A battery of age-appropriate musical tasks was designed for this study. All tests of musical ability were presented to the child as “musical games” and included positive feedback after each trial to increase motivation. Two types of musical abilities tasks were included: musical perception and musical production tasks. These were developed based on previous research on the assessment of musical abilities in children (e.g., Gordon, 1986; 1989; Peretz et al., 2013) and adults (e.g., Law & Zentner, 2012) and on a considerable number of studies examining music preferences, perception and production abilities in young children (e.g., Fancourt, Dick, & Stewart, 2013; Jensen & Neff, 1993; Morrongiello & Trehub, 1987; Trehub & Hannon, 2009). Musical perception tasks included the assessment of pitch, melody, rhythm and tempo perception, while musical production tasks included the assessment of singing and tapping along to a beat.
A pilot study was conducted between January and March 2015 in two of the participating nurseries to evaluate feasibility of the musical measures. The following two sections elucidate the construction of the tests and report relevant pilot study observations.
2.2.2.1 Design of music perception tasks.
All music perception tasks (pitch, melody, rhythm and tempo perception) followed the same format. Insights sought in the initial pilot exploration of the task format included feedback on: [a] designing the tasks in a format that would not be cognitively challenging for children of this age group (i.e., the majority of 3- and 4- year-old children would be able to understand and perform the task), [b] making the tasks pleasant and fun for the children in order to motivate them to complete the tasks.
The format that was explored was based on a 2+1 oddity paradigm that has been successfully used with children between the ages of 3.5- and 4-years (e.g., Jensen & Neff, 1993; White, Dale, & Carlsen, 1990). In this type of task children are presented with three aural examples: a standard tone or short melody followed by two alternatives. Three identical colored shapes are presented with the tones/melodies and the child is required to select the one that sounds different (or the same) as the standard stimulus. This procedure has been shown to be a powerful method of non-verbal assessment of auditory discrimination in young children (Jensen & Neff, 1993; White et al., 1990).
The task was designed using E-prime software. To test whether children followed the format of the task, short and undemanding musical stimuli (i.e., three note melodies where the “change” stimulus included changes in both contour and pitch direction) were embedded into the tasks. Each auditory stimulus (either environmental or musical sound) was matched to the position of a colorful shape on the screen (all three shapes/colors were identical in each trial to avoid responses based on visual preference) and children were required to point to the shape that sounded “different”. To ensure that children understood the concepts of “same” and “different” a series of visual stimuli where the child had to identify the “same” and the “different” one was presented prior to the auditory task. More specifically the picture of e.g., a dog was presented at the top of a piece of paper while the picture of the same dog and the picture of a cat was presented at the bottom (see Figure 2.1). The child was then asked to point to the one that is the same and then the one that is different from the picture at the top.
Twelve participants (7 boys) between the ages of 3.4 and 4.7 years were tested in this format. According to the observations gathered during the exploration of this paradigm: [a] many of the children tested would appear to loose interest overtime and respond randomly, [b] positive verbal feedback provided by the experimenter and visual feedback on the screen (colorful “thumbs up” picture) did not appear to increase motivation or spark enthusiasm.
With the aim of creating a more appealing task with stimulating feedback the following format was conceived while all perception tasks (pitch, melody, rhythm and tempo perception) followed the same procedure: In each trial the child listened to a musical element corresponding to the drawing of a little girl named Maggie that appeared at the top of the computer screen (see Figure 2.2; Maggie’s picture appeared and the melody or pitch was heard simultaneously). Two identical shapes would then appear successively on the lower left and right sides of the screen corresponding to
another two melodies or pitches. One was always the same while the other was always different to Maggie’s and the child was asked to point to the one that sounded the same. Position of appearance (whether the ‘same’ item would appear on the left or right) and order of appearance (whether the ‘same’ item would be heard first or second) was counterbalanced. All stimuli were presented at a standard volume of 75db. This volume was chosen to ensure that the stimuli would be well above the volume threshold for all participants and was similar to previous studies with young children (Jensen & Neff, 1993; Thompson, Cranford, & Hoyer, 1988). Stimuli in each trial were separated by 1- second silent intervals while inter-trial intervals varied in order to ensure that the child was attentive before each trial (each trial was initiated by the experimenter). Order of trials in all tasks was randomized across participants. Positive feedback included the image of Maggie clapping while little stars appeared out of her hands. Furthermore, children were provided with a small card at the beginning of the task on to which they were allowed to apply a star sticker of their choice each time they responded correctly. Negative feedback included the image of Maggie frowning slightly. The help of an expert animator (Middlesex Art & Design student Eleonora Quario) was sought for the creation and animation of the Maggie character. All tasks were designed and run using E-prime software. A score of 1 was assigned after each correct response and a score of 0 was assigned after an incorrect response.
