On the origins of Germanic heroic poetry: a case study of the legend of the Burgundians

ON THE ORIGINS OF GERMANIC HEROIC POETRY A Case Study of the Legend of the Burgundians* The following paper attempts to reassess the value of current theories regarding the origins of heroic poetry in the Germanic languages. It is based on a thorough consideration of the evidence for heroic poetry and historical or memorial traditions in the early middle ages, the period when this poetry are supposed to have originated. I differentiate between three stages of reception for the extant legends: the original audience, the peoples whose histories the legends purport to tell and among whom they might have been composed; the various societies that kept these legends alive in the centuries between postulated original composition and extant textual record; and finally the audience of the existing texts. This paper is concerned with the first two levels of reception. I begin by examining whether there is indeed evidence that the legends were composed at some point during the so-called migration period among the peoples whose his- tories they tell, and what function they might have had in this first stage. This is followed by an examination of the possible causes for and func- tions of the intermediate transmission, which doubtless took place at least partly amongst societies not directly linked to the originators of the leg- ends, if the latter did indeed originate in historical fact as a form of oral history. Based on a reassessment of the historical evidence, I shall argue that the extant works in Germanic languages have no value for the under- standing of any previous stage of oral culture; they do, however, provide evidence for the existence of this culture, and for a (very limited) continu- ity within the oral traditions in the Germanic vernaculars of the middle ages. I conclude with some suggestions that might, I hope, help lay the foundations for more historically plausible theories regarding the form, function and origins of heroic poetry in Germanic languages than those that are current in literary scholarship. The legend of the fall of the Burgundians will be examined in detail with the example of the Old Norse ›Atlaqviða‹, the sole witness to this * I would like to express my profuse thanks to Andy Orchard for his help with this paper. I am also deeply indebted to Nick Everett and Sandy Murray for their many criticisms. Thanks are also due to Hartwig Mayer and Markus Stock. Remaining infelicities remain my sole responsibility. Brought to you by | University of Toronto-Ocul Authenticated Download Date | 7/27/15 9:31 PM  ON THE ORIGINS OF GERMANIC HEROIC POETRY 221 legend in any vernacular that can be read as a completely independent artefact, unconnected with the story of the dragon-slayer (Sigurð/Sıˆvrıˆt) or that of Ermaneric’s death. It should be made clear at the outset that this paper is by no means an interpretation of any specific text; the ›Atlaqviða‹ provides a convenient example against which the historical evidence may be compared. I The ›Atlaqviða‹ may be summarised as follows:1 Atli, the Hunnish king, sends a messenger to Gunnar, whose hall is in the courts of Giu´ci, to invite Gunnar to visit Atli at the stað i Danpar (›banks of the Dnieper‹) and receive gifts from him. Gunnar asks the advice of Ho§ gni, stating that he cannot conceive of treasures greater than what they already possess. Ho§ gni indicates the wolf’s hair wound in the ring sent by Guðru´n, and concludes that it is meant as a warning. Although no one urges Gunnar on, he, sem konungr scyldi (›just as a king ought‹), announces his intention to travel to Atli. Before leaving, he states that the arfr Niflungar (›inheritance of the Niflungs‹) will be ruled by wolves if he does not return. Gunnar and his companions go to Atli’s court, which is guarded by men apparently expecting a fight. Gunnar’s sister Guðru´n sees them first, and, indicating through her words that they are unarmed, says that they should not have come, and that the snake-pit awaits them. Gunnar says it is too late to look for an army from the Rosmofioll Rı´nar (›red cliffs of the Rhine‹). The Huns, vinir [following the manuscript] Borgunda (›relatives of the Burgundi- ans‹), put Gunnar in fetters, apparently without a fight; Ho§ gni fights but is also captured. Atli offers to ransom Gunnar for gold, but Gunnar (called geir-Niflungr, ›spear-Niflung‹), refuses. He alone has the secret of the hodd Niflunga (›treasure of the Niflungs‹), and before he is put to death, he says that Rı´n scal ra´ð a (›the Rhine shall rule‹) the inherit- ance of the Niflungs. Following the deaths of Ho§ gni and Gunnar, amidst much drinking of ale in the hall, Guðru´n kills her sons by Atli and feeds them to Atli, after which she kills him. The last lines state that she brought death to three þio´dkonunga (›kings of peoples‹) before her own death. A Burgundian settlement existed along the Rhine in the early fifth century, and a Burgundian king, Gundicharius, is attested as having fought against 1 The edition used is: Edda. Die Lieder des Codex Regius nebst verwandten Denkmälern, vol. I: Text, ed. by G. Neckel, 5th ed. by H. Kuhn, Heidelberg 1983 (Germanische Bibliothek Reihe 4). For orientation on the poems of the ›Elder Edda‹ manuscript, see J. Harris, Eddic poetry, in: Old Norse-Icelandic literature. A critical guide, ed. by C. J. Clover and J. Lindow, Ithaca 1985 (Islandica 45), pp. 68Ð156. See also the introduction and commentary in U. Dronke, The Poetic Edda, vol. I: Heroic poems, Oxford 1969. All translations are mine. Brought to you by | University of Toronto-Ocul Authenticated Download Date | 7/27/15 9:31 PM 222 SHAMI GHOSH the Huns in the 430s. It is not completely clear whether the Huns con- quered the Burgundians or whether this was done by a (possibly) Hunnish army acting under or for Aetius. The testimony of the earliest sources (from the mid fifth to the early sixth century) is as follows:2 Eodem tempore Gundicharium Burgundionum regem intra Gallias habitantem Aetius bello obtrivit pacemque ei supplicanti dedit, qua non diu potitus est, siquidem illum Chuni cum populo suo ab stirpe deleverint (›At the same time Aetius conquered Gundicharius, the Bur- gundian king living in Gaul, in war, and granted him peace, for which he was pleading. He did not enjoy this peace for long, since the Huns destroyed him and his people root and branch‹);3 Burgundiones qui rebellauerant a Romanis duce Etio debellantur (›The Burgundians who had rebelled were destroyed by the Romans under the command of Aetius‹); Aetio duce et magistro militum Burgundionum caesa XX milia (›Twenty thousand soldiers of the Burgundians were cut to pieces by Aetius, the general and master of soldiers‹);4 Bellum contra Burgun- dionum gentem memorabile exarsit, quo universa / paene gens cum rege per Aetium | deleta (›A remarkable war raged against the Burgun- dian people, in which almost the whole people along with their king was destroyed by Aetius‹);5 Burgundiones victi ab Aezio patricio (›The Burgundians were conquered by Aetius the patrician‹).6 The ›Lex Burgundionum‹ (510–520) records Gibica, Gundomar, Gislaha- rius and Gundaharius as the ancestors of the Burgundian king Gundobad 2 The principal historical sources are conveniently assembled and analysed by Dronke [n. 1], pp. 29Ð36, esp. pp. 34Ð36; on the historical background, see I. N. Wood, Gentes, kings and kingdoms Ð the emergence of states. The kingdom of the Gibichungs, in: Regna and gentes. The relationship between late antique and early medieval peoples and kingdoms in the transformation of the Roman world, ed. by H.-W. Goetz, J. Jarnut and W. Pohl, Leiden [et al.] 2003 (Transformation of the Roman world 13), pp. 243Ð269. Cf. also idem, Ethnicity and the ethnogen- esis of the Burgundians, in: Typen der Ethnogenese unter besonderer Berücksichtigung der Bayern, vol. I, ed. by H. Wolfram and W. Pohl, Vienna 1990 (Veröffentlichungen der Kommission für Frühmittelalterforschung 12), pp. 53Ð 69; K. F. Stroheker, Studien zu den historisch-geographischen Grundlagen der Nibelungendichtung, in: DVjs 32 (1958), pp. 216Ð240. A good synthesis of the Burgundians’ history is provided by R. Kaiser, Die Burgunder, Stuttgart 2004 (Kohlhammer Urban-Taschenbücher 586). 3 Prosperi Tironis Epitoma Chronicon, in: Chronica minora saec. IV. V. VI. VII, vol. 1, ed. by T. Mommsen, Hanover 1892 (MGH AA 9), pp. 341Ð349, 1322. 4 Hydatii Limici Chronica Subdita, in: The Chronicle of Hydatius and the Consula- ria Constantinopolitana. Two contemporary accounts of the Roman Empire, ed. by R. W. Burgess, Oxford 1993, pp. 1Ð172: 99; 102. 5 The Gallic Chronicle of 452: a new critical edition with a brief introduction, ed. by R. W. Burgess, in: Society and culture in late antique Gaul: revisiting the sources, ed. by R. W. Mathisen and D. Shanzer, Aldershot 2001, pp. 52Ð84, p. 79. 6 The Gallic Chronicle of 511: a new critical edition with a brief introduction, ed. by R. W. Burgess, in: Society and culture [n. 5], pp. 85Ð100: 97. Brought to you by | University of Toronto-Ocul Authenticated Download Date | 7/27/15 9:31 PM  ON THE ORIGINS OF GERMANIC HEROIC POETRY 223 (474Ð516).7 This Gundaharius is generally accepted as being identical with the Gundicharius named by Prosper Tiro, and as the historical ante- cedent of the Gunnar/Gunther figure in the extant versions of the legend. Gibica is thought to be the same as the Old Norse Giu´ci. The ›Nibelungen- lied‹ has a Giselher (Gislaharius?), brother to Gunther (Gunnar); other lays in the eddic corpus refer to a brother called Guthorm/Gothorm (Gun- domarius?).8 The relationship of the Burgundians to the Huns is less straightforward. Atli corresponds to the Hunnish king Attila, who is not attested in histori- cal sources as playing a part in the destruction of the Burgundians. He died in his bed around 453. Conflicting accounts state that he died of a nosebleed or by the hand of a woman. At any rate, there is drinking, blood, and a woman present at his death: Attila rex Hunnorum Europae orbator provinciae noctu mulieris manu cultroque confoditur. quidam vero sanguinis reiectione neca- tum perhibent (›Attila, king of the Huns and the bane of the European region, is pierced at night by the hand and knife of a woman. Some however say that he died by the throwing up of blood‹);9 Qui, ut Priscus istoricus refert, exitus sui tempore puellam Ildico nomine decoram valde sibi in matrimonio [. . .] socians eiusque in nuptiis hilaritate nimia resolutus, vino somnoque gravatus resupinus iaceret, redun- dans sanguis, qui ei solite de naribus effluebat, dum consuetis meati- bus impeditur, itinere ferali faucibus illapsus extinxit [. . .] sequenti vero luce cum magna pars diei fuisset exempta, ministri regii triste aliquid suspicantes post clamores maximos fores effringunt inveni- untque Attilae sine ullo vulnere necem sanguinis effusione peractam puellamque demisso vultu sub velamine lacrimantem (›Who [i.e. At- tila], as the historian Priscus relates, at the end of his days took in marriage [. . .] an extremely pretty girl with the name of Ildico;10 and exceedingly exhausted in the nuptial celebrations and weighed down by wine and sleep, as he lay on his back, the blood that normally flowed out of his nostrils choked him, since its accustomed passage was im- peded, and having flowed on a deadly path through his throat, killed 7 Liber Constitutionum sive lex Gundobada, in: Leges Burgundionum, ed. by L. R. von Salis, Hanover 1892 (MGH LL II/1), pp. 29Ð122, III, p. 43: [. . .] regiae memo- riae auctores nostros, id est: Gibicam, Gundomarem, Gislaharium, Gunda- harium (›[. . .] our ancestors of royal memory, that is: Gibica, Gundomar, Gisla- harius, Gundaharius‹). 8 ›Sigurðarqviða in scamma‹, 20, 1; 22, 5; ›Brot af Sigurðarqviðu‹, 4, 3. 9 Marcellini v. c. comitis chronicon in: Chronica minora saec. IV. V. VI. VII, vol. 2, ed. by T. Mommsen, Berlin 1894 (MGH AA 11/2), pp. 37Ð108 at p. 86. 