Among all of the tennis strokes, the stroke characteristics of the volley are seldom investigated. This is surprising considering its importance as an offensive weapon in men's singles and both men's and women's doubles play. One important characteristic of this stroke is that it is often performed under a significant time stress. Because the distance between the players is reduced, anticipation, fast reactions, and fast movements are critical. Although descriptive data of the tennis volley are useful for teaching and instruction purposes, there is little information about the process of transferring weight and propelling the body for the volley or about detailed movement descriptions of the stroke. Furthermore, the common patterns or principles used for volleying at a variety of ball speeds, locations, and heights have never been investigated.
A review of literature indicates few attempts to study the biomechanical aspect of the tennis volley. Elliott et al. (5) examined the movement patterns during volleying from the service line and from closer to the net. Relatively slow ball speeds (an average pre-impact ball velocity of 15.7 m·s−1) were used and the position of the ball was not controlled. The findings indicated that players produced a significant backswing, moving the racquet behind the hitting shoulder for both forehand and backhand strokes. The upper limb and racquet tended to move as a single unit, but movement occurred at the shoulder, elbow, and wrist joints. Van Gheluwe and Hebbelinch (7) collected data from three male subjects and reported the activity of nine muscles and the ground reaction force during the forehand volley. These investigators divided the volley into four phases-introductory backswing, forward acceleration, impact, and follow-through. Because only one force platform (60 cm × 90 cm) was used, the subjects were asked to keep both feet on the platform at the beginning of the stroke. As a result, the ground reaction forces recorded were the resultant forces acting on both feet and were very difficult to interpret. The other limitation of this study was that the experiments were not conducted on a tennis court and the subjects did not use their own racquets when the data were collected.
Because of an apparent lack of biomechanical data on the tennis volley technique, the purpose of this study was to examine selected reaction and movement characteristics of the volley stroke under different ball height, speed, and lateral location conditions. Specifically, this study focused on the temporal characteristics and the ground reaction forces/torques during the tennis volley.
Seven skilled male tennis players with no known musculoskeletal disorders were recruited as subjects (age: 24.7 ± 7.1 yr, height: 180.0 ± 7.8 cm, weight: 758.1 ± 67.4 N). All subjects were right-handed and had played competitive tennis for many years. They were in excellent physical condition and used their own tennis racquets when the data were collected. Four subjects were current varsity tennis team members of the University of Illinois. The other three subjects were former NCAA Division I collegiate tennis players and were teaching professionals/coaches at the time this study was conducted. All subjects were all-court players and had competed extensively in both singles and doubles and were proficient in executing the volley stroke. They signed informed consent documents before their participation.
Experimental setup. All trials were conducted in an indoor gymnasium. A tennis court was marked off using white duct tape on the hardwood flooring. The net was setup and the net poles were securely fastened to the floor. Two force platforms (Advanced Medical Technology, Inc., Newton, MA, AMTI model OR6-3, 50.8 cm × 46.4 cm surface area) were installed midway between the net and the service line (Fig. 1). The distance between the centers of the two platforms was 66.7 cm. Two high-speed video cameras (Peak Performance Technologies, 120 Hz, Englewood, CO) located about 4 m above the floor were stationed near the corners on the opposite side of the court. Small pieces of white tape were placed on the head of the tennis racquet so it would be more visible on the videorecordings.
A microphone was placed near a ball machine located behind the baseline. The sound produced by the ball machine as the ball was projected activated an event synchronization unit (Peak Performance Technologies). The output signal from the synchronization unit was sent to the cameras (a white square mark appeared on the video images) and the force platform recording system. This allowed for the synchronization of all recording systems.
