Is diachronic conlanging worth it?

[GoshDiggityDangit]

I want to make a naturalistic language, but I don't want to make one full language for the time and effort of two. Do I need a proto-language (or two {or three [or four]}) to make a good naturalistic conlang? Do they need to be fully fleshed out? Or can I make something naturalistic without one (or two {or three...})?

[eldin raigmore]

It is up to you.

[elemtilas]

Much will depend on your end goal (if any, apart from "making a naturalistic language").

You don't néed a protolanguage. Most people on Earth get along very well, learning & using their moderately naturalistic languages, without ever learning about or even considering the existence of that language's proto-. What a proto-language will give you is the sense of time depth. All those curious changes that happen to a langauge over time: sound changes, grammatical fluctuation, lexical drift and also the sense that these things are not orderly. Earlier layers of language can affect later layers through borrowing and ressurexion of lost words.

An invented language does not need to be fully fleshed out. But again, it might need to be. And then, you'd need to consider what means "fully fleshed out". In order to answer these questions, again, much will depend on your end goal. Do you want to keep a diary? Do you want to translate the Bible? Do you want to devise an invented culture with its own histories, its own religions and create an entirely novel scripture? Do you want to recreate English literature over the last thousand years or are you happy with writing the equivalent of Basic English?

Basically what eldin said; but we really can't help you unless you give us a better idea where you're headed! This is like asking "can I walk to Berlin from here?"

[Dormouse559]

It's turtles all the way down. So you have to pick somewhere to start when it comes to nautalistic a priori conlangs, and it will be arbitrary. Whether you set that starting point 20,000 years in the past or just 20 is your decision. I gather from your post you're not interested in developing a past version of the language, so in that case, don't. It may occasionally be helpful to think about how certain features have changed over time, but you don't have to create a full-fledged language for that, or even a decent sketch.

[Salmoneus]

I think naturalism is more about processes than about end-points. We invoke proto-languages not just because it makes a nicer "end product", but because the process of thinking diachronically, making ourselves explain things through their history, is just part of the process. It's... fun.

Is it worth it? Well, is conlanging of any kind worth it?

But in terms of the final language, not using diachronics means you'll end up with one of the following:
a) a language without many interesting bits - quirks, oddities, irregularities, counterintuitive things
b) a language that has many interesting bits, but of a kind that plainly aren't meant to have plausible diachronic origins
c) a language that has many interesting bits that look like they're meant to have diachronic origins, but it's not clear what they might be
d) a language that has many interesting bits that do indeed look like they have plausible diachronic origins.

If you have a), many people will find it a boring and somewhat unnatural language (though, of course, natural languages do vary in how quirky they are). If you have b), it my be interesting, but it's not naturalistic - whether that matters is up to you. If you have c), it looks like a bad attempt to make a naturalistic language.

And if you do d), you can make a language just as interesting and persuasive as any diachronically-derived conlang. However, I think most would agree from experience that the quickest and easiest way to make it look like you've done the diachronics is just to do the diachronics...


Of course, as dormouse points out, "diachronics" doesn't necessarily mean "proto-language". Doing diachronics does imply a proto-language, but it doesn't necessarily require that you begin with an exhaustive description of a single, coherent proto-language.

[Zekoslav]

I concur with Salmoneus: for me the process of doing diachronics is even more fun than the end result... I find it more fun to play with the development of a language than to flesh out any of the stages of that development. And it's worth it, because as you decide one some changes, other changes suggest themselves... ever really weird changes. Proper diachrony really is a good way to generate interesting but plausible features. And it's enough for the proto-language to be a sketch, or even very similar to the descendant language... you don't need to do Proto-Indo-European to Modern Albanian to be realistic, Latin to Old Italian is more than enough.

In fact, you can just start working on the descendant language and the history of it's features will suggest itself: For example I made a language with a very simple phoneme inventory which consisted of /p t k f s x m n ŋ r/, and decided to add /t͡s/ to make it less symmetrical and more interesting. Then I got reminded of the High German consonant shift and thought, could /f/, /t͡s/ and /x/ derive from earlier /pʰ/, /tʰ/ and /kʰ/ ? Then I remembered that there way an universal that said languages with aspirate consonants usually have /h/ as well, so I decided to add /h/ which was later lost. Then I rembered rhinoglottophilia and said, hey, couldn't some instances /ŋ/ derve from that /h/, so that I don't derive it from consonant clusters like English did? Then I looked up on WALS and saw that distinctive aspiration was very rare in languages with small consonant inventories, and remembered Grimm's law, and derived /pʰ/, /tʰ/ and /kʰ/ from earlier /p/, /t/ and /k/ and /p/, /t/ and /k/ from earlier /b/, /d/ and /g/. So I've derived a proto-language sketch from my descendant languages sketch, no problem:

