Thus, for instance, the word can-opener or the phrase Here’s Johnnie! are conventionally entrenched for millions of American English speakers. The sanc- tion they receive from well-entrenched [Object-Process-er]or[Here’s Name!] schemas reinforces their legitimacy as parts of the English language. Those same schemas will also sanction such nonestablished (perhaps more accurately, not-yet- established) structures as beetle-smasher or Here’s Hortense! These structures do not sanction themselves, but the sanction they receive from the schemas qualifies them as acceptable English. For the (presumably novel) word beetle-collector, there is also direct sanction from the elaboratively closer schemas Small.Item-collector and (perhaps) Insect-collector, 23 and partial (indirect) sanction from butterfly- collector and (perhaps) bug-collector and others, making it more strongly sanc- tioned than beetle-smasher would be. Similarly, Here’s Jennie! will receive sig- nificantly more sanction from the established Here’s Johnnie! than will Here’s Hortense!, because the sanction is more nearly (though not fully) direct. This is the mechanism by which Cognitive Grammar accounts for the occur- rence of novel formations. Schematic patterns sanction both established and novel structures, and a novel structure is automatically acceptable to the degree that it directly elaborates a well-established, elaboratively close schema or set of schemas. The sanction it receives ‘‘is [the] measure of [its] well-formedness, i.e. how closely it conforms to linguistic convention’’ (Langacker 1987a: 66). Although a schema in such a case is a kind of rule, it is not the schema but the speaker who, taking ad- vantage of the sanction afforded by that rule, creates the new structure. ‘‘Creative’’ as opposed to ‘‘rule-governed’’ production of new forms will be evidenced by novel structures which depend more on partial than on full sanction, or whose sanction comes only from elaboratively distant sources. In extreme cases, there may even be no clear sanction, and a structure will simply be invented out of the blue and through constant repetition become established. Much more com- monly there is some degree of sanction. If someone were to say Over there’s Her- man! it would be rather odd, because there is not clear sanction for usage of over there as opposed to there or (better) here in a presentational structure. But it would still have some sort of indirect sanction from such structures with here and there.If it were Over there’s Johnnie!, especially if said with the proper intonation and tim- ing, the sanction would be significantly stronger and the usage, though creative and norm-bending, more nearly in line with the norms for English. Of course, if such a structure is used enough, it will become established in its own right and can become a source of sanction for other, even more divergent, structures. 4.6. Analogy under a Schema-Based Model Under many other models, rule-based and analogy-based accounts of linguistic creativity are seen as strict alternatives, theoretically distinct, and relegated to dif- ferent modules of the grammar. Under Cognitive Grammar, the difference between them is one of degree, and the two types may often be simultaneously active. 100 david tuggy For example, the word ink-jet was originally coined to designate a device that squirted droplets of ink onto paper. 24 This was the central and distinctive mech- anism for a particular type of printing device, which was then called an ink-jet printer, often shortened to ink-jet and dehyphenated to inkjet. It contrasted with such other types as dot-matrix, daisy-wheel, and thermal-paper printers and was superior to most of them in its ability to quickly and quietly print graphics- intensive copy on standard paper. Later, the terms deskjet and laserjet were coined on the analogy of inkjet as names for particular brands of printers; 25 a deskjet is actually a kind of inkjet, but a laserjet is not. Assuming that deskjet was the first of the two new formations, 26 we can represent what happened as an extension from the established inkjet to the nonestablished deskjet, with the relationship of partial schematicity mediating sufficient sanction to warrant the new formation. This is represented below in figure 4.10a. It was a ‘‘creative,’’ norm-bending formation, but it caught on and became established. In accordance with what was said in section 3 and diagrammed in figure 4.1, this partial sanction entailed the activation of the specifications common to the two structures, thus facilitating the estab- lishment of the schematic structure (figure 4.10a.i) consisting of those specifi- cations. Also in Figure 4.10a is represented the subsequent extension from this group of structures to the novel laserjet, with the concomitant activation of a new schema for the whole category. This schema (4.10a.ii), which designates a high- tech, graphics-friendly printer with a name Noun-jet, can, to