1. Introduction: integrational linguistics and distributed cognition

Proponents of integrational linguistics are notably short of allies in the field of traditional academic linguistics. However, in recent years a seemingly willing dialogue partner has emerged from within cognitive science, cognitive anthropology, and the philosophy of mind in the form of various advocates of distributed cognition/mind.^{Footnote 1} The grounds for potentially fruitful exchange and collaboration between integrationists and distributors (for short) have been mooted on a number of occasions and the issue has even formed the subject matter of a number of conferences and special journal editions. Although authors on both sides have been quick to point to certain areas of incompatibility between the two research programmes,^{Footnote 2} a measure of reciprocal interest has been expressed. Love (Reference Love2004, p. 526), an integrationist, notes for example that:

Of the distributors, Sutton (Reference Sutton2004, p. 503) has stated that “Distributed Cognition and Integrational Linguistics have much in common. Both approaches see communicative activity and intelligent behaviour in general as strongly context-dependent and action-oriented, and brains as permeated by history.” Cowley (Reference Cowley2004, p. 587) pushes things a little further and claims that integrational linguistics in fact needs to incorporate a theory of distributed cognition if it is to avoid remaining “locked into a critical tradition that neither illuminates semiogenesis nor the many ways in which multimodal activities are orchestrated”. Cowley is also a leading advocate of the derived view that language is itself ‘distributed’, although quite what this claim amounts to in onto-epistemological terms is not always clear. I shall return to this question later on.

The most notable exception to the circumspect enthusiasm which has attended the prospect of incorporating insights from distributed cognition into integrational linguistics has been, perhaps unsurprisingly, the latter’s founding and central figure, Roy Harris. Harris argues that in endorsing the proposition ‘the mind is distributed’ the distributors are the perpetrators of a ‘category mistake’^{Footnote 3} in the sense of the term used by Ryle (Reference Ryle1949) in his attack on Descartes’ classical view of mind, a view which the distributors are themselves seeking to demolish. The irony of this is not lost on Harris (Reference Harris2004, p. 730) who asks: “Having got rid of one category mistake […] why set up another in its place?” Harris (2004, pp. 728–729) sets out a characteristically strong rhetorical stall by invoking a number of examples intended to bring into focus what he sees as the oddness or absurdity of the distributors’ position.

Just as a tie or any other item of clothing simply does not fall into that category of things or entities capable of grieving or harbouring grief, for Harris “the mind – or mind, if you prefer it without the article – just isn’t the kind of thing that can get distributed. I live in a world in which there is in practice no alternative to treating my mental activity as sui generis” (2004, p. 729). It is clear that for Harris the term ‘distributed mind’ is to be treated as synonymous with ‘extended mind’ as used by Clark and others, although it should be pointed out that ‘distributed mind’ is not a term Clark himself uses, nor in general do advocates of socially distributed cognition (see below). In any case, ‘distributed mind’ is certainly a notion which has become widely associated with Clark in particular. Whether Harris can be said to have coined the term himself is a matter of debate.^{Footnote 4} Now, terminological matters aside, it would be unfair to dismiss the kind of examples Harris mentions above as rhetorical cheap shots on his part, for some distributors have explicitly endorsed quite such notions. In the famous case of the senile Otto trying to find his way to the museum, Clark and Chalmers (Reference Clark and Chalmers1998), wielding a metaphysical Occam’s Razor, do not make the intuitive common-sense claim that Otto’s notebook functions as an aide-mémoire and that Otto believes that it can tell him where the museum is, but rather that the notebook actually is the repository of his memory and belief as to the museum’s location. Elsewhere, Chalmers (Reference Chalmers and Clark2008, p. ix) writes of his recently purchased iPhone that:

One might instead ask whether Chalmers’ iPhone has not merely taken over the ‘central functions’ of any previous address-recording or number-calculating technologies he may have possessed (address book, pocket calculator, abacus, etc.) rather than those of his brain. Chalmers’ position logically entails that he would also be willing to regard anything from a wall calendar to a copy of the Yellow Pages as part of himself and his mind under the right circumstances. Can minds also extend into one another? Suppose I ask another person to remember a piece of information for me (telephone number, address, etc.) and I then subsequently ask him/her to relate it to me. In doing so does that person, however ephemerally, become part of my mind, part of me? It would seem so given that that person is effectively performing the same function as my address book, and in their seminal paper Clark and Chalmers (Reference Clark and Chalmers1998) make an explicit appeal to such explanatory simplicity in justifying their extended mind thesis. Any other type of explanation would, in their words, be ‘needlessly complex’. Now, if the distributed answer to the foregoing questions is indeed affirmative, it clearly torpedoes common-sense understandings of both mind and individual selfhood and furthermore points to a deep divide between the distributed position and that of Harris. The divide can perhaps best be regarded as ensuing from a conflict between a theory which makes a virtue of terminological innovation in pursuit of a particular kind of scientific rationality and one which seems to favour a common-sense or lay discourse about the mind and mental acts.

