the mouth movements in this game are just random, different types of mouth movements that don't sync up to any audio. This is also true for your character whenever you type something in linkshell etc. You say something like "hi!" and your chara makes about 5-6 extra syllables. lol

So really, in the voiced cutscenes, they're moving their mouths much in the same way they are in the un-voiced ones...randomly generated movements.

I don't work with SE, but being familiar with the animation industry, I can almost certainly say that this was done for simplicity sake. This isn't a console release, like FFXIII, where the English localization actually went back into the motion capture studio to re-record mouth movements to match the English audio. This is an MMO, and as such, they cut corners where they can.

If they did as requested and synced the mouth movements to the audio, which one would it be? Remember, everyone in the world uses the same game files, just different "text" based on the language of choice. If they sync to JP, English, French and German lips would still be off...etc for the others.

Syncing all 4 languages would mean much more development time...heck just doing motion capture for ONE language would be time consuming enough. They would have to go in and actually change the game files...meaning each language would need to have their own client..or have extra files in there somewhere. It's not as easy as just selecting the audio option in your settings. Pretty sure this is why they left it the way it is currently....with random mouth movements that don't match up to any language.

Just a little tidbit: in version 1.0, the only selectable audio was English, with the subtitle text changing according to client language (even for the Japanese, imagine that!). Since only English was recorded, the lip movements DID match their speech. It's most visible in the starting cutscenes for the three cities (Gridania's has a few closeups of Yda when she's talking). Take a look on youtube if you're bored.

On the subject of actually animating lip movements themselves, this is a fairly simple process for a big company like SE. They need only set up their motion capture studio for facial capture, clean up the data, and apply it to the 3D models. If done right, there is hardly any manual adjustments that need to be made.