The Language Instinct: How the Mind Creates Language

by Steven Pinker

  The really ancient superfamilies, like Nostratic, Amerind, and Eurasiatic (to say nothing of Proto-World), are still dismissed by most linguists. So is the idea associated with the human geneticist Luca Cavalli-Sforza that genetic families and language families should coincide. This is sometimes true for very large racial and ethnic groupings, like the speakers of Semitic, Bantu, San, and European languages in Africa, but is for from true in the general case. The reason is that languages, unlike genes, are not always transmitted vertically from parents to children but often are transmitted horizontally from conquerors to the conquered, from majorities to immigrants, and from prestigious speakers to déclassé ones.

  In 2004, two geneticists, Alec Knight and Joanna Mountain, came up with a startling claim about Proto-World, the ultimate mother tongue spoken by the first modern humans, namely that it was a click language. Their hypothesis, though certainly not accepted by most linguists, is not crazy speculation but was based on four observations. First, the two main families of click languages spoken in Africa (by the Hadza in Tanzania and the San in the Kalahari Desert) are linguistically unrelated, in the sense of having no common ancestor within the past 10,000 years. Second, the San and Hadza peoples are genetically unrelated to each other. Third, each group has levels of genetic diversity suggesting that the two are descendants of the ancestors of all living humans. Fourth, linguists have documented that languages with clicks often lose them, but languages without clicks never originate them. A simple explanation is that the first modern humans, who lived around 100,000 years ago, had a language with clicks, which survived in these two African peoples but were lost in all the other descendant groups.

  Language death is, of course, still a major concern of linguistics. Two organizations that support and document endangered languages are the Foundation for Endangered Languages (www.ogmios.org/home.htm) and the Endangered Language Fund (www.endangeredlanguagefund.org).

  Chapter 9: Baby Born Talking—Describes Heaven. A nice introduction to how babies talk is a book called How Babies Talk by the psychologists Roberta Golinkoff and Kathy Hirsh-Pasek.

  The evidence that young brains are better than older brains at learning and creating language has been piling up in the last dozen years, and there is evidence for a gradual decline in the ability to master an accent beginning as young as two. Neuroimaging studies suggest that a second language acquired in childhood is processed in the brain in a different way than a second language acquired in adulthood: in the former case, the two languages completely overlap; in the latter, they stake out distinct adjacent regions.

  At the same time it’s been hard to prove that there is a discrete “critical period” for language acquisition. The linguist David Birdsong has suggested that people simply get worse as they get older: children are better than adolescents, who are better than twenty-somethings, who are better than thirty-somethings, and so on. Birdsong endorses the hypothesis in this chapter that age effects on language are part of the general process of senescence. The issue is not so easy to resolve, because people learn a second language under a wide variety of circumstances and motivations, which could blur any blips or elbows in an age curve.

  Another complication is that age may play a clearer role in acquiring a first language than a second. Psychologists have long suspected that adults might do a pretty good job at learning a second language by falling back on their first language as a crutch, learning the second in terms of how it differs from the first. A nifty study by Rachel Mayberry, reported just a bit too late for me to have included it in The Language Instinct, bears this out. She found that congenitally deaf people who learned American Sign Language as a first language in adulthood did far worse than people who lost their hearing from an accident or disease and learned it as a second language in adulthood. (Congenitally deaf adults who learned ASL as children were, as one would expect, best of all.) This confirms that adults are much worse than children at acquiring a language, but that the difference is masked by the fact that most adults are learning a second language, not a first one.

  Age effects in language acquisition have figured in a controversy in American educational policy that might be even more contentious than the reading wars. Until recently, many American states had an eccentric version of bilingual education in which children of immigrants were taught in their native language (usually Spanish), and English was introduced only gradually, with immersion often delayed until they were on the verge of adolescence. Many academics support these programs (partly because they had a vague aura of being pro-minority and pro-immigrant), despite the opposition of many immigrant parents, the lack of good data showing that the programs helped children, and their underlying assumption that older children are better at acquiring a language than younger children. Ron Unz, the activist who spearheaded plebiscites curtailing the programs, pointed out that this is like assuming that rocks fall upward.

  Chapter 10: Language Organs and Grammar Genes. Aside from the fallout of the genomic revolution, the biggest scientific advance connected with language has been the analysis of the brain through neuroimaging, particularly fMRI (functional magnetic resonance imaging) and MEG (magnetoencephalography). The classic language areas generally come through in these studies, though the overall picture is now far more complicated. An example close to home is this image of your brain on language, which my former student Ned Sahin and I obtained when we scanned people as they read words on a screen and repeated them silently or silently converted them to the plural or the past tense:

  The computer has “inflated” this rendering of the left hemisphere, so that the sulci (grooves) are visible as the darker gray blobs. The hot spots represent increased blood flow, and you can see that many of the areas are similar to the ones I depicted in Chapter 10 based on the data available at the time (autopsies and CAT scans). At the bottom rear (lower right in the picture) you can see the primary visual cortex. To the immediate left of it is a newly discovered region called the “visual word area,” where word shapes are detected. Above that is a region in the vicinity of Wernicke’s area, which is involved in recognizing the word. The huge forest fire at the center of the frontal lobe (on the left side of the picture) includes Broca’s area (which is involved in grammatical computation) and areas involved in controlling the mouth. A part of this inferno extends downward into the large groove beneath the frontal lobe (the insula) and may reflect the programming of articulation (even though our subjects did not actually speak, because the head movements would have blurred the image). The activity seen in the long horizontal groove high in the parietal lobe at the back of the brain (at the top right of the picture) reflects people’s attention to the visual display.

  Unfortunately, beyond this coarse geography, the bounty of neuroimaging data on language of the last decade has not led to a neat picture of what parts of the brain do what kinds of work in language. But there have been a few attempts to organize the mess. Peter Hagoort has argued that the left inferior frontal lobe (the large region on fire at the left of the picture, which includes Broca’s area) is charged with “unifying” bits of linguistic knowledge (words, rules, sounds, constructions) into a coherent and meaningful sentence. Within this region, Hagoort suggests, there is more processing of meaning at the lower front end, more processing of sound and articulation at the upper hindmost end, and more processing of grammar in the middle. Another scheme, by David Poeppel and Greg Hickok, focuses on the comprehension side. They suggest that word understanding begins in the general vicinity of Wernicke’s area and then splits into two streams. One heads down and forward in the temporal lobe (the elongated lobe at the bottom of the picture) and connects to meaning; the other extends up and then forward to the frontal lobes and connects to articulation.

  The development of neural circuitry of the brain in utero, which I described in purely hypothetical terms, is another area of biology that has mushroomed in the past dozen years. In his highly readable book The Birth of the Mind, my former student Gary Marcu
s explains the basic science behind embryonic brain development, and speculates on how the circuits underlying language and cognition might be laid down by genes and early neural activity.

  Chapter 11: The Big Bang. The evolution of language, which didn’t exist as a field when I wrote The Language Instinct, is now the subject of many books, conferences, and research programs. A recent volume called Language Evolution: States of the Art contains position statements by the major players; I have an essay in it arguing (as I did in this chapter) that language is an adaptation to the “cognitive niche,” in which humans use language to negotiate cooperative relationships and to share technical know-how. The burgeoning field is even the subject of a new popular science book by the journalist Christine Kenneally.

  The most exciting new developments in the field come from the genomic revolution. Several genes or genetic loci with a role in language have been identified, confirming that language is genetically complex and not the result of a single lucky mutation. Even more remarkably, there are new techniques that can analyze genetic variation and distinguish genetic changes which have been naturally selected from those which spread by chance. One method is to see whether the nucleotide changes that affect a protein product (and hence are visible to natural selection) are more numerous than the changes that have no function, and hence must be random evolutionary noise. Another is to see whether a gene shows less variability within the members of a single species than it does between different species. Not only does the FOXP2 gene show these fingerprints of selection, but so do several genes involved in auditory processing in humans (but not in chimpanzees), presumably because of the demands of understanding speech.

  Another important development is that computational evolutionary linguistics is no longer a one-man enterprise. My colleague Martin Nowak has developed several mathematical models that add teeth to the intuition that some of the basic design features of language confer selective advantages to intelligent social agents. These include syntactic rules that express complex meanings and the so-called duality of patterning, in which phonemes are combined into words and words are combined into sentences.

  In 1995, I took part in a conference at UCLA in which Sue Savage-Rumbaugh announced that someday she expected Kanzi (the pygmy chimpanzee she had trained with symbol systems) to be giving her talks for her. We’re still waiting. Though I think that Kanzi and other bonobos can understand and use words with greater reliability than had been shown when I wrote The Language Instinct, their ability to combine them remains rudimentary. Indeed, the striking achievements in animal communication have been seen in species that are far more distantly related to us than chimpanzees. The most receptive trainee for an artificial language with a syntax and semantics has been a parrot; the species with the best claim to recursive structure in its signaling has been the starling; the best vocal imitators are birds and dolphins; and when it comes to reading human intentions, chimps are bested by man’s best friend, Canis familiaris. This pattern bears out my advisory that it’s a mistake to ask about language in “animals,” as if there were some evolutionary gradient with humans at the top and chimpanzees one rung down. Instead, animals at different positions in the tree of life evolved the cognitive and communicative abilities that are useful to them in their ecological niches. Humans are still the only species that naturally develops a communicative system with a combinatorial syntax and semantics, befitting our unique occupation of the cognitive niche.

  In an unusual collaboration, Chomsky wrote a paper in Science in 2002 with the comparative psychologists Marc Hauser and Tecumseh Fitch that sought to bridge the rift between linguistics and animal behavior research. The authors distinguished between language in a “broad sense,” namely the entire set of abilities that go into speaking and understanding (concepts, memory, hearing, planning, vocalizing), and language in a “narrow sense,” namely the abilities that are unique to language and unique to humans. They suggested that the broad language faculty contains many abilities we share with other animals, but that the narrow language faculty consists only of syntactic recursion. As I have mentioned, Ray Jackendoff and I were unpersuaded, and expressed our reservations in a debate in the pages of Cognition.

  Chapter 12: The Language Mavens. This was by far the most widely noticed chapter in the book. Despite my statement to the contrary, many readers assumed that I was opposed to any kind of encouragement of standard grammar or good style. Some assumed that I was advocating a 1960s-style attitude of doing your own thing, letting it all hang out, and taking a walk on the wild side. As a radical language libertine, I even made an appearance as a character in David Foster Wallace’s novel Infinite Jest. In fact the chapter simply publicized what everyone who has studied the history of English soon discovers: many prescriptive rules, despite being cited with an air of dogmatic certitude and haughty one-upmanship, are pure twaddle and have no basis in logic, style, clarity, or literary precedent.

  I do feel bad at having had some fun at the expense of the witty writers Richard Lederer and William Safire, especially after The New Republic excerpted the chapter with a title and cover that took some gratuitous digs at Safire. When I met him a year later, he was gracious about the episode, and he has occasionally consulted with me for his columns since then. The same cannot be said for John Simon, who surmised in the National Review that I was trying to excuse the bad grammar of my uneducated parents.

  My call for a language maven who thinks like a linguist has been answered by Jan Freeman, who writes an unfailingly insightful column called “The Word” in The Boston Globe (http://www.boston.com/news/globe/ideas/freeman/). And my call for linguists who address style and usage has been answered by Geoffrey Pullum and Mark Liberman in their delightful blog called “Language Log” (http://itre.cis.upenn.edu/~myl/languagelog/), with occasional contributions from Geoffrey Nunberg, John McWhorter, and other linguists. (See also the book by Liberman and Pullum that I recommend in “Read on.”)

  Chapter 13: Mind Design. I have expanded the content of this chapter into two books: How the Mind Works, which is about the other cognitive and emotional instincts that make up human nature, and The Blank Slate, which is about the idea of human nature and its political, moral, and emotional colorings.

  Read on

  Author’s Picks Suggested Reading

  If you liked The Language Instinct, I think you’ll like these…

  Steven Pinker, How the Mind Works (1997), Words and Rules (1999), The Blank Slate (2002), The Best American Science and Nature Writing (2004), and The Stuff of Thought (2007). Shameless self-promotion.

  David Crystal, The Cambridge Encyclopedia of Language (2nd ed., 1997) and The Cambridge Encyclopedia of the English Language (2nd ed., 2003). Not really encyclopedias, but lavishly illustrated, easily browsable, and thoroughly addictive collections of essays on every aspect of language you can imagine.

  Judith Rich Harris, The Nurture Assumption (1998) and No Two Alike (2006). The mystery of what makes us what we are. It’s not just the genes, but it has even less to do with the way our parents brought us up.

  John McWhorter, Word on the Street: Debunking the Myth of “Pure” Standard English (2001), and The Power of Babel: A Natural History of Language (2005). More on language, from a linguist with expertise in Creoles, Black English Vernacular, and the relation of language to culture.

  Mark Liberman and Geoffrey K. Pullum, Far from the Madding Gerund, and Other Dispatches from Language Log (2006). Hilarious, erudite blog postings on linguistics and public life.

  Diane McGuinness, Why Our Children Can’t Read and What We Can Do About It: A Scientific Revolution in Reading (1997). Not just a book on pedagogy but a history and explanation of the remarkable invention we call the alphabet.

  Thomas Sowell, Late-Talking Children (1998) and The Einstein Syndrome: Bright Children Who Talk Late (2002). An unexpected interest of the economist, columnist, historian, and father of a late-talking child.

  Bill Bryson, The Mother Tongue: English and
How It Got That Way (1991). An entertaining history of the language from the well-known humorist and travel writer.

  Roger Brown, Words and Things (1958). One of the inspirations for this book, from my graduate school adviser.

  Rebecca Wheeler (Ed.), The Workings of Language (1999), and Stuart Hirschberg and Terry Hirschberg (Eds.), Reflections on Language (1999). Essays by linguists and journalists on many aspects of language in the public sphere, including “uptalk,” accents, sex differences, Ebonics, the reading wars, literary style, and the English-only movement.