Navigation – Plan du site

Mechanisms for characterizing kinds and classes1

Sandeep Prasada
p. 45-62

Résumés

Cet article examine brièvement les recherches récentes relatives aux mécanismes conceptuels permettant de représenter les connaissances génériques qui sous-tendent et contraignent le langage générique. Il apparaît que nos systèmes conceptuels fournissent deux manières formellement distinctes de représenter les multiplicités – en tant qu’espèces et en tant que classes. Les espèces s’expriment via des catégories nominales alors que les classes s’expriment via des syntagmes nominaux. Par ailleurs, alors que les espèces peuvent être caractérisées par des connexions de sens, des connexions statistiques et des connexions causales, les classes peuvent être caractérisées seulement par ces deux dernières. Nous avançons que les espèces sont représentées via un mécanisme génératif capable, en principe, de générer une infinité dénombrable de représentations. Nous avançons également que la distinction entre espèces et classes rend compte des phénomènes habituellement analysés en termes d’espèces bien établies ou non.

Haut de page

Texte intégral

1. Introduction

  • 1 This work was supported by PSC-CUNY 38, GRTI, and startup grants to the author. It also received in (...)

1It is a remarkable fact about human cognition that despite the fact that we encounter only particular entities, we generally think and speak of them as instances of one or another kind (e.g. a dog, a tree, a table). Furthermore, we are able to think and talk about the kinds as such. Thus the statements in (1-3) are interpreted generically and are understood to characterize various kinds of things.

(1)

Dinosaurs are extinct.

(2)

Ducks lay eggs.

(3)

Mosquitos carry malaria.

2Recent research suggests that children appear to use noun phrases both generically and non-generically to talk about kinds as well as instances of kinds right from the start (Pappas & Gelman, 1998; Gelman & Tardif, 1998; Goldin-Meadow, Mylander & Gelman, 2005; Cimpian & Markman, 2008; Gelman, Goetz, Sarnecka & Flukes, 2008), suggesting the capacity to make generic and non-generic reference are closely tied in development. These simple facts raise a host of important questions concerning the conceptual capacities that underlie and constrain generic language including the following.

(I) What are the characteristics of the mechanism(s) that allow us to think of particular entities as instances of kinds and kinds as such?

(II) How are we able to characterize kinds? Do we have more than one way in which we can characterize kinds? What kinds of representational mechanisms are implicated by different ways of characterizing kinds?

(III) Are all generic noun phrases mapped onto kind representations? Or does the mind have more than one way of representing multiplicities, each of which afford distinct ways of characterizing them?

3This article briefly reviews some recent research that bears on these questions. Collectively, the research uncovers some of the conceptual mechanisms that underlie and constrain generic language. In so doing, the research lends support to the idea that generics give voice to our ways of conceiving of and characterizing kinds (Leslie, 2007, 2008, this volume; Prasada, 2000, 2010; Prasada & Dillingham, 2006, 2009; Gelman, 2003, 2010; Hampton, 2010; this volume).

2. Mechanisms for thinking of things as instance of kinds and kinds as such

4What sort of mechanism underwrites our ability to think of things as instances of a kind? In thinking of something as an instance of a kind, we implicitly think of it as being the same as indefinitely many other things with respect to the kind of thing it is. Some of these other instances may actually exist, while indefinitely many others exist only potentially. Together they constitute the kind. As such, representing something as an instance of a kind appears to require a generative mechanism such as (4) which is capable of generating indefinitely many new representations (5), each of which allows us to think of a thing as a different instance of a kind (Prasada & Dillingham, 2009; Prasada, 2010).

(4)

Ki

(5)

K1 K2 K3

5The fact that, at their core, kind concepts such as dog and tree are generative mechanisms such (4) that are capable, in principle, of generating a discrete infinity of representations has been overlooked by traditional research on conceptual representation. Most research has focused on the conditions under which something is considered to be an instance of a kind without explicating what is involved in representing something as an instance of a kind and thus the mechanisms implicated by such an ability. Traditional approaches propose that in order for something to be considered an instance of a kind, the thing must possess characteristics that (i) are necessary and sufficient for belonging to the kind (definitional approach); or (ii) are similar enough to a prototype which represents the average or ideal characteristics of the kind (prototype approach); or (iii) are explainable via the intuitive theory of the kind (theory-based approach). All of these approaches require a mechanism such as (4) which can generate representations such as (5) which can represent those things that meet the criteria for being an instance of kind proposed by each approach. As such, the mechanisms for representing kinds and characterizing kinds being developed here, and in related work (Prasada & Dillingham, 2006; 2009; Prasada, 2010: Prasada, Hennefield & Otap, 2012) are a necessary complement to traditional theories of conceptual representation and underlie our ability to think of things as instances of kinds as well as kinds as such.

6The form of the mechanisms needed for thinking and talking about kinds and instances of kinds also explains why the ability to make generic and non-generic reference appear together in development. The mechanisms needed for thinking and talking about instances of kinds (5) are generated by the mechanism needed for thinking and talking about kinds (4), which makes implicit reference to instances of kinds. As such, it is natural that the ability to make generic and non-generic reference should appear together in development.

3. Characterizing kinds via principled and statistical connections

7Though mechanisms such as (4) may underlie our capacity to think about kinds, we don’t usually simply think about kinds, but characterize them in one or another way. Given that a kind is constituted by indefinitely many instances, and any one of us has experience with only a limited number of instances of a kind, characterizing a kind requires that we come to see the connection between the kind and a property as being non-accidental and thus extendible to indefinitely many instances that have yet to be encountered (Goodman, 1955). The question then becomes whether our conceptual systems distinguish more than one form of non-accidentality (Greenberg, this volume), and thus provide more than one basis for characterizing kinds.

