Grammar and Poetry in Late Medieval and Early Modern Wales: The Transmission and Reception of the Welsh Bardic Grammars

By Michaela Jacques

The medieval Welsh bardic grammars were composed and transmitted during a period of intense social and political change in Wales. These documents, which contain both a highly Latinate description of the Welsh language and a treatment of the strict poetic metres, began their life as essentially vernacular artes poetriae. However, from the early fourteenth century to the end of the sixteenth, they were recopied and revised over and over by bards, bureaucrats, antiquarians, humanists, and the readers and reciters of poetry. At different times they served as practical handbooks, official regulatory documents and attempts to realign the Welsh texts with contemporary Latin and English scholarship. This book weaves a close textual analysis of the revisions made to the text into a broader consideration of the historical contexts that gave rise to each subsequent version. The resulting narrative offers insight into the development of Welsh bardic and scholarly practices over the course of two centuries.

• Language: English
• Publisher: University of Wales Press
• Release date: Mar 15, 2024
• ISBN: 9781837721016

Book preview

IllustrationIllustrationIllustration

© Michaela Jacques, 2024

All rights reserved. No part of this book may be reproduced in any material form (including photocopying or storing it in any medium by electronic means and whether or not transiently or incidentally to some other use of this publication) without the written permission of the copyright owner except in accordance with the provisions of the Copyright, Designs and Patents Act 1988. Applications for the copyright owner’s written permission to reproduce any part of this publication should be addressed to the University of Wales Press, University Registry, King Edward VII Avenue, Cardiff CF10 3NS.

www.uwp.co.uk

British Library Cataloguing-in-Publication Data

A catalogue record for this book is available from the British Library.

ISBN     978-1-83772-099-6

eISBN    978-1-83772-101-6

The right of Michaela Jacques to be identified as author of this work has been asserted by her in accordance with sections 77 and 79 of the Copyright, Designs and Patents Act 1988.

The University of Wales Press gratefully acknowledges the funding support of the Books Council of Wales in publication of this book.

The publisher has no responsibility for the persistence or accuracy of URLs for any external or third-party internet websites referred to in this book, and does not guarantee that any content on such websites is, or will remain, accurate or appropriate.

Cover image: Oxford, Jesus College, MS. 15, page 5. © Jesus College, Oxford, used under a Creative Commons Attribution (CC-BY 4.0) licence

Cover design: Olwen Fowler

For Caroline Jacques

Contents

List of Tables

Abbreviations

Acknowledgements

Introduction

Chapter 1 A Welsh ars poetriae

Chapter 2 Tools for Reading

Chapter 3 ‘Bardic’ Grammars

Chapter 4 Official Documents

Chapter 5 Bardic Humanism

Conclusion

Appendix: Translation of the Bardic Grammar in the Red Book of Hergest

Notes on the translation

Bibliography

Notes

List of Tables

Table 1.1 Terminology for grades of comparison

Table 3.1 Labels for cynghanedd in Peniarth 126

Table 3.2 Selection of references to bardic grammars in bardic marwnadau

Table 4.1 Manuscripts containing a version of the 24 mesur text

Table 4.2 Musical contents of manuscripts containing the 24 mesur cerdd dafod text

Table 5.1 Rhetorical terminology in 16th- and 17th-century glosses of the bardic grammars

Abbreviations

Acknowledgements

I am grateful to acknowledge the Centre for Medieval Studies and the School of Graduate Studies at the University of Toronto, who hosted and funded the postdoctoral fellowship that allowed me to prepare this manuscript. Additionally, much of the research for this project was completed during my doctoral dissertation, for which I received funding from the Celtic Languages and Literatures Scholarships Committee, the Harvard Sheldon Travelling Fellowship, the Medieval Academy of America and the Richard III Society.

