Version classiqueVersion mobile

Langues indo-européennes

 | 
Françoise Bader

Chapitre 13. Germanic (Le germanique)

Jay H. Jasanoff

Texte intégral

1. Introduction

1The Germanic languages of the present day occupy a large expanse of northern and central Europe, including Scandinavia and the British Isles; one language of the family, English, has spread in recent times to cover the greater part of North America, most of Australia, and smaller areas elsewhere. But like Romance, the other Indo-European language group with a worldwide distribution, Germanic started from insignificant beginnings in an area remote from the centers of European civilization. Early in the first millennium B. C., Proto-Germanic, the source of the modem languages, was spoken by a small population of Iron Age agriculturalists in southern Scandinavia and the adjacent Baltic and North Sea coasts. In the centuries that followed, Germanic tribes slowly spread to the south and southeast, largely displacing or absorbing the Continental Celts, who had occupied much of central and eastern Europe a half millennium earlier. Before the beginning of the Christian era, Germanic peoples stood across the Rhine from the fields of Roman Gaul, and intermittently menaced the prosperous Greek settlements on the shores of the Black Sea.

  • 1 Another descendant of Old Norwegian is Faroese, spoken by about 43,000 people in the Faroe Islands.

2The culture of the ancient Germans, at the moment of their first appearance in history, was distinctive and homogeneous. Germanic society was characterized by a shared religion, a shared legal and political system, a shared ars poetica and, of immediate interest to us here, a shared language. Interesting glimpses of this world are afforded by the Germania of Tacitus, written around the year 100 A. D. By that time, the Germanic dialects had been diverging for three or four centuries. The tribes which, starting from what is now western Poland, had moved toward the Danube basin and the steppes of southern Russia, constituted an incipient East Germanic linguistic group, which differed, only slightly at first, from the North Germanic of southern Scandinavia and the West Germanic of Jutland, Germany proper and the non-Celtic part of the Low Countries. The latter group was itself an aggregate of closely related dialects, of which modem scholars often recognize three: North Sea Germanic or Ingvaeonic, Rhine-Weser Germanic and Elbe Germanic, corresponding to tribal divisions recorded by Tacitus and other Roman authors. East Germanic, now extinct, gave rise to Gothic, the language of the Arian bishop Wulfila’s translation of the Bible; the substantial surviving fragments of this fourth century text constitute our earliest extensive remains of any Germanic language. A late form of Gothic is believed to persist in Crimean Gothic, the dialect of a short sixteenth-century word list. North Germanic is first attested in short inscriptions in the Runic alphabet (cf. PGmc. *rūnō « secret », several of which go back to the beginning of the third century or even earlier. The language of these meager texts is usually termed Primitive Norse (Urnordisch); until the sixth century it was more conservative than Gothic, and hardly distinguishable, in phonology at least, from Proto-Germanic itself. The earliest North Germanic literary languages, often collectively known as Old Norse, are Old Icelandic and Old Norwegian, together constituting West Norse, and Old Danish and Old Swedish, together constituting East Norse. All are attested from the twelfth century; Modem Icelandic, Norwegian, Danish and Swedish (from c. 1500) are their direct descendants1. The present-day languages of the Scandinavian mainland are structurally quite different from Old Norse, having participated for nearly a thousand years in the general western European drift toward morphological simplification. Icelandic, on the other hand, has changed but little from its medieval form, and is by far the most archaic Germanic language still spoken.

  • 2 Here too belongs Afrikaans, spoken in the Republic of South Africa by the descendants of the origin (...)

3The West Germanic languages present a more complicated picture. The Ingvaeonic group is best represented in the early medieval period by Old English, formerly called Anglo-Saxon, which was brought to Britain, apparently amidst scenes of great violence, by invaders from the Continent in the fifth and sixth centuries. Old English is attested from the eighth century; after c. 1100 it is usual to speak of Middle English, and after c.1450 of Modern or New English. Closely related to English is Frisian, the earliest documents of which, in Old Frisian, date from the thirteenth century. Old Saxon, attested from the early ninth century, and its descendants Middle and Modern Low German, are mixed dialects, with an Ingvaeonic core overlaid by elements from the neighboring non-Ingvaeonic speech of the Franks. The Franks, a tribal confederation first mentioned in the third century, were the principal bearers of Rhine-Weser Germanic. Their northernmost dialect, poorly attested at the end of the tenth century as Old Low Franconian, became the basis of Middle and Modern Dutch (from c. 1500)2. The remaining Frankish dialects, Middle Franconian, Rhenish Franconian, and East Franconian, collectively constitute the “Middle German” (mitteldeutsch) component of High German, or German proper. Old High German, documented from c. 750, is an assemblage of Middle German dialects and dialects of the “Upper German” (oberdeutschi) type, the latter consisting chiefly of Bavarian and Alemannic, representing the old Elbe Germanic group. Around 1050 Old High German reached the stage known as Middle High German, which in turn developed into Modern (or New) High German around 1350. Modem literary German is based primarily on East Franconian, the dialect of Martin Luther’s sixteenth-century translation of the Bible. Another descendant of Old High German is Judeo-German or Yiddish, with a history extending back to the eleventh century.

4The map of the Germanic languages is thus, from the beginning of the historical record, crisscrossed by dialect divisions divisions whose interpretation and significance have traditionally been a source of controversy. The old view that North and East Germanic are more closely related to each other than to West Germanic is no longer widely held. The special points of resemblance between Gothic and Old Norse are either common retentions, like the ending -t in the 2 sg. preterite (cf. Go., ON namt vs. OHG nāmi « you took », or structurally insignificant innovations, like the occlusion of *ww- and *-jj- to *-ggw- and *ddj-/*-ggj- (cf. Go. twaddje, ON tveggja vs. OHG zweiio « two [gen. pl.] »). More impressive are the innovations common to North and West Germanic, such as the change of *ē to (cf. ON Sáð OHG sát vs. Go. se] os « seed », the development of an s-extended demonstrative with the meaning « this » (cf. ON pessi, OHG desēr), and the creation of a seventh class of ablauting strong verbs (cf. ON heita, pret. hét, OHG heizan, pret. hiaz vs. Go. haitan, pret. haihait « call »). Another debate centers about the status of West Germanic itself, which some scholars have refused to recognize as a historically valid linguistic grouping. Such doubts are probably exaggerated. While Old English, Old Saxon and Old High German are far less alike than, say, Old Icelandic and Old Swedish, they show too many common developments (e.g., the change of *ð to *d, the gemination of consonants before *-j-, the loss of final *-z after unstressed vowels) for the assumption of a Common West Germanic period to be dispensed with entirely. West Germanic is best thought of as a cluster of closely related dialects, with the deepest isoglosses separating Ingvaeonic from the others.

5The contentious state of Germanic dialectology makes it easy to lose sight of the fact that the early Germanic languages are all very similar. The Germanic peoples of the early Middle Ages were probably in some sense aware of their kinship. Their traditional heroic literature knows no tribal bounds: the Old English epic Beowulf mentions the Gothic king Ermanaric, who lived in the Ukraine in the fourth century, while the Old High German Hildebrandslied evokes the fifth-century Italian wars of Odoacer and Theodoric. Thanks to the closeness of the languages and the richness of their attestation, scholars have been able to arrive at a remarkably detailed picture of Proto-Germanic, despite the absence of contemporary written sources. The technique that has made this possible is the comparative method, which has also been successfully applied to the reconstruction of Proto-Celtic, Proto-Italic, Proto-Greek, Proto-Indo-Iranian and Proto-Indo-European itself. It is probably safe to say that, despite some abiding problems, Germanic is one of the best understood branches of the IE family.

6What kind of language, then, was Proto-Germanic? Like Celtic and Italic, its neighbors to the south and west – but unlike Balto-Slavic, its neighbor to the east – Germanic belongs to the centum division of IE languages; that is, it prehistorically merged the PIE “palatal” sounds * *ĝ and *ĝh (phonetically perhaps /ky/, /gy/, /gyh/) with the velars *k, *g and *gh, keeping the resulting *k, *g and *gh distinct from the laviovelars *ku, *gu and *gh. In other respects, such as the merger of PIE *a and *o and the substitution of *m for *bh in the ending of the dative (< instrumental) plural, Germanic accords with Balto-Slavic. Phonetic and grammatical agreements of this kind are matched by lexical isoglosses: the Germanic words for « nourish » (OE alari) and « seek » (Go. sokjari) are shared only with Celtic and Italic (cf. OIr. alid, Lat. alere; OIr. saigid ( « makes for »), Lat. sāgīre), while the name of the aspen tree (OHG aspa) and the word for « bare » (OHG bar) recur only in Balto-Slavic (cf. OPr. abse, Russ, osina; Lith. bãsas, OCS bosũ). Such special correspondences, which can be multiplied, are normal between geographically contiguous related languages and in no way compromise the position of Germanic as an autonomous branch of Indo-European.

2. Phonology

2.1. Consonants

7On the phonological level, the most conspicuous development of the period between the breakup of Proto-Indo-European and the formation of Proto-Germanic was the Germanic Consonant Shift, often known, in honor of one of its co-discoverers, as Grimm’s Law. This consisted of three distinct changes:

  1. the PIE voiceless stops *p, *t, *k, *k*u, when not preceded by an obstruent, became the voiceless fricatives *ʄ, *] o, *x (> later *h in most environments), *χw (> later *hw);

  2. the PIE voiced stops *b (rare), *d, *g, *gu became the voiceless stops *p, *t, *k, *kw; and

  3. the PIE voiced aspirated stops *bh, *dh, *gh, *g*uh became the voiced fricatives *ƀ, *đ, *ǥ, *ǥw, which often gave voiced stops in the attested languages.

  • 3 An aberrant (dialectal?) development of to *b seems to underlie Eng. bid (Go. bidjan; PIE *g*uhedh-(...)

