Sue Savage-Rumbaugh
and friend
This is a study by Sue Savage-Rumbaugh et al. where the ability of non-human primates to acquire language is investigated.
Speaking personally, I do not understand the need to find out if a species of chimpanzee can communicate with humans or not. What are you going to talk about? The price of bananas? O.k. rant over, let's get back to the study.
It is known that chimpanzees in the wild do communicate with each other, by both gesture and "voice". There have been many previous attempts to study language acquisition by chimpanzees. Some studies have tried to teach American Sign Language, e.g. Gardner & Gardner with Washoe. Other studies used geometric shapes or tokens.
It has been shown that when a chimpanzee is taught the communicative skill of requesting, labelling, and comprehending, the chimpanzee can then use the skills they have acquired to indicate behaviours that they are going to engage in, in advance of actually carrying out those behaviours.
Nevertheless, it is still the case that chimpanzees do not learn language skills in the same way that a human child learns language. Chimpanzees require special training sessions where they are reinforced for using the correct term for an object.
One ape, Loulis, (an infant adopted by Washoe) has been observed to produce a number of hand gestures without human training. However, although he uses signs, it has not been tested if he uses these signs correctly. It has been observed that chimpanzees gesture in the wild (one of the reasons for teaching chimps a gestural language) and in captivity too, and it may be that Loulis has just copied Washoe without any "understanding" of what the gestures mean.
Contrasting the way that human children acquire language to the way that chimpanzees acquire language is interesting. It is not really clear how children come to associate an object with a sound. Chimpanzees require a lot of training. Children learn to use sounds and are able to refer to objects that are not present, e.g. a mother might ask a child "Where is your teddy". In response to this the child might start to look around the room, if they cannot find it they might have to go to another room in the house to retrieve their teddy. In this instance the word "teddy" has moved from a mere association to a mental representation of what a teddy is, and if not present, to find it. If chimpanzees can use language to refer to things that are not present, then it would then be possible to say that chimps can use language. In order to get to this stage it has been found that chimpanzees require special training.
It is true that chimpanzees do not have the vocal apparatus to speak. However, one question that might be important is: "How much are they able to understand?" Some workers in this area have said that their participants in the study were able to understand more than they were able to produce, but to empirically test this has been found to be quite difficult. A chimpanzee called "Ally" was taught to produce behaviours in response to 10 verbal cues, e.g. "Ally bring me a pillow". Sadly, despite being in a language using environment from birth, the number of correct response remained low and tests were not always done without contextual cues. A lowland gorilla called "Koko" was claimed to have much better comprehension skills, but he showed an unusual pattern of results. He was given the test called "Assessment of Children's Language Comprehension". This is a test where the questions contain four line drawings and a sentence, and the sentence has to be matched to one of the line drawings. When a child does this test, they frequently stop when the questions become too difficult. Koko would attempt every question no matter of its difficulty, and scoring 50%. However it was difficult to interpret Koko's results because she did not differentiate the same way that a child would.
To get your own copy of the study, here is the reference:
"Spontaneous Symbol Acquisition and Communicative Use by Pygmy Chimpanzees (Pan paniscus)", Sue Savage-Rumbaugh; Kelly McDonald; Rose A. Sevcik; William D. Hopkins and Elizabeth Rubert. Journal of Experimental Psychology: General, 1986, Vol. 115, No. 3, pages 211-235. (Quite a long study)
​That is enough for now, let's get on with the study:
Method:
The method used was a Case Study. This is an intensive study of a small number of participants, usually over a long period of time. This study lasted for 10 years.
Sample used:
The main participants were a males bonobo chimpanzee called Kanzi, who was born in captivity at the Yerkes Regional Primate Research Centre on the 28th October 1980. He was first engaged in the study at the age of 6 months. His mother, Matata, who was born in the wild, and has participated in other studies of language acquisition. His younger sister, Mulika, born 22nd December 1983, she was temporarily separated from her mother for four months due to having an eye infection.
The Bonobo's performance was compared to 2, male, pan troglodytes (chimpanzees) called Austin and Sherman. these started their training at ages 1.5 years and 2.5 years respectively. They were removed from their mothers before the start of the study.
One basic difference between pan paniscus (bonobo or pygmy chimpanzee) and pan troglodytes is that the bonobo is smaller than the chimpanzee, and the bonobo tend to be lees aggressive and more sociable than the other type of chimpanzee.
Procedure:
The main method of communication with all of the participants was a visual symbol system. This consisted of a panel of buttons with geometric symbols that light up when pressed. This action is treated as having the same meaning as speaking a word. once it was clear that Kanzi comprehended many words, his system included a speech synthesiser that uttered the word once the corresponding button was pressed. Several presses would mean the same word was repeated the same number of times the button was pressed. This system was designed to compensate for the anatomical and neurological limitations that prevent chimpanzees from using spoken words.
EARLY REARING AND EXPOSURE TO LEXIGRAMS
Kanzi was first exposed to the use of graphic symbols, and human speech when he was 6 months old, while he was still being cared for by his mother. Kanzi remained with his mother until he was 2.5 years old, during which time attempts were made to teach Matata eight symbols. Initially, she did not make the required progress and after a while she was trained in the use of the symbols in a more formal manner, that had proved effective with other chimpanzees, i.e. Sherman and Austin. No attempts to teach Kanzi the use of the symbols while his mother was in training because it was not feasible to separate them without causing distress. Kanzi seemed disinterested in requesting food, preferring instead to steal his mother's food.
