On the Oral-Formulaic Theory and its Application in the Poetic Edda: The Cases of Alvíssmál and Hávamál

 First of all, I will give a short introduction to the Oral-Formulaic Theory, also referred to as ‘Parry-Lord Theory’, from the names of its two major – and first – proposers. Their approach to literature was a totally new one, and changed radically the questions literature scholars have asked themselves ever since. It was originally intended to be a contribution towards the solution of the age-old Homeric question: for centuries, Homeric scholars had been debating over the identity of Homer and the authorship of the Iliad and Odyssey, which now came to be better understood as rooted in traditional oral poetry, and hence excluding any modern-like one-man literary composition (at least ex novo). The concept of originality, thus, had to be fully redefined, for there can be no real originality without authorship. The term compiler, then, started being used referring to written texts, and the tradition conceived as the sum of the contributions of single exponents, or singer of tales – who were the ones who kept it alive and dynamic – took over the role of the author.

These conclusions move from a theoretical apparatus based on an analogy, namely that the Yugoslav oral epic tradition as researched by Parry and Lord in the first half of the 20th century could offer a model for reconstructing other, and even much older, traditions, such as the one of Homer, which we cannot know much about directly. Crucial was also the particular technique used by the so-called singers of tales, which could be observed and understood in their natural environment, possibly resembling previous stages in other cultures, that is to say, in a context very often lacking the knowledge of reading and writing. Singers were expected to be fully illiterate, practising their art by the only means of mnemonic techniques of learning and performing by recreating what they had learned; the performance was thus a moment in the tradition, coinciding with the composition and the transmission, being handed over in that very moment to the audience (Lord:5). To be able to compose while performing, the singer of tales had to learn a specialised poetic language whose mastery did not allow him to “move any more mechanically within it than we do in ordinary speech” (36). Constituents of this language (or ‘poetic grammar’, as Lord calls it) and consequently essential features of oral poetry, are rhythm, formulas and themes.

The concept of formula was probably the most revolutionary one introduced by Lord and Parry. Parry’s definition of formula as “a group of words which is regularly employed under the same metrical conditions to express a given essential idea” (Parry:80) supplanted once for all a huge amount of inadequate modus appellandi for the Homeric “repetitions”, or “recurrent phrases” (such as “stock epithets”, “epic clichés” and “stereotyped phrases”, to name those quoted by Lord (30)). Formulas are “the phrases and clauses and sentences” of the poet’s specialised poetic grammar which he learns “by hearing them in other singers’ songs”; the process of memorisation is thus unconscious, and “follows the same principles as the learning of language itself” [Lord:36]. Oral poetry came therefore to be primarily conceived as formulaic poetry, in that an extensive use of formulas building up a paratactic narration could allow the poet to shape a song while singing (any use of the term ‘improvisation’ in this context (such as Paul Acker (xiii)) does not take into account Lord’s distinction, albeit rather obscure, between improvising and composing while singing; see also Harris (117), and, since it seemed somehow clear that poets’ memory was sometimes inevitably likely to fail, to fill in the metre with pre-made words combinations without having to interrupt the singing to think.

The ancient epos was also thought to have been constructed in oral composition in relatively short episodes, each one of the same status, interest and importance (this does not imply anything on the length of the performance itself though, which could be made up of many episodes in a row, for among the Yugoslavs a poet could sing for a whole night, Lord points out (14)). This was done by means of themes, being “the groups of ideas regularly used in telling a tale in the formulaic style of traditional song” (Lord:68).

