2.2.1 Deliberate deference to the linguistic community

Suppose that an interdisciplinary wild bunch are working frantically on a taxonomy of linguistic deference. For several hours now they have been discussing similarities and differences between certain examples of default deference and borderline cases of deliberate deference. All the participants, A, B, C and D, agree on a common characterization for these terms and are now trying to tie up the remaining loose ends. The debate seems never-ending. At a critical moment, realising that lunchtime is almost over, the most obstinate, A, tells the others:

(7) All right, let’s say that, in deference to you, I’ll accept your argument.

We assume that the source language of their discussions is a local dialect that conforms to the definitions on which they had previously agreed. But, if A’s utterance is understood by B, C and D, they will not think that A is deferring linguistically or epistemically to any of them, but rather that she is accepting their argument out of respect for them. However, respect is not what the word “deference” would mean in the source language of this context: it is a meaning it has in a different language, namely Standard English.

In (7), the speaker again exploits contextual features in order to make it manifest that she means to shift out of the source language (the deferentialists’ dialect) and into a target language that is Standard English. The co-text plays a central role: in its source-language technical sense, the noun “deference” does not collocate with “in ____ to you”. This alone should be enough to induce recognition of the shift. All in all, this example shows that deliberate deference does not necessarily rest on language-shifts to an idiolect or a sociolect. One can defer deliberately to the linguistic community.

2.2.1.1 Deference, polysemy and Humpty-Dumpty

This, at any rate, is the conclusion if our analysis is the right one. Yet, we are aware of another possible account for (7): it could be said that, in uttering (7), A simply exploits the polysemy of the word “deference”. In other words, where our analysis posits a language-shift from a technical dialect into Standard English, others might see no shift at all. Their argument, then, would be that the technical dialect of the interdisciplinary team is nothing but an extension of the standard language. In this extension, the ordinary senses of “deference” (respect and compliance with another’s judgment) are inhibited, while a technical sense is highly activated. On this view, all A does in uttering (7) is reactivate an ordinary sense of “deference”.

We have some sympathy for this analysis. However, we think that its implications are not so straightforward as they look. First, notice that if polysemy is involved in (7), then it is polysemy of a special kind, for the sense that “deference” has in the source language (the technical local dialect) is not (yet) one that is recorded in the lexicon of the target language (Standard English). The problem here is that the deferentialists’ work results in “deference” acquiring a new meaning. Neologisms and meaning-creations always originate in the margins of the common language. Sometimes they catch on, sometimes they don’t. But, if they do, it is always because some aspects of the language spoken by a small group become incorporated into the common core. Until that happens, those aspects cannot be said to belong to the common language. Actually, as some lexicographers have shown (e.g. Rey-Debove 1978, 283-286), new words and new lexical meanings, when they occur in utterances of the common language, are often set off by quote marks or special prosody, indicating that they still feel like words in another language. Our analysis in terms of deliberate deference provides an explanation for the diachronic process by which lexical creations may become part of the common core. In the case of new meanings, this will lead to increased polysemy, but only after the process of extension of the common language has been completed.

Our analysis is less susceptible to accusations of Humpty-Dumptyism than an account strictly in terms of polysemy. On our view of deliberate deference, a speaker does not decree that this or that expression is to be ascribed a new meaning. Rather, she uses expressions which have already acquired a meaning in a given language (be that a common language, sociolect or idiolect). The only decision the speaker makes is to exploit contextual features in order to induce the appropriate language-shift. That is not Humpty-Dumptyism. By contrast, those who reject the deferential account and argue that examples like (7) exhibit plain polysemy can be suspected of Humpty-Dumptyism. On their view, a single language underlies the interpretation of (7), namely an extension of the common language. This extension includes a new meaning of an already existing term. It seems then that, merely as a result of their theoretical debates, the deferentialists have succeeded in creating a new meaning for “deference”. This means that they have acted pretty much like Humpty-Dumpty in the Alice story.

2.2.1.2 Two more examples

We have shown how our account could accommodate the intuition that polysemy is somehow involved in (7). However, we believe that there are examples similar to (7) for which a polysemy-based account is not even a likely contender. We present two such cases below. The first illustrates deliberate deference to the linguistic community, while the second shows that a speaker can deliberately defer to another common language.

