How word came to us

"Speak and I will baptise you," Cardinal Polignac is supposed to have said in the 18th century to an orang-outan newly delivered to the King of France's zoo. Every human being speaks at least one of the 4 000 languages that have been inventoried on Earth. Enquiring into what makes us human beings requires us to reflect on the appearance of language, both in the course of evolution and in that of a child's development.

And what if learning by imitation was just as effective, if not more, than oral learning, for example to transmit skills like tool-making, which has existed since pre-historical times. Here, a Mesolithic flint from the Fond des Blanchards (Gron-Yonne-FR). © CNRS Photothèque/INRAP/Jérôme Chatin
And what if learning by imitation was just as effective, if not more, than oral learning, for example to transmit skills like tool-making, which has existed since pre-historical times. Here, a Mesolithic flint from the Fond des Blanchards (Gron-Yonne-FR). © CNRS Photothèque/INRAP/Jérôme Chatin

We have known for more than a century that the production of language requires the integrity of a region of the brain's left cortex. Neurologist Paul Broca discovered this area, which today carries his name, when examining a patient suffering from aphasia, the autopsy of whom revealed the destruction of this region of the brain in a vascular accident.

Modern neuroimaging methods confirm that Broca's area is activated when we speak. It would therefore be tempting to tackle the question of the birth of language in terms of the appearance of this cerebral zone, which does not exist in primates. Tempting, but difficult, because the brain consists of soft matter that does not fossilize. Examining moulds of hominid crania does not therefore allow us to say with certainty whether this all-important Broca's area did or did not exist in our ancestors' brains. Some scientists believe that it existed from the Homo habilis stage (4 million years ago) onwards, while others say that it appeared only with Homo sapiens (a hundred thousand or so years ago), with earlier species having only a rudimentary protolanguage at their command.

The position of the larynx

Palaeontologists have, however, found another way of tackling the question about the advent of articulated language. Speaking requires Broca's area, but it also calls for a vocal apparatus consisting of the tongue, the larynx (the membranous folds of which form the vocal chords) and the pharynx, which draws the air from the larynx towards the mouth and the nose. The longer the pharynx, the longer the air can vibrate, and the greater the possible range of sounds. In adult humans, the larynx is situated low down, at the bottom of the throat. In the great apes, on the other hand, it is situated at the top. "In this way humans can form vowels by modifying the shape of the tongue in two dimensions - vertically, at the base of the tongue at the bottom of the throat and horizontally, at its extremity in the mouth - which increases the range of sounds," explains James Steele of the Institute of Archaeology at University College London, who coordinates the Hand to Mouth(1) project.

Does this make the descent of the larynx to the bottom of the throat the anatomical signature for the appearance of language? It is on this hypothesis that researchers are working. By reconstructing in a computer model the shape of the vocal tract of hominid fossils, they hope to be able to date the appearance of a larynx sufficiently low to permit the production of articulated language.

Just how useful is language?

Another way of postulating the problem is to ask in what way language could have been useful to our distant ancestors. Speech has disadvantages as well as benefits. With a low larynx, air and food can circulate at the bottom of the throat, with a risk of suffocation if either takes the wrong path. From an evolutionary viewpoint, this danger has therefore to be counterbalanced by other benefits. Which? For the Hand to Mouth teams, this could be the production of tools. "The manufacturing of tools is a social activity that needs to be passed on from generation to generation by teaching," James Steele continues. "We are seeking to understand whether this skill is acquired more effectively by mere imitation, or by oral teaching." The question is widely debated. Japanese university professors have taught their students to produce carved stone tools, both by demonstrating the processes in silence, and by giving them precise oral instructions. The outcome: both groups had comparable results, in both cases very poor, given the complexity of the task. It is this type of experiment, until now inconclusive, that the Hand to Mouth project is looking to pick up again using the expertise of its archaeologists and anthropologists. The theoretical context of this research has been renewed with the recent discovery of mirror neurons - which are activated only when a subject reproduces an observed action - in a region of the brain that is involved in speech.

For James Steele, "this discovery suggests the appearance of certain properties of human language to have been dependent on the preexistence of neuronal circuits which serve to read other people's behaviour by regarding their movements."

Wakening to language

The appearance of articulated language has probably necessitated a series of anatomical changes to the brain and the vocal apparatus during evolution. And in small children? In a baby, as in the great apes, the larynx is situated high up, allowing it to suckle and breathe simultaneously. It then moves down quickly, whereby baby's first burbling noises become articulated words. This wakening to language thrills parents as much as it fascinates scientists.

In specialist terminology human language is a generative system, which makes it possible to construct an infinite number of sentences from a finite number of words (50 000 to 100 000 in an average adult vocabulary), the meaning of which is fixed by convention. If we don't know the meaning of a word, we look it up in a dictionary. On the other hand, we can understand the meaning of any new combination of words within a sentence, because this combination is governed by a set of rules, known as syntax. From age 3 or 4, children master the essentials of this syntax, without learning it.

We never learn at school that in the sentence "This child has a ball. That one also has one", that ‘that' designates another child and the second ‘one' another ball. Hence the idea, advanced in the 1950s by American linguist Noam Chomsky, of a human genetic predisposition towards language learning. Since then, hundreds of researchers have attempted to decrypt the foundational basis of this innate ‘universal grammar', whose existence was posited by Chomsky, but whose nature remains enigmatic.

Consonants and vowels

"The sound signal of the word does not contain any evident information relative to the lexicon or grammar of the language," notes Jacques Mehler, a specialist in cognitive sciences at the Scuola Internazionale Superiore di Studi Avanzati in Trieste (IT) and coordinator of the Calacei project. "Even if we presuppose the existence of very powerful innate structures, we still have to explain the relationship between the linguistic structure and the perceived signal. Recent research shows that the signal is richer than we thought, containing ample statistical information on the distribution of certain fundamental elements which are perhaps detected unconsciously when we learn to speak." Hence the hypothesis being tested by Mehler and his team that consonants are used by the brain to make out words in prosody whilst vowels are used in particular to make out the syntax.

The Calacei researchers are also interested in the way a newborn child learns its native language. Earlier work at Trieste had established that babies are sensitive to the rhythm of words from birth. If a baby is made to listen to different languages and we measure its attention by observing its eye or head movements, we discover that it is already capable of segmenting words into consonants/vowels in order to determine certain rhythmic properties of languages. The Calacei team has been able to show that a four day-old baby can already distinguish the repetition of A-B-B type syllables from A-C-C type ones. This faculty implies an activation of the Broca's area, which matures very early. From the age of three months, we observe that a child reacts to a recording in its native language; but not the same recording when played backwards. Even more surprising is the fact that the detection of these incongruities in successions of syllables elicits a reaction of surprise from the baby, as if it were expecting something to follow. This predictive faculty constantly develops with the acquisition of language... and the appearance of humour. As humourists will tell you, one of the strongest sources of comedy consists of verbal incongruities which throw off balance a brain that had been expecting something else.

Wasn't it François Rabelais who said, in the 16th century, that "laughing is specific to humans"?

Mikhail Stein

  1. The Hand to Mouth and Calacei projects are part of the European initiative Nest Pathfinder, What it means to be human.