There are many scholars whose generous assistance over the course of this project has been invaluable. I am deeply indebted especially to Catherine McKenna, Barry Lewis and Paul Russell, whose incisive and thoughtful commentary greatly improved the final product. Any remaining errors are of course my own. Thank you to Brent Miles for supporting and supervising the postdoctoral fellowship during which this book was completed. For their assistance over the course of the research for this book, I am also grateful to Jan Ziolkowski, Jerry Hunter, Peredur Lynch, Ann Parry Owen, Bethan Miles, Cameron Wachowich, Celeste Andrews, Joseph Shack and Kate Leach.

Finally, for their love, patience and support, thank you to my family, Ann, Karl, Frank, Leigh, Jilly, Lily, Jake, Lolo and Leonard; and to my long-suffering non-Celticist friends, Sophie, John, Stephanie, Al and Gemma. Special thanks to Carl, whose love and encouragement made the preparation and submission of this manuscript possible.

Introduction

This is a study of the bardic grammars, Middle Welsh treatises that fuse Latin grammatical learning with the sounds and metrical systems of Welsh poetry. From the early fourteenth century to the end of the sixteenth, these treatises were recopied over and over by bards and scholars. Each individual copy contains variations and revisions, ranging from minute changes in orthography to complete overhauls of the base text. While certain alterations were made to improve usability, these revisions often reflect the introduction of contemporary scholarship or new regulations into bardic thought and practice. The gradual evolution of these texts therefore offers us valuable insight into the social and intellectual history of the poets in late medieval and early modern Wales. It is also a case study for scholars of medieval linguistic history, an example of how Latin scholarship was adapted and re-adapted to fit the specifications of a vernacular language over the course of a long period of transmission.

For the scholar of pre-modern Wales, the bardic grammars are an example of a highly malleable text – a text that could be adapted, re-shaped and re-purposed by a variety of different users to a variety of different ends. By tracing how a single text was adapted and altered over the course of two and a half centuries, it is possible to come to a better understanding of how its producers and editors operated within a society that was in the process of rapid social and linguistic change. The bardic grammars are one lens through which to view the massive shift that took place in the late medieval and early modern period in Wales. As we move from the fourteenth century to the sixteenth, it is possible to observe the way in which a society progressed from the medieval to the early modern, as the bardic grammars themselves move from carefully penned texts in expensive anthology manuscripts, to roughshod personal copies, to official documents.

Though sometimes intended as pedagogical texts, the bardic grammars do not necessarily offer a guide to poetic training from the fourteenth century to the sixteenth; as shall be discussed, there are omissions and misrepresentations that mean that they are of limited utility as training documents. Instead, they represent a record of some of the ways in which the Welsh bards thought about their products and their social role in the late medieval and early modern period, and they serve as an indication of the types of external scholarship and contemporary thought that had begun to make their way into the minds of the bardic order. The bardic grammars alone cannot tell us how to compose poetry, but they may tell us a little bit about how to approach the extant corpus.

For the scholar of the history of linguistics, or even the history of scholarship and education more broadly, the bardic grammars offer something different: they are a window into how mainstream scholastic ideas were appropriated and refracted on a small scale, in a vernacular context. They are a study in how grammatical scholarship in Latin and English was received and adapted to meet the needs of a Celtic language, from the first attempt to describe the Welsh language using technical grammatical and poetic vocabulary, through to the repeated subsequent attempts to refine this description.

Despite their significance, these texts have barely been discussed outside the field of Welsh and Celtic studies. When the Welsh grammars are mentioned at all by non-specialists, there is a tendency to rely on secondhand information from a select few English-language articles, mostly by Ann Matonis.1 Matonis’s contributions are extremely valuable, but for some time have been one of few English-language treatments of the grammars widely available outside the field of Celtic studies.

