This interface will soon cease to be publicly available. Use the new interface instead. Click here to switch over now.
Already as early as the 1220s, Snorri Sturluson, in his Háttatal, commented on the practice of cliticisation in skaldic poetry, that is, the suffixation (or prefixation) of an unstressed, monosyllabic word onto another, usually with the loss of an intervening vowel. He states (SnE 2007, 8): Fjórða leyfi er þat at skemma svá samstǫfur at gera eina ór tveim ok taka ór annarri hljóðstaf. Þat kǫllum vér bragarmál ‘The fourth licence is to shorten syllables in such a manner as to make one out of two and remove the second vowel. We call that bragarmál (“poetic speech”)’. As an example of this licence, Snorri cites Þmáhl Máv 1/1V Varðak mik, þars myrðir, literally ‘defended-I myself where murderer’ (rendered in ms. Tˣ(48r) as ‘Vardac mic þars myrdir’; here and elsewhere emphasis has been added). Here the personal pronoun ek ‘I’ and the particle es have been cliticised onto the verb varða ‘defended’ and the adverb þar, literally ‘there’ respectively (varða ek > varðak; þar es > þars). Although some manuscripts preserve bragarmál in their renditions of skaldic poetry, such contractions are usually reproduced as two words. In ms. 448(38) of Eyrbyggja saga, for example, Þórarinn’s line is given as ‘Varda ek mic þar er myrdir’, creating a hypermetrical verse with eight rather than six syllables.
In 1878 Eduard Sievers collected the instances in which superfluous syllables cause unmetrical lines in syllable-counting dróttkvætt, and most subsequent editors of skaldic poetry, including the editors of the SkP volumes, have adopted his principles (note that some of them obtain only in poetry composed prior to 1200 or 1250). Sievers’s editorial guidelines are subsumed under the categories below.
When required by the metre, cliticised forms are used even though the orthography of the manuscripts may not reproduce these forms. Hence the relative particle (e.g. þeim es > þeims dat. pl. ‘who’, sá es > sás m. nom. sg. ‘who’ etc.), conjunctions (e.g. svá at > svát ‘so that’, því at > þvít ‘because’ etc.), es ‘is’, the 3rd pers. sg. pres. indic. of vesa ‘be’ (hér es > hérs ‘here is’, hann es > hanns ‘he is’, etc.), as well as ek ‘I’ (varða ek > varðak ‘I defended’) are given in their contracted forms when necessary (see Sievers 1878, 477-9, 489-95, 497-504). As far as the relative particle and the 3rd pers. sg. pres. indic. of vesa (both es) are concerned, these words are not contracted in the editions of post-1250 poetry (when ‑s has been rhotacised to ‑r). In the lemmata in the Readings sections of the SkP editions, the diplomatic forms are given in parentheses (e.g. ‘ þars (‘þar er’)’).
B. Deletion of superfluous pronouns
Later redactors and scribes frequently insert superfluous words (such as personal pronouns) for syntactic simplification (see Sievers 1878, 467, 512-13). If such pronouns result in unmetrical lines, they are deleted. In Grani Har 2/8II vel njóti þess — Jóta, for example, the Morkinskinna version (Mork(9r)) reads ‘vel nioti hann þess iota’, literally ‘well may-enjoy he that of Jótar’. Here a scribe inserted the extrametrical pronoun hann ‘he’, creating a heptasyllabic line. That is also the case in Valg Har 5/3II farðir goll ór Gǫrðum, where ms. H(28v) reads ‘færðir þv gvll or gorðvm’, literally ‘brought you gold from Russia’ (with extrametrical þú ‘you’), and in the Morkinskinna (Mork(16v)) version of SnH Lv 6/3II sýnts, at sitk at Ránar, rendered as ‘synt er at ek sitc at ranar’, literally ‘clear is that I sit-I at Rán’s’, ek ‘I’ has been added. In the latter instance, the pronoun, which was originally cliticised onto the finite verb (sitk lit. ‘sit-I’) was retained, but another, unmetrical ek ‘I’ was added by a scribe (cf. the Mork(24r) version of Mberf Lv 5/5II ‘An ec þott ec eigi finnag’, literally ‘love I although I not find-I’). When a pronoun was added by a later scribe, the pronoun that originally cliticised onto the finite verb was sometimes dropped, and the normalisation entails the deletion of the extrametrical pronoun and the restoration of the enclitic ‑k (see Sievers 1878, 504) as in Bǫlv Hardr 7/1II Heimil varð, es heyrðak, literally ‘Granted was, as heard-I’, where the main manuscript Kˣ(537r) reads ‘Hemil varð er ec heyrða’, literally ‘Granted was as I heard’.
