Language VisualizationLanguage Visualization

Language Visualization
and Multilayer Text Analysis

Objectives

The first objective of this investigation was the creation of a prototype tool that can study language/discourse phenomena in three-dimensional space. The idea was to develop a tool which allow a researcher to explore interactively the structures and typologies of discursive formation in large samples of textual data and develop new techniques for reading and interpreting text space. With great support from theVisualization Group here at the Cornell Theory Center, this first objective was successfully accomplished.

IBM Data Explorer (DX) The second objective, the actual study of textual data, is still unfolding. My interest is in studying the structures and typologies of discursive formation in large samples of textual data, and developing new techniques for reading and interpreting text. It has taken about a year of trial and error to get to a stage where actual anaylysis can be performed effectively. My first analysis was of Donna Haraway's "Manifesto for Cyborgs" and other cyberspace related documents. While this tool has already offered me important clues about discourse formation and about future strategies for language visualization, it should remain clear that this line of research is still very young and that this is a very limited example of material analyzed and of techniques employed.

Tools and Methods

The raw material for this work is ASCII text. Project Gutenberg and other electronic archives provide valuable sources for literary and cultural texts. All the texts that I tested, with the exception of Donna Haraway's "Manifesto for Cyborgs" (which I scanned in), came from public electronic archives. A C program converts ASCII files into IBM Data Explorer (DX) files. The ASCII program also regulates the size of the pages, and for the moment I am using an 8.5" x 11" standard. DX then reads the text as data points on a X,Y,Z axis and plots them according to the spacing in the ASCII2DX program. While this program can work on both IBM RS6000's or SGI workstations, most of the images shown here were produced on a SGI Onyx computer.

There are two basic parts to this tool, the pageview and the imageview, which appear simultanously on the screen every time the program is executed.

Pageview Imageview

The Pageview:

For the Pageview, DX can position a camera view to take a slice of the image and position it in a separate space so that the viewer can observe the context in which the words appear. As shown in the image, the words selected appear highlighted in the page. This pageview can be placed anywhere in the text and sometimes appears in the imageview as the white page.

The Imageview:

The imageview contains the entire text or section of text used in the analysis. Examples of the different methods of viewing the text are shown below. Since the purpose is to observe patterns of specific word/concepts, sometimes it not necessary (and less computer intensive) to show all the text. The use of the tubes to connect words can serve as a visual aid for tracing the word patterns. Occasionally the tube patterns can get too complicated to provide any valuable information. At the bottom of the imageview the selected words are shown next to the number of times they appear in the text.

Next:Techniques

El diseño de la página y las imágenes son