Revisiting Xunzi’s Philosophy of Language

Chris Fraser’s essay “Language and Logic in the Xunzi,” which was published in the Dao Companion to the Philosophy of Xunzi (2016) that I edited, gives an excellent introduction and overview of the views about language that appear in the Xunzi, with a focus on chapter 22 of the text, “Zhengming” (正名, “Correct Naming”).¹ Even so, I disagree with some of the claims that Fraser makes, and so on this occasion I want to defend a different interpretation of the Xunzi. At the end, I will suggest that we should understand the views presented in the “Zhengming” chapter in a more restricted way than they are commonly understood.

In particular, one of Fraser’s more striking claims about Xunzi’s view is that “His theory of names [i.e., words or terms] is purely extensional” (Fraser 2016, 315).² Likewise, in a footnote, Fraser states that Xunzi “shows no awareness of a theoretical notion akin to the meaning or intension of a word” (Fraser 2016, 302n21). It is these claims that I would like to challenge here, and then use that challenge to reorient how we see the “Zhengming” chapter. After first presenting some background information and discussing some methodological considerations, I will review what I take to be Fraser’s reasons for his position, and then proceed to my argument against him.³

For readers who may not be familiar with the technical notions of “extension” and “intension” as used in recent philosophy of language, it will perhaps be helpful to begin with a brief explanation of the distinction between the two, so that they may better appreciate what is at stake in Fraser’s thesis, and why his claim is potentially controversial. To put it simply (a little too simply), the “extension” of a term is that set of objects to which that term may be correctly applied, and the “intension” of a term is that set of attributes that all and only those objects possess and that serves as the criterion that is used explicitly or implicitly by speakers of the language for applying the term to them, and that is conveyed in such usage.⁴ So, for example, the extension of the term “bridge” will include the Golden Gate Bridge, the Brooklyn Bridge, the Tower Bridge, the Ponte Vecchio, the Jade Belt Bridge, the Gwangandaegyo, and so on. The intension of “bridge”—or more precisely, one intension of the term—would be a “structure forming or carrying a road, path, or (in later use) a railway, and others, which spans a body of water, a roadway, a valley, or some other obstacle or gap, and allows a person or vehicle to pass unimpeded over or across it” (Oxford English Dictionary Online).⁵ Each of those bridges I just listed is such a structure, and because of that feature, each may be called a “bridge.” In turn, when I say of some X that “It is a bridge,” its being such a structure is the information conveyed about X. It is common for the intension of a term to be called its “meaning,” and for the extension of a term to be called its “referent(s),” and Fraser himself follows this practice as well.

Two further points about intension and extension are worth noting, because they will be relevant for the discussion to follow. First, a term’s intension is standardly taken to determine its extension, rather than extension determining intension, because (among other reasons) for any given set of objects, there are potentially many attributes that they might share in common (e.g., is visible at time t, is located on the planet Earth, is younger than the Sun, and so forth). So, from the mere fact that a term is applied to some set of objects, one cannot automatically infer which shared attributes, if any, are the basis for applying it to them, and hence derive the term’s intension. On the other hand, given a criterion provided via an intension for applying a term or withholding it, one can, at least in principle, identify what falls within the term’s extension.

Second, while many terms, especially what are called “general terms” or “class terms” (such as “bridge”), will have both an intension and extension, not all terms will have both. On the one hand, a term such as “unicorn” may have the intension “a magical animal with the body of horse and single horn growing on its head” but lack an extension, since there is nothing in the world that actually fits that description.⁶ On the other hand, one could simply make up a term and use it to refer to a set of objects without thereby committing to the identification of some attribute(s) shared among them that is conveyed in the use of the term and governs its application by those who use it, in which case the term would have an extension without an intension. For example, I might decide to call a group of files on a computer server that I share with others the “2az7q43bbv” files, and I may have grouped them together because they are similar in some way, but the term “2az7q43bbv” itself need not presuppose any particular identification of those similarities, such that it conveys mere membership in that group (and not any shared features that may form the basis for membership in the group), and similarly its use is governed only by grasping the members of that group (and not necessarily the basis for membership in the group). In such a case, we can say that “2az7q43bbv” is meaningless (it has no intension) but still has a set of referents.⁷ For the sake of convenience, let us call cases of terms with extensions but no intensions “mere labels.”

