In a recent conversation with a number of other conlangers, the question came up of how different people approach conlanging, and how much diversity there is between different conlangers' preferred processes. Having heard that and thought about it a bit, I figured it might be worthwhile to write a brief overview of how I personally conlang, and touch on a few ideas and strategies that have influenced my work.
Given the limitations of time and endurance, I will not expound on how I create specific types of languages (although that would make for an interesting blog post series at some point) — only on the most general aspects of how I approach conlanging as an art. When reading this, please bear in mind that while I currently create mostly naturalistic or semi-naturalistic artlangs, I've gone through several serious engelanging phases, whose influence cannot be ignored even if engelangs are no longer my primary creative focus.
Many, many of my conlanging projects got their start when I first encountered a particular interesting natural language, and decided to build a conlang around a few features (be they phonological, grammatical, or other) of it. Some of these projects were, in hindsight, either laughable or boring — stealing the surface texture of a particular language without even attempting to replicate the underlying complexity that gave it that texture, and making up the rest of the language a priori and sometimes quite randomly — but a few have stuck around, and I come back to them from time to time when I feel like it.
What works best, it seems, is to choose a feature or two of an existing natural language — the vowel system, for instance, or some particular aspect of the verbal morphology — and graft that onto several other features taken from different language families, smoothing over any rough edges. Even so, however, this is simply a starting point, and as the language grows, it should (and must) eventually take on a life of its own. Furthermore, the more experience I gain conlanging, the more I find myself drawing only the smallest bits of inspiration from natural languages and instead building languages from the most basic of linguistic building blocks, without following any particular model (and only occasionally checking or particularly caring about ANADEWs).
Just about every single one of my conlangs contains at least a handful of words that are or contain references to some real-world person, place, or thing. I won't spoil any of them here, but this is something that I definitely do — if you've seen a word that looks familiar, you may well be right about its origin. Bad puns and other linguistic jokes are something that I am a huge fan of in the real world, and thus it is unsurprising that this comes out in my conlanging as well.
That said, there is such a thing as overdoing it. I rarely create languages as saturated in jokes as e.g. Klingon is, and I often disguise the joke to the point where it can't really be called a joke anymore, and is at best inspiration for that particular word — and in many cases, I no longer remember what that inspiration was. I've thought about creating a language or two composed entirely of jokes, and have started projects to that effect on a handful of occasions, but as of the present none of them have gone particularly far. Maybe someday I will revive some of them, but at the moment this isn't a major priority — humor and references are much more useful as inspiration when I know I need to create a word but just can't think of one, regardless of the language.
While sometimes I have a conlanging idea that feels too good to risk forgetting, those are the edge cases. Most of the time, I will let the seeds of a new language germinate in my head for days or even weeks on end, and only start writing things down once they have begun to take shape in a more concrete way. In this particular manner, my approach to conlanging is very similar to my approach to music or to writing fiction: I often prefer to imagine a significant portion of the work before starting to "fix" it by recording it; once "fixed", the threshold for making changes can seem higher than at the imagination stage and thus I may feel stuck with what I have already written down, even if it no longer fits.
While doing this, I may iterate through the same set of patterns again and again, trying to find a way to make them all fit together, even long after my conscious mind might have concluded that the problem is intractable and I need to take a different approach. For example, I might spend hours or occasionally even days mentally listing all of the phonemes that I'm prepared to add to a particular language and trying to organize them into a neat grid covering the different values of one or more grammatical categories, while adding and discarding various clusters or secondary articulations. At times, this can feel almost compulsive, like I've been seized by a need to solve that particular problem with no option to set it down for the time being, although thankfully this is relatively rare these days.
Many of my conlanging projects have started from the smallest of seeds: just a few words that I thought up one day, or a sentence, or the barest skeleton of a phonology that happens to inspire a particular aesthetic. For instance, Ruapai grew from the chorus of a lullaby, whereas Tolaviizh was ultimately built around the word for "face mask" (unsurprisingly during the first year of the pandemic). Although this is not a universal part of my conlanging by any means, it is common enough to bear mentioning in its own section.
As hinted at above, Ruapai — until recently aptly referred to as simply "The Lullaby Language" — first took shape as the first line of a lullaby, which I found myself singing to myself while I was having trouble falling asleep one bright Finnish summer night. The line kiraj tekturiu i required many hours of careful thought and examination to turn from a sequence of meaningless phonemes into an analyzable and meaningful sentence, but from it eventually developed the rest of the chorus of that lullaby, as well as some basic verbal morphology; more followed as I fleshed out the lullaby and eventually did my first Lexember in the language. However, despite its key role in the development of the language, I did not actually finish the lullaby until immediately before the LCC 2023 Relay, in which it served as a torch for one ring.
The situation with Tolaviizh may have been less poetic, but was ultimately similar. I had seen floating around on the internet a meme about people who didn't know how to wear their covid masks depicting a mask with the text "it goes over your nose!"; after seeing that, it occurred to me that "it goes over your nose" wouldn't be a bad way to translate 'face mask' in a verb-oriented, polysynthetic conlang with a syntax (and most importantly, nominalization strategy) partially inspired by Algonquian languages (and, in hindsight, also the verb-noun nominalization strategy in English that gave us terms like "cutthroat" and "pickpocket"). Again, this led to me fleshing out a phonology and morphology, and soon enough I had another conlanging project to think about.
If I had to name one of my worst traits as a conlanger (and as a hobbiest in general), it would be my talent for abandoning projects before they are anywhere near finished. While a conlang, like any artistic endeaver, is never truly done, I do tend towards the "lots of sketches and half-fleshed out ideas" end of the spectrum, compared to other conlangers who might devote years to a single language and develop it to a point comparable with a conlang that people actively use, such as Klingon. I've been trying to change this in recent years, but it's still very easy to go off and chase the most recent shiny new thing, and forget about the projects I'd been working on before that.
There are probably a million other random things that I could mention here, but I want to both keep this post to a manageable length and actually get it up on the blog in less than several months' time. Perhaps, as time and energy allow, I may expound on this topic more in the future.