Are languages digital codes?

Language Sciences 29 (2007) 690–709

www.elsevier.com/locate/langsci

Are languages digital codes?

Nigel Love

Department of Linguistics, University of Cape Town, 7701 Rondebosch, South Africa

Abstract

Language use is commonly understood to involve digital signalling, which imposes certain con-straints and restrictions on linguistic communication. Two papers by Ross [Ross, D., 2004. Metalin-guistic signalling for coordination amongst social agents. Language Sciences 26, 621–642; Ross, D.,this issue. H. sapiens as ecologically special: what does language contribute? Language Sciences 29]are discussed in this connection. It is evident that the particular limitations of digital language thatRoss is interested in depend on the claim not just that language is (partly) digital but that languages

are digital codes. But it is questionable whether languages are codes at all. The idea that they are mayderive some force from the fact that the most commonplace and familiar semiotic devices we call‘codes’ are digital in character. If codes are digital and linguistic units are in some sense or degreedigital, that may explain the temptation to think of languages as digital codes. But closer examina-tion of the digitality of linguistic units offers no support for the digital-code idea, for language use, itis argued, is in its essence fundamentally analogical.� 2007 Published by Elsevier Ltd.

Keywords: Carr, P.; Dennett, D.; Digital code; Distributed cognition; Language and cognition; Ross, D.; Se-miology

1. Introduction

Language is bound to be a focus of intense interest in any attempt to understand humancognition. It is a trite truism that the fact that the cognitive abilities of Homo sapiens out-run those of other species and are in certain respects unique is connected in some way with

0388-0001/$ - see front matter � 2007 Published by Elsevier Ltd.

doi:10.1016/j.langsci.2007.01.008

E-mail address: [email protected]

mailto:[email protected]

N. Love / Language Sciences 29 (2007) 690–709 691

the fact that Homo sapiens uses language. But exactly what it is that language does for us,and how it does it, are questions to which there are no agreed answers.

The general issue of ‘language and cognition’ may be at once separated into two ques-tions or clusters of questions: (i) how we cognise1 language and (ii) how language facili-tates certain cognitive powers distinctive of human beings. Underlying both, of course,is the prior question how language itself is to be conceptualised. Not only does that ques-tion have priority, I think that trying to answer it may illuminate the other two. I shallapproach it here via an analysis of the proposition that languages are digital codes.

2. Language and digitality

Communication by means of language is commonly understood to involve the deploy-ment and interpretation of signals that are discrete, arbitrary and systematically combinable.‘Discrete’ here implies e.g. that ‘communication’ in the previous sentence is analysable as aninstance of the English word communication as distinct from the French word communica-

tion or the English words communion, community, commutation and other items it might incertain circumstances be confused with. Whereas in contrast the difference between thenon- or quasi-verbal utterances representable as ‘mm’, ‘mmmmm’, ‘hmm’ etc. is non-dis-crete: despite the fact that these may have different and readily differentiable meanings,there is no determinate analysis available to settle the question whether we have three dis-tinct units of the signalling system or instances of a single unit capable of continuous var-iation in one or more dimensions. (That indeed is why there is doubt whether to count suchutterances as verbal.) ‘Arbitrary’ makes the familiar Saussurean point that there is ulti-mately no intrinsic reason why e.g. a word meaning ‘communication’ should take thatform.2 ‘Combinable’ has already been adequately illustrated by (what would standardlybe taken as) the various instances so far of the word communication: the main contrast hereis with signals that are ‘semantic isolates – waving a hand, or a handkerchief or a newspaperto attract someone’s attention for example, is perfectly meaningful, but it is not integratedin any structured way with comparable signals’ (Harris, 1984, p. 132).

Discreteness, arbitrariness and systematic combinability are the three main characteris-tics that decisively set off digital from analog signalling. If I am in pain I may grunt andgroan. If I am in extreme pain I may shout and scream. These signals are neither discretenor arbitrary. Grunting and groaning, like shouting and screaming, are not determinatelydistinct from each other,3 and as signals they are universally used by all human beings withthe same meaning or range of meanings. They may to a limited degree be combinable withother signals, but they are not systematically combinable.

The difference between shouting and screaming vs grunting and groaning signals thegreater intensity of the pain. The higher pitch and greater loudness of the vocalisationsiconically represents, or analogically models, the greater painfulness. But the semantic dif-ference signalled by the difference between a mere groan and a piercing scream is a featureof the message as a whole. By contrast, digital signalling allows the separation of differentsemantic elements: one signal for pain, say, and another for the greater intensity. Just as

1 As will be explained, ‘cognise’ here carries no Chomskyan implications.2 That is not to say that, taking its place in a web of synchronic analogies, it may not be ‘relativement motive’

(de Saussure, 1922 [1916], p. 181).3 Any more than the two pairs are.

692 N. Love / Language Sciences 29 (2007) 690–709

when we insert the adjective intense in front of the noun pain and say ‘I am in intense pain’.Intense and pain are two discrete digits, or combinations of discrete digits, of the signallingsystem constituted by the English language. In actual oro-aural communicative use, it isperhaps needless to say, digital signals are combined with analog signals of all sorts,and indeed at the margins the two categories are not always clearly distinguishable.4

Given how obvious it is that the digitality of language and languages enormouslyenhances the communicational possibilities open to creatures possessed of such a signal-ling system, it is perhaps remarkable that the limitations of digital language should be sucha common topos among both philosophers and linguists. Roy Harris observes (Harris,1984, p. 131) that ‘human beings like to think of language as some kind of supreme featof communicational engineering. But in some respects it is extremely crude, clumsy andinflexible.’ Daniel Dennett, in a passage (Dennett 1993 [1991], p. 302) invoking remarksmade by Justin Leiber (1991), likens a natural language to a high-level programming lan-guage in respect of how it ‘constrains the brain’, making certain things easy to say, but atthe price of making certain other things hard to say. Or consider A.N. Whitehead’s remarkthat ‘language foists on us exact concepts as though they represented the immediate del-iverances of experience’ (Whitehead, 1917, quoted by Moore and Carling (1988), which isa whole book devoted to aspects of this topic). Bryan Magee states flatly that the

4 Seesomethbatsmatriumpthis ex

plain fact is that none of our direct experience can be adequately put into words. Andthis is true not only of our sensory experiences of the external world. Going on insideme all the time is a complex and dynamic flow of every-changing awareness, mood,response, reaction, feeling, emotional tone, perceptions of connections and differ-ences, back references, side references, with flickering thoughts and glimpses andhalf-memories darting in and out of the various interweaving strands. I might be ableto imagine this being translated into some kind of orchestral music but certainly notinto words. (Magee 1998 [1997], p. 97)

