Tuesday, 14 November, 2000
L&C VI � Animal Language
When philosophers and theologians search for a human quality that elevates us above animals, they most often turn to language. The complexity of language, with its endless permutations and incredible expressive power allows us to plan and interact in ways that set us apart from the mute beasts. Yet at the same time, great efforts have always been made to scale the language barrier dividing us from various highly intelligent marine mammals and primates.
What criteria do we look for when we talk about animals acquiring language? The actual vocal production seems irrelevant � certainly we don�t deny that deaf-mutes are language-users. Indeed, we wouldn�t even require that we can communicate with the animal, just that members of the species can communicate with each other with a level of subtlety similar to ours. However, we would still want to be able to judge how complex an animal communication system is, even if we can�t understand it (yet). Communication (especially at an animal level) is about conveying meaning to another agent to influence its behaviour. This is true of distress calls, infants� hunger cries, ants� pheromone trails and human discussion. We could measure the complexity of a communication system by the complexity and range of behaviour it can trigger. In the example of ant�s pheromone trails, the behaviour may be complex, but the range of behaviour it can cause is limited to path-following, perhaps also conveying some idea of what lies at the end. A humpback whale�s song may last for minutes and minutes, but the relatively invariant patterns and pitches form a series that conveys relatively little information.
In order for communication to be linguistic, we look for arbitrarily-defined symbols to be bound by a syntax, a set of rules governing how the symbols can be combined which give rise to infinite variations of meaning. For instance, the shape of the pheromone trails cannot be modified in order to convey the ant�s mood or the weather - because the ants lack the syntactic richness, their communication is sub-linguistic and very restricted in the forms of cooperation it can facilitate.
Given that it seems certain that humans have at least some degree of hard-wired language skills, it may be foolish to expect animals to talk to us on our terms. And if they have some degree of innate linguistic capability, it is likely to take a very different form and structure to ours. Therefore, it makes sense to me to look first at what animals are capable of in the wild � if communication between them appears rudimentary, then it is expecting even more of them to try and bring them towards our delicately and laboriously evolved language.
So we would expect the adaptability and level of cooperation of a group to indicate the complexity of their communication mechanism. As vocabulary grows and the rules combine to generate more abstract meanings, language becomes a means for bypassing the slow learning mechanisms that evolution is equipped with, and generations are able to accumulate and build upon the knowledge of their ancestors. It is tempting to get embroiled in an evolutionary debate about whether language preceded tool use or vice versa, but certainly the two drove our evolution of larger brains.
Whalesong has proved a source of fascination, as well as a few surprise chart-toppers. Certainly, a casual listener cannot make any sense out of the cacophony of pops, squeaks and whistles of differing pitch and pattern. Yet Roger and Katy Payne�s 20 years of recordings of humpbacks in Bermuda have shown that their song can be broken down in some ways. Humpback whales tend to migrate to the same set of breeding grounds every year, travelling in pods. The breakthrough came when they realised that there was a particular song of up to half an hour�s length that was being repeated by all the whales in the breeding ground. Although the song evolves slowly but constantly over a period of months and years, every whale in the breeding ground will keep up with the latest changes. It has proved extremely difficult to detect whether this change is led by any particular dominant whale(s) or not. They have broken the song into lengthy themes, then phrases and then units, as a means of comparing different songs. They found that although the themes always came in a particular order, they could be skipped sometimes. Further work has been done, trying to relate the whales� behaviour (a complicated pattern involving �aunties�, escorts travelling with a mother and calf) to the sounds they make, e.g. aggressive activity between two escort males will lead to a variety of slaps, bubbles, vocal social-sounding calls, song segments, trumpeting and grunts before up to fifteen other whales stream to join, trying to jostle the principal escort. The humpbacks don�t require a particularly complex communication system, since they are promiscuous grazers, so they make fewer social calls (in contrast with the gibbon duet � see below) and don�t use group-wide cooperation for hunting like dolphins, for instance. The richness of the sounds of whalesong seems convincing evidence that there is more to it than simply a shibboleth or spacing mechanism. Perhaps comparing it with our own music (like a serenade) might be nearer the truth, since animal song plays such an important role in mating in many species.
