128141 FS FS Research Seminar I / II (2022S)

Information theory may be unfamiliar to many linguists, but it lies at the heart of many widely used linguistic notions. For example, a _collocation_ is a syntagmatic relation between words that occur more frequently together than expected by chance; in other words, knowing that the first word of a collocation was used, hearers would expect the second word to appear with a higher probability than in a random context. In information-theoretic terms, the first word carries non-trivial "(pointwise) mutual information" about the second word. So a traditional linguistic term can also be recast as an information-theoretic one. Such recasting is useful because it opens the way for deeper and/or more automatic analysis of textual data.

In the theoretical part of the course, we will introduce some fundamental notions of information theory as relevant for linguistics, and study several state-of-the-art articles that employ broadly information-theoretic analytical tools for linguistic analysis. For an article example, cf.: https://www.frontiersin.org/articles/10.3389/frai.2020.00073/full For an example of a visualization tool, cf.: http://corpora.ids-mannheim.de/diaviz/royalsociety.html

In the practical part of the course, you will need to study a complex article on your own and react to it in writing; provide feedback to another student doing the same; use one of the analytical tools we will have introduced to study a specific linguistic phenomenon (e.g., the behavior of a single word or a collocation); finally, you will need to form a research group with other students, capitalizing on the different skills that different students have, and then carry out a research project, present its results, and write them up in a research report.

Background requirements:
-- Mathematics: high-school level mastery of mathematics; if you also have been exposed to probability theory before, you might have it a bit easier. However, this is not a mathematics class: the aim is to learn how to use an analytical apparatus that is _based_ on math, not to learn the relevant math _itself_.
-- Computational skills: depending on your background and the skills of the students in your research group, the research project may or may not involve significant coding work. In other words, do not worry if you don't have programming experience, as long as you are generally comfortable with using computer-based analytical tools and are ready to collaborate with other students who have the skills you currently lack.
-- Linguistics knowledge: you do not need to have background in linguistics, but if you lack it, you do need to be open to working with other students in the class who can complement your skills.