Design features: human language vs. animal communication systems
Chimp studies: can chimps learn language?
Evolution: some implications for language
Summary: "innateness" unit
Why study animal communication systems?
Similarities, differences between animal, human systems may help determine:
Whether or not humans are unique, and if so,
Which aspects of language are uniquely and definingly human.
Design features of language, animal communication systems
prerequisites to comparing the two systems:
--what counts as communication in animals (or in humans)? hiss of a cat? batting of eyelashes? and some animal phenomena which seem to be communicative are not interpretable (cf. discussion of dolphins in Akmajian)
start with last problem. cf. File 10. work by Charles Hockett.
some additional parameters added by O'Grady et al.?
In human languages, VAC is primary means of communicating. Why? (Evolutionary speculation: leaves the rest of the body free to carry on other things.)
--writing (but writing clearly a late invention, subordinate to speaking: many speakers of human languages are illiterate);
Some animals use visual channel for communication: bees, stickleback (fish) uses postures; or chemical: moths (Files).
Vocal-auditory channel (VAC) clearly not unique to humans or even mammals: birds use VAC. (crickets, rattlesnakes, woodpeckers send sound signals, but not through vocal tract.)
So use of vocal-auditory channel not a defining characteristic of human language.
arbitrariness of symbols; i.e. symbols are not iconic, not inherently meaningful, do not resemble in any way real world referent
possible exception: onomatopoeic words (sound symbolism) but very limited number of such words in any language, and if sound was truly iconic, then variation among languages shouldn't exist (see Files 22, 23). and would not expect to find variation in a particular language.
"Onomatopoetic" verbs in Witsuwit'en. All have structure: t-l-root
he/she is talking without teeth.
he/she is shouting, talking in a loud, angry voice.
he/she is talking with a lisp.
he/she is whistling.
he/she is screaming.
he/she is coughing.
they're crying, wailing.
he/she is talking fast.
he/she is talking while walking around.
they're walking around whistling.
it (animal) makes noise, calls from a long distance.
it (bear) is grunting.
it's pecking on wood.
it (bird) is chirping.
the squirrel is chattering.
it (squirrel) is chattering.
frog is croaking.
frog is croaking.
it (marmot) is squeaking.
it (rodent) is squeaking.
his/her heart is pounding, thumping.
it's popping (from heat).
it (air) is hissing while escaping.
(inanimate) is squeaking, creaking. (new shoes, floor, snow underfoot)
(inanimate) is squeaking, creaking. (louder noise than tlcets)
the tree is creaking, cracking from cold weather.
trees are creaking, rubbing together, making noise.
they're swishing. (sound of dry leaves swirling)
he/she is knocking.
other animals may use arbitrary or non-arbitrary (iconic) symbols:
--crabs indicate aggression by raising a leg: size(??) of leg indicates degree of hostility
--opening mouth and baring teeth to threaten: Japanese macaque in Figure 16.3
--responses of Vervet monkeys to 'eagle', 'leopard' and 'snake' predators (Figure 16.12)
--gulls indicate aggression by turning away from opponent and pulling grass out of the ground (arbitrary)
arbitrariness not unique to humans, despite File 10
Cultural transmission (tradition)
i.e. is handing down of system from generation to generation necessary? does a learner learn only from a more experienced user? humans: Critical Age cases show that learner cannot learn in absence of native speaker input. Chomsky: basic capacity for language is innate; details depend on exposure to a particular language (nature+nurture)
. i.e., how much learning exists in animal language?
'dialects' possible evidence for learning in animals, but some dialects are innate, some learned.
--bees: dialects are learned: hybrids of Austrian and Italian honeybees (latter are ones with intermediate distance Sickle dance) do dance of parent they most resemble physically; also bee raised in isolation has no problem doing dance
--birds: varies (as discussed in Ak.)
