Combining Cognitive Stylistics and Computational Stylistics

Studies in the computational analysis of texts have been successful in distinguishing between authors and in linking anonymously published texts with their authors but computational tools have yet to be accepted as mainstream techniques in literary analysis. Criticisms are generally centred around the belief that computational analyses make false claims about scientific objectivity and are in fact no less subjective than any other critical approach. This is perhaps because computational projects are in conflict, at a fundamental level, with contemporary post-structuralist notions of subjectivity, meaning and the arbitrary nature of language. This paper will argue
that these objections rest on assumptions about language that need to be examined in light of developments in
linguistics and cognitive psychology and that cognitive linguistics has the potential to bring a more interpretive framework to computational stylistics, a practice that has traditionally been applied in fairly narrow, empirical way.
Whilst computational analysis points to the possibility of subjectivity that is more coherent than some theoretical approaches imply, it does not necessarily diminish the role of culture and context in the formation of texts and subjectivity as highlighted by materialist readings. The application of cognitive linguistics in a computational study provides a model of syntax and semantics which is not independent of context but deeply bound up in context. Cognitive linguistics can explain the existence of computational results in a way that Saussurean based theories can not. It can offer a rich interpretive model that does not neglect the importance of author, reader, or context through its approach to language and literature as an expression of an innately constrained and embodied human mind.
Computational stylistics, particularly in studies of
attribution, generally makes use of function words in order to distinguish between texts. As Craig (2004)
explains, independent variables like genre, are compared
with counts of internal features, or dependent variables, like function words. Correlation of these two kinds of variables is the primary tool of computational stylistics (275-76). Critics of stylistics, most notably Stanley Fish, tend to privilege individual instances of particular words
in interpretive communities over more general rules, and question the validity of the stylistic project. From a cognitive perspective, however, language possesses
universal features because it emerges from the interaction
of “inherent and experiential factors” that are “physical, biological, behavioural, psychological, social, cultural and communicative” (Langacker 1). Langacker claims that each language “represents a unique adaptation to common constraints and pressures as well as to the
peculiarities of its own circumstances” (1). Computational
stylistics of the kind undertaken in this study provides
us with evidence of the peculiarities and creative
adaptations of an individual user, and also highlights more general trends which can be used for comparative purposes.
Our attitude to what we can say about a text depends
largely on our account of language. Widely shared
post-structuralist assumptions about language and
indeterminacy have contributed to the lukewarm reception
of computational stylistics in literary interpretation. In cognitive linguistics “Semantics is constrained by our models of ourselves and our worlds. We have models of up and down that are based on the way our bodies actually
function. Once the word “up” is given its meaning
relative to our experience with gravity, it is not free to “slip” into its opposite. “Up” means up and not down” (Turner 7). Cognitive stylistics views a text as the product of a human action and it therefore carries the mark of that
action. The cognitive belief that language and conventional thought emerge from “our perception of a self within a body as it interacts with an environment” suggests that meaning is somewhat constrained and that “some form of agency is fundamental to language” (Crane 22).
The idea of authorial agency is one that is rejected by structuralist and post-structuralist critics. In proclaiming the death of the author Barthes suggests that the text becomes an ‘open sea’, a space of ‘manifestly relative
significations, no longer tricked out in the colors of an eternal nature’ (Barthes 170). The notion of the
“transcendental signified” rejected by post-structuralist critiques is not, however, the notion of agency proposed
by cognitive stylistics. The view of the “Author”
rejected by Barthes, Derrida and Foucault is as, Seán Burke explains “a metaphysical abstraction, a Platonic type, a fiction of the absolute” (27). Cognitive stylistics tends not to deal in absolutes.
Stylistics is one way of getting evidence and making sense of texts as human actions. Through its approach
to thought and language, cognitive stylistics points to issues that are of concern to scholars of literature, such
as “subject formation, language acquisition, agency and rhetoricity” (Richardson 157). Cognitive philosophy
claims that the mind is embodied and that concepts are therefore created “as a result of the way the brain and body are structured and the way they function in
interpersonal relations and in the physical world” (Lakoff and Johnson 37). The links between the brain and the body mean an objective reality is impossible given the role our sensorimotor system plays in perception. But
as Lakoff and Johnson explain, it is our sensorimotor system’s role in shaping conceptual systems that keeps these systems in touch with the world (44).
The embodied cognitivism of Lakoff and Johnson, also known as second generation cognitivism, argues that our access to the external world is mediated through cognitive processes. Cognitivism provides a framework in which
we can still legitimately engage with psychoanalytic
interpretations, gender focused readings, and the material
conditions of production while using computational and cognitive techniques of analysis. Computers enable us to draw together instances of common forms, and other
features of a text, in a way that would be simply impossible to an individual human reader. Cognitive linguistics provides
a theoretical justification for paying attention to common
forms in the first place, and reveals a way in which the features highlighted by computational approaches can contribute something of value to traditional literary
analysis.
References
Barthes, Roland. On Racine. Trans. Richard Howard. New York: Octagon Books, 1977.
Burke, Seán. The Death and Return of the Author: Criticism and Subjectivity in Barthes, Foucault and Derrida. Edinburgh: Edinburgh University Press, 1988.
Craig, Hugh. “Stylistic Analysis and Authorship
Studies.” A Companion to Digital Humanities. Eds. Susan Schreibman, Ray Siemens and John Unsworth: Blackwell, 2004. 273-88.
Crane, Mary Thomas. Shakespeare’s Brain: Reading
with Cognitive Theory. Princeton and Oxford:
Princeton University Press, 2001.
Lakoff, George, and Mark Johnson. Philosophy in the Flesh: The Embodied Mind and Its Challenge to Western Thought. New York: Basic Books, 1999.
Langacker, Ronald W. Foundations of Cognitive
Grammar: Descriptive Applications. Vol. II. Stanford, California: Stanford University Press, 1991.
Richardson, Alan. “Cognitive Science and the Future of Literary Studies.” Philosophy and Literature 23.1 (1999): 157-73.Turner, Mark. Death Is the Mother of Beauty. Chicago: University of Chicago Press, 1987.