Conlanging and the Novel

1. Introduction

The hobby of “conlanging”, or creating “constructed languages” or “conlangs”, has a fairly sizeable following. One major function played by conlangs is to play a part in novels and other works of fiction. I would argue, however, that conlanging for this purpose is subject to specific constraints and ought to some degree be considered separately from conlanging in general. This essay, which is aimed at reasonably experienced conlangers as its primary audience, will discuss the reasons for this claim.

2. Realism vs. accessibility

It seems reasonable to claim that a major aim of many conlangers is something we might call “realism”. This most often surfaces in the attempt to create a language that is on one hand strongly distinct from the (Indo-)European languages with which their creators are most familiar, but on the other could nevertheless reasonably occur as an actual human language. Let us take two extreme, hypothetical examples of “bad” conlanging to prove the point.  Firstly, imagine a language which has the consonant inventory of English, the vowel inventory of Spanish, and translates the first few lines of the Tower of Babel story as follows:

Ken al entia darum aki an kiti e an deri spoki.
Now the whole world had one language and a common speech.

Kar peruple dati evar, em tuti an plon de Shinar e mati da.
As men moved eastward, they found a plain in Shinar and settled there.

Em spoki a sem ader: “Ven, kab ma ten piked e kob em dokiali.”
They said to each other: “Come, let us make bricks and bake them thoroughly.”

This sort of language would receive criticism for being far too much like English. On the other hand, a language with the consonant inventory /b q m m̥ n ŋ ŋʷ ç ɬ ɰ k͡ʘ/ and nothing corresponding to the concept of the verb might be equally criticised as being totally implausible in real life.

Such an approach is reasonable enough in conlanging done for its own sake. It is the approach usually assumed by “how-to” guides such as Rosenfelder’s Language Construction Kit. But I would like to suggest there is at least one area in which “realism” must compete with other concerns. The area I have in mind is the domain of language creation for the purposes of including that language in a novel, or other work of fiction. (This applies especially to the written medium; conlangs made for film or (hypothetically) theatre are perhaps not subject to quite the same restraints.) Conlangs have occurred frequently in fantasy and science fiction novels for some time; the most influential and well-known novelist-conlanger has been J.R.R. Tolkien, but any work that involves made-up words or names from another language can be seen as indulging in conlanging to a degree.

The main competitor with realism as the ultimate aim of the novel-oriented conlang is, in my mind, accessibility to the reader. Whilst other conlangs are made for personal enjoyment, and possibly some degree of dissemination to an audience with some linguistic knowledge of their own, the conlang in the novel (assuming the author intends to be published) must satisfy first and foremost the predominantly linguistically naive audience of that novel’s readership. That this provides conflict with the quest for realism can be illustrated by the following example from the actual human language Abkhaz:

                Sará apħəs asap’ən sxarp alalsərdz:a'yt’.
                I got the woman to wash my shirt with soap.

Abkhaz is an actual language, and furthermore it is non-Indo-European; if someone, therefore, were to present it as a conlang and not get caught out, they would probably be considered to have done extremely well in achieving the goal of “realism”. Yet it would probably not do as a language to be included in a novel. The reason for this is primarily phonological: the reader presented with the sentence above would have very little idea of how to pronounce it properly. Admittedly, one might be able to help matters by altering the orthography, e.g. to get rid of characters with which readers would be unfamiliar, but something like Saraa aphhes asap’en skharp alalserdzzayt’ isn’t likely to result in a terribly accurate pronunciation either, if readers even bothered trying.

A suggested remedy might be able to include an appendix giving pronunciation details, as Tolkien does in The Lord of the Rings. But this may be more difficult than it sounds: describing the pronunciation of ejectives or [ħ] in a readable (I would suggest both brief and understandable) way is not an easy matter. Tolkien’s pronunciation appendix is helped greatly by the fact that the sounds in his languages generally exist either in English or in other languages (German, Spanish) with which his readers might well be familiar; he does not have to give detailed phonetic information. There is also the issue that many readers may not pay much attention to a list of pronunciations, particularly if they only encounter it after reading the rest of the book. In Tolkien’s work this is not a great problem, because his spellings are largely intuitive – only <c> for /k/ is likely to lead the reader astray in pronouncing most of the names approximately correctly, and there are no symbols like <ħ> or <ə> that are likely to flummox the reader entirely.

From this, I conclude that a constructed language in a novel ought to be fairly intuitive orthographically, something which has knock-on effects on the phonology – one must choose only sounds that one can represent in intuitive ways. Note that I do not believe the orthography of a novel-oriented conlang need be totally intuitive: this would essentially restrict the writer to only the phonemes of English or a subset of them, and thus rule out cross-linguistically common sounds like non-diphthongised [o] or [e] or trilled [r]. In all fairness, it does not seem like much of an issue if a reader mangles a word like pékero [pekero] to [peɪkɛɹoʊ]. On the other hand, if that same reader pronounces aphhes [apħəs] as [æfhɛz] – or even worse, struggles to pronounce it at all – one might conclude that the phonology and orthography of the language is too far from their linguistic comfort zone to be of much value.

A related point: the conlang used in a novel may potentially use orthographic conventions that would have little purpose in other contexts. Some examples based on Tolkien’s work may again be instructive here. A language which does not have any diphthongs has little need to show that a sequence <ae> is intended to be pronounced with a syllable break, /a.e/. But the English-speaking reader will probably not read <ae> as /a.e/, and therefore it might be useful to use a diaeresis <aë> to point that reader closer to the correct pronunciation (an alternative would be to avoid words with consecutive /a/ and /e/ in the novel entirely). Similarly, one might use an acute accent or a diaeresis to show or remind the reader that final <e> is intended to be pronounced.

