Teaching Sign Language to a
Chimpanzee.
R.A Gardner and B.T. Gardner (1969).
Preliminary Considerations
The Chimpanzee as a subject.
A chimp chosen for this study because it is an intelligent and social
animal. The main disadvantage of using a chimp is that it does not possess
vocal apparatus that would allow the production of human speech. Hayes and
Hayes (1951) tried this; their chimp, 'Vicki', produced only four sounds in six
years! Chimps do use their hands a lot in their natural habitat, so it was
decided to build upon this natural ability by training the author's chimp,
'Washoe', to use American Sign Language (ASL).
American Sign Language.
Each symbol will vary in its degree of abstraction (e.g. sign for 'flower'
involves putting fingertips together and touching both nostrils. This is
iconic, as the sign mimes a flower being held to the nose. Other signs are
quite abstract, for example 'dog' is signed by slapping the thigh!). The
performance of Washoe can be compared with that of deaf children.
Washoe.
Caught in the wild and received by the Gardners when she was between 8 and
14 months. Chimps are completely dependent until two years of age, and
semi-dependent until the age of four. Full adult growth is reached between 12
and 16 years.
Laboratory conditions.
The Gardeners tried to make Washoe's environment as similar as possible to a
[deaf] human infant. Many helpers were used. There was always somebody in
attendance during Washoe's waking hours. Every helper communicated with Washoe
by using ASL, rather than with the spoken voice (the use of which was
minimised). Helper's acted as friends and companions to Washoe, making use of
various games and activities to make the learning experience enjoyable.
Training Methods.
Imitation.
As with chimpanzees in general, Washoe naturally imitated. Washoe signed the
sign for 'toothbrush' spontaneously upon entering the Gardener's bathroom and
noticing toothbrushes. There seems to have been no obvious motive, except to
communicate.
Babbling.
'Babbling' here does not mean vocal babbling, rather the untaught signs used
by Washoe to communicate a desire. Washoe used a begging gesture, which was not
too different from the ASL signs for 'give me' and 'come'.
Instrumental conditioning.
Humans could not learn a language, purely by instrumental conditioning,
although it seems likely that the 'trick vocabulary' of early childhood could
be acquired in this way. Instrumental conditioning was one strategy used with
Washoe. Tickling was used as a reward. The sign for 'more' was learnt by
instrumental conditioning. This sign was later applied to a variety of relevant
situations.
Results
Vocabulary.
A Sign was added to a checklist when reported by three independent
observers. The sign had to occur in an appropriate context and without
prompting. The checklist was used to record the frequency of a sign. A sign had
to be used at least once a day for 15 consecutive days, before it was deemed to
have been acquired. Alternatively a sign had to be used at least 15 days out of
30 consecutive days. By the end of the 22nd month of the project, thirty-four
signs had been learnt.
Differentiation.
The sign 'more' used in many different situations until a more specific sign
had been learnt. The sign for 'flower' used inappropriately for the idea of
'smell'. After additional training eventually Washoe was able to differentiate
between 'smell' and 'flower'.
Transfer.
Although the same object was presented for each learning trial (e.g. the
same hat), Washoe was able to use the sign for other similar objects (e.g.
other hats).
Combinations.
Washoe was able to combine two or three signs in an original way, examples
being 'gimme tickle', 'open food drink' (meaning 'open the fridge'), 'please
open hurry', etc.
Concluding observations.
We can not be certain that Washoe has acquired 'language'. This is because
it is difficult to draw the line between what counts as 'possessing a language'
and 'not possessing a language'. If Washoe were to communicate information to an
observer, whom had no other source of information, then this would be strong
evidence to support the claim that Washoe possesses 'language'.
Commentary.
It is not just the setting up of valid dependent variables that are
important, perhaps what is more important, is the interpretation of the
dependent variables. Does Washoe's use of signs constitute language? Chomsky
(1957) believed that language is unique to humans, claiming that we have a
Language Acquisition Device that enables us to appreciate readily the
grammatical structure of language, enabling the infant to make sense of what
would otherwise be a chaotic mass of sound. Although the chimpanzee's vocal
apparatus can not produce speech sounds, this does not mean that the animal can
not communicate effectively non-orally. Premack (1971) used plastic shapes.