The relative neglect of the bardic grammars is unfortunate, as they are an important witness both to the way in which mainstream grammatical learning was adapted into non-mainstream conditions, and to the shifting organisational patterns of late medieval and early modern Welsh society. However, the neglect is perhaps unsurprising given the fact that the most recent complete edition of the grammars dates to 1934, and is written in Welsh; the medieval bardic grammars have never been translated in full.2 This book only begins to answer that gap: first of all, by providing an English translation of a bardic grammar in the Appendix, and secondly, by offering a discussion of the copies of the grammars in later manuscripts that have to this point largely been ignored. For the first time, the various versions of the bardic grammars are situated in their proper contexts throughout all periods of circulation. These texts are incredibly rich, and the more one picks away at them, the more they reveal. I hope this is only a small contribution to a much longer conversation that will be taken up by other scholars, both within the field of Welsh and Celtic Studies and without.

Part I: Background

Latin and vernacular grammar

Like many other vernacular grammars and poetic manuals produced in the medieval and early modern West, the Welsh bardic grammars were based in, and emerged from, the study of Latin.3 Long before the bardic grammars were composed, Latin had ceased to be spoken as a native language in Europe, but continued to enjoy a robust second life as a ‘prestige’ language throughout the medieval West – the international language of scholarship, administration and the Church.4 This meant that Latin had to be learned as a second language by anyone who wished to seriously participate in written intellectual discourse or study of any kind, and knowledge of Latin was a necessary precursor to any more complex studies. It has been said that the study of Latin grammar ‘was the precondition for having a literate culture at all’.5 Latin grammar was the centrepiece of the medieval trivium – Grammar preceded Rhetoric and Logic, and was the entry-point into any kind of higher-level studies, including the study of canonical literature – with the analysis of Scripture as the ultimate end goal.6 Before one could approach any literate text, it was first necessary to have a firm grasp on the basics of language, beginning with letter and sounds, moving through to words, and eventually sentences. Only once all this was understood could more complex pieces of literature be analysed.

The formalised study of Latin as a second language naturally entailed the production of educational materials to support this study, in the form of Latin grammars, written guides to the formal features of the Latin language (accidence, morphology, syntax). Many of these texts were designed to teach Latin to non-native speakers. One of the most enduring antique grammars (the Institutiones Grammaticae, discussed further in Chapter 1) was written by the grammarian and teacher Priscian in the sixth century, as a pedagogical aid to pupils whose first language was Greek.7 A little later, Insular grammarians in early Britain and Ireland too had to confront the problem of teaching Latin – a language that was not only foreign, but belonged to a completely different linguistic family than the language they spoke at home. This presented a particular challenge in Britain and Ireland: as Vivien Law wrote, ‘an Italian or Spaniard who had studied no grammar would write bad Latin: an Irishman or Anglo-Saxon without grammar could write no Latin at all’.8

The centrality of the Latin language to literacy and education in the medieval period meant that it tended to be the blueprint for linguistic thought of any kind, including linguistic descriptions in and of vernacular languages that did not derive from Latin. However, vernacular languages themselves were generally not the objects of serious study in the medieval West, and linguistic descriptions of vernaculars were relatively rare. More commonly, vernacular languages could be used as the medium for instruction in Latin – a sometimes-reluctant concession to students’ greater comfort in their own mother tongue. In the Old English grammar he composed to teach Latin to speakers of English, Ælfric apologised for his own use of the vernacular, explaining ‘I know that many people will reproach me for having chosen to occupy myself with such studies, i.e. translating grammar into English. But I consider this subject to be suited to ignorant children, not greybeards.’9 In this instance, Old English was a practical tool intended to be used to teach Latin grammar to children, and not the end-object of study in itself.

