1. See especially Krifka, Manfred, “Common Nouns: A Contrastive Analysis of Chinese and English,” in The Generic Book, ed. Carlson, Gregory N. and Pelletier, Francis Jeffry (Chicago: University of Chicago Press, 1995), 398–411 Google Scholar.

2. Cikoski, John S., “Towards Canons of Philological Method for Analyzing Classical Chinese Texts,” Early China 3 (1978), 25 Google Scholar.

3. Hansen, Chad, Language and Logic in Early China (Ann Arbor: University of Michigan Press, 1983), 33, 176n.5Google Scholar; Dobson, W.A.C.H., Late Archaic Chinese (Toronto: University of Toronto Press, 1959), 21–22 CrossRefGoogle Scholar; Graham, A.C., review of Hansen, Language and Logic, Harvard Journal of Asiatic Studies 45 (1985), 694–95Google Scholar. William Boltz may intend a similar view when he calls classical Chinese nouns “neutral”; see Boltz, “Desultory Notes on Language and Semantics in Ancient China” (review of Hansen, Language and Logic), Journal of the American Oriental Society 105 (1985), 309–10CrossRefGoogle Scholar. Graham's views shifted in response to the analysis by Christoph Harbsmeier that I criticize below, though with reservations. For the shift, see Graham, , Disputers of the Tao: Philosophical Argument in Ancient China (La Salle: Open Court, 1989), 402 Google Scholar; for his reservations, see Graham, , “Reflections and Replies: Harbsmeier,” in Chinese Texts and Philosophical Contexts: Essays Dedicated to Angus C. Graham, ed. Rosemont, Henry Jr. (La Salle: Open Court, 1991), 275–76Google Scholar.

4. Harbsmeier has published his analysis three times: Harbsmeier, Christoph, “Marginalia Sino-Logica,” in Understanding the Chinese Mind: The Philosophical Roots, ed. Allinson, Robert E. (Oxford: Oxford University Press, 1989), 155–61Google Scholar; Harbsmeier, Christoph, “The Mass Noun Hypothesis and the Part-Whole Analysis of the White Horse Dialogue,” in Chinese Texts and Philosophical Contexts, ed. Rosemont, , 49–66 Google Scholar; and Harbsmeier, Christoph, Science and Civilisation in China, vol. 7, pt. 1Google Scholar, Language and Logic (Cambridge: Cambridge University Press, 1998), 311–21Google Scholar. “Marginalia Sino-Logica” was preliminary; the latter two publications differ only in detail.

5. Harbsmeier, , “The Mass Noun Hypothesis,” 60 Google Scholar; Language and Logic, 320.

6. The expression is Quine's. See, for example, Quine, W.V.O., Word and Object (Cambridge, Mass.: MIT Press, 1960), 90–95 Google Scholar.

7. One might try to make this criterion purely syntactic by giving an exhaustive list of the syntactic frames in which nouns must divide their reference, as in Bunt, Harry C., Mass Terms and Model-Theoretic Semantics (Cambridge: Cambridge University Press, 1985), 12–15 Google Scholar, but that would be of limited use for comparative work, since we would need a way to determine which classical Chinese contexts distinguish between count nouns and mass nouns. I assume that we can often tell whether a noun divides its reference on a particular occasion, and that we can distinguish between classical Chinese syntactic frames on that basis.

8. For ease of expression I write as if each noun is associated with, at most, a single principle of individuation. As I make clear in section 5, this is a simplification, but avoiding it would not affect my arguments in any significant way.

9. I treat as formal any feature of a thing that goes beyond what the thing is made of. In this sense, formal features include all sorts of organizational, structural, functional, and even quantitative features.

10. I learned the distinction between functions and classes from Cikoski, John S., Classical Chinese Word Classes (Ph.D. diss., Yale University, 1970), 11–16 Google Scholar; and Boltz, , “Desultory Notes,” 312 Google Scholar.

11. I will not give any detailed justification of this claim, but will mention two common ways in which nouns that we would intuitively classify as mass nouns can be used as count nouns. First, most can be used to refer to kinds, as when the expres-sion “three wines” is used to refer to three kinds of wine. Second, many that refer to food or drink can be used to refer to individual servings or portions, as when “three wines” refers to three glasses of wine. For more on this issue, see the many examples in Ware, Robert X., “Some Bits and Pieces,” in Mass Terms: Some Philosophical Problems, ed. Pelletier, Francis Jeffry (Holland: D. Reidel, 1979), 15–29 Google Scholar; and the discussion of the “universal grinder” in Pelletier, Francis Jeffry, “Non-Singular Reference: Some Preliminaries,” in Mass Terms, 5–6 Google Scholar. See also Allan, Keith, “Nouns and Countability,” Language 56 (1980), 546–47CrossRefGoogle Scholar; and Bunt, , Mass Terms and Model-Theoretic Semantics, 9–12 Google Scholar.

