The Kinematics of the Return of Serve in Tennis: The Role of Anticipatory Information

Abstract

Visual anticipatory information from early periods of ball flight is thought necessary to intercept the ball in many sports. This study analyzed the temporal characteristics of returning a tennis serve by manipulating the amount of visual information available to the receiver. The movements of tennis players receiving ‘serves’ were measured on court. Participants received serves when playing against a ball machine or an actual server during full vision conditions and also during partial vision occlusion (i.e., early ball flight, second third, last third of ball projection). We measured the moment of the receiver’s movement initiation; the back swing duration; and the forward swing duration. There were no consistent differences in these movement characteristics between the ball machine and the server up to the projection speed of 125 km.hr-1. There were differences in the duration of the forward swing during the partial vision conditions. Initiation of the forward swing occurred earlier and the swing duration was increased when the first third of ball flight was occluded. Important anticipatory information about when to initiate the forward swing is present during the first third of ball flight. When receiving moderately fast serves up to 125 km.hr-1, the receiver does not appear to use information from the server’s action to modify the timing of their response.

Introduction

The demands of a fast ball sport like tennis place heavy emphasis on anticipation and perception. There are severe spatial and temporal requirements that are intricately combined. Glencross and Cibich (11) argue, that perceptual anticipation is essential in fast ball sports, because of inherent human limitations in reaction time and the movement time would cause the decision to be made too late for an effective response. In a classical study, Bahill and LaRritz (3) revealed that expert baseball players are not able to track a ball during its whole flight due to human visual limitations. They also argue that even the best athletes are not able to track a fast ball which is less than 1.5m away. To hit a fast moving ball a receiver needs to identify: where to swing in order to hit the ball (spatial accuracy); when to swing (temporal accuracy);and how to time the length of the swing. When receiving a fast serve in tennis(up to 250 km.hr-1) elite players have as little as 400 ms from the server’s racquet-ball contact to make their response. The task for the tennis player who is trying to return a serve includes: anticipation and timing, ball flight trajectory prediction in space, and on-line modification of racquet-ball alignment (20).

Schmidt (20) suggested we can functionally categorise anticipation into spatial and temporal components. Accordingly previous anticipation research in tennis has mainly adopted laboratory approaches using either spatial or temporal occlusion techniques. Research has typically recreated visual information by projecting screen videos or film frames of tennis players (21,22). In 2005, Shim et al. (23) argue that it is possible to anticipate the type of stroke (i.e., topspin lob, ground stroke), but not the direction of the outgoing ball. Other researches, Abernthy and Zawi (1) or Shim et al. (22)compared groups of novice players and expert players in fast ball sports. They show that the groups focus upon different visual cues and that experts were more accurate at anticipating ball flight particularly when they were tested with a representative environment (i.e., on court rather than watching a video or point light display).

In other in situ tennis research (19, 24) eye movements were examined to determine which cues the eyes focus on during the anticipation of the opponent’s motion. Day (6) used a helmet which occluded the receiver’s vision at the moment when the ball hit the server’sracquet. The results showed that skilled tennis players were able to make predictions based on pre-contact cues. Hence most in-situ research has been concerned with visual anticipation of ball direction. Williams (26) says that the player can rely on this deep prediction more reliably than on-line visualinformation from early parts of the ball flight. Despite the apparent importance of anticipatory cues from the server’s action, players regularly practice using ball serving machines (in which anticipatory cues are largely absent). In 2007, Renshaw et al. (18) showed differences in the initiation times of back swing in cricket, back swing and downswing duration while facing the bowler and the ball machine. In 2011, Pinder et al. (16) is proposing ways how to optimize development programs in fast ball games and in which situations to use the ball machine. It is not clear how important prior vision of the server’s action actually is for the timing of a receiver’s movements.

Clearly, ball trajectory information is also important for the returning player in order to successfully intercept it. For the period of time after the opponent hits the ball, research has mainly focused on spatial anticipation.For example in some studies, participants estimate where the ball would bounce on the court (10, 25). Williams (25) pointed out that expert players spend less time watching the ball, while in their remaining time they are able to watch the subsequent movements of their opponent. Novice players spent more time watching the ball but were still less successful predicting the future ball contact point with the ground.

