1. Introduction
The dating of Beowulf, an Old English epic poem preserved in a single manuscript copy produced around the year 1000 (Dumville Reference Dumville1988, Fulk et al. Reference Fulk, Robert and John2008:xxv–xxxv), is a longstanding controversy in Anglo-Saxon studies. Scholars have used various cultural, historical, paleographical, and linguistic arguments to date the composition of the poem as early as ca. 625 and as late as ca. 1025 (Bjork & Obermeier Reference Bjork, Anita, Robert and John1997, Evans Reference Evans1997:41–63, Neidorf Reference Neidorf2014b). Because the evidence of language is generally regarded as more decisive than other forms of evidence, linguistic arguments for the poem’s date of composition have typically received the most exacting scrutiny (Fulk Reference Fulk2014). For instance, Fulk’s claim that the poem’s adherence to Kaluza’s law indicates composition prior to 725 has been minutely and contentiously examined in a series of subsequent publications (Fulk Reference Fulk1992:§§12, 33, 40, 179–183, 406–421; Suzuki Reference Suzuki1996; Hutcheson Reference Hutcheson2004; Frank Reference Frank2007a; Weiskott Reference Weiskott2012; Neidorf & Pascual Reference Neidorf and Rafael2014). Likewise, there has been much discussion of the tendency in Beowulf for the weak adjective to be used without a determiner, a feature regarded by some scholars, though not all, as a sign of the poem’s relative antiquity (Lichtenheld Reference Lichtenheld1873, Chambers Reference Chambers1959:105–106, Amos Reference Amos1980:110–124, Fulk Reference Fulk2014:27–28, Yoon Reference Yoon2014, Neidorf Reference Neidorf2017:§173). It is therefore surprising that almost no detailed discussion has attended the claim, propounded by Roberta Frank in publications spanning three decades (1979, 1981, 1987, 1990, 2007b, 2008), that Old Norse influence is discernible in the language of Beowulf. If Frank’s contention were correct, it would require the poem’s composition to be situated no earlier than ca. 890, by which time a permanent Scandinavian presence had come to be established in England.Footnote 1
Several brief critiques of Frank’s argument have appeared, each one amounting to no more than a few pages, in which disagreement is registered for the most part on general theoretical grounds (Fulk Reference Fulk1982:343–344, 2014:21–22; Andersson Reference Andersson1983:295–297; Townend Reference Townend2000:357–358, 2015:4–5). It has been pointed out, for instance, that the evidence Frank construes as indicative of Old Norse influence on Beowulf is equally susceptible of an obverse interpretation, that is, as evidence for Old English influence on Old Norse poetry. It has also been objected that lexical and syntactic parallels between Beowulf and Old Norse poetry are better regarded, in principle, as shared inheritances from the Common Germanic period rather than byproducts of Anglo-Scandinavian contact during the Viking Age. Yet no sustained examination of the linguistic material in Beowulf that Frank interprets as signs of Old Norse influence has yet been undertaken. Three monumental books on Anglo-Scandinavian language contact have appeared in the past two decades (Townend Reference Townend2002, Dance Reference Dance2003, Pons-Sanz Reference Pons-Sanz2013), yet none of them engages with Frank’s contention that Nordicisms are to be found throughout Beowulf. The omission of Frank’s evidence from these books amounts perhaps to its tacit rejection, yet there are signs elsewhere of credence in her claims. Of particular significance is the fact that the principal article in which she enunciated her views (Frank Reference Frank1981) has been reprinted twice in anthologies of Beowulf criticism edited by Peter S. Baker (Reference Baker1995, Reference Baker2000). Her arguments have, moreover, recently received assent in a monograph on the dating of Beowulf by Helen Damico (Reference Damico2015:8–13). In two surveys of scholarship on Anglo-Scandinavian literary relationships, Frank’s views are rehearsed without endorsement, but also without any explicit statement of their improbability (Bjork Reference Bjork, Philip and Elaine2001, Dance Reference Dance2004).
