Do they originate in particular dialects or languages that influenced Latin?

Is the question even answerable? With any degree of certainty?

Just curious.

  • 2
    This is an interesting question, but I'm not sure what would constitute an answer. Perhaps a comparison between related languages to see if five declensions was a Proto-Italic thing, comes all the way from PIE, or something else?
    – Joonas Ilmavirta
    Commented Apr 27, 2017 at 22:00
  • 1
    @JoonasIlmavirta, One possible answer that occurs to me is: There is no evidence that allows us to know anything. Another possible answer might be: We find similar endings in other, seemingly related languages. Another possible answer might be: We find exactly those endings in other languages that are not Latin---they were directly imported. Commented Apr 27, 2017 at 22:16
  • @JoonasIlmavirta And I've wondered if the five noun declensions might have resulted from five tribes, all with their own languages, mixing and merging over centuries. I don't think there's any way to know a priori what would constitute an answer. I think the only thing to do is look into it and see what one finds (or what other people have found).
    – Ben Kovitz
    Commented Apr 28, 2017 at 8:19

3 Answers 3


The different declensions started in Proto-Indo-European. Latin regularized and simplified them, giving the five somewhat-regular patterns you're familiar with.

PIE nouns came in a few different types:

Stems ending in *-eh₂

*h₂ was a "laryngeal" sound, so-called because we don't have a better name for it. Scholars can't agree on what its actual sound was; there are arguments to be made for various different sounds, though I'm partial to the voiceless velar fricative as in "Bach" (/x/).

As Proto-Indo-European broke apart, all the laryngeals disappeared. But when they were next to a vowel they left "color" on it before vanishing. *h₂ in particular tended to turn e into a.

So these turned into the Latin "a-stem" nouns: the first declension.

Stems ending in a "thematic vowel"

Some nouns were formed using suffixes ending in o. These were the only true "vowel-stem nouns" in PIE, and the "thematic" o occasionally became e (such as in the vocative singular). If you know Ancient Greek, you'll recognize this intermittent o/e immediately from verb conjugations.

In Old Latin, short o tended to turn into u in many environments. So the original forms -os, -om became -us, -um. This formed the second declension. (Note that this pattern didn't affect long ō, which persisted in the accusative plural.)

Stems ending in *-y

PIE didn't distinguish much between the vowel i and the consonant y (IPA /j/), a trend which persisted into Latin. So when endings were applied to nouns ending in *-y, it tended to become a vowel i.

These became the "i-stem nouns", a subset of the third declension. These are the ones like turris which have -im in the accusative singular.

Stems ending in *-w

Like with y/i, PIE didn't have much of a distinction between u and w. (Same as Latin: compare auspex with a vowel and avis with a consonant, from the same stem au-.)

Nouns whose stems ended in *-w became the Latin "u-stems", the fourth declension. Unlike the second declension, the u that appears here actually comes from a PIE w/u, so you don't see it alternating with o.

Stems ending in laryngeals (?)

I'm not sure of the etymology of this class, actually. And someone with a better knowledge of Proto-Indo-European should feel free to correct me.

But I believe stems ending in "laryngeals" other than *h₂ became the fifth-declension, with the preceding vowel becoming the characteristic e.

Stems ending in anything else

PIE was fine with consonants smashed together, but most of its descendants reversed this tendency. So when the endings involving consonants were put onto the stems involving consonants, extra vowels often got inserted in between.

In Latin, these ended up becoming "consonant-stems", which make up the majority of the third declension. These are the ones like rex which have -em in the accusative singular.

A few exceptions

Nouns which didn't fit cleanly into any of these groups were generally shoved into the nearest one. Diēs, for example, became the only masculine in the fifth-declension, while domus can't decide whether it's second or fourth. But almost no nouns actually remained "irregular"; Latin was very good at forcing them into these five categories.

And thus, the variety of PIE nouns became Latin's five-declension system.

  • 2
    Excellent answer.
    – cmw
    Commented Apr 28, 2017 at 4:24
  • 1
    bit of trivia on the lack of distinction between v and u in Latin: there are quite many Latin professors (at least in Italy) that won't tolerate the pronunciation of a modern v, but will pretend the u sound any time a u/v is encountered in a Latin text
    – Federico
    Commented Apr 28, 2017 at 14:54
  • 2
    @Federico Are you referring to the classical pronunciation of v as /w/? If so, that's actually pretty universal among classics who use restored pronunciation.
    – brianpck
    Commented Apr 28, 2017 at 15:43
  • 1
    Nice answer. On the fifth declension, its origin is a bit unclear, but it doesn't seem to have to do with laryngeals. Rather, it probably has to do with the two words dies and res, which underwent some specific changes from PIE to Latin that made them look different from the other four declensions, and later other nouns were drawn into that new class too. (This is based on the admittedly brief discussion in Weiss.)
    – TKR
    Commented Apr 28, 2017 at 20:41
  • 1
    @Draconis: so Latin has actually simplified something?! There's a first--what was it like prior this miracle?
    – tony
    Commented Feb 20, 2018 at 9:50

The declensions are historical and developed from Proto-Indo-European.

Per Sihler's New Comparative Grammar of Greek and Latin:

enter image description here

See also Quiles/Lopez-Menchero's Grammar of Modern Indo-European:

enter image description here

  • 1
    I love these funky PIE symbols. Perhaps because I don't understand them. Do they bear any resemblance to noun the declensions of Latin? Or does the resemblance lie in the fact that PIE and Latin have the same number of noun cases? Commented Apr 27, 2017 at 23:09
  • 1
    @DavidCharles There is indeed a connection. I'll try to tidy up this post and add additional information later tonight, and hopefully someone else can chime, too.
    – cmw
    Commented Apr 27, 2017 at 23:18
  • 2
    @DavidCharles There's also not universal agreement on the number of noun-classes in PIE: much like how -um neuters and -us masculines could be either separated or combined in Latin.
    – Draconis
    Commented Apr 28, 2017 at 1:20

For what it is worth, declensions have arisen independently in different language families. As mentioned, the Proto Indo-European language had 8 cases, which reduced to 5 in Ancient Greek, 6 in Latin, 6 in Russian, 8 in Sanskrit (but only 3 in Hindi), and 4 in German (and 0 in English, French, etc.).

Regarding other language families ... The classical Semitic languages (Arabic, etc.) have 3. Quechua, the language of the Incas, has 7 - 19 (depending on how exactly you define a case). Japanese has 7 (although they are marked with separate particle words). Hungarian has 7. Somali has 4.

  • 2
    The question was not about the five (or so) cases in Latin, but the five different declensions (classes of nouns or other words).You make a valid point, but I'm not sure how it's related to the question. Can you elaborate?
    – Joonas Ilmavirta
    Commented Feb 19, 2018 at 19:46

Your Answer

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

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