Don Ross’s version of the ‘limitations of language’ thesis points up what is necessaryand desirable about limitations, with particular reference to the role of digitality in enforc-ing them. The importance for Ross of the fact that languages are (partly) digital lies essen-tially in the communicational constraints that digitality imposes. By imposing constraintsdigitality facilitates co-ordination between interlocutors: ‘I don’t know whether I meanquite the same thing as you do when I use the English word ‘‘democracy’’ . . . The digitalcharacter of our communication system, however, locks us into a tacit agreement to try tocoordinate our respective conceptions round this particular fixpoint of the system . . . weresort to social forces independent of either of us to stabilize this coordination; neitherof us gets to introduce endless complications for the sake of strategic advantage’. But this‘stabilizing power’ is bought at the cost of renouncing the ‘flexibility of fine coordination,the ability to try to achieve equilibrium by mutual antes of infinitely small change. To dothat we must exploit signalling devices from outside the digital system – inflexions, facialexpressions, historical and musical properties of words and their relations that poets

e.g. Harris’s discussion (1981 p. 176 ff.) of cricket-field vocalisations that take the form ‘[zaaaa]’ oring like it, given the formula (‘How’s that?’) laid down by the Laws of Cricket for appealing for an’s dismissal. Is ‘[zaaaa]’ to be taken as a distorted rendering of digital how’s that?, an analog cry ofh, etc., or as something between the two? (This is not exactly the point Harris himself wants to make aboutample.)


exploit as art. That this is a trade-off is shown by the fact that we can’t exploit thesedevices without limit in an exchange unless we’re prepared to undermine the value ofour digital apparatus altogether. Practical conversations carried on entirely in poetry orin nudges and winks are material for comedy’ (Ross, 2004, p. 624).

In this paper Ross illustrates the constraining function of the digital apparatus with ref-erence to the expression of emotions. Emotional signalling systems, making use of, inter

alia, ‘variations in facial expressions, pitch and loudness of voice, posture, physical dis-tance, degree of movement, rate of smiling, extent and duration of eye contact’ are essen-tially analogical in nature: ‘people do not go about in affectless baseline states which theythen periodically modulate by flashing emotional phase-shift signs’. However, ‘interpreta-tion problems confronting signalling in such a system are formidable’, hence the impor-tance of the fact that ‘agents who share a substantial body of cultural conventions willhave at their strategic disposal a range of labels for emotional state-types encoded in theirpublic language, which neither controls’, because that allows them to ‘stabilize dynamicsin bargaining with analog signalling systems, by attaching conventional labels to somesalient, recurrent elements in these dynamics using resources drawn from a digital publiclanguage’ (Ross, 2004, pp. 633–635).

Turning to Ross (this issue), we find that the digitality of language is now identified asthe chief factor in what makes human beings ‘ecologically special’. Once again, the essen-tial contribution of digitality is its power to constrain communicative interaction in waysthat facilitate its integration into, and the integration of, a continuous flow of ongoingactivities. Following Dennett (1993 [1991]), Ross argues that one of the functions of dig-itality is to enable us to narrate ‘selves’ into being.5 A self is a narrative construct similar toa character in a novel. It is ‘a virtual object ‘‘spun’’ in public discourse and closely mon-itored and tended by the organism that hosts and takes the lead in constructing it’. Early inthe process many possibilities remain open, but as people get older the ‘possibility space’becomes narrower. The reduced space is the self, says Ross; the reduction is what enablesothers, along with the subject himself, to predict what he will do across a range of situa-tions, and thereby successfully coordinate their behaviour. Describing their behaviour inthe terms made available by a natural language forces people to sort the data into ‘cate-gorical spaces of lower dimensionality’ than would be used by a neuroscientist or behav-ioural scientist who was trying to be objective. Thus there is a kind of dimensionalcompression through translation into a digital representational format. People have tointerpret and report the data in the terms made available by the culturally determined‘report form’ they find in the environment – the local language. They ‘force their thoughtsto conform to evolved digital categorisation spaces by continuously narrating accounts oftheir behaviour and interpreted mental processes, both ongoing and in retrospect’.

It is evident that the particular limitations of digital language that Ross is interested independ on the claim not just that language is (partly) digital in character but that lan-

guages are digital codes. Ross is concerned to draw attention to the advantages accruingfrom the fact that, to put it crudely and simply, you can only say what there are wordsfor – that by and large you have no option but to express yourself in the words that havebeen codified for you as constituting your particular language. And indeed the words code,

5 Ross explicitly says that ‘the property of language to which Dennett’s account appeals is not that it has syntaxor phrase structure but that it is digital’, but Dennett (1993 [1991]) does not in fact use the term.


coding etc. are prominent in Ross’s discussion. ‘Coding information in a digitally struc-tured way is at least one of the things human languages – and perhaps no other languages– do’ (Ross, 2004, p. 622). ‘Human communication often involves an element of digitalcoding’ (p. 624). He takes ‘the existence of efficient communication with digital vehiclesfor granted, rather than doubting . . . that the codes in question really are in fact (partly)digital’ (p. 623). So it may be apposite to consider the whole question whether or in whatsense languages are codes at all.

3. Are languages codes?

In Love (2004) I presented some general reasons for denying that languages are codesor are usefully to be likened to codes. Since there is no consensus as to what might bemeant by claiming that they are (a fact that may in itself be relevant to answering the ques-tion), debate on this matter is perpetually at risk of collapsing into a mere wrangle aboutwords. What is important here is not what the word code means or what its permissiblemetaphorical extensions might be, but how languages work semiologically. My contentionis that significant points about how languages work semiologically may be made by con-

trasting languages with signalling systems that are uncontroversially understood to becodes in the relevant sense.

The Morse code, I take it, is a code. Familiar systems of road traffic lights or railwaysignal lights constitute codes. Indeed they are digital codes. The essential idea is that someset of physical phenomena, in themselves semiotically empty, encode information, mean-ings, concepts . . . that the user familiar with the system proceeds to decode. Dots anddashes, or their sonic equivalents, in themselves mean nothing. A red light is just a redlight, and even when used as a signal does not necessarily mean ‘stop’. On the cricket fielda red light indicates that the third umpire has given the batsman out.6 The continuouslyflashing red lights on one wingtip of an aircraft serve (by contrasting with the continuouslyflashing green lights on the other) to inform the pilot of another aircraft encountering it bynight whether it is heading towards or away from him.

A fundamental point made in Love (2004) is that language does not stand in any com-parable encoding relationship to anything outside language. The red traffic light, whenoperative in its normal context, might be said to encode the meaning ‘stop’. Whereas anutterance ‘dog’ does not encode anything. That is not to say that it may not be meaningful.It is to say that nothing outside itself can be identified as its meaning. To put it another way,the meaning of dog is ‘dog’. Of course one can attempt to enlighten someone who doesn’tknow what dog means by providing verbal formulae that hint at what it means, or comeclose to characterising what it often means, or commonly serve as translation equivalentsin other languages. But no alternative verbal formula constitutes its meaning. It may insome circumstances be helpful to say that dog means ‘canine animal’. But this is an approx-imation at best. Canine animal is not synonymous with dog; it means ‘canine animal’.

