This interface will soon cease to be publicly available. Use the new interface instead. Click here to switch over now.

Cookies on our website

We use cookies on this website, mainly to provide a secure browsing experience but also to collect statistics on how the website is used. You can find out more about the cookies we set, the information we store and how we use it on the cookies page.

Runic Dictionary

login: password: stay logged in: help




(Vol. I. Poetry for Scandinavian Rulers 1: From Mythological Times to c. 1035 > 2. General Introduction > 5. The diction of skaldic poetry > 5.1. Kenning > 5.1.1. Defining the term)

5.1.1. Defining the term

(subheadings only)

A. A modern definition of the kenning

In modern research on kennings, there have long been two opposing definitions: a broader one proposed originally by Rudolf Meissner (1921), in which the kenning is viewed as a circumlocution of more than one part; and another, narrower definition recognising only a metaphorical kenning, that is, one based on a metaphor. Its advocate was Andreas Heusler (1922). Some scholars of Old English have used the term kenning even more broadly (Bode 1886; Rankin 1909-10; Marquardt 1938; Singer 1933). Scholars of Old Norse literary history have paid special attention to the metaphorical kenning, finding in it the visual imagery and vividness that skaldic poetry otherwise seemed to lack. However, Meissner’s definition, founded on the extensive material of his own corpus, has prevailed. But his definition covers more than just the circumlocutory structure. The following definition can be gleaned from the introduction to his book on kennings: a kenning replaces a noun of ordinary discourse, consists of at least two parts and follows typical circumlocutionary patterns, on the basis of which the actual kennings are formed by variation. A similar definition is found in Krause (1930).

It is generally recognised that a kenning has multiple parts. A basic kenning consists of two elements that can appear in various syntactic and semantic arrangements. These constituents are traditionally referred to in English as the base-word and the determinant. The base-word assumes the syntactic position of the element being paraphrased. The two elements may form a genitive construction like brjótr baugs ‘breaker of the ring’ (Rloð Lv 8/1, 2VIII (Ragn 24)), or a compound such as hringbrjótr ‘ring-breaker’ (Egill Hfl 17/4V (Eg 50)), and the two structures are absolutely equivalent in skaldic style, both referring to a generous man. The choice may be determined in some cases by metre. Further, both structures can be developed to form extended kennings of three or more parts, described below.

Essential to the nature and poetic effect of the kenning is its substitutive character. It replaces a common word or proper noun in a sentence, as when the concept ‘giant’ is represented as gljúfrskeljungr ‘gully-whale’ (Ggnæv Þórr 1/4III). Constructions that use a kenning as a predicate (e.g. those linking it with verbs meaning ‘be’ or ‘be called’) infringe this substitutive rule and are rare. The same is true of kennings that appear in apposition to another word.

Scholars agree that there is essentially only a group of about one hundred referents for which kenning patterns or types exist, as Meissner’s 1921 corpus also shows. A kenning pattern can be described as a combination of two terms referring to a single referent, for example ‘fire of the sea/of water’ for ‘gold’. Specific kennings are developed by variation on a pattern, e.g. in this case eldr lagar ‘fire of the ocean’ (SnSt Ht 69/7III) for ‘gold’, and see further below Referents from a military milieu, such as warrior, weapons, wounds, corpses and ships, are particularly common, but so too are concepts like gold, earth, sea, sun, moon, animals, humans (as men or women) and body parts. As well as these generic concepts, both individual human beings and gods are also referred to using kennings. In Christian skaldic poetry new kennings were created for the new content according to the old paradigms (see §1.2 above).