European cuckoo--innate: birds reared in isolation sing song of species;
male bullfinch--learned: bullfinch can learn canary song
chaffinch--part learned, part innate: year-long development period
some birds reared in isolation don't sing right; critical period for song acquisition mirrored in critical period for language acquisition
cultural transmission of system not unique to humans
other experiments with animals have shown this is true for other innate capacities, like vision
interchangeability of roles of sender and receiver of message. Humans can send any messages that they receive from other humans, and vice versa. Some limits: adults and babies don't communicate by using the same kinds of signals; also obviously have to be talking about humans that speak the same language.
some male and female differences in language. Witsuwit'en:
-yez 'her son, daughter'
-ts 'his daughter'
-yi 'his son'
Many animal communication systems don't have interchangeability:
--female vs. male differences in moth communication (Files).
but many animal systems do have interchangeable roles, so interchangeability not definitive
do signs of the system convey meaning? do the signs refer to specific or general objects
in animal communication systems, more limited range of meanings seems to be expressed
eg. of vervet monkey in Ak: danger calls which refer to specific types of danger--adder, lion, eagle, human (but may refer simply to degrees of danger)
semanticity (in sense of reference) a possible distinguishing characteristic
talking about things removed from here and now. only humans? (yes--Files)
bees can talk about sources of food that are not in immediate environment.
but bees can only talk about source of nectar just visited, not one visited yesterday, or two trips earlier
primates "do not appear to communicate about imaginary pasts or futures"
but even bee system, which contains displacement, is not creative:
von Frisch experiment: put dish of sugar water straight at top of radio beacon, positioned hive underneath the dish. bees who found the food did a round dance, but couldn't indicate straight up--sources of food typically found on ground, not in clouds
bird calls show no productivity. "birdsong shows evidence of recombination...but it is doubtful whether these recombinations transmit novel messages."
eg. of Buzz and Doris the dolphins
Duality of patterning
only combinations of sound units (eg.s), not units themselves, are meaningful. units can be combined in different ways to form meaningful signs. meaningful signs can be combined. so sound patterns below the level of the sign (morpheme); patterns above the level of the sign (combinations of morphemes). phonology and syntax.
/k/, /æ/, /t/: /kæt/, /ækt/, /tæk/
/tæk kæt/, /kæt tæk/
birds also have this: combinations of notes
O'Grady: "There is no evidence of this type of patterning in any known animal communication system."
Language is hierarchically organized. Arguments for constituent structure in syntax. Most easily seen with regard to syntax.
best defining chars. of language are creativity, structure-dependence, possibly semanticity, displacement. insofar as no animal system has these, lg is unique to humans.
O'Grady adds additional characteristics: prevarication (the system enables the users to talk nonsense or to lie), learnability (a user of the system can learn other variants), reflexiveness (the ability to use the communication system to discuss the system itself)
Caution: preceding defining characteristics of language apply only to adult, human language. children may lack:
interchangeability (don't speak what they hear),
displacement (typically talk about here and now, egocentric),
looks like humans are innately predisposed to learn a structure-dependent, creative language, and as we will see, if this capacity isn't exercised, it atrophies
Comparing chimp and child language acquisition
Chronology of studies
Long interest in subject, but first explored in 20th century
1931--Gua. reared by Prof. Kellor and his wife along with their son David. Treated like a baby. Gua could learn to understand over 70 words, never learned to produce anything. Gua showed that it's not just lack of opportunity that prevents chimps from learning language
1947---Viki. Reared by Hayes and Hayes. Extensive coaching, learned to say 4 words: papa, mama, cup, up
why these attempts failed: chimps not capable of uttering human sounds. can respond to spoken commands, but not acquire speech. (chimps in the wild are generally silent, communicating non-vocally. later experiments use sign language, token manipulation, pressing buttons.
1966--Washoe. Allen and Beatrice Gardner. acquired when about 1X months old. taught to use ASL, reared in 'natural' human language environment, surrounded by humans who communicated with her and each other using only signs--no rigorous training schedules, explicitly tried to develop conversational behavior.
other spinoffs--Lucy and other chimps (Roger Fouts): Washoe had adoptive son Loulis. Fouts studying transfer of signing from one generation to another.
1966--Sarah. Premacks. Reared in a cage, not home environment. No 'conversations'. All done by testing. Trained to manipulate plastic tokens. Rewarded with chocolate. Associated tokens with simple nouns. Complex concepts such as color of, same, different.
Project still continues, but focused on cognitive rather than linguistic abilities of animals.
1974--Nim Chimpsky--Terrace. ASL. Sterile lab environment, different teachers, long hours. Terrace concluded that most of NC’s signs were prompted by his teacher or imitative of his teacher.
