Chapter 4: Early recording analysis and its relevance to understanding portamento in Schubert's sound world and today

Introduction

The earliest recordings of singers and instrumentalists at the turn of the 20^th century can give us great insight into the expressive techniques and styles of the second half of the 19^th century and even earlier to Schubert’s time. Many of the first recorded musicians were middle-aged or retired when recording at the start of the 1900s, so both their education and performing career took place during the mid-19^th century and most were thus closely linked to musicians in Schubert’s circle or his contemporaries. Therefore, it is interesting and important to understand their expressive language in order to gain better insight into both the historical evidence of portamento use and how to incorporate portamento into modern-day performance.

For this study I analysed 14 vocal recordings and 7 violin recordings, recorded between 1902 and 1920. For the purposes of this study, as recording analysis is time-consuming work, I focused my analysis on recordings of singers and violinists who were most relevant and connected to Schubert and the early-19^th-century sound world. Regarding the 14 vocal recordings, I analysed both the original vocal recordings and Sarah Potter’s emulations of these recordings and used spectrogram images of Potter’s emulations instead of the original vocal recordings, as the recording quality of Potter’s recordings are far higher than the original recordings and the portamento shapes, speeds and placements in her emulation recordings matched the originals so closely that her emulations are a strong representation of historical portamento execution. Throughout this chapter I include spectrogram images of Potter’s emulations of portamenti in order to illustrate aspects of portamento types and include excerpts of each of the original audio recordings of each singer for historical reference. I analysed more singers than violinists, as the access to Potter’s emulations compared with the original recordings provided more detailed information about vocal portamento execution, thus making it easier to judge the nature of portamenti in each vocal recording.1 Furthermore, we have recordings of multiple singers performing the same piece, which makes comparison of style and technique easier, and allows for assessment of variation between and within pieces.

Early recordings from the turn of the 20^th century have been analysed by some of the leading 19^th-century performance practice researchers, such as Clive Brown, Neal Peres da Costa, Anna Scott, David Milsom, Sarah Potter, Kai Köpp, John Potter, Emlyn Stam and Daniel Leech-Wilkinson. These scholars have used recording analysis techniques and software such as Sonic Visualiser to investigate the expressive devices of 19^th-century performance practices, such as tempo flexibility, deviation from the score, expressive vibrato, tempo rubato and portamento. However, so far none have gone into a great deal of detail when looking specifically into early-recorded portamento use. While Potter (2014) annotated scores of early-recorded singers and marked some portamenti, she did not categorise the different types heard and even left out many altogether.2 Furthermore, Köpp (2015) used Sonic Visualiser to create a list of the types of portamento he has heard in early-recorded singers and instrumentalists that relate to 19^th-century written sources,3 which Stam is currently developing in more detail in his forthcoming PhD thesis.4 Stam is comparing early-recorded violists and singers in detail and notes that they share a common stylistic language and way of producing portamento, but that singers use a greater range of portamento types, as violists (and string players in general) have limited fingering choices and string crossings that impact portamento execution.5 However, despite this more detailed analysis and categorisation of early-recorded portamento types, the categories that both Köpp and Stam developed unfortunately do not account for all types heard in the recordings I examined.

Therefore, in order to gain a greater understanding of the varied portamento use in early vocal and violin recordings, for each recording I use Sonic Visualiser software to collect and analyse data for the following aspects of performance (full data can be seen in Appendix 4.1):6

• The types of vocal and violin portamento used in each piece, including total number of portamenti in relation to the number of bars of singing/playing, the percentage of total ascending compared to descending portamenti in each piece, and the most frequent types of portamento used by singers and violinists

• Spectrogram images to show the shape, rate of change of pitch, and frequency and density of sound for different types of portamento

• Types and amount of portamenti present in relation to character of piece/section

• Tempo graphs of most pieces to show use of rubato and tempo fluctuation

Based on these findings, I then compare portamento use in early recordings with the historical written documentation of the early-19^th century in order to explore both the relevance of analysis of early recordings when understanding Schubert’s sound world, as well as the unique expressive qualities of each performer. Many 19^th-century performance practice researchers, including Scott, Peres da Costa, Stam, Potter and Leech-Wilkinson, argue that performers in the 19^th- and early-20^th-centuries had a spontaneous and unique way of singing and playing. They have all found substantial variation in use of expressive devices, but agree that each performer used a common expressive language.7 Scott even argues that the “imperfections” of “style are themselves research outcomes.”8 These researchers have also noted that, in order to fully understand each specific aspect of 19^th-century performance practice, including portamento, it is essential to analyse all aspects and their relationships to each other, as these expressive devices were interconnected and part of an overall expressive language. Therefore, I use Sonic Visualiser to analyse sonic patterns of not only the various portamento types used in early recorded singers and violinists, but also other aspects of 19^th-century performance practices, such as rubato and rhythmic flexibility, expressive vibrato, and legato singing and playing styles, and their relationships to portamento use.

Methodological approach

There are a number of technical aspects, both positive and negative, to note concerning the use of Sonic Visualiser. It is important to understand the capacity and constraints of the technology, so that appropriate inferences can be made concerning its outputs.

• Positive aspects of Sonic Visualiser include:

  • Visual representation of audio, such that small details that the ear cannot hear can be represented graphically. This is especially important if early recordings suffer from a lot of extra noise

  • The dynamics of each note can be represented visually, by variation in colour (usually redness) and density of the spectrogram

  • The width and speed of vibrato can be measured accurately

  • The shape of portamenti and rate of change of speed and pitch of each slide can be quantified

  • The recordings can be slowed down considerably, such that the operator can hear subtle or fast slides that are difficult to hear, and the specific moment when bows or syllables change

  • The frequency range of vibrato and portamento can be measured (as measured in standard physical units: Hz)

  • The beat placement, and thus overall tempo fluctuation in a tempo graph, can be indicated

• There are, however, some negative aspects, which have required an adaptive approach. These include:

  • With early recordings, both the audio and visual can be distorted as a consequence of extra sounds from the recording equipment. This is more pronounced in the violin recordings than in the vocal recordings. (Patti’s recordings are best for spectrogram analysis, as she sang very purely, and the recording quality was relatively high)

  • Therefore, while I look at the shape and mathematical qualities of slides, I do not measure the exact rate of change of speed of each portamento as it is too difficult to get an accurate reading of each frequency- and time-point for each point on the graph, due to the poor recording quality

  • When there is melodic dislocation between players, making decisions about where to put beats is difficult, and is therefore difficult to notate the rubato use

    • Therefore, I have made educated estimates for beat placement in each piece that gives a general idea of fluctuation in tempo flexibility

  • Sometimes it is difficult to see and/or hear when bows change, but usually if the recording is slowed to half-speed then an accurate analysis can be made

    • Therefore, I have made some educated estimates in counting different portamento types when bow changes were ambiguous in both the audio and visual representations of recordings

• General:

  • I have judged obvious versus subtle portamenti based on the relative audibility of each portamento within each piece

  • When marking the changing fingers on/sliding to same note portamento, I put these in the ascending and descending categories depending on whether they slide into the next note or away from the old note

  • For each spectrogram image used in this Chapter, I have indicated beat placement with a vertical purple line and the amount of time between each beat is shown by a white horizontal or diagonal line connecting the purple lines, where horizontal means no tempo fluctuation, diagonal ascending white lines indicate acceleration, and diagonal descending white lines indicate rallentando

  • In the spectrogram images, the degree of redness of the fundamental pitch does not necessarily give an accurate reading for dynamic or air/bow pressure as it is often distorted by the sound of the piano. Hence, it is best to look at the density of green lines in the upper partials for each soloist to judge air and bow pressure and dynamic more accurately

• Sonic Visualiser also allows the user to describe the shape of the portamento curves mathematically. To this end I:

  • Compare each portamento slide shape to their equivalent mathematical graph shapes, such as linear (in mathematical terms, positive: y=x, or negative: y=-x), exponential growth curves (in mathematical terms, positive: y=e^x, negative: y=-e^x) or exponential decay curves (in mathematical terms, positive: y=e^-x, negative: y=-e^-x). Note that while Leech-Wilkinson describes portamento curves as “parabolic,”9 I believe that linear or exponential curves are more suited to the different portamento curves I analyse using Sonic Visualiser. Parabolic shape is not necessarily wrong, but parabolas can drastically vary in shape, whereas exponential and linear curves more accurately match the rate of change of pitch and speed in different portamenti

  • Provide example plot images of these cure shapes in Appendix 4.2, illustrating the basic outputs of the software 

• Scientific Pitch Notation:

  • I use Scientific Pitch Notation (SPN) to describe pitches in each example

Descriptions of recordings analysed

Singers

I have chosen to analyse recordings by the following singers: Adelina Patti (1843-1919), Nellie Melba (1861-1931), Amelita Galli-Curci (1882-1963), Blanche Marchesi (1863-1940), Emma Albani (1847-1930), Marcella Sembrich (1858-1935), Suzanne Adams (1872-1953), and Gustav Walter (1834-1910). Table 4.1 below lists the vocal recordings examined, using both the original recordings and Potter's emulations for reference. These singers were born between 1834 and 1882 so the majority were educated by artists closely related to Schubert’s time and their performing careers were during the second half of the 19^th century. Leech-Wilkinson argues that a performer’s style solidifies in the first two decades of their life, which would mean that the older singers, such as Patti, Melba, Albani and Walter, would have been greatly influenced by the performance styles of the 1840s-1860s, which is closely related to Schubert’s sound world.10

Of the 14 vocal recordings, I have chosen to analyse 7 more specifically in order to create equal same sizes to compare with the 7 violin recordings. These are marked in yellow in Table 4.1 below. Results from the full analysis are presented in Appendix 4.1; the major findings are summarised below.

