Performance Practice Review

Volume 1
Article 5
Number 1 Spring/Fall

The "Not-so-precisely Measured" Music of the

Middle Ages
Hendrik van der Werf

Follow this and additional works at: http://scholarship.claremont.edu/ppr

Part of the Musicology Commons, Music Performance Commons, and the Music Practice
Commons

van der Werf, Hendrik (1988) "The "Not-so-precisely Measured" Music of the Middle Ages," Performance Practice Review: Vol. 1: No.
1, Article 5. DOI: 10.5642/perfpr.198801.01.5
Available at: http://scholarship.claremont.edu/ppr/vol1/iss1/5

This Article is brought to you for free and open access by the Journals at Claremont at Scholarship @ Claremont. It has been accepted for inclusion in
Performance Practice Review by an authorized administrator of Scholarship @ Claremont. For more information, please contact
[email protected].
Expanse of the Field: Medieval to Contemporary

The "Not-so-precisely Measured"

Music of the Middle Ages
Hendrik van der Werf

In the area of performance practice few issues have been debated as

fervently and as dogmatically as has rhythm and meter in medieval song.
For more than a century researchers have tried to reconstruct the
original manner in which these songs were performed and, by the middle
of the century, almost every conceivable theory had been broached. In a
book published in 1962 Burkhard Kippenberg subjected the existing
theories to a thorough scrutiny and came to the conclusion that no
logical and decisive evidence had been brought forth for any of them.
In addition to the objections voiced by Kippenberg one should mention
the monolithic approach as a serious weakness of most research in this
area: it has been (and sometimes still is) taken for granted that one type
of rhythm governed either all songs in a given language or all songs of
the entire medieval period.

Several thousand non-liturgical songs have come down to us. For the
majority only the texts have been preserved, but we have both text and
melody for well over 2,000 of them, the precise figure depending upon
one's own definitions of the terms "non-liturgical," "song," and "Middle

1. Burkhard Kippenberg, Der Rtythmus im Minnesang; eine Kritik der literar-

und musikhistorischen Forschung mil einer Obersic/u iiber die musikalischen Quellen
(Munich, 1962).

42
 "Not-so-precisely Measured" Music 43

Ages." It may be regretted that so many of the songs with music have a
French text, but this accident of history does have a fortunate aspect in
that one genre, often called "trouvere song,"2 or simply "chanson," has
been preserved in such abundance that we can learn much about
performance practice from the texts and melodies themselves, from the
manner in which they were preserved, and especially from the manner in
which a given song varies from one manuscript to another. Although we
all would prefer a more direct approach, we have little choice but to take
the best known genre as a vantage point from which to study the lesser
known repertories. With great caution we can try and determine in what
respects the trouvere songs differ from, and in what respects they
resemble the other ones. For now, we may restrict our attention to
monophonic song in French and Occitan. We must bypass plainchant
altogether, and leave other "not-so-precisely-measured" genres for a
future occasion.

The Texts

Our study of melodic rhythm must begin with the texts. One of the
major principles for versification of troubadour and trouvere songs (and
of most subsequent poetry in Romance languages) is the "syllable count."
A given line, or verse, normally has a fixed number of syllables in each
strophe of a given poem, but there is no fixed position for accented
syllables in places other than the rhyme and, in a limited number of
poems, the caesura. An occasional line may seem trochaic or dactylic,

2. For the sake of this study the term "trouvere" has both the advantage and
disadvantage of being vague. Some scholars, especially experts on medieval poetry, have
used the term sparingly and usually in the meaning to be proposed here. Others,
especially authors of textbooks on the history of music, have used it in reference to all
French poet-composers of the twelfth and thirteenth centuries. According to their
contents the manuscripts allow a conveniently narrow interpretation of the term. The
medieval songs in Old-French have been preserved in a dozen large manuscripts (and in
many small and fragmentary collections) dating from the middle of the thirteenth through
the beginning of the fourteenth century. They contain a rather wide variety of poems and
melodies, but one fairly homogeneous group prevails. The poems in this group are
strophic in form and content, and most of them deal with fin' amor, nowadays often called
"courtly love." Only the first strophe of the poem is provided with a melody. I shall
restrict the term "trouvere song" to members of this group.
3. The term "troubadour song" will be used for songs in the Provencal, or
Occitan, language.
4. Christopher Page, Voices and Instruments of the Middle Ages: Instrumental
Practice and Songs in France 1100-1300, (London, 1987) comes to largely the same
classification, and considers troubadour and trouvere songs to be in a "high style," the
others in "lower styles." With more or less the same results, we can also take attribution
as a criterion for categorization because most of the songs that are attributed to a specific
author belong to the first group; conversely, many of the songs outside of it are
anonymous. Fearing that uninitiated readers may see a value judgment in the terms
"high" and "low," I prefer not to use them.
44 Hendrik van der Werf

but there is no case in which an entire poem or entire strophe has a

regular alternation of accented and unaccented syllables. I may recall
here that there is such an alternation in many Latin poems of the same
time, including many (not all) Latin motet texts and even some French
motets. Traditionally, the syllable count is referred to as the "meter" of
the poem. In discussions of melodic meter this could lead to confusion,
and one might be tempted to reason that, if there is meter in the poem,
there must have been meter in the music. This apodictum is flawed
because the term "meter" is used in two meanings. A fixed number of
syllables in a poem does not necessarily mean that the melody had a
fixed alternation of accented and unaccented or long and short units.
Except when the context calls for a general term, and hi discussions of
"modal rhythm," I will avoid the terms "rhythm" and "meter" in
preference for less ambiguous ones, such as "syllable count,"
"accentuation," and "duration."

