How sentence processing sheds light on mixed language creation

Abstract

Auer (1999) and O’Shannessy (2012) suggest that codeswitching can become conventionalized, i.e., what is dubbed a “fused lect”, and eventually give rise to a “mixed language”. In this paper I propose that the study of sentence processing can shed light on this process. I discuss recent experimental data from a typologically rare form of mixing, variably termed “fused lect” (Adamou 2010) and “unevenly mixed language” (Adamou and Granqvist 2015), characterized by the conventionalized use of Turkish verbs together with Turkish morphology in a Romani environment. Specifically, Adamou and Shen (2019) conducted two on-line experiments, a picture choice task with sentence auditory stimuli (37 participants) and a word recognition task in sentence context (49 participants). Results from these experiments indicate that language switching costs depend on the degree of conventionalization and support usage-based approaches to language processing. I argue that these findings also lend support to the categorization of fused lects as an intermediate form between codeswitching and mixed languages.