, Volume 74, Issue 1, pp 115–129

Misleading Appearances: Searle on Assertion and Meaning


    • F.R.S.-FNRS, Laboratoire de linguistique textuelle et de pragmatique cognitiveUniversité Libre de Bruxelles
Original Research

DOI: 10.1007/s10670-010-9229-z

Cite this article as:
Kissine, M. Erkenn (2011) 74: 115. doi:10.1007/s10670-010-9229-z


John Searle’s philosophy of language contains a notorious tension between a literalist view on the relationship between sentences and their meanings, and what—at the first glance—appears to be a virulent defence of contextualism. Appearances notwithstanding, Searle’s views on background and meaning are closer to literalism than to contextualism. Searle defines assertion in terms of the commitment to the truth of the propositional content. In absence of an independent criterion to delimit the asserted content, such a definition overgenerates—hence Searle’s commitment to literalism. His position is untenable—and this is the general lesson of the paper—, because sentence meaning cannot be used to determine the asserted content.

1 Introduction

This paper aims at clarifying some important conceptual issues underlying contemporary debates about the relationship between sentence meanings and speech act contents. Since, ultimately, it is the very notion of ‘literal meaning’ that proves to be the matter of dispute, it is wise to begin by laying down the three conceptions of this term that are explored in this paper.

The first one is defended by Searle (1969). According to him, the literal meaning of an utterance corresponds to the speech act any literal and felicitous utterance of the uttered sentence would constitute. What the speaker S means depends on which speech act she is performing by her utterance. If S is speaking literally, this speech act matches the conventional meaning of the corresponding sentence type. The literal meanings of sentence tokens are thus matter of speakers’ intentions, but, at the same time, they are determined by linguistic conventions.

The second conception is defended by ‘literalist’ theories of Cappelen and Lepore (2005) or Soames (2002). Here the utterance’s literal meaning does not depend on the speech act the speaker is performing. The literal meaning of sentence tokens is entirely determined by the semantic interpretation of the corresponding types.1 The crux of such theories is that all tokens of the same sentence type share one and the same literal meaning—the same semantic content—, even though these tokens correspond to the performance of speech acts with distinct content and/or force.

Finally, ‘contextualist’ theories, like Sperber and Wilson (1995), Carston (2002) or Recanati (2004), claim that the structure of sentence types radically undetermines the literal meanings of their tokens. The literal meaning of any utterance depends on extra-linguistic factors. This, we will see, seems to also be Searle’s (1978) position.

We thus find Searle in two theoretical trends. According to the first, whenever the speaker uses a sentence literally, what she means (the literal speech act she performs) is determined by sentence meaning. According to the second, sentence meaning radically undermines what the speaker literally means on any occasion of utterance. For this reason, Searle’s philosophy of language has been said to contain a notorious tension between a minimalist, ‘literalist’ view of the relationship between sentences and their meanings, and what appears to be a virulent defence of contextualism (see e.g. Carston 2002, pp. 69–70; Recanati 2003). Yet, opposing previous interpretations, I will claim that, appearances notwithstanding, Searle’s allegedly contextualist credo conceals a staunched literalist hypothesis, in that it requires his Expressibility Principle—discussed below—to be interpreted in the strongest way, viz. that the meaning of every sentence token has a literal meaning that is not affected by extra-linguistic facts.

This view is different from literalist theories of Cappelen and Lepore or Soames. For these authors, the literal meaning of any sentence token corresponds to the output of the semantic interpretation of its syntactic structure; literal meaning, in these theories, differs from speech act content. Limiting the discussion to assertion, for literalists the linguistic structure of sentences does not determine the content of assertions utterances of these sentences correspond to. By contrast,—and this a point that has been scarcely touched upon in the literature—Searle’s theory of assertion forces him to accept that it is the content of literal assertions that is entirely determined by the linguistic structure of the uttered sentence. The reason for this, I will show, is that Searle takes the commitment to the truth of the propositional content as definiens in his essential condition of assertion.

I should like to note that the aim of this paper does not reduce to offering a new exegesis of Searle’s philosophy—although a clarification of these intricate aspects of his writings is welcome per se. Discussing Searle’s theory of assertion and meaning will lead to the following, more general claims:
  1. (a)

    Meaning dependence on the background and hypotheses like Searle’s Expressipility Principle are compatible only if literalism is applied to semantic content, leaving the determination of the asserted content to pragmatic factors;

  2. (b)

    Defining assertion in terms of truth-commitment requires an independent method in order to determine the asserted content. Searle uses linguistic meaning—the structure of sentence types—to this end. However, the dependence of utterance interpretation on the background—which is acknowledged by literalists and contextualists alike—is incompatible with this strategy.


