every once in a while, we run into the same discussion in the Nijbo development server: there's something weird about the way Lojban's positional case tags work.
under CLL, BPFK, and generally with the language as it stands now, the way the FA-series particles work is that any unmarked arguments after them count up (CLL 9.3). to illustrate this, they give a simpler example, but what interests me is the more contrived example 3.8:
mi klama fi la atlantas le dargu fe la bastn le karce
I go from Atlanta via the road to Boston by car.
(in case you don't remember all the places of klama, it's '… goes to … from … via … by means of transport …'. whew!)
here, ‹mi› fills the first place (it's in the regular, pre-verb subject position). then, ‹la atlantas› fills the third place of klama (the 'from'), because it's marked with ‹fi›. then ‹le dargu› comes next, but it's not marked with a case tag, so it automatically fills the next place — here the fourth one, the 'via' place. then comes ‹fe la bastn›, which is tagged with ‹fe›, so it fills the second 'to' place. then — ŭedaisai — the next noun phrase, ‹le karce›, is untagged, so it fills… the next unfilled place, which in this case is the fifth one. this feels like an absolute proof that Lojban is meant for computers instead of people.
the CLL does admit that "such a convoluted use of tags should be avoided [...] the rules stated here are merely given so that some standard interpretation is possible", and they are right in that no Lojbanist would do this unless they wanted to play pranks on their listener.
but the counting-up rule adds convolution to sentences that don't need it. consider the construction:
a'y dunda fi by fe cy
'A gives B C'
the first ‹fi› tag is needed under any interpretation of the case tags. if I just said ‹a'y dunda by boi cy› instead, ‹by› would fill dunda-2, and ‹cy› would fill dunda-3. in other words, I would be saying 'A gives B to C', not 'A gives B C' [= A gives C to B].
the second case tag here though — ‹fe› — is only needed because of Lojban's "FA prepositions count up" rule. if you say this instead:
a'y dunda fi by boi cy
you'd end up filling the (here nonexistent) fourth place of ‹dunda›. Lojban FA ends up prioritizing higher places of verbs, that often don't even exist, and requires you to say ‹fe› every time you want to switch back to the x2 — which in most verbs with "naturalistic action frames" is the direct object! if you forget the ‹fe›, you'll end up filling some bizarre by-standard or under-conditions bloatplace half of the time that doesn't even deserve to exist. [you really have to hand it to the Toanists for restricting their verbs to three places at most.]
is this a real problem though, Mati? — I hear you ask
it is! everywhere I go — in a metaphorical online sense, at least, as Lojbanists don't touch grass — I hear people say things like:
i mi ei retsku ‹fi› la srasu ‹fe› losedu'u sy kansa djica jicu zanru lonu vimcu sy le sivni casnu
I have to ask srasu if he wants to join or if he's fine with [us] removing him from the private conversation
u ka'e skuna ‹fi› do ‹fe› le'e tcila caku
but I can tell you about the details right now
(I saw these two sentences back to back in the RJG chatlogs. all examples in bold are from attested sources — from conversations on roljbogu'e, actually, unless I specify otherwise. examples I wrote to illustrate things are in italics.)
[...] ko fi mi benji fe su'o mupli snavei [...]
send me some sound recordings
i lo prenu cu cusku fi mi fe lo simsa be lu [...]
this person said to me something like, "[...]"
and so on and so on. this ‹fi› followed by ‹fe› thing is very common, and I feel like it only exists because Lojban's founders really thought we would be filling the fourth and fifth places of verbs way more often than we really do.
there's also the converse: it's quite common to see a redundant ‹fo› tag right after the x3 was stated with ‹fi›:
‹fi› ko preti ‹fo› le bradi fa lu [...]
ask your opponent if [...]
(here you have the problem of preti having a weird frame as well. retsku is usually less confusing, but ko retsku fi le bradi fe lu [...] runs into the other problem from above with fi and fe.)
i mi fanva ‹fi› lo jbobau [‹fo› sa'a lo jugybau] fe «la finpe be vi la berxas»
I translated into Lojban [from Chinese] "there is a fish in the North Sea"
(in this example, the English can't copy the word order from the Lojban. fronting indirect objects like this just doesn't work this time… but you get my point.)
anyway, what is the solution to this? is there a better way, that doesn't require you to go ‹cusku fi … fe lo se du'u …›, and that respects more naturalistic intuitions of which arguments should be able to go untagged?
it turns out there is. and it's not even completely novel — solpa'i used to do something similar (though not identical), and my proposed solution is, in fact, the same one that is used in modern Loglan.
