A piece of music can h ave an ut t erly consist ent, relent less pulse and y et give t h e same impressions of rest and mot ion, pause and accelerat ion. T he ere is never a moment wh en an eigh-not e durat ion is not sounded somewh ere in the t ext ure of t
A piece of music can h ave an ut t erly consist ent, relent less pulse and y et give t h e same impressions of rest and mot ion, pause and accelerat ion. T he ere is never a moment wh en an eigh-not e durat ion is not sounded somewh ere in the t ext ure of t
A piece of music can h ave an ut t erly consist ent, relent less pulse and y et give t h e same impressions of rest and mot ion, pause and accelerat ion. T he ere is never a moment wh en an eigh-not e durat ion is not sounded somewh ere in the t ext ure of t
Author(s): Joseph P. Swain Source: Music Theory Spectrum, Vol. 20, No. 1 (Spring, 1998), pp. 48-71 Published by: on behalf of the Society for Music Theory Stable URL: http://www.jstor.org/stable/746156 . Accessed: 05/10/2014 03:17 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp . JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org. . Oxford University Press and Society for Music Theory are collaborating with JSTOR to digitize, preserve and extend access to Music Theory Spectrum. http://www.jstor.org This content downloaded from 193.144.2.35 on Sun, 5 Oct 2014 03:17:03 AM All use subject to JSTOR Terms and Conditions Dimensions of Harmonic R h y t h m Joseph P. Swain It is one of t h e more deligh t f ul paradoxes t h at a piece of music can h ave an ut t erly consist ent , relent less pulse and y et give t h e same impressions of rest and mot ion, pause and accelerat ion, t ension and resolve t h at f reer rh y t h ms migh t give. Consider t h e opening of t h e Bach A Major Prelude (WTC II) given in Example 1: t h ere is never a moment wh en an eigh t h -not e durat ion is not sounded somewh ere in t h e t ext ure. Indeed, such is not at all unusual: t h ere must be t h ousands of composit ions, ch ief ly in Baroque st y le but also in ot h er st y les, t h at could make t h e same claim, including most of t h e ot h er pieces in t h e t wo volumes of Th e Well- Tempered Clavier.1 Yet it is one of t h e great ach ievement s of Bach 's art t h at no sense of const raint at t ends t h is eigh t h - not e obsession. On t h e cont rary , t h e dy namic of t h e Prelude's t ensions mocks it s rh y t h mic unif ormit y . Wh en t h e dominant h armony of t h e second measure moves back t o t h e t onic in 'Bach is t h e t ext book case f or t h is ef f ect . Maury Yest on writ es "Th e consist ent rat e of at t acks in regular sixt eent h not es is t h e reason t h at Bach 's music h as so of t en been used t o illust rat e t h e principle t h at a sequence of st eady pulses must be dif f erent iat ed by some ot h er crit erion in order t o creat e aest h et ically signif icant rh y t h m." Maury Yest on, Th e St rat if icat ion of Musical R h y t h m (New Haven: Yale Universit y Press, 1976), 40. t h e t h ird, t h ere is barely any of t h e expect ed resolut ion. In- st ead, a subt le t ension begins t o build over t h e t onic pedal sounding t h rough out , a sense of somet h ing h eld back wh ich y earns t o proceed, a power suddenly loosed in t h e f ourt h measure int o a brief glory t h at quickly subsides int o recom- mencement in t h e dominant key . To be sure, t h e t h ird bar h as t wo voices sh aring t h e eigh t h - not e pulse f or t h e f irst t ime; t h e f ourt h h as it t raded bet ween t wo voices on every dot t ed-quart er beat f or t h e f irst t ime; and t h ere is somet h ing about t h e bass line st art ing in m. 4, using rest s as part of it s melody f or t h e f irst t ime, t h at implies f ast er ch ord ch anges. But a lit t le t inkering, as suggest ed in Example 2, t o put t h e same f eat ures in t h e opening bars produces no ef f ect like Bach 's; t h ese explanat ions of t h e paradox are at best part ial, if not simplist ic. And t h e long t onic pedal in t h e alt o (m. 3) h ardly suggest s t h e t ension t h at seems t o build t h e longer t h e not e is h eld. Crit icism of Baroque music h as languish ed wit h out an an- aly t ical t ech nique f or h armonic rh y t h m. Virt ually any music could prof it f rom a bet t er underst anding of t h is complex ef - f ect , but in t h e case of Baroque music, and perh aps f or Bach 's music above all, t h e underest imat ion of h armonic rh y t h m blocks essent ial underst anding of t h e st y le. Th e musical lan- guage t h at evolved in t h e sevent eent h cent ury depends upon a consist ent rh y t h mic pulse-mot or rh y t h m-as it s ch ief means of cont inuit y , it s primary means of connect ing local This content downloaded from 193.144.2.35 on Sun, 5 Oct 2014 03:17:03 AM All use subject to JSTOR Terms and Conditions Dimensions of Harmonic R h y t h m 49 Example 1. J. S. Bach , Prelude in A Major f rom Th e Well-Tempered Clavier, Book II, mm. 1-6 I rI Text ure JIJ J~ J Tm JT3 | r Ph enomenal BassPit ch J bJ J JJ b| J R oot 1 J J) J b J);; J. A EA E A b A c f E Densit y R oot 2 I|:~Terf h i T M f L- U1rL r3 m7m T h lm 6m m JrJ JJ SbJ [ Im JJ- mmm J~. f J dIJ f l g$ A A g A g A b D A gi A gJ A gJ A bD A I I I I I I I I I I I I I I I I I I I I I 0o A o. E J. A Funct ion J . J J J J |J J I, J. A: I V I V I V I IV V 0. A: I J dlJ IV V I I 0. V -bJ DJ-;7J.- V I V V I V J. IV This content downloaded from 193.144.2.35 on Sun, 5 Oct 2014 03:17:03 AM All use subject to JSTOR Terms and Conditions 50 Music Th eory Spect rum Example 1 [cont inued] 4ext ur J r STe 7im m m \JI* m m m mm Ph eno ea BassPit ch J J J J) R oot I J J* J17Ji D b c b gI D J J 71J 7 gi ct E A f l c f b D J 3 J TJ J JJ g E b A b Ag AE B go E b A b A go E A E B Densit y .. . . . .. I I I R oot 2 J . J. . J. J. J. J iJ. D b go A f t gl A gt A E Funct ion A: IV VIV V I I IV I IV V I IV J. v I IV I V I V J. J. IJ. J. J. J. V I IV V I V 'J. A: IV This content downloaded from 193.144.2.35 on Sun, 5 Oct 2014 03:17:03 AM All use subject to JSTOR Terms and Conditions Dimensions of Harmonic R h y t h m 51 Example 2. R evision of Example 1, mm. 1-2 A M'- J- - -f bLf ^ , ._ . _ , I st ruct ures and building percept ible h ierarch ies wit h t h em.2 Because t h e primary element t o organize t h at pulse, along wit h met er, is f unct ional h armony , an underst anding of Ba- roque h armonic rh y t h m amount s t o not h ing less t h an an un- derst anding of st ruct ures and musical ef f ect s at t h e most ac- cessible level of t h e list ening experience.3 Considering t h e long t radit ion of h armonic t h eory in t h e West , it is surprising t h at t h e explicit concept of h armonic rh y t h m is but a h alf -cent ury old, at t ribut ed t o Walt er Pist on.4 Here is Pist on's def init ion of h armonic rh y t h m in t h e 1944 edit ion of t h e Harvard Dict ionary of Music: 21 use "musical language" not only as a met aph or, but in t h e t ech nical sense t h at I describe in Musical Languages (New York: Nort on, 1997), es- pecially ch apt er 8. Very brief ly , I t ake musical language t o be a mode of h uman cognit ion, h eld in common by members of a musical communit y , t h at ent ails an invent ory of discret e sounds (ph onology ) wit h rules of combinat ion (sy nt ax) t h at t oget h er produce ranges of semant ic ef f ect s wit h in specif ic prag- mat ic cont ext s (genres). 3I use t h e problemat ic t erm "st ruct ure" h ere and elsewh ere in t h is paper ch ief ly in it s connot at ion as "organizat ion," as somet h ing t h us responsible t o h uman percept ion and cognit ion. See Joseph P. Swain, "Wh at is Meant by 'Musical St ruct ure'?" Crit icus Musicus 2/12 (1994): 20-44. 4Mary Irene Arlin, "Harmonic R h y t h m in Select ed Fugues f rom Th e Well-Tempered Clavier, Book I" (M. M. t h esis. Indiana Universit y , 1965), 6. Th e rh y t h mic lif e cont ribut ed t o music by means of t h e underly ing ch anges of h armony . Th e pat t ern of t h e h armonic rh y t h m of a given piece of music, derived by not ing t h e root ch anges as t h ey occur, reveals import ant and dist inct ive f eat ures af f ect ing t h e st y le and t ext ure. Ch ief of t h ese are t h e f requency of t h e h armonic ch ange, and t h e rh y t h mic qualit y of t h at ch ange. It is a f airly broad def init ion, encompassing not only t h e ex- pect ed "root ch anges" and "f requency of h armonic ch ange," but also "rh y t h mic qualit y ." Clearly , h owever, t h e ch ange of root was t h e aspect of t h e def init ion wit h t h e most immediat e t ech nical applicat ion and relevance in Pist on's own t ext - books, as well as in t h e f ew st udies of h armonic rh y t h m t h at h ave appeared since t h en.5 Th e lit erat ure of h armonic rh y t h m is small, despit e t h e undeniable signif icance of Pist on's insigh t , because inh erent complexit ies conf ound it s applicat ion. Th e t roubles wit h ot h er kinds of rh y t h mic analy sis come t o mind: bey ond t h e list ing of durat ions and t h eir obvious role in mot ivic con- st ruct ion, no one h as f ound a consist ent sy nt ax of durat ions, per se, except met er. Th e t h eory of met er and it s percept ion 5See ch apt er 1 of Arlin, "Harmonic R h y t h m," f or an overview of t h ese st udies t o 1965. This content downloaded from 193.144.2.35 on Sun, 5 Oct 2014 03:17:03 AM All use subject to JSTOR Terms and Conditions 52 Music Th eory Spect rum h as encouraged a signif icant lit erat ure indeed,6 but alt h ough h armonic f unct ion is of t en ment ioned as a f act or in est ab- lish ing a met er, t h e act ual h armonic rh y t h m h as f ound no role in met ric t h eory . Ironically , it is t