Three themes in the work of Anna Wierzbicka inform the present study and the predecessor work upon which it rests. The first is her long-term engagement with prototypes in semantics. Rejecting the conventional dichotomy between prototypes and semantic invariants, she has long insisted that a prototype can be ‘part of the semantic invariant itself’ (Wierzbicka 1996: 167). Second, Wierzbicka has always recognised that lexical polysemy is normal, indeed ubiquitous, in everyday language and that it is impossible to reach valid generalisations about words, meaning and grammar without taking polysemy seriously. This means not only adopting a rigorous methodology whereby one can reliably distinguish genuine polysemy from generality or vagueness of meaning, but being prepared to put in weeks or months of painstaking effort to disentangle complex polysemic networks of intertwined lexical, phraseological and grammatical meanings.