Research on intercepting a temporally-occluded object is common in various ball-catching studies. In 2009, Dessing et al. (8) showed that catching movements were initiated significantly later, when the first part of the ball trajectory was occluded, compared to no- and late occlusion. In a laboratory study, Port et al. (17) participants were trying to intercept a partially occluded moving object on a computer screen under various visual movement conditions. Trials, where the object was intercepted more than 100 ms before the target actually arrived were categorized as early errors; more than 100 ms after the target were considered as late errors. It was concluded that a large majority of early errors had occurred in trials with decelerating targets, and their percentage had tended to increase with longer target motion times; and late errors occurred in relation to all three target acceleration types(accelerating, decelerating and constant speed), and their percentage tended to decrease with longer target motion types.

Researchers who have focused on the relative importance of different parts of ball flight indicate contradictory results. Carlton (5) argues that vision is focused upon the ball in the last half of its trajectory. Bahill and McDonald (4) claim that baseball batters are using only very little information from the middle part of the ball trajectory and that the first and the last part provide essential information for an accurate swing. However in their study of experienced batters, DeLucia and Cochran (7) suggested that the necessary interception information could be obtained from any part of the ball trajectory. Visual occlusion immediately prior to the ball-hand contact adversely affected hand position in inexperienced adult catchers (13), but experienced catchers were not affected. It was suggested that skilled catchers were able to determine more accurately the ball flight characteristics from the earlier parts of the ball trajectory and consequently use that information to predict more precisely the correct interception point. The contrasting results from these different studies are likely a consequence of the different task constraints examined (i.e., hitting vs. catching) and also the different methods of visual occlusion employed. To date no research has examined the relative importance of different phases of ball flight when returning a serve in tennis.

In this study we wished to examine how tennis players use anticipatory information from the action of the server and subsequent ball flight to inform the timing of their movement responses. To present a representative task environment relevant to tennis players, ball speeds of 100 and 125 km.hr-1 were recreated by a ball machine and an actual server on court. Whilst elite tennis players can serve much faster than our top speed (above 200 km.hr-1) the ball machine was not able to reliably produce such high velocities. Three periods of ball flight (‘early’, ‘middle’, and ‘late’)were occluded whilst the receiver attempted to return the ball. It was predicted that experienced tennis players would move later when receiving from the ball machine than the actual server and therefore their stroke would be shorter in duration (i.e., more hurried). Given the high spatio-temporal demands of the task we also predicted that the occlusion of early ball flight would result in an early initiation of response and a longer racquet swing.

Material and Methods

Participants: Two groups of 7 right-handed male tennis players participated in this experiment (N=14). Group [1] with a mean age of 23.3 years(SD = 2.28) faced balls served from a ball machine. Group [2] had a mean age of25.3 (SD = 4.19) and faced a real tennis player. The participants were assigned into the groups randomly. All participants were national tournament players ranked in the top 200 players of the Czech national ranking system. All participants were free from injury at the time of testing and none had corrected vision.

Apparatus: The research proceeded in an indoor court in order not to be influenced by extraneous variables (i.e., wind, distractions, quality of light, etc.). For group 1, the ball machine was placed on the base line of the tennis court, 1 m to the right of the center service line. The muzzle of the ball machine was placed at a height of 2.8 m. The ball machine was calibrated to serve the balls in one direction only with a minimum spin on the ball. A pair of liquid crystal glasses (Plato, Translucent technologies) were placed on the participant’s head. There were photo diodes installed on the end of the ball machine muzzle which triggered the occlusion when the ball passed through. The ball machine was connected through a computer to the occlusion glasses. The wires connecting the occlusion glasses to the computer did not limit the participant’s movements. The action was recorded by video camera so that the recordings could be evaluated using two dimension analysis.

For Group 2, the ball machine was replaced by a skilled tennis player, who was serving the balls. The photo diodes were placed at the height 2.3 m on the base-line of the tennis court, where a tennis player was serving. The photo diodes triggered the occlusion, when the server swung through with his racquet. Like the ball machine, the server performed flat serves with minimum spin on the ball. The server was a former pro-circuit player and he is now playing national competitions and working as a tennis coach. He was able to produce required serves with a suitable speed and placement in 85 % of cases(the rest served wide or hit the net). A consistent background of a dark green curtain was used in both conditions.