Table 4.1: List of vocal recordings analysed. Yellow highlighted rows indicate the subset of seven recordings analysed. See Appendix 4.4 for complete recordings of each piece.11

Character descriptions of vocal recordings

Mozart: Voi, che sapete sung by Patti (1905), Melba (1907) and Melba (1910)

In the second act of Le Nozze di Figaro Cherubino (a male role traditionally played by a female soprano) enters and is asked to sing the song he wrote for the Countess before being dressed up as a woman to lure the Count to confirm his infidelity. The aria laments his confusion about his lustful habits and uncontrollable emotions and is a charming and somewhat playful aria. Both Patti and Melba (in both her recordings) use portamento throughout the piece but in different ways. Patti uses the most portamenti (61 in 71 bars of singing) while Melba uses 24 in 71 bars of singing in both recordings. Both singers favour sliding during syllables under slurs and before changing syllables, but Patti does more in general and also does more portamenti after the destination syllable. Each recording has different portamenti in different places, which shows the unique expressive voice of each performer, and that no two performances were the same.

Bishop: Home, Sweet Home sung by Melba (1905), Patti (1905) and Galli-Curci (1917)

This song was adapted from the American actor and dramatist John Howard Payne’s 1823 opera, Clari, or The Maid of Milan. The song’s melody became extremely popular and widespread and was quoted frequently by other composers throughout the 19^th century. This sentimental song tells of the pleasures of a humble home, concluding, “there’s no place like home!” Consistent with written 19^th-century suggestions of frequent portamento use in slower sentimental pieces, each recording uses many portamenti throughout the piece (at least one per bar on average), with Melba using 57 in 49 bars of singing, Patti using 65 in 50 bars of singing, and Galli-Curci using 57 in 50 bars of singing. The most frequent type of portamento used is the slide before syllable change but like in the Mozart, there is varied use of different types of portamento throughout each recording, which enhances the overall expressive effect of each recording.

Mozart: Batti, batti sung by Patti (1904) and Sembrich (1904)

Zerlina’s aria in Mozart’s opera, Don Giovanni, begins with Zerlina seducing Masetto by teasing him to hit (batti) and abuse her in a child-like, yet plotting, manner, and, as he falls under her spell, she rejoices in a lively and celebratory 6/8 dance singing pace (peace). Portamento use in this aria by both singers is varied and expressive, with slower portamenti emphasising both the vulnerability and deception of the first section, and faster types enhancing the joyful and playful celebration of the 6/8 section. Patti uses a total of 55 portamenti in 98 bars of singing, while Sembrich only uses 12 in 100 bars of singing, using her portamenti at important moments, rather than as a fundamental aspect of her singing, like Patti. Having studied and sung this piece in my late teens, I can relate to Patti’s and Sembrich’s different uses of expressive portamento in order to enhance both the teasing plotting first section, as well as the joyful 6/8 dance section. While my singing teacher encouraged me to practice this aria by singing only vowels and sliding between each note to develop a legato singing technique, the only portamento I was “allowed” to use, however, was a subtle one at the end of the fermata in bar 61, after a scale-like ornamentation to connect the F₅ to C₅ on “core.”

Schubert: Am Meer sung by Walter (1904)

The twelfth Lied from Schubert’s collection of songs, Schwanengesang (D957), this slow, lamenting, troubled song lends itself to dramatic expressive uses of portamenti, which Walter uses frequently. He uses 59 portamenti in 34 bars of singing, mainly on the most tormenting lyrics or expressive notes, which is consistent with Garcia’s suggestion of using portamenti in “passionate” places.12 While I have not personally sung this particular Schubert Lied, I have sung many others during my late teens and undergraduate years, and I always found it difficult and unnatural to suppress my urge to use some sort of portamento when connecting notes throughout each song, especially when singing Lieder by Schubert. However, while I did use a legato singing technique, doing similar exercises as those used when practising the Mozart aria above, I made sure to eliminate obvious portamenti in my singing in performance, due to my modern singing training and the expectations to fit into a “clean” modern musical aesthetic. Through my current research, I now realise that my natural urge was much like the early-recorded singers I have studied.

Chaminade: L’été sung by Marchesi (1906) and Albani (1914) and Gounod: Jewel Song sung by Adams (1902/3), Sembrich (1906) and Melba (1906)

Both Chaminade’s L’été and Gounod’s Jewel Song are fast lively pieces and therefore do not have as many portamenti present in each recording, compared with the other slower pieces listed above. For the Chaminade, Marchesi uses a total of 19 portamenti in 131 bars of singing, and Albani uses a total of 15 in 65 bars of singing. For the Gounod, Adams uses a total 18 portamenti in 85 bars of singing, Sembrich a total of 22 in 123 bars of singing, and Melba 30 in 123 bars of singing. Although the frequency of use of portamento is significantly less in these faster songs, portamenti are still very present at key expressive and dramatic moments. This shows that portamento was not only limited to slower or plaintive pieces, but to lively ones as well, they were just used in different ways, as will be explored further in the next Chapter. Ironically, while the recordings of these pieces had the least portamenti present, it was via French pieces of similar character to these that I was encouraged as a young singer to include portamenti. From my observations, it seems that today, portamento in modern singing (and, indeed in violin playing) is only acceptable in late-19^th-century French repertoire, and only then occasionally for special effect.

Violinists

Like the singers, the early-recorded violinists I examined have strong links to contemporaries of Schubert, such as Spohr. I analysed recordings by Joseph Joachim (1831-1907), Leopold Auer (1845-1930), Hugo Heermann (1844-1935), Marie Soldat (1863-1955) and Marie Hall (1884-1956). Table 4.2 below lists the violin recordings examined. Apart from Hall, these violinists were born between 1831 and 1863 so presumably had similar stylistic influences, from the 1830s-1850s, to the singers examined.

Both Joachim and Soldat are highly significant figures to analyse when exploring 19^th-century performance practices. Joachim was a student of Joseph Böhm (1795-1876) and Ferdinand David (1810-1873), and both were closely connected to Schubert through Spohr—with Böhm having premiered Schubert’s Piano Trio in E flat in 1828. Joachim wrote his own method of violin playing with Andreas Moser (1859-1925) in 1905 (which praised the use of portamento and imitating the voice) and was close friends with the Schumanns and Brahms from the 1850s.13 Joachim and Augustus Pott (a student of Spohr’s) taught Soldat.14 This makes both recordings of Soldat interesting, given her connection through pedagogical lineage to both Spohr and Schumann (through Joachim). Soldat’s recording of Spohr’s Violin Concerto No. 9: Adagio was seen as “not only the most convincing performance but [was] particularly valuable historically because she studied this concerto with a pupil of the composer.”15 She was also seen as “represent[ing] the Joachim school at its best period, and [was] imbued with all the traditions of the great classical school.”16 Furthermore, Milsom argues that Soldat’s adaptations of Sophr’s markings “seem entirely congruent with what we can discern of German nineteenth-century performance from other sources, such as annotated editions.”17 Heermann also studied with Joachim and was the first violinist to premier Brahms’s Violin Concerto in Paris, New York City and Australia. Hall is the youngest violinist examined and is not directly related to Schubert, but she was still a highly-regarded violinist in the early-20^th century and her recordings give us insight into the later uses of portamento in violin playing, compared to the younger recorded violinists examined.18 Her recordings also show that early-19^th-century performance practices were still being employed by younger players in the first two decades of the 20^th century.

Table 4.2 List of the seven violin recordings analysed. See Appendix 4.4 for complete recordings of each piece. 19

Character descriptions of violin recordings

Brahms: Hungarian Dance No. 1 played by Joachim (1903)

This folk-inspired show piece alternates between a lilting melodic section and a lively virtuosic section with fast runs and double-stops. Joachim uses a total of 27 portamenti in 167 bars of playing and uses the majority of these in the slower lilting sections.

Joachim: Romanze in C played by Joachim (1903)

While Joachim’s Romanze in C is not a particularly slow piece, it has a beautiful expressive melody that lends itself well to portamento use. Joachim’s recording has a total of 80 obvious portamenti in 170 bars of playing and he also uses the fundamental Ferdinand David shifting technique of audible guide-note shifting which creates connection of notes throughout, even when expressive portamento are not present.

Tchaikovsky (arr. Wilhemj): Melodie from Souvenir d’un lieu cher played by Auer (1920)

Tchaikovsky’s Melodie is slow and nostalgic in character and Auer’s recording is heavily laden with portamenti; he uses a total of 65 portamento in 81 bars of playing.

Ernst: Nocturne in E played by Heermann (c1910)

Ernst’s Nocturne in E begins as a typical lullaby piece that lilts the listener into a calm slumbering state, but as the piece builds the violin part becomes denser and more dramatic with many cadenzas, which are perfect opportunities for free use of tempo rubato and portamenti. Unsurprisingly, Heermann uses a total of 51 portamenti in 39 bars of playing.

Spohr: Violin Concerto, Op. 55, No. 9: Adagio played by Soldat (1920)

The Adagio movement of Spohr’s Violin Concerto No. 9 consists of a combination of serene melodic sections and virtuosic cadenza-like sections, and Soldat uses frequent portamento and melodic dislocation from the piano part throughout. In accordance with these musical expressive characteristics of the piece, Soldat uses a total of 96 portamenti in 97 bars of playing.

Schumann: Abendlied played by Soldat (1920)

Schumann’s Abendlied (Night Song) is a very slow tranquil piece that has many expressive intervals that invite portamento use. Fittingly, Soldat uses a total of 28 portamenti in 28 bars of playing. In both recordings of Soldat, she favours guide-finger shifting before the bow, and this preference may have been passed down from Spohr himself.

