Chapter 3: The Foundations of Language
Where did language come from?
The “ding-dong,” “heave-ho,” or “bow-wow” theory explains that language came from imitation since
many words are representative of the sound they make (cuckoo for birds).
Most obvious idea = evolved as beneficial adaptation shaped by natural selection; it is still a controversial
Alternative = language arose as a side-effect of the evolution of something else, that was much more
important, such as the increase in size of the brain and general intelligence.
Arguments in favour of this theory:
• Not enough time for something as complex as language to evolve since the evolution of humans
diverged from that of other primates
• Grammar cannot exist in any intermediate form
• Possession of a complex grammar has no obvious selective advantage, evolution could not have
selected for it
Recent years, hypothesis that language evolved by Darwinian natural selection as an advantageous
adaptation has largely won, because it provides a well-understood general mechanism for how language
could have arisen (natural selection) and because the objections do not hold much water.
Now, it is apparent that
• there was sufficient time for grammar to evolve,
• it evolved to communicate existing cognitive representations
• the ability to communicate using a grammar-based system is a big evolutionary advantage (you
want to tell the other person that this area is inhabited by a lot of enemies)
The capacity for language and symbol manipulation must have arisen
• as the human brain increased in size
• When Homo sapiens became differentiated from other species between 2 million and 300,00
A structure corresponding to Broca’s area was present in the brains of early hominids 2 million years ago.
• Shape of human skull changed over time, enabling better control of speech
• Articulatory apparatus has not changed significantly over the last 60, 000 years
Early humans that communicated through grunts could not have directly evolved to have a complex
language that uses grammar idea of a protolanguage (Bickerton, 1990, 2003) emerged.
• Arose with evolution of Homo erectus about 1.6 million years ago
• Has vocal labels attached to concepts
• Does not have a proper syntax
• Primates taught sign language, very young children, children deprived of early linguistic input,
speakers of pidgin language could be said to use protolanguage.
What pressures selected for language? Perhaps the social set-up. Rich communication is a huge
• Associated with important aspects of language, especially grammar
• In animals = coordinating sensory and motor information, and skilled complex movements
• Damage in humans leads to difficulty in acquiring language normally • current structure of the gene arose through a mutation within the last 100,000 years, which lead to
greater development of Broca’s region and an enhanced ability to coordinate complex sequences
o Corballis flowering of human culture, art, and technology, and the expansion of Homo
sapiens about 40,000 years ago were due to this mutation and development of language
o Mutation = speech became fully autonomous, no longer relied on gestures; enabled better
Current controversy = extent to which the evolution of language depended on the hands, and whether
grammar arose from the use of manual gestures.
Paget (1930) = language evolved in intimate connection with the use of hand gestures so that gestures
developed to expand the available repertoire
Corballis (1992, 2003, 2004) = evolution of language freed the hands from having to make gestures to
communicate, so that tools could be made and used simultaneously with communication
• Argues that language arose from primate gestures
• Additional information = imaging studies show that the brains of great apes are specialized in a
very similar way to humans
o Chimpanzees and gorillas, like humans, show an asymmetry between the left and right
hemispheres of the brain with Brodmann’s area 44* being particularly enlarged on the
left. Area involved in gestures.
o *Brodmann’s area 44 = Broca’s area in humans, involved in producing speech
o One explanation, brain of great apes became specialized to enable sophisticated gestures,
human brains kept specializing with speech arising from gestures.
Mirror neurons play a particular role in the evolution of language, with manual gestures rather than vocal
communication driving evolution:
• Mirror neuron system for grasping imitation allowed early manual signs to develop
As gesture-based language evolved, vocalizations became incorporated into the gesture system, leading to
specialization and lateralization of the language and gesture systems and the right-handed preference in
Elman (1999) language arose from a communication system through many interacting “tweaks and
Deacon (1997) language and brain co-evolved in an interactive way; frontal cortex of humans grew
larger symbolic processing became more important linguistic skills became necessary to manage
symbolic processing development of speech apparatus to implement these skills demand and enable
further symbolic processing abilities.
Fisher and Marcus (2006) language was a complex reconfiguration of several systems that became
adapted to form language.
Do animals have language?
