RECONSTRUCTING NEW SERBIAN LEXICAL BLENDS: (SOCIO)LINGUISTIC EVIDENCE

Abstract

Lexical blends are unconventional products created by combining (parts of) two or more source words, often involving segment loss and structural overlap, which makes their constituent elements difficult to recognize. This structural opacity poses challenges for speakers attempting to reconstruct the source words, a topic that has been only sparsely investigated in Serbian, particularly with regard to speaker comprehension. The aim of the present paper is twofold: (a) to investigate whether relative morphotactic transparency and overlapping contribute to native speakers’ success in reconstructing Serbian blends and (b) to investigate the influence of several sociodemographic factors (sex, age, education level, occupation, and area of residence, specifically rural vs. urban) on the informants’ success in blend reconstruction. The results of the quantitative analyses of the informants’ correct answers show that complete-telescope blends are most successfully reconstructed, whereas the reconstruction of complete-inclusive blends is found to be least successful. It has also been shown that overlapping blends are more successfully reconstructed than non-overlapping ones. The results of the sociolinguistic analysis show that none of the four sociodemographic factors (i.e., sex, age, education level, and area of residence) has significant effect on the informants’ success in recognizing the source words in a blend, but that there are statistically significant differences in the average number of correct answers between certain occupational groups, possibly due to varying degrees of linguistic creativity and exposure to language innovation in these occupational settings.