Download as pdf
Download as pdf
You are on page 1of 14
Henry KLUMPENHOUWER Network Analysis and Webern’s Opus 27/III In Anglo-American music theory, the analysis of atonal music over the last 25 years has been dominated by Forte Set Theory. This essay presents a fairly recently developed alternative methodological construct called Klumpenhouwer Networks, and argues for its analytical usefulness, In particular, the essay demonstrates how the Network model is able to extend many more meaning- ful relationships than Set Theory can, pointing out analytic possibilities arising out of this ability. The first twelve measures of the last movement of Webern's Piano Variations (opus 27) provide the context for this discussion to take place. The essay presents two different interpretations of the musical excerpt, both developed within the framework of Network Analysis In the past few decades, analyses of atonal music have special reference to the beginning of Webemn’s opus been carried out with reference toa single model of _27/III. The structure of the essay is fairly straight- harmonic structure, the pitch-class collection. This _forward: it works through from the beginning of the Paper discusses a fairly recent alternative model piece, explaining various important features of the ‘called Klumpenhouwer Networks, and argues forits model along the way." analytical usefulness. The model is presented with Rai Mesend J =e 0 ir Brample There are other introductions to this analyte methedelogy, but they set out to demonstrate diferent features ofthe model. See in par- ‘eur David Lewin, "Klumpenhouwer Networks and Some lzgraphies that Involve Ther", Muse Theory Spectrum 2 (0950), 3, 'p. 8320; Henry Klumpenhouwer, “Aspects of Row Structure and Harmony in Martine’ Impromptu Number 6”, Penpecvs of New ‘Music 29 (1981), 2, pp. 38-354: David Lewin, “A Tutorial on Klumpenhouwer Networks, Using the Chorale in Schoenberg's Opus 1, No", Joural of Musie Theory 38 (194, PP. 79-40% oo Tyosenrirr voor Munieerttont, jaancane SmunenT (198) TyOSeWaIrT voor MUZIEKTHEORIE Example r gives the first page of the score. The ‘opening section of the piece lasts until the Eb in measure r2, where the tempo slows, and the texture ‘changes. The pitch structure parses easily into three r2-tone rows, indicated on example r by the reference labels J, Kand L, Example 2a abstracts the three rows; appropriate measure numbers appear at the right of each row. Fh chk coo example ab Example 2b depicts in network format the relation- ships among the three rows. The arrows extending between the row names are labeled with various pitch-class operations and serial operations. The arrow from J to K tells us that applying Ig to the pitch: class content of row J, and maintaining serial order, yields cow K. The arrow from K to Lis labeled by a Rabi Niedendd =a 80 1s Eomple pitch-class operation and an order operation. I, 38 before, shows that the pitch-class content of K is, transformed via that operation to yield the pitch-class content of L; the letter R indicates that the serial order of the transformed pitch-class elements is reversed. ‘The arrow extending from J to L, labeled simply by the letter R, represents the product ofthe operations that label the arrows from J to K and K to L. Ok FAY Gf mis Bh och AL Bh ams Dy Bh Bk Bis mm g-12 Accordingly, it shows that reversing the serial order of the pitch-class content of row J yields row L. In the following analysis, I shall concentrate on the ‘music corresponding to rows J and K. Much of the music in measures 1-9 is melodic and distinctly motivic, and I shall pursue two different strategies for partitioning the rows into melodic patterns and 2 Network Anatysis ano Weeern's Ovus 27/IN harmonic structures. Example 3 investigates the first of these strategies. It divides the music corresponding to rows J and K into cight three-note collections, labeled Jr-J4, and Ki-Kq, for reference. The first, second, and fourth collections in each row have a similar rhythmic profile: along note followed by two quarter notes. The third collection in each row serves sa motivic foil for the surrounding collections. n a B gb Bk bb | Dk cH ck | FE Bk Gh 1013) (012) 013) 34 31 32 Ki Ka Ka KE Gk OF | BY FR OPH | Ch Dh Bk (015) (012) 1013) 34 3-1 32 Sanpes Example 4 abstracts the eight collections from the score in example 3, Below each three-note collection, the example provides the relevant prime form and set- designation from Forte's catalog." We see that