Compositional Devices in Steve Reich's Octet




Brent Heisinger





                     In the spring of 1982, New Sounds San Jose featured a Saturday evening program of music by Steve Reich with the composer in attendance.  Rehearsals of Octet with San Francisco musicians, many of whom had played the work under John Adams, and San Jose State University faculty were scheduled on a Friday.  The day before, following a rather urgent plea from organizers, I agreed to conduct the piece[1] under the condition that I meet with the composer to get from him an overview of the work and insights into potential rehearsal problems.  That meeting and subsequent concentrated study of the score and ECM recording led to a musical "internalizing" that was refreshingly unique to me, a type of experience that I had not encountered with other pieces.  At once, Octet felt natural, accessible, stylistically consistent and well-crafted, and with each playing, new relationships revealed themselves.  The following observations are the result of my interest in knowing more specifically how this was achieved by Reich.


                    Written for two pianos, string quartet, and two woodwind parts which call for Bb clarinets alternating with bass clarinet, flute and piccolo, Octet is based on the repetition of various ideas of different lengths.[2]  It is in five at a fast tempo ( q = c. 184) and projects two constant levels of pulse, the eighth-note and quartet-note.  Contrasted with these faster pulses are very slow changing string parts.  Though Reich changes key in the first three sections, the work is essentially pandiatonic with each section projecting a somewhat obscure tonic and respective mode:  Section I - C# Dorian:  II - D# Aeolian:  III - E flat Dorian:  IV - Db Major:  V - A flat Mixolydian.[3]






                    The work is in five sections clearly marked by Reich in the score with Roman numerals.[4]  Well into Section III, however, a sudden shift in the pianos to a wider range produces perhaps the most distinct change in the work.  This two-part division of Section III is labeled here as Sub-sections III(a) and IIIb).  Each of Reich's five sections is a little over three minutes in length.


                    Form and texture evolve from three strands, two of which are present throughout:  a propelling canon-ostinato in the pianos, and a drone effect in the strings.  Various melodic increments enter at different times creating a third strand either distinct from the rest of the texture or doubling parts completely or partially.  Found in the woodwinds and strings, they appear in different registers, are of different lengths, and vary as to their tunefulness.  As can be observed in the general diagram of Example 1, Sections I and III are related in that they contain a drone-canon  in the violins and similar melodic increment  entrances.  Sections II and IV are alike in that they include a drone-ostinato  in the strings and similar melodic increments   entrances.  Section V combines elements of both groups.




                                             Example 1: Formal structure of Reich's Octet


                    In keeping with the concept of subtle change, Reich deliberately avoids distinct, abrupt changes of sections.  Instrumental parts overlap (note Example 1), new keys are closely related and carefully anticipated, and changing lines  fade in and out.  The most obvious gesture signaling each new section is the change of range and register, and that which marks each new section is the entrance of a part in the abandoned register.  Typical of this "modulating" process is the transition between Sections I and II parts of which are shown in Examples 2 -a/b/c.


                    Thirty measures before Section II, the bass melodic increment in the cello begins its ten-measure fade-out.  This is followed by a six-measure fade-out of the left hand parts of both pianos (Example 2-a) thus narrowing the overall range from C#2 - D#6 to G#3 - D#6[5] and leaving only the flute melody, the right hand parts of both pianos, and the two-violin double stops.  The upper notes of the right hand piano parts are then shifted unnoticeably to the left hand parts while the violins begin fading out (Example 2-b).  At this point in the music, no E's are present.  So when the right hand part of Piano 1 introduces a new transposition beginning on D#4 and including E3's (Example 2-c), no uncharacteristic dissonance is created and the modulation of key from five to six sharps is achieved without fanfare.  This narrowing of the range upward clears the lower register for the lower parts of the canon-ostinato in Sections III and V, and the lower parts of the string drone-ostinato in Sections II, IV and V.  Assisting the gradualness of this transformation is the flute melody which overlaps twenty-eight measures into the new section.


                         Example 2a: Reich's Octet mm. 127-128                          Example 2b: Reich's Octet mm. 137-138.

                                                                  Example 2c: Octet mm. 143-144.






                    The most generative of the three textural strands is the two piano canon-ostinato  which produces the unchanging quarter-note and eighth-note pulses and is the source for most melodic increments.  Each hand of the Piano 1 part serves as the dux  of a two-measure group-circle canon, the comes  parts of which can be found, as shown in Example 3, in the Piano 2 part in the respective hands at the unison.


                                                                          Example 3: Section I, mm. 57-68.


