A question without fossils

Writing is only about 5,000 years old. Spoken language is far older, but sounds disappear the moment they are made, and the soft tissue that produces them rarely survives. Estimates of when full language appeared differ by hundreds of thousands of years.

In the 19th century, this gap was filled with so much speculation that the Société de Linguistique de Paris wrote a ban into its rules in 1866: it would accept no papers on the origin of language. (The same rule also banned papers on creating a universal language, a story told in Why do people invent languages?)

Interest returned in the late 20th century, when biology, genetics and animal research began to supply real evidence. That evidence does not give a full answer, but it has overturned some popular ideas.

The voice box: special, but not unique

In adult humans, the larynx (the voice box) sits low in the throat. In newborn babies it sits high, and it moves down during the first years of life.

For a long time, this low larynx was seen as the key to speech. The linguist Philip Lieberman and others argued that it gives humans a vocal tract in two parts, the mouth and the throat, which makes the full range of vowels such as "ee", "ah" and "oo" possible.

In this view, the low larynx came at a price. Charles Darwin had already noticed that in us, as in other air-breathing animals, everything we swallow passes over the opening of the windpipe. Lieberman argued that a low larynx makes choking on food more likely, although how much more is debated.

Later research complicated this picture:

  • Other animals have a low larynx too. Tecumseh Fitch and David Reby (2001) showed that male red deer and fallow deer have a lowered larynx and pull it down even further when they roar. Lions and tigers have one as well. A longer vocal tract makes a call sound deeper, as if it came from a bigger animal, so a low larynx is not only about speech.
  • Monkey throats could make many speech sounds. In 2016, Fitch and his colleagues filmed macaques with X-rays while they ate, yawned and called, and modeled the sounds their vocal tracts could make. The range of vowels was wide enough for clear speech. They even produced a computer version of "Will you marry me?" with a macaque's vocal tract.

Their conclusion: monkeys have a vocal tract that is ready for speech, but not a brain that can control it for speech. Anatomy still matters, and researchers continue to debate how much. But the brain's control over the throat, tongue and lips matters at least as much as their shape.

FOXP2: the gene the press called the "language gene"

That brings us to the brain, and to a British family known in research as the KE family. About half of its members, across three generations, have serious difficulty speaking clearly. Researchers began studying them in the 1990s, and in 2001 a team led by Cecilia Lai and Anthony Monaco found the cause: a change in a single gene, FOXP2.

The affected family members find it hard to make the fast, precise sequences of mouth and face movements that speech needs, and they also have problems with some aspects of language beyond pronunciation. Newspapers soon called FOXP2 the "language gene".

The real picture is more modest:

  • FOXP2 is a gene that switches many other genes on and off. It is active in the developing brain, and also in organs such as the lungs.
  • Many animals have it. In zebra finches, reducing it in a brain area used for song makes young birds copy their tutor's song less accurately (Haesler et al. 2007).
  • The human version differs from the chimpanzee version by two small changes in the protein (Enard et al. 2002). Neanderthals had the same two changes (Krause et al. 2007), so they are older than our own species.

FOXP2 is best seen as one gene among many that help the brain learn and control fast movement sequences, including those of speech. It is not a gene "for" words or grammar.

Guessing the first words

Before such evidence existed, 19th-century scholars could only guess what the first words were. The German-born philologist Max Müller, who doubted these guesses, gave two of them the nicknames still used today, partly in mockery:

  • The "bow-wow" idea: the first words imitated sounds in nature, such as a dog's bark or the crack of a branch.
  • The "pooh-pooh" idea: the first words grew out of cries of feeling, such as "ouch" or "oh".
  • The "yo-he-ho" idea, a later proposal: language grew out of the rhythmic calls people make when they work together, like sailors hauling a rope.

Each explains a few words at most. As part 1 showed, even imitative words differ from language to language, and they make up only a small part of any vocabulary. Darwin had yet another idea: that something like singing came before speech.

Protolanguage: words before grammar?

A more recent proposal comes from the linguist Derek Bickerton (1990). He suggested that before full language, early humans used a protolanguage: words, but no grammar to combine them. A message might have been little more than "Bear. River. Run!"

Bickerton pointed to similar systems today: toddlers in the two-word stage, pidgins (simple contact languages that adults improvise when they share no language), and apes taught to use signs. All of them communicate with words and very little structure.

This is a hypothesis, not general agreement. Other researchers think the first units were whole messages that only later broke into separate words (Wray 1998), or that music-like calls came first, as Darwin suggested. Whether Neanderthals, our closest extinct relatives, had full language is also an open question, one we come back to in part 3.

The parallel with learning holds in one sense. A toddler saying "more juice" and a beginner saying "Coffee. Two. Please." both get a message across with words and almost no grammar. Words alone can carry a lot of meaning. Grammar makes it precise.

What's next

A pile of words is not yet a language. How did humans start combining words into sentences, and why? In part 3, we look at one surprising answer, gossip, and at syntax, the part of language that lets a few thousand words express an unlimited number of ideas.