Statistics in Linguistics and Communication
Course Description
This course focuses on statistical analysis of research data, centred around today’s standard method, which involves linear models, especially those with mixed effects (i.e., both “fixed effects” and “random effects” (e.g., participants drawn randomly from a population of people, and often also words or sentences drawn randomly from a language). If you are already familiar with some statistical techniques, you may have seen t-tests, correlation tests, and analysis of variance. During the course, these concepts turn out to be specific simplifying cases of mixed-effects modelling.
The course also addresses design issues that make the analyses suitable for your experiment, such as sampling, data collection, and reliability and validity of measurements. You apply these concepts, together with the analysis techniques, to theoretical, typological and applied research in your linguistic subdisciplines. Special attention is paid to statistical inference, i.e. correct use and formulation of statistical results.
You do not need prior knowledge of statistical methods. If you do have such knowledge, you may in fact have to unlearn some internalized ways of thinking.
Objectives
Upon successful completion of this course, you will
- have gained insight into the application of statistical and experimental methods in empirical research in linguistics;
- be able, given your data table, to compute the best mixed-effects model that describes your data by applying a technique called linear regression or a technique called logistic regression. The software you will be able to use for accomplishing that is R.
Further information can be found here.
NB: This is a course mainly meant for MA students. As a result, the amount of places for PhD researchers is extremely limited. Based on interest, we may organise a course specifically for PhD researchers next semester.
Lecturers
This course is taught by Titia Benders and Paul Boersma.