                     The composite eighth-note pulse of the two-piano canon does not change throughout the work, nor does the rhythmic placement of the dux  parts of Piano 1 within the two measures.  However, the comes  parts in Piano 2 shift with each section (Examples 4-a to 4-e) providing subtle variation of pitch, rhythmic and dynamic accents, and it is primarily from these changing parts that Reich's melodic increments find their source.  In addition, the interval between the right hand and left hand parts changes with each section.  Note in Example 3 from Section I, the interval between the first pitch of each hand in the Piano 1 part.  This compound P5 changes to a P5 in Section II (Example 4-a), which is followed by intervals of a compound M6 and compound m3 in section III (Example 4-b, Example 4-c), P4 in Section IV (Example 4-d), and compound m3 in Section V (Example 4-e).  This shifting produces noticeable changes of sonority, especially brightness and fullness. 




                Example 4a: Section II, mm. 171-174                                     Example 4b: Section III(a), mm.343-346


                        Example 4c: Section IIIb, mm. 403-386                                  Example 4d: Section IV, mm. 473-476




                                                                   Example 4e: Section V, mm.637-641


                    When the complete canon is heard, always the same register and range is employed in both pianos, creating a constant underlying homogeneous sound.  And with its angularity (primarily perfect fifths), rhythmic structure, and accents produced by the three triads in each hand of each part, the canon-ostinato projects what Reich refers to as "sub-melodies."  Like a kaleidoscope, these figures change with subsequent shifting of the comes  parts along with changes of register, range, dynamics, transposition and, on occasion, melodic intervals and contour.  This will be illustrated later.





                    Reich contrasts this very active strand with a sustained drone effect in the strings.  This strand is produced in two ways - what is referred to here as a drone-canon  in Sections I and II using double stops in the two violins, and drone-ostinato   in Sections II and IV using (eventually) the complete string quartet.  Section V begins with the overlapping from Section IV of the drone-ostinato which is juxtaposed with the drone-canon.  Both the drone-canon and drone-ostinato are lengthened in Section V from two- to four-measure repeats, the drone-canon then to eight-measure repeats, and both finally to ten-measure repeats with the complete quartet playing double stops to the end of the piece.


                    Always the dux  of the drone-canon is introduced in the first violin part in diminution, two measures in length (Example 5).  This is treated in canon at the unison and third quartet-note by the second violin.  It is not until several repeated measures, however, that through an additive process the relatively non-contrapuntal canon is fully stated.  The comes  (Violin 2) part begins with the last of the three intervals of the dux   (Violin 1) part; then in reverse, the second interval is added in alternation followed by the first until the canon is completed.  Example 5 illustrates how this is accomplished in Section I.  It is not until after this point that the eight-measure elongated version of the drone-canon is heard along with the first melodic-increment.  After four repeats the parts are then extended to ten measures.



                                                                                  Example 5: mm. 1-56


                    The entrance of the comes   part after ten beats and placement of tenuto marks generally enhances the symmetry of two-measure groupings.  This appears to be an important alignment, since with the eight- then ten-measure repeat, the comes   ends up to be three beats short of the dux.  Nevertheless, the drone-canon in its ten-measure form generates the most sustained and thus contrasting textural strand.[6]





                    The drone-ostinato (Sections II, IV, and V) is a four-measure repeated figure which after several hearings takes on its own slower asymmetrical metric identity.[7]  In Sections II and IV it begins with the first violin followed by entrances of the remaining string parts; in Section V it is stated from the outset in the viola and cello parts.  In all cases, the last two measures of the ostinato are introduced first, followed by the addition of the first two measures, not unlike the introduction of the drone-canon.  Delaying a full musical statement is clearly a significant stylistic device in Octet.


                    Each of the three drone-ostinatos differs in pitch and rhythm content.  The alternating of similar two-chord progressions appears in Sections II (B 7 - D#m11, see Example 6) and IV (G flat 9 - Bbm11).  It is at these points in the work that harmonic momentum is particularly felt.  The combination of the drone-canon and drone-ostinato in Section V suggests a third two-chord progression (Ebm11 - Ab9 and later Ab13) which is less pronounced due to the non-chordal nature of the drone-canon.  Example 6 shows how Reich arrives at the full statement of the drone-ostinato in Section II, a procedure used similarly in Section IV.


