Edited by: Jennifer MacRitchie, Conservatorio della Svizzera Italiana, Switzerland
Reviewed by: Richard Ashley, Northwestern University, USA; Joe Wolfe, UNSW Australia, Australia
*Correspondence: Alex Hofmann, Institute of Music Acoustics (IWK), University of Music and Performing Arts Vienna, Anton-von-Webern-Platz 1, 1030 Vienna, Austria e-mail:
This article was submitted to Cognitive Science, a section of the journal Frontiers in Psychology.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
This paper investigates the production and perception of different articulation techniques on the saxophone. In a production experiment, two melodies were recorded that required different effectors to play the tones (tongue-only actions, finger-only actions, combined tongue and finger actions) at three different tempi. A sensor saxophone reed was developed to monitor tongue-reed interactions during performance. In the slow tempo condition, combined tongue-finger actions showed improved timing, compared to the timing of the tongue alone. This observation supports the multiple timer hypothesis where the tongue's timekeeper benefits from a coupling to the timekeeper of the fingers. In the fast tempo condition, finger-only actions were less precise than tongue-only actions and timing precision of combined tongue-finger actions showed the higher timing variability, close to the level of finger-only actions. This suggests that the finger actions have a dominant influence on the overall timing of saxophone performance. In a listening experiment we investigated whether motor expertise in music performance influences the perception of articulation techniques in saxophone performance. Participants with different backgrounds in music making (saxophonists, musicians not playing the saxophone, and non-musicians) attended an AB-X listening test. They had to discriminate between saxophone phrases played with different articulation techniques (legato, portato, staccato). Participants across all three groups discriminated the sound of staccato articulation well from the sound of portato articulation and legato articulation. Errors occurred across all groups of listeners when legato articulation (no tonguing) and portato articulation (soft tonguing) had to be discriminated. Saxophonists' results were superior compared to the results of the other two groups, suggesting that expertise in saxophone playing facilitated the discrimination task.
Producing expressive sound on single-reed woodwind instruments is a highly sophisticated motor task, requiring coordination between the fingers, aural cavity, and respiration (Scavone et al.,
To perform expressively on woodwind instruments, the player may use a range of parameters to shape longer sequences of tones such as onset timing and tempo or the loudness of individual tones. An important dimension in woodwind performance is tongued articulation, thus referring to the way the tongue controls the shape of tone onsets, tone offsets, and the connections between tones (see Krautgartner,
Articulation techniques on saxophone can be grouped in two main types: tongued articulation techniques and articulation without tonguing.
Timing precision in the execution of complex movement patterns is essential for musicians to produce rhythm in a sequences of tones. Palmer et al. (
Perception and action in human motor control are strongly connected. The motor theory of speech perception argues that human understanding of speech-based auditory stimuli is based on the ability to recognize related vocal tract movements required to produce equivalent sounds (Liberman and Mattingly,
In this paper, we investigate articulation techniques in saxophone performances in two experiments. In a production experiment, we examine timing measures in relation to effector combinations (tongue, finger, and both), finger movement directions (pressing for tone onsets vs. releasing for tone onsets), and different articulation techniques (legato, portato, staccato). In a second experiment, we test whether motor expertise in a particular field (i.e., saxophone performance) influences the perception of different articulation techniques in recorded saxophone sounds.
Seven female and twelve male graduate saxophone students from the University of Music and Performing Arts Vienna (
Two isochronous 24-tone melodies were designed for the experiment. Both melodies consisted of the same elements (Figure
The experimental set up consisted of a sensor-equipped alto-saxophone, a microphone, a digital metronome, and a multi-channel recording device. Strain gauge sensors (2 mm, 120 Ohms) attached to synthetic saxophone reeds (by Légère Reeds, Ltd.) were used to capture the bending of the reed during performance (Figure
The experiment was conducted in accordance with the Declaration of Helsinki: Participants gave written consent prior to the experiment, played under normal performance conditions, and received a nominal fee at the end of the experiment.
