Langacker, Ronald W. 1999. Losing control: Grammaticization, subjectification, and transparency. In Andreas Blank and Peter Koch, eds., Historical semantics and cog- nition 147–75. Berlin: Mouton de Gruyter. Langacker, Ronald W. 2005. Construction grammars: Cognitive, radical and less so. In Francisco J. Ruiz de Mendoza Iba ´ n ˜ ez, and M. Sandra Pen ˜ a Cervel, eds., Cognitive Linguistics: Internal dynamics and interdisciplinary interaction 101–59. Berlin: Mouton de Gruyter. Levinson, Stephen C. 2001. Covariation between spatial language and cognition, and its implications for language learning. In Melissa Bowerman and Stephen C. Levinson, eds., Language acquisition and conceptual development 566–88. Cambridge: Cambridge University Press. Michaelis, Laura A. 2004. Type shifting in Construction Grammar: An integrated approach to aspectual coercion. Cognitive Linguistics 15: 1–67. Noonan, Michael. 1985. Complementation. In Timothy Shopen, ed., Language typology and syntactic description 2: 42–140. Cambridge: Cambridge University Press. Sanders, Jose ´ . 1994. Perspective in narrative discourse. PhD dissertation, Tilburg University. Sinha, Chris. 1999. Grounding, mapping and acts of meaning. In Theo Janssen and Gisela Redeker, eds., Cognitive linguistics: Foundations, scope, and methodology 223–55. Berlin: Mouton de Gruyter. Slobin, Dan I. 1996. From ‘‘thought and language’’ to ‘‘thinking for speaking.’’ In John J. Gumperz and Stephen C. Levinson, eds., Rethinking linguistic relativity 70–96. Cam- bridge: Cambridge University Press. Stein, Dieter, and Susan Wright, eds., 1995. Subjectivity and subjectivisation: Linguistic perspectives. Cambridge: Cambridge University Press. Sweetser, Eve. 1990. From etymology to pragmatics: Metaphorical and cultural aspects of semantic structure. Cambridge: Cambridge University Press. Talmy, Leonard. 1978. Figure and ground in complex sentences. In Joseph Greenberg, ed., Universals of human language, vol. 4, Syntax 625–49. Stanford, CA: Stanford Uni- versity Press. Talmy, Leonard. 1988. The relation of grammar to cognition. In Brygida Rudzka-Ostyn, ed., Topics in cognitive linguistics 165–205. Amsterdam: John Benjamins. Talmy, Leonard. 2000a. Toward a cognitive semantics. Vol. 1, Concept structuring systems. Cambridge, MA: MIT Press. Talmy, Leonard. 2000b. Toward a cognitive semantics. Vol. 2, Typology and process in concept structuring. Cambridge, MA: MIT Press. Thompson, Sandra A. 2002. ‘Object complements’ and conversation: Towards a realistic account. Studies in Language 26: 125–64. Tomasello, Michael. 1999. The cultural origins of human cognition. Cambridge, MA: Har- vard University Press. Tomasello, Michael, Josep Call, and Brian Hare. 2003a. Chimpanzees understand psy- chological states—the question is which ones and to what extent. Trends in Cognitive Sciences 7: 153–56. Tomasello, Michael, Josep Call, and Brian Hare. 2003b. Chimpanzees versus humans: It’s not that simple. Trends in Cognitive Sciences 7: 239–40. Traugott, Elizabeth Closs. 1989. On the rise of epistemic meanings in English: An example of subjectification in semantic change. Language 65: 31–55. Traugott, Elizabeth Closs. 1995. Subjectification in grammaticalisation. In Dieter Stein and Susan Wright, eds., Subjectivity and subjectivisation: Linguistic perspectives 31–54. Cambridge: Cambridge University Press. 80 arie verhagen Traugott, Elizabeth Closs, and Richard B. Dasher. 2002. Regularity in semantic change. Cambridge: Cambridge University Press. van Hoek, Karen. 1997. Anaphora and conceptual structure. Chicago: University of Chicago Press. van Hoek, Karen. 2003. Pronouns and point of view: Cognitive principles of coreference. In Michael Tomasello, ed., The new psychology of language: Cognitive and functional approaches to language structure, 2: 169–94. Mahwah, NJ: Lawrence Erlbaum. Verhagen, Arie. 1995. Subjectification, syntax, and communication. In Dieter Stein and Susan Wright, eds., Subjectivity and subjectivisation in language 103–28. Cambridge: Cambridge University Press. Verhagen, Arie. 2000. Concession implies causality, though in some other space. In Eli- zabeth Couper-Kuhlen and Bernd Kortmann, eds., Cause—condition—concession— contrast: Cognitive and discourse perspectives 361–80. Berlin: Mouton de Gruyter. Verhagen, Arie. 2001. Subordination and discourse segmentation revisited; or, Why matrix clauses may be more dependent than complements. In Ted Sanders, Joost Schilper- oord, and Wilbert Spooren, eds., Text representation: Linguistic and psycholinguistic aspects 337–57. Amsterdam: John Benjamins. Verhagen, Arie. 2005. Constructions of intersubjectivity: Discourse, syntax, and cognition. Oxford: Oxford University Press. construal and perspectivization 81 chapter 4 SCHEMATICITY david tuggy 1. Introduction One of the most intellectually fertile concepts of Cognitive Grammar has been that of schemas. 