For Harris, it is not so much the case that talk of a distributed or extended mind and the assigning of cognitive status to inanimate material objects is straightforwardly false, but rather that in his view it does not make sense to talk in such terms. Harris (2004, p. 727) declares his “extreme diffidence” in using the terms “distributed mind” and “distributed cognition” because, as he puts it, “I do not know what meaning to attach to them”. In essence then, Harris appears to be advancing an ‘Ordinary Language’ argument against the notion of distributed/extended mind. Indeed, his position is strongly reminiscent of that of Bennett and Hacker (Reference Bennett and Hacker2003) in their criticism of cognitive neuroscience and the habit of certain authors – Daniel Dennett is a particular target of their displeasure – of ascribing psychological attributes to the brain rather than to the human being as a whole. For Bennett and Hacker, to say that the brain feels, thinks, decides, believes, etc. is to commit the mereological fallacy and to simply not make sense, as in their view such psychological notions can only be sensically predicated of individual persons. In support of their position, they cite Wittgenstein’s remark from the Philosophical Investigations that “Only of a human being and what resembles (behaves like) a living human being can one say: it has sensations; it sees, is blind; hears, is deaf; is conscious or unconscious.” Harris, however, encounters a difficulty here, and the force of his argument is somewhat drained by his own philosophy of language. Bennett and Hacker – clearly in thrall to the ‘language myth’ (Harris, Reference Harris1981) and the idea of a language as a fixed code shared by all members of the linguistic community – base much of their argument on an appeal to the supposed rules of linguistic usage, claiming that “nonsense is generated when an expression is used contrary to the rules for its use” and that nonsense ensues from the use of “a form of words that is excluded from the language” (Bennet & Hacker, Reference Bennett and Hacker2003, p. 6). Now, as an integrationist, Harris cannot appeal to such arguments based on norms of discourse since there simply are no rules for the use of expressions, and ontologically there is no language from which words or expressions can be excluded. For integrationists, it is individual language-makers who determine the meaning of words and expressions, not the language to which they allegedly belong. As a consequence, it would seem at most that Harris is only able to invoke this kind of ordinary language argument in respect of himself, and that talk of a distributed or extended mind makes no sense for him personally. This has the effect of depriving his category mistake allegation of much of its rhetorical and philosophical potency, since there clearly are many – intelligent – people for whom talk of a distributed mind and any related or derived claims are reasonable and do make good sense.

A retrospectively notable and unfortunate aspect of Harris’ conceptual critique of distributed cognition, particularly in view of the discussion of ‘distributed language’ below, is the failure to distinguish between the view of the individually distributed or extended mind as advanced most notably by Clark and Chalmers (see also Menary’s edited volume, 2010a) and the notion of a socially distributed cognitive system, the seminal account of which can be found in the work of Hutchins and his study of the collective cognitive tasks performed by the crew in the navigation of a US Navy warship (1995a, 1995b). In fact, unlike the extended/distributed mind view, the distributed cognition perspective does not focus on mental activity at all, but rather on how practical tasks are accomplished through social interaction. And indeed, Harris does not address the latter. This, however, is largely understandable, since at the time of his article in the 2004 Language Sciences Special Issue on ‘Distributed Cognition and Integrational Linguistics’, what was to become the distributed language view still drew heavily on Clark’s individual-centred account of cognition. However, following the formation of the Distributed Language Group in 2005, the influence of Clark appears to have waned considerably, due in part to his problematic views on language and continuing adherence to a ‘naïve realist’ position (Steffensen, Reference Steffensen and Cowley2011). In fact, Harris (Reference Harris2004, p. 733) had already identified Clark as a “reocentric surrogationist”, the pinnacle of integrationist condemnation (see also Wheeler, Reference Wheeler2004, for another representationalist or code-based view of language). Challenging appeals to an individual language faculty, the current distributed language movement instead promotes a social, collective view of language and human interactivity (see Cowley & Vallee-Tourangeau, Reference Cowley2013) in which individual beings and actions are not the primary units of analysis, and it therefore owes far more to Hutchins’ concept of socially distributed cognition. The ontological status or importance assigned to individual minds within this strand of theorizing is not altogether clear, although one can note affirmative references to the notions of ‘distributed mind’ or ‘extended mind’ from authors who would otherwise also advocate social, ecological, or dialogical views of language and cognition, notably Cowley and Spurrett (Reference Cowley and Spurrett2003), Linell (Reference Linell2009), Ross (Reference Ross, Ross, Spurrett, Kincaid and Lynn Stephens2007), and Menary (Reference Menary2010b). Indeed, Menary (Reference Menary2010b, p. 227) notes that the “extended-mind-style arguments” of Clark and Chalmers are “allied” to those of distributed and embodied cognition associated with, amongst others, Hutchins (Reference Hutchins1995a) and also Gallagher (Reference Gallagher2005).^{Footnote 5} Hutchins (Reference Hutchins2000, p. 4) himself has also written that “the cognition of an individual is distributed cognition too”.

