No text extract is returned ("extract": "...") when there is an abbreviation/definition in the text that has terminology that is translated by Lingo. If Lingo is disabled, text extract works as prescribed. Is this a bug? is
Topic on Extension talk:TextExtracts
I know it's a terrible workaround but you can temporarily disable and re-enable Lingo extension using cron or task scheduler. (i.e. comment out Lingo and remove comment tag)
My current setup is to add "//" (comment) in front of Lingo extension line every day 3am, before running the backup scripts.
I'm not familiar with Lingo and it doesn't even work with recent MediaWiki. You should probably discuss this with its developers, sorry.