8Examples are very numerous: cf. Eng. foot (Go. fotus), three (Go. ] orija), horn (Go. haurn), what (Go. /υiata) beside Lat. pēs, trēs, cornu, quod; Eng. tooth (Go. tun] ous), corn (Go. kaurn), quick (Go. qius) beside Lat. dēns, grānum, uīuus (< *gwīwos); Eng. bite (Go. beitan), dare (Go. (ga)-daursan), goose (NHG Gans), warm (NHG warm, with w- < *gw-) beside Skt. bhid- « split », dhṛṣ- « be bold », háṃsa- (< *ĝh-) « (wild) goose », gharmd- (< *guh-) « hot drink »3. These changes were not restricted to word-initial position, but also took place word-intemally. Here, however, the voiceless fricatives produced by Grimm’s Law, together with the inherited voiceless sibilant *s, were under certain conditions subject to the later rule known as Verner’s Law. The effect of Verner’s Law was to convert *f *] o, *x, *xw and to the corresponding voiced fricatives *ƃ, *đ, *g, *gw and *z when the preceding vowel did not bear the pre-Germanic (= PIE) movable accent. Thus, the Germanic word for « father », which was originally end-stressed (cf. Skt. pitár-, Gk. patēr), became *fa] oēr by Grimm’s Law and *fađēr (> Go. fadar, OE fœder, OHG fater) by Verner’s Law; the originally root-accented word for « brother », on the other hand, retained its *–] o- < *-t- until the end of the Proto-Germanic period (cf. Go. bro] oar, OE brōðor, OHG bruoder; Skt. bhrātar-, Gk. phrātēr). The conditioning environment for Verner’s Law was lost when the pre-Germanic accentuation system, inherited from Proto-Indo-European, was replaced by the later Germanic system of fixed initial stress.

9Grimm’s and Verner’s Laws are justly celebrated beyond the narrow circle of Germanicists and Indo-Europeanists. The discovery of Grimm’s Law in the early nineteenth century furnished dramatic proof that sound change was a systematic process, thereby stimulating the search for other recurrent correspondence patterns and fostering the growth of the fledgling science of comparative philology. Later, in the 1870’s, the discovery of Verner’s Law helped to show that even the apparent exceptions to regular sound changes could be systematically explained – a recognition eventually enshrined in the famous principle that « sound laws have no exceptions ». Yet despite many years of careful, and sometimes even obsessive, study, it is remarkable how little is known for certain about the mechanism or chronology of Grimm’s and Verner’s Laws. According to a widely held view, the initial step in the Germanic consonant shift was the passage of the voiceless stops to voiceless aspirates (*p*ph, *tth, etc.), a change which left the voiced stops free to lose their voicing while the aspirated stops, both voiced and voiceless, developed further to fricatives. This is a plausible scenario, and it can be paralleled, at least typologically, by a similar shift in Armenian; but in the last analysis it is only a hypothesis. Other theories continue to command respect as well, including the view that the Germanic voiceless stops go back to glottalized *p’, *f, etc., rather than to voiced *b, *d, etc., in the parent language (the “glottalic theory”). As for the absolute chronology of these changes, all that can be said is that Grimm’s Law took place before the bulk of the dialectal Germanic borrowings from Latin, but later than the acquisition of such loanwords as Go. reiki, OHG rīhhi « kingdom » (< Celtic *rīgio-; cf. OIr. ríge « id. ») and Go. paida, OHG pfeita « cloak » (Gk. baítē « shepherd’s cloak », of foreign origin), which seem to have undergone the shift.

10The effect of Grimm’s and Verner’s Laws was to leave early Proto-Germanic with an obstruent system that was far richer in fricatives than in stops. This imbalance, “unnatural” from a typological point of view, was partly remedied by subsequent changes, the most important being the development of the voiced fricatives to stops in certain environments. By the end of the Proto-Germanic period, *ƃ, *đ, *g, *ǥw had become *b, *d, *g, *gw after nasals, and *b and *d had probably become *b and *d in word-initial position as well. Go. bindan « tie » (cf. Skt. bándhati < *bhándhati) was thus phonetically [bindan], although the contrast between [b], [d] and [ƃ], [đ] remained phonologically non-distinctive. Further fricative > stop rules were introduced in the individual Germanic dialects.

  • 4 A possible exception is to be seen in forms like OE naca, ON nǫkkvi “boat” (< PIE *neh2ṷ-ón-?) and (...)

11The remaining PIE consonants were the three “laryngeals” *h1, *h2, *h3, and the six sonorants *r, *l, *m, *n, *ḭ, *ṷ, the latter with syllabic allophones *, *, *, *ṇ, *i, *u. As in most other IE languages, the laryngeals were normally lost in Germanic4, although their earlier presence can be detected from such effects as compensatory lengthening and the “coloring” of adjacent vowels (cf. Go. ietan « let, release » < PGmc. *lētanan < *lēd- < PIE *leh1-d-; Go. bro] Oar < PGmc. *brō] Oēr < *bhrā- < PIE *bhrah2- < *bhreh2-). Among the sonorants, the liquids (*r, *l) and the nasals (*m, *n) were preserved unchanged in most positions; new instances of consonantal *r, *1, *m, *n were introduced by the change of syllabic *ṛ, *, *, *n to *ur, *ul, *um, *un (cf. Go. wulfs, OE wulf « wolf » < *ṷlk*uo-/*ṷḷpo-; Go., OE hund « hundred » < *k̂ṃtóm). The treatment of the glides and *u was more complicated. *ṷ was stably maintained as PGmc. *w, but *ḭ, which survived as *j in word-initial position and after consonants, was apparently lost after vowels other than *i (cf. Go. salbon, OHG salbōn « anoints », with *-ō- < *-āḭe/o- < *eh2-ḭe/o-). The much-discussed Verschärfung or “hardening” of intervocalic *-i- and *-u-to PGmc. *-jj- and *-ww- seems to have been phonologically regular when pre-Gmc. *-j- was preceded by * or an i-diphthong (i.e., *-ijja- < *-īja-, *-ajja- < *-aija-, etc.), or when pre-Gmc. *-w- was preceded by *ū or a u-diphthong (i.e., *-uwwa- < *-ūwa-, *-awwa- < *-auwa-, etc.). Such sequences frequently arose through the loss of a laryngeal, as in PGmc. *hawwanan « hew, chop » (ON, hǫggva, OE hēawan) < *hauwa- < *hau ’a- < *kauh2-o-.

  • 5 It is here assumed that *kw, *hw and *gw, the reflexes of the PIE labiovelars, patterned phonologic (...)

12Proto-Germanic, then, had the following consonant system:5

13This inventory was preserved in Gothic, where the only important phonological innovations were the devoicing of word-final voiced fricatives (cf. nom. sg. dags « day » < *daǥaz) and the change of *-ww- and *-jj- to -ggw- and -ddj- (cf. triggwa « faithful » < *triwwa-, twaddje « two [gen. pl.] » < *twajjēn). In Old Norse the contrast between voiced and voiceless fricatives was lost altogether: *z became r (cf. dagr « day »), while *f and *] o were voiced to [v] and [đ] after vowels and liquids, thus partly effacing the results of Verner’s Law (cf. bróđir and fađir, both with – ð.) Old Norse also witnessed important changes to the glides: *h, no longer synchronically interpreted as a voiceless fricative, was restricted to word-initial position (cf. sld « slay » vs. Go. slahan); single *w and *j were lost in a number of conspicuous environments (cf. ormr « worm », ungr « young » vs. Go. waurms, jaggs’); *-ww- and *-jj- became -ggv- and – ggj- (cf- tryggvi (= Go. triggwa), tveggja (= Go. twaddje)). Geminate consonants, already firmly established in Proto-Germanic (cf. *fulla- « full » < *fulna- < PIE *pḷ h1 -nó-; *wissa- « known » < PIE *ṷit stó- < *ṷid-tó-), became much more common, especially in West Norse, through such rules as the assimilation of *– n] o, *-l] o-to -nn-, -ll-, the assimilation of *-nt-, *-nk- to -tt-, -kk-, and the change of *-kj-, *-ǥj- to -kkj-, -ggj-.

14Some of these developments were paralleled in West Germanic. Here, as in Old Norse, *z was lost as an independent phoneme, disappearing in final position (cf. OS dag « day ») and falling together with *r elsewhere (cf. OS wārun « were » < *wēzun). Word-internal *f merged with *b in part of the West Germanic area, but *d became *d everywhere, thus remaining distinct from *] o even in languages like Old English, where the voiceless fricatives were subject to secondary voicing (cf. brōðor vs. feeder). An important West Germanic novelty was the gemination of all consonants except *r before a following *j – a development which led to the rise of characteristic morphophonemic alternations between single and double consonants (cf. OS tellian « say » < *taljan; pret. teliđa < *taliđōn). The most conservative language of the group, at least from a phonological point of view, was Old Saxon. Old English and Old Frisian, among other innovations, palatalized the velar consonants to affricates before front vowels and *j (cf. OE licgan [-ǰǰ-], OFr. lidza « lie » < WGmc. *liggjan), reducing *g before front vowels to the status of a palatal glide. In Old High German the West Germanic consonant system was utterly transformed by the High German Consonant Shift. This complex of changes, which began in the Upper German area in the sixth century, eventually spread in a less general form to the East, Rhenish and Middle Franconian dialects. The most characteristic feature of the shift was the passage of the voiceless stops *p, *t, *k to affricates, which became long tense fricatives in postvocalic position. Examples are legion: cf. OHG pfeffar beside Eng. pepper; OHG zehan (<z-> = a As-like affricate), wazzar (←zz→ = a sibilant distinct from /s/) beside Eng. ten, water; OHG chuninc (<ch-> = [kx] (Upper German only)), mahhōn beside Eng. king, make. The voiced sounds were also affected: WGmc. *d became *t in Upper German and part of Middle German (cf. tag « day »), while WGmc. *b/ƀ, *g/ǥ and early OHG *d < *] o tended to become p, k and t in Upper German.

  • 6 Here and below, “PIE *a” and “PIE *o” refer indifferently to “true” *a and *o and to the secondary (...)