h e h armonic element t h at provides t h e ent ree f or rh y t h m int o a more mult i-leveled sy nt ax.7 But t h e main problem t h at h as blocked all progress in t h e analy sis of h armonic rh y t h m is t h e def init ion of a h armonic event . Calculat ing durat ions is a simple mat t er, but wh at are we t o count as a ch ord? Th e applicat ion of Pist on's simple f ormula-ch anges of root -t o Bach 's Prelude in A major runs int o t rouble almost immediat ely . Th e f irst measure present s no problems t h rough t h e f irst eigh t eigh t h -not e pulses, but we t h en encount er an all-t oo-f amiliar dif f icult y . Sh all t h e B-D dy ad in t h e bass and alt o on t h e nint h eigh t h -not e pulse be considered a new root ch ange, wit h a missing f if t h ? Th at view would be consist ent wit h t h e pat t ern of ch anges t h us f ar, but we must t h en ignore t h e soprano (a pedal t one?) and t h e dy ad's pat ent linear cont inuit y . Th e inescapable quest ion, t o be f aced again and again in t h is piece and t h ousands like it , is: wh at are t h e rh y t h mic ef f ect s of st ruct ural versus ornament al h armony , and h ow t o record t h e dif f erence? Th is leads direct ly t o t h e weigh t ing problem so f amiliar t o f ormal t h eorist s. Durat ion alone is uncomplicat ed: durat ions of single not es are one-dimensional values and are t h eref ore 6A f ew recent t h orough going t reat ment s of t h e problem of met er are: Wallace Berry , St ruct ural Funct ions of Music (Englewood Clif f s, N.J.: Prent ice-Hall, 1976), ch apt er 3; Fred Lerdah l and R ay Jackendof f , A Gen- erat ive Th eory of Tonal Music (Cambridge, Mass.: MIT Press, 1983), ch apt ers 2 and 4; Joel Lest er, Th e R h y t h ms of Tonal Music (Carbondale and Ed- wardsville: Sout h ern Illinois Univerit y Press, 1986), ch apt ers 3-4; and William E. Benjamin, "A Th eory of Musical Met er," Music Percept ion 1/4 (1984): 355-413. 7In a sense t h at is t h e approach carried out in Yest on's St rat if icat ion of Musical R h y t h m, wh ere Sch enkerian crit eria are used t o abst ract h igh er levels of rh y t h mic act ivit y . See especially ch apt er 3. commensurat e by nat ure. A h alf not e is t wice t h e durat ion of a quart er and t h at is all t h ere is t o it . Percept ibilit y does not ent er int o t h e quest ion; it is assumed t h at dif f erent du- rat ions are equally comput able. But h armonies not only h ave lengt h ; t h ey also h ave st rengt h and dif f erent iat ed f unct ion. Th e f irst t h ree A major h armonies in t h e Prelude, f or in- st ance, all h ave a quart er-not e lengt h , but t h e second is in- vert ed and t h e t h ird h as t h e lowest bass. Are t h ey rh y t h mi- cally equivalent , t h en? Th e concept of h armonic rh y t h m presumes percept ibilit y , but h armonic t h eory h as long sup- posed, t acit ly or ot h erwise, t h at pit ch simult aneit ies are not equally salient . Wh at t h en of dissonant ch ords, such as sev- ent h s, or suspensions? Does it mat t er t h at a root ch ange is accomplish ed by t h ree moving voices inst ead of one or t wo? How are t h ese dif f erences t o be reckoned? Finally , t h e h armonic qualit y of ch ord f unct ion, it s es- sent ial sy nt act ic role in ph rasing, also demands considerat ion, cert ainly in music of t h e h igh Baroque. If we discount t h e weak-beat passing t ones f or t h e moment , m. 4 h as a D major ch ord on t h e f irst dot t ed quart er beat and a b minor on t h e second. R oot analy sis would sh ow a ch ange, but f unct ional analy sis would sh ow t wo ch ords of subdominant f unct ion. Can t h ese alt ernat ives be reconciled in one analy sis? How about t h e pedal t ones in t h e opening t h ree bars? Is t h is ev- idence of h igh -level f unct ion? In sh ort , a h armony in t h e West ern t radit ion h as mult iple propert ies wit h mult iple ef f ect s. A simple def init ion of h ar- monic rh y t h m as "pat t ern of ch ord ch anges" is def icient be- cause it f ails t o speak t o t h ose mult iplicit ies and t h ereby f ails t o account f or t h e f act t h at a h armony may ch ange in some of it s propert ies wh ile h olding const ant in ot h ers. If past st ud- ies of h armonic rh y t h m seem t o be ch aract erized by arbit rary ch oices made in t h e f ace of such dilemmas, it is only because t h e t ools of analy sis were t oo crude f or t h e job. West ern h armony is a complex ph enomenon wit h many f acet s, and it s h armonic rh y t h m calls f or a much f iner dissect ion. This content downloaded from 193.144.2.35 on Sun, 5 Oct 2014 03:17:03 AM All use subject to JSTOR Terms and Conditions Dimensions of Harmonic R h y t h m 53 THE SEVER ALDIMENSIONS OF HAR MONIC R HYTHM I propose t o dif f erent iat e an analy sis of h armonic rh y t h m along t h e lines of six h armonic propert ies, h ere called di- mensions. Th is t axonomy of h armonic rh y t h m recognizes t h e individual cont ribut ions of root ch anges at t h eir several pos- sible levels, of course, but also t h e cont ribut ions of h armonic f unct ion, t h e bass voice, "non-h armonic" t ones and ch ords, and t h eir relat ions in t h e f ramework of t h e wh ole t ext ure. In t h is way , many of t h e uncomf ort able ch oices f orced by a Pist on-t y pe analy sis of a work such as t h e Bach Prelude are avoided, but not sidest epped. Alt ernat ive h armonic values sh ow up at some place in t h e analy sis. Th e B-D dy ad is recognized bot h as a "real" h armony wit h h armonic f unct ion and as a melodic prolongat ion of an A major t riad, each realit y h aving it s ef f ect on t h e list ener's experience. Th e derivat ion of t h e analy ses varies according t o t h e na- t ure of t h e dimension at h and. Some, including t h e dimen- sions of ph enomenal h armonic rh y t h m and bass pit ch rh y t h m, proceed in st raigh t f orward f ash ion like an arit h met ical cal- culat ion; f ormalized rules could easily be writ t en f or t h em, if any one want s t o. Ot h ers will resist any f ormalizat ion t h at complet ely excludes part icular judgment s because of t h e un- predict able and unavoidable ef f ect s of cont ext .8 Since alt ernat ive h armonic values are represent ed at dif - f erent places in t h e analy sis, t h e f ull benef it of such repre- sent at ion obt ains only wh en t h e various dimensions are com- 8In f act , t h e larger cont ext of t h e crit ical mat t er under inquiry sh apes decisively any result ing analy sis. Some quest ions may require only t h e analy sis of but one or t wo dimensions, a part ial pict ure of t h e h armonic rh y t h m. Th e t ech niques t h at f ollow are t o a large ext ent separable, and wh ile it is t rue t h at t h eir conjunct ion allows cert ain kinds of crit ical int erpret at ions t h at a less compreh ensive view migh t not , t h ere is no need f or an exh aust ive analy sis of every dimension if t h e mat t ers under invest igat ion do not require it . Th e point is t o ref lect t h e complexit y of t h e experience of h armonic rh y t h m, it s mult i-f acet ed real percept ions, as t h e basis f or an inf ormed crit icism of t h e music. Such complexit y may not alway s be at issue. pared. Indeed, t h e inf ormat ion given by cert ain dimensions could be more easily f ound by direct inspect ion of t h e score. We sh all see, h owever, t h at t h e ef f ort of drawing all t h e dimensions severally y ields h igh dividends-an ent irely new view of poly ph onic t ext ure t h at , like a spect rograph , sh ows at a glance t h e dy namics of rh y t h mic-h armonic t ension and ot h er relat ionsh ips in t h e poly ph ony t h at would be ot h erwise h idden. I. THE R HYTHM OF THE TEXTUR E Th e f undament al st andard f or comparing t h e dimensions in any composit ion is t h e rh y t h m of t h e t ext ure. Th e rh y t h m of t h e t ext ure is t h e pat t ern of durat ions con- sist ing of t h e f ast est moving rh y t h ms in any voice at any given moment . Ot h er t h eorist s h ave called t h is composit e rh y t h m.9 St rict ly speaking, rh y t h m of t h e t ext ure is not h armonic rh y t h m at all, since it f requent ly h appens t h at a h armony will be reit erat ed wit h out ch ange in any cont ribut ing voice, as in t h e beginning of Vivaldi's "Wint er" Concert o (see Example 3). It is t h e st andard of comparison of all ot h er h armonic dimensions, h owever, because it is t h e sum of all rh y t h mic act ivit y . Any new pit ch at t ack regist ers in t h e rh y t h m of t h e t ext ure.10 In many Baroque composit ions, part icularly t h ose 9Joel Lest er discusses t h is concept in Th e R h y t h ms of Tonal Music, 6. Wallace Berry (St ruct ural Funct ions, 204) uses t h e t erm "t ext ural rh y t h m" but in a more general way , most ly t o indicat e t h e ef f ect s of voices ent ering and exit ing t h e t ext ure. l?Th is t enet can be derived logically once t h e ot h er dimensions of h ar- monic rh y t h m are def ined, as will be seen. I can t h ink of only one possible except ion: wh en a consonant or implied h armony of unst able f unct ion, such as a dominant ch ord or dominant pedal, is sust ained f or so long t h at it acquires a local st abilit y in t h e list ener's percept ion. In t h at case t h ere is indeed a ch ange in h armonic f unct ion wit h out any corresponding ch ange in t h e t ext ural rh y t h m. But not e t h at t h is ch ange is unlike any ot h er rh y t h mic ch ange: be- cause it is impossible t o ident if y a precise moment wh en t h e ch ange t akes place, it h as no discret e arit h met ical value, as do all ot h er mat t ers of durat ion. This content downloaded from 193.144.2.35 on Sun, 5 Oct 2014 03:17:03 AM All use subject to JSTOR Terms and Conditions 54 Music Th eory Spect rum Example 3. Ant onio Vivaldi, Concert o Op. 8, No. 4 "Wint er," I, mm. 1-12 Ir 'r -r 'Er simile -A I A A A A I.AAAA A..AA .A .