In the previous chapter we argued that the geyi-method could be considered a very general model applicable to all interpretation. In this chapter we discuss major theories of interpretation from the side of analytic and continental philosophy, as well as input from Chinese philosophy. In our view, all too often these theories are enshrined in the ideal language paradigm, in order of increasing degree: Gadamer’s common language, Habermas’ ideal speech situation, Quine and Davidson’s reliance on subject-predicate logic as a model for a language. Then we present a model of “interpretation across philosophical traditions,” including all variants of intercultural philosophy as reviewed in chapter 7.
Both Davidson and Gadamer have acknowledged similarities between their theories. Davidson shares with Gadamer an attempt to provide an account for the general possibility of understanding.1 However, in granting similarities between Gadamer’s hermeneutic interpretation and Davidson’s radical interpretation, one needs to be aware that these similarities can also be exaggerated, as the following passage from Gadamer illustrates:2
It really seems as if there is a real proximity between Davidson’s concept of “radical interpretation” and my own extension of the concept of hermeneutics. … [However,] … What is fundamentally at issue is not primarily science and epistemology but, … the “ontology” of life communicating itself through language. The form of the proposition, and here I stay close to Hegel, is not suited to express speculative truths. Even the model proposition that Davidson employs—“snow is white”—seems strange to me from this viewpoint. Who uttered this, even if it is true?
There are two fundamental features that we believe every theory of interpretation should include. One is the feature of holism(s) and hence indeterminacy or underdetermination (stressed by Quine in particular, but equally acknowledged in some form by Davidson, and reflected in Gadamerian hermeneutic circles). The second feature is the recognition that something like a principle of charity or principle of humanity as discussed in analytic philosophy or the assumption of a shared truth of the subject matter in the hermeneutic tradition is needed as a quasi-transcendental condition of possibility for translation and interpretation.
We start with Quine’s thought experiment of radical translation, which is a behaviorist model of how to go about translating a strange language in a first contact situation, followed by Davidson’s thought experiment of radical interpretation, which is less behaviorist. Radical interpretation aims to tell us what an utterance means whereas, strictly speaking, radical translation cannot do this. It achieves no more than correlating utterances.
Quine first proposed his thought experiment of radical translation in the 1960s.3 It was based on his familiarity with the practice of field linguists at the time: how to proceed in order to edit a dictionary and to fashion a grammar for a language for which no interpreters or any other information is available by engaging in fieldwork communicating with monolingual speakers. Although Quine emphasizes that this concern is a philosopher’s thought experiment, given the sources that he uses, we are on safe ground to claim that Quine’s thought experiment is the theoretical counterpart of the first contact situations.
What Quine learned from the field linguists is such a procedure:4
1.Begin by getting the informant to call out names for some concrete objects, but be aware that even in the naming of ordinary objects languages differ in their classification of things. Even if the investigator and the informant are looking at the same object, they may focus on different aspects or parts of it.
2.Then get short utterances and account for as much of the linguistic and extralinguistic context as one can imagine might condition the meaning of each utterance, such as preceding or following up statements, person speaking, and person(s) spoken to. This information is all the more necessary because much of the original glosses will be incorrect.
3.Build on a large enough stock of phrases and draw your own conclusions about the meaning of the foreign forms. Think of them in their own terms.
4.Even when one is not asking for explicit terms, one can obtain corrections or alternative ways of saying things, because people seem to have an irresistible impulse to translate unnatural speech into their natural speech, and some of this—unless the society’s etiquette forbids such behavior—is verbalized.
What Quine was up to was to investigate whether meaning could be defined via a definition of “same meaning.” First he showed that a definition of same meaning via synonymy is not possible because all relevant notions are interconnected. A definition via substitution salva veritate does not work either; nor does a definition of the same meaning for observables via the same stimuli. A possible definition of the same meaning could be what is shared by translations of the “same” utterance.5 But it turned out that this was not possible either. Quine’s analysis had considerable impact in analytic philosophy, apparently showing that “same meaning” could not be defined.
Quine’s thought experiment of radical translation runs as follows: Assume that a field linguist visits a people whose language is completely unknown to outsiders. How would the investigator go about making a translation manual for their language? Assume that the linguist has all the time in the world and the locals are well disposed; then the linguist should proceed as follows:
1.Observe native utterances and their concurrent observable circumstances, in particular what one thinks might be observation sentences. There will be greetings, questions, and commands as well, but they are noted down more provisionally.
2.Identify indications of assent and dissent.
3.Try out native observation sentences you have collected in the “right” circumstances to see whether you can confirm your hypotheses.
4.Identify logical connectives and quantifiers (notions such as “and” and “all”).
5.Try out nonobservation sentences in conversations.
6.Identifying particular words, rather than complete utterances, will come later.6
Although Quine draws on the actual practice of field linguists, he formulates the thought experiment in such a way that the relevant features are made more explicit (such as the distinction of word and sentence; observation and nonobservation; the role of assent and dissent; and the need to presuppose logic). An important difference with the linguist’s advice is that Quine takes utterances instead of words as basic. We agree with Quine that at the most basic level of interpretation, the focus has to be on utterances. Words (concepts) can only be identified as constituting parts of CS’s when already quite something is known about the strange language.
Quine is not very explicit about his assumptions, but in order to be successful in providing a translation handbook that works in practice, he seems to assume (1990a: 42–46):
1.“Empathy dominates the learning of language” (42).
2.People have short and simple expressions for what is conspicuous to them. The grammar of their language is learnable.
3.Broadly speaking, the strangers have similar interests as other humans. They note the same saliences in the environment, at least often enough.
4.The beliefs of the local people are usually true (according to the standards of the interpreter), in particular beliefs expressed in observation sentences.7
Radical translation works in practice, but there are two theoretical problems that undermine the idea that meanings are well-determined and therefore undermine the idea of “same meaning” shared by translations. There is indeterminacy of translation at the level of sentences and indeterminacy of reference at the level of words. Usually it is said that for Quine, the point of indeterminacy of translation is to challenge the notion of synonymy and hence the reification of ideas (Gaudet 2006: 63). This is correct. However, for our interests, the following remark of Quine is more directly relevant:8 “What the indeterminacy thesis is meant to bring out is that the radical translator is bound to impose about as much as he discovers.”