At age 1.5 years Kanzi started to show interest in the symbols. He would try to grab them when they appeared on the video screen or the projectors. While Matata was using the keyboard, Kanzi would press keys and then run to the vending machine, demonstrating that he had learned that pressing the buttons caused food to be dispensed. Kanzi would often interrupt his mother's training sessions by sitting on her, the keyboard, he would jump on her hand as she was about to reach for a piece of food, grab the food, and run off. Matata tolerated his behaviour, though sometimes she would shoo him away.
Kanzi was separated from his mother at age 2.5 so that she could join the breeding colony elsewhere in the University. Matata became pregnant and returned to the Language centre 4 months later. Kanzi was allowed to spend as much time as he wanted with his mother but spent most of his time with his human companions. Mulika was born 9 months later. Kanzi was interested in Mulika from the time of her birth. When she was 1 month old Matata would allow Kanzi to carry her around. At 4 months of age Mulika developed an eye infection and was separated from Matata for treatment. On her return, like Kanzi she seemed to prefer her human companions, but unlike Kanzi she did not see​ Matata during symbol training, but could observe Kanzi.​
​​
REARING AND EXPOSURE TO LEXIGRAMS FOLLOWING SEPARATION FROM MATATA
In many respects, Kanzi and Mulika's rearing environment, following separation from Matata, was very similar. They were all with people who used the visual symbol system around them throughout the day for a variety of communication needs.
Kanzi and Mulika began (a) using symbols without specific training, (b) identified the correct symbol regardless of location, (c) did not tend to confuse symbols for similar items (e.g. apple, orange and banana) and (d) comprehended spoken words, no formal training was ever attempted with them. It is important to note the food reward was not dependent on symbol acquisition.
Staff were instrumental in symbol acquisition as they would model symbol use for all of their communication with each other and with Kanzi and Mulika. During daily activities people commented on the activities both vocally and using the symbols. Sometimes Kanzi and Mulika would watch what the staff were doing, and sometimes they would not.
NATURALISTIC OUTDOOR ENVIRONMENT
During the warmer months of the year, food was dispersed at 17 locations within a 55 acre forest that surrounds the laboratory. Food was not available in the laboratory during this time meaning that the chimps would have to travel around to find food, much as they would have to do in the wild.
While Kanzi only knew a few symbols, he was provided with photographs of the food so that he could indicate what food he would like to eat. These photographs were also at the location where that food was available. Not all food was available at every location outside, e.g. the "lookout" was where the bananas and juice were, and the tree house was where the peaches could be found. Kanzi would indicate what food he wanted and would travel to the location he knew where it could be found. Within 4 months, Kanzi had learned all of the locations of the food, he could not only select a photograph when he wanted to go to a specific location, but could also guide others to the appropriate location. At the start, he would carry the photograph the entire distance, frequently pointing at it and vocalising. Later, he would use the keyboard to announce where he wanted to go, and gradually the use of the photographs ceased. It is important to remind readers that correct performance with any of the equipment would not mean that Kanzi would be denied food.
Typically, Kanzi would initiate travel to a particular location and obtains food on arrival. If he does not indicate where to go, then travel is initiated by staff. If on arrival Kanzi does not want the food, he is asked if he wanted to put it in the backpack, a positive gesture, such as pushing the food towards the backpack, if he ignored the request, food would be left where it was. After a half days travel there would be some items in the backpack. If Kanzi indicated that he wanted an item that was in the backpack he would touch the backpack, but otherwise he is indicating he wants to go to that location.
Four months after the food had initially been placed outside, when Kanzi was 3 year old, he was tested with a member of staff who had never been in the wooded area at all and did not know where each location was, what food was available at each location, where the paths were, or how to get back to the laboratory. She was completely reliant on Kanzi's knowledge of the area that he had been using every day before that. The test was conducted across two afternoons. During that time she travelled with Kanzi, at Kanzi's initiative. Kanzi selected each location from a photograph or by pointing at a lexigram. Each location had 6 to 10 alternatives, either photographs or lexigrams representing both food and locations. Kanzi would guide the staff member to the location or where the food may be found.
INDOOR ENVIRONMENT
When inside, during the day, Kanzi and Mulika would be asked to join in with some of the tasks, such as changing bed linen, doing the laundry, and preparing food. They would frequently try to engage others in games such as: hide and seek, Marco Polo, chase, grab, tickle or play bite, they would often spontaneously help with simple tasks like mopping up spills, washing dishes, spaying the hose, scrubbing the floor and moving objects from one location to another. They would move around the centre and play with clay, toy dolls, and bubbles. They would like to watch videotapes of people they know doing interesting things In the evening, Kanzi liked to watch TV, a number of videos with lexigram overlays had been prepared and were of interest to the chimpanzees.
DATA RECORDING
From the point when Kanzi started using the lexigram, at age 2.5 years, a complete record was kept of his utterances for 17 months. All lexigrams used by Kanzi and Mulika were automatically recorded by the computer when indoors. When outside the usage of lexigrams were recorded by hand and fed into the computer in the evening along with contextual notes. All data summaries are based on all of Kanzi's lexigram use from age 30 months of age (when he was separated from his mother) to 47 months of age. Mulika's use of lexigrams is recorded from age 11 month to 21 months.