10 Although the name might be Germanic, speculations on her potential relation- ship to the Guðru´n of legend have no basis. For less sceptical views, cf. e. g. Dronke [n. 1], p. 32 sq.; W. Haubrichs, Die Anfänge: Versuche volkssprachlicher Schriftlichkeit im frühen Mittelalter, 2., durchgesehene Auflage, Tübingen 1995 (Geschichte der deutschen Literatur von den Anfängen bis zum Beginn der Neu- zeit I/1), pp. 95 sq. Brought to you by | University of Toronto-Ocul Authenticated Download Date | 7/27/15 9:31 PM 224 SHAMI GHOSH him [. . .] And the following morning, when the better part of the day had passed, the attendants of the king, sad and suspecting that some- thing was wrong, after calling out loudly, broke down the doors and found the slaughter of Attila had been accomplished by the effusion of blood without any wound, and the girl was weeping under the covers with a dejected face‹).11 These sources were composed about a century after the events; Jordanes, however, claims to be following Priscus, who was contemporary. Two later texts, Poeta Saxo’s narrative on Charle- magne (9th century), and the ›Quedlinburg Annals‹ (11th century), pro- vide somewhat embellished versions of the story: Poeta Saxo adds to what we know from the earlier texts that the woman (here unnamed) killed Attila in revenge for the murder of her father; the annals state further that she was taken by force after her father was killed.12 Be- cause of their late date, they are of less relevance to our purposes, but we should note that they might furnish evidence of an oral tradition regarding the death of Attila intruding into Latin, literate culture; they could, however, also have been influenced by Paul the Deacon’s account of Alboin. It has been suggested that Attila’s wife was of Burgundian origin, and the finds of artificially elongated skulls in graves along the Rhine and in south- eastern Gaul in the Rhoˆne valley have, on somewhat dubious grounds, been taken as evidence of cultural links between the Huns and Burgundi- ans, including intermarriage.13 Beyond these few grave-finds, the only con- temporary indication of any cooperation between Huns and Burgundians comes from Sidonius Apollinaris, who states that the Hunnish army that fought against Aetius under Attila at Chaˆlons in 451 included Burgundi- ans.14 Some Burgundians are mentioned as joining the settlement in Sa- paudia later in the additions to the ›Liber Constitutionum‹, but Stroheker’s assertion that these had lived long under Hunnish rule cannot be substan- tiated. That they might have brought with them material on the death of Attila, which then became a part of the legend of the Rhine catastrophe, is therefore questionable.15 Whether either these Burgundians or those 11 Jordanis Romana et Getica. De origine actibusque Getarum, ed. by T. Mommsen, Berlin 1882 (MGH AA 5/1), c. 49 at pp. 123 sq. 12 Poetae Saxonis annalium de gestis Caroli magni imperatoris libri quinque, in: Poetae latini aevi Carolini, vol. 4/1, ed. by P. von Winterfeld, Hanover 1899 (MGH Poet. IV/1), pp. 1Ð71 at III, ll. 25Ð34 (l. 17 might suggest that the events de- scribed are a part of a living memorial tradition); Die Annales Quedlinburgenses, ed. by M. Giese, Hanover 2004 (MGH SRG 72), p. 415. 13 Dronke [n. 1], p. 30; Stroheker [n. 2], pp. 216 sq.; 224. 14 Gai Sollii, Apollinaris Sidonii, Epistolae et Carmina, ed. by C. Lütjohann, Berlin 1887 (MGH AA 8), pp. 173Ð264, Carm. VII, ll. 316Ð356: the Huns invade Gaul along with other barbarian groups, among whom ›the Burgundian urges on the Scirian‹ (Scyrum Burgundio cogit, l. 322). 15 Leges Burgundionum [n. 7], ›constitutio extravagans‹, § 21, 12: a Burgundioni- bus, qui infra venerunt (›from the Burgundians, who came later‹); Stroheker [n. 2], pp. 229 sq.; similarly Kaiser [n. 2], p. 201; cf. Wood, Gibichungs [n. 2], p. 260. Brought to you by | University of Toronto-Ocul Authenticated Download Date | 7/27/15 9:31 PM  ON THE ORIGINS OF GERMANIC HEROIC POETRY 225 mentioned by Sidonius Apollinaris have anything to do with the descend- ants of those defeated in 436 is debatable, as is the relationship of the (supposed) Burgundians with elongated skulls on the Rhine to those in southern Gaul. The death of Attila as recorded in the legend has dubious historical foundations: there is no evidence of any marital relationships between the Huns and the Burgundians, even if the some groups of the latter might have been a part of a Hunnish army; and there is no contem- porary or near-contemporary source that provides a narrative of Attila’s death sufficiently close to that found in the later material. From the testi- mony of the Middle High German ›Nibelungenlied‹, Poeta Saxo, and the ›Quedlinburg Annals‹, it is apparent that Attila’s death might not even have been a fundamental part of the Burgundian legend, and given the lack of historical evidence, it is not considered here in further detail. Comparing the contemporary or near-contemporary historical evidence with the narrative of the ›Atlaqviða‹, we find that there are four important details that the lay has in common with at least one early source (though not necessarily all these details are contained in any single contemporary record). First, the names of the kings are or could be historical, bearing significant similarities to the latinized forms recorded in the contemporary sources. Second, these kings are said to be Burgundian; the names of specific kings are, in other words, associated with a specific ›ethnic‹ desig- nation, which is in turn linked to a specific geographical area. Third, the Gibichung or Gju´kungar kings of the Burgundians from somewhere along the Rhine are said to have fought against the Huns (thus bringing in an- other ethnic designation). The last common detail is that this battle leads to the destruction of the Burgundians. All these details are also present in the other extant vernacular versions of the legend. Although this might not seem like much Ð and I stress that none of the later Germanic material provides any information of historical value not contained in some contemporary source Ð it is, I believe, more historical fact than is common for most of the corpus of Germanic heroic litera- ture.16 Given that the extant versions of our legend, dating from roughly eight or nine centuries after the events, are built around a core of histori- cal fact, it seems reasonable to accept that this historical core reached the composers of the extant poems in some form, whether oral or written; in other words, not all that we have in the extant monuments was purely poetic creation. It remains, then, to determine how far back in time we 16 That relatively little historical information is contained in heroic legends was pointed out already by A. Heusler, who also stressed, correctly, that the plot of the extant legends can have no real explanation in history. See his Geschichtli- ches und Mythisches in der germanischen Heldensage [originally pub. 1909], in his: Kleine Schriften, ed. by S. Sonderegger, Berlin, 1969, vol. 2, pp. 495Ð517. Brought to you by | University of Toronto-Ocul Authenticated Download Date | 7/27/15 9:31 PM 226 SHAMI GHOSH may trace the possibility of a poetic form being given to our core historical facts. A group of Burgundians was settled in south-eastern Gaul in the second half of the fifth century by Rome.17 This settlement grew into a fairly prominent kingdom, and the Burgundian kings were, by the end of the century, heavily involved in imperial politics at Rome, and patrons to members of the senatorial aristocracy of southern Gaul.18 It is generally accepted in literary scholarship that the destruction of the Rhine kingdom was commemorated through song, most probably by the surviving Bur- gundians, following a tradition of oral heroic poetry thought to be current among the Germanic tribes during this period (if not already several cen- turies before). This is thought to be the origin of the legend as we have it in the eddic poems, and with some (often significant) variations, in the ›Vo§ lsunga saga‹ and the ›Nibelungenlied‹.19 How plausible is this assumption? What is the evidence that there was any surviving oral tradition in the later Burgundian kingdom that com- memorated the destruction of the kingdom of Gundaharius? Before at- tempting to deal with this question, we must make a digression to another one, fundamental to the subject of this paper: what was ›Burgundian iden- tity‹? The evidence seems to show that the Burgundian kings in Savoy understood themselves as Roman officials (magistri militum; patricii) as well as reges gentium;20 their regnum consisted not only of members of a gens Burgundionum, but also of the Gallo-Roman populace of the region over which they ruled. It is extremely unclear to what extent they differentiated between their roles as rulers of their own gens and regents of a ›Roman‹ populace, and, at least by the time of the codification of their laws, it appears that the differences were not held to be essential in 17 Chronicle of 452 [n. 5], p. 80: Sapaudia Burgundionum reliquiis datur cum indigenis diuidenda (›Sapaudia is given to those left of the Burgundians to be shared with the native people‹); cf. Wood, Gibichungs [n. 2], p. 246. It is impor- tant to note that the settlement was probably not a single event, but an ongoing process over several decades; this instability might well have had consequences for the sense of identity (or lack thereof) among the settlers. It is quite possible that people settling in Sapaudia at different times might have had somewhat different traditions (cf. Wood, Ethnicity [n. 2], pp. 65Ð69). 18 On the social structures of the Burgundian kingdom, cf. Kaiser [n. 2], pp. 38Ð56 and pp. 102Ð132, and esp. pp. 49Ð52 and pp. 121Ð124 on the kings’ involvement in Roman politics and their patronage of Romans respectively. 19 For the consensus of scholarship, see e. g. Dronke [n. 1], pp. 29Ð31; 40Ð42; Hau- brichs [n. 10], p. 92. 20 Cf. P. Amory, Names, ethnic identity, and community in fifth- and sixth-century Burgundy, in: Viator 25 (1994), pp. 1Ð30 at pp. 11Ð12; Wood, Gibichungs [n. 2], p. 251; Kaiser [n. 2], pp. 49Ð52. Brought to you by | University of Toronto-Ocul Authenticated Download Date | 7/27/15 9:31 PM  ON THE ORIGINS OF GERMANIC HEROIC POETRY 227 constituting the regnum.21 From the career of Gundobad, it is apparent that at least he and his successors were heavily romanised,22 and the extent to which they maintained any kind of Germanic culture (especially in the linguistic sense of the term) is debatable. There is little evidence to suggest that the kings or their advisors in lawmaking were concerned to stress the identity of the kings as specifi- cally Burgundian in any ethnic or even cultural sense, and while the army probably consisted primarily of descendants of pre-settlement Burgundi- ans, the kings probably did not see themselves as having a distinct ›ethnic‹ identity, separate from that of those they ruled. The Burgundian royalty definitely exerted much effort to ›romanise‹ themselves, and it seems hasty to assume that there could have been a conflict between ›Germanic‹ and ›Roman‹ traditions or ethnic identities (although there were some variances in shades of Burgundian identity among those ruled by the Bur- gundian kings).23 Based on the evidence, all we can safely say is that ›Burgundian‹ normally referred to a political (and sometimes legal) entity, with uncertain links to any ethnic or cultural traditions. This entity, more- over, was ruled by kings who adopted the Christian religion, Latin, and (at least to some extent) Roman law; it would appear hasty to assume too much adherence to ›Germanic‹ cultural identity in the face of the extant evidence. It is clear that group identity was a complex matter in this period, and that people and peoples might have subscribed to multiple and changing identities.24 The discourse on ethnic or national identity could have had 21 P. Amory, The meaning and purpose of ethnic terminology in the Burgundian laws, in: Early Medieval Europe 2 (1993), pp. 1Ð28 esp. at 8Ð10, 24Ð26; Wood, Gibichungs [n. 2], pp. 255Ð262. 22 Amory [n. 20], pp. 10Ð13; D. Shanzer, Two clocks and a wedding: Theoderic’s diplomatic relations with the Burgundians, in: Romanobarbarica 14 (1998), pp. 225Ð258. 23 Cf. Wood, Gibichungs [n. 2], pp. 260Ð264; Amory [n. 21], p. 24; Kaiser [n. 2], pp. 111Ð114; 126Ð135. 24 The literature on the subject is vast; see the most recent critiques collected in A. Gillett (ed.), On barbarian identity. Critical approaches to ethnicity in the early middle ages, Turnhout 2002 (Studies in the early middle ages 4), especially the papers by Gillett and Kulikowski. Other recent works with contrasting views include Amory [nn. 20 and 21]; H.