The Inverted Swallow in Artistic Gymnastics and Its Related Preconditioning Strengthening Exercises: Electromyographic Analysis, Muscle Synergies and Training Implications

Abstract

The Inverted Swallow (IS) is a rare isometric upper-body skill in artistic gymnastics. Athletes use Preconditioning Strengthening Exercises (PSEs) to develop the strength necessary to hold the IS at competition rings. However, the specific muscle excitation and synergies typical of the IS and its PSEs remain unexplored. Thus, this study analyzed muscle excitations and synergies during the IS and its five common PSEs. Muscle excitation was recorded from the pectoralis major, latissimus dorsi, triceps brachii, infraspinatus, trapezius, serratus anterior, biceps brachii and posterior deltoid muscles in six elite calisthenics athletes (age: 24.5 ± 3.0 years; height: 168.6 ± 5.8 cm; body mass: 65.0 ± 4.7 kg). Non-Negative Matrix Factorization was used to extract synergies. The results showed a predominant role of the posterior deltoid and triceps brachii during the IS. Significant differences were observed in the muscle excitation of the triceps brachii between the IS and its PSEs (p ≤ 0.001; W = 0.765), and in the posterior deltoid (p = 0.002; W = 0.629). Muscle synergy patterns also differed in simultaneous muscle activation between the IS and its PSEs. These findings provide the first detailed evidence of the characteristics of the IS element, providing insights for optimizing strength training and gymnastics performance.

1. Introduction

The rings are one of the six apparatuses featured in the men’s Olympic artistic gymnastics program. A significant portion of the skills performed on the rings consists of slow-moving or static hold elements, such as the swallow, the iron cross and the support scale, which require a long training process to be achieved [1]. These elements deeply influence the athletes’ final competitive score on the rings [2]. In order to strengthen the muscle groups involved in these elements, athletes typically incorporate into their training routines a series of exercises defined as Preconditioning Strengthening Exercises (PSEs), performed through the use of pulley systems, dumbbells, barbells or loop bands [3]. Athletes typically perform these exercises with a load corresponding to at least 75% of their body weight, as this is considered the minimum threshold required to sustain holding elements on the rings [3]. Bernasconi and colleagues [4] reported that the use of a counterweight system (with a load of 18 ± 9% of the athlete’s body weight) could replicate the technique of the swallow, but with a lower muscle excitation of the pectoralis major compared to the main element. Additionally, special supports that reduce the arm lever have been used to reproduce the iron cross with a lower muscle strain, a reduced shoulder asymmetry and a reduction in unnecessary and incorrect movements and positions [5,6]. Similarly, Rosaci et al. (2025), studying the muscle excitation and synergies of the support scale and its PSEs, reported that the use of pulley systems helped athletes to replicate the correct technique, emphasizing the position of the scapulae, while lifting two dumbbells in a standing position may be useful to strengthen the action of the shoulder flexor muscles that are typically involved on the competition rings [7]. Thus, analyzing muscle excitation and synergies during holding elements in gymnastics could help coaches and researchers to identify the specific characteristic of the technical elements and to determine which exercises should be included in training sessions based on the primary aim of the workout.

The Inverted Swallow (IS) represents a more advanced upper-body skill in artistic gymnastics and calisthenics compared to the aforementioned hold elements. This exercise, indeed, is rarely performed alone or in combination with swing elements during rings competitions [8]. The IS consists of a horizontal supine body position with the arms fully extended at rings height (Figure 1). The limited execution of this element makes it difficult to fully understand its specific muscular demands and to identify the most effective exercises to replicate muscle excitations and synergies. Athletes typically use different PSEs in their attempts to learn or improve the IS; however, to the best of our knowledge, the excitation of the main muscle groups during the IS and the effectiveness of the PSEs used in training routines have not been investigated. Therefore, this study aimed to analyze and compare the muscle excitations and synergies during the Inverted Swallow and during five different PSEs commonly used in training by athletes. The authors hypothesized different muscle excitation between IS and its PSEs and that employing a counterweight pulley system may enable athletes to more accurately reproduce the technical position and muscle synergies of the main competitive exercise, as compared to the other PSEs.

