Pygmy chimps are used (Pan paniscus)
2 chimps that have learnt language from other chimps.
Can produce agent-verb-recipient combinations.
Common chimpanzees (Pan troglodytes) are unable to do this. Use simple verbs only referring to the addressee.
Associational learning – symbol enables the acquisition of a particular object.
Representational learning – comes after associational learning and means that the ape can label objects and refer to objects that are not present.
Disappointing results from previous research.
Research based on Pygmy chimps Kanzi mainly and younger half-sister Mulika.
They are highly social animals and thus are more likely to have a predisposition for acquiring language.
Pan Paniscus – reared with access to mother. Kanzi 4 years old male, born in captivity whereas mother (Mutata) was caught in the wild. Started learning at 6 months. Also Mulika the younger half-sister who was attached to humans.
Pan Troglodytes – reared with language between 1975 to 1983. Austin and Sherman. They started language training at 1˝ and 2˝ years respectively.
Lexigram board – computer with voice synthesiser, but pointing board was used in the field owing to the computer board not being robust enough. Also 100 ASL (American Sign Language) signs were used. The pointing boards were not so good for use by the Pan troglodytes as they use broad hand movements.
Matata (mother) was trained whilst she was looking after Kanzi. Kanzi was not trained at this time but could observe his mother training. Kanzi often interrupted. At 1˝ years Kanzi showed an interest in the keyboard, randomly pressing keys to obtain food. Mulika (younger sister of Kanzi) did not observe her mother training but did observe Kanzi. Both Kanzi and Mulika became more attached to humans after a brief separation from their mother (for medical reasons).
The Pygmy chimps Kanzi and Mulika were trained in a similar way to the Common Chimpanzees Sherman and Austin, except that the common chimps could not use a portable keyboard and they did not understand human speech, so tones were used every time a symbol was pressed. Kanzi and Mulika had synthesised speech on their keyboards.
Kanzi and Mulika did not have formal training because:
A) they began using symbols without training.
B) They identified symbols regardless of the position on the board.
C) They did not confuse symbols for similar items.
D) They understood spoken English.
Trainers modelled spoken English and symbol use in a variety of natural settings.
Kanzi and Mulika were trained to collect food from locations around a wood. Started training with pictures of food. Kanzi could select the picture that represented the food that he wanted to travel to. Upon retrieval the chimps were given the opportunity to carry their food in a backpack if they responded positively to the question. If Kanzi chose food that was already in his backpack he was then asked whether or not he wanted to eat the food in the backpack or travel to the location where the food was stored. After 4 months of training Kanzi (aged 3 years) was able to lead a ‘blind’ experimenter to the correct locations of chosen foods.
Kanzi and Mulika, like Austin and Sherman join in household chores and enjoy entertainments including specially prepared videos with lexigram subtitles.
Records for Kanzi from 2˝ years old until 17 months after that. Symbol use was automatically recorded indoors, but researchers took notes when outdoors with the portable lexigram board. Mulika was studied from 11 until 21 months. Utterances were recorded as spontaneous, imitated or structured (which means replies to questions from the researchers).
A lexigram was said to have been learnt if it occurs nine times out of 10 spontaneously and was verified (concordance between sign and behaviour). There was strict criteria. If for example Kanzi selected “banana” and pointed to the rucksack which contained a banana this was not adequate for concordance. A sign could be dropped from the chimp’s vocabulary if it was not used correctly 90% of the time. Four and a half hours of videotape was checked by an independent observer. The original observer (real-time observer) was not aware of the checking procedure. There was 100% agreement. Nine utterances were noted by the independent video observer that were not noticed by the real-time observer. This suggests that many signs are missed in real-time.
Tested informally in everyday situations. At the end of 17 months Kanzi was tested under control blind lab conditions. Mulika was tested between the ages of 18 and 21 months. The testing took place in the afternoon with between 20 and 14 trials in one session. Three or four alternatives were used. Alternatives were not used on consecutive trials. All of the alternatives were selected from targets.
Photograph to lexigram- shown a photograph and have to choose one of three lexigrams. The alternatives were selected by a second blind experiment and the choice was not observed by the original observer (behind a folding blind).
Spoken English to photograph- “Kanzi, can you show me X?”
Sherman and Austin were tested on their vocabularies out without the spoken English tests. They were rewarded with food unlike Kanzi and Mulika.
Between 6 and 16 months Kanzi and Mulika used iconic gestures for travel and actions. For example pointing, twisting for undoing lids, cracking nut gestures, blowing up a balloon. These were often accompanied by affective vocalisations. Sherman and Austin only used gestures when being given full human attention, and would not persist. They did not elaborate or give specific gestures but just generally pointed.
After a separation from the mother Kanzi’s use of the keyboard was less random and more purposeful. Could indicate particular foods, or ball to play with. Mulika began to show signs at 12 months. Tended to use “milk” for a variety of activities. After two months symbols began to be selected more appropriately. After “milk” the second symbol was “surprise” which indicates a randomly selected reward. Kanzi learnt the symbol “strawberries” whilst eating mushrooms and at first only used this symbol in this context.
