Why is there variation rather than nothing?

Language Sciences 39 (2013) 31–53 Contents lists available at SciVerse ScienceDirect Language Sciences journal homepage: www.elsevier.com/locate/langsci Why is there variation rather than nothing? Bernard Laks Université Paris Ouest Nanterre La Défense, Laboratoire Modyco, Institut Universitaire de France, France a r t i c l e i n f o Article history: Available online 6 April 2013 Keywords: Variation Heterogeneity Corpus linguistics Darwinian linguistics Biolinguistics Functional communication a b s t r a c t Although variation and heterogeneity are generally recognized as a characteristic of linguistic usage, they are hardly conceived as a fundamental dimension of grammar. Introducing a distinction between an exemplum linguistics and a datum linguistics, we show that the history of the language sciences, both conceptually and descriptively, is organized around opposing conceptions of what a linguistic fact is. While Generative Grammar and its practitioners are clearly examples of the first, corpus linguistics illustrates the second conception. The upsurge of datum linguistics and of usage based models since the beginning of the 21st century allows for a fundamental reassessment of the functional role of variation in language. We show that while the biolinguistic program rests on the negation of the sociocultural framing of the language faculty and denies any functional impact of communicational competence in shaping language and grammars, constructional approaches and usage-based linguistics make room for a Darwinian sociocultural conception. In such a conception, communication is a leading force in the emergence and stabilization of human linguistic competence, both diachronically and synchronically. In this approach, variation and heterogeneity are to be seen as core factors in shaping language. Thus, as argued some 50 years ago by Weinreich, variation and heterogeneity are to be regarded as structurally functional, and as fundamental dimensions of language. Ó 2013 Elsevier Ltd. All rights reserved. Everyone knows that language is variable. Two individuals of the same generation and locality, speaking precisely the same dialect and moving in the same social circles, are never absolutely at one in their speech habits. Sapir (1921, p. 147) 1. Approaches to variation: historical perspectives In the recent history of linguistics, Weinreich et al.’s publication (1968) was a historical turning point. For many young linguists of the moment, this seminal paper marked the birth of modern sociolinguistics and was a decisive influence on their intellectual career. Of course, in 1968 the term ‘‘sociolinguistics’’ had already been around for several years.1 Currie created it in the 1950s in order to explain, among other things, the relation between languages and castes in India (Currie, 1952). Though sociolinguistics and the sociology of language progressively shaped their identity during the 1960s through a series of major international conferences which marked the field,2 the sociolinguistic perspective really came to the fore in 1968, combining in an original way a structural dialectology, in particular urban structural dialectology, and research in the field of contacts between languages and pidgins on the one hand and historical linguistics on the other.3 E-mail address: Bernard.laks@u-paris10.fr Currie (1952) presents the term sociolinguistics as a personal lexical creation. He strongly claimed its paternity in 1981 (Currie, 1981). 2 Cf. Koerner (1991) for a historical analysis (caveat the initial reference to Currie is misdated in it). 3 On the influences on sociolinguistics, Cf. Weinreich for structural dialectology (1954), Labov for urban dialectology (1966), to Weinreich for language contacts and pidgins (1951, 1953), to Lehman for historical linguistics (1962). 1 0388-0001/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.langsci.2013.02.009 32 B. Laks / Language Sciences 39 (2013) 31–53 Such convergence is hardly surprising. Meillet had announced it half a century earlier in his inaugural lecture at the Collège de France (Meillet, 1921), where he argued that all the dimensions of linguistic change (in the geographical, social, stylistic or historical area) stem from a single source, the social institution character of language. Hethus reminded linguists that all observable phenomena of mixing, evolution and fragmentation, in short everything that brings language to life according to Whitney (1875), always comes from an external source. That is the reason why linguists, as long as they adopt the internal point of view only, can indeed describe these phenomena, but only in terms of what they are: they definitely fail to explain them.4 Meillet also emphasized that in order to reach a causal explanation, it is necessary to adopt an external and social point of view. Relating linguistic facts to their cause and thus explaining them, requires recognition of the driving role played by variation in linguistic phenomenology. Such variation is linked to the intrinsic social dimension of language.5 Such was Meillet’s lesson, which was taken up by Weinreich half a century later. The objective of the present paper is to show that the lesson was on the whole ignored by 20th century linguistics, but that it is currently making its way back to the foreground with a renewed acuteness. In 1968, in their manifesto encouraging an empirical analysis of linguistic change, Weinreich and his two PhD students,6 thus placed variation at the heart of linguistic phenomena. To this end, they combined several questions. Working on the various models of linguistic change put forward since the Neogrammarians, they notably discussed the phenomena of gradual or sudden evolution, the comparative role of phonetic rules and of lexical scattering, the existence of residues and mutation exceptions, etc.7 This led them to analyse the intergenerational transmission models as a source of change. Paul’s suggestions (1909), like those of Chomsky and Halle (1968), were criticised in the light of the internal variability of grammars and the social embedding of language in speech communities and peer groups where linguistic identity is forged. As underlined in the third point of their conclusion, while all internal variation and heterogeneity within a given language does not necessarily lead to a change in the said language, all processes of change necessarily stem from socially marked, assessed and promoted internal variation and heterogeneity. Languages are thus considered as systems which are unstable, open, plastic, deformable and porous. Related to the question of the contact between languages, the notions of mixing and interlanguage, which at the very least since Schuchardt (1909, 1922) have established creolisation as a principle in the evolution of all languages, are thus reasserted. Overall, it is in the intrinsic social character of language, by way of the intimacy of the link between language and the socially qualified speech community, that Weinreich et al. (1968) locate the initial source and driving force of linguistic change. They highlight the fact that a speech community is a concrete social organisation. It is therefore, ex definitio, deeply heterogeneous, divided, hierarchical and structured by an antagonistic social dynamics. In such case, the implication is that linguistic variation and heterogeneity on the one hand and social heterogeneity on the other are simply the two aspects of the same social reality. It is because no perfectly stable homogeneous community can ever exist that no perfectly stable and invariant homogeneous language can exist. Thus, the intimate link between language and social structure, affirmed by all the Masters of 20th century modern linguistics is here apprehended through the prism of change. However, Weinreich, Labov and Herzog drew from this a series of very general linguistic conclusions, which, through the foundation of variationist (socio)linguistics would profoundly mark the theoretical landscape. They wrote that in order to progress and achieve a comprehension of linguistic phenomena, linguistics must break the link between structure and homogeneity and systemic organisation and invariance. All societies, all cultures and all human organisations show strong internal differentiations, structured hierarchies and more or less conflicting heterogeneities, and the very essence of the social resides in the organisation of these differences in what are constantly changing dynamic systems.8 Social heterogeneity and variability cannot therefore be considered as parasitic, accessory or abnormal dimensions. They are the very dimensions of the social. As I will show in this paper, it follows that variation and heterogeneity must be located at the very heart of linguistic systems of which they constitute the organising and functional principle. Variationist linguistics thus places variation at the very heart of the linguistic model. It thus relegates the invariant approaches to the category of homogeneous grammars. A homogeneous and invariant grammar is merely a grammatical system which has been standardised by the very same social it claims to ignore. Thus, if all languages are heterogeneous and variable, the grammar which claims to describe and model it must also be so. But above all, variation systemicity and heterogeneity organisation are to be seen as the very source of what creates the structure and the system of languages. It is indeed the existence of various communication modes and systems which leads us to grammaticalise and systematise them, as we will see below. This is precisely what the Master of Geneva was saying when he declared that language as a system is the result of various forms of speech being simultaneously perceived by the same collective consciousness. It follows from 4 One can appreciate at its fair value Meillet’s condemnation of internal grammar founded on the formal categories of logic alone. ‘‘L’ancienne grammaire générale est tombée dans un juste décri parce qu’elle n’était qu’une application maladroite de la logique formelle à la linguistique où les catégories logiques n’ont rien à faire’’ (Meillet, 1921, p. 15). 5 ‘‘Le seul élément variable auquel on puisse recourir pour rendre compte du changement linguistique est le changement social dont les variations du langage ne sont que les conséquences parfois immédiates et directes, et le plus souvent médiates et indirectes’’ (Meillet, 1921, p. 17). 6 William Labov and Marvin Herzog both defended a PhD in Columbia under the direction of Uriel Weinreichin in 1964. 7 For a more recent detailed discussion Cf. Labov (1981, 1994, 2001). 8 Without entering here a post-Hegelian debate over history, its dynamics and rationalities, as, for instance, Fukuyama (1992) thought he could synthetize, we shall oppose this dynamic conception of the social to the perfectly sterilised, ideal and static approach which Chomsky (1965, p. 12) thinks he can adopt for purely heuristic purposes by choosing an ideal speaker–listener (Cf. infra). For a criticism of such a claim to be able to extract oneself from history, societies, their dynamics and the very hubris of the social, Cf. Bourdieu (1997). B. Laks / Language Sciences 39 (2013) 31–53 33 this that, for Saussure, the linguistics of speech (variable) is the preliminary condition for all language linguistics. (Saussure, 1916, p. 2759, 2001, p.83).9 Fifty years on, Cartesian idealism was to enter an objection, through Chomsky (1966), arguing that, in order to be cognitively instantiated, linguistic structures and grammar itself are by their formal nature necessarily homogeneous, monotonous and invariant, and that this is due to the immanent character of thought and logic on which they are grounded. Such apposition, that of the Grammarians of homogeneity and of linguistic statics since long before Port Royal,10 scotomises all the social dimensions of grammar. Chomsky, for instance, who invokes the linguistic competence of speakers, considers that all analysis of human competences as socialised competences is null. We are thus encouraged to believe that what Bourdieu (1980) called ‘‘The Theory of Practice’’ is insignificant. In contrast, the sociologist believes that competences are the product of the internalisation of an exteriority which is always socially qualified. These competences are put into practice in specific social contexts, like the exteriorisation processes of an internality that is itself socialised. This theory of habitus, of which the linguistic habitus and the practical communicative competences associated to it represent what is simply a specific instance, explains the key place that is occupied by socially motivated and controlled variation and heterogeneity at the heart of linguistic structures. It resonates with the most recent works of the so-called current of ‘‘cultural cognition’’ (Tomasello, 1999, 2008a), with its extensions in contemporary linguistics.11 I will come back to this in more detail further on. This approach postulates that in an actual speech community, which is therefore socially structured, the capacity to communicate is necessarily grounded on a cognitive capacity to master linguistic heterogeneity and variation, a mastery which operates apropos and in situation, just like the habitus. Actually, much more than grammars and linguistic structures themselves, it is this cultural and cognitive ability to master heterogeneity which is transmitted from generation to generation, thanks to a cultural and social reproduction process (Maniglier, 2008). In this perspective, the issue of intergenerational change raised by Weinreich, Herzog and Labov finds a transparent explanation. Thus the hypothesis of relating heterogeneity and linguistic variation to the social and cognitive processes which govern ‘‘la vie des signes au sein de la vie sociale’’, Saussure (2001, p. 34) allows us, because they are being related to their function, to understand both their scope and their limits. Indeed, it is precisely because it exists and because it constitutes a system that linguistic variation is never erratic. On the contrary, it is always constrained and organised. Structural heterogeneity is thus grammaticalised through the play of two opposite dynamics: the requirements of intra-community understanding limit heterogeneity and differentiation, whereas the existence of a differentiated social organisation limits linguistic homogeneity and structural invariance. It is the eternally shifting play and balance of these two opposite dynamics which structure variation and give rise to linguistic change. Of course, the observation of variation in its systematicity and the presentation of heterogeneity as a structured phenomenon require the use of an appropriate methodology. It is common knowledge that a decontextualised observation destroys the systematicity of variable phenomena, making them appear random.12 Indeed, the observation of facts outside the social ecosystem which conditions them destroys everything that the practice in question owes precisely to its practical character.13 It is for this reason that the analysis of systemic variation necessarily begins with a critical study of the observables. 2. Datum and exemplum, two approaches to data Language Sciences, just like all sciences, must be based on a clear empirical foundation. This means that the construction of linguistic observables necessarily plays a key role in it. Therefore, against what Bachelard calls ‘‘common opinion’’ (1938, pp. 14–16), it is to be recalled that there is no scientific data which is not the result theoretical questioning and precise epistemological construction: it is the point of view which creates the object,14 and as he reiterates: ‘‘les instruments ne sont que des théories matérialisées. Il en sort des phénomènes qui portent de toutes parts la marque théorique’’ (ibidem.) It is in the light of this critical epistemology that the very notion of linguistic observable, of language fact or language data needs to be tried. The notion is not a self-evident one, and in the field of contemporary linguistics two major approaches can be contrasted in this respect, depending on whether they take variation into account or whether they destroy it. Usage based linguistics (Barlow and Kemmer, 2000), to which variationist linguistics belongs,15 considers attested linguistic usages as facts to be taken into account in a grammatical system. The latter is thus a system which necessarily integrates heterogeneity and variation. Echoing the Durkheimian principle, it thus deals with linguistic and social facts as if they were 9 On these issues Cf. Laks (2011b), Béguelin (1990), and Bouquet (1997). Cf. Arnauld and Lancelot (1660) and infra. As we will see, cultural cognition is closely related to construction grammars, examplarist grammars and more generally to usage-based grammars. Cf. Tomasello (2008c), Goldberg (2006), and Barlow and Kemmer (2000). 12 On the observer’s paradox Cf. Labov (1975, 1976). On the linguistic market and on observation effects Cf. Encrevé (1976, 1982). 13 Cf. Bourdieu (1994, 1993). 14 ‘‘Et quoi qu’on dise, dans la vie scientifique, les problèmes ne se posent pas d’eux-mêmes. C’est précisément ce sens du problème qui donne la marque du véritable esprit scientifique. Pour un esprit scientifique, toute connaissance est une réponse à une question. S’il n’y a pas eu de question, il ne peut y avoir connaissance scientifique. Rien ne va de soi. Rien n’est donné. Tout est construit’’ (Bachelard, 1938, p. 14). 15 With Langacker (1987), Tomasello (2008b) suggests the more theoretical term of functional cognitive linguistics to designate approaches based on usage observation which do not take for granted the preconstructed grammatical classifications of Western grammatical tradition but aim at reconstructing cognitive linguistic analysis from the examplarist, static and probabilistic structure of usages. I will be using the more common term of usage based linguistics. 10 11 34 B. Laks / Language Sciences 39 (2013) 31–53 things, rejecting introspection and its preconceived classifications16 in favour of a inventorying of attested language productions. Such a method of investigation necessarily addresses the heterogeneity and the variation of language, the two features which it must apprehend as properties which are systematic and fundamentally structural. In contrast, the grammarian conception, illustrated by Generative Grammar under its various avatars (Minimalism, Universalist comparativism, Biolinguistics) builds its system by basing itself on examples or on grammaticality judgements, disregarding what can be observed in real speech communities. All variation or heterogeneity is thus related to dimensions, which are exterior to the language, and is thus eliminated from the observable thus constructed. Considered as a dimension of pure performance, variation is denied any kind of systematicity and grammatical relevance. In terms of the abstraction to which the Cartesian conception of language and cognition lays claim, ‘‘Linguistic theory is concerned primarily with an ideal speaker–listener, in a completely homogeneous speech-community, who knows its language perfectly and is unaffected by such grammatically irrelevant conditions as memory limitations, distractions, shifts of attention and interest, and errors (random or characteristic) in applying his knowledge of the lingual in actual performance.’’. The necessity of such abstraction, which Chomsky (1965, p. 3) claims to be the very condition of scientific rationality,17 is concomitant with the denial of any corpus linguistics.18 He considers that empiricism is of no more interest for linguistics than the collection of moths is for science in general.19 As I have shown elsewhere (Laks, 2008), beyond a classical epistemological opposition between idealism and empiricism, the opposition between variationist linguistics and generative linguistics reactivates the cleavage, which is also classical, between the sciences of datum and those of exemplum. Indeed, for as long as they have existed, linguistics and philology have opposed these two perspectives, which condition the specific relation to the data which they deal with. In order to clarify the contemporary debate over the place of variation and heterogeneity in grammars, a brief examination of the issue is worthwhile. 