Investigating heart rate and rhythm changes in an infant’s music education course: A case study

Participants

Six (five female, one male) healthy, full-term infants, taking part in a music education course in Greece specifically designed for their ages (see below), were studied. Verbal and written informed consent was obtained from the parents before the start of the experiment, explaining both the educational and the data collection processes. The ages of the infants were from 14 to 25 months. Two age groups were formed for the needs of the study, following the aforementioned ‘Dynamic Skills Theory’ segregation of cognitive development: the ‘younger toddlers’ (1Ba; 2Ba; 3Ba) and the ‘older toddlers’ (1N; 2N; 3N); 14–18 and 21–25 months for each group, respectively. For the younger toddlers group (M = 16, SD = 2.0 months of age), all subjects had a two-parent household history showing, according to the parents’ statements, an average income and life background (OECD, 2013), typical development (not known disorders) and no prolonged injuries. A 95.83% rate of attendance (SD = 0.0721) for the 10-episode research period was recorded for the younger toddlers. For the older toddlers group (M = 23, SD = 2.0 months of age), all subjects had a two-parent household history and an average income and life background (OECD, 2013), typical development (no known disorders) and no prolonged injuries as well. A 91.67% rate of attendance (SD = 0.1443) for the 10-episode research period was recorded for the older toddlers.

Music course and study design

A toddlers’ music education course was specifically designed for the study. An educator, the toddlers, plus one of their parents participated in each one of the episodes of the course. An active and experienced primary and toddlers’ music education professional (15 years of practical classroom experience) being naïve of the study’s hypothesis, collaborated with the researcher to develop content for a 10-episode music education course, using 10 activities, targeting the ages of 0.5 to 3 years. The specific age range was employed in order to compensate for cognitive maturation differences, as according to the Dynamic Skills Theory paradigm followed in this study, the second tier of cognitive maturation the ‘older toddlers’ group belong to may extend slightly beyond the first 24 months of life.

In accordance with the research design, parents were not expected to play a vital intercommunicative role. They were asked to just follow the educator’s rhythmical lead as if they were on their own in the relevant educational context, without hence interfering with their child’s performance or reactions. This latter directive was, of course, waived in the case of other nursing reasons irrelevant to the course content or process (i.e., excessive crying or feeding issues). According to the aforementioned outline, parents were approached as observers/mimicking followers of the course content, and thus an independent variable in the research design system. They were therefore expected to move along with the educator’s and toddlers’ intercommunicative system without providing any exogenous force to it. For this latter reason, too, the parents’ reactions or involvement was not studied as a separate research element.

During the course design process, the Greek cultural context was considered, whilst certain safety and health elements were also maintained. A classroom, 9 × 6 m², was used, which was well-ventilated, with no loose or sharp-ended objects. The course site was fully covered with baby-mats. The aim for each episode was to not exceed 50 min, while 7–10 min of time before and after each episode was also included in the course design to enable the collection of biorhythmic (cardiac rhythm), nutrition and sleep data.

The 10 episodes were implemented once per week in a consecutive 8 + 2 week period (interrupted by Christmas holidays). Episodes were run in parallel for both groups, taking usually place in the late morning–afternoon (10:30 a.m. to 1:00 p.m.), except the last two episodes, which took place in the late afternoon (5 p.m. to 7 p.m.). The researcher controlled for a relaxed (resting) HR state to occur in the immediate period before each episode for all participants; therefore no HR data distortion was expected due to the aforementioned timing differentiation. The participants were invited to arrive on-site 15 min prior to the implementation of the episode to avoid any unnecessary HR fluctuations. As there is no relevant evidence in literature suggesting different resting HR numbers in humans in different parts of the day, it was expected that the toddlers would reach their resting HR in these 15 min, hence starting from a similar cardiac state baseline before each episode.

As far as the course content was concerned, a student-centred approach (O’Neill & McMahon, 2005) was followed, providing ample opportunities to infants for hands-on, sensory-motor, and emotional experiences as well as for educator-infant interactions. The 10 activities comprising each episode of the course were common for both groups and implemented in the same order for each episode. A set of different musical instruments and sound-producing objects or techniques (n = 21) ² relevant to primary and preschool education were used during the episodes’ delivery.