8It has long been noted that in certain cases, there appears to be a non-accidental statistical connection between a kind and a property (6), whereas other cases (7) appear to involve some sort of non-accidental non-statistical connection between the kind and the property that has variously been characterized as being a definitional, essential, or analytic connection (Lawler, 1973; Dahl, 1975; Burton-Roberts, 1977; Carlson, 1977, 1995; Krifka et al., 1995; Cohen, 2001).

(6)

Tires are black.

(7)

Dogs are four-legged.

9More recently, Prasada & Dillingham (2006) have sought to investigate and characterize the special non-accidental connection on its own terms rather than to try to assimilate it to notions such as definition, essence, and analyticity that had been developed for other purposes. They dubbed the special connections principled connections and identified principled connections as involving properties that instances of kinds are understood to have in virtue of being the kinds of things they are (e.g., having four legs for a dog). Properties with a principled connection to a given kind were dubbed k-properties, whereas properties that had merely a statistical connection to a kind were dubbed t-properties. Principled connections were further characterized as involving an explanatory dimension, a normative dimension, and a statistical dimension.

10The explanatory dimension of principled connections is evident in the fact that it seems perfectly natural to respond to a question such as (8) which seeks an explanation for the presence of a k-property, by citing the kind of thing it is (9). Such responses are not possible where a mere statistical connection holds between the kind and the property (10)-(11).

(8)

Why does that have four legs (pointing to a dog)?

(9)

Because it is a dog.

(10)

Why is that black (pointing to a tire)?

(11)

#Because it is a tire.

11The explanatory dimension of principled connections is a consequence of the fact that common sense concepts are not merely devices for sorting things into categories, but instead provide ways of understanding what things are and why they have the properties they do (Carey, 1985; Murphy & Medin, 1985; Keil, 1989; Gopnik & Meltzoff, 1997). The mode of formal explanation on display in (8)-(9), whereby a property of a thing is explained by reference to the sort of thing it is was common in ancient science, but fell out of favor as an acceptable form of explanation during the scientific revolution (Moravcsik, 1991; 1995). Nevertheless, the formal mode of explanation appears to be central to our common sense conceptions of things and their properties (Chomsky, 1975; Moravcsik, 1981, 1998; Prasada, 2000, 2005; Prasada & Dillingham, 2006, 2009; Pustejovsky, 1995).

12The normative dimension of principled connections is reflected in the fact that statements such as (12) strike us as true. The properties that things have in virtue of being the kinds of things they are, are properties that we think they should have.

(12)

Dogs, by virtue of being the kind of thing they are, should have four legs.

(13)

#Tires, by virtue of being the kind of thing they are, should be black.

13Though instances of a kind may lack some of their k-properties, such instances are understood to be defective, immature or incomplete in some way (Prasada & Dillingham, 2009). These normative expectations are lacking in cases where there is a mere statistical connection between a kind and a property (13).

14Finally, principled connections have a statistical dimension whereby we expect that instances of a kind will generally possess the properties that have a principled connection to the kind. Consequently, generics such as (7) may be paraphrased as (14).

(14)

In general, dogs have four legs.

(15)

In general, tires are black.

15In this respect, generics involving principled connections are similar to those involving statistical connections and thus (6) may be paraphrased as (15).

16In a series of experiments using stimuli such as those illustrated in (6)-(15), Prasada & Dillingham (2006) provided evidence that principled and statistical connections are distinguished by our conceptual system in the manner described above. Principled connections between a kind and a property involve an explanatory dimension that supports formal explanation, as well as a normative dimension and a statistical dimension, whereas statistical connections lack both the explanatory and normative dimensions. It is important to note that though the examples given above illustrate principled connecitons via an example of a natural kind and statistical connections via an example of an artifact kind, Prasada & Dillingham’s (2006) stimuli included items from a wide range of content domains (natural kinds, artifact kinds, social kinds) with items involving principled and statistical connections in each domain. Principled and statistical connections were found to be distinguished from one another in the same manner in each domain. This suggests that the representation of principled and statistical connections is not limited to items of one or another kind, but may be represented for any kind of thing. Finally, virtually identical results were found for a subset of the stimuli that were matched on the average prevalence of the k- and t-properties, showing that the results could not be due to differences in the prevalence of properties that have a principled or statistical conneciton to a kind.

17In sum, it appears that our conceptual systems provide at least two distinct ways of characterizing kinds—by representing a statistical connection between a kind and a property, and by representing a principled connection between a kind and a property.

4. Mechanisms for representing principled and statistical connections

18Turning to the question of what mechanisms are needed to represent principled and statistical connections, it appears that a simple associative mechanism in which the strength of the association varies as a function of the prevalence of the property in instances of the kind is likely to be adequate for representing statistical connections. As such, no modification of the mechanisms in (4) and (5) is needed in order to represent statistical connections between a kind and a given property.