2. Harris and the ‘vulgar concept of mind’

Harris’ principal objection to the notion of distributed or extended mind is that it conflicts with and indeed appears to involve a significant reconceptualization away from a “vulgar concept of mind” (Hampshire, Reference Hampshire, Wood and Pitcher1971, p. 20). Such a concept is underwritten by what Harris (Reference Harris2008, pp. 1–5) terms ‘vulgar mindspeak’ – a province of ‘ordinary language’ which serves common-sense or folk-psychological intuitions about the mind and mental activities. Such language comprises the everyday use of verbs such as think, intend, believe, imagine, remember and nouns such as idea, belief, memory, etc., and according to Harris (Reference Harris2008, p. 2) “allows one to talk, in short, as if there were no doubt whatever that human beings had minds, and that minds were where most of human thinking was done”. Harris contrasts vulgar mindspeak with what he mischievously refers to as ‘cognobabble’ – what he regards as the arcane and obfuscatory jargon employed in many discussions of the mind and particularly from within the field of cognitive science. According to Harris (Reference Harris2008, p. 4):

Harris’ support for the vulgar concept of mind has found little favour amongst the distributors, and indeed is the source of some impatience and bewilderment. Spurrett (Reference Spurrett2004, pp. 500–501), for example, is deeply suspicious of appeals to common sense over and above the claims of scientific enquiry.

Spurrett would seem to be implying that Harris’ appeal to a lay notion of mind is on a par with, say, the privileging of folk theories of medicine at the expense of the empirically tested claims of medical science. However, it is important to appreciate certain subtleties in Harris’ position here. His point is not that vulgar mindspeak represents some kind of “transcendental wisdom about mental activities”, and he acknowledges that “it can be thoroughly misleading when its distinctions are promoted to the status of theoretical truths” (2008, p. 148). His main target here is clearly theorists, including certain proponents of distributed cognition, who promote erroneous telementational accounts of communication. For Harris (2004, p. 729), such accounts in part find their origin in theorists taking too seriously idiomatic lay expressions such as conveying one’s thoughts. It is not the case then that vulgar mindspeak provides an infallible theory of mind or shows conclusively what (and where) the mind really is. Indeed, it might even be questioned whether vulgar mindspeak can actually be said to imply or constitute a theory of mind at all, let alone a coherent or consistent one. This is the view of Melser (2004, p. 184), who writes that:

Melser makes a compelling point here, although one might respond by arguing that vulgar mindspeak at least provides some basis for an implicit theory of what the mind is not or cannot be, which would presumably include something which can be located outside the human body in material artefacts or any other elements of the local environment. However, for Harris, the category mistake which he attributes to the distributors would seem to lie not so much in their locating the mind outside the skull or the body, but in their seeking to locate it at all. As such, the issue of whether the mind and cognition is distributed or not is itself misconceived. For Harris, the question of location therefore simply does not or ought not to arise.

Set against this is the fact that vulgar mindspeak, which often overlaps and intersects with vulgar brainspeak, would intuitively appear to locate the mind in the head, and indeed Harris is quick to warn against investing any ‘blind faith’ in the vocabulary of vulgar mindspeak. Yet, at the same, he still “cannot see it ever being replaced by a better or more ‘scientific’ one” (2008, p. 148). Here we come to the crux of the matter. The question is then whether the distributors are able, as they implicitly claim, to offer a viable scientific account of mind and mental phenomena which improves upon lay or common-sense conceptions. Harris is deeply sceptical about theories of mind constructed under the banner of ‘cognitive science’. Indeed, his position would seem to question whether there can be any such thing as a viable ‘science of the mind’, or at least one which in his view does not simply involve a gratuitous stipulative redefinition of what counts as either ‘mental’ or ‘scientific’. He notes of such theories that:

Harris here touches upon an important and powerful argument for proponents of distributed cognition to contend with, and which bears similarities to that advanced by Button (Reference Button2008), who also levels the charge of non-verifiability. Button notes, rightly, that certain proponents of distributed cognition draw conclusions about inner computational-type processing in the brain on the basis of publicly observable behavioural data. Yet such internal processing is not observable or measurable. Instead, it is simply postulated as the result of an a priori adherence to a computational model of mind.

It is not necessary to accept every prong of Button’s argument against distributed cognition to see that he raises a crucial issue here. In fact, the linguistic strand of his argument, which takes as its inspiration Wittgenstein’s rejection of the possibility of a private language in order to question the validity of an inner world/outer world dichotomy, is unlikely to hold much appeal for the integrationist, the reason being that the denial of such a dichotomy would seem to conflict sharply with the individual’s own lived experience. As Harris (2008, p. 158) notes, “there is no single, common, undivided world that ‘we all’ live in. I don’t even know who ‘we all’ are.” Moreover, for Harris it is the role of vulgar mindspeak to articulate the connections and integrations between these inner and outer worlds. It would therefore be misplaced to criticize distributed cognition on account of its affirmation of an inner realm. The difficulty instead arises in relation to the kind of claims which are made as to the nature of that inner realm and what can be known about it.