2.1. Exemplum sciences The prescriptive or grammatical approach operates by way of a series of normed (proto)types which are used as exempla. In Greek grammatical techné, in the work of Dionysius Thrax, or in the subsequent development which Apollonius Dyscolus was to bring to the method, leading to what was already a systematic syntax, and also in the Latin Arsgrammatica from Varro to Donatus and Priscian, language data are constantly manipulated and regarded as examples. Whether they illustrate constructions or lexical formations, these examples are drawn from writers of prose or poems who are considered as classical. The examples are listed and form a virtually stable corpus, to be passed on and taken up, from one grammarian to another.20 The strong link between grammar and pedagogy in both Greece and Rome is well known. The grammatical example therefore becomes established as a natural tool for memorisation exercises and as a support for inductive reasoning. Grammatical art consists in discovering the underlying principles of these lists of examples. Perfectly logically, the notions of paradigm and systematic regularity follow from it. Thus the link between the example and the rule to whose discovery it leads has been, right from the origins of classical grammar, particularly strong.21 In the science of the exemplum, grammatical knowledge is thus built on the emblematic use of quotations taken from those who are already qualified as the ‘‘best authors’’. However the teaching of grammar and rhetoric also belongs to propaedeutics. It is closely related, even more so during the Middle Ages and in the Classical period, to the teaching of logic and the learning of right-thinking, in the same way that spelling consists in the learning of right-writing (and right-reading). By what is right is meant what is regulated, that is, what is compliant with the immanent rules which grammarians and philosophers discover by contemplating their corpora of examples, rules which constitute the very foundation of true thinking.22 16 ‘‘Qu’est-ce en effet qu’une chose? La chose s’oppose à l’idée comme ce que l’on connaît du dehors à ce que l’on connaît du dedans. Est chose tout objet de connaissance qui n’est pas naturellement compénétrable à l’intelligence, tout ce dont nous ne pouvons-nous faire une notion adéquate par un simple procédé d’analyse mentale, tout ce que l’esprit ne peut arriver à comprendre qu’à condition de sortir de lui-même, par voie d’observations et d’expérimentations [. . .]. Traiter des faits d’un certain ordre comme des choses, [. . .] c’est en aborder l’étude en prenant pour principe qu’on ignore absolument ce qu’ils sont, et que leurs propriétés caractéristiques, comme les causes inconnues dont elles dépendent, ne peuvent être découvertes par l’introspection la plus attentive’’ (Durkheim, 1927, preface to the second edition XIII). 17 ‘‘La notion de langue est en elle-même un haut niveau d’abstraction. Les linguistes ont toujours à juste titre, procédé à une idéalisation: donnons nous disent-ils, l’idée d’une communauté linguistique homogène. C’est le seul moyen de procéder rationnellement. (. . .). Vous devez abstraire un objet, vous devez éliminer les facteurs non pertinents. Du moins si vous voulez en faire une étude non triviale’’ Chomsky (1977, pp. 74–75) (quoted from the French original version). 18 Aarts (2000, p. 5) questions Chomsky: ‘What is your view of modern corpus linguistics?’ NoamChomsky’sanswerisclearcut: ‘It doesn’texist.’ 19 ‘‘La sociolinguistique qui est censée naître de la sociologie et de la linguistique, ne tirera rien de la sociologie [. . .]. Vous pouvez aussi collectionner des papillons et faire beaucoup d’observations. Si vous aimez les papillons c’est très bien; mais cette activité ne doit pas être confondue avec la recherche rationnelle. [. . .] La lutte contre l’idéalisation est la lutte contre la rationalité; elle signifie n’ayons pas de travail intellectuel significatif’’ Chomsky (1977, pp. 74– 75). 20 On the emergence of the grammatical theory and the successive use of the treatises by various authors Cf. Baratin et al. (1981). 21 ‘‘Les exemples, reconnus comme tels, permettent de faire l’économie de la règle qu’ils symbolisent’’ Holtz (1981, p. 109). I owe to my friendship with Madeleine Keller the numerous references on ancient grammarians and on the research of the Ars Grammarita group she organises. Cf. Grammatica (2005), also Keller (2009). 22 ‘‘Art de penser et non point art de bien penser, parce qu’un art a toujours pour tâche de donner des règles; que les règles définissent toujours une action correcte et qu’il n’y a pas plus d’art de mal penser qu’il n’y a de règles pour peindre mal. La pensée incorrecte est une pensée sans règle; et une règle qui ne serait « point bonne » ne saurait en aucune manière être considérée comme une véritable règle’’ (Foucault, 1967, p. 7). B. Laks / Language Sciences 39 (2013) 31–53 35 In the framework of this purely grammatical approach, language facts are not in themselves consistent. They only intervene as mere supports for hypothetico-inductive abstract thought. Corpora of examples thus enable the grammarian to build, thanks to idealisation and abstraction, a language theoria, hexqi9 a, which in Greek thought was precisely conceived of as a way of seeing, of contemplating things, through an inductive reasoning, in order thus to grasp the very essence of things and thus gaining access to Knowledge. According to its very definition, an example is thus speculative; it enlightens because it is a call to contemplation, out of which theoria arises through illumination. A vehicle on the path to knowledge for the Master, it is also a path to truth for the student whose progress it is helping towards the mastery of right-thinking by imitation.23 It is a school and, as such, its virtue consists in transcending the initial imitation so that, thanks to mental integration, it can lead to edification (ædificatio). The value of the example is therefore not only a function of its exemplary value for imitation. It is valuable also for its inferential driving function.24 As Valery noted appositely (1941), the example is never used directly, it is applied to the paradigm and leads to the rule, which it names and thus helps us memorise: ‘‘Quia nominor leo ne veut pas dire ‘‘parce que je me nomme lion’’, mais ‘‘je suis une règle de grammaire’’’’. 2.2. Exemplum linguistics today: Generative Grammar Generative Grammar falls within this tradition. Indeed, from the moment that it rightly rejects in principle the variable data of usage and chooses to found its reasoning on corpora of examples created thanks to grammaticality judgements, it is led to adopt an epistemological framework which Chomsky (1966) theorised as that of Cartesian linguistics. During the progressive emergence of this theoretical framework, the issue of linguistic data was essential. Regarding 1950s linguistics, the structuralist corpus was explicitly understood as a compendium of attested usages. It was precisely through his vigorous critique of this that Chomsky broke away from Harris. By objecting to performance data and by breaking away from datum linguistics, Generative Grammar explicitly rallied exemplum linguistics. Criticism of the syntagmatic model (Chomsky, 1957) and instrumentalisation of the Gold theorem (1967) converged to define a new linguistic and cognitive approach (Chomsky, 1965, 1968) for which the data to be taken into account are no longer constituted by the spontaneous productions by speakers in communication situations, but are the result of an introspective study of constructions removed from their context, to be used as exempla (Cf. Supra note 16 the criticism of introspection by Durkheim). It is therefore entirely logical that Chomsky (1995) should end up by laying down the internal language of thought as the object of Generative Grammar, as opposed to the external observable language, thus resuming the path which was opened by Port Royal.25 From then on, Generative Grammar presents itself as a state of the mind theory, which intends to model the cognitive functions specific to the human language faculty. Its purpose is no longer to analyse usage facts (performance) but is to (re)construct the logic of the mental processes involved in the syntactic organisation of sentences (competence). As Milner wrote (1989), Generative Grammar, in order to characterise these linguistico-cognitive processes, endeavours to define what is the possible but especially what is the impossible of language and languages. The judgement as to the grammaticality of sentences is thus the instrument of this partition. It constitutes the real observable which linguistic theory must account for. Thanks to grammaticality judgement, a corpus of examples pertaining to the formally correct (or incorrect) functioning of the cognitive capacity specific to a language is constructed. The formalisation of this particular grammar (PG), with specific parameters in turn enables the grammarian to infer and formalise the principles of universal grammar (UG) which characterises the language faculty specific to the species. In this sense, Generative Grammar is a Cartesian theory of mind, the object of which is categorically not the system of inter-individual communication between social individuals, but the (universal) language of human thought. The ideal and theoretical speaker–listener, the source of grammaticality judgements and examples, is not therefore in any sense a social individual. As Encrevé (1986) vigorously noted, the Chomskyan speaker is deaf and dumb, does not communicate and does not have any interindividual relations. He is involved in a radical solipsism, with no perspective of dialogue whatsoever. Ultimately, UG and PG are based on a corpus of examples with no direct link to the usages observable in a given speech community. Chomsky’s ostentatious repudiation of the inventorying of facts and of linguistic surveys, an essential component of datum linguistics, thus becomes easier to understand. Objecting to the taxonomic basis of scientific thought, he grounds his approach on a strictly hypothetico-deductive epistemology of sciences. He scornfully underlines that usage 23 One can recognise the triptych Contenplatio, imitatio, illuminatio, which in Christian mysticism, for instance in Saint Bonaventure’s work, leads to the ædificatio. One can also be reminded of role played, beyond the actual work of Thomas a Kempis, by L’imitation de Jésus-Christ. 24 ’’Ce n’est plus l’imitation qui est requise de l’apprenant par production et confrontation des modèles, mais l’intellection. On lui demande de comprendre la rationalité du système de la grammaire, les ‘‘causes’’ comme disent les maîtres. Sanctius, Sanchez de Las Brozas: De causis linguae latinae: ‘‘Qui a jamais dit: Ego amo deum et deus amatur a me? Dira-t-on aussi: facioorationem do tibidamnum et bien d’autres tournures du même genre. C’est auprès des meilleurs auteurs qu’il faut apprendre le latin et non chez les grammairiens. La grammaire n’apprend pas à parler latin, mais elle renvoie la langue latine à un art de sorte qu’ensuite, par imitation du latin, on puisse parler’’ (Chevalier, 2007, p. 155). 25 ‘‘We are concerned, then with states of language faculty, which we understand to be some array of cognitive traits and capacities, a particular component of the human mind/brain. The language faculty has an initial state, genetically determined; in the normal course of development it passes through a series of states in early childhood, reaching a relatively stable steady state that undergoes little subsequent change (. . .). We call the theory of the state attained its grammar and the theory of the initial state universal grammar. (. . .) When we say that Jones has the language L, we now mean that Jones’s language faculty is in the state L [. . .] To distinguish this concept of language from others, let us refer to it as I-language, where I is to suggest ’internal’, individual’, and intentional. The explanatory model outlined deals specifically with language acquisition under the idealized conditions of an homogeneous speech community. (. . .) The (acquisition) process is (viewed) as if it were instantaneous’’ Chomsky (1995, pp. 18–19). 36 B. Laks / Language Sciences 39 (2013) 31–53 data do not, any more than the collecting of butterflies, provide the foundation for a truly scientific research practice (Cf. Supra note 19). 2.3. Datum sciences Regarding linguistics and philology, the description of usage is as old as the grammatical perspective I have just mentioned. Where exemplum linguistics only considers language facts as an inferential support, datum linguistics is built as a minute observation and description of usages attested in their diversity, heterogeneity and variation. The notion of usage therefore takes over from the exemplarity of observed data. In the works of Cicero, Horace and of the great rhetoricians,26 and later in that of Quintilian after them, usage always comes first.27 For such instances of datum linguistics, the grammarian’s first task is thus to draw up lists of language facts which can more precisely be called corpora of observed usages. It is these inventories, which are the foundation of linguistic regularities, insofar as, just as Meigret (1542) stipulated at the dawn of the history of the French language, ‘‘les règles sont dressées sur l’usage et façon de parler lesquels ont toute puissance, autorité et liberté’’. Rules are not therefore inferred from examples as from immanent principles; they are deduced ‘‘de la commune observance qui comme une loi nous les a tacitement ordonnées’’.28 We may note that we are actually very close to what was theorised several centuries later under the expression ‘‘speech community’’, a notion, which, as we have already seen supra is articulated in Saussure’s work as a collective awareness supporting a shared common linguistic knowledge. It remains nevertheless that while the datum grammarian is a collector of usage facts to which he tries to give meaning,29 the theoretical oppositions remain strong. Among contemporaries, the various positions are clear-cut, because if the inventorying of usage constructs reference corpora, the latter, implicitly or explicitly, disclose what are highly diverging construction and conclusion principles. So who governs usage, and what is the prescriptive value of such usage? Such are the questions which arose, and we may note that the field of descriptive usage-based linguistics during the 16th and 17th centuries was rife with conceptions opposing the linguistic of the datum. At a time when Modern French and its first grammars were being created, the debate thus opposed prescriptivist normativists and certifying descriptivists. In line with Meigret, Ramus (1562), to take an example, confers linguistic authority on the people, that is to say the entire community of speakers, for them to rule language,30 whereas Vaugelas restricts such quasi-judicial power to a few chosen elites.31 It thus becomes clear that the normative and prescriptive options, just like the descriptive ones, are also compatible with datum linguistics. The difference lays in the corpora construction and conclusion principles, and in the explicitly normative quality, which grammarians confer upon the data. Though the issue is not, or is only marginally a matter of debate in the case of exemplum linguistics, where it is always possible to regard any example built by means of a qualitative judgement or a selective digest as a practical internalisation of the intended norm (Cf. Bourdieu, 1982), for datum linguistics on the contrary, prescription and description represent antithetic choices. Ultimately, it is not only the type of survey which draws the line. Much more crucial is the question of the theory of the linguistic survey carried out. Either the survey aims at finding proper usage by selecting within a speech community the legitimate representatives, or it aims at describing the linguistic object that is active in the said community. In the first case, the corpus shows the correct forms it considers as patterns and incidentally stigmatises the most vulgar solecisms32 and in the second case, it describes usages in their attested heterogeneity and variability. Of course, any linguistic data is always natural, when it is produced by a speaker, including such instances where it is constructed by linguists or by a speaker by means of a grammaticality judgement. The only requirement is to know how and for what purpose it has been produced. In answer to Milner, who had postulated the sole primacy of grammaticality judgement, Bourdieu reasserted the datum character of any linguistic production, including errors or impossible sentences made by grammarians for argumentative purposes, once their social and pragmatic production context is taken into account.33 He thus considers all data to be the product of a situated observation or of a questioning: of a survey. It is therefore 26 ‘‘Licuit semperque licebit Signatum praesente nota producere nomen. [. . .] Multa renascentur quæ jam cecidere, cadentque Quæ nune sunt in honore vocabula, si volet usus, Quem penes arbitrium est et jus et norma loquendi’’ Horace (457/1944). 27 ‘‘Consuetudo uero certissima loquendi magistra, utendumque plane sermone ut nummo, cui publica forma est. Omnia tamen haec exigunt acre iudicium, analogia praecipue, quam proxime ex Graeco transferentes in Latinum proportionem uocauerunt’’ Quintilianus (90/1842, I, 6, 3). 28 ‘‘Je confesse que cela serait raisonnable, si les règles qu’on fait de grammaire, commandaient à l’usage: vu qu’au contraire les règles sont dressées sur l’usage et façon de parler’’ Meigret (1542, p. 46). On these issues Cf. Glatigny (1982). 29 Just as Glatigny noticed (1982, p. 104) ‘‘Alors que Palsgrave prescrit, que Sylvius déduit, [Meigret] explique.’’ 30 ‘‘Le peuple est souverain de sa langue et la tient comme un fief de franc alleu, et n’en doit reconnaissance a aucun seigneur. L’école de cette doctrine n’est point es auditoires des professeurs hébreux, grecs et latins en l’Université de Paris; elle est au Louvre, au Palais, aux Halles, en Greve, à la place Maubert’’ (Ramus, 1562). 31 ‘‘De ce grand Principe, que le bon usage est le Maitre de notre langue, il s’ensuit que ceux-là se trompent, que en donnent toute la juridiction au peuple [. . .], C’est la façon de parler de la plus saine partie de la Cour, conformément à la façon d’écrire de la plus saine partie des Auteurs du temps. Toutefois quelque avantage que nous donnions à la Cour, elle n’est pas suffisante toute seule de servir de règle, il faut que la Cour et les bons Auteurs y concourent, etce n’est que de cette conformité qui se trouve entre les deux, que l‘Usage s‘établit’’ (Vaugelas, 1647/1934). 32 It is not to be forgotten that good usage treatises are usefully completed with prescriptions ‘‘ne dites pas, mais dites’’ based on the model of the Appendix Probi appended to the Instituta Artium. 33 ‘‘La sociologie se donne pour première tâche le recensement des formes d’acceptabilité, c’est-à-dire de la relation entre une phrase et les situations où elle est acceptable. [. . .] Donc, premier travail du sociologue: recenser. Ensuite, il reste à faire la science des conditions de production de la phrase et des conditions de son acceptabilité’’ Bourdieu in Bourdieu et alii 1977, p. 45). B. Laks / Language Sciences 39 (2013) 31–53 37 always necessary to indicate explicitly what was the survey situation in which a datum was inventoried, because it is only by interlinking such data, be it incorrect or absurd, to its social production and reception conditions that one can build an explanatory analysis (Cf. Supra Meillet’s position). Contrary to exemplum grammar, which hypostatises grammaticality judgement, ‘‘La sociologie (ou, si l’on veut, la sociolinguistique comme branche de la sociologie) [. . .] s’accommode de toutes les formes d’acceptabilité; son datum, c’est la relativité absolue de l’acceptable et non l’absolutisation d’une forme particulière d’acceptabilité. [. . .]’’, (Bourdieu in Bourdieu et al., 1977, p. 45). As a result, datum (socio)linguistics is ultimately based on the survey theory which, implicitly or explicitly, is its founding principle. Thus, as Encrevé proved (1976, 1982), it is by starting out from a deconstruction of the observer paradox and by aiming at a vernacular state of language that Labov (1966, 1976, 1979) builds the specific datum on which his (socio)linguistics is established. 