12. My conclusion is close to that of Allan, , “Nouns and Countability,” 545–47Google Scholar, which concludes that the count/mass distinction applies to noun phrases, rather than to lexical entries (that is, more or less, word-types). This is usually true in English, since in English the noun phrase generally includes enough information to determine whether the head noun is a count noun or a mass noun. However, the situation is different in classical Chinese, because it gives determiners much less importance than does English, and its quantifiers are not typically constituents of the relevant noun phrase. Thus, in classical Chinese it is often necessary to consider more than the noun phrase to determine whether a noun divides its reference on a particular occurrence.

13. The argument starts with Hilary Putnam's and Saul Kripke's well-known arguments that something that is not H₂O could not count as water, whatever its macroscopic properties, and add the truism that it is at least partly in virtue of formal features—namely, its arrangement—that some collection of atomic stuff counts as H₂O. See Putnam, Hilary, “The Meaning of ‘Meaning’,” in Mind, Language and Reality, vol. 2 of Philosophical Papers (Cambridge: Cambridge University Press, 1975), 223–25CrossRefGoogle Scholar; and Kripke, Saul, Naming and Necessity (Cambridge, Mass.: Harvard University Press, 1982), 116–29Google Scholar.

14. My view is that when paradigmatic English count nouns are made to function as mass nouns in unusual contexts, they lack structural thresholds. I take this to be one of the lessons to be drawn from Pelletier's “universal grinder” (see Pelletier, “Non-Singular Reference,” 5-6). The grinder is a machine that reduces anything to its component stuff. If you put books in one end, out the other end spurts the stuff those books had been made of. Pelletier makes the plausible claim that it is appropriate to refer to the product of the grinding as book, where “book” (used as a mass noun) refers to the unstructured stuff that the books had been made of, and is not associated with a structural threshold. Unless we accept some version of the ambiguity thesis, we are left concluding that the normal English noun “book” is not associated with a structural threshold. By contrast, if the mass noun hypothesis is true, then a classical Chinese noun such as pian 篇 “bamboo scroll” will have to be associated with a structural threshold even when it occurs as a mass noun, since when used as a mass noun it has the same extension it has when used as a count noun. (When pian is used as a mass noun, it refers to all and only individual pian; it does not refer to the stuff that pian are made out of regardless of what form it is in.) This entails that Pelletier's “universal grinder” argument could not go through in classical Chinese, since it would be in-appropriate to say that when you put pian into the universal grinder, pian comes out the other end: what comes out the end is not pian, but the stuff that the pian was made of. (We get the same effect in English if we imagine what the grinder produces when we pass furniture into it. It certainly isn't furniture that comes out the other end, though it may be a mixture of chair and table.)

15. Allan, Keith, “Nouns and Countability,” 548–65Google Scholar, pursues this sort of analysis and distinguishes eight levels of countability among English common nouns.

16. Strictly speaking, in both of these contexts a noun can function in a third way, as a kind term. (When a noun functions as a kind term, it refers to its entire extension, but construes it as a kind, as in “Horses have four legs” and 聖人不仁 “Sages are unkind.”) Here and in the rest of the article I ignore this function; taking it into account would complicate expression without undermining my basic analysis.

17. A classical Chinese example of this pattern might be 島魚皆生於陰 “birds and fish are both born of yin” (Huainanzi 淮南子 4/35/13; citations are to pian, page, and line in A Concordance to the Huainanzi [Institute of Chinese Studies Ancient Chinese Texts Concordance Series], ed. D.C. Lau 劉殿爵and Chen Fong Ching 陳方正[Hong Kong: The Commercial Press, 1992]). If “birds” and “fish” are count nouns in this sentence, it is probably not because of the jie 皆 “both.”

18. Harbsmeier, , Language and Logic, 318 Google Scholar.

19. Harbsmeier, , “The Mass Noun Hypothesis,” 55 Google Scholar, and Language and Logic, 316. He cites these expressions from Guoyu 3.1/39/20 (citations are to section, page, and line in cordance Series], ed. D.C. Lau and Chen Fong Ching [Hong Kong: The Commercial Press, 1999]), Shangjunshu 2/3/18 (citations are to pian, page, and line in A Concordance to the Shangjunshu [Institute of Chinese Studies Ancient Chinese Texts Concordance Series], ed. Lau, D.C. and Ching, Chen Fong [Hong Kong: The Commercial Press, 1992])Google Scholar, and Hanfeizi 46,948 (citations are to pian, and to page number in Qiyou, Chen 陳奇猷, Hanfeizi jishi 韓非子集釋 [Shanghai: Renmin, 1974])Google Scholar, respectively.