In contrast, grammars of the vernacular (that is, grammars describing the features of a vernacular language) were scarce before the twelfth or thirteenth century. Counterexamples certainly exist from an earlier period (most notably the Old Irish Auraicept na nÉces, parts of which may date to as early as the seventh century10), but these were far from the norm. Rather, most vernacular grammars emerged after around 1300, alongside the increasing production of vernacular literature.11 Most of these grammars were specifically concerned with features of the vernacular poetry, which could not be adequately described using Latin terminology alone, and therefore demanded separate (and vernacular) treatment.12 The fourteenth century saw the compilation of the Welsh bardic grammars; around the same time or a little earlier, scholars working in other vernaculars produced the Old Icelandic First Grammatical Treatise (middle third of the twelfth century), the Donatz Proençals (second quarter of the thirteenth century), and the Occitan Leys d’Amors (1332).13 As both Douglas Kelly and Vivien Law have noted, it is difficult to generalise about vernacular grammatical writing in the medieval period: each individual grammar was the product of a distinct tradition and set of historical circumstances, and each had a distinct motivation behind its composition.14 Although the Welsh bardic grammars are exemplary of a wider trend of grammatical writing in the vernacular, the interest for scholars lies in their unique expression of grammatical concepts (as applied to a Celtic language and shaped by the demands of Welsh poetry), and in the particular circumstances that gave rise to their production and reproduction over the course of several centuries.

Latin and bardic education

Most of the recipients of a medieval Latin education would have begun their studies using the same text: Aelius Donatus’ Ars Minor, a short treatise on the eight parts of speech (noun, pronoun, verb, adverb, participle, conjunction, preposition, interjection), which has been called ‘the most successful textbook in the history of Western culture’.15 At a higher level, one might expect familiarity with more difficult grammatical texts, like Donatus’ Ars Maior or Priscian’s Institutiones.16 Such an educational background is reflected in the content of the bardic grammars, especially the grammatical segment, which is essentially an adaptation of material resembling Donatus’ Ars Maior or Priscian’s Institutiones Grammaticae (for a more detailed discussion of this relationship, see Chapter 1).

Because the content of the bardic grammars is so strongly rooted in elementary, pedagogical Latin texts, any discussion of the Welsh treatises should begin by considering the state of elementary education in medieval Wales; the compiler was using texts that would have been most frequently encountered by schoolboys learning Latin for the first time. Our knowledge of education in medieval Wales is limited, and the surviving evidence is sparse – but in all accounts, the church looms large. As was the case elsewhere in Europe, the monasteries, especially Cistercian monasteries, played a central role in the provision of elementary Latin education to pupils in Wales, and many pupils who learned Latin would have done so as a part of their clerical training.17 This association has led scholars to emphasise (perhaps to overemphasise) the possible clerical background of the compiler of the bardic grammars.18 But an education rooted in Latin grammar is only half the story: after all, the bardic grammars are also bardic.19

The professional bards who wrote throughout the period examined in this book are usually referred to as either the cywyddwyr (‘men of the cywydd’), denoting the metre in which they most frequently wrote, or the Beirdd yr Uchelwyr (‘Poets of the Noblemen’), to distinguish them from their predecessors, the Beirdd y Tywysogion (‘Poets of the Princes’).20 The bardic order to which the Beirdd yr Uchelwyr belonged was (at least in later periods) a guild-like professional society21 that regulated the poets who were paid to write in the ‘strict metres’ (the cywydd, the englyn and the awdl22). The bards’ bread and butter was praise poetry and eulogy, which were the most popular commissions from their noble patrons, but they also wrote poems of request, poems of thanks, religious poetry, love poetry, prophetic poetry, poems of reconciliation, debate poems, satires, healing poems, nature poems, erotic poetry and poems to the nobleman’s court, among other more minor categories.23