Mendelssohn: Violin Concerto in E minor: Allegretto non troppo played by Hall (1905)

The third movement of Mendelssohn’s Violin Concerto in E minor begins with a plaintive melody with sparse accompaniment in B minor, in which Hall uses portamenti sparingly at expressive moments. The movement then makes a sudden juxtaposition of character, where the orchestra plays a dotted march-like figure in E major that the violin answers with flourishing arpeggiations to high harmonic notes, and then continues in this playful jovial manner until the end of the concerto. Hall uses a total of 30 portamenti in 235 bars of playing, where the majority of portamento use is in the first 14 bars during the slower plaintive section. Hall’s recording of this concerto shows some more subtle expressive uses of portamento and also shows that portamento use was still common among the younger generation of early-recorded violinists in the early-20^th century.

Comparison of recording findings to late-18^th- and early-19^th-century written documentation

Many similarities can be seen between the use of portamento in the early recordings examined in this Chapter and the written documentation relating to Schubert’s era discussed in the previous Chapter. I first discuss the comparison between legato singing and violin playing style, and then make detailed analyses of the portamento types found in early recordings, comparing them to historical written sources.

Legato singing style and 19^th-century connected violin shifting style

Connection between notes, or legato singing style, as the basis of good singing, was well-documented throughout the 18^th and 19^th centuries by singers such as Tosi, Corri, Garcia and Lanza.20 This connection can be seen clearly in every vocal recording examined, as lines between notes can be seen in the spectrogram even when the singer is not doing a portamento. One example can be seen in Fig. 4.1 where Patti does not use portamento between the 3 notes (A₄ – G#₄ – A₄) on “with” in bar 17 of Bishop’s Home, Sweet Home, but we can clearly see that each note is connected.21

Figure 4.1: spectrogram image of connected notes (with linear/constant rate of change of pitch), and corresponding score from Potter’s emulation of Bishop’s Home, Sweet Home sung by Patti (1905), bar 17 (“with”), along with original recording excerpt for reference.22 Note that the vertical purple lines indicate crotchet beat placement and the white descending diagonal line indicates slowing of tempo, and the note names have been added to show changes in pitch in the image. It is also interesting to note Potter's emulation of Patti’s use of sparing vibrato (indicated by the wavy green lines in the upper partials of the spectrogram image that become more pronounced as she sings the first A₄). The orange slur in the annotated score indicates the legato singing.

Legato violin playing can also be heard throughout the early-recorded violinists examined. However, unlike in singing, due to the nature of violin technique, where the left-hand fingers stop the string to change pitch, a true continuous consistent legato cannot be achieved, unless the violinist only ever plays on one string with one finger. Therefore, the early-recorded violinists examined use a larger variety of types of portamento and shifting techniques to create the illusion of consistent smooth legato playing, as we will see throughout the examples below.

Una corda, or keeping passages on one string, allows for consistency of tone, and legato playing, as in legato singing, was a fundamental part of 19^th-century violin technique. Spohr states that, “if two tones lying at a distance are to [be] drawn together in one bowing … the leap from one tone to another cannot be made without the sliding of the hand being heard.”23 This connection of notes can be compared to the legato singing style of the 19^th century, as shifting was most often an audible event. Spectrogram images of the violin recordings discussed throughout this Chapter show that the majority of shifts in early recorded violinists were connected and even through some are not audible through the extraneous sounds of the recording technology, these connected lines in the images show that there was almost always connection between the instrument and the left-hand fingers during shifts. In fact, when the recording is slowed to about half speed, even the fastest shifts are consistently and clearly audibly connected. This shows a clear link between the shifting techniques advised in 19^th-century treatises, such as those of Spohr and David, as shifts were advised to be conducted with guiding notes in order to connect them.24 There is also debate among scholars as to how much early recording technology hides the intricacies of sounds of each performer, such as portamenti and audible shifts. Kai Köpp and his team in Bern are currently re-enacting recordings with period technology in order to judge the audibility of aspects of early recordings, and it may be that shifts and portamenti were much more obvious in live violin playing than is audible in early recordings.x Baillot states in his treatise that violinists chose different fingerings for passages “depending on the feeling they wanted to give to their passages.”25 I have found in early violin recordings that passages were generally played on one string where possible, which increases the use of audible shifting (portamenti). This is consistent with the evidence of una corda fingerings throughout the 19^th century, examples of which can be seen in Chapter 3.

Types of portamento found in early vocal and violin recordings

From my analyses, I have created lists of vocal and violin portamento types heard in the early recordings examined. These have been developed from Köpp and Stam’s lists (see Table 4.3 for Stam’s portamento types list) and are designed to account for all types of portamento present in each recording I have analysed. Köpp and Stam use string playing terminology that does not always relate to singing technique and the intervals indicated for the types in Köpp’s list also do not account for all variants on each type of portamento found in early recordings.26 Therefore, I have gone into more detailed explanation of each type heard in the vocal and violin recordings examined. I have categorised six main types of portamento in the early recorded singers studied (Table 4.4) and six types in the early recorded violinists studied, with sub-categories for the violin types based on shifting-finger and bow-change placement (Table 4.5).27 I have also compared the vocal portamento types to their equivalent violin types (Table 4.6). These all share similarities with the documentation of early-19^th-century portamento use, as discussed in the previous Chapter.

Table 4.3: Emlyn Stam’s list of portamento types found in his analyses of early recorded singers and violists, based on the list made by Köpp.28

Portamento Techniques in 19^th-century string and vocal practice29

• PL: (Portamento Langsam) Sliding with one finger during a slur

(Small intervals up to a perfect fourth according to violinist Louis Spohr 1836 - 1859)

• PS: (Portamento Schnell) Sliding with two different fingers during a slur

(Large intervals of a perfect fifth or greater; Spohr prefers the guide finger rather than the arrival finger)30

• I: (Intonazione) Sliding into the beginning of a phrase

(Small intervals, with the arrival finger)

• C: (Cercar della nota) Sliding with the arrival finger after a bow change

(Small intervals)

• A: (Anticipazione della nota) Sliding with the arrival finger before the bow change

(Small and large intervals)

• L: (Librar la voce) Changing fingers on the same note

(Small intervals)

Table 4.4: My own terms and explanations of portamento types in the early-recorded singers examined. Types marked with ⁺ are pre-existing types noted by Köpp/Stam, types marked with * are my own discoveries and types marked with ^ are my own elaborations on existing types that allow for more detailed comparison.31

Table 4.5: My own terms and explanations of portamento types in the early-recorded violinists examined. Types marked with ⁺ are pre-existing types noted by Köpp/Stam, types marked with * are my own discoveries and types marked with ^ are my own elaborations on existing types that allow for more detailed comparison. I have created main categories with corresponding sub-categories based on shifting-finger placement and bow-change placement.

Table 4.6: My own comparison of vocal portamento types compared to violin types present in the early recordings examined.

Portamento type 1: sliding on one syllable under a slur in singing or same-finger slides under a slur in violin playing

Interestingly, Garcia and de Bériot both concurred that sliding “between two notes connected by the same syllable” was the “best manner of employing portamento,”32 but in the 14 early vocal recordings examined only 17% of all portamenti used were on the same syllable under a slur, whereas 58% were between syllables. Lolli, Haydn, Beethoven and Schubert all indicated same-finger shifts under a slur in their works, which can also be heard in early recordings.33 Six of the seven violin recordings examined had between 1 and 15 same-finger slides under a slur per recording, with Joachim using the most (15) in the recording of his Romanze in C (see Fig. 4.2). Similarly, vocal slides on the same syllable under a slur are illustrated in Fig. 4.26, 4.29 and 4.33 later in the Chapter.

Figure 4.2: spectrogram image of same-finger (4^th-finger) slide under a slur, with corresponding score and recording of bar 39 of Joachim’s Romanze in C played by Joachim (1903), bar 39.34 The portamento is negative (descending) linear (y=-x) with constant rate of pitch change, indicated with a green line in the annotated score.

Vocal portamento types 2 & 3: sliding before or after the destination syllable in singing

While Garcia’s and Bériot’s preference seemed to be for slides under a slur, Bériot also stated that portamenti may be employed between syllables “when the expression requires it.”35 When sliding between syllables, the general consensus from the written sources of the 19^th century seemed to be that sliding before the syllable of the following note was the desired way to execute a portamento, as seen in treatises by Corri, Vaccai, Bayly, Fröhlich, Croker and Garcia. Corri, Vaccai and Garcia explicitly stated that the voice should be carried with the syllable of the initial note, rather than that of the destination.36 This preferred type of portamento can be seen throughout each recording analysed and is by far the most common type of portamento. Of the 14 vocal recordings analysed, I noted a total of 203 portamento where the slide was before the destination syllable, compared with sliding afterwards, of which there were only 92 instances (Fig. 4.3). Only two recordings used more slides after the destination syllable than before, but the difference was not large.37

Figure 4.3: graph comparing number of portamento slides before and after syllable changes in the 14 early vocal recordings examined clearly shows that sliding before a syllable change was the preferred portamento type.38

One clear example of sliding before a syllable change is by Patti in Mozart’s Batti, batti (Fig. 4.4), where we see a clear exponential growth of rate of change of pitch (y=e^x) and a slight overshoot of the destination pitch before settling, i.e. Patti starts the slide slowly and quickens her pitch change as she ascends to the destination note.

