I thank the members of the research colloquium of the Chair for Theoretical Philosophy at the University of Mannheim and the participants of the 2021 San Raffaele School of Philosophy for valuable feedback. My special thanks go to Wolfgang Freitag, Dorit Bar-On, Mitchell Green, Maximilian Philipps, Helge Rückert and two anonymous reviewers of this journal for very helpful comments on previous drafts. I am grateful to Christopher von Bülow for proofreading and editing and to the Fritz Thyssen Stiftung for financially supporting this research.
- 1 Proponents of avowal descriptivism are, e.g., Rosenthal (1986, p. 75; 1993, pp. 49–50; 1995; 2010) (...)
1We regularly avow being in mental states. Avowals are serious and competent utterances of present-tense self-ascriptions of mental states, like “I believe that it’s raining”, “I am annoyed” or “I like you”. The most straightforward analysis of avowals is descriptivism. According to avowal descriptivism, avowals of the form “I ψ ((that p)/(o))”, with the dominant verb ψ referring to a psychological state, p for a proposition and o for an object, function as assertions about the speaker’s mental state of ψ-ing and are expressive of the belief that the speaker is ψ-ing. Descriptivists further deny that avowals are expressive of the speaker’s (first-order) ψ-ing. That is, on a descriptivist view, serious and competent utterances of “I believe that it’s raining”, “I am annoyed” and “I like you” function as reports about the speaker’s mental states.1
2There is a challenge to this position which motivates a different account of avowals. David Rosenthal – an avowal descriptivist – formulates it as follows:
[W]hen I say ‘I doubt it’ […] it seems natural to take me to be expressing my doubt, and not just reporting what mental state I am in. […] If you ask me whether it is raining and I say ‘I think so’, it would be bizarre to take me to be talking about my mental state, rather than the weather. (Rosenthal, 1993, p. 69)
- 2 For expressivist proposals, see, e.g., Wittgenstein (1953), Finkelstein (2003), Bar-On (2004), Fre (...)
3Considerations such as these (paired with the idea that avowals exhibit some kind of first-person authority) have led some to propose avowal expressivism: the thesis that serious and competent utterances of the form “I ψ ((that p)/(o))” are typically expressive of the (speaker’s) first-order mental state of ψ-ing. On expressivist views, with a serious and competent utterance of “I believe that it’s raining”, “I am annoyed” or “I like you” the speaker typically expresses a first-order belief about the weather, annoyance or affection for the addressee, respectively.2
- 3 Rosenthal uses “I think that p”. Since he seems to use thought in the sense of belief, I use “beli (...)
4Rosenthal admits (as do other descriptivists) that we often use “I believe that p”3 instead of “p” – we use the two interchangeably. While the expressivist takes this to show that, in these cases, a speaker’s avowal “I believe that p” serves the same expressive function as her utterance of “p”, namely expressing the speaker’s belief that p, Rosenthal denies this thesis. He wants to uphold avowal descriptivism, yet concludes that utterances of “p” and “I believe that p” are performance-conditionally equivalent:
- 4 According to Rosenthal, the relevant qualification is that “the sentence ‘I think it’s raining’ ca (...)
Let’s assume the commonsense view that the speech act ‘I think it’s raining’ reports, but does not also express, one’s thought that it’s raining, and that the remark, ‘It’s raining’, does express that thought. […] Nevertheless, the two remarks have, with a qualification that does not matter for our purposes, the same conditions of use, that is, the same conditions for the appropriate performance of speech acts that use those sentences. The speech acts are, we can say, performance conditionally equivalent. (Rosenthal, 2010, p. 28)4
- 5 I assume that Rosenthal uses “saying” in the thick sense of the word, meaning “asserting” and not (...)
