Note: This paper is a sequel to my "Conceptual Relativism and Translation" (which appeared in Language, Mind and Epistemology, Preyer et al (eds.), Kluwer Academic Publishers, 1994).  Sections 1.1-1.2 below summarize parts of that paper which are directly relevant to my discussion in the present paper.  I am currently preparing the paper for publication and expect to expand parts of sections 2-5 at the expense of section 1.  Comments are warmly welcome.

    Certain problematic philosophical positions can, I believe, be traced back to an excessively narrow conception of linguistic understanding.  On that conception—which I describe as a 'theory'-theory of language understanding—our understanding of other people's speech is to be explicated as fundamentally a theoretical process of figuring out on the basis of evidence, tacitly perhaps, the correct interpretation of the sounds they make in speaking.<1>   I think it can be argued, for example, that this conception plays an important role in Quine's arguments for his indeterminacy thesis.<2>   I also think that getting away from the 'theoretical' model of linguistic understanding may allow us to reject Dummett's argument against truth-conditional semantics (and the global anti-realism he bases on that argument).<3>
    In this paper, I want to argue that certain mistakes committed by both proponents and opponents of so-called Conceptual Relativism can be also traced back to a commitment to the 'theory'-theory conception of linguistic understanding.  A less artificial, `non-theoretical' conception would allow us to make intuitive sense of the possibility of genuine conceptual diversity.  But acknowledging this possibility does not force us to join the contemporary trend of postmodernist relativism.  Accepting the possibility of conceptual diversity, I will argue, is consistent with rejecting relativism.

1. On the Very Idea of Conceptual Differences

    Let me begin by formulating four different claims that have been made in connection with Conceptual Relativism.

    (I) There can be substantive differences among conceptual 'frameworks' or 'schemes' or 'world views' across different cultural, social,
            historical, etc. groups.
    (II) We cannot adjudicate among divergent conceptual schemes.
    (III) The correctness of a conceptual scheme is relative to culture, society, historical period; there is no absolute, culturally- (or socially-)
            independent standard for assessing conceptual correctness.

    (IV) The worlds in which possessors of different conceptual schemes live are themselves different from one another (since "[a] world exists
            only relative to an imposition of concepts").<4>

1.1 The Denial Strategy

    The `denial strategy' can be presented in a general argument form as follows:

            P  If two individuals possess genuinely divergent conceptual schemes, then there must be failures of intertranslatability between the languages they speak.
    But   Q  There can be no failures of intertranslatability.
    So,    C  There can be no genuine divergence of conceptual schemes.<11>

P presents untranslatability as a necessary condition for genuine conceptual diversity.  Indeed, attempts to establish conceptual diversity are often supported by furnishing examples that allegedly meet this condition—actual cases where translation of another culture's discourse is impossible.<12>   We can think of the intuitive idea connecting untranslatability and conceptual diversity as follows.  Possession of a language requires possession of a set of concepts.  The words and phrases of the language serve to express those concepts.  Where we find failures of translation between two languages, we can locate expressive gaps in one or the other language, or expressive mismatches between the two languages.  To the extent that the expressive gaps signal the presence of incommunicable concepts, and to the extent that the expressive mismatches reveal incommensurability of concepts, the notion of divergent conceptual schemes—and with it the threat of conceptual inaccessibility—can get a foothold.  In this way, the limits of translatability—if there were such—would seem to mark the limits of our ability to 'see the world through the eyes' of those whose discourse we're unable to translate.  The more pervasive untranslatability between our languages is, the wider the conceptual chasm between us.<13>
    Now, exploiting this kind of link between untranslatability and genuine conceptual divergence, the anti-relativist might then deny that there can be genuine failures of intertranslatability—that is, argue for Q above.<14>   However, Q seems implausible on its face.  I think there are plenty of examples to show that failures of translation are quite common; this is a direct consequence of the syntactic, semantic, and pragmatic mismatches that exist pretty much between any two extant natural languages.  Here I can only mention some examples briefly.  The examples are actually designed to show that denying untranslatability is both implausible and unnecessary for rebutting Conceptual Relativism.

1.2 Failures of Translation<15>

    Lexical mismatches are mismatches at the level of lexicon.  Every language has lexical items that stand for idiosyncratic elements in the environment, history,  culture, or society of its speakers, and that may be completely missing from the environment, etc. of speakers of other languages.  For example, the term [kabary] names a special kind of formal speech given only on certain types of Malagasi ceremonial occasions.  Languages may also "package" differently elements that are present in the background of their speakers.  For example, Vietnamese reportedly has a single lexical item standing for [someone who leaves to go somewhere and something happens at home so that he has to go back home.]  And there are more radical examples of incompatible divisions of the color spectrum.  These kinds of mismatches are a constant source of translation difficulties.
    There are also various grammatical and pragmatic mismatches.<16>   For two simple cases, consider these.  First: in Hebrew, pronouns, verbs, adjectives, and nouns constructed out of verbs and adjectives are marked for gender.  This relatively superficial grammatical feature could get in the way of translatability.  Just imagine a context in which an English speaker says: "I have just broken up with my lover" having no intention of disclosing the lover's gender.  Second: a translator of the sentence "You are sick" into French, would have to have background information of which the ordinary English-speaking audience need not be aware, in order to decide whether to translate it as "Tu es malade" or "Vous etes malade" (viz., as to whether the speaker is sufficiently familiar with the addressee).  Where the information is not available—or where it's important that it not be conveyed—the translation is bound to be misleading, distorting, or to disclose too much information to the French audience.  (More examples will be given later.)
    Now, in these sorts of cases, translators are usually in a position to provide an adequate gloss that conveys the content of the untranslatable material.<17>   But such glosses will often involve metalinguistic descriptions of the relevant languages, or very lengthy socio-cultural explanations, or even ungrammatical sentences.  These widely available explanatory devices do not by themselves enable us to produce translations; rather, they enable us to justify particular judgments about failures of translation.  This is important, because it undercuts one popular anti-relativist strategy, which proceeds by arguing that, any time a relativist tries to convince us that a bit of discourse is untranslatable, she must defeat her claim by unwittingly providing a translation.  The reason this strategy does not work is that not every explanation of untranslatable material can count as a translation of it.  At the same time, however, where the possibility of providing such explanations exists, the effort to establish conceptual inaccessibility would indeed seem thwarted.
    The above examples are designed to support the claim that there is no direct route from untranslatability to conceptual inaccessibility (or even to conceptual difference—witness the gender marking example). But, if this is so, then we should be suspicious of any attempt to connect failures of intertranslatability too tightly with divergence between conceptual schemes.  Languages can fail to be intertranslatable even though their speakers do not diverge in their conceptual schemes.  I also think that the reverse is possible.<18>   I think we must conclude that intertranslatability (or failure thereof) is an innocent by-stander that gets shot in the debate over Conceptual Relativism.  What is really at issue is the possibility of genuine conceptual 'distance' and the possibility of conceptual inaccessibility, as well as the relationship between them.  By shifting the discussion from these possibilities to questions of intertranslatability, and by playing fast and loose with the notion of translation, both proponents and opponents of Conceptual Relativism end up committing themselves to implausible positions.
    Examples of the sort given above can be explained and described in our language.  This by itself may give the lie to the claim that they illustrate the possibility of genuine conceptual difference.<19>   The anti-relativist may be willing to acknowledge the existence of schemes of concepts that are alternatives for us: ones we can comprehend and explicate, or represent, using our own conceptual resources.  The 'very idea' the anti-relativist wants to reject is that of alternatives to us: conceptual schemes that are supposed to be incommensurable with ours, or are otherwise inaccessible to us.<20>   The examples of failures of translation we have seen, though, presuppose broad conceptual similarity between the others and us.  So the anti-relativist may insist that they are besides the Conceptual Relativist's point.

