Synonymy and rank in alliterative poetry

Sign Systems Studies 40(1/2), 2012 Synonymy and rank in alliterative poetry Jonathan Roper Department of Estonian and Comparative Folklore, University of Tartu e-mail: [email protected] Abstract: This paper addresses the high sonic demands of alliterative metres, and the consequences of these demands for sense: the semantic stretching of common words and the deployment of uncommon (archaic, ‘poetic’) words. The notion of alliterative rank is discussed as an indicator of such consequences (examples are given from English and Estonian verse) and the range of onsets found for synonyms of key notions in verse traditions is remarked upon. 0.0. The metres that develop within a particular language will typically involve phonological features key to that language. 0.1. Languages with initial syllable stress will typically develop metres that feature sound-patterning involving initial syllables, especially the onsets of such syllables. Examples include the alliterative metres found in the Finnic, Germanic and Mongolian languages. 0.2. Such metres and the verse in them may form relevant comparanda regardless of any lack of genetic connection, of differences in the extent and nature of their attestation, of differences in genres recorded in these metres, etc. Good comparanda possess both similarity and difference. 0.3. Concepts developed within the research tradition of a particular alliterative verse tradition may well prove to be of value in the research of other alliterative verse traditions. 1.0. There are stylistic features common to verse in Finnic and Germanic alliterative metres. Such an alliterative style includes parallelistic restatement and the use of synonyms. Such a slow-moving style is not simply the outcome Synonymy and rank in alliterative poetry 83 of the presumably oral delivery of much of such verse; it can be found even in the most ‘modern’ of alliterative verse traditions, i.e. those used in fourteenthcentury England. This applies more particularly to the use of synonyms than to the use of parallelism, because a stock of synonyms for key notions is needed to meet the high demands of alliteration. That a single poem may not have recourse to a large number of synonyms does not disprove the existence of such a stock of synonyms, as a system of synonymy will rather be evidenced throughout a tradition. On the other hand, such a system does require an active tradition, and thus such a style will not necessarily be found in post- (or indeed pre-) traditional alliterative verse, for which Auden’s mid-twentieth century The Age of Anxiety (1947) might stand as an example. 1.1. Phonemic demands have semantic implications. Sound repetitions, whether alliteration, rhyme or other, constrain the poet’s choice of words. For example, the constraint on word choice and arrangement in a rhymed pentameter is (if we assume an average of 4 content words per line) that the nucleus and the coda of the two rhyming syllables must match. In other words, 4 phonemes of the eight content words are constrained, an average of one phoneme per two content words. The constraint involved in composing alliterative verse (again if we assume an average of 4 content words per line) is that at least two of the onsets match per line. This makes a similar-looking constraint of (at least) one phoneme per two content words. 1.2. The above figures are somewhat rough – the rate of constraint will be higher if there are fewer than four content words in a rhyming line, and will also be higher if there are three rather than two alliterating words in an alliterative line. Nevertheless they give us a benchmark figure, which gives us the impression that the semantic implications of the phonemic constraints are similar in both cases. 1.3. Such an impression is incorrect. While the degree of constraint is similar, its distribution is not: the number of words it affects, and hence the freedom of word choice, is at least twice as high in alliterative verse. Sound plays a larger role in restricting the making of sense in English (and other) alliterative metres than it does when compared to rhymed metres. This is what gives rise to the ‘haze’ of alliterative verse (Sarv 1999) and to observations such as Sisam’s 84 Jonathan Roper (1953: 60) that “alliterative verse is not a good medium for precise expression”. It also leads poets to have recourse to synonyms. 