The metrical and ornamental requirements of Welsh poetic metres are incredibly complex, and it is unlikely they could have been learned without significant instruction. It is therefore generally thought that the bardic order must have had some kind of professional training; however, the nature of that system remains opaque. By analogy with the situation in Ireland, previous scholars have written about ‘bardic schools’ in Wales, in which presumably young poets would have been guided through an oral, Welsh-language education or apprenticeship in order to learn how to compose poetry.24 The problem with the bardic schools, of course, is that there is little evidence of their existence even in medieval Ireland, and none at all in Wales.25 The evidence becomes clearer as one moves forward in time: by the sixteenth century the bardic grammars had become practical documents for use in bardic training (this phenomenon is discussed in Chapter 3), and a bardic ‘curriculum’ is eventually laid out in the fifteenth-century document known as Statud Gruffudd ap Cynan (discussed in Chapter 4). However, the preponderance of evidence for the later period has perhaps led us to anachronistically attribute the functions of the later bardic grammars to the earlier copies. This function is far from obvious at the outset, and what we know about medieval bardic education consists almost entirely of assumption.

The education of the bards (whatever it may have been) has been taken by some scholars to be separate from the education enjoyed by members of the clergy, and by extension divorced from a basis in Latin learning. G. J. Williams argued that Welsh bards understood little Latin, apart from a few words here and there, using as evidence the faulty Latin examples contained in a revision of the bardic grammars from the fifteenth century, and dismissing the presence of Latin vocabulary in the work of Meilyr Brydydd and Cynddelw Brydydd Mawr.26 The scholar Eurys Rowlands wrote that ‘although we have plenty of evidence of monastic patronage of the bards, we have none at all that the bardic education system was connected with the monasteries in the late Middle Ages in Wales’.27 This may well be true, and bardic training probably did not demand any specific knowledge of Latin. But the boundary between poetic and Latin education may have been more porous than has been assumed.

The very existence of the bardic grammars, which combined Latin grammar with bardic poetry, mitigates against a stratified view of clerical and bardic education. There are specific examples that suggest this overlap in knowledge is not without precedent: some clerics may have had bardic training;28 likewise, some poets may have been more conversant in Latin than previously imagined.29 Certain bards must have had at least a limited exposure to Latin: the professional classes in medieval Wales were relatively fluid, and bards could be drawn from the same class, and even the same families, as lawyers or clerics – for whom Latin was requisite.30 Nicholas Orme has recently expanded our understanding of schooling in medieval Wales beyond just the monastic schools, and has written that the type of pupil who would have learned Latin grammar was essentially ‘boys with some social status and access to wealth. It is likely that they included the sons of the lay aristocracy and the prosperous elites of the towns.’31 Certainly, not every bard went through a Latin grammar school, and it was not a required part of bardic training – but there is no reason to imagine that the pupils in these schools should not have included some future bards, and no reason to insist on such a strong boundary between Latin and poetic knowledge. In spite of this, the idea that the ‘author’ of the bardic grammars must have been a cleric has persisted, in part because of the names that came to be associated with these texts.

Part II: The bardic grammars

Authorship

A great deal of scholarship about the bardic grammars has been devoted to identifying their authors, who are usually named as Einion Offeiriad (‘Einion the Priest’),32 Dafydd Ddu (‘the Black’) of Hiraddug33 or, occasionally, Cnepyn of Gwerthrynion.34 All these figures were connected to the bardic grammars at an early stage. In two of the three earliest manuscript versions of the bardic grammars (the Red Book of Hergest and Llanstephan 3), Einion Offeiriad is listed as the originator of three invented metres (§81);35 in the bardic grammar in Peniarth 20, the name Dafydd Ddu is substituted.36 These metrical forms do not appear in the poetry of the gogynfeirdd, who wrote in the period immediately preceding the emergence of the bardic grammars.37 Rather, the additional metres seem to have been introduced to round out the list of metres to twenty-four.38 Dafydd Ddu is also cited as the author of the section on metrical faults in the Gramadeg Gwysanau, a short grammatical fragment related to (but distinct from) the bardic grammars (see discussion on p. 34). In the same document, Cnepyn Gwerthrynion is connected with the definition of the Welsh metres.39 None of the early examples list any author of the tract as a whole; Einion and Dafydd are only named as ‘authors’ of the grammars beginning the early modern period, with the first such attribution made by the antiquarian Robert Vaughan.40 In later periods, Dafydd Ddu rose to the fore as the author of the bardic grammars, replacing Einion as the dominant figure in the early historiography of the codification of Welsh poetry until his resurgence in the twentieth century.41

