Acta Psychologica North-Holland
269
81 (1992) 269-286
The role of vision in the temporal and spatial control of handwriting Robert
R.A. van Doorn
Free University, Amsterdam, Accepted
March
and Paul J.G. Keuss
The Netherlands
1992
The general observation that handwriting is not noticeably impaired by the withdrawal of vision can be explained in two ways. One might argue that vision is not needed during the act of writing. Micro-analyses should then reveal that spatial as well as temporal writing features are identical in conditions of vision and no vision. Alternatively, it is possible that vision is needed during the act of writing, but that without vision possible errors and inaccuracies have to be prevented. Assuming that the latter would place an extra demand on movement control, this should be revealed by an increase in processing time. We have found evidence for the latter view in the present study in which 12 subjects wrote a nonsense letter sequence with and without vision. Close examination showed that writing shapes remained equally invariant under both vision conditions. suggesting that spatial control was unaffected by withdrawing vision. The prediction that invariance of shapes is preserved in the absence of vision at the expense of processing time increments was confirmed. The increase of reaction time observed when visual guidance was withdrawn suggests that more processing time was needed prior to the movement start. Moreover, the RT increment was larger when a short writing duration was instructed. The present findings will be discussed in light of the remarkable flexibility of writing as a motor skill in which writers appear to be able to employ specific strategies to preserve shape in the absence of visual guidance.
Introduction Experimental findings generally have led to the agreement that visual feedback serves a specific function in the performance of skilled movements (Keele 1986; Schmidt 1988). However, movement skills differ as to the impact of vision on their output. Studies on ball interception in sports suggest that movement output and visual input Correspondence to: R.R.A. van Doorn, Vrije Universiteit, Faculteit der Psychologie en Pedagogische Wetenschappen, Vakgroep Psychonomie, De Boelelaan 1111, 1081 HV Amsterdam, The Netherlands.
OOOl-6918/92/$05.00
0 1992 - Elsevier
Science
Publishers
B.V. All rights reserved
are strongly related (Bootsma and Van Wieringen 1990; Savelsbergh et al. 1991). For the latter type of movements, moment-to-moment visual control of the approaching ball is considered necessary. Handwriting movements, however, are considered to be far less dependent on a moment-to-moment visual guidance as they are assumed to be prestructured and highly organized (Van Galen 19901. In handwriting, vision is believed to function as a background monitor (Van Galen et al. 1989) which suggests a rather weak relationship between movement output and visual input. One would expect that when motor control does not strictly depend on peripheral information, performance should not be heavily impaired when this information is unavailable. But, in handwriting, performance is not totally independent of visual feedback. On the contrary, it has been found that during handwriting more writing errors occur (omissions, repetitions and transpositions of letters and strokes) when visual feedback is absent (Smyth and Silvers 19871. The mere fact that writing errors occur, indicates the involvement of visual feedback in a normal writing situation. Yet, since writing errors are rare, one might argue that even without visual guidance, writing is not affected drastically. According to this view, vision is not a strict prerequisite for handwriting because, apart from an occasional writing error, writing is not altered by the withdrawal of vision. Alternatively, one could argue that the rare occurrence of writing errors without vision indicates that a larger number of errors is prevented. One might infer that having to prevent errors under a no-vision condition should become manifest in writing performance. A typical demonstration of the latter view is provided by Smyth (19891. Without the availability of visual feedback subjects minimized pen lifts during the writing of capital letters. During the production of capital letters, pen lifts consist of visually guided repositioning movements above the writing surface towards an anticipated spatial location on paper. Without visual guidance this type of repositioning is prone to errors. Thus, in order to avoid inaccurate repositioning which results in spatial errors, the use of pen lifts is minimized in the absence of vision which means that normal writing is altered. Besides capital letters, adults often use connected lower-case script which results in cursive handwriting. In contrast to capital letter production, cursive letter production requires hardly any pen lifts. Thus, without vision, repositioning errors due to pen lifts are hardly to
R.R.A. mn Doom, P.J.G. Keuss / The role of rision in handwriting
271
be expected in cursive handwriting. Nevertheless, shape preservation of strokes might still be threatened by the withdrawal of visual guidance, resulting in accuracy variations. However, apart from the observed writing errors, Smyth and Silvers (1987) did not report any changes in letter shapes, which suggests that either writing accuracy remained unaffected by the withdrawal of vision, or that subjects actively prevented letter deformations by employing compensatory strategies. In order to verify the existence of this type of prevention, its measurement is required. In fact, the act of preventing errors might become manifest in writing behaviour only by means of minor adjustments. In this sense, adjustments, and thus alterations of writing, are accepted by subjects for the sake of constant spatial accuracy. The present paper investigates whether cursive writing without vision occurs without a loss of accuracy and whether to this end adjustments are realized. Therefore, cursive letter sequences are studied which are written with and without the availability of visual feedback. The question rises as to how adjustments will become apparent in cursive writing under a no-vision condition. In other words, what kinds of writing features will change? Like in many other studies on motor control, recent handwriting research has mainly been focussing on temporal features (Van Galen et al. 1990). An increase in task demands has been found to become manifest in increased latencies or reaction times CRT), and increased movement times (MT). These measures have been used to study the characteristics of the processes underlying handwriting (see Van Galen, 1991, for a recent overview). The present paper continues this tradition by exploring the temporal aspects (RT and MT) of writing patterns produced with and without visual feedback. Indeed, temporal features have been found to alter when visual feedback is withdrawn. An increase of movement time throughout a writing sequence was found under a condition without visual feedback Wan Galen et al. 1989). The latter finding suggests that processing during the output phase takes more time. Since processes prior to the output phase might also be involved, RT should be included in the analyses to get a more complete picture. In aiming movements, RT and MT have been shown to be negatively correlated. Phillips and Glencross (1985) found that a decrease of MT was accompanied by an increase of RT. This finding suggests that a decrease of processing load during the output phase leads to an increase of processing load
272
R.R.A. am Doom.
