Determining the structure of the sentence in Proto-Germanic presents problems, because most of the early texts are either poetic or translations. The Runic texts are exceptions, but they have requirements of style and arrangement on stones so that not all of them are without problems, and many are very short. Nonetheless, as a result of their antiquity and native tradition, in determining the basic structure of the sentence we rely heavily on them as texts that appear to be least modified by literary conventions or introduction of non-native patterns through translation. Evidence based on them, as well as on early texts that are not translations, leads to the conclusion that the order was subject-object-verb (SOV, simply represented as OV).
It is generally agreed that the Gallehus Runic inscription provides a typical pattern of the Proto-Germanic sentence structure. Dated to the fifth century A.D., it is a straightforward statement framed as an alliterative poetic line.
ek hlewagastiz holtijaz horna tawido.
I Hlewagastiz of Holt horn I-made
'I, Hlewagastir of Holt, made the horn.'
The subject ek is followed by two appositional nominals; these in turn are followed by the object, which occupies the most important place in the alliterative poetic line, and finally by the verb. This inscription and others provide strong support for the conclusion that the structure of the sentence in Proto-Germanic was SOV. Moreover, its intonation pattern may be assumed to have the principal stress on the major alliterating syllable, which is generally the first accented syllable in the second half-line. The alliterating syllables of the first half-line would have had less stress, and the final stress would have been weak, as indicated by its lack of alliteration. It would have been accompanied by weakening and downward pitch, indicated here by #. The pattern is represented as 2 – 3 – 1#, usually given as 231#.
A longer Runic inscription on the Tune stone supports this conclusion even though it is not in accordance with poetic requirements, although the first half-line is somewhat comparable to a line of alliterative poetry. Making considerable use of apposition, the two sentences are verb-final.
ek wiwaz after woduride witadahalaiban worahto
I Wiwar after Woduridar Breadward produced [this]
[meR] woduride staina þrijoz dohtriz dalidun
(for) me Woduridar stone three daughters arranged
I, Wiwar, produced this for Woduridar, the Breadward.
For me, Woduridar, [my] three daughters arranged the stone,
the chief inheritors among the heirs.
Many additional examples of SOV order could be supplied from the earliest literary texts, such as the first three lines of the Old English Beowulf. Moreover, for other syntactic constructs than word order these are more useful than the runes, which as in the examples cited above are generally short and straightforward with few modifiers.
Hwæt, wē Gār-dena in gēardagum
þēodcyninga þrym gefrūnon,
hū ðā æþelingas ellen fremedon.
Listen, we have heard of the glory of the Spear-danes,
of the kings of the people, in days of old,
how the heroes performed deeds of valor.
The two verbs stand at the ends of their clauses, as well as of the poetic lines. In the second and third lines the objects þrym and ellen occupy the major alliterative position. The two genitives precede the noun they modify. These lines and many others that might be cited from the early poetry provide strong support for the conclusion on OV order, and on the intonation of simple clauses, also evident in the Runic examples.
Among other patterns characteristic of OV sentence structure are comparative constructions attested in the early texts. Like clauses with objects of verbs, they are based on transitivity. In OV comparative constructions, the noun being compared, as the "standard" in the construction, precedes the adjective, much as objects precede verbs; it is usually in the dative case, but occasionally in the genitive. Numerous examples are attested, such as these in Old Norse and Old English:
sal sér hon standa, sólo fegra, (Vǫluspá 64.1-2)
hall sees he stand from-sun fairer
'He sees a hall standing (there), fairer than the sun.'
Among examples in Beowulf is line 1850:
þæt þē Sǣ-Gēatas sēlran næbben
that than-you the Sea-Geats better-one not-have
that the Sea Geats do not have a better one than you
Among examples in later verse are the following from the Old English poem Elene:
Hēo wǣron stearce, stāne heardran
They were strong harder than stone (Elene 505) and
sunnan lēohtra brighter than the sun (Elene 565)
Examples are also found in prose texts, but for the most part these no longer have OV order. Moreover, the use of the dative preceding the standard was gradually lost. Small (1929:83) concluded that it was no longer used after the second half of the tenth century, citing the translation of Latin fortior nobis in Exodus as strengre þonne we.
Another OV pattern in the early texts is the use of postpositions rather than prepositions. Examples may be cited from verse, such as:
Nástrǫndo á, norðr horfa dyrr;
at Nastrond, the door facing north; (Vǫluspá 38.3)
Scyldes eafore Scedelandum in
Scyld's offspring in the Scandinavian lands (Beowulf 19)
frēawine folca Frēslondum on
lord of the people in Friesland (Beowulf 2357)
eorlas on elne; ic him æfter sceal.
the warriors to courage; I shall follow them. (Beowulf 2816)
Instances of the use of postpositions with nouns may be considered residues, for even in Beowulf prepositions are the most common adpositions, as in the first half-line of Beowulf line 2816. They are standard in Old English prose texts, such as The Voyages of Ohthere and Wulfstan,
Hē cwæð þæt hē būde on þǣm londe norþweardum wiþ þā Westsǣ.
He said that he lived on the land to the north facing the West Sea.
And the comparative construction found in this text is comparable to that in use today, for example:
Sēo is brādre þonne ǣnig man ofer sēon mæge.