2.4. Datum linguistics today: structuralism Just like exemplum linguistics, datum linguistics is to the forefront in contemporary approaches. As I have shown elsewhere (Laks, 2008), structuralism, whether European or American, founds its epistemology on the inventorying and systematic listing of observed data. Like all the great modern scientific projects, from Linnaeus (1735) and Buffon (1749) to Lamarck (1809) and Darwin (1859) it is a taxonomic science. In its formal expression, whether it takes the form of a structural (Troubetzkoy, 1939; Hockett, 1942), a functional (Martinet, 1962), or a transformational model (Harris, 1951), the aim is always to establish a system based on the reasoned taxonomy of a corpus of linguistic facts. Such a corpus is the product of a computation and therefore of a survey. In this survey, the notion of speech community is essential. Contrary to the Chomskyan approach, which reduces speech community to a single virtual point of which the ideal speaker–listener represents to some extent the eponym, the structuralist survey gives an operative force to the notion of communication relation. The latter enables the linguist to define the speech community as the network of potential interlocutors. Indeed, for structuralists, at least since Bloomfield, the first object of the linguistic survey is not language but the speech community: language is what can be observed within a given speech community as a means of communication between individuals. The social and cultural nature of language, as understood by Whitney and Saussure, is thus the foundation for the inventory of linguistic observables within the perimeter of all speakers using the same code. Imagine, says Bloomfield (1933, p. 46), that each time a speaker addresses a linguistic occurrence to another speaker, a line is drawn between these two social individuals. After a while, thanks to the density of its black, grey and white areas, this virtual graph would enable us to map both the speech community and the language which organises it, while showing the relative communication and exchange densities. We thus see that the notion of communication network is an integral part of this approach. Of course, these networks are not sociologised. Communication is still considered here as asymmetric and reciprocal articulation, while the social relations of domination and prestige, which always render exchange unequal, are ignored. However as Labov did in his early work, you only have to add a sociological dimension to the linguistic survey (Labov, 1966), or, even better, as the later Labov would do,34 to deconstruct the single speech community, taking into consideration the asymmetry of social relationships (Labov, 1972). Or, alternatively, like Milroy (Milroy and Milroy, 1985), it suffices to question the social reticulation of networks, in order to overcome the structuralist linguistic survey and build a specific sociolinguistic kind of survey. In all cases, as Hymes underlines (1972, p. 43): ‘‘The natural unit for sociolinguistic taxonomy is not the language but the speech community’’.35 What marks the difference between structural sociolinguistics and classical dialectology is precisely the attention brought to the internal social organisation of the speech community. Where dialectology saw variation and change as signs of the disintegration of the original community and of its language, and therefore placed the NORMS36 at the heart of its linguistic investigation, modern sociolinguistic approaches seethe speech community as being fundamentally heterogeneous, divided, hierarchised and structured. They thus conclude that sociolinguistic variation is indeed at the heart of truly living, that is to say socialised, languages. According to this approach, the speech community never shows any uniformity of usage and linguistic practices. On the contrary, the latter are as stratified and diversified as the social structure itself. Communication relationships are thus governed by a body of social and linguistic norms that are recognised by everyone who assesses and hierarchises these differentiated practices in a way that is common.37 Dissensus of usages and therefore of practices, and consensus of norms and of assessment systems: we are close here to the Durkheimian consensus as it was reformulated by Saussure: the social dimension of the linguistic treasure put in the brain of each individual. The speech community, which is at the heart of structural linguistics and, on a broader level, of all datum linguistics, is therefore primarily the place of heterogeneity, differentiation and contrast, antagonism and conflict. As long as one considers it at the individual level, it appears to be scattered and variable. However as soon as one considers it as a social structure and 34 On the distinction between the first and the second Labov, Cf. Encrevé (1976). For a detailed analysis of the concept of speech community Cf. Patrick (2002). 36 Acronym created by Chambers and Trudgill (1980) to designate the ‘‘Non-mobile, Old, Rural, Male Speakers’’ which represent the reference speakers of classical dialectologists. 37 ‘‘The speech community is not defined by any marked agreement in the use of language elements, so much as by participation in a set of shared norms. These norms may be observed in overt types of evaluative behavior, and by the uniformity of abstract patterns of variation which are invariant in respect to particular levels of usage’’ (Hymes, 1972, p. 120). 35 38 B. Laks / Language Sciences 39 (2013) 31–53 as soon as individuals are apprehended as social agents who are caught up in this canvas, the speech community presents itself as a system for the functional control and organisation of such heterogeneity, a system whose keystone is the recognition of common norms, which does not mean that they are equally shared. In the terminology of Bourdieu’s sociology terms, heterogeneity and the social differentiation of agents are fragmentation principles, while the legitimate recognition of social categorisation and organisation principles reconstitutes the uniqueness of the social. This is why social conflicts can be considered as conflicts about the legitimate definition of what legitimately applies to all of us: norms, classes, hierarchies. What at the individual and interindividual level appears to be variation and heterogeneity, manifests itself, at the social level, that of the speech community, to be functional and organised. It is thus in the speech community that the act of the speaker, an eminently labile act, is constituted as a functional social fact, that is to say as datum for (socio) linguistics.38 3. Datum and exemplum: the turn of the 2000s The landscape of contemporary linguistics is still structured in terms of these fundamental theoretical oppositions. In the 1960s, the break with structuralism and the advent of Cartesian linguistics (Chomsky, 1965, 1966) solidly established exemplum linguistics, which Generative Grammar is, as the dominant linguistic paradigm. This led to a negation of the structuring character of interpersonal communication. Indeed, as we have just noted, Generative Grammar, like all exemplum linguistics, substitutes an ontological individual for the socialised speaker–listener, thus denying any linguistic relevance to very concept of speech community and is inevitably led to deny any functional character to interpersonal communication. As Chomsky forcefully reminds us in his most recent analyses, for Cartesian linguistics, human language is not defined or shaped by communication. An approach in terms of the social or communicative function is rejected de jure and de facto: neither from a phylogenetic point of view nor from an ontogenetic point of view is language constituted for functional needs, be they communicative or otherwise.39 Chomskyan language is defined on foundations which are exclusively cognitive and mental and in a strictly individual approach.40 As Labov points out (1987), in the idealist conception of Generative Grammar, language is an ontological property of the individual, in the sense that he belongs to the homo species, whereas in the materialistic conception, language is a functional property of the social groups formed by the socio species.41 This inevitably leads to a marked divergence of approaches to what constitutes, for the linguist, a linguistic fact (Cf. Labov, 1975), a divergence where the datum/exemplum opposition just mentioned reemerges. Thus the issue is not only that of the lack of substance of the grammaticality judgement, which is very often evident in situations of sociolinguistic variation (Cf. Labov, 1996). At issue is the very characterisation of what, in terms of the linguist’s procedures of investigation, is defined as a linguistic fact. Having emphasised the fact that the functionalist approach to language as a communication tool in various social groups is inevitably led to take into consideration the differences which separate languages, so as to present a systematic description of it, by way of a structural method, Berwick and Chomsky (2011)42 thus underline that the non-functionalist mentalist approach, which sees language as a property of the human mind, is on the contrary led to reduce the surface or actualisation differences, in order to highlight, by way of formal induction, the unity and singularity of the cognitive functioning involved.43 It thus becomes evident that the opposition between structural linguistics and generative linguistics is not based on any technical or methodological arguments concerning the power of explicative models of grammatical facts or on the descriptive or predictive capacity of the analyses proposed. The opposition stems from what are radically differing approaches to what must be considered as a speaker, a language fact, the role of communication, or the very existence of communities of speakers. With the Cartesian turn in the 1960s, Chomskyan linguistics brought to the foreground an idealist and abstract mentalist approach which, as in the 17th century, necessarily depended on exemplum linguistics, excluding ex hypothesis all structural variation and heterogeneity. 38 ‘‘A SpCom is defined in functionalist terms as a system of organized diversity held together by common norms and aspirations. . . Members of such a community typically vary with respect to certain beliefs and other aspects of behavior. Such variation, which seems irregular when observed at the level of the individual, nonetheless shows systematic regularities at the statistical level of social facts’’ (Gumperz, 1982, p. 24). 39 ‘‘It has been conventional to regard language as a system whose function is communication. This is indeed the widespread view invoked in most selectionist accounts of language, which almost invariably start from this interpretation. However, to the extent that the characterization has any meaning, this appears to be incorrect, for a variety of reasons to which we turn below.’’ Berwick and Chomsky (2011, p. 25 also pp. 35, 36). ‘‘Accordingly, any approach to evolution of language that focuses on communication [. . .] may well be seriously misguided’’ Chomsky (2011, 61). Cf. also Hauser et al. (2002, p. 1569). 40 ‘‘The word ‘‘language’’ has highly divergent meanings in different contexts and disciplines. In informal usage, a language is understood as a culturally specific communication system (English, Navajo, etc.). In the varieties of modern linguistics that concern us here, the term ‘‘language’’ is used quite differently to refer to an internal component of the mind/brain (sometimes called ‘‘internal language’’ or ‘‘I-language’’. We assume that this is the primary object of interest for the study of the evolution and function of the language faculty. However, this biologically and individually grounded usage still leaves much open to interpretation (and misunderstanding)’’ (Hauser et al., 2002, p. 1569). 41 ‘‘What is language? The idealist conception is that language is a property of the individual, a species-specific and genetically inherited capacity to form rules of a particular type, relatively isolated from other activities of the human intelligence. The materialistic conception is that language is a property of the speech community, an instrument of social communication that evolves gradually and continuously throughout human history, in response to a variety of human needs and activities’’ (Labov, 1987, p. x). 42 It is quite rare to find in Chomsky’s work a characterisation, even critic, of structural linguistics and of linguists which preceded him. The reference to Sapir, Boas and Bloomfield, just as that to European structuralism via Troubetzkoy and the highlighting of Harris, all deemed as linguists of linguistic diversity whereas he defends linguistic uniqueness, deserves to be underlined. 43 Such uniqueness and such singularity reside in the recursion principle of which the MERGE unification function is an example. I will come back to this later. B. Laks / Language Sciences 39 (2013) 31–53 39 3.1. Empirical re-grounding However as early as the end of the 1990s and the start of the 2000s, the international field of linguistics was to undergo considerable changes and datum linguistics would recover the place it had occupied 50 years earlier. In all fields of linguistic analysis, the volume of large-scale descriptive research multiplied and the empirical basis for analyses was thus increased. The very notion of linguistic data corpora, consistently castigated within the generative framework, was thus reasserted. No domains remain unaffected by this empirical re-grounding, which has led to an explicit critique of the limits and the dead-ends, which a procedure of analysis, based on a very limited number of examples taken up from one author to another,44 leads to. To take a single, supremely emblematic example, since the 2000s, the Innatist hypotheses of Generative Grammar, based on the thesis of the poverty of stimulus, which is thus incapable of leading to the acquisition of language (Berwick and Chomsky, 2008; Chomsky, 1968), can be usefully reappraised, in the light of the CHILDES observation corpora, which totals 44 million words in 32 languages and which amounted to 750 gigabits in 2004.45 To date, over 3200 publications refer to it. More generally, the considerable development of storage and information processing capacities, the emergence of new technological tools and the development of data sharing cooperative networks made possible by internet, has resulted, since the turn of the new century, in what is a patent clear empirical reorientation of linguistics which, to borrow the dichotomy formulated by Berwick and Chomsky (2011) reorientates language sciences towards the documentation of the diversity of phenomenologies at each level of analysis, as a prerequisite to all constructions of abstract hypotheses concerning the unity of the cognitive phenomenon. For the exponents of the generative paradigm, it is true that this constitutes a methodological reorientation which is devoid of any significant consequences, insofar as it impacts all the levels (phonology, morphology, syntax, pragmatic semantics, etc.) which they regard as peripheral, whereas their reductionist Innatist model concerns only the universal syntactic core (see above). However such a generalised empirical re-grounding has an enduring impact on the theoretical landscape of linguistics. Major figures of linguistic thought, who for a long period had been criticised or marginalised, Harris for instance, are making their way back to the centre of the inter-theoretical debate, with a surprising acuity (Nevin, 2002). The rehearsal of old characterisations, which, somewhat disdainfully, considered that the theoretical framework of transformational grammar (Harris, 1951) merely represented a methodological approach,46 are no longer sufficient to counter what a rereading of these propositions can offer to 21st century linguistics (Goldsmith, 2005b). As I have demonstrated elsewhere (Laks, 2008), and contrary to what continues to be maintained in the generative framework, the methodological approach to linguistic diversity and the descriptive analysis that is built on large sized corpora do not in any way preclude either the construction of general and abstract hypotheses or even the construction of an explanatory theoretical model. On the contrary, what has been reaffirmed since the turn of the century is the classical conception, subscribed to by science historians, which regards the procedure of systematic description, the reasoned classification of facts and the progressive construction of ascending systemic hypotheses as the very movement of modern scientific thought, leading to the construction of increasingly comprehensive explicative theories. Such a turn, to what Goldsmith (in press) defines as a neo-empiricism, is therefore marked by the emergence of many large-sized corpora and by the construction of a number of remarkably powerful methodological and theoretical processing tools. As he also demonstrates, the recent improvements in the fields of statistics, probabilistics and, more generally, in mathematic stochastics renew the formal and modelling approaches. Corpus linguistics, which constitutes the empirical and descriptive basis required for the improvement of these analyses, is thus becoming predominant in all the areas of research on languages and speech, including the field of formal grammars or the psycholinguistics of learning.47 Within the framework of such neo-empiricism, the issues of data variation and heterogeneity, along with the question of the variability and plasticity of models capable of accounting for it, questions which were for long had been excluded by exemplum linguistics, are now coming back to the foreground. Labovian variationist linguistics can thus no longer be regarded, as Chomsky did (1977, pp. 75), as a peripheral dialectology, devoid of theoretical interest. The quantitative analysis of large corpora and the systemic approach to the internal variation of the languages and grammars, which represent the heart of his scientific programme, are thus brought back to the forefront of current research (Labov, 2004).48 Such neo-empir44 Thus the first analysis of the liaison and the voiceless e in French in the generative framework (Schane, 1965) suggests 41 methodical rules based on 73 examples (Cf. Laks, 2011a). On the contrary, the ‘‘Phonologie du Français Contemporain’’ programme suggests, for the same phenomena, a basis of 190,000 and 47,500 relevant phonological sites which enable to completely reverse phenomenology. It is therefore possible to suggest for these classical questions on the phonology of French radically new analyses (Cf. Durand et al., 2011). 45 For a general presentation Cf. MacWhinney (2000, 2007).For a global overview Cf. Gleason and Thompson (2002). The figures of the Talkbank corpus which weight 450 gigabits for 55 million words in 18 languages need to be added, Cf. http://childes.psy.cmu.edu/. 46 ‘‘The publication that was the foundation of American structural linguistics in the 1950s, Zellig Harris’s Methods in Structural Linguistics (1951), was called ‘‘methods’’ because there seemed to be little to say about language beyond the methods for reducing the data from limitlessly varying languages to organized form. European structuralism was much the same. Nikolai Troubetzkoy’s classic introduction to phonological analysis was similar in conception. More generally, structuralist inquiries focused almost entirely on phonology and morphology, the areas in which languages do appear to differ widely and in complex ways, a matter of broader interest, to which we will return’’ (Berwick and Chomsky, 2011, p. 2). 47 Cf. for instance Goldsmith and Aris (2009) for an quantitative and formal analysis of learnability tabula rasa of phonological classifications. 