1 The aim of this chapter will be to characterize this concept, relate it to some of the other concepts discussed in the surrounding chapters of this book, and illustrate some of the many ways it is used under Cognitive Grammar. Particular attention is given to how it allows Cognitive Grammar to explicate such traditional concepts as polysemy, syntactic categories, rules, analogy, figurative language, head- ship and valence, and composition, in useful and intuitively satisfying ways. These phenomena under other models must be handled by separate mechanisms, but recognizing them as manifestations of schematicity allows Cognitive Grammar to handle them in an integrated manner. The concept in itself is not a novelty attributable to Cognitive Grammar, but some of its applications are, and especially novel is the theoretical unification it affords. In particular, Cognitive Grammar handles, by this single cognitive mech- anism, both phenomena which are linguistic in the strict sense (the categories and generalizations in speakers’ minds which constitute part of languages) and meta- linguistic phenomena (such as the categories linguists use to talk about language and languages). 2. TheNatureofSchematicity 2.1. The Basic Idea The use of the term in Cognitive Grammar has numerous historical roots, 2 but the basic idea is an ancient, commonsensical one. Briefly, a schema is a superordinate concept, one which specifies the basic outline common to several, or many, more specific concepts. The specific concepts, which are called elaborations or instan- tiations or subcases of the schema, fill in that outline in varying, often contrastive ways. Both Langacker’s and Lakoff’s usages of the term have been quite influen- tial in Cognitive Linguistics circles; they will be examined briefly in the next two sections. 2.2. Langacker’s Characterization Langacker considers the ability to generalize, which he equates with the extraction of schemas, to be one of the most central human cognitive capabilities. It involves the recognition of core commonalities, abstracting away from less important (for the cognitive task at hand) details which may differ from one concept or cognitive experience to another. This ability may be operative in any domain or combination of domains of cognition (Langacker 1987a: 132), and it in fact pervades our thought relative to them all. The relationships of schematicity thus established are one of the main kinds of relationships that structure the ‘‘inventory of conventional lin- guistic units’’ which constitutes a language (73–75). 3 The notion of schematicity pertains to level of specificity, i.e. the fineness of detail with which something is characterized; the notion always pertains, primarily if not solely, to precision of specification along one or more parameters, hence to the degree of restriction imposed on possible values along these parameters. A schema is thus abstract relative to its elaborations in the sense of providing less information and being compatible with a wider range of options. The differ- ence is akin to that between representing a structure by plotting it on a fine grid (where even minor features show up) and on a coarse grid (where only gross features are preserved). Our cognitive ability to conceptualize situations at varying levels of schematicity is undeniable. It is manifested, for instance, linguistically in the existence of terms for superordinate as well as subordi- nate terms. The linguistic significance of this ability is hard to overstate. (Langacker 1987a: 132–35) Schemas are constituted as such by virtue of their relationship to their elabo- rations, the specific subcases that give the same information at a higher level of detail. It does not make sense to call a concept a ‘‘schema’’ or say it is ‘‘schematic’’ schematicity 83 except in the context of specific cases relative to which it is abstract or whose information it represents at a coarser level of detail. Similarly, it makes no sense to speak of an ‘‘elaboration’’ except in the context of a concept which is schematic for it. All human concepts are schematic in some degree, abstracting away from the differences in the particular experiences or thoughts on which they are based. 4 They ‘‘allow a range of variation rather than pinning things down to an exact value. Without this inherent imprecision and the flexibility it affords, language could hardly have become a viable instrument of thought and communication’’ (Lan- gacker 1987a: 132–33). The ‘‘terminal nodes’’ or most specific concepts we can ex- press are not different in kind, but only in degree, from the relatively abstract and even very highly abstract concepts with which we think and which we commu- nicate on a day-to-day basis. Since schematicity is a relative matter and all concepts communicated lin- guistically are schematic in some degree, it should not surprise us to find hierar- chies of schematicity, with one concept schematic relative to others, but itself serving as an elaboration of yet more highly schematic concepts. Thus, Langacker gives tall ? over six feet tall ? about six feet five inches tall ? exactly six feet five and one-half inches tall,orthing ? animal ? mammal ? rodent ? squirrel ? ground squirrel,ormove ? locomote ? run ? sprint (1987a: 132–35). 