                    Violin 1 begins by fading in the last two measures of the ostinato, followed by the cello setting up the B 7 - D#m11 progression.  Before the other string parts enter, however, the full length four-measure ostinato is formed, the cello widening the range by inverting its double stops.  This gesture noticably richens the sonority of the passage.  At this point, the drone-ostinato takes on its slower asymmetrical metric quality with the sustained notes giving rise to a 5-4-4-2-2-3 grouping of quarter-note pulses.  This is in conflict with the clear grouping of 5 generated by the composite eighth-note values in the other strands.  Reich highlights this contrasting pattern with a stepwise viola melody that fades in to a forte, then reinforces the harmonic progression with the fading in of the bass clarinet carrying the roots of the two chords (B and D#).  The groupings of the four-measure drone-ostinatos of Section IV (3-2-3-2-3-2-2-3) and Section V (5-5-3-2-2-3) are especially emphasized by the cello part and coincide more with the meter of the piece.



                                                                                        Example 6: mm. 147-286


                    The three drone-ostinatos have in common the metric grouping of the last two measures, 3-2-2-3 (omitting the ties over the bar of mm. 2 and 3 in Section II).  This is the same metric combination as the diminution form of the drone-canons heard in Sections I (Example 5) and III, and makes particularly compatible the juxtaposition of the two "drones" in the first part of Section V (note Example 1).



Melodic Increments


                    The third textural strand is comprised of various melodic increments that can be found in four settings - (1) as a two-measure direct doubling of a canon-ostinato part, (2) as a two-, eight-, or ten-measure melody derived from the canon-ostinato, (3) as a bass doubling of a slower more sustained string part, and (4) as an independent string melody within the drone-ostinato.  Melodic increments in Octet clearly are extensions of the texture.


                    One of the most striking features in this work is the entrance in all sections of a very fragmented woodwind melodic increment that with subsequent repeats becomes increasingly active rhythmically.  This is brought about through the addition of notes that emerge, along with the Piano 2 part, as a comes   line of the canon-ostinato.  Reich varies pitch and rhythmic order and instruments with each appearance, but the procedure remains the same.  The entrance of the bass clarinets in Section III will serve to illustrate.  Example 7  shows four two-measure units, each of which adds new notes until the complete comes part is stated in the fourth two-measure group.  As can be observed, this strand is the doubling of the Piano 2 left hand part (except, of course, for the full triad).



                                                                       Example 7: mm. 319-344


                    Reich's additive process is revealed in Example 8 which is the Piano 1 left hand part or dux of these bass clarinet parts.  Beat one of the comes part is the ninth beat of its dux  (arrow).  The first two-measure repetitions, four times, begin with the notes linked to circle 1 (when the triad appears, Ab and Eb are the notes taken by the bass clarinets).  The next two-measure repetitions add the notes of circle 2, the third two-measure group the notes of circle 3, and finally the Bb and Eb are taken from the triad of circle 4 to complete the statement.


                                                                     Example 8: mm. 25-26


                    Most often Reich's melodic increments are not exact doublings of one canon-ostinato part, however.  Rather, they are derived from the piano parts by (1) taking notes from different parts of the canon-ostinato, or (2) momentarily doubling a single part but changing it later in the process of generating a longer melody.  The first process is demonstrated below (Example 9) with a passage from Section II.  Here the piccolo and bass clarinet lines are taken from the right hand piano parts and left hand piano parts respectively.




                                                                                             Example 9: mm. 373-376.


                    There are three passages that contain particularly noticeable ten-measure melodies which find their source in the canon-ostinatos - a melody in the flute and later a juxtaposed piccolo melody in Section V.  In the first two measures of these melodies, pitches and/or rhythmic ideas are taken from a piano part.  These ideas are then brought back, often with subtle changes that distance the melody somewhat from that part.  In all cases, the melodies are limited to five pitches.


                    By far the most tuneful and independent of these ten-measure lines is the flute melody of Section I, measure three and four of which appear as a two-measure repeated unit near the end of Section II.  The projection of G#m in measures 1-4 followed by D#m in measures 5-10 divides the melody into two parts, each part consisting of two measure groups that are linked as shown in Example 10.


                                                                                            Example 10: mm. 87-96


                    Measures 1-2, 5-6 and 9-10 possess the same rhythm and nearly the same contour (mm. 5 and 9 are identical); measures 3-4 and 7-8 are sequentially related.  The structure according to these two-measure units, then, is A   B   A1   B1   A2   - the practice of returning to or altering ideas rather than continuously repeating them underlies the melodic construction of all these longer melodies.  In contrast to the melodic increments mentioned earlier, the pitches of this melody are not inherited from the canon-ostinato.  And with a prominence of thirds and narrower range, it is less disjunct and thus considerably disparate from the rest of the texture.  However, the

 first two measures take from the canon-ostinato the rhythm of beats 1-4 of measure one :                                                                                                                                                   

and beats 4-5 of measure two:  see Example 3.  It is this rhythmic doubling that establishes the two-measure grouping of the melody.


