As we have demonstrated repeatedly, incommensurability is a semantic phenomenon closely related to the problem of cross-language communication: Whether successful linguistic communication is possible between two sufficiently disparate P-language communities. At the extreme end of the spectrum of different degrees of communication breakdown is the failure of mutual understanding. Understanding the members of one’s own language group, even within one’s own family, can be taxing. Understanding cross-differences of classes, races, genders, religions, cultures, and traditions can seem all but impossible. However, we find that understanding an unknown P-language with incompatible M-presuppositions turns out to be more challenging and problematic. It is not just the vocabulary and grammar of the language that is unknown, but also the way of thinking and justification, the mode of reasoning, the categorical framework, and the cosmology, all of which constitute the linguistic context from which the language derives its life, that are incomprehensible to us. The failure of mutual understanding between two incompatible P-language communities suffices to break down the cross-language communication between them.
We have contented ourselves so far with vague and pragmatic pronouncements on the notions of understanding and communication. To give a sufficient interpretation of the concept of incommensurability, the time has come to clarify the notions of the failure of cross-language understanding and cross-language communication breakdown. Presumably, to do this we need a theory of cross-language understanding, which would specify both necessary and sufficient conditions of cross-language understanding, and a theory of cross-language communication. However, considering the vagueness and difficulties that go with the notions of understanding and communication, to produce a general account of cross-language understanding and communication would be much more than we can achieve here. Fortunately, such a general account is not needed either considering our limited purpose here. What concerns us are not the notions of understanding and communication in general, but specific senses of understanding and communication: effective cross-language understanding and successful cross-language communication. Furthermore, to explain the failure of cross-language understanding or cross-language communication breakdown, it would be sufficient if we are able to locate some essential semantic and/or conceptual obstructions between two P-languages and thus to identify a significant necessary condition of effective cross-language understanding and that of successful communication between two language communities. We will deal with the failure of cross-language understanding in this and the next chapter (chapter 13) and leave the topic of communication breakdown for chapters 14, 15, and 16.
The notion of effective cross-language understanding will become clear in due course. Nevertheless, a few preliminary remarks about the concept, the subject, and the object of effective cross-language understanding should be helpful here. First, what do we mean by ‘to understand a (declarative) sentence or a language’? ‘To know or comprehend its meaning’, of course, one may say. But what is supposed to be included in its meaning? What is the meaning of a sentence anyway? To answer these questions, we need to know the basic functions of linguistic expressions to be understood. German psychologist K. Bühler (1934) put forward a tripartite schema of language functions that places the linguistic expression in relation to the speaker, the world, and the hearer/interpreter: The speaker comes to an understanding with the hearer about something in the world. According to Bühler’s semiotic model of linguistic expression, a linguistic expression functions simultaneously as symbol (correlated with states of affairs), as symptom (depending upon the speaker’s intention), and as signal (its appeal to the hearer). J. Habermas1, following K. Bühler, J. Austin, and J. Searle, identifies those three aspects of linguistic expressions as the three structural components of speech acts or the three dimensions of the meanings of utterances:
(a) The propositional content: What is said literally and explicitly, not implicitly, with a linguistic expression that is supposed to represent states of affairs and can be defined in terms of its truth conditions in the case of a sentence.
(b) The expressive content: What is intended with a linguistic expression by the speaker.
(c) The illocutionary content: What is used in a speech act to enter into a relationship with the hearer/interpreter.
Presumably, a comprehensive understanding of a linguistic expression should deal with all these aspects of meaning. More specifically, understanding a written declarative sentence usually involves not only a comprehension of its propositional content, but also, if not always, appreciation of the intention of the author, the ‘illocutionary force’ with which it is issued, and the conversational implicatures of what is said in some specific contexts, an appreciation that goes far beyond knowing its propositional content. Following the convention, we can separate the propositional content of a declarative sentence from the other two contents, i.e., the expressive and the illocutionary content, and call the former the semantic content and the latter the pragmatic content.
Accordingly, we can identify at least two notions of understanding. On the one hand, to understand a sentence could mean to grasp its propositional content (not propositional attitude), the thought expressed, or its literal meaning (conventional meaning, namely, the meaning determined by the linguistic conventions of a language), which we can call propositional understanding?2 Following the dominant view these days in the philosophy of language, we need to distinguish propositional attitude, including the content of a belief, expressed by a speaker in terms of a sentence from the propositional content asserted by the sentence. The content of beliefs and other kinds of propositional attitude are characteristically intensional: The content of a speaker’s belief that ‘A is F’ may differ from the content of another of his belief that ‘B is F’ even if ‘A is B’ This is because belief is taken by many to be a relation between believers and propositions or ‘modes of acquaintance with propositions or ways in which a believer may be familiar with propositions’ (N. Salmon, 1986, 441). Hence, the content of beliefs might be essentially the property of particular persons, so that different interpreters could apprehend the same state of affairs in similar ways, though never in the same way. In contrast, the propositional content and the literal meaning are characteristically extensional: If A and B are co-referential, then the literal meaning or the propositional content of the sentence ‘A is F’ and ‘B is F’ is the same. Subsequently, (literal) meaning and propositional content are objective, or at least intersubjective, in the sense that they are not in the head of an individual believer, but are properties of a language that are publicly accessible and communicable.
On the other hand, to fully understand a sentence could mean, besides to grasp its propositional content, to capture the illocutionary force that the sentence induces on the hearer, the speaker’s particular intention in uttering the sentence on a particular occasion, or the speaker’s particular beliefs and other propositional attitudes associated with the utterance. Clearly, since the propositional content of a sentence is the semantic foundation on which its other dimensions of meaning depend, propositional understanding constitutes the central core of any notion of understanding.
