Speaker’s Reference, Semantic Reference and Public Reference

Kripke views Donnellan’s misdescription cases as cases where semantic reference and speaker’s reference come apart. Such cases, however, are also cases where semantic reference conflicts with a distinct species of reference I call public reference , i


The public referent of a designator
Intuitively, the problem in the Donald case seems to be that Bob's intention to say something about Donald is simply not manifest to his audience in the same way as in Donnellan's traditional misdescription cases.It is incumbent upon a speaker to make his communicative intention manifest to his audience by tailoring his speech act so as to allow his audience to recover this intention from the publicly available evidence as to this intention2 .In the traditional cases the speaker manages, despite his error, to do so, while in the Donald case this does not happen.This idea allows us to define a distinct species of reference, call it 'public reference'.
The public referent of a designator is the object that best fits all the relevant public evidence available at the time of utterance as to the speaker's referent of the designator.
Note that, provided we adopt the popular view of communication according to which communication is a matter of a speaker conveying information to a hearer, a need for the notion of the public referent of a designator arises naturally from any construal of communication.In any communicative interaction there will be potential cues immanent to the situation that the speaker can employ in order to communicate.These potential cues will include the linguistic conventions of whatever public language speaker and hearer share (i.e. the cues that determine semantic reference), but will also include extra-conventional factors like objects that happen to be visually salient, information that the speaker and hearer can take to be common knowledge between them, and so on.Any rational speaker will try to tailor his speech act so that his communicative intention is recoverable from such (conventional and extra-conventional) cues, any rational hearer will know he is trying to do so, any rational speaker will know that any rational hearer will know that he is trying to do so, and so on.It is the role of such cues in communication -not merely Donnellan's misdescription cases -that motivates the introduction of the notion of the 'public referent', i.e. the object that best fits all publicly available cues3 .
The issue concerning non-conventional cues used to communicate has received extensive discussion in the literature concerning bare demonstratives 4 .Note, however, that while authors writing on bare demonstratives have recognised the role played by extra-conventional evidence in facilitating communication, the issue addressed here is distinct.The issue in the demonstratives literature concerns whether such extra-conventional factors determine semantic reference.In the current discussion, however, the semantic reference of the designator in ( 1) is not at issue as it is straight-forwardly determined by considering the linguistic conventions governing (1) and the facts concerning the situation5 .I will not claim that such extra-conventional cues influence the semantic reference of (1); rather I will claim that recognising the role played by such cues in facilitating communication forces us to recognise a species of reference (that is determined by both conventional and extraconventional cues) and is distinct from both semantic reference and speaker's reference.
In typical acts of communication the public referent, semantic referent and speaker's referent will coincide.It can, however, always turn out that the speaker has misjudged which object best fits the publicly available cues, and so public reference and speaker's reference can come apart6 .Furthermore, semantic reference and public reference can come apart in at least three ways.First, public reference and semantic reference can come apart when the charitable interpretation of a speaker's words implies that the speaker chose a designator due to the false belief that the designator is applicable to the speaker's referent.In other words, cases like where someone uses the designator 'the richest man in the world', yet it seems obvious that he wishes to talk about Bill Gates, not (the somewhat wealthier) Carlos Slim.In such a case the semantic referent of the utterance is Carlos Slim.The public referent, i.e. the person that a rational third party would, based on the public evidence, think that the person is trying to talk about, is Bill Gates.If the public evidence is reliable, i.e. if the person is actually trying to talk about Bill Gates, then Bill Gates is both the speaker's referent and the public referent of the utterance.We could, however, have a case where the public evidence is misleading.The speaker could wrongly believe that Warren Buffett is the richest man in the world and so, despite appearances to the contrary, be trying to talk about him.In such a case Carlos Slim would be the semantic referent, Bill Gates the public referent and Warren Buffett the speaker's referent of the utterance.
Secondly, public reference and semantic reference can come apart when the charitable interpretation of a speaker's words implies that he is using some linguistic convention in a non-standard way, for instance if he appears to have some mistaken7 linguistic belief.
Consider, for instance, someone who uses the designator 'most famous imaginary number', but it seems obvious that he wishes to talk about π and is merely using the term 'imaginary number' incorrectly.This could, if the public evidence is reliable, be a case where public reference differs from semantic reference, but coincides with speaker's reference.
Alternatively, we could imagine a case where the public evidence is misleading as the speaker is trying to talk about something else entirely.Suppose that the speaker actually has Euler's number in mind.If we assume that i is the most famous imaginary number, then i would be the semantic referent, π would be the public referent and e would be the speaker's referent of the utterance.