Tyack enlisted the help of the US Navy with more success, using playbacks of the social calls, which brought the singing whales to home straight in on the loudspeaker at 12km/hour. He considered that the song may be used as a spacing mechanism for courting males, much like the larger baleen whales in their scattered herds. These use very pure and very loud 20Hz blips, originally identified as being whale sounds by the pattern of 2� minute periods of rest every quarter of an hour, the breathing pattern of a large, slow-swimming whale. Schevill believes that these sounds are used to locate each other over tremendous distances, perhaps even 3000 miles (if they make use of deep sea channels to focus the sound), since fin whales travel in very widely scattered herds. Such a low frequency signal would be ideal for long-distance communication in water, bouncing off the sea bottom and being propagated through the surface-frozen polar ice. Payne and Webb have calculated that the increase in underwater low frequency noise from human sea traffic may have had unimagined consequences for the breeding patterns of these creatures, since their range may now be down to as little as a few hundred miles.
Even the rapid clicking of the Californian gray whale and the sperm whale is probably used largely for navigating and locating food. However, these morse-like sounds appear to play an important role in the sperm whale�s complex social structure (which would fit in with the use of higher frequency sounds for close interaction), as a form of identification, and maybe even allowing a dominant male to lead.
However, despite the indecipherability of whalesong, its functions tend to be fairly basic. Though it may be tempting to imagine that their huge brains are indicative of a rich mental life belied by clumsy appendages and an environment ill-suited to tools, initial observations point to a less entrancing reality. Whales are able to keep in contact over tremendous distances and indicate emotional state, distress, identity and the like, whalesong does not appear to be far removed in function than the gibbons� duets to each other when demarcating their territory. However, it is difficult to see why it should be adaptive to keep each other constantly abreast of changes to the communal song, though it did allow Southern right whales to identify and ignore synthesised or humpback variations on their songs.
Dolphins� neocortex comprises 98% of their cortex, compared to 95% of man�s and 69% of kangaroos� for example, although theirs is thinner than ours. So as you would expect, dolphins� communication definitely appears more complex than most whales. They have an extremely varied vocal repertoire of both pulsed and unpulsed vocalisations, comprising squawks, whistles, squeaks, blips, groans, clicks, barks, rattles, chirps and moans. Echolocation requires the dolphin to produce and decode the echoes from up to 700 units of sound per second (which coagulates into about 20 or 30 units when we listen) amidst the noise of the underwater world. Norris demonstrated the acuteness of this echolocation sense when it was found that a blindfolded dolphin could dodge obstacles, pick up food, differentiate between copper and aluminium plates of the same colour and between hollow and solid objects. This incredibly difficult task might seem to be the reason they have huge brains � yet, as it turns out, bats manage a similar task flying at 40mph, and their brains weigh less than a gram.
Norris and Brownley studied the social structure and behaviour of groups of Hawaiian spinner dolphins in the wild in the Pacific, distinctive for their twisting leap. Certain patterns have become very clear. They stay within a tight group, during the day when socialising and playing, and at night, when hunting cooperatively. Early in the day, they swim slowly and close, making a few clicks. During the afternoon, when they are becoming highly active, they begin to swim in zig zags, using higher frequency sounds (whistles and burst-pulses). The whistles and overall noise grow to a frenzy, until the group suddenly leaves in unison to hunt (using clicks alone for navigation and echolocation).
It is clear that the sounds being made play an important role in the group�s activities. The excited whistles may indicate their emotional state, as well as identifying the sender. Whistling is also associated with feeding � indeed, bottlenosed dolphins in captivity whistle in synchrony with feeding and training routines. The raucous whistling previous to the hunt may be a way of readying and rallying the group together. The clicks seem to serve a more functional purpose, aiding navigation and echolocation, but probably also identifying each other and indicating positions, just as the Californian gray and sperm whales use clicks when the group comes close together. Attempts to correlate sounds with emotional content have linked pulsed yelps with courtship intimate chuckling during caressing, squeaks during training in captivity with frightened or alarmed squeaks, aggressiveness with buzzes and non-vocal jaw-clapping or tail-slapping.
Indeed, when the continuous bouts of whistling from a tropical spotted male dolphin that had been caught were played back to his school, they fled instantly. However, when the same alarm calls were played back to a different school, they showed only curiosity, indicating that dolphins� local dialects vary as much as killer whales�. Apocryphal evidence of dolphins avoiding boats of a similar shape to one that was used to kill a dolphin the week before supports the claim that dolphins in the wild can communicate more information than simply their own characteristics and emotional state. Indeed, it seems difficult to imagine that dolphins would have evolved the capability to produce such a wide variety of sounds and to use them so much in their interaction with other dolphins if the information content of their chattering could be conveyed by just a few stereotyped tones.
