Compound or Term Features? Analyzing Salience in Predicting the Difficulty of German Noun Compounds across Domains

Proceedings of the 10th Conference on Lexical and Computational Semantics, pages 252–262 August 5–6, 2021, Bangkok, Thailand (online) ©2021 Association for Computational Linguistics 252 Compound or Term Features? Analyzing Salience in Predicting the Difficulty of German Noun Compounds across Domains Anna H¨atty1,2, Julia Bettinger2, Michael Dorna1, Jonas Kuhn2, Sabine Schulte im Walde2 1Robert Bosch GmbH, Corporate Research, Renningen, Germany 2Institute for Natural Language Processing, University of Stuttgart, Germany {anna.haetty, michael.dorna}@de.bosch.com, {julia.bettinger,jonas.kuhn,schulte}@ims.uni-stuttgart.de Abstract Predicting the difficulty of domain-specific vo- cabulary is an important task towards a bet- ter understanding of a domain, and to en- hance the communication between lay people and experts. We investigate German closed noun compounds and focus on the interaction of compound-based lexical features (such as frequency and productivity) and terminology- based features (contrasting domain-specific and general language) across word represen- tations and classifiers. Our prediction experi- ments complement insights from classification using (a) manually designed features to charac- terise termhood and compound formation and (b) compound and constituent word embed- dings. We find that for a broad binary distinc- tion into easy vs. difficult general-language compound frequency is sufficient, but for a more fine-grained four-class distinction it is crucial to include contrastive termhood fea- tures and compound and constituent features. 1 Introduction In times of a constant growth of domain-specific data, it is more important than ever to analyse char- acteristics of domain-specific vocabulary. Domains are topically restricted subject fields containing domain-specific vocabulary that encode domain knowledge. The more technical the terminology in the domain vocabulary, the more difficult it is per- ceived by lay people unfamiliar with the domain. Predicting the difficulty of domain-specific vocabu- lary is therefore an important task for enhancing the communication between lays and experts. A promi- nent example in this respect is the medical domain, where the prediction of difficulty of medical terms can enhance the communication between doctors and patients, e.g. by simplifying medical texts (Abrahamsson et al., 2014; Grabar and Hamon, 2014; Wandji Tchami and Grabar, 2014). While the medical domain represents a well-researched focus, the problem of miscommunication appears across domains. Previous research on automatic term difficulty prediction already explored a large number of pa- rameters, but as to our knowledge there is yet no study that investigated how difficulty can be at- tributed to complex phrase formation processes (a language phenomenon) in interaction with domain specialization (a domain phenomenon). The cur- rent study investigates these aspects, goes beyond domain peculiarities (such as Latin words in the medical domain), and performs analyses across three rather different domains: Cooking, DIY (’do- it-yourself’) and Automotive. While we choose a diverse set of domains, we otherwise focus on a special phenomenon within domain-specific vocabulary: German closed com- pounds. Closed compounds are complex expres- sions that consist of several lexemes and are writ- ten in a single string of characters. An example is Bremsfl¨ussigkeit ’brake fluid’, which is com- posed of the two simple words Bremse ’brake’ and Fl¨ussigkeit ’fluid’. By focusing on closed com- pounds, the boundaries of the phrases …

Content truncated to 500 words