This is a topic I've been carrying around with me for a while, but which I never got to write down, because it was too abstract in my mind. I knew what I wanted it to accomplish, but I didn't see an immediate way to get there until now.
First of all, what does proprietor fronting mean? Well, it's a term I made up for the concept of extracting part of a proposition and placing it before said proposition, which then turns into a property. Just keep reading.
Let's look at {djuno}: x1 knows that x2 (du'u) is true about x3
The x2 could be a property of the x3, but it would make it impossible to know what {mi djuno lo ka klama} means without also mentioning the x3 of djuno, which would be fine the x3 came first, but it doesn't. There is another gismu like that, which even uses a property, but has the proprietor coming after it, which is very annoying: {cpedu} x1 asks for action x2 (ka) from x3. For example:
mi cpedu lo ka ce'u sidju mi kei lo pulji
"I asked [to help me] from the police." (literally)
This order of arguments makes sense from the standpoint of wanting to make {cpedu} a gismu of the {cpacu}/{lebna} family, but can be annoying otherwise. Of course, {cpedu} might be a special case, since we often like to say (in English anyway) "I asked for help", in which case the x3 is irrelevant. Let's ignore {cpedu} for the time being and let us return to {djuno}.
It would make sense to have {djuno} be "x1 knows x2 to have property x3 (ka)", so that:
mi djuno do lo ka jifyjunxu'a
"I know that you are lying."
But that's not how {djuno} is defined. However, it shows exactly what proprietor fronting is. Proprietor fronting makes it possible to go from:
lo du'u ko'a broda
to
ko'a lo ka ce'u broda
in a systematic way, and optimally without performing any crazy operations on the selbri.
In current Lojban, the only word that does something similar to this is {jai}, but {jai} is not optimally equipped for this task, in my opinion. I've always considered {jai TAG} the more basic version of {jai}, with {jai broda} being the "odd" one out. Coupled with my dislike for SE-strings and {jai} being quite a lot like SE, I'm starting to wonder if {jai broda} should be removed. I don't exactly dislike {jai broda}, but I do dislike {se jai broda}, and it only gets worse with {se jai se broda}. To me, that's the opposite of an elegant construction. Another "problem" is that {jai} acts on the x1 by default. {tu'a} can act on any sumti you want without having to rearrange the entire selbri, just like proprietor fronting should be able to be used freely on any abstraction place.
With proprietor fronting, there would be no need for {djuno} to be redefined. And {djuno} isn't the only relevant gismu. The same things apply to dozens and dozens of other gismu, even ones without du'u-places but with nu-places (think for example {kanpe} or {pacna}). The method I've used until now was to make separate lujvo or zi'evla with the new place structure, e.g. {kairju'o} (or its zi'evla {ckajiju}) "x1 knows x2 to have property x3". But that would mean having to create an insane number of new, but systematically related, brivla. Why not introduce a new mechanism that can do the same by operating on the existing gismu?
I've known for a while now that I wanted to have such a mechanism. But how would it work?
Let's pretend for now that {djuno} is "x1 knows x2 (du'u)" and forget about the x3. We want to have a regular mechanism to make the following transformation:
mi djuno lo du'u do badri -> mi djuno do lo ka badri
and equally:
mi kanpe lo nu do klama -> mi kanpe do lo ka klama
I can see two possible ways to do this. I will list them using an experimental cmavo kai'a (in parallel universe Lojban, it would be {kai}).
sumti to the next property on the right, skipping anything that isn't property, including selbri and other
sumti. 2b is compatible with 2.
Examples: mi do kai'a djuno lo ka badri
mi do kai'a pu kanpe lo ka klama
ko curmi mi kai'a ca lo nu mi ba'o lumci lo palta kei lo ka kelci lo kelskami
"Allow me, after I have cleaned the plates, to play with the game console."
no da kai'a lakne sei badri lo ka cinri mi == no da zo'u lo nu da mi cinri cu lakne (this example also shows how {jai} is just a very special case of what {kai'a} does)
cfipu fa lo ctuca kai'a ro lo tadni lo ka so'i roi cenba lo ka casnu ma kau
"The teacher annoys all the students by frequently switching subjects."
This shall suffice to present the general idea behind proprietor fronting. Chances are that by now you know whether or not you like it. I would prefer not to make up an experimental cmavo, as I dislike their shape, but what other options are there, really. ;)
If only {kai} were free...
First of all, what does proprietor fronting mean? Well, it's a term I made up for the concept of extracting part of a proposition and placing it before said proposition, which then turns into a property. Just keep reading.