Each utterance (Use of the lexigram?) is classified as correct or incorrect, spontaneous, initiated or structured. Spontaneous utterances are those initiated by Kanzi or Mulika with no prior prompting or any other behaviour designed to elicit a response. Initiated utterances are those that include any part of a companion's keyboard utterances. Structured utterances are those that are initiated by questions, requests, or object showing behaviour on the part of a companion. These are usually preceded by a question such as "What is this?" In these circumstances the response is known before it is actually made. Such questions are usually used to find out if Kanzi or Mulika know how to answer the question.
VOCABULARY ACQUISITION CRITERION
For the correct use of a symbol be counted as correct the method for assessing this differs from some previous research. In previous research using the correct symbol or gesture appropriately on a specific number of occasions counted as having learned the gesture or symbol. However, it has been said that it is possible for comprehension appropriate responses to be shown where comprehension is not also present. The measure used in this study was that Kanzi or Mulika had to use the correct symbol on 9 out of 10 situations, these usages of symbols had to be spontaneous. For example, if Kanzi requested to go to the treehouse, he would be told "Yes, we can go to the treehouse". If he led the staff to the treehouse, this would be an appropriate use of the symbol for "treehouse", leading the staff anywhere else would be incorrect. If Kanzi asked for a banana, he would be presented with a variety of food, including a banana. Selecting the banana would count as a correct use, anything else would be incorrect. On some occasions, Kanzi would be asked to lead the staff member to the location where a banana might be found. Leading the staff member the correct way to the correct location would result in a correct use of the term "banana".
In the situation where Kanzi requested a banana and pointed at the backpack, if there was only a banana in the back pack, this would not be counted as a correct use of the symbol for banana, but the situation would be treated no differently, because Kanzi had used the symbol appropriately.
TESTS OF PRODUCTIVE AND RECEPTIVE CAPACITIES
In addition to recording the use of lexigrams, that were spontaneous, Kanzi and Mulika were tested informally, in everyday situations. For example, if Kanzi was playing with some keys, he would be asked what they were. Conversely, if he had scattered several objects around the floor, he would be asked to pick them up, sometimes he would be told to pick them up in a certain order, e.g. ball first, then the blanket. If Kanzi seemed unsure, he was shown what to do. Such requests were not drills and they were not rewarded with food. Additionally ,such interactions were only done using symbols that Kanzi and Mulika already knew.
When Kanzi was between 46 and 47 months old he was tested on all the items in his vocabulary. The same tests were given to Mulika when she was 18 and 21 months of age. These test tried to control for things such as contextual setting, inadvertent glances and so forth. The order of presentation and the location of stimuli were carefully controlled in order to preclude any knowledge on the part of the experimenter that might inadvertently bias the response.
(Prior to the test results reported here, Kanzi was also tested at 32 months of age on receptive skills, at 35 months of age on naming skills and at 45 months of age on both naming and receptive skills.) Test trials were done in the afternoons with between 20 and 40 trials in one session, depending on the amount of time spent playing between trials. No test item was ever repeated during a given session and no two trials were the same. The test trials and alternatives were randomly determined on each trial with the requirements that neither the alternatives nor the test items were repeated on consecutive trials.
For example, Kanzi might be presented with the lexigrams: green bean, apple and tomato, and asked, in spoken English to, "Show me the tomato lexigram." The tomato lexigram was not repeated as the target item during that session, unless it was tested in a different manner, e.g. Kanzi might be shown the tomato lexigram and then presented with photographs of a banana, some M&Ms, and a tomato. However, two trials with the same target item would never follow one another under an circumstances. Additionally, the same alternatives were never paired more than once with the same target item.
Photograph to lexigram​​
During this test, Kanzi was shown a photograph and was asked to select, from a set of three alternatives, the proper lexigram for that photograph. The alternatives were set behind a folding blind, and were not visible to the experimenter. The placement of the alternatives in the blind by a second experimenter to eliminate the possibility that Kanzi might get a hint by the experimenter looking one way or another. Kanzi would make his selection by pointing to a particular lexigram. The response was observed by the experimenter who then dropped the blind to determine whether or not the choice was correct. The second experimenter did not observe what happened in the test, after he had set up the lexigrams. Also the English name of the photograph was not mentioned.
Spoken English to photograph
In this test, Kanzi listened to the spoken English word and selected the appropriate photograph from a set of three alternatives. As before the alternatives were set up by a second experimenter so that the experimenter could not see the alternatives and could not give a clue to the correct response. The English word was usually resented in a sentence and then repeated, e.g. "Kanzi can you show me the banana. Banana" After Kanzi had made his choice, the experimenter lowered the blind to determine if the response was correct or incorrect.
Synthesised speech to lexigram
This followed the same procedures as above, but the word was spoken by the speech synthesiser that was connected to the system that Kanzi used in the laboratory to communicate. The word was produced twice, and Kanzi was presented with four alternatives, and Kanzi pointed to the correct lexigram. This test was devised so that Kanzi would not get any clue from the intonation of a human voice. Some words produced by the synthesiser are difficult to understand, if you are not used to it. This was not a problem for Kanzi, and this test was only performed on Kanzi.
Results
UNTUTORED GESTURAL USAGE: SOME OBSERVATIONS​
Between the age of 6 and 16 months of age, both Kanzi and Mulika spontaneously started to use gestures to communicate their needs. For example the used an outstretched arm and hand to point toward an area that they wished to be carried; they made twisting motions towards containers with twist top lids that he wanted help with; they made hitting movements towards nuts when they wanted other to crack for them; they would request that another would pick something up for them by gesturing to the person and then to the item. Vocalisations would often accompany gestures to orient the listener's attention and to convey their mood.