the extent that it becomes established, be used to directly sanction such new formations as Design- Jet, PaintJet, and OfficeJet. Further extensions and schematizations allow Verb-jet formations like ThinkJet and (taking design and paint as verbs) DesignJet and PaintJet, Adjective-jet formations like QuietJet, and names for not-only-printer and nonprinter computer peripherals like CopyJet and ScanJet. These extensions are represented in figure 4.10b. ‘‘Analogy’’ is most clearly to be invoked where there is no preestablished schema to directly sanction the newly coined structure. But the very notion of analogy implies that the ways in which the new structure is analogous, or similar, to the old Figure 4.9. Sanction of established and novel structures schematicity 101 are crucial to the new formation. That is, there must be some basis for the analo- gy; and that basis will constitute an incipiently established schema. Such are the schemas represented in figures 4.10b.iii–vi. To the extent that such schemas become entrenched and begin to participate directly in sanctioning the formation of new structures, the mechanism of rule-based creativity is active. Such is the case of 4.10b.ii (a more entrenched 4.10a.ii) vis-a ` -vis OfficeJet, PaintJet, and DesignJet. But as long as such partially schematic relationships as those from deskjet or laserjet to PaintJet and OfficeJet are also important for establishing the latter, the mech- anism of analogy is also at work, reinforcing the entrenchment of 4.10b.ii in the process. The distinction between a schema-based and an analogy-based account of novel formations ‘‘comes down to whether the schema [rule] has previously been extracted, and whether this has occurred sufficiently often to make it a[n estab- lished] unit.’’ ‘‘If the notion of analogy is made explicit, and if rules are conceived as schemas, there is no substantial difference between analogical and rule-based descriptions. The model therefore achieves a significant conceptual unification’’ (Langacker 1987a: 447). 4.7. Figurative Language Under Cognitive Grammar, figurative usages of language involve the same sorts of structures we have been seeing repeatedly. Thus, metaphors (this volume, chapter 8) and metonymies (this volume, chapter 10) involve extension from a standard (the ‘‘literal’’ sense) to a target (the ‘‘figurative’’ sense). Their whole configuration, including both senses, will constitute the semantic pole of the expression in its figurative usage. Thus, in the cat (is) out of the bag the literal meaning cat out.of bag is extended to mean information out.of concealment, as represented in figure 4.11a (see Langacker 1987a: 93). Figure 4.10. Examples of analogy 102 david tuggy The categories of metaphor and metonymy overlap and grade into each other and often coincide, but the major difference between them in the clear cases is that for metaphors the designated cognitive configuration (the profile in Langacker’s terms) can be seen as holding steady while the cognitive background (base) shifts dramatically, and in metonymy the base holds steady while the profile shifts. This has the result that for metaphors the extraction of a coherent schema with both literal and figurative senses (standard and target) as subcases, as in figure 4.1,is generally possible, whereas for metonymies it is more problematic (see the dis- cussion in the next section). For instance, for most speakers of modern American English, the primary sense of dish is closely synonymous with the primary sense of plate, designating the physical, usually round and slightly concave, object on which food is typically placed for eating, and the sense food prepared in a certain manner is an extension from it. The profiling shifts from the object on which food is placed to the food (bounded in domains of quality and manner of preparation instead of, or typically besides, in space). Figure 4.11b diagrams this: it also provides a couple of additional examples of metaphor, where an activity one excels at or a good-looking girl or woman is called a dish. 4.8. Domains Some concepts expressed in alternative terms are easily reducible to a schema: thus girl/woman in figure 4.11b is intended as a shorthand for (physically mature) female human. It is not similarly easy to conceive of a schema containing the Figure 4.11. Examples of metaphor and metonymy schematicity 103 common essence of plate and food. A common base for the two concepts can be characterized, however, and the dashed-line schema in 4.11b may be taken as identifying such a concept. (Perhaps we might paraphrase it as ‘Thing prominent in the typical scenario of well-prepared food being offered to humans for con- sumption’.) This is problematic, however: at least prototypically, the profiling of subcase and schema must match for there to be full schematicity, and we do not have that in this case. Rather, either the base is devoid of profiling, or it has some sort of alternative profiling. 