19A different picture emerges when we consider how principled connections are represented. Unlike statistical connections, which are mathematical in nature and thus are entirely promiscuous and can hold between any two things (e.g. colors and sounds, shapes and sizes, etc…), principled connections are formal connections that are limited to holding between conceptual representations of kinds and properties. Furthermore, it is part and parcel of representing something as an instance of a kind to think that some of its properties are true of it by virtue of its being the kind of thing it is. Consequently, specifying how principled connections are represented will help us more fully specify the mechanisms that underlie our ability to think and talk about kinds and instances of kinds. Ideally, the way principled connections are represented would help explain why they support formal explanation, why they support the types of normative expectations they do, and why most instances of a kind are expected to have their k-properties.

  • 2 The term aspect was left undefined. The intuitive interpretation of the term, meaning roughly part (...)

20The aspect hypothesis does this (Prasada & Dillingham, 2009). According to the hypothesis, representing a principled connection between a kind and a property requires representing the property as one aspect of being that kind of thing.2 The aspect hypothesis helps explain the key characteristics of principled connections in the following way.

21The aspect hypothesis allows us to exploit a part-whole principle to provide formal explanations of k-properties. For example, if having four legs is represented as one aspect of being a dog, then the four-leggedness of any given dog may be explained by citing the fact that it is a dog because the existence of a whole presupposes the existence of its parts, and thus the existence a k-property may be rendered intelligible by identifying the kind of thing of which it is an aspect.

22The aspect hypothesis also affords the possibility of exploiting a principle of perfection or completeness to ground normative expectations concerning the presence of k-properties. If having four legs is understood to be one aspect of being a dog, then a dog should have four-legs, otherwise it would be incomplete and/or have something wrong with it.

23Finally, the aspect hypothesis can explain the statistical aspect of principled connections. Insofar as a given property (e.g. having four legs) is an aspect of being a given kind of thing (e.g. a dog), all instances of the kind are expected to have the property. However, many kinds of things (e.g. dogs), by virtue of being the kinds of things they are, are material entities, and thus interact causally with their environments. In some circumstances, the causal interactions some instances of the kind may happen to be subject to may prevent a k-property from developing or being present in those instance. Because there is no reason to suppose that these special circumstances will generally prevail, instances of the kind are expected to generally have their k-properties.

24The aspect hypothesis thus provides a natural explanation for the key characteristics of principled connections. More importantly, Prasada & Dillingham (2009) report five experiments in support of the hypothesis. As such, we have good reason to think that principled connections between a kind and a property require representing that property as an aspect of being that kind of thing. Does this mean that the concept that represents a k-property of a given kind is represented as a consituent of the kind concept of that kind? So, for example, is the concept yellow represented as a constituent of the concept canary as it is in the phrasal concept yellow canary? Prasada & Dillingham (2009) argue against this possibility for two reasons. First, every token of a phrasal type must have the properties denoted by their constituents (e.g. every yellow canary must be yellow), but instances of a kind can lack their k-properties (e.g. there can be non-yellow (albino) canaries). Second, whereas it is clearly impossible to have the concept yellow canary without having both the concept yellow and the concept canary, one could certainly have the concept canary without having the concept yellow (Fodor, 1998).

25Prasada & Dillingham (2009) point out that the part-whole relation is between the things that are thought about via the concepts for the k-property (e.g. yellow) and the kind (e.g. canary) rather than between the concepts for the k-property and the kind. They propose that a mechanism of the sort shown in (16) is needed for representing principled connections.

(16)

ki a1, a2… pa1

In this representation, the kind of thing something is projects what we may call the kind’s aspect structure a1, a2,… Each element in the aspect structure (e.g. a1) is capable of indicating the sort of property (e.g. color, form of locomotion) that instances of that kind of thing may have by virtue of being that kind of thing. The representation of a principled connection requires co-indexing one of these aspect representations with the concept P that represents a property that has a principled connection to the relevant kind. For example, the principled connection between being a canary and being yellow is represented in the following manner (17).

(17)

canaryi → a1, a2… yellowa1

26Aspect representations, do not, in and of themselves, represent a specific property and can only direct attention to a specific property once it has been co-indexed to a representation that represents a property (P) in and of itself. Furthermore, aspect representations do not have an identity or existence that is independent of the kind representations of which they are a part. As such, they are like parameters or argument positions in argument structures, but unlike constituents which do have an identity and existence that is independent of their being constituents of a phrase or complex concept.

27The aspect hypothesis and its embodiment in the mechanism in (16) suggest that kind representations provide a principle of unity in two senses. First, kind representations unite aspect representations. Each aspect represents part of what is involved in being the relevant kind of thing without exhausting what it is to be that kind of thing. For example, being green is part of being a frog, but not all there is to being a frog. Similarly, for having four legs, ribbiting, being able to hop, and various other aspects of being a frog. Second, kind representations unite the indefinitely many instances that show this form of unity into an abstract unit (the kind). In sum, when we think of something as an instance of a kind, we think of it as possessing an intrinsically general form of unity which is possessed by indefinitely many things which together form an abstract unit (a kind) which is also an object of thought.

28The mechanism in (16) captures these core formal characteristics of our thinking about kinds and instances of kinds. One might wonder whether the formal characteristics attributed to this mechanism may be consistently realized within a formal system? The answer seems to be that they cannot be consistently realized within set theory, but may be consistently realized within category theory (Asher, personal communication, December 5, 2011). This strikes me as a good reason to endorse the call for exploring category theory as the relevant formal tool for capturing the formal aspects of language and cognition (Macnamara, 1994; Magnan & Reyes, 1994; Asher, 2011).

5. Are principled and statistical connections sufficient for characterizing kinds?

29Returning to the question of the means available to our conceptual systems for characterizing kinds, we may ask whether principled and statistical connections are sufficient for characterizing kinds in the ways that we do? Some reasons to question their sufficiency comes from the types of generic knowledge expressed by sentences such as (2) and (3) repeated here as (18) and (19).