5He generalizes from this instance and claims that there is a performance-conditional equivalence “between saying that p and saying that one thinks that p” (Rosenthal, 2010, pp. 31-32). Thus, Rosenthal claims that assertions5 of sentences of the form
(1) “p”,
and
(2) “I believe that p”
are performance-conditionally equivalent. He bases this generalization on utterances of
(3) “It’s raining”,
and
(4) “I believe that it’s raining”.
6While serious and competent utterances of (3) and (4) can be, and indeed often are, used interchangeably – they often serve the same function in a discourse – I will argue that assertions of forms (1) and (2) are not performance-conditionally equivalent. After briefly introducing Rosenthal’s performance-conditional equivalence thesis in Section 1, I present counterexamples to this generalization in Section 2. Based on Freitag’s theory of explicit expressives, Section 3 explains what goes wrong in the counterexamples and offers an expressivist explanation of the specific performance-conditional equivalence of utterances of (3) and (4). The discussion strongly suggests a necessary condition for the performance-conditional equivalence of utterances of sentences of forms (1) and (2): the respective expressive acts need to have identical expressive content; that is, they need to be expressive of the same mental state. That avowal expressivism explains performance-conditional equivalence has also been noted by Dorit Bar-On (2010, pp. 54-55, 59). My expressivist account of performance-conditional equivalence is in the same spirit, and I aim to support this thesis by providing examples where the performance-conditional equivalence thesis and the assumption of avowal descriptivism lead one astray. I conclude that the performance-conditional equivalence thesis does not hold under a descriptivist analysis of avowals but rather requires avowal expressivism.
7Let’s start with serious and competent utterances of (3) and (4). Usually, “I believe that it’s raining” functions to convey the same information as “It’s raining”: With both utterances the speaker conveys information about the weather – that it is raining – as well as about herself, namely that she believes that it’s raining. “I believe that it’s raining” and “It’s raining” can thus often be used interchangeably. We can similarly observe that “Ouch!” and “I am in pain” (or “It hurts”) can usually be used interchangeably, just as much as “Thank you!” and “I am grateful” (Rosenthal, 2010, p. 28; 1986, p. 74). Based on these observations, and with avowal descriptivism in the background, Rosenthal proposes the following principle of performance-conditional equivalence:
- 6 This is a slightly modified version of how Rosenthal puts it in 1995, p. 322. Compare Searle (1979 (...)
(P) (Typically,) speech acts and assertions of self-ascriptions of those mental states of which the respective speech act is expressive are (roughly) performance-conditionally equivalent.6
8The performance-conditional equivalence described in (P) is qualified in two ways. First, there is a restriction to typical contexts. For instance, sometimes one might say “I believe that p” without being willing or able to say “p”. Consider the following case of psychotherapy: After several sessions, her psychotherapist has finally convinced Sally that she believes that her brother is angry with her (e.g., as a result of some quarrel back in their childhood). Sally now believes that she believes that her brother is angry with her. Thus, Sally would aptly report on her mental state by asserting “I believe that my brother is angry with me”, thereby expressing her higher-order belief that she believes that her brother is angry with her. She would, however, not assert “My brother is angry with me”, for the therapist has also convinced her that, although she has this deep-rooted belief, that belief is mistaken and her brother is in fact not (or no longer) angry with her (for a similar example, compare also Bar-On, 2010, p. 54). Rosenthal cannot explain this mismatch but just dismisses it by claiming that these contexts are quite rare and restricts performance-conditional equivalence to typical contexts (2010, p. 29).
- 7 For an overview of mitigation, compare Thaler (2012). For an overview of hedged assertion, compare (...)
9Second, according to (P), if the two acts are performance-conditionally equivalent, they are not fully but only roughly equivalent. This restriction is supposed to account for the phenomenon of hedging. For instance, if “I believe that p” is used to assert that p, which it can be but need not be, the assertion that p has only weakened assertive force – it is a hedged assertion.7
- 8 Let’s pause here. While (P) seems plausible for the pairs “Thank you!” and “I am grateful (to you) (...)