2.0. Synonyms can be simplex or complex. For example, the notion sea could be expressed by the simplex ‘brim’ (“An old poetical word for the sea; also, ‘flood’, ‘water’”, according to the Oxford English Dictionary) or by the complex ‘hron rād’ (whale road, Old English). Likewise, the notion snake could be expressed by the simplex ‘tõuk’ (caterpillar, Estonian) or the complex ‘mõtsa nõgil’ (forest needle, South Estonian). We can also distinguish a third class of synonym: agentive nouns. Thus the notion sea might be expressed by agentive noun ‘ship-breaker’ or the notion snake could be expressed by the agentive noun ‘mätastemagaja’ (hummock-sleeper). These three categories of synonym correspond to those that Snorri defined in the Skáldskaparmál as heiti, kenning and kent heiti. 2.1. Researchers of Estonian alliterative verse have produced work on the use and nature of synonyms. Peegel (2004) has compiled a dictionary of nominal synonyms in Estonian Kalevala-metric verse in which he, presumably following Snorri, groups them under the categories of ‘lihtsõnad’ (simple words), ‘liitsõnad’ (compound words) and ‘tegijanimed’ (agentive nouns). Labi (2006) has discussed the verbal semantics with some attention to synonymy. In addition, Peegel’s interesting essays (1997) often touch on the question of synonyms and synonymy. 2.2. Research into English alliterative verse has proceeded somewhat differently. While there is the useful work of Marquardt (1938) on kennings, no work as significant as, say, the creation of a dictionary of nominal synonyms of all types for both Old and Middle English has been achieved. Perhaps the chief contribution of such researchers has been the development, and subsequently honing, of the concept of alliterative rank (Brink 1920, Borroff 1962, Cronan 1986). 3.0. Not all notions expressed in verse occur with the same frequency (whether type or token). One measure of the importance of a notion (or, more broadly, of a semantic field) might be the number of synonyms it is expressed by. Synonymy and rank in alliterative poetry 85 3.1. Here we will examine which notions are the most synonymized in the Estonian and English alliterative verse traditions, albeit with several provisos. Firstly, we shall only look in detail at synonyms for nouns, though we shall take a side-glance at simplex synonyms for verbs. Synonyms for adjectives and adverbs, areas where a poet potentially has even more room for manoeuvre, will not be discussed here. Secondly we shall restrict ourselves to simplex synonyms, which will also lend the results a certain roughness. For example, from Peegel (2004) we can see that in Estonian tradition the notion goose is expressed by only one simplex synonym, while it can also be expressed by thirty complex synomyms. Nevertheless, in most cases, as the following results show, this can characterise a tradition in a broad-brush manner. 3.2. Following Cronan’s analysis, we can say that in Old English verse, the notions with the most simplex synonyms are (in this order): Lord/King, Sea/Water, War/Battle/Fight, Boat/Ship, Man/Warrior/Retainer, Sword, Woman, Fire, Hall, Warrior/Hero, Son/Young Man, Death, and Hall. Here, the heroic character of Old English verse is revealed: journey, battle, fire, the hall, etc. 3.3. Semantic fields which are most often expressed by synonyms in Middle English alliterative verse cover, according to the categorization used in Brink (1920), similar themes to those in Old English alliterative verse (battle, weapons, hall, etc.). If there is a difference, it is that there are now also some more ‘sophisticated’ elements, as a new-found emphasis on (and development of sets of synonyms for) splendour, politesse, festivities, clothing and armour reveals. 