Niet-Intentioneel Gebruik van de Eerste Taal in Mondelinge Tweede-Taalproduktie
Theo Bongaerts | Vakgroep Toegepaste Taalwetenschap en Methodologie Katholieke Universiteit Nijmegen
The results are presented of a study of unintentional switches to the L1 in L2 speech production, and the relevance of the findings for current views on bilingual speech production is discussed. The data for the study are 771 unintentional switches to Dutch at the word level in a 140,000 word corpus of L2 speech produced by 45 Dutch learners of English at three different proficiency levels.
The learners' use of switches to the L1 appeared to be related to their L2 proficiency level. Function words were more often involved in unintentional codeswitching than content words, particularly in the case of the less advanced L2 learners. These findings are interpreted as providing support for current spreading activation accounts of lexical access in bilingual speech production in which the relative frequency of L1 and L2 words in the learner's repertoire plays an important role.
Following Levelt (1993), it is assumed that the frequency effect plays a role at the lexeme or word form level rather than at the lemma level. Arguments are advanced that the unintentional use of L1 words in L2 production can be viewed as the result of errors, or slips, in lexical access at the lexeme level. Most switches were not in any way adapted to the L2 system, but there were a few instances of morphological (12) and phonological (10) adaptation to the L2. The cases of morphological adaptation all involved the use of a Dutch base form (stem) without the appropriate Dutch inflectional morpheme. In all cases the unintentionally used Dutch words were adapted to English in the sense that, in accordance with the rules of English, no inflectional morphemes were added to the stem. The absence of Dutch inflections offers support for Myers-Scotton's (1993) Matrix Language Frame Model, which predicts that in L2 + L1 constituents, in which L2 is the matrix language and L1 the embedded language, inflectional morphemes are supplied by the L2.
Finally, on the basis of the occurrence of certain adaptations in the data and the absence of certain other adaptations, it is, be it tentatively, concluded that inflected words must be stored in the mental lexicon both fully and in decomposed form and that there is a checking device which intercepts forms that are not represented in the mental lexicon.
Article language: Dutch