Savage-Rumbaugh et al (1980) used a special computer that produced geometric
shapes that represented words. Others have used ASL (e.g. Patterson 1978 taught
a gorilla, Terrace 1979 taught a male chimp. In all these cases it has to be
decided whether the animal concerned possesses language. Aitchison (1983)
proposes these 10 criteria for language:
- Use of the
Vocal-Auditory Channel
- Arbitrariness (use of
neutral symbols (i.e. words) to denote objects, etc.)
- Semanticity (the use
of symbols to mean or to refer to objects/actions)
- Cultural transmission
(handing down from generation to generation)
- spontaneous usage
- turn-taking
- Duality (a
combination of different sounds and the order in which these sounds are
placed)
- Displacement
(references to things not present in time or space)
- Structure-dependence
(e.g. the structure of grammar and the relevance of word order)
- Creativity
An argument that we meet when considering the work of Piaget is relevant
here. The Discontinuity theory of language considers the differences between
human and animal language as qualitatively different, rather than
quantitatively different.
Evaluation
- Washoe demonstrates semanticity
(e.g. 132 signs learnt, able to generalise). There is some evidence of
displacement (e.g. 'all gone cup' and 'more milk'). Washoe combined words
showing evidence of creativity. Regarding structure-dependence, Washoe was
not always consistent with word order. This could be due to one or more of
the following reasons:
a) the Gardeners rewarding Washoe for correct sign production regardless
of order,
b) sign order is not as consistent as word order, this is evidenced by
deaf people, who also produce signs in an inconsistent order,
c) this was an intermediate stage,
d) Washoe could not understand the importance of the correct ordering of
the signs she was producing.
- Sarah and Lana were trained
to use symbol order correctly and coped to a limited extent. Koko did not
use any particular order. Nim Chimpsky preferred certain sign orders, but
failed to demonstrate that he understood any rules.
- Petitto and Seidenberg
(1979) conclude that apes use repetitive, inconsistently structured
strings. Nim's longest recorded utterance is 'Eat drink, eat drink, eat
Nim, eat Nim, drink eat, drink eat, Nim eat, Nim eat, me eat, me eat'.
- We must question whether
the apes demonstrated true semanticity. Are they really using the signs as
words? Also, are the apes being spontaneous? Chimps try to 'acquire'
objects rather than idly converse about them.
- Children can understand
that 'dog chase cat' has a different meaning from 'cat chase dog', unlike
apes. We could compare the Mean Length of Utterance (MLU) for children with
the MLU for apes. This comparison would be misleading owing to the
repetitive nature of the strings created by apes. Ignoring this, children
quickly move from a MLU of 11/2 to a MLU of 4 words. Nim's MLU did not
rise, holding steady at around 1.1 to 1.6. Nim did not sign spontaneously
on his own initiative. Trainer prompted most signings. Nim would often
start to sign before the trainer had finished signing; this suggests that
he is not having a genuine conversation.
- Gardener's attack Terrace's
upbringing of Nim, saying that he was subjected to routine drills rather
than being brought up as a child would be.
- Eysenck (1984) points out
that the Gardners concentrated on only one component of an ASL sign, that
being just the hand configuration. Movement, orientation and location were
neglected. ASL does not have the grammatical structure of English, so we
should be careful about equating signing ASL with speaking a language.
Nonetheless, deaf people are using ASL as a language, so we should accept
that if a chimp uses ASL then he or she is using a language also. Brown
(1986) suggests the chimp should use ASL in such a fashion that would
convince us that a human had the capacity to use ASL. However, we must be
wary of anthropomorphizing (viewing animal behaviour as if it were acting
as a human).
- Chomsky (1980) believes
that apes lack the capacity to understand the 'rudiments of the
computational structure of human language'. Aitchison (1983) believes that
the relative ease at which humans acquire language point to an innate
ability. Eysenck (1984) believes that language in its complete form is
unique to humans.
