Topics

A research group led by Professor Okanoya has discovered a statistical structure in bird songs that is similar to that of human language.

A research group led by Professor Kazuo Okanoya of Advanced Comprehensive Research Organization at Teikyo University applied an analytical method inspired by the mechanism by which human infants find word boundaries from continuous sounds to the songs of the Bengalese finch, revealing that a statistical structure similar to that found in human language exists.
A research group led by Professor Okanoya analyzed 30 song recordings from six male Bengalese finches collected in Japan, recorded at five developmental stages. Their findings revealed statistically cohesive subsequences, defined by boundaries where syllable connections significantly weaken. The frequency of these subsequences closely matched a Zipf distribution, where a small number of units are frequently used and many units appear rarely. This structure could not be explained solely by syllable frequency biases, but rather by a sequence structure exceeding two adjacent syllables. Furthermore, the Zipf distribution was present from the early stages of development, and the cohesiveness of subsequences became clearer with growth. In addition, "Zipf's Law of Conciseness," which states that more frequently used subsequences are shorter, was also confirmed.
This research was published in the academic journal "Science Advances" on August 8, 2026 (Saturday).

The press release is available here.
Click here for information about Specially Appointed Professor Okanoya.

To the topic List