2. Materials and Methods

2.1. Participants

Six elite calisthenics athletes (age: 24.5 ± 3.08 y; height: 168.6 ± 5.8 cm; body weight 65.0 ± 4.7 kg; years of experience: 6.6 ± 1.8 y; training time per week: 11.8 ± 2.5 h) voluntarily participated in this study. The inclusion criteria were the ability to perform the IS element on rings with horizontal shoulders and hip and foot alignment. In addition, athletes reporting injuries that occurred in the six months prior to the study were not included. Each participant signed an informed consent form before the beginning of the investigation, and the study was approved by the local Committee of Bioethics (n.0139105; 20 June 2022).

2.2. Procedures

Standardized general and specific warm-ups were performed before the assessments. The general portion of the warm-up consisted of 5 min on a cycle ergometer against a light resistance, 10 bodyweight squats, 10 bodyweight walking lunges, 10 dynamic walking hamstring stretches, 10 dynamic walking quadriceps stretches and 5 pushups [9]. Following the general warm-up, each athlete performed 5 min of self-administered specific exercises. At the end of the specific warm-up, the IS and the five different PSEs were executed in a random order with 5 min recovery time between the exercises. The random order of the exercises was enabled by the Research Randomizer web service (Urbaniak, G. C. & Plous, S.; Version 4.0). The execution technique of each exercise was assessed by qualified gymnastic coaches (>10 years of experience). The PSEs performed were as follows: IS at cable pulley (IS_CP), IS with counterweight (IS_CW), IS with supported legs (IS_SL), IS at cable system (IS_CS) and IS with dumbbells (IS_D). These exercises are illustrated and described in Table 1. Each exercise was held for a minimum of two seconds. The length of the ring cables was 1.5 m, while the diameter of the rings was 32 mm. The same pulley system (G-Force Training System, Turin, Italy) was used to perform the IS_CP and the IS_CW. With the exception of the IS_CP exercise, which involved a cable pulley system designed to reduce the athletes’ body weight by approximately 50%, the load used for the other exercises was set to correspond to 75% of each athlete’s body weight. In particular, for the IS_SL exercise, the athletes placed their legs on a force platform to quantify the actual load they were supporting. If the initial estimate of the supported weight was inaccurate, the athlete repeated the trial, adjusting their leg placement to achieve the target value. Furthermore, if an athlete was unable to maintain proper posture or alignment during the first attempt of any exercise, a second trial was permitted with a reduced load to ensure correct execution. The load pulled with the correct technique was 43.2 ± 6.4 kg (65.6 ± 10.5% of their body mass) for IS_CS and 33.8 ± 3.2 kg (52.2 ± 5.6% of their body mass) for the IS_D exercise.

2.3. Instruments and Data Analysis

A wireless bipolar surface electromyograph (Cometa systems inc., Milano, Italy; resolution: 16-bit; gain: 1000; impedance: 10⁵ Ω) was used to record the muscles’ signals at a 2000 Hz sampling frequency. The skin areas under the EMG electrodes were shaved, cleaned and prepared with conductive cream. After a careful preparation of the skin, two Ag/AgC1 electrodes were placed on 8 trunk/shoulder muscles on the left side of the athletes’ bodies. Data from the biceps brachii (BIC), middle trapezium (TRA), posterior deltoid (DEL) and triceps brachii long head (TRI) were collected following European SENIAM procedure [10]. EMG activities of the pectoralis major sternocostal part (PEC), serratus anterior (SER), infraspinatus (INF) and latissimus dorsi (DOR) were collected based on previous literature [11,12,13,14]. Physical verification tests were executed to ensure accurate electrode positions. The raw signals were filtered with a 10–500 Hz band-pass filter. The root mean square (RMS) values of the EMG signals were calculated for each muscle in each exercise with a window of 25 ms and a step of one sample. The ratio between the RMSs of the PSEs and the IS, expressed in percentage, was used to calculate the muscular excitation of the PSEs relative to the main element [7].

