Skáldskaparmál - 01.01.1997, Síða 234

232
Matthew James Driscoll
the same way as were the south slavic epic poems and other types of traditional
epic narrative studied by Parry and Lord and their disciples.21 Their metrical
construction is far too rigid to allow for this kind of improvisation.
This is not, however, to say that their texts were wholly fixed, and an
examination of the transmission of any of the rímur reveals that a good deal of
variation was in fact possible.25 If we compare the Wolfenbiittel and AM
manuscripts of Skikkjurímur (leaving for the moment the utterly worthless
Stockholm manuscript), it is plain for a start that verb tense was variable,
particularly where this involved no difference in the number of syllables, for
example with words like segja — sagði, fara - fóru, er— var, there are some twenty
examples of differences of this type between the two manuscripts. Word order
could also vary where the metre was unaffected: 1.46, 3 reads in 22 hvorki varþar
hark néglammf' but in Wþá var hvorki hark néglamm (here also with þá instead
of þar)\ III.51, 4 is in 22 böndin gjörvöllsluppu úr, but böndin sluppu gjörvöll úr
in W, to take only two examples - there are many more. There is also a great deal
of variation as regards the presence or absence of extra unaccented syllables,
particularly at the beginning of the line. These are frequently words whose
semantic content is limited, such as the example we have just seen (þá — þar)-,
other words of this type include svo, nú, því, að, þó. It seems also to matter little
which unstressed word is actually used, núorþá, semov er, ogor en. Othervariables
include the presence or absence of the definite article, especially where this has
no effect on the metre, e.g. 11.12, 3 tíðindi (22), tíðindin (W), and the form of
the negative adverb eigi, ei, or ekki. In most of these cases there are no compelling
reasons to prefer one reading to another. Neither, I would argue, is there any
reason to assume that as far as these particular words are concerned the text need
ever have been fixed.
Few if any of these words are important to the rhyme or alliteration. Words
that are should, in theory at least, be locked into the puzzle in such a way as to
I refer here to the “oral-formulaic theory”. The canonical texts are TheMakingofHomeric Verse:
The CollectedPapers ofMilmati Parry, ed. Adam Parry (Oxford, 1971; repr. New York, 1980),
and Albert Bates Lord, The Singer of Tales (Cambridge, Mass., 1960). For a review and
bibliography of scholarship in this field see John Miles Foley, Oral-Formulaic Theory and
Research: An bitroduction andAnnotated Bibliography (New York, 1985), regularly updated in
the journal Oral Tradition (1986—).
25 A similar conclusion with regard to skaldic poetry has recently been advanced by Russell Poole
in ‘Variants and Variability in the Text of Egill’s Höfuðlausn , The Politics of Editing Medieval
Texts, ed. Roberta Frank (New York, 1993), pp. 65-105. There has been a fair amount of
discussion along these lines with reference to the Middle English romances; see in particular
William A. Quinn and Audley S. Hall, Jongleur: A Modified Theory ofOral Improvisation and
Its Ejfects on the Peformance and Transmission ofMiddle English Romance (Washington, D.C.,
1982), who argue, p. 88, that ‘the “text” of a medieval romance, though not so ephemeral as
that associated with oralpoetry proper, may not be so fixed as that associated with a written (i.e.,
printed or “edited”) composition’. The first chapter of Murray McGillivray’s book Memoriza-
tion in the Transmission of the Middle English Romances (New York, 1990) provides an excellent
survey of recent work in this area.
26 All citations are given here in standard modern lcelandic orthography.