In this paper it is investigated what happens in a culturally evolving communication system if children learn faster than adults. An agent based computer model is investigated which contains a population of agents in which vowel systems emerge when individuals play imitation games. These vowel systems are found to be preserved better in the case that there is an age structure in the population. When there is a difference in learning ability between children and adults, the vowel systems are more stable, the mutual intelligibility stays higher and a higher complexity can be maintained. Recent advances in sociolinguistics indicate that changes in language learning ability while growing up can account for the different ways that linguistic changes spread from community to community. They found that the structure of the system is better preserved when speakers of all ages are involved in language contact as opposed to only adult to adult contact. Our findings therefore seem to be in line with observations in real language evolution and we provide additional evidence for the fact that it is important to consider the consequences of language acquisition age effects in the study of language preservation and change.