FIRST APPEARANCE OF LEXIGRAM USAGE FOR COMMUNICATIVE PURPOSES
Immediately following his mother's removal for breeding purposes, Kanzi would approach the keyboard much more often to light symbols. He would often stop and search for a particular symbol on the keyboard, instead of flailing his hands toward the board. Although it was assumed that Kanzi had learned (from watching his mother) that lighting the symbols was a way of obtaining food, it was not assumed that he had formed any referential relation between particular symbols and particular foods, or that using particular symbols could be used to ask for a particular food.
Even though it was not thought that Kanzi was purposefully selecting a particular lexigram to ask for a particular food or activity, people responded to such keyboard utterances as though they were intentional. They were asked to do this as language is viewed as a process of "guided reinvention" in which it is the interpretation of a child's communication as intentional long before they actually are, that facilitates the development of language for intentional communicative purposes. It was made clear to all staff that the purpose of these "rich interpretations" is to promote communicative development but not to evaluate it.
Examples of Kanzi's behaviour after separation from his mother are given blow:
-
Kanzi requested "apple", by touching the symbol on the keyboard, and one was retrieved from the refrigerator for him. He took a few bites from it and dropped it onto the floor and began to do some playful spins and other gymnastic manoeuvres, which he often liked to do, while grinning broadly. Kanzi suddenly stopped his spinning and touched "apple" again, picked up the apple from the floor and took a bite from it. He then touched "chase" and looked toward the experimenter, who responded by chasing Kanzi around the room.
-
After eating an apple and drinking juice, both of which he had requested from the keyboard, Kanzi again touched "juice", then picked up the juice and poured it out, and began stomping the apple and juice together while displaying a "play" face.
-
Kanzi touched "ball" and then began looking all over the room. Upon finally finding the ball, which had rolled under his keyboard, he displayed a "play" face, grabbed the ball, and started slapping it vigorously.
Mulika started using symbols at 12 months of age, much earlier than Kanzi had. However she had her own particular was of using the keyboard, in that she used one symbol "milk" for many different things. These included: requests to be picked up; requests for attention; requests to be taken to different places; requests for food; and requests for milk. Around this time, Mulika would imitate people when they used other symbols to tell her the name of something, but whenever she used the keyboard spontaneously, she would always select milk. To make her intentions clear she would often behave in a way to indicate what she wanted. For example, she would select "milk", climb onto someone's shoulders and point in the direction of travel. If there was a selection of food items available, she would touch "milk" on the keyboard, and then point to the food she wanted, e.g. an apple. Simply pointing to the apple and vocalising would have been enough for Mulika to be given an apple, this is what had happened before use of the keyboard. It did not seem that simply touching any of the symbols was a way to make things happen. Like Kanzi, Mulika did not experience any difficulty identifying lexigrams in new positions, on other keyboards, or off keyboard. From the time she began using "milk" she could select it from an array of lexigrams.
This use of the "milk" lexigram lasted for about 2 months. After that time Mulika would use a range of different lexigrams appropriately. The second lexigram that she would use was "surprise". This happened at 14 months of age, she had requested "milk" and gestured toward the refrigerator. After being given the milk, she drank some and refused to drink any more, instead she went back to the keyboard and started searching very carefully. She then touched the "surprise" lexigram and looked to see if anyone was watching. An unusual unnamed food was found in the refrigerator and given to Mulika. She ate this with relish. In the next few months "surprise" "Matata", "peanut", "hot-dog", "coke", "mushroom", "melon", "cherry", "banana", "jelly", "go", and "blueberry" began to appear, all were used for specific and appropriate things. "Milk" became confined to the instances when she really did want milk, except when she appeared to be distracted and uncertain and she would revert to using "milk" and gesturing her request.
Both Mulika and Kanzi would initially use a symbol within a usage routine. For example, the symbol for "strawberry" was introduced to Kanzi while he was eating mushrooms on mushroom trial. He was told that someone knew of a place where there were strawberries and everyone expressed excitement. Kanzi imitated the use of the symbol for strawberries and travelled with his companions. When they arrived Kanzi readily consumed the strawberries. For a number of days later, he would only use the "strawberry" symbol only in the situation where he first learned it, when he was eating mushrooms on mushroom trail. It would appear that here would remind him of strawberries. He would then go to the keyboard and use the symbol for "strawberries" and then gesture that he wanted to go and get them. Upon arrival he did not use the symbol to request or name the strawberries, even if he was asked to do so. He also showed that he understood phrases such as "hide strawberries" as he would grab them before they could be hidden. After a while, the use of the symbol for strawberry, moved out of the situation where it initially occurred and Kanzi performed tests where this symbol was selected in response to the spoken word.
This way of learning new symbols, in a situation first and then becomes used more generally, is probably how apes and children acquire new words. Often, a symbol's use may expand to an increasing number of contexts, e.g. if Kanzi touched the symbol for strawberry that could mean that he wanted to go to where they can be found, when he is asking for one to eat, and when shown a photograph of strawberries. As we have seen, combinations of utterances are understood by Kanzi e.g. "hide strawberry".