1977--Lana, Sherman, Austin--chimps. Kanzi--pygmy chimp. Rumbaugh and Savage-Rumbaugh manipulated computer-generated lexigrams, geometric symbols. Kanzi learned as an observer when the Rumbaughs were teaching his mother. S-R and R found that reversability is not automatic: chimps had to be trained to recognize symbol as well as punch symbol. Lexigram system extended to work with profoundly retarded humans. Georgia State.
--Lucy and other chimps (Roger Routs).
1981--Koko, Michael (gorillas)--Patterson.
1981-86—Loulis, Washoe’s adoptive son. Washoe taught Loulis 50+ ASL signs. Also, Loulis imitated some of Washoe’s signs. Fouts studying transfer of signing from one generation to another. Fouts claims that the chimps are learning an infinite number of signs.
General problems with the studies
--inherent bias of researchers. chimps have very human qualities. researchers overemphasize human qualities. too much anecdotal rather than formal experimental evidence.
--Clever Hans phenomenon: experimenter may unconsciously reveal expected or hoped-for behavior through kinesio signals. Clever Hans story. Performance reduced to cuing or imitations. Oskar Pfungst exposed Clever Hans. showed impossible to avoid cuing. even if experimenter knows about danger of cuing.
In ape studies by Rumbaugh et al., chimps trained to produce proper hand signs when words spoken by humans. observed that when contextual information, speakers' gestures eliminated, chimps did little better than chance.
Double blind experimental design can eliminate possibility of cuing animals. (Film) Washoe repeatedly looked to human experimenter for confirmation, perhaps cuing. Reliable test for Washoe's vocabulary, but syntax. Persons recording data are unaware of what right answer should be. but more difficult to use than other designs. not as widely used. Studies of language development in children rarely use double-blind method. Much evidence concerning Koko the gorilla, such as fact that she has discussed abstract concepts like death, is anecdotal.
Summary of chimps' possible accomplishments
and possible reinterpretation of results of these studies.
--chimps can learn to associate referents (some quite complex, abstract) with arbitrary symbols--can do this with many symbols in double-blind experiments
but even this disputed: is this using language or are chimps merely learning behavior that will get them a reward (recall film sequence of Hayes and Hayes rewarding Vicki). some problems with interpreting it as language. Rumbaugh et al. suggest that learning meaning of a word required chimps to be trained in several subskills. Often chimp learns to punch the keyboard symbol for 'apple' when shown an apple, but when human punched symbol for apple, chimpanzee can't pick out apple from other fruits. suggests chimps have not associated symbol with referent.
--can learn to use language spontaneously. Washoe signing to herself--[quote from Gardner in Lieberman p. 239]. tooth-brush example. but no spontaneous signing of questions by Washoe. Terrace reported Nim Chimpsky spontaneously signalling 40% of time.
recently reported work of pygmy chimp Kanzi is most convincing. case of this (Rumbaugh and Savage-Rumbaugh) in NYT, Jan. 24, 1985. most chimps require long, arduous training period, but Kanzi said to be acquiring symbols spontaneously and comprehending spoken English. Even Terrace impressed. pygmy chimps thought to be smarter than ordinary chimps. more human in appearance (less of a snout). walks upright more than ordinary chimps. perhaps more evolutionarily advanced (cognitively? than other chimps) 'spontaneous' acquisition. Kanzi apparently learned to manipulate symbols while playing in lab as his mother was being trained on computer. Kanzi started to use language spontaneously around 1/2.
--can learn to use language creatively
new combinations of signs. Washoe's 'water bird' for 'swan' but Terrace: 'Washoe may have simply been identifying a body of water and a bird, in that order', but Fouts (quote from Liberman, p. 241):
Neologistic 'water bird' (for 'swan') not an isolated occurrence.
'Listen drink' (for chimpanzee Moja, referring to Alka seltzer in a glass)
'Metal hot' (for 'cigarette lighter')
'metal cup drink coffee' (for a thermos bottle)
'dirty good' (potty chair)
and Lucy (Fouts):
'smell fruits' (citrus fruits)
'cry fruit' or 'hurt fruit' (radish)
'drink fruit' or 'candy fruit' (watermelon)
Should be able to tell whether these are combinations or simply two single-word utterances from whether or not Washoe drops her hands in between signs
--totally new signs. Washoe's sign for 'bibs' (Ak.): Lucy invented a sign for a leash used to walk her along a busy highway.