1.3 Failures of Interpretation

    At this point, it could be pointed out that what should matter to the anti-relativist argument is not translation, but interpretation.  On a widely accepted conception (due to Davidson), the interpretation of a sentence in an object language consists in providing its truth-conditions in a metalanguage.<21>   So, you have given the interpretation of, say, the French sentence "La neige est blanche" when you've specified in a language you understand what the world would have to be like for that sentence to be true.  Importantly, the constraints on (Davidsonian) interpretation can be regarded as in some ways less stringent than the constraints on translation (ordinarily understood).<22>   For example, a book-length description of the conditions in the world that have to obtain in order for a Malagasi sentence containing the term [kabary] to be true would constitute an interpretation of that sentence; but it would not qualify as a translation.  (Similarly for the Hebrew gender-marking example.)  These examples show that not all translation failures are interpretation failures.
    It might then be suggested that the sharing of a conceptual scheme should be tied to inter-interpretability, not to intertranslatability.  If so, establishing conceptual divergence or inaccessibility would require giving examples where interpretation is blocked.  And the anti-relativist would insist that this is what cannot be done.  This last claim is partially supported by the fact that, so often, exotic examples produced by way of illustrating conceptual divergence are so often accompanied by specifying truth-conditions for the exotic utterances in our language.  Even if we grant that such specifications do not amount to translations, it still seems far from clear that the relativist can meet the challenge to illustrate genuine interpretation failures.
    I think we should also be cautious about linking conceptual divergence or inaccessibility too closely to failure of interpretation.  First, the explanatory devices actually used in making sense of untranslatable discourse sometimes fall short of, and other times outrun interpretations (read as specifications of truth-conditions).  Secondly, it can be argued that there are conceivable cases where even interpretation fails, yet where conceptual access is still possible and vice versa.  So substituting 'interpretation' for 'translation' would not help very much the anti-relativist's argument against conceptual diversity.  Furthermore, the denial of failures of interpretation seems no more necessary to a rejection of conceptual relativism than the denial of translation failures.
    Let me briefly illustrate how interpretation might fail.  I shall first give a somewhat technical example that comes from an early work by Davidson, "Truth and Meaning."<23>   Using Tarski's theory of truth, and adopting his solution to the Liar paradox, Davidson accepts that the truth-predicate for each natural language is not expressible in that language, though it is expressible in other languages.<24>   We cannot specify in English truth-conditions for sentences involving the predicate "true in English", on pain of paradox; though we can speak about these sentences in French using, say, the predicate "vrai en Anglais".  The point of the example is that, at least on Davidson's own semantic treatment, it will turn out that any French sentence containing the predicate "vrai en Anglais" cannot be interpreted in English.  So this illustrates a failure of interpretation, and a principled one at that.<25>   However, I would argue that the concept expressed by "vrai en Anglais" is not inaccessible to us as English speakers. <26>   Intuitively, it seems wrong to think of each language's own truth-predicate as hiding a kind of 'conceptual blind-spot'.  As long as we have mastery of the relevant formal apparatus, there seems to be nothing to block our access to the concept which the French speaker expresses using "vrai en Anglais" (or vice versa), failure of interpretability notwithstanding.  Here, conceptual access seems secured by our understanding of the semantic goings-on, where this understanding is not, as it happens, manifestable in our ability to interpret.  So I take this example to illustrate both the possibility of interpretation failure and the fact that interpretation and conceptual access may pull apart.<27>
    Let us consider a less technical example where interpretation fails.  Certain Australian languages (e.g., Walbiri) are reported by linguists to have a rather limited numerical language.  They have names for one, two, a few, many.  Now whereas we can presumably interpret Walbiri sentences involving number words in English, there is no way to interpret infinitely many English number sentences in Walbiri as it stands now.  (Just consider English sentences such as "the cube root of ten is not a rational number.")<28>   So we get here systematic (though partial and one-way) interpretation failures between English and Walbiri.
    An anti-relativist might deny that this type of interpretation failure supports any exciting thesis of conceptual divergence, because in describing the mathematical limitations of the Walbiri case (he might say) we presuppose a lot of conceptual similarity between the Walbiri speakers and ourselves.  So, even if we have succeeded in showing some conceptual differences between the Walbiri and ourselves—viz., that we possess, and they lack a certain range of mathematical concepts, the conceptual divergence illustrated is of a very limited and local sort.  I have two comments in response.  First, how limited the divergence is in this case will depend on the extent to which mathematical thinking permeates other areas of thought and action.  Genuine differences in mathematical concepts may indeed turn out to lead to a rather interesting case of conceptual divergence.<29>   Second, if the anti-relativist resorts to denying that the kind of interpretation failure illustrated by the Walbiri's inability to interpret our mathematical discourse does yield an exciting enough case of conceptual divergence or inaccessibility, it looks like we are no better off with interpretation than we were with translation.  In both cases partial failures will be said to come in two varieties: the exciting and the not exciting.  And we still do not have a principled way of separating the cases, so as to be able to use the im/possibility of interpretation as a test for conceptual distance.

2. A `Theory'-Theory of Linguistic Understanding

    [THIS SECTION IS TO BE EXPANDED.]

2.1 Linguistic Understanding as Theory

    The conception of understanding I see at work here can be described as a kind of 'theory'-theory.<30>   The basic idea behind this conception is that full understanding of what other speakers say and mean, at home as well as abroad, is the outcome of a theoretical explanation of their behavior (verbal and non-verbal).  Our entry into others' speech, both as novices and as mature interpreters, is provided by observation of speakers' behavior.  The behavior serves as our data, or evidence on the basis of which, with the help of various hypotheses, we infer what they mean, think, etc.  We have achieved understanding of what they mean when we have matched with the sounds that come out of their mouths with the interpretation dictated by our theory.<31>
    Under the broad canopy of this 'theory'-theory one finds strange bed-fellows, such as Quine and Davidson, on the one hand, and Fodor and Chomsky, on the other (and there are many others who join).  Both Quine and Chomsky, for instance, invite us to think of a child learning a first language as in crucial respects like an amateur theorist of the language spoken in his community.  In probing a language, Quine tells us, both child and radical linguist use as their data "the concomitances of .. utterance and observable ... situation" (1969:81).  And Chomsky presents "[t]he problem for the linguist, as well as for the child learning the language" as that of "determin[ing] from the data of performance the underlying system of the rules" that govern the language, which requires having "a method for devising an appropriate grammar, given primary linguistic data." (1975: 4, 25).  On this picture, the task for both the child and the linguist is to settle on a theory of the language at hand.  In the child's case, this (obviously tacit) theory is expressible in the more or less stable set of linguistic judgments that can serve to characterize her as a competent speaker.  In the linguist's case, the theory may be explicitly expressed in terms of a set of rules for the language under study.<32>
    The child-linguist analogy is not a modern invention.  In an early passage of the Philosophical Investigations, Wittgenstein ascribes it to Augustine:

