So, I'm trying to make something with a naturalistic feel, even though the current phonetic system distinguishes about 38 phonemes... Let's say I'll tackle this later. I have issues with grammar... I want to use Proto-Kartvelian as a root source but make a language that has Indo-European grammatical structure. So... When I tried making noun declensions I generally end up making something that turns into agglutinitive suffixes... E.g. if we have a masculine noun ending in -o, I simply use consonants after the -o to express cases, then change the -o to -a for plural and repeat. E.g. -o/ -og/ -ot/ -om/ -ol vs -a/ -ag/ -at/ -am/ -al... It is too regular. How do I change certain things so that it looks less artificial, but still maintain some logic... I need advice.

  • 1
    That can't fit in a simple answer. To add natural-looking irregularities, read textbooks about historical linguistics and sound changes; get used to see languages as historical processes. Lyle Campbell's manual is a good start. If you're a fantasy fan, try to check out what Tolkien did about the evolution of Quenya and Sindarin and the irregularities thus generated.
    – melissa_boiko
    Mar 17, 2020 at 15:09
  • 1
    If you want it to be I-E and not agglutinate, you need declensions and conjugations full of paradigms, which should partially but not sufficiently overlap, so that a given root can represent a totally different noun or verb in several inflection classes. Like the perfect tense in Latin.
    – jlawler
    Mar 17, 2020 at 23:17
  • You might find this question and its answers and this one, too to be of relevance. Mar 18, 2020 at 11:50

3 Answers 3


One possibility: perhaps your "-a- for plural" derives from an ancient word for "many" that was of a different declension, having different endings, so you could posit:

S -o / -og / -ot / -om / -ol
P -a / -az / -asis / -amis / -ay

Another possibility: the historical importation of a foreign declension pattern into words of certain classes, which practice later infects native words of other classes either in whole or in part.

Consider the Latin words that have Greek declension and extrapolate from there.

A third possibility: the later "regularisation" of ancient defective words actually introduces an irregularity. For example, your modern word for "mob" might be an ancient plural form of "person", so you could posit:

S -a / -ag / -at / -am / -al
P -a / -az / -asis / -amis / -ay


Introduce sound changes.

For a simple example, say that the sequence -om- becomes -um- in all situations at one point in the language's history, and -at- becomes -it-.

This would (using your examples) mean that instead of

  • -o/ -og/ -ot/ -om/ -ol
  • -a/ -ag/ -at/ -am/ -al

you'd have:

  • -o/ -og/ -ot/ -um/ -ol
  • -a/ -ag/ -it/ -am/ -al

Now introduce another one: in the case of back vowels (o and u), if they're followed by a voiced stop, the stop becomes devoiced.

  • -o/ -ok/ -ot/ -um/ -ol
  • -a/ -ag/ -it/ -am/ -al

So, irregularity but obviously still based on what was originally a completely regular system.


One thing I should have noted is that the two sound changes only effect the affixes, but that's not how sound changes work. Once you take into account what the affix is attached to, you can introduce even more irregularity.

Now let's introduce two masculine nouns, say minano and nukugo. Under your original system they'd be:

  • minano / minanog / minanot / minanom / minanol
  • minano / minanag / minanat / minanam / minanal
  • nukugo / nukugog / nukugot / nukugom / nukugol
  • nukuga / nukugag / nukugat / nukugam / nukugal

After the first sound change:

  • minano / minanog / minanot / minanum / minanol
  • minana / minanag / minanit / minanam / minanal
  • nukugo / nukugog / nukugot / nukugum / nukugol
  • nukuga / nukugag / nukugit / nukugum / nukugal

Okay, no problem, still sort of regular, no difference between the two nouns, the only change has been to the affix.

And after the second:

  • minano / minanok / minanot / minanum / minanol
  • minana / minanag / minanit / minanam / minanal
  • nukuko / nukukok / nukukot / nukukum / nukukol
  • nukuka / nukukag / nukugit / nukukum / nukukal

Now add a third sound change: If you an unvoiced consonant between two identical vowels, that consonant is dropped and the vowels become one lengthened one

  • minano / minanok / minanot / minanum / minanol
  • minana / minanag / minanit / minanam / minanal
  • nuuko / nuukok / nuukot / nuuum / nuukol
  • nuuka / nuukag / nuugit / nuuam / nuukal

Okay, now we have a situation where you've got nuuum. Let's suppose the phonotactics of your language don't allow vowels that long or the same vowel repeated that often, and if that situation ever arises, that third vowel is replaced by a glottal stop followed by a schwa (that I'll write as 'e).

  • minano / minanok / minanot / minanum / minanol
  • minana / minanag / minanit / minanam / minanal
  • nuuko / nuukok / nuukot / nuu’em / nuukol
  • nuuka / nuukag / nuugit / nuuam / nuukal

Final sound change: if a word final -t or -l is preceded by -ku- or -ko-, the vowel is dropped.

So here's the final outcome of the two words:

  • minano / minanok / minanot / minanum / minanol
  • minana / minanag / minanit / minanam / minanal
  • nuuko / nuukok / nuukt / nuu’em / nuukl
  • nuuka / nuukag / nuugit / nuuam / nuukal

The first has remained pretty regular, the second not so much.


"Indo-European grammatical structure"—I presume you are referring to the highly inflecting state of most ancient languages. This will be quite an undertaking, given that Sanskrit nouns have eight cases and three numbers and several declension classes, given the complexity of the Ancient Greek verbal system. Study those systems to some detail and than decide what you want to take from them.

Also look at derivative morphology: There are common verbal prefixes among Sanskrit, Latin, Greek, and Germanic (at least), and there are common methods to form compound nouns. Also several means to create nouns from verbs or verbs from nouns go far back in time and can be reconstructed for the protolanguage.

I wish you a lot of fun and good look with you conlang project!

  • I once heard it said that Greek had no regular verbs. Mar 19, 2020 at 3:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.