Quine showed that it is possible to redistribute the meaning of words in sentences without changing the latter’s truth value. This is called indeterminacy of reference. The following “didactic” example may get across the nature of the highly theoretical issue concerning the indeterminacy of reference of words in a sentence. Consider the phrase “five cows” in English and the correlated expression in modern Chinese:9 wu tou niu 五头牛. In Chinese, three characters are needed; in English, two words. The first Chinese character, wu, correlates with five and the third character, niu, with cow(s). There are at least two ways to explain the middle character. One interpretation is that tou modifies the use of the word for five (for example by indicating that the reference is to five big things). A second interpretation is that niu is a mass noun, that is, it could be rendered by the neologism cow-cattle. On the latter reading the translation of the Chinese expression would be something like “five heads of cow-cattle.”10 On the former reading, the translation would be something like “five cows (which are big animals).” In the second interpretation, the distinction between count nouns and mass nouns in English fades away. Both explanations are consistent with other uses of the character tou. According to Quine’s thesis of the indeterminacy of reference, there is no way to decide on the “true” meaning of the parts of the Chinese expression. Relative to translation into English, the meaning of Chinese words is indeterminate. The reference of English words is indeterminate relative to translation into Chinese. The phenomenon of indeterminacy of reference is no barrier to successful communication. It does not matter whether one has the “right” meaning in one’s head (if this were to make sense), as long as one utters the right words in the right circumstances. The truth value of sentences in which the phrase “five cows” or wu tou niu occurs remains the same.
Indeterminacy of translation is analogous to indeterminacy of reference, the former at the level of sentences and the latter at the level of words. Quine’s analysis and arguments show that his thought experiment implies that two radical translators may come back with very different translation manuals, both of which are equally supported by the empirical evidence. Both manuals would be equally right in giving a consistent picture of the meaning of all utterances in the language at issue and are equally useful for talking to people.11
On the basis of these and similar considerations, Quine concludes:
1.One cannot learn the meaning of words one by one: there is meaning holism.12
2.Strictly speaking the meaning of words is indeterminate (given their occurrence in a particular sentence with a particular truth value): indeterminacy of reference.
3.In theory more than one translation manual is possible—each of them equally well supported by the evidence: indeterminacy of translation.
4.The above applies to one’s own language as well. “Radical translation begins at home” (1969: 46).
5.“Success in communication is judged by smoothness of conversation, by frequent predictability of verbal and nonverbal reactions and by coherence and plausibility of native testimony” (1990a: 42).
Quine’s “proof” of the indeterminacy of meaning is for his thought experiment, which brings out the fundamental features of translating an unknown language. However, because of the close similarity of the thought experiment with the actual practice of field linguists, we suggest the two indeterminacies also apply to every concrete interpretation and form the most fundamental feature of the underdetermination of an interpretation by the evidence.
One needs to distinguish three “kinds” of indeterminacy or underdetermination: indeterminacy of reference and translation as explained above; underdetermination of every interpretation or theory by the evidence; and incommensurability. The former and the latter might be considered as two extremes in the following sense:
•Many equally correct translations are possible (indeterminacy).
•Strictly speaking no translation is possible. This is related to the thesis of incommensurability often ascribed to Kuhn and Feyerabend.
The underdetermination of a theory by the data (the evidence) is sometimes called the Quine-Duhem thesis. It says that every theory (and hence every interpretation) is underdetermined by the evidence in the sense that there is always the possibility of finding an alternative interpretation (theory) that fits the data equally well. The Quine-Duhem thesis is mainly discussed in the philosophy of science. However, as we consider interpretation to be similar to a (scientific) theory, the Quine-Duhem thesis applies here as well.13
Davidson’s thought experiment of radical interpretation has enriched radical translation by adding belief-desire-meaning holism to Quine’s meaning-meaning holism.14 Davidson describes the radical interpretation situation as follows: prior to interpretation, the beliefs, and so on, that underlie a fellow-speaker’s utterances, as well as the meanings attached to the latter, are treated together as parameters the values of which are unknown to an interpreter. For instance, she as an interpreter cannot find out what he as a speaker means by the words he uses without making certain determinate assumptions about his beliefs (at the time of the utterance).15 However, the only way she can ascribe beliefs to the speaker is via the attribution of meanings to his words. Hence, she seems to be caught in a circle. A similar difficulty is encountered when she tries to construct the hierarchy of his preferences, or when trying to disentangle his beliefs from his desires—or whatever classification of his attitudes she favors. The only way out of the circle is to assign values to all parameters at once, which is subject to epistemic virtues like coherence and empirical adequacy as well as to the principle of charity (to be discussed in detail in the next chapter). One feature of the latter principle is that if she wants to know what a speaker means, she has no choice but to treat him on the whole as a speaker of truth and having reasonable beliefs.16 Such a principle is a necessary precondition for constructing a theory of meaning for a language. Unless she interprets a person as having largely true beliefs, she would fail to understand what his beliefs are about.
Davidson’s theory of interpretation aims to provide the minimal requirements to be able to construct a theory of meaning for the language of a speaker. An empirical theory of meaning of the speaker’s language tells us what each sentence of the speaker’s language means (i.e., when it is true). In order to carry out the project of explaining meaning in terms of truth, Davidson employs the apparatus of Tarski’s semantic definition of truth. Tarski is using the notion of meaning to define truth. Davidson assumes the latter in order to define the former.
David Lewis (1985) and other critics of Quine and Davidson have argued that their conclusion of indeterminacy is achieved by leaving out constraints, which would prevent underdetermination. For example, one should not exclude from the radical interpretation situation (scientific) knowledge of the speaker considered as a physical system and of his physical environment. Hence, (provisional) ascription of beliefs is possible without knowing any meanings.17 Similarly, McGinn (1984) argues that the perceptual system is autonomous with respect to the belief system (also called categorical perception), following Fodor’s modularity thesis (1983). Therefore, radical interpretation should start with assigning perceptual experiences to the subject, on the basis of the observable physical facts about his environment and his relation to it, and only secondly go from what has been established at the first stage to an ascription of beliefs and meanings. We disagree with Lewis and McGinn. Both of them have too much confidence in cognitive universals.18
Although we borrow insights from Quine as well as Davidson, in particular concerning issues of holism, indeterminacies, and possibly necessary preconditions for interpretation, we disagree with their major presuppositions. As a result, we allow even more holism and underdetermination.