- ^ ... * F iz= rrrr rrrr rrrrrrFrr rrrr rrrr rr r r r FFFFFFFI f - F- T rrrr ^ " -- k -I.I I? . I I I 6L - I "' Text ure r rrrf r Ir rrrrlrrrrf rrlrr r rr rrr rrrlrrrrrrr I Ph enomenalo io |o Jo 0| ? BassPit ch o Jo- I o o| o R oot I o |o lo | o |,o |o ? f e- f Densit y | R oot 2 o [o lo Lo o 1 o-I f Funct ion o Io Jo Io, o Lo 1 f :I V I o Lo1 1o I Ioo Io? l I f : I --m-----_ .. ....... _ _ _ _ _ _ _ _ . This content downloaded from 193.144.2.35 on Sun, 5 Oct 2014 03:17:03 AM All use subject to JSTOR Terms and Conditions Dimensions of Harmonic R h y t h m 55 Example 3 [cont inued] 7 Text u re' f c' t r f l I f r f rrrL Ph enom. o I| o r r Bass o o Pit ch R oot ? |o o I b~ F G Densit y , I R oot o o , -o F Funct ion co bf : [I] IV c f : IV c: [IV] V _ C,o f : I c: [IV] O o o f l- Grr cL=Llr G c I . I 0 lo G 40 1 i: V I -43- 1 o I o c Io I # CrR rrr rr rUrR r rrr rrrrrrrrr r f LIT- J- - - - This content downloaded from 193.144.2.35 on Sun, 5 Oct 2014 03:17:03 AM All use subject to JSTOR Terms and Conditions 56 Music Th eory Spect rum of Bach , t h e t ext ural rh y t h m will prove t o be a st ream of some const ant value, like t h e relent less eigh t h not es in t h e A major Prelude. Th at is t h e ext reme expression of t h e Baroque sy n- t ax of mot or rh y t h m. Ot h er composit ions, of course, will h ave more variet y . Wh at ever t h e case, t h e rh y t h m of t h e t ext ure is t h e base f rom wh ich t o compare any and all of t h e ot h er dimensions of h armonic rh y t h m. 2. PHENOMENAL HAR MONIC R HYTHM Th e A major Prelude of Bach not only sh ows an at t ack on every eigh t h -not e, producing an ut t er consist ency of pulse, but also sh ows t h at f or each of t h ose at t acks at least one voice h as ch anged pit ch , producing a new simult aneit y . If we f orget all about t y pe and f unct ion f or a moment , wh at remains of "h armony " is t h e f undament al concept of t ones sounding t oget h er. Any ch ange in t h at combinat ion, wh et h er or not a dif f erent ch ord is creat ed, is a rh y t h mic event of h armonic aspect , t h e dimension of ph enomenal h armonic rh y t h m. Bach h as coupled h is mot oric t ext ure wit h an equally mot oric ph e- nomenal h armonic rh y t h m, not at all an inf requent f eat ure of h is art , and never one t o be underest imat ed. Th e analy sis of t h is dimension involves import ant deci- sions bet ween pit ch and pit ch -class orient at ions. In t h e passage f rom t h e "Et in unam sanct am" aria f rom Bach 's B Minor Mass sh own in Example 4, f or inst ance, sh all we count t h e second and t h ird eigh t h -not e combinat ions of m. 4 as a single quart er-not e durat ion because t h e t h ird is just a re- ordering of t h e pit ch -classes of t h e second, or sh all we view t h em as t wo dist inct vert ical ph enomena because t h ey are clearly dist inct set s of pit ch es? In my own analy ses I h ave t aken a st rict "ident ical pit ch es only " approach , ruling out simple oct ave t ransposit ions as ident it ies. Th ere is t oo much evidence of t h e percept ual dif f erences made by oct ave t rans- posit ions t o ignore, despit e t h e undeniable role of oct ave equivalence in most west ern sy nt axes.1' Besides, t h is equiv- alence, and t h e pit ch -class orient at ion in general, can be re- f lect ed in t h e dimension of root h armonic rh y t h m lat er on. Th at is one of t h e virt ues of t h is t axonomy : t h e loss of a cert ain h armonic aspect in one dimension can be rest ored in anot h er. Once t h e decision about pit ch classes is made, t h e analy sis of ph enomenal h armonic rh y t h m is complet ely st raigh t f or- ward: simply record every durat ion of every vert ical com- binat ion. A glance at t h e analy sis of t h e excerpt f rom Vi- valdi's "Wint er" Concert o sh own in Example 3 sh ows t h at t h ere can be enormous dif f erences bet ween t h e rh y t h m of t h e t ext ure and t h e ph enomenal h armonic rh y t h m and t h e root ch anges bey ond. In Bach , it is of t en t h e case t h at bot h t ex- t ural rh y t h m and ph enomenal h armonic rh y t h m are so con- sist ent , and of t en ident ical, t h at combining t h em in one graph is best f or visual clarit y -as I h ave done in Examples 1 and 4-so long as t h ey are not f orgot t en wh en int erpret ing t h e annot at ions. 3. BASS PITCH R HYTHM Th eorist s h ave long recognized t h at t h e bass voice exer- cises more inf luence on t h e propert ies of a ch ord t h an any ot h er, t h ough it is not necessarily t h e ch ord root .12 One f a- mous reason f or dist inguish ing bet ween t h e bass voice and root movement is t h at in so doing t h e bass is able t o maint ain it s own melodic int egrit y . But anot h er reason is t h at t h e bass est ablish es a dimension of h armonic act ivit y dist inct f rom st rict root movement . "Th e lit erat ure on t h is subject is enormous. For a review of some of it , see David But ler, Th e Musician's Guide t o Percept ion and Cognit ion (New York: Sch irmer, 1992) ch apt er 7; and W. Jay Dowling and Dane Harwood, Music Cognit ion (New York: Academic Press, 1986), ch apt ers 5-6. '2Joel Lest er, Composit ional Th eory in t h e Eigh t eent h Cent ury (Cam- bridge: Harvard Universit y Press, 1992), 25. This content downloaded from 193.144.2.35 on Sun, 5 Oct 2014 03:17:03 AM All use subject to JSTOR Terms and Conditions Dimensions of Harmonic R h y t h m 57 Example 4. J. S. Bach , Mass in B Minor, "Et in unam sanct am," mm. 1-13 . L >r^ m > r nr1 nrTr P-~ r7w-1 " I.1 --7 l Im \r I r 6ot bw I \I r I r 'm JlJ' IJ-M' -J' I''- Iz- m Im-r m -n' Im I A g# -AE A g# - A J J b I lJ mJ h lr'j A1j. J JIJ. J E A b E A g# - ,,J=_ ...J I,'J 7 A E c# g# c# g# A A E c# g# c# g# A Densit y I I I R oot 2 J _ J | J. A g# - A Funct ion J3 J J)| J) A:IIV I V I AJ. L A: I I I II I I II i....... J EAU J AJ. J A J. b EA b E A g# - A E IJ. c# J. E I A:V I VE:[I] J, J , I E: I IV V I V I J A 1J. J AJ. J. IJ:. J h i;J IJ IV V I IV V I V "' E: I V IV V d. i Jl Text ure Ph enomena Bss pit ch R oot I I~~~~~~~- .-- j MLt y l7J3' jTj->j ~-JTMJt e JJ< ~! ii+; ~ - rt l b !,-,.j J 'J Ai J-. I M r- R P'# # XD J Wir MJ 5r M J - d7 I d# - E d# - E 7 $ 7 A J)J A d# -E I A: V IV V I IV This content downloaded from 193.144.2.35 on Sun, 5 Oct 2014 03:17:03 AM All use subject to JSTOR Terms and Conditions 58 Music Th eory Spect rum Example 4 [cont inued] 8 Text ure :7 :n Jm Ph enomenal Bass pit ch J c f I m B R oot I J A f J - BJ A c# f # d# - B mm r- Jm m m, I, m ; 7- 7 J 7 |J j. Il r, IJ. E A d# -E B E c# 7 J | 1 J 7 J * IJ J, J. I d_ m ,lJ. E Ad# - E BE Densit y ,, i * i R oot 2 J. J. IJ. , . iJ. . . I J. J. I . . I J. A f # B E A B c# E A B E iIVV I v I I IVI lIV v r1 v I I I J . . I. IJ I . V I I V V I V J. IJ V I Funct ion A: I J. E: IV nE:VV JE: VIV VI / ..~ h # n,..--, iX ._ I_ _ I I1 1- 1 r jr -_ I rJ L J J 'Sj'^ ^ r ^ r ^ r Aj IJ I Jo gr - This content downloaded from 193.144.2.35 on Sun, 5 Oct 2014 03:17:03 AM All use subject to JSTOR Terms and Conditions Dimensions of Harmonic R h y t h m 59 Example 5. Giocch ino R ossini, II Barbiere di Siviglia, "La Calunnia," mm. 42-45 42 BASILIO 4 - 1t n^ TT F - I Dal-la boc-ca f uo-ri u- scen-do \as~zr-i np a 7 _ _ _ Th e dimension of bass pit ch rh y t h m recognizes t h e psy - ch ological signif icance of mot ion in t h e bass voice. Simply put , any ch ange in t h e bass not e suggest s a ch ange in t h e h armony .13 Th is is somet h ing bey ond t h e ph enomenal, of t en independent of root ch anges. Th is ef f ect is perh aps best h eard in a rapid alt ernat ion of ch ord f act ors sounding under a single h armony , such as degrees 1 and S under a t onic or I and S under a dominant ; Example 5, f rom music of R ossini, demonst rat es. Endless use of such a t ech nique may be f ound in ninet eent h -cent ury It alian opera arias, in piano rags wit h t h eir st riding lef t h and, and in much salon music. It is prob- ably no coincidence t h at examples f rom t h e ninet eent h - cent ury most easily come t o mind, f or such a bass allowed composers t o simulat e a h armonic rh y t h m wh ile maint aining in t h e act ual root movement t h e slower mot ion so import ant t o t h e R omant ic musical language. 