Testing quasi-lexemes with theories of compounding enables me to analyze their specificity and limitations. Quasi-lexemes have a semantic content like complete lexemes do, but are constrained by their morphological incompleteness. Despite the fact that their lexical categories are not readily discernible, there is evidence for derivational lineages in Greek and Latin, and these derivations account for the division between primary and synthetic patterns in compounding. For that reason, quasi-lexemes can legitimately be said to obey the constraints of complete lexemes in terms of the syntactic and semantic relations. The only problem is their belonging to three lexical categories only (verb, noun, adjective), which compels me to defend that, although quasi-lexemes form compounds and are liable to undergo paradigmatic extension opening up the way for productivity, at the same time, they tend to participate in compounding in a more restrained and conventional way than complete lexemes. |