INTRODUCTION—
CRAFT, SENTIMENT, AND MECHANISM IN THE MEDIEVAL LYRIC
The troubadours of medieval southern France still have an audience: not just of scholars, but also of poets, musicians, and others pursuing an interest in poetry or music. Until recently, an unduly large proportion of our attention to the troubadours (as well as to their northern French counterparts, the trouvères) has been directed at the amatory theories they elaborated.[1] Even to scholars, their love terminology seemed systematic, and its apparent precision seemed to afford glimpses into a spiritual knowledge beyond our own. As a result, their poetry was overrun by a scholarly quest for the meaning of love, their usual "theme." Still, love per se remained unelucidated by these inquiries, and post-Romantic scholars blamed the poets' insincerity for the mere words, to them a disappointing shadow, they found at the center of troubadour and trouvère lyric, where the object of the quest—amors —should have been finally snared.
With the announcement in 1949 that "the theme is but a pretext" for the interplay of conventions and forms in Old French love lyric (Guiette [1949] 1960, 15), medievalists began to learn new and productive ways of reading trouvère poetry. In turn, this formal approach developed by Guiette, Dragonetti, and Zumthor has inspired better readings of Old Provençal lyric. We can now admire songs that nineteenth- and early-twentieth-century critics found repellent, for example, the songs of Raimbaut d'Aurenga, whom Alfred Jeanroy called "the first of our funambulist poets" and blamed for offering "a foretaste of the nerve-wracking ses-
tina" (1934, 43). The modern reader sees artistic strategy where Jeanroy saw impropriety—where formal argumentation, elaborate sound patterns, and conscious weaving of the familiar with the original draw attention to the poem itself. For Jeanroy, whatever cast doubt on the sincerity of the poet's amorous homage also rendered suspect its delicacy and good taste and, thus, even its literary value. Today, beauty need not equal biographical truth, and we accept poetry made more of words than of emotion.
One of the major contributions of this formal approach is to expose and discourage anachronism: the superimposition of favorite biases, filters through which even the best-trained eyes often view the lyrics of the troubadours. Chief among the older biases is a view of poetry derived from romanticism: poetry is private, personal, and autobiographical. That this view has been sustained by the vidas and razos (the fictional "lives of the poets" and the "stories" "explaining" their work) probably reflects the outlook of their thirteenth-century biographers rather than of the twelfth-century troubadours (see Poe 1984). With the aid of the formal approach, which takes medieval lyric composition as a professional game of words, not as the embodiment of personal passions, we can outgrow the unacknowledged fancy that the troubadour Jaufre Rudel's "faraway love" was modeled after Dante's Beatrice; we can realize that troubadour poetry (unlike "all good poetry," according to William Wordsworth) was not "the spontaneous overflow of powerful feelings"—at least, not if we tentatively accept the poet whose "love is a pretext" and compare him to the poets whom Wordsworth most detested: "Poets, who think that they are conferring honour upon themselves and their art, in proportion as they separate themselves from the sympathies of men, and indulge in arbitrary and capricious habits of expression, in order to furnish food for fickle tastes, and fickle appetites, of their own creation" (Knight 1896, 49).
Wordsworth's accusation hits an unexpected mark with the poetry of "programmatic accumulation of motifs"—motifs that are themselves mere "forms of content" attached to "forms of expression." The poet does not sing, but "actualizes":
A given song of the troubadour Peire Vidal is entirely composed, as if in a game, of a programmatic accumulation of motifs which the poet must expound. The probability of the motifs, then, depends less upon their choice than upon their mode of realization. Several among them are nothing more
than forms of content, to each of which are attached several forms of expression, with greater or lesser chances of actualization.
(Zumthor 1972, 230)
The discovery of a "network of lexical fields" in the Northern grand chant courtois has, in a sense, restored trouvère song to intelligibility for our late-twentieth-century minds. Zumthor's implied analogy with electronic circuitry familiarizes poetry based on permutation and combination, on the endless recurrence of standard courtly motifs "which activate one another," just as constant use familiarizes the heartless machinery we depend on. Probability and randomness make themselves at home in the literary critical terminology applied to medieval poetry. The mechanistic view of literature is, in fact, becoming one of our own computer-age reading biases for literature, as much as the emotional view dominated nineteenth- and early-twentieth-century criticism. This is not necessarily a disaster, as long as we do not begin to suppose that word frequency, actualization, or probability were matters of concern to medieval poets.