Kanzi
seven words at 30 months. 44 words at 46 months (productive) (over 50 receptive) but five words dropped out of use. Combinations at 46 months were approximately 800. Kanzi’s receptive mode higher than productive mode. 80% were spontaneous.
Mulika
“milk” at 17 months. Six words at 22 months.
Kanzi’s keyboard grew as the vocabulary grew, whereas Mulika had to contend with a large keyboard from the start.
6% of utterances. Reduced combinations without repetition, such as “ice water go” meaning get some ice water, rather than “play me Nim play” (Terrace et al 1979). Kanzi produced 2,540 utterances compares to 19,000 for Nim. Kanzi used combinations for games or to benefit others, whereas Nim’s combinations were for food for himself only.
Kanzi invented combinations that asked researchers to chase each other, which was not demonstrated in advance.
Kanzi and Mulika’s utterances were truly spontaneous whereas Sherman and Austin’s were often prompted by some activity of the researcher (e.g. placing food).
Kanzi and Mulika understood the abstract nature of these tests. No rewards or activity would follow the choice. Sherman and Austin expected to be rewarded according to their choice.
Sherman and Austin were unable to do the Spoken English tests and would show signs of frustration (scratching all over their bodies). They would instead attempt to mimic spoken English unsuccessfully.
As a control for the possibility that Kanzi might be responding to voice intonation he was tested on a Votrax speech synthesizer. Each symbol was presented three times and was paired with two alternatives. Kanzi did well on this task but not as well as with human speech, but Kanzi failed only on specific synthesised words, which even the researchers found difficulty in comprehending.
Kanzi learned the location of 17 food sites within 4 months.
During a blind test of Kanzi’s knowledge of the foraging sites, he selected a photograph on five occasions and a lexicon on seven occasions; both the photograph and a symbol were chosen on three additional occasions. Each time he then led the experimenter to the correct location, sometimes travelling over 30 minutes to reach the previously selected destination. On all but one trial, Kanzi took the most direct route possible. In the single instance that he did not, he directed the blind experimenter to take him to the back of the 55 acres, an area where he is not normally allow to go. Presumably, Kanzi took advantage of the blind experimenter’s naivety to go places that he did not normally get to travel. Kanzi did not suggest travelling 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 un-visited locations. Kanzi obliged, again taking the most direct possible route. This ability cannot be attributed to a set travel plan or to a routine, because his travel patterns varied considerably from day-to-day. This test was not given to Mulika because she was not comfortable travelling in the woods without Kanzi who generally led the way anyway.
Kanzi mainly, and to an extent Mulika, generalise their symbols to other related objects. For example ‘coke’ is generalised to all dark fizzy drinks or ‘rain’ is used to refer to water from a garden sprinkler. Kanzi uses the food names to indicate where he wants to go in the woods. Remember foods are located in various parts of the wood and juice, for example, is located at the tree house. So instead of indicating the tree house, Kanzi indicates ‘juice’ but upon arrival at the tree house he is not interested in drinking the juice. Kanzi sometimes practices on the keyboard alone. For example, he might indicate ‘pine needle’ and then proceed to pick up pine needles. If interrupted by researchers Kanzi would cease the activity. When punished by being kept indoors Kanzi indicates ‘no colony’ or ‘no play yard’. Once he was punished for eating wild mushrooms he indicated ‘no bite mushroom’ and when asked whether he would be good outdoors he gave positive vocalisations.
It is recognised that the sample size of 2 pygmy chimps and 2 common chimps is small. Also the environments were different for the two pairs of chimps. Nonetheless it was felt that the empirical evidence was so strong as to discount these possible confounding variables.
1. Kanzi and Mulika learnt rapidly whereas Austin and Sherman needed a lot of coaching. Kanzi and Mulika spontaneously recognised that lexigrams can stand for objects and events. They did not need to differentiate between naming and requesting and their receptive skills preceded their productive skills (they understood words before they used them). Interestingly, their mother who started to learn late found it more difficult to learn a language and performed more poorly than Austin and Sherman; this suggests that there is a critical period in which chimps can efficiently acquire language. To control for the effect of different environments the researchers are now raising common and pygmy chimps side by side.
2. Kanzi and Mulika understand spoken English whereas Austin and Sherman do not, they relied more on pointing to indicate which object was to be retrieved, for example. Also Kanzi and Mulika did not require strong contextual clues. Because of the ability to understand spoken English it is thought that the learning of the lexigrams would be facilitated and that they learnt these lexigrams in an entirely different way to how Austin and Sherman learnt them.
3. Kanzi and Mulika could be more precise with their symbols. Austin and Sherman would confuse ‘juice’ and ‘coke’ and could only handle broad definitions, whereas Kanzi and Mulika readily differentiated between lexigrams for similar items. Kanzi only appeared to make ‘within category errors’ when there wasn’t the specific lexigrams on the board.
4. Kanzi could ask for A to act on B, where neither A nor B was Kanzi, whereas Austin and Sherman could only form requests that involved themselves.
Assuming that Kanzi and Mulika are representative of typical pygmy chimps it is concluded that pygmy chimps have a far greater propensity for the acquisition of symbols than other apes.