The forms follow the reconstruction of PG used on Wiktionary (which I think might be a bit anachronistic in some aspects, but that probably doesn't matter here). There are no passive forms or dual forms since they are not needed.
This is the strong class 4 verb *beraną ‘to carry, to bear’. The endings of other strong verbs are very similar. The weak verbs may require some thinking, though.
Present indicative:
*berō
*birizi
*biridi
*beramaz
*birid (*beradiz)
*berandi
This is the same as regular PG. The alternative 2pl is actually the older PG dual. The stem *ber- might be analogically restored to forms that show i-umlaut, so *birizi > *berizi. This happened in Proto-Norse.
Present subjunctive:
*beraų
*beraiz
*berai
*beraim
*beraid
*berain
Again, no difference from regular PG. I think *aų was, at least originally, a disyllabic sequence rather than a diphthong. It is form earlier *ajų.
Imperfect indicative:
*berą
*biriz
*biri
*beram (*berum)
*birid (*berud)
*beran (*berun)
The forms are inherited directly from the PIE thematic imperfective, i.e. they are formed from the present stem with secondary endings. The alternative plural endings in parenthesis are borrowed from the perfect.
Imperfect subjunctive/Personal infinitive:
There's no natural form for the imperfect subjunctive. On the other hand, it apparently only survived in the Romance languages in Sardinian, and in Portuguese and Galician where it was analysed as a personal infinitive. So at least for Portuguese/Galician, maybe you could form a personal infinitive by adding the imperfect endings to the infinitive stem. Note that the 1sg *beraną would be identical to the infinitive so this could be the starting point. Perhaps something like *wiljų beraną ‘I want to bear’ was analysed as ‘I want that I bear’, so that speakers started to (at least optionally) use *wilīz beraniz ‘you want to bear/you want that you bear’.
*beraną
*biraniz
*birani
*beranam (*beranum)
*biranid (*beranud)
*beranan (*beranun)
Perfect indicative:
*bar (*bebar)
*bart (*bebart)
*bar (*bebar)
*bērum (*bebrum)
*bērud (*bebrud)
*bērun (*bebrun)
This is the regular PG past, which basically goes back to the PIE perfect. The form in parenthesis is an alternative reduplicated form (which would of course also be used for other forms based on the perfect stem). It is possible that more strong classes showed reduplication in earlier PG, and that reduplication was progressively lost (note that reduplication was almost completely lost in attested branches other than Gothic).
Perfect subjunctive:
*bērį̄
*bērīz
*bērī
*bērīm
*bērīd
*bērīn
This is the regular PG past subjunctive.
Pluperfect indicative
*barą
*bariz
*bari
*baram (*barum)
*barid (*barud)
*baran (*barun)
This is an innovative form based on the perfect stem but with thematic secondary endings. I'm not quite sure what grade the stem should be in but since it is a thematic inflection, I think it should have the same grade throughout and it seems reasonable that it would share a stem with the singular of the indicative perfect.
Pluperfect subjunctive:
*baraų
*baraiz
*barai
*baraim
*baraid
*barain
Again, this is an innovation. This is based on the perfect stem but with the same endings as the PG present subjunctive. Just like the pluperfect subjunctive, it uses the thematic endings (where the past/perfect subjunctive goes back to an the athematic optative).
Future perfect indicative
*barō
*barizi
*baridi
*baramaz
*barid (*baradiz)
*barandi
Another innovativ form. This is just the perfect stem with present endings, which might make some sense. The Latin future perfect apparently survived as a future subjunctive in some Romance languages, while I don't think the regular future survived in any language.
Imperative:
——
*ber
*beradau
*beramaz
*birid (*beradiz)
*berandau
The 1pl (not found on Wiktionary) and the 2pl is the same as the indicative. There is no 1sg.
Infinitive:
*beraną
Gerund/Present participle:
*berandz > *berandaz
This may start to be declined like a regular adjective.
