How Not To Conlang?

[Salmoneus]

The original form of the verb is teozok. [Words in the distant past had a regular CVCVC form, as in Austronesian, but the stressed open first syllable lengthened, and then the long vowels broke into a variety of diphthongs].

There are originally three persons (-/, -t, -k) and two numbers: singular, and plural (-i). But clusters become geminates, and geminates merge with singletons, so person is no longer marked. As the language is VSO, but moves to SVO, subject pronouns bond the verb to replace the old person marking (this is why person is marked on the 'wrong' side of the number suffix). This has a couple of oddities: the 1st person singular pronoun is not merged, and the third person plural is itself formed from the singular with a plural suffix (so plurality is double-marked in this form). Also, slightly confusing, the second person plural pronoun (and hence now affix) was -i, the same as the old plural marker. This gives us the paradigm:

teozok
teozok-ur
teozok-un
teozok-i-na
teozok-i-i
teozok-i-uni

From this, sequences of identical vowels are simplified, giving a second person plural form 'teozok-i'. This seems a pretty reasonable paradigm, I think?

Now, stress is placed on the the antepenultimate mora (or preceding vowel for coda sonorants). Hence:

téozok
teozókur
teozókun
teozókina
teózoki
teozokíuni

However, it's illegal to have more two unstressed syllables in a row at the beginning of a word, so a rhythmically-produced secondary stress becomes primary for the third person plural: téozokiuni

All well so far. Now things get a little complicated. Final vowels drop after sonorants. Umlaut creates diphthongs (uCi > ui, etc). Unstressed diphthongs simplify (in this case, eo > e, iu > i, oi > i). Thus:

téozok
tezókur
tezókun
tezóikin
teóziki
téozikin

Still OK, right? Now, final nasals drop. That's followed by the loss of unstressed vowels, with two exceptions: vowels forming the second element of diphthongs are not lost, and in sequences of two unstressed vowels, only the second is lost. Thus we have:

téozk
tzókr
tzókn
tzóik
tózik
téozik

From here, it's no surprise at all when same-syllable clusters are simplified - the first element of a word-onset cluster is dropped, while the second element of any coda cluster is dropped. The diphthong -oi- shows height levelling to become -oe-. And of course the loss of initial unstressed vowels has made stress non-contrastive. Hence:

teoz
zok
zok
zoek
tozik
teozik

The second person singular is replaced by the plural - due to politeness constraints, strengthened by the need to disambiguate following its merger with the third person singular. And final fricatives are lost. Hence:

teo, tozik, zok, zoek, tozik, teozik. Quod erat producendum!

All perfectly regular, naturalistic and above board.

[your pronunciation guide suggests some further, very believable changes after the orthography became fixed: /e/ > /j/ as a diphthong onset, /tj/ > /s/, and the high vowel /i/ undergoes breaking > /@i/ > /ai/].


Anyway, the same processes can operate throughout the verbal paradigm. I'd also suggest that since, coincidentally, eo-o and iu-u verbs happen to be the most productive categories, their vowel alternations, which superficially resemble metathesis in the first person plural, are then generalised as metathesis to the rarer verb classes. So, for instance, 1p of kuatak should etymologically be **taik (or taak, depending how far the diphthong levelling goes), it actually ends up tauk instead, by analogy. [forming a fun inversion-alternation with the 1s form, kuat].

[alice]

Gosh, I'm flattered. Unfortunately none of these reconstructions would work for any other verbs, but never mind.

For those of you that care: https://web.archive.org/web/20060923075 ... ahaha.html. Deity-VOC, this brings back memories.

[Salmoneus]

Gosh, I'm flattered. Unfortunately none of these reconstructions would work for any other verbs, but never mind.

Well, it works for any EO-I or IU-I verb, and many other verbs can do similar things through analogy.

Your verb 'maaron' could be made to work similarly either by giving it an original long vowel in the second syllable, or by assuming a second conjugation with o-umlaut instead of i-umlaut, somehow (perhaps original long 'u' is umlauting and short 'u' isn't?). That way you could have a conjugation maa(r), maron, man, maan, maron, maaron.

[Imralu]

I'll posit one rule:

• Don't make a conlang you don't like!

Even that has a caveat though - it can kind of be fun or funny to make something deliberately awful. But you're still getting enjoyment from it somehow.