The course was philosophically-based on the GIML method for infants’ and toddlers’ music development (http://giml.org) adapted thereafter for the project’s specific population. The activities in each episode presented rhythmical structures, covering sound exemplifications, singing, imitation, rhythm games, scales, spatiotemporal movements coordination, body mapping through sound and movement coordination, examples of speech prosody and wording, cradling and relaxation (see Table 1 for full list of the 10 activities, plus the two examples in what follows). It was not expected that the toddlers should fully imitate or follow the content, although maximum exposure to it was a desirable aim.

OPEN IN VIEWER

Table 1.

Episodes 10 activities common to each group.

Name	Content	Tool/s used
Forte and Piano in context	Playing with Forte and Piano terms, exploiting rhythmical structures; alteration of dB levels in a rhythmical context.	Boomwalkers, Bongos, Rhythm Sticks, Drums, Wooden crow sounders with mallets, Tambourines
Staccato and Legato in context	Playing with Staccato and Legato; alteration of a sound’s duration in a rhythmical context.	Harmonicas, Kazoos, Boomwalkers
Rhythmical scales	Scale playing; rhythmical representation of various frequencies.	Xylophones, Piano, Harmonicas
Space and rhythm	Rhythm embodiment activity; moving in space rhythmically.	Body movements
Phonemes and syllables	Rhythmical representation of speech sounds; syllables and phonemes enhanced by the sound of instruments	Speech and singing, Wooden blocks, Scarves, Teddy bears
Slow and fast	Representation of identical rhythmical structures in various tempi; focus on attentional performance	Triangle, Maracas
Story telling	Presenting speech rhythm and prosody	Narrative approach of rhythm
Melodic line	Presentation of various melodic lines (synthesis of rhythm and frequencies)	Xylophones, Piano
Singing intervals	Songs presented in between the activities as relaxation intervals	Piano, Xylophone, Bongos, Body movements
Relaxation	The end of each episode; a slow song offered at the end of each episode	Singing

Activity example 1

The first example is ‘rhythmical phonemic games’, where phonemes or syllables like /α:/, /Ʊǝ/ and /dα:/ were presented by the educator in different rhythmical groupings (e.g. three crochets and a pause in a 4/4 measure) or single rhythmical modes (e.g. a phonemic stimulus matched to a semi-quaver) for a few minutes in each episode. The activity started with its ‘pure’ version, where the phonemes were individually presented in a specific pitch and rhythmical mode. After a few minutes and repetitions, the same phonemes were combined in mixed rhythmical groupings following the same pitch, while in the enriched version towards the end of the activity, there was a mixed grouping of phonemes and rhythm, changing pitch randomly while co-ordinating the phonemic presentation with the sound of an instrument or movement. This instrument or movement – like the wood blocks or the movement of a scarf in space using one hand or a foot – was progressively introduced in rhythmical coordination with the phonemic sound, enhancing multimodally the rhythmical stimulus. The core aim of this activity’s content was to cover different sound exemplifications through gradually evolving rhythmical games. The parent present in this activity was helping towards a maximum stimulus exposure by fully imitating the educator’s presentation.

Activity example 2

Another example is the ‘illustrating song’ which was combined with specific body movements, aiming to get the toddlers to reflexively imitate the different sound/rhythm movements in play. This activity exploited specific movements mainly belonging to the Central Pattern Generators (CPG) neuronal network (Hooper, 2000). This neuronal network underlies the production of most rhythmic motor patterns in mammals, while it is suggested that for repeating them, there is no need for an intrastimulus evoked potential – that is, a cognitive decision. In the ‘illustrating song’ activity, when, for example, a rhythmical structure of repeatable crotchets was found in the song, it was matched to hand slapping the floor using an analogous repetition rate and rhythm (vertical hand movement belonging to the CPG network; that is, movement of the hand during walking). If another group of crotchets was found after the first one, then the other hand would be used, alternating hand movements accordingly. If a group of quavers was found, this was then matched to the movement of clapping both hands (clapping mode of the CPG network) following the same pattern across the whole song.

Data registration rationale

As it is hard to fully isolate the exogenous rhythmical elements of the course content (i.e., random movement sounds, random speech, crying etc.) in a non-laboratory based music study, it was decided for all episodes to be fully videotaped, digitally recording their continuous rhythmical soundscapes for further investigation and analysis (see the Data measurements and analysis section for a detailed presentation of analysis).

