Relevanssi has problems with highlighting terms when non-ASCII alphabet are used. The highlights don’t work with Cyrillic alphabet, for example.
This is a problem with the regular expression I’m using to find the terms in the text. The regular expression uses word boundaries to find terms that are at the beginning or at the end of a word. That way a term that appears inside another word doesn’t get highlighted. If you search for ‘cat’, Relevanssi would highlight ‘catastrophe’ or ‘cat‘, but not ‘dedicated’.
However, the word boundaries are only recognised in ASCII characters (a-z, 0-9, pretty much). This means that Relevanssi can’t find any terms to highlight in Cyrillic text or other non-ASCII alphabets.
Relevanssi 2.8 added a feature to fix this (and Premium 1.4 will add it too). It’s a checkbox that eases the restriction a bit: if you uncheck it, Relevanssi won’t care about word boundaries. It’ll highlight every occasion of the search term, also inside other words. It’s not as clean as the original highlighting regexp, but it works.