Since we are mainly dealing with scientific languages, which are public languages composed of declarative statements, we can, for the sake of clarity, focus on their semantic dimension only. In fact, recall that Kuhn and others discover that a failure of cross-language understanding between two scientific languages cannot be simply taken as evidence of the interpreter’s limitation of knowledge or lack of interpretative skills. It is a failure of mutual understanding between two P-language communities as a whole, rather than between some individual speakers with different dialects, intentions, attitudes, or conflicting interests. It clearly involves some deep semantic and/or conceptual obstructions between two substantially different languages. Therefore, in our following discussion, we will be only concerned about the essential core of any notion of understanding, i.e., propositional understanding. More precisely, corresponding to the speaker’s act of expressing a thought via a sentence to the interpreter about a state of affairs, there stands the interpreter’s apprehension of the thought expressed:
The interpreter is able to understand a sentence S if and only if he or she can grasp/apprehend its propositional content.
Second, who is supposed to be the subject of propositional understanding? We are not interested here in how each individual member of a language community understands another language, but rather how a language community as a whole understands it. Thus, a speaker’s (as an individual speaker of a language) meaning or intention on a particular occasion and an interpreter’s (as an individual interpreter of a language) personal apprehension of a sentence have no place in our discussion. Whenever I mention ‘the speaker” or ‘the interpreter’ of a language, I mean all the speakers or interpreters as a whole or the representative speaker or interpreter, which roughly corresponds to U. Eco’s ‘Model Reader’ who agrees to abide by the rules and conventions of a linguistic game in order to reach a coherent understanding, rather than some specific individual speaker or interpreter with different transient psychological states, such as Eco’s empirical reader with a specific personal encyclopedia or world knowledge. Based on a commonly accepted assumption about linguistic competence—the competent speakers of a language can and do effectively understand the language and fully communicate with one another—the so-called literal meanings of sentences of a language are actually the meanings understood and used by the speaker of the language. In this sense, ‘the literal meanings’ of the sentences of a language are actually what I might call, more precisely, ‘the speaker’s language-meanings’ (not an individual speaker’s meaning), or ‘language-meanings’ for short to emphasize the fact that the meanings are conventionally established by the language. Although I will still, following the convention, talk about the literal meaning in the discussion below, please do bear in mind that I am actually referring to the speaker’s language-meaning.
Finally, ‘what to be understood’ is a language as a whole, not some isolated words or sentences of a language. Of course, the language to be understood is not just any kind of language, but rather a P-language as we have identified.
A variety of accounts of cross-language propositional understanding have been proposed. None of them, however, are able to identify a significant necessary condition of understanding. I have discussed one account extensively in chapter 2, that is, the translational account of understanding, according to which the interpreter can understand an alien language by translating it into the interpreter’s own language. I argued that translation is neither a necessary nor (necessarily) a sufficient condition for understanding. Understanding an alien language is always possible through language learning no matter whether the interpreter can translate it into his or her own home language.
Others contend that one can understand a language by understanding its components (the compositional account of understanding). But understanding a language is obviously different from understanding its words or sentences. In fact, in any given case of cross-language understanding, there would always be a failure to grasp or understand some of its specific parts. The experience of understanding tells us that sometimes the interpreter can understand a sentence of another language without understanding some of the terms involved. For example, for a sentence of phlogiston theory, ‘phlogiston is phlogiston’, failing to understand the term ‘phlogiston’ does not keep the interpreter from understanding the sentence itself. To understand each word of a sentence is not necessary for understanding the sentence. More to the point, it is quite uninteresting to learn that understanding a language requires a comprehending of its parts. Even if understanding the parts of a language is necessary for understanding it, it is too trivial a position to be useful as an explication of the failure of cross-language understanding. In addition, as I will argue later, understanding each word of a sentence is not sufficient to understand the sentence as a whole.
One might argue that these problems with the compositional account of understanding lie in an ill-formed view of meaning and understanding, namely, that a word or a sentence could have its meaning in isolation. But to understand one part, words or sentences, of a language we have to understand the language as a whole, since the meaning of a part is determined by its place in the whole language. Therefore, others who follow the holistic view on meaning and understanding correctly insist that to understand a sentence of a language one needs to identify the unique role or the function of the sentence in the language. However, how can we identify the unique role or the function of every sentence in a language? Presumably, we need a recursive procedure to identify the roles of infinite sentences from some limited subset if language is learnable. As D. Davidson points out, the role of a sentence in a language can be determined by constructing a theory of truth for the language based on Tarski’s semantic notion of truth. Then, the above functional account of understanding can be reduced to the truth-conditional account of understanding.
Among many other accounts of propositional understanding that I will not discuss here, Davidson’s truth-conditional account of understanding is by far the most appealing one (Davidson, 1984). According to it, the Tarskian semantic notion of truth plays the most essential role in linguistic understanding due to a necessary conceptual connection between the knowledge of (Tarskian) truth conditions and understanding. To understand a sentence, it is necessary and sufficient to know its truth conditions; to understand a language is to know the truth conditions of any sentences of the language.
Many have challenged Davidson’s basic doctrine that Tarski’s semantic notion of truth is central to a theory of understanding by casting doubt on the effectiveness of the knowledge of Tarskian truth conditions to understanding. Some raise doubt about the thesis that the knowledge of truth conditions is necessary for cross-language understanding. They claim that the requirement is too strict since Tarskian truth conditions of many sentences are either unknowable or unavailable to the interpreter. Other alternatives are offered that do not appeal to truth conditions, such as M. Dummett’s (1993) verificationist conditional account, which focuses on assertability conditions, instead of truth conditions, and W. Sellars’ and G. Harman’s (1984) conceptual role account. On the other hand, for some, the knowledge of truth conditions is too trivial to be sufficient for understanding. In my opinion, whether the knowledge of Tarskian truth conditions is sufficient for understanding depends on what counts as qualified knowledge of the truth conditions. If we strengthen the knowledge of the truth conditions strictly enough— for example, if ‘to know the truth conditions of a sentence S in a language L’ means ‘to know the proposition expressed by the T-sentence of S in terms of knowing the T-sentences for all other sentences in L’—I fail to see why one cannot understand a language by knowing the truth conditions of all its sentences.
