* Saturday, 2014-10-18 at 22:27 -0300 - Jorge Llambías <jjllambias@gmail.com>:
> On Sat, Oct 18, 2014 at 8:36 PM, Martin Bays <mbays@sdf.org> wrote:
> > But it still feels very counter-intuitive to me that these things which
> > look like functions shouldn't just be functions.
>
> There's no doubt that they are functions, the question is whether they are
> primitive or derived.
>
> And if we take them as primitive, the strange thing is that they form
> a small closed class, so you could not do with most functions what you
> could do with them.
Could you rephrase that? I don't see what you mean.
But anyway: technically the class isn't actually closed, because of {na'u}!
> > li no pi'i mo'e ro namcu du li no
>
> ("cu" before "du" is so often forgotten, that it might be worth considering
> moving "du" to its own selma'o with the grammar of "cu du". One drawback
> though would be that we'd no longer have "lo du", or at least it would
> require additional changes.
Putting it (and maybe {me}?) in selbri2 would make some sense, and would
preserve {lo du}. But maybe {du} has reasonable uses in tanru; there are
a fair few lujvo using its rafsi.
> > li xy mleca li re .e se ni'i bo ci
>
> BTW, do we know what the expansion of "broda .i jek tag bo brode" is? Is it
> something like "broda .i jek brode .i je lo du'u/nu broda cu brodi lo
> du'u/nu brode" with a suitable brodi?
Something like that, I'm sure.
We can get at that brodi using {xo'i} (a neat experimental cmavo due,
iirc, to selpa'i): it's {xo'i [tag]}, or {se xo'i [tag]} if the tag is
a tense.
(So e.g. {xo'i bai} ~= {bapli}, {xo'i ba} ~= {se balvi}, and {xo'i fi'o
broda} ~= {broda} for a binary broda.)
It's important that the events/facts which are the arguments to {xo'i
[tag]} correspond precisely to the propositions being connected. So just
repeating {broda} and {brode} isn't quite right.
There are also scope questions. Probably we don't want
{ro danlu cu jbena gi'e ba bo morsi} to be equivalent to
{ro danlu cu jbena .i ba bo ro danlu cu morsi}?
If the logical connective isn't {je}, probably the tag relation should
only apply when the connectands are both true, i.e. corresponding
events/facts occur/hold. This seems sensible, and is supported by CLL
Chapter 10 Verse 17.10.
Firstly, we need a way to assign an event/fact variable to a proposition
without changing its semantics. {fi'o du} lets us do this; {fi'o du ko'a
broda} means that broda occurs/holds and ko'a is equal to the
event/fact of this. Let's make this a primitive in the logic, writing it
as "=.", so e.g. "{ko'a}=. broda()". (So technically "[term]=." is
a modal operator.)
(I write the above paragraph as if I'm sure it makes sense, but I'm not.
If there are many events of brodaing in the situation, is {fi'o du ko'a
broda} true when ko'a is any of those events, or only when it's the
"intended" one? The below makes sense in either case, but with subtly
different results. If instead tags work such that {fi'o du ko'a broda}
isn't true for *any* ko'a in such a situation, then there's a problem!)
Then we can handle tagged conjunction by quantifying over events:
ro danlu cu jbena gi'e ba bo morsi ->
FA x1:(danlu(_)). EX x2. (x2=. jbena(x1) /\ (ba)(x2). morsi(x1))
ro da poi ke'a danlu ku'o su'o de zo'u ge fi'o du de zo'u da jbena
gi ba de zo'u da morsi
, which is I think a natural and useful way to interpret it.
To be clear, that's intended to be equivalent to the following more
symmetric form:
FA x1:(danlu(_)). EX x2. EX x3. ((x2=. jbena(x1) /\ x3=. morsi(x1))
/\ {ba}(x2,x3))
ro da poi danlu ku'o de di zo'u ge ge fi'o du de da jbena gi fi'o du
di da morsi gi de xo'i ba di
Now for connectives other than conjunction, I'm not seeing a neater
solution than to "skim off" the TT case (if there is one); so e.g. to
consider {broda .i na ja ba bo brode} to be equivalent to
{ na broda .i ja broda .i je ba bo brode}, where the tagged conjunction
is interpreted as above.
Not very pretty. I'd be happy to hear of alternative possibilities.
(When working on tersmu, I somehow convinced myself that the above
handling of tagged conjunction would go through unchanged for other
tagged connectives. So that's what's implemented. But thinking about it
now, I see that this is quite wrong.)
> > The first one is also true with the relational semantics, but I'm not
> > sure it expresses the same thing (I'd translate it as "the thing which
> > is 0 times anything is 0", so the fact that such a thing exists becomes
> > a presupposition rather than a statement).
> It's odd however to have to read "li no pi'i mo'e ro namcu cu du li no"
> differently than "lo pilji be li no bei ro namcu cu du li no"
Is it? I don't know.
To analogise: I think the english pair
"zero times any number is equal to zero" and
"the product of zero and every number is equal to zero"
conveys the two meanings, using roughly the same structures.
> > Is there a reason not to declare that {na ku zo'u tu'e broda .i brode}
> > is equivalent to {na ku zo'u ge broda gi brode}?
>
> In natlangs we have the option of not specifying what the logical
> connection between clauses is, e.g.: "Fool me once, shame on you; fool me
> twice, shame on me". That's four clauses with their connections left up to
> context. I'm not sure we should eliminate the possibility of doing things
> like that in Lojban by forcing an obligatory conjunctive interpretation.
True... but even there, I'd say that the boolean connectives involved
are all conjunctions; this doesn't seem like the kind of conditional to
be analysed using ganai-gi.
Perhaps there can be more to an {i} than *just* conjunction, but then
perhaps the same could be said of {i je}.
Attachment:
signature.asc
Description: Digital signature