--creative solutions to problems: Kanzi reported to use 2-3 word sentences often without prompting. quite creatively (at least from anecdotes) when friend Austin the chimp was moved out of compound, Kanzi got longely, typed 'Austin TV' to request videotape of Austin
--limted acquisition of syntax. hard to tell how much syntax acquired from 2-3 word utterances. existence of 2-sign utterances for Nim questioned by Terrace. often a short interval between signs---really 2-sign utterances. (but others have criticized Terrace for artificiality of his language environment for Nim. long hours, sterile environment, many different teachers---of course results worse for Nim.)
Washoe's word order never consistent. first acquired single words. possible explanations for inconsistent word order (1. importance of strict order might not have been emphasized by trainers, is not essential to communications; 2. harder to maintain strict order in signed language (deaf adults don't adhere rigidly to ordering)
Sarah--structure-dependence in signing, no spontaneity, creativity. But possible evidence of ability to carry out structure-dependent operations. Was able to carry out with difficulty 'Sarah insert apple red dish banana green dish', 'Sarah cracker candy yellow dish cracker blue pail insert'
'Sarah cracker candy yellow dish cracker blue pail insert'
but could argue that only shows Sarah knows what order to perform certain operations in in order to get reward. What would Sarah do with ‘blue cracker and pail’: would she recognize the ambiguity in this phrase? Did she ever produce these kinds of sentences?
conclude chimps capable of learning some aspects of language, but not predisposed to learn it like humans. chimps don't have latent capacity for human language. most predisposition towards language found in Kanzi. pygmy chimp Kanzi supposed to have learned how to comprehend oral language, as well as produce keyboard 'language'
Chimps vs. children
Is chimp language creative, rule governed, as children's language is?
stages similar: one-word---multi-word (but 2-3 words is end of line for chimps). Terrace: Nim's utterances don't steadily get longer. length consists of repetitions. children: from 2-word stage on, utterance length increases rapidly
chimps, like children, describe new objects in terms of descriptive properties or functional attributes of the new objects (cf. Washoe's new signs, like 'water bird' for 'swan')
overgeneralize referents [quote from Fouts in Lieberman, p. 242]
syntax not as complicated. trouble learning word order, structure-dependent operations. severely mentally handicapped children do better at learning language.
possible evidence that chimps' language is creative, spontaneous (signing to self) (but chimps don't spontaneously sign questions--cf. children continually asking 'what that?'), exhibits displacement possibly structure-dependence (Sarah)
chimps seen capable of acquiring language only at a level found in children. skills of Kanzi--most advanced of chimps reported--seen comparable only to those of a child 1-2 years old.
--is human language fundamentally different from animal communication? (continuity vs. discontinuity theories of linguistic evolution, other theories discussed in Ak. Ch. 8)
have seen similarities between human language and animal communications. are these accidental or systematic similarities. how did language develop?
discussion about origin of language is speculative. not possible to reconstruct exact form of first language (comparative method works only for families of time-depth up to 8000 years)--no fossil record (earliest writing much later) but can say something about how the ability to produce language has evolved.
modern humans clearly related to other primates, exp. gorillas and chimpanzees.
several evolutionary milestones which distinguish humans from other hominoids
--walking on two feet, leaving hands free for tool use
--increas in size and cognitive capacity of brain
--changes in shape of vocal tract
will discuss last two of these here. bipedal locomotion prob. didn't contribute much to language development.
1. increase in brain size
primates generally well equipped in terms of brain capacity wrt body size. within primates, brain-body ratios increases from prosimians to monkeys to hominoids
increase in brain size true of evolution in general
1. progressively greater brains found in more complex animals. amphibians to reptiles to mammals. increase in brain usu. reflect jumps in complexity of animals lives, accompanied by greater learning capacity. Biggest ratios found in marine mammals
brain-body size ratios:
2. typically, species get bigger over time, both brain and body. brain of earliest hominoids around 400 cc (as is chimps today) averages 1360 cc. today. some of increase attributable to increase in body size, but not all.
not just increase in brain size (dolphins have large brains too. important, but difference in overall organization of brain. increase in brain primarily in cortex, which controls voluntary activities, and higher functions like cognition, coordination. lobes involved with memory (temporal), sensory integration and association (parietal), and motor control and emotional behavior (frontal) got bigger. lobe involved with visual functions (occipital) got smaller; in apes, pattern is reversed (occipital got bigger)
why increase in brain size? one theory: to deal with increasingly complex world. thought that what is so complex and unpredictable about lives of humans is behavior of other humans. greater intelligence needed to deal with social life, not subsistence activities. note: all primates are intensely social.
note increase in social complexity---increase in intelligence---better communication. likely that appearance of even primitive language---perhaps beginning with the 1- or 2-word stage that children go through---would greatly increase survival value. so great selective pressure in favor of langauge.
