This ending is common among PG words, but is not present in any descendent or ancestor. Take, for example, this:


There is no *-az or anything like it in any daughter language so how was it reconstructed?

  • It's not true it was never preserved. Finnish has loanwords such as 'kuningas' and 'ruhtinas' that preserved the ending.
    – Someone211
    Commented May 17 at 15:40
  • 4
    Two things to consider, apart from the point made by @Someone211 above: (1) not all Germanic nouns ended in -az, only nouns in specific declensions; (2) nouns in similar declension types in other Indo-European languages often have comparable endings, such as Latin -us, Greek -os, Lithuanian -as, Sanskrit -aḥ (< -as), etc. Commented May 17 at 15:43

1 Answer 1


We can tell that the ending *-az existed in Proto-Germanic on the basis of direct attestation, internal comparative, external comparative, and loanword evidence.

This ending is directly attested in our earliest records of the Germanic languages, the Elder Futhark inscriptions. The language of these is conventionally referred to as Proto-Norse where "proto" is understood in the archaeological sense, and not a linguistic one, as this is not a reconstructed language.

In Proto-Norse we see the strong masculine nominative singular ending -ᚨᛉ -az. This is also often transcribed -aR as the sound represented as z/R merges with *r in the North Germanic languages, but is clearly kept distinct from r by the people making these inscriptions. It is not clear whether the sound had already transformed from something more like [z] to something more like [r] at this stage, so it is unclear which transcription is a more accurate representation phonetically.

Even without the runic evidence however we would need to reconstruct this ending on internal comparative evidence.

First, we have two well-attested old Germanic languages that show a reflex of a consonant from this ending, although both have lost a vowel. These are Old Norse (where it is reflected as -r) and Gothic (where it is reflected as -s).

Note that in Continental North Germanic this -r has since been lost, and in Insular North Germanic (i.e. Icelandic and Faroese) an epenthetic vowel has been inserted before it.

Regardless, we must have some ending here featuring a consonant that is reflected in Old Norse as r and Gothic as s.

We already know of just such a consonant from various other positions in the word: Proto-Germanic *z (which developed from the application of Verner's Law to Proto-Indo-European *s).

From this we would require an ending with the shape *-(V)z(V).

We also see however that this ending frequently triggers a-mutation (a vowel change where *u, and possibly *i, are lowered to *o, and *e, before a following *a, with the extent of this change and frequency of apparent exceptions varying between different Germanic varieties), so the first vowel in the ending must be an *a.

That leaves us with the possible shapes *-az(V) and *-za.

We can then note that *z only occurs in quite a restricted number of positions (generally after a vowel) which rules out the second shape.

At this stage we can't rule out the possibility of an additional second vowel, or even further consonantal material, but they would be unnecessary to explain the evidence we have, and so we would reconstruct *-az.

Approaching from the other direction, from external comparative evidence, we know that Proto-Indo-European had a thematic animate (masculine in Late Proto-Indo-European) nominative singular ending *-os (also occurring stressed as *-ós, especially in thematic adjectives). Noting that Proto-Indo-European *o regularly develops into Proto-Germanic *a, and that following Verner's Law an *s following an unstressed vowel evolves into *z, we would expect this ending to be reflected in Germanic as *-az (possibly alongside a Verner alternant *-as, which would be more common in adjectives).

We also see early loans into the Finnic languages that preserve it explicitly. Most clearly, the Finnish word kuningas "king" almost exactly reflects the reconstructed Proto-Germanic *kuningaz (noting that Proto-Finnic voicing was largely allophonic and later only morphophonological, to the extent that modern Finns still often struggle to produce voiced obstruents in foreign languages, and so we would expect a foreign z to be borrowed as *s).

As such, the external comparative and loanword evidence vindicates our reconstruction on the basis of the internal comparative evidence, and align exactly with the direct attestation.

Either direct attestation or the internal comparative evidence would be sufficient to posit the ending at the Proto-Germanic stage.

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.