31]As Donnellan puts it, 'It might be that an omniscient observer of history would see an individual related to an author of dialogues, that one of the central characters of these dialogues was modeled upon that individual, that these dialogues have been handed down and that the speaker has read translations of them that the speaker's now predicating snub-nosedness of something is explained by his having read those translations.  This is the sort of account that I have in mind by a "historical explanation".' [Donnellan 1977, p. 230]