Arguably the largest single feature (after the loss of vowel length) in this family is palatalization. /k/ followed by a front vowel has caused three of the four languages to have phonemic split (note that /k/ continues to appear in other contexts in all languages). In Chele, /k/ > /?/, the /t/ representing a full fronting to the alveolar ridge or teeth. In Thele, this fronted sound is retained, but without the following palatal (which we might assume was originally there). With Shel, on the other hand, /k/ has become /?/. It is possible that this change had /?/ as an intermediate stage. Kelu has not had phonemic split, although the /ie/ diphthong (rather than /?/) suggests that elements of the changes are ‘bubbling under’ (of which, more in a second).

In all the daughter languages /li/ or /lj/ have become fully palatalized at /?/. In Kelu, this sound is long at boundary with a stressed syllable.

Two languages – Shel and Thele – have intervocalic voicing of /p/ to /b/, even across syllable boundaries. This is a classic example of lenition.

In Thele and Kelu, original /?/ has diphthongized to /ie/. This may be related to palatalization. It may be that the other languages also had this development earlier in the process.

Proto-Kelo had ‘full vowels’ in unstressed final syllables. Chele and Thele have maintained these forms. Shel, on the other hand, has either reduced the original vowels to schwa (with */-a/) or to nothing (with */-o/ and /-i/). Kelu has retained these vowels, but all have risen or (in the case of /-i/) diphthongized, the latter being due to there being no further opportunity to rise.

All languages reduce unstressed initial syllables to schwa.

It’s quite difficult for me to talk about potential relationships between languages, since they were created by me and I therefore know what I intended with them. However, here is an honest attempt.

Kelu has a number of ‘peculiarities’ which make it seem less closely related to the other three (although of those three it is perhaps most like Thele). Chele, Shel and Thele seem quite closely related, although Shel seems in some ways to be more closely related to one and then closer to the other. We could assume that the three languages lie on a continuum, with Shel in the centre, although we also have to accept that Shel is far less conservative morphologically than the other two. It’s quite difficult to represent this type of relationship on a family tree.
In terms of similarities with ‘real world’ language, the varieties shown resemble (although they do not replicate) changes in the Romance languages. Changes in Chele and Shel resemble Italian and French respectively, while Thele is similar to (although not the same as) Spanish (in Spanish, /k/ > /?/ (or in some varieties /s/), not /tj/, before front vowels, but I felt that the latter might provoke more interesting discussion). In a sense, Kelu is ‘like’ Sardinian, since the latter is far more conservative with many features than any other of the western Romance languages. I have made it more like Thele than Sardinian is like Spanish, however.

The changes in the final vowels in Kele resemble the Great Vowel Shift. Needless to say, Sardinian doesn’t have this set of changes; but I wanted a chain shift to be present.