Vohp posted some really good stuff in the thread on Socialese syntax and I thought it ought to be preserved, so I'm posting some of it here.
Jeff on the duration of proximate and obviative:
Assignment of the proximate/obviative persons is set for a "discourse segment" and stays constant throughout the segment; but what constitutes a segment can vary--it's a vague and amorphous term, by necessity; sometimes it can mean a single sentence, or series of related sentences, or an entire conversation; a discourse segment is what any given pair of speakers wants it to be. The proximate/obviative assignments can be changed at will, for any number of reasons, or kept constant.
Jeff on compound agents or patients:
I don't recall compound participants being an issue for me. When only one participant is compound, it's easy:
"Bill and Ted ate peyote."
For this, you'd mark the verb with a 3rd person plural animate agent and a 4th person singular patient; both Bill and Ted would be marked as 3rd/proximate--they're treated as a unit, as part of the same thing (compound subject/agent), with regard to the marking. The same would be true if the object/patient was compound instead--it would likely be obviative, and you'd mark it on the verb as a plural obviative patient (while obviating any free-standing words that represent the compound participants).
Dual compound participants is a little trickier:
"Bill and Ted killed John and Adam"
For this, you'd incorporate John and Adam as a compound object/patient into the verb; and mark Bill and Ted as described above.
These rules, or something like them, should allow you to handle any manner of complex subjects or objects.
The N. grammar has this to say on the subject (this was written sometime in 2004, according to a note):
There are two exceptions to the above prohibition on two animate or inanimate third persons in an expression. First, when two or more identical (in terms of animacy) third persons are part of a complex subject or agent--i.e., a subject or agent composed of more than one distinct entity--they will all be third person; obviation will not occur among them. Of course, as above, in the context of complex participants obviation may be reversed in an expression, so that the subject or agent is obviated over the object or patient when one or the other must undergo obviation; if this occurs, a complex object or patient behaves just like and is subject to the same rules as a complex subject or agent. Also, when any one member of a complex participant is obviated, all members will likewise be. Second, in cases of dual subject agreement with identical (in terms of animacy) subjects, both participants are considered to be the same, usually third person.
An animate and an inanimate person in a complex participant is treated as a plural animate with regard to the verb marking.
If you're wanting to use free-standing participants with a pronominal in a complex participant (e.g., "Ted and I"), you'd simply use a form of "we" here, and not use "Ted" as a free-standing nominal; you'd make Ted's identity and presence understood by context or earlier sentences.
Jeff on relative clauses:
Mohawk doesn't distinguish these at all, morphologically, except perhaps in using a different mood/aspect combination for them (the fromer would be aorist, since it's expressing something generally true, while the latter would be imperfective). "He who X" doesn't exist in Mohawk; it's more direct--"He Xes"; and it can encompass the meaning of the former. (A lot of Amerindian names get translated into English as "he who does such and such"; but in most cases, the "who" is an English addition and not present in the original.)
Complex sentences and syntax:
I can show you how Mohawk would handle that complicated sentence you used earlier (subsituting 'meat' for 'malt', because I don't know the Mohawk word for the latter):
tsi ehlal lawatakotelonhv~i. ne takohs latsinokelyo?s. ne tsinowe~ sako?owalak?s. ne o?walonka sawalaokanonhsakonye~?s sak.
'This dog made the cat afraid. The cat had killed the rat. The rat had eaten the meat. The meat had lain in Jack's house.'
(A more literal translation: 'This nearby dog cat-frightened. The cat had rat-killed. The rat had meat-eaten. The meat had in-the-house-(of)-lain Jack.')
Mohawk, being constrained in its clause-forming abilities, can't handle complicated clauses like the above as single sentences, not anymore than English can produce a single word like sawalaokanonhsakonye~?s 'it had lain inside his house'. A similar method would be used to handle a long possessive construction--it'd be broken down; there'd be no need for 20 or 30 different sets of pronominal and possessive markers. I've never seen or heard such a long possessive construction in Mohawk, but that's how it would be translated.
Mohawk word order marks topicality and importance; and it's certainly used rhetorically. But it isn't associated with grammatical relationships, as far as I can see; who is doing what to whom, and who belongs to whom, are marked by verb prefixes. When we say "free word order," we typically mean free with regard to expressing grammatical relationships; but there are other pragmatic principles at work in Mohawk syntax, as I've said before.
I don't think you're getting my point tho. -3sSEN and 3sSEN.poss don't do anything to mark the car as being possessed by the man, without needing to ASSUME that. I mean, consider:
Bob-3sSen ... Frank-3sSen ... car-3sSen.poss ... bike-3sSen.poss
There's no way to tell, just by the idea that the words agree in gender, number, and person, that it's Bob's car and Frank's bike, or vice versa.
I can only speak for the languages I know, but Mohawk or Cheyenne would handle this by obviating the possessed thing (Cheyenne) or switching the gender of one of the 3rd person singulars (Mohawk), so that, e.g., a 3S masculine nominal would match up with the 3S masculine possessive; they'd be identified as the same person. Rarely in a Mohawk or Cheyenne sentence are you going to have three or more 3rd (or 4th) persons of identical animacy, gender and number. In Cheyenne, you can theoretically have up to five distinct 3rd persons: an animate 3rd singular, an inanimate 3rd singular, an animate 3rd plural, an inanimate 3rd plural, and an obviative (4th person); Mohawk has three numbers and four genders, so the possibilities there are even larger. And in cases where there are multiple persons that fall into identical pronominal categories, remember what I said earlier: they'd break down such a complicated or confusing clause into smaller pieces. (It's likely that obviation in Algonquian and the multiple numbers and genders of Iroquoian evolved from a need to disambiguate 3rd persons; so the pronominal distinctions that these language families possess are well-suited to the uses and needs of the speakers--the languages have as many pronominal distinctions as they need for everyday conversation; when things get more complicated, there are ways to work around that, just as English has its share of work-arounds.) (Another strategy of Mohawk/Iroquoian here is to incorporate the patient or object into the verb; when you do this, you don't mark the incorporated participant on the verb with a prefix; this typically frees up a pronominal category you can then use elsewhere in the sentence.)