5 An arrow is used to graphically represent the schematicity relationship, with the schema at the tail and its elaboration at the head of the arrow; thus ? can be read as ‘is schematic for’, and / as ‘is an elaboration of’. At each step, alterna- tive elaborations are possible; for instance, instead of locomote above we might have had contract or wave or fall; instead of run we might have had walk or crawl or (purposely) roll; instead of sprint we might have had jog or trot. Note also that schematicity is a ‘‘transitive’’ concept, in the logical sense: A ? B and B ? C logically necessitates that A ? C; thus move ? sprint and ground squirrel / thing. 6 In sum, for Langacker any concept that abstracts away from differences among similar subcases may be properly called a schema. 2.3. Lakoffian ‘‘Image Schemas’’ Lakoff rarely speaks of schemas in this general sense, but often uses the related term ‘‘image schema’’ (which he credits Langacker with helping to elucidate; Lakoff 1987: 68) in a more restricted sense. Image schemas are relatively simple structures that constantly recur in our everyday bodily experience: containers, paths, links, forces, balance, and in various orientations and relations: up-down, front-back, part-whole, center-periphery, etc. These structures [image-schematic together with ‘‘basic- 84 david tuggy level’’] are directly meaningful, first, because they are directly and repeatedly experienced because of the nature of the body and its mode of functioning in our environment. (Lakoff 1987: 267–68) These are certainly schemas in the Langackerian sense, but perhaps the only characteristic necessary for making them so is that they are ‘‘relatively simple.’’ The characteristics that draw Lakoff’s attention are things like their constant recur- rence, their basis in bodily experience and thus their direct meaningfulness, their gestaltish nature (1987: 272), their ‘‘preconceptual structuring’’ (292–93), their uni- versality in human experience (302, 312), and their ubiquity in language use (272), particularly as structuring concepts for the metaphors so central to human un- derstanding (283). Thus, for Lakoff, image schemas are ‘‘central truths’’ (296). Many, doubtless most, Langackerian schemas will not exhibit these qualities to any high degree: there are multitudinous concepts in the minds of speakers of every language under the sun which are simple relative to other concepts but recur relatively rarely, are not based in bodily experience, are not directly meaningful, are limited to one or a few cultures, and so on. They also are well worth investigating and considering—Lakoff agrees, saying of such ‘‘noncentral truths’’ that ‘‘to me, this is the most interesting kind of truth’’ (297). In the rest of this chapter, we will follow the Langackerian definition. This is not to discount in any way the importance of the concepts that Lakoff is exam- ining, the subset of schemas which are in fact experientially basic, directly mean- ingful, and so on. They are in fact the theme of Oakley (this volume, chapter 9). But the commonality of these with all other direct abstractions is significant and worth discussing. 2.4. The Ubiquity of Schematicity As noted above, many particular schemas, under the Langackerian definition, are far from universal. For instance, the concept of an opening, in chess-nuts’ usage, is a schema including king-pawn and queen-pawn openings (along with such less common ones as king’s knight or queen’s rook–pawn openings) and such cross-classifying concepts as gambit; and each of these is schematic over many different families of openings (e.g., queen-pawn opening ? queen’s gambit ? queen’s gambit declined ? cambridge springs defense, etc.), and each of these in turn has many subpatterns over which it schematizes. None of these schemas can be expected to exist in all the world’s languages, much less among all the speakers of all those languages (though, of course, as the culture of chess spreads, they may be expected to spread with it). What is ubiquitous in the world’s languages is this kind of relationship: that is, schematicity itself. Every language will have some concepts which are relatively specific and others which designate the same sort of entity but are less specific as to details. The following discussion will serve to illustrate this contention. schematicity 85 3. Schematicity and Similarity; Full and Partial Schematicity Schematicity relations arise when cognizers compare mental structures and per- ceive similarities between them. The act of comparison is asymmetrical, comparing a target structure to a standard. The resulting judgment of similarity or nonsimi- larity can be thought of as a sort of vector relationship, in which the degree to which the standard can be recognized in the target is a major parameter of magnitude along which different comparisons may differ. The human cognitive apparatus is apparently of such a nature that as this degree approaches complete recognition, the system experiences a state of heightened excitation; we notice (whether con- sciously or not) when the standard’s specifications are