# Talk:poi'i

## Discussion

• John Cowan:
• Can you provide a concrete example of such an abstraction, and an x1 that would make it true? I don't understand this.
• And Rosta:
• It's a utility.
• 1. It allows de facto prenexes without need for goi:
• mi viska la djan = la djan goi ko'a zo'u mi viska ko'a = la djan poi'i mi viska ke'a
• 2. It allows for reflexives:
• mi poi'i ke'a viska ke'a = I see myself.
• mi poi'i ke'a jinvi tu'o du'u ke'a melbi = I believe myself to be beautiful.
• 3. It allows sumti tail formation in cases that can otherwise be difficult to handle:
• le poi'i la djan jinvi tu'o du'u ke'a melbi = certain ones who John believes to be beautiful
• le poi'i ke'a viska ke'a = certain ones who see themselves
• The basic idea is a NOI converted into a NU.
• .djorden.:
I had trouble understanding at first because I missed that its selma'o is NU and not NOI. Maybe it should use a different cmavo than ke'a so you don't confuse those used to poi? (zo ke'ai zo'o)
(For the record I don't like or advocate this cmavo)
• And Rosta:
Such ambiguities already arise with NOI within NOI, and the official solution to that problem, viz xi-subscripting, could be used here, without having to create yet another cmavo. Indeed, when ce'u was created I was opposed to it, advocating the use of ke'a instead, and I hold to that view still.
• xorxes:
I agree that ke'a/ce'u are essentially the same thing. I used ke'u with ka before ce'u was created. I like the flexibility that poi'i affords, but I prefer seka for that meaning.
• pc:
Ummh, can someone explain to me what similarities there are between ke'a, a restricted anaphora, and ce'u a bound variable in abstractions?
• One way to see the similarity is to replace poi with pe sekai le ka (or noi with ne sekai le ka), so for example le broda poi ke'a brode is le broda pe sekai le ka ce'u brode. In both cases, relative-clause = NOI subsentence /KUhO/, tanru-unit = NU subsentence /KEI/, the respective KOhA keeps a slot in the subsentence open. One important difference is that multiple uses of ke'a collapse to a single slot, while multiple uses of ce'u correspond to different slots.
• Ahah! When spun out in certain contexts they give the same resultant claim, not that they are conceptually at all similar. Sorta like ka and du'u, without the grammatical similarities.
• And Rosta:
I agree. I think I must have proposed poi'i because ka has no x2 (and IMO nor should it). But that leaves se ka meaningless, and a reasonable interpretation it to take se ka as forcing a predicate that is is a verion of ka with an x2. Okay, then - see SE + x2-less brivla.
• .maten.:
Is there any circumstance in which poi'i can't be replaced by du DA poi (or du lo du poi), being only one or two syllables longer and non-experimental?
 mi du da poi ke'a viska ke'aI am one who sees themself.
 le du be da poi ke'a viska ke'aCertain ones which equal something which sees itself.
• xorxes:
• Yes, poi'i is approximately du (be) da poi, therefrom its form, but it doesn't use up a variable. Many cmavo can be paraphrased in terms of other cmavo, but it is still convenient to have the compact forms. In your example, a difference would appear if you were to write the sumti after the selbri: Compare poi'i ke'a broda kei fa ro de with du su'o da poi ke'a broda ku'o fa ro de. In the second case, ro de is within the scope of su'o da. (poi'i is probably fully equivalent to du (be) tu'o da poi.)
• mudri:
• Isn't this just new voi? I guess poi'i is better for backwards compatibility.
• la gleki:
• This is in fact the original cmavo. voi instead of poi'i was proposed much later.