Ball machine. A tennis ball machine (Apollo Wizard, Gloucester, VA), located 14.5 m away from the force platforms, was modified to meet the needs of this study. The objective was to create a unit that would allow the experimenter to control the speed, the height, and the lateral location of the ball's trajectory. The ball machine was first mounted on a base board (127 cm long × 61 cm wide × 2 cm thick plywood) that pivoted at the front end (Fig. 2). This allowed the experimenter to control the lateral location of the ball's trajectory by rotating the base board. Marks were placed on the floor near the rear right corner of the base board showing proper orientation of the ball machine for each lateral location, allowing replicable ball placement. Projection tubes of different lengths that could be easily mounted on and removed from the ball machine were used to achieve the speed control. The height control was achieved by a support plate placed under the projection tube. The support plate was free to slide along a rail that aligned with the longitudinal axis of the projection tube in an overhead view. Marks were placed on the base board to signify the proper placements of the support plate for different experimental conditions.
To disguise the ball machine and to prevent the subject from anticipating ball placement, a screen (124 cm high × 91 cm wide) with an opening (49 cm high × 53 cm wide) was positioned in front of the ball machine. A piece of black cloth attached to the screen was draped over the ball machine. This prevented light from entering the enclosure and made it impossible for the subject to see any part of the ball machine. The screen opening remained covered by a 75 cm × 100 cm panel until the moment before the ball was released from the machine. Thus, the subject did not have a clear view of the projection tube of the ball machine and could not predict the direction of the oncoming ball. However, the subject could anticipate the ball release because of the distinct noise (buildup of pressure inside the ball machine) immediately before the ball release.
Trials. The variables manipulated in this study included: 1) the speed of the ball as it left the ball machine, 2) the ball height, and 3) the lateral location of the ball when it reached the vertical plane containing the centers of the two force platforms:
* Three heights: 1.7-2.0 m (high), 1.25 m (middle), and 0.25-0.7 m (low) above the floor. For the fast speed condition, the projected ball would have landed beyond the baseline if the height was higher than 1.7 m. The ball just cleared the net in all low height trials and the difference in the low height for different speed conditions resulted from the difference in trajectory under different speed conditions.
* Five lateral locations: left (2 m from the center line of the court), left-middle (1 m), middle (0 m), right-middle (1 m), and right (2 m). Since all subjects were right-handed, the left location was on the backhand side of the subjects.
* Three speeds: 26.3 m·s−1 (58.8 mph) (fast), 23.1 m·s−1 (51.7 mph) (medium), and 16.5 m·s−1 (36.8 mph) (slow). These three speeds were the only speeds available for the ball machine used in this study.
As a result, there was a total of 45 experimental conditions (3 heights × 5 lateral locations × 3 speeds). To ensure consistent ball trajectories in different experimental conditions, a new can of tennis balls (Penn Championship, regular-duty felt, optic yellow) was used in each data collection session. In terms of the ball location when it reached the vertical plane containing the centers of the two force platforms, results of reliability tests showed that 68% of the balls (1 SD) landed within 5.2 cm horizontally and 3.8 cm vertically from the target locations. The conditions were presented in a random order. The rest interval between trials was about 1.5 min. Each data collection session lasted for about 3 h.
In all trials the subject adopted a ready position with one foot located on each force platform and was asked to volley the ball so that it would land on the middle backcourt. In each trial, collection of force platform data was triggered before the ball release and the data were sampled at 200 Hz for 5 s. At the end of each trial, the subject was asked to rate his own performance using a 5-point scale (5 = excellent, 1 = poor) based on the pace of the ball after impact and the landing location. A rating of zero was automatically assigned when the subject was not able to hit the ball (e.g., in trials of fast ball speed coupled with wide ball location).
The performance ratings for different experimental conditions were first analyzed (Table 1). Because the average scores for the left- and right-middle lateral location trials were higher than the left and right trials and the movement characteristics of volleys made in the middle trials are quite different from those that made to the sides, only the trials for the left- and right-middle lateral locations were selected for subsequent quantitative analysis. Although the left, right, and middle lateral location trials were not analyzed, these trials were important because they increased the degree of uncertainty in terms of ball condition. Seven trials were not analyzed because one subject made a short forward hop before the ball was released and his feet were partially outside of the force platforms or the event synchronization unit was not activated for unknown reasons. As a result, the total number of trials analyzed (119) was less than 126 (7 subjects × 18 conditions (3 heights × 2 lateral locations × 3 speeds)).
