20th WCP: The Structure of Chinese Language and Ontological Insights: A Collective-Noun Hypothesis*

1. Introduction

Through a comparative case analysis regarding the Chinese language, this paper discusses how the structure and functions of a natural language would bear upon the ways in which some philosophical problems are posed and some ontological insights are shaped. In so doing, I suggest and argue for a mereological collective-noun hypothesis about the denotational semantics of Chinese nouns. By this case analysis, the paper aims to contribute to the elucidation of the relation between language and philosophy in this regard.

My discussion begins with a puzzle: why the classical Platonic one-many problem in the Western philosophical tradition has not been consciously posed in the Chinese philosophical tradition and why, generally speaking, classical Chinese philosophers seem less interested in debating the relevant ontological issues. (1)

One suspects that the structures and uses of different languages might play their roles in pushing philosophical theorization in different directions; the ways of speaking and writing of the Chinese language might reveal and reflect Chinese folk ideology and then influence the ways in which certain philosophical questions are posed and certain ontological insights are formed. This puzzle is significant because it is concerned with a fundamental philosophical question about the relation between thought and language.

The problem of relating Chinese thought to the structure and functions of the Chinese language has for generations tantalized sinologists and those philosophers who are concerned with the problem. Nevertheless, in the last decade, some significant progress has been made in this regard. In his book Language and Logic in Ancient China, (2) Chad Hansen advances a novel and provocative theory about the nature of the classical Chinese language. (3) The central thesis of Hansen's theory is his mass-noun hypothesis. Its main ideas are these: (1) the (folk) semantics of Chinese nouns are like those of mass-nouns (i.e., those nouns referring to the so-called interpenetrating stuffs, like 'water' and 'snow'), and naming in Chinese is not grounded on the existence of, or roles for, abstract entities (either on the ontonic level or on the conceptual level) but rather on finding 'boundaries' between things ("to know a word is simply to be able to discriminate" (4)); (2) influenced by the mass-noun semantics, the classical Chinese semantic theorists and ontological theorists view words in ways that are natural to view mass nouns rather than count nouns, and Chinese theorists tend to organize the objects in the world in a mereological stuff-whole model of reality. In this way, according to Hansen, the language theory of classical Chinese philosophers differs fundamentally from the language theory of Western philosophy.

To my knowledge, Hansen's mass-noun hypothesis of the classical Chinese semantics and ontology,since being put forward, has been challenged mainly in two ways. One way is to challenge the mass-stuff model from the perspective of a holographic process ontology. (5) The other way is to directly challenge Hansen's mass noun hypothesis, arguing that there is a clear grammatical distinction in classical Chinese between count nouns and other nouns (generic nouns and mass nouns). (6) As far as the second way to challenge Hansen's mass noun hypothesis is concerned, Hansen thinks that it misses the his point. (7) I think that, although Harbsmeier's criticism does miss Hansen's point to some extent, it is still somehow relevant to evaluation of his hypothesis. Nevertheless, for the sake of space, I will not further discuss Harbsmeier's challenge here. The challenge from the holographic perspective is indeed a genuine challenge to Hansen's mereological mass-stuff hypothesis. Now the real question is whether the former conflicts with or is compatible with, or even an elaboration of, the latter. Nevertheless, in this work, I will not directly pursue this issue. Rather, I will respond to Hansen's view in the same semantic referential approach and within the same mereological-analysis track.

In this paper, I suggest and argue for a collective-noun hypothesis. Its main ideas are these: (1) Chinese nouns typically function, semantically and syntactically, in the way collective-nouns function, and the folk semantics of Chinese nouns are like those of collective-nouns; (2) their implicit ontology is a mereological ontology of collection-of-individuals both with the part-whole structure and with the member-class structure, which does justice to the role of abstraction at the conceptual level; and, (3) encouraged and shaped by the folk semantics of Chinese nouns, the classical Chinese theorists of language take this kind of mereological nominalism for granted; as a result, the classical Platonic one-many problem in the Western philosophical tradition has not been consciously posed in the Chinese philosophical tradition, and the classical Chinese philosophers seem less interested in debating the relevant ontological issues. This mereological collection-of-individuals model of reality, I believe, would provide a more reasonable interpretation of the semantics of classical Chinese nouns and the classical Chinese ontological theory.