Imagine a guru who, though using the spelling, the grammar and large chunks of the English lexicon, nevertheless chooses to redefine a whole class of key terms (say, “life”, “love”, “devotion”, etc.) in such a way that the ordinary senses of these terms no longer have currency in the language of the guru’s community. One can hardly say here that the guru’s language is a mere extension of Standard English. Now imagine that the guru is preaching to his flock and that his sermon is broadcast on his own satellite TV channel. For a while, he talks directly to his live audience. At one point, however, he looks straight at the camera and, addressing ‘the rest of the world’, says things like:

(8) You may experience ‘love’ and ‘devotion’ in your hearts, but these are just debased forms of true love and true devotion.

Our suggestion is that, at least for the interpretation of “ ‘love’ ” and “ ‘devotion’ ”, the guru shifts into Standard English. This, we indicate by means of scare quotes, to reflect the fact that the language-shift engineered by the guru is a deliberate one.

Examples (7) and (8) belong with a class of utterances which display an intrasentential shift into another common language, as in:

(9) Barthes described the book as “un choc historique” and “un repère nouveau et un départ pour l’écriture”. (Times Literary Supplement, 03/05/02 : 9)

The shift here may be for the sake of accuracy in quoting, or for local color, or meant as a display of one’s linguistic skills. However that may be, this is a deliberate language-shift into a common language: French. Although we acknowledge that (9) is different from the previous two examples, it provides further evidence suggesting that deliberate deference is not systematically to idiolects and sociolects.¹⁰

2.2.2 Deliberate deference without language-shift?

We now consider a class of utterances that seem to fall under the same category as the previous ones. Yet, they turn out to lack one important property exhibited by the various examples of deliberate deference studied so far.

Let us assume that Kate, who has no training in law, is attending a trial. Both the judge and the defendant’s counsel use terms of art with which she is not familiar. For instance, it is not obvious to her whether the defendant committed a felony, an offence or a misdemeanor.

During a break, while talking about the proceedings with other members of the audience, Kate is trying to determine the sort of crime that the defendant is guilty of. In so doing she says things like:

(10) I don’t think what he did was a felony. I’d say it was a misdemeanor.

Since she realises that her understanding of these terms is at best sketchy, she often supplements her utterances with a metalinguistic comment, or articulates them with a special intonation pattern, of the sort that can be rendered by means of scare quotes:

(10’) I don’t think what he did was a felony, as the judge put it. I’d say it was a misdemeanor, if I understand the lawyer’s distinction.

(10’’) I don’t think what he did was a ‘felony’. I’d say it was a ‘misdemeanor'.

These comments and extra markers indicate that we are not dealing with instances of default deference. So, are we dealing with deliberate deference, and, if so, to whom? In an externalist framework such as ours, it is generally accepted that the meaning of legal terms is fixed by members of the legal profession for the whole of the linguistic community. There should therefore be no difference between the meaning that the judge ascribes to “felony” and “misdemeanor” and the meaning that these terms have in the lexicon of English. And if there were a difference, Kate, as a non-expert member of the audience, would probably choose to trust the norms of the linguistic community (as fixed by the body of experts alluded to above). This suggests that, when Kate utters (10), (10’) or (10’’), she is not (just) deferring to this judge or lawyer, or even to the legal profession, but to the norms of the linguistic community.

It is tempting to conclude that examples (10)-(10’’) are a further illustration of deliberate deference to the linguistic community. But, as hinted above, these examples lack one significant feature exhibited by the other cases: they involve no language-shift. In (10)-(10’’), the language with respect to which terms such as “felony” and “misdemeanor” are interpreted is none other than the source language set by default. This entails that, unlike what can be observed in (6) and (7), Kate’s deferring turns out to have no impact on the truth conditions of her utterances. Still, there is a major difference between (10)-(10’’) and genuine cases of default deference. Kate resorts to metalinguistic comments or special intonation patterns in order to make the language parameter of the context salient. This does not happen in cases of default deference, where the speaker typically has no communicative intention to bring the language of interpretation into the foreground.

