This video is brought to  you by language.

Literally. It’s what I’m using right now  to transmit information to you.

Language is one of the fundamental,  uniquely human things about us. But where did it come from? Language doesn’t exactly fossilize,  so to answer that, we have to look   at it like any other biological trait  and try to determine how it evolved.

Here’s just a few of the ways  scientists have tried to talk it out. [♪ INTRO] Figuring out when we developed language  is really important for a lot of reasons. For starters, plenty of anthropologists  and linguists think that language was one   of the key factors that allowed  modern humans to develop complex   societies and other behaviors that make us unique. It’s a lynchpin in social interaction,   sharing information, formulating worldviews,  and developing cultural ideas like myths.

But like all big questions in research,  this one has proved to be contentious. So much so that in the 1860s, the Linguistic   Society of Paris forbade its members from  discussing the origin of language at all. And a few years later, the equivalent  British group followed suit.

They may have been worried that these  contentious discussions could lead to   researchers exchanging more than just words. And while at least researchers are able to  study it now, there’s still a lot of debate. But if we’re going to talk about when language  evolved, first we have to define what language is.

More specifically, we need  to define what makes human   language different from the ways  that other animals communicate. Bees do dances to tell their  coworkers where food is,   prairie dogs have different alarm  calls to warn of different predators, and a parrot named Alex reportedly had  a vocabulary of over 100 words that he   was able to use to identify  things and answer questions. But there’s a lot more to a language than that.

While there are differences of opinion  over the best definition of a language, one of the features that mostly everyone  thinks a language needs is recursion, or being able to embed one piece of  grammar in another of the same type. For example, in the sentence  “Becca told me that Russell   baked cookies,” the words “Russell baked cookies” could function as a whole sentence on their own. But by putting them inside another sentence,   we have a new idea that we couldn’t  have had without that recursion.

Being able to stack phrases like this allows   us to create an essentially  infinite number of sentences. We can do this, and understand people when  they do it, because language has syntax, or rules that determine how pieces and parts  are put together for specific meanings. So language requires two skills.

The cognitive ability to  understand these recursions,   and the physical ability to express them. Which brings us to the question  of how we got those skills. There are two basic ideas, both of which  have their supporters and detractors.

The first is that language is so  complex that it had to develop   in teeny tiny increments over  a super long period of time. In this version, language and our cognitive skills  co-evolved, with natural selection acting on a   series of pre-language traits, eventually  evolving into a fully fledged language. That could have looked like having a small  vocabulary of individual word-sounds,   then combining two words  together to mean something new, then eventually developing syntax and  recursion and language at the same time.

Baby steps. The other idea is that the evolution  of language happened a lot faster,   and we basically went from no  language to language all at once. In this scenario, natural selection would have   worked with traits that had  nothing to do with language.

So maybe our brain size evolved  because it favored tool-making,   and our inner ears evolved to be extra  sensitive to the pitches most common in nature. And once we had evolved all of those physical  and cognitive traits for other purposes,   all it took were tiny tweaks to put all  those together and make a language.. And to figure out which of  these hypotheses is correct,   we have to turn to physical evidence to  let us peek into evolutionary history.

Which is really squishy, because language  itself doesn’t leave much evidence behind. Sure, there’s writing, and the oldest example of   that is Mesopotamian cuneiform,  which is about 5000 years old. But we’re pretty sure that language  has to be older than writing itself,   and probably by a long shot.

Because many of the things that humans have done   just wouldn’t be possible without  a way to communicate big ideas. Like, how do you tell another person  how to build a structure without words? So we have to look for indirect evidence  of language that predates the written word.

Language isn’t really possible without symbols  and the cognitive ability to process them. A word can represent an object,  but it’s not the object itself. This kind of symbolic representation means that  language lets us talk about things that aren’t   directly in front of us, plus abstract ideas  or concepts like fear or love or extinction.

And because these things aren’t  exact representations of concepts, there needs to be some way of ensuring that  everybody knows what a specific symbol represents. They would have had to talk about it. So if we look for evidence of symbolic  thought, we can assume they also had language.