proto-language:
/p t k b d g s h m n r/

p t k > pʰ tʰ kʰ > f t͡s x
b d g > p t k
h > ŋ or nothing

descendant language:
/p t t͡s k f s x m n ŋ r/

Of course when I started doing phonotactics I invented more changes, so that the protolang had more allowed consonant clusters than the descendant language, and this leads to plausible, naturalistic allomorphy when suffixes are added to roots... (e. g. tn > nn but t͡sn > ss)

[eldin raigmore]

.... However, I think most would agree from experience that the quickest and easiest way to make it look like you've done the diachronics is just to do the diachronics...
Of course, as dormouse points out, "diachronics" doesn't necessarily mean "proto-language". Doing diachronics does imply a proto-language, but it doesn't necessarily require that you begin with an exhaustive description of a single, coherent proto-language.

Very to this.

Especially to the “proto-conlang needn’t be a complete conlang”.
For example, the protolanguage could be completely regular; it could be without sandhi or allophony; it could skip vocabulary for non-ancient nouns or verbs or modifiers or relators; and so on.

[sangi39]

I'm going to agree with basically everyone in the support for Sal's reply to this question.

1) Diachronic conlanging can, in and of itself, be pretty fun on its own (I enjoy it particularly because it can go along with world-building and history. Having a proto-language (or several) means you can flesh out the world as you go, without necessarily starting from scratch each time you want to create a new conlang, say, for the neighbouring country.



2) It's not necessary, but it can be helpful. I think I read somewhere that conlangers like D. J. Peterson have stated that a proto-language or diachronics are necessary in order to create a language that appears naturalistic (although I'm not sure to what they might mean when they say "proto-language". They could mean "relatively recent older stage" in the same vein as the sort of diachronics suggested by Dormouse above), but I disagree with that view.

As with a lot of conlanging (or any "craft" really), a lot of it comes down to research and experience. If your goal is "naturalism" (in the sense of "attested at least somewhere") then a really good starting point is just, well, to read, do some digging, look at everything you can. Eventually you get a feel for what sorts of things are naturalistic, and I think it's perfectly plausible to create a conlang that appears as if it has some features that look like they developed over time (so Sal's option d). I think at that point, though, you're probably thinking to yourself "this could have come from this" anyway, just not committing that thought to paper.



3) As Eldin points out in the first reply, it really is up to you. If you're just looking to create a quick conlang, then a small hint of diachronics might be worth the effort. If you're looking into creating an entire world, then throwing the start point back a couple thousand years might serve your purposes. It's also down to which option you might find more fun. If you enjoy diachronics, then go crazy. If not, then just think of it as a handy tool you can pick up if something's not working the way you'd hoped.

[k1234567890y]

it can be worthy, however, I tend to use the “proto-language” mainly to control the outcome of the modern vocabulary so that they would look more plausible and more coherent as words of the same language.

[Curlyjimsam]

The one really big advantage of having a proto-language is that it allows you an easy way of coming up with related languages, or of representing texts from earlier stages of the language if you decide to delve into the history of your culture. Trying to work backwards and create realistic relatives or historical forms with the "present-day" form as the starting point is really difficult. Even if you don't want to create relatives or a history just now, it could still be that you decide you want to do so in future.

(Having two or more related languages also lets you do other interesting things, like have these languages borrow words from one another.)

You don't have to flesh out all the details of the proto-language. I tend not to go into less detail with syntax, for example, as a lot of things I'm going to change anyway in ways that won't leave much trace of what was there originally. I keep the vocabulary quite small and more restricted to more everyday vocabulary, and derive or borrow most less frequent words. And I tend not to give my proto-languages much if any irregular morphology, on the assumption that it will be levelled out by the time of the "present-day" language. (Plus, proto-irregularities only lead to questions like "so where did that come from?")

How far back you choose to go is another question. I tend to aim for "around or just before the first written records" or "just before the earliest reconstructable point" (where you've got multiple related languages). The amount of work you end up doing will be related to what choices you make in this regard.

[Shemtov]

Having a protolang is not necessary, pee se, but having an idea of what the protolang looked like, and what any relatives look like, will help with naturalism
It doesn't have to be written down. My current project has many irregularities, but that's based on what I think the protolang looked like.