within each row no two collections belong to the same set- lass. Consequently, the four collections cannot be related under pitch-class transposition or inversion. However, we shall sce that under the network model of chord structure, we can relate all the collections in rows J and K, and do so in an analytically interesting way. Example 5 introduces the new model by investigating Js the first collection in row J. Example 5a shows the three pitch-classes of jas the contents of nodes in a amples Nework interprettons of ‘Transformational Network. The arrows and the pitch- class operations that label them tell us how to con- ceive the various intervals extending between the pitch-classes in Jt, The intervals are directed intervals: that is, they are defined with respect to both magnitu- de and direction. Accordingly, example 5a asks us to construe Jr as a network of three relationships: an interval from B4 to Eb via the pitch-class operation T, a4 Ak Gt 1014) 33 lr Ks | ct ak toa) 33 ‘an interval from B4 to Bb via the operation Ty,, and an interval from Bb to Eb via the operation T;, I want to stress here that the Transformational Network displayed in example 5a is an interpretation of the pitch-class collection [1 - which is, of course, itself an interpretation. Furthermore, example 52 is one of many possible interpretations of Jz. Example 5b is another. Comparing example sa and example sb, wwe see that both networks construe an interval from Bi to Bb via T,,. However, this is where the similarity ends. Where example 5a construes an interval from Bito Eb via T,, example 5b construes an interval from Eb to B4 via Ts; where example 5a construes an interval from Bb to Eb via T,, example 5b construes 2. The StuctureofAtonal Musi (New Haven, Yale University Press: 1973), Allen Forte's catalog ofa possibe collections of ptchclie tons and his discussion ofthat use inthe analysis of atonal muse, remains a central tet for Angle American “Set Theor.” Readers unfamiliar with the basics ofthe approach can develop enough sophistication by reading section through section 1.2 (pp 6). 2% an interval from Eb to Bb via T,, The networks in examples 5a and sb use only trans- position operations to interpret relationships between pitch-classes. We can also use pitch-class inversion to interpret these relationships. There is only one stipulation: the constituent pitch-class operations of a network interpretation must obey the group theoretic rules for combining transposition and inversion operations.’ Examples sc, sd, e, and sf are network interpretations of fx that use both transpositions and inversions as artow labels. Example sf, for instance, construes - a8 did examples sa and sb - an interval from B4to Bb via Try The I,-arrow in example sf asserts that Eb inverts around an axis of two pitch-classes whose associated integers sum to 1. Since inversions are reflexive operations - that is since they are always their own inverses.-the I, arrow can be construed in either or both directions: this is represented on the diagram by a tworheaded arrow. So we can also conceive of Bb inverting under I; to yield Eb. Example sf interprets the relationship B4/Eb as 1; it inverts Eb around an axis whose associated integers TyDscHuirr voor WUZIEKTHEORIE ‘sum to 2 to yield BA, and vice versa. The network in ‘example sf obeys the rules for combining trans- positions and inversions. (For instance, Ty, followed by I, produces 1,,) All of the networks displayed in example 5 are equally ‘well-formed. They do not exhaust all potential relevant network structures; they do, however, show the range of possible interpretations of Jz. None has inherent analytical value over any other: a given inter- pretation may be more or less suggestive or useful depending on a particular analytical context Example 6 sets out to establish more context. The example displays 2 number of different network inter- pretations of J4, the last trichord in row J. The first ‘two networks, exemples 68 and Gb, interpret the structure of J4 through transpositions only. The rest ‘use both transposition and inversion operations. ‘Comparing the networks in example 5 and example 6, ‘we notice certain similarities between the two net- works in example sf and example Gc. Both networks have an arrow labeled T,,, an arrow labeled I,, and an arrow labeled 13, nm a4 mi om. 45 Sample? 