It seems that if the meaning of a linguistic unit has to be conceived of as a verbal itemof some kind (and it is hard to see what the alternative might be), no verbal item willultimately do to state that meaning accurately other than the linguistic unit in question

6 When first introduced the red light meant the opposite, i.e. that the batsman was not out, presumably on thebasis that a red signal light in the road or railway context implies ‘stay where you are’. But this analogy soon gaveway to a broader semiological interpretation of red vs green that caused them to exchange meanings.

Fig. 1.


itself. Dog just is ‘dog’. In this sense it doesn’t encode its meaning but wears it on its face. Itdoesn’t follow, however, that what it wears on its face automatically determines its contri-bution to a given semiotic event.

The possibility of overlooking this point perhaps accounts for the tension sometimesdetected between the stability and potential permanence of writing and the claim that‘signhood is a transient, not a permanent, property of a sign’ (Love, 2004, p. 531). Con-sider the two road signs7 in Fig. 1.

On the left we have a STOP sign, of a design standard in many countries round the world,consisting of a white-bordered red octagon bearing the word ‘stop’ in white capital letters.On the right we have a sign that may be less common but is at any rate frequently encoun-tered on South African roads. Despite what may well be the immediate reaction of somewho are not familiar with it, it is not a STOP sign. Rather, it gives warning that a STOP signis imminent. It does so (i) by constituting in effect a picture of a STOP sign, in that the octa-gon is completely surrounded, or framed, by a red triangle, and (ii) because on road signs ared triangle is the conventional indication of a warning (as opposed to e.g. a mandatoryinstruction). The subtlety with which this message is conveyed lies, I suggest, near the limitsof what such a sign can do without accompanying verbal assistance. Very likely there aremany road-users of at least average attentiveness who, encountering it for the first time,take it for a mere variant of the standard STOP sign. Whereas no one is fooled into takinga picture of a red light for a red light. The reason the STOP-sign warning sign teeters onthe brink of inefficacy is that a heavy semiotic burden (cancellation of the usual significance

7 ‘Sign’ here and in the immediately following discussion refers to certain material objects, which depending oncontext and circumstance may or may not be operative as the vehicles of signs in the semiological sense.


of ‘stop’ on a road sign) is placed on nothing more substantial than the arrangement of thevarious elements in the overall design. The word stop itself remains, and will be read assuch. And stop means ‘stop’. At the same time, it no more means ‘stop’ here than the wordstop on the sign on the left means ‘stop’ once the sign has been uprooted and thrown away.In contrast, whether or not a coloured light, or a sequence of dots and dashes, has anymeaning at all, in any sense, depends entirely on the communicational context.

Carr (this issue) agrees that languages are not codes. He objects to the view that the useof a language involves encoding concepts ‘in phonic or visual substance, which is transmit-ted and then decoded back into conceptual content’. His reason for doing so is not thatconcepts, if they are anything, are from the outset items that take verbal form, but thatthe code idea is ‘deeply mystical’, in that ‘it claims that conceptual content can be trans-

mogrified into phonic substance (‘‘converted’’ and ‘‘transduced’’ are also used in the liter-ature)’. He proposes that this notion be replaced by the notion of physical representation,whereby conceptual content is (non-iconically) represented in linguistic events.8

It is hard to see what this objection, or the proposal for circumventing it, amount to. Inthe first place, is it intended to apply solely to the idea that languages are codes, or to theidea that there can ever be anything that is properly called encoding and decoding? Forwhatever is mystical about the ‘transmogrification’ whereby conceptual content is encoded,it is presumably the process by which traffic lights, for instance, come to have the meaningsthat they have: a meaning (e.g. ‘proceed if the road ahead is clear’) is transmogrified into theswitching-on and shining of a green light. If intended to apply solely to languages, perhapsthe objection is that we have no account or clear idea of what is involved in carrying out thealleged encoding and decoding operations. (Whereas no comparable problem arises in thecase of traffic lights because, whatever it is that happens, it takes place in the public world.)But that wouldn’t suffice to show that language-use doesn’t involve something usefullycalled encoding and decoding (if there were good reasons to think that it does) and that lan-guages are therefore not properly regarded as codes. All it would show, or rather state, isthat we were ignorant of how the trick was done. And if that is the problem, the alternativeproposal – that in using a linguistic form we ‘represent’ a conceptual content – hardly getsus out of the wood. Not only are we still stuck with the idea that a conceptual content (i.e. ameaning) is somehow separable from the linguistic form that ‘represents’ it, it is unclearwhy, if the notion of encoding a conceptual content is mystical, the notion of representing

it is any less so, especially if, as Carr allows, the representing is non-iconic. Using dog torepresent whatever its conceptual content might be must in some sense involve attachingor associating the conceptual content to or with the linguistic form. But we have no moreliteral understanding of any such attachment or association process than we do of the con-version or transduction process that Carr sees as fatally involved in thinking of dog asencoding its conceptual content. The fact is that, given our present knowledge of the rele-vant cognitive phenomena, these are just alternative figures of speech. And if figures ofspeech are all that are on offer, one may as well take one’s pick. As matters stand it seemsas pointless to argue about whether we encode meanings or represent them where languageis concerned as it would be outside the linguistic realm. Having lost the white queen frommy chess set I can use a shirt button to represent it. Alternatively, I can encode the meaning‘white queen’ into my shirt button. But ‘represent’ and ‘encode’ here are being used quite

8 See e.g. Burton-Roberts and Carr (1999) for further details.


loosely and vaguely. They are not naming specific cognitive operations, let alone cognitiveoperations well understood and known to be distinct. It is true that the latter description ofhow – to use another figure of speech – I transform my shirt button into the white queenmay seem, misleadingly, to imply some more elaborate process than the former. The pointis, though, that however I choose to describe it, what we are able to say about what I actu-ally do (announce to the world, or to my opponent, or just to myself, that the shirt buttonwill stand in for the missing queen) remains the same.

If languages are indeed not properly to be regarded as codes, it is for reasons quite dif-ferent from those Carr suggests. As well as the fundamental reason already mentioned,thinking of a language as a code crucially involves abstracting from the vagaries and inde-terminacies of actual speech events and reifying the abstracta. It demands that a givenutterance be taken as an utterance of the abstract invariant that constitutes the relevantunit of the code (cf. Love, 2004, pp. 538–539). Vindicating the code idea would thereforeseem to depend on determinately identifying the units of which a given code consists. Love(2004) offers reasons for doubting that this is possible – for doubting that there is, to takeconcrete illustrations, any ‘context-neutral fact of such matters as whether aluminum andaluminium are or are not the ‘‘same’’ word, or whether an utterance ‘‘[LgEgLgE]’’ is or isnot an utterance of helicopter, that it is the linguist’s professional privilege to discoverand divulge’ (Love, 2004, p. 540).

Although he rejects that idea that language-use involves encoding and decoding concep-tual content, Carr nonetheless thinks of a language as, if not a code, then a ‘system’ ofdeterminately identifiable units, and that it is indeed his professional privilege and dutyto determine what, as a matter of fact, the units of a given system are. Accordingly hetakes issue with the case to the contrary presented in Love (2004). He analyses that caseas having three prongs. He identifies (i) an argument from phonetic variability, (ii) anargument from ambiguity, (iii) an argument from child mispronunciations.