'It is broader than any can see across.'
A further OV pattern places titles after the name, as in Beowulf 2430:
Hēold mec ond hæfde Hrēðel cyning,
King Hrethel protected and kept me,
Similarly, in the 'Wars of Alfred the Great' he is referred to as Ǽlfrēd cyning.
Among Old Norse examples is the following from the Hrolf saga:
Kømr nú þessi fregn fyrir Hrólf konung...
This information comes now to King Hrolf...
This use also with titles is clear from the position of 'priest' in the first sentence of Ari's Libellus Islandorum:
Íslendingabók gørða ek first biskupum várum Þorláki ok Katli,
ok sýnda ek bæði þeim ok Sæmundi presti.
I first prepared the Iceland book for our bishops Thorlak and Ketill,
and I showed it both to them and to the priest Saemund.
The order may be the basis of the use of characteristic designations with notables, as in Ari's statement on the year 870 A.D.:
Ísland bygðisk fyrst ór Norvegi á dǫgum Haralds ins Hárfagra,
Hálfdanarsonar ins Svarta...
Iceland was first colonized from Norway in the days of Harold the Fair-haired,
the son of Halfdan the Black.
Equational sentences, in which the verb is the copula, also had final order of the verb in Proto-Germanic, as in an inscription on the Kragehul lance shaft:
ek erilaz asugisalas em
I erilaz of-Ansugisalaz I-am
I am the erilaz of Ansugisalaz.
But this position for the copula may not have been maintained for long in Proto-Germanic; as in other early documents, the copula in Beowulf is positioned like other verbs.
There is less evidence for retention of OV structure in the basic modifying patterns — relative clauses, adjectives, and genitives. In OV languages these precede the noun modified. Even in Beowulf and the other early texts, however, relative clauses consistently follow the noun they modify. But the use of participles in clauses preceding the noun modified suggests that such clauses may have continued the use of OV relative clauses in Proto-Germanic. Examples are given in section 5.3.1 below.
Moreover, particles, with or without a demonstrative pronoun that is in accordance with the noun in gender and number and with the other members of the clause in case, may have been a further development to the construction that was later introduced by a relative pronoun. These differ among the dialects, providing a further indication that the postposed pattern of relative clauses was a late development, e.g. Gothic ei, Old English þē/þe, Old Norse er and es.
The following Gothic passage illustrates the use of ei as a relative marker, as a conjunction, and as a relative marker with a demonstrative pronoun:
... und þana dag ei waírþái þata, duþē ei ni galáubidēs waúrdam meináim, þōei usfulljanda in mēla seinamma. (Luke 1:20)
... until the day that these things shall be fulfilled, because thou believest not my words which shall be fulfilled in their season.
The Old Norse particles es, later er, and then sem fulfill similar functions, as indicated in the following passages from Ari's Libellus Islandorum:
Eiríkr inn Rauði hét maðr Breiðfirzkr, er fór út heðan þangat ok nam þar land er síðan er kallaðr Eíríksfjǫrðr.
Eric the Red was the name of a man in Broadfirth, who went out from here to there (Greenland) and took up land there that has since been called Eric's Inlet.
In Old English the particle þe, þē, also spelled ðe, was widely used for all relative clause patterns, among them supplementing demonstratives, and also conjunction, as in the following examples from Beowulf:
worolde wilna, þē ic geweald hæbbe. (950)
of joys in the world over which I have control.
sē þe ēow wēlhwylcra wilna dohte. (1344)
(he) who provided for you all favors.
In Gothic the demonstrative pronouns sa, sō, þata plus the particle ei were the basis of the relative pronouns saei, sōei, þatei with reference to third person nouns, presumably to provide a more specific reference than that of the simple particle. The ei particle was also used with first and second person pronouns e.g. ikei '(I) who', þūei '(thou) who', as in 1 Corinthians 15.9:
ik áuk im sa smalista apaústaúle ikei ni im waírþs ...
I actually am the least of the apostles, who am not worthy...
In Old Norse the particle was not combined with the pronoun, as in the following example from Ari's Libellus Islandorum:
Þá váru hér menn Kristnir þeir er Norðmenn kalla papa.
Then there were Christian men here whom the Norwegians call papa.
The particle then came to be omitted. Already in Beowulf forms of the demonstrative pronoun sē, sēo, þæt were used without a particle to introduce relative clauses, as in line 453:
beaduscrūda betst, þæt mīne brēost wereð,
the best of military garments, that protects my breast,
The development is the same in Old High German, as illustrated by Hildebrandslied 15b:
(liuti) dea erhina warun
people who were earlier
Direct evidence of the earlier OV pattern is also given by limiting adjectives with weak inflection that precede nouns without definite articles. Seventy-five are found in Beowulf, e.g. gomela Scylding 'the old Scylding' [= 'Swede'] (2487; cf. Klaeber 1950:xcii). Weak adjectives are n-stems; these indicated a specific person or entity in Proto-Indo-European, as also in Latin where they gave rise to proper names, e.g. Varrō 'Varro', in contrast with the descriptive adjective o-stem vārus 'knock-kneed'. Line 1859 of Beowulf illustrates their use to indicate a specific entity: þendan ic wealde wīdan rīces 'as long as I rule over this wide kingdom'. They were not maintained in English, but they have made up a separate declension in German, where they alone no longer provide their original meaning but follow elements that do, such as definite articles.
