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

On Sun, Jun 1, 2014 at 6:54 AM, Gleki Arxokuna <gleki.is.my.name@gmail.com> wrote:

Before that we need to check if some proposition and property places can also take events.

It seems to me that properties (i.e. incomplete propositions) are in a sense incompatible with anything else. The only way a property place could also take something else is through overloading. That's not necessarily a bad thing, but it means using the same word for two strongly related but different predicates.

E.g. kakne2 was an event place that could lead to an object place (mi kakne lo bajra) there. Now it's a property place (mi kakne lo ka ce'u du lo bajra).

And in the opposite direction, some people prefer binxo2 as a property, so that "mi binxo lo bajra" becomes "mi binxo lo ka ce'u bajra".

Propositions can be seen as properties (mi kakne lodu'u lo no'a cu du lo bajra). What is chosen in this or that place seems to be rather arbitrary to me.

Incomplete propositions are not complete propositions, but since places for incomplete propositions are normally tied to another place that contains what is required to complete the proposition, then yes, in principle either way could be chosen to define the predicate. This doesn't apply so easily to cases where the property is tied to two places at once though, because there's no immediate proposition that could replace the property.

Most predicates with places tagged as "du'u" also have an additional "about" place, which means they could be replaced with property places: "mi djuno lo ka ce'u nelci lo cakla kei do". The problem here is that in these predicates the "about" place comes after the proposition place, and for most property predicates the thing-with-property place comes before the property. So a more natural order would be "mi djuno do lo ka ce'u nelci lo cakla".

So the first step is to clearly separate events vs. propositions+properties.

An incomplete proposition does not describe an event. A complete proposition can and often does describe an event. If we wanted to, we could easily make do with a single NU for all subordinate clauses, since it's not really necessary to point out explicitly that a proposition describes an event, and we don't strictly need more than "ce'u" to mark an incomplete proposition. So yes, the assignment of ka/du'u/nu to argument places is rather arbitrary.

du'u: introduces a complete proposition, unless it contains an explicit ce'u, in which case it introduces an incomplete proposition.

ka: introduces an incomplete proposition. It may but need not contain an explicit ce'u, because if not explicit an implicit one is assumed.

nu: introduces a complete proposition that describes an event. Same as du'u with respect to ce'u.

Some propositions are just too abstract to be said to describe an event e.g. lo du'u li vo sumji li re li re, where or when would that take place? Not really an event.

I think "sound" is way too specific to be a type. Something like "living organism", or even "container" would have many more instances than "sound". And if sance1 and zgike1 are "sound", shouldn't tonga1, voksa1, savru1 and rilti1 be as well?

rilti1 or rilti2? Or both? Also sanga2.

I was only looking at the x1 column, but yes, those would seem to be sounds too.

The problem for me is that krixa2, cmoni2 can probably be texts. And text and sound often go together.

Semantic categorization currently puts them in different classes (communication, non-linguistic utterances, Music/sound...)

Can I cusku a selsanga? That's why I wanted "sound" to be a separate type that could include both sounds and "text".

If a text is anything with linguistic meaning, then some sounds are also text. But there are sounds that are not text, and texts that are not sound. cmoni2 is explicitly described as non-linguistic.

mixre

changed to "x1 (object, event) is a mixture including x2 (set of objects or events)"

/gunma

changed to "x1 (object, event) is a joint mass, team of components that are x2 (object, event)"

I think those should be "any type".

/kamni

kamni3 is probably a "set".

In my opinion no place should be marked as "set".

rilti is marked as a sequence but again it should be just a group
Any example of rilti1?

It's easy to provide vague examples like

lo sutra cu rilti lo nau zgike

"A fast one is the rhythm of this music."

but I guess you want a more specific description like "3/4 is the rhythm of this song". I can already hear the objections, but I think I would go with "li ci fi'u vo cu rilti lo nau zgike".

What is a group according to you?

One that consists of many. E pluribus unum.

porsi1 includes {ce'o}
casnu1 includes {ce} or {jo'u}.

I think porsi1 takes the one (lo gunma) while casnu1 takes the many (lo se gunma). But these are not types in our number/object/event/proposition/property typology. Both the one and the many could be of any of those types.

ckilu/sidbo are marked as "(concept)". A sidbo is a du'u, it's something that could potentially become a fact. I don't think it needs a special type. And ckilu has little to do with concepts, as far as I can tell.

What to fill ckilu1 with if not with {si'o kei}?

I don't exactly know, but I have a lot of difficulty extracting a scale from a proposition.

mu'o mi'e xorxes