Now, if on one hand this theoretical system has revolutionised the way to approach many different kinds of traditional literature, it has been widely criticised and came to be adopted as a source of methodological material, rather than a method itself. The problem was not its content, which has been widely accepted, but rather its claim to be a valid term of analogy. As John Miles Foley says in the article Orality, Textuality, Interpretation (1991), “the standard Parry/Lord approach to oral literature has significant but limited applicability to the texts with which they are most familiar”; that is due to the fact that “the Yugoslav oral epic tradition, especially the Muslim tradition on which Milman Parry and Albert Lord founded their analogy, may in important ways reflect the dynamics of, for example, the Anglo-Saxon and ancient Greek traditions, but it differs in that it arose and developed in a nearly textless environment, whereas “all medieval and classical works have of course reached us only in written form” (34). A text, according to Parry and Lord, does not really exist in purely oral cultures, but at the same time when someone transcribes an oral song into a text, it remains oral. This can occur in different ways and degrees, and although the poetic Edda is most certainly not a (direct) transcription, it is a shared opinion that traces of some original orality can be found in the written text. Because of this feature shared by most of European (and not only) folk epics (and related genres), as Acker reminds (surprisingly within brackets, as if it were an irrelevant clarification), “the term oral-formulaic now often refers not to the oral composition of a given work, but rather to its employment of formulas that are the inheritance […] of an oral culture” (xiv).

This led to probably the most important modification of the theory: the concept of (oral) formula, indeed, had been perceived as as much enlightening and outfitted in the beginning as ill-suited or even improper straight afterwards. The problem was mainly its close reference to and dependence on oral composition, or, as Acker states, that formulas were initially seen “as a necessary by-product of the improvisational composition of oral epic verse” (85). Lord’s functionalistic thought, indeed, did not allow the required attention to formulaic expressions in literary (non-oral) works, and, most importantly, in genres other than epics (whose common definition and boundaries could in my opinion be highly questioned, though) (ibid.). It has even been pointed out that formulas, intended as ‘bound phraseology’ can occur in ordinary speech (‘oral prose’) too, existing as ready-made entries in the poetic lexicon (Kiparsky:82) (kennings too can be somehow considered bound phraseology, though).

Parry himself was persuaded by a deeper analysis of the functionality of the formula to introduce the notion of formulaic system: “we may say that any group of two or more such like formulas make up a system, and the system may be defined in turn as a group of phrases which have the same metrical value and which are enough alike in thought and words to leave no doubt that the poet who used them knew them not only as single formulas, but also as formulas of a certain type” (Parry:85). This definition was then entirely applied by Magoun to Anglo-Saxon poetry, and he was followed by D.K. Fry, who highlighted the priority of the system at the expense of the simple formula. Criticism concerning these applications has been thoroughly expressed by Teresa Pàroli [615], who outlined the substantial differences between the Homeric and the Germanic poetictraditions in these terms:

The crux of the problem lies […] quite in the relationship between formula and metrical structure of the composition. The Homeric hexameter was so strict that the possibility of a variation is restricted to the substitution of a spondaic to a dactilic passage or vice versa, so that the link between the formulaic element and the metrical locus is in effect very tight […]. The situation in the field of Germanic alliterative poetry, on the other hand, is remarkably different; here the connection between the two hemistichs of a verse is given by an alliterating phoneme, and it is most of the times – but not always – carried by nominal elements, whereas the metrical model of each single colon appears to be much simpler, being it liable to be preceded by unstressed elements (anacrusis), to feature a highly variable number of syllables and even sometimes a number of strong cola higher than the ordinary two in those expanded verses attested in Anglo-Saxon poetry and so frequently occurring in Heliand. Not only, thus, does the Germanic situation present the possibility of formulaic variations which are impossible in Homeric poetry, but it also appears to be, in relation to the latter, extremely heterogeneous from the point of view of the metrics. This means that to apply without the proper regards definitions of formula and formulaic systems obtained by research on Homer to an epics that is metrically, formally and stylistically far from that one is not only fruitless, but also illegitimate[1].



