Gothic attests a few Germanic forms of the suffix *-gaz with *-haz:
... this seems to show that at least some forms of pre-Germanic *-kos showed a process of drawing the accent away from the initial syllable to the following:
In present use, the aorists show full-grade in the singular and zero-grade elsewhere, as in 3sg. *kwimidi, 3pl. *kumandi; *knididi, *knudandi; *swifidi, sufandi, *trididi, *trudandi:
| Person | *kwemaną | *knedaną | *swefaną | *tredaną |
|---|---|---|---|---|
| 1 Sg | *kwemō | *kneþō | *swefō | *treþō |
| 2 Sg | *kwimizi | *kniþizi | *swifizi | *triþizi |
| 3 Sg | *kwimidi | *kniþidi | *swifidi | *treþidi |
| 1 Pl | *kumamaz | *knudamaz | *sub/famaz | *trudamaz |
| 2 Pl | *kumiþ/d | *knudiþ/d | *sub/fiþ/d | *trudiþ/d |
| 3 Pl | *kumanþ/di | *knudanþ/di | *sub/fanþ/di | *trudanþ/di |
The early or pre-Germanic root aorist in past use leaves its traces in the preterites of stative-present *kunnaną and *unnaną, and also somewhat differently in the verbs *hlaþaną, *standaną and *waldaną.
In the former, the irregular preterite stems *kunþ- and *unþ- presumably go back to *ǵnéh₃-t > *ǵń̥h₃-t, *h₃én-t > *h₃ń̥-t, where the zero-grade seems to have spread from the non-singular *ǵn̥h₃-ént, *h₃n̥-ént.
In the latter, the preterites *hlōþ, *stōþ and *wulþ (< *kléh₂-t, *stéh₂-t and *h₂wĺ̥h₁-t), seem to have influenced the creation of present forms also containing a dental. Although in the case of *waldaną, it's not improbable that the present was replaced by the compound *h₂wolh₁-dʰh₁- as also found in Balto-Slavic.
The lengthened grade -ē- found in the oblique of class 4 and 5 strong preterites probably spread from the singular indicative where it was originally confined, presumably in a Narten fashion of 3sg. **kwēm, 3pl. **kwemun. This is corroborated by the abnormal "o-grades" found in some deverbal formations where -a- is expected: *nēmą, where **namą (< *nemaną) is expected, *ētą for **atą (< *etaną), in a small class of o-grade deverbal neuter a-stems; numerous causative-iteratives which notoriously take the o-grade, some with traditional o-grade variants:
... and some with no traditional o-grade variant at all:
Eventually the singular -ē- was replaced by the regular o-grade -a- in all verbs except one, *etaną, which somehow preserved the transitional phase as 3sg. *ēt, 3pl. *ētun.
A list of formations found in Germanic that point to earlier, presumably more complex, formations.
Stressed
The second -a- in *ana- likely resulted from a residual laryngeal schwa, requiring stress to survive. This could show that pre-Germanic preverbs/prefixes carried stress. Otherwise, it represents a secondary *an- + *-o as in *aba-, *uba-.
Unstressed
The PIE thematic present middle endings should presumably have given the following in Proto-Germanic, assuming no other changes:
| Person | Indicative | Subjunctive |
|---|---|---|
| 1 Sg | *-ōi | *-ai |
| 2 Sg | *-idai | *-aid |
| 3 Sg | *-idai | *-aid |
| 1 Du | *-awid | *-aiwid |
| 2 Du | *-idą | *-aidą |
| 1 Pl | *-amid | *-aimid |
| 2 Pl | *-idu | *-aidu |
| 3 Pl | *-andai | *-air |
Instead, we see the analogical change of 2sg *-idai/-aid to *-izai/-aizau, replacement of subj 3pl *-ro with *-nto, spread of the marker *-ai from the 2sg/3sg/3pl to all forms in the indicative, thus the corresponding spread of the marker *-au (from whatever source) through the subjunctive. Presumably the unsyncretized Proto-Germanic passive should have been the following:
| Person | Indicative | Subjunctive |
|---|---|---|
| 1 Sg | *-ai | *-au |
| 2 Sg | *-izai | *-aizau |
| 3 Sg | *-idai | *-aidau |
| 1 Du | *-awidai | *-aiwidau |
| 2 Du | *? | *? |
| 1 Pl | *-amidai | *-aimidau |
| 2 Pl | *-idwai | *-aidwau |
| 3 Pl | *-andai | *-aindau |
The 2nd dual forms are harder to predict, as depending on the date of the spread of the markers, we could expect: very early *-idamai/-aidamau, early *-idanai/-aidanau or late *-idai/-aidau. The latter might be preferable, being less marked, contributing in part to the syncretism shown in the Gothic passive.
