您现在的位置:首页  逻辑新论

THE LOGIC OF NATURAL KIND TERMS

作者:张力锋文章来源:现代逻辑与逻辑应用研究所点击数:10更新时间:2017-04-24



本文刊载于The Philosophical Forum Vol. 45, No.3 (2014): 199-216



THE LOGIC OF NATURAL KIND TERMS



LIFENG ZHANG



     The essentialist thesis of Saul Kripke and Hilary Putnam takes the truth of necessary a posteriori propositions as one of its main premises. The latter requires that both proper names and general names be rigid designators. The rigidity of general names for natural kinds has stimulated much wider debates than that of proper names. Applying the indeterminacy thesis to the whole process of natural kind terms’ reference-fixing, Chenyang Li argues that natural kind terms are not rigid designators and then it is unwarranted that there are necessary a posteriori truths about natural kinds. My aim in this paper is to explain the indeterminacy phenomena in fixing the reference of natural kind terms articulated by Li, to argue that natural kind terms are rigid in a relative sense, and to elaborate their rigidity mechanism. In section I, I will examine Li’s argument in detail and present his conventionalist conception of natural kind terms’ reference. Then in section II, I will clarify the indeterminacy thesis and argue in what sense terms are determinate. Finally in section III, I will illuminate the rigidity mechanism of natural kind terms and defend an updated version of the causal theory.

 I. DOUBTS ABOUT THE RIGIDITY OF NATURAL KIND TERMS: A TYPICAL OBJECTION

     Some philosophers have studied natural kind terms and tried to undermine the plausibility of Kripke-Putnam essentialist thesis for natural kinds, by extending the notable Quinean thesis of indeterminacy, and hence denying the possibility of necessary a posteriori propositions representing the essences of natural kinds.

 Chenyang Li is one such case. He argues that natural kind terms are indeterminate in reference. “In naming a natural kind,” he writes, “[…] there is always […] an inevitable indeterminacy as to what is named.” His case is that “[…] this feature of natural-kind naming determines that true identity statements of natural kinds can never express necessary truth.”1 Li’s idea is that naming a natural kind is quite different from naming an individual: people do not have the “concrete” species or genus in front of them. Their only recourse is to pick out some paradigmatic instances and name a kind they represent by ostension.

 But here Li thinks we have a problem: which natural kind exactly do these instances represent? It is well-known that an individual can be a member of a genus but can also belong to a species of the genus in a taxonomic framework. Since both species and genus are natural kinds, an instance can in principle be paradigmatic of several natural kinds. Li takes an example: a particular Red Delicious apple can be an instance of the natural kind Red Delicious apple; meanwhile, it can also be of the natural kind apple. When people point to some Red Delicious apples and name ostensively a natural kind of which these Red Delicious apples are paradigmatic, what kind of thing are they naming?

     To strengthen the power of his argument, Li models himself on Willard Van Orman Quine and extends the naming of natural kinds to the case of radical translation. He conceives of a possible world W in which there is no such thing as a fruit and consequently no corresponding natural kind term or its species name. If a few Red Delicious apples are brought to W, then the people in W will point to them and name that kind of thing “ABC”. But does the natural kind term “ABC” refer to the kind “Red Delicious,” or to the more general kind “apple,” or perhaps to an even broader kind such as “fruit”? This question cannot be answered by the people in world W: when you pass a number of Macintosh apples or pears to the residents in W and ask them whether these things are ABCs, they are not in a position to decide an answer merely by virtue of what they had in their minds at the time of naming. For the Macintosh apples, says Li, pose the following difficulty: “[…] on the one hand, Macintoshes share sufficient similarities with Red Deliciouses to be one kind; on the other hand, they also bear sufficient dissimilarities with Red Deliciouses to be a different kind.”2 Without a criterion for the natural kind ABC, all the W people can do is make a decision or stipulation as to whether a Macintosh is an ABC. Li specifically mentions “this kind of decision is also a decision about what nature or sameness relation is involved and what kind of thing ABCs are.”3

 Li goes even further. The indeterminacy of reference possessed by natural kind terms in cases of radical translation, Li stresses, is actually also what people are confronted with in everyday life. One of his examples is a natural kind term “xiang”, the Chinese counterpart of “elephant”. Presumably, he guesses, “xiang” only designated the Asian elephant initially, and was later also used to refer to the African elephant. He points out that when people already have a natural kind term listed in their vocabulary, if they come across a newly emerging individual, then it is not a necessary decision whether they will regard it as the referent of that natural kind term or not; otherwise for a species, there would be in two distinct languages an exact one-one correspondence between the natural kind terms used to name that species, if any. To illustrate this point, Li invites us to observe two Chinese terms “yan” and “er” and an English term “goose”. From a Chinese perspective,yan ander stand respectively for two quite distinct species natural kinds: wild goose and domestic goose; the “decision” of the English speakers, however, is that they are simply of one genus natural kind: goose. As far as the universal indeterminacy in the naming of natural kinds or the reference of natural kind terms is concerned, Li summarizes the phenomena as follows:  

 Generally speaking, when we come across a new object O that makes us seriously wonder whether it belongs to a kind (which we have already named in the past without further focusing to determine whether objects like O are or are not instances of the kind), we need to decide whether it is of the kind.