Under (i), taking up one of my examples, Carr observes that aluminum and aluminium

have the same denotation. Therefore, ‘in stating that the foil in the kitchen is made of alu-min(i)um, British and American speakers are accessing pretty much the same thought. Thegrounds we have for claiming this are based on similarity of usage, crucially involving sim-ilarity of context of utterance’. ‘If there are enough situations in which we can identifyAmericans and Brits uttering phonetically similar utterances in sufficiently similar con-texts, we can assume that those utterances, considered as physical representations, repre-sent very similar/overlapping conceptual content’ (Carr, this issue).

Let us accept that we can assume that. The same might be said of pairs such as air-

plane � aeroplane, flammable � inflammable, mom � mum, or, if we are not restricted spe-cifically to American–British variation, among � amongst, kitty � kitten, pants � pantiesand countless others. All these pairs differ phonetically in no more than a segment orso and have very similar or overlapping meanings. The question is whether they arethe same words. All Carr is claiming (and contentiously at that) is that it is possibleto impose a certain criterion that enforces a certain answer. Other criteria might givea different answer. For instance, airplane and aeroplane manifestly involve different pro-cess of word-formation, in that the former appears to be compounded of two Englishwords, while the latter has the medial -o- that indicates the conjoining of Greek or Latinroots. The point is that in cases like these English-speakers will be in doubt or divided onthe ‘same’ vs ‘different’ question, and that is a reason for denying that any determinate‘system’ is in play.


Under (ii) Carr objects, or appears to object, to my claim that there is no unequivocalanswer to the question whether I didn’t leave because I was angry when it means ‘because Iwas angry I didn’t leave’ is the same sentence as I didn’t leave because I was angry when itmeans ‘it wasn’t because I was angry that I left’. Carr’s response is to point out that thereare two distinct thoughts here, and that ‘we are able to use one and the same visual rep-resentation to represent those unambiguous thoughts [. . .] while, in speech, it is possible toproduce two distinct physical representations of the two distinct thoughts’.

This analysis is fair enough, as far as it goes. It explains why the question arises. But itleaves entirely untouched the issue of what would answer it. In effect Carr offers two com-peting conceptualisations of ‘sentence’, in terms of one of which the two readings of I

didn’t leave because I was angry would be readings of the same sentence, while in termsof the other they would be different sentences. But this is precisely to admit that there isno unequivocal answer to the question whether I didn’t leave because I was angry is oris not a unit of the code.

Under (iii) my question was whether the notion that alleged attempts, no matter howprimitive or incompetent, to instantiate a given unit of the code (e.g. the English-acquiringchild who says ‘[LgEgLgE]’), can be taken actually to instantiate it (in this case, the Englishword helicopter) without by implication doing violence to the fundamental idea that a code(or system, if Carr prefers) functions in virtue of the possibility of identifying the units ofthe code in question. In response Carr dismisses what is at issue here by saying that suchmispronunciations are on a par theoretically with ‘adult phonetic variability’: in both casesthere is systematicity, and ‘clear limits on variability’.

But, setting aside what we may or may not know of the speaker’s intentions in the spe-cific case of very young children’s speech, Carr simply begs the question. The issue iswhether an utterance ‘[LgEgLgE]’ stands in the same relationship to the word helicopter

as we might take it that an utterance ‘[helIkÅptE]’ does – i.e. by constituting per se whatwould generally be acknowledged as an utterance of that word. Claiming, as Carr presum-ably would, that [LgEgLgE] falls within the limits as to how far the pronunciation of heli-

copter may vary seems to take it for granted that we know from the outset that [LgEgLgE] isindeed a pronunciation of helicopter. The question is why this should be taken for granted.It is true that in certain highly restricted circumstances ‘[LgEgLgE]’ may be understood asmeaning ‘helicopter’,9 but it does not follow from that, plus the fact that the child whoutters it is learning English, that it somehow just is an instantiation of the word helicopter,however helpful to the progress of the English-learning child imposing such an interpreta-tion might be.10

9 But it would be no good my informing you in the course of general conversation that yesterday I went for athrilling ride in an [LgEgLgE] and expecting you to latch on, immediately or perhaps at all, to what I meant.10 When a character in Kingsley Amis’s Take a Girl Like You ‘hears himself say’, in an extremity of drunkenness,

‘Hallo, I parry stashed a nowhere hermes peck humour speech own. June I haggle unction when donned ring goneoh swear’, his interlocutor’s response is, not surprisingly, ‘Pardon? What did you say?’ (Amis 1962 [1960], p. 219).The significant point here is that what the speaker ‘hears himself say’ is not represented as what he intended tosay. On the account that Carr would apparently give, because the utterance means ‘Hallo, I’m Patrick Standishand now we’re home I expect you must be Joan. Julian and I had lunch and went on drinking elsewhere’, that issimply what was said (however Amis chooses to represent it). And on the basis of knowing that that is what wassaid, we can presumably go on to establish the relevant variability limits (in this case, for drunken adult BritishEnglish).


Ross, too, appears to believe that languages are codes – indeed, that it is important thatthey should be so conceptualised. Nowhere in Ross (2004) do we find an account of whathe means by calling a language a code, but in Ross (this issue) linguistic codehood is saidto consist in the fact that ‘similar public linguistic representations cue similar behavioralresponses in individuals with similar learning histories, as a result of conventional associ-ations established by those similar histories’.

While perhaps not choosing to express it thus, many will find this statement unexcep-tionable. As an account of what a linguistic code is, though, it seems problematic in a num-ber of respects, of which the most piquant is that it is manifestly self-stultifying. Howeversimilar their learning histories, few if any English-speakers confronted with a linguistic‘representation’ transcribable as ‘code’ will suppose that Ross’s formulation has anythingto do with what might be meant by calling a language a code. For ‘linguistic representa-tions’ in the quotation from Ross you can substitute any X you like (lavatories, post offi-ces, railway stations . . .) and come up with an acceptable statement. Doubtless any set ofrecognisably similar public institutions, or local instantiations of some generic public insti-tution, will trigger similar responses . . . But a national ‘network’ of public houses, banks,cricket grounds. . . does not constitute a ‘code’ in any sense of the word except one thatRoss has gone outside the code to invent.

More seriously, one notes that ‘similar’ is doing an awful lot of work here. And that isproblematic, for as Ross himself has said elsewhere (Ross, 1993, p. 2), ‘the idea of ‘‘recog-nizing similarity’’ is philosophically mysterious’. It is also philosophically fundamental, forone reason because recognising similarity is crucial for any form of inductive reasoning.Ross quotes Searle (1979, p. 106) as having put his finger on the difficulty: ‘Similarity isa vacuous predicate: any two things are similar in some respect or other’. Ross’s contexthere is an exploration of the possibility of finding a way round this problem as it arises forany account of the processing and understanding of metaphorical expressions based on theidea – which Ross endorses – that metaphors establish (without overtly stating) a similar-ity between two entities. Ross is sanguine about the prospects for an eventual solution,which is no doubt why he believes that we might ultimately be able to generalise it tothe totality of the workings of whole languages and ‘explain stabilities in communicationby reference to objectively measured similarities . . . even if there is no one dimension ofsimilarity that is always privileged, or even if the set of stabilizing dimensions varies fromcase to case’ (Ross, this issue).