However, because of the difficulties of identifying the source of vocalisations underwater and observing behaviour, experimenters have focused on dolphins in captivity. Bastian (1964) conducted a famous experiment with two dolphins, able to hear but not see each other. The female was taught to push a paddle for food, and within a short space of time, the male had learned to do the same, presumably with the aid of the noises the female was making. To argue that the male was able to do this without there being an existing mode of communication would require that the male decode the female�s sounds from scratch, which would be even more impressive and implausible. In a 1974 Russian study, two bottlenoses which couldn�t see each other communicated with an opening self-identifying section followed by a more complex series of clicks which appeared to convey the message. This pattern of communication has parallels with birdsong, and with a little imagination, even with the packets of data transferred by TCP/IP, the transmission control protocol used by modems � each packet has an identifying header containing information about the sender and destination, and the message is sandwiched inside.
Some small progress has been made in dolphin communication. The Flipper Sea School, home of the eponymous TV hero, constructed a whistle language, similar to that used by sheepdogs. Hemen, Wolz and Richards went further with two female bottlenose dolphins. Taking advantage of dolphins� good auditory memory for long lists of sounds, they were able to marry visual and verbal signals with tasks. Most encouragingly of all, the dolphins actually started to talk back using these signals. Such interchange seems a vital hallmark of language. John Lilly probed dolphins� propensity to mimic by getting them to follow a count up to ten and recognisably say simple English words. Although the dolphins sounded significantly less human than mynah birds and budgerigars, the difference is that the dolphins were able to learn new words almost instantaneously.
Killer whales, like dolphins, porpoises, sperm and pilot whales, are odontocetes. They travel in pods of less than 50 which don�t change or mingle very much. Fisher and Ford have shown that the sounds of killer whale pods are very stable, comprised of about 12 distinct, stereotyped calls, that they use to communicate when spread over a couple of miles. It has proved difficult to link sounds with particular meaning or behaviour, beyond possibly position, identity, emotional/activity state and pod identity. The stability of killer whale pods may be linked to the stability of their sounds � in contrast, humpback whales with their choice of breeding grounds and inter-mingling have a constantly evolving song. Van Heel taught Gudren the killer whale to associate a certain object with a specific sound, then added action words (verbs) to create new combinations. Apparently Gudren began to give the signals back, though with less success than some of the higher primates taught ASL like children.
To summarise broadly the case for marine mammals, baleen whales use their deep pulses as a means of keeping in contact over huge distances, dolphins use sounds to socialise and cooperate in hunting, various marine mammals use songs as a means of spacing out or keeping together a group and we can draw approximate links between some types of sounds and the context they are almost always used in, e.g. distress calls, courtship. The case of marine mammals might then seem a relatively clearcut one. It is tempting to argue that many of the most complex sounds that we can see structure in are largely stereotyped and do not vary enough to carry a range of meaning, and that when a wide repertoire of sounds is encountered (e.g. socialising dolphins), that it is the overall volume and type of sounds that signifies a group emotional or activity state, rather than that each series of calls and clicks could one day be translated as �Hey Jim, has your fin healed?� and �Yes, it�s fine. Look how much further out of the water I can jump today!�, or words to that effect. However, some of the examples of more complicated human-killer whale and dolphin-dolphin interaction in captivity hint that these animals are partially capable of comprehending and assembling constructions with quite specific meanings.
One good reason to think that the first progress we will make in breaking down the language barrier separating humans from the rest of the animal kingdom will come with primates is that we share their sensory apparatus and environment. We can treat chimps and apes literally like human children, allow them to learn language like human children, give additional help where needed 24 hours a day, interact with objects using similar bodies and with modified ASL (American Sign Language), even teach them a language as rich as our own verbal one.
We see parallels of almost all of the marine mammal communication mechanisms in primate vocalisations, and also in birdsong. Fiercely territorial gibbons and Madagascan indris sing a tremendous duet at dawn to delineate each other�s territories and cement the pair bond. New World Howler monkeys howl loudly in packs to space each other out, though their territories are not fixed; the same is true of the widow tits, perhaps because their more barren environment requires them to range further than can be defended as territory. More social monkeys use stereotyped distress calls, sometimes to indicate specific predators. But it is long-term studies with individual monkeys living with their researchers that have seen potentially the most exciting progress.