Interestingly, Harris is not the only author to explicitly accuse distributed cognition of comprising a category mistake. Nardi (Reference Nardi2002), for instance, detects a ‘major category error’ in distributed cognition’s ascription of cognitive status to non-human elements and artefacts.^{Footnote 6} She argues that to talk of ‘non-human cognitive agents’ is to pervert the meaning of cognition beyond recognition and usefulness by eradicating from its definition certain key elements which are central to human cognition and cannot be shared by non-human elements. She notes that “[a]s far as I know, distributed cognition does not attribute awareness and judgment to ‘non-human cognitive agents.’ A theory of cognition that deletes awareness and judgment from the meaning of cognition has reduced its scope to the point of changing the very meaning of the word” (Nardi, Reference Nardi2002, p. 274). This echoes the point made by Harris that whatever is going on inside the machines and artefacts with which a person interacts, such objects and their operations do not thereby constitute part of that person. Perhaps, granted a certain degree of metaphorical licence, it might seem reasonable to talk of one’s smartphone or calculator as being part of oneself and one’s mind, but, pace the distributors, one cannot say that it just simply is the case. Again, though, Nardi’s ordinary-language-style category mistake argument falls down, or rather fails to hit home, on precisely the same grounds as the one advanced by Harris. That is to say that there is simply no such thing that is the meaning of the word ‘cognition’ and no language-immanent rules for its use which have been or indeed can be violated. In fact, there seems to be a striking lack of unanimity as to the question of precisely what is meant by ‘cognition’ and many of its derived terms and notions, as is starkly illustrated by Menary’s (Reference Menary2010b, p. 230) observation that there is “no real agreement in the cognitive science community on a definition of what a cognitive process is nor of what the vehicles of cognition are”. It is interesting, however, that this failure to agree on a common definition has not led more cognitive scientists to question the scientific status of their discipline. We would certainly be surprised and no doubt disturbed if neuroscientists were to cheerily announce that they were unable to reach any sort of consensus on what the brain is or what constituted a neural process. After all, it is precisely an analogous difficulty which has led integrationists to question the self-proclaimed scientific status of modern orthodox linguistics, i.e., its failure to satisfactorily and non-arbitrarily identify and define its purported objects of study, namely individual languages. Even Saussure recognized that the failure to define the object of study was a mortal threat to modern linguistics’ claim to be a science, although his solution to the problem was ultimately an unsatisfactory one.

3. Harris’ (integrated) mind

Given that distributed cognition is unable to show (as opposed to infer) what – and hence also where – the mind is or might be, does Harris have a more compelling positive (as opposed to positivist) account of mind on offer? This is a matter open to debate, although one can concede that it is an account unlikely to win many converts from within cognitive science. First, it should be made clear that Harris has little truck with eliminativist theories of mind, and is quite clear in stating that he believes himself (and presumably other people) to have a mind (Harris, Reference Harris2008, p. 155). However, he immediately adds the disclaimer that this belief is not based on any arguments or evidence and that he finds it unlikely that he will “ever find much more about it than [he] can gather from the humdrum experiences of daily life”. Harris’ argument for the existence of his mind is essentially a linguistic one based on the affordances of English vulgar mindspeak. He admits as much, noting that “the mind I think I have is largely a linguistic construct […] I can imagine that if I had been brought up to speak a quite different language, I might have learnt quite different ways of talking about my inner and outer worlds – perhaps it might have been a language in which there was no word for mind at all” (Harris, Reference Harris2008, p. 158). Indeed, he even expresses a measure of empathy with the view that the existence of mental activities does not logically commit one to the postulation of a mind in which those activities occur. Yet, for Harris, such a view nevertheless constitutes a “verbal cop-out” since the “way vulgar mindspeak works in practice hardly allows us to admit to mental activities and in the same breath deny that we have minds”.

Harris’ reasoning here is, however, perhaps not especially compelling, and he could be equally vulnerable to accusations of a ‘cop-out’, since his position suggests that vulgar mindspeak is or ought to be beyond any possibility of revision or reform (see below). However, the question remains as to whether mental activities necessarily need a mind in which to occur, and it is one which cannot be answered by any analysis of linguistic usage, ordinary or otherwise. It is essential not to conflate the semantics of the word mind with questions of ontology. Whether we are actually in possession of real-world things which we might call minds is independent of any linguistic conventions which may make reference to them.