As if in reaction to this mechanistic view, we are now seeing a timely emphasis on the voice as the "body" of poetry (Zumthor 1982, 1987; Ong 1982). The voice, as the nearest thing to our sense of touch, is much more "physical" than the visual representations of words; in this sense, poetry created for oral performance stands in something of a "physical" (more than intellectual) relationship with its audiences. It is also active, in the sense that through speech—and even more through song—one body physically acts on another; the "performative" aspect of troubadour lyric thus takes on new significance.
It is tempting to map the successful formal approach to trouvère lyric directly onto the quite different terrain of troubadour poetry. And yet, though it may prove genuinely valid, we can impose a poetics of the northern chanson on the southern canso only after testing its appropriateness. Not merely a difference in language separates the troubadours and the trouvères, but a difference in the cultural climate for both poetics and politics. The social position of women, for instance, differed greatly in that women in the south could legally own land. "Langue d'oc" expresses a poetic tradition other, and slightly older, than the derived traditions of "langue d'oïl." When Zumthor argues for a high degree of correspondence in the vocabulary of the two languages ("one could easily establish a lexicon of these equivalencies"; 1972, 191), he encourages the reader to generalize a theory intended to describe the songs of the trou-
vères. Yet few would maintain that a description of the Old French conventional love lyric can be generalized to Occitan lyric before 1150–1170.[2] The period of "duality of form and, especially, of content" (1972, 190), which Zumthor bypasses, actually offers a variety that is not necessarily dual. The "programmatic" stage has not yet been reached, and troubadour poets debate vigorously and ingeniously over the merits of potential "programs."
To include Occitan poetry in the poésie formelle described by Guiette, Dragonetti, and Zumthor, we would have to ignore a fundamental difference between the troubadours and the trouvères: originality and individuality were of prime importance to the troubadours, whereas the trouvères strove primarily to refine convention. The works of Marcabru, Peire d'Alvernhe, Giraut de Bornelh, Raimbaut d'Aurenga, and Arnaut Daniel scarcely resemble the poetry of the trouvères: "Major poets did not only not seek to conform to tradition, but attempted to shape a developing literature to their own individual concepts of eloquence" (Paterson 1975, 6). As we shall see, these poets whom Paterson singles out for their individuality are important figures in a movement favoring authorial originality and the conservation of "the legitimate versions" of their songs. Still, even troubadours who open their works to adaptation and re-creation emphatically contradict the conception of the vanishing author, whose voice "is stifled in a composite, neuter, oblique text, destructive of personal identities," and whose stylistic signatures are swept away, along with his name, in the wave of medieval anonymity: "The author has disappeared: there remains the subject of the enunciation, a speaking insistence integrated in the text and inseparable from its functioning: 'that' speaks" (Zumthor 1972, 69).
We need only recall Peire d'Alvernhe's "Cantarai d'aquestz trobadors" to realize how far from this je impersonnel were the troubadours of Peire d'Alvernhe's generation, thus confirming Paterson's view:
Cantarai d'aquestz trobadors
que canton de maintas colors
e·l peier cuida dir mout gen;
mas a cantar lor er aillors
q'entrametre·n vei cen pastors
c'us non sap qe·s mont'o·s dissen.
D'aisso mer mal Peire Rotgiers,
per qe n'er encolpatz primiers,
car chanta d'amor a presen;
e valgra li mais us sautiers
en la glieis'o us candeliers
tener ab un gran candel'arden.
(P d'Alv 12, 1–12)[3]
I will sing about these troubadours, who sing in many "col-
ors," and even the worst of them thinks he recites very nicely;
but they will have to sing elsewhere, for meanwhile I see a hun-
dred shepherds such that not one of them knows whether he is
going up or down.
Thus Peire Rogier is unfortunate, because he will be the
first accused, since he sings about love in public; and he would
be better off with a psalter in church, or holding a candlestick
with a large burning candle.
