The modern field of linguistics dates from 1786, when Sir Willian Jones, a British judge sent to India to learn Sanskrit and serve on the colonial Supreme Court, realized just how similar Sanskrit was to Persian, Latin, Greek, Celtic, Gothic, and English (yes, he really spoke all of those). He concluded that the similarities in grammar were too close to be the result of chance. The only reasonable explanation, he claimed, was the descent of these languages from some ancient progenitor.
This ancestor language is now awkwardly known as Proto-Indo-European (PIE). It and the people who spoke it are the subject of David Anthony’s book The Horse The Wheel And Language 1. I picked up the book hoping to learn a bit about really ancient history. I ended up learning some of that, but this is more a book about linguistics and archeology than about history.
Proto-Indo-European speakers produced no written works, so almost all of their specific history is lost. The oldest products of their daughter languages – like the Rig Veda – date from well after the last speakers of the original language passed away.
Instead of the history that is largely barred to us, this book is really Professor David Anthony attempting to figure out who these speakers were and what their lives looked like, without the benefit of any written words. He does this via two channels: their language, and the physical remains of their culture.
Unfortunately, there is at least one glaring problem with each approach. Their language is thoroughly dead and there was (at the time of writing) no scholarly consensus on where they originated.
Professor Anthony is undaunted by these problems. It turns out that we can reconstruct their language and from that reconstruction, determine where they most likely lived. If both approaches are done properly, it should be possible to see archeological details reflected in their language and details of their language reflected in their remains.
The first problem to solve then is the reconstruction of PIE. How does one do this?
Well it turns out that all languages change in similar ways. The way we pronounce consonants often shift, with hard sounds sometimes changing into soft sounds, but very rarely the reverse. How we say words also changes. Assimilation occurs because we tend to omit difficult to pronounce or inconvenient middle syllables (this has led to the invention of contractions in English) and addition happens because we add syllables in the middle of difficult tongue movements (compare the “proper” and colloquial ways of pronouncing the word “nuclear” or the difference between the French athlète and the English athlete).
It would be very odd for an additional syllable to be added in an area where tongue movements aren’t particularly hard, or a syllable to be removed from a word that is typically enunciated. Above all, these changes are regular because they rely on predictable laziness.
Changes tend to happen to many words at once. When people began to hear the Proto-French tsentum (root of cent, the French word for 100) as different from the Latin kentum, they had to make a decision about how exactly it would be pronounced. They chose a soft-c, a sound Latin lacks, but that is easier to say. This change got carried over to every ts-, c-, or k-, that had previously made the same sound as kentum/tsentum, except those before a back vowel (like “o”), presumably because a soft sound there is actually harder to say2.
There’s one final type of change that Anthony mentions: analogy. This is where a grammatical rule used in a single place (e.g. pluralization with -s or -es) is expanded to encompass many more words or cases (most English nouns were originally pluralized with other suffixes, or with stem changes like “geese”; it was only later that people decided -s and -es would be the general markers of plural nouns).
If you have a large sample of languages descended from a historical language (and with Proto-Indo-European, there really is no lack), you can follow a bunch of words backwards through likely changes and see if they all end up in the same place.
If you do this for the modern words for “hundred” from many PIE daughter languages, you’re left with *km’tom (an asterisk is used before sounds where there is no direct evidence). All words for hundred in modern descendants (as well as dead ancient descendants that we know how to speak) of Proto-Indo-European can be derived from *km’tom using only well-attested to and empirically observed rules of language change.
(I occasionally got chills reading reconstructed words. It’s amazing how some words that our distant ancestors spoke thousands upon thousands of years ago are fairly well preserved in our modern speech.)
This is pretty cool, because it allows us to start seeing which words were common enough in Proto-Indo-European to be passed down to all daughters and which words were borrowed in.
With a reconstructed vocabulary of about 1,500 words, we can figure out some things that were important to Proto-Indo-Europeans. They seem to have words for relatives on the male side, but not the female side. This suggests that after marriage, the wife moved in with the groom. Less domestically, they seemed to have a word for cattle rustling, suggesting that they weren’t unfamiliar with increasing their wealth at the expense of their neighbours’.