Thirdly, public reference and semantic reference can, at least on one understanding of such phenomena, come apart where linguistic conventions underdetermine what is communicated.
If someone utters a phrase like 'the tennis player who is ready' 8 , but does not specify what the person is supposed to be ready for, then, arguably, such an expression has no conventionally determined semantic referent.Yet the phrase may well have a public referent, i.e. the information publicly available at the time of utterance may seem to render a clear verdict concerning who the utterer wishes to talk about by uttering the phrase 9 .We could, once again, have a case where this public evidence is reliable, and so public reference and speaker's reference coincide.Or, alternatively, it could be the case that the public evidence is misleading and the speaker wishes to talk about someone else entirely.In such a case the utterance has no semantic referent, but does have a public referent, distinct from its speaker's referent.
All theorists, of course, do not agree that a phrase like 'the tennis player who is ready' should be taken to have no semantic referent.For the purposes of this paper, however, I will not commit to any position concerning the semantic reference of devices like demonstratives, definite descriptions that may be said to contain unarticulated constituents, and so on, or commit myself to any specific construal of the distinction between semantics and pragmatics.
Such issues are orthogonal to present concerns, as shown by the fact that theorists can be in perfect agreement concerning the public reference of an utterance, yet fundamentally at odds about how to understand the semantic reference of an utterance.Consider the example above: 8 Theorists who argue that sentences like 'Bob is ready' do not express full propositions with determinate truthconditions include, among others, Bach (1994) and Carston (2008). 9Underdetermination cases should be taken to include cases of purely pragmatic communication, i.e.where no linguistic conventions whatsoever are involved in determining which object a communicative action is about.If my friend is standing at a bar and asks if I would like a beer by drawing my gaze to a beer and giving me a questioning look, then the notion of semantic reference simply does not apply to the situation.Yet the beer can be the public referent of my friend's communicative action as he made it evident that he wished to communicate something about the beer to me.See Carston (2008: 61 -63).
one theorist may claim that the utterance of the phrase 'the tennis player who is ready' has no semantic referent, another may claim that it does have a semantic referent as the speaker's intention co-determines semantic reference, while another may claim that it does have a semantic referent as some individual is contextually salient and this fact co-determines semantic reference.Yet all three can agree on the identity of the public referent, i.e. the object that best fits all the relevant public evidence available as to the speaker's referent of the designator.This is so, despite the fact that different semantic theories held by different theorists may lead them to differ on whether public reference and semantic reference coincide in a specific case.Inasmuch as such differences are due to different semantic theories, this is irrelevant to the nature and viability of a theory of public reference.Such disputes do not serve to undermine the viability and utility of the notion of public reference, provided that the theorist believes that there are at least some linguistic devices that can be used to communicate speaker's reference, and that can be uttered in circumstances that include at least some non-semantic contextual clues as to the speaker's intention.As long as this is the case, then semantic reference and public reference are distinct in theory and can come apart in practice.I know of no theorist that violates such a constraint.Public reference is determined by all the publicly available evidence as to the speaker's referent of an utterance.Such public evidence straightforwardly includes semantic factors like linguistic conventions.Beyond this, theorists will differ about which of the factors determining public reference also serve to determine semantic reference, but such a dispute need not imply any dispute about the nature of public reference.We have a somewhat parallel situation with speaker's reference; some theorists may incorporate elements of speaker intention into their semantic theory, while others may not.This could lead to it being the case that these theorists differ about whether, in some specific case, speaker's reference and semantic reference coincide.Yet such disagreement, inasmuch as it concerns semantics, is orthogonal to issues concerning the nature and viability of a theory of speaker's reference.
The 'Donald' case and the three types of cases outlined above indicate that we should recognise speaker's reference, semantic reference and public reference as three distinct species of reference.I do not mean to suggest that all three are somehow ontologically basic, i.e. not analysable in terms of one of the others.As my definition of public reference indicates, I take public reference to be a kind of assessment as to speaker's reference and hence a notion that is parasitic on the more basic notion of speaker's reference 10 .In fact, I am 10 Of course, someone who rejects the notion of speaker's reference can still agree that there is some relation, definable in terms of reference and which includes non-semantic publicly available criteria, that holds between Bob's utterance and the happy teetotaller and can usefully be called 'public reference', but claim that it need not be construed in terms of speaker's reference.sympathetic to arguments that semantic reference is a similarly parasitic notion 11 .My use of 'species of reference' is supposed to indicate no more than that all three notions are essential to analysing the process of communication, that each determines an object in a unique way and that each must be defined in terms of the notion of reference.

