The Battle of Maldon, page 11
Appendices
I
‘OLD ENGLISH PROSODY’
[In ‘The Tradition of Versification in Old English’, Tolkien alludes to the belief that ‘metre was to the ancient poets a recipe independent of the pudding’. The excerpt from ‘Old English Prosody’ (MS. Tolkien A 30/2, fols. 6–20) below extends this cookery metaphor in developing a ‘recipe’ for the craft of alliterative verse, one which no doubt served Tolkien in the composition of new works like The Homecoming. The text is numbered 1–29 and dates to the early 1930s; the first lecture in a series on Old English Prosody was delivered in the Examination Schools at Oxford on 13 October 1932.]
This analysis is intended to be primarily practical. The quarrels of metrists, which are more notorious than those of theologians, are largely due to the failure to distinguish clearly between an elaborate theory of rhythms, and an analysis of the content of lines – no two lines in the same metre being, of course, ever precisely identical from this point of view – on the one hand, and a recipe on the other, a pattern or scheme for the use of poets, a mould into which they can pour, according to skill and taste, the infinitely varied matter of actual words.
It is the recipe we want – especially in Old English where it provides an essential and enormously valuable critical tool. It is the recipe which must exist, explicit or unconscious in both the mind of poet and audience (or student) if metrical skill or performance is to be appreciated more rationally than the vague pleasure found, say, by those ignorant of musical technique in listening to music. It is the recipe which must exist if a poet in fact is to write verse that is capable of any schematic analysis at all. The fact that OE verse is readily capable of such analysis is sufficient proof, even if Norse traditions did not exist to support us, that some such recipe did once, more or less explicit and teachable exist, even though all tradition and instruction in these matters (which was in all probability never committed to writing) has perished in England without trace.
The proof of the pudding is not only in the eating but in the making or at any rate reproduction. Only a correct recipe – though it may be expressed in words and in manner quite different from those of the original cook, and even perhaps unintelligible to him – will produce again the same pudding. Only using a recipe founded on Sievers’ analysis – with modifications perhaps, but not fundamental alteration – can Old English verse be written: by which I mean can anyone who knows the OE verse language write new matter in it, which is not only a string of half-lines actually found in our records (this can be done without any metrical knowledge or theory at all!), and which does not only contain some lines of a pattern actually found, but also contains no lines which are not found.
This is sufficient to show that, however expressed or inculcated, the recipe of the poets in the classical period (say the eighth century) was the same as that expressed in our own odd way in our analysis.
We will therefore attempt our recipe. Do not let us be troubled by the cry that this is ‘artificial’. Of course it is! Verse is. Language is enormously complex and varied. It is impossible to find any two groups of words that are precisely equal from all points of view, that might be of metrical significance (which is all phonetic points of view, since ‘metre’ may take into account all points of phonetic analysis – length, stress, intonation, consonantal and vocalic structure, sequences of these, and so on; or it may be content to select from those, and leave the rest to taste). But ‘prosody’ – verse-planning – consists not in analyzing the intricacies of language, but in devising a pleasing pattern (more or less fitted to the tendencies of the language, no doubt) and fitting words to it. It is this interplay between the fluidity and variability of language, and the relative rigidity of the conscious and controlled pattern that constitutes at once the skill and the pleasure of verse-writing and verse-hearing.
Some recipes are extremely simple. This does not say that the verse written to them is necessarily simple or monotonous. It means that of the complexity of the phonetic structure of the medium the conscious and deliberate pattern has elected to regard only a few features or even one salient fact. In these cases the practice of a poet can hardly be fully understood from the bare recipe. From eggs flour butter and sugar very different puddings can be made according to various recipes (and various cooks). In such cases we are only content to say so and so is the ‘metre’, if we know for some reason or other that his simple pattern was all that the poet actually fully-consciously used, and that other important parts of his practice were due to less fully conscious processes, which we dub ‘ear’ or ‘feeling’ or instinct …
The OE recipe is nonetheless not one of the simpler ones, because its scheme took into consideration nearly all the phonetic facts of the language and bound them intricately together. Only vocalic structure (rhyme) was left ‘unconsidered’ – as part of the scheme – and handed over to the individual ‘cook’ as a seasoning to be employed according to need and taste.[1] OE recipe included length of syllable (time duration), stress (loudness) – of which at least three grades were consciously distinguished primary, subordinate, atomic – alternation in rhythmic pattern of line syllables, and alliteration on opening sounds (of loud syllables).
