17
Stadler Elmer, Stefanie, KV1 Stadler Elmer, S. (2011). Structural aspects of early song singing. In A. Baldassare (ed.), Music – Space – Chord – Image. Festschrift for Dorothea Baumann‘s 65th birthday (pp. 765-782). Bern: Lang. Structural aspects of early song singing Stefanie Stadler Elmer Abstract ͳ Early song singing is a natural phenomenon still rarely studied. It reveals a child’s access to both music and language within a cultural setting, on the basis of predisposed human faculties. In this study a young (1 year and 8 months) girl’s singing of a traditional song was analysed with computer-aided methods. The results show that she produced relevant features in structured time to create a song that resembles the model in many respects. The comparison with the song model revealed some of her strategies in solving the problem of producing a stream of ordered sounds. Musical elements such as melodic contour, phrase segmentation, periodic stress (or metric) pattern, repetition and variation, and onomatopoeic syllable formation are major constituent elements, whereas linguistic features such as pronouncing syllables or forming words are less clear or even absent. The results of this case study are congruent with the singing-before-speaking or the musical- origins-of-language hypothesis. 1. Introduction For most children in any culture, songs are the first comprehensive musical form they encounter, mostly directed to them in form of lullabies and other songs sung by their parents or caregivers. Probably in all cultures, parents discover that singing makes it possible to induce or enhance feelings of comfort and wellbeing in both the singer and listeners. In recent decades, the affect-regulating functions of song singing have received more attention in research. Caregivers’ infant- and child-directed song singing has become a growing subject of interdisciplinary interest for infant 1 I am very grateful to Zvi Penner and Dorothea Baumann for inspiring discussions on this subject. - As a collaborator in the AIRS (Advancing Interdisciplinary Research in Singing) MCRI (Major collaborative Research Initiative) supported by the Social Sciences and Humanities Research Council of Canada (SSHRC), I also express my appreciation for the scholarly opportunities provided at the international level.

Structural aspects of early song singing

Embed Size (px)

Citation preview

Stadler Elmer, Stefanie, KV1

Stadler Elmer, S. (2011). Structural aspects of early song singing. In A. Baldassare (ed.), Music – Space – Chord – Image. Festschrift for Dorothea Baumann‘s 65th birthday (pp. 765-782). Bern: Lang. �

Structural aspects of early song singing

Stefanie Stadler Elmer

Abstractͳ �

Early song singing is a natural phenomenon still rarely studied. It reveals a

child’s access to both music and language within a cultural setting, on the basis of

predisposed human faculties. In this study a young (1 year and 8 months) girl’s

singing of a traditional song was analysed with computer-aided methods. The results

show that she produced relevant features in structured time to create a song that

resembles the model in many respects. The comparison with the song model

revealed some of her strategies in solving the problem of producing a stream of

ordered sounds. Musical elements such as melodic contour, phrase segmentation,

periodic stress (or metric) pattern, repetition and variation, and onomatopoeic

syllable formation are major constituent elements, whereas linguistic features such as

pronouncing syllables or forming words are less clear or even absent. The results of

this case study are congruent with the singing-before-speaking or the musical-

origins-of-language hypothesis.

1. Introduction �

For most children in any culture, songs are the first comprehensive musical form

they encounter, mostly directed to them in form of lullabies and other songs sung by

their parents or caregivers. Probably in all cultures, parents discover that singing

makes it possible to induce or enhance feelings of comfort and wellbeing in both the

singer and listeners. In recent decades, the affect-regulating functions of song singing

have received more attention in research. Caregivers’ infant- and child-directed song

singing has become a growing subject of interdisciplinary interest for infant

������������������������������������������������������������1 I am very grateful to Zvi Penner and Dorothea Baumann for inspiring discussions on this subject. -

As a collaborator in the AIRS (Advancing Interdisciplinary Research in Singing) MCRI (Major collaborative Research Initiative) supported by the Social Sciences and Humanities Research Council of Canada (SSHRC), I also express my appreciation for the scholarly opportunities provided at the international level.

Stadler Elmer, Stefanie, KV1

researchers,2 linguists,3 and musicologists, and those biologists4 interested in music

perception and cognition from an evolutionary perspective. The infant is equipped to

perceive features relevant for language5 and music processing.6 The auditory systems

start functioning as early as about three month before birth.7

Children not only listen and react to songs; they also learn to reproduce them.

Spontaneous song reproduction and song invention can be observed during the

second year of life.8 Very early, children discover the regularities in these musico-

linguistic events and are motivated to combine the various elements involved

playfully to create primitive versions of a coherent song. When they do, children

tend to follow some of the general rules concerning the temporal organisation of

melodies and lyrics.