Task and procedures: Each participant was allowed to warm up.Participants were told that they would be videotaped. They hit the balls only with the forehand stroke. The balls were delivered to the same receiving location consistently so that the participants did not have to leave the starting position to reach the ball. They were instructed to hit the ball with a full swing (not to set the racquet nor block the ball without any swing). They were also told to hit the ball down the line. There was a target on the opposite side of the court, which they tried to aim at. The initial position for every player was the same. This position was marked on the tennis court 0.5m behind the base-line and 0.7 m to the left of right side line. We told all participants to start every trial from this position.

There were eight conditions based on the ball speeds [2] and visual occlusion conditions [4]. The two average ball speeds were approximately 100km.hr-1 (27,8 m.s-1) and 125 km.hr-1 (34.7 m.s-1) . In the full visual condition the average duration of total ball flight was 1260 ms at 100km.hr-1speed and 960 ms at 125 km.hr-1 speed. The speed was checked by radar.The speed range was no more than ± 10 km.hr-1. If the ball had been outside this range; hit the net; was wide or out (especially when a server was serving)the trial was repeated till the ball had satisfactory speed, direction and landed in the correct field. In every situation participants saw the initial0.2 seconds of ball flight followed by a subsequent occlusion of one-third of the remaining trajectory.

Each participant received 3 practice trials without any occlusion.Approximately 0.5-1.5 second before each serve a signal “action”was called (only in the case of the ball machine). Consequently 5 trials proceeded at the speed 100 km.hr-1and 5 trials at the speed 125 km.hr-1 in the full vision condition. Next, there followed 30 trials with 5 repetitions for every combination of these 2 factors. The participants did not know which part of the ball flight trajectory would be occluded, nor the ball speed. The sequence of ball occlusion and ball speed was set randomly and the same sequence was used for all participants. There were short 3 minute breaks after every 10 trials.

Data analysis: The research was evaluated from two-dimensional analysis of the video recordings. There were 3 different dependent variables in this experiment as the outcomes. Movement initiation time was measured at the beginning of the racquet back swing. Specifically, this was the time between the ball appearance from the ball machine (or when server struck the ball) and the initial movement of the players forearm to start the forehand back swing. For the second dependent variable we measured the duration of back swing i.e., the time elapsed between initial backward movement of the player’s forearm to start the back swing and the initial forward movement to start the forward swing. The third dependent variable was the duration of forward-swing i.e., the time between the initial forward move of player’s forearm to start the forward swing and racquet-ball contact. For the data analysis we used descriptive statistics to compare number of errors; repeated measures ANOVA2×2×4 (Bonferroni) to find the effects and the interaction among the effects and for the pair effects comparison.

The results were divided into the two categories. 1) The participant hit the ball. 2) The participant missed the ball. The video tapes were reviewed by three independent raters and an expert evaluation was performed to determine if the missed ball resulted from temporal or spatial error – also was used by Haller and Clark (13). We gave the raters the same evaluation instructions and criteria, i.e., whether the participant made the stroke swing too early or too late. The inter-rater reliability calculated between the independent scoring of the trials was 0.953. For the forward swing duration, we were not able to analyze the missed balls as there was no racquet-ball contact.

Results

For Group 1, 280 balls were sent from the ball machine (40 balls per person). 27 balls out of 280 were missed (9.6%). 11 misses occurred at 100km.hr-1 and 16 occurred at 125 km.hr-1. Most errors were classified as temporal in nature (85%). Early errors tended to be most common when the second third of the ball trajectory was occluded (see table 1). 4 missed balls were evaluated as spatial errors.

For Group 2, 278 balls were served from the tennis player (there were two trials missing). 37 serves were missed (13.3%). 11 balls were missed at the speed of 100 km.hr-1 and 26 balls at the speed of 125 km.hr-1. Most temporal late errors and early errors occurred during the first (44.8%) and second(51.7%) third of occlusion (see Table 1). 8 missed balls were evaluated as a spatial errors.

Table 1

Both groups comparison of number of trials containing temporal and spatial errors on missed balls.

Table 1

Receiver’s movement initiation

Mean scores for all the conditions are displayed in Figure 1. The receiver responded later during the occlusion phases when facing the server in comparison to ball machine, however this trend was not statistically significant. 2×2×4 analysis of variance revealed only a significant main effectof speed (F1,6 = 16.2, p < 0.05). The average initial movement times of the125 km.hr-1 speed are generally lower than for the 100 km.hr-1 speed. All other effects were non-significant.