Let's look at {djuno}: x1 knows that x2 (du'u) is true about x3
The x2 could be a property of the x3, but it would make it impossible to know what {mi djuno lo ka klama} means without also mentioning the x3 of djuno, which would be fine the x3 came first, but it doesn't. There is another gismu like that, which even uses a property, but has the proprietor coming after it, which is very annoying: {cpedu} x1 asks for action x2 (ka) from x3. For example:
mi cpedu lo ka ce'u sidju mi kei lo pulji
"I asked [to help me] from the police." (literally)
This order of arguments makes sense from the standpoint of wanting to make {cpedu} a gismu of the {cpacu}/{lebna} family, but can be annoying otherwise. Of course, {cpedu} might be a special case, since we often like to say (in English anyway) "I asked for help", in which case the x3 is irrelevant. Let's ignore {cpedu} for the time being and let us return to {djuno}.
It would make sense to have {djuno} be "x1 knows x2 to have property x3 (ka)", so that:
mi djuno do lo ka jifyjunxu'a
"I know that you are lying."
But that's not how {djuno} is defined. However, it shows exactly what proprietor fronting is. Proprietor fronting makes it possible to go from:
lo du'u ko'a broda
to
ko'a lo ka ce'u broda
in a systematic way, and optimally without performing any crazy operations on the selbri.
In current Lojban, the only word that does something similar to this is {jai}, but {jai} is not optimally equipped for this task, in my opinion. I've always considered {jai TAG} the more basic version of {jai}, with {jai broda} being the "odd" one out. Coupled with my dislike for SE-strings and {jai} being quite a lot like SE, I'm starting to wonder if {jai broda} should be removed. I don't exactly dislike {jai broda}, but I do dislike {se jai broda}, and it only gets worse with {se jai se broda}. To me, that's the opposite of an elegant construction. Another "problem" is that {jai} acts on the x1 by default. {tu'a} can act on any sumti you want without having to rearrange the entire selbri, just like proprietor fronting should be able to be used freely on any abstraction place.
With proprietor fronting, there would be no need for {djuno} to be redefined. And {djuno} isn't the only relevant gismu. The same things apply to dozens and dozens of other gismu, even ones without du'u-places but with nu-places (think for example {kanpe} or {pacna}). The method I've used until now was to make separate lujvo or zi'evla with the new place structure, e.g. {kairju'o} (or its zi'evla {ckajiju}) "x1 knows x2 to have property x3". But that would mean having to create an insane number of new, but systematically related, brivla. Why not introduce a new mechanism that can do the same by operating on the existing gismu?
I've known for a while now that I wanted to have such a mechanism. But how would it work?
Let's pretend for now that {djuno} is "x1 knows x2 (du'u)" and forget about the x3. We want to have a regular mechanism to make the following transformation:
mi djuno lo du'u do badri -> mi djuno do lo ka badri
and equally:
mi kanpe lo nu do klama -> mi kanpe do lo ka klama
I can see two possible ways to do this. I will list them using an experimental cmavo kai'a (in parallel universe Lojban, it would be {kai}).
- Place {kai'a} before the first sumti. It will be treated as the proprietor of the following property abstraction.
Examples: mi djuno kai'a do lo ka badri
mi kanpe kai'a do lo ka klama
The two sumti can be moved around independently, as long as they remain in order. Example:
mi kai'a do kanpe lo ka klama
This could be both an advantage and a disadvantage. - Place {kai'a} between them like a connective. Here, the order wouldn't matter much, as long as there is one property connected with one proprietor.
Examples: mi djuno do kai'a lo ka badri
mi kanpe do kai'a lo ka klama
If {kai'a} works like a connective, then it won't be possible to move {do} in front of the selbri, which is a disadvantage. However, instead of making it a true JOI, it could be a long-distance connective, which placed after the first argument and connects it to the next sumti no matter how far away it is in the sentence, thereby being able to bridge across the selbri, and possibly even other intermittent sumti. This would be solution 2b:
sumti to the next property on the right, skipping anything that isn't property, including selbri and other
sumti. 2b is compatible with 2.
Examples: mi do kai'a djuno lo ka badri
mi do kai'a pu kanpe lo ka klama
ko curmi mi kai'a ca lo nu mi ba'o lumci lo palta kei lo ka kelci lo kelskami
"Allow me, after I have cleaned the plates, to play with the game console."
no da kai'a lakne sei badri lo ka cinri mi == no da zo'u lo nu da mi cinri cu lakne (this example also shows how {jai} is just a very special case of what {kai'a} does)
cfipu fa lo ctuca kai'a ro lo tadni lo ka so'i roi cenba lo ka casnu ma kau
"The teacher annoys all the students by frequently switching subjects."
This shall suffice to present the general idea behind proprietor fronting. Chances are that by now you know whether or not you like it. I would prefer not to make up an experimental cmavo, as I dislike their shape, but what other options are there, really. ;)
If only {kai} were free...