The ambivalent reception of Frank’s arguments—which are not widely credited, but not considered wholly discredited either—is a probable consequence of the uncertainty attending the array of linguistic peculiarities she identified in Beowulf to be explained under the hypothesis of Old Norse influence. Frank called attention to a number of interesting features of the language of Beowulf that distinguish it from other Old English works and seem, in her view, to be paralleled exclusively or most prominently by features of Old Norse language and literature. Even if her hypothesis were rejected on general theoretical grounds, these linguistic peculiarities would still remain in need of a superior explanation. The central aim of the present article is thus to reassess the data adduced in Frank’s studies and determine what, if any, significance this material might possess for the dating of Beowulf. Closer examination of the data in the light of evidence not considered by Frank reveals that much of it is better regarded as indicators of archaic composition than as indicators of late Scandinavian influence. Beowulf differs from the rest of the corpus of Old English, while exhibiting affinities with texts recorded in other early Germanic languages, not because its author consciously imitated the speech of foreign Germanic speakers, but because the early composition of this poem resulted in its conservation of a wider array of features that were probably characteristic of Proto-Germanic usage.
2. Alleged Nordicisms
Before entering into our analysis of Frank’s alleged Nordicisms, the reader should note that we are not dealing here with items that could be identified as Norse borrowings on verifiable phonological grounds, such as Old English sweġen ‘young man’, the vocalism of which indicates that it is a borrowing of Viking Age Norse +swainʀ (classical Old Norse sveinn), descended from Proto-Germanic +swainaz, which also yielded the native Old English cognate swān (Pons-Sanz Reference Pons-Sanz2013:28–31). Such items could not figure into Frank’s arguments because, as Matthew Townend observed of Beowulf, “its 3,182 lines contain not a single clear loanword from Old Norse” (2000:357). What we are dealing with, rather, are cases in which the language of Beowulf has allegedly been crafted in imitation of Norse diction for the literary purposes of characterization or allusion. In Frank’s view, the Beowulf poet was a late author who used his familiarity with the speech and the poetry of Scandinavians living in his contemporary England (ca. 890–950?) to represent more faithfully the Scandinavians inhabiting the 6th-century world of his poem. As Frank (Reference Frank and William1979:10) puts it:
The Beowulf poet … may have known what to do with a Nordicism when he found one; an occasional Danish idiom, placed carefully and significantly in his text, could be used to capture the flavor and texture of the sixth-century Scandinavian society he was describing.
Our aim in scrutinizing her alleged Nordicisms is thus to determine whether this literary explanation of each phenomenon possesses sufficient explanatory power to prefer it to the null hypothesis or an alternative hypothesis.
2.1. Missan (+ Genitive Object) and Mǣl is Mē tō Fēran
Discussing the verse miste merċelses ‘he missed the mark’ (l. 2439a), Frank (Reference Frank1981:132) contends that here the Beowulf poet “inserts what seems to be a Nordicism: missan (ON missa) ‘to miss, not hit’ with a genitive of the object, a usage common in Old Norse but otherwise unknown in this sense in Old English.”Footnote 2 Frank is right to identify Beowulf as the only Old English work in which missan unambiguously takes a genitive object, but her explanation of its uniqueness appears rather doubtful in the light of certain factors. One significant consideration not mentioned in Frank’s discussion is that cognates of missan govern the genitive case not only in Old Norse, but also in Old Frisian (Bremmer Reference Bremmer2009:100) and Old High German (Schützeichel Reference Schützeichel1969:129). There can thus be little doubt that the genitive was the case governed by this verb in Proto-Germanic. Consequently, the usage of missan in Beowulf requires no special explanation: It simply reflects the conservation of the original regimen of the verb. Outside of Beowulf, the word missan is recorded in the corpus of Old English only in three works, The Canons of Edgar, Instructions for Christians, and The Life of Saint Giles, none of which exhibits the verb taking an unambiguous genitive object.Footnote 3 In view of this distribution, the anomaly to be explained is not why missan should govern the genitive in Beowulf, but why it governs other cases in late Old English works. The simplest explanation for the discrepancy is to suppose that missan always governed the genitive when it meant ‘to fail to hit an object’, but governed other cases when it meant ‘to escape the notice of’. If this explanation were not credited, however, then the only other plausible way to make sense of the evidential distribution would be to see miste merċelses as a morphosyntactic archaism preserved in an early poem, but lost before the late Old English period.
