The sociosemiotic commitment

1 The social turn in Cognitive Linguistics

Cognitive Linguistics in the new millennium is characterized by a growing attention for the social and cultural aspects of language, on three levels of analysis. These three levels are chacterized by an increasing schematicity (Geeraerts and Kristiansen 2015). The first level considers variation within languages: to what extent do the phenomena that are typically focused on in Cognitive Linguistics exhibit variation within the same linguistic community? The research conducted within this approach links up with the research traditions of sociolinguistics, dialectology, and stylistic analysis, using the same meticulous empirical methods as these traditions: see Kristiansen and Dirven (2008); Geeraerts et al. (2010); Pütz et al. (2012). The next level is that of variation among languages and cultures, taking the form of cultural and anthropological comparisons or of historical investigations into changing conceptualisations across time periods (Palmer 1996; Sharifian in press). The third level, beyond intralinguistic and interlinguistic variation, is that of language as such: here we can situate foundational studies that emphasize and analyze the way in which the emergence of language as such and the presence of specific features in a language can only be adequately understood if one takes into account the socially interactive nature of linguistic communication (Sinha and Jensen de López 2000; Geeraerts 2005; Zlatev et al. 2008; Croft 2009; Harder 2010; Zlatev 2010; Schmid 2016). Although the label “cognitive sociolinguistics” is most strongly associated with the first level of analysis, it will be used in this paper as shorthand for the social turn in Cognitive Linguistics in general.

This social turn, it should be pointed out, is not a complete novelty in the history of Cognitive Linguistics, given that, for instance, the notion of ‘cultural model’ played a significant role in the emergence of the new framework (Holland and Quinn 1987). It is therefore best characterized as a deliberate strengthening and foregrounding of initially secondary features. At the same time, the social turn in Cognitive Linguistics is strictly speaking not yet a social turn of Cognitive Linguistics, to the extent that it is not the dominant perspective throughout the various branches of Cognitive Linguistics. All the more important then to ask the question what motivates the shift. This paper explores some of the fundamental reasons for introducing a social perspective in Cognitive Linguistics, and more specifically argues that the original inspiration and the distinctive theoretical characteristics of Cognitive Linguistics prime for the development of cognitive sociolinguistics. This is not a trivial issue, for how can such a social conception be reconciled with the very cognitive orientation of Cognitive Linguistics? Doesn’t the self-styled cognitive nature of Cognitive Linguistics automatically imply a commitment to a psychologizing rather than social perspective? Or should we conclude that Cognitive Linguistics is a misnomer?

The question will be answered in three steps. First, the issue will be situated against the philosophical background of Popper’s ontological pluralism. The next two sections show how two defining features of Cognitive Linguistics – its cognitive nature and its usage-based nature – naturally lead to a social approach, while at the same time requiring an empirical methodology. A concluding section then briefly explores the challenges ensuing from the social turn.

2 Language as a “third world” phenomenon

The epistemological status of a social view of language can be insightfully discussed against the background of Popper’s ontological pluralism, which assumes the existence of three worlds:

Without taking the words “world” or “universe” too seriously, we may distinguish the following three worlds or universes: first, the world of physical objects or of physical states; secondly, the world of states of consciousness, or of mental states, or perhaps of behavioural dispositions to act; and thirdly, the world of objective contents of thought

(Popper 1972: 106).

Language, in Popper’s view, firmly belongs to the third world, as a symbolic system for representing knowledge and building up argumentations. It is not disconnected from the other worlds, though:

Language itself, like a bird’s nest, is an unintended by-product of actions which were directed at other aims. – How does an animal path in the jungle arise? Some animal may break through the undergrowth in order to get to a drinking-place. Other animals find it easiest to use the same track. Thus it may be widened and improved by use. It is not planned – it is an unintended consequence of the need for easy or swift movement. This is how a path is originally made – perhaps even by men – and how language and any other institutions which are useful may arise and how they may owe their existence and development to their usefulness

(Popper 1972: 117).