To ensure that the children fully understood the procedure prior to the task, they were first administered four practice trials in the musical perception task format. The 4 practice trials included easily identifiable sound stimuli (e.g. a dog barking, cat going miaou) and the child had to complete 3 out of 4 trials correctly. Participants were excluded from the study if she/he did not meet this criterion on the first session. Only one child failed to reach the criterion. Moreover, three practice trials were included before each musical perception task. The child had to identify 2 out of 3 practice trials correctly in order to move on to the test trials.
Figure 2.1. Example of picture presented to the child to ensure that the concepts of “same” and “different” are understood.
Figure 2.2. Visual configuration appearing in the four music perception tasks (shapes and colours differ among trials).
It is important to note that Audie’s test (Gordon, 1989) being the only published test suitable for this age group was piloted in the present investigation (n = 10) in parallel to the aforementioned tasks. An observation of children’s manner of responding to the task however, revealed a proneness to inattention and difficulty in maintaining interest (e.g., children often initiated conversations during the task). Furthermore, some of the children who exhibited good performance in the computerized odd-one-out task created for this study, responded randomly to Audie’s test (i.e., 5/10 correct). Given that audiation is a concept that has not been empirically tested, there is a possibility that contrary to Gordon’s prediction, this task requires children to hold a melody in memory across 10 trials and compare it to new stimuli thereby imposing a significant cognitive
load on young children, causing their attention to wane. This assumption can be corroborated by empirical research suggesting that the introduction of new tonal stimuli may mask recall of previously introduced tones and melodies (Allen, 2013; Deutsch, 1970; Massaro, 1970)
Design of stimuli
In the melody, rhythm and tempo perception tasks, computer-generated melodies composed by the author were used as stimuli. The piano sound from Garageband software was used due to its familiarity to most listeners of this age group and for its clear and brisk timbre. Individual pitches (pure tones) were used in the pitch task. During piloting, different versions of the stimuli in each task were presented to the children and approximately 5 to 10 children were tested on each version. This was done to ensure that the stimuli constructed would be neither too easy nor too difficult for the children, in other words, to ensure that the stimuli in each task would be sensitive enough to identify individual differences. More specifically, in each perception task we would expect most children to perform [a] above chance i.e., respond correctly to more than 50% of the trials and [b] the majority of children (60-70%) to respond correctly to 50% to 90% of the trials while a smaller number of children (10 to 20%) should respond correctly to >90% of the trials. This calculation was based on score distributions of several standardized subtests of the Wechsler Preschool and Primary Scale of Intelligence IV (WPPSI-IV, Wechsler, 2012) where the theoretical values of the normal distribution (i.e., percentile ranks) can be used to calculate the proportion of children scoring at different levels. Note that this was a rough calculation used in the pilot phase with the aim of determining test sensitivity based on very small samples. This calculation took into account the need to ensure that the children scored above chance in the tasks. Therefore only distributions of subtests where the majority of scores (80%) was above the .50 cut off representing chance performance were examined.
