Conlang Random Thread

[linguistcat]

Considering making a "reverse Japanese" conlang for the main non-human species in the space fantasy I'm writing.
...

I have specific questions now
Would prefixes be more likely to show up than suffixes in this language? Or could either occur? Or would suffixes actually be the most common affix?
What about auxiliary verbs? Would they be more likely to show up before the main verb? Or could the main verb of a sentence/verb phrase be the head and the aux verb come after it, later becoming a suffix of sorts?
In the vein of "reverse Japanese," would something like "sentence final particles" be more likely to go at the start of a sentence to alter it, or could they still go at the end of a sentence? Are there any examples of something like this in a VSO language either way?

Thanks for any answers I might get.

[jal]

Would prefixes be more likely to show up than suffixes in this language? Or could either occur?

Given that many languages have both pre- and suffixes, I'd say that either could occur. It also depends on the origina of your pre/suffixes: were they once stand-alone words, becoming clitizised? If so, what type of words were they, and what was their normal location (before or after the head)?


JAL

[linguistcat]

Would prefixes be more likely to show up than suffixes in this language? Or could either occur?

Given that many languages have both pre- and suffixes, I'd say that either could occur. It also depends on the origina of your pre/suffixes: were they once stand-alone words, becoming clitizised? If so, what type of words were they, and what was their normal location (before or after the head)?

JAL

Thank you for the reply. The language I'm planning would be strongly head first, so prepositions would become prefixes but most other word-combos that turned into some form of derivation would become suffixes. I'm not sure how aux verbs act cross linguistically, but I'd prefer they became suffixes that conjugated the main verb stem. But if that would be unlikely in this case I could work around it.

[jal]

I'm not sure how aux verbs act cross linguistically, but I'd prefer they became suffixes that conjugated the main verb stem. But if that would be unlikely in this case I could work around it.

A quick google showed that the position of auxiliary verbs strongly depends on whether the language is OV or VO. In OV languages, it's mostly VAux, while in VO languages, it's mostly AuxV (see e.g. here (PDF, page 25)). Also, it seems that cross-linguistically, the auxiliary verb is treated as the head of the VP, so that in a strongly head-first language the order is AuxV (which coincides to a high degree with VO; see e.g. here (PDF, page 2)).


JAL

[linguistcat]

Thanks, I appreciate it. I will check out both pdfs you linked. If I have further questions I'll probably start a separate thread.

[jal]

Thanks, I appreciate it. I will check out both pdfs you linked. If I have further questions I'll probably start a separate thread.

Sure, good luck with your conlang :).


JAL

[Jonlang]

Would it be naturalistic to have both the dative case and the allative case in a language? My L conlang is going to have a Finnish-like case system, but when doing some research into Finnish cases I realised it doesn't have a dative but uses the allative instead. I would rather use both, even if it means the dative is some sort of corrupted/reduced form of the allative or otherwise etymologically linked. I don't particularly want the allative to have the dative function, but I may if it's the naturalistic option.

[Moose-tache]

I think Hungarian has both? The allative isn't super common as a separate case (i.e. not a postposition), but when it does occur it doesn't seem like the dative is rare. Nenets has dative but no allative, so I'm guessing Finnish and Estonian lost the dative secondarily after the allative developed, implying that there may have been a time when even these languages had both.

[Jonlang]

I think Hungarian has both? The allative isn't super common as a separate case (i.e. not a postposition), but when it does occur it doesn't seem like the dative is rare. Nenets has dative but no allative, so I'm guessing Finnish and Estonian lost the dative secondarily after the allative developed, implying that there may have been a time when even these languages had both.

Thanks, I'll take a look into those.

------------------------------------------------------------


So I have another question: is it naturalistic for a case-heavy (a few more than Finnish) to not inflect adjectives to agree with the noun's case? Because, in the conlang I propose that the adjective will never be separated from the noun anyway, like even though the word order is pretty free (but has an underlying SOV order) the adjective and noun need not really be separated, so I see no reason why it would always need to agree except in number.

[bradrn]

So I have another question: is it naturalistic for a case-heavy (a few more than Finnish) to not inflect adjectives to agree with the noun's case? Because, in the conlang I propose that the adjective will never be separated from the noun anyway, like even though the word order is pretty free (but has an underlying SOV order) the adjective and noun need not really be separated, so I see no reason why it would always need to agree except in number.

Huallaga Quechua has 10 cases, and as far as I can tell it has no case concord (Weber 1989, A Grammar of Huallaga (Huánuco) Quechua) — not quite as many cases as you’re asking for, but still reasonably large. You may also be interested in this:

Mark Norris, A typological perspective on nominal concord, concludes that 30% of languages with concord within the noun phrase, and 25% of languages with morphological case, have case concord. (I didn't see a reference to areal or genetic patterns, but I also skimmed pretty ruthlessly.)

[Travis B.]

So I have another question: is it naturalistic for a case-heavy (a few more than Finnish) to not inflect adjectives to agree with the noun's case? Because, in the conlang I propose that the adjective will never be separated from the noun anyway, like even though the word order is pretty free (but has an underlying SOV order) the adjective and noun need not really be separated, so I see no reason why it would always need to agree except in number.

I have heard it posited (don't call me on this) that Finnic borrowed case concord from IE myself, and have heard it stated that case concord is not normal for case-heavy languages.

