This seems like a variation on regression to the mean. If you take a group of people who have had unusually bad circumstances (language + cultural barriers) and

This seems like a variation on regression to the mean. If you take a group of people who have had unusually bad circumstances (language + cultural barriers) and compare with a similar group (their children) odds are that things will improve on the next iteration (assuming that the bad circumstances are transient and not endemic).