-W. Goetz, Regna and gentes: conclusion, in: Regna and gentes [n. 2], pp. 597Ð628; idem, Gens. Terminology and perception of the »Germanic« peoples from late antiquity to the early middle ages, in: The construction of communities in the early middle ages. Texts, resources and arte- facts, ed. by R. Corradini, M. Diesenberger and H. Reimitz, Leiden [et al.] 2003 (Transformation of the Roman world 12), pp. 39Ð64; W. Pohl, Tradition, Ethno- genese und literarische Gestaltung: eine Zwischenbilanz, in: Ethnogenese und Überlieferung. Angewandte Methoden der Frühmittelalterforschung, ed. by K. Brunner and B. Merta, Vienna 1994 (Veröffentlichungen des Instituts für österrei- chische Geschichtsforschung 31), pp. 9Ð26; Wood, Ethnicity [n. 2]. S. Reynolds, Brought to you by | University of Toronto-Ocul Authenticated Download Date | 7/27/15 9:31 PM 228 SHAMI GHOSH specific, normally political functions (though it did not necessarily always do so), which would not remain static over time; there should not, there- fore, be any assumption of an unchanging tradition any more than of an unchanging Germanic (or, for that matter, Burgundian) identity. It is also extremely unclear what value was placed on ethnic terms by early medie- val writers, what relationship the Latin terminology (for example, words such as gens, regnum, natio) might have had to the understandings of those who had no Latin, and whether, in fact, ethnic or national identity was considered to be of much importance among either the Latinate or non-Latinate people of the time; we must also remember that our sources reveal nothing about what the barbarians themselves thought about their own identity at this period. If a Burgundian oral tradition in a Germanic dialect survived, and if the legend we are concerned with here was a part of this tradition, it definitely coexisted with other discourses within the Burgundian cultural sphere, which were much influenced by Roman, Latin, and Christian tradition. That there appears to have been no official, recorded history that was anything like what is contained in the legend, and that the Burgundian kings Ð who do state that they are descendents of the kings of the leg- end Ð also identify themselves as a part of a Roman elite, might seem to indicate that the origins of this legend were probably not in the Burgun- dian royal family and its self-representation. That the kings seem to have made no attempt to discover or create a past, or to record genealogies as forms of legitimation, might also indicate a lack of interest in a specifically Burgundian tradition. The self-representation of the Burgundian kings ap- pears to have been more oriented towards projecting a ›Roman‹ rather than a Burgundian identity, though it is important to note that the evidence is largely e xt er na l self-representation; what form of identity formation was used by the royal family among other Burgundians is unknown. However, although our legend deals with kings, it is not necessary that any oral memorial tradition was the sole preserve of the royalty, nor is it indeed necessary that the royal family could not participate in multiple discourses of history and memory. While the royal family might have seen itself at least partly as Roman, it is quite possible that the rest of the Burgundians settled in Gaul propagated non-Roman traditions as a part of their cultural memory, especially if »soldiering in Burgundy as in the Our forefathers? Tribes, peoples, and nations in the historiography of the age of migrations, in: After Rome’s fall: narrators and sources of early medieval history (FS W. Goffart), ed. by A. C. Murray, Toronto 1998, pp. 17Ð36, presents a nu- anced and balanced perspective. Brought to you by | University of Toronto-Ocul Authenticated Download Date | 7/27/15 9:31 PM  ON THE ORIGINS OF GERMANIC HEROIC POETRY 229 late Empire was a hereditary profession«.25 This might not have been an appropriate form of royal self-representation in codified, Latin laws, which were aimed at the whole of the populace, but the kings might well have participated in this tradition as well. Alternatively, these traditions could have been maintained independent of royal self-representation; different levels of society in the Burgundian kingdom might have practiced different forms of memoria and identity formation.26 It may seem unlikely that a tradition of Germanic poetry reinforcing some kind of Burgundian identity was maintained when one considers the Latin sources of the period, but it should be noted that these sources are, firstly, written and in Latin, therefore inherently inaccessible to at least a sizable proportion of the settlers in the first decades of the settlements; and secondly, most extant sources that have anything to say about the Burgundians are forms of representation to those who are clearly not ›Burgundian‹: they are written either by or for outsiders, primarily ›Ro- man‹ outsiders. The laws are the exception to this; they do seem to make some distinction (however vague and flexible) between Burgundians and Romans. They are, however, not really a form of self-representation. The primary reference to what might be surviving oral tradition among the Burgundians comes from Sidonius Apollinaris, who mentions Burgundians singing in German. This is not sufficient evidence for the existence (or non-existence) of Burgundian historical poetry.27 What mode of identity formation was used by the social classes below the kings is not known; it seems indubitable that it was in the vernacular, and included an oral tradi- tion. That the extant sources make no mention of this means little: unless one assumes either that the Burgundians had no sense of history or that all records were lost, an oral tradition probably formed a part of their memorial practices. This tradition might also have included a lay concern- ing the destruction of the Rhine kingdom, though there is no reason to believe that beyond the four basic details enumerated above, any such 25 Amory [n. 20], p. 27; cf. also Amory [n. 21], p. 24; Wood, Gibichungs [n. 2], p. 262. We should note, though, that the Burgundians were apparently not solely a mili- tary group; cf. Kaiser [n. 2], pp. 31; 34 sq.; 138Ð142. 26 Cf. Reynolds [n. 24], pp. 22 sq. (emphasis added): »Medieval ideas of hierarchy and custom allowed for layers of political authority and community. Just as lords at every level had rights against their superiors, so each lord’s subjects could envisage themselves as a co mm un it y w it h i ts ow n r ig ht s a nd cu s- t om s within the wider community of the kingdom. People owed loyalties at each level, so that c on fl ic ts co ul d a ri se [. . .] between those owed to king- doms and those owed to the other communities to which people belonged«. 27 Sidonius Apollinaris [n. 14], Carm. XII, ll. 4Ð7. Kaiser [n. 2], p. 201, believes with- out evidence that the songs referred to are ›(Helden-)Lieder‹. Brought to you by | University of Toronto-Ocul Authenticated Download Date | 7/27/15 9:31 PM 230 SHAMI GHOSH ›original‹ lay had anything in common with extant Germanic poetry from later periods. From the preceding discussion, it will be apparent that any Burgundian memorial poetry that might have existed would probably have been, within three generations of the Rhine catastrophe, a part of a larger his- torical discourse embracing Latin, Roman identity and history as well. Ger- manic traditions were probably less important for the highly romanised ruling class than for other sections of the population. It also seems very unlikely, given the intermingling between the settlers and the native popu- lace in Gaul, that any memorial tradition of the earlier Rhine kingdom would have survived uncontaminated for more than two or three genera- tions at the most; the function Ð and therefore the content Ð of a lay commemorating the Rhine catastrophe would undoubtedly have been quite different in the context of the first generation of settlers (possibly survivors of the events of 436) and that of their descendants three or four generations later (with no link to the living memory of those events), who had now probably married into the local populace, adopted their (Romance) language, and were ruled by heavily romanised kings. We should note too that intermarriage between Romans and Burgundians was allowed, and Romans were also allowed to serve in the military;28 even specifically military traditions, therefore, are likely to have been rapidly romanised. The Burgundians eventually fade from historiography as a specifically distinguished people.29 The Burgundian kingdom became a part of the Frankish domain.30 Furthermore, although the Burgundians who settled in southern Gaul must have spoken a Germanic language, by the end of the eighth century at the latest, and probably far earlier, that region spoke some form of Romance vernacular. Given that the majority of the popula- tion in Sapaudia and along the Rhoˆne remained native Latin/Romance speakers after the settlement of the remnants of the Rhine Burgundians, and given the lack of laws preventing intermarriage, the move away from a Germanic language must have taken place quite early.31 I f our legend 28 Leges Burgundionum [n. 7], § 12, 5; § 100; Lex Romana Burgundionum, in: ibd., pp. 123Ð163, § 45, 3. 29 See Goetz, Terminology [n. 24], tables 1 and 2 on pp. 62Ð63. 30 On the complex historical events leading to the end of Burgundian independ- ence, see Kaiser [n. 2], pp. 57Ð74. We should note, however, that the Burgundian region was a self-conscious and quasi-independent entity under Frankish rule at least until around 700 (cf. Kaiser [n. 2], pp. 191Ð200); native traditions could potentially have flourished here, although it is unlikely that they would have done so in a Germanic language until such a late date. 31 Kaiser [n. 2], p. 100, suggests (on somewhat shaky evidence) ›sprachliche Unsi- cherheit‹ and ›sprachliche Assimilation‹ among the Burgundians as early as the late 5th century. Brought to you by | University of Toronto-Ocul Authenticated Download Date | 7/27/15 9:31 PM  ON THE ORIGINS OF GERMANIC HEROIC POETRY 231 originated among the Gallic Burgundians of the mid-fifth century, it must have transferred to some other Germanic-speaking group by the end of the eighth at the latest. The Franks are an obvious candidate; we shall now turn to the possibility that the legend might have been kept alive among the Franks, or perhaps even have originated among them. II That the Franks might have had something to do with the genesis of the Burgundian legend has been noted before.32 The earliest records of a cog- nate of niflungr as a place-name appear much later, but from the eighth century, Nibelunc is recorded as a personal name among Frankish fami- lies; whether the latter had any relationship with any Burgundians is un- known.33 The Burgundian king Sigismund’s cousin Clothild married the Merovingian king Clovis, and there was thus from a very early period a close link between the Burgundian and the Merovingian royal families; other prominent Burgundian and Frankish families were also soon linked by marriage. Under the Franks, the word Burgundia appears to be used solely as a territorial term. Fredegar (late seventh century) mentions Bur- gundaefarones, but he seems to use the term to denote the Burgundian aristocracy connected to the court of the (Merovingian) king, without nec- essarily any specifically ethnic connotations.34 A family of Burgundiofar- ones are recorded in the late ninth century ›Vita Faronis‹,35 but as the family concerned lived mainly in northern Gaul, it is unclear whether there is any link to the people mentioned by Fredegar.36 The Burgundian legend is closely connected, in all extant vernacular sources, with that of Brynhild and Sigurð. There was a historical Brune- childe, married to the Burgundian king Sigibert, who ruled Burgundy as regent for her grandson in the first decade of the sixth century.37 Fredegar 32 For instance by Stroheker [n. 2], esp. p. 234 33 Dronke [n. 1], p. 37; Kaiser [n. 2], pp. 202 sq. 34 Chronicarum quae dicuntur Fredegarii scholastici. Libri IV cum continuationi- bus, ed. by Bruno Krusch, Hanover 1888 (MGH SRM 2), pp. 1Ð193, Book IV, 41; 44; 55; cf. Wood, Ethnicity [n. 2], pp. 54Ð55; A. C. Murray, Germanic kinship structure. Studies in law and society in antiquity and the early middle ages (Stud- ies and texts 65), Toronto, 1983, p. 93. 35 Vita Faronis episcopis Meldensis, in: Acta sanctorum ordinis S. Benedicti, ed. by Jean Mabillon, 3rd edition, vol. 2, Paris 1936, pp. 606Ð625. 36 Wood, Gibichungs [n. 2], p. 247. 37 For a detailed study of Brunechilde, see J. L. Nelson, Queens as Jezebels: the careers of Brunhild and Balthild in Merovingian history, in: Medieval women. Dedicated and presented to Professor Rosalind M. T. Hill on the occasion of her 70th birthday, ed. by D. Baker, Oxford 1978 (Studies in church history: subsidia 1), pp. 31Ð77. Brought to you by | University of Toronto-Ocul Authenticated Download Date | 7/27/15 9:31 PM 232 SHAMI GHOSH mentions the Burgundians as being extremely hostile to this queen.38 There can be no direct link between her and our legend, but is possible that names, and potentially the hostility attached to Brunechilde, current in the early sixth century were transposed onto the dragon-slayer legend and then attached to the Burgundian story.39 The absorption of Burgundy into Merovingian Francia, the retention of a Germanic language in part of the Frankish realm, and the similarity of names between some historical Frankish figures and those of legend could be enough to suggest that the Burgundian legend lived on and was modi- fied among the Franks (one might assume that the joining with the Bryn- hild legend took place at this stage). There was plenty of contact between the descendants of the Burgundian settlers and the Frankish aristocracy, and given the geographical proximity, it seems most plausible that if the legend survived in a Germanic language (in other words, if it was not later translated back into a Germanic language from Latin or Romance), it did so among the Franks. This would imply that it also travelled north fairly early. More interesting in this context is that at least by the Carolingian pe- riod, the Franks were interested in pasts, and specifically in Germanic pasts. Two eighth-century texts, the ›Chronicon universale Ð 741‹, and the ›Passio Sigismundi‹, confer on the Burgundians a Scandinavian past (thus distinguishing them from most Frankish origin myths);40 the ›Vita Faronis‹ gives the family of the Burgundiofarones a history in the Roman provin- ces of the Rhine, and like the ›Passio Sigismundi‹ derives the name derived from the Roman fortifications there (the ›Vita Faronis‹ has a somewhat more detailed but essentially similar account of the settlement and ety- mology to that of the ›Passio Sigismundi‹).41 38 Fredegar [n. 34], IV, 41. 