Both Davidson and Quine assume the unproblematic identification of sentences or assertions, even before ascribing particular beliefs, meanings, desires, intentions, and such like. They also assume that the identification of assent and dissent,19 as well as the universality of standard (first-order predicate) logic as the basic grammar of all languages, are unproblematic. Instead we suggest that each assumption about the other person and his or her language should be equally open for revision and contestation.20
Davidson’s theory of interpretation is more sophisticated than Quine’s theory of translation. Among Davidson’s assumptions, the most fundamental one is that of a universal descriptive metaphysics.21 For example, Davidson says:
Truth, knowledge, belief, action, cause, the good and the right, are the most elementary concepts we have, concepts without which (I am inclined to say) we would have no concepts at all. Why then should we expect to be able to reduce these concepts definitionally to other concepts that are simpler, clearer, and more basic? (2005: 20)
We argue that this assumption, although an obvious part of the background of the modern interpreter, has to be modified in the course of interpretation.
While Quine’s thought experiment of radical translation and Davidson’s more sophisticated theory of radical interpretation can be seen as approaching the problem of (intercultural) communicative interaction by seeking the conditions under which meanings can be ascribed to utterances of a speaker, Habermas’ later work can be seen as a linguistic reformulation of his earlier contributions to the Frankfurter Schule. This school (its later developments also being referred to as critical theory) advocated a rethinking of the relation between truth and freedom. But that raises the question: What sort of truth justifies the stance of the critical theorist? Horkheimer said that theory should be motivated by an interest in a future in which people make their own history in a society that is rational in a noninstrumental sense. But how can this goal be distinguished from other “ideological” goals? If “everything” (we, communication, and so on) is alienated and power relations govern all social interactions, how can there be a reasonable rationality? These are the leading motivating questions of Habermas’ theory of communicative action. How can one provide the normative basis of all communicative and other kinds of social interaction, capable of explaining systematically disturbed communication as well? Habermas (1970, 1981, 1998) still sees room for change by understanding Enlightenment ideals of rationality better.
According to Habermas, there are three aspects to communicative interaction: truth (the objective), validity or rightness (the social), and authenticity or sincerity (the subjective). Developments in the Western traditions have pushed validity and authenticity to the margins.22 Rationality has shriveled to instrumental rationality in terms of means and ends. In the everyday world (manifest life-forms), all three aspects of communicative interaction are equally present. But “the systems” (such as the institutions of the state, science, finance), which originally were instituted ostensibly for the well-being of people, became independent, reducing action to instrumental rationality, and thus becoming a threat to the lifeworld by undermining its freedom and meaningfulness. The way rationality worked and still works in the lifeworld, in terms not only of truth, but also of validity and sincerity, must imbue the systems, too. This would require major adjustments of the political organization of society. Interpreting or understanding is never just receiving or mirroring meaning or grasping a correct description of the world. Whatever the context, there is always evaluation, assessment, and negotiation of truth, validity, and authenticity claims.
In chapter 2 we criticized Habermas in that he adheres to a variant of the ideal language assumption. Habermas acknowledges that, normally, speech acts do not fulfill the requirements of the model of pure communicative action; that is to say, communicative discourse that is not disturbed by inequalities and power relations that limit the right of participants to raise questions about any statement, the right to bring up any statement for discussion, and so on. But he argues that one cannot avoid acting, counterfactually, as if “pure” communicative interaction is possible. Human communication rests on this unavoidable fiction or transcendental idea of an ideal speech situation, says Habermas.
There are similarities between the views of Habermas (1998) and Apel (1998) on these issues. Both Apel and Habermas assume that there is a universal human communicative capacity, which can be approached with a transcendental method. According to Apel, the idea of a possible and necessary consensus cannot be refuted. Every attempt to do this is by definition a performative contradiction. Apel puts much greater weight on a transcendental Letztbegründung (final grounding in a priori philosophy) than Habermas, who draws on empirical theories and aims at something like a social theory of formal structures rather than the more traditional a priori philosophy of Apel.
Not all commentators are convinced that Habermas has succeeded in distinguishing between pure and impure communicative interaction, true and false consensus, free and unfree consciousness, and so on. Dussel (1988) has challenged Apel and Habermas by asserting that before there can be “ideal discourse,” there must be life in the sense of secure conditions of preservation. If the latter condition is not met, “ideal discourse” would be an empty generalization.23 Young (1997) and others have criticized Habermas for not having taken into account the “conversational realities” (Shotter 1993) of today, that is, the radical plurality of the polis. Every attempt to achieve reciprocity threatens to undermine difference. We have to find a form of “antisymmetrical reciprocity” (Young 1997: 343). Further, Habermas implicitly reproduces an opposition between reason and desire or feeling, because he devalues and ignores the expressive and bodily aspects of communication. He ignores such language use as greeting, rhetoric, storytelling, and so on (as do the other major theories of interpretation). By focusing on the ideal of a free argumentative discourse for all, other features of communicative interaction are neglected.
When the views of Habermas and Apel are applied to issues involving different traditions. more criticisms arise. For example, Habermas associates premodern forms of life with nostalgia and irrationality (Linkenbach 1986: 292). Therefore, his discourse ethics and theory of communicative interaction is almost completely irrelevant for cross-cultural interpretation. We agree with these criticisms that Habermas neglected the intercultural side. In particular Habermas’ assumption of a tradition-transcending rationality cannot be accepted. For him, concepts such as truth, rationality, and justification play the same role in every language community, even if they are interpreted differently and applied in accordance with different criteria. Such a position does not leave enough room for major divergences.