48 It is certainly in the field of phonology, specifically optimalistic, that the relation between phonological modelling and variationsim was recently the most productive. Cf. for instance Antilla (2007), Antilla and Cho (1998), and Boersma (1998). For an approach to the relation between phonetics and phonology in this regard Cf. Hayes and Cziráky Londe (2006) and Hayes et al. (2004). Cf. also the acts of ‘‘Workshop on Variation, Gradience and Frequency in Phonology’’ (Stanford, CA, 2007): http://www.stanford.edu/dept/linguistics/linginst/nsf-workshop/workshop-july-2007.html. 40 B. Laks / Language Sciences 39 (2013) 31–53 icist reorientation, which Labov calls materialistic, has not however abandoned the theoretical ambition which is proclaimed by Cartesian linguistics. On the contrary, by taking up the functionalist concepts of situated interpersonal communication, of speech community and of interaction, the years 2000 have witnessed the emergence of new theoretical and analytical approaches to the language phenomenon. 3.2. From corpus linguistics to usage-based linguistics These new empiricist approaches, though based on a certain number of structuralist theses (functionalist approaches, the role of socialised interpersonal communication and the circumscription of speech communities) find their origin in what was an old division within Generative Grammar at the time of its emergence. As Goldsmith and Huck demonstrate (1995), the debate over ‘‘generative semantics’’, far from constituting a merely anecdotal or surface historical epiphenomenon, was in fact the actualisation, in the middle of the 1960s, of an extremely sharp fundamental rift within the emerging generative and transformationalist current. Even though at the end of the 1970s the current of generative semantics appeared to have definitively broken up and though the debate gradually was exhausted, the leading participants in the questioning of the orthodox generativist model remained very active and most of them found themselves at the heart of theoretical re-compositions which took place at the strategic points of cognitive linguistics in the 2000s. Insofar as these models play a central role in the current debate, it is necessary to take a brief look at what was a crucial moment of rupture, when a significant number of linguists broke with the standard Chomskyan model.49 Indeed, when one examines the main themes promoted by generative semanticists and the main criticism they make,50 one is struck by the contiguity with many contemporary issues. For Lakoff (1973a) for instance, it is absolutely necessary to take into account the social, cultural and interactional context within which language unfolds. His criticism of the Cartesian turn reasserts the functional character of language and he deems it necessary to place interpersonal communication at the heart of the actual functioning of language. One of the results of this period was the recognition of the necessity for a solid empirical and descriptive basis. Against Cartesian Innatism, Lakoff defends a social and cultural conception of language facts, a position which leads him to give primacy to the semantic and pragmatic content of linguistic occurrences when analysing the form and also the characteristics, both specific and universal, of language. Thus, a certain number of driving forces, which structure our most contemporary debate, were already active in the debate over generative semantics in the 1960s and 1970s. In the course of the 1990s, most of these actors would gravitate towards the cognitive linguistic movement. Today, cognitive linguistics represents a gathering point for many contemporary currents, including semantics and pragmatics, but also construction grammars and sociolinguistic variationism and connectionism. These currents, while retaining their specific orientations, cooperate and converge on a certain number of broad orientations. Langacker (1987) brought these currents together under the common name of ‘‘usage-based linguistic models’’. Such a convenient synthetic appellation designates a wide sector of international linguistic research sharing common positions, without its being united around a rigid common theory. Indeed, in stark contrast to Generative Grammar, this contemporary current of research, which defends the remarkable profusion of usage data advocates a certain austerity of the conceptual apparatus and the theoretical and formal framework.51 Within this framework, usage does indeed capitalise a considerable practical linguistic knowledge, active in all areas pertaining to the production and interpretation of occurrences, whereas the abstract and general cognitive principles, reduced to the minimum, are not specific to the field of language, corresponding to what are major and general cognitive functions. In these approaches, the status of linguistic practices is that of situated observables, where it is the richness of the background social and cultural context which largely determines their common interpretation strategies (Barlow and Kemmer, 2000, p. XXVI). Thus, in a framework structured by interpersonal relations and the communication function, language practices and their semantic and pragmatic interpretation are fundamental.52 Grammar is no longer a condition of the 49 Lakoff quotes Ross and MacCawley as constituting the initial core which Fillmore, Talmy and Langacker joined later, then in the emergence of cognitive linguistics Fauconnier, Rosch, Kay, MacDaniel. In the 1980s and 1990s, they converged with the defendants of connectionism Rumelhart, Feldman, etc. Lakoff (1973a) thus establishes a link between linguists who, over around forty years, were successively active in various circles: generative semantics, pragmatics, casual grammars, construction grammars, the theory of metaphor, connectionism, etc. 50 An analysis of the theoretical basis of the debate is not part of my topic. For a historical analysis, refer to Harris (1993) and to Goldsmith and Huck (1995) for a more epistemological analysis. The consolidated texts of generativist semanticists of which the subject is not a technical point or a specific argument are quite rare. I refer you however to Lakoff (1973a,b). The second constitutes an answer to the analysis, itself critical, of the debate between the generativist semanticists and the advocates of the standard model by Searle (Searle, 1972). 51 ‘‘I have argued, both on methodological and on empirical grounds, that the principle of generality have received in linguistics a commonly accepted interpretation that is in fact not appropriate to its subject matter. Current doctrine favors a minimalist account of linguistic knowledge, described in accordance with complex array of theoretical apparatus featuring specialized devices for the various ‘components’ of the linguistic system. By contrast, cognitive grammar pursues a maximalist account of linguistic knowledge, and tends toward austerity in adoption of theoretical constructs; it seeks a unified treatment of the various facets of linguistic structure, attributing their differences to the content of the domain in question rather than the basic constructs invoked to handle them’’ (Langacker, 1988, p. 160). 52 ‘‘A usage-based model is one in which the speaker’s linguistic system is fundamentally grounded in ‘usage events’: instances of a speaker’s producing and understanding language [. . .]. In this view, it does not make sense to draw a sharp distinction between what is traditional called ‘competence’ and ‘performance,’ since performance is itself part of a speaker’s competence. Instead of viewing language processing as something external to the system, which happens only to the outputs of competence, processing is rather to be seen as an intrinsic part of the linguistic knowledge system, which cannot he treated separately from it?’’ Barlow and Kemmer (2000, pp. VIII–IX) . B. Laks / Language Sciences 39 (2013) 31–53 41 production of language events. On the contrary, it is the routinised and impoverished sub-product (Barlow and Kemmer, 2000, p. XI) of these events. By granting a central place to the grammaticalisation process, seen both as cognitive and diachronic, usage-based approaches to linguistics entirely reverse the classical Chomskyan competence/performance dichotomy, granting performance a generalised status as the practical and functional drive of communication and relegating competence to a position as an internal sedimented sub-part of communication (Langacker, 2000, pp. 6–9). Echoing Benveniste and the discourse analysis current which emerged from it, Langacker clearly posits discourse in interaction as the source of language and of its grammatical routinisation. He thus adopts the neo-empiricist perspective mentioned above.53 As I have demonstrated elsewhere (Laks, 2011b), this was already Saussure’s position when he defended the primacy of speech linguistics as the sine qua non condition of a language grammar. By apprehending language practices in their ecological, social and cultural context, usage-based linguistics once again grant a central status to linguistic description, to distributional, statistical and frequency analyses and are unquestionably located within the movement of corpus linguistics.54 They are datum linguistics and, in correlation with this, they recognise internal variation and heterogeneity for what they are. But above all, in usage-based models, such sociolinguistic variation, both synchronic and diachronic, recovers a systemic and functional status.55 Lastly, both from the point of view of cognitive instantiation of linguistic regularities and the point of view of the situated acquisition of language and grammar, these approaches are parallel to the connectionist and neuromimetic approaches. Seen in more general terms, they participate in the dynamic modelling of language. With all their variants and related approaches, construction grammars, examplarist and occurrentialist grammars, discursive, neural and cognitive grammars, stochastic and probabilistic grammars, etc.,56 usage-based models and datum linguistics have profoundly changed the international linguistic field and, by marginalising Cartesian linguistics, they represent, as we shall now see, the dominant paradigm of the early 21st century. 3.3. Grammar and usage, a critical review In his presidential address before the Linguistic Society of America’s congress, Newmeyer (2003), who is renowned for being a fervent defender of Chomskyan linguistics (Cf. Newmeyer, 1988), confirmed such a shift and drew up a somewhat disabused picture of the international landscape of linguistic research. He confirmed that, internationally, the paradigm of Generative Grammar has become the minority, whereas usage-based models are now dominant. For Newmeyer, generative semantics, which integrated cognitive linguistics, eventually succeeded, with the assistance of usage-based linguistics, in imposing the rejection of all clear distinction between linguistic knowledge and usage of language, a distinction which is nevertheless at the heart of the Chomskyan paradigm (op. cit., p. 683). The cognitive autonomy and stability of grammar were seriously challenged and were cast aside in favour of a logic of transitory dynamics, partial structuring and stochastic organisation. These various approaches postulate an important storage of concrete forms and occurrences. They minimise the scope and effect of formal grammatical constraints (op. cit., pp. 683–684).57 These approaches have been very widely diffused and have extended their influence even to the field of optimality models, though the latter were initially close to generativism. Connectionism and neuromimetic systems, which in the 1980s came in for severe criticism (Pinker and Mehler, 1989), have flourished, providing a computational back-up for these approaches. Such was the assessment of the state of research by the President of the LSA in 2003. No field of research remains unaffected by this reference paradigm shift. Automatic language processing and formal grammar research have been impacted, along with phonology, syntax and semantics.58 Furthermore, the Chomskyan 53 To found the primacy of discourse, it is often quoted from Benveniste (1966, p. 131), without always seeing that the sentence he created for such purpose, ‘‘nihil est in lingua quod non priusfuerit in oratione.’’ is only the copy of the perfectly Aristotelian scholastic sentence ‘‘nihilest in intellectu quod non priusfuerit in sensu’’, which is the essence of the sensualist empiricism of Hume of Locke. 54 ‘‘A usage-based theory, whether its object of study is the internal or external linguistic system, takes seriously the notion that the primary object study is the language people actually produce and understand. Language in use is the best evidence we have for determining the nature and specific organization of linguistic systems. Thus, an ideal usage-based analysis is one that emerges from observation of such bodies of usage data, called corpora. But even if not based primarily on such data, at a minimum, analyses must ultimately be at least consistent with production data [. . .] ‘‘The importance of frequency: Because the system is largely an experience-driven one, frequency of instances is a prime factor in its structure and operation. Since frequency of a particular usage pattern is both a result and a shaping force of the system, frequency has an indispensable role in any explanatory account of language’’ (Barlow and Kemmer, 2000, pp. XV, IX). 55 ‘‘[There is an] intimate relation between usage, synchronic variation, and diachronic change: Patterns in usage data are in general patterns of variation along different dimensions of various kinds, from formal to social. In a cognitive usage-based model, variant linguistic forms can be thought of as alternate possibilities licensed by the linguistic network. The selection of a given entrenched variant for activation is governed by a complex set of motivating factors, including system-internal as well as contextual, situational factors. As observed in the seminal work of Labov, variation is highly structured, not only in the individual’s system, but across groups of speakers. The effects of usage on the linguistic system [. . .] lead us to expect that speakers’ language will be influenced by the productions they hear in particular speech communities of which they are members. [. . ..]’’ the more speakers talk to each other the more they will talk alike, and so linguistic variation will pattern along lines of social contact and interaction’’ (Barlow and Kemmer, 2000, XVII). 56 Goldberg (2006, 1995), Bybee (2001, 2006), Feldman (2006), Lakoff and Johnson (1999), Chater and Manning (2006), Goldsmith and Aris (2009), and Manning (2003). 57 Cf. Langacker (1998) mentioned supra note 52. 58 ‘‘I am quite sure that Christopher Manning is right when he writes that ‘[during] the last 15 years, there has been a sea change in natural language processing (NLP), with the majority of the field turning to the use of machine learning methods, particularly probabilistic models learned from richly annotated training data, rather than relying on hand-crafted grammar models’ (Manning, 2002b: 441)’’ (Newmeyer, 2003, p. 682). 42 B. Laks / Language Sciences 39 (2013) 31–53 competence/performance distinction has itself become obsolete59 and the importance of a strictly generative psycholinguistics has drastically declined.60 Finally, the functionalist and communicative option once again dominates the linguistic landscape and it is beyond any doubt that a coherent and substantial case has been put forward in support of usage-based linguistics.61 With this new paradigm, variation, heterogeneity and sociolinguistics, which previously were consistently criticised as epiphenomenal (Cf. supra), are making their way back to the foreground of linguistic theory. Moreover, if the very notion of the sentence is called into question, and if the complete disparity between what grammar predicts and what usage attests is henceforth generally admitted, what is there which remains of Cartesian linguistics and Generative Grammar?62 As is usual when the Chomskyan paradigm is the object of a large scale attack, instead of envisaging a debate with its critics, or the option of an evolution or adaptation which might integrate the arguments of its opponents, what the 2000s witnessed was the complete overhaul of the reference framework of Generative Grammar.63 With the emergence of Biolinguistics, the reorganisation of the generative theory has endeavoured to impose a radical change in the terms of the debate. Henceforth the issue would appear to be located in the field of the evolution of species, where Chomskyan Biolinguistics sets out to prove the specificity of the faculty of language specific to humankind. Referring to the distinction between the two extensions of the concept of the faculty of language, which were previously postulated in the minimalist programme (Chomsky, 1995) and reformulated in the new biogenetic evolutionist framework (Hauser et al., 2002), Newmeyer opts for a defence and illustration of Generative Grammar on the basis of a radical distinction between grammar and usage. In the context of an argumentation which concedes everything to usage-based models, he thus takes up the Chomskyan line of argument that Generative Grammar and Biolinguistics are not linguistics, in the sense that their subject is not language in the commonly accepted sense of the term. Their only interest is in the cognitive disposition, which is the principle of language. Thus, thanks to a scorched earth policy, Newmeyer concedes that datum linguistics and usage-based models satisfactorily cover the entire span of the linguistic field, but he goes onto say that grammar remains, and that in its principle it must be radically distinguished from usage. Such a cognitive and grammatical remnant is thus the very object of Generative Grammar and Biolinguistics, whose only concern is the faculty of language in the narrow sense, and not at all in the broad sense. Indeed, the Faculty of Language in the Broad sense (FLB) is by no means specific to humankind. It covers all the communicational, interactional, social and cultural aspects. As Chomsky (1995) had already noted, after Fodor (1983a), it comprehends all the peripheral systems and covers language phonetics, phonology, morphology, semantics and pragmatics. In short, everything that is not syntax. And yet the Faculty of Language in the Narrow sense (FLN), defined in opposition to it, does not cover all syntax, since its exclusive concerns are core syntax or, more precisely, one of its computational principles: recurrence, as it expresses itself through the MERGE principle. In his endeavour to save the very heart of generative grammar, Newmeyer thus concedes all FLB to usage-based linguistics. The aim is, in collaboration with Hauser et al. (2002), thus to focus Biolinguistics on the analysis of the FLN.64 In the debate between datum linguistics and exemplum linguistics, between usage-based models and Cartesian linguistics, we thus come back in fine to the cognitive issue. 59 ‘‘I believe that the great majority of psycholinguists around the world consider the competence-performance dichotomy to be fundamentally wrongheaded’’. (Newmeyer, 2003, p. 682) 60 The observation had already been stated by Tomasello (1995, p. 135). ‘‘The list [of innate aspects of language] contains things that no nonlinguist would ever recognize: such things as the projection principle, the empty category principle, the subjacency constraint, and the coordinate structure constraint. All of these universals are described in linguistically specific terms such that it is very difficult to relate them to cognition in other psychological domains’’. 61 ‘‘First and most importantly, there is the evidence that has mounted n the past quarter-century that significant aspects of grammars are motivated by considerations of use. Functional linguists and generative linguists with a functional bent have provided (to my mind) incontrovertible evidence that grammars are shaped in part by performance considerations’’ (Newmeyer, 2003, p. 683). 62 ‘‘Reinforcing skepticism about classical generative models is the disparity between sentences generated by these grammars and actual utterances produced by language users. This disparity has led some linguists to conclude that grammar itself bears no relation to the proposition-like structures posited by formal linguists; structures specified by formal rules that take the sentence to be the basic unit of grammar, where sentences are in a rough mapping with propositions, verbs with predicates, and noun phrases with logical arguments. The priority of the sentence is dismissed by some critics of the generative program as a carryover from the Western logical tradition, reinforced by the conventions of written language’’ (Newmeyer, 2003, p. 683). 