1 The Received Wisdom on What is Said

In philosophy of language, the dominant view regarding indexicals’ contribution to content, inherited from the work of David Kaplan (1989), has it that what is said by a speaker using a sentence that contains an indexical pronoun (or the ‘content’, in Kaplan’s terms) differs from the lexical meaning of the sentence (or its ‘character’) in that, on the one hand, it includes the specification of the pronoun’s referent (as determined in the context of utterance), but, on the other, does not include the lexically encoded conditions that help determining the referent, such as the condition of being female in the case of ‘she’, or the condition of being a speaker in the case of ‘I’. Suppose that on Friday May 15, 2012, at noon, Alma says:

(1):

I am hungry

According to the Kaplanian view, the meaning of the sentence in (1) is a function from contexts to contents which, given a context, returns the proposition that the speaker of that context is hungry at the time of the context, while what is said by (1), or, equivalently, its content, is the output of that function as applied to the context of (1), hence the proposition that Alma is hungry on May 15, 2012, at noon.

Let me clarify from the outset that my target in this paper is not Kaplan’s semantic account of indexicals, but rather, an account of what is said that builds directly upon Kaplan’s notion of ‘what is said’. In other words, the issue is whether the Kaplanian notion of content is able to account for the content of assertion and for what is said, where the latter is taken to be what calls for analysis, rather than just another technical notion.Footnote 1 Since I do not want to engage into Kaplan scholarship, I shall speak of the Kaplanian account, whether or not Kaplan himself would have subscribed to it.

There are two intertwined motivations for the Kaplanian account. Suppose that at the time at which Alma utters (1), Bruce, pointing at her, says:

(2):

She is hungry

Then we can truly report Bruce as having said in (2) what Alma said in (1). Indeed, they both said that Alma was hungry. Or, suppose that on Saturday May 16, Chris says:

(3):

Alma was hungry yesterday at noon

Again, it seems that one may truly report Alma and Chris as having said the same thing on the grounds of their utterances of (1) and (3).Footnote 2

The previous examples involve speakers who same-say using sentences whose meanings are not the same. The second motivation for the Kaplanian view concerns speakers who use sentences with the same meaning, yet fail to same-say. Suppose that on Saturday, May 16, 2012, Chris says:

(4):

I am hungry

Those who share Kaplan’s intuitions would insist that what is said by Alma in (1) and what is said by Chris is (4) are different things:

What is said in using a given indexical in different contexts may be different. Thus if I say, today, “I was insulted yesterday,” and you utter the same words tomorrow, what is said is different […] There are possible circumstances in which what I said would be true but what you said would be false. Thus we say different things (Kaplan 1989: 500).Footnote 3

2 De Se Assertion

It has been pointed out many times (e.g. Feldman 1980: 80, Lewis 1980: 97) that cases such as (1)–(4), in which the sentence “I am hungry” is uttered by different speakers, are as much of a problem for the Kaplanian view as they are a motivation for it. On the one hand, a proponent of the Kaplanian view is arguably right to say that, in some sense, what Alma says in (1) is different from what Chris says in (4). As we have seen, Alma’s utterance of (1) says, in some sense, the same thing as (2) or (3), but there seems to be no sense in which Chris’s utterance of (4) says the same thing as either (2) or (3). But on the other hand, there is, to use Lewis’s words, an “equally legitimate” sense in which Alma in (1) and Chris in (4) do say the same thing. Indeed, each says that he or she is hungry.

The intuition that Alma in (1) and Chris in (4) same-say is further supported by our linguistic practices of reporting what is said. Consider the following dialogue:

(5):

Chris: I am hungry

(6):

Bruce: Alma said that, too

Bruce’s report in (6) is ambiguous. It can be understood as reporting Alma to have said that Chris was hungry, or that she herself was hungry.Footnote 4 This ambiguity is very similar to the well-known syntactic ambiguity with VP-ellipsis. Suppose that Bruce says:

(7):

I love my wife, and so does Chris

On its “strict” reading, (7) says that Chris loves Bruce’s wife, while on its “sloppy” reading, it says that Chris loves his own wife. Given this apparent similarity, I will use the ‘strict’ vs. ‘sloppy’ terminology for the ambiguity that we find with reports of same-saying such as (6).Footnote 5

