Volume 2 (1999)             PIPA Home | Volume 1 (1998) | Submissions | IPA Home

Literary Assumptions and Linguistic Analysis in Poetics

-- Toby Griffen

One of the most striking aspects of Welsh literature is the degree to which the linguistic rules of poetics have been analyzed and maintained. We find detailed description and discussion of poetics not only in works on the literature,¹ but even in linguistic analyses of the language per se.² Indeed, the literary scholar cannot even approach the poetry without first gaining a rigorous knowledge of the phonological, grammatical, and semantic patterns inherent to this, the oldest continuous poetic tradition in Europe. Nor can the linguist approach analysis without reference to the poetry that the people themselves consider to be the finest expression of their language.

This cozy relationship between literature and linguistics is, however, by no means restricted to Welsh. In all linguistic analysis, including that of English, linguists rarely analyze what is actually spoken, in spite of their insistence that the spoken language not the literary language is the object of their study. Whether we choose to say that we are studying langue rather than parole, competence rather than performance, we rather abruptly dismiss real language utterances and concentrate more upon what the speaker intended to say that is, what the speaker would have written.

While the very existence of the study of poetics demands the interaction between literature and linguistics, we must be extremely careful that this interaction not become so close as to stifle understanding. As perhaps no other language typifies this close interaction as does Welsh, it is not surprising that we should find cases in which our understanding has actually been diminished by the traditional assumptions of poetics.

The Alliteration of Cynghanedd Poetry
Modern Welsh poetry has been dominated throughout its history by the intricate patterns of alliteration found in cynghanedd ( correspondence ) poetry. In this poetic form, consonants in one part of the line correspond in order to those in another part of the line. For example, the following line exemplifies one of the simpler of such patterns:

These patterns are, however, far from simple. Since the Golden Age of cywydd poetry in the late fourteenth to the early sixteenth centuries, the prescribed patterns have been determined by working backwards through the great poets lines from the stress accent of the last emphasized word. The placement of this accent and the patterns deriving from it are succinctly summarized by Eurys Rowlands as follows:

1. cytbwys acennog          ´            ´

2. cytbwys diacen                   ´              ´

3. anghytbwys acennog   ´             ´

4. anghytbwys diacen            ´               ´

That the overriding importance of basing the patterns upon the stress accent should date from the particular period of the cywydd is by no means surprising. By this time, Late Latin had shifted the basis of its metrical patterns from quantity (long vowels as opposed to short vowels) to stress the emphatic increase in amplitude associated with the beat. Further supporting the primacy of the beat in poetry throughout Europe at this time was the ascendance of isotonic music, in which the stress accent was required to coincide with the beat of the music.⁵ Of course, this development was certainly supported by the stress accent in the English language adopted after the various invasions by the now dominant low-land Britons (those who had spoken Latin during the Roman period).

Thus, it is hardly surprising that the tradition should arise in Welsh poetics in which the stress accent was indeed considered to be the basis of the metrical systems. Consequently, literary scholars and linguists alike have long recognized the stress accent as primary.

From a linguistic point of view, however, such a description is incomplete. There is another accent in Welsh a pitch accent rising throughout the word and culminating in the final syllable. In most cases, this syllable follows the stress accent, but it can also coincide with it when the stress accent falls on the last (or only) syllable.⁶

If we place the tradition aside and consider the possible role of the pitch-accented syllable, we find that all of the complexity suddenly disappears. Basing the patterns on the final pitch accent, we can restate the complicated rules above as follows:

Superior as this single rule may appear, it is all the more compelling due to the fact that the stress-based rules admit faults in which a consonant before the stress accent may correspond with one after the stress accent.⁸ In spite of their complexity, the old rules break down in the end. The pitch-based rule, on the other hand, not only eliminates the fault, but it reveals through the stress-based conception of a fault that the correspondences must indeed start one syllable later before the pitch accent, not before the stress accent.

Anapaestoids in Old Welsh
That the final syllable played a major role in Old Welsh is beyond doubt. While some linguists claim that the final syllable maintained pitch and stress accents and that the stress then shifted to the preceding syllable, others claim that the final syllable maintained pitch accent and that stress accent later developed on the preceding syllable. However, we choose to approach the question (and there are some compromise positions as well), we can all agree that there was a pitch accent on the final syllable.⁹

In his monumental Cerdd Dafod ( The Art of Poetry ), Sir John Morris Jones speculates on how this poetry may have been organized and then traces its development into the stress-based meters of Middle and Modern Welsh.¹⁰ Of course, he assumes what the literature has long maintained: Welsh poetry is inherently based upon stress accent. Yet, from the analysis of cynghanedd poetry above, this assumption certainly appears far less secure than we had once believed it to be.

