I recall reading, years ago, about a study of words for numbers -- one, two, three, etc. The idea here is that such words should have remained rather stable during transitions from one language to another. And such is the case for Latin-based languages. But (if I recall correctly) it's totally different for distant language groups, e.g., Japanese.
Instead the *grammar* may have singular, dual, trial, paucal and plural instead of a simple singular/plural like English.
Here's Glenn Morton showing the distribution of **tik and **akwa ("water" - which I think may be onomatopoetic gurgling)
And here's a version of Ruhlen's list.
I find all this fascinating.