Proto-Germanic grammar

The a-stems descended from the PIE thematic inflection and were by far the most common type of noun in Proto-Germanic.[2] Although they could originally be any gender in PIE (as could be seen in Latin), in Proto-Germanic they were restricted to either masculine (ending in -az) or neuter (ending in -ą). The two genders differed only in the nominative, vocative and accusative cases; the other three cases were identical for both. There were two smaller subgroups within the a-stems, ja-stems, and wa-stems. These were declined the same as regular a-stems, but with a suffix -(i)j- or -w- before the ending. It was only in the daughter languages that the wa- and (especially) ja-stems began to diverge significantly from the regular a-stems.

The origin of the genitive singular ending is unclear, as it does not reflect the reconstructed Proto-Indo-European ending *-osyo. Its final voiceless consonant is unique, where normally only *z appears word-finally in inflectional endings. Sometimes *-asa or *-isa is reconstructed, with an extra vowel to account for this.

For the nominative and vocative plural, the northern West Germanic languages attest a voiceless *-s in the ending also. The origin of this is not clear.

The ō-stems descended from the thematic eh₂-stems in PIE,[3] but there were also examples that descended from originally neuter collective nouns that were reinterpreted as feminine singulars. They were the feminine equivalent of the a-stems and were the most common type of feminine noun, with a nominative singular ending in -ō. There were also jō-stems and wō-stems, declined identically to the regular ō-stems but with a suffix before the ending.[4]

The ī/jō-stems descended from athematic/ablauting nouns in *-ih₂/yéh₂-. In Proto-Germanic, they had almost fallen together with the jō-stems, except that they had a nominative/vocative singular form in -ī rather than -jō. They did not survive as a distinct class in any language except possibly Gothic.

The i-stems descended from PIE nouns in -is, many of which were feminine abstract nouns in -tis. They were reasonably common and appeared in all three genders, although neuter i-stems were very rare with only a handful of reconstructible examples. The masculine and feminine i-stems were declined the same, with a nominative singular in -iz. The neuters ended in -i. [5]

The u-stems descended from PIE nouns in -us, many of which were masculine nouns in -tus. They were formally parallel to the i-stems, but because of sound changes they had become a distinct class. They were less common than the i-stems overall and appeared in all three genders, although neuter u-stems were likewise very rare. The masculine and feminine u-stems were declined the same, with a nominative singular in -uz. The neuters ended in -u, but since there are no neuter plurals attested, their plural inflection can only be guessed. [6]

The an-stems were a common group of noun, descended from various kinds of PIE n-stem, all of which were athematic. They were either masculine or neuter, although neuters were rare. Their nominative singular forms ended in -ô. There were also jan-stems and wan-stems, which were declined mostly as regular an-stems. The an-stems correspond with Latin 3rd declension nouns such as homō (gen. hominis) "man" and nomen (gen. nominis) "name." They are also the source of many modern German weak nouns.

The masculine nominative singular ending cannot be reconstructed with confidence, as both North and East Germanic reflect a rather different ending. Old Norse -i and Gothic -a can conceivably come from an ending *-ē, but the source of such an ending is unknown. Kroonen also disputes the genitive singular ending, having the ending terminate in -az instead of -iz;[7] for the sake of relevance, the -az genitive ending termination will be used in discussing declensions of zero-grade–ablauting an-stems in this article.

Some nouns, such as *namô "name," preserved an alternate ablaut pattern, with the suffix in the zero grade in some forms.