These were not isolated but indissolubly connected – stress and length were only considered together, and alliteration was infused in connexion with both, sometimes dictated by them, sometimes colouring the words, and determining as it were the rhythm in doubtful cases. This function of alliteration is most important, and is frequently overlooked by critics who perceive the interconnexions of our artificially isolated types. But as a matter of fact enormous numbers of OE half-verses are susceptible of various analysis if the alliteration is unknown. Where the full line is set out and the alliteration known only in rare cases is there any doubt – except purely as to nomenclature.
We shall not expect a simple recipe then. But one thing must be noted – the poet’s practice was founded solidly on the facts of natural, if formal and ceremonious, speech. In that he had an infallible guide in the classic period. It was only later when the established formulae and verses of the old poets became a convention at variance with the facts of speech that we get confusion and inconsistency …
For though deliberate employment of rhythm can be observed clearly in Beowulf, and changes in rhythm for different purposes, this is not part of the ‘recipe’, but a grace, colouring, or ‘seasoning’ specially employed by the poet …
OE verse is built entirely of blocks balanced against one another and a common rhythmic pattern was not achieved, nor aimed at. I will call its fundamental principle weight: this is length of syllable but not as measured in a machine … but as appreciated by the ear, and mind, easily influenced by concomitants high tone, loudness, actual importance in significance (for meaning), and suggested importance (the apparent importance given to words by form, portion in line, rhyme, alliteration, etc. – it is the constant attempt of OE to make actual and apparent importance coincide, but this is not of course with so tricky a thing as language always achievable). The weight of a verse element is in fact substantially time duration, but coloured by loudness (stress) and significance. OE verse is built like a wall or tower of solid chunks of more or less equivalent weight, each independent, but cemented by alliteration, piled course on course, or line on line. The basis is the weighty syllable and its normal concomitant the light syllable – this is the actual basis of the language. The simplest form of recipe for a single line is then this:
2 blocks (quite independent metrically) consisting each of 2 full elements or ‘feet’. A foot is a heavy (weighty) syllable – a long and stressed (loud, significant) syllable – and a concomitant light syllable – a short and unstressed (quiet, insignificant) syllable. The cement is provided by the rule that the first heavy syllable of each block must begin with the same consonantal sound – which we will call, following the custom borrowed from Norse, staves.
II
THE TRADITION OF VERSIFICATION IN OLD ENGLISH [CONTINUED][1]
I will deal first with certain lines in which corruption is indubitable, whether its emendation is obvious or not. First of all 1 is defective because the MS is capite mutilus.[2] 109 has a word grimme omitted at the beginning (haplographically owing to its similarity to grundene). 172 has the second half missing owing to some accident (probably scribal).
More debatable lines may now be examined.
(a) Probably or certainly corrupt
1. 75: wigan wigheardne, se wæs haten Wulfstan. This is a case of the misplacement of the head-stave. There are two other apparent cases of this somewhat serious fault: 45 and 288. But whatever may be thought of these others, it is difficult to believe that 75 is recorded as it left the author. Its defect is so simply remedied, and the corruption of so easy and frequent a kind. The author probably said: Wulfstan haten, which was still a current way of putting it, and so remained until much later. In which case we have in our text substitution of a wordier and more prosy equivalent. But in any case the received order is probably wrong (and may be due to transcription): the more natural order, even using the relatival expression, is se Wulfstan wæs haten. Compare: An preost wes on leoden; Laȝamon wes ihoten, he wes Leouenaðes sone [Layamon’s Brut].
