B3: Information Density and Fragments in German
Project B3 focuses on the predictability of non-sentential utterances (called fragments) like “Was zum Kaffee?” vs. their more fully expanded variants (like e.g. „Möchtest du was zum Kaffee?”), taking into account various factors such as script knowledge, text type, type of construction and surrounding linguistic context. Following the assumption of uniform information density, the main hypothesis pursued is that the use of a fragment is governed by the desire to avoid significant troughs and peaks in information density. In order to determine precisely what a predictive context is, the corpus analysis will investigate the use of fragments according to different text types as well as varying linguistic contexts. In addition, planned psycholinguistic studies include reading time experiments designed to establish whether fragments are processed more rapidly in predictive contexts and are thus relatively easier to process than their fully expanded counterparts.
On the omission of articles and copulae in German newspaper headlines Journal Article Forthcoming
Linguistic Variation, 17 (2), Forthcoming.
Sentential or not? - An experimental investigation on the syntax of fragments Inproceedings Forthcoming
Proceedings of Linguistic Evidence 2016, Tübingen, Forthcoming.
Optimal encoding! - Information Theory constrains article omission in newspaper headlines Inproceedings
Proceedings of EACL 2017, Valencia, 2017.
The Fragment Corpus (FraC) Inproceedings Forthcoming
Proceedings of the 9th International Corpus Linguistics Conference, Birmingham, UK, Forthcoming.
Dipper, Stefanie; Neubarth, Friedrich; Zinsmeister, Heike (Ed.): Proceedings of the 13th Conference on Natural Language Processing (KONVENS 2016), pp. 125-127, Bochum, 2016.