Under the controversial Kluge's Law, a paradigmatic n in an an-stem immediately succeeding a stop consonant may fuse to create a geminate stop, in the same context as Verner's Law, seen most prominently in the genitive singular. The dative singular of ablauting an-stems would also be affected by Verner's law itself, causing the voicing of an originally voiceless root-final consonant. This voiceless geminate may then be reduced to a single voiceless stop after a heavy syllable. Furthermore, some an-stems contained root ablaut across their paradigms. All these morphophonological alternations would create multiple stems for a single noun's declension paradigm. The descendants of these ablauting an-stems in the daughter languages were derived from one morpohonological alternant generalized across a paradigm; paradigmatically leveled derivatives would be generally derived from the nominative singular stem, genitive singular stem, dative singular stem, accusative plural stem, or some chimera of these.[7]

Kroonen reconstructs the following masculine an-stems incorporating Kluge's Law (of which he is an ardent supporter) and usually assuming the amphikinetic accent pattern. [7]

One zero-grade–ablauting an-stem in particular, *hrīmô ("hoarfrost"), would thus gain a uniquely irregular declension of its own, due to a combination of Kluge's law alternations, post–heavy-syllable geminate shortening, and the loss of an original root-final f in several forms, such as the nominative singular. This left the word with a nominative singular of *hrīmô and a genitive singular of *hrīpaz.[7]

The ōn-stems were an innovated formation, created by attaching n-stem endings to older feminines in -ō. They were likewise always feminine, and acted as the feminine counterpart of the an-stems. They probably ended in -ǭ, but that is not certain. There were also jōn-stems and wōn-stems, declined identical to regular ōn-stems.

Kroonen additionally reconstructs a single root-ablauting ōn-stem noun, *bijǭ ("bee") (which he reconstructs as *bīō in his published work), which he states to have an oblique stem in *bin-.[7]

This group of nouns contained only a single type of abstract noun, formed by attaching an īn-suffix to adjectives. They were always feminine, and were essentially identical to ōn-stems, with ī replacing ō in all the forms. The nominative singular ended presumably in -į̄.

The r-stems were limited only to a group of five close kinship terms: *fadēr ‘father’, *mōdēr ‘mother’, *brōþēr ‘brother’, *swestēr ‘sister’ and *duhtēr ‘daughter’. Their inflection was somewhat unusual, but generally resembled other consonant stem nouns.[8]

The z-stems descended from PIE acrostatic neuters in -os/es-. They were fairly rare and always neuter. They were formed similarly to an-stems, but with z replacing n. Their nominative singular forms ended in -az.[10]

Nouns in this group are usually just called 'consonant stems'. It was mostly a class of remnants, consisting of PIE root nouns (nouns with no suffix) and nouns with a suffix ending in a consonant other than n, r or z. There are few reconstructible neuters; those that can be reconstructed were irregular. It is possible that many reconstructible a-stem neuters originally belonged to this class, however.

The nominative singular cannot be reliably reconstructed. Purely etymologically, *-s would be expected, or perhaps *-z after a voiced consonant. However, it is likely that the final consonant cluster would have undergone simplification, as it did in other Indo-European languages such as Latin. This is not attested in any Germanic language, so this remains speculative. [11]

There were several irregular nouns in this class. The most prominent are *tanþs ‘tooth’ and *wrōts ‘root’, which preserved ablaut and Verner alternation in the root.[12] The noun *mili ‘honey’, one of the few reconstructible neuters, had lost its stem-final -t but kept it as -d- in the oblique cases.[13] The nominative singular of the noun *mann- ‘human, man’ has so far not been conclusively reconstructed.

These were by far the most common type of adjective, and even in the oldest languages (except Gothic) there was a tendency for all adjectives to be declined alike. The adjective *gōdaz "good" is given here as an example. [16]

Although these three classes were originally different, sound changes had made them largely identical in the feminine forms. Only the nominative singular differed: it was -jō for the ja/jō-stems, but -ī for the i- and u-stems. The masculine and neuter forms are uncertain, but may have been identical to the ja-stem adjectives already by analogy with the feminines (as they are in Gothic). Only the nominative singular forms remained distinct, the masculines ending in -jaz, -iz, -uz and the neuters in -ja, -i, -u.[17]

The inflection of present participles in -nd- is likewise difficult to reconstruct. The feminines inflected as i- and u-stems, ending in -ī in the nominative singular but -jō- in the other forms. The masculines and neuters may have already acquired ja-stem endings, but it is certain that the nominative ended in -ndz or -nds.[18]