15The modem Germanic languages have unusually rich and complex vowel systems. The vowel system of Proto-Germanic, however, was notably simpler than most of its descendants. Proto-Indo-European had five short vowels – an inventory which Germanic reduced to four by merging PIE *a and *o as *a (cf. Go. akrs « field » beside Lat. ager, Gk. agrós « id. »; Go. ahtau « eight » beside Lat. octō, Gk. oktō)6. (PIE « *ә », in reality a subphonemic epenthetic vowel that developed in the vicinity of cluster-internal laryngeals, yielded *a in initial and final syllables (cf. « father », ultimately < *ph2tḗr) but disappeared elsewhere (cf. Go. dauhtar « daughter » < *dhuĝh2tēr)). There was also confusion of the front vowels *e and *i in some environments. *e was raised before a following *i or *j (cf. OHG ist « is » < *esti), before tautosyllabic nasals (cf. OHG hring « ring » < *hringaz < *hrengaz (> Finnish rengas « id. »), and in non-initial syllabes (cf. OHG (gi) nerit « saved (past ptcp.) » < *naziđaz < *nosetós). These rules affected the inherited diphthongs as well. PIE *ai, *oi and *au, *ou fell together as PGmc. *ai and *au (cf. Go. haihs « one-eyed » (: Lat. caecus « blind »), wait « knows » (: Gk. (w) oĭde « id. »), while PIE *ei developed via *ii to *ῑ (cf. OHG bīz (z) an, Go. beitan « bite » (<ei> = [Ī]) < *bheid-). PIE *eu, the only remaining diphthong, was raised to *iu before a following i-vowel (cf. OHG biotan « offer » < *beuđan, 3 sg. biutit < *biuđi] o).

16Just as the short vowels *a and *o merged as PGmc. *a, the PIE long vowels *ā and *Ō fell together as PGmc. (cf. Go. bro] oar (<o> = [ō]) beside Lat. frāter, Gk. phrᾱtēr; Go. flodus « tide » beside 0Gk. plōtós « floating »). There was no confusion, however, between PIE *ē and *ῑ; *ē consistently yielded a long open front vowel (phonetically probably [ǣ]), which was lowered further to *ā in North and West Germanic (cf. Go. 3 pl. setun (<e> = [ē]), OHG sāzzun « sat »). The long diphthongs *ēi, *ōi, etc. were shortened before consonants in Germanic, as in most other IE languages ( “Osthoff’s Law”), but probably maintained their existence in absolute final position until the end of the Proto-Germanic period.

  • 7 The corresponding Gothic form, haurn, is ambiguous, since PGmc. *u would regularly have given au ( (...)
  • 8 The Neogrammarian derivation of PGmc. *hē2r from PIE * k̂ēi-r is almost certainly incorrect; see fu (...)

17Although the loss of the distinction between a- and o-vowels reduced the stock of early Proto-Germanic vowels to four shorts and four longs, the complexity of the system was increased by subsequent changes. Toward the end of the Proto-Germanic period – or immediately thereafter, according to some scholars – the number of short vowels was again brought to five by the lowering ( “a-umlaut”) of *u to *o before an *a or *ō in the following syllable. The effect of this rule can be seen, e.g., in the word for « hom », which appears in Old Norse, Old High German, Old Saxon and Old English as horn, even though the original preform was clearly *hurnan< *k̂ṛnó-7. *i was occasionally lowered to *e in the same environment, as in OHG, OE nest “nest” < *nizdó- (: Skt. nīḍá.–, Lat. nīdus, etc.). A Proto-Germanic change which was purely phonetic, but which had phonological consequences in the daughter languages, was the development of the long nasalized vowels n, *ῑn and n from earlier *an, *in and *un before *h (cf. Go.] oahts [-ā-], OHG (gi) dāht, OE] oōht « thought [past ptcp.] » < *] oanhta-). Also important was the addition to the system of « 2, » a closer vowel than the old *ē < PIE , from which it remained distinct in every Germanic language except Gothic. The origin of 2 is one of the most controversial issues in Germanic phonology. Very few lexical items can be shown to have had this vowel in Proto-Germanic, and most of them were either borrowings, like Go. Kreks « Greek » (cf. OE pl. Crēcas), or words of unclear etymology, like Go. fera, OHG fēra, fiara « side ». A central role in discussions of the problem is played by the word for « here » (Go. her, ON hér, OHG hēr, hiar, OE hēr), the exact formation of which remains obscure8.

18Late Proto-Germanic thus had the following vowels in non-final syllables:

19Gothic, eminently conservative in its treatment of the Germanic consonant system, was more innovative in its treatment of the vowels. The distinction between the short high vowels and short mid vowels was for all practical purposes lost: *e and *o became i and u, which were then lowered ( “broken”) to [Ɛ] and [ɔ], written <ai> and <au>, before r and h (cf. wair] oan « become » < *wir] o>- < *wer] o-). Similarly, *eu and *iu fell together as iu, while the remaining diphthongs, *ai and *au, were monophthongized to [Ɛ̄ε] and [ͻ̄ε] – a change not indicated in Wulfila’s orthography (haitan “order”: [hεɛtan]). PGmc. *ē and 2 merged as [é], written <e>. The nasalized vowels lost their nasality, establishing /ā/ as a new phoneme distinct from / ō/ and /ɔɛ/.

  • 9 It will be noted from this example that the “long” syllabic sonorants *, *, *ṃ and * fell togeth (...)

20Old Norse maintained many of the phonological features that were lost in Gothic, though often with significant phonetic alterations. The five short vowels of Proto-Germanic were in principle retained; all, however, were subject to umlaut rules of various kinds (see below), and *e was sometimes “broken” to ja or , as e.g., in hjalpa « help » (: OE helpan, Go. hilpan). The long vowels generally preserved their identity, with *ē becoming ā (<á>), as everywhere outside Gothic. Of the diphthongs, only *au survived as such; *ai became ei (cf. steinn « stone » beside Go. stains), while *eu and *iu transfered their syllabicity to the second element, giving and jú (cf. bjóða beside Go. biudan). Massive reductions took place in unstressed syllables. Here length distinctions were obliterated, diphthongs were monophthongized, and all syllabic nuclei that resisted syncope were ultimately reduced to a, i/e, or u/o. Thus, e.g., the PGmc. 2 sg. form *wakaiz (i) « you are awake » (= Go. wakais) yielded ON vakir, while the corresponding 1 sg. pret. *wakaiāōn « I was awake » gave ON vakða New instances of n, *īn and n were produced from short vowels before *-ns-. All nasalized vowels, old and new, eventually lost their nasal component, but and n were first lowered to n and n, whence é and ó (cf. ósk « wish » < *wūnskō < *wunskō < *ṷṇskā < *ṷṇh1sk-)9.

21Of the West Germanic languages, Old High German was the most conservative. Except when subject to umlaut, the short vowels were fairly well preserved, even in unaccented syllables. Syncope, far less pervasive than in Old Norse, was sensitive to syllabic weight: pre-OHG *hōrita « I heard » became hórta (cf. ON heyrða), but *lagita « I laid » remained as legita (contrast ON lagða). The most important changes were those affecting the long vowels and diphthongs. WGmc. *ē (< *ē2) and *ō showed an early tendency to diphthongize to ia, ie and ua, uo (cf., e.g., bruoder beside Go. bro jar), while a new ē and ō were generated from *ai and *au in unstressed and, under certain conditions, stressed syllables (cf. lōn « wage » beside Go., ON laun). Where they survived as diphthongs, *ai and *au became ei and ou (cf. houbit « head » beside Go. haubi] o); *eu generally passed to io, remaining distinct from iu < *iu. Most of these rules also operated in Old Saxon, except that here the diphthongization of *ē and *Ō to ia and uo was confined to one of the two main dialects, and the change of *ai and *au to ē and ō was universal (cf. stēn « stone », hōƃ id « head »). Unlike Old High German, Old Saxon sporadically developed nasalized vowels before the voiceless fricatives *s, and *] o – an Ingvaeonic rule that applied with greater consistency in Old English (cf. OS sāfto « gently » < sōnjto < *sanfto). Both Old High German and Old Saxon merged the nasalized vowels with their oral counterparts before the beginning of the literary period.

  • 10 The precise environments for this rule were complex and varied considerably from dialect to dialect

22In Old English the most important changes to the short vowels were the fronting of *a to œ (cf. dœz “day” (<z> = [j]) and the subsequent “breaking” of œ, e and i to the short diphthongs ea, eo and io before liquids and glides (cf. beorg « mountain » < *berg; slēan « slay » < *sleahan < *slœhan < *slahari)10. The long vowels, when not umlauted, were generally maintained, but WGmc. (< PGmc. *ē) was fronted to œ̄ (cf. wœron « werc » = OS wāruri), and a new ā was produced by the monophthongization of *ai (cf. stārt « stone »). The zz-diphthongs remained diphthongal and contrastively long: *au developed via *æw and *eaw to ēa (cf. hēafod « head »), while *eu and *iu gaveēo and ῑo, which later fell together in the dominant West Saxon dialect. As in Old Norse, there were far-reaching shortenings, reductions and losses in non-initial syllables, the effect of which was to leave a, u/o and e [ə] as the only completely unstressed vowels in the language. Nasalized vowels were generated before *-ns-, *-n] o- and *-nf- as well as before *-nh-; *ān was rounded to nand became ō with the loss of distinctive nasalization (cf. sōfte beside OS sāfto). The treatment of the vowels in Old Frisian was broadly similar.

23All the Germanic languages other than Gothic show the effects of umlaut rules, which caused a vowel, normally stressed, to assimilate in one or more features to a vowel or consonant that followed. A phonetic tendency toward such assimilation was probably already a feature of Proto-Germanic; early illustrations can be seen in the raising of *e to *i before *i and *j and the lowering of *u to *o before *a and *ō. By far the most conspicuous type of umlaut in the daughter languages, however, was i-umlaut proper, which converted rz-like vowels to e-like vowels, and back rounded (u-like and o-like) vowels to front rounded (ü-like and ö-like) vowels before *i, *ῑ or *j in the next syllable. There is no trace of this rule in Gothic or early Primitive Norse, but its workings can be seen in every later Germanic language. In general, -umlaut was slower to establish itself in the continental West Germanic dialects than elsewhere. In early Old High German the process was still purely phonetic: vowels were fronted before a preserved z or j, but not before an etymological *i which was lost by sound change (cf. gast « guest » < *ǥastiz, but nom. pl. gesti < *ǥastTīz). In most cases the effects of i-umlaut were not shown in Old High German spelling, but began to be graphically represented in Middle High German, where the weakening of unstressed i to e [ə] rendered the new umlaut vowels contrastive (cf. scōni « bright, fair » > MHG schœne > NHG schön). By the middle of the eleventh century all the Old High German back vowels and diphthongs had acquired phonologically distinctive fronted counterparts, as was also the case in Old Saxon/Low German. In Old Low Franconian and Dutch -umlaut remained inconspicuous and purely phonetic, eventually disappearing with the loss of i in unaccented syllables.

