I have put together a collection of musical examples based on playing the Seek Well “game” with myself. The object of the game was to play various tonal melodies and record the continuations I expected to hear at the point where each melody left off. More importantly, the reasons behind each expected continuation were examined in the context of the musical forces of gravity, inertia, and magnetism (along with other traditional musical factors such as key, meter, and harmony). The leaping-off point for my thought processes was the multi-level computer model in each case, although I was not assuming that I had been given a multi-level analysis of the melody. Instead, I gave some thought to how the interpreted structure of the initial cue melody would arise during listening. I also was thinking about the examples loosely in terms of a Letter Spirit-type model of cognition, where the interpretation would be governed by various musical factors interacting with different levels of urgency depending on the context.
To clarify my descriptions below, I should point out that I had some trouble deciding what the appropriate hierarchical structure of each melody would be in each case. In particular, I didn’t have any melodies where more than 2 levels of hierarchy were apparent. Thus, I usually refer to the “surface” when I mean all the notes of the melody but I say “background” when I’m thinking about the next level of structural importance. In some cases this should probably have been called a “middle level” with a true background level hidden deeper in the structure.
The musical examples are available on my web site at the following URL:
http://www.cs.indiana.edu/~epnichol/seekwell/
In the examples, the first line of music is the “cue”, while the other lines of music are possible continuations. The background level of the analysis is also shown in most cases.
After the examples I present several conclusions and things I have learned from the experience.
Examples of Expected Melodic Continuations
This is a repeat of the example we went through while I was
in
In listening to this melody, the key is established easily by the ascending scale through the tonic triad, with the emphasis on C and E.
The presence of the repeated E, where repetition is not the norm in the rest of the cue, suggests that the second E falls on a strong beat. This suggests 3 beats per measure.
Given the meter and key signature, the background structure can be extrapolated from the strong beat of each measure. In this case it gives C followed by E.
Now, to compose the expected continuation, the multi-level model would examine the structure at level B. At this level, the C and E notes suggest a reference alphabet of the tonic triad, with the tonic and octave comprising the goal alphabet. Inertia and magnetism both point to continuing to G instead of giving in to gravity and falling back to C, so the next note of the prediction is G followed by the inertial continuation on to the upper C. This completes the background level completion.
At the surface level, the given rules determine that the major scale is the reference alphabet, while the tonic triad is the goal alphabet. To fill in the surface structure an analysis of the other notes besides the strong C and E is necessary. In both measures, the first note is followed by two ascending steps through the reference alphabet, so this pattern is recorded and used to complete measure 3 which is already known to start with G, producing the G-A-B continuation. Finally, in measure 4, we note that both the forground and background levels have reached a goal pitch – C in both cases – so this is an appropriate place to end the expected continuation.
A second expected continuation is also possible. Consider that after analyzing the given cue in the same manner as above the listener entertains the possibility of continuing on to G and then on to C in the background level as in staff b. However, due to the expectation for the 2nd and 3rd beats of each measure to continue rising by step, this background implies a melody that continues to ascend in the same direction for a whole octave. This is a rather long span to continue in the same direction without interruption, so pressure to turn around in the middle is introduced.
A simple idea to “turn around” the melody would be to start measure 3 with a C, giving in to gravity at the background level shown in staff d. This idea might be workable, although the jump back down from G to C in staff E might sound out of place after the stepwise ascending beginning of the melody; this expectation of stepwise continuity at the surface introduces another pressure.
Instead of modifying the background level, another idea is to modify the surface structure in staff e. Retaining the G in the first beat and discarding the idea of continuing upwards to A and B, the initial analysis of beats 1 and 2 in the first two measures might be generalized to say “continue by stepping twice in the same direction for two beats”. With the constraint of going upwards thus removed, the continuation could proceed downwards through F back to E. Following this lead, the background structure might also “turn around” in measure 4, going back down to E just in time for the E to be repeated (following the pattern of the 1st beat of each measure repeating the 3rd beat of its preceding measure). Finally, the background could continue due to pressure from inertia and gravity back down to C, as would the surface structure, coming to rest on the goal of the initial C in measure 5.
This potential continuation, unfortunately, lacks the strength of the first one (staffs b-c) due to its mundane symmetry (with a repeated high note in the very middle of the phrase) and the implied tonic harmony throughout. The shift in direction in measure 3 weakens the background structure which was able to climb to the upper C in the first version. The first version also implied a shift of harmony to the dominant in measure 3, cadencing on I on the last note; no cadential pattern is in effect at the conclusion of this melody.