3 Saying the Same Thing Versus Using the Same Words

In the previous section, we saw that when different people say “I am hungry,” there is a sense in which they are saying the same thing, for each is saying that he or she is hungry. A response readily available to the Kaplanian is to point out that the same sentence is being uttered, and that this could explain why we are inclined to hear the two speakers as saying the same thing—for, after all, they are uttering the same words. In this section, I will show that this is not the right approach to the problem of de se assertion. Although the use of one and the same sentence may partly account for the intuition that the same thing has been said, that cannot be the end of the story, since, as I will show, it is neither necessary nor sufficient for the speakers to be using the same sentence in order to be reported, literally and truly, as same-saying. I will first provide cases in which speakers are using sentences that have different meanings as well as different contents (relative to their respective contexts), yet there is a sense in which they are saying the same thing, as robust as in the case of (1)–(4). Then I will provide cases in which two speakers are using the same sentence, but when we report them as same-saying, the report comes out false.

3.1 Different Meanings, Different Contents, Same Thing Said

Imagine a situation in which Alma, Bruce and Chris have attended Prof. Cheng’s lecture on Montague on Monday evening. During the lecture, Bruce tells Chris:

(8):

I really like this lecture

Several days later, in a conversation with Chris about recent lectures that they have attended, Alma says:

(9):

I really liked Prof. Cheng’s lecture on Montague last Monday

Chris may then truly reply to Alma:

(10):

That’s what Bruce said, too

The sentences used by Alma and Bruce are different, and so are their lexical meanings. What is more, the differences at stake are quite significant: where (8) contains a demonstrative, (9) contains a complex definite description, and while (8) contains present tense, (9) is in the past tense. The Kaplanian contents associated with (8) and (9) are also different, the one involving Bruce, and the other, Alma. Despite all this, the report in (10) is ambiguous and has a true, sloppy reading, on which Bruce is reported to have said that he liked Cheng’s lecture. This reading will be dominant if, for instance, it is common knowledge in the context of (10) that Bruce has no idea who Alma is and could not have said anything explicitly about her.

3.2 Same Meanings, Different Things Said

Just as using the same sentence is not required for same-saying, it is not enough either. Consider the following pair of dialogues:

  1. (i)

    de se assertion

    (11):

    I am a fool. (Alma talking to Chris)

    (12):

    I am a fool. (Bruce talking to Chris)

    (13):

    That’s what Alma said, too. (Chris’s reply to Bruce)

  2. (ii)

    de te assertion

    (14):

    You are a fool. (Prof. Cheng talking to Alma, overheard by Chris)

    (15):

    You are a fool. (Chris talking to Bruce)

    (16):

    (?) That’s what Prof. Cheng said, too. (Chris talking to Bruce again)

There is an interesting asymmetry between the 1st person and the 2nd person pronoun in how they behave in reported speech. Consider (13). As it stands, it has two readings: a strict reading, which reports Alma to have said that Bruce was a fool, and a sloppy reading, which reports her as having said that she herself was a fool. If it is, say, common knowledge in the context of (13) that Alma would have never said such a thing about Bruce, the immediately available reading of (13) will be its sloppy reading, and (13) will be true in virtue of Alma’s having uttered (11). However, if we try the same sort of sloppy report by simply replacing ‘I’ by ‘you’, such a report will be typically unavailable. Unless the reporter notes that the reportee was talking to someone else, or unless there is something peculiar about the context of the same-saying report, (16) will not be ambiguous, but downright false (assuming that Prof. Cheng never said that Bruce was a fool).

The asymmetry between the 1st person pronoun and the 3rd person pronoun is even more striking. Consider the following case, minimally different from (i) or (ii):

  1. (iii)

    de re assertion

    (17):

    She is a fool. (Prof. Cheng, talking of Alma)

    (18):

    She is a fool. (Chris, pointing at Daisy)

    (19):

    (?) That’s what Professor Cheng said, too. (in reply to Chris)

Again, in an ordinary context, (19), as it stands, is not ambiguous. Only one reading seems to be available, namely, the one on which Cheng said that Daisy was a fool.

