The aim was to identify which criteria children use to decide on the category membership of native and non-native vowels, and to get insight into the organization of phonological representations in the bilingual mind.

The study consisted of two cross-language mispronunciation detection tasks in which L2 vowels were inserted into L1 words and vice versa. In Experiment 1, 10- to 12-year-old Dutch-speaking children were presented with Dutch words which were either pronounced with the target Dutch vowel or with an English vowel inserted in the Dutch consonantal frame. Experiment 2 was a mirror of the first, with English words which were pronounced "correctly" or which were "mispronounced" with a Dutch vowel.

Analyses focused on extent to which child and adult listeners accepted substitutions of Dutch vowels by English ones, and vice versa.

The results of Experiment 1 revealed that between the age of ten and twelve children have well-established phonological vowel categories in their native language. However, Experiment 2 showed that in their non-native language, children tended to accept mispronounced items which involve sounds from their native language. At the same time, though, they did not fully rely on their native phonemic inventory because the children accepted most of the correctly pronounced English items.

While many studies have examined native and non-native perception by infants and adults, studies on first and second language perception of school-age children are rare. This study adds to the body of literature aimed at expanding our knowledge in this area.

The study has implications for models of the organization of the bilingual mind: while proficient adult non-native listeners generally have clearly separated sets of phonological representations for their two languages, for non-proficient child learners the L1 phonology still exerts a strong influence on the L2 phonology.