Video recordings. For each trial, the video field numbers of four critical instants were identified using a Peak Motion Measurement System (Peak Performance Technologies): 1) ball release: the first video field showing the synchronization mark, 2) initial racquet movement: the instant when the subject started to move the racquet sideward after the ball release, 3) end of backswing (not observed in every trial): the beginning of a distinct forward racquet motion before the ball impact, and 4) ball impact: the first video field showing the impact between the ball and racquet.
The instant of ball release was identified first and the field number was reset to zero (t = 0, Fig. 3). The last three instants were determined by qualitative inspection (no digitizing were performed) and the field numbers were converted relative to the instant of ball release. The instants of initial racquet movement and ball impact were determined using both camera views. In the event of disagreement in field number, the average of the two numbers was used. The instant of the end of backswing was determined from the recordings of the camera location on the contralateral side.
Force platform recordings. The ground reaction force (GRF) and ground reaction torque (GRT) data were normalized to the body weight of the subject. The signs of selected data were reversed so that the same terminologies can be applied to both feet (e.g., medial/lateral instead of left/right for sideward direction). As a result, positive values indicate horizontally lateral, horizontally backward, or vertically upward GRF, and medially rotated GRT (Fig. 1). For each trial analyzed, three critical instants were identified from the GRF data: 1) ipsilateral foot off, 2) ipsilateral foot on-a side step of the foot on the same side of the oncoming ball before the crossover step of the other foot (not observed in every trial), and 3) contralateral foot off-the instant of takeoff for the crossover step of the foot on the opposite side of the oncoming ball.
For the purpose of this study, seven phases of a volley were defined using the critical instants identified from the video and GRF recordings (Fig. 3). Four of these phases, including the ready, reaction, pushing, and stroke phases, were observed on every trial. The remaining three phases (ipsilateral side step, backswing, and forward swing) were only observed on some trials. The backswing and forward swing were absent on trials when subjects moved their racquet laterally to intercept the ball without a distinct backswing or forward motion before ball impact. For the purpose of discussion, this type of volley motion was referred to as "blocking" in this study. In other trials there was a distinct backswing followed by a forward racquet motion before ball impact. This movement pattern has been referred to as "punching" in the tennis literature (3).
For each experimental condition, means and SD were computed for all phase times except the ready phase time. Mean and SD were determined for the average and maximum normalized GRF in the medio-lateral (M/L), antero-posterior (A/P), and vertical directions and GRT in the reaction, stroke, and pushing phases. One-way analysis of variance (ANOVA) with repeated measures was performed to test for the effect of ball lateral location, speed, and height on selected temporal and GRF/GRT parameters. A confidence level of P ≤ 0.05 was chosen to indicate statistical significance.
RESULTS AND DISCUSSION
One of the limitations of this study was the fact that the subjects were required to execute the volleys from the same court location, which was midway between the net and service line, in all trials. In game situations, the volley location depends on a number of factors. Players often position themselves toward the side on which the approach shots landed and vary their distance from the net depending on their quickness, size, and specific match circumstances. The subjects were also required to position their feet on the two force platforms before the start of the trial. Therefore, the present study did not mimic conditions where the player is moving forward on the court such as a volley executed after a serve. Furthermore, skilled tennis players often anticipate the trajectories of the passing shots and start to move toward one side before the ball is returned. In spite of these limitations, the present study represents the first major effort in quantifying the general movement characteristics of a volley under different ball flight conditions.
