Encyclopedic meaning and its organization

If linguistic meaning is learned, and moreover, learned by a non-modular cognitive system, its necessary property is that it cannot be distinguished by any strict criteria from other categories of human knowledge (FCG-1: 155–

156).

For Cognitive Grammar, meaning equals conceptualization (CGBI: 30);

conceptualization, in turn, may accommodate anything that is experientially associated with a concept through the contexts of its instantiation. This view of semantics is referred to as encyclopedic (for a comprehensive description, see FCG-1: 154–166). The meaning of a linguistic item cannot be condensed into a description in propositional form in the way of a dictionary item, or by feature matrices as in structural semantics; instead, meaning is seen as consisting of multiple nodes of information that are interconnected by, and organized hierarchically according to, their centrality for a given concept, its semantic variant or its specific instance.

Network model precludes definite boundaries of meaning: centrality is a gradient feature, a relative value of which may easily be overruled by contextual factors. Similarly, a usage-based grammar necessarily involves gradation from pragmatics to semantics (e.g. Langacker 1988: 57, CGBI: 39–

42). A semantic unit acquires its unit status gradually through use but also continues to be elaborated and adjusted by the pragmatic contexts in which it is instantiated (FCG-1: 157). The result is that the boundary between conventional and non-conventional conceptual meaning is indeterminate both for a semantic change as a process, as well as for an encyclopedic meaning as a structure.

Langacker (CGBI: 40), however, denies the interpretation that the view of Cognitive Grammar would preclude the existence of pragmatics altogether.

Rather, context-dependent or pragmatic meaning is the logically necessary prerequisite of conventional semantics. Semantics and pragmatics are thus presented in Cognitive Grammar as separate concepts, which, however, do not correspond to strictly separate ontological categories. Even from the perspective of Cognitive Grammar, one can point out phenomena that are clearly either semantic or pragmatic (ibid.). The mere existence of intermediate cases nevertheless entails that the categories these notions evoke are ontologically fluid.

The notion of gradation from pragmatics to semantics is a logical extension of Cognitive Grammar’s theoretical premises. The specific formulation of this gradation has nevertheless certain weak points that need to be addressed. First

of all, as is extensively argued by Itkonen (e.g. 1978, 1981, 1997, 2003, 2008a, 2008b) sufficient linguistic communication requires that it is carried out under rules that are socially constituted and thus commonly known, therefore transcending the spatiotemporal scope of a single instance. Conversely, the knowledge of language is of normative character, i.e. it involves socially defined criteria of correctness. In so being, linguistic knowledge of speaker of a language is unfalsifiablea priori (Itkonen 1997: 58), and it cannot be reduced into a contingent fact about the context in which is instantiated. Furthermore, the correctness is typically a binary feature, which suggests a rather well-defined category of context-independent linguistic knowledge.

Now, the application of this social definition to knowledge of linguistic meaning is not totally straightforward (for discussion see Itkonen 1997, 2008a, forthc). The vagueness and flexibility of meaning-construction in actual discourse precludes any specific predictions about what a linguistic unit will mean on specific instance; just consider the unlimited human creativity in how we establish novel animal analogies in our attempt to insult our conspecifics. It can be argued, however, that such creativity supposes the conventional semantics, from which the mutually understandable insults are derived from. Pragmatic meaning can therefore be defined, at least to a substantial extent, as dependent on conventional meanings.

It thus seems necessary that semantics and pragmatics are posited as separate yet interacting ontological categories. On the other hand, an analytical distinction between semantics and pragmatics is necessary for any systematical description of meaning: any categorical judgment about a semantic regularity presupposes the exclusion or abstraction of contextual variables – or the detection of a contextual constant. The relevant question for a theory of encyclopedic semantics is how these two distinctions, the ontological and the analytical, should correspond to each other.

The most practical answer is prefaced by defining the ontological distinction as ideally definite but empirically indefinite. There can be no partially falsifiable knowledge, so any category constituted by unfalsifiable knowledge has to have quite precise boundaries ideally, whether the mapping of these boundaries is a possible task or not. This position, in turn, defines the nature of the analytical distinction. We cannot to distinguish in any absolute or comprehensive way between the categories of conventional and non-conventional meaning for a specific expression. Rather, we need to assume that the distinction exists and try to map out the parts of the boundary that are relevant for the present issue. As the semantic/pragmatic distinction is in any case operationalized by semantic analysis, the analytical distinction is best seen as systematic approximation of the ideal goal.

The analytical distinction between semantics and pragmatics is therefore best applied so that the latter refers to any part of meaning that is defined, elaborated or disambiguated by any part of the context. This approach makes

the distinction largely perspectival.¹⁶ Any actual instance of a semantic unit involves both semantically and pragmatically defined content, and the conventional content itself may pertain to a semantic variant of a polysemic concept disambiguated by the context. Furthermore, Östman (2005) demonstrates how the pattern according to which a conventional semantic unit becomes specified from context may itself entrench into a conventional pattern of symbolization, that is, a “discourse construction” (ibid. 130). Given the abundance of potentially conventional discourse structures, it is safest to define the semantic/pragmatic distinction as an analytical tool, the specific import of which depends on the level of analysis vis-à-vis the organization of discourse.¹⁷

As faithful to its theoretical proclamations, Cognitive Grammar nevertheless avoids applying the distinction between semantics and pragmatics into its descriptive praxis. To prevent the concept of semantics from dissolving entirely, Cognitive Grammar must present a principled view of how different encyclopedic knowledge associated with a concept is organized in its acquisition and usage.