Theme = both animal communication systems and attempts to teach human-like language to animals,
particularly chimpanzees. Why?
1. Provides focus for the issue of what we mean by the term language
2. Informs the debate about the extent to which aspects of language might be innate in humans and
have a genetic basis
3. Might tell about which other social and cognitive processes are necessary for a language to
develop 4. Great intellectual interest
Animal communication systems
Many animals possess rich communication systems.
Communication is much easier to define than language; it is a transmission of a signal that conveys
information. The signal is the means that conveys information (sound, or smell). Communicative signals
have an element of intentionality in them whereas signals that are informative do not (eg, a cough informs
the person has a cold but it is not a communication. The person telling you they have a cold is
• Ants = pheromones (chemical messengers)
• Honey bees = complex waggle dance
• Primates = visual, auditory, tactile and olfactory signals. Use a wide variety of calls to symbolize
a range of features of the environment and their emotional states. They communicate about
stimuli for which they do not already possess signals communicative system has an element of
No evidence suggests that dolphins (nor whales) employ sequences of sub-units that convey particular
messages, in the same way as we combine words to form sentences to convey messages.
• Hard to define
• Dictionary = human speech; artificial system of signs and symbols with rules.
To some extent, aim of modern theoretical linguistics is to offer an answer to this question.
Hockett (1960) listed 16 general properties or design features of spoken human language, with emphasis
on physical characteristics of spoken language. However not all are necessary defining characteristics:
written language does not display “rapid fading” yet is a form of language.
It is a useful framework for thinking about how animal communication systems differ from human
laHockett’s 16 design features (box 3.1, page 56)
1. Vocal-auditory channel (producer speaking, receiver hearing)
2. Broadcast transmission & directional reception (signal travels out in all directions, can be localized
in space by hearer)
3. Rapid fading (signal rapidly disappears)
4. Interchangeability (can be both receivers and transmitters)
5. Complete feedback (access to everything about their productions)
6. Specialization (amount of energy in signal is unimportant, word means the same whether shouted or
7. Semanticity (signals mean something)
8. Arbitrariness (symbols are abstract)
9. Discreteness (vocabulary made of discrete units)
10. Displacement (refer to things remote in time and space)
11. Openness (invent new messages)
12. Tradition (can be taught and learned)
13. Duality of patterning (combinations of otherwise meaningless units are meaningful, applying at both
level of sounds and words, and words and sentences)
14. Prevarication (ability to lie and deceive)
15. Reflectiveness (communicate about the communication system)
16. Learnability (speaker of one language can learn another) All communication systems possess some of the features. Some of the characteristics are more important
• Duality of patterning
These all relate to the fact that language is about meaning. We can add other features that emphasize the
creativity and meaning-related aspects of language.
Marshall (1970) language is under our voluntary control. Creativity of language stems from our ability
to use syntactic rules to generate a potentially infinite number of messages from a finite number of words
using iteration and recursion.
Syntax has 5 important properties:
1. Language is a discrete combinatorial system; meaning of words do not blend into each other,
keep their identity
2. Well-ordered sentences depend on ordering syntactic categories of words in correct sequences
3. Sentences are built round verbs
4. We can distinguish words that do the semantic work of the language (content words) from words
that assist in the syntactic work of the language (function words).
5. Recursion enables us to construct an infinite number of sentences from a finite number of rules.
No animal communication system has these properties.
All nonhuman communication systems are different from language; many animals possess rich symbolic
communication systems that enable them to convey messages to other members of the species, that affect
their behaviour, that serve an extremely useful purpose, and that possess many of Hockett’s design
features. They lack the richness of human language.
Can we teach language to animals?
Some animals have the biological and cognitive apparatus to acquire language but have not needed to do
so in their evolutionary niche VS other animals are incapable of learning languages.
• Knew labels of over 200 items
• Would fetch even when he could not see the owner
• When faced with a new name, would infer that the name applied to a novel object
Suggests that general learning mechanisms might go some way to explaining early word learning
Pepperberg (1981, 1983, 1987) African grey parrot Alex:
• 80 words, object names, adjectives and some verbs
• Produce and understand short sequences of words
• Classify 40 objects according to colour and what they were made of
• Understand concepts of same and different
• Count up to 6 Herman, Richards, and Wolz (1984) trained two bottle-nosed dolphins, Phoenix and Akeakamai,
artificial languages. One used gestures of the trainer’s arms, the other acoustically based using computer-
generated sounds transmitted underwater. However, they only tested comprehension of language, not
production. They showed no evidence of being able to use function words.
