User:Maiku/MLL3 design notes

MLL3 is a codename meaning Maiku's loglang, version 3. This page contains some random notes collected from the Logical Languages Lodge Discord server and perhaps in the future other things.

Evolution of MLL3 Syntax from Xorban

A long time ago, from Xorban I got the idea to go with SOV, but I was still hemming and hawing on many aspects of basic syntax, so late last year (i.e. 2020) I went back to Xorban for inspiration. To explain what I came up with, I will first present a Xorban refresher/crash course.

In Xorban, a verb is composed of a stem followed by one or more variables separated as needed by "-k-".

A stem is one or more consonants buffered as needed with the special vowel "ë". A variable is one or more regular vowels separated as needed by an apostrophe. I'll double-underline variables to make them stand out better in glosses. Some variables are pre-assigned, e.g. a'a means "I/me". For example:

sërnaka'a

sërn-a-k-a'a

STEM-VARIABLE-SEPERATOR-VARIABLE

pertain-a-SEPARATOR-1.SG

"a pertains to me."

A clause is a verb or a clause preceded by zero or more noun phrases.

A noun phrase is composed of a determiner followed by a clause, in this position called a restriction clause, while the clause that the noun phrase is connected to is called an assertion clause.

A determiner is composed of a consonant plus a variable. A variable specifically in this position is what I call an assignandum (that which is to be assigned a referent).

A conjunction joins two following clauses ("Polish" prefix notation).

To take a simple sentence:

la je sërnaka'a mëlta xëkra

DET CONJ CLAUSE CLAUSE CLAUSE

l-a je sërn-a-k-a'a mëlt-a xëkr-a

the-a and pertain-a-,-1SG cat-a black-a

(a, such that ..., is or are such that ...) (both ... and ...) (a pertains to me) (a is a cat or are cats) (a is or are black)

a, such that both a pertains to me and a is a cat, is such that a is black

"My cat is black."

That's the crash course. Now let's fiddle with this syntax. If you try to write this sentence using Lojban words, treating variables as separate words and creating a totally different syntax based closely on Xorban you might come up with something like this:

{lo my. je srana my. ["-k-"] mi mlatu my. xekri my.}

...except now -k- is not needed because of the CV shape of the variables. Note I am using {my.} "M" instead of {.abu} "A" for a.

I had already wanted to make some changes to this.

1. I had settled that postpositions would serve as core case tags and peripheral argument markers (typologically congruent with SOV). In addition, for various reasons, sometimes you will want to add a NP to a clause without linking it to a case role, so there is also topic-marking (more accurately, pendent-marking) postposition that can serve that function; here we will recycle {zo'u} for this.
2. By making postpositions mandatory for NPs, you allow determiners to be optional (you only need to delimit one side of a NP). Since the semantics of articles are fuzzy, while case marking is relatively clearcut, I'd rather make postpositions the mandatory NP-delimiter and make determiners optional rather than vice-versa.
3. Whether you use Polish notation (as in Xorban) or terminators (like Lojban {ku}), either way, if you center-embed heavy NPs you are going to run into human mental-processing-ability limitations. So my maneuver is to eliminate center-embedding of heavy NPs; only pronouns (variables) can be embedded (exactly as shown above); in its place, we will connect heavy NPs either using a genitive postposition, or by topicalizing (or pendentizing) a NP and using its assigned variable (i.e. a resumptive pronoun) later in the clause.
4. The need for the {je} prefix is eliminated by (3); since contentives aren't allowed to be adjacent elsewhere we can drop it without ambiguity. To round out the system, VPs will take a mandatory aspect-marker in order to mark the end of the clause/sentence.

Making these changes, the old sentence

{lo my. je srana my. mi mlatu my. xekri my.}

becomes:

{my. srana my. mi mlatu my. zo'u xekri my. vau}

where the first {my} is an assignandum (same as before) and {zo'u} has been inserted as required as a topic marker and {vau} standing in as the required aspect marker; we have dispensed with {lo} and {je} which are no longer needed.

Now here's the key maneuver: {my} occurs four times; three times it occurs as the first argument of a contentive. What if we dropped the first argument? Well, looking at the NP, the assignandum is the same as the first argument of {srana} and {mlatu}, so it seems we could make the first argument implicit; in cases in which the assignandum is supposed to match a non-first argument, we can create voice-suffixes to force the desired argument to be first. In the case of {xekri} though, we have a snag; there is no assignandum. However, the spot for the assignandum is vacant. Perhaps the left-position of a contentive can serve two purposes; in a NP, it is the assignandum position while in the VP/matrix it is simply the first argument (assignatum): And that's what I have decided. Rewriting with these new rules:

{my. srana mi mlatu zo'u my. xekri vau}

m pertain 1SG cat PENDENT m black VERB

"My cat is black."

The first {my.} is assigned; the second used.

And now we've arrived at how my language currently works (except I have a shortcut for English "my = pertaining to me"), if you want explicit variables and you want to be very formal. If you want to speak in a less highly formal (i.e. standard colloquial) way, you can change the marker {zo'u} into a case marker and remove the variable entirely:

{srana mi mlatu fa xekri vau}

pertain 1SG cat NOM black VERB

"My cat is black."

Here {fa} is a nominative case postposition that connects its NP into the first argument slot of the VP.

(I know this is a lot to digest, but hopefully it wasn't too opaque.)

TODO: explain semantics of serial contentives (each having its own x2+ pronouns while sharing x1); provide examples illustrating features of this syntax e.g. pendents, case markers, resumptive pronouns, etc.