Starting from a critique of Jakobson's model of communication on the grounds of intentionality and decontextualisation, I show how a revised version of the model by Jean-Jacques Lecercle addresses these problems but is still largely a print-based literary model. For our purposes it doesn't take into account the digital apparatus - computer hardware and software – nor the performance required to access and "realise" the digital text. I suggest a bi-planar model for digital literature which superposes a digital text model superimposed over the literary. But for this to work as a model the two planes need to connect. This connection I argue is provided by the concept of performativity. As a first run at testing this model, I examine how performativity integrates the natural language of the literary text with the artificial language of the digital apparatus. For this to happen first requires an exposition of integrational linguistics and speech act theory, both of which view language as profoundly performative. It is a short step from there to the performativity of code. However, although this may be a successful move, it may be that a more dynamic, less fixed, model of digital text is required; one that is itself digital.

When future generations are quite accustomed to sitting at a keyboard and "typing" an audio-visual product that incorporates sounds, letter-forms and pictures systematically interrelated, they will have acquired a new concept of writing, a new concept of literature, a new concept of language.1

With this paper I am returning to familiar ground in an attempt to further clarify and establish a central role for performativity in relation to digital literature. This paper develops the ideas outlined at the Chercher le texte/ELO Conference in Paris in September 2013 in combination with a paper delivered at the Digital Materialities Conference organised at the University of Coimbra in May 2015. Both papers begin with the question of whether digital literature needs a theory of language, and if so, what it might look like. Or another way of putting this is, does digital literature presuppose a theory of language and if so, which one? The ultimate goal of this endeavour is to suggest an integrated model of digital literature. This might be described, following Deleuze, as a "collective assemblage of enunciation" The aim then is to give a full account of all the elements of this "collective assemblage"; to place as many of the disparate elements of digital literature as possible under a single concept – that of performativity. There is a particular risk involved in this sort of totalising project, namely that in its drive for coherence, it "… unavoidably exercises a kind of conceptual violence with regard to the abundance of phenomena2". Whether or not this is the case will be dealt with at a later date. Before we can decide that we need to see if a case for coherence can be made.

As a starting point for this investigation, I am going to refer to Jakobson's standard communication model. My guess is that many, consciously or unconsciously, apply this model to the reading of a print text, where the author stands in for the addresser/speaker, the text for the message and the reader for the addressee/speaker.This is a basic encoding/decoding model where language or text acts as a simple "conduit" between a mental event in the mind of the writer and a corresponding mental event in the mind of a reader, such that the former is reproduced in the latter. This is also known as a telementational model of communication, one which, as Roy Harris puts it, "… guarantees semantic invariance as between speaker and hearer". This model can be criticised on a number of levels – fixed code, single author, author/reader distinction and intentionality – and I will come back to some of these criticisms later on in the paper.

In literary/writerly terms, Jakobson's is very much a model based on the figure of the single author of a print text communicating with a clearly defined reader. As such the model also raises a number of questions around the problem of intention. In response to these shortcomings, I would begin by adjusting the model in the manner suggested by Lecercle in his Interpretation as Pragmatics3. The changes are subtle but highly significant. In this revised model, the central position of the writer as begetter of the text is replaced by the centrality of the text which 'interpellates' the writer and the reader into subject positions.Lecercle's revised model fits much better with the mode of digital writing where the clear-cut roles of writer and reader/sender and receiver have collapsed, and where there is now a role for various contextual elements under his designation of 'encyclopaedia'. The revision also deals effectively with the problem of intentionality by sidelining it without ignoring it. Now intentionality resides in the text itself rather than in the mind of the author. In other words, "the text is intended to be this sort of text" rather than "the author intended this sort of text". Finally this revised model moves away from the telementational model. The text no longer simply acts as a transparent conduit by which the reader supposedly has direct access to mental events (whatever they might be) in the mind of the author via a process of encoding and decoding.

The most significant shortcoming of this revised model from the point of view of digital writing is that is still largely a model based on the print text. What it's not capable of giving is an account of the temporality, dispersion and performativity of the eventilised digital text. This essentially boils down to simple questions of time and space. Given the central role of temporality in digital writing – a position argued most convincingly be N. Katherine Hayles among others – where does time appear in this model? How does it account for what Hayles refers to as the "…the time of performance for an electronic text versus the time of production for print4" (Her italics). Secondly, how does it account for the dispersed and multi-site nature of digital text? Digital text is displayed across a variety of platforms, in different spaces. Each of these make it perform differently each time it is accessed. Its materiality shifts and changes. This sense of mutability should be recognized and written into the model in some form. Thirdly, where is the machine in all this? When it comes to e-lit/digital text we cannot simply ignore the computer, especially its analog components, as a medium for writing and publication in the way that the literary might choose to ignore the substrate of the book.