Other adjectives, as well, precede nouns in Old English, as they do in Modern English, in contrast with a stricter VO language like French with regularly postposed adjectives:
Þæt wæs gōd cyning! He was a good king. (Beowulf 11)
Similarly in Old High German, as in the Hildebrandslied:
fohem wortum in few words (10)
friuntlaos man friendless man (24)
Many genitives also precede nouns in Old English and Old High German in accordance with OV patterning, such as:
þēodcyninga þrym the glory of the kings of people Beowulf 2
wuldres Wealdend the Lord of glory Beowulf 17
Scyldes eafera a descendant of Scyld Beowulf 19
Comparable examples are attested in the early texts of the other dialects, such as the Old High German Hildebrandslied:
Heribrantes sunu the son of Heribrant 7
Otachres nid the hatred of Otacher 18
degano filu a large number of warriors 19
And similarly in Old Norse, as in the Vǫluspá (40:7-8)
tungls tiúgari í trollz hami
of-sun destroyer in troll's appearance
destroyer of the sun in the appearance of a troll.
The Runic inscriptions provide few examples for the OV patterning of nominal modifier constructions, in part because of their conventions as in the use of apposition, in part because of their brevity, which among other features leave little possibility for relative clauses. The late Eggja inscription, cir. 700 A.D., includes a descriptive adjective before a noun in the phrase viltiz mænnz 'wild men'. And the Stentoften and Björketorp inscriptions, cir. 675 A.D., end with a verb-final clause saR þat barutR 'who breaks this'. Accordingly, the modifying constructions as well as the order of objects with regard to verbs indicate that the Proto-Germanic sentence structure was verb-final.
On the other hand, the basic structure of sentences may be modified to emphasize elements, especially in poetry, as in the following examples where æfter as well as mit geru and man are placed for the purpose of emphasis.
Ðǣm eafera wæs æfter cenned Beowulf 12
To-them son was after born
A son was then born to them.
mit geru scal man geba infahan Hildebrandslied 37
with spear should one gift receive
One should receive a gift with the spear.
The Old Icelandic example in section 5.1.6b provides another example, with verb and adverb preceding the subject. And Ari begins his work with the statement:
Íslendingabók gørða ek fyrst biskupum várum Þorláki ok Katli...
The Iceland Book I prepared first under our bishops Thorlak and Ketill...
Imperatives may be considered marked patterns, and typically they have the verb in first place, as in modern English, e.g.
Gesaga him ēac wordum, þæt hīe sint wilcuman (Beowulf 388)
Tell them also in words that they are welcome.
Forgip mir in dino ganada rehta galaupa. (OHG Wessobrunner Gebet)
Give me in your grace correct belief!
Like marked constructions in general, these do not confute the conclusion on the basic OV order, but they may be expected for highlighting relevant passages.
Questions without interrogative marker have the verb in first position, as in the Old High German baptismal vows:
Do you forsake the devil?
Questions of this order may include the enclitic -u, as in Gothic:
Maguts-u drigkan stikl...? Can you drink the cup...? John 19:39
When questions include an interrogative pronoun or adverb, it is initial, both in direct and indirect questions, as in the following examples:
Hvat's þat drauma? What sort among dreams is that? Eiríksmál 1
Hwanon ferigeað gē fǣtte scyldas Beowulf 333
Whence do you bear ornamented shields?
her fragen gistuont He began to ask
fohem wortun, hwer sîn fater wari in few words who his father was,
fireo in folche among the people of men Hildebrandslied 8b-12a
We assume that these patterns had been maintained from Proto-Germanic. Except for the absence of enclitic particles like -u, they were also continued in the later dialects.
Subordinate clauses may be distinguished as those that modify nouns, i.e. relative clauses, and those that complement verbs, either as objects or as adverbial clauses. Both reflect the OV structure of the earlier language, so that they may be reconstructed in Proto-Germanic. They also illustrate the various types of compound sentences in the language.
In an extensive article, Windisch (1869) concluded that, in Proto-Indo-European, relative clauses were introduced by the particle yo. His conclusion was widely disputed in view of the various means for introducing relative clauses in the dialects and also in lack of treatment of Proto-Indo-European as an OV language. But when Hittite was read, the form of relative clauses or, more precisely, their earlier form, became clear. They arose from a sequence of clauses in which the noun or pronoun modified was indicated by a marker in an initial clause; this clause was followed by the principal clause that often included an anaphoric pronoun referring to the item modified. The Old Latin inscriptions provide examples, such as:
Quei ager commutatus est, de eo agro siremps lex esto
some field changed is from that field exempt law is-to-be
The law will not apply to a field that has been changed.
The Old High German sequence in Otfrid 2.13.9 provides an example for Germanic:
ther brut habet, ther scal ther brutigomo sin
that-one bride has, that-one shall the bridegroom be
Who has the bride will be the bridegroom.
When the VO pattern became dominant, the principal clause occupied first position, and the earlier preposed element was placed after its antecedent, as in:
That one shall be the bridegroom who has the bride.