20. Harbsmeier is inconsistent concerning the elements that can occur only with count nouns. “The Mass Noun Hypothesis,” 52, and Language and Logic, 313, list all four terms, but “The Mass Noun Hypothesis,” 55, and Language and Logic, 316, mention only shu 數 “several.” My impression is that the former statement is more authoritative, and I rely on it. Harbsmeier, , “Marginalia Sino-Logica,” 159 Google Scholar, includes ge 各 “each” and jian 兼 “all, both” as distinctive of count nouns, and does not mention mei 每 “every” at all. There are many quantifying expressions that Harbsmeier does not treat. Some produce count contexts, while others produce neutral contexts. I do not know of any quantifying expression in classical Chinese that produces mass contexts. The most obvious candidate, jin 盡 “exhaustively, all,” occurs with a count noun in the sentence 桓公兼此數節者而盡有之 “Duke Huan brought together these several opportunities and jin ‘exhaustively’ had them” (Xunzi 荀子 7/25/15; citations are to pian, page, and line in A Concordance to the Xunzi [Institute of Chinese Studies Ancient Chinese Texts Concordance Series], ed. D.C. Lau and Chen Fong Ching [Hong Kong: The Commercial Press, 1996]).

21. Graham, , “Reflections and Replies,” 275 Google Scholar.

22. The expression occurs at Lüshi chunqiu 呂氏舂秋 18.8/117/27 (citations are to section, page, and line number in A Concordance to the Lüshi Chunqiu [Institute of Chinese Studies Ancient Chinese Texts Concordance Series], ed. Lau, D.C. and Ching, Chen Fong [Hong Kong: The Commercial Press, 1994])Google Scholar. See Harbsmeier, , “The Mass Noun Hypothesis,” 64n.63Google Scholar, and Language and Logic, 317n.9. The same Lüshi chunqiu passage also refers to the two officials as er li 二吏 without the classifier at 18.8/118/2.

23. Harbsmeier, “The Mass Noun Hypothesis,” 57, tentatively accepts wu “thing” as a generic noun, but the suggestion does not occur in Harbsmeier, Language and Logic.

24. Graham, , “Reflections and Replies,” 275–76Google Scholar.

25. In support of Graham, wu 物 “thing” does seem to divide its reference into kinds in the sentence 萬物之生而各異類 “As for the production of the ten thousand wu 物 ‘things,’ each is of a different kind” (Huainanzi 4/35/17).

26. There is a typographical error in Harbsmeier, , Language and Logic, 316 Google Scholar, resulting in the mistaken attribution of the expression shu min 數民 “several people” to Mengzi 孟子 1.2/1/14 (citations are to section, page, and line in A Concordance to the Mengzi [Institute of Chinese Studies Ancient Chinese Texts Concordance Series], ed. Lau, D.C. and Ching, Chen Fong [Hong Kong: The Commercial Press, 1995])Google Scholar. The expression used there is shu min 庶民 “numerous people,” which is cited accurately in Harbsmeier, , “The Mass Noun Hypothesis,” 55 Google Scholar.

27. Graham's failure to draw the stronger conclusion may be based on a misreading. Graham takes Harbsmeier to be claiming that the use of container classifiers is dis tine-tive of mass nouns in two ways: mass nouns require these classifiers when counted, and only mass nouns can occur with them ( Graham, , “Reflections and Replies,” 275 Google Scholar). But Harbsmeier makes neither of these claims. He explicitly recognizes that one can directly count mass nouns in expressions such as san jiu 三酒 “three wines = three kinds of wine.” In “The Mass Noun Hypothesis” (to which Graham was responding), he says nothing about phrases such as yi ju ren 一車人 “one cartful of people,” but in Language and Logic, 318, he allows that they are probably acceptable, with ren 人 “person” functioning here atypically as a mass noun. Thus, besides his rule 2, the only syntactic difference Harbsmeier posits between count nouns and mass nouns is that the former can, and the latter can't, occur with certain quantifying expressions.

28. If these nouns behave differently in neutral contexts, that would also be a semantic difference. But because Harbsmeier does not see the importance of neutral contexts, he does not raise this possibility. I turn to this issue in section 6.