Children’s early song singing is not only amazing, it also raises important

interdisciplinary questions such as: How does a child organise linguistic and musical

elements? What does a young child express about what he or she discovers about the

regularities in song singing? How does the singing conform to and diverge from

traditional or conventional forms? What kinds of regularities are to be found in early

performances? Is there a kind of hierarchy of dispositions, or of a developmental

sequence with respect to competences in gaining access to auditory stimulations for

monitoring vocalisations? What do performed structures tell about the child’s

perception and cognition of musical and linguistic rules?

As far as the relationship between the musical and the linguistic domain is

concerned, one old idea, expressed, for instance, by Jean-Jacques Rousseau in 1781,9

������������������������������������������������������������2 E.g. Elena Longhi, Songese’: maternal structuring of musical interaction with infants, in: Psychology

of Music, 37/2 (2009), 195–213. 3 E.g. Simone Falk, Musik und Sprachprosodie. Kindgerichtetes Singen im frühen Spracherwerb. Berlin

2009. 4 E.g. Aniruddh Patel, Music, language, and the brain. Oxford 2008; Bjorn Merker, Guy S. Madison, &

Patricia Eckerdal, On the role and origin of isochrony in human rhythmic entrainment, in: Cortex 45 (2009), 4–17.

5 Peter W. Jusczyk, The discovery of spoken language. Language, speech, and communication. Cambridge Mass. 1997.

6 Sandra Trehub, Infants as musical connoisseurs, in: The child as musician. A handbook of musical. Ed. Gary E. McPherson. Oxford 2006, 33–49.

7 Jean-Pierre Lecanuet, Prenatal auditory experience, in: Musical beginnings. Origins and development of musical competence. Ed. Irène Deliège and John Sloboda. Oxford 1996, 3–34.

8 E.g. Patricia E McKernon, The development of first songs in young children, in: New directions for child development. Ed. Dennie Wolf. San Francisco (CA) 1979, 43–58; Mechthild Papousek and Hanus Papousek, Musical elements in the infant's vocalization: Their significance for communication, cognition, and creativity, in: Advances in Infancy Research 1 (1981), 163–224.

9 Jean-Jacques Rousseau, Essai sur l'origine des langues, où il est parlé de la Mélodie, et de l'Imitation musicale. Geneva 1781, http://www.uqac.uquebec.ca/zone30/Classiques_des_sciences_sociales/index.html (accessed December 5, 2008).

Stadler Elmer, Stefanie, KV1

is singing capacities are pre-adaptations that enable or facilitate language. Similarly,

as well as many others, Vaneechouttee and Skoyles argue, “children learn spoken

language by means of innate melody recognition capacity.”10 They postulate a Music

Acquisition Device or MAD, in the sense that human infants are naturally equipped

with the capacity to recognise and reproduce pitch and stress patterns. In early

musicality, the roots for language can thus be found, or put differently, language has

some of its origins in music. Musical abilities may help to explain how children learn

a large vocabulary, and create and acquire linguistic grammar.

In this context, early song singing can be considered very interesting behaviour

since it relies on both musical and linguistic elements and rules. According to the

singing-before-speaking hypothesis, we can assume that the musical dimensions of

song singing are more easily accessible for the child than the linguistic ones. In

contrast to this hypothesis, in the past researchers used to postulate a linguistic

primacy theory that claimed words appear first in both song acquisition and singing

development, followed by rhythm, contour, and intervals, in that order.11 This

hypothesis led to progress in singing being described as moving from speech-like

chanting of the song text to singing in a limited pitch range.12 I have critically

discussed this theory elsewhere13 and will not address it further here. Rather, I want

to find out what early song singing reveals about the role of musical and linguistic

features and rules that a child is able to create and recreate.

Bearing the singing-before-speaking-hypothesis in mind, this chapter is limited

to one case study. It concerns a girl’s spontaneous singing of a traditional song at the

age of one year and eight months. Before analysing her singing (section 4), we need

to define song singing with its basic and general features and some rules (section 2),

and discuss the methodology for analysing this behaviour (section 3). Because the

vocalisations of young children do not yet fit into a particular cultural system with

specific linguistic and musical rules, analysing and describing them is an intricate

enterprise.

The aims of this chapter are to provide an example of a micro-genetic analysis of

very early spontaneous song singing, to analyse its structural characteristics, to

������������������������������������������������������������10 Mario Vaneechoutte and John R. Skoyles, The Memetic Origin of Language: modern humans as

musical primates, in: Journal of Memetics 2 (1998), 84–117, http://cfpm.org/jom-emit/1998/vol2/vaneechoutte_m&skoyles_jr.html1998, (p. 6) (accessed June 18 2010).

