Canadian Journal for Traditional Music (1976)
text Underlay in Gagnon's Collection of French-Canadian Songs
Observers have long been impressed with the rhythmic quality of French folk song in general and its Canadian branch in particular. Thus, Werner Danckert began his account of French folk song with an extensive discussion of its rhythm and what he termed its "élan vital."1 And Marius Barbeau emphasized the "charm,. . . rhythm, and lilt" of Canadian song.2 The songs these writers seem to be referring to are marked by a strictly periodic tempo giusto rhythm and tend to be almost entirely syllabic. Gagnon's collection consists almost exclusively of such songs, which, in their incisive rhythms and plain, unornamented style, contrast with the highly melismatic, parlando rubato melodies of many so-called complaintes.3
The rhythmic appeal of folk songs from many parts of the world can be described as "vital," "charming", "lilting", etc. But one can recognize a folk song as French on the basis of rhythm alone. What, then, is the source of this distinctive rhythmic character? The thesis of the present study is that the rhythmic vitality and elan of many French-Canadian folk songs results from the special way in which the syllables of the texts are matched to the durations of the tunes, that is, from a special system of text underlay. The corpus upon which this thesis is based consists of the songs in Ernest Gagnon's collection Chansons populaires du Canada, 1865-67.4 Not surprisingly, the same system is seen to be at work in more recent collections of French-Canadian song and antecedents for it are found in the popular songs of France which flourished around 1500.
There are a number of advantages to studying the songs in Gagnon's collection. First, it comprises a large number of songs: more than a hundred distinct texts in about two hundred musical settings. Within this large collection, many genres of French-Canadian folk song which we know to have been extant in the nineteenth century are represented: narratives and lyrics, voyageur songs, songs in strophic as well as refrain forms, children's songs including rondes, enumerative songs, and so forth. Thus, all other things being equal, the great regularities which one finds in Gagnon's publication could not be attributed to mere coincidence or to a poor sample.
Secondly — and this is most important — Gagnon appears not to have been aware of the system of text underlay and metre which we are about to describe. Although songs conforming to this system can be notated in the metrical system of nineteenth-century musicians, the theory of accentuation underlying the nineteenth-century system is not precisely the same as the one we will describe. At the same time, Gagnon seems not to have forced the songs which he collected into the "normal," "regular," or "standard" measures of his day. Quite the contrary, he scrupulously attempted to adapt the system to the "irregularities" of the songs. For example, in "Mon per' n'avait file que moi (verse). Mon cocur est en age (refrain)", there seems to be either an "extra" beat or a "missing" beat. Gagnon accordingly inserts a "long" measure of 3-4 into the otherwise 2-4 tune.
In a similar case ("Derrièr' chez nous ya-t-uri étang (verse)! J'aimerai jamais ces amants volages (refrain)"), he inserts a "short" measure of 2-8 into the otherwise 2-4 tune.
Since Gagnon does not force the tunes into standard measures, one can be very certain that his notations are metrically as faithful as possible to the songs as he heard them.
Finally, Gagnon appears not to have forced the texts into the standard scansion system of French poetry. In art poetry, the e-mute sound is pronounced. In folk song, it is frequently elided as it would be in the prose of speech. Gagnon frequently renders his texts with elisions as follows:
Jacquot Hug's (not "Hugues") s'est présenté, G: 272
Encor' (not "Encore") sur la mer il m'envoi' (not "m'envoie"), G: 181
Gagnon's scrupulousness, about notating elisions is just one manifestation of his apparent attempt to render the songs as he heard them, an attempt evident throughout his collection.
General Features of French-Canadian Folk Song
Before considering the way in which syllables of text are co-ordinated with the rhythms of melody, a number of general observations on the prosody and music of the songs in Gagnon's book are in order. First, a distinction should be made between the prosody of recited French verse and that of French folk song. The scansion of recited or read French poetry has traditionally been based on the number of syllables in each line. In such metrical systems as those of Classical, German, and English poetry, there are patterns of stressed and unstressed, or long and short, syllables within a line. These patterns break the line into shorter units: half-lines and feet. In recited French verse, such patterns are not typical. Accordingly, there are no generally systematized units between a syllable and a line.5 However, we are dealing here with sung poetry, and, as we shall see, musical durations do give rise to such intervening units as the half-line and foot.
