This paper addresses the need for text visualization systems that can support discourse analysis. While fruitful work has been undertaken exploring the paradigmatic axis in systemic functional linguistics, there is much scope for complementary modeling along the syntagmatic axis. However, in exploring long-range text patterns we face the problem that it is difficult for a human analyst to track complex and unfolding relationships in what is high-dimensional data. I introduce AppAnn, a system designed to support linguists in undertaking such long-range analysis and demonstrate how it can be used to visualize patterns of appraisal in texts.
This study aims to shed some light on the role of evaluative language in the process of persuasion in the newly emerging genre of 'online debate'. Drawing on the APPRAISAL framework within Systemic Functional Linguistics, this study investigates the distributional patterns of APPRAISAL choices and co-choices in a corpus of widely viewed online debate texts (ODTs). Based on the voting results of each ODT, textual parts of the corpus were segmented into two main categories: 'more persuasive' and 'less persuasive' debaters. Supported by two specially designed software tools, the ODT corpus was manually annotated for APPRAISAL features, and frequencies of choices and co-choices were extracted automatically. In line with previous research, the findings of this study revealed significant APPRAISAL patterns associated with the ODT debaters, in addition to unique co-patterns characteristic of the 'more persuasive' and 'less persuasive' debaters. These findings are discussed in terms of potential implications, limitations and directions for future research.
One of the fundamental underpinnings of systemic functional linguistics (SFL) is that the relationship between language-as-system and language-as-text is modelled probabilistically in relation to the cline of instantiation. This offers a spectrum of new ways to approach several SFL concepts quantitatively. This paper falls within that spectrum as it proposes that the relatively recent concepts of coupling and syndrome can be redefined quantitatively in relation to instantiation through two statistical methods – namely log-linear analysis and multiple correspondence analysis (MCA) The application of these two methods is illustrated through an analysis of a corpus of twelve online voting-based online debate texts (ODTs) The results and discussion sections of this paper show that the methods can identify and quantify significant couplings and syndromes from both probabilistic and statistical perspectives. Both methods illustratively highlight eleven couplings and four syndromes associated with the more persuasive and less persuasive ODTs writers.
This study investigates least delicate patterns of appraisal in two diachronic corpora of UK Parliament and U.S. Congress speeches over the last two centuries, focusing on diachronic changes and trends of systemic probabilities of least delicate engagement and attitude polarity. Based on computational algorithms that automatically extract appraisal instances and intersections from the two corpora, the comparative analysis carried out in this paper incorporates several statistical methods, including homogeneity or ‘change-point’ tests, Mann-Kendall trend analysis, and time-series Correspondence Analysis. The results indicate that, in both corpora, probabilities of monoglossic as well as attitudinal patterns (as opposed to neutral ones) follow statistically significant upward trends. In addition, positive polarity is increasing steadily, especially in the U.S. Congress. appraisal intersections are also dynamically changing depending on changes in sociopolitical circumstances. More specifically, in the formative and early years during which party conflicts were intensified, heteroglossic patterns are favored. In war and post-war periods, monoglossic patterns are more associated with neutral polarity. In recent decades, during which political polarization hit a peak, monoglossic patterns begin to favor attitudinal polarity. These findings are discussed in terms of possible causal and correlational interpretations, limitations and directions for future research.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
customersupport@researchsolutions.com
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.