When looking at individual primates in captivity being taught rich, full language systems like ASL, more appropriate and rigorous criteria are needed for defining �language use� than gauging how precisely animals can affect each other�s behaviour by communicating. Hockett (1960, 1966) devised a set of 13 linguistic universals, criteria which he believed had to be fulfilled in order to say that language was being used. I have listed what I believe are the seven key properties (ignoring: duality of patterning; auditory-vocal channel; broadcast transmission and directional reception; discreteness; rapid fading):
semanticity and arbitrariness � the symbols "convey meaning through their stable association with realworld situations" and not through any inherent connection to their referents
productivity � the ability to produce sentences that haven�t been heard or used before, i.e. to use language to express something new that can be understood by the listener/audience
interchangeability � conversation/dialogue between speakers, both taking the roles of speaker and listener
specialization � can say something or use language instead of performing an action, e.g. give a verbal warning rather than a physical attack or show of aggression
displacement � can talk about things that aren�t there at the moment
cultural transmission � information passed on to next generation, i.e. learning can be conducted by language without experience
These leave a very broad picture of language. They don�t explicitly talk of syntax though. This may or may not prove universal to languages in general; human languages are certainly wholly dependent on the application of rules to subtly alter the meaning of a sentence (e.g. compare �I hit Jamie� with �I was hit by Jamie�). However, I am reluctant to stipulate that our notion of universal language requires the perceptual stream (whether of gestures, sound, stench or whatever) to be divisible into discrete units, since I discuss below how a definition of syntax would be stretched by the very different (physiological) mechanisms involved in animal or alien language. It should be interesting to see how the results of various primate studies measure up to these criteria. It should be noted that ASL, for instance, would not have qualified as language according to Hockett�s full list of universal criteria (duality of patterning, auditory-vocal channel).
Washoe, brought up from the age of about 10 months by Allen and Trixie Gardner like a human child, was the first really serious contender for a linguistically-competent primate. Robert Yerkes had experimented in the 1920s with teaching apes to speak, but had concluded that it is impossible. Indeed, probably the most successful was Viki, a chimpanzee taught by Keith and Cathy Hayes to speak four words (Gardner & Gardner, 1989). It wasn�t until almost forty years later that Yerkes� suggestion to try some form of sign language instead, in recognition of primates� manual dexterity. They taught Washoe ASL, which replaces speech in deaf humans, all of whom are indisputably card-carrying language-users. They taught her by using sign language whenever in her presence, as well as playing fun and familiar games and physically moulding her hands into the required shapes, rewarding language-learning success with food treats and approval. This long-term relationship-building has proved crucial, as with human children, in the more successful studies.
By the age of five, she had mastered 133 signs. She had been spontaneously combining signs after learning only 8 or 10. She had learnt to use the sign for �more� in various contexts like �more food�, having been taught it with reference just to �more tickling� (Gardner & Gardner, 1979). Indeed, a 1974 video clip seemed to capture Washoe responding to a baby doll inside a cup with �Baby in my drink�.
By this time, word of Washoe�s ability had reached skeptical ears. Herbert Terrace set out to settle the issue once and for all with Project Nim, and its hero, Nim Chimpsky. Nim was raised in a similar way, though he was rewarded with approval rather than food, which Terrace later recognised may have adversely affected Nim�s motivation to sign (Terrace 1979). Nim learned many words, used signs in the absence of their referents (displacement) and learnt that signed warnings could be heeded without accompanying aggression (specialisation). Critically however, Nim appeared not to generate new sentences himself � reviewing the video evidence suggested that on the occasions when Nim appeared to pair words in a meaningful-seeming way, his trainers had been responsible for prompting him somehow. Terrace concluded reluctantly but adamantly that Nim�s and Washoe�s achievements could be explained in Skinnerian behaviourist terms, and that use of a generative grammar was beyond them.
