COMMON UNDERLYING TENETS

The central idea in usage-based models is that a user's language emerges as a result of exposure to numerous usage events (Kemmer & Barlow, Reference Kemmer, Barlow, Barlow and Kemmer2000), that is, situated instances of the language user understanding or producing language to convey particular meaning in a specific communicative situation. When a speaker engages in communication, it is assumed that she is attempting to achieve specific interactional goals using intentionally chosen linguistic strategies aimed at members of her speech community. Tying usage events to particular speech communities reflects the understanding that actual language use is culturally and contextually embedded.

All usage-based approaches adhere to five key tenets: (a) The primary purpose of language is communicative; communicative use shapes language itself. As Tomasello (Reference Tomasello2003) argued in relation to first language learning, “language structure emerges from language use” (p. 5). (b) Natural language always occurs in context, and the user's choices in crafting an utterance are influenced by an array of contextual factors. Context itself is a complex construct that transcends a compositional approach in which additional dimensions can simply be added without recognizing subtle, interacting linguistic reflexes that occur in relation to multidimensional context. For instance, all usage-based models have recognized the audience or the participants in an interaction as a major aspect of context (for genre, see Martin & Rose, Reference Martin and Rose2008; for listener expectation, see Chafe, Reference Chafe1994; for “ground,” see Langacker, Reference Langacker1987, Reference Langacker1991b; Taylor, Reference Taylor2004). Subtle changes in the relationship between the speaker and the audience result in changes in the speaker's language choices. (c) Language is learned. Patterns of usage, including collocational information (at the level of grammatical patterns as well as the word level) and frequency information, are held to be central to the learning of the system. This tenet entails a rejection of a separate language module with a preestablished grammatical configuration that is activated or triggered by exposure to salient language data. (d) Meaning is not housed solely in lexical items. Grammatical patterns are meaningful in and of themselves (e.g., Givón, Reference Givón2001; Langacker, Reference Langacker, Robinson and Ellis2008a; Martin, Reference Martin2002). (e) Language can be accurately and fully accounted for through a monostratal model, that is, all syntactic patterns can be accounted for without assuming two (or more) levels, such as deep structure and surface structure. Alternative syntactic patterns, such as active versus passive, are separate, meaningful patterns, not transforms of each other. (See Zyzik, Reference Zyzik2009, for further discussion of usage-based approaches, especially on the importance of input frequency.)

SYSTEMIC FUNCTIONAL LINGUISTICS

SFL began with the work of M. A. K. Halliday in the 1970s, largely in response to more formal, structuralist approaches that discounted the centrality of communication and context in shaping language. The main goal of Halliday's model has been to account for culturally situated linguistic communication that is understood as a transmitter not only of propositional information but also of larger cultural values, and as a tool for getting things done (e.g., Byrnes, Reference Byrnes2009; Halliday, Reference Halliday1994). In this model, language is inextricably bound to human social behavior, culture making, and culture transmittal, including transmission of status and power relations.Footnote ² A central goal is to understand how ordinary language as it is used in everyday practice transmits systematic patterns of culture. (Halliday, Reference Halliday1978; Halliday & Matthiessen, Reference Halliday and Matthiessen2004).

Much of more recent SFL work views language through the filter of genre (Martin, Reference Martin2002, Reference Martin2009). Genre is understood as a specific text type, aimed at a narrowly specified audience (e.g., scientific academics) with specific communicative goals (e.g., persuading, informing) in particular formats (e.g., scientific journals). The emphasis is on analyzing text produced by speakers and writers as they communicate in a given genre. Thus, SFL clearly situates all analysis in a defined sociocultural context. Every language has numerous genres, and every genre is language specific. Each genre is held to utilize slightly different, conventionalized patterns of linguistic resources, and the patterned occurrence of these linguistic bundles determines whether a text represents an appropriate example of that genre.

Below the overarching level of genre, SFL posits a tripartite division of context: (a) field, which concerns what is being talked about and what is going on in terms of the social activity of a situation; (b) tenor, which refers to how social roles and relations are being enacted; and (c) mode, which relates to the opportunities and constraints on deployment of linguistic resources associated with the situated text, including resources and constraints inherent in spoken versus written discourse (e.g., Byrnes, Reference Byrnes2009; Halliday & Matthiessen, Reference Halliday and Matthiessen2004; Martin, Reference Martin2009). For example, consider how these three contextual elements might play out in an article about the battle of Gettysburg written for an academic journal: The information about the battle (and argumentation concerning its import to the Civil War) as well as the components of description (informing) and persuasion (presenting an opinion and attempting to convince the audience it is an opinion worth considering) would represent the field; the positioning of the author as an academic authority on the subject communicating to an informed readership who share similar expectations about how such information and argumentation is presented would represent the tenor; and the expectations and stylistic constraints of the academic journal would represent the mode. Each of these dimensions of context is in turn related to a metafunction, described in the following paragraph, which is realized in terms of specific aspects of linguistic expression. Of the three usage-based approaches, SFL has developed the most delineated constructs concerning how specific aspects of context relate to the linguistic realizations found in a situated text (e.g., Martin, Reference Martin2002, Reference Martin2009; Martin & Rose, Reference Martin and Rose2008).

In terms of actual linguistic expression, the deployment of various linguistic resources or patterns is argued to enact three different communicative metafunctions, all of which exist simultaneously and interactively in any text: ideational, what is being done and what the text is about or what meaning the text is expressing; interpersonal, linguistic cues indicating speaker stance and relationship to audience; and textual, or what makes a stretch of language hang together as a coherent whole and how participants are tracked through the discourse, information flow, and so on. (The constructs of theme/rheme are classic examples of the textual metafunction; e.g., Halliday, Reference Halliday1978; Halliday & Matthiessen, Reference Halliday and Matthiessen1999; Martin & Rose, Reference Martin and Rose2008).