The production of messages takes place in the second world, driven by subjective communicative intentions, but as a product, language belongs to the third world: language as a communicative instrument transcends the individual. As such, it is largely autonomous with regard to the individual language user: language users precisely become users in a language that is already given – regardless of the fact that their own activity as language users may subsequently contribute to changes in the language. Or, to continue Popper’s analogy: language users walk along existing paths, but they may widen them, change their course, use them so unfrequently that they are soon overgrown, or simply create new ones. More systematically, language may be situated in the three worlds at the same time: it has a material existence (in various forms, ranging from brain states and anatomical speech processes to books and text files), it has a psychological reality in the mind of speakers and hearers, and it has an objective existence going beyond the individual, mental level. (Cognitive Linguistics, as in Verhagen 2005, would probably talk of an “intersubjective” rather than an “objective” existence. The distinction deserves a philosophical discussion, ^[1] but in the present context, we focus on the “third world” nature that is captured by both terms alike.)

A view like Popper’s is not unique in the history of linguistics and the philosophy of language. It looms large in Itkonen’s conception of language, to which we will return below, and it corresponds broadly with the picture drawn by Keller (1994) of language as a “phenomenon of the third kind” – the unintended product of an intentional action. It fits well into the currently popular conception of language as a Complex Dynamic System (e. g., Beckner et al. 2009), specifically through its insistence on the dialectic relationship of individual behavior and supra-individual outcomes: the individual user acquires a ready-made language, but contributes to it by perpetuating and developing it. And of course, it squares with the Saussurean point of view that language, as langue, is primarily a social semiotic that goes hand in hand with purposeful individual behavior at the level of parole. This list and each of its elements could undoubtedly be expanded, but the point will be clear: points of view that oppose a reduction of language to a psychological phenomenon constitute a respectable strand in twentieth century thinking about language.

Why then would Cognitive Linguistics be attracted to such a “third world” conception of language? The question may be answered by having a closer look at the defining features of Cognitive Linguistics: its cognitive nature, of course, but also its usage-based nature and its recontextualizing nature. In the next two sections, the impact of the first and second feature are explored in depth. For reasons of brevity, the third feature will not be spelled out in this paper, but its impact will be clear; see Geeraerts (2010a). (Generative grammar decontextualizes grammar by disassociating what is considered to be the core of linguistics from the discursive context of performance and language use, from the social context of interaction and variation, and from the cognitive context of meaning and experience. Cognitive Linguistics is a recontextualizing approach in that it reincorporates these contextual domains into the scope of the grammar – including, needless to say, the social perspective).

3 Language as cognition and the Cognitive Commitment

The characterization of Cognitive Linguistics as “cognitive” has basically (but largely implicitly) received two slightly different interpretations in the history of the framework. On the one hand, Cognitive Linguistics has been presented as adhering to the Cognitive Commitment, i. e., “a commitment to make one’s account of human language accord with what is generally known about the mind and the brain, from other disciplines as well as our own” (Lakoff 1990: 40). According to this view, cognitive linguists should take into account the empirical findings of disciplines like cognitive psychology, developmental psychology, and the neurosciences, and atune their linguistic models accordingly. On the other hand, the editorial statement of the launching issue of the journal Cognitive Linguistics – the same issue, in fact, in which Lakoff formulated the Cognitive Commitment – defines Cognitive Linguistics in terms of a conception of language “as an instrument for organizing, processing, and conveying information” (Geeraerts 1990: 1). According to this view, language is studied not as if it were a separate and autonomous cognitive module, but as a reflection of general conceptual organization, categorization principles, processing mechanisms, and experiential and environmental influences. Interdisciplinary openness of the type championed by the Cognitive Commitment is not absent from this alternative interpretation, but rather than being formulated as a principle on its own, it results from the idea that language is intrinsically integrated with the human cognitive capacities at large.

The distinction between both formulations is subtle but real: the Cognitive Commitment is more schematic than the alternative formulation, and is basically non-committal with regard to the content of “what is generally known about the mind and the brain”. In principle, it leaves open the possibility that that general knowledge takes the form of a modular, formalist conception of language as common in Chomskyan approaches. This is surely not what Lakoff presupposes, but it is only the alternative formulation that makes the presupposition explicit. It can be argued, in that respect, that the latter, more substantiated interpretation gives a better account of what Cognitive Linguistics does in practice than the former, more schematic one. This is the case for two reasons.