However, the real trouble with the truth-conditional account is that even if we were able to achieve understanding by strengthening the knowledge of the truth conditions, we would have to give up the hope that the same knowledge could serve as a necessary condition for understanding. This can be shown briefly as follows. Suppose that an ordinary English speaker Jenny understands a sentence S in English,
(S) Smith is cowardly.
That means that Jenny grasps the proposition expressed by S in English. But for Jenny to know the truth conditions of S, she has to know the proposition expressed by the T-sentence of S,
(Ts) ‘Smith is cowardly’ is true in English if and only if Smith is cowardly.
To know what is expressed by Ts, Jenny has to know that it is verified as true in terms of a theory of truth for English that gives all T-sentences to other sentences in English. Jenny needs a few strong assumptions to derive the propositional knowledge of Ts from the propositional knowledge of S, such as the assumptions that Jenny tacitly knows Tarski’s semantic notion of truth and that she tacitly knows how to construct a Tarski-type theory of truth for English. I do not think an ordinary competent English speaker can and should master these assumptions in order to understand English. A speaker may understand a sentence well without knowing its truth conditions. Without specifying a significant necessary condition for understanding, the truth-conditional account cannot explain the possible failure of understanding between two substantially disparate languages.
Please notice that I do not intend to argue here that the knowledge of truth conditions cannot be a necessary condition for understanding. As I have mentioned above, it depends upon what counts as the knowledge of truth conditions. If we interpret the knowledge of truth conditions weakly enough, it could be a necessary condition for understanding. For example, if we interpret the content of the ‘know that’ clause as a propositional attitude of Jenny’s, then for Jenny to understand an utterance of ‘Smith is cowardly’, it is at least necessary for her to know that it is true if and only if Smith is cowardly. For her to know that theorem, it is not necessary for her to have any ideas about Tarski’s semantic notion of truth and of how to construct a Tarski-type theory of truth for English.
More importantly, as to be argued later, it is the knowledge of truth-value conditions, not the knowledge of truth conditions, that plays an essential role in cross-language understanding.
A P-language is underlain by a particular cosmology loaded with a set of existential presumptions and universal principles, embodies a unique mode of reasoning, and is embedded with an exclusive lexical structure. It is not too much of an exaggeration to claim that a P-language as such reflects a form of life. The conceptual richness of a P-language determines the depth and inclusiveness of understanding it. A P-language is fully intelligible, and its purported justification is adequately understood by an interpreter only if its underlying M-presuppositions, including its cosmology and its mode of reasoning, are fully comprehended. Any interpreter who fails to comprehend the underlying M-presuppositions cannot see the ‘point’ of the language, and thus cannot become an engaged communicator. Consequently, effective cross-language understanding cannot be achieved. This is the reason why in so-called ‘abnormal discourse’ (to borrow R. Rorty’s terminology) effective mutual understanding across two conceptually disparate P-languages is problematic and difficult.
Consider the following three core sentences from three different P-languages:
(4) The association of the yin and rain makes people sleepy.
(30) Element a contains more phlogiston than element b.
(31) Electrons have eternally hidden nuclei.
Recall that (4) is a sentence from the language of traditional Chinese medical theory (CMT); (30) from the language of phlogiston theory; and (31) from the language of an imaginary physical theory. Now consider Dr Smith, an interpreter who is educated within contemporary Western scientific tradition. Suppose that he is familiar neither with phlogiston theory nor with CMT and its underlying pre-modern Chinese mode of reasoning. Is Dr Smith able to effectively understand (4), (30), and (31)?
Dr Smith can understand each word of (31). After all, he knows what an electron is, what it means for something to be eternally hidden, and what it means to say that a particle possesses a nucleus. Certainly one may be forgiven for thinking that the hypothesis about the eternally hidden nuclei no more involves ‘an empty play of words’ than claims about permanently confined quarks seriously discussed and defended in contemporary physics. Besides, Dr Smith realizes that (31) is in good syntactic and semantic order. Contrary to M. Schlick and others who believe that (31), like claims of a pseudoscience, is senseless, (31) does make sense to Dr Smith. Presumably, as long as a sentence does not contain any meaningless words and is not ill-formed, it makes sense to the interpreter. In fact, it is logically possible to have a context in which (31) could be used to say something true or false.
However, even if Dr Smith can make sense of (31), he is still unable to understand what is being said by it. The trouble is not that (31) involves meaningless words or combines meaningful words in an illegitimate way, but rather that it is odd to the point of being unintelligible (in the sense of failing to have a point) in contexts that are conceptually recognizable to him. For Dr Smith, there is no context, as far as he can tell, in which such a claim can be deemed true or false. He can, as M. Schlick (1991) argues, convince himself that if he asked a question of the speaker, ‘What do you actually mean by the presence of this nucleus?’ then the speaker would have to admit, ‘everything would be exactly as before’. Thus he can justifiably conclude that the speaker ‘had not succeeded in conveying to us the meaning of the hypothesis that electrons have eternally hidden nuclei’. In addition, a sentence can be used to make a point only within a suitable context. (31) could be said to have a point only when it is put forward within the context that Dr Smith is obliged to reject as conceptually unsuitable or impossible. Thus, (31) fails to make a point for him as an interpreter. Consequently, Dr Smith can neither understand what is being said nor grasp the thought expressed by the sentence.
The importance of the above observation is that although the interpreter from another language can know the meaning of each word of a sentence of an alien P-language, he or she might still not be able to effectively understand it if he or she is not aware of its point or is unable to grasp its propositional content. So declaring that one can make sense of a sentence is altogether different from declaring that one can understand what the sentence is saying.
Our second observation is more significant. We find that although both (4) and (30) are unintelligible to Dr Smith, it is important to make a distinction between the two. An old P-language may be forgotten, but can still be made intelligible to the modern reader who is willing to spend the time relearning it. In contrast, some P-language—especially when embodied within a substantially disparate intellectual or cultural tradition—indicates such a radically disparate mode of thinking and/or categorical framework as to require something far more complicated than mere learning of the language itself. In order to understand it, one has to learn the whole form of life behind it (with the mode of reasoning and the category system as its cores). Two distinct languages of phlogiston theory and CMT are good examples of this contrast.
