Scott (Yvain) did this in his fictional world. For example:

• Assertions require probability statements. There’s no grammatically correct way to say “The dog is in the blue house” without adding some modifier for how certain you are of this (probably along the lines of “tautologous, near-certain, probably true, uncertain, probably false, near-certainly false). There is also the option of elegantly expressing particular numerical possibilities.

• Assertions about plurals require quantifiers. For example, it’s ungrammatical to say “Atheists break the law”. One has to clarify this by adding “all”, “most”, “at least one”, “a disproportionate number”, etc. It’s pretty hard to stereotype in Kadhamic without meaning it.

Incidentally, learning a new language isn't required for this.

One can, for example, adopt the habit of saying "I want X to work" or "I expect X to work" or "I would be happier if X worked" or "I would be happier if I expected X to work" instead of "X should work" while continuing to speak English.

Put differently: the habit of setting trigger-points around certain words ("should," "think", "want", "can", :will", etc.) to ensure that I actually know what I'm saying when I say them is useful.

Don't go overboard with that -- IIRC, extremely few people succeed in becoming fluent in Lojban. IOW, think twice before flouting a linguistic universal.

(it can mean "I want it to work", an 'ought' statement)

Not everyone would agree that “I want it to work” is a correct restatement of deontic modality. (The one I use when wanting to avoid the ambiguity of “should” is “it had better work”.)

This would likely also have the same effects as thinking in any foreign language.

That effect is due to the fact that you're forced to use your System 2. It probably disappears after you become too fluent in the language (for example, FWIW, I don't ‘feel’ that happening with English).

Also, +1 or ygert's suggestion to read The Language Construction Kit, and you may want to check out the resources I mentioned in my reply.

Certainly a ridiculous munchkin idea! It's a cool idea, although I would estimate that the actual difficulty of getting it working is very high. If you do manage, that would be quite awesome though. If you are serious about actually trying this, check out The Language Construction Kit . It's a pretty cool website giving tips and advice on language construction. Perhaps it could be useful.

Around two years ago, I tried devising a language for roughly this purpose. I concluded that it wasn't a worthwhile use of time; devising it is easy, but becoming fluent takes way too much time, especially since there's no corpus (or a very small corpus, if you use something like Lojban).

I write down and regularly review all my ideas, experiences, etc., and I've found it very useful to invent my own words (interspersed in normal English) for concepts that need annoying circumlocutions in normal English. I also use the derivational morphology of Esperanto and my own conlangs.

For an interesting example of a personal language created as a psychological experiment, see gjâ-zym-byn.

How does Korean relate to this? I speak it semi-fluently and none of those three things happen in it. I have however found its folding in of adjectives into verbs one of several useful toeholds for learning Lojban, though.