May 24, 2009

Musings on Chad Hansen’s Logic and Language in Ancient China: Part 2, Chapters 2-5

In my earlier entry, I posed a single criticism against the first chapter of Hansen's book, and only in later chapters do my stronger objections to Hansen's text arise, more specifically, with his denial of certain Western philosophical problems in light of the nature of the ancient (and arguably modern) Chinese lexicon.

The bread and butter of Hansen's text is his defense of what he finds to be the overarching philosophical perspective of ancient Chinese philosophers, which he dubs behavioral nominalism. For the Chinese view, he states, the ancient Chinese philosophical perspective is behavioral "because, in the place of internal mental representations of particulars and properties, the Chinese view of mind (heart-mind) [心], is dynamic; the mind is the ability to discriminate and distinguish 'stuffs' and thereby to guide evaluation and action" (31), and further claims that it is behavioral "because the Chinese philosopher is not committed to any entities other than names and objects. There is no role in Chinese philosophical theories like that played by terms such as meaning, concept, notion, or idea in Western philosophy" (31). Chinese theory of reference, claims Hansen, is really just a one-to-one tie between names (名) and the stuffs (實) to which names refer, forging further and asserting that the ontology of ancient China is mereological. In fact, Hansen spends a great amount of text returning to the mereological nature of Chinese ontology by appealing to the mass noun (or noncount noun) hypothesis, his most popular and controversial view of ancient Chinese language. The challenge for Hansen is to undo what he views as Platonic (non-mereological) presumption in the interpretation of Chinese philosophical texts by appealing to the structure of the ancient Chinese language, to connect the virtually universal prominence of noncount nouns in ancient Chinese noun vocabulary to the rejection of the class-member-to-set manner of predication that inspires the one-many problem, the prompter of Platonic dilemma. "A naming paradigm in conjunction with count nouns, but not mass nouns, generates the traditional one-many problem of philosophy and explains the appeal to Platonism," he writes.

In fact, Hansen is challenging the assumption of abstractionism in all of pre-Han China.

"I want to use these observations about grammar and this hypothesis about the intuitive picture of language and the world appropriate to Chinese grammar to analyze another old question -- does Chinese have abstractions?

"Again, the tides of academic favor seem to have a generational history. The generation which first encountered Chinese thought was struck by the absence of abstract metaphysics and suggested that Chinese language could not form abstractions. The later generation, sensitive to what seemed to them the implicit intolerance of this view, has mostly rejected it. My view is that the question has been misconstrued by both sides. The question is not what language can do, but what theories do do. Classical Chinese philosophical theories had no roles for abstractions.

"I am inclined, in a qualified way, to side with the earlier generation on the question of abstraction (as in a qualified way I supported the later generation on the question of Chinese logic). I would like to argue for the claim that no Chinese philosophical system of the classical period in China was committed to the existence of or had roles for abstract (universal) entities in any of the traditionally important ways that Western semantics, epistemology, ontology, or philosophy of mind had roles for abstractions" (37-38).

Hansen goes through a brief Western history of abstractionism, channeling most of his Chinese antithesis to Plato's impact on ancient and modern Western philosophy, to give some encouraged words to the Englightenment nominalists who unfortunately stuck to the idea of a thing as its intended meaning. He then follows:

"Now my denial of abstraction in China amounts to a denial that there is any similar interlocking set of philosophical theories. I argue that we can satisfactorily interpret Chinese philosophical writings without attributing a philosophical commitment to abstract or mental entities. To know a word is simply to be able to discriminate...If mereological interpretation of the philosophical writings of the period is possible, then given the explanatory significance of the grammar of Chinese nouns we should prefer nominalist interpretations to traditional, Western-style, abstract interpretations" (38-39).

Hansen, at least on the account of at least minimal abstractionism. On page 40, Hansen provides an example (in Wade-Giles), which really can't be appreciated without the traditional Chinese there, as well. It's best for us to pose it as a question:

Take a very, very basic sentence of ancient Chinese, say, "大非小也."

How should we interpret it?

  • "Greatness is not pettiness."
  • "Bigness is not smallness."
  • "Big is not small."
  • "Big is not small."
  • "Small things are not big things."
  • "Being great is not being small."

All of the above? None of the above? That's the hideous game of ancient Chinese translation and interpretation: no markers for use-mention distinction, no articles, no abstract noun suffixes (like the English -ness or -hood), no linguistic characteristics in word order that allow for abstraction (transliteratred, "大非小也," is, "big not-so small [end-sentence]"), no grand formalism that clarifies one part of speech domain from another. We get serialization, topicalization, and even a few other helpful hints on predication of subjects, but that's slim pickings when compared to other ancient languages, or even modern Chinese!

