Conlang Random Thread

[cedh]

https://footballbatsandmore.wordpress.c ... ogoglyphs/

At 350 glyphs...working on more

Very nice. Put your glyphs into a font, because this is inspiring me to make a simple online input method editor for conlang logograms. Copy-paste the key:value pairs of romanization:unicodecharacter, specify your logogram font in your computer (just the name, no need to upload it), and bam!, instant input method editor.

What I believe some people (including me, and quite possibly masako too) would be interested in is the same thing but with image files instead of (or in addition to) a font as the input. Of course, the images should be predictably named, so you could e.g. specify a path like "/script/script_<KEY>.png", where <KEY> is a variable. Bonus points if the thing could variously output as HTML code or as an image file, the latter optimally with a custom writing direction (left-to-right, right-to-left, top-to-bottom...) and a specifiable row/column length...

[Kuchigakatai]

Very nice. Put your glyphs into a font, because this is inspiring me to make a simple online input method editor for conlang logograms. Copy-paste the key:value pairs of romanization:unicodecharacter, specify your logogram font in your computer (just the name, no need to upload it), and bam!, instant input method editor.

What I believe some people (including me, and quite possibly masako too) would be interested in is the same thing but with image files instead of (or in addition to) a font as the input. Of course, the images should be predictably named, so you could e.g. specify a path like "/script/script_<KEY>.png", where <KEY> is a variable. Bonus points if the thing could variously output as HTML code or as an image file, the latter optimally with a custom writing direction (left-to-right, right-to-left, top-to-bottom...) and a specifiable row/column length...

That does sound like a desirable utility. I just spent 15 minutes researching whether this is possible in JavaScript (I've never done *any* kind of image processing before), and it turns out it is. (I was particularly worried I wouldn't be able to access the image files in a user's computer without having the user upload them to a server, but it turns out there's a way.) So it sounds like a project to me.

I expect it to be somewhat slow: if you have a large logography (say, my conlang Nakka and its ~750 glyphs, including punctuation) and the images are big in size, there will be disk reading dragging the program down, plus all the constant real-time resizing and updating. Images should be as small as possible. Slow is better than non-existing nevertheless!

By the way, my original idea was to allow many-to-many relationships between key and glyph, because logographies are always like that. In Mandarin, 好 can be hao3 or hao4, and hao4 alone can be 好, 號, 浩, 昊 and a bunch of others. Filenames could be whatever you want because you'd need to define the key-glyph relationships elsewhere anyway. But maybe I could implement both my many-to-many way and your simpler one-to-one way... Urgh, I'm getting some scope creep already.

Regarding your bonus points, what did you have in mind when you mentioned "HTML code" output? Bare <img> tags you could copy-paste into your own website?

[cedh]

Regarding your bonus points, what did you have in mind when you mentioned "HTML code" output? Bare <img> tags you could copy-paste into your own website?

Yes. Basically, the tedious part in preparing a text for publication (some of this is of course possible with search-and-replace in a text editor, but that's not very sophisticated and already needs hook characters to the left and right of each glyph identifier). After having defined my input method, I would like to simply write my conlang text in a glyph-for-glyph transliteration and get working code for showing the text on a website.

That said, I'd personally even prefer wikicode over HTML because I usually publish my conlang documentation through MediaWiki. But there you'd have to deal with the wiki's internal file system, and HTML is much more useful elsewhere. Maybe the best way to approach this would be to allow the user to provide their own surrounding code for each image (including the use of variables like KEY, VALUE, maybe image size, etc.), and have a few standard options predefined, among them HTML <img> tags and possibly MediaWiki [[image]] tags. And since the generated code might also be useful here on the ZBB, why not phpBB [img] tags too?

[Salmoneus]

I've detailed eight strong verb classes! Only another eight or so to go!
...and then only five or so weak verb classes, probably.

...seriously, Germanic, what is wrong with you!?

[Ælfwine]

I've detailed eight strong verb classes! Only another eight or so to go!
...and then only five or so weak verb classes, probably.