     Therefore, the ultimate reference (or the scope) of a natural kind term is pending and hence indeterminate as long as the entire process of naming is not finalized.4

 In other words, according to Li, natural kinds in human cognition are indeterminate forever, and even what could count as a natural kind, to a great extent, is something done contingently by a linguistic community.  

     From the referential indeterminacy of natural kind terms, Li progresses to arguing for the impossibility or unwarrantedness of the necessary a posteriori truths alleged involving natural kind terms. As regard to the identity propositions of natural kinds, their general form is: the natural kindK is a natural kind with the characteristici (call itI). According to Li’s thesis, being natural kind terms, both “K” and “I” are indeterminate in reference. Since neither of them designates the same natural kind rigidly, the identity betweenK andI is not necessary. One such type of identity truth is “Water is H2O”, which is supposed to reveal the essence of water. It should not be a necessary one, suggests Li. In Putnam’s thought experiment of Twin Earth, is XYZ on Twin Earth water or not? Li contends, in order to judge whether XYZ is water, there should be a definitional standard to determine the level at which water is posited in a taxonomic system, which in turn depends partially on whether XYZ is water. Therefore, since the natural kind term “water” is referentially indeterminate as it were, it is up to the linguistic community to decide whether XYZ is water. But according to Li’s previous suggestion, now that such a decision is only something conventional, it is not necessarily the case; thus, it is quite possible that XYZ is water. “Water is H2O”, the necessary a posteriori proposition in the minds of Kripke and Putnam, would be no more necessary. With the conclusion extended to any natural kind term in the ordinary sense, there would be no necessary a posteriori identity proposition of natural kinds in Li’s opinion. Moreover, any proposition representing the essence of a natural kind would not be necessary.  

     On the other hand, concerning some defenses of the thesis of necessary a posteriori truth, Li points out that even granted the intrinsic logic of the Kripkean thesis, no necessary a posteriori truth of natural kind is obtainable. He considers an example from the proponents of Kripke: there is an objection to the effect that if XYZ is indeed water, what his own analysis proves is nothing but that the proposition “water is H2O” is false. According to Kripke’s original idea, however, if the proposition in question is true, then although discovered empirically, it is necessarily so. Consequently, such a counterexample does not constitute a refutation to necessary a posteriori truth. In this situation, claims Li, the discussion of the necessary a posteriori truth of “water is H2O” turns on the truth condition for such a type of proposition. Nevertheless, because of the referential indeterminacy of natural kind terms, he insists, people can never know whether two involved natural kind terms have the same ultimate reference, thus it is impossible for a related true identity proposition to be available. So in light of the above interpretation, the Kripkean thesis would seem to be meaningless, for we could not get access to the alleged necessary a posteriori statements at all. Analogously, given the truth of some propositions concerning natural kinds in an empirical perspective, it is possible for them to be false after all by virtue of the indeterminacy of natural kind terms. As a result, these propositions are not necessary either. With consideration of the two aspects, Li believes that the impossibility of necessary a posteriori truth intrinsically follows anyhow from the Kripkean thesis on a posteriori necessity.

 Because of the alleged impossibility of necessary a posteriori truth of natural kinds, there is no truth about natural kinds’ essences. Any attempt to pursue the essences of natural kinds is doomed to failure, concludes Li.  

 II. AN INQUIRY INTO THE INDETERMINACY THESIS

 Quine’s indeterminacy thesis is perhaps the most celebrated and influential philosophical doctrine during the second half of the 20th century. It is an organic body consisting of three interconnected, inter-supportive components, namely, the indeterminacy of translation, the inscrutability of reference and the underdetermination of scientific theory. Quine illustrates the indeterminacy of translation with the radical translation from the putative jungle language into English. He imagines a scenario in which a field linguist comes to a completely unknown native tribe, studying its language hitherto unknown to the outside world. In Quine’s account of language acquisition, human language understanding begins with observation sentences. This imagined field linguist finds out that whenever in the situation of a rabbit’s running by a native will say such a sentence as “Gavagai!”. With the information in mind, he tries imitating the natives to utter the jungle sentence “Gavagai!” when a rabbit scurries by, and consequently, his linguistic act in that circumstance will prompt the assent of a native speaker to that sentence; whereas, pronouncing “Gavagai!” without the presence of rabbits can only prompt the dissent of the informant from that sentence. Thus, it can be seen that “Gavagai!” is an occasion sentence observable to subjects and the natives can easily reach agreement on utterances of “Gavagai!” in different situations. Since according to Quine, “an observation sentence is an occasion sentence that the members of the community can settle by direct observation to their joint satisfaction,”5 we can judge that “Gavagai!” is a typical observation sentence. It is because of its intersubjective observability that “Gavagai!” gets understandable and translatable, and further is produced its linguistic “meaning”.  

 In the language of English, meanwhile, whenever there is a scene in which a rabbit runs by, the English speakers will usually say “Rabbit!”, which is a one-word sentence. Considering the same situation under which the jungle speakers are prompted to utter “Gavagai!” while the English speakers “Rabbit!”, the field linguist tries to translate the jungle sentence “Gavagai!” into the English sentence “Rabbit!”. The translation practice of the field linguist manifests Quine’s empirical view of language. According to him, linguistics is an empirical discipline, and the meaning of a sentence is simply its empirical content. So the professed linguistic “meaning” is nothing but stimulus meaning. It is the perceptual experiential content that a sentence has as its meaning. What uttering a sentence expresses, conveys and can be understood by others is none other than the sort of stimulus meaning.

 But observation sentences such as “Gavagai!” contain empirical content, i.e. the scene of a rabbit’s scurrying by, which could absolutely motivate the field linguist or any beginner of the jungle language to produce different translations or understandings in the logical sense. For example, notes Quine, the field linguist might be motivated quite well by the scene and represent the empirical content as “Rabbit-hair!”, “Running-rabbit image!” and so on; even the empirical content could be represented as “Rabbit-fly!” in Donald Davidson’s opinion. Along with that, these sentences become the translation candidates of the observation sentence “Gavagai!”: there emerges the indeterminacy of translation. Being the translations of “Gavagai!”, “Rabbit-hair!”, “Running-rabbit image!” and even “Rabbit-fly!” have the same epistemic status as “Rabbit!”; the reason for this is during the course of translation, the conceptual scheme, the habit of thinking, the social-cultural background of the linguistic community he belongs to, the psychological character and so forth occupy a privileged place of preconceptions in the mind of a translator, they are endowed to the understanding of the natives’ language, and consequently, there will be generated different translations or interpretations for the same sentence of the jungle language, varying with the translators.  

 There is no distinction between right and wrong among the different versions of translation for according to Quine, there is no fact of the matter to translation. As people’s language translation or learning starts with observation sentences, or in other words, observation sentences are the foundation of understanding, the translation or understanding of a language seems indeterminate as a whole in Quine’s eyes with the indeterminacy in principle during the course of radical translation of observation sentences. It should be noted that the language here is not limited to alien one but including one’s own native language. We may infer, the process of infants’ learning their mother tongue is virtually similar to that of radical translation, and several children can have their own respective understandings of a same sentence, which could be even logically incompatible with one another but preserve the same stimulus meaning.

 Quine also maintains that as components of sentences terms have inscrutable reference, coupled with the indeterminacy of translation. Since the term “gavagai” is the sole component of the one-word sentence “Gavagai!”, it naturally follows in the logical sense that the reference of the former is indeterminate due to the latter’s indeterminacy of translation: in some modes of translation, “gavagai” designates the kind of thing rabbit; in other modes, it might as well refer to the kind of thing such as rabbit-fur, running-rabbit-image or rabbit-fly. In his late years, Quine even says it is the inscrutability of reference that is his initial motivation to design the thought experiment of radical translation:

 Ironically, indeterminacy of translation in a strong sense was not what I coined the word “Gavagai” to illustrate. Seen as a term, the word illustrated inscrutability of reference […] Translation of “Gavagai” as “(Lo, a) rabbit” is insufficient to fix the reference of “gavagai” as a term; that was the point of the example.6

 In exactly the same circumstances as the indeterminacy of translation is, the inscrutability of reference manifests itself not only in the foreign terms such as jungle words, but also in one’s native terms. In the course of learning its native language, an infant may well assign different referents to a term under the same stimulation of a scene.  

 It is important to remark that even Quine himself realizes that there is a need to request some restriction on the theses of indeterminacy and inscrutability under consideration, viz. in the extreme case of language learning, and that they are not to be extended to the interior of a linguistic community with a specific social-cultural background. Quine argues the relativity of inscrutability of reference via reductio ad absurdum, namely the “ontological relativity” he later elaborates. If it is indeterminate whether the jungle word “gavagai” means the same thing as the English word “rabbit”, then it is either indeterminate whether the jungle word “gavagai” refers to rabbits; if it is indeterminate whether the jungle word “gavagai” refers to rabbits, then it is either indeterminate whether the English-speaking neighbor refers to rabbits with the word “rabbit”. Along this route, Quine infers the absurd conclusion that it is either indeterminate whether he uses the word “rabbit” to refer to anything.  

 I have urged in defense of the behavioral philosophy of language, Dewey's, that the inscrutability of reference is not the inscrutability of a fact; there is no fact of the matter. But if there is really no fact of the matter, then the inscrutability of reference can be brought even closer to home than the neighbor's case; we can apply it to ourselves. If it is to make sense to say even of oneself that one is referring to rabbits and formulas and not to rabbit stages and Gödel numbers, then it should make sense equally to say it of someone else. After all, as Dewey stressed, there is no private language.7

     Such a conclusion is apparently ridiculous:  

 […] we seem to be maneuvering ourselves into the absurd position that there is no difference on any terms, interlinguistic or intralinguistic, objective or subjective, between referring to rabbits and referring to rabbit parts or stages; or between referring to formulas and referring to their Gödel numbers. Surely this is absurd, for it would imply that there is no difference between the rabbit and each of its parts or stages, and no difference between a formula and its Gödel number. Reference would seem now to become nonsense not just in radical translation but at home.8

 Through the above reductio ad absurdum, Quine is actually attacking an absolute view of reference without any background theory. Only in terms of the view of absolute reference, holds the alleged nonsense of reference. He maintains an eliminativist stance toward absolute reference. It is just according to the ordinary conception of absolute reference that there follows the absurd conclusion of no ontological difference among rabbits, undetached rabbit parts and temporal stages of rabbit, hence no such thing as reference: no term refers to anything. From such a somewhat macroscopic theoretic perspective, the inscrutability of reference is not an isolated thesis, which is an important link of the eliminativism toward absolute reference. Being an extensionalist, Quine could not bear the nihilism of reference, and hence will definitely reject the ordinary idea of absolute reference, for the sake of which he proposes the doctrine of the relativity of reference and ontology and substitutes a relative reference concept for the absolute one.  

 According to the conception of relative reference, if people say “gavagai” designates rabbits, this formulation must be relative to some translation manual. Specifically speaking, first translate “gavagai” into the English common noun “rabbit” in light of some translation manual, second get rabbit as the referent of “rabbit” via disquotation, and finally determine the reference of “gavagai” as rabbit. As far as the terms of one’s own native language are concerned, the first step could be skipped and one could directly apply disquotational principle to the determination of their reference. During the course, reference is no more isolated from a linguistic or theoretic framework but always relative to a world theory underdetermined by perceptual experience (namely, the mechanism of disquotational reference) and a bilingual manual of translation as the background; relative to some specific background interpretation or translation manual, reference is no more indeterminate. Scott Soames evaluates this doctrine in perspective:  “the key feature of the doctrine is supposed to be that reference is nonsense, if thought of ‘absolutely,’ but it is not nonsense if understood as relativized to some sort of background theory or language.”9 Quine himself sketches the thesis of referential and ontological relativity incisively as follows:  

 I can now say what ontological relativity is relative to, more succinctly than I did in the lectures, paper, and book of that title. It is relative to a manual of translation. To say that “gavagai” denotes rabbits is to opt for a manual of translation in which “gavagai” is translated as “rabbit,” instead of opting for any of the alternative manuals […] And does the indeterminacy or relativity also extend somehow to the home language? In “Ontological Relativity” I said it did, for the home language can be translated into itself by permutations that depart materially from the mere identity transformation […] But if we choose as our manual of translation the identity transformation, thus taking the home language at face value, the relativity is resolved. Reference is then explicated in disquotational paradigms analogous to Tarski's truth paradigm (section 33); thus “rabbit” denotes rabbits, whatever they are, and “Boston” designates Boston.10

     As I understand, both translation manual and disquotational reference reflect an important character of the semantic notion of reference: the relativeness to a linguistic community, which is underdetermined by the social-cultural and psychological characters of the community the language speakers belong to, say, the linguistic habits. For a stable linguistic community, the experts of translation will always try their best to compose a public recognized translation manual for an alien language in compliance with their own conceptual schema. Conceptual schema comes into being in a long-term social-historical process, with a distinctive feature of culture and psychology. Of course, from a logical point of view, it is quite likely that even based on the same conceptual schema, the translation experts will compile a few translation manuals with great differences, which is the so-called inscrutability of reference Quine’s thought experiment of radical translation is intended to illustrate. Nevertheless, once one of the translation manuals is adopted as the generally acknowledged, we might have to admit that alien words should have their determinate reference in the interior of the community.  

     In terms of the disquotational reference for native languages, it is even more so. “Rabbit” denotes rabbits, “water” designates water, in which there is not any indeterminacy. The specific social, historical and psychological background formed in a long period of time, and the a prior conceptual schema of the linguistic community members the former causes, I suppose, make the word “rabbit” denote the kind of small mammal emerging before the speaker which is good at running, alert and grazing on grass, but not its undetached parts or temporal stages. It refers to the species which it does as a matter of fact. Logically speaking, in spite of the indeterminacy of reference for words in infants’ language learning, they will dissolve this sort of indeterminacy with instructions from their parents and under the influence of the environment of the linguistic community, and take the option of the general reference mechanism adopted by the community, which can be regarded as a special form of translation manual merely with their own native words as items being translated while entities as items translating the former, to understand those words in an intersubjective way. Reference is hence fixed.

     Up till now, the inscrutability of reference has been clarified a little bit more comprehensively. As I take it, the indeterminacy of natural kind terms proposed by Li is just a particular instance of the indeterminacy of reference. Therefore principally, since inscrutability is a production of the notion of absolute reference, which is beyond linguistic communities, emerges in between different linguistic communities and cannot be extended to the interior of one’s native or any linguistic community, the use of a natural kind term in any specific linguistic community is not indeterminate any more, but fixed in reference. Surely, the use of natural kind terms in a community has its own particularity, which is a rather complicated issue and needs further detailed investigations.