What is going on here? As already suggested, what is usually meant by calling a lan-guage a code involves reification. A language is taken to consist of abstract entities thatremain the same despite differences in their precise instantiations in different mouthsand minds, at different times and places. On an account of this kind Ross’s philosophicalmystery is circumvented by conjuring up what is indeed a privileged dimension of similar-ity among utterances – namely, that in virtue of which certain similar utterances count asutterances of the same abstract linguistic unit(s). The set of abstract units constitutes thecode and, on this story, what Ross calls ‘stabilities of communication’ are achieved in sofar as communicators use the same code. Ross’s appeal to mere similarities (rather thanunderlying samenesses) appears to deny that there are linguistic codes in this sense. Atthe same time, it seems clear that he takes for granted the existence of linguistic codesin this sense, as when he refers to the English word democracy as a ‘fixpoint of the system’(Ross, 2004, p. 624) or speaks of ‘a range of labels for emotional state-types’ (Ross, 2004,p. 643) encoded in one’s public language. What makes democracy a fixpoint of the system,


if it is,11 can only be that it has been codified as such, by a process that had nothing to dowith measuring similarities among utterances.

4. Language and digits

The most commonplace and familiar semiotic devices we call ‘codes’ are digital in char-acter. Language and languages, too, appear to work by virtue of a large element of digi-tality in their semiological constitution. If codes are digital and languages are digital, doesthat explain the temptation to think of languages as digital codes? It may be helpful at thispoint to consider what light is shed by the use in this connection of the word digital itself.Why should the units of the code be called ‘digits’?

Let us look at a set of digits:

0 1 2 3 4 5 6 7 8 9

The Arabic figures constitute a familiar example of what semiologists sometimes call anemblematic frame (see e.g. Harris, 1995, pp. 137, 168f.; Harris, 2000, pp. 106–110). Anemblematic frame is a closed, finite set of symbols with certain ready-made relationshipsamong the members, but which have no intrinsic semiotic content. The most familiarexamples in our culture are sets of items used in playing games. Take dominoes. A setof dominoes is an emblematic frame. The only given relationships among them are prece-dence and equivalence. Taking your turn at dominoes, in the usual game played withthem, requires laying down a piece one end of which has an equivalence relationship withthe available end of a piece already played. The equivalence is usually symbolised as amatching number of pips. There is also a precedence relation – the higher the precedenceof the unplayable pieces left in your hand at the end of the game, the worse off you are. Butthe emblematic frame in question is not intrinsically wedded to playing that particulargame, or to playing games at all. A slightly more complex example would be a pack ofplaying cards. The cards themselves ‘mean’ nothing: they simply offer a system of abstractrelationships, again of precedence and equivalence, on which you can impose any signif-icance you like. The standard pack of 52 cards in four suits of thirteen could readily bepressed into service for calendrical purposes, for instance, in a society that attached specialimportance to marking the weeks and seasons of the year. In fact, apart from divination,we mostly use them for playing games. But note the huge variety of kinds of card games. Isthere any structural similarity between snap and bridge, or patience and poker? They areall played with the cards, but the system of emblems is being exploited in very differentways.

The Arabic figures constitute an emblematic frame in this sense; in fact they form a sub-part of the playing card frame. In this case the only given relationship among the membersis precedence, or priority. There is an established ordering, as given, but its significance isnot fixed. In particular, the figures are not necessarily used for counting. A group of peopleeach pick blindly from a set of tokens marked from 0 to 9. Highest goes first, wins theprize, or whatever. All that matters here is, precisely, that the symbols have an established

11 Is fixpoint a fixed point of the system or a lexical innovation of Ross’s own devising?


order of precedence. Nothing is being counted. Very simple emblematic frames, like thisone, when pressed into service for purposes of writing, constitute notations.

Using the figures for counting, by treating them as representing numbers, is one of themost frequent uses we have for them, and it may be that historically their more semiolog-ically primitive functions are, as a matter of fact, actually derived from this. But as sym-bols, as digits, as a notational system, they have no necessary connection withmathematics, and using them for mathematical purposes requires the imposition of aset of rules. That is to say, the notation has to be used in a particular script. Althoughthe established ordering may determine the arithmetical value to be attached to the tensymbols themselves, if they are to be used in mathematical writing for representing num-bers beyond nine, conventions have to be established:

13 + 6 = 73102 � 56 = 631

In this script units tens and hundreds are written from left to right rather than from rightto left. It is arbitrary what the rules are, but there have to be rules of this kind if the nota-tion is to be used as a script.

I take the use of this notation to yield scripts for mathematical writing as exemplifyingthe basic idea of a digital code. The set of digits becomes a code if and when there are rulesfor its use in systematically encoding semiotic values. The question is whether there is anyparallel or analogue in language.

The obvious candidate parallel is alphabetic writing:

A B C D E F G H I J K L M N O P Q R S T U V W X Y Z

Once again we have a notation, this time with 26 members, in this particular version of theRoman alphabet, with standard precedence relations as given, reading from left to right.This notation shares some of its more primitive uses with the Arabic figure notation, andin those uses is in fact interchangeable with it. (Hence the term ‘alphanumeric’.) So youcan indifferently ‘number’ the points in your argument 1, 2, 3, 4, or, if you prefer, a, b, c,d. In some systems of vehicle registration so-called ‘numbers’ and telephone so-called ‘num-bers’ letters and figures are combined. Note that in this kind of use even the basic precedencerelations are nullified – a telephone exchange may historically have allocated its ‘numbers’ innumerical order, and at one time if no longer you could tell roughly how old the British civilaircraft you were flying in was by how close to the beginning of the alphabet its registrationletters were – but all that matters is that your ‘number’ be a combination of digits differentfrom everyone else’s. So from the semiologist’s point of view the alphabet is no more intrin-sically wedded to representing language than the Arabic figures are to representing numbers.Indeed, one familiar use of a subset of digits from the alphabetic notation (I, V, X, L, C, M,. . .) is mathematical. The ‘Roman numerals’, as it happens, illustrate the possibility thatthere may be indeterminacies in the rules for using a notation as a script. So, for example,4 can be either IIII or IV. And if 4 can be IV, and 9 is IX, can 49 be IL, and if not, whynot? Here the rules seem to have been formulated with a looseness that allows in a rangeof cases for variant representations of the same number.

The question is whether, when the alphabet is used in connection with language, there isany parallel with the way figures are used in mathematical writing. In other words, is thealphabet, as a linguistic script, capable of use as a digital code?