The metrical behavior of mǣl ‘time, occasion’ in the verse mǣl is mē tō fēran ‘it is time for me to go’ (l. 316a) is another singularity in Beowulf to which Frank calls attention. Normally, a noun in this metrical position would be expected to alliterate and receive ictus, yet the line’s alliteration on /f/ requires mǣl is mē tō fēran to be scanned as a Type A3 verse with mǣl treated as a nonictic particle in a clause-initial dip.Footnote 4 Elsewhere in Beowulf and the rest of the Old English poetic corpus, mǣl exhibits the metrical behavior of a noun, but in this instance the poet appears to have regarded it as an adverbial particle equivalent to nū ‘now’ (see Suzuki Reference Suzuki2004:263, Fulk et al. Reference Fulk, Robert and John2008:137). Frank explains the peculiar metrical behavior of mǣl in this instance by construing it as a sign of Old Norse influence. She identifies two parallel verses in eddic poetry—mál er mér at ríða ‘it is time for me to ride’ (Helgakviða Hundingsbana II, l. 49) and mál er at þylia ‘it is time to perform’ (Hávamál, st. 111)—and uses them to interpret mǣl is mē tō fēran as an allusion to Old Norse poetry, a verse intended to lend “a kind of pagan Norse colouring” to the passage in Beowulf (Frank Reference Frank1981:132). The most serious obstacle facing Frank’s interpretation is that it requires belief in perfect intelligibility between speakers of Old English and Old Norse. In the scenario she has envisioned, the Beowulf poet and his audience possess no mere passing familiarity with Scandinavian speech, but such a firm command of the language that they are able to consume Old Norse poetry with full understanding of its metrical structure. This scenario appears rather unlikely in the light of recent research into Anglo-Scandinavian language contact, discussed below in section 3.1, which indicates that the degree of mutual intelligibility between speakers of Old English and Old Norse was adequate and limited, but not perfect.
An alternative explanation for the singularity of mǣl is mē tō fēran is to regard it as a poetic archaism. Two independent considerations support this interpretation. First, it is significant that the parallel verses adduced by Frank are taken exclusively from eddic rather than skaldic poetry, since the diction and metrics of eddic poetry establish it as the more conservative poetic tradition, descended from the same Proto-Germanic parent tradition that yielded the diction and metrics of Beowulf (Russom Reference Russom1998, Townend Reference Townend2015). Second, the presence of an uninflected infinitive (tō fēran) militates against the supposition that this verse represents a late coinage, since the distribution of the uninflected infinitive in the corpus of Old English suggests that it is a morphological archaism (Sievers Reference Sievers1885:255, 312, 482; Brunner Reference Brunner1965:§363, note 3; Pope Reference Pope1966:237; Bliss Reference Bliss1967:§44; Fulk Reference Fulk1992:§3; Neidorf Reference Neidorf2017:§45). In view of these considerations, it is reasonable to interpret mǣl is mē tō fēran as the reflex of a poetic formula that had taken shape during the Common Germanic period. The treatment of mǣl as an adverbial particle would then be limited to Beowulf simply because later Old English poets were less aware of this poeticism. That awareness of the variable metrical behavior of certain words should diminish over time is confirmed by the decline in instances of nonparasiting and noncontraction in late Old English poetry (Fulk Reference Fulk2007b). Just as the Beowulf poet knew that wuldor ‘glory’ and ġeþēon ‘prosper’ could possess variable metrical weight (compare ll. 1136a, wuldortorhtan weder, without parasiting, and 665b, hæfde kyningwuldor, with parasiting; and ll. 25b, man ġeþēon, uncontracted, and 910b, ġeþēon scolde, contracted), he also knew that earlier poetic practice licensed the treatment of mǣl as both an ictic noun and a nonictic particle depending on the metrical context. The simplest way to explain the Beowulf poet’s ostensibly unique possession of such knowledge is to assume that he operated before the authors of most of the other Old English poems.
2.2. Postposition of Þone
Postpositional use of the demonstrative pronoun sē is another remarkable feature of Beowulf, which distinguishes its language from that of the rest of the corpus of recorded Old English. Beowulf contains six verses that exhibit this feature, each of which consists of a masculine compound in the accusative case followed by demonstrative þone:
(1)
Frank explains the extraordinary postpositioning of þone in these verses by regarding them as “imitations of normal Norse word order,” artful attempts to recall varieties of postpositioning that she believes a late Beowulf poet could have heard in 10th-century Scandinavian speech (1981:133–134). She buttresses her claim by reasoning as follows:
… the postponed demonstratives of Beowulf might have sounded northern and darkly heathen to an Anglo-Saxon ear, the way the inversion in ‘Castle Dangerous’ somehow calls up the atmosphere of a Gothic romance, and that in ‘Eggs Florentine’ an elegant menu.