39 On the genesis of the dragon-slayer legend see W. Haubrichs, Sigi-Namen und Nibelungensage, in: Blütezeit (FS L. P. Johnson zum 70. Geburtstag), ed. by M. Chinca, J. Heinzle and C. J. Young, Tübingen 2000, pp. 175Ð206, with exten- sive references; neither Haubrichs nor the older work he cites provide conclu- sive arguments. I cannot go as far as Kaiser [n. 2], p. 202, who believes not only that the Brynhild and Kriemhild figures are derived from the historical Brunechilde, but also that the events of the 14th aˆventiure of the ›Nibelungen- lied‹ are derived from Brunechilde’s conflicts with Queen Fredegund. 40 Chronicon universale Ð 741, in: Supplementa tomorum IÐXIII, pars I, ed. by G. Waitz, Hanover 1881 (MGH SS 13), pp. 4Ð19 at 4; Passio S. Sigismundi regis, in: Fredegar [n. 34], ed. by W. Wattenbach, pp. 329Ð340, c. 1. The Scandinavian ori- gin could derive from Fredegar [n. 34], III, 65; cf. Wattenbach’s note to the ›Pas- sio‹. 41 Vita Faronis [n. 35], c. 8. The Burgundians inhabited Germania according to this text. The etymology of the name might be derived from Orosius or Isidore, the Scandinavian origin possibly from Fredegar, Jordanes or Paul the Deacon, whose works were in circulation in Carolingian libraries: cf. Wattenbach’s note Brought to you by | University of Toronto-Ocul Authenticated Download Date | 7/27/15 9:31 PM  ON THE ORIGINS OF GERMANIC HEROIC POETRY 233 There are no precedents in fifth-, sixth- or seventh-century sources for a Scandinavian origin of the Burgundians, and the etymological explana- tions seem to derive from learned sources rather than native material. It seems certain that the interest in this ancient and potentially Scandinavian past was thought appropriate in the eighth century and after (though it should be noted that even in these later histories, the Burgundians are put together as a people by Rome). It is not possible, however, to determine whether or not the Burgundians mentioned in these texts (and presumably derived ultimately from the testimony of Orosius) have anything to do with the Gibichungs of the fifth-century chronicles: the Rhine catastrophe is not mentioned, and the settlement in southern France is portrayed very differently from the contemporary chronicles. Nor do any of the Latin sources of this period share anything in common with our legend in the extant vernacular forms.42 Otherwise too, it appears that the Carolingians had some interest in a barbarian (and possibly Germanic) past, as is apparent from a number of statements in ninth-century texts:43 Ninth-century evidence includes Einhard’s famous statement that Char- lemagne recorded what might be barbarian heroic or historical poetry;44 there is no indication that the kings mentioned need have been exclu- sively of Charlemagne’s own family, or even Frankish. Charlemagne’s son Louis the Pious also apparently learnt pagan songs in his youth, which appears to indicate a current interest in the court at the time, even if he repudiated them later.45 Whether these songs were ›Ger- manic‹ or not is difficult to tell Ð it is conceivable that they might have been Aquitanian rather than Germanic.46 Charlemagne’s interest in The- oderic is evident from the testimony of Agnellus of Ravenna, who tells to c. 1 of the ›Passio Sigismundi‹ [n. 40]; and Orosius, Historiae adversus pa- ganos, ed. by M.-P. Arnaud-Lindet, Paris 1990 sq., VII, 32, 1. On the ›Passio Sigis- mundi‹ and the ›Vita Faronis‹, see also Wood, Ethnicity [n. 2], pp. 56Ð57. 42 Paul the Deacon mentions that Attila kills Gundicarius, perhaps evidence of his knowledge of some version of our legend, but given that he does not elaborate, it is not possible to know at what stage of development he might have encoun- tered it. See Paul the Deacon, Historia Romana, ed. by H. Droysen, Berlin 1879 (MGH AA 2), pp. 4Ð224 at p. 202. 43 On the following, see W. Haubrichs, Veterum regum actus et bella Ð Zur sog. Heldenliedersammlung Karls des Großen, in: Aspekte der Germanistik (FS H.-F. Rosenfeld zum 90. Geburtstag), ed. by W. Tauber, Göppingen 1989 (GAG 521), pp. 17Ð46; and D. Geuenich, Die volksssprachige Überlieferung der Karolin- gerzeit aus der Sicht des Historikers, in: DA 39 (1983), pp. 104Ð130. 44 Einhardi Vita Karoli Magni, ed. by O. Holder-Egger, Hanover 1911 (MGH SRG 25), c. 29. 45 Theganus, Gesta Hludowici imperatoris, ed. by E. Tremp, Hanover 1995 (MGH SRG 64), pp. 167Ð278 at c. 19, p. 200. 46 Cf. Astronomus, Vita Hludowici imperatoris, ed. by E. Tremp, Hanover 1995 (MGH SRG 64 [n. 45]), pp. 279Ð555 at c. 4, p. 294; it is not entirely clear what is meant by peregrinorum mores in this context. Brought to you by | University of Toronto-Ocul Authenticated Download Date | 7/27/15 9:31 PM 234 SHAMI GHOSH us that the emperor took from Ravenna a statue of Theoderic.47 The Old High German ›Hildebrandslied‹48 is also set in the context of Theoderic’s battles, though it contains no information that is strictly historically accurate. It is impossible to ascertain exactly what value this figure had for the Carolingians, but it seems manifest that Theoderic was already a revered ancient ruler (though whether he was thought to be ›Ger- manic‹ is unclear, as is the value of such a term in the period). There are also some references to written Germanic poetry in library cata- logues.49 In addition, Flodoard of Rheims makes a clear reference to a written vernacular narrative about Ermaneric.50 Other ninth-century references to what might be Germanic heroic poetry seem to be over- whelmingly about oral performance. In addition, there is some evidence that in the ninth century, some tradition of historical vernacular poetry dealing with relatively recent events was still current: Poeta Saxo refers to (probably vernacular) songs about Charlemagne’s ancestors;51 de- spite dependence on Einhard, Poeta Saxo’s words seem to indicate that a contemporary audience would not have found the idea of songs about the ancients implausible. In addition, the existence of the ›Ludwigs- lied‹,52 a poem of uncertain genre that praises the exploits of a young Carolingian king, is testament to a still existing v e r n a c u l a r tradition of historical poetry in the 880s (I do not mean to imply that the ›Ludwigs- lied‹ is an example of an unchanging, ancient Germanic tradition Ð manifestly, it is not, being composed in end rhymes and not alliterative verse).53 The astonishing existence of a Latin ›Germanic‹ epic, the ›Waltharius‹, possibly composed in the ninth century, is further testament to an interest 47 Andreas Agnellus, Liber Pontificalis Ecclesiae Ravennatis, ed. by O. Holder-Eg- ger, Hanover 1878 (MGH SRL VIÐIX), pp. 265Ð391 at c. 94, p. 338: On Theoderic in this period, cf. especially H. Löwe, Von Theoderich dem Großen zu Karl dem Großen. Das Werden des Abendlandes im Geschichtsbild des frühen Mittelalters, in: DA 9 (1952), pp. 353Ð401; and M. Innes, Teutons or Trojans? The Carolingians and the Germanic past, in: The uses of the past in the early middle ages, ed. by Y. Hen and M. Innes, Cambridge 2000, pp. 227Ð49 esp. pp. 242Ð245. 48 Althochdeutsches Lesebuch, ed. by W. Braune, 17th ed. by E. A. Ebbinghaus, Tübingen 1994, pp. 84 sq. 49 See P. Lehmann (ed.), Mittelalterliche Bibliothekskataloge Deutschlands und der Schweiz, vol. I, München 1969, p. 248 (Reichenau catalogue of 821 sq.); and p. 260 (Reginberts catalogue from between 835 and 842). On written vernacular poetry, see E. Hellgardt, Zur Mehrsprachigkeit im Karolingerreich. Bemerkungen aus Anlaß von Rosamond McKittericks Buch »The Carolingians and the written word«, in: PBB 118 (1996), pp. 1Ð48 at pp. 36Ð38. 50 Flodoard, Die Geschichte der Reimser Kirche (Historia Remensis Ecclesiae), ed. by M. Stratmann (MGH SS 36), Hanover 1998, IV, 5 (Que regibus quibusdam Folco direxerit scripta) at p. 383. 51 Poeta Saxo [n. 12], V, ll. 117Ð120 and ll. 545Ð546. 52 Althochdeutsches Lesebuch [n. 48], pp. 136Ð138. 53 On other Germanic praise-poems in relation to the ›Ludwigslied‹, see H. Beck, Zur literaturgeschichtlichen Stellung des althochdeutschen Ludwigsliedes und einiger verwandter Zeitgedichte, in: ZfdA 103 (1974), 37Ð51; he rightly suggests that there need not be any opposition between a Germanic tradition and Chris- tian/Latin traditions. Brought to you by | University of Toronto-Ocul Authenticated Download Date | 7/27/15 9:31 PM  ON THE ORIGINS OF GERMANIC HEROIC POETRY 235 in and knowledge of Germanic heroic legend Ð even in monastic and/or clerical circles.54 This poem is of especial interest because it also contains the historical figures of Attila, Gundaharius, Gibica, as well as the un- historical Hagano, who corresponds to the Old Norse Ho§ gni, Hagen in the ›Nibelungenlied‹. Gibica is here, however, a Frank (ll. 13Ð14). There is a conflict with the Huns, but the kingdom of Gibica is not destroyed; he sends a hostage to the Huns, and at the end of the poem Gundaharius rules a still-existing kingdom; the legend of the fall of the Burgundians, if related to the ›Waltharius‹, has turned into something very different. That the historical Burgundian king is now a Frank seems to support the view that Burgundian legend travelled among the Franks Ð and, it seems, was even made a part of their history in some way. It should be noted that the ›Waltharius‹ must have occupied a very different place in the historical consciousness of its audience than vernacular, oral legends. The ›Waltharius‹ could only have a primary audience among the Latinate and literate, and with its many clear references to the inherited Latin literary tradition, it would have been perceived very differently from ver- nacular material, even when the latter was received by the same audience. That apart, vernacular legends would also have had a wide audience with no access to such epics as the ›Waltharius‹. The ›Waltharius‹ competes with a flourishing tradition of Latin prose historiography, with more exact- ing claims to accuracy, in which no connections are made between the Burgundian Gibica and his kin, Attila, and the Frankish kings. For this reason alone, it is unlikely that the ›Waltharius‹ was understood by its audience as being as historically accurate as the prose historiography. Such a conflict need not have been the case among people who would have had no access to the Latin historiography because of lack of literacy and/or Latin Ð but how different any vernacular version of the legend may have been from the extant Latin text is impossible to ascertain. However, given that some form of the legend of Walter travelled beyond this period, possibly in the vernacular,55 and remained connected to the figures of the Burgundian legend (though in the ›Nibelungenlied‹, these are Burgundians, not Franks), the poet was probably drawing on an exist- ing oral tradition. As is the case with the Brynhild/Sigurð figures, the Gibi- chungs are connected with characters (Waltharius and Hiltgund) for whom no clear historical model exists. If the suggestion that the Brynhild 54 Waltharius, ed. by K. Strecker, mit Unterstützung von Otto Schumann (MGH Poet. VI, 1), Weimar 1951, pp. 1Ð85; cf. P. Klopsch in 2VL 10 (1997), cols 627Ð 638; and P. Dronke, Waltharius Ð Gaiferos, in: U. and P. Dronke, Barbara et antiquissima carmina, Barcelona 1977, pp. 27Ð79. 