IX. Intelligent communication

“ Taffy dear, the next time you write a picture-letter, you’d better send a man who can talk our language with it, to explain what it means”

“How the First Letter was Written”

by Rudyard Kipling

A hero of Carlo Collodi’s “Adventures of Pinocchio”, the carpenter Gerpetto, made his wooden marionette in silence in a room until he made the mouth of it, and no sooner was it finished than it began to laugh and poke fun at him. Starting from very early experiments on Animal Intelligence, we have considered them in silence and now it’s time to repay means of communication to our subjects and to try to understand what they can say to each other and perhaps to experimenters as well.

In this part we will consider a question of how addressed signalling work in animal societies. It is intuitively clear that many of social species have to possess complex communications. The question of the existence of a developed natural “language” in non humans remains so far obscure. Of course, just as we have seen the same problem when considering animal cultural behaviour, how to treat language behaviour in animals is heavily influenced by our decision on which definition of language to adopt. Many definitions in literature make language trait only humans’. Even following broad definitions of language, animal experts are oscillating between questions “Why animals don’t have language? ” (Cheney and Seyfarth, 1997) and “What’s so special about speech? ” (Hauser, 2001). The main difficulties in the analysis of animal “languages” appear to be methodological. In this part of the book we will shortly consider the main experimental approaches for studying complex communication in animals. What is of no doubt is that studying communicative means of different species is a good tool to judge about their cognitive abilities.

This part contains analysis of experimental approaches for studying animal language behaviour. The term “language” being attributed to animals is a point at issue so I prefer to use the notion of “language behaviour” rather than “language” in those cases when descriptions of complex forms of communication do not require the use of definite terms. We will concentrate on methodological aspects of studying animal language behaviour and on the role of language behaviour in animal intelligence.

At least three main approaches to a problem of animal language behaviour have been applied recently. First, it is a direct dialog with animals based on language-training experiments. Being applied to apes and one grey parrot, this approach has revealed astonishing mental skills. It is important to note that this way to communicate with animals is based on adopted human languages. Surprisingly few is known yet about natural communication systems of those species which were involved in language-training experiments. The second approach is aimed at direct decoding of animal signals. The third approach to study animal communication has been suggested basing on ideas of information theory (Ryabko, 1993; Ryabko and Reznikova, 1996). The main point is not to decipher signals but to investigate just the process of information transmission by measuring the time duration which animals spend on transmitting messages of definite length and encoding. Application of this approach has already allowed demonstrating that a few highly social ant species possess possibly one of the most intricate forms of known animal communication, and it is a great challenge to extend these experimental schemes and approaches for studying other species.

28. CAN ANIMALS EXCHANGE MEANINGFUL MESSAGES?

Understanding “languages” of animals seems to be an attractive and hardly achievable skill for humans with which many legends are associated. The title of Lorenz’ (1952) book “King Solomon Ring” refers to a legend about King Solomon who possessed a magical ring which gave him the power of speaking with animals.

In 1661 Samuel Pepys wrote in his diary about what he called a " baboon": " I do believe it already understands much English; and I am of the mind it might be taught to speak or make signs" (as cited in Wallman, 1992).

More than two hundreds years later, Garner (1892) experimentally tried to “learn the monkey tongue very much in the same way men learn the language of a strange race of mankind”. In his book “The speech of monkeys” Garner (1892) argued that the reasoning of monkeys differs from that of man in degree, but not in kind, and if it be true that man cannot think without words, it must be true of monkeys. Garner spent much time in zoological gardens of Washington and Cincinnati conducting experiments in which he showed or gave different things to monkeys and then took them away recording by a phonograph all sounds which animals emitted during repetitive events. Garner examined several species including a chimpanzee, a spider-monkey, several rhesus macaques and others. With the help of the phonograph Garner conducted first playback experiments. He artificially changed elements of animal’s “speech” and then compared how animals react to variants of reproduced “words”. In particular, Garner insisted that he succeeded to distinguish a word that meant “food” in capuchin’s language from other ones which meant “bread” and “vegetables/fruits” (Garner argued that capuchins sound similarly when they express a desire to get a carrot, an apple, or a banana). The word “food”, as Garner gave this, was also used by capuchins as friendly greeting.