2. 7: he let him þa of handon leofne fleogan. This is defective in giving the stave to the weaker word (let) only in the first hemistich. The older language actually made finite verbs in real speech subordinate normally to concomitant nouns; and this fact was in consequence recognized in verse-making. It was not a metrical rule. The metrical rule was that of the two lifts in a half-line the loudest and most audible must bear the stave (the other might join in). The rule was obeyed if the most emphatic word did alliterate – the part of speech might vary in different languages or times, without breach of metrical rule. It is probable, both from the alliteration of later poems and from linguistic development, that the subordination of finite verbs became less and less marked in English, as a general rule. We thus find verbs taking the stave instead of nouns in various lines of Maldon. These are dealt with later. But in this line we have a verbal form and use that was specially weak. A preterite of this kind – especially in ‘auxiliary’ use before a governed infinitive – was often a mere ‘dip’ in the older verse, indicating pronunciation on a very low tone. Letan in this use is moreover shown to have been weak by the development of forms with shortened stem vowel in ME (and other languages). We must then regard line 7 as defective. But it is at least likely that the ordinary prose word handon has been substituted for the poetical equivalent folman. A similar process is several times observed in Beowulf, e.g. 965 (handgripe MS for mundgripe shown by head-stave M.). That it has occurred here is further suggested by 108–9: Hi leton þa of folman feolhearde speru; and 150 fleogan of folman. The resulting crossed alliteration would according to the technique of Maldon suffice to correct the fact that the answer to the head-stave was still on a weak syllable. Compare he gehleop þone eoh þe ahte his hlaford 189 which is a parallel in reverse. Crossed-alliteration is a definite feature of Maldon. Clear cases of its normal form (abab) are 24, 63, 68, 98, 170, 255, 256, 320, and probably 285; of the form (abba) 159, 167, 189, 289. More fugitive cases of alliterative echo are 34, 75, 130, 151, 197, 262, 318.[3]
3. It is also difficult to believe that in 224: he wæs ægþer min mæg and min hlaford, we have precisely what was originally composed. The first hemistich is passable; the second must either be given a stress on min which is clean contrary to the natural emphasis and the opposition here of mæg and hlaford, or else neither scan nor alliterate. But it must be admitted that though (in a generally competent poem) corruption in such a line is highly likely – especially since it is so easy still to express precisely the required sense metrically and idiomatically that a native speaker of Anglo-Saxon can hardly have been metrically stumped – emendation, of the satisfactory kind that carries with it the explanation of the corruption, is not obvious.[4]
4. The absence of all alliteration in 183: Ælfnoð and Wulmær begen lagon has been generally recognized as due to corruption. The emendations possible are[5] bewegen ‘slain’ corrupted to begen under the influence of begen in the preceding line, which in itself makes begen in 183 suspect. Or mere omission of words: e.g begen (on wæle) lagon.
(b) Alliteration on ‘weak’ words
There remain some other lines in which the alliteration might be regarded as imperfect. The alliteration on a verb in preference to a noun has already been alluded to. Cases which are probably perfectly genuine are those already alluded to above: 128 Hogode to wige (contrast Yfeles hogode 133); 242; 127; 240; 189. The last as pointed out above is assisted by crossed alliteration. These may be parsed not simply as freer technique, but actually as conforming to the rules owing to the increase of verbal stress – especially as in four of these five cases when it preceded its subject. But in wénde þæs formóni man þa he on méare rád 239 the emphasis on moni naturally lifts it above the other words.[6]
The occasional stressing of a verb is actually found in the older poetry, and examples are also seen in the Chronicle poems (in strict metre): e.g. þæs þe us Secgað bec (Brunanburh); beFæste þæt rice (E. The Confessor). Probably sufficiently alliterated is 282 Sibyrhtes broðor and swiðe mænig oþer. Thus interpreted[7] it is in scansion passable according to the system of Maldon, with elongated types discussed above. But here we have probably the adjunct of rhyme. This will be discussed below with reference to line 271.