In both recited and sung French poetry, there is an important distinction to be made between masculine and feminine lines. Typically, a masculine line, one in which the last word ends in an accented syllable, rhymes with another masculine line; and a feminine line, one in which the last word ends in the pronounced e-mute sound (as in femme or retournent), rhymes with another feminine line. Thus, the functions of masculine and feminine lines are kept separate. However, this does not mean that masculine and feminine lines cannot be compared. For the purposes of comparison, the final e-mute sound of feminine lines is discounted. Thus, a feminine line of eight syllables (including final e-mute sound) is counted as "7+" syllables. It is considered to be equal in length to a masculine line of seven syllables.6 Indeed, in folk song, as opposed to recited verse, feminine lines are often "masculinized" by dropping the final e-mute sound entirely. In many cases, such "masculinized feminine" lines rhyme with truly masculine lines (see Example 1, above).
Thirdly, a distinction can be made between lines with an even number of syllables and lines with an odd number of syllables. In the folk songs being considered here, verse lines generally range from five to eight syllables in length. However, lines of six or eight syllables are by far more frequent than lines of five or seven syllables. This distinction between even and odd numbers of syllables is also found in classical French prosody, where lines of ten or twelve syllables are accounted for but not lines of nine or eleven syllables.7
In folk songs, the number of syllables in a line is not as rigidly fixed as it is in recited poetry. For example, the first line of a verse might have five syllables in the first stanza and six syllables in a subsequent stanza. (See, for example G: 295). Although the prosody of folk song is more flexible than that of recited classical verse, the flexibility is determinate: lines of functionally equal length which occur in parallel positions within a stanza are found to differ by one syllable, but rarely by more than that. Folk prosody also allows for isolated syllables to be repeated immediately in stuttering fashion with no effect on the count of the line involved. For example, in one variant of "A Saint Malo, beau port de mer," (G:24), the following lines occur in the refrain:
Nous irons sur l'eau,
Nous y prom' promener.
Thus, the prosody of folk song is not as rigid as that of classical poetry, but at the same time, there are limits to its flexibility.
Finally, an important distinction must be made between what happens in verses and what happens in refrains. These two parts of a stanza scan differently: typically, verses have five to eight syllables in a line, whereas refrains can vary from two to eleven or more syllables in a line. In both refrain and verse sections, lines appear in pairs of approximately equal size, such that (a) one line is feminine and another masculine:
Dans les prisons de Nantes, (feminine)
Lui ya-t-un prisonnier. (masculine) G:27
or (b) the two lines of a pair are joined by rhyme or assonance:
Mon pere avait un beau champ de pois;
J'en cuillas deux,j'en mangea trois. G: 114
Ah! je m'en vais entrer en danse:
C'est pour amant chercher.
Je me retourn',je me revir',
J'en n'ai pas trouvé de mongré. G: 217
(Note that the last example combines the procedure of the first two examples; pairs are formed by alternation of masculine and feminine lines, and the two pairs are joined by rhyme).
Verses are marked by a strong tendency toward isosyllabism. In refrains, this tendency is weaker, lines with different numbers of syllables being found frequently:
verse Mon pére aussi m'a marié. 8 (+)
refrain Gai lon la, je m'en vais rouler. 8
verse Un incivil il m'a donné. 8
refrain Je me roule,je me roule; 7 (+)
refrain Gal ion la, je m'en vais rouler 8
refrain En filant ma quenouille. 6 (+) (G: 214)
Thus, verses — as opposed to refrains — tend to be highly redundant structurally, symmetrical, balanced, normative, and "regular" in construction.