- Savage-Rumbaugh believes
that the difference between human and chimp language is quantitative. She
trained bonobo chimps to use a 'lexigram', in a naturalistic setting. This
consisted of a portable board that could be carried to wherever the chimps
were, and used regardless of whatever they were doing. Each time a
geometric shape on the board was pressed a word would sound. The board was
used to allow the chimp to communicate in much the same way as a child
would use spoken language. The board was used successfully, with the
chimps displaying relative ease of learning the symbols, and by their
using the symbols spontaneously.
- The Gardners didn't see
Washoe for 11 years, yet when she saw them Washoe immediately signed
'Come, Mrs G.', and led her to an adjoining room and began to play a game
with her. Can we justify taking an animal away from its natural
surroundings? Should we try to teach them an unnatural language? We should
consider the possible psychological damage that might be caused by all of
this teaching. The Great Ape Project, supported by eminent people,
believes that apes should have many of the rights of humans.
Questions
- What is anthropomorphism?
[2 marks]
- Give two other methods used
by other chimps as alternatives to signing. [2 marks]
- What is the differences
between the chimp's and the children's use of language? [8 marks]
- Can apes acquire language?
[7 marks]
- Discuss the programme of
teaching language to chimps from an ethical standpoint. [6 marks]
Links
Jane
Goodall's article on Chimps
Computer
masters baby talk
Your questions
2) Comment on what Gardner & Gardner tells us about the human behaviour
and experience it investigated.
The development of language in the chimp parallels the
development of
language in children (at least deaf children). Chimp language just takes
longer. The acquisition of language seems to require much interaction
with
caregivers and is developed through relating language to everyday events
that are shared between the chimp and the caregivers; just as in humans!
4.Gardner
and Gardner: Chimpanzee study
Finally, I
was wondering if you could give me one ethical issue-apart from the issue of
implications for the rights of Apes-testing, and of the removal from its
natural environment. Also there was 2 other issues (not ethical) which were
found within the study.
Ethics:
- Washoe was indoctrinated into the ways of humans and
then mercilessly put into a cage once the Gardner's had finished with her.
- Taken from the wild.
- Deprived of the companionship of other chimps.
- Strain of being encouraged to learn things that were not
natural to the animal.
Methodology:
- Experimenter bias. Although this could well have been an issue, the
Gardners did take steps to avoid this. The Gardners used blind observers. Those reporting
Washoe's answers never knew what referent Washoe was looking at. This
eliminates the Clever Hans Error that Terrace and others ignore.
- Imitation rather than learning. This was Terraces ‘big conclusion’. Terrace used rigorous conditioning
procedures, unlike
the Gardners, which led to Nim begging for treats and not signing for the
sake of signing. A study by O'Sullivan and Yeager (1989) with Nim showed that there was much
less imitation in a relaxed conversational setting compared to a setting
where Nim was rewarded for signing. By the way, imitation is not diametrically
opposed to learning. Many humans and non-humans learn through imitation. There
is much imitation in speech of children. A type of sentence that occurred
in the signing of the chimps and occurs in humans is expansion. Repeating
a word or sign of the other interlocutor then adding more words or signs to
add to the information.
Results:
- Poor rate of learning, although the rates of learning
are similar to children up to a certain age. Patterns of acquisition are similar
- No grammar. Although American Sign
Language is an inflected language and the chimpanzees used inflections appropriately
(see Rimpau, Gardner, and Gardner, 1989).
- Could not claim to show that chimps
could learn a language that was on a par with ours. This is debatable as ASL is a human language.
Symbol language is not natural and must be trained more rigorously. If you
measure the signing rates of the cross-fostered
chimpanzees Tatu and Dar vs. the lexigram production of Kanzi (a bonobo
trained and studied by Sue Savage-Rumbaugh) over an hour, Tatu and Dar
sign over a hundred signs while Kanzi produces less than 15 lexigrams.
Last updated 3rd January 2003