13R oger Sessions seems t o agree: "In individual inst ances, it is t rue, a ch ange of posit ion may somet imes become rh y t h mically equivalent t o a real h armonic ch ange; generally wh en t h is is t rue it is t h e result of a st riking skip in t h e bass, or of some ot h er kind of vigorous cont rapunt al movement ." R oger Sessions, Harmonic Pract ice (New York: Harcourt , Brace and Company , 1951), 79. Once again t h e issue of pit ch versus pit ch class arises, now wit h part icular relevance due t o t h e f requency of oct ave leaps. Are t h e oct ave E's in m. 2 of t h e A major Prelude (Example 1) t o be regarded as a cumulat ive eigh t -beat not e or not ? Again, I pref er t h e st rict int erpret at ion of pit ch es, not pit ch classes, and f or much t h e same reasons as bef ore. To argue ot h erwise would be t ant amount t o accept ing t h at t h e f ive E's could be in t h e same oct ave wit h out delet erious ef f ect s. "Wh at is t h e bass voice?" is a more t icklish quest ion. Here t h e easy way out - "t h e lowest sounding pit ch " -is t h e wrong way out . Th e not ion of "bass voice" invokes a sy mbolism t h at , like h armony it self , encompasses a number of dif f erent aspect s. One of t h ese is it s associat ion wit h a specif ic range, of t en wit h a specif ic inst rument , t h at ident if ies it as a special melody wit h special f unct ions in t h e t ext ure. Th e bass is perf ect ly capable of dropping out of t h at t ext ure, and in f act , t h at exit and subsequent re-ent rance can creat e signif icant rh y t h mic art iculat ions f or t h e t ext ure as a wh ole. It would be quit e silly t o look f or bass pit ch es among t h e oboes d'amore wh en t h e cont inuo is rest ing in mm. 6 and 7 of "Et in unam sanct am" (Example 4), just t o f ash ion a smoot h cont inuit y wit h wh at precedes. On t h e cont rary , t h e absence of t h e bass This content downloaded from 193.144.2.35 on Sun, 5 Oct 2014 03:17:03 AM All use subject to JSTOR Terms and Conditions 60 Music Th eory Spect rum allows t h e oboes' cont ramet ric progressions t o become a kind of insert ion, a f unct ional parent h esis wit h in t h e h armonic cont inuit y of t h e real bass, barely maint ained by it s almost insouciant int errupt ion on t h e downbeat of m. 7. It is a de- ligh t f ul means of producing rh y t h mic variet y wit h in an un- f lagging st ream of eigh t h s. So t h e analy st must answer t h e quest ion "wh at is t h e bass?" wit h sensit ive considerat ion of t h e cont ext . Somet imes t h e answer is not uncont roversial. Is t h e bass st ill a bass wh en it leaps an oct ave in t h e t h ird measure of t h e A major Prelude and begins t o sh adow t h e soprano's melody a t ent h below? Once t h e bass voice h as been isolat ed, derivat ion of t h e bass pit ch rh y t h m is quit e simple, as long as t wo t h ings are remembered. One is t h at t h e graph sh ows t h e rh y t h m of t h e bass pit ch es, not necessarily t h e individual bass not es; wh en t h e bass voice does repeat an exact pit ch , as in m. 2 of t h e Prelude, t h e durat ional values of t h e repeat ed pit ch es are t ied t oget h er. Th e ot h er is t h at , because t h e bass can drop out of t h e t ext ure, rest s f ind t h eir righ t f ul place in t h e graph , t oo. Cognit ive ret ent ions of root ident it y and h armonic f unct ion are recognized in ot h er dimensions. 4. HAR MONIC R OOT/QUALITY R HYTHM It is principally t h e root ch anges t h at int erest ed Walt er Pist on.14 His analy t ical sy mbols are t h e t radit ional R oman numerals, wh ich ident if y t h e root in t h e cont ext of an es- t ablish ed key , of course, but t h ey also h int at t h e h armonic 14Again R oger Sessions essent ially agrees: "By 'ch anges of h armony ' is, of course, meant ch anges wh ich involve real h armonic cont rast -t h at is, ch anges of root . A mere ch ange of posit ion, or a progression wh ich includes only dif f erent inversions of t h e same root ch ord, involves movement of a dif f erent t y pe" (ibid.). I h ave t ried t o account f or h is second t y pe wit h my dimension of bass pit ch rh y t h m. aspect of f unct ion. Funct ion is clearly relat ed t o root , but is just as clearly a separable mat t er.15 Since t h e general st rat egy h ere is t o compreh end t h e var- ious h armonic aspect s by t easing t h em apart , h armonic f unc- t ion, a complex mat t er wit h it s own levels of art iculat ion, is lef t f or anot h er dimension. Th is is a great simplif icat ion; it obviat es any need t o declare t h e key in ef f ect and relat e t riad root s t o it . Th e dimension of root lqualit y h armonic rh y t h m indicat es but t wo aspect s of any ch ord: it s root and it s t riad qualit y (major, minor, diminish ed, or augment ed). R ecog- nizing t h e t riad qualit y is quit e essent ial: a ch ange in t h e t h ird f act or of a sust ained t riad can h ave t remendous ef f ect , t h ough t h ere is no ch ange of root . Each durat ional sy mbol in t h is line of my examples is accompanied only by a simple indicat ion of t h e root pit ch wit h a sign f or qualit y : upper case if t h e t riad is major, lower case if minor, an added plus sign (+) if augment ed, or an added minus sign (-) if diminish ed. In t h e analy ses present ed h ere it h as not been necessary t o add sevent h s or h igh er f act ors. (Th e addit ion of a sevent h may of course ch ange t h e h armonic f unct ion, and t h at is so indicat ed in anot h er dimension.) However, in some musical languages t h e qualit y of t h e sevent h const it ut es import ant rh y t h mic art iculat ions, and t h e sy mbolic not at ion would h ave t o be amplif ied t o indicat e t h em. It is rarely in t h e Baroque composer's int erest t o disguise t h e ident it y of a root , but in lat er musical languages ambiguit y h as powerf ul expressive pot ent ial. To a great ext ent , graph ing t h e f orce of clarit y in t h e arrival of a new h armony is given '5Th ey are not ent irely separable, of course, nor does a decision about root necessarily precede one about f unct ion. As I argue in ch apt er 4 of Musical Languages, circular or f eedback relat ions are inevit able consequences of per- cept ual sy st ems. My decision t o record t h e last eigh t h -not e beat of m. 1 in t h e Bach Prelude as a root ch ange is based part ly on my h earing t h at ch ord as a weak subdominant f unct ion t h at proceeds t o a st rong dominant . Th ere is not h ing t o apologize f or in ent ert aining such cross inf luences; t h ey are part of crit ical judgment . This content downloaded from 193.144.2.35 on Sun, 5 Oct 2014 03:17:03 AM All use subject to JSTOR Terms and Conditions Dimensions of Harmonic R h y t h m 61 over t o ot h er dimensions of h armonic rh y t h m st ill t o come-- densit y and h armonic f unct ion-but even in Baroque music t h e mat t er of ident if y ing root s may arise in cases of single pit ch es and monoph onic lines. Th ere is no all-purpose rule. In one case, such as t h e f irst f our not es of "Et in unam sanct am" (Example 4), t h e analy st may judge t h at , t h ough no t riads are sounded, t h e sy mbolic import of scale degrees i, 7, and S is st rong enough t o elicit f amiliar h armonic pro- gressions on at least one and perh aps t wo levels. In ot h er cases, an unaccompanied not e may clearly cont inue t h e t riad last sounded; t h is would need t o be ref lect ed in t h e durat ional sy mbol f or t h at t riad. In y et ot h ers, t h ere may be no t riadic implicat ion at all, part icularly in analy ses of non-t onal music; t h is dimension t h en drops out of t h e graph and we f all back on t h e t ext ural and ph enomenal h armonic rh y t h ms. Baroque music, h owever, cert ainly h as it s sh are of am- biguit ies regarding wh at will count as a "real" t riad, including all t h e at t endant mat t ers of inversions, t h eir relat ive weigh t s, and so on. Some of t h ese complexit ies are solved by repre- sent ing t h eir ef f ect s in ot h er dimensions; ot h ers are expressed by graph ing root /qualit y ch anges on more t h an one level. In t h e f irst measure of my analy sis of t h e A major Prelude in Example 1, f or inst ance, t h e f irst level of root /qualit y ch ange (R oot 1) sh ows movement by quart er and eigh t h not e and includes any h armony t h at is sounded and spelled as an au- dible t riad. Th e second level (R oot 2) views t h at ent ire mea- sure as a prolongat ion of a single A major h armony , graph ed as a dot t ed wh ole not e, t h e lengt h of t h e measure. Th us we capt ure t h e af orement ioned B-D dy ad wit h in t h e t h ird beat as bot h a "real" t riad, wit h palpable f unct ion t o be graph ed lat er on, and as melodic elaborat ion of a governing A major t riad. Th is f amiliar solut ion, of course, brings wit h it a f amiliar problem: est ablish ing t h e crit eria f or h ierarch ical analy sis. Af t er more t h an a h alf cent ury of cont roversy and experiment over t h is point , t h ere is cert ainly no def init ive solut ion h ere. In t h ese analy ses I adopt but t wo guiding principles: con- sist ency of t reat ment , and audibilit y of t h e h ierarch y imag- ined. Consist ency simply means making t h e same decisions in similar sit uat ions. Treat ing t h e B-D dy ad as a real t riad on t h e f irst level in m. 1 implies t h at similar dy ads are similarly t reat ed in measures t h at f ollow. Of course, t h e principle is h ardly ironclad; ch anging cont ext will h ave it s way . Th us t h e t est of audibilit y . Mult i-level graph ing implies t h at a passage can be h eard or imagined in mult iple modes, generally as melodic prolongat ions of a root , at t h e same t ime, each mode ref lect ed in a level of t h e graph . Th is working rule implies t h at t h e number of levels may ch ange in t h e course of an analy sis, as t h e t ext ure and h armonic procedures ch ange, and t h ere is cert ainly no prescript ion f or a cert ain number of levels t h at one must h ave, as such prescript ions of t en f ab- ricat e st ruct ures wh ere none can be h eard.16 So a complex piece like a prelude or an aria of Bach may present mult iple root st ruct ures as a mat t er of course, wh ile a simpler con- cept ion such as t h e ch orale sh own in Example 6 calls f or but one level of root /qualit y ch ange. Th is is not h ing more t h an a common-sense ref lect ion of t h e f act t h at musical t ext ure may be simple as well as complex. Th e part iculars of t h e h ierarch y arise f rom t h e part iculars of t h e composit ion at h and. Hear h ow caref ully Bach insist s upon t h e cognit ive realit y of t h e f irst level of root /qualit y ch ange in t h e opening of h is Prelude. Th ere are only t wo voices, but t h e h armonic ident it ies are absolut ely clear. Not h - ing suggest s t h at any t one is not a f ully f unct ioning t riad f act or unt il t h e sevent h beat , wh en t h e soprano begins t o acquire a pedal qualit y , and wh en some pit ch es in t h e lower voices begin t o t ake on more ambivalent f unct ions. But by t h en, even as t h e h igh er level becomes subst ant ial, it is much t oo 16For a more f ormal t reat ment of h ierarch ical h armonic rh y t h m t h at does ext end t o quit e abst ract levels, see Lerdah l and Jackendof f , Generat ive Th eory , ch apt er 11. This content downloaded from 193.144.2.35 on Sun, 5 Oct 2014 03:17:03 AM All use subject to JSTOR Terms and Conditions 62 Music Th eory Spect rum Example 6. Ch orale Herr Got t , Dich Loben Allen Wir f rom t h e Genevan Psalt er (1551) 0~ ?