We've had questions on constructed languages (conlangs) before--their inventories, sound changes and phonotactics--but we rarely if ever talk about the grammar change. When designing a naturalistic conlangs grammar, you usually base it off of actual languages and mix/match features to find a balance you like, but what comes next?

How do you make a conlangs grammatical evolution feel realistic, rather than engineered?

  • $\begingroup$ Star trek borrowed from obscure languages for some of theirs, especially when only a few words were spoken by supporting characters. Pashto was used for some of the names ( ferrenghi = foreigner, etc) $\endgroup$ – pojo-guy Feb 3 '18 at 15:13

Don't try to go into details without first acquiring a base in historical linguistics; and it obviously depends on your purpose:

  • If you are designing a constructed language intended to be used as an auxiliary intercommunication language then you generally consider it static.

  • If you are designing a language intended to stand for the naturally evolved language of a fictional linguistic group then you don't "mix [and] match [grammatical] features to find a balance you like", but rather you mimic the features of an attested language or two; and then you follow on with an immitation of attested processes of linguistic change.

For example, in the Indo-European family the general tendencies over the last three or four millennia were

  • Reduction and regularization of inflectional paradigms:

    • Three numbers (singular, dual, plural, e.g. in Sanskrit and Ancient Greek) to only two numbers.
    • Three grammatical genders (masculine, feminine, neuter) to only two grammatical genders, or even no grammatical gender (e.g., English or Persian).
    • Many nominal case forms to few nominal case forms, or to one (as in English or French, where nouns are no longer declined for case).
    • Many mood and tense verbal forms to few mood and tense forms, or, as in the case of English, very few.
    • Elimination of "irregular" paradigms (which themselves were the remnants of old regular patterns) in favor of simpler patterns.
  • In parallel with the simplification of morphology, a shift from free word order (made possible by the rich morphology) to strict word order (to compensate for the morphological simplification). We say that ancient Indo-European language were more synthetic and modern languages more analytic.

This evolution in grammar was obviously superperposed with sound changes, and interconditioned by them; they cannot be separated.

The general trends are just that, trends. Occasionaly, in the history of a specific language, they could be reversed; for example, in French the future tense is a newly evolved synthetic form derived from an analytic form (infinitive + avoir) which had replaced the old Latin synthetic form, which itself was descended from an analytic form (verbal stem + *bhu)...

But all this is useful only if the work is ample enough to follow the evolution of one or more languages over millennia; for example, Tolkien did it with his Middle-Earth elves: Common Eldarin -- Quenya (and Telerin) -- Sindarin. But then Tolkien knew what he was doing: he was a professional historical linguist (he worked on the Oxford English Dictionary) and a professor of Anglo-Saxon literature.


In the Indo-European language family the general trends in grammatical evolution over the last 3 or 4 millenia were:

  • Synthetic structure → analytic structure (with English exhibiting a trend towards a fully isolating language)
  • Complex morphology → simple morphology → almost no morphology (e.g., English)
  • Reduction in morphological categories (for example, English has almost completely lost the entire category of grammatical gender) (complicated by the constant emergence of new categories, which in turn become candidates for elimination; consider for example English verbal aspect)
  • Elimination of irregular paradigms (complicated by the constant emergence of new paradigms, which in turn become irregular due to sound changes)
  • Free word order → limited freedom → fixed word order

Omnis ars naturae imitatio est (all art is an imitation of nature). Seneca, Ad Lucilium Epistulae Morales, 65.3, published around 65 CE.

  • 4
    $\begingroup$ Tolkien knew what he was doing Probably the most important point here (not that the rest isn't good). $\endgroup$ – StephenG Feb 2 '18 at 10:47
  • $\begingroup$ @StephenG Agreed. This used to be a major problem when people started to make Tolkien-style fantasy after the Lord of Rings became a success and did not accept that an ordinary author is NEVER going to match Tolkien in inventing languages for their people and NEVER should even try. Making cultures and languages for your people is fine but you need to accept it will not end up as good as the things Tolkien managed to create and use it accordingly. $\endgroup$ – Ville Niemi Feb 2 '18 at 12:20
  • 1
    $\begingroup$ @VilleNiemi: "Never" is a long time. There are many people who could, if they wanted to, describe the evolution of a fictional language family as well as Tolkien, or better. Historical linguistics is still alive. Linguists with a second inclination for fantasy are not that rare -- in the 19th century they had the Brothers Grimm, in the 20th century there was Tolkien, in the 21st century there may will be TrEs-2b... $\endgroup$ – AlexP Feb 2 '18 at 12:27
  • $\begingroup$ @AlexP Which is why I qualified it with "ordinary author". If you have the background and the inclination, fine. But if you do not, which is ordinary, you won't live long enough to compensate for that lack. ie. "never" actually is capped by your life expectancy. (Science might of course extend life spans to the point where I would be wrong in the future.) $\endgroup$ – Ville Niemi Feb 2 '18 at 13:01

I think irregularity is a key factor when giving a conlang a realistic look. Natural languages have plenty of exceptions, inconsistencies and rules that simply don't make sense.

The general rule is that languages tend to simplify, but also constructions that are frequently used do rarely change. We have three or four words for many verbs such as cook/cooks/cooked (for example), but many more for more common verbs like be, go, come... What is more, former irregular verbs tend to regularity such as dreamt/dreamed or burnt/burned. A good example are some plurals which used to be irregular, like brother/brethren, whose plural nowadays is simply brothers. You can also add some defective verbs, like can (which doesn't have a singular third person mark).

Another thing to take into account are idioms, expressions and proverbs, which also rarely change. These features give realism to a language, a sensation of being alive. For instance, the expression safe and sound is rather usual, but you don't usually say you are sound (at least in an informal context).

My advice is to incorporate a handful of exceptions and incongruencies to give your language that feeling of having evolved rather than having been designed.


Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.