The primary cause for Dr Smith failing to understand (30) lies in the meaningless term ‘phlogiston’. As long as he learns the meaning of the term and the corresponding existential assumption (the existence of phlogiston), he is able to understand (30). Although Dr Smith does not believe there is such a substance as phlogiston, he can work it out and understand the point of what Priestley is saying when presenting his phlogiston theory. This is mainly because Priestley’s phlogiston theory, lying within the same intellectual tradition as modern science, is conceptually recognizable to Dr Smith. After being given the meaning of ‘phlogiston’, he is able to identify and comprehend a metaphysical presupposition of the language, i.e., the existence of phlogiston. He can thus fully recognize the truth-value conditions of (30).
For contrast, turn to CMT. As we have pointed out in chapter 5, Dr Smith’s failure to understand CMT is not due to the difficulty of translation of the language of CMT into modern English or a corresponding modern scientific language. For example, his failure of understanding (4) does not just lie in the meaningless term ‘yin’. He can understand the meaning of ‘yin’ by being given a plain definition. But even if he could make sense of (4) (in the sense that it does not involve any meaningless term and the terms are not combined in any illegitimate way), he would still be left in a fog. What is the point of what is being presented or argued for by (4)? Or what is the thought that (4) expresses? To know the point of what is being said by (4), Dr Smith has to know the proper contexts in which (4) can be used to say something true or false, namely, to know its truth-value conditions. But such contexts are not conceptually recognizable to him, for he does not comprehend the conceptual framework (consisting of the pre-modern Chinese mode of reasoning, its underlying yin-yang cosmology, and the related medical category system) within which these possible contexts are constructed. To grasp the thought expressed by (4), Dr Smith needs to comprehend the pre-modern Chinese mode of reasoning and the related category system that are central to the thought and presupposed by the proposition expressed by (4).
Unfortunately, it is the pre-modern Chinese mode of reasoning and the related category system that are totally alien and scarcely comprehensible to Dr Smith. The goal of attunement to nature was so highly valued in the pre-modern Chinese culture that it became a necessary condition of understanding and rational justification. Then it is possible that what ancient Chinese thought to be rationally justifiable and perfectly intelligible is not at all rationally justifiable and intelligible in an intellectual tradition that severs the connection between understanding and attunement. This is what actually happens when Dr Smith encounters CMT. As C. Taylor points out, the world for the European intellectual tradition ceased to be a possible object of attunement after the rise of modern science. Instead, the world became alienated from human beings and became the object of investigation, experiment, and control. The original connection between understanding and attunement was severed, dismissed as mere projection onto the world order of things human beings find meaningful (Taylor, 1982). Therefore, the pre-modern Chinese mode of reasoning, which values attunement so highly, is totally alien to the modern Westerner and thus hard to understand. This explains why many substantial sentences of the language of CMT such as (4) sound so strange to Dr Smith that he cannot fully grasp them.
It seems to be clear that the understanding of CMT is, for Dr Smith, an entirely different exercise from relearning phlogiston theory. Of course, it does not mean that Dr Smith cannot learn the language of CMT. However, for Dr Smith, the language acquisition process involved in learning CMT is different from that involved in learning phlogiston theory. Using a metaphor, we can say that the former is a ‘wholesale’ learning process while the latter is a ‘retail’ learning process. CMT reflects a unique belief system and embodies a specific form of life. By studying the Chinese intellectual tradition, Dr Smith should be able to comprehend the pre-modern Chinese mode of reasoning, the underlying yin-yang cosmology, and the categorical system presupposed by the language of CMT. After such a ‘wholesale’ learning, he is able to understand effectively the theory and even talk in the pre-modern Chinese way. However, Dr Smith at best can start the pre-modern Chinese way of speaking only if he becomes alienated or dissociated from the thought and the way of speaking used in the modern Western intellectual tradition.
The essential role of M-presuppositions in cross-language understanding is observed by many others. Recall that Hacking emphasizes the essential role of a style of reasoning in understanding a scientific tradition, such as Paracelsan medical theory: ‘Understanding is learning how to reason’ (Hacking, 1982, p. 60). Until one has learned how to reason in the speaker’s way one cannot understand effectively what the speaker is saying. This is especially true for the understanding between two P-languages within two substantially disparate traditions or cultures. ‘Understanding the sufficiently strange is a matter of recognizing new possibilities for truth-or-falsehood, and of learning how to conduct other styles of reasoning that bear on those new possibilities’ (Hacking, 1982, p. 60). While Hacking emphasizes the necessary role of modes of reasoning in effective cross-language understanding, Kuhn, in a similar spirit, explores the role of a categorical framework in effective understanding. To understand effectively a scientific language, one has to familiarize oneself with its lexical structure. For instance, the Aristotelian assertions in which terms like ‘force’ and ‘void’ play an essential role are perfectly intelligible with the Aristotelian lexicon in place. But apparently similar assertions are barely intelligible to a Newtonian who is unfamiliar with the Aristotelian lexicon. This was why Kuhn became lost when he first encountered Aristotelian physics.
For A. Maclntyre, ‘a language may be so used ... that to share in its use is to presuppose one cosmology rather than another ...’ (Maclntyre, 1985, p. 7). A language reflects a way of life. Choice between two languages is to choose between two alternatives and, sometimes, incompatible sets of beliefs and ways of life. ‘Moreover each of these sets of beliefs and ways of life will have internal to it its own specific modes of rational justification in key areas and its own correspondingly specific warrants for claims to truth’ (Maclntyre, 1985, p. 8). In this way, we are unable to find application for the concepts of truth and justification that are independent of the standards of one language community or the other. Therefore, the language of a culture or tradition is fully intelligible and its purported justification is adequately understood for the interpreter from another radically different culture or tradition only if its underlying cosmology, its way of life, and its scheme of beliefs embodied in the language are fully comprehended (Maclntyre, 1985, p. 13). Similarly, J. Habermas contends that understanding a language is sharing a form of life (a lifeworld). Forms of life cannot be described fully in another incompatible language. One who does not share the form of life cannot see the ‘point’ of the language and cannot put oneself in the position of being an engaged communicator. Consequently, effective understanding cannot be achieved.