63 As it has been pointed out elsewhere (Encrevé, 2000), the Chomskyan current is often inclined to a hagiographic rewriting of its own history. Thus, because Chomsky (2007, p. 9) dates the creation of the term biolinguistics back to the critical review which Piatteli-Palmarini made in 1974 on the debate he had in Royaumont with Piaget, all the advocates of the new framework date its creation back to this period (for instance Di Sciullo and Boeckx, 2011; Berwick, 2011; Boeckx and Grohmann, 2007). The historians which are said to have noticed the absence of the term in the title of the French or English version of the debate (Piatelli-Palmarini, 1980; Piatelli-Palmarini and Noizet, 1979) and to have underlined a more constant reference to the issue of acquisition in a Innatist vs. Constructivist framework and therefore to have suggested a historically inaccurate reading. Chomsky (2007, p. 9) goes further and dates the biolinguistic framework to his first works in the 1950s, Van Riemsdijk is somewhat dubious: ‘‘I sort of said jokingly that in retrospect, if you tell Chomsky that he only came up with that stuff later, he would probably deny it and say that it had been clear to him right from the start — and, you know, who am I to say that he would be lying? All I’m saying is there was no real evidence in the writing that that was the main goal he was pursuing. It would actually be interesting at some point to ask him this question’’ (Grohmann, 2007, p. 138). 64 We cannot agree with Newmeyer when he quotes Saussure to this end. As I have shown elsewhere with many others (Laks, 2011b), and contrary to the vulgate of the Cours, Saussure clearly distinguishes the grammar of the linguist and that of the speaker. The second is usually impenetrable because it constitutes a formation of compromises of usage variation. For Saussure, absolutely, Language Sciences are on the side of usage analysis (what he calls speech linguistics) in their social and cultural dimensions (i.e. semiological). B. Laks / Language Sciences 39 (2013) 31–53 43 4. Heterogeneity and variation: the cognitive issue As we have just seen, with the paradigm of Biolinguistics, Chomskyan linguistics has adopted a programme which is new only in appearance. Since the Minimalist programme, the distinction between FLB and FLN was in place, with the focus being placed on the individual cognitive faculties and the internal representational states via the distinction between E-language and I-language.65 What is new is the exclusive focus of Cartesian linguistics on the FLN and I-language. All other dimensions are abandoned and left to usage-based models and to datum linguistics. Sociologists of scientific knowledge will have no option but to regard it as the effect of an inter-theory balance of power, such as was observed by Newmeyer supra. 4.1. Innatism, essentialism and rationalism By reframing its programme, Biolinguistics only needs to address its attentions to what is an extremely limited object. It therefore no longer needs to take into account or answer arguments other than those related to its specific focus: the faculty of language in the narrow sense.66 But Chomskyan reductionism is even more radical: the FLN only contains properties which are strictly specific to human languages, and among these, only those which cannot be acquired by exposure to external data. These specific properties can actually be reduced to a single one, the computational mechanism of recursion, and even the latter is apprehended under the very specific form of the MERGE syntactic principle.67 All the rest pertains to the FLB and is therefore external to biolinguistic investigation and can thus be left to usage-based linguistics. The latter can very well be envisaged within a Darwinian and Evolutionist framework in which the emergence of a superior and complex communication form provides a decisive selective advantage to the species. MERGE cannot do this. For Chomsky, such an operation, which is the essence of all core syntax, does not consist in a gradual adaptation but in an abrupt break. Against the classical Darwinian axiom, Berwick (2011) strongly underlines that natura [syntax] facit saltum, with the result that the emergence of core syntax is thus void of any Evolutionist determination.68 In accordance with Cartesian rationalism, Chomskyan biolinguistics therefore postulates MERGE as the ultima ratio in which the (restrictively) human nature of mankind resides. Indeed, if, as Berwick and Chomsky (2011) admit, the computational mechanism of recursion is found in various domains of animal and human cognition (numeration, structuration of visual scenes, etc.),69 MERGE does not directly result from it, it rather constitutes an exaptation from it. The term is not used here in its original meaning (Gould and Vrba, 1982), where the reutilisation for new objectives of an existing cognitive function is carried out under the adaptive pressure and is therefore a response to the selective process, but in a purely discontinuist and catastrophic sense: the recursion function of the human being, active in various cognitive domains, suddenly and for no reason whatsoever, finds itself being applied to existing communication systems. This non-motivated break, which takes place in the brain of a single individual, is decisive: syntax, and grammar along with it, the human language and the homo species, are born. The resulting selective advantage is considerable, but it is not what motivates the process.70 This point is decisive. Indeed, if syntax only results from an adaptive evolution, if there is no decisive break and if such adaptation takes place within the group, in order to consolidate its gregarious, social, cultural and symbolic organisation, then it is the group’s evolution which is the driving force. In other words, what is implied is a functionalist approach, which places the interpersonal communication function at the heart of the linguistic disposition and which makes the emergence of human language a moment of evolution, both led and motivated by the selective advantage resulting from a better social organisation and a better symbolic regulation of the group. On the basis of such hypothesis, the FLB covers the entire field, 65 ’’We are concerned, then with states of language faculty, which we understand to be some array of cognitive traits and capacities, a particular component of the human mind/brain. The language faculty has an initial state, genetically determined; in the normal course of development it passes through a series of states in early childhood, reaching a relatively stable steady state that undergoes little subsequent change (. . .) To a good first approximation, the initial state appears to be uniform for the species. (. . .) we call the theory of the state attained its grammar and the theory of the initial state universal grammar. (. . .) The initial state is in crucial respects a special characteristic of humans, with properties that appear to be unusual in the biological world. (. . .) When we say that Jones has the language L, we now mean that Jones’s language faculty is in the state L (. . .) To distinguish this concept of language from others, let us refer to it as I-language, where I is to suggest ’internal’, individual’, and intentional’’ (Chomsky, 1995, pp. 18–19). 66 ‘‘We hypothesize that FLN only includes recursion and is the only uniquely human component of the faculty of language. We further argue that FLN may have evolved for reasons other than language, hence comparative studies might look for evidence of such computations outside of the domain of communication (for example, number, navigation, and social relations)’’ (Hauser et al., 2002, p. 1569). 67 ‘‘In this sense, there is no possibility of an ‘intermediate’ language between a non-combinatorial syntax and full natural language syntax — one either has Merge in all its generative glory, or one has effectively no combinatorial syntax at all, [. . .], in a sense there is only a single grammatical operation: Merge. Once Merge arose, the stage for natural language was set. There was no turning back.’’ Berwick (2011, p. 99). Cf. also Berwick and Chomsky (2011), Chomsky (2004, 2011), and Hauser et al. (2002). 68 ‘‘However, unlike Linnaeus’ and Darwin’s slogan shunning the possibility of discontinuous leaps in species and evolution generally—natura nonfacitsaltum — we advocate a revised motto that turns the original on its head: syntax facitsaltum — syntax makes leaps — in this case, because human language’s syntactic phenotype follows from interactions amongst its deeper components, giving it a special character all its own, apparently unique in the biological world’’ (Berwick, 2011, p. 6). 69 ‘‘Only FLN is uniquely human. [. . ..] we hypothesize that most, if not all, of FLB is based on mechanisms shared with nonhuman animals. In contrast, we suggest that FLN—the computational mechanism of recursion—is recently evolved and unique to our specie’’ (Hauser et al., 2002, p. 1572). 70 ‘‘Unbounded Merge (hence displacement) must have arisen from some rewiring of the brain, hence in an individual, not a group. The individual so endowed would have had many advantages: capacities for complex thought, planning, interpretation, and so on. The capacity would be partially transmitted to offspring, and because of the selective advantages it confers, it might come to dominate a small breeding group, though as with all such novel mutations, there is an issue about how an initially small number of copies of such an allele might survive, despite a large selective advantage’’ (Berwick and Chomsky, 2011, p. 13). 44 B. Laks / Language Sciences 39 (2013) 31–53 including usage-based linguistics. The FLN and the rationalist paradigm are no longer needed as such. Moreover, as we have already mentioned supra, if we are on the side of usage and datum, heterogeneity and variation are features which find themselves at the heart of the disposition. Even better, it is the existence of social heterogeneity and variation which requires a communicational control method of the latter. Ergo, as Whitney and subsequently would Saussure would state, language is therefore indeed a social institution. Such a conclusion must at all costs be refuted. To this end, Hauser et al. (2002), who have perfectly grasped what is at stake, endeavour to reject the arguments in favour of the communicational functionalist hypothesis by contesting the fact that the increasing needs for organisation, cooperation and exchange within the group can motivate exaptation of recursion in MERGE. However, all the arguments in favour of a break in continuity having led to the catastrophic emergence of core syntax, remain extremely debatable and even the hopes placed in the discovery of the ‘‘language gene’’ (FOXP2) have been disappointed.71 Ultimately, as Fodor underlines (2001), the Chomskyan argument is of a strictly epistemological nature. Criticising those who believed they could open a Darwinian path in Generative Grammar precisely by referring to a functionalist reading of human communication and its adaptive advantages (Cf. for instance Pinker, 1997; Pinker and Bloom, 1989), Fodor reaffirms that in a correctly understood rationalist framework, this option cannot be defended and that it is ultimately by referring to Plato that Chomsky argues and defends the innate character of the language faculty.72 Thus language, fundamentally, does not have anything to do with interpersonal communication and has everything to do with the nativist theory of knowledge, at least as regards its key mechanism. In the biolinguistic approach, the universality of MERGE, just like the impossibility of deriving it from the communication function by way of a gradual Darwinian evolution are axiomatic stipulations. As we have just seen, if the principles of core syntax could stem from interpersonal communication, the entire biolinguistic construction would be ruined. The same applies to its universal character. It is nevertheless to be noted that if the character of axioms postulated a priori cannot be proved for both these theses, each of them can be empirically contested. In practice, many approaches have posited the derivation of the cognitive principles of core syntax from cognitive functions which are not specific to the linguistic domain and are active in other domains of human intelligence or in interpersonal communication relations.73 Against these functionalist and communicational hypotheses, the position which we have just formulated testifies to the radicalism of the Chomskyan answer. This answer appears even more categorical, if that is possible, when the universality of the recursion of languages is called into question. The controversy over the non-recursion of Pirahã is a good example. Everett (2005) cast doubt on the universality of recursion on the basis of his analysis of the Amerindian language, of which he is a specialist. He drew from his analysis certain general consequences about the relations between language and culture, which invalidated the biolinguistic position.74 The invalidating character of such criticism did not escape the attention of commentators.75 Nevins et al. (2009)therefore endeavoured to prove that the syntactic analysis of the phenomena isolated by Everett did not in any way challenge Hauser, Chomsky and Fitch’s analysis (2002) of the relation between language and culture, providing one postulates a sufficiently abstract and general characterisation of MERGE.76 In his detailed answer, Everett (2009, p. 439) rightly underlines that this once again amounts to a restriction of the falsifiable character of the principles which are postulated for universal grammar insofar as ‘‘this [more abstract] version of MERGE can neither be supported nor criticized by facts because it is definitional and therefore not falsifiable’’. We can therefore conclude that for Biolinguistics and Chomskyan rationalism, the property which is indefectibly specific to language, that of the principles of core syntax, their absence of correlation to any social, 71 ‘‘If so, then the entire FOXP2 story, and motor externalization generally, is even further removed from the picture of core syntax/semantics evolution. [. . .] Summarizing, FOXP2 does not speak to the question of the core faculty of human language because it really has nothing to do with the core language phenotype, Merge and syntax. [. . .] To be sure, FOXP2 remains a possibly necessary component of the ‘language system’ [. . .] But it is not human language tout court. If all this is so, then the explanation ‘for’ the core language phenotype may be even more indirect and difficult than Richard Lewontin (1998) has sketched’’ (Berwick and Chomsky, 2011, pp. 12–13). 72 ‘‘Chomsky’s ideas about innateness would have been intelligible to Plato; and they would have been intelligible in much the terms of the present debate. This is because Chomsky’s nativism is primarily a thesis about knowledge and belief; it aligns problems in the theory of language with those in the theory of knowledge. Indeed, as often as not, the vocabulary in which Chomsky frames linguistic issues is explicitly epistemological. [. . .] much of the knowledge that linguistic competence depends on is available to the child a priori (i.e., prior to learning). [. . .]; it is, to repeat, primarily epistemological nativism that Chomsky shares with the rationalists. When Plato asks what the slave boy knows about geometry, and where on earth he could have learned it, it really is much the same question that Chomsky asks about what speaker/hearers know about their language and where on earth they could have learned that’’ (Fodor, 2001, p. 10). 73 On the issue of the origin of languages Cf. Laks et al. (2007). Cf. also Bickerton (1998), Bickerton and Calvin (2000), Dessalles (2000), Dessalles et al. (2006), and Pinker (1997, 1999) for different approaches. 74 ‘‘[My] article also offers more detailed argumentation for the hypothesis that culture can exert an architectonic effect on grammar. It concludes that Pirahã falsifies the single prediction made by Hauser et al. (2002) that recursion is the essential property of human language’’ (Everett, 2009, p. 455). 75 In their defence of Chomskyan biolinguistics, Nevins et al. (2009, p. 671), thus wrote: ‘‘The New Scientist (March 18, 2006) suggested, for example, that Pirahã might constitute ‘the final nail in the coffin for Noam Chomsky’s hugely influential theory of universal grammar’; and the Chicago Tribune (June 10, 2007), under the headline ‘Shaking language to the core’, reported that Everett had ‘fired a volley straight at the theory when he reported that the Brazilian tribe he was studying didn’t use recursives [sic]’. More recently, the Times of London (October 24, 2008) has characterized Everett’s claim that ‘Pirahã lack the grammatical principle of recursion’ as an ‘astonishing find’. If the conclusions in NP&R are correct, of course, Pirahã presents us with no nail, no coffin, no volley, and no astonishing find.’’ 76 ‘‘As NP&R (no. 11) pointed out, no construction in a given language (be it English or Pirahã) constitutes a demonstration of recursion or its absence independent of the analysis that this construction receives in the context of a particular theory. Hauser et al. (2002) presupposed, rightly or wrongly, an approach to syntactic structure in which all phrase structure—not just clausal embedding or possessor recursion—serves as a demonstration of recursion. We had this in mind when we noted in NP&R that if Pirahã really were a language whose fundamental rule is a non recursive variant of Merge, no sentence in Pirahã could contain more than two words’’ Nevins et al. (2009, p. 679). I underline this. B. Laks / Language Sciences 39 (2013) 31–53 45 cultural or informational dimensions and their nativism thus represent the non-questionable beginnings of an approach to human nature which is fundamentally essentialist, an approach which supposes a progressive discontinuity and is, from this perspective, close to metaphysical creationism. 