Examining the weaknesses of the second continuation might lead to several different improvements. Two of these in particular seem promising as places for departure: the lack of harmonic changes and the lack of an implied cadential pattern.
Considering these ideas we can return to deciding how to continue the background structure in measure 3. Because the G has already been tried both times before, we again search for a new note. One of the pressures at the surface level – reversing the direction of the melody – is still in force here, so a note below G, such as E, seems in order. However, we also want to change the harmony, possibly to V in order to establish a cadence. V gives us the most likely choice of G again, which has already been used effectively in the first attempt; B and D are too far away from the G at the end of measure 2 to be useful while preserving stepwise continuity. This leads us to consider other chords such as V7. Indeed, the F in V7 seems promising, because it is just a step away from the melody’s G and also reverses direction as desired. (Staff f)
Several possibilities emerge here, such as continuing with F-G-A in measure 3 -- although this implies the harmony IV instead of V7 it seems reasonable. However, because we have reversed the direction of the melody at the background level it seems natural to reverse the pattern analyzed for beats 2 and 3 in the first measures as well. Where the previous pattern was “continue up for 2 steps” this would be effectively swapped for “continue down for 2 steps”, leading to the continuation “F-E-D”. This also has the virtue of implying the V7 intended (or perhaps ii or vii, of course).
Finally, to complete the phrase we need to continue the background level, because F is an unstable tone to stop on. The most likely candidate is E, predicted by giving in to gravity and magnetism at the background level. This also implies a tonic harmony in the 4th measure which makes sense following the dominant harmony in the previous measure. This continuation has reached a point of repose here at this imperfect cadence, so this is a good point to stop the continuation.
Measures 1-4(b1) are analyzed just as in Example 1. However, in measure 4 beats 2 and 3 are a neighbor figure and then measure 5 repeats the C but with the embellishment suffix “go down for 2 steps” for beats 2 and 3.
For the expected continuation in m. 6, it follows by inertia that the pattern will continue: G-F-E. For m. 7, we turn to the background structure where by inertia, gravity, and magnetism, the line will continue down to E, and on down to C by inertia and gravity from there. Filling in the line gives E-D-C for measure 7 with a final C in measure 8.
However, this continuation suffers a similar problem as the second continuation in Ex. 1: the line is extremely symmetrical. Also, the harmony is mostly tonic, except for the implied dominant in m. 3. A cadence seems missing in m. 7-8, especially considering the simple antecedent-consequent 3 measure phrases implied here. Finally, a subtle “hiccup” is evidenced in the surface structure at measure 7: the E is repeated across the barline in a way inconsistent with measures 5 and 6. Although this repetition happened in m. 1-3, it seems out of place here due to the smoothly descending melody coupled with the lack of a harmonic reason for the repeated note. The harmony doesn’t change from tonic here; otherwise the repetition might signify a suspension or similar accented nonharmonic tone.
Given these problems, it is easy to improve the surface structure by modifying measure 7, where the “hiccup” of the repeated E occurs: by changing the harmony to G in this measure, we can continue down to a D on beat 1 (see staffs d and e) based on inertia, gravity, and magnetism. This measure can then proceed using our inertial rule “down 2 steps”, giving D-C-B followed by the C in measure 8. This is a more satisfying ending due to dominant->tonic cadence and the smooth effect brought about by “dipping” below the tonic to the leading tone which pushes the melody back up to the tonic. The change of direction completes the feeling of closure.
A similar fix to measure 7 based on V7 harmony provides the ending F-E-D-C, where the change in direction in measure 7 adds interest. However, the ending that includes the leading tone may be slightly preferable because of the improved feeling of closure at the end of the phrase.
When listening to the start of the given melody, the first five notes seem to indicate G major due to the metric prominence of the Bs and the G. The rest of the melody helps to confirm the key due to the presence of the leading tone F# and another G and B. Additionally, it sounds like duple meter due to the position of all the Bs and Gs as well as the change to a non-tonic triad tone, C, on the 9th note. Thus we can assume 4 notes per measure.