With these points in mind, the controversial character of Fraser’s view may now become clearer. In particular, his claim that Xunzi’s understanding of terms is “purely extensional” is tantamount to saying that Xunzi thinks of all or nearly all terms as being mere labels or at least highly akin to mere labels, which is a rather odd view on the face of it. Moreover, many of the terms that Xunzi seems to want to cover with his discussions in chapter 22 are the sorts of “general” or “class” terms that would typically be analyzed as having both an intension and extension. Therefore, Fraser’s interpretation would present Xunzi as strangely blind to an aspect of such terms that many thinkers have considered to be extremely important, a strangeness that can be brought out by saying that on Fraser’s reading (as he himself says more or less) Xunzi is highly concerned with terms—but pays no attention to their meanings. Furthermore, many or perhaps even most philosophers would hold that a purely extensional understanding of terms simply cannot provide an adequate account of language, and from that perspective Xunzi’s view will then seem not just to be wrong, but to be wildly wrong in a way that seems rather out of character with the keenly observant mind that one sees in so much else of the Xunzi. For such reasons, many scholars are likely to want to resist Fraser’s interpretation.

If one is going to dispute his reading, though, there are certain methodological considerations that should be noted. As a preliminary point, I readily grant that we find in the Xunzi no explicit discussion that distinguishes intensions from extensions of words. However, just because Xunzi himself does not explicitly draw such a distinction does not, by itself, preclude him from having a notion of intension implicitly. This point is a variation on a problem that has been raised in many other (Western) discussions of Chinese thought (and cross-cultural studies more broadly): someone will object to a given analysis of some Chinese text or thinker on the basis that the analysis relies on some notion for which there is no equivalent word in Chinese. It is a fallacy, though, to think that lacking a word for something amounts to lacking the concept of that thing, as Bryan Van Norden has argued succinctly (Van Norden 2007, 21-23).⁸

Of course, the absence of any explicit discussion of intension in the Xunzi does pose an interpretive hurdle for anyone who wants to maintain that Xunzi does have a notion of intension. Namely, one still has to provide positive evidence that such a notion is at work in Xunzi’s thought, even if only implicitly, and that one needs to appeal to such a notion in order to adequately explain what one sees in the text. Otherwise, even if attributing a notion of intension to Xunzi were compatible with the textual evidence, there would be no strong reason to do so—such an attribution would be interpretively superfluous and might even generate unnecessary problems for understanding Xunzi’s thought. This last point is, I think, the most charitable way to understand Fraser’s objection to attributing a notion of intension to Xunzi: such an attribution is unsupported by the text, and even if it can be made to fit, it is neither necessary nor helpful for understanding Xunzi’s views of language.

Before reviewing the evidence Fraser offers for his “purely extensional” understanding of Xunzi’s view, let me note one further methodological complication. Fraser’s claims are about Xunzi’s theory of language, and not about Xunzi’s own use of language.⁹ For this reason, even if it turns out that Xunzi at times seems to speak in ways that presume a notion of intension, that may show no more than that Xunzi’s own linguistic practice is simply inconsistent with his theory of language (if Fraser is right that Xunzi has a purely extensional theory of terms). Indeed, if—as mentioned earlier—a purely extensional theory of terms simply cannot in fact be an adequate theory of language, then actually we should expect Xunzi’s own practice to be inconsistent with his theory, because his theory would be flawed to begin with. Of course, it would be nice if Xunzi’s own linguistic practice turned out to be completely compatible with his theory of language, but there is no antecedent reason to suppose that such must be the case, and the principle of charity (at least as I would construe it) does not, in general, require us to favor those interpretations that render his theory of language consistent with his own linguistic practices.¹⁰ Hence, in order to challenge Fraser’s claims, one needs to be careful and focus on evidence from the text that relates to Xunzi’s theory of language in particular, as opposed to evidence that pertains just to Xunzi’s use of language.

