Ugric languages

[6] The consonant cluster *lm is in the Ugric languages mostly reduced to plain /m/ (e.g. PU *śilmä 'eye' → Hungarian szem, Mansi сам /sam/, Khanty сем /sem/).

A peculiar exception is the numeral '3', in which Hungarian (három) and Mansi (хурэм /xuːrəm/) point to an original cluster *rm, whereas the rest of the Uralic family suggests original *lm (Khanty холәм, Finnish kolme, Estonian kolm, Inari Sami kulma, Erzya колмо, etc.)

The reverse has also been suggested—Hungarian and Mansi retaining the original form of the numeral, whereas Khanty and all the Finno-Permic languages would have innovated /l/ for some reason.

The three Ugric varieties also share the lateralization of Proto-Uralic *δ to *l (as do the Permic languages), but it is possible this postcedes the emergence of retroflex *ɭ from PU *l in Khanty.

This includes both basic vocabulary, e.g. 'fire' (Hungarian tűz, Mansi таўт /taːwt/, Khanty тут /tut/) as well as more specialized terminology, particularly the word for 'horse' (H ló, lov-, M луў /luw/, Kh лав /law/) and related items such as 'saddle' (H nyereg, M нагэр /naɣər/).

A common derivational innovation is seen in the word for 'louse': Proto-Uralic *täji → *tä(j)-ktVmV → H tetű, M такэм, Kh тевтәм.

Holopainen (2023)[8] argues that many known loanwords and suspectable substrate vocabulary show too much irregularity in sound correspondences to be derived from a common Ugric proto-language, and may have been borrowed independently into Hungarian and Ob-Ugric, or even all three of Hungarian, Mansi and Khanty; while for others, it is unclear if they were actually innovated, or represent rather common retention from Proto-Uralic.