Amid the contemporary wars about the nature of semantic content and of asserted content, I believe this clarification of the logical space is very much worth making. Furthermore, the point in (b) puts an important constraint on theories that define assertion in terms of truth-commitment with respect to their propositional content: such approaches need to find a method to determine the asserted content that does not rely on linguistic meaning (and that does not make truth-commitment quadefiniens redundant).

Here is how I will proceed. The next section starts by a brief presentation of Searle’s Principle of Expressibility; combining it with Searle’s view on the background of meaning will lead me to argue that, under its most coherent interpretation, this Principle grounds a literalist theory of utterance meaning. In Sect. 3 I will discuss Searle’s theory of assertion; it will be argued that, as such, Searle’s definition of assertion does not allow one to delimit the asserted content adequately. In Sect. 4 we will see that Searle’s view implies that the content of any literal assertion is determined by sentence meaning. In Sect. 5, I will argue that Searle’s position leads to absurd consequences, because literalism cannot be applied to the content of speech acts.

2 The Principle of Expressibility and the Background of Meaning

Since Speech Acts, Searle defends the Principle of Expressibility:

[…] for any speaker S whenever S means (intends to convey, wishes to communicate in an utterance, etc.) X then it is possible that there is some expression E such that E is an exact expression of or formulation of X. (Searle 1969, p. 20).

The claim seems to be that it is possible, although not necessary, that every intention to communicate a certain content can be satisfied by uttering a sentence which is the exact expression of this content. The following passage corroborates such a reading.

Of course, a given language may not be rich enough to enable speakers to say everything they mean, but there are no barriers in principle to enriching it. (Searle 1969, p. 68)

In other terms, any human language is, in principle, powerful enough to provide the speaker S willing to express a certain meaning T with T’s exact expression E. In some cases, E does not belong to the language L; however, if L is a natural language, L’s structure allows it to be enriched with E. Furthermore, even though E already belongs to L, it is not necessary that the speaker S will be able to use E. In other terms, given S’s intention to express T, the existence of an actual or a possible sequence E of L qua the exact expression of T is necessary; what is not, is S’s capacity to use E in order to satisfy her intention, wish,… to express T.
In Searle’s view, saying that literal meaning of S’s utterance is T boils down to saying that the speech act S performed by her utterance is the one determined by the linguistic meaning of the uttered sentence. In other words, literal meanings correspond to speech acts. Accordingly, the Principle of Expressibility applies to the relationship between linguistic elements and the speech acts they serve to perform.

[The Principle of Expressibility] enables us to equate rules for performing speech acts with rules for uttering certain linguistic elements, since for any possible speech act there is a possible linguistic element the meaning of which (given the context of the utterance) is sufficient to determine that its literal utterance is a performance of precisely that speech act. (Searle 1969, pp. 20–21)

At this point, it is important to emphasise that, thus interpreted, the Expressibility Principle can still be read in (at least) two ways. (For comprehensive discussions, see Recanati 1987, pp. 219–224, 2001, 2003, 2004, pp. 83–86; Carston 1988, 2002, pp. 30–42, 64–70.) The first, ‘absolute’ reading does not relativise the Principle to contexts of use. If what is literally meant by an utterance of E is T (i.e. a certain speech act), viz. if E is the exact expression of T, then the literal meaning of any utterance of E will be T, whatever the context of use.

The second reading relativises the Expressibility Principle to contexts of use. For any meaning TS would be willing to express (viz. for any speech act S would be willing to perform), and for a context C1, the language L allows, in principle, to construct an expression E, such that E is the exact expression of T in C1. However, nothing guarantees that in another context C2E would still be adequate for allowing S to literally express T (viz. to perform the same speech act as in C1).

The surprising observation is that, in some other writings, Searle seems to be virulently opposed to such interpretations of the Expressibility Principle. To take one of his favourite examples, consider the two occurrences of the word cut in ‘Bill cuts the grass’ and in ‘Sally cuts the cake’:

Though the occurrence of the word ‘cut’ is literal in [both] utterances …, and though the word is not ambiguous, it determines different sets of truth-conditions for different sentences. … One way to see this is to imagine what constitutes obeying the order to cut something. If someone tells me to cut the grass and I rush out and stab it with a knife, or if I am ordered to cut the cake and I run over it with a lawnmower, in each case I will have failed to obey the order. That is not what the speaker meant by his literal and serious utterance of the sentence. (Searle 1980, pp. 222–223)

Searle (1980, p. 222) points out a sentence like (1) is perfectly acceptable.
  1. (1)

    General Electrics has just announced the development of a new cutting machine that can cut grass, hair, cakes, skin, and cloth.

