¹ This broad statement, whilst true as such, needs two qualifications: despite the precision, none of the sources conveys a fully accurate text for certain songs (such as ‘Pas de tor’ (B31)); and despite the consistency, the word setting in some individual cases (such as ‘Helas! pour quoy’ (R2)) may be highly variant from copy to copy. The scribal techniques on which textual precision and consistency in the Machaut Manuscripts depended are definitively analysed in Lawrence Earp, ‘Scribal Practice, Manuscript Production and the Transmission of Music in Late Medieval France: The Manuscripts of Guillaume de Machaut’, Ph.D. diss., Princeton University (1983).

² Scribes were accustomed to copying the words first and writing the notes in afterwards over the top. Judging the placement of syllables in very melismatic music rested on estimation and was often flawed. There are frequent errors in the manuscripts. For a sensitive consideration of this issue in relation to a difficult source, see Upton, Elizabeth Randell, ‘Aligning Words and Music: Scribal Procedures for the Placement of Text in the Chantilly Codex’, in A Late Medieval Songbook and its Context. New Perspectives on the Chantilly Codex (Bibliothèque du Château de Chantilly, Ms 564), ed. Plumley, Yolanda and Stone, Anne (Turnhout, 2009), 115–32Google Scholar.

³ Machaut conceived his creative works as an oeuvre; and the creation, copying and transmission of them in dedicated codices was a function of this. He is thought to have overseen the activity. Williams, Sarah Jane, ‘An Author's Role in Fourteenth-Century Book Production: Guillaume de Machaut's “Livre ou je met toutes mes choses”’, Romania 90 (1969), 433–54CrossRefGoogle Scholar. Earp, Lawrence, ‘Machaut's Role in the Production of Manuscripts of his Works’, Journal of the American Musicological Society, 42 (1989), 461–503CrossRefGoogle Scholar.

⁴ Dömling, Wolfgang, Die mehrstimmigen Balladen, Rondeaux und Virelais von Guillaume de Machaut (Tutzing, 1970), 25–32Google Scholar; Dömling's observations were built on and developed by Rose Lühmann, Versdeklamation bei Guillaume de Machaut, Ph.D. diss., Munich (1978). Earp, Lawrence M., ‘Declamatory Dissonance in Machaut’, Citation and Authority in Medieval and Renaissance Musical Culture: Learning from the Learned, ed. Clark, Suzannah and Leach, Elizabeth Eva [Woodbridge, 2005], 102–22Google Scholar; idem, ‘Declamation as Expression in Machaut's Music’, in A Companion to Guillaume de Machaut, ed. Deborah McGrady and Jennifer Bain (Leiden and Boston, 2012), 209–38. A declamation model similar to that proposed by Lühmann can be used as an editorial tool for reconstructing the word setting of Machaut's songs. Maw, David, ‘“Bona cadentia dictaminum”: Reconstructing Word Setting in Machaut's Songs’, Music and Letters 94 (2013), 383–432CrossRefGoogle Scholar.

⁵ Gieber, Robert L., ‘Poetic Elements of Rhythm in the Ballades, Rondeaux and Virelais of Guillaume de Machaut’, Romanic Review 73 (1982), 1–12Google Scholar.

⁶ Göllner, Marie-Louise, ‘Musical and Poetic Structure in the Refrain Forms of Machaut’, in eadem, Essays on Music and Poetry in the Late Middle Ages (Tutzing, 2003), 181–98Google Scholar, at 198. The essay was first published in 1989.

⁷ Marie-Louise Göllner, ‘Interrelationships between Text and Music in the Refrain Forms of Guillaume de Machaut’, in eadem, Essays on Music and Poetry in the Late Middle Ages, 213–27. The essay was first published in 1994.

⁸ Attention is restricted here to Machaut's refrain songs. The hypothesis that is advanced relates also to his lais and motets; but to include consideration of them would exceed the scale appropriate for an article of this sort.

⁹ See de Coussemaker, Edmond, Scriptorum de musica medii aevi nova series, 4 vols. (Paris, 1864–76), 3: 128Google Scholar; and Staehelin, Martin, ‘Beschreibungen und Beispiele musikalischer Formen in einem unbeachten Traktat des Frühen 15. Jahrhunderts’, Archiv für Musikwissenschaft 31 (1974), 237–42CrossRefGoogle Scholar. The passages are given in full with translations, commentary and discussion of their application to Machaut's songs in David Maw, ‘Words and Music in the Secular Songs of Guillaume de Machaut’, D.Phil. thesis, University of Oxford (1999), 1: 17–26.

