Wiktionary talk:About Proto-Germanic

Definition from Wiktionary, the free dictionary
Jump to navigation Jump to search

Indicating nasalisation in entry names[edit]

Our current practice is to not indicate nasal vowels in entry names, but to write a following -n instead. The original reasoning for this was that most sources use this convention, and so it's what most people will end up adding to etymologies. But it has the downside that it doesn't distinguish final nasal vowels from final -n, even though this distinction was phonemic in Germanic. Compare *tehun with *nahtų, the accusative of *nahts. I'm wondering now whether we should use the ogoneks in the entry titles as well. A related question concerns the combination -nh- for the same reason. -anh- was actually pronounced -ą̄h-, and similarly for all other vowels. We could still redirect the following-n names to the new names. —CodeCat 20:42, 29 January 2012 (UTC)

I think that showing the distinction between the sounds is important in the headword form, even if the pronunciation key makes this clear. I would have no problem using ogoneks in entry titles. The only issue for me would be other materials which forgo this use and use -n instead, how to distinguish it myself here, but I presume that there are clear rules determining which is which? Leasnam 15:11, 30 January 2012 (UTC)
Final -n is fairly rare in Germanic, because it normally disappeared after unstressed vowels. One case where it didn't disappear is when another sound followed it at the time it was lost. In most cases this sound was -t, which was lost after -n in time. So, usually final -n goes back to an earlier -nt. This is the case in some numbers like *tehun but also in the 3rd person plural forms of verbs. The sound was also retained in single-syllable words such as *þan and *in. But as a rule, the nominative singular forms of neuter a-stem nouns and feminine n-stem nouns will have a nasal vowel, as will the verb infinitive form. —CodeCat 15:21, 30 January 2012 (UTC)
Because the nasal vowel is common in PGmc, I think they should be included in the entry titles. They would certainly help to distinguish words with nasal vowels from words where -n was not lost, too --Proto-Germanic Fan 18:16, 30 January 2012 (UTC)
I agree with the former posts that the nasal vowels should be included into the entry names, because firstly we already include it into the content page, and secondly the nasalisation of vowels is phonemic in Proto-Germanic. I am in favour of using the ogoneks in the entry names, but I have another proposal too, because when I am in the library the ogoneks do not display on the computer, which is a huge disadvantage for me, because I am there often for study. I would thus propose to use the sign ɴ for nasalisation, e.g. *nahtuɴ (which looks like *nahtuN) for *nahtų. The second advantage of this is that one does notice the nasalisation surely, and thirdly it looks a little more convenient than all those signs under and above some vowels, e.g. *hūsôɴ for *hūsǫ̂. I write the nasalisation in this way for Old Norse, and it really works well. --Dyami Millarson 13:21, 4 February 2012 (UTC)
Another convention would be to use superscript n, as in nahtuⁿ. That letter is commonly used to indicate nasalization in a variety of languages, very widely supported and shouldn't have any interoperatibility issues. -- Liliana 14:01, 10 February 2012 (UTC)
That's also a good idea yes... but it would be a bit more work. —CodeCat 17:51, 16 February 2012 (UTC)

Proto-Germanic given names[edit]

From the few attested PGmc given names, it seems the *-jaz in the second element becomes *-ijaz, ex. *Bidawarijaz from Nøvling, Denmark. That would render *Raginaharjaz as *Raginaharijaz. --Victar (talk) 03:49, 1 May 2013 (UTC)