In the following, first, I will explain why Chinese nouns do not typically function in the count-noun pattern but in a certain none-count-noun pattern. Second, I will explain why the semantics and implicit ontology revealed and reflected by the semantics and syntax of Chinese nouns is a mereological nominalist ontology of collection-of-individuals which classical Chinese philosophers take for granted.

2. The Structure of Chinese Nouns and Ontological Models

In contrast to those typical common nouns, the so-called count nouns like 'person' and 'horse', in Indo-European languages such as Greek and English, the syntax of Chinese nouns appears strikingly similar to the syntax of those uncountable nouns either 'collective nouns' (such as 'people', 'cattle' and 'police') or 'mass nouns' (such as 'water' and 'snow'). The reason that count nouns are so called is a semantic reason: their direct referents are countable separate individuals: one person, nine horses, five mountains. The grammatical representation of this semantic point is this: count nouns have their plural forms, or, they can be pluralized. In this way, when count nouns are used in Indo-European languages, they refer either to one single individual or to many separate individuals. By contrast, collective nouns and mass nouns do not take pluralization. For, when they stand alone, collective nouns and mass nouns are not supposed to stand for countable individuals but for a whole, either a collection-whole or a mass-whole. It seems that Chinese nouns do not typically function in a count-noun pattern but in some non-count-noun pattern. For, grammatically, classical Chinese nouns have no plural forms, and even the nouns in modern Chinese have no plural form at least in the sense of 'plural forms' in the Indo-European languages. (8) This linguistic fact seems to reveal and reflect the following semantic point: a Chinese noun, when standing alone, typically denotes a whole of many things or a whole of much stuff rather than one individual.

One tendency regarding the relation between linguistic expressions and their referents seems to be common both in the Chinese semantic tradition and in the Western semantic tradition (at least during their respective classical periods): almost all words were treated as names, and naming was regarded as the main semantic relation. Naturally, naming is taken as a one-to-one relation; that is, one name refers to some one single thing, whatever it is and whether or not it actually consists in many things or in much stuff. However, the one-to-one naming paradigm, when connected with different noun-syntactic patterns, would reveal and reflect different semantic orientations and different ontological tendencies. When the one-to-one naming paradigm goes with the count-noun syntactic pattern, names typically behave like count nouns; that is, it seems that they are typically supposed to pick up countable, separate, and self-sufficient things rather than to refer to wholes which themselves consist of many countable things or of much stuffs. However, when the one-to-one naming paradigm goes with the non-count-noun syntactic pattern (either with the collective-noun pattern or with the mass-noun pattern), names do not behave like count nouns; they are typically supposed to denote wholes of many things or wholes of much stuffs. The folk semantics connected with a non-count-noun syntactic pattern tends to organize the objects basically under the part-whole relation and hence makes its implicit ontology have a mereological character.

Now, if the nouns in the Indo-European language in which Western philosophy expresses itself typically function as count nouns while the Chinese nouns typically function as non-count nouns (either mass nouns or collective nouns), and if the folk semantics and implicit ontology of nouns in a language would contribute to shaping the intuitive understanding of the language-world relation in a philosophical tradition which uses the language to express itself, then the different ways of noun-functioning would encourage and motivate different ontological models of reality. With the count-noun function-pattern and one-to-one naming paradigm in mind, one might be encouraged and motivated to think that the world consists of countable self-sufficient things both at the particular level and at the universal level when one looks at the structure of the world. Under the count-noun functioning pattern in the Indo-European language, the Platonic one-many problem with the following presupposition seems to be quite natural: there is one single, self-sufficient universal entity which is common or strictly identical across all the particular individuals which share the same name. Those philosophers with this presupposition in mind have searched for such a single entity and tended to identify it either with one ontonic universal instantiated by particulars (Platonic realism or some other versions of realism regarding universals) or with one conceptual entity shared by minds (conceptualism).

However, if the folk semantics of Chinese nouns, whether it goes with the collective-noun function pattern or with the mass-noun function pattern, tends to organize the objects basically under the part-whole relation and hence makes their implicit ontology have a mereological character, the classical Chinese philosophers who use Chinese nouns to express themselves would be encouraged to look at the world in terms of mereological ontology, and they would be discouraged from posing the Platonic one-many problem with the presupposition aforementioned. For the classical Chinese philosophers, the common names raise no Platonic one-many problem at all. That, I believe, is why the classical Platonic one-many problem has not been consciously posed in the Chinese philosophical tradition and, generally speaking, the classical Chinese philosophers seem less interested in debating the relevant ontological issues.

