Britten, Physics and Phonics

While I was at the International Congress of Voice Teachers this past summer, I happened upon a seminar given by an old friend of mine, Dr. Kevin Hanrahan, who is Associate Professor of Voice at the Glenn Korff School of Music of the University of Nebraska-Lincoln, and Matthew Clegg, a DMA voice student at the same university. The seminar was entitled Benjamin Britten: A Study in Vocal Acoustics, and focused on Britten's Serenade for Tenor, Horn and Strings, and A Charm of Lullabies for mezzo-soprano, looking into how the composer set the text, to see if the singer simply sang exactly as the text (specifically the vowels) as composed whether it would match the dynamics and expression that Britten intended. It was so fascinating that I asked him to share some of their findings with us!

Q: When you set up this experiment, what were you trying to prove/discover?

A: After working with Kenneth Bozeman at a Nebraska NATS workshop where he discussed what he calls, "Practical Vocal Acoustics", I had the idea that perhaps Britten was using some of the principles that Bozeman was discussing in the workshop and in his book, Practical Vocal Acoustics.  So, what we were looking to discover was whether or not Britten set the text in such a way as to exploit “passive vowel modification” strategies resulting in a timbre that would coincided with the dynamic making, thus making the music vocally and expressively easier for the singer.  Passive vowel modification differs from Active Vowel modification in that passive vowel modification results from a change of pitch resulting in a harmonic/formant interaction without a change of vowel shape, where active vowel modification is when there is a change in vowel shape so as to create a harmonic/formant interaction regardless of pitch.  The main difference is that the singer has to "actively" change or modify their vowel as opposed to "passively" letting the vowel modify as a result of the pitch. In the end it is a matter of whether you move the pitch (passive vowel modification) or move the formant (active vowel modification).

Q: What made you choose Benjamin Britten as the example?

A: Both Matthew and I are tenors, and we both have a fondness for Britten and his music.  One of the reasons we love Britten so much is that we find it to be so easy to sing from a vocal standpoint, and so easy to sing expressively. Knowing that he was very particular about how he composed his vocal works, and that he composed them with specific individuals in mind, we thought he was a natural choice to explore the relationship between the timbres resulting from passive vowel modification and dynamics.

Q: How can we use physics to prove this?

A: Physics doesn’t necessarily “prove” anything, but it does help us explain what we observe.  What we now know is that for "covering" to occur the second harmonic (H2 or 2F0) has to rise above the first vowel formant (F1) or F1 has to be lowered below H2 usually by lengthening the vocal tract which can occur from depressing the larynx, puckering the lips, changing a vowel shape, etc.  This passing of the H2 through F1 is what Bozeman calls "Closed" Timbre. So, using [a], which has a F1 at approximately G5 (middle C is C4), as an example, if one were to sing a tenor "high A" or A4 on an [a] vowel one would only need to maintain the [a] vowel shape to achieve "cover", the crossing of H2 (A5) above F1 (G5).  If one wanted to actively modify by puckering slightly, this would then shift F1 to a lower pitch, perhaps (E5 or F5) and increase the "Closed" or "Covered" quality of the sound. If F1 is continually lowered to where it is just above or in tune with H1, the fundamental, then a new quality takes over, what Bozeman calls “Whoop”.  “Whoop” is a very fluty, light quality that is often associated with soft dynamics or falsetto sounds.  In contrast to “Whoop” is a quality Bozeman calls “Yell” which occurs when F1 is tuned to H2, and as its name suggests is associated with aggressive and loud dynamics.  The last timbre that Bozeman labels is “Open” which also has an association with loud dynamics, and occurs when F1 is above H2.  Now these qualities can be associated with different laryngeal modes, for example, Mode 1 is associate with “Open”, “Close”, and “Yell” timbres, where Mode 2 is associate with “Whoop”.  Looking at the research of Fabiani and Friberg, their study looked at the influence of pitch, loudness, and timbre on instrument dynamics.  They report that dynamics can even be recognized if the loudness (not intensity) is the same, therefore loudness can only partially explain dynamic recognition. They argue that timbre is that missing factor. So, if you put these pieces together then, you have timbres that coincide with dynamics via established vocal registers or modes.  Using the F1/H2 interaction, you can then predict what the likely resulting timbre or vocal register/mode is based on the vowel and pitch, and therefore, the likely resulting dynamic.  If the predicted dynamic matches the dynamic indicated by Britten, then the expressiveness of the singing would be more naturally and easily accomplished.  Furthermore, if the timbre, dynamic, and pitch corresponded with the most conducive vocal register or mode, it would also be vocally easier to sing.

Q: How did you set the parameters?