Figure 4.4: spectrogram image of sliding before the destination syllable during “-li-” of “agnellina”, with corresponding score from Potter’s emulation of Mozart’s Batti, batti sung by Patti (1904), bar 7 (“-lina”), along with original recording excerpt for reference. The portamento curve is an ascending positive exponential growth (y=e^x), as she starts the pitch ascent slowly and increases rate of change of pitch as she approaches the destination note, indicated with a blue curve in the annotated score. Note that the vertical purple lines indicate crochet beat placement and the white diagonally descending line shows decreasing tempo (declining=slowing).39

Violin portamento types 2.1, 2.2 & 2.3: sliding with the guide finger (under a slur or before or after bow change) in violin playing

Spohr was clear in his preference for shifting with the guide (initial) finger as opposed to the destination finger to avoid “unpleasant howling.”40 This preference can also be heard in early-recorded violinists as I found that 59% of the total portamenti used in the 7 early recorded violinists studied were slides with the guide finger (including under a slur and before and after the bow change) and only 15% were with the destination finger (including under a slur and before and after the bow change). Guide-finger shifting was overwhelmingly the preferred shifting technique across all portamento types (Fig. 4.5). Also, of the guide-finger shifts between bow changes, most were undertaken before the bow change, as indicated in Spohr’s treatise.

Figure 4.5: graph of number of guide-finger portamenti before bow changes as compared with number of destination-finger portamenti after bow changes in the seven early violin recordings examined. This clearly shows that, like in singing, sliding with the guide-finger before a bow change was the preferred portamento type.41

It is particularly interesting to note that in her recording of the Spohr Violin Concerto No. 9: Adagio (1920), Soldat uses the largest amount of the guide-finger types of shifts (70) out of the 7 recordings I examined, which perhaps shows her connection to the great Master himself through her pedagogical lineage (see Fig. 4.6 and 4.7 for examples).

Figure 4.6: spectrogram image of slide with guide finger before the bow change from C₄ to C₅, with corresponding score and recording from Spohr’s Violin Concerto: Adagio played by Soldat (1920), bar 26.42 The portamento is positive linear (y=x) with constant rate of ascending pitch change, indicated with a green line in the annotated score.

Figure 4.7: spectrogram image of slide with guide finger under a slur from F₅ to F₆, with corresponding score and recording from Spohr’s Violin Concerto: Adagio played by Soldat (1920), bar 22.43 The portamento is positive linear (y=x) with constant rate of ascending pitch change and Soldat uses increased bow pressure during the shift (which can be seen in the redness of the upper partials of her notes), as it is marked with a crescendo (<), indicated with a green line in the annotated score.

It is also interesting to note that shifting with a guide finger was present in violin teaching methods throughout the 19^th century and well into the early-20^th century. Similar to Spohr and David’s shifting exercises, The Yost System for Violin by Gaylord Yost (1888-1958) comprises shifting exercises based on the use of a guide note (Fig. 4.8). This shows that shifting with the guide finger was still a fundamental part of violin technique from Schubert’s time and throughout the 20^th century. This technique of shifting is in fact still taught today (I was also taught to use a guide note when shifting between positions with different fingers in my modern violin studies), but, as was the case in my own education, once the guide-tone shifting practice has been completed by the student for left-hand control and intonation accuracy, then the aim of modern violin playing is very much to hide the sound of the shift by releasing left-hand finger and bow pressure during each shift. This outcome is the opposite to the connected and audible shifting techniques I have found in 19^th-century violin treatises.

Figure 4.8: Yost’s advice on the nature of shifting with the guide finger to a guide note before playing the destination note.44

Violin portamento types 3.1, 3.2 & 3.3: sliding with the destination finger (under a slur or before or after bow change) in violin playing

Sliding with the destination finger was generally discouraged during the early-19^th century with only one exception in Spohr’s treatise, where he stated that sliding with the destination finger to a harmonic was acceptable. In my analysis of early-recorded violinists, I found that every shift to a harmonic was made with the destination finger. Again, Soldat carries on Spohr’s shifting technique legacy in Fig. 4.9 below.

Figure 4.9: spectrogram image of slide to harmonic on D₅ with destination finger under a slur, with corresponding score and recording from Spohr’s Violin Concerto: Adagio played by Soldat (1920), bar 7.45 The portamento curve is positive linear (y=x) with constant rate of ascending pitch change, indicated with a green line in the annotated score.

Surprisingly though, given the abundance of written 19^th-century sources advocating for guide-finger shifting,46 various types of destination-finger shifts to a stopped (non-harmonic) note can be heard in most early recordings I examined. This may be due to the “modern French school” creeping into German violin playing,47 but it was still seen as a “perverted mannerism” in 1889 by Hermann Schröder and generally not a tasteful type of portamento in the German school.48 Auer uses the most of this so-called later French style of shifting out of the 7 violin recordings examined (see Fig. 4.10 for an example), which may be because he is playing a late-19^th-century piece (Tchaikovsky’s Souvenir d’un lieu cher, from which Melodie was arranged, was first published in 1879). However, he still uses more guide-finger shifts than destination-finger ones, which is consistent with the written evidence of common shifting styles of the 19^th century.

Figure 4.10: spectrogram image of a destination-finger shift (2^nd-4^th finger) after the bow change from G₅ to D₆, with corresponding score and recording from Tchaikovsky’s Melodie played by Auer (1920), bar 22.49 The portamento curve is an ascending positive exponential growth (y=e^x), as he starts the pitch ascent slowly and increases rate of change of pitch as he approaches the destination note, indicated with a blue curve in the annotated score.

Vocal portamento type 4: Corri’s “anticipation” graces in relation to the anticipation portamento in singing

Similarities to Corri’s “superior” and “inferior” anticipation graces can be heard throughout the vocal recordings analysed (see Fig. 4.11 in Chapter 4 for all of Corri’s graces).

Figure 4.11: Corri’s Graces from page 32 of his method.50

Every singer examined uses the anticipation portamento either before or on the beat at various points in each recording. Some are obvious, and some are very subtle—sometimes almost inaudible in the recording but visually present in the spectrogram images. Fig. 4.12, 4.13 and 4.14 show examples of obvious and subtle anticipation portamenti in vocal recordings.

Figure 4.12: spectrogram image of obvious ascending anticipation before the beat during “Che” (with positive linear/constant rate of change of ascending pitch, y=x), with corresponding score from Potter’s emulation of Mozart’s Voi che saptete sung by Patti (1905), bar 7-8 (“Che cosa e amor”, what is love in English), along with original recording excerpt for reference. Note that the vertical purple lines indicate crochet beat placement and white diagonal lines indicate tempo fluctuation, where Potter’s emulation of Patti speeds up between “Che” and “co-” and then slows considerably during “cosa e a mor”.51 Note that Patti also uses a negative (descending) linear (y=-x) portamento before the syllable change between “Che” and “co-”, indicated with a green lines in the annotated score.

Figure 4.13: spectrogram image of obvious ascending anticipation on the beat during “cho-,” with corresponding score from Potter’s emulation of Mozart’s Voi, che sapete sung by Patti (1905), bar 29 (“ch’ora”), along with original recording excerpt for reference. The portamento curve is an ascending negative exponential decay, (y=-e^-x) as she starts the pitch ascent slowly and decreases rate of change of pitch as she approaches the destination note, indicated with a red curve in the annotated score. Note that the vertical purple lines indicate crochet beat placement and the white diagonal lines indicate tempo fluctuation (Patti speeds up at the beginning of the phrase after the slower piano interlude, emulated by Potter). The undulations in the green upper partials of Potter's emulation indicate her control of vibrato and strongly match Patti's. Also note that she does not use vibrato during the portamento and eases into oscillations once she lands on the destination note.52

Figure 4.14: spectrogram image of subtle ascending anticipations on the beat during the “s” of “senza” and “p” of “pal,” with corresponding score from Potter’s emulation of Mozart’s Voi, che sapete sung by Patti (1905), bar 48-50 (“sospiro, e gemo senza voler pal pito”), along with original recording excerpt for reference. The portamento curves of each anticipation are all basically positive linear with constant rate of change of ascending pitch, indicated by the green lines in the annotated score. Note that the vertical purple lines indicate crochet beat placement and the ascending diagonal white lines indicate the speeding up throughout this phrase. Also note the subtle dip in pitch on “sos” during the anticipation portamento, which is partially covered by consonant noise from each s.53

Corri advises that, “the distance of the intervals of the Graces … should be consonant with the Key.”54 This is generally the case in the early recordings examined, although it is sometimes difficult to determine from which note exactly the singer anticipates. Two examples of anticipating from a “consonant” note can be found in Melba’s recording of Bishop’s Home, Sweet Home (Fig. 4.15 and 4.16).

