What is language? It’s easy to identify language… … but not so easy to define it What is special about human language? How does it differ from forms of.

What is language?

• It’s easy to identify language…

• … but not so easy to define it

• What is special about human language?

• How does it differ from forms of communication used by other animals?

Defining human language

• Human language can be defined by:– Symbolic reference: abstract relation of

meaning to symbol– Grammar: use of symbols in combinations

governed by a syntactic ordering system

Grammar

• What is grammar?– The meaning of a signal doesn’t just depend

on the symbols in the signal– It also depends on how the symbols are put

together

• There is no evidence that any form of animal communication has a grammar

Big Brain only?

• Language an advanced form of animal communication?

• Is the difference a matter of brain power?• Counterarguments:

– If this was the case, we’d expect to see simple languages in other animals.

– We don’t.– If this was the case, the complexity of animal

communication would be dependent on intelligence– It isn’t.

Animal communication systems

• A huge range of systems:– Vervet monkeys– Whales– Bees– Ants

Vervet monkey: Studies of Seyfarth and Cheney

Historical Theoretical Positions

Two different views of language development in humans

Skinner: behavioral psychologistChomsky: theoretical psychologist

Skinner: proposed that language, like all behaviors was an operant behavior that is formed by reinforcement and shaping. Infants learn a language like rats learn to get a reward by pressing a bar.

Chomsky: proposed a mechanism that included innately specified constraints on the forms that language could take. Infants had innate constraints on language that included specification of a universal grammar and universal phonetics.

Three critical components of language acquisition:a. initial state of knowledge

Skinner: no innate info necessaryChomsky: innate knowledge was a core

componentb. mechanisms responsible for developmental

changeSkinner: rewards guide changeChomsky:growth or maturation of the

language modulec. the role played by ambient language input

Skinner: language input did not cause language to emerge

Chomsky: language input triggers a particular pattern from those innately stored. 

Components of language acquisition

New emerging view of language acquisition does not support either an operant conditioning model, or a selectionist model. The new model suggests that infants engage in a new form of learning, in which language is mapped in detail by the infants brain.

New view of language acquisition

Parsing the auditory world in to language units

What is the problem that an infant must solve?

Infants must learn how to parse the auditory world, by learningto perceive the phonetic units of speech from the formant frequencies and learn to parse words from running speech.

Evidence: Neural Homologies • Vocal calls in apes are much less under voluntary

control than their hand movements.

• Broca’s area most concerned with production of speech

• Wernicke’s area most concerned with speech comprehension

• Broca’s area is NOT homologous with cingulate cortex (controlling calls) but with areas in pre-motor cortex for manual control, action recognition, and imitation: possibly a ”mirror neuron system”

Tomasello’s Theory of Language Evolution and Cultural Influence

• The ability to use symbols evolved

• These symbols were used in “discourse” patterns

• Over historical time, the “discourse” patterns were grammaticalised into syntax

Symbols are a product of biological evolution, grammar is a product of cultural evolution

How to account for language evolution

• Modern human language is possible because of modern human genes.

• So one is tempted to explain the evolution of those genes.

• But what selects for language-learning genes.

• People already speaking!• But then…language should not evolve—it’s a

Catch 22…

But it isn’t

• Not if you imagine language emerging little by little.

• Somebody invents the first, most primitive, linguistic innovation.

• Others copy it.

• Once it is stable it acts as a selection pressure on genes that will make learning this easier and faster.

• And then on to the next innovation, and so forth.

But what exactly do we have to explain

• Many who ponder the evolution of language spend a lot of time thinking about the articulatory and auditory mechanisms.

• But we need to explain what happened first.

• And that can’t be it because our articulatory mechanisms will not be selected for unless people are already speaking.

But what exactly do we have to explain?

• Humans have a dramatically lowered larynx.– puts the base of the tongue in the throat

cavity– allows movement of the tongue to modify

simultaneously the shape of the throat and mouth

• hence: exquisite vowel and pitch control

• But it also causes us to choke to death!

Descent of larynx and lengthening of laryngopharynx

• Adult human larynx is lower in the throat than that of non-human primates and human infants

chimpanzee humanchimpanzee human

A classic theory based on these observations

• Lieberman and Crelin (1971; Lieberman 1984) reconstructed Neanderthal vocal tract

• Proposed: Neanderthals had a higher larynx than modern humans, which would not have permitted them to articulate the point vowels

• This and other speech inadequacies contributed to their extinction

But what exactly do we have to explain?

• You can teach chimps a little bit of symbol use.– they can do this with their hands– but not with their mouths.

• Humans can learn to sign as easily as they learn to speak– even though in all of recorded human history there has

been no signing– by contrast, reading is tough

• Brain areas most associated with language are also associated with motor movement.

But what exactly do we have to explain?

• Thus, the most reasonable hypothesis is that language, in its origin, was gestural.

• This means a theory of language origins needs to worry most about the cognitive changes, not the physical changes that made vocal, articulated speech possible.

Why not grammar?

• Many theorists of language evolution argue that grammar, not symbolic reference, is what sets humans apart.– As if grammar could be of any use without

reference!

• Certainly human grammar is quite amazing, but it will not be selected for unless there is something to talk about.

Assume True Imitation

• Assume that our capacity for imitation improved beyond what we see in modern chimps.

• Assume that humans are capable of copying technique after observation.– it is something that chimps cannot do, so, if

this explains language, it also explains why chimps don’t have it.

True Imitation model ranking

• True Imitation will select for copiers who rank potential models and prefer the most skilled.

• In indirect forms of learning (e.g. local enhancement, or goal emulation) this will not be true.– Only when I obtain information directly from my

model can the distribution of skill among potential models be tapped.

• If all you get from a model chimp is the association between using reeds and eating termites.

• If you can’t learn the specifics of that model’s technique

• then you should pay no attention to skill differences among the models.

• But if you can directly acquire the model’s specific motor patterns (his/her individual technique), you will benefit from ranking the models according to skill.

• You should pick the most skilled fisher as your model.

What does this all mean for the invention of symbolic reference?• First, both model and apprentice benefit

from linguistic innovations.– The client will learn faster and better.– Given that, other clients are bound to notice,

and so the model will get a bigger posse of clients.

• In other words, the occasional model-apprentice genius pair has a clear incentive to get reference started.

What does this mean for the stability of the trick?

• Since the invention of reference by such a genius pair will improve the quality of the learning experience…

• This allows that model to capture most if not all of the client market in the population.

• And so the innovation quickly spreads to the whole group and is therefore stable.

• The innovation does not die with the two genius inventors.

What about the evolution of language genes?

• Given the intergenerational stability of reference, it can act, through a Baldwinian process, as a selection pressure acting on genes that make individuals fast learners of the ‘trick’ of reference.

• After that it is all a series of Baldwinian feedback loops and the rest of the history of language evolution unfolds mechanistically.

Capacities required for linguistic innovation

Linguistic Innovation

Spread to populational fixation

Selection for genes that make learning the innovation easier

(Baldwin effect)

First step