(18)

Ducks lay eggs.

(19)

Mosquitos carry malaria.

30It appears that ducks can be characterized as egg-layers (18) despite the fact that less than half of them (i.e. only fertile, mature female ducks lay eggs) actually do so. Examples of this sort, which Leslie (2007, 2008) calls minority characteristic generics seem like generics involving principled connections, however they clearly differ from the generics studied by Prasada & Dillingham (2006, 2009) on the statistical dimension and thus it is unclear if they involve principled connections or some third kind of connection between kinds and properties.

31More problematic still is the type of generalization expressed in (19) in which mosquitoes are characterized as carriers of malaria even though a very small percentage of mosquitoes do so (Cimpian, Brandone, & Gelman, 2010; Leslie, 2007, 2008). Such low prevalence generics have been dubbed striking generics as they seem to always involve very striking (good or bad) properties (Leslie, 2007; 2008). Given that striking generics can be true when a small percentage of instances have the property in question and they also do not seem to involve an expectation that most instances will or should have the property, they suggest a type of connection between kinds and properties that is neither principled nor statistical.

32Prasada, Khemlani, Leslie, & Glucksberg (under revision) tested whether these two classes of generics involve principled connections by investigating if (i) the relevant properties are represented as aspects of being that kind of thing (20); (ii) the relevant properties can receive formal explanations (21), and (iii) we have normative expectations concerning the relevant properties (22).

(20)

Laying eggs is an aspect of being a duck.

(21)

Why does that (pointing to a duck) lay eggs? Because it is a duck and laying eggs is one aspect of being a duck.

(22)

Ducks are supposed to lay eggs.

33Comparing the responses participants gave to the minority characteristic items to those given to items that clearly involved principled and statistical connections, it was clear that the minority characteristic items differ from the statistical connection items in the same manner that the principled connections items did. This suggests that minority characteristic generics such as (18) are represented via principled connections. On the other hand, the striking items patterned just like the statistical connection items and differed from the principled connection items in the same way that the statistical connection items did. This suggests that striking generics clearly do not involve principled connections. The results do not allow one to conclude that striking generics involve statistical connections as the experiment was not designed to test whether this was the case.

34According to Leslie (2007, 2008) and Prasada (2010), the properties that characterize kinds in striking generics have a causal connection to the kind’s material nature. To test this idea, Prasada, Khemlani, Leslie & Glucksberg (under revision) investigated whether the striking items differed from the statistical items on statements such as (23)-(24).

(23)

There is something about Ks that causes them to P.

(24)

The parts and functional organization of Ks cause them to be P.

35As predicted, striking generics received significantly higher ratings on these items than the statistical generics, suggesting that striking generics involve a causal connection between the kind and the property.

36In sum, Prasada, Khemlani, Leslie & Gucksberg (under revision) found evidence that our conceptual systems provide at least three ways of characterizing kinds—via principled connections, via statistical connections, and via causal connections. Are these three ways of characterizing kinds exhaustive? This is obviously an empirical question, however, Prasada et al. provide a reason to think that they may be. If we focus on the formal dimension of a given kind, we attend to the form of unity possessed by instances of the kind and thus to the principled connections displayed by instances of the kind. If, on the other hand, we focus on the material dimension of a given kind by attending to the material constitution of the instances, their causal dispositions to behave in one or another manner in appropriate circumstances come to the fore. Finally, we can charaterize a kind without attending to either the formal or material characteristics of the kind, but simply note the statistical connections that exist between instances of the kind and various properties. The remaining logical option is to simultaneously focus on both the formal and material dimensions of a kind, but there is no reason to think that doing so should involve some new way of characterizing kinds. As such, it appears that principled, causal and statistical connections may exhaust the possibilities for characterizing kinds in and of themselves.

6. Characterizing kinds and classes

37Thus far, we have considered the different ways in which kinds may be characterized, but are there different ways of representing the mutiplicities characterized in generic sentences? The assumption within psychology is that whereas categories may differ in how “kind-like” they are, there is a single way of representing categories and the differences in kind-likeness are due to differences in how coherent and richly structured the members of the categories are (Keil, 1989; Markman, 1989; Murphy & Medin, 1985; Shipley, 1993). Prasada, Hennefield & Otap (2012) refer to this position as the Uniform Representation Hypothesis. A version of this hypothesis also seems to be operative in linguistics, where the distinction between “well-established” and non “well-established” kinds is made because only “well-established” kinds may appear in definite generics (25)-(26) (Krifka et al., 1995), however, it is generally assumed that the distinction between “well-established” and non “well-established” kinds is not a formal distinction (Zamparelli, 2011).

(25)

The Coke bottle is beautiful.

(26)

#The green bottle is beautiful.

38On this view “well-established” and non “well-established” kinds are formally identical, though they differ on at least some of their non-formal characteristics.

39An alternative possibility articulated by Prasada, Hennefield & Otap (2012) is the Distinct Representations Hypothesis according to which our conceptual systems formally distinguish two types of representations of categories, kind representations and class representations, which provide distinct ways of thinking and talking about categories that are not determined solely by their category structures. One reason to think that this is the case comes from Prasada & Dillingham’s (2006) observation that while both (27) and (28) may be paraphrased in a manner that reveals a characterization in terms of statistical connections (27a), (28a), only (27) can be paraphrased in a manner that reveals a characterization in terms of principled connections (27b), (28b).