3.4. In Estonian alliterative verse (based on a count of the simplex synonyms for each entry in Peegel 2004), we find that the notions with the most simplex synonyms are somewhat different. The synonyms are for these notions (in this order): girl (‘neiu’), bride (‘pruut’), woman (‘naine’), daughter (‘tütar’), child (‘laps’), boy (‘poiss’), orphan (‘vaeslaps’), man (‘mees’), male suitor (‘peigmees’), sister (‘õde’), daughter-in-law (‘minia’), brother (‘veli’/‘vend’), singer (‘laulik’), son (‘poeg’), snake (‘uss’), mother (‘ema’), young woman (‘noorik’), rich person (‘rikas’), estate bailiff (‘kubjas’), old man (‘vanamees’). There are many more simplex synonyms in Estonian than in Old English tradition (the top notion in Estonian tradition has 214 simplex synonyms as 86 Jonathan Roper compared with 19 for the top notion in Old English tradition), but such figures accord with the size of the surviving corpora of these two verse-traditions. From these results, we can see that in the Estonian material, family relations, romantic relations and the theme of betrothal come to the fore. 3.5. Peegel (2004: 134) notes that ‘naine’, ‘neiu’, ‘noorik’, ‘minia’ and ‘pruut’ (woman, girl, young woman, bride) are difficult to distinguish in practice, and we can also note that it may also be hard to distinguish in practice between some of the denotations for man in Old English. If we combine these overlapping and confusible categories, we then come to the conclusion that over half of the simplex synonyms in Old English are for male actors, and over half of the simplex synonyms in Estonian are for female actors. Given that these figures include all most synonymizable entities, including fire, ship, snake, etc., these are significant results. That Old English alliterative verse is a male tradition and Estonian alliterative verse a female tradition should be seen as related to the (gender of the) singers and their audiences. 3.6. Examining synonyms for verbs we find little difference between the two verse traditions. In the English material the main verbal semantic fields with a notable number of synonyms are those of moving, seeing, hearing, and speaking. These are precisely the same three areas that Labi (2006) deals with when discussing verbal synonymy: speaking, movement and perceiving. This need not be particularly surprising given that speaking, moving and perceiving are the main things that characters do in narratives. Nevertheless, this parallel as regards synonymized semantic fields for verbs makes the difference in focus of English and Estonian nominal synonyms especially striking. 4.0. Not all synonyms are equal in terms of the frequency with which they stave, i.e. appear as a stressed and alliterating element within the line. Staving differs from alliterating in that the latter simply involves the repetition of word-initial sounds in whatever form of poetry or prose, whereas the former is a strictly metrical concept referring to the repetition of word-initial sounds involving metrically-relevant stressed syllables. For example, a word in the final position of an English alliterative line or an unstressed word (such as a grammatical word) in such a line may be said to alliterate with other words, but it cannot be said to stave. Synonymy and rank in alliterative poetry 87 4.1. Alliterative rank is the measure of how frequently a particular word staves, and is usually expressed as a percentage. For example, in Sir Gawain and the Green Knight, the word ‘court’ appears 17 times. Each and every time it appears it staves in a line running on /k/ – so in this poem ‘court’ has an alliterative rank of 100%. The word ‘table’, on the other hand, appears 12 times. It staves only twice on those appearances: its alliterative rank is just 16⅔%. 4.2. Although none of those who devised or developed the notion of alliterative rank (Brink 1920, Borroff 1962, Cronan 1986) discuss word-choice in other phonemically-constrained verse forms, e.g. rhymed verse, we might talk of the rank of words in these other systems. Thus a word in a rhyming prosody will have its own rhyming rank. In the Sonnets of Shakespeare, ‘love’ (as base form of both verb or noun) occurs 82 times, and the word ‘prove’ thirteen times. ‘Love’ rhymes thirteen times, while ‘prove’ rhymes on ten of its occurences, thus the former has a low rhyming rank of 16% and the latter has a high rhyming rank of 77%. 4.2. These figures are very often consistent across the various poets in a verse tradition: if a word has a high alliterative rank in one poem, it is likely to have it in others as well. And likewise if a word has a low alliterative rank it is likely to have one in other poems too. Consider, following Cronan, the words knight and lede (and the variants of both words). In Morte Arthure, the former has a rank of 18%, and the latter of 100%, in Gawain and the Green Knight, the former has a rank of 39%, whereas the latter again has a rank of 100%, and in William of Palerne, the former word has a much higher rank of 78%, but, once again, relatively speaking this is low when compared with the latter word, which once again has a rank of 100% (as do gom and frek). In each of these cases ‘knight’ alliterates much less frequently than other synonymous words: it ranks lower. 