entirely preserved in the target concept. It follows for the same reasons that schematicity at a smaller elaborative dis- tance, where the schema has many specifications which are recognized and the target structure adds few details, is likely to be more salient (produce higher ex- citation) than an elaboratively distant schematicity. Thus, for example, rodent ? squirrel will naturally be a more salient schematicity relation than thing ? squirrel. Full schematicity (represented by the previously mentioned solid arrow from standard to target, S ? T) occurs in just the case when all the standard’s features are preserved in the target, that is, when there is 100 percent coincidence. When there is not such full coincidence, where there is omission, contravention, or distortion of the standard’s specifications, some degree of partial schematicity or extension obtains (represented by a dashed-line arrow, S " T). Most comparisons, obviously, yield judgments of partial rather than full schematicity; very many involve so much distortion that there is little reason to talk of even partial sche- maticity. But as they approach the limiting case of full schematicity, their cognitive (and linguistic) importance increases. 7 Nothing prohibits a simultaneous or subsequent converse comparison, taking the erstwhile target as standard and the erstwhile standard as target. When there is partial schematicity in one direction, there may well be the same in the other (A " B and B " A may both obtain). Where there is full schematicity (A ? B), the converse comparison predictably yields partial schematicity (B " A) except in the limiting case of identity or correspondence, where each concept’s specifications are fully exhibited in the other. Thus, since run ? sprint and the two concepts are not identical, it is predictable that a converse comparison will yield the judgment sprint " run: some of sprint’s specifications are omitted from run. As mental comparisons and schematicity judgments of these sorts are repeated, especially repeated saliently (forcefully), in a person’s thinking, they become en- trenched in his or her mind, and their ease of reactivation is thereby enhanced. As usage events that presuppose or even assert them occur, their conventionality is 86 david tuggy established, and they become part of that subset of the person’s cognitive repertoire which constitutes the language he or she shares with other speakers; see the dis- cussion of entrenchment and salience in Schmid (this volume, chapter 5). Consequently, nonlinguistic cognitive structures start to become linguistic as soon as they are used as part of a phonological or semantic structure, that is, the minute language users start to talk about or with them. They are unlike more cen- tral structures only in their lesser degree of entrenchment and/or conventionali- zation. (They cannot be conventionalized without being entrenched, though they can be entrenched without being conventionalized.) Langacker points out (1987a: 372–75) that any act of comparison which yields a judgment of partial schematicity necessarily involves activation of the specifi- cations that the compared entities have in common. To the extent that those specifications form a coherent concept, it will tend to be schematic for both the compared entities, and to the extent that it recurs, and especially as it proves useful in other contexts, it will become cognitively entrenched and conven- tionalized. In this way, the relationship A " B tends strongly to facilitate the establishment of C, the schema subsuming A and B, in the cognitive network which constitutes the language. 8 And, of course, the establishment of C facili- tates its use for communicative purposes, which in turn establishes its conven- tionality and further entrenches it. By repeated occurrences of this sort of pro- cess, quite extensive and complex subnetworks can be built up (see section 4.1, figure 4.3a). Schemas have the important and paradoxical property of being immanent to their elaborations. Since all the specifications of C are, by definition, fulfilled in A and in B, whenever A or B is activated, C is being activated as well. (Any time language users think or speak of a squirrel, they are thinking of or mentioning a rodent,amammal,ananimal, etc.) Thus, the representation in figure 4.1b, where the curved and dashed lines of correspondence indicate identity, is entailed by 4.1a. This renders obvious an awkwardness or inaccuracy of diagrams using the arrow conventions. For analytic purposes, we are representing in separate boxes (using the container metaphor, in fact), on a piece of paper, structures which are not as discrete as the representation might suggest, whose link to each other is much more like identity. It is important not to let this analytic convenience unduly influence the way we understand the relationship. Langacker’s comments in chap- ter 17, page 433, though prompted by the specific issue of polysemy, are apropos for all schematic networks. It is also important to remember that the arrows used in diagrams of this sort are a notational