On animal communication
Some cognitive abilities of animals
1. Introduction
Studies of trained animals in a laboratory environment:

Non-human primates



Washoe and other chimpanzees (Gardner & Gardner 1969; 1978)
Sarah and three other African-born chimpanzees (e.g., Premack 1976)
Kanzi, the bonobo (e.g., Savage-Rumbaugh & Lewin 1994)
 Koko, the lowland gorilla (Patterson 1978a; 1978b)
 Chantek, the orangutan (Miles 1978; 1983)
 Nim, the chimpanzee (e.g. Terrace 1983)

Non-primates




Squirrel monkeys (Burdyn & Thomas 1984)
Bottle-nosed dolphins (e.g., Herman 1987; 1989)
Californian sea lions (Schusterman & Krieger 1984; Schusterman &
Gisiner 1988)
Alex, the African Grey parrot (e.g., Pepperberg 1999b)
Communication systems:

Chimpanzee Washoe


American Sign Language (ASL)
Raising in an environment appropriate for a human child



Chimpanzee Sarah

Object-word relationship


living quarters with furniture, a kitchen, bedroom, bathroom, toys, tools, etc.
days made up of meals, naps, baths, play, schooling, and outings
pieces of plastic, backed with metal adhering to a magnetized slate, each
standing for some English word
Bonobo Kanzi

Yerkish, an artificial lexigram system

portable, folding posterboard containing printed lexigram symbols

Orangutan Chantek

“Pidgin” Sign English




Chimpanzee Nim


“Pidgin” sign language
Californian sea lions Rocky & Bucky


based on gestural signs of ASL
English word order
little grammatical morphology
Artificial gestural language
Bottle-nosed dolphins Phoenix & Ake

Artificial languages

acoustic language and "dolphinized" version of a gestural language
Studies of animals in their natural behavior and environment:

Vervet monkey species

Cheney and Seyfarth (1992)



Continuous period of 11 years
Kenya's Amboseli National Park
Zuberbühler (2002), Zuberbühler, Cheney & Seyfarth (1999)

Rainforest of the Ivory Coast
2. Animals’ linguistic abilities
Question:
Which traits in animal behavior may relate to languagerelated cognitive abilities?
2.1 Communicative intentions
Claim:
Non-human animal communication is restricted to
imperatives.
But what is the communicative intention e.g. of the leopard,
the eagle, and snake alarm calls of vervet monkeys?
Levels of intentionality
i.
Zero-order
i.
ii.
First-order
i.
ii.
iii.
No beliefs or desires at all
Beliefs and desires
No beliefs about beliefs
Second-, third-, or higher-order
i.
Some conception about one's own and other individuals’ states of mind
The communicative system of vervet monkeys (Cheney &
Seyfarth 1990; 1992)

Four phonologically distinct and arbitrary calls
→ representational signals standing for an (even invisible) object:
Wrr → given when neighboring group has first been spotted
 Snake call



Leopard call


reaction: stand bipedally and peer into surrounding grass
reaction: vervets on the ground run into trees
Eagle call

reaction: look up or run into bushes
Observations:

Vervet monkeys’ calls are

suggestive of first-order intentionality


reason: calling monkeys want others to run into trees, bushes, etc.
not necessarily suggestive of second-order intentionality

in that case they would want others to think that there is a leopard, eagle, or
python nearby
Most important motivation of animal communication is manipulative
rather than declarative!
Terrace (1985) calls this motivation acquisitive:
 Communicative behavior of animals gears primarily at expressing
requests


96 % of the lexigram utterances made by bonobo Kanzi
86 % of all utterances (65 % for objects and 18 % for actions) in Rivas’
(2005) sample on five chimpanzees
 Chimpanzee signing = mostly acquisitive in nature

Note: Motivations of apes cannot be reduced to being manipulative!


found signing when alone, e.g. looking at pictures in magazines
imaginative play using signing, or talk to themselves
 signing’s purpose unlikely to have been request-oriented!
But: Animal communication is not exclusively acquisitive!

