Show Summary Details

Page of

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, LINGUISTICS ( (c) Oxford University Press USA, 2016. All Rights Reserved. Personal use only; commercial use is strictly prohibited. Please see applicable Privacy Policy and Legal Notice (for details see Privacy Policy).

date: 22 January 2018

Cognitively Oriented Theories of Meaning

Summary and Keywords

There are two main theoretical traditions in semantics. One is based on realism, where meanings are described as relations between language and the world, often in terms of truth conditions. The other is cognitivistic, where meanings are identified with mental structures. This article presents some of the main ideas and theories within the cognitivist approach.

A central tenet of cognitively oriented theories of meaning is that there are close connections between the meaning structures and other cognitive processes. In particular, parallels between semantics and visual processes have been studied. As a complement, the theory of embodied cognition focuses on the relation between actions and components of meaning.

One of the main methods of representing cognitive meaning structures is to use images schemas and idealized cognitive models. Such schemas focus on spatial relations between various semantic elements. Images schemas are often constructed using Gestalt psychological notions, including those of trajector and landmark, corresponding to figure and ground. In this tradition, metaphors and metonymies are considered to be central meaning transforming processes.

A related approach is force dynamics. Here, the semantic schemas are construed from forces and their relations rather than from spatial relations. Recent extensions involve cognitive representations of actions and events, which then form the basis for a semantics of verbs.

A third approach is the theory of conceptual spaces. In this theory, meanings are represented as regions of semantic domains such as space, time, color, weight, size, and shape. For example, strong evidence exists that color words in a large variety of languages correspond to such regions. This approach has been extended to a general account of the semantics of some of the main word classes, including adjectives, verbs, and prepositions. The theory of conceptual spaces shows similarities to the older frame semantics and feature analysis, but it puts more emphasis on geometric structures.

A general criticism against cognitive theories of semantics is that they only consider the meaning structures of individuals, but neglect the social aspects of semantics, that is, that meanings are shared within a community. Recent theoretical proposals counter this by suggesting that semantics should be seen as a meeting of minds, that is, communicative processes that lead to the alignment of meanings between individuals. On this approach, semantics is seen as a product of communication, constrained by the cognitive mechanisms of the individuals.

Keywords: cognitive semantics, conceptual spaces, frame semantics, idealized cognitive models, image schemas, prototype theory, semantic maps, visual cognition

1. What Is Cognitive Semantics?

1.1 Realism Versus Cognitivism in Semantics

There exist two fundamentally different answers to the question of what constitutes the semantics of language, one realistic and one cognitive (or conceptualistic). According to the realistic approach, the meaning of a word (part of speech) or expression is something out there in the world. For example, the word horse denotes all horses in the world. According to the cognitivist view, meanings are mental entities. For example, the word unicorn denotes an animal that looks like a horse with a horn on its forehead, independently of whether such animals actually exist. This article presents some cognitively oriented theories of meaning. However, before this can be done, some general remarks concerning the goals of a semantic theory are in place.

There are two classical realist approaches. One approach, deriving from Aristotle, delineates the meanings of the predicates (a cover term for all content words) in terms of necessary and sufficient conditions. The second approach describes meanings in terms of truth conditions based on a mapping between elements of a language and the external world. The first developed theory of this type is Frege’s (1892) semantics. Both approaches bracket the role of the language users in the connection between a language and its meaning.

In contrast, the core idea of a cognitive semantics is that meanings of linguistic expressions are cognitive constructions. These constructions must not necessarily have correspondences in the real world. People communicate with surprising success about centaurs, witches, and the grin of a Cheshire cat. The use of language is seen as one among many cognitive activities and not a process with independent standing. Thus, a cognitive semantics puts the minds of the language users in the center.

1.2 Lexical Meanings and Construals

Within cognitive semantics the emphasis is on lexical meaning rather than on the meanings of sentences (see Geeraerts, 2017). Most analyses concern relations between words and representations of concepts (categories).1 Since Aristotle, philosophical analyses of concepts have aimed at providing necessary and sufficient conditions for the application of a concept. Instead, many cognitively oriented theories rely on prototype theories of concepts. According to the Aristotelian view, concepts are represented by natural kinds that have a fixed reference. In contrast, within cognitive semantics, concepts are seen as more flexible and dependent on the context.

Many words seem to be polysemous, that is, they appear have a number of different meanings that are only distantly related (see Vicente & Faljum, 2017). There are two main ways of handling this phenomenon in a semantic theory (Lakoff, 1987, p. 420; Tyler & Evans, 2001, pp. 727–733; van der Gucht, Klaas, & De Cuypere, 2007; Zlatev, 2003). One is full specification, where each meaning of a word is represented separately in the lexicon, but where semantic relations between the different meanings can also be specified. The other is minimal specification, where one meaning of a word is considered to be central and where other meanings are derived from the central one by additional information from the context or by semantic transformations (Gärdenfors, 2014, section 11.1.1; Wilson & Carston, 2007).2

Lakoff (1987, p. 422) argues in favor of the full specification interpretation and Tyler and Evans (2001, pp. 731–737) argue for a weaker form they call principled polysemy, the main difference being that Tyler and Evans provide criteria for identifying separate meanings. Principles of cognitive economy, however, count against the full specification position. People’s minds do not fabricate new meanings ad hoc, but adapt cognitive constructions that are already available (cf. Jackendoff, 1983, pp. 118–189). Another disadvantage of the full specification approach is that our memory would be strained if, for example, we were to have separate encodings of all the 24 meanings of over that Lakoff (1987) identifies. In contrast, remembering a prototypical meaning and then using some general semantic principles for creating other meanings is more cognitively economical.

On the other hand, the minimal specification approach must supply principles for how new meanings of a word are generated from the central one. A general proposal is presented by Wilson and Carston (2007), who discuss mechanisms of lexical narrowing (drink used to mean ‘alcoholic drink’), approximation (‘south-facing’ is used to mean ‘in the general direction of south’), and metaphor. Using the relevance theory of Sperber and Wilson (1986), they argue that these forms of lexical extensions can be given a unified treatment.3

When meaning elements are combined to form meanings of complex concepts and sentences, cognitive semanticists often speak of construals. A construal is a mental model of an event with a particular focus of attention (often called topic) (see Croft, 2012, section 1.4; Croft & Wood, 2000; Givón, 2001; Langacker, 1987, section 3.3; Langacker, 2008, ch. 3). There are, however, other aspects of how a construal is formed (see Croft & Wood, 2000; Langacker, 2008, ch. 3, for a survey). One aspect is perspective: For example, if you and I are located on two sides of a house, I can say that you are behind the house, if I put myself in the center, or I can say that you are in front of the house, if I put the house and the direction of its main side in focus. Another aspect is categorization: A construal must select a level of generality to describe an object, for example, terrier, dog, mammal, or animal. Yet another aspect is the relation to the common ground in the communicative situation (see Section 7). For example, when selecting whether to use a pronoun, noun, or name to refer to an individual, the speaker must consider whether the individual or the name is part of the common ground.

In this article, two main features of cognitive semantics are presented. The first concerns connections between meaning and perceptual and bodily experiences. Section 2 discusses the connections between meaning structures and other cognitive processes. Some of the main perceptually oriented semantic theories are presented in section 3. Force dynamics is introduced as a foundation for an analysis of the role of actions and events in semantics (sections 4 and 5). Semantic maps are briefly presented in section 6. The second feature concerns socio-cognitive aspects of meanings that are built up during discourse. Section 7 is devoted to how socio-cognitive aspects of language use influence how meanings of expressions are created.