PROGRESS DURING THE 17 MONTHS AFTER THE INITIAL SEPARATION FROM MATATA
Kanzi made rapid progress over the 17 months following his separation from Matata. Words acquired during this time are given below:
​
CUMULATIVE LIST OF KANZI'S SYMBOL ACQUISITION
​
Symbols Age Acquired (months)
​
Orange
Peanut
Banana
Apple
Bedroom
Chase
Austin 30
Sweet Potato
Raisin
Ball
Cherry
Peaches
Coke
Bite 31
​
Melon
Jelly
Tomato
Orange drink 32
​
Trailer
Milk 33
​
Key
Tickle 34
​
Coffee
Juice
Bread 35
​
Groom 37
​
Egg 40
​
Hamburger
Water
M&M
Surprise 41
​
Clover
Matata
TV
Orange Juice 42
​
Mulika
Carrot 43
​
Grab
Treehouse
Blanket
Blackberry 44
​
Mushroom Trail 45
​
Refrigerator
Hot-dog 46
Although Mulika started using symbols much earlier than Kanzi, her initial progress was much slower than Kanzi, suggesting again that Kanzi had learned many things that were not evident prior to the departure of his mother. Mulika's progress may have been underestimated as she was not tested as much as Kanzi was. However, testing of Mulika had started as an earlier age than with Kanzi.
A table of all the words that Mulika had learned during the ages of 12 to 22 month is shown below:
CUMULATIVE LIST OF MULIKA'S SYMBOL ACQUISITION
​
Symbols Age Acquired (months)
​
Milk 17
​
Key
T-room
Surprise 20
​
Juice
Water 22
This is a picture that I "stole" from Wikipedia. It shows Kanzi, interacting with Sue Savage-Rumbaugh, who can be seen on the right edge of the photograph.
COMBINATIONS
Kanzi's multi-symbol utterances, or combinations, appeared very early within the first month of spontaneous keyboard use. These were far less frequent than single symbol utterances throughout the 17 month period covered by this report. Across the 17 month period, Kanzi produced 2,540 non-imitated combinations and 265 prompted or partially imitated combinations. All but 10 of the combinations were judged to be appropriate to the context and interpretable by his human companions, Of the total number of the combinations 764 combinations were used only once. Combinations only accounted for 6% of the total communications. Nevertheless, when Kanzi used combinations, all of the symbols that he use typically added new elements of information to the situation. For example he might utter "ice water go" (with "go" conveyed by gesture) to ask someone to get ice water for him.
Many of Kanzi's two word utterances seem simply to be merged double item requests, that is, requests for two distinct items that are uttered together. For example, on arriving at the food site where both hot-dogs and coke were located, Kanzi often wanted to consume both foods and asked for them by indicating "hot-dog coke". Ono other occasions he would use gesture to indicate the he wanted to have them mixed together.
The most interesting aspect of Kanzi's three word utterances was that he used many of them to specify individuals other than himself as the agent or beneficiary of some actions. Of Kanzi's three word combinations 36% were used to specify a beneficiary other than himself. Thus the most frequent use of three word utterances resulted from Kanzi's attempts to initiate games such as chase, grab or tickle between various persons by indicating "grab" or "chase grab" at the keyboard and then taking one persons hand and pushing it toward the second person. In this way, he designated who was to be the agent of chaser and who was to be the recipient or chasee.
The people who interacted with Kanzi did not paly game of chase grab and bite with each other until Kanzi began to make his interest in observing these activities evident. It is interesting to note that until there is reason to specify an agent and beneficiary other than himself, there is no need to make agent and beneficiary communicatively explicit. As long as the chimpanzee is the recipient of all food, tickling, trips outside and so forth, there is no reason to specify himself as recipient nor the addressee as agent.
KANZI'S MOST FREQUENT COMBINATIONS
​
2 ITEMS
​
Chase person (g)
Person (g) chase (g)
Chase (g) person (g)
Person (g) pat (g)
Chase bite
Chase Kanzi
Peron (g) come (g)
Tickle ball
Bite person (g)
Come (g) chase (g)
Ball Tickle
Chase Sue
Kanzi chase
surprise money
Bite chase
Pat (g) person (g)
Kanzi grab
Grab person (g)
Chase bite
Pat (g) this (g)
Chase come (g)
Person (g) go (g)
Ball pat (g)
Person (g) bite
Chase tickle
​
3 ITEMS
Chase person 1 (g) person 2 (g)
Person 1 (g) pat (g) person 2 (g)
Person 1 (g) person 2 (g) pat (g)
Person 1 (g) chase person 2 (g)
Person 1 (g) grab person 2 (g)
Person 1 (g) chase (g) person 2 (g)
Person 1 (g) person 2 (g) chase
Kanzi chase person (g)
Chase bite person (g)
Person (g) chase Kanzi
Person 1 (g) grab (g) person 2(g)
Chase grab person (g)
Person 1 (g) person 2 (g) bite
Chase Kanzi person (g)
Person 1 (g) tickle person 2 (g)
Person (g) Kanzi chase
Kanzi person (g) chase
Chase five Kanzi
Chase person (g) Kanzi
Pat (g) person 1 (g) person 2 (g)
Bite chase person (g)
Sue bite person (g)​​
THE ISSUE OF IMITATION
It is informative to compare Kanzi's and Mulika's rates of imitated utterances with those of normal children. This can be seen below:​
IMITATED VERSUS SPONTANEOUS UTTERANCES
​
SUBJECT PROPOTION OF PROPORTION OF
IMITATED UTTERENCES SPONTANEOUS UTTERANCES
Kanzi 0.11 0.80
Mulika 0.20 0.67
Stage 1 child 0.18 0.82
Like human children, Kanzi and Mulika imitate most often when they are learning new words. It seems reasonable to assume that language learners that imitation is a strategy used by language learners when they are not sure what to say in a given situation. Moreover, if the language model realizes this, the model will typically show them what to say, thereby setting the stage for more imitation to occur.
