Register in Czech: Designing an MDA-based experimental study

Abstract

There are no conventionalized ways to investigate the results of multidimensional analysis (MDA) from the perceptual perspective in an experimental setting. An MDA of the Czech corpus Koditex by Cvrček et al. (2020) established eight dimensions of variation based on 122 linguistic features. The first two dimensions, which explain the largest proportion of shared variance, are labeled as 1. dynamic (+)/static (-) and 2. spontaneous (+)/prepared (-). In our study, we investigated if some situational contexts of language use evoke stronger associations with the poles of the two dimensions than others. Furthermore, we aimed to explore the impact of the mode of language use and the properties of the interlocutor on the ratings. Czech native speakers (n=107) rated various situational contexts on 7-point Likert-like scales representing the MDA-based dimensions. The items were balanced in the formality of the interlocutor’s name (Mr. or Mrs. in connection to surname/first name), the interlocutor´s gender, and the mode of language use (spoken/written). The statistical analysis uncovered a significant effect of the formality of the interlocutor's name and the mode of language use on the ratings. Using first names and spoken mode resulted in ratings closer to the positive poles of the dimensions. The comparison of individual items showed that some situations, mainly those representing the negative poles of the dimensions, are rated more consistently than others. The results of our study offer insight into how native speakers evaluate situations of language use on the scales of preparedness, subjectivity, and interactivity.