I don't exactly have a name for this proposal. I could call it "Loglan-style FA", as my friend janbe does. more funnily, ninlat has suggested the name ŭokfa (from "woke FA"), as recently in roljbogu'e there has been a tendency of calling novel grammatical reforms "woke". I have called it "solpa'i FA" before, but Ntsékees says that solpa'i's interpretation of FA is not the same as mine. I'll leave this idea unnamed for now; if we even need a name, it's community discussion that will decide what this is going to be called.
anyway, let's get to it.
how I think new FA should work
the point of a FA preposition is to change the word order of the inherent arguments of the verb (so not arguments incidental to it that are tagged with prepositions like gau or ma'i). as I see it, the way this should happen is that only the argument that is tagged with the FA preposition is moved out of place — so, no counting up.
this is easiest to see with an example:
i fi su da ra na tolsimsa la ĭubabas ku ne la tci'iros [mi za'o citno · solpa'i]
in a way, she isn't too different from Yubaba from [the movie] Chihiro.
note how after the tagged ‹fi su da›, the following argument, ‹ra›, fills the x1 of the verb ‹tolsimsa› instead of filling a nonexistent x4. (‹simsa› only has three places: '… is similar to … in aspect …'.)
likewise:
e'o ko dunda fi mi lo va cukta
can you give me that book over there, please
would have the x3 of ‹dunda› be filled by ‹fi mi›, which is tagged explicitly, and then the following noun phrase, ‹lo va cukta›, fills the x2 of the verb instead of the nonexistent x4. once again, "new FA" prioritizes lower places of verbs — especially places one and two — as opposed to the higher places, which either don't exist or aren't even used for most verbs.
and even when verbs have higher places, like fanva or klama, it's often clearer to tag all places numbered 3 and above. the examples above with redundant fo after fi also show this — it's not very intuitive to say, for example:
mi fanva fi lo jbobau lo glibau
for 'I translate to Lojban from English' — I'd expect a lot of people to just reach for the officially redundant, but better for clarity, fo here. on the other hand:
mi fanva fi lo jbobau lo lisri
'I translate to Lojban a story' (under new FA)
is something that I suspect is reasonably common to want to say, but the language fights against you here (officially, at least) by forcing you to say ‹… fi lo jbobau fe lo lisri›.
even for those who enjoy their higher places, there is an advantage to new FA: you can front a higher place without having to "reset" the count with a preposition right after to say it, as in, for example:
I sent a picture to you on Discord.
the noun phrases and the main verb here are:
I → mi
sent → benji, which has the frame:
x1 sends x2 to x3 from origin x4 via means x5 (a rare quinary verb, which arguably deserves all its places)
a picture → lo pixra
you → do
Discord → la diskord
let's suppose that here you want to front the 'on Discord' part. in English, this would look like "on Discord, I sent a picture to you" (which sounds quite unnatural, but it gets my point across). under official FA, you would have to say:
fu la diskord fa mi benji lo pixra do
once again, the language punishes you for fronting a higher place, and requires you to "reset back the clock" by saying ‹fa› after the ‹fu la diskord› argument. this should be unacceptable — after all, what can even come after ‹fu›? an x6? x6 doesn't even exist outside of the most monstrous kurtyvla!
under new FA, you can move the ‹fu la diskord› x5 wherever you want, and it won't mess with anything else:
fu la diskord mi benji lo pixra do
'on Discord, I sent a picture to you'
mi benji fu la diskord lo pixra do
'I sent, on Discord, a picture to you'
mi benji fi do lo pixra fu la diskord
'I sent to you a picture on Discord'
all of these require one less FA tag to express under my proposed new rules than they do under the existing standard.
rarely does one actually use all places of a quaternary or quinary verb at once. you usually only use one or two of the main ones (the x1, the subject, and the x2, the first object, usually), and then you move to the extra place with an explicit FA tag (unless it's a sentence-final x3).
the existing standard, though, fights against this tendency, by assuming that you are going to fill all verb places in order, and any deviation from this needs to be marked by skipping to the new place and then "de-skipping" back to the old place!
in practice though, most verbs just fill one or two places, with some filling three. having to "roll back" every time you want to fill the occasional anomalous fourth place, or needing to say ‹fi … fe …› every time you want to say ‹… cusku fi mi fe lo se du'u …›, is just the language fighting against the speaker. higher places should be marked with prepositions, and when I want to go back to a more core lower place I shouldn't have to "set the clock back" with a ‹fe›!
oh, verb-initial word order still works, by the way
the most common trick with FA in Lojban is to say:
broda fa x1 x2 x3…
because (especially in a language where every content word is a verb) you often want the effect of different word orders, or you want to avoid center embedding in subclauses. this is especially common with words like zgadi '… should be the case', or cumki '… is possible'. even in English, we often say 'it is possible to do x' more than we say 'doing x is possible', especially in longer clauses.
this trick still works under new FA. the ‹fa x1› marks that this x1 indeed fills the first slot of the verb. all other slots of the verb go as if there was no ‹fa›-tagged argument before them — so, since they come after the verb, they start counting from x2.
I think this covers most of what I have to say about my idea on new FA. I could have gotten into how Loglan does it, but I am no Loglanist — janbe surely knows better than me; all I know is that it works the way that I propose it should work. there are also details about how it's easier to implement the new FA from a formal semantics and syntax standpoint, but that's something Zearen knows more than me about.
whew! (I wrote this whole blog post in a single day.) thanks for reading, take care, have a nice day, and see you next time!