Turning now to Fraser’s reasons for adopting the interpretation he does, I discern in his essay two main arguments for a “purely extensional” construal of Xunzi’s understanding of terms: one explicit and one implicit. The explicit argument is based on Xunzi’s remarks about the role of perception in naming. After quoting and expounding a section of chapter 22 that discusses perception, Fraser writes:

A signal facet of this compact theory of perception . . . is that it depicts the sense organs as directly differentiating their objects, rather than producing mental representations of them that the heart or mind distinguishes. . . . Xunzi’s account of perception—like that of the Mohist “Dialectics”—ascribes no role at all to mental contents such as sense data, mental images, or ideas. The absence of such semantic or epistemic intermediaries is one likely reason that the distinctions between appearance and reality or between phenomena and noumena play no role in classical Chinese thought and why classical Chinese thinkers were not troubled by sense skepticism. Also significant is that in tying the right use of names directly to sense discrimination, Xunzi makes no appeal to intensional concepts. The use of words is explained completely by appeal to speakers’ ability to distinguish their extensions according to shared norms.¹¹ Indeed, nothing in Xunzi’s theory corresponds to the notion of the meaning or intension of a word, although, as we will see, his treatment of ci 辭 (“expressions,” “phrasings”) introduces a notion similar to speaker’s meaning. (Fraser 2016, 301-302)

I have reservations about Fraser’s claims concerning Xunzi’s views of perception here but pursuing them at length would require a detailed analysis of the passages on perception in the Xunzi that would take us far afield from the questions about language that I wish to pursue. So for now, it will suffice merely to note the basic structure of Fraser’s argument: if words are generated by naming objects that are differentiated through perception, and if that differentiation of objects occurs without the need for “sense data, mental images, or ideas,” then intension plays no role here (presumably because a notion of intension requires “semantic or epistemic intermediaries” of such a sort), and instead the only function of words is to designate the set of things that constitutes their extensions.¹²

Fraser’s other main argument for his position, which is mostly implicit, seems to rest on how he construes some of Xunzi’s terminology in chapter 22. To see this implicit argument, we should look first at the following remarks from Fraser:

[On Xunzi’s view] The main threat to . . . orderly regulation of language is miscreants who engage in the “great depravity” of “splitting phrases and recklessly inventing names in order to disrupt right names” (HKCS 22/108/4-5, 144). Xunzi here refers to those who, like Hui Shi 惠施, Deng Xi 鄧析, and Gongsun Long 公孫龍, are known for confusing, paradoxical sayings, but also those who advocate ethical or psychological theses he rejects, such as Song Xing 宋鈃. He is in effect claiming that a major factor explaining the mistaken doctrines of his philosophical opponents—whether frivolous, such as Gongsun Long’s logic-chopping claim that a white horse is not a horse, or sincere, such as Song Xing’s pacifist doctrine that a person can be insulted without thereby being disgraced—is that they muddle the proper referents of words. (Notice that Xunzi does not say they garble the meanings of words. Like the Mohists’, his theory does not explicitly treat the meaning or intension of terms, but their reference or extension.) (Fraser 2016, 293-94)

Here, with his list of those Xunzi is criticizing, Fraser is alluding to the refutation of unacceptable sayings that occupies a substantial portion of chapter 22. What is significant, though, is that there Xunzi’s complaint about those sayings and the people who make them is that it is framed in terms of the ways that those sayings confuse people about ming 名 (names, words) and their shi 實 (objects, things, substances). On Fraser’s description, this complaint is equated to the idea that such sayings “muddle the proper referents of words,” which indicates that Fraser is taking the shi to designate only referents and construing the relationship between ming and shi in a purely extensional fashion.

Why does Fraser understand Xunzi this way? Again, though he does not provide an explicit defense of that approach, a hint can be gleaned from his paraphrase of another passage from that section of chapter 22, where he reviews Xunzi’s account of “the purpose of having names.” Fraser says:

[Xunzi] holds that the wise regulate the names used to refer to things, so as “to clarify noble and lowly” and “to distinguish similar from different,” such that intentions can be conveyed smoothly and tasks accomplished effectively (HKCS 22/108/12-14, W 145-46). (Fraser 2016, 295)