...seriously, Germanic, what is wrong with you!?

Tell me about it...

[Salmoneus]

If you told someone about Proto-Germanic (and most of its descendents), they wouldn't believe you.

"So, verbs follow one of approximately twenty-six conjugational patterns, not counting the irregular ones. There's only two basic paradigms of inflectional endings, though those show consonant gradation in some conjugational classes, and may result in a number of modifications to the verbal root in others, while the verbal root itself also often shows a pattern of consonant gradation in most of the conjugational classes, which does not signify any coherent distinction but is just there seemingly for fun. The verb root itself in most classes follows a pattern in which up to five seemingly unrelated vowels alternate throughout the paradigm phonemically (usually fewer, but which forms share vowels changes depending on the conjugational class), while allophonic affection (which becomes phonemic in all daughter languages) increases that tally to seven vowels (in some daughters eight). These vowel changes are particularly confusing given that many verbs are derived from other verbs by similar vowel changes (and, as with the consonant gradation patterns, many verbs therefore share some but not all forms with other verbs with different meanings). There are also many classes of verb that undergo partial reduplication, although depending on the stage of the language this may for some verbs be reduced into a semi-predictable pattern of vowel and consonant changes throughout the root, while other verbs of the same class remain unaffected. Verbs distinguish only three persons and three numbers, as well as two voices and three moods, although amusingly some verbs have decided to use all their past tense forms for their present tenses and now have no past tenses, while other common verbs have replaced large sections of their paradigms through suppletion (sometimes borrowing from the same source verb as other verbs so that, again, verbs share entire sets of forms), as you do. Vowels, incidentally, distinguish three values of length, but obviously only one for /a/, and two for /o/ because there's no short /o/, of course, and they also distinguish nasalisation. The clusters /hn/ and /wr/ are considered a perfectly reasonable way to start a word.
Now, about all these noun classes, some of which show vowel alternation, consonant alternation and metathesis simultaneously..."

[Kuchigakatai]

Regarding your bonus points, what did you have in mind when you mentioned "HTML code" output? Bare <img> tags you could copy-paste into your own website?

Yes. Basically, the tedious part in preparing a text for publication (some of this is of course possible with search-and-replace in a text editor, but that's not very sophisticated and already needs hook characters to the left and right of each glyph identifier). After having defined my input method, I would like to simply write my conlang text in a glyph-for-glyph transliteration and get working code for showing the text on a website.

That said, I'd personally even prefer wikicode over HTML because I usually publish my conlang documentation through MediaWiki. But there you'd have to deal with the wiki's internal file system, and HTML is much more useful elsewhere. Maybe the best way to approach this would be to allow the user to provide their own surrounding code for each image (including the use of variables like KEY, VALUE, maybe image size, etc.), and have a few standard options predefined, among them HTML <img> tags and possibly MediaWiki [[image]] tags. And since the generated code might also be useful here on the ZBB, why not phpBB [img] tags too?

I just spent some time writing a specification of how this project is going to look and run. It is... large and ambitious, and it will take me a while to get where I want to get.

So meanwhile, as it's going to take me some time to finish it, I made a simple webpage with a JavaScript program that should be good enough for your particular specific use case (replacing codes with values in a one-to-one correspondence).

https://ser-be-etre-shi.neocities.org/u ... lacer.html

I provided a short input example, but here's some instructions on how to use it nevertheless:

1. Enter the code-to-result correspondences. The program interprets the first equals sign = as the separator. Spaces are not handled, as I use them to separate tokens in the input text area.
2. Press the "Load correspondences" button. This loads the correspondences in the RAM of your computer.
3. Type stuff in the "Input code" field. You will see correspondences being chosen on the fly, along with a formatted output at the bottom.
4. You can edit the formatting being used, using {CODE} for the keys and {RES} for the values.
5. You can copy the outputs to your clipboard by pressing buttons.