Turning 'to musical regularities, one fmds that the durations of the songs in Gagnon's collection are organized in a way that is periodic and 'overwhelmingly commetric rather than syncopated. Thus, for example, if the following durations appear:
the succeeding durations will maintain the established periodic sequence of quarter notes, for example, as follows:
rather than interrupting it, as Iouows:
(where x indicates a time-point implied by the preceding succession of periodic quarter-note durations, but not realized by the attack or beginning of an event).
And fmally, when periodic units (such as that represented by a quarter note) are divided into parts, the relationship between the first part and the following parts is uniformly as follows: parts a and b succeed each other, where a b or a >b. Thus, the followjng divisions of a quarter note appear:
as does the following division of a dotted quarter note:
In divisions of a dotted quarter note, the two basic parts may be subdivided further, but the relationship of" "or">" still holds:
In summary, one expects that the relationship between syllables and lines of text on the one hand and musical durations on the other hand will (a) reflect the distinctions between masculine and feminine lines, odd and even numbers of syllables, and verse and refrain; and (b) mirror the periodic and commetric style of the music, the basic dichotomy between simple and compound times, and the scheme of subdividing musical durations.
The System of Text Underlay
The system of text underlay manifests itself most simply in cases involving 'simple duple metre' and masculine lines with odd numbers of syllables. In such cases, one can clearly perceive that syllables are paired as follows: first and second, third and fourth, etc., until one reaches the last syllable which is not paired, which, in fact, stands alone. Thus, syllables are paired as follows:
The durational framework imposed on' these pairs is such that the first syllable (0) is longer than or equal in length to the second syllable (1):
The total durational value of such a pair of syllables can correspond to an eighth, quarter, or half note. Thus, the total durational value of one pair is equal to, one half of, or one quarter of, the value of another pair. In Figure 1, above, all pairs are equal in duration. In the following instance, one pair is twice as long as the other (Figure 2).
A complication arises when verse lines with odd numbers of syllables are feminine. In such cases, the last and penultimate syllables are treated as though they formed a pair in' the main body of the line rather than at its end (Figure 3).
The same possibilities for pairs of durations to be related to each other as equals (as in Figure 3, above) or in the relations half-to-whole or quarter-to-whole, exist as in the case of masculine lines.
Another possibility for feminine lines consists in the penultimate syllable being shorter than the last syllable, but the pair adding up to a duple durational unit. For example, in the following line (Figure 4), the last syllable is curtailed, being the duration of a dotted eighth rather than a full quarter note.
There is, thus, a contrast between masculine and feminine lines. In a feminine line, the grouping in pairs of syllables sometimes continues throughout the entire line. In a masculine line, the grouping in pairs is always interrupted at the end of the line.
When a line includes an even number of syllables, the first stands alone and the remaining syllables are paired. This can occur in one of several ways. The duration of the first syllable might be shorter than that of the following pair: an eighth, quarter, or half of their duration as in Figure 5.
And it also happens that a nine-syllable line is treated as though it were a feminine line of four syllables plus a masculine line of four, as in Figure 8.
Such exceptions are rare, and the handling of them only goes to "prove the rule."
Refrains, though they follow the rules of text underlay, are not so restricted as are verses. Quite often, rhymes and text underlay conspire to produce much shorter lines than appear inverses (Figure 9).
Occasionally, it seems that exceptional underlay in a verse line stems from imitating the rhythms of the refrain. The refrain of "A Bytown, c'est un' joli' place" contains a line which is broken into halves by its underlay (Figure 10):
confusing and ambiguous, these structures are still determinate: there is never any doubt that a rhythmic interpretation is possible; rather two or more interpretations compete for our attention. in verses, only one interpretation usually seems appropriate.
The relationship between texts and durations in compound metres is much the same as in simple metres. Here, however, the division of periodic units is not an exact, rational bisection (as is the case, for example, when is divided into but an intervallic approximation to bisection:
is divided intc Figure 1 5 illustrates pairing of syllables in verse lines featuring compound metre:
When two dotted quarters are combined there results a measure of six
Six eighth-notes can be interpreted as 2 groups of 3 eighth-or as 3 groups of 2 eighth-notes: Some
durations are commetric in 6-8 but syncopated in 3-4:
They, thus, imply 6-8. Conversely, some patterns of durations are commetric in 3-4 but syncopated in 6-8:
They, thus, imply 3-4. However, some patterns of durations are commetric in both 6-8 and 3-4:
These imply 6-8 or 3-4.