                    The second ten-measure melody (piccolo - Section III) is more derivative of the canon-ostinato.  As can be observed in Example 11-a, the first two measures double much of the Piano 1 right hand part with measure 1 providing ideas for subsequent measures, namely, eighths 1-4, 8-9.  After measure 1, (marked x) the melody settles into a sequence of alternating measures each of which might be thought of as a derivation of one or the other nearly identical measures provided in Example 11-b.  What follows are non-identical pairings of measures the first measure of which is articulated by the F7 (measure 3 is therefore marked a).  This two-measure organization suggests a variations structure:  A   A1   A2   A3   A4  .  Unlike the flute melody of Section I, this line is angular, relatively non-tuneful and without a sense of harmonic progression.


                                                                      Example 11a: mm. 403-412



                                                                      Example 11b: mm. 403-412


                    Near the end of Section IV, the piccolo and flute enter with a two-measure doubling of the second dux  (Piano1, left hand) and comes  (Piano 2, left hand) parts respectively.  At the beginning of Section V and after sixteen measures of repeat, the two parts then fade out.  These two measures, however, reappear (with piccolo and flute parts reversed) as the first two bars of a ten-measure extended version heard twenty measures before the end of the piece.


                    Prior to this (forty measures before the end), the flute enters doubling exactly the first two measures of the right hand part (dux) of Piano 1 (Example 12-a).  Measures 3-4, 6, and 7-9, however, move away from doubling.  Twenty measures later, in similar fashion, a piccolo melody enters with an exact doubling of the first two measures of the right hand part (comes) of Piano 2 (Example 12-a).  This line remains more faithful to the canon-ostinato with all but ten notes doubling the piano part.  Though the piccolo melody is not strictly imitative of the flute line after the first two measures, points of exact doubling of the canon-ostinato direct the listener to canon-like relationships especially with the recurrences of the familiar rhythmic patterns of beats 1 and 2 of measure one and beats 4 and 5 of measure two of the original canon theme.  Because the location of the comes   part of the canon-ostinato in this section begins on beat 4, measure two of the dux  (see also Example 4-e), the piccolo line tends to de-emphasize the two-measure groupings of the flute melody.




                                                                          Example 12a: mm. 677-696



                                                     Example 12b: Imitative Structure


                    On a larger scale, a comparison of the two ten-measure lines at their closest proximity, i.e., the measures of the piccolo line matched as closely as possible with the flute part (Example 12-b),[8] reveals that five of the ten measures are strictly imitative and two others are rhythmically identical - d  and d1 .  This alignment is achieved by considering the last four eighth notes of the repeated piccolo part (see arrow in Example 12-a) as beats 1 and 2 of a "comes " part.  Though each line possesses its unique return structure, a similar grouping of measures (4-2-4) is implied in both parts yielding the feeling of a ten-measure melodic statement.


                    A conspicuous feature of this canonic passage is the exchange between parts of figures derived from the high Db, Ab and C of measures 1 and 2.  They are especially pronounced because of their high register and appearance in one form or another in all ten measures.


                    This final passage is the culminating statement of the generative canonic process of the piece.  The two comes  parts in Piano 2 are finally aligned (Example 4-e), the ten-measure drone-canon is in place, and dux and comes  parts introduced in the opening one-measure canon are extended to ten measures and highlighted via two-part melodic counterpoint.  In addition, the extended drone-ostinato returns from Sections II and IV to round off the form.


                    Least prominent of all the melodic increments are the four entrances that compliment the drone-ostinato.  Reich calls on a bass clarinet to provide the bass line in Section II and draw out an inner part in Section IV, and uses the viola to supply relatively conjunct melodies in the middle register in Sections II and IV.[9]  These parts typically fade in as notated in Example 6.



                    Finally, one of the most unique features of Steve Reich's style is the appearance in some of his works of what he calls "sub-melodies."  (Because some in Octet are more rhythmic than melodic in nature and one is distinctly harmonic, I use the term "sub-patterns.")  These Reich considers "the unintended by-products of the intended process."[10]  In this piece, the angularity, register placement, doubling, rhythmic placement, and transposition shifts of the two-piano canon-ostinato, give rise to various non-predetermined yet desired patterns that are strong enough perhaps to be considered a fourth strand.  These two-measure sub-patterns add rhythmic/melodic/harmonic counterpoint to an already active texture, and in the ECM recording they ring through at times with such strength that it sounds as though they are being performed on a third piano.  The strength of projection, it should be noted, will vary with performance irregularities such as balance, performance hall acoustics, player mannerisms, tuning, recording idiosyncrasies, etc.