24In Old English and Old Norse the phonologization of -umlaut was very early. The word for « guest » in these languages shows umlaut in the singular as well as the plural (OE nom. sg. ziest, ON gestr < *-iz; cf. Prim. Norse -gastiR); umlaut is also found in preterites of the type OE hīerde, ON heyrða « heard » < *hauziƌōn, where the -i- that triggered the rule was syncopated after heavy syllables. Old Norse generally maintained its newly acquired rounded vowels and diphthongs (y, ý, ø, ǿ, ey), but Old English (West Saxon) kept only y and ý, unrounding the umlaut products of ō and io to ē and īe (cf. fēt (dialectal fœ̄t, ON fǾtr) “feet” < *fōtiz). A different but related process in Old Norse was labial umlaut (often separated into “u-umlaut” and “w-umlaut”), which induced rounding effects in stressed vowels similar to the fronting effects of i-umlaut (cf. nom. pl. lǫnd « lands » < *landu < *landd’, s0kkva « sink (tr.) » < *sekkwa < *sankwijan (= Go. sagqjan, OHG senken)). The term “u-umlaut” is also applied to a special breaking process in Old English (cf. siolfor, seolfor « silver » < *siluƀra-).

25The far-reaching changes of the immediate post-Germanic period left the medieval Germanic languages with exceptionally complex, “crowded” vowel systems. It is no accident that in the later development of Germanic these configurations were often unstable. “Chain shifts”, marked by fresh diphthongizations, monophthongizations, frontings and/or raisings, dot the history of the modem languages. The best known example is the so-called Great Vowel Shift in English, which began in the fifteenth century, with the diphthongization of ĭ and ū to [ei] (later [ai]) and [ou] (later [au]), and the raising of the Middle English high mid vowels ē and ō to fill the resulting structural gap. No other language shows precisely the same sequence of changes, but Standard German implemented a similar set of rules, diphthongizing MHG ī, ū and ü to [ei], [ou] and [öü] (later [ai], [au], [ɔi]), and then producing new instances of ī, ū and ü from the old diphthongs ie, uo and üe. An entirely different chain shift, triggered by the fronting of u to [ü], took place in Swedish and Norwegian.

26The treatment of final syllables raises especially difficult problems in Germanic. Germanic was one of the branches of Indo-European which, like Balto-Slavic, Greek and Indo-Iranian, distinguished between two types of long final syllables. “Acute” finals, better termed “bimoric”, were normally subject to shortening; thus, the bimoric accusative singular of o-stem nouns, which originally ended in *-ōn (< *-ōn < *-ām < PIE *-eh2m), gave forms in (short) -a in Gothic (cf. giba « gift (acc.) ») and -a in Old High German (geba « id. »), corresponding to Greek -an (-ān under the accent) and Sanskrit -ām. Contrasting with these were “circumflex” or “trimoric” syllables, which maintained their length with greater tenacity. Such a trimoric ending was the *n – note the use of the circumflex accent – of the Proto-Germanic genitive plural (cf. Gk. -ōn (-ōn), Skt. – ām, -aam), which survived as (long) -o in Gothic (gibo) and as -o in Old High German (gebono). Trimoric finals arose in late Proto-Indo-European and early Proto-Germanic from earlier sequences containing a hiatus, typically the result of a lost laryngeal. The introduction of the bimoric: trimoric opposition in final syllables marked a complication in the Germanic phonological system vis-à-vis that of the parent language, but the number of possible contrasts in final syllables was reduced in other ways. All final obstruents other than and *z were lost, and final nasals disappeared with nasalization of the preceding vowel. This left *s, *z, *r, *l and *n (< *-nt) as the only consonants permitted to stand at the end of a Proto-Germanic word.

27Final syllables were lost or reduced in every Germanic language, but the chronological details are unclear. The apocope of short vowels other than *u in absolute final position seems to have had its beginnings in late Proto-Germanic (cf. Go. wait « knows », ON veil, OHG weiz, OS wēt, OE wāt < *wait (i) < *ṷóide). On the other hand, the syllable types *-vɛn and *-VC were retained into the dialectal period, as can be seen from forms like Prim. Norse horna « horn » < *hornan and -gastiR < *ǥastiz. Gothic instituted a Dreimorengesetz ( « Law of Three Moras »), by which most final syllables were abbreviated by one mora: short vowels disappeared, bimoric long vowels were shortened, and trimoric long vowels were reduced to ordinary longs (cf. dags « day » < *-az, giba « gift » < *ō, galeiko « likewise » < *-ō). North and West Germanic weakened (bimoric) *-ō to *-u soon after their separation from Gothic; this vowel, like all short-vowel endings that survived the first wave of truncations, was eliminated or retained according to circumstances in the individual dialects (cf. ON gjǫf vs. ΟΕ ziefu < ǥeᵬu (*-ō, but after a heavy syllable ON run « rune, secret », OE run < *rūnu < *-ó). The trimoric endings were eventually reduced as well, but later; no North or West Germanic language completely elided the *-dn of the genitive plural (ON -a, OHG -o, OS -o, OE -a) or the *-ō of adverbs like Go. galeiko (ON -a, OHG -o, OS -o). In every language except Icelandic and Faroese the tendency of vowels in final syllables to weaken or disappear persisted into the modem period, and went hand in hand with the decline of inflectional morphology.

3. Morphology and syntax

3.1. Declension

28In comparison with the turbulent history of Germanic phonology, the historical morphology of Germanic seems almost prosaic. Particularly in its nominal system, Proto-Germanic was a fairly typical IE dialect, not unlike, e.g., Latin, its approximate contemporary. There were three genders (masculine, feminine and neuter) and two numbers (singular and plural), the dual having been lost except in the personal pronouns. Of the eight PIE cases, Germanic preserved the nominative, accusative, vocative, genitive and instrumental; the locative, perhaps with some admixture from the dative, gave the synchronic “dative”, while the ablative survived only in adverbs. No individual language, however, actually retained six cases. The vocative, proper to the singular alone, disappeared everywhere except in Gothic; the instrumental survived as a case-form distinct from the dative only in a few noun classes in West Germanic, and here again only in the singular. Noun paradigms in the older Germanic languages typically consist of a nominative, accusative, genitive and dative, sometimes with the addition of a vocative or instrumental in the singular.

29The inflection of PIE nominal stems was a complex process, commonly involving accent movement and stem-internal ablaut (apophony) as well as the addition of case endings (cf. e.g., nom. sg. *mén-ti-s (later *mṇ-tí-s) « mind », gen. *mṇ-tei-s, loc. *mṇ-tē). Most of this apparatus was eliminated in Proto-Germanic. Accentual mobility was lost by sound law; all words acquired fixed initial stress. The role of ablaut was drastically reduced, partly through the expansion of the apophonically invariant a-stems and ō-stems (= PIE o-stems and ā-stes), and partly through the synchronic reanalysis of inherited forms with suffix ablaut (cf. Go. dat. sg. gamundai « mind », historically < *mṇ-tḗi-Ø but synchronically -mun-dai). Descriptively speaking, Proto-Germanic nouns fell into five major and several minor declensions, each characterized by a distinctive set of synchronic endings and conventionally named for the vowel or consonant that historically ended the stem (zz-stems, d-stems, z-stems, zz-stems, zz-stems). Following a venerable tradition initiated by Grimm, the vocalic stems are sometimes called “strong”, while the n-stems are called “weak”.

30A good impression of noun declension in the older Germanic languages is afforded by the Gothic, Old Norse and Old High German paradigms of the masculine zz-stem *daǥaz « day »:

  • 11 The PIE instr, pl. of o-stems ended in *-ōis; PGmc. *-amiz was created on the model of the other st (...)

31These forms, corresponding etymologically to the Latin and Greek second (o-stem) declension, can in most cases be easily traced to Proto-Germanic and PIE antecedents. In the singular, the nominative, accusative, vocative and instrumental go directly back to PGmc. *-az (PIE *-os), *-an (PIE *-om), *-i (PIE *-e) and *-ō (PIE *-ō < *-oh1), respectively. The gen. sg. in Old High German and Old Norse continues PGmc. *-asa < PIE *-oso, a variant of *-osḭo (= Gk. -oio, Skt. -asya); Go. -is is a borrowing from the pronominal declension, where, e.g., the gen. sg.] ois « of this, of that, eius » continues PIE *teso. The dat. sg. in Gothic probably represents the old instrumental; in Old High German and Old Norse (where degi underwent the rule of “g-umlaut”) the corresponding ending is best taken from the PGmc. loc. sg. in *-aĩ (trimoric?) < PIE *-oʻi. Of the plural forms, the Gothic and Old Norse nominative and accusative go back to *-ōz (or *-<ōz; PIE *-ōs < *-o-es) and *-ans (PIE *-ons), one or both of which presumably also underlies the somewhat mysterious OHG -ā . The gen. pl. in Old Norse and Old High German continues PGmc. *ōn (PIE *-oHom?), replaced under obscure circumstances by *n in pre-Gothic. The dat. pl. forms are, as always in Germanic, instrumentals in *-mi (z) < dialectal PIE *-mi (s), with *-m- for *-bh- as in Balto-Slavic11.

32There were also neuter a-stems, which maintained the Indo-European peculiarity, common to all neuters, of making no distinction between nominative and accusative (cf. Go. nom.-acc. sg. land « land » < *-an < *-om, nom.-acc. pl. landa < *-ō < *-ā; ON land, pl. lǫnd < *-u; OHG lant, pl. lant < *-u). A few neuter a-stems in West Germanic had plurals in *-ir (u) (cf. OE cild “child”, pl. cildru; OHG lamb « lamb », pl. lembir (= NHG Lämmer)); these were originally s-stems (cf. Lat. genus « race », pl. genera (< *-esa)) in which the formative element -ir- < *-iz- < *-es- was synchronically reinterpreted as a number marker. Very important from a descriptive point of view were the ja-stems, both masculine and neuter, which continued PIE stems in *-io- and *-iḭo-. In Proto-Germanic the distribution of *-ja- (< *-io-) and *-ija- (< *-iḭo-) was metrically regulated: light syllables were followed by *-ja-, while heavy syllables and polysyllabic stems were followed by *-ija-. The difference between the two suffix-variants, ultimately a reflex of the PIE rule known as Sievers’ Law, appears clearly in Go. harjis « army » (< *korḭo-) vs. hairdeis « shepherd » (< *k̂erdh.iḭo-). ja-stem inflection was productive in Germanic, being associated, e.g., with the borrowed suffix *kārija- < Lat. -ārius, which was later exploited as a device for forming agent nouns (cf. Go. bokareis « scribe », OE bocere « id. », OHG lērāri (= NHG Lehrer) « teacher »).

