An essay on diachronics in fictional language construction

Originally published in issue 2 of the Conlanger.com Journal, July 2006.


If I could give one piece of advice to a newbie conlanger looking for realism, it would very likely be this: make sure your language contains at least some historical information.

There are a number of reasons for this opinion. Perhaps most obviously, it makes things far, far easier when one decides one wishes to realistically create some relatives – or even dialects – for one’s language. I know I’m not the only conlanger to have spent many years creating a language, decided I wanted to create relatives, and ended up faced with the horrible tasks of working out sound changes and so forth backwards – Mark Rosenfelder (alias ‘Zompist’) is probably a well-known example. In my case, I actually gave up in the end, even with the help of the incredibly useful ‘Reverse Sound Change Applier’, and decided to completely rework the language (Vixen, in case you’re wondering) from its historical ancestor forwards – though maintaining a fairly significant amount from the language’s previous incarnations. On the other hand, there is a certain amount of good feeling associated with successfully working out changes backwards and then forwards again – I’ve probably never got a bigger kick out of conlanging than when I successfully worked out the first sentence of Vixen’s relative Monronese, after what must have been several hours’ work. Yes, several hours, for one short (five-word) sentence. I, for one, don’t really feel that the unthinkably vast amount of work that would have been necessary to create the language in something close to entirety would have been worth the jubilation I’d have got out of it.

My second reason for stressing the importance of diachronics is that, quite frankly, they add realism. I’m not sure it’s possible to really come up with a truly realistic language with no idea of the history behind it. It doesn’t matter if you want to create relatives or not – a language’s past makes it what it is today. Diachronics help to explain the irregularities and the oddities. Why does English /dɪvaɪn/ have /dɪvɪnɪti/ as a derived form? Why are the singular and plural forms the same for the second-person pronoun? Where the hell did all those strong verbs come from? You won’t find the answers to these questions in the present – you’ll find them in the past. A conlang that doesn’t contain similar ‘weirdness’ lacks reality; but if it contains the weirdness without the (historical) explanation it lacks reality as well, because these things generally come from somewhere – though obviously, something had to be just ‘invented’ at some point. This should be the case in your languages, as well. There has to be a point where things are either (a) completely regular and ‘normal’, or else (b) the irregularities were pulled out of thin air. But if you want as much reality as possible, put that point far, far back in the past, so that the current incarnation of your language hopefully has most of its peculiarities drawn from somewhere. Because, to reiterate, the peculiarities with history are generally superior to those without – certainly, you don’t want too many of the latter (people will sometimes do strange things with no apparent motivation, of course).

You may well ask, however: ‘if I’m going to have to display alanguage at some point in history that isn’t wholly realistic, why not just make that language in the present?’ My answer to this is that you don’t have to display the most artificial stage of your conlang. For me, a protolanguage is very much something to be kept private – one analogy I like to think of is that however much you publicly display your babies, you probably won’t go around conceiving children outside of the privacy of your own home. You create the oldest stage of the language in private, and only publicly display what it grows into. Neither do I tend to advertise the changes between different stages of a language or language family – I will only display different synchronic incarnations of related languages or a language at various different points (generally only those within recorded history, and not anything that in real life would require reconstruction), and not what happened between those incarnations. Of course, if you wish to do it differently, that’s up to you – but my view is, if people really want sound changes or a proto-lang, they can bloody well work them out themselves. I will, however, occasionally reference a certain change or ‘reconstructed’ proto-word in grammar documents, to help illustrate a point – but these are never more that tiny windows into the totality of my creation.

The third (and final, due to time and space constraints) reason I will give for the significance of language history is quite simply that it can be quite enjoyable. Though I cannot locate the exact quote, I seem to remember Mark Rosenfelder writing of the enjoyment arising from creating a language with its own unique ‘feel’ simply by applying a certain set of changes, and I myself have had this experience. On the other hand, I must admit that the novelty of inventing long lists of sound changes has now worn off, and it does tend to get rather boring (morphosyntax is more fun – I reference the memorable afternoon of ‘beating up’ Proto-White-Vixorian’s fairly complex morphology through application of phonological change and analogy) – but try not to let that put you off. Even with what can seem at the time unnecessarily hard work, the end result is just as pleasurable, if not more so.

Remember that diachronics is about more than just sound changes. Certainly, they are a big part, but I have read ‘advice’ that seems to suggest that all one needs to do to create a language’s descendant is to stick its lexicon through a computer program, work out how the morphology would be affected, and make sure the name is changed throughout the grammar document. Changes in pronunciation cannot adequately describe the differences between Old and Modern English – there are plenty of other things at work, like lexical borrowing, semantic shift and analogy. And remember also that although sound changes are almost completely regular, it is only almost. Always allow for a few exceptions to your sound change code.

One thing that might put people off diachronics is that it can make revision an awful lot harder. You don’t just change one set of word endings, you change dozens. But I generally find that as long as grammar documents are laid out coherently, making changes across a whole language family can be relatively easy – certainly when compared to having to create a proto-language backwards. Don’t worry about it.