Você está na página 1de 18

Sports Biomechanics

January 2008; 7(1): 5471

The effects of differential and variable training


on the quality parameters of a handball throw
LLER
HERBERT WAGNER & ERICH MU
Department of Sport Science and Kinesiology, University of Salzburg, Salzburg, Austria

Abstract
Our aims were to undertake a comprehensive temporal, effective, and practical training study (variable
and differential learning) that would offer athletes the opportunity to increase their performance, and to
analyse the effects by measuring kinematics and quality parameters. Two participants of differing
standards a player of the first Austrian League and an Olympic and World Champion but of similar
anthropometric characteristics were recruited. One of the participants (Austrian League) was tested on
five different occasions (pre-test and four retests) to measure the effects of four different training phases
using kinematic analysis. The results of the study indicate an increase in ball velocity within the
differential training phases (first, second, and fourth phases), different proximal-to-distal sequences of
the participants, and a change of movement pattern during training measured by the segment velocities
and the angle time courses.

Keywords: Differential and variable training, handball throw, motor learning, movement pattern,
proximal-to-distal sequence

Introduction
In elite team handball, shooting on goal is one of the most important aspects of the game. For
a shot to be successful, it requires maximum ball velocity and precision as well as an element
of surprise for the defensive players and goalkeeper. But what factors influence maximal ball
velocity and precision in a handball throw, and what kind of training should be undertaken to
increase ball velocity and precision to optimize the throw?
Van den Tillaar and Ettema (2004) reported that 67% of ball velocity at ball release can be
explained by the summation effects from the velocity of elbow extension and internal
rotation at the shoulder. Joris and colleagues (Joris, Edwards van Muyen, van Ingen
Schenau, and Kemper, 1985) showed that a high ball velocity depends on an optimal
proximal-to-distal sequence, but Fradet et al. (2004) revised this thesis based on their results
with French handball players: maximal linear speed of the shoulder occurred after maximal
linear speed of the elbow. Wagner and colleagues (Wagner, Klous, and Muller, 2006)
measured the kinematics of the upward jumping throw performed by handball players of
varying skill. They found that the main reason why top players produced higher ball
velocities than less proficient players was the velocity of the shoulder, especially shoulder
flexion, together with elbow extension and ulnar deviation at the wrist. When summarizing
Correspondence: E. Muller, Department of Sport Science and Kinesiology, University of Salzburg, Rifer Schlossallee 49, A-5400
Hallein-Rif, Austria. E-mail: erich.mueller@sbg.ac.at
ISSN 1476-3141 print/ISSN 1752-6116 online q 2008 Taylor & Francis
DOI: 10.1080/14763140701689822

Differential and variable training in the handball throw

55

the results of these studies, it is clear that it is important to optimize the movement of the
throwing arm, in particular the velocity of the shoulder, elbow, and wrist.
It is usual for handball-specific training to be used to optimize the throw in handball.
Trainers provide instructions and corrective feedback. They plan handball-specific strength
training by throwing with a lighter/heavier, smaller/larger ball or with an additional weight or
training under game-specific conditions (e.g. against one or more defensive players or with
one or more offensive players).
To determine the training that is most appropriate to optimize ball velocity and precision
for the handball throw, it is important to know the standard at which the athlete performs.
For low-performance athletes, it is important to keep the conditions as constant as possible
to stabilize the movement pattern and to avoid neural overload (cf. Roth, 1989; Schollhorn,
2000). In contrast, for high-performance athletes, it is important to vary the movement
pattern to ensure adequate reaction to changing conditions and therefore to stabilize the
movement. At the elite standard, it is often necessary to develop training methods that offer
athletes the possibility to improve their performance further. Therefore, we chose a method
that offers an athlete individual optimization of certain movement patterns in contrast to the
theory of imitating the movement of a model to improve performance. We selected a variable
training method based on the variability of practice hypothesis (cf. Roth, 1989; Schmidt and
Wrisberg, 2001) and a differential training method modelled on differential learning
(cf. Kelso, 1997b; Schollhorn, 2000; Zanone and Kelso, 1997). Comparing these training
methods for the handball throw would be useful because little research has been conducted
on the acquisition of team handball skills, apart from a few studies on variable practice (Roth,
1989; Schmidt and Lee, 1999; Wagner, 2005) and differential training (Schollhorn, 2001,
2003).
Variable training
The Variability of Practice theory predicts that practicing a variety of movement outcomes
with the same program (i.e., by using a variety of parameters) will provide a widely based set
of experiences upon which a rule or schema can be built (Schmidt and Lee, 1999, p. 373).
For throw training in handball, the desired schema defined by invariant elements should
involve experience of as many different combinations of parameters as possible that require
changes in variant features within a class of skills to optimize the movement. The athlete
must learn how to alter his or her schema to achieve a particular outcome in different
conditions. That is, following Schmidt (1975, 1976, 1988), for various starting situations
(X) and result conceptions (Z), the appropriate parameters (Y) must be measured. For
throw training in handball, Roth (1989) recommends varying the following programme
parameters: action speed and overall duration, fast or slow throw execution, jump assistance
or handicaps, overall force, and changing the throw strength or spatial parameters such as
point of release or release angle. Therefore, we used lighter or heavier, smaller or larger sport
devices to vary the parameter of absolute force, special training devices to vary the parameter
of movement duration, and the participants were required to throw with different foot
positions, release angles, and different points of release during variable training.
In addition to variation of the programme parameters, the arrangement and sequence of
the exercises relative to contextual interference effects can also play a role in the success of
learning. Lee and Magill (1983) reported that serial or random practice was more effective
than blocked practice for success in a retention test. Similar results were reported by Shea
and Morgan (1979), Shea and colleagues (Shea, Kohl, and Indermill, 1990), and Wulf and
Lee (1993). In a serial or random practice condition, different exercises (e.g. A, B, C, and D)

56

H. Wagner & E. Muller

are performed one at a time either in a specific sequence or in a completely random order.
With blocked training, exercise A is repeated several times before moving on to exercise B,
then exercise C, and finally exercise D. To optimize the variable training, it was important to
establish either a serial or a random practice schedule. Therefore, we first arranged the
exercises for variable training under methodical standpoints and then randomized them.