4.3. A high alliterative rank may indicate more than the usefulness of a particular word to the poet searching for a head-rhyme. 4.4. Just as there are ‘literary’, ‘poetic’ and ‘archaic’ words to be found in rhymed verse rima causa, there are unusual words to be found in alliterative verse alliterativa causa. 88 Jonathan Roper 4.5. A high-ranking word may be undergoing semantic stretch: such a word may be used in a non-prototypical sense (in a metaphorical, metonymic, or symbolic) sense. Borroff terms these senses ‘idealising’, ‘generalising’ or ‘elevated’ senses (Borroff 1962: 81), but this is really a subset of the broader class of non-prototypical senses, as we can also imagine words can be stretched in a derogatory rather than an elevated direction. 4.6. Given that when poets are sometimes pressed by the demands of sound to grope for obscure words, and given that they will sometimes come up with original solutions but often resort to existing (i.e. now traditional) solutions, then words with a high alliterative rank may be archaic-poeticisms, i.e. words long used chiefly in verse. If cognates to these archaic-poeticisms are found in alliterative verse within the same language group, we might be able to use alliterative rank as a reconstructive tool in historical poetics. 4.7. In his discussion of rank as a diagnostic of archaic-poeticisms, Cronan is cautious, but he could have been more cautious still. After having presented an intriguing set of cognates that share an unusually high alliterative rank in both Old English and Old Norse (e.g. Old English ‘secg’ (man) and its Old Norse cognate ‘seggr’), he concludes that “a word that is descended from the common Germanic poetic tradition is likely to alliterate more frequently than one that is not” (Cronan 1986: 150). Note, that his caution here lies in not claiming the reverse, namely, that a word which alliterates more frequently is likely to be a word descended from common Germanic poetic tradition. But a tradition need not have a long line of descent just because it is now common, and a widespread feature within a tradition need not be old. The commonality may be a result of independent solutions to similar problems using similar material, or the commonality may be a result of a borrowing from one tradition to another (this is particularly possible given the contacts between speakers of Old English and Old Norse over centuries in north-east Atlantic area). In either case, independent developments or borrowings will date from later than (in some cases, very much later than) the common Germanic period. That some poeticisms date from that time is not impossible, but high alliterative rank is not a sufficient justification for such a dating. 4.8. Given that a high alliterative rank suggests a word is undergoing semantic stretch, and may also suggest a word is unusual (a poeticism or even an archaic- Synonymy and rank in alliterative poetry 89 poeticism), research into the alliterative rank of words in the Baltic-Finnic and other alliterative metres would be desirable. 5.0. Not all onsets are equal in terms of the frequency of their occurence. The alliterative poet in English will be searching for words to stave with /b/ and the alliterative poet in Estonian will be searching for words to stave with /k/ more often than either of them will be searching for words to stave with /č/. 5.1. Over time a set of synonyms may develop so as to cover all the chief onsets found within a verse tradition. 5.2. In Old English alliterative verse, the notion man had the following synonyms: ‘beorn’, ‘freca’, ‘guma’, ‘leod’, ‘rinc’, ‘secg’, ‘scealc’ and ‘wiga’. Three centuries later these same words were still in use (albeit in forms such as ‘burne’, ‘freke’, ‘gome’, ‘lede’, ‘renk’, ‘segge’, ‘schalk’, and ‘wyge’) and they had been joined by two more synonyms – ‘tolk’, the original meaning of which was not man but translator, and ‘haþel’, a nominalisation of an adjective denoting ‘noble’. In both cases we can see the sense has been stretched to a degree to meet the demands of sound. 