Please see the Conlang Manifesto Manifesto before subscribing to any advice here. Unless you are me, this doesn’t really apply to anyone, least of all those writing conlangs as amusing reference grammars, auxlangs or conlangs to support the writing of a fictional novel.
However, my particular list may have applicability to revival of dead languages, especially the remnant sort, like Beothuk.
I’m writing this because it’s good to have some guiding principles when creating something.
A conlang should not have a conculture. The culture of a conlang is the culture of the creator and it’s fans.
A conlang should be finishable. If it takes machine generation to get the dictionary out the door, do it! Any feature or technique that gets in the way of publishing the materials necessary for a community member to start using the language should be dealt with as a top priority.
A conlang should be finished. A conlang is finished when you can write any thought that comes to mind, although it may be laborious or require following the languages instructions regarding word and phrase building, it should be possible.
A conlang should have nothing fictional driving its design. The fictional bits of a conlang will be rejected by your brain like bacteria. It shouldn’t rely on a fictional social order or a fictional mental model of the world (or at least nothing so fictional that even you don’t believe it).
A conlang designer should build a community. A conlanger should not sabotage his community by changing the language too fast after it has committed fans.
A conlang designer should innovate with care. The parts that *are* safe to change are the parts that no one is using. If your extended anaphora system is ignored, then go ahead and propose a new one.
A conlanger should not suffer from delusions of grandeur. Your language will not be language spoken globally and studied by scholars and toasted by diplomats and statesmen. It will not have a wikipedia section. It will never be taught in Universities.
A conlanger should not suffer from delusions of worthlessness. A conlang is valuable for the thoughts you can think with it, the peripheral skills you learn, for it’s potential use as augmented communication system (i.e. a conlang for the deaf & blind). A language doesn’t have to be the most successful language in the world, like English, or the most successful artificial language, like Esperanto to be valuable.
A conlanger should not worry about derivatives. The appearance of derivatives is the sign of a healthy conlang community.
If a language has fans, it is valuable– the community will speak, the community will be right.
A conlang is for humans. It should not have features specifically added to make it humanely impossible to use, such as requiring two mouths.
A conlang community should grow primarily through recruitment outside of the conlang enthusiast community. This isn’t to say they should be excluded, but they shouldn’t be the focus of marketing efforts. The conlang community has a glut of languages relative to the number of people in that community. The rest of the world is a monolingual wasteland of the under served. Get fans and the members of the conlang enthusiast community will follow without much more encouragement.
A conlang should not compete with existing conlang communities. Existing conlang communities are infintesmally small, often less than 5 active members at any one time. Likewise, designing your conlang to be too similar to an existing conlang leads to poaching, schism and worst of all– it isn’t a new way of thinking.
A conlang should be a collaborative project. As soon as the recipes for extension are done, the language should be released to the forces of nature.
A conlang shouldn’t put features design before community design. And by design I am speaking of something fuzzier than club rules, I’m talking about your personal plans of action regarding showing up in chat rooms, on skype, on twitter and so on.
A conlang should strive to enable an alternate way of looking at things. The more a conlang draws on existing languages, the less it has to offer. A perfect example would be a conlang that confers a good sense of direction by requiring everything to be place according to the cardinal directions instead of egocentric direction words. Variations on existing languages, unless they have some striking new feature, can’t help but to let its speakers stay carry on their existing mental habits.
A conlang should be easy in some sense. A language is like a tool. The designers of tools must consider their users when picking features. If it isn’t dead simple, it doesn’t serve the users. On the other hand, if too much of the simplicity comes from similarity to the target audience’s
A conlang community should designed for resilience. The language should have no dependency on it’s creator, any one website. It should not have it’s core materials lock up behind copyrights or trademarks, lest that it become illegal to freely distribute them.
A conlang should have no fictional linguistics. Claims of features should be tested. Reference grammars are the guidelines, but the corpus and the test performed on the corpus is the final word in what a conlang is.
A conlang should consider economics. It shouldn’t take so long to learn that only the independently wealthy can afford to spend time on it. It should be usable in any endevour, commercial or otherwise. If a conlang is used in a fictional work, and fictional cruft it picks up is only relevant to that fictional work and is irrelevant to the “canonical” form.