Masters’ Academy

Prof. Vahram Atayan, Heidelberg University

What data tells – or rather shows: Data visualization in linguistics

The last decades have witnessed a major paradigm shift in linguistics towards empirical approaches in research. The compilation of large corpora and the development of a wide range of corpus and computational linguistic methods opened up entirely new perspectives for contrastive linguistics and translation studies. Computational linguistic tools allow the automatic annotation of different linguistic categories on the language surface. Moreover, machine learning methods provide access also to more abstract semantic categories (such as sentiment and argumentation). Nevertheless, humans are still much better than automated systems at analyzing highly abstract phenomena. This is the reason why we consider the combination of manual annotations of abstract phenomena with statistical evaluation a particularly promising approach in linguistics and in the humanities in general. Since humans are usually not very adept in processing numerical data, which constitutes the standard output of statistical analysis, we also believe that the visualization of statistical results is of great importance. In our talk, we will discuss some types of research questions concerning nominal data and different scenarios for statistical evaluation and data visualization.

BIO: Vahram Atayan is Full Professor of translation studies for French and Italian at the Institute of Translation and Interpreting at Heidelberg University. He studied physics in Yerevan (Armenia), then computer science and translation in Saarbrücken (Germany). In his PhD Thesis, he analyzed the macrostructures of argumentation in German, French and Italian. His research interests cover translation studies, contrastive linguistics, languages for special purposes, temporal semantics and corpus-based discourse linguistics.