3 The combination of pitch las transposition and inversion may be summarized inthe following four rules 1.Tpy flowed by Ty e9uls Tey 2. followed by Ip €qUa5 I 3p fllomed by Ty, 5 In 4. followed yy 8818 Thr ‘All addon and subtraction are modulo 2, 4 Throughout the essay | ute Baboit’s notation for pitch-lass inversion. a Nerwoux Anauysis ano Weaenn's Opus 27/IIt Example 7 isolates the two network-interpretations of Juand J4 given by example sfand example 6c. The ifhustration indicates that the graphs of Jz and J4 are stricly identical, By ‘graph’ I mean the configuration of nodes and arrows and the operations that label the arrows. Corresponding arrows travel in the same direction, and have the same operation. In other words, except for the contents of their nodes - that is, except for their pitch-classes - the networks of J1 and J4.are the same. Networks related in this way are called ‘strongly isographic’. a a“ Om a" De Example 8 investigates the strong isography in more detail. The example copies over the two networks from example 7 and extends arrows from the three nodes of Jr to the corresponding nodes in the network ‘of J4. The three arrows are labeled either “+a" or ‘2’, representing two semitones in the conventional positive or negative senses for pitch-classes. BYand Bb, the pitch classes in Js related under the trans- position operation T,,, both move through the same interval in order to reach their corresponding pitch- classes in J4, At and G#, respectively. They move, that is, in parallel motion, Eb and Bb, the pitch-classes in Jtrelated via I,, move up two semitones and down two semitones, respectively, to achieve Ft and Gf, their corresponding pitch-classes in J4. They move, that is, in symmetrical contrary motion. BY and Eb, the pitch- classes in J related via 1,, move down two semitones and up two semitones, respectively, to achieve At and F}, their corresponding pitch-classes in ]4. They like- ‘wise move in symmetrical contrary motion. Example 8 is paradigmatic in the following respects: ‘Tytelated arrows suggest symmetrical parallel motion; I, related arrows suggest symmetrical contrary motion. This is an important feature of the network model: it describes harmonic structure in a ‘way that engages voiceeading as well (on) 2 Emample 9 in 7, n wo” Example g elaborates example 8, connecting the abstract structure given by example 8 to the motivic profile of the music itself. The long, sustained Eb, ‘which begins Jr, moves ‘up by two' to the long, sustained F, which begins |4. The slurred quarter- note figure BYBb, which concludes Jr, moves ‘down by two’ to the slurred quarter-note figure ALG#, ‘which concludes J4. Melody and accompaniment thus move in symmetrical contrary motion from the left side of example 9 to the right side. And the sym- metrical contrary motion is evidently connected with the preservation of the specific I-relations involved, from the left side of the example to the right side. Example 9 makes example 8 especially audible, because the motivic profile of the music strongly projects the abstract network-relationships. Example ro displays six network interpretations of J2, the second three-note collection in row J. Examples 10a and 1b use only transpositions to label their constituent arrows. The other four networks in example ro are labeled with both inversions and ‘transpositions: all obey the group-theoretic rules for combining pitch-class operations. Examining the six networks, we see that one of them - the network in example 10¢ - relates suggestively to the network interpretations of Jr and J, studied earlier in example 7. Like those two networks, the network in example 10 comprises a T,,-arrow, an L,-arrow, and an I, arrow. Example 11 abstracts the network from, example roe and places it between the network inter- pretations of Jr and J4 studied in example 7. The net- ‘work interpretation of Jz given in example roe has been visually rearranged here, to bring out its similarity to the networks of Jr and J4. As a result, vertical page position, for the central network in example 11, does not correspond to relative registral position in the music. The example shows that al three networks are strongly isographic: all have the same arrangement of nodes and arrows, and corresponding arrows are labeled by the same operations. Based on the strong isographies among the three net- ‘works, we can extend correspondences among their constituent pitch-classes. The visual alignment of net- ‘works in example 11 suggests, for instance, that Eb in Jt, Clin J2, and Fin J4 play analogous or corre- ‘sponding roles in their