(27)

Polar bears eat seals.

a. In general, polar bears eat seals.

b. Polar bears, by virtue of being polar bears, eat seals.

(28)

White bears eat seals.

a. In general, white bears eat seals.

b. White bears, by virtue of being white bears, eat seals.

40It appears that despite the fact that polar bears and white bears are practically co-extensional in our experience and thus whatever correlational or causal knowledge we have of polar bears is also true of white bears, we nevertheless think and talk about them differently and characterize them differently. This suggests that the phrasal nominal white bear and the lexical nominal polar bear may be mapped onto formally distinct ways of representing multiplicities.

41Prasada, Hennefield & Otap (2012) report four experiments which suggest that phrasal nominals are mapped onto class representations whereas lexical nominals can be mapped onto kind representations and that class and kind representations are formally distinguished in a number of ways. Most important for present purposes, was the finding that participants endorsed the claim that the multiplicity picked out by lexical nominals constitute a single kind (29), but this is not the case for phrasal nominals (30). Conversely, they rejected the possibility that the instances that a lexical nominal picks out need not constitute a single kind (31), but allowed that this may be the case for phrasal nominals (32).

(29)

Sunflowers are a single kind of flower.

(30)

#Decorative flowers are a single kind of flower.

(31)

#Sunflowers are not necessarily a single kind of flower.

(32)

Decorative flowers are not necessarily a single kind of flower.

42This pattern of results suggests that lexical nominals are mapped onto kind representations, as it is an intrinsic property of kind representations that the things that they apply to constitute a single kind. Class representations, on the other hand, only specify the description that members of a class must satisfy, and thus leave open the possibility that instances of different kinds can satisfy the same description and thus be members of the same class.

43This difference between the representations of kind and class concepts that underlie lexical nominals and phrasal nominals provides an explanation for why generics that characterize a class (28) can be understood as providing a statistical characterization of the class (28a), but do not support characterization via principled connections (28b), whereas generics that characterize a kind (27) support both types of characterization (27 a,b). Members of a class (e.g. white bears) cannot be assumed to be a single kind of thing and thus to share a single form of unity which may be characterized by reference to aspects of being that kind of thing. This suggests that the mechanism needed for thinking about classes has the form in (33).

(33)

[white [beari → a1, a2…]]

44Whereas the kind representation projects an aspect structure, the same is not true for the class concept white bear. The reason why it does not project an aspect structure is that it cannot be assumed that white bears constitute a single kind and thus have a single form of unity which may be characterized by reference to its aspects. Statistical characterization remains a possibility as this mode of characterizing makes no assumptions about the formal or material properties of the members of the class.

45A similar explanation can be given for the contrast between (25) and (26). A single kind of bottle is being characterized in (25) whereas the phrasal nominal green bottle does not allow one to assume that a single kind of bottle is being characterized in (26) and thus there is no way to decide to what, if anything, the predicate beautiful applies. If, on the other hand, a bare plural is used (34), then the class of green bottles may be characterized statistically even if the class consists of many kinds of bottles.

(34)

Green bottles are beautiful.

46If this is correct, then the contrast between (25) and (26) may be explained without recourse to the distinction between “well-established” and non “well-established” kinds. Instead, the contrast relies on two formally distinct ways of representing multiplicities provided by our conceptual systems, as kinds and as classes.

47As discussed above, both kinds and classes allow statistical characterization, but they differ in that kinds allow characterization via principled connections, but classes do not. This raises the question as to whether classes, like kinds, allow characterization via causal connections? The answer appears to be that classes do allow characterization via causal connections with the conditions that specify membership in the class playing the role of additional conditions in the causal context. For example, though there is no causal connection between celery and softness, there is a causal connection between boiling, celery and softness (35).

(35)

Boiled celery is soft.

48In sum, kind and class representations differ in that the former, but not the latter, allow characterization via principled connections. Linguistically, kind representations are given voice via lexical nominals, whereas class representations are given voice via phrasal nominals.

49In addition to the differences between kind and class representations discussed above, Prasada, Hennefield & Otap (2012) identify the following differences. Kinds are understood to be intrinsically general and thus are constituted by indefinitely many instances, whereas classes may be finite. For example, the books that are on my bookshelf at this moment constitute a perfectly good class. A related difference is that kind representations require a modal distinction between actually existing members and the indefinitely many potentially existing members, whereas no such distinction is implicated by class representations. To represent a class, we construct a description that specifies the conditions that members of the class must meet in order to belong to the class. This requires no assumptions as to whether anything exists that meets the description, nor do we posit the existence of the class itself. In contrast, when we represent a kind, we posit the existence of a kind constituted by indefinitely many instances, some of which may actually exist, whereas indefinitely many others exist only potentially. Kind representations also do not specify conditions that must be met in order for something to be an instance of the kind. Instead, they specify properties (k-properties) that have a principled connection to the kind and thus are understood to be aspects of being that kind of thing (Prasada & Dillingham, 2006; 2009). As such, kind and class representations embody different perspectives from which we can represent and reason about multiplicities and their properties.