Figure 1

Average times of the players’ initial move during different speed and type of server (ball machine or server).

Figure 1

Back swing duration

Mean scores for all the conditions are displayed in Figure 2. 2×2×4 repeated measures ANOVA showed significant main effects of speed (F1,6 = 66.79, p <0.05), and of occlusion (F1,6 = 5.27, p < 0.05). There were longer durations of back swing for the 100 km.hr-1 serves in comparison to the 125 km.hr-1 speed.There was no difference between the ball machine and the server except 100km.hr-1 speed during the 2nd and the 3rd third occlusion. Simple Main Effects ANOVA showed significant effect of type of server (between the ball machine and server) during the 100 km.hr-1’s second third occlusion (F1,6 = 6.54, p< 0.05). Values at the 125 km.hr-1 speed were similar. Significant effect of occlusion also existed between the 125 km.hr-1 server full vision condition and the 125 km.hr-1 server second third occlusion (F1,6 = 7.29, p < 0.05).

Figure 2

Average duration of players’ back swing during different speeds and types of server (ball machine or server).

Figure 2

Significant differences (p < 0.05)

1Between the server and the ball machine during the 100 km.hr-1’ssecond third occlusion.

2 between the 125 km.hr-1 server full vision condition and the 125 km.hr-1server second third occlusion.

Forward swing duration

Times during full vision conditions and during the final third of occlusion were similar at the same speed (Figure 3). 2×2×4 repeated measures ANOVA revealed significant main effects of speed (F1,6 = 23.44, p < 0.05), and interactions between speed and occlusion (F3,18 = 7.61, p < 0.05), speed and type of server (F1,6 = 13.9, p < 0.05), and between speed, occlusion and type of server (F1.3,7.78 = 17.51, p < 0.05). Simple Main Effects ANOVA showed some significant effects which are labelled on the next graph.

Figure 3

Average times of players’ move between the initial forward move of player’s forearm to start the fore swing and racquet-ball contact during all conditions.

Figure 3

Significant differences (p < 0.05)

1 Between the 100 km.hr-1 ball machine full vision condition and the 100 km.hr-1 ball machine first third of occlusion (F1,6 = 15.25, p < 0.05)

2 Between the 125 km.hr-1 ball machine full vision condition and the 125 km.hr-1 ball machine first third of occlusion (F1,6 = 8.85, p < 0.05).

3 Between the 100 km.hr-1 server full vision condition and the 100 km.hr-1 server second third of occlusion (F1,6 = 15.25, p < 0.05).

4 Between the 100 km.hr-1 ball machine first third of occlusion and the 100 km.hr-1 server first third of occlusion (F1,6 = 21.17, p < 0.05)

5 Between the 100 km.hr-1 ball machine second third of occlusion and the 100 km.hr-1 server second third of occlusion (F1,6 = 8.2, p < 0.05).

Discussion

During the full vision conditions no temporal or spatial errors occurred. The ball machine was able to send the balls accurately and without an error. The server was able to reach the required speed and direction, but there were a couple of errors during the trials, as he hit the net or served wide (15 % of the serves were fault). The participants knew which direction the service was going, so the task was a little simplified as they didn’t have to make a decision whether to use a forehand or backhand stroke.

The results showed that in Group 1 (ball machine) most temporal errors occurred during the second third of the occlusion, i.e., the period when the ball was bouncing onto and off the ground. In Group 2 (server), the temporal errors structure was different. At the speed of 100 km.hr-1 most errors were during the second third of occlusions. At the 125 km.hr-1 speed almost the same number of errors occurred during the first and second third of occlusion. There occurred only 1 temporal error in all the cases during the 3rd third of the occlusion. In both groups there were only a few spatial errors.

These results support Carlton’s study (5) which shows the visual information of the second half of the ball trajectory to be very important. Our results showed the visual importance of the second third the of ball trajectory.

The effect of occlusion showed that catching movements were initiated significantly later, when the first part of the ball trajectory was occluded,compared to no and late occlusion (8). In our study the players could always see the initial part of the ball flight trajectory, so that there was no difference during the occlusion in the initial movement times.