(1) is a good piece of evidence that cut is not polysemous. Take, for instance, Cruse’s (1986, p. 21) textbook example of polysemy (2).
  1. (2)

    Arthur and his driving license expired last Thursday.

Unlike (1), (2) sounds like a pun—a zeugma—because expire can be taken as a different lexical entry when predicated about a human or about an official document. By contrast, in (1), cut—as used in cutting the grass, cutting the cloth, etc.—is not ambiguous between different lexical entries, even though its denotation is contextually influenced. A related argument is that, in contrast with (1), it is impossible to translate (2) in many other languages (for instance, in Russian); this reveals, once again, that expire in (2) is linguistically ambiguous while cut in (1) is not. This is not to say that English verb cut is not polysemous; arguably, cut corresponds to distinct lexical entries in cutting the grass and cutting the prices. The point is that there is no polysemy involved in (1). Compare (1)—where no polysemy is involved—with (3), where cut is polysemous, and which is similar to (2).
  1. (3)

    Arthur cut the grass in his garden and the prices in his shop.2

Furthermore, the acceptability of (1) shows that the context of utterance does not determine one single content for the occurrence of cut within a given sentence, as happens, for instance, with the indexical I (see Cappelen and Lepore 2005, p. 102, fnt). In (4), I is assigned one single referent which is the agent of waking up, getting out of bed, etc.
  1. (4)

    I woke up, got out of bed, dragged a comb across my head…

In (1), by contrast, cut is not assigned one single denotation: cutting bread is still different from cutting hair.
Could we explain away the alleged indeterminacy of cut by lexical composition? The claim would be that cut has no full-fledged meaning in isolation, but acquires one by composing with an NP. Under such a conception, in (1) each elided occurrence of cut would be said to constitute, in conjunction with the corresponding NP, a distinct semantic unit. However, Searle emphasises that the background assumption about what it means to cut the grass, as opposed to cutting the cake, can be cancelled.

Suppose you and I run a sod farm where we sell strips of grass turf to people who want a lawn in hurry …. Suppose I say to you ‘Cut half an acre of grass for this customer’; I might mean not that you should mow it, but that you should slice it into strips as you could cut a cake …. (Searle 1980, pp. 224–225)

The cancellation of such background assumptions is an important obstacle to the semantic solution just outlined, for it entails that cutthe grass does not always denote the same kind of activity (see also Recanati 2004, pp. 94–95).

It follows that (a) cut can express different meanings across contexts of use (e.g. as applied to grass or to bread); (b) that for a given context of use cut can express different meanings (e.g. as applied to grass in normal circumstances vs. as applied to grass in a sod farm).

Generalised to every expression, (a) undermines the ‘absolute’ version of the Expressibility Principle, and (b) its second, ‘context-relative’ weaker version. Take again the indeterminacy of the meaning expressed by cut. If the Expressibility Principle, under the readings considered so far, is correct, it follows that there is no meaning such that, for instance, ‘Bill cuts the grass’, is its exact expression, for the meaning of cut is undetermined for any context of use.

However, as such, examples like the ‘cutting the grass’ one do not falsify the Expressibility Principle. As we have seen, this principle entails that to every meaning we intend to express corresponds a (possible) exact linguistic expression. But, so far, nothing implies that the converse is true. If an utterance has a literal meaning T, the Principle of Expressibility does not entail that (in the context of conversation) the uttered sentence is the exact expression of T.

A reviewer put forth an objection along precisely these lines. “Many sentences can be used on different occasions to express different meanings. This is the norm, not some exception. However, given a fixed background, it should be possible to formulate (to articulate) a sentence which in the circumstances exactly expresses one literal meaning. Perhaps, with respect to a given background, for any meaning that can be intended, some sentence can be formed which exactly expresses that meaning. Those sentences will be uncommon, and cumbersome […].”

Such an idea faces a quite well known problem of finding sentences corresponding to the cases where the utterance’s meaning is not exactly expressed by the sentence uttered.3 What is the exact expression corresponding to what is meant by ‘Bill cuts bread’ (uttered in a context where we can safely assume that if Bill cuts the bread, he does it in a usual way)? ‘Bill cuts the bread with a knife’? But then there exist fake knives, which are not suitable for cutting bread. ‘Bill cuts the bread with a knife suited for cutting the bread’? But what counts as suitable for cutting the bread depends on many background assumptions. And so on, and so forth. In fact, Searle’s point is precisely that there are no such paraphrases; furthermore, he intends his point about the indeterminacy of meaning to be extended to every natural language sentence (see, especially Searle 1978).

