[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [lojban] Specifying sumti types: another revision of gimste is complete






2014-06-04 2:40 GMT+04:00 Jorge Llambías <jjllambias@gmail.com>:

On Tue, Jun 3, 2014 at 1:43 PM, Gleki Arxokuna <gleki.is.my.name@gmail.com> wrote:

2014-06-03 3:09 GMT+04:00 Jorge Llambías <jjllambias@gmail.com>:

When you say that casnu1 and simxu1 are "sets", you mean "plural", right? As in "lo ci prenu cu casnu lo cukta", "lo re prenu cu simxu lo ka ce'u prami ce'u". Or do you mean "set" as in "lo'i"/"lu'i"? Or both?

Moving on to the x2's:

You also have sets for mixre2, porpi2, spoja2, lanxe2, jbini2, fenso2, konju2, liste2, kampu2, ralju2, lanzu2, bilma2, kancu2, linji2, plita2, sarni2, jinga2 (why?), misno2, natmi2, pesxu2, ransu2, terdi2, gredile2, kombitu2, vlamei2;

That suggests that you do mean "plural" by "set". But then there's cmima2, which shouldn't be a set by that rule.
BTW, is there a difference between (set) and (set of any type)? 

Well, I'm losing the track of this discussion. What are your suggestions on naming those places?

It depends on what the goal is.

I want a gimste that would teach 1. te sumti interactions 2. what type can go into each place. 3. how plurality is formed inside each place if applicable.

If you can see other goals (like e.g. improving semantic categorization) you are free to add new columns or ask me if it requires some automation.


Are we just specifying types for each argument place, or something else besides types? You started by saying you were specifying mutually incompatible types. For this, the first thing to do is to list all the possible types, so we know what we have to choose from. "Mutually incompatible" is realatively easy to achieve for the more abstract types (proposition, property, relation, number), but for the more concrete types (event, sound, text, object) it is not always so easy to see them as mutually incompatible, since there's a lot of overlap.   

Another difficulty is that many of the specifications (usually inherited from the official gimste) are not about permanent types at all, but either about roles (agent, patient, instrument, observer, place, medium) or about distributivity ("set", "mass", "individual"). I put those in scare quotes because there isn't even an agreed definition for what they mean, so using them to explain something else is always risky.

 
Of course I can change all of them to just "object".  Still I wish a formalized explanation was given for each place of what connective to choose. E.g. using {jo'u} for porsi1 would be strange.

It all depends on how you use it. Since "porsi" can just as easily mean "are sequences" as "is a sequence", there shouldn't be a problem in using a plural sumti formed with "jo'u" in x1. You just have to know that "jo'u" doesn't create an emergent single thing like ce/ce'o (and "joi" with some of its definitions) do.

Out of {ce'o}, {jo'u}, {joi} casnu1 can accept {jo'u} but not {ce'o}, porsi1 can accept {ce'o} but not {jo'u} (unless that jo'u connect sumti then connected with {ce'o} again, right?

This is what I also want to reflect in the gimste.


The important thing for porsi1 is that it has to be something that consists of other things, so that it can make sense to say that those other things are in some order. But "something that consists of other things" is not a type, at least not a type in the above typology, because all the concrete types there can consist of other things, and probably most of the abstract types too. If we make "set" another abstract type, and the only type allowed for porsi1, then we connot say such elementary things as "mi viska lo porsi be (fi) lo manti" because abstract sets are not the type of thing that can be seen. 

Most of the places marked as "set" are usually plural and non-distributive, i.e. it doesn't make much sense to fill them with a sumti that refers to only one thing, and when filled by a sumti that refers to more than one thing you cannot distribute the predication for each of the things, it applies to all of them together. But that's independent of the type of the things.

If I define
porsi1 as "x1 (any type; {ce'o} for showing sequence)"
casnu1 as just "x1 (any type)"
gunma as "x1 (any type; {joi} for showing mass)"
would it be okay?


To sum up: "type", "role" and "distributivity" are three different and mostly independent properties of argument places. 

(observer) is not a type in the same sense that proposition/property/number/etc are types.

Okay I can change that from "x3 (observer)" to ''observer x3 (object)".

That works for me. One problem with the word "object" though is that sometimes it includes, but sometimes it is used in contrast with, sentient beings. I guess this should be clarified somewhere. These "objects" will almost always be people.
If you have a better term than "object"...
 
 
 
  {cinza} is not a body part. They are tweezers.

Hmm, right, it's mainly a tool:

 "x1 is a/are tong(s)/chopsticks/pincers/tweezers/pliers [tool/body-part] for x2 to pinch x3"

But it doesn't describe an action. "ko'a ca'a cinza ko'e ko'i" doesn't mean that ko'i actually gets pinched, just that ko'a actually is for pinching ko'i. If "cinza" was an action, the expected definition for me would be "x1 pinches x2 with x3 (at locus x4)". So I don't think x2 is an agent. I can see how x3 fits with the tool definition, but I don't really see why there is an x2 at all though, other than to make it a body part..

 
One last question.
nelci - x1 is fond of/likes/has a taste for x2 (object/state).

Can we say:
mi nelci lo plise
mi nelci lo nu do limna

I have no problem with those two.
 
mi nelci lo ka limna

I can accept this one, but to me it entails a separate meaning for "nelci" if it is to mean that I like to swim, as opposed to just liking the abstract property of being a swimmer:

If it is to mean "I like to swim" then the place would be
"x2 (any type, property of x1)"

 
mi nelci li mu
?

And that's why you always bet on it when playing roulette?

mu'o mi'e xorxes


--
You received this message because you are subscribed to the Google Groups "lojban" group.
To unsubscribe from this group and stop receiving emails from it, send an email to lojban+unsubscribe@googlegroups.com.
To post to this group, send email to lojban@googlegroups.com.
Visit this group at http://groups.google.com/group/lojban.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "lojban" group.
To unsubscribe from this group and stop receiving emails from it, send an email to lojban+unsubscribe@googlegroups.com.
To post to this group, send email to lojban@googlegroups.com.
Visit this group at http://groups.google.com/group/lojban.
For more options, visit https://groups.google.com/d/optout.