Garner anticipated some findings of present days. For instance, applying a similar experimental paradigm, Hauser and Marler (1992) found that rhesus macaques produce five acoustically distinctive vocalisations when they find food. Three vocalisations (“warble”, “harmonic arch” and “chirp”) are restricted to the discovery of high quality, rare food items, whereas the other two (“grunt”, “coo”) are given while waiting for and eating lower quality, common food items. An individual’s hunger level is positively correlated with call rate, but not with the type of call produced. That is, the type of vocalisation produced is influenced by the type of food discovered, and not by the discoverer’s hunger level (Hauser, 2000).

Attempts to identify categories in animals’ communication are based on a natural idea that the complexity of language behaviour should be connected with high levels of sociality and cooperation in animals’ societies. In sixties and seventies elegant but ambiguous experiments were conducted with highly social intelligent animals such as chimpanzees and dolphins that were asked to pass some pieces of information to each other.

In Menzel's (1971, 1973 a, b, 1975) classic experiments a group of chimpanzees living in an enclosure searched for hidden food (see also Chapter 12). The experimenter placed a piece of food hidden in a box in view of a leader. An informed animal succeeded in leading others to the reward by drawing attention to it through actions such as tapping others on the shoulders or repeatedly glancing at them while heading in the direction of the food. Eventually, those chimpanzees who were naï ve to the location of the bait seemed to have learned to recognize individuals most competent in finding food and followed them until rewarded. Menzel did not explicitly test whether the followers understood that the leader knew the location of the hidden food. However, he argued that chimpanzees somehow learn from the leader what kind of the reward was hidden. They more readily searched for fruits than vegetables, and displayed signs of exciting and fear in anticipation to find a scaring toy (a snake) hidden in the box. Menzel suggested that chimpanzees possess means for transferring information about both location and properties of objects but it remained unclear how they do this.

The dolphin experiments carried out by Bastian (1967) and Evans and Bastian (1969) concerned cooperative behaviour of two young dolphins which could involve intelligent communication. Researchers tried to teach a male dolphin, Bazz, and a female, Doris, to communicate through an opaque barrier. First of all, while they were still together, the dolphins were taught to press paddles when they saw a light (session A). If the light was kept steady, they had to press the right-hand paddle first; if it flashed then the left-hand one. When they did this correctly they were rewarded with fish. At the second stage (session B) of the experiment, they were taught to follow a sequence: Doris had to wait for Buzz to press the signal first. When the light came on under those conditions and stayed on, Doris waited for Buzz to hit his signal. When he didn't, Doris eventually made a sound, and soon after Buzz pushed the right signal. Doris followed suit, and they got their fish. As soon the dolphins had learned this manoeuvre, they were separated (session C). Now they were situated in adjusted pools that allowed them to hear but not see each other. The paddles and light were set up in the same way, except that the light which indicated which paddle to press first was seen only by Doris. In order to get fish both dolphins had to press the paddles in the correct order. Doris had to inform Buzz which this was, as only she could see the light. The dolphins demonstrated essentially perfect success over thousands of trials at this task. It seemed that Doris was transferring novel information to Bazz by means of acoustic signals. Indeed, the transmission of this type of information would imply that dolphins have the ability to create acoustic symbols to represent something as arbitrary as a flashing or steadily lit light.

Appreciation of Bastian’s paradigm came from Markov and co-authors (Markov and Ostrovskaya, 1990; Zanin, Markov and Sidorova, 1990) who used adult female dolphins in an experiments modelled closely but not exactly after Bastian's. They replaced light signals with balls of different size and incorporated role reversals successfully into the same experiment. In Bastian's experiment, the role reversal segment was not completed, but enough data was accumulated, before they shut down the experiment, to show that there was not an easy exchange of roles and thus probably no " words" that were easily used by both dolphins. In experiments of Markov and co-authors the dolphins did cooperate in joint work with a high percentage (83%) and in 90% of all the joint work pressed the same paddle. After the high percentage of Session A, though, the response percentages began to decline until the dolphins refused to " talk" to each other in the session C. This breakdown in communication was due to one of the dolphin's (Jenny) dominance over the other (Kora) and her actions stopped the role reversal (where Kora would “tell” Jenny which ball to press). Nevertheless, the obtained results showed that dolphins can co-ordinate the behaviour of each other. The relative ease with which the dolphins reversed roles enabled the researchers to suggest that the dolphins were using an existing equivalent of a " word" from their own communication code.