First, a “language as cognitive tool” view makes it easier to understand some of the core features of Cognitive Linguistics: the primacy of semantics in linguistic analysis, the encyclopedic nature of linguistic meaning, and the perspectival nature of linguistic meaning. The primacy of semantics in linguistic analysis follows in a straightforward fashion from the cognitive perspective itself: if the primary function of language involves knowledge and communication, then meaning in the broadest sense must be the prime focus of linguistic attention. The encyclopedic nature of linguistic meaning follows from the categorial function of language: if language is a system for the categorization of the world, there is no need to postulate a systemic or structural level of linguistic meaning that is different from the level where world knowledge is associated with linguistic forms. The perspectival nature of linguistic meaning implies that the world is not objectively reflected in the language: the categorization function of the language imposes a structure on the world rather than just mirroring objective reality.

Second, a “language as cognitive tool” view allows for a clear demarcation between Cognitive Linguistics and generative grammar. (This is a comparison that was probably more important when Cognitive Linguistics emerged than it is now, but it is instructive in the present context.) The cognitive linguist is interested in our knowledge of the world, and studies the question how natural language contributes to it. The generative linguist, conversely, is interested in our knowledge of the language, and asks the question how such knowledge can be acquired, given a cognitive theory of learning. As cognitive enterprises, Cognitive Linguistics and generative grammar are similarly interested in those mental structures that are constitutive of knowledge. For the cognitive linguistic approach, natural language itself consists of such structures, and the relevant kind of knowledge is knowledge of the world. For the generative grammarian, however, the knowledge under consideration is knowledge of the language, and the relevant mental structures are constituted by the genetic endowment of human beings that enables them to learn the language. Whereas generative grammar is interested in knowledge of the language, Cognitive Linguistics is so to speak interested in knowledge through the language.

But then, given the two interpretations of cognitive in Cognitive Linguistics, how easily could they be reconciled with a social, “third world” view of language? The Cognitive Commitment is consonant with a “third world” view if it is taken to imply that the “first world” and “second world” characteristics of language constrain but do not exhaust its “third world” nature. In the context of Popper’s analogy: the constitution of the soil, the slope of the terrain, and the density of the vegetation will have an effect on the path that actually emerges, but the path in itself is not simply the conjunction of these features, it is the cumulative result of purposeful human activity restricted but not caused by the environmental conditions. In a similar way, natural language is shaped by features such as the nature of the speech organs and the neurophysiology of the brain (i. e., embodiment in the most literal sense), but language as such does not coincide with the combination of those features. Examples of such an approach are found in multivariate quantitative grammar studies that incorporate psychological factors (like persistence or processing cost) alongside grammatical and semiotic variables, discourse-related features, and possibly also lectal characteristics: see for instance the probabilistic grammar model (Bresnan and Ford 2010; Szmrecsanyi and Hinrichs 2008; Szmrecsanyi 2009), or the lectally enriched multivariate grammar studies illustrated by studies like Grondelaers et al. (2002), or Levshina et al. (2013), and discussed in Heylen et al. (2008).

Next, the alternative, substantiated interpretation fits a “third world” interpretation even better, because it zooms in on precisely the cognitive function of language as a tool for organizing and exchanging knowledge – just like the role attributed to language in Popper’s ontological pluralism. The “language as cognitive tool” interpretation of the label Cognitive Linguistics, then, is consonant with a “third world” interpretation if “language as cognition” encompasses shared and socially distributed knowledge and not just individual ideas and experiences. An early example (early in the history of Cognitive Linguistics, that is) of a universalist “second world” view contrasting with a culturally and historically informed “third world” view may be found in the debate about the basis of the anger is heat metaphor. Lakoff and Kövecses (1987) interpreted the metaphor in terms of universal physiological experiences: increased body heat is taken to be a physiological effect of being in a state of anger, and anger is metonymically conceptualized in terms of its physiological effects. Geeraerts and Grondelaers (1995), however, pointed out that quite a number of the expressions illustrating the anger is heat pattern are lexical relics with a historical basis in the theory of humours, the highly influential doctrine that dominated medical thinking in Western Europe from antiquity to early modern times. In recent years, taking a cultural and historical perspective in metaphor research has been gaining momentum and recognition, also among the major spokesmen of Conceptual Metaphor Theory: see Kövecses (2005).