33The other strong declensions present few problems. The ō-stems, like their PIE counterparts, the a-stems, were exclusively feminine. Many were abstracts, including the common deadjectival nouns in *-i] oō (cf. Go. mildi] oa « mercy », OHG miltida beside OHG milti « mild, merciful »). A distinct subclass were the -stems, some of which preserved an archaic nom. sg. in PGmc. *-ī, recalling the Sanskrit type devī « goddess », gen. devyās (cf. Go. bandi « bond », gen. bandjos). The i-declension originally contained both masculines and feminines, which had identical paradigms in the parent language. Over time, however, the two diverged; in every Germanic language the masculine i-stems were analogically influenced by the a-stems, while the feminines, which included the common deverbal abstracts in *-ƌiz (*-tiz), tended to keep their inherited endings (cf. Go. gasts, gen. gastis [m.] vs. gahugds « thought », gen. gahugdais [f.]). Both genders retained -i- or -j- in the endings of the plural, a fact ultimately responsible for the rise of the umlaut plural types Gast: Gäste and Kraft: Kräfte (< OHG kraft [f.] « strength », pl. krefti) in Modem German. The u-stems, containing words of all three genders, were at the outset a relatively small class. In Gothic, however, the inherited stock of u-stems was considerably augmented by loanwords (e.g., aggilus « angel ») and former root nouns, which developed an -u- by sound change in the acc. sg. (cf. fotus « root », acc. fotu < *-un < *-un < *-ṃ).

34Most of the early Germanic dialects preserve traces of consonant-stem inflection in the kinship terms (r-stems; cf. Go. bro/oar, gen. bro] ors, dat. bro] or), in petrified present participles (nt-stems; cf. Go. fijands « enemy », dat. fijand, nom. pl. fijands) and in scattered root nouns (cf. Go. baurgs « city », gen. baurgs, dat. baurg). None of these groups, however, compares in importance with the n-stems, the expansion of which was one of the most characteristic morphological innovations of Germanic. The masculine n-stems were an inherited class, originally marked by an ablauting suffix *-on-/*-en-/*-n-, with a nom. sg. in PIE *-ō < pre-PIE -ōn (cf. Lat. homo, – inis « man »). (A less numerous subtype had only *-en- and *-n-, with *-en in the nom. sg. (: Lat. pecten, -inis « comb »)). In pre-Germanic the nom. sg. in *-ō was remade to *-ōn, which became PGmc. *n; the other cases added the consonant-stem endings, which were mostly of the form *-V, *-V n or *-VC, to the *-n- of the stem. The result was a paradigm in the daughter languages in which the nom. sg. ended in a vowel and the oblique cases tended to end in simple -n, with no desinence proper (cf. Go. guma “man”, gen. -ins, dat. -in, acc. -an, nom.-acc. pl. -ans; OHG gumo, -en, -en, -on, -on; OE guma, -an, -an, -an, -an). Each language had its own peculiarities: the West Germanic dialects collectively substituted trimoric *-Ōnfor n in the nom. sg., probably under the influence of the weak adjectives (see below); Old English generalized the o-grade suffix-form -an- to all the oblique cases; Old Norse replaced *-dn by *-en in the nom. sg. and lost *-n by sound change elsewhere (gumi, -a, -a, -a, pl. -ar). The neuter u-stems were likewise an old type, the inherited membership of which included the PIE word for « name » (cf. Go. namo < *n, gen. -ins; cf. Lat. nomen, -inis). Wholly new in Germanic, on the other hand, were the feminine n-stems, which were of two types: a) n-extended forms of original ō-stems (< ā-stems), such as Go. tuggo « tongue », gen. tuggons, dat.-acc. tuggon (= OHG zunga, -un; OE tunge, -an; cf. OLat. dingua, Skt. jihvā); and b) similarly extended forms of original feminines in *--, which served as adjectival abstracts (cf. Go. hauhei « height », gen. -eins; OHG hōhī, gen. hōhī (n)). In the later West Germanic languages the n-sterns were the starting point for the common plural types in -en (cf. Eng. oxen, NHG Zungen, etc.).

35The declension of pronouns in Proto-Germanic, as in every other IE language, was full of anomalies. The personal pronouns were highly irregular, but retained such archaisms as separate dual forms in the first and second persons (cf. p. 257, fn. 4) and a suppletive paradigm based on the PIE stems *uei- and *nes- in the 1 pl. (cf. Go. weis « we » vs. uns « us », recalling Skt. váyam: nas). The possessive adjectives *mīna- « my », *] OĪna- « thy », *sīna- « his, her, their (own) », *unsara- « our », *izwara- « your » were distinctively Germanic creations. Among the demonstratives, Germanic retained the PIE *so-/*to- pronoun (cf. Go. nom. sg. masc. sa « this, that », fem. so, nt.] oata), which became a definite article in West Germanic, as well as the pronoun *e-/ *i-, which furnished personal pronouns, especially in Gothic and Old High German (cf. Go. is, ita « he, it », OHG er, iz). Most Germanic languages also made significant use of the PIE demonstratives *k̂e-/*k̂i- (cf. OE, OS hē « he », ON hann « id. » and *eno-, which became a definite article in Old Norse (nom. sg. masc. -inn). All these shared a number of inflectional peculiarities, such as the presence of a t-element in the nom.-acc. sg. neuter (Go.] oata, OHG daz; cf. Skt. tád), an ending *-nōn in the acc. sg. masc. (Go.] Oana, OHG then), a stemenlargement *-mm- < *-zm- in the dat. sg. masc. and nt. (Go.] oamma, OHG themu; cf. Skt. tásmai), and an enlargement *-z- in the gen. and dat. sg. fem. and gen. pl. of all genders (Go.] oizos, ] oizai, ] oize/] oizo; cf. Skt. tásyās, tásyai, téṣām/ tāsām)). The interrogative stem *hwa- (Go. /υαs « who », OHG (h) wer, etc.), though lacking a plural, displayed many of the same features. There was no separate relative pronoun in Germanic, the PIE relative *ḭo- (cf. Gk. hós, Skt. yá-) having disappeared.

  • 12 Comparative adjectives, made with the zero-grade of the PIE suffix *-ḭos-/ *-ḭes-/ *-is-, lacked st (...)

36Adjectives were provided with a dual inflection in Proto-Germanic. Depending on their immediate syntactic environment, they either remained “strong”, being declined as a/ō-stems, ja/-stems, i-stems or (occasionally) zz-stems; or they became “weak”, acquiring the declension of zz-stems. The weak declension of adjectives arose from the use of *-on-/*-en-/*n- as an “individualizing” suffix in late Proto-Indo-European – a function seen in Greek and Latin personal names of the type Gk. Strábōn, gen. Strábōnos « Squint-eyes » < strabós « squint-eyed » and Lat. Cato, -ōnis « Smarty » < catus « smart, sharp ». PGmc. *ǥōđaz thus originally meant simply « good [nom. sg. masc.] », while *ǥōđn (i.e., *gōđa- + -ōn) meant « the good one ». The distinctive innovation of Germanic was to reinterpret the semantic difference between the two in purely syntactic terms, so that *gōđōn became obligatory in the inherently individualizing environment of a preceding demonstrative, even when a noun followed (*sa ǥōđōōn gumōn« the good man »). In this way every adjective acquired a complete set of n-stem forms in all three genders. Predicate adjectives and adjectives in indefinite noun phrases were unaffected by the new rule and continued to inflect as vocalic stems (*ǥōđaz gumd n « a good man »), though their individual declensional forms were in many cases borrowed from the pronouns (cf. Go. dat. sg. masc. godamma, acc. sg. godana, gen. sg. fem. godaizos, etc.; cf. NHG gutem, guten, guter)12 It is of interest to note that a typologically similar distinction between “definite” and “indefinite” adjectives was created, though with very different formal means, in Balto-Slavic.

3.2. Conjugation

3.2.1. Strong verbs

37The history of the Germanic verbal system is for the most part a history of simplification and regularization. Proto-Indo-European had a present and three preterite-like tenses: the imperfect, built from the present stem and thus marked for durativity or iterativity (cf. Gk. éthnéiskon « I was dying »; the aorist, properly denoting a punctual, and hence non-present, action (éthanon « I died »); and the perfect, originally denoting a state but frequently taking on the value of a resultative preterite (téthnēka « I am dead » > post-class. « I have died »). Germanic merged all three into a single category, known simply as the preterite. The preterite and the present, which served also as a future, were the only tenses in the Proto-Germanic verbal system; there were no periphrastic formations of the type found in the medieval and modern Germanic languages (cf. Eng. I will do, I have done, etc.). In its treatment of the inherited categories of mood, voice, person and number, Germanic was an averagely conservative branch of the IE family. Of the moods, the subjunctive was lost, but the indicative, imperative and optative were retained. The PIE mediopassive was preserved as a grammatical passive, at least in the present (cf. Go. bairada < *bhérotoi; cf. Gk. pheretai); a trace of the older middle value survives in Go. 1, 3 sg. haitada « am, is called », OE hātte « id. », Prim. Norse 1 sg. haite (< PGmc. *haitōi) « am called ». Corresponding to the special dual pronouns of the first and second persons, Proto-Germanic retained distinct first and second person dual forms of the verb. In the historical period, however, the passive and the dual were confined to Gothic. The earliest North and West Germanic languages expressed the passive by syntactic means and used plural verb forms with dual subjects.