Various researches in various sports or tasks are bringing different results. E.g. in baseball – Bahill and McDonald (4) say, that baseball batters are using only very little information from the middle part of the ball trajectory and that the first and the last part provide essential information.DeLucia and Cochran (7) suggested that the necessary interception information for the softball batters could be obtained from any part of the ball trajectory. However during a batting cricket study, Golby (12) says the middle third section of the ball’s flight trajectory was being crucial in maintaining normal performance levels. Our results are similar to Golby (12). The middle ball trajectory segment was crucial for the stroke timing from our occlusion phases. We can interpret this as the player need to see the initial ball flight trajectory after the ball bounces, which is very important for the fine tune forward swing timing. In 2009, Pinder et al. (15) report that cricket batters initial movement began later when they were facing the bowling machine as the players needed to assimilate ball flight information. However our study has not shown any differences in the movement initiation between the ball machine or real server. Nevertheless, there was a difference in the forward swing time initiation. The forward swing initiation began later when the players where facing the ball machine as these players had longer back swing time. In a soccer study, Dicks et al. (9) say the player is tracking the ball in its initial part of trajectory, in which the player is collecting the information about the ball flight characteristic (e.g. spin, direction, speed)and so that determine its impact location onto the ground. In a ball catching tasks, Haller and Clark (13) say that losing sight immediately prior to the ball-hand contact adversely affected hand position in inexperienced adult catchers. Due to disparity in previous studies in various sports and tasks; it will be necessary to separate ball trajectory estimation studies in the fastball games; [1] without the ball impacting the ground; [2] with its impact with the ground and following bounce; [3] with a multiple bounce, where every single change of the ball trajectory may contain essential information for its estimation and the research should try to be as close as possible to the real conditions of the sport.

In our study the participants could always see the initial ball flight trajectory for 200 ms and then came the thirds of occlusion. There is 200 ms time delay in human motor control system which is typical for all human movements (14). The players are collecting the essential information for the ball trajectory anticipation in its initial part (10). The use of a bowling machine resulted in batters converging on nonspecifying variables, delaying thedevelopment and attunement to specifying variables (2). Pinder et al. (15) says using ball machines affects movement coordination of skilled cricket batters and other athletes. We can support this finding as there were differences inthe forward swing initiation when the players where facing the ball machine.

Conclusion

Tennis players need to see the initial part of the ball trajectory so they can determine its direction, speed and early spin on time. For tennis players,the information of the ball impact place with the court is not the only crucial thing, but the information about the initial part of the ball trajectory after the ball bounces is also crucial. We tried to be very close to real conditions in this study. Even professional players sometimes have the second service speed around 130 km.hr-1. This study was limited with the number of participant in the groups, however these results should show some overview and these results should be compared with other similar studies.

There appears to be no difference in the initial time to start the back swing during any of our occlusions. There was a little difference in the second third of occlusion during the phase between the start back swing and the start of forward swing. The biggest differences were between the first and second third of occlusion during the phase between the start of forward swing and racquet-ball contact. This means that the stroke timing was different. The players missed the information when exactly to start the forward swing and this appeared more during the lower speed (longer ball deliver time).

Applications in sport

Timing of tennis receiver’s movement (forward swing) is altered with ball machine. Player is acting differently while using ball machine comparing to a real server – we recommend not to use ball machines too often.Occlusion could be a practical technique to train players to watch different parts of ball trajectory or the server’s action or the ability to watch opponents’ action while incoming ball. Skilled tennis players are clearly very good at altering the coordination to deal with changes in environment– they are often facing different players on different surfaces (slow,medium, fast) and playing with various types of the balls in different weather conditions (wind, rain) etc.

Acknowledgments

The project was supported by Czech Republic’s Ministery of Education Youth and Physical Education MSM 0021620864.

References

1. Abernethy, B., & Zawi, K. (2007). Pickup of kinematics underpinsexpert perception of movement patterns. Journal of Motor Behavior, 39,353-357.

2. Araújo, D., Davids, K., & Passos, P. (2007). Ecological validity,representative design and correspondence between experimental task constraintsand behavioural settings. Ecological Psychology, 19, 69-78.