Augustine describes the learning of human language as if the child came into a strange country and did not understand the language of the country; that is, as if it already had a language, only not this one.  Or again: as if the child could already think, only not yet speak.  And 'think' would here mean something like 'talk to itself'. (1953: #32)

A speaker is a mapping from messages onto wave forms, and a hearer is a mapping from wave forms onto messages.  The character of each mapping is determined, inter alia, by the conventions of the language that the speaker and hearer share.  Verbal communication is possible because the speaker and hearer both know what the conventions are and how to use them: ... The exercise of their knowledge .. effects a certain correspondence between the mental states of speaker and hearer. ... The speaker ... has a certain [message] in mind and the hearer can tell [what message] it is. (1975: 108)

really is like a theory at least in this, that it is derived by wit, luck and wisdom from a private vocabulary and grammar, knowledge of the ways people get their point across, and rules of thumb for figuring out what deviations from the dictionary are most likely. .. [This is like] the process of creating new theories to cope with new data in any field—for that is what this process involves (1986: 446).

2.11 Quine&Davidson v. Chomsky&Fodor

    Following Quine, Davidson upholds, whereas Chomsky and Fodor reject, a "Publicity Requirement" for language:

The correct understanding of linguistic expressions by their users must be in principle possible.  The meanings of linguistic of expressions must be publicly available to and discoverable by the learners of a language.<35>

    Fodor and Chomsky, on the other hand, want to bring the metaphysics of language fully on a par with a realist metaphysics in other domains, and believe there is no more room for skeptical anxiety in studying language and mind than there is in any other area.  On their view, facts about meaning consist in facts about certain kinds of internal goings-on in the 'minds/brains' of speakers (to use Chomsky's locution).  The assignments of meanings we ordinarily make when interpreting others' speech are in effect hypotheses about those internal goings-on.  As interpreters of others' speech, we may, at least in principle, be as wrong about what they mean as we can are about other internal goings-on (say, in regarding processes in people's stomachs).  Meaning assignments are thus assimilated to theoretical conjectures in other areas where evidence may underdetermine theory.  And they are equally in good realist standing.
    The Quine/Davidson reasoning that leads to the indeterminacy thesis relies on the claim that public evidence is insufficient to determine (intuitively distinguishable) meanings.  The realism advocated by Chomsky and Fodor consists in insisting that hidden facts about goings-on in speakers' minds-brains can serve to determine what public evidence leaves undetermined.  But note that there is another way to reject Quinean skepticism about meaning.  One could deny that public evidence is insufficient to determine meaning assignments.<37>   Elsewhere, I have argued that the support for this claim can itself be traced to a tacit acceptance of the 'theory'-theory  conception of linguistic understanding, a conception that I am claiming the Quineans share with their Chomskian opponents.  If we can free ourselves from that conception, it may turn out that we need neither resign ourselves to indeterminacy nor embrace meanings as hidden, hypothesized internal goings-on.

2.2 'Theory'-Theory and Conceptual Relativism

    There is another point of convergence between Davidson and Fodor on the issue of Conceptual Relativism.  We have seen that Davidson denies the possibility of genuine, philosophically interesting conceptual differences across cultures.  We can now see this denial in the context of his acceptance of a 'theory' theory of language.  For Davidson, understanding others is a matter of devising a theory for interpreting their utterances.  But the methodology we must employ in devising such a theory has built into it a rather strong Principle of Charity, which guarantees that we must see them as conceptually like us in all fundamental respects.<38>
    Interestingly, Fodor arrives to the same idea of fundamental conceptual similarity beneath the surface of apparent cross-cultural diversity, though by a rather different route.  For him, as we saw, understanding is a matter of translating into one's language of thought.<39>   This means that genuine conceptual difference could arise only if translation into the language of thought were in particular cases impossible.  But, at least for human languages, the possibility of such translation is guaranteed by the empirical fact that any normal human speaker is born capable of learning any human language.  (Indeed, this is part of the empirical support Fodor cites for his innate language-of-thought hypothesis.)  Fodor would, then, treat any practical obstacles in matching the expressions of two natural languages as owing to merely superficial differences in conventional ways of encoding of mental messages.  Human thought (as opposed to speech) is conducted in a single innate language which is universal in Tarski's sense: it is capable of expressing anything expressible in any natural language, past, present and future.<40>
    Whether we take Davidson's transcendental route or Fodor's empirical route to the denial of conceptual diversity, we end up in roughly the same place.  Actually, it is not any one specific place.  In a sense, we end up all over the place, since the kind of position Davidson and Fodor converge on finds our basic concepts, our basic beliefs, everywhere.  Depending on your perspective, you could see this position as a form of either cognitive imperialism (or colonialism), or cognitive parochialism.<41>   Such a position does not seem to me very plausible, since I think the possibility of interesting conceptual diversity, and even inaccessibility, is intuitively a perfectly intelligible one.

3. 'Non-Theoretical' Understanding and Conceptual Diversity

    So far, I have tried to connect the denial of diversity with a 'theory'-theory conception of understanding others.  I will now connect the acceptance of diversity with what I consider to be a more intuitive, non-theoretical conception of understanding.