Figure 4.15: spectrogram image of obvious anticipation before the beat from a chosen “consonant” note (G#₄) during “though”,55 with corresponding score from Potter’s emulation of Bishop’s Home, Sweet Home sung by Melba (1905), bar 5 (“though”), along with original recording excerpt for reference. The portamento curve is an ascending positive exponential growth (y=e^x), indicated with a blue curve in the annotated score. Note that the vertical purple lines indicate crochet beat placement and the diagonally ascending white line shows an increase in tempo during bar 5.56

Figure 4.16: spectrogram image of obvious anticipation before the beat from a chosen “consonant” note (G#₄) during “there’s,” with corresponding score from Potter’s emulation of Bishop’s Home, Sweet Home sung by Melba (1905), bar 8 (“there’s”), along with original recording excerpt for reference. The portamento curve is a positive exponential growth (y=e^x) as she starts the pitch ascent slowly and increases rate of change of pitch as she approaches the destination note, indicated with a blue curve in the annotated score. Note that the vertical purple lines indicate crochet beat placement and the basically horizontal white line shows an almost steady tempo.57

Violin portamento type 4: anticipation portamento in violin playing

One of the few occasions where vocal and violin portamento use differ is in regard to anticipation slides. Unlike in singing, violinists rarely used anticipation slides to start a phrase. Of the 7 recordings I examined, only 4 recordings had anticipations and there was a total of only 7 obvious and 3 subtle anticipations found across the violin recordings examined, as opposed to 63 obvious and 53 subtle anticipations in the 14 vocal recordings examined.58 The anticipations in violin playing were usually during a phrase between notes that were separated by a break in sound (Fig. 4.17). This seems to be due to the fact that singers can use consonants that are easy to subtly scoop, and because singers have to “make” each note, there is always the feeling of a singer approaching a note from below, even if there is no obvious portamento present. In contrast, violinists just have to put a finger on a string to make a note so in order to have an anticipation to a note they actively have to “make” the slide and it is often difficult to do this in a subtle and tasteful manner across all characters of music.59

Figure 4.17: spectrogram image of anticipation with destination finger and intense bow pressure (due to redness of pitch lines) between phrase marks, with corresponding score and recording from Tchaikovsky-Wilhemj’s Melodie played by Auer (1920), bar 7. The portamento is positive linear with constant rate of ascending pitch change (y=x), indicated with a green line in the annotated score. Note the complete stopping of sound from the violin in between the F₄ and Eb₄ as Auer moves his hand back down the neck of the violin in order to scoop up to the Eb₄after his bow change.60

Portamento type 5: changing fingers on a repeated note in violin playing or scooping back up to a repeated note in singing

Changing fingers on a repeated note can be seen in David’s edition of his Introduction and Variations, sur un thème de Schubert, Op 15: Tema (Fig. 4.18) and Ernst’s Élégie (Fig. 4.19).

Figure 4.18: changing fingers on repeated notes in bars 8, 10, 12, 13, 18, 20, 22, and 23 of David’s edition of his Introduction and Variations, sur un thème de Schubert: Tema, Op 15.61

Figure 4.19: shifting fingers on repeated notes in Ernst’s Élégie (composed between 1829 and 1838), bar 29-30.62

Examples of this shift can be heard in Heermann’s recording of Ernst’s Nocturne in E (c1910) (Fig. 4.20), where Heermann slides up to the E₅ using a same-finger shift under a slur and then shifts from 2^nd finger to 1^st finger each time he repeats D₅ and C#₅ in the chromatic descent.

Figure 4.20: spectrogram image of shifting fingers on repeated notes (finger indicated in brackets in image), with corresponding score and recording from Ernst’s Nocturne in E played by Heermann (c1910), bar 31. Note the slowing of tempo (indicated by the descending white diagonal line), used to emphasise the shifts and diminuendo marking. Due to poor recording quality it is somewhat difficult to see the shifts on the spectrogram, but it is audibly obvious, and you can in fact see slight dips in pitch during each shift, with linear/constant rate of change in pitch, indicated with green lines in the annotated score.63

This technique was also used in singing and can be heard in Patti’s recording of Mozart’s Voi, che sapete (1905) (see Fig. 4.21).

Figure 4.21: spectrogram image of Patti scooping during “l” of “diletto” to slide back up to repeated F₅ (with rubato), with corresponding score from Potter’s emulation of Mozart’s Voi, che sapete (1905), bar 29-30 (“ch’ora e diletto”), along with original recording excerpt for reference.64 The portamento curve is an ascending positive exponential growth (y=e^x) as she starts the pitch ascent slowly and increases rate of change of pitch as she approaches the destination note, indicated with a blue curve in the annotated score.

Other characteristics of portamento from written and recorded evidence as heard in early recordings

Chromatic glissando

Woldemar talks of the Mestrino glissando portamento and often portamento are notated as chromatic scales between notes in both violin and vocal treatises, such as Woldemar (violin) and Novello (voice). However, I have not come across this way of executing portamenti in the early recordings examined.65 Potter adds a chromatic glissando in her emulation of Galli-Curci's recording (see Potter in Fig. 4.22) but this portamento does not correspond to my analysis of the original recording.66

Fig 4.22: spectrogram image of chromatic glissando, and corresponding score and recording from Potter's emulation of Bishop’s Home, Sweet Home sung by Galli-Curci (1917), bar 54 (“like home!”). Note the clear jagged lines (albeit all connected by portamenti) that look like a staircase during Potter’s chromatic glissando in combination with extreme slowing of tempo (indicated by the white diagonally descending line between beats).67 This creates a basically negative linear portamento (y=-x) with constant rate of descending pitch change, indicated with green lines in the annotated score. Although this does not match the original recording, it is an interesting use of this type of portamento and may have historical grounding. However, further research should be conducted into this type of portamento in early recorded singers.

Portamento in relation to vibrato use

Violinist Charles de Bériot lamented the abuse of portamento when used in conjunction with too much vibrato. It is difficult to judge how much vibrato was “too much” during the 19^th century and it is equally difficult to gain a true sense of the use of vibrato in early recordings, as vibrato was often distorted due to primitive recording technology. This aspect was not within the scope of the present study, and further research needs to be undertaken in this area to link portamento with vibrato use. However, general observations of pitch variation in spectrogram images of the early-recorded violinists examined seems to indicate sparing use of vibrato and a much thinner version of vibrato than current modern standard. Soldat and Heermann, like the pure mostly vibrato-less voice of Patti, tended to use very little vibrato in their recordings, especially when using portamento, and both would often increase vibrato during a note, but start each note with a pure tone.

Tessitura and register influence on portamento use in singing

Garcia states that using portamento in higher registers is easier and therefore should be more frequent. Whether he was referring to the registers within each voice type or just to singers in higher registers (such as sopranos and tenors) is unclear. What we can see from early recording analysis is that there were many more portamenti used in female singers as opposed to males, which is why I have included early recordings of mainly female sopranos, with the exception of Walter, who was a male tenor. A thorough examination of the ratio of higher to lower register portamenti is not possible within the scope of this study but I found that most vocal portamenti on the recordings studied were between notes in the upper range of the “fausset” and “tete” registers, as indicated by Garcia in 1840 (see Fig. 4.23).

Figure 4.23: Garcia’s “fausset” and “tete” registers (indicated by red circles) in comparison to other vocal registers throughout the 19^th and 20^th centuries.x Most portamenti I have found in early-recorded singers lie within these pitch registers.

From Henrich’s findings, it appears to me that the lower the register the more work the vocal folds and larynx have to do, as they are thicker when lower and therefore have to work harder to oscillate, which makes pitch change more difficult in general. Therefore, it is potentially harder to slide gracefully or quickly between notes in lower registers, which might explain why there are more portamenti found in the higher registers of female sopranos and male tenors than the lower registers of female altos and male bass and baritones.68

Pitch direction and speed of portamenti in early-recorded singers and violinists

In both vocal and instrumental treatises of the 19^th century, the general consensus was to use fewer portamenti when descending.69 In singing, Lablache, Corri and Garcia, along with others, advised to use more portamento when ascending and less when descending. This is consistent with the early recordings analysed: 10 out of the 14 vocal recordings analysed had more ascending portamenti than descending (Fig. 4.24) and there were clearly more ascending portamenti used than descending in the early recorded violinists I examined, with a range of 42%-71% of all portamenti used being ascending (Fig. 4.25).70

Figure 4.24: Graph of total ascending and descending portamenti evident in the 14 vocal recordings examined. Note the numbers (1-14) on the y axis correspond to the numbered list of recordings at the beginning of the Chapter.71

Figure 4.25: Graph of total ascending and descending portamenti evident in the 21 vocal and violin recordings examined. Note the numbers (1-21) on the y axis correspond to the numbered list of recordings at the beginning of the Chapter (the violin recordings are indicated from 15-21).72

Fröhlich, in his vocal and instrumental treatise of 1811, also suggested that descending portamenti should be faster than ascending. While a complete statistical analysis of the speed of each ascending and descending portamento in each recording is beyond the scope of this study, it is apparent from my analyses that the speed of slide in each portamento is dependent on the character of the piece and their placement within that piece, expressively and dramatically, as will be discussed in further detail below.

Vocal use of rubato with portamento

In singing, Bayly associated portamento with “dragging” (rubato) and every vocal and violin recording I have analysed has rubato in association with the majority of the portamenti present, with 52-95% of the total portamenti in each vocal recording analysed and 56-94% of the total portamenti in each violin recording analysed being used in association with rubato. I have made tempo graphs of each main recording analysed, which can be viewed for comparison in Appendix 4.3.

Melba uses portamento to slow down and speed up parts of phrases, as well as to dislocate the melody from the accompaniment. In this example (Fig. 4.26) Melba uses both the subtle anticipation of G#₄ on the “g” of “give” and slides before the destination syllable from “give” to “me” to slow down the whole phrase. The pianist also arpeggiates the chords underneath her, which is another key 19^th-century expressive technique that further distorts the “true” beat.73

Figure 4.26: spectrogram image of a subtle anticipation slide on G#₄ and a positive linear (y=x) slide before the destination syllable from G#₄ to A₄,74 with corresponding score from Potter’s emulation of Bishop’s Home, Sweet Home, sung by Melba (1905), bar 32 (“give me”), along with original recording excerpt for reference.75 Both the anticipation and same-syllable portamento curves are positive linear (y=x), indicated with green lines in the annotated score.

Similarly, Albani uses portamento to slow over an expressive bar line in Chaminade’s L'été (Fig. 4.27).

Figure 4.27: spectrogram image of slide before destination syllable with considerable slowing down (as seen in the steep descending diagonal white line), with corresponding score from Potter’s emulation of Chaminade’s L'été sung by Albani (1914), bar 24-25 (“qui dore,” who guilds in English), along with original recording excerpt for reference.76 The portamento curve is a descending positive exponential decay (y=e^-x) as she starts the pitch descent quickly and decreases rate of change of pitch as she approaches the destination note, indicated with a red curve in the annotated score.