Faced with these facts, we believe that the right theoretical choice consists in maintaining that (10)-(10’’) involve deliberate deference. Accordingly, we must relax criterion (iii) of paradigmatic instances like (6), so as not to require the presence of a language-shift in the strict sense.¹¹ We therefore propose the following definition:

S performs an act of deliberate linguistic deference if and only if:

(a) S produces an utterance u;

(b) S exploits certain contextual features in order to make salient the linguistic parameter L for the interpretation of u or some segment of u;

(c) S wants her exploitation of contextual resources to be recognized as part of her communicative intentions by the audience.

Although our definition does not include any requirement for a language-shift, cases of deliberate deference can still be represented by means of the deferential operator. For instance, what happens in (10) can be captured by the following formula:

(10₁) I don’t think what he did was a R_{StandardEnglish} (felony). I’d say it was a R_{StandardEnglish} (misdemeanor).

The deferential operator indicates that the expressions “felony” and “misdemeanor” are to be interpreted with respect to Standard English. In cases of language-shifts, the only difference is that the value of “x” in “R_x (s)” is distinct from the source language. Deliberate deference with a language-shift is nothing more than an important sub-category of deliberate deference.

3.      Non-linguistic deference and other related phenomena

In this section, we study the notions of epistemic deference and imperfect mastery. Epistemic deference should be carefully distinguished from linguistic deference, and our first comments will be focused on justifying this distinction. We then show that the notions of epistemic deference and epistemic evidence, though related, must be kept well apart. Finally, we discuss the phenomenon of imperfect mastery and its relationship to linguistic and epistemic deference.¹²

3.1 Epistemic deference

Deference is an issue of interest not only to linguists and philosophers of language, but also to epistemologists and philosophers of science. It is generally admitted that a lot of the knowledge that we possess is acquired deferentially, by testimony. But deference does not only affect the things we know: it also affects our beliefs, beliefs we are none the less ready to act upon. We receive information from many different sources, and we make choices as to which information to accept and which to reject. Imagine a lady with a rare disease who wants to gather different opinions about her illness before undertaking a medical treatment. Every doctor she meets gives her an opinion, based on evidence and other considerations. If the diagnoses differ, she will have to decide which doctor to trust above all others. But, underlying our beliefs and actions are not just other people's judgments on issues for which there is a fact of the matter. We also defer to others on issues that are largely a matter of personal opinion. Suppose that Takeshi has been told by one friend that Sakura is the best sushi-bar in town, and by another that Mikado is the best. If he wants to take his fiancé(e) for a date, Takeshi will have to decide which friend to trust, whose judgments of taste are more reliable.

In what follows, we will focus on those instances of epistemic deference that underlie assertions, because it is in these cases that epistemic deference may be most easily confused with linguistic deference. We will say that a person who makes an assertion is deferring epistemically when she bases her claim, partly or completely, on someone else’s opinion. Typically, a person who asserts that she has arthritis is epistemically deferring to the doctor on whose diagnosis she relies. We argue below that this phenomenon is distinct from linguistic deference. Furthermore, it cannot be reduced to the notion of epistemic evidence, even if a certain correlation exists.

3.1.1 Epistemic and linguistic deference

Whereas linguistic deference is involved in fixing the meaning of a term, epistemic deference occurs when a person defers to someone else concerning a particular judgment. Whether a speaker is deferring epistemically or not is independent of whether she is deferring deliberately or by default for the use of the words occurring in her utterance.

To begin with, it is easy to realize that default linguistic deference must be independent of epistemic deference. As we have argued at length, default linguistic deference is a ubiquitous phenomenon. Epistemic deference, on the other hand, occurs when we rest a claim upon other people’s opinions. It is not surprising, then, that default linguistic deference can, but need not, co-occur with epistemic deference. For example, suppose that Tim goes to see a doctor who, having examined him, tells him: “You have myositis.” Tim does not know what myositis is. He only understands that it is related to the pain he is feeling in his calves. Back at home, he tells his mother:

(11) I have myositis. It is nothing serious. I should just rest for a while.

