Re: [engelang] Xorban Development - engelang messages

On Thu, Aug 23, 2012 at 6:58 PM, Jorge Llambías <jjllambias@hidden.email> wrote:

On Thu, Aug 23, 2012 at 12:29 AM, Mike S. <maikxlx@gmail.com> wrote:

> I was brainstorming a bit last night. Observing that "l-" was simply a
> structure-word, and reasoning that the binary-operator "lX CCCX" was going
> to get used a whole lot, I was thinking that maybe CCCX by itself under
> certain conditions could be interpreted as an equivalent unary-operator.

("lX CCCX" is itself a unary-operator, even though I confusingly call
the construct 'binary-operator' in the grammar, the binary-operator
part is just "lX".) The problem is establishing those "certain
conditions", since CCCX looks like a predicate (or nullary operator)
not like a unary-operator.

When I first glanced at your grammar, I thought you made a mistake in the sentence production because of the way you formulated your operator productions. Because the whole thing is so minimal, I just figured it was just a quick sketch. A day later I realized you had defined the whole language precisely in ten production rules.

In your latest it might be slightly clearer to rename things:

sentence := illocutionary-operator? formula
formula:= predicate | operator formula

I would suggest something better than "operator" if I could think of something, but I cannot. Maybe semiformula or formula-starter.

>> na sa xrja vfla ... na se nlceka
>> It is not the case that for some x which is a pig, x flies ... It is
>> not the case that for some y, y is a wing of x (=pigs).
>> No pig flies. They have no wings.
>
> Okay, it took me a while to get this. Yes, that seems to work the way we
> want.
>
> The only question is about using the commonly used variables of shape "V"
> - if we are also doing things like using free variables to create the
> passive voice so to speak, we have to be careful about that. The sentence
> "na la xrja nlceka" intended as "pigs aren't liked", under a certain context
> might mean "the woman with the red hat doesn't like pigs". So maybe we
> should set aside or some "V'V" or "Vy"or "aw" for the purpose of an
> explicitly unbound variable indicating something like "zo'e".

We could have one of the V'V (say "e'u") implicitly restricted to
"people". So "pigs aren't liked" ("people don't like pigs") would be
"la xrja na nlce'uka" (or equivalently "na la xrja nlce'uka"). That's
not a general solution for the passive voice, but it covers a lot of
ground. Maybe for a better passive we can resort to "ï'o" (from Lojban
"zi'o").

I just thought we needed a variable like "o'e" that was always going to bound to the "l- sm-" constant. I suppose "e'u" doesn't hurt anything, but it seems to me that peoplishness is pretty implicit in ncl-x1, so I question the need to tap into V'V-space for "e'u".

I was also thinking of adding "g-" to the K-separators as a way of
adding an agent argument to any predicate.

You haven't yet stated what Xorban argument structures are going to be like, but I assume that slots will be defined judiciously. One or two in most cases; common slot-types added via -K- suffixes. Maybe change -k- to -t- versus -g-.

> In any case, discourse topics should probably get a more marked variable
> than "V".

Would that be more a matter of style rather than prescription?

Yes, I suppose so; my ideas are shifting on this. This language is a little unusual on how it overtly and dynamically assigns and reassigns variables for every predication that in other languages would be covert, and I have been wondering how that's going to play out. For gut-reasons I thought that the discourse topic should get its own variable. I also had this strange passing idea of having say "o'a" as a book-level topic, "o'e" being a chapter-level secondary topic, o'i being a paragraph level tertiary topic, etc. This would be a good way to structure information in longer formal works. In regular speech, "o'a" would probably be more than sufficient for the current discourse topic. "a" would probably be the sentence-level topic as needed.

>> Here I note I've already used "nlc" both for "x1 likes x2" and for "x1
>> is a wing of x2". We will need some vocabulary construction if we
>> don't want this kind of homonymy.
>
> It's probably not expected or acceptable for an engelang of this sort to
> have a sentence that means both "they have no wings" and "no one likes
> them". We have plenty of room for roots. I assume that epenthetic schwas
> are permitted and that we are not being asked to master Georgian
> phonotactics...(?)

Yes, that's the idea. Schwas can be inserted at will between consonants.

Are there going to be any rules on where they will be inserted in the "standard" dialect and how they will be written, metalinguistically if not otherwise? FWIW I prefer /mëlt-/ to /mlët-/ for "cat".

> I am unsure about the phonotactics of /y w/ so I will
> leave those out:
>
> CC: 17^2 = 289
> CCC: 17^3 = 4913*
> CCCC: 17^4 = 83521*
>
> *minus "nmC(C)" and similar series. I assume CC for special things like
> generics and case tags. CCC for regular vocab and CCCC for jargon.

Roughly yes.

> No
> compounding or derivation.

No formal rules for that, at any rate, no.

> Some effort should be made so that changing one
> phoneme doesn't result in a valid root within a similar semantic category.

Yes. Although with numbers, that's just what happens.

If we are going by Lojban, p/b are pretty close, as are m/n and s/z. Probably should assign three new numbers, or bolster (some or all) numerals to CC.

> Depending on how many variables we need, I almost want to suggest "e" for
> roots where needed and "o" for compounding. That would leave a, i, u and all
> 25 V'V for variables, and eliminate the schwa phoneme.

It would make words longer though, since we would need to introduce a
number of restrictions on permissible consonant clusters. I prefer to
keep the simple CCC* for now with no restrictions.

The rules would be the same for /ë/ as they are now - insert them where needed, i.e. no rules.

root :=C ë? C (ë? C)*
stem:=root (hyphen root)*

The hyphen could be any "V ( '? V)*"
If "o" were a hyphen then "brost" could be a compound of br- and st-.

Just ideas.

> /y w/ could be used
> in root onsets, but I would reserve /ay/ etc. as variables.

I was originally thinking of them as ordinary consonants, but I can
see how they can cause trouble when they are followed by another
consonant. Maybe I will remove them from the current portion of the
grammar and perhaps reserve them for stuff not yet considered, like
interjections and illocutionary indicators.

I think they could safely form PyëC and TwëC sequences where P is a non-coronal obstruent and T is a non-labial obstruent. I think putting them at predicate-edges position is iffy because I don't like /iy yi uw wu/ arising at word boundaries. Probably not worth the headaches.

> Compounds can
> be both dictionary entries and nonce expressions. Compounds are not
> particularly logical, but they can be helpful in constructing vocab.

Predicates are an open class, so the dictionary will not contain them
all, and yes some will be nonce.