This joke about Peire Rogier's originality, the insinuation that he should resort to singing from a psalter instead of composing his own songs, can be funny only if originality was the norm. At the same time, as a bit of literary criticism, the stanza acknowledges a difference in intention between persona and poet, both of them caricatured: the image of Peire Rogier as a choirboy "holding a large burning candle" suggests a contrast between the apparent innocence of Peire's lyrics and their implicit sexual content, intimidated but no less "ardent" in sacred territory—granted that sometimes a candle is just a candle. Peire d'Alvernhe creates just such a distinctive portrait for each poet; some of these vignettes are known to be based on lines from their subjects' songs, and the caricatures mock the way each represents himself to his audiences.[4]
"Individual concepts of eloquence," defended by poets "que canton de maintas colors," were expressed partly through personalities; the special trademark of each poet's "performing self" is more than a mark of genre specialization. It is the stamp of authorship, more or less firmly imprinted on every troubadour poem. When transmitters did not know who composed a poem, they would usually attribute it to someone: out of 2,542 troubadour songs only about 250 come down to us as "anonymous"; we know the names of some 460 troubadours (Pillet and Carstens 1933).
One peculiarity of medieval poetry that might be presumed to depend completely on authorial self-effacement and anonymity—the "essential mobility of the medieval text"—is if anything more pronounced in troubadour poetry than in other medieval lyric traditions. Zumthor's general
description of medieval manuscripts applies in every detail, and in the extreme, to the songs of the troubadours:
Variants in the flower of the text: words, isolated turns of phrase, variants bearing on more considerable fragments, added, omitted, modified, substituted for others; alteration or displacement of parts, variants in the number and sequence of elements. (1972, 71)
For a body of poetry so preoccupied with individual style, how can this general trend in medieval literature hold true? How can it happen that "the notion of textual authenticity seems to have been unknown" (Zumthor 1972, 71)? Yet Zumthor's concept of mouvance, of the changeable medieval text or, as he puts it, "the text creating itself" (p. 73), becomes more and more relevant to troubadour poetry the more we learn about its transmission. The question then becomes, how did these adamantly individualistic poets reconcile themselves to the fact (if they knew it) that their songs would weather and change and be restored, like barn murals, perhaps within their own lifetimes?
The "openness" of texts governed throughout their transmission by mouvance, as opposed to the "closure" of texts resisting its tyranny, is one of the central distinctions addressed here. "Open texts" are relatively unfamiliar to twentieth-century readers, whose assumptions were shaped mainly by print media rather than by oral or electronic modes of "publication." Yet it will become apparent that the "open text" was a norm against which certain troubadours—rather unsuccessfully—rebelled. It was also a norm that many troubadours embraced and fostered.
This book concerns itself with the style of troubadour poetry as it might have been influenced by available modes of transmission. An attentive reader soon realizes that transmission preoccupies the troubadours: even when the phrase "to send a messenger" is absent, the closing lines (tornada, equivalent to the French envoi ) "send" the song to one or several specific persons or places, or to a general audience. Conscious as they were of their songs' destinations, the poets anticipated their songs' destinies as well. The idea that some troubadours might have tried to interfere with mouvance —to stop the "moving text"—suggests intriguing possibilities: for example, I will show that trobar clus (closed poetry) was a style associated with the effort to control the quality of circulation and thus (with only moderate success and intermittent application) to allow poets to compose fixed texts. By contrast, some poets evidently welcomed
mouvance . Jaufre Rudel seems to invite revision by future singers (Pickens 1977). This finding, when we pursue its implications through a large body of evidence, leads to the important discovery that Jaufre was not alone among his contemporaries and successors in adopting this attitude, so alien to the modern view of poetry. As we shall see, many other poets offer the same invitation and encourage, or at least facilitate, the recasting in performance of their poetry.
If the unity and the permanence of perfection associated with fixed poetry were not valued by the troubadours, then we must ask what the troubadours did value as essential to their lyric composition. To answer these questions, I examine several sources of knowledge about the circulation of troubadour song.
The transmitters reveal their methods in at least two ways. The compilers who made the anthologies, as well as the theoreticians who from the thirteenth century gave expert advice to young poets, occasionally offer straightforward comments on their experience in learning, composing, or collecting songs. Other transmitters—anonymous performers, scribes, and compilers—have left us only indirect evidence: traces of their work surviving in manuscript variants. I survey the more direct evidence in Part One, "Making and Sending: A View from Within the Literary Texts."