Anyway, since conlangs are mostly only appreciated by their authors, our own opinions matter more than anyone else's. Yes, we do discuss our conlangs with each other and tell each other what we find interesting and what we don't and we give each other ideas and teach each other about possibilities and whatnot, and that does lead in certain directions that a lot of us agree on, and if you're looking for approval within the group, you might be disappointed, but if that's not important to you ... ultimately, unless you have some specific other goal in mind (creating the IAL that changes the world, for example, earning the ZBB's Gold Star of Brilliant Conlanging) the whole reason most of us have this this funny hobby is our own enjoyment. If you really enjoy your relex of English with an unnatural lopsided phonology, yay! The rest of us might not find it that cool, but if you enjoy it and you haven't set any other purpose for it than personal wheeeeee, you are enjoying conlanging and is that not the goal?

[Pabappa]

• Don't make a conlang you don't like!

Even that has a caveat though - it can kind of be fun or funny to make something deliberately awful. But you're still getting enjoyment from it somehow.

Yeah, I've actually been doing that since my very beginning. Every language that I liked had a foil where the features I liked were missing or deliberately broken in such a way that they were "worse than having nothing". (Like marking tense on all verbs, but having just enough collisions that auxiliary verbs need to be used as well).

Every once in a while, though, a "hate lang" like that develops a feature that I like and isnt something I can shoehorn into a language I like. For example, Dreamlandic developed noun prefixes a- and i- from loss of /h/ and /g/ and vowel collisions, and I like that so much that I think it could develop into a new type of noun class system that I cant use in the other languages because there are either too many noun classes or too few (animacy only or animacy and gender) .

[Alon]

My biggest suggestion for conlangers is "know what you're intending to use the language for."

I imagine the vast majority of conlang usage is for SF/F media, especially stories we write and RPGs we run. If it's ever planned to be written, aim for a phonology that's reasonable to Romanize. For example, use 2 or 3 vowel heights and not 4 as in Portuguese or Bengali, unless you're running a game for players who speak a language where <é ó è ò> will be understood readily (remember, the vast majority of readers will forget the accents). It's fine to have things that are not English phonology: Cthulhu is not a legal word in English phonotactics and yet Anglophone geeks know and remember the name. But it should be transcribable, somehow.

You can go further away if it's a language that's explicitly intended to be hard for the POV culture. For example, I have a phonological outline for a language used by dragons, who, with their long snouts, have so many places of articulation that there are 192 consonant phonemes and another 96 clusters. It's meant to be arcane, and the more down-to-Earth speakers of Draconic, the lizardpeople, have had to greatly simplify the consonant inventory to be able to pronounce it and invent new expressions to distinguish words that to them are homophonous, yielding a Vulgar Draconic variant that they feel is inferior to the original version. Here the difficulty of nailing down Draconic - if it's ever written in a story it will probably have a bunch of diacritics on the consonants - is intended to reflect how humans and humanoids find the language and how lizardpeople have an inferiority complex toward it.

Because conlangs tend to be used for naming more than for anything else, you can go nuts with the syntax. You can make learners memorize a ton of different inflections and irregular forms, because for the most part these are Easter eggs for the story: "aha, there is an etymological connection between the castle of Dayafa and the name Defe."

[gestaltist]

Don't get bogged down in complex phonology and fail to progress any further into your language. Phonology can be a roadblock. But it's not the be-all end-all.

Unfortunately I see that sometimes.

I needed to hear that today. Thanks.

[Pabappa]

I think another thing to avoid is badly mixed conlangs. If you like French and Japanese, that's great, and if you want to make a language like French and a language like Japanese, that's great too, but try not to make a language that's about half of each unless you really know what you're doing. Especially with the phonologies. About ten years ago there was a book series called the Wheel of Time by Robert Jordan, and his language was called the Old Tongue. It wasnt the worst conlang in the world, to be sure, but the author puffed it up as if it was one of the best. He even considered his work to be on par with Tolkien's.

One thing we agreed on that was wrong with the Old Tongue was that it tried to imitate several different natlangs at the same time, so that some words looked like French, others looked like Arabic, and others looked like English. It just didnt work. I think a lot of people's first conlangs may be like that, but a lot of us started out when we were ten years old ... to put something like that in a finished work and then consider it to be the pinnacle of human achievement shows that you really don't know what you're doing.

edit: i see Robert Jordan is no longer with us .... I think this was a little further back since we wouldnt have been bashing someone after he was dead. All criticism still stands, as it applies to the people who defend his conlang as being among the all-time best, but I'm not sure that's such a heated issue these days.