However, we absorb a number of Habermas’ ideas for our XYZ-model of interpretation (see below). We have already indicated that our notion of form(s) of life can appropriate Habermas’ account of the lifeworld. One of the most important contributions by Habermas is his emphasis that truth/meaning is inseparable from validity/rightness and authenticity/sincerity. This insight has been incorporated into our principle of mutual attunement (see chapter 10). Apel and Habermas’ model of rational argumentative discourse, although not acceptable in general, may be used as a model for the academic dialogue among interpreters, except that no dialogical consensus should be required. The intrinsic underdetermination of interpretation requires productive dissensus.24
Gadamer did not have much to say about interaction with non-Western traditions, but there are quite a few self-claimed successors of the hermeneutic tradition in the circle of Chinese philosophy, as well as intercultural discourse in general. We start with some comments on Gadamer, as he is the one who introduced prevalent hermeneutic notions such as fusion of horizons, prejudice (Vorurteil), and dialogic consensus (Verständigung).25
Gadamer’s hermeneutics starts from the conditions of possibility for actual interpretative practices. It stresses the truth-disclosive, self-transformative function of the fusion of horizons and, following Heidegger, the priority and inescapability of the question of being. A certain kind of relativism is unavoidable, but it does not constitute a hindrance that needs to be overcome. Rather, it is a necessary ontological precondition for being-in-the-world.
Gadamer agrees with Wittgenstein that the limits of language are the limits of the world. There is no view from nowhere outside dialogical interaction, nor is there a transcendental definition of rationality (as Habermas and Apel maintain). Gadamer would also agree with Wittgenstein that form(s) of life, language (games), and world are interrelated concepts.26
The understanding of a language is … an accomplishment of life. For you understand a language by living in it. (1989: 346–347).
Learning to speak does not mean learning to use a pre-existent tool for designating a world already somehow familiar to us; it means acquiring a familiarity and acquaintance with the world itself and how it confronts us. (1976: 62–63)
Gadamer’s paradigm is a dialogue between two beings, an “I” and a “Thou,” which constitute the primitive case of understanding. Dialogue (in Plato’s sense) is the model for every kind of interpretation: everyday meaning-making, textual analysis, translating, participating in a market economy, and so on, are all objects for the hermeneutic approach. Hence, for Gadamer, hermeneutics is a general feature of all human activity. Gadamer also applies his dialogical model of conversation to the interpretation of texts, although “one partner in the conversation, the text, speaks only through the other partner, the interpreter” (1989: 389).
Everything we have said characterizing the situation of two people coming to an understanding in conversation has a genuine application to hermeneutics, which is concerned with understanding texts. (387)
The reciprocal relationship that exists between interpreter and text … corresponds to the reciprocity involved in reaching an understanding in conversation. (389)
Gadamer continues:
Thus it is perfectly legitimate to speak of a hermeneutical conversation. But from this it follows that hermeneutical conversation, like real conversation, finds a common language. (389)
We agree with Gadamer that there is a reciprocal relation between interpreter and text. However, the “common language” Gadamer speaks of is still the language of the interpreter (even though it may have changed during the process of interpretation). As indicated in chapter 2 we take Gadamer’s stress on a “common language” to be a trace of the ideal language assumption. Gadamer writes: “Every conversation obviously presupposes that the two speakers speak the same language” (387). But the last sentence is obviously not what is to happen of necessity.27
Gadamer says interpretation “is like a real conversation in that the common subject matter is what binds the two partners, the text and the interpreter, to each other [and] … are thus bound to one another in a new community” (389).28 We do not think, in particular with respect to interpretation across traditions, one can speak so easily of a common subject matter.29 “Only that translator can truly re-create who brings into language the subject matter that the text points to” (389). Perhaps this might be defended for interpretation within one and the same tradition, but not in general.
We borrow Gadamer’s notion of the dialogical method of interpretation for interpreting texts, but dispense with the affiliate notions of a common language, a fusion of horizons, and a truth that would unite all partners beyond their (initial) horizons.30 Truth always remains relative to a conceptual scheme or language; truth may even be the wrong notion to invoke in an intercultural context.31 Gadamer focuses on different opinions (Standpunkte), which will be fused relative to the truth of the subject matter they share.32 We focus on understanding before the issue of (dis)agreement arises. We agree that “a conversation has a spirit of its own” and has an inner infinity and no end,33 but disagree “that the language in which [a conversation] is conducted bears its own truth within it” (385).34
Gadamer writes: “The text brings a subject matter into language, but that it does so is ultimately the achievement of the interpreter. Both have a share in it” (390). This may be true in isolated cases, but it is not the aim of interpretation when interpreting ancient or “far away” texts, because in the community of interpreters, there would be alternative proposals for bringing the (assumed) subject matter into language. Gadamer is more correct when he speaks of a “fundamental gulf between the two languages.”35 This “gulf” image applies in particular to a contemporary scholar Z interpreting an ancient text X, but in addition there are also “gaps” in the community of contemporary Z’s.
Perhaps most importantly, Gadamer is too optimistic when it comes to the capacities of a translator or interpreter to find “the best solution” by “weighing and balancing possibilities” and to reach “a compromise” in “finding a language that is not only his [i.e., the interpreter’s] but is also proportionate to the original” (389). Gadamer’s dialogical model is envisioned on the primal case of an individual Z translating and interpreting X and/or Y. It needs to be re-envisioned on the primal case of a group of Z’s.