All of these figures held some kind of bardic authority, and as a result were swirling around in the folkloric background of the bardic grammars – whether they were responsible for the authorship of the entirety, or a portion, or nothing at all of the tract itself. Any conclusions beyond this must be tentative. The question of authorship has been thoroughly plumbed in the secondary literature over the course of the past century or more, and no definitive answer has yet been supplied. Further discussion on this point would be of limited utility. For this reason, I resist naming an author in my own treatment of the early versions of the bardic grammars, and refer instead to its compiler or reviser, as appropriate.

Date

Because the original authorship of these texts is so uncertain (and the historical record not always clear), it is not possible to base our dating of the bardic grammars on when Einion Offeiriad and Dafydd Ddu lived. Instead, it has generally been repeated that the grammars were first composed in response to the Edwardian Conquest of 1282,42 the result of either a ‘conservative impulse’43 (comparable to the impulse that produced the Hendregadredd manuscript44) or perhaps an attempt to re-define the role of panegyric poetry in the wake of the loss of its traditional patrons.45

This post-Conquest dating has been made in part on the basis of the metrical examples in the bardic grammars, extracts of poetry intended to showcase a particular metre. One such extract, taken from a poem by Gwilym Du of Arfon, was dated by J. Beverley Smith to 1316–17.46 The difficulty here is that the metrical examples were the easiest parts of the text to substitute with newer or just alternative snippets of verse, and in later editions these metrical examples were frequently subject to wholesale replacement.47 Metrical examples cannot therefore offer any kind of secure dating for the text as a whole. All that can be stated with absolute certainty is that the bardic grammars were in existence by the time the earliest manuscript copy, Peniarth 20, was written, which Daniel Huws places around the year 1330.48 Any history of the bardic grammars that predates 1330 is necessarily speculative: my own speculative prehistory is offered in Chapter 1.

Contents

The bardic grammars can be divided into two segments, one dealing with the grammar of the Welsh language, and the other with poetic, or bardic, material. The grammatical segment has its basis in the tradition of Latin grammatical writing (especially the writings of Priscian and Donatus),49 and treats the basic foundations of language from the smallest individual component to the largest: the letters, the syllable, the parts of a sentence, sentences and figures of speech. In the early versions of the bardic grammar, the treatment of the parts of a sentence closely resembles the same section in Priscian’s Institutiones (it divides all parts of a sentence into two: nouns and verbs).50 The section on syllables combines Latinate material (like the definition of the syllable) with an idiosyncratic and apparently non-Latinate system of describing diphthongs and syllables. This is the only part of the grammatical segment that has direct bearing on the actual composition of poetry, as it is vital to grasp different types of diphthongs for the correct formation of rhyme and cynghanedd.51

The bardic segment deals with the native Welsh metres, poetic faults, attributes worthy of praise and, finally, the triads of poetry. The description of the native Welsh metres includes a brief definition of each metre, followed by an example stanza; next, the section on the faults offers an explanation of the various errors that a poet can make in a poem (grammatical, syntactical and metrical); following this, there is a section that lists the various descriptors that can be used to praise different classes of people (e.g. religious men and women; secular men; married women); finally, there is a section of triads, which both summarise the grammars in lists of three, and sometimes add new content. Curiously, the early versions of the bardic grammars all omit a description of cynghanedd or cymeriad,52 two requisite ornamental features of Welsh poetry written in the strict metres.53 This omission would not be corrected until the sixteenth century (discussed in Chapters 3 and 5).