4 Usage and variation

A second important feature of Cognitive Linguistics that suggests a social perspective is it usage-based nature. The observational data that will enter into a usage-based approach will often be drawn from a corpus, and as such they will most often not be internally homogeneous: because the texts collected for the corpus come from various sources, it will not always be known in advance whether the variation that may be observed in the corpus is due to lectal factors or not. Filtering out the effects of such factors will then be a methodological requirement for any cognitive linguistic attempt to analyse the usage data – even if the analysis is not a priori interested in lectal variation. The existence of variation within a language – variation that may be socially structured or that may simply be individual – has a crucial impact on the methodological consequences of a social orientation in Cognitive Linguistics. Without such variation, the language as a semiotic system used in a given community is the same for each and every member of that community, and as such, each language user belonging to that linguistic community constitutes a representative sample, in statistical terms, of the language. Accordingly, investigating a native user’s knowledge of the language (as the case may be, the investigating linguist’s own native knowledge of the language) is a legitimate basis for doing linguistics. This methodological stance is obviously well known from Chomskyan grammar, but then, Chomskyan grammar takes a “second world” view of language, targeting an alleged Universal Grammar that belongs to the biological constitution of mankind. That is to say, the methodological motivation for introspection of the Chomskyan type is not so relevant for a social conception of Cognitive Linguistics, given the fact that the basic assumptions in Popperian terms are so different.

However, some additional discussion is necessary here. Itkonen (2003, 2008 and many other publications) argues for a ‘third world’ view of language relying on intuition rather than introspection. According to Itkonen, linguistics – explicitly defined as the study of language as a social phenomenon – is based on intuition rather than observation, because the social nature of language takes the form of a system of social norms. Norms have a specific epistemological status which is revealed by the distinction between rules and regularities: a rule, as describing what ought to be done, differs from a regularity, which describes tendencies in what is actually done. A hypothesis about observable regularities may be incorrect (and may thus be falsified) but a rule cannot be falsified, it can only be broken. It follows that the methodology of studying norms is different from the method of the empirical, descriptive sciences. Referring to the same Popperian ontology that we discussed earlier, Itkonen introduces an epistemological distinction between respectively observation, introspection and intuition. Intuition, the appropriate method for the investigation of norms, is basically an inspection of one’s own knowledge of the linguistic conventions: “When one is describing one’s own native language, data-gathering consists, not in experimentation/observation but in trying to remember something that one in principle knows already. Notice also that one does not try to remember what someone has said in fact, but what ought to be said” (Itkonen 2003: 40–41). Itkonen’s terminological contrast between intuition and introspection corresponds with his rejection of the Chomskyan conception of language. Itkonen argues that Chomsky’s psychologistic conception is an ontological as well as an epistemological category mistake. Language belongs to the world of social norms, not to the world of psychological states – and hence intuition rather than introspection is required.

But such an intuition-based perspective could easily be associated with the assumption that on the normative level, there is no variation, or at least, that the existing variation is by and large irrelevant. Without denying the importance of normativity in language nor the role of intuition in the empirical cycle of linguistic research (Geeraerts 2010b), such an assumption needs to be challenged from a philosophical and a linguistic perspective (for an extended discussion, see Geeraerts 2005).

First, as Itkonen admits, norms may exhibit cases of uncertainty and competition. But such cases will then ask for an observational approach. In the case of uncertainty, if individuals hesitate which norm applies, then that is an indication that their intuition (and a fortiori, that of the researcher) does not always provide an answer. In the case of controversy and normative clashes, the existence of competing norms will often be linked to a struggle or antagonism between groups, i. e., to social variation of precisely the kind that we think should be studied. In general, what is needed here is a theory of norms of the kind developed by Bartsch (1987), which explicitly recognizes possible competition between alternative norms, and specifically also the vital role such competition may play in explaining the dynamics of language.