Let us collect all these insights that we have revealed so far about the role of M-presuppositions in cross-language understanding. Theoretically, ‘what a sentence of an alien language means’ can be used to refer either to the meanings of the words used in the sentence, or to the thought expressed by it. To understand what a sentence of an alien language means is not just to know the meanings of its words. A good dictionary can help us with that. But it cannot help us understand the thought expressed by the sentence. To effectively understand a sentence of an alien language is not just simply to make sense of it, but rather to grasp the thought expressed by it. To know the thought expressed by a sentence, it is necessary to know that it is assertable or that it has a point, and to know what it asserts or what its point is. As S. Cavell notes, ‘we can understand what the words mean apart from understanding why you say them; but apart from understanding the point of your saying them we cannot understand what you mean’ (1979, p. 206). If a sentence is comprehensible to the interpreter, he or she has to understand the point of what is being said, being presented, or being argued.
Whether or not a sentence of a P-language, when considered within the context of the interpreter’s language, can be used to make an assertion (has a truth-value) is language dependent. More precisely, as I have argued in chapter 9, the assertability of a sentence of a language is determined by the M-presuppositions of the language. A sentence that is apparently the same could be used to assert something or have a point within the context of one language but without a point in a rival one. This establishes the fact that in order to capture the point of what is being said by a sentence of an alien P-language, it is necessary to comprehend its M-presuppositions. There is a conceptual bridge that connects the two referents of the expression of ‘what a sentence of an alien language means’, namely, ‘the meanings of the words used by the sentence’ on the one hand and ‘the thought expressed by the sentence’ on the other. The connection is not established by universals, propositions, or rules, but rather by the M-presuppositions that make certain syntactical utterances become assertions. In conclusion,
A P-language is fully intelligible to the interpreter only if he or she can conceptually recognize and comprehend the M-presuppositions underlying the language, including the mode of reasoning with its underlying cosmology, the categorical framework, as well as other universal principles and existential assumptions.
Thus, we have so far identified a significant necessary condition of effective understanding of a P-language, i.e., the knowledge of its M-presuppositions.
It seems to be questionable for some whether one needs to share M-presuppositions of an alien language in order to conduct some ordinary linguistic acts. For example, one could order a bowl of gavagai stew from a native whether or not one shares the M-presuppositions of the native language in which ‘gavagai’ could mean rabbit, undetached rabbit parts, rabbit time slice, etc. To respond, I think we should distinguish how to effectively understand a language from how to conduct an ordinary linguistic act. If one wants to understand the native language in which ‘gavagai’ plays an essential role (as ‘phlogiston’ does in the phlogiston theory), then one needs to recognize and comprehend the specific lexical taxonomy about ‘gavagai’. In contrast, to conduct a rather ordinary linguistic act of ordering a bowl of gavagai stew from the natives, it is not necessary for one to share the lexical taxonomy. Within a concrete linguistic context, there are some other more direct ways to perform such an act. One could simply point to the stew that looks like rabbit stew and say, ‘I want this’. However, suppose that ‘gavagai’ means different things to the natives at different times (it means ‘rabbit’ in the morning, but ‘rabbit time slice’ in the afternoon). If one says ‘I want gavagai stew’ in the afternoon, the native would be quite confused.
Returning to our initial question, Dr Smith is able to make sense of, but is unable to effectively understand (4), (30), and (31) because he cannot comprehend the respective M-presuppositions underlying these sentences. After relearning the meaning of ‘phlogiston’, he can quite well understand (30). However, to fully understand (4), he has to learn the pre-modern Chinese form of life—with the pre-modern Chinese mode of reasoning, the yin-yang cosmology, and its category system as its core—together with Chinese medical theory itself.
It has been argued above that effective understanding of a sentence in an alien language should be distinguished from making sense of the sentence. But what exactly does it mean ‘to make sense of a sentence on the one hand and ‘to effectively understand’ it on the other? It seems safe to say that one understands a sentence if one knows what it means. According to this everyday manner of speaking, the phrase ‘to understand’ is an abbreviation of the phrase ‘to know the meaning of. But to make sense of a sentence seems no more than to make it meaningful. If so, the distinction between effective understanding and making sense seems to be blurred.
Suppose someone was now to say sentence (16), ‘The present king of France is bald’. No one would deny that (16) is cognitively significant. The problem is how to explain the significance of such an obviously vacuous sentence. For B. Russell, only sentences with truth-values could be significant. A sentence with a nondenoting subject like (16) has a truth-value (it is false) and thus is significant (Russell, 1905). By contrast, for P. Strawson, a sentence could be both significant and truth-valueless. For example, a vacuous sentence is truth-valueless (due to the failure of its presupposition) although it is obviously significant. Strawson thinks that the alleged connection between significance and bivalence should be severed based on his distinction between a sentence and the use of a sentence. The same sentence can be used, by different persons in different linguistic contexts, to make different assertions; different sentences can be used to make the same assertion. Significance is a semantic property of sentences while truth-value is the function of the use of sentences (or a semantic property of assertions).
To give a meaning of a sentence is to give general directions for its use in making true or false assertions. ... For to talk about the meaning of an expression or sentence is not to talk about its use on a particular occasion, but about the rules, habits, conventions governing its correct use, on all occasions, to refer or to assert. So the question of whether a sentence or expression is significant or not has nothing whatever to do with the question of whether the sentence, uttered on a particular occasion, is, on that occasion, being used to make a true-or-false assertion or not. (Strawson, 1950, p. 220; italics as original)
For example, sentence (16), uttered by someone today, is certainly significant, since every word of it is meaningful and it follows correct grammatical conventions. Nevertheless, this does not mean that any particular use of (16) has to be true or false. If it were uttered by someone today, it is truth-valueless (since one of its presuppositions, (16a), is false). But if the same sentence were uttered by someone in the reign of Louis XV, it had a truth-value. Therefore, whether a sentence is significant should be separated from whether it has a truth-value in some specific context. Specifically, according to Strawson, to say that a sentence is significant is to say that it has a possible truth-value. Or, more precisely, as long as the sentence satisfies linguistic conventions governing its correct use, it could be used in certain contexts to say something true or false.
