WORDS IN MOTION
Follow a sound through time.
Trace reconstructed words down historical branches—or release a sound into an illustrative landscape of possible change.
DESCENT TREE
One root, several outcomes
STOCHASTIC LANDSCAPE
420 descendants from one frozen ancestor
Illustrative model · not calibrated forecasts
Words and short phrases both work. Deep time mixes sound change, analogy, boundary drift and social fashion; the other modes retain the narrower Proto-Germanic model.
LIVE INTERVENTIONS
Nudge the language while it is moving
Press play, then inject a social pressure at the current date. Change the input afterward and the same seeded history—including every nudge—will be applied to the new word.
PARALLEL DESCENDANTS
The same word in six possible worlds
Rule-based transformations · illustrative only
Historical view
A compact teaching dataset built around familiar Indo-European cognates. It preserves the broad descent and major sound laws while collapsing dialects, overlapping periods and many intermediate stages.
Possible-futures view
A transparent coupled-feature model. Each descendant carries changing levels of lenition, palatalisation, vowel reduction, consonant loss, tone and analyticity. The Word Lab adds a separate five-millennium sandbox for phrase boundaries, analogy, contact and social fashion, with interventions that alter later outcomes. Percentages and paths are demonstrations, not empirical forecasts.
Sources to extend it
Glottolog for topology; Wiktionary and specialist etymological dictionaries for forms; Index Diachronica and historical grammars for rules; calibrated transition models for probabilities.