One can conceive of the kenning patterns as a kind of corpus for poets, but it was also one that was indispensable to the listener’s understanding of the poetry. Poets created kennings by varying these patterns. Variation can be described linguistically as paradigmatic substitution. For each element of the pattern, a sort of paradigm of available substitutes exists for the poet to select from according to his needs. This paradigm consists not only of synonyms, but also of words belonging to a logical taxonomical hierarchy, and of kennings. That is to say, one word is replaced by another, related word that is often only similar in meaning, or by a kenning. (It must be emphasised here that – a few exceptions aside, to be discussed below – substitution with a kenning is only possible for determinants, not base-words.) For example, in a kenning like unnvigg ‘wave-horses’ (ÞjóðA Frag 1/8II), all synonyms for ‘horse’ (e.g. hestrfákrviggrdrasill) could stand in for the base-word ‘horse’, as could words for every other four-legged animal such as ‘elk’, ‘buck’ or indeed even ‘bear’ or ‘boar’. The same is true for the determinant unnr ‘wave’, which can itself be replaced by any term for water such as ‘sea’, ‘river’, ‘surf’ and so on.

Further complexity is possible, however, since the determinant (and very rarely the base-word) can be substituted by a kenning, thus producing an extended kenning. An example is viðir bála elfar ‘trees of the fires of the river [gold > men]’ (Sigv ErfÓl 1/1, 2I). Here ‘trees of gold’ refers to men, while the determinant, ‘fires of the river’, standing for gold, is connected to the base-word ‘tree’ through a genitive construction. Another example is viðir hjǫrþings ‘trees of the sword-assembly [battle > warriors]’ (Ólbjarn Lv 1/4V), where the base-word viðir ‘trees’ are again men, and the determinant kenning is a compound (‘sword-assembly’ standing for ‘battle’). Snorri refers to this technique as tvíkent ‘doubly paraphrased’ or ‘doubly modified’, and the term is also used in modern scholarship. However, the technique can be taken further still, when the determinant of the determining kenning is replaced by yet another kenning, producing complex kennings that Snorri calls rekit, literally ‘driven’. A kenning from Egill Lv 23/1-2V (Eg 30) will serve as an example: bǫrr ljósheims fjarðǫlna lyngs ‘tree of the bright world of the fjord-fish (plural) of the heather’. This kenning is resolved as follows: ‘fjord of the heather’ is land, its ‘fish’ are snakes, the ‘bright world of snakes’ is the gold snakes lie upon (as in the legend of Sigurðr and Fáfnir), and the ‘tree of gold’ is a man.

In rare instances, however, a kenning’s base-word can also be replaced by a kenning. Two very common kenning-patterns exemplify this observation. In the first example a warrior can be referred to as a ‘feeder of wolves or birds of prey’, wherein the base-word ‘feeder’ can appear either as a simplex, e.g. fœðir døkks hrafns ‘feeder of the dark raven [WARRIOR]’ (ÞjóðA Sex 16/2, 4II) and nistir varga ‘feeder of wolves [WARRIOR]’ (ÞhamMagndr 2/1II), or be paraphrased with a kenning, as shown by the following examples, in which ‘destroyer of hunger’ stands for the base-word ‘feeder’: eyðir gráðar ulfa ‘queller of the greed [FEEDER] of wolves [WARRIOR]’ (Arn Hryn 7/1, 2II), hungrdeyfir ins hvassa hrafns ‘hunger-appeaser [FEEDER] of the fierce raven [WARRIOR]’ (Steinn Nizv 6/1-2II). A further example of this device occurs in mythological kennings like farmr arma Sigynjar‘cargo of the arms [HUSBAND] of Sigyn <goddess> [= Loki]’ (Þjóð Haustl 7/2III).