Second, and more importantly, it is doubtful whether languages can indeed be thought of as homogeneous linguistic systems. That idea itself is a cornerstone of structuralist thinking, adequately captured by Meillet’s formulation of a language as “un système où tout se tient” (a.o. 1937: 475). It is also a highly influential idea: throughout the twentieth century, mainstream linguistics has thought of languages in terms of coherent systems represented by their grammar. Linguistic communities are accordingly defined by the fact that they share a grammar. Such an approach does not deny variation, but the unit of variation was a language, a linguistic system. A common way of speaking, for instance (one that most young linguists will still come across in the course of their training) is to say that dialects are full-fledged languages because they have their own grammar.

There are two ways of showing that this notion of homogeneity (a single coherent grammar shared by a group of language users) is problematic: first, by looking at the language users, and second, by looking at the linguistic phenomena that constitute the alleged coherent systems. With regard to the former, there is increasing evidence for individual differences between language users: see e. g., Sharma et al. (2008) or Dąbrowska (2012), Divjak et al. (2016). Given that a traditional approach can model individual variation in terms of users possessing different grammars, i. e., participating in different communities, individual differences are specifically relevant when they involve users of what would traditionally be recognized as a single system. Such differences actually even occur on the “first world level”; studies like those of Mielke (2015) e. g., reveal by means of ultrasound techniques how the position of the tongue in the pronunication of the same phoneme may differ considerably across speakers. With regard to the latter, linguistic phenomena do not coincide neatly if their lectal distribution is mapped out. If languages exist as coherent systems, then all the elements of a given system would occupy the same position in lectal space (i. e., in the space defined by the social, geographical, contextual dimensions that might structure the variation). But already in the nineteenth century, linguists learned that dialect boundaries are not neat bundles of isoglosses: different linguistic phenomena have different geographical distributions, fuzzying the boundary between one system and the other. In a similar way, lectal isoglosses in general need not coincide (see Geeraerts 2010c). The idea that each linguistic feature may in principle have its own lectal distribution revives the older view of Jaberg that each word may have its own particular history: “In Wirklichkeit hat jedes Wort seine besondere Geschichte” (1908: 6).

If we now tie some of the strands together, it appears that a social conception of Cognitive Linguistics occupies a distinct position in the history of linguistics. The two dimensions that we have so far identified to characterize it, each embody a fundamental change with regard to a position that defined an earlier stage in the development of linguistics. First, the social interpretation of cognitive in Cognitive Linguistics squarely puts linguistics back in Popper’s third world, in contrast with the psychologicalization of language in Chomskyan linguistics. It is a return, in that sense, to the pre-Chomskyan Saussurean conception of language as a social semiotic. But second, the usage-based character of Cognitive Linguistics reverses the Saussurean and structuralist emphasis on langue, as an internally homogeneous system, in favor of parole as intrinsically varied.

In other words, the social turn in Cognitive Linguistics is a return to a Saussurean conception of language as a social semiotic without the Saussurean assumption of the internal homogeneity of language systems. The three relevant stages in the theoretical development of linguistics can therefore be organized in a systematic pattern: see Table 1. (Admittedly, the picture only holds for Cognitive Linguistics to the extent that it takes shape as a cognitive sociolinguistics – but that is a point to come back to in the concluding section.)

It follows that the method of a socially inspired Cognitive Linguistics, like its object, needs to go beyond the level of the individual. It needs to give an intersubjectively reliable description of the intersubjective reality of language. This explains the importance for cognitive sociolinguistics of quantitative methods of the kind that are common in the exact sciences, characterized by the replicability of results, the explicit operationalization of concepts, the statistical testing of hypotheses. In this respect, there is an affinity between the descriptive “social turn” in Cognitive Linguistics and its methodological “quantitative turn” (Janda 2013).