27 It may, in the end, be a matter of definition, but it seems reasonable to posit that people can, as one of the ‘‘focal adjustments’’ they make to a conceived scene (Langacker 1987a: 116–37), disengage the profiling from any particular entity within the scene. Such a construct in the case of plate and food certainly abstracts from the differences and retains the commonality of the two concepts, which makes it rather difficult to deny that it is a schema with respect to them. Accepting it as such would make it natural for us to view metaphor and metonymy as similar cognitive phenomena, yet the difference in the kind of focal adjustment (despecification of significant parts of the base for metaphor; despecification of the identity of the profile for metonymy) will allow us to appropriately distinguish the two phenom- ena as well. Such conceptual unification of closely related phenomena is surely a desirable result. This sort of schema (if that is the proper name for it) is probably rightly to be equated with the idea of a cognitive domain, which in turn is very close if not identical to what people mean by such terms as ‘‘script,’’ ‘‘scene,’’ ‘‘frame,’’ ‘‘ICM,’’ ‘‘scenario,’’ ‘‘semantic field of potential,’’ or ‘‘mental space’’; for some, this is a, if not the primary, meaning for the term ‘‘schema’’ (see Schank and Abelson 1977; Adams and Collins 1979: 3; Chafe 1987: 29; Cienki, this volume, chapter 7; Fau- connier, this volume, chapter 14). For Langacker at least, domains are ubiquitous: ‘‘Semantic units are characterized relative to cognitive domains, and any concept or knowledge system can function as a domain for this purpose’’ (1987a: 63, 147– 82). This echoes Fillmore’s (1975: 124) statement that a scene can be ‘‘any kind of coherent segment of human beliefs, actions, experiences or imaginings.’’ For other authors, only a subset of concepts, generally highly schematic and ubiquitous in cognition, or otherwise especially prominent, cognitively independent, and so on, may be deemed worthy of being called a domain (or frame, etc.) (see Croft 1993: 337–45). 28 By whatever name, what we are talking about is a coherent set of in- terrelated concepts within which or in relation to which entities may be singled out for profiling. It is a specialized kind of schema with no profiling prespecified. 29 It does not follow that schemas of this type are of high salience. It is difficult in most cases to think of them as meanings of the structures involved. Although I can entertain the schematic concept of the food-served-on-a-plate scenario, devoid of any profiling, it is significantly more difficult to think of it as a meaning of dish. It would seem that such profile-less concepts are difficult to maintain as objects of conception and, as a result, unlikely to be entrenched in specific cases of meton- 104 david tuggy ymy. The partial schematicity relation involved in the same configuration (i.e., in the example, plate " food) suffers under no such disadvantages and is more likely to be well entrenched. The requisite domain or scenario for establishing a metonymy commonly characterizes not the prototypical meaning of a lexical item, but a more elaborate but less prominent subcase. It is only a subcase of bag, and not a very salient one for most people nowadays, that features the notion of ‘bringing game home from the field in a bag’. But that base scenario is the one that underlies the metonymic change to the verb bag meaning ‘to successfully hunt or capture’, or to the noun bag meaning ‘the game’ in such a case. When we speak of domains we use nominal structures (e.g., nouns such as domain, scenario, etc.) which profile the domain as a whole. That is a rather dif- ferent thing from the kind of profile-less concept we have been discussing. It is much like profiling a place: a (profiled) location is not schematic for the things in it. There is a similar difficulty in conceiving of a domain or situation which one has just named and thereby profiled as schematic for the elements which can be located in it. Whether or not profile-less domains are to be considered schemas for the elements in them, profiled domains are not. This is particularly relevant to cases of part-to-whole or whole-to-part me- tonymy (synecdoche or meronymical metonymy), where one of the two concepts involved is largely coextensive with the common base for the two concepts. Thus, in the case of wheels " vehicle, the common base would be a vehicle with its wheels, but it would be a vehicle which is neither profiled itself nor has any subpart profiled. Such a concept could still be claimed to be activated as a schema for the two metonymically related meanings, but the meaning vehicle, which profiles the whole, is a different concept precisely because it is profiled. 