¹⁰ Both virelai and ballade had duplex versions in which the first part of the virelai or the second part of the ballade were also repeated (see, for example, ‘Puis que ma doulour’ (V7) and ‘Dous amis’ (B6)). In songs of these sorts, the difference between the two forms (which had in any case a common origin in the late thirteenth-century ballette) is reduced. It is then the nature and recurrence of the refrain that establishes most clearly a distinction between virelai and ballade.

¹¹ Legrand's ‘Des rymes et comment se doivent faire’ is thought to have been written shortly before 1405. Earnest Langlois, ed., Recueil d'arts de seconde rhétorique (Paris, 1902), 1–10; the dating and circumstances are discussed on xv–xviii; ballade form is described at 7–9.

¹² Page, Christopher, ‘Johannes de Grocheio on Secular Music: A Corrected Text and a New Translation’, Plainsong and Medieval Music 2 (1993), 17–41CrossRefGoogle Scholar, at 29.

¹³ See the formal analyses in Maw, ‘Words and Music’, 1: 234–9.

¹⁴ The unstressed syllable of a paroxytonic rhyme was a reduced ‘e’ (schwa), described by Legrand as ‘when this vowel e is pronounced imperfectly and quietly’ (‘quant ce voyeul e se prononce imparfaitement et faintement’, Langlois, ed., Recueil d'arts de seconde rhétorique, 3). Legrand referred to paroxytonic words as ‘femenine’. In his view, which is adopted in modern accounts of this verse, such syllables are not counted in the length of the line when they occur in the rhyme; thus the lines of the refrain of V1 are all counted as hexasyllables, even though the first four of them end in a paroxytonic rhyme comprising an additional seventh syllable. There was no unanimity on this amongst poetic theorists of the time. Eustache Deschamps observed rather the differences in length between such lines: ‘the lines should not have the same number of syllables, but ought to be of 11 or 10, of 7, 8 or 9, as it might please the poet, without making them all of the same length, for the ballade made of equal lines is neither so pleasing nor of such a fine disposition’ (‘les vers ne soient pas de mesmes piez, mais doivent estre de .ix. [surely ‘.xi.’] ou de .x., de .vii. ou de .viii. ou de .ix., selon ce qu'il plaist au faiseur, sanz, les faire touz egaulx, car la balade n'en est pas si plaisant ne de si bonne facon’). Raynaud, Gaston, ed., Oeuvres complètes d'Eustache Deschamps, 11 vols. (Paris, 1891), 7: 276Google Scholar. For a full exposition of fourteenth-century theory of poetic technique and its application to Machaut's lyric verse, see Maw, ‘Words and Music’, 1: 8–23. The reduced ‘e’ in rhymes is addressed at 8–11. The terms ‘stress’ and ‘accent’ are sometimes used interchangeably and sometimes contrasted with one another in different ways. Within this article, I use ‘accent’ to refer to an element in a sequence of syllables or notes that is inherently stronger than those surrounding it. ‘Stress’ is then a formal category of metre, a position within a sequence that is expected to be stronger than those surrounding it. The stress pattern of a metre needs a corresponding pattern of accents to manifest it, but once established, may be understood to continue even in the absence of an accent or if an accent is displaced, hence the contrast between the terms.

¹⁵ Throughout the article, poetic form is represented by the pattern of its constituent lines, as defined by their rhymes (given alphabetically starting at ‘a’ and capitalised for a refrain) and the number of syllables up to and including the stressed rhyme syllable. The presence of an additional unstressed rhyme syllable is shown by an apostrophe (’).

¹⁶ Machaut preferred to call the virelai ‘chanson baladee’ (i.e. danced song), so the idea of dance was prominent in his thinking for the genre. All three formes fixes had their origins in dance-song, but their stylistic development during the fourteenth century obscured these origins. Earp, Lawrence, ‘Genre in the Fourteenth-Century French Chanson: The Virelai and the Dance Song’, Musica Disciplina, 45 (1991), 123–31Google Scholar; idem., ‘Lyrics for Reading and Lyrics for Singing in Late Medieval France: The Development of the Dance Lyric from Adam de la Halle to Guillaume de Machaut’, in The Union of Words and Music in Medieval Poetry, ed. Rebecca A. Baltzer, Thomas Cable and James I. Wimsatt (Austin, 1991), 101–29.

