How dissimilar must case endings be to each other?

Question

I'm asking this question for a conlang. I know that's normally too subjective for this forum, but honestly, I think this may be an exception.

I tend to try to keep things as terse as possible. Often times my case endings are just the 5 or so vowels of the language. Nothing more. But I was wondering, would this be too minimalist for common use? Not that I intend my conlang to actually be used, but it would just seem weird to me for the fictional people who speak my language to use something that would be unusable in the real world.

You can't have things be too minimalistic, you need redundancy. Humans don't catch every phoneme pronounced, and this problem is even worse over a phone. So in general its a bad idea to have distinctions that differ only in one phoneme. Not that this doesn't happen (English's plural and possessive endings are an example, also there's only one phoneme of difference between 'can' and 'can't'). But looking at languages that do have case systems, they don't just change the final vowel and leave it at that. They often add consonants, sometimes entire syllables, when declining their nouns. Also, I've noticed that languages seem to dislike having endings that are just a single vowel be anywhere close to eachother. For example, in Latin, there is no declination where two of the cases are marked with -i and -e. And in Finnish the only cases that end in a vowel are -a and -i, all the others are differentiated by consonants too. Russian does allow nouns to take an -i or -e ending for different cases, even in the same declination, though in Russian an -e always palatalizes the previous consonant, creating a redundancy.

But looking at Latin and Finnish, I have noticed something. They do allow cases to be differentiated by just a consonant. For example, the Latin first declination singular ending for the nominative is -a, and -am for the accusative. And looking at the second declination again, I see that there is a long -i ending that contrasts with a short -e ending, in the second declination. Still, the different lengths is a redundancy.

I find this interesting that I can't seem to find a language with cases differentiated with a single vowel (or close ones at least), but languages seem to have nothing against cases being differentiated by a single consonant.

Is this universal? If there was a language where the case endings were just -a, -e, -i, -o, and -u, would speakers find these too similar to each other? Would they require some extra redundancy, like a consonant being added to one of the front vowel cases to differentiate it from the other, lengthening one, or just letting the two cases merge into one?

Essentially, this is a question as to how much redundancy is necessary. Obviously, redundancy is necessary, but how little can you get away? What is the minimum redundancy that is necessary? Does anyone even know the answer to this?

Answer 1

If there was a language where the case endings were just -a, -e, -i, -o, and -u, would speakers find these too similar to each other?

Consider modern Russian, which has six cases: nominative, genitive, dative, accusative, instrumental, and prepositional. In the first declension these are indicated for the most part by single vowels.

kniga = a book
knigi = of a book
knige = to a book
knigu = a book (object of a verb)
knigoj = using a book
knige = a book (object of a preposition)

Answer 2

You don't even have to have case endings at all. The overwhelming majority of Niger-Congo languages don't, neither do Salishan, Chinese or most Romance languages. Classical Arabic did just fine with the case ending -i, -u, -a (modern dialects dispensed with that, not because it was uninformative but because final vowels got deleted). Most Russian case endings are just a vowel; all of the case markers in Hawrami are just a vowel.

Answer 3

In my understanding, you are trying to set the limits for redundancy, regarding the case endings.

I think, the direct answer to this question depends on your conlang's articulatory stress patterns:

• if your conlang allows the stress to lean towards the last syllable (or even better, be able to shift towards the end in noun cases other than Nominative), the single-proneme case endings are just fine — simply because even a "lazy" speaker will be forced to produce it clearly (that's what the articulatory stress is!);
• however, if your words (in Nominative) can be long enough, and the stress patterns lean towards the 1st syllable, you probably need more redundant case endings.

This question is largely relevant:

Why did English lose declensions while German retained them?

"The loss and weakening of unstressed syllables at the ends of words..... had disastrous effects on the inflectional system, since many endings now became identical." — (Barber, 1993: 157)

Answer 4

As mentioned by user6726, Classical Arabic (and also Akkadian) distinguish nominative accusative and genitive just by final vowel (-u, -a, -i respectively) so distinguishing case solely by vowel is definitely possible

For an even more minimally marked case system, something like Irish is a good example where lots of cases are distinguished only by whether the final consonant is broad or slender, or whether the word triggers mutation of a following (syntactically close) word and if so, what kind of mutation