~. I:J W- ' 801 O ::O 8 J d::y ' loF%er0oI 88~~~~~~~~~~~~~~~~~~~~~~~4 . ' I , Text ure o Io o Io Io I Ph enom. I J J 1o o o o o1 Bass pit ch o J J J I o o o o R oot 1 eJ | bo 0 L G D e b e D G I I . J J J oI o o I I J Jloolo J J Jl oo lo o IH J Iooloo J J Joo o o IJ J J I?o olo J Joolo IoI lJ J J JI D e C G D e DGD G C f # - G o j J J J J jo o J JJ J Jo o o J J J J o o0 D Ge D af # -G D G I I I I I I II I I I I I I I I I I I I I I I I I Jo I V I Ir o IV I io o J J Ii V I o io? o4o V I J J J I V I o 0j o o JoJo. oolo IV I V IV V I IV V I JJJJo IV V IVV o J 0. 0 o V I IV V lat e t o quest ion t h e int egrit y of t h e f irst level. Th at is wh y Bach can compose t h e second bar in a much dif f erent way , t h e oct ave leaps in t h e bass nearly causing t h e f irst level t o meld wit h t h e second, only t o break f ree again in t h e t h ird bar. Th is is but a glimpse of t h e kind of soph ist icat ed rh y t h mic art iculat ion t o be revealed wh en we allow h armonic st ruc- t ures t o come in and out of f ocus in analy sis. 5. R OOT CHANGE DENSITY It seems a t ruism t o say t h at h armonic ch anges can be emph at ic or subt le, unmist akable or barely not iceable, and y et previous analy ses of h armonic rh y t h m h ave communi- cat ed only t h e f act of ch ange wit h out indicat ing t h e salience of t h at ch ange. In music af t er t h e Baroque, emph asizing an import ant h armonic ch ange can be as simple as bringing in t h e f ull orch est ra or crash ing down on t h e piano key board- any kind of dy namic cont rast will do. But passages of earlier music of t en go a long way wit h out ch anging dy namics or t ext ure t o any signif icant degree, making t h e weigh t ing of h armonic ch anges more dependent on subt ler f orces. We ac- count f or t h ese subt let ies in t h e dimension of root ch ange densit y . R oot ch ange densit y is simply t h e number of voices t h at ef f ect a ch ange of root .17 Ot h er crit eria could be considered - 17Yest on discusses t h e "densit y of simult aneous at t acks or t h e densit y of simult aneous pat t erns" (St rat if icat ion of Musical R h y t h m, 46), and ot h er writ - Densit y Funct ion o I ol,J I This content downloaded from 193.144.2.35 on Sun, 5 Oct 2014 03:17:03 AM All use subject to JSTOR Terms and Conditions Dimensions of Harmonic R h y t h m 63 inst rument at ion, f or example-but I h ave f ound t h at t h e number of cont rapunt al voices in t h e t ext ure gives t h e most f ait h f ul ref lect ion of t h e ef f ect of marking h armonic ch anges. In t h e A major Prelude, f ull of pedals and suspensions, Bach generally avoids t h e maximum of t h ree-voiced root ch anges unt il h e leads int o t h e modulat ion in m. 5. It is as if t h at progression, at sigh t rh y t h mically unremarkable, must be sin- gled out f or special emph asis, t h us cont ribut ing t o t h e sense of recommencement in m. 6. Wh en a root ch ange arrives wit h less t h an maximal den- sit y , t h ere is eit h er a common t one h eld over f rom t h e last t riad, a st rong-beat dissonance such as a suspension or ap- poggiat ura, or, as in t h e Prelude, pedal t ones. But wh at about dissonance? Surely it could be argued t h at a dissonant t one lends salience t o a root ch ange wit h at least as much power as a consonant , albeit new t one does. Th ere is, h owever, a t rade-of f in t h is mat t er, f or salience is as much a product of cert aint y about t h e ident it y of t h e new root as of ph enomenal emph asis on it s arrival. Any disso- nance, wh et h er arising f rom ch ord sevent h s or h armonic f igurat ion, necessarily makes t h e ident it y of t h e ch ord less explicit , since it present s t h e possibilit y of anot h er root in- t erpret at ion wit h t h e dissonant not e it self being a t riad mem- ber. On t h e downbeat of m. 6 in Bach 's Prelude, f or example, eit h er not e in t h e acoust ic dissonance A-B could t urn out t o be a ch ord t one upon resolut ion: t h e result ing t riad could be A major if t h e B resolves upward, in wh ich case t h ere would be no root ch ange f rom t h e previous beat , or E major, t h e one act ually inst ant iat ed wh en t h e A resolves downward. Indeed, any dissonant combinat ion of t ones allows at least ers h ave addressed t h e issue, but not , t o my knowledge, wit h t h e specif ic h armonic applicat ion and def init ion t h at is h as h ere. Lest er comes closest wh en h e writ es "One t y pe of t ext ural accent is caused by at t acks in many or all voices of a t ext ure. Th ese point s of densit y are accent ed in relat ion t o t h ose point s at wh ich only one or a f ew voices h ave at t acks" (Th e R h y t h ms of Tonal Music, 29). one ot h er int erpret at ion of t h e root : obscurit y of ident it y t akes away f rom wh at acoust ic h arsh ness would emph asize. Th e immense complicat ions of weigh t ing dissonance (f or in- st ance, are sevent h s above t h e root "real" dissonances?), h ave persuaded me t o leave t h is issue t o be ref lect ed in t h e comparison of root ch anges wit h t h e ph enomenal dimension, wh ere all cont ribut ing pit ch es get t h eir due. A number of not at ions could indicat e root ch ange densit y , including numerals. A musical st af f calibrat es t h e bar graph I use in my analy ses. If only one voice est ablish es a new h armony , no mark is made. If t wo voices est ablish a new h armony , a bar ext ends up t o t h e second line; f or t h ree voices, it ext ends t o t h e t h ird line, and so on. Th e pat t ern of t h e bars sh ows t h e t rends at a glance. Th is view is generally t h e most inf ormat ive, f or one can t oo easily read t oo much int o local densit y f luct uat ions. R oot ch anges by t h e int erval of a second obviously h ave a h igh er probabilit y of relat ively h igh densit y because t h ere are no common t ones. St ill, sus- pensions can alway s reduce root ch ange densit y , and t h ere are plent y of inst ances wh ere more voices move t h an would be absolut ely necessary t o ef f ect t h e new root . Th e endings of Bach f ugues come t o mind, wh ere t h e composer of t en abandons a smoot h er t ext ure of sust ained t ones f or a climax of dense h armonic rh y t h m. Example 7 sh ows t h ree such end- ings. Densit y , it seems, is a means of amplif y ing or at t enuat ing t h e h armonic ef f ect s produced by root mot ions and t h eir f unct ions, a kind of volume cont rol on h armonic rh y t h m t h at act s in t wo explicit dimensions wit h in it self : st rengt h , sh own by t h e h eigh t of t h e bars, and pace, sh own by t h e placement of t h e bars wit h in t h e normal rh y t h mic not at ional graph . In t h e densit y graph of t h e Prelude (Example 1), wh at st ands out is t h e consist ency of t h e st rengt h of t h e root ch anges-all made by t wo voices save t h e alt o ent rance in m. 1-and t h e accelerat ion in m. 3. It is curious t h at in t h e "Et in unam sanct am" rit ornello (Example 4) t h e densit y of t h e This content downloaded from 193.144.2.35 on Sun, 5 Oct 2014 03:17:03 AM All use subject to JSTOR Terms and Conditions 64 Music Th eory Spect rum Example 7. J. S. Bach , conclusions t o f ugues f rom Th e Well-Tempered Clavier, Book I: Ct major; D minor; D# minor ,,.. r 19 I d "I ' IR ~ ~ ~ ~ I a J .n.. _ - 6 s 1Z n n Jb ; Text ure JJJ JT | JJ J J J Ph enomenal Bass pit ch J c It G CJ R oot i J IJ n J Gt Ci dl GO C JTJJJJJI J. JJT *r JInJ J. IJ. :J7J J J. d De-D e- D d7n f J7gt IJ dl f -gl a?gl b -cx-d I A . I I I .11 I I I I 1.111 I I I I progression declines radically as t h e cadence is approach ed. Th is comes of Bach 's reduct ion of t h e t wo independent oboes d'amore int o a single voice. It is as if h e want s t o undermine t h is cadence even as h e is making it . Th is passage recurs six more t imes in t h e aria, and t h e t reat ment of t h e oboes is t h e same in all but t wo, in wh ich t h ey are rest ored t o indepen- dence, doubling t h e densit y of t h e cadent ial approach : one is a cadence in E major (mm. 58-60) at t h e aria's midpoint and t h e ot h er is t h e last cadence of t h e piece. Surely t h e h igh ligh t ing ef f ect of h armonic densit y cont ribut es t o t h e h igh -level art iculat ion of t h e aria's f orm. 6. HAR MONIC FUNCTION Insof ar as f unct ional h armony , perh aps more t h an any ot h er h armonic aspect , art iculat es t h e ebb and f low of musical t ension t h at is essent ial t o any musical sy nt ax, t h e dimension of h armonic f unct ion most f ait h f ully addresses t h e sh ape of a ph rase in t radit ional musical languages. For Pist on, f unct ion is t ied t o root ident it y and it s movement . Th at view, h owever, requires int erpret ing t h e sy nt act ic roles of t h e root s wh ile simult aneously observing t h eir movement . But f unct ion is not sy nony mous wit h root ident it y : it of t en proceeds at a dif f erent Densit y i This content downloaded from 193.144.2.35 on Sun, 5 Oct 