Clearly a large proportion of Kanzi's and Mulika's utterances were truly spontaneous and were not elicited by the teacher creating situations that required an utterance. This means that, most of the time, they are adding new information to the situation as opposed to simply saying what is expected of them in order to achieve a "way out".
FORMAL TESTS
Formal tests required a very different sort of symbol orientation than Kanzi and Mulika normally encountered in their day-to-day activities because such tests are, by their very nature, noncommunicative. Thus, instead of touching "juice" because they wanted to travel to a location in a wood where juice was found, formal tests required Kanzi and Mulika to select juice because someone held up a photograph and asked them what it was. Touching the symbol for juice in this context did not result in anything different or significant happening, as was the case when juice was used in a communicative context.
In spite of these differences, Kanzi and Mulika did well on formal tests from their first administration. They seemed to understand that the experimenter were not communicating about something that was going to happen as they touched a symbol, but rather were posing a specific question. Kanzi's and Mulika's abilities to select lexigrams in response to the spoken English word, to select photographs in response to the spoken English word, and to select photographs when shown lexigrams, are the sorts of things they were asked to do in formal tests. A portion of the results are shown below.
VOCABULARY TEST: KANZI
ITEM
MATCH
SYMBOL TO
ENGLISH
MATCH PHOTO TO ENGLISH
MATCH PHOTO TO SYMBOL
MATCH SYMBOL TO SPEECH SYNTH.
A-Frame X C C C
Apple C C C C
Ball C C C C
Banana C​ C C C
Blackberry C C C C
Blanket C C C C
Bread C C C C
Campfire C C C C
Carrot C C C C
Chase C NP NP C
Cherries C X C C
etc.
​
Kanzi was tested on a total of 67 items, each item in four different ways, making a total of 268. Of these, NP=not presented due to not having a suitable photograph (14 or 5.2%), X=incorrect response on 2 or 3 of 3 test trials (23 or 8.5%), the remainder were C=correct (231 or 86.3 %).
I'd like to put a picture of Mulika in this space, but I have searched the internet and not been able to find one. If anyone has one of these they would like to share, you could make an old man very happy.
VOCABULARY TEST: MULIKA
ITEM
MATCH
SYMBOL TO
ENGLISH
MATCH PHOTO TO SYMBOL
MATCH PHOTO TO SYMBOL
Apple C C C
Ball C C C
Blackberry C C C
Blueberry C C C
Bubbles C C C
Bunny C C C
Cereal C C C
Cherry C C C
Clover C C C
Can opener C C C
Coffee C C C
etc.
​
Mulika was tested on a total of 42 items, each item in three different ways, making a total of 126. Of these, NP=not presented due to not having a suitable photograph (1 or 0.8%), X=incorrect response on 2 or 3 of 3 test trials (7 or 5.5%), the remainder were C=correct (118 or 93.7 %).​
Clearly, these tests confirmed that Kanzi and Mulika had associated various referents with lexigrams, that these associations were bidirectional, and that spoken English words were as closely linked to lexigrams as were the items that the lexigrams represented.
At the time these tests were given to Mulika, it was thought that she only knew only a few lexigrams well enough to accurately select them out of context in a formal test. However, her test results quickly revealed that her capacities had been underestimated and illustrated that she knew 42 symbols, a number of which she had not yet used herself.
In order to rule out the possibility that Kanzi's ability to respond to spoken English was based on some nonphonemic aspect of speech (such as the rhythm or intonation pattern of speech) the English word comprehension test was readministered using a Votrax speech synthesiser. Each symbol was presented three times during this test and was paired with two different alternative on each presentation. Speech produced by the synthesiser is often difficult for individuals unaccustomed to it to interpret. Also, words vary in their clarity, with some being more interpretable than others. Kanzi's accuracy with synthesised speech was slightly lower than with normal speech, however, his difficulties were not global, but linked to particular words. It was because of these difficulties, this test was not used on Mulika. The authors experienced difficulties understanding many of the same words.
ANNOUNCEMENT AND VERIFICATION OF TRAVEL PLANS
When traveling long distances (over 20 m.) Kanzi typically rode on a person's shoulders and extended his hand to indicate which way to go. If the person did not respond he would turn their head in the appropriate direction. As he became older, he was encouraged to walk greater distances. Kanzi learned the location of 17 different food sites within 4 months after they were introduced. He seemed to understand within a few days that the photographs (recall that several photographs were typically laid out from which Kanzi was to indicate/choose a destination) served as a means of communicating to the people where he would like to go, and at times he would go out of his way to make sure that the person had seen the photograph that he had selected.
During a blind test of Kanzi's knowledge of the foraging sites, he selected a photograph on five occasions, and a lexigram on seven occasions; both a photograph and a symbol were chosen on three occasions. Each time he then led the experimenter to the correct location, sometimes traveling over 30 minutes to reach the previously selected destination. En route, Kanzi often pointed to the photograph or symbol, as though to remind himself and the experimenter where they were headed. As Kanzi travelled from one place to another, many opportunities arose for him to take a wrong turn and to thus go somewhere else other than the destination that he had indicated. On all but one trial, Kanzi took the most direct route possible. In the single instance that he did not, he directed the naïve experimenter to take him to the back of the 55 acres, an area that he is not normally allowed to go. He directed the experimenter to the very end of the enclosure before returning. Presumably, Kanzi took advantage of the experimenter's naiveté to go places he did not normally get to travel. This out of the way side trip lasted 30 minutes and took the experimenter off the trails and through some of the densest brush possible. After exploring this area, Kanzi led the experimenter back to the trail and on to the correct location.
