André Cluytens recorded 'Les Contes d'Hoffmann'
twice. The first time was in 1948 with a cast of French singers
in a production based on the Opéra-Comique. This second
recording dates from the 1960s and is a deliberate attempt to
create a star vehicle. Both versions used the Choudens version
with Guiraud recitatives. Neither version attempts to present
the work with spoken dialogue or to explore what Offenbach originally
wrote. But there remains a very big difference between them, and
this later one is perpetually having to compete with its older
sister. Where the older recording scores is that it presents a
fine array of the leading French singers of the 1940s. Almost
uniquely, it is a recording of the work with an idiomatic French
sound and this is something that few subsequent recordings have
captured. The cast on this later recording is more international
and no matter how idiomatic their French, something is lost.
'Les Contes d'Hoffmann' suffers from similar
textual problems to Bizet's 'Carmen' and for a similar reason
- the composer's death. In fact the librettists of 'Carmen' also
wrote libretti for Offenbach's operettas (though not 'Les Contes
d'Hoffmann') and both operas were written for the Opéra
Comique, so spoken dialogue was essential even if the works are
of a more serious bent. And both libretti are shot through with
stock Opéra Comique characters such as students, smugglers
and the like. But 'Carmen' has a distinct textual advantage over
'Les Contes d'Hoffmann'. Bizet survived to see the opera through
to performance and the vocal score through to publication. Offenbach
did not. Granted he did complete a vocal score and Michael Kaye
and others have done sterling work in reconstructing his lost
original. But just as Bizet made numerous changes in rehearsal,
so Offenbach was wont to alter and cut during the rehearsal process.
So his vocal score cannot be regarded as the definitive opera.
It would surely have changed had he lived. 'Les Contes d'Hoffmann'
and 'Carmen' have another thing in common, problems with length.
The composers' first thoughts amounted in each case to a long
opera. 'Carmen' was cut during the rehearsal process and still
ran to nearly three hours (not including intervals) on its opening
night. 'Les Contes d'Hoffmann' has a similar problem and the tendency
to restore material cut or lost simply makes the opera even longer.
The version used here is basically someone's early 20th
Century pragmatic attempt to create a coherent piece. It is not
what the composer wrote but, to a certain extent, it works. Where
this traditional version falls down is in the use of sung recitative.
It makes the work easier for non-French speakers, but slows the
whole opera down, making Offenbach's speedy prologue into a sluggish
Act I.
The problem with the recitatives is that they
make the whole piece seem sluggish. This is particularly true
in Act I where the students can easily outstay their welcome.
After all this is only a prologue to the main story. On this recording
Act I takes 30.34 minutes whereas in Cluytens’ earlier recording,
which uses substantially the same edition, this act takes 26 minutes.
The Bonynge recording Act I (called the Prologue there) takes
only 25 minutes. It is a pleasure to turn to either of these recordings.
This recording uses three singers and an actress
for the heroines and three singers for the four villains. George
London sings two of the villains. As Lindorf, Nicola Ghiuselev
sounds very Russian with a focused but rather curdled tone. He
lends a suitable darkness to Lindorf, but suffers some strain
in his opening aria. I never really felt that he managed to endue
the character with a sense of evil. He is also rather at a loss
in the little satirical duet with Hoffmann, where Nicolai Gedda
judges Hoffmann's tone nicely. For Bonynge, Gabriel Bacquier demonstrates
just how Lindorf should be sung. Bringing a wonderful French style
to the role, he makes the music count and hides well any sense
of strain. And he goes on to sing all of the villains.
Act 1 suits Gedda very well as his tone lends
credibility to the character's youth. But in Act 2, in his opening
Romance, where he needs to open up in the upper register, his
voice fails to respond with the requisite warmth and vibrancy
of tone. Gedda, of course, never sounds strained and his Hoffmann
remains so stylish. But Placido Domingo, despite his wayward French,
sings the role for Bonynge with such burnished tone that it is
difficult not to be seduced.
As Spalanzani, Michel Sénéchal
is a model of style and I only wish he had more to do. This also
goes for Jean-Christophe as a baritone Niklausse. He delivers
Niklausse's little song in Act I with style and wit, one only
wishes he had been allocated a proper baritone part.
As Coppelius, George London sounds fuzzy and
rather unfocused. Neither he nor Ghiuselev sound completely at
home in the idiom and certainly London fails to make the passage
work in Coppelius's aria count for much. He also sounds rather
strained.