11 E.g. Helmut Moog, Das Musikerleben des vorschulpflichtigen Kindes. Mainz 1968; David Hargreaves, The developmental psychology of music. Cambridge 1986. See for details: Stefanie Stadler Elmer, Kinder singen Lieder: Über den Prozess der Kultivierung des vokalen Ausdrucks. Münster 2002.

12 Graham Welch, A developmental view of children's singing, in: British Journal of Music Education 3/3 (1986), 295–303.

13 Stefanie Stadler Elmer, Die Entwicklung des Singens: Eine kritische Diskussion der Beschreibungs- und Erklärungsansätze, in: Zeitschrift für Entwicklungspsychologie und Pädagogische Psychologie 28/3 (1996), 189–209.

Stadler Elmer, Stefanie, KV1

compare them with the traditional song, and finally, to reconstruct the child’s access

to the musical and linguistic rules of song singing and to evaluate their role. This

research should shed more light on children’s early capacities to express vocally

musical and linguistic units, and on the relationship between these two domains.

2. Song singing �

In many modern media during the past couple of years, the notion song has often

been used for a piece of any kind of music. Here I adhere to the traditional word

meaning, where song is related to voiced musical sounds.14 In a minimal sense, a

song is a vocally produced sequence of prolonged vowels whose pitches are

modulated. This minimal version is typical of infant vocalisation before it separates

out into more singing-like and more speaking-like modes. In addition to prolonged

vowels, the infant’s early more singing-like vocalisations can be characterised by

their connection to the child’s emotional state and their feelings of comfort,

wellbeing, or playfulness, as well as to particular movements. In contrast, the more

speaking-like vocalisations also include the expression of negative emotions and the

signalling of requests. The ability to sing can be considered to be an innate capacity.

For parents and caregivers, the infant’s vocalisations contain very important

information about the child’s basic needs and emotional states.

The distinction between pre-musical and pre-verbal vocalisations is rather vague

since, as contextualised listeners, we tend to attribute culturally specific features that

might not be verifiable, and also might not be intended by the infant. It is only

towards the end of the first year and around the beginning of the second that

situations can be observed where the child seems to be able to switch between modes

to express a clear intention. Our analysis so far has shown that the speaking-like

mode seems differentiated later than the infant’s singing mode. In other words,

extending vowels and thereby modulating pitch to form a melody is far easier for an

infant than articulating syllables and combining them into words. Nevertheless, for

the infant, both modes are strongly intertwined, and contextual conditions are

essential to identify the infant’s or child’s intentions. We need concepts of a minimal

or pre-version of song singing to be able to categorise the continuum between

speech-like and singing-like vocalisations, and to follow their transition into

gradually elaborated and culturally specific forms.

Children achieve some differentiation in their vocalisations when their primitive

singing version extends to a sequence of syllables (lyrics) that are patterned by

������������������������������������������������������������14 The analysis of 13 sources (dictionaries, encyclopedias) yielded throughout key words such as

singing, vocal activity, sung poetry, sung melody, chanted poetry, and vocal composition. All definitions are related to using the human voice.

Stadler Elmer, Stefanie, KV1

periodic accents and bound to more-or-less discrete pitch categories. This leads to the

next definition: song singing always consists of some kind of linguistic and musical

elements, of basic elements of both domains, which essentially are vocal sounds

differing in frequency, duration, loudness, and timbre. Apart from these basic

elements, linguistic and musical elements have culturally specific features and follow

conventions. Briefly said, they concern the organisation of time by meter on the basis

of a regular pulse that is periodically accentuated by strong and weak beats, of pitch

by tonal rules (scales etc.), and of syllables aligned with the beats or notes, forming

words and rhymes. These song-singing rules are exemplified in more details in

section four where the song model the child intended to sing is described.

3. Method to analyse children’s singing �

When analysing the basic parameters, pitch, timing, and syllables, in infants’ or

children’s vocalisations, a combination of auditory control and acoustic analysis

seems optimal. Baumann has also proposed such a combination as a general

methodological procedure.15 To analyse children’s singing, we designed a method

using our two freeware computer programs.16 The first, the Pitch Analyzer, offers

two different algorithms for extracting pitch,17 and the second, the Notation Viewer

provides a graphic representation of data based on the acoustic signals

(http://mmatools.sourceforge.net/), both with detailed instructions. Other programs

for analysing pitch are also available,18 but they are not specially designed for this

kind of signal. This method is the first to analyse the organisation of several

parameters simultaneously configured in a sung song on an acoustic basis. Here, I

only summarise the main steps.