The average ratings for the left- and right-middle locations were similar (Table 1), indicating that the subjects in this study could handle volleys on both sides with the same degree of success as long as the oncoming balls were "within reach." When they needed to reach out to execute a volley, they performed better on the forehand side. One possible explanation is the backhand side has a shorter reach than the forehand side because the subject must reach across his body. As expected, the performance rating is significantly related to the ball speed (correlation coefficient, r = −0.394, P < 0.001, df = 313). In terms of the height of contact, the subjects performed better in the middle height trials than in the high and low height trials. The average ratings for the left location (backhand)-fast speed trials was 0.90 ± 1.6 (N = 21). The subjects were not able to reach the ball (no-hit) in 67% of these trials. Although the fastest ball speed used in this study (26.3 m·s−1) was not as fast as the ball speeds resulting from ground strokes (4), it did provide sufficient challenge to the subjects.
The average reaction time for different conditions ranged from 205 to 226 ms (Table 2). Whereas executing a volley with uncertain ball location and speed is more complex than a simple two-choice reaction task, the recorded reaction times were slightly shorter than the two-choice reaction times reported in the literature (e.g., 226 ms reported by Leonard (6)). The difference probably results because the subjects in this study could anticipate the time of ball release from the machine. The time of stimulus presentation is unpredictable in typical reaction time studies. It is also likely that good tennis players have shorter reaction times than average individuals.
The significant difference in reaction time between the forehand and backhand (P = 0.038) indicates that the subjects reacted more quickly in the backhand trials than the forehand trials. It is possible that reaction times were faster for backhand trials because backhand volleys require a greater degree of shoulder and trunk rotation than forehand volleys (5). Subjects must react sooner to get their body in position for the volley. The significant difference in reaction time across different speed conditions (P = 0.043) indicates that the subjects initiated the racquet movement earlier with increasing ball speed. During the reaction phase subjects may process ball speed information and delay the initiation of movement when ball speed and ball location do not require a particularly fast action.
The sum of the reaction and stroke times for different speed conditions indicated that, on average, the times for the ball to reach the subjects were 587, 671, and 1,028 ms for the fast, medium, and slow speed conditions, respectively. The stroke times demonstrate that skilled tennis players could complete a volley (from initial racquet movement to ball impact) successfully in shorter than 400 ms. Because stroke times are dependent upon the flight times of the ball from release to impact, the significant differences in stroke time found across different speed (P = 0.000) and height (P = 0.017) conditions are expected results.
Comparisons between the durations of the stroke and pushing phases under the different speed conditions (Table 2) revealed that the contralateral foot was still in contact with the ground at the instant of ball impact for fast speed trials. This is one of the reasons the subjects could not reach or missed wide balls in this experiment. The opposite was true for the slow speed trials. This suggest that players seldom have enough time to initiate the crossover step before making contact with the ball unless the oncoming ball is slow. Even when the velocity of the oncoming ball was slow, the subjects timed the contralateral pushoff so that the crossover step was not completed until after the ball impact. In other words, the ball impact occurred before the landing of the contralateral foot and the body was still moving forward at that instant.
Significant differences were found in the backswing time across different speed (P = 0.000) and height (P = 0.000) conditions. Significant differences were found in the forward swing time between the forehand and backhand (P = 0.001) and across different speed (P = 0.013) conditions. When the average forward swing time is expressed as a fraction of the average backswing time, the ratios for different conditions were 40.3% (backhand), 30.8% (forehand), 31.3% (fast speed), 31.6% (medium speed), 39.1% (slow speed), 26.1% (high height), 33.5% (middle height), and 50.5% (low height). These results illustrate the effects of the three variables on the temporal characteristics of the backswing and forward swing during a punching action.
Figure 4 shows the GRF/GRT data from a trial where an ISS was observed. All ISS were initiated after the initial racquet movement and completed before the contralateral pushoff. Significant differences were found in the ISS time between the forehand and backhand (P = 0.001) trials. It was possible that the subjects took a longer ISS to gain lateral reach in the backhand volleys.