Even though some scholars still do not share the fact that the only way to find an applicability for the Oral-Formulaic Theory in written texts such as the Poetic Edda is to subsume it to an ur-poetry (Acker: 86) for which we can use Lord’s and Parry’s model (mainly because of its inevitably axiomatic character), the evidence represented by the so-called pan-Germanic formulas as well as the common Germanic alliterative metre is nevertheless both striking and luring. For even if we assume that works such as the Poetic Edda merely ‘imitated’ or used the register of oral composition for stylistic reasons, it is important not to forget that “one cannot imitate a poetry that is not there”; “if there is no original, that is, no oral epic in formulaic language, there can be no imitation in written form” (Haymes:48). Now, the question is: is it possible to apply the Oral-Formulaic model on written composed texts which use ‘oral-like’ poetic language, typical themes identical to the ones which can be hypothesized for oral poetry, and the common Germanic metre? The dichotomy, then, that we find ‘oral’ formulas in demonstrably written works, does not take into account the mixed nature of the cultures of the Middle Ages (Acker: 86), where monasteries were oxbows of literacy floating in a wide ocean of orality, and therefore somehow indirectly witnessing the    surrounding         orality    (see    also      Kellogg’s       ‘quasi-literary period’ (Kellogg and Scholes:2)).

This was by no means possible, according to Albert B. Lord. According to the initial formulation of the Theory, in effect, literacy and orality are not only distinct, but mutually exclusive, and therefore allowing no intermingling such as the one we find in the Middle Ages. Lord believed that, as soon as oral songs in oral cultures began being memorised, this meant the killing of the ‘tradition’, regardless whether memorisation occurred through writing or purely mnemonic techniques. He therefore regarded as ‘written cultures’ any cultural environment where any kind of fixed text arose, even if texts were composed and passed on without the use of writing. This led some to aberrant monstra linguistica such as the concept of “written composition without writing” (Stolz and Shannon:176). It is evident that what Lord was thinking of was a state of mind literate cultures have (comparable to the fact that written cultures too can show oral-like behaviours) and which oral cultures too can happen to have, rather than what can be literally understood. However, it is also evident not only that this interdicts any serious analysis of works such as the Poetic Edda – which ultimately must be somehow considered in terms of orality – but also that such purely oral a culture is extremely unlikely to be found, because relatively short fixed texts or text portions can – and certainly do – arise and be memorised in just every culture.

Concerning this particular issue, in his article Literacy and Orality in the Poetic Edda (1991) Robert Kellogg claims that “in an oral tradition, poetic narratives of the Eddic sort exist as ‘texts’ only at the moment of performance”, whereas when they are not performed they exist “only as an abstract cultural competence”, i.e., as saussurean langue as opposed to the parole (96-7). This is doubtless very interesting in some particular cases – and I will come back to this – but, as we have seen, such a view of the matter derives ultimately from an outdated reception of the initial Lord-Parry system. Such an utterance can only be completely true if we assume that the poems were, as Lord wanted us to believe, continuously re-created out of the traditional stock of formulas and themes. However, we have no reason not to believe that at least some of the Eddic poems or rather portions of them could have existed as ‘fixed texts’ even before their Verschriftlichung. This is the case, I believe, of Alvíssmál: it is in my opinion very likely that this and other heitatö? could have circulated as a didactic text (Acker:64) and therefore been learnt by heart and used for educational purposes possibly in a time when the society was still predominantly oral. To deny this possibility would be to be blind before the evidence of fixed memorised texts – such as proverbs, prayers, or magic spells and charms, which usually are fossilised texts because of the belief in the effectiveness of the exact words, but also songs, which can be easily found in any culture.