C. Deletion of syllables and restoration of earlier forms
Scribes often failed to understand an earlier archaic form of a word, such as a finite verb with the enclitic negation ‑(a)t (see Sievers 1878, 495-7), and replaced it with a more transparent construction. For example, in the Morkinskinna version (Mork(37v)) of ESk Ingdr 4/1II Myndit seima sendir, ‘Myndi eigi seima sendir’, literally ‘Would not gold’s dispenser’, an illicit disyllabic negation (eigi) has replaced the negative suffix ‑t, whereas the Fagrskinna Aˣ version (FskAˣ(384)) ‘Myndi at sæima senndir’ provides the older negation (although not cliticised Myndi-t). Sometimes scribes would replace old, short adverbial forms with the more familiar long forms, as in the Fagrskinna Bˣ version (FskBˣ(65r)) of Valg Har 7/7II brast ríkula ristin, which reads ‘brast rikuliga ristinn’, ‘split richly engraved’. Scribes also confused the old unstressed prepositions ept ‘after’, fyr ‘before’, und ‘under’ and of ‘over, above’ and the adverbs (or stressed enclitic prepositions) eptir ‘after’, fyrir ‘before’, undir ‘under’ and yfir ‘over, above’, creating unmetrical lines (see Sievers 1878, 479-86), as in the Hrokkinskinna rendition (Hr(54va)) ‘hardgiædr undir midgardi’ of Arn Hardr 16/2II harðgeðr und Miðgarði ‘harsh-minded under Miðgarðr’. Later scribes also added suffixed definite articles to nouns (see Sievers 1878, 513), as in Morkinskinna’s (Mork(24r)) version of Sjórs Lv 2/2II veldr því karl í feldi ‘velldr þvi carl ifelldinom’, literally ‘causes that man in cloak-the’, and such articles are deleted if the metre warrants this. In general, cliticisation of the definite article does not occur until the thirteenth century – the only example from the eleventh century, Arn Hryn 15/3II Yggjar veðr, meðan heimrinn byggvisk, literally ‘Yggr’s wind-storm, while world-the is-peopled’ is suspect, because a scribe may well have replaced the archaic expletive particle of with the more familiar enclitic definite article (see Finnur Jónsson 1901, 80).
D. Reintroduction of hiatus forms
Words that contain two vowels in hiatus, that is, two adjacent vowels, are usually contracted in the manuscripts, and the old forms are reintroduced if the later forms produce hypometrical lines (e.g. blám > bláum dat. pl. ‘blue’; see Sievers 1878, 514-17). It is not clear exactly when such words were contracted, but judging from Snorri’s comments in Háttatal (SnE 2007, 7) and the examples he provides in SnSt Ht 7III, this must have happened by 1220 (see also ANG §130). Accordingly, such a pentasyllabic line as the Kˣ(544v) version of Þfagr Sveinn 9/8II, which reads ‘ofꜵl bǫndr dvꜵldo’, literally ‘unsaleable farmers prevented’, has been normalised to ófǫl búendr dvǫlðu.
In poetry composed in metres that are not syllable-counting and contain hypermetrical lines, such as málaháttr and ljóðaháttr, these principles do not obtain.