[Moose-tache]

I think any language where adjectives class more with verbs than nouns will lack concord. I'd have to check, but I'm guessing something in NEA might fit the bill.

[Travis B.]

I think any language where adjectives class more with verbs than nouns will lack concord. I'd have to check, but I'm guessing something in NEA might fit the bill.

I would tend to think the same myself.

[zompist]

Imbabura Quechua has 10 cases, and as far as I can tell it has no case concord (Weber 1989, A Grammar of Huallaga (Huánuco) Quechua) — not quite as many cases as you’re asking for, but still reasonably large.

Sounds like you're conflating dialects or sources there— Imbabura is in northern Ecuador, Huallaga is in north central Peru.

But the point is correct: Quechua adjectives aren't inflected at all.

FWIW Wikipedia lists 18 cases for Cusco Quechua. Quechua is full of suffixes and I'm not convinced that all of Wikipedia's list are in fact cases. E.g. -lla, which they call an exclusive case, is also applied to verbs.

[bradrn]

Imbabura Quechua has 10 cases, and as far as I can tell it has no case concord (Weber 1989, A Grammar of Huallaga (Huánuco) Quechua) — not quite as many cases as you’re asking for, but still reasonably large.

Sounds like you're conflating dialects or sources there— Imbabura is in northern Ecuador, Huallaga is in north central Peru.

Yes, you are entirely correct. I checked an Imbabura grammar first, then found what I was looking for in Weber’s grammar of Huallaga. Should be fixed now.

[Jonlang]

Thanks for the feedback, folks! Very helpful.

[Ares Land]

I have a question... OK lots of questions for creators of logographic scripts, syllabaries and other non-alphabetic scripts.

If you draw these on the computer, do you use any specific programs, ie do you use Photoshop/GIMP or a font creator?
And if you have a font creator, do you have any recommandations (also, where do you put the glyphs in Unicode?)

And finally, do you have any good methods for logograph input to share?

[Kuchigakatai]

I have a question... OK lots of questions for creators of logographic scripts, syllabaries and other non-alphabetic scripts.

If you draw these on the computer, do you use any specific programs, ie do you use Photoshop/GIMP or a font creator?
And if you have a font creator, do you have any recommandations (also, where do you put the glyphs in Unicode?)

And finally, do you have any good methods for logograph input to share?

I have no experience whatsoever in this area, but rehashing things I've heard...

Clawgrip used to recommend shelling out the USD $150 for the standard licence of High Logic FontCreator, but I don't know why (it costs $200 if you want to use automated kerning, which seems nice). I once asked a professional designer who does Korean and Chinese fonts (and occasionally Greek for fun), and she recommended High Logic FontCreator due to its "opentype features" (which she said would come in handy if I were to make an Arabic font—my interest at the time) and mid-range cost, or if you have more money to spend, FontLab.

Many people seem to use Inkscape (which is free) to do the vector drawing when it comes to the glyph shapes, but I think both FontCreator and FontLab come with their own vector drawing tools anyway.

I recall Guitarplayer saying that he found it shameful that Unicode in practice is very much intended to be used with natlangs, and that the best way to encode a conscript is to put the glyphs in something with similar properties. So if you're using right-to-left, use the Arabic range. I don't know why this was or even if I'm misremembering something (shouldn't the Private Use Area allow just about any property?). (EDIT: Actually, I'm vaguely remembering it may have had something to do with the not-so-widely-supported Graphite technology he was using.)

Clawgrip found it easiest to just map his logographic conscript to Japanese kanji with similar meanings, as he knew Japanese anyway and could use a Japanese input method. I once heard some Classical Chinese nerds saying they were looking into Linux Rime to be able to map Classical Chinese characters to Middle Chinese pronunciation, using a Middle Chinese romanization as the "pinyin". Rime is unfortunately meant for Chinese only, which means the documentation is only available in Mandarin... If you know how to code you could probably whip up your own romanization-to-conscript GUI converter though, in theory you just need to load a key-value mapping.

[Kuchigakatai]

As for free-of-cost font-making software, FontForge exists but even in 2022 I've still heard people complaining about its learning curve. Birdfont, I've heard, is easier but buggier. And GlyphrStudio is rather laborious in terms of UX (though it's entirely online, a curious convenience!), with kerning supported rather indirectly due to limitations in the free library it uses (you need to export your project as an SVG font and then convert it to OpenType with something else). There's others.

[Richard W]

I recall Guitarplayer saying that he found it shameful that Unicode in practice is very much intended to be used with natlangs, and that the best way to encode a conscript is to put the glyphs in something with similar properties. So if you're using right-to-left, use the Arabic range. I don't know why this was or even if I'm misremembering something (shouldn't the Private Use Area allow just about any property?). (EDIT: Actually, I'm vaguely remembering it may have had something to do with the not-so-widely-supported Graphite technology he was using.)

Yes, the PUA should, but it doesn't. The problem is with loading the properties of the points. The interfaces that exist tend to be at the code level, and not accessible at run time. Furthermore, PUA settings should be at the font level, as without a font the PUA is pretty meaningless. The PUA works if the default CJK properties are OK, so it can work reasonably well for a left-to-right script, but if it is complex you will probably need a HarfBuzz renderer, though just possibly the rlig feature will be enabled. (I think in general, if Graphite is available, you are dealing with a HarfBuuzz renderer anyway.) Right-to-left PUA is likely to be a world of pain.