Although it dates only from about the thirteenth century, the Black Book of Carmarthen is nonetheless generally regarded as the oldest manuscript in the Welsh language.¹¹ The original composition is, of course, considerably older than the manuscript, dating probably from the late Old Welsh to the early Middle Welsh periods between the tenth and eleventh centuries. Nearly all of the book consists of poetry, mostly in the form of religious, vaticinatory, panegyric (elegiac), and legendary poetry, including englynion. There have been several editions,¹² as well as smaller samples.¹³

In the first half of the first poem Ymddiddan Myrddin a Thaliesin ( The Conversation of Myrddin and Taliesin ), we find regular lines in length, but an irregular meter. We should recall that while the pitch accent is always on the final syllable, the stress accent is usually on the preceding syllable, but occasionally on the final syllable. As it is the final syllable that rhymes, we can mark one stanza (spoken by Myrddin) with underlines for rhymes and acute accent marks for stress, as follows:

To account for the obvious irregularity with accented syllables rhyming with unaccented syllables, Jarman posits different meters for different lines.¹⁴ Unfortunately, there is no way of predicting which line will have which meter. As a mnemonic and rhythmic device, then, the meter is no help at all, in spite of the fact that the purpose of meter is to provide such a mnemonic and rhythmic device.

If we shift our attention to the pitch accents though, we find that the word and phrase accents happen to coincide with the rhymes. Thus, we can simply remove the accent marks above. Moreover, we find that the emphasized internal accents rhyme internally, and the emphasized external accents rhyme externally.

Borrowing the traditional quantitative notation from Latin, let us represent unaccented syllables with the micron () and accented syllables with the macron (-­). Noting that the resulting pattern reflects the native Welsh pitch accent patterns, we find the following:

--    --    --     12

--     --    --     13

--     --    --     14

--     --    --     15

--     --    --     16

What we see here are regular, pitch-based anapaests. The meter is regular, anapaestic trimeter throughout. This would in fact provide the poet and reciter with a device that is both mnemonic and rhythmic a device that makes sense in the context of poetry based upon a native oral tradition.

If we examine the rest of the first half of the poem,¹⁵ we find that the anapaestic trimeter holds with great regularity. Moreover, every other poem in the collection that has internal rhymes maintains with just as great a regularity what we might term hyperanapaestic trimeter lines with three unaccented syllables before the tonic final syllable of the foot.

When we recall that the final pitch accent was the one noncontroversial and regular mark of Old Welsh, the notion that poetry surviving through the oral tradition with expected mnemonic and rhythmic devices may have relied upon anapaestoid meter comes as no surprise. After all, if the accent is realized on the final syllable, it would make sense that the meters would be based upon the final syllable as well an anapaestoid as opposed to an dactyloid pattern.

Conclusion
If we approach poetics with assumptions based upon the tradition of literary scholarship while eschewing fresh innovative analyses, then whatever faults may lie within the traditional assumptions will continue. In Welsh, this means that modern poets write their cynganeddion according to tortuous rules learned by rote, so that critics will be pleased with the result. It means that the native literary tradition must go undetected under layers of imposed requirements. It means that we will never again experience a Golden Age.

Such a dreary outlook is made all the more ludicrous by noting the source of our traditions the bardic grammarians. As Thomas Parry points out,

Thus, we base our assumptions upon historical literary assumptions known to be suspect.

Lest we simply sit back and cluck our tongues at poor misguided Welsh scholars, let us examine our assumptions in English, French, German, and other languages more widely taught in the institutions of learning in the State of Illinois. How many of us can even name the assumptions we use but do not question? How many of us have examined the language of literature we profess to study free of these assumptions? How many of us, if we did indeed see a discrepancy as large as those noted for Welsh, would dare even suggest the possibility of analyzing English afresh, of starting over with French, of striking out on a new path in German?

There is also a tradition in Celtic studies allowing traditions to be challenged. To be sure, in the more commonly taught literatures, the assumptions challenged in poetics tend to reflect theoretical linguistic positions such as structuralism, metrical phonology, etc. But these challenges are mere secondary issues compared to the assumptions we do not challenge. Have they and we become such cozy bedfellows that we do not even know that they are there?

To give a brief example in German, the Alemannic and Bavarian dialects of the south were dominant in the Middle Ages, and literature written in these dialects is considered to be Middle High German. By the modern period, however, East Central dialects became dominant, and these dialects have come to represent New High German.¹⁷ Throughout all of this, the poetic form known as the Spruch has maintained its popularity in Swabian literature from the Middle Ages to today.¹⁸ In our literary histories, however, we say that the Spruch was popular in Middle High German but did not survive in New High German¹⁹ --   a statement that totally ignores the fact that we are shifting our attention not only in time, but also between regional dialects with their own literary styles. Even linguists posit rules to explain how a form in Middle High German changed to a corresponding form in New High German, without mentioning the facts that the Middle High German form may well be extant in modern Bavarian and that the New High German form was probably derived from something in Upper Saxon never even similar to the contemporary Bavarian form from which it supposedly changed.²⁰

In German, are we ready to make an issue of this? Or do we decide that this is a rather cozy bedfellow after all one perhaps best left in undisturbed slumber?