The child’s singing and the social context are either recorded on audio- or

videotape. The analysis starts with the analysis of the social interaction, where

significant stimulators of singing are the social context and, even more important, the

vocal interaction prior to the child’s singing. When studying children’s acquisition of

a new song, which is defined as the model song, the teaching is adaptive. Any event

related to the model is reported, including the presentation of the model, joint

singing, and a child’s solo singing, to reconstruct the dynamics of the learning

process and to position a child’s solo productions within the sequence of the events.

������������������������������������������������������������15 Dorothea Baumann, Können wir unseren Ohren trauen? Hörerfahrung und Messresultate müssen

sich ergänzen, in: Schweizer Musikzeitung 1 (1998), 1–9. 16 Stefanie Stadler Elmer, und Franz-Josef Elmer, A new method for analyzing and representing

singing, in: Psychology of Music 28/1 (2000), 28–42. 17 Wolfgang Hess, Pitch determination of speech signals: Algorithms and devices. New York 1983. 18 E.g. praat by Paul Boersma and David Weenink, Praat: Doing Phonetics by Computer. University of

Amsterdam, Spuistraat 210, 1012VT Amsterdam, available at www.praat.org.

Stadler Elmer, Stefanie, KV1

In the case of spontaneous singing, whenever possible the social context should be

recorded and reported as well.

After the interaction or context analysis, each child’s sung solo is stored as a

sound file in the computer and then analysed with acoustic tools. The Pitch Analyzer

is shown in figure 1. The rich data given by this acoustic analysis is then reduced to a

limited set of categories and symbols in a new notation system. The sung pitch

qualities are coded by symbols as summarised in table 1. These codes and data on

phrases, pitch endings and beginnings, the onset-times, and syllables are structured

for the visualising program (Notation Viewer). It yields figures 3 and 4, where the

child’s organisation of the following parameters is shown: syllables, their pitches and

timing. Such visualisations are the basis for the next step where the diagrams are

analysed with respect to a person’s strategy or recurrent patterns in making songs.

Fig 1: The Pitch Analyzer calculates pitch with two different algorithms. Results are given in Hz and cents.

Stadler Elmer, Stefanie, KV1

Tab. 1: Symbols used to denote measures and additional features in the diagrams.

Code Symbol Description

1 stable pitch

2

stable pitch, ending with an upward or downward glissando

3

stable pitch, starting with an upward or downward glissando

4

unstable pitch, but a clear upward or downward glissando

4. Early song singing – the case study with Ulla (1 year, 8 months)

A traditional song as a model and the context of singing

The context of Ulla’s singing was ritualistic in the following sense. Her family had

a wooden rocking horse and whenever a child sat and rocked on it, the parents or the

caregiver sung the traditional song Hopp, hopp, hopp, Pferdchen lauf Galopp. This song

suits the scenario with the rocking horse well because the lyrics mention a horse, and

the binary meter accords well with the rocking movements. The song is given in

figure 2. The temporal organisation is a binary meter counted as two quarters, with a

strong accent on the first beat and a medium one on the second beat. In most phrases,

this basic pulse is subdivided into eight notes. The song is segmented into three parts

as ABA, and each one is divided into two sub-parts, so that it has twelve bars

altogether. The phrase A and its slightly varied version A’ have clearly marked

boundaries with pauses between the sub-parts, whereas phrase B consists of four

bars with eight notes without a pause. The melody is in F major, and the repetitive

motive is organised within the range of a fifth in all parts either as a tonic triad or as

a scale. The melody follows the basic rule of marking the key in the first bar and of

ending with the tonic. The lyrics have three rhyming pairs positioned

correspondingly at the end of each sub-phrase. Altogether, the boundaries of the sub-

phrases and phrase are well marked by the pauses, rhymes, and melodic motive of a

fifth.

Stadler Elmer, Stefanie, KV1

Fig. 2: The traditional song Hopp, hopp, hopp, Pferdchen lauf Galopp. The first line shows the score and the second line the lyrics. The rhymes at the end of each sub-phrase are marked with a bracket. The third line visualises the periodic accents with vertical lines with three different weights: strong, medium, and weak. Finally, the phrases are marked with brackets and denoted with AB1B2A’.

One of these horse-rocking-and-singing rituals was video-recorded. It shows Ulla

rocking on the wooden horse, when the caregiver next to her has started to sing the

song. The child joins in singing, and after one or two phrases, the caregiver stops

singing and lets Ulla sing on her own. Ulla continues singing the phrases AB nine

times, but she never sings the entire phrases consisting of AB1B2A’ (see figure 2).

