麻豆传媒

DO LINGUISTS need to get away from the library and spend more time talking? Ever since Noam Chomsky revolutionised linguistics in the 1950s, research has focused on the structures underlying grammatically correct sentences and on our astonishing ability to both generate and understand an infinity of expressions. Now two new books separately argue that the to and fro of conversation is key to understanding language. If they are right, the idea of language as a computational system needs to make some room for that of language as a tool for cooperative communication.

Daniel Everett of Bentley University, Massachusetts, and Nick Enfield at the University of Sydney start from different places. Everett鈥檚 wide-ranging How Language Began is rooted in his 30 years working with tribal groups in the Amazon as an anthropologist. Enfield鈥檚 How We Talk comes out of the lab and speeds through extraordinary experiments on the fast-paced, interactive flow of conversation.

Everett has already had a famous debate with Chomsky and his colleagues, related in Everett鈥檚 bestseller Don鈥檛 Sleep, There Are Snakes. He set out as a missionary to live with the Pirah茫, a group of Amazonian hunter-gatherers, aiming to learn their language and translate the New Testament into it. Instead, his encounter with this happy culture, which values 鈥渋mmediacy of experience鈥� over unsubstantiated tales, led him to lose his faith and to discover that their language was unlike anything studied before.

Pirah茫 appears not to use recursion, the way in which other languages tuck phrases inside other phrases to build more complex sentences. That seemed to challenge core ideas about language and sparked a huge controversy. Now it is more accepted that languages can be expressive without necessarily using recursive forms.

鈥�Homo erectus, who lived over 60,000 generations before us, led the way: we are merely an upgrade鈥�

Surprisingly, the hero of How Language Began is not modern humans, but Homo erectus, who lived over 60,000 generations before us. 鈥淭he greatest hunter. The greatest communicator. The most intrepid traveller. Perhaps the greatest distance runner on Earth鈥� is how Everett sees our ancestor. H. erectus led the way: we are merely an upgrade.

Everett鈥檚 speculative account of our forebear鈥檚 life and travels are the most remarkable parts of a remarkable book. I felt I had been born 1.8 million years too late. Although my brain would have been just two-thirds of its modern size, a pristine world would have lain before me.

H. erectus left their home in Africa and spread throughout Europe and Asia, ever onwards. Their tools, settlements, use of fire and crossing of waterways in boats or rafts all shout to Everett that it was 鈥渦pright man鈥� who first had symbolic thought and culture, and must have invented the first forms of language, too.

Everett begins his argument with the life of H. erectus and the origins of symbolic thought, moving through millennia of 鈥渦pgrades鈥� to our brains, vocal apparatus and language in all its varieties. We travel from H. erectus鈥榮 simple exchanges to our ability to pour out over 100 words per minute in conversation.

By pushing language鈥檚 origins back so far, Everett again collides with Chomsky, who sees language emerging in Homo sapiens about 80,000 years ago. But Everett doesn鈥檛 merely dispute the 鈥渨hen鈥� of language, but also why it was invented and how it changed. By 鈥渋nvented鈥�, Everett means just that: language was a creation of culture, growing out of humans鈥� developing social cognition and capacity to use symbols. For him, language is a communicative tool and not the product of sudden genetic change that gave H. sapiens new powers of thought, as Chomsky argues.

What kind of language might H. erectus have spoken? The earliest languages might simply have had a series of loosely ordered words and gestures. The shared context and culture would have been enough 鈥渢o determine the interpretation needed鈥�, writes Everett. More complex languages, with hierarchy and recursion, could have evolved from there.

At each step, language would have to fit our cognitive and perceptual limitations, auditory range, vocal apparatus and brain structures. At the same time, the advantages of communicating successfully would create new selection pressures on humans: brains evolved for languages, and languages evolved for brains. But, stresses Everett, language didn鈥檛 begin with the first word but 鈥渢he first conversation, which is both the source and the goal of language鈥�. Language is about interaction, about the simplest exchange in a shared culture.

Everett鈥檚 鈥渂aby steps鈥� version of how language evolved is at odds with Chomsky, who explains in Why Only Us, his 2016 book with Robert Berwick, that language did not emerge for communication, but from a 鈥渟light rewiring鈥� giving some individuals new powers of thought. Chomsky is adamant that 鈥渢here is no room in this picture for any precursors to language 鈥� say a language-like system with only short sentences鈥�. There is a clear choice: communication and conversation came first and grammar later, or a language of thought came first and was later 鈥渆xternalised鈥�, as Chomsky puts it.

Conversation is central for Enfield, too. His sparkling book, How We Talk, sets out to show that the power of the 鈥渃onversation machine鈥� is as astonishing as any of our grammatical achievements, and that how we direct our fast-paced exchanges challenges how we think about human nature.

The action, Enfield explains, is concentrated in a 1-second window that opens as soon as someone stops speaking. Taking an average from 10 languages, an answer to a question will appear in around 200 milliseconds 鈥� the time it takes to blink. English speakers average 236 ms, while the Japanese manage it in an astonishing 7 ms and the deeply thoughtful Danes in 468 ms.

We soon discover the timing of replies can be manipulated. 鈥淒o you want to go for a drink?鈥� Reply 鈥測es鈥� inside 200 ms and you are off to the bar; delay another 400 ms and you will be asked if you really want to go. But choosing 鈥渘o鈥� in the first 100 ms signals flat rejection. Say nothing in that 1-second window and you may face a frown for your lack of cooperation and be asked again.

Then there are the 鈥渦hs鈥� and 鈥渦ms鈥� of English: 鈥渦h鈥� says hold on for a short time (about 250 ms), while 鈥渦m鈥� indicates you should expect a longer delay (about 700 ms). Both delays can either signal 鈥淗old on, I鈥檓 processing what you鈥檝e said鈥�, or warn that a negative response is coming.

Finally, there is 鈥渉uh鈥�. In 2013, Enfield and his colleagues found that this, or something sounding very similar, was the one universal 鈥渨ord鈥� in all languages. It is, he explains, the quickest, easiest sound humans can make, and a speedy sound is needed to interrupt fast when you lose track of a conversation and need help.

There is a deeper significance here, too. 鈥淭his little word, like the turn-taking system it operates in, suggests a moral architecture to communication,鈥� says Enfield. 鈥淗uh鈥� symbolises 鈥渦niversal cooperation鈥�, the commitment people make to 鈥渉elp each other, where necessary and possible, to stay on track in conversation鈥�. That requires social cognitive skills unique to our species, writes Enfield, for even 2-month-old babies love to take turns in their interactions with adults.

Enfield closes his slim book with an inspiring chapter. We are, he argues, 鈥渙n the verge of a full-blown scientific revolution in research on the human capacity for language鈥�. He lays out the two current big ideas: that language can鈥檛 exist without specifically human forms of social cognition and interaction; and that language is a 鈥減rivate, purpose-specific computational system for operating upon information鈥�.

For Enfield, though, there is no need for one idea to triumph over the other. Rather, we can draw on all sorts of research. We can, say, welcome what has been learned about the structure of language while looking at how the flow of conversation may have shaped grammar. Despite the dismissive and brutal arguments common in linguistics, he is optimistic a new generation will build bridges.

I suspect Everett would welcome that future too.

Book information

This article appeared in print under the headline 鈥淭he amazing power of the conversation machine鈥