There is an interesting epilogue to Washoe�s story. The Gardners taught several other chimps to sign, but they wanted to know whether Washoe was able to pass on her linguistic knowledge to offspring. Unfortunately, both Washoe�s infants died. Roger Fouts (who was now looking after her in her spacious new home at the Central Washington University) tells a heart-rending story of how he had to tell her the news about her second infant:
And so I went in, and she came up to me, her eyes lit up. She came up to me and she said, `Baby, holding, holding.' And it was a question, she was saying, basically, `Where's my baby?' And I had to tell her, I said, `He's dead. He's finished.' And with that, the baby sign literally dropped into her lap, her head dropped, and she moved away into the corner and stopped signing.
The Fouts were able to find Loulis, a 10 month old infant, for her to adopt. Over the next five years, no sign language was used by humans in Loulis� presence, yet he was still able to learn 51 signs just from the other chimpanzees, more by picking them up like a child learning language than by any active teaching (Ingersoll). This cultural transmission led the Gardners to conclude that "once introduced, sign language is robust and self-supporting, unlike the systems that depend on special apparatuses such as the Rumbaugh keyboards or the Premack plastic tokens" (Gardner & Gardner 1989: 25).
Lana the chimpanzee was taught by Duane Rumbaugh to use an electronic keyboard with specially-designed lexigram keytops, rather than ASL, as a means of producing sounds in response to human speech. Using a keyboard as Lana�s means of expression was a means of addressing criticisms made of Washoe�s �sentences� that she was simply wiggling her fingers hopefully and that her trainers were projecting meaning onto them. Rumbaugh also discusses its usefulness in being able to vary the training procedures in some way (Rumbaugh, Warner & von Glasersfeld, 1977). Lana spontaneously used �this� to refer to things for which she had no name (Rumbaugh & Gill 1977), as well as using �no� as a protest, having been taught it as a logical negation. Lana�s trainers make the statement:
we cannot assume that Lana 'understands' the meaning of every word she uses as we do, but the consistent appropriateness of her novel sentence constructions ... support the conclusion that she has conceptual meanings for many of them and also for their relationships (Rumbaugh & Gill 1977).
The reluctance to ascribe meaning to Lana seems to illustrate Stuart Shanker�s complaint that linguists are prejudiced against chimp linguistic skills like putting together a noun and a verb to form a two-word sentence, although they would consider these to be nascent linguistic abilities in a very young child. In the same way, we should not look to subjective feelings of what a word means and demand that of primates in some way � not only can we not test for this, but we should not deny that our hairy cousins have a basic but real grasp of a word�s meaning if their use of it is appropriate (in context and syntax) and can be used and understood when conversing.
However, probably the most promising candidate for inter-species communication is Kanzi, a member of the rare Bonobo monkey species (genetically closest to humans). Kanzi is the adopted son of Matata, a female Bonobo. Matata, originally the subject of the study, had not taken well to using the �talking keyboard� Savage-Rumbaugh was teaching her, and had mastered few lexigrams in two years (at 10 years old, she was probably too old for language acquisition). On the other hand, Kanzi seemed fascinated by it, and in the same length of time had started using the keyboard to initiate games of �chase�. He had picked up a few of the keys just by listening to the teaching given to his adopted mother before even being officially trained. After a year and a half, he had amassed a vocabulary of fifty symbols, and was regularly combining words. These multi-word utterances appeared to demonstrate genuine comprehension, and 90% were spontaneous. He also used three word utterances to indicate a recipient of an agent�s action such as tickle or chase.
The next major breakthrough was Kanzi�s apparent understanding of human speech, even though no attempt had been made to train him to be able to do this. Washoe had been reportedly able to understand human speech, but Terrace�s criticisms of Project Washoe had all but dismissed the claim. This transcript of Kanzi�s first ever telephone conversation with Janine Murphy seems undeniable evidence that Kanzi�s comprehension of English is high, certainly far in excess of his vocabulary on the Talking Keyboard:
Ms. MURPHY:� Kanzi, this is Janine. Would you like any food? Tell me what food you'd like.
TALKING KEYBOARD:� Food surprise.
Ms. MURPHY:� Some food surprise?
TALKING KEYBOARD:� Food surprise.
Ms. MURPHY:� Kanzi, would you like a juice, or some M&Ms, or some sugar cane?
TALKING KEYBOARD:� M&Ms.
Ms. MURPHY:� You like M&Ms? Okay. Kanzi, is there any other food you'd like me to bring in the backpack?
TALKING KEYBOARD:� Ball.
Ms. MURPHY:� A ball? Okay.