No, it's because of Sievers' law. —CodeCat 10:23, 1 May 2013 (UTC)
Could you elaborate? --Victar (talk) 15:55, 1 May 2013 (UTC)
w:Sievers' law is a Germanic sound law that operated before, during, and sometime after the Proto-Germanic period. It says that -j- and -ij- are in an allophonic (or allomorphic) distribution, and which of the two appears depends on the length of the preceding morpheme. A syllable is short (light) when it contains a short vowel followed by a single consonant, and long (heavy) otherwise - that is, if it contains a long vowel, a diphthong, or ends in more than one consonant. -j- appears after a short syllable, -ij- after a long syllable. In the case of *harjaz, the preceding syllable *har- is short, so the alternant -j- appears. The distinction between morphemes and words is also important, because this rule continues to apply even in compounds like the ones you named. This implies, then, that *Bida-warijaz must have a long ā: -wārijaz, because that is the only way to account for the -ij-. This ā must have developed from an earlier ē. *harjaz, on the other hand, is known to have had a short -j- because the word *harjaz is attested separately. —CodeCat 16:05, 1 May 2013 (UTC)
Right, but *warjaz is attested as short, so this must have been some exception to the rule. --Victar (talk) 16:25, 1 May 2013 (UTC)
On the *warjaz entry page, you wrote, "The compound form of this noun was conflated with *-ārijaz in many languages, eventually causing both to be treated as one." Couldn't that be applied to *harjaz as well? --Victar (talk) 16:44, 1 May 2013 (UTC)
How do you mean? Are you suggesting that *harjaz in compounds was confused with *-ārijaz? —CodeCat 16:45, 1 May 2013 (UTC)
I'm suggesting all PGmc personal names that ended in *-arjaz became *-ārijaz due to this confusion. Or perhaps it wasn't a confusion at all and just an effect of *-ar-. --Victar (talk) 16:51, 1 May 2013 (UTC)
What evidence and motivation is there for such a change? —CodeCat 17:00, 1 May 2013 (UTC)
*Bidawarijaz is evidence and confusion is a possible motivation. --Victar (talk) 17:11, 1 May 2013 (UTC)
OK, found some more evidence. The w:Skåäng Runestone reads harija ÷ leugaz, so it could be argued that *harjaz was rendered as *harijaz in the second element of PGmc names as well. --Victar (talk) 18:10, 1 May 2013 (UTC)

Proto-Germanic verbs in -ēną[edit]

Hi. I've touched on this a few times in the past, but I am still uncertain how we are dealing with these verbs here: using Gerhard Koebler's Germanisches Wörterbuch et al. (Bosworth & Toller, etc.), there is a class of verbs (stative) in which the ending is -ēną or -Vną. In OHG and ODT, these appear as -en/-ēn (a good example is *swibēną (to move to and fro, float). Are these what we are currently showing as -āną verbs? I doubt it, but that is the closest verb class I can find to this. I have been avoiding creating entries for these verbs solely due to the fact that I am unsure. Leasnam (talk) 15:38, 13 May 2013 (UTC)

We have mostly followed Ringe's reconstruction of class 3, but it is still rather controversial and unclear. The main issue with -ēną is that -ē- normally develops into -ā- in Old High German, so it can't be the actual origin of the -ē- found in those verbs. Rather, the most accepted explanation is that the -ē- of OHG is from earlier unstressed -ai-, which is present in some of the endings but was later generalised to all the forms on the model of class 2's uniform -ō-. However, Ringe reconstructs two subclasses: a larger -ja/ai- class consisting mostly of stative verbs, and smaller -ā/ai- class of denominatives (which only survives in Gothic I believe). I think that model is gaining acceptance but there are still many disputed details. The main points of controversy are:
  • Why the ja-class has few reflexes in the Germanic languages, despite being much larger than the ā-class.
  • How the ja-class arose within late PIE, in particular given the evidence of cognates in other PIE languages.
  • How to account for certain other irregularities within Germanic that can't be explained easily with a simple two-way split.
There is a good discussion about the issue here. —CodeCat 16:42, 13 May 2013 (UTC)

Ogonek RFM[edit]


The following discussion has been moved from Wiktionary:Requests for moves, mergers and splits.

This discussion is no longer live and is left here as an archive. Please do not modify this conversation, but feel free to discuss its conclusions.