2. changes in shape of vocal tract
one of 1st attempts to teach a spoken human language to an ape (Vicki) was unsuccessful. chimps can't produce range of sounds that humans can. vocal tract of adult humans different from that of other mammals, newborn humans.
fossil record, comparison with other related mammals indicates that adult human pattern is innovative. how and why are humans so different?
long observed law of biology is "ontogeny [development] recapitulates phylogeny [history of species]"---young of dif. species are more similar to each other than adults are.
newborn humans (up to 3 months, more similar to newborns of other primates in shape of suprlaryngeal vocal tract)
--larynx is higher in newborn humans, other primates: larynx and epiglottis (sits on top of larynx) acts as a tube from lungs to nose, permitting simultaneous swallowing, eating. larynx in adult humans is considerably lower.
--therefore less bend in airway (larynx to lips in newborns, other animals) more efficient for breathing--less airflow resistance, better for running
--also shape of mouth is longer in newborns, other animals---more efficient for chewing (greater tooth area)
--adult humans have thick muscular, mobile tongue: can vary size of mouth cavity. Cf. long thin tongues of monkeys
--adult human vocal tract more hazardous for swalling. larynx must be pushed out way by muscles of vocal tract
so why descent of larynx? adult humans have setup which is less efficient for chewing, breathing and more hazardous for swallowing. would seem to be evolutionary pressure against it
change in vocal tract possibly due to more erect posture
also possibly due to change i shape of jaws, no snout on humans. cf. apes: but tongue same shape and size. recessed jaws pushed tongue and larynx further back in neck.
note consequence of lowered larynx
dramatic dif. in shape of vocal tract. adult humans have a 2-tube tract. newborns, animals have a 1-tube vocal tract. more efficient for breathing. humans have to compensate for 2-tube inefficiency by breathing through mouth when increase in aerobic intake is necessary---other animals, newborns breathe only through nose. (??dogs panting)
but a single tube vocal tract can't produce basic vowel sounds of human language /i u a/
other evolutionary change in shape of vocal tract---restructuring of human basicranium
--leading to lowering of larynx
must have been high evolutionary advantage in having lowered larynx to compensate for increased danger of dying due to choking and less efficient breathing and chewing. changes in cognitive capacity of brain thought to permit high rate of speech vocalization. language of higher survival value to species than safe breathing.
Summary---are humans the only animals with language?
...the language faculty does appear to be a unique human possession. Other organisms have their own systems of communication, but these have properties radically different from human language, and human language is far more than a mere system of communication: Language is used for expression of thought, for establishing interpersonal relations with no particular concern for communication, for play, and for a variety of other human ends. In the past years there have been numerous efforts to teach other organisms (forexample, chimpanzees and gorillas) some of the rudiments of human language, but it is now widely recognized that these efforts have failed, a fact that will hardly surprise anyone who gives some thought to the matter. The language faculty confers enormous advantages on a species that possesses it. It is hardly likely that some species has this capacity but has never thought to use it until instructed by humans. That is about as likely as the discovery that on some remote island there is a species of bird that is perfectly capable of flight but has never thought to fly until instructed by humans in this skill. Although not a logical impossibility, this would be a biological miracle, and there is no reason to suppose that it has taken place. Rather, as we should have expected all along, the evidence suggests that the most rudimentary features of human language are far beyond the capacity of otherwise intelligent apes, just as the capacity to fly or the homing instinct of pigeons lie beyond the capacity of humans.
limited recoverability of language function if language-dominant hemisphere removed. infants with left vs. right hemisphere removed. later in life, infants with left hemisphere removed have trouble processing syntactically complex sentences
lowered larynx enhanced vocalization at expense of safe breathing. Kanzi's reported ability to understand, but not produce, spoken language suggests possibly that ability to comprehend evolved before ability to produce. cf. children---ontogeny recapitulates phylogeny.
Accomplishments of chimps vs. children
Chimps, with enormous teaching, don't get past early multi-word stage. questionable creativity, spontaneity.