What one finds is that the text underlay patterns which only imply 3-4 do not appear. Ones which imply 6-8 or 3-4, or only 6-8, however, do appear. Figure 16 represents an instance where only 6-8 is implied:
Here, syllables are found in pairs of durations such that the first is as long as, or longer than, the second. This, of course, was also the case with simple metres.
The first syllable of a line with an even number of syllables can be as long as, or one-third the length of, the following pair. The next line of the, same song shows a case where a 6-8 or 3-4 interpretation is possible (Figure 17):
Which is more appropriate? In this case, the 3-4 interpretation is more consistent with the text underlay system for simple metres. For in the 3-4 interpretation, the beats of the measure coincide with grouping in pairs:
This brings us to a consideration of how the songs in Gagnon's collection might be barred. In simple metres, the groupings of syllables into pairs suggest alternations among measures of 2-16, 2-8, and 2-4. Thus, the example in Figure 2, above, might be barred as follows:
Such a notational convention would reveal the underlying rhythmic and metrical structure of the songs. And in songs with compound metre, alternations among 3-6, 6-8 and 3-4 often seem called for as in Figure 18 which is based on the examples of Figures 16 and 17, above:
Thus, though the rhythmic organization of many songs is isometric in its sub-or background-structure, the groupings of the texts suggest a heterometric surface or foreground-structure. It is this distinctive interplay between surface and background rhythmic structures which, I think, contributes much to the peculiar rhythmic quality of the songs.
Finally, the refrains of songs in compound metre must be considered. Though the underlay of refrains in simple duple metre songs was exceptional in that it created ambiguous patterns that could be interpreted in more than one way, it never transgressed completely the bounds of the "rules" for underlay which were found to be uniformly applicable to verse sections. In the refrains of compound time songs, however, the rules are in many cases thrown over completely. Sometimes, these transgressions are minor as in the following instance, where the rhythmic organizations of a verse line and the following refrain line are "run-on" (Figure 19):
And frequently, a feminine ending breaks up a refrain line as in Figure 20:
But the most common type of exception is of major importance. It involves the use of "extra" syllables in refrain lines: without such syllables, the refrain's underlay would be regular; with them, the whole complex of rules breaks down (Figure 21):
In summary, one can state that the distinctions between masculine and feminine lines, odd and even numbers of syllables, verse and refrain lines, and simple and compound times manifest themselves not only in the overall style of the songs in Gagnon's collection, but also in the matching of the text to durational values.
Text Underlay in Other French Songs
It must be emphasized that the system for co-ordinating text and music which we have described applies to every line of every song in Gagnon's collection. Since the system is so regularly adhered to, one wonders whether it has antecedents or descendants. The evidence from Continental French popular songs around 1500 suggests that the system does indeed have predecessors, and the evidence of more recent collections of French-Canadian folk song suggests that it did not die out in the nineteenth century .10
The question of text underlay in French Renaissance song is a thorny one. The problem stems from the notational practice of the time which does not "ligate" (or "join") the tones which are to be sung to a given syllable. Thus, one can only be sure of the proper text underlay in cases where the setting is syllabic.. An examination of the approximately 250 monophonic French songs which have been preserved from this period shows that about 90 per cent of the syllabically set lines adhere to the system described for Gagnon's collection. And if one attempts to apply this system to the non-syllabic lines, one finds that it works in about 80 per cent of the cases.
This correspondence between the two repertoires is not so surprising as it may seem at first. Both repertoires stem from the same linguistic tradition, both appear to have been transmitted primarily by oral means, and both display the flexible approach to prosody which was described above.1 1 How this correspondence is to be interpreted is more problematical. One could consider the system found in Gagnon's collection a survival from past centuries. Or the finding might indicate that the problems of co-ordinating a prosody based on syllable-count with commetric musical metre was solved in the same way twice. Since we lack a proper musical history of French song in oral tradition, the answer remains in doubt.