For some time we've had the situation where Proto-Germanic words that have a final nasal vowel were spelled with a following -n, but in links and headwords the display was overridden to show an ogonek instead. The rationale given for it (by me) at WT:AGEM was that many sources write the final -n, and that it would therefore be more convenient. However, now that we have lots and lots of Proto-Germanic words, it doesn't really seem so convenient anymore, but instead an unnecessary nuisance. We can't just drop the ogoneks because the distinction between regular vowel, nasal vowel and vowel + n was phonemic in Proto-Germanic and the three have distinct outcomes in the descendants. In retrospect, sources may use -n but they may also use some other representation, so we can't follow all of them. However, probably the most prominent Proto-Germanic source is currently Ringe 2006, and that uses ogoneks. So if we are to follow any source at all, that would be a very strong candidate. Therefore I propose to move all of the Proto-Germanic pages to the ogonek form when applicable, and to update all the links pointing to those entries. (One note: while you may argue that ogoneks make the words harder to type, consider that Proto-Germanic also uses macrons, circumflexes and the letter þ, so adding ogoneks would barely add any further difficulty) —CodeCat 02:20, 21 March 2013 (UTC)

Oppose purely out of selfish motivation. Macra and circumflexes are easy for me to type; thorns much harder but much rarer. Ogoneks are worse than any of those for me. In general, circumflexes are a widely available character (for example, they are common enough to be in Mac's U.S. character set) and macra and thorns are used in representing several older Germanic languages and will likely be easier for Germanicists to input. Ogoneks do have some Germanic use, but it is sparing to say the least, and they are an uncommon character group in general. —Μετάknowledgediscuss/deeds 03:39, 21 March 2013 (UTC)
That may be true, but since when have we shied away from hard-to-input characters when we felt them to be more appropriate (putting the smart quote debate aside)? I mean, we don't have any problems inputting IPA characters, or any of the many awkward symbols in PIE, do we? —CodeCat 03:44, 21 March 2013 (UTC)
Yeah we do. I never argued against PIE characters because I don't know much about it, but I find it very frustrating sometimes. As for IPA, the day should come very soon when people can input X-SAMPA in the {{IPA}} template and depend on a backwards-version of Michael's Lua converter (easy to write, a little harder to implement). —Μετάknowledgediscuss/deeds 04:05, 21 March 2013 (UTC)
Unrelated to the main topic but I use "Character Palette" on Mozilla Firefox, which let me type any character I want without opening a second tab (except for certain diacritic symbols, esp. Arabic but I use Windows On-Screen keyboard for that). I don't need to have multiple tabs and copy-paste and I can have a few palettes if I run out of room on my toolbar. E.g. ɛŽžČ芚ʹʺÁáÉéÍíÓóÚúÝýɛ́āēīōūǎṯḥḏṣḍṭẓʿġʾâğäöüçşïıñôţḑŗŕǎ̀ʻøј fits on one panel - they have spaces on the panel and clicking on one symbol (they act like buttons) adds it to a clipboard. Other browsers should have something similar and there could be better tools. --Anatoli (обсудить/вклад) 04:14, 21 March 2013 (UTC)
My inclination is to support a switch for the reasons CodeCat outlines. - -sche (discuss) 04:21, 21 March 2013 (UTC)
  • @Μετάknowledge, there's also Edittools (when it loads). :) -- Eiríkr Útlendi │ Tala við mig 05:35, 21 March 2013 (UTC)
    • I use the US international layout, which has support for a lot of diacritics on Linux, including macrons, ogoneks, breves, hačeks and so on. On Windows it's a bit more meager, but there is a keyboard layout editor for Windows that you could use to add the extra characters to your layout. For characters that are not in my keyboard layout I either use the edittools, or I look up the unicode codepoint for the combining diacritics (when I want to type ā́ for PIE, I type shift+altgr+3, a, ctrl+shift+u, 0301, space). —CodeCat 13:34, 21 March 2013 (UTC)
      • On Win XP and Win 7, I've had good luck using the AllChars app, a little utility that lets you define behavior closer to how the Compose key works in Linux, even using text-based config files for very easy configuration. I'm not affiliated with the project at all, just a happy user. [1] :) -- Eiríkr Útlendi │ Tala við mig 16:19, 21 March 2013 (UTC)
Support: having to care for a cosmetic display-form is one less thing to remember to do Leasnam (talk) 16:25, 21 March 2013 (UTC)
  • Support. Ogoneks and ⁿ are the two ways I've most often seen PGmc nasalized vowels represented. —Angr 21:46, 22 March 2013 (UTC)