respective networks: all are ‘TyOSeHRIFT VooR MUZIERTHEORIE Biample 10 \Nework interpretations off2 involved in both I, and 1, relationships. Moreover, all __ inversions as arrow-labels. Examining the six net- are uniquely involved in those relationships within ‘works, we see that none is strongly isographic to the their respective networks. Similarly, Bb in Jr, Cftin _ networks of jr, J2, and J4 studied earlier. There are, J2, and G# in ]4 also play analogous or corresponding however, certain consistencies between the graph of roles in their respective networks: all are involved the network in example 12d and the graph common (again, uniquely) with theT,, relationship and the I, _to the networks of 1, J2, and J4. Example 33 abstracts relationship. Finally, we can correlate Bin Jr, Diin _on its right side the network of example 12d for J2, and A¥in J4, on the basis of their analogous net- further study; on its left side the illustration includes work positions. the graph of Jr, J2, and J4. The two graphs are dis- Example 12 shows six network interpretations of J3, _played to emphasize certain similarities in trans- the third trichord in row J. The first two networks use formational structure, and not to study the registral only transpositions to label their arrows; the re- positions of the pitch-classes involved. Comparing the ‘maining four networks use both transpositions and _two structures, we see that Ty, labels corresponding, pa 4 @ le 1 @ 10 @ \a)% \-)% es) ‘ea 4 ° 5 ds rg {2s i , i ho | @ 2 @ To( @ Ney it \ar i Yee tt & es) ee) rome 2 Nerwrk interpretation 3 » NeTwork Anatysis AND Weasan's Orus 27/I 1-32-14 +3 example arrows on both graphs. The corresponding inversion operations, however, are not the same: where the ‘graph on the left has an arrow labeled I,, the graph on the right has an arrow labeled by Typ; and where the graph on the left has an arrow labeled I, the ‘graph on the right has an arrow labeled Tyo, Yet, while the corresponding inversion operations on the ‘wo graphs are not identical, their indices do differ by a constant amount, namely, 9 (mod 12). Accordingly, ‘we can imagine an operation that captures the difference between the two graphs. This operation, ‘which is conceptually related to pitch-class trans- position, has the following effect: where the graph on the left has an arrow labeled by a transposition operation, the graph on the right has an arrow labeled by the same transposition operation; where the graph con the lefthas an arrow labeled by an inversion opera- tion with index n, the graph on the right has an arrow labeled by an inversion operation with index (n + 9) (mod 12), which we shall notate as . The angle brackets distinguish this relationship, an operation on pitch-class operations, from ordinary-T, an operation on pitch-classes. is a relationship that obeys Lewin's first rule (of five) for isography among ‘Networks of the kind we are using. The appendix to the examples reproduces Lewin’s first and second rules for network isography, the rules that will be pertinent to this paper. Example 14 studies the network interpretations of all four trichords in row J. The networks are arranged visually to emphasize corresponding elements of ‘transformational structure, and not to reflect the registral positions of constituent pitch-classes. Underneath the networks, the diagram provides the appropriate isographic relationships: represents strong isography; represents the isography investigated in connection with example 15; , the inverse of , isthe isography that yields the graph of J4 when applied to the graph of J3. ‘The display in example 14 suggests correlations among the pitch-classes of Jr through J4 under the various given isographies: Eb in J1, Clin J2, Ein J3, and Fin J4 play analogous roles in their respective networks; 50 do Bb in Jz, Cif in J2, Fein J3, and G# in. J4. And so do BAin Jr, Diin Ja, Gtin J3, and Adin J4, Example 15 investigates registral permutation through the four networks under the correlations brought about by network structure shown in example 14. The example displays the pitch-lasses of the four net- works to reflect their relative registral positions. Three registral positions are represented on the left by ‘numerals: numeral t represents the highest pitch within each chord; numeral 2, the second highest pitch; numeral 3, the third highest pitch. Underneath each arrow arrangement the example provides the relevant permutation of registral positions 1, 2, and 3 brought