[Boşkoventi]

gvue du xqa kuordhiar pon huo xefti ohcee nyiwi terkvaer

hnqwpzit kwov takw oqw

[malloc]

The trouble is that the the animacy distinction actually works rather well in most cases and eliminating it would cause significant hurdles. With the sentence "Mary made coffee for John", the distinctions for animacy and obviation make it simple enough to parse. Without the animacy distinction, the language must resort to extra degrees of obviation all the time. It really is the edge cases where all the participants are inanimate that difficulties arise.

[Ælfwine]

If you told someone about Proto-Germanic (and most of its descendents), they wouldn't believe you.

"So, verbs follow one of approximately twenty-six conjugational patterns, not counting the irregular ones. There's only two basic paradigms of inflectional endings, though those show consonant gradation in some conjugational classes, and may result in a number of modifications to the verbal root in others, while the verbal root itself also often shows a pattern of consonant gradation in most of the conjugational classes, which does not signify any coherent distinction but is just there seemingly for fun. The verb root itself in most classes follows a pattern in which up to five seemingly unrelated vowels alternate throughout the paradigm phonemically (usually fewer, but which forms share vowels changes depending on the conjugational class), while allophonic affection (which becomes phonemic in all daughter languages) increases that tally to seven vowels (in some daughters eight). These vowel changes are particularly confusing given that many verbs are derived from other verbs by similar vowel changes (and, as with the consonant gradation patterns, many verbs therefore share some but not all forms with other verbs with different meanings). There are also many classes of verb that undergo partial reduplication, although depending on the stage of the language this may for some verbs be reduced into a semi-predictable pattern of vowel and consonant changes throughout the root, while other verbs of the same class remain unaffected. Verbs distinguish only three persons and three numbers, as well as two voices and three moods, although amusingly some verbs have decided to use all their past tense forms for their present tenses and now have no past tenses, while other common verbs have replaced large sections of their paradigms through suppletion (sometimes borrowing from the same source verb as other verbs so that, again, verbs share entire sets of forms), as you do. Vowels, incidentally, distinguish three values of length, but obviously only one for /a/, and two for /o/ because there's no short /o/, of course, and they also distinguish nasalisation. The clusters /hn/ and /wr/ are considered a perfectly reasonable way to start a word.
Now, about all these noun classes, some of which show vowel alternation, consonant alternation and metathesis simultaneously..."

Now try adding grammatically induced initial mutations and sandhi to that pattern and you got a winner.

[Xwtek]

What do you call this relativization strategy in my Asent'o language.

I call it relative pronoun. However, it is not declined for case (There is no case marking here, postposition is used instead). Basically I use special pronoun and special version of verb. These two words delimits the relative clause. However, clause internally, that pronoun serves as argument of clause (or inner clause of that clause). Example

I eat the food.
peka no-pok-a
Food 1SG>3SG-eat-R
It is expensive (lit. that's price is high)-
ika a-kundeta a-temb-us-a
DIST 3SG-POSS 3SG-Expensive-COP-R
Combined into:
se a-kundeta a-temb-ur-o peka no-pok-a
I eat the food that is expensive.

There is also another strategy, by simply making the argument to be relativized proximal and conjugating the verb to participle mode. But it only works for some arguments, like Subject, Object, and Dative (There is no locational applicative and genitive applicative). When more complex need arises, the "relative pronoun" strategy is used instead. The above strategy only clauses containing the relative pronoun proximalized (And also if it is clause in clause, not only clause have to be proximalized, the relative pronoun inside inner clause is proximalized, too.) The relative pronoun can also be followed by postposition.

[akam chinjir]

What do you call this relativization strategy in my Asent'o language.

Yeah, relative pronouns: if you've got postpositions moving with the pronoun to the head of the clause, that's the same as case-marking as far as relativisation strategies go. (WH-movement, pied-piping included, is what's supposed to define relative pronouns.)

Relative pronouns are super rare outside of SAE languages and languages that have adopted them from SAE languages, as you probably know. To be honest I'm not sure if they're even attested in prenominal relative clauses (though on that point I might be confusing them with resumptive pronouns).

[Xwtek]

What do you call this relativization strategy in my Asent'o language.