3. The Collection-of-individuals as a Mereological Whole

My argument in this part will be given in the following steps: first, I will further examine some relevant linguistic (syntactic and semantic) evidence for the collective-noun hypothesis; second, I will spell out why the collective-noun hypothesis has a stronger explanatory power than the mass-noun hypothesis does in several respects; and third, I will explain how it is possible for a mereological whole to be collection-of-individuals as a class in terms of certain contemporary conceptual resources.

Now what is at issue is this: Does the function of Chinese nouns go with the collective-noun function pattern or with the mass-noun function pattern? Although the function of Chinese nouns examined so far appears to be compatible with both patterns, and although Chinese nouns in both patterns denote mereological wholes, the semantics and implicit ontologies in the two cases are significantly different. A collection-whole and a mass-fusion whole have different ontological structures: the former consists of (many) separate individuals, while the latter consists of (much) inseparable and interpenetrating stuff; and they have different part-whole structures to be discussed. (9)

If one wants to give a mass-noun interpretation of the folk semantics of Chinese nouns, there should be convincing linguistic evidence regarding the semantics of Chinese nouns. Hansen provides further linguistic evidence to support his mass-noun model. In English, a count noun can be directly preceded by (cardinal) numerals or indefinite articles 'a' or 'an', whereas a mass noun cannot be directly preceded by numerals or indefinite articles. Thus, phrase 'two horses' is well-formed while the phrase 'two water' is ill-formed. Instead, one has to associate mass nouns with certain other expressions (called 'measures' or 'sortals') so as to turn what they denote into countable units. For example, in English, we say 'two cups of water'. On the other hand, after the Han dynasty, Chinese nouns, like mass nouns in English, cannot be directly preceded by numerals or indefinite articles or demonstratives; each noun is associated with appropriate measures or sortals. Thus, in Chinese, instead of saying "san-mao [three cats]", "wu-ren [five persons]" and "yi-gou [one dog]", one says "san-ZHI- mao", "wu- KOU-ren" and "yi-TIAO-gou". Note that Hansen's point of resorting to this syntactic feature of modern Chinese nouns is not purely grammatical but semantic: the words like 'zhi', 'kou' and 'tiao' stand for some quantity-measures like a cup, and 'mao', 'ren' and 'gou' refer to mass-stuff; the folk semantics of Chinese nouns are like those of mass-nouns. Hansen indeed notices that the classical Chinese nouns are directly preceded by numerals or demonstratives, and sortals or measures were not grammatically necessary for the classical Chinese noun. For example, one simply says "san-ren" ['three persons'], "wu- ma" ['five horses'], and "ci ren" ['this person']. But Hansen takes this linguistic fact in his favor, suggesting that the semantics of mass-nouns in the pre-Han classical Chinese "may have affected Chinese language development": "[g]radually nouns came to have a more uniform mass-noun syntax", and "[t]he number-sortal forms became standardized during the Han dynasty". (10)

This linguistic analysis, I believe, is problematic or at least controversial. The reasons are these. First, consider those nouns in modern Chinese corresponding to count nouns in the Indo-European languages. Those words which go between numerals and the nouns are not really measures or the quantity-indicators; rather, they are what I call 'individualization-indicators'. These individual-indicators, unlike such genuine measure-expressions as 'cup', indicate certain features of individuals rather than certain quantity-measures. (11) The genuine function of those individualization-indicators, in my opinion, is to individualize the collections denoted by, say, 'mao', 'ren', and 'gou' so as to count the individual(s) in the respective collections. This point will be highlighted by their following characteristic usage: if one cannot find a suitable individualization-indicator which would best match the noun in question, one could just use the Chinese word 'ge' (meaning individual) instead. For example, instead of saying "san-ZHI-mao", "wu-KOU-ren", and "yi-TIAO- gou", one can simply say "san-GE-mao", "wu-GE-ren", and "yi-GE-gou". The function of 'ge' here performs the same function which those substituted individualization-indicators are supposed to perform: to individualize the collections denoted by, say, 'mao', 'ren', and 'gou' so as to count the individual(s) in the respective collections. The meaning of 'ge' here highlights the function of individualization.

