Supporting 10 different languages in regex is a drop in the ocean. The regex can be generated programmatically and you can compress regexes easily. We used to have a compressed regex that could match any placename or street name in the UK in a few MB of RAM. It was silly quick.
I think it will depend on the language. There are a few non-latin languages where a simple word search likely won't be enough for a regex to properly apply.
woah. This is a regex use I've never heard of. I'd absolutely love to see a writeup on this approach - how its done and when it's useful.