To forestall a possible misunderstanding, I am not claiming that there is no sense whatsoever in which Cheng in (14) and Chris in (15) could be taken to have said the same thing. For instance, we may take them to be same-saying insofar as they are both saying of their addressee that he or she is a fool. But if this should serve as grounds for reporting what they said as being the same, the mere report in (16) won’t do. In general, what is further required is that the reporter should make it explicit that the addressee was someone else, as in:

(20):

That’s what Professor Cheng said, too, to Alma. (Chris talking to Bruce)

The report in (20) is, again, ambiguous between a sloppy reading, on which Cheng said to Alma that she was a fool, and a strict reading, on which Cheng is reported as having said to Alma that Bruce was a fool.

With the 3rd person pronoun, one can similarly report that the same thing has been said, provided that one makes it explicit that it was said about different people. Thus the following report, based on Cheng’s utterance of (17), becomes correct:

(21):

That’s what Professor Cheng said, too, about Alma

There is, then, an interesting asymmetry between de se assertion and the other cases, since the former, unlike the latter, is such that the reporter does not have to make it explicit that the reportee was talking about herself or himself.Footnote 6 This asymmetry raises the following problem. Suppose that Kaplanian contents play the role of what is said. Then, (11) and (12) have different contents, and still, in an important sense, they say the same thing: in both cases, the speaker is saying of herself or himself that she or he is a fool. This sense of same-saying is further reflected in the fact that, properly disambiguated, the report in (13) is true. Now, a Kaplanian might think that this is because the sentences uttered in (11) and (12) are the same. However, this explanation won’t work. Take (14)–(15) and (17)–(18). There, too, the sentences uttered are the same, but we do not get a sloppy reading for either (16) or (19). In ordinary contexts, such reports will not be ambiguous, but false.

To bring the point home, what the cases discussed in this section suggest is that something is still missing in the account that the Kaplanian view gave us for the de se cases in the first place. Recall how the dialectics goes. The Kaplanian view predicts that if different speakers say “I am hungry”, they are saying different things. Faced with the observation that we often hear, and report, such speakers as same-saying, the Kaplanian notes that this must be because the speakers, although saying different things, are using the same sentence, and are thereby prone to being perceived as same-saying. Now, if this were the whole story, one would expect that in other cases in which the speakers are using the same sentence, as in (14)–(15) or in (17)–(18), they would be just as prone to being perceived, and reported, as same-saying. Yet, as we saw, this is not what happens. What the Kaplanian still owes us, then, is an account of the asymmetry in reporting same-saying when the target sentences use the 1st person pronoun rather than the 2nd or the 3rd person pronoun. What is more, we have pointed out that there are cases of de se assertion, viz. (8)–(9), in which we hear, and report, the speakers as same-saying, even though they are not using the same sentence (and not expressing the same Kaplanian contents either). Since the same-sentence explanation fails to apply to such cases, the Kaplanian owes us another explanation. To be sure, neither of these two issues shows that the Kaplanian account is wrong or untenable; what it shows is that, qua an account of the intuitive notion of what is said, it is in need of further development.

4 De Se Assertion Versus the Other Senses of ‘What is Said’

It is possible that what I have described as a received view may well no longer be one. There is a growing number of Kaplan’s followers who, while fully endorsing his semantics, do not want to identify semantic content with what is said. Among such semi-Kaplanians one finds e.g., Salmon (1986) and Soames (2002), to mention only those two. The idea, which seems to be more and more widespread, is that the notion of what is said is simply too versatile and too heavily context-dependent to be possible to capture by means of the notion of content.

One of the first to have pointed out that Kaplan’s identification of semantic content with what is said was unwarranted was David Lewis:

Unless we give it some special technical meaning, the locution ‘what is said’ is very far from univocal. It can mean the propositional content, in Stalnaker’s sense (horizontal or diagonal). It can mean the exact words. I suspect that it can mean almost anything in between (1980: 97).

In addition to the senses mentioned by Lewis, there is also was Ziff (1972) called the implication sense of what is said. Suppose that Bruce is thinking of enrolling in Prof. Blanchet’s logic class and asks Alma what she thought about it. She says:

(22):

I don’t think I’ve ever been in such a boring class in my whole life

Then he asks Daisy, and she says:

(23):

That class is a sheer waste of time!