55 On the various versions and the possible modes of transmission, cf. Dronke [n. 54]. Brought to you by | University of Toronto-Ocul Authenticated Download Date | 7/27/15 9:31 PM 236 SHAMI GHOSH legend began to be joined to the Burgundian one from some time in the seventh century can be accepted, then there are two cases of legends of unclear historical accuracy being linked, by the ninth century, to one that does appear to have a clear and definite historical core. The only vernacu- lar heroic lay extant on the continent, the ›Hildebrandslied‹, dates to this period, and, like the ›Waltharius‹ and the Brynhild legend, cannot be clearly related to any specific historical events, although it too does make mention of some historical characters. It seems fair to suggest, on the basis of the little evidence we have, that legends concerning migration- age figures had by the ninth century begun (and perhaps completed) a move from being fairly historical to being rather unhistorical (in content, though not necessarily in function). A well-known example of a story from the migration period, which was actually written within decades of the event it concerns, transforming into a legend, is that of Alboin and Rosamund. There are a number of different Latin versions of the story recorded within a few decades of Alboin’s death:56 Hoc anno Albuenus rex Langobardorum a suis, id est, Hilmaegis cum reliquis consentiente uxore sua Verona interfectus est: et supra scrip- tus Hilmegis cum antedicta uxore ipsius, quam sibi in matrimonium sociaverat, et omnem thesaurum, tam quod de Pannonia exhibuerat quam quod de Italia congregaverat, cum partem exercitus, Ravennae rei publicae se tradidit (›In that year Alboin, the king of the Lombards, was killed in Verona by his own followers, that is, by Hilmaegis and others, with the consent of Alboin’s wife. And the aforementioned Hil- megis, along with the aforementioned wife of Alboin, whom he married, and all the treasure Ð both what he [Alboin] had brought forth from Pannonia as well as what he had collected in Italy Ð surrendered him- self to the Roman state at Ravenna‹);57 Mortua autem Chlothosinda, uxore Alboeni, aliam duxit coniugem, cuius patrem ante paucum tem- pus interfecerat. Qua de causa mulier in odio semper virum habens, locum opperiebat, in quo possit iniurias patris ulcisci; unde factum est, ut unum ex famulis concupiscens, virum veninu medificaret (›With Chlothosinda, the wife of Alboin, having died, he took another spouse, whose father he had killed a short time perviously. For this reason the woman, who always hated her husband, awaited an opportu- nity when she could avenge the injustice to her father. Therefore it hap- pened that she, desiring one of his entourage, administered poison to 56 Cf. O. Gschwantler, Die Heldensage von Alboin und Rosimund, in: H. Birkhan (ed.), Festgabe für Otto Höfler zum 75. Geburtstag, Vienna 1976 (Philologica Germanica 3), pp. 214Ð247 at pp. 217Ð224 for dates and analysis of these and other sources; I do not fully agree with his interpretations. See also W. Goffart, The narrators of barbarian history (A.D. 550Ð800). Jordanes, Gregory of Tours, Bede, and Paul the Deacon, Princeton 1995, pp. 386 sq.; 290Ð293. 57 Marii Episcopi Aventicensis chronica, in: Chronica minora saec. IV. V. VI. VII., vol. 2 ed. by T. Mommsen, Berlin 1894 (MGH AA 11/2), pp. 225Ð239 at p. 238. Brought to you by | University of Toronto-Ocul Authenticated Download Date | 7/27/15 9:31 PM  ON THE ORIGINS OF GERMANIC HEROIC POETRY 237 her husband‹);58 Albuenus Chlodesindam, Chlotharii regis filiam, ha- buit uxorem; qua defuncta, aliam duxit coniugem, cuius patrem in- terfecerat. Ipse vero eiusdem mulieris fraude venino perit (›Alboin was married to Chlodesinda, the daughter of king Clotharius. After she was dead, he took another spouse, whose father he had killed. Alboin him- self perished by venom through the deception of the same woman‹);59 Aluinus Langobardorum rex factione coniugis suae a suis nocte inter- ficitur (›Alboin the king of the Lombard is killed at night through the plotting of his wife by his own followers‹);60 uxoris suae Rosemundae regis Conimundi filiae dolo apud Veronam interfectus est auxiliante sibi Elmigisilo, cum quo adulterari credebatur: quod postea manifes- tum est, dum eum sibi in loco mariti tam coniugio quam etiam regno copulare conata est (›he was killed by the cunning of his wife Rosa- mund, the daughter of king Cunimund, at Verona, with the help of Elmi- gisilus, with whom she is thought to have been having an affair; this later became known, for she tried to join herself to him in place of her husband both in marital union and in the rule‹).61 The key thing to note for our purposes is that the earliest form of the story remained close to fact: all versions agree, for all their variations, that Alboin was murdered with the connivance of his wife and her helpers from his own people. There is nothing innately unlikely about this; from what we know of the history of the period it seems quite plausible that Alboin could have married the daughter of a vanquished king, who might have wished to exact her revenge on him for this and been aided by or used the aid of one of his followers to accomplish this deed.62 The legend has not, therefore, undergone the transformation that the Burgundian leg- end has in all extant versions, where the destruction of a people is turned into a private conflict between Atli (or his wife) and the Burgundian kings. Paul the Deacon, writing in the late eighth century, recounts the same story, but with substantial additions to the testimony of the contemporary sources: In Paul’s version the story of Alboin and Rosamund takes on a full- blown legendary form: Alboin incites his wife to revenge by asking her, when he is drunk, to drink out of the skull-cup made from her father’s 58 Gregor von Tours, Historiarum libri decem, vol. 1: libri IÐV, ed. by R. Buchner, Darmstadt 1977, IV, 41. 59 Fredegar [n. 34], III, 65Ð66 (a compressed retelling of Gregory). 60 Iohannis abbatis Biclarensis chronica, in: Chronica minora [n. 57], pp. 207Ð220 at p. 213. 61 Continuatio Havniensis Prosperi, in: ibd., vol. 1, ed. by T. Mommsen, Berlin 1892 (MGH AA 9), pp. 298Ð339 at p. 337 sq. 62 For a survey of violence, revenge and feud in the early middle ages, see G. Halsall, Violence and society in the early medieval west: an introductory sur- vey, in: Violence and society in the early medieval west, ed. by G. Halsall, Wood- bridge 1998, pp. 1Ð45. On the role of revenge in the career of Brunechilde, see also Nelson [n. 37], pp. 39Ð44. Brought to you by | University of Toronto-Ocul Authenticated Download Date | 7/27/15 9:31 PM 238 SHAMI GHOSH head. She attempts to plot with Helmechis, but he advises her to seek Peredeo’s help. The latter refuses to help her, so she lies in the bed of her chambermaid, with whom Peredeo is having an affair; after he has unknowingly slept with her, she threatens to tell Alboin of their inter- course unless he helps her. He agrees, and while Alboin sleeps, removes his arms and ties up his sword; following Peredo’s advice, Rosamund lets Helmechis, the murderer, into the room. Alboin is unable to defend himself, and is killed.63 Here we note that the sparse story recorded by the contemporary histo- rians has been much embellished, and provided with a number of more personal details. Somewhat like the Burgundian legend, there is a differ- ence between the contemporary historical records and the later version. While it is possible (though by no means necessary) that Paul’s additions derives from a vernacular, oral narrative, what is more important for our context is that there is a clear embellishment and shift towards a narrative more motivated by immediate personal considerations in the later work. I f Paul’s source was vernacular oral tradition (as he appears to claim: he states that king was celebrated in songs among diverse peoples, all of whom, we should note, spoke Germanic dialects; Paul does not, however, explicitly state that his own version derives from any of these songs),64 it seems most plausible to assume that it represented the story as told in his own time; Paul’s version says little about what might have been in oral tradition immediately after the events. I would suggest that if there was an orally transmitted story soon after Alboin’s death, it was closer to that recorded by the earlier historians (and might have been a source for them). If this be the case, we would have here evidence that a narrative, originally rather brief and primarily historical in content, becomes trans- formed by the Carolingian period into one that appears to be more charac- teristic (according to modern scholarship) of heroic legend. The narrative of Alboin and Rosamund from its earliest extant forms to Paul the Deacon’s version, and perhaps the embellishments of the account of Attila’s death provided by Poeta Saxo and the ›Quedlinburg annals‹, are examples of stories that did deal with very recent events being trans- formed as those events receded into the distant past. This is not direct evidence for Burgundian practice, but it could be a support for the general argument that historical material from the fifth and sixth centuries was 63 Pauli historia Langobardorum ed. by L. Bethmann and G. Waitz, Hanover 1878 (MGH SRL), pp. 12Ð187, II, 28 at pp. 87 sq. Paul’s version of the story takes up two pages in this edition, in contrast to the few lines of the contemporary historians. 64 Historia Langobardorum [n. 63], I, 27, p. 70. We should note that this mention of songs occurs after another story about Alboin, and it cannot be certain that Paul intends us to believe that the king’s death was equally celebrated. Brought to you by | University of Toronto-Ocul Authenticated Download Date | 7/27/15 9:31 PM  ON THE ORIGINS OF GERMANIC HEROIC POETRY 239 turned into more legendary matter by the ninth century. If the Burgundian legend originated among some class of Burgundians soon after 436, and survived (probably with some transformations already) through to the end of the fifth century, it would at this time (like the Lombard story at the end of the sixth century) probably still have been seen as ›historical‹, dealing with fairly recent events, which would have been in the living memory of at least the grandparents, if not the parents, of the generation towards the end of the century. Like the story of Alboin and Rosamund, it dealt with a recent, identifiable past, and like this story, it is probable that its form and content was quite different from what we know from later sources: not so removed from historical fact, and closer in generic function to poems such as the ›Ludwigslied‹. The Burgundian legend, however, would have been superseded as cur- rent historical poetry by other narratives, dealing with more recent events. Perhaps because of the dramatic nature of its narrative, the Burgundian legend could have been retained as a part of inherited lore, and modified and transformed as Burgundian identity itself changed over generations, with some elements of recent historical poetry regarding the Frankish royal family perhaps being added, as these too lost their current relevance. Initially, however, the legend would have been fairly close to fact. The Lombard examples also show, however, that a record of historical fact (whether oral or written Ð we cannot be certain how the various early historians received their information) can travel beyond its origin quite fast (if this was among the people who experienced the historical events), and thereby possibly undergo transformations both in content and func- tion. It is impossible to know how soon a legend might have been trans- formed, how far it might have travelled, and how the changes might have been brought about; on the basis of the evidence examined, it seems rea- sonable to suggest that changes that removed the legend further from the realm of historical plausibility in the minds of an audience that had a living memory of the events would have taken place at least after the passing of eyewitnesses, and thus probably at least a few generations after the recorded events. III Past literary scholarship has expended much effort on elucidating the processes by which historical events were transformed into a poetic form that often distorted history. Such heuristic efforts appear to operate on the assumption that the legends achieved their present lack of historical Brought to you by | University of Toronto-Ocul Authenticated Download Date | 7/27/15 9:31 PM 240 SHAMI GHOSH veracity fairly close to the time of the events they tell of.65 For this reason, scholars such as Andreas Heusler believed that there was a sharp distinc- tion between praise-poetry, which did not contain a narrative and was concerned with current events, and heroic legend, which was concerned with the past (and specifically the past of the migration period), and did contain a narrative and plot (›Fabel‹). Many scholars would agree that heroic legend served as a form of collective memory, a belief borne out by comparisons with other, surviving oral traditions.66 The comparative evidence is impressive indeed, and suggests that it is reasonable to assume that the pre-literate cultures of Europe would have had some sort of oral tradition that probably functioned as a means of identity formation, group cohesion and collective memory.67 Even though more recently scholars 65 The classic (and still influential, if controversial) theories remain those of A. Heusler, Die altgermanische Dichtung, 2. , neubarbeitete und vermehrte Aus- gabe, Potsdam 1943; see also his Lied und Epos in germanischer Sagendichtung, Dortmund 1905, and his work cited at n. 16. The most significant rejoinder to Heusler was provided by W. Haug, Andreas Heuslers Heldensagenmodell: Prämissen, Kritik und Gegenentwurf, in: ZfdA 104 (1975), pp. 273Ð292. An im- portant response to Haug is presented by T. M. Andersson, Walter Haug’s Hel- densagenmodell, in: Germania. Comparative studies in the Old Germanic langua- ges and literatures, ed. by D. G. Calder and T. C. Christy, Woodbridge 1988, pp. 127Ð141. Of the recent work assessing the various theories see especially A. Ebenbauer, Heldenlied und »Historisches Lied« im Frühmittelalter Ð und davor, in: Heldensage und Heldendichtung im Germanischen, ed. by H. Beck, Berlin, New York 1988 (Ergänzungsbände zum RGA 2), pp. 15Ð34; and idem, Hat das »Nibelungenlied« eine Vorgeschichte? Eine Polemik, in: 6. Pöchlarner Helden- liedgespräch. 