Then she starts again with phrase A (the tenth time), and stops after the sub-phrase

/bob, bob, bob/. Then the caregiver continues with sub-phrase A /Pferdchen lauf

Galopp/. Ulla joins in after two notes, and they continue with B1B2A’ and end

together. Then the caregiver starts talking to Ulla, and the recording ends. This entire

scene lasts 137 seconds. Of the complete audio file, 30 seconds of Ulla’s singing were

analysed as shown in two continuous parts in figures 3 and 4. In addition to these

figures, table 2 shows Ulla’s production of her syllables during 53 seconds. Of these,

the first 30 seconds are included in figures 3 and 4.

Stadler Elmer, Stefanie, KV1

Fig. 3: The first part of Ulla’s singing. The thin line represents the song model. It was sung in B major. The scale’s tonic triad notes (do, mi, so) are given as dashed horizontal lines. The points and solid lines tied by dashed lines represent Ulla’s vocal productions. The y-axis represents the pitch-continuum on which the Western pitch categories are depicted. The x-axis represents time in seconds, and the ticks below the x-axes show Ulla’s onset-times of syllables. The syllables she sung are given again in table 2.

Fig. 4: Ulla’s singing, part 2, is the immediate continuation of the first part in Fig. 3.

Stadler Elmer, Stefanie, KV1

Overall, this early song singing is remarkable because the analysis shows that a

very young child is able to simultaneously organise melodic, temporal and linguistic

features into a coherent unit that is easily recognisable by listeners as this particular

traditional song.

The pitch dimension in Ulla’s singing

As far as melodic features are concerned, there are several striking occurrences:

The analyses in both diagrams show that Ulla’s melody matches the contour of the

model song well. There is a great resemblance between the sung song and the model.

When listening to the recording, the melody is clearly identifiable.

As mentioned earlier, the fifth is a prominent melodic feature in the song model

(see figure 2). All melodic patterns have the range of a fifth, as do the two sub-

phrases of A and A’, and each of the two bars in B1 and B2. Did Ulla discover this

recurrent melodic pattern? Does her singing show that some pitches - the tonic triad -

are more important than other ones? In figure 3 we see that, in her first two sub-

phrases, Ulla extends the contours, yielding a larger interval than a fifth, whereas the

contours of the next two sub-phrases of B are both clearly smaller than a fifth. The

next sub-phrase, starting at second 13, her second sub-phrase of A with the syllables

/bob, bob, bob/, not only yields a perfect fifth, but also a perfect tonic triad with three

absolute pitches matching the model. This is remarkable. Her second sub-phrase of A

is shifted lower as a whole, but almost covers a fifth. The next two contours of sub-

phrases B in figure 4 are again smaller than a fifth. The last two sub-phrases of A in

the same figure are also smaller than a fifth, but they have rather stable contours and

each one starts and ends at almost the same pitch.

Before considering the question about her access to tonal regularities, the

following issue has to be addressed: Ulla’s two phrases B, beginning at second 8 in

figure 3 and second 0 in figure 4 raise a problem: the caregiver spontaneously sung

the song in B major. The model’s lowest pitch is F#3 in phrase B, and this is too low

for a young child’s vocal range. How does she deal with this? She seems to sing as

low as possible, but in both cases she cannot produce a stable or almost stable pitch.

Her lowest pitch region is clearly below A3, but this is still too high to match the

given song model. This problem might have made her narrow the pitch range of

phrase B in both subparts. What is also interesting is her ‘mistake’ in phrase A in

figure 3 at second 5, where she repeats a pitch that is almost C4. She produces this

pattern again at the same pitch level at the beginning of the next phrase B at second

8. Only the second time this pattern corresponds with the model. It seems that, at

second 5, Ulla anticipated the pitch pattern that would soon follow at exactly the

same pitch level. Had she discovered the stabilising functions of the pitches of the

tonic triad? To try to answer this question, I compared the matches between her

Stadler Elmer, Stefanie, KV1

melodic pitches and the three horizontal lines indicating do, mi, so and found that

they matched only sometimes: At the ending of the first sub-phrase, at the beginning

and ending of the second sub-phrase, the ‘perfect’ second sub-phrase of A (all in

figure 3), and at the beginning and ending of phrase A (figure 4). All these matches

concern the tonic and the dominant (the fifth). It seems that Ulla sometimes uses

these important scales notes as tonal centres while organising the melody. Although

the melodic contour as a whole fits well with the song model, it is only some

beginnings and endings of sub-phrases that indicate some tonal stability. Few of her

pitches suggest a general, stable organisation.