Yeah, relative pronouns: if you've got postpositions moving with the pronoun to the head of the clause, that's the same as case-marking as far as relativisation strategies go. (WH-movement, pied-piping included, is what's supposed to define relative pronouns.)

Relative pronouns are super rare outside of SAE languages and languages that have adopted them from SAE languages, as you probably know. To be honest I'm not sure if they're even attested in prenominal relative clauses (though on that point I might be confusing them with resumptive pronouns).

1. Indeed. That feature is indeed inspired by Indo-European language. I just fitting it on a left branching saliency based language, instead of nominative-accussative language. In many iteration, Asent'o have 3 cases, and currently I'm thinking about 2 cases. But neither of them is Subject vs Object.
2. Asent'o is spoken along central-western part of mataka continrnt in a conworld, so there is no IE there. Is it still realistic enough

[missals]

Another kind of horrible thing I just thought of:

A language where every vowel has a consonant counterpart and vice versa - it only permits CV(C) syllables, and words are strictly consonant-initial, meaning that if a prefix (with an odd number of phonemes) is added, the entire thing is resyllabified, e.g.

/a/ = /k~h/, /i/ = /t~s/, /u/ = /p~f/, /e/ = /j/, /o/ = /w/

/aea/ ‘dog’ = [keh] + /ii/ ‘diminutive’ = [kehis] ‘little dog’

/u/ = definite article

/uaeaii/ ‘the little dog’ = [pajasi]

[Xwtek]

Another kind of horrible thing I just thought of:

A language where every vowel has a consonant counterpart and vice versa - it only permits CV(C) syllables, and words are strictly consonant-initial, meaning that if a prefix (with an odd number of phonemes) is added, the entire thing is resyllabified, e.g.

/a/ = /k~h/, /i/ = /t~s/, /u/ = /p~f/, /e/ = /j/, /o/ = /w/

/aea/ ‘dog’ = [keh] + /ii/ ‘diminutive’ = [kehis] ‘little dog’

/u/ = definite article

/uaeaii/ ‘the little dog’ = [pajasi]

Sorry, I have seen loads of this and this is overdone

[Nortaneous]

gvue du xqa kuordhiar pon huo xeafti ohce nyiwi tearkvaer

hnqwpzit kwov takw oqw

easu

eahe siwea
ease sitea
eaxqe xeinqe

[Xwtek]

How does nounlike adjective works? Probably because of my native language, I associate adjective as verb. As I want to make Rkou grammatically strange, I decide to make adjective nounlike. (In contrast of Asent'o that adjective is conjugated for person, number, aspect, and mood). However, Rkou's declension is limited to singular-plural.

[akam chinjir]

How does nounlike adjective works? Probably because of my native language, I associate adjective as verb. As I want to make Rkou grammatically strange, I decide to make adjective nounlike. (In contrast of Asent'o that adjective is conjugated for person, number, aspect, and mood). However, Rkou's declension is limited to singular-plural.

The easiest thing is to require a copula when using adjectives as predicates. You can also make your adjectives agree with their head nouns for things like gender, number, and case.

[masako]

I finished developing glyphs for family terminology and a novel way to render various relatives.

Using these glyphs:

These can be found on this page

You generate these terms:


Note the use of -ha (augmentative) and -hi (diminutive) to denote relative age. The handwritten versions of each example are syllables, not logographic.

[akam chinjir]

I finished developing glyphs for family terminology and a novel way to render various relatives.

Nice!

For each of the family terms, the glyphs actually seem to be functioning both as ideographs and as syllables. E.g., taya husband = ta + ya, but also = tlaka male + ntaya spouse. Is the idea that these are or derive from compounds? So that the ta in taya means or derives from something that meant male, for example. Can the two uses ever come apart? On the one hand, suppose you had a word for a male relative that didn't include the element ta. Would you still write that word using the tlaka glyph? On the other hand, could you ever use that glyph to write the syllable ta, without regard to meaning? (Or, for a subtler case, what if you used the wrong ya---ntaya spouse rather than tlaya marry---to write ntaueya cousin-in-law?)