The first clause of Fraser’s sentence here corresponds to a clause in the original that reads “知者為之分別制名以指實,” and which in my translation of the Xunzi is rendered as “the wise person draws differences and establishes names in order to point out their corresponding objects” (Hutton 2014, 237). Fraser’s wording “the names used to refer to things” strongly suggests that he takes the word zhi 指 (lit. “point out,” “point to”) to designate the act of referring. Since in the relevant passage Xunzi is explaining the purpose and process of naming, we can thus reconstruct Fraser’s thinking as follows: Xunzi says that the purpose of words is to zhi 指 some shi 實, but if the linguistic action zhi is the act of referring, then the shi that are “pointed out” by the words must be the referents of those words, and so Xunzi’s account of words can be understood in an entirely extensional manner.

Based on my reconstruction so far, readers may have already identified various places in Fraser’s account that could be targeted to undermine his claims. I will not attempt a thorough attack on them here, as that would require more space than is available on this occasion, but I do want to point out briefly that the textual evidence on which he bases his claims may not provide conclusive support for his view after all. In particular, recall his claim that “in tying the right use of names directly to sense discrimination, Xunzi makes no appeal to intensional concepts” (Fraser 2016, 302). Fraser derives this claim from the following piece of chapter 22:

So then on what grounds do we deem things similar or different? I say: On the grounds of the sense organs. As to any creatures of the same kind, with the same affects, how their sense organs detect things is similar. So they converge in how they model things as resembling each other. This is the means of reaching consensus on conventional names by which to indicate things to each other. (HKCS 22/108/14-16, Fraser’s own translation from Fraser 2016, 300)¹³

The passage talks about sense discrimination in the first four sentences, and then offers the remark that “This is the means” (shi suo yi 是所以) to name things appropriately. In that respect, the passage does transition directly from sense discrimination to naming, but that may not quite amount to “tying the right use of names directly to sense discrimination.” The reason is that the expression “this is the means” need not always function as an exhaustive account of how something is or should be done, but can rather serve to point out the primary means by which something is or should be done, while omitting subsidiary means. For comparison, consider the following passage from chapter 4 of the Xunzi:

Being a good son and younger brother, conscientious and honest, restrained and hardworking, carrying out one’s tasks without daring to be lazy or arrogant—these are the means by which the common people (shi shu ren zhi suoyi 是庶人之所以) obtain warm clothing, plentiful food, and long life, avoiding punishment and execution. (HKCS 4/14/12-13; Hutton 2014, 26)¹⁴

Note that “the means” here are said to provide “warm clothing” and “plentiful food,” but the list of behaviors that are identified as “the means” to achieve these benefits do not actually include farming, textile production, and sewing. Nor should we assume that such activities just are the “tasks” (shiye 事業) mentioned here, since otherwise the claim made in the passage would not apply to craftsmen who do not grow food or make clothes, but who are most likely meant to be covered by this claim as well. In that case, the passage moves directly from being a good person and a hard worker (in general) to having warm clothing and plentiful food, but—on pain of saddling Xunzi with an absurd view otherwise—that does not entail that warm clothing and plentiful food are obtained directly from being a good person and a hard worker without need for farming, textile production, and sewing (or making money from other work and using it to buy those items). By the same token, even if the passage from chapter 22 moves directly from sense discrimination to naming, without mention of intensional concepts, that need not entail that Xunzi thinks one can dispense with intensions altogether—they might be part of the process that is simply elided in that passage, just as farming, textile production, and sewing are elided in the chapter 4 passage.¹⁵ In a similar vein, even if one were to grant (as my reconstruction above of Fraser’s second argument has it) that the purpose of terms is identified by Xunzi solely in terms of referring to things, that would not preclude the possibility that intensions are nonetheless still part of the mechanism by which terms accomplish that goal.

