Thank you very much, Taronyu, - again - for your quick response! I feel a bit ashamed about the million "thank you"s, so I'd like to send them back.
Of course: The part of speech for many of the words are undoubtful. My remark was just about words like "touch" or "wrong" or "dance" or "begin" and so on.
As a latin teacher, who studied classics (of course), I find it even a bit amusing, that in the development of the Na'vi-Vocabularies we can see the same problems, which always appear, when one tries to reconstruct an ancient text. There is always the primary source (in most times lost or unknown), than this source was copied (Copy A), than this copy was the source for further copies (Copy B) which were the source for other copies and so on. And every time a text is copied, it is (necessarely) altered, so that in our times it is a really hard job to reconstruct, what was in the original text, what Caesar or Cicero or Tacitus really meant.
As for the Na'vi-Dictionaries I think we have the following "Line of tradition":
Frommers notes only known to himself (mother source) --> Survival guide (primary source) --> Internet-list by "matt" (Copy A) --> Pocket guide (Copy B) --> your dictionary (Copy C)
One thing you can observe is, that, if one of the copyists doesn't share an information (on purpose or by mistake), the following texts will not include this information any longer (deletion). This is the case about all the words, which didn't make it from the mother to the primary source yet, and this is the case e. g. about the word 'awkx (cliff), which is in the primary source (right between 'aw and 'awpo), but not in the Copy A (I suppose, matt just copied not very carefully). And since then the word is lost (Copies B and C doen't include it).
The other thing is, that if a copyist alters the text (because he thinks, that he is correcting an error, or that it would be better for the readers to know also this or that), this alteration will be part of the further copies (augmentation). This is the case with all the stress-marks and syllable-dots, which were added to Copy A by Karyu Amawey. Now they are not only in his Pocket guide, but made it also in your dictionary.
The problem about all this is, that a person, who just has one of the copies has no choice as to take this copy's information for granted. But this information may be wrong (compared to the original text). And since we don't know much about the mother source (that means: Frommers grammarbook and dictionary - in case they exist), there is allways the problem, that someday we - hopefully - will get access to that mother. And than it maybe, that many informations, which we took for granted, will not fit to the mother.
So - as well as Prrton did often about grammar-points - I just want to stress out, that it is quite tricky to add information to the material we got and declare things to be so or so, when the only foundations of all this are (more or less plausible) guesses. People may take this things as granted and memorize them during their language learning, and have to UNLEARN things, when the mother will be accessibe. And unlearning things, which seemed to be clear, isn't that easy. So I would suggest more caution and precision in compiling offsprings, so that anybody can clearly separate the things we really know (because they were released officially) from the things we are just guessing.