Despite these supportive experiments with a reversal stage completed, the question about dolphins’ natural “language” still remains unclear. One can agree with Morton’s (1971) sentence that if dolphins do have a language they are going to extraordinary lengths to conceal the fact from us.

Firmly resolved to discover what animals really conceal from us, we should first define the term “language” and identify a field of comparative analysis of human and animal languages.

29. COMMUNICATION, SPEECH AND LANGUAGE: WHAT FALLS TO THE SHARE OF NON HUMANS?

In this chapter we will consider some notions and definitions which help us to specify limits for reasoning about animals’ language behaviour as one the most intelligent forms of relationship.

Communication. The term " communication" enjoys a wide variety of meanings, which is of no wander because communication is a diverse and widespread phenomenon that serves as a substance of any social behaviour. It is difficult to imagine both social behaviour of any level lacking communicative means and transferring of information that does nothing with social relations.

Some definitions include codes of specific signals which animals use to communicate with each other. For instance, Vauclair, (1996) defines communication as following: " Communication consists of exchanges of information between a sender and a receiver using a code of specific signals that usually serve to meet common challenges (reproduction, feeding, protection) and, in group-living species, to promote cohesiveness of the group." < /small>

In group living animals communication serves many important functions such as (1) to advertise individual identity, presence, and behavioural predispositions; (2) to establish social hierarchies; (3) to synchronise the physiological states of a group during breeding seasons; (4) to monitor the environment collectively for dangers and opportunities; (5) to synchronise organised activities (migration, foraging).

Many species have well-developed forms of inter-individual exchange of information based on specific social signals related to courtship, defence of territories, ranking, care about offspring and other forms of social inter-relations. There is a huge body of examples concerning the use of expressive signals within a wide variety of species. For instance, lizards have gestures such as submissive and aggressive circular forelimb waving, tail lashing, head bobbing, back arching, and so on.

To identify complex forms of animal communication that may be attributed to language behaviour, the important feature is purposiveness. It is often difficult to decide whether animals intend to share information with conspecifics or they use inadvertent signalling. We can consider language behaviour the most complex form of communication which take place when animals advisedly transfer the information to each other.

Speech. It is intuitively clear that communication it too broad concept and speech is too narrow for considering as the form of information transferring in animals. Speech is a specially designed system for language and serves as a very effective mean of communication. According to Kimura (1989) there are two major forms of language in humans: speech and the manual sign language of the deaf. Besides, many ways of signalling languages exist in human culture such as Morse code, dram-dram signalling, flag signalling, whistle, semaphore, smoke buoy-signalling and so on. Evidently, human beings universally and preferentially employ speech to communicate.

In the great majority of scientific literature speech is defined as a form of communication specific to humans. Phonetic assurance of speech in humans is supported by the coordinated actions of several articulators (jaw, lips, tongue, velum, and larynx) and by specific brain structures. The task for a child learning to speak is to reproduce, or to imitate, the patterns of articulatory gesture specified by the acoustic structure of the word heard. A capacity to imitate vocalisations is confined to a few species of songbird, certain marine mammals, certain monkeys, and humans. Speech development has both universal and language-specific aspects. Perceptually, speech already has a unique status for the human infant within a few hours or days of birth. Neonates discriminate speech from non-speech, prefer speech to non-speech, and prefer their mother’s voice to a stranger’s.

Perceptual studies of infants, from 1 to 6 months of age, based on the habituation- dishabituation experimental paradigm have shown that infants can discriminate virtually any speech sound contrast on which they are tested, including contrasts not used in the surrounding languages (these experiments have been described in Part II as an expressive example of studying habituation: the amplitude of suckling baby's dummy and several other behavioural cues were used as characteristics of infants’ reactions to novel stimuli). Over the second 6 months of life infants gradually cease to discriminate non-native sound contrast. During a period in which an infant is perhaps first attending to the different contexts in which words are used, it also gradually sifts speech sound contrasts that are functional in the surrounding language from those that are not. (Studdert-Kennedy, 1983; Eimas, 1985; Menyuk et al., 1986). A real world problem facing the human infant is how to segment the continuous acoustic stream of speech into functional units such as phonemes, words, and phrases. It has been shown that infants are equipped with mechanisms that enable them to extract abstract rules that, subsequently, may form the foundation upon which grammars are constructed (for a review see: Hauser, Chomsky and Fitch, 2002).