This declension class was not a separate class of adjectives. Rather, adjectives could sometimes take this declension instead of their own strong declension. The weak declension was identical to the an-stem and ōn-stem declensions of nouns. Comparatives and ordinals used an alternative variety of the weak inflection, in which the feminine forms were not those of the ōn-stem nouns but of the īn-stems. [18]

Proto-Germanic had a demonstrative *sa ‘that, those’ which could serve as both a demonstrative determiner and a demonstrative pronoun. In daughter languages, it evolved into the definite article and various other demonstratives. [19]

Proto-Germanic possessed a general anaphoric pronoun *iz ‘he, she, it, etc.’ that was used as a third-person personal pronoun. It was inflected as follows:[20]

The interrogative pronoun *hwaz ‘who, what’ was inflected likewise, but without plural forms. The feminine forms were probably rarely used, only if the person or thing being asked about was known to be feminine.[21]

There were several other pronouns and determiners in use, such as *jainaz ‘yon, that over there’, *aljaz ‘other’ and *allaz ‘all’. These were declined as strong adjectives, usually with no weak form. The proximal demonstrative *hiz ‘this’ was inflected as *iz. Neither pronoun survived in Old Norse, both survive in Gothic, and the two were eventually conflated in West Germanic, with the northern languages using the forms with h- (as English he) and the southern languages those without (German er). The Old Norse formation of the masculine and feminine singular pronouns is not fully understood, but appears to go back to a form *hanaz.

The inflection of the first- and second-person pronouns was very different from any other kind of nominal, with case forms that did not match those of other nominals. As in Proto-Indo-European, the case system was markedly reduced - only four cases were distinguished. The dative and instrumental were identical, and the nominative was used as a vocative as well. The paradigms were suppletive, with different roots for the singular and dual/plural, and also with different roots for the nominative and non-nominative.

Unique within Germanic was that the pronouns of the first and second person retained distinct dual forms, which referred specifically to two individuals. Verbs also retained distinct dual forms in the first and second person, which agreed with the pronouns. The dual was lost in other nominals, and therefore the third-person dual of verbs was lost as well since verb–subject agreement was no longer possible. Although the dual pronouns survived into all the oldest languages, the verbal dual survived only into Gothic, and the (presumed) nominal and adjectival dual forms were lost before the oldest records. As in the Italic languages, it may have been lost before Proto-Germanic became a different branch at all.

¹ – Unstressed variant

The genitive forms of the first and second person pronouns were inflected as strong adjectives, unlike other genitive forms which were uninflected.[23]

Strong verbs used ablaut (i.e. a different vowel in the stem) and/or reduplication to mark the past tense. Almost all Proto-Germanic strong verbs were thematic verbs formed directly from a verbal root, although a few relic verbs exhibited other formations such as j-present (from PIE -ye/yo-) or n-infix verbs.[24]

Strong verbs were divided into seven main classes, distinguished by the ablaut pattern, which in turn was usually determined by the shape of the root. The first six classes formed the past tense with ablaut alone, and the seventh class formed it through reduplication. However, there were some class 7 strong verbs that showed both ablaut and reduplication. The reduplication itself was performed by taking the first consonant of the root and prefixing it to the stem, with the vowel e in between. If the word began with s- followed by another consonant, that consonant was also duplicated.[25]

The initial s- of the original stem apparently underwent Verner's law alternation when the reduplicating prefix was added, and became -z- (as in Gothic saizlēp for the past tense of slēpan, and the four Icelandic ri-verbs).[26]

Verner's law alternations (f-b, þ-d, h-g, hw-gw, s-z) generally occurred in the final stem consonant of all strong verbs. The first and second principal parts showed the voiceless alternant, and the third and fourth showed the voiced alternant. This phenomenon is known as Grammatischer Wechsel, and survives in several West Germanic languages up to today (as in English was and were).[27]