To see this, let us take another of his examples. Searle (1978) famously observes that an utterance of (5) seem to make sense only if some facts about gravitation belong to the background.
  1. (5)

    The cat is on the mat.

If the cat and the mat float in outer space, outside the Milky Way galaxy, the truth-valuation of (5) is (almost) impossible.
First, assumptions about gravitation do not belong to the literal content of (5), nor assimilate to propositional presuppositions whose truth is required for (5) to be interpretable. The background can be manipulated in such a way as to insure interpretability even in an environment without gravitation.

… as we are strapped in the seats of our space ship in outer space, we see a series of cat-mat pairs floating past our window. Oddly, they come in only two attitudes [from our perspective either the cat is just above the mat, or the cat is floating upside down, just under the mat]. “Which is it now?” I ask. “The cat is on the mat”, you answer. Have you not said exactly and literally what you meant? (Searle 1978, p. 212)

Second, Searle argues that background assumptions cannot be enumerated. He gives the example of a cat that is suspended to a system of strings in such a way as it is in contact with the mat, but without exerting any pressure on it. Without a specific background—suspending the assumption that the cat exerts on the mat the force corresponding to his weight (on the Earth)—it becomes again impossible to interpret (4).

The upshot of the foregoing is that it is impossible to build a sentence such that, for a given context, it would express exactly the meaning of (4) in that context. Such a sentence should comprise all the background assumptions that determine what S means by (4). Yet these assumptions do not belong to the content of (4), nor can they be explicitly enumerated. In other words, the background cannot be propositional (see also Carston 2002, pp. 69–70).4 (This is certainly the reason why Searle positions his notion of background in line with Wittgenstein’s (1989) writings on certitude, whose central theme is that in the domain of practical knowledge doubt is unconceivable.)

To be sure, the Expressibility Principle is compatible with the hypothesis that even though there is no principled obstacle to constructing the exact linguistic expression of any meaning, nobody has managed so far to generate any exact expression using natural language—as is demonstrated by the radical dependence of utterance interpretation on the background. Obviously, such a hypothesis would be unfalsifiable, and offers little interest for the study of natural language.

However, there is another, stronger interpretation of the Principle of Expressibility, which can solve the apparent tension exposed in the foregoing. (I should like to stress that I am not saying that this is Searle’s position; incidentally, neither do I believe that views like the one to be exposed now are correct (cf. Kissine 2007, forthcoming).) Under this stronger interpretation, any well-formed sentence E corresponds to the exact expression of a certain propositional meaning T.5 (Note that this still leaves open the possibility that some propositions cannot be expressed exactly.) In an adequate context, E can be used to express its exact and literal meaning T. However, even though E is used in order to convey some other meaning, or is used in a way that makes the propositional meaning T impossible to access, it does not follow that E’s literal meaning is not T anymore. This thesis would imply that in all of its occurrences, cut keeps its literal meaning—however, this literal meaning is not the meaning that is communicated by S. As for (5), the difficulty to interpret it in some contexts does not affect the exact meaning associated by the linguistic rules to the expression ‘The cat is on the mat’. Let us consider this, obviously radical view in more detail.

The doctrine according to which speakers do not necessarily know which proposition is literally expressed by their utterances has been recently popularised by Cappelen and Lepore (2005). These authors agree that examples like Searle’s ‘Cat on the mat’ show that in absence of implicit background assumptions we cannot decide what S meant by her utterance. However, for radical literalists like Cappelen and Lepore, it does not follow that the literal meaning of the sentence used depends on background assumptions, viz. on the context of utterance. Nothing implies, argue Cappelen and Lepore, that the absence of adequate background cancels the semantic content of, say, (5). Our intuitions of context dependency stem from the fact that, owing to the rupture with the background, it is totally unclear which meaning S intends to convey by using this sentence. However, for a radical literalist, the sentence in (5) still corresponds to the exact and literal expression of a certain meaning—the literal meaning which is determined by the structure of the sentence type ‘The cat is on the mat’, and which is inherited by the totality of its tokens. Now, what exactly such a meaning corresponds to in the real world is not, according to Cappelen and Lepore, the subject matter of semantics of natural language, and not the object of speakers’ intuitions, but rather the job of metaphysics.

