Vous êtes connectés en tant qu'administrateur. Cette page est en cache jusqu'à Fri, 20 May 2022 20:41:35 GMT. Prévisualisez la dernière m-à-j en cliquant Rafraîchir.

The Language that Machines Read

3:45 PM, mardi 14 août 2018 (1 heure 15 minutes)

We take it for granted that machines read. Computers have read in and out put since the earliest days of post-Turing computation. But when we say that machines read, this unambiguously begs the question of what it is that they read and leaves unexamined the even more fundamental question of what reading is, as a process, as a behavior. Reading is constitutive of language. This statement means that something called reading brings language into being. As far as we know, language is something that only humans have and thus reading, in this sense, must be something that only humans do. The contentious contradictions entangled in these statements will be reconfigured by the proposal that what machines read is what Walter Ong would call a grapholect, "a transdialect language formed by deep commitment to writing." In fact, they read encoded transcriptions of grapholectic languages (plural, if we do not constrain our attention to the global

English grapholect). This reading is not the type of reading that humans do and, ultimately, we might not want to recognize it as reading at all. It is a distinctly différant (the Derridean form is deliberate and will be elaborated) sub-dialectic practice of reading that, at best, may help to constitute a linguistic grapholect. These circumstances and their current configuration – under the regime of computation and in the context of predominant vectoralist instrumentalism and solutionism – must be analysed and therapeutically addressed before we can move on to entertain the possibility that machine reading, as we know it, could be productively and generatively deployed in order to deform, reform, transform or otherwise reconfigure the texts of digital language arts or digital humanities practices.

Brown University