I conclude from Ulla’s melody that she had already discovered some melodic

rules:

a) Produce melodies with stable pitches (unless a problem occurs).

b) Maintain the direction of a melodic contour.

The next two rules appeared to apply sometimes, but not yet in a stable fashion:

c) Focus absolute pitches.

d) Identify repetitions of same pitches within and across contours.

e) Use the tonic and the dominant as tonal centres.

The time dimension in Ulla’s singing

Turning to the temporal organisation, we can examine the phrasing, durational

units with subdivisions, and accents. Ulla reproduces the phrase-structure as

presented to her, except for the sub-phrases of B (beginning of figure 4), where she

halves the sub-phrases again. By maintaining the symmetry, she still follows the rule

that breathing should not interrupt the metrical pattern. This is not trivial, since some

children may occasionally break this rule.19 All her phrases are coherent and contain

the correct number of notes or syllables. As mentioned above, throughout the

recorded scenario, on her own, she did not sing the entire phrases consisting of

AB1B2A’, but only phrases A and B. Ulla omits the second part of B, and she replaces

A’ with A. Note that A’ is a melodic variation of A. What is striking here is the fact

that Ulla does not omit elements at the lowest level of the song’s organisation, the

notes and syllables, but she omits and replaces entire phrases. What does tell us

about how she creates the phrases?

As shown in figure 2, the phrases are kept together by the dense and intertwined

organisation of melodic and linguistic elements synchronised by metrical rules. Each

sub-phrase has clear boundaries marked by pauses, rhymes, melodic contours, and a

symmetrical number of bars. Metrical rules concern the duration of the pulse, on the

one hand, and the periodic accents on the other.

������������������������������������������������������������19 E.g. Stadler Elmer, Kinder singen Lieder (as note 11).

Stadler Elmer, Stefanie, KV1

Ulla repeats the first part of phrase A (Hopp, hopp, hopp), and uses it to substitute

the first sub-phrase of A’, although it would theoretically appear only at the song’s

beginning. What makes it so prominent? Only this sub-phrase has long notes,

consists of a repetitive syllable, and ends with a pause. Interestingly, she only clearly

pauses before and after this sub-phrase /bob, bob, bob/ in-between phrase A, and not

between her sub-phrases of B, and she even ignores the pause between A and B. Yet,

she clearly produces the phrase-structure by breathing as proposed by the model, but

she seems to be attracted by the contrastive features of the first sub-phrase with its

repetitive syllables, long notes, and a pause. In addition, there is the rhyme-pair

/Hopp-(Ga-)lopp/ at the end of the each sub-phrase of A, which indicates another

structuring feature across the two sub-phrases of A. While she discovered the sub-

boundaries and boundaries of phrase A, this is not the case for phrase B: here she

omits the second part of B. We can speculate that she had not yet grasped the rhyme-

pair /Steine – Beine/ that would allow cross-binding of both ends of the sub-phrases

of B. The pattern of the first sub-phrase of A, with its repetitive syllables /Hopp/ and

the rhyme-pair /Hopp-(Ga-)lopp/, seems to be more easily accessible for Ulla.

From Ulla’s timing of the song’s phrases, we can conclude that she segments the

song into coherent units, she uses contrastive features to mark the boundaries, but

she still fails to identify a relevant similarity between two dissyllabic words at the

ends of the sub-phrases B that would allow cross-structuring of the sub-phrases. Her

solution is to ignore the two parts of phrase B and produce only one.

Next, we focus on the durational regularities of Ulla’s singing. Ticks on the x-

axes in figures 3 and 4 represent the onset time measures of the notes or syllables. We

see that Ulla repeatedly prolongs the first three notes (quarter notes in the model),

and within each of these patterns, the timing is locally regular. But comparing the

patterns, we see that their tempos differ. All the subsequent notes should

theoretically be last half the duration (eight notes). But Ulla’s timing is again rather

irregular, except for two parts: her fourth sub-phrase in figure 3 (second part of B)

and part B in figure 4. We conclude that Ulla did not yet know about maintaining a

steady beat. Some notes are regularly subdivided and some parts are regularly

timed, but others are not.

The organisation of the second feature of the meter, the accents, should be studied

together with the linguistic elements. Periodic accents allow the metrical pattern to

be identified perceptually. Accentuation is through one or by combinations of the

following three parameters: loudness, pitch, and duration.20 Accents only work in

relation to other constituents at the same level and never in isolation.

������������������������������������������������������������

.20 E.g. Matthew Y. Chen, Toward a grammar of singing: Tune-text association in Gregorian chant, in: Music Perception 1 (1983), 84–122.