Cappelen and Lepore explain away the intuition that, for instance, ‘Bill cuts the grass’ means different things in a context where Bill mows the lawn and in the sod farm scenario by arguing that infinity of propositions can be asserted by the tokens of the same sentence type. In each of different contexts of use, the utterance of ‘Bill cuts the grass’ corresponds to the assertion of a different proposition [Bill mows the lawn], [Bill slices off strips of grass], etc.; but in each of these contexts, the literal content, i.e. the proposition expressed semantically, is simply [Bill cuts the grass], whatever this means. According to radical literalism, speakers’ intuitions bear on speech act contents, not on semantic, literal contents.

To repeat, most probably, Searle would reject such an interpretation of his Expressibility Principle. Nevertheless, I believe it is worth considering for two reasons. First, even though Searle’s writings on the background—e.g. Searle (1978, 1980)—are often quoted as examples of hard-core contextualism, viz. of the claim that the content of any sentence token is radically undetermined by the linguistic structure of the corresponding type, it is instructive to see that once these views are examined in combination with the Expressibility Principle, it emerges that they are closer to radically literalist theories of the sort defended by Cappelen and Lepore (2005), than to contextualism (see Stanley 2005 for a similar suggestion). Second,—and this will be the topic of the next two sections, reading the Expressibility Principle as coupling well-formed expressions with literal meanings is, in fact, part and parcel of Searle’s theory of assertion. This, as we will see, makes Searle’s position even more radical than claiming that speakers have no intuitions about the literal, semantic content of the sentences they utter.

3 Searle’s Definition of Assertion

Searle (1969, p. 67) distinguishes between the essential condition of assertion, and its preparatory and sincerity conditions. In order to successfully assert that p, the speaker (S) must:
  • Perform by her utterance an undertaking to the effect that p represents an actual state of affairs [essential condition];

  • Have grounds or reason for believing that it is true that p, and that it is not mutually manifest that A knows that p [preparatory conditions];

  • Believe that p [sincerity conditions]

If the essential condition is violated, the utterance cannot be considered as an assertion. By contrast, if a preparatory condition and/or the sincerity condition are/is violated, the assertion is still successful, although defective (also Vanderveken 2005, pp. 717–718). For instance, the liar violates the sincerity condition, but her utterance is still a successful assertion, albeit a misleading one.6
According to Searle, an utterance u counts as an assertion, if, and only if, it counts as an undertaking to represent the world as it is. In other words, u commits S to the truth of a certain propositional content—the content p of the assertion Ass(p) that corresponds to u.7 The problem is that, as such, this view implies that S asserts every proposition the truth of which she commits herself to by u. As emphasised by Alston (2000, pp. 104, 117–118), if S asserts every proposition the truth of which she commits herself to by her utterance, there is an obvious danger of inflating the asserted content. For instance, as is revealed by Moore’s paradox, in asserting that pS publicly commits herself to believing that p. Yet, the content of an assertion that p is not p and S believes that p. There is an obvious reason for that. Take (6) and (7):
  1. (6)

    My car is parked in front of the library.

  2. (7)

    My car is parked in front of the library, and I believe that my car is parked in front of the library.

If by uttering (6), S asserts (7), then the sincerity conditions commit her to believing (7). Hence, if by uttering (6) she is committed to believe (7)—and if S assets every proposition the truth of which her utterance commits her to, then by uttering (6) S actually asserts (8). Infinite regress looms large.
  1. (8)

    My car is parked in front of the library, and I believe that my car is parked in front of the library, and I believe that I believe that my car is parked in front of the library.


To sum up, as it stands, Searle’s definition of assertion does not allow us to delimit the propositional content of assertions in an adequate way. It tells us that an utterance u is an assertion if, and only if, by uS commits herself to the truth of the asserted propositional content; yet, we do not know which proposition among those the truth of which S is committed to by u should count as the asserted content.

Alston (2000), who also defines assertion in terms of commitment to the truth of the propositional content, is well aware of this problem. His solution is to claim that every constituent of the asserted proposition maps directly onto a syntactic constituent of the sentence used.8 Transposing Alston’s claim in Searle’s terminology, if the Expressibility Principle means that every well-formed sentence type has a fixed meaning, inherited by the totality of its tokens, the propositional content of literal and direct assertions is determined by the linguistic structure of the sentence used. For instance, take the examples (6) and (7). Clearly, they instantiate two different sentence types. Therefore, if the propositional content of assertions is determined by the linguistic structure of the corresponding sentences, asserting (6) is not equivalent to asserting (7). Furthermore, even though by asserting the content in (6) S commits herself to the truth of the content in (7), nothing commits S to the truth of (8).