about by the isographic correspondences studied in example 14. The registral arrangement of the pitches in Jz permutes under ‘(r25)'to the regis- tral arrangement of J2. The expression (123) denotes that the pitch in registra position ‘’ of J relates to the pitch in registral position ‘2'ofJ2, while the pitch in registral position ‘2’ of Jr relates to the pitch in registral position ‘3’ of J2, and the pitch in registral ire z ia aoe ein 3 naa) 2 0B) comple as E eee AN 33) a position ‘3’ of Jr relates to the pitch in registral position ‘r’ of J2. The symbol (123) concisely represents the registral rotation among the three registral positions from Jr to Ja. Example 15 suggests that network structure (as inter- preted in example r4) interacts with registral struc- ture in a consistent way through the four collections in row J. Under the various isographies, the network structure rotates registrally from Jr through J2 and J3, ‘until achieving the original alignment again in J4. a comple 16 Example 16 presents the interaction in a different way. It depicts the durational profile and melodic con- tour structure of the four collections in row J. Relative registral position within each collection is shown by vertical position on the page. In each collection the first note is the longest and the highest sounding; the second note is shorter than the first and is the lowest. sounding; the third note has the same duration as the second note and lies registrally between the first two notes. Accordingly, each collection has precisely the same melodic contour, when contour is defined by the interaction of relative temporal position and ‘TYDSCHAIFT YooR MUzteKTHEORIE relative registral position within a collection of pitches. Focusing on the T,,-arrow in each network, wwe see as we read across from Jr through J2 and J3 to J4 the rotation of network structure against registral ‘order (described earlier in example 15), and temporal order. Earlier, in example 9, we noted that the motivic profiles of Jz and |4 strongly articulated the similari- ties in their network structure. Example 16 elaborates on that hearing: the correspondence of motivic struc- 4 o— ae ee | “Oty » ture and network structure is changed from Jz to J2, ‘changed again from J2 to J3, and changed yet again from J3 to J4. Each of the three changes involves the registral permutation (r23), as noted in example 15. ‘The total effect of the three changes is a complete registral tour; accordingly, J4 has the same relative registral profile as Jr. Indeed, the music for J4 has the same relative durational profile as well. With respect to dynamics, too, the music returns at J4 from forte to piano, the level of jx. All of these aspects of the music strongly articulate a retum at [4 to the original formation of J. n Nerwons Anatysis ano Wanenn's Onus 27/)i! Example 17 investigates the four trichords that consti- tute row K. The network interpretations of the four trichords in row J are copied over from example 14 for reference. The trichords in row K are all rhythmically identical to theit corresponding collections in row J- Indeed, each pitch in row K has precisely the duration of ts counterpart in row J, and occurs four measures and one half note after its counterpart in row J, Each pitch-class and each collection of pitch- ‘The network interpretations of the trichords in row K relate suggestively to the network interpretations in row j. Comparing the networks in example 17, we see certain similarities between the graphs of Kr, K2, and Ks and the graph of J3 All of these graphs have an arrow labeled Iyo and an arrow labeled Iyy. They differ only in their transposition operation: the graph of Kr, Ka, and K has an arrow labeled T,, while the ‘graph of J3 has an arrow labeled T,,. We see analo- ous similarities between the graph of Jr, J2, and J4, and the graph of Kj: Al of these graphs have an arrow labeled I, and an arrow labeled [,. They differ only in their transposition operation: the graph of Jr, J2, and ]4 has an arrow labeled T,,, while the graph of Ky has an arrow labeled T,. In example 18, the three-note collections J5 and K3 have been reinterpreted. The transposition arrow has been reversed on the networks of these collections, and the transposition operation has been appropriate- ly replaced by its inverse-related or complementary operation: T,, on the network of [3 of example 17 becomes T, om the network of J3 in example 18; T, on the network of K3 in example 17 becomes T,, on the network of Kj in example 18. On example 18 all trans- position arrows point forward in time when related to their corresponding structures in the music: this is not true of the previous network interpretations of [3 and Kj in example 17. 