
Full text loading...
Abstract
Discourse Completion Tasks (DCTs) have been one of the most popular tools in pragmatics research. Yet, many have criticized DCTs for their lack of authenticity (e.g., Culpeper, Mackey, & Taguchi, 2018; Nguyen, 2019). We propose that corpora can serve as resources in designing and evaluating DCTs. We created a DCT using advice-seeking prompts from the Q+A corpus (Baker & Egbert, 2016). Then, we administered the DCT to 33 participants. We evaluated the DCT by (1) comparing the linguistic form and the semantic content of the participants’ DCT responses (i.e., advice-giving expressions) with authentic data from the corpus; and (2) interviewing the participants about the instrument quality. Chi-square tests between DCT data and corpus data revealed no significant differences in advice-giving expressions in terms of both the overall level of directness (χ2 [2, N = 660] = 6.94, p = .03, V = .10) and linguistic realization (χ2 [8, N = 660] = 17.75, p = .02, V = .16), and showed a significant difference but small effect size in terms of semantic content (χ2 [6, N = 512] = 30.35, p < .01, V = .24). Taken together with the interview data, our findings indicate that corpora are useful in designing DCTs.
Article metrics loading...
Full text loading...
References
Data & Media loading...