These findings suggest that our species is highly predisposed for development of speech as the specific form of communication, and they are in harmony with Chomsky’s (1968) point of view that there is an innate apparatus in humans generating a universal grammar.

Studies of Santos et al. (1999) have shown that the same method of studying reactions to speech sounds can be used with human infants and non-human primates. Instead of suckling response used as the behavioural cue in human infants, for monkeys, researchers recorded a head-orienting response in the direction of the concealed speaker. The habituation- dishabituation experimental paradigm provides a tool to explore similarities and differences in perceptual mechanisms.

Results of play-back experiments have shown that such animals as cotton-top tamarins not only attend to isolated syllables but also attend to strings of continuous speech. Tamarins demonstrated the ability to discriminate sentences of Dutch from sentences of Japanese in the face of speaker variability which enables to suggest that they are able to extract acoustic equivalent classes. Specifically, having been exposed to continuous acoustic stream of syllables, tamarins, like human infants, were able to compute relevant statistics (Hauser and Fitch, 2003).

It is important to note that neuroanatomically, there is a direct connection from the cortical face area to the laryngeal motoneurons in humans but not in monkeys. Bilateral destruction of this area distorts human vocal utterances severely, while those of the monkey remain unaffected. These and other findings suggest that the cortical face area participates in voluntary articulation in humans. At the same time, discrimination experiments showed that there is a left hemisphere (right ear) advantage for the recognition of species-specific (but not others) calls in monkeys. So the principal neural prerequisites for decoding speech sounds seem to be already present in the monkey (Jü rgens et al., 1982; Weiss et al., 2002).

As we have seen, speech is just one form of language, the most complex and likely to be confined to humans. Unlike the broad notion of communication, and the specific notion of speech, the concept of language should be more useful for reasoning about animal communication.

Language. The use of language is one of the most complex human skills and it is no wonder that many scientists define language in such a way that only human beings can be said to be capable of it. For instance, Hornby (1972) defines language as “Human and non-instinctive method of communicating ideas, feelings and desires by means of a system of sounds and sound symbols”. Chomsky (1986) considers language prerogative of humans that is aimed to facilitate free expression of thought and clarify one's ideas, as well as to help establish social relations and to communicate information. Behaviourally, language can be defined as a system of self-generated movements, composed of definable units, which can arbitrary represent some object, event, or intention on the part of the mover (Kimura, 1979). Human language is the most sophisticated communicative system which includes thousands of spoken and written languages. As a reflective communication, language can be used to transfer ideas, feelings, to lie and to exchange fantastic imaginations. It is a debated question whether any but human animal share these features. As with most questions about animal cognition, there is a problem of methodology and interpretation. As we have already seen in this book, many species share with humans such cognitive abilities as abstraction, categorisation, and classification. Animals’ ability to lie and sympathise with others will be discussed in the next Part. All these abilities are essential for the development of language but it is still difficult to decide whether they are sufficient. What can we interpret to be equivalent to human language depends on how we identify language and to what extent our concept of language allows comparisons between species. < /small>

It is now agreed that language cannot be described or defined by one single feature; it is a polymorphus concept. Hockett (1960, 1963) identified a range of characteristics that described essential features of language. Hockett’s longest list includes 16 features. Some of them are only applicable to human spoken language and are not relevant to all forms of language in general, including human sign language. However, some of the design characteristics are seen as essential features of any form of language and are useful criteria to assess claims about animal language. Several authors have used Hockett’s set of characteristics to complete a list of criteria that an act of communication must fulfil if it is to be regarded as language (Aitchison, 1983; Anderson, 1985; Pearce, 2000).The following list provides a useful framework for evaluating the linguistic skills of animals.

1. Interchangeability: language is a two-way process that involves both

sending and receiving the same set of signals. This is distinctive from some animal communications such as that of the stickleback fish. The stickleback fish make auditory signals based on gender (basically, the males say " I'm a boy" and the females say " I'm a girl"). However, male fish cannot say " I'm a girl, " although it can perceive it. Thus, stickleback fish signals are not interchangeable. In many social species any member of a group can produce and perceive universal signals, so their communication systems meet this criterion.

