A corpus based analysis of TED talks and academic lectures

Abstract

TED talks are a contemporary digital genre via which a wide range of speakers, some of whom are academics, communicate their ideas to a global audience. There is evidence to suggest that TED talks are being used as learning materials in educational contexts including as imitation academic content in EAP classrooms. Therefore, in order to inform the discussion on the use of TED talks in EAP, the current thesis characterises and compares TED talks and academic lectures. In doing so, this thesis contributes to our knowledge of spoken presentation genres, an under-researched area in corpus studies. Two corpora of over five million words each were compiled for this study, the TED Talk Corpus (TTC; n = 2463) and the Open Yale Courses Lecture Corpus (OYCLC; n = 708). Both corpora contain a range of academic disciplines representing soft-pure, soft-applied, hard-pure, and hard-applied areas (Biglan, 1973a; 1973b). The current thesis conducts three rounds of analysis. First, multi-dimensional (MD) analysis is conducted on both corpora using Biber’s (1988) original six-factor model. Second, Exploratory Factor Analysis (EFA) is conducted on the TTC to create a new MD model, which is then applied to the OYCLC. Third, the coverage of academic lexis and the frequency of academic formulas are measured across both corpora. When the TTC and OYCLC were compared in terms of Biber’s (1988) dimensions, both corpora were found to be involved, non-narrative, non-abstract discourses with tendencies towards context-independent reference, overt expressions of persuasion, and on-line information elaboration. Significance testing found that the corpora were different for each factor with small effect sizes, except for the dimension of abstract information, which found a medium effect size. In comparison to Biber’s (1988) registers, the TTC and OYCLC were consistently close to each other, although the dimension of narrative versus non-narrative discourse had a great degree of disciplinary variation. EFA on the TTC resulted in a six-factor model. These factors were interpreted to represent four dimensions and two functions. The dimensions were: condensing information versus dispersing information, human-world oriented versus object/concept oriented, overt statement of fact versus subverting expectations, and passive versus active. The two functions were: denial and epistemic stance. When compared to the OYCLC, no difference was found for Dimension 1, but significant differences were found for all other factors. Conclusions are drawn which challenge traditional disciplinary groupings, particularly in terms of philosophy and economics. OYCLC lectures were found to have significantly more academic lexis and formulas compared to TTC talks, with small effect sizes for the three lists of lexis and a medium effect size for formulas. When academic tagged TED talks were compared to OYCLC lectures, these differences were reduced, particularly in terms of high frequency academic lexis. However, the difference for the AFL was only marginally reduced. When academic disciplines are considered, it appeared that the lists had biases, particularly finding low levels of academic lexis in humanities related lectures and talks. Findings are drawn together in the concluding chapter with a discussion on pedagogical applications, particularly concerning academic listening.