The best strategy I can think of to avoid this radical conclusion draws upon a distinction made by Emma Borg. In her 2004 book Minimal Semantics, Borg distinguishes between minimal semantic understanding, i.e., knowledge of what she calls “liberal” truth conditions, on the one hand, and knowing how to “verify” (or knowing what would make it the case) that the truth condition is met, on the other hand (Borg 2004, p. 238). Thus, the members of the social group who only know (what we would call) completely red-skinned and completely green-skinned apples are unable to know whether the truth condition of the sentence “this [the speaker refers to an apple] is red” is met regarding a not fully ripe apple, but they nevertheless know the truth condition—namely that the object the speaker wants them to attend to be red—whatever the latter may require in the case at hand. They have full semantic knowledge but lack background know-how. However, the latter is only required for “verification,” or
knowledge of the proposition p stated (i.e., knowledge that p),
but not for the less demanding
knowledge of which proposition was stated (i.e., knowledge that p is the proposition literally expressed by the speaker).
The latter is sufficient for semantic knowledge regarding the statement.
I like this answer to the contextualist challenge, but I think that it eventually leads to a more moderate version of contextualism, rather than to a full-scale rejection: it leads to a version that makes room for semantic knowledge without background assumptions or know-how, knowledge whose content can indeed be investigated by formal semantics.
Clearly, the advocate of the present answer needs to explain how one can understand a sentence while lacking the kind of background know-how regarding which Searle would claim that in the absence of such capacities the “notion of the meaning of the sentence” has no clear “application” at all (see quotation above). Searle would stress that in the absence of appropriate background assumptions or know-how we have no clear idea of how to understand a sentence like “This (apple) is red;” which manifests itself in the fact that we do not, for instance, know how to follow the corresponding order “Bring me the red apple!” (cf. Searle 1983, p. 147). In the light of Borg’s distinction, this can be described as lack of knowledge about “verification,” but what about the strong intuition that in the absence of such background know-how the sentence fails to express a content that can be evaluated in terms of truth and falsity? To strengthen this intuition, consider Searle’s examples S2–S4 (cf. Searle 1983, Ch. 6):
(S2) Bill opened the mountain.
(S3) Bill opened the grass.
(S4) Bill opened the sun.
These sentences are syntactically well-formed and contain meaningful English expressions; yet they do not express clear semantic content—unless we imagine some background know-how regarding what it means to open a mountain, the grass, or the sun.[10] The mere combination of the literal meaning of the verb “opened” with the literal meanings of other English expressions in accordance with the English syntax does not seem to be enough to produce a clear truth-evaluable content, despite the fact that “to open” does not look like an indexical that yields as reference a unique behavioural relation (or type of action) referred to as “opening,” for a neatly defined type of context—in the way that “I” always yields the speaker of the utterance context as its referent. Borg would disagree; she says about an analoguous example by Searle (“John cut the sun”):
If the competent language user understands all the parts of the sentence (she knows the property denoted by the term ‘cut’, she grasps the meaning of the referring term ‘John’ and she understands the meaning of the definite description ‘the sun’) and she understands this construction of parts, then she knows that the utterance of this sentence is true just in case […] John stands in the cutting relation to the sun. Now clearly any world which satisfies this condition is going to be pretty unusual (and there may be some vague cases [...]) but there will be, it seems, some pretty clear cases on either side of the divide. For instance, any world where John’s actions do not have any effect on the physical status of the sun is clearly going to be a world where the truth-condition is not satisfied. While any world where John’s actions do result in some kind of severing of the physical unity of the mass of the sun is a world where the truth-condition is satisfied. (Borg 2004, p. 236)
This reply to Searle is unconvincing for at least two reasons.
First: To begin with, Borg here equates semantic knowledge concerning the verb phrase “cut” with knowledge of the property it denotes (see the first brackets in the quotation). But arguably this phrase does not denote any property in isolation; it only does so in the context of a sentence (by the “context principle”).[11] And Searle’s parallel point about “opened” is that this verb phrase denotes quite different properties in S2–S4, respectively, without being ambiguous. That the verb is unambiguous in these cases becomes intuitively plausible if we apply the “conjunction reduction” test (cf. Searle 1992, pp. 178-179). Instead of asserting the conjunction of S2–S4 we can just as well say: “Bill opened the mountain, the grass, and the sun” and perhaps add: “he used a secret universal device for the task recently developed by NASA.” This may be a weird example, but its weirdness does not seem to be due to the ambiguity of “opened.” Rather, unlike the imagined NASA devisors we simply have no background know-how that would enable us to assign truth conditions to this sentence.