Lastly, Lord’s claim that authorship plays no active role in oral and oral-derived poetry cannot be for many reasons any longer accepted. Indeed, the Lord-Parry strictly synchronic approach to South Slavic epics defined the poet’s poetic grammar as a static technical language, meant to be nearly totally impersonal. This was also the main argument against the theory: poems such as the Iliad, the Odyssey and Beowulf had to respond to aesthetic categories – suggesting an exceptionally gifted personality behind them, which the Oral-Formulaic Theory in its initial formulation simply disregarded. It has now become clearer that, even considering such poems within an oral framework, the language of the tradition is never completely monolithic, but obeys to diachronical, diatopical, diastratical and, in a word, idiolectal needs, which ultimately change from poet to poet (Foley:38). Even moving within the tradition he inherited, a poet can have his own ‘custom’ poetical grammar, just as ordinary speakers of the same language all have their own idiolect. The poet is not a seer mystically possessed by a god-like tradition, but an independent individual who can also express his personality and his own aesthetic views through his art (Pàroli:617). This is particularly important in the context of Eddic poetry, which is nothing but an unitary work, each of whose poems presents unique characteristics of time (and perhaps place: see Bugge’s claim that the Edda was brought to Iceland from the British Isles!) of composition, occasion, genre, metre, register, and which ultimately hints back at different authors.

As we have seen, in the field of Eddic studies it is extremely important to find a way to apply what has been achieved so far in terms of Oral-Formulaic possibilities of approaching texts which takes into account a wider scenario than that first outlined by Parry and Lord. Firstly, because these texts have to be considered in their context, and secondly because a tailor-made solution would be in this case as much tailor-made as Parry-Lord’s approach to South Slavic poetry.

We have seen that not only could part of the Eddic material have been composed orally already in a more or less fixed form, but that the role played by individual poets who were not simply representatives of the tradition but of themselves as well could have been decisive in the process of fixing the text. It seems that precisely this kind of texts has been examined by Paul Acker in his survey of formulaic material in Alvíssmál[2], which, as he says, “offers an unusual opportunity to observe a poet’s variations within a highly schematic formal structure” (66). The poem, composed in ljóðaháttr, consists of a wisdom test the dwarf Allvíss (all-wise) must pass in order to get to marry Þórr’s daughter, Þrúðr; he is therefore interrogated by the god on the names or heiti (poetic synonyms) for thirteen things, until the dawn comes and he is apparently petrified by the sunlight, according to Þórr’s intentions. The interrogation and the answers follow a very strict pattern with minimal variations, so that most of the times only the concept-word (sú iörð, sá eldr, sá himinn, etc.), filling a specific metrical slot, and few other elements change in the stanza. I would like to highlight here the particular importance of such other elements, in that they build up a “slot-filler system” (Acker:64) which can perfectly be filled ad libitum and potentially ad infinitum by the poet. This provides him with a very useful tool to remember not only poetic synonyms (and kennings) for substantives, but also the verbs and phrases which are most likely to accompany them in a context of oral composition (iörð : liggr, eldr : brenn, etc.). This is exactly the kind of poem which, although on the one hand featuring a very fixed structure, is on the other hand extremely flexible, in that it can be reduced or extended according to the needs of the poet, and its constituents can be substituted with anything the poet considers relevant, leaving him a great autonomy and possibility of personal expression – or rather exercise, in this particular case.

A poem like Alvíssmál would have been considered in a Lordian approach as a ‘common heritage’ of a given oral civilisation, for because of its schematic form its origin can be hardly traced back to a particular individual who wanted to express his own aesthetic views. However, as we have seen, it allows the poet willing to make use of it to do so, and in this direction, I believe, one is to interpret the variations – albeit minimal, in this case – which are offered in the extant text. Ultimately, not directly Alvíssmál but rather, perhaps, its model (which we cannot expect to be much different from it) is doubtless to be considered as a training aid for poets, which may well have its origin in the oral world of poetry. In this particular case I can well agree with Kellogg (96-7) in considering a hypothetical ur-Alvíssmál (consisting of the mere basic formulaic stanzas) as an ‘abstract entity’, or langue, whereas the extant poem seems to be its concrete realisation brought into being by an individual conscious of his own status of poet.