Before we can conclude that apes have learned language, need to show that they have mastered both the
ability to associate a finite number of words with meanings or concepts and to use rules to combine these
words into an infinite number of sentences.
What are the other cognitive abilities of chimpanzees?
Cognitive abilities of chimpanzee named Viki aged 3 ½ years were generally compatible to those of a
child of a similar age on a range of perceptual tasks, such as discriminating and matching similar items;
not on tasks involving counting. Another chimp, Sarah, performed at levels close to that of a young child
on tasks such as conserving quantity as long as she could see the transformation occurring.
1. Suggests that for many basic cognitive tasks language is not essential
2. Suggests that there are some non-cognitive prerequisites to linguistic development
3. Suggests that cognitive limitations in themselves might not be able to account for the failure of
apes to acquire language.
Talking chimps: Gua and Viki
Kellogg and Kellogg (1933) raised female chimpanzee named Gua (type of rearing called cross-
fostering or cross-nurturing); only understood a few words, never produced any that were recognizable.
Hayes (1951) raised Viki as a human child, attempted to teach her to speak. After 6 years, could
produce four poorly articulated words using her lips.
The vocal tracts of chimps are physiologically unsuited to producing speech, this alone could account for
lack of progress. We cannot conclude anything about the general language abilities of primates from these
Later attempts to teach apes languages involved manipulating artificially created symbols or sign
• Female chimpanzee caught when she was 1 year old
• Taught American Sign Language (ASL) has words and syntax
• Age 4 = could produce about 85 signs, comprehend more
• Few years later, vocabulary increased to about 150-200 signs
• Signs were nouns, verbs, adjectives, negatives and pronouns
• When she did not know a sign, she could create a new one water bird for duck
• Combined signs and used them correctly in strings up to five items long
• Could answer some questions that use WH-words (what, where, why, who)
• Washoe’s adopted son Louis both spontaneously acquired signs form Washoe and was seen to be
taught by Washoe cultural transmission
• Unclear whether it is a language that has been transmitted or just a sophisticated communication
system • At first sight, Washoe appears to have acquired use of words and meanings and some sensitivity
to word order in both production and comprehension.
• Chimpanzee trained in a laboratory setting to manipulate small plastic symbols that varied in
shape, size and texture
• Symbols could be ordered in certain ways according to rules formed language called
• Less memory load is required in this case
• Produced mainly simple lexical concepts
• Could produce novel strings of symbols, usually at the level of substituting of one word for
• Produced sentences that were syntactically quite complex (If… then)
• Showed metalinguistic awareness (reflectiveness) in that she could talk about the language
• Little evidence that Sarah was grouping strings of symbols together to form proper syntactic units
Nim and others
• Chimpanzee taught language based on ASL
• Learned about 125 signs
• Researchers recorded over 20,000 utterances
• Regularity of order in two-word utterances, usually a place was the second thing mentioned;
longer utterances were characterized by many repetitions.
• Rarely signed spontaneously
• 90% of his utterances were in reply to trainers and concerned immediate activities like eating,
drinking and playing
• 40% of utterances were simply repetitions of signs
• O’Sullivan and Yeager said that the type of training limited Nim’s linguistic abilities; he
performed better in a conversational setting than formal training session.
Other attempts: Savage-Rumbaugh, Rumbaugh, and Boysen (1978) taught chimpanzees Lana,
Sherman, and Austin language using a computer-controlled display of symbols structured according to an
invented syntax called Yerkish (symbols that serve as words = lexigrams).
