Following Saussure (1916), it is conventional to view language as a system of 'signs'. Morphological approaches have traditionally been distinguished with reference to the locus of the units that they treat as minimal signs within a grammatical system. Initial versions of morphemic analysis regarded morphemes as signs. Word and paradigm models, whether realizational or implicational, identified ... [Show full abstract] lexemes as minimal signs. However, as morphemic models have assumed a more realizational character, the fundamental semiotic divide has shifted, from a contrast between morphemic and lexemic models to an associative/discriminative split within the class of lexemic approaches.