In CX1 entering 'gu' in lang search for 'gujarati' language looks clear:

CX2 will display the entered 'gu' as though there is some other characters beneath it:

Note 1: As far as I could see, the issue is present for languages with Eastern Nagari script (Bengali - bn,
Assamese -as). The issue is not present for Devangari script languages and Javanese.
Note 2: there one more minor issue - when search input is deleted manually-one letter at a time, some text will be displayed momentarily flashed under "Search for a language" pre-filled text. But this issue is present in both cx1 and cx2:

• enter one letter in 'Search for a language' box
• manually delete that letter (by backspace/delete button)
• some text under "Search for a language" will be displayed