Language Log has a fascinating article about creole languages and birdsongs:
Zebra finches are among the songbirds who learn their songs by imitating adults, just as human children learn their language by interaction with those who already know it. Male songbirds raised in isolation, without any conspecific adult models during the critical period for song learning, are handicapped for life: they develop only an ill-organized, infantile "subsong". From the example of abused or feral children like Genie, we know that something similar happens with human children.
In both cases, this raises a sort of chicken-and-egg question: if normal development requires an adult model, then which came first, the pupil or the tutor?
One obvious possibility is that the normal pattern is implicit in the species genotype, but requires a combination of cultural evolution and infant learning, repeated over several generations, to develop completely.
The cited work by Olga Feher et al. demonstrates this kind of "multi-generational phenotype" (Ofer's phrase) experimentally, in a colony of zebra finches whose founder was an isolate. As each succeeding generation learns songs from the preceeding one, the effects of biases in the learning process accumulate, so that after a few generations, normal zebra finch songs have re-emerged.
Read the whole thing.