Kanzi did not suggest traveling to two places in the field during this test. Consequently, when Kanzi no longer appeared to want to initiate travel, the experimenter used spoken English to ask that Kanzi lead her to the two remaining unvisited locations. Kanzi obliged, again taking the most direct possible route. Kanzi's ability to announce his desire to travel to a particular location, and then to lead someone there, cannot be attributed to a set travel plan, or to a routine, because his travel patterns varied considerable from day to day.
This test was not given to Mulika, because at time of writing she was not comfortable travelling in the woods without Kanzi along; in addition, Kanzi generally insisted on leading the way.
GENERAL OBSERVATIONS REGARDING KANZI AND MULIKA'S SYMBOL USAGE
Kanzi seems able to generalise beyond the typical referent of a given symbol, as have other apes. He has, for example used "coke" to refer to all brands of dark carbonated drinks, "rain" to refer to the sprinkling produced from the spray of a hose, "tomato" to refer to a variety of small, round red fruits (e.g. strawberries and cherries), and "bread" to refer to all varieties of bread including taco shells. Kanzi has used "hamburger" to refer to all cuts of beef, raw or cooked, "hide" to refer to hiding, finding, and exploring new areas of the forest, and "chase" to refer to following, as well as games of tag. Some symbols are used for a number of different referents, the dimension of similarity was not recognised by the staff. For example, he has used "clover" to refer to the specific plant, but he also uses it to refer to parsley that grows in tight clusters on the ground and to red bud blossoms that grow in tight cloverlike clusters on a tree. By contrast, other terms applying to outdoor vegetation, such as "pine needles" and "velvet plant" have been used only for these specific plants. Similarly, Mulika has used "apple" to refer to plums, "paint" to refer to crayons, and "peas" to refer to green beans.
Kanzi is also able to use symbols in different ways depending upon the context. For example, because many of the foods are located in the woods, he often uses a food name such as "juice" to indicate that he wants to go to the location (treehouse) where juice is typically located. Upon arriving there, he may show no interest in the juice, but play on the ropes, and in the treehouse located at this particular area. Yet, even while Kanzi is playing, if he is asked to select the juice from a group of foods, he will do so. If a bottle of juice is held up and he is asked to name it, he will do so readily and then continue playing.
When particular foods are not associated with particular locations. Kanzi has nonetheless learned to use the location name. However, the area of space to which he applies the symbol often seems to be much broader that the area to which the symbol refer. For example, Kanzi's use of "Sue's Office" refers not only to the indoor area of this office, but also to a large area outdoors that is around the office. Similarly, "play yard" is a term used to refer to the outdoor caging area occupied by Matata. Kanzi has used this term to refer to the entire grassy area behind the lab, and as a means of requesting both to go into and out of, Matata's outdoor cage. Also, if Kanzi is out in the woods, a long distance from the lab, he may use a number of words such as "Sue's office", "bedroom", or "child side" to indicate that he wishes to travel back toward the lab. Upon arriving at the lab, however, Kanzi will then specify a particular part of the lab that he did not mention out in the woods. It is as though the farther he is from the lab, the more general is the referent of the location terms that are used. However, once he is within the lab, he then uses terms with greater specificity.
Kanzi frequently takes the keyboard and goes off by himself to use it. He seems to be practising in such circumstances. He has, for example, pointed to a lexigram such as "pine needles" and then gathered together a pile of pine needles. He has touched the symbol for "rock" and then piled tiny pebbles on top of the rock lexigram. He often points to "hide" and then covers himself with the keyboard or blankets. When experimenters attempt to interact with him during such activities, he abruptly terminates the interaction and departs, as though such things are not meant to be social activities. Similar behaviour has not yet been observed in Mulika.
Kanzi also has used the keyboard to express a willingness to behave differently. On occasion, if he is taken inside when he does not want to go he will respond "no colony" or "no play yard" to indicate that he does not want to be where he is. Once, when he was disciplined by being confined indoors for repeatedly eating wild mushrooms. Kanzi emphatically touched "no bite mushrooms". When he was asked if he would be "good outdoors" he responded with positive vocalisation and then behaved appropriately when allowed to go back outdoors.
Discussion:
It is important to note that the data in the present report were compiled from the daily notes and reflect all of Kanzi's and Mulika's utterances during the period covered by this report.
Although we recognise that the number of subject involved in the present study is small, some preliminary species contrasts are in order. Among the many performance differences that have emerged between the common and the pygmy chimpanzee reared by people and exposed to a graphic symbol communication system, the most striking to those who have worked daily with both species is the ease with which Kanzi and Mulika have comprehended that lexigrams could be used as a mode of symbolic communication to communicate about absent referents and events. Sherman and Austin (common chimpanzees) did not spontaneously form associations between lexigrams and objects as did Kanzi and Mulika. Even after such associations were carefully taught, further training was required to separate the communicative acts of naming and requesting, and still further training was required to develop proper receptive responses to lexigrams used by others. Not until Sherman and Austin had received 3 years of training did they begin to evidence a clear concordance between what they said and what they then did. Kanzi and Mulika seemed to spontaneously grasp the idea that the lexigrams stood for objects and for events and that one used them to communicate information about those objects and events. They did not need to be taught to differentiate between naming and requesting, nor did they need to be taught receptive skills. Unlike Austin and Sherman, Kanzi's and Mulika's receptive skills typically preceded their productive skills. Because they learned words before they began to use them, when usage did appear, it was usually appropriate from the start, just as is this case with normal children. By contrast, Sherman and Austin (and other language trained apes) have gone through a period of new-word usage in which the initial usages are incorrect and drop out, as the correct associations become formed through production.