An important genre that combines hermeneutics and Chinese philosophy is Cheng Chung-ying’s onto-hermeneutics, which has been presented in numerous publications, both in Chinese and in English. The aim of Cheng’s onto-hermeneutics is a critical mutual engagement of Eastern and Western thinking in order to rehumanize humanity, drawing on Eastern and Western resources. As to the “onto” part, the assumption is that metaphysical reality is both rationally accessible and intuitively experienced. Human understanding is ultimately rooted in experience of reality. It is assumed that the interpreter “seek[s] an understanding in terms of the meaningful connections of basic notions of a given text, with both intended and factual references to a reality which one could also experience and appeal to independently of the text” (2000: 33). That is to say, the reader lives, inhabits, and experiences the very reality that is textually represented (Ng 2007). The reader is ontologically conceived to be an intrinsic part of the reality that the text describes. Therefore, onto-hermeneutics can be considered as a proposal for a world philosophy, presupposing the language independent accessibility of “the reality that the text describes,” requiring various language independent universals, for example, a strong sense of empathy across traditions: “one can use one’s feelings to feel the feelings of the others and one can expect that others could reciprocate in the same way” (Cheng 2000: 42). Therefore, “knowledge would be a matter of this congruence of mutual feelings reached from this reciprocating process of feeling projection and feeling response.”36
Notwithstanding his universalistic presuppositions,37 Cheng has highlighted a large number of differences between Western and Chinese philosophy. For example, according to Cheng, scholars of past generations have translated Western “ontology” as bentilun 本体论, and then reversely stipulated benti in terms of Western ontology, without thinking of the meaning of benti.38 Proceeding from the etymology of the Chinese language, ben 本, literally meaning root, refers to origin, and thus has the connotation of historicality, temporality, and interiority. Ti 体 [體], literally meaning body, refers to integrity, spatiality, and exteriority. Benti thus refers to what is ultimate in the whole cosmos. It embraces multiplicity while preserving integrity. Cheng argues that Western scholars pursue methodology at the expense of ontology, and cherish knowledge while ignoring understanding.39
Liu Xiaogan has criticized Chinese scholars who exclusively use “intellectualistic” hermeneutic methods, while forgetting about society and self. He argues that Gadamer’s fusion of horizons has been dogmatically applied to the research of Chinese philosophy.40 An example might be Zhang Longxi (1992), who invokes a blanket “universality of the hermeneutic phenomenon” to justify putting Western and Chinese sources “in a sort of critical dialogue” (xiii). But this alleged dialogue is actually a monologue that imposes on classical Chinese poetics a Western concern with things such as language’s “inherent metaphoricity, ambiguity, and suggestiveness; and its implication for the author and the reader” (xiii).
A good example of the defense of the hermeneutic approach in sinology and intercultural philosophy is Roetz (1993). Roetz stresses the familiar hermeneutic suggestion that the “process of understanding is best thought of as a dialogue … a time-transgressing conversation” (102). Other features Roetz mentions are the relevance of the implicit background-presuppositions of the text, the uniqueness of its language, and its historical embedding in specific, contingent circumstances.
Roetz emphasizes that the sense of the text is not comprehensible by an explanation of causes, but only by understanding the reasons of the author. Reasons are explicitly included in our XYZ-model of interpretation presented below. But we should emphasize that ascribing reasons is different from assuming an “authorial intention.”41 As Gadamer remarks:
It sounds at first like a sensible hermeneutical rule, generally recognized as such, that nothing should be put into a text that the writer or the reader could not have intended. But this rule can be applied only in extreme cases. For texts do not ask to be understood as a living expression of the subjectivity of their writers. (396)
Roetz stresses the focus on motivations (i.e., reasons) in hermeneutics, which he considers to be missing in Hansen’s analytic philosophy, but Hansen also ascribes motivations to literati such as Laozi and Mencius.42
Habermas (1986) has argued that Gadamer’s position is relativistic and that he leaves no room for criticizing other cultures. Taylor (1985, 1995) has defended Gadamer saying that Gadamer actually is much more aware of ethnocentric risks. This is illustrated in his emphasis on a critical attitude toward the historical contingency of one’s own tradition, on ensuing Vorurteile and on a much more acute awareness than Habermas of factors that tend to reduce alterity. Habermas’ critics may be correct, but there may be a more unperceptible (and therefore perhaps more pernicious) ethnocentrism in Gadamer’s universal hermeneutics in that he assumes the possibility (at least as an ideal) of fusion of horizons, of a shared language, and of the universality of reason that transcends the local use of language. Because Gadamer cannot give universal criteria of whether and how ideals such as Verständigung and fusion of horizons are achieved by the hermeneutic interpreter, it seems that it is the interpreter who has the last word: in the end, otherness is eliminated.
Many writers have applied Gadamer’s notion of the fusion of horizons to intercultural exchanges. Therefore, it needs to be stressed that Gadamer does not often use the expression fusion of horizons. When he does so, it is only with respect to interpretation within one tradition (i.e., only for interpretation within the Western traditions). He says such things:
Part of the hermeneutical approach [is] to project a historical horizon that is different from the horizon of the present. (305)
We regain the concepts of the historical past in such a way that they also include our own comprehension of them. (376)
In a tradition this process of fusion is continually going on. (305)
In contrast, Taylor argues,
Understanding the other is to be construed as fusion of horizons. (2011: 37)
Gadamer’s account of … the fusion of horizons applies also to our attempts to understand quite alien societies and epochs. (2011: 38)
The “fusion of horizons” operates through our developing new vocabularies of comparison, by means of which we can articulate these contrasts. (1994: 67)
However, as already pointed out, this fusing of horizons is taking place in the language of the interpreter and it is an open question to what extent the horizon of what is interpreted has entered the picture. We hold that the belief that one can find a (shared) truth at the horizon is unwarranted and easily leads to various kinds of asymmetries. Insofar as one can speak of a fusion of horizons at all, it should be understood as pragmatic consensus in concrete situations for the purpose of continuing the exchange, but this will not remove asymmetries. It is not enough to say, as Simpson does, that,
the common language within which this encounter comes to expression need not be seen as hegemonically usurping the original languages.
Simpson’s text continues as follows:
Rather, think of it as a dialect in the making, or perhaps as a specialized vocabulary of each. … Such a common language can serve as the basis for a community structured upon an ever expanding shared vocabulary for discussing, but not for prescribing, moral and cultural identity. (2000: 434–435)
No reasons are given why the “specialized vocabulary” is not subject to hegemonic forces.