It looks like the consensus is leaning towards supporting, so I will start the move. —CodeCat 13:14, 9 April 2013 (UTC)

  • The only question I have is how easily discoverable the ogonek spelling will be. If users enter the spelling with the final "n", will they be redirected to the ogonek spelling? If so, then great, I'm all for this move. If no, then how do we tweak things to make sure that users can find the ogonek spelling? -- Eiríkr Útlendi │ Tala við mig 01:56, 10 April 2013 (UTC)
    • The nasal vowels always occur at the end of the word, so by the time someone has typed in all the other letters of the name, there will only be one or two search suggestions left, including the final nasal if it exists. I don't think that is even a huge problem though, because I think that most users will end up on those pages by clicking on a link in an etymology. And because we hide these pages away in appendixes, they aren't actually meant to be very discoverable, I imagine. —CodeCat 02:13, 10 April 2013 (UTC)
      • <head scratch> Don't we usually include multiple spelling variants, with (in most cases) one selected for maintenance and the others as alternative spelling? - Amgine/ t·e 01:04, 13 April 2013 (UTC)
        • That works when the spelling is actually used, but these are entirely unattested forms. The spelling of attested words is therefore subject to CFI while Proto-Germanic words are of a very different nature. So what exactly does "alternative" mean when there are no attestations? Reconstructions are always made based on some kind of representation, and we've just decided on one particular representation. That doesn't mean that other representations are different terms, they're just different ways of representing the same reconstructed phonemes. It's a bit like deciding that we use straight quotes instead of angle quotes, or no macrons for Latin... the angle quotes aren't in any way an "alternative spelling", they're just a different representation of the same thing. —CodeCat 01:15, 13 April 2013 (UTC)
    All pages have been moved, AFAIK. —Μετάknowledgediscuss/deeds 22:21, 14 August 2013 (UTC)

Kluge's Law in Etymologies[edit]

Guus Kroonen cites this law many times in his 2013 dictionary. Perhaps it should be discussed further before citing any more etymologies requiring it. Already, some class-2 verbs cite this law in their etymologies explaining the gemination of their consonants; however, The entries for *arô "eagle", *sternǭ "star", and many class-4 strong verbs may need some explanation on the presence of their unassimilated consonants.Nayrb Rellimer (talk) 04:23, 29 March 2015 (UTC)


Presumably, and this is an assumption, people didn't call each other *Harjagastiz and *Þeudōberhtaz every day. Do we have any idea of what nicknames they might have had? Thinks like Harja or Hari, Þeō or Berti. UtherPendrogn (talk) 22:28, 2 November 2016 (UTC) UtherPendrogn (talk) 22:28, 2 November 2016 (UTC)


CC:ing some thoughts about hiatus transcription from Reconstruction talk:Proto-Germanic/fōr:

Is *fuïniz etc. really a good notation to use here? Germanic languages usually use ¨ as umlaut rather than diaresis, so this might seem to suggest something like [ɸuɨniz] to readers, rather than the intended [ɸu.iniz]. (I certainly did a double take upon seeing this form mentioned in an entry.)
I would suggest using IPA-compatible notation: *fu.iniz etc.