¹⁷ Cadence is understood in this article as a function of phrasing and without prejudice as to chord progression, a definition sufficient for the present purpose.

¹⁸ This device is common in Machaut's music, whereas the rhymes in his ballades are seldom ‘rich’ in this sense; rime riche is found more often in his rondeaux. Poirion, Daniel, Le Poète et le Prince: L’évolution du lyrisme courtois de Guillaume de Machaut à Charles d'Orléans (Geneva, 1978), 433Google Scholar.

¹⁹ Maw, David, ‘Meter and Word-Setting: Revising Machaut's Monophonic Virelais’, Current Musicology (2002), 69–102Google Scholar, at 78–9.

²⁰ I expand here upon the conception of dissonance introduced by Lawrence Earp to his discussion of Machaut's word setting (see Earp, ‘Declamatory Dissonance’). He was primarily concerned with structurally unusual declamation (exceptions to binary grouping); my proposal makes the concept central to the dislocation of the poem and its musical mimesis in Machaut's practice. Earp's more recent essay (‘Declamation as Expression’) views Machaut's practice in terms of normativity and deviance. I adhere to the concept of dissonance as it conveys the active friction between the formal implications of the verse and its setting.

²¹ Maw, ‘“Bona cadentia dictaminum”’, 396–7.

²² Strictly speaking, the caesura is a syntactic break, as after jolis in ‘Gais et jolis, lies, chantans et joieus’ (B35). In practice, a break does not invariably occur; but the caesura always marks a word boundary, and the fourth syllable is usually a linguistic accent: ‘Et puis que Dieus m'a fait si eureus’ (also from B35). Daniel Poirion has argued against the caesura as a position of fixed stress on account of the occurrence of the lyric caesura (accented third and unaccented fourth syllables): ‘de ma dame pour qui sui amoureus’ (again from B35). He proposes that the position be regarded simply as a pause. Pause gives agogic accent, so that the idea of a fixed stress is not vitiated by the idea. Poirion, Le Poète et le Prince, 441–2. Maw, ‘Words and Music’, 1: 9–10.

²³ Lühmann, Versdeklamation, 69–70. There is no need to avoid the generally understood term ‘chord’ when discussing medieval music, despite its absence from music theoretical writings of the time. The concept was not used in elementary teaching of polyphony, but compositional practice, an area of musical thought scarcely touched by theorists then, clearly used chordal thinking. A distinction between chord-notes and non-chord-notes (i.e. between the harmonic structure and its embellishment) is essential for a satisfactory account of fourteenth-century polyphony (the concepts of consonance and dissonance used by fourteenth-century theorists become problematic in this undertaking). Use of the word ‘sonority’ to denote a chord impoverishes the language. Sonority refers to the quality of sounds employed in musical performance: the same chord changes its sonority if it is sounded by different instruments or sung to different syllables.

²⁴ Machaut played-off the convention with more subtle procedures in certain rondeaux. In ‘Dame, mon cuer’ (RF 7), the cantus cadences on d at the end of both parts despite the different rhymes. The two notes are differently harmonised, however, as G and D chords respectively. ‘Se vous n'estes’ (R7) and ‘Comment puet on’ (R11) cadence on D chords at the end of both parts, but the tonal meaning of these is different (the first tends towards a C chord, whilst the second is tonical). The two chords are distinguished melodically (on a and on d) in the cantus parts.

