deriving conlangs, generally

[LinguistCat]

When deriving a conlang from an existing natural language or a reconstruction of a proto lang (natural or not), how do you go about it? Do you focus on getting a specific sound, or having certain features (ex: deriving gender from a language that didn't have gender), or do you choose sound changes you think you would like and reanalyze things as you go?

Do you focus on certain word types to have a specific change you're looking for, and then just apply the same rules to other words and see what happens?

Does it depend on the language and goals you have for the language or depend on the language you are deriving it from?

I realized I'm having problems with direction for my current project. Part of it is that I've never liked how languages I've made ended up coming out when I started from nothing, but I've also never derived a language from something already existing. But I also recognize that my goals for the language (sound "cat-like" and be related to Japanese) were extremely vague. I'm just trying to see how other people have tackled this so I can get a better handle on how I want to try it

[Zaarin]

I usually have an idea of what I want the daughter languages to look like before I create a proto-lang, then I create the proto-lang, then I create the sound changes necessary to give me what I originally planned for the daughter languages (or something like it anyway).

[Frislander]

I tend to go in totally different directions with a priori versus a posteriori. With a priori I tend to start with the daughter and work on the parent either alongside or after I'm creating the daughter. With a posteriori I prefer to read all about the proto-language and the isoglosses found in the daughters and then I decide what I like from the daughter languages (e.g. my Algonquian language is a mix of my favourite bits from Arapaho and to a lesser extent Cheyenne, as well as a few innovations of my own, while my Iroquoian language shares a good deal with Tuscarora, Huron and Onondaga).

[mèþru]

I've never done such a language to the scale of more than a few sound changes and then forgetting about it. alice wrote that there are two methods: the historical method, in which one applies the target languages historical sound changes on the source language; and the adaptive method, in which one uses sound changes that approximate how a native speaker of the target language would render the source language. Since there is no target "cat language", the adaptive method is the only one that can work; you can try to adapt the sounds of some historical Japonic to the vocal system of your cat people. After doing a search here and on CBB for what you have previously written, it seems like it would make more sense for them to speak an a priori language using typical cat communication (cat phonemes, cat body language, etc., regularised with syntax) when in cat form among themselves and the local Japanese dialect when in human form. Interestingly, the first appearance of cat beings in Japanese literature is a journal from August 2, 1233 (where a mountain cat creature is said to have killed and eat several people in Nara), but Chinese literature about them goes all the way to the Sui dynasty. The early Japanese source may have in fact been referring to dog or wolf with rabies, but other sources from the same century also talk about cat monsters. Perhaps the divergence could be with Middle Japanese or Early Middle Chinese, although, again, they probably had some language before human contact. You could justify having a common linguistic ancestry if they originated from humanity in some manner (a curse, a blessing, undeath, etc.)

[mèþru]

@Frislander: is the Iroquian language part of Conspiration des poudres?

[mèþru]

Cat's mouths are smaller, so they might be used to not that much sound distinction within a region of the mouth. Also, their lips seem to be less exaggerated than humans and therefore might be less useful for making sounds. Most sounds I have heard from cats seem to come from alveolar to palatal area. Maybe they would tend to allophony between palatals and all other dorslas and the guttural places of articulation and also overuse labials.

Idea:
With Early Middle Chinese, this could result in a merger of dentals, retroflexes and alveo-palatals into a single coronal category while palatals, velars and laryngeals merge into a single dorsal category. This would cause many words to become homonyms, so it would develop compound forms much before other Chinese varieties. The cats could then slowly travel from Jinhua, reaching Nara by the 13th century. During that time, cat Chinese would fade away into a substrate of the cat's own language.

[Jonlang]

I start with a phonetic inventory of the language I want to make. Then I find out realistic ways for those sounds, in the clusters I want, to occur. I then use this to come up with a phonetic inventory for a proto-lang, which is usually a fair bit larger than its daughter langs. For building words I create bases, from these bases I apply seemingly meaningless phonemes, I then apply the sound changes to get a word I like.

Most bases are (C)V(C) and can sometimes, but not often, be (C)VCV(C).

I then add "suffixes" to these. The meanings behind these suffixes are not discernible by the time the daughter conlang word is formed, sometimes it may even disappear. There are also other "manipulations" which can be done to the base, i.e. nasalisation, the placing of a nasal before a plosive; nasal-strengthening, the placing of a voiced plosive after a nasal; fricitising, the placing of /s/ before a plosive.... among others. The bases have a "feel" for meaning rather than a definition.

The base LEK has a "feel" of "wood", so from it I will form the words for "wood", "wooden", "tree", "(tree) bark", "carpenter", etc.

LEK > add a simple noun ending -wa via vocalic extension -e-wa = LEKEWA > manipulate the base by adding /s/ before the /l/ = SLEKEWA. Now I have the Proto-PQL word 'slekewa' for 'wood'.

After the sound changes this pops out as 'lhagu' /ɬagu/ in the "P" language and 'eslequë' /ɛslɛkʷwɛ/ in the "L" language. (I think this is accurate, I haven't checked).