If the argument I have just given is right, then Fraser’s interpretation of Xunzi as having a purely extensional theory of terms may not be necessitated by the textual evidence we have, but it might still be the best available interpretation, and Fraser’s objection could still stand: attributing to Xunzi an implicit notion of intension is unsupported by the text, and even if it can be made to fit, it is neither necessary nor helpful for understanding Xunzi’s views of language. So at this point, let me now review some other textual evidence in order to provide grounds for being dissatisfied with Fraser’s interpretation and seeking an alternative. In particular, consider the following passage from chapter 3 of the Xunzi, which plays on different senses of the word zhi 治 (“put in good order,” “orderly”):

The gentleman puts in good order what is orderly. He does not put in good order what is chaotic. What does this mean? I say: Ritual and yi are called orderly. What is not ritual and yi is called chaotic. So, the gentleman is one who puts in good order the practice of ritual and yi. He does not put in good order what is not ritual and yi. That being so, if the state is in chaos will the gentleman not put it in good order? I say: Putting a chaotic state in good order does not mean making use of the chaos to put it in order. One gets rid of the chaos and replaces it with order. Bringing cultivation to a corrupt person does not mean making use of his corruption in order to cultivate him. One gets rid of the corruption and supplants it with cultivation. So, the gentleman gets rid of chaos and does not put chaos in good order. He gets rid of corruption and does not cultivate corruption. The way that “put in good order” functions as a term is like when one says that the gentleman “does what is orderly and does not do what is chaotic, does what is cultivated and does not do what is corrupt.” (HKCS 3/10/12-16; translation modified from Hutton 2014, 19)¹⁶

For present purposes, the last line of this passage is the most important. The sentence begins with the rather unusual phrase zhizhiweiming 治之為名 (more literally, “the way 治 is a name”). This phrasing serves to mark out that zhi 治 is being discussed as a word, rather than as a phenomenon. (Or in other words, this is a “mention” rather than a “use” of zhi 治, and the addition of zhiweiming 之為名 after zhi 治 is needed to make this clear, because ancient Chinese did not have quotation marks that serve this purpose in modern languages, as in my translation above.) By the same token, though, the phrase implies that what follows it is said from a reflective standpoint about language, the kind of standpoint that also serves as the basis for the discussion of naming in chapter 22. It is significant, then, that the remainder of the sentence clarifies “the way 治 is a name” by pointing to its use in another sentence. Such an approach does not fit with a purely extensional theory of names, because on a purely extensional theory, the only salient fact about names is the objects to which they can be applied, and so appealing to the way a name is used in a sentence, which is a matter of its relation to other words, is to focus on an irrelevant feature.¹⁷ Rather, if the way zhi 治 is used in a sentence provides significant information for understanding that word, that must be because the word has some other aspect—an intension—that is revealed by that use.¹⁸ To put the point another way, the Xunzi passage is giving roughly what Richard Robinson (1950, 106-108) has labeled an “implicative” or “contextual” definition, where a term’s meaning is gleaned from how it appears in another sentence, but such a process both presupposes and specifies intensions of words. This instance in the Xunzi is also stipulative in nature, in that here the text is not merely reporting how the word zhi 治 is commonly understood, but also demanding that it be understood (and then used) in a particular way. To that extent, the passage is, additionally, an exercise in “correct naming” or “the rectification of names” (zhengming 正名) that is the focus of chapter 22, and thus provides a data point that may be relevantly compared with the discussion there.

In sum, if the argument I have just offered is correct, then the chapter 3 passage presents a relatively clear instance where Xunzi, when thinking reflectively about language, relies on the intensional aspect of a word, though he has no explicit label for it. That suggests, contrary to Fraser’s view, that Xunzi’s theory of terms is not purely extensional.

Further motivation for dissatisfaction with Fraser’s interpretation can be found when we examine a passage from chapter 23, the “Xing’e” chapter, but it will take some effort to make it clear. In particular, there we find the following criticism of Mengzi:

Mengzi says: “People’s learning is [an instance of] their nature being good.” I say: This is not so. This is a case of not attaining knowledge of people’s nature and of not inspecting clearly the division between people’s nature and their deliberate efforts. In every case, the nature of a thing is the accomplishment of Heaven; it cannot be learned, it cannot be worked at. Ritual and yi are what the sage produces. They are things that people become capable of through learning, things that are achieved through working at them. Those things in people which cannot be learned and cannot be worked at are called their “nature.” Those things in people which they become capable of through learning and which they achieve through working at them are called their “deliberate efforts.” This is the division between nature and deliberate effort. (HKCS 23/113/16-19; translation modified from Hutton 2014, 249)¹⁹

The basic point Xunzi is making is clear enough: namely, Mengzi’s claim is incorrect, because it treats as an instance of human nature something that is not, and hence it not only misconstrues the category of human nature, but also misuses the word “nature,” as the last part of the passage implies.