2 ‘The revised networks displayed in example 18 are analytically suggestive. The third collection in each. row - the collection that stands out with respect to motivic structure - also stands out in the matter of its network relationship with the other collections in that row. Moreover, the network of J3 is strongly isograp- hhic to the networks of Kr, Ka, and K4, while the net- ‘work of K3 is strongly isographic to the networks of Tu J2,and Ja. ame 19 Example 19 abstracts the two different graphs of example 18, The graph appropriate to Jt, 2. J4 and Ky appears on the left of the diagram. The graph appropriate to J3, Kx, Ka and K4 appears on the right. Comparing thern, we see that they do not relate under a relationship of the form . Taey do not label corresponding arrows with the same transposition operation. And the indices of corresponding inversion operations do not differ by a constant amount. There is, however, a regularity in the way the two networks differ. While the graphs do not label corresponding arrows with the same transposition operations, they do label corresponding arrows with inverse-related ‘transposition operations -T, and T,,. Moreover, while the indices of corresponding inversion operations do not difer by a constant amount, they do sum to a constant amount. The indices of L, and ly ssum to 0; so do the indices of l, and I,,- Accordingly, we can imagine an operation that captures the difference between the two graphs. This operation, which is conceptually related to pitch-class inversion, hhas the following effect: the transposition operation in the graph on the left ‘inverts’ to the transposition ‘operation in the graph on the right -it becomes its complement; and the two inversion operations in the ‘etaph on the left invert around an axis whose elements sum to 0 to yield the two inversion operations in the graph on the right. We'll notate this particular operation as . The relationship obeys the second rule (of five) for isography, which is reproduced ia the appendix to the examples. The angle brackets, distinguish the relationship, an operation on pitch- class operations, from ordinary Ian operation on pitch-classes. The analysis displayed in examples 17 and 18 has much to commend it. It relates all eight collections in the two rows even though they belong to four different Forte set-types, and it provides detailed structure for each collection. Moreover, by ‘way of the specific network interpretation of the third collection in each row, the analysis casts the two presented rows in a very special relationship. aig MieBend J = a. 80 Vee M N BBE Bb Dy | C8 Ch FE ER Gk | P (0145) Sf 101367) “7 S19 P Q KE Gk GE ER | FE FE Ch Dy By | Ff (014s) SF (or367) “7 519 Eomple 21 Example 20 considers another way of parsing the music in measures 1-9. It groups the pitches from ‘cach row according to the pattern of dynamic marking and the placement of rests, a strategy that has been discussed in Chris Hasty’s analysis ofthis piece By this articulation, each row comprises a four- TYBSCHRIFT vooR MuZiexTHEORIC note collection, a five-note collection, and a three-note collection. The six collections are labeled M through R for reference. Example 21 abstracts the six collections from example 20; the example includes the associated dynamic marking, and the pertinent set-class from Forte's catalog. The following analysis takes some of Hasty’s ° Fr OAN GE P (014) 33 R ch oak Bb P (014) 33 interesting observations as a point of departure and applies the network model of chord structure ‘explained in the first part of this paper. Example 22 {isolates the collections M and N, including the appropriate measure numbers and the associated dynamic marking. 5 Christopher Hasty, “Rhythm in Pos-Tonal Music: Preliminary Questions of Duration and Motion” Journal of Musle Theory 25 (19) v2, pp 183.206. 