7. Conclusions and future directions

50Until recently, there had been little effort to characterize the formal dimension of common sense concepts. The research reviewed here suggests that investigation into the formal dimension of common sense conception is likely to yield interesting insights into the language-conceptual system interface. An important question for future research is to develop a theory of the types of aspect structures projected by kind concepts in different domains. Though the focus here has been on how conceptual mechanisms may underlie and constrain generic language, the finding that kind representations are formally distinct from class representations and that the former are given voice via lexical nominals whereas the latter are given voice via phrasal nominals is potentially relevant to other linguistic phenomena where a sharp distinction is found between the behavior of lexical items and phrasal expressions. For example, the interpretation of so called weak definite NPs (Carlson, Sussman, Klein & Tannenhaus, 2006), or bare singulars in English (Burns & Soja, 2000; Stvan, 1998).

Haut de page

Bibliographie

Des DOI (Digital Object Identifier) sont automatiquement ajoutés aux références par Bilbo, l'outil d'annotation bibliographique d'OpenEdition.
Les utilisateurs des institutions abonnées à l'un des programmes freemium d'OpenEdition peuvent télécharger les références bibliographiques pour lesquelles Bilbo a trouvé un DOI.
Format
APA
MLA
Chicago
Le service d'export bibliographique est disponible pour les institutions qui ont souscrit à un des programmes freemium d'OpenEdition.
Si vous souhaitez que votre institution souscrive à l'un des programmes freemium d'OpenEdition et bénéficie de ses services, écrivez à : access@openedition.org.
Format
APA
MLA
Chicago
Le service d'export bibliographique est disponible pour les institutions qui ont souscrit à un des programmes freemium d'OpenEdition.
Si vous souhaitez que votre institution souscrive à l'un des programmes freemium d'OpenEdition et bénéficie de ses services, écrivez à : access@openedition.org.

Asher, Nicholas (2011). Lexical Meaning in Context: A Web of Words. Cambridge: Cambridge University Press.
DOI : 10.1017/CBO9780511793936

Format
APA
MLA
Chicago
Le service d'export bibliographique est disponible pour les institutions qui ont souscrit à un des programmes freemium d'OpenEdition.
Si vous souhaitez que votre institution souscrive à l'un des programmes freemium d'OpenEdition et bénéficie de ses services, écrivez à : access@openedition.org.

Burns, Tracey C.; Soja, Nancy N. (2000). Children’s acquisition of NP-type nouns: Evidence for semantic constraints on productivity. Language and cognitive processes 15: 45-85.
DOI : 10.1080/016909600386110

Format
APA
MLA
Chicago
Le service d'export bibliographique est disponible pour les institutions qui ont souscrit à un des programmes freemium d'OpenEdition.
Si vous souhaitez que votre institution souscrive à l'un des programmes freemium d'OpenEdition et bénéficie de ses services, écrivez à : access@openedition.org.

Burton-Roberts, Noel (1977). Generic sentences and analyticity. Studies in Language 1: 155-196.
DOI : 10.1075/sl.1.2.02bur

Carey, Susan (1985). Conceptual Change in Childhood. Cambridge, MA: MIT Press.

Carlson, Greg N. (1977). Reference to Kinds in English. PhD. Dissertation. University of Massachusetts, Amherst. Published 1980 by Garland Press, New York.

Carlson, Greg N.; Sussman, Rachel.; Klein, Natalie.; Tanenhaus, Michael (2006). Weak definite NPs. In Davis, C.; Deal, A. R. & Zabbal, Y. (eds.) Proceedings of NELS 36: 179-198. UMass/Amherst: GLSA/Chicago.

Chomsky, Noam (1975). Reflections on Language. New York: Random House.

Cimpian, Andrei; Markman, Ellen M. (2009). Information learned from generic language becomes central to children’s biological concepts: Evidence from their open-ended explanations. Cognition 113/1: 14-25.

Format
APA
MLA
Chicago
Le service d'export bibliographique est disponible pour les institutions qui ont souscrit à un des programmes freemium d'OpenEdition.
Si vous souhaitez que votre institution souscrive à l'un des programmes freemium d'OpenEdition et bénéficie de ses services, écrivez à : access@openedition.org.

Cimpian, Andrei; Brandone, Amanda C.; Gelman, Susan A. (2010). Generic statements require little evidence for acceptance but have powerful implications. Cognitive Science 34/8: 1452-1482.
DOI : 10.1111/j.1551-6709.2010.01126.x

Format
APA
MLA
Chicago
Le service d'export bibliographique est disponible pour les institutions qui ont souscrit à un des programmes freemium d'OpenEdition.
Si vous souhaitez que votre institution souscrive à l'un des programmes freemium d'OpenEdition et bénéficie de ses services, écrivez à : access@openedition.org.

Cohen, Ariel (2001). On the generic use of indefinite singulars. Journal of Semantics 18: 183-209.
DOI : 10.1093/jos/18.3.183

Dahl, Östen (1975). On generics. In Kennen, E. (ed.) Formal Semantics of Natural Language: 99-111. Cambridge: Cambridge University Press.

Fodor, Jerry A. (1998). Concepts: Where Cognitive Science Went Wrong. New York: Oxford University Press.

Gelman, Susan A. (2003). The Essential Child: Origins of Essentialism in Everyday Thought. London: Oxford University Press.

Format
APA
MLA
Chicago
Le service d'export bibliographique est disponible pour les institutions qui ont souscrit à un des programmes freemium d'OpenEdition.
Si vous souhaitez que votre institution souscrive à l'un des programmes freemium d'OpenEdition et bénéficie de ses services, écrivez à : access@openedition.org.