Stadler Elmer, Stefanie, KV1

Some linguistic aspects

The metrical pattern of the target model song in figure 2 is visualised with

vertical lines. As a rule in singing children’s songs, each note matches a syllable.

Hence, the stress pattern of the musical meter has to be congruent with the prosodic

stress pattern. There are three kinds of accents: strong, medium, and weak. In all

Western music pieces and songs, each bar starts with a strongly accentuated beat.

Accordingly, the words of the lyrics are positioned with respect to accents:

phonologically strongly accentuated syllables are synchronised with strong and

medium musical beats, and weakly accentuated syllables coincide with weak beats.

We have already seen in the previous sections that Ulla was already able to

organise pitch and time according to some rules. Otherwise, the song would not have

been identifiable as a particular song. The problem for a child at the very young age

of one year and eight months is the lyrics of the song. At that age, a child has not yet

discovered all the prosodic rules that govern word formation.21 She or he will still

produce monosyllables as words, and reduce or ignore combinations of consonants

that are difficult to articulate.

In the present context, I have to restrict this complex phonological issue to the

following question: How did Ulla produce the linguistic elements of this song at a

development level before minimal word formation occurs?

Table 2 contains the song’s lyrics, the syllables subsequently produced by Ulla,

and a synopsis of her phrases A and B. The song’s lyrics and Ulla’s mother tongue

are both German. As described above, Ulla omitted an entire phrase, but she did not

omit or add either a syllable or a note. Hence, the number of notes and syllables are

congruent according to the stress pattern rule. There is one small exception: in line 10

in table 2, the second of her repetitive /bob/ is incomplete. We hear how she missed

the modelled pronunciation, but she kept on going without hesitation. Possibly she

wanted to maintain the meter.

We next investigate how she created the syllables that would result in German

words. A comparison of the first two lines show that she replaced the sound /h/ with

/b/, she omitted /pf/, /ch/, /n/, /l/, /f/, and the second part of the diphthong /au/ of /lauf/,

the vowel /u/. Her reduced words retain mainly the vowels, and she correctly placed

the stress accents. What is striking is how she used the trochaic stress pattern,

consisting of a strong accent on the first syllable followed by the Schwa-syllable in

her approximation of /Pferdchen/ with /`e-gäi/ or something similar. She correctly

pronounced another accent in /`e-ga `la ga-`lop/, where she stressed the one-syllable-

word /la/, which she substituted for the 3-moraic /lauf/ and /bob/ instead of /Hopp/.

������������������������������������������������������������21 Zvi Penner and Christian Krügel, Von der Silbe zum Wort: Rhythmus und Wortbildung in der

Sprachförderung. Bildung von Anfang an. Troisdorf 2006.

Stadler Elmer, Stefanie, KV1

She also correctly accentuated the syllables that correspond with the strong beat

of the musical meter, hence creating a sequence of contrasting strong and weak

syllable-to-note patterns. All the five A-phrases she produced were transcribed and

are listed in the left column in table 2. This list shows that she made only minimal

variations across the productions of this phrase A. As mentioned above, the

repetitive syllable or word /Hopp/ may have served as a kind of scaffold.

Phrase B is more complicated because we do not know which parts - B1 or B2 -

Ulla intended to create. All her syllables forming phrase B are also listed as a

synopsis in the left column in table 2. Each of the verse lines consists of eight

syllables, divided by two sub-parts with four syllables. Each of these sub-parts starts

with a stressed syllable, followed by a weak syllable, /`e-gä/ or a small variation on

this, and then the same pattern again, but with large consonantal variations and

some variations in the vowels. The stress pattern is strikingly even and regular.

However, the variations in the syllables in the second parts of the sub-verses reveal

no clues about which of the B-phrases she intended to produce.

There are two possible interpretations: a) as mentioned above, Ulla had not yet

discovered the rhyme-pair /Steine/ - /Beine/ (see figure 2). This rhyme could have

served as a scaffold for producing a repetition and for elaborating further the

structuring of the entire verse lines. Alternatively, b) apart from still missing the

rhyme in her part B, Ulla may have made the variations in order to fill in the

temporal slots and to keep the meter going. Her repetitive production of phrases AB

may indicate that one of her focuses was on maintaining the temporal framework

with its periodic metrical pattern, and on filling it with trochaic syllable-patterns. Her

approximation of the song’s language was influenced partly by her producing

syllables that tend to adopt the vowel of the target syllables, by her using the pattern

of strong and weak accents given by the binary musical meter, and by her omitting,

reducing or substituting the articulation of consonants. This phonological level has

not yet been exhaustively analysed.