8 | Nerwonn Anatysis ano Wenenw's Opus 27/0 M N Th Bh Bb De chock FB EW Gh Pm. te FS xm23 ample 22 » Dy « Dy e FF ot Bo Bk Bh Bock ae 3 M N Example a3 Examples 23a and 23b are based on Hasty's discussion of M and N. In his analysis, he divides each collection into two parts, labeled alpha and beta on example 23. In both M and N the alpha-element involves a semitone relation: Eb/Di in example 23a; F#/Giin example 23b. The beta-element has the intervallic structure of the set-type [014]: it contains single instances of the interval classes 1, 3, and 4, a5 depicted in the diagrams. Example 23a shows that Hasty conceives of Df in measure 2 as two notes: the note Dias an element of alpha, and the note Dias an. clement of beta. » ( oD ub Ts & O-O>8 6: extends arrows labeled by inversions. Example 24b interprets the N-collection in the same way: Within alpha and beta the network extends arrows labeled by transpositions; between alpha and beta the network extends arrows labeled by inversions. Again, the dia- ‘gram is designed to emphasize similarities between networks; accordingly, page position does not reflect registral or temporal position in the music. ‘Comparing the two networks in the example, we see certain regularities. The networks have the same node /arrow configurations, which is to say, they have the same number of nodes and arrows, and corre- » ee vamp 24 Example 24 takes Hasty's division of each collection {nto two parts and invokes the network model to study the resulting structure. Example 24a is a net- work interpretation of the M-collection. Within alpha and beta the network extends arrows labeled by trans- positions; between alpha and beta the network ™ « O——o nf ty » @-0-@ mh % sponding arrows extend in the same direction, Corresponding arrows are labeled with the same ‘transpositions (that is, within each alpha-element and beta-element). The indices of corresponding inversions all differ by 6. Accordingly, we can define an isography between the two networks (but not a rcmple as u strong isography) under the relationship . ‘The last unit of row J is O (measures 4-5). While this unit only has three notes, its till possible to under. stand it within the framework of Hasty’s alpha- elements and beta-elements, together with our net: ‘work analyses of those structures, Example 242 took the pitch-class D§ to be a part of both alpha and beta: in like manner, example 25 interprets both Git and A} as members of both alpha and beta. The network interpretation of © conceives of two relationships between Aiand F: a Tg-arrow from Alto within beta; and an [,-arrow between alpha/Al and beta/F4. Example a5 includes for reference the networks of M and N, Comparing the network interpretations of M and O, we see that the two networks have a number of related elements. Where M has an arrow labeled by a transposition, O has an arrow labeled by the inverse- related transposition; and the indices of correspond- ing inversion operations in M and O sum to 6. Consequently, we can define an isography from M to O under the relationship <[¢>. Carrying out an ana- ogous comparison of the networks of N and O, we ssee that is the relationship appropriate to the isography between those two networks. ‘As [have already pointed out, the networks given in ‘example 25 are not displayed according to their pre- sentation in the music, but rather to emphasize similar elements of network structure between them. Bearing this in mind, we can make a number of observations about how the three networks in ‘TYDSCHRIFT VOOR MUZIEKTHEORIE example 25 engage with the music they interpret. In all three networks the contents of the right-most nodes of alpha and beta -that is D4 and Dt in the M network; F#+and E¥in the N network; Ak and Fin the (O network - always sound together in the music, though they do not necessarily attack at the same time, In M, the alpha and beta elements share one constituent member (D4); in N alpha and beta share no constituent members - they are articulated as com- pletely distinct elements. In O, alpha is entirely ‘contained within beta. Accordingly, through measures 15, alpha becomes more prominent and independent, then ‘disappears’ into beta, Example 26 elaborates the previous example, connecting the abstract structure given by example 25 to the motivic profile of the music itself: Example 26 hhas two staves: the alpha-clement of each network is provided on the upper staff, the beta-element on the lower staff. Accordingly, the arrows within each staff are labeled by transpositions, while the arrows between the staves are labeled by inversions. The arrow-relationships given in example 26 are just those studied in example 25. Bearing the networks of example 25 in mind, we see that from M to N the alpha clement - the T,_ arrow - is reversed; and that from N to O the beia element is reversed, so that from M to O - that is, from the