3.1 A Bit of Conceptual Diversity

    Part of the 'denial strategy' is to downplay all potential candidates for conceptual diversity.  Thus, Davidson tells us that the existence of "a common co-ordinate system on which to plot" the conceptual differences "belies the claim of dramatic incomparability" (1984: 184).  Fodor might make the same claim, taking the 'common co-ordinate system' to be the language of thought.  But let us consider again the Walbiri number example.  It seems that whereas we can represent the Walbiri number system, compare, and contrast it with ours, using the expressive resources of English, Walbiri speakers cannot do the same.  We may allow that Walbiri speakers could undergo serious retraining and learn our mathematical concepts, and that Walbiri could evolve so as to allow its speakers to talk about them.  But this does not detract from the intelligibility or interest of the claim that right now, the Walbiri mathematical thinking is significantly different from ours, and that at present a whole range of our mathematical concepts is beyond the Walbiri conceptual access.  To the extent that mathematical concepts permeate numerous aspects of our thought and action, the differences between us may be far more reaching than seems at first.<42>
    The Walbiri example illustrates an asymmetrical case involving one-way conceptual inaccessibility, where we may assume that the inaccessibility could be remedied by retraining or re-education.  (Using previous jargon, we might say that, whereas the Walbiri system is only an alternative for us, our system is an alternative to them.)  I think even more extreme examples can be conceived, ones where there is threat of mutual inaccessibility, which cannot be remedied so easily.  We can conceive of intelligent, conversing beings who possess radically different sense organs from ours.  Imagine that, unlike us, they are completely incapable of perceiving colors.  Or, imagine that they (like bats, and unlike us) possess an ecolocation sonar system for detecting the presence of objects in total darkness.  Let us further suppose that we share enough science with the aliens so that we can fully interpret their experience-discourse and vice versa.  That is, we can make enough discoveries about each others' brains (as well as about relevant environmental inputs and behavioral outputs), so as to be able to match each others' experience utterances with truth-conditions.
    Let us assume, with physicalists, that all this puts us and the aliens in a perfect position to have theoretical knowledge of all the facts about each others' experiences, and thereby to achieve a certain kind of mutual understanding—theoretical understanding.<43>   We are able to develop correct theories about the experiences of the aliens based on the available evidence—we can explain and predict their behavior, and vice versa.  Still, it does seem intuitively that our understanding of each others' experiential discourse is incomplete.  The reason, it seems, is that, because our sensory experiences are so radically different, we and the aliens diverge in our conceptualization of colors and sounds (at least).  We possess, if you will, different 'experiential concepts'.<44>   In the familiar case of incompatible color divisions, we can easily master the others' color concepts (and they ours) and, with suitable practice, apply them so as to classify things by their colors as they do.  By contrast, in the present case, given that we are physically unable to undergo each others' sensory experiences, the mutual accessibility of our 'experiential concepts' would seem blocked.<45>
    Let me remark that my reason for claiming that mutual understanding between the aliens and us is compromised is not that experiences, in general, are essentially 'private' or ineffable.  We can suppose that members of each group can communicate about their experiences, and fully know about and understand each other's experiences.  If members of each group could somehow come to have the sense organs of the members of the other group, they could fully understand their experiences as well.  Furthermore, I am not relying on the idea that there are 'subjective' properties of experiences (or facts about them) that are inaccessible from an observer's 'objective' point of view.  At issue here are concepts, not properties.  All that is required for my present point is a distinction between experiential concepts and properties of experiences.  I take it that this is not a controversial distinction: e.g., the concept WATER can be different from the concept H2O, even if the property of being water is taken to be one and the same as the property of being H2O.
    Now, if the theoretical conception of understanding sketched earlier were exhaustive, the possibility of mutual interpretation should have guaranteed our mutual conceptual access and full understanding of our respective discourses.  But it does not seem to.  I submit that our intuitive assessments of conceptual distance and the acknowledgement that genuine conceptual diversity is possible rely on a broader conception of understanding, to which I now wish to turn.

3.2 'Non-Theoretical' Understanding

    We have seen that, on an intuitive conception of understanding, interpretation (at least in the Davidsonian sense) is no guarantee of full understanding or conceptual access.  We may be in a position to provide truth-conditions for someone's discourse without possessing all the concepts they employ when engaging in the discourse, where these concepts may even be unavailable to us, in the sense that we may be unable to acquire them.  This is what the case of aliens was supposed to illustrate.  We have also seen that there could in principle be understanding and conceptual access even where interpretation is not possible.  As the truth-predicate case illustrates (above, p.xx), there may be principled limitations on the expressive resources of languages, which would prevent us from specifying the truth-conditions of certain utterances, even though we do have mastery of all the concepts involved in making the utterances, and are therefore in a position to understand them.
    I take this to suggest that the commonsense conception of understanding is in some ways broader than what I've called the 'theoretical' conception.  Commonsense seems to recognize more ways of understanding than is dreamt of by `theoretical' philosophers.  Can we make sense of commonsense here?  What else could there be to understanding others besides the theoretical understanding that is captured by the philosophical notions of translation, interpretation, and theoretical explanation?
    Let us go back to our examples.  In the aliens' case, we should consider that, given our respective physiological make-ups, the aliens and we would not be able to employ each others' sensory concepts, despite the fact that we can have a theoretical grasp of the conditions under which the aliens apply them.  Since we could not adopt the aliens' sensory concepts, we might say that their sensory 'conceptual framework' constitutes an alternative to ours; it is not an alternative for us.  In the truth-predicate case, what allows us to understand the French speakers' utterances involving the predicate "vrai en Anglais", despite our inability to specify their truth-conditions in English, is the fact that we are able to employ the concept expressed by the French predicate.  The semantic thinking of the French is not an alternative to our own.
    Whether or not a (portion of a) conceptual framework can count as an alternative to (a portion of) ours may not depend on whether we can have what I have been calling 'theoretical understanding' of it.  It may depend instead on whether or not we could use the scheme to get around in the world.  A conceptual framework is, as the common metaphor goes, a way of 'seeing the world'.  We have access to another's conceptual framework if we can 'see the world through the other's eyes'.  I am suggesting that 'seeing the world (or part of it) through others' eyes' is a matter of using their concepts.  It requires having your own thinking of things—your mental acts of perception, categorization, etc.—structured and guided by their concepts.  Sometimes we can recognize our ability for this despite our inability to find words in our language with which to translate the other's speech or to specify the truth-conditions of their utterances.  And, other times, we can recognize our inability to do so despite the availability of translation or interpretation.  (Here we have the aliens case, but perhaps sometimes closer to home: the mentally retarded, the criminally insane, or the multiple personality.<46>)
    What plays a crucial role in our recognition of conceptual difference in such cases is a perceived failure to project ourselves into the other person's position and act or think as s/he does or would.  In such cases, we may well be able to theorize; what we are unable to do is empathize.  This inability may account for the perceived incompleteness of understanding or limited access.  Empathy, then, is an ingredient missing from the theoretical conception of understanding.  Theoretical understanding may offer only limited help when our aim is to understand understanders.  A full understanding of understanders is empathetic; it seems to require being able to employ their concepts.<47>
    Empathy's role in understanding has been increasingly appreciated by, of all people, Quine, whom we presented earlier as a leading 'theory'-theorist in language.  Already in Word & Object, Quine pointed out that in the actual practice of radical translation as well as in the ordinary attribution of propositional attitudes at home, we resort to an 'essentially dramatic idiom': "we project ourselves into what, .., we imagine the speaker's state of mind to have been, and then we say what, in our language, is natural and relevant for us in the state thus feigned" (1960: 219).  Still adhering to the child-linguist analogy, Quine has explicitly mentioned empathy as what "dominates the learning of language, both by child and field linguist."  The parent volunteers utterances and encourages the child's utterances based on the parent's assessment of the child's perceptual orientation.  And the linguist "imagines himself in the native's situation as best he can."  Similarly, the ascriber of a belief or desire etc. that such-and-such "projects [the content] empathetically to the creature in the attitude," be it another human subject, or an animal.  In all this, "practical psychology is what sustains us" and its "method ... is empathy".<48>
    When speaking in a theoretical vein, however, Quine's position concerning 'practical psychology' is still skeptical.  In another recent book, he describes empathy as coming into play when objective considerations give out; its use signals that we have moved away from solid ground and into infirm territory, where almost anything goes, objectively speaking.<49>   Quine, then, would seriously doubt that the deliverances of our commonsense empathetic understanding could constitute a legitimate subject matter for scientific study.<50>   But other contemporary authors seem much more optimistic, and have been trying to develop empirically workable alternatives to the 'theory' theory which incorporate the idea of empathy.<51>   Of course, to ground their optimism one would need a full articulation of the alternative 'no-theory' conception of understanding, which is not something I have provided here.  Instead, I have tried to motivate a search for an alternative by considering the debate over Conceptual Relativism.<52>

4. Conceptual Divergence and Conceptual Relativism

        [THIS SECTION IS TO BE REVISED.]