Evaluation of early attempts to teach language to apes
At first, attempts look convincing:
• Important design features of Hockett appear to be present
• Specific signs are used to represent particular words (discreteness)
• Apes can refer to objects that are not in view (displacement)
• Issue of semanticity = controversy
• However, learned associations between objects and events and responses
• Sarah could discuss the symbol system itself (reflectiveness)
• Signs could be combined in novel ways (openness)
• Reports of apes passing sign language on to their young satisfy feature of tradition • Signs are combined according to specified syntactic rules of ordering; apparently acquired a
Many problems with some of this research; 2 sources of debate: methodological criticisms of the training
methods and the testing procedures, and argument over how the results should be interpreted.
Methodological criticisms ASL not truly symbolic, many of the signs are icons standing for what is
represented in a non-arbitrary way; Premack’s plastic symbols do not use ASL, also some ASL signs are
iconic but many are not, deaf people use ASL in a symbolic way.
ASL is different from spoken language = condensed this might affect the way in which animals use the
In Washoe’s case, great proportion of her signing seemed to be based on signs that resemble natural
gestures. Also possible her trainers over-interpreted her gestures. Deaf native signers observed marked
discrepancy between what they thought Washoe had produced and what the trainers claimed.
Furthermore, Sarah’s performance deteriorated with a different trainer.
Problems arise over the fact that no one produced an exact corpus of all the signs a signing ape made and
in what context. Also, the data presented are reduced so that repetitions are eliminated. Imitations of
humans’ preceding signs abound, whereas genuinely creative signing is rare. Thompson and Church
(1980) produced computer program simulating Lana’s acquisition of Yerkish. She learned to associate
objects and events with lexigrams and used one of a few stock sentences depending on situational cues.
No evidence of real understanding of word meaning or syntactic structure.
Differences between apes’ and children’s language behaviour:
Utterances are mainly in the here-and now Utterances can involve temporal
Lack of syntactic structure displacement
Little comprehension of syntactic Clear syntactic structure and consistency
relationships between units Ability to pick up syntactic relationships
Need explicit training to use symbols between units
Cannot reject ill-formed sentences Do not need explicit training to use symbols
Rarely ask questions Can reject ill-formed sentences
No spontaneous referential use of symbols Frequently ask questions (to find out more
Spontaneous referential use of symbols
• Pygmy chimpanzee
• Vital step in spontaneously acquiring the understanding that symbols refer to things in the world,
behaving like a child
• First acquired symbols by observing his mother being trained on the Yerkish system of lexigrams
• Interacted with people in normal daily activities and was exposed to English
• Comprehend English as well as Yerkish was studied + compared with ability of young children;
performed as well as or better on number of measures than a 2-year-old child.
• 30 months = learned at least 7 symbols
• 46 months = 50 symbols and 800 combinations
• Sensitive to word order and understood verb meaning
• Spontaneous utterances formed more than 80% of output
• Why so successful? Perhaps because of early exposure Critics:
Seidenberg and Petitto (1987) Kanzi understands names in a different way from humans; Kanzi’s
acquisition of apparent grammatical skills was much slower than that of humans and his sentences did not
approach the complexity displayed by a 3-year-old child.
Savage-Rumbaugh (1987) & Nelson (1987) critics underestimated abilities of chimpanzees and
overestimated the appropriate linguistic abilities of very young children.
Kako (1999a) does not appear to be able to use morphology; does not modify his language according
to number; no clear evidence Kanzi uses recursive grammatical structures.
Evaluation of work on teaching apes language
Rivas (2005) chimpanzees used mainly signs for actions and objects; little evidence of either syntactic
or semantic structure in their signing. Concluded signing of apes showed many differences from early
language of children.
Pigeons can be conditioned to peck differently when two different words are presented; does that mean
that they learned what the word means just like humans? Do they know how a certain word (tree) is
conceptually related to other words? No. Pigeons can be conditioned to say that the word ‘tree’ looks
more like ‘tee’ than ‘horse.’
Is the use of signs by chimpanzees more like that of pigeons or of humans? 2 questions need to be
1. Can apes spontaneously learn that names are constant across contexts? A strawberry remains a
strawberry whether it is in your bowl or attached to a strawberry plant.
2. Do these primates have the same understanding of word meaning as we do? No unequivocal
answers to these questions.
It is debatable whether the chimpanzees have learned the meaning of the symbols in the way that we
know the meanings of words. They can sometimes learn effectively; Kanzi and Panbanisha (another
bonobo chimpanzee) could learn new words naming objects very quickly, with only a few exposures to it,
sometimes learn by observation.
