Project D12-N (2014-2018)

Principal Investigator: Sabine Schulte im Walde
Researchers: Maximilian Köper, Sylvia Springorum

Associated researchers: Eleri Aedmaa, Jeremy Barnes, Sai Abishek Bhaskar, Evangelia Kiagia, Daniela Naumann, Kim-Anh Nguyen, Dominik Schlechtweg

Student researchers: Alla Abrosimova, Reem Alatrash, Delia Alf, Alexander Frey, Anna Hätty, David Helbig, Glorianna Jagfeld, Ingrid Kasimir, Nana Khvtisavrishvili, Florian Lux, Monique Marquez, Daniela Naumann, Alisa Noha, Alexandra Schweins, Sabrina Tistl

The goal of D12 is to identify regular meaning shifts of German particle verbs (PVs) with prepositional particles, such as abstottern, aufbrausen, nachweinen. Our D12 hypothesis of meaning shifts assumes that there are regular mechanisms in meaning shifts from one domain to another when a base verb (BV) is combined with a particle, and that these shifts apply across a semantically coherent set of verbs. For example, the syntactically similar BVs brummen 'hum' and donnern 'rumble' both describe a displeasing loud sound. Combined with the particle auf, they result in near-synonyms roughly meaning 'forcefully assigning a task' (in one of their senses).

As typically not all senses of a PV undergo meaning shifts (for example, there are at least two senses of aufsprudeln, of which one is literal (meaning 'bubble up'), and one results from a regular meaning shift ('become angry'; shared with aufbrausen and aufkochen), a major prerequisite to identify such regular meaning shifts is sense discrimination. The two major parts of D12 therefore comprise

1. sense discrimination for particle verbs, particles and simplex verbs (including the base verbs with regard to the chosen particle verbs); and
2. identification of regular meaning shifts and the conditions that trigger the meaning shifts.

Concerning (i), our work is interdisciplinary, combining (A) theoretical considerations about salient meaning components with (B) empirical evidence from psycholinguistic experiments and (C) computational models with a strong focus on distributional information and soft clustering methods to represent and disambiguate contextual ambiguity. Concerning (ii), we aim to incorporate the empirical sense properties into computational models that identify regular polysemy regarding PV-BV meaning shifts via analogy. Importantly, we are not only interested in the identification of the meaning shifts per se, but in addition we aim to determine the conditions of the meaning shifts, i.e. the meaning components of the particle verbs and their constituents that trigger the meaning shifts, such as subcategorisation behaviour at the syntax-semantics interface; contextual indicators of abstract vs. concrete contexts; domain information; aspect; etc. As a side-product of meaning shift identification, we will address compositionality: one type of meaning shift is a non-shift with strong PV-BV overlap and a prototypical particle meaning.