The most prominent notes in this melody are the initial B, the G at the start of m2, and the harmony-changing C in m3. Assuming that these notes for a deeper level of structure, we turn to examining how each of these notes is embellished. The first measure has an upper-neighbor tone followed by a descent to link with the G in the next measure. Measure two is similar, with a lower neighbor tone (perhaps this was due to inertia in the downward-moving line) followed by a skip up to B to link smoothly with the C in the next measure.
To start an expected continuation, it is necessary to figure out where the background like is leading. Because the skip from B down to G was followed by a skip back up to C, it doesn’t seem clear what should follow. Perhaps the G is heard as leaving an auralized trace so that the background should return to fill in the gap created; perhaps by moving to A. Another possibility is that the G would be discounted because the B-C ascending line is more coherent, in which case gravity and magnetism might pull the C back down to B (see staff c). In this case the upper-neighbor figure is used on beats 1-3 (instead of the lower-neighbor as in m2) due to the upward inertia created by the ascending G-B-C line that precedes it.
Alternatively, the B-C ascending background line might continue due to inertia, moving on as B-C-D (staff d). Here the upwards inertia is invoked in a strong fashion; upwards inertia is present at both the background level and the foreground level by the time we reach beat 2 of m3. The upwards inertia overcomes the pattern-inertia that was causing up to expect an upper-neighbor tone, and instead the melody continues upwards to E before finally turning around to connect smoothly with the following D via a skip to C.
A final possibility is shown as staff e: In this case, the background level is expected to continue even past B-C-D in order to reach the goal of the tonic. In this example continuation the background level simply jumps on up to the tonic G instead of moving through all the intermediate notes (B-C-D-E-F#-G). This quickening of the underlying pace may be due to the expectation of a phrase that ends around m4 or 5. One other interesting feature of this continuation is the next-to-last note, the final D in measure 4. This note arises due to several pressures: if m4 were written D-E-F#-G-G, the ending would be a bit weakened by the anticipation of G, and this repeated note pattern is not characteristic of the rest of the melody that has come before; most connections are by step, with a few skips of a third. Second, the third measure sets up a pattern of stepping up through a third before dropping back to the first note; this little motive can be reused in m4 to good effect because it avoids the repeated note G and slightly prolongs the effect of the leading tone before the final tonic.
Key and rhythm are established together for the listener, based on an initial attempt to use 4 beats to the measure. This guess appears to work well because the second measure follows the same neighbor-tone pattern as the first, transposed. The initial G followed by the E suggests a triadic harmony incorporating these notes, such as E minor or C major. The E minor is ruled out by the F-natural, so C-major seems like the best choice.
Having established the key and rhythm, we see that the G in measure 1 and the E in measure two figure prominently due to their strong metrical position and membership in the implied tonic triad. The neighbor-tone figure also highlights the importance of these tones. Thus, the background structure is inferred as a single G in measure 1 followed by the E in measure 2. Given this, the background can be easily predicted to contain a C in measure 3 due to inertia, gravity, and magnetism applied to the tonic triad, with the tonic-dominant frame as the goal alphabet. Finally, the foreground level might be filled in as C-D-C-B in measure 3 with the final C in measure 4 completing the prediction. (staff d)
A few variants might be tried in measure 3 to add interest instead of repeating the figure in measures 1 and 2 exactly. For instance, the forces of inertia and gravity might take over at a surface level to pull the melody all the way down to A before rising back to the final C (see staff c).
Finally, this continuation in Staff e may be expected if the listener hears the start of an antecedent phrase and anticipates an entire period for the continuation. Instead of continuing down to the tonic C by inertia and gravity, the music is heard to be driving toward a half cadence, perhaps on V. The melody continues by modifying the neighbor tone pattern slightly by repeating the upper-neighbor F (a passing tone in context at beat 4) in order to reach a point of repose on the G in measure 4. Even with the requirement of only using notes of the same duration, the listener might expect a pause here in the form of a longer duration for the G, rests, or a fermata. In any case, the consequent phrase would then be expected to continue on down from the G to the C, perhaps copying one of the earlier expected continuations of the first phrase.
The first obvious characteristic of this example is the key: 3 of the first 4 notes make up a C major chord, and indeed all the notes of the C major diatonic scale are present here, with no extra notes.
The meter is a little harder to figure out: the jump to the C in beat 4 might suggest triple meter, but the skip from G down to E occur 4 notes later rather than 3, so this isn’t the strongest theory. Digging a little deeper, as noticed before the harmony implied by the first 4 notes is a C major triad, while the next four notes seem to be in a different harmony – iii or perhaps V – due to the presence of the B on the 5th note.