SFL represents linguistic patterning as involving networks of closed systems of options offered by the particular language for the purpose of communication. These are the systematic bundles of linguistic forms associated with the field, tenor, and mode of each genre.Footnote ³ Halliday clearly saw meaning as the shaper of language: “[I]n a systemic grammar every category (i.e., theoretical concept) is based on meaning: it has a semantic, as well as a formal, lexico-grammatical reactance” (Halliday & Matthiessen, Reference Halliday and Matthiessen1999, pp. 3–4). Choice of lexicogrammatical form is understood as a meaning-making choice. The focus is on identifying linguistic patterns that are typically employed to create a situated, coherent text (as perceived by the intended audience) that matches a particular communicative context (defined in terms of genre with specifically defined audience and register). Although audience and audience expectation have been identified as part of tenor and the interpersonal dimension, the emphasis has been on meaning being primarily a property of the text itself rather than deriving through dynamic interpretation and inferencing on the part of the listener.

Halliday believed that the form of language is explicable almost solely in terms of communicative function; he rejected “the psychological or physiological” as motivating factors (Halliday & Matthiessen, Reference Halliday and Matthiessen1999, p. 7). Indeed, he stated, “Instead of explaining language by reference to cognitive processes, we explain cognition by reference to linguistic processes” (p. 7). An example of this focus is his definition of theme/rheme; theme is simply seen as the departure point of the sentence. The speaker's choice of theme is primarily related to the textual metafunction, which has to do with information flow and how the speaker chooses to lay out the message. Importantly, the analysis of theme/rheme does not address underlying cognitive or neuro-physical motivations. Halliday's almost exclusive concern with language as a social-cultural phenomenon and disregard for cognitive and physical-neural dimensions is an important factor that sets SFL apart from discourse functionalism and CL.

Halliday's analysis of the passive is tied to his understanding of theme and reflects his belief in the unimportance of cognitive motivation. He has noted that English speakers have an awareness of the meaning expressed when something is in first position and, moreover, that theme is usually associated with the actor role unless there is good reason not to do so. He further noted that the passive allows the speaker to disassociate theme from actor in order to change focus or omit the actor, but his analysis offered no explicit discussion of general cognitive motivation for these choices.

The emphasis on using genre as the filter through which to analyze syntax has tended to lead SFL linguists away from general descriptions of grammatical constructions as they are used to communicate across a range of text types. The close ties between narrow context and the situatedness of linguistic choices are another factor that distinguishes SFL from CL. It should be noted that a relatively new development in SFL has involved close analysis of the discourse of multiple areas and an increased understanding of generic structures and intertextual relations among various genres (Martin, Reference Martin2002).

The majority of the applications of SFL to issues in L2 have come from the work of Halliday's followers in Australia. In Australian educational linguistics, genres have been defined as “staged, goal-oriented social processes” (Martin, Reference Martin and Christie1999, p. 127). Fuller understanding of the linguistic and discourse patterns generally employed within a particular genre has served as the basis for genre-based instruction that addresses the notion that genres unfold in stages and that experienced readers or listeners are likely to experience a sense of frustration or negative evaluation when phases do not unfold as expected. Christie (Reference Christie, Whittaker, O'Donnell and McCabe2006) also emphasized the importance to L2 teachers of the knowledge about language (KAL), including knowledge of grammar (as long as grammar is understood in terms of making appropriate grammatical choice in order to create effective, situated meaning) as an important result of genre-based teaching. Research into field, or the particular activity that is being accomplished through discourse, has explored several kinds of school and workplace discourse, including the areas of mathematics, science, geography, history, and English for academic purposes (see Martin, Reference Martin2009, for a fuller discussion). In general, there has been a recent shift to analyze longer texts as a series of smaller genre complexes that unfold in stages (see Christie, 1999; Martin, Reference Martin2009, for classroom discourse). These insights have been used to teach both L2 written discourse and spoken discourse. For overviews of application of genre to L2, see Byrnes (Reference Byrnes, Galloway and Cothran2006, Reference Byrnes2009).

In the United States, Byrnes and her colleagues (e.g., Byrnes, Reference Byrnes2009; Byrnes, Maxim, & Norris, in press) have successfully developed a genre-based curriculum for advanced learners of German as a foreign language. The focus is on close reading of genre-defined texts and analysis of the situated choices writers make in order to create meaning. Thus, the reading-writing-speaking connections are of central concern. The close reading involves intensive analysis of genre-based usage of lexicon, grammatical choices, and rhetorical structure in reading and writing. Texts are presented as “models for the meaning-making resources available in the language system; texts are instances of linguistically construed cultural situations that themselves are reflective of the entire cultural system” (Byrnes et al., in press). Schleppegrell and her colleagues have also applied SFL to issue in L2 writing (e.g., Achugar, Schleppegrell, & Oteíza, Reference Achugar, Schleppegrell and Oteíza2007; Schleppegrell, Reference Schleppegrell2004, Reference Schleppegrell and Bazerman2008; Schleppegrell, Greer & Taylor, Reference Schleppegrell, Greer and Taylor2008). Colombi (Reference Colombi2009) has used concepts from SFL in her work with heritage language learners.