summary over correspondences between the structures involved. A more complete (and potentially more confusing) version of figures 4.1a and 4.1b would be like figure 4.1c, 9 where specifications x, y, z, and q correspond in all three boxes, but A and B have other specifications which are not matched in each other or in C. For many purposes, notably for teasing out the specifics of blending mechanisms (see section 4.12), it may be necessary to attend carefully to those individual correspondences or noncorrespondences. schematicity 87 4. Functions of Schemas in the Cognitive Grammar Model A number of phenomena which other theories treat in quite disparate ways are claimed, within Cognitive Grammar, to be manifestations of schematicity. This fundamental insight of Cognitive Grammar, that all these phenomena are, at bot- tom, the same thing, affords a conceptual unification that is an attractive feature of the model. 4.1. Categorization: ‘‘Classical’’ and Prototype-Based Categories The relationship of schematicity is central to the characterization of categories of any sort in the Cognitive Grammar model. ‘‘Classical’’ categorization (to use Taylor’s 1995: 21–37 term for it) has, since Aristotle’s day, assumed categories with fairly rigid and predictable boundaries, in- cluding all and only those structures which meet their definitions. They are defined either by a single abstract characterization or, in some versions, by a combination of abstract features. Thus, the category man (i.e., human) can be defined as consisting of all and only featherless bipeds, or, nearly equivalently, all and only those entities which exhibit the combination of features [–feathered] and [þbiped]. Figure 4.1. Extension tends to facilitate establishment of schemas 88 david tuggy Such categories can be easily modeled using schemas. All that is necessary is to restrict one’s attention to relations of full schematicity and ignore relations of extension. Thus, in figure 4.2a, the relations of full schematicity from C to A and B mean they are members of the classical category defined by C, while the relations of partial schematicity from C to D and E mean they are not. In figure 4.2b, the two schemas C and C' are the functional equivalents of features: each defines a classical category, and A and B would be in the classical category defined by the overlap of the two categories they define. 10 Classical categories have no gradations of membership: all members have equal claim to their status as such (Taylor 1995: 24). This again can be modeled in, or read off, structures such as those in figure 4.2. However, as Schmid indicates in chapter 5 (this volume), structures are expected under Cognitive Grammar to vary in their degree of salience or cognitive prominence (i.e., the energy with which they occur in the mind, generally closely paralleling the degree of their entrenchment). This parameter of differentiation means that when a category is activated, some mem- bers of it are likely to be more strongly or inevitably activated than others. In our diagrams, we will represent differences of salience by increasing the thickness of the box lines for cases of relatively high salience and by the use of dashed lines for cases of relatively low salience. Thus, in figure 4.1, A is more salient than B, which is more salient than C. Ignoring such differences, as the represen- tation in figure 4.2 does, gives a ‘‘flat’’ structure like that assumed by classical cat- egorizations; including them means that some members are more highly entren- ched, and thus more salient, than others. It is natural for comparisons to be made from what is familiar to what is less so; it is therefore quite common for a strongly entrenched and highly salient concept to anchor many relations of full or partial schematicity. In such cases, this strongly entrenched concept serves as the center of a category constructed on the prototype model (see Lewandowska-Tomaszczyk, this volume, chapter 6). By repeated ap- plications of the process represented in figure 4.1, there may come to be a single schema uniting the whole or a small number of schemas covering large overlapping parts of the category, but they will tend not to be as salient as the prototype and will thus be less important cognitively and linguistically. This will be despite the nat- ural salience that they gain from the fact that the relationships they anchor are Figure 4.2. Classical categories modeled by schemas schematicity 89 . concepts of Cognitive Grammar has been that of schemas. 1 The aim of this chapter will be to characterize this concept, relate it to some of the other concepts discussed in the surrounding chapters of. so on. They are in fact the theme of Oakley (this volume, chapter 9). But the commonality of these with all other direct abstractions is significant and worth discussing. 2.4. The Ubiquity of Schematicity As. expected under Cognitive Grammar to vary in their degree of salience or cognitive prominence (i.e., the energy with which they occur in the mind, generally closely paralleling the degree of their entrenchment).