Analyzing the background structure helps to settle the matter of rhythm. It is hard to find consistent background level melodic line if we assume triple meter. It is also hard to see a coherent melodic background if we assume 2 beats (this might make sense in Phrygian mode, however…), and every 4 beats doesn’t work very well either. However, if we look at every 8 notes, we see the start of a linear melodic segment: E-D, where the next note in the sequence is predicted to be C by gravity, inertia, and magnetism. Thus, we expect to hear this C after 2 measures of the continuation. (see staff b).
The 8-note pattern established in m1-2 seems to be a rising line with a leap to a chord tone followed by a descent back to the initial note. Following this pattern we might expect something like m3-4 in staff d of the example, where the last note of m4 is changed to the dominant to help establish the V-I sound at the end of the phrase. Note that the tritone leap leaves me feeling unsettled, perhaps because it leaves an auralized trace, even though the leaps were through chord tones if we analyze m3-4 as V7. This might explain the alternate continuation shown in staff d: instead of ending on the tonic the trace left at the surface by the F-B leap disrupts the expectation of the C at the background level; instead there is pressure to “fill the gap” below the B. This leads to the idea of moving to A rather than C at the background level, which makes harmonic sense because it would imply a V-vi deceptive cadence.
It should be noted that these completions could probably be improved; in both cases the tritone leap sounds a bit out of place.
The first challenge in developing an expected continuation to this melody is understanding the rhythmic structure. After hearing the first 4 notes the best guess as to the meter would be to assume a regular duple meter, but as soon as the melody drops down to the A on the seventh note the first 6 notes begin to be heard as one unit. The following notes, B-A-B-C, follow the same pattern as the first notes did, reinforcing the idea that the regular meter is based on 6 notes. It might be possible to hear this in 3 instead of 6, but the implied triads make this less likely: if the first 3 notes were in one measure and the next 3 notes in another measure, the 4th note of the phrase, F, would be an accented passing tone in between two chord tones of an implied C major or A minor triad. Hearing the metrical accents on the 1st, 3rd, and 5th notes here gives more support to the notes of the implied triad, and seems like a more natural way to hear the passage. The same argument follows for the 2nd measure (the A-B-A-B-C-E figure).
At this point is seems possible to begin a possible analysis of the background level. The upper-neighbor and upper-third figure in both measures, along with the repetition of the first pitch of each measure, establishes the prominence of the first pitch of each, so these pitches are retained at the background level (E in measure 1 and A in measure 2). It still is unclear which key is implied at this point, although C Major and A minor seem the most likely choices. Additionally, C major is preferable because there are no explicit notes present from the melodic minor scale.
The next step is considering the background analysis to see if it can be predicted clearly. However, because there are only 2 notes separated by a leap, there is no obvious pattern although it is interesting that the leap is down a 5th. If the background level were to have a predicted completion at this point it would make sense to relate the next note to the initial E because the leap to A leaves an auralized trace. Gravity would suggest moving downwards, perhaps by step to D.
At the point that moving to D is noticed as a possibility, several musical ideas might be brought to the surface of attention: the E-A-D pattern proposed at the background level is an obvious movement by 5ths that makes musical “sense”. This suggests the possibility of a melodic sequence at the surface level, and indeed the first two measures sound sequential because of the similarity of the first 5 note pattern in each measure. Finally, when reexamining the surface melody it becomes apparent that stepping down to the note D would follow nicely after the leap up to E at the end of the 2nd measure.
The confluence of all these factors leads us to write D at the start of the 3rd measure, filling in the 3rd measure melody as D-E-D-E-F in the pattern of the previous measures. The last note for the measure is not as obvious because in the first measures it functioned more as a connective between the measures. D might make sense here to keep this measure a transposed version of measure 1. In addition, if the next note were to be found below D the D would help connect to the next measure through gravity and inertia. In particular, the next note may be predicted to be a low G due to the sequence by 5ths (and the background note in measure 5 would be C, finally bringing the sequence to rest on the tonic).
