Post History

Not to downplay the examples asked about, but when I look at a few somewhat randomly sampled languages not covered by the collection, I encounter mostly quite different forms.  For example:

* Finnish: kieli
* Mandarin: 舌头 [shétou]
* Turkish: dil
* Basque: mihi
* Inuktitut: ᐅᖃᖅ [uqaq]

[Here is a map](https://beautiful-basque-country.tumblr.com/post/750343530207199232/mapsontheweb-the-word-for-tongue-in-various) showing some of the diversity close to where I live.

A single lexical item is not enough to base entire language superfamilies on; if you have to consider over 10,000 years of language evolution, there will be not only inheritance and shifts of sound and meaning within the individual subfamilies, but also a massive opportunity for horizontal contact to account for.

(When a [similar question](https://linguistics.stackexchange.com/questions/48206/could-the-initial-d-in-the-word-for-tongue-be-originally-a-prefix) was asked previously at another site, specific potential horizontal contact between Latin and Etruscan was hinted at in an answer.  But my point isn't about Etruscan.  My point is about language contact in general.  And archeogenetics is only starting to connect reconstructed proto-languages to known excavations; prior to that, prehistorical language contact opportunities were exceedingly difficult to reason about with any confidence.)

Time for a minor semantic nitpick:

The question mentions three "English" meanings: "tongue", "lip", and "language".  In English, "tongue" and "language" are close to each other both semantically and etymologically, with "lip" standing apart, but quite a few languages of the world actually have "lip" and "language" as homonyms, with a shared etymology.  From my perspective, there seems to be a very widely encountered tendency for the metonymic connection between an organ of speech and "language", which could be explained by mouths predating typewriters, but whether the anatomical correlate is a tongue, a lip, or lips as a pair, or the whole mouth, seems less universal.  Even in the question's own examples, Bunuban "djälän" could, depending on the specific nasal consonant represented, mean "tongue" or "mouth". So I'd advise not to take "tongue" and "language" as having a universal/privileged semantic connection during research of this kind.

After my own admittedly limited exercise, I feel that some languages of the world have somewhat similar words for "tongue" between themselves, and others less so.  As far as I can tell, the cross-family distribution could even still be mostly accidental or horizontal rather than genetical, with each well recognized language family having its own independent story involving both internal evolution and external contact.

There are 100+ still extant language families and a highly motivated linguist could probably check out how the phenomenon scales to a larger sample and also look at how cohesive any identified club is archeologically (in time, in space) and statistically.  (Is the Turkic "dil" a member of the club, or is it already too distinct?)  Proper sampling would however have to avoid the methodological error of making the sample - or the similarity threshold - fit a particular preconceived hypothesis.

**TL;DR:** I don't expect a single, cohesive reason behind this data.