2. Specialization: language is not a by-product of some other biological

function; it has a special function for communication only. Human enjoys the most specialised system of communication. Returning to the stickleback, the male reacts to biological aspects of females’ messages such as swollen belly, and the female reacts to the change of the male’s colouration, so the males’ message is more specialised.

3. Discreteness: the basic units of language (such as sounds) can be organised into discrete units and categorized as belonging to distinct categories. There is no gradual, continuous shading from one sound to another in the linguistics system, although there may be a continuum in the real physical world. Thus speakers will perceive a sound as either a [p] or a [b], but not as blend, even if physically it falls somewhere between the two sounds. It is important that discrete units can be broken apart to form new signals.

4. Arbitrariness of units: language is composed of discrete units, and the form of the signal does not depend on a referred thing. Messages consist of arbitrary units but not from icons describing an object. This important property returns us to the epigraph to this part of the book. Taffy, the girl from Kipling’s story, sent her mother a letter consisting of icons, and wrong interpretation of pictures resulted in a great conflict.

5. Displacement: language can be used to refer to things that are not present in space (here) or time (now). These requirements are known as “spatial displacement” and “temporal displacement” respectively. Human language allows speakers to talk about past and future. Speakers can also talk about things that are physically distant (such as other countries and other planets). They can even refer to things and events that do not actually exist. Many researchers consider honey bee’s Dance Language to meet this criterion. Bees are able to refer to objects that are distant in space and even in time. For instance, Lindauer (1960) reported that in the night bees performed dances signalling feeder sites that they had visited the previous day. Apes easily meet this criterion with the use of intermediary artificial language but we know little about the potential of their natural communication.

6. Semanticity. Communication meets this criterion if signals (or words) convey specific meanings. We presume that for an animal a signal has meaning if it can activate a representation of the event to which it relates. Thus vervet monkeys appreciate the meanings of alarm calls they hear (Cheney and Seyfarth, 1980, 1990). It is likely that honey bees may appreciate the meaning of communications they receive. In Chapter 12 results of Dyer’s (1991) experiments were described. Bees left the hive when the returning scout indicated that the food was beside a lake. But they did not leave the hive when they were informed that food was near the middle of the lake. Thus honey bees appear to interpret the meaning of the dance – possibly by identifying the potential location of food on cognitive map – and then decide whether it is worth making the journey (Pearce, 2000).

7. Productivity: language can be used to produce an infinite variety of new messages from a limited vocabulary. Productivity is also called “creativity”. In particular, productivity of language enables the use of analogies, and by the use of analogies children adopt rules of grammar, or syntax. Productivity is a powerful property of language that makes it an open-ended system. The use of intermediary languages enables animals to meet this criterion. It is challengeable to explore whether any of natural animal communications meets this requirement.

8. Traditional (cultural) transmission: human language is not inborn, or at least, completely inborn. Although humans probably have a genetically-based capacity for language, they must learn, or acquire, their native language from other speakers. This is different from many animal communication systems where the animal is born knowing their entire system. At the same time, as it was described in Parts VII and VIII, cultural transmission plays a great role in the development of communication in some species of birds, monkeys, whales and may be some others. We have also seen from Part VII that Cultural Transmission does have limits in some species. For example, with humans and finches, language acquisition does not happen easily or ever fully after a certain age. Finches have been tested by removing them from other finches until they are grown; the test finches do make calls (so it is somewhat innate) but never to full capacity (so it is somewhat transmitted culturally). With humans, several cases with children grown in isolation due to unfortunate concourse of circumstances show that although some language can be learned, not nearly as much can be once a certain age is passed. It is known that language acquisition occurs best at an early age, primarily before puberty, and the capacity for it decreases after that age. Some communication systems are fully innate, however, with no cultural transmission. For example, cow birds (who lay their eggs in other birds’ nests), when born, will give cow bird calls, not the calls of whatever bird they are raised by.

30. DIRECT DIALOG WITH ANIMALS: LANGUAGE-TRAINING EXPERIMENTS

The use of intermediary languages for studying “linguistic” and intellectual potential of animals can be considered a revolutionary approach which has changed the general concept of animal intelligence. Only thirty years ago it could be difficult to imagine that animals can learn to associate arbitrary signs with meaning, to generate new symbols with new meanings, and to use these signs to communicate simple statements, requests, and questions; to refer to objects and events displaced in time and space; to classify novel objects into appropriate semantic categories, and to transmit their knowledge to peers and offspring.