Walter also uses rubato in the majority of his portamenti (88%), which are especially effective at key dramatic moments, such as in Fig. 4.28. Walter slides before the destination syllable on the descending octave from “wie-“ to “-der” while releasing air pressure and slows down considerably while doing so.

Figure 4.28: spectrogram image of descending portamento before the destination syllable, with corresponding score and recording of Schubert’s Am Meer sung by Walter (1904), bar 17 (“wieder”).77 The portamento curve is negative linear (y=-x) with constant rate of change of descending pitch, indicated with a green line in the annotated score. It could be slightly exponential in shape, but the spectrogram image is not clear enough to distinguish this.

Although tempo fluctuation and melody dislocation were obviously very common in association with portamenti in slower or more sentimental pieces, I found that early recorded singers also used portamento with rubato in key moments of faster joyful pieces, such as Chaminade’s L'Été (Fig. 4.29).78 In this piece Albani uses a descending portamento with positive exponential decay of pitch (y=e^-x) to slow down the beat as she descends from F#₅ to C#₅ on “chan,” indicated with a red line in the annotated score. She then further slows down while drawing out the chromatic descent E₅-D#₅-D₅ on “-tez” by connecting the chromatic notes with subtle negative linear (y=-x) portamenti that are visually clear in the spectrogram but almost inaudible (unless the recording is slowed down to about 60% speed), indicated with green lines in the annotated score. She releases her breath pressure slightly as she slides between the chromatic notes (as seen in the decrease in density of colour in the upper partials of the spectrogram), which further emphasises the sensuous nature of the chromatic descent.

Figure 4.29: spectrogram image with corresponding score from Potter’s emulation of Chaminade’s L’été sung by Albani (1914), bar 45-46 (“chantez ai”/“sing”), along with original recording excerpt for reference.79 The portamento curve is a descending positive exponential decay (y=e^-x), as she starts the pitch descent quickly and decreases rate of change of pitch as she approaches the destination note, indicated with a red curve in the annotated score. Albani also uses fast subtle portamenti to connect the chromatic descending notes on “tez,” which Potter also emulates, indicated with green lines in the annotated score.

Hairpin markings (< >) in relation to rubato use with portamento in violin playing

Historical written evidence of hairpin markings was not only used to indicate bow pressure and speed, but also often used to indicate tempo rubato, which is clearly evident in early recordings. While we cannot know exactly what edition of pieces were being used in each recording, we can see that, like in singing, rubato in association with portamento use was extremely common. Between 56% and 94% of portamenti used in the 7 violin recordings analysed were connected with obvious rubato (an average of 74% per recording). This shows that portamento did not generally stand alone in the expressive language of the 19^th century and was mostly used in combination with other aspects of that sound world, including expressive vibrato, arpeggiation of chords in piano playing, and tempo rubato. These figures also show that while each performer shared a similar expressive language, as the average of rubato use with portamento is high (74%), there was much fluctuation in the ways in which performers used these expressive tools, shown by the range of percentage of portamenti with rubato use between recordings examined (56%-94%). Larger-scale rubato was also present throughout each violin recording examined and a tempo graph of each recording can be seen in Appendix 4.3 for comparison of rubato use between violinists.

Air pressure when executing portamenti in singing

Garcia suggested using “continued pressure of air” during a portamento, while Corri argued that the strength of the slide should be regulated depending on the distance of the intervals.80 As Corri stated: “In descending, drop the Grace into the Note, and in ascending, swell the Note into the Grace,” i.e. decrescendo during a descending portamento and crescendo during an ascending portamento.81 Patti favoured continuous and sometimes increased air pressure throughout her portamenti (see Fig. 4.30), often increasing air pressure during ascending portamenti. The subtlety of her air use can be seen much more easily than other singers, as she sang with an extremely pure tone than makes the aural and visual analysis of her singing much easier.

Figure 4.30: spectrogram image of sliding after the destination syllable with continuous air pressure (shown by consistent green lines connecting notes and redness of fundamental pitches) with rubato shown by white diagonally descending line (declining=slowing), with corresponding score from Potter’s emulation of Mozart’s Batti, batti sung by Patti (1904), bar 34 (“ba-”), along with original recording excerpt for reference.82 The portamento curve is positive linear with constant rate of change of ascending pitch (y=x), indicated with a green line in the annotated score.

Similarly, Melba also often favours steady or increased air pressure when ascending compared with descending, although her use of fast consistent vibrato and a less pure tone than Patti’s sometimes makes judging aspects of her portamenti more difficult. An example of portamento with increased air pressure can be seen in Fig. 4.31, where Melba’s increase in air pressure and dynamic while sliding with the destination syllable (“no”) to the E₅ is shown by the increase of redness and consistency of green lines in the spectrogram image.

Figure 4.31: spectrogram image of sliding after destination syllable with increased air pressure (due to the redness of pitch lines) and tempo rubato (quickening), with corresponding score from Potter’s emulation of Bishop’s Home, Sweet Home sung by Melba (1905), bar 22-23 (“There’s no”), along with original recording excerpt for reference. The portamento curve is ascending negative exponential decay (y=-e^-x), i.e. Melba starts her pitch ascent quickly and slows down as she approaches the destination note, which Potter emulates, indicated with a red curve in the annotated score. Note the increase in tempo during the portamento as indicated by the diagonal ascending white line.83

Similar to Corri’s view of varied dynamic and air pressure during portamenti, Croker states that a variety of approaches to portamento execution is possible, but that in general it is advised to crescendo when ascending and decrescendo when descending in portamento execution (Fig. 4.32).84

Figure 4.32: Croker’s advice for portamento execution, first with plain notation and then with annotation advising nature of ascending and descending slides.85

As seen throughout this Chapter, there are many examples of Corri’s and Croker’s suggestion to crescendo when ascending and decrescendo when descending in early recordings. Similarly, Fig. 4.33 shows Patti’s release of air pressure as she sings the sighing descending fourth on “core”, where the redness of the fundamentals decreases, and where a slight distortion of the green lines occurs towards the end of the slide.

Figure 4.33: spectrogram image of release of air pressure during the “sighing” descending fourth, with corresponding score from Potter’s emulation of Mozart’s Batti, batti sung by Patti (1904), bar 56-57 (“hai core”), along with original recording excerpt for reference.86 The portamento curve is negative linear with constant rate of change of descending pitch (y=-x), indicated with a green line in the annotated score. Note that the redness of the fundamental pitch does not give an accurate reading for Potter’s emulation of Patti’s air pressure as it is influenced by the dynamic of the piano accompaniment, so a better indicator is the decrease in density of green lines in the upper partials to see the release in air pressure.

While I found that there were clearly more ascending slides with basically continuous or increased air pressure and descending slides with releasing air pressure, there are examples of varying use of air pressure in descending slides, such as in Patti’s 1904 recording of Mozart’s Batti, batti (Fig. 4.34). In this example she first releases air pressure from “ver” to “mil” at the end of bar 20 and then uses continuous air pressure to slide down from “cri” to “ne” in bar 21. The differences in air pressure can be seen in the visibility of green lines in the spectrogram image.

Figure 4.34: spectrogram image of different air pressure in descending slides, with corresponding score from Potter’s emulation of Mozart’s Batti, batti sung by Patti (1904), bar 20-21 (“cavarmi il crine”/“tear my hair”), along with original recording excerpt for reference.87 Both descending portamenti are positive exponential decays (y=e^-x), as they start the pitch descents quickly and then slow the rate of change of pitch descent on approach to the destination notes, indicated with red curves in the annotated score.

Bow pressure and speed in executions of portamento in violin playing

As discussed above, vocal treatises of the 19^th century gave differing advice on the use of air pressure when executing portamenti. Stam notes that early-recorded violists often use an even bow speed in general but then use a variety of bow techniques when executing different portamenti in different characters.88 This observation concurs with Corri’s indication of varied air pressure depending on the character when executing portamenti. The hairpins (< >) marked in association with portamenti in Spohr’s and David’s respective treatises (usually marking “crescendo” for ascending portamenti) could be interpreted as indicating increased bow pressure and/or speed when executing ascending portamenti and decreased for descending portamenti. There is not enough written evidence to support this theory conclusively, but it is an interesting aspect of the technique to consider, and evidence of increased bow pressure on an ascending portamento can be seen and heard in Soldat’s recording of Spohr (Fig. 4.7 above), and of Schumann (Fig. 4.35 below), as well as later in this Chapter in Heermann’s recording of Ernst (Fig. 4.40).

Figure 4.35: spectrogram image with corresponding score and recording from Schumann’s Abendlied played by Soldat (1920), bar 8.89 The portamento curve is an ascending negative exponential decay, (y=-e^-x), as she starts the pitch ascent quickly and decreases rate of change of pitch as she approaches the destination note, indicated with a red curve in the annotated score. Soldat uses increased bow pressure on her ascending shift, which is marked with a crescendo as part of the hairpin (< >) marking.