Having noted that this construction is confined to the final third of Beowulf, Frank reasons further: “The Beowulf poet may have copied a prominent linguistic feature of the Danelaw in order to colour the last third of his poem Geatish” (1981:134). There is, however, a glaring discrepancy between the creative freedom Frank would attribute to the Beowulf poet and the stereotyped character of the actual occurrences of the postposed demonstrative in the poem. That is to say, if the Beowulf poet were freely emulating a contemporary Nordic colloquialism, it is difficult to believe that he would incorporate it into his poem only in verses that feature masculine compound nouns in the accusative case. It is likewise doubtful that the poet would exercise this creative freedom by generating six verses that make use of only three different second compound elements: -hlem in 1a,e, -weard in 1b,f, and -wong in 1c,d. Of course, the simplest way to make sense of these apparently arbitrary restrictions on the poet’s freedom to postpone demonstrative sē is to suppose that all six of the verses in question reflect the poet’s awareness of one archaic formula.Footnote 5 Beowulf would then differ from the rest of the corpus on account of its chronological anteriority and its correspondent retention of archaic features lost before the composition of other works.
Indeed, there are additional reasons to believe that postposed þone is a syntactic archaism. Frank, in her argument for regarding this feature as a Nordicism, notes that similar postpositioning occurs in Norwegian and Swedish runic inscriptions from the 7th century on (1981:134). It should be noted, however, that there are also examples of postposed demonstratives and other determiners in much earlier North-West Germanic runic inscriptions, which reflect the most immediate ancestor of both Old English and Old Norse. Examples can be cited from three pertinent inscriptions (Antonsen Reference Antonsen1975:54, 50, 38) from the 4th and 5th centuries:
(2)
The postpositioning of determiners in these three inscriptions—hali hino ‘stone this’, mago minino ‘son mine’, and swestar minu ‘sister mine’— is precisely what is to be expected from an SOV language, such as North-West Germanic evidently was (see, among others, Lehmann Reference Lehmann, Frans and Herbert1972:244–246, Antonsen Reference Antonsen1981, Orton Reference Orton1999:300). In view of Greenberg’s (Reference Greenberg and Joseph1963) well-known demonstration of the correlation between post-positioning and SOV word order (see Dryer Reference Dryer1992), it is significant that 70% of the transitive clauses in the corpus of North-West Germanic runic inscriptions, which spans the 3rd to the 7th century, exhibit OV syntax (Antonsen Reference Antonsen1975:§7.1.1, Lass Reference Lass1994:220). Accordingly, the post-position of þone in Beowulf is better interpreted as a syntactic relic of the parent language embedded in a moribund formula than as a late borrowing between daughter languages. The contrary proposition is particularly difficult to credit since historical Old English was already in the process of transitioning to SVO order, which favors prepositioning rather than postpositioning of modifiers (see, among others, Mitchell Reference Mitchell, Malcolm, Douglas and Terry1994:164, Pintzuk & Taylor Reference Pintzuk, Ann, Ans and Bettelou2006, Ecay & Pintzuk Reference Ecay, Susan, Leonard, Rafael and Tom2016:154).
2.3. Miscellaneous Lexico-Semantic Data
Frank identifies various lexical and/or semantic parallels between Beowulf and Old Norse poetry. The precise significance of these parallels is not always made clear in her discussion. In most cases, she appears to interpret the parallel as a sign that the Beowulf poet either borrowed a word from Old Norse poetry or used a word in a particular sense in order to allude to Old Norse poetry. This section seeks to determine whether the hypothesis of late Norse borrowing is genuinely required to explain the various lexical and/or semantic peculiarities of Beowulf that figure into Frank’s argument.
First, we consider the word lofġeornost. Beowulf is exceptional in the corpus of Old English for using the word lofġeorn in an ostensibly positive sense. In other works, this word carries a negative meaning, such as ‘ostentatious’ or ‘vainglorious’, yet context suggests that the final word of Beowulf, which is lofġeornost (l. 3182b), is a term intended to eulogize the dead protagonist and should therefore be translated as ‘keenest for fame’ (Alexander Reference Alexander2003) or ‘most honor-bound’ (Fulk Reference Fulk2010).Footnote 6 Frank discerns Nordic influence in this semantic peculiarity. She observes: “Lofġeorn is documented in an unambiguously good sense in Scandinavian verse, however; Old Norse lofgjarn occurs in an eddic stanza praising the Vǫlsung hero Sigurthr as he passes through encircling flame” (1981:135, Neckel Reference Neckel and Hans1983:322). It is more economical, however, to interpret the peculiar use of lofġeornost in Beowulf as an archaism rather than a Nordicism. Since the positive sense of lofġeorn reflects its basic etymology, the negative sense it acquired in Old English is doubtless due to the pejoration it underwent in Christian contexts after the conversion of the Anglo-Saxons. Frank’s suggestion of Scandinavian influence is thus implausibly complex, as it supposes that lofġeorn underwent pejoration in Old English and then regained its original meaning through a process of amelioration induced by contact with Old Norse poetry.