"The absence of a -ness-like suffix does not prove that there was no theory, nor could its presence prove that there was such a theory. In a sense, the absence of any novel conceptual apparatus at all is evidence that there was no such philosophical theory since usually in the spelling out of such a theory one creates linguistic forms. In the absence of of either theory or conceptual apparatus there is no reason to attribute subconscious reference to the nonexistence abstract objects. Borrowing a metaphor from the Tao Te Ching [Daodejing, 道德經], the language and the theory 'give birth to each other'" (41).

Hansen wants interpretation to favor Quinean semantic ascent (40), but his warrant of it is not exactly convincing. A famous quote from the first chapter of the Daodejing, "無名萬物之母," is clearly not semantic ascent because we would not treat 名 as the verb of the sentence. The same with, "父父," by Confucius. These sentences are really just examples of noun phrases that, paired in serialization, presume a sentential completeness. It's just that noun phrases don't seem to make a sentence to the English reader. There was a good review of Hansen's book which undoes some interpretive concerns by reminding that we interpret dual-noun sentences, especially when the terms are identical, by treating the second noun phrase as a predicate, thus giving us, "Without-name is ten thousand things' mother," and, "Father fathers," respectively. Hansen may have given an argument to something a bit better, really an argument for the indeterminacy of part of speech tagging in ancient Chinese, a rather compelling supplement to Quine's indeterminacy of translation in a real-world, full-language context. For Hansen, such a view would promote a mereological preconception of the world, one so strong that it enveloped even the syntax of the language. I'm somewhat confident that serialization pulls enough weight for such a conception of Chinese mind and language.

My most distant departure from Hansen's second chapter is his claim that the one-many problem simply does not occur. No one-many problem? It's strange to me only in that Hansen himself gives the one-many problem under Chinese semantics, and while the "one" and the "many" may differ in a very important respect, the real problem of language that the one-many problem reoutlines, "What does X name really?" (36) is still present in identical form.

"Baby Susie learnes to utter 'doggie' in the presence of Fido (the family dog -- a collie) and the neighbor's German shepherd and a few other occasional mongrels as examples. However, the first time she sees Uncle Harry's Afghan hound, she promptly chirps, 'Doggie!' How did she know? We tend to say she has learned to abstract from particular examples -- learned abstract thinking. She has abstracted from all the particular dogs she had encountered the features common to all dogs. Seeing that the Afghan hound had these features, even though quite different in other respects, she correctly classifies it as a dog. This classification depends on her having an abstract idea.

"Baby Mei-ling, on the other hand, has learned to use the word '狗' for that stuff which she encounters again and again at Uncle Jang's. But the story does not involve any abstracting. Rather one says that she has acquired the ability to distinguish dog-stuff from non-dog-stuff. She is, in effect, not seeing a different object, but a different part of the same stuff. The problem of learning for Mei-ling is how she is able to reidentify the same stuff. But expressing the problem in that way makes us less likely to talk of abstracting properties from different objects. As we shall see (see pp.127-37), the philosophical problem corresponding to the possibility of abstract knowledge generated by the Chinese picture is rather how we can possibly know or love some mereological whole rather than just knowing or loving those parts we encounter in our vicinity" (52).

The problem for Chinese language has merely taken a step forward in this metaphysical jumble. It is plainly arguable from the Platonic view that Mei-ling's understanding of '狗' arises from the temporal instances of her exposure to the same dog, while the ancient Chinese could say of Plato what Hansen has suggested, that all she did was put a label, dog, on a differently perceived "stuff" at some odd time. There's no reason to believe that the one-many problem disappears under this reasoning. Rather, it appears the (pseudo-)problems just dance on different squares.

Perhaps the genius of Chinese and early Analytic tradition is the revelation of the illusion of the one-many problem here as one built on a false dilemma: "Do we get the universal from the particulars, or do we divide the universe into particulars?" The processes are congruent. At the same time that we're dividing the world with our senses, we're also grouping them together with our language. Hence, a term's definition is set by both the way the universe is "cut" and by the way the universe is "grouped." It is a false dilemma to assume that a speaker of a language must gain his knowledge by one means or the other exclusively.

To get the intuitions pumping on the identity of the problem, let's simply pick another word, something much, much more general. How about the word something? Well, do we go from the particulars to the universals, or do we go from the universals to the particulars? In neither case does a member-to-set or mereological metaphysical conjecture help explain the process. Baby Susie points at something and, lo and behold, it is something, and gathers from all of her knowledge that everything is something. Baby Mei-ling points at something and, lo and behold, it is also something, and all she learned was to distinguish it from nothing. The problem for Susie is that she has no reason to form every single thing as something by individual reference, and that's the problem of induction. The problem for Mei-ling is that she has no means of distinguishing something from nothing, and that's Meinong's flaw. The metaphysics on both sides leave explanatory holes that are bigger than the problem it aims to resolve, the exact opposite of an epistemic gain.

