Vocal learning is the ability to modify acoustic and syntactic sounds, acquire new sounds via imitation, and produce vocalizations. "Vocalizations" in this case refers only to sounds generated by the vocal organ (mammalian larynx or avian syrinx) as opposed to by the lips, teeth, and tongue, which require substantially less motor control. A rare trait, vocal learning is a critical substrate for spoken language and has only been detected in eight animal groups despite the wide array of vocalizing species; humans, bats, cetaceans, pinnipeds (seals and sea lions), elephants, and the three distantly related bird groups songbirds, parrots, and hummingbirds. Vocal learning is distinct from auditory learning, or the ability to form memories of sounds heard, a relatively common trait which is present in all vertebrates tested. For example, dogs can be trained to understand the word "sit" even though the human word is not in its innate auditory repertoire (auditory learning). However, the dog cannot imitate and produce the word "sit" itself as vocal learners can.
Classification
Historically, species have been classified into the binary categories of vocal learner or vocal non-learner based on their ability to produce novel vocalizations or imitate other species, with evidence from social isolation, deafening studies, and cross-fostering experiments. However, vocal learners exhibit a great deal of plasticity or variation between species, resulting in a spectrum of ability. The vocalizations of songbirds and whales have a syntactic-like organization similar to that of humans but are limited to Finite-State Grammars (FSGs), where they can generate strings of sequences with limited structural complexity. Humans, on the other hand, show deeper hierarchical relationships, such as the nesting of phrases within others, and demonstrate compositional syntax, where changes in syntactic organization generate new meanings, both of which are beyond the capabilities of other vocal learning groups.
Vocal learning phenotype also differ within groups and closely related species will not display the same abilities. Within avian vocal learners, for example, zebra finch songs only contain strictly linear transitions that go through different syllables in a motif from beginning to end, yet mockingbird and nightingale songs show element repetition within a range of legal repetitions, non-adjacent relationships between distant song elements, and forward and backward branching in song element transitions. Parrots are even more complex as they can imitate the speech of heterospecifics like humans and synchronize their movements to a rhythmic beat.
Continuum hypothesis Even further complicating the original binary classification is evidence from recent studies that suggests that there is greater variability in a non-learner's ability to modify vocalizations based on experience than previously thought. Findings in suboscine passerine birds, non-human primates, mice, and goats, has led to the proposal of the vocal learning continuum hypothesis by Erich Jarvis and Gustavo Arriaga. Based on the apparent variations seen in various studies, the continuum hypothesis reclassifies species into non-learner, limited vocal learner, moderate vocal learning, complex vocal learner and high vocal learner categories where higher tiers have fewer species. Under this system, previously identified non-human vocal learners like songbirds are considered complex learners while humans fall under the "high" category; non-human primates, mice, and goats, which are traditionally classified as non-learners, are considered limited vocal learners under this system. Recent work, while generally acknowledging the usefulness of this richer view of vocal learning, has pointed out conceptual and empirical limitations of the vocal learning continuum hypothesis, suggesting more species and factors should be taken into account.
Evidence in various species
Known vocal learners
Birds
The most extensively studied model organisms of vocal learning are found in birds, namely songbirds, parrots, and hummingbirds. The degree of vocal learning in each specific species varies. While many parrots and certain songbirds like canaries can imitate and spontaneously combine learned sounds during all periods of their life, other songbirds and hummingbirds are limited to songs learned during their critical period.
Bats The first evidence for audio-vocal learning in a non-human mammal was produced by Karl-Heinz Esser in 1994. Hand-reared infant lesser spear-nosed bats (Phyllostomos discolor) were able to adapt their isolation calls to an external reference signal. Isolation calls in a control group that had no reference signal did not show the same adaptation.
Further evidence for vocal learning in bats appeared in 1998 when Janette Wenrick Boughman studied female greater spear-nosed bats (Phyllostomus hastatus). These bats live in unrelated groups and use group contact calls that differ among social groups. Each social group has a single call, which differs in frequency and temporal characteristics. When individual bats were introduced to a new social group, the group call began to morph, taking on new frequency and temporal characteristics, and over time, calls of transfer and resident bats in the same group more closely resembled their new modified call than their old calls.