One possible solution to some of these questions might be to create a second, parallel layer to the model, which would look something like this:In this new model, it is the machine, with its hardware and software components, that occupies the central position and calls the writer and reader into subject position, and in this case, "author" would refer not simply to the writer of the final "literary" text, but to any of the programmers and coders who write the scripts for the machine itself. It is also important to make the point that these positions refer to "sites", so to speak, rather than individuals, and there is no reason why the same individual should not occupy each site simultaneously. This is often true of digital text where the functions of reader and writer (or "maker") of the text overlap. In order to maintain a distinction between the elements of the two models, the second layer would have to be transposed over the first, in effect turning it into a three dimensional model. This is a double articulation, the model becomes bi-planar, as below.Obviously, at this stage the fit is not ideal. The relationship between the contextual notion of "encyclopaedia" and that of "hardware" is by no means evident, unless one is dealing with a highly expanded notion of what constitutes hardware. However, this does draw attention to another, more general question with regard to this bi-planar model: what is the relationship between the first layer and the second? In other words, how are the connections between the two layers of the model formed and maintained? If this is to function effectively as an integrated model, there has to be a high level of connectivity and coherence.

Here I want to argue that the connection between the two layers of the model is made via the concept of performativity (or perhaps 'mediation' as it is articulated by Alexander Galloway5, although Galloway's notion is itself shot through with performativity) and furthermore, that this performativity is a vital component of any model of digital literature.

One way of articulating this claim is to give an account not of whether, but in what way performativity links together the respective elements of the bi-planar model. The task then would be to show how performativity works as a 'linking' device. For our purposes here, I am going to focus on the 'language' panels in both the layers of the model. In other words, how does performativity link the (natural) language of text with the (artificial) language of the machine – at the same time recognising that this natural/artificial dichotomy is not a hard and fast one?
In order to make this argument I will outline a specific theory of digital language, one which is rooted in pragmatics and amalgamates some of the central tenets of integrational linguistics with those of speech act theory. This is an unorthodox approach, but one which is coherent and, I think, particularly relevant to digital literature, especially with regard to the notion of and need for integration.

So, let's begin with a quick survey of the conceptual landscape I am operating within. Integrational linguistics was developed mainly in the 1980s by a group headed by Roy Harris. Its foundational text is his The Language Myth in which he sets out a critique of standard orthodox linguistics based on what he sees as two fundamental fallacies of which have underpinned practically all mainstream thinking about language in the Western tradition. The two fallacies go under the names of the "fixed-code fallacy" and the "fallacy of telementation". The first of these fallacies erroneously, according to Harris, views language as a set of internally structured systems whose units are invariable and regular (i.e. grammatical) and that as a consequence there exists such a thing as language which is susceptible to scientific analysis. This mistaken view of language is at the heart of the Saussurean notion of 'langue' and Chomsky's notion of 'competence'. From this emerges the idea of telementation – the second fallacy. Knowledge of these fixed codes, which are theoretically shared across members of a language community, means that members of that community can neatly transfer the contents of their mind as mental acts coded in language to the mind of other individuals. This is precisely the basis of the Jakobson model referred to above.

In opposition to these two 'myths', integrationism understands the linguistic sign as radically indeterminate. In other words, signs do not exist as decontextualised items with set meanings which pre-exist their instantiation (performance) in a communication. There is no such thing as a decontextualised sign. There is no text without a context, and signs do not bring their invariant ready-made meanings with them. Rather they are made and remade only in the communicational time-based episodes in which they are performed. Furthermore, there is no cast-iron guarantee that those involved in the communicational context will recognise the same signs as occurring since all contextualisation is individual. This approach calls into question the whole basis of the telementational model. For the integrationist, there can be no intersubjective sharing of fixed meanings.
For Harris a number of consequences flow from this; primarily, the rejection of the Saussurean notion of the linguistic sign as a biplanar entity which links a determinate form (signifiant) with a determinate meaning (signifié) – a notion which underpins nearly all mainstream approaches in modern linguistics (structuralist, poststructuralist, generativist etc.).
Further, Harris questions other fundamental assumptions, namely: "(i) that the linguistic sign is arbitrary; (ii) that the linguistic sign is linear; (iii) that words have meanings; (iv) that grammar has rules; and (v) that there are such things as languages6". For the integrationist, as Jon Orman puts it, communication involves '... the integration of human activities for which signs (linguistic or otherwise) act as interfaces' within a highly contextualised temporal flow. Language is made through performance. With regard to (v) Harris's position is that "languages" do not exist as first-order phenomena, but as second-order constructions which are an amalgamation of the linguistic behavior of a particular language group and as such are constantly subject to change7". In other words language is highly performative.