Presumably the pre-Germanic tense system had developed into the following:
| Present (imperfective/perfective) | Perfect (stative/resultative "present") | |
|---|---|---|
| Imperfect (imperfective past) | Aorist (perfective past) | Pluperfect (stative/resultative past) |
The perfect describing a present state resulting from a past action, while a pluperfect would have described a past state resulting from a further past action.
The exact route this took from pre-Germanic to Germanic is quite complex.
The irregular verb do seems best at preserving morphological traces of this system:
| Person | Present | Imperfect | Aorist | Perfect | Pluperfect |
|---|---|---|---|---|---|
| 3rd sg. | *dōþi < *dʰóh₁ti | *dedē < *dʰédʰeh₁t | *dēþ < *dʰéh₁t | *dedō < *dʰedʰóh₁e | *didi? < dʰedʰh₁et? |
| 3rd pl. | *dōnþi < *dʰóh₁n̥ti | *dedun < *dʰédʰh₁n̥d | *din? < *dʰh₁énd | *dedun < *dʰedʰh₁ń̥d | *dedan? < *dʰedʰh₁ond? |
The present stem reflects an athematic o-grade present, for whatever reason replacing the expected reduplicated athematic present.
The expected aorist non-singular stem was probably already regularized to *du- (*dun < *dʰh₁ń̥d). Eventually even this was replaced by the newer stem *dēd- after the reinterpretation of *dēþ as a bare stem. Compare *stōþ > *stōdun, *hlōþ > *hlōdun for similar formations.
Old English dyde seems to reflect a stem *dud-, possibly from a thematic pluperfect with analogical introduction of secondary zero-grade from aorist non-singular *du-, giving 2sg. *dudiz, 3sg. *dudi.
That *du- had become a secondary zero-grade stem is perhaps further proved by the secondary zero-grade Old English dyde, "a deed", < *dudiz replacing earlier *daþ/diz. Compare Latin conditiō, Celtic *kondatis, Ancient Greek θέσις. Given this, it is not much of a leap to assume that the inherited to-stem participle had become *dudaz from earlier *dadaz.
Therefore, a possible rendering of the early Germanic active 'do'-paradigm could have appeared as follows:
| Tense | Mood | 1st Sing. | 2nd Sing. | 3rd Sing. | 1st Dual | 2nd Dual | 1st Plur. | 2nd Plur. | 3rd Plur. |
|---|---|---|---|---|---|---|---|---|---|
| Present | Indicative | *dōmi | *dōsi | *dōþi | *dōs | *dōþiz | *dōmaz | *dōþ | *dōnþi |
| Subjunctive | *dōį | *dōis | *dōi | *dōiw | *dōiþiz | *dōim | *dōid | *dōin | |
| Imperative | - | *dō | *dōþau | - | *dōþiz | - | *dōþ | *dōnþau | |
| Imperfect (Derived from o-grade) | Indicative | *dǭ | *dōs | *dō | *dōw | *dōþiz | *dōm | *dōþ | *dōn |
| Imperfect (Inherited) | Indicative | *dedę̄ | *dedēz | *dedē | *dedū | *dedudiz | *dedum | *dedud | *dedun |
| Aorist | Indicative | *dēþ | *dēst | *dēþ | *dēdū | *dēdudiz | *dēdum | *dēdud | *dēdun |
| Subjunctive | *dēdį̄ | *dēdīz | *dēdī | *dēdīw | *dēdīdiz | *dēdīm | *dēdīd | *dēdīn | |
| Perfect | Indicative | *dedō | *dedōt | *dedō | *dedū | *dedudiz | *dedum | *dedud | *dedun |
| Subjunctive | *dedį̄ | *dedīz | *dedī | *dedīw | *dedīdiz | *dedīm | *dedīd | *dedīn | |
| Pluperfect | Indicative | *dudą | *dudiz | *dudi | *dudau | *dudadiz | *dudam | *dudad | *dudan |
A little more speculatively, it seems possible that each tense could at one time have been given respective participles:
It is possible that Germanic weak verb classes, especially those of the third, fourth and lost stative *-ēną-class, played a larger role in describing verb transitivity to a more extensive extent in earlier Germanic.