P.J. G. Keuss / The role
of r Sm
in hundwriting
prior to the start. A similar mechanism might be involved when handwriting is performed without visual feedback. The increase of MT under a no-vision condition can be thought of as an attempt to enlarge the total processing time. The core question is whether the lengthening of MT has consequences for RT prior to the start of the movement sequence. Two outcomes are feasible regarding RT when visual information is unavailable before and during the writing output phase. (1) By an increase of MT, sufficient processing time is obtained and RT need not change. (2) The increase of MT does not suffice. Total processing time has not reached an appropriate level and thus, an increase of RT will be needed to reach that level. In order to be able to differentiate between these two situations, we will not only withdraw visual feedback but we will also vary the writing duration instructions. In the present study subjects are instructed to write a letter sequence in a normal, comfortable pace, and in a very high pace requiring minimum movement duration. When more processing time is needed under a no-vision condition, an additional increase in RT is expected in the short-duration condition. The assumption is that if more processing time is needed, but this cannot be accomplished by an increase of MT alone, then RT will have to be lengthened as well. Lengthening of latencies indicates a longer processing time. Without vision these longer processing times might be needed for adjustments of the writing act aimed to prevent inaccurate writing. Therefore, to be sure that the lengthening of latencies is aimed at the prevention of spatial effects resulting from the absence of vision, the spatial accuracies of writing under both visual feedback conditions need to be compared. Writing accuracy has received little attention in research. Accurate writing has been related to the invariance of the trajectory shapes. Generally, it is assumed that shapes remain unaffected under a variety of conditions (Maarse et al. 1989; Teulings et al. 1986; Thomassen and Teulings 1988). There are, however, some indications that constant shapes are not always realized. Recently, it was found that writing in different sizes causes horizontal and vertical letter spacing to be resealed in unequal proportions (Wann and Nimmo-Smith 1990). The issue in the present paper is whether shapes remain invariant when
R. R.A. uan Doom, P.J.G. Keuss / The role of vision in handwriting
213
visual feedback is unavailable. It is reasonable to assume that shapes will vary more when the absence of visual feedback is too demanding. In our line of reasoning this will be the case when temporal alterations do not create a sufficient amount of processing time needed to maintain the spatial invariance. The prediction is that spatial accuracy of written letter sequences will not suffer if lenthening of RT and MT suffices to prevent the deteriorating effects of the absence of visual feedback. If, as in the condition of a shorter instructed movement duration, demands are too high to maintain shape invariance, a decline in accuracy should be the result. This should be even more pronounced when visual feedback is absent, and an increase of MT is prevented by a short duration instruction.
Method Subjects Twelve right-handed students from the Free University participated as paid subjects. The selection criterion for taking part was their familiarity with cursive script production. Apparatus and data recording Data were recorded with a ‘Summagraphics Bitpad One’ tablet shaped digitizer, connected with an Olivetti M 280 personal computer. Recording was done with a sampling rate of 95.2 Hz and a spatial reliability of 0.1 mm. The recorded writing patterns were translated into the frequency domain (by means of a forward Fast Fourier Transform, FFT) after which all frequencies above 12.5 Hz were cut off. Subsequently, a backward translation into the spatial domain (backward FFT) was undertaken (Teulings and Maarse 1984). This method resulted in smooth and reliably interpolated writing patterns. Procedure, task and design Subjects were seated in front of a table, in the centre of which the digitizing tablet was positioned. Because the digitizer was slightly tilted, the table was adjusted to obtain a flat writing surface. Subjects were free to use their habitual writing posture. Writing had to be performed on a blank sheet without lineation taped on the tablet. A normal ball-point pen was used. The pen was attached to a flexible wire which did not interfere with normal writing. Subjects wrote the letter sequence lenehele, being a nonsense word in Dutch, under two conditions of Vision (vision and no-vision) and
two conditions of Instructed Duration (normal and short). In the vision condition, subjects were able to look at their writing performance. In the no-vision condition, an opaque box was placed over the tablet in order to preclude perception of the writing hand and stylus, without disturbing the writing movements. In the short Instructed Duration condition subjects had to write the letter sequence within exactly 2.0 seconds. In the normal Instructed Duration condition subjects were permitted to finish the entire sequence within 2.5 seconds. Note that these instructions differ from speed instructions which are generally used in experiments with speed instructions. In the present experiment we specified the durations of the time interval in which the letter sequence had to bc finished in the two conditions. Each subject performed two blocks of trials which differed with respect to Instructed Duration. Each block consisted of two series of 15 trials. A series was related to the availability of Vision which was either present or absent. The reason for nesting the feedback blocks within the Instructed Duration blocks was to get subjects acquainted with one specific Instructed Duration that was fixed throughout both conditions of Vision. Randomization of Instructed Duration across trials would have been confusing and prone to violation of the instruction. Blocks and series were counterbalanced across subjects. Before each Instructed Duration block, the semantically and phonologically simple letter sequence lenehele was presented in printed capital letters on the computer screen approximately 1.0 m in front of the subject. Each trial block started with 20 training trials to get the subject acquainted with writing the letter scquencc according to the instructions A trial started with a low-pitch warning signal, whereupon the subject had to position the stylus on paper. A high-pitch tone signalled the subject to commence as quickly as possible with writing the letter sequence. Subjects were instructed and trained having completed the sequence when a second high-pitch tone was presented, so that writing duration was specified by the time interval between the two high-pitch tones. The letter scquencc consisted, in its cursive form, of 10 upstrokes and 10 downstrokes. Given the estimated minimum stroke duration of 100 ms (Teulings and Maarse 1984) which people generally riced in fast handwriting, the short Instructed Duration condition (2 s) required subjects to write at a maximum speed. The time interval between trials was 2500 ms. The simplicity of the letter sequence ensured that writing errors as they were reported in the study of Smyth and Silvers (1987) were not likely to occur in the present study. Analyses