Medieval Writings Concerning Measurement

Medieval writings about the poetry of the troubadours and trouveres are
devoid of information concerning melodic rhythm. Among the treatises
dealing with music, only the one by Johannes de Grocheio contains a few
remarks that may pertain to trouvere songs. In order to evaluate his
remarks, we must keep in mind the development of the discipline called
musica. St. Augustine is one of several authors who define musica as
the "art of measuring" or the "art of measuring well." Some of those who
were interested in measuring, including St. Augustine, seem to have
found great delight in studying the numbers in those Latin poems that
have long and short syllables in a ratio of 2:1. In the fifth chapter (the
fifth "book" as medieval people called it) Augustine discusses numerical
equalities that we can perceive with our intellect, but not with our senses.
In the last chapter the real purpose of Augustine's study emerges when
he turns to the harmony found in God. Other learned authors were
fascinated with the numbers or ratios found among pitches in the scale.
Some transferred these numbers to the universe and organized the
heavenly bodies as pitches in an octave. In an almost inscrutable
development, the meaning of the term musica widened and came to
include everything that we now call music. For some time this
development went hand in hand with a confusing non-sequitur: since

5. For two recent and rather different interpretations of musica in reference to

duration see van der Werf, The Emergence of Gregorian Chant: A Comparative Study of
Roman, Ambrosian, and Gregorian Chant, vol. 1,1 (published by the author, Rochester,
N.Y., 1983), 22-30, and John Stevens, Words and Music in the Middle Ages: Song,
Narrative, Dance and Drama, 1050-1350 (Cambridge, 1986), 413-434.
 "Not-so-precisely Measured" Music 45

musica concerns itself with precisely measurable phenomena, everything

discussed under the heading musica was assumed to be precisely
measurable. In the second half of the thirteenth century Franco of
Cologne and Johannes de Garlandia discuss in detail measurements in
motets, while explicitly stating that plainchant was immeasurable.
Johannes de Grocheio disagrees with their latter statement. In his
opinion calling "music" immeasurable was contrary to the tradition of
musica. Circumventing the problem in typical medieval fashion, he
concedes that plainchant was "not so precisely measured.11 Although he
is not explicit concerning trouvere songs, Grocheio seems to place them
among the "not-so-precisely-measured" genres. Even if I am wrong on
the last point, we must reckon with the possibility that not all music of
the thirteenth century was precisely measured. Above all, we must be
very cautious in taking at face value medieval statements about the
measurability of any genre of music.

The Musical Notation

Most musical scribes of troubadour and trouvere songs used the square
notation which we know from Gregorian chant and which bears no
indications of duration aside from double notes. One of the distinctive
features of this notation was its use of simple and compound neumes. In
the latter one notational symbol comprises several pitches sung to one
syllable. Sometime in the thirteenth century the neumes acquired
mensural meaning. Not only the presence or absence of a stem, but also
the shape of the compound neume, also called a ligature, determined the
duration of its individual pitches. We do not know to what extent the
scribes of troubadour and trouvere sources were familiar with this
innovation, but it is obvious that non-mensural notation was the norm for
the chansons. In a manuscript that almost exclusively contains works by
Adam de la Hale mensural notation was used for his rondeaux and
motets, all of them polyphonic, while non-mensural notation was used
for the chansons and jeux-partis, all of them monophonic. The scribe of
the manuscript now known as the Chansonnier Cangi demonstrated his
familiarity with mensural notation by using it for the major part of a
motet for two voices, while his notation of most chansons is decidedly
non-mensural. Almost as if to confuse us, he gives the impression of
having used some half-hearted form of mensural notation in a limited

6. For more details on the semi-mensural notation in the Chansonnier Cangi

see van der Werf, The Chansons of the Troubadours and Trouvires a Study of the Melodies
and Their Relation to the Poems (Utrecht, 1972), 36-37 and 139-146. See also the
photographic reproduction, transcription, and discussion in Jean B. Beck, Le Chansonnier
Congi, 2 vols. (Paris, 1927).
46 Hendrik van der Werf

number of chansons by clearly distinguishing between stemmed and

unstemmed single notes without giving mensural meaning to his
ligatures. Furthermore, the alternation of stemmed and unstemmed
single notes varies from completely regular in some melodies to
absolutely meaningless in others. This leaves us with the problem of
distinguishing between real and make-believe mensuration in this
particular manuscript. All in all, the scribal habit of giving chansons in
non-mensural notation strengthens the idea that duration in chansons
was "not so precisely measured."

Multiple Versions

By a stroke of luck the troubadour and trouvere repertories, especially

the latter, present us with a source of information that we have just
begun to explore. In many instances a given song has been preserved in
more than one manuscript; and some songs occur in up to a dozen
sources. The multiple versions are rarely identical. For a long time it
was assumed that the extant readings were copied, directly or indirectly,
from the author's autograph and that copyists were to be blamed for the
many discrepancies. Early in this century another explanation emerged,
as literary scholars recognized that initially the songs were disseminated
by word of mouth, and that only in the mid thirteenth century was
dissemination through writing juxtaposed against a continuing oral
tradition. Most importantly, the realization arose that, for both
performers and scribes, requirements for faithful transmission were
much looser than they are in our print-dominated society. In other
words, both the persons involved in the oral tradition and those who
preserved songs in written form felt free to vary certain aspects of the
texts. Consequently, editors of the poems gave up their attempts at
reconstructing the original version of a poem. It is now general practice
to select one version as basis for an edition and to list variants from other
sources in the critical apparatus.

