Both birdsong and human language are learned, requiring complex social input. New findings show, however, that bird populations 'seeded' with aberrant song input transform it to normal song in a few generations.
Around 650 BC, the Egyptian king Psammeticus reputedly had two children raised by a mute goatherd, isolated from all linguistic input, to determine what language they would speak spontaneously1. As legend has it, the children's first word was bekos, so the king judged the 'original language' to be Phrygian, in which bekos means bread (that bekos represented the bleat of a goat was apparently not considered). Today, such an experiment seems both unethical and naive, for we know from numerous examples that children deprived of linguistic input will not spontaneously speak any language. To acquire language, humans need rich linguistic input early in life2,3. In a study published on page 564 of this issue, Fehér and colleagues4 greatly improve on Psammeticus's experiment, using laboratory-raised songbirds.
When isolated from normal input, humans develop only basic communication systems. Deaf children raised in hearing homes spontaneously create simple communication systems termed home-sign5. Lacking semantic and grammatical complexity, such systems cater only to simple, specific needs. Similarly, adults thrown together with no common language develop pidgins — communication systems with small vocabularies and simple grammar. But intriguingly, in both cases, later generations transform such simple systems into true languages, with a full range of grammatical expressiveness. Pidgins can become complex, stable languages known as creoles in just a few generations6,7: in Nicaragua, deaf children transformed a set of simple home-sign systems into the full Nicaraguan sign language (NSL) within a few decades, under the watchful eyes and video cameras of linguists8.
But such fascinating examples raise more questions than they answer. Do the systems developed by isolated children reflect an original, biologically given form of language, as Psammeticus believed? Or is cultural transmission itself a necessary part of human language? Do children require rich lexical and grammatical input to fully activate the language-acquisition system? How do the neural networks underlying isolated languages such as home-sign differ from those that mediate communication with full languages? For ethical reasons, linguists cannot address these questions with controlled experiments, and interpretations of 'natural experiments' such as NSL remain controversial. A more biologically tractable model of cultural transmission is therefore needed.
One possibility involves investigating birdsong. Biologists have known since Darwin's time that, to sing properly, young songbirds require song input, typically during an early sensitive period preceding adulthood9. Deprived of such input, they sing only an abnormal song, with fewer syllable types and less structure than normal.
Darwin further recognized that learned birdsong nullifies traditional nature/nurture dichotomies. Birdsong is neither an instinct present at birth, nor an arbitrary cultural construction: it rests on biological foundations, but also requires specific inputs to develop properly. Songs are transmitted and transformed over several generations of learners, forming birdsong 'dialects' that vary geographically within a species10. Thus, both birdsong and language are generated by a species-typical 'instinct to learn' that constrains, but does not fully determine, the final outcome11. Birdsong acquisition provides, in this sense, an animal equivalent of human cultural transmission.
In their study, Fehér et al.4 first raise young male zebra finches in isolation. At adulthood, these birds sing only a raspy, arrhythmic isolate (ISO) song. The authors then pair these adult male 'tutors' with young males. These first-generation young learners successfully imitate ISO song, but already begin to transform it in the direction of normal — wild-type — song. Iterating this process, using first-generation song as input to a second generation of birds and so on, the authors follow the birds for up to five generations and find that each generation's song moves steadily closer to normal zebra finch song.
Fehér and colleagues performed these initial experiments with otherwise-isolated males, tutored one-on-one. To explore the consequences of a more typical social environment, the authors4 established a colony of females, socially housed and seeded with a single ISO male. Again, the songs of the descendant males transformed steadily in the direction of normal song. Intriguingly, however, these learner males' songs also incorporated their male siblings' song errors and female whistles. This 'cultural' experimental condition is more reminiscent of human sign languages such as NSL, in which a critical mass of young signers is needed to spark the development of a rich lexicon and grammatical complexity typical of true human language12.
Of course, zebra-finch song is not language: birdsong is not used to express complex meanings; has only a simple syntax; and is typically produced only by males. But the techniques and model Fehér and colleagues introduce exploit a mechanism — cultural transmission of vocally learned signals — shared by both systems13,14. This new approach will allow detailed mechanistic explorations into how iterating the process of song acquisition can lead to something more diverse and structured than any individual bird creates on its own. What are the differences in brain circuitry between ISO birds and their fourth-generation descendants? In later generations, will we see better-developed song circuitry, or different activation patterns of the genes involved in vocal learning? What about females: do they prefer normal song, regardless of their upbringing, or does song resembling their father's ISO song excite them more?
Ultimately, addressing these issues will allow us to uncover the detailed biological bases of songbirds' instinct to learn song. If discoveries concerning genetic similarities in birdsong and human speech15 are any guide, such work may also provide intriguing hints about the biological mechanisms underlying the language-acquisition system in humans. Thus, Fehér and colleagues' birdsong studies provide a pioneering approach for exploring the biology of culturally transmitted systems in the laboratory.
About this article
Nature Reviews Genetics (2011)