Gelman, Susan A. (2010). Generics as a window onto young children’s concepts. In Pelletier, F. J. (ed.) Kinds, Things, and Stuff: The Cognitive Side of Generics and Mass Terms. (New Directions in Cognitive Science v. 12.): 100-121. New York: Oxford University Press.
DOI : 10.1093/acprof:oso/9780195382891.003.0006

Gelman, Susan A.; Tardif, Twila Z. (1998). Generic noun phrases in English and Mandarin: An examination of child-directed speech. Cognition 66: 215-248.

Format
APA
MLA
Chicago
Le service d'export bibliographique est disponible pour les institutions qui ont souscrit à un des programmes freemium d'OpenEdition.
Si vous souhaitez que votre institution souscrive à l'un des programmes freemium d'OpenEdition et bénéficie de ses services, écrivez à : access@openedition.org.

Gelman, Susan A.; Goetz, Peggy J.; Sarnecka, Barbara S.; Flukes, J. (2008). Generic language in parent-child conversations. Language Learning and Development 4: 1-31.
DOI : 10.1080/15475440701542625

Format
APA
MLA
Chicago
Le service d'export bibliographique est disponible pour les institutions qui ont souscrit à un des programmes freemium d'OpenEdition.
Si vous souhaitez que votre institution souscrive à l'un des programmes freemium d'OpenEdition et bénéficie de ses services, écrivez à : access@openedition.org.

Goldin-Meadow, Susan; Gelman, Susan A.; Mylander, Carolyn (2005). Expressing generic concepts with and without a language model. Cognition 96: 109-126.
DOI : 10.1016/j.cognition.2004.07.003

Gopnik, Alison; Meltzoff, Andrew N. (1997). Words, Thoughts, and Theories. Cambridge, MA: MIT Press.

Greenberg, Yael (this volume). Genericity and (non)accidentalness.

Format
APA
MLA
Chicago
Le service d'export bibliographique est disponible pour les institutions qui ont souscrit à un des programmes freemium d'OpenEdition.
Si vous souhaitez que votre institution souscrive à l'un des programmes freemium d'OpenEdition et bénéficie de ses services, écrivez à : access@openedition.org.

Hampton, James A. (2010). Stability in concepts and evaluating the truth of generic statements. In Pelletier, F. J. (ed.) Kinds, Things, and Stuff: The Cognitive Side of Generics and Mass Terms. (New Directions in Cognitive Science v. 12.): 80-99. New York: Oxford University Press.
DOI : 10.1093/acprof:oso/9780195382891.003.0005

Hampton, James A. (this volume). Generics as reflecting conceptual knowledge.

Keil, Frank C. (1989). Concepts, kinds, and Conceptual Development. Cambridge, MA: MIT Press.

Krifka, Manfred; Pelletier, Francis Jeffrey; Carlson, Greg N.; ter Meulen, Alice; Link, Godehard; Chierchia, Gennaro (1995). Genericity: An introduction. In Carlson, G. N. & Pelletier, F. J. (eds.) The Generic Book: 1-124. Chicago: University of Chicago Press.

Lawler, John (1973). Studies in English Generics. University of Michigan Papers in Linguistics 1:1. Ann Arbor, MI: University Press.

Format
APA
MLA
Chicago
Le service d'export bibliographique est disponible pour les institutions qui ont souscrit à un des programmes freemium d'OpenEdition.
Si vous souhaitez que votre institution souscrive à l'un des programmes freemium d'OpenEdition et bénéficie de ses services, écrivez à : access@openedition.org.

Leslie, Sara-Jane (2007). Generics and the structure of the mind. Philosophical Perspectives 21: 375-403.
DOI : 10.1111/j.1520-8583.2007.00138.x

Format
APA
MLA
Chicago
Le service d'export bibliographique est disponible pour les institutions qui ont souscrit à un des programmes freemium d'OpenEdition.
Si vous souhaitez que votre institution souscrive à l'un des programmes freemium d'OpenEdition et bénéficie de ses services, écrivez à : access@openedition.org.

Leslie, Sara-Jane (2008). Generics: Cognition and acquisition. Philosophical Review 117: 1-47.
DOI : 10.1215/00318108-2007-023

Leslie, Sara-Jane (this volume). Generics as cognitive defaults.

Macnamara, John (1994). Logic and cognition. In Manamara, J. & Reyes, G. E. (eds.) The Logical Foundations of Cognition. Vancouver Studies in Philosophy, Volume 4: 11-34. Oxford: Oxford University Press.

Magnan, Francois; Reyes, Gonzalo E. (1994). Category theory as a conceptual tool in the study of cognition. In Manamara, J. & Reyes, G. E. (eds.) The Logical Foundations of Cognition. Vancouver Studies in Philosophy, Volume 4: 57-90. Oxford: Oxford University Press.

Markman, Ellen M. (1989). Categorization and Naming in Children. Cambridge, MA: MIT Press.

Format
APA
MLA
Chicago
Le service d'export bibliographique est disponible pour les institutions qui ont souscrit à un des programmes freemium d'OpenEdition.
Si vous souhaitez que votre institution souscrive à l'un des programmes freemium d'OpenEdition et bénéficie de ses services, écrivez à : access@openedition.org.

Moravcsik, Julius M. E. (1981). How do words get their meanings. Journal of Philosophy 78: 5-24.
DOI : 10.2307/2025394

Moravcsik, Julius M. E. (1998). Meaning, Creativity, and the Partial Inscrutability of the Human Mind. Stanford, CA : CSLI Publications.