2. Connections Between Meaning Structures and Other Cognitive Processes

A primary task for a cognitively oriented theory of meaning is to describe the cognitive structures that carry the meanings (cf. Jackendoff, 1983, p. 16, cognitive constraint). If the use of language is grounded in cognition, it will be influenced by other cognitive processes, such as attention, perception, and memory. Therefore the nature of human perceptual and cognitive systems is highly relevant to the study of language itself. A central challenge for a cognitive semantics is to ferret out the connections between language and human cognition.

2.1 Parallels Between Semantics and Visual Processes

A paradigmatic example of such a connection is how people can talk about what they see. It turns out that there are clear parallels between semantics and visual processes, many of which one is normally not aware of (Jackendoff, 1987; Landau & Jackendoff, 1993; also cf. Chafe, 1995; and Gärdenfors, 2004). In brief, semantics and perception, vision in particular, cannot be separated. As a consequence, the structure of perceptual processes will constrain semantic representations.

An interesting example of a mental process that mirrors vision is fictive motion that shows up in many linguistic constructions. Consider the following examples:


Cognitively Oriented Theories of Meaning


Cognitively Oriented Theories of Meaning


Cognitively Oriented Theories of Meaning

These constructions do not describe any processes that occur in the real world. Instead they represent people’s understanding of how they ‘perceive’ in their inner worlds: you follow the fence with your gaze, your gaze moves along the landscape, and the ‘vision’ of the camera is directed to a particular area (cf. Matlock, 2004).

A related notion is discussed by Langacker (1987, Section 3.1.2), who argues that one can perform a mental scanning of various forms of spatial structures. For example, the two following sentences describe the same scene:


Cognitively Oriented Theories of Meaning


Cognitively Oriented Theories of Meaning

The difference between (4) and (5) is the direction of the mental scanning.

Talmy (2000) writes that a perspective is selected from which the speaker’s ‘mental eyes’ look out when a scene is described.


Cognitively Oriented Theories of Meaning


Cognitively Oriented Theories of Meaning

In (6), the perspective is from inside the room, while in (7) an external perspective is adopted.

Two other visual processes are zooming in and zooming out. These processes have clear parallels also in semantics. For example, the location of an object can be described as zooming in:


Cognitively Oriented Theories of Meaning

The same task can also be achieved by zooming out:


Cognitively Oriented Theories of Meaning

Cognitive semantics is influenced by ideas from Gestalt psychology. The distinction between figure and ground reappears as the distinction between trajector and landmark in the works of Lakoff (1987); Langacker (1987); and Talmy (1988), for example. The trajector (figure) can be interpreted as the focus of attention, while the landmark (ground) is the entity that the trajectory is contrasted with. Note that this kind of attention is directed to the ‘mental space’ of representation, not to the ordinary space as in visual or auditory attention. A tacit assumption, though, is that the two kinds of attentional processes are of the same kind.

The general conclusion of these examples is that any cognitive theory of meaning should account for the parallels with perceptual processes. A strong proponent of the perceptual grounding of semantics is Barsalou (1999), who argues that word meanings should be understood as perceptual symbols that are dynamic patterns of neurons functioning as simulators that combine with other processes to create conceptual meaning.

2.2 Metaphors and Metonymies

Cognitive linguistics has brought forward the importance of metaphors and metonymies as strong tools for expanding the meanings of words (see Geeraerts, 2017). Also these semantic constructs have perceptual correspondences. It is natural to describe a metaphor as the transfer of a pattern from one domain to another. For example in ‘sharp comment’ and ‘uneven discussion’ the words sharp and uneven are transferred from the shapes domain to the more abstract domain of conversation or argumentation (Lakoff & Johnson, 1980). The notion of a pattern is most easily connected to the visual modality, but patterns can also be found in auditory and tactile experiences. A metaphor can thus be used to refer to a structure in a new domain by reusing the meaning of a word from an old domain. In this way one can account for how a metaphor transfers information from one conceptual domain to another; what is transferred is a perceptual pattern rather than domain-specific information.

Metonymies can, in general, be described as processes of refocusing. Classical types of metonymy include pars pro toto, where a part of an object is focused instead of a whole object (“there are three sails on the lake,” i.e., sails instead of boats), and totum pro parte, where the focus is put on something that contains the object as a part (“Paris announces shorter skirts this season,” i.e., the city is made the focus instead of the fashion designers in it). These processes correspond to the process of zooming in (on the sails) and zooming out (to the city of Paris) in visual attention. In contrast to metaphor, metonymy is based not on the similarity between two domains but on meronomic (part-whole) and other contiguity relations within the same domain (Lakoff & Turner, 1989, p. 103; Langacker, 2008, p. 69; Peirsman & Geeraerts, 2006, p. 27). In brief, metaphors refer to mappings between domains; metonymies refer to meronomic and contiguity relations within domains (see also Gärdenfors & Löhndorf, 2013).

3. Meaning Structures

An assumption of semantics is that most (non-grammatical) words refer to categories. So a basic question for a cognitive theory of meaning is how categories are mentally represented. In this section some of the most influential answers are presented.

3.1 Prototypes

Within psychology, the prototype theory developed by Rosch and her collaborators was presented as an alternative to the classical Aristotelian view of categories (see, for example, Lakoff, 1987; Mervis & Rosch, 1981; Rosch, 1975). The main idea of prototype theory is that within a category of objects, certain members are judged to be more representative than others. For example, robins are judged to be more representative of the category bird than are ravens, penguins, and emus; and desk chairs are more typical instances of the category chair than rocking chairs, deck chairs, and beanbag chairs. The most representative members of a category are called prototypical members. Furthermore, representation via prototypes also entails that categories show graded membership, determined by how representative the members are. This is in contrast to the classical theory of categories where all members of a category have equal status (Smith & Medin, 1981).

One problem of prototype theory is that the structures proposed are not rich enough to handle the basic semantic task of explaining compositions of meanings. For example, a pet fish is neither a prototypical fish, nor a prototypical pet. An early proposal was to use fuzzy sets to compute the prototype of a combination of concepts from the prototypes of its constituents. However, Osherson and Smith (1981) demonstrate that this approach results in incorrect results for many types of combinations. Kamp and Partee (1995) try to circumvent these problems for the prototype theory by applying the notion of supervaluations. But a straightforward application of their theory cannot handle examples like striped apple and porcelain cat. The conclusion is that prototype theory does not seem sufficient for developing a cognitive semantics.

3.2 Frame Semantics

An early proposal to model cognitive representations of meanings is frame semantics, proposed by Fillmore (1968, 1976, 1982). A frame consists of a number of attributes (dimensions), each of which has a set of possible features (also called values). For example, the frame for car has attributes such as engine, wheels, motor, fuel, drive, each of which can take on different features—a motor can have the values 4-cylinder or 6-cylinder, etc. (Barsalou, 1992). Löbner (2014) generalizes the notion of a frame and suggests that it can be used for representing syntactic structure as well as cognitive meaning. Frame semantics will be compared to conceptual spaces in Section 3.4.