It should be noted that adhering to these constraints does not necessarily restrict one to creating Western European style phonologies. Japanese is perhaps a prime example of a non-Indo-European language that nevertheless can be represented in an orthography that allows English monolinguals to closely approximate its pronunciation. In general, languages with smaller phonemic inventories are more likely to only have sounds that are also found in Western European languages, though this is not to say one cannot attempt a fairly large inventory in a conlang intended for a novel.

One should also consider the effects of accessibility in deciding phonotactics and when creating words for the conlang’s lexicon, particularly if those words are to be used frequently. Just because a word is made up of a sequence of accessible sounds, it may not be easy for a reader to pronounce: a word like trplkm is likely to cause problems. This applies, however, even when words do not violate the phonotactic constraints of the reader’s own language, particularly in the case of lengthy words or those with the same sound used repeatedly: arururududararuru could potentially be an English word, but that does not mean it is easy to say aloud on first glance.

It seems to me that the major constraints of accessibility on realism are, however, restricted to phonology, orthography and the forms of words in the lexicon; in morphology, syntax, semantics etc. the novel-oriented conlanger generally has as much freedom as any other. Extremely detailed morphology or syntax, however, may be redundant for other reasons, however, which I shall discuss in the next section.

3. Extent of inclusion

Personally, I would advocate that a “less is more” approach is the best one when incorporating a conlang into a novel: that is to say, one should aim to include as little of the conlang as possible. There is little point in including lengthy texts in one’s conlang, as most readers will simply not read them. However, some authors have succeeding in doing differently – there are several fairly extensive samples of Tolkien’s languages in The Lord of the Rings – so I will not claim this should be a hard-and-fast rule.

This ties in closely to the accessibility issue: the more of a conlang included in a work, the less accessible that work is going to be. In general, I would suggest that conlangs in novels are restricted to: (1) names of people and places, (2) concepts which cannot be translated easily into the language in which the bulk of the novel is written, and (3) very short utterances in languages which the viewpoint characters do not understand. Longer utterances in languages not spoken by these characters are simply going to come across to them as a babble of indistinguishable sounds, and so there is no point in representing them in detail to the reader; utterances in languages which the viewpoint characters do understand (particularly where they are these characters’ first languages) might as well be provided in English. I say this because I do not believe most readers are interested in extensive language samples; if they are too long, they will probably just skip them, and may even be frustrated by their presence.

Thus we return to the redundancy of detailed morphology or syntax that I mentioned at the end of the previous section. If all samples of the language are restricted to (say) three words at the very maximum, there is probably little point as far as the novel is concerned in stipulating the rules of relative clause formation or going into detail on the intricacies of the ablative case; these things are simply not very likely to turn up in the text. At the most extreme, a novel-oriented conlanger can work out the grammar (and vocabulary) of the language as they go along, only creating that which is actually required for the story. This is not to say, of course, that conlanging for novels ought to avoid detail as a matter of principle. But it should be noted that the vast majority of detail one might want to put into a conlang will probably not surface in the novel itself; if one does want to go into detail, one must be aware that one is doing so purely as a personal project.

4. Aesthetics

 A third short point I would like to make is on the relevance of aesthetics to novel-oriented conlanging. Many conlangers would argue that the look and sound of their languages are important anyway, but I feel that aesthetics are perhaps an especially important consideration when one is targeting a wider audience. Firstly, it is necessary to consider the impressions a language will create on others, not just on its creator. A conlanger might think that shopbop mitzakhkhakhra is the most beautiful sentence ever written, but if they want to include that sentence in their novel they must also be aware that other people might think differently.

Secondly, the aesthetics of the conlang will contribute to the general aesthetic of the whole novel. If the work is supposed to be dark and serious, then setting it in the town of Gombolstiltskin in the Monong-gong-gong province of Ladiladiladila is possibly not the best idea; on the other hand, these names may be more fitting if the work is comic. A specific case of the aesthetics of a conlang interacting with other elements of the work occurs in the naming of places or characters; the sound of names can contribute a great deal to the overall perception of the thing named. If one wants to create a character who is supposed to ooze heroism or villainy, then it generally helps if that character’s name (whether it is in a conlang or not) will sound to the intended readership like a fitting name for a hero or villain, or at the very least is not completely inappropriate for such a function. Similarly, an empire of evil is probably best served by the author if it isn’t given a name that will just make the reader laugh.

5. Conclusion

In summary: when one is creating a conlang for a novel, there are reasons to believe that one ought not simply follow exactly the same processes as one would when creating a conlang for some other purpose, but ought to bear other considerations in mind as well. The actual extent to which the conlang is displayed in the novel may be quite limited.

This has at least one important knock-on effect not discussed previously: it may be dangerous to take a pre-existing conlang, created for some other purpose, and incorporate it unchanged into one’s work of fiction. It would perhaps be wiser to consider the arguments made here and either be prepared to revise a pre-existing conlang as necessary or to start a new language from scratch specifically for the purposes of inclusion in the novel. When a conlanger decides to become a novelist, or vice versa, the two crafts must not be carried out in isolation but rather one should inform the other. 

Copyright © J. Baker 2013