In an experimental study of language switching and selection, bilinguals na
med numerals in either their first or second language unpredictably. Respon
se latencies (RTs) on switch trials (where the response language changed fr
om the previous trial) were slower than on nonswitch trials. As predicted,
the language-switching cost was consistently larger when switching to the d
ominant L-1 from the weaker L-2 than vice versa such that, on switch trials
, L-1 responses were slower than in L-2. This "paradoxical" asymmetry in th
e cost of switching languages is explained in terms of differences in relat
ive strength of the bilingual's two languages and the involuntary persisten
ce of the previous language set across an intended switch of language. Nami
ng in the weaker language, L-2, requires active inhibition or suppression o
f the stronger competitor language, L-1; the inhibition persists into the f
ollowing (switch) trial in the form of "negative priming" of the L-1 lexico
n as a whole. (C) 1999 Academic Press.