29. Harbsmeier, , “Marginalia Sino-Logica,” 159 Google Scholar.

30. Ci shu huan 此數患 “these several disasters” refers to individual disasters at Zhuangzi 莊子53/20/38 (citations are to page, pian, and line in Zhuangzi yinde 莊子引得, Harvard-Yenching Institute Sinological Index Series, Supplement No. 20 [Peiping: Harvard-Yenching Institute, 1947])Google Scholar, but qi huan 七患 “seven disasters” refers to kinds of disaster in the title and opening passage of Mozi 墨子5; see Mozi 4/5/1 (citations are to page, pian, and line in Mozi yinde 墨子引得, Harvard-Yenching Institute Sinological Index Series, Supplement No. 21 [Peiping: Harvard-Yenching Institute, 1948])Google Scholar. Harbsmeier discusses this example in “The Mass Noun Hypothesis,” 52. When it occurs as a count noun, ma “horse” usually refers to individual horses, but the expression liu ma 六馬” six horses” refers to six kinds of horses at Zhou li 周禮 4.51/60/13 (citations are to section, page, and line in A Concordance to the Zhouli [Institute of Chinese Studies Ancient Chinese Texts Concordance Series], ed. Lau, D.C. and Ching, Chen Fong [Hong Kong: The Commercial Press, 1993])Google Scholar, Harbsmeier, mentions this case in “The Mass Noun Hypothesis,” 52 Google Scholar, and Language and Logic, 319. Ma 馬 probably also divides its reference into kinds of horses in the stock expression liu ma 六馬 “the six horses,” which refers to the horses that take up six positions on a team, as in Hanfeizi 34.717.

31. Mozi 25/16/23-26/16/45.

32. Harbsmeier, , “The Mass Noun Hypothesis,” 52 Google Scholar, and Language and Logic, 314.

33. Xunzi 2/8/9.

34. Xunzi 6/23/6.

35. Xunzi 9/36/10.

36. Xunzi 12/60/20.

37. Zuo zhuan 左傳, Duke Xiang 3.4/230/16 (citations are to reign, reign year and paragraph, page, and line in A Concordance to the Chunqiu Zuozhuan [Institute of Chinese Studies Ancient Chinese Texts Concordance Series], ed. Lau, D.C. and Ching, Chen Fong [Hong Kong: The Commercial Press, 1995]Google Scholar; Bojun, Yang 楊伯峻, Chunqiu Zuo zhuan zhu 春秋左傳注, rev. ed. [Beijing: Zhonghua, 1981]Google Scholar, uses the same paragraph divisions). Wu 物here refers to accomplishments or more specifically to promotions.

38. Zuo zhuan, Duke Zhao 7.14/340/24. The wu 物here seem to be various lengths of time defined by natural cycles, though this passage is subject to commentarial disagreement. (I have based my translation on commentaries cited by Bojun, Yang, Chunqiu Zuo zhuan zhu, 1297.Google Scholar)

39. Zuo zhuan, Duke Zhao 25.3/387/8. The “two things” are probably husband and wife, or perhaps their roles. Legge, James, The Chinese Classics, Volume 5: The Ch'un Ts'ew with the Tso chuen (Hong Kong: Hong Kong University Press, 1960), 708 Google Scholar, treats husband, wife, inner, and outer as four distinct things. Presumably the two things would then be the two distinctions in question.

40. Zuo zhuan, Duke Zhao 31.5/403/30. Here the “two things” are the two narratives, or the lessons they exemplify.

41. Mozi 69/4Z/41. Here “one reality” most likely refers to one kind of thing. The stock example for the situation described here is dogs, who can be referred to both as gou 狗and as quan 犬. All dogs taken together count here as one reality. See Graham, A.C., Later Mohist Logic, Ethics, and Science (Hong Kong: Chinese University Press, 1978), 218, 335 Google Scholar; cf. Mozi 67/41/27 and Graham, , Later Mohist Logic, 408–9Google Scholar.

42. Xunzi 22/109/11. Shi 實 here are distinguished by spatial separateness. Thus shi 實 “reality” is now dividing its reference into individual things.

43. Lüshi chunqiu 7.3/35/13. Here one shi 實is just one reality, as distinct from the way it is described, and the individual/kind distinction doesn't apply. We might get a similar effect in English by saying that attacking and defending are “really the same thing” or that they “amount to the same thing.” (Chris Fraser brought this example to my attention.)

44. I adopt this phrasing in order to allow for the fact that most nouns can function as kind terms in some neutral contexts.