(c) Misplaced Head-stave
More serious are breaches of the rule with regard to the place of the head-stave in the second hemistich. A cardinal rule of the older practice (Norse and English) was that the head-stave must be borne by the first stressed syllable (or lift) of the second hemistich. This is not broken in good verse, because it is essential to the structure of the verse. Disregard of this rule alters its character altogether. In Maldon we find nonetheless (beside the case of 75 dismissed above)
gehyrst þu, sælida, hwæt þis folc segeð 45
raðe wearð æt hilde Offa forheawen 288
Editorially we must probably retain these, since their emendation is not obvious, and Maldon to an editor is primarily the fragment as received. But for the metrical historian they may be regarded as corrupt. They are the only clear cases of the non-alliteration of the first lift in the second hemistich: and this is a bad fault. To attribute them to the capable author of the poem is difficult – unless his poem was extemporized, and never revised; both are less likely than corruption through repetition. They seem to me fairly clear cases of damage by repetition (if not by scribe) according to the process considered above. For in both cases there is no difficulty whatever in expressing the thought metrically, even with words that are employed elsewhere in the fragment; still less with OE material generally. An Anglo-Saxon would need hardly a minute’s thought each if we asked him now to put these lines right.
In 45 the poet had flotan (72, 227), lidman (99, 164), leode (23, 50) not to mention other synonyms for sælida or folc to work with; or a trifling rearrangement hwæt segeð þis folc, which while not so good would have been a natural order to him, and sufficient to his technique. Of course on the normal editorial principles we are naturally reluctant to replace a good poetical word like sælida, because it has become an almost axiomatic assumption (certainly having some support in MSS of the older poetry) that scribes usually substituted prose-words if they substituted anything. But we have not only scribes to reckon with.[8] Here we have probably reciters contemporary with the poet who knew sælida just as well as he.
In 288 we cannot well reshuffle the order. Placing Offa at the end would rectify metre, but is quite unnatural. If there is corruption, it is due either (a) to Offa having crept in unwanted from 286, causing dislocation and possibly word-loss: a very likely event in oral repetition; or (b) to æt hilde having been substituted (e.g. after forheawen æt hilde 223) for some other words, not necessarily expressing the same idea: again a very likely event in repetition. In the former case we can compare 113–5: If Wulmær had crept in on 115, ousting he 114 and swiðe 115 the alliteration would have been destroyed. In the latter we may suppose a word for battle, or weapons, has been altered; or other suitable sense, beginning with a vowel has been altered: e.g. mid ecgum (cf. ecg 60 and mid billum 114).[9]
(d) Minor irregularities of alliteration
The poem as received has several other lines that present minor irregularities of alliteration. The following contain, probably, not so much contravention of old rules, as evidence for a slight shift in language.
Thus: 308 unwáclìce wæpna neotan shows an unstressed un- in contrast with, say, únwàclìcne (Beowulf 3138). But un (originally in pre-Germanic an unstressed form) showed a natural tendency to lose its stress, or to vary it (as still). Unstressed use is found even in the older verse: cf. unMúrnlìce madmas dæleþ (Beo. 1756) contrasted with eteð angenga Únmùrnlìce 449.
But 57 is more unusual: ùnbefóhtene nu ge þus feor hider. The alliteration only of Lift 2 is quite in order, granted that Lift 2 is not of inferior emphasis to Lift 1. The treatment of un, rhythmically separated by a weak syllable from the stressed stem, as not superior to it is less common. Contrast unforcuð 51. But this treatment is probably in accordance with actual language, and due to the alteration of the strict subordination of the second of two stressed syllables which generally obtained in the older language. As of similar kind therefore are to be regarded 298 þurstanes Sunu wið þas Secgas feaht; and 80 Ælfere and Maccus Modige twegen. Though easily improved by transposing of sunu and Maccus they are probably genuine. Similar also are 266 he wæs on Norðhymbron heardes cynnes and 242 scyldburh tobrocen abreoðe his angin. The latter is easily re-written with transposition (though the resulting type is not good) or substitution of the synonyms bordhaga, bordhreoða (neither of which however occur else in Maldon). The former is improved by the form Norðanhymbrum (in which case it becomes parallel to 80, 298). But though neither are strictly regular according to the old language, they are probably genuine: 266 as testimony to an actual pronunciation norþhymbre (cf. Northumberland); 242 to a special emphasis on tobrocen. In that case 266a is not C (a type which when genuinely present is naturally intolerant of alliteration only on the second lift) but A with a slight lift and even alliteration on he. The b in burh assists in smoothing 242.[10]