A second component of my performative view of natural language is based originally on the work of ordinary language philosopher, J. L. Austin. Writing in the 1950s, Austin developed a theory of language in his book How to Do Things with Words8, which began to undermine the predominance of truth-conditional semantics. Whereas much of the early 20th century "turn to language" in philosophy had focused on questions of meaning and truth, Austin made the point that many utterances in language have nothing to do with truth and falsity. They don"t state or describe; they do things. In saying something, he argued, you are doing something. A clear example of this is the uttering of a wedding vow or the formula used in launching of a ship. These locutions Austin referred to as "performatives". In terms of analysis, questions revolve not so much around whether a statement is true or false, but rather whether it is 'felicitous' and 'infelicitous', terms which Austin used to designate the extent to which a performative is successful in achieving what it was intended to do. Austin suggests a tri-partite description of performatives; the locutionary act (the particular form of a spoken utterance together with gestures, body language, etc.) the illocutionary act (an indication of what sort of utterance is performed, ie warning, suggestion, order, bet, etc.) and the perlocutionary act (the consequences which flow from the uttering of the perfomative). Allied to this is a notion of a 'force' which is attached to each of these utterances.

This work was later expanded upon and developed by philosophers such as John Searle in their work on speech act theory, such that any utterance, indeed language itself, has to be understood as performative in some sense. Central to this performative understanding of language again is context. For the speech act theorist, as for the integrationist, context is not something which simply modifies the already given meaning of an utterance. Context establishes its meaning in the moment of performative communication.

At the risk of adding confusion, it might be worth adding another term to this discussion which might be used in our model as a substitute for "language", and that is "languaging". This is how linguists J. Normann Jørgensen and Kasper Juffermans provide a gloss on this term. They write: "A languaging perspective sees language in actual practice not as bounded, countable entities that are given in the world, but as dynamic, creative potential to produce meaning through the use of arbitrary signs. A languaging perspective conceptualizes language as a verb (as practice or behavior), rather than as a noun (a thing or object) and places the activity and the agents (languagers) in focus rather than the linguistic system (languages)9".

Alongside Harris, they see the conventional view of languages as no more than socio-cultural abstractions, rather than the naturally occurring phenomena that conventional linguistics assumes them to be. By contrast, we are positing a view of the language here as a time-based, highly contextualised performance of a collective assemblage of signs, of which the linguistic is the primary but by no means the only sign. And in fact, this view fits well with our experience of digital text. It would seem to resonate with textual variations across browsers, animated poetry, written and recorded voice hybrids, scrolling text, digital text with performance, digitally projected text, indeed with the very display of language on an 'eventilised' electronic screen. From this perspective the performative linguistics of the language of digital text is not aberrant or avant garde, but a manifestation, or a paradigm, of how natural language actually functions from an integrationist point of view.

If we accept the integrationist view of natural language as highly contextualized and performative, then the follow-up question is, how might the performativity of the natural language of the creative digital text link to the artificial language of code and programming? The answer would seem to be very straightforward. Active codes perform the digital text, or rather allow the digital text to be performed. The programming language causes events, linguistic and otherwise, to take place. It is by its very nature performative. We don't ask of computer code what does it mean. We ask what does it do. This is the central question that Austin asked. And this is the argument of Geoff Cox in his book, Speaking Code. Cox argues that code in particular is a form of speech and in this he makes explicit use of John Searle's speech act theory10 and, by implication, Austin's theory of performatives. More importantly he does not separate out artificial and natural languages, but rather calls for a mode of analysis which would show how they interpenetrate each other and act integrationally. He writes:

Once code is likened to speech, it also provides the possibility of new forms of criticism that combine natural and artificial languages into new speech acts.11

This idea of integration mentioned here brings us back to Harris's understanding of language and the central concern of my work.

