Ralph Kirkpatrick's odd rendition of D. Scarlatti's Sonata in G, K. 455
November 9, 2024 1:37 AM Subscribe
Here's a youtube recording of Ralph Kirkpatrick playing Domenico Scarlatti's Sonata in G, K. 455. I think there's something very peculiar in this performance and I'm hoping someone here agrees and can explain it. I'll describe it below the fold so you can listen to the piece first without bias. If you're not familiar with the piece, here's Scott Ross's version, which doesn't have the peculiarity.
Here's the peculiar thing I hear: About 27 seconds into Kirkpatrick's recording, there's an extra half measure, so if you were counting along 1-2-1-2, you'll suddenly find yourself counting 2-1-2-1. About 7 seconds later, there's another surprise half measure and we're back to 1-2-1-2. It happens again when the A section repeats. Anyone else hear it?
I've been listening to many versions of this piece, and this is the only one that does this. All the scores I've seen agree: Kirkpatrick is wrong. But he's the K in K. 455! Surely he knows what he's doing.
I haven't seen a manuscript, so maybe Scarlatti actually did write those half measures. But is there any other example of Scarlatti engaging in such metrical shenanigans?
I really don't know what to think. Your thoughts?
Here's the peculiar thing I hear: About 27 seconds into Kirkpatrick's recording, there's an extra half measure, so if you were counting along 1-2-1-2, you'll suddenly find yourself counting 2-1-2-1. About 7 seconds later, there's another surprise half measure and we're back to 1-2-1-2. It happens again when the A section repeats. Anyone else hear it?
I've been listening to many versions of this piece, and this is the only one that does this. All the scores I've seen agree: Kirkpatrick is wrong. But he's the K in K. 455! Surely he knows what he's doing.
I haven't seen a manuscript, so maybe Scarlatti actually did write those half measures. But is there any other example of Scarlatti engaging in such metrical shenanigans?
I really don't know what to think. Your thoughts?
Neat feature on youtube I didn't know about until recently: you can listen at 25% speed. Which I just did, with the IMSLP score, and....I don't know quite what's going on there. It doesn't sound to me like he's playing quite what's written but I'm lousy at reading music and he is an actual paid harpsichordist. You might try it, though.
Bonus, just to compare with a piano recording which was bound to be a bit more succinct, I listened to the Horowitz which, much as I love a harpsichord and period instruments in general, Horowitz in Scarlatti is always the best.
posted by less-of-course at 5:21 AM on November 9 [2 favorites]
Bonus, just to compare with a piano recording which was bound to be a bit more succinct, I listened to the Horowitz which, much as I love a harpsichord and period instruments in general, Horowitz in Scarlatti is always the best.
posted by less-of-course at 5:21 AM on November 9 [2 favorites]
I just tried playing along with these, and while Scott Ross is following the cpdl score, Kirkpatrick does something different starting measure 29 through the cadence in 31, that involves adding two beats (giving it the phase-shift feel you noted).
Where the score's left hand runs:
29 F# D G G
30 A G F# D
31 G E A A
Kirkpatrick plays
29 F# E F#D
30 G E A G
31 F# D G E
32 A A
Kirkpatrick's version is consistent with the same underlying chord progression, and I actually think I like his version better, since on reflection it feels weird and foreshortened for that section to only go 7 bars.
posted by Bardolph at 6:32 AM on November 9 [3 favorites]
Where the score's left hand runs:
29 F# D G G
30 A G F# D
31 G E A A
Kirkpatrick plays
29 F# E F#D
30 G E A G
31 F# D G E
32 A A
Kirkpatrick's version is consistent with the same underlying chord progression, and I actually think I like his version better, since on reflection it feels weird and foreshortened for that section to only go 7 bars.
posted by Bardolph at 6:32 AM on November 9 [3 favorites]
Agreeing with @lownote on faster, looser, clumsy
I also wonder if some of the difference is the instruments themselves
posted by falsedmitri at 7:07 AM on November 9
I also wonder if some of the difference is the instruments themselves
posted by falsedmitri at 7:07 AM on November 9
Best answer: Professional harpsichordist here.
Well spotted.
First, what exactly happens?
In bar 29 (and the identical bar 36), in both original sources (also available on IMSLP. Venice manuscript book 11 and Parma manuscript book 13, the second sonata in both), the harmonic progression speeds up from one melodic step per bar to two steps per bar.
In other words, the top melody that initially changed every bar at the bar line now moves upwards every half bar: Bars 29/36: top not a and b. Bars 30/37: top note c# and d. Then the cadential twiddle comes in bars 31/38. So this is how it's written in the music (and every modern edition I've seen).
What does Kirkpatrick do? He plays bars 29/36 analogous to the preceding bars (as opposed to the next-following ones) that is, progressing with the melody by only one step per bar. So here, the a lasts the entire bar 29/36, and the b comes half a bar too late, on the downbeat of bar 30.
As noted in the original question here above, this manoeuvre adds a half bar in the second half of bar 29 that only gets caught up with in bar 36, where the entire phrase is literally repeated, including another added half bar.
Second, why would something like this happen?
I had to exclude the most natural explanation (something that happened when I recorded my first few Scarlatti sonatas back in the day), namely that the producer who did the editing of the (at the time:) tape made an editing mistake and added a half bar. First, that kind of mistake would only have happened once, not twice in the first half and twice in the repeat. Second, the left hand in the altered bars 29/36, f#-e-f#-d, matches the altered melody but not anything else on the page. So, intentional or improvised, the version we hear here is not an accident purely in terms of harmony and voice leading.
So, Kirkpatrick likely did one of two things:
1) he mis-remembered the music. Playing from memory, which he perhaps did, would have introduced this kind of risk. He wouldn't be alone either; there are some fabulous blunders in recordings of other world-famous keyboard players. It is true, when we record our repertoire, we usually grab the score just to make sure, and to be able to re-do botched passages on cue (i.e. bar number), but if we feel safe with the repertoire, perhaps we also don't. Kirkpatrick had researched Scarlatti thoroughly and knew his stuff inside out, so perhaps he left his music book at home, who knows...
2) and just as likely, he altered the text on purpose. The two-volume Kirkpatrick Schirmer edition of selected Scarlatti sonatas (which sadly doesn't include this sonata) is full of small textual changes, things that perhaps sound better, added and omitted notes, re-written embellishments and adjusted passagework (as compared with the existing sources). So he was definitely ready to add his own stamp on his interpretations, no matter his reputation as a Scarlatti scholar.
PS: there's nothing clumsy about his playing.
posted by Namlit at 11:29 AM on November 9 [6 favorites]
Well spotted.
First, what exactly happens?
In bar 29 (and the identical bar 36), in both original sources (also available on IMSLP. Venice manuscript book 11 and Parma manuscript book 13, the second sonata in both), the harmonic progression speeds up from one melodic step per bar to two steps per bar.
In other words, the top melody that initially changed every bar at the bar line now moves upwards every half bar: Bars 29/36: top not a and b. Bars 30/37: top note c# and d. Then the cadential twiddle comes in bars 31/38. So this is how it's written in the music (and every modern edition I've seen).
What does Kirkpatrick do? He plays bars 29/36 analogous to the preceding bars (as opposed to the next-following ones) that is, progressing with the melody by only one step per bar. So here, the a lasts the entire bar 29/36, and the b comes half a bar too late, on the downbeat of bar 30.
As noted in the original question here above, this manoeuvre adds a half bar in the second half of bar 29 that only gets caught up with in bar 36, where the entire phrase is literally repeated, including another added half bar.
Second, why would something like this happen?
I had to exclude the most natural explanation (something that happened when I recorded my first few Scarlatti sonatas back in the day), namely that the producer who did the editing of the (at the time:) tape made an editing mistake and added a half bar. First, that kind of mistake would only have happened once, not twice in the first half and twice in the repeat. Second, the left hand in the altered bars 29/36, f#-e-f#-d, matches the altered melody but not anything else on the page. So, intentional or improvised, the version we hear here is not an accident purely in terms of harmony and voice leading.
So, Kirkpatrick likely did one of two things:
1) he mis-remembered the music. Playing from memory, which he perhaps did, would have introduced this kind of risk. He wouldn't be alone either; there are some fabulous blunders in recordings of other world-famous keyboard players. It is true, when we record our repertoire, we usually grab the score just to make sure, and to be able to re-do botched passages on cue (i.e. bar number), but if we feel safe with the repertoire, perhaps we also don't. Kirkpatrick had researched Scarlatti thoroughly and knew his stuff inside out, so perhaps he left his music book at home, who knows...
2) and just as likely, he altered the text on purpose. The two-volume Kirkpatrick Schirmer edition of selected Scarlatti sonatas (which sadly doesn't include this sonata) is full of small textual changes, things that perhaps sound better, added and omitted notes, re-written embellishments and adjusted passagework (as compared with the existing sources). So he was definitely ready to add his own stamp on his interpretations, no matter his reputation as a Scarlatti scholar.
PS: there's nothing clumsy about his playing.
posted by Namlit at 11:29 AM on November 9 [6 favorites]
Response by poster: Thanks all for your input!
@HearHere, glad you like the piece. I was thinking about making an FPP about it. Still am.
@less-of-course, help me decide which of the Horowitz recordings on youtube to put in the FPP.
@Namlit, I was hoping you'd pop in. Thanks for your insights. I'll just consider Kirkpatrick's version a "fabulous blunder"!
@Bardolph, glad to hear you prefer it. You remind me that not everyone doesn't like what I don't like :D
posted by mpark at 6:07 PM on November 9
@HearHere, glad you like the piece. I was thinking about making an FPP about it. Still am.
@less-of-course, help me decide which of the Horowitz recordings on youtube to put in the FPP.
@Namlit, I was hoping you'd pop in. Thanks for your insights. I'll just consider Kirkpatrick's version a "fabulous blunder"!
@Bardolph, glad to hear you prefer it. You remind me that not everyone doesn't like what I don't like :D
posted by mpark at 6:07 PM on November 9
I also wonder if some of the difference is the instruments themselves
There's a difference in pitch that may be confusing to the listener. Kirkpatrick plays in modern pitch, Scott Ross in 415Hz. a semitone lower.
The recordings were made more than ten years apart (Kirkpatrick 1971, Ross 1984-85) which, in the context of the history of recording harpsichords in the second half of the 20th century, actually matters. Technicians started to go further away from the instrument as time went on, creating a less harsh and 'overtony', more natural sound. Ross's recording is still too direct to my taste, but it sounds more like the real thing.
The instruments: for this sonata a keyboard up to g''' is needed, which isn't, and wasn't, Baroque standard, and is a little hard to come by, even today.
Kirkpatrick used on his recording an unspecified instrument by Rainer Schütze (likely a Flemish model of sorts with a good measure of Schütze himself in it). His registration in this piece is two 8' (normal pitch) stops and a 4' stop (an octave higher), all sounding together (so, no variation in registration). The octave stop results in an overall squeakier sound than Ross's recording.
Ross played this particular sonata on a French-style harpsichord made by Anthony Sidey, using a registration of just two 8' stops, no octave. But this instrument appears to be rather strongly voiced (i.e. the plectra are quite sturdy) so it carries a bit of a punch even without the octave.
posted by Namlit at 12:14 PM on November 10 [1 favorite]
There's a difference in pitch that may be confusing to the listener. Kirkpatrick plays in modern pitch, Scott Ross in 415Hz. a semitone lower.
The recordings were made more than ten years apart (Kirkpatrick 1971, Ross 1984-85) which, in the context of the history of recording harpsichords in the second half of the 20th century, actually matters. Technicians started to go further away from the instrument as time went on, creating a less harsh and 'overtony', more natural sound. Ross's recording is still too direct to my taste, but it sounds more like the real thing.
The instruments: for this sonata a keyboard up to g''' is needed, which isn't, and wasn't, Baroque standard, and is a little hard to come by, even today.
Kirkpatrick used on his recording an unspecified instrument by Rainer Schütze (likely a Flemish model of sorts with a good measure of Schütze himself in it). His registration in this piece is two 8' (normal pitch) stops and a 4' stop (an octave higher), all sounding together (so, no variation in registration). The octave stop results in an overall squeakier sound than Ross's recording.
Ross played this particular sonata on a French-style harpsichord made by Anthony Sidey, using a registration of just two 8' stops, no octave. But this instrument appears to be rather strongly voiced (i.e. the plectra are quite sturdy) so it carries a bit of a punch even without the octave.
posted by Namlit at 12:14 PM on November 10 [1 favorite]
You are not logged in, either login or create an account to post comments
2) I just listened to both versions a few times. Ross's performance is a little slower and a little steadier with the tempo. Kirkpatrick's is both faster and looser.
I have two possible theories to explain what you are hearing:
1) The measures in question (31 and 38) are hard. Kirkpatrick is just a tiny bit clumsy navigating them.
2) This section is actually a 7 bar phrase (starting in measure 25 and repeating in measure 32). If you're counting the measures but not the beats, which would very understandable given the tempo, you're noticing that the phrase ended a bar earlier than our brains would expect. Once it it repeats you're back on track. You notice it in Kirkpatrick more because the tempo is faster.
posted by lownote at 4:48 AM on November 9 [2 favorites]