Primates Evidence of non-human primates acquiring new sounds is very limited. Chimpanzees in both captivity and in the wild have been recorded producing novel sounds to attract attention. By puckering their lips and making a vibrating sounds, they can make a "raspberry" call, which has been imitated by both naïve captive and wild individuals. There is also evidence of an orangutan learning to whistle by copying a human, an ability previously unseen in the species.
Some primates change their vocalizations to sound like their conspecifics- a process called vocal accommodation. After forming a pair bond, marmosets (Callithrix jacchus) adjust their calls to match those of their mate. One study found that marmosets with very disparate calls altered them more quickly, resulting in vocal similarity regardless of what they initially sounded like. A similar phenomenon was shown in an experiment where marmosets were moved to a different area. Their calls became more similar to those of nearby monkeys.
Bonobos (Pan paniscus) also show signs of vocal accommodation. A study on bonobos in zoos found that genetically unrelated individuals living together had similar barks. Individuals living in different locations had dissimilar barks. Similar results have been found in Campbell's monkeys (Cercopithecus campbelli) and Guinea baboons (Papio papio). In other species, it is unknown if regional differences in calls are simply a result of genetic variation rather than vocal learning.Snowdon, CT. (2009). "Plasticity of communication in non-human primates," in Advances in the Study of Behavior, eds M. Naguib and V. M. Janik (Burlington, NJ: Academic Press), 239–276.
Some experiments have tried to test vocal learning in primates via cross-fostering. In 1989, Masataka and Fujita cross-fostered Japanese and rhesus monkeys in the same room and demonstrated that foraging calls were learned directly from their foster mothers, providing evidence of vocal learning. However, when another independent group was unable to reproduce these results, Masataka and Fujita's findings were questioned. A cross-fostering experiment with marmosets and macaques showed convergence in pitch and other acoustic features in their supposedly innate calls, demonstrating the ability, albeit limited, for vocal learning.
Cetaceans
Whales Male humpback whales (Megaptera novaeangliae) sing as a form of sexual display while migrating to and from their breeding grounds. All males in a population produce the same song which can change over time, indicating vocal learning and cultural transmission, a characteristic shared by some bird populations. Songs become increasingly dissimilar over distance and populations in different oceans have dissimilar songs.
Whale songs recorded along the east coast of Australia in 1996 showed introduction of a novel song by two foreign whales who had migrated from the west Australian coast to the east Australian coast. In just two years, all members of the population had switched songs. This new song was nearly identical to ones sung by migrating humpback whales on the west Australian Coast, and the two new singers who introduced the song are hypothesized to have introduced the new "foreign" song to the population on the east Australian coast.
Beluga whales are nicknamed "canaries of the sea" for their ability to produce a wide range of vocalizations. They can manipulate both the melon and nasal tract to make sounds such as cackles, whistles, trills and squawks. One captive male beluga named NOC was documented spontaneously imitating the vocalizations of adult humans, several octaves lower than typical beluga sounds. Researchers determined NOC intentionally hyper-inflated his vestibular sac, normally used to keep water out of the blowhole, so he could build up sufficient pressure to produce these frequencies.
Dolphins Vocal learning has been observed in orcas (Orcinus orca). Two juvenile orcas, separated from their natal pods, were seen mimicking cries of California sea lions (Zalophus californianus) that were near the region they lived in. The calls of these two juveniles were also different from their natal groups, reflecting more of the sea lion calls than that of the orcas.
Captive bottlenose dolphins (Tursiops truncatus) can be trained to mimic sounds. One study used an underwater keyboard to demonstrate that dolphins are able to learn various whistles in order to do an activity or obtain an object. Complete mimicry occurred within ten attempts for these trained dolphins. Other studies of dolphins have given even more evidence of spontaneous mimicry of species-specific whistles and other biological and computer-generated signals.
Such vocal learning has also been identified in wild bottlenose dolphins. Bottlenose dolphins develop a distinct signature whistle in the first few months of life, which is used to identify and distinguish itself from other individuals. This individual distinctiveness could have been a driving force for evolution by providing higher species fitness since complex communication is largely correlated with increased intelligence. However, vocal identification is present in vocal non-learners as well. Therefore, it is unlikely that individual identification was a primary driving force for the evolution of vocal learning. Each signature whistle can be learned by other individuals for identification purposes and are used primarily when the dolphin in question is out of sight. Bottlenose dolphins use their learned whistles in matching interactions, which are likely to be used while addressing each other, signalling alliance membership to a third party, or preventing deception by an imitating dolphin.