4.2. On the non-acquisition of languages It has not been sufficiently noted that since the Cartesian turn of the mid-1960s, Chomskyan linguistics was presented as a theory of the non-acquisition of languages. Such a thesis was further reinforced with the Principles and Parameters approach and the Minimalist Programme (Chomsky, 1995). Acquisition is here reduced to a parametrical setting of the principles of universal grammar (UG). The setting is validated via a language acquisition device (LAD) which confronts UG with the obvious data through which a particular grammar (PG) can be created. Acquisition is thus modelled according to the renowned formula: ðUG  LADÞðdataÞ ¼ PG In this formula, two of the three factors are considered to be innate and are thus given, right from the start. It follows that language acquisition, so to speak, does not correspond to anything other than a local specification of mechanisms that are genetically present in the speaker’s mind. Such a theory, typical of Cartesian rationalism, which considers that logic and universal grammar principles are immanent (Arnauld and Lancelot, 1660), is that which Plato used when demonstrating to Meno that his slave already had a perfectly and innate knowledge of the geometry of the square.77 In the corpus of Generative Grammar, this nativist theory is however never presented as a philosophical positioning but as the conclusion of an empirical reasoning. Such reasoning, which actually is exactly like Plato’s, is based on the postulated poverty of the stimulus accessible to the child during language acquisition.78 Given its key importance for the balance of the Chomskyan theory, we might expect it to have been the subject of many studies based on long-term empirical observations and to have been soundly proved. However in the Chomskyan context, it is patent that this has not been the case. The theory of the poverty of the stimulus represents one of the most recurrent items of generative literature. It is nevertheless one of items with the least empirical support, in this regard. This is not the case within the framework of usage-based models. In datum linguistics, many, particularly voluminous corpora have been built specifically to test the hypothesis of the poverty of stimulus. The quantitative and qualitative analyses, with regular longitudinal monitoring of children’s language acquisition are legion. For many different languages, children’s productions, ranging from precocious babbling to consolidated competence, are very well documented79 (Cf. Supra the CHILDES cooperative programme). And yet these analyses lead precisely to a radical questioning of the Chomskyan hypothesis. Newmeyer supra has relayed these findings. The extremely dynamic domain of acquisition psycholinguistics thus offers many empirical refutations of the hypothesis of the impossibility of acquiring such or such syntactic function, on the sole basis of the data available in the environment.80 The linguistic data with which a speaker is confronted on a daily basis are both numerous and resonant of a social, cultural and interactional context which both reinforces and validates them.81 Mehl et al. (2007) thus evaluate the daily production of a speaker at around 16,000 words, with of course a large dispersion around this average. As for native language acquisition, Morgan (1989, p. 352) reckoned that a child acquired its language after having been confronted with around 4,280,000 sentences. Such orders of magnitude immediately suggest a testing of the statistic and probabilistic models in order to account for acquisition. De facto, many analyses based on Bayesian neuro-computational techniques or using other stochastic tools have exposed the weaknesses of the theory of the poverty of stimulus, with reference to a succession of syntactic phenomena and have convincingly invalidated the nativist approach.82 The nativist approach of Generative Grammar is not founded solely on the argument of the poverty of stimulus which, as I have just shown, is a highly questionable one. A second and, this time, formal argument, is mentioned in support of the child’s necessity, when learning a language, to dispose of a rich repertory of a priori linguistic knowledge. This is the Gold theorem (1967), which I have already mentioned. As was already the case with the theory of the poverty of stimulus theory, the Gold theorem is often referred to as a final proof in generative literature, but it is very rarely commented on or analysed in its practical 77 Meno questions Socrates: ‘‘Mais qu’est-ce qui te fait dire que nous n’apprenons pas et que ce que nous appelons le savoir est une réminiscence ?’’. After the geometrylessongiven to the young slave, Socratesconcludes: ‘‘C’est donc que ces opinions se trouvaient déjà en lui. N’est-ce pas vrai? [. . .] S’il ne les a pas acquises dans la vie présente, il faut bien qu’il les ait eues dans un autre temps et qu’il s’en trouvât pourvu d’avance’’ Platon (1849 Trad. Victor Cousin, 80d–86c). 78 ‘‘Likewise, the central problem of language acquisition arises from the poverty of the ‘‘primary linguistic data’’ from which the child effects this construction; and the proposed solution of the problem is that much of the knowledge that linguistic competence depends on is available to the child a priori (i.e., prior to learning) (Fodor, 2001, p. 10). 79 As young Chomsky wrote in his review of Skinner: ‘‘The manner in which factors operate and interact in language acquisition is completely unknown. It is clear that what is necessary in such a case is research, not dogmatic and perfectly arbitrary claims, based on analogies to that small part of the experimental literature in which one happens to be interested.’’ Chomsky (1959, p. 43). For a review of empirical arguments against the thesis of the poverty of the stimulus Cf. for instance Pullum and Scholz (2002) and Sampson (2002). For a defence of such thesis Cf. Berwick et al. (2011) 80 Many syntactic functions, coordination, subordination and subject inversion, etc. are analysed in this perspective Cf. Parisse (2005) for a presentation. 81 What Mufewene (2001) rightly called language ecology. 82 Cf. for instance Elman and Lewis (2001) for a neuro-computational approach, Perfors et al. (2006) for a Bayesian analysis of the position of the auxiliary in interrogative clauses, Foraker et al. (2009) for a Bayesian analysis of anaphoras, Reali and Christiansen (2005) for a static analysis of position of the auxiliaries in polar interrogative clauses. 46 B. Laks / Language Sciences 39 (2013) 31–53 implications. As Johnson pointed out (2004), Gold’s concern was far removed from any agenda in linguistic research and is even further away from any hypothesis on human learning processes. The Gold theorem is a strictly mathematical demonstration within the framework of the general theory of formal languages. Gold does not address any issues related to language acquisition or to grammar selection by a speaker. As its title indicates, his theorem deals exclusively with the identification of a formal grammar, among all those belonging to a class capable of generating the same set of symbol sequences. Having meticulously criticised each of the interpretations of the Gold theorem and after proving their largely interpretative character, Johnson concludes that the theorem is devoid of any relevance to the cognitive debate.83 It follows that the Gold theorem does not bring any arguments to the theory of the poverty of stimulus theory. It thus leaves the Innatist theory for what it is: a purely epistemological construction that is closely linked to the rationalist orientation of exemplum linguistics. But if the Gold theorem does not enable generative theory to found the non-learnability of human language grammars, it remains that given a corpus of linguistic occurrences, there is no single solution to the challenge of constructing a formal model. In the framework of neo-empiricism he defends, Goldsmith (in press) refers to Marcken’s research (1996) on unsupervised automatic learning. By clearly distinguishing between the field of cognitive argumentation and that of the formal models of language acquisition, he proves that the application of the statistical principle of the Minimum Description Length – MDL (Rissanen, 2007) would enable researchers to bring about the convergence and to optimise a device for the automatic learning of morphology (Goldsmith, 2005a, 2011), without any a priori knowledge. Precisely because he rejects the metaphorical interpretation we have seen in operation in the exploitation of Gold’s theorem, Goldsmith rightly underlines that the cognitive interpretation of MDL and of the results obtained in unsupervised automatic learning remains as yet an undecided issue in linguistic theory. 4.3. Usage-based models and construction grammars: an alternative cognitive model As I have already said, with usage-based grammars, the cognitive issue can be addressed within a new framework. The questions on variation and heterogeneity we have formulated here find their proper place. For the last 30 years, the field of cognitive anthropology has developed spectacularly. The collaboration of neuropsychologists, sociologists, anthropologists, ethnologists, linguists and psychologists has brought about decisive advances in the understanding of the genesis and evolution of symbolic, cultural and social systems.84 The comparative analysis of these systems, such as they exist in the animal world and in mankind, including the analysis of their phylogenetic development and their development in human groups, has enabled research to draw firm conclusions in a field that is mainly dominated by what Changeux (1983) called a neuronal Neo-Darwinism. The questions that are posed by the functioning of linguistic, cultural and social systems and by their intergenerational reproducibility had already been tackled in an evolutionist and Neodarwinian perspective by Dawkins (1976), who suggested they be treated metaphorically as alleles of a new specific gene which appeared with the modern human species. These debates and suggestions are of course of the utmost importance in a research perspective such as that of Biolinguistics. Most of them have however been ignored by those operating within the generativist approach. The field of cognitive anthropology is indeed largely dominated by a paradigm with which, as we have just seen, the latter is incompatible. In accordance with its functionalist dimension, this paradigm seeks to discover in the analysis of the functions fulfilled by a device the motivations for its existence. In its interactionist and communicational dimension, it grants a central place to communication and to interpersonal relationships. And yet, as I have already pointed out above, Hauser et al. (2002), Berwick and Chomsky (2011), and Di Sciullo and Boeckx (2011), along with the linguists involved in the biolinguistic current, go to great lengths to refute all functional approaches to language and to deny that communication is in any way involved in the genesis and in the development of the language faculty specific to mankind. Chomsky (2007) is very clear on this point: the recursion function specific to the FLN is first of all totally endogenous. It generates an internal language of thought which only becomes externalised as a basis for interpersonal communication in a totally secondary and accessory way.85 Functional and communicational approaches are therefore rejected ex definitio. In sum if, as it is generally the case in cognitive anthropology, the group’s gregarious organisation function and the regulating communication function are granted an initial motivation status, we effectively abandon the rationalist framework in favour of paradigm of datum descriptive linguistics. The variety and heterogeneity of these systems thus find themselves fore grounded in the field of research. From them on, what is required is a methodology of systematic analysis that will be able to give an account 83 ‘‘In fact, as long as the notion of identifiability in the limit from any environment has no obvious psychological interpretation. there is little of psychological interest to be concluded from Gold’s Theorem [. . .] Despite its simplicity, many authors have taken Gold’s Theorem to threaten some fundamental views about the mind, and they have responded with various criticisms. However many of these attacks are misguided for largely formal reasons. But a look at the details shows that Gold’s Theorem is still of questionable direct relevance to cognitive science’’ (Johnson, 2004, p. 587). 84 In the volume extract from an interdisciplinary symposium coordinated by the Fyssen Foundation for anthropology, one can find a recent situation: Levinson et al. (2006). 85 ‘‘Emergence of unbounded Merge in human evolutionary history provides what has been called a ‘‘language of thought,’’ an internal generative system that constructs thoughts of arbitrary richness and complexity, exploiting conceptual resources that are already available or may develop with the availability of structured expressions. If the relation to the interfaces is asymmetric, as seems to be the case, then unbounded Merge provides only a language of thought, and the basis for ancillary processes of externalization. [. . .] The capacity would be transmitted to offspring, coming to dominate a small breeding group. At that stage, there would be an advantage to externalization, so the capacity would be linked as a secondary process to the sensorimotor system for externalization and interaction, including communication’’ (Chomsky, 2007, pp. 22, 23). B. Laks / Language Sciences 39 (2013) 31–53 47 of their extent and variety.86 This explains why usage-based linguistics and psychologists who adopt this framework have been so much involved in the recent emergence of cognitive anthropology. It also explains why this has not been the case for generativist linguists. It is within the framework of a generalised cognitive anthropology that, with the backing-up of the results of usage-based linguistics,87 Tomasello has, over the past 15 years, been devising a linguistic and cognitive model within the general line of evolution, as envisaged by the Neodarwinians. He therefore fits very explicitly within the framework of datum linguistics.88 As many cognitivist anthropologists have underlined, interpersonal verbal communication and the development of very sophisticated culturally and socially regulated forms of behaviour have provided the human species with a decisive selective advantage. However this communication capacity is not particular to human beings. There are numerous prodromes in all social animals, from those farthest removed, such as insects, to those closer to mankind, such as primates and apes. It has been suggested that reciprocal altruism (Trivers, 1971, 2002), as it is illustrated for instance in complex delousing strategies or resource sharing, is one of the motivations of social regulations, whose complexity grows with evolution. Taking into account the degree of anticipatory calculation which it presupposes, coupled with the behaviour adjustments and individual strategies within groups, a certain number of sociobiologists have seen in it the origin of complex social behaviour forms particular to mankind.89 For many anthropologists however, such processes are inadequate to explain the break in continuity that defines the locus of the human. While acknowledging the considerable role which the development of reciprocal altruism has played in the evolution towards mankind, Tomasello bases his analysis on the fundamental continuity break which the emergence in mankind of an intentionality attribution90 factor introduces within the, already highly sophisticated, nexus of social systems and communication regulation modes. Such intentionality attribution radically modifies the functionalities of reciprocal altruism. It is specific to the human species and it is motivated by the reinforcement of the gregarious function.91 From the ontogenetic point of view, Tomasello sees its substrate in the checking, pantomime and mimetic activities of young children and in other activities involving shared deixis, which regulate the common denomination of objects. The reading of intentional states, a new functionality, which develops in children between the age of 9 and 12 months, enables them to build a genuine mind theory, the basics of abstract thought (Tomasello, 2003, p. 3). It is on this basis that communication is analysed as a function which allows a person to manipulate symbolically the intentional and mental states of the people with which he or she is interacting. Language is thus conceived as a ‘‘structured inventory of conventional symbolic units’’ (Langacker, 1987, p. 73) which takes on this function in the manipulation of mental states. The double orientation, communicational and functional, of the modalities is therefore very clear. Their anchorage in social structuration and in the cultural and symbolic organisation of human groups is, from this point of view, fundamental.92 The intentionality attribution factor, along with the reading of intentions, behaviour and events is, in this regard, itself the generator of a form of abstract thought, because it immediately focuses on their practical meaning in the interpersonal relationship. Confronted with the remarkable variability of acts and objects, the human mind thus develops a specific ability to recognise partial similarities. The search for comparable patterns, in terms both of the perceptual and conceptual aspects, leads to what is a spectacular development of the capacity to classify, that is to say to simplify, to extract abstract recurring 86 Positions that Chomsky (2007, p. 1) explicitly attributes to Skinnarian behaviourists and more specifically to (American and European) structuralists mainly concerned about corpora and data recollection methods, on the foreground he quotes Harris and his work Methods in structural linguistics. Langacker does not object to this observation when he writes: ‘‘language has two basic and closely related functions: a semiological function, allowing thoughts to be symbolized by means of sounds, gestures, or writing, as well as an interactive function, embracing communication, expressiveness, manipulation and social communion. A pivotal issue in linguistic theory is whether the functions language serves should be taken as foundational or merely subsidiary to the problem of describing its form. The recognition of their foundational status is the primary feature distinguishing functionalist approaches to language from the formalist tradition (notably generative grammar)’’ (Langacker, 1998, p. 1). 87 In the two volumes he gathered under the sufficiently explicit title ‘‘The new psychology of language: cognitive and functional approaches to linguistic structure’’ Tomasello (1998, 2008c) thus presents the contributions of linguists working in the framework of usage-based models, many of which participated in the debates of generative semantics and of linguistics which emerged from it: Langacker, Talmy, Fillmore, Fauconnier, Givon, but also Croft, Hopper Bybee, Goldberg, Haspelmath, Van Valin, etc., which quite clearly identifies the linguistic current to which he refers. 88 ‘‘In diametric opposition to [generative] methodological assumptions, cognitive-functional linguists take as their object of study all aspects of natural language understanding and use, including unruly idioms, metaphors, and irregularities. They [. . .] take as an important part of their data not disembodied sentences derived from introspection, but rather utterances or other longer sequences from naturally occurring discourse’’ (Tomasello, 2008c, p. XIII). 89 On the issue of ‘‘cheaters’’ emerged an interrogation on the dissymmetry between benefits and obligations. Information on the developments of reciprocal altruism in moral and politic sociology, in the game theory and also in mathematic modelling of behaviours and markets for instance in Clavien’s work (2010). 90 ‘‘Specifically, human cooperation is structured by what some modern philosophers of action call shared intentionality or ‘‘we’’ intentionality In general, shared intentionality is that is necessary for engaging in uniquely human forms of collaborative activity in which a plural subject ‘‘we’’ is involved: joint, intentions, mutual knowledge, shared beliefs-all in the context of various cooperative motives’’ (Tomasello, 2008a, p. 7). 91 ‘‘Specifically, human beings cooperate with one another in species-unique ways involving processes of shared intentionality [. . .] This fundamentally cooperative process makes human communication utterly different from the communicative activities of all other species on the planet’’ (Tomasello, 2008a, pp. 72, 99). 92 ‘‘At some point in human evolution, Homo Sapiens evolved the ability to communicate with another symbolically. [. . .] These transformations of linguistic structure occur as a result of social-interactive processes’’ Tomasello (2008a), ‘‘And what about language? The current hypothesis is that it is only within the context of collaborative activities in which participants share intentions and attention, coordinated by natural forms of gestural communication, that arbitrary linguistic conventions could have come to existence evolutionarily [. . .] this perspective on human communication and language thus basically turns the Chomskian proposal on its head, as the most fundamental aspects of human communication are seen as biological adaptations for cooperation and social interaction in general, whereas the more purely linguistic, including grammatical, dimensions of language are culturally constructed and passed along by individual linguistic communities’’ (Tomasello, 2008a, pp. 9, 11, 163). 