Only humans can use recursion – where phrases can include phrases of the same type.
Has it been demonstrated that apes can combine symbols in a rule-governed way to form sentences?
sentences are simply generated by frames; nothing more than a sophisticated version of conditioning,
does not show the creative use of word-ordering rules.
Monkeys can learn very simple grammar; cannot learn more sophisticated, human-like grammars that use
hierarchical structures where there are long-distance dependencies between words (if… then).
Cotton-top tamarins learn which sequences of sounds tend to occur often together can discriminate
words from nonwords. They are unable to learn more sophisticated artificial grammars that use
Hauser & Chomsky (2005) recursion is the only uniquely human component of language; an
immensely powerful one. Pinker and Jackendoff many more aspects of language: properties ofw rods and grammar, and the
anatomy and control of the vocal tract, unique to humans. FOXP2 gene is unique to humans and is
involved in the control of speech and language.
Lastly, the Piraha language of the Amazon does not seem to use any recursion but it is a human language.
Why is the issue so important?
1. Led to a deeper insight into the nature of language and what is important about it
2. Worth noting that although the cognitive abilities of young children and chimpanzees are not very
different, their linguistic abilities are. language processes are to some degree independent of
other cognitive processes.
3. Language is species-specific and has an innate basis; only humans possess a language acquisition
device (LAD) that enables us to acquire language.
The biological basis of language
Are language functions localized?
Parts of the brain are specialized for specific tasks. Brain-imaging techniques indicate which parts of the
brain are active when we do tasks such as reading or speaking (and anything else for that matter).
2 hemispheres of the brain are in part specialized for different tasks:
• In most right-handed people:
o left hemisphere = analytic, time-based processing
o right hemi = holistic spatially based processing
o 96% language functions are predominantly localized in left hemi dominant
• Rasmussen and Milner (1977) even 70% of left-handed people are left-hemi dominant
• Just the same left-hemi brain regions are activated in people producing sign language with both
Early work on the localization of language
Most of the evidence on the localization of language comes from studies of patients with brain damage.
Aphasia = impairment in language production or comprehension as a result of brain damage
Broca’s aphasia = Damage to cortex of left frontal lobes resulted in impairment in ability to speak,
despite vocal apparatus being intact and the ability to understand language unaffected
Part of the brain responsible for speech production = Broca’s area.
Wernicke’s aphasia: localized in temporal gyrus; characterized by fluent language that makes no sense
and inability to understand language. Hearing is unaffected.
Wernicke’s area plays a central role in language comprehension.
The Wernicke-Greschwind model
Wernicke’s model: sound images of object names stored in Wernicke’s area of the left upper temporal
cortex of the brain. When we speak, sent along arcuate fasciculus to Broca’s area in left lower frontal
cortex where sound images are translated into movements for controlling speech
Wernicke-Greschwind model = elaboration of Wernicke’s Language flows from areas at the back to the front of the left hemisphere.
• When we hear a word, information is transmitted from the part of the cortex responsible for
processing auditory information to Wernicke’s area
• If we speak that word, information goes to Broca’s area & passed on to motor area responsible
• If word is to be spelled out, auditory pattern is transmitted to the angular gyrus. If we read a
word, visual area activates angular gyrus then Wernicke’s area.
Damage to arcuate fasciculus is a disconnection syndrome; it consists of difficulties in repeating
language, while comprehension and production are unimpaired.
The angular gyrus plays a central role in mediating between visual and auditory language.
The model is too simple:
1. Language functions are not restricted to the left hemisphere; some language functions are
processed in the right hemisphere: prosody, deep dyslexia
a. Eg., Right cerebellum becomes significantly activated when we process the meaning of
2. Even in left hemisphere, other areas than Wernicke’s and Broca’s areas are important entire
superior temporal gyrus
3. Brain damage does not have those clear-cut effects. Complete destruction of areas rarely results
in permanent aphasias.
4. Almost all with aphasia have anomia (difficulty in finding the names of things) regardless of
5. Selective stimulation of Broca’s and Wernicke’s area does not produce the simple, different
effects that we might expect.