Moravcsik, Julius M. E. (1991). What makes reality intelligible? reflections on Aristotle’s theory of Aitia. In Judson, L. (ed.) Aristotle’s Physics: A Collection of Essays: 31-48. Oxford: Oxford University Press.

Moravcsik, Julius M.E. (1995). Philosophic background of Aristotle’s Aitia. In Sim, M. (ed.) The Crossroads of Norm and Nature. Essays on Aristotle’s Ethics and Metaphysics: 237-246. Lanham, MD: Rowman & Littlefield.

Murphy, Gregory L.; Medin, Douglas L. (1985). The role of theories in conceptual coherence. Psychological Review 92: 289-316.

Format
APA
MLA
Chicago
Le service d'export bibliographique est disponible pour les institutions qui ont souscrit à un des programmes freemium d'OpenEdition.
Si vous souhaitez que votre institution souscrive à l'un des programmes freemium d'OpenEdition et bénéficie de ses services, écrivez à : access@openedition.org.

Pappas, Athina; Gelman, Susan A. (1998). Generic noun phrases in mother-child conversations. Journal of Child Language 25: 19-33.
DOI : 10.1017/S0305000997003292

Format
APA
MLA
Chicago
Le service d'export bibliographique est disponible pour les institutions qui ont souscrit à un des programmes freemium d'OpenEdition.
Si vous souhaitez que votre institution souscrive à l'un des programmes freemium d'OpenEdition et bénéficie de ses services, écrivez à : access@openedition.org.

Prasada, Sandeep (2000). Acquiring generic knowledge. Trends in Cognitive Sciences 4: 66-72.
DOI : 10.1016/S1364-6613(99)01429-1

Format
APA
MLA
Chicago
Le service d'export bibliographique est disponible pour les institutions qui ont souscrit à un des programmes freemium d'OpenEdition.
Si vous souhaitez que votre institution souscrive à l'un des programmes freemium d'OpenEdition et bénéficie de ses services, écrivez à : access@openedition.org.

Prasada, Sandeep (2010). Conceptual representations and some forms of genericity. In Pelletier, F. J. (ed.) Kinds, Things, and Stuff: The Cognitive Side of Generics and Mass Terms. (New Directions in Cognitive Science v. 12.): 36-59. New York: Oxford University Press.
DOI : 10.1093/acprof:oso/9780195382891.003.0003

Prasada, Sandeep; Dillingham, Elaine M. (2006). Principled and statistical connections in common sense conception. Cognition 99: 73-112.

Prasada, Sandeep; Dillingham, Elaine M. (2009). Representation of principled connections: A window onto the formal aspect of common sense conception. Cognitive Science 33: 401-448.

Prasada, Sandeep; Khemlani, Sangeet; Leslie, Sara-Jane; Glucksberg, Sam (under review). Conceptual distinctions amongst generics.

Prasada, Sandeep; Salajegheh, Anna; Bowles, Anita; Poeppel, David (2008). Characterizing kinds and instances of kinds: ERP reflections. Language and Cognitive Processes 23: 226-240.

Format
APA
MLA
Chicago
Le service d'export bibliographique est disponible pour les institutions qui ont souscrit à un des programmes freemium d'OpenEdition.
Si vous souhaitez que votre institution souscrive à l'un des programmes freemium d'OpenEdition et bénéficie de ses services, écrivez à : access@openedition.org.

Prasada, Sandeep, Hennefield, Laura, & Otap, Daniel (2012). Conceptual and linguistic representations of kinds and classes. Cognitive Science. DOI: 10.1111/j.1551-6709.2012.01254.x
DOI : 10.1111/j.1551-6709.2012.01254.x

Pustejovsky, James (1995). The Generative Lexicon. Cambridge, MA: MIT Press.

Format
APA
MLA
Chicago
Le service d'export bibliographique est disponible pour les institutions qui ont souscrit à un des programmes freemium d'OpenEdition.
Si vous souhaitez que votre institution souscrive à l'un des programmes freemium d'OpenEdition et bénéficie de ses services, écrivez à : access@openedition.org.

Shipley, Elizabeth F. (1993). Categories, hierarchies, and induction. Psychology of Learning and Motivation 30: 265-301.
DOI : 10.1016/S0079-7421(08)60299-6

Stvan, Laurel (1998). The semantics and pragmatics of bare singular noun phrases. Ph.D. dissertation, Northwestern University.

Zamparelli, Roberto (2011). Building (subkinds). Talk given at the Genius 3 Conference Generics: From Morphology to Cognition. École Normale Supérieure, Paris, December 5-6.

Haut de page

Notes

1 This work was supported by PSC-CUNY 38, GRTI, and startup grants to the author. It also received infrastructure support from RCMI grant RR03037 from the National Center for Research Resources (NIH) to the Gene Center at Hunter College.

2 The term aspect was left undefined. The intuitive interpretation of the term, meaning roughly part as in not all of, is all that the AH, as well as the tests of the AH conducted by Prasada & Dillingham (2009), required.

Haut de page

Pour citer cet article

Référence électronique

Sandeep Prasada, « Mechanisms for characterizing kinds and classes », Recherches linguistiques de Vincennes [En ligne], 41 | 2012, mis en ligne le 01 septembre 2014, consulté le 29 septembre 2016. URL : http://rlv.revues.org/2058 ; DOI : 10.4000/rlv.2058

Haut de page

Auteur

Sandeep Prasada

Department of Psychology/ Program in Linguistics Hunter College/ Graduate Center, CUNY

Haut de page

Droits d’auteur

© Presses universitaires de Vincennes

Haut de page