10It’s not entirely clear how Rosenthal explains the performance-conditional equivalence specified in (P). Using a factive, causal notion of expression (1986, p. 78; 1993, p. 56; 1995, pp. 316, 324; 2010, p. 25), Rosenthal holds that sincere speech acts express those mental states by which they are (partly) caused. A speaker’s sincere “Thank you!” expresses gratitude, a speaker’s sincere promise “I’ll be there” expresses an intention to show up and a speaker’s sincere “Ouch!” expresses pain. In light of this correspondence between (sincere) illocutionary acts and expressions of mental states, Rosenthal assigns the corresponding self-ascriptions of mental states the same use conditions (Rosenthal, 2010, p. 32): Whenever one says “Thank you!”, one is “roughly equally disposed to say” (p. 29) “I am grateful”, and whenever one promises to be there, one might just as well have said “I intend to be there”.8 Nevertheless, according to Rosenthal, serious and competent utterances of self-ascriptions of mental states have assertive force and thus express the speaker’s belief that she herself is in the mental state named. So, while, for instance, “I am grateful” can be used instead of “Thank you!”, the former does not function as an expression of gratitude but is expressive of the speaker’s belief that she herself is grateful.
11With (P) and avowal descriptivism in the background, Rosenthal concludes that assertions of forms (1) and (2) are generally performance-conditionally equivalent:
[T]he speech acts of asserting that p and asserting that I think that p, though they differ in respect to their truth conditions, have roughly the same conditions of assertibility. Any circumstances in which I could say that p are circumstances in which I could say I think that p. (Rosenthal, 1995, p. 320)
12This generalization is incorrect. In the next section, I provide examples of utterances of sentences of forms (1) and (2) for which (P) together with the assumption of avowal descriptivism predicts performance-conditional equivalence even though, in the most plausible contexts, the respective pairs are not performance-conditionally equivalent. (P) together with the assumption of avowal descriptivism predicts performance-conditional equivalence where there is none.
13Far from denying the connection between certain illocutionary acts and the expression of corresponding mental states, I think it is precisely because of this connection that Rosenthal’s general thesis of the performance-conditional equivalence of assertions of sentences of forms (1) and (2) is incorrect. Utterances of sentences of forms (1) and (2) are performance-conditionally equivalent only if the utterance of an indicative sentence of form “p” is indeed used as an assertion, or at least as an expression of the belief that p, and the utterance of (2) serves to express that very same belief. Only if that’s the case, as it usually is for the pair of (3) and (4), can sentences of forms (1) and (2) be used interchangeably. However, it is not the case that all serious and competent utterances of an indicative sentence “p” are assertions, or even expressions of a belief that p. This is especially vivid in those cases where “p” is a present-tense mental state self-ascription. Consider the following pairs:
(5) “I like you.”
(6) “I believe I like you.”
(7) “I am afraid.”
(8) “I believe I am afraid.”
(9) “I hope that it’s raining.”
(10) “I believe I hope that it’s raining.”
- 9 Consider also the following example inspired by a remark of Wittgenstein’s. Wittgenstein notes tha (...)
14All three pairs have the syntactical form of pair (1) and (2). Thus, if Rosenthal’s avowal descriptivism is correct and if (P) holds for assertions of sentences of forms (1) and (2), then (P) also holds for the above pairs. That is, it should be possible to use the members of each pair of sentences interchangeably; their utterances should be performance-conditionally equivalent. But utterances of the members of these pairs are not performance-conditionally equivalent – at least not typically.9
15Consider this for (5) and (6). A newly won friend wants to hear me say (5) “I like you” and is rather disappointed if I just tell her (6) “I believe I like you”. What she wants to be assured of is that I like her and not merely that I believe I like her. After all, I might be mistaken in my belief and not like her at all. Sensitive to this possibility, she will only take assurance of my affection for her from an utterance of (5) but not of (6) – even if (6) does not convey any hesitance on my part. And if I repeatedly use (6) instead of (5), showing no regard for the difference, she might become exceedingly cross with me (compare Williams, 2013, p. 1126, for similar examples). (Similar considerations hold for the other pairs.)