Taylor has proposed the following epistemic virtue of comprehensiveness for intercultural or world philosophy: “a more comprehensive account … fuses more horizons,” covering “a wider band of perspectives” (2011: 32). Instead of the bird’s-eye point of view of nomothetical science, “we substitute the ideal of languages that allow for the maximum mutual comprehension between different languages and cultures across history.” Drawing on both Gadamer and Wittgenstein, Taylor writes:43
The adequate language in which we can understand another society is not our language of understanding or theirs, but rather what one would call a language of perspicuous contrast. This would be a language in which we could formulate both their way of life and ours as alternative possibilities in relation to some human constants at work in both … in which the possible human variations would be so formulated that both our form of life and theirs could be perspicuously described as alternative such variations. (1985: 125)
Taylor’s “language of perspicuous contrast” is nothing else than a particular language (that of the interpreter) raised to the level of an ideal language.44 As an epistemic virtue, comprehensiveness is relevant, but Taylor’s language of perspicuous contrast is not very different from more explicit proposals for a universal ideal language based on “constants” to be shared across humanity.
Gadamer himself leaves little room for non-Western traditions. However, occasionally he speculates about the possibility for Eastern languages to rise above the determination of philosophical presumptions due to the Graeco-Latin grammatical structures. The Eastern figurative expression, he says, “acquires its expressive power from the reciprocal reflection of what is meant and what is said.” (1976: 239). This remark conforms to a generally shared assumption that, for example, as ideography, the references of the Chinese language are somehow mirrored in its scripts.
Still, Gadamer wonders whether this language might be an alternative mode of utterance “within one and the same universal, namely within the essence of language and reason” (ibid.). He seems to suggest a double enrichment of our understanding of Western and Eastern languages, which prima facie take contrasting orientations. On the one hand, concept and judgment figuring in Western languages should be understood in relation to their embedment in the life of meaning; on the other hand, the aspect of conation that undeniably exists in Eastern languages can be “drawn into the hermeneutical movement that creates common understanding” (240).45 Gadamer seems to define such a movement as aufheben, which means, as he explains, “to take up and use.” The essence of language as such resides precisely in fulfilling this function of sublation.
In this section, we propose a model of cross-cultural interpretation that is applicable to intercultural philosophy, which is a complex of multiple intercultural real-life or imaginary dialogues.46 The model applies to interpretation of an ancient text from another tradition, as well as to face-to-face intercultural encounter between two (groups of) human beings (as in first contacts or in intercultural philosophical dialogue). Suppose an interpreter, Y, is interpreting the words of a philosopher or of another human being, X. Y uses observations of X to provide interpretations of X’s sayings, inscriptions, and other actions.47 The observations may include:
•observing similarities in forms of life; recognizing unfamiliar human practices as being (somewhat) similar to practices that Y is already familiar with;
•observing X’s (un-interpreted) utterances (including inscriptions);
•observing X’s actions (described in Y’s language);
•observing circumstances (the Umwelt) in which X is situated (both socio-historical context and physical environment).
These observations are not neutral, but depend on the interpreter’s background, which consists of:
•beliefs, expectations, values; concepts employed in forming beliefs, interests that explain the why of these concepts; values that guide interests and underlying practices; relevance of different epistemic virtues; and in particular:
•knowledge of what has been said about X’s utterances and X’s embedment in specific circumstances, including X’s language community.
Usually, and typically in intercultural philosophy, the interpreter or translator Y is already familiar with a wide range of human practices and conceptual schemes, which is part of Y’s background. In cases such as first contacts, X’s behavior and circumstances would play a more important role than the interpreter’s background knowledge (except for a general idea of what might be expected of a human being); while in cases of comparative philosophy, the interpreter’s background knowledge of what the history of commentators have already said about X and X’s contemporaries would be dominant. Observations and background knowledge together allow the interpreter to generate hypotheses concerning:
•meanings of X’s utterances (i.e., translations in terms of Y’s language);
•beliefs, desires, motivations, illocutions, perlocutions, and other attitudes of X;
•presupposed background (language games, practices, conceptual schemes) of the meanings, beliefs, and so on just mentioned (i.e., X’s background);
•and, on the basis of the above: hypotheses concerning (clusters of) conceptual schemes of X, which do not show easy family resemblances with Y’s conceptual schemes and which Y has to learn and elucidate in Y’s language.
The last item on the list can be divided into three stages of complexity of learning X’s conceptual schemes (allowing for all kinds of intermediate cases):
1.Learn X’s concepts by FR-extension of the interpreter’s own concepts; for example extending youxi 游戏 to games or games to youxi, thus provisionally committing oneself to a quasi-universal.48
2.Learn some of X’s concepts from scratch. That is to say, one can learn the extension of a strange word by pointing and communicative interaction. An example would be learning the extension of the word vovetas, for which no extendable FR-concept is readily available in English. Similarly, from written texts concepts can be learned from the context in which they occur, assuming the behavior of the physical laws are the same everywhere (water flows downhill),49 and recognizing similarities in the large in the human practices described in the text.
3.For the more sophisticated stages of interpretation, Y has to learn how X’s concepts are embedded in clusters of X’s concepts and language games. That is to say: one has to learn the sense of a strange concept by learning its use in many different contexts, which are partly described using alien concepts that have already been interpreted and partly described using FR-extension.
Observing similar practices is part of observation. Learning similar conceptual schemes by FR-extension is part of interpretation. X’s conceptual schemes can be explicitly presented in the writings of X or his contemporaries, plausibly extracted from the language X uses, or ascribed as implicit. Ascribing CS’s is always underdetermined by the relevant data and subject to hermeneutic circles.