Based on the sample sizes of the backswing and ISS phase times (square brackets in Table 2), the frequencies of occurrence for the blocking, punching, and ISS were computed and presented in Table 3. Overall, a punching action occurred in 75% of the trials and these were evenly distributed between forehand and backhand trials. It occurred more often in the slow speed and middle height conditions. It is obvious that whenever possible punching is preferred to blocking, at least for the highly skilled players, because the forward swing allows the players to have better control on the ball placement and to increase the speed of the ball after impact.
The ISS occurred more often in the forehand (45%) than in the backhand (34%) trials. Because similar data from beginners and intermediate players are not available, it is not certain whether the ISS is a discriminative characteristic between skilled and unskilled players. It is noteworthy that the ISS is rarely mentioned in the instruction of a volley stroke. In terms of the footwork for volleying, most instructions provided in tennis texts recommend that the heel of the ipsilateral foot should be used as the pivot point for the axial rotation of the body (1,2). It seems that more research is needed to determine the significance and role of ISS in the initiation of lateral movement.
Ground Reaction Force/Torque Parameters
The maximum and average GRF and GRT values during the reaction, stroke, and pushing phases are presented in Tables 4-6. Because the initiation of movement and weight shifting in the lateral direction were the prime interests of this study, ANOVA statistics were performed only on the average M/L and vertical GRF, and GRT in the reaction, stroke, and pushing phases.
Reaction phase. Relatively small medially-directed lateral GRF were observed on both feet during the reaction phase (Table 4). The average resultant vertical GRF (sum of the vertical forces acting on the feet) was about 95% of the body weight (BW) during this phase. Also, both the maximal and average contralateral vertical GRF were greater than the corresponding ipsilateral forces. These observations suggest that the unloading and shifting of body weight toward the ipsilateral side were initiated before the initial racquet movement.
Significant differences were found between the forehand and backhand trials in the average ipsilateral M/L force (P = 0.002), average ipsilateral vertical force (P = 0.035), average contralateral M/L force (P = 0.001), and average contralateral torque (P = 0.000). Significant differences were also found across the speed conditions in the average contralateral M/L force (P = 0.030) and average contralateral vertical force (P = 0.017). Some of these significant differences involved small values that may not carry any significant meaning although the differences are statistically significant. The significant difference in ipsilateral M/L force between the forehand and backhand trials suggest that the shifting of BW was more pronounced in the forehand trials.
Stroke phase. Although the stroke and pushing phases are not identical, the major findings were similar (Tables 5 and 6). To avoid redundancy, results of the statistical analysis performed on stroke phase GRF/GRT parameters are reported without discussion. Significant differences were found between the forehand and backhand trials in the average contralateral torque (P = 0.017), across the speed conditions in the average ipsilateral vertical force (P = 0.000), average contralateral M/L force (P = 0.000), average contralateral vertical force (P = 0.000), and average contralateral torque (P = 0.038). Significant differences were also found across the height conditions in the average ipsilateral vertical force (P = 0.029).
Pushing phases. As expected, the maximal and average contralateral GRF were generally greater than the corresponding ipsilateral forces (Table 5). It is apparent that the forward and lateral movements during a volley are the results of the ipsilateral unloading and contralateral pushoff. The average resultant vertical GRF (102% BW) was only slightly greater than the BW during this phase. The average resultant vertical GRF during the pushing phase for different lateral location and speed conditions ranged from 99.3 to 104% BW. It appears that ball lateral location and speed have little effect on average resultant vertical GRF. However, large differences were found across ball height conditions. The average resultant vertical GRF for the high, middle, and low height conditions were 111.2, 98.4, and 94.0% BW, respectively. If one neglects the racquet weight acting on the right hand of the subject, the vertical forces acting on the subject during a volley were his own body weight and the vertical GRF. Because the body weight and vertical GRF acted in opposite directions, vertical motion of the center of gravity of the subject resulted if the resultant vertical GRF was not equal to the body weight (impulse-momentum relationship). The resultant vertical GRF indicate that the subjects elevated and lowered their centers of gravity notably in the high and low height trials, respectively.
