I’ve set aside plot-work for The Curse of Steel for the moment, so I can once and for all get the constructed-language work for that story knocked out. The idea is that my protagonist is going to encounter not only her own culture but several others as well, most of them somewhat related to her own in linguistic terms. Kind of like an Iron-Age Celt visiting Latin-speaking or Greek-speaking areas; the languages wouldn’t be intelligible to her, but names and some bits of vocabulary would sound hauntingly familiar. I’m also aiming for the reader to feel comfortable with the names they find in the story, which suggests not wandering too far from the Indo-European tree.

The procedure I’m working is to develop a partial constructed language that’s somewhat reminiscent of Proto-Indo-European (PIE), and then to apply a consistent set of sound-change laws and grammatical changes to generate words in two or three daughter languages.

Not at all difficult, especially once I’ve developed some computer tools to automate the process, but it is kind of detail-driven and time-consuming. The biggest potential pitfall is trying to imitate PIE too closely. One thing we do know about the reconstructed PIE language is that it made Classical Greek or Sanskrit look simple in comparison. My constructed languages for this project are going to be a lot less fiddly and complex. They’re just going to be naming languages, for the most part, so I don’t need to have a bunch of linguistic complication, I just need to be able to hint at it in a plausible manner.

So far I’ve developed a tool (a Perl script of about 120 lines) to generate all the “legal” word roots in the ur-language (about 150,000 of them, more than I’ll ever use). I’ve dumped all of those into an Excel spreadsheet which now serves as my master list for future lexicon-building.

I’m currently working on a partial description of the ur-language, with special attention to morphology: just how do you form verbs or nouns from word roots, how do the verbs conjugate, how do the nouns decline, and so on. As soon as that’s more or less finished, I’ll be building another Perl script (or maybe two) to automatically generate verb conjugations or noun declensions as needed.

The last step will involve developing two or three sets of sound-change laws, so I can take completed words in the ur-language and create daughter-language words from them. Another Perl script for that, I think.

Once all this work is done, my constructed-language workflow will get a lot simpler, and hopefully more consistent. Do I need a name or a bit of exotic vocabulary? Build a word in the ur-language, by selecting a legal root from the list and applying the defined morphology rules. Then run that through the sound-change script to generate final lexicon entries. Everything goes into a set of Excel spreadsheets, so I can sort and massage the results as needed. If I ever feel like tweaking the structure a bit, it becomes easy to modify the scripts and re-run everything.

I find when working on world-building, having the technical knowledge necessary to produce plausible detail isn’t the most important thing. That part is relatively easy. The hard part is scoping the task so that you can plan it out from start to finish, and then having the discipline to finish the task and move on. World-building is a neat hobby. If your goal is to actually write stories, you can’t permit yourself to get buried in the world-building. As I’ve learned to my cost.