Table 1. The Precondition Strengthening Exercises commonly used by athletes to improve Inverted Swallow performance on the gymnastics rings.

Image	Descriptions
	Inverted Swallow with cable pulley (ISCP) The athlete performs the IS using a counterweight system with two pulleys. One end of each cable is connected to an external weight, while the other ends of the cables are connected to the athlete’s body by a specific belt.
	Inverted Swallow with a counterweight system (ISCW) The athlete performs the IS using a counterweight system with two pulleys. One end of each cable is connected to an external weight, while the other ends of the cables are connected to the athlete’s body by a specific belt.
	Inverted Swallow with supported legs (ISSL) The ISSL consists of performing the Inverted Swallow on rings with the legs supported by a step at the same height as the shoulder line and with the hands to the side of the pelvic bones.
	Inverted Swallow at cable station (ISCS) In a supine position, the athlete holds two weights connected to the rings through a cable system with their hands at their side, in line with their pelvic bones.
	Dumbbells in standing position (ISD) In the DSD exercise, the athlete lays in a prone position with their arms straight at the side of their pelvis and abducted at 45°. The athlete holds two dumbbells with their hands to the side of their pelvic bones.

Table 1. The Precondition Strengthening Exercises commonly used by athletes to improve Inverted Swallow performance on the gymnastics rings.

Image	Descriptions
	Inverted Swallow with cable pulley (ISCP) The athlete performs the IS using a counterweight system with two pulleys. One end of each cable is connected to an external weight, while the other ends of the cables are connected to the athlete’s body by a specific belt.
	Inverted Swallow with a counterweight system (ISCW) The athlete performs the IS using a counterweight system with two pulleys. One end of each cable is connected to an external weight, while the other ends of the cables are connected to the athlete’s body by a specific belt.
	Inverted Swallow with supported legs (ISSL) The ISSL consists of performing the Inverted Swallow on rings with the legs supported by a step at the same height as the shoulder line and with the hands to the side of the pelvic bones.
	Inverted Swallow at cable station (ISCS) In a supine position, the athlete holds two weights connected to the rings through a cable system with their hands at their side, in line with their pelvic bones.
	Dumbbells in standing position (ISD) In the DSD exercise, the athlete lays in a prone position with their arms straight at the side of their pelvis and abducted at 45°. The athlete holds two dumbbells with their hands to the side of their pelvic bones.

The signal-to-noise ratio was controlled using the Agostini procedure [15]. When the signal-to-noise ratio value guaranteed the good behavior of the synergy extraction algorithm [16], the muscle synergies were identified for each muscle and exercise following the procedure suggested by Ghislieri and colleagues [17,18]. After applying the previously described band-pass filters, the EMG signals underwent a series of processing steps, including high-pass filtering (8th-order zero-lag IIR Butterworth digital filter, cut-off frequency of 35 Hz), rectification and low-pass filtering (4th-order zero-lag IIR Butterworth digital filter, cut-off frequency of 12 Hz). The EMG envelope was then amplitude-normalized to the maximum value recorded across all muscle signals for each participant. To extract a set of N time-invariant muscle synergies that minimized the reconstruction error, the Non-Negative Matrix Factorization algorithm was applied [18]. The “nnmf” MATLAB^® function was executed, varying the number of muscle synergies from 1 to 8 and using parameters optimized in previous studies: the multiplicative update algorithm, a function tolerance of 1 × 10⁻⁶, 50 replicates and a maximum of 1000 iterations [18]. For each number of muscle synergies, the R² similarity between the original and the reconstructed EMG signals was computed. The optimal number of synergies required for an accurate reconstruction of the original signals was determined using the “elbow” criterion, which identifies the point at which the R² curve shows the most pronounced change in curvature [7]. Only the time-independent muscle synergy weight (W) was considered, using the mean value across all participants. When a single synergy was analyzed, the contribution of a muscle was considered relevant if its activation exceeded a defined threshold. This threshold was calculated as half the range of the activation values observed across the participants for a specific muscle and exercise. All the EMG analyses were executed using MATLAB software (version 23.2.0).