Both in theory and in application, SFL has a strong tendency to focus on the surface patterns of the text, such as use of nominalizations, hypotactic versus paratactic sentence structure, amount of pronoun usage, ellipsis, lexical repetition, and semantic networking. While recognizing audience expectation and participants as part of culturally situated context, the assumption seems to be that, for the most part, meaning exists in the text (Halliday, Reference Halliday1994, Reference Halliday and Webster2002; Halliday & Hasan, Reference Halliday and Hasan1976; Hoey, Reference Hoey1991). Issues such as linguistic elements underspecifying the intended interpretation, the importance of inferencing, and the use of background knowledge in both the interpretation and production of discourse are not addressed. This lack of emphasis on the listener as an active creator of meaning, as well as lack of emphasis on human cognition (such as human category formation, organization of memory, and use of background knowledge) as central to shaping language and to meaning making, have been key distinctions between SFL and the other two usage-based approaches considered here.

DISCOURSE FUNCTIONALISM

While SFL represents a model of language clearly distinguishable from discourse functionalists and CL, these last two are more appropriately seen as models located on a continuum. Historically, the two approaches have exerted considerable influence over each other, and some of the same scholars have been associated with both (e.g., Wallace Chafe and Paul Hopper).

The discourse functionalists have sought to develop a relatively detailed description of grammar that attends closely to the relationships among grammar, discourse, and the producers of the discourse. Grammar usage is seen through the lens of discourse and meaningful speaker choice. Their understanding of the discourse producers also includes an explicit recognition of the cognitive dimension, in addition to the social-cultural. For instance, Chafe's (Reference Chafe1994, Reference Chafe, Butler, de los Ángeles Gómez-González and Doval-Suárez2005, Reference Chafe, Zelazo, Moscovitch and Thompson2007) explanation of English speakers' choice of determiners is largely based on the role of memory (including information retrievable from schema-based knowledge) and attention. In his analysis of spoken English, Chafe (Reference Chafe, Schiffrin, Tannen and Hamilton2001) noted that language is produced in relatively short intonation units, which he hypothesized are constrained by the one new idea per unit constraint, which is in turn tied to short-term memory and processing constraints.

Although there are a number of important researchers associated with discourse functionalism, here I will focus on Givón's contributions as representative of the approach. Givón (e.g., Reference Givón1995, Reference Givón2001) recognized the importance of discourse context, although he has not developed a formalism for the connections between context and grammar as found in SFL. In distinction from SFL, in considering context, he does not recognize genre as an overarching construct which is central to understanding the connection between syntax and meaning. In further contrast to SFL, he has argued for the importance of taking more than just the social-cultural-communicative dimensions into account in order to fully understand why language is as it is. Givón (Reference Givón2001) has taken what might be understood as a broadly biological approach to functionalism. He argued that biologists long ago recognized that they could not adequately understand a particular part of an organism, for instance, the human heart, without understanding how it functioned within the organism as a whole. On analogy, he has argued that it is impossible to adequately understand language (or any particular formal structure found in language) without understanding how language functions in terms of the purposeful social and communication activities that its producers use it to perform. Moreover, language cannot be understood apart from the full human organism itself. At a more specific level, Givón looks to general cognitive processes as important in shaping language. He sees cognition and the brain (including issues of language processing), as well as communication and social interaction, as some of the central, natural parameters that shape language (Givón, Reference Givón1995, Reference Givón2001). A good example of integrating cognitive motivation into his explanation of grammar is his discussion of an animacy hierarchy and its relation to thematic roles. In terms of cognitive saliency rooted in evolution and survival, humans are especially concerned with their own actions and the forces that impinge upon them; this is reflected in sentence structure and thematic roles. At the most basic level, agents are usually animate, preferably human; patients or themes are more likely to be inanimate. Nonanimate forces, such as earthquakes, which affect humans in ways that are similar to animate beings, are less typical but possible agents.

Givón's (Reference Givón1995, Reference Givón2001) analysis of iconicity, or the notion that a resemblance exists between the shape of language and the concepts being represented, also illustrates his inclusion of the cognitive dimension as an important factor in determining the structure of language. He noted that formalists tend to dismiss iconicity in language as untenable because of the tremendous morphosyntactic variation found across languages. One of the seeming problems with iconicity has been the basic claim that word order essentially corresponds to the order of thought. On the face of it, this claim appears false, given that languages demonstrate many word orders, while human conceptualization or thought order is presumably universal. Givón has argued that the existence of competing pragmatic and communicative principles accounts for the variation. For instance, he argued that the universal principle of focused elements coming first (or iconically, the information the speaker finds most pressing coming at the beginning of the sentence) is often in competition with the well-documented principle of given information coming at the beginning of the sentence (iconically, starting with what is known). Competition between these two communicative and cognitive forces accounts for certain patterns of variation. Givón explained the failure to find obvious iconic universals, such as a single word order for all languages, by drawing on the notion that humans represent complex biological systems that have evolved and adapted to multiple, often competing pressures over time. Because of the complexity of the process of evolutionary development in response to multiple forces, simple one-to-one relationships between form and function are rare. He further argued that if one views iconicity at a more abstract level and takes competing forces into account, iconicity can be seen as a fundamental force shaping language. Note that Givón's explanation of the speaker's choice of what information comes at the beginning of a sentence was essentially a reanalysis of theme/rheme. Givón's attribution of speaker choice being influenced by cognitive principles contrasts with SFL.