Harris’ standpoint would seem somewhat tantalizing and probably also deeply unsatisfactory for some, for in effect it amounts to saying that “yes, there may really be no such thing as the mind, but ordinary language (in English at least) gives us little option other than to talk as if there were such a thing”. Is Harris skirting close to a deterministic position here? It is perhaps significant that Harris’ account is largely in the first person. If his point is merely that our ordinary language used to refer to our mental activities is so permeated with terms and expressions implying that there are such entities as minds that he himself is willing to go along with it for the most part – all the while conceding that his mind to which he refers is largely a linguistic construct – there would appear few grounds for dispute. However, given that Harris regards his mind – and presumably those of others – as an essentially linguistic construct, one might counter by arguing that, this being the case, one is therefore free to define and construct it as one chooses and that consequently his attribution of a category mistake to those who choose to construct it differently falls down. What linguistic grounds are there for discouraging one from constructing the mind such that it extends beyond the skull and body of the individual? If you choose to regard certain exosomatic objects as part of you and your mind for particular purposes, who am I to argue? To parry this objection, however, one could question what sense it makes to talk of the location of linguistic constructs given that they are – almost by definition – not real-world objects or entities which occupy physical space. When viewed in this way – i.e., as a linguistic construct – mind would seem to belong to the same order of such concepts as personality or identity. Now, the question ‘Where is your personality/identity located?’ seems an odd one and I find that, upon personal reflection at least, I have no particular view on the matter and indeed have trouble making sense of the question. One might respond here by claiming that if my mind or personality must, as Harris would maintain, be part of me, it must in some sense also be in me. But in what sense exactly? Again, here no clear answer emerges. Perhaps figuratively or metaphorically, but as Harris (Reference Harris2008, p. 63) himself has noted, it is far from clear that the distributors are indulging in metaphor. The question then is whether reference to the distributed or extended mind is anything more than just a rhetorical device, a ‘way of talking’ to attract attention to the wider research programme, or whether they are claiming that the mind is something more than a linguistic construct and really is distributed or extended.

What Harris is in effect doing is denying the possibility of a ‘real definition’ of mind, that is to say a definition which states the essence of the thing ‘mind’ and which pre-exists its naming. Providing a real definition of mind is therefore not to be regarded as a parallel enterprise to giving such a definition of a natural substance like salt or copper, for there is no equivalent empirical enquiry which can be conducted into the matter. If a definition is to be provided for some particular purpose then it must be of the word mind, which requires one to look at and select from the multiplicity of ways in which it used. But is a definition of mind even necessary other than for the purposes of academic theorizing or compiling dictionaries? After all, lay speakers are presumably able to successfully integrate the term mind and its various cognates into patterns of communicative behaviour without having any firm conviction as to what ‘the mind’ actually is, for there is rarely cause to reflect on the matter. Does the question of whether humans have minds, let alone where they might be located, even crop up in lay discourse? When I tell someone that some matter or other has been “on my mind all week”, I would be surprised to say the least if I received a reply along the lines of “well, can you be sure that you even have a mind and if so, where is it?” After all, mind is manifestly a term which originates in ordinary, lay language, much like its conventional translation equivalents in other European languages – Geist, geest, esprit, etc. Use of the term does not presuppose a scientific account or theory of mind. An instructive analogy can here be made with the integrational critique of orthodox linguistics, a central thrust of which is that mainstream academic linguistics represents the attempted (and indeed ultimately futile) scientization of the lay concept of ‘a language’ (Love, Reference Love2004, pp. 528–529). A similar accusation could be made against attempts from within cognitive science to define the mind on a more ‘scientific’ (i.e., surrogationalist) basis. After all, from an integrationist perspective, the semantic footing of ‘scientific’ language is no more (or less) secure than any other form of discourse, i.e., meaning is equally indeterminate (Harris, Reference Harris and Bhaskar1990, Reference Harris2005).

Just as Harris (Reference Harris1996a, p. 148) sees no prospect of or point in reforming lay metalanguage in order to do away with the various mythical assumptions about language it may harbour, he also has no ambition to do likewise with vulgar mindspeak, regarding “anyone who has such an agenda […] as a quack prescribing linguistic pills for philosophical ills’ (Harris, Reference Harris2008, p. 158). Such a position will clearly prove utterly unacceptable for anyone seeking to define the mind on a reocentric basis, whereby the use of a term is validated through its correspondence to an object in the external world. For anyone so inclined, it would consequently be quite intolerable to talk as if one had a mind purely as a consequence of lay linguistic convention. For Harris, however, the challenge is rather to avoid becoming captive to language and to see the myths and misconceptions which it engenders for what they are. Nevertheless, it is interesting to speculate – since he does not address the question directly – on the extent to which Harris would be prepared to go in the direction of favouring vulgar mindspeak over any rival mindspeaks, given its inconsistency. How about religiously or mystically tinged vulgar mindspeak?^{Footnote 7} Would he be willing to sanction talk of the ‘soul’ or ‘spirit’ in the same way he does ‘mind’? Given the indeterminacy and open-endedness of what could be deemed to constitute vulgar mindspeak, it is by no means clear where one is to draw the line.

I should perhaps again make it clear at this point, if it was not already so, that my purpose here is not to call into question or undermine the entire programme of research which goes under the banner of distributed cognition. My limited objective has been merely to examine, from an integrationist perspective, the theoretical cogency and explanatory utility of the proposition that the mind is or can be distributed. However, this leads on to another important point raised by Harris, namely that of whether the distributors’ adherence to the metaphysical contention that the mind is distributed is actually necessary in the context of their wider programme of research. Harris clearly believes not, describing it as a “self-inflicted injury” and “masochistic” (2004, p. 730), and proposes instead to talk of an ‘integrating’ or ‘integrated’ mind.