Strawson’s above distinction tells us that meaningfulness and sensefulness are conceptually connected with the possible truth-value, instead of with the actual truth-value of a sentence. ‘Sensefulness’ may be defined based on logically possible truth-value:
A sentence S of an alien language L is senseful to the interpreter who speaks language L1 if and only if S when considered within the context of L1 has a logically possible truth-value.
Recall that, as we have clarified in chapter 9, a sentence has a logically possible truth-value if it could be used to say something true or false in certain logically possible contexts. Usually, as long as a sentence is in good semantic and syntactic order, it could be used to say something true or false in certain logically possible contexts. Most of the so-called pseudo-science assertions identified by logical positivism belong to this category. For instance, sentence (31) is in a good semantic and syntactic order such that there are some logically possible contexts in which (31) could be used to say something true or false. Hence, (31) has a sense. In contrast, the following sentences are senseless:
(38) Loves Jenny falling Bob in.
(39) Three sapdlaps sat on a bazzafrazz.
(40) The number two is red.
(38) is not a well-formed sentence. (39) contains meaningless terms. (40) is little different from the first two. Having fixed on the natural language meanings of the subject and predicate, we find that (40) is neither true nor false since its sortal presupposition, i.e., ‘A number can be colored’, is false. It is not just truth-valueless in some possible worlds, it is rather neither true nor false in any possible worlds in which we continue to use these words with their ordinary meanings. In other words, sentence (40) could not be used, in any context, to make an assertion.
Accordingly, the notion of ‘making sense of or common-sense understanding (i.e., ‘understandingc’) may be defined as follows:
The interpreter can make sense of or understandc S if and only if (a) S is senseful to him or her and (b) he or she knows the sense of S.
However, to say that a sentence has a logically possible truth-value does not mean that the truth-value is conceptually recognizable to the interpreter, since the interpreter may not be able to recognize its truth-value conditions. For example, our Dr Smith could not specify the truth-value conditions of (4). The sentence has no conceptually possible truth-value to him. To distinguish making sense from effective understanding, it is useful to distinguish meaningfulness from sensefulness based on conceptually possible truth-values defined in chapter 9:
A sentence S in an alien language L is meaningful to the interpreter who speaks language L, if and only if S when considered within the context of L1 has a conceptually possible truth-value.
Recall that S has a conceptually possible truth-value for one if and only if one can recognize and comprehend its truth-value conditions. According to Convention P, one can know the truth-value conditions of S only if one is able to comprehend its presuppositions. One can know the truth-value conditions of core sentences of a P-language only if one can know their shared absolute presuppositions. Those presuppositions are nothing but what we call the M-presuppositions of the language. As we have argued in the previous section, to comprehend the M-presuppositions of a P-language is necessary for one to effectively understand the language. Thus, we can define the notion of effective understanding as follows:
The interpreter who speaks language L1 can effectively understand a sentence S of an alien language L if and only if (a) when considered within the context of Li S is meaningful to the interpreter, and (b) the interpreter knows the literal meaning (the propositional content) of S.
The notion of effective understanding so defined suggests that the understanding of a sentence S of a language is actually a two-staged cognitive process. First, is S meaningful (or does S have a conceptually possible truth-value) to the interpreter? S has a conceptually possible truth-value to the interpreter if and only if the interpreter can recognize and comprehend the truth-value conditions of S. Second, if S is meaningful to the interpreter, then what is its meaning? According to Davidson’s truth-conditional theory of meaning, to know the truth conditions of S is sufficient to know its literal meaning. Notice that the truth conditions of concern here are not Davidson’s truth conditions of a sentence with an actual truth-value, but rather the truth conditions of a sentence with a conceptually possible truth-value. To indicate this distinction, we may call the latter the possible truth conditions of S. The interpreter knows the literal meaning of S if he or she knows its possible truth conditions.
Knowing the truth-value conditions of a sentence is a prerequisite to knowing its truth conditions. For this reason, the account presented above may be referred to as the truth-value conditional theory of understanding. According to this theory, the interpreter can effectively understand a sentence only if he or she knows its truth-value conditions. Accordingly, the interpreter is able to effectively understand a P-language only if he or she knows the truth-value conditions of its core sentences. Furthermore, as was argued earlier, the interpreter can know the truth-value conditions of the core sentences of a P-language only if he or she is able to comprehend its M-presuppositions. This brings us back to the same conclusion drawn from the previous section: It is necessary for the interpreter to be able to identify and comprehend the M-presuppositions of a P-language in order to understand the language effectively.
It might be asked what counts as to comprehension of the M-presuppositions of a P-language? Or what are the conditions for comprehending them? An M-presupposition of a P-language (i.e., ‘phlogiston exists’) is usually a statement or a set of statements that are accepted as either true or false by both sides of the communication. Based on Davidson’s truth-conditional theory of understanding, it can be comprehended by knowing its Tarskian truth conditions.
At last, we notice that some sentences of an alien language are meaningful to the interpreter only when they are put forward within certain possible contexts that the interpreter is obliged to reject as unsuitable. This is because one may be able to identify and comprehend a conceptually possible context in which a sentence says something true or false but does not regard the context as a genuine one. For instance, although the sentence (30) is meaningful for modern chemists who happen to know the existential assumption behind it, they simply deny the truth of the assumption. Similarly, we can understand the story of Snow White very well by imagining a world in which the story tells us something true or false. But we deny the actual reality of such an imaginable world. In this case, sentences can have conceptually possible truth-values (be meaningful for the interpreter), but lack actual truth-values. We say that those sentences are not factually meaningful to the interpreter. By definition,
A sentence S of an alien language is factually meaningful to the interpreter who speaks language L if and only if S when considered within the context of L has an actual truth-value.
