setStrength(Collator.PRIMARY) means to treat only differences in base letter as significant (a different to b, but same as a-grave and A). Accents and capitalisation are both ignored. On reflection, Collator.SECONDARY may be more appropriate - that treats letters as different if they are accented, but not if they are capitalised. On further reflection, I don't think the collator strength matters too much at all, as even a full-strength collator will order differences of accent and capitalisation within the same base letter (abCde comes between abcda and abcdz, for instance). Strength is more useful for equality tests.

As you can tell, I only pretend to be an expert at this stuff! The important thing is to use a Collator rather than using Unicode code-point order (which I suspect JEmplode is currently doing). And it might be better to use the user's Locale (Locale.getDefault()) rather than Locale.US...