Bonobo Kanzi


Chimpanzees






4 % of his utterances = indicatives or statements
4 % of their utterances = answers on questions by humans
2 % = to name or label objects and pictures
8 % = communicative intention not evident
largest categories of speech acts = action requests and naming
almost ¼ of chimpanzee’s communications = other kinds of sign acts
Gorilla Koko

comments about state of the environment via signing

E.g., signing of LISTEN QUIET when alarm clock stopped ringing next door
2.2 Concepts
Conceptual abilities of non-human animals

Wildborn squirrel monkeys

Claim: Squirrel monkeys have "natural concepts" and are capable of
using working memory with conceptual information


Laboratory pigeons (Herrnstein, Loveland & Cable 1976)

Reliable and discriminate responses to exemplars of water (pictures of
puddles, streams, lakes, etc.), trees, people, leaves, and fish


remembrance of conceptual information for at least 16 seconds in order to
make a conceptual choice
→ even when new pictures were presented
Wild Diana monkeys (Zuberbühler, Cheney & Seyfarth 1999)

Referential abilities: Females react with their eagle alarm call not only
on eagle shrieks but also on males’ eagle alarm calls
Obligatory abilities in order to understand a concept:
a) Understanding that different referents are instantiations of one and
the same entity.
b) Understanding that such instantiations include referents that are
outside the here and now of a given situation.
c) Use of learned entity in new contexts.
d) Reconstruction of the presence of a conceptual entity.
e) Production of novel instances of instances of that entity.
f) Relating of different conceptual entities to one another on the basis
of size, shape, color, etc.
Extension
Animals are able to “over-generalize”, that is, to extend the
use of a form-meaning pairing to referents beyond the ones
canonically associated with that unit.
→ conceptual reasoning

Orangutan Chantek

Use of the sign LYN not only for his caregiver Lyn Miles but for all
caregivers
 Use of the sign DIRTY to refer to bad things, until BAD was learned

Lowland gorilla Koko

Extension of the sign STRAW, learned initially with reference to drinking
straws, to plastic tubing, cigarettes, and a car radio antenna
 Extension of the sign NUT, learned initially as a name for packaged
nuts, to roasted soybeans, sunflower seeds, and pictures of nuts

Bottle-nosed dolphins

Generalization of the signal HOOP, taught initially as a name for a
particular, large, octagonal floating plastic hoop, to hoops of different
sizes, shapes, etc., as well as to sinking instead of floating hoops
 Extension of the sign WATER, taught initially to refer to a thin stream
flowing from an ordinary garden hose, to a waterfall when WATER was
used in a sentence for the first time at that new location
Same vs. Different
There is conceptual knowledge at least in some animals!

Squirrel monkeys:


Association of triangularity with choosing same and heptagonality with
choosing different
Dolphins and sea lions:

Response to novel combinations of attribute + object labels as well as to
novel combinations of actions + object labels (Pepperberg 1992)
Pepperberg’s Grey parrot Alex


Comparison of objects with one another on the basis of relational
concepts of 'same' and 'different'
Discrimination between objects on the basis of color, shape, and
material


E.g., when asked: “What color is (item designated by shape-X and
material-Y)?” → distinctions on known as well as on novel objects
Distinction between relative differences in objects’ size:



Transfer of size relationships to objects not involved in training
Transfer of his knowledge to items of novel colors, shapes, and sizes
Indication of situation of items not differing in size
2.3 “Lexicon“
Which types of form-meaning pairings can animals distinguish?

E.g., artificial languages used to train bottle-nosed dolphins: objects,
actions, and properties

Contra: human-chimpanzee interaction (Rivas 2005; Gardner &
Gardner 1978)


Acquisition of a large range of ASL unit types
Presentations of only the following classes of signs: objects, actions,
request markers (GIMME, HURRY), deictic sign THAT/THERE/YOU,
chimpanzee's own name sign
Numbers
Can animals represent numerosity?
a)
b)
c)
Can they identify a property of the stimulus that is
defined by the number of distinguishable elements it
contains?
Can they count?
Can they use numerical representations recursively?
a) Can animals identify a property of the stimulus that is defined
by the number of distinguishable elements it contains?
Discrimination of stimuli differing in number:
 See Brannon & Terrace (1998): e.g., pigeons, parrots, rats, dolphins,
monkeys, and chimpanzees
 See Koehler (1943; 1950), Pepperberg (1987a), Boysen & Berntson
(1989):

Ravens and jackdaws succeeded on numerical match-to-sample tests
on quantities up to 8
 Chimpanzee Sheba: demonstration of ordinality and labelling of the sum
of two arrays separated in time and space
 Grey parrot Alex: correct reply to question ‘How many?’, production of
vocal numerical labels for sets of 2 to 6 objects, abilities in handling
numerical quantities

E.g., when presented two pieces of cork or five pieces of wood →
responses: two cork and five wood
b) Can animals count?
c) Can animals use numerical representations recursively?
Neither do non-human animals show a concept of counting, nor do
they appear to have the capacity to create open-ended generative
systems.

