An Automatic Measure of Cross-Language Text Structures
- PDF / 659,716 Bytes
- 14 Pages / 439.37 x 666.142 pts Page_size
- 14 Downloads / 180 Views
An Automatic Measure of Cross-Language Text Structures Kyung Kim1
© Springer Science+Business Media B.V. 2017
Abstract In order to further validate and extend the application of GIKS (Graphical Interface of Knowledge Structure) beyond English, this investigation applies the GIKS to capture, visually represent, and compare text structures inherent in two “contrasting” languages. The English and parallel Korean versions of 50 expository and 50 narrative texts from Newsweek, Popular Science and Vogue magazines were converted into Pathfinder network graphs for analysis, based on key concepts and their relative proximity relationships in the texts. Results indicate that the novel text structures obtained by this approach reveal unique, useful, and interesting linguistic differences for the English and Korean texts, confirming prior linguistic literature (Sohn 2001). Particularly, the analyses show that the expository text structures are somewhat similar between these two languages while the narrative text structures are quite different. These findings demonstrate the utility of the GIKS for representing the nature of texts as manifested in different languages. If further confirmed in other languages, this computer-based approach could offer a new way for students and instructors to interact with lesson texts by providing a visual representation of text structure, regardless of language. Keywords Text structure · Expository and narrative text · Knowledge structure · GIKS
1 Introduction Over the past 30 years, contrastive rhetoric has examined similarities and differences in texts across languages and cultures, based on the hypothesis that texts are not merely static products but are functional parts of dynamic cultural contexts (Crossley and McNamara 2012); for example, texts written in English by non-English speakers might reflect the & Kyung Kim [email protected] 1
Learning, Design, and Technology, Department of Learning and Performance Systems, The Pennsylvania State University, University Park, PA 16802-1303, USA
123
K. Kim
patterns valued in their home culture because “texts have schematic structures which are culturally variable” (Connor and Kaplan 1987). From Kaplan’s (1966) initial theory on contrastive rhetoric to the present, the bulk of research has demonstrated that different culture-linguistic systems affect thought and perception in a different way (see for review, Taboada 2006). The implication is that language patterns reflect perceptional patterns, and perceptional patterns are reflected in text artifacts. Until now, many computational text analysis tools have been developed that make comprehensive studies of text variables feasible (see for review, Kim 2012a). These technologies attempt to leverage many different text variables from surface-(e.g., lexical elements) to deep-levels (e.g., semantic elements) contained in a text in order to better apprehend its explicitly structured representation. Although the MITOCAR (Text Model Inspection Trace of Concepts and Relations, Ifenthaler et al. 2010) func
Data Loading...
