Collaborative Research: Learning new sounds - understanding and overcoming phonological interference in songbirds and humans

How do people communicate successfully with other people? Part of being understood is pronouncing words so they are not confused with other words. The difference between back and pack in English, for example, is just a single sound, but if someone does not get that difference right they may not be understood. Although young children learn such differences in their first language easily, adults learning a new language typically do not. Is it possible to improve adults' learning of sounds in other languages so that people will be able to communicate more easily and more successfully? This study uses information from experiments with songbirds to advance a new hypothesis about how learners – whether they are humans or birds – learn new sounds. Just as humans can be bilingual, some bird species can be “bilingual” as well: they are capable of learning new songs after early life. This study proposes that a similar way of learning sounds has evolved in both songbirds and humans, and combines “second song” learning experiments in zebra finches and second language learning experiments in adult humans. The research will help to better understand how vocal communication systems evolve across different species and how to improve language learning outcomes in human adults. The project will also train high-school and undergraduate students in interdisciplinary research skills through involvement in birdsong and speech learning experiments. This project also advances NSF’s priorities in Artificial Intelligence. Both songbirds and humans can learn multiple songs or languages, an important adaptation to life in complex social groups. In humans this ability has a paradoxical downside: the sounds that are learned first, in one’s native language (L1), can interfere with learning similar sounds as an adult in a second language (L2). Three important features of phonological interference remain ill-understood: what the mechanisms are that cause such interference, whether phonological interference is uniquely human or shared with other animal vocal learners like songbirds, and how to reduce interference and promote effective learning. This collaborative study will advance the understanding of the mechanisms underlying phonological interference, the generality of interference across species, and ways to reduce or eliminate it. The study investigates the hypothesis that the mechanism responsible for subtle changes in native sound categories that occur throughout life (e.g., to accommodate variability across and within speakers) directly interferes with learning new sounds. The additional hypothesis is that a similar mechanism has evolved convergently in highly social songbirds, causing interference in learning new songs. To test these hypotheses, and to derive an auditory training method for reducing interference in both birdsong and speech learning, the study combines experiments inducing second song learning in zebra finches, with a multi-session study of L2 sound learning in adult humans. The study will yield insights into how life-long adaptation of previously learned skills affects de novo learning in humans and other animals and can yield the first songbird-derived method for improving L2 speech sound learning. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria. NSF Award ID: 2536872 | Program: 01002627DB NSF RESEARCH & RELATED ACTIVIT | Principal Investigator: Virginia Valian | Institution: CUNY Hunter College, NEW YORK, NY | Award Amount: $174,073 View on NSF Award Search: https://www.nsf.gov/awardsearch/show-award/?AWD_ID=2536872 View on Research.gov: https://www.research.gov/awardapi-service/v1/awards/2536872.html