The general features of the model are supposed to hold no matter how close or how far the “distance” is between X and Y. The particular subdivision of the observations, background, and interpretation is not rigid. Finer discriminations and more parameters can be added, but the model is sufficiently complicated to pin down most of what is important for interpretation. First of all, there is not just one thing that is being interpreted, but numerous things are involved in the process of interpretation at the same time, thus forming a holistic whole. For example, beliefs of X and meanings of X’s words (and other things) are always interpreted at the same time, contributing to the underdetermination of interpretation further discussed in the next section. Further, results of interpretation can be added to or replace part of the background (of Y and X respectively), but the original background of an interpreter (Y or X) cannot be eliminated or transcended completely. This may be called hermeneutic relativity, arising from the Gadamerian notion of Vorurteil.50 Gadamer writes:
To try to escape from one’s own concepts in interpretation is not only impossible but manifestly absurd. To interpret means precisely to bring one’s own preconceptions into play so that the text’s meaning can really be made to speak for us. (Gadamer 1989: 398)
We would rephrase this remark slightly; instead of “can really be made to speak for us,” we would say: “can be made to speak at all.” Y’s background will determine what things the interpreter will find salient, relevant, interesting, important, and so on. Each interpretation is always relative to a theoretically informed, cultural, and linguistic background; hence relative to Y or a group of Y’s. Radical translation and radical interpretation cannot avoid hermeneutic relativity either. As Quine remarks:51
A struggling radical translator will naturally and rightly impose familiar grammatical concepts on the alien language where practicable, and will even warp them a bit and extend the familiar terminology as required. Similar tendencies have facilitated translation down through the ages and are probably responsible for much of the apparent similarity of structure from language to language. (1990b: 166)
Hermeneutic relativity cannot be avoided; it cannot even be avoided by “going native.”52
With some modification the XY-model can be applied to interpreting text X by interpreter Y. Usually a large number of Y would be involved in the project of interpreting X. Some may focus more on philological reconstruction of the text; others on its translation and/or interpretation. Some would claim to have access to the “original meaning” of X and claim to speak on behalf of (the author of) X.53 Others may aim for an interpretation that is relevant to and directed at a particular audience. X’s “words” may already have been transmitted for a long time, to which all kind of commentaries have been added. The text may have been translated into foreign languages and received various interpretations in other traditions. Deceased commentators have to be represented by contemporary interpreters. X may well have contributed to all kinds of contemporary philosophical discussions via the “original” text and via later interpretations, commentaries, and translations. Thus, interpreting text X can be understood as a dialogue among all philosophers, translators, commentators involved.
The model to be proposed for comparative philosophy is slightly more complicated. Nevertheless it is based on a similar dialogical model. Consider the comparison of texts, philosophers, or traditions represented by the letters X and Y. For either X or Y, what has been said in the previous paragraph applies: a group of philosophers, Zj(X), are involved in interpreting X, being in dialogue among one another concerning their interpretations; and another group, Zk(Y) are interpreting Y. A comparison of X and Y can be given the form of an imaginary dialogue between X and Y, being constructed by a group of contemporary philosophers Zi.54 So, the primitive model of comparative philosophy is one in which interpreters, Zi set up an imaginary dialogue between some X and some Y. Both X and Y are represented by a group of interpreters, Zj(X) and Zk(Y) respectively, each of these groups being in dialogue among one another. The group of interpreters setting up such an imaginary dialogue, Zi, are in dialogue with the groups Zj(X) and Zk(Y), as well as in dialogue among one another.55
Although we give the notion of dialogue a central place, this notion should not be idealized. In daily practice invoking an ideal notion of dialogue may not help anything in a situation of serious conflict. Gadamer’s dialogical consensus, Habermas’ (or Apel’s) discourse ethics, or Heidegger’s “genuine dialogue” are all too idealized.56 However, for academic purposes (as in exchanges among Zi), Habermas’ laws of argumentation can perhaps serve as a (required) quasi-universal of communicative interaction.
Interpreting a text, a philosopher, or a tradition is always underdetermined by the evidence (the data). Meanings, beliefs, intentions, and such like are all interpreted at the same time. No specific entity can be interpreted independently of numerous other interpretations. Every interpretation is relative to a context or background that cannot be described completely (the problem of complete description). Each particular facet of interpretation is connected to other facets of interpretation by hermeneutic circles (or holistic relations). That is to say, changing the interpretation of the one would change the interpretation of the other, and vice versa. For example, if Y notices an inconsistency in the beliefs of X, the conclusion of inconsistency (in terms of Y’s criteria) may be well supported, but it may be possible to disclaim the inconsistency by assuming that the meanings of some of X’s utterances are different.
In the context of Davidson’s theory of interpretation the holism of meaning and belief is highlighted in particular. Beliefs can be attributed only holistically through their location in a structural pattern. This leads to the more general holism of language (meanings) and thinking (beliefs about the world) and the holism of text and surrounding forms of life. In translating books such as Alice in Wonderland (by Lewis Carroll) and Gödel, Escher, Bach (by Douglas Hofstadter) into modern Chinese, the translators realized that the point of such books lies in the overall effect of the text and they did not hesitate to make numerous local changes in the meaning of the text in order to preserve the global charm of the original in translation.57
Another form of holism, implicit in Davidson’s theory of interpretation, is as follows. Knowing the meaning of the characters occurring in, say, the phrase bai ma fei ma 白馬非馬 (white-horse not horse) consists in knowing the meaning of these characters in all other utterances in which one or more of these characters occur. One strand in analytic philosophy (Frege, Dummett) restricts the thesis of holism to the level of an utterance. Davidson assumes holism across all sentences of a language.58 We take the view that, although Davidson is strictly speaking correct, in practice one may often work with a plurality of wholes (larger than one utterance) rather than a single whole. This agrees with Hacking (1986: 457): “Why have one big whole, rather than lots of lesser wholes?”59 Only in exceptional cases, related to high-level meta-CS’s, would the interpretation of a very large number of conceptual schemes be affected.60 For example, the meta-CS of “material object” may be presupposed as a quasi-universal in virtually every conceptual scheme associated with contemporary languages.61
The analytic philosopher’s holism can be considered as a counterpart of continental philosophy’s hermeneutic circle (also called the principle of reciprocity between what grounds and what is grounded). Hermeneutic efforts are directed at finding an interpretation that can both make sense of the individual parts of a text and integrate them into a consistent whole. Here the “consistent whole” is called the horizon of the text and may refer to the whole text (work), the whole corpus of a philosopher or a school, or an even larger semantic horizon at a particular time. But this account is still too simplistic. There are numerous hermeneutic circles. The hermeneutic circles or the “holisms” the interpreter has to deal with cause an indeterminate number of indeterminacies of underdetermination. Whatever detail of the XYZ-model one zooms onto, one would encounter more varieties of holisms or hermeneutic circles, such as:
•Meanings of words are interpreted synchronically, not consecutively. There are no basic concepts that can be exempted from holism.
•There is already a hermeneutic circle going from the parts to the whole and vice versa “inside” a Chinese character.