Differential training
Learning may take the form of a phase-transition process that involves stabilization of the
required pattern as an attractive state of the coordination dynamics (Kelso, 1997a, p. 175).
This phase transition that is, the change in movement pattern from one to another stable
state was demonstrated experimentally using rhythmic finger movements (cf. Kelso, 1981,
1984) and transformed into a mathematical model (see Haken et al., 1985) by calculating
the relative phase between the two involved fingers. The most important finding in this
experiment was that by changing a control parameter in line with the movement frequency
starting from a certain critical frequency, the fluctuations increase and system changes are
self-organized. In this case, it is not arbitrary; rather, the change is from one stable state
(attractor; anti-phase 1808) to another, whereby the second attractor (in-phase 08) is
more attractive. This effect can be recognized also with complex movements when a newly
learned movement pattern reverts to an old movement pattern during competition. By
increasing movement velocity (i.e. altering the control parameter), the system becomes
unstable and change is self-organized to the more attractive attractor.
The ability to change from a bi-stable to a tri-stable regime of pattern dynamics has been
shown by Zanone and Kelso (1992a, 1992b, 1997). Participants with bi-stable dynamics,
with a stable behaviour at 08 and 1808, were assigned to practise a 908-phase. After 5 days of
training, a new attractor (the standard deviation of 908 relative phase decreased) existed,
whereby the stability of this new attractor depended on the pre-existing attractors, according
to Zanone and Kostrubiec (2004). Furthermore, the symmetry pattern (2708 phase) of the
to-be-learned pattern became an attractor state too, although such a pattern had never been
practised, which could be interpreted as a transfer within an effector system. Whether a
transfer between effector systems is also possible was tested by Kelso and Zanone (2001).
They observed that the practised 908 phase of a rhythmic arm movement also became a
stable state for the leg and vice versa, although the legs did not practise such a pattern. This
could be interpreted as a transfer of learning across two effector systems.
To apply the results of these studies to the training methodology of a complex, wholebody, acyclic, highly dynamic movement such as the handball throw, the following points
need to be considered:
1. The movement velocity during throw training should not be maximal because the
experiments of Zanone and Kelso (1992a, 1992b, 1997) were performed below the
critical frequency.
2. In the view of coordination dynamics, motor learning is associated with the building of a
new attractor. The to-be-learned pattern should reach an attractive and stable state
(Zanone and Kostrubiec, 2004).
3. Non-linear phase transitions play a critical role in the control and learning of purposeful
coordination skills (Walter, 1998).
4. Fluctuations probe the stability of coordinative states and allow the system to discover
new coordinative states (Kelso, 1997b).

Differential and variable training in the handball throw

57

It must also be considered that:


1. Temporal stability may be lost when parameters independent variables in the
language of experimental design are varied continuously over a sufficiently wide range
(Kelso, 1997b).
2. Motions that phenomenally appear to be isomorphic are never identical (Hatze, 1986).
Therefore, movement variability is part of every training session whether it is desired or not.
3. Human actions are controlled autonomously (Walter, 1998). Following Wagner and
Blickhan (1999), allowance is made for the peripheral self-organization by the human
system. This means that muscle parameters are set through the selection of exercises for
achieving a specific goal such that the higher areas of the central nervous system are not
recruited.
Coordination training in handball demands varying movement parameters over a
sufficiently wide range (cf. Hertz, Krogh, and Palmer, 1991), focusing mainly on
destabilization of an existing attractor and the building of a new movement pattern.
The differences are selected such that the expected values lie within the chosen extreme
values (principle of interpolation; cf. Figure 1). In this context, Schollhorn (2000) speaks of
the differential learning approach. According to Schollhorn (2000), practising with different
exercises also offers the ability to react continuously to new situations in a rapid and
appropriate way.
Variations in the movement pattern of the handball throw (cf. Fradet et al., 2004; Van den
Tillaar and Ettema, 2004; Wagner et al., 2006) and the principles of movement variability in
general (cf. Newell and Corcos, 1993; Schollhorn, 2000), in the context of the handball
throw, result in the variations and differences listed in Figure 2. These possible variations
serve as a basis for the conception of the individual training units. For the differential training
approach, the principle of contextual interference was also used.

Variable and differential training in practice


In previous studies of variable (cf. Catalano and Kleiner, 1984; McCracken and Stelmach,
1977; Roth, 1989) and differential (cf. Schollhorn, 2001, 2003) training, only lowperformance participants were used and the duration of the training was temporally limited.
Therefore, the aim of this study was to conduct a comprehensive temporal, effective, and
practical training study that would offer athletes the opportunity to increase their
performance, and to analyse the effects by measuring kinematics and quality parameters.

Figure 1. Interpolation of neural systems, demonstrated by the elbow angle at the reverse point. The expected values
(optimal ankle) should lie within the chosen extreme values (maximum: 1808; minimum: 458). The dashed line is
equivalent to the position of the arm at the release point.

58

H. Wagner & E. Muller

Figure 2. Possible variations and differences in handball throws (translated from Wagner, 2005, S. 141).

