We study the role of "text as data" in the Political Sciences (as one prominent case of Social Sciences) and compare their typical research paradigms to those that are common in the CompLinguistics/NLP community. Text genres of interest will include parliamentary debate, party manifestos, public documents from NGOs, etc. Prominent methods that we will cover are the analysis of sentiment and argument, topic modeling, and "scaling", i.e. mapping documents to scales like the "left-right spectrum". The course has two parts: In phase 1 we read papers, to get acquainted with the PolSci research perspectives, and in phase 2 students will work on term projects; these can be either software projects or empirical studies.