The alternations were somewhat more complicated in verbs containing labiovelars (hw, kw or gw), since these underwent further changes in Germanic. Generally, labiovelars were delabialised and became h, k or g when they stood next to -u-, or were preceded by -un-. This applied both to stem-initial and stem-final labiovelars, and could be seen in verbs such as *strīkwaną "to stroke", where the -u- of the ending in the past nonsingular indicative (but not subjunctive) triggered delabialisation. In *kwemaną "to come", the -u- of the past participle stem itself triggered the change.[28]

Since gw (when it had not been delabialised at an earlier stage) generally became w in Germanic unless preceded by n,[29] verbs ending in gw or hw (which became gw through Verner alternation) were particularly complex. If gw was preceded by n, it remained as such but the usual delabialisation rules applied. This could be seen in *þrinhwaną "to press". If gw was not preceded by n, then it became w unless next to u, in which case it was delabialised and appeared as g instead. The verb *sehwaną "to see" demonstrates this three-way consonant alternation.

The 2nd person singular past indicative ending -t generally triggered assimilation of the preceding consonant according to the Germanic spirant law:[30]

• bt, pt > ft
• dt, tt, þt > st
• gt, kt, hwt > ht

Strong verbs had four principal parts, from which the remaining forms could be derived. Principal part 1 was the present tense, part 2 was the past singular indicative, part 3 was the remainder of the past tense, and part 4 was the past participle. If the vowel of part 1 contained -e-, it became -i- when the following ending began with -i- through i-mutation; this occurred in the 2nd and 3rd person singular forms, and the 2nd person plural form. This also happened in the entire present tense of all j-present verbs.[31]

An example verb of each class is shown here, using different forms to demonstrate different principal parts. The first principal part is demonstrated by the infinitive without i-mutation, and by the present third-person singular indicative with i-mutation. The past third-person singular indicative is used to demonstrate the second principal part, and the second-person form with its ending -t is shown as well. The past third-person plural is used for the third principal part, and the past participle for the fourth.

An example verb *nemaną "to take" is shown here to illustrate the inflection of strong verbs. Other strong verbs were inflected analogously, but with different vowels in the root and/or reduplication of the initial consonant(s). The j-present verbs were inflected like weak class 1 verbs in the present tense, but dropped the j-suffix in the past tense and then inflected like regular strong verbs.

Weak verbs used a suffix containing a dental consonant to mark the past tense. This suffix is now generally held to be a reflex of the reduplicated past imperfect of PIE *dʰeh₁- originally "put", in Germanic "do". They were generally derived verbs, being formed from nouns, verbs or adjectives (so-called denominal, deverbal and deadjectival verbs). A few underived verbs that were originally strong j-present verbs acquired weak past tenses in Proto-Germanic.

Weak verbs were divided into five main classes, of which four survived into the distinct history of the daughter languages. One class was formed by a few relic verbs which have no present-tense suffix, but the verbs in the other classes had a present-tense suffix, which varied from class to class. The past tense endings were always identical, and resembled those of strong verbs, but prefixed with a dental infix. They were as follows:[32]

The initial consonant of the suffix was normally -d-, but the class 1 j-present verbs, the suffixless weak verbs and the preterite-present verbs had -t- if the ending consonant of the stem was an obstruent, in which case the obstruent assimilated to the dental.

Already a small relic class in Proto-Germanic, at most three verbs are reconstructible: *bringaną "to bring", *brūkaną "to need", *būaną "to dwell". Their present tense was identical to the present of strong verbs, and their past tense was formed with no linking vowel; their third-person singular past indicative forms were *branhtē, *brūhtē and *būdē respectively. This class did not survive as a distinct class in any daughter language, and its verbs were moved to other classes of verb.[33]

Class 1 consisted of verbs ending in -(i)janą, and has a past tense in -id-.[34] The present tense suffix varied between -ja/ija- and -i/ī-.