This view is based essentially on the concepts of figure/ground segregation and cognitive domain. Cognitive domain, in brief, is a function of a concept defined by Cognitive Grammar (FCG-1: 488) as a “coherent area of conceptualization relative to which semantic units may be characterized”. The different pieces of information associated with a concept are described as cognitive domains or substructures within cognitive domains, which themselves are usually free-standing concepts of their own. The rationale is that, similar to the way in which different perceptual objects are categorized in comparison to other objects, the informational substructures of a concept are categorized in comparison to matching elements within different cognitive domains. For instance, the meaning of [BALLOON]¹⁸is based on our experience with balloons, which can be decomposed into different realms of experience;

for example, the shape, feel, typical color, function, physical behavior, cultural function etc., which all become meaningful as related to different cognitive domains (e.g. the basic domain of [SPACE] or the complex domain of [PARTY]).

Figure/ground segregation, in contrast, is a gestalt-psychological core principle, according to which the perception of an object always involves its demarcation out of whatever serves as a background (FCG-1: 120–122, CGBI:

58). Cognitive Grammar (following Talmy’s Cognitive Semantics: 1972, 1978, 2000a: chap. 5) claims that this principle is active within conceptual structure

16 This approach is further motivated by the sheer number of possible semantic/pragmatic interfaces (Turner 2009).

17 In my own analysis of discourse-level construal phenomena, the level of pragmatics is constituted primarily by intra-textual conceptual relations, especially those between co-referential expressions.

18 Concepts and cognitive domains are not only treated in a uniform manner by Cognitive Grammar, but they are also expressed using the same notation. Brackets stand for unit status, and capital letters mark semantic content.

as well. The linguistic manifestation of figure/ground segregation is referred to as profile/base-structure (FCG-1: 183–189). Put as simply as possible, a profile is an “entity designated by a semantic structure” (FCG-1: 491). CG’s argument, however, is that profile is not enough to describe the semantic value of an expression. Rather, “[t]he semantic value of an expression […] derives from the designation of a specific entity identified and characterized by its position within a larger whole” (FCG-1: 183). What this means in practice is that conceptual information is seen as inherently relational: the concept of a three-dimensional object consists of configuration or a region within a conceptual domain of three-dimensional space.

Thus, whereas the profile is the “entity designated by a semantic structure”, the base is the requisite cognitive structure within which the profile delineates a sub-part. According to Langacker (FCG-1: 486), it is “[t]he cognitive structure against which the designatum of semantic structure is profiled”. As in the case of [BALLOON], certain properties are more central to a concept’s meaning either by default or under pragmatic conditions: the appearance and use of a balloon are relatively more salient within the conventional meaning of balloon than the knowledge of the balloon’s capacity of being charged with static electricity, or the sound it makes when it deflates. The combination of the most salient properties stands out as the concepts profile, which is thereby delineated within the concept’s semantic scope. Most concepts are complex enough to conflate information from different cognitive domains; the typical organization of a conceptual base is thus a combination of different cognitive domains, denoted by concept’s matrix (CGBI: 44).

The definition of profile is straightforward, but it is obvious that its case-specific application does not yield an exhaustive description of the “entity designated by a semantic structure”. The profile/base distinction rather introduces a principled way of describing the vagueness of meaning: what is profiled or un-profiled is not definitive or rigid, but (once again) subject to various pragmatic effects. This adds flexibility to the theory, making Cognitive Grammar able to describe the variability of lexical meaning in terms of an alteration of conventional organization within the matrix. The visual appearance of an inflated balloon, for instance, is thus quite peripheral to the meaning ofballoon when the concept is entertained by a technician working at the balloon factory.

The profile/base distinction has yet one important function. It is the key-concept in Cognitive Grammar’s effort to provide the main grammatical categories with a semantic description. What is profiled can be either a “thing”

or a “relation”. Whereas things are designated by a nominal predication (i.e.

nouns), relations divide into processes (designated by verbs) and atemporal relations (designated, for example, by adjectives, adverbs and grammatical cases). Typically, most concepts involve in relationally organized information in their bases. The point Cognitive Grammar makes is that it is the profiled entity that distinguishes major conceptual classes from each other: a region in a conceptual space or a relation between multiple such regions.

Categorization based on schemas and prototypes, the encyclopedic conception of meaning and its organization into cognitive domains according to the figure/ground-organization are at the basis of semantics in Cognitive Grammar. This type of semantics blurs the distinction between perceptual and conceptual information both in positing cognitive mechanisms and properties shared by these two realms and in allowing experiential meaning to conventionalize into semantic units. This is further elaborated by the positing of so-called basic domains, which relate to distinct perceptually grounded or otherwise conceptually non-reducible types of information, e.g. the basic domain of [SPACE]. The most important analogy between perceptual and conceptual information posited by Cognitive Grammar, however, is the analogy between the expressive flexibility of natural language and the viewing or experiencing of a situation from different viewpoints or perspectives. This capability is explicated by the dimensions of construal, and I will return to it after discussing how these dimensions depend on a representative conceptualizer.