While firmly arguing for meaning, discourse, communicative, and cognitive motivations as the genesis of grammar, Givón did not reject formal grammatical structure as a reality of language. He argued that competing motivations (communicative, cognitive, social-cultural, etc.) often lead to a situation in which grammar becomes established as a conventionalized system (e.g., 2001). In terms of analysis of specific grammatical constructions, Givón's work on the role of passive versus active is representative. Having analyzed multiple instances of discourse in several languages, Givón argued that passive accomplishes several functions: (a) de-emphasizes the agent, (b) places focus on the theme/patient, (c) allows for continued focus on a topical participant that may not be agentive, and (d) tends to make the action more static.

Taking a strong comparative linguistic typological approach, many of his studies consider the frequency distribution of features in different languages, striking a balance between what is common across languages and their individual characteristics. In distinction from SFL, discourse functionalism focuses on syntax as coding meaning in a way that is generalizable across discourse contexts rather than concentrating on the typical syntactic patterns that co-occur in a particular genre. Discourse functional analyses do not focus on issues such as patterns of ellipsis or tendency to use nominalizations in a particular text type.Footnote ⁴

We turn now to applications of discourse functionalists to issues in second language learning and teaching. Setting aside work which focuses on speech acts, it is probably fair to say that a majority of discourse applications to issues in L2 learning have been influenced by discourse functionalists. Celce-Murcia (Reference Celce-Murcia, Hinkle and Fotos2002) acknowledged the important contributions of Givón in her discussion of teaching L2 grammar from a discourse/functional perspective. Tomlin's (Reference Tomlin and Odlin1994, Reference Tomlin, Nuyts and Pederson1997) influential red fish/blue fish experiments, which demonstrated that native English speakers' use of active constructions when the focus is on the agent and passive when the focus in on the patient grew out of Givón's analysis. This understanding of active versus passive has been widely accepted by L2 discourse analysts (e.g., Celce-Murcia & Larsen-Freeman, Reference Celce-Murcia and Larsen-Freeman1999; Celce-Murcia & Olshtain, Reference Celce-Murcia, Olshtain and Hinkle2005; Larsen-Freeman, Reference Larsen-Freeman, Hinkle and Fotos2002; McCarthy & Carter, Reference McCarthy, Carter, Hinkle and Fotos2002). Many studies by Celce-Murcia (e.g., Reference Celce-Murcia, Hinkle and Fotos2002) and her students have analyzed speaker choice between competing constructions, such as used to versus would, in naturally occurring discourse; these studies assumed that speaker choice in lexical constructions and syntax has a general communicative meaning across a wide range of text types. Carter, Hughes, and McCarthy (Reference Carter, Hughes and McCarthy2000) provided analysis of numerous grammatical patterns, from aspectual choices to use of relative clauses, in terms of general communicative functions. They pointed out a range of text types as good sources of naturally occurring uses of a particular pattern, but did not tie the pattern to a particular genre. For instance, their discussion of modal usage included analysis of newspaper horoscopes and print advertisements, two quite different genres.

COGNITIVE LINGUISTICS

CL inherits many basic assumptions from discourse functionalism. For instance, there has been general recognition of the importance of discourse-level phenomena, although relatively little discourse analysis has actually been carried out by cognitive linguists. As Sanders & Spooren (Reference Sanders, Spooren, Geeraertz and Cuyckens2005), in their review of the discourse-CL connection, noted, “At the moment, the cognitive linguistic study of discourse is still more of a promising challenge to linguists and students of discourse, rather than a well-established part of everyday cognitive linguistic practice” (p. 917).Footnote ⁵ (Exceptions to this trend include the work by, e.g., Cameron, Reference Cameron, Tyler, Kim and Takada2008; Dancygier, Reference Dancygier, Tyler, Kim and Takada2008; Moder, Reference Moder, Achard and Kemmer2004, Reference Moder, Tyler, Kim and Takada2008). CL also recognizes the role of iconicity, particularly as it relates to metonymy and some aspects of primary metaphor.

What sets CL apart from the discourse functionalist approach is its primary commitment to developing a model of language that reflects human neural-physical architecture and human interaction with the social-physical world. By extension, CL assumes that reflexes of the same neuro-cognitive principles should operate in multiple areas of language. This is what Lakoff (Reference Lakoff1990) termed the “generalization commitment” (p. 39). Thus, it is of theoretical import for CL that we find occurrences of polysemy in lexical items, such as give (Newman, Reference Newman1996); functional morphemes, such as tense markers (Tyler & Evans, Reference Tyler, Evans, Pütz, Dirven and Niemeier2001) and -er (Panther & Thornburg, Reference Panther, Thornburg, Pütz, Niemeier and Dirven2001); and grammatical constructions (Goldberg, Reference Goldberg1995). Moreover, CL pays a great deal of attention to evidence from psychology and neuroscience (e.g., Coventry & Garrod, Reference Coventry and Garrod2004; Coventry & Guijarro-Fuentes, Reference Coventry, Guijarro-Fuentes, Robinson and Ellis2008; Gibbs, Reference Gibbs2006; Taylor, Reference Taylor2003; von Stutterheim, Reference von Stutterheim and Foster-Cohen2003) concerning human category formation, memory capacity and the role of input frequency in automaticity (e.g., Bybee, Reference Bybee, Janda and Joseph2003, Reference Bybee2006; Schmid, Reference Schmid, Geeraerts and Cuyckens2007) and in first language acquisition (e.g. Tomasello, Reference Tomasello2003; Lieven & Tomasello, Reference Lieven, Tomasello, Robinson and Ellis2008). Tomasello (Reference Tomasello2003) also argued for a unique genetic endowment involving joint attention and ability to develop a theory of the mind as essential social components required for symbolic communication and hence language.