16If serious and competent utterances of (5) and (6) are not performance-conditionally equivalent and the pair has the form of pair (1) and (2), then assertions of sentences of forms (1) and (2) are not performance-conditionally equivalent – or utterances of self-ascriptions of mental states are not (always) assertions (without any first-order expressive content) after all. Thus, given the lack of performance-conditional equivalence for utterances of (5) and (6), (P) needs to be restricted or avowal descriptivism must be given up.
- 10 One way to go might be to insist that sameness or difference in performance-conditions (also) rest (...)
17Some might attempt to save avowal descriptivism by reference to hedging. They might argue that with (6) the speaker is weakly asserting that she likes the addressee. There are two ways to go from here: Either one insists that utterances of (5) and (6) are performance-conditionally equivalent after all – with (6) one just asserts shyly that one likes the addressee. Or one agrees that utterances of (5) and (6) are usually not performance-conditionally equivalent but attributes this to the latter functioning as a hedged assertion that the speaker likes the addressee. (Corresponding considerations hold for (7) and (8), and (9) and (10).) Let’s assume that the avowal descriptivist grants that utterances of (5) and (6) are typically not performance-conditionally equivalent. While there is indeed some weakening of illocutionary force if (6) is used as an assertion that one likes the addressee, the same holds for assertions of (2) and (4). Anyone taking this route while agreeing that utterances of (5) and (6) are typically not performance-conditionally equivalent and upholding (a restricted version of) (P), would need to explain why hedging or hesitation precludes performance-conditional equivalence for utterances of (5) and (6) but not of (3) and (4).10
18In light of counterexamples (5)–(10), I propose to give up avowal descriptivism, which also leads to a modification of (P). In the next section, I offer an expressivist explanation for why utterances of (5) and (6) are usually not performance-conditionally equivalent while utterances of (3) and (4) are. The problem with utterances of (5) and (6) is not that the speaker makes assertions with the same content, albeit of different assertive strengths (here I echo Rosenthal, 2010, p. 28, who claims that “degrees of confidence won’t be relevant”). Instead, utterances of (5) and (6), irrespective of the strength of their respective illocutionary force, if they function as illocutionary acts at all, usually have different expressive contents. That’s why they are typically not performance-conditionally equivalent.
19The counterexamples strongly suggest that avowal descriptivism and, thus, also (P), as specified above, are incorrect. While there is, under certain circumstances, a performance-conditional equivalence between speech acts and serious and competent utterances of present-tense self-ascriptions of those mental states of which the respective speech acts are expressive, the lack of performance-conditional equivalence in pairs (5)–(10) points to the fact that, often, serious and competent utterances of self-ascriptions of mental states are simply not assertions about one’s mental states (or they are, at least, also expressive of the mental state named). Avowal descriptivism is incorrect and (P) needs to be modified.
- 11 In many respects, Freitag’s theory of explicit expressive acts shares the spirit of Bar-On’s neo-e (...)
20I propose to go expressivist. The central expressivist thesis with respect to avowals – going back at least to Wittgenstein – is that in uttering a present-tense self-ascription of a mental state, the speaker typically expresses the first-order mental state named (see, e.g., Bar-On, 2004, p. 330). As Wittgenstein holds, in affirming “I hope he’ll come” (1953, PI §585), a speaker expresses hope, with “I am in pain” (1953, PI §244) she expresses pain, and with “I am afraid” (1953, PI II ix) she expresses fear. I will now give a very rough sketch of Wolfgang Freitag’s theory of expressives, which I follow in this paper (for a more comprehensive introduction, see Freitag, 2018; Freitag and Yolcu, 2021; and Freitag and Kraus, 2022).11
- 12 Some, however, hold that expression is factive and maintain that with an insincere illocutionary a (...)