(IV) The worlds in which possessors of different conceptual schemes live are themselves different from one another (since "[a] world exists only relative to an imposition of concepts").

[reality is] a vast production…something humanly produced and maintained. (Carey (1989:26).)

[A]ll facts are discourse specific [(since no fact is available apart from some dimension of assessment or other)] (Fish (1980: 199).)

[W]e do not make stars as we make bricks; ... The worldmaking mainly in question here is making not with hands but with minds, or rather with languages or other symbol systems. Yet when I say that worlds are made, I mean it literally... (Goodman (1980: 213).)

4.1 Construction, Adjudication, and Relativity

    I think that an anti-relativist confronted with a constructivist would do well to go after the Constructivist claim itself, and the way it is supposed to be supported by the Impossibility of Adjudication and Relativity claims, rather than denying the possibility of Conceptual Diversity.  And there are plenty of things to say here.
    To begin with, the constructivist seems guilty of a confusion between concepts and things conceptualized, a version of the mention-use confusion.  The constructivist idea seems driven by considering a rather extreme case of diversity: the case of genuine alternative conceptual schemes or frameworks—ones that are in some sense or other globally incommensurable.  Using this idea, constructivists often try to support their position by putting forth the claim that we could not adjudicate among alternative schemes.  This is the Impossibility of Adjudication claim:

(II) We cannot adjudicate among divergent conceptual schemes.

(III) The correctness of a conceptual scheme is relative to culture, society, historical period; there is no absolute, culturally- (or socially-) independent standard for assessing conceptual correctness.

5. Conceptual Schemes and Reality<57>

    [THIS SECTION IS TO BE REVISED.]

1 A word of caution.  In recent years, there has been a very lively debate in the philosophy of mind/psychology between proponents of the so-called 'theory'-theory, and their opponents, who often subscribe to the so-called Simulation Theory.  (For discussion, see the papers in Davies and Stone (1995a and 1995b).)  What I shall be referring to as a 'theory'-theory of linguistic understanding bears some relations to various views that have been grouped under that same title in that debate.  However, I do not take myself here to be engaging directly in that debate, nor do I take myself to be offering direct reasons in favor of a 'simulation theory' of linguistic understanding, since I am not sure, at this stage, what such a theory would amount to.
    I discuss some of the connections between what the 'theory'-theory of linguistic understanding and the 'theory'-theory of mind in "Semantic Eliminativism and the `Theory'-Theory of Linguistic Understanding" (ms.).  In Linguistic Understanding: Theory and Practice (a monograph in progress), I offer further discussion of these connections, and take some steps toward developing a more positive, 'no-theory' conception of linguistic understanding.

2  I argue this briefly in Bar-On (1992).  The argument is developed more fully in "Semantic Eliminativism …"  I touch on it briefly in 2.11 below.

3  For a critical assessment of Dummett's argument, see Bar-On (1995).

4  For this statement of constructivism, as well as criticism, see Devitt (1991: 157, and passim).

5 Or, at any rate, its analogue in the moral case would not seem to be a claim properly separable from the relativity claim.  The moral relativist analogue of (IV) would be the claim that the moral realities in which possessors of different moral codes live are themselves different from one another.  But what is there to the claim that the moral realities of these groups are different, over and above the claim that the groups adhere to different (and irreconcilably conflicting) moral codes?  The problem here is that the idea of a moral reality that is independent of any moral values/standards makes little sense.  What would make such a reality distinctively moral?

6  COMMENT ON WHAT BECOMES OF THE RELATIVITY CLAIM ON THIS LINE.
    Obviously, what I have offered here is a very rough, if not caricaturist sketch of the constructivist line of thought.  CITE SOME DIFFERENT VERSIONS OF CONSTRUCTIVISM.  CF. CARNAP's "Linguistic Frameworks".  REF GOODMAN, PUTNAM, RORTY.

7  The status of such findings is far from controversial.  It is important to recognize that the attribution to another culture of adherence to the said principle will depend on our interpretation of their various practices, as well as their moral vocabulary.  CITE FOOT ON A KIND OF 'CHARITY PRINCIPLE' BUILT INTO THE INTERPRETATION OF MORAL VOCABULARY AND HOW IT IMPACTS CLAIMS OF MORAL DIVERSITY.  REFs on MORAL DIVERSITY.  (RACHEL, EDEL's NEW BOOK.)

8  REF.  We shall return to this case below, p. x.

9  Consider the term "blue" in English.  English speakers describe both the sky on a clear day and periwinkles as blue.  In Hebrew, by contrast, there is a sharp division within the 'blue range' that corresponds to the division yielded by "pink" and "red" in English.  The term [kakhol] is reserved for the darker shades of blue, whereas the term [t'chelet] is reserved for the lighter shades.  Thus, it would be as incorrect for a Hebrew speaker to describe the sky on a clear day as [kakhol] as it would be to describe a pink flower as "red" in English.  Though this is not as extreme a case of incompatible color divisions as the one I envisage in the text, it can lead to problems of translation.

10  Notice that, on the face of it, the conceptual relativist is in a peculiar position compared to the moral relativist with respect to the diversity claim.  The moral relativist typically presupposes a prior understanding of the moral discourse of the allegedly different culture.  And she tries to support the claim of moral diversity by describing in our language the others' actions and moral pronouncements or beliefs taken as understood.  However suspect such a presupposition of understanding may be in the moral case, in the conceptual case it would seem detrimental.  How can you presuppose that you understand the other culture's discourse, while at the same time claiming to establish that its members are genuinely different conceptually?  REF DAVIDSON.
    This, by the way, may help explain why considerations about language and translation figure more prominently in discussions of conceptual relativism than they do in discussions of moral relativism.  ALTHOUGH CITE FOOT AGAIN.