CL is deeply rooted in the notion of embodied meaning (e.g., Gibbs, Reference Gibbs1994, Reference Gibbs2006; Lakoff, Reference Lakoff1987; Lakoff & Johnson, Reference Lakoff and Johnson1999; Langacker, Reference Langacker1987, Reference Langacker1991b). Cognition itself is understood as being shaped by humans' unique neural and physical architecture in interaction with the social-spatial-physical environment. Consider one aspect of humans' unique physical architecture, its consequences for human conceptualization of the world, and one of its reflexes in language: Humans stand on their hind legs. This means that beginning in infancy, humans are acutely aware of balance, the force of gravity, and the salience of being “up” for being able to see more, grasp objects more freely, move more quickly, and so on. Being able to stand upright makes “up-down” and resistance to gravity meaningful to humans in ways they are simply not to, say, snakes or hummingbirds. Standing, the quintessential human experience of up-ness and resistance to gravity, figures into many metaphors that have to do with persistence and self-esteem, for example, stand up for yourself, stand for something you believe in, this aggression will not stand. Grady (Reference Grady, Geeraertz and Cuyckens2007) argued that a primary metaphor found in many languages is “persistence is remaining erect.” The salience of up versus down for humans and its reflex in language was well documented in Lakoff and Johnson's (Reference Lakoff and Johnson1980) study on orientation metaphors.

In terms of human perceptual systems, since the Gestalt psychologists, it has been established that the human visual system is configured such that, when viewing a spatial array, humans tend to focus on smaller, more mobile entities (focus) that are located in terms of larger, more stable entities (ground). CL has emphasized that this aspect of the human visual system is reflected in language (Talmy, Reference Talmy2000a, Reference Tannenb), as in the prepositional system (The bike is by the school) in which the smaller, more moveable “bike” is the focus and located in relation to the larger, less mobile “school” (the ground). Consider the oddness of the sentence? The school is by the bike, in which the focus/ground relationship is reversed. Talmy (Reference Talmy2000a) argued that figure/ground relations account for many patterns of clause ordering in complex sentences. Langacker (Reference Langacker1987, Reference Langacker1991b) also argued that the focus/ground relation is ubiquitous in larger language structures, for example, the sentential subject acts as the focus element in relation to the predicate, which acts as the ground.

Basic human interactions with the physical and social world provide a foundation for human conceptual and cognitive representations which are in turn reflected in language. Humans use their understanding of the external, physical world as a framework for representing internal phenomena—emotions, self-reflective concepts, and more abstract concepts. The physical underpinning for conceptualization is reflected throughout language. In other words, humans regularly think and talk about internal phenomena in terms of the external, sociophysical. This is central to conceptual metaphor theory, one of the constructs most closely associated with CL (e.g., Grady, Reference Grady1997, Reference Grady, Gibbs and Steen1999, Reference Grady, Geeraertz and Cuyckens2007; Lakoff & Johnson, Reference Lakoff and Johnson1980, Reference Lakoff and Johnson1999). Within CL, metaphor is understood as a ubiquitous part of thought and language. Rejecting the traditional analysis that metaphor is based solely on resemblance, Grady has argued that much of metaphor is based in fundamental human experiences of the world in which two independent phenomena repeatedly co-occur, such that they become closely associated in memory. Once this association has been established, humans appear to think about the more abstract phenomena in terms of the more concrete. A straightforward example is the everyday observation that as more liquid is added to a container, there is a simultaneous rise in the level of the liquid. This ubiquitous observation leads to a tight mental association between the concepts of “amount” and vertical elevation, that is, “up-down.” As a result, we think and talk about an increase in amount in terms of an increase in vertical elevation (Grady, Reference Grady, Gibbs and Steen1999; Lakoff & Johnson, Reference Lakoff and Johnson1999), as in The price of potatoes is up/rising or Executive compensation remains elevated. Grady (Reference Grady, Gibbs and Steen1999) termed this type of metaphor primary metaphor, which is based on experiential correlation. After examining more than 20 languages, many of which are not historically related, Grady (Reference Grady, Gibbs and Steen1999) argued that the primary metaphor, “more is up,” (p. 87) along with a dozen other primary metaphors, appears to be universal. CL's emphasis on cognition and embodied meaning allows exploration of many linguistic phenomena which are of no theoretical import to SFL nor discourse functionalists, such as recurring instances of the same cognitive processes at different levels of language, metaphor, and polysemy.

CL maintains that the multiple, rich, interacting cognitive capacities of humans are sufficient to account for language. In the general area of classification and organization of memory, some of the key cognitive capacities include humans' incredibly well developed abilities at classifying, pattern finding, and a huge, highly structured memory capacity. See Taylor (Reference Taylor2003, Reference Taylor2004) for a detailed discussion of categorization and CL. Humans are particularly adept at forming categories, that is, at perceiving similarities over several specific instances of entities or events which may differ substantially in terms of surface manifestations (e.g., a wide range of entities are perceived as chairs even though they differ in terms of shape [a desk chair versus a bean bag chair], materials [wood versus soft fabric], and numerous other physical properties). Moreover, human memory is richly patterned and networked. Work by Rosch and her colleagues (e.g., Rosch, E. & C. Mervis, Reference Rosch and Mervis1975) has established that one key aspect of human categorization (and hence memory structuring) involves centrality or prototype effects and radial categories. Radial categories are organized around a set of attributes abstracted from best exemplars of the category (this is the central prototype); other entities that are perceived to resemble the prototype in important ways (including how humans interact with an entity or how a type of event affects humans) are part of the category but understood to be less good examples. The classic case is that for English speakers from North American the general attributes of a small songbird, such as a robin, represent a prototypical bird while penguins and ostriches represent less good exemplars. Goldberg (Reference Goldberg1995) argued that the same principles of categorization and centrality effects apply to syntactic constructions and the particular verbs that occur in the construction. For instance, in the double object construction (Subj-X Verb Obj-Y Obj-Z), the central meaning is X caused Y to receive Z, and the verb that most closely matches this meaning is give. Corpus analyses have revealed that give is the most commonly occurring verb in the double object construction. Thus a sentence such as Yun gave Hiro the paper represents a prototypical instance of the construction. A sentence such as Mari willed Miko her house can be explained through systematic principles of extension, but is a less central exemplar of the construction, as it does not entail physical transfer of an entity from one participant to another. Goldberg further argued that the predominance of give in the double object construction in motherese provides the young child an anchor for learning the meaning of the construction. First, the most central case is learned; extended uses are gradually added to the child's repertoire as they encounter adult use of less prototypical verbs, such as bake or fax in the construction.