That’s not all we can get from their words. Linguists also believe that Proto-Indo-Europeans had chiefs, who in turn had patrons. They worshipped a male sky deity and sacrificed horses and cattle to him. They formed warrior bands. They avoided speaking the name of the bear. They drove, or knew of, wagons. And they had two words that we could translate as sacred, “that which is forbidden” and “that which is imbued with holiness”.
(There are many more minor cultural touchstones scattered throughout the book. I don’t want to spoil them all.)
We also know the animals and plants they had words for. Reconstructed PIE has words for temperate trees, horses and cows, bees and honey.
These give us clues to where they lived, in the same way that knowing the words “shinney”, “hockey”, “Zamboni” and “creek” are spoken somewhere might help you make a guess as to where that somewhere is.
And while these words help us rule out the Mediterranean and the deserts, they don’t give us much in the way of a specific location without a when, which requires two different methods.
First, we can figure out the approximate death of Proto-Indo-European, the approximate century or millennium when it was entirely splintered into its daughters, by using what linguists have discovered about the rate of language change.
While most vocabulary changes rather quickly, making this a poor tool for dating very old languages, there are a group of words, the core vocabulary, that change much more slowly. The core vocabulary of any language is only a couple hundred words, but they’re some of the most important ones. Normally, core vocabulary includes the words for: body parts, small numbers, close relatives, a few basic needs, a couple of natural features or domesticated animals, some pronouns, and some conjunctions.
English, a prolific borrower, has borrowed 50% of its total vocabulary from the romance languages. It’s core vocabulary, however, is largely free of this borrowing, with only 4% of core vocabulary words borrowed from romance languages.
Core vocabulary changes by about 14-19% every thousand years depending on the language. It’s also known that once two dialects differ by more than 10% of their core vocabulary, they are more properly thought of as separate languages.
Here’s where written language comes in handy. By comparing written inscriptions with known creation dates in different daughter languages, we can make a guess as to when the languages diverged.
The oldest inscriptions in a PIE-derived language are in the Anatolian languages (which were spoken in what is now Turkey). However, Anthony chooses not to use these, because they entirely lack many grammatical innovations that are otherwise common in daughter languages. This leads him to believe that they split away much earlier than other daughters. The presence of later shared innovations means that at the time of the Anatolian split, Proto-Indo-European was probably still a living language and still evolving.
Better candidates are archaic Greek and Old-Indic, both of which have inscriptions dated to around 1,450 BCE. By comparing the differences in wording and grammar between these two and using known rates of change, Anthony dates the end of Proto-Indo-European at around 2,500 BCE. This means that after 2,500 BCE, it doesn’t make sense to speak of a single unified Proto-Indo-European language.
Second is the birth date, the other half of the critical window. To find it, Anthony looks for words that have a known date of invention, specifically “wool” and “wagon”. Getting broadly useful amounts of wool from sheep wasn’t possible until a mutation made sheep coats much larger. We know roughly when this mutation occurred, because sheep suddenly became a larger portion of herds around 3,500 BCE, displacing goats (which produce more milk). The only reasonably explanation for this event is the advent of wool producing sheep, which were very valuable as a source of clothes.
Similarly, wagons have left physical evidence (both directly and in preserved images) and that evidence has been carbon dated to 3,500 BCE3.
Since all Proto-Indo-European languages outside of the Anatolian branch have related words for both “wagon” and “wool” that show no evidence of borrowing from other languages, it seems reasonable to conclude that some form of the language existed when wagons and wool first began to reshape the pre-historic world. That means the language had to exist by 3,500 BCE.
There is, I should note, one competing theory that Anthony outlines, in which PIE and Indo-Hittite languages split around 7,500 BCE. This theory requires several unlikely things to happen however; it requires the word for wagon to evolve from the same verb meaning “to turn” in both branches (five similar verbs existed), it requires the PIE speaking people to disperse over all of Europe and become the dominant culture then (this would have been very hard pre-horse domestication, when material cultures were small and language territories tended to be much smaller than modern countries), and all of this would have to happen while material cultures were becoming very different but languages (supposedly) weren’t evolving.