•There is holism between the language of the text and the language in use at the time.
•Interpretation of passage and work, of work and corpus, of corpus and Umwelt are interdependent.
•There is holism between motivations and background.
•There is holism between the background of Y and the ongoing interpretation.62
•There is holism between logic, grammar, and semantics of the language of X.
•There is a hermeneutic circle between object and method of interpretation.
•There is holism of epistemic virtues, both for one interpreter and across interpreters.63
•There are hermeneutic circles within and between the necessary preconditions and constraints of interpretation.
•There is holism of different hermeneutic circles.
Chinese philosophy (or Western philosophy for that matter) can be considered as a living example of the thesis of underdetermination of interpretation. As Tan Sor-hoon (2005) observes in a review of the Encyclopedia of Chinese Philosophy (Cua 2003), there are few really authoritative or representative views in Chinese philosophy that are supported by broad consensus. Surely there are “writers who have made original contributions” (to the development of Chinese philosophy). But key terms in classical texts are translated and interpreted differently, sometimes even leading to straightforward contradictions (in English). From a scientific point of view, this may seem wrong, but it can also be seen as positively right, because central concepts like ren 仁 (benevolence?) and li 理 (pattern?) have many dimensions and “the different ways in which authors employ them and the different characteristics, contexts, and philosophical stakes they pay attention to enrich one’s understanding [of these notions]” (112). This is one possible reason for the great variety of interpretations: the concepts attached to such characters, being understood differently in different sources (different X). Another reason for the variety is the way in which different interpreters (different Y) manage the holism of relevant hermeneutic circles.
We conclude that first contacts, Quine’s arguments, the practice of comparative philosophy, and the arguments in this section show that underdetermination is an intrinsic feature of interpretation. Underdetermination implies that there is no such thing as the correct interpretation. There is an indefinite manifold of more or less plausible interpretations—the degree of plausibility primarily depends on epistemic virtues that are favored.
A concrete example of balancing beliefs, motivations, meanings, and background (of Y and X) is the discussion on Gongsun Long’s “white-horse not horse.” Gongsun Long 公孙龙 famously put forward such a seemingly self-contradictory statement:64
白 馬 非 馬
bai ma fei ma
[a] white horse(s) [is/are] not/no [a] horse(s)
Most interpreters assume that the utterance is (meant to be) true and Gongsun gives reasons for it being true. But in Chinese (and in English) the statement seems to be straightforwardly false. If I am standing in front of a white horse I am standing in front of a horse. If Gongsun believes the statement to be true, it cannot be an empirical proposition, but must be a logical or other a priori proposition, for example: the concept “white horse” is not the concept “horse.” If Gongsun believes that the statement is false, it could be a simple empirical (false) statement; but what could be the reason for arguing for such a trivially false statement? Alternatively, the statement might be intentionally ambiguous, in which case one should try to preserve the ambiguity of the statement in translation.
In what might be called the common sense reading of Gongsun’s text, Graham suggests that Gongsun starts with a “dubious analysis of the white horse into two mutually pervasive parts” (i.e., the color appearance part and the horse gestalt part) and proceeds with “relentlessly logical working out … the implications of its suspect premises” (1989: 86, 90).65
It is important to note that all interpretations and translations have to fit the same data: 白馬非馬 (bai ma fei ma). The data allow either a realist, or conceptualist, or nominalist interpretation (Cheng Chung-ying 2007). The distinction between concrete and abstract object, as well as between general and singular term, is independent of stimulus meaning (Hansen 2007: 479). Hence, the same data allow for different philosophically laden interpretations of the text of Gongsun:
•“Horseness” and “white-horseness” are two different universals (Feng Youlan).
•The attributes “white-horse” and “horse” are two different attributes (Hu Shi).
•Gongsun dismisses reference and relies on sense (Lai Whalen 1995).
•What Gongsun with the word fei denies is solely identity, while his attackers take him to deny class inclusion (Chiemelsky).66
•Ma 馬 is the name of the mereological set of horses. Hence horse-stuff is an object (substance or kind-thing) scattered in space-time (Hansen 2007), which object includes all kinds of nonwhite horses.67
•Fung Yiu-ming (2007) has argued that if one interprets Gongsun’s dialogue as an argument between a theory of direct reference and a theory of descriptional reference, bai ma fei ma is both plausibly true and plausibly false.
Even the logical structure of Gongsun’s argument is a matter of dispute, even among analytic philosophers.68
Among commentators there is considerable variety in how the context is drawn: the phrase, the text, the corpus, or how the text fits into ongoing issues of debate at the time the text was composed. For example, as to the latter, it has been suggested that Gongsun was responding to the confusion caused by Confucius and Mozi defining what is right differently (Lai Whalen 1995). Contrary to tradition (according to which the sage kings define what is right), Mozi and Gongsun would have believed there must be a “rational” way of establishing what is right. Therefore, Gongsun would have tried to rectify natural language, concluding that common sense is wrong to say (a) white horse is (a) horse.
If one restricts the focus to the few texts of Gongsun that have survived, one may argue for or against a hermeneutic circle between Gongsun’s bai ma fei ma text and an opaque text, which Graham gave the title “Pointings and Things,” in which the character zhi 指 plays a central role (Graham 1989: 90–94).69 Interpretations may also be influenced by a prejudice concerning the “status” of the author of the text. In this case, the question of sincerity and motivation again enters into hermeneutic circles. Compare such different comments:
Gongsun Long stands out as a philosopher of language and its ontology. … being an analytic philosopher in our times. (Cheng 2007: 537–538)70
Since we know he was an intellectual prankster, we cannot assume the texts will present cogent arguments for a well-reasoned philosophical position. (Fraser 2009: §6.1)71
If the assessment by Fraser were correct, it would be more plausible to expect ambiguities in Gongsun’s text.
The question of translating and interpreting bai ma fei ma illustrates that (depending on a holistic web of knowledges and judgments of interpreters) a range of plausible interpretations are possible of the same data. This holistic web includes how the variety of hermeneutic circles involving the data are assessed and includes the views of X’s contemporaries X may be responding to, as well as the knowledge concerning the views of earlier and contemporary commentators.