Another word may use the same -wa noun deriver but use no base manipulation, or a different one:

LEK + -wa (no vocalic extention) + nasalisation = LEŋKWA. The Proto-PQL word 'leŋkwa' means 'tree'.

This pops out as 'lamp' /lamp/ in the P language and as 'lenquë' /lɛŋkʷwɛ/ in the L langauge.

[Dē Graut Bʉr]

For my Iropo-Antilonian language family (basically my conworld's Indo-European), I have a general idea of how each branch develops, though I don't develop any descendants until I get more specific ideas. When that happens, other sound changes and grammatical changes tend to suggest themselves, and then a whole new language appears more or less out of the blue.

Incidentally, Proto-Iropo-Antilonian is the result of years of trying to figure out what the ancestors of Bearlandic were like, not unlike what Zompist did with Proto-Eastern and Verdurian.

[Curlyjimsam]

I think I tend to have a mix of having specific end goals in mind and just letting the sound changes run their course. An example of this would be the language family whose sound changes I worked on most recently, where I wanted one set of daughters to have more fricatives and front vowels, and another to have more stops and back vowels. But other than this, and the general property of the family that there are a lot of complex consonant clusters, I didn't have any particular end goals in mind - I just tried to work in sound changes that seemed particular natural in the context of the systems taken as wholes, and threw in various others for good measure.

Other families go a lot further back, so I can't promise my memories are reliable. I think the prevalence of syllabic consonants in Koshenian was a deliberate decision, but I didn't think much about what the phonological character of the language should look like in advance beyond this. For Greater Atlian, I'm pretty sure I deliberately went for the overall strong preference for CV syllables, and the development of the /ʃ/ and /ɬ/ phonemes was definitely planned in advance, being something I wanted to preserve from an earlier version of the language (which lacked any diachrony). But many other features just arose as I went along.

For Viksen, going back even further, there were also a number of features I think I wanted to preserve from earlier versions, e.g. palatal consonants, and if I remember correctly things like a preference for CVC and CVCVC words and voicing of most medial plosives, so I worked these into the sound changes. But many other features weren't planned in this way - e.g. an awful lot of words begin with /j/ or /w/, which wasn't something I planned, it was just a consequence of the sound changes as I ended up arranging them.

With morphology and syntax I often have a clearer idea of where I'm going - e.g. I deliberately arrange things to destroy inflectional paradigms or create an ergative system. But even there a lot of things just emerge as I'm going along.

[LinguistCat]

Thank you all for this input.

@mèþru: I do like the idea of them having come from China to Japan, as well as carrying a Chinese based language to Japan that would later affect their Japanese and/or the language they develop alongside Japanese. I was actually considering that they came over earlier than the first reports of their existence in the 1200s, but I'm not sure how much earlier would be reasonable. I'd have to do a lot more research about Middle Chinese since I really don't know anything about it outside of some effects on Japanese and of course a little bit from borrowings. I'll have to look into that unless I could have a good reason the cats in Japan would pick up Japanese but not Chinese like their precursors.

On a non-conlang note, there are stories of humans, for various reasons and at various times, becoming cat youkai, usually nekomata. So this could definitely affect things linguistically for the cats and the people becoming youkai.

[Frislander]

@Frislander: is the Iroquian language part of Conspiration des poudres?

Possibly, though frankly I kinda have no clue what's happening with it, it's kinda hard to place in the family (it's definitely Northern Iroquoian, but apart from that is also kind of its own thing), but I suppose it could be, particularly if I decide to completely redo it to make it more "traditional" by Iroquoian standards.

[Salmoneus]

What's key for me is finding a particular feel that I'm interested in. That applies to phonology, but also to grammar.

To take Wenthish, for example: years ago, repeatedly, I wanted to make a sister language to English, derived from Old English. But I never found something that really grabbed my attention. Then at some point, I had the core idea for Wenthish: a Germanic language in the north atlantic, reminiscent of English, but not necessarily its closest relative.

The setting gave me a few ideas to narrow down the 'feel' - there could be a nordic influence, there could be eth. The setting gave me more of an idea: windswept islands, deforested (mostly), rocky outcroppings, rain, fog, more wind. That didn't necessarily tell me what the language was going to be like, but it gave me, as it were, a measuring stick. With previous attempts, I never knew who spoke the language, so I never had anything concrete to aim at. Now I could keep in mind this measuring stick to see how much I intuitively felt the language matched the people.

Then I read up a lot on historical sound changes in Germanic. I basically started with the changes leading to Old English, and played around a bit - what happens if you skip this change, or expand that one, or shift the order around a little? I found there were a few changes that I wanted to keep from real life. I wanted intervocalic fricative voicing, for instance, and I kind of liked the idea of extending it to initials, as in West Country or Dutch. I liked the 'rural' feel of that. And I also found some interesting shifts I could add. The biggest one sounds small: redefining the allophonic rules governing early voiced stops/fricatives, so that there is no hardening after nasals. That matters because it means the ingvaeonic spirant law has a much wider effect than in English. That helps give the language its distinctive feel: Germanic, knotty, archaic, but also sort of soft.

So for me, it's not exactly fixing an objective and reaching it. It's having a general idea, and then trying to find developments that didn't just match that, but that helped solidify it and make it more specific.