Second, if one agrees that the grammar of classical Chinese nouns is not a mass-noun grammar but at the same time insists that the semantics of classical Chinese nouns is like those of mass-nouns, and if those grammatical features of Chinese nouns do reveal certain semantic points, then we are owed an explanation of why there were the tensions between the non-mass-noun grammar of classical Chinese nouns and their mass-noun semantics and between the non-mass-noun grammar of classical Chinese nouns and the alleged mass-noun grammar of modern Chinese nouns. On the other hand, it seems that the tensions do not go with the collective-noun hypothesis. For one thing, as discussed in the last section, collective nouns in classical Chinese can be directly preceded by numerals or indefinite quantifiers. For another thing, as explained above, the grammar of those nouns in modern Chinese corresponding to count nouns in the Indo-European languages should not be regarded as a mass-noun grammar.

Third, if a Chinese noun corresponding to a count noun in English is a pictograph to indicate something concrete, the pictograph, etymologically speaking, typically depicts the common characteristic shape or feature of those separate individuals in a collection rather than of an inseparable stuff-whole; the countable units are taken as separate individuals rather than bits of inseparable and interpenetrating stuff.

The available linguistic evidence thus suggests that, first, given that the function-pattern of Chinese nouns is a non-count-noun pattern, a collective-noun hypothesis seems to be a more reasonable interpretation of the folk semantics of Chinese nouns than a mass-nouns hypothesis; second, what the folk semantics of Chinese nouns encourages and shapes, if any, seems to be a collection-of-individuals paradigm of reality rather than a mass-stuff paradigm of reality. (12)

Furthermore, theoretically speaking, it seems implausible to say that a Chinese noun, say, 'horse' refers to a mereological horse-stuff whole in the way in which a genuine mass term like 'water' refers to water-stuff. There are two reasons. First, the horse-whole and the water-whole, for instance, have different ontological structures: the horse-whole, in fact, is a collection of many separate individual horses, while the water-whole is a collection of much inseparable and interpenetrating stuff. Why this difference is significant is to be seen below. Second, assuming that a whole is divisible into parts above the molecule level, one can find that the transitivity of the parthood would highlight the different part-whole structures of the horse-whole and of the water-whole. If a horse is a part of the horse-whole, then, since a horse-leg is a part of a horse, a horse-leg is also a part of the horse-whole because the parthood is transitive; however, it is obvious that the noun 'horse' is supposed to refer to the horse rather than to the horse-leg. Nevertheless, on the other hand, the genuine mass noun, say 'water', would not be confronted with the same problem when what it refers to is taken as a mereological object, so long as water is divisible above the molecule level. (13) These two reasons suggest two difficulties with the mass-noun hypothesis in regard to its explanatory power.

There is the third difficulty which is not necessarily connected with the mass-stuff model but with Hansen's interpretation of it: in my opinion, it is implausible to say that people discriminate one type of stuff from another or one particular part of horse-stuff from other parts without any abstraction in mind. For, according to our own cognitive experience, any cognitive process which involves distinguishing one stuff from another stuff cannot be free of abstraction. Hansen say: "...horse can be held to be the name of a mereological object---a thing-kind, and we can explain our ability to distinguish horse-stuff from ox-stuff as our ability to distinguish characteristic shape"; (14) This remark is not clear enough. As said before, many horses are recognized as the same kind of things according to their characteristic shapes. However, drawing the characteristic shapes of kind from the characteristic shapes of many individual horses has to resort to some abstract generalization regarding the similarity of their shapes; only on the basis of having already recognized the two kinds of stuff (horse-stuff and ox-stuff) respectively can one distinguish one kind of stuff from other kinds of stuff.

The collection-of-individuals model can effectively solve the first problem. For the model directly regards a noun as denoting a collection-whole of individual things and takes the case of the mass noun to be a special case: it denotes a collection of one single inseparable and interpenetrating stuff. Moreover, the collection-of-individuals model can directly solve the third problem. For it takes a collection as a class; a class is understood as the extension of a concept which consists in some abstraction.

