How language-like are animal communication systems?

How language-like are animal
communication systems?
 Primate gestures feature semantics but
no productivity or syntax
 No animal communication system
features recursive syntax or
productivity at multiple levels
 So what happens when you try to teach
language to animals…?
Intervention Studies
 “Suppose an anthropoid were taken into a
typical human family at the day of birth…
Suppose he were fed upon a bottle, clothed,
washed, bathed and given a characteristically
human environment; that he were spoken to
like the human infant from the moment of
parturition; that he had an adopted human
mother and an adopted human father… The
[optimal] experimental situation should be
attained if this technique were refined one
step farther by adopting such a baby ape into
a human family with one child of
approximately the ape’s age.”
 --Winthrop Kellogg
Intervention Studies
methodological
issues
the ‘Clever Hans’ effect
trainer overinterpretation
conditioning vs.
understanding
Intervention Studies
early studies
Gua
1951)
(Kellogg & Kellogg, 1933);
Viki
(Hayes,
chimpanzees raised with human children
failure to acquire spoken language due to
limitations of the chimpanzee vocal tract
American Sign Language
(ASL)
Washoe
(Gardner & Gardner, 1969)
chimpanzee trained on
vocabulary, simple syntax
acquired 100-200 signs, some
word order
claims of novel combinations
(e.g., ‘water bird’),
comprehension of whquestions
American Sign Language
(ASL)
Gardner & Gardner 1969
Vocabulary test
Two observers
interpreting gestures
Performance ranged
from 54% to 88%
Gardner & Gardner 1984
Koko (Patterson, 1978)
 Gorilla trained on ASL
 Very little data; only consists of single examples (after
32 months of data collection)
 K: You me there (indicating the refrigerator)
B: What want?
K: Eat drink.
B: Eat or drink, which?
K: Glass drink.
B: That? (Gets out juice.)
K: That.
 “Cookie rock” for stale roll, “eye hat” for mask… But how
many uninterpretable examples were there?
 The gestures resemble ASL only superficially
Petitto and Seidenberg 1979
Question: Koko are you going
to have a baby in the future?
Koko signs: Pink
Patterson explains: We had
earlier discussion about
colours today.
Question: Do you like to chat
with people?
Koko signs: Fine nipple.
Patterson explains: Nipple
rhymes with people, she
doesn't sign people per se, she
was trying to do a "sounds
like..."
Question: Does she have hair?
Or is it like fur?
Koko signs: Fine.
Patterson explains: She has
fine hair.
Question: Koko, do you feel
love from the humans who
have raised you?
Koko signs: Lips, apple give
me.
Patterson explains: People give
her her favourite foods
‘Can an ape create a sentence?’
 Did Washoe and Koko acquire a
grammar?
 Insufficient data; no report of whether
“more tickle” was more common than
“tickle more”
 Was “water bird” a novel meaning
produced by a combination of signs? Or
did Washoe just name the water, then
the bird?
Terrace et al. 1980
‘Can an ape create a sentence?’
Nim Chimpsky
(Terrace et al., 1979)
ASL-trained chimpanzee
systematic analysis of over 20,000 signs
some systematic ordering of two-sign
utterances
longer utterances
➔ less structure,
more repetitions
‘Can an ape create a sentence?’
Terrace et al. 1980
‘Can an ape create a sentence?’
 Many of Nim’s
multi-word
utterances
seemed to be
imitations of the
teachers
 This may have
been true for
Washoe and
Koko as well
Terrace et al. 1980
Symbol Systems
Sarah
(Premack, 1972)
arbitrary plastic symbols
trained on labels,
sequencing
acquired sequencing,
some substitution
Premack 1972
Premack 1972
Is this “semantics”, or just an elaborate chain of stimulus-response
associations for reward?
Tests were forced choice
Premack 1972
Symbol Systems
Lana, Sherman, & Austin
(Savage-Rumbaugh & Rumbaugh, 1978)
Yerkish
Symbol Systems
 Use of symbols was overly general, yet
inflexible
• Continued responding after machine ran
out of food
• No differentiation between GIVE and POUR
• Tended to keep using last correct symbol or
use the symbol for a high-preference food
• More easily made associations between
symbols and positions than between
symbols and objects
Savage-Rumbaugh & Rumbaugh 1978
Kanzi
Symbol Systems
(Savage-Rumbaugh et al., 1993)
bonobo (pygmy chimp) began to
learn Yerkish from mother
use of >50 symbols in >800
combinations by 46 months
word order comprehension
verb meaning
Kanzi’s word order
 Learned from observation:
 Action-agent: 119 examples, agentaction: 13 examples
 Goal-action: 46 examples, action-goal: 10
examples
 Invented:
 Gesture follows lexigram (182/249
examples)
 Very few three-word examples
Greenfield & Savage-Rumbaugh 1990
Kanzi
Sentences in comprehension experiment:
Can you put the apple in the hat?
Put the ball in the oil
Knife the sweet potato
Make the doggie bite the ball
Put the ball on the doggie
Put the doggie on the vacuum
Performance was around 80% on reversible sentences (requiring
comprehension of word order)
Human two-year-old scored around 70%
Savage-Rumbaugh et al., 1993
Symbol Systems
production
comprehension
Kanzi
(Savage-Rumbaugh et al., 1993)
Other Animals
Alex, the grey parrot
(Pepperberg, 1996)
trained using social modeling technique
word and feature labels
Same/different
counting
feature conjunctions
Other Animals
Rico, the border collie
(Kaminski, Call, &
Fischer, 2004)
trained to fetch objects
based on spoken labels
vocabulary of 200+
words
evidence of ‘fast
mapping’ in word
learning
Other Animals
bottlenose dolphins
(Herman et al., 1984)
artificial gestural
language
‘inverse grammar’
2-5 word sequences
syntactic categories
Syntactical rules learned by dolphins
 Object + Action (WINDOW TAIL-TOUCH)
 Modifier + Object + Action (LEFT PERSON MOUTH)
 DO + Action + IO (HOOP FETCH PIPE)
 Modifier + IO + DO + Action (RIGHT BASKET PIPE FETCH)
 Modifier + DO + Action + Modifier + IO(P) (SURFACE PIPE
FETCH BOTTOM HOOP)
 Dolphins could generalize to new examples
 Performance ranged from 60 to 100%
 Dolphins refused to perform invalid sentences (Hermann et
al. 1993)
 Dolphins could repeat last action (Mercado et al. 1998)
Hermann et al. 1984
What’s Missing?
recursion
(Hauser, Chomsky, &
Fitch, 2002)
joint
attention/intentions
(Tomasello, 2006)