In the present edition, the following principles apply to the explanation of these kennings. For a two-part kenning, the referent or paraphrased concept is given in small capital letters inside square brackets, e.g. Gerðr gollhrings ‘the Gerðr <goddess> of the gold ring [WOMAN]’ (Hharð Gamv 1/7, 8II). If the kenning refers to an identifiable human or mythical personage, the following notation is used: hefnir leifs ‘avenger of Óláfr [= Magnús]’ (Arn Hryn 14/1, 2II). For three-part (tvíkent) kennings, an arrow (>) denotes the further reference, e.g. lýsibrekka íss leggjar ‘the bright slope of the ice of the arm [SILVER > WOMAN]’ (Hfr Lv 15/1-2V (Hallfr 18)) or (in the case of an inverted kenning) randa rymskyndir ‘hastener of the noise of shield-rims [(lit. ‘noise-hastener of shield-rims’) BATTLE > WARRIOR]’ (SnH Lv 3/3II). This should be read as ‘the noise of shield-rims’ = ‘battle’, and then as ‘hastener of battle’ = ‘warrior’. The complex kenning cited above from Egill Lv 23V would receive the following explanation in the translation: [LAND > SNAKES > GOLD > MAN]. As seen here (and cf. §2. 4.), explanations of proper names such as Gerðr <goddess> or Hlǫkk <valkyrie> and literal translations of inverted kennings are added where appropriate.

The technique of substitution provides the skald with innumerable ways not only of finding the appropriate words to satisfy requirements of metre and rhyme, but also of highlighting content through allusion. Numerous scholars have regarded the variation on kenning patterns as the skalds’ main poetic achievement (Meissner; Mohr 1933; Krijn 1927; Einar Ól. Sveinsson 1947; Gutenbrunner 1963b; Marold 1983). Because the kenning is based on traditional paraphrasing patterns, and is not primarily the product of individual poetic inspiration, it was not highly regarded by literary scholars from the Romantic era until the mid-twentieth century. Holding up an individual style as the ideal, scholars bemoaned the kenning’s lack of originality and its stereotypical quality. However, modern literary studies have come to recognise a stereotypical ornamental language as both acceptable and interesting.

B. The terminology of Snorra Edda

The evidence of Snorra Edda is important to a discussion of Old Norse poetic terminology but difficult to interpret. The earliest use of the noun kenning in the context of a discussion of poetics is found in this treatise. It can also mean ‘emotion’, ‘feeling’, ‘knowledge’ and ‘teachings’ in Old Icelandic. In addition to the simplex kenning Snorri also uses compounds like sannkenning and viðkenning and terms related to kenning like heiti andkent heiti. There is also a verb kenna, along with prepositional constructions like kenna til or kenna við. Unfortunately there are differences between the way the words are defined and used in Skáldskaparmál and Háttatal, and they sometimes even seem to contradict one another (see Marold 1994b and 1995).

If one looks at the structure of Skáldskaparmál as a whole, one finds a clear twofold opposition drawn between periphrases of two or more parts on the one hand and simple poetic synonyms on the other. The classification of poetic terms either as kent ‘paraphrased’ or as ókent ‘non-paraphrased’ – found only in ms. U – corresponds to this opposition. However, not only does the introduction to Skáldskaparmál in mss R, W and contradict this, but so does the system described at the beginning of Háttatal. Both make a clear threefold distinction, but these, too, differ from one another.

Skáldskaparmál in R, W and begins by dividing the poetic devices according to whether they refer to metre or to diction. This also corresponds to the presentation in Háttatal. But then comes the following sentence, one that has had repercussions throughout skaldic studies (SnE 1998, I, 5): Þrenn *er grein skáldskaparmáls. Hver? Svá: at nefna hvern hlut sem heitir; ǫnnur grein er sú er heitir fornǫfn; in þriðja málsgrein er kǫlluð er kenning ‘“There are three categories of the language of poetry.” “What are they?” “To call everything by its name; the second category is the one called fornǫfn; and the third category of language is what is called kenning”’ (Translation modified from Faulkes 1987, 64). The method of kenning formation is then exemplified with kennings for Óðinn such as Sigtýr ‘Victory-Týr’, Hangatýr ‘Hanged-Týr’, and Farmatýr ‘Cargo-Týr’. But these are then referred to as Óðinn’s heiti and called kent heiti ‘paraphrased heiti’. Clearly, Snorri’s usage differs from that of modern Old Norse poetics, where heiti has established itself as a word denoting a poetic term that is a simplex. In Snorri’s work, even a kenning can be referred to as heiti.

