Bilingual children and adults differ in their semantic categorisation decisions for cognates and non-cognates.

Two experiments investigated priming effects of cognate and non-cognate translations in Greek and English. Experiments 1 and 2 involved adults and children, respectively, with either Greek or English as their dominant language who participated in a masked priming semantic categorisation task. The experimental stimuli were cognate and non-cognate translations or unrelated pairs where either the prime was in the dominant language (L1) and the target in the non-dominant (L2) or vice versa. In Experiment 1, priming for cognate translations was observed relative to the control baselines only in the L1-L2 direction (priming asymmetry). There were no effects for non-cognates. In Experiment 2 the task was the same but participants were 10 to 11 year old bilingual children, with Greek or English as their dominant language. Priming effects were observed for cognates in the L2 - L1 direction, whereas non-cognates did not show any effects. The results for adults are in congruence with previous research with unbalanced bilinguals and are discussed within current models of bilingual semantic organisation. The results from the children experiment are discussed with reference to task demand characteristics and effects of new methods of teaching L2 to early learners which allows exploitation of their concepts.