Nils Ringe, Department of Political Science, University of Wisconsin-Madison personal website
Problem/Project Description:
The Language(s) of Politics: Multilingual Policy-Making in the European Union
Politics today increasingly takes place in international or regional organizations, yet existing research tends to ignore that it is conducted in multilingual environments by national politicians who interact with one another either in non-native languages or rely on translators as intermediaries. This oversight is striking considering that language is at the core of politics and that an increasing body of research in the cognitive sciences indicates that the use of a foreign language affects people’s political attitudes, inferences, risks perceptions, and moral judgments. This project treats multilingualism as a key features of European Union politics and investigates multilingual politics and policy-making in the EU’s four main institutions.
How is this problem at the intersection of human behavior and machine learning?
A key proposition drawn from almost 100 interviews I conducted in the main EU institutions is that EU actors rely on their own “type” of English as their primary shared working language. This “EU English” is a special purpose language that reflects the particular needs of participants in EU policy-making and is characterized by a shared (but limited) vocabulary, syntax, and semantics. It also tends to be more neutral, technical, and utilitarian than language used in national policy-making, which implies that multilingualism depoliticizes EU politics. I am hoping to find a collaborator with the skills to empirically test these propositions using computational science or machine learning tools.
What part of your project requires a Machine Learning Student:
I have secured data that would allow for the analysis of “EU English,” namely 15 hours of video footage of negotiations between key lawmakers from different parties in the European Parliament. This footage has been transcribed and is ready for analysis and comparison to transcripts from similar negotiations in two legislatures that conduct their business in native English. It is thus possible to compare EU English to “regular” English used in legislative politics with regard to variables such as vocabulary size; collocations; sentence length; number of verb phrases, adjectives, and adverbs; grammatical and pronunciation errors (and correction thereof); disfluencies; metaphors; and use of non-English terms.
Specific Tasks and Expectations of the Graduate Student (label % time for each task):
(Guideline for timeline: one summer, one semester and roughly 8-10hrs/week or 120 hours total)
I am hoping to find a collaborator for the fall semester 2018, but could start as early as August 1, 2018. I don’t believe the task described above will be particularly time consuming, but I also have additional ideas if the collaboration is productive.
Anticipated Deliverables:
I am writing a book on multilingual politics in the EU, but the EU English chapter would lend itself to publication as a co-authored stand-alone article in a political science journal. I am also open to using the data in other ways that my collaborator would deem promising.
Ability to fund student consultants: Yes/No
Student consultants will be funded on an hourly basis.
Contact information:
Nils Ringe, Department of Political Science, UW-Madison