mungojelly@ixkey.info wrote:
coi ro do .uiI just got the clue that the UI2 (pe'i/se'o/etc) and UI3 (mu'a/la'a/etc) are some sort of UI! They bind to the thing to their left, then, eh?.i mi pu viska lo nu ko'a za'a la djan. ti'e tavla I saw ko'a talking to someone who I heard was John.So I can say "ko'a za'a" I observed that it was ko'a and also "la djan. ti'e" I know through hearsay that it was John. Right?mi'e bret.
I would interpret this as saying that you saw someone talking to someone else, and that you observed that the talker was someone previously mentioned in the discussion (ko'a), and you perhaps heard that talker label the other person as "djan" and thus infer that ko'a was talking to John.
Yes these are all UI, and potentially can be mixed up in all sorts of creative ways, many of which have not yet been explored. The groupings into UI2, UI3, etc, was to try to keep together those words that had similar intended function, and thus apply a suborganization to the UI list in the cmavo. But they are all UI.
Remember that the gismu and cmavo lists were originally intended as inputs for LogFlash, and thus we wanted people to learn similar/contrasting words together. They weren't going to learn all of UI in one sitting.
Members of UI attach to whatever is on their left, assuming it has any semantic weight itself. Thus groups of UI join to each other, and to whatever is on the left of the UI. If what they attach to is a marker of a grammatical boundary (a terminator, or ni'o, or .i and a few others) then they apply transitively to the entire marked grammatical structure.
The internal structure of a string of UI is usually undefined, but a couple of the UI words have been specifically defined to modify the UI word immediately to their left, if they appear after a UI.
lojbab