The Italian scholar Teresa Pàroli, referring to a number of specific works on the topic, wrote about the implications of the use of verba dicendi in Eddic poetry. Unfortunately, her interesting analysis cannot be applied, I believe, to Alvíssmál, which though makes a large use of formulae containing this kind of verbs. Indeed, all Þórr’s recurrent enquiries in the poem begin with the imperative phrase segðu mér þat, Allvíss, containing the verbum dicendi segia. Her main point is that Eddic poems containing this particular in formulaic direct-speech introductions can be considered more recent than those in which qveða and/or mæla are predominant (Pàroli:60). In the case of Alvíssmál, in fact, there are no direct-speech introductions at all (that is why she does not include it in the chapter dedicated to Eddic poetry), and the phrase with which þórr’s enquiries begin is already part of a direct-speech conversation; moreover, the presence of the dative mér and the use of the imperative mood would perhaps in any case, I think, require the verb segia. That means that in the case of Alvíssmál, unlike in other Eddic poems, formulae of this kind cannot provide us with any sure information about the time of its composition.

As we have seen, the difference between Alvíssmál and other poems such as Hávamál, which can rather be considered gnomic, consists in its being a practical tool for poets, which eventually can have been given the status of an all-accomplished poem at the moment of its Verschriftlichung. However, Hávamál presents some interesting formal similarities with Alvíssmál, which have to be taken into consideration in the framework of our oral-formulaic approach.

The structure of Hávamál, which has been defined as a “synthesis of social wisdom and Odinic myth” (Larrington:19), is more complex and protean than that of Alvíssmál, and each of its part requires therefore special attention. According to a widespread scholarly practice, the poem can be devided into five sections, according to the style and subject-matter of the strophes:

I. The Gnomic Poem, 1-103.
II. The “Gunnlöð Episode”, 104-110.
III. Loddfáfnismál, 111-37.
IV. Rúnatal, 138-45.
V. Ljóðatal, 146-64.

Each of the parts of which the poem is made up can be attributed to different times and authors. Even though the subject-matter variates consistently, the genre to which the poem can be ascribed, that of wisdom poetry, gives a rather coherent and unitary shape to smaller units which apparently follow no clear logical progression: wisdom poetry, indeed, “has no prescribed form; no narrative or chronological principle, by which the poet may order his gnomes, is inherent in the material”, as Larrington points out (65).

As far as our purposes in this essay are concerned, I will briefly comment on only two parts of the poem, i.e., what is usually referred to as ‘The Gnomic Poem’ and Ljóðatal. The first probably originated from a basic sequence of ljóðaháttr verses already present in folk tradition and already linked by verbal repetitions, or by theme (Larrington:18); it displays a catalogue of maxims building up a sort of archetypical story or a situation familiar to everyone in medieval Iceland, that of a traveller arriving as a guest at a farmhouse hall teeming with strangers. Interestingly, there is no real narration, but rather a clever juxtaposition of sayings depicting such typical and anonymous situation. The latter is “a list of eighteen spells, whose contents are briefly sketched, but whose text is never given” (62); here there is no (pseudo-) narrative framework as in the Gnomic Poem, but the spells are organised by enumeration, a structural device common in wisdom poetry. In the Ljóðatal, “the poet skilfully uses the alliterating ordinal to provide a mnemonic hook for the key word of the spell’s subject” (62-3); an example of this formulaic device is 1551-3 (the alliterating words are in italics):

þat kann ek it tíunda:

ef ek sé túnriðor

eika lopti á,

These two parts of Hávamál are interesting from an oral-formulaic point of view because of the different techniques of memorisation: the narrative frame on the one hand, and the enumeration plus alliteration on the other. As we have seen in Alvíssmál, these two texts can represent a later stage of oral wisdom poetry whose form could possibly be somehow already fixed in a hypothetical previous oral stage. In the particular case of Ljóðatal, then, it may well be that the the spells are not ‘genuine’, and thus the expression of literacy, as Larrington suggests (63) , in the sense that they were not directly mentioned. However, the very fact that such an allusive text arose could mean that at that stage of the civilisation which produced them the belief in the indirectly mentioned spells was still alive. The Christian compiler of the Old English Charms or the Old High German Merseburger Zaubersprüche, in fact, could perhaps transcribe the charms without any fear of their effectiveness, whilst an Icelandic poet, I believe, would have needed a mnemonic tool useful to remember them, but at the same time carefully avoiding unleashing their power.