2.4. Statistical Analysis

The Shapiro–Wilk test was used to assess the normal distribution of the data. Differences in muscle excitation were calculated using the Friedman test non-parametric ANOVA for each muscle [19]. If the ANOVA result was significant, Conover’s post hoc test using Bonferroni’s correction was used to compare the muscle excitation of the IS with the PSEs. The T-Stat was used to assess differences in ranked scores, with higher values indicating greater differences while lower values suggested lower differences [20]. The significant alpha level was set at 0.05 for all analyses. All statistical analyses were performed using SPSS software (version 28.0, IBM, Armonk, NY, USA).

3. Results

Four synergy patterns (W1, W2, W3, W4) were extracted during the IS. The analysis of the muscle synergies during the IS element showed the simultaneous activation of the SER, INF, TRA, TRI and DEL in W1, of the BIC, DEN, TRI and DEL in W2, of the PET and BIC in W3 and of the SER and INF in W4.

3.1. Comparison Between Preconditioning Strengthening Exercises (PSEs)

The Friedman test non-parametric ANOVA revealed several differences in muscle excitation between the IS and its PSEs for the following muscles: the PEC (p = 0.024; X²_F = 12.952; Kendall’s W = 0.432), BIC (p = 0.029; X²_F = 12.476; Kendall’s W = 0.416), SER (p = 0.003; X²_F = 18.095; Kendall’s W = 0.603), DOR (p = 0.006; X²_F = 16.190; Kendall’s W = 0.540), INF (p = 0.003; X²_F = 17.810; Kendall’s W = 0.594), TRI (p < 0.001; X²_F = 22.952; Kendall’s W = 0.765) and DEL (p = 0.002; X²_F = 18.857; Kendall’s W = 0.629). No significant differences in the TRA (p = 0.204; X²_F = 7.238; Kendall’s W = 0.241) muscle was found between exercises. The results of the detailed Conover post hoc tests are reported in the specific subsection of each exercise. Figure 2 and Figure 3 report muscle excitation (including statistical differences) and muscle synergy (W1, W2, W3, W4) comparison between the IS and its five PSEs, respectively.

3.2. Inverted Swallow with Cable Pulley (IS_CP)

The EMG muscle analysis of the IS_CP exercise showed a significantly lower excitation of the PEC (−48%; p = 0.009; T = 3.924), SER (−52%; p < 0.001; T = 4.919), DOR (−41%; p < 0.001; T = 4.983), INF (−36%; p = 0.001; T = 4.640), TRI (−42%; p < 0.001; T = 7.847) and DEL (−56%; p < 0.001; T = 4.854) compared to in the IS. The synergy analysis showed the simultaneous muscle activation of the PEC, SER and TRA in the W1 synergy, the BIC, SER, INF and TRI in the W2 synergy, the PET, BIC, SER, DOR TRA and TRI in the W3 synergy and PEC only in the W4 synergy.

3.3. Inverted Swallow with Counterweight Pulley System (IS_CW)

The muscle excitation registered in the IS_CW exercise did not significantly differ to the IS for all muscles (>67%; p > 0.427; T < 2.08). The synergy analysis showed the simultaneous muscle activation of the PEC, TRA in the W1 synergy, the BIC, SER, TRI and DEL in the W2 synergy, the PET, BIC and SER in the W3 synergy and the SER, DOR and TRI in the W4 synergy.