The attempt to establish guidelines for the normalisation of ninth- to fourteenth-century poetry is problematic because it is difficult to pinpoint exactly when a particular phonological or morphological change took place – changes occurred gradually over time and at different times in different parts of Scandinavia. In fact, scholars often derive the dates for many phonological and morphological changes from the skaldic corpus; in particular from words carrying internal rhyme (see, e.g. Konráð Gíslason 1895-7, II, 145-209, 297-305; Finnur Jónsson 1901). Furthermore, the poetic language was conservative, and archaic forms continued to be used in poetry long after they had disappeared from everyday speech. With these caveats in mind, it nonetheless seemed both prudent and necessary to develop a set of guidelines that attempt to capture the chronology of phonological and morphological changes in Old Norse-Icelandic from the ninth to the fourteenth centuries. For the sake of convenience, and based on the linguistic evidence, we distinguish between the following periods: the earliest period (c. ninth century-1200), the period 1200-50, the period 1250-1300 and the period 1300-1400. Standard normalisations that have been adopted across these periods involve the unstressed endings ‑i / ‑i- and ‑u / ‑u- (not ‑e / ‑e- and ‑o / ‑o-) and initial <i> in the definite article (inn, ‑inn, etc. not enn, ‑enn, etc.).
A. The earliest period (c. ninth century-1200)
Most of the phonological changes that occurred during this period are evidenced in stressed syllables carrying internal rhyme. As far as the vowels are concerned, long vowels followed by a consonant cluster were shortened (e.g. mínn > minn ‘my, mine’, háski > haski ‘peril’, see ANG §127; Finnur Jónsson 1901, 76). Poetry composed during the eleventh century contains both long and short vowels in that position, for example Arn Þorfdr 21/4II mínn auðgjafa sína, literally ‘my wealth-giver their’ versus Arn Hardr 4/4II hugi minn es þat sinni, literally ‘thought mine is that his’ (both c. 1065-6) and Arn Þorfdr 11/2II bráskat þat dœgr háski, literally ‘ceased-not that day peril’ (c. 1065-6) versus Sigv Lv 23/6I vask til Rúms í haska, literally ‘was-I to Rome in peril’ (after 1030). Long [i:] occurs in poetry as late as 1150-1200, e.g. Gamlkan Has 60/6VII hug mínn siðir þínir, literally ‘mind mine virtues yours’, but because this poem otherwise consistently rhymes inn : inn (e.g. Has 2/4VII, 14/4VII, 18/4VII, 51/2VII, 53/2VII, 57/2VII) and ín : ín (e.g. Has 1/8VII, 5/4VII, 7/4VII, 8/8VII, 19/8VII, 23/6VII, 55/6VII, 58/6VII), this isolated occurrence could represent a poetic licence and has been treated as such in SkP VII. In the poetry predating 1200, the SkP editions use the short vowels in such words as minn, þinn, etc. unless warranted by internal rhyme. The [o:] in gótt ‘good’ does not appear to have been shortened until the thirteenth century (see LP: góðr).
Stressed > ó when nasalised (e.g. ntt > nótt ‘night’, ANG §116). This change is attested as early as 1050 (Anon (HSig) 2/6II vask í nótt fyr óttu, literally ‘was-I last night before dawn’) and seems to have been well established by c. 1070, for example in Steinn Óldr 5/8II Óláfr borinn sólu, literally ‘Óláfr born sun’, although that poem also has one example of the older form láfr (Óldr 7/6II láfr konungr hla, literally ‘láfr king certainly’). Again, the SkP editions use the forms as dictated by internal rhyme, but Óláfr is used consistently after 1100.