Anthony doesn’t give this theory much credence.
With a rough time-range, we can begin looking for our Proto-Indo-Europeans in space. Anthony does this by looking for evidence of very old loan words. He finds a set coming from Uralic, which also has a bevy of very old loanwords from PIE4.
Uralic (appropriately) probably first emerged somewhere near the Ural Mountains. This corresponds well with our other evidence because the area around the Urals (where borrowing could have taken place) is temperate and home to the flora and fauna words we know exist in PIE.
The PIE word for honey, *médhu (note its similarity with the English word for a fermented honey drink, “mead”5), is particularly useful here. We know that bees weren’t common in Siberia during the time when we suspect PIE was being spoken (and where they were common, the people weren’t herders), but that bees were common on the other side of the Urals.
Laying it all out, we see that PIE speakers were herders (there’s an expansive set of words relating to the tasks herders must accomplish), who lived near the Urals but not in Siberia. The best archeological match for these criteria is a set of herder people who lived in what is now modern-day Ukraine and it is these people that Anthony identifies as the Proto-Indo-Europeans.
If this feels at all dry, I want to assure you that it wasn’t when I read it. I felt that the first section of the book was the strongest. Anthony provides an excellent overview of linguistics, archeology, and some of the crazy stuff he’s had to invent to help him in his studies.
For example, he believes that horses were ridden much earlier than was commonly thought, perhaps around or before 3,500 BCE. To prove this, him and his wife embarked on a study of how bits wear teeth in horses’ mouths, which culminated in empirical studies with a variety of bit types (including rope) done on live horses that had never been previously given bits, assessed using electron microscopy. The whole thing is a bit bonkers, but it has resulted in a validated test that allows archeologists to determine if a given horse was ever ridden, as well as vindication for Anthony’s chronology of domestication.
Unfortunately, a lot of the rest of the book was genuinely dry. There was a dizzying array of cultures inhabiting the Eurasian steppes in the period Anthony covers, each with their own house type, pottery type, antecedents, and descendants. Anthony goes through these in excruciating detail. It’s the sort of thing that other archeologists love him for – a lot of these cultures are very poorly described outside of Russian language publications – but it’s hard for a lay-person to follow. I may have pulled it off if I built a giant flow chart, but as it was, I mostly felt overwhelmed.
(Anthony has to go through them all to explain how PIE-derived languages ended up everywhere we know them to have. People of Europe don’t speak PIE-derived languages just because of Latin. Many people the Romans conquered spoke languages that were distantly related to the invader’s tongue. Those languages need to be accounted for in any theory about Proto-Indo-Europeans.)
This is disappointing, because the history started off so engagingly. Anthony outlines how the earliest ancestors of the Proto-Indo-Europeans had persistent cultural frontiers with hunter-gatherers on the Urals on one side and the farmers in the Bug-Dniester valley on the other.
The herding and farming economies required a moral shift from previous hunter-gatherer practices, one that would see agriculturalists harden their hearts to their own children starving, if the only thing that could assuage their hunger was their last few breeding pairs or their seed grain. This is the first time I saw someone lay out the moral transformation necessary to accept agricultural and having it laid out so starkly made it much easier to understand why not every pre-historic group was willing to adopt it.
(I had always thought the biggest moral change was accepting accumulation of wealth, but this one is, I think, more important.)
This is not to say that the herders and farmers were exactly alike; their different ways of life meant they were culturally distinct. In addition to their dwellings and material culture, they differed in funeral customs and probably in religion. Everything we know about early-PIE speakers suggest that they worshipped a sky god of some sort. The farmers who lived next door decorated their houses with female figurines, figures that never show up in any excavation of herder camps or grave sites.
I was also shocked at the amount of long distance trade and the wealth acquisition that was going on 6,000 years ago. There are kurgans (circular rock topped graves) with grave goods from Mesopotamia dating from that long ago, as well as one kurgan where someone was buried with almost 4 kilograms of gold ornamentation.