No natural ability to discriminate numerosity


Rhesus monkeys (Brannon & Terrace 1998)



Utilization as a "last resort", if other bases for discrimination, such as
shape, color, size, frequency, or duration of a stimulus, are eliminated
Spontaneous representation of the numbers of novel visual stimuli
Extrapolation of an ordinal rule to novel numerosities
Grey parrot Alex: skills in handling numbers

No capability of counting or of a recursive understanding of numbers
2.4 Functional items
Location

Chimpanzees


Gorilla Koko


Learned red and dish, then understood command 'Insert the apple (in)
red dish'
Signs for prepositions on, out, and up
Grey parrots

Concepts of in versus on



to get an item that is in another object, one type of manipulation must be
used
to get an item that is on another object, manipulation is different
Bottle-nosed dolphin Phoenix

Ability to link an action term ('fetch‘) with a transport object, a
destination object, and spatial terms, being asked to carry a frisbee
through, over, or under a hoop
Deixis

Washoe

Distinction of personal deixis:




Gorilla Koko


Signs for 'me' and 'you'
Orangutan Chantek


sign for 'me' = by tapping one's own chest
sign for 'you' = by pointing away from the chest toward the addressee
"wild card sign" = frequently used pointing sign THAT/THERE/YOU
Distinction of 'I' and 'you'
Chimpanzee Sarah

Spatial deixis



comprehension of difference between ‘Sarah take this’ and ‘Sarah take that’
production of ‘Give Sarah this’ vs. ‘Give Sarah that’
production of demonstratives pronominally and attributively → ‘Give Sarah
this cookie’ vs. ‘Give Sarah that cookie’
Negation
Do animals have, or can they be trained to distinguish
negative concepts such as
(a) rejection,
(b) non-existence, and
(c) denial?
(a) Rejection
→ Most animal species that have been appropriately
trained know to handle the notion of rejection.
(Pepperberg 1999b)
(b) Non-existence

Washoe and others:


Other chimpanzees:


Comments upon the absence of a familiar object at a customary
location
Correct responses to questions about objects not present for sensory
reference
Grey parrot Alex:
Reaction to an object’s absence: nuh to refuse an object offered instead
of the requested
 Limited use of the concept of non-occurrence or absence

(c) Denial
Difficulties in understanding this concept:

Chimpanzee Sarah

E.g., "Red on green ?" (= 'Is red on green?')


referring to the relationship of two colored cards
Alteration of relationship on 30 % of the occasions to make it possible
for her to answer 'Yes' rather than 'No'
Questions

Chimpanzees


Gorilla Koko


Creation of a sign for polar questions by use of gestural intonation
Grey parrot Alex




Concept of polar (yes-no) and WH-questions
Comprehension of vocal questions
Extraction of relevant categories from word questions
Ability to produce questions by use of what
Bottle-nosed dolphin Ake


Polar interrogative form: sign for an object + question sign
Productive response to signal presence vs. absence of an object
2.5 Compositionality
Combination of form-meaning pairings in trained non-human primates:
→ Chimpanzee Washoe after 10 months of training
→ Chimpanzee Moja at the age of 6 months
→ Gorilla Koko after 4 months of sign language training
Questions:
Can animals combine form-meaning pairings with each pairing
retaining its meaning constant?
Do they understand that utterances can be broken up into concepts
and can be combined productively and in a principled way?
Are they able to use at least two paradigms of linguistic forms
productively in a sequence?

Alex

Recombination of beginnings and ends of existent labels


Kanzi, the bonobo

Perception of Tickle ball as consisting of two distinct units rather than of
one unanalyzed entity


suggested by his ability to re-combine these units
Bottle-nosed dolphins

Interpretation of PIPE TAIL-TOUCH as a command to perform an action
on an object


E.g., utterances such as "banacker" (banana-cracker)
that is, interpretation as an expression consisting of two discrete entities
Wild Diana monkeys

Combinatory rule suggestive of a concatenation of form-meaning
pairings and of a grammatical function:

Sequences of two calls are not productive since there are no other formmeaning pairings to which they could be applied.
2.6 Argument structure
Are animals able to form sentences that can be said to be
homologs or analogs of what we find in human language?
Do they have the ability to acquire an argument structure?
Chimpanzee Sarah
 Distinction between case functions ("subject" vs. "object")
 Comprehension of ‘name of’ when used as part of an ACC phrase