Final sets of stimuli for all music perception tasks are available in Appendix B. Pitch perception stimuli. Ten stimuli were created based on previous pitch perception tasks for adults, young children and infants (Law and Zentner, 2012; Maxon & Hochberg, 1982; Olsho, 1982; 1984). The stimuli were sinusoids, or pure tones generated using the Audacity software. Pure rather than complex tones were used because pitch changes in complex tones can be perceived as a result of harmonics rather
than fundamental frequency (Licklider, 1954). Following Maxon and Hochberg’s (1982) experiment with 4-year-old children, the duration of the pure tones was 400ms with a 25ms linear onset and offset ramp. The standard stimulus was 1000 Hz. This frequency was chosen based on previous work with infants indicating that discrimination ability is superior for high rather than low frequencies (Olsho, 1984). Furthermore, it has been used as a reference frequency in a number of experiments both with children (e.g., Agnew, Dorn, & Eden, 2004; Bobin-Bègue & Provasi, 2005; Thomson et al., 1999) and with adults (e.g., Paraskevopoulos, Kuchenbuch, Herholz, & Pantev, 2012; Tillmann, Janata, & Bharucha, 2003; Weiss, Granot, & Ahissar, 2014). Comparison stimuli differed in frequency (lower or higher) and were pivoted around 1000 Hz. Fifty per cent of the comparison stimuli were at a lower pitch, while 50% of the comparison stimuli were at a higher pitch compared to the standard stimulus. The first comparison stimulus represented the easiest trial and differed from the standard by 120 Hz while the difference between comparison and standard stimuli in the remaining nine trials ranged from 60 Hz to 12 Hz (the difference decreased across nine stimuli). Three training trials preceded the task where comparison stimuli differed by 200 Hz from the standard stimulus. This range of differences in pitch height represented a middle ground between a difference threshold range of 6 to 57 Hz for infants (Olsho, 1984; note that infants were tested with implicit head-turn procedures that are more sensitive in tapping perception) a threshold range of 25 to 64 Hz in 5-year-old children (Thomson et al., 1999) and a threshold of 12 Hz for 4-year-old children (Maxon & Hochberg, 1982). Adult levels of pitch discrimination can range from differences of 7 to 12Hz (Law & Zentner, 2012).
Thirteen children (7 girls, ages ranged from 3 years 6 months to 4 years 9 months) tested with this set of stimuli produced variability in their scores (nine children scored between 6 and 8 out of 10, two scored ≥9 out of 10 and three children scored ≤5 out of 10), suggesting that this set of stimuli is sensitive enough to differentiate between young children with different levels of pitch discrimination ability. Interestingly, 4 out of 13 children who failed to detect the easiest difference (120Hz) were at the lower end of the score distribution (≤6).
Melody perception stimuli. Twelve melodies were composed by the author for the melody discrimination task. Melodies were 3 to 5 tones long ranging from 1 to 3 seconds in overall duration. To make stimuli more engaging, each melody was
originally composed in C major but was then transposed into a different musical key (all major scales were used). Melodies were generated at a tempo of 140bpm, which is close to the spontaneous motor tempo (i.e., manual tapping task at the most comfortable tempo) of children in this age group (Gérard and Rosenfeld, 1995; Provasi & Bobin- Bègue, 2003). This is thought to coincide with an optimal sensitivity zone or referent period where processing of musical intervals becomes more accurate (Jones, 1976; Jones & Boltz, 1989).
Differences to be detected in the comparison stimuli consisted of one-tone changes. Difficulty was manipulated across two levels: [a] the length of the melodies (stimuli included six 3- and six 5-note melodies) and [b] changes in comparison stimuli were either contour-violating or contour-preserving. This was based on previous work showing that 4- to 6-year-old children can more readily identify contour-violating compared to contour-preserving transformations in short melodies (Morrongiello et al., 1985). Because children of this age might have acquired differential levels of key membership (i.e., understanding which notes belong in a key and which do not) and implicit harmonic knowledge (i.e., understanding which notes are more likely to follow others in a musical key; see Corrigall & Trainor, 2013) and to avoid any possibility that changes are more salient for some children due to out-of-key or out-of-harmony violations, all changes were kept within the key and harmony of the standard melody.
Another issue of concern was the position of changed tones within the melody. Research findings regarding the role of the position of tones in memory retention has been inconclusive, with some adult studies suggesting that initial and final tones are easier to remember than middle tones (Ortmann, 1926; Williams, 1975; Silverman, 2010) and other research demonstrating that memory in both children and adults is superior for final compared to initial tones (Bentley, 1966; Ross, Olson, Marks, & Gore, 2004; Siegel, 1974). Overall, it appears that the final position is clearly advantageous for memory retention, whereas it is less clear whether memory is facilitated for the first tone of a melody. Since changes in the last tone would interfere with the tonality of the sequence, changes in comparison melodies were limited to one tone in the middle of the melody (i.e., 2nd tone in 3-note melodies and 2nd, 3rd or 4th tone in the 5-note melodies).
Pitch interval changes ranged from 3 to 12 semitones and included both upward and downward changes. The average pitch interval changes were equivalent across 3- and 5-note melodies with a mean of 6.5 (range: 3-12) and 6.2 (range: 3-12) semitones respectively. Average pitch interval changes were, however, different between contour-
violating (mean of 9.3 and range of 8-12 semitones) and contour-preserving stimuli (mean of 3.8 and range of 3 to 5 semitones). Out of six children tested on the set of stimuli (5 boys, ages ranged from 3 years 5 months to 4 years 2 months), one child had very good performance (identified >90% of trials correctly), four identified 60% to 80% of the trials correctly and one child exhibited poor performance (<50%), suggesting that the stimuli might be sensitive in identifying different levels of melody discrimination ability.