38Presents were formed in a variety of ways in the parent language. Some verbal roots, for example, made presents without any kind of affixation (e.g., *h1ḗd-ti [root *h1ed-] « eats »); others made reduplicated presents (e.g., *dhédheh1-ti [root *dheh1-] « puts » or nasal-infix presents (e.g. *bhi-né-d-ti [root *bheid-] « splits »); still others made “thematic” presents in *-e/o- (e.g., *bhér-e-ti [root *bher-\ « carries ») or *-ḭe/o- (e.g., *bhudh-ḭe-tor [root *bheudh-] « is alert, notices »). Germanic, like most of its sister branches, greatly extended the range of the thematic formations; in particular, it regularized and generalized the simple thematic type with e-grade of the root (*bhéreti). The overwhelming majority of primary verbs were provided with such presents, which often replaced older formations. Thus, e.g., *h1éd-ti was, in effect supplanted by *édeti (Go. itib « eats »); *bhi-né-d-ti became *bhéideti (Go. beiti] O « bites »); *bhudh-ḭé-tor became *bhéudheti (Go. -biudi]O « offers, orders » < « remarks »). Thematic presents were also built to roots which had no synchronic e-grade, such as PGmc. *an- « breathe » (cf. Go. -ani] O < *áneti < *h2enh1- or PGmc. *skaƀ- « scrape » (cf. Go. skabi]O < *skábheti < *skabh-). Remaining vestiges of the non-thematic present classes were generally thematized, as, e.g., Go. standib « stands », with both nasal infix (cf. OE pret. stōd, Eng. stood) and stem-final *-e/o-. The only competing present formation of any prominence was the type in *-ḭe/o-, which was especially common in verbs with Germanic a-vocalism (cf. Go. hafji/o, OHG hevit « lifts » (= Lat. capit < *káp-ḭe-ti « takes »).

39The inflection of simple thematic presents in the Germanic dialects can be illustrated by the following forms of the verb « to become » (root PGmc. *wer] o- < PIE *ṷert- « tum »):

40The endings of the Proto-Germanic indicative forms were best preserved in Gothic. Gothic retained the original distribution of e- and o-color in the thematic vowel, with -i- < *-e- in the 2 sg. (PIE *-esi), 3 sg. (*-eti) and 2 pl. (*-ete), and -a- < *-o- elsewhere (1 sg. *-ō < *-o-h2, 1 pl. *-ome, 3 pl. *-o-nti). The 2 sg. in -is and 3 sg., 2 pl. in -i] o go back to *-iz and *-iđ, with voicing by Verner’s Law, as shown by the preserved -d of the 3 pl. in -and. Voiced endings were also generalized in Old Norse, as is clear from the 2 sg. in -r < *-iz; a striking Old Norse innovation was the substitution of -r for *-ð in the 3 sg., a change datable to the seventh century. Old High German preserved a Proto-Germanic feature not found in Gothic or Old Norse – the raising of the root vowel -e- to -i- in the 2 sg. and 3 sg., where the following ending contained an -i-. (A later extension of the same process brought about the change of *wer/ou to wirdu in the 1 sg.) The OHG 1 pl. in -umēs is noteworthy: the development of -u- from – a- in medial syllables before -m- was phonologically regular in North and West Germanic (cf. ON -um), but the sequence -mēs remains unexplained. Old English 2 sg. wierst and 3 sg. wier] o presuppose earlier *wir/ois and *wir/oi/o, with the voiceless variants of the endings (cf. OS -is, -iđ); the addition of -t in the 2 sg., a feature later found in other West Germanic dialects as well (cf. MHG -st), was due in part to the influence of the 2 sg. pronoun. An Ingvaeonic trait, common to Old English and Old Saxon, was the generalization of the 3 pl. in *-an] o (> OE -a] o, OS -ađ) to the other plural persons.

  • 13 Gothic has a 3 sg. impv. in -adau and a 3 pl. impv. in -andau, the exact relationship of which to t (...)

41Of the non-indicative forms, the 2 sg. imperative goes back to the bare thematic stem in *-i < *-e, as in most early IE languages13. The optative forms point to Germanic prototypes in *-ai- (< PIE *-o-ih1-), which survived as (graphic) -ai- in Gothic and underwent monophthongization and/or weakening elsewhere. The endings of the optative were partly distinct from those of the indicative, reflecting the fact that the optative in the parent language was associated with the “secondary” endings, which lacked final *-i. The Gothic 3 sg. in -ai thus corresponds exactly to Skt. -et (bháret) and Gk. -oi (phéroi), while the 3 pl. in -aina shows the secondary addition of a final vowel, presumably taken from the 1 pl., to an original ending *-a (j) in < *-oḭent (cf. Gk. phéroien, Avestan baraiiǝn). The Ingvaeonic languages generalized the ending of the 3 pl. to the other persons, as in the indicative; Old High German kept the plural forms distinct, but with -m, not -mēs, in the 1 pl. All the Germanic languages inherited a present participle in *-and- < *-ont- (cf. Gk. pheront-), which was enlarged to an n-stem in Gothic and Old Norse (Go. nom. sg. wair] Oands or -anda, gen. -andins, dat. -andin, etc.), but to a ja-stem in West Germanic. The attested infinitives in -an (ON -a < *-an) go back to a PGmc. verbal noun in *-anan < *-onom, comparable to the Sanskrit verbal abstracts in -anam (e.g., bháraṇam « carrying »).

42The preterites of the primary verbs in *-anan and *-janan – or “strong” verbs, as they are called – were based on the PIE perfect. Originally a stative category of restricted distribution, this formation became productive in Germanic, encroaching upon and eventually usurping the functions of the imperfect and aorist. The perfect in the parent language was marked by three distinctive characteristics: reduplication, a special set of personal endings, and a distinctive ablaut pattern that opposed an accented o-grade root in the singular to an unaccented zero-grade root in the plural. Of these three features, reduplication was usually lost in Germanic, but the ablaut, accent and endings of the PIE perfect were retained. The result was the Germanic strong preterite, of which the verb *wer] Oanan again furnishes a typical example:

43With the exception of the 2 sg. in West Germanic, which probably continues the old perfect optative used as politeness form (*wurđīz), the indicative singular goes back to PIE 1 sg. *ṷeṷórta (< *-h2a < *-h2e; cf. Skt. vavárta « I turned »), 2 sg. *ṷeṷórt sta (< *-th2a < *-th2e; Skt. vavárttha), 3 sg. *ṷeṷórte (Skt. vavárta). The other forms are based on the “weak” perfect stem *ṷeṷŗt-, which was followed by accented endings in the parent language (cf., e.g., Skt. 1 pl. vavṛtmá). Both the change in vocalism and the accent shift were faithfully preserved in Germanic, the former being seen in the substitution of PGmc. *-ur- (< *--) for *-ar- (< *-or-), and the latter in the “grammatical change” (grammatischer Wechsel) of *–] o- to *-d- – a feature lost in Gothic. The *-u- of the plural endings was probably generalized from the 3 pl. in *-un < *-ṇt, itself the replacement of an earlier r-ending (cf. Skt. vāvṛtúr < *-ṛs). The preterite optative rests on the PIE perfect optative, with *-ī- (< *-ih1-), the variant of the mood sign originally proper to the plural and dual, generalized to the singular as well. The past passive participle of strong verbs points to a formation in *-anaz (var. *-inaz) akin to the Sanskrit perfect middle participle in -āná- (< *-onó-; cf. vāvṛtāná-); before this suffix the root vowel *-u- was lowered to *-o- in North and West Germanic.

44The inflection of strong verbs in Proto-Germanic was thus organized around four principle parts – the present, conventionally represented by the infinitive (*wer] oanan); the preterite singular (1 sg. *war] o < *war] Oa); the preterite plural (*wurđum < *-ume (?)); and the past participle (*worƌanaz). This system was maintained in all the medieval languages, but broke down in the early modem period, when the distinction between singular and plural stems in the preterite was generally abandoned. Traditional Germanic grammar recognizes six “classes” of ablauting strong verbs, of which the first five represent special cases of a single pattern. Class I consists of verbs with a root-internal i-diphthong: these originally had *-ei- in the present, *-oi- and *-i- in the preterite, and *-i- in the past participle, whence, e.g., Go. beitan: bait: bitum: bitans and OHG bīz (z) an: beiz: bizzum: (gi) bizzan. Class II, completely parallel, contains verbs with a u-diphthong (cf. Go. biudan: bau]O: budum: budans, OHG biotan: bōt: butum: (gi) botari). Class III, the type of *wer] oanan, includes verbs with a root-internal sonorant that yielded -ur-, -ul-, -urn- or -un- in the third and fourth principle parts; when the sonorant in question was a nasal, the present regularly had z-vocalism and there was no lowering of *-u- to *-o- in the past participle (cf. OHG bintan: bant: buntum: (gi) buntan « bind »). Classes IV and V consist of verbs with roots ending in a single consonant, either a sonorant (class IV; cf. Go. bairan: bar: berum: baurans, OHG beran: bar: bārum: (gi) borari) or an obstruent (class V; cf. Go. giban: gaf: gebum: gibans, OHG geban: gab: gābum: (gi) geban). The *-ē- of the preterite plural in class IV was borrowed from class V, replacing *-u(*burum < * (bhe) bhṛ-mé), which was analogically extended to the participle on the model of classes II and III. In class V itself the source of the *-ē- in *gēbum is unclear. An analogical origin has been suggested, starting, e.g., from forms like *sētum « we sat » (< *se-zd-?) and/or *ētum « we ate » (< *h1e-h1d- or *h1ēd-). The participle *ǥeƃanaz shows the substitution, probably ancient, of e-grade for the inconvenient zero-grade *gƃ-.

45Class VI, containing verbs built to roots of the form CaC-, with a-vocalism, stands apart from the other types. The nucleus of this category was made up of verbs with old o-grade presents (e.g., Go. malan: mol: molum: malans, OHG malan: muol: muolum: (gi) malan « grind », cf. Lith. málti « id ») as well as verbs with etymological *-a- (e.g., Go. hafjan: hof: hofum: hafans, OHG heffen: huob: huobum: (gi) haban). The *-ō- of the preterite thus continues both PIE *-o- and *-ā-. The closest comparanda are Latin perfects like scābī (pres, scabō) and fōdī (pres, fodiō « I dig »), with corresponding forms in Greek and Celtic.

  • 14 The “neo-diphthong” *ei is not, of course, to be confused with PIE *ei, which yielded in Germani (...)