Endnotes

1.  See, for example, J. Loth, La Métrique galloise (Paris, 1900); John Morris Jones, Cerdd Dafod (Oxford, 1925), J.J. Evans, Llawlyfr y Cynganeddion (Cardiff, 1951); Eurys I Rowlands, Poems of the Cywyddwyr: A Selection of Cywyddau c. 1345-1525 (Dublin, 1976).

2.  Perhaps formost among these is John Morris Jones, A Welsh Grammar: Historical and Comparative, Phonology and Accidence (Oxford, 1913). See also Toby D. Griffen (ed.), The Linguistics of Welsh Literature, special issue of Language Sciences, 15:2, 1993.

3.  Rowlands, Poems of the Cywyddwyr, p. xxix.

4.  Rowlands, Poems of the Cywyddwyr, p. xxviii-xxix.

5.  See Reinhard Strohm, The Rise of European Music, 1380-1500, (Cambridge, 1993).

6.  On the accent in Welsh, see especially D.M. Jomes, "The Accent in Modern Welsh," Bulletin of the Board of Celtic Studies 13 (1949), pp. 63-4; John Morris Jones, A Welsh Crammar; T. Arwyn Watkins, Iethyddiaeth: Agweddu ar Astudio Iaith, (Cardiff, 1961); T.D. Griffen, "On Phonological Stress in Welsh," Bulletin of the Board of Celtic Studies 28 (1979), pp. 106-12; Briony J. Williams, Stress in Modern Welsh, (Bloomington, 1989).

7.  A fuller justification for this rule is found in Toby D. Griffen, "A Single Accent Rule for Cynghanedd," Journal of Celtic Lingustics 5 (1997), pp. 00-00.

8.  Compare Rowlands, Poems of the Cywyddwyr, p. xlvi.

9.  See, for example, T. Arwyn Watkins, "The Accent in Old Welsh -- its Quality and Development," Bulletin of the Board of Celtic Studies 25 (1972), 1-11; also, Toby D. Griffen, "Epenthesis and the Old Welsh Accent Shift." Studia Celtica 26-7 (1991/92), 163-74.

10. Morris Jones, Cerdd Dafod, pp. 310-18.

11. Meic Stephens (ed.), The Oxford Companion to the Literature of Wales (Oxford, 1986), p. 41.

12. See especially W.F. Skene (ed.), The Four Ancient Books of Wales, (Edinburgh, 1868); J. Gwenogvryn Evans (ed.), Black Book of Carmarthen, (Evans, 1906); A.O.H. (ed.), Llyfr Du Caerfyrddin, (Cardiff, 1982).

13. Most notably these include A.O.H. Jarman (ed.), Ymddiddan Myrddin a Thaliesin (O Lyfr Du Caerfyrddin) (Cardiff, 1967); and Merion Pennar (ed. trans.), The Black Book of Carmarthen (Felinfach, 1989).

14. Jarman, Ymddiddan, pp. 5-8.

15. The second follows a different tradition, perhaps reflecting a scribal conflation of texts or a shift in attention to a more modern figure. On the argument for a conflated edition, see J. Gwenogvryn Evans, Black Book of Carmarthen, p. 161; also Jarman, Ymddiddan Myrddin a Thaliesin (pp. 3-5). The argument for a change in topic is given in Toby D. Griffen, Names from the Dawn of British Legend: Aneirin, Taliesin, Myrddin/Merlin, Arthur (Feinlach, 1994), pp. 64-5.

16. Thomas Parry (ed.), The Oxford Book of Welsh Verse, (Oxford, 1962), pp. ix-x.

17. See John T. Waterman, A History of the German Language (Seattle: University of Washington Press, 1966), p. 102.

18. See, for example, Peter Schalack, Schbruchdeidl (Stuttgart: Peter Schlack Verlag, 1977); Wilhelm König, Hond ond Kadds: Schwabäbische Gedichte, Sprüche und Aphorismen (Reutlingen: Karl Knödler Verlag, 1982); among many.

19. See, for example, Gero von Wilpert, Sacwörterbuch der Literatur, 4th ed. (Stuttgart: Alfred Kröner Verlag, 1964), pp. 667-68.

20. Compare the tables of correspondences on such books as Richard von Kienle, Historische Laut- und Formenlehre des Deutschen (Tübingen: Max Niemeyer Verlag, 1969). Such correspondences can even be found on Waterman, A History of the German Language.