800 Jahre Nibelungenlied. Rückblick Ð Einblick Ð Ausblick, ed. by K. Zatloukal, Vienna 2001 (Philologica Germanica 23), pp. 51Ð74; more recently and cautiously, cf. K. Reichl, Singing the past: Turkic and medieval heroic poetry, Ithaca, NY [et al.] 2000, which provides a very valuable comparative perspective. A conservative statement of the consensus is in W. Haubrichs [n. 10], pp. 106Ð 114 and 127Ð133. For important views contrary to almost all other literary schol- arship, see R. Frank, Germanic legend in Old English literature, in: The Cam- bridge companion to Old English literature, ed. by Malcolm Godden and Michael Lapidge, Cambridge 1991, pp. 88Ð106; and W. Goffart, Conspicuously absent: martial heroism in the Histories of Gregory of Tours and its likes, in: The world of Gregory of Tours, ed. by Kathleen Mitchell and Ian N. Wood, Leiden [et al.] 2002 (Cultures, beliefs, and traditions. Medieval and early modern peoples 8), pp. 365Ð393. 66 See e. g. L. Hedeager, Migration period Europe: the formation of a political men- tality, in: Rituals of power, from late antiquity to the early middle ages, ed. by J. L. Nelson and F. Theuws, Leiden [et al.] 2000 (Transformation of the Roman world 8), pp. 15Ð57; J. Heinzle, Die Nibelungensage als europäische Heldensage, in: Die Nibelungen. Sage Ð Epos Ð Mythos, ed. by J. Heinzle, K. Klein and U. Obhof, Wiesbaden 2003, pp. 3Ð28 at pp. 18Ð21; Reichl [n. 65], esp. at pp. 55Ð 73, 133Ð151; Haubrichs [n. 10], pp. 107Ð112; 132. 67 On the importance of memorial practices in the middle ages, see the stimulating works of O. G. Oexle, Memoria in der Gesellschaft und in der Kultur des Mittelal- ters, in: Modernes Mittelalter. Neue Bilder einer populären Epoche, ed. by J. Heinzle, Frankfurt a. M. [et al.] 1994, pp. 297Ð323; idem, Memoria als Kultur, Brought to you by | University of Toronto-Ocul Authenticated Download Date | 7/27/15 9:31 PM  ON THE ORIGINS OF GERMANIC HEROIC POETRY 241 such as Ebenbauer and Reichl have argued for less sharp distinctions between the genres of panegyric and heroic verse, the consensus still appears to be that heroic poetry was in some way formed at a time close to the events it narrates, and that not just a few historical details, but also the plot structure and narrative motifs (and, some views, even the ›heroic ethos‹) of the extant texts are representative of a form of poetry dating back to the so-called migration period. It is generally accepted that the productive period for Germanic legend lies between the late fourth and the beginning of the seventh century: all the historical events that could correspond to what is contained in extant heroic legend in any Germanic vernacular belong in this period Ð a time in which, it should be noted, many Germanic peoples were making the move from being pre-literate to being at least partly literate cultures.68 The consensus of literary scholarship is that the tradition of composing ›heroic‹ lays based on history was no longer productive after this point. That a tradition of oral poetry survived through this period seems certain, though since no ›historical‹ works exist dealing with later events as past history, it is probably safe to assume that the oral tradition from the sev- enth century onwards consisted of poems about events of the migration period or of poetry concerned with current or fairly recent events. The (implicit or explicit) assumption, therefore, is that there was in the migra- tion period a current tradition of vernacular oral poetry that served some identity-forming function as a form of historical consciousness, and fur- thermore, that the legends probably originated among those peoples whose histories the poems narrate. In the case of the Burgundian legend, it would have originated amongst the surviving Burgundians as an ideal- ised memory of their past glory, and would have been handed down over the generations in a more or less fixed form, having a lr ea dy made the transformation from ›history‹ to ›legend‹, whether through an unconscious process innate in oral culture or at the hands of an individual poet. However, given the available evidence, we can never know if there was any sort of heroic poetry in the fifth century, nor whether the Burgundian legend enunciated any kind of heroic ethos, and transformed the historical events into a personal, family conflict. It seems absurd to me to argue against the existence of any oral historical tradition whatsoever, but in: Memoria als Kultur, ed. by O. G. Oexle, Göttingen 1995 (Veröffentlichungen des Max-Planck-Instituts für Geschichte 121), pp. 9Ð78, and from a rather differ- ent perspective, J. Fentress and C. Wickham, Social memory, Oxford [et al.] 1992; more broadly on the use of memory, see the seminal work of J. Assmann, Das kulturelle Gedächtnis. Schrift, Erinnerung und politische Identität in frühen Hochkulturen, Munich 1992. 68 Cf. Heusler [n. 16], pp. 495 sq.; and more recently, Hedeager [n. 66], pp. 31Ð37. Brought to you by | University of Toronto-Ocul Authenticated Download Date | 7/27/15 9:31 PM 242 SHAMI GHOSH whether or not this was ›heroic‹ is a moot point; all that we can say is that given the commonness of some form of heroic mentality in oral- traditional cultures around the world, some form of heroic ethos quite possibly existed among some section of the Burgundians too, though we cannot know how much it might have had in common with what is ex- pressed in the extant narratives. There is no evidence for an extant tradition of heroic poetry among the Burgundians, and if they composed any poem commemorating their de- feat in 436, it travelled fairly soon, and most probably underwent a number of transformations in the course of its journey. The generic function of the original lay was in all likelihood very different from that of the extant versions. I would like to propose that extant Germanic heroic poetry de- rives from h is to ri ca l poetry of the migration period, which might not have been as ›heroic‹ as the surviving texts. I suggest that the original poems were probably closer to historical fact, and functioned in much the same way as later works such as the Old High German ›Ludwigslied‹ or the Old English ›Battle of Brunanburh‹ and ›Battle of Maldon‹,69 free rela- tions of recent history, in which the characteristics of heroic poetry Ð the transformation of the political into the personal, the ›epic‹ chronology that allows personages of different centuries to mingle with each other Ð were probably only very minimally present, if at all (I stress that I make no broader claims with regard to similarity of form, plot, ethos, style or con- tent between these three poems and my postulated original Burgundian lay). In other words, the chronology was probably fairly accurate (the living witnesses to the events would have ensured that figures generations apart would not have been brought together as contemporaries), and while there might have been some intrusion of personal elements, this would have been on a much smaller scale than in the extant later legends, and would have corresponded to what was historically plausible (as is the case in the early reports concerning Alboin’s death). Thus, the motivation for the conflict between Burgundians and Huns would probably have been portrayed as primarily (and probably solely) political. Furthermore, the much discussed heroic ethos Ð with its stress on honour, exorbitant be- haviour, martial heroism etc. Ð need not have been in evidence in these original poems; there is little evidence either way. The supposed dichotomy of content between heroic and historical po- etry is in fact far from clear when one considers poems such as the ›Lud- 69 Das Ludwigslied [n. 52]; The Battle of Brunanburh, ed. by A. Campbell, London 1938 (discussion of the historical background at pp. 43Ð80); The Battle of Maldon, ed. by D. G. Scragg, Manchester 1981 (an overview of the historical background at pp. 8Ð23). Brought to you by | University of Toronto-Ocul Authenticated Download Date | 7/27/15 9:31 PM  ON THE ORIGINS OF GERMANIC HEROIC POETRY 243 wigslied‹, ›The Battle of Brunanburh‹ or ›The Battle of Maldon‹ (despite many similarities, I exclude skaldic verse from my analysis because of the insuperable problems involved in dating, transmission and authenticity). In terms of narrative content and closeness to historical fact, there is actually not very much difference between the ›Atlaqviða‹ and these three ›historical‹ poems: like the ›Atlaqviða‹ or other heroic poems, none of these three works actually present very much in terms of historical detail, and can scarcely be understood on their own as historiography without recourse to other, independent accounts. We should note that both ›Maldon‹ and the ›Ludwigslied‹ are poems in which historical events, while perhaps not distorted, are nevertheless clearly reported selectively to present a particular perspective, and both poems have fictional elements and a clear narrative (›Fabel‹). Neither can fit easily and exclusively into either category of praise-poem or heroic poem. None of these three poems present the kind of considerations of personal conflict and motivation common in the heroic works such as the ›Atlaqviða‹ or the ›Nibelungen- lied‹. All the events presented in the ›Ludwigslied‹, ›The Battle of Brunan- burh‹ or ›The Battle of Maldon‹ may be explained solely by what we know from other sources about the historical situation (even though ›Maldon‹ blurs the lines here to some extent, all the action is explicable by the actual historical context); this is impossible for the extant versions of the Burgundian legend.70 It appears fair to state that there is what might be called a generic continuum between straightforward historical/praise poetry dealing with contemporary/recent events, and heroic poetry dealing with the past, rather than simply these two extremes; the primary distinction in content appears to be that historical poetry deals with the present or recent past, whereas heroic verse deals with a distant, and sometimes rather mixed- up past. This in turn explains much of the chronological and psychological difference between the two types. Although it is true that strict genre distinctions are modern inventions and do not necessarily correspond to the evidence,71 the social function and possibly context of a poem that deals with a recent event in which one’s own king or his identifiable pre- decessors took part, and of one that deals with past events which took place in a time that might not have been ›our past‹, must be quite different. The main difference in function, deriving from the content, is that historical/ praise poetry is firmly rooted in a contemporary political context, whereas 70 On ›Maldon‹, ›Brunanburh‹ and the ›Ludwigslied‹, in addition to the notes and introductions in the editions cited above, see Beck [n. 53]; Reichl [n. 65], pp. 56Ð 60; 70Ð73; 177 sq.; and Ebenbauer, Heldenlied [n. 65], on the theoretical aspects of the preceding paragraph. 