The modification of relative clause structure took place in the different Indo-European dialects or dialect groups, as indicated by the different relative particles among them. Some of these were adapted from forms of *yo-, such as Sanskrit yá- and Greek hó, others from interrogative markers, such as Latin quis, quid, and still others from demonstrative markers, such as Old English sē, sēo, þæt, e.g. Beowulf 1267b with sē referring to Grendel:
heorowearh hetelīc, sē æt Heorote fand
a hateful outcast who found at Heorot...
As here, the relative clause follows its antecedent in the later dialects, and its verb typically stands in final position. The comparable relative pronoun is also so used in Old High German, as in the Hildebrandslied 16 with reference to the plural liuti 'people':
alte enti frote, dea erhina warun
old and wise, who were earlier.
In Gothic the relative clause was generally introduced by the particle ei, as in Luke 1:20:
und þana dag ei wairþai þata
until that day that this will come to pass
The source of ei has been disputed. One origin that has been proposed is as a reflex of a form of PIE yo, like Slavic ijo, cf. 4.7; it then would be a direct continuation of the yo marker proposed by Windisch. Its variant þei is used as relative marker after indefinite pronouns, as in Mark 6:22:
bidei mik þishʷizuh þei wileis
ask me whatever (that) you wish
In Old Norse the particle es, later er and sem, had a similar function, as in the sentence from Leif's Voyage:
Hann mælti þá á Norrœnu, er stund leið...
He said then in Norse, (which) as the time passed...
Note also som in the following sentence from Hrafnkels saga:
sá maðr, som fyrir gekk, heilsar þeim fyrri ok spyrr, hverir þeir væri.
The man who went ahead greets them first and asks who they might be.
The different particles provide evidence that relative clauses and other subordinate clauses as well had only a general marker in Proto-Germanic, as noted in section 5.1.6a above. In view of the differences it is difficult to decide on the most likely element, but because of the Indo-European use of *yo, an earlier form of Gothic ei is most likely.
Although these examples illustrate the various structures of relative clauses in the later languages, reflexes of an alternative pattern that may have existed earlier are attested in Beowulf as well as in the Gothic translation of the Bible:
Hwīlum flītende fealwe strǣte
at-times contending on-yellow street
with horses traversed. Beowulf 916-917a
They, who competed from time to time, proceeded
down the sandy road with their horses.
The same pattern is found in Gothic, with a participle corresponding to the finite verb in Greek, as in Luke 7:44:
eisêlthón sou eis tḕn oikían
I entered into your house...
Atgaggandin in gard þeinana wato mis ana fotuns meinans ni gaft
To me entering into your house you did not give water for my feet.
Many other such preposed participles are found in the Gothic text; although they often correspond to the same construction in the original Greek, nevertheless they provide evidence that the construction was not entirely alien to Proto-Germanic.
The Gothic particle ei is used not only for introducing relative clauses but also for object clauses and adverbial clauses, as in Matthew 5:17 where the clause is object of hugjáiþ:
Ni hugjáiþ ei qēmjáu gataíran witōþ...
Do not think that I have come to destroy the law...
In Luke 2:3 it introduces a result clause:
Jah iddjēdun allái, ei melidái wēseina
And they all went, so that they might be registered.
It is also used to introduce conditional clauses, as in Mark 9:42 where it corresponds to the Greek conditional conjunction ei:
Gōþ ist imma máis ei galaígjáidáu asiluqaírnus ana halsaggan is
It is better for him if a millstone were laid on his neck.
These and other instances indicate that in Gothic and possibly in other dialects, a general particle was used for introducing subordinate clauses.
The Old Norse particle es, with voiced corresponding form er, had similarly wide uses, as in the following sentence re: the introduction of Christianity into Iceland.
Hann sende hingat til lands prest, þann es hét Þangbrandr ok hér kende
mǫnnom kristne ok skírþe þá alla, es viþ trú toko.
He sent here to the country a priest, whose name was Thangbrand, and he taught
the people Christianity and baptized all of them who accepted the faith.
In time modifications were introduced to provide conjunctions that indicated more specifically the relationships between the principal clause and the modifying clause. The most frequent such compound conjunction in Gothic was þatei after verbs indicating thought, belief and the like, as in John 11:13, although the simple ei was also used after such verbs:
Iþ jáinái hugidēdun þatei is bi slēp qēþi.
But they thought that he had spoken about sleep.
Similarly, Matthew 9:28:
Ga-u-láubjats þatei magjáu þata táujan?
Do you believe that I can do that?
Other such conjunctions in Gothic are based on prepositions and adverbs, for example fáurþizei 'before', miþþanei 'while, as', sunsei 'as soon as', swāei 'so that', þadei 'whereto', þandei 'because', þanei 'that', þarei 'where', þaþrōei 'from where', þēei 'through that'. As the meanings of the compound conjunctions indicate, the conjunctions and adverbs were added to ei for greater precision.
As a further development, the ei component may have seemed unnecessary. It was often omitted in Gothic, as were similar particles in the later dialects.
jah warþ, þan ustáuh Iesus þō waúrda
and became when finished Jesus these words
And the effect was, when Jesus finished these words... Matthew 7:28
Such adverbial clauses in the later language often include subjunctive forms of the verb as well as the introductory conjunction, as in Beowulf 452:
Onsend Higelāce, gif mec hild nime...
