Robert Sapolsky on Language: this video starts an exploration of human language. It covers language as a biological behavior and looks at universals of the roughly 6,000 or so human languages estimated to exist. It also explores the neurobiology of language, the acquisition and development of language, and a comparison with animal communication. The exploration of animal communication allows Sapolsky to suggest an answer to the question of what is unique about human language in the animal kingdom.

A deficiency of Sapolsky's biology-focused presentation is that it misses the tenor of the importance of language in human culture. So I'm motivating my notes with Wade Davis' poetic definition of language: "A language, by definition, is a flash of the human spirit. It's a vehicle through which the soul of each particular culture comes into the material realm. Every language [...] is an old growth forest of the mind, a watershed of thought, an ecosystem of social, spiritual, political possibilities."

That is even more poignant when you consider the fact that 90% of human languages are expected to go extinct this century. One of the most devastating signs of the rapacious side of globalization and religious missionaries.

Behavior of Language. Universals across all human languages.

Semanticity: breakup of the continua of sounds into discrete "buckets" or units of meaning (words).

Embedded clauses: to express conditionals and circumstances.

Recursion (generativity): there is a potentially infinite number of combinations of words.

Displacement from emotions: ability to talk about the past and future; here and there; emotionally distant circumstances. Animal communication is overwhelmingly about their current emotional state.

Arbitrariness of language: words are not tied inextricably to their meaning: an arbitrary relationship between the signal and the message.

Meta-communication: the ability to talk about language.

Motherese or baby-talk: speaking to children in a higher-pitch voice, slow and clear, emphasizing melodic structure and vowels, repeating phrases, and focusing on the baby's face.

Is it about emotion or instruction? Pet talk is less slow and clear and doesn't focus on the face so it is not about instruction.

Is language more about cognition or the conceptual structure or about the generation and reception of sound (motor system)?

Gallaudet University in Washington, D.C. "was the first school for the advanced education of the deaf and hard of hearing in the world" (quote from

Sign language suggests that language is mostly a cognitive faculty: it has many of the properties of language but uses different brain systems. Observe that at around 9 months, deaf and hearing babies begin to babble (practice fragments of whatever communication system they are exposed to). Pathologies of the brain affect ASL speakers in similar ways to people using spoken language.

Prosody refers to aspects of language like tone and emphasis that are not encoded in grammar or vocabulary. Sign language also has prosody.

Deaf people where ASL is their native language, use their auditory cortex when they "listen" to signs. Is that really evidence that language is about the underlying cognitive structures?

There are regional accents, poetry, and puns in ASL even though it doesn't use lips, tongue, or ears. If ASL is a second language it is "stored" in the same brain regions as a second spoken language.

Neurobiology of language.

How modular is the function of language in the brain? How separate is language from the rest of the cognitive functionality of the brain?

Williams syndrome is a genetic disease of very expressive, warm, affectionate, and verbally adept people whose IQs are near the borderline of intellectual disability. Their effective language skills imply that language is a separate faculty of the brain and not just one of many skills the brain develops. In addition, there are people with normal IQs who have various heritable language disabilities. Further supporting the idea that language is a modular function of the brain.

Nevertheless, people with Williams although fluent with words, do not communicate much meaning in their sentences. Moreover, many of the people with other heritable language disabilities are at the lower measures of intellectual function which suggests that language is probably connected in with general brain function (not so modular).

In conclusion, there is some evidence that language is a general function of the brain with quite a bit of localizations. Probably another "both/neither" rather than a clear cut situation.

The language function of the brain is lateralized (located on one side of the brain) to the left brain in 90% of people. In general, prosody is localized to the right brain in 90% of people.

The Wada test uses barbiturates to test for lateralized language and memory brain functions. Modern brain imaging will probably obsolesce this test.

Aphasia is the general term for being unable to communicate effectively.

There are three main regions of the brain associated with language: Broca's area, Wernicke's area and the arcuate fasciculus.

Broca's area is the edge of the brain region that handles motoric function (moving your body). It is about moving your lips and larynx. A Broca's aphasia or production aphasia affects the ability to produce language.

Wernicke's area is near the auditory processing region and is about language comprehension. A Wernicke's aphasia subject often generates a meaningless word salad (

The arcuate fasciculus is the collection of axons connecting Wernicke's with Broca's and so connects comprehension to language production.

Almost all stroke damage affects all three regions, but sometimes localized damage will affect one area or piece within it giving us clues to localized function.

Deaf people who are native speaker's of ASL use these same areas of the brain for production and comprehension and the communication between them. It appears that ASL and verbal language people use the same regions in the same ways. Whistled languages also use the same neurobiological centers in the brain.

Alexia: inability to read language.

Agraphia: inability to express language in writing.

Pictograph languages like Chinese have slightly different localization centers.

A fourth region of the brain, the basal ganglia, is used for motor control which is needed for both spoken and signed languages. Gesticulation appears to be tied into the motor output of language even with blind speakers which suggests that emotional inputs could be connected in the basal ganglia. Why?

The rest of the limbic system (see my notes on a video on this part of the brain: is deeply connected to all these areas. Speech therapists will often recommend emotive singing for production aphasias since activating the emotional limbic system can sometimes help speaker's overcome damage in Broca's area. Tourette syndrome sometimes involves cursing uncontrollably which implies limbic inputs to language production. Talking can often help us feel better (the basis for psychotherapy is that there are deep connections between language and emotion). Stimulation of the brain can sometimes result in an emotional utterance; stimulate the same region in a rhesus monkey and it will give a call. A lot of the emotional input is directed to the right hemisphere where prosody is generally centered.

Other primates have very small "thickenings" where the Broca's and Wernicke's areas are located, but basically humans are the only species with these brain structures. There is evidence in lateralized facial expressions that some primates have a connection between their limbic systems and expressions.

Language acquisition and development.

Roger Brown and colleagues published a landmark study "A First Language: The Early Years" which detailed the early stages of language development in children. At 15-20 months of age, children start saying combinations of words which they've never heard before. There are days when we can learn 10 new words a day so that by 18 one's vocabulary is in the 60,000 word range.

BF Skinner, leader of the behaviorist school of psychology (which emphasized that reward and punishment shapes learning language and all other human behavior), and Noam Chomsky, a famous linguist, debated their views on language acquisition in the 1960s and 70s. Chomsky argued that kids can generalize the rules of language to acquire its generative structure despite imperfect, partial examples from others. This "poverty of stimulus" argument suggests that there must be an innate faculty of language to explain the nature and speed with which language is acquired. There must be some prepared learning (a form of instinct in ethology; see for more on instinct, ethology, behaviorism and Skinner).

In addition to prepared learning, there appears to be statistical learning, that is, picking up rules from statistical patterns. There are studies showing that really young kids appear to notice statistical patterns in language.

"Any kid can learn any language." Even though different languages have different vowels and consonants. Between 1-3 months of age babies show roughly equal response to phonemes, the building blocks of sounds in a language. "As you begin to acquire your own [language], you start to be unable to hear the difference between other sounds".

By 3 months, kids can distinguish language from nonsense. Between 3-6 months, they start tuning in to vowel sounds. By 8-9 months, babbling starts and they can recognize word combinations. By 11 months they can generate vowels with consonants following soon thereafter.

In most cultures kids pick up language by observation. Only recently has language been explicitly taught to kids by presenting them with simple words and grammatical structures. But there is not yet good evidence that this helps them acquire language more effectively.

A second language learned after age 12 or so typically involves having a distinct accent. When people learn multiple languages before age 6, the languages are housed in overlapping regions of Broca's and Wernicke's. Languages learned later are housed in peripheral parts of those areas. Brain damage effects give the evidence: a language learned after age 6 might be affected while not other languages.

Sapolsky says it is widely believed in the field that "Languages are invented by kids". This surprises me: Klingon and Esperanto and Arabic (suggested by Richard Bulliet) and written Turkish are adult-designed languages. What about poets and other writers: don't they change the language? What about Public Speakers, Comedians, theater and TV, etc.? Don't each of us invent our language all the time? It isn't just kids. But is it really predominantly kids?

He cites evidence from Nicaraguan schools for the deaf where the kids invented their sign language. He suggests they invented Nicaraguan sign language, but is that a dialect or a whole language? Can languages get invented in whole cloth by kids? He said the Nicaraguan had teachers, so it was a joint project at least, right? He argues that older people rarely learn the new language being developed by kids. The transmission goes from older kids to younger ones. But didn't the teachers learn and adopt this new sign language?

Apparently it took 3 generations of students to devise a fully formed grammar and the first generation kids never fully adopted the more sophisticated form.

What are the dynamics of language development at the societal level?

Neurology of language acquisition

At 12-16 weeks of age in the fetus, evidence of lateralization in gene expression is found some of which appear to have a connection to families with language disorders. By 30 weeks, the structural asymmetry of the brain is observable. Metabolic lateralization takes several years after birth before it is pronounced. Kids tend to myelinate their Wernicke's area about 9 months of age which is 3 months before they activate Broca's (about 1 year).

More on myelination in these notes:

Judith Rich Harris emphasized the importance of peer instead of parental influences in children especially in language acquisition: kids grow up with the accent of the community around them and not the accent of their parents. Peers are extremely important in human social development. We seem to be wired to pick up language from our slightly older peers. Kids don't want to talk to their parents in the home language. Values in the local culture and language are very important: Tolstoy got upset when his nurse addressed him in an informal verb tense.

More about Judith Rich Harris' work is in my notes on Sapolsky's lecture "Behavioral Genetics I":

In some languages, words for kinship relationships inherently involve the speaker's relationship. Why does this imply more intertwined social relations?

In egocentric languages, like English, directions are given relative to one's body ("here" or "there"). In many other languages, directions are a function of external landmarks. Aboriginal Australians use compass directions. Lera Boraditsky tells how this requires them to employ a system of dead reckoning to know where true South is at all times. For English speakers we need a compass! In English stories progress from left to right. In these aboriginal languages, stories begin in the East where the Sun rises. So the IQ test for placing tiles depends on the compass direction in which the player faces, not left or right!

Sapir-Whorf hypothesis: the nature of your thought is deeply shaped by the language you speak. Sapolsky talks about two Amazonian tribes whose language affects their ability to do math. The Pirahã have no words for ordinal or cardinal numbers: their counting is effectively 1, 2, many (; the Munduruku have a more complex system, but they only have cardinal numbers through 5 ( They are not a cognitively challenged people because they demonstrate an astute understanding of plants and other advanced capabilities needed to survive in the rainforests of the Amazon.

Since early in the course, Sapolsky has been emphasizing the importance of the ethology (animal behavior) perspective: "ethology is interviewing an animal in its own language". But now he observes that that soundbite must be taken very cautiously "because communication is so intertwined with mind, with values, with meaning". So it is very problematical "to interview an animal in its own language". Most people who speak multiple languages find that their emotive styles, expressivity, analyticalicity, etc., are different in their different languages.

How are thought and language related? Even though these examples suggest that language shapes our thinking. For words to have been invented in the first place requires an a priori thinking pattern that gets captured in the invented language. Is language a culture's capturing of important-to-them thinking patterns? But language is so complex and expressive does it allow the expansion of thought into domains that could not be accessed without language? Is language simply a tool for thought? Did thought just invent language to communicate its content? But the act of listening itself suggests language evokes thought.

Is the vicious circle between thought and language resolvable or are they like yin and yang inextricably intertwined aspects of the human experience? Will the connection between thought and language always be tantalizingly ineffable? Or do we just need to invent the right word (or experiment) to communicate and clarify this relationship?

Animal communication

Some animals show evidence for semanticity (that discrete sounds can have meaning). For example, Vervant monkeys have discrete calls (sounds) for danger above or danger below. A word is a stable concept transmitted independent of emotion. These calls are words (Wiktionary: "The smallest unit of language which has a particular meaning and can be expressed by itself; the smallest discrete, meaningful unit of language."). Chickens also have unique calls for different kinds of dangers and some synonyms.

Multimodal communication (such as facial expression and calls) are observed, for example, by showing an incongruous facial expression and a call affecting a response in an observing animal.

Intentionality is also seen among animals: For example, vervant monkeys give alarm calls more readily when their relatives are around (squirrels too). He mentions that this assumes some theory of mind faculties (the recognition that there are other actors who can have different information than I do: see ).

What is unique to humans seems to be dramatic displacement (about things that are far away or imaginary), arbitrariness, embedded clauses, recursiveness, and the capacity to lie (which is related to arbitrariness). Dogs cannot lie about their fear because it is communicated pheromonally (by the chemistry of the breakdown products of stress hormones), so it puts its tail between its legs to try to reduce it pheromone emissions.

Attempts to teach language to other species

Viki the chimp was trained using behaviorist methods to speak English. She learned four words: mama, papa, up, and cup (

Luella and Winthrop Kellogg raised the chimp Gua with their son Donald for nine months ( When Donald started making chimp vocalizations his parents stopped the experiment. Oops!

Realizing that chimps lack the anatomy to produce human speech, Allen and Beatrix Gardner taught Washoe ASL ( Washoe combined the ASL signs for water and bird to invent the term water-bird or duck. She babbled before going to sleep. She lied in ASL. David Premack tested Sarah ( and found her to have embedded clauses and other elements of grammatical language. Koko the gorilla was taught ASL by Penny Patterson. She could talk about dreams, gossip, and lie.

Herbert S. Terrace led ASL experiments on Nim Chimpsky ( which were reported in a landmark paper "Can an ape create a sentence?" Science 23 November 1979: Vol. 206 no. 4421 pp. 891-902. Terrace argued in detail how nothing any of the chimps including Nim did fit the criteria for language. No word invention, random word orderings, more words did not imply more meaning, non-spontaneous utterances (not communication, just social responses). Patterson critiqued Nim's training: no continuity of trainers whose knowledge of ASL was minimal, communication assumes a deeper interest in the parties than a bunch of grad students tag-team teaching Nim.

Sapolsky concludes with the case of Kanzi, a Bonobo chimp ( The researchers are being much more careful in their data collection than previous researchers and they are finding that he can construct embedded clauses, logical constructions, analogy, his mistakes tend to be in semantic categories instead of random distributions.

Can animals do language? Can they communicate with nearly the same facility as humans? Do they have enough cognitive capacity or are their smaller Wernicke's and Broca's areas inherently limiting of their ability to master anything like the complexity and richness of human language?