The herders and farmers didn’t live next door in harmony forever. Changes to their stable arrangement happened as a result of one of the Earth’s period historical climate fluctuations (which caused a collapse among many of the farmers and may have led to more raiding from the early-PIE speaking herders) and later the adoption of horse-riding (which made raiding easier) and wagons (which allowed herders to bring water with them and opened the inner steppes up to grazing).
Larger herds and changing boundaries led to clashes among the herders (we’ve found kurgans where the bodies bear marks of violent deaths) and to raids on agriculturalists (we’ve found burned villages peppered with arrows), although interestingly, never the farmers directly adjacent to the steppes. It may be that the herders didn’t want to disrupt their trading relationships with their neighbours and so were careful to raid dozens of kilometers away from their own borders (a task made easier with horses).
The farmers were no pushovers; some of their towns held up to 10,000 people by the third millennium BCE. These towns were bigger than the cities of Mesopotamia, but lacked the civic organizational features of the true cities of the Fertile Crescent.
And it was at about this point in the narrative where the number of cultures proliferated beyond my ability to follow and I began writing down interesting facts rather than keeping track of the grand narrative.
Here are a few that I liked the most:
- About 20% of corpses in warrior graves (those with weapons and other symbols of membership in warrior society) whose gender is known are female. This matches the percentage in much later steppe graves. As Kameron Hurley said, women have always fought.
- Contrary to popular stereotypes, the cultures of the Eurasia steppes weren't reliant on cities for manufactured goods. They had their own potters and metalsmiths and they made many mining camps. In fact, by the 2000s BCE, it seems that Mesopotamian cities were dependent metal mined on the steppes,
- In the early Bronze Age, tin was worth its weight in silver. When tin wasn't available, bronze was made with arsenic.
- Horses were probably domesticated because they winter better than the other animals that were available in Eurasia at the time. Cows will starve to death if grass is hidden by snow, while sheep and goats use their nose to move snow off of grass (which means that they're helpless once it's covered in ice). Sheep, cows, and goats are all unable to drink water that is covered in ice. Horses break ice and move snow with their hooves, making winter no real inconvenience to them. Mixing horses with cows can allow cows to eat the grass that horses uncover.
- Disaffected farmers may have been attracted to the herding economy because wealth was much easier to build up. Farmland is hard to acquire more of without angering your neighbours, but herds given good pasture will naturally grow exponentially. A lot of the spread of the herding economy into Europe probably used some sort of franchise system, where locals joined the PIE culture and were given some animals, in exchange for providing protection and labour to their patron.
I’ve struggled through a lot of books that are clearly meant for people more knowledgeable in the subject than I am. It might just be a function of how interested I am in archeology (that is to say: only tolerably interested) that this is the first of them that I wish had an abridged edition. If you aren’t deeply interested in archaeology or pre-history, there’s a lot of this book that you’ll probably end up skimming.
The rest of it makes up for that. But I think there would be market for Anthony to write another leaner volume, meant for a more general audience.
If he ever does, I’ll probably give it a read.
David Anthony is very sensitive to the political ends that some scholars of Proto-Indo-European have turned to. He acknowledges that white supremacists appropriated the self-designation of “Aryan” used by some later speakers of PIE-derived languages and used it to refer to some sort of ancient master race. Professor Anthony does not buy into this one bit. He points out that Aryan was always a cultural term, not a racial one (showing the historical ignorance of the racists) and he is careful to avoid assigning any special moral or mythical virtue to the Proto-Indo-Europeans whose culture he studies.
White supremacists will find nothing to like about this book, unless they engage in a deliberate misreading. ↩
This is why the French côte is still similar to the Latin costa. ↩
Anthony identifies improvements in carbon dating, especially improvements in how we calibrate for diets high in fish (which contain older carbon, leading to incorrect ages) as a major factor in his ability to untangle the story of the Proto-Indo-Europeans. ↩
Uralic is the language family that in modern times includes Finnish and some languages spoken in Russia. ↩
While looking up the word *médhu, I found out that it is also likely the root of the Old Chinese word for honey, via an extinct Proto-Indo-European language, Tocharian. The spea ↩