This is not just minor terminological quibbling on Harris’ part. As he sees it, the advantage of talking of an integrated/integrating mind rather than a distributed mind is that the former makes no claim about the location of the mind and hence does not commit the category mistake which he identifies in the latter. Furthermore, it would also not seem to require one to abandon the vulgar or lay concept of mind, which the notion of distributed mind clearly does. As such, it is consistent with integrational linguistics’ wider concern to be a lay-oriented form of enquiry. This, of course, is not likely to prove a particularly attractive argument for the distributors, whose approach is manifestly not lay-oriented and may even be said to be hostile to such an orientation. Sutton (Reference Sutton2004, p. 517), for example, sees the avowed ‘lay orientation’ of integrational linguistics as a serious obstacle to the advancement of a viable research programme, noting that “Harris’s restriction of linguistics to (suitably purified versions of) common-sense understandings threatens to paralyze enquiry”. This hostility to a lay-oriented approach is revealed perhaps most starkly in the jargonistic terminology frequently employed by the distributors. For example, instead of ‘people’ or ‘individuals’ we find terms such as ‘embodied brains’, ‘brains-bodies’, and a pervasive predilection for what Harris terms ‘computational rhetoric’. Indeed, it is the peculiar idiom employed in much of the distributed literature which constitutes one of its most inaccessible features from an integrationist point of view. Button (Reference Button2008) – not an integrationist, although making a point with which Harris would likely sympathize – argues against distributed cognition on the grounds that it gratuitously re-describes the socio-cultural world in the ‘arcane language’ of cognitive science (see also Jones, Reference Jones2010). Yet this language of cognitive science itself often draws heavily from the language of computer science, and so what we in effect see is the socio-cultural world described at double remove in a kind of second-hand computerese. It would be an interesting exercise to see whether the distributors would be able to reformulate the full gamut of their claims in something resembling ‘ordinary language’, and eschewing any computational jargon. Whether they would be inclined to do so is another matter. Spurrett (Reference Spurrett2004) makes a telling observation when he notes that integrational linguistics’ aversion to computational talk in relation to the study of language and the mind points to a ‘deep divide’ with distributed cognition and scientific approaches more generally. If, as Spurrett claims, acceptance of a computational model of mind is indeed an “admission criterion for the distributor club” one is unlikely to see too many integrationists filling out membership forms.

A more promising candidate for reconciliation with Harris’ idea of an integrated mind might – on initial inspection at least – be found in the notion of ‘cognitive integration’ advanced by Menary (Reference Menary2007, Reference Menary2010b). Rejecting more traditional internalist and connectionist accounts of cognition, Menary emphasizes the interplay and integration of an organism’s body and bodily processes with certain features of its environment in accomplishing cognitive tasks. For Menary (Reference Menary2007, p. 13) a cognitive task includes such acts as “perceiving the world, remembering things about the world and employing things remembered in making inferences, problem solving and the like”. Menary (Reference Menary2010b, pp. 229–230) writes that “the primary motivation for cognitive integration is the brute fact of our embodiment, especially our bodily manipulation of environmental vehicles”. A chief concern of Menary is to establish that the bounds of cognition and hence the mind lie beyond the brain and body, in contrast to the view of internalists such as Adams and Aizawa (Reference Adams, Aizawa and Menary2010), for whom the mind is very definitely ‘in the head’. Again, there is a focus on locating cognition and the mind in spatial terms. Harris rejects the validity of debates on the location of the mind altogether, so it is important that by questioning externalism he is not immediately categorized as an internalist alongside Adams and Aizawa. Indeed, Harris explicitly rejects the view that the mind is bounded by the brain, The one area in which Menary’s and Harris’ otherwise quite kindred notions of ‘integration’ perhaps do not quite coincide is in regard to which elements of a process either is prepared, for the sake of argumentation, to regard as cognitive. However, given the quite vague agreement on what the term means, this is ultimately a minor matter given the more important shared insight, namely that of the body’s and brain’s necessary and unavoidable integration with elements of the external physical environment in performing a wide range of contextualized human activities.

4. Integrational linguistics and distributed language

An outgrowth of the increasing reciprocal interest between advocates of distributed cognition and certain researchers in the language sciences has been the emergence of a would-be research paradigm centred upon the notion of ‘distributed language’.^{Footnote 8} Like integrationists, proponents of distributed language are seeking to combat the code-based view of language and languages all but ubiquitous in mainstream linguistic theorizing, as well as traditional approaches to linguistic cognition founded upon notions of inner symbolic representation. Distributed language research typically focuses on how individuals coordinate their interactions with each other and the local environment in first-order communication or ‘languaging’. According to Thibault (2011, p. 5), the rationale behind the use of the term ‘languaging’ as opposed to the more common ‘language’ or ‘language use’ is that “what we habitually and unthinkingly call ‘language’ is an open-ended meshwork of interlinked functioning components founded on material dynamics that know no single stable state based on abstract forms. If we assume that language is a stable synchronic state, we end up with the idea that people ‘use’ a stable language system and its concomitant that individual persons are ‘language users’.” Here we see a reformulation of the integrationist insight that humans are more properly regarded as ‘language-makers’ (Harris, Reference Harris1980) rather than users. Admittedly, this is little or no mention of the term ‘languaging’ in the integrationist literature, but there is clearly a great deal of common ground here based on the rejection of a Saussurean-type linguistics concerned with the description and analysis of form-based abstracta.