5.3. Why words beginning with /t/ and /h/ might be particularly good candidates for bending or stretching to represent the notion man is to be understood by the frequency of occurrence of onsets in Old and Middle English. The synonyms for man (and the word ‘man’ itself) covered the eleven most commonly occurring onsets: /b, f, g, h, l, m, r, s, š, t, w/ (/h/ was also considered to alliterate with vowel-onsets). The less common onsets /č, d, j, dȝ, n, p, θ, ð, v, z/ were not assigned a traditional synonym for the notion man in this tradition. This is a highly optimal system that presumably developed over many centuries, and the stretching of the senses of ‘tolk’ and ‘haþel’ is part of the development of this system. 5.4. Finlayson’s (1963: 381) suggestion that words of high rank tend to have “wide notion spheres” can be taken further to suggest that words within certain semantic fields may be more prone to be taken up and used to stave than others. But the examples of ‘tolk’ and ‘haþel’ reminds us that the role of sound is also important here. 90 Jonathan Roper 5.5. If we compare these synonyms for man with the synonyms for the new concept of politesse, we can find that the words used in this latter semantic field display much phonemic duplication. Of the sixteen words, 4 begin with /k/, 2 with /h/, 2 with /l/, 2 with /p/, 2 with /m/ and 2 with /r/. This distribution, 16 terms but only 6 onsets, is a highly non-optimal one. It might be imagined that had the alliterative tradition continued in English verse, then, given a number of additional centuries, the choices and inventions of generations of poets would have resulted in a much more optimal distribution for this semantic field. 5.6. While there may not be the time-depth of documentation to produce such findings regarding developments in the onsets of synonyms for particular notions over time in the Estonian alliterative verse tradition, we can at least note that it is suggestive that this tradition frequently has synonyms covering the most common onsets /k, p, t, v/, while onsets such as /j/ and /n/ are poorly represented. A time-depth of attestation is important in such investigations as Middle English scribal practices show that it is precisely such no longer easily comprehensible synonyms of high alliterative rank (such as ‘freke’) that are likely to be replaced by understandable words (such as ‘man’), scribes of course being less attentive to sound than poets. 5.7. This is implicitly accepted by modern editors, who all have to deal with alliterative poems that do not survive in their original manuscript. Take line 1038 of The Siege of Jerusalem. In manuscript it runs: Fynde couþe þe no man þat on þe feet couþe. Here there are only two of the three staving words we would expect in a line of Middle English alliterative verse. Hanna and Lawton (2003: 78) amend the line to read: Fynde couþe þe no freke þat on þe feet couþe. This gives us three staving words in a line running on ‘f’ (‘fynde’, ‘freke’, ‘feet’). Implicitly, they assume that at some stage in the manuscript transmission of this poem, a scribe replaced the no longer easily understandable word ‘freke’ with an understandable word with the same meaning, ‘man’. At the date this Synonymy and rank in alliterative poetry 91 happened, the (diachronic) semantic stretch was at breaking point, and the scribe chose to sacrifice the sound to retain the sense. 5.7. Many of the words for the notion man used in Middle English verse and discussed in 5.2 are not found in Middle English prose. While for the contemporary auditor of Middle English verse they may thus have had an archaic, or so-called “poetic”, feeling to them, diachronically speaking they are effective solutions to the demands of alliterative verse, and the satisfaction they give in terms of sound trumps any semantic bending out of shape. Just as scribes can give up on old words that meet the demands of sound but fall short on sense, a new generation of poets may no longer follow the example of their elders in using particular words, though the singing and speaking poets may hold on to them for longer than the writing scribes. 