Another component of category formation is the ability to recognize patterns and perceive resemblances among varied entities, events, and even attributes that are simultaneously understood as different. This is the heart of conceptual metaphor, conceptualizing one entity in terms of another. When we say a person is a twig, we are reflecting our ability to see a resemblance between the fleshless structure of a tree branch and a human with exposed bone structure due to very little body fat. Goldberg (Reference Goldberg1995, Reference Goldberg2006) argued that syntactic constructions can also be extended metaphorically. For instance, Lou told Chris a story is analyzed as a metaphoric extension of the double object construction; the acceptability of verbs of communication in this construction is accounted for by the conduit metaphor for communication (Reddy, Reference Reddy and Ortony1993), by which English speakers perceive a similarity between exchanging information and exchanging physical entities.

Memory is also organized in terms of flexible, dynamic schemas that involve abstract, hierarchical generalizations built up over a number of specific instances; both specific instances and the abstracted generalizations are represented in memory. Psychologists have represented schemas in various ways from scripts for specific types of recurring sociocultural events, such as going to a restaurant, to development of abstract concepts such as hot, to knowledge of the general rhetorical structure and language of fairy tales. The point is that human memory is richly, systematically patterned. Moreover, any one entity or event or experience can be part of multiple schemas, thus memory is massively interconnected and redundant. Humans use established, schematic patterns to help interpret and organize new information. Richly structured, massively interconnected memory has numerous reflexes in language. Langacker (e.g., Reference Langacker1987, Reference Langacker2008a) has argued that our knowledge of sentence structures is schematic, with some representations of a particular pattern being quite specific and lexically (and phonetically) filled in while others are highly abstract and general. In other words, humans have both vast lists of sentences stored in memory (including information about the contexts in which the sentences occurred) and generalizations or “templates” representing abstract sentential structure. Crucially, these templates are not represented as being disconnected from meaning. For example, a template for an active, transitive sentence might be represented as something like [Subject-X Verb Object-Y] connected to a general meaning of [X-Actor/Initiator causes some effect on Y-Patient]. The argument that syntactic patterns are best understood as high-level schematic abstractions is also at the heart of construction grammar, Adele Goldberg's particular take on syntax (Goldberg, Reference Goldberg1995, Reference Goldberg2006). Humans can tap into the schema at whatever level of abstractness is most useful at the moment. Words are argued to act as underspecified prompts into encyclopedia memory (Evans & Green, Reference Evans and Green2006; Langacker, Reference Langacker1987, Reference Langacker1991b; Taylor, Reference Taylor2004). Langacker (e.g., Reference Langacker1987, Reference Langacker1991a, Reference Langacker2008a) and cognitive linguists generally argue against strict divisions between lexical items and syntactic patterns, as well as strict divisions between semantics and pragmatics. Fauconnier and Turner (Reference Fauconnier and Turner2002) argued that many linguistic forms are best understood as prompts from the speaker to the listener to access pertinent schemas and construct particular relational configuration among concepts, such as analogy, role and identity, causality, and so forth. The general claim, then, is that words and the syntactic patterns in which they occur always underdetermine the rich interpretation regularly assigned to any utterance. Thus, humans' well-developed inferencing abilities, in conjunction with richly structured memory, are also seen as a central part of how language works.

Part of the ability to form schema is importantly connected to humans' large memory capacity and sensitivity to linguistic input and input frequency. In fact, studies of frequency effects (e.g., Bybee, Reference Bybee, Janda and Joseph2003, Reference Bybee2006; Bybee & Moder, Reference Bybee and Moder1983; Ellis, Reference Ellis2002a, Reference Ellis2002b, Reference Ellis, Tyler, Kim and Takada2008a, Reference Ellis, Robinson and Ellisb) have shown that humans are quite sensitive to the frequency with which they have encountered words and the constructions in which words are encountered, down to evidencing reaction time sensitivity to how likely they are to hear a verb in the present tense versus the past tense. The growing body of evidence amassed by Tomasello (e.g., Reference Tomasello2003) and his colleagues (e.g., Lieven & Tomasello, Reference Lieven, Tomasello, Robinson and Ellis2008) indicates that first language learning is item-based and frequency driven. Their work shows that young children are highly attuned to the specific exemplars of language in the input and the context in which they are used. For instance, children appear to constrain their use of a particular verb heard in an intransitive construction to intransitive use until they also hear that verb used in a transitive construction, even though they use other verbs in transitive constructions (Tomasello & Brooks, Reference Tomasello and Brooks1998). As we have seen, Goldberg (Reference Goldberg1995) exploited sensitivity to frequency in her explanation of the co-occurrence of particular verbs in syntactic constructions. SFL and discourse functionalism also assume that input is central to learning language and the shape language takes. However, CL's commitment to putting established cognitive processes at the center of the theory and its dedicated explorations of exemplar based learning, abstraction over numerous instances, and automaticity as a result of entrenchment (based primarily on input frequency) sets it apart from the other two approaches.Footnote ⁶