In the previous section I argued that Searle’s theory of background—when it is taken together with the Expressibility Principle—implies that the meaning of sentence tokens is entirely determined by the linguistic structure of the corresponding types. We see now that Searle—and, in fact, any theory of assertion based on the notion of truth-commitment—needs a clear distinction between the asserted content and other propositions the truth of which S’s assertion commits her to. In order to operate such a divide, Alston takes the structure of the sentence type to determine the propositional content of any literal assertion performed by each of its tokens.

This strategy is literalist in spirit, in that it links meaning to the linguistic structure. Yet, it differs in an important aspect from theories like Cappelen and Lepore’s. As we have seen in the previous section, Cappelen and Lepore (2005) or Soames (2002) explain away the contextual or background dependence of meaning by locating it at the level of the speech act performed; it is sentence meaning, not the asserted content, that is, in their view, determined by the linguistic structure. By contrast, claiming that the asserted content is determined by sentence meaning amounts to applying literalism to the speech act content. Before assessing the consequences of such a move, it is important to see why literalism is actually used by Searle in order to determine the asserted content.

4 Literalism and Assertion

For Searle, the conditions that govern assertion (and, more generally, any other illocutionary act) correspond to constitutive rules that underlie the use of sentences and fix their meanings.

S utters the sentence T and means it (i.e., means literally what he says) =

S utters T and
  1. (a)

    S intends (i-I) the utterance U of T to produce in H the knowledge (recognition, awareness) that the states of affairs specified by (certain of) the rules of T obtain. (Call this effect the illocutionary effect, IE.)

  2. (b)

    S intends U to produce IE by means of the recognition of i-I.

  3. (c)

    S intends that i-I will be recognised in virtue of (by means of) H’s knowledge of (certain of) the rules governing (the elements of) T. (Searle 1969, pp. 49–50)


As such, this does not mean that speaker meaning, recovered by the (ideal) addressee, corresponds to the sentence meaning fixed by linguistic conventions; it might be the case that sentence meaning is only a step towards recovering what the speaker meant. However, for Searle, speech acts constitute the minimal units of communication. In particular, as we have seen, the Expressibility Principle applies to sentence/speech act relationship; in any context where S uses the expression E with its exact and literal meaning, this utterance corresponds to the direct and literal performance of a particular speech act.

A speech act has, for Searle, the form F(p), F being the illocutionary force and p the propositional content. Limiting the discussion to assertions, the propositional content is determined by certain truth-conditions, and the assertive force by the fact that the satisfaction of these conditions requires the words to fit the world. The conventional meaning of a sentence typically used for asserting thus reduces to Ass(p), where p is the propositional content under the scope of the assertive force.9

To be sure, context is needed in order to know whether or not a speech act has been performed. From the fact that whenever an expression E is used to express its literal meaning, the corresponding utterance is the assertion that p, it does not follow that every utterance of E constitutes the assertion that p. What does follow, however, is that knowing which literal illocutionary act has been performed by an utterance of E is knowing the linguistic meaning of the uttered sentence.

Speaker’s meaning, according to Searle (1983, 2007), consists in the intentional imposition of conditions of satisfaction on utterances. An assertion that p is satisfied if, and only if, it fits the world; that is, if, and only if, p is true. The same conditions of satisfaction correspond to the belief that p. This belief is the belief that S must have if the sincerity conditions of her assertion that p are to be satisfied. However, the essential difference between beliefs and assertions is that only the latter commit S to the truth of the propositional content. This is so, according to Searle, because of the double role of linguistic conventions. Linguistic conventions—sentence meanings—provide S with a reliable and repeatable means to perform the assertion that p.

The speaker intentionally produces an utterance, he intends that the utterance should itself have conditions of satisfaction, for example truth-conditions. But, and this is the next crucial point, if he is to succeed on a regular basis, then there has to be some socially recognised conventional device, some repeatable device, the production of which can be regularly and conventionally taken by his interlocutors to convey the message. […] the first phenomenon is essential to the performance of speech acts, and the second phenomenon, the repeatable devices, consists typically of words and sentences of a language. (Searle 2007, p. 23, emphasis in the original text)

The conventional meaning, viz. the linguistic structure, enables the hearers to know which speech act has been performed whenever the utterance is serious and literal. Furthermore, by making the truth-conditions of the asserted proposition—of the corresponding belief—public, linguistic conventions commit S to the truth of this proposition. The linguistic meaning of a sentence determines the assertion to the performance of which the literal utterance of this sentence corresponds, and, thereby, the truth-commitment carried by this utterance.