In retrospect, it is difficult to believe that in the early 1960s I was the first
to use differences and similarities among multiple versions as a source of
information about musical characteristics, especially the rhythm of
troubadour and trouvere melodies. Melodic variants are more numerous
(and probably more significant) than textual ones. Moreover, oral
transmission can be proven much more convincingly for the music than
for the poetry, thanks to the many songs in which the music for the first
and the second verse are repeated for the third and the fourth verse (i.e.
AB AB X). Almost invariably these first four lines differ less in multiple
versions than do the subsequent lines. If there had been a written
 "Not-so-precisely Measured" Music 47

transmission from poet-composer to extant manuscripts, we could have

explained this phenomenon only by assuming that musical scribes turned
sleepy and sloppy at the beginning of the fifth verse, but returned to
fairly accurate copying when they started the next tune. In an oral
tradition, however, a person learning a song from listening to it would
hear the A and B melodies twice as often as the X section, and thus
retain the former better than the latter. Accepting oral transmission as
the normal process does not imply that scribes entered the songs directly
into the extant sources either from memory or upon hearing. On the
contrary, there are ample indications that they (or someone else) made
what we might call "a rough copy" that was used as the model, or
exemplar, in the production of a collection in book form. Another
feature provides valuable information concerning the scribes. Four of
the trouvere sources preserve long groups of songs, often by a single
author, in the very same order and with few differences. Clearly, these
four scribes had access to the same exemplars, and it is encouraging to
learn that they could copy very precisely. Obviously, they made some
errors, but it is of more importance to note that they also must have
made some deliberate changes.

When evaluating similarities and differences among multiple versions, we

must choose between two assumptions. Putting things in black and
white, we can assume that the transmitters were connoisseurs who left
the essential features of a song intact, or that singers and scribes knew
virtually nothing about the subtleties of troubadour and trouvere art, and
corrupted both text and music. The texts of the chansons help us solve
this riddle in that they suggest that troubadour and trouvere poetry is
unlikely to have appealed to the masses. Their dissemination is not likely
to have been accomplished by footloose and unsophisticated entertainers
who eked out a living as jugglers and storytellers in city streets and town
squares. Instead, the poems are rather esoteric and are likely to have
been appreciated only by afficionados. It may be argued that these
connoisseurs, many of them troubadours or trouveres themselves, were
responsible for the transmission. At the beginning of the oral tradition
stood the poet-composer himself. Regardless of whether he sang in
order to teach his creation to someone else or in order to present it to
his peers, the author, as a song's first performer, established the manner
of presenting it to an audience. If the transmitters were experts and
connoisseurs, they are not likely to have altered any characteristics that
were essential to either the genre or the individual song, although they
may have varied other features within the boundaries of the poetic and
musical customs of the time. In this process the author may also have
been the first to vary his song from one presentation to another. If this
48 Hendrik van der Werf

was the case, the similarities and differences among multiple versions
afford us a valuable source of information about the manner in which the
songs were actually performed.

The "Approximate11 Equality of Individual Notes

The differences between multiple versions make it impossible to

reconstruct the original melody in all its details, while the similarities
assure us that what the scribes left us must have been closely related to
what the poet-composer sang at the "world premiere" of a given chanson.
The multiple versions of many songs agree fairly well on the pitches for a
given passage, but differ on how the pitches are to be distributed over the
text.7 Taken all together, such differences inescapably lead to a strong
but negative conclusion: they could not have come about, if all chansons
always had been performed in modal rhythm or in any regular
alternation of long-short or accented-unaccented units. Fortunately, the
study of similarities and differences also allows a positive conclusion,
albeit a vague one: the variants could have come about only if essentially
all pitches were of more or less equal importance to the flow and the
character of the melody. This gives new meaning to Grocheio's remark
that pitches in trouvere songs were "not so precisely measured," i.e. that
they were of more or less equal duration, with emphasis on, and great
uncertainty about the degree of "more or less." In a performance in
which pitches do not have a fixed duration and do not come in a fixed
sequence of stress and unstress, any text could be sung as the performer
desired. The poet-composer as first performer more or less determined
the "not so precisely measured" duration of a pitch, a syllable, and a
word.

Although this conclusion is vague, it does solve some problems, e.g., why
variants in the choice or order of words can bring about variants in the
placement of textual accents. In this type of free rhythm, strophes with
different distributions of accents can be performed to the same melody,
without injustice to either the text or the music. Even better, in this free
rhythm both text and melody can receive proper attention and neither is
subservient to the other. This gives meaning, too, to the "double note,"
i.e., the immediate reiteration of a pitch over a syllable. It seems

7. Since my conclusions are based upon the overall situation, I would rather not
refer to specific examples and instead suggest the study of many songs in multiple
versions. For this purpose see van der Werf, The Extant Troubadour Melodies:
Transcriptions and Essays for Performers and Scholars, Gerald A. Bond, text editor
(published by the author, Rochester, N.Y., 1984) and Trouvires-Melodien, Monuments
Monodica Medii Aevi, vote. XI and XII, Bruno Sta'blein, general editor (Kassel, 1977-
1979).
 "Not-so-precisely Measured" Music 49

reasonable that a double note represents a pitch that is longer than the
one respresented by a single note. And the duration of a given pitch
very likely varied from one performer to another and from one strophe
to another. Thus, it should be not surprising that two scribes, notating
the same melody, might disagree on whether a given pitch was
represented by a double or single note. At the present stage of the
research, it is risky to draw conclusions from the fact that double notes
occur exclusively as part of a compound neume, i.e., over syllables sung
to two or more pitches.