 III. THE MEANING AND REFERENCE OF NATURAL KIND TERMS

 As for a comparatively stable linguistic community (of course, a community with a same language forms a cultural community), there is a unified translation manual in which natural kind terms such as “water” and “goose” have context-determinate reference. Being a communicational tool of a group, language inevitably has a social character, which implies that only by the linguistic labor division and cooperation among individuals of the group, can sentences and words get their meanings fixed. As a result of the interaction of such factors as society, culture, history, individual interests and psychology, different linguistic communities will produce different reifications of the same stimulus meaning. Specifically, in the putative original naming ceremony it is random to some extent whether the Chinese people used the natural kind term “pingguo” to designate a species apple which some paradigmatic instances represent, or a broader genus fruit. For it is a consequence of the linguistic users’ choice with such special psychological factors as intentions in the specific context.  

 Li correctly notices the predicament the theory of direct reference is confronted with, and demonstrates the relative arbitrariness of naming natural kinds. The naming of an individual creates a one-one correspondence relationship of direct reference between the individual and its name. Since there is no need to distinguish multiple reference relationships in the sense of logical possibility, even without any knowledge of the proper name, members of the linguistic community could trace back to the initial naming in order to fix its reference. For this reason, according to Kripke, the reference of proper names can get fixed without the medium of meaning; describing properties or states of the object referred to, if any, meaning is neither sufficient nor necessary for determining the reference of a proper name. Proper names have no meaning. In contrast, it is impossible for there to be a straightforward correspondence relation between a natural kind term and its bearer similar to the one between a proper name and the object in question. No matter how many samples there are for a natural kind, they can never exhaust all the elements in the extension of the natural kind term. The paradigmatic instances are not equal to the natural kind itself, and the abstract natural kind cannot be set before the namers like the individuals. Even if this straightforward correspondence relation could obtain for natural kind terms, the theorists of direct reference still face the more serious plight pointed out by Li: from a logical point of view, a natural kind term could be used to name several kinds of things. Thus, only appealing to the naming ceremony for samples is not sufficient to determine the reference of a natural kind term.  

 Despite the blind spot, it does not mean that the theory of direct reference is irretrievable. This blind spot could be overcome after all. The crucial point is how to settle the problem of representativeness of samples or paradigmatic cases. The process for people to know natural kinds is roughly like this: first to know sufficiently numerous relevant individuals, then to contrast, generalize and abstract among the properties or states in question, and finally to group them in classes or categories so as to form the concept of natural kind. Therefore, naming natural kinds is based on specific characterization of properties or states, which is the important ground for cancelling the doubt of referential indeterminacy of natural kind terms. As Timothy McCarthy notes,  

 […] it is not only its paradigms but also certain properties associated with a natural kind term that play a role in fixing its reference. In conjunction with its paradigms, such a set of properties fixes a kind if there is a property exemplified by the paradigms that plays an appropriate role in explaining why they exemplify the properties in the set (or exemplify most of these properties, suitably weighted).11

 Being the intermediate link in fixing reference, such characterization of properties or states certainly constitutes part of a natural kind term’s meaning. Now, what on earth is the meaning of a natural kind term?

 As argued in section II, according to Quine’s thesis of translation indeterminacy, the meaning of a sentence or word is relative to a translation manual; only relative to some specific linguistic community is determinate the meaning, which manifests the conventionality or sociality of languages. Then, what mechanism is the relative determinacy of linguistic meaning grounded in to emerge? Quine does not offer an answer. Meaning is a means to fix reference: Gottlob Frege distinguishes two elements of a linguistic symbol as early as at the beginning of contemporary analytic philosophy,  

 […] it is natural, now, to think of there being connected with a sign (name, combination of words, letter), besides that to which the sign refers, which may be called the reference of the sign, also what I should like to call the sense of the sign, wherein the mode of presentation is contained.12

 In terms of the delineation of standard semantics, the reference of a word is an individual or a class of individuals (n-tuple set of individuals), and the reference of a sentence is a truth-value (truth or falsity), hence the meaning or sense of a word is something by which the individual or the class of individuals is determined as reference, and the meaning or sense of a sentence is the condition under which the truth-value of the sentence is determined. Being words of a special sort, natural kind terms take natural kinds as their reference respectively, so that their meanings are of course the ways in which to determine the natural kinds. According to the traditional description theory, the means to fix the reference of a word is a description or a cluster of descriptions about the properties or states. Whatever individual or class of individuals satisfies the description or the cluster is to be the reference of the word. With meaning construed in this narrow sense, the descriptionalists practically regard natural kind terms as predicates or indefinite descriptions. If we employ such a theory of meaning, then the only way for us to understand words is to make the meanings of words objective in the sense of intersubjectivity. Only in this way could words legally have the function as communicational tools. For this purpose, Frege even tries to substantialize meanings so as to cut the connection between them and psychological activities, which have the distinctive character of subjectivity. However, the cut is too simplified and idealized. Without psychological activities, it is impossible for language users to understand words or capture their meanings. When grasping a descriptive meaning, a language user cannot help but be in a psychological or mental state. Contemporary philosophy of language acknowledges that even with subjective features, such things as mental states could be none the less intersubjective. It won’t have the consequence Frege worries about to take meaning as mental state, i.e. meaning will never be public. As Putnam observes, “for psychological states are 'public' in the sense that different people (and even people in different epochs) can be in the same psychological state”.13