--Tropylium (talk) 09:35, 3 November 2016 (UTC)

Since PGmc doesn't have a diphthong ui, why not just write *fuiniz? It can only be three syllables. —Aɴɢʀ (talk) 10:32, 3 November 2016 (UTC)

Germanic descendant order[edit]

@CodeCat, Leasnam, Angr, Metaknowledge, -sche, after some edits I made, @Theudariks pointed out to me the order of Germanic descendants mentioned under Entries, categories and templates. Our general principle in the rest of Indo-European has been a slightly modified alphabetization (We normally count Old English as English for sorting purposes). It seems to me, however that the ordering of Germanic entries has little to do with the alphabet or even the importance of the language (why is Gothic at the bottom?). May we standardize the ordering to something more sensible? —JohnC5 01:13, 14 November 2016 (UTC)

The ordering that was adopted at the time was meant to put similar languages closer together, as well as to have English first. So the natural progression from English, the northernmost West Germanic language, is north-south towards Upper German. The same was done for North Germanic, but going from west to east. Gothic is an outlier among the Germanic languages, so it came last. —CodeCat 01:44, 14 November 2016 (UTC)
That makes some sense. It still seems fairly arbitrary, and I would prefer alphabetization for consistency. —JohnC5 01:53, 14 November 2016 (UTC)
Alphabetic order (with special preference given to English, since this is en.wikt) does seem a lot more intuitive, I for one would support that. — Kleio (t · c) 03:56, 14 November 2016 (UTC)
Hmm, not to be set in my ways, but I kind of like it the way it is for exactly the reasons given by CodeCat: grouping should show progression away from English Leasnam (talk) 04:59, 14 November 2016 (UTC)
I would also suggest that we could start breaking it out by Eastern, Western, and Northern Germanic as we to with Slavic. That would allow us to alphabetize and keep similar forms together. —JohnC5 05:24, 14 November 2016 (UTC)
I don't have terribly strong feelings about it, but I don't think we need to put English at the top of the list just because this is en-wikt. I'd separate East Germanic, North Germanic, and West Germanic (in alphabetical order), and then alphabetize by main language name (Old Dutch, Old English, Old Frisian, Old High German, Old Saxon and so on) within each group. —Aɴɢʀ (talk) 08:46, 14 November 2016 (UTC)
Angr's proposal represents my preferred layout. —JohnC5 15:29, 14 November 2016 (UTC)
I also like that idea, especially adopting the Slavic sorting system. Does leave me wondering what would be done with Frankish and Proto-Norse in terms of alphabetic position in those rare entries that do feature those languages. — Kleio (t · c) 15:50, 14 November 2016 (UTC)
Frankish and Proto-Norse (to keep it with Old Norse). —Aɴɢʀ (talk) 16:01, 14 November 2016 (UTC)
:::::::::Well, I'd still prefer to see English and Frisian grouped together, so maybe 1). An East, North, West Division, then 2). A subdivision in North (East & West) and in West (Anglo-Frisian-Saxon, Dutch, German-Yiddish) then alphabetised ? Like so:
  • East:
    • Gothic
  • North:
    • East:
      • Danish
      • Gutnish
      • Norwegian Bokmål
      • Swedish
      • Westrobothnian
    • West:
      • Faroese
      • Icelandic
      • Norn
      • Norwegian NyNorsk
  • West
    • Ingvaeonic
      • Anglo-Frisian
        • Anglic
          • Old English
            • Middle English
              • English
              • Scots
              • Yola
        • Frisian
          • Old Frisian
            • North Frisian
            • Saterland Frisian
            • West Frisian
      • Old Saxon
        • Middle Low German
          • Low German
    • Irminonic
      • Old High German
        • Middle High German
          • German
    • Istvaeonic
      • Frankish
        • Old Dutch

Is this attainable ? Am I asking too much ? ...on second thought, it looks more complicated than the current flow...Just tell me how to do it :) Leasnam (talk) 16:19, 14 November 2016 (UTC)