Circumstantial evidence is investigated in Part Two, "The View from Without: Performance and Poetics Reflected in the Chansonniers ," where, through a statistical analysis of the works of twenty-three troubadours, I trace some important causes of stanzaic transposition, the feature of troubadour mouvance that most seriously disarms literary interpretation. Part Two looks at the transmission of troubadour lyric from the outside, from the perspective of our time as we attempt to recapture the medieval text. It begins to sort out what can be deduced from the number and formal properties of extant copies. Here the "chirographic folk," as Ong (1982) would call them, record the creations of a performing art and appropriate it to the newly ascendent culture of the book. Discrepancies among the recorded versions of the same "poem" illuminate its life as performance as well as the circumstances of its encryption in script.
Chapter 4, "Mouvance in the Manuscripts," discusses the results of a statistical survey aimed at correlating such features as manuscript survival, rhyme complexity, and stanza length with the mutability of a given poet's work. The combination of evidence from transmitters and evidence from poetry is enlightening, once we can assume an early transmission
that was largely oral. When transmission shows that a particular rhyme scheme was "easy to learn," then the poet can display willingness to accommodate the transmitters either by mentioning that his song is leu ad aprendre (which may be a true statement or an ironical one) or by using the "easy" schema without comment. Chapter 5, "Rhyme and Razo: Case Studies," evaluates the stabilizing effects of certain rhyme schemes, using both statistics and detailed analysis of exemplary songs whose manuscript transmission shows how these schemata fared in practice. In some cases these songs tell us explicitly what kind of "stability" the poet hoped to give his song. The recasting of songs was so predominant a feature of transmission that it was accepted even by those who could conceive of the idea of a closed text.
More valuable than the testimony of the transmitters is that of the poets themselves. Part Three, "Poetics and the Medium," explores the vocabulary of metaphors and images with which the poets allude to mouvance and textual integrity, to the "closed" and "open" text, to the genesis of a song's perfection or of its flaws. In seeking the poet's "intention" as to the way a song would circulate, the modern reader can never be sure that a given text represents the "authentic," original song in the poet's "own words."[5] Yet that text can at least offer the stylistic statement of someone who represents himself (or herself) as the author, who dons the persona of the original poet. At worst, it is the statement of a transmitter active at a time when the song was still a "living thing."
Troubadour songs tell us, subtly or overtly, what the poets expected from their transmitters; I focus on those statements of style that display the poets' awareness of their mode of transmission and their efforts to adjust to it or rebel against it. Instructions, criticisms, and praises for their transmitters: these constitute the poets' most direct "testimony" and thus provide key evidence for Part One. The poets also answer the question of who the primary transmitters were, allowing us to assess the role assigned by poets to professional and nonprofessional singers. Taking into account the rhetorical setting for such statements of style, one can still find there the reflection of actual performance practice.
Part Three returns to literary questions. Here I apply what has been learned about transmission to reexamine twelfth-century controversies about poetics. If we can now share some of the poets' assumptions about transmission, we can also read more clearly their references to it. I address in this way not only the set of literary terms surrounding trobar clus and trobar natural with their concurrent issues of "closure" and "legiti-
macy" or "authenticity," but also an array of images that stand metaphorically for notions of textual integrity and mutability.
The book culture that arose in the mid-thirteenth century and had by the fourteenth transformed the idea of the "text" from a verbally woven thing to a visually fixed object comprised part, but not all, of the transmitting cultures responsible for the texts that survive in manuscripts today. This new book culture began to produce the chansonniers about a century after the poets in my study composed their lyrics. Thus, many of the conclusions drawn here will not apply to later poets, from 1250 onward, who lived to see their works anthologized and for some of whom the songlike quality of poetry became perhaps rather a topos than a reality. It is likely that a performing tradition carried on twelfth-century troubadours' songs long after their deaths, independent of manuscript makers who captured on parchment the versions that they "gathered from the air" from time to time in the period when manuscripts were being made. The present study thus applies primarily to poets of the twelfth century and to the transmitters who succeeded them.
The study of the visual aspect of songbooks as books and as artifacts offers us new perspectives on the reception of troubadour lyric among literate clerics who made the parchment anthologies (see Huot 1987). The copyists' conception of what a lyric is does not necessarily conform in every detail to the poets' conception or to the audiences' conception. But to the extent that we can learn how a given anthologist understood his task, we discover a link in the chain of transmission and its influence over the form in which he passed down troubadour songs.