As far as consonants are concerned, ð > d in the environment | [+ long syllable] l, n- (e.g. hvílð > hvíld ‘rest’, ANG §238.1b). This change is pre-literary and attested as early as the ninth century (cf. Bragi Rdr 4/4III hendr sem fœtr of kenndu, literally ‘hands as well as feet recognised’). The <d> spelling has been adopted in all SkP editions. In the twelfth century we see the first examples of the rhotacised forms of the verb vera ‘be’ (older vesa), the first being ESk Lv 1/3II ert (svát eigi skortir), literally ‘you are (so that not lacks)’ (c. 1115; see also ESk Hardr II 2/2II). This change seems to have occurred earlier in Norway than in Iceland (Finnur Jónsson 1901, 93), and Snorri (c. 1223) uses both the rhotacised and the non-rhotacised forms (SnSt Ht 58/1III, 82/5III, 87/7III). The SkP editions retain the non-rhotacised forms (vesa ‘be’, es ‘is’, est ‘are’, vas ‘was’, etc.) in poetry pre-dating 1250, unless the rhotacised forms are required by internal rhyme. The non-rhotacised form of the relative particle (es) is also used in the editions of poetry predating 1250.
B. The period 1200-50
During this period (or a little before 1200), a, o, ǫ, and u in stressed syllables were lengthened in the environment | ‑lf, lg, lk, lm, lp (occasionally ‑ln, ls) in Icelandic (ANG §124.3), causing such changes as halfr > hálfr ‘half’, holpinn > hólpinn ‘helped’, ulfr > úlfr ‘wolf’. Around the same time, all geminates (except ll, mm, nn, rr | –l, m, n, r; gg, kk | –j, w) were shortened before a consonant (ANG §284), producing such forms as kendu (< kenndu ‘knew’), alt (< allt ‘all’). For nasalised > ó, see above.
C. The period 1250-1300
Around 1250, coalesced with á (srum > sárum ‘wounds’), œ with æ (bœr > bær) and ǫ with ø, represented orthographically as ö (mjǫk > mjök) in stressed syllables in Icelandic (ANG §§107, 120, 115.2). The consonant cluster [ts] represented orthographically as <z> (Gizurr) becomes [ss] (Gissurr; ANG §274.2). For the rhotacisation of ‑s / ‑s- in forms of the verb vesa / vera ‘be’ and in the relative particle, see above. In terms of morphology, there is an increased use of the mediopassive ending ‑z(k) (< ‑sk), although ‑sk is often retained in monosyllabic mediopassive verbs (e.g. komsk ‘comes’) in manuscripts after 1250. In poetry post-dating 1250, the SkP editions adopt the ending ‑z(k) uniformly (for the mediopassive, see ANG §§543-4 and Kjartan G. Ottósson 1992).
Note that the principles for the normalisation of fourteenth-century poetry are outlined in §9 of the Introduction to SkP VII, and they will not be addressed here. For the normalisation of the language of the runic corpus, see Introduction to SkP VI.
The discussion below does not attempt to be exhaustive; rather, it gives an overview of the most important changes.
Note that this date, as most others, is approximate.
The SkP editions routinely render the spellings of personal names and other Old Norse-Icelandic words in the skalds’ Biographies (including the sigla for skalds and poems), as well as in the Translations and Notes, according to the norms of thirteenth-century (pre-1250) orthography. Hence the spelling of many names that appear in the Text of an edition will differ from the spelling of the names in the English Translation and Notes. In Sigv Berv 1/2II, for example, the Text reads Sighvatr hefr gram lattan but the Translation uses the form Sigvatr (literally ‘Sigvatr has lord dissuaded’), because medial ‑h- was lost in compounds when the h formed the onset of the second element (e.g. Sig-hvatr > Sig-vatr, ANG §294). Similarly, long vowels were shortened before consonant clusters (see above), which resulted in such pairs as Þórmóðr and Þormóðr, Oddi lítli and Oddi litli. Conversely, short vowels were lengthened in certain environments (see above), yielding Þjóðolfr and Þjóðólfr, Kalfr and Kálfr, Ulfr and Úlfr. For the different forms of the name Óláfr, see above. Scandinavian ethnic names are also normalised according to these principles, hence Hǫrðar, Þrœndir, but Hólmbúar (< Holmbúar). For the treatment of place names in the Translations, see §2.4 above; elsewhere in the edition, place names are given in their Old Norse-Icelandic or modern forms, or both, as appropriate.