Portamento use in relation to musical character and context in vocal and violin recordings and written sources

There are a number of other general comparisons that can be made between written 19^th-century documentation and early-recorded vocal and violin style. The consensus of written documentation of the early-19^th century was that a clean onset or start to a phrase was best. While the opposite was often seen in early vocal recordings, there were very few anticipations at the beginnings of phrases found in violin playing, and the majority of anticipated slides up or down to notes in violin playing were during a phrase. Bériot often marked portamento between the anacrusis and downbeat of pieces in his treatise, an example of which can be heard in Melba’s portamento during the anacrusis in Fig. 4.31. He also indicated portamento in cantabile passages, which is also evident in early recordings, as the most portamenti were used in slower, more singing pieces, such as in Bishop’s Home, Sweet Home (57, 65, 57), Schubert’s Am Meer (59), Joachim’s Romanze in C (80), Tchaikovsky’s Melodie (65), Spohr’s Violin Concerto: Adagio (96) and Schumann’s Abendlied (28). Similarly, Hall uses more portamenti in the slower espressivo section of Mendelssohn’s Violin Concerto: Allegretto non troppo than in the faster scherzando section (8 portamenti in the first 14 bars, compared to the total of 22 in the 221 bars of the Allegro molto vivace section). Both written and recorded documentation of portamento use shows us that character appropriateness was a key consideration when executing portamento: Joachim warned against using tasteless portamento as it created “unhealthy music-making,”90 and as discussed above and in Chapter 5, Garcia and Bériot give weight to character appropriateness of portamento use, advocating for more portamento use in slower or passionate places.91 Similarly, Spohr’s Violinschule editions of his Violin Concerto No. 9 and Rode’s Violin Concerto No. 7 show fingerings indicating frequent portamenti in the slow movements, which is consistent with other writings of 19^th-century portamento use. Soldat’s recording of Spohr’s Violin Concerto No. 9: Adagio (1920) shows fingerings almost completely consistent with Spohr’s markings, thereby further linking her use of portamento to Spohr’s, in a style which would have also been familiar to Schubert.

The written sources of the 19^th century generally advised to use portamento in moderation, but the early recordings examined show many more portamenti than expected throughout each piece, so it is hard to quantify “moderation.” It is clear from the early recordings that there were more portamenti used in slower or sentimental pieces than in faster or lively pieces. In the recordings of Chaminade’s L'Été and Gounod’s Jewel Song we see far fewer portamenti per bar than the recordings of Mozart’s Voi, che sapete and Batti, batti, Bishop’s Home, Sweet Home, and Schubert’s Am Meer, all of which are slower and sometimes more sentimental or nostalgic pieces, lending themselves to more frequent use of portamento. However, we must remember that obvious and deliberate portamenti were present in every single piece analysed. Rather than multiple portamenti per bar, as in the slower pieces, we tend to see portamenti used in key moments and in lighter and more playful ways in pieces with a jovial character: in these works, slides are often faster linear slides or the more desirable ascending exponential growth or descending exponential decay slides, and often with more release of breath during the slide.

As we know, Garcia advised to use portamenti in more passionate places and we can definitely see lots of portamento used in each passionate or dramatic place in all of the early recordings examined. These instances also tend to have portamento with consistent or increased air pressure, so the slide is even more obvious than in a normal phrase setting. While the consensus of 19^th-century written sources was that a clean onset is best, we see many anticipations throughout every vocal recording. However, some singers only used anticipations a few times per piece and some were so subtle they were almost inaudible under the extra layer of recording sounds. It is difficult to make conclusions about the nature of 19^th-century singers’ onsets, as most notes have a slight swoop up to them on the spectrogram, but they are sometimes hardly audible, and Garcia may have been talking only about using obvious anticipations with large intervals sparingly. What we do know though is that every singer used anticipations in varying ways.

Croker advises portamento use to connect phrases smoothly (Fig. 4.36), which can be heard in both Melba’s 1907 and 1910 recordings of Mozart’s Voi, che sapete (Fig. 4.37 and 4.38), even to the so-called detriment of the phrase structure. In both recordings Melba slows down considerably into the recapitulation of the opening phrase and connects the “-si” of “cosi” with “voi” in a slide before the destination pitch, even though there is a comma between the words, which is consistent with Croker’s annotation. Croker’s close connection to Patti through his teacher, Albert Visetti, who also taught Patti, and the wide dissemination of this treatise at the turn of the 20^th century, shows the relevance and importance of his writings on vocal performance practices of the time.92

Figure 4.36: Croker’s annotation of a slur, indicating portamento use in order to bind two phrases together, even though there is a rest and exclamation mark indicated in the original notation.93

Figure 4.37: spectrogram image with corresponding score from Potter’s emulation of Mozart’s Voi, che sapete sung by Melba (1907), bar 54-56 (“cosi, Voi”), along with original recording excerpt for reference. The first descending portamento is negative linear (y=-x) with constant rate of descending pitch change, indicated with a green line in the annotated score, and the second portamento curve is an ascending positive exponential growth, (y=e^x) as she starts the pitch ascent slowly and increases rate of change of pitch as she approaches the destination note, indicated with a blue curve in the annotates score. Note Melba’s deliberate slowing during “si” on F₄ and the portamento up to “Voi” on Bb₄ (shown by the diagonally descending white line in Potter’s emulation).94

Figure 4.38: spectrogram image with corresponding score from Potter’s emulation of Mozart’s Voi, che sapete sung by Melba (1910), bar 60-62 (“cosi, Voi”), along with original recording excerpt for reference. The portamento shape is positive linear (y=x) with constant rate of ascending pitch change, indicated with a green line in the annotated score. Again, note Melba’s deliberate slowing during “si” on F₄ and the portamento up to “Voi” on Bb₄ (shown by the diagonally descending white line in Potter’s emulation).95

In violin playing, Bériot, based on his close connection with 19^th-century singing style through Garcia, gave examples of different types of portamento for different characters and his markings show an interesting correlation to the mathematical representation of slides in both the violin and voice recordings. Bériot’s “lively” portamento, indicated by a straight line, corresponds to the positive linear/constant rate of change of speed and pitch (y=x) in the ascending portamento in Heermann’s playing (Fig. 4.40), and the “soft” portamento of Bériot, indicated by a curved line, corresponds to the negative exponential decay of rate of change of pitch (y=e^-x) in Heermann’s playing (Fig. 4.40), i.e. he slows the rate of pitch change at the end of the slide as he approaches the D₄.96

While there are examples of slow ascending portamenti throughout these early recordings (see Joachim in Fig. 4.39 for an example), there are also examples of descending portamenti being deliberately drawn out for dramatic effect, such as in Heermann’s recording of Ernst’s Nocturne in E (Fig. 4.40).

Figure 4.39: spectrogram image of a common slow ascending guide finger shift under a slur with steady bow pressure and speed, with corresponding score and recording from Joachim’s Romanze in C played by Joachim (1903), bar 79. The portamento curve is a positive exponential growth (y=e^x), as he starts the pitch ascent slowly and increases rate of change of pitch as he approaches the destination note, indicated with a blue curve in the annotated score. Note the exponential slow ascending portamento is used in association with slowing of overall tempo (indicated by the descending diagonal white line between beats).97

Fig. 4.40 shows that Heermann takes 0.22 seconds to slide up to the B5 harmonic and 0.39 seconds to slide down to D₄, using more bow pressure and less bow speed during this descending slide to create a more intense and deliberate sound thus heightening the dramatic moment.

Figure 4.40: spectrogram image of a faster ascending speed of slide (basically liner: y=x), indicated with a green line in the annotated score, and slower descending (positive exponential decay: y=e^-x) speed of slide, indicated with a red curve in the annotated score, with corresponding score and recording from Ernst’s Nocturne in E played by Heermann (date unknown), bar 35. Note the overall slowing of tempo during the bar with slow descending portamento (as indicated by descending white diagonal line), which corresponds to Heermann’s exponential slowing of pitch descent in his descending portamento from B₅ to D₄. The light purple lines and figures above show the time range between start and end of each portamento.98

Portamento at fermatas

JF Schubert encouraged singers to use portamento at fermatas as another form of “decoration” and these decorations can be seen throughout the examined early recordings.99 Patti slides down to an undefinable note after the fermata in bar 71 of Mozart’s Voi, che sapete (Fig. 4.41) and adds a portamento at the end of the fermata in bar 61 of Mozart’s Batti, batti (Fig. 4.42).

Figure 4.41: spectrogram image of portamento sliding to undefined note at the end of a fermata, along with corresponding score from Potter’s emulation of Mozart’s Voi, che sapete sung by Patti (1905), bar 71 (“-te”), along with original recording excerpt for reference.100 The ascending portamento shape is negative exponential decay (y=-e^-x), where Patti starts the slide quickly and then slows the rate of change of pitch as she approaches the destination note, indicated with a red curve in the annotated score of Potter’s emulation. The descending portamento is descending negative exponential growth (y=-e^x), as she starts the slide slowly, then quickens the rate of change of pitch as she approaches the destination note, indicated with a blue curve in the annotated score. Note that Potter’s annotation of the score indicates that Patti slides to a Bb₄ but I find it quite difficult to define the landing pitch of her slide.

Figure 4.42: spectrogram image of portamento at the end of a fermata, with corresponding score from Potter’s emulation of Mozart’s Batti, batti sung by Patti (1904), bar 61 (“core”), along with original recording excerpt for reference.101 The portamento curve is a descending negative exponential growth (y=-e^x), as she starts the pitch descent slowly and increases rate of change of descending pitch as she approaches the destination note, indicated with a blue curve in the annotated score. Note the released air pressure during the descending portamento, as it acts as a way of “releasing” the held note.

Adams adds a descending slide before the destination pitch at the final fermata in Gounod’s Jewel Song (Fig. 4.43). Although she uses sparing portamento in her recording due to its fast and lively nature, she does use portamento at key points in the piece, and especially to assist in her fluctuation of tempo, like all other singers studied.

Figure 4.43: spectrogram image and corresponding score from Potter’s emulation of Gounod’s Jewel Song sung by Adams (1902-3), bar 92 (“-age”), along with original recording excerpt for reference.102 The portamento is negative linear (y=-x) with constant rate of descending pitch change, indicated with a green line in the annotated score.