To discover how troubadour style might have been shaped by the poets' expectations about jongleurs, audiences, and the system of circulation, one needs to examine an abundance of poets' comments on style and transmission. For this reason I concentrate on a period in the development of this poetry when self-conscious artifice prevailed, when the poets participated in frequent exchange of literary ideas, when they liked to explain their poetics to audiences capable of appreciating such information. The generation of 1170 is ideal for this purpose. Like trobadors before and after them, they send poems to one another for approval, they deliver harangues in person, they engage in contests, they exchange ten-zones, and they borrow stanzaic patterns from friends in hopes of outdoing them. Their "individuality" is abetted rather than isolated by the awareness of active competition and controversy: the starling returns
with a reply rather than flying off "lonely as a cloud" into the private outlands of independence. Undeniably by 1170, many troubadours did "cultivate a traditional poetry of form"; however, they prized freshly made forms instead of moving toward a small canon of fixed forms. At this time, poets became especially liberal with allusions to their own art: its techniques, its purposes, its hazards. Their individuality springs from their participation in a literary community, with each of their maintas colors contributing to the spectrum.
I have chosen some of the poets essential to this inquiry from among those Peire d'Alvernhe teases in his satire of their "many colors." Between June and September 1170, according to Walter T. Pattison's hypothesis (1933), there converged at or near Puivert (Aude) Eleanor II's wedding party, a large group of noblemen, and a following of troubadours:[6]
Lo vers fo faitz als enflabotz
a Puoich-vert, tot iogan rizen.
(P d'Alv 12, 85–86)
The poem was made for the stuff-bellies[7] at Green Peak, all in
fun and laughter.
The poem lets us glimpse a bright literary constellation as it once actually convened. It is the sense of literary comradeship reflected in Peire's satire, as well as his assurance that all will join in criticizing and appreciating the unique style and character of each singer, that leads me to begin with this group of poets.
Of the twelve trobadors mentioned aside from Peire d'Alvernhe himself, only four names are recognizably those of famous poets. I have already drawn attention to Peire's "inculpation" of Peire Rogier. This poet has been dated to the third quarter of the twelfth century, partly because he spoke as an "experienced senior" to Raimbaut d'Aurenga (born ca. 1144) as early as 1167.[8] He may have been the eldest at the gathering and thus earned first place (and an ironic portrait as an innocent choirboy); however, the descriptions in "Cantarai d'aqestz trobadors" provide no real evidence for a poet's age.
Everything about the passage introducing us to Giraut de Bornelh suggests age: he is described as "a dry wine-skin in the sun" and compared to "an old woman carrying a bucket" looking sadly into a mirror. Yet Giraut, "master of the troubadours," outlived all the other known poets in the group: his documented activity spans nearly forty years (fl. 1162–1199). With these dates, he could have been one of the younger poets
gathered in 1170. It is well known that for the troubadours, jovens (youth) was a frame of mind rather than a time of life; Peire d'Alvernhe's comments on Giraut, then, tell us more about Giraut's style than about his seniority at the gathering.
The high esteem in which Bernart de Ventadorn was held by his contemporaries can perhaps be gauged by the diminution he undergoes in Peire's satire. "A handsbreadth shorter than Bornelh," Bernart is described not as an adult but as the child of his mother and father. The "ardor" of Bernart's love songs is implicitly ascribed to the heat of his mother's baking-oven. Again, his "youth" is figurative: Bernart's activity as a poet is documented between 1147 and 1170, the gathering at Puivert being his last recorded appearance. Bernart is without doubt the most famous troubadour today, so much so that an inordinately large proportion of discussion of "the troubadours" in general is based on the favorite two or three of his poems.