3. Bahill, A.T., & LaRitz, T. (1984). Why can’t batters keep theireyes on the ball? American Scientist, 72, 249-253.

4. Bahill, A. T., & McDonald, J. D. (1982). Zero-latency tracking of baseballs and other targets. Investigative Ophthalmology and VisualScience, 22, supplement of abstracts), 265.

5. Carlton, L.G. (1981). Processing visual feedback information for movement control. Journal of Experimental Psychology: Human Perception andPerformance, 7, 1019-1030.

6. Day, L. J. (1980). Anticipation in junior tennis players. InProceedings of International Symposium on Effective Teaching of Racquet sports 1980. Ed. Groppel J., Sears, R. Champaign: University ofIllinois.

7. DeLucia, P.R., & Cochran, E.L. (1985). Perceptual information forbatting can be extracted throughout a ball’s trajectory. Perceptualand Motor Skills, 61, 143-150.

8. Dessing, J.C., Wijdenes, L.O., Peper, E., & Beek, P.J. (2009). Adaptations of lateral hand movements to early and late visual occlusion.Experimental Brain Research, 192, 669-682.

9. Dicks, M., Davids, K., & Button, C. (2009). Representative task designs for the study of perception and action in sport. InternationalJournal of Sport Psychology, 40(4), 506-524.

10. Féry, Y., & Crognier L. (2001). On the tactical significance of game situations in anticipating ball trajectories in tennis. Research Quaterlyfor Exercise and Sport, 72, 143-149.

11. Glencross, D., & Cibich, B. (1977). A decision analysis of games skills. Australian Journal of Sports Medicine, 9, 72-5.

12. Golby, J. (1989). The relative importance of perceptual information throughout the sections of a cricket ball’s trajectory. Current Psychology: Research & Reviews, 8(3), 188-199.

13. Haller, C.F., & Clark, J., E. (1990). Effects of occluding a ball’s trajectory on the interception performance of adults and children.Current Directions in Motor Development, 3, 80-90.

14. McDonald J. D., & Bahill A. T. (1983). Zero-latency tracking of predictable targets by time-delay systems, International Journal of Control, 38, 881-893.

15. Pinder, R. A., Renshaw, I., & Davids, K. (2009). Information–movement coupling in developing cricketers under changing ecological practice constraints. Human Movement Science, 28(4),468-479.

16. Pinder, R. A., Renshaw, I., Davids, K., & Kerherve, H. (2011). Principles for the use of ball projection machines in elite and developmental sport programs. Sports Medicine, 41(10), 793-800.

17. Port, N. L., Dassonville, D. N. P., & Georgepoulos A. P. (1997). Manual interception of moving targets. Experimental Brain Research,116(3), 406-420.

18. Renshaw, I, Oldham, A. R., Davids, K. W., & Golds, T. (2007). Changing ecological constraints of practice alters coordination of dynamic interceptive actions. European Journal of Sport Science, 7(3),157-167.

19. Rippol, H. (1989) Uncertainty and visual strategies in table tennis.Perceptual and Motor Skills, 68, 507-512.

20. Schmidt, R.A. (1991). Motor learning and performance; From principles to practice. Human Kinetics, Champaign.

21. Shim, J., Carlton, L.G., Chow, J.W., & Chae, W. (2005). The use ofanticipatory visual cues by highly skilled tennis players. Journal of Motor Behavior, 37, 164-175.

22. Shim, J., Carlton, L.G., & Kwon, Y. (2006). Perception of kinematic characteristics of tennis strokes for anticipating stroke type and direction.Research Quaterly for Exercise and Sport, 77, 326-339.

23. Shim, J., Miller, G., & Lutz, R. (2005). Visual cues and informationused to anticipate tennis ball shot and placement. Journal of SportBehavior, 28, 186-200.

24. Singer, R. N., Williams, A. M., Frehlich, S. G., Janelle, C. M.; Radlo,S. J., Barba, D. A., & Bouchard, L. J. (1998). New frontiers in visualsearch: An exploratory study in live tennis situations. Research Quarterly for Exercise and Sport, 69, 290-296.

25. Williams, A. M., Davids, K., Burowitz, L., & Williams, J. G. (1994). Visual search strategies in experienced and inexperienced soccer players.Research Quaterly for Exercise and Sport, 65, 127-135.

26. Williams, A. M., Davids, K., & Williams, J. G. (1999). Visual perception & action in sport. E & FN Spon, London.

 

Print Friendly