2014 03:17:03 AM All use subject to JSTOR Terms and Conditions Dimensions of Harmonic R h y t h m 65 pace and deserves a separat e analy sis t h at can t h en be com- pared wit h an analy sis of pure root movement . Th e dist inct ion is clearly display ed in t h e f ourt h measure of t h e Bach Prelude (Example 1). R oot movement s on bot h lower and h igh er levels (R oot 1 and R oot 2 in t h e example) are evenly paced, ref lect ing t h e inexorable drive of Bach 's count erpoint . But t h e movement of h armonic f unct ions, de- pict ed at t h e bot t om of t h e ch art , is much more varied, re- vealing an exquisit e t ension in Bach 's ch oice of h armonies t h at gives t h e impression of a composit ional f reedom bound up in t h e clearest , most orderly pat t erns. Th e f unct ions are Hugo R iemann's: t h e st able t onic (I); t h e mobile, t ense dom- inant (V); and t h e mediat ing subdominant (IV). Th ey are assigned by considering t h e t radit ional associat ions of root wit h R iemann's f unct ions in conjunct ion wit h cont ext ual f ac- t ors such as ch ord disposit ion, met ric st rengt h , durat ion, po- sit ion in t h e ph rase, and so on. Th e analy sis of levels of h armonic f unct ion, wh ile anal- ogous t o t h at of root movement , is signif icant ly more com- plex. An analogy wit h sy nt act ic embedding in language is appropriat e.18 Wh en we claim t h at t h e second h armony of t h e A major Prelude h as a dominant f unct ion on one level and is cont ained in t h e t onic f unct ion on anot h er, we t read close t o cont radict ion; y et t h at is wh at h appens in speech percep- t ion, and f luent list eners can easily accept a dualit y of f unct ion f or t h at sound in Bach 's f irst measure. In addit ion t o em- bedding, h owever, t h ere is a dist inct h ierarch ical relat ionsh ip called parent h esis, as suggest ed in mm. 6 and 7 of t h e "Et in unam sanct am" rit ornello, h igh ligh t ed wit h vert ical bracket s in Example 4 (t h e h orizont al bracket s sh ow discont inuous embedded f unct ions). In parent h esis t h e progress of h ar- monic sy nt ax is act ually suspended wh ile anot h er unrelat ed '8Such analogies of sy nt act ic embedding h ave been described by Allan Keiler, "Th e Sy nt ax of Prolongat ion I," In Th eory Only 3/5 (1977): 3-27, and Lerdah l and Jackendof f , Generat ive Th eory . ph rase is insert ed. Wh en t h e parent h esis is f inish ed, t h e su- perordinat e sy nt ax resumes. Th e connect ions bet ween t h e f unct ions of t h e main ph rase--IV in m. 6, V and IV in m. 7-are real and direct ; wh at is insert ed is int egral only t o it self . Th at is wh y t h e rh y t h mic values do not f ill t h e measure. Th e rh y t h mic ef f ect is pat ent ly audible, int riguing, and com- plex, creat ing a special kind of t ension in t h e rit ornello's cent er. Because f unct ions operat e and t ake t h eir very def init ions wit h in a specif ic t onal cent er, local t onicizat ions of h armonies and h igh er-level modulat ions not only creat e embedded lev- els of h armonic f unct ion, t h ey cause t h e percept ion of f unct ions of ch ords gone past t o be revised. Th e opening of Vivaldi's "Wint er" Concert o (Example 3) provides a con- cent rat ed sampling of t h e analy t ical problems t h at may arise. Th e long F in t h e bass is a t onic pedal f or most of it s lif e, but at some point it must t urn int o a subdominant , since t h e f irst cadence is in C minor. Wit h in t h at cadent ial approach a number of ot h er t onicizat ions present t h e list ener wit h sev- eral occasions of ret rospect ive list ening. Th e B, minor t riad of m. 7, at t h e moment of h earing, count s as a weak sub- dominant , but wh en t h e bass F is reint erpret ed as t h e root of a V7/iv in t h e next measure, t h e B b ch ord t urns out t o h ave been a i6 in disguise; a new t onal orient at ion arises. Similar reorient at ions pile up rapidly in t h e next t wo bars: t h e h ar- mony of m. 9, a dominant of C, f orces reh earing m. 8 as a subdominant ; t h e t ent h bar implicat es G as a t onal cent er, but h as a much weaker f unct ional connect ion wit h t h e G dominant sevent h just h eard. Th ere is lit t le doubt t h at t h e working memory of h uman cognit ion can manage such ret rospect ive reint erpret at ions of f unct ional sy nt ax,19 but t h at does not mean t h at it can alway s '9Est imat es of t h e t ime unint erpret ed sensory inf ormat ion remains in working memory run f rom t wo t o f our seconds, plent y of t ime f or review af t er t h e f act . R est rospect ive list ening is an essent ial component of Eugene This content downloaded from 193.144.2.35 on Sun, 5 Oct 2014 03:17:03 AM All use subject to JSTOR Terms and Conditions 66 Music Th eory Spect rum do so complet ely wit h regard t o h armonic rh y t h m. All t h e h armonic rh y t h ms discussed so f ar t ake t h eir measure f rom at t ack point s locat ed precisely in t h e st ream of an underly ing pulse of const ant durat ion unit s, wh ich in Baroque music is usually explicit in t h e rh y t h m of t h e t ext ure. So wh en t h e eigh t h measure of Vivaldi's concert o f orces a reconsiderat ion of t h e f unct ion of m. 7, we need not only t o reimagine t h e qualit y of t h at f unct ion, but also t h e locat ion of it s onset precisely eigh t pulses bef ore. Th at is likely t oo great a st ret ch f or working memory .20 Wit h more rapid root mot ion (see t h e last ch ords of "Et in unam sanct am," Example 4, m.5), such precision is possible, but ot h erwise it seems best t o represent t h e ret rospect ive f unct ion in bracket s wit h out assigning a rh y t h mic value. Inst ead, t h e rh y t h mic ef f ect of reint erpret ing is concent rat ed on t h e moment wh ere t h e reint erpret at ion is f orced, t h e downbeat of m. 8. Embedded levels of act ivit y drop in and out wit h out re- spect t o any predet ermined number of levels. Th e rh y t h mic ef f ect of adding or subst ract ing a level of h armonic f unct ion resides in t h e int ensit y of rh y t h mic art iculat ion, f elt as t h e h eigh t ened cognit ive act ivit y required f or t h e processing and appreciat ion of t h e h armonic sy nt ax.21 In t h e Vivaldi t h e in- Narmour, Th e Analy sis and Cognit ion of Basic Melodic St ruct ures: t h e Implicat ion-R ealizat ion Model (Ch icago: Universit y of Ch icago Press, 1990), 203-204. 20Th e limit f or t h e number of it ems in a cognit ive "ch unk" of inf ormat ion h as been est imat ed f rom f our t o seven. For a summary of t h e relevant st udies, see Joseph P. Swain, "Th e Need f or Limit s in Hierarch ical Th eories of Mu- sic," Music Percept ion 4/1 (1986): 121-147. 211 t ake t h is cognit ive ef f ort in t h e int erpret at ion of embedded h armonic f unct ions t o be analogous t o t h at required f or t h e int erpret at ion of embedded sy nt ax in speech , on wh ich psy ch olinguist s are generally agreed. See Trevor A. Harley , Th e Psy ch ology of Language (East Sussex: Erlbaum, Tay lor & Francis, 1995), 163-165. It sh ould also be clear by now t h at I do not agree wit h Lerdah l and Jackendof f 's assumpt ion t h at "t h e ear seeks, insof ar as possible, a regular underly ing h armonic rh y t h m" (Generat ive Th eory , 130). creasing complexit y of t h e f unct ional graph depict s t h e st eadily rising t ension, support ed by great er h armonic den- sit ies and f ast er root movement s, t h at prepares t h e explosive ent rance of t h e violin soloist . METER Wh at of h armonic rh y t h m's relat ionsh ip t o met er? "Har- monic ch ange is t h e single most powerf ul met er-producing f act or" writ es Joel Lest er unequivocally , and on t h is t h eorist s are generally agreed.22 Th e very act of h armonic ch ange can so easily accent beat s t o creat e t h e requisit e h igh er-level pat - t ern. Th e A Major Prelude t h us est ablish es it s lower-level t riple met er unmist akably f rom t h e out set in it s coincidence of bass pit ch , root movement , and f unct ion. Funct ion it self of t en indicat es accent ed beat s wit h t onic f unct ion and un- accent ed beat s wit h dominant -convent ional associat ions t h at are nonet h eless ext remely powerf ul and ef f icient .23 So t h e inf luence of h armonic rh y t h m on met er is unde- niable. But t h e not ion of a "h armonic met er" t h at is ent irely dist inct f rom t h e met ric pat t ern of a passage will not f ly . Met er requires t wo element s: beat s of consist ent lengt h and Harmonic rh y t h m is not a met er, and it s f lexibilit y is it s great power t o art iculat e. 22Lest er, Th e R h y t h ms of Tonal Music, 66. Lerdah l and Jackendof f in- dicat e t h at h armonic rh y t h m is t h e st rongest f act or in t h eir model of met rical st ruct ure (Generat ive Th eory , 84-85). For a compreh ensive summary of eigh t eent h - and ninet eent h -cent ury t h eorist s' views about t h e relat ionsh ip of h armony and met er, see William Earl Caplin, "Th eories of Harmonic-Met ric R elat ionsh ips f rom R ameau t o R iemann" (Ph .D. diss., Universit y of Ch i- cago, 1981). 