 But is a mental state about a property or state or a cluster of those things in a position to determine uniquely a certain natural kind? For example, people usually have such a stereotype about water as follows: colorless, tasteless, odorless, transparent, thirst-quenching liquid and so on. When one is in a mental state describing the above properties and states, can she use the word “water” to refer to the uniquely determinate natural substance water? Putnam poses the famous Twin Earth modal argument to refute the description theory of natural kind terms. There is a liquid on Twin Earth, extremely similar to water in the aspect of external properties and states. The Twin Earth people also call it “water”. Accordingly, at the time using the natural kind term “water”, the Twin Earth people are in the same mental state as us. The liquid the Twin Earth people call “water”, though, has an inner molecular structure quite different from that of the earth water in reality, whose chemical formula is XYZ, not H2O; it is a natural substance other than the earth water. It is thus clear, in Putnam’s view, that mental states describing properties or states cannot determine uniquely a certain natural kind.  

     Undeniably, meaning is the way to fix reference, but we can never hastily identify it as a description or a cluster of descriptions in the narrow sense. There are many elements participating in the fixing of words’ reference, such as knowledge about the reference, interdependence relation among language users, selection of paradigmatic cases, etc. Therefore, being something understandable principally by all members of a linguistic community, meaning should be seen as a social institution.  

     As far as natural kind terms are concerned, naming is surely the first phase for the formation of reference relation, and must have something to do with meanings of natural kind terms. While some members of a linguistic community dub a natural kind with a term, they are always with some intentions, namely significant signs the natural kind exhibits to be distinctive from other kinds of things. The samples used in the ostensive baptism are picked out as representatives of a certain natural kind just on account of those significant signs they show. As a proverb says, “bird of a feather flocktogether”, the reason why they are grouped into a kind is that they share common attributes or characters, without which it is unlikely to form a conception of kind. The significant signs or stereotype is generally crude and prescientific, perhaps by no means counts as an ultimate unique criterion to distinguish members of the natural kind in question, but nevertheless an important constituent of the meaning. If the least stereotype for a natural kind is not captured, fluent conversations on related topics can hardly be carried out.  

     For instance, in order to be a qualified member of a linguistic community, one needs at least to know that water is normally a kind of fluid or liquid, otherwise she cannot communicate with others smoothly in topics relevant to water if ignorant of any stereotype for water. A is a speaker such that she does not even know that water is a liquid. When she says to B, another qualified member of the linguistic community,  “birds are not water after all, although they have legs”, which seems to implicate in the present context that water has legs, B will certainly protest and indicate the categorical error A has committed: “water” is not a creature with legs, but a liquid.

     What is more, ordinary languages are full of lots of phrases formed out of natural kind terms, such as “coffee-colored”, “minty”, “salty” and “rose-red”. This phenomenon is sufficient to reveal that there are some stereotypes for natural kinds universally accepted by the linguistic community. Yet since, as we know, several samples could share countless common properties or states, why is none other than certain of them singled out to make up the stereotype? I reckon that this is a problem up to particular contexts. When some members of the English community select a few apples as paradigmatic cases and name a kind they represent “apple”, the stereotype meaning of “apple” is determined by their interests or intentions at that moment. If they do that for edible purposes and already have the conception of “fruit”, then such characters as color, luster, moisture content and texture will automatically compose the stereotype. If they perform the linguistic action out of the same motive but with no intention yet to distinguish different fruits, these samples are no more widely representative of the species apple, or in other words, they are partial as paradigmatic cases; even so, people could still abstract “plant fruit”, “rich in moisture content” and “alleviating hunger” as stereotype from all the common properties or states, thus “apple” literally designates the natural kind of fruit. The specific context at the moment of naming determines the stereotype meaning of a natural kind term, which gets richened and revised with the spread of the natural kind term, but the core content of which will not have great changes.

     Owing to the presence of namers’ particular intentions in a specific context, the rough imprecise stereotype plays an important auxiliary role in the naming activity, determining approximately a natural kind the samples are representative of. In the scenario Li describes as a counter-example, whether some members of a linguistic community are dubbing Red Delicious, apple or fruit with “ABC” can be handled properly. If in the naming context the reason why those members select some Red Deliciouses as paradigmatic instances is that they possess the stereotype usually imposed on apple nowadays, then notwithstanding the restricted conditions (for example, there grow no other kinds of apples except Red Delicious in the local area of naming) and hence no universal representativeness of these samples, the word “ABC” still denotes the natural kind apple. With the naming relation (of course, together with the stereotype meaning attached) passed and accepted by the community, people will not “decide” or “stipulate” whether it is an ABC when glimpsing a pear or a Macintosh. On the contrary, they will judge the latter is an ABC while the former not in virtue of the direct reference relation fixed auxiliarily by the stereotype information attached, although either bears sufficient similarities with those Red Delicious samples to the same extent as it does dissimilarities. Another circumstance is that the area where the naming takes place abounds with all kinds of apples. In order to emphasize or highlight the unique taste or other distinct features for Red Delicious to be itself, the namers deliberately select a few Red Deliciouses with the typical properties or states mentioned above as paradigmatic instances. Put differently, there is yielded only one kind of apple in the local region, namely Red Delicious, and what the residents are interested in is the specific natural kind Red Delicious qua which those samples are Red Deliciouses or themselves, not the genus apple which they belong to. In this situation, “ABC” does not designate the genus apple, but the natural kind Red Delicious. Hence spontaneously neither a pear nor a Macintosh is an ABC.  