For a seond intuition, we may consider the word dog, and then consider what Susie and Mei-ling are really looking at. Let's say the things Susie and Mei-ling are perceiving aren't dogs, but are genetically modified species of a few different sorts, but that eventually, the /dog term is successfully applied to more regular cases of dogs. Did she really ever learn the Form of Dogness or to distinguish dog-stuff from non-dog stuff? My suspicion is to say, "No," and then follow, "but that's irrelevant to the question that the one-many problem purports to aim to solve."

This is the erroneous move I see Hansen making, playing into metaphysics when he need not do so. Hansen can quite plainly accept both manners of language acquirement for whatever language. His defense for mereology would have to be based, instead, on the proximity of ancient (and modern) Chinese to more formal logical syntax, the primitives of which I believe most definitely favor mereological interpretation. Meanings of words are complex, and their meanings are constantly salvaged by the wide nets of disjunctions and revisions given new data, and in light of Duheim dilemma over a similar matter, our understanding of definition (and in large part, determinacy in translation) is all too naive if we force ourselves to assume, Platonically, that some "Form" is built on some intersecting "properties," or mereologically, that the "stuff" is cut up into "parts" and "instances."

Hansen's remaining text is largely defense of his mereological take on linguistic and philosophical considerations of ancient Chinese philosophical texts, starting from Confucian and Daoist sources (Chapter 3), going then to Neo-Mohist sources (Chapter 4), and finally, to Gongsun Longzi's 白馬非馬 (Chapter 5). The commentary is right in broad strokes, in its citations that Chinese philosophy was more geared toward pragmatics and activism over epistemic explanation; that Chinese philosophers worried more about regulative than descriptive use of language (also citing Fingarette's concern with "the force of speech acts" (60)).

Hansen persists with the ideas that Chinese language is highly mereological, that its power of language in ancient China was to "divide or cut the universe of discourse into portions or opposites" (61). I agree with this take of all languages, and I agree from a basis of formal, not natural languages, so don't see this as being particularly Chinese. Anything can only be understood as being the opposite of all of the things that are not it, and by that reasoning, knowing a term is equally a process of eliminating bad conditions leading to appropriate naming as it is a process of absorbing better conditions. Hansen attributes Chinese thinkers, as I also believe personally, the tendency to strong conventionalism, that not only do the sounds and marks for terms arise out of social convention, but that the actual differentiation of referents is dictated largely by social convention, as well.

Most of his positive evidence in favor of noncount nouns in epistemology was found in the following portion:

"Nominalism in the areas of epistemology and philosophy of mind in China is reflected in pre-Han Chinese grammar of propositional attitudes: believes that, thinks that, knows that, and so on. Traditional empiricist Western epistemology rested on a widely articulated theory of mind. The mind was viewed as a container of items called thoughts or ideas which potentially correspond both to words and to objects or states of affairs in the world. That commonsense philosophy of mind is easily correlated with our own grammar of propositional attitudes -- of know and believe. Both verbs take sentences or that-phrases as their 'objects.' Similarly, we take the sentence or proposition to be the 'content' of the belief or knowledge state -- the content of the mind. Hence a mind's believing is just its containing thoughts or ideas. The thoughts are things which, like sentences, may be true or false -- depending on whether or not they correspond to the way the world actually is. This picture of inner, conscious, or mental states suggests that belief is a subjective representation and that knowledge is the representation's corresponding to the way things are. The belief-knowledge distinction is a prominent feature of English grammar.

"Now ancient Chinese has two quite different and grammatically complex expressions which are routinely translated into the propositional attitudes -- chih [zhi] 知 'know' and i wei [yi wei] 以為 'believe'. Because the two expressions were grammatically quite different, Chinese theories of knowledge virtually never used that contrast to formulate skepticism. Instead, knowing was presented as a kind of skill. Think of it as a skill in applying names (discriminating according to community practices). Propositional belief, similarly, was a disposition of a speaker to apply such expressions to objects in a particular way and then to behave in the ways conventionally associated with that predicate; for example, to believe Nixon is evil is to 'evil' (apply the term evil to) Nixon and to vote against him or demonstrate in the streets.