45. Xunzi 1/2/5.

46. There are many cases where it is not clear that the two potential interpretations of a sentence differ in any significant way. In these cases, the noun occurs as a mass noun, since dividing its reference would not alter the meaning of the sentence. (I consider cases of this sort in section 7.)

47. Zhuangzi 1/1/1.

48. Ming 名 “name” also occurs in a neutral context; I will discuss it below. Kun 廳 “fish eggs” and bei ming 北冥 “northern darkness” may both be proper names. Even if they are, both still have descriptive content, and my arguments in the main text are relevant to determining what that content is. For example, they entail that the fish would have as a name a mass noun (Fish Roe) rather than a count noun (Fish Eggs). (I thank Donald Harper for suggesting this idea.) If, as many commentators suggest, kun 鲲 is the name of a species of large fish rather than a proper name, then it would be occurring as a kind term, and not as either a count noun or a mass noun.

49. Zhuangzi 7/2/94.

50. Zhuangzi 24/10/26.

51. Mengzi 1.1/1/3.

52. Zuo zhuan, Duke Zhao 16.3/364/15-365/2.

53. The ring is introduced with the sentence 宣子有環，其一在鄭商 “Xuanzi had jade rings, one of them was held by a merchant in Zheng” (Zwo zhuan, Duke Zhao 16.3/364/15). It is referred to as yi huan 一環at Zhao 16.3/364/17, and as yi yu 一玉at Zhao 16.3/364/22.

54. Zuo zhuan, Duke Zhao 16.3/364/22,23,30; and 365/1,2.

55. This paragraph responds to a concern raised by an anonymous referee for Early China. I address a related issue in section 10.

56. Mengzi 11.8/59/11.

57. Xunzi 1/1/12.

58. Xunzi 1/2/10.

59. Zhuangzi 42/17/9. Da 大 also has a moralistic use in which it does not force individuation.

60. Mengzi 2.11/12/7.

61. The Mohists seem to mark this distinction in Mozi 76/44/33:

苟是石也白，敗是石也盡與白同。是石也唯[=雖]大，不與大同。 If this stone is white, then broken this stone is all the same as the white. If the stone was big, it is not the same as the big thing.

62. Mengzi 11.7/59/7.

63. See, for example, the sentences 以偷生反側於亂世之閒 “In order to make off with their lives they switch sides in a disorderly age” (Xunzi 4/14/14) and 夢之中又占其夢焉 “Within a dream—one even interprets one's dream there” (Zhuangzi 6/2/82).

64. Xunzi 1/1/3.

65. Commonsense individuation is required to interpret sentences such as金重於羽 “Metal is heavier than feathers” (Mengzi 12.1/62/3), where the point is surely not that some number of coins is heavier than an equal number of feathers. On the general issue of modification and individuation, see Bunt, Mass Terms, 197-211.

66. Note that we are prevented from drawing this conclusion in the English case only because the lack of a determiner and a plural morpheme forces us to treat “wood” as a mass noun even in the expression “straight wood.” There is no similar pressure to treat the classical Chinese mu 木 “wood” as a mass noun in zhi mw 直木 “straight wood.”

67. The semantic neutrality of English count nouns obscures the fact that, syn tactically speaking, they are marked, while mass nouns are neutral. (Allan, “Nouns and Countability,” 545, states that there is no language in which mass nouns are syntacti-cally marked, but count nouns are not.)

68. Hansen, , Language and Logic, 32–37 Google Scholar.

69. Hansen, Chad, A Daoist Theory of Chinese Thought: A Philosophical Interpretation (London: Oxford University Press, 1992), 46–48 Google Scholar, relates the mass noun hypothesis instead to the problem of change, or the relation between a thing and the stuff it is made of. I believe that this has more to do with structural thresholds than it does with the count/mass distinction, and leave the issue aside here.

70. Harbsmeier argues against Hansen's claims on the ground that they attribute to Warring States philosophers a highly unnatural and abstract view. This assumes that the universal/particular view is not merely familiar, but is also in some sense objectively more natural, or less theoretical. As Graham, , “Reflections and Replies,” 276–77Google Scholar, points out, this begs the question against Hansen. See Harbsmeier, , “The Mass Noun Hypothesis,” 49–50 Google Scholar, and Language and Logic, 311-12.

71. See especially Hansen, Language and Logic, chap. 4.

72. A singular term is a term that construes its reference as a single object. (Proper names are often offered as examples of singular terms.) Singular terms contrast with general terms, which can refer to two things of the appropriate kind without construing them as a single thing. See Hansen, , Language and Logic, 35 Google Scholar, and A Daoist Theory, 49, notes * and **.