A simpler explanation for the unique retention of the etymological meaning of lofġeorn in Beowulf is to suppose that this poem was composed close enough to the conversion for the semantic consequences of Christianization to have not yet taken full effect. Indeed, there are several other words in Beowulf that exhibit their etymological or pre-Christian meanings: synn, fyren, and bealu consistently possess their earlier senses of ‘violence’ or ‘hostility’ rather than the senses of ‘sin’ or ‘evil’ that they developed in Christian discourse (Robinson Reference Robinson1985:55–57); hrēow is used in the older sense of ‘sorrow’ rather than the theological sense of ‘penance’ (Shippey Reference Shippey and Antonio1993:173–175); scucca and þyrs remain terms for the material monsters of Germanic folklore rather than the spiritual devils of Christian demonology (Pascual Reference Pascual2014). Most apposite to the case of lofġeorn are the cases of ġylp and wlenċo: The former came to mean ‘boast’, but is used in its earlier sense of ‘vow’; the latter came to mean ‘arrogance’, but can still mean ‘bravado’ in Beowulf (Cronan Reference Cronan2003:400–401). In view of these related semantic archaisms, the hypothesis that lofġeorn reflects Scandinavian influence appears ad hoc. The more holistic way to interpret the positive valence of lofġeorn is to construe it as one of several indications that the composition of Beowulf antedated the semantic changes brought about by the gradual development of an institutionalized Christian vernacular discourse.
Another word that figures into Frank’s argument is þēodcyning. Noting that Old Norse þjóðkonungr is used in eddic and skaldic poetry in reference to royal overlords or national kings, Frank contends that the word þēodcyning in Beowulf is a “political catchword” linked to 10th-century developments, which the poet borrowed from Scandinavian sources (1981:129–130). The notion that þēodcyning is a late borrowing appears highly improbable in light of this word’s attestations in the corpus of Old English. Outside of Beowulf, it appears in Genesis A, The Fates of the Apostles, Soul and Body I, Riddle 67, Judgment Day II, The Death of Edward, The Letter of Alexander to Aristotle, and Napier Homily 29; there is also an Old Saxon cognate, thiodcuning, which is attested in the Hêliand (Townend Reference Townend2015:2–4). In view of this distribution, there can be little doubt that þēodcyning derives from a poetic compound that had been present in the North-West Germanic poetic lexicon, if not in the Proto-Germanic poetic lexicon. It cannot be regarded as a Nordicism; its use in Beowulf requires no special explanation.Footnote 7
Benċþelu ‘bench-plank’, an Old English compound attested exclusively in Beowulf, is another alleged Nordicism. Its Old Norse cognate, bekkþili, is attested exclusively in Eiríksmál, a skaldic poem probably composed in England around 955 to commemorate Eiríkr Bloodaxe, king of Northumbria. On account of the English connections of Eiríksmál, Dietrich Hofmann interpreted bekkþili as a borrowing of benċþelu (1955:45–46). Frank disputes his claim and suggests that the direction of borrowing should be reversed (1981:124, note 7). In our view, there does not appear to be any particularly compelling reason to regard either word as a borrowing. Yet if there were some reason to believe that one of the two terms must be the product of borrowing, the restriction of bekkþili to a single Old Norse poem that was probably composed on English soil must tilt the scales in favor of Hofmann’s position.
In a study of poetic conservatism in Old English poetry, Dennis Cronan (Reference Cronan2004) identified various words in Beowulf as lexical or semantic archaisms on account of their restriction to a corpus of six conservative poems, which consists of Beowulf, Genesis A, Exodus, Daniel, Maxims I, and Widsith. Fourteen words figured into Cronan’s study, including missere ‘half-year’, þenġel ‘prince’, eodor ‘lord’, fær ‘ship’, and heoru ‘sword’. In an article responding to Cronan’s study, Frank rejects his conclusions and argues that these five words are Nordicisms, not archaisms, since their Old Norse cognates “appear in 10th- and 11th-century skaldic verse, some of it composed and recited on English soil” (2008:9).
