There's no written record. No audio recording. No living speaker. Yet linguists have reconstructed thousands of words from Proto-Indo-European — a language spoken roughly 6,000 years ago that nobody alive has ever heard. It is the ancestor of English, Hindi, Russian, Greek, German, Persian, Lithuanian, and many more — together representing about 3 billion native speakers today.

The reconstruction works through systematic sound correspondences known as sound laws. Compare the Latin pater, German Vater, English father, and Sanskrit pitar. All share a common origin, but the sounds drifted apart over millennia. Linguists trace each sound forward through systematic shifts: where Latin has p, Germanic languages reliably have f. Where Latin has t, Germanic has th. Where Latin has a long e, Germanic has a long i.

Run those laws backwards, and the parent form emerges: pəter — the reconstructed Proto-Indo-European word for father, marked with an asterisk by convention to indicate it is reconstructed rather than attested.

This methodology was first systematized in the 1800s by linguists including August Schleicher and the Junggrammatiker — Neogrammarian — school in Germany. The work has continued for two centuries. Modern linguists now have reconstructed Proto-Indo-European words for kinship, tools, animals, weather, and even abstract concepts. The current reconstructed lexicon includes more than 1,500 roots.

The reconstruction is detailed enough that scholars have written entire stories in PIE. One famous text — The Sheep and the Horses, composed by Schleicher in 1868 — has been revised repeatedly over the past century as understanding deepened.

Where the speakers lived is still debated. Most evidence — words for the horse, the wheel, kinship structure — points to the Pontic-Caspian steppe in modern Ukraine and southern Russia, around 4500 BCE. Their entire material world has been reconstructed from words left in living languages.