Fluctuation in the duration of pitches must have been so pervasive that

even as staunch a believer in musica as Grocheio was unable to measure
them as precisely as he would have liked. On the other hand, something
must have enabled scribes to distinguish between pitches of average and
longer than average duration, and to consider most of them average.
Probably, the duration of pitches was sufficiently close to equal to create
a prevailing, although "not precisely measured" unit of time.

Application to a Specific Chanson

Clearly, the above conclusions pertain to troubadour and trouvere songs,

in general. In respect to a specific melody, the search for the original
rhythm has a disappointing result, for we are unable to reconstruct
precisely how it was performed seven or eight centuries ago. However,
two aspects of our general conclusions add up to valuable guidance for
today's singer of early musk. I am optimistic enough to think that we
can come reasonably close to an authentic rendition of troubadour and
trouvere songs by applying the general conclusion to a specific chanson.
Playing the schoolmaster, I may suggest that performers begin by
studying the text (not reciting it without the music). In the next step, sing
the entire song (not one strophe at a time) making all single notes fairly
equal to one another, and making double notes more or less twice as
long as single ones. By concentrating on getting the text across to an
audience, one is likely to develop small differences in the duration of
individual pitches and, especially, almost continuous but subtly executed
fluctuations in tempo. A brief elaboration on the two caveats in the
above suggestion may be helpful. Beginning the learning process of a
song with reciting the text without music may result in a rendition in
which the syllables are of equal duration. For basically syllabic songs,
this may not be a serious shortcoming but, for more ornate ones, this
type of rhythm may fail to do justice to the melody. If one were to learn

8. In existing "precisely-measured" transcriptions such lengthening often

conflicts with the meter selected by the editor.
50 Hendrik van der Werf

a song a strophe at a time, the first strophe might get fixed so strongly in
one's mind that its rhythm gets transferred to subsequent strophes.
Needless to say, such uniformity will fail to do justice to the differences
of meaning and textual flow in individual strophes.

Perhaps, these theories can be elucidated a bit more with the help of a
personal note or two. I am not a trained singer and have virtually no
experience in reciting poetry to an audience. I do not give recitals but
often sing some chansons to illustrate a lecture. I am convinced that one
can do justice to both poem and melody. By trial and error, one can
learn how to sing several pitches over a seemingly insignificant syllable,
and to sing one pitch and one "not so precisely measured" unit of time for
the most important syllable of the sentence. In the process one is likely
to develop a great appreciation for what at first may appear a duality in
medieval songs: one comes to enjoy singing pitches that have primarily
melodic meaning at the same time one is "reciting" a poem. Text and
melody may not be wedded as they are in songs by Schubert or Faur6, or
in an aria of Mozart, yet they go far beyond the point of coexisting or of
merely tolerating one another.

The first time I published the above conclusions I used the terms
"declaim" and "declamatory" in an unwise attempt to capture a manifold
theory in a single word. Paving more attention to the single term than
to the entire theory, some fellow medievalists rejected my conclusions as
valid only for syllabic passages. Andrew Hughes wrote that my "theory
fails to help with melismatic passages." And as the following excerpt
(with a quotation from my book) indicates, my choice of terms seems to
have misled John Stevens as well:

M y . . . task will be to examine some of the deeply held, if not always

deeply questioned, beliefs which are current about words-and-music
in medieval song. The assumption that it is the words which
substantially determine the rhythm of a song is shared not only by
convinced adherents of the 'modal' theory but by its strongest
opponents, such as Appel, Monterosso and van der Werf. The
practical results they come to are very different; but at root they do
have certain beliefs in common Van der Werf is more ready to
subordinate the melodies to the exigencies of the text: 'the rhythm

9. Most notably in van der Werf, "Deklamatorischer Rhythmus in den

Chansons der Trouveres," in Die Musikforscfumg 20 (1967): 122-44.
10. Andrew Hughes, Medieval Music: the Sixth Liberal Art (Toronto, 1974), 160.
 "Not-so-precisely Measured" Musk 51

in which one might declaim the poem without the music' will shape
the musical interpretation.

I apologize to those who were led astray by an isolated term but derive
solace from those who understood the total theory. More importantly, I
hope that this and other recent discussions of rhythm in chansons rectify
whatever wrong impressions I may have given on earlier occasions.

Precisely Measured Melodies

Although they do not seem to be very numerous, there are exceptions to

the above generalizations. In fact, it has been known for a long time that
there are some trouvere songs in which the pitches are precisely
measured. Unfortunately, most studies of them, even the most recent
ones, are flawed by preconceived notions. Too often, it is taken for
granted that, until the middle of the thirteenth century, modal rhythm
was the only form of precise measurement for both monophonic and
polyphonic music. In addition, the occurrence of modal rhythm in some
chansons is still taken as evidence that all of them are in modal rhythm.
In some other cases the researcher seems preoccupied with the "saving"
of modal rhythm for the troubadour and trouvere repertories. What
we need is an objective study of all songs that show signs of durational
measurement. For the sake of the present discussion we may divide such
songs into three groups.

