Literature, Talk, and Computing

With the advent of computing, the role and function of talk, writing, and literature has changed.

Historically, writing has been a key inscription and storage technology across many cultures. With the “bottleneck of the signifier” broken, writing is no longer the dominant inscription technology.¹ Instead it is the “universal machine,” with its binary logic and symbol manipulation, that forms perhaps the most significant inscription and storage technology of our time. Clearly this has repercussions – though not necessarily negative – for the ways in which writing and literature, its aesthetic complement, are conceived and function.

This shifting media ecology also affects perceptions and uses of talk. Firstly talk itself is newly capturable. While the tape recorder revolutionized the inscription of talk in the 1960s,² digital recording and effective speech recognition software have rendered it capturable and manipulatable across multiple media modes and for multiple outputs: sound, writing, code. Talk is no longer ephemeral.

Secondly, this newfound inscriptibility upends many of talk’s long-standing cultural meanings. As a technology of orality, talk sits on the lower rung of the oral/written hierarchy that has dominated much discussion on the topic in Western cultures. The changing affordances of talk disrupt these traditional binaries between writing and orality, in productive, if often unpredictable ways.

One of the more pervasive, if surprising, disruptions surrounds the perceived relationship between writing and orality in digital environments. We do not live in a postwriting media ecosystem; writing is still a pervasive inscription technology and continues to mediate talk in many settings. However, there has been a change in the relative status of writing and orality as it is orality, rather than writing, that is held to be the model for digital communication. Our communications via email, chat applications, comment boards, and alike continue to utilize writing, but they often draw on orality as the standard by which these communications are measured.

In addition to this changed status, there has also been a growth in the deployment of talk, orality, and conversation as the means by which humans interact with software, as Talking Interfaces demonstrates. Talk’s perceived affordances clearly have newfound appeal in digital contexts: whether it’s informality, immediacy, interactivity, or even its ephemerality. This appeal might be based on the assumption that talk is less mediated than writing – a talking interface as being one step further towards the holy grail of brain-machine interfacing (where the computer can “read your thoughts”). For Sherry Turkle meanwhile the appeal of online talk is a distraction from the profoundly adversarial relationship that exists between digital technologies and “open-ended and spontaneous” conversation.³ In either case, this assumption that talk is less mediated than writing is worth unpicking. Talking interfaces do constrain some forms of talk, as Turkle worries; however they also afford others. Relatedly, face-to-face conversation itself affords (as Turkle notes) and constrains certain interactions, undermining any simple digital/analog binary. If we are living in a more talkative time (a return perhaps to the eighteenth century?), we would do well to scrutinize the forms and protocols that shape said talk and reflect on the cultural implications of this shift.

One way into thinking about these issues is to place talk’s mediation into a longer historical narrative. To do so, I turn to literature. While potentially existing in an antithetical relationship (writing versus orality), literature has long been a key means for mediating talk. This is a point that I have explored in depth in my research into interviews and the ways in which literary culture has conceived of and utilized the practice and form.⁴

I turn to literature here for a couple of reasons. Firstly, writers have long represented talk in their work; literature thus offers us a useful if not exclusive corpus of talk and its protocols across time and cultures for study. Further, the act of representing talk has prompted many writers to reflect on the specific function of writing and literature. Talk becomes a useful foil, encouraging writers to reflect on the mediative, communicative, and interactive aspects of reading and writing. Talk’s appearance in literature is often therefore overdetermined and suggestive for our purposes, helping to elucidate the long-term perceived affordances and constraints of talk and thus position contemporary Talking Interfaces within their historical trajectory.

My second reason concerns methodology. I am by training a literary critic and seek to bring the tools and questions of literary criticism to bear on an unorthodox object, the talking interface, in order to enrich our understanding of the communicative forms that surround us and contribute to their critique. I follow (at a distance!) the many path-breaking humanities scholars who have brought their insights to the field of computing. In so doing, this project attends to those situations, texts, and case studies in which literature, computing, and talk coalesce.

The choice of material presented under the banner of “Literature, Talk, and Computing” varies in format and aim. Two biographical case studies follow.

The first focuses on British linguist and natural language processing (NLP)/machine translation pioneer Margaret Masterman. Wrestling with how to build translation models based around semantics, Masterman experimented with using computers to write poetry. For Masterman, poetry writing offered a means by which to garner insight into natural language and develop appropriate processing models. The fact that Siri does not deploy Masterman’s insights today indicates the degree to which the industry, for so long, sidelined her contributions .

The second example concerns the author J. M. Coetzee and provides an intellectual history of his thinking around literature, talk, and computing. Before gaining global renown as a writer, the South African embarked upon a career in programming in the 1960s and completed a PhD in early humanities computing. Later, and across his illustrious career as a writer, Coetzee has engaged deeply with the question of what literature might do in an age of computation. In particular, he has emphasized interactivity in his understanding of both literature and programming: conversation forms a topic in a number of his works as he seeks to present the experience of reading literature as resistive in an age governed by what he sees as the binary YES-NO logic of digital tech.⁵

Both vignettes demonstrate the work that biographical narratives – in particular those of women and citizens of countries other than Britain and the US – can perform. Biography is a foundational form of life writing, but one that exists in an uneven relationship to literary studies where it is both widely consulted (even after the denigrations of New Criticism) and often undervalued as a form of academic labor. Yet as the work of scholars such Janet Abbate, Mar Hicks, Julianne Nyhan, and others have vitally demonstrated through their work collecting oral histories of women's contributions to the early history of computing, life writing can perform critical interventions.^6,7,8 These two pieces of interdisciplinary writing aim to contribute to wider attempts to expand computer history both along gender lines and into the Global South.

If these examples explore wider questions surrounding communication, literature, and computation, my discussion of literary experimentation with chatbots is much more explicit in its engagement with talking interfaces. I further demonstrate what literary studies might offer to the analysis of chatbots as a sociocultural phenomenon. Here methods, tools, and sensitivities from literary studies offer ways of reflecting more generally on the cultural import of language modeling that renders some talk executable and some not; some talk interfaces accessible and others less so.

Additionally, I present A Literary Guide to Natural Language Processing. An act of interdisciplinary explication or pedagogical explanation, the section attempts to demystify NLP for the uninitiated. The section also reflects, in the process, on how literary studies and computing might productively interact with each other to develop their own field specific analysis: talking interfaces as enabling creative interdisciplinary discussion.

Where to now?

Biographical Case Study: J. M. Coetzee

Biographical Case Study: Margaret Masterman

A Literary Guide to Natural Language Processing

Talking Interfaces