Recent models of how language is related to the brain
Ullman (2004) D/P (declarative/procedural) model of how language relates to the brain. Language
depends on two brain systems:
a. Mental dictionary, lexicon, depends on declarative memory in left temporal lobe.
b. Mental grammar depends on procedural memory based on neural system involving frontal lobes,
basal ganglia, cerebellum, and regions of the left parietal lobe.
Distinction is one between linguistic rules (syntax) and words.
Some portions of the brain are more important for language functions than others; there are multiple
routes in the brain.
Hickok and Poeppel (2004) focus on auditory comprehension. Early speech perception involves the
superior temporal gyrus bilaterally (more on the left); then splits into two streams
a. Dorsal (towards back and top of brain), concerned with mapping sound onto a representation
involving articulation and relates speech perception to speech production, output – integration of
auditory and motor information
b. Ventral (towards front and bottom of brain), concerned with turning sound into meaning.
Sex differences and language
It is difficult to establish the direction of causality for findings, especially since some differences can be
the result of cultural causes. Yet, girls are better on some verbal tasks, have better verbal memories and
are better readers and spellers. Males Females
Greater lateralization Relatively less aphasia after damage to left
Greater right-ear left-hemisphere advantage for hemisphere, they recover faster
perceiving speech sounds During language task, girls show activation in both
During language task, boys show activation only in left & right pre-frontal cortex
left pre-frontal cortex Significantly more likely to be interrupted
Significantly more likely to interrupt Fluent, producing more words, longer sentences,
More likely to suffer from clinical disorders like fewer errors in a given time
Is there a critical period for language development?
Lenneberg (1967) Critical period hypothesis:
1. Certain biological events related to language development can only happen in an early critical
period, particularly hemispheric specialization
2. Certain linguistic events must happen to the child during this period for development to proceed
normally; for instance songbirds need to listen to the song within 15 days of their birth to acquire
the song normally.
Evidence from the development of lateralization
A considerable amount of development takes place after birth, throughout childhood and possibly
adolescence maturation. The cortex shows some plasticity: after damage it can somewhat recover and
Lateralization emerges throughout childhood; damage to left hemisphere does not lead to permanent
disruption of language abilities. ol to language), either can specialize; critical period hypothesis is best-
known version of this hypothesis
1. Irreversible determinism (invariance) hypothesis: left hemisphere is specialized for language at
birth; the right hemi only takes over if something happens to the left.
2. Emergentist account: the two hemi are characterized at birth by innate biases in types of
information processing that are not specific to language processing.
Lenneberg argued lateralization occurs as a result of maturation during the ages of 2 and 5 years, slows
down, complete by puberty.
The entire function of left hemisphere can be taken over by the right if the child is young enough
hemidecortication = an entire hemisphere is removed (for instance, in the case of exceptionally severe
epilepsy; young child will almost completely recover.)
Dennis & Whitaker (1976) children who had had the entire left cortex removed had particular
difficulties in understanding complex syntax, compared with children who had had the right cortex
removed. However, critics of the study mention that IQ should have been matched for.
Evidence from studies of lateralization in very young children
Entus (1977) used the sucking habituation paradigm to examine whether 3-week-old infants respond
differently to sound stimuli. Sucking rate increased when speech stimuli were presented to the right ear
(left hemi advantage) and when non-speech stimuli where presented to the left ear (right hemi advantage).
Molfese (1977) measured evoked potentials, found hemispheric differences to speech and non-speech
stimuli in 1-week-old infants, with left hemi preferring speech.
Mills, Coffrey-Corina and Neville (1993) data suggest that right hemi plays an important role in early
language acquisition. Unknown words elicit electrical activity across right hemisphere. Infants who show early left-hemi processing of phonological stimuli show better language abilities
several years later.
The data also suggest that the left hemi has an affinity for language, contradicting the view that the two
hemi are equipotential.
Evidence from second language acquisition
Johnson & Newport (1989) 2 hypotheses about language learning
Maturational State Hypothesis Exercise Hypothesis
Capacity disappears or declines as Unless the capacity is exercised, it is lost
maturation progresses As long as the 1 language is exercised
Children are better at acquiring the second during childhood, the ability to acquire
language others remains intact