Kripke's analysis is, at best, incomplete
In Kripke's traditional case the teetotaller is both the speaker's referent and the public referent of the utterance.Hence the traditional case is a case where speaker's reference clashes with semantic reference, but also a case where public reference clashes with semantic reference.This raises the question: which distinction captures 'what is really going on' in such cases?I will construe this question as: which species of reference is it that causes our intuition that the utterer of (1) in the traditional case 'said something true' of the teetotaller?Three options can be distinguished.
First, we could claim that Kripke's construal of the Donnellan cases is complete and correct.
In other words, we could insist that Kripke's claim that the intuitive appeal of the idea that the speaker has 'said something true' in the Donnellan cases is fully explained by the fact that such cases involve a clash between speaker's reference and semantic reference and that the matter of public reference is completely irrelevant.One way to do so would be to claim that, in the Donald case, we do intuitively feel that Bob has said something true about Donald.I fail to see any merit in following this option; personally I feel no appeal to say that Bob has, in any sense, said something true about Donald, nor have I encountered any semanticist who has such an intuition.Furthermore, on such a view it would merely be the most fantastic coincidence that all the traditional cases just happen to be cases where speaker's reference and the public reference coincide, despite it seeming obvious that the cases were identified in virtue of this fact.For these reasons I feel safe in dismissing this variant of option one.
Another way to defend option one would be to admit that we do not feel compelled to say that Bob said something true about Donald, but to deny that this has anything to do with Donald not being the public referent of the utterance.In other words, it could be claimed that the traditional cases and the Donald case are also different in some other way that does not involve public reference and that this difference, once noticed, somehow implies that we need not abandon a Kripkean construal of the traditional misdescription cases.One way of doing so would be to suggest that our intuitions in the Donald case are generated in virtue of the fact that the example invites us to consider the case as hearers of the utterance, whereas in 11 Gail Stine (1978: 324) and Mark Sainsbury ( 2014) have defended such views.
as omniscient third parties apprised of all the facts.If this is correct, then we are the victims of a kind of parlour trick when we think the Donald case casts doubt on Kripke's construal of the misdescription cases.
While I cannot conclusively dismiss the above possibility, I do not see any positive reason to suppose that it is true.The misdescription cases are standardly explained from the perspective of a third-person narrator.This places us, at least as far as the stories go, in the position of an omniscient third party and I see no reason to suppose that we place ourselves in an epistemically constrained position when our intuitions are solicited.Admittedly it is the case that the Donald case -and the notion of public reference -brings the perspective of the hearer12 'into the story' in some basic way.This, however, as mentioned before, is also true of the traditional cases; it is surely no accident that the traditional cases are cases where public reference and speaker's reference coincide, i.e.where the speaker's reference of the utterance is obvious to those who hear the utterance.The original cases were clearly selected with the perspective of the hearer in mind, and so I see no reason to suppose that the Donald case involves hearers in a way that the original cases do not.
The above, of course, is only one way of arguing that there is some relevant difference between the original cases and the Donald case that serves to disqualify the Donald case from casting doubt on Kripke's analysis.Perhaps other ways of making such a case are possible.
Nothing obvious, however, springs to mind, and so I will treat this option as dismissed until a positive case for such a contention is made.
A second option would be to say that Kripke is partially correct, but that his analysis needs to be supplemented by acknowledging the role of public reference.On this analysis, then, the Donnellan cases should be seen as cases where our intuitive judgment that the utterer 'said something true' tracks the fact that speaker's reference and public reference coincide, i.e. on such a view our intuitions are responsive to the coincidence of speaker's reference and public reference.On this view 'what is really going on' in the traditional cases is that public reference and speaker's reference coincide, but clash with semantic reference.This would imply that Kripke's analysis is partially correct, but incomplete as it misses the relevance of public reference.
A third, more radical option, would be to say that Kripke's analysis is simply wrong, i.e. that the traditional Donnellan cases should be analysed as cases where public reference and semantic reference differ, with the matter of speaker's reference being altogether irrelevant.
This would be the case if someone thinks that the Donald case is like the traditional cases in all relevant ways, i.e. for those who intuitively feel that Bob 'said something true' about the happy teetotaller in the Donald case, despite the fact that he was trying to talk about Donald (and despite the fact that some unrelated person was the semantic referent of his utterance).