To evaluate possible improvements after completing this training, we analysed a world-class


athlete with similar anthropometric characteristics to a training participant for comparison.
In this context, it was of interest to establish if the movement pattern of the training
participant approximated that of the world-class athlete. That is, can the difference between
the actual and desired value of a certain model be reduced?
The training programme should involve three different phases. In the first phase,
preferable differentiated exercises should allow the system the possibility to optimize

Differential and variable training in the handball throw

59

self-organization of a movement pattern. This movement pattern could be a new one or a


stabilized old one with the positive effect of an increasing movement quality. In the second
training phase, the dynamic of the movement should be improved by changing the external
conditions as well as increasing ball velocity. Since in this phase both external and internal
forces will vary during the movement, this training allows the athlete to improve the internal
forces to optimize the movement. In this context, Bernstein (1967) speaks of the highest
stage of movement coordination that can only be realized by top-class athletes. In the final
training phase, the contents of the preceding training phases should be combined to increase
the chances of improving performance.
Methods
Participants
Two right-handed athletes of differing ability but similar anthropometric characteristics
(training participant: age 30 years, mass 91 kg, stature 1.85 m; model participant: age 34
years, mass 91 kg, stature 1.83 m) were recruited for the present study. The training
participant had been a runner-up in the First Handball League of Austria (first league),
whereas the model participant was a Olympic Champion, World Champion, and twice
World Handball player of the Year. The training participant was tested on five different
occasions (pre-test and four retests). Both participants were familiar with traditional
handball training, but had no practical experience with differential or variable training in
handball.
Training intervention and testing
As previously explained, the main aim of this study was to conceive a training programme
that would noticeably improve ball velocity and accuracy of a high-standard athlete within
one year. To measure the effects of the individual training methods, the training intervention
was divided into four phases (see Figure 3). We chose four phases instead of three because we
divided the first phase into two further phases. In the first of these phases accuracy was the
focus, whereas the second was used to increase ball velocity. The length of the individual
training phases (at least 6 weeks) was selected in such a way that a change of the
measured parameters after finishing the training phase could be ascribed to that phase.

Figure 3. Design of programme for the training participant. T phase 1 differential training for maximum
accuracy; T phase 2: differential training for maximum ball velocity; T phase 3: variable training for maximum ball
velocity; T phase 4: complex differential training for maximum accuracy and ball velocity. The model participant
undertook one test session (MP: pre-test) only.

60

H. Wagner & E. Muller

It must be considered that the effects of the completed training phase could influence the
following phase; however, because of the length of the training phases, this influence should
play a minor role.
In the first two training phases, the principle of differential learning was applied with the
aims of maximizing accuracy (first training phase) and ball release speed (second training
phase). The third phase of the training intervention was pursued according to the principle of
variable training with the goal of increasing ball velocity. The fourth and final training phase
was aimed at maximizing accuracy but also ball velocity by means of a very complex
structured training method based on the principle of differential learning. In the following
sections, the four training phases are explained in detail.
First training phase (goal: maximize accuracy). Muller and Loosch (1999) examined the
connection between remark variability and result stability using a dart throw. The results
from these investigations were transferred to the handball throw and from this the condition
factors for aiming accuracy were obtained:
.
.
.
.

the coordinates of ball release point (x -, y-, and z-coordinates)


vertical ball release angle
horizontal ball release angle
ball velocity

In addition to the possible variations in these condition factors (throwing position, throwing
direction, ball release position, and movement velocity), the following were varied in this
phase (Figure 1): stated foot position, step sequence, non-throwing position, joint position,
and joint movement. In terms of differential training (Schollhorn, 2000), it was also
necessary not to repeat a movement, which meant that one exercise (e.g. ball reception above
head level) was combined with a second exercise (e.g. with different step sequences) and thus
during each trial new initial conditions were accomplished (cf. Figure 4). To create a further
difference, all throws were randomized within a training unit (approx. 50 60). The ten
training units in the first training phase provided a total of 498 different throws.
So as not to strain the capacity of the central nervous system, attention was paid at the
beginning of all training phases to accomplish separate individual exercise forms and to
combine them only in further consequence.
Second training phase (goal: maximize ball velocity). To distinguish this phase of training from
the first training phase, all throws were executed against a neutral wall to exclude visual
perception of aiming accuracy; that is, the variations in the ranges of throwing position and
direction in this phase were not accomplished. In contrast, the range of movement rhythm
was integrated into the training, for example a slow arm cocking with a fast acceleration

Figure 4. Combination of two different exercises (overhead throw with four different step sequences).

Differential and variable training in the handball throw

61

Figure 5. Sequence of a throw forced by the throwing slingshot (the arm was forced by a rubber cord to reduce the
overall duration).

phase or vice versa. Since this training phase focused on increasing ball velocity, it was also
varied within the range of movement velocity; that is, a slow extension of the elbow joint with
an explosive inflection of the wrist in the final phase of the throw. Also, the complexity of the
task was continuously increased: up to three different exercises were combined and, as in the
first training phase, all throws within a training unit were randomized. In this second training
phase, 15 training units were completed for a total of 788 throws.
Third training phase (goal: maximize ball velocity). In this phase of the training, specially
developed training devices were introduced that supported the throw movement in the sense
of the theory, with the aim of increasing velocity speed without changing the essential
structure. Using a throwing slingshot training device, the throwing arm was additionally
accelerated in the arm acceleration phase to reduce the overall duration, or restrained to
increase overall force (cf. Figure 5). For the additional acceleration and inhibition of wrist
flexion, a wrist cuff with a similar function to the throwing slingshot was used. In this phase of
training, different balls were also used. In contrast to the differential training, a throwing
movement was accomplished several times within one training unit in the variable training.
Furthermore, a serial rather than randomized methodology was selected. In this phase of
training, two training blocks, each lasting 6 weeks, resulted in 2450 throws.
Fourth training phase (goal: maximize ball velocity and accuracy). The aim of this fourth and last
training phase was to maximize both ball velocity and accuracy. For this reason, the exercises
from the first two training phases were combined (up to five different exercises) and thus the
complexity of the task increased. In the fourth training phase, 2374 different throws were
accomplished within 40 training units.
Both before and after each respective training phase, movement technique and quality
were assessed using kinematic analysis. This allowed a connection to be made between the
training phases and their effects on the execution of movement using time-series
representation.
Apparatus
For determining the image coordinates, we used two NTSC 180-Hz (640 480) highspeed cameras (HSC-250, Motion Analysis Co.). The cameras were positioned at an angle of
908 to one another and at a distance of approximately 10 m to the right side of the training
participant. The two cameras were interfaced to a computer and automatically synchronized.
For the measurement of accuracy, a third camera was used (JVC 120 Hz digital camera) to
film the instant of balltarget contact. This third camera was positioned behind the
participant. For image processing and evaluation of the video recordings, SIMI Motion was