23Wallace Berry post ulat es t h at wh at h e calls t onal f unct ion "is in and of it self met rically neut ral," but h e of f ers a very implausible example (St ruct ural Funct ions, 330). To t h e cont rary , wh en I h ave play ed f or st udent s t h e be- ginnings of pieces such as t h e March f rom Fidelio, wh ich begins wit h a dom- inant bass on t h e not at ed downbeat and maint ains t h is pat t ern f or t wo bars, t h e st udent s invariably sense t h at t h e piece begins wit h an upbeat . This content downloaded from 193.144.2.35 on Sun, 5 Oct 2014 03:17:03 AM All use subject to JSTOR Terms and Conditions Dimensions of Harmonic R h y t h m 67 a regular pat t ern of accent ing t h ose beat s.24 Harmony sup- plies neit h er one. Since t h e wh ole mat t er of h armonic rh y t h m is grounded in h armonic ch anges, a "h armonic met er" would demand t h at h armony ch ange at some consist ent t ime durat ions-t h e h armonic beat -ot h er t h an t h ose marked by t h e at t ack point s of t h e melodies t h at const it ut e t h e t ext ure. Th is is impossible, since it is t h e melodic ch anges t h at creat e t h e h armonic ones. Indeed, t h e f ascinat ion wit h h armonic rh y t h m is in t h e variet y of h armonic lengt h s, not t h eir con- sist ency . If t h ere is no "h armonic beat " ot h er t h an t h e one coincident wit h t h e beat of t h e t ext ure, t h en t h ere can be no regular accent ed beat , except one coincident wit h t h e t ext ure. Is it possible t h at a "h armonic met er" could t h en exist out of ph ase wit h t h e normal t ext ural met er? Th is h appens in some sense in Bach 's rit ornello (Example 4). Once a met er is est ablish ed f irmly and becomes almost self -maint aining ac- cording t o t h e Gest alt principle of cont inuat ion, h armonic f unct ions can act against it (mm. 6-7).25 But even h ere, t h e displaced accent s t h at creat e t h is parent h esis arise as much f rom t h e melodic grouping in t h e oboes d'amore (B-C# - D# -E) as f rom t h e f unct ional pat t ern of t h e h armony . It is 24Th ere is h igh consensus on t h is point . Lest er writ es, "Two separat e component s are t h us necessary f or t h e exist ence of a met er: a st ream of beat s or pulses, and an organizat ion of t h ose beat s or pulses int o accent ed and unaccent ed ones" (Th e R h y t h ms of Tonal Music, 45). Ot h er t h eorist s pref er t o place t h e t wo element s in a h ierarch ical sy st em. Yest on cont ends t h at "met er is an out growt h of t h e int eract ion of t wo levels-t wo dif f erent ly rat ed st rat a, t h e f ast er of wh ich provides t h e element s and t h e slower of wh ich groups t h em" (St rat if icat ion of Musical R h y t h m, 66). See also Lerdah l and Jackendof f 's model in Generat ive Th eory , ch apt ers 2 and 4. 25Lest er's ch aract erizat ion of t h is cont inuit y is apropos h ere: "Once a met er h as been est ablish ed, it s power is so great t h at it can overrule t h e act ual occurrence of a h armonic ch ange and h ave us int erpret t h at ch ange as eit h er an ant icipat ion or a delay ... of t h e 'act ual' point of ch ange" (Th e R h y t h ms of Tonal Music, 79). A f ormal approach t o such int eract ions is evident in Lerdah l and Jackendof f 's rule of Met rical St abilit y (Generat ive Th eory , 164-165). viable but f or a brief t ime, only as long as t h e list ener can maint ain t h e prevailing met er in working memory , h ere aided by Bach 's bass accent s. Th e ef f ect is a special kind of sy n- copat ion, a sust ained met ric dist urbance. If ext ended t oo f ar, such h armonic t ricks simply est ablish a new met er alt oget h er. I cannot see much pract icalit y in t h eorizing about a "h ar- monic met er" independent of t h e rh y t h mic one. Inst ead, we sh ould pay at t ent ion t o t h e ef f ect s of h armonic mot ion int eract ing wit h t h e prevailing met er. At lower st ruc- t ural levels such int eract ions can creat e rat h er acut e and par- t icular t ensions. At t h e h igh er levels t h e sensat ion is more subt le, as wh en Bach unexpect edly moves t o t h e subdominant f unct ion on t h e weak beat in t h e t h ird measure of h is Prelude and t h en t ies t h e h armony across t h e barline, creat ing a del- icat e rh y t h mic-met ric t ension t h at powers t h e f ast er mot ions and sequences t o f ollow. INTER PR ETATION OF HAR MONIC R HYTHM Wh at insigh t s can an analy sis of t h e dimensions of h ar- monic rh y t h m provide? To begin wit h , t h e easy comparison and t h e composit e pict ure present ed by several or all of t h e dimensions at once of f ers a new perspect ive on musical t ex- t ure.26 In t h e case of "Et in unam sanct am" (Example 4), t h e pict ure f ills out Wilh elm Fisch er's t ripart it e rit ornello model of Vordersat z, Fort spinnung, and Epilog.27 We see at a glance t h e st abilit y of t h e int roduct ory f irst f our measures, t h e t ra- dit ional h armonic f unct ions allied st rongly wit h a h igh -level duple met er, h igh densit y reinf orcing t h e clarit y of t h ose 26I t ake "t ext ure" t o mean t h e qualit y of t h e combinat ion of simult a- neously sounded melodies and t h eir relat ions. Like h armony , it h as many aspect s. 27Fisch er proposed t h is model in 1915. See Joh n But t , Bach : Mass in B Minor (Cambridge: Cambridge Universit y Press, 1991), 60. This content downloaded from 193.144.2.35 on Sun, 5 Oct 2014 03:17:03 AM All use subject to JSTOR Terms and Conditions 68 Music Th eory Spect rum ch anges, t h e nearly perf ect congruence of t h e various di- mensions. Even t h e saucy beginning in t h e cont inuo f inds an explanat ion: unh armonized, it is t h e merest suggest ion of low-level progressions t h at come alive lat er on. Th en f ollows t h e relat ively acut e t ension of t h e parent h esis, wit h it s sud- denly f ast er root movement , and it s release int o measures eigh t and nine, wh en bass pit ch rh y t h m becomes congruent wit h t h e t ext ure/ph enomenal f or t h e f irst t ime. Th e Epilog h as y et anot h er ch aract er, wit h st ill a new bass pit ch rh y t h m and a curiously low densit y wit h no st rong t onic f unct ions unt il t h e cadence. Th e pict ure reveals t h e magnif icent variet y of Bach 's rh y t h m, a variet y most sensible but h idden and unif ied under t h e cover of an absolut ely consist ent t ext ural and ph e- nomenal h armonic rh y t h m. It conf irms not just Fisch er's ar- t iculat ions of t h e rit ornello, but subt le means of t ransit ion bet ween t h em. Not ice h ow t h e embedded f unct ions compli- cat e t h e t ext ure at t h e end of t h e f irst ph rase, bef ore t h e parent h esis, and h ow t h e st eady second-level root mot ion, so clearly audible, t ies t h e suspension passage af t er t h e paren- t h esis int o t h e Epilog wit h out int errupt ion. Th is new and det ailed analy t ical met h od of f ered h ere sh ould be especially powerf ul in t h e analy sis of t h ose t ext ures t h at ent ail a soph ist icat ed h armonic sy nt ax. A musical sy nt ax it self ent ails t h e cont rol of musical t ension,28 and h erein lies anot h er kind of insigh t provided by t h e analy sis of h armonic rh y t h m. I propose t wo working assumpt ions: 1) t h at rh y t h mic t ension of one kind rises wit h increase in speed of ch anges; and 2) t h at rh y t h mic t ension of anot h er kind rises wit h t h e decrease of congruence - coincidence of mot ions - among t h e dimensions.29 Bot h of t h ese common-sense assumpt ions de- rive f rom t h e simple f act t h at at t h e end of a composit ion, 28See Swain, Musical Languages, ch apt er 2. 29In St ruct ural Funct ions of Music (1, 86), Wallace Berry , wh ose principal concern is t h e descript ion of t h e course of "int ensit y " in music, essent ially makes t h e same assumpt ions. wh en t ension is resolved, t h e mot ion in all dimensions ceases and t h eref ore congruence among t h em is at a maximum. Th e congruence principle may be t h e more cont roversial of t h e t wo, but it explains wh y t h e t ensions in t h e Example 6 ch orale seem so simple and placid. Th e correspondence among t h e dimensions approach es complet e unif ormit y ; only t h e h igh er-level f unct ion sh ows some variance wit h wh at h ap- pens above, and even so f alls int o line by t h e end of every ph rase. Almost all t h e rh y t h mic t ension derives f rom simple speed, f ast er mot ion in t h e middle of t h e ph rase, and it is perh aps not coincident al t h at t h e only passing mot ion is re- served f or t h e end. Th is is an ext reme case, of course. Freer composit ions will sh ow much more variet y , requiring a more relat ive and con- t ext ual underst anding of speed and congruence. In h is "Win- t er" Concert o, Vivaldi creat es a disjunct ion by t h e simplest means possible: h ammering a ch ord repeat edly wit h out let - t ing any t h ing move, so t h at t h e t ext ural rh y t h m is eigh t t imes f ast er t h an any t h ing else. Th is calls t o mind Leonard Mey er's aest h et ic rule about expect ing ch ange; Vivaldi h ere push es t h e limit s of our t olerance. Despit e t h e ext remit y of t h e dis- crepancy , t h e eigh t h -not e mot or is not t ruly a part of t h e h armonic rh y t h m it self (alt h ough t h e incongruous mat ch be- t ween t h e ph enomenal rh y t h m and every t h ing below is surely part of t h e ef f ect ). Th e rise in t ension h ere is powered by increasing speed in t h e root movement and most of all by t h e mult iple int erpret at ions of f unct ion t h at list eners must per- f orm, a t ension deriving direct ly f rom t h e st rain on our f ac- ult ies. Vivaldi f ash ions t h is climax by beginning wit h a signif icant discrepancy bet ween t h e t ext ural rh y t h m and t h e ot h er h ar- monic dimensions. Such a t ension is magnif ied in Bach wh en t h e more immediat e dimensions of ph enomenal h armonic rh y t h m and bass pit ch move at t h e same pace as t h e t ext ure, f or t h en it is not simply it erat ed pit ch at t acks against all of h armony , but rat h er t wo massive, incongruent st reams of This content downloaded from 193.144.2.35 on Sun, 5 Oct 2014 03:17:03 AM All use