What do you want me to do?

[Kì'eyawn]

Kaltxì ma smuk.  So, this is a deceptively simple sentence in English, but i'm uncertain how to translate it into Na'vi.  Here is my attempt:

Pehem[ìri?] ngal new futa oe fìkem si?

What do you think?  Suggestions?  How would you translate it?

[Hawnuyu atxen]

Maybe:
Pehem ngal oeti new si?

It may be wrong (most likely), but about "si": i didn't added <iv>, because here it isn't the verb "you" want, but "me" (to do it).

[kewnya txamew'itan]

Ma hawnuyu, definitely not right, the <iv> is required even in the sentence you give.

You're also missing the point of si and how simple this sentence actually is:

nga new oe pehem sivi

[Hawnuyu atxen]

In this case i don't really see why... i know that verbs after modals have to have <iv>, but here the two doesn't have anything to do with each other... if it still needs the subj, than i learned something new today too

[kewnya txamew'itan]

Yes they do have something to do with each other.

in this sentence you don't want me, you want the action to be carried out by me so the <iv> is still very much an essential part of the phrase.

I think this is also one of the reasons new has the long form, it makes this a lot easier, in the long form we get:

ngal new futa oe pehem sivi

to make it the short form we just drop the -l and futa giving us:

nga new oe pehem sivi

[Hawnuyu atxen]

Okay, i get it now, irayo!

[eanayo]

Gah. Next time refresh the browser before posting.

Yeah, what everyone said, though I'd personally stick to the futa version to avoid any ambiguity.

[kewnya txamew'itan]

What ambiguity is there?

I can see that it is easier to figure out the grammar to use when saying it using the long form but I can't see any ambiguity in hearing the short form or any extra difficulty in translating when you see the short form.

[eanayo]

Well, since (who am I telling this, you're the expert ) word order doesn't give any information about who wants what...
What do you want me to do or
What do I want you to do.
Yes?

Granted, the latter is not terribly likely to be the actual meaning, but I like being precise

[kewnya txamew'itan]

Ahh, you need to read annis' explanation of this here.

Free word order doesn't mean free word order, only free constituent order. 

The sentence's constituents are clauses whose constituents are the verb, any adverbs, a subject and possibly a direct and/or indirect object etc., each object or subject has as its constituents the original noun and any genitives, adjectives or relative clauses. With modals, the subordinate verb and its clause is also a constituent of the first clause (taking a similar place to the object).

You're getting the ambiguity by mixing the clauses which can't be done. The phrase nga new oe pehem sivi can only be split into clauses as {nga new {oe pehem sivi}} and not nga} new oe {pehem sivi because otherwise you're leaving a clause split and open ended which doesn't seem to be allowed.

[omängum fra'uti]

You actually can "split" a clause.  Or really, embed a clause inside another.  (There have been Frommer examples but I can't think of any off the top of my head.)

Oel ikranit a {tsaw neto tswolayon} tse'a
I see the ikran which flew away.

In the case of "nga new oe pehem sivi" there's no grammatical way it couldn't be ambiguous.  But if it were written as "new oe pehem sivi nga" then it could be either "new {oe pehem sivi} nga" or "new oe {pehem sivi nga}" - but the second one seems odd, and contextually most likely only the first would make sense.

To understand why the phrase as written isn't ambiguous, lets write out the long form.

Ngal new fì'ut a {oe pehem sivi}

Ignoring case endings, there's no way you could split that such as anything but "oe" is "pehem sivi".

[NeotrekkerZ]

It's not a verbatim translation, but how about

Kempe nga new oeta?  What do you want from me?

Using ta as a suffix eliminates the ambiguity completely and is fairly succinct as well.

*if it sounds a little idiomatic, I suppose you could throw in tel:  Pehem nga new tivel oeta?  What do you want to receive from me?

[Kì'eyawn]

Tewti, irayo seiyi oe ayngaru nìwotx.  This has been very interesting to read this discussion.

I guess what confused me is that, while i know "si-verbs" can get split up to some extent, i didn't think i could break up "kem si" across clause boundaries—so i didn't know what to do, since "what (action") clearly needs to go with "nga new" but "do (action)" needs to be with "oe".  Thank you for all your help, ma smuk!

[omängum fra'uti]

It's not really splitting it across clause boundries.  Rather, if the word that "si" goes with can be understood from context, it can be left off.  In this case, it's pretty clear that it's going with the "pehem", so there's no need to repeat "tsakem".

[kewnya txamew'itan]

You actually can "split" a clause.  Or really, embed a clause inside another.  (There have been Frommer examples but I can't think of any off the top of my head.)