The suggestion that these five words are late borrowings from skaldic poetry appears improbable for several reasons. First, there is no phonological reason to regard missere, þenġel, eodor, fær, or heoru as borrowings of missari, þengill, jǫðurr, far, and hjǫrr; in fact, it seems rather doubtful that speakers of Old English possessed the linguistic knowledge required to convert hjǫrr into heoru. Second, the merely adequate level of mutual intelligibility between Old English and Old Norse (see section 3.1) raises grave doubts about the ability of an English audience to consume skaldic poetry at all, much less to borrow its poetic vocabulary. Third, the technical and quotidian character of nearly all of the words that Old English borrowed from Old Norse during the Viking Age renders it doubtful that five words drawn from the realm of poetic diction should have been borrowed as well.Footnote 8 Fourth, if these words were late borrowings, it is incredible that they should be attested in no verifiably late compositions, such as 11th-century annals of the Anglo-Saxon Chronicle, but are instead confined to poems that are widely assumed to be early compositions on account of their linguistic and metrical conservatism (Cable Reference Cable1981, Fulk Reference Fulk1992, Russom Reference Russom, Donka and Robert2002, Bredehoft Reference Bredehoft2014). A fifth and final factor militating against Frank’s position is that missere is used predominantly in poetic formulae: Seven of its eight attestations occur in the formulae of misserum frōd ‘wise in years’ (2× in Genesis A), fela missera ‘many years’ (2× in Beowulf, 1× in Exodus), and hund missera ‘fifty years’ (2× in Beowulf) (Cronan Reference Cronan2004:40). If this word were a late borrowing, one should not expect to find it in a formula (fela missera) known to two different poets, who presumably composed independently of each other. The more natural explanation for the appearance of fela missera in Beowulf and Exodus (and no other poems) is that this formula had a secure place in ancient poetic tradition and was consequently available to two early poets, but was forgotten at some point before the composition of later poetry.
The near confinement of missere to formulae undermines Frank’s position, while bolstering Cronan’s conclusion that this word, along with the others discussed in his study, are lexical or semantic archaisms. The attestation of their cognates in Old Norse poetry is to be understood as evidence not for late borrowing, but for the presence of the common ancestors of these words in the North-West Germanic poetic lexicon (Townend Reference Townend2015).
In summary, we conclude that Beowulf does not appear to contain a single linguistic feature that can reasonably be regarded as a sign of Old Norse influence. Most of the morphological, syntactic, lexical, and semantic peculiarities that had been explained by Frank as consequences of Scandinavian contact are much more economically and holistically explained as linguistic archaisms (mǣl is mē tō fēran, postposition of þone, lofġeornost, missere, þenġel, eodor, fær, heoru; possibly the construction missan + genitive object). Other items in question seem to possess no chronological significance whatsoever (þēodcyning, benċþelu). The absence of genuine Nordicisms in Beowulf is consistent with the conclusion reached on independent grounds (see section 3.2) that Beowulf was composed well before the first bands of Vikings had reached the British Isles.
3. Theoretical Considerations
Close examination of each alleged Nordicism in Beowulf eliminates the evidential basis for the hypothesis of Old Norse influence. In the present section, we adduce theoretical considerations that provide additional reasons to doubt this hypothesis.
3.1. Mutual Intelligibility
In his study of Anglo-Scandinavian language contact, Matthew Townend built on the theoretical models of Hockett (Reference Hockett1987) and Milliken (Reference Milliken1988) to reach conclusions that bear directly on the possibility that Old Norse could have influenced the language of Beowulf in the manner envisioned by Frank. His nuanced conclusion is worth citing at length:
… the available evidence points fairly unequivocally to a situation of adequate mutual intelligibility between speakers of Norse and English in the Viking Age. This intelligibility was adequate in the sense that it would seem to have been sufficient to preclude the need for interpreters or widespread bilingualism; this is not to argue that it was either perfect or instantaneous, and indeed it is important that ‘adequate intelligibility’ is under no circumstances equated with perfect intelligibility. (Townend Reference Townend2002:182)
The ambiguous character of Townend’s conclusion derives from evidence pointing in two directions: There is evidence, on the one hand, that speakers were able to establish phonemic correspondences between Old English and Old Norse; and there is evidence, on the other hand, that speakers generally failed to interpret the morphological configurations of words correctly. The success of phonemic identification is most clearly manifested in cognate substitution found in the Scandinavianization of toponyms and the Anglicization of Ohthere’s account of his voyages to King Alfred (Townend Reference Townend2002:43–68, 89–109).Footnote 9 The failure of speakers to analyze the morphology of words, meanwhile, is evident in the array of loanwords absorbed from Norse into English in which original grammatical suffixes have been reinterpreted as part of the stem, such as (a)thwart, scant, want, bask, and busk (Townend Reference Townend2002:200).Footnote 10 What the two types of evidence conspire to indicate is that speakers of Old English and Old Norse could communicate adequately without learning the other language in a systematic way, as they would need to do in the case of a non-Germanic language such as Latin, French, or Welsh. Such adequacy would have discouraged speakers from learning the other language well enough to obtain advanced proficiency in it.