Cognitive linguists exploit findings from psychology indicating that concepts deriving from human interaction with the social-physical world, such as the spatial relations coded by prepositions or our conceptualization of a horse, crucially involve sensory-motor imagery. CL has argued concepts are better represented as being gestaltlike and schematic in nature, rather than represented as linguistic propositions or semantic feature bundles (e.g., Langacker, Reference Langacker1987). A tremendous amount of psychological evidence exists demonstrating sensory images are an essential part of human conceptualization. One simple example is the vast reaction time literature documenting the effect of imageability on how quickly readers process words.

The connections between spatial conceptualization and language have provided powerful theoretical insights (e.g., Langacker, Reference Langacker1987; Talmy, Reference Talmy2000a, Reference Tannenb). Talmy extended notions of spatial boundedness, which derive largely from visual perception, to explanations of mass versus count nouns as well as aspect. One particularly important aspect of spatial conceptualization is the human ability to conceptualize an event or state as a spatial scene. A related facet involves humans' ability to manipulate their perspective on a scene, that is, humans have the ability to construe or view a single scene from several perspectives. The ability to conceptually construe a scene from multiple perspectives has many reflexes in language. One central reflex is the various syntactic constructions found in language that convey similar propositional content but provide a different speaker stance or perspective, such as the active construction which puts the focus on the agent versus the passive construction that defocuses the agent and instead puts focus on the patient/theme. Note that Givón's analyses recognize general principles of cognitive processing, such as the tendency to start with what is given because this provides the listener with a shared reference point, but makes no reference to the human perceptual system and its effect on cognition, which is in turn reflected in language.

Talmy (Reference Talmy2000a) also discovered cross-linguistic typologies involving differences in how individual languages “divide up” their conceptualization of a spatial scene. He argued that while all humans see the same elements in a motion scene, for example, they all see the participants, the manner in which they move and the path of their motion, different speech communities have developed distinct, conventionalized patterns of packaging the information. Perhaps the most well known is the packaging difference between a satellite framed language (like English in which manner is often incorporated into the motion verb and path indicated by a separate particle construction, as in the sentence The woman strolled into the room) versus a verb-framed language (such as Spanish, in which motion and path tend to be conflated and manner expressed in a separate phrase, as in the sentence The woman entered the room strolling). This distinction has been a major focus of Slobin's (Reference Slobin, Shibatani and Thompson1996a, Reference Slobin, Gumperz and Levinson1996b, Reference Slobin, Gentner and Goldin-Meadow2003, Reference Slobin and Gathercole2009) cross-linguistic studies and his theory of thinking for speaking.

Applying CL to issues in L2 research and learning is still in its infancy. Although a handful of books and overview articles have been dedicated to the connection, the majority of the work has focused on potential insights rather than reporting on classroom applications or experiments (Achard & Niemeier, Reference Achard and Niemeier2004; Boers & Lindstromberg, Reference Boers and Lindstromberg2008; De Knop & De Rycker, Reference De Knop and De Rycker2008; Pütz, Niemeier, & Dirven, Reference Pütz, Niemeier and Dirven2001b; Langacker, Reference Langacker, Robinson and Ellis2008b; Robinson & Ellis, Reference Robinson and Ellis2008; Tyler, Kim, & Takada, Reference Verspoor, Tyler, Ritchie and Bhatia2008; Verspoor & Tyler, Reference Verspoor, Tyler, Ritchie and Bhatia2009). Even though the body of applied work is relatively small, space limitations allow for just a brief mention of some of the applications of CL. As I noted earlier, CL conceptualizes lexical items, phrases and morphosyntax on a continuum rather than as distinct components of language. As an organizational device, I will discuss the applications by dividing the studies into two main areas: vocabulary and morpho-syntax. This should not be taken as an indication of a theoretical divide between the lexicon and morphosyntax.

Vocabulary teaching, broadly defined, has been the area of L2 learning to which CL has been most frequently applied. This is probably due to the fact that some of the central, unique constructs from CL, such as conceptual metaphor theory, polysemy networks, and embodied meaning, are particularly useful in explaining some of the most difficult aspects of word meaning. The volume edited by Boers and Lindstromberg (Reference Boers and Lindstromberg2008) presented recent empirical findings on the applications of metaphor and metonymy in teaching general vocabulary, idioms, figurative language, and phrasal verbs. MacArthur and Littlemore (Reference MacArthur, Littlemore, Boers and Lindstromberg2008) and Walker (Reference Walker, Boers and Lindstromberg2008) presented effect-of-instruction studies based on CL in conjunction with corpus-based instruction. In general, the findings over the past 10 years indicate that alerting subjects to L2 patterns of metaphor and metonymy (and how they relate to semantic networks) is often effective in increasing learners' ability to learn and retain new words. For instance, Boers & Lindstromberg (Reference Boers, Lindstromberg, Kristiansen, Achard, Dirven and Ruiz de Mendoza2006) found that having students consider salient consequences of up-down led to an increased ability to interpret and retain unfamiliar metaphoric uses of verbs such as soar, skyrocket, and plunge (such as My spirits soared), as well as phrases and idioms. See also Boers, Eyckmans, and Stengers (Reference Boers, Eyckmans and Stengers2007) and Littlemore & Low (Reference Littlemore and Low2006).