Note: previously, ‘name of’ always being confined to NOM phrase
Transfer of quantifiers from NOM phrase to ACC phrase

E.g., production of sentences such as 'Some cracker is round,' 'All
cracker is PL square,' etc., followed by correct performance when
instructed 'Sarah take some cracker,' etc.
Grey parrot Alex
 Use of structures [wanna go + location unit] and [want/wanna +
object unit] with a range of form-meaning categories
 Distinction of


volitional propositions (want X, wanna go X) and
commands (e.g., go X, you tickle me)
2.7 Linear arrangement
Are animals capable to develop a way of consistently and
productively ordering paradigms of form-meaning pairings?
In other words, do animals have something corresponding
to word order in human language?

Regular patternings

Chimpanzees Ally and Koko




Linear arrangement in bottle-nosed dolphins

Inference of thematic object roles (transport vs. designation) from
syntactic position


Ability to distinguish between PIPE FETCH HOOP 'Fetch the pipe to the
hoop' and HOOP FETCH PIPE 'Fetch the hoop to the pipe‘
Iconic ordering patterns


preference for order “DEM – N" in 92 % of two-sign constructions
preference for order “SUBJ – V – OBJ" in 89% of three-sign constructions
preference for order “ADJ – N” in 75 % of attributive phrases in two-sign
utterances
E.g., orders like tickle-bite and chase-hide
Ordering distinctions

Orangutan Chantek


object-GIVE: when object referred to was present
GIVE-object: when object referred to was absent
2.8 Coordination
NO subordination, BUT patterns of coordinating concatenation!

Chimpanzee Sarah

Acquisition of ability to conjoin nouns

Combination of two to three nouns having the same syntactic function in the
same sentence: objects of one and the same verb


E.g., "Mary give Sarah apple banana orange."
Bottle-nosed dolphin Phoenix

Instruction to act on a sequence of two propositions each consisting of
an object and an action, e.g.,


Instruction: PIPE TAIL-TOUCH PIPE OVER
Response: swimming to the pipe, touching it with her tail flukes, and jumping
over it
2.9 Taxonomic concepts
Do animals have, or can they acquire relational categories, in the
sense that they perceive and/or describe one concept in terms of
another concept?

Tomasello & Call (1997)

Primates → strong evidence in favor of taxonomic concepts:


A number of animals, including avian species, are able to both perceive and
produce predications on the basis of sameness and differences between
objects
Premack (1976)

Chimpanzees distinguish between first and second order relations:

Observation e.g. that the relation between ‘red’ and ‘red’ is same, as that
between ‘grey’ and ‘grey’ is same, and that the relation between ‘red’ and
‘red’ is the same as that between ‘grey’ and ‘grey’

the latter being a relation between relations (→ second order relation)
(1) Hierarchical taxonomic relations
a)
b)
c)
d)
e)
f)
Inclusion: A is a kind/type of B
(e.g., An apple tree is a kind/type of tree).
Property relationship: B has property A
(a red ball)
Partonomy (or meronymy): A is a part of B
(A finger is part of a hand)
Social relationship: A is a relative of B
(Anne’s father, husband, etc.)
Possession: A has B
(Anne’s car, name, etc.)
Location: B is located at A
(the book on the table)
(1a) Inclusion

Common domain: modifying compounding


no clear instances of it in non-human animals, nor of productive
compounding in general
Interpretation of reported combinations of form-meaningful pairings:



unitary, non-compositional meanings
combinations of free forms
Chimpanzee Washoe


described a swan by signing WATER BIRD
Savage-Rumbaugh et al. (1980)


Chimpanzees interpret symbols (lexigrams for specific foods or tools) in
order to then label them with lexigrams for the hypernyms ‘food’ or ‘tool’
Chimpanzees have ability to treat ‘food’ and ‘tool’ as representational labels,
and to expand the use of these labels to novel exemplars

Reason: training encouraging the appearance of functional symbolic
communication between chimpanzees
(1b) Property relationship

Gorilla Koko


Acquisition of distinct signs for 16 modifiers, including 'big', 'clean', 'cold', etc.
Trained African Grey parrots

Classification of objects on the basis of physical properties



Sentences of the target form "Sarah take red dish"
Question markers as in “Red ? apple” and “Yellow ? banana”
Orangutan Chantek


Extension of the word rock as a label to the property 'hard'
Chimpanzee Sarah



color (blue, green, grey, etc.), shape (2-corner, 3-corner, 4-corner, etc.), and material
(chain, hide, key, etc.)
Use of attributes, as in red bird and white cheese food eat
Sea lions & bottle-nosed dolphins

Modifiers for color (BLACK, WHITE, GREY), size (LARGE, SMALL), locations
(WATER, LAND), and locative modifiers LEFT, RIGHT, BOTTOM, and SURFACE
(1c) Partonomy
Claim: Apes are capable to understand the nature of part-whole
relationship.