Presumably the third class originally supplied a general inchoative aspect, as seen in the difference between *dugaiþi, "to become useful > to help" against *duganą, "to be useful". The role of simple stative present in relation to the inchoative seems to have been supplied by preterite-presents, j-stem verbs and probably also at one time by the *-ēną-class of verbs attested only in the derivatives *armēǭ and *fagēþiz. Competition between the three seems to have resulted in the almost complete disappearance of the latter verb class.
This interpretation of the third class, one of an inchoative linked with a stative companion, would help explain certain confusion between forms in the daughter languages, including verbs that show umlaut/lack of umlaut, gemination/lack of gemination and so on, as eventually the inchoative verbs themselves gradually developed into simple stative verbs in competition with their original companion.
| Inchoative | Stative |
|---|---|
| *armaną "to become miserable (for... ) > to pity" (wk3) | *armēną "to be miserable" (stative -ē-) |
| *duganą "to become useful > help" (wk3) | *duganą "to be useful" (pp) |
| *faganą "to become pleased" (wk3) | *fagēną "to be pleased" (stative -ē-) |
| *habaną "to become in possession of" (wk3) | *habjaną "to be in possession of, have" (wk1) |
| *libaną "to become left, survive" (wk3) | *libjaną "to be left > to remain/live" (wk1) |
| *munaną "to become thought of(?) > to be minded" (wk3) | *munaną "to think, consider" (pp) |
The third class might have also served to derived inceptive verbs from activity verbs:
| Activity | Inceptive |
|---|---|
| *burgijaną "to keep, preserve" (wk1) | *burganą "to begin to keep > hold onto" (wk3) |
| *fulgijaną "to follow" (wk1) | *fulganą "to begin to follow > follow after, pursue" (wk3) |
| *sagjaną "to tell, recount" (wk1) | *saganą "to begin to recount > say" (wk3) |
The fourth class might have served for anticausative aspect, derived from causative/transitive verbs.
In Germanic there seems to be a class of nouns that declines in the "strong" cases with initial stress, Cv́C-uC-, while receiving oxytone or paroxytone stress in the "weak" cases, CvC-íC-; while additionally being marked by a -u- in the originally unstressed syllable in the former, but with an -i- in the originally stressed syllable in the latter.
The best theory to come to one's mind at the moment that would make sense of all the forms and accentual shifts, would be that this type of inflection/derivation was grammaticalized from earlier collective formations like like CéC-l̥/r̥ or CóC-l̥/r̥ (obl. CC-én-), which would result in Germanic forms like CeC-uR- (obl. CeC/G-iR-, where G is a mutable consonant by Verner's law). This would also help explain its perceived collective capacity in some formations.