4.9. ‘‘Elaboration Sites’’ and Syntactic Coherence Relationships of schematicity are, in the Cognitive Grammar model, important for syntagmatic valences. Always some (sub)structure in one entity is identified with the neighboring entity or a substructure of it. Usually when the whole of the one entity is identified with a substructure of the other, there is a clear relationship of schematicity between the two. The schematically characterized substructure is in such cases called an elaboration site or e-site. Most typically, a central participant in a Relation functions as an e-site for a Thing. Thus, the process ate has as central participants a schematically characterized eater and some schematically charac- terized food. In figure 4.12a, those substructures are identified with, and elaborated by, the Things John Wayne and the toast, respectively. To the extent that an e-site is salient within the meaning of a structure and its elaborative distance from its target (its syntagmatic partner) is great, the structure schematicity 105 containing that e-site is said to depend on its syntagmatic partner. Thus, in figure 4.12a ate depends quite strongly on John Wayne and on the toast. 30 Depen- dence is a central element for characterizing the range of kinds of valences (Lan- gacker 1987a: 298–310). 4.10. Profile Determinance and the Complement-Modifier Distinction Figures 4. 12b and 4.12c represent two possible results of combining the toast with John Wayne ate, 31 preserving the same dependence relationship (of John Wayne ate on the toast). In one case, the composite structure designates the same entity as John Wayne ate, and in the other the same entity as the toast.In Cognitive Grammar terms, John Wayne ate is profile determinant in the one case and the toast is profile determinant in the other. Profile determinance is the major factor in what has traditionally been called headship. 32 Thus, John Wayne ate is the head of the construction in figure 4.12b (the semantic pole of John Wayne ate the toast) and the toast in figure 4.12c (the semantic pole of the toast John Wayne ate). Profile determinance amounts to a schematicity relationship in which the composite structure elaborates the profile determinant component. Figure 4.12. The elaboration of e-sites, profile determinance 106 david tuggy When the profile determinant within a construction depends on its syntag- matic partner(s), as in figure 4.12b, a head-complement construction obtains. Con- versely, when the nonhead depends on the profile determinant, as in figure 4.12c, it is a head-modifier construction (Langacker 1987a: 309–10). Schematicity is thus central to the definitions, within Cognitive Grammar, of these important syntactic notions. 4.11. Schemas and the Component-Composite Relationship The profile determinant, in clear cases, is fully (or very nearly fully) schematic for the structure of which it is a component. It thus sanctions the formation of the composite structure. But nonprofile determinant components also sanction par- ticular substructures or aspects of the composite structure. Thus, in figure 4.13a, which represents in slightly greater detail the same structure as figure 4.12b, the nonhead is represented as sanctioning a subpart of the composite structure. In figure 4.13a, the toast is virtually, if not totally, the same within the com- posite structure as in its solitary state as a component. It is thus quite easy to con- strue it as simply embedded within the structure or as added to its syntagmatic companion to form that structure. This is a very common and a prototypical kind of component-composite relationship, and its predominance is what makes plau- sible the commonly assumed building-block metaphor, which construes complex structures as composed completely and exclusively of the components, much as a brick wall consists entirely and exclusively of bricks. One of the implications of this mental model is that the bricks (e.g., lexical items) and their patterns of integration (syntax) are very difficult to construe as anything but completely different sorts of entities. Typically there are small discrepancies, however. In figure 4.13b(¼ 4.12c), there is such a discrepancy: the structure John Wayne ate comes with a strong ex- pectation that a phonologically subsequent noun phrase will elaborate the concept of the eaten substance; that is, the phrase is strongly transitive, but its counterpart in the composite structure is not. It is for that reason that the relationship between the component structure and its counterpart in the composite structure is re- presented by a dashed rather than a solid-line arrow. This difference, though in a sense minor and quite understandable, even fully expectable, 33 is a very mild case of something that can be seen more clearly in other cases where the independent meaning of a structure differs significantly from its meaning as a component in a composite structure. For instance, the noun toast alone designates (at least for most American English speakers) sliced (and other- wise initially untreated) bread the surface of which has been toasted, that is, browned by