Reaction Time (RT) was defined as the duration of the interval between the first high-pitch tone (imperative stimulus) and the first increase of the tangential velocity of the writing movements. RT was measured for each trial and averaged over trials per condition. Preliminary checks of the data showed that subjects did not produce any writing errors but that in the more demanding conditions (No-Vision and short Instructed Duration) they were not always able to comply with the Duration Instruction. In 32% of the trials, subjects failed to product the sequence in time, due to which final letters
R.R.A. fun Doom, P.J.G. Keuss / The role of l,ision in handwriting
275
were not recorded. In order to analyze a comparable data set across subjects and conditions, the final three letters of each sequence were excluded from analyses. Movement time, trajectory size and spatial variability of the sequence (consisting of the initial five letters) were determined. Moreover, the same measures were calculated for each of the initial four strokes of the letter sequence, separately. Borders between individual strokes were defined by tangential velocity minima (Lacquaniti et al. 1984). To analyze the spatial variability across replications, each stroke was normalized in size, time and orientation. Spatial variability was determined by the sum of spatial distances of each stroke from the average stroke, obtained via orthogonal projection, across the replications per condition (Maarse et al. 1989). Spatial variability of the trajectory consisting of the ‘initial five letters was aquired by accumulating the spatial variabilities of the initial 14 strokes. Measurements were averaged across 15 replications, which was done for the initial four strokes and across the initial five letters. We were also interested in the variation of movement duration and trajectory size across replications within each condition. For that purpose, we calculated for each of the initial four strokes separately the relative standard deviation of movement time across replications, i.e. the standard deviations divided by the average across replications. A similar procedure was followed for trajectory size. Finally, for each subject, correlations were determined between RT and MT of the first stroke, and between RT and spatial variability of the first stroke. Subsequently, the correlation coefficients were converted into Fisher z-scores and analyzed by means of an ANOVA. Within-subject repeated measurement ANOVA’s on each of the dependent variables were carried out with the individual means of the dependent variable for each condition and subject as cell entries. Dependent variables, measured for the initial four strokes were entered into separate ANOVA’s according to an Instructed Duration (short/normal) X Vision (yes/no> X Letter (l/e) X Stroke (up-/downstroke) factorial design. The dependent variables that were determined across the initial five letters, and the Fisher z-scores, were analyzed according to an Instructed Duration (short/normal) x Vision (yes/no) factorial design.
Results
Correlations
Correlations between RT and MT of the initial stroke varied between -0.30 and 0.80. Correlations between RT and the shape variability of the initial stroke varied between - 0.40 and 0.70. No effects of Vision and Instructed Duration were found on the Fisher z-scores which were derived from the correlation coefficients between RT and MT of the initial stroke and the correlation coefficients between RT and spatial variability of the initial stroke (a > 0.1).
276
R.R.A. van Doom, P.J.G. Keuss / The rvle of vision in handwriting 2oo !??action
Time (ms)
Vision
Fig. 1. Mean reaction time (vertical axis) as a function of Vision (horizontal axis) and Instructed Duration (black bar = normal Instructed Duration, shaded bar = short Instructed Duration).
Reaction time
As a main factor, vision contributed to the observed variation in RT (F(1, 11) = 16.52, p < 0.01). As fig. 1 illustrates, RT was lengthened by 20 ms in the no-vision condition as compared to the vision condition. Instructed Duration had no effect on RT (F(1, 11) = 0.73, ns). However, the RT increment under no vision was larger when the Instructed Duration was short, but the interaction was only marginally significant (F(1, 11) = 4.15, p = 0.066).
Mol~ement lime
Fig. 2 shows the cffccts on MT across the initial five letters as a function of Vision and Instructed Duration. Movement time was identical in both vision conditions (F(1, 1I) = 0.03, ns). Fig. 2 also illustrates that movement time was clearly shorter under the condition of short Instructed Duration, F(1, 11) = 9.27, p < 0.05, indicating that, in general, the subjects obeyed the duration instructions. However, subjects failed to comply to the short Duration Instruction in the no-vision condition, because in this condition movement time was 56 ms longer than allowed (i.e. 1400 ms for the initial 14 strokes that were selected for the analyses). The interaction between Instructed Duration and Vision was significant, F(1, 11) = 6.45, p < 0.05. Variations in MT of the initial four strokes (constituting the initial two letters) as a function of Vision and Instructed Duration arc shown in fig. 3. Movement time was shorter when Instructed Duration was short (F(1, 11) = 366.06, p < 0.001). There was no main effect of vision on movement time (F(1, 11) = 0.29, ns). Fig. 3 shows that movement time decreased with Stroke position, indicated by the significant effects of Letter. F(1, 11) = 138.97, p < 0.001, and Stroke. F(1, 11) = 50.90, p < 0.001). In the first letter (I), the differcncc between the durations of up- and downstrokes was more
277
R.R.A. uan Doom, P.J.G. Keuss / The role of Gsion in handwriting
Movement ,800~--_-__
Time (ms) -~
I
Instructed Duration
Vision Fig. 2. Mean movement time (vertical axis) of the first five letters of the letter sequence lenehele as a function of Vision (horizontal axis) and Instructed Duration (black bar = normal Instructed Duration, shaded bar = short Instructed Duration).
outspoken than in the second letter (e), as indicated by the significant interaction Letter X Stroke, F(1, 11) = 8.18, p < 0.05. The interaction between Instructed Duration and Vision for the initial four strokes was not significant (F(1, 11) = 2.21, p < 0.1) indicating that the Instructed Durations were obeyed well in the initial part of the sequence even when Vision was withdrawn
Movement
Time
(ms) Letter
e Vision
Instructed Duration
175:
I
\ normal .
150
r
normal
" short '\ b
125l
t
short 1
F 100
75
. I1
2
Stroke
1
2
Number
Fig. 3. Mean movement time (vertical axis) of the initial four strokes ing the initial two letters of the sequence lenehele as a function Duration.
(horizontal of Vision
axis) constitutand Instructed
70
Trajectory
Size (mm)
Yes
Vision Fig. 4. Mean trajectory
a function
of Vision
size (vertical (horizontal Duration.