This class might also by the basis for the collective suffixes *-idją, *-iþją, from the quite numerous dental oblique stems plus the normal collective *-ją. When the stems are collectivized with *-ją, they seem to prefer the oblique stem, *apilją, *aritją, *alizją
| Case | *falud | *hafud | *halud | *rakud | *werud |
|---|---|---|---|---|---|
| Nominative | *falud | *hafud | *halud | *rakud | *werud |
| Vocative | *falud | *hafud | *halud | *rakud | *werud |
| Accusative | *falud | *hafud | *halud | *rakud | *werud |
| Genitive | *faliþiz | *haubiþiz | *haliþiz | *rakiþiz | *wiriþiz |
| Dative | *faliþi | *haubiþi | *haliþi | *rakiþi | *wiriþi |
| Instrumental | *faliþē | *haubiþē | *haliþē | *rakiþē | *wiriþē |
Animals
Material/Substance
People/Body part
Places
Plants
Tools/Structures
The Germanic class 3 weak verbs are a merging of at least two pre-Germanic verbal paradigms, a denominal formation and a middle-esque formation.
The first is an unproblematic formation of factitive/causative verbs cognate with Ancient Greek -όω, an innovated denominal formation which kept the nominal thematic *-o- vowel instead of replacing it with *-e-, as is represented by the first weak class. So an adjective like *bataz > *bʰodós would be derived as *bʰodóyeti, as opposed to to *hailijaną, "to make whole" from *hailaz, "whole", < *koyl-e-yeti < *koylos. From this formation, past participles ended in *-adaz.
The second is, according to Jasanoff, a middle formation that's reflected in Sanskrit duhe, a competing form to dugdhe, which represents PIE *dʰugʰ-ói, "... is useful" > with peculiar Indic semantics "providing milk". To verbs of this formation a suffix *-yéti was added, giving pre-Germanic *-o(i)yéti. In Proto-Germanic the forms *-oyé- and *-oyó- developed into *-ai- and *-ā-, respectively. From this formation, past participles ended in *-aidaz.
This type of deponent middle formation probably indicated stative, inchoative and reflexive semantics: stative *wakāną, "to be vigilant" > "to be awake"; inchocative *kaldāną, "to become cold"; reflexive *armāną, "to become as (treat oneself as) poor" > "to show mercy".
Verbs that have confused the reconstruction of this class, namely *habāną, *sagāną, *libāną, are to be understood as derived and secondary:
Subsequently the stem *-ai- > *-ē- was leveled through the paradigm in Old High German, probably because the stem alternant *-ā-, unstressed *-a-, was indistinct with respect to other verb classes.
While the third weak class survives in Old Norse, it is far less used than the other two surviving weak classes, 1 and 2. This is very likely because the development of the North Germanic mediopassive formation in -sk took on much of its semantic responsibility.
In the other West Germanic languages, the 3rd weak class was merged with weak class 2 verbs, possibly because that class was a very productive destination for verbs from other classes, or perhaps the class had been reconstituted as *-ā-j-aną, just as weak class 2 was as *-ō-j-aną, with the result ending up being indistinguishable due to phonological developments.
In Proto-Germanic, when adverbs and prepositions were attached to nominals and adjectives, they would expect to bear stress, and therefore show a non-Verner voicing; on the other hand, when they were attached to verbs, they bore no stress. In the North-West Germanic languages, almost all prefixes/preverbs show overwhelming Verner's voicing, with only a few forms without, showing that this branch had largely leveled out the stressed/non-Verner alternants. In East Germanic, this had gone the other way, with the stressed/nominal alternate forms leveling out the unstressed/verbal forms, except for ga-, bi-, and possibly uz-, which secondary became unstressed, but does show assimilation to following -rs, cf urrinnan.
Below is a list of most, if not all prefixes/preverbs in the Germanic lexicon, showing both the voicing forms, with non-Verner's on the left, and Verner's on the right. Forms that show no difference either way are listed singly.
Some forms show not just a voiced alternant, but possibly also an ablaut variant.