11  A prime example of this strategy is Donald Davidson's line in his well-known article "On the Very Idea of a Conceptual Scheme," henceforth "OIVC"  (1984: 183-198).  In that article, Davidson sets out to reject Conceptual Relativism by undermining the 'very idea' of a conceptual scheme.  He tries to do that by denying the possibility of significantly different conceptual schemes—that is, by denying genuine conceptual diversity.  And he does that by, in effect, denying the possibility of untranslatability between languages.
    Davidson can be seen as supporting P by an identification he proposes early on in "OVICS" between conceptual schemes and sets of intertranslatable languages (1984: 185).  Davidson also explicitly attributes the necessity claim to proponents of Conceptual Relativism: "The failure of intertranslatability is a necessary condition for difference of conceptual schemes" (op. cit. 190); "The test of [conceptual] difference remains failure or difficulty of translation" (op. cit. 191).  The context is one in which he is attacking the proponents' attempt to make sense of the idea of alternative conceptual schemes; and so it may seem as though he does not himself wish to endorse the necessity claim.  However, the necessity claim follows from the identification of intertranslatability and sameness of conceptual scheme, which Davidson does endorse.
    For a fuller treatment of Davidson's argument see my (1994).  1.1-1.3 summarize those parts of that paper which will be relevant to my discussion here.

12  REF WHORF.

13  This intuitive idea is behind the following quotations:

Words ... constitute, in the words of Humboldt, "a veritable world which the mind, ... must interpose between itself and objective reality". (von Wartburg, 1969: 169)
Language is the expression of the form in which the individual carries the world with him. (von Wartburg, 1969: 162)
The difference in languages ... implies a different conception of the world. (von Humbolt, in Leitzman 1905: 27)

14  Notice that, on the face of it, the conceptual relativist is in a peculiar position compared to the moral relativist with respect to the diversity claim.  The moral relativist typically presupposes a prior understanding of the moral discourse of the allegedly different culture.  And she tries to support the claim of moral diversity by describing in our language the others' actions and moral pronouncements or beliefs taken as understood.  However suspect such a presupposition of understanding may be in the moral case, in the conceptual case it would seem detrimental.  How can you presuppose that you understand the other culture's discourse, while at the same time claiming to establish that its members are genuinely different conceptually?
    This, by the way, may help explain why, as we shall see below, considerations about language and translation figure more prominently in discussions of Conceptual Relativism than they do in discussions of moral relativism.  ALTHOUGH RECALL FOOT, FN 7.

15  For relevant references, more examples, and discussion, see Bar-On 1987: Chapter V) and (1993).

16  For an example of a grammatical mismatch, consider:

        English:      (i) Rabe put the basket on/under the table.
                         (ii) The table on top of/under which Rabe put the basket was damaged.
        Malagasy:  (i') (similarly) [Rabe put the basket on/under the table.]
              but:      (ii') [The table which was basket?put by Rabe was damaged.]

    When the Malagasi relative clause is extracted and put in the subject position, the exact locative relation which "table" bears to the verb is lost.  Keenan argues that proposed translation of English (ii) into Malagasi will fail to be exact.  I'll skip his reasons, and point out instead the problem that arises when you consider translation in the other direction.  Any natural English candidate for translating (ii') will inevitably contain more information than the original Malagasi sentence, and, to this extent, will be an inexact translation.  Note also that the Malagasi (ii') can be asserted in ignorance of whether Rabe put the basket under, on top of, next to, etc. the damaged table, but it doesn't imply such ignorance; the sentence simply remains silent on the issue.  You might want to try: "The table under which or on top of which or next to which or ... Rabe put the basket was damaged".  But such a sentence would clearly suggest that the speaker doesn't know which is the case, and so would be at best misleading.  REF KEENAN + Bar-On (1995).

17  For a much more extensive discussion (including more examples and references) see Bar-On (1987: Chapter 5) and (1993), where I defend at length the claim that we should not be so permissive as to count just any content-conveying gloss as a translation.  But we shall come to this point shortly.

18  Take English-speaking flat-earthers, or (true) Berkeleyian idealists, or committed astrologists.  By ordinary standards, they share our language, so (again, by ordinary standards) no issue of intertranslatability arises.  Yet they seem to differ from us conceptually.  They clearly have fundamentally different conceptions from us of some fairly basic matters.  Furthermore, the radical differences between our beliefs should raise the suspicion that we diverge in some key concepts.
    Committed Davidsonians might deny that we can assume they speak the same language as ours (meaning earth by "earth", etc.), all appearances to the contrary.  But it seems the only way to do that is to make—question-beggingly—(apparent) conceptual difference a sufficient condition for non-homophonic translation.  Furthermore, this maneuver would be useful only if we can guarantee that no other conceptual differences (generated by other fundamental disagreements in beliefs) would surface when we produce non-homophonic translations of their discourse.  This is, I believe, highly questionable.  (For relevant discussion and examples, see, e.g., Aune (1987), Rescher (1985) and Devitt (1990: Ch.s 9 & 13).)

19  Such examples, as Davidson puts it, "can be explained and described using the equipment of a single language. … Different points of view make sense, but only if there is a common co-ordinate system on which to plot them; yet the existence of a common system belies the claim of dramatic incomparability."  (1984: 184)

20  The distinction between alternatives for us and alternatives to us is due to Lear (1982).
    Davidson's "OIVC" can be seen as an attempt to argue that there can be no alternatives to our conceptual schemes.  His argument trades on the identification of shared conceptual schemes with sets of intertranslatable languages which we have seen reason to reject.  Davidson says: "we must say much the same thing about differences in conceptual scheme as we say about differences in belief: we improve the clarity and bite of declarations of difference, whether of scheme or opinion, by enlarging the basis of shared (translatable) language or of shared opinion."  (1984: 197, my emphasis.)  And elsewhere, where Davidson summarizes his "OVICS" argument, he also puts it in terms of translation: "if translation succeeds, we have shown there is no need to speak of two conceptual schemes, while if translation fails, there is no ground for speaking of two."  (1980: 243).  (Note that this formulation of the argument remains misleadingly silent on whether what is at issue is total or partial failure of translation.)

21  Where the object language can—but need not—be the same as the metlanguage).  For Davidson's theory of interpretation, see Essays 1-5 and 9-11 in his (1984).

22  In (1987: Ch.V) & in (1993), I argue that our ordinary assessments of translations are highly contextual and appeal to an equivalence relation which is different from the relation of sameness of truth-conditions.  Often it is more stringent.  But in other cases it is looser.

23  Reprinted in Davidson (1984: 17-36).

24  Indeed, the truth-predicate for each natural language L is expressible in any natural language other than L, provided it has the appropriate semantic vocabulary.

25  The same will hold for other semantic predicates such as "refer".  Of course, the success of the example will depend on one's particular approach to the Liar.  For a critical review of a number approaches to the Liar, as well as an anti-hierarchical approach, see Simmons (1993).  The point of the example is simply to illustrate how failures of (Davidsonian) interpretation could arise, and how even if they did, this should not provide grist to the conceptual relativist's mill.

26  Davidson himself presents this case as illustrating relative conceptual deficiency: "there may in the nature of the case always be something we grasp in understanding the language of another (the concept of truth) that we cannot communicate to him" (1984: 29).  He takes failure of interpretation of this kind to establish that there is at least this much conceptual divergence between ourselves and speakers of any other language: they possess, whereas we lack, concepts expressed by (certain) semantic predicates of their language, and vice versa.  I beg to differ.  (And, arguably, there are reasons why Davidson himself should not take the case this way.)
    Incidentally, if we were to agree with Davidson, we should be able to devise a 'twin-earth' case in which two individuals differed conceptually (with regard to certain semantic concepts), even though there were no differences between their ('narrow') psychologies or their environments.

