Please note that this message does not really bear on my proposal, but only on lexical design and stylistic considerations.
On Fri, Aug 31, 2012 at 3:29 PM, And Rosta
<and.rosta@hidden.email> wrote:
I think this proposal is better than any earlier proposals.
One thing I need to ponder more is
plnaki "A uses I" => plnabeki "A does something to E using I"
I feel it would be better
plnabi "A uses I" => plnabike "A uses I on E"
The direct object of English verb {use} has the syntax & force of a patient so I share your feeling. The reason that I chose to represent {plnabeki} as I did here is because I anticipate that {plnabeki} (or another stem like it) will be frequently used in composed predicates with the force of {[do something] with} rather than {use} in sentences like:
la tcmpza le cnke ju li kksi plnabeki spfabe.
"The chimp is breaking windows with coconuts."
... in which, and ("" = informally speaking) the "instrument" {li kksi} is an "oblique" argument of the "clause", and the modifier {ju li kksi plnabeki} can be dropped without changing the pragmatic meaning a great deal:
la tcmpza le cnke spfabe.
"The chimp is breaking windows."
The idea that I have would be to maximize argument "parallelism" so that composed predicates like plnabeki and spfabe would frequently (but by no means always--much less necessarily) have both the same agent and same patient; "obliques" would be largely k1. The patient arg -be- in plnabeki is so obvious in this case that it could probably be elided, especially since with a parallelist approach {spfabe} would strongly suggest it.
But it would work fine with no loss or gain of syllables if you insisted that the "instrument" should be the formal patient of "pln-". Xorban is very, very flexible in this regard. We could coin *two* stems for {use}, too. In short it is largely a matter of style, and it's up to the lexicon designer to make these choices.
It doesn't matter much for that one predicate, but I'm thinking of parallelisms with:
VkV >> VbVkV
liquid X soaks Y >> A makes liq X soak Y
stuff X fills Y >> A makes stuff fill X
(patientive) ckceki "P is soaked with K"
-> ckcabe "A soaks P"
-> ckcabeki "A soaks P with K"
(patientive) tsneki "P is filled with K"
-> tsnabe[ki] "A fills P [with K]"
sharp thing X cuts Y >> A makes sharp thing X cut Y
"Sharp thing" feels a lot like an instrument here doesn't it? I doubt you could pull that off in most languages other than English and coincidentally Lojban.
(agentive) ktnabe "A cuts P"
-> ju li knli plnabeki ktnabe "A cuts P with sharp thing"
Xorban probably makes it worthwhile to judiciously add instruments to the argument structure of some action-predicates:
-> li knli ktnabeki "A cuts P with sharp thing"
And if you insist on calquing the original:
-> li knli ktnabibe "A2 makes a sharp thing cut P"
rough thing X abrades Y >> A makes rough thing X abrade Y
-> ju li rfsi plnabeki gskabe "A abrades P with rough thing"
-> li rfsi gskabeki "A abrades P with rough thing"
or
-> li rfsi gskabibe "A2 makes rough thing abrade P"
stuff X covers Y >> A makes X cover Y
(patientive) gcreki "P is covered with K"
-> gcrabe[ki] "A covers P [with K]"
light X shines on Y >> A makes light X shine on Y
***
(patientive) gscrkmeki "P shines on K"
-> li gsni gscrkmabeki "P shines light on K"
-> la slra gscrkmaki "The sun shines on K"
For each polyadic stem bcd-, the lexical entry for bcd- will specify the role of the explicit argument in bcdV.
For each triadic stem bcd-, bcdVkV would always mean that the Agent is omitted (right)?
An agent can be present without a -b- present, e.g. cskaki "A says K". -k- always marks the k-arguments k1, k2, ... in order.
But bcdVbV would be potentially ambiguous: which of the two nonagent arguments does the second V stand for? This too could be specified in the lexical entry for bcd-. However, if an extra C could be spared for this purpose -- -x- ([G]), say -- then bcdVbV vs bcdVxV vs bcdVkV would allow any of the three args to be implicit without ambiguity.