      It is worthwhile to note that stereotypes stem from the intentions and interests of the linguistic communities, varying with linguistic communities. Consequently, an alleged natural kind in one linguistic community often has no counterpart in another one. In other words, it has not been named yet over there. Li presents several excellent cases in this respect, such as no proper English natural kind terms translatable for the Chinese ones “yan” and “er”, no Chinese natural kind terms manifesting the difference between the English ones “mouse” and “rat”. What these cases illustrate, however, is not the indeterminacy or arbitrariness of natural kind term’s reference. On the contrary, I figure that they show negatively that it is the presence or absence of some stereotype at the moment of naming that leads to and hence accounts for the difference in the cognition of a natural kind. Because the members of the English community did not consider the difference between domestic and wild animals when picking out paradigmatic instances, there is the natural kind goose as a genus alone while no such species asyan ander in this community; because they took into account the analogous difference in this respect of stereotype without noticing the common features between the two sorts of paradigmatic instances, there are two parallel natural kind terms in English: “mouse” and “rat”, whereas corresponds no English natural kind term to the Chinese one “laoshu”, which is a genus name.

 While stereotype meaning is an important component in the semantic network of natural kind term, it is not the whole. On the one hand, it is not sufficient to discern exactly the membership of a natural kind merely by virtue of the stereotype. For instance, we usually take lemons to be yellow acidic fruits with elliptic shape. Generally, we could tell lemons from other fruits by such a stereotype; but as to some particular individuals of lemon, this “criterion” might not function properly and we probably might make wrong judgments: an unmatured cyan lemon will be judged mistakenly not to be a lemon. Therefore, here is Putnam’s remark: “to say that something is a lemon is […] to say that it belongs to a natural kind whose normal members have certain properties; but not to say that it necessarily has those properties itself.”14 On the other hand, as mentioned above, since stereotype is a cluster of external phenomenal features in general, other stuff may possess the stereotype of lemons in some special circumstances, so that be misjudged reversely: some artificial food with most of lemon’s stereotype will be identified erroneously as lemon. With the emergence of a mistaken judgment grounded in stereotype, there need to be people rectifying the mistake. Appealing to the work of the “corrector”, the ordinary members of the linguistic community can socially fix the ultimate reference of natural kind terms. The role of these “correctors” is another important constituent of the reference mechanism of natural kind terms. According to Putnam’s hypothesis,  

 [E]very linguistic community exemplifies the sort of division of linguistic labor just described: that is, possesses at least some terms whose associated ' criteria' are known only to a subset of the speakers who acquire the terms, and whose use by the other speakers depends upon a structured cooperation between them and the speakers in the relevant subsets.15

 The problem at present is whether the experts knowing the recognition criterion in a linguistic community are dogmatists, that is to say, whether this portion of social meaning for a natural kind term distributed to experts is decided arbitrarily. I think the scientific criteria offered by experts in related areas are explanatory and could explain rather sufficiently why things exhibit those stereotypes in theory, hence indispensableessential characteristics of them, not the products of experts’ personalfavors. Let us take lemons for example again. Botanists or horticulturist could satisfactorily explain why those lemon samples exhibit the above stereotype from a view of plant ecology and internal components, hence render a reasonable precise identification criterion, which is easily operable; meanwhile, the criterion with explanatory function is what a lemon has to satisfy and lemon instances lacking most characteristics in the criterion are unimaginable. According to the identification criterion provided by the experts, unmatured cyan lemons and other abnormal lemon individuals can also be taken as members of the natural kind lemon; the reason is that they fit the criterion in fact: presumably, when the normal conditions are fulfilled, such as the requirements of certain growth phase and appropriate weather, they will exhibit the stereotype or the most. Alternatively, according to the same criterion, artificial lemons are excluded from the natural kind, for they do not have some essential characteristics stipulated by the criterion: for example, they are not made of cells at all, or do not possess the most basic characteristics of life.

 The criterion for the extension of a natural kind term the experts of a linguistic community stipulate is essential, different from the stereotype. A stereotype ordinarily stands for the features of appearance, which could be shared by different kinds of individuals in principle. The criterion proposed by experts is a representation of the natural kind’s metaphysical essence and hence the epistemological basis of the belief in the natural kind held by the linguistic community. Therefore, given that an object is not a member of a certain natural kind, even if exhibiting most or all of the stereotypical features, it must be somewhat different from things of that natural kind at the level of the expertized criterion. In ordinary life, we often say some one is essentially a good guy. As an essence in character, good or evil will be bound to manifest itself in great questions about right versus wrong, otherwise the discourse of essence will be vacuous. The identification criterion of a natural kind set by experts is analogous to such fundamental attributes.  