Partonomy concept in chimpanzee Sarah



Ability to match pieces and names of fruit to intact fruit
Comprehension of “name of" when used as part of an “ACC” phrase
Partonomy concept in chimpanzees Peony & Elizabeth

ability to sort



plant parts (leaves, stems, seeds, and flowers) into a plant class
animal parts (fur, teeth, hair, and bones) into an animal class
Partonomy concept in spider monkeys (Tomasello & Call 1997)

categorization of fruit trees in their environment on the basis of the
particular type of fruit they bore
(1d) Social relationship
Observation: A wide variety of non-human primates have access to rich
knowledge of who is related to whom, as well as who is dominant and
who is subordinate.
In fact, primate understanding of relational categories is said to have
evolved first in the social domain to comprehend third-party social
relationships.
Kinship relationship in non-primate species:

Captive bat mothers (Dechmann 2005)


Chimpanzees (Hare et al. 2000)



Recognition of pups, although pups do not recognize their mothers
Understanding of subordinate-dominant relations
Ability to play different roles
Different primate species (Tomasello & Call 1994; Tomasello 2000a)


Understanding of third-party social relationships
Understanding and forming of categories of third party social
relationships
 Redirected aggression:

A1 (or A's kin) is attacked by B, A retaliates by attacking B's kin
(1e) Possession
Fitch, Hauser & Chomsky (in press):
 animal “ownership” is influenced by dominance, priority of access,
value of resource, and species-specific rules and exceptions

“ownership” concept in some animals with overlap with our own
No indication of a taxonomic relation possessor-possessee in any of
the animals that have been studied!
→ e.g., Gorilla Koko: production of combinations such as KOKO
PURSE or HAT MINE
(1f) Location
Bottle-nosed dolphins Ake and Phoenix

Understanding of some kind of locative modifier-object construction
(Note: certainly not sufficient evidence)

Relational sentence: transport word or destination word + modifier



E.g., PERSON LEFT FRISBEE FETCH (for Ake) and FRISBEE FETCH BOTTOM
HOOP (for Phoenix)
Arrangement patterns of the form [O1 + O2 + R]
Use of LEFT and RIGHT as locative modifiers (M) before object term,
hence [M + O]



O1 and O2 = object terms
R = relational action term
Exposure (without any training) to expanded structures [M + O1 + O2 + R]
and [O1 + M + O2 + R]
Conclusion
Taxonomic relations are a requirement for there to be noun
modification and also for noun phrase recursion.
The animals concerned are said to
 both comprehend and produce conglomerations of features and
assign them to objects



‘red’ and ‘banana’ in the case of the chimpanzee Sarah
rose and paper in the case of the Grey parrot Alex
have abstract concepts such as ‘red-ness’ or ‘banana-ness’
3. Language-like abilities in animals
(2) Language-like abilities of some trained non-human animals
a)
b)
c)
d)
e)
f)
g)
h)
i)
to understand salient characteristics of concepts
to distinguish form-meaning pairings ("words")
to acquire form-meaning pairings of more than one hundred items,
including items denoting objects, actions, and some numbers
to handle functional items for negation and interrogation
to have an elementary understanding of the notion of deixis
to use an elementary argument structure
to acquire some understanding of linear arrangement of formmeaning pairings
to conjoin propositions and/or form-meaning pairings
to acquire some basics of taxonomic hierarchy as it manifests itself
in inclusion and part-whole relations
4. Conclusion



The mechanisms underlying human speech perception were largely
in place before language evolved.
Several of the core properties of human syntax lie within the grasp
of other animals.
Structural features of human languages concern primarily apes’
cognitive abilities rather than their communicative abilities.

The two are not entirely mutually incompatible since there is clearly an
overlap area:

These abilities correspond at least to one layer of human language evolution