In the beginning of the experiment, each player had to choose a synthetic saxophone reed out of four different reed-strengths (Légère: 2.0, 2.25, 2.5, 2.75). All saxophonists were allowed to use their own mouthpiece but played on the same alto-saxophone (77-SA, by Stagg). The metronome provided the synchronization signal on each quarter-note beat. The introduced tempi were 120 beats per minute (slow, IOI for eighth notes = 250 ms), 168 bpm (medium, IOI = 178.6 ms), and 208 bpm (fast, IOI = 144.2 ms). All participants got a 5 min warm-up, to practice the melody with the metronome at a slow tempo. For the experiment, each participant played both melodies in legato, portato, and staccato articulation. They synchronized with the metronome for two repetitions and continued playing when the metronome was muted, until the melody had been played 6 times in total. We recorded two trials per tempo condition, ordered from the slowest to the fastest. The experiment lasted for approximately 1 h per participant. In total 4644 tones were recorded per player (2 melodies x 2 trials x 3 tempi x 3 articulations; containing 145 tones for portato and staccato and 97 tones for legato). After their performances, the participants filled in a questionnaire about their musical background and the experiences with the sensor saxophone.
Sensor equipped saxophone reeds were used to capture the bending of the vibrating reed during human performance. Figure
The data captured during the experiment contained more than 88,000 played tones, which makes a manual transcription impossible. A multiresolution analysis (MRA) based on wavelet methods has been used successfully for the analysis of various time critical signals, ranging from medical data (i.e., ECG time series, Percival and Walden,
The reed signal was decomposed using the Maximal Overlap Discrete Wavelet Transform (MODWT) of level
To evaluate the quality of the LDF, it was tested on a small data set which contained 2020 manually annotated landmarks. Starting from the annotated ground truth, the existence and number of detected landmarks around the annotated events was checked. The standard measures
To examine timing of the produced sequences, we calculated inter-onset intervals (IOI, in ms), as the time interval between two subsequent TRR (onset) landmarks (
The average signed timing error of all performances during the synchronization phase was close to zero (
The melodies (Figure
We grouped parts of the melodies according to the effectors required for playing and compared onset timing between these parts. A Two-Way repeated measures ANOVA on timing error by effector combination and tempo, indicated a significant main effect of the executing effector [
Fingers only | −0.03332 | −0.03430 | −0.00095 | −0.01010 | −0.01170 | −0.01402 |
Fingers w. Tongue | −0.02675 | −0.02760 | ||||
Tongue only | −0.00666 | −0.00521 | 0.04549 | 0.03438 | 0.04636 | 0.04446 |
A Two-Way repeated measures ANOVA on temporal precision (CV) by effector combination and tempo, showed a significant effect of tempo [
To play descending tone sequences on the saxophone, keys have to be pressed, while ascending sequences require fingers to open tone holes. To see if the direction of finger movements (pressing down vs. lifting up) influences the timing of the performance, we contrasted (legato) sequences with a focus on key depression (Melody 1) to those focussed on lifting the keys (Melody 2). A Two-Way repeated measures ANOVA on timing error showed no significant effect of the direction of finger motion nor any interactions with tempo. The same ANOVA on timing precision showed no significant effects. Similar observations have been reported for clarinet performances by Palmer et al. (
We recorded reed signals of melodies with legato, portato, and staccato articulation (Figure
In the questionnaire, we asked how comfortable the participants felt while playing the sensor-equipped saxophone reed. The reed quality had to be rated between 1 (very good) and 7 (very bad). Results showed that the reed quality was evaluated as medium quality (
We were interested in the abilities of listeners with different expertise in music performance to discriminate between common articulation techniques in the sound of the saxophone. Furthermore, we were interested whether motor expertise in saxophone performance would facilitate the perception of saxophone articulation.
Nineteen female and twelve male (
In a 3 × 3 × 2 × 2 × 2 (3 articulations × 3 intervals × 2 registers × 2 players × 2 listening blocks) design, we tested which articulation techniques our participants were able to discriminate. We recorded note transitions of three different pitch intervals (major second, major sixth, major sixth including register change), with legato, portato, and staccato articulation, within two registers by two different players (one of them also participated in the production experiment, the other is the first author of this paper) on the same alto-saxophone (YAS 32, by Yamaha Corp.), using their mouthpieces (AL3, by Vandoren; Original 7*3, by Claude Lakey). Recordings were made in an anechoic chamber using a microphone (C414, by AKG Acoustics) and Labview hardware and software (DAQ LabView 2011, by National Instruments Corp.) for recording the stimuli (44.1 kHz sampling rate, 16 bit resolution). Both players used synthetic sensor equipped saxophone reeds (section 2.1.3), to ensure tongue-reed contact in the portato and the staccato playing conditions. During the recordings, both players heard a metronome click on headphones (108 bpm for larger intervals, 120 bpm for small intervals), to produce consistent timing in the stimuli. We recorded three eighth-notes (two note transitions, see Figure
Stimuli were presented to the participants in the form of an ABX listening test on a laptop computer. A java-based software program enabled the participants to click on one of 3 buttons (A-Button, B-Button, X-Button) to play back one stimulus. Buttons A and B contained two note transitions played with different articulation techniques. Button X contained a third recording that matched the articulation used in either A or B. The question our participants had to answer was: “Does X sound like A or B ?” Listeners had to decide whether X was more similar to A or B. Responses and reaction time were recorded by the software.