Send off to Higelac, if the battle should take me...
The development of adverbial clauses in the Germanic languages, then, is clear. In early Proto-Germanic there was no subordination. Later, particles like Gothic ei were added to indicate relationships between clauses; such particles were then combined with adverbs and prepositions that made the relationship more specific. Later yet, the appended particle was omitted and the conjunction consisted entirely of the element that indicated the specific relationship of the subordinate clause to the principal clause, which might be further specified by use of the subjunctive mood.
Negation is expressed by means of the particle ne, as also in Proto-Indo-European and some of the other dialects. It may be found in its full form with reference to an entire clause where ne is commonly placed before the verb, as in:
Hē bēot ne ālēh, He didn't fail to perform the promise, Beowulf 80
When used with nominal and adverbial elements it is prefixed and generally unaccented, accordingly in its zero grade form, PGmc un-, as in the adjective Go. unkunþs, ON ūkūðr, OE uncūþ, NE uncouth, and similarly in nouns, Go. unkunþi 'lack of knowledge', Go. unhráinei 'uncleanliness' and numerous others throughout the dialects; cf. also Gothic ni hwas and ni áins 'no one'. From the second of these combinations, English none and German nein developed. Under primary stress un- maintained this form in the West Germanic languages, but in Icelandic it became ó- as in ó-vitr 'ignorant'; cf. ON ólīkr, NE unlike, and many other such forms.
When negating verbs or clauses, it has not survived in the modern dialects. As is widely found with negatives, words were often added to strengthen the negation; the strengthening element is commonly a noun, as in Old English and Old High German wiht 'being, thing', but a form of the negative itself may also be added:
Nē hīe hūru winedrihten wiht ne lōgon, Beowulf 862
They did not blame their lord a whit,
The cognate of wiht is also used in this way with the negative, in other dialects, e.g. Go. ni waíht, OHG niwiht, OE nā-with, as in the Gothic clause from Matthew 10:26:
Ni waíht áuk ist gahuliþ...
For not a thing is hidden...
The two words have been conflated in OHG niht, NHG nicht, NE naught, not. In Old Norse, on the other hand, the noun by itself has come to mean 'nothing', as in:
át vætr Freyja átta nóttum Þrimskviða 108
Freyja ate nothing for eight nights
Because the combination or its result in Old Norse is found in all the dialects, it and other such combinations, as illustrated below, may have been used already in Proto-Germanic.
Another such strengthening element is Gothic áiw 'length of time', as in:
ni áiw swā unkunþ was in Israela. Matthew 9:33
(at) no time so unknown was in Israel
Never had such a thing been seen in Israel.
The corresponding combination in German has given rise to nie 'never'. Cf. also Old English nealles, as in:
Nealles mid gewealdum wyrmhord ābræc Beowulf 2221
Not at all of his own accord did he break into the dragon's hoard.
A postposed enclitic used to support negation is also attested in several Germanic dialects, e.g. Gothic -hun:
jah ni fralaílōt áinōhun izē miþ sis afargaggan. Mark 5:37
and not he-let any-one of-them with himself follow
and he did not let any of those with him follow him
The form was voiced in accordance with Verner's law in Old Norse -gi, and in Old English -gen, Old High German -gin, as in Old Norse hvergi, Old English hvergen, Old High German hwergin 'nowhere, not at all'. Still other clausal and verbal negatives are used in the early texts, but they must be left to the grammars of the dialects.
Forms of the Indo-European middle inflections survived in Gothic in the present indicative and subjunctive, but with passive meaning, as in:
Aþþan atgaggand dagōs þan afnimada af im sa brūþfaþs. Mark 2:20
But the days will come when the bridegroom is taken from them.
Jah jabái gards wiþra sik gadáiljada... Mark 3:25
And if a house is divided against itself...
jah þu, barnilō, praúfētus háuhistins háitaza... Luke 1:76
and you, child, shall be called a prophet of the Highest...
uha haite (Old Norse, Kragehul spearshaft)
I am called Ūha — My name is Ūha
Oddly, the active forms of ON heita have taken on the middle meaning in the other dialects, but the middle form in Old English hǽtte has not been maintained.
Ingolfr hét maðr nórœnn. Íslendingasaga 1
Ingolf was the name of a northern man.
Similarly in Modern German, for example: Er heisst Wilhelm 'His name is William' from 'He is called William'. We therefore assume the form for Proto-Germanic and also conclude that the middle survived into Proto-Germanic.
The shift in use of the middle to the passive, and the increased use of the passive in the Germanic dialects as well as in the other Indo-European dialects, is in accordance with the shift of late Proto-Indo-European as well as the dialects from active structure to transitive structure. Some residues of the earlier structure are maintained in Gothic, as in the intransitive use of transitive verbs like dáupjan 'baptize, wash' and bimáitan 'circumcise'. Streitberg (1920:191-192) cites examples of such use in translations of Greek middle forms, e.g. Mark 7:4 niba daupjand for the Greek middle form baptísōntai 'unless they wash [themselves]', an alternate form in the passage for hrantísōntai 'they sprinkle themselves'.