Several of the manuscripts saving primarily chansons contain a separate

section with motets. This does not prove much more than that

11. Stevens, Words and Music, 493-494. I cannot understand how Stevens, p. 502,
footnote 28, concludes that I speak "of the concept of 'approximate equality' as applying
only to syllabic (i.e. single note) progression."
12. See also van der Werf, The Extant Troubadour Melodies, 75-83, and my
contribution to the forthcoming Handbook of the Troubadours, Ron Akehurst and Judith
M. Davis, eds.
13. See my review of Chanter m'estuel: Songs of the Trouveres, Samuel N.
Rosenberg and Hans Tischler, eds. (Bloomington, 1981) in Journal of the American
Musicological Society 35 (1982): 539-54. See also Tischler's reaction to my review in the
same, 36 (1983): 34H4, and my response in 37 (1984): 206-208. In a review of two of my
books in Journal of the Plainsong and Medieval Music Society 8 (1985): 59, David Hiley
called my evaluation of Tischler's theories "one of the best available descriptions (and
refutations) of the idea that modal rhythm provides a key to the interpretation of
troubadour songs."
14. This seems to be the case with some of Theodore Karp's publications, e.g.
Three Trouvere Chansons in Mensural Notation" in Gordon Athol Anderson: in
Memoriam (Henryville, Pa., 1984), 474-94.
15. See especially mss. Paris, B.N. f.fr. 12615 (known as "Chansonnier de
Noailles") and Paris B.N. f. fr. 844, published as Le manuscrit du mi, 2 vols., Jean B. Beck,
ed. (Philadelphia, 1938).
52 Hendrik van der Werf

collectors who were primarily interested in chansons did not necessarily

shun motets, and that modal rhythm was not anathema to the
connoisseurs of the "not so precisely measured" chansons. Of more
interest is the "split personality" behaviour of those pieces that appear as
monophonic songs in chansonniers and as motets for two voices in other
collections. In the latter, the relation between tenor and upper voice
suggests that they were conceived and normally performed in modal
rhythm. It does not seem strange that, at least occasionally, certain
motets were performed without tenor, but we can only guess at what
happened to their modal rhythm in a monophonic rendition. The type of
double occurrence discussed here is not the only form in which an extant
chanson is related to an extant motet, and a thorough study of all such
cases will teach us quite a bit, not only about the compositions involved
but also about the differences and similarities among chansons and
motets in general.

Some of the more interesting and more elusive exceptions to be

mentioned occur exclusively in the Chansonnier Cang6 and are
anonymous. As discussed above, the scribe of this manuscript
occasionally made use of what some have considered an early form of
mensural notation, but which I prefer to call a semi-mensural notation,
because only the single notes, not the ligatures, appear to express
mensuration. As I have shown before, in most instances the rhythm
suggested by the semi-mensural notation is flatly contradicted by
differences and similarities among multiple versions of the chansons
concerned. But in a few cases good reasons exist for accepting that the
chansons combine the modal rhythm of motets with features more
typical of chansons,19 one being that no multiple versions can be found
that might argue against these songs having been conceived in modal
rhythm. But the strongest reason is that these songs resemble motets in
the way the modal rhythm fits the melody. Before we can determine the
place of such hybrid songs in the poetry and music of the thirteenth
century, we need to know more about this particular manuscript.

16. See van der Wcrf, The Chansons, example 12,134-38.

17. Hie same type of notation is used in the motet collection Paris, B.N. nouv.
acq. fr. 13521, published in photographic reproduction by Fricdrich Gennrich, Ein
altjranzosischer Mottetenkodex...La Clayeae (Darmstadt, 1958) and by Luther Dittmer,
Paris 13521 & 11411 (Brooklyn, 1958) and in transcription by Gordon A Anderson, Motets
of the Manuscript La Clayeae (Rome, 1975).
18. The Chansons, 36-40.
19. The Chansons, 144-47. Some reservation must be made about my following
the standard procedure of maintaining one rhythmic mode throughout a composition. In
the case of semi-mensural notation it is impossible to determine how strictly the composer
adhered to a given rhythmic mode. As I hope to show in the near future, there are several
indications that mixing modes was not uncommon.
 "Not-so-predsely Measured" Music 53

Someone involved in its compilation seems to have been interested in

making it look like a motet collection, and in making its chansons look
tike motets. The semi-mensural notation is the most striking
consequence of this attempt. The order in which the chansons are
entered may be another one. This is the only trouvere chansonnier to
present the songs in alphabetical order (exclusively according to the first
letter of the text), while at least two motet collectors similarly organized
their manuscript.

Finally, some chansons contain internal features atypical of their genre

compelling us to consider whether they are more precisely measured
than is normal. In a chanson by Blondel de Nesle the uniformity in its
multiple versions, including the placement of word accents and
distribution of pitches over the text, is unusual for the genre. A close
examination of these features made me "conclude that if this chanson
was meant to be performed in one of the rhythmic modes known to us, it
probably was performed" in the third mode. None of the rather diverse
commentary upon my transcription has brought us any further to a
solution of the problems posed by such atypical chansons. I am still
not convinced that it was conceived and normally performed in one of
the rhythmic modes known to us, but still consider it likely that its
pitches were measured more precisely than was usual for chansons.