subject to JSTOR Terms and Conditions Dimensions of Harmonic R h y t h m 69 h armonic mot ion, one f ast and one slow, pulling against one anot h er. Th at is wh y it is so import ant t h at Bach nearly al- way s unif ies h is surf ace mot or rh y t h m wit h const ant ph e- nomenal ch ord ch anges, and f requent ly wit h a st eady bass as well. Th e new and signif icant t ension arising f rom t h is t ext ure is h is personal st amp on t h e Baroque musical language. Th ere is a kind of paradox built int o t h e musical language of a Baroque composer, if t h ese assumpt ions about rh y t h mic t ension are righ t t o any ext ent . Th e basic pulse of t h e mot or rh y t h m will nearly alway s move f ast er t h an ot h er h armonic dimensions, part icularly in a cont rapunt al t ext ure. As t h e end is approach ed, t h en, t h e composer can only bring t h e di- mensions of h armonic rh y t h m int o congruence by speeding t h em up, t h us convert ing one kind of rh y t h mic t ension int o anot h er, or by composing a plagal cadence, wh ich is essen- t ially a h armonic-rh y t h mic cush ion t h at absorbs t h e t ension of dif f erent iat ed count erpoint wh ile t h e t onic is maint ained and t h e remaining voices unif y . We h ave seen h ow Bach can accelerat e and unif y t h e mot ions in all h is voices at t h e end of some of h is f ugues. He does somet h ing quit e similar at t h e end of h is "Et in unam sanct am" aria, sh own in Example 8. It is of t en at t h is point in h is arias t h at Bach someh ow sculpt s t h e cont inuit y of h is surf ace rh y t h m int o a moment of special f inalit y t h at proclaims t h at t h e rit ornello about t o f ollow is t h e last . Wh en h is singer int roduces a diminish ed sevent h arpeggio as t h e unison oboes f inish t h eir f ormula, t h e ex- t ension creat es anot h er cadence made wit h out t h eir h elp t h at is more st riking and conclusive t h an any t h ing t h ey h ave play ed in t h is piece. Th e ch art sh ows t h e new element s: f irst , t h e diminish ed sevent h builds a prolonged dominant f unc- t ion, t h e f irst such f unct ion in t h e aria, t o amplif y t h e last cadence; more import ant , suddenly all t h e dimensions move at about t h e same speed. One kind of t ension-discrepancy of mot ions-convert s t o anot h er kind-sh eer speed-t h at is more suscept ible t o pinpoint cont rol. It is Bach 's part -writ ing t h at accomplish es t h is t ransf ormat ion, as h is bass sounds wit h powerf ul f inalit y t h e root of every ch ord wit h maximum den- sit y in rapid f ire. Th e rit ornello t h at f ollows is a welcome conf irmat ion of t h is resolut ion, not just a bookend t o t h e aria. Such variat ion among t h e dimensions is put t o work wit h marvelous result s in t h e A major Prelude. In t h e f irst measure (see Example 1) t h ere is a h igh degree of congruit y among t h e dimensions; only t h e most abst ract levels of root mot ion and h armonic f unct ion are laggard, and t h ey are barely es- t ablish ed h ere.30 Indeed, t h ere is no ch ange in t h em as y et , so t h ey can h ardly be said t o const it ut e a rh y t h mic pat t ern at all. But t h e second measure t ransf orms t h is easy and uni- f orm swing wit h out t h e sligh t est sign of ef f ort or violence t o t h e t ext ure. Every dimension except t h e mot oric t ext ure/ ph enomenal is slower; even t h e bass pit ch adopt s a t h robbing sy ncopat ion in place of it s t roch ees. Th is brings t h e pedal t ones int o st ruct ural prominence, and makes t h e more ab- st ract mot ions much more palpable. Wh en t h e t h ird measure ret urns t o t h e f ast er mot ion-compare t h e densit ies-and even accelerat es it in t h e t h ird beat and drives it wit h t h e bass, t h ere is t h at ch aract erist ic st rain t o go f orward, an elast ic t ension t h at is released in t h e f ourt h bar, ant icipat ed beau- t if ully by t h e ch ange of h armony on t h e upbeat . And t h e most signif icant det ail of t h is st raining t ension is t h at long, long, deceiving t onic pedal in t h e alt o, because it st ubbornly pro- claims t h e discrepancy in t ext ure among t h e voices t h at is f lesh ed out in h armonic act ivit y graph ed below.-Th e f ourt h and f if t h measures convert t h is st raining t ension int o mobile t ension, a release and y et a propulsion at t h e same t ime. Th e dimensions now sh ow a gradat ion of durat ions-part icularly import ant is t h e mediat ing root movement of dot t ed quart ers, marking t h e h igh -level beat s as ot h er dimensions mark t h e 300ne f ut ure ref inement of t h is met h od is t h e addit ion of a means of indicat ing t h e relat ive salience of t h e various dimensions. Were t h e voice- leading not so persuasive h ere, one migh t ch oose not t o indicat e any rh y t h ms at t h is level f or t h is measure. This content downloaded from 193.144.2.35 on Sun, 5 Oct 2014 03:17:03 AM All use subject to JSTOR Terms and Conditions 70 Music Th eory Spect rum Example 8. J. S. Bach , Mass in B Minor, "Et in unam sanct am," mm. 126-132 126 j[q - - si am,u-nah n san " ct am ca- t h o-li-cam et a- po - st o -i - cam- ec-cle si - am. l a r ^ r ^ rl _ I _ r 1r . _ ~i I I I Ir I1 . i.^ L Text ure Ph enom. Bass J 7 J 7 IJ R oot 1 JU T J I J. Dg# -A E D f # 7 J 7 J 7 J 7 I mJ .7 3|,7 ,/ -J7 "J7 | ,T J. I - 7 I- JJ. d i# - A 7,D IJ A h A Dg# - f # A E A d# - B d# -E A E f # Db E A E Densit y R oo2 J. J. . J. DJ. J. I J. D E f # A D E d# - IJ . J. I|J7- TIJ. B E AE f # Db E A Funct ion iA:IVV I V I IA:IVV I V IV I I Imml I . J. IIVV IV I I V I :V J li; J iJ)J. 11A:V IV I IV V I J. J. I. IJ. A: IV V IV J. I J. IJ. - IVIVJ VI I V IV V I m [ ' IJJJJT3-'ml-T m ._ _ | m 'mJ7"l3|7l"T| | This content downloaded from 193.144.2.35 on Sun, 5 Oct 2014 03:17:03 AM All use subject to JSTOR Terms and Conditions Dimensions of Harmonic R h y t h m 71 lower. Such a gradat ion allows t errif ic variet y wit h out cre- at ing t h e incongruous mot ions of t h e previous measure. So many dif f erent mot ions cannot creat e so explicit a cont rast . Bach 's pedal t ones compared t o Vivaldi's h ammering (Ex- ample 3) suggest an assy mmet rical f eat ure of t h e congruence h y pot h esis: wh en surf ace event s are signif icant ly f ast er t h an h armonic rh y t h m, as in t h e "Wint er" concert o, t h e result ing incongruence creat es an almost placid, even t ension, but wh en t h e reverse case is t rue, as in Bach 's pedals, t h e re- sult ing t ension is dy namic, rapidly becoming more acut e wit h each passing beat . In part , t h is is because Bach 's idea is a cont radict ion of common experience. Wh en h armony moves, so must t h e melodies in t h e t ext ure, because it is t h e ch anges in pit ch collect ions t h at creat e new h armonies. How can me- lodic mot ion be f rozen wh en t h e h armony is driving? Anot h er part of t h is sensat ion is owing t o our int uit ive percept ion of t ech nique, an essent ial aspect of our appreci- at ion of musical sy nt ax.31 Vivaldi's idea is pat ent ly ef f ect ive, t h ough it t akes no great skill t o reit erat e a ch ord. But look again at t h e t en beat s of measures t h ree, f our, and f ive in t h e Prelude, wh ere Bach not only aligns t h e t ext ural rh y t h m wit h t h e ph enomenal in unrelent ing eigh t h -not e beat s, but also t h e root mot ion (R oot 1). To ach ieve t h ese const ant mot ions, Bach must insure t h at t h e voice-leading is sensible, t h at t h e result ing vert icalit ies y ield suf f icient pit ch inf ormat ion t o sound a clear t riad, and f urt h ermore, t h at t h e f unct ions of t h ose t riads occur in t h e righ t place wit h respect t o t h e met er. 31See Swain, Musical Languages, Ch apt er 2. Th is coordinat ion, as any one wh o h as t ried it knows, de- mands t h e h igh est skills of composit ional art . (Compare again t o Vivaldi, wh ere voice-leading problems in t h e viola creat e inconsist encies in t h e ph enomenal h armonic rh y t h m.) To command in addit ion t o all t h ese t h e h armonies of t h e ot h er dimensions and t h e pedal t ones seems lit t le sh ort of mirac- ulous. Th e t ension is aest h et ic: we h old our breat h , wait ing t o h ear wh et h er h e can really pull it of f . Th at t onic pedal does not f igure in t h e graph s, but it cer- t ainly f igures in our appreciat ion of t h e musical t ension of t h at moment . We are t h us reminded t h at , as rich as t h e pict ure of h armonic rh y t h m may be, it is insuf f icient by it self . As t h e sense of h armonic rh y t h m arises f rom and engages ot h er as- pect s of t h at experience, so does any evaluat ion of h armonic rh y t h m y ield it s best insigh t s wh en it engages many way s of crit ical list ening. ABSTR ACT Alt h ough undeniably a most signif icant aspect of musical t ext ure, h armonic rh y t h m h as not been precisely concept ualized and ana- ly zed since Walt er Pist on f irst def ined it . Th e new analy t ical t ech - nique present ed h ere at t acks t h e cent ral problem-ident if y ing t h e proper pit ch simult aneit ies wh ose durat ion pat t erns are at issue-by dissect ing t h e various propert ies of h armony int o separat e compar- at ive analy ses: t ext ural rh y t h m, ph enomenal h armonic rh y t h m, bass pit ch , root /qualit y , densit y , and f unct ion. Th e result ing analy sis pro- vides a new, det ailed view of musical t ext ure and, in conjunct ion wit h t wo h y pot h eses, a means t o underst and t h e dy namic t ension of h ar- monic rh y t h m. This content downloaded from 193.144.2.35 on Sun, 5 Oct 2014 03:17:03 AM All use subject to JSTOR Terms and Conditions