To be sure, certain sentences can be used as the exact expression of several literal speech acts types; but if, as Searle contends, the performance of each of these speech acts corresponds to the expression of the exact meaning of the sentence E (or of the conjunction of sentences E1,… En), then E is paired with several literal meanings, i.e. E is linguistically ambiguous. However, saying that a sentence has several meanings because it is linguistically ambiguous is very different from saying that different tokens of the same sentence type can have different meanings because the literal meaning of each token depends on something else than its conventional meaning.

Cases of ambiguity thus set aside, it follows that whenever it is used appropriately and literally, every token of a (grammatically declarative) sentence type E constitutes the same literal assertion Ass(p). By her performing Ass(p), S commits herself to the truth of the propositional content p. Therefore, the literal utterance of any token of E commits S to the truth of one and the same proposition p. We see again that if the Expressibility Principle is interpreted as coupling sentence types with certain meanings, we have a criterion for identifying the content of assertions. It seems coherent to ascribe to Searle the view that the content of any literal and direct assertion is entirely determined by the conventional meaning of the sentence used.

5 Why Literalism About Assertion Content is Problematic

From the previous sections we see that Searle’s theory of assertion implies that the conventional meaning of every (declarative) sentence type determines a unique assertion type (ambiguity set aside). However, contemporary literalists, like Cappelen and Lepore (2005) and Soames (2002), maintain—and for good reasons—that the content of assertions (and more generally, of speech acts) is not determined by sentence meaning.

Let us consider a case where the absence of a relevant background renders the truth-valuation impossible. For instance, S says ‘The cat is on the mat’ in a context where it is impossible to decide whether this utterance is true or not. In such a case, as we have seen in Sect. 2, both Searle and literalists like Cappelen and Lepore have to accept that linguistic conventions still determine a literal meaning, but that speakers are denied epistemic access to it. But it is one thing to say—like Cappellen and Lepore—that this meaning is the semantic content associated with the sentence ‘The cat is on the mat’, and quite another one that the literal meaning of ‘The cat is on the mat’ is the assertion any literal and appropriate utterance of ‘The cat is on the mat’ corresponds to.

For Cappelen and Lepore sentence meaning corresponds to the semantic content that is determined by the linguistic rules, but whose truth-conditions may well lie out of (naïve) speaker epistemic reach. In other words, knowledge of the semantic content of a sentence does not suffice (and, possibly, is not even necessary) to understand the speech acts performed by the use of this sentence.

By contrast, Searle has to maintain that whenever the utterance performed is a literal and a serious one, knowing the linguistic meaning suffices to recover the speech act this utterance corresponds to (for the same reading of Searle, see Recanati 2003). Linguistic conventions determine the literal performance of assertions, that is, they determine the commitment associated with these assertions. Limiting the discussion to assertion, this means that linguistic conventions tell us that whenever S uses the expression E literally and seriously, she commits herself to the content p of the assertion her utterance E constitutes.

But, in cases where background conditions are violated, like Searle’s ‘The cat on the mat’ example, discussed in Sect. 2, it is impossible to tell what is the proposition the truth of which S commits herself to by her utterance. If we do not know what commitment her utterance imposes on S, we do not know what assertion her utterance corresponds to. Since the linguistic conventions determine a single literal assertion type as the literal meaning of each token, it follows that, in cases where the background fails to guarantee an adequate interpretation, speakers suddenly loose their knowledge of the linguistic, conventional meaning of the sentence.

This is clearly an absurd consequence. In theories like Cappelen and Lepore’s, ordinary speakers are denied epistemological access to (most) semantic contents—qua invariant sentence meanings. Therefore, cases of rupture with the background have no bearing, in their view, on semantic meanings; they illustrate merely the fact that sometimes we do not understand which speech act has been performed. For Searle, by contrast, recovering the literal assertion performed by S involves nothing more than the knowledge of linguistic conventions. Such knowledge should allow the hearers to access the literal meaning, that is, the assertion performed, even though the background is too poor. It follows that whenever it is impossible to tell what S means by saying ‘The cat is on the mat’, linguistic rules—which we, as competent speakers of English, know—tell us that S commits herself to the truth of the propositional content any literal and appropriate assertion of ‘The cat is on the mat’ would commit her to. Yet, in the case in hand, we are unable to indentify this content, whereas in the case of appropriate and literal use, we do know what S asserts by saying ‘The cat is on the mat’.

