The perfectly balanced bilingual is a myth. In over 30 experiments using a variety of methodologies I have never found a bilingual who behaved like two monolinguals in one head. In addition, I have found evidence that the first language is not always the dominant language and that high levels of fluency in a second language are possible. This has led me to conclude that language learning is a highly dynamical process that is sensitive to the external conditions in which each language is used. Evidence in favor of this view includes:

• Early Spanish-English bilinguals tested in my laboratory are dominant in their second language, English. This is revealed in the amount of priming of a sentence context to a visual word (Hernandez, Bates, & Avila, 1996), the number of pictures (a maximum of 60) that can be named on the Boston Naming Test (Kohnert, Hernandez, & Bates, 1998) and on the speed and accuracy for naming familiar pictures. In addition, these individuals show a large degree of variability in the first and second language, suggesting that other factors aside from age of acquisition are contributing to fluency in both languages.
• Bilingual children show improvement (indexed by both faster RTs and more accurate performance) in picture naming from age 5 to 18 (Kohnert, Bates, & Hernandez, 1999). However, this improvement is marked by a dramatic change from Spanish dominance at age 5 to clear English dominance by age 14. Second, bilinguals who learned their second language after the age of 16 can attain a high level of fluency in their second language even when it is learned in adolescence. I am currently continuing this line of research by looking at both early and late bilinguals across both semantic and syntactic processing tasks.
• When making a decision about the gender of a word, early and late bilinguals show increased activity in Broca’s area for difficult items. In late bilinguals, the increase in activity extends into areas anterior to Broca’s area. This suggests that late bilinguals perform deeper lexical and syntactic analyses when making gender decisions for difficult items. However, early bilinguals require more complex syntactic analyses for difficult items in the same task.