Such a conversation with someone not physically present (displacement) about events in the future had been thought beyond the abilities of chimps, and indeed non-humans in general.
Savage-Rumbaugh took advantage of 55 acres of forest to attempt to create a more ape-like environment for him, in the hope that it would help him develop the cognitive capacities that would be exercised in the wild. She created certain feeding areas, and Kanzi travels from place to place, talking about where they were going and doing next and what kind of food they were going to eat. This is intended to tap into what she sees as an innate primate desire to predict what is going to happen, using language. She highlights the importance of enculturation, the accumulution of nonverbal cues, attention-directing maneuvers, and repetition, very important aspects of the way we speak to young children.
After Kanzi passed a simple test of identifying vocalised words with pictures with 90% accuracy, Savage-Rumbaugh put together 600 deliberately unpredictable sentences, and measured Kanzi�s ability to understand them by his ability correctly perform them. She compared his performance on them with Janine Murphy�s two-year old daughter, Alea. Both chimp and child were correct about three quarters of the time, though their strengths and weaknesses varied from sentence to sentence. They concluded that Alea and Kanzi�s linguistic comprehension was roughly on a par.
Savage-Rumbaugh believes that it was at roughly the point that bonobos are at now when comprehension so far outweighs vocal production, that we evolved the ability to say consonants (critical in differentiating sounds) to match. Savage-Rumbaugh believes that it is almost certainly because Kanzi is a bonobo chimp that he is so far ahead of previous studies with apes and common chimps:
Results indicate that the propensity of the pygmy chimpanzee for the acquisition of primitive language skills is considerably in advance of that yet reported for other apes. Language acquisition in the pygmy chimpanzee seems to be accompanied and facilitated by the ability to understand spoken English.
Bonobos are the most visually human-seeming of all apes, in stride, stance, resting postures, gestures and facial expressions. They are notably less aggressive and more social than common chimps, and have been observed congregating regularly in groups of two or three hundred, conducting constant vocal exchange, almost like gossiping. Whether or not you want to accept the claims made about Kanzi�s teetering on the brink of human language depends on whether you emphasise comprehension or production in your view of linguistic competence.
Lastly, it makes sense to examine briefly birds� claim to language use, since their intelligence and capability for vocalisation is considerable. Bird brains are large in relation to body mass; some bird brains are 6-11 times larger than like-sized reptiles, and up to 9% of body mass, a similar proportion to higher mammals. A further sign of intelligence is their tool use, at a comparable level with primates in Egyptian vultures, New Caledonian crows and bowerbirds, for example. Intelligence does vary widely from species to species: the New Zealand alpine parrots, the kea, have been filmed solving a puzzle in minutes that has taken tits, pigeons and crows anywhere from days to months to complete (Beynon, BBC). Their propensity to mimic is amply demonstrated in the well known example of the stealing of cream from the top of milk bottles by British blue and great tits. First observed in the 1920s, within 30 years the learned behaviour had spread throughout virtually the entire country.
Irene Pepperberg and her African grey parrot, Alex, are evidence that some birds are also able to learn to communicate with probably as much success as chimps. Alex has the major advantage of being able to vocalise English words, and the 20-year-old bird can ostensibly count to six and recognize and name 100 different objects, along with their color, texture and shape. Alex does occasionally seem to use phrases without meaning them, but he certainly seems to mean most of what he says, such as �Wanna go back� preceding climbing onto the back of a chair. This conversation ensued when he was presented with a metal key and a green plastic one (Scientific American, 1996).
Pepperberg teachers Alex and his younger companions, Griffin and Kyaaro in a threesome � herself, the bird and a student. One person holds up an object, the other names and then receives it. The bird receives the new toy by listening and learning the word. Pepperberg now uses Alex in the second teaching role instead of a student � �he rarely makes mistakes when in this role, and Kyaaro and Griffin learn faster from him than from humans�.
Pepperberg does not claim that all parrots or even all Greys are as capable as Alex, but that the psittacine capacity for language is greater than expected, especially with appropriate support. She notes that the reputation parrots have for mimicry without meaningfulness may be based on unsuccessful experimental techniques in the 1950s and 1960s (e.g. Mowrer, 1950). In the wild, it seems that vocalisations are important in social interactions between mated pairs and among flock members, and that (like vervets), both physical structure and usage of these vocalisations appears to be learned.