However, after writing measure 4 as a low G-A-G-A-B…, the complete melody to this point may begin to sound monotonous – a good rule of thumb states that after 3 repetitions a melodic figure can start to lose its interest. The sequential figure is quite plain due to its repetition of the first note of each measure, and the circle of fifths is perhaps too easy to predict. Several ideas might spring to mind here about how to liven up the melody in measure 4. Due to the fact that we are due for the note G and the tonic is predicted in measure 5, the sequence may be considered more generally in its harmonic form rather than as strict melodic transposition. Notes of the dominant G major triad, G, B, and D (and also F) are possibilities, although measure 3 already used D and we already tried G. B seems more likely than F for the next choice for several reasons. First, we earlier noted that it was a bit hard to establish the key definitively as C major, and the presence of the leading tone might help drive the key home here at the V-I cadence that ends the phrase. Additionally, if we reconsider the background level, the pattern E-A-D-B-C is striking: The repeated leaps – there is no stepwise movement here – leaves auralized traces that split the notes into two separate lines: the E-D line on top appears to be progressing under the force of gravity by step, to be drawn on down to C by gravity along with inertia and magnetism. The lower A-B line, however, is progressing upward through the scale by step, and inertia and magnetism will pull this line up to a unison of the two lines on the final C. The compelling nature of this background leads us to complete the melody with something like B-C-B-C-D-E—C, where the E at the end of the measure was predicted by inertia.
In this longer example, the key of C minor is easily apparent due to the collection of tones in use, which cover the entire melodic C minor scale. The half step down to B natural on the 8th note is a particularly strong clue as to the key. Considering the meter solidifies this key choice: the measures are readily perceived to have 4 beats due to the tonic minor triad outlined on notes 1, 5, and 9, along with the G on beat 13. These notes are also easily perceived as structurally significant due to their metric position, so they are retained at the background level in the embellishment hierarchy.
The surface pattern of the given melody does not fit a specific pattern in relation to each background level note, but the melody does have a certain smoothness to it due to the continuous stepwise motion and lack of any immediately repeated notes. This accounts for the movement back up to F in the first measure, for instance, contrasted with the second half of the given melody that rises by step through a diminished seventh without ever turning around in direction.
The expected continuation in this case is driven by the expectation of a consequent phrase, where the given melody is the antecedent phrase. This expectation arises due to the focus on the dominant in m4; this sounds like a half cadence on V. At the background level, a simple continuation of the phrase is generated by expecting to hear the G-Eb-C tonic triad outlined again followed by a concluding goal tonic C to end the period. (see staff b). The expected G in measure 5 also arises from the voice leading at the surface in m4; the last note is an F, which might be expected to continue down to Eb by gravity and inertia. However, the pattern of continuing with stepwise motion also gives G as a possibility, and due to the desired parallelism with the first phrase this choice makes sense.
There are several possible completions given this expected background structure; one example is given in staff c. Here m5 is different from m1 in order to provide some variety in the phrase, although paradoxically this causes m4 and m5 to be identical, which could probably be improved. However, this pattern in m.5 effectively connects the G to the Eb at the start of m6. Similarly, this exact neighbor pattern is reused in m6 to connect smoothly to m7. However, in m7 there are a few pressures to change the pattern. First, the repetition of the neighbor tone pattern a fourth time might be too monotonous. Second, there is the expectation of a V-I cadence at the end of the phrase, and a C-D-C-B-C pattern doesn’t convincingly demonstrate the V chord. Instead, m7 may be expected to sound like I on beats 1 and 2 and V on beats 3 and 4. This causes the “rule” of consistent stepwise motion to be broken, allowing the C-Eb leap on the first 2 beats followed by the D-B leap on beats 3 and 4, establishing the cadential chords while minimizing the size of the leaps. This measure still flows smoothly. Additionally, the slight variation (allowing jumps) causes the measure to stand out a bit more and prepare the listener for the end of the phrase.
A second possibility is given in staff d. This might arise due to a similar argument as above if the background level of the continuation were expected to proceed by step through the diatonic scale instead of through the tonic triad. Note, however, that here the descent from G to C is hastened at the end in the rush to get to the end of the phrase in m. 8 as implied by the 4-measure first phrase. The background line is G-F-Eb-C. Of course, the D is heard, but it’s not found at the same metrically significant spot at the other notes in the background.
The initial challenge provided by this example is determining the key. Because there is no obvious way to determine the key it makes sense to continue on to examine the implied meter, which may be expected to help shed light on the key problem. As usual, in the absence of other information we can safely assume duple meter. In this case, the 5th note stands out because of the leap to and away from the note, so it would make sense for it to be on a strong beat. Additionally, the pattern of notes in each measure (if we assume 4 beats per measure) is consistent; there is a pattern involving descending notes with identical 2nd and 3rd notes in each measure.