What, though, are we to make of the proposition that language is distributed, and what theoretical status are we to assign to it? While Linell (Reference Linell2013) is quite prepared to talk of Distributed Language Theory, the position statement of the Distributed Language Group^{Footnote 9} makes the claim that “[t]here can, we believe, be no theory of distributed language”, which itself of course is very much a theoretical proposition. What then is the motivation for viewing language as distributed? According to Rączaszek-Leonardi (Reference Rączaszek-Leonardi2009, p. 670), “Far from being reducible to a system ‘in the brain’, language must be viewed as radically heterogeneous and as spread across space, time and bodies”, while Steffensen (Reference Steffensen2009, p. 684) similarly claims that “language is a heterogeneous set of physical, cognitive and social activities that unfold in real-time on many time-scales. It arises as we adjust to each other and coordinate our life worlds with each other, behaviourally and cognitively.” The key point of emphasis then in the distributed account would seem to be the non-local or, as Linell (2013) would prefer it, the non-localizable character of language.^{Footnote 10} The parallels with the notion of the distributed mind are clear. If language is not just to be found in the brain and about the body, it must be therefore be seen as extending into the local environment and incorporating or co-opting non-human elements of it. This non-localizability leads Cowley (Reference Cowley2014, p. 68) to claim enigmatically that language is both “nowhere and everywhere”. The immediate suspicion which again arises though, from a Harrisian perspective at least, is whether the same kind of category mistake is not once again being made. While an integrationist would certainly agree that linguistic communication involves the interaction of mind, body, and environment, does this warrant the conclusion that language is itself distributed? Does non-localizability entail distributedness? Do the questions ‘Where is language?’ or ‘Where does language take place?’ even make sense? They are certainly not the sorts of questions one can envisage cropping up in lay discourse.

What does not emerge especially clearly from the distributed language literature is a unified sense of what ‘language’ is or is taken to be. Perhaps the most succinct definition is that offered by Thibault (Reference Thibault2011, p. 7) who states that “[f]irst-order languaging […] is whole-body sense-making activity that enables persons to engage with each other in forms of coactions and to integrate themselves with and to take part in social activities that may be performed either solo or together with other agents”. This is a definition with which an integrationist, although unlikely to put it in quite those terms, could not find too much fault, as it highlights the important integrationist point that language is not limitable to mere verbal output and that no definitive dividing line can be drawn between the linguistic and non-linguistic components of a communicational episode. To this extent, the distributed and integrational approaches to language would seem to largely coincide as regards their views of what first-order linguistic activity actually involves. The more contentious issue would seem to centre more upon the terms and concepts invoked in its description and analysis. As with the notion of distributed mind, an integrationist will again be inclined to pose the question of what benefit there is to be gained from describing language as distributed as opposed to integrated. After all, for integrationists the term ‘language’ refers chiefly to the human faculty or capacity for verbal sign-making which comprises both the physical production and interpretation of signs. In first-order communication, this capacity is integrated with other biomechanical and cognitive capacities as well as with features of the local environment. This much is uncontroversial. However, to accept the view that language is distributed would seem to require something of a redefinition away from this view of language, since it hardly makes sense to say that my capacity to create signs extends beyond my brain and body, beyond me in effect. According to Cowley (Reference Cowley2009, p. 497), “language emerges as bodies engage with both each other and the world”. The thing is, though, almost any form of human (inter)activity could be described in this way – cricket emerges as bodies engage with each other and bat and ball, fishing emerges as bodies engage with rod and line, etc. Are cricket and fishing also distributed? If so, it is not especially clear what is to be gained from describing them thus. What would it contribute to our understanding to say that cricket is distributed across players, balls, stumps, etc.? What such activities manifestly are though is integrated and integrating. From the distributed perspective then, it seems language is not merely to be equated with human sign-making but expanded to also include the material artefacts and products which may facilitate and result from such sign-making.^{Footnote 11} Is there any temptation for the Harrisian integrationist to view language in this way? It seems doubtful, since the notion of integrated language alone performs the necessary theoretical task – i.e., that of emphasizing the interaction of mental, physical, and environmental phenomena – without requiring one to subscribe to the rather arcane metaphysics inherent in the distributed view. If we are to regard language as a form of semiologically driven behaviour or activity, it is surely a category mistake to assign linguistic status to non-human material artefacts. To do so is to confuse their material form with their semiological function, and in order to function semiologically they require the presence of a sign/language-maker, that is to say someone who integrates them into a pattern of activity. Viewed in this way, the question of the location of the sign as a semiological phenomenon – be it localized or distributed – is a red herring and just does not arise.