48 B. Laks / Language Sciences 39 (2013) 31–53 forms and to gather within equivalence classes objects or acts that are partially different but are equivalent from the functional point of view. This ascending process of abstraction and conceptualisation goes with a very large statistic and probabilistic sensitivity to the recurrence of the same, to the recognition of complex distributional patterns, beyond similarities and differences and over very long time sequences. We can thus observe that far from being abnormal, according to this general pattern, the existence of structured heterogeneity and the intrinsic variation of objects and acts is actually what leads and motivates this ascending work of abstraction, pattern creation and classification. Indeed, if the variable and heterogeneous occurrences need to be interpreted in order to reconstruct their underlying practical intentionality, it becomes necessary to analyse the partial similarities and differences in order to classify them as equivalent, by way of the construction of what is a more or less abstract system of categories. We may remark that this is very precisely the taxonomic thesis of the system which I mentioned above when presenting datum sciences. In this approach, the cognitive functions which are implicated, contrary to Fodor (1983b) and Chomsky’s (1986) theory, are not in any way specific. Sensitivity to partial similarities, pattern extraction, categorisation, routinisation and statistic sensitivity correspond to what are very general functions of human cognition, which find application in the field of verbal interactions but are not specific to the latter.93 With the exception of intentionality attribution and the social and cultural dimensions which it introduces, the field of language is not at all specific. The same higher cognitive functions are applicable to interpersonal relations as elsewhere. Discontinuity in evolution is thus considered here to find its source in the learner’s reading of the mental states of alter egos and in the complete change which is thus introduced into interpersonal relationships, into their practical control, their routinised organisation and the abstract conceptual categorisation, which their surface variability causes to be implemented so as to deal with them from the cognitive point of view. As can be seen, the approach is a patently neoempiricist one, in which communicative practice is first compared with the codes which will eventually regulate and organise it. This is consistent with datum sciences and with usage-based linguistics, which consider the language system as secondary in relation to language practices. Because language is approached as a routinised social and cultural modality, the grammar which constitutes its systemis seen as a sub-product of communicational activity. Grammar is thus no more than the social and cultural product of a progressive grammaticalisation process of speech acts. Contrary to generativist approaches, which consider it to be an aborigine condition of language, in this approach, just as in all cultural and social phenomena, grammar is seen as the historical product of a social activity.94 The same is true for the grammatical categories and functions which, far from pre-existing their instantiation, as in the Cartesian approaches, are simply progressive cognitive constructions, taxons and super taxons, enabling us to gather in a single group various and variable occurrences. Thus, as Langacker (1991) underlines for instance, the ‘‘verb’’ category does not pre-exist its effective instantiations. What constitutes a verb in a given language is nothing other than the set of those elements which regularly behave as such and which are therefore cognitively and memorially gathered under a definite abstract category which has been constructed from the concrete, heterogeneous and variable functionalities manifesting themselves in a discourse. As I have emphasized, it is the perception of functional similarities along with differences in forms, leading to the perception of heterogeneity and variation, which motivates this work of taxonomic abstraction leading to the construction of taxons and super taxons, which grammarians call categories.95 We can therefore understand better why Tomasello (2003) builds his ontogenetic and phylogenetic approach on the basis of the explicit support of usage-based linguistics and we can see that the opposition between datum and exemplum linguistics, Cartesian linguistics and usage-based linguistics, ultimately between rationalism and empiricism, is in evidence at all levels of linguistic and cognitive analysis. The variability and heterogeneity of language facts, whether they are denied or asserted as fundamental, occupy a central place in these theoretical constructions. We can also understand why these approaches are instances of a neostructuralism where the taxonomy and the progressive emergence of an explicative system are built on the analysis of observable data (Laks, 1996). The corpus, providing its scope is sufficiently wide and representative, is a rich one. The taxonomy which emerges directly from it, along with the statistical analysis of its organisation, inform the theory which we can build from it. In this perspective, linguistics is manifestly a data processing method, such as Harris (1951), to take an example, theorised and systematised it. The relation between linguistic analysis and cognitive analysis is not, as is the case in generative grammar, posited as a thetic a priori.96 It is rather 93 One will have recognised the thesis defended by cognitive linguistics and usage-based linguistics (Langacker, 1987, 1991), Fauconnier (1997) and Lakoff and Johnson (1999). Of course, this is not a coincidence. 94 ‘‘If grammatical structures do not come directly from the human genome, as above-reported data suggest they do not, and if children do not invent then de novo, as they clearly cannot, then it is legitimated to ask, Where do grammatical structures come from? The answer is that, in the first instance they come from processes of grammaticalization in language history. [. . .]’’ Even so, grammaticalization by itself is not enough because, it does not account for the abstractness of linguistic structures. [. . ..] children make this contribution in more extended developmental processes in which they apply their general cognitive, socialcognitive, and vocal-auditory processing skills to the historical products of grammaticalization’’ (Tomasello, 2008a, p. 163). 95 We face here a debate which made its way through the entire history of grammar and philosophy: are the understanding categories (e.g. grammatical categories) a priori data or are they build through cognitive experience? This debate was recently revived in the field of mathematics: whereas Connes defends a Platonic position of the a priori character of elements and mathematical laws which the searcher only (re)discovers, Changeux defends that they are mental constructions which are born to the relation between higher cognitive functions and the date from experience and context. Cf. Changeux and Connes (1989). 96 ‘‘We use the term ‘grammar’ with a systematic ambiguity. On the one hand, the term refers to the explicit theory constructed by the linguist and proposed as a description of the speaker’s competence. On the other hand, we use the term to refer to this competence itself [. . .] To summarize, then, we use the term ‘grammar’ to refer both to the system of rules represented in the mind of the speaker-hearer, a system which is normally acquired in early childhood and used in the production and interpretation of utterances, and to the theory that the linguist constructs as a hypothesis concerning the actual internalized grammar of the speaker-hearer’’ (Chomsky and Halle, 1968, pp. 3–4). B. Laks / Language Sciences 39 (2013) 31–53 49 the result of the convergence between what cognitive psycholinguistics constructs independently as a functional and acquisitional pattern, on the one hand, and what structural linguistics proposes as a systematisation of the corpus on the other. As I have previously underlined contra all the vulgates of the Cours de Linguistique Générale (Laks, 2011b), this precisely was Saussure’s position.97 5. Conclusion: Variatio omnibus Variation holds a central place in the thought of Darwin, who thus begins his milestone opus (Darwin, 1859) with a first chapter entirely dedicated to variation, its origin, its effects and its mechanisms. Of course, as Hoquet states (2009), Darwin at the time did not yet have a totally articulated genetic model which could enable him to analyse in detail the mechanism of the random generation of diversity of which he describes the effects. Such decisive progress was to be carried out several years later by Wallace. Mayr (1982) nevertheless emphasises in his history of biological thought how, and in what terms, the three moments which are postulated by Darwin, variation, selection and heritage, remain central for genetics and for the modern theory of evolution. We now know very precisely how this random generator of diversity actually works: all reproduction of living organisms is imperfect, the copying and recombination of the genome introducing random differences at each generation. With sexual reproduction, these imperfections are even more important, insofar as two genomes are recombined. Evolution depends directly on this variability by selecting certain characters that are randomly produced by the generator. Those which confer upon the individual and its kin a decisive adaptive advantage are thus selected and reproduced. We thus come to understand that during evolution, diversity and variability increase, the human species itself being marked by a large-scale variability, in every possible respect. The reason why I insisted on recalling the central place, which variability occupies in the modern theory of evolution, was to underline that, contrary to how Chomskyan rationalism considers it, it is not in any way an abnormal or disruptive modality. On the contrary, variation appears in the realm of the living as a fundamentally dynamic and structuring modality. Concerning language and cognition, the logic is no different. Recently, Labov (2001, pp. 3–33) recalled that regarding linguistic change, of the 16 criteria on which Darwin himself had established his parallel between the evolution of species and the evolution of languages, 15 were appropriate within the terms of the current state of variationist research. The 16th, which is about the adaptive advantage and the improvement of communicational yield which linguistic change ought to bring, constitutes what Labov calls the Darwinian paradox. Indeed regarding linguistic change, one cannot put forward a selective process which would select a form, a structure or a linguistic function on the basis of the improvement it would produce. Therefore, the main concept of Darwinism, that of the random generator of diversity, coupled with the tension between the uniqueness of a species, notably as measured by the faculty to combine two genomes in a non-sterile line of descendants (i.e. the species barrier) and the heterogeneity of the phenotypes, can be applied to the cultural and social systems which languages are: an extreme variability of forms and occurrences, nevertheless limited by the necessities of inter-comprehension and the marking of one’s belonging to a same speech community. However the importance of the concept of variation and structured heterogeneity is not limited to the phenomena of linguistic change. What I have reaffirmed here is that its impact is much larger and that it affects the entire field of linguistics, to say nothing of the cognitive sciences. If exemplum linguistics is constantly eliminating all heterogeneity and all variation, datum linguistics, in contrast, usage-based models and cultural cognition theory are fundamentally based on its structuring potentialities and on the processes which it introduces. As we have seen, it is the variability of forms in usage which constitutes the driving force and the motivation of their increasingly abstract taxonomic organisation and the progressive crystallisation of functional processing categories. This variability does not affect only inter-individual differences. Because it is one of the fundamental dimensions of all usage, and because grammar is built on usage, such variability and internal heterogeneity also affects the cognitive and practical modalities of the linguistic disposition of all speakers, both synchronously and diachronously. In this approach, the linguistic competence of a situated speaker is neither stable nor homogeneous. Insofar as it is a practical competence, socially constituted and socially used, it is indeed, both in terms of its individual and its social aspects, a historical and cultural product. What guarantees its relative unity and what contains it within the limits of given variability, is precisely what constitutes the historical and social unity of human communities, the unequal but restraining sharing of norms, rules and routines, in a word, the sharing of a same culture. It also resolves, incidentally, a problem which was never actually addressed within the Cartesian approach: the supposedly strict identity of the grammars of all speakers of a language, a position which leads to the famous issue of the convergence of acquisition procedures, and which indirectly motivates the use, metaphorical as we have seen, of the Gold theorem. In usage-based linguistics, nothing imposes such a convergence of acquisition or such a strict identity of mental grammars. Interpersonal communication in a real community, interwoven with structural variation and heterogeneity, does not posit the strict identity of competences. On the contrary, as we can see for cultural and social procedures, it is the sharing of the same norms along with common procedures for their evaluation which, even if social individuals were to have available to 97 ‘‘La langue ne peut pas procéder comme le grammairien, elle a un autre point de vue et les mêmes éléments ne lui sont pas donnés, elle fait ce qui par le grammairien est considéré comme des erreurs mais qui n’en sont pas, car il n’y a de sanctionné par la langue que ce qui est immédiatement reconnu par elle. [. . .] Entre l’analyse subjective des sujets parlants eux-mêmes (qui seule importe!) et l’analyse objective des grammairiens, il n’y a donc aucune correspondance, quoiqu’elles soient fondées toutes deux en définitive sur la même méthode (confrontation de séries)’’ Saussure (1916 édition Engler 1968, p. 2759). 50 B. Laks / Language Sciences 39 (2013) 31–53 them a non-egalitarian and differentiated usage of the latter, ensures the coherence of the social fabric and which constrains its internal heterogeneity at the same time as it limits the variation of practices and usages. While it does not postulate a complete convergence and a perfect identity of stabilised states at adulthood, the in situ acquisition of communicational, linguistic, cultural and social competences does not stipulate a homogeneity of the data on which it feeds either. We know that statistical modelling used in the various automatic, symbolic and sub-symbolic or connectionist systems cannot converge if the data are too regular and homogeneous. A certain amount of noise or uncertainty is always necessary in the systems. Sometimes, their presence needs to be guaranteed through their explicit introduction, by way of form of disequilibrium. As I pointed out, following Tomasello (2008a), human cognition is extremely sensitive to regularities and differences, to their recurrence and to their organisation in time; in short, the human intelligence of phenomena and acts largely belongs to the statistico-probabilistic type. This is why, on the contrary, far from representing a break or a breach, the existence of a constrained, limited and structured variability represents a formal advantage and a decisive factor of cognitive facilitation. In addressing the arguments on which the idealisation of the speaker–listener and the a priori homogenisation of speech communities can be grounded, Chomsky (1980, pp. 27–28), for want of any empirical argument or any factual observation, writes: ‘‘So we are left with what must be the crucial question: does the idealization so falsify the real world that it will lead to no significant insight into the nature of the language faculty, or does it, on the contrary, open the possibility for discovering fundamental properties of the language faculty? In short, is the idealization legitimate? Suppose that someone takes the negative stand on this question. Such a person is committed to one of the following two beliefs: 1. People are so constituted that they would be incapable of learning language in a homogeneous speech community; variability or inconsistency of presented evidence is a necessary condition for language learning. 2. Though people could learn language in a homogeneous speech community, the properties of the mind that make the achievement possible do not enter into normal language acquisition in the real world of diversity, conflict of dialect, etc. I cannot believe that anyone who thinks the matter through would really maintain either of these beliefs. In fact, each seems hopelessly implausible. Suppose, then, that we reject them. Thus we accept that humans have some property of mind which would enable them to learn the language of a homogenous speech community were they to be placed in one.’’ The panorama of contemporary linguistic research, which I have outlined in this paper, based on the epistemological distinction between exemplum linguistics and datum linguistics, leads us to question, one point after the other, this reductio ad absurdum. On the contrary, when one takes seriously the enormous mass of usage data and the functioning of real communities, structured variation appears as a motivating and a driving force in the processes of acquisition and as an organiser and regulator of interpersonal communications. In a functionalist perspective, we are thus currently returning, 50 years later, to what was Weinreich’s great intuition: ‘‘The solution, we will argue, lies in the direction of breaking down the identification of structuredness with homogeneity. The key to a rational conception of linguistic change—indeed, of language itself—is the possibility of describing orderly differentiation in a language serving a community. We will argue that native like command of heterogeneous structures is not a matter of multidialectalism or ‘‘mere’’ performance, but is part of unilingual linguistic competence. One of the corollaries of our approach is that in a language serving a complex (i.e., real) community, it is absence of structured heterogeneity that would be dysfunctional’’. Weinreich et al. (1968, p. 96) References Aarts, B., 2000. Corpus linguistics, Chomsky and fuzzy tree fragments. In: Mair, C., Hundt, M. (Eds.), Corpus Linguistics and Linguistic Theory. Rodopi, Amsterdam/Atlanta, pp. 5–13. Antilla, A., 2007. Variation and optionality. In: Lacy, d.P. (Ed.), The Cambridge Handbook of Phonology. Cambridge University Press, Cambridge, pp. 519–536. Antilla, A., Cho, Y.-m.Y., 1998. Variation and change in optimality theory. Lingua 104, 31–56. Arnauld, A., Lancelot, C., 1660. Grammaire générale et raisonnée, contenant les fondements de l’art de parler. Républications Paulet 1969, Paris. Bachelard, G., 1938. La formation de l’esprit scientifique; contribution à une psychanalyse de la connaissance objective. J. Vrin, Paris. Baratin, M., Desbordes, F., Hoffman, P., Pierrot, A., 1981. L’analyse linguistique dans l’Antiquité classique. I, Les théories. Éditions Klincksieck, Paris. Barlow, M., Kemmer, S., 2000. Usage Based Models of Language. CSLI, Stanford Cal. Béguelin, M.