71 Reichl [n. 65], pp. 70Ð73. Brought to you by | University of Toronto-Ocul Authenticated Download Date | 7/27/15 9:31 PM 244 SHAMI GHOSH heroic verse is more universal in its meaning: since the contemporary situation has receded in time, the narrative of the poem can (and perhaps must) focus more on personal qualities and motivations (not necessarily historically accurate), and is freed from more immediate historical and political concerns. Because it is free of the political context of the events it relates, and even of living memory of it, chronological details may be altered to suit the psychological needs of the narrative. In the poetry rooted in a contemporary context, much may be left out of the narrative, since a knowledge of the background could be assumed; once such knowl- edge receded from the common memory, for a legend to enjoy a produc- tive reception, it might have been necessary to provide it with a more universal context, which in turn would require a focus on the personal over the political. Such a distinction between two types of historical poetry might well have existed in the migration period too. However, while for later times, the migration period functioned as a ›heroic age‹, to which people looked back with wonder and awe, and which provided some form of exemplary entertainment, in the migration period itself, when the events of the later heroic poems were happening or were of the very recent past, they were looked upon in a very different light. Just as in the Carolingian period it appears that there were poems dealing with the recent past, as well as ›heroic‹ literature dealing with an ancient past, so too this might have been the case in the migration age. If there was a ›heroic ethos‹ in the migration period and in the early medieval centuries, it was nourished by a different heroic age from that of the Carolingians, and one that is lost to us; no evidence shows that the extant heroic poems derive their ethos from the migration period. There is no reason to believe that the Burgundian legend, in its original form close to the events it reported, presented a narrative that diverged significantly from the historical record in terms of the motivation for the conflict or chronology; nor need we believe that the ›heroic‹ elements (acceptance of fate, dying heroically, etc.) belonged in the original story. Any transformations from a narrative with a historical motivation, regard- less of how little it contained in terms of factual content, to one in which the plot was moved by Heusler’s ›Privat-menschliches‹, either through agency of an individual poet (Heusler) or by modifications caused by Haug’s literary ›Muster‹, must have taken place at a later stage. It is pos- sible that the transformations of the narrative of Alboin and Rosamund, and perhaps also the few late references to the death of Attila, are witness to this change from historical narrative motivated by political concerns, to ›heroic‹ narrative, motivated by personal conflict. Although it is apparent that similar transformations of political into more personal events are Brought to you by | University of Toronto-Ocul Authenticated Download Date | 7/27/15 9:31 PM  ON THE ORIGINS OF GERMANIC HEROIC POETRY 245 present in heroic legends from many parts of the world, I know of no evidence to suggest that this transformation takes place while living wit- nesses to the events, who would be able to contradict both the recurring theme of ›personal conflict as motivation‹, and the garbled chronology (and potentially also the expression of an ›heroic ethos‹) are alive. We know well enough in what ways heroic epics differ from the facts, when contemporary chronicles exist; we cannot claim to k no w how, why, or when these differences came into being. While my theory outlined here is, of course, impossible to prove, it accords better with what we know of historical traditions in the early middle ages, and Ð equally important Ð our massive ignorance of the nature of these traditions. It follows from my proposal that the functions of migration-age and early medieval historical poetry had nothing (or very little) in common in that period itself with the functions of the heroic poetry it later became. In other words, whether or not any (no longer existing) heroic poetry of the fifth century had something in common with extant heroic poetry, the archetype of the Burgundian legend served a very different function from the extant texts derived from it. Theories concerning the function of he- roic poetry that are based on the extant texts might perhaps be applicable to the latter, but they cannot explain their migration-age archetypes. T he l it er ar y a na ly si s o f t he su rv iv in g m on um en ts is us el es s w it h r eg ar d t o t he ro le of th ei r m od el s i n t he so ci et ie s i n w hi ch t he y o ri gi na te d. Whether one considers them to be realistic and typical portrayals of the time, exempla of honourable action, fascinating tales of exorbitant characters Ð we have no way of knowing if these assessments might have applied to the historical poetry of the migration period as well. It is admittedly often hard to fit the extant heroic poetry into what is known of the cultural context in which it was written, for which reason explanations are sought in the heroic ethos of the past; but such an ethos cannot legitimately be sought in texts written several centuries after the passing of the postulated heroic mentality they are supposed to be evi- dence for. The Burgundian legend, originally a historical narrative, was trans- formed over time into a tale of family conflict between a gold-hungry Attila and the three Burgundian kings. Its original function, like that of the ›Ludwigslied‹, was to record the memory of a battle, possibly with some identity-forming function for survivors of that conflict; its later func- tion might have been to serve as an example of good (or bad) conduct for a king, perhaps even to serve as a means of interpreting the (distant) past; ultimately, it would have become something close to what it is for us now: a work of art and entertainment. Like any report of a historical event, the original version must have been coloured by a particular perspective (and Brought to you by | University of Toronto-Ocul Authenticated Download Date | 7/27/15 9:31 PM 246 SHAMI GHOSH we should note how ›Maldon‹ attempts to find a moral victory in the face of defeat: something similar might have been present in the original Bur- gundian legend), but this perspective is completely lost to us, and cannot be recovered by recourse to any extant text. As in the case of the Lombard legend, there is a transformation over time Ð and it is probable that much of this transformation had taken place by or took place during the Carolin- gian period. All the evidence examined above (the ›Passio Sigismundi‹, the ›Waltharius‹, the ›Hildebrandslied‹, and the Lombard legend, perhaps even the names of Brynhild and Sigurð and the narratives concerning Attila’s death) seems to point to an increasing departure from historical fact and admission of more legendary elements into narratives of the mi- gration age from end of the seventh century onwards. By the Carolingian period, the legends were no longer a part of a spe- cific group’s historical consciousness; in other words, they no longer had any kind of identity-forming function. The stories were by now more »a matter of folklore, a set of widely diffused common stories with stock characters, which served almost as parables, examples of morality in ac- tion [. . .]«.72 This also allows (perhaps even invites) the admission of ele- ments almost certainly extraneous to the original narrative: in our case, apart from the transformation into a conflict between kin, the intrusion of the Ermaneric legend, and the linking with the Brynhild story. Since the legend of the Burgundians d oe s have a historical core it seems most likely that there was indeed a vernacular legend about these events, which was current before, during, and after the ninth century (by which point it probably served an exemplary rather than identity-forming function), but one which was not created then. I would suggest that this legend must have originated close to the time of the historical events Ð when it w as a part of a historical consciousness, that of the group whose history it tells. The possibility that the story originated among another more or less disinterested contemporary group cannot be ruled out. It is highly likely, however, that the Burgundians did have some form of historical poetry, and therefore that they did commemorate the events of 436 in some poetic form. It seems simpler to assume that the surviving texts derive from this tradition, rather than postulating an independent origin outside the theatre of events in addition to a Burgundian tradition, with the former being the one to survive, rather than the latter. Of course, it is extremely likely that if there were more than one contemporary story about this event, cur- rently extant versions would have been influenced not just by one, but by several ›originals‹. We should note also that n on e of these potential prototypes need have been formal poems; they may just have been stories 72 Innes [n. 47], p. 248. Brought to you by | University of Toronto-Ocul Authenticated Download Date | 7/27/15 9:31 PM  ON THE ORIGINS OF GERMANIC HEROIC POETRY 247 passed on informally, not necessarily in the context of any kind of per- formance or ceremonial setting.73 There is plenty of evidence that songs about ancient deeds in the ver- nacular were sung between the sixth and the thirteenth centuries, so it seems highly unlikely that the various legends w ri tt en in the later period have nothing to do with what was s un g before. While the oral/lay and literate/clerical cultures of the early middle ages were not mutually exclu- sive cultural spheres, cases of vernacular borrowing from written, Latin material on the continent are exclusively confined, until the end of the Carolingian period at least, to religious literature, and arise largely from conscious clerical efforts to disseminate the message of Christianity. It is impossible to establish direct textual relationships between any extant Germanic legend and any extant contemporary Latin text; this is not the case for the many vernacular religious works of this period. It seems most likely, therefore, that the transmission was in the vernacular all the way. This brings us to the crucial question of language: the only continental Germanic legend to survive in Romance in the region of the historical figures it contains is the legend of Walter of Aquitaine. In all of those regions where the historical equivalents of the Germanic heroes lived, the language spoken by the ninth century was no longer a Germanic one. Moreover, with the same exception of the legend of Walter, Germanic legends do not survive in the vernacular Romance languages. This is surely not a coincidence. It might be explained by a Carolingian interest in a Germanic past, but if one assumes that Germanic legend originated in this period in the imagination of poets, such an interest would also require a fairly ambitious enterprise of research into somewhat obscure chronicles from several centuries beforehand, and translation into the ver- nacular Ð something for which there is absolutely no evidence at all. Given that there is plentiful evidence of some form of vernacular poetry in the Carolingian period, some most probably containing stories about the past, it is surely easier to assume that our legend derived from a living oral tradition rather than a reading of Latin history. This should not imply, however, a reversion to the position that the Germanic legends remained largely unchanged (in content, function, or ethos) over the centuries: pre- suming that a s to ry was kept alive in oral tradition does not imply lack of transformation any more than postulating a creative translation of Latin historiography. With regard to the Burgundian legend, I stress again that while some continuity seems to have existed, this bore only the four basic 73 Cf. H. Kuhn, Heldensage vor und außerhalb der Dichtung, in: Zur germanisch- deutschen Heldensage: Sechzehn Aufsätze zum neuen Forschungsstand, ed. by K. Hauck, Darmstadt 1961 (WdF 14), pp. 173Ð194. Brought to you by | University of Toronto-Ocul Authenticated Download Date | 7/27/15 9:31 PM 248 SHAMI GHOSH historical facts that the extant vernacular works share with contemporary historical record: these facts, however, could have been manipulated in any number of ways, and embellished with any number of potential motifs, conflicts, and moral or ethical messages. Given the state of the evidence, it seems that if we are to seek the origins of our extant vernacular legends, they may be sought in the historical events only insofar as the few factual details are concerned; the origins of their transformation from historical record to narratives driven by personal conflict, with mythic and ›heroic‹ motifs and telescoped chronology, can probably be better traced to the eighth and ninth centuries, or at any rate, to a period several generations after the events. IV The preceding examination of evidence shows that we can know little about Germanic oral tradition beyond the fact that it probably existed; its content and function remain opaque, and parallels to other cultures can only be used with caution.74 While it is certain that there was such a tradition, we can know little about what it contained and what function it served. Modern anthropologists may ask their subjects about their notions of identity; since we see them only through the lens of Latin writing (laws, letters, poetry and histories), which brings with it its own baggage, the group identity of the peoples among whom Germanic oral poetry probably flourished is quite opaque to us. Since nothing survives of the ›oral culture of the barbarians‹, it seems very likely that at least some of what survives was created later as a form of retrospective use of the past Ð and it is in fact most unlikely that there was no interference from later periods. What is extant cannot be seen as any evidence about the culture of ›Germanic antiquity‹, or even as evidence that such a thing as Germanic antiquity existed. We should also remember how soon at least some sections of Burgundian society began to adopt ›Roman‹ customs, and eventually a Romance language; this should caution us about assumptions of cultural stability and the importance and preservation of traditions from which our extant texts might be derived. Discourses of memory in the early middle ages most probably had a significant identity-forming function, but given the lack of clarity about 74 Germanic heroic poetry is preserved among peoples with no relation to its he- roes; this is, to my knowledge, never the case in other cultures. The identity- forming function of other heroic traditions cannot, therefore, simply be trans- ferred by analogy to Germanic poetry. Brought to you by | University of Toronto-Ocul Authenticated Download Date | 7/27/15 9:31 PM  ON THE ORIGINS OF GERMANIC HEROIC POETRY 249 the nature of early medieval social groups, it remains unclear to us how they used their traditions and their past. This applies equally to the surviv- ing poetry that is thought to originate in this period. Traditions and memo- ries could be manipulated and possibly invented, as well as carefully se- lected and extinguished. Moreover, legends could, and manifestly did, travel, and could be used by people who had no relation to any past that was in the legends; this need not indicate any sense of ›Germanic‹ kinship. The possible functions of extant legends in later contexts say nothing about their functions in their original contexts, which were certainly dif- ferent. As noted, with one exception, there are no Romance legends containing material also in Germanic languages. On the other hand, heroic legends dealing with what were once current historical events did come into being in Romance (for example, the ›Guillaume‹ cycle and the ›Chanson de Ro- land‹ from the twelfth century, both dealing with events from several cen- turies earlier). This is further indication that among both the Germanic and the Romance peoples a current tradition of historical poetry was maintained; that ›Germanic‹ legends did not survive among the now ›Ro- mance‹ peoples is not because the latter no longer cultivated such poetry. The legends dealing with a ›Germanic‹ past were widely transmitted among Germanic peoples not because of a shared past Ð the past of the legends was shared by the Romance speakers more than by the Germanic speakers Ð but because of closely related languages, and common forms of poetry (using, for example, similar alliterative metres or kennings). The legends were probably carried to the peoples among whom a Germanic language continued to be spoken by the end of the seventh century. Hav- ing been composed by Germanic-speaking peoples, they died out among those peoples when their language changed from Germanic to Romance: hence the survival only among distant and unrelated peoples. This ex- plains also the fact that Romance languages preserve for us no legends from before the shift from Germanic to Romance that are also preserved in a Germanic language. In their new environment, our narratives would have been received and passed on n ot as a form of maintaining ›our past‹, but simply as a part of inherited tradition, ›folklore‹, with a purpose quite different from that of current historical poetry. By the ninth century, although a tradition of historical poetry dealing with the relatively recent past remained current, there existed also a tradition of such inherited manifestations of culture, the value of which did not lie in their power to help form identity by identification with common ancestors. The same legend, with the same ›factual‹ core, therefore, could now have served a very different function from that of its original context. Brought to you by | University of Toronto-Ocul Authenticated Download Date | 7/27/15 9:31 PM 250 SHAMI GHOSH The lack of ›ethnic‹ identities stable over centuries might have been, paradoxically, the root of a ›pan-Germanic‹ culture: if people’s ethnic iden- tities were not well-defined or static, and if people were often absorbed into different groups, but simultaneously maintained some cultural practi- ces (such as the telling of specific legends), these would have then been carried into new situations, where any previous relevance of these legends in terms of identity formation b y m ea ns of a l in k t o a sh ar ed pa st was lost. This does not mean that the traditions lost all cultural relevance. There may have been no ›Germanic‹ ethnic or national identity, but this did not preclude some similarities, including, especially, common poetic techniques and themes, as well as somewhat similar social structures (by no means exclusive to Germanic cultures), and most crucially, closely related languages. The growing awareness of the kinship of languages may have been connected to a shared antiquarian interest in stories in the related languages. The sharing of poems and/or legends with a base in a particular group’s history does not indicate that the groups with different histories who shared had any sense of a common past; it could mean that cultural traditions overlapped, which permitted historical traditions to do the same as groups merged and changed over time. The distant past did not need to be ›our past‹ to have some value in a people’s historical con- sciousness, if the value lay elsewhere than in forming identity through i de nt if ic at io n with the past. A past might have cultural uses without being thought of as one’s own. This might also explain, for instance, the joining of what seem to be (from a historical point of view) the very different traditions of the Sigurð legend (with its very strong mythic elements), the fall of the Burgundians, and the death of Ermaneric. With each movement of the legend, the new groups adopting and adapting it might have brought their own historical traditions that melded and lost their immediate historical value as the group identities overlapped. Ultimately, the sense of a particular group’s past need have little to do with what we might consider to be the real past, judged by biological or direct political links. For the purpose of cultural traditions, an ›invented‹ past is as good as a real one: beyond a certain lapse of time, no one was likely to have a fixed memory of any events, and the memorial traditions that were propagated would have changed considerably from the historical facts, taking on new traditions and facts and merging them in completely unhistorical ways. For their audiences, however, this did not make the legends any less ›authentic‹ Ð by which I mean viewed as having some p er ce iv ed correlation to fact. Heroic poetry, therefore, cannot be viewed simply as a manifestation of traditions about a shared past; one must always be careful to consider whose past, and the value of the tradition for the people who maintained Brought to you by | University of Toronto-Ocul Authenticated Download Date | 7/27/15 9:31 PM  ON THE ORIGINS OF GERMANIC HEROIC POETRY 251 it. The wanderings and transformations of legend, coupled with the in- creasing and competing claims of Latin historiography, might indicate that for some, at least, the legends might have had no credibility as fact. But some form of vernacular historical tradition probably retained credibility among the many who had no access to Latin historiography, and the line between fiction and history was probably drawn fairly late, and is unlikely to have been pervasive at all levels of society until at least the end of the middle ages. I stress that o ur conceptions of history and myth are very different from those of the middle ages; it seems likely that there existed, even until later in the middle ages, differing conceptions of history that might have allowed for an equal credibility regarding both what we con- sider ›historical‹ information, and heroic traditions. The Latinate and literate, as well as the vernacular-speaking and illiter- ate sections of society might well have participated in the same discourse of oral tradition Ð though possibly for different reasons. While for those with no access to Latin history and literate notions of historiography, the oral traditions probably still had some credence as ›fact‹, for both this social layer as well as the elites, the tradition of heroic legends, in their performance, probably also had a function »more generally as a cultural symbol. Its significance for a (however sociologically defined) community consists also in the fact that it expresses cultural values, reinforces them, and comments upon them«.75 The cultural values of the Carolingians (or of thirteenth-century Icelanders or Bavarians) were certainly not identical to those of the fifth-century Burgundians. If these legends were perceived as an emanation of a heroic ethos in the ninth or the thirteenth century, this says nothing either about the origins of the legends, or their original moral position, which, like the content, could have changed quite consid- erably and probably did. The cultural values these legends represented, moreover, need not necessarily be related to any sense of ethnicity or group origins; the heroic legends might have value as inherited, ancient tales of a distant past, with possibly some moral message (which too ought not to be understood as unchanging over the course of transmis- sion), even if that past was not ›our past‹ for the audience. It may be objected that I am simply substituting a ›Germanic‹ past with a ›distant‹ past; but there is a difference. An Englishman with a public school education in the nineteenth century would have counted both Rich- ard the Lionheart and Caesar as a part of his cultural inheritance, and located both in the past. This does not mean that he would have identified the past of Caesar as being his own, as he probably would have done with that of King Richard. Similarly, a German-speaking courtly audience in 75 Reichl [n. 65], p. 177. Brought to you by | University of Toronto-Ocul Authenticated Download Date | 7/27/15 9:31 PM 252 SHAMI GHOSH the early thirteenth century would probably have viewed the figures of Germanic legend as well as Charlemagne and Alexander the Great as be- ing heroic figures of a glorious past, though not necessarily in all three cases a past to which their own group could, would, or wished to trace its lineage. Of course, medieval nostalgia of the sort proposed by Frank (n. 65) could well have played a part Ð but there is little evidence for it before the beginning of the late eighth century. An image of a common Germanic past, if created in the eighth century, cannot explain the survival of legend until this point; I would argue that it alone probably cannot explain it even after this point. Even before the Carolingian period, I would propose, Germanic legends would have made the transformation from tales of the recent past, with some form of identity-building function, to narratives set in the distant past, now an element of inherited cultural tradition, but not necessarily with any identity-forming role. Understanding the oral memorial traditions of pre-literate Germanic- speaking peoples, and how they relate to textual historical records and extant heroic legends would require a more detailed study of not just one, but all of the legends, and unfortunately not all other legends provide us with the same relative richness of early source material as the Burgundian story. Such detail is impossible here, but I hope some of the methods by which such a study might proceed have been illustrated. The foregoing reflections have, I hope, both presented a step towards a better under- standing of the social context of heroic legend in the many centuries be- fore it was written, and provided a chapter of the prehistory of Germanic heroic poetry that is more compatible with the historical record than has often been the case. TORONTO SHAMI GHOSH Brought to you by | University of Toronto-Ocul Authenticated Download Date | 7/27/15 9:31 PM