However, one might immediately object: if the collection is viewed as a mereological whole, the collection-of-individuals model would be in the same boat as the mass-stuff model when being confronted with the problem about the horse-leg aforementioned. For, although one can say that a horse-leg is not an individual specified by the concept of horse and hence is not a member of the horse-collection as a class, it is still a part of an individual horse; If a horse is taken as a part of the horse-collection whole, then, since parthood is transitive, it follows that a horse-leg is a part of the horse-collection whole. Let me call this objection 'the horse-leg objection'. One might further object: if a collection is taken as a class, it is simply not a mereological whole. Let me call this objection 'the no-mereological-class objection'.

I believe that the no-mereological-class objection is really a dogma. David Lewis's recent work on mereological classes (15) shows that a class could be a mereological whole in a certain consistent way. Below, by resorting to some conceptual resources in Lewis's formal account of mereological classes, I explain the characteristic features of the collection-whole model and how the model can deal with the horse-leg objection.

The crucial differences that distinguish the mereological interpretation of the collection-of-individuals model from the mereological interpretation of the mass-stuff model are these:

(1) a is a member of A if and only if A is a class and the singleton {a} of a is part of A (16) : that is, a collection A to which a noun refers is taken both as a mereological whole and as a class; what is taken as a part of the whole A is not individual a but its subclass---singleton {a}; the part-whole relation here is the relation between class and subclass.

(2) In this model of reality, reality divides exhaustively into individuals and classes as mereological fusions, (17) i.e., collection wholes, rather than into mass-stuff wholes. Note that there is a fundamental difference between a collection-whole and a mass-stuff-whole: as far as the ontological status is concerned, the parts of the former could be separable individuals while the parts of the latter could not.

(3) An individual a and its singleton {a} have the exactly same ontological status but differ only in our different ways of understanding the classification of a at the conceptual level. (18)

(4) What really distinguishes a horse as a part from a horse-leg as a part is a certain concept which characterizes the membership of the class; since the horse-leg of a horse is not a subclass of the horse's singleton though it is indeed part of the horse, the horse-leg is not a part of horse-whole.

The fact that we can give a consistent interpretation of the classical ontological insight in terms of a modern account of mereological classes does not mean that the classical Chinese philosophers have those aforementioned technical conception regarding mereological classes. Nevertheless, the fact does suggest: on the one hand, the collection-of-individuals model revealed and reflected by the semantics of Chinese nouns could be a consistent ontological insight and be precisely defined in terms of some contemporary conceptual resources; on the other hand, a certain natural way of speaking or the folk semantics of a natural language might motivate and push philosophical theorization in a convincing and promising direction.

It is noted that, in my preceding discussion, for the sake of convenience, I focus on such Chinese nouns like ma (the collection of horses) or shui (the collection of water); the references of those collective nouns are either individual physical objects or physical stuff. However, what constitutes a mereological collection in the collective-noun hypothesis is not necessarily something physical, static or individual-object-like. It might well be something dynamic or process-like. In this regard, a hexagram, as a noun-like ideographic symbol serving in the philosophical context of Yi-Jing, quite explicitly presents us another typical case of Chinese ideographic nouns under the collective-noun hypothesis. That is, the denotational reference of a hexagram is a collections of dynamic processes with shared changing pattern which is ideographically symbolized by its graphic shape. In my another writing, (19) I have shown that, at the interface between the sub-system of hexagrams in Yi-Jing and other parts of Chinese ideographic language, a hexagram could serve syntactically and semantically as a collective noun; and, in the philosophical context of Yi-Jing, a hexagram does behave as a characteristic collective noun. In this way, the ideographic semantic structure of the hexagram would constitute not only a strong support and effective illustration of the collective-noun hypothesis but also its significant extension regarding dynamic process reference. (20)

Notes

* A longer version of this paper is forthcoming in Philosophy East and West Vol. 48, No. 4. This version appears with the permission of the Editor of Philosophy East and West.