21Given classical speech act theory (Austin, 1962; Searle, 1969), linguistic acts generally have an illocutionary and an expressive dimension. A serious and competent utterance of “It is raining”, for instance, provides twofold information. The speaker describes the world – she asserts that it is raining. But she also gives information about herself – she expresses a belief that it is raining. A serious and competent utterance of “It is raining” thus also functions as a non-explicit expressive act (Freitag, 2018, p. 59). So far, that’s standard and fairly uncontroversial.12 The theory of expressives takes this part of expressivism as given.
22The central expressivist thesis is that there are also explicit expressive acts (Freitag, 2018, p. 59): utterances in which the mental state expressed is made explicit in the utterance. Let’s consider this for doxastic avowals. A serious and competent utterance of “I believe that it is raining” typically gives the same two-fold information as “It is raining”. With “I believe that it is raining” the speaker typically expresses a belief that it is raining as a direct, explicit expressive act and she might assert that it is raining as an indirect illocutionary act. With both utterances – “It is raining” and “I believe that it is raining” – the speaker expresses a belief that it is raining; by uttering “It is raining” she does so non-explicitly and by uttering “I believe that it is raining” she does so explicitly.
- 13 Compare Searle (1983, p. 6).
- 14 Since, on my view, expression is non-factive, the speaker might not be in the mental state express (...)
23Expressive acts have expressive contents: in performing an expressive act, the speaker expresses a mental state. Expressive contents C are made up of a psychological mode13 and a mental state content (at least for intentional mental states). For instance, with β for belief, p for that it is raining and S indicating the speaker, the non-explicit expressive act performed in seriously and competently uttering “It is raining” and the explicit expressive act performed with “I believe that it is raining” have the same expressive content, namely βS(p).14 Similarly, with γ representing gratitude, the non-explicit expressive act performed with “Thank you!” and the explicit expressive act performed with “I am grateful to you” both have γS(you) as their expressive contents. And, with π representing pain, the non-explicit expressive “Ouch!” and the explicit expressive “I am in pain” have identical expressive contents, namely πS.
- 15 Note that avowal expressivism, as presented here, is not a semantic thesis. The semantic content o (...)
24Of course, Freitag’s theory of expressives does not deny that self-ascriptions of mental states can be used to report on one’s own mental states. Self-ascriptions of mental states of the form “I ψ ((that p)/(o))” have two uses: depending on context, they can be used as explicit expressive acts or as non-explicit expressive acts (Freitag, 2018, p. 66; compare Bar-On, 2004, p. 303). In the latter case, the relevant avowal is indeed expressive of the belief that the speaker is ψ-ing and functions as an assertion that she is ψ-ing.15
25Armed with the theory of (explicit) expressive acts, let’s return to the (alleged) performance-conditional equivalence of serious and competent utterances of sentences of forms (1) and (2). What goes wrong for utterances of (5) and (6)? Typically, utterances of (5) and (6) cannot (adequately) be used interchangeably, for they have different expressive contents. With a serious and competent utterance of (5) “I like you” the speaker usually does not assert that she likes the addressee. Nor does she express a belief that she likes the addressee. Instead, “I like you” serves to express the speaker’s liking the addressee. An utterance of (6) “I believe I like you”, however, does not serve to express the speaker’s liking the addressee. Instead, in its explicit-expressive use it is expressive of the speaker’s belief that she likes the addressee, and in its non-explicit-expressive use it is expressive of the speaker’s second-order belief that she herself believes that she likes the addressee. That’s why serious and competent utterances of (5) and (6) are typically not performance-conditionally equivalent: they usually function as expressive acts with different expressive contents.
