Vector Space Models and the usage patterns of Indonesian denominal verbs --- A case study of verbs with meN-, meN-/-kan, and meN-/-i affixes

This paper demonstrates a computational approach of Vector Space Model (VSM), combined with Hierarchical Agglomerative Clustering, to identify semantic (dis)similarity and cluster between a set of Indonesian denominal verbs with meN-, meN-/-kan and meN-/-i affixes. We contextualise the study within the hypotheses that some -kan/-i verb pairs exhibit indistinguishable as well as distinct semantics. Our VSM-based cluster analysis captures derivational families that do cluster together and those where -kan/-i pairs are separated, reflecting their distinct semantics. We also found verbs of different roots and morphologies forming coherent semantic clusters (i.e., MOTION, COMMUNICATION, and PSYCH verbs). Our quantitative corpus-based study sheds a new light on how forms with these three morphological affixes differ in their semantic distribution, providing some support to the qualitative view of semantic differences and similarity between -i and -kan derivatives.