6 Conclusion

Defining the essential condition of assertion as committing oneself to the truth of the propositional content is feasible only if this content can be delimited by appealing to some external criterion. The reason for this is that while asserting certain propositions we, eo ipso, commit ourselves to the truth of many other propositions. Searle’s strategy is to adopt a literalist point of view, and to affirm that the propositional content of any assertion is determined by the linguistic meaning of the uttered sentence. Like any literalist theory of meaning, this one has to explain the fact—emphasised by Searle himself—that the contents of our utterances seem to depend on the context of utterance. The usual strategy is to distinguish the semantic content, determined by the linguistic meaning, from the asserted proposition, determined by pragmatic factors (Cappelen and Lepore 2005; Soames 2002). But this strategy is out of Searle’s reach, for he maintains that the content of our assertions is determined by linguistic meaning. Therefore, a very reasonable conclusion for this paper is that a theory of assertion that relies on the commitment to the truth of the propositional content should not tie this content to linguistic conventions.

To be sure, nothing prevents us—at least a priori—to combine a definition of assertion in terms of truth-commitment with some other theory of asserted content. For instance, one could argue, in a Gricean vein, that the asserted proposition is the proposition S intends A to believe or the proposition S intends to make manifest to A. However, there is a great risk that such a method to determine asserted content will suffice to define assertion itself. For instance, if the asserted content is to be defined with respect to S’s intention to make A believe a proposition or to make a proposition manifest to A, performing an assertion could well boil down to making manifest one’s intention to make A believe a proposition or to making manifest one’s intention to make a proposition manifest to A. And this would render the mention of truth-commitment superfluous. I leave to further research the question whether or not this pessimism about theories of assertion in terms of truth-commitment is justified.


Usual qualifications must be made for a restricted set of indexicals, whose linguistic meaning requires a contextual parameter for the assignation of semantic content.


I owe this example to an anonymous referee.


This problem is related to the old question of Quine’s ‘eternal’ sentences; for a discussion, see, for instance, Recanati (1994).


Searle is quite inconsistent on this topic. In Searle (1983, chapter 5), he distinguishes the background of implicit know-hows from the network of intentional states such as beliefs. In Searle (1992, Chap. 8) he includes the network of intentional states within the background. Yet, Searle (2001) gets back to affirming that the background is not propositional.


Restricting the discussion to declarative sentences and to assertions.


Searle and Vanderveken (1985) propose a slightly different description of the structure of speech acts. Every illocutionary act is defined (a) by its illocutionary point; (b) the mode of achievement of the illocutionary point; (c) the degree of strength of the illocutionary point; (d) propositional content conditions; (e) preparatory conditions; (f) sincerity conditions; (g) the degree of strength of the sincerity conditions. The illocutionary point is equivalent to Searle’s essential condition—it is “the purpose which is internal to [the utterance’s] being an act of that type” (Searle and Vanderveken 1985, p. 37). The illocutionary point of an assertion is to represent an actual state of the world. Since assertions have no specific mode of achievement, no specific conditions on propositional content conditions and that the degrees of strength of their illocutionary point and of their sincerity conditions are neuter (Searle and Vanderveken 1985, p. 183), we can safely use Searle’s (1969) more informal presentation.


A reviewer pointed out that Searle does not “accept abstract propositions that are expressed by some sentences, or some utterances […]. He thinks that propositional acts are abstract components of some illocutionary acts, or, perhaps, are some illocutionary acts abstractly considered.” This is correct; actually, it reinforces the point to be made below, i.e. that the Principle of Expressibility, under its strongest interpretation, applies to speech acts.


Or, in case of ellipsis, onto a constituent of the non-elliptical cognate of the uttered sentence. Note that if, as claimed by Stainton (1998, 2005), there are genuinely sub-sentential assertions—irreducible to syntactic ellipsis—Alston’s solution would face an important obstacle.


While Searle (1983) maintains that representing intentional states amounts to performing illocutionary acts, he also emphasises that the intention to represent an intentional state is conceptually distinct from communicating that representation.



I’m extremely grateful to Marc Dominicy for his detailed remarks on a previous draft. Two anonymous referees for this journal provided insightful criticisms that helped me to improve this paper considerably. My research is supported by a post-doctoral researcher grant from the Fonds de la Recherche Scientifique de la Communauté Française de Belgique (F.R.S.-FNRS). The results presented here are also part of the research carried out within the scope of the ARC project 06/11-342 Culturally modified organisms: “What it means to be human” in the age of culture, funded by the Ministère de la Communauté française—Direction générale de l’Enseignement non obligatoire et de la Recherche scientifique.

Copyright information

© Springer Science+Business Media B.V. 2010