Borg would probably reject the context principle (thus paying a high price for her view) and answer that there may be vague cases in which we do not know whether the opening relation obtains or not, but that “there will be […] some pretty clear cases on either side of the divide” (Borg 2004, p. 236); after all, in the preceding quotation she makes a parallel claim about the example “John cut the sun.” But this answer is, again, unconvincing (as is Borg’s parallel claim). One might just as well argue that both S5 and S6 describe the same relation, the opening relation, as obtaining between different objects.
(S5) Bill opened his hand.
(S6) Bill opened the door.
But opening a hand is an intentional bodily movement, while opening a door is a more advanced or complex action that merely involves such bodily movements. These are different kinds of behavioural relation. Of course, clear examples of the obtaining of both of these relations may have something in common, but this common feature does not seem to constitute a common type of action. And what (if anything) is the verb phrase in S5 and S6 supposed to denote, if not a type of action?
Nor is the verb phrase in this pair of sentences ambiguous. This is made plausible by the conjunction reduction test: it is perfectly fine to abbreviate the conjunction of S5 and S6 as follows: “Bill opened the hand and the door.”
The (to my mind) false impression that the unambiguous verb phrase in S2–S4 denotes the same behavioural relation or feature as in, say, S6, merely comes from the fact that we tend to think of established uses of “a opened b” sentences (or “a cut b” sentences) when trying to construct an interpretation for cases like S2–S4 that we do not really understand. But there is no such use in these cases (see the next paragraph but one).
Second: Moreover, Borg’s claim that “any world where John’s actions do result in some kind of severing of the physical unity of the mass of the sun is a world where the truth-condition [of ‘John cut the sun’] is satisfied” is simply false. If John causes an explosion whose effect is that the physical unity of the mass of the sun is severed (such that it breaks into, say, two halves),[12] he does not thereby cut the sun. I suppose that any attempt to secure a minimal truth condition for S4 (and S2–S3, for that matter) is doomed to failure. In order to have at least a slight chance of getting off the ground, any such attempt will have to mention something that can be done using sharp-edged tools (or devices simulating such tools),[13] and it seems impossible to define (let alone imagine) a procedure of this type that could in principle be applied to the sun.
To anticipate the alternative approach I am going to take, in cases like S2–S4 there is no established sentence-use because there is no appropriate background know-how to be found in the relevant social group (including its late members), hence no group of (current or former) “producers” (see below), and hence no relation conventionally denoted by the verb phrase that could enter the respective truth condition. Therefore, these sentences have “literal meaning” (as Searle puts it) but lack semantic content. Literal meaning is not usage (in the current sense), nor does it require a particular usage—unlike respective meaning.
On similar grounds (to return to the last example about S1), if in the envisaged social group there is no background know-how regarding certain apples that we would readily classify as “red,” against that background, the sentence S1 has no clear application to such apples in the language use of that group, and it does not have the same truth condition as in ours. An interpretation problem occurs. I am attracted by an interpretation-theoretical principle proposed by Timothy Williamson, probably inspired by Gareth Evans, which Williamson calls the principle of knowledge maximization (as opposed to the principle of truth maximization to be found in traditional hermeneutics, and endorsed by Donald Davidson):
The shift from conventions of truthfulness to conventions of knowledgeableness also has repercussions in the methodology of interpretation. The appropriate principle of charity will give high marks to interpretations on which speakers tend to assert what they know, rather than to those on which they tend to assert what is true […]. (Williamson 2000, p. 267)
The right charitable injunction for an assignment of reference is to maximize knowledge, not to minimize ignorance. (Williamson 2007, p. 265)
According to the principle of knowledge maximization, an interpretation is correct to the extent that it maximizes “the number of knowledgeable judgements, both verbalized and unverbalized, the speaker comes out at making” (McGlynn 2012, p. 392). To motivate this principle, although in a somewhat modified version, imagine that in the above example about the box there are in fact two objects in the box—a red ball and a yellow apple—but that we know that the speaker does not know about the ball, which was already hidden in the box before we put the apple into the box while the lightning was such as to make the apple look red.[14] The speaker, who observed how we put the apple into the box, mistakenly believes it to be red and exclaims: S1 (“this is red”). No doubt, if this utterance has any truth condition, it involves an apple rather than a ball, and the utterance is false. A suitably modified version of the principle of knowledge maximization yields this result as the correct interpretation, while the principle of truth maximization fails to do so. After all, the speaker would only give voice to a true belief here if her statement concerned the ball rather than the apple. However, this belief would not qualify as knowledge, in the described situation, and by assumption the speaker lacks any other knowledge regarding that ball. By contrast, the speaker possesses some knowledge about the apple, which is in fact yellow. In Evans’ terms, the speaker has opened a mental dossier (a dynamic system of beliefs) about the apple, which contains quite a number of (correct) information about it, even though the addition of the belief that the apple is red does not enlarge that body of knowledge. Thus, the speaker ought to be interpreted as giving voice to that false belief, Davidson and traditional hermeneutics notwithstanding. This interpretation takes into account more relevant knowledge on the part of the speaker than the other.
