An Inquiry into the Semantic Transparency and Productivity of German Particle Verbs and Derivational Affixation

This study addresses the relation between morphological productivity and semantic transparency. Us- ing distributional semantics, we compare German word formation with particles with derivational word formation. We observed that derivational suffixes, but not particles, tend to make strong independent semantic contributions to their carrier words. In two-dimensional t-SNE maps, complex words show clus- tering by affix, but not by particle. Furthermore, the semantic vectors of suffixed words are predictable from their base words with higher accuracy than is possible for particle verbs. This holds irrespective of whether the semantic vectors of complex words are predicted using a generalization of the FRACCS model, or whether a more recent, and more precise, additive model (CosClassAvg) is used. For particle verbs, but not affixed verbs, a greater semantic similarity within the set of complex words correlated negatively with the number of types. Only for particle verbs, a greater number of types predicted a reduced probability of unseen types. Considered jointly, the particle verbs are characterized by quan- titative properties that are very similar to those observed for Mandarin compounds. We propose that particle verbs, just as compounds, primarily serve the onomasiological function of labeling, resulting in highly distinct semantic vectors for particle verbs. By contrast, derivational affixation groups complex words closely together in semantic space while maintaining strong and consistent semantic relations with the base words. This enables these words to serve not only as labels, but also allows them to be used for syntactic recategorization.