Consequently, an implication of Townend’s study is that the success of much of the communication between speakers of Old English and Old Norse would have depended upon the use of a simplified syntax. This would have been the natural way for speakers to compensate for their inability to establish correct phonemic correspondences between inflectional endings. In view of Milliken’s (Reference Milliken1988:1) conclusion that “the degree of intelligibility between any two dialects correlates with the system of phonological correspondences that exists between cognate lexical items in those dialects,” it is significant that the most unpredictable divergences between Old English and Old Norse resided in the realm of inflectional endings. The centralization of vowels in unstressed syllables to schwa in Old English resulted in the loss of many inflectional distinctions, which would have obscured their genetic connection with the corresponding inflections in Old Norse (see note 10). Because the ability of speakers to identify cognate material was limited to the lexemic level, the flexible word order enabled by a rich inflectional system that would have characterized communication in a monolingual context would have needed to be abandoned in contexts of Anglo-Scandinavian language contact. An important consequence of this conclusion is that, although speakers of Old English and Old Norse would not have found it difficult to convey basic information to each other, they would have experienced considerable, and probably insurmountable, difficulties if they attempted to consume sophisticated works of poetic art featuring elevated diction and complex syntax.
Advances in the study of Anglo-Scandinavian language contact thus render untenable some of the foundational assumptions informing Frank’s studies. Several of Frank’s arguments for Old Norse influence on the language of Beowulf are predicated upon the belief that speakers of Old English avidly and expertly consumed Old Norse poetry. She writes, for instance, of “the poet’s and the audience’s acquaintance with Norse literary genres,” which provides, in her view, the literary rationale behind the implementation of many of the alleged Nordicisms (Frank Reference Frank1981:137). Similarly, she posits that “the Beowulf poet seems to count on his listeners’ familiarity with a contemporary skaldic genre, on their ability to recognize the changes he rings” (Frank Reference Frank1981:134). Considering the well-known syntactic complexity of skaldic poetry, it is improbable that speakers of Old English could understand skaldic poetry at all, much less that they could understand it well enough to observe subtle allusions to its diction.Footnote 11
Finally, the evidence considered above encourages the rejection of the most fundamental assumption behind Frank’s argumentation: “In tenth-century Anglo-Saxon England, two vernacular literary cultures— Old Norse and Old English—were in constant interaction, just as Anglo-Norman and Middle English were in the centuries following” (Frank Reference Frank and William1979:13). We reject this assumption for two reasons: First, the loanwords from Old Norse into Old English, unlike those from Anglo-Norman into Middle English, are of a predominantly nonliterary character (see note 8); second, unlike Middle English authors, who would have learned Anglo-Norman as a foreign language and acquired literacy in it, speakers of Old English would have felt no need to learn Old Norse as a foreign language and obtain the advanced proficiency required to consume sophisticated literary works at the speed of recitation. Accordingly, Old English literature was much less susceptible to Old Norse influence than Middle English literature was to Anglo-Norman influence.
3.2. Onomastic Conversion and Linguistic Archaisms
Because recent work on mutual intelligibility suggests that speakers of Old English and Old Norse could adequately convert personal names from one language into the other, it might be thought to lend support to the notion that the legendary material in Beowulf could have been transmitted to the poet by Scandinavian informants during the Viking Age. This possibility is often entertained or advocated in arguments for Old Norse influence on Beowulf. Frank (Reference Frank1981:214; compare Stanley Reference Stanley1981, Niles Reference Niles, Joanna and Hans2011), for instance, contends that the Beowulf poet’s “interest in and knowledge of things Scandinavian was the result of the Danish settlements in England and not part of a distant folk memory imported by the Anglo-Saxons from their continental homeland.” She notes, in support of her view, that skaldic poets were perfectly capable of converting the Old English name Æðelstān into its equivalent Old Norse form, Aðalsteinn, and then supposes that the Beowulf poet could easily have rendered Halfdanr into Healfdene and Hugleikr into Hyġelāc (Frank Reference Frank1981:124).