²⁵ Langlois, ed., Recueil d'arts de seconde rhétorique, 7–8. Legrand's exposition is occasionally unclear through inconsistency or uncertainty in the use of terminology. The word vers initially refers to a single line of a poem; but in the descriptions of the virelai and ballade it clearly refers to a group of lines. Such semantic bivalency seems to have been widespread: the sources of Johannes de Grocheio's treatise use versus in both senses (Christopher Page has proposed a correction to distinguish versus (stanza) from versiculus (line), ‘Johannes de Grocheio on Secular Music’, 27–8); and for Deschamps, the word ver generates an outright confusion, as when he says that ‘les deux vers [=groups of lines] après … doivent estre de .iii. vers [=lines] ou de deux et demi’ (Raynaud, Oeuvres complètes d'Eustache Deschamps, 7: 281). The word ‘verse’ has still these two senses today. Legrand does not explain what he means by the word couples, but it may well be two or more lines that rhyme (like the modern ‘couplet’). In that sense, then, each of the two vers of B37 (shown below) contains two half couplets that are completed by the rhyming lines in the other vers. See Langlois's discussion, Recueil d'arts de seconde rhétorique, 5–6. For Deschamps, the term couple meant stanza (Raynaud, Oeuvres complètes d'Eustache Deschamps, 7: 278, 281, 287–8 and 290).

²⁶ Deschamps uses the terms without comment in his description of the virelai in a way that suggests their borrowing from musical form. Raynaud, Oeuvres complètes d'Eustache Deschamps, 7: 281–3.

²⁷ A similar idea is conveyed in Johannes de Grocheio's use of the word consonantia for rhyme (Page, ‘Johannes de Grocheio on Secular Music’, 26–9).

²⁸ The existence of harmonic tonality in music pre-1600 and the general utility of the ‘major-minor-phrygian nomenclature’ in the discussion of this music are argued in Caldwell, John, ‘Some Aspects of Tonal Language in Music of the Fifteenth and Sixteenth Century’, Proceedings of the Royal Musical Association 110 (1983–4), 1–24CrossRefGoogle Scholar. Although ostensibly focused on a later period, Caldwell's account includes mention of earlier music with special attention to Machaut (7–9, 13–15). Recognition that Machaut's music can be usefully classified by major and minor keys was argued by Gilbert Reaney and has been subsequently developed by Peter Lefferts and Yolanda Plumley. Reaney, Gilbert, ‘Modes in the Fourteenth Century, in Particular in the Music of Guillaume de Machaut’, in Organicae voces: Festschrift Joseph Smits van Waesberghe, ed. Fischer, Pieter (Amsterdam, 1963), 137–43Google Scholar. Lefferts, Peter M., ‘Signature Systems and Tonal Types in the Fourteenth-Century French Chanson’, Plainsong and Medieval Music 4 (1995), 117–47CrossRefGoogle Scholar; Plumley, Yolanda, The Grammar of Fourteenth-Century Melody: Tonal Organization and Compositional Process in the Chansons of Guillaume de Machaut and the Ars Subtilior (New York and London, 1996)Google Scholar. A useful discussion of the complexity of tonal action in Machaut's work is given in Bain, Jennifer, ‘“Messy Structure”? Multiple Tonal Centers in the Music of Machaut’, Music Theory Spectrum, 30 (2008), 195–237CrossRefGoogle Scholar.

²⁹ The fifth phrase has different cadences (open and closed) on repetition and has thus been counted twice.

³⁰ This technique is often assimilated into the concept of musical rhyme; but there is just as important a difference between a repeated figure and a repeated phrase as there is between a repeated syllable and a repeated line, so the distinction needs to be made. See Maw, ‘Words and Music’, 1: 25–6.

³¹ See Maw, ‘Words and Music’, 1: 25 for a comprehensive list of the songs in which they occur and 220–3 for a detailed discussion of the phenomenon.

³² ‘Nes qu'on porroit’ (B33) and ‘De triste cuer’ (B29) are similar, but the word setting is not identical between the two occurrences in these cases. Where the refrain is the final phrase or phrases of a part (as in B35), there is a similarity to the repeated material that ends each of the puncta of an estampie. As the puncta of an estampie are repeated with ouvert and clos cadences, it is tempting to view that genre as influential on the practice in Machaut's songs.

³³ Refrains are sometimes included within the body of the two parts. In ‘Tres douce dame’ (B24), it is the penultimate phrase of the first part that recurs, ending the oultrepasse.

³⁴ Pope, Mildred, From Latin to Modern French with Especial Consideration of Anglo-Norman (Aberdeen, 1934; rev. edn, London, 1952)Google Scholar, 103 (§223). The enduring importance of linguistic accent in French poetry from the Middle Ages to the present day is argued by Pensom, Roger, Accent, Rhythm and Meaning in French Verse (Cambridge, 2018)CrossRefGoogle Scholar. Machaut is considered at 50–4.