The bardic grammar can be subdivided into sections as follows (the numbers refer to the section numbers in the translation at the back of this book):

§§ 1–48: Grammatical segment

§§ 49–162: Bardic segment

Each of these subsections is contained in the earliest versions of the bardic grammars, although as we will see, as they were edited and abbreviated over the course of the fourteenth, fifteenth and sixteenth centuries, some of these sections could be lost, while others (like cynghanedd) were added.

Versions

The critical concern with authorship, and a focus on the earliest manuscript copies, has meant that a simple two-stage model of revision has dominated scholary discussions of the transmission history of the bardic grammars to date, with Einion Offeiriad as primary composer, and Dafydd Ddu as later reviser. Ifor Williams, the first to propose this model, imagined a scenario in which sometime between 1322 and 1326 (or possibly, he thought, after 1355), Einion Offeiriad composed a bardic grammar for his patron Rhys ap Gruffudd. Slightly later, his younger contemporary Dafydd Ddu came along and made a revision of the grammar – inserting, for instance, a reference to his homeland of Hiraddug in an example for the compositional fault called carnymorddiwes (‘overtaken by hoof’) (§101).54 Other scholars offered similar interpretations of the relationship between Einion and Dafydd.55 This book essentially follows this division of the early versions of the grammars into two recensions, although without any strong reliance on authorship as evidence of the relationship between the texts. It is indisputable that the bardic grammar in Peniarth 20 contains additions that are absent in both the Red Book and Llanstephan 3 (the so-called ‘early’ versions of the text). However, it must be kept in mind that both the Red Book and Llanstephan 3 copies of the bardic grammars post-date Peniarth 20 by at least half a century: when we refer to the ‘original version’, therefore, we are referring to a hypothesised text that does not exist, but might well resemble the Red Book or Llanstephan 3 copies.

The next major revision of the bardic grammars did not occur until the middle of the fifteenth century, and the resultant product was not properly a bardic grammar. In 1455, the bard and scholar Gutun Owain re-wrote the section on the parts of a sentence to more closely align it with a Middle English grammatical tract known as the Accedence, which was itself a translation of Donatus’ Ars Minor.56 The result was that the number of parts of a sentence expanded from two to eight, and new vernacular grammatical terminology was inserted into the text. This version, called the Dwned, is not properly a bardic grammar (it omits the poetic material) and so is not treated in this book; however, this revision had a major effect on the transmission of the bardic grammars, as all bardic grammars subsequent to this tract incorporate it into the section on the parts of a sentence.

In the sixteenth century, the neat division into recensions broke down. The bardic grammars became fragmented, and certain sections began to travel independently. The most significant of these were the 24 mesur (‘Twenty-four metres’), which consists of a sparser version of the section on metres (usually only the title of a metre and an example verse), and the 15 bai cyffredin (‘15 common faults’), a list of fifteen metrical or grammatical errors that can be made in a poem. Unlike the 24 mesur text, this list was not fixed, and the fifteen named errors varied from one manuscript to the next. Both texts are discussed in Chapter 4.

The final and most significant revision of the bardic grammars occurred at the end of the sixteenth century. Pum Llyfr Kerddwriaeth (‘Five Books of Poetry’) was the most comprehensive of the bardic grammars to date, and contained important additions – like the introduction of a section on cynghanedd. Almost all copies of the bardic grammars from about 1570 on contain at least elements drawn from the Pum Llyfr, if not the text in its entirety. This version is the subject of Chapter 5.

That said, this is a simplification of a much more complicated transmission history, especially as we move into the later periods of copying. This complicated history is the subject of chapters 3–5.

Manuscripts

Over the course of some 300 years, the Welsh bardic grammars were heavily recopied. The text survives today in approximately eighty different manuscripts. I have listed below only those manuscripts discussed in this study (twenty-nine different copies): an almost complete list of all manuscripts can be found in GP, at pp. xiii–xvi (earlier manuscripts), and pp. xlvii–lviii (later manuscripts). All information regarding the dates of the manuscripts and the identities of their copyists comes from Daniel Huws’s Repertory of Welsh Manuscripts and Scribes. Many of the manuscripts are composites; as much as possible, I have listed the date and copyist of the relevant section on bardic grammar, and not the manuscript as a whole. I also list the roman numeral attached to the text in the GP edition where one exists, and give references to other editions wherever relevant. Note that this list is in alphabetical order, and does not reflect the chronology of the manuscripts or their order of appearance in this book.