Rhythm perception stimuli. The rhythm perception stimuli were created based on the Montreal Battery of Evaluation of Musical Abilities (MBEMA) (Peretz et al., 2013) for 6- to 8-year-old children. The same melodies as the melody discrimination task were used in this task. Differences to be detected in the comparison stimuli consisted of changes in the duration of adjacent tones. This manipulation altered the rhythmic grouping of the comparison melody while preserving the number of notes, overall duration and meter of the standard melody (see also Peretz et al., 2013). Difficulty of the stimuli was manipulated across three levels: [a] the length of the melodies (as in the melody discrimination task), [b] by changing duration of either two or three tones in a sequence and [c] changes in duration occurred either on the downbeat (easy trials, 50% of stimuli) or on the upbeat (more difficult trials, 50% of stimuli) of the melody’s meter. Out of the five children tested on this version of the stimuli (two boys, ages ranged from 3 years 7 months to 4 years 8 months), one child had very high performance (identified >90% of trials correctly), three children identified 60% -80% of the trials correctly and one child exhibited low performance (50%). This suggested that this set of stimuli might be sensitive in identifying variability in rhythm discrimination ability.
Tempo perception stimuli. Ten 4-note melodies were composed by the author for the tempo discrimination task. Four, rather than 3-note melodies (as in part of the rhythm and melody discrimination tasks) were used to ensure that the children adequately registered the tempo of each melody. All melodies were composed in the key of C major but were then transposed into different musical keys to make the stimuli more engaging (10 major scales were used). The difficulty level of this task was manipulated by varying the degree of the differences in tempo between the standard and the comparison stimulus. In other words, all standard stimuli were 100bpm whereas comparison stimuli were either slower or faster, with differences in tempo rates
decreasing linearly. The rate of 100bpm for the standard stimuli was chosen based on previous experiments showing that this is the optimal sensitivity zone for tempo discrimination in both infants (Baruch & Drake, 1997) and adults (Baruch, Panissal- Vieu & Drake, 2004). Furthermore, the only study that has so far explored tempo discrimination thresholds in children of 3- and 4-years of age has used 100bpm as a reference against which discrimination thresholds were estimated (Bobin-Bègue & Provasi, 2005). This is a very useful reference for the creation of the current task, given that it provides a guide for determining tempo differences in the comparison stimuli. According to the results of this study, an absolute difference of > 20bpm can readily be detected by the majority of children between 3 and 4 years of age (both 3- and 4-year- olds performed above chance in detecting this difference) but children’s success rates decreased with increasing difficulty (i.e., as differences in tempo decrease). Furthermore, 3-year-old children performed at chance in detecting an absolute difference of 15bpm but were above chance performance in detecting a 25bpm difference (Bobin-Bègue & Provasi, 2005). Based on these results, a first draft of stimuli was created where the differences in comparison stimuli ranged from 25bpm to 8bpm, revolving around a standard tempo of 100bpm. The rationale was that a set of stimuli with differences close to the discrimination threshold for this age group would efficiently identify variability in tempo discrimination ability.
The first draft of stimuli was tested on five children (4 girls, ages ranged from 3 years 7 months to 4 years 4 months). Their performance ranged from 30% to 60% correct (3 out of 5 children performed ≤ 50%) suggesting that this set might have been too challenging for children of this age group. Another set of stimuli was then created with differences in tempo between standard and comparison stimuli ranging from 18 to 50bpm. Seven children were tested in this version of the stimuli (1 girl, ages ranged from 3 years 5 months to 4 years 8 months) with 4 out of 7 children performing at chance or below (≤ 50%) while no child responded correctly to 90% of the trials or above. This suggested that tempo discrimination in these groups of children might not follow the same thresholds as in the Bobin-Bègue & Provasi (2005) study, presumably because of differences in the nature of the tasks. Their task differed in the following ways: [a] the stimuli used were 10-tone sequences of identical tones rather than melodies, [b] each trial presented an individual sequence which was either faster or slower than 100bpm and the child had to respond by pressing the left or the right button.