5 The sociosemiotic commitment

The various reasons we have identified for a social turn in Cognitive Linguistics may now be translated into a Sociosemiotic Commitment. ^[2] To complement the Cognitive Commitment, we define a commitment to make one’s account of human language accord with the status of language as a social semiotic, i. e., as an intersubjective, historically and socially variable tool, and to base that account on a methodology that likewise transcends the individual. In the current stage of development of Cognitive Linguistics, this is not yet a highly entrenched commitment: even though (or precisely because) Cognitive Linguistics initially defined itself in opposition to generative grammar, it understandably carries along influences from the mainstream tradition of modern linguistics as synthesized in the previous sections. To round off, then, we may list the main habits that need to be abandoned or altered by linguistic approaches adhering to the Sociosemiotic Commitment:

1. an exclusive reliance on introspection (or intuition)

2. a bias towards universal features of language, to the detriment of attention for the historicity and variability of language

3. a belief in languages as internally homogeneous systems

4. a predilection for the psychological “second world” status of language, to the disadvantage of its intersubjective “third world” status.

The first two of this list have a weaker position than the other two. Corpus methods and experimental approaches (which – it should be emphasized – include introspection and intuition as part of the empirical cycle; see Geeraerts 2010b) have acquired a very strong, perhaps even dominant position in linguistic research, even if their further expansion is somewhat hampered by an absence of appropriate methodological training in many linguistic curricula. Similarly, the attention that the label “cognitive sociolinguistics” has been able to generate (see the references in the first section), or the interest in historical research within Cognitive Linguistics (Winters et al. 2010; Diaz Vera 2015) illustrate that the attention of cognitive linguists is most certainly not exclusively centered on universals.

The final two habits in the list seem to be more tenacious, though. They are also connected to each other: if you believe that the grammar takes the form of a single internally homogeneous system (in the structuralist sense) that is uniformly shared by all members of a linguistic community, then the psychological reality of language and its social reality basically coincide: what is in the head is just the same as what shapes the behavior of the community. It is then even attractive to assume (entirely in line with the generative tradition of thinking about language) that the psychological reality is in some sense more basic, because it is, through its biological and neurological substrate – its literal embodiment – somehow more tangible than the social reality of language. But if you take into account the social and individual variation within a linguistic community, the rules and tendencies shaping the behaviour of the community and the individual mental representation of those rules may differ. Individual linguistic behavior will be a partial or divergent reflection of the aggregate behavior, and (as the ultrasound data show) identical external behavior may even be internally produced in different ways by different individuals. Accordingly, getting a good idea of the behavior in the community becomes a separate goal from investigating mental representations – and an indispensable one at that for a proper interpretation of the psychological phenomena.

This also implies that the focus of Cognitive Linguistics on the Cognitive Commitment needs to be relaxed. Yes, we need cognitively realistic grammars, but we also need systematic attention for the intersubjective, social, historical reality of language. A fair amount of modesty may be recommended here: cognitive linguists had better admit that they do not know yet how exactly to reconcile the two commitments. In very general terms, all manifestations of cognitive sociolinguistics mentioned throughout this paper provide starting-points for further developments. In one direction, psychological factors like processing cost or priming effects may be integrated in variationist studies. In the other, lectal and pragmatic variation may be built into models of grammatical knowledge. And mediating between both directions, a well-ordered and balanced model such as Schmid’s (2016) may function as an organizational backdrop against which to situate the multiple aspects of the enterprise. But beyond these immediate gains, open questions – specifically, open questions relating to meaning as the core of Cognitive Linguistics – lie waiting. If meanings are in fact social construals, ^[3] they may exist more in the community than in the head, but if they are less in the head than cognitive linguists may have assumed, going through the head to identify meaning is also less efficient than we may have believed. Thinking about meaning as not being primarily in the head is nothing new in the philosophy of language, though, and as such, a confrontation with philosophical semantics should be high on the agenda of our rethinking of the fundamentals of Cognitive Linguistics. Compared to the challenge inherent in such a confrontation, it is insignificant whether what comes out might eventually still be called ‘Cognitive Linguistics’…