According to the notion of actual truth-value defined in chapter 9, a sentence of an alien language has an actual truth-value for the interpreter if and only if the interpreter holds its underlying presuppositions to be true. That means that S is factually meaningful to the interpreter if and only if S’s presuppositions are held to be true. For the advocate of phlogiston theory, (30) is factually meaningful, no matter whether it is actually true or not, since the existential presupposition of (30) is held to be true by the advocate. When the interpreter regards the core sentences of a language as factually meaningful, he or she commits him or herself to the truth of its M-presuppositions.
We have concluded that a P-language is fully intelligible to the interpreter only if he or she can conceptually recognize and comprehend the M-presuppositions underlying the language. The interpreter who fails to grasp the M-presuppositions of the language could not effectively understand it. However, M-presuppositions, such as the mode of reasoning or the lexical taxonomy of a P-language, are not easily identifiable. This poses a problem in knowing whether the interpreter is able to identify and comprehend the M-presuppositions of a P-language. It would be helpful if we could locate a clearly identifiable semantic indicator for the failure of cross-language understanding.
As observed earlier, for the Western physician, due to the inability to identify and comprehend the underlying pre-modern Chinese mode of reasoning and its associated yin-yang cosmology, the core sentences of traditional Chinese medical theory are simply not candidates for truth or falsity. In other words, to the Western physician, they do not possess conceptually possible truth-values. Similarly, Hacking has noticed a strong linguistic correlate of our failure in understanding Paracelsus. Numerous Paracelsan sentences, when considered within the context of modern scientific theories, do not have any conceptually possible truth-values, because we cannot comprehend the Renaissance mode of reasoning underlying the language. Kuhn concurs. When the modern interpreters find Aristotelian sentences hard to understand, the trouble is not that they think Aristotle wrote falsely, but that they cannot attach truth or falsity to many Aristotelian core sentences since the Aristotelian lexical taxonomy presupposed by the sentences is totally alien to them.
Our case study of the Newton-Leibniz debate on the absoluteness of space (chapter 5) tells us a different story. We found that the confrontation between the Newtonian and the Leibnizian languages of space is different from the confrontation between Chinese and Western medical languages. Not like the Western physician who cannot identify and comprehend the M-presuppositions of the Chinese medical theory, the Leibnizian is able to identify and understand the underlying M-presupposition of the Newtonian language, namely, the existence of Newtonian absolute space. Hence, for the Leibnizian, the set of sentences SN-SL has conceptually possible truth-values (i.e., is meaningful), and thus the Newtonian language is perfectly intelligible. Nevertheless, the Leibnizian categorically rejects the assumption of absolute space, which makes the sentences in SN-SL actually truth-valueless (lack of actual truth-values). Therefore, it is the lack of conceptually possible truth-values, not the lack of actual truth-values, that is semantically correlated with the failure of cross-language understanding.
In fact, we could derive such a semantic indicator from our truth-value conditional account of understanding. When the interpreter fails to comprehend the underlying M-presuppositions of an alien P-language, based on Convention P, the core sentences of the language would be rendered conceptually truth-valueless to the interpreter since he or she cannot identify and comprehend the truth-value conditions of those sentences. Lack of conceptually possible truth-values makes those sentences meaningless to the interpreter, and results in his or her failure of effective understanding of the alien language. We can hence use the lack of conceptually possible (not actual) truth-values as a strong semantic indicator of the failure of effective understanding on the interpreter’s part. In general, if the speaker of a P-language PL1 is unable to recognize and comprehend the M-presuppositions of an alien P-language PL2, then the core sentences of PL2 when considered within the context of PL1 will lack conceptually possible truth-values. Suppose the speaker of PL2 faces the same situation when encountering PL1. Then a conceptually possible truth-value gap occurs between PL1 and PL2, which signifies semantically the failure of mutual cross-language understanding between the two language communities.
It has been argued that the interpreter can effectively understand a P-language only if he or she is able to identify and comprehend its M-presuppositions. M-presuppositions are language dependent. Hence, effective understanding is language dependent. Just as it is not useful to ask whether a sentence itself is true or false, but only whether its specific use within a linguistic context is true or false, so it is not useful to ask whether or not a sentence in isolation is meaningful. We can only ask whether, when considered within the context of a specific language, it is meaningful, and what its meaning is. Therefore, the core sentences of a P-language that are meaningful and can be understood in the contexts of its own or some other compatible P-languages (languages with compatible M-presuppositions) might not be fully understood when considered within the context of some incompatible P-languages (languages with incompatible M-presuppositions).
We need to emphasize that the essential role of contexts in effective understanding is not that the same sentence is meaningful in one context, but meaningless in the other. Rather, it is the difference between using a sentence to say something true or false in one context when the same sentence says nothing at all in the other context. It is not a contrast between affirmation and counter-affirmation, but a contrast between affirmation and silence. When a sentence is considered within an unsuitable context, it could sound unintelligible to the point where what is being said by the sentence totally eludes the interpreter.
The language-dependent feature of effective understanding has a significant impact on cross-language understanding. Presumably, the degrees of difficulty and the types of understanding involved when the interpreter encounters an alien P-language depend upon the relation between the alien language and the interpreter’s own language, especially the familiarity of the interpreter to the M-presuppositions of the alien language. First, there will be cases in which the M-presuppositions of the alien P-language are compatible with those of the interpreter’s own language, which we will call normal discourse. Within normal discourse, we may make a further rough-and-ready distinction based on whether the interpreter is familiar with the alien language. When the interpreter is familiar with the language, understanding happens without a glitch. When the language is initially unfamiliar to the interpreter but either lies in the recent past of the interpreter’s own cultural or research tradition or comes from a closely related cultural or research tradition, the interpreter could still have a measure of tacit familiarity with its M-presuppositions by virtue of his or her own historical, cultural, or research tradition. Thus, the interpreter can still comprehend the M-presuppositions of the alien language based on his or her historically formed tacit ‘pre-understanding’. The understanding of the language could be readily obtained.