On this view, then, our intuitive judgment that the utterer 'said something true' in Donnellan's cases was responsive to facts about public reference, not speaker's reference (as Kripke seems to think) or the coincidence of public reference and speaker's reference (as on the second option).Such a construal amounts to claiming that, when Kripke identified speaker's reference as the non-semantic species of reference operative in the Donnellan case, he picked the wrong one of the two non-semantic species of reference that coincide in such cases.
I confess that my own intuitions concerning the Donald case are not definite enough to make it obvious which facts such intuitions are tracking.On the grounds afforded by such intuitions I am inclined to accept option three, but not without also seeing some merit in option two.
Once we consider matters beyond intuitions, however, I think it becomes clear that we should prefer option three.I base this on the fact that the notion of public reference is so basic to communication that we would expect our intuitions to track it.I explain why I view it as basic below.

The importance of public reference
When communicating, the hearer of an utterance of a designator needs to try and determine the speaker's referent of the designator.He can do no better than rationally exploiting the public evidence as to speaker's reference that is available at the time of utterance.A rational speaker will recognise that a hearer is in such an epistemic situation, and is therefore obliged to tailor his communicative action so as to make his intention recoverable via the exploitation of such evidence.A rational hearer will recognise that the speaker intends for his intention to be recoverable in virtue of such cues, a rational speaker will recognise that a rational hearer will recognise that he intends for his intention to be thusly recoverable, and so on.The publicly available cues that facilitate this process include the conventional rules of the language known by all competent speakers of the language.These rules serve to determine semantic reference.These conventional cues, however, are only a subset of the cues that can be rationally employed to communicate, for communication is also facilitated by nonconventional, contextual 13 factors.Hence arises the need to talk about the public referent of a 13 These cues are contextual cues, but in the epistemic sense, not in the constitutive sense.Their relevance depends on whether they can be evidence as to the speaker's reference of an utterance.Note that such relevance will always be relevance for a specific 'public', i.e. audience.The specific 'audience' will be determined in the designator, i.e. the object that best fits all public cues available at the time of utterance.The categories of semantic reference and public reference are alike in that both concern cues that speakers use to communicate, that hearers will know that speakers use to communicate, that speakers will know that hearers will know that speakers use to communicate, and so on.This reasoning, if correct, implies that communication is only intelligible if both speaker and hearer have some intuitive grasp of the notion of public reference, just as they have some intuitive grasp of the notion of semantic reference.
Our practices serve to vindicate the claim that speakers and hearers have an intuitive grasp of the notion of public reference.Suppose, for instance, that I am speaking to a friend about a mutual acquaintance who has recently much improved his golf game and I call the acquaintance 'Tiger Woods'.Such a communicative act only works if both parties exploit certain conventional and non-conventional interpretive cues, i.e. the conventional linguistic rule that 'Tiger Woods' refers to Tiger Woods and non-conventional aspects of context such as it being common knowledge between the relevant parties that Tiger Woods is a fine golfer, that the actual Tiger Woods is not conversationally salient, that the friend who has recently much improved his golf game is conversationally salient, and so on.Note, furthermore, that such a communicative act is only intelligible if the parties to the communicative act grasp that a sincere speaker can rationally violate the general norm of trying to make sure that speaker's reference and semantic reference coincide.However, it is never similarly rational for a sincere speaker to violate the general norm of trying to make sure that speaker's reference and public reference coincide.Such an action would amount to sincerely trying to communicate, i.e. make an intention manifest, and also trying to intentionally create a situation where the overall public evidence as to the intention is misleading, i.e. actively trying to not make the relevant intention manifest.Such action cannot be rational.In fact, it is precisely because pragmatics is more basic than semantics, at least in this sense, that I can rationally and sincerely violate linguistic convention in the 'Tiger Woods' case.In such cases I rationally depend on it being common knowledge between myself and the hearer that extra-conventional public evidence can, in principle, 'overrule' the conventionally determined semantic verdict.This presupposes that both speaker and hearer have, as was predicted by theory, some grasp of the notion of public reference.
