Zebra finches transform random songs to exhibit linguistic laws

Linguistic laws are increasingly used as markers of efficiency in non-human communication, but it remains unclear how rapidly these patterns can emerge. In this re-analysis of experimental data from James and Sakata (1), I assessed whether zebra finches tutored with random songs that have flat frequency distributions and fixed lengths transform them to exhibit three linguistic laws associated with efficiency in human language. Menzerath’s law and Zipf’s rank-frequency law are present in the learned songs to a similar extent as in human language, while there is only weak support for Zipf’s law of abbreviation. These results suggest that some measures of language-like efficiency can emerge extremely rapidly, while others may require iterated social learning.