I feel like I have to comment at least a little on Na'vi, the language of the Na'vi people in the new movie Avatar. I haven't seen the movie yet, but I'm quite interested to do so soon, though I'm assured it does not revolutionize cinema as much of the press seems to claim. Paul Frommer was in charge of creating the language, and recently did a guest post on Language Log about some linguistic aspects of Na'vi. Since he's the creator, I see no reason to give a summary of the language (his description is well worth reading), but there are a couple things I can comment on.
One thing I like about Na'vi is that it uses ejectives. I don't like it for some scientific reason like "most languages have ejectives" (only about 15% do); I just like ejectives. They're fun to pronounce. I'm a bit nonplussed as to why Frommer chose to represent the glottalization with an "x", though: /p'/ is represented as px, /t'/ as tx, and /k'/ as kx. My guess would be that it makes the language look alien, which is as important a consideration as any when you're coming up with an alien language for Hollywood. Clearly Frommer has put a lot of thought into Na'vi; he even goes into restrictions on syllable structure. And these aren't just any random restrictions, but logical ones actually utilized in many natural languages. He remarks that only /f/, /s/, and /ts/ can appear as the first member of a consonant cluster. Now, I'm a bit dubious about the naturalness of this class in terms of actual occurrence, but at least theoretically it makes good sense; it's an exhaustive collection of the language's voiceless fricative phonemes (of course, /ts/ is not a fricative, but we can lump it under an ad hoc collection of "fricative phonemes" if we assume that affricates display edge effects, and since these sounds are the first members of clusters, the relevant edge of /ts/ would be the /s/ part). On the other hand, I don't think there really are any languages that do this. Some languages do allow only fricatives as the first members of complex clusters, but usually this is a class like /s/ and /hl/ (the lateral fricative; don't make me dig up my Unicode chart, I'm using Haida practical orthography), as in Haida. This is perhaps a more natural class because these are both coronal fricatives. Frommer's generalization is certainly theoretically warranted by some assumptions, because there seems to be something special about voiceless fricatives and clusters. But (as I argued in my M.A. thesis) it seems to be something special about coronal voiceless fricatives; I don't think we should expect to find, e.g., an extrasyllabic /f/ at the beginning of a complex cluster as we find extrasyllabic /s/ (claimed for English by Roca & Johnson, claimed for Blackfoot by me). On the other hand, for those who don't aspirate the /t/ in "fifteen", there's always the question of whether they syllabify it as fif.teen or fi.fteen.
Hopefully none of this seems like criticism of Frommer's language, because I certainly don't mean it as such. In a world populated by underdeveloped, Indo-European-influenced conlangs, it's nice to see someone as knowledgeable and dedicated as Frommer take the time to give us an artificial language that's interesting. I'd be interested in finding out a bit more about Frommer himself. All I can find is that he's in the business school as USC, and that he's referred to as a "linguist", but I haven't been able to dig up what linguistic research he's done, or what he looked at for his Ph.D. or M.A. (both from USC), if indeed either of those degrees were in linguistics. There are a lot more interesting things about Na'vi that you should read about in his Language Log guest post, but I've already run on for a while, so I'll end it here.