27  Indeed, it may be insisted that it is manifestable through our ability to translate the relevant expressions.  For, I would argue that our own English predicate "true in English", while inadequate to interpret, in the Davidsonian sense, her "vrai en Anglais," is perfectly adequate to translate it.  This would be just another example where a perfectly good translation does not preserve extension, or truth-conditions.  Such examples are prevalent.  (See my (1993: section 1).)

28  See Keenan (1978: 174f.).  Keenan refers his readers to a manuscript by Ken Hale, "Gaps in Grammars and Cultures" (1971) for supporting data and analysis.

29  REF RESNIK.  It is best to think about the case by considering what the users of the richer set of concepts can think which users of the impoverished system simply are unable to think.  Thinking of small children is this regard may be useful.

30  But see footnote 1 above.

31  I recognize that this characterization needs elaboration.  In particular, there may be different versions of this conception, pertaining, for instance to whether the view concerns the process of the acquisition of language understanding, the actual practice of linguistic interpretation (radical or not), or the character of the relevant concepts (e.g., meaning, reference, synonymy, etc.).  REF DAVIES AND STONE's INTRO (1995).  I allow myself some liberty here, since the authors I cite below as representatives of what I am calling a 'theory'-theory of linguistic understanding explicitly treat language acquisition, as well as intra-linguistic understanding, on a par with the interpretation of foreign speech.
    I do attempt a fuller characterization in my "Semantic Eliminativism …"  I leave fuller discussion for my Linguistic Understanding: Theory and Practice (in progress).

32  For further articulations of this 'theory'-theory by these authors, see Chomsky (1975: Chapters 1&2), and Quine (1960: Ch.2), (1974), and (1976: 57f.).
     REWRITE.  Over the years, Chomsky has gradually moved away from attributing to the 1st-language learner a process of selecting among alternative grammars.  His present picture is one according to which the child's innate "universal grammar" places heavy constraints on the set of possible grammars the child can acquire, leaving certain `surface' linguistic features (characterized by "parameters" of the universal grammar) undetermined.  These 'parameters' become 'fixed' upon exposure to the idiosyncratic linguistic input of the particular language the child ends up acquiring.  REFs.  While this may seem like a denial of the 'theory'-theory, it really is not.  For, on the present picture, we are still to think of the linguistic input as evidence on the basis of which the child can determine (unconsciously, as always) inductively, the surface features of her language.  This determination is what Chomsky (still) counts as constructing a theory of the language the child is acquiring.  The point of enriching the innate universal grammar, so as to leave very little unsettled, is to make the process of theory construction by the child possible, given the limited amount of data he has to work with.  (This is Chomsky's way of solving the "poverty of the stimuli" problem.)  ANALOGY TO PERCEPTION?  For relevant discussion of Chomesky's view, see, e.g., REFs.  See also Bar-On (1992) and (1997).

33  I say "surprisingly", since it's well known that Davidson would have no truck with Fodor's language of thought, and differs greatly from him on so many other issues in the philosophy of language and mind, metaphysics, and epistemology.

34  Davidson distinguishes a 'prior' theory from a 'passing' theory.  "For the hearer, the prior theory expresses how he is prepared in advance to interpret an utterance of the speaker, while the passing theory is how he does interpret the utterance.  For the speaker, the prior theory is how he believes the interpreter's prior theory to be, while his passing theory is the theory he intends the interpreter to use."  (1986: 442)
     For some critique of Davidson's view as presented in the article from which these quotations come, see Bar-On and Risjord (1992b)

35  Cf. e.g. Davidson (1990: 301, 314).  For discussion of the Publicity Requirement, see my (1992), (1996) and (1997).  REF FODOR AND LEPORE + COMMENT.

36  Quine sees this as reason for being skeptical about so-called 'intuitive semantics'; whereas Davidson claims that the indeterminacy he recognizes is harmless.  REFs.  For some discussion see REFs + Bar-On (1990) and (1995).

37  Indeed, there is a way that Chomsky and Fodor would agree.  For, they do think that, by and large, speakers achieve a great deal of success in figuring out the speech of others, much greater than the success ordinary folks achieve in theorizing about other matters of fact.  Chomsky thinks it is a remarkable fact that we do figure out the grammar of the language we are born into on the basis of relatively meager public evidence.  He takes this success as attesting to the fact that we are born equipped with very rich innate mental endowments.  REF FODOR & LEPORE.  However, Chomsky and Fodor take the further step of taking the 'real' meaning facts to consist in facts about internal goings-on that are implicated in speech and interpretation.  This is the step that I think can be obviated.

38  REFS on Charity.  For Davidson to be able to deny the possibility of conceptual diversity, he must argue (as indeed he tries to) that there can be no uninterpretable speech.  See Bar-On and Risjord (1992) for an argument that, properly understood, Charity cannot afford this strong result.
     Moreover, even a strong Charity principle (such as is needed to support Davidson's "OVIC" argument) would at best help rule out only radical divergence of beliefs.  Given Davidson's "inscrutability of reference" thesis (see 1984: Essays 15 & 16), an interpreter could interpret someone's simple vocabulary items as being about, say, rabbits-on-a-rainy-day, rather than rabbits, consistently with making most of the interpretee's beliefs turn out to be true.  But this means that an interpreter could understand the interpretee as operating with different basic concepts.  The Davidsonian methodology of interpretation does not force us to 'read our own concepts' into the interpretee's mind.  This seems to leave plenty of room for conceptual divergence.
    The right Davidsonian response to this may be to endorse Quinean skepticism: beyond what can be settled by the methodology of interpretation, there is no fact of the matter as to what concepts the interpretee possesses, and thus no factual import to questions of conceptual similarity/difference.

39  REFS FODOR (1975), (198x).
    An interesting twist: Fodor adamantly objects to the view that in good part drives Davidson's argument against relativism, namely, semantic holism.  He says: "I hate holism.  Because holism always leads to relativism, and I really hate relativism" (1991: 299).

40  Given this claim, we are to see the diverse natural languages we are familiar with as essentially like notational variants of each others—each language exhibits its own idiosyncratic phonetic labels of the very same Mentalese items.  The only conventionality of natural languages that is admitted is the uninteresting "Saussurian" arbitrariness.  As Chomsky has declared: 'there is really only one human language.'—and … (I have heard it added) no one speaks it!

41 The latter term is used in Ramberg (1989: 69,85).

42  Although no sharp lines can be drawn here, it could be argued that the Walbiri example is different from our earlier example of the Malagasi term [kabari].  For, introducing English speakers to the concept expressed by the Malagasi term [kabari] would presumably not require re-educating or retraining them.  On the other hand, re-education would be required for members of a pre-scientific culture to gain access into our scientific concepts.  (So Davidson's dismissal of the Kuhnian examples of radical scientific changes may be unwarranted.)