A more substantial and, in my view, potent integrationist objection to the distributed language approach might run as follows. As a self-avowed ‘scientific’ approach to human interaction, the distributed language approach needs a determinate, objective object of analysis. Indeed, a unit of analysis. In passing, one can note the rather rich irony of Harris’ integrational linguistics serving as a theoretical ‘springboard’ (Steffensen, Reference Steffensen and Cowley2011, p. 204) for a supposedly scientific approach to human interaction, given his repeated contestation of linguistics’ claim to scientific status. Clearly, in seeking to identify such a unit of analysis, the phenomenological Saussurean dictum that, as far as language is concerned, it is the point of view which creates the object, will not do in this regard. Now, of course, by endorsing an integrationist view of language and communication which denies the linguistic sign any sort of determinacy, one is forced to seek determinacy elsewhere. The distributed language view does so by locating it in the postulated distributed socio-cognitive system or ‘extended ecology’, which it sees as responsible for producing human interactional behaviour. Steffensen’s (Reference Steffensen and Cowley2011, p. 205) remark that “in order to fully grasp the complexities of human interaction and cognition, the entire extended ecology will have to become our unit of analysis” is highly revealing in this regard. The “entire extended ecology” is clearly held to be a determinate entity. This may well be the case, of course, if it amounts to nothing more than the claim that the workings of the universe are fundamentally determinate and regular. Integrational linguistics also has no place for magic or mystification. The more important question is whether this “entire extended ecology” is in any realistic or practical sense determinable, especially when it includes, as it must surely do, a temporal dimension (Uryu, Steffensen, & Kramsch, Reference Sutton2014). How does one go about determining all the events and arrangements of matter which have contributed to an episode of human interaction? How does one identify and rank relevant causality within that process? It is rather like retrospectively explaining the weather. How far back does one go? How can one be sure one has all relevant data? Explanations of everything come to an end somewhere. Different human purposes call for different levels of explanation, different units of analysis. The problem with the “entire extended ecology” is that it is so unwieldy a notion as to be barely credible or operable when conceived of as a unit. The notion of a unit implies countability, and therefore also discreteness. How does one go about identifying the boundaries of and between different extended ecologies? If there are none, or it cannot practically be done, then we can hardly be said to be dealing with a unit in any recognizable sense of the term. Just as Saussure was never actually able to identify with any conviction any actual, real-world examples of a langue in his technical sense of the term as an ontologically real social unit of the human language faculty, it seems as if a similar difficulty may attach to the extended ecologies of the distributors. What is interesting in both cases is that a communal social system, admittedly of a quite different type and emphasis, is posited to which the individual person appears wholly subordinate. Under both the Saussurean and distributed models, it is the system which ultimately determines an individual’s behaviour as well as his/her phenomenological experience. Hence Steffensen’s (Reference Steffensen and Cowley2011, p. 205) remark that “in an extended ecology […] biology and phenomenology cease to be individual”. Now, of course, pointing out similarities between Saussure and the distributed language approach may be seen as a provocative manoeuvre, given the latter’s wholesale rejection of the former. However, it seems clear that under both approaches the individual qua individual is required to fall out of view in order to establish a determinate, albeit highly abstract, unit/object of analysis.

From an integrationist viewpoint, in seeking to understand human communication, the solution to the quandary thrown up by the indeterminacy of the linguistic sign is not merely to relocate determinacy or to go and seek it elsewhere. The only semiological determinacy an integrationist is willing to recognize is a phenomenological one based on the individual’s own experience. This is not to affirm the view that individuals are merely doing and saying what they claim to be doing and saying, although this subjective dimension is clearly lost sight of to a considerable extent in the distributed language view. Lay (meta)linguistic knowledge does not hold a trump card in this sense. The point rather is that the semiological dimension of a communicative encounter cannot be ‘got at’ from a third-party, outsider perspective other than through an essentially and ultimately inadequate behaviouristic account. While the distributed view sees the integrationist preoccupation with the individual’s private semiological experience as unduly narrow, one might counter by asking to what extent viewing language naturalistically as socially coordinated whole-body sense-making activity underplays the role of individuals’ second-order metacommunicative beliefs and conceptualizations in shaping their interactional behaviour (dynamics vs. symbols). Once again, though, one is left faced with the seemingly intractable problem of how to go about investigating such a question empirically.

Distributed language claims to be concerned with human interactivity in its broadest sense. An important point to consider, however, is that not all human interactivity is semiological activity. Not all integrations involve contextualized sign-making activity. The most obvious examples here are purely sensory experiences and the biomechanical responses they draw forth (Harris, Reference Harris1996b, pp. 176–178). Now, in claiming that “language is real-time metabolic activity” and that “language functions as airborne synapses” (2011, p. 205), Steffensen would seem effectively to be eradicating any distinction between the semiological and non-semiological. Language certainly would not be possible without metabolic activity, but it is altogether another matter to claim that language is metabolic activity. Playing the piano is also not possible without metabolic activity, but it hardly advances our understanding of what it involves and requires of individuals to say that it just is metabolic activity. A conception of language removed from semiology and which is theoretically unable or unwilling to acknowledge the distinction between those forms of interaction mediated by the real-time contextualization of signs by individual sign-makers and those which do not is an impoverished one and explanatorily inadequate. Language is in a very important sense then not like breathing or thermoregulation.