Frame semantics originated with case grammar (Fillmore, 1968): A case frame is described as “characterizing a small abstract ‘scene’ or ‘situation,’ so that to understand the semantic structure of a verb it was necessary to understand the properties of such schematized scenes” (Fillmore, 1982, p. 115). An example is the ‘commercial transaction’ frame, the elements of which include buyer, seller, goods, and money. The attributes of this frame are these ‘participant roles’ and their features are the various persons and objects that can fill the roles. When explaining the semantics of verbs such as buy, sell, pay, charge, etc., different elements of the frame and their relations are exploited. For example, (10) expresses an acceptable relation between the attributes, while (11) does not:


Cognitively Oriented Theories of Meaning


Cognitively Oriented Theories of Meaning

3.3 Image Schemas and Idealized Cognitive Models

A concept that to some extent derives from frame semantics is that of an image schema. Image schemas also have clear connections to perceptual processes; in particular, they have an inherent spatial structure. For example, Lakoff and Johnson (1980), Johnson (1987), and Lakoff (1987) argue that schemas such as container, source-path-goal, and link are among the most fundamental elements of meaning.

However, the term image schema is used in different ways by different cognitive semanticists (Zlatev, 1997, pp. 40–44). Holmqvist (1993, p. 31) connects directly to perception when he defines image schemas as “that part of a picture which remains when all the structure is removed from the picture, except for that which belongs to a single morpheme, a sentence or a piece of text in a linguistic description of a picture.” Mandler relates image schemas to topological representations:

An image of a container … must have a particular shape, and the material inside it either conforms to the shape of the container or not, but a topological representation of this relation eliminates this information, leaving only the topological relation of a bounded space with an inside and an outside. In this sense image schemas are topological: They simply do not include some of the information that might be in an image. (Mandler, 2004, pp. 81–82)

The most condensed account comes from Gibbs and Colston (1995, p. 349), who define image schemas as “dynamic analogue representations of spatial relations and movements in space.” This definition focuses on the dynamics of the representations.

Most of the analyses within cognitive linguistics concern relations between words and categorizations, that is, mental representations of concepts. A notion that combines prototype theory with image schemas is Lakoff’s (1987) idealized cognitive models. An idealized cognitive model is a mental representation of some aspects of the world—idealized because it abstracts across a range of experiences. However, idealized cognitive models are typically richer in detail than the basic image schemas that were introduced by Lakoff and Johnson. Similar ideas were also proposed by Vandeloise (1986), Herskovits (1986), Langacker (1987), Talmy (1988) and others, and they all use some form of graphic representations of the mental models. There exists, however, no general theory concerning what counts as a cognitive model and how the models are to be identified and described.

A common assumption in the kinds of lexical analyses performed in cognitive linguistics is that a word or an expression has a prototypical meaning, which then can be extended by different transformations. For example, Brugman’s (1981) and Lakoff’s (1987) central model for over is depicted in Figure 1.4 The content of the schema can be formulated entirely in terms of spatial dimensions as the trajector moving along a horizontal path in a position vertically higher than a landmark. A typical example is the following:


Cognitively Oriented Theories of Meaning

Cognitively Oriented Theories of MeaningClick to view larger

Figure 1. The Central Image Schema for Over

(Source: Lakoff, 1987, p. 426).

To give another example of how a cognitive model is represented, Langacker (1987) argues that the meaning of a verb is represented as a process in time. In a schematic way, such a process can be described as involving three stages: one schema part for the beginning, one part for the middle and one part for the end of the process. A basic cognitive model for climb, for example, is depicted as in Figure 2.

Cognitively Oriented Theories of MeaningClick to view larger

Figure 2. Image Schema for Climb

(Source: Langacker, 1987, p. 311). tr = trajector, lm = landmark.

In this figure, the time axis is drawn below the three stages. It is marked as a thick line since the temporal aspect of climb is in focus. The landmark is supposed to be vertically extended and the trajector, the thing doing the climbing (the small circle), is assumed to be in physical contact with the landmark.

Also, grammatical elements are analyzed as having schematic meanings, albeit more abstract than those of content words. Consider the morpheme –er that, in English, is used to turn a verb for an action into a noun representing the agent executing the action (agentive nominal). According to Langacker, the schema for the verb climb can be turned into a schema for climber by using the same dimensions, objects, and relations. Only the focus of the schema is shifted from the time dimension to the trajector (see Figure 3). The thick line around the schema is how Langacker represents an object (in contrast to a process where the time line is emphasized).

Cognitively Oriented Theories of MeaningClick to view larger

Figure 3. Image Schema for Climber

(Source: Langacker, 1987, p. 311).

The transformation can be viewed as the mental correspondence of refocusing. This kind of change has an obvious parallel in vision, where looking at the same scene can generate very different mental construals, depending on which aspects of the scene are focused on.

Some researchers have tried to derive all cognitive schemas from a fundamental set. One such theory is Jackendoff’s (1990) conceptual semantics (for a related program, see Goddard & Wierzbicka, 1994). Jackendoff assumes a finite set of semantic primitives and a finite set of rules governing their interaction. A problem with the theory is that the primitives, for example plural, are highly abstract and it is unclear to what extent such primitives have a grounding in human cognition.

3.4 Conceptual Spaces

Instead of image schemas that represent abstract patterns, other cognitively oriented theories of meaning use tools from topology and geometry. Some early models, based on the mathematical catastrophe theory, have been developed by Petitot (2011) and others. These theories have, however, only been applied to a limited set of semantic areas. Another geometric approach is the vector models of Zwarts (1997), which are used in analyses of the semantics of prepositions.

Gärdenfors (2000, 2014) has developed a theory of conceptual spaces that is proposed as a cognitive foundation for the semantics of natural languages. A conceptual space consists of a number of quality dimensions. Examples of such dimensions are temperature, weight, brightness, pitch, and force, as well as the three ordinary spatial dimensions of height, width, and depth. The force dimension is essential for the analysis of actions and events and thereby for the semantics of verbs. Quality dimensions correspond to the different ways stimuli can be judged similar.

Dimensions are bundled into domains that represent different classes of properties, for example color, taste (sweet, salty, sour, and bitter dimensions). and sound (pitch and volume dimensions). A conceptual space is defined as a collection of quality dimensions divided into domains. Most domains can be assigned a metric, so that one can talk about distances in the conceptual space. Such distances indicate degrees of similarity between the objects represented in the space. However, there are domains that are of a qualitative nature, such as kinship relations or biological classifications.

To illustrate such a geometric structure, consider the color space. Human cognitive representations of color can be described along three dimensions. The first is hue, represented by the familiar color circle going from red to yellow to green to blue, then back to red again. The second dimension is saturation, which ranges from grey as a zero point in the center, to increasingly greater intensities of color at the border of the color circle. This dimension is isomorphic to an interval of the real number line. The third dimension is brightness, which varies from white to black, and thus is also isomorphic to a bounded interval of the real number line. Together, these three dimensions—one circular, two linear—constitute the color domain. It can be illustrated by the so-called color spindle (see Figure 4).

Cognitively Oriented Theories of MeaningClick to view larger

Figure 4. The Color Space

(Source: Gärdenfors, 2014).

The notion of a domain has been discussed extensively in cognitive linguistics (e.g., Croft, 1993; Croft & Cruse, 2004; Evans & Green, 2006; Langacker, 1987). Langacker (1987, pp. 152–154) distinguishes between ‘locational’ and ‘configurational’ domains, where locational means being located within dimensional space, while configurational concerns the relations between the parts of objects (see also Gärdenfors & Löhndorf, 2013). Besides basic domains, Langacker talks about abstract domains, for which identifying the underlying dimensions is more difficult. The domains correspond roughly to the ‘lexical fields’ of earlier semantic theories (Lyons, 1977; Trier, 1931). However, the geometric structures of the domains bring in richer representational possibilities.