There are many excellent books and reviews written by researchers who carried out projects on teaching sign languages to apes (Savage-Rumbaugh, 1986; Savage-Rumbaugh and Lewin, 1994. Gardner et al., 1989; Fouts, 1997; Patterson and Linden, 1981); dolphins (Herman et al., 1984) and an African grey parrot (Pepperberg, 1999). Here I briefly describe how this method has influenced development of studying of animal language behaviour and intelligence.

Attempts to enter into negotiations with animals began from dialogs between humans and apes. As it was mentioned in Part I, Robert Yerkes made a suggestion that was not followed up on for forty years: perhaps apes could learn gesture sign language. Yerkes failed to teach chimpanzees to speak and he concluded that apes cannot speak because they lack the tendency to reinstate auditory stimuli - in other words to imitate sounds. Indeed, several attempts to teach apes to repeat human words were almost unsuccessful. Furness (1916; see review, Ristau and Robbins, 1982) taught a female orangutan to produce vocally, “papa”, “cup” and “th” over 11 months of instruction. Hayes and Hayes (1952) achieved a little more with their chimpanzee Viki who was taught to pronounce a few words, mama, papa, cup and up, by the use of positive reinforcements. At first, Viki was rewarded with food for making any vocalisation, and this way she was taught how to produce sound on demand. It took 5 months before Viki was able to produce vocalisations that were recognisable in any way to humans, the first one being a hoarse " ah" sound. This was later developed to " mama." However, the Hayes recorded that Viki had considerable difficulty pronouncing words, would often become confused and pronounce or use the words incorrectly. Also she held her lip while " talking". It has been revealed later that upper respiratory systems with a vocal apparatus enabling human speech differ in apes and humans and, what is a not less important, neural mechanism developed for encoding and decoding phonemic communication are specific and much different from those in apes (Lieberman, 1984).

Once it was apparent that apes could not learn to speak, other variants of direct dialog have been elaborated. It is worth to note that such dialogs were based on apes’ propensity for categorization, one of the most important properties of carriers of the developed language.

30.1. “Token language”

One of variants of a mute dialog with animals was the “token language” elaborated by Wolf (1936). In general, tokens have been described as secondary rewards that can be exchanged for any primary reward (such as food, drink, toys etc.). They have mainly been used in studies of operant behaviour. Nevertheless, in his classic experiments Wolf (1936) demonstrated chimpanzees as possibly being able to understand symbolic meaning of tokens and use them to reach specific goals. In his experiments tokens were differently awarded and thus differed by their purchasing power. For instance, inserting a blue token into a slot an animal received two bananas, while a white token gave only one. A black token could be exchange for any kind of food whereas a yellow one for drink only. In other series of experiments blue tokens being inserted into the hole gave apes a possibility to return to their living cage from enclosure, and the yellow one gave a possibility to play with a human tutor. When a rat appeared in an enclosure, chimpanzees who hated rats abandoned all their entertainments and used blue tokens quickly in order to get to the home cage. Cowles (1937) revealed that chimpanzees can “work” for tokens and they can even accumulate several tokens before exchanging them. Later “token dialog” was successfully used in several experiments (Kelleher, 1958; Schastnyi and Firsov, 1961; Firsov, 1972).

Recently Sousa and Matsuzawa (2001) have followed the same paradigm using a computer setting. Three chimpanzees were provided with a vending machine for inserting tokens (Japanese coins) and three touch-sensitive panels. The panels were connected to computers that controlled a discrimination task using tokens as a reward. Discrimination tasks were intellectually costly and included discrimination within several sets each of 10 visual stimuli. Apes demonstrated a high level of accuracy, suggesting that the tokens were almost equivalent to direct food rewards. The apes also saved tokens before exchanging them for food. The chimpanzees performed tasks to collect tokens with the objective of exchanging them for food and thus planning their proximate behaviour. The authors suggest that chimpanzees use tokens as specific tools. In sum, these results support Wolf’s suggestion that apes can understand symbolic meanings of tokens.