Returning to the question of key, our metric analysis has provided some useful insight. First, due to the collection of pitches present, we can assume C major, G major, or A minor are the most likely choices. Next, if we just consider the strongest beats – the first beat of each of the two measures – we see a rising C-D movement. This might seem to be a typical way to begin a melody in C major or a minor, but it seems out of place for G major. Looking at the notes another way, we might notice an overall descending line including C-B-A-G. This also sounds promising for C major, and perhaps a bit more characteristic of C than A minor. Finally, perhaps there is a slight preference for major keys when there is a lack of a definite answer; there is no indication of the melodic minor scale here, for instance. Thus, we can proceed with an idea of being in C major, even though with the right continuation we might be able to make a good A minor melody.
The next question is one of determining the current embellishment pattern in order to predict the continuation in the following measures. The most obvious bit of pattern is the repeated 2nd note in each measure. Additionally, it is clear that each note after the first in each measure is lower than the previous. In m1 the pattern is stepwise but in m2 there is a leap down away from the first note, followed by the stepwise descent.
A sample continuation would take the background C-D line into account and write something like E-D-D-C for m3. For m4, the C-D-E line might keep rising up to F based on inertia, but leaping down in the C triad back to C also sounds plausible due to voice leading because the surface structure is a melodic line leading down to C (see staffs b-c). As the tonic, C is certainly a goal tone and this might be a phrase ending.
A slightly different continuation might give thought to harmony, and if it decided to retain tonic harmony then m3-4 might turn out the same as above. However, if we expected a half cadence on V then we might expect the background level to continue to a note of the V chord in m4. In this case, D seems plausible based on a gravity prediction from the E above and a magnetism prediction based on moving to V harmony. In hearing the continuation based on the background C-D-E-D line, we might also employ the device of m. 2 to leap down between the 1st and second beats, because this has the benefit of emphasizing C major, which wasn’t clear enough in the beginning measures; this would lead to E-C-C for the first 3 beats of m3. The last note could continue down to B; however, there is a force expecting it to move up to D for improved voice-leading with the D expected in m4. This involves relaxing the pattern noticed in the first 2 measures by requiring only a stepwise motion to beat 4 instead of a step downwards. (see staffs d-e).
These two continuations notwithstanding, there is still another possibility, based on seeking an explanation for the pattern of repeated notes in the middle of each measure combined with the enticing C-B-A-G descending line hidden away in the opening 2 measures. The repeated notes on beats 2 and 3 sound too mechanical for some reason, but if we heard the second as an accented nonharmonic tone it might be more interesting. Indeed, if we shift our perception of the measures so that the 2nd note of each repeated pair falls on the downbeat, we hear the first two notes as a pickup measure and the C-B-A-G line is more prominent. Ignoring the pickup measure for the moment, we might also imagine the familiar idiom of a series of 4-3 melodic suspensions, implying a bass line such as F-E in measures 1 and 2, along with a possible analysis of ii6 and I6 in the first measures.
Considering the background structure, we now see a prominent C-B-A line if we include the C in the pickup measure (even though it’s not in the same metric position as the B and A). This line has the obvious continuation down to G, so we can predict measure 3 to go something like A-G-C-G (see staffs f-g) where the jump up to C is expected by reexamining the embellishment pattern in the 1st measure and seeing that the 3rd note is a leap to and away from a harmonic tone. This continuation in m. 3 also helps solidify the feeling of C major.
Finally, m. 4 and 5 might drive towards a cadence on V to help clarify the key. The sample continuation is a bit hard to explain, although I think it’s due to an expectation to keep the background descending line moving instead of stopping on G; it leaps down to the next note in the V chord, D. This inspires the voice leading in m. 4 where the 4th note is predicted to be E instead of F in order to move stepwise down to the final D. In addition, the 3rd note of that measure might be predicted to be B (the pattern here is a fourth above the 2nd note), but the desire to avoid the augmented fourth leap causes the B to be replaced with a smaller leap up to A. This also preserves the leap of a fourth downward from beat 3 to beat 4, as well as causing the leap to be to and away from a harmonic tone (as in the previous measures) if we analyze this measure as the ii chord.
Throughout the examples above, some ideas occurred repeatedly. I have collected those that seem the most significant here for use in future reflection: