Everybody already knows what I'm going to say but by golly it is the right answer: test which one performs better for your site. 2 + 2 = 4, in the time it takes two highly paid people to debate "Submit!" versus "Sign up" you can have the A/B test already running and be back doing productive stuff.
A/B tests are only useful when you can gather statistically significant amounts of data from them. For a lot of small websites, or infrequently used features on larger websites, that is not the case.
Without statistically significant data, there is no hope for a statistically significant margin gain in the arena of form submissions. That is, why not just switch up the language? It's a trivial commit and will be an obvious change in conversions or no change. It would have taken less time to test than this comment to be written; just do it -- the parent comment's point.
haha, this sounds like a piece of advice straight out of Starcraft 2. If you're anywhere but in the top 10% of ladder or above, you'll win more games and get better faster by just focusing on building more stuff than your opponent.
Unit composition, racial balance, specific attack/scouting timings, etc. profoundly don't matter if you can't scrape together a hundred food army and an expansion 10-15 mins into a game. If you can, you can start strategerizing.