Nevertheless, it should be noted here that the study’s specific intention was not to individually record, analyse and compare the ten activities’ beat and tempo components, rather, it tried to approach the episodes as a continuous real-life music education environment and construct, thus registering the episodes’ summative rhythmical handling trajectory from start to end. In the recording and data analysis process, therefore, all other possible manifestations of beat and tempo elements and fluctuations were allowed and included, even if originating from random speech, crying, movements, discussions, and random beat and tempo occurrences emanating from the outer classroom surroundings, during or in between the activities. Based on this environmentally holistic rhythmical registration rationale, the pre- and post-episodic registration of HR was also used in preference to the pre and post activity HR registration. The aim was to capture the true – non lab-based, that is – summative impact on the infants’ HR of the whole-episode rhythmical soundscape.

Instrumentation

A digital Toshiba (Camileo BW10) video camera was used for the videotaping process, having attached a condenser microphone (Audio-Technic®, ATR25 condenser microphone) for a more detailed raw sound recording. The camera and microphone were placed in the same physical spot in the room for each of the episodes in order to avoid sound-recording differentiations and distortions. Video was decoded in an H264 format, for future acoustical and behavioural analysis of the episodes. A sole audio signal was extracted from the digital video recording in an off-time sequence using Audacity software. All audio data was decoded in – and used as – WAV-Lossless files.

In order to measure the infants’ heart rates, a photoelectric finger oximeter (EDAN H10) was used. The specific transducer was properly placed on the right hand index finger for every infant before and after each episode.

Acoustical analysis of the episodes was performed using Sonic Visualiser software (Cannam, Landone, & Sandler, 2010). Sonic Visualiser is a program for viewing and exploring audio data for semantic analysis and annotation. In order to extract two specific audio features in this research project (tempo and beats), two different VAMP plug-ins were used on Sonic Visualiser. ‘A Vamp plugin is a chunk of compiled program code that carries out analysis of a digital audio signal, returning results in some form [picture] other than an audio signal’ (Cannam, n.d., p. 1). In this case, BBC Rhythm-Tempo and Queen Mary, University of London Tempo and Beat Tracker plug-ins were installed on the software. Specific configurations were implemented on certain software parameters to illustrate and then analyse the audio signal through the specific plug-ins. ³ These are shown in detail in the next section.

Data measurements and analysis

In order to validate or reject the study’s hypothesis and answer its related research questions, the researcher decided to perform a two part comparative-descriptive statistical analysis: (a) in between the summative rhythmical data collected from each episode for each group, and (b) in between the summative rhythmical data collected from each episode for each group and the infants’ HRs for each group.

Ten episodes were videotaped and recorded during the course delivery, although seven of them were finally analysed. Out of the 10 episodes, the first two were excluded for reasons of ‘calibration.’ These were perceived during the analysis as the accustomization period for the infants and the educator in the new educational environment. The last of the 10 episodes was also excluded from the analysis owing to the infants’ low attendance in both groups. The final seven episodes studied were parallel in the course delivery timeline. For the ‘young toddlers’ group, 5 h and 26 min were recorded in total through the seven episodes, while for the ‘older toddlers’ cohort, 5 h and 9 min. For the ‘young toddlers,’ a 46 min 41 s average episode duration was calculated (SD = 8.6 min), whilst for the ‘older toddlers,’ a 44 min and 8 s average episode duration was recorded (SD = 11.24 min).

According to the study’s approach on rhythm perception and rhythmical handling, ¹ the following measurements were collected for each group and episode separately:

In order to find and cross-correlate the in-between groups’ rhythmical trajectories, as well as the groups’ rhythmical trajectories to the summative HR trajectories for each group for the seven episodes (hence validating or rejecting the study hypothesis), a beat analysis was initially performed using Sonic Visualiser. The ‘Beat Tracker’ VAMP plug-in was employed with the following parametrical configurations: Beat Tracking Method [New]; Onset Detection Function Type [Complex Domain]; Adaptive Whitening [No]; Window Size [1114]; Window Increment [557]; Window Shape [Hamming].

In order to follow a common basis for each episode, a 5-min continuous and consecutive epoch segregation and analysis approach was chosen (Figure 1). After segregating each episode in 5’ epochs, the number of beats found in every epoch was averaged. The beats trajectory (increase/decrease) for each episode was then presented based on these averaged 5’ epochs. Scatter plots were used and the fitting regression line (R²) was calculated to project the beats trajectory for each episode (Figure 2).

OPEN IN VIEWER

Figure 1.

Example of the ‘beat tracker’ vamp analysis for a whole episode. Data are segregated in 5’ epochs.

OPEN IN VIEWER

Figure 2.