Recent Performances and Recordings

Until recently, it was customary to perform all troubadour and trouvere

songs in strict modal rhythm, in which syllables were the primary
durational units in a ratio of either 1:2 or \-2\iP For syllabic passages,
this resulted in an uninspiringly regular alternation of long and short
syllables. Judging by recent recordings, performers of early music have
completely abandoned this aspect of modal rhythm. For neumatic and
especially for very ornate passages, "modal" performance yielded rhythms

20. This is the case for ms. Bamberg, Staatl. BibL, Lit 115, published in
photographic reproduction and transcription by Pierre Aubry, Cent motets du xiiie sticle, 3
vols. (Paris, 1980), and published in transcription by Gordon A. Anderson, Compositions
of the Bamberg Manuscript (Stutgart, 1977). It also is the case for several sections of the
ms. known as W2, published in photographic reproduction by Luther Dittmer,
Wolfenbuttet 1099 (1206) (Brooklyn, 1959).
21. See van der Werf, The Chansons, 42 and 100-103. See also my remarks
concerning chanson R620 by Blondel de Nesle in Trouveres-Melodien 1,559 and 26-32.
22. Stevens in Words and Music, 448, Charlotte Roederer in Schirmer History of
Music (New York, 1982), 72-73, and Theodore Karp, Three Trouvere Chansons in
Mensural Notation," in Gordon Athol Anderson: in Memoriam (Henryville, Pa., 1984), 491-
94.
23. See footnote 13.
54 Hendrik van der Werf

which, depending upon one's point of view, were either "fascinating" or

"weird"; they often were contrary to the style of motets. In current
practice many singers make the duration of individual pitches quite
unequal. In ornate passages this interest in unequal lengths is combined
with an apparent desire to make the duration of the syllables close to
equal. There seems to be no published defense of this practice, so that
one cannot help but wonder whether it arose under the influence of the
earlier practice wherein the syllable was the controlling factor, and
"fascinating" or "weird" rhythms thereby became associated with the
troubadours and trouveres. One may also wonder whether the interest
in making individual pitches unequal represents an attempt to make
secular songs radically different from Gregorian chant. In two respects
this desire is without ground. Firstly, our desire for marked differences
between religious and secular music is a relatively modern phenomenon
which came about slowly well after the Middle Ages, and seems to have
gained its greatest impetus during the nineteenth century. In addition,
the often quoted pronouncement of Johannes de Grocheio is not the
only indication that, throughout the Middle Ages, duration in plainchant
was not precisely measured and may not have resembled the equalistic
performance advocated by Andr6 Mocquereau and his fellow monks of
Solesmes.

Dancing Songs

It was for practical, not for ideological reasons, that my publications on

non-Uturgical music of the Middle Ages have almost exclusively
concerned chansons of the troubadours and trouveres. Going beyond
those repertories is difficult, almost risky, because of the low number of
extant melodies for the other songs. The troubadour sources contain
many poems that clearly fall outside of the genre discussed thus far.
Alas, only a very few of them are preserved with a melody, and most of
these seem to stem from the late thirteenth or early fourteenth centuries.
For songs with French texts fate has been more considerate, but still not
generous enough to provide us with multiple versions for many of them.
Thus, I may be forgiven for making primarily cautionary remarks in the
next few paragraphs.

24. Stevens's book, Words and Music, advocating "isosyllabic" performance of

many genres of medieval song was published long after singers went in this direction. For
an evaluation of Stevens's theories see my review of his book in the forthcoming issue of
Journal of Musicological Research.
25. Concerning accentuation and duration in plainchant see van der Werf, The
Emergence of Gregorian Chant: a Comparative Study ofAmbrosian, Roman and Gregorian
Chant (published by the author, Rochester, N.Y., 1983), vol. 1,1,22-42
 "Not-so-precisely Measured" Music 55

At first glance, it would appear that pitches in dancing songs were not
only precisely measured but also contained regular alternations of stress
and unstress. Unfortunately, we do not know enough about medieval
dances to either corroborate or contradict this notion. Probably because
modal rhythm is associated with ternary meter, medieval dance songs are
traditionally transcribed and performed in some kind of "waltz" rhythm,
even though we do not know enough about either the dances or the tunes
to exclude binary meter from consideration. To make things worse, we
even have difficulty identifying dancing songs. Almost the only
undeniable cases occur in narratives in which we are told that (certain)
people danced to a song, and here the text may be given but the music
usually is lacking. Furthermore, such a dancing song is given various
labels, such as some form of the words "rondeau" or "carolle," or
something like "chanson de carolle," or simply "chanson." As generic
terms were used in the Middle Ages, we should be wary of taking for
granted that every text said to be a "rondelet" or "carolle" is a dancing
song.

About a dozen poems have been preserved, without music, under the
heading "estampie," but there is no indication whatsoever that they are
dancing songs, or that they were performed to a clearly measured tune.
We also have a number of tunes without text called "estampie" that may
be dance tunes, but they stem from the fourteenth century. As far as I
know, we have only one case in which both text and music have been
preserved for a song called "estampie." Moreover, the melody occurs
with both a French and an Occitan text; the former is anonymous, the
latter is attributed to Raimbaut de Vacqueiras. Nevertheless, we have
no clear indication about duration and accentuation in the melody, per
se. A confusing factor in the study of the estampie is that it clearly is
related to the Latin sequence, the German Leich, the French descort and
lyric lai, none of which seem to have had anything to do with dancing.
Clearly, we need an extensive study not only of the estampies, but also of
the various members of the large sequence family. This research must
be without preconceived notions; e.g., ft must not start with the premise
that the Latin sequence is the ancestor of this group.

There is a theory that the rondeau, the virelai, and the ballade either are,
or derive from dancing songs. This usually goes together with the

26. See also Christopher Page, Voices and Instruments, especially 77-87.
27. See also van der Werf, "Estampie," in New Grove Dictionary of Music and
Musicians, Stanley Sadie, ed.
28. For a brief discussion and complete transcription of both songs see van der
Werf, TheExuint Troubadour Melodies, 291-93.
56 Hendrik van der Werf

assumption that their refrains derive from a practice that dance songs
were intoned by a soloist, some of whose verses were repeated by the
(other) dancers. It is a thankless task to try to disprove a theory that has
never been proven. It may suffice to give the most pertinent facts. It
should not surprise anyone to learn from the narratives that medieval
people did dance, and (occasionally or often?) did so to a song. From
some narratives we may also conclude that an alternation between a
soloist and others occasionally occurred in dancing songs. Forms of the
word "rondeau" occasionally appear as labels for a dancing song. The
noun "ballade" seems to be related to the verb battare, meaning "to
dance." Some late entries in Le Manuscrit du Rot, which in their form
resemble the virelai, have the title danssa. Beyond that, there is little
or nothing to connect all rondeaux, virelais, and ballades to dancing.