5.8. We can also look at the rank of different (or shifted) senses of the same word. Drawing on Cronan (1986) we can see that ‘ecg’ (edge) when used to cover the notion of ‘sword’ alliterates 96% of the time. This contrasts with the low rank ‘ecg’ has when it is expressing its straight and unstretched meaning of edge. While the figures are dramatic, the finding is not counter-intuitive: it is precisely when there is such semantic stretch that we should suspect the word is being used alliterativa causa. 5.9. In the case of both ‘archaisms’ (5.6) and ‘figurative uses’ (5.7) the words in question are not used because they are poetic. They have become poetic by being repeatedly in verse. Arguments that words with high rank are morely likely to be colloquial than poetic (Krishna 1975) confuse words that are stylistically high (e.g. king, knight, eorl, etc.) for words that are stylistically unusual (e.g. freke, burne, etc.) or semantically stretched (e.g. ‘king’ used to denote a badger or a shopkeeper). 6.0. As we have seen, line-internal alliteration has high semantic demands, higher than those of end-rhyme. Often in alliterative verse when sounds are being matched, meanings are being stretched. Resort to synonyms is an example of this semantic stretch (there being no such thing as perfect synonymy), and alliterative rank is good indicator of which words (and which senses of individual words) have undergone such stretching. The concept of rank might be employed more thorough-goingly to further investigate the 92 Jonathan Roper practice of alliterative verse in Germanic tradition, and in other alliterative traditions, whether Balto-Finnic tradition, Mongolian or Somali (Frog, Stepanova 2011; Kara 2011; Orwin 2011), and indeed the mentalities expressed within them (Frog, Roper 2011). References Auden, Wystan Hugh 1947. The Age of Anxiety: A Baroque Eclogue. New York: Random House. Borroff, Majorie 1962. Gawain and the Green Knight: A Stylistic and Metrical Study. New Haven: Yale University Press. Brink, August 1920. Stab und Wort im Gawain. Halle: Max Niemeyer. Cronan, Dennis 1986. Alliterative Rank in Old English Poetry. Studia Neuphilologica 58: 145–158. Finlayson, John 1963. Formulaic Technique in Morte Arthure. Anglia 81: 372–393. Frog; Roper, Jonathan 2011. Verse versus the ‘Vanir’: Response to Simek’s ‘Vanir Obituary’. Retrospective Methods Newsletter 2: 29–37. Frog; Stepanova, Eila 2011. Alliteration in (Balto-) Finnic Languages. In: Roper, Jonathan (ed.), Alliteration in Culture. London: Palgrave, 195–218. Kara, Gyorgy 2011. Alliteration in Mongol Poetry. In: Roper, Jonathan (ed.), Alliteration in Culture. London: Palgrave, 156–179. Krishna, Valerie 1975. Archaic Nouns in the Alliterative Morte Arthure. Neuphilologische Mitteilungen 76: 439–445. Labi, Kanni 2006. Eesti regilaulude verbisemantika. Tartu: Tartu University Press. Marquardt, Hertha 1938. Die altenglischen Kenningar. Halle: Max Niemeyer. Orwin, Martin 2011. Alliteration in Somali Poetry. In: Roper, Jonathan (ed.), Alliteration in Culture. London: Palgrave, 219–230. Peegel, Juhan 1997. Kuld on jäänud jälgedesse. Tartu: Eesti Kirjandusmuuseum. – 2004. Nimisõna poeetilised sünonüümid eesti regivärssides. Tallinn: Eesti Keele Sihtasutus. Sarv, Mari 1999. Regilaul: Clearing the Alliterative Haze. The Electronic Journal of Folklore 10: 126–140. Sisam, Kenneth 1953. Studies in the History of Old English Literature. Oxford: Clarendon Press. Synonymy and rank in alliterative poetry 93 Синонимия и ранг в аллитеративной поэзии Настоящая статья посвящена требованиям созвучности в аллитеративных размерах и соответствующим последствиям для смысловых требований: трансформация лексики из общесловарного фонда и использование специальных (архаических, «поэтических») слов. Обсуждается понятие ранга аллитерации как индикатора таких преобразований (рассматриваются примеры из английского и эстонского стиха), рассматриваются аллитерирующие начала синонимов для ключевых в данной поэтической традиции слов. Sünonüümia ja aste alliteratiivses luules Käesolev artikkel on pühendatud allitereeruvate värsimõõtude kõrgetele kõlanõuetele ning tagajärgedele, mis neil nõuetel on tähendusele: tavaleksika semantika avardumisele ja ebatavaliste (arhailiste, ‘poeetiliste’) sõnade kasutamisele. Alliteratsiooniastme mõistet käsitletakse kui nende tagajärgede näitajat (näiteid tuuakse inglise ja eesti värsist); samuti kommenteeritakse värsitraditsioonides leiduvate võtmemõistete sünonüümide alguste skaalat.