62

H. Wagner & E. Muller

used and for calibration a 26 pass point cube (2.5 2.5 2.5 m; PEAK Company) was
utilized. To check the accuracy of the systems used, an error analysis of point appointment,
repeated digitizing, and segment length appointment was conducted. Precision of point
appointment (eight selected control points) to determine the accuracy of the coordinates of
the cube yielded a mean deviation of 3 mm in all three spatial dimensions. The reliability of
the point determination was calculated using 5 digitization of a selected trial yielding over
all 22 segments a mean absolute error of 2.5 mm (s 3) for the space coordinates and 2.58
(s 1.5) for the segment angle. For final validation of the kinematic data, segment length
determination was performed whereby for five selected trials the calculated segment lengths
for the total course of motion were compared with the anthropometrically (upper arm:
360 mm; forearm: 288 mm; pelvis: 394 mm; thigh: 459 mm; shank: 428 mm; foot: 300 mm)
obtained segment lengths and thereby an in-medium mean absolute error of , 10 mm was
determined.
Task and procedures
Of the multiple complex throwing motions in Olympic handball, the 7-m throw was selected.
This allowed comprehensive empirical investigation under standardized conditions. The 7-m
throw involves a complex form of movement whereby the total movement consists, for the
most part, unconscious elements of motion given its short duration, predominantly without
direct correction by neural reaction, and comprising through its spatial extension a large
number of possible degrees of freedom.
For each scheduled test, the kinematic characteristics, accuracy, and initial velocity for 17
20 throws were recorded. The training participant completed all five test sessions, before and
after each training phase, while the model participant completed only one test session. As
explained earlier, the model participant served only for comparative purposes. For each
executed throw, the participants first took their starting position, which was the same as in
competition. Then, the measurement equipment was switched into record mode and the
command to throw was given, which was comparable to the whistle of a referee in competition.
The participant then had the task of throwing the ball at a target 7 m away at maximal speed
and with the highest possible precision. The goal for all throws of both participants was to hit
the centre of the target. As in competition, the fixed foot was allowed to move only after release
of the ball. A further restriction during the throws was that the release should occur at the
highest possible point. Deviations of up to 100 mm, which were measured with SIMI Motion,
were tolerated to allow for the natural variation in execution of this motion. This default was
important to ascribe possible variability to movement variability and not to another throwing
technique. In this context, the abduction of the throwing arm could be interpreted as a change
in technique of shoulder throw. Subsequent evaluation used only those throws (the first ten
throws) that met the above criteria, whose deviation from the centre of the target in the x- and
y-directions was less than 0.5 m, and for which all data were available. At the end of the test
series, the participants were able to view the completed throws on a screen but there was no
possibility of checking the quality of the result.
Data analysis
The angle, angular velocity, and segment velocity time-courses were calculated (see
Appendix) using the Peak Motus 9.0 analysis system, after digitization and data filtering with
a low-pass Butterworth filter at a cut-off frequency of 8 Hz. Dorsal flexion of the wrist, elbow
flexion, and shoulder abduction were calculated as the vector angle between three points

Differential and variable training in the handball throw

63

from one to the other bordering segment. External shoulder rotation was defined as the angle
between the forearm and the anterior direction of the shoulder in the transverse plane of the
upper arm, according to Fleisig and colleagues (Fleisig, Nicholls, Elliott, and Escamilla,
2003): Because external rotation is calculated indirectly by the forearms angular motion
about the upper arm, the accuracy of the calculation may diminish as the forearm and upper
arm segments approach parallel to each other (p. 54). To provide comparability between the
individual measurements, all throws were time-normalized over the reverse point of the
elbow segment (a 0 m/s2) and the release point identified as the last contact between finger
and ball. The beginning of the motion was determined as the point in time 100 ms before the
reverse point and the end-point of the motion was set exactly five frames after the release
point. Further normalization of the data series was performed using a spline function
(Quintic Spline) at 50 intervals (t 50).
The quality of the motion is defined using ball velocity (VBall) and accuracy (PTarget). Ball
velocity was defined as the value of the arc velocity curve at the release point and, like the
angle and angular velocity paths, was calculated using three-dimensional video analysis.
Accuracy was calculated by the normal distance of the impact point from the centre point
and determined by a two-dimensional video analysis.
Differences in the characteristics of the individual training phases of the training
participant were calculated using analysis of variance and an independent samples t-test
(training participant vs. model participant).
Results
Quality of movement
Increasing the quality of the movement (i.e. ball velocity and accuracy) was one of the main
goals of the training intervention. The results of measuring these parameters are shown as a
bar diagram (mean values and standard deviations) in Figure 6. Repeated-measures analysis
of variance with time as the main factor yielded highly significant differences between retests
1 and 2 (P 0.002**) and retests 3 and 4 (P 0.000**). There was only a tendency for the
remaining phases of the training participant (pre-test, retest 1, and retest 3) to be
differentiated (pre-test to retest 1: P 0.273; pre-test to retest 2: P 0.738; retest 1 to 2:
P 1.000). For ball velocity, there were significant differences between the training and
model participant at all measurement times (P 0.000**) except the post-test
(P 0.042*). This exception is attributable to the training participant increasing
performance during the course of training by more than 10%, or the gradient from pretest to post-test (retest 4) in comparison to the model participant diminished from 16% to
5%. In contrast to ball velocity, no significant differences were observed for accuracy, the
second characteristic of interest. A tendency for increased accuracy of the model participant
compared with the training participant was evident. Also, retest 1 and the post-test were
different to the other phases for the training participant.
Proximal-to-distal sequence
Following Joris et al. (1985), the handball movement passes through a kinematic chain
involving the whole body. The impulse should be transferred from one segment to the next
(proximal-to-distal) so that the ball finally reaches maximal speed. Figure 7 shows the speed
profile of the hip, shoulder, elbow, wrist, middle hand, finger, and ball of the training
participants fastest shot. As shown in Figure 7A, maximal speed increases from one segment