                    The canon-ostinatos in Octet project noticeably different sub-patterns.  Visual samples of these appear in Examples 13-a to 13-f which show composite piano parts of all the positions of the two-measure canon-ostinato.  The open note-heads indicate the sub-patterns that can be heard in each section in the ECM recording.  (It should be added that the strength of sub-patterns occasionally fluctuates within the section.)




                 Example 13-a: Section 1, mm. 57-58                                Example 13-b: Section II, mm. 171-172





            Example 13-c: Section III(a), mm. 343-344                        Example 13-d: Section III(b), mm. 403-404





            Example 13-e: Section IV, mm. 473-474                             Example 13-f: Section V, mm. 637-638


                    According to these examples (admittedly based on my hearing), four basic sub-patterns can be traced through the work:  figure 1 - the repeated ascending P5; figure 2 - the repeated descending P5; figure 3 - the four-note ascending bass line; and figure 4 - a non-melodic two-measure rhythmic pattern in the upper register.  The most penetrating sub-pattern is that of Sub-section III(b).  Here in particular the melodic and harmonic major sixths (Ab5 - F6) brought about by the abrupt expansion of range (Gb2 - Ab6 to Eb2 - C7) produce a striking "bell-like" quality (any listener not hearing this element as an exceptionally eloquent addition to the musical moment misses what this style and aesthetic position is all about).  Other figures might well be heard, nevertheless, the presence of sub-patterns in one form or other exists adding to the texture a wealth of musical material into which the ear can journey.


                    I consider Octet to be one of Steve Reich's most substantial works.  My judgment might well be influenced by my close association with the work and the composer, still, I find to be particularly exquisite not only his choices of pitch and rhythmic materials and his practice of generating sub-patterns, but his timing and control of subtlety levels of change. It is this dimension of minimalist composition that often serves me best in measuring artistic craft.


     [1] After the scheduled two two-hour rehearsals, it was mutually agreed the piece was not ready for performance, consequently, it was dropped from the concert program.

     [2] Commissioned by Radio Frankfurt (Hessischer Rundfunk), Octet was written in 1979 and premiered the same year by the Netherlands Wind Ensemble.  The piccolo was included during rehearsals of Reich's own ensemble later that year.  With Reich's blessing, Ransom Wilson, in 1983, re-scored Octet to include eight violins, four violas, four cellos, and a double bass part drawn from cello and bass clarinet lines; the four wind parts were given six players for reinforcement (Solisti of New York, Angel Records:  DS-37345).  A second arrangement by Tibor Szemzo of Hungary's Group 180 (which performs only repetitive music), includes an oboe, trombone, Fender bass, and bassoon, which Reich claims "adds a note of humor."  It was first performed in Hungary in 1984 (Group 180 II, Hungaroton:  SLPX 12799).

     [3] Reich mentioned to me that he grew up with the diatonic system and still enjoys working with it.  It was because of this preference that prompted one of his teachers, Luciano Berio, to suggest that he forgo 12-tone writing and compose tonal music.

     [4] The score used for this article was given to me by Steve Reich for the (1982) rehearsals, the notation is his; it was his wish that it remain in the San Jose State University Library.  A few discrepancies between the score and the ECM recording were discovered during the study - two repeated measures (mm. 451-452) are not on the recording; the string parts at m. 481 in Section IV differ from the recording; a saxophone appears to be playing the Bb clarinet part at m. 153 in Section II and m. 455 in Section IV; the Bb bass clarinet part at m. 195 is out of range and apparently was recorded on a contra-bass clarinet; and the viola at m. 183 sounds an octave higher than written.

     [5] Octave identification:


     [6] In an interview with Jonathan Cott, Reich explains that when asked some years ago why he doesn't write any slow music, he replied, "In my Octet, are you going to concentrate on listening to pianos . . . or to the strings, which are playing much more spaciously?" (record notes, The Desert Music, Elektra/Asylum/Nonesuch Records, 1985).

     [7] The drone-ostinato is taken directly from the vocal part of his Music For A Large Ensemble (1978).

     [8] Measures of the piccolo part in Example 12-a begin with the identifying letter e so that they are properly shown to be in contrast to the flute part.  Example 12-b, however, is based on the "imaginary" relocation of the bar line treating the "new" measures as the comes  part, hence, the use of letters that correspond to the flute line.

     [9] The score does not show the viola part with an independent melody in Section IV.  However, a distinct melodic line exists in the recording.  Since the viola was given this part in Section II, it is presumed here that it was given the same role in Section IV for the recording.


     [10] Reich, Steve, Writings About Music, New York University Press, New York, 1974, p. 10.