A significant subclass of class 1 weak verbs were causative verbs. These were formed in a way that reflects a direct inheritance from the PIE causative class of verbs. PIE causatives were formed by adding an accented suffix -éye/éyo to the o-grade of a non-derived verb. In Proto-Germanic, this suffix survives as -j/ij-, and is affixed to the stem of the strong past tense with Verner's Law voicing applied (originally due to the placement of the accent on the suffix). Examples:[35]

• *bītaną (I) "to bite" → *baitijaną "to bridle, yoke, restrain", i.e. "to make bite down"
• *rīsaną (I) "to rise" → *raizijaną "to raise", i.e. "to cause to rise"
• *beuganą (II) "to bend" → *baugijaną "to bend (transitive)"
• *brinnaną (III) "to burn" → *brannijaną "to burn (transitive)"
• *frawerþaną (III) "to perish" → *frawardijaną "to destroy", i.e. "to cause to perish"
• *nesaną (V) "to survive" → *nazjaną "to save", i.e. "to cause to survive"
• *ligjaną (V) "to lie down" → *lagjaną "to lay", i.e. "to cause to lie down"
• *faraną (VI) "to travel, go" → *fōrijaną "to lead, bring", i.e. "to cause to go"
• *faraną (VI) "to travel, go" → *farjaną "to carry across", i.e. "to cause to travel" (an archaic instance of the o-grade ablaut used despite the differing past-tense ablaut)
• *grētaną (VII) "to weep" → *grōtijaną "to cause to weep"
• *lais (I, preterite-present) "(s)he knows" → *laizijaną "to teach", i.e. "to cause to know"

An example class 1 weak verb *dailijaną "to divide" is shown here in the present tense. Note that because of Sievers' law, there are two possible endings in the present tense. One set, the one shown here, is used for long stems, and has -ij- and -ī-. The other set, used for short stems, has -j- and -i-. In the past, both sets have -i-.

A few irregular verbs in this class were j-presents, and had the suffix only in the present tense, analogous with the strong j-presents. Their past tense replaced the initial -d- with -t-, with the preceding consonant assimilating to the suffix according to the Germanic spirant law:

Class 2 consisted of verbs ending in -ôną. They had a past tense in -ōd-, and the present tense suffix was trimoric -ô-. It was originally a denominative subclass of class 1, formed from nouns that ended in -ō. However, because of the loss of -j- between vowels, the surrounding vowels contracted, creating a distinct class. Already in Proto-Germanic, new verbs of this class had begun to be formed from nouns of other classes. It would later become the primary denominative class in most of the daughter languages.[36]

The verb *salbôną "to anoint" is given here as an example. Notice that because of the vowel contraction, the indicative and subjunctive have mostly become alike.

This class had two subclasses, which were mostly different in the forms, but they shared the suffix -ai- in some. The two subclasses merged into one in all the later languages for reasons that are not quite clear, but presumably the fact that they shared some endings may have had something to do with it.

The first and larger subclass formed stative verbs. They had an infinitive in -(i)janą and a past tense in -d- with no linking vowel (but generally with no assimilation either). The present tense suffix varied between -ja/ija- and -ai-. These verbs were statives. The verb *sagjaną "to say" is shown here. Like class 1 weak verbs, the -j- varied with -ij- depending on the length of the stem.[37]

The second subclass formed factitive verbs. They had an infinitive in -āną and a past tense in -ād-, with -ā- having contracted from earlier -aja- after loss of intervocalic -j-. The present tense suffix varied between -ā- and -ai-. These verbs were factitives, similar to the first class of weak verbs. It was already a small class in Proto-Germanic, though it may have remained marginally productive. The verb *þewāną "to enslave" is shown here.[38]

This class has been notoriously difficult to reconstruct, but some details are known. The infinitive ended in -naną, and the past tense was formed with -nōd-. The present tense forms are uncertain, but probably varied between -ō- and -a-. These verbs were inchoatives, and indicated a change of state or the process of coming into that state. As a result, they were always intransitive, and had no passive forms or a past participle. The verb *liznaną "to learn" is given here, but note that these reconstructions are very uncertain.

Preterite-present verbs were in principle weak verbs, since they formed their past tense with the weak dental suffix. However, they were unique in that their present tense forms were not those of the other verbs; rather, they were identical to the past tense forms of strong verbs. They descended from old Indo-European verbs that retained their stative meaning rather than being reinterpreted as past tense forms.