A number of insights have emerged from analyses that combine metaphor theory with the notions of prototype effects and polysemy networks, for example, Csabi (Reference Csabi, Achard and Niemeier2004), for teaching the distinction between hold versus keep and their multiple meanings by using the metaphor “possessing is holding” (p. 234); Lowie & Verspoor (Reference Lowie, Verspoor, Achard and Niemeier2004), for teaching English prepositions; Rudzka-Ostyn (Reference Rudzka-Ostyn2003), for teaching phrasal verbs; and Verspoor & Lowie (Reference Verspoor and Lowie2003), for teaching multiple senses of polysemous words. Tyler and her colleagues (reported in Tyler, Reference Tyler and Pütz2008b) have successfully used Tyler & Evans's (Reference Tyler, Evans, Pütz, Dirven and Niemeier2001, Reference Tyler and Evans2003, Reference Tyler, Evans, Achard and Neimeier2004) analysis of the semantics of English prepositions (which draws heavily on polysemy networks, metaphoric extension, and embodied meaning), in effects-of-instruction experiments to teach the many meanings of over.

Metaphoric extension and embodied experience are central to Sweetser's (Reference Sweetser1990) analysis of the multiple meanings associated with all English modals (i.e., the social world-speech act or root meaning versus the logical reasoning or epistemic meaning). For example, You should take a sweater (social) versus That should be Eve at the door (reasoning). Sweetser's basic analysis was that the meaning of the modals is based in force dynamics.Footnote ⁷ Understanding of physical forces that affect forward motion have been extended first to the social world and then to the world of reasoning. Tyler converted Sweetser's descriptions of the various types of force dynamics associated with each modal into a set of simple figures, depicting various forces as being internal or external to a mover. In a series of experiments, Tyler and her colleagues (reported in Tyler, Reference Tyler, Robinson and Ellis2008a, Reference Tyler and Pütz2008b) found that subjects receiving instruction on the semantics of modals from a CL perspective used modals in a more nativelike way than subjects who were in an input flood situation but did not receive a CL explanation.

A number of studies have successfully applied CL insights to aspects of learning L2 morphology. Hamrick and Attardo (in press) exploited Talmy's analysis of boundedness, in conjunction with prototype theory, to successfully teach learners of Italian appropriate auxiliary selection in the passato prossimo. Niemeier (Reference Niemeier, Boers and Lindstromberg2008) also used boundedness to create teaching materials for English progressive aspect, as well as mass-count distinction. Huong (Reference Huong2005) drew on boundedness in the explaining mass-count distinctions in English nouns in relation to learning the English article system. Drawing on the primary metaphors “here is now/there is then” and “physical closeness is control,” Tyler and Evans (Reference Tyler, Evans, Pütz, Dirven and Niemeier2001) created pedagogical materials that systematically explained unusual uses of English past and present tenses, such as the use of past tense to indicate politeness, as in I was wondering if you'd like to have lunch, where the wondering is clearly taking place in the moment of speaking. Using notions of semantic extension and polysemy, Liamkina (Reference Liamkina2006, Reference Liamkina, Tyler, Kim and Takada2008) carried out a longitudinal study examining advanced foreign-language German learners' mastery of the multiple meanings and uses of German dative. Those who received explanations based on prototype theory and polysemy networks outperformed those receiving a traditional approach. Panther and Thornburg (Reference Panther, Thornburg, Pütz, Niemeier and Dirven2001) created a systematic, pedagogically oriented account of English -er nominals, drawing on conceptual schema theory, prototypes, and conceptual metaphor.

Complex syntactic points have also been addressed. Cadierno (Reference Cadierno, Robinson and Ellis2008) reviewed several studies that focused on learning motion verb constructions; these studies applied Talmy's verb-framed/satellite-framed analysis, as well as Slobin's (Reference Slobin, Gumperz and Levinson1996b) insights on thinking for speaking. The concept of construal has been fruitfully applied by several researchers. Achard (e.g., Reference Achard, Robinson and Ellis2008) investigated naturally occurring uses of definite and partitive articles in French. He argued that a speaker's choice between the two was a matter of construal and conventionalized cultural schemas; he successfully applied that analysis to L2 instruction. See Verspoor and Tyler (Reference Verspoor, Tyler, Ritchie and Bhatia2009) and Achard and Niemeier (Reference Achard and Niemeier2004) for discussion of other applications of construal to language teaching.

Recall that CL argues that syntax itself is meaningful and that syntactic patterns are templates abstracted over many instances of utterances. Moreover, there are no derivations or transformations of a more basic syntactic pattern to a second, derived pattern. Thus, Goldberg represented the double object construction (Subj-X Verb Obj-Y Obj-Z) as meaning X caused Y to receive Z, as in Hiro gave Yun the book. This construction contrasts with the caused-motion construction (Subj-X Verb-Obj Y Oblique object Z) which means X caused Y to move to Z, as in Lissa threw the football to Mike. A few studies demonstrating the usefulness of Goldberg's construction grammar (Goldberg, Reference Goldberg1995, Reference Goldberg2006) in L2 learning are beginning to emerge: Mellow (Reference Mellow2006) for teaching relative clause constructions; Manzanares and Rojo Lopez (Reference Manzanares, Lopez, De Knop and De Rycker2008) for double object constructions; Strauss, Lee, and Ahn (Reference Strauss, Lee and Ahn2006) for Korean completive constructions.