[Salmoneus]

I think the problems with the Old Tongue were primarily:
a) it was indecipherable. It had the common (see GRRM) Faentaesye Voewels Proebleim, made worse with constant apostrophes (that sometimes looked like they might be glottal stops, but other times were clearly just apostrophes). It made it very difficult to guess how anything was meant to be said. [it's OK if people pronounce your conlang 'wrong', but if they don't know how to pronounce it then it's like a little slap in their faces every time they encounter it]

b) it was bland. I wouldn't go so far as to say it looked Arabic (beyond the al' prefix) or French, it just looked generic fantasy. It had no particular distinctive feel, in my opinion. Which I guess is another way of saying different samples were in conflict, but I never actually felt there was a clear conflict between different, as it were, modes - more that there was a crisis of identity throughout the language in which it never really made itself either coherent or distinctive.

[Vardelm]

b) it was bland. ... It had no particular distinctive feel, in my opinion. ...there was a crisis of identity throughout the language in which it never really made itself either coherent or distinctive.

I wonder if conlang guides could help more in this area. I don't think I've read much (anything?) on how to develop an "identity" or "distinctive feel" for a language.

This is something I've tried to do a little more in my recent projects. I wanted a pan-African/Arabic feel for my Jin language, and I feel like it worked out well (at least for the main audience: me. ) When I looked at what I had for Devani, it felt incredibly bland, so that will become more Indic. Dwarvish is taking inspiration from Celtic & Nordic langs.

Coming up with something that has a unique identity - one that doesn't copy real languages - seem non-trivial. IMO combining aspects of 2-3 different natlangs seems like a good way to go. The main point it to not overdo it so that there are a few key features that can stand out rather than being swallowed by too much noise.

[Raholeun]

In short; think carefully about your phonology and the romanization. Make your phonotactics coherent and the transcription plausible yet distinctive.

[gestaltist]

In short; think carefully about your phonology and the romanization. Make your phonotactics coherent and the transcription plausible yet distinctive.

I feel the two things necessary to get that coherent feel are morphophonological processes and diachronic sound change. Both are a lot of work, and from what I've seen posted on the internet, most conlangs don't go very deep in either direction.

Thinking about morphophonology requires diving a bit into phonotactics, cluster resolution, and (possibly) prosody. Having diachronics introduces those lovely irregularities, but also cognates that don't look quite alike but "feel" connected.

(EDIT: I don't think going very deep on either is required, and there are instances where you might go lighter on one of the two - eg., a language with a high level of analogy might obliterate much of diachronic irregularity. But it's more common for conlangs to not have enough than to have too much in this area...)

[alice]

IMO combining aspects of 2-3 different natlangs seems like a good way to go.

Heh. I do that all the time!

The main point it to not overdo it so that there are a few key features that can stand out rather than being swallowed by too much noise.

It's a bit like cooking a steak with flavouring so that the flavour doesn't all end up at one end. How's that for an analogy?

Back to Bad Conlangs: I get the impression from people's replies that it's probably impossible to create a conlang which everyone will consider to be Bad. Although it might be an interesting exercise (but not for me!)

[akam chinjir]

Back to Bad Conlangs: I get the impression from people's replies that it's probably impossible to create a conlang which everyone will consider to be Bad. Although it might be an interesting exercise (but not for me!)

Relex Esperanto?

[alice]

Back to Bad Conlangs: I get the impression from people's replies that it's probably impossible to create a conlang which everyone will consider to be Bad. Although it might be an interesting exercise (but not for me!)

Relex Esperanto?

Well, there is that, but surely you want some sort of a challenge?

[Imralu]

Well, there is that, but surely you want some sort of a challenge?

Relexing esperanto with a priori vocab would actually go a long way towards fixing its eurocentrism

[Alon]

Back to Bad Conlangs: I get the impression from people's replies that it's probably impossible to create a conlang which everyone will consider to be Bad. Although it might be an interesting exercise (but not for me!)

Volapük!

[mèþru]

Hey I like Volapük. Minus the sexism

[Salmoneus]

Everyone likes Volapuk. Volapuk is a great language. And unlike the conlangs of everybody else here, it was once learnt by millions of people.

[Raholeun]

It was once learnt by millions of people.

[citation needed]