I would leave out as many headers that aren't languages as possible, besides the East/West/North Germanic division mentioned above. The Northwest/Northeast gmc., Anglo-Frisian, Anglic, Frisian, Ingvaeonic etc. headers seem to merely turn the list into an unnecessarily complicated labyrinth of nested languages. Just use the basic East/West/North scheme and nest languages alphabetically within those. — Kleio (t · c) 16:36, 14 November 2016 (UTC)
Yeah. I agree. Leasnam (talk) 16:41, 14 November 2016 (UTC)
How about grouping it by the sound changes of the particular words instead of the language tree, like this:


  • /u/
    • Dutch: put
    • ...
    • -> /i/
      • Old English:
        • English: pit
      • Old Dutch:
        • West Flemish: pit
      • ...
This would be more true to how words evolved, like a Wave model, although without a geographical location. e.g some Dutch words are Ingvaeonic and the current model listing by relation doesn't always work for this. However this change cannot be automated by bots :( --Eliot (talk) 18:24, 14 November 2016 (UTC)


I have some Proto-Germanic and Old English words I'd like to add. How should I go about it? ÞunoresWrǣþþe (talk) 15:55, 9 December 2016 (UTC)

Are they real Old English words, that the Anglo-Saxons used? I'm asking because we sometimes get people who like to invent words. —CodeCat 16:31, 9 December 2016 (UTC)
Yes, they're real. They're red on a lot of modern-day words' etymologies. Some of them are even absent from this dictionary, being quite rare (moraþ, mulled wine). ÞunoresWrǣþþe (talk) 16:35, 9 December 2016 (UTC)
Is this edit correct? https://en.wiktionary.org/wiki/mora%C3%BE ÞunoresWrǣþþe (talk) 16:57, 9 December 2016 (UTC)
It looks ok. Judging from the pronunciation though, the a is long, so there should be a macron on it in the written forms. Also, þ is not an IPA character. —CodeCat 17:02, 9 December 2016 (UTC)
Bosworth-Toller lists it as "móraþ, mórod", suggesting that the o rather than the a is long. —Aɴɢʀ (talk) 18:46, 9 December 2016 (UTC)
Koebler also has mōraþ. oh nvm, you already edited the entry — Kleio (t · c) 18:53, 9 December 2016 (UTC)

*e > *i cases[edit]

I noticed that Kroonen does not reconstruct the *i-umlaut of *e to *i for PGmc yet (see Reconstruction talk:Proto-Germanic/heminaz). Should we follow this, or use the same logic as with *ei > *ī — even if Kroonen is right (though he presents no real argument either way), nothing in the Germanic data itself allows disambiguating *i from *e in these cases, so maybe we should still give reconstructions with *i.

(We follow the same principle with e.g. Proto-Finnic (*š > *h is post-PF but pan-Finnic) and Proto-Slavic (*ǣ *æ *a *ā > *ě *e *o *a is post-PSl but pan-Slavic); not with Proto-Indo-Iranian, however.)

There's a similar argument to be made with *eN > *iN. This seems to postdate NWG *a-umlaut of *i. E.g. in Old Norse: *benda- > binda, *blendaz > blindr, *brenna- > brinna etc., though to my knowledge this chronology is not very widely accepted (these cases are usually treated as "blocking of umlaut"). --Tropylium (talk) 13:15, 16 April 2018 (UTC)

Also an advance warning: I know a guy about to release a PhD thesis (several papers are already out) where he argues that North Germanic distinguished *i and *e even after all the raisings (*e > *i when unstressed, *eN > *iN, *e-i > *i-i), roughly as *ɨ and *i: this is demonstrated by that *i < *e always triggers later *i-umlaut, but *ɨ < *i in various positions does not (*stadiz > staðr); and *i < *e can be umlauted to y (*sengwa- > syngva), but *ɨ < *i cannot (*liþuz > liðr). --Tropylium (talk) 14:30, 16 April 2018 (UTC)