Good point, but while {bu} eats one word, *hypothetical cmavo* should eat one selbrisle.
(Of course assuming that a {ke}-{ke'e} bracket counts as selbrisle.)
That wouldn't parse here. Think about it in terms of a grammar:
jufra = sumti* selbri sumti* / sumti*
sumti = LE selbri
selbri = KE brivla* KEhE? / brivla
{lo blanu stizu} parses as:
jufra
|
+---------+
| |
sumti selbri
| |
+--+ brivla
| | |
LE selbri |
| | |
| brivla |
| | |
== ===== =====
lo blanu stizu
(regular Lojban: lo blanu cu stizu)
(Apologies for the cheesy ASCII tree, but anyways.)
{lo ke blanu stizu}:
jufra
|
sumti
|
+--+
| |
LE selbri
| |
| +--+------+
| | | |
| KE brivla brivla
| | | |
== == ===== =====
lo ke blanu stizu
(regular: lo blanu cu stizu)
Your {lo blanu broda broda} doesn't parse - it'd parse to this point:
jufra
|
+---------+------+
| | |
sumti selbri sumti
| | |
+--+ brivla LE
| | | |
LE selbri | |
| | | |
| brivla | |
| | | |
== ===== ===== =====
lo blanu broda broda!?
That doesn't match (as you can see), since {broda} isn't among the selma'o BRIVLA
(let's pretend BRIVLA is a selma'o).
If we add NU, then your proposed thingy would work in context.
It's also possible for an unmatching brivla to start a new bridi, with the {.i} elided:
selsku = (I* jufra)*
Then:
selsku
|
+----------------+
| |
jufra jufra
| |
+---------+ selbri
| | |
sumti selbri brivla
| | |
+--+ brivla |
| | | |
LE selbri | |
| | | |
| brivla | |
| | | |
== ===== ===== =====
lo blanu broda broda
(regular: lo blanu cu broda .i broda)
Hope it makes sense.
~ mi'e la uakci mu'o re'i
.i .e'e ma'a jbokei gi'e nai jboke'i ~