26Avowal expressivism does not only explain why utterances of (5) and (6) are usually not performance-conditionally equivalent. It can also account for cases where (5) and (6) can indeed be used interchangeably (these are admittedly very special contexts). Utterances of (5) and (6) are performance-conditionally equivalent only if (5) is used as a non-explicit expressive act expressive of the speaker’s belief that she likes the addressee and (6) is used as an explicit expressive act expressive of the very same belief. Only then do utterances of (5) and (6) serve to express the same expressive content. (A similar reasoning applies to (7) and (8), and (9) and (10).)
27This analysis carries over to utterances of (3) and (4). They are performance-conditionally equivalent only if (3) is used as a non-explicit expressive act expressive of the belief that it is raining and (4) is used as an explicit expressive act expressive of the belief that it is raining as well. If (4), however, is used as a non-explicit expressive act expressive of the speaker’s belief that she herself believes that it is raining, the expressive contents of utterances (3) and (4) diverge and the two utterances are not performance-conditionally equivalent.
- 16 Here, I focus on the expressive dimension of speech. Performance-conditional equivalence might als (...)
28The expressivist proposal, thus, has more explanatory power than (P) and avowal descriptivism. It can explain when utterances of sentences of forms (1) and (2) are performance-conditionally equivalent and when they are not, as it is usually the case for utterances of (5) and (6). The discussion strongly suggests that utterances of sentences of forms (1) and (2) are performance-conditionally equivalent only if the utterance of a sentence of form (1) is an expression of the belief that p and the utterance of a sentence of form (2) functions to express that very same belief. That is, utterances of sentences of forms (1) and (2) are performance-conditionally equivalent only if (1) is used as a non-explicit expressive act, expressing the speaker’s belief that p, and (2) is used as an explicit expressive act, also expressing the speaker’s belief that p. Only if an utterance of (2) makes explicit what is implicit in an utterance of (1), namely that the speaker believes that p, can (1) and (2) be used interchangeably – their utterances then function to convey the same information: that the speaker believes that p.16
29But if an utterance of a sentence of form (1) does not serve to express the speaker’s belief that p but a different mental state, as is usually the case for utterances of (5), (7) and (9), and that mental state differs from the one expressed when uttering a sentence of form (2), then the two utterances are not performance-conditionally equivalent. Accordingly, I propose to revise (P):
(P*) Speech acts and serious and competent utterances of self-ascriptions of those mental states of which the respective speech act is expressive are (roughly) performance-conditionally equivalent only if their expressive contents are identical.
- 17 (P*) generalizes beyond the doxastic case. With the same structure, we can give a necessary condit (...)
30Performance-conditional equivalence of speech acts of forms (1) and (2) requires identical expressive content. The expressive contents of serious and competent utterances of sentences of forms (1) and (2) are identical if and only if the former is a non-explicit and the latter an explicit expressive act. Hence, performance-conditional equivalence of speech acts of forms (1) and (2) requires avowal expressivism.17
- 18 If the mark of performance-conditional equivalence is identical expressive content, we might not n (...)
31I have presented counterexamples against the general performance-conditional equivalence of assertions of forms (1) and (2). In response, I have offered an expressivist way of distinguishing between utterances of sentences of forms (1) and (2) which are performance-conditionally equivalent and such which are not. Only if the expressive content of the expressive act performed with an utterance of a sentence of form (1) matches that of the expressive act performed with an utterance of a sentence of form (2) are the two utterances performance-conditionally equivalent.18 The phenomenon that utterances of (3) and (4) can often be used interchangeably, which Rosenthal correctly identifies, can thus be explained by reference to a theory of expressive acts: With a serious and competent utterance of (3) the speaker non-explicitly expresses the belief that it’s raining. And an utterance of (4) usually serves to express that very same belief explicitly. Only if we have this match of implicit and explicit expression of the very same belief are utterances of (3) and (4) performance-conditionally equivalent. Thus, avowal expressivism does not only explain the frequent interchangeability of speech acts of forms (1) and (2) and gives a necessary condition for their performance-conditional equivalence, but the thesis of performance-conditional equivalence even provides an argument in favor of avowals having first-order expressive content.
RSS feed