43  This kind of theoretical understanding is not available to the Walbiri speaker vis-a-vis our mathematical concepts.  (In this respect, the Walbiri example illustrates a more radical case than the alien beings case.)
     It should be noted that the lack of theoretical understanding in the Walbiri case is matched by the impossibility of the interpretation of certain sentences in Walbiri.  There may, then, be some interesting link between what I have called 'theoretical understanding' and interpretation (though we should still bear in mind the truth-predicate case).

44  REF LOAR on PHENOMENAL CONCEPTS.

45  For a recent attempt to derive a duality of properties of experience from a duality in concepts, see Chalmers (199x).  I think this attempt is seriously flawed, but cannot argue this here.

46  In these latter cases, we legitimately use the homophonic translation, assigning to utterances truth-conditions disquotationally, yet we come to realize gradually and partly in that way, that these people see things very differently from us, sometimes unfathomably so.

47  The charge that 'theory'-theorists fail to take proper account of the role played by projection and empathy will not be new to anyone familiar with the debate between 'Theory'-Theory and Simulation Theory in the philosophy of mind.  (REF GORDON, HEAL, GOLDMAN.)  I am here putting the charge to a particular use, in the context of defending the possibility of conceptual diversity.  The connections between my claims regarding the role of empathy and those of Simulation theorists remain to be explored.

48  Here are fuller quotations from which some of the remarks in this paragraph are excerpted (See Quine's 198x (Pursuit of Truth):

 Empathy dominates the learning of language, both by child and field linguist.  In the child's case it is the parent's empathy.  The parent assesses the appropriateness of the child's observation sentence by noting the child's orientation and how the scene would look from there.  (p. 42)

 Empathy guides the linguist still as he rises above observation sentences through his analytical hypotheses, though there he is trying to project into the native's associations and grammatical trends rather than his perceptions.  And much the same must be true of the growing child.  (p. 43)

 Practical psychology is what sustains our radical translator all along the way, and the method of his psychology is empathy:  he imagines himself in the native's situation as best he can.  (p. 46)

 Martha empathizes Tom's perception that it is raining just as the field linguist empathizes the native's perception that a rabbit has appeared.  Learning a language in the field and teaching it in the nursery are much the same at the level of observation sentences:  a matter of perceiving that the subject is perceiving that p.  (p. 62-3)

 The evidence is not assembled deliberately.  One empathizes, projecting oneself into Tom's situation and Tom's behavior pattern, and finds thereby that the sentence `The train is late' is what comes naturally.  Such is the somewhat haphazard basis for saying that Tom perceives that the train is late."  (p. 63)

 Empathy is why we ascribe a propositional attitude by a content clause. . . .  The content clause purports to reflect the subject's state of mind rather than the state of things.  (p. 68)

 The language [in which a content clause is specified] is that of the ascriber of the attitude, though he projects it empathetically to the creature in the attitude.  The cat is purportedly in a state of mind in which the ascriber would say `A mouse is in there'.  The quotational account reflects the empathy that invests the idioms of propositional attitude from `perceives that' onwards.  (p. 68-9)

We all have an uncanny knack for empathizing another's perceptual situation, however ignorant of the physiological or optical mechanism of his perception.  The knack is comparable, almost, to our ability to recognize faces while unable to sketch or describe them.  (p. 42-3)

49  Quine remarks that the need to 'read ourselves into the minds of others' arises "the farther we venture away from simple discourse about familiar concrete things," since then "the farther apart the checkpoints tend to be spaced"; that is, the less testable our hypotheses are.  He thinks we naively harbor "an exaggerated idea of" successful communication.  Whereas objectively speaking, beyond "simple discourse on familiar concrete things, ... [t]he miracle of communication... is a little like the miracle of transubstantiation: what transubstantiation?" (Quid., p.29).  (I think that in these passages, Quine may be conflating empathy with the principle of charity.)

50  As for the possibility of conceptual differences, Quine would abide by his earlier, Word & Object position: "There is no telling how much of one's success with analytical hypotheses is due to real kinship of outlook on the part of the natives and ourselves, and how much of it is due to linguistic ingenuity or lucky coincidence. ... We alternately wonder at the inscrutability of the native mind and wonder at how very much like us the native is, where in the one case we have merely muffed the best translation and in the other case we have done a more thorough job of reading our own provincial modes into the natives speech."  (1960: xx) [NOTE ON COMMON MISUNDERSTANDING OF QUINE AS ADOPTING CR; CF. KATZ]

51  SEE FN 46 above. MORE REFs

52  I undertake such a search in "Linguistic Understanding: Theory and Practice."

53  See his (1972), especially pp. 135-140.  Ziff's discussion clearly allows for the intelligibility of conceptual diversity and also alludes to the connection between understanding and empathy I made earlier in an attempt to fund diversity.

54  REF ON MAPS

55  REF PUTNAM, RORTY, etc.

56  REF UNDERDETERMINATION, INC BAR-ON (1986).

57  This section is based on my "Pains, Stains, and Automobiles: Concepts and Reality (A Reply to John Heil’s 'Levels of Reality and the Reality of Levels')" delivered at the Greensboro Symposium in Philosophy on Current Issues in Ontology (3/31-4/2 2000).

58  A fuller characterization would attempt to capture the fact that concepts that are elements of a conceptual framework are, in some sense, not optional.  (This, in contrast with 'made-up' concepts.)

59  REF FODOR.

60  REF.

61  I do not mean to suggest that this conception is cogent.  As remarked in section 4 above, I believe it is rife with confusion.

62  REF HEIL.

63  HEIL: The extension in such a case is held together by "a sprawling, somewhat unruly family of similar properties."  The sharing of a single property, Heil maintains, is not the only legitimate ontological glue or bond; similarity among items' properties will often do too.

64  Could there be bedges if no one did (or would) judge anything to be a bedge?  Note that, as we raise the question, we ourselves are using the concept BEDGE.  It is up to the world whether anything gets caught in the BEDGE-net in the counterfactual situation envisaged.  Though there may be cases where the answer is No, the case at hand does not seem to be among them.
REWRITE The line I am pursuing here may be further clarified by considering how concepts that have 'non-homogenous' extensions behave with respect to the Tarskian truth-schema (which some consider to be the hallmark of realism).  A relevant instance would be: "'That bedge is crumbling" is true iff that [contextually salient] bedge is crumbling."  Properly understood, the schema has a mentioned sentence on the left-hand side, and a used sentence of our own language on the right-hand side.  On my understanding, using the sentence "That bedge is crumbling" assertively does not suffice for committing you to there being a distinct property—being a bedge—that all and only bedges share.  The schema is about the semantic relation of truth, not about ontology.  It should be seen as a semantic principle, rather than a way of ushering in a suspect metaphysics.  So I can use the concept BEDGE to pick out an item that bears only certain similarities to other items properly called "bedge," so as to say something true—namely, that that one is crumbling.