The experimental procedure complied with the Declaration of Helsinki: Participants gave written consent prior to the experiment. All participants worked on the same laptop computer (by ASUSTeK Computer Inc.) in a quiet environment and used the same studio headphones (K121, by AKG Acoustics). The could adjust the playback volume to a comfortable level. Each participant had 5 practice trials to learn how to navigate the ABX listening test software. A pop-up on the screen indicated when the actual experiment began. The experiment was grouped into two blocks, each containing all 36 stimuli in four different listening orders. Participants were allowed to play back the stimuli as often as required to make an assured judgment. After the first block was done the participants filled in a questionnaire about their musical background. Afterwards the participants made another set of judgements. The entire experiment lasted for about 30 min per participant.
Overall, participants from all three groups were able to accomplish the listening test with over 87% of correct answers. A Chi-squared test revealed no significant difference on correct answers between the two repeated listening blocks [χ2(1) = 2.59,
To convert dichotomous response data to an interval-scale level, we computed the percentage of wrong answers per participant collapsing across listening blocks and players. A Two-Way ANOVA on percentage of wrong answers, with articulation (type of articulation to discriminate) as within-subjects and listeners expertise as between-subjects revealed a significant effect of the articulation [
Focussing on articulation,
Concering the listeners expertise, Figure
This study investigated the production and the perception of articulation on the saxophone with two experiments. For the production experiment we built a sensor equipped saxophone reed to monitor tongue-reed interaction in alto-saxophone performance, while participants performed melodies at three tempi with different articulation techniques. The captured sensor-reed signals showed that for portato articulation, the tongue-reed contact duration was independent from the given tempo, whereas for staccato articulation the gap between the tones was relative to the given tempo. In legato articulation, no tongue stroke occurred, and tone transitions were initiated by a change of the fingerings. Such coordination tasks occur with all wind instruments, where different effectors (tongue and fingers) are required to produce one tone (e.g., flute, clarinet, trumpet). It is also the case for string instruments that the player has to coordinate different effectors to produce one tone. Bowing movements with the right arm have to be coordinated with left hand fingerings. Baader et al. (
In our study we looked into temporal effects of saxophone performances under different tempi, which were produced by different effectors (fingers only, tongue only, combined tongue-finger actions). We found that at the slow tempo, tone onsets produced by tongue-only actions were significantly less precise than tone onsets produced by fingerings only. Highest precision was archived for combined tongue-finger actions. This corresponds to our hypothesis that timing precision improves for combined tongue-finger actions. However, we did not expect to see that in the fast tempo condition, tonguing alone was more precise than finger-only actions and combined tongue-finger actions showed a high timing variability, at approximately the same level as finger-only actions. This finding suggests that fingers play a dominant role in the overall timing of saxophone performances.
In woodwind performance finger actions usually do not receive the same attention as with piano playing, where the finger movements directly produce the sound. Our observation that there is a strong influence of finger timing on the overall timing in woodwind performance may put a new focus on further investigations of finger movements, finger trajectories, and finger forces in this domain. With the help of sensor-equipped wind instruments and the development of new customized sensors, useful advice for music education may be gained in future research.
In the listening experiment, we observed that the articulatory sound modifications (legato, portato, staccato) were mostly perceivable for non-musicians, musicians (not playing saxophone) and professional saxophonists. Only the sound of portato tongue-reed strokes was difficult to discriminate from that of non-tongued legato tone-transitions. There are two possible reasons for this. First, a brief damping of the reed vibrations does not immediately stop the standing wave in the resonator and thus only slightly modifies the radiated sound. Not all listeners notice that the reed has been stopped. Second,
This research was supported by the Austrian Science Fund (FWF): P23248-N24.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
We would like to thank Dominica Knapp for conducting the listening experiment. Further we thank Caroline Cohrdes, Gerald Golka, and Laura Bishop for fruitful discussions. The authors made extensive use of free software under GNU/Linux.
1A, B, and X stimuli contained three different articulations: no correct answer was possible.