Abstract
One strain of research in psycholinguistics focuses on morphological processing and how speakers parse and store lexemes. Experimental data gleaned from modern languages suggest that speakers register morphemic structures and that words that share root morphemes interact in the mental lexicon and form families.
This study applies the same framework to Ancient Greek and explores the implications of morphological parsing for the acquisition of semantics. We propose that speakers tentatively connected novel words to their possible etymological relatives because of formal similarities and that such a lexical neighbourhood supplied cues to the intended meaning of linguistic signs.
These types of cues, however, were missing for words that were synchronically isolated and could not be matched with any related lexemes. In turn, this increased the possibility for semantic reanalysis, which can be registered as semantic change from the diachronic perspective.
This hypothesis has been tested on two datasets consisting of Ancient Greek nominal lexemes created using the suffixes -μο- and -ρο-, respectively. The experimental groups comprised words for which we have predicted diminished formal transparency and hence a greater chance of being synchronically isolated; the control groups contained words more likely to have been formally transparent.
We have observed the prevalence of semantic change in both groups of the two datasets and corroborated this part of the analysis by consulting the Late Antique Greek etymologica. The subsequent statistical analysis has revealed that the words belonging to the groups of the potentially isolated lexemes were more likely to undergo semantic reanalysis than their counterparts from the control groups.