At this stage, this paper is little more than a surface-scratching exercise. Greater time and space would allow for an analysis of the other areas of the model in relation to performativity in such a way that, rather than seeing performativity in terms of linkage between the specific areas of the model, it may be simpler to think about it as the environment in which the whole apparatus of digital writing (both as process and output) functions. Performativity permeates the whole assemblage and integrates the disparate elements which constitute this complex digital writing environment.

Any future development of the model would have to deal with some immediate objections however. Firstly there is the nature of the model itself. It may be bi-planar and apparently three-dimensional, but it is still very much static, pinned to the two-dimensional surface of the page. Why would we want to represent a performative process in anything other than a dynamic form? Shouldn"t the digital affordances of the computer be used to model the complex system of digital literature? If the major element of this model is the insertion of temporality, should the model itself not be in motion, susceptible to change over time? And to what extent can an explanation of the behavior of the parts be sufficient to explain the behavior of the whole? This in turn raises the question of whether, where digital literature is concerned, the standard discourse of the "literary" should be replaced and/or augmented by the vocabulary of systems dynamics, - forces, direction, flows, feedback and temporality – in order to understand the non-linear behavior of complex systems.

A second objection to this extended performative model would be simply "where does it get us in literary analytical terms?" As with any performative text, it is not enough for digital literature to concern itself simply with the words, with the language of the text. Given this insistence on mutability, dispersal, performance, context, indeterminacy and time-based media, how can we say anything meaningful about the literary quality of digital text work? Traditionally, literary/linguistic studies deal with literary texts sub specie aeternitatis, by attempting to squeeze time out of a linguistic structure, thereby allowing some sort of immutable truth to be expressed "…in the transitory vehicle of words12". Where digital literature is concerned, however, it would seem to demand that we all become Heraclitans. Just as you can never step into the same river twice, is it not also the case that you can never read the same piece of digital writing twice? This is close to the position that Socrates arrives at in the Cratylus if one accepts the fundamental indeterminacy of language:

But we cannot even say that there is any knowledge, if all things are changing and nothing remains fixed; for […] if the very essence of knowledge changes, at the moment of the change to another essence of knowledge there would be no knowledge, and if it is always changing, there will always be no knowledge, and by this reasoning there will be neither anyone to know nor anything to be known.13

One solution to this is to somehow extract the literary text from its context, from its embedding in a complex digital assemblage, and treat it as a separate element – and I have often seen this done especially in relation to a "close reading" of digital writing. But in these cases of course what we are reading is no longer digital literature; it is print literature masquerading as digital literature. Proffering answers to these questions will certainly require more research and possibly a complete change of direction.

Notes

1. 1. Harris, Roy. Rethinking Writing London. Athlone. 2000. p. 242.back

2. 2. Krämer, Sybille. Medium, Messenger, Transmission: An Approach to Media Philosophy Amsterdam. University of Amsterdam Press. 2015. p. 165.back

3. 3. Lecercle, Jean-Jacques. Interpretation as Pragmatics London. Macmillan. 1999.back

4. 4. Hayles, N. Katherine "The Time of Digital Poetry: From Object to Event" in New Media Poetics. Contexts, Technotexts, and Theories, Eds. Morris A. and Swiss T, MIT Press. 2006. p. 185.back

5. 5. Galloway, Alexander. The Interface Effect Cambridge. Polity. 2012.back

6. 6. Harris, Roy. The Language Myth London. Duckworth. 1990. p. 45.back

7. 7. Orman, Jon. "Linguistic Diversity and Language Loss: A View from Integrational Linguistics" Published online at Language Sciences June 2013. 2013. Viewed at https://www.academia.edu/3487135/Linguistic_diversity_and_language_loss_a_view_from_integrational_linguisticsback

8. 8. Austin, John Langshaw. How to do Things with Words Cambridge, Mass. Harvard University Press. 1975.back

9. 9. Jørgensen, Norman and Juffermans, Kasper. "Languaging" Viewed online the 03.09.2015 at https://www.academia.edu/1808400/Languagingback

10. 10. Searle, John. Speech Acts. An Essay in the Philosophy of Language. Cambridge. CUP. 1969.back

11. 11. Cox, Geoff. Speaking Code. Coding as Aesthetic and Political Expression. Cambridge, Mass. MIT Press. 2012. p. 14.back

12. 12. Harris, Roy. http://www.royharrisonline.com/INP26.html. 2010. Accessed August 2015back

13. 13. Plato. Cratylus Harmondsworth. Penguin Classics. 1972. p. 85.back