It is easy to imagine Bruce replying to Daisy:

(24):

That’s what Alma said, too

Reports such as (24) are very natural and ubiquitous, but the relevant sense of same-saying cannot be captured by either lexical meaning or Kaplanian content, or even any combination of the two. This presses even further the worry whether one could ever account for the notion of what is said by pinning it down to something as stiff as the notion of semantic content.Footnote 7

In the context of the present discussion, the plurality of senses connected with the locution ‘what is said’ raises the following worry: the sort of cases that I brought up in Sect. 3.1. against the Kaplanian view, aren’t they just another garden-variety of the many senses of ‘what is said’? In the remainder of this section, my goal will be to show that there is a significant difference between reports of same-saying in the case of de se assertion and reports such as (24), which rely on the implication sense of what is said.

Reconsider (24). Even if we take this report to be true in the context at stake, someone who witnessed Alma’s utterance of (21) may justifiedly challenge Bruce (the reporter) as follows:

(25):

No, Alma didn’t quite say that. She only said that she had never been in such a boring class. That need not mean that the class is a waste of time

Presented with such a challenge, Bruce will normally retract, or at least somehow qualify his report, for instance as follows:

(26):

OK, she didn’t quite say that, but that’s what she meant

This retraction doesn’t necessarily show that the report in (24) was false.Footnote 8 But what it shows is that the sense of same-saying relevant to the truth of such reports is not its most literal sense, but rather, a looser sense. Now, compare this with the case of de se assertion:

(27):

Alma: I am hungry

(28):

Chris: I am hungry

(29):

Bruce: Alma said that, too

(30):

Daisy: (?) No, she didn’t quite say that. She only said that she was hungry

Faced with Daisy’s challenge, Bruce will not retract his report in (29)—rather, he will point out that he was precisely reporting Alma as saying that she was hungry, and that Daisy simply failed to disambiguate his report properly. Once again, there is a striking similarity with VP-ellipsis:

(31):

Chris: I love my wife

(32):

Alma: So does Bruce

(33):

Daisy: (?) No, he doesn’t. He only loves his own wife

Daisy’s objection in (33) is off the target, just as it was in (30). In both cases, her attempt at denying her interlocutor’s report merely emphasizes the fact that the report was ambiguous—she denies the reading not intended by the reporter precisely in order to assert the intended reading, and this is why (30) and (33) sound infelicitous.

The lesson to be drawn from these examples is that reports of same-saying in the case of de se assertion pattern differently from such reports in the other cases known from the literature. In those other cases, there is evidence that the locution ‘what is said’ is used loosely rather than literally.Footnote 9 On the other hand, the sense of same-saying deployed in reports of de se assertion belongs squarely among literal uses.Footnote 10

5 Concluding Remarks

My aim in this paper has been to point out some problems for the received view regarding indexicals’ contribution to the asserted content, on which regardless of whether we use the first, second or third person pronoun, the content, or what is said, will be the same provided that the referent is the same. I have argued that the received view can only handle a limited range of cases, and that there is no straightforward amendment that would allow it to handle the rest. In particular, the idea that speakers who say “I am hungry” may be truly reported to have said the same thing because they have used the same sentence is unsatisfactory, as I argued in Sect. 3. On the one hand, speakers who say “She is hungry” may be truly reported to have said the same thing only if their uses of ‘she’ refer to the same individual (or else, if the reporter makes it explicit, or if the context makes it sufficiently clear, that they said it about different individuals). On the other hand, there are cases of de se assertion in which speakers are truly reported as same-saying even though they are using sentences with sometimes very different lexical meanings. In Sect. 4, I considered another possible approach to the notion of what is said, which holds that this notion, as well as the relation of same-saying, are just too heavily context-dependent to be analyzable by means of the notion of semantic content. While acknowledging that we may often report same-saying in cases in which neither the Kaplanian contents nor the de se contents are the same, I showed that there was still a significant difference between these reports, which are easy to challenge and thus suggest that the relevant sense of ‘what is said’ is a loose one, and reports of same-saying in the case of de se assertion, which, when challenged, reveal that there was genuine ambiguity in the report, and thus fall together with the (more) literal uses of the locution ‘what is said’.

By way of a final remark, let me note that the data that I have presented, and the related problems for the Kaplanian view that I have raised, need not be seen as insuperable obstacles. Perhaps there are amendments to the view that would enable it to handle the data. However, the most straightforward ones fall short of accounting for the cases discussed, and for what seems to be a privileged status that the first person pronoun has not only in the realm of thought (as has been long known from the literature on de se attitudes and on the essential indexical), but also in language.