Of Raimbaut d'Aurenga, Count of Orange and the ninth poet in Peire's gallery, much more is known because of the importance of his family, whose financial doings—testaments, property transfers, oaths of allegiance, receipts of rent—are comparatively well documented.[9] Peire accuses Raimbaut of overvaluing the pipautz ("derelict" singers) who come to him for charity. It is true that Raimbaut entertained at his castle Courthézon at least two other troubadours—Peire Rogier and Giraut de Bornelh. But we must not suppose him a wealthy patron; he spent most of his life in debt and died at an early age (about twenty-nine) in 1173, probably from the epidemic of influenza in that year. The jongleur with the beautiful voice, Levet, was present at the signing of his testament. So was another young poet—and his presence at Raimbaut's deathbed has escaped the notice of both troubadours' editors—whose imaginative invention owed much to Raimbaut, both in his versification and in his comic style. That poet was Guilhem de St-Didier, who deserves much more attention than he gets from modern scholars.[10]
Finally, Peire d'Alvernhe describes himself; there are two versions, and in neither of them does Peire spare himself the satirical treatment he has applied to the others. His voice is so bad, he says, that he sings "like a frog in a well." Alternatively, he sings "both above and below." Either way, hardly anyone can understand his words:
Peire d'Alvernge a tal votz
que chanta con granoill' en potz,
lauza·s mout a tota gen;
pero maistres es de totz,
ab c'un pauc esclarzis sos motz,
c'a penas nuils horn los enten.
(P d'Alv 12, 79–84)[11]
Peire d'Alvernhe has such a voice that he sings like a frog in a
well, and he praises himself much before all kinds of people;
but he is the master of all, if only he would clarify his words a
little, for scarcely anyone understands them.
Peire's documented literary career stretches from 1149 to 1170; lake Bernart de Ventadorn, he makes his last recorded appearance in this literary "gallery."
Bernart Marti, whose poetry much resembles that of the influential older poet Marcabru (fl. 1130–1149), perhaps attended the Puivert gathering under the name Bernatz de Saissac (Roncaglia 1969a). His dispute with Peire d'Alvernhe over the practicality of "whole poetry" (vers entiers ) is only one of many signs that he participated in working out the literary issues confronting the poets of 1170. In Peire's self-description, he is still defending himself against Bernart's accusation of wrongly claiming de totz maistria (mastery of all); even now Peire affirms that he is maistres de totz, although others (especially Bernart Marti) have misunderstood his words of self-praise—he should have "clarified his words a little." These stylistic controversies, and others like them, are discussed in Chapters 6, "Nature Enclosed" (on trobar clus, trobar natural, and related problems), and 7, "The Metaphorical Vocabulary of Mouvance and Textual Integrity".
I also include in my study several near-contemporaries of the poets in Peire's gallery, especially those who either contributed to the stylistic debates of circa 1170 or profited from them, whether by providing models for innovation or by borrowing and refining the techniques of others. The most important of these contemporaries are Rigaut de Berbezilh (fl. 1141–1160),[12] Guilhem de St-Didier (fl. 1165–1195), and two Catalan poets, Pons de la Guardia (fl. 1154–d. 1188) and Guilhem de Berguedà (fl. 1138–1192).
Two figures have been included not because of contemporaneity with the poets of 1170 but because of the methods of their modern editors: Jaufre Rudel (fl. 1125–1148) and Peire Vidal (fl. 1183–1204). Both Rupert T. Pickens and d'Arco Silvio Avalle address questions of transmission, textual instability, and writing; the two editors' opposing hypoth-
eses about the poets' efforts to "preserve their songs" reflect a division far more important for the issues dealt with here than the chronological gap between the two poets. While Jaufre Rudel's editor suspects that the poet invited revision of his poems, Peire Vidal's editor claims that the poet compiled his own book to fix his texts for posterity. The two styles of editing depend on two opposite ways of understanding "the lyric text," the mode of transmission, and the importance of authenticity. It will make a great difference for the editing of Provençal song whether other poets, by their own account and their transmitters', tend to resemble Jaufre Rudel as Pickens sees him or Peire Vidal as Avalle sees him.
Finally, no investigation of troubadour poetics can be complete without Arnaut Daniel (fl. 1180–1195). A case can be made for his association with the poets of circa 1170, especially with Raimbaut d'Aurenga, but even without such "literary contacts" he would be useful for this study.[13] He perfects many of the techniques proposed and debated in 1170, and he enjoys describing the goals and processes of composition.
My statistical survey includes twenty-three poets, from Guilhem IX, the first troubadour, to performers of the mid-thirteenth century. After that time there is a historical burgeoning of the influence of literacy and the book culture on poetry, even poetry that relied on conventions developed by the troubadours. My thesis, that the twelfth-century troubadours were ambivalent toward the concept of the fixed text and tried alternately to promote it and to undo it, no longer applies when later poets take reading as the model of reception and writing as the model of composition.