Kanzi's and Mulika's mother, Matata, did not acquire symbols spontaneously. Although she did begin to learn with training, she nonetheless still evidenced far more difficulty than did Sherman and Austin. Her difficulties suggest the there may be a critical age in the pygmy chimpanzee, beyond which acquisition of a symbol system is very difficult. Such a limiting effect suggests that perhaps other rearing variables might account for the observed differences between Sherman and Austin as contrasted with Kanzi and Mulika. Differences in rearing environments did exist and are difficult to quantify. Nonetheless, the observations that Sherman and Austin needed training and that they did not comprehend spoken English well, are in agreement with those of others who have reared common chimpanzees in human environments from infancy. This makes it unlikely that Sherman's and Austin's specific environment led to these difficulties. However, infants of both species are presently being reared side by side in the same environment as Kanzi and Mulika to answer this question more thoroughly.
A second major difference between the species is Kanzi's and Mulika's ability to comprehend English words. Sherman and Austin responded to English only in situations that provided them with contextual information, including the nonverbal glances and gestures of the speaker. Given the cues of context, intonation, and nonverbal gestures, they reacted appropriately to simple commands such as "don't do that", "wait", "it's your turn", or "open the door". However, if asked to retrieve a specific object, they became confused when more than one object was available and they were only given the English word as input. They then relied on pointing gestures to indicate which object they were being asked to retrieve. These observations are in accord with previous work using home reared Pan troglodytes chimpanzees.
Kanzi's and Mulika's ability to comprehend single words is clearly not context dependent. It is important to note that at no time did anyone try to teach Kanzi or Mulika to respond to English commands, or to utter single words over and over to them. People spoke normally around them at all times. Their ability to respond to specific words, which they extracted from complex phrases, was completely unforeseen when the study started. The speech synthesiser was added to the keyboard after data were obtained that revealed comprehension of spoken words. To the degree that Kanzi and Mulika can comprehend even single words as communicative referential utterances, they are demonstrating a basis for language comprehension that was not available to Sherman or Austin. Understanding that people use spoken words to stand for things and understanding what many of those words are give Kanzi and Mulika a decided advantage when they encounter lexigrams being used as symbols. Instead of having to learn what symbols are and how they function communicatively, they only have to learn to "read" the lexigrams or to pair them with the spoken words they already know. As Mulika's data shows, this process can occur at a very early age and can predated the fluent usage of the symbols. The way that pygmy chimpanzees acquire graphic symbols seems to be fundamentally different from the way in which common chimpanzees learn them.
A third difference is found in the specificity with which lexigrams are associated with referents. Sherman and Austin were readily inclined to acquire broad differentiations, such as between playing, eating, drinking, or grooming, but differentiation within these categories were not something that they spontaneously treated as important. Only when the training environment required attention to such differentiations did they become reliable. Much has been made of the ape's ability to form general categories, yet it was far easier for Sherman and Austin to learn general categorical terms than it was to form discreet item-symbol association within categories. Additionally, they tended to drift toward a single term for all items within a category. Had their training begun with symbols whose referents were maximally discriminable such as eat, groom, play and outdoors, instead of such items as orange, banana, and M&M, their early learning surely would have been more rapid. Conceivably, they may not have required training to learn such broad-based symbols, particularly if the position of the lexigram was permitted to be confounded with its appearance. However, as discrimination within such categories were attempted, it is likely that they would have experienced considerable difficulty.
By contrast, differentiations between such items as juice and coke were made spontaneously by Kanzi and Mulika and no special effort was required to maintain them. When Kanzi did generalise the use of his symbols, it was typically because there was no lexigram for the similar item, not because he made "within category errors" Once coke and juice were learned, Kanzi did not use these symbols interchangeably, nor was any review or drill required to help Kanzi maintain the proper association of each symbol with it referent.
A fourth different between Kanzi and Sherman and Austin was Kanzi's ability to request that A act on B, when he was neither A nor B. Sherman and Austin never formed requests in which someone other than themselves was the beneficiary of the request. The ability to conceptualise and then to symbolically initiate complex forms of interaction between others would seem to be the precursor of syntactical structure, if not for the basis itself for the occurrence of syntax.
Conclusion:
The main problem with Case Studies such as this is that you can gather a huge amount of data, but then the ability to apply this data to other individuals is extremely limited. That is the case here, the authors of the study recognise that with such a small sample can they assert that the differences they found can be applied universally. They point out that what was found to be sufficiently interesting to carry out further research to sort out the variable in rearing of the different types of chimpanzees.
Presently, infants of both species are being reared together to address this question effectively.
Why should two such closely related species of chimpanzee differ so greatly in their ability to acquire a functional symbolic communication system? Presently, there is no answer to this question, but it is hoped that the further research planned will provide some answer to this.
There is no question that the individuals in this study are representative of their species, the pygmy chimpanzee appears to possess a far greater propensity for the acquisition of symbols than other apes.
Designed by Tony; powered by Wix.com