The above reasoning also explains why, when we say that the happy teetotaller is somehow relevant to Bob's utterance, this is not something irrelevant that we 'impose' on Bob.Any sincere and rational speaker can be expected to, in typical cases, have a standing intention to same way as 'public reference' is determined and for the same reason; the relevant audience would be the audience that it is most rational, based on public evidence, to take to be the intended audience.use designators correctly, i.e. to make sure that semantic reference and speaker's reference coincide.The same goes for how a speaker exploits non-conventional cues in order to communicate.Just like the hearer of an utterance will expect the utterer to be trying to use linguistic rules correctly in order to communicate, so a hearer can expect that the utterer will be trying to employ non-conventional cues 'correctly'.For example, if the communicative success of an utterance depends on the attention of both parties being drawn to the most visually salient object, then the hearer can expect the utterer to make the utterance only if the utterer believes that one object clearly qualifies as the most visually salient, that there is good reason to believe that the parties will identify the same object as most visually salient, and so on.In general we can say that a rational and sincere speaker can be expected to have a standing intention to make sure that the public referent of a designator upon an occasion of use coincides with the speaker's reference of the utterance.This intention amounts to no more than the intention to communicate in such a way that the extra-conventional, public cues relevant to the utterance help to make the speaker's intention manifest 14 .If a speaker can be attributed such intentions, then the semantic referent and the public referent of a designator are relevant to his utterance in virtue of norms that he himself is committed to in virtue of attempting to communicate.
The above point about the basic nature of the notion of public reference also suggests that we should not make the mistake of thinking that the notion of public reference is only applicable to exceptional cases where conventions are used in a non-standard way.The notion is relevant to all communicative contexts that feature a designator.This can be shown by considering a case of seeming misdescription.Stipulate that the facts are as Kripke says they are in his modified Donnellan case, but, this time, the speaker does intend to speaker-refer to the man drinking champagne.In other words, stipulate that some thought worthy of expression concerning the champagne drinker had suddenly occurred to the speaker, but that nothing in his past behaviour indicated that he is aware of the champagne drinker's existence.Further stipulate that the teetotaller is standing close by and has, just seconds prior to the utterance of (1), let out a yell of delight after receiving a phone call, hence making himself -and his happiness -salient.In such a scenario speaker's reference and semantic reference coincide, but a rational observer would still think that the speaker wishes to speaker-refer to the teetotaller.Something has gone wrong; the mistake is a matter of the speaker's misjudgement of the communicative context.The speaker violated an obligation that goes beyond making sure that he means what his words mean; he violated a reasonable obligation to make sure that 14 Such rational constraints on communication can be motivated in terms of more general principles or in terms of established theories of pragmatics, e.g.Relevance Theory.My formulation is intentionally generic as my argument does not depend on any particular view of pragmatics.
it is evident that he means what his words mean 15 .Hence, even when the speaker is speaking in a standard way, the extra-conventional evidence as to speaker intention available at the time of utterance still places some minimal constraint on what can usefully be uttered, and remains something that rational speakers and hearers should be mindful of.
The above considerations strongly suggest that the category of public reference is, as is the case with the categories of speaker's reference and semantic reference, among the 'natural kinds' of communication.It forms a basic part of the scaffolding that makes communication possible; hence we should expect people to have some intuitive grasp of the notion of public reference.I see no similar considerations that indicate some utility in tracking the coincidence of speaker's reference and public reference.Simply put, there is some point in tracking public reference (option three), but no point in tracking the coincidence of speaker's reference and public reference (option two).On the assumption that our intuitions track entities that are useful to track, I endorse option three.In other words, I do think that any lingering inclination to say that the speaker 'said something true' in the Donnellan cases should be seen as an intuitive response to facts about public reference, not speaker's reference.
Note, however, that if the reader is unconvinced by the above reasoning and endorses option two, it remains the case that Kripke's analysis missed something important, namely the role of public reference in the Donnellan cases.This conclusion is interesting in itself.Both option two and three also imply the more important claim 16 that there is a species of referencepublic reference -that is fundamental to communication, is tracked by our intuitions and that our intuitive judgments are responsive to17 .Kripke's paper showed how easy it is to conflate semantic reference and speaker's reference.His paper also played a large role in sensitising the profession to the general importance of this distinction.The present considerations show that there is at least one more species of reference that should similarly be at the forefront of our minds.
15 Or, at least, he violated some obligation to make sure that he is not in a situation where there is strong evidence to upset the default assumption that he means what his words mean. 16Of course, the construal of the misdescription cases offered in this paper do not affect the conclusion Kripke reaches in his argument against Donnellan.Kripke's conclusion, namely that the oddity of the Donnellan cases reflect a non-semantic species of reference, holds good even if we construe Donnellan's cases as concerning public referrence and not speaker's reference.