In saying (11), Tim is deferring by default to the linguistic community concerning the meaning of the term “myositis”, and, at the same time, he is deferring epistemically to the doctor, the truth of whose diagnosis he takes for granted. But when Tim tells his mother: “My calves hurt badly”, he is certainly not deferring epistemically to the doctor, for he is best placed to judge whether a part of his own body hurts or not. However, Tim will still be deferring by default to the linguistic community concerning the meanings of the words that he is using to report the pain in his calves, such as “calves”, “hurt”, etc.

Somewhat more interesting are the connections between epistemic deference and deliberate linguistic deference. Let us approach these through various examples. We have already seen that epistemic deference occurs independently of default linguistic deference. When Tim sincerely asserts that he has myositis, he is deferring epistemically to the doctor, but from a semantic point of view, he is deferring by default to the entire linguistic community. This is even more obvious in his assertion that “it is nothing serious”, which is again epistemically based on the doctor’s judgment, but involves only terms that Tim, a native English speaker, fully masters.

Conversely, deliberate linguistic deference occurs independently of epistemic deference. This is clear from our analysis of example (5) above:

(5) Jan, I have this bad case of arthritis. Would you close the curtains and hand me some aspirin?

Though Tineke is borrowing the bogus doctor’s deviant definition of “arthritis”, she is not deferring to any medical diagnosis made by that doctor, or even to any opinion that he might have regarding her condition.

The mutual independence of linguistic and epistemic deference can be given a more complex, and more subtle, illustration. Think again of the doctor who intentionally uses the word “arthritis” with the deviant meaning that his patient attributes to it. Suppose that this doctor asked for a specialist’s opinion regarding the symptoms in his calves. Diagnosed with inflammation, which is precisely the condition for which the misguided patient uses the word “arthritis”, he tells the colleague with whom he has that patient in common: “My calves hurt. It is arthritis.” Although he does not defer epistemically for the claim that his calves hurt, the doctor defers epistemically to the specialist for the claim that his condition is “arthritis,” i.e. inflammation of his calf muscles. At the same time, he is deferring deliberately to their patient’s idiolect, for the semantic question of what counts as “arthritis.” In this case, deliberate linguistic deference occurs together with epistemic deference, but with distinct deferees.

In sum, linguistic deference and epistemic deference are distinct and mutually independent phenomena, though they can combine in various ways, as has been amply illustrated in previous sections.

3.1.2 Epistemic deference and epistemic evidence

The examples given above might suggest that epistemic deference occurs as a direct result of there being insufficient epistemic evidence for making a claim. And it is true that epistemic deference is quite often a matter of the amount of epistemic evidence that one has for making a certain statement. Thus, if you have no independent evidence to assert p, but have been told by someone you trust that p, you are likely to assert p, simply because you rely on that person’s judgment. We say in such a case that you are epistemically deferring to that person. On the other hand, when you have the best possible epistemic grounds of your own for asserting p, then in asserting p, you will probably not want to rely on someone else’s judgment.

However, lack or poverty of epistemic grounds are neither a sufficient nor a necessary condition for epistemic deference. Someone who has all the evidence that can be had may still choose to defer epistemically to someone else. Thus consider a doctor who happens to be the greatest expert on arthritis, but lacks self-confidence. It is plausible to say that, when he tells a lady patient “You have arthritis,” he is epistemically deferring to his colleagues on the issue of whether that woman’s condition is indeed arthritis, even though he has enough of his own evidence for this claim. To indicate that he is doing so, he might say “We believe that what you have is arthritis.” Or, imagine that Naïma is a first-rate scientist whose research shows how to achieve cold fusion, but is very shy and insecure. She is doing tests in her lab when a senior researcher, whom she deeply respects and admires, tells her: “You are wasting your time. Believe me, cold fusion is something impossible!” Out of sheer insecurity, she decides to defer to his opinion, even though it directly contradicts a claim that she has excellent evidence for, namely, that cold fusion is possible.