However, this argument also has some rather curious features that, as far as I am aware, have not received attention from scholars. First, the criticism begins with the remark that Mengzi’s claim “is a case of not attaining knowledge of people’s nature and of not inspecting clearly the division between people’s nature and their deliberate efforts.” The latter problem is clearly addressed at the end of the passage, where Xunzi distinguishes the different sets of things that are labeled as “nature” or “deliberate efforts,” and the final statement “This is the division between nature and deliberate effort” is simultaneously about the difference between those sets and the terms that refer to them. Yet, insofar as the start of Xunzi’s criticism mentions “not attaining knowledge of people’s nature” and ostensibly treats that as a distinct point from “not inspecting clearly the division between people’s nature and their deliberate efforts,” then it implies that there is an additional element to the argument here that is not simply identical to what occurs at the very end of the passage.

Indeed, it is not too hard to see that something else is going on in the argument here, which makes for a second curious feature of the passage. Namely, near the start of the passage, Xunzi says, “In every case, the nature of a thing is the accomplishment of Heaven; it cannot be learned, it cannot be worked at,” and then near the end he says, “Those things in people which cannot be learned and cannot be worked at are called their ‘nature.’” The closeness in wording of these two sentences makes them look like they might be simply reiterating the same point, but in a slightly different order, and if the former is supposed to be support for the latter, then it may even seem as if Xunzi is arguing in a circle. So, how are we supposed to understand what is going on here?

As a solution to this puzzle, I want to suggest that Xunzi is not simply saying the same thing twice and is not arguing in a circle. Rather, we should see the former statement as elucidating the complaint that Mengzi does not understand people’s nature. In particular, Xunzi is starting from a commonplace in early Chinese thought, and which even Mengzi would have accepted, namely that a thing’s nature (xing 性) is a product of Heaven. Xunzi then articulates what he takes to be a clear implication of the idea that the xing is produced by Heaven, which is that it does not come about through any human efforts, and hence “it cannot be learned, it cannot be worked at.” It is that implication, I gather, that Xunzi thinks Mengzi fails to grasp. When the passage then goes on to say that “Those things in people which cannot be learned and cannot be worked at are called their ‘nature,’” it is noting that in light of this implication, the word “nature” is to be used as a label for only those things in people that cannot be learned and cannot be worked at.

As an argument against Mengzi, this line of attack may not be very persuasive, but that is not the important point here. Rather, my reason for focusing on this passage has to do with the transition that occurs in it. By the end, in the discussion of the different items to which the labels “nature” and “deliberate effort” are to be applied, the topic is clearly about (at least) the extensions of those terms, but then the contrast with the beginning part of the passage and how Xunzi describes it suggests that the starting point of the argument is not (simply) about extension. If I am right about the way Xunzi is using an inference from a commonly held belief about xing to articulate the extension of the word “xing,” then the process mirrors that by which an intension determines an extension. Or in other words, the idea that xing is a product of Heaven is a conceptual point about xing that limits the use of the word “xing,” and hence even in the absence of an explicit term for an intension, that idea seems to be playing intensionlike role. (It is hard to see how the argument could be explained on a purely extensional view of terms, since on such a view, terms have no meaning that could support inferences about their extensions. For example, if “gaahhk” is a simply a label for the extension constituted by some set of things X, then in the absence of any intensional content for “gaahhk,” we cannot make any inferences based on the notion of “gaahhk.” We could still make inferences based on the extension of “gaahhk,” such as if all members of set X have some characteristic Y, and characteristic Y is incompatible with some other characteristic Z, then would could infer that “Nothing that is gaahhk has characteristic Z.” However, that presupposes that the extension of “gaahhk” as a term has already been fixed at the outset, and the characteristic Y plays no role in determining its extension, but such is not the way the argument about xing unfolds here.)

