Theoretical Grounding for Computer Assisted Expert Text reading (CASTR)

Jean Guy Meunier, Pierre Poirier, Jean Danis, Nicolas Payette


Digital humanities’ technology has mainly focused its development on scholarly text digitalization and text analysis . It is only recently that attention has been payed to the activity of reading in a computerized environment. Some main causes of this have been the advent of the Ebook but more importantly the massive enterprise of text digitalization .(Gallica, Google Books, World Wide library, etc.)

In this new environment, the classical scholarly reading of text (be they manuscripts or codex) is radically transformed. The practices that scholars had slowly developed, refined and consolidated throughout the centuries and disciplinary constraints are suddenly confronted to the new types of digital text presentations, scrolling, visualisations, hypertexting, networking, etc., and maybe more importantly, to new modes of parsing, commenting, annotating, synthesizing and, finally, writing and dissemination. Even the solitude of text meditation is now broken by the ocean of related texts, interactive comments, social scholarly interventions, etc. In order to assist in this transformed type of reading, some digital technologies have emerged. Among them, one can find QDAMiner In Vivo, Atlas and more recently Pliny. We are also involved in this type of technology (Danis, Payette and Meunier, 2008).

In this paper, we analyze, in a very exploratory manner, three main dimensions of computer assister scholarly reading of text (CASROT) : the cognitive, the computational and the software dimension. The cognitive dimension of scholarly reading pertains not the nature of reading as a psychological activity but to the complex interpretative act of going through argumentations, narrations, descriptions, demonstrations, dialogs, themes, etc. that are contained in a text. This interpretative activity is not in itself a rule governed behaviour   albeit a text may present some strong regularities: a genre (Rastier, 2001), a rhetorical structure (Mann and Thomson, 1988), a logical structure (Hobbs, 1990, Bringsjord and Zenzen, 2003). It is closer to what the hermeneutic tradition call the “understanding“ (verstehen) of a text : A cognitive process that is rooted in linguistic, semantic, inferential, pragmatic, cultural, social, structures, etc. And even after this as been dug into, the text does not reveal itself in totality. A new reading is always possible. Every text is polyvalent. It varies its meaning on reception (Jaus, 1978) and reader (Eco, 1965) and does not necessarily reveal itself in a first reading or even in a first analysis (e.g. the Bible or the Koran). The hermeneutic tradition has been repeating this for at least the last century (Schleiermacher, 1987; Heidegger, 1962; Gadamer, 1996; Habermas, 1984, 1987). Contemporary theories of reading adhere to this thesis (Iser, 1985; Fish, 1980; Barthes 1970 Gervais 1993) and a technology that wishes to assist a scholarly reading must respect its complex cyclic nature.

The computational dimension is even more challenging. As the hermeneutic tradition insists, reading is a kind of dialog with a text. The content changes throughout the interaction with the reader. More so, the reader itself changes his own  background knowledge and attitude as he travels through the text

This interaction implies a modification of computational models of text reading. A computational model cannot be a sort of well formedness or pattern recognition process, or a kind of algorithmic extraction of “content” procedure. This particular interaction goes in hand with new reading strategies that are complementary to the typical linear reading process.( quantitative reading/hypertextual reading etc.) Technicaly, this imply computational models that allow dynamic interactions between micro-textual reading and global structuring  processes. All of this interaction should at the same time allow interpretation based on a changing structures that are constantly modified throughout reading process.   

Because of the constant interventions of the reader in the core of the text by annotating commenting, rewriting, social linking, etc., the computation itself has become interactive computation (Van Leeuwen & Wiederman, 2000) or what now has been called hypercomputation. (Clelland 199x, Bringjord & Zenzen 2003, Syropoulos 2008) A technology that wishes to assist this interaction must innovate on the computational level of the technology.

Finally, the software dimension pertains to the material computer implementation of these cyclic and interactive properties of scholarly reading. The cyclic dimension requires high flexibility in the myriad of functions. The interactive dimension opens up on constant introduction of new layers to the text and even more so with the social intervention of other readers.


reading, textual analysis, professional reading environments

Full Text:


Comments on this article

View all comments

New Knowledge Environments
© University of Victoria