Another residue of the earlier contrast is the distinction in Gothic between perfective forms with ga-prefix in contrast with the imperfective use of the simple verb as treated by Streitberg (1920:196-198), where he cites among examples the passage on the death of Lazarus in John 11:11: Lazarus ...gasaízlēp 'Lazarus has fallen asleep / is sleeping' for the Greek with its middle form: Lázarus ...kekoímētai.
Instead of the forms with ga-prefix, the other dialects make use of reflexive pronouns, as in the following Old High German clause:
gurtun sih iro suert ana Hildebrandslied 5
they girded (for themselves) their swords on
In Old Norse the combination of reflexive pronoun and verb has given rise to a new inflection, with -mk suffixed in the first person singular and -sk in all the other persons. The forms have various uses, from reflexive and reciprocal to middle and passive, as in the following examples:
Often the forms in -sk correspond to simple verbs in English, for example:
Þá sættusk þeir á þat... İslendingabók 1
Then they agreed on that...
ok hafði allt farizk vel at... İslendigabók 5
and everything had fared well...
Passive meaning was expressed chiefly by compound forms consisting of auxiliaries accompanying the past participle. The auxiliaries vary. In Old Norse the auxiliary was vera, as in the following sentence from the section in Libellus Islandorum on the introduction of Christianity:
Þá var þat mælt... Then it was announced...
And a few lines later in the present tense:
es kallaþr es Vellankatla which is called Vellankatla
Passive forms with the auxiliary werden that is currently used in German to form the passive are attested in the early Old High German works like Muspilli, as in the subjunctive form in line 12:
ze uuederemo herie si gihalot werde
to which group it (the soul) will be taken
Also in the corresponding indicative present form in line 39 with future meaning:
denne uuirdit untar in uuic arhapan.
then a battle will be raised among them
The corresponding auxiliary is also used in Old English to make the passive with transitive verbs, as in Beowulf 6-7:
egsode eorl[as] syððan ǣrest wearð
he terrified warriors since he first was
But a few lines later in line 12 the auxiliary wesan was used, as in later English. It is used with both intransitive and transitive verbs:
Ðǣm eafera wæs æfter cenned
To them an heir was later born
In view of the several auxiliaries in the dialects, we may assume that the compound passive forms were developed after Proto-Germanic. In the proto-language both passive and middle meanings were expressed by reflexes of Proto-Indo-European middle forms. But the forms were maintained only in Gothic, and as shown in section 3.8, they were replaced by compound forms in the other dialects.
The two Proto-Germanic indicative tenses, present and preterite, primarily expressed present and past time, as in citations provided earlier such as Vǫluspá 64: Sal ser hon... 'He sees a hall'; and the Gallehus inscription ...tawido '(I) made'. When subjunctive forms are used in accordance with clause structure of subordinate clauses, they may also express present and past action, as in Beowulf 452 cited above: gif mec hild nime 'if the battle should take me' as a general statement rather than expression of a future action, and in the Gothic Mark 5:17 with its expression of past time: Ni hugjeiþ ei qemjau gataíran witōþ... 'Do not think that I came to destroy the law...'
The present tense is also used to express future time, as in Gothic, Mark 2:20: aþþan atgaggand dagōs... 'but the days will come...'
While tense is the primary meaning expressed by inflections of verbs, residues of aspect are also attested, though chiefly with verb forms prefixed with ga-. For example, the form of feallan in Beowulf 2100 clearly indicates completed action of the wounded Grendel:
mōdes geōmor meregrund gefēoll
sorrowful in mind he fell to the bottom of the sea
By contrast, the unprefixed forms in line 772 express continued action:
Þā wæs wundor micel, þæt se wīnsele
wiðhæfde heaþodēorum, þæt hē on hrūsan ne fēol,
That was a great wonder, that the winehall
withstood the brave battlers, that it was not falling to the ground,
The contrast is evident in other verbs as well, such as bīdan 'wait, stay, remain' versus ge-bīdan 'live to see, experience'.
The contrast is also attested in Gothic, but less frequently; cf. háusjan 'hear', gaháusjan 'perceive'; saíhʷan 'see', gasaíhʷan 'catch sight of', as in Matthew 11:4:
Gaggandans, gateiheiþ Iohannē þatei gaháuseiþ jah gasáihʷiþ.
Going, tell John what you perceive and notice.
In High German the use of the prefix leads to more extensive contrasts, as in Middle High German bern 'carry', gebern 'give birth', and also further in present-day German bieten 'offer', gebieten 'command', fallen 'fall', gefallen 'fall to someone's share, please'. While past participles indicate completed action in Old Norse, they are not prefixed by an equivalent to ga-, nor is the prefix used before other forms of verbs.
We may conclude that the distinction between imperfective and perfective aspect was maintained into Proto-Germanic, and was expressed by means of the ga- prefix, but that it was primarily expressed in the past participle, also when this was not prefixed as in Old Norse.
Uncertainty was expressed in the early dialects by verb forms in the subjunctive, and modality by some of the modal auxiliaries, which however are wider in scope, as shown below.
As noted above, the Germanic subjunctive is formally a reflex of the Indo-European optative; reflexes of the Indo-European subjunctive forms are not attested. A typical means of expressing uncertainty is found in indirect questions, with the subjunctive form wāri in the Hildebrandslied 8b, also cited in section 5.2.
