Contact x time

Abstract

Investigations of the language behaviour of immigrant communities usually find that the degree of maintenance of the heritage language or shift to the language of the host country differs both between immigrant communities and between individuals. On the one hand, group comparisons between migrants and controls typically show signs of attrition among the experimental populations. On the other, investigations of individual linguistic development within one group of immigrants also show great variation between speakers where the degree of L1 maintenance or deterioration is concerned. Two decades ago, an explanation for such differences in L1 proficiency was proposed which invoked an interaction of two factors: time spent in emigration and amount of contact with the L1. In a study of Dutch migrants in France, De Bot, Gommans & Rossing (1991) established that there was a correlation between attrition effects in free speech and length of residence, but that this effect was only apparent for those speakers who had very little contact with their L1.The present investigation attempts to replicate this finding in a large-scale study of the attrition of L1 German in an L2 English and an L2 Dutch setting. It finds that, while length of residence has no explanatory validity when assessed across the entire population, a differential investigation of subgroups of speakers with different amounts of contact does show an impact of time with respect to performance on formal tasks, perceived foreign accent, and accuracy in free speech. Interestingly, with the exception of accuracy measures, this correlation obtains not only for those migrants who have the least contact with L1, but also for those with the most. It is argued that both very frequent and very infrequent use of the L1 can accelerate attrition, either through contact-induced change within a bilingual migrant community, or through lack of rehearsal.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 52 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.