³⁵ Maw, David, ‘Accent and Metre in Later Old French Verse: The Case of the Polyphonic Rondel’, Medium Aevum, 75 (2006), 46–83Google Scholar.

³⁶ Maw, ‘Words and Music’, 1: 29–31, 46–50 and 240–8. Studies by Steven Guthrie suggest that poetic metre became progressively more abstract towards Machaut's time, encouraging ‘experimentation’ in lyric verse. Such experimentation appears to have been more intense in the ballades that Machaut set to music than those that remained in La Louange des dames. Guthrie, Steven, ‘Machaut and the Octosyllabe’, Studies in the Literary Imagination, 20 (1987), 55–75Google Scholar; idem., ‘Meter and Performance in Machaut and Chaucer’, in The Union of Words and Music in Medieval Poetry, ed. Rebecca A. Baltzer, Thomas Cable and James I.Wimsatt (Austin, 1991), 72–100. See also the excellent account in Graeme Boone, Patterns in Play: A Model for Text Setting in the Early French Songs of Guillaume Dufay (Lincoln and London, 1999), 45–78. I am inclined to attach more significance than Boone to the possibility of ‘regular alternation … as a kind of subtle background scheme’ (66), and in particular with regard to Machaut, writing a century earlier than Dufay; but sight must not be lost of the ‘variability and mobility’ (74) of tonic accent in the writing of individual lines against such a background.

³⁷ There can be a difficulty in deciding when an accentual pattern such as this is a variation within the ubiquitous binary metre and when it constitutes a change to that metre. In this case, the triple grouping is not present in the second stanza (‘pensee ne me laist’: -/-/-/) but is in the third (‘qui ainsi me refait’: --/--/). The fact of its adoption in the musical setting suggests that it is a metrical rather than merely rhythmic change. As such, it effects a transition between the preceding heptasyllable and the following decasyllable, both of which are binary. Like the heptasyllable (/-/-/-/), the ternary decasyllable has four stresses (-/-/--/--/) as compared with the five of the binary decasyllable (-/-/-/-/-/). I have drawn attention to a similar shift from binary to ternary metre as a mediation between heptasyllables and pentasyllables in the ballette poem ‘Vostre douz viaire cler’ (Maw, ‘Accent and Metre’, 46 and 71). The implication of that example is that poetic metre in French around the beginning of the fourteenth century could be both stress-timed and syllable-timed.

³⁸ The other is ‘Quant j'ay l'espart’ (R5). See Maw, David, ‘“Trespasser mesure”: Meter in Machaut's Polyphonic Songs’, Journal of Musicology, 21 (2004), 46–126CrossRefGoogle Scholar, at 94 and 96.

³⁹ See the refrain of ‘De desconfort’ (B8) for another example of this sort. The setting of common lexis (i.e. words such as douce, dame, amour and so on) in the genre of courtly love poetry against its intrinsic linguistic accent may have been generically apt. See Maw, ‘Accent and Metre’, 69.

⁴⁰ The exception in this possibility is the refrain, as recurrent phrases in the musical setting could never correspond exactly to the function of the refrain line(s) of the poem in any of the song forms.

⁴¹ See, for example, O'Neill, Mary, Courtly Love Songs of Medieval France: Transmission and Style in the Trouvere Repertoire (Oxford, 2006), 84–5Google Scholar, 89, 91, 199–202 for musical rhymes; 67, 80–1, 89, 92, 187–90 (II, XXI, XXV, XXVIII, XXII, XXXVI) for musical refrains (musical phrases repeated in the B-section or between the A- and B-sections of a bar-form stanza); 90 shows an instance of pre-Machaldian cross-cadencing between lines 3 and 6.

⁴² ‘Après vint Philippe de Vitry … Après vint maistre Guillaume de Machault, le grant retthorique de nouvelle fourme, qui commencha toutes tailles nouvelles, et les parfais lays d'amours.’ Langlois, Recueil d'arts de seconde rhétorique, 12.

⁴³ Elizabeth Eva Leach argues for the self-consciousness of inventing polyphonic song as an element in the composition of the first ballades in ‘Death of a Lover and the Birth of the Polyphonic Balade: Machaut's Notated Ballades 1–5’, Journal of Musicology, 19 (2002), 461–502CrossRefGoogle Scholar.