Starting from the assumption that the Slavic aspect (perfective : imperfective) opposition builds on stem derivation, we investigate changes in the inventory of so-called aspect triplets in Czech since 1750. Triplets arise from aspect pairs consisting of an imperfective simplex stem and its perfective derivative whose prefix overlaps in meaning with a semantic component of the simplex (Natural Perfectives; Janda et al. 2013).
If such pairs acquire another imperfective stem via suffixation, aspect triplets result (e.g., dělit RIGHTWARDS DOUBLE ARROW roz-dělit RIGHTWARDS DOUBLE ARROW rozděl-ova-t 'separate'). Many triplets "survive", but they do not form a homogeneous class w.r.t. some parameters.
Based on a database for the period 1750-2017, we explore frequency-based patterns in the formation and persistence of Czech aspect triplets; these patterns are established for six subperiods by employing time series clustering with non-Euclidean distance measures. We then inquire how consistently such patterns are associated with their "age" and part-of-speech provenance.
Moreover, the obtained clusters provide a starting point for case studies of selected typical representatives that examine changes in lexical semantics, collocations, and argument coding.