An example of averaged 5’ epochs, showing the line of regression for the specific episode.

In order to measure the tempo aspect of the rhythmical data in each episode, the BBC Vamp plugin was employed, directly providing – through the electronic analysis on the Sonic Visualiser – a number equal to the average tempo for every episode (Figure 3). The following parametrical configurations were used for the tempo analysis: Sub-Bands [7.000]; Threshold [1.000]; Moving Average Window Length [200.0000 frames]; Onset Peak Window Length [6.000 frames]; Minimum BPM [12.000 BPM]; Maximum BPM [300.0000 BPM]; Window Size [1024]; Window Increment [256]; Window Shape [Hamming].

OPEN IN VIEWER

Figure 3.

Tempo averaging example on Sonic Visualiser.

The aim of the tempo analysis was to provide metadata (Table 2) to statistically assess whether or not the overall tempo implementation differed significantly or not in between the episodes. Statistical analysis should show if the overall rhythmical structure of the episodes – usage of beats – had been followed by chance (or not) by the educator. The same process was followed for the 5’ epoch SD of beats for each episode (Table 2), finally cross-correlating the results of the tempo and SD beats for the two groups.

OPEN IN VIEWER

Table 2.

Beats and tempi datasets for all episodes for the two groups (where #a = young toddlers, #Ν = older toddlers).

	Average tempo	Average tonic pitch	Average beat tracker per 5’	SD beat tracker per 5’
Episodes
1a	132	6D	122	11.904
2a	126	7D	119	11.346
3a	105	7D	131	5.837
4a	95.3	7D	121	11.139
5a	121	6D	119	9.386
6a	105	7D	121	10.403
7a	91.5	6D	117	15.236
1N	115	6D	119	14.998
2N	86.5	6D	122	13.344
3N	108	7D	123	11.207
4N	107	6D	123	11.141
5N	111	7D	130	14.664
6N	136	7D	124	11.906
7N	137	7D	121	9.436
		6D	1174.6595 Hz
		7D	2349.319 Hz

In order to fulfil the second part of the study, heart rates for all six infants were registered for all 10 episodes, while a statistical analysis was carried out on the two groups’ HR data by performing a t test. For reasons mentioned above, only seven episodes’ measurements were included in the analysis. Heart rate levels are displayed before (upper reading for each subject) and after (lower reading for each subject) each episode in Table 3. Before performing the t test on the aforementioned data, a MANOVA on the dependent variables (tempo; beats, HRs) and the two independent variables (groups) was first performed, controlling first for significant interactions among the variables.

OPEN IN VIEWER

Table 3.

HRs (bpm) before (upper line) and after (lower line) each episode.

Participant		1st episode	2nd episode	3rd episode	4th episode	5th episode	6th episode	7th episode
1Ba	before	110	118	116	116	111	120	114
	after	112	120	123	116	120	137	118
2Ba	before	100	99	120	97	109	123	144
	after	110	115	135	112	120	130	132
3Ba	before	114	112	111	113	109	121	119
	after	120	121	118	125	119	122	123
1N	before	102	100	109	113	110	109	109
	after	112	106	123	115	117	111	111
2N	before	118	103	101	119	116	118	113
	after	119	104	112	115	115	129	124
3N	before	112	110	117	109	98	114	110
	after	108	120	102	102	100	109	120

Limitations

This study may have produced different results if more groups had been available. It is not suggested, however, merely to include more infants in every available group, but instead to increase the actual number of the groups under investigation. It has already been shown that (especially in educational contexts) group dynamics play a major role in the way individuals perceive their environment, hence learning approaches change according to the level of interdependence links available in the group (Forsyth, 2009, p. 8). Thus, more infants in the group might have structured completely different variables of intercommunication and stimuli exposure for each individual, providing a completely different physiological or acoustical (rhythmical) result. On the other hand, having more groups – hence cross-individual data – to investigate would have certainly provided a wider cross-correlation basis.

Furthermore, it should be noted that there was no control over the wider infants’ exposure to the specific stimuli and learning approach. The specific developmental context included both children and parents in its process. Thus, there is a possibility that some or all of the parents used the sessions’ content to ‘improve’ their parenting. If this was the case, it may or may not have had a certain effect – either behavioural or physiological – on the participating infants during the delivery of the course. The study did not account for this particular research angle, and it was assumed from the beginning a similar base line of exposure for all infants. Still, this is a limitation to be considered, and one should proceed with a certain caution towards generalization even for the specific group of infants.