GerEO: A Large-Scale Resource on the Syntactic Distribution of German Experiencer-Object Verbs

Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022), pages 3391–3397 Marseille, 20-25 June 2022 © European Language Resources Association (ELRA), licensed under CC-BY-NC-4.0 3391 GerEO: A Large-Scale Resource on the Syntactic Distribution of German Experiencer-Object Verbs Johanna M. Poppek, Simon Masloch, Tibor Kiss Linguistic Data Science Lab, Ruhr-Universit¨at Bochum Universit¨atsstraße 150, 44801 Bochum, Germany {johanna.poppek, simon.masloch, tibor.kiss}@rub.de Abstract Although studied for several decades, the syntactic properties of experiencer-object (EO) verbs are still under discussion, while most analyses are not supported by substantial corpus data. With GerEO, we intend to fill this lacuna for German EO-verbs by presenting a large-scale database of more than 10,000 examples for 64 verbs (up to 200 per verb) from a newspaper corpus annotated for several syntactic and semantic features relevant for their analysis, including the overall syntactic construction, the semantic stimulus type, and the form of a possible stimulus preposition, i.e. a preposition heading a PP that indicates (a part/aspect of) the stimulus. Non-psych occurrences of the verbs are not excluded from the database but marked as such to make a comparison possible. Data of this kind can be used to develop and test theoretical hypotheses on the properties of EO-verbs, aid in the construction of experiments as well as provide training and test data for AI systems. Keywords: experiencer-object verbs, psych verbs, database, corpus data, German 1. Introduction Psych verbs and their properties in multiple languages have ignited discussions among linguists for several decades. Usually, the term ‘psych verb’ is used in the literature for verbs where one argument expresses the experiencer of some psychological state, see e.g. Landau (2010). Experiencer-object (EO) verbs are those psych verbs whose experiencer argument is linked to the object position in the canonical case. Despite several pro- posed analyses, the properties of EO verbs remain un- der discussion (for a short overview of the literature, see Rozwadowska et al. (2020)). This holds also for the question whether they display distinctive characterist- ics that distinguish them from ‘regular’ transitive verbs. Since Belletti and Rizzi (1988), psych verbs are often considered syntactically deviant, although this has oc- casionally been called into question (e.g. by Grafmiller (2013) and ˙Zychli´nski (2016)). In the canonical pattern, the role of the stimulus (STM) is assigned to the subject, and the role of the experi- encer (EXP) is assigned to the object. Verbs that are known to occur in this particular setting are classified as EO verbs, although some of them are known to also license other syntactic patterns, e.g. the reflexive one described in section 4.1. Argument structure altern- ations have been a matter of debate from the begin- ning in the case of the passive, and especially the re- flexive pattern has stirred interest recently (Alexiadou and Iord˘achioaia, 2014; Pijpops and Speelman, 2017; Hirsch, 2018; Rott et al., 2020, among others). However, there is only little corpus-based work on the topic and, to our knowledge, little to non available data- sets containing larger portions of annotated data on the …

Content truncated to 500 words