"The grammar of the belief context in Chinese is that of a three-place predicate: x i y wei F (where x is a person, y is an object and F a property), for instance. 'Nancy i [以] John wei [為] the most handsome man in Kansas.' Translators are sometimes careful to replicate this structure yielding the translation, 'x deems y to be F.' The difference between chih and i wei can be represented as analogous to that between a disposition or habit and an acknowledged skill. A skill has a success component; it is done correctly. The disposition may or may not also be a skill. The success element in the meaning of chih is what makes the translation knowledge (versus belief) work. Knowledge, too, has a success or objective component which belief lacks. But the background philosophical account of what this component is in Chinese and Western systems is rather different. The simple word-by-word association hides that underlying structure which shapes the different formulations of skepticism in the two traditions. The Chinese case leads quite naturally to a skepticism based on conceptual relativism where the classical empiricist belief-knowledge distinction led to skepticism of the senses. The 'correctness' or skill in the application of an expression to an object is much more closely dependent on the conventions of a language than is the notion of a correspondence between ideas or thoughts and reality. So the Chinese concept of knolwedge (chih) [zhi 知] is of a skill that is relative to some practice or institution of distinguishing and naming -- that is, to some language" (63-65).

In the Neo-Mohist, chapter, however, Hansen explores the semantic correspondence to a variety of other terms, most pertinently those of idea 意 as counterexample to the claims he makes against abstractionism and of class 類 as counterexample to his denial of Platonic member-to-set relation:

"In the Neo-Mohist chapters, i [yi] 意 'idea' does not, in fact, play the semantic role of explaining the meaning of general terms at all. It is used of images or memory or imagination. The eighteenth century view that meanings are images is a separate assumption which we do not need to attribute to the Mohists. The only direct semantic role of i [yi 意] 'idea' is in explaining the use of sentences for pragmatic communication -- where its translation is 'intent' rather than 'idea.' Lei 類, on the other hand, cannot be the technical equivalent of 'class' since, in the spirit of the Taoist contrast theory of language, the mathematical notion of class is 'born together' with the distinctions of subclass and member. Without any such distinction, lei [類] could only be regarded as a mereological class -- using (as we shall see) only the distinction of part-whole. This conclusion, obviously, is reinforced by the masslike grammar of the nouns called lei-names and the tendency to explain their semantics in the same way one explains the semantics of proper names. We will notice that the use of lei makes it clear that it is mainly used to assert similarity between thing-kinds and only derivatively to describe a thing-kind based on similarity of its parts. Neither term, in its use in the dialectical chapters, contradicts the basic nominalist character of the semantic theory" (113-114).

"I [yi] 意 'idea-image' occurs in several other places in the Canon in various contexts. Graham has noticed that i [yi] 意 'idea-image' is used in connection with hsiang [xiang] 想 'image'. In these contexts, i does seem to be mainly a picture of a remembered or envisioned object. There are other occurrences of a verbal use of i (as in the above quotation) -- roughly, 'to imagine a thing.' It does have an epistemic use, that is, as envisioning, contemplating, and so on, but it is not used in explaining the meaning of words as such. I [yi] never functions as an abstraction from concrete images nor as a general idea corresponding to general terms" (114).

"The term lei 類 'class' is more central to the semantic project in the dialectical chapters. Chmielewski is the main proponent of the view that it means 'class' in the logical or mathematical sense. If that is true, then it is an abstraction called for by the same problems that generate universals for the Platonic tradition. The suggestion has already been rebutted in part by Graham. He observes that objects are never viewed as members of a class, but as, in his words, of a kind or not of a kind. That is close to my own objection: to suggest there is the notion of 'mathematical class' without the notion of 'member' is paradoxical. These are theoretical 'brothers' that, like the Taoist opposites, are 'born together.' The concept of 'class' or 'species' is generated in a theory which also includes 'member' or 'specimen' as distinct from 'subclass' or 'subspecies.' Without these companion concepts (which Chmielewski himself claims are absent), it is impossible to understand how lei 類 could correspond to the mathematical or semantic term class.

"Graham's objection points to another curious feature of the use of lei [類] in Chinese. The concept seems to stand at least one step above 'natural kind.' Thus the Mohists say that ox and horse are t'ung lei [tong lei] 同類 'same lei' but they never say that of white-horse and horse, where we would expect if if lei did correspond even to our ordinary nonmathematical use of kind or class. T'ung lei [tong lei] (same lei) is used to assert a similarity of things in different natural classes, not to represent a relation between objects comprehended by a natural kind term" (116).

Overall, Hansen's behavioral nominalism is quite defensible. However, his defense does not undo similarly Platonic problems, but just considers them from a different perspective from which neither side can claim truth as both bicker on two ends of a common fallacy. For the argument in the presumption of Chinese mereological language to play as both Hansen and I believe it should, I think Hansen would have done much better sticking to his syntactic and semantic artillery, from which his work is much better prepared to drill into the aptness of Chinese to suit modern logical syntax, argue for equal consideration of mereologically gained understandings of terminology, and then shoot holes into the Platonic metaphysical problem from the stance of its fallacy, not the incompleteness of its account. For the linguistic analysis to creep into metaphysics without the full intent of its destruction is, for my tastes and rigor, an unwarranted pacifism.

No comments:

Post a Comment