If one examines twentieth-century collections of French-Canadian song, one again finds that the majority of lines adhere to the system of text underlay described for the songs in Gagnon's nineteenth-century collection. For example, of the forty-two songs in Marius Barbeau's Jongleur Songs of Old Quebec, every line of every song adheres to the system with seven exceptions. Of these seven exceptional cases, I was unfortunately able to locate the original copies of only one. This was Mme Jean F. Bouchard's rendering of "Blanche comme la neige." 2 When the original recording and transcription were compared with the version of the melody published in Jongleur Songs, discrepancies were found only in the exceptional lines:
"Blanche comme la neige/Belle comme le jour." In the publication, exceptions to the system occur; in the original recording and transcription, the system is followed precisely. This, coupled with the fact that obvious errors in rhythmic notation occur in thirteen of the forty-two songs, leads one to wonder how accurate our modern publications are. Accordingly, I would advocate a check (based on original recordings) of many of the more recent publications of French-Canadian song. For text underlay, the vital relationship between the syllables of text and their musical rhythm is a source of much of the song's expressive power.
University of Toronto,
Toronto, Ontario
1 Werner Danckert, Das europaische Volkslied, (Berlin, Bernhard Hahnefeld, c. 1939), (repr. Bonn, H. Bouvier und Co., 1970), Pp. 228 f.
2 Marius Barbeau, Jongleur Songs of Old Quebec (Toronto: Ryerson, 1962), p. xvi.
3
4 Quebec, Bureau du "Foyer canadien," 1865-67. For a list of editions of Gagnon's collection, see Conrad Laforte, La chanson folkiorique et les ecrivains du XIXe siecle, (Montreal: Hurtobise HMH, 1973), pp. 134 f. The present study employs the seventh printing based on the re-edition of 1880.
5 For the prosodic systems of Classical, German, English, and French poetry, see Edmund Gosse, "Verse," The Encyclopaedia Britannica (London and New York: The Encyclopaedia Britannica, 13th ed., 1911, vol. 27-28), columns 104147.
6 This convention is followed, for example, in Patrice Coirault, Formation de nos chansons folkloriques (Paris: Editions du Scarabee, 1953).
7 This is discernible, for example, in the prosodic treatises in Ernest Langlois (ed.), Recueil d'arts de seconde rhetorique (Paris: Imprimerie nationale, 1902 [Collection de documents inedits sur l'histoire de France v. 182]).
8 To students of French Renaissance music, this so-called "long-short-short" pattern is familiar as the opening of innumerable chansons and instrumental pieces based on the chanson style. See Gustave Reese, Music in the Renaissance (New York: Norton, 2nd ed., 1959), p. 292.
9 In a rational scale of measurement, each quantity can be represented as a rational number and can be divided exactly into halves: 2 = 1+1; 1 = ½+½; etc. In an intervallic scale of measurement, each quantity can be represented as a whole number and can only be divided exactly into halves when that number is even. Otherwise, when it it is odd, only an approximation is possible: 3 = 2+1 or 5 = 3+2 or 2+3; 7 = 4+3 or 3+4; etc. Note that for text underlay in the songs of Gagnon's collection, the only odd numbered durational value that occurs is 3, and it is always divided into 2+1 (i.e., sequences of 1+2 (for example, where are never found).
l O Both of these repertoires are discussed in the author's "Melodic and textual types in French popular song, ca. 1500" (Ph.D. dissertation, Columbia University, in preparation), chapters 4 and 8.
11 Correspondences between the two repertoires extend to tonal structures and form as well. See ibid., chapter 8.
12 This recording was made available to me through the great kindness of Renée Landry and Roxanne Carlisle of the Canadian Centre for Folk Culture Studies, National Museum of Man, Ottawa.
© Canadian Journal for Traditional Music