In psycholinguistics lexicalisation is the process of going from meaning to sound in speech production.
The most widely accepted model, speech production, in which an underlying concept is converted into a word, is at least a two-stage process. First, the semantic form (which is specified for meaning) is converted into a lemma, which is an abstract form specified for semantic and syntactic information (how a word can be used in a sentence), but not for phonological information (how a word is pronounced). The next stage is the lexeme, which is phonologically specified.[1]
Some recent work has challenged this model, suggesting for example that there is no lemma stage, and that syntactic information is retrieved in the semantic and phonological stages.[2]