Exploring their origin a bit further, I suggest that if the rondeaux,

virelais, and ballades of Guillaume de Machaut were descendants of
dancing songs, they are at least as far removed from their origin as
Beethoven's scherzos are removed from the courtly minuet. From yet a
different point of view, we may recall that playing with recurrent
thoughts, words, and sentences is a favorite habit among poets in so-
called "primitive" as well as "high" cultures, just as playing with recurrent
melodic ideas and phrases is popular among composers in notationless
cultures. In the fourteenth century these favorite features became
stylized and standardized, and this process seems to have been started
before the beginning of that century. It is beyond question that the
rondeaux (or the rondets, as they are called in their sole source) by
Guillaume d'Amiens are different from his chansons and from chansons
by other trouveres, but no indication has been found that they have
anything to do with dancing. Their melodies may have been measured,
but there is no reason to consider ternary meter and modal rhythm
axiomatic. Since these are the only extant monophonk rondeaux, and
since they occur in only one manuscript, we do not have much material
for research on duration in their melodies. Despite all this uncertainty, I
urge that some systematic and unprejudiced study be made of all the
songs that may have had anything to do with dancing, however remote
that connection may have been, even though we may never acquire
precise knowledge of rhythm or meter in each individual case.

29. For further information see van der Werf, "Estampie."

30. For a discussion of the "balete" preserved (without music) in ms. Oxford,
Bodleian Library, Douce 308, sec van der Werf, The Chansons, 153 and 158-59.
31. For an edition of these rondets see Friedrich Gennrich, Rondeaux; Virelais,
und Ballades (Dresden, 1921), 30-38.
 "Not-so-precisely Measured" Music 57

Narrative Songs

Some two hundred French poems can be typified as "narrative songs."

This does not include large epics, known as "chansons de geste"; it
concerns relatively short, more or less strophic songs, most of which
were published a century ago by Karl Bartsch under the general labels
"romances" and "pastourelles. Several attempts have been made to
subdivide this large group into concise and easily recognizable types.
Style and content seem to have been the primary criteria for
categorization. To some extant the form of the strophe was also
considered, but the style and form of the music played no role. Thus,
past research has failed to yield, reliable information to those interested
in reviving these songs in their original rhythm. My own limited research
makes me wonder whether a typological study, taking into account all
pertinent aspects of text and music, will yield significantly fewer
categories than there are narrative songs.

As a group, the narrative songs combine formal aspects of text and music
in a manner that links them with both the trouvere song and chanson de
geste. The former is strophic in the usual meaning, in the latter the
counterpart of the strophe is normally called a laisse, the length of which
may vary widely. The subdivision of the narrative songs under
consideration is usually called a strophe even though the individual
strophes of a given song may vary somewhat in their number of lines.
Five or six strophes is almost standard for trouvere songs, while the
number of laisses in a chanson de geste is unlimited, and the total epic
may have a few thousand lines. In the narrative songs the number of
strophes ranges from five to ten in most cases, to twenty or thirty in
others. In trouvere songs a ten-syllable line has either ten or eleven
syllables, depending on whether the rhyme consists of one or two
syllables. This curious form of arithmetic is due to the tradition that in
French and Occitan poetry the second (unaccented) syllable of a so-
called feminine rhyme is not included in the syllable count. In epic
poetry and in many narrative songs a ten-syllable line may have ten,
eleven, or twelve syllables. The increase in arithmetical problems is due
to the fact that, in addition to an unaccented syllable in the rhyme, there
may be an unaccented syllable in the caesura, that is in the "break" after
the fourth or (less often) the sixth syllable. Beyond the frequently
occurring AB AB opening, trouvere songs have relatively little repetition
of entire melodic lines. In the narrative songs, however, we often find
three- or four-fold repetition of the very first phrase, which may well be

32. Karl Bartsch, Abfranzoasche Romanzen and PastoureUen (Leipzig, 1870).

58 Hendrik van der Werf

related to a tradition of singing all the lines of a laisse, or of an entire

chanson de geste, to essentially the same melodic phrase. Differences
in choice of rhyme schemes among the three genres are important but
not of direct relevance to questions of melodic rhythm and meter.

It seems inconceivable that a chanson de geste was performed in modal

rhythm or in any other alternation of long and short syllables. If "not so
precisely measured" duration was useful in any genre, it must have been
in epic poetry. Almost every narrative song turns out to have some
formal characteristics of both the chanson de geste and of the trouvere
chanson. Although this does not give incontrovertible evidence for their
rhythm, it does justify the speculation that duration in them was "not so
precisely measured." Assuming that duration in all troubadour and
trouvdre songs, in chansons de geste, in all romances and pastourelles,
and in all of plain chant was "not so precisely measured" does not
necessarily mean that it all sounded alike. On the contrary, this rhythmic
freedom allows a wide range of differences in expressiveness and rhythm,
however subtle the fluctuations may turn out to be. Let us hope that
someone will study the entire group of narrative songs, considering each
song on its own merits. The purpose of the research should not be to put
each song into a category, to deduce the rhythm for one or two in each
category, and to decide that all songs in that category had the same type
of rhythm or meter. For the present, however, we are left with no
alternative but to consider them all as belonging to one category, and
assume that duration in a given narrative was "not so precisely
measured," unless evidence to the contrary can be found.