Domains are related to the attributes of frame semantics. The main contrast with frame semantics is that the theory of conceptual spaces puts greater emphasis on the geometrical structure of the concept representations. The definition of word meanings in conceptual spaces makes it possible talk about meanings being close to each other and about objects being more or less central representatives of a category. Thus conceptual space can be seen as combining frames with prototype theory, although the structure of the conceptual spaces makes possible predictions that can neither be made in frame semantics nor in prototype theory (see e.g., Jäger, 2010). On the other hand, frame semantics can be applied to a broader range of phenomena than lexical semantics (see e.g., Löbner, 2014).

A limitation of the theory of conceptual spaces that it shares with frame semantics is that the set of possible domains (attributes) is not well defined, and their structure may be unknown. For perceptual domains, such as the color domain, the structure can be established via psychophysical experiments, but for more abstract domain other methods must be sought.

Gärdenfors (2000, 2014) proposes that meanings of words from open word classes can be represented as convex regions of a conceptual space. That a region is convex means that, if some objects located at x and y in some domain are both examples of the meaning of a word, then any object that is located between x and y will also be an example of the meaning (see Figure 5). For example, the meaning of red is represented as a convex region of the three-dimensional color space. Convexity may seem a strong assumption, but it is a remarkably regular property of many perceptually grounded categories: for instance, colors, vowels, and shapes. Although a main argument for convexity is that it facilitates the learnability of concepts, it is also crucial for assuring the effectiveness of communication (Jäger & van Rooij, 2007; Warglien & Gärdenfors, 2013).

Cognitively Oriented Theories of MeaningClick to view larger

Figure 5. Illustration of a Convex and a Non-Convex Set.

Analyzing word meanings as convex regions provides a motivation for some aspects of prototype theory. When word meanings are defined as convex regions in a conceptual space, prototype effects are to be expected. Given a convex region, one can describe positions in that region as being more or less central. Conversely, if prototype theory is adopted, then the representation of word meanings as convex regions is to be expected. Assume that some quality dimensions of a conceptual space are given: for example, the dimensions of color space; and that the intention is to decompose it into a number of categories, in this case, colors. If one starts from a set of prototypes—say, the focal colors—then these prototypes should be the central points in the categories they represent. The prototypes can then be used to generate convex regions by stipulating that any point within the space belongs to the same category as the closest prototype. This rule will generate a certain decomposition of the space: a so-called Voronoi tessellation that always generates a convex partitioning of the stimuli space.

The structures introduced by conceptual spaces can be used to represent the meanings of different classes of words. First, Gärdenfors (2000, 2014) proposes that the meaning of an adjective can be represented as a convex region of a single domain such as a color (red, green, etc.), shape (round, rectangular, curved, etc.), or size (big, small, tall, etc.). He conjectured that color terms in natural languages denote convex regions of the color domain. This means, for example, that there should be no language that has a single word for the colors denoted by ‘green and orange’ in English (and that includes no other colors), since such a word would represent a non-convex area in the color space. Jäger (2010) has provided strong support for this conjecture. He studied color classification data from 110 languages and found 93.6% correct classifications in an optimally convex partitioning of the color space.

Second, the meaning of a noun is typically a concept represented as a complex of properties from a number of domains. For example, the meaning of apple corresponds to regions of the color domain, the shape domain, the taste domain, the nutrition domain, etc. Different types of nouns can be classified by the domains they presume. For example, the difference between count nouns and mass nouns can be defined thusly: mass nouns do not include the shape domain. Similarly, abstract nouns do not include the spatial domain.

Third, many prepositions express spatial relations, but there are also prepositions that refer to the time domain, such as before and after, and some that refer to the force domain, such as against and on (Gärdenfors, 2014, pp. 201–230). Spatial prepositions can be grouped into two classes: locative, indicating where something is, and directional, indicating where something is going. Zwarts and Gärdenfors (2016) show that the meanings of locative prepositions can be represented by convex regions of the spatial domain and directional prepositions by convex sets of paths. However, a special assumption for this analysis is that the metric of space is represented in polar coordinates.

Most cognitively oriented theories of meaning have focused on lexical semantics. However, all real communication involves composing simple meaning components (words) into larger structures (see Michaelis, 2017; Pelletier, 2016). A remarkable feature of human thinking is our ability to combine concepts and, in particular, to understand new combinations of concepts. Nobody has problems grasping the meaning of combinations like pink elephant, striped apple, and cubic soap bubble, even if one will never encounter any object with these properties. An important criterion for a successful theory of concepts is that it should be able to explain the mechanisms of meaning combinations. Although there exist some proposals for such models that cover certain types of compositions of meanings (e.g., Gärdenfors, 2014, 241–252; Hampton & Winter, 2017; Kamp & Partee, 1995), this area still lacks comprehensive modeling.

4. Force Dynamics

Much of the focus in cognitive semantics has lain on the spatial structure of the image schemas. Lakoff (1987, p. 283) goes as far as putting forward what he calls the “spatialization of form hypothesis,” which says that the meanings of linguistic expressions should be analyzed in terms of spatial image schemas plus metaphorical mappings.

As a complement, a more dynamic and embodied view of mental models has been developed. In Piaget’s sensory-motor schemas, which were introduced for modeling cognitive development and not for semantics, motor patterns are central. These can be seen as a special case of the dynamic patterns that form our fundamental understanding of the world. Following Johnson (1987) (also Barsalou, 2008; Casasanto, 2014; Clark, 1997; Lakoff, 1987), the claim is that semantic structures also depend on kinaesthetic experiences and on emotions.

Johnson (1987) was among the first to bring out the role of forces in image schemas. He argues that forces form perceptual gestalts. He writes: “Because force is everywhere, we tend to take it for granted and to overlook the nature of its operation. We easily forget that our bodies are clusters of forces and that every event of which we are part consists, minimally, of forces in interaction” (1987, p. 42). Johnson presents a number of ‘preconceptual gestalts’ for force. These gestalts function as the correspondence to image schemas but with forces as basic organizing features rather than spatial relations. The force gestalts he presents are ‘compulsion,’ ‘blockage,’ ‘counterforce,’ ‘diversion,’ ‘removal of restraint,’ ‘enablement,’ and ‘attraction.’

Experiments on how subjects perceive the movement of persons and other objects (e.g., Giese & Lappe, 2002; Johansson, 1973) support this position. The results suggest that the kinematics of movement contain sufficient information to identify the underlying dynamic force patterns. Runesson (1994, pp. 386-387) claims that one can directly perceive the forces that control different kinds of motion. The process is automatic—one cannot help but see the forces.

Also Talmy (1988) emphasizes the role of forces and dynamic patterns in image schemas in what he calls force dynamics. He recognizes the concept of force in expressions such as these:


Cognitively Oriented Theories of Meaning


Cognitively Oriented Theories of Meaning

Forces are taken as governing the linguistic causative, extending to notions like letting, hindering, helping, etc. In (13), the word kept indicates that the described situation does involve any counterforce that stops the rolling of the ball. The force domain can be extended metaphorically to social or psychological forces, for example commands, threats, persuasions, and seductions. In (14), John may be prevented from leaving the house by some physical force (e.g. the door being locked), but the preventing may also be of a mental nature (e.g., that he will be punished is he leaves the house). Talmy develops a schematic formalism that, for example, allows him to represent the difference in force patterns between expressions such as:


Cognitively Oriented Theories of Meaning



Cognitively Oriented Theories of Meaning

Talmy’s dynamic ontology consists of two directed forces of unequal strength, of which the focal one is called ‘agonist’ and the opposing one ‘antagonist,’ each having an intrinsic tendency towards either action or rest, and a resultant of the force interaction, which is also either action or rest. For example, in (16), the stiff grass exerts a counterforce (antagonist) to the inertia (agonist) of the rolling ball. Saying that the ball kept rolling is analyzed as that the agonist is stronger than the antagonist. A limitation of his analysis, however, is that it only involves situations with two participants.