“’
axis) of the first five letters of the letter sequence lenehelr
axis) and Instructed
Duration
shaded bar = short Instructed
(black
bar = normal
as
Instructed
Duration).
and Instructed Duration was short. The difference between short and normal Instructed Duration was larger for the first and second stroke constituting the letter 1, as compared to the first and second stroke of the letter e, as indicated by the interaction Instructed Duration X Letter, F(1, 11) = 12.92, p < 0.001. The relative standard deviation of movement time did not vary as a function of Instructed Duration (F(1. 1I) = 1.68. ns), Vision (F(1, 1 I) = 0.05, ns). Letter (Hl, 11) = 0.05, ns), and Stroke (FCI. 1 I) = 0.41. ns).
Trujectory
size
Variations in the mean Trajectory Size are shown in fig. 4 (across the initial five letters of the writing sequence) and fig. 5 (for the initial four strokes constituting the first two Ictters). Fig. 4 shows a larger mean Trajectory Size of the initial five letters in the no-vision condition than in the vision condition (F(1. 11) = 11.20, p < 0.05). A similar main effect of vision was found for the initial four strokes (F(1. 11) = 8.81, p < 0.05). Trajectory Size of the initial five letters and of the initial four strokes did hot vary as a function of Instructed Duration (RI, 1I) = 0.34, ns), and (F(1, I I) = 0.10, ns). respectively). No interactions between Instructed Duration and Vision on Trajectory Size across the initial five letters (F(1, I I) = 0.42, ns) and across the initial four strokes (F(1, 11) = 0.17. ns) were observed. Fig. 5 illustrates that the effect of Vision was larger in the letter 1 than in the letter e (F(1, I I) = 4.26, p = 0.063). Morcovcr, Vision had a stronger impact on upstrokes than on downstrokes, as indicated by the main effect of Stroke, F(1, 11) = 5.73, p < 0.05, which might be related to the fact that in the present experiment an upstroke is always an initial stroke of a letter. These issues will bc focussed on in the discussion section.
R. R.A. uan Doom, P.J. G. Keun / The role of cision in handwriting Size (mm) ,. Trajectory ~_~ ~ __~__ -~~~ __ Letter
1 Letter
J
279
,
~~
I
e
654 3
Fig. 5. Mean trajectory Size (vertical axis) of the initial four strokes (horizontal axis) constituting the initial two letters of the sequence lenehele as a function of Vision and Instructed Duration.
The relative standard deviation experimental variables (p > 0.2).
of Trajectory
Size did not vary as a function
of the
Spatial uariabili& Across the initial five letters of the sequence, spatial variability decreased when subjects had to perform the task according to the short Instructed Duration (see fig. 6), which was statistically confirmed (F(1, 11) = 16.45, p < 0.05). However, in the
Spatial Variability 507
~
~~~~
---1
-~
45-
I
Instructed Duration
no
Yes
Vision Fig. 6. Mean spatial variability measures (vertical axis: depicted in arbitrary units resulting from normalisation in time, size and orientation) across the initial five letters of the letter sequence lenehele as a function of Vision (horizontal axis) and Instructed Duration (black bar = normal
Instructed Durations shaded bar = short Instructed Duration).
2x0
R.R.A. ~mn Doom.
8 Spatial Letter
P.J.G.
Keuss
/
The role
of ~,ision in handwnting
Variability P
Letter
4
7
Vision t
6
/
I
/
5
Yes
normal
.
“0
normal
Yes
short
.
“0
short
A 4
2
I
1 1
1
Instructed Duration
I 2
Stroke
1
2
Number
Fig. 7. Mean spatial variability (vertical axis: depicted in arbitrary units resulting from normalisation in time, size and orientation) of the initial four strokes (horizontal axis) constituting the initial two letters of the sequence lerzehele as a function of Vision and Instructed Duration.
initial four strokes (see fig. 7) this effect was not found (F(1, 11) = 0.07, ns). Fig. 6 suggests a slight increase of spatial variability, measured across the initial five letters, in the no-vision condition as compared to the vision condition. This effect seemed most pronounced in the normal Instructed Duration condition. Fig. 7 does not show such increments for the initial four strokes. Statistical tests revealed, however, that neither the main effects of vision on spatial variability measured across the initial five letters (F(1, 11) = 0.38, ns) and across the initial four strokes (F(1, 11) = 0.07, ns), nor any interaction between Vision and Instructed Duration across the initial five letters (F(1, 11) = 1.17, ns) and across the initial four strokes (F(1, 11) = 0.01, ns) became significant. The spatial variability of the letter e was larger than of the letter I (F(1, 11) = 9.42, p < 0.05).Also, across the initial four strokes, shapes of downstrokes were more variable than of the shapes of upstrokes, as indicated by the main effect of Stroke (F(1, 11) = 12.24, p < 0.01). This was more outspoken in the letter e, as indicated by the significant interaction Letter X Stroke, F(1, 11) = 8.33, p < 0.05.