Conversely, there are situations in which people may form and express a firm judgment even on an issue for which they have no good epistemic grounds. Consider a woman whose partner tells her “You have arthritis” just out of some inner conviction. He is not, then, deferring epistemically to anyone. People do make claims for which they have no good evidence, and which do not reflect other people’s opinions. Such claims –people’s best guesses, as we might put it– exemplify the case where one lacks epistemic evidence, and yet abstains from deferring epistemically.

3.2 Imperfect mastery

Many philosophers hold that there are concepts, and that concepts are very much like mental files in which information gets stored. Consider the concept that Carmelia has of a certain particular, say François Recanati. Her concept contains three main types of information: perceptual information, e.g. that the particular concerned by this concept is that guy, whom she sees talking right there in front of her, descriptive information, e.g. that he is the author of Literal Meaning, and metalinguistic information, e.g. that he is called “François Recanati”. Our concepts of universals, too, mostly combine those three types of information. But in many cases, the concepts that we associate with words that we use, like “arthritis”, “elm” or “hydrogen”, are fairly poor, and the information they contain does not enable us to decide on any given occasion whether the word correctly applies to something we are presented with, or to draw certain inferences that someone more knowledgeable could draw. If the concept that a person associates with a term is poor or, at any rate, not as rich as the concept that experts associate with it, we talk of imperfect mastery. Note that mastery is very much a matter of degree, and that it is not obvious that anyone ever achieves perfect mastery. But to bring the issue home, one might want to know how the phenomenon of imperfect mastery relates to linguistic and to epistemic deference, and ask questions like the following. What information must be present in a concept for one to be able to defer, deliberately or by default, using the associated term? Conversely, could the presence of some information make deferring impossible? Does epistemic deference arise whenever we make assertions using concepts that we do not perfectly master? And will the wealth of information in our concepts prevent us from deferring epistemically?

By way of giving a single answer to these questions, we hold that imperfect mastery is a phenomenon that must be kept separate from linguistic deference and from epistemic deference. In other words, whatever a person’s mastery of the concept associated with some term, whatever the amount and quality of the information contained in the mental file, the following options all remain viable: the person will defer by default when using the term, or she will defer deliberately to some contextually salient dialect, be it or not the dialect from which she picked up the term. Likewise, the person may or may not defer epistemically for assertions that she makes using the term.

3.2.1 Imperfect mastery and linguistic deference

Consider a medical expert whose concept of “arthritis” is as rich and determinate as can be. Does such a person defer linguistically when she uses the term “arthritis”? The intuition is that the meaning of such terms is determined precisely by such experts. So if we say that the expert defers in turn, who could she possibly defer to?

Though there may be a grain of truth in this intuition, our account of default deference does not require the speaker to have the intention to defer, or to know which source language is contextually selected. This means that even our medical expert defers by default to the linguistic community when using “arthritis”. It just happens that she is among the experts who ultimately determine the meaning of the term.

We have just shown that “perfect” mastery is compatible with deference by default. Is it also compatible with deliberate deference? Again, the answer is “Yes”. To see this, just recall our example of the doctor who suffered from an inflamed calf muscle. This doctor could be assumed to know as much about arthritis as possible. Yet, this did not prevent him to wittily exploit the ignorance of a patient and tell his colleague “It must be arthritis.”

In a similar way, imperfect mastery allows both for default and deliberate linguistic deference. A woman who knows virtually nothing about arthritis, except that there is something called “arthritis”, can use this word to say true or false things, whether or not she has any intention to defer linguistically at all. Or she may defer deliberately, indicating the source from which she got the word and making it clear that she intends to apply the word to whatever it is that her source applies it to, even though she might have no idea what that is.

3.2.2 Imperfect mastery and epistemic deference

Our level of mastery of a given concept can neither force us to defer epistemically nor prevent us from doing so. Someone who has perfect mastery can still choose to defer epistemically, like the shy scientist who endorses her senior colleague’s opinion that cold fusion is impossible. Most often, though, people with excellent mastery of a concept make assertions without deferring to other agents, provided that they have strong enough epistemic grounds for their assertions. It is true, too, that, if we know hardly anything about myositis, we are unlikely to go around making unwarranted claims about it. Thus, if we report that Tim has myositis, we will typically do so because someone whose judgment we trust told us that Tim had myositis, or because we read it in Tim’s medical file. In those cases, we defer epistemically. But others with the same level of mastery may make the very same claims without deferring epistemically, e.g. out of some inner conviction, however odd this may seem. In any case, even cautious speakers aware of their poor mastery of a given concept will be ready to make certain assertions about myositis without deferring epistemically. For example, they will confidently assert that myositis is a condition called “myositis”, or that they would not like to be diagnosed with myositis, even though they have no idea what that is.