 Take the liquid XYZ on Twin Earth for example. Although XYZ is almost the same as water in the respect of external characteristics, so far as chemical reactions with the function of explaining the stereotype are concerned, they cannot be the same. Water will be resolved into two volumes of hydrogen and one volume of oxygen by electrolysis, which is the extensional criterion that relevant experts set for the natural kind term “water” in some contexts; in contrast, according to the modern theories of chemistry, XYZ cannot have such a chemical change due to its molecular formula; therefore, XYZ is not the same kind of liquid as water on the earth at ordinary temperatures. If the linguistic community urges that the liquid on Twin Earth be called “water” simply because XYZ exhibits almost the same stereotype as water on the earth, with no care about the explanatory extension criterion offered by the experts, then in this case “water” is only a natural kind term in the nominal sense, and what it refers to is not a unified natural kind but a loose type of substance. Consequently, the paradigmatic instances of the alleged water do not share a common metaphysical hidden structure either, and they merely have a nominal essence, without a real essence. Even in such a circumstance, people will usually consciously distinguish these two sorts of liquids, and call them two different kinds of water. Hence, as to the problem whether the posteriori proposition “Water is H2O” is necessary, there are two cases to be considered. In the first case, “water” designates particularly water on the earth, so the proposition in question is actually an abbreviation of “Water on the earth is H2O”, which is obviously necessary. In the second case, “water” refers more widely to the nominal natural kind. Now that there is no real essence governing the stereotype and playing an explanatory role, such a sentence representing its essence as “Water is H2O” is impossible, meaningless and needless.

 In summary, natural kind terms have meanings, which are the modes or ways of fixing the rigid reference of natural kind terms construed in a wide sense. The meanings are distributed to different groups in a linguistic community, including naming (related to such elements as naming contexts and intentions of namers), stereotype meaning that ordinary members have in mind, identification criterion meaning functioning as an explanation of stereotype and grasped only by relevant experts and the like. By means of the complex mechanism of reference, we could rather sanely meet the challenges Li poses to the theory of direct reference concerning natural kind terms.



 I would like to thank Steve Wykstra, who helped me improve my English expression in Section I of this article. Additionally, the financial support for this study came respectively from Jiangsu Provincial 333 High-level Talents Cultivation Project, Jiangsu Overseas Research & Training Program for University Prominent Young & Middle-aged Teachers and Presidents, a research project of the National Social Science Foundation of China (12AZX008) and another one of the Ministry of Education of the People’s Republic of China (11JJD720001).



1 Chenyang Li, “Natural Kinds: Direct Reference, Realism, and the Impossibility of Necessary A Posteriori Truth,” The Review of Metaphysics 47 (1993): 261-76, 262.



2 Chenyang Li, “Natural Kinds: Direct Reference, Realism, and the Impossibility of Necessary A Posteriori Truth,” The Review of Metaphysics 47 (1993): 261-76, 267.



3 Chenyang Li, “Natural Kinds: Direct Reference, Realism, and the Impossibility of Necessary A Posteriori Truth,” The Review of Metaphysics 47 (1993): 261-76, 268.



4 Chenyang Li, “Natural Kinds: Direct Reference, Realism, and the Impossibility of Necessary A Posteriori Truth,” The Review of Metaphysics 47 (1993): 261-76, 270.



5 Willard Van Orman Quine, “Three Indeterminacies,” in Perspectives on Quine eds. Robert B. Barrett and Roger F. Gibson (Oxford: Blackwell, 1993): 1-16, 2.



6 Willard Van Orman Quine, “Three Indeterminacies,” in Perspectives on Quine eds. Robert B. Barrett and Roger F. Gibson (Oxford: Blackwell, 1993): 1-16, 6.



7 Willard Van Quine, “Ontological Relativity,” in Willard Van Quine, Ontological Relativity and Other Essays (New York: Columbia University Press, 1969): 26-68, 47.



8 Willard Van Quine, “Ontological Relativity,” in Willard Van Quine, Ontological Relativity and Other Essays (New York: Columbia University Press, 1969): 26-68, 47-8.



9 Scott Soames, “The Determinacy of Translation and the Inscrutability of Reference,” Canadian Journal of Philosophy 29 (1999): 321-70, 347.



10 Willard Van Orman Quine, Pursuit of Truth (Cambridge, MA: Harvard University Press, 1992), 51-2.



11 Timothy McCarthy, Radical Interpretation and Indeterminacy (New York: Oxford University Press, 2002), 128.



12 Gottlob Frege, “On Sense and Reference,” in Translations from the Philosophical Writings of Gottlob Frege, eds. Peter Geach and Max Black (Oxford: Basil Blackwell, 1960): 56-78, 57.



13 Hilary Putnam, “The Meaning of ‘Meaning’,” in Hilary Putnam, Mind, Language and Reality: Philosophical Papers, Vol. 2, (Cambridge: Cambridge University Press, 1975): 215-71, 222.



14 Hilary Putnam, “Is Semantics Possible?” in Hilary Putnam, Mind, Language and Reality: Philosophical Papers, Vol. 2, (Cambridge: Cambridge University Press, 1975): 139-52, 141.



15 Hilary Putnam, “The Meaning of ‘Meaning’,” in Hilary Putnam, Mind, Language and Reality: Philosophical Papers, Vol. 2, (Cambridge: Cambridge University Press, 1975): 215-71, 228.



23