Some writers on the subject seem to think that the mere fact that it consists of a set ofdigits automatically makes the alphabet a digital code. Thus the literary theorist FlorianCramer says that whereas ‘sounds and images are not code by themselves, but have tobe turned into code in order to be computed, any written text already is code’ (Cramer,2001). But this seems to mean just that written text consists of alphanumeric characters.Doubtless that makes them easy to encode, for computing purposes: you assign the char-acters values in a particular, e.g. binary, digital code for use by the computer, such as theASCII code. You can call a set of characters such as those constituting an alphabet ‘code’if you like. In which case there is an important difference between ‘code’ and ‘a code’. Andit is the latter that we are interested in here.

What might the alphabet encode, where language is concerned? One possible, indeedtraditional, answer is: speech sounds. The reason this answer seems tempting is that allknown uses of the alphabet in connection with natural languages are phonographic in ten-dency. To different degrees in different cases there is a discernible correlation between let-ters and sounds. This comes across strikingly where it so happens that the samecombinations of elements of the alphabetic notation are used for different purposes in dif-ferent scripts. For instance, alter, chair, sail, truth are, respectively, a German word mean-ing ‘age, epoch’, a French word meaning ‘flesh’, and Welsh words meaning ‘foundation,basis’ and ‘falsehood’. Alternatively, they may be read as English words. Fortuitous trans-lingual homographies of this kind demonstrate very clearly the difference between a nota-tion and a script. Notationally there are four items, but scriptically there are eight.

What is remarkable here is that, given that the alphabet is after all just a notation, theseidentically written but completely unconnected words in different languages, with nothingin common semantically, nonetheless show a certain phonetic similarity. Moreover, aspart of turning the alphabetic notation into a script some languages impose an extra layerof structure on the notation itself in order to facilitate the correlation with sounds.

A B C CH D DD E F FF G NG H I L LL M N O P PH R RH S T TH U W Y

This is the Welsh version of the Roman alphabet. No J, K, Q, V, X or Z, but nonetheless28 ‘letters’, because consonantal digraphs count as separate units of the script. They havetheir own names and their own place in alphabetical order. It is as if English-speakersthought of, say, shop as a three-letter word beginning with ‘esh’.

So the connection with phonography is there. That is why large stretches of broad tran-scription in a phonetic alphabet based on attaching to the Roman letters Standard Aver-age European sound values are perfectly readable by an SAE speaker who has neverlearned to use the phonetic alphabet as such: bat [bat], hat [hat], mat [mat], sat [sat] . . .

Nonetheless it is obvious that the phonographic tendency is just that: a sort of partialkeeping in touch with the idea of representing sounds. One reason that is all it is is thatspeech neither consists of nor is determinately analysable as a series of discrete sound seg-ments – a problem that phonetic alphabets, specifically and overtly intended to representsounds, have always come up against, let alone ordinary ones.12 Although the linguisticuse of the alphabet may have its historical origins in some sort of analysis of the phoneticstructure of syllables, for many or most alphabetically written languages that is conspic-

12 See e.g. Port and Leary (2005) for recent extensive discussion of implications and consequences of the fact that‘linguistics has mistakenly presumed that speech can always be spelled with letter-like tokens’ (p. 927).


uously not how it actually works now. English is an especially notorious instance. We cansafely conclude that if a language is a digital signalling system, the digits we are looking forare not to be equated with the individual letters of alphabetic writing.

So what does alphabetic writing represent, if anything, and how? Very roughly speak-ing, alphabetic scripts appear to require, minimally, that words or word-level units shouldbe individually recognisable. This requirement is independent of whether, associated withthe use of the alphabet for writing a given language, there are established rules of orthog-raphy. In the case of English, for instance, such rules are a recent development, but in anyevent they are not at issue here. Shakespeare is said to have written his name in a numberof different ways, none of which corresponds to the modern convention, but the require-ment that the word be recognisable in isolation sets limits to the possible variation. It issometimes said that all that is required for legibility, at any rate in a society accustomedto an orthography, is that what would be the first and last letters of a word as correctlyspelt should remain in position, as in

Spraehpsake is siad to hvae wirettn his nmae in a nmebur of dferfeint wyas

but this applies only in certain contexts, such as where there is a combination of wordsallowing the resolution of specific difficulties in the light of an interpretation of thewhole.13 Spraehpsake would probably not be a useful spelling in a context that offers nomore interpretative clues than ‘I like Spraehpsake’.14

Nonetheless, alphabetic writing does not require any systematically consistent represen-tation of words.15 We can certainly set up standard spelling systems if we like, althoughthey are by no means universal, and we can occasionally get a standard spelling systemto make useful word-level distinctions, as with

ritewriteright

wright

But just as often we do not bother, as with

port (not starboard)port (not sherry)

port (not harbour)

And in any case we can do perfectly well without using any such standard system, even ifthere is one available:

13 Ross has a term for people like me who find the idea problematic that languages are digital codes. He calls us‘processing holists’ (Ross, 2004, p. 623). But if something like this can processed at all, ‘holistically’ is a prettygood word for how it’s being done. One’s sense of what is going on here is a function of seeing the whole lottogether, i.e. holistically14 Fcuk, though, is interpretable in isolation (granted a context in which English is known).15 And surely right there we see an important disanalogy with canonically digital signalling. Make the slightest

error when keying in e.g. an email address and the stupid machine sends the message straight back to you.

Cristes mæssecristesmessechrist-massekryst-masse

cristemescristemasse

cristmescristmascrysmascristimas

ChristmasseChristmass


These are all historically attested, and a lot of them, and others you could produce, are stillin common use, especially among children sending out home-made Christmas cards.

Doubtless we all know, or are capable of coming to know, that these are variant spell-ings of the same word – or words. The question is how we know that, if the system forsignalling words is supposed to be digital. The fact is, writing these out required my com-puter to generate different digital encodings for each one. That is how it keeps them apart.Anyone who wants to say that they all in some sense encode ‘the same thing’ has a prob-lem identifying the level at which the sameness gets encoded.

The difficulty is that there seems to be no higher level of abstraction to which we can goif we want to vindicate the idea that the sameness underlying the variants is itself somehow

signalled digitally. Of course there is no difficulty in finding an ad hoc graphic representa-tion of what we are looking for here – for instance, we can represent the English word ofwhich all the above are variant spellings like this:

Take the current standard spelling and put in authoritative underlined Gill Sans caps to
show that this is a metarepresentation of ‘the word itself’ and not just another variant.The trouble is that, play as you will with the typography, it is just another variant. (Downthere at the bottom of the list.)
The writing conventions of English, like those of all other languages, offer no superor-dinate system of metarepresentations consistently establishable as such. Nor are we anybetter off – in fact we’re worse off – if we turn to spoken English. Here of course we findindefinitely many variant signals associated with all or any one of these written forms.

The fact is that these written forms, considered as representations, are indeterminate as

to the scope of what they represent. Take any one of them and ask: does that somehowintrinsically stand for the word itself, or merely instances of that particular version of it?If I took one of the ‘incorrect’ ones at random, e.g.