By contrast, Háttatal offers a threefold distinction of another kind entirely: the verb kenna is introduced as part of the art of breyta/skipta með máli ‘varying the language’ (SnE 2007, 5). The basic concept is familiar from Latin rhetoric, which views poetic devices as variations on ordinary expressions. Háttatal lists three devices: 1. kenning (with the sub-categories kenttvíkent and rekit for kennings of two, three or more parts), which serves to ‘increase the vocabulary’ (SnE 2007, 5-7); 2. sannkenning (with the sub-categories sannkenningstuðning and tvíriðit, indicating the number of elements). Sannkenning is introduced here as a combination of a noun with a corresponding adjective or of a verb with a corresponding adverb. The role of the sannkenning is ‘to beautify and fill out the language’ (SnE 2007, 7); 3. nýgervingar ‘new constructions’ or ‘new creations’: These can be described as extensions of the metaphorical domain of one kenning into another and/or into the verb of the sentence. Nýgervingar are meant to demonstrate ‘knowledge and skill at speech’ (SnE 2007, 7).

The two terms fornǫfn (from Skáldskaparmál) and sannkenning (from Háttatal) appear once more in Skáldskaparmál (SnE 1998, I, 107), but here they seem to have a completely different meaning: Enn eru þau heiti er menn láta ganga fyrir nǫfn manna. Þat kǫllum vér viðkenningar eða sannkenningar eða fornǫfn ‘There are also those terms that are put in place of men’s names. We call these viðkenningar … or sannkenningar … or fornofn…’ (Translation modified from Faulkes 1987, 152). The viðkenningar, as is further explained, paraphrase persons through things they own, their relatives, or their social relationships (e.g. friend, enemy). The explanations and examples of sannkenningar show that they paraphrase individuals on the basis of their personal qualities. Two linguistic devices are employed for this: compounds with -maðr (e.g. spekimaðr ‘wise man’), or nouns derived from adjectives such as orðspekingr ‘wisely-speaking man’. In this case, the similarity to the definition in Háttatal is clear (cf. Marold 1995).

The main problem here is the term fornǫfn, which is also used at the beginning of Skáldskaparmál (see further Marold 1995). Neither there nor anywhere else do we find the term explained nor examples of it given. Some scholars (Brodeur 1952, 145; Halldór Halldórsson 1975, 24; Clunies Ross 1987, 24) take fornǫfn to be an umbrella term for sannkenningar and viðkenningar, but this leads to problems with the passage from the beginning of Skáldskaparmál, because the examples given there under kenning are actually viðkenningar, so there would then no longer be a clear distinction between the second group (fornǫfn) and the third group (kenningar). It thus seems plausible to follow Halldór Halldórsson’s (1975, 25) suggestion and view fornǫfn as a calque on the less common Latin term pronominatio for the more frequent antonomasiaViðkenning could then be viewed as a sort of loose translation of Latin antonomasia. Latin antonomasia encompasses both devices – epitheton and periphrasis – which correspond exactly to sannkenning ‘description (of a person) in terms of their qualities or essence’ and viðkenning ‘description of a person in terms of his possessions, relatives, or social relationships’. While viðkenning can be seen as a two-part periphrasis in the same sense as kenning,sannkenning cannot, because the examples of sannkenning given in Skáldskaparmál are derivations and compounds – and as such they are simplices. The same is true of the attributive use of adjectives as described inHáttatal. That explains why there are no sannkenningar among the kennings listed in Skáldskaparmál. These irregularities can possibly be explained if the distinction between kent heiti and ókent heiti in ms. U is taken to represent the original concept, which also corresponds to the structure of Skáldskaparmál. The original concept was altered in a later version, in which the author attempted to apply the concept of antonomasia from Latin rhetoric to Old Icelandic poetry with his calques viðkenning and fornǫfn (Marold 1994b). Altogether, it must be said that the terminology of Snorra Edda, because of its contradictory nature, does not lend itself to use in a modern definition of skaldic diction.

© 2008-