The dependency distance hypothesis for bilingual code-switching
This paper addresses the questions why and where, i.e. in which dependency relations, multilingual speakers are likely to code-switch. Code-switching (CS) is the linguistic behaviour of producing or comprehending language which is composed of lexical items and grammatical structures from two (or more) languages. This paper proposes that long dependency distances between syntactically related units facilitate code-switching (Distance Hypothesis DH). Dependency distance is defined as the number of words intervening between a head and a dependent. The DH is tested on two data sets from typologically different language pairs: a 93,235 word corpus of German/English monolingual and code-switched conversations analyzed in Word Grammar (WG), and a 19,766 word Treebank of Chinese/English mono- and bilingual speech. The data sets support the DH in general and on specific syntactic functions. In ongoing work the DH is being tested further on Welsh/English and Spanish/English corpora and experimentally.