(1) The Platonic one-many problem begins with the following observation: objects around us share features with other objects; and many particular individuals, say, horses bear the same name 'horse'. The Platonic one-many problem presupposes that there is one single universal entity which is common or strictly identical across all those particular concrete horses and by virtue of which many individual horses bear the same name 'horse'; the single universal entity is labeled 'horseness'. (Cf., Plato: Phaedo, 78c-e, 104a; Republic, 596a) The Platonic one-many problem is how to characterize the status of universals and the ways by which particulars share universals. Under the presupposition aforementioned, there is the debate among Platonic realism, other versions of realism, and conceptualism In the Western philosophical tradition, nominalism appears as a challenge to the very presupposition of the Platonic one-many problem: there are no universals except for general words like 'horse' and 'dog'. The debate between Platonic realism and nominalism is thus metaphilosophical.

Note that the Platonic one-many problem is not the same as a neutral question of how a common name refers to many things. The latter issue is neutral because the question itself does not presuppose any specific ontological position, whether Platonic realism or conceptualism or nominalism. This neutral issue can be investigated from different perspectives and in different ways. For example, it could be approached from a mathematical perspective: in group theory, the structure of one-to-many mapping relation, homomorphism, is given a precise formal presentation. However, the Platonic one-many problem is not ontologically neutral but goes with the presupposition in question.

(2) Hansen, Chad (1983): Language and Logic in Ancient China (Ann Arbor: The University of Michigan Press). Also see his subsequent further defending writings; e.g., his (1985) : 'Chinese Language, Chinese Philosophy, and "Truth" ' in The Journal of Asian Studies, Vol. XLIV, no.3 (May 1985), pp. 491-519; and his (1992): A Daoist Theory of Chinese Thought (New York: Oxford University Press).

(3) Cf., Hansen, 1983, pp. 30-54; especially, pp. 37-42; 1992, Chapter 2.

(4) Ibid., p. 39.

(5)Cf., Chung-ying Cheng (1987): 'Logic and Language in Chinese Philosophy', in Journal of Chinese Philosophy 14, pp. 285-307; David Hall and Roger Ames (1987): Thinking Through Confucius(Albany: SUNY Press), especially pp. 261-268. According to this way is concerned, although some scholars also emphasize the implicit ontology of Chinese language, they focus on the case analysis of the typical philosophical nouns or terms, such as 'tai-ji ', 'wu', 'yin-yang', 'wu-xing', which constitute the basic lexicon (vocabulary) of Chinese metaphysical systems as found in the writings of the early Confucianists, the early Daoist, and Neo-Confucianists; they argue that those nouns stand for interpenetrating wholes and parts in a quite different sense from Hansen's: the individual things behave in the on-going patterns and in the events or processes of interaction among them, and the universe behaves as an organic whole with parts exemplifying the structure of the whole; they claim that Chinese words in general share this ontological feature of combing universality and particularity, abstractness and concreteness, activity and the result of activity. In this way, some writers prefer to consider the relations of 'parts' and 'wholes' in terms of the model of 'focus' and 'field' and take Chinese ontological views as holographic rather than mereological. (Cf., David Hall and Roger Ames, 1987, pp. 262-264.)

(6) Cf., Harbsmeier, Christoph (1989): "Marginalia Sino-logica", in Allinson, Robert E. ed. (1989): Understanding the Chinese Mind: The Philosophical Roots (Oxford University Press), pp. 125-166; (1991): "The Mass Noun Hypothesis and the Part-Whole Analysis of the White Horse Dialogue", in Henry Rosemont, Jr. ed. (1991): Chinese Texts and Philosophical Contexts (La Salle, Illinois: Open Court), pp. 49-66.

(7) Cf., Hansen, 1992, one of footnotes on p. 48.

(8) In English, count nouns have another feature: they cannot stand alone as a noun phrase unless they are in plural form or preceded by articles, demonstratives, numerals, and so forth. For example, 'Horse is white' is not grammatical in English. In contrast, collective nouns and mass nouns can stand alone as noun phrases in sentences. For example, 'People like it' and 'Snow is white' are perfectly grammatical English sentences. Like collective nouns and mass nouns in English, Chinese nouns by themselves are complete noun phrases. Nevertheless, in Greek, a count noun can stand alone as a noun phrase, by default denoting one individual. So, I take the grammatical feature of a noun per se being a complete noun-phrase in the above sense to be necessary, rather than sufficient, for functioning as a non-count-noun.