Significant differences were found between the forehand and backhand trials in the average contralateral torque (P = 0.015). The greater GRT in the backhand trials are probably associated with the greater body rotation required in the backhand volleys. The significant differences across the speed conditions in the average ipsilateral M/L force (P = 0.012), average ipsilateral vertical force (P = 0.001), average contralateral M/L force (P = 0.001), and average contralateral vertical force (P = 0.000) suggest that the greater the ball speed, the stronger the contralateral pushoff. The greater ipsilateral and smaller contralateral vertical GRF in the slow speed trials hint that the subjects leaned toward the side of the oncoming ball instead of using a strong ipsilateral unloading/contralateral pushoff in controlling the lateral motions with slow ball speeds. This indicates that even when the oncoming ball is slow, the subjects did not try to complete the backswing or crossover step as soon as possible and wait for the ball to arrive. They timed their movements so that the stroke could be completed in a smooth continuous sequence. Significant differences were also found across the height conditions in the average ipsilateral vertical force (P = 0.003) and average ipsilateral torque (P = 0.019). The large increase in ipsilateral vertical forces and minimal increase in contralateral vertical force from low to high height trials implies that the height gain in executing high volleys mainly resulted from the strong ipsilateral pushoff during the pushing phase.
The major findings of this study are summarized as follows:
1. According to the self-reported performance ratings, the subjects could handle the volleys on both sides with the same degree of success as long as the oncoming balls were "within reach." When they needed to reach out to execute a volley, they performed better on the forehand side.
2. The average reaction time (from ball release to the instant of initial racquet movement) for different experimental conditions ranged from 205 to 226 ms. These values are slightly shorter than the two-choice reaction times reported in the literature probably because of the difference in subject sample.
3. The stroke time (from initial racquet movement to ball impact) values demonstrate that skilled tennis players could complete a volley successfully in less than 0.4 s.
4. Under the experimental conditions tested, subjects seldom have enough time to initiate the crossover step (takeoff of the contralateral foot) before making contact with the ball unless the oncoming ball is slow. Even when the ball speed is slow, the subjects timed the contralateral pushoff so that the crossover step was not completed until after the ball impact. In other words, the ball impact occurred before the landing of the contralateral foot and the body was still moving forward at that instant.
5. For the 18 experimental conditions (2 lateral locations × 3 speeds × 3 heights) analyzed, a distinct forward racquet motion immediately before ball impact occurred on 75% of the trials and was evenly distributed between forehands and backhands. Forward racquet motions also occurred more often in the slow speed and middle height trials.
6. The ipsilateral side step (ISS-a side step of the foot on the same side as the oncoming ball before the crossover step of the other foot) occurred more often in forehand (45%) than in backhand (34%) trials.
7. The ground reaction force (GRF) data during the reaction phase suggest that the unloading and shifting of body weight toward the ipsilateral (same side as the oncoming ball) side were initiated before the initial racquet movement and the shifting was more pronounced in the forehand trials.
8. The GRF data during the pushing phase suggest that 1) the forward and lateral movements during a volley are the results of the ipsilateral unloading and contralateral pushoff, 2) the subjects elevated and lowered their center of gravity notably in the high and low volleys, respectively, 3) the subjects completed the lateral movement by leaning sideward in the slow speed trials and by a vigorous pushoff of the contralateral foot in the fast speed trials, and 4) the height gain in executing high volleys resulted mainly from the strong ipsilateral pushoff during the pushing phase.
RECOMMENDATIONS FOR FUTURE STUDIES
The present study provides some baseline data on the temporal and ground reaction force/torque characteristics of the tennis volley under different ball conditions. To gain further insights into the optimal techniques of this stroke, we recommend the following topics for study:
1. Kinematic analysis of the tennis volley under different ball conditions including the body segment and racquet kinematics.
2. Identification of the differences in volley stroke mechanics among advanced, intermediate, and novice tennis players.
3. Investigation of the significance and role of the ipsilateral side step in initiating lateral movement in the tennis volley.