Mesopotamian cuneiform developed  out of a system of physical tokens,   which were used by merchants  to keep track of merchandise. Each different type of token  corresponded to a specific item,   like a jar of oil or a basket of grain. That’s symbolic representation, so we  can say that people had to have used   language as long ago as the oldest of these  tokens existed, which was 10,000 years ago.

But that’s where our ability to directly  translate symbols pretty much ends. Before that, we have things that we  believe are linked to symbolic thought,   but we don’t necessarily know  what the symbols represent. There’s plenty of examples of  objects that had no literal   utilitarian function, but we still find  them in places where humans hung out.

There are these geometric patterns at a   site called Blombos Cave in South  Africa which are 75,000 years old. A similar decorative pattern is found  over and over again at that site,   suggesting that it meant something  to the people etching it. And in Morocco, researchers have found shells   with little holes drilled into  them that are 82,000 years old.

The holes suggest the shells were  worn as beads, or, ornamentation. Unfortunately, those are pretty much the  oldest physical artifacts of symbolic   thought that researchers can agree on, and the  evidence going further back just gets murkier. Like, there’s this pigment made from clay  called ochre, which had functional uses,   but also could have been used to  color things for symbolic reasons.

Archaeologists have found flakes of pigment in   layers of dirt dating back  as far as 400,000 years ago, but it’s tough to say which things  were colored for functional purposes,   and which of them were symbolic. That’s about as far as we can  get with just looking at stuff. So now let’s turn to the evidence we get  from language-users, and talk about anatomy.

But before we get into that, we’re going  to use our language skills on a quick ad. This SciShow video is supported by JMP,   a statistical analysis software designed  to help you with all your analytics needs. If you’re working with a data set that has  what feels like way too many variables,   you could use JMP’s predictor screening  to highlight the most meaningful variables   and avoid wasting your energy on the other ones.

You could also untangle which of  those variables have the biggest   impact on your dataset using  JMP’s multivariate approaches. Principal Components Analysis and Partial   Least Squares regression sort  out those questions for you. There’s no need to struggle through data  that you can’t make heads or tails of.

JMP is here to help. And you can test it out with a 30-day free  trial for anyone, anywhere at jmp.com/scishow. Most of the languages in the  world are spoken out loud.

And speaking requires breath control,   and that requires fine control of  the muscles from a bunch of nerves. Fossilized remains show that the duct that  those nerves passed through is much larger   in Homo sapiens and Neanderthals  than it is in earlier species, suggesting that early humans and  Neanderthals had greater control   of their lungs and were better  equipped to produce speech. The vocal tract is also critical for speech.

The human vocal tract includes our voice  box that’s about halfway down the throat,   the sinuses and palate, and our round tongue. And the location of that voice box  raises some evolutionary questions,   because while that’s good for speech, it’s  also bad if you don’t want to die from choking. Being able to produce sounds isn’t enough, though.

You also need others to be  able to hear those sounds. So some archaeologists have looked at the  bones of the inner ear that let us hear. These bones are tiny, so  they can be hard to recover.

But fortunately, newer discoveries  of well-preserved ear bones give   us a peek at how they’ve changed through time. Modern humans have ear bones that are  especially sensitive to the frequency   range of human speech, which suggests that  they were adapted to listening to others talk. But there’s also evidence that Neanderthals  had these same changes to the throat and ears,   which has two tantalizing explanations.

Either Neanderthals independently evolved the  ability to chit chat with their neighbors,   or that our ancestors were ready to talk and  listen before our two lineages split apart,   which was over 400,000 years ago. So physically, most of the requirements for  speech production were in place by that time. That’s all on the production part  of language, or at least speech.

But you don’t necessarily  need speech for language. Gestures and sign languages are just as  linguistically complex as spoken ones,   and would have required a  lot of brainpower to use. Which means that the evidence of ear  and throat changes might not be able to tell   us the furthest possible origin date for language.

But what we can look at is brains. Or, the closest thing to them  that we get in the fossil record. Endocasts are fossilized impressions of the  inside of the skull, which let researchers study   the shape of various regions of a brain even  when that brain has long since decayed away.

They’re not perfect, since they can’t really tell  us how specific neurons connected with each other   or sent signals, or what the structures  on the inside of the brain looked like. But they are a starting point. Some researchers have considered brain volume  when looking for signs of language capacity.