Strong verbs derive primary verbal adjectives in at least three ways:
The semantics of these suffixes and formations can at least be split into two:
Characterized entirely by the "gerundive"-type of adjective. Rather than term it "gerundive" however, a word which is characterized by more meanings than the one that characterizes this class, it would be more preferable to term this either "passive derverbal" or "patientive deverbal", since semantically it seems to be somewhat more patientive or passive, indicating what a verb can or could do, often to the subject. Resembles Latin deverbal adjectives in -bilis
Semantically this type of formation seems to imply simple activity, or "doing ... (verb)", similar to Latin deverbal adjectives in -īvus
With the suffix *-ra-:
To this we add forms with the suffix *-ula-:
In Latin, where both PIE *ǵn̥h₁-tós, *ǵn̥h₃-tós are expected to have given homophonous gnātus and *gnātus, Latin (and probably Italic) instead dissimilated the terms to gnātus and gnōtus, introducing the full-grade secondarily to the latter form. As these roots and their derived forms are very basic and common, it seems very likely that speakers would attempt to dissimilate them where possible.
In Germanic, we see that speakers did the very same thing, but in a different fashion. Rather than introduce a full-grade to a root with expected zero-grade, which goes against the regularity of Germanic ablaut formations, especially with regards to verbs, Germanic speakers instead opted for contrast in accent, which secondarily gave contrast in voicing. Therefore, where *ǵn̥h₁-tós and *ǵn̥h₃-tós are expected to have given homophonous *kundaz and *kundaz (from earlier *kuntós and *kuntós), Germanic speakers instead opted for contrasting accent (*kuntós, *kúntos) between the two, which with Verner's alternation gave way to voicing contrast, *kundaz, "born", against *kunþaz, "known". This voicing also shows itself spreading to related formations, *kindiz/*kundiz, "kind", and *kinþiz/*kunþiz, "knowing", and probably most noticeably to the preterite stem of *kunnaną: *kunþē, *kunþēdun, although this isn't too surprising, as the preterite root is based on the participle stem.
Several Germanic words reflect an e-grade root, with or without oxytonic accent, with which several extra-Germanic cognates can provide evidence for original o-grade. Additionally are added forms that show secondary e-grade within Germanic itself, although these vary in certainty.
Within Germanic there are a small subset of verbs that feature full grade only in the singular present, with zero-grade in the non-singular present:
| Person | *kumaną | *knudaną | *subaną | *trudaną |
|---|---|---|---|---|
| 1st Singular | *kneþō | *kwemō | *swefō | *treþō |
| 2nd Singular | *kniþizi | *kwimizi | *swifizi | *triþizi |
| 3rd Singular | *kniþidi | *kwimizi | *swifidi | *triþidi |
| 1st Plural | *knudamaz | *kwimizi | *subamaz | *trudamaz |
| 2nd Plural | *knudiþ | *kwimizi | *subiþ | *trudiþ |
| 3rd Plural | *knudanþi | *kwimizi | *subanþi | *trudanþi |
While there is no certainty on how other roots came to belong to this group, some verbs like *knudaną and *trudaną above came from aorist-presents; that is, aorists being re-classed as the present form of a verb, rather than taking the usual root towards becoming preterite.
There seems to be another subset of verbs that act in the same manner, albeit with a singular stem extended with *-j- earlier *-y- or *-i- (keep in mind -ji- resolves to -i-):
| Person | *hafjaną/*habaną | *ligjaną/*leganą | *sitjaną/*setaną |
|---|---|---|---|
| 1st Singular | *hafjō | *ligjō | *sitjō |
| 2nd Singular | *hafizi | *ligizi | *sitizi |
| 3rd Singular | *hafidi | *ligidi | *sitidi |
| 1st Plural | *habamaz | *legamaz | *setamaz |
| 2nd Plural | *habiþ | *ligiþ | *sitiþ |
| 3rd Plural | *habanþi | *leganþi | *setanþi |
This would explain why the latter two show up as if from *leganą and *setaną in Gothic, while also explaining the widespread confusion between *haf- and *hab- in the daughter languages. Again, how they came to be conjugated as such is uncertain to one at the moment; at least the root *sed- formed a root aorist, from which it might have carried on, but the other two don't seem to have had such formations. Perhaps *leg- acquired it by semantic association with *set-.