Similarly, in early-recorded violin playing, portamenti were also used as an ornament at fermatas in cadenza passages. Heermann uses portamenti throughout his cadenzas in Ernst’s Nocturne in E (Fig. 4.44) and while there are no specific cadenzas marked in Spohr’s Violin Concerto, No. 9: Adagio, there are multiple cadenza-like passages throughout the movement where Soldat uses a great deal of portamento.

Figure 4.44: Recording clip and score of the final cadenza of Ernst’s Nocturne in E played by Heermann (c1910).103

Portamenti used to emphasise chromatic movement in violin playing

Portamenti (often, but not exclusively same-finger shifts) were used to emphasise minor seconds in David’s edition of Beethoven’s Violin Sonatas and Ernst’s Élégie (see Fig. 4.45).104

Figure 4.45: same-finger (1^st-finger) shifts under slurs to emphasise the minor seconds between Ab₄ and G₄ marked in bar 15 of Ernst’s Élégie (written between 1829 and 1838).105

Examples of this type of shift can be seen in Marie Soldat’s recording of Schumann’s Abendlied (Fig. 4.35 earlier in the Chapter) and Joachim’s of his Romanze in C (Fig. 4.46). Soldat uses a subtle but audible guide-finger shift from Ab₅ to G₅ under a slur to emphasise the minor second. She then increases bow speed and pressure to slide with the guide finger from G₅ to D₆, possibly because of the marked hairpin, to create an ascending slide with negative exponential decay of rate of change of pitch (y=-e^-x), i.e. she gradually slows down as she slides into the D₆ during the “crescendo” of the hairpin, which can also be seen due to the decrease in white line tempo indication.

Similarly, Joachim uses a same-finger shift under a slur to emphasise chromatic ascent (Fig. 4.46).

Figure 4.46: spectrogram image with corresponding score and recording from Joachim’s Romanze in C played by Joachim (1903), bar 16. Note the increase in overall speed during the ascending chromatic slide in association with the crescendo marking.106 The portamento is positive linear (y=x) with constant rate of ascending pitch change, indicated with a green line in the annotated score.

Sliding between chromatic notes can also be heard in early-recorded singing, an example of which can be seen in the spectrogram image of Albani’s recording of Chaminade earlier in this Chapter in Fig. 4.29.

Portamento use in orchestral playing

General written advice in the early-mid 19^th century was not to use portamento in orchestral playing unless specified (as seen in Spohr, Gassner and Meyerbeer).107 While this was condemned throughout the century, clear portamenti can be heard in the recordings of Elgar conducting Nimrod from his Enigma variations in 1926 with the Royal Albert Hall Orchestra and in the orchestral introduction of Heermann’s recording of Ernst’s Nocturne in C (namely, destination after bow change, guide under slur and guide before bow shifts).

Discussion and Conclusions

From my examinations of 21 early recordings of singers and violinists, I have been able to illustrate patterns of 19^th-century performance style and draw connections between the use of portamenti in early recordings and those documented in written sources of the early-mid-19^th century. I used Sonic Visualiser to help identify and categorise the different types of portamento found in both vocal and violin recordings, including several hitherto academically undescribed types. I found that many portamento types were common to both vocal and violin use and concurred well with 19^th-century written sources of portamento use. The desired legato singing and playing style of the 19^th century was heard and seen throughout the vocal and violin recordings, and the common portamento types from 19^th-cenutry treatises, such as sliding under a slur/syllable, sliding before a bow/syllable change, anticipation slides, and scooping back up to/changing fingers on a repeated note, were also used throughout all recordings in ways consistent with 19^th-century writings. While there was some variation in portamento use between vocal and violin recordings, due to obvious differences between singing and violin techniques, the general musical intention of different types of portamento were similar in both vocal and violin uses of the device. I found that, consistent with 19^th-century writings, there were more ascending than descending portamenti, and often singers and violinists would crescendo when ascending and decrescendo when descending, but not always, as it was dependent on the expressive context of the phrase.

Generally, in both vocal and violin recordings, I found that portamento slide types were either linear (usually faster portamenti) or were ascending portamenti with positive exponential growth, i.e. the rate of change of pitch increased during the approach of the destination pitch. Bériot indicated portamento slide shapes in his treatise (as discussed in Chapter 3, see Fig. 3.23 for example). Linear portamento shapes corresponded to Bériot’s “lively” portamento, which could be used on notes “thrown with grace or hurled with energy,” and this can be heard in various recordings examined in both character types. The ascending positive exponential growth portamento shape matched the shapes of “soft” and “drawn-out” portamento types as indicated by Bériot, which were indicated for moments of “tender” or “plaintive or sorrowful expression,” respectively.108 While this type of portamento shape is fitting for tender, plaintive and sorrowful characters, I would also argue that it is relevant for any character if executed in a thoughtful way. The ascending positive exponential growth shape seems to be the most common and tasteful type of portamento slide heard in the early vocal and violin recordings examined (as illustrated in Fig. 4.4, 4.10, 4.15, 4.16, 4.21, 4.37 and 4.39). I have found this portamento type to be the least obtrusive way to execute expressive portamenti, as it is more natural to hold the initial (lower) note longer, starting the ascending slide slowly then quickening into the destination pitch, than to draw out the final part of the slide up into the destination pitch. It is also interesting to note that the majority of guide-finger shifts in the violin recordings were either linear or ascending exponential growth slide shapes, as it is natural to start an ascending guide-finger shift slowly and then quicken into the destination pitch as you put down the destination finger. In contrast, it is easier to slow down into the destination note when using a destination-finger shift; the slowing of rate of change of pitch ascent helps the violinist find the correct pitch with the sliding destination finger. This association gives us insight into the reasons behind the preference for guide-finger shifting by early-19^th-century violin pedagogues, such as Spohr and David, as, in most settings, the slide shape of guide-finger shifting sounds and feels more natural, both aurally and mathematically, than destination-finger slide shapes. The equivalent descending type of portamento slide, which is descending exponential decay, also works well as a relatively unobtrusive type of portamento, as illustrated in Fig. 4.27 and 4.29, where both Marchesi and Albani respectively use this type to enhance the sensual French character of Chaminade’s L’été. Similarly, Patti uses this type in Fig. 4.34 to emphasise the phrase, “tear my hair,” in Mozart’s Batii, batii and Heermann uses it in Fig. 4.40 after a held note in a cadenza-like passage. Phrases that have no particular or overly dramatic intention are often musically enhanced by these subtle and natural portamento types, thus these types of portamento are especially useful when aiming to create a musical phrase with unobtrusive portamenti.

On the other hand, ascending portamenti with exponential decay of pitch change, tend to be more obvious or obtrusive to the ear and lend themselves more to moments of expression and drama that call for more intensity. This portamento type, which could be seen as the less desirable type of slide, can be heard throughout Patti’s recordings (e.g. in Fig. 4.13 and 4.41) and the frequency of this type of portamento may be a contributing factor to some of the current negative over-indulgent associations with Patti’s singing, as this type of portamento is much more obvious than others and sometimes sounds “out of character” in calmer musical contexts. However, this type of slide can be used to great effect. Soldat uses this type of portamento slide at a moment indicated with a hairpin in Fig. 4.35, and because of her relatively fast slide and the placement of the portamento at the “peak” of the hairpin, the portamento type fits within the musical context. Similarly, Melba in Home, sweet home (Fig. 4.31), uses this type in a portamento after the destination syllable, which also fits within the nostalgic and sentimental character.

Due to this variety in use of portamenti, I found it essential to also consider other 19^th-century expressive devices, such as expressive vibrato, tempo rubato and piano arpeggiation, when analysing early-recorded portamento use, as these expressive elements cannot be effective when standing alone. The majority of portamenti were used in association with some form of tempo rubato, as portamenti tend to distort the “real” beat, and often singers and violinists would vary their use of vibrato depending on the musical context of a given portamento. I found a variation in speed and intensity of slide throughout each recording, which was dependent on the character and expressive intention of the context of each portamento. 19^th-century written sources advised more portamento use in passionate or slow, cantabile passages, which was consistent with these early recordings. However, I also found portamenti used in faster and playful pieces at expressive or dramatic moments, which shows that, instead of only being used in one particular character, portamento was a fundamental part of any musical expression, with the type of portamento used being varied to fit the expressive context.

One inconsistency between written and recorded evidence of portamento use is that written documentation from Schubert’s time often suggested portamento use in moderation, whereas early-recorded singers and violinists used portamento frequently. However, the contemporary written sources from the early-recordings era also urged portamento and vibrato use in moderation; both Joachim and Andreas Moser’s Violinschule of 1905 and Auer’s Violin Playing As I Teach It (1921) urge for this restraint in portamento use,109 but we have clearly heard in this Chapter that both Joachim and Auer used frequent and varied portamenti throughout their recordings. Substantial research has been undertaken into the inconsistencies between written and recorded evidence of late-19^th-century performance practices. Peres da Costa concludes that “the comparison between early recordings and contemporaneous written texts has exposed striking contradictions time and time again.”110 Therefore, we must not take written evidence as prescriptive rules, as the many “delicate shades in music” that cannot be described in writing and must “be learnt and felt by the genius and practise … of a performer.”111

Through this analysis of early-recorded singers and violinists, I have explored the many different types of portamento and their relevance to early-19^th-century performance practices. But most importantly, I was struck by the spontaneity and individuality of each recording and the different ways in which portamento and other expressive devices were used. These performers were obviously speaking a collective expressive language but also using their own expressive dialects to convey their individual musical voices. Naturally, the next step in this exploration of portamento use within 19^th-century performance practices is to experiment with the nuances and variety within this language in current HIP singing and violin playing, and this is explored in Chapter 5.