If we assume there’s some minimum size threshold  for language, we have to assume that we meet it. So if a species had brains that were the same size   as ours, we could say that it was  possible they used language too. That keeps Neanderthals in the  running, but it also means that   an even older hominin like Homo erectus  could have gossiped around the campfire,   as far back as 2 million years ago, even if they  were using more of a gestured language to do so.

It’s important to note two things here. First, brain volume doesn’t totally  correlate with intelligence. And second, being really intelligent  doesn’t necessarily mean you have language.

That’s a specific type of smarts. So instead of looking at the whole brain,   other research has focused on areas that we  know are needed for language in modern humans. Fortunately, two of the most famous language  areas in modern humans, Wernicke’s area and   Broca’s area, are right on the surface of  the brain, so we can see them on endocasts.

Looking at our extended family,   regions that are physically similar to these  two areas show up in great apes and baboons, which suggests that this particular feature was  actually present before apes and Old World monkeys   split off from each other, somewhere  in the range of 20-30 million years. But those species don’t have the  ability to use their own languages,   so obviously those other animals  aren’t using them for speech. So it’s likely that they evolved for other  purposes, and then, at some point, got reassigned.

In apes, these areas may be related  to gesturing and vocalizations,   forms of communication but not language. So that’s score one for the “language was a   byproduct of evolution of unrelated  traits” folks I mentioned earlier. At this point, we’ve narrowed the  development of language down to   somewhere between 2 million years ago, when  brains evolved to be at least as big as ours,   and 82,000 years ago, when  ornamental beads show up.

Which… it’s not nothing, but it’s such a big  time span that it doesn’t really tell us much. Fortunately, there’s another  area to explore: Genetics. Researchers have identified a gene  that appears to code for some aspects   of vocal production and learning  in a bunch of vertebrate species.

For instance, we have a mutation in  a speech-related gene called FOXP2   that sets us apart from other living species. But this same mutation was  also present in Neandertals,   meaning that it’s at least 400,000 years old. But the mutation alone doesn’t tell us  anything about the cognitive aspects   of language, since it’s more about  vocalizing than it is about grammar.

So instead of looking for relationships  between genes and a specific behavior,   in 2025, an international team  of researchers decided to look   at the genetic relationships  between groups of early people. And instead of using archaeological evidence   to try to pinpoint when language  developed, they used migration. As I’ve mentioned, language is uniquely human.

Which means that, outside of some very specific  circumstances with individuals who have cognitive   or physical differences, language is something  that all humans have an innate ability to use. Every single human community on earth  has a language, and some have multiple. That’s more than 7,000 languages  with complex rules for how words,   sentences, and meanings are constructed.

This suggests that language, or at least  the capacity for it, developed back when   Homo sapiens were living in close quarters,  before we’d started migrating out of Africa. All modern humans are descended from those  different waves of migrants, so all the   traits that we have in common across the whole  species most likely existed before we left Africa. Researchers are pretty sure that the first  group of people to separate from the rest   of Homo sapiens were the ancestors of the  Khoisan people found in Southern Africa.

Genomic data allows us to estimate when they  split from the rest of the population by looking   at DNA and identifying the places where there  are differences from other groups of people. See, mutations in DNA happen at a pretty  consistent rate, so by comparing how many   have built up between two groups, scientists  can estimate how long ago the groups separated. And that let the researchers estimate that the  separation happened roughly 135,000 years ago.

Now, this doesn’t mean that  language evolved 135,000 years ago. It means that the capacity for  language was in place by then. We don’t have a ton of evidence of symbolic  behavior that suggests we were actually using   language at that point, and we won’t confidently  see that for at least another 50,000 years.

So one possibility is that our  ancestors evolved the physical   structures for language back before Homo  sapiens and Neandertals split apart. Those cognitive skills continued to  develop in the Homo sapiens lineage,   and we had basically the full toolbox by the time  some of us hit the road to the rest of the world. And then somewhere around 80,000 years  ago, we put it all together, and boom.

Language and modern human behavior. But that’s just one idea of  what might have happened. So while we may never know exactly when  our ancestors acquired the gift of yapping,   that won’t stop us from talking about it. [♪ OUTRO]