Talmy also notices the possibility in language to choose between what he calls force-dynamically neutral expressions and ones that do exhibit force-dynamic patterns, as in the following:


Cognitively Oriented Theories of Meaning


Cognitively Oriented Theories of Meaning

Sentence (18) involves forces since it expresses that Maria has the power to close the door, but decided not to exert her power.

5. Events, Actions, and the Semantics of Verbs

5.1 Cognitive Theories of Events and Actions

In addition to space and objects, events are fundamental cognitive entities (see e.g., Zacks & Tversky, 2001). Within psychology there is an extensive research tradition investigating how events are perceived and cognitively organized (presented in Radvansky & Zacks, 2014). A particular focus in this tradition is the hierarchical organization of events. The purpose of this section is to outline how cognitive analyses of actions and events have prepared the way for new approaches to the semantics of verbs.

Croft (2012) presents a causal model of events based on two vector spaces (showing many similarities with conceptual spaces). An example of the model is presented in Figure 6.

Cognitively Oriented Theories of MeaningClick to view larger

Figure 6. Croft’s Representation of “Jack Broke the Vase”

(Source: Croft, 2012, p. 212; Figure 5.2).

Two sub-events are represented in this figure: the lower scheme involves two dimensions, q (force) and t (time), where Jack’s action is represented as a momentary change in the q-dimension. The upper scheme also involves two dimensions: q (qualitative change) and t (time), where the change of the vase is represented by a change of level along the q-dimension. The arrow from the lower to the upper scheme represents the causal chain. Croft, (2012, p. 9) emphasizes that his model is based on geometrically (as opposed to diagrammatically or symbolically) represented components since it involves dimensional spaces.

According to Croft, his model demonstrates that “events can be decomposed in three distinct ways: temporally, in terms of the temporal phases; qualitatively, in terms of the states defined on the qualitative dimensions for each participant’s sub-event; and causally, in terms of the segments of the causal chain” (Croft, 2012, pp. 216–217). He points out that dividing an event into sub-events accounts for its causal structure, while the nature of the qualitative changes covers the aspectual structure.

Similar ideas have been developed by Gärdenfors and Warglien (Gärdenfors, 2007; Gärdenfors, 2014; Gärdenfors & Warglien, 2012; Warglien & Gärdenfors, 2013; Warglien, Gärdenfors, & Westera, 2012). First, vectors are used to represent the forces involved in an action.5 For many actions, for example moving and lifting, a single force vector may be sufficient, but for some, such as walking and swimming, a complex of forces is involved. More generally, an action can be defined as a pattern of forces since several force vectors are interacting.

Second, an event is represented in terms of two components—the force of an action that drives the event, and the result of the application of force. (In contrast to Croft’s model, time is not explicitly represented, but is implicit in the dynamics of the event.) An event is built up from an agent, an action, a patient, a result, and possibly other ‘thematic roles’ such as instrument, recipient, and beneficiary (Dowty, 1991; Levin & Rappaport Hovav, 2005). Agent and patient are individuals or objects that have different properties. The result of an event is modeled as a change vector representing the change of properties of the patient before and after the event (see Gärdenfors, 2014). For example, when somebody (the agent) pushes (the force vector) a table (the patient), the forces exerted make the table move (the result vector). When somebody bends a stick, the result may be that the stick breaks. When the result involves no change, then the event is a state. For example, when a person is leaning against a wall, the forces and counterforces balance each other so that there is no movement.

In this framework, an event is represented by a mental model, a construal, that contains the force and result vectors. In contrast to Croft’s (2012) model and the psychological theories presented in Radvansky and Zacks (2014), an event in this model always contains both force and result vectors.

5.2 Semantics of Verbs

The cognitive models of actions and events have been used to develop a semantic representation for verb meanings (Croft, 2012; Gärdenfors, 2014; Warglien et al., 2012). Languages contain two types of verbs (Levin & Rappaport Hovav, 2005; Rappaport Hovav & Levin, 2010). The first type is manner verbs that describe how an action is performed. In English, some examples are run, swipe, wave, push, and punch. The second type is result verbs that describe the result of actions. In English, some examples are move, heat, clean, enter, and reach. When expressing an event in a sentence, either the force or the result vector is put in focus in the construal, and this focus determines whether a manner or result verb is used in the sentence.

Warglien et al. (2012) and Gärdenfors (2014) propose a thesis for the semantics of verbs that says that the meaning of a verb (verb root) is a convex set of vectors that depends only on a single domain. For example, push refers to the force vector of an event (and thus the force domain), move refers to changes in the spatial domain of the result vector and heat refers to changes in the temperature domain. The thesis entails that there are no verbs that mean ‘walk and burn’ (multiple domains), and there are no verbs that mean ‘crawl or run’ (not convex). The proposal can also explain similarities between verb meanings. For example, the meaning of walk is more similar to that of jog than that of jump because the force patterns representing walking are more similar to those for jogging than those for jumping.

In neurolinguistics, Pulvermüller’s (2003) results provide important indications of the connections between the meaning of verbs and bodily actions. He focuses on the motor aspects of meaning schemas. When the brain understands a verb, it prepares an action. For example, Pulvermüller has shown that when one reads the word kick, the same part of the motor cortex is activated as when one actually kicks. An interpretation is that the brain simulates the action it reads about.

6. Semantic Maps

Another way of representing meanings in a geometrical fashion is semantic maps (for example, Croft, 2001; Haspelmath, 1997, 2003; van der Auwera & Plungian, 1998; Zwarts, 2010). Like conceptual spaces, semantic maps are spatial representations of how a class of meanings are related so that closeness on the map means that meanings are similar. The maps have mainly concerned grammatical expressions (parts of language belonging to closed classes), but they can also be applied to open class content words. An early example is van der Auwera and Plungian’s (1998) map of different forms of modality.

Zwarts (2010) distinguishes between two approaches to semantic maps. In one approach, the map is ‘induced’ from a cross-linguistic lexical matrix in a data-driven fashion. In the other, the map is defined on cognitive grounds, independently from cross-linguistic data, and then the map is confronted with the lexical matrix.

The purpose of constructing a map is often typological in that the maps are supposed to represent meanings that are common to a group of languages or even universal to all languages. In other words, the hypothesis is that languages can differ in how they map the words from a particular domain onto a semantic map, but they cannot differ with respect to the underlying semantic map. If successful, the map suggests a universal underlying cognitive structure.

Cognitively Oriented Theories of MeaningClick to view larger

Figure 7. Haspelmath’s (1997) Semantic Map of Indefinite Pronouns.

A clear example is Haspelmath’s (1997) map of indefinite pronouns (see Figure 7) that is proposed as a universal map. The first distinction among the meanings is whether the pronoun is specific or non-specific. Meanings (3)–(9) are all non-specific. The distinction between (1) and (2) is determined by whether the speaker knows the referent or not. Case (3) covers the case when the speaker is not committed to the existence of the referent, etc., for the remaining cases (see Haspelmath, 1997, section 3.2, for a description of the nine cases). Haspelmath shows, for example, that the meaning of somebody in English covers meanings (1)–(5), nessuno in Italian covers (4), (6), and (7); dhipote in Greek covers (5), (8), and (9); and quelconque in French, and irgendein in German cover (2)–(6), (8), and (9). An explicit assumption of the methodology of semantic maps is that the meaning elements of the maps must be organized so that, for each domain and for each language considered, the region corresponding to the meaning of a term must be connected. Croft (2001, p. 96) calls this the semantic map connectivity hypothesis. This hypothesis is valid for all the 40 languages that Haspelmath has used as a basis for his map.