Christmass

and asked you to ‘copy that down’, what would you do? I think you are as likely as notautomatically to ‘correct’ the spelling. In other words, the level of abstraction at which youare supposed to take the inscription in question is not somehow there to be read off itsface.

So, if languages are digital codes, where are the digits?

5. Language as interpretatively terminal

What is above all else special about language – and what we are up against with theChristmass problem – is that it is interpretatively terminal. There is nothing that standsto language in the relation that language stands in to everything else. This is why, inthe end, the digits turn out to be mysteriously elusive. Real digital codes require to beinvented and their use explained – in some higher-order language that has a semiotic flex-ibility that is simply incompatible with its being itself any kind of established code. It isonly because linguistic signs are radically indeterminate with respect to their identity asunits of a prior code that natural languages can meet the open-ended, unpredictable com-municational demands that we impose upon them.

Language can be used to investigate and talk about anything under the sun, not exclud-ing the sun itself. But what about language itself? A particular kind of use of language totalk about language – that is, a particular way of exploiting the reflexivity of language wecall ‘linguistics’. And the fact that linguistics is language about language has often madelinguists uneasy. For instance, the linguist J. R. Firth remarks that ‘the reflexive characterof linguistics, in which language is turned back on itself, is one of out major problems’(Firth, 1957 [1948], p. 147). But Firth never attempted to solve the problem, or even tostate precisely what he took it to be.

The problem is that if we turn the medium of inquiry back on itself it becomes an object

of inquiry, and to envisage treating linguistic phenomena as objects is, in and of itself, topropose a distorted account of them. There are no (first-order) linguistic objects of anykind. Language is a temporally situated, ongoing process – the process of making andremaking signs in contextualised episodes of communicative behaviour. And if that isaccepted then, apart from the provision of anecdotal accounts of the specifics of particularcommunicative episodes, one might be inclined to conclude that pointing this much out is


where linguistics ought to stop. Whatever it might be to go further, if going further impliesengaging in the kind of retrospective talk about linguistic signs that requires that they beidentified in abstracto, it is not and cannot be a matter of reporting on objectively givenfirst-order realities. In a sense, therefore, linguistics is logically impossible. Identifying asign involves decontextualising the unique communicative episode within which and forpurposes of which the sign was created, abstracting and reifying some aspect of that epi-sode, and presenting the reification for inspection and analysis as ‘the sign’ in question.But whatever is thus presented cannot be the sign in question. For the sign has no existenceoutside its unique communicative episode.

Analytic discourse about language – which involves identifying and citing the linguisticunits recurrently signalled in the digital code – requires decontextualisation, abstractionand reification. And the ultimate basis for these operations lies in the familiar, everydaymetalinguistic act of repetition. What is said or written can be repeated. Someone asksme ‘Did you say ‘‘bat’’?’ My answer is: ‘No, I said ‘‘hat’’ ’. But when I say ‘No, I said‘‘hat’’ ’, I am not somehow identifying the abstract unit of the English vocabulary of whichmy first utterance was a particular instance. I am simply repeating what I said – i.e. pro-ducing another utterance. The repetition will not of course be an exact replica. In manycases it won’t necessarily be anything objectively like the original. It will merely be similarto the first in whatever dimensions of similarity I think contextually relevant for usefully

answering my questioner’s question. What those dimensions are will vary from occasionto occasion. Compare these exchanges:

Did you say [k¤IsmEs]/ No, I said [k¤EsmEs].Did you say [k¤IsmEs]/ Yes, I said [k¤EsmEs].

There is no one, universal, context-neutral dimension of relevant similarity. Ross concedesthis point, but nonetheless insists that the existence of a code can be established by objec-tive measurement of similarities. But how this is to be achieved is entirely obscure. Forwhat counts as ‘similar’ to what is a matter to be decided on the spot, in the light ofthe particular communication situation, by the participants themselves. Knowing no GreekI once managed to find my way out of a maze-like public building in Athens when itdawned on me that if I transliterated a certain sign on the wall I got something very likethe English word exodus. I was working with similarities all right, but presumably not theones the Greek-speakers responsible for putting the sign there had in mind. What wouldRoss propose to measure here?

For certain metalinguistic purposes we do indeed entertain the idea of a context-neutraldimension of relevant similarity, which can be generalised and projected as the basis for athoroughgoing, self-consistent reification of a whole language, giving us a determinateidentification of the morphemes, words and higher-level units that constitute the linguisticsystem. For a sketch of how and why that idea arises, in literate societies, see Love (2004).The core of the argument presented there is that writing is indeed, as Ross says, a mile-stone in human cultural evolution, but not for the banal reason usually offered, that itmakes messages durable and portable. The reason it is a milestone is that it allows for sys-tematically reconfiguring our conception of language itself,16 in such a way as to give sub-

16 Cf. Menary’s discussion in this issue (Menary, this issue) of how writing reconfigures thinking in general.


stance and effect to the idea that utterances are utterances of something more abstract.What is curious about this idea is that its hold on us is simultaneously very powerfuland yet surprisingly feeble. On the one hand we take it entirely for granted that there is

a determinate analysis of our utterances and inscriptions in terms of the words, sentencesand so on, that they instantiate. That of course is the basis for the invention of codes andciphers of all kinds – the very idea of codes, I suggest, is parasitic on the notion that nat-ural languages work by encoding recoverable meanings in fixed, determinately identifiableforms. We retroject that idea on to first-order language itself and elaborate metalinguisticcodifications of languages – i.e. grammar books and dictionaries. While at the same timewe find ourselves completely unfazed by the banal ease with which we can call in questionor change our minds about our criteria for identifying linguistic units – which do indeed,as Ross says, ‘evolve’. We don’t really believe in this story after all.

What do I have to do in order to use language to communicate? As a speaker-writer Ihave to call on my past linguistic experience and creatively deploy parts of it so as to max-imise the likelihood that you will perceive contextually relevant similarities between myutterance or inscription here and now, in this communication situation, and other utter-ances and inscriptions you are likely to have come across. As a hearer-reader I have todo my best to understand you in the light of that same experience. This process doesn’t

require either of us to be in possession of the deliverances of some particular codification

of the language in use, whether private or public.

Most of the time I simply do not have to worry about whether when you say [tEmeI|ou]and I say [tEmAtE¨] or you say [eE(¤)pleIn] and I say [eE¤EpleIn] the dictionary, or the lin-guistic theorist, or anyone else would say they are the same words or different words,although these differences might be communicationally salient and significant in some cir-cumstances. In fact I don’t have to identify the linguistic units either of us use at all. Anymore than the first humans who achieved semiosis by interpreting some vocalisation asnon-iconically meaningful had to identify the linguistic units. They couldn’t have hadto, because there weren’t any. To interpret recognising that two things are similar as amatter of recognising one thing of which both are instances is just that – an interpretation.It is not one that we automatically resort to in other areas of life, nor is it one we need toresort to for communication by means of language.

