Invited Talk

The Speaker

Elke Teich
Department of Language Science and Technology
Saarland University

Elke Teich is a full professor of English Linguistics and Translation at the Department of Language Science and Technology, Saarland University, Saarbrücken, Germany. Since 2014 she has been the head of the Collaborative Research Center “SFB 1102 Information Density and Linguistic Encoding” funded by the German Research Foundation (DFG). She is currently a principal investigator on two projects in SFB 1102, one on diachronic language change and one on human translation, as well as the Saarbrücken Cluster of Excellence Multimodal Computing and Interaction (MMCI) and the German CLARIN project (Common Language Resources and Technology Infrastructure). Elke Teich is an editorial board member of several journals and book series, including ‘Languages in Contrast’ (Benjamins) and ‘Linguistics and the Human Sciences (Equinox)’. She is a regular reviewer for national and international funding agencies, including Deutsche Forschungsgemeinschaft (DFG), Humboldt Foundation, Schweizer Nationalfonds and the Finnish Academy.
Teich’s expertise ranges from descriptive grammar of English and German over (multi-lingual) register analysis with a special focus on scientific language to translatology. She worked on machine translation, automatic text generation, corpus linguistics and the digital humanities at the following academic institutions: Gesellschaft für Mathematik und Datenverarbeitung (Fraunhofer), Information Sciences Institute (ISI)/USC Los Angeles, University of Sydney, Macquarie University and Technical University Darmstadt. Her research focus in the last 10 years has been on developing computationally based approaches to modelling language variation and change.

The Talk

Linguistic variation and the dynamics of language use

It is widely acknowledged that linguistic variation is a core feature of language, affecting all linguistic levels from the phonetic to the semantic level.  Linguistic variation emerges and is reinforced through language use in context, continuously adapting to social and cognitive constraints.  Language use thus provides excellent data for studying (changing) socio-cultural practices as well as the (general) mechanisms of human communication.

In my talk I focus on two opposing but complementary effects to be observed in the dynamics of language use:  innovation and conventionalization.  Innovation leads to an expansion of linguistic options by new linguistic coinages, e.g. new words entering language or known words being used in new contexts. Conventionalization leads to a reduction of options by convergence in linguistic usage, i.e. the tacit agreement on “how to say things” often associated with a specific style or register. I will show that while innovation and conventionalization pull in different directions, they interact in specific ways to keep language intact for communication.

The underlying approach is corpus-based, using data-driven methods. Language models (e.g. word embeddings) are combined with selected information-theoretic measures (entropy, surprisal), providing models of language use and indices of linguistic variation (here: with special regard of innovation and conventionalization).  I will focus on the domain of scientific writing (English) from a diachronic perspective with side glimpses at translation in the domain of European Parliament.