Oel ikranit a {tsaw neto tswolayon} tse'a
I see the ikran which flew away.

In the case of "nga new oe pehem sivi" there's no grammatical way it couldn't be ambiguous.  But if it were written as "new oe pehem sivi nga" then it could be either "new {oe pehem sivi} nga" or "new oe {pehem sivi nga}" - but the second one seems odd, and contextually most likely only the first would make sense.

To understand why the phrase as written isn't ambiguous, lets write out the long form.

Ngal new fì'ut a {oe pehem sivi}

Ignoring case endings, there's no way you could split that such as anything but "oe" is "pehem sivi".

With the nga afterwards then it could be ambiguous. But if you have it in the form nga new oe pehem sivi then you'd have to split the embedded clause which would be a no-no.

Edit: oops, misunderstood you.

It's not a verbatim translation, but how about

Kempe nga new oeta?  What do you want from me?

Using ta as a suffix eliminates the ambiguity completely and is fairly succinct as well.

*if it sounds a little idiomatic, I suppose you could throw in tel:  Pehem nga new tivel oeta?  What do you want to receive from me?

You don't really receive the action though do you?

I like you first one provided you used fa not ta, also you'd have to put nga in the ergative and kempe in accusative as new is being used transitively.

Kempet ngal new oefa looks ok to me (although still possibly a tad idiomatic).

[eanayo]

Ahh, you need to read annis' explanation of this here.

Free word order doesn't mean free word order, only free constituent order.  

Ok, I have been aware of the difference between word order and clause consistuent order (my apologies for using the incorrect terminology).
However, I was completely unaware that new automatically introduces a new clause. That explains why I thought it could be ambiguous.

Learn something new every day May I just ask if there is a source for this, and if there are other words (apart from the obvious a and tsnì and whantnot you have) that do such things? edit: nevermind. Reading the wiki helps. And now if you'll excuse me, I'll be hiding under a rock for the next couple of years.
Thanks for being so patient with me!

[NeotrekkerZ]

You don't really receive the action though do you?

Not necessarily, however, you do want the other person to perform an action, so in that sense you're "receiving" the person's cooperation.  TBH, I would never personally write it this way with tel, just wanted to present another option.

I like your first one provided you used fa not ta, also you'd have to put nga in the ergative and kempe in accusative as new is being used transitively.

Hmm, this got me thinking:  Are we supposed to use endings with question words?  I always viewed them as things that were not nouns in which case they wouldn't be subjected to the erg/acc rule.  I mean if you go completely literally, then yes, I'd agree; but in the completely literal case it is being used as a noun.  Do we have any attested examples of a question word with an ending? 

The reason I bring it up is I can't seem to think of a sentence with "what" where it isn't used transitively, yet I've never seen peut/peur/'upet/'uper, only peu/'upe.

[omängum fra'uti]

The only examples we have of questions from the corpus are, I believe...  Pelun/lumpe which tends to be adverbial in nature anyway, pefya/fyape which also tends to be adverbial in nature, and kempe which is being used with the auxiliary verb (kempe si nga).  So we have no examples of it being used in a transitive situation and not getting a case ending.  That said, it would be ambiguous without.

[kewnya txamew'itan]

You don't really receive the action though do you?

Not necessarily, however, you do want the other person to perform an action, so in that sense you're "receiving" the person's cooperation.  TBH, I would never personally write it this way with tel, just wanted to present another option.

I like your first one provided you used fa not ta, also you'd have to put nga in the ergative and kempe in accusative as new is being used transitively.

Hmm, this got me thinking:  Are we supposed to use endings with question words?  I always viewed them as things that were not nouns in which case they wouldn't be subjected to the erg/acc rule.  I mean if you go completely literally, then yes, I'd agree; but in the completely literal case it is being used as a noun.  Do we have any attested examples of a question word with an ending? 

The reason I bring it up is I can't seem to think of a sentence with "what" where it isn't used transitively, yet I've never seen peut/peur/'upet/'uper, only peu/'upe.

1. Whilst it is undoubtedly good to have a second opinion, there is no physical receiving so any use would have to be idiomatic so you'd be walking on eggs (if you'll pardon the Spanish idiom)

2. There is no grammatical reason not to inflect question words, for example, Spanish distinguish between where and to where so there is no reason not too. As omängum says though, other than with peu/'upe it isn't normally going to be necessary.

[NeotrekkerZ]

Yeah, that's probably why I never considered it with the kempe/peu.  I'll watch my "whats" more carefully from now on.