64

H. Wagner & E. Muller

Figure 6. Mean values (standard deviations) for the training participant (pre-test, retests 1, 2, 3, and post-test) and
the model participant (pre-test).

to the next in a continuous fashion. In addition, the instant in time where each segment
reaches its maximum increases from proximal to distal. Whereas this transfer can be noted
for all shots (only the maximum speed differs; see Table I) of the training participant, the
model participant does not show this transfer in any shot. The elbow reaches maximum
speed after the shoulder segment.
Maximal joint velocity
To assess optimal execution of the movement on the basis of movement quality, the maximal
joint velocities of the finger, middle hand, wrist, elbow, shoulder, and hip were calculated (see
Table I). For the training participant, a global significant difference between phases for the
factor time was observed for all joints (finger: P 0.006**; middle hand: P 0.000**; wrist:
P 0.000**; elbow: P 0.001**; shoulder: P 0.000**) except the hip (P 0.223). For
the training participant, maximal joint velocity was highest when ball velocity was highest
(post-test), which could be due to an optimal proximal-to-distal sequence. Lower maximal joint
velocities (significant differences for the finger: P 0.029*; middle hand: P 0.002**; elbow:
P 0.003**; shoulder: P 0.003**) were observed for retest 2 than retest 1, although the ball
velocity was higher (P 0.001**). Comparing the throws of the model participant and all
Table I. Maximum speeds of selected joints and ball (m/s) (mean ^s)

Pre-test
Retest 1
Retest 2
Retest 3
Post-test
Test (MP)

Ball

Finger

Middle hand

Wrist

Elbow

Shoulder

Hip

21.1 ^ 0.4
21.7 ^ 0.6
22.5 ^ 0.5
21.6 ^ 0.6
23.6 ^ 0.6
25.1 ^ 0.8

17.5 ^ 0.6
19.5 ^ 1.0
18.6 ^ 0.8
19.3 ^ 1.3
19.7 ^ 1.1
22.9 ^ 0.9

14.7 ^ 0.3
17.0 ^ 0.6
15.7 ^ 0.6
16.0 ^ 1.0
16.5 ^ 0.7
19.1 ^ 1.0

11.5 ^ 0.2
13.6 ^ 0.2
13.0 ^ 0.2
13.1 ^ 0.5
13.9 ^ 0.4
15.3 ^ 0.7

9.1 ^ 0.2
9.1 ^ 0.2
9.6 ^ 0.3
9.6 ^ 0.3
9.8 ^ 0.2
10.1 ^ 0.4

4.2 ^ 0.2
4.5 ^ 0.1
5.0 ^ 0.2
5.0 ^ 0.2
5.3 ^ 0.2
5.5 ^ 0.5

2.1 ^ 0.1
2.2 ^ 0.2
2.2 ^ 0.2
2.3 ^ 0.2
2.1 ^ 0.2
2.4 ^ 0.2

MP model participant.

Differential and variable training in the handball throw

65

throws of the training participant, significant differences were observed for ball velocity as well
as maximal joint velocity of the finger, middle hand, and wrist (P 0.000**). In contrast, for the
elbow, shoulder (between the model participant and post-test of the training participant:
P 0.084 and P 0.394 respectively), and hip (between model participant and retest 3 of
training participant: P 0.323), no significant differences were observed. Since no significant
differences between the tests of the training participant could be determined for maximal
joint velocity of the hip, calculation of the angle and angular velocity time courses were limited to
the joints located distal from the hip (shoulder, elbow, and wrist of the throwing arm).
Angle ranges
As the aim of the acquisition phase was to optimize the joint movement (flexion extension,
abduction adduction, internal external rotation) we analysed the angle ranges to show this
effect. For the training participant, a globally significant difference between tests for the
factor time was observed for all angle ranges (elbow extension: P 0.000**; shoulder
abduction: P 0.005**; shoulder external rotation: P 0.000**) except dorsi-flexion of the
wrist (P 0.236). As shown in Table II, the higher range of elbow flexion (retest 2: 99 ^ 68;
post-test: 104 ^ 108) of the training participant might have had a positive effect on ball
velocity (retest 2: 22.5 ^ 0.5 m/s; post-test: 23.6 ^ 0.6 m/s), although the values of the
model participant (53 ^ 98) are at odds with this notion. The range of external shoulder
rotation increased significantly from the pre-test (95 ^ 288) to retest 1 (318 ^ 428), after
which it settled down to that of the model participant (204 ^ 68) (retest 2: 241 ^ 218; retest
3: 240 ^ 118; post-test: 209 ^ 128). For dorsal wrist flexion, no significant differences were
observed between tests of the training participant or between the training participant and
model participant.
Maximal angular velocity
During the acceleration phase of the throw, internal shoulder rotation (for external rotation
we observed negative values, thus we chose internal rotation) results in movement of the
throwing arm in the direction of motion and thereby the final phase of the throw up to ball
release. As shown in Table III, the model participant realized very high values
(v 8130 ^ 12008/s), which exceeded those of top Austrian players (v 5610 ^ 9308/s;
see Wagner et al., 2006) as those of professional baseball players (v 7240 ^ 10908/s; see
Fleisig, Barrentine, Zheng, Escamilla, and Andrews, 1999). Conversely, the training
participant recorded much lower values (v max. 3620 ^ 9908/s) and the differences
between the model participant and all values of the model participant were highly significant
Table II. Angle range of selected angle time courses (mean ^s)