46Not all strong verbs were capable of ablaut. Presents containing root-internal *-ai-, *-au-, *-ar-, *-al-, *-am- and *-an- (i.e., the structural counterparts, with a-vocalism, of classes I-III) might have been expected to change *-a- to *-ō- in the preterite, thus producing forms of the type *hōit (a) to *haitanan « order », *stōut (a) to *stautanan « push » and *hōlđ (a) to *hađdanan « hold ». Proto-Germanic, however, did not permit long vowels in this position, having shortened tautosyllabic before glides and sonorants to *a by Osthoff’s Law. The distinction between the present and preterite stems was therefore marked in an exceptional way, namely, by the retention of reduplication. Reduplication was also preserved in strong verbs with *-ō- in the present (e.g., *wōpjanan: *wewōp (a) « cry ») and, by analogy to these, in verbs with *-ē- in the present and *-ē- or *-ōin the preterite (e.g., *lētanan: *lelōt (a) « let »). The PGmc. reduplicated preterites maintained their existence in Gothic, where they were inflected regularly (cf, e.g., haihait, lailot, pl. haihaitum, lelotum, opt. 3 sg. haihaiti, lailoti, ptcp. haitans, letans). Traces of similar forms survive also in Old Norse (e.g., sera « I sowed » < *sez.ō- (cf. Go. saiso) and West Germanic (e.g., OE heht « called » (cf. Go. haihait). For the most part, however, North and West Germanic converted the reduplicating verbs to ordinary strong verbs of a new type ( “class VII”), characterized by the replacement of *-aby *-e- in the second and third principal parts. Thus, the preterite of OHG stōzan (< *stautan) was remade to stioz (< *steut), while that of OHG heizan (< *haitari) became hiaz (< *hē2t < *heir;14 cf. OE hēt, ON hét). The vowel *-ē2- was also consistently introduced into the preterite of verbs with ē-vocalism (cf. OHG liaz, OE lēt, ON lét (: PGmc. *lētanan)); the same process took place dialectally in verbs of the structural type *halđan (cf. OHG hialt vs. ON helt (*-ě-, OE hēold (*-eu-)). The genesis of these new preterites is hotly debated. It seems likely, however, that their nucleus was supplied by abridged reduplicated forms (cf. OE wēpan « cry », pret. wēop < *weup < *wew (ō) p, like heht < *heh (ai) t) or, alternatively, by reduplicated forms with contraction (cf. ON auka « increase », pret. jók < *euk < *e-auk (= Go. -aiauk)). The hypothesis of a source distinct from the reduplicated perfect, much favored in the older literature, is hardly likely.

3.2.2. Weak verbs

  • 15 The Proto-Germanic status of the Gothic fourth class (type ga-fullnan « be fulfilled », pret. ga-fu (...)

47Like every IE language, Proto-Germanic had a variety of derived present types, including causatives, iteratives, duratives and other semantically marked formations. These went back to well-known prototypes in the parent language, from which they inherited the property of being exclusively presential, i.e., of lacking associated aorist and perfect stems. Germanic was thus unable to furnish ablauting or reduplicated preterites to its secondary presents. Instead, it equipped them with preterites of a new type, the chief characteristic of which was a dental consonant, usually *-đ-. Such preterites, and the verbs that formed them, are traditionally called “weak”. Three main classes are recognized15:

48Class I: the present stem originally contained the thematic suffix *-éie/o-, which gave first *-iji-/*-ija- and then, by a kind of converse of Sievers’ Law (cf. p. 266), *-ji-/*-ja- after light syllables. In a small minority of cases, which may once have been strong (e.g., *sōk (i) janan « seek », Go. sokjari), the present suffix was *-ḭe/o-, which merged with *-éie/o- by becoming *iji-/*-ija- after heavy syllables. The difference between the two treatments is clearest in Gothic: cf. nasjan « save », 3 sg. nasji] o vs. hausjan « hear », 3 sg. hausei] O, like harjis vs. hairdeis. North and West Germanic lost *-j- before *-i-; instances of *-j- that remained caused gemination in West Germanic (cf. OS tellian « say » vs. 3 sg. teliƌ, OHG zellen, zelit). The present inflection was otherwise virtually identical with that of strong verbs. The nucleus of class I consists of PIE causatives, iteratives and (usually factitive) denominatives (cf., e.g., Go. satjan, OS settian, OE settan « set, cause to sit » (= Skt. sādayati), Go. fulljan, OE fyllan « fill » (: PGmc. *fullaz « full »).

49Class II: the original suffix was PIE *-āḭe/o- (< *-eh2-ḭe/o-), which contracted to *-ō- in Proto-Germanic. From a descriptive point of view, PGmc. *-ō- thus replaced the thematic vowel in all present forms (cf. Go. salbo, -os, -o] o, etc., opt. 3 sg. salbo, inf. salbon « anoint »). In Old English the a-variant of the thematic vowel was restored through the interposition of a secondary *-j-, borrowed from class I (cf. OE 3 sg. sealfap < *-ō] o vs. 1 sg. sealfie < *-ōju, inf. sealfian < *-ōjari). Old High German and Old Saxon remade the 1 sg. in PGmc. *-ō to -ōm (> OS -on) on the model of the irregular West Germanic verb *dōn « do, put », where the 1 sg. *dōm (OHG tuom, OE dōm), represented pre-Gmc. *dhedhōmi (earlier *dhedhēmi). Class II verbs, like their Latin counterparts of the first conjugation, continued a variety of inherited formations, with denominatives greatly preponderant (cf. OHG OE sealf, OHG salba (ōstem) « ointment »).

50Class III: the recurring formal characteristic is a 3 sg. in *-ai/o, with an associated 2 sg. in *-ais and 2 pl. in *-ai] o (cf. Go. habai] o « has », OHG habēt, OE hæfe] o; ON vakir « is awake »). Otherwise the individual languages show little agreement. Old High German has a regular paradigm, doubtless secondary, in -ē- (3 pl. habēnt, inf. habēn, etc.); Gothic and (usually) Old Norse intersperse forms in *-ai- with forms in *-a- (Go. 3 pl. haband, inf. haban; ON 3 pl., inf. vakdy); Old English and Old Saxon, where the type is vestigial, have *-ja- where Gothic has *-a- (OE 3 pl. habba/o, inf. habban, OS habbiad, habbian). Most class III verbs are stative or inchoative, and thus invite comparison with the Latin statives in – ēre (e.g., tacēre « be silent ») and Balto-Slavic statives in *-ē- (e.g., Lith. budėti « be awake »). Traditional opinion favors a derivation from PIE *-ē- (< *-eh1-), *-ēie/o- or *-aḭe/o- (< *-h1-ḭe/o-), none of which, however, can easily account for the attested forms. An alternative view takes the 3 sg in *-ai] o from a 3 sg. middle in *-ai < PIE *-oi (cf. Skt. duhé “milks”), with *–] o added from the other present classes.

51The past participle of weak verbs was made with the PIE suffix *-tó-. In the case of a few primary verbs this morpheme was added directly to the root (cf., e.g., ON sōttr « sought », OHG (gi) suoht, OE sōht < PGmc. *sōhtaz); normally, however, it was substituted for *-ḭe/o- in the pre-Germanic present stem (cf. Go. nasi/os, OHG (gi) nerit, OE nered < PGmc. *naziđaz < *nose-tó-s, like Lat. monitus « warned » < *mone-to-; Go. salbo/os, OHG (gi) salbōt, OE sealfad < *salđōdaz < *solpā-tó-s, like Lat. cantātus « sung »). Class III verbs interpose *-ai- before the dental in Gothic and Old High German (habai] Os, (gi) habēt « had »), but OS (gi) habd and OE hcefd (< *habdaz) probably preserve an earlier state of affairs.

52The origin of the Germanic weak preterite is perhaps the most thoroughly studied morphological problem in any branch of the IE family. There is no generally accepted explanation of this category. In its Proto-Germanic form, the weak preterite was made by adding endings, wholly or partly distinctive, to a stem ending in a dental and closely resembling the stem of the past participle (*sōht-, *naziđ-, *salƀōđ-, *haƀđ-). The reconstruction of the endings is problematic. In the plural and optative most of the languages have the same terminations as in the strong preterite (cf. e.g., ON sóttum, -uð, -u, opt. 3 sg. sótti; OHG suohtum, -at, -tin, opt. 3 sg. suohti). Gothic, however, inserts an element -ed- before the normal endings, producing forms in – (d) edum, – (d) eduf, – (d) edun, – (d) edi, etc., the antiquity of which is disputed. Of the singular paradigm, only the first person in *– (đ) ōn can be reconstructed with certainty (cf. Go. – da, Prim. Norse -do (tawido « I made »; later ON -ða), OHG -ta, -de). The 3 sg. is often assumed to have anded in PGmc. *– (ƌ) ē, largely on the strength of Prim. Norse -de (tawide « (he) made », later ON -ði), but this is phonologically doubtful and contradicted by the still earlier runic form talgidai « (he) carved », apparently < PGmc. *-dai. The 2 sg. forms are flatly incompatible: Gothic has – (d) es, Old High German and one dialect of Old Saxon point to *– (đ) ōs (OHG -tōs), while the other languages favor *– (đ) ais (ON -δir, OE -des, OS (one dialect) -des, -đas). Many attempts have been made to interpret these facts. Adherents of the “composition theory” have explained the weak preterite as a periphrastic formation containing the verb “to do” (WGmc. *dōn). The strongest argument in favor of this analysis comes from the Gothic forms in -dedum, etc., which recall OHG tātum, -ut, -un « we (you, they) did »; the strongest argument against it comes from the vocalism of the singular endings, and of the 3 sg. in particular. Other, less attractive periphrastic explanations have been proposed, some involving the past participle itself. Probably the most plausible alternative to the do-theory remains the view, due to Collitz, that takes the weak preterite from a nucleus of inherited perfect middles in 3 sg. *– (đ) ai < pre-Gmc. *– (t) oi.