being held close to a source of radiant heat. In French toast, the com- posite structure designates sliced bread which has been browned, but only after it has been dipped in a milk-and-egg batter, and the manner of the browning is by schematicity 107 being placed on a hot surface (i.e., the bread is fried rather than toasted). Toast is still usefully identified as the head (profile determinant) of the composite structure, but its relation to the composite meaning is one of partial rather than full sche- maticity. The other component, French, does not correspond clearly to anything at all salient in the composite structure. Most speakers will suppose that this method of preparing bread for eating originated in France, but that is a quite peripheral and even a somewhat doubtful specification. The building-block metaphor does not work very well in such a case, and of course there are even more egregious examples: English horn, for instance, which is a kind of large oboe (one of the least horn-like of wind instruments) and which is not particularly English in origin or distribution, or eavesdrop (see figure 4.13c), which is an action of listening to what is not addressed to one and has nothing obvious to do with eaves or with dropping (though since drop designates a process it is more nearly schematic for the com- posite and thus identifiable as head). 34 Such discrepancies between the components and the composite are entirely un- problematic under the Cognitive Grammar model. Instead of the building-block Figure 4.13. Sanction by components 108 david tuggy model, it is helpful to adopt a scaffolding metaphor: ‘‘component structures are seen as scaffolding erected for the construction of a complex expression’’ (Lan- gacker 1987a: 461). Their structural specifications and modes of integration will generally parallel and suggest the shape of the composite structure, but (even from the beginning and vastly more so as it becomes entrenched in its own right) it exists independently of them and may vary from them in significant ways (Langacker 1987a: 460–64). By accounting for composition in terms of schematicity relations, the Cog- nitive Grammar model handles such variations of compositionality with no fur- ther machinery. The prototype for composite structures has a clear and fully sche- matic relationship from the head to the composite structure and identical or fully schematic relationships from nonhead components to clearly identifiable subparts of the composite structure. The prototypicality of this configuration accounts for the plausibility and pervasiveness of the building-block model. But the existence of such structures does not preclude others where the headship is less easy to deter- mine and where the contributions of the components to the composite structure are difficult to recognize—to the point where it might be posited that they are not components at all. The differences between these kinds of constructions are all matters of degree rather than differences of kind: no new syntactic or lexical ma- chinery is needed to account for the full range of attested types. 4.12. Blends A powerful theoretical tool wielded by many practitioners of Cognitive Linguistics has been the idea of blending mental spaces to achieve a new kind of combined space with emergent properties (see Fauconnier, this volume, chapter 14; Turner, this volume, chapter 15). At least in clear cases, blending structures are easily seen as a particular kind of schematic network. Coulson and Oakley (2000: 178) describe the conceptual integration network as central to conceptual blending theory: These networks consist of two or more input spaces structured by informa- tion from discrete cognitive domains, a generic space that contains structure common to all spaces in the network, and a blended space that contains selected aspects of structure from each input space, and frequently, emergent structure of its own. Blending involves the establishment of partial mappings between cognitive models in different spaces in the network, and the projection of con- ceptual structure from space to space. This can be expressed in a schematic network as in figure 4.14a. Figures 4.14b–d represent three kinds of blends. Figure 4.14b represents a high-level blend refer- enced in several publications by Fauconnier (e.g., 1997), in which the concepts of schematicity 109 . with the neighboring entity or a substructure of it. Usually when the whole of the one entity is identified with a substructure of the other, there is a clear relationship of schematicity between the. (the ‘‘figurative’’ sense). Their whole configuration, including both senses, will constitute the semantic pole of the expression in its figurative usage. Thus, in the cat (is) out of the bag the. particular entity within the scene. Such a construct in the case of plate and food certainly abstracts from the differences and retains the commonality of the two concepts, which makes it rather