7. Cognitive-Pragmatic Aspects of Meaning

A heavy attack against the very possibility of cognitive semantics was launched by Putnam (1975). He writes that, in his mind, the words beech and elm have the same reference, but he knows that the words refer to different kinds of trees. On the basis of this, he claims that meanings cannot be determined from individuals’ mental representations. The lesson to be learned from Putnam’s argument, however, is not that cognitive semantics is impossible, but that it has downplayed the social structure of language.

The theory of image schemas (presented in Section 3.3) does not account for how image schemas can be compared between individuals. It is implicitly assumed that all individuals within a language community use the same schemas. However, this assumption is difficult to combine with any reasonable theory of how the image schemas or concepts are learned. A similar argument applies to frame semantics (presented in Section 3.2) and to the theory of conceptual spaces (presented in Section 3.4), because it may happen that different individuals associate a word with different frames or different regions of a space.

Instead of returning to a realist version of semantics, a satisfactory cognitively oriented theory of meaning must take the social aspects of language into account. The question is how it can be determined whether other people have ‘corresponding’ references in their mental construals when they use the same words? To attack this problem, a theory of meaning must also take the social and pragmatic aspects of language use into account. According to a social view, a theory of meaning should not primarily concern words or sentences, but rather communicative acts, in particular speech acts (see e.g., Dor, 2015).

As a reply to Putnam’s challenge, a semantic theory based on alignment or meeting of minds has been proposed (Pickering & Garrod, 2013; Warglien & Gärdenfors, 2013). The central claim is that the social meanings of the expressions of a language are indeed determined from their individual meanings, that is, the meanings the expressions have for the individuals; but the meanings will be adjusted within the heads of the individuals as a result of successful or failed communicative interactions. Another way of expressing this is to say that words do not ‘have’ meanings, but they evoke meanings (Paradis, 2005). According to this view, the meanings of expressions do not reside in the world or in the image schemas or concept representations of individual users, but they emerge from the communicative interactions of language users. A consequence of taking a cognitive perspective is that there will be no sharp boundary between pragmatics and semantics—semantics can be characterized as conventionalized pragmatics (Langacker, 1987, section 4.2).

Two basic types of meetings of minds can be distinguished: one slow and one faster (Gärdenfors, 2014, section 1.5). The slow one concerns how a community adjusts uses of words so that they obtain relatively fixed meanings within the community that are largely independent of any particular communicative context. This is the traditional realm of lexical semantics. The fast process concerns the development of a shared world—common ground—during a dialog or a similar exchange of communicative acts (Clark, 1992, 1996; Pickering & Garrod, 2004, 2013; Stalnaker, 1978). The kind of process concerns expressions—such as pronouns, demonstratives, determiners—that obtain their meanings during a communicative interaction.

Clark (1992) says that that process of constructing a common ground is characterized by the following properties:

  1. (i) The participants in a conversation work together against a background of shared information.

  2. (ii) As the discourse proceeds, the participants accumulate shared information by adding to it with each utterance.

  3. (iii) Speakers design their utterances so that their addressees can readily identify what is to be added to that common ground. (1992, pp. 4–5)

In this process, the meanings of the community from the slow process are taken as a starting point, but to this are added the meanings of noun phrases, pronouns, demonstrative, determiners, etc., that depend to some extent on the communicative environment. The fast process rapidly accumulates new meanings that the exchange can build upon in later stages.

A way of modeling such an accumulation of referents is Discourse Representation Theory (Kamp, 1981; see also Heim, 1982 for a similar approach). According to this theory, a discourse representation structure contains two main components: First is a set of referents that have been introduced or made salient during the discourse; second is a set of conditions representing information about the referents that the communicators share. These conditions are expressed in a formalism that is similar to predicate calculus. The semantics that has been proposed for discourse representation theory is an extension of classical model theory. Hamm, Kamp, and van Lambalgen (2006) suggest that discourse representation theory is compatible with a cognitive approach to semantics. On the other hand, Lakoff (1987) and others argue that a fruitful cognitively oriented theory of semantics cannot be constructed in terms of truth conditions.

8. Cognitive Semantics—An Interdisciplinary Endeavor

The central problem for a cognitive theory of semantics is to present an account of the mental structures that carry the meanings of words and linguistic constructions. The account should be supported by evidence available from psycholinguistics, neurolinguistics, and historical linguistics. This article has presented some of the main cognitively oriented theories of semantics. These theories have led to many successful applications, expanding our understanding of word meanings and other semantic phenomena. As can be seen from the survey, however, there is no general agreement on how mental structures are to be described.

Future developments of our understanding of cognitive processes such as perception, attention, and causal reasoning will hopefully guide semanticists into more grounded judgments concerning which models of mental structure best explain the meanings of language. Apart from understanding human communication, an increasingly important application of such models is the automatic language processing that takes place on computers, mobile phones, and other technical devices.

Further Reading

Croft, W. (2012). Verbs: Aspect and argument structure. Oxford: Oxford University Press.Find this resource:

Croft, W., & Cruse, D. A. (2004). Cognitive linguistics. Cambridge, U.K.: Cambridge University Press.Find this resource:

Gärdenfors, P. (2000). Conceptual spaces: The geometry of thought. Cambridge, MA: MIT Press.Find this resource:

Gärdenfors, P. (2014). Geometry of meaning: Semantics based on conceptual spaces. Cambridge, MA: MIT Press.Find this resource:

Haspelmath, M. (2003). The geometry of grammatical meaning: Semantic maps and cross-linguistic comparison. In M. Tomasello (Ed.), The new psychology of language (Vol. 2, pp. 211–242). Mahwah, NJ: Lawrence Erlbaum.Find this resource:

Jackendoff, R. (1987). On beyond zebra: The relation of linguistic and visual information. Cognition, 26, 80–114.Find this resource:

Lakoff, G. (1987). Women, fire, and dangerous things. Chicago: University of Chicago Press.Find this resource:

Langacker, R. W. (2008). Cognitive grammar: A basic introduction. Oxford: Oxford University Press.Find this resource:

Talmy, L. (2000). Toward a cognitive semantics: Vol. 1, Concept structuring systems. Cambridge, MA: MIT Press.Find this resource:


Barsalou, L. (1992). Frames, concepts, and conceptual fields. In A. Lehrer & E. F. Kittay (Eds.), Frames, fields, and contrasts (pp. 21–74). Hillsdale, NJ: Lawrence Erlbaum.Find this resource:

Barsalou, L. (1999). Perceptual symbol systems. Behavioral and Brain Sciences, 22, 577–609.Find this resource:

Barsalou, L. W. (2008). Grounded cognition. Annual Review of Psychology, 59, 617–645.Find this resource:

Brugman, C. (1981). Story of over. Bloomington, IN: Indiana Linguistics Club.Find this resource:

Casasanto, D. (2014). Bodily relativity. In L. Shapiro (Ed.), The Routledge handbook of embodied cognition (pp. 108–117). London: Routledge.Find this resource:

Chafe, W. (1995). Accessing the mind through language. In S. Allén (Ed.), Of thought and words: The relation between language and mind (pp. 107–125). London: Imperial College Press.Find this resource:

Clark, A. (1997). Being there: Putting brain, body and world together again. Cambridge, MA: MIT Press.Find this resource:

Clark, H. (1992). Arenas of language use. Chicago: University of Chicago Press.Find this resource:

Clark, H. (1996). Using language. Cambridge, U.K.: Cambridge University Press.Find this resource:

Croft, W. (1993). The role of domains in the interpretation of metaphors and metonymies. Cognitive Linguistics, 4, 335–370Find this resource:

Croft, W. (2001). Radical construction grammar: Syntactic theory in typological perspective. Oxford: Oxford University Press.Find this resource:

Croft, W. (2012). Verbs: Aspect and argument structure. Oxford: Oxford University Press.Find this resource:

Croft, W., & Cruse, D. A. (2004). Cognitive linguistics. Cambridge, U.K.: Cambridge University Press.Find this resource:

Croft, W., & Wood, E. J. (2000). Construal operations in linguistics and artificial intelligence. In L. Albertazzi (Ed.), Meaning and cognition: A multidisciplinary approach (pp. 51–78). Amsterdam: John Benjamins.Find this resource:

Dewell, R. (1994). Over again: Image-schema transformations in semantic analysis. Cognitive Linguistics, 5, 351–380.Find this resource:

Dor, D. (2015) The instruction of imagination. Language as a social communication technology. Oxford: Oxford University Press.Find this resource:

Dowty, D. (1991). Thematic proto-roles and argument selection. Language, 67, 547–619.Find this resource:

Evans, V., & Green, M. (2006). Cognitive linguistics: An introduction. Edinburgh: Edinburgh University Press.Find this resource:

Fillmore, C. (1968). The case for case. In E. Bach, & E. T. Harms (Eds.), Universals in linguistic theory (pp. 1–88). New York: Holt, Rinehart, and Winston.Find this resource:

Fillmore, C. (1976). Frame semantics and the nature of language. Annals of the New York Academy of Sciences: Conference on the Origin and Development of Language and Speech, 280, 20–32.Find this resource:

Fillmore, C. (1982). Frame semantics. In Linguistics in the morning calm (pp. 111–137). Seoul, South Korea: Hanshin.Find this resource:

Frege, G. (1892) Über sinn und bedeutung. Zeitschrift für Philosophie und philosophische Kritik, NF, 100, 25–50.Find this resource:

Gärdenfors, P. (2000). Conceptual spaces: The geometry of thought. Cambridge, MA: MIT Press.Find this resource:

Gärdenfors, P. (2004). Visualizing the meanings of words. In Y. Eriksson & K. Holmqvist (Eds.), Language and visualization (pp. 51–69). Lund, Sweden: Lund University Cognitive Studies 119.Find this resource:

Gärdenfors, P. (2007). Representing actions and functional properties in conceptual spaces. In T. Ziemke, J. Zlatev, & R. M. Frank. (Eds.), Body, language, and mind: Vol. 1, Embodiment (pp. 167–195). Berlin: Mouton de Gruyter.Find this resource:

Gärdenfors, P. (2014). Geometry of meaning: Semantics based on conceptual spaces, Cambridge, MA: MIT Press.Find this resource:

Gärdenfors, P., & Löhndorf, S. (2013). What is a domain? Dimensional structures versus meronymic relations. Cognitive Linguistics, 24(3), 437–456.Find this resource:

Gärdenfors, P., & Warglien, M. (2012). Using conceptual spaces to model actions and events. Journal of Semantics, 29, 487–519.Find this resource:

Geeraerts, D. (2017). Lexical semantics. Oxford Research Encylopedias.

Gibbs, R. W., & Colston, H. L. (1995). The cognitive psychological reality of image schemas and their transformations. Cognitive Linguistics, 6, 347–378.Find this resource:

Giese, M. A., & Lappe, M. (2002). Measurement of generalization fields for the recognition of biological motion, Vision Research, 42, 1847–1858.Find this resource:

Givón, T. (2001). Syntax, Vol I. Philadelphia, PA: John Benjamins.Find this resource:

Goddard, C., & Wierzbicka, A. (1994). Semantic and lexical universals: Theory and empirical findings. Amsterdam: John Benjamins.Find this resource:

Hamm, F., Kamp, H., & van Lambalgen, M. (2006). There is no opposition between formal and cognitive semantics. Theoretical Linguistics, 32, 1–60.Find this resource:

Hampton, J. A., & Winter, Y. (Eds.). (2017). Compositionalilty and concepts in linguistics and psychology. Berlin: Springer.Find this resource:

Haspelmath, M. (1997). Indefinite pronouns. Oxford: Oxford University Press.Find this resource:

Haspelmath, M. (2003). The geometry of grammatical meaning: Semantic maps and cross-linguistic comparison. In M. Tomasello (Ed.), The new psychology of language, Vol. 2 (pp. 211–242). Mahwah, NJ: Lawrence Erlbaum.Find this resource:

Heim, I. (1982). The semantics of definite and indefinite noun phrases. (Doctoral dissertation) University of Massachussetts, Amherst, MA.Find this resource:

Herskovits, A. (1986). Language and spatial cognition: An interdisciplinary study of the prepositions in English. Cambridge, U.K.: Cambridge University Press.Find this resource:

Holmqvist, K. (1993). Implementing cognitive semantics. Lund, Sweden: Lund University Cognitive Studies.Find this resource:

Jackendoff, R. (1983). Semantics and cognition. Cambridge, MA: MIT Press.Find this resource:

Jackendoff, R. (1987). On beyond zebra: The relation of linguistic and visual information. Cognition, 26, 80–114.Find this resource:

Jackendoff, R. (1990). Semantic structures. Cambridge, MA: MIT Press.Find this resource:

Jackendoff, R. (2012). Language as a source of evidence for theories of spatial representation, Perception, 41, 1128–1152.Find this resource:

Jäger, G. (2010). Natural color categories are convex sets. Amsterdam Colloquium 2009, LNAI 6042, 11–20.Find this resource:

Jäger, G., & van Rooij, R. (2007). Language structure: Psychological and social constraints. Synthese, 159, 99–130.Find this resource:

Johansson, G. (1973). Visual perception of biological motion and a model for its analysis. Perception and Psychophysics, 14, 201–211.Find this resource:

Johnson, M. (1987). The body in the mind: The bodily basis of cognition. Chicago: University of Chicago Press.Find this resource:

Kamp, H. (1981). A theory of truth and semantic representation. In J. A. G. Groenendijk, T. M. V. Janssen, & M. B. J. Stokhof (Eds.), Formal methods in the study of language (pp. 277–322). Amsterdam: Mathematical Centre Tracts 135.Find this resource:

Kamp, H., & Partee, B. (1995), Prototype theory and compositionality, Cognition, 57, 129–191.Find this resource:

Lakoff, G. (1987). Women, fire, and dangerous things. Chicago: University of Chicago Press.Find this resource:

Lakoff, G., & Johnson, M. (1980). Metaphors we live by. Chicago: University of Chicago Press.Find this resource:

Lakoff, G., & Turner, M. (1989). More than cool reason: A field guide to poetic metaphor. Chicago: University of Chicago Press.Find this resource:

Landau, B., & Jackendoff, R. (1993). “What” and “where” in spatial language and spatial cognition. Behavioral and Brain Sciences, 16, 217–265.Find this resource:

Langacker, R. W. (1987). Foundations of cognitive grammar, Vol. 1. Stanford, CA: Stanford University Press.Find this resource:

Langacker, R. W. (2008). Cognitive grammar: A basic introduction. Oxford: Oxford University Press.Find this resource:

Levin, B., & Rappaport Hovav, M. (2005). Argument realization. Cambridge, U.K.: Cambridge University Press.Find this resource:

Löbner, S. (2014). Evidence for frames from human language. In T. Garnerschlag, D. Gerland, R. Osswald, & W. Petersen (Eds.), Frames and concept types: Applications in language and philosophy (pp. 23–67). Dordrecht, Netherlands: Reidel.Find this resource:

Lyons, J. (1977). Semantics. Cambridge, U.K.: Cambridge University Press.Find this resource:

Mandler, J. M. (2004). The foundations of mind: Origins of conceptual thought. New York: Oxford University Press.Find this resource:

Matlock, T. (2004). Fictive motion as cognitive simulation. Memory and Cognition, 32, 1389–1400.Find this resource:

Mervis, C., & Rosch, E. (1981). Categorization of natural objects. Annual Review of Psychology, 32, 89–115.Find this resource:

Michaelis, L. A. (2017). Meanings of constructions. Oxford Research Encyclopedias.

Osherson, D. N., & Smith, E. E. (1981). On the adequacy of prototype theory as a theory of concepts. Cognition, 12, 299–318.Find this resource:

Paradis, C. (2005). Ontologies and construals in lexical semantics. Axiomathes, 15, 541–573.Find this resource:

Peirsman, Y., & Geeraerts, D. (2006). Metonymy as a prototypical category. Cognitive Linguistics, 17, 269–316.Find this resource:

Pelletier, F. J. (2016). Semantic compositionality. Oxford Research Encyclopedias.

Petitot, J. (2011). Cognitive Morphodynamics. Bern: Peter Lang.Find this resource:

Pickering, M. J., & Garrod, S. (2004). Toward a mechanistic psychology of dialogue. Behavioral and Brain Sciences, 27, 169–190.Find this resource:

Pickering, M. J., & Garrod, S. (2013). An integrated theory of language production and comprehension. Behavioral and Brain Sciences, 36, 329–347.Find this resource:

Pulvermüller, F. (2003). Neuroscience of language: On brain circuits of words and serial order. Cambridge, U.K.: Cambridge University Press.Find this resource:

Putnam, H. (1975). The meaning of “meaning.” In K. Gunderson (Ed.), Language, mind and knowledge (pp. 131–193). Minneapolis: University of Minnesota Press.Find this resource:

Radvansky, G. A., & Zacks, J. M. (2014). Event cognition. Oxford: Oxford University Press.Find this resource:

Rappaport Hovav, M., & Levin, B. (2010). Reflections on manner/result complementarity. In M. Rappaport Hovav, D. Doron, & I. Sichel (Eds.), Lexical semantics, syntax, and event structure (pp. 21–38). Oxford: Oxford University Press.Find this resource:

Rosch, E. (1975). Cognitive representations of semantic categories. Journal of Experimental Psychology: General, 104, 192–233.Find this resource:

Rubio Fernandez, P. (2007). Suppression in metaphor interpretation: Differences between meaning selection and meaning construction. Journal of Semantics, 24, 345–371.Find this resource:

Runesson, S. (1994). Perception of biological motion: The KSD-principle and the implications of a distal versus proximal approach. In G. Jansson, S.-S. Bergström, & W. Epstein (Eds.), Perceiving evens and objects (pp. 383–405). Hillsdale, NJ: Lawrence Erlbaum.Find this resource:

Smith, E. E., & Medin, D. L. (1981). Categories and concepts. Cambridge, MA: Harvard University Press.Find this resource:

Sperber, D., & Wilson, D. (1986). Relevance: Communication and cognition. Oxford: Blackwell.Find this resource:

Stalnaker, R. (1978). Assertion. Syntax and Semantics, 9, 315–332.Find this resource:

Talmy, L. (1988). Force dynamics in language and cognition. Cognitive Science, 12, 49–100.Find this resource:

Talmy, L. (1996). Fictive motion in language and “ception.” In P. Bloom, M. A. Peterson, L. Nadel, & M. F. Garrett (Eds.), Language and space (pp. 211–276). Cambridge, MA: MIT Press.Find this resource:

Talmy, L. (2000). Toward a cognitive semantics: Vol 1, Concept structuring systems. Cambridge, MA: MIT Press.Find this resource:

Trier, J. (1931). Der deutsche wortschatz im sinnbezirk des verstandes. Heidelberg: Winter.Find this resource:

Tyler, A., & Evans, V. (2001). Reconsidering prepositional polysemy networks: The case of over. Language, 77, 724–765.Find this resource:

Vandeloise, C. (1986). L’espace en français: sémantique des prépositions spatiales. Paris: Editions du Seuil.Find this resource:

Van der Auwera, J., & Plungian, V. A. (1998). Modality’s semantic map. Linguistic Typology, 2, 79–124Find this resource:

Van der Gucht, F., Klaas, W., & De Cuypere, L. (2007). The iconicity of embodied meaning. Polysemy of spatial prepositions in the cognitive framework. Language Sciences, 29, 733–754.Find this resource:

Vicente, A., & Faljum, I. L. (2017). Polysemy. Oxford Research Encyclopedias.

Warglien, M., & Gärdenfors, P. (2013). Semantics, conceptual spaces, and the meeting of minds. Synthese, 190, 2165–2193.Find this resource:

Warglien, M, Gärdenfors, P., & Westera, M. (2012). Event structure, conceptual spaces and the semantics of verbs. Theoretical Linguistics, 38, 159–193.Find this resource:

Wilson, D., & Carston, R. (2007). A unitary approach to lexical pragmatics: Relevance, inference, and ad hoc concepts. In N. Burton-Roberts (Ed.), Pragmatics (pp. 230–259). London: Palgrave.Find this resource:

Zacks, J., & Tversky, B. (2001). Event structures in perception and conception. Psychological Bulletin, 127(1), 3–21.Find this resource:

Zlatev, J. (1997). Situated embodiment: Studies in the emergence of spatial meaning. Stockholm: Gotab.Find this resource:

Zlatev, J. (2003). Polysemy or generality? Mu. In H. Cuyckens, R. Dirven, & J. R. Taylor (Eds.), Cognitive approaches to lexical semantics (pp. 447–494). Berlin: Mouton de Gruyter.Find this resource:

Zwarts, J. (1997). Vectors as relative positions: A compositional semantics of modified PPs. Journal of Semantics, 14, 57–86.Find this resource:

Zwarts, J. (2010). Semantic map geometry: Two approaches. Linguistic Discovery, 8, 377–395.Find this resource:

Zwarts, J., & Gärdenfors, P. (2016). Locative and directional prepositions in conceptual spaces: The role of polar convexity. Journal of Logic, Language and Information, 25, 109–138.Find this resource:


(1.) In the literature, the notions of concept and category are used in different ways. Here, categorization is taken to be a rule for classifying objects, while concept is used more generally for mental representations.

(2.) Van der Gucht, Klaas, & De Cuypere (2007) derive these positions historically from Locke and Leibniz respectively.

(3.) For empirical support for this position see Rubio Fernandez (2007).

(4.) Dewell (1994) proposes a slightly different schema.

(5.) A vector has a starting point, a direction, and a length. In physics, vectors are used to represent forces, where the length of the vector stands for the strength of the force, and changes, where the length of the vector stands for the amount of change.