Second, by contrast, there will be cases in which the interpreter encounters an alien P-language whose M-presuppositions are incompatible with those of his or her own language, which we will call abnormal discourse. Even within an abnormal discourse in which the M-presuppositions of an alien language are substantially distinct from those of the interpreter’s own, the interpreter could still be familiar with the language if the interpreter can gain access to it through interpretation already essayed by historians or anthropologists (such as the language of Aristotelian physics) or if two languages coexist within a broad intellectual tradition (such as the Newtonian language and the Leibnizian language of space). However, the real challenge arises when the interpreter faces an alien P-language with which he or she is not familiar and has not even a tacit familiarity because it belongs to a cultural, research tradition other than his or her own (such as traditional Chinese medical theory for a Westerner), or because it exists in the distant past and lacks a continuous history of interpretation and appreciation within the interpreter’s own cultural tradition (such as the Paracelsan medical language for a contemporary Westerner).
Obviously, normal discourse scarcely concerns us here; our central concern in explication of incommensurability is with abnormal discourse. When two disparate P-language communities confront one another, each with its own body of M-presuppositions, but lacking a knowledge of each other, the speaker of one P-language often falls into the temptation of approaching the other unknown P-language by imposing, reading into, or projecting the categories, beliefs, or the mode of reasoning embodied in the speaker’s own language upon the other. Each community will usually represent the beliefs of the other within its own tradition, in abstraction from the relevant tradition of the other. This was the strategy used by Kuhn in that hot summer of 1947 when he attempted to ‘approach Aristotle’s texts with the Newtonian mechanics’ (Kuhn, 1987, p. 9). This is a phenomenon frequently encountered by historians or anthropologists. For lack of an alternative historians or anthropologists are tempted to understand an old or alien text as they would if it had occurred in either contemporary discourse or in their own culture or tradition. As we have discussed, Davidson defends such a projective understanding by presenting his version of the principle of charity, which requires, in interpretation, to put the speaker in general agreement with the interpreter according to the interpreter’s view. The principle is obviously powerful and could at least be utilized to jump-start the interpretation process. However, the principle is too ‘powerful’ to the extent that it is vulnerable to being abused for ethnocentric ends. The principle tells us that to understand others, we have to render them so that as much as possible they speak the truth, make valid inferences, and so forth according to our notion of truth, truth conditions, and logical rules. We have to make them think like us, attribute our own beliefs to them, and interpret their utterances within the system of our own language.
There is a hidden assumption behind the above projective approach to understanding: Others are basically like us, sharing the same linguistic conventions, belief systems, and, most importantly, M-presuppositions. This assumption is a manifestation of absolutism in cross-language understanding, which is based on a basic conviction that there is or must be some permanent, ahistorical, culture-transcendent matrix or framework (R. Rorty) to which one can ultimately appeal in determining the nature of rationality, intelligibility, truth, reality, and morality. To make the discourse of others intelligible and rational one needs to be able to find some area of agreement. Within the analytic tradition, the desired agreement has often been imagined to lie in some common language. Specifically, in the discussion of cross-language understanding the agreement manifests itself as shared or compatible M-presuppositions. Absolutists believe that there is or should be a common or third language between any two languages if understanding between them is possible. Such a third language would have two central features: (a) It does not presuppose allegiance to either of the two sets of M-presuppositions associated respectively with the two languages in question, nor presuppose any other set of M-presuppositions that might compete for allegiance with those two; (b) It must be able to provide semantic and ontological resources for an accurate representation of the two languages, including the M-presuppositions embodied in each language.
The projective approach to understanding is justifiable only when the above absolutist assumption is sound. In normal discourse in which the M-presuppositions of an alien P-language are compatible with that of the interpreter’s own language, the projective approach can proceed without much difficulty. The interpreter is able to understand the other language since he or she can recognize its M-presuppositions by way of analogy or by simply taking his or her own as the other’s. Normal discourse is supposed to be contrasted with abnormal discourse in which both sides speak P-languages with incompatible M-presuppositions. Roughly, normal discourse proceeds at the stage of Kuhnian normal science while abnormal discourse happens during Kuhnian scientific revolution.3 However, the projective approach would ensure the failure of cross-language understanding in abnormal discourse. Projecting the M-presuppositions of the interpreter’s own language upon an alien P-language would suspend or distort the M-presuppositions of the latter. Suspending the alien P-language’s M-presuppositions would make its core sentences conceptually truth-valueless and hence suspend all empirical contents of meaningful statements of the language. By distorting them, the interpreter puts the original meaningful statements out of their appropriate contexts and hence causes them to lose their original meanings. Either way prevents recognition and comprehension of the M-presuppositions of the alien language. Lack of knowledge of the M-presuppositions of an alien P-language is sufficient to preclude effective understanding of the language. From each point of view, the key concepts and core statements of the other—just because they are presented apart from the linguistic context constituted by its own M-presuppositions from which they draw their conceptual life—will necessarily appear without context, lack justification, and hence become meaningless and unintelligible. Therefore, in abnormal discourse, the projective approach to understanding is doomed to failure. This is the real source of the failure of cross-language understanding that we have often experienced between two conceptually disparate languages.
When the speakers of two P-languages cannot understand one another, no communication acts can be carried out. The communication between the two P-language communities breaks down completely, as is often experienced by the Chinese medical community and the Western medical community. I call this kind of communication breakdown between two P-language communities caused by lack of effective understanding complete communication breakdown.
1 Habermas, 1992, p. 57; 1987, p. 62.
2 In the following discussion, I will treat the literal and the propositional content of a declarative sentence and the (objective) thought expressed by it roughly as the same conception and ignore the subtle distinction between them.
3 The terms ‘normal discourse’ and ‘abnormal discourse’ are borrowed from R. Rorty (1979, Ch. 7), but I use them in a different way here.