A: For the timbre prediction we used Kenneth Bozeman’s book, Practical Vocal Acoustics.  To determine the vowels for the text we used an online IPA generator set to American English.  We did this because Received British English tended to favor more mixed vowels and not the purer, if you can call American vowels pure, American vowels.  For example, the word “blow” in American English is [bloʊ] but in British English is [bləʊ]. We confirmed this by listening to British tenors who were well-known for their interpretation of Britten’s music.  When the transcription software didn’t produce a transcription, we looked up the word in the Oxford English Dictionary.  Following standard classical diction rules, i.e. Madeleine Marshall’s Singer’s Manual of English Diction, vowel and “r” combinations were all muted, i.e. changed to [ə].  For the formant pitches we adapted Scott McCoy’s suggested pitches in conjunction with Lindsey’s combination of Well’s and Jones’ vowel chart.  To determine when one timbre was truly different from another, we looked at research about pitch perception and harmonics. We determined that a plus or minus of a major 2^nd was sufficient to properly identify a timbre.  For example, if F1 was C5, then if H2 was between Bb4 and D5, we would conclude that H2 was tuned to F1 resulting in “Yell” timbre. Finally, we associated the timbres with dynamics using established descriptors for the vocal registers, and dynamics shaping was approximated by increasing or decreasing the dynamic level one level, i.e. p to mp, every two beats.  The predicted dynamic shaping was compared to a recording of a well-known tenor and mirrored the tenor’s dynamic shaping quite well.

Q: How can you tell whether the vowels and the pitches are a good match?

A: We assigned various dynamics to the timbres, so pp-p was “Whoop”, mp-f was “Closed”, f-ff was “Open” and f-fff was “Yell”.  Then if the predicted timbre matched the marked dynamic, it was a good match.  For example, if the dynamic was forte, then “Closed”, “Open” or “Yell” were considered to be a good match. Then, if the timbre was used in what would be considered an appropriate pitch range, so for a tenor one would expect “Closed” around G4 or higher, that further strengthened the match. We also looked at groups of dynamics/timbres, meaning how often was “Whoop” used correctly with soft dynamics or incorrectly with loud dynamics. What we are doing now is looking a little more carefully at when there wasn’t a good match or when the dynamic isn’t used correctly with the timbre to see if there is a reason why Britten may have chosen to mismatch the timbre and dynamic.  At the moment it looks like this mismatch is occurring to aid projection. 

Q: What was the upshot of the experiment?

A: What we found was that overall, the predicted timbre matched Britten’s dynamic only 25% of the time.  There were occasions, however, when the predicted timbre matched the dynamic as much as 50%, but it depended on the individual song.  Furthermore, we found that Britten often assigned the “correct” vocal register or mode for the dynamic when in the upper portion of the voice.  We also looked at occasions when F2 happened to correspond to H3 or H4.  The supplemental amplification of F2 is a common technique for tenors, and has the effect of aiding diction clarity.  We found the F2 supplemental tuning was occurring over 37% of the time.  It was also frequently being used in the upper portion of the voice and when the dynamic was soft in conjunction with louder timbres which we interpret as being Britten’s attempt to aid the singer in projection at soft dynamic levels.  Finally, we noticed that when the predicted timbre and dynamic did not match, the timbre shifts mirrored the dynamic shifts, i.e. a shift from “Closed” to “Open” would correspond to a dynamic change of p to mp, etc., as if Britten were cueing the singer as to how a line should be shaped.

Q: How is this useful for singers? Should it change their approach to the text/performance?

A: The big take away for singers here is that with Britten, if you just sing the pitch and the vowel he asks, you will likely sing the dynamic or dynamic shaping he wants.  In other words, Britten has programmed by his choice of pitch and vowel, how to sing his music.  We think this type of analysis would be beneficial to singers, at least on a macro level.  Both of us teach singing, and we frequently see singers making their lives more difficult by imposing an interpretation on a piece.  If they would just take a moment and consider what might happen if they just sing what is on the page, they may find that they don’t need to work as hard to provide a musically pleasing product.  Of course, this is not always true, but it would seem logical, at the very least, to begin from that point of view.  We think this type of analysis also has a tremendous impact on the practice of transposing art songs, particularly. It has become very popular for women to sing song repertoire that was originally composed for men and vice versa. If you have a composer, like Britten, who was very specific about what colors or timbres they wanted from the singer and was also adept at building that into their composition, transposing them or simply singing them up or down an octave would destroy the composer’s artistic concept, not to mention making it very difficult vocally to sing the music.  For example, in Britten’s “Nocturne” from Serenade, he has very carefully chosen his pitch/vowel/dynamic combinations, a 50% match and 64% match above C4.  To have that piece sung by soprano an octave higher would completely destroy those combinations resulting in either the soprano not accomplishing what Britten has asked for or accomplishing it with great difficulty and vocal manipulation. It is really wise to do that?  With some composers, who are less intentional and adept, this may not be an issue, but it is important to consider.  So, we would say that this analysis definitely is useful in aiding interpretation, and vital when considering singing repertoire in transposition.

**********

In case we needed another reminder that singing the text and the proper vowels is important, here we have it! 

--Ellen