The evidence suggests that the mutual conversion of these particular names would indeed have been possible, since the process would have depended on the mere identification of regular phonemic correspondences. Yet there are several names in Beowulf whose Old English forms diverged considerably from their Old Norse cognates. Names such as Hrōðulf, Hrōðgār, Ēadġils, Onela, and Ongenþeo constitute decisive counterevidence to the proposition of Viking Age borrowing, since it is exceedingly doubtful that any speaker could have reconstructed these etymologically correct Old English forms from their Old Norse equivalents (Hrólfr, Hróarr, Aðils, Áli, Angantýr).Footnote 12 The authenticity of the proper names in Beowulf indicates that the poem’s legendary material was transmitted to England during the migration period, not the Viking Age.
Furthermore, while evaluating the hypothesis of Old Norse influence on Beowulf, the overall balance of archaisms and neologisms in the poem’s language must be borne in mind. Fulk’s (Reference Fulk, Christopher and Geoffrey2007a) survey of pertinent features reached the conclusion that no unambiguous neologism is structurally integral to Beowulf, while a wide array of conservative features distinguishes its language from that of works known to have been composed during the 9th, 10th, and 11th centuries. Archaic features of Beowulf include the following: its conservation of an archaic lexical stratum (Cronan Reference Cronan2004, Neidorf Reference Neidorf2013–2014); its adherence to Kaluza’s law (Fulk Reference Fulk1992, Neidorf & Pascual Reference Neidorf and Rafael2014); its high incidence of nonparasiting and noncontraction (Fulk Reference Fulk1992, Reference Fulk2007b); its regular use of the weak adjective without a determiner (Yoon Reference Yoon2014); its retention of the earlier i-stem genitive desinence in winia and deniga (Fulk Reference Fulk1992:§§279– 281); its preference for OV over VO word order (van Kemenade Reference Kemenade, Donka and Robert2002, Ecay & Pintzuk Reference Ecay, Susan, Leonard, Rafael and Tom2016:154, Russom Reference Russom2017). Some nonstructural (that is, orthographical) features of the poem are also notable archaisms, such as the spellings <ec> (for later <ecg>) and <þeo> (for later <þeow>) (Fulk Reference Fulk2014, Neidorf Reference Neidorf2017:§56). Indeed, when evaluating any hypothesis pertinent to the dating of Beowulf, one must take into account the now considerable body of textual evidence, in the form of errors and anomalies, that can be satisfactorily explained only under the hypothesis that the poem existed in written form centuries before the production of its extant manuscript (Lapidge Reference Lapidge2000, Neidorf Reference Neidorf2018). In view of the linguistic and textual indicators of archaic composition, the incompatible hypothesis of Old Norse influence on Beowulf would require overwhelming and unambiguous substantiation for it to be credible. At the same time, those same indicators of antiquity establish a strong initial probability of the correctness of our hypothesis that many of the alleged Nordicisms are actually additional archaisms.
4. Conclusion
Sara M. Pons-Sanz has argued that 20th-century scholarship on Anglo-Scandinavian language contact is often marred by a tendency to discern Old Norse influence in places where it is not genuinely to be found (2013:273–279). We concur with her judgment and extend her critique to the study of Beowulf. The three major recent works on interactions between Old English and Old Norse, namely, Townend Reference Townend2002, Dance Reference Dance2003, and Pons-Sanz Reference Pons-Sanz2013, have each, in their own way, eliminated many superfluous conjectures and painted a more realistic picture of the mechanisms and consequences of language contact. The present article follows their lead by offering the first systematic examination of Frank’s contention that Old Norse influence is evident in the language of Beowulf. Our analysis of each alleged Nordicism in the poem yielded no instance in support of this hypothesis, and our discussion of theoretical considerations bearing on its probability identified strong additional reasons to discount it. We conclude that Beowulf should under no circumstances be thought to exhibit Old Norse influence or result from contact with Scandinavian settlers during the Viking Age. In all likelihood, this poem was first composed and committed to parchment close to the year 700.