Gianna d'Angelo is no more than adequate as Olympia..
Certainly her roulades are passable, if a little smudged, but
she is not in sufficient control to make them count dramatically
and musically, the way Sutherland does. And d'Angelo does not
ornament Olympia's second verse. This may be authentic, but for
those of us brought up on Sutherland's Olympia, this seems a missed
opportunity.
The Venice act (performed second) highlights
the peculiarity of having Niklausse sung by baritone, Jean-Christoph
Benoit. No explanation is given in the booklet for this curious
event. Some adjustment of the text has to be made and the Barcarolle
has to be sung by another (mezzo-soprano) singer, though in fact
this may be more authentic.
As Giulietta, Elisabeth Schwarzkopf makes a very
aristocratic courtesan. Sounding disdainful rather than smouldering
she remains, nevertheless, very stylish. And what she actually
sings, is of course, very beautiful. After all Schwarzkopf never
made a bad recording.
The rather darker Hoffmann of this act does not
entirely suit Gedda's voice type, but he remains stylish and convincing
dramatically. As Dapertutto, Ernest Blanc is very stylish, displaying
a fatal charm and in 'Scintille Diamant' he copes with the taxing
tessitura confidently. He makes a most fascinating villain and
I only wish that he had been allowed to sing more (or even all)
of them.
In Act 4, Victoria de Los Angeles does not completely
convince as Antonia. Her interpretation is convincingly youthful
and girlish, but she does rather show some strain in the upper
register. Still, she is a fine performer and her duets with Gedda
are some of the loveliest things on the record as well as being
convincingly idiomatic.
Rather than having the 4th villain
sung by another singer, Dr. Miracle is played by George London
who is no more convincing in this role than as Coppelius. Though
the role seems to suit him better, his performance remains effortful.
So the climactic trio with Miracle, Antonia and her mother, is
something of a disappointment with the strain showing, in different
ways, on both de Los Angeles and London.
In the Epilogue, the role of the Muse is spoken
by a French actress and she does sound wonderful. Being as this
is the traditional Choudens edition, the role of the Muse is reduced
to a minimum. This has the effect of splitting the role off completely
from Niklausse which reduces, even further, the sense of the plot.
Giving the role of Niklausse to a baritone has
the effect of making the vocal texture of the opera a little bottom
heavy. It means that the heroine is usually the only female voice
in each act. So inevitably, on this recording, the spotlight goes
onto the three heroines. But splitting the soprano role into three
has the effect of over spotlighting the individual sopranos. With
only one role to sing there is no possibility of appreciating
other aspects of their approach. It does highlight the weakness
of each performance though both de Los Angeles and Schwarzkopf
do contribute much to delight.
Nicolai Gedda's performance in the title role
is a tour de force. He is about the only foreign singer on this
recording to sing in anything like an idiomatic French manner.
He has a good feel for the style of the piece, but I must confess
that I would have preferred a rather richer tenor voice, with
a little more depth to it. But that is being rather hyper-critical.
Whereas the heroines and the villains are split
between singers, Jacques Loreau is the only one who doubles roles
in the correct manner and he sings all the character tenor parts
brilliantly. All the smaller roles are sung idiomatically by French
singers. But instead of providing a secure background for the
international singers, they do seem to rather show them up.
When looking at other recordings of the opera,
length, pacing and edition become a very big issue. Cambreling's
performance of the complete Oeser edition (which adds music from
other Offenbach pieces) has a mammoth 214 minutes of music. The
two recordings based on Michael Kaye's critical edition both have
nearly 3 hours of music. Jeffrey Tate's now deleted recording
uses as much of Kaye's edition as was available when the recording
was made and uses spoken dialogue. Whereas Kent Nagano uses Kaye's
complete edition preserving the recitatives. He balances this
by taking quite a brisk view of the tempi and the pacing of the
piece thus does not suffer too much. Of course each of these recordings
offers a different selection of items and some include extras
as appendixes. Whether Offenbach would have wanted the opera to
go anywhere near the 3 hour barrier is a moot point, lovely though
some of the new music is. There is probably still room for a good
new recording, with dialogue, taking into account Michael Kaye's
discoveries and with a single soprano as all the heroines. This
is especially important if the role of Stella is expanded in the
Epilogue. Until that happens, I shall continue to return to Sutherland
and Bonynge, and the older Cluytens for its unique sense of style.
Robert Hugill
see also review
by Tony Haywood