Aberystwyth – National Library of Wales

NLW MS 3029B

c.1615x1630, copied by the antiquarian Robert Vaughan. Two complete bardic grammars (pp. 1–30; 31–70), the Dwned (pp. 71–111) and related texts, including a copy of William Salesbury’s treatise on figures and tropes (pp. 131–61). Huws, Repertory I, p. 185. Discussed in Chapter 5.

NLW MS 17116B

(Gwysaney 28) c.1560, copied by scribe X128, apparently an associate of Gruffudd Hiraethog.57 Mostly musical material; also contains the 24 mesur text (ff. 55–8v). Not in GP. Huws, Repertory I, pp. 293–4. Discussed in Chapter 4.

NLW MS Llanstephan 3

s.xv1. Complete bardic grammar, ff. 33–56v. II in GP. Edited in GP, pp. 19–37. Huws, Repertory I, p. 61. Discussed in chapters 1 and 2.

NLW MS Llanstephan 28

c.1455–6, copied by the bard and scholar Gutun Owain. Contains two relevant texts: a metrical fragment with innovations by Dafydd ab Edmwnd (pp. 10–16) and the text on the eight parts of a sentence, known as the Dwned (pp. 33–69). Both are important revisions of the earlier grammars. XXII in GP. Grammatical segment (the Dwned) edited in GP, 67–88; section on the metres edited in a separate article by G. J. Williams.58 Huws, Repertory I, p. 66. Discussed in Chapter 3.

NLW MS Llanstephan 45

s.xvi2. Complete bardic grammar, pp. 1–56. XXVIII in GP. Huws, Repertory I, p. 70. Discussed in Chapter 3.

NLW MS Llanstephan 55

c.1579, copied by Siôn Dafydd Rhys. Complete bardic grammar, pp. 167– 98. XX and LXXIII in GP. Section on the poetic faults and the metres edited in GP, pp. 59–65. Huws, Repertory I, p. 72. Discussed in chapters 3 and 4.

NLW MS Llanstephan 169

s.xvi2, copied by X16. Contains a copy of the 24 mesur, pp. 76–7. Huws, Repertory I, p. 95. XXXII in GP. Discussed in Chapter 4.

NLW MS Peniarth 20

c.1330, copied by scribe X88. The major fourteenth-century revision of the bardic grammars. III in GP, pp. 305–49. Edited in GP, pp. 39–58. Huws, Repertory I, pp. 342–3. Discussed in chapters 1 and 2.

NLW MS Peniarth 51

c.1460–80, copied by the bard Gwilym Tew. Re-attaches a shortened version of the Dwned to the section on letters and syllables and diphthongs, pp. 63–113. Edited by Anne Jones in a PhD dissertation.59 XXV in GP. Huws, Repertory I, pp. 359–60. Discussed in Chapter 3.

NLW MS Peniarth 56

s.xvi1; c.1543; s.xvi/xvii. Composite manuscript. Contains three sections on bardic grammar, written in three different hands. Main hand (s.xvi1) copied the section on the diphthongs and the parts of a sentence, corresponding to Peniarth 20 (pp. 1–24) and a section on the englynion (pp. 42–7). Second hand (also s.xvi1) copied a section on the parts of a sentence (pp. 31–41), discussed in Chapter 3. Third hand, Richard Longford (1543), copied the section on the metres (pp. 103–28). XVI and XXX in GP. Huws, Repertory I, p. 362. Discussed in Chapter 4.

NLW MS Peniarth 62

1582–c.1600, copied by the