Discussion The present experiment confirmed the impact that vision has on latencies of letter production in cursive handwriting. Without visual feedback more time is needed to commence the writing sequence. In the spatial domain, trajectories become larger under the no-vision condition, while spatial variability across replications remains unaf-
R.R.A. uan Doom, P.J.G. Keuss / The role of Laisionin handwriting
281
fected. What do these results mean for a psychomotor theory of handwriting? According to the often proposed explanation concerning the invariance of shapes, spatial variability should not differ as a function of vision (Maarse et al. 1989; Thomassen and Teulings 1988). We argued that under a demanding condition like no vision, strategies may be employed to ensure accurate shape reproduction. Overt manifestations of strategies should become apparent in the temporal domain of the act of writing. A possible strategy is the increase of processing time. In the present paper, the relation between RT and MT was considered in this context. The idea is that the total processing time needed to perform a motor task is distributed over RT and MT (Phillips and Glencross 1985). Moreover, RT and MT are believed to be linked, which means that a strong decrease of one results in an increase of the other variable so that the total processing time remains constant. Also, under demanding conditions more processing time might be needed leading to increments of the latencies. Writing without vision might be such a demanding condition. Indeed, the finding that without vision the duration of writing movements increased (Van Galen et al. 1989) suggests an increment in processing time during the output. Our prediction that RT should increase when the MT increase is not enough to fulfill the total processing time requirement, could be confirmed, although it was reflected by only a marginally significant interaction between Instructed Duration and Vision on RT (p = 0.066). Nevertheless, the identified increase of RT suggests that the adjusted processing time during the output phase (MT) did not suffice to cope with the absence of vision and that prior to the movement output additional processing was required, The finding of Phillips and Glencross (1985) that in simple aimed movements RT negatively correlates with MT could not be confirmed for writing sequences in the present experiment. In our data, RT and MT were not negatively correlated. However, in the vision condition, a short Instructed Duration resulted in a shorter RT. The present results are not at odds with findings in studies on movement sequences (Hulstijn and Van Galen 1983; 1988; Klapp and Wyatt 1976; Sternberg et al. 19781, in which positive correlations between MT and RT were found. Apparently, due to the more complex nature of the involved movement sequence, being a cursive letter sequence in the present experiment, a shorter movement duration made an earlier
282
R.R.A. c’an Doom, P.J.G. Keum / The role of Lision in handwriting
movement start possible. We may add that the positive relation between MT and RT in movement sequences only holds as long as vision is available. In fact, we could confirm the prediction that the RT increment due to no vision is more pronounced when short movement duration was instructed. The apparent ease with which writers start a sequence under a vision condition when a short duration is expected, vanishes when vision is precluded. Our interpretation of this finding is that additional processing time is needed under the no vision condition. Since the increase of processing time is not possible during movement output in the short Instructed Duration condition, the processing time increment is realized prior to the movement start. We argued that the assumption of a prolonged processing time as a reaction on increased demands in writing, is only valid when the main objective of the writing performance, namely shape consistency, is achieved. Although no correlation was found between RT and shape variability of the initial stroke, the results show that withdrawing visual feedback did not affect the spatial variability of trajectory shapes. Also, in a condition in which the increase of MT was prevented by a short Duration Instruction, writing without visual feedback did not affect spatial variability. To the contrary, the results suggest that subjects try to suppress the variability of shapes under demanding conditions. Under a short Instructed Duration condition spatial variability of the trajectory comprising the initial five letters decreased under both Vision conditions. Still, although in general, subjects proved to be able to cope with the time contraints imposed on their performance; under combined conditions without visual feedback and short writing duration, they were unable to comply with the Instruction Duration (cf. the significant interaction Vision X Instructed Duration on movement time). Apparently, in order to preserve the consistency of shapes they had to violate the Instructed Duration and preserve a minimum of processing time. Time-pressure effects of short Instructed Duration, being the violation of the short Duration Instruction and decrements of spatial variability with short Instructed Duration, were not detectable within the initial four strokes of the sequence, but became manifest only after a larger part of the sequence was achieved. These results might suggest that the shapes of the letters were not all equally vulnerable to external demands, which would explain that in our data only the
R.R.A. mn Doom, P.J.G. Keuss / The role of c&ion in handwriting
283
summated values across different letters were affected. Alternatively, initial strokes of a sequence might be exceptional in that time pressure only had an effect on strokes of subsequent letters. It is possible that initial strokes were affected by starting the sequence, as indicated by their rather long durations (see fig. 3) under the short Instructed Duration condition. It is unclear, whether these start-up effects were related to biomechanical factors or whether programming factors (Portier et al. 1990) were involved. The initial four strokes were dissimilar regarding trajectory size (fig. 5) and spatial variability (fig. 7). The former measure decreased with stroke number, whereas the latter measure increased when writing progressed from the initial stroke to the fourth stroke. Whether these effects were related to stroke positions or whether they might be attributed to letter type differences (1 vs. e) is already being focussed on in a follow-up study. The present results suggest that a strict invariance of shapes does not exist in writing. Since a certain, but limited, spatial variability exists, one might argue that writing shapes are allowed to vary within a tolerance range, which as such constrains shape variability. In this view, writing can be viewed as a rather flexible skill. Certain changes in the temporal domain of writing are allowed as long as (or perhaps so that) this tolerance region is not exceeded. As a consequence, shapes will remain relatively consistent under a variety of conditions (Maarse et al. 1989; Teulings et al. 1986). Note that consistency of shapes was presently determined for repeated writing within each condition separately. Although in the no vision condition shapes were equally consistent as in the vision condition, it is still possible that shapes changed in an absolute sense. The finding of Wann and Nimmo-Smith (1990) that horizontal and vertical sizes change in unequal proportions when larger writing is required is therefore not at odds with the present data. Writing larger than normal might affect horizontal and vertical sizes differently without a decrease of shape consistency. We are presently investigating this idea in follow-up research. However, in the present data, a change of shapes is suggested by the increase of trajectory size in the no-vision condition. Note that with a constant spatial variability, the absolute change of shapes could certainly not result from a loss of control, which is also reflected by the constant (relative) standard deviation of trajectory size in vision and no vision conditions. The question arises to what end trajectory size increases when
284
R.R.A. can Doom, P.J.G. Kmss / The role of vision in handwriting
visual guidance is not possible. It might be that our subjects wrote sloppily, because they were unfamiliar with the no-vision condition. Still, larger trajectories could not have been resulting from a decline of spatial control, as Spatial Variability did not increase in no vision. It might be reasonable to argue that sizes were increased with a specific purpose. Although our subjects were not aware of their enlarged script when vision of their ongoing performance was precluded, they said they relied heavily on what they felt. Since vision was precluded, a fair assumption is that kinesthetic information was used to evaluate shapes. The kinesthetic information mode as compared to the visual mode might use a different shape criterion which could explain why shapes in an absolute sense seem to alter. This is, however, no compelling reason why trajectory sizes should grow. A probable explanation is substantiated by a finding provided by physiological research that finger movements increase in sensitivity when the excursions are made larger (Clark and Horch 1986). In writing it might well be that, in order to employ kinesthetic feedback adequately in absence of vision, subjects increase their writing size. According to this idea an equal amount of spatial information should become available across a larger trajectory. The larger distance might enhance the spatial resolution of the involved kinesthetic receptors, without the requirement of more processing time. In summary, this study has shown that writing under conditions of absence of vision and short Instructed Duration exerts certain increases in latency and movement time to preserve a certain amount of processing time so that trajectory shape output remains consistent. Moreover, handwriting, although a very complex skill, can be performed with a remarkable flexibility. When demands become too heavy, duration instructions are simply discarded in order to fulfill a far more important goal, relative consistency of shape output. Although adult writing undergoes specific changes, without vision, control is not impaired. Therefore, we concur with Smyth and Silvers (1987) that vision is not a strict prerequisite for movement control and the consistent production of shapes in handwriting.