Moreover, one also notes that, in English, count nouns go with 'many', while mass nouns go with 'much'; however, Chinese nouns lack a many/much distinction. Nevertheless, a lack of many/much distinction does not mean the assimilation of many/much into much. So, this linguistic fact is not necessarily in favor of the mass-noun hypothesis, although it might suggest that, in Chinese, there seems a more uniform way to deal with measure-expressions.

(9)Furthermore, the mass-noun hypothesis and the collective-noun hypothesis have different epistemological implications. If the mass-noun hypothesis is right, one appears to be not obliged to commit oneself to any kind of abstraction: not only the abstract object at the ontonic level regarding what exists but also the abstraction at the conceptual level regarding people's cognitive process. For, according to Hansen, once one wants to refer to a particular part of this stuff-whole, one just uses one's mind "not as repositories of weird [abstract] objects called ideas, but as the faculty encompassing the abilities and inclinations to discriminate stuffs from each other." (Nevertheless, I will explain why such discrimination could not be free of abstraction.)

However, if the collective-noun hypothesis is right, one has to do justice to the role of abstraction as the ability of generalization at the conceptual level, although one is not necessarily committed to any abstract entities at the ontonic level. For, in this case, there is still a question of why and how certain particular individuals as members are classified into, or are viewed as, one collection-whole as a class. In such a cognitive process, some abstraction at the conceptual level is unavoidable. For example, many horses are recognized as the same class or the same kind of things according to, say, their characteristic shapes. However, drawing the characteristic shapes of kind from the characteristic shapes of many individual horses has to resort to some abstract generalization regarding the similarity of their shapes.

(10) Hansen, 1992, p. 49.

(11) A further etymological analysis, I believe, would reveal this. For example, 'kou' in 'wu-kou-ren' means mouth (each person has one mouth), while 'bei' in 'yi-bei-shui [one cup of water]' means cup which has nothing to do with any feature of water.

(12) Although generic nouns, when being preceded with numerals (such as 'bai-shou' ['the hundred kinds of animal'] and 'wu-gu ' ['the five classes of cereals']), would count kinds rather than individuals and hence cannot be directly fall under the individual-collection paradigm, their naming paradigm can be eventually reduced to the individual-collection paradigm. For, if a generic noun refers to a second-order collection, the kinds which are counted are collections of individuals.

(13) Indeed what the standard mereological analysis requires is a sum principle (If X is horse and Y is horse, then X + Y are horse) rather than a subtraction principle (If X and Y are parts of horse, then X is horse and Y is horse). The points here are these: first, the horse-whole has a significantly different part-whole structure from that of the water-whole; second, perhaps a version of mereology which could do justice to the member-class relation (or to a subtraction principle) is expected in some cases.

(14) Hansen, 1983, p. 54.

(15) David Lewis (1991): Parts of Classes. (Oxford: Basil Blackwell).

(16) Cf., ibid., p. 16.

(17) Cf., ibid., pp. 1, 7, and 73. By definition, a fusion is a sum of some things iff it has all of them as parts and has no part that is distinct from each of them.

(18) When saying that a is one member of the class A, we have such a background-assumption before we assign the membership of the class A to a: the membership of the class A is talked about here; when claiming that {a} is part of the mereological whole A, we do not have the background-assumption aforementioned but explicitly assign the membership of the class A to a by the notation {a}. In this way, to say that a is one member of the class A is just equivalent to saying that {a} is part of the mereological whole A.

(19) Mou, Bo: "An Analysis of the Ideographic Nature and Structure of the Hexagram of Yi-Jing: From the Perspective of Philosophy of Language," Journal of Chinese Philosophy (forthcoming).

(20) Earlier versions of this paper were presented at Conference on Philosophy and Language at the SUNY at Buffalo (March 1996), at a conference on mind, language and matter at Rutgers University (April 1996), and at the ACPA (Association of Chinese Philosophers in America) group meeting session "In What Ways Is Chinese Philosophy Different From Western Philosophy?" at the 1996 Eastern Division meeting of the American Philosophical Association (Atlanta, December 1996). I am grateful to Mark Berman, Rolf Eberle, Carl Gillett, Jingmei Yuan and the audience at the above conferences for their helpful comments and criticism of the paper. I am especially grateful to Professor Chad Hasen for his helpful and stimulating comments and criticism of an earlier draft of this paper. Finally, I am thankful to Yenching Library of Harvard University for providing a grant for my relevant research.