beginning to the end of row J -both alpha and beta elements are reversed. Example 27 carries out network interpretations of the analogous structures in row K. The network of M, N, Nerwonn Asacysis ano Weneat's Opus 27/0 and O are included for reference. All six networks are isographic under some angle bracket relationship. ‘Comparing M and P, N and Q, and O and R, we see that each network in row J is isographic to its counter- part in row K via . There are yet other interesting consistencies among the networks. N and R are strongly isographic networks; $0 are O and Q. In both pairs, the strong isography runs counter to the moti- vic and serial associations between the rows. Strong isography relates the network of N, a collection that presents alpha and beta as distinct elements, to the network of R, a collection that does not articulate alpha as a separate motivic element. There is a parallel relationship between O and Q. @, © ‘@ te fae foe Oa Oa ® emp 2t Example 28 arranges the six networks of rows J and K {in a way that emphasizes certain consistencies in their interrelationships. The diagram divides the six networks - represented by their letter names - into ‘wo groups: M, , and Q appear along the top of the diagram; P, N, and R appear along the bottom of the diagram. The example includes the dynamic mar- kings pertinent to each collection. Each ‘group’ has. the same intervallic structure: an arrow labeled , and an arrow labeled . to represent strong isography. Moreover, all corresponding networks in the two groups relate under . Finally, dynamic ‘markings invert between corresponding -related networks: networks labeled piano relate to networks labeled forte, and vice versa. Example 29 displays the same information (minus the dynamics) in a different format. The illustration lists the labels ofthe six networks horizontally according to temporal order in the music. Only the seven relationships studied in example 28 are given in example 29. The resulting diagram in example 29 helps us to conceive with respect to the passage of the ‘music the relationships discussed in example 28. Yet ithas a certain appeal in its own right. The network of relationships among M, N, O and P is symmetrical: ‘And Q and R are strongly isographic to the center of that symmetrical structure. Example 3o is yet another diagram of the relation- ships among the six networks. Among the networks, there are only four different graphs: under strong isography, N and R have the same graph; so do O and Q. Example 30 arranges the four graphs horizontally in the order they first appear. Each is represented (at ‘a higher level) by the node that appears beneath it in the diagram. The node corresponding to the graph of network M is labeled ‘start’, since this is where the ‘music begins in measure r. The arrows trace the path taken in measures 1-9 through the circuit of the four graphs shown at the top of the diagram. The arrows themselves are labeled by relationships appropriate to Example 30 6 TyOScHRIrT vooR MUZIEXTHEORE the graphs contained in the nodes. There are only two different arrow labels, and . (Both operators are their own inverse or complement, s0 ‘they can label arrows traveling in both directions.) ‘The diagram places the graph of M and graph of P at ‘opposite ends of the circuit: the remaining graphs may be seen to mediate between the two poles. All of these features induce a ‘motivic’ structure among the networks via the graphs they contain, a structure that is analytically suggestive. Appendix “Rules or norapy of Kumpenhouwer Networks 1. Klumpenhouwer Networs (a) and (b), sharing the same configuration of nodes and arrows, wil be fographic under the circumstance that + each Tenumber of Network (bs the same asthe coresponding Tnumber of Network a), and + each Lumber of Neseore (8 i exact j more than the eomesponding number of Nett (2) The pertinent automorphism of the Tl goupis Fx Fue Cn) =Tai Fete) Insp 2. Klumpenhouwer Networks (3) and (2), that + each T.number of Network (b) is the complement ofthe comesponding T-number in Network (), and + each numberof Network () i exactly j more than the complement ofthe corresponding num in Network (3) ‘The pertinent automorphism ofthe T/lgroupis Fen: Fane (iq) Tent FONE) = Lia” ing the same configuration of nades and arrows, willbe iographic under the ccumatance Excerpted from: David Lewin, “Kumpenhouwer Networks and Some sographies Among Them", Mini Theory Spectrum 12 (1990) 7, 288 a

You might also like