Chromatic Alterations

Finally, we must turn to two "technical" aspects of the performance of

medieval songs. Musica ficta, or in plain English chromatic alterations,
may well form the most elusive problem in the performance of medieval
music. Although no all-encompassing study of this phenomenon has
been undertaken, many strong opinions have been voiced. The
troubadour repertory is small enough that a complete survey can be
made of the sharp, natural, and flat signs in the four sources involved; at
the same time, it is large enough to give meaningful data. For scholars
these data are of great significance, but they fail to offer precise

33. In a forthcoming publication I hope to explore the ramifications of the theory

that in a chanson de geste one line of music was repeated for every textual verse.
34. It is unclear to me why Richard H. Hoppin transcribes narrative songs in
modal rhythm in his Medieval Music (New York, 1978), 292.
35. For these data, see van der Werf, The Extant Troubadour Melodies, 38-61.
 "Not-so-precisely Measured" Music 59

prescriptions to performers. To begin with, we have no certainty as to

how long a given sign of alteration is valid, although some indication
exists that it often pertains only to the pitches on the rest of the staff on
which it stands. Beyond that, the variants among multiples versions of
troubadour and trouvere songs show that changes were made in the
diatonic or chromatic nature of some, but by no means many of the
melodies; still they neither reveal who made them nor whether chromatic
alterations were added or deleted. Medieval performers do not seem to
have been as concerned about note-for-note retention of a melody as are
present-day musicians. This attitude appears to have affected chromatic
alterations as well as other melodic aspects. Scribes contributed to these
differences by adding (or omitting) accidentals in accordance with
criteria unknown to us. Since all large chansonniers contain at least
some chromatic alterations, we can be fairly sure that troubadours and
trouveres themselves did not shun them, but we do not know whether a
given composer left us exclusively diatonic melodies, whether he altered
a pitch frequently, or whether he seldom did so. Thus, when it comes to
determining the diatonic or chromatic state of a melody, it is left to the
performer to make decisions where the scholar can only plead ignorance.

Instrumental Accompaniment

Instrumental accompaniment is another thorny issue for scholars and

performers alike. For a long time, it was widely accepted that the songs
of the troubadours and trouveres were always performed to instrumental
accompaniment. As many scholars must have done before me, and as
especially Christopher Page did after me, I have searched in vain for
information about accompaniment. We know that instruments existed in
the time of the troubadours and trouveres, and obviously that they were
used. It appears that some troubadours and trouveres could play
instruments. We also know that the Middle Ages were not unacquainted
with the phenomenon of accompanied song. For example, Tristan often
accompanied his own singing. But no evidence is present that
troubadour and trouvere chansons were accompanied. It is significant
that in troubadour and trouvere poems, as well as in the medieval
literature concerning them, we find numerous references to singing, but
none to accompanying. The mere fact that, occasionally, a musical
instrument is mentioned in a poem is no evidence for instrumental

36. The fact that both Tristan and Isolde played instruments is a welcome
antidote to the myth that in the Middle Ages playing an instruments was considered a
base occupation. Tristan and Isolde were highly admired and the story tellers would not
have portrayed them as superb instrumentalists, if playing an instrument would have been
in conflict with their noble birth.
60 Hendrik van der Werf

accompaniment, and even if we could find evidence for accompaniment

of a troubadour or a trouvere song on a certain occasion, we would still
have no evidence that they were habitually accompanied. In the only
extensive study to date of the medieval use of instruments. Page
essentially confirmed what I wrote some twenty-five years ago. He
went far beyond the troubadour and trouvere repertories so that we now
begin to have some information on what genres actually were
accompanied. One thing has not changed: we still do not have any
manuscript that actually preserves the accompaniment to any song prior
to the fourteenth century. Recital and recording situations being as they
are, it may be difficult for performers to abandon instrumental
accompaniment completely, but I express pleasure at noting that
percussion instruments seem to be losing ground in recordings of
medieval music.

On more than one occasion I have written that medieval performers of

chansons must have sung expressively, but that we do not know just how
dramatic their renditions were. We may be able to draw one more
conclusion from the nature and the number of variants among multiple
versions. As mentioned above, the variants in the melodies are not only
more numerous but also more significant than the variants in the poems.
Perhaps one can conclude that the similarities and differences among
multiple versions suggest that medieval singers concentrated on
presenting the poetry but were rather free in their treatment of the
music. In this respect we may raise a crude question: are troubadour
and trouvere chansons poems that happen to have a melody or are they
musical compositions that happen to have a text? We may not want to
answer either question with a simple "yes" or "no," but we can safely
assume that the poet-composers wanted their texts to be understood.
Thus, we may have an objective criterion by which to judge authenticity
of present-day performances, and hold that, from a historical point of
view, something is seriously wrong when a song is performed in a rhythm
and with an accompaniment that obscure the words. For modern
performers there is a painful irony in this conclusion. A performer who
sings without instrumental accompaniment, who makes the pitches more
or less equal, and who gives a perfect rendition of the text, may
encounter less audience appreciation than the one who sings in jumpy
rhythms, who is accompanied by an ensemble of odd-sounding
instruments, and who is dressed in medieval garb.

37. Christopher Page, Voices and Instruments.