-J., 1990. Conscience du sujet parlant et savoir du linguiste. In: Liver, R., Werlen, I., Wunderli, P. (Eds.), Sprachtheorie und Theorie der Sprachwissenschaft. Festschrift für Rudolf Engler, Gunter Narr, Tübingen, pp. 208–220. Benveniste, É., 1966. Problèmes de linguistique générale. Gallimard, Paris. Berwick, R.C., 2011. Syntax facit saltum redux: biolinguistics and the leap to syntax. In: Di Sciullo, A.M., Boeckx, C. (Eds.), The Biolinguistic Enterprise. New Perspectives on the Evolution and Nature of Human Language Faculty. Oxford University Press, Oxford. Berwick, R.C., Chomsky, N., 2008. Poverty of the Stimulus’ Revisited: Recent Challenges Reconsidered. CogScI, Washington. Berwick, R.C., Chomsky, N., 2011. The biolinguistic program: the current state of its development. In: Di SCiullo, A.M., Boeckx, C. (Eds.), The Biolinguistic Enterprise. Oxford University Press, Oxford, pp. 18–42. Berwick, R.C., Pietroski, P., Yankama, B., Chomsky, N., 2011. Poverty of the stimulus revisited. Cognitive Science 35, 1207–1242. Bickerton, D., 1998. Catastrophic evolution: the case for a single step from protolanguage to full human language. In: Hurford, J.R., Studdert-Kennedy, M., Knight, C. (Eds.), Approaches to the Evolution of Language. Cambridge University Press, Cambridge, pp. 341–358. Bickerton, D., Calvin, W.H., 2000. Lingua ex Machina: Darwin and Chomsky Reconciled. MIT Press, Cambridge Mass. Bloomfield, L., 1933. Language. H. Holt and Company, New York. Boeckx, C., Grohmann, K.K., 2007. The biolinguistics manifesto. Biolinguistics Open Journal Systems, 1–8. Boersma, P., 1998. Functional Phonology: Formalizing the Interactions between Articulatory and Perceptual Drives. LOT, Amsterdam. Bouquet, S., 1997. Introduction à la lecture de Saussure. Payot, Paris. Bourdieu, P., 1980. Le sens pratique. Éditions de Minuit, Paris. Bourdieu, P., 1982. Ce que parler veut dire: L’économie des échanges linguistiques. Fayard, Paris. Bourdieu, P., 1994. Raisons pratiques: sur la théorie de l’action. Seuil, Paris. B. Laks / Language Sciences 39 (2013) 31–53 51 Bourdieu, P., 1997. Méditations pascaliennes. Seuil, Paris. Bourdieu, P. (Ed.), 1993. La Misère du monde. Editions du Seuil, Paris. Bourdieu, P., Rey, A., Milner, J.C., Delesalle, S., Encrevé, P., Fauconnier, G., 1977. Table ronde « linguistique et sociologie du langage ». Langue Française 34, 35– 51. Buffon, G.-L.L.d., 1749. Histoire naturelle, générale et particulière, avec la description du Cabinet du Roy., Texte établi, introduit et annoté par Stéphane Schmitt avec la collaboration de Cédric Crémière. ed. Éditions Honoré Champion, 2007–2009, Paris. Bybee, J., 2001. Phonology and Language Use. Cambridge University Press, Cambridge. Bybee, J., 2006. Frequency of Use and the Organization of Language. Oxford University Press, Oxford. Chambers, J.K., Trudgill, P., 1980. Dialectology. Cambridge University Press, Cambridge [Eng.], New York. Changeux, J.-P., 1983. L’homme neuronal. Fayard, Paris. Changeux, J.-P., Connes, A., 1989. Matière à penser. Odile Jacob, Paris. Chater, N., Manning, C.D., 2006. Probabilistic models of language processing and acquisition. Trends in Cognitive Sciences 10, 335–344. Chevalier, J.-C., 2007. Les exemples et la norme dans les grammaires: étude historique. In: Siouffi, G., Steuckardt, A. (Eds.), Les linguistes et la norme. Peter Lang, Berlin, pp. 151–163. Chomsky, N., 1957. Syntactic Structures. Mouton, La Haye. Chomsky, N., 1959. A review of B.F. Skinner’s verbal behavior. Language 35, 26–58. Chomsky, N., 1965. Aspects of the Theory of Syntax. M.I.T. Press, Cambridge, Mass. Chomsky, N., 1966. Cartesian Linguistics: A Chapter in the History of Rationalist Thought. Harper & Row, New York. Chomsky, N., 1968. Language and Mind. Harcourt Brace & World, New York. Chomsky, N., 1977. Langue, linguistique, politique: dialogues avec Mitsou Ronat. Flammarion, Paris. Chomsky, N., 1980. Rules and Representations. Columbia University Press. Chomsky, N., 1986. Knowledge of Language: Its Nature, Origin, and Use. Praeger, New York. Chomsky, N., 1995. The Minimalist Program. The MIT Press, Cambridge, Mass. Chomsky, N., 2004. Biolinguistics and the Human Capacity (Talk delivered at MTA, Budapest, May 17, 2004). <http://www.chomsky.info/talks/ 20040517.htm>. Chomsky, N., 2007. Of minds and language, biolinguistics. Open Journal Systems, 9–27. Chomsky, N., 2011. Some simple evo-devo theses: how true might they be for language? In: Larson, R.K., Déprez, V., Yamakido, H. (Eds.), The Evolution of Human Language, Biolinguistic Perspectives. Cambridge University Press, Cambridge, pp. 43–63. Chomsky, N., Halle, M., 1968. Sound Pattern of English. Harper and Row, New York. Clavien, C., 2010. Je t’aide, moi non plus: biologique, comportemental ou psychologique, l’altruisme dans tous ses états. Vuibert, Paris. Currie, H.C., 1952. A projection of sociolinguistics: the relationship of speech to social status. Southern Speech Journal 17, 28–37. Currie, H.C., 1981. Sociolinguistics and American linguistic theory. International Journal of the Sociology of Language 31, 29–41. Darwin, C., 1859. L’origine des espèces au moyen de la sélection naturelle ou la préservation des races favorisées dans la lutte pour la vie. Flammarion, Traduction Edmond Barbier, Nouvelle édition Daniel Becquemont, 1992, Paris. Dawkins, R., 1976. The Selfish Gene. Oxford University Press, New York; Oxford. Dessalles, J.-L., 2000. Aux origines du langage: une histoire naturelle de la parole. Hermès, Paris. Dessalles, J.-L., Picq, P., Victorri, B., 2006. Les origines du langage. Le Pommier, Paris. Di Sciullo, A.M., Boeckx, C., 2011. The Biolinguistic Enterprise. New Perspectives on the Evolution and Nature of Human Language Faculty. Oxford University Press, Oxford. Durand, J., Laks, B., Calderone, B., Tchobanov, A., 2011. Que savons nous de la liaison aujourd’hui? Langue Française 169, 103–135. Durkheim, É., 1927. Les règles de la méthode sociologique. F. Alcan, Paris. Elman, J.L., Lewis, J.D., 2001. Learnability and the statistical structure of language: poverty of stimulus arguments revisited. In: Skarabela, B., Fish, S., Do, A.H.-J. (Eds.), 26th Annual Boston University Conference on Language Development. Cascadilla Press, Boston, pp. 359–370. Encrevé, P., 1976. Présentation In Labov William 1976: Sociolinguistique. Editions de Minuit, Paris. Encrevé, P., 1982. A propos du marché linguistique. In: Dittmar, N., Schlieben-Lange, B. (Eds.), Die Soziolinguistik in romanischsprachigen Ländern. Narr, Tübingen, pp. 97–104. Encrevé, P., 1986. Variation et structure, études de phonologie et de pragmatique sociolinguistiques. Thesis Paris VIII, Paris, 2 Volumes. Encrevé, P., 2000. The old and the new: some remarks on phonology and its history. Folia Linguistica XXXIV, 56–84. Everett, D., 2005. Cultural constraints on grammar and cognition in Piraha: another look at the design features of human language. Current Anthropology 46, 621–646. Everett, D., 2009. Piraha culture and grammar: A response to some criticism. Language 85, 405–442. Fauconnier, G., 1997. Mappings in Thought and Language. Cambridge University Press, Cambridge. Feldman, J.A., 2006. From Molecule to Metaphor. A Neural Theory of Language. MIT Press, Cambridge Mass. Fodor, J., 1983a. The Modularity of Mind. MIT Press, Cambridge MA. Fodor, J., 1983b. The Modularity of Mind: An Essay on Faculty Psychology. MIT Press, Cambridge Mass. Fodor, J., 2001. The Mind Doesn’t Work that Way: The Scope and Limits of Computational Psychology. MIT Press, Cambridge Mass. Foraker, S., Regier, T., Khetarpal, N., Perfors, A., Tenenbaumb, J., 2009. Indirect evidence and the poverty of the stimulus: the case of anaphoric one. Cognitive Science 33, 287–300. Foucault, M., 1967. La Grammaire générale de Port-Royal. Langages 7, 7–15. Fukuyama, F., 1992. La fin de l’histoire et le dernier homme. Flammarion, Paris. Glatigny, M., 1982. La notion de règle dans la « grammaire » de Meigret. Histoire Epistémologie. Langage 4, 93–106. Gleason, J.B., Thompson, R.B., 2002. Out of the baby book and into the computer: child language research comes of age. APA Review of books 47, 390–394. Gold, M.E., 1967. Language identification in the limit. Information and Control 16, 447–474. Goldberg, A.E., 1995. Constructions: A Construction Grammar Approach to Argument Structure. University of Chicago Press, Chicago. Goldberg, A.E., 2006. Constructions at Work: The Nature of Generalization in Language. Oxford University Press, Oxford. Goldsmith, J., 2005a. An algorithm for the unsupervised learning of morphology. Natural Language Engineering 12, 1–19. Goldsmith, J., 2005b. Review: the legacy of Zellig Harris: language and information into the 21st century. Language 81, 719–736. Goldsmith, J., in press. Towards a new empiricism for linguistics. <http://hum.uchicago.edu/~jagoldsm//Papers/empiricism1.pdf>. Goldsmith, J., 2011. The Linguistica Project. <http://linguistica.uchicago.edu/linguistica.html>. Goldsmith, J., Aris, X., 2009. Learning phonological categories. Language 85, 4–39. Goldsmith, J., Huck, G., 1995. Ideology and Linguistic Theory: Noam Chomsky and the Deep Structure Debates. Routledge, New York. Gould, S.J., Vrba, E., 1982. Exaptation – a missing term in the science of form. Paleobiology 8, 4–15. Grammatica, G.A., 2005. Présentation du De Adverbio de Priscien. Histoire Epistémologie Langage 27, 7–28. Grohmann, K.K., 2007. An interview with Henk van Riemsdijk biolinguistics. Open Journal Systems, 137–149. Gumperz, J.J., 1982. Discourse Strategies. Cambridge University Press, Cambridge. Harris, R., 1993. Linguistic Wars. Oxford University Press, Oxford. Harris, Z., 1951. Methods in Structural Linguistics. University of Chicago Press, Chicago. Hauser, M.D., Chomsky, N., Fitch, W.T., 2002. The faculty of language: what is it, who has it, and how did it evolve? Science 29822, 1569–1580. Hayes, B., Cziráky Londe, Z., 2006. Stochastic phonological knowledge: the case of Hungarian vowel harmony. Phonology 23, 59–104. 52 B. Laks / Language Sciences 39 (2013) 31–53 Hayes, B., Kirchner, R., Steriade, D., 2004. Phonetically-Based Phonology. Cambridge University Press, Cambridge. Hockett, C.F., 1942. A system of descriptive phonology. Language 18, 3–21. Holtz, L., 1981. Donat et la tradition de l’enseignement grammatical: étude sur l’Ars Donati et sa diffusion (IVe-IXe siècle) et édition critique. Centre national de la recherche scientifique, Paris. Hoquet, T., 2009. Darwin contre Darwin: comment lire ‘‘L’origine des espèces’’? Éd. du Seuil. Horace, 457/1944. L’Art poétique ou Épître aux Pisons, Paris, Garnier, 1944. Hymes, D.H., 1972. Models of the interaction of language and social life. In: Gumperz, J.J., Hymes, D.H. (Eds.), Directions in Sociolinguistics. Holt, Rinehart and Winston, New York, pp. 35–71. Johnson, K., 2004. Gold’s theorem and cognitive science. Philosophy of Science 71, 571–592. Keller, M., 2009. Exemples et citations chez Priscien: examen de deux passages du livre xv « De aduerbio » des Institutiones grammaticae (GL III, 70, 4–71, 6 et 77, 7–12). In: Bortolussi, B., Keller, M., Minon, S., Sznadjer, L. (Eds.), Traduire, transposer, transmettre: dans l’Antiquité gréco-romaine. Picard, Paris. Koerner, K., 1991. Toward a history of modern sociolinguistics. American Speech 66, 57–70. Labov, W., 1966. The Social Stratification of English in New York City. Center for Applied Linguistics, Washington. Labov, W., 1972. Sociolinguistic Patterns. University of Pennsylvania Press, Philadelphia. Labov, W., 1975. What is a Linguistic Fact? Peter de Ridder Press, Lisse. Labov, W., 1976. Sociolinguistique. Editions de Minuit, Paris. Labov, W., 1979. Le parler ordinaire. Editions de Minuit, Paris. Labov, W., 1981. Resolving the Neogrammarian controversy. Language 57, 267–309. Labov, W., 1987. Some Observations on the Foundation of Linguistics. <http://www.ling.upenn.edu/~wlabov/Papers/Foundations.html>. Labov, W., 1994. Principles of Linguistic Change: Internal Factors. Blackwell, Oxford. Labov, W., 1996. When intuitions fail. In: McNair, L., Singer, K., Dolbrin, L., Aucon, M. (Eds.), Papers from the Parasession on Theory and Data in Linguistics. Chicago Linguistic Society, Chicago, pp. 77–106. Labov, W., 2001. Principles of Linguistic Change: Social Factors. Blackwell, Oxford. Labov, W., 2004. Quantitative analysis of linguistic variation. In: Ammon, U., Dittmar, N., Mattheier, K.J., Trudgill, P. (Eds.), Sociolinguistics. An International Handbook of the Science of Language and Society. La Haye, Berlin, pp. 6–22. Lakoff, G., 1973a. Deep Language. Review of Books, New-York. Lakoff, G., 1973b. Interview with Herman Parret. In: Parret, H. (Ed.), Discussing Language. Mouton, The Hague, pp. 151–178. Lakoff, G., Johnson, M.H., 1999. Philosophy in the Flesh: The Embodied Mind and Its Challenge to Western Thought. Basic Books, New-York. Laks, B., 1996. Langage et cognition: l’approche connexionniste. Hermès, Paris. Laks, B., 2008. Pour une phonologie de corpus. Journal of French Language Studies 18, 3–32. Laks, B., 2011a. La phonologie du français et les corpus. Langue Française 169, 3–17. Laks, B., 2011b. La phonotactique saussurienne: Système et loi de la valeur s Langages 185, 91–109. Laks, B., Cleuziou, S., Demoule, J.-P., Encrevé, P., 2007. Origin and Evolution of Languages: Approaches, Models, Paradigms. Equinox, London. Lamarck, J.B.P.A.d.M.d., 1809. Philosophie zoologique, ou, Exposition des considérations relative à l’histoire naturelle des animaux. Dentu, Paris. Langacker, R., 1987. Foundations of Cognitive Grammar I: Theoretical Prerequisite. Stanford University Press, Stanford. Langacker, R., 1991. Foundations of Cognitive Grammar II: Descriptive Applications. Stanford University Press, Stanford. Langacker, R., 1998. Conceptualization, symbolisation and grammar. In: Tomasello, M. (Ed.), The New Psychology of Language: Cognitive and Functional Approaches to Linguistic Structure. Lawrence Erlbaum, Mahwah, NJ, pp. 1–39. Langacker, R., 1988. A usage-based model. In: Rudzka-Ostyn, B. (Ed.), Topics in Cognitive, Linguistics, pp. 127–165. Langacker, R., 2000. A dynamic usage-based model. In: Barlow, M., Kemmer, S. (Eds.), Usage Based Models of Language. CSLI, Standford University, Standford, pp. 1–65. Lehmann, W.P., 1962. Historical Linguistics: An Introduction. Holt, Rinehart and Winston, New York. Levinson, S.C., Jaisson, P., Fyssen, F., 2006. Evolution and Culture: A Fyssen Foundation Symposium. MIT Press, Cambridge, Mass. Linné, C.v., 1735. Systema naturae, per regna tria naturae: secundum classes, ordines, genera, species cum characteribus, differentiis, synonymis, locis. Bibliothèque Nationale de France, Paris. MacWhinney, B., 2000. The CHILDES Project: Tools for Analyzing Talk. Erlbaum, Mahwah, NJ. MacWhinney, B., 2007. The Talkbank project. In: Beal, J., Corrigan, K., Moisl, L. (Eds.), Creating and Digitizing Language Corpora: Synchronic Databases, vol. 1. Houndmills, Basingstoke, Hampshire, Palgrave-Macmillan. Maniglier, P., 2008. Processing culture: new trends in artificial intelligence and linguistics in the light of structuralism. In: Franchi, S., Bianchini, F. (Eds.), Toward an Archeology of Artificial Intelligence. Springer, Berlin. Manning, C.D., 2003. Probabilistic syntax. In: Rens, B., Hay, J., Jannedy, S. (Eds.), Probabilistic Linguistics. MIT Press, Cambridge, MA, pp. 289–341. Marcken, C.d., 1996. Unsupervised Language Acquisition. Ph.D. MIT, Cambridge Mass. Martinet, A., 1962. A Functional View of Language. Clarendon Press, Oxford. Mayr, E., 1982. The Growth of Biological Thought: Diversity Evolution, Inheritance. Belknap Press, Cambridge MA. Mehl, M.R., Vazire, S., Ramirez-Esparza, N., Statcher, R.B., Pennebaker, J.W., 2007. Are women really more talkative than men? Science 317, 82. Meigret, L., 1542. Traité touchant le commun usage de l’écriture françoise. Republications Slatkine Genève, 1972, Lyon. Meillet, A., 1921. Linguistique historique et linguistique générale. Champion, Paris. Milner, J.-C., 1989. Introduction à une science du langage. Editions du Seuil, Paris. Milroy, J., Milroy, L., 1985. Authority in Language: Investigating Language Prescription and Standardisation. Routledge & Paul, London. Morgan, J.L., 1989. Learnability considerations and the nature of trigger experiences in language acquisition. Behavioral and Brain Sciences 12, 352–353. Mufwene, S.S., 2001. The Ecology of Language Evolution. Cambridge University Press, Cambridge. Nevin, B., 2002. The Legacy of Zellig Harris: Language and information into the 21st Century: Philosophy of Science, Syntax and Semantics. John Benjamins, Philadelphia. Nevins, A., Pesetsky, D., Rodrigues, C., 2009. Piraha exceptionality: a reassessment. Language 85, 355–404. Newmeyer, F.J., 1988. Linguistics: The Cambridge Survey. Cambridge University Press, Cambridge. Newmeyer, F.J., 2003. Grammar is grammar and usage is usage. Language 79, 682–797. Parisse, C., 2005. New perspectives on language development and the innateness of grammatical knowledge. Language Sciences 27, 383–401. Patrick, P.L., 2002. The speech community. In: Trudgill, P., Chambers, J.K., Schilling-Estes, N. (Eds.), The Handbook of Language Variation and Change. Blackwell, Malden Mas, pp. 573–599. Paul, H., 1909. Prinzipien der sprachgeschichte. M. Niemeyer, Halle. Perfors, A., Tenenbaum, J., Regier, T., 2006. Poverty of the stimulus? A rational approach. In: Sun, R. (Ed.), 28th Annual Conference of the Cognitive Science Society. Erlbaum, pp. 663–668. Piatelli-Palmarini, M., 1980. Language and Learning: The Debate between Jean Piaget and Noam Chomsky. Harvard University Press, Cambridge, Mass. Piatelli-Palmarini, M., Noizet, Y., 1979. Théories du langage, théories de l’apprentissage: le débat entre Jean Piaget et Noam Chomsky. Éditions du Seuil, Paris. Pinker, S., 1997. How the Mind Works. Norton, New York. Pinker, S., 1999. L’instinct du langage. Editions Odile Jacob, Paris. Pinker, S., Bloom, P., 1989. Natural language and natural selection. Behavioral and Brain Sciences 13, 707–784. B. Laks / Language Sciences 39 (2013) 31–53 53 Pinker, S., Mehler, J., 1989. Connections and Symbols. MIT Press, Cambridge, Mass. Platon, 1849. Menon, de la vertu. Traduction Victor Cousin Paris, Rey. Pullum, G.K., Scholz, B.C., 2002. Empirical assessment of stimulus poverty arguments. The Linguistic Review 18, 9–50. Quintilianus, F., 90/1842. Institutio Oratoria, Paris, Trad. M. Nisard. Ramus, P., 1562. Gramère, Paris. Reali, F., Christiansen, M., 2005. Uncovering the richness of the stimulus: structure dependence and indirect statistical evidence. Cognitive Science 29, 1007– 1028. Rissanen, J., 2007. Information and Complexity in Statistical Modeling. Springer, Berlin. Sampson, G., 2002. Exploring the richness of the stimulus. The Linguistic Review 19, 73–104. Sapir, E., 1921. Language, An Introduction to the Study of Speech. Harcourt Brace and Company, New York. Saussure, F.d., 1916. Cours de linguistique générale. Payot, Paris. Saussure, F.d., 2001. Ecrits de linguistique générale. Gallimard, Paris. Schane, S.A., 1965. The Phonological and Morphological Structure of French, Modern Languages and Linguistics. MIT, Cambridge Mass. Schuchardt, H., 1909. Die Lingua Franca. Zeitschrift für Romanische Philologie XXXIII, 441–461. Schuchardt, H., 1922. Hugo Schuchardt-Brevier. Ein vademekum der allgemeinen Sprachwissenschaft. Max Niemeyer, Halle. Searle, J.R., 1972. A Special Supplement: Chomsky’s Revolution in Linguistics. Review of Books, New-York. Tomasello, M., 1995. Language is not an instinct (review of Pinker 1994). Cognitive Development, 131–156. Tomasello, M., 1998. The New Psychology of Language: Cognitive and Functional Approaches to Language Structure. Erlbaum, Mahwah, NJ. Tomasello, M., 1999. The Cultural Origins of Human Cognition. Harvard University Press, Cambridge, Mass. Tomasello, M., 2003. Constructing a Language: A Usage Based Theory of Language Acquisition. Harvard University Press, Cambridge Mass. Tomasello, M. (Ed.), 2008a. The New Psychology of Language, vol. 2. Psychology Press, New York. Tomasello, M., 2008b. Origins of Human Cognition. MIT Press, Cambridge, Mass. Tomasello, M., 2008c. Some surprises for psychologists. In: Tomasello, M. (Ed.), The New Psychology of Language. Lawrence Erlbaum, Mahwah, NJ, pp. 1–15. Trivers, R., 2002. Natural Selection and Social Theory Selected Papers of Robert Trivers. Oxford University Press, Oxford; New York. Trivers, R.L., 1971. The evolution of reciprocal altruism. The Quarterly Review of Biology 46, 35–57. Troubetzkoy, N.S., 1939. Grundzüge der Phonologie. Valéry, P., 1941. Tel quel. Edition Oeuvre II, Paris, Gallimard, Bibliothèque de la Pleiade, 1960. Vaugelas, F.d.C., 1647/1934. Remarques svr la langue françoise, utiles à ceux qui veulent bien parler et bien escrire. Droz, Paris. Weinreich, U., 1951. Research Problems in Bilingualism, with Special Regard to Switzerland, Ph.D. Columbia University, New York. Weinreich, U., 1953. Languages in Contact: Findings and Problems. Mouton, The Hague. Weinreich, U., 1954. Is a structural dialectology possible? Word 10, 268–280. Weinreich, U., Labov, W., Herzog, M., 1968. Empirical foundations for a theory of language change. In: Lehmann, W., Malkiel, Y. (Eds.), Directions for Historical Linguistics. University of Texas Press, Austin, pp. 95–188. Whitney, W.D., 1875. La vie du langage. Librairie Germer Baillière et compagnie, Paris.