The role of the author (or rather compiler) of the Hávamál is therefore to be primarily identified in the literate person who organised such an heterogeneous material into the coherent poem we know today, according to the leading topic of folk wisdom associated to and personified in Óðinn. To the single parts of the poem, which have their clear root in oral tradition, however, we can (albeit carefully) attempt an application of an oral-formulaic approach, resulting in the interesting implications we have seen.


Acker, Paul: Revising oral theory : formulaic composition in Old English and Old Icelandic verse. Garland studies in medieval literature vol. 16. Garland reference library of the humanities. vol. 2104. New York : Garland, 1998.

Foley, John Miles: Orality, Textuality, and Interpretation. In: Vox intexta. Orality and Textuality in the Middle Ages, edited by A.N. Doane & Carol Braun Pasternack. Madison: The University of Wisconsin Press, 1991. p.34-45.

Harris, Joseph : Eddic poetry. In: Old Norse-Icelandic Literature. A Critical Guide., edited by Carol Clover and John Lindow. Ithaca and London: Cornell University Press, 1985. p.68-156.

Haymes, Edward R.: The Germanic Heldenlied and the Poetic Edda: Speculations on Preliterary

History. In: Oral Tradition vol. 19, n.1 (http://journal.oraltradition.org/issues/19i/haymes.pdf)

Kellogg, Robert : Literacy and Orality in the Poetic Edda, 1991. Madison, The University of Wisconsin Press In: Vox intexta. Orality and Textuality in the Middle Ages, edited by A.N. Doane & Carol Braun Pasternack. p.89-101.

Kellogg, Robert and Scholes, Robert: The Nature of Narrative. I serie: A Galaxy Box. London : Oxford University Press, 1968.

Kiparsky, Paul: Oral Poetry: Some Linguistic and Typological Considerations . In: Oral Literature and the Formula. Ed. Benjamin A. Stolz and Richard S. Shannon. Ann Arbor: Center for the Coordination of Ancient and Modern Studies, University of Michigan, 1976. pp. 73-106.

Larrington, Carolyne: A store of common sense : gnomic theme and style in Old Icelandic and Old English wisdom poetry. Oxford English monographs. Oxford : Clarendon Press, 1993.

Lord, Albert B.: The Singer of Tales. Harvard Studies in Comparative Literature, 24. Cambridge, MA: Harvard University Press. Rpt. New York: Atheneum, 1968 et seq. Rpt. Harvard University Press, 1981.

Neckel-Kuhn = Neckel, Gustav (hrsg.von): Edda : die Lieder des Codex Regius nebst verwandten Denkmälern. Germanische Bibliothek. 4. Reihe, Texte. Heidelberg : Carl Winter, 1983.

Parry, Milman: Studies in the Epic Technique of Oral Verse-Making. I. Homer and Homeric Style. Harvard Studies in Classical Philology, 41:73-147. Rpt. In The Making of Homeric Verse: The Collected Papers of Milman Parry. Ed. Adam Parry. Oxford: Clarendon Press, 1971. pp. 266-324. Rpt. in German in Homer: Tradition und Neuerung. Ed. Joachim Latacz. Wege der Forschung, Band 463. Darmstadt: Wissenschaftliche Buchgesell-schaft, 1979. pp. 179-266.

Pàroli, Teresa: Sull’elemento formulare nella poesia Germanica antica. Biblioteca di ricerche linguistiche e filologiche, no. 4. Istituto di Glottologia, Università di Roma, 1975.

Stolz, Benjamin and Shannon, Richard (eds.): Oral literature and the formula. Ann Arbor : Center for the Coordination of Ancient and Modern Studies, University of Michigan, 1976.

[1] All translations from Italian henceforth are mine, D.F..

[2]All quoted passages from the Poetic Edda are from Neckel-Kuhn’s edition.