Birdsong may eventually be shown to be far more complex and informative than we realise now. Indeed, its complex vocalisations may give birds an advantage over other animals as an advanced vehicle for communication. However, progress in inter-species communication like with Alex is likely to lag behind primate studies for the most part since most birds (like the intelligent marine mammals) lack the capacity to converse with either speech or ASL.
The more phylogenetically distinct animals are, the more different their bodies, motor systems, perceptual apparatus (especially their auditory world) and brains will be. Whales and dolphins, though mammals, are obviously very different, and live in a genuinely different environment. This raises a legitimate question as to whether we could really identify a complicated linguistic-like system if we heard it in action. Earlier, we decided to use the degree and manner in which information communicated by animals to each other was able to influence their behaviour as our operationalist criterion. However, we must admit that our ability to observe is biased and limited by our inability to comprehend their communications. Learning animal language is compounded by the fact that we don�t know what we are listening for, unlike infants, whose speech perception appears to home in on the universal human phonemic inventory then pare that down to the phonemes of the language they are exposed to. Just as, for instance, we find it difficult to define life because we only have the example of terrestrial life to look at, an evolutionary tree with possibly only one root, attempting to abstract to the universals of language from one specimen may be dangerous.
Dolphins, for example, are capable of producing and perceiving a wide range and rapidity of sounds that overlaps but extends beyond our auditory capabilities. This is not simply a case of talking faster or in a higher pitch, or even using a mixture of clicks and whistles in place of our phonemes � it may be that the form a dolphin language does (or might eventually) take could be very different to ours. One of Hockett�s most anthropocentric criteria of language was that it be composed of discrete units, which we can divide up in human language at the level of words, morphemes or even phonemes. This may not be the case with dolphins. It may be that their sounds overlap in some way, forming a series from which distinct units cannot be extracted. We can see this to a limited extent in the difficulties of developing computerised speech recognition � although the phonemes we hear sound stereotyped and distinct, in fact they blend into one another as our tongue and articulatory mechanisms take shortcuts in moving from the production of one sound to the next. Or perhaps the temporal patterning gives a clue � at the most basic level, frequency is pitch. But we assume that a conversation conducted in rapid chatter conveys the same meaning as a laboured or leisurely discussion that dolphins with their far more acute auditory temporal resolution will not take advantage of it. Similarly, birds that are able to produce two sounds at once may employ this capability in their syntax in some way.
When Wittgenstein said �If a lion could talk, we could not understand him� (Philosophical Investigations), he was referring to the way that linguistic meaning is bound up with what it is like to be human. More recently, Douglas Adams quipped along similar lines that although dolphins turn out to be as intelligent as us and live equally complex lives, we never realised it because we couldn�t understand why they didn�t build cars and buildings and machinery � they never realised it about us because we did. This consideration goes a step beyond the difficulties of bridging the gulfs between our respective perceptual apparatuses discussed above � it might be that we can never understand what a very different animal to us is saying because we can never relate the meanings of its words to the humanness of our lives. Lion-, dolphin- or chimp-language is so inextricably tied to being a lion, dolphin or chimp that we could never translate or interpret their language in terms of our experiences except in such broad terms that we lose all the intricate meaning being conveyed.
We might imagine aliens adapted to live in a Jovian atmosphere whose closest sense to vision is a sensitivity to magnetic fields and whose language appears to rely on complex tonal harmonies. With almost no natural predators but a phenomenally complex social structure, we are able to observe with difficulty through the thick, gaseous haze that this alien song is closely linked to their emotional/activity state and courtship. We identify it as a pre-linguistic sort of pairing bond or serenade, and cannot see that its subtle poetry is heavily dependent on the context of exactly what has been said before, and that amplitude-modulation within a very very small range differentiates meaning.
This rather extreme thought-experiment is merely intended to bring home that we still can�t be too sure about marine mammals, and there�s definitely something (relatively primitively) linguistic in chimps� use of ASL and talking keyboards when communicating with us and with each other. Perhaps the reason that Savage-Rumbaugh�s slightly exaggerated claims about breaching the ape-human divide are so derided is a political and emotional one � when language is no longer the stronghold of humanity, then our ethical anthropocentricism will necessarily be regarded in the same light that slavery is now (Cavalieri & Singer 1993).
 Strictly speaking, French physician Julien Offray de La Mettrie had first conceived of teaching apes sign language in the 18th century.