References Bootsma, R.J. and P.C.W. van Wieringen. 1990. Timing and attacking forehand drive in table tennis. Journal of Experimental Psychology, Human Perception and Performance 16. 21-29.
R.R.A. L’WIDoom, P.J.G. Keuss / The role of vision in handwriting
285
Clark, F.J. and K.W. Horch, 1986. ‘Kinesthesia’. In: K.B. Boff, L. Kaufman and J.B. Thomas (eds.), Handbook of perception and human performance, Vol. 1: Sensory processes and performance. London: Wiley. Hulstijn, W. and G.P. van Galen, 1983. Programming in handwriting: Reaction time and movement time as a function of sequence length. Acta Psychologica. 54, 23-49. Hulstijn, W. and G.P. Van Galen, 1988. ‘Levels of motor programming in writing familiar and unfamiliar symbols’. In: A.M. Colley and J.R. Beech (eds.), Cognition and action in skilled behaviour. Amsterdam: Elsevier. Keele, S.W., 1986. ‘Motor Control’. In: K.R. Boff, L. Kaufman and J.P. Thomas (eds.), Handbook of perception and human performance, Vol. II: Cognitive processes and performance London: Wiley. Klapp, S.T. and E.P. Wyatt, 1976. Motor programming within a sequence of responses. Journal of Motor Behavior 8, 19-26. Lacquaniti, F., C. Terzuolo and P. Viviani, 1984. ‘Global metric properties and preparatory processes in drawing movements’. In: S. Kornblum and J. Requin (eds.), Preparatory states and processes: Proceedings of the France-American Conference, MI, Aug. 1982. Hillsdale, NJ: Erlbaum. pp. 357-370. Maarse, F.J., G.P. van Galen and A.J.W.M. Thomassen, 1989. Models for the generation of writing units in handwriting under variation of size, slant, and orientation. Human Movement Science 8, 271-288. Phillips, J. and D. Glencross, 1985. The independence of reaction and movement time in programmed movements. Acta Psychologica, 60, 209-225. Portier, S.J., G.P. van Galen and R.G.J. Meulenbroek, 1990. Practice and the dynamics of handwriting performance: Evidence for shift of motor programming load. Journal of Motor Behavior 22, 474-492. Savelsbergh, G.J.P., H.T.A. Whiting and R.J. Bootsma, 1991. Grasping tau. Journal of Experimental Psychology: Human Perception and Performance 17, 315-322. Schmidt, R.A., 1988. Motor control and learning: A behavioral emphasis. Champaign, IL: Human Kinetics. Smyth, M.M., 1989. Visual control of movement patterns and the grammar of action. Acta Psychologica 70, 253-265. Smyth, M.M. and J. Silvers, 1987. Functions of vision in the control of handwriting. Acta Psychologica 65, 65-73. Sternberg, S., S. Monsell, R.L. Knoll and C.E. Wright, 1978. ‘The latency and duration or rapid movement sequences: Comparisons of speech and typewriting’. In: G.E. Stelmach (ed.), Information processing in motor control and learning. London: Academic Press. Teulings, H-L. and F.J. Maarse, 1984. Digital recording and processing of handwriting movements. Human Movement Science 3, 193-197. Teulings, H-L., A.J.W.M. Thomassen and G.P. van Galen, 1986. “Invariants in handwriting: The information contained in a motor program’. In: H.S.R. Kao, G.P. van Galen and R. Hoosain (eds.), Graphonomics: Contemporary research in handwriting. Amsterdam: Elsevier. Thomassen, A.J.W.M. and H.-L. Teulings, 1988. ‘Grafische productie: De motoriek van schrijven en tekenen’. In: P.J.G. Keuss, G. ten Hoopen and A.J.J. Mannearts (eds.), Menselijke motoriek. Amsterdam: Swetz & Zeitlinger. Van Galen, G.P., 1990. Effects of phonological and motoric demands in handwriting: Evidence for discrete transmission of information. Acta Psychologica 74, 259-277. Van Galen, G.P, 1991. Handwriting: Issues for a psychomotor theory. Human Movement Science 10, 165-191. Van Galen, G.P., M.M. Smyth, R.G.J. Meulenbroek and H. Hylkema, 1989. ‘The role of short term memory and the motor buffer in handwriting under visual and non-visual guidance’. In:
286
R.R.A. can Doom.
P.J. G. Keuss / The role
of vision in handwriting
R. Plamondon, C.Y. Suen and M.L. Simner (eds.), Computer recognition and human production of handwriting. Singapore: World Scientific. Van Galen, G.P., R.R.A. van Doorn and L.B.J. Schomaker, 1990. Effects of motor programming on the power spectral density function of finger and wrist movements. Journal of Experimental Psychology: Human Perception and Performance 16, 755-765. Warm, J.P. and I. Nimmo-Smith, 1990. Evidence against the relative invariance of timing in handwriting, The Quarterly Journal of Experimental Psychology 42A, 105-119.
269
81 (1992) 269-286
The role of vision in the temporal and spatial control of handwriting Robert
R.A. van Doorn
Free University, Amsterdam, Accepted
March
and Paul J.G. Keuss
The Netherlands
1992
The general observation that handwriting is not noticeably impaired by the withdrawal of vision can be explained in two ways. One might argue that vision is not needed during the act of writing. Micro-analyses should then reveal that spatial as well as temporal writing features are identical in conditions of vision and no vision. Alternatively, it is possible that vision is needed during the act of writing, but that without vision possible errors and inaccuracies have to be prevented. Assuming that the latter would place an extra demand on movement control, this should be revealed by an increase in processing time. We have found evidence for the latter view in the present study in which 12 subjects wrote a nonsense letter sequence with and without vision. Close examination showed that writing shapes remained equally invariant under both vision conditions. suggesting that spatial control was unaffected by withdrawing vision. The prediction that invariance of shapes is preserved in the absence of vision at the expense of processing time increments was confirmed. The increase of reaction time observed when visual guidance was withdrawn suggests that more processing time was needed prior to the movement start. Moreover, the RT increment was larger when a short writing duration was instructed. The present findings will be discussed in light of the remarkable flexibility of writing as a motor skill in which writers appear to be able to employ specific strategies to preserve shape in the absence of visual guidance.
Introduction Experimental findings generally have led to the agreement that visual feedback serves a specific function in the performance of skilled movements (Keele 1986; Schmidt 1988). However, movement skills differ as to the impact of vision on their output. Studies on ball interception in sports suggest that movement output and visual input Correspondence to: R.R.A. van Doorn, Vrije Universiteit, Faculteit der Psychologie en Pedagogische Wetenschappen, Vakgroep Psychonomie, De Boelelaan 1111, 1081 HV Amsterdam, The Netherlands.
OOOl-6918/92/$05.00
0 1992 - Elsevier
Science
Publishers
B.V. All rights reserved
are strongly related (Bootsma and Van Wieringen 1990; Savelsbergh et al. 1991). For the latter type of movements, moment-to-moment visual control of the approaching ball is considered necessary. Handwriting movements, however, are considered to be far less dependent on a moment-to-moment visual guidance as they are assumed to be prestructured and highly organized (Van Galen 19901. In handwriting, vision is believed to function as a background monitor (Van Galen et al. 1989) which suggests a rather weak relationship between movement output and visual input. One would expect that when motor control does not strictly depend on peripheral information, performance should not be heavily impaired when this information is unavailable. But, in handwriting, performance is not totally independent of visual feedback. On the contrary, it has been found that during handwriting more writing errors occur (omissions, repetitions and transpositions of letters and strokes) when visual feedback is absent (Smyth and Silvers 19871. The mere fact that writing errors occur, indicates the involvement of visual feedback in a normal writing situation. Yet, since writing errors are rare, one might argue that even without visual guidance, writing is not affected drastically. According to this view, vision is not a strict prerequisite for handwriting because, apart from an occasional writing error, writing is not altered by the withdrawal of vision. Alternatively, one could argue that the rare occurrence of writing errors without vision indicates that a larger number of errors is prevented. One might infer that having to prevent errors under a no-vision condition should become manifest in writing performance. A typical demonstration of the latter view is provided by Smyth (19891. Without the availability of visual feedback subjects minimized pen lifts during the writing of capital letters. During the production of capital letters, pen lifts consist of visually guided repositioning movements above the writing surface towards an anticipated spatial location on paper. Without visual guidance this type of repositioning is prone to errors. Thus, in order to avoid inaccurate repositioning which results in spatial errors, the use of pen lifts is minimized in the absence of vision which means that normal writing is altered. Besides capital letters, adults often use connected lower-case script which results in cursive handwriting. In contrast to capital letter production, cursive letter production requires hardly any pen lifts. Thus, without vision, repositioning errors due to pen lifts are hardly to
R.R.A. mn Doom, P.J.G. Keuss / The role of rision in handwriting
271
be expected in cursive handwriting. Nevertheless, shape preservation of strokes might still be threatened by the withdrawal of visual guidance, resulting in accuracy variations. However, apart from the observed writing errors, Smyth and Silvers (1987) did not report any changes in letter shapes, which suggests that either writing accuracy remained unaffected by the withdrawal of vision, or that subjects actively prevented letter deformations by employing compensatory strategies. In order to verify the existence of this type of prevention, its measurement is required. In fact, the act of preventing errors might become manifest in writing behaviour only by means of minor adjustments. In this sense, adjustments, and thus alterations of writing, are accepted by subjects for the sake of constant spatial accuracy. The present paper investigates whether cursive writing without vision occurs without a loss of accuracy and whether to this end adjustments are realized. Therefore, cursive letter sequences are studied which are written with and without the availability of visual feedback. The question rises as to how adjustments will become apparent in cursive writing under a no-vision condition. In other words, what kinds of writing features will change? Like in many other studies on motor control, recent handwriting research has mainly been focussing on temporal features (Van Galen et al. 1990). An increase in task demands has been found to become manifest in increased latencies or reaction times CRT), and increased movement times (MT). These measures have been used to study the characteristics of the processes underlying handwriting (see Van Galen, 1991, for a recent overview). The present paper continues this tradition by exploring the temporal aspects (RT and MT) of writing patterns produced with and without visual feedback. Indeed, temporal features have been found to alter when visual feedback is withdrawn. An increase of movement time throughout a writing sequence was found under a condition without visual feedback Wan Galen et al. 1989). The latter finding suggests that processing during the output phase takes more time. Since processes prior to the output phase might also be involved, RT should be included in the analyses to get a more complete picture. In aiming movements, RT and MT have been shown to be negatively correlated. Phillips and Glencross (1985) found that a decrease of MT was accompanied by an increase of RT. This finding suggests that a decrease of processing load during the output phase leads to an increase of processing load
272
R.R.A. am Doom.
P.J. G. Keuss / The role
of r Sm
in hundwriting
prior to the start. A similar mechanism might be involved when handwriting is performed without visual feedback. The increase of MT under a no-vision condition can be thought of as an attempt to enlarge the total processing time. The core question is whether the lengthening of MT has consequences for RT prior to the start of the movement sequence. Two outcomes are feasible regarding RT when visual information is unavailable before and during the writing output phase. (1) By an increase of MT, sufficient processing time is obtained and RT need not change. (2) The increase of MT does not suffice. Total processing time has not reached an appropriate level and thus, an increase of RT will be needed to reach that level. In order to be able to differentiate between these two situations, we will not only withdraw visual feedback but we will also vary the writing duration instructions. In the present study subjects are instructed to write a letter sequence in a normal, comfortable pace, and in a very high pace requiring minimum movement duration. When more processing time is needed under a no-vision condition, an additional increase in RT is expected in the short-duration condition. The assumption is that if more processing time is needed, but this cannot be accomplished by an increase of MT alone, then RT will have to be lengthened as well. Lengthening of latencies indicates a longer processing time. Without vision these longer processing times might be needed for adjustments of the writing act aimed to prevent inaccurate writing. Therefore, to be sure that the lengthening of latencies is aimed at the prevention of spatial effects resulting from the absence of vision, the spatial accuracies of writing under both visual feedback conditions need to be compared. Writing accuracy has received little attention in research. Accurate writing has been related to the invariance of the trajectory shapes. Generally, it is assumed that shapes remain unaffected under a variety of conditions (Maarse et al. 1989; Teulings et al. 1986; Thomassen and Teulings 1988). There are, however, some indications that constant shapes are not always realized. Recently, it was found that writing in different sizes causes horizontal and vertical letter spacing to be resealed in unequal proportions (Wann and Nimmo-Smith 1990). The issue in the present paper is whether shapes remain invariant when
R. R.A. uan Doom, P.J.G. Keuss / The role of vision in handwriting
213
visual feedback is unavailable. It is reasonable to assume that shapes will vary more when the absence of visual feedback is too demanding. In our line of reasoning this will be the case when temporal alterations do not create a sufficient amount of processing time needed to maintain the spatial invariance. The prediction is that spatial accuracy of written letter sequences will not suffer if lenthening of RT and MT suffices to prevent the deteriorating effects of the absence of visual feedback. If, as in the condition of a shorter instructed movement duration, demands are too high to maintain shape invariance, a decline in accuracy should be the result. This should be even more pronounced when visual feedback is absent, and an increase of MT is prevented by a short duration instruction.
Method Subjects Twelve right-handed students from the Free University participated as paid subjects. The selection criterion for taking part was their familiarity with cursive script production. Apparatus and data recording Data were recorded with a ‘Summagraphics Bitpad One’ tablet shaped digitizer, connected with an Olivetti M 280 personal computer. Recording was done with a sampling rate of 95.2 Hz and a spatial reliability of 0.1 mm. The recorded writing patterns were translated into the frequency domain (by means of a forward Fast Fourier Transform, FFT) after which all frequencies above 12.5 Hz were cut off. Subsequently, a backward translation into the spatial domain (backward FFT) was undertaken (Teulings and Maarse 1984). This method resulted in smooth and reliably interpolated writing patterns. Procedure, task and design Subjects were seated in front of a table, in the centre of which the digitizing tablet was positioned. Because the digitizer was slightly tilted, the table was adjusted to obtain a flat writing surface. Subjects were free to use their habitual writing posture. Writing had to be performed on a blank sheet without lineation taped on the tablet. A normal ball-point pen was used. The pen was attached to a flexible wire which did not interfere with normal writing. Subjects wrote the letter sequence lenehele, being a nonsense word in Dutch, under two conditions of Vision (vision and no-vision) and
two conditions of Instructed Duration (normal and short). In the vision condition, subjects were able to look at their writing performance. In the no-vision condition, an opaque box was placed over the tablet in order to preclude perception of the writing hand and stylus, without disturbing the writing movements. In the short Instructed Duration condition subjects had to write the letter sequence within exactly 2.0 seconds. In the normal Instructed Duration condition subjects were permitted to finish the entire sequence within 2.5 seconds. Note that these instructions differ from speed instructions which are generally used in experiments with speed instructions. In the present experiment we specified the durations of the time interval in which the letter sequence had to bc finished in the two conditions. Each subject performed two blocks of trials which differed with respect to Instructed Duration. Each block consisted of two series of 15 trials. A series was related to the availability of Vision which was either present or absent. The reason for nesting the feedback blocks within the Instructed Duration blocks was to get subjects acquainted with one specific Instructed Duration that was fixed throughout both conditions of Vision. Randomization of Instructed Duration across trials would have been confusing and prone to violation of the instruction. Blocks and series were counterbalanced across subjects. Before each Instructed Duration block, the semantically and phonologically simple letter sequence lenehele was presented in printed capital letters on the computer screen approximately 1.0 m in front of the subject. Each trial block started with 20 training trials to get the subject acquainted with writing the letter scquencc according to the instructions A trial started with a low-pitch warning signal, whereupon the subject had to position the stylus on paper. A high-pitch tone signalled the subject to commence as quickly as possible with writing the letter sequence. Subjects were instructed and trained having completed the sequence when a second high-pitch tone was presented, so that writing duration was specified by the time interval between the two high-pitch tones. The letter scquencc consisted, in its cursive form, of 10 upstrokes and 10 downstrokes. Given the estimated minimum stroke duration of 100 ms (Teulings and Maarse 1984) which people generally riced in fast handwriting, the short Instructed Duration condition (2 s) required subjects to write at a maximum speed. The time interval between trials was 2500 ms. The simplicity of the letter sequence ensured that writing errors as they were reported in the study of Smyth and Silvers (1987) were not likely to occur in the present study. Analyses
Reaction Time (RT) was defined as the duration of the interval between the first high-pitch tone (imperative stimulus) and the first increase of the tangential velocity of the writing movements. RT was measured for each trial and averaged over trials per condition. Preliminary checks of the data showed that subjects did not produce any writing errors but that in the more demanding conditions (No-Vision and short Instructed Duration) they were not always able to comply with the Duration Instruction. In 32% of the trials, subjects failed to product the sequence in time, due to which final letters
R.R.A. fun Doom, P.J.G. Keuss / The role of l,ision in handwriting
275
were not recorded. In order to analyze a comparable data set across subjects and conditions, the final three letters of each sequence were excluded from analyses. Movement time, trajectory size and spatial variability of the sequence (consisting of the initial five letters) were determined. Moreover, the same measures were calculated for each of the initial four strokes of the letter sequence, separately. Borders between individual strokes were defined by tangential velocity minima (Lacquaniti et al. 1984). To analyze the