53Among the scattered archaisms that survived the restructuring of the PIE verbal system in Germanic, special mention must be made of the “preterito-presents”, a small but important Proto-Germanic class that partook of the properties of both strong and weak verbs. These were in origin PIE or pre-Germanic perfects which resisted the normal semantic evolution of the perfect and retained their value as stative presents. From a Germanic point of view, the synchronic presents of the preterito-presents were thus formally similar to strong preterites, while their synchronic preterites and past participles were weak. More than a dozen such verbs occur in Gothic, including witan « know » (1 sg. wait: pl. witum, pret. wissa; cf. Skt. véda: vidmá; Gk. oîda: ídmeri), ga-daursan « dare » (gadars: ga-daursum, pret. ga-daursta; cf. Skt. dadhárṣa: dadhṛṣmá) and munan « think » (man: munum, pret. munda; cf. Gk. mémona: mémameri). Comparable forms occur in the other early dialects, but like ablauting verbs generally, the preterito-presents tended to decline in the later history of Germanic. In the modem languages they survive mainly as auxiliaries (cf. Eng. shall, can, may; NHG sollen, können, mögeri), joined in this function by another “irregular” verb, the old athematic present optative *wilῑ- « wish, want » (Go. 3 sg. will, 1 pl. wileima; cf. Eng. will, NHG will, wollen).

3.3. Syntax

54The syntax of Proto-Germanic was typologically similar to that of the other early IE languages, especially Italic, Greek and Balto-Slavic. It will suffice here to note a few special features.

55Agreement of adjectives with their head nouns, and pronouns with their antecedents, in number, gender and case was a general IE feature, but it was a peculiarity of Proto-Germanic that plural noun phrases of mixed gender induced neuter agreement (cf., e.g., Go. jah qemun] oan a] oei is jah bro] orjus is jah uta standandona [nt. pl.] insandidedun du imma, haitandona [nt. pl.] ina « and his mother and brothers came and, standing outside, sent in to him, calling him » (Mk. 3.31)). This rule had its origin in the accidental merger and confusion of the old masculine dual ending *– ō < PIE *-o-h1, which became obsolete in Germanic, with the productive neuter plural ending *-ō < *-ā < *-e-h2 .

56Germanic was relatively conservative in its treatment of the PIE prepositional adverbs (e.g. *pro « in front (of) », *eni « inside (of) », *ko (m) « together (with) », etc.). As in the classical languages, these had a twofold development, yielding prepositions on the one hand and proclitic verbal prefixes on the other. The univerbation of the new prefix + verb complexes, however, was not yet complete in Proto-Germanic, as can be seen from occasional instances of tmesis like Go. uz-uh-hof « and (-uh-) lifted up » (us-hafjan « lift up ») and ga-u-/va-se/vi “whether (-u-) he might see anything (–/va-)” (ga-sai/van « see »). Such cases of “Wackemagel’s Law” – the movement of enclitics to second position in the sentence – are wholly confined to Gothic. Old Norse abandoned the older type of verbal composition entirely, while the West Germanic languages carried the process of fusion to completion (cf. NHG erheben = Go. us-hajjari). All the Germanic languages show a later tendency to create new compound-like phrases ( “separable compounds”) using stressed adverbs of place or direction – the type of Eng. go in, ON ganga inn, NHG (hin) eingehen, etc.

57Proto-Germanic word order was “free” in the same sense as in other highly inflected IE languages; that is, a change in the order of the constituents in a given sentence often affected its emphasis or focus, but not its grammaticality. There is reason to believe that in the earliest Germanic dialects the basic, “unmarked” word order in both main and subordinate clauses was subject – object – verb (SOV), as is still the case in the earliest Primitive Norse inscriptions. Even in Proto-Germanic, however, main clauses freely employed the alternative order subject – verb – object (SVO) to focus attention on the direct object. SVO order predominates in main clauses in the early literary languages and their medieval and modem descendants, in some of which (e.g., English) it eventually spread to subordinate clauses (cf. The man saw the woman; when the man saw the woman...), while in others (notably German) it contrasted with SOV order in subordinate clauses (cf. Der Mann sah die Frau; als der Mann die Frau sah...). Relative clauses, marked only by a particle in Proto-Germanic (cf. Go. sa « he, that one », saei « (he) who »; ON : sá er; OHG der: der dar; OE se: se] Oe), patterned in this respect like other subordinate structures. Subject-verb inversion, generating sentences with VS (O) order, was obligatory in main clauses introduced by sentence-connective particles like ON /oā, OE and OHG da (cf. ON] oá gekk hann inn « then went he in », etc.). Such “verb second” or “V2” effects have been explained as an indirect reflex of Wackernagel’s Law, since finite verbs in main clauses were enclitic in Proto-Indo-European.

58The synchronic study of Germanic syntax – relativization, topicalization, inversion, and “movement” generally – is a subject of great current interest in theoretical linguistics. If it proves possible to advance from synchronic to diachronic insights, the rich evidence of the Germanic languages may yet play as important a role in the development of a mature theory of historical syntax as it once played, in the time of the Neogrammarians, in the development of a theory of historical phonology.

Bibliographie

Bibliographic

Bach A. (1965). – Geschichte der deutschen Sprache5 Heidelberg, Quelle und Meyer.

Bammesberger A. (1986). Der Aufbau des germanischen Verbalsystems, Heidelberg, C. Winter.

Baugh A. (1978). A history of the English language3, Englewood Cliffs, Prentice-Hall.

Fourquet J. (1938). – L’Ordre des elements de la phrase en germanique ancien, Paris, Belles Lettres.

– (1956). – Les Mutations consonantiques du germanique; essai de position des problemes, Paris, Belles Lettres.

Haugen E. (1976). – The Scandinavian languages: an introduction to their history, Cambridge, Mass, Harvard University Press.

Jolivet A. and Mossé F. (1947). – Manuel de l’allemand du Moyen Age, des origines au xive siècle, Grammaire. Textes. Glossaire. Nouvelle édition. Paris, Aubier.

Krahe H. (1966-1967). – Germanische Sprachwissenschaft 6 . I. Einleitung und Lautlehre. II. Formenlehre. III. Wortbildungslehre, von W. Meid, Berlin, De Gruyter.

Krause W. (1971). – Die Sprache der urnordischen Runeninschriften, Heidelberg, C. Winter.

Lehmann W. (1986). – A Gothic etymological dictionary, with bibliography prepared under the direction of Helen-Jo J. Hewitt, Leiden, E. J. Brill. Based on Sigmund Feist, Vergleichendes Wörterbuch der gotischen Sprache3 (1939).

Lockwood W. B. (1965). – An informal history of the German language; with chapters on Dutch and Afrikaans, Frisian and Yiddish, Cambridge, England, W. Heffer.

Meillet A. (1930). – Caractères généraux des langues germaniques4, Paris, Hachette.

Mosse F. (1958). – Esquisse d’une histoire de la langue anglaise2, Lyon: I. A. C.

– (1942). – Manuel de la langue gothique, Grammaire. Textes. Glossaire, Paris, Aubier.

Nielsen H. (1989). – The Germanic Languages – Origins and Early Dialectal Interrelationships, Tuscaloosa, London, University of Alabama Press.

Ramat P. (1981). – Einführung in das Germanische, Tübingen, Niemeyer.

Seebold E. (1970). – Vergleichendes und etymologisch.es Wörterbuch der germanischen starken Verben, The Hague, paris, Mouton.

Weerman C. (1989). – The V2 conspiracy: a synchronic and a diachronic analysis of verbal positions in Germanic languages, Dordrecht, Providence, Foris.

Wessen E. (1968). – Die nordischen Sprachen, Berlin, De Gruyter.

Notes

1 Another descendant of Old Norwegian is Faroese, spoken by about 43,000 people in the Faroe Islands.

2 Here too belongs Afrikaans, spoken in the Republic of South Africa by the descendants of the original Dutch settlers.

3 An aberrant (dialectal?) development of to *b seems to underlie Eng. bid (Go. bidjan; PIE *g*uhedh- « ask ») and bane (OE bana, OHG bano « slayer »; PIE *g*µhen- « slay »).

4 A possible exception is to be seen in forms like OE naca, ON nǫkkvi “boat” (< PIE *neh2ṷ-ón-?) and Go. unkis “us two (dat. acc.)” (< *unkwis < PIE *ṇh3-ṷé?), where *h2 and *h3 seem to have given PGmc. *k before *ṷ when the accent followed.

5 It is here assumed that *kw, *hw and *gw, the reflexes of the PIE labiovelars, patterned phonologically as clusters in Proto-Germanic.

6 Here and below, “PIE *a” and “PIE *o” refer indifferently to “true” *a and *o and to the secondary *a and *o that resulted from the coloration of *e by a neighboring laryngeal. Similarly, the “long vowels” are loosely understood to include etymological short vowels that were compensatorily lengthened when laryngeals disappeared in early pre-Germanic.

7 The corresponding Gothic form, haurn, is ambiguous, since PGmc. *u would regularly have given au ( [ͻ]) in this position. It is possible that a-umlaut was a common North and West Germanic innovation not shared by Gothic.

8 The Neogrammarian derivation of PGmc. *hē2r from PIE * k̂ēi-r is almost certainly incorrect; see further p. 274.

9 It will be noted from this example that the “long” syllabic sonorants *, *, *ṃ and * fell together with their short counterparts in Germanic.

10 The precise environments for this rule were complex and varied considerably from dialect to dialect.

11 The PIE instr, pl. of o-stems ended in *-ōis; PGmc. *-amiz was created on the model of the other stem-classes.

12 Comparative adjectives, made with the zero-grade of the PIE suffix *-ḭos-/ *-ḭes-/ *-is-, lacked strong forms entirely (cf. Go. batiza « better », ON betri, OHG bezziro). Superlatives had the usual twofold declension.

13 Gothic has a 3 sg. impv. in -adau and a 3 pl. impv. in -andau, the exact relationship of which to the corresponding forms in Sanskrit and Hittite remains to be clarified.

14 The “neo-diphthong” *ei is not, of course, to be confused with PIE *ei, which yielded in Germanic.

15 The Proto-Germanic status of the Gothic fourth class (type ga-fullnan « be fulfilled », pret. ga-fullnoda) is uncertain.

Auteur

Cornell University, New York, États-Unis

Le texte et les autres éléments (illustrations, fichiers annexes importés) sont sous Licence OpenEdition Books, sauf mention contraire.

Cette publication numérique est issue d’un traitement automatique par reconnaissance optique de caractères.
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search