3.4. Inverted Swallow with Supported Legs (IS_SL)

The EMG muscle analysis of the IS_SL exercise showed a significantly lower excitation of the SER (−48%; p = 0.001; T = 4.696) and TRI (−29%; p < 0.001; T = 5.231) compared to the IS. No other differences between the muscle activation in the IS_SL and IS were registered (p > 0.05). The synergy analysis showed the simultaneous muscle activation of the BIC, SER and TRA in the W1 synergy. The INF, TRI and DEL were active in the W2 synergy, while the PEC, BIC, SER, INF and TRA were active in the W3 synergy. The W4 synergy showed the activation of the BIC, SER and DOR muscles.

3.5. Inverted Swallow at Cable Station (IS_CS)

The muscle excitations of the INF (−23%; p = 0.024; T = 3.536), TRI (−34%; p < 0.001; T = 6.103) and DEL (−37%; p = 0.005; T = 4.160) registered in the IS_CS exercise were significantly lower than in the IS element. The muscle synergies revealed the simultaneous activation of the TRA and TRI in W1 and the activation of the BIC, DOR and DEL in W2. The W3 synergy showed the activation of the PEC, BIC, DOR, TRI and DEL, while the W4 synergy indicated the activation of the BIC and DOR only.

3.6. Inverted Swallow with Dumbbells (IS_D)

The IS_D exercise showed a lower excitation of the PEC (−38%; p = 0.037; T = 3.363) and SER (−43%; p = 0.012; T = 3.801) compared to the IS, only. The W1 synergy showed the activation of the BIC, SER, INF, TRA and DEL. The W2 synergy consisted of the activation of the PEC, SER and DEL. The W3 synergy showed the simultaneous activation of the PEC, BIC, DOR, INF and TRI, while the BIC, DOR TRI and DEL were included in the W4 synergy.

4. Discussion

This study aimed to evaluate the muscular excitation and synergies of the IS and its PSEs in highly skilled male athletes. The results confirmed the hypothesis and showed that different muscle excitation patterns could be detected in the IS compared to its PSEs.

The main finding of the study is that the TRI and DEL were found to have very important roles in the execution of the IS. The large excitation of these two muscles (TRI: 2286.6 ± 460.3 µV; DEL: 1351.3 ± 612.6 µV) and their synergistic activity in the W1 and W2 synergies during the execution of the IS may be associated with the extension of the humerus [21,22]. Therefore, the main muscles involved in the maintenance of the IS differ from those used in other whole-body horizontal hold elements in the still rings specialty of artistic gymnastics. Additionally, the triceps represent bi- articular muscles that extend the elbow joint and the shoulder, together with the posterior deltoid [23]. During the W1 synergy, these two muscles (TRI and DEL) are co-active with the TRA (333.1 ± 114.2 µV), SER (337.8 ± 83.8 µV) and INF (548.9 ± 115.5 µV). During the execution of the IS, the TRA muscle promotes scapular adduction (essential for correct scapular alignment during the isometric hold), the SER controls scapular stabilization on the thorax and the INF maintains the external rotation of the humerus. Unlike other hold gymnastics elements, the role of the INF muscle in this element is likely limited to the maintenance of humeral external rotation [4,24]. In addition, despite the relatively high excitation of the DOR muscle registered (838.2 ± 338.7 µV), its action is probably less relevant than the roles of the TRI and DEL, as the DOR did not show any synergistic activation with other muscles. Finally, the low excitation of the BIC (97.3 ± 24.6 µV) and PEC (77.2 ± 50.49 µV), active in the W3 synergy only, suggests that the principal action of these muscles is to stabilize the humerus and control the movement of the rings [25].The present study showed that muscle excitations and synergies were different in the IS compared to its PSEs. Specifically, the reduced muscular excitation recorded in many muscles during IS_CP compared to the IS may be attributed to the decreased load and increased stability provided by the cable pulley system. This finding is consistent with previous investigations conducted on the support scale [7], and confirms that muscle excitation may be lower when the load is reduced by the cable pulley system and when a stable condition is compared to an unstable one [26]. Furthermore, the muscular synergies observed in the IS_CP suggest a low degree of similarity between this element and the IS performed on regular rings. None of the extracted muscle synergies included the activation of the DEL muscle, and the W2 synergy only appeared to include some muscles active during the W2 synergy of the IS. Conversely, holding 75% of an athlete’s body weight in IS_CW did not result in significant differences in muscle excitation compared to the IS. This is consistent with the findings of Bernasconi and colleagues [4], who reported similar muscle excitations of the PEC, BIC, SER, DOR and TRI when the swallow element was reproduced using a counterweight system. Additionally, although the W1 synergy led to the activation of the PEC and TRA only, the IS_CW was the only exercise showing similar muscle synergies compared to those of the IS element. Specifically, the W2 and W3 synergies showed comparable muscle activation, while the W4 synergy was distinguished by the additional activation of the DOR, which was not observed in W4 in the IS element. These findings suggest that a 25% reduction in body weight resulting from the use of a counterweight pulley system can maintain the muscle actions typical of the IS element. Similarly, the IS_SL performed at 75% of body weight induced higher muscle excitation in all the detected muscles, except for the SER and TRI. Notably, the INF and TRA showed the highest excitation in IS_SL compared to the IS (+3% and +45% for INF and TRA, respectively). This increased excitation, in addition to the synergistic action of the TRA and INF observed in the W2 synergy, suggest that this exercise allows athletes to achieve greater scapular adduction and enhanced humeral abduction and external rotation.