The principle of knowledge maximization needs to be modified in terms of, or supplemented by, a more traditional theory of justification in order to yield this result. To see this, let us first consider another example, inspired by Husserl (cf. Husserl 1987, p. 212), which I have used in earlier writings to motivate my “neo-Husserlian,” moderately externalist reconstruction of his view on respective meaning and intentional content.
Let's assume that at a time t1 Ed points at a certain table in the seminar room where he has just been lecturing and exclaims:
[(S7)] This table wobbles.
One of the students is prepared to take Ed to the caretaker, to make sure that the table gets repaired immediately. The way from the seminar room to the caretaker's office is rather complicated. But they manage to find it. The caretaker asks Ed to take him to the seminar room with the wobbling table. The student has other things to do. So Ed has to take the caretaker to that room by himself. Finally, they arrive at a seminar room that Ed falsely believes to be the room with the wobbling table. At t2 Ed points at a certain table, which he regards as that wobbling table, and once again declares [S7]. The caretaker investigates the table and contradicts Ed—who reacts somewhat irritatedly. (Beyer 2001, pp. 284-285)
It is unclear which referent (table 1 or table 2) the interpreter is supposed to assign to the demonstrative term “this” according to the (unmodified) principle of knowledge maximization. After all, both of these assignments would lead to an ascription of knowledge to the speaker: knowledge about table 1 (to the effect that it wobbles) and table 2 (to the effect that it is a table he takes to be wobbling), respectively.
To decide the issue, the interpreter needs to take a closer look at the speaker’s epistemic motivation for making the judgment given voice to in her utterance of S7 at t2—he needs to consider the experience(s) with recourse to which the speaker can justify her claim to knowledge. If the judgment is motivated by a perception the speaker is having, thus qualifying as an observational judgment, it will be about the object of that perception: that object is perceived as thus-and-so and for this reason (on this ground) judged to be thus-and-so. This is what happens at t1: the speaker perceives table 1 as wobbling and is sincerely giving voice (in the narrow sense) to an accordingly motivated judgment to the effect that it wobbles. However, at t2 the epistemic situation is different. The speaker's judgment is motivated by a memory of table 1 rather than by her current perception of table 2. It is this memory that rationalizes her judgment, and could be self-ascribed by the speaker when justifying her judgment. Therefore, the speaker gives voice, in the narrow sense, to a judgment about table 1, namely that it wobbles. I have elsewhere called this epistemically-determined truth condition the utterance's internal truth condition.[15] According to the neo-Husserlian approach, respective utterance meaning determines the internal truth condition.
So in order to yield interpretations that adequately reflect the meaning intentions actually given voice to by the speaker, and thus the respective meanings of their utterances, the principle of knowledge maximization needs to be supplemented by (or reformulated in terms of) a more traditional theory of justification, drawing upon notions like observation, memory and testimony (referring to sources of justification). Note that the present approach to reference supports a version of the context principle: it is only in the context of a judgment that a referent can be assigned to a mental act of reference given voice to by a singular term.
Let us finally return to the example about the two objects in the box. In this example the speaker gives voice to a judgment about the yellow apple rather than the red ball in her utterance of S1, because she (falsely) remembers that ball as being red, having opened, on an earlier occasion, a mental dossier about it containing the (incorrect) information that the ball is red, while she neither remembers nor perceives, nor has heard about the ball that also happens to be in the box. Thus, the judgment given voice to can only be motivated by, and justified with recourse to, that memory—even if it does not yield knowledge in the case at hand. And that memory concerns the apple rather than the ball, because it belongs to the speaker's body of information about the apple. Thus, on a version of the principle of knowledge maximization modified in accordance with the foregoing neo-Husserlian approach to reference assignment, the utterance in question concerns the apple, if anything.
Now by the principle of knowledge maximization (in both versions), if there is no background that enables members of a social group to express knowledge by a sentence like S1 in a situation where we would readily apply that sentence—on the basis of our own knowledge and background know-how—, then the following conclusion recommends itself: in the language use of this social group the sentence lacks the determinate truth-evaluable meaning it expresses in our own language use, in a given context. (Contrast what Borg says about “John cut the sun;” see the above quotation from Borg 2004, p. 236.)
This speaks in favour of contextualism. However, it does not speak in favour of a radical version of contextualism, which would not allow for a notion of minimal semantic knowledge that can indeed be possessed in the absence of personal background know-how—a version that thus ignores the above-described difference between two types of knowledge regarding truth conditions. In what follows, I shall sketch a more moderate version of contextualism that does take this difference into account.