In sum, even though there is probably a correlation between an agent’s imperfect mastery of a certain concept and her being inclined to defer epistemically, epistemic deference and imperfect mastery are distinct phenomena, irreducible to one another.

4.      Conclusion

In this paper, we have argued for the mutual independence of three related phenomena, namely linguistic deference, epistemic deference and imperfect mastery. One of our initial questions has been what kind of framework could accommodate instances of falsity-due-to-misunderstanding and cases in which a speaker overtly chooses to use an expression the way someone else uses it. Our answer has been that, in both types of cases, an expression or segment of discourse is used deferentially. This is what grounds the category of linguistic deference. Within this category, we have distinguished two varieties, default deference and deliberate deference, which, together, cover a significant proportion of the examples addressed in the literature.

Default deference has been shown not to be restricted to those cases where the speaker defers to the linguistic community as a whole. We have supplied examples of default deference to a sociolect and even to a local dialect. As for deliberate deference, which is usually understood as deference to an idiolect or local dialect, we have given evidence that it ain’t necessarily so. Our examples suggest that speakers sometimes defer deliberately to the norms of the common language. The account we provide differs in one further respect from the picture that emerges from what little literature has been written on the subject. Deliberate deference does not always involve a genuine language-shift.

Concerning the related notions of epistemic deference and imperfect mastery, we have contended that they are distinct from each other and orthogonal to linguistic deference. Someone defers epistemically when they base a claim on someone else’s judgment, but this does not entail that they are deferring for the meaning of the words they are using. With respect to imperfect mastery, we have shown that the partial understanding of a concept does not constrain a speaker to defer either linguistically or epistemically.

Providing a taxonomy of the various cases of deference discussed in the literature is like drawing a map of the tip of an iceberg. However accurate the map, it is insufficient. Just as safe navigation requires awareness of what lies under the water surface, any stable theory of deference requires awareness of the philosophical and linguistic issues of titanic proportions that underlie it. In the case at hand, the submerged part of the iceberg comprises issues such as quotation, simulation, echoic uses, irony, polysemy, knowledge acquisition, justification, cognitive architecture and concepts.

References

Burge, T. (1979). Individualism and the Mental. Midwest Studies in Philosophy. Volume IV. Studies in Metaphysics. P. French, T. E. Uehling and H. K. Wettstein. Minneapolis, University of Minnesota Press

Loar, B. (1990). Personal References. Information, Semantics and Epistemology. E. Villanueva. Oxford, Basil Blackwell: 117-133

Recanati, F. (2000). Oratio Obliqua, Oratio Recta : an essay on metarepresentation. Cambridge, Mass., MIT Press

Recanati, F. (2002). "Unarticulated Constituents." Linguistics and Philosophy 25: 299-345

Rey-Debove, J. (1978). Le Métalangage: étude linguistique du discours sur le langage. Paris, Le Robert

Woodfield, A. (2000). "Reference and Deference." Mind & Language 15(4): 433-451

Notes

10 There are plenty of examples like (9), and they are usually taken to be related to quotation. (See issue 17 of The Belgian Journal of Linguistics for various discussions.)

11 Another option would be to leave the criterion for deliberate deference unaltered. As a result, examples like (10)-(10’’) would come under a third category of linguistic deference, intermediate between default and deliberate. In our view, however, these examples have much more in common with deliberate deference; hence we shall not pursue this line of reasoning further.

12 Let it be clear from the outset that our goal in this section is not so much to make a new contribution to the existing literature on epistemic and cognitive issues related to deference, as to clearly distinguish those issues from the ones that arise in relation to linguistic deference.