The analysis of the IS_CS showed the reduced muscular excitation of the DEL, TRI and INF compared to the IS, while the excitation of the BIC was more elevated (+15.1%) compared to the IS and represented the highest excitation level registered among all the exercises studied in the present investigation. The observed muscle synergies revealed that the IS_CS was the only exercise in which the DOR was active in three of the four synergies considered, while the INF was not involved in any synergistic actions. As a result, this exercise does not seem to fully replicate the typical muscular actions of the IS element, showing a greater involvement of shoulder extension and adduction driven by the DOR (with both DOR and BIC active in W3) and a facilitated external rotation of the humerus. These results are probably influenced by the 45-degree angle between the cable and the ground.

The use of dumbbells resulted in significantly different muscle excitations for the PEC (−38%) and SER (−43%) muscles only. The high levels of muscle excitation detected in the DEL, TRA and INF are consistent with previous studies focused on neutral grip rows with fully extended elbows [27]. In addition, and similarly to IS_CS, the DOR contributes in the W3 and W4 synergies. Notably, the IS_D exercise is the first exercise in which the INF is involved in synergy with the DOR (W3). This may be attributed to the prone position adopted by the athletes during the execution of the exercise, which, similarly to the swallow, support scale and neutral grip rows, provides the hypothesis that the synergy between the INF and DOR muscles enhances shoulder stabilization and prevents the antero-posterior displacement of the humeral head [4,27]. A limitation of the present study is found in its small number of participants. However, only a few athletes have the technical and strength characteristics required to perform the Inverted Swallow on the rings correctly. In addition, future research should investigate the effects of different training protocols based on these PSEs to evaluate and monitor gymnasts’ performance across various preparation and competition phases.

5. Conclusions

This is the first study to identify the activity of the main muscles involved in the IS element and its related PSEs, providing a foundation for optimizing strength training protocols and enhancing performance in artistic gymnastics through targeted exercise selection. Due to the lower excitation in comparison to the IS, the IS_CP may be included in warm-up routines. In contrast, the IS_CW appears to be the most appropriate option for replicating both the muscle excitation and synergistic activation patterns of the IS, making it a suitable choice during both preparatory and competitive periods. The IS_SL may be used to specifically target humeral abduction and external rotation; however, due to the reduced involvement of the TRI, it is advisable to include additional exercises aimed at strengthening this muscle. In addition, the IS_D appears to be particularly effective for enhancing back musculature, while the IS_CS exercise does not seem to replicate the IS pattern and should be used by athletes to strengthen humeral adduction only. The inclusion of specific exercises may enhance the different components that lead to the proper execution of the main competitive positions.