Based on Diewald (2009, 2015a, 2017),
it is claimed that the notion of paradigm needs to be introduced
into constructional accounts, as otherwise the essence of
grammaticalization cannot be properly captured in construction
grammar. It is suggested to define grammatical paradigms as a new
node type, a “hyper-construction”, which represents the categorical,
non-gradient specifics of grammatical meaning. This line of
argumentation is supported by a discussion of changes in the German
and English modality, tense, determiner, and number paradigms.
Independent arguments for the psychological reality of paradigms
come from the study of implicational relations in inflectional
morphology and diachronic phenomena such as layering, suppletion, and
paradigm pressure.