Her fragen gistuont
fōhem wortum, hwer sin fater wāri
He began to ask
with few words, who his father might be
The subjunctive is also widely used in other subordinate clauses when indication of greater uncertainty is intended, as Heusler (1920:131) points out for Old Norse in the contrast between the indicative vill finna hann 'wants to speak with him' and the subjunctive vile finna hann 'would like to speak with him'.
Conjunctions introducing subordinate clauses developed independently in the different dialects, as their forms indicate; jabai 'if' introduces conditional clauses in Gothic, ef in Old Norse, gif in Old English, wenn in Old and New High German. We may then further assume that, in Proto-Germanic itself, the conditional clause did not need an initial conjunction, as is still possible in the current languages, e.g. 'Should your right eye offend you...' Its initial position may be supported by examples in Gothic, e.g. Matthew 5:29:
Jabai augo þein þata taihswo marzjai þuk, ustagg ita
If your eye, the right one, offends you, pluck it out.
But in later prose texts the conditional clause frequently follows the main clause, as in this sentence from Ari's Libellus Islandorum:
Sagði at honum þótti þá komit hag manna í ónýtt efni, ef menn skyldi eigi
hafa allir lǫg ein á landi hér.
He said that it seemed to him that the affairs of men would come to an evil plight,
if all men would not have one law in the country here.
Shorter expressions may also be inserted in the middle of sentences, such as the formula used twice in Beowulf:
gif mec hild nime 452, 1481
if the battle should take me
Constructions in which uncertainly is expressed by subjunctive forms include that-clauses, for which different conjunctions are found in the various dialects. In Ari the missionary's account, Thangbrand uses the preterite subjunctive of the auxiliary mono 'become' in his report to Olaf Tryggvason concerning the possibility of success in introducing Christianity to Iceland:
lét ørvǽnt, at hér myndi Kristni enn takask.
left it hopeless that Christianity might ever be accepted here.
Modal auxiliaries, like the form of mono used here, are reflexes of roots that are treated as full verbs in other Indo-European dialects; cf. Go. wiljan 'want, will', with cognates in the other Germanic dialects, based on the Indo-European root *wel-, from which regular verbs are formed in other dialects, such as Skt vṛnīté 'chooses', Lat. velle 'wish'. They filled the role of the Indo-European subjunctive, which did not survive in Germanic. It is difficult to account for the process of replacement, given the absence of continuous texts. We can simply state that they were treated like auxiliaries, as exemplified in the sections below.
Various uses of modal auxiliaries are cited below, with their meanings.
Among the modalities, possibility is expressed by the auxiliaries magan and kunnan, as in Bede's account of the poet Caedmon's response to the man who appeared to him in a dream and asked him to sing something:
Ne con ic nōht singan. I can't sing anything.
But the man replies:
Hwæðre þū meaht mē singan. But you may sing to me.
A comparable use in the preterite indicative may be cited from The Voyages of Ohthere and Wulfstan, where Ohthere tells King Alfred:
Þā fōr hē þā gīet norþrihte swā feor swā hē meahte...
Then he traveled further northward as far as he might...
A use of the preterite subjunctive of Old Norse mono was cited above; another occurs somewhat later in the account, when Gizarr contradicts Thangbrand about the possibilities for Christianity in Iceland, saying that there is no other expectation...
an þar myndi hlýða.
than that it would succeed there.
Forms of willan are used to express wishes or desires, as in the Hildebrandslied 40b:
wili mih mit dinu speru werpan
you want to strike me with your spear.
And similarly in the often quoted couplet from the Old Icelandic account of the introduction of Christianity:
vilk at goð geyia, I don't want to bark at a god,
grey þykkjumk Freyia. (but) Freyja seems to me to be a bitch.
The preterite has the same meaning, as in an example from The Voyages of Ohthere and Wulfstan:
Hē sæde þæt hē æt sumum cirre wolde fandian...
He said that at some time he wanted to try to find out...
Obligation or the somewhat stronger notion of necessity is expressed by the auxiliary skulan, as in the question of Caedmon:
Hwæt sceal ic singan? What should I sing?
Numerous examples are found in the account of Christianity coming to Iceland, such as
aller menn skylde kristner vesa
all men should be Christians
Similarly, in the Hildebrandslied 27, Hadubrand says:
mit geru scal man geba infahan
With a spear one should receive gifts
Causation is expressed by the use of lātan as auxiliary, as in this sentence from the Heimskringla dealing with Haakon the Good:
Aþalstein konungr lét skíra Hókon ok kenna rétta trū ...
King Adalstein had Haakon baptized and taught the correct faith...
The same pattern with an infinitive is found in the Hildebrandslied 63:
do lettun se ærist asckim scritan
then they first caused their lances to glide forth
The imperative generally indicates commands, as in the account of the poet Caedmon:
Berað mē hūsl tō. Bring me communion.
Similarly, in the Old High German Wessobrunner Gebet:
Forgip mir in dino ganada rehta galaupa Give me in your grace correct belief
And in the Old Norse story of Auþun and his polar bear:
kom þā til mīn come then to me
Other modal auxiliaries have been listed in section 3.8 among the preterite-presents. Syntactically their uses are comparable to those of the modals exemplified here.