Pre-test
Retest 1
Retest 2
Retest 3
Post-test
Test (MP)

Wrist angle [8]


(dorsal flexion)

Elbow angle [8]


(flexion)

Shoulder angle [8]


(abduction)

Shoulder angle [8]


(external rotation)

54 ^ 4
49 ^ 6
51 ^ 10
50 ^ 11
55 ^ 18
49 ^ 5

59 ^ 5
62 ^ 4
99 ^ 6
76 ^ 10
104 ^ 10
53 ^ 9

73 ^ 10
64 ^ 7
52 ^ 9
51 ^ 7
31 ^ 11
59 ^ 10

95 ^ 28
318 ^ 42
241 ^ 21
240 ^ 11
209 ^ 12
204 ^ 6

MP model participant.

66

H. Wagner & E. Muller


Table III. Maximal angular velocities of selected angular velocity time courses (mean ^s)

Pre-test
Retest 1
Retest 2
Retest 3
Post-test
Test (MP)

Wrist angle [8/s]


(dorsal flexion)

Elbow angle [8/s]


(flexion)

Shoulder angle [8/s]


(abduction)

Shoulder angle [8/s]


(internal rotation)

1840 ^ 300
1300 ^ 110
1400 ^ 210
1500 ^ 300
1530 ^ 500
1440 ^ 330

1260 ^ 140
1470 ^ 130
1600 ^ 190
1680 ^ 340
1820 ^ 210
1070 ^ 180

520 ^ 40
440 ^ 30
480 ^ 60
450 ^ 110
480 ^ 100
540 ^ 90

1860 ^ 1400
3620 ^ 990
1910 ^ 370
3150 ^ 880
2110 ^ 230
8130 ^ 1200

MP model participant.

(P 0.000**). For shoulder abduction and dorsal flexion of the wrist there were no
significant differences between the pre-test of the model participant and those of the training
participant except shoulder abduction for retest 1 (P 0.009**) and dorsal wrist flexion for
the pre-test (P 0.010*). The highest maximal angular velocity of elbow flexion was
measured in the post-test of the training participant (v 1820 ^ 2108/s). This value is
much higher than those of the model participant (v 1070 ^ 1808/s) and the difference
between them was also highly significant (P 0.000**).
Discussion and implications
Quality of movement
The main focus of the training intervention was to increase ball velocity. In each training
phase, ball velocity increased continuously from the pre-test (21.1 ^ 0.4 m/s) to post-test
(23.6 ^ 0.6 m/s) of the training participant, except retest 3 (21.6 ^ 0.6 m/s). This dramatic
increase in performance (more than 10% increase in ball velocity) of a high-standard athlete
is remarkable. Therefore, the applied training method is very effective for this complex,
acyclic, highly dynamic movement. However, considering the wave-like course of ball
velocity of the training participant over the complete training period and the decreased ball
velocity seen at retest 3 after variable training practice, it could be interpreted that this type of
training would be counterproductive for this participant. Since the method of variable
training arose from schema theory (cf. Schmidt, 1975, 1976, 1988), these results are
comparable to those of Schollhorn (2001, 2003), who suggested differential training is more
effective than traditional training. But in this context we should clarify the limitations of the
study. Because only one participant was used during training, we were able to create a
practical study that involved intensive continuous training at a high standard over one year.
We are unable, however, to categorically whether differential training is more effective than
variable training in general. This question should be addressed in future research.
Proximal-to-distal sequence and maximal joint velocity
Joris et al. (1985) showed that a high ball velocity depends on an optimal proximal-to-distal
sequence. This proximal-to-distal sequence was seen in all throws of the post-test of the
training participant. The maximal speed and the instant in time at which each segment
reached its maximum increased from one segment to the next in a continuous fashion.
It would appear that the training participant produced great trunk torsion due to a tension
that is resolved in the acceleration phase, and thus transfers the impulse from segment
to segment. The advantage of this shooting technique is a relatively high end speed due

Figure 7. Velocity courses of selected joints and ball velocity course (A: Training participant: post-test, 6th throw; B: Model participant: pre-test, 6th throw)

Differential and variable training in the handball throw


67

68

H. Wagner & E. Muller

to a longer acceleration distance. The relative time to the reverse point is about 400 ms (cf.
Figure 7A). This movement in handball is preferred in particular in jump shots from a
distance and is similar to the tennis serve (cf. Fleisig et al., 2003). The model participant, in
contrast, rotated his trunk quickly in the movement direction which led to a high torque of
the shoulder joint. This external rotation in the shoulder results in an acceleration of the
shooting arm in the direction of motion. Therefore, the instant of maximum speed is
delayed. In relation to the shoulder, the time of the whole movement is shortened (300 ms;
cf. Figure 7B). This form of execution is usually undertaken for shots from the back court
when the circumstances afford a fast shot, or to surprise the defence (in accord with the
results of Fradet et al., 2004). It can thus be concluded that the two participants prefer
different throwing techniques to reach maximal ball velocity. It must be noted that, during
training, the training participant was never given instructions about which throwing
technique he should use. This occurred on a self-organized basis.
The maximal segment speeds (hip, shoulder, elbow, wrist, middle hand, finger) and ball
velocities of both participants (cf. Table I) accentuate the need to reach a maximal segment
speed from proximal to distal to release the ball as fast as possible, independently of
performance standard (high values for the model participant as well as versus the post-test of
the training participant). These findings are in line with those of other studies of handball
(Fradet et al., 2004; Joris et al., 1985, Van den Tillaar and Ettema, 2004; Wagner et al.,
2006). But some of the currect data (retest 1) also suggest higher joint velocities do not
automatically produce a faster ball velocity. The transfer of the impulse from the fingers to
the ball might not be optimal. It could be that the ball scrolls over the fingers, which occurs
during competition if the players do not use sufficient glue.
Angle range and maximal angular velocity
Regarding the increasing range of external shoulder rotation and elbow flexion from pre-test
(shoulder rotation: 95 ^ 288; elbow flexion: 59 ^ 58) to post-test (shoulder rotation:
209 ^ 128; elbow flexion: 104 ^ 108), our results show the importance for the training
participant to increase these joint amplitudes to increase movement quality. The increase in
range of elbow flexion might correspond to an individually optimal movement solution for
the training participant because it is at odds with that of the model participant. However, this
was purposefully provoked by the exercises during the differential training phases, especially
by the variation in joint movement. The decreasing range of elbow flexion after retest 3
(variable training) might be the cause of the observed reduction in ball velocity at retest 3.
The discussion thus far has demonstrated increasing ball velocity with training for the
training participant. But what were the differences between the two participants? The answer
lies in the values for maximal angular velocity, especially that of internal shoulder rotation of
the model participant, which was markedly higher (more than 100%) than that of the
training participant. These maximal angular velocities might be the main reason for the
higher ball velocities of the model participant versus the training participant. Furthermore,
there were no significant differences for dorsal wrist flexion, elbow flexion or shoulder
abduction. These results are similar to those of Van den Tillaar and Ettema (2004), who
emphasized the need for maximal internal shoulder rotation to realize a high ball velocity,
and Wagner et al. (2006), who pointed out that optimal coordination of the shoulder is the
most important factor for high ball velocity of top-class players versus inefficient players.
Interpreting the results of this study and the results of Van den Tillaar and Ettema (2004)
and Wagner et al. (2006), it can be concluded that increasing the maximal angular velocity of
internal shoulder rotation should produce an increase in ball velocity. But to increase internal

Differential and variable training in the handball throw

69

shoulder rotation requires many years of training with a combination of different training
methods such as handball-specific coordination and strength training.
The training participant, however, achieved a high ball velocity through fast elbow flexion
(same results for the angle range), because at the post-test, where ball velocity was highest
(23.6 ^ 0.6 m/s), the maximal angular velocity of elbow flexion was also highest
(1820 ^ 2108/s). The angular velocity of shoulder abduction and internal rotation might
have played a subordinated role for the training participant (cf. Table III). For maximal
angular velocity of dorsal wrist flexion and shoulder abduction, no significant differences
were observed between the throws of the model participant and those of the training
participant. To identify a connection between the training intervention and the measured
variables, it should be noted that maximal angular velocity of elbow flexion increased
significantly between retest 1 and 2; but during this second training phase (differential
training to increase ball velocity), maximization of elbow flexion angular velocity was also
purposefully trained.

Practical implications and recommendations for coaches


Based on the results of this study, we suggest introducing differential training into the normal
training process. To obtain optimal results, it makes sense to train the differential method
separately; but, it is also possible that this training method could be part of another training
session as long as one does not overstrain the athletes. The authors recommend two training
sessions of 6 8 weeks (2 3 sessions per week with a maximum of 60 repetitions). In the first
training session, not more than two skills should be combined. Only in the second session
can more than two skills be combined. The athlete, however, should be relaxed. As with
other training modes of specific coordination training, a systematic structure, from easy to
complex skills, is necessary to enable correct execution of all exercises. Because there are no
clear guidelines for the execution of this training method, as there is for endurance or
strength training, success is apart from the athletes desire strongly linked to the
knowledge of the exact movement sequence by the trainer. The role of the trainer changes
from one of observer (fixing and controlling lactate, heart rate, and time limits) to an active
and central role in training.

Conclusion
Differential training can be recommended as a mode of training to increase the throwing
ability of high-performance athletes, whether elite handball athletes or athletes in other
sports where optimal coordination behaviour determines the standard of performance.

References
Bernstein, N. A. (1967). The coordination and regulation of movement. Oxford: Pergamonn Press.
Catalano, J. F., and Kleiner, B. M. (1984). Distant transfer in coincident timing as a function of practice variability.
Perceptual and Motor Skills, 58, 851856.
Fleisig, G. S., Barrentine, S. W., Zheng, N., Escamilla, R. F., and Andrews, J. R. (1999). Kinematic and kinetic
comparison between baseball pitching and football passing. Journal of Biomechnaics, 12, 207224.
Fleisig, G., Nicholls, R., Elliott, B., and Escamilla, R. (2003). Kinematics used by world class tennis players to
produce high-velocity serves. Sport Biomechanics, 1, 51 71.
Fradet, L., Botcazou, M., Durocher, C., Cretual, A., Multon, F., Prioux, J. et al. (2004). Do handball throws always
exhibit a proximal-to-distal segment sequence? Journal of Sports Sciences, 5, 439 447.

70

H. Wagner & E. Muller

Haken, H., Kelso, J. A., and Bunz, H. (1985). A theoretical model of phase transitions in human hand
movements. Biological Cybernetics, S1, 347356.
Hatze, H. (1986). Motion variability. Its definition, quantification and origin. Journal of Motor Behavior, 18, 516.
Hertz, J., Krogh, A., and Palmer, R. G. (1991). Introduction to the theory of neuronal computation. New York: Addison
Wesley.
Joris, H. J. J., Edwards van Muyen, A. J., van Ingen Schenau, G. J., and Kemper, H. C. G. (1985). Force, velocity
and energy flow during the overarm throw in female handball players. Journal of Biomechanics, 6, 409414.
Kelso, J. A. S. (1981). On the oscillatory basis of movement. Bulletin of the Psychonomic Society, 18, 63.
Kelso, J. A. S. (1984). Phase transitions and critical behaviour in human bimanual coordination. American Journal of
Physiology, 15, 10001004.
Kelso, J. A. S. (1997a). Dynamic patterns: The self-organization of brain and behavior (2nd edn.). Cambridge: MIT
Press.
Kelso, J. A. S. (1997b). Relative timing in brain and behaviour: Some observations about the generalized motor
program and self-organized coordination dynamics. Human Movement Science, 16, 453460.
Kelso, J. A. S., and Zanone, P. G. (2001). Coordination dynamics and transfer across different effector systems.
Journal of Experimental Psychology: Human Perception and Performance, 28, 776797.
Lee, T. D., and Magill, R. A. (1983). The locus of contextual interference in motor skill acquisition. Journal of
Experimental Psychology: Learning, Memory and Cognition, 9, 730 746.
McCracken, H. D., and Stelmach, G. E. (1977). A test of the schema theory of discrete motor learning. Journal of
Motor Behavior, 9, 193 201.
Muller, H., and Loosch, E. (1999). Functional variability and an equifinal path of movement during targeted
throwing. Journal of Human Movement Studies, 36, 103 126.
Newell, K. M., and Corcos, D. M. (1993). Variability and motor control. Chicago: Human Kinetics.
Roth, K. (1989). Taktik im Sportspiel (Tactics in the sport game). Schorndorf: Hofmann.
Schmidt, R. A. (1975). A schema theory of discrete motor skill-learning. Psychological Review, 82, 225260.
Schmidt, R. A. (1976). The schema as a solution to some persistent problems in motor learning. In G. E. Stelmach
(Ed.), Motor control: Issues and Trends (pp. 4166). New York: Academic Press.
Schmidt, R. A. (1988). Motor control and learning: A behavioral analysis. Champaign: Human Kinetics.
Schmidt, R. A., and Lee, T. D. (1999). Motor control and learning. Champaign: Human Kinetics.
Schmidt, R. A., and Wrisberg, C. A. (2001). Motor learning and performance: A problem-based learning approach
(2nd edn.). Champaign: Human Kinetics.
Schollhorn, W. (2000). Practical consequences of systems dynamic approach to technique and strength training.
Acta Academiae Olympique Estonia, 8, 2537.
Schollhorn, W. (2001). Discrete and continuous effects of traditional and differential sprint training. In J. Mesters,
G. King, H. Struder, E. Tsolakidis, and A. Osterburg (Eds.), Book of abstracts of the 6th ECSS Congress in Cologne
(p. 311). Cologne: ECSS.
Schollhorn, W. (2003). Coordination dynamics and its relevance to sport and exercise. In E. Muller,
H. Schwameder, G. Zallinger, and V. Fastenbauer (Eds.), Book of abstracts of the 8th ECSS Congress in Salzburg
(p. 170). Salzburg: ECSS.
Shea, C. H., Kohl, R., and Indermill, C. (1990). Contextual interference: Contribution of practice. Acta
Psychologica, 73, 145157.
Shea, J. B., and Morgan, R. L. (1979). Contextual interference effects on the acquisition, retention and transfer of a
motor skill. Journal of Experimental Psychology: Learning, Memory and Cognition, 5, 179 187.
Van den Tillaar, R., and Ettema, G. (2004). Force-velocity relationship and coordination patterns in overarm
throwing. Journal of Sport Science and Medicine, 3, 211 219.
Wagner, H. (2005). Optimierung komplexer Bewegungsmuster bei Wurfbewegungen. Aachen: Meyer & Meyer.
Wagner, H., and Blickhan, R. (1999). Stabilizing functional of skeletal muscle: An analytical approach. Journal of
Theoretical Biology, 199, 163179.
Wagner, H., Klous, M., and Muller, E. (2006). Kinematics of the upward jumping throw in handball: Comparison
of players with different level of performance. In H. Schwameder, G. Struzenberger, V. Fastenbauer,
S. Lindinger, and E. Muller (Eds.), Proceedings of the XXIV International Symposium on Biomechanics in Sports
(pp. 161164). Salzburg: Department of Sport Science and Kinesiology, University of Salzburg, Austria.
Walter, C. (1998). An alternative view of dynamical systems concepts in motor control and learning. Research
Quarterly for Exercise and Sport, 4, 326333.
Wulf, G., and Lee, T. D. (1993). Contextual interference in movements of the same class: Differential effects on
program and parameter learning. Journal of Motor Behavior, 25, 254 263.
Zanone, P. G., and Kelso, J. A. S. (1992a). Evolution of behavioral attractors with learning: Nonequilibrium phase
transitions. Journal of Experimental Psychology: Learning, Memory and Cognition, 18, 403421.

Differential and variable training in the handball throw

71

Zanone, P. G., and Kelso, J. A. S. (1992b). Learning and transfer as dynamical paradigms for behavioral change.
In G. E. Stelmach and J. Requin (Eds.), Tutorials in motor behaviour (pp. 563582). Amsterdam: Elsevier.
Zanone, P. G., and Kelso, J. A. S. (1997). The coordination dynamics of learning and transfer: Collective and
component levels. Journal of Experimental Psychology: Human Perception and Performance, 23, 14541480.
Zanone, P. G., and Kostrubiec, V. (2004). Searching for (dynamic) principles of learning. In V. K. Jirsa and
J. A. S. Kelso (Eds.), Coordination dynamics: Issues and trends (pp. 5789). Berlin: Springer.

Appendix: Variables selected for kinematic analysis


Based upon the digitized points shown in Figure 8, joint angles were calculated for the
shoulder (abduction and external rotation), elbow (flexion), and wrist (dorsal flexion) of the
right arm, and the velocity time courses for the finger (second distal phalanx), middle hand
(head of the second metatarsal), wrist, elbow, shoulder (right arm), and hip segments.

Figure 8. Marker locations and joint angles (wrist: dorsal flexion; elbow: extension; shoulder: abduction and external
rotation) of the upper body.

Você também pode gostar