Apart from the way that this passage seems to involve a notion of intension, my reason for focusing on it is that it presents a highly relevant example for comparison with the discussion of chapter 22. For, like the passage from chapter 3 that we considered earlier, Xunzi here, too, seems to be engaged in an exercise of rectifying or correcting names. Moreover, unlike the example from chapter 3—but like chapter 22—Xunzi’s discussion of language in this passage from chapter 23 is in the service of repudiating a rival philosophical position.

Now insofar as these passages from outside chapter 22 seem to show that at least some of Xunzi’s discussions of language implicitly invoke a notion of intension, then comparing them with the views presented in chapter 22 raises at least two possibilities. One is that we should perhaps revise our view of the chapter 22 material and allow that even there, Xunzi may implicitly incorporate a notion of intension. Another possibility is that Fraser is correct to understand the view in chapter 22 in purely extensional terms, but Xunzi is simply inconsistent in his views of language, and hence we see a different view in other chapters—and note that here the inconsistency would not be the sort of inconsistency whose evidential relevance I warned about at the beginning of this paper, namely an inconsistency between Xunzi’s use of language and his theory of it. Rather, the inconsistency would be in his understanding of language itself, as revealed by those moments when he is explicitly addressing the proper usage of terms. If that were the right way to view the situation, then the inconsistency might be explained in any of the usual ways that others have attempted to explain inconsistencies in the Xunzi, such as suggesting the Xunzi was simply careless about these matters, or that he changed his mind over time, or perhaps even that the inconsistent passages were written by different people.

Of the two possibilities just mentioned, I am inclined toward the former, since I think that out of charity, we ought to avoid accusing Xunzi of inconsistency unless there is overwhelming evidence in favor of such an accusation. However, to make the case for re-reading chapter 22 as incorporating a notion of intension (at least implicitly) would take more time and space than I have available here. So instead of doing that, I want to spend the remainder of this paper discussing a third sort of possibility, a kind of compromise position that is suggested by the passage from chapter 23 that we were considering. The compromise can allow that even in the absence of a worked-out re-reading of chapter 22, the apparent differences between the views of language that we see in that chapter and what we see in other chapters need not be regarded as a simple inconsistency.

In order to see this third possibility, we can start by noticing that in the middle part of chapter 22, where Xunzi diagnoses and rebuts three different sets of sayings that involve corruption of naming practices, the sayings he discusses are those of Song Xing, Yin Wen, Hui Shi, Deng Xi, Gongsun Long, and the Mohists.²⁰ Mengzi’s view about the goodness of human nature is strikingly absent from that collection, despite the fact that Xunzi criticizes Mengzi explicitly and implicitly in multiple places in the text. This raises a question: are the analyses of those problematic sayings and the rebuttals to them meant to apply to Mengzi’s claim about human nature as well, even though it is not explicitly mentioned there?

One feature of this section might suggest that the analyses and rebuttals are indeed meant to apply to Mengzi, as well as almost anyone else with a philosophical position that Xunzi opposes, for at one point, Xunzi remarks, “In every case of deviant sayings and perverse teachings that depart from the correct Way and recklessly innovate, they will belong to one of these three classes of confusion” (HKCS 22/109/20-21; Hutton 2014, 240).²¹ Furthermore, a little later in the chapter, he says that when one grasps the true Way, then “One will use the true Way to discriminate what is vile just like drawing out the carpenter’s line in order to grasp what is curved and what is straight. Thus, deviant sayings will not be able to cause disorder, and the hundred schools will have nowhere to hide” (HKCS 22/110/9-10; Hutton 2014, 241).²² Especially the reference to the “hundred schools” makes it sound like Xunzi would include Mengzi among those covered by his remarks in this section, as one of those who falls prey to one of the “three confusions.” If that is right, and if Xunzi’s understanding of names in chapter 22 is taken to be purely extensional, then there does seem to be a problem with squaring Xunzi’s analysis there with the example from chapter 23 that I have provided.