Stadler Elmer, Stefanie, KV1

Tab. 2: The song’s lyrics and the syllables produced by Ulla during the 53 seconds recorded. The syllables during the first 30 seconds are also depicted in figures 3 and 4. Her pronunciation is transcribed with normal letters (not phonetic transcription), and her accents are marked.

The song’s lyrics Lyrics sung by Ulla 1;8 during 53.2s

A: Hopp, hopp, hopp,

Pferchen lauf Galopp

B1: Über Stock und über Steine

B2: Aber brich dir nicht die Beine

A: Hopp, hopp, hopp,

Pferchen lauf Galopp.

Synopsis of Ulla’s parts A and B

Part A:

1) A: bob bob bob

2) `e-gäi `ja ga-`lop

4) A: bob bob bob

5) `e-gäl `la ga-`lop

7) A: bob bob bob

8) `e-gal `la pal-`lop

10) A: bob bh bob

11) `e-ga `la ba-`lop

13) A: bob bob bob

14) `e-gäl `la ga-`lop

Part B :

3) `e-gä `nai-nä `e-gä `lai-jä

6) `e-di `ma-jä `e-gä `mai-jä

9) `e-gal `lai-jä `e-ga `pa-po

12) `e-ga `lai-jä `e-ga `lai-ja

15) `e-gä `nei-jä …

1) A: bob bob bob

2) `e-gäi `ja ga-`lop

3) B : `e-gä `nai-nä `e-gä `lai-jä

4) A: bob bob bob

5) `e-gäl `la ga-`lop

6) B: `e-di `ma-jä `e-gä `mai-jä

7) A: bob bob bob

8) `e-gal `la pal-`lop

9) B: `e-gal `lai-jä `e-ga `pa-po

10) A: bob bh bob

11) `e-ga `la ba-`lop

12) B: `e-ga `lai-jä `e-ga `lai-ja

13) A: bob bob bob

14) `e-gäl `la ga-`lop

15) B: `e-gä `nei-jä …

Stadler Elmer, Stefanie, KV1

5. Final remarks

Early song singing is a normal phenomenon in human children all over the

world. Song singing as a musico-linguistic event exhibits just one among other

symbolic and cultural tools a child grows up with. But it is one of the earliest

instances of all the arts. Children’s songs follow rules concerning the dimensions of

time (phrases, duration, and accents) and pitch (selecting and sequencing more or

less stable pitch categories), and the lyrics normally consist of syllables forming

words, rhymes, and text. It would be very interesting to find out how a young child

applies these rules while making a song, and whether such song singing provides

clues about whether the musical or linguistic features and rules prevail.

This case study of a girl’s song singing at the age of one year and eight months is

just an example of the potential of research in such a stimulating environment. Using

a micro-genetic and acoustically based analysis, it was possible to show that she was

able to create a song that follows the main conventions of children’s song singing.

Ulla’s song could clearly be identified as related to the song model, mainly because

of the following features:

x the shaping of the melodic contour was similar to that of the model;

x the way she segmented the phrases and marked their boundaries using pauses or

rhymes;

x her adoption of contrasts between long and short notes;

x her use of regular stress patterns with strong and weak syllables or notes;

x her approximation of the target syllables using vowels in a onomatopoeic

manner;

x her application of some intra- and inter-phrasal rules to enhance coherence;

x her use of repetitive patterns at the levels of pitch, syllables, phrases, and

contours.

She applied all these rules sufficiently well to create coherence, with, however,

still many instabilities and inconsistencies. As expected, the main challenge or

weakness was how she articulated syllables and words. For Ulla, these linguistic

elements were clearly less differentiated than the musical ones. Ulla ignored,

omitted, or replaced consonants still difficult to pronounce. She also missed

identifying two words as a rhyming pair that would have provided a structure to the

endings of two sub-phrases. The phonological analysis of Ulla’s sung syllables is not

yet exhaustive, but it seems she paid most attention to vowel sounds and their

accents, and her syllables were not yet forming words or word-like units.

Stadler Elmer, Stefanie, KV1

What is also striking was her motivation to learn this song, and repeat it. She

challenged herself with the complicated task of bringing together simultaneously

several features according to some rules. One of the basic rules seems to have been to

frame phrases temporally with periodic accents and contrasting durations, and

another was to approximate the melodic contour. Both rules are organised in

symmetrical forms that are filled in with variable linguistic features. Her early

mastering of these rules may indicate that the rule-based system of song singing

facilitates the synchronisation of musical rhythm with prosodic rhythm. It seems that

the temporal and melodic frameworks can act as a scaffold to make the production

keep going and to shape the beats with pitch patterns and syllables.