1.3 Computers and Critics 7
1.3.1 View of the TextA computer's view of a text is quite different from that of a human being. For the human the text exists on at least three distinct levels: the medium - usually ink marks on paper, the signifier - the character or letter 'A', and the signified - the meaning 'A'. While we normally work in larger groupings of characters - words, sentences, and paragraphs - we can see the clear distinction between the different levels if this is required. The computer can not; its total 'awareness' of the text is limited to its internal representation of the text as a sequence of codes. Because of this, the notion of signified is missing from the computer's awareness. For it, everything collapses onto the level of the signifier; all concepts of 'meaning' must be supplied by the researcher.
This unavoidable simplification is not as limiting as it first appears. Because of the generality of the computer, multiple sequences of other symbols may be encoded together with the text. These are most often drawn from a number of categories, supplying information that is implicit for us in the encoded text, but which the computer cannot detect. These categories can be encoded in symbols either as a parallel sequence to the text, or as a separate table built up from it. In either case, the text can now be viewed as a list of words with a variety of information encoded for each word. Some of this information might describe whether the word belonged to any of the categories of, for instance, image vs. non-image, descriptor vs. functor, and so on.
Other information might be position within the line, line number, page number, chapter, grammatical type, and more. The use of parallel sequences can help the computer to differentiate between words that are the same in one or more of the sequences. For instance, the category 'grammatical type' would distinguish the sequence of characters 'b o w' - the noun, from 'b o w' - the action. In this way, the text can be converted to a hierarchical structure, where each unit is successively defined in terms of lower units, ranging from chapters to characters. Under this system "concepts of form, structure, and meaning relate to patterns along, across, and among these various strata" 8 .It is only in this way that any concept of 'meaning' can be analyzed by the computer, and even then the meaning that is analyzed has been supplied by the critic.
These very concepts of materiality of text and primacy of category to define and characterize form are central to several formalist schools of criticism - among them the Russian Formalists and their second generation, the Prague School. To quote Smith: "at the level of word or figure, the Russians placed considerable emphasis on liberating the word from its fixed conventional connotations so that its full richness could be seen" 9 The concept of category also plays a varied and pervasive role, although it is not always immediately apparent. As an example, take Vladimir Propp's Morphology of the Folk Tale 10 , one of the more successful large scale projects. Propp reduced the plots of some 479 fairy tales to sequences of basic actions, or functions, revealing their universal narrative structure. As his functions represent actions described in the narrative by configurations of words and phrases, they can be viewed as a higher level category of the text, with themes as a still higher level composed of sequences of functions. Similar to this study is Tzvetan Todorov's Grammaire du Decameron; a highly abstract study of narrative sequence in which a universal conceptual grammar (The Grammar of Narrative) is proposed. Under this scheme, different types of textual segments are factored into various categories of 'actions', at distinct levels. These can then be manipulated by the grammar, or described by it, as required.
The concept of category is also central to the Prague School. A prominent example of the use of this for analysis is Jakobson and Levi-Strauss' analysis of Les Chats by Baudelaire 11 . Here the various categories consist of phonic, syntactic, morphemic, semantic, and metrical units. These categories are structured in interlocking levels, and the patterns and relations within, and between them, are central to the analysis of the sonnet, and are vital for the overall critical interpretation.
More directly associated with these assumptions of a material text and the notion of categorical strata is French Structuralist criticism, possibly best described in Roland Barthes' "The Structuralist Activity" 12 . For Barthes, all descriptive units can be classified into categories, which he calls paradigms. He is a strong advocate of the primacy of these categories when undertaking critical analysis: "this notion of paradigm is essential ... if we are to understand the structuralist vision" 13 . Only by careful dissection of a text into "mobile fragments" can its structure be adequately articulated.
The formalist group closest to this view of text is the London School, centered on J.R. Firth, but most articulately developed by M.A.K. Halliday. Halliday sees the building block of the text as lexis. This he defines as the "set of substantives that occupy the places in the sequence of categorical units within a stratum" 14 . At its lowest level this consists of the graphic symbols on the page or the phonemes in utterances. Larger units, such as words, phrases, or syntactic patterns, are produced by collocations, formal patterns of co-occurrence, which may be enumerated to form sets. This scheme, only given here in crude outline, comes very close to a strictly categorical view of text.
Thus several recent structuralist schools "share the assumptions of an autonomous, material text; they encourage examination of the text with the aid of stratified levels of conceptual categories; they differ widely, however, in the formality with which such strata are defined and linked to one another" 15 . The use of the computer requires a high degree of formality to function; therefore, it encourages a more precise description of categories. Since these categories must be described in functional terms for the computer, computer- assisted criticism presents no inherent bar to the differing basic critical perspectives of each group. All of the structuralist schools have been limited both by a lack of any detailed methodology and the problems associated with applying their theories to any sizeable body of text. Because of its plastic functional generality, the computer could be instructed to recognize or create any categories required, and to process substantial works once they were in machine-readable form. This would make it a very useful methodological adjunct to conventional Formalist/Structuralist approaches, providing in effect a general research tool of potentially unbounded application.
Last modified: Monday, 11-Dec-2017 14:42:04 AEDT