However, other evidence suggests that despite the remarks in chapter 22 that make it sound as though all rival philosophers could be classed as succumbing to one of the “three confusions,” Xunzi recognizes that in fact not all philosophical errors stem from corruption of language. For example, Xunzi criticizes the Mohist practice of frugality in chapter 10 of the text (HKCS 10/44/20-10/46/4), and there it is clear that he takes Mozi’s error to consist both in misdiagnosing the greatest source of trouble in the world—namely, as insufficiency—and in thinking that a radical frugality would have the consequence of resolving the problem. In that case, it is clear that on Xunzi’s account, what has gone wrong in Mozi’s view has nothing to do with language but is rather primarily a matter of misunderstanding causes and effects.

Insofar as Xunzi does not treat all philosophical errors as involving corruption of language, perhaps one way to understand the discussion of the “three confusions” in chapter 22 is that the discussion there is limited to a particular class of philosophical errors, namely those that involve not simply mistaken claims, but moreover an abuse of language.²³ If so, then it would not be surprising that outside chapter 22, Xunzi sometimes speaks in ways that imply a different understanding of language. That, then, is the third possibility for understanding the relation between, on the one hand, the examples from chapters 3 and 23 that I have discussed, and on the other hand, the discussion of naming in chapter 22. In that case, one could conceivably still take the treatment of names in chapter 22 to be purely extensional, but that discussion would be formulated around and limited to responding to abuses of language, with the response drawing on just part of Xunzi’s view of naming, rather than representing Xunzi’s complete account of the subject. Such an interpretation is why I have called this third possibility a kind of “compromise” position that one still might adopt in the absence of an argument for seeing chapter 22 as incorporating a notion of intension.

Since I cannot work out that argument on this occasion, I want at least to recommend the idea that chapter 22 does not necessarily represent Xunzi’s full conception of naming. However, note that that could be true, even if an argument were provided to show that one can find a notion of intension in chapter 22’s account of language, so that there is no inconsistency between that chapter and the examples from chapters 3 and 23 that we have examined. Moreover, it would still remain the case that chapter 22’s analysis and rebuttal of the “three confusions” does not capture Xunzi’s overall understanding of philosophical error. So, we must be careful about the lessons we want to draw from chapter 22 in these regards, and that is the final suggestion with which I conclude.

Acknowledgments

For their comments and questions on an earlier version of this essay, I thank Eirik L. Harris, Philip J. Ivanhoe, Al Martinich, Justin Tiwald, and Bryan Van Norden, and two anonymous referees for this journal. I also want to thank Lin Hongxing 林宏星 of Fudan University and Xie Xiaodong 谢晓东 of Xiamen University for inviting me to present earlier versions of this paper at their institutions, where I received helpful feedback from them and the audiences they assembled, and from Wang Kai 王楷 of Beijing Normal University, who served as commentator for my presentation for Fudan. In the present version I have not, regrettably, been able to fully address all the suggestions I was given.

References

• Copi, Irving, and Keith Burgess-Jackson. 1996. Informal Logic, 3rd ed. New Jersey: Prentice Hall.
• Fraser, Chris. 2016. “Language and Logic in the Xunzi.” In Dao Companion to the Philosophy of Xunzi, edited by Eric L. Hutton. Dordrecht: Springer. [https://doi.org/10.1007/978-94-017-7745-2_10]
• Fraser, Chris. 2021. “Representation in Early Chinese Philosophy of Language.” Philosophy East and West 71 (1): 57-78. [https://doi.org/10.1353/pew.2021.0004]
• Hutton, Eric L., trans. 2014. Xunzi: The Complete Text. Princeton: Princeton University Press.
• Lau, D. C., and F. C. Chen, eds. 1996. A Concordance to the Xunzi 荀子逐字索引. Hong Kong: The Commercial Press 商務印書館. (Cited here as “HKCS.”)
• Oxford English Dictionary Online, 3rd ed. n.d. s.v “bridge, n.1.” Accessed July 30, 2021. https://www.oed.com/view/Entry/23233, .
• Prudovsky, Gad. 1997. “Can We Ascribe to Past Thinkers Concepts They Had No Linguistic Means to Express?” History and Theory 36 (1): 15-31. [https://doi.org/10.1111/0018-2656.00002]
• Robinson, Richard. 1950. Definition. Oxford: Clarendon Press.
• Van Norden, Bryan. 2007. Virtue Ethics and Consequentialism in Early China. New York: Cambridge University Press. [https://doi.org/10.1017/CBO9780511497995]