Adverbs are very prominent in the early texts. Examples from poetry illustrate that prominence. They often fill the third accent in the poetic line, as in Beowulf 31:
lēof landfruma lange āhte.
The beloved king had long ruled.
Formed with various suffixes, they referred to time, place and manner. Those with a nasal suffix indicated action away from a point, as in Beowulf 91:
Sægde sē þe cūþe
frumsceaft fīra feorran reccan,
He said, he who could
tell the origin of men from far in the past,
Those with an -r- suffix often indicated action to a point, as in Beowulf 370:
sē þǣm heaðorincum hider wīsade.
who those warriors directed hither.
Adverbs of manner were based on various suffixes, some with u-stems, as is still clear in OE gearwe, as in Beowulf 265:
gamol of geardum; hine gearwe geman
old in years him readly recalls
Klaeber pointed to a notable feature in the use of adverbs of place (1950: xciii). They often provide "instructive instances of the characteristic fact that in the old Germanic languages the vivid idea of 'motion' was predominant in many verbs which are now more commonly felt to be verbs of 'rest'." One of the passages he cites is Beowulf 1805b and 1806:
wolde feor þanon
he wished far from there
cuma colenferhð cēoles nēosan.
visitor bold ship seek
the bold visitor
wished to fare afar in his faithful ship, (R.P.M. Lehmann, Beowulf 1988:69).
In other passages, nēosan simply means seek. Here the adverb þanon clearly prompts an interpretation of activity. Another example involves the verb 'to shine', indicating the sun's activity in directing its rays from the south rather than in the south.
sunne sweglwered sūþan scīneð! Beowulf 606
the sun, clothed in radiance, shines from the south!
Examples from Old Norse illustrate similar prominence, though those given here from Ari's Libellus Islandorum accompany verbs indicating activity.
Ingólfr hét maðr Norrœnn, er sannliga er sagt at fœri first þaðan til Íslands...
Ingolf was the name of the Norwegian man, who accurately is said to have traveled first from there to Iceland...
Eiríkr ... hét maðr, er fór út heðan þangat...
Eric was the name of a man who traveled out from here to there...
En svá er sagt, at þat bæri frá, hvé vel þeir mæltu.
And thus it is said, that that was wondrous, how well they spoke.
The various adverbs need little comment, but it is clear that they are introduced to make statements more explicit, such as sannliga in the first example above, and even more so svá in the last. Similarly the adverbs indicating movement! And the adverb meaning 'from, concerning' with the verb bera 'to bear' has led to an idiom. It is also clear that adverbs like út and frá may be comparable to particles.
We may ascribe both the prominence of adverbs and their promotion of activity with verbs indicating place to their position at the Active-Stative stage of the proto-language. During that stage they were particles, or like OE súþan they consisted of suffixed nominals that qualified verbs. Relationships in the sentence were determined by agreement rather than transitivity so that relationship of nominals as well as particles to verbs resulted from semantic correspondence. As Proto-Germanic became an accusative language, the nominals either were interpreted as nouns with inflections or as adverbs. In the course of time the adverbs became less prominent and the basic elements of the predicate were nouns in various roles.
The early texts, such as the runes, indicate clearly that the order of Proto-Germanic sentences was OV, as also in Proto-Indo-European. Moreover it was highly paratactic, as the first two lines of the Eggja stone illustrate; the stone is dated around 700 A.D., indicating that parataxis was long in force. It is given here in conventional Old Norse notation, with a gap indicated in parentheses. The two last lines have gaps so that they consist of phrases rather than sentences; the forms are modified in accordance with standard Old Norse (Krause 1966:227-235):
Ni's sólo sótt ok ni saxe stæin skorenn; ni (læggi) manna nækðan, niþ rinnR ni viltiz mænnz læggi a[b].
It is not struck by the sun nor is the stone scored with a knife. No one should lay it out uncovered, when the moon wanders (over the sky) nor should wild men lay [it aside].
Other early texts, as late as Ari's Libellus Islandorum of approximately 1120-1133 A.D., are also largely paratactic.
But modifications were introduced, as by relative clauses. In the Tune stone of around 500 A.D., apposition was used to qualify a noun as in many later Germanic texts; the text is given in Old Norse as by Krause (1966:162-167):
þrijoz ðohtriz ðalidun arbija arjostez arbijano
Three daughters shared the inheritance, the closest relatives of the heir
It would then have been a small matter to modify the apposition to a relative construction, such as those in Beowulf where the relative clauses follow their referent, an assumption supported by introductory particles (e.g. OE sē, Go. ei, ON es) that were associated with demonstrative pronouns as in Ari's sentence on the discovery of Greenland:
Land þat er kallat er Grœnland fannsk ok bygðisk af Íslandi.
The land that is called Greenland was found and colonized from Iceland.
Subsequently, specific relative pronouns were introduced, as were conjunctions. Auxiliaries became common, as in this quotation, also for expressing the passive and modalities. Such innovations are in keeping with the gradual shift from OV to VO structure. The dialects, therefore, came to differ considerably from Proto-Germanic in their sentence structure.