What is the relevance of all this? First, it eliminates the need to treat cognising languageas something special or sui generis, for instance a la Chomsky. The Chomskyan paradigmis based on an essentially rhetorical argument about how quickly a language is ‘acquired’,i.e. on how quickly we allegedly master the alleged code (contentiously characterised asconsisting at its core of a recursive syntax). It couldn’t be done, so the argument runs,unless we were genetically primed for it. This whole line of inquiry rests on thinking ofa language as a code.

Secondly, my story requires that we conceive of language and its use in a way thataligns it with what distributed cognitionists tell us about the mind and its interaction withthe body and the world. On my story, pace Ross, we do not proceed where language isconcerned by selecting from a mental storehouse the digits or combinations of digits thathave already been prepackaged for us as encoding what we want to say, or by identifyingthe same digits in other people’s utterances and referring to the mental storehouse in orderto decode them. (It may feel like that to some of us, sometimes, or perhaps most of us mostof the time, but that is because we’ve been educated to think about language in that way.)Instead, I suggest, the task is to work out on the hoof what semiotic significance to confer


on certain phenomena (vocal noises, marks on paper, etc.) in order to operate relevantlyon the world in accordance with the requirements of the unique real-time communicationsituation we find ourselves in.

6. The analogical creation of linguistic digits

It can now be seen what was wrong with the initial characterisation of linguistic signal-ling as digital in that the units involved are discrete, arbitrary and systematically recom-binable. The problem is that the units have no antecedent existence, but aremetalinguistic products of the essentially analogical processes involved in linguistic com-munication. This is not to say that there cannot be perceived in any particular utteranceunits that are discrete, arbitrary and systematically recombinable. But it is the communi-cation process itself that confers on them these properties. Suppose I am a learner of Eng-lish unfamiliar with the degree of colloquiality that allows for utterances such as‘[sgE¨na¨]’, ‘[sfaI/fE¤I/]’. To interpret these in some context of utterance it would be help-ful to perceive the similarity between certain elements, e.g. [gE¨], [na¨], [faI/] and otherutterances I have in the past produced or encountered. On that basis I come to be wellplaced to attach an appropriate significance to the initial [s], which thereby becomes a dis-crete, arbitrary and systematically recombinable digit of the system, as is confirmed when Itry out an utterance of my own, e.g. ‘[swÅtStelitEnaI/]’, and get a satisfactory response. Butnone of this implies any conscious process of deploying a unit of the code, let alone enter-taining an analysis of that unit, e.g. as the (well known?) English clitic (prefix?) s�.

We come back to the importance of relevant similarities. ‘Relevant similarities’ hereimplies analogies. Suppose someone says to me ‘there’s a Rock of Gibraltar’, as the Rockof Gibraltar hoves into view, for the first time in my experience, round a bend in the road.On any straightforwardly digital account this is anomalous: everyone knows there is onlyone Rock of Gibraltar; the indefinite article is out of place. Or is it, granted that thespeaker can assume prior familiarity with images of the Rock, and that the experienceof seeing it in reality but from a distance is in certain respects like encountering anothersuch image? What the utterance means, roughly and in part, is ‘there’s the Rock of Gibral-tar, as seen from a viewpoint that makes it look like a familiar representation of itself’. Animportant analogy might be with ‘there’s an Eiffel Tower’ or ‘there’s a Taj Mahal’, asuttered, say, in the presence of plastic models of these and other monuments. In saying‘there’s a Rock of Gibraltar’ one is analogically representing (i.e. re-presenting) otherutterances, parts or fragments of other utterances; one is alluding to or invoking the rel-evant portions of a whole linguistic experience. ‘How do you do?’, ‘hello, how are you?’,‘nice day today’ and other stereotypical examples of phatic communion show this in itspurest form. Such phatic utterances function simply by drawing one’s interlocutor’s atten-tion to his or her experience of other similar utterances.

None of this is to deny that ‘there’s a Rock of Gibraltar’ has here been interpreted as asequence or combination of English digits, or indeed that such an analysis seems cruciallyrelied on in the foregoing account of what it means. The point is simply that as a whole itdoes not bear or yield a meaning antecedently allowed for by the deliverances of any cod-ification of English known to me in advance of being called upon to interpret it. Wherelanguage is concerned we make analogical use of language itself, a process that confersan inherently unstable digitality on the bits of language thus incessantly recycled and


transmuted. And that is why the digits themselves dwell in the shadow world of the virtual,the indeterminate and the perpetually revisable.

References

Amis, K., 1962 [1960]. Take a Girl Like You. Penguin Books, London.Burton-Roberts, N., Carr, P., 1999. On speech and natural language. Language Sciences 21, 371–406.Carr, P., this issue. Internalism, externalism and coding. Language Sciences.Cramer, F., 2001. Digital code and literary text. ArcHive, 4.Dennett, D.C., 1993 [1991]. Consciousness Explained. Penguin Books, London.de Saussure, F., 1922 [1916]. Cours de linguistique generale. Payot, Paris.Firth, J.R., 1957 [1948]. The semantics of linguistic science. In: Firth, J.R. (Ed.), Papers in Linguistics 1934–1951.

Oxford University Press, London.Harris, R., 1981. The Language Myth. Duckworth, London.Harris, R., 1984. Must monkeys mean? In: Harre, R., Reynolds, V. (Eds.), The Meaning of Primate Signals.

Cambridge University Press, Cambridge, pp. 116–137, Reprinted in N. Love (Ed.), The Foundations ofLinguistic Theory: Selected Writings of Roy Harris. Routledge, London, 1990, pp. 158–179.

Harris, R., 1995. Signs of Writing. Routledge, London.Harris, R., 2000. Rethinking Writing. Athlone Press, London.Leiber, J., 1991. Invitation to Cognitive Science. Blackwell, Oxford.Love, N., 2004. Cognition and the language myth. Language Sciences 26, 525–544.Magee, B., 1998 [1997]. Confessions of a Philosopher: A Journey Through Western Philosophy. Phoenix,

London.Menary, R., this issue. Writing as thinking. Language Sciences. doi:10.1016/j.langsci.2007.01.005.Moore, T., Carling, C., 1988. The Limitations of Language. Macmillan Press, London.Port, R., Leary, A.P., 2005. Against formal phonology. Language 81, 927–964.Ross, D., 1993. Metaphor, Meaning and Cognition. Peter Lang Publishing, New York.Ross, D., 2004. Metalinguistic signalling for coordination amongst social agents. Language Sciences 26, 621–642.Ross, D., this issue. H. sapiens as ecologically special: what does language contribute? Language Sciences.

doi:10.1016/j.langsci.2006.12.008.Searle, J., 1979. Metaphor. In: Ortony, A. (Ed.), Metaphor and Thought. Cambridge University Press,

Cambridge, pp. 92–123.Whitehead, A.N., 1917. The Organisation of Thought: Educational and Scientific. Williams & Norgate, London.

http://dx.doi.org/10.1016/j.langsci.2007.01.005

http://dx.doi.org/10.1016/j.langsci.2006.12.008

Documents

Are languages digital codes?