Supine/Past participle:
*buranaz > *boranaz
Note the a-umlaut causing the shift *u > *o.
Might I suggest *z > r medially but *z > s finally? This might make the distribution of s and r more similar to Latin, where medial s was not that common, due to the Pre-Classical Latin rhotacism. It also helps distinguish the endings *-izi and *-iz in languages that lose the final vowel, so it helps distinguish the present form the imperfect. Also note that in West Germanic, final *z was generally lost while medial *z rhotacized.ixals wrote:z s
I agree that intervocalic voicelessness with initial voicing is very unexpected.Creyeditor wrote:1. I think if you look at naturalistic sound changes, intervocalic /d/ and initial /t/ would be more natural if you view them as intervocalic lenition and word initial fortition. On the other hand having a intervocalic lenition apply twice would be a shameixals wrote: 1. When changing Proto-Germanic words to make them sound more Latin, I have a sound change that changes /θ/ to either /d/ or /t/ depending on which sounds better for the word. I had the idea to actually specify the rules for that. I'm thinking about changing it to /d/ at the beginning of a word and to /t/ everywhere else. Changing it to /t/ between two vowels for examples is better than to /d/ because of the loss of intervocalic /d/ in a lot of Romance languages. But then changing it to /t/ would revert previous Germanic sound changes so I'm not sure about this. Opinions?
Maybe there is some other possibility. /θ/ could become /t/ at the end of a word and /d/ in all other positions? I really don't know
Having it become /t/ only finally is almost the same thing as having it become /d/ in all cases. There are actually very few instances of final *þ in Proto-Germanic. One reason is that Pre-PG, final *t (> PG *þ) became *d (> PG *t), which is true of many IE branches. This final *t was lost in late PG. This is why you have words like *alu ‘ale, beer’ with a stem *aluþ-. The nom-acc.sg was originally *alut.
Perhaps it could become /t/ in all cases? That would make the language look more Romance, e.g. tū instead of dū.
Alternatively, it could become voiceless only if it comes before or immediately after a stressed (i.e. mostly initial) vowel. So somewhat like Verner's law, although not quite. Grammatical words like pronouns can count as unstressed, as in Mainland Scandinavian where initial /θ/ became /d/ in pronouns (du, det etc.) but /t/ in content words (tunn), or as in English (thou, that vs. thin).
Also note that there is often an alternation in Germanic between *þ and *d due to Verner's law, so that means you can pick which one you want and generalize that form.
All attested Germanic languages (possibly excluding some early inscriptions but that may be controversial) pretty regularly lost the second vowel in disyllabic inflectional endings if it was short. So *-aną > *-an, *-amaz > *-amz, *-adiz > *-adz, *-iniz > *-inz etc. The trigger could have been strictly phonological at some point, with analogy restoring some endings, or maybe it was always sensitive to morphology. The change must have spread to all Germanic languages only after they split up, since some languages (notably Old English) show indirect evidence for the final vowel from conditional sound changes.Creyeditor wrote:2. I thought the infinitives in Germanic languages were shortened in every Germanic branch, so I think irregularly doing something to produce nicer outcomes is totally okay.ixals wrote:2. I'm thinking about changing something about the infinitive. In Proto-Germanic it is *-aną most of the time and it would be -ana in VG. But I don't like the outcomes for the Germance languages because its too long in my opinion, so I'm thinking about changing it irregularlytoo *-ane in VG. So the ending in Germanic Spanish (or Germañol) would change from -ana to -án and in Germanic Portuguese (or Germuguês) from -am to -ão. Opinions?
Perhaps there could be a similar shift of final *ą (maybe also *a) to *e if it is in the second syllable of an inflectional ending, or just in the third syllable of a word. So it could be a regular change, just very rare. Note that PG didn't have unstressed short *e except before *r, so some conditional changes to create unstressed *e might be good to have.