Conlang Random Thread

[Lērisama]

I'd be curious in participating in one, but looking at the previous ones makes me feel we'd need more people.

[bradrn]

I'd be curious in participating in one, but looking at the previous ones makes me feel we'd need more people.

Yeah, that was the problem. Hardly anyone from the other team actually submitted any languages, and even on our team the only reason we had much to show at all was that foxcatdog made a lot of languages.

[Ahzoh]

Learning Ojibwe, I felt inspired that Vrkhazhian should have a "conjunct" form in contrast to an independent one. It would be used for relative and subordinate clauses.

[jal]

Learning Ojibwe, I felt inspired that Vrkhazhian should have a "conjunct" form in contrast to an independent one. It would be used for relative and subordinate clauses.

A form of what? Verb?


JAL

[Ahzoh]

Learning Ojibwe, I felt inspired that Vrkhazhian should have a "conjunct" form in contrast to an independent one. It would be used for relative and subordinate clauses.

A form of what? Verb?


JAL

A form of verb

[malloc]

The quest to numerate my minimalist project continues. Assuming that I settle for boring old decimal, I still have many interesting options for how numbers work. The obvious approach is digits one through nine and roots for powers of ten. However that sounds surprisingly bloated for a language that only needs six color words. An alternative approach is biquinary, building numbers six through nine (nice) from five plus one through four. An additional improvement would involve deriving numbers from existing roots somehow, perhaps "five" from "hand" and "ten" from "body" and so forth.

That said, this approach also raises some significant challenges. The heavy reliance on the element "five" to form numbers would result in considerable repetition and make numbers longer than other approaches. The form "sixty seven" would become the unwieldy "five and one tens and five and two" or something along those lines. Deriving numbers from existing elements saves elements but also invites confusion. Imagine the difficulties of parsing "hand and two bodies" as "seventy".

[bradrn]

What about binary? That’s as minimalist as you can get…

[/ˌnɐ.ˈɾɛn.dɚ.ˌduːd/]

... six through nine (nice) ...

nice.

... "sixty seven" ...

never mind, the door's over there.

[malloc]

What about binary? That’s as minimalist as you can get…

Maybe so, but just using one and zero would result in fantastically long number words. I have experimented with binary based on successive powers of two which avoids this problem to some degree. Consider the difference between "one zero zero zero zero" and a basic root for sixteen. Granted roots for multiple powers of two would lose the minimalism of pure binary.

Decimal has kind of grown on me despite its arbitrary nature, though, precisely because it reflects human anatomy. With art increasingly under siege from computers, it feels increasingly important to stress the humanity of art. Giving my project numbers based on human anatomy sends the message that it's fundamentally for humans rather than AI.

[Travis B.]

I personally am a fan of using biquinary for 1-9, vigesimal for 20-90, and roots for powers of ten for 10, 100, 1000, and 10000, and combinations of roots for powers of ten for powers of ten greater than 10000.

[Travis B.]

If you want to base things on binary, though, why not use hexadecimal or octal? Then you can fundamentally base things in binary, while having it being far more human-friendly than lots of ones and zeroes.

[malloc]

If you want to base things on binary, though, why not use hexadecimal or octal? Then you can fundamentally base things in binary, while having it being far more human-friendly than lots of ones and zeroes.

That could certainly work. I have some promising ideas for hexadecimal numerals that would fit neatly with the conscript for my language.

Overall it feels like optimizing the number system in one way requires significant downsides elsewhere. There is a difficult balance between conceptual elegance, practical usability, lexical minimalism and so forth. Minimizing the number of roots requires more complex rules for combining them to form numbers of a given size. More arithmetically logical alternatives to decimal don't fit existing measurement systems very well. It is quite the conundrum. Consider the difference between the following:

six-thousand-seven-hundred-eight-ten-nine
five-one-thousand-five-two-hundred-five-three-ten-five-four

The latter requires fewer lexical items because higher digits are formed from lower digits, but it requires more morphemes for each number. It also has considerably more repetition since the element for "five" recurs with each digit. Furthermore it also shows more formal complexity since the digits themselves are constructed from smaller elements.

[malloc]

Here is a rough draft of proposed hexadecimal numerals. There are four basic strokes (shown in the top row) representing successive powers of two. The other digits are built by combining these and summing their respective values. For instance the digit 0xF (15 in decimal) contains strokes for one, two, four, and eight.

[WeepingElf]

Those make good sense and look nice!

[Travis B.]

Those make good sense and look nice!

Seconded!

[Raphael]

Those make good sense and look nice!

Seconded!

Yes!

[jal]

Here is a rough draft of proposed hexadecimal numerals. There are four basic strokes (shown in the top row) representing successive powers of two. The other digits are built by combining these and summing their respective values. For instance the digit 0xF (15 in decimal) contains strokes for one, two, four, and eight.

Quite cunning. You say it's hexadecimal, but the underlying system is binary. Nice!


JAL

[Ahzoh]

I feel uneasy about my case system. My cases feel too patternistic or monotonous. It just looks like all the affixes are -VC in the singular and -V̄Ca in the plural. The only variation away from this pattern is the vocative singular and equative case. Otherwise, not much variation in affix "shape" or affix vowels.

Sadly, my limited knowledge of sound changes results in my morphological systems being diachronically shallow. My constraints, which are important, also limit my options; specifically the constraints where hiatus is banned and that there cannot be more than two unstressed light syllables in a row.

I don't know, do these look naturalistic?

Animate a-stem Singular Plural Nominative -am -āya Vocative -ī -āya Accusative -as -āsa Genitive -an -āna Equative -īli -īli

Animate i-stem Singular Plural Nominative -im -īya Vocative -ī -īya Accusative -is -īsa Genitive -in -īna Equative -īli -īli

Inanimate a-stem Singular Plural Nominative -an -āna Vocative -an -āna Accusative -an -āna Genitive -aḫ -āya Equative -īli -īli

Inanimate i-stem Singular Plural Nominative -in -īna Vocative -in -īna Accusative -in -īna Genitive -iḫ -īya Equative -īli -īli

Moreover, I am not sure how best to assign the declension classes. I would like it to not be completely random (e.g. all kinship nouns and names are i-stem) but without accidentally creating noun class subsystems (e.g. like in Polish).

[bradrn]

I don't know, do these look naturalistic?

They look fine to me. Of course, I can’t say anything about your phonaesthetic preferences.

[Ahzoh]

I don't know, do these look naturalistic?

They look fine to me. Of course, I can’t say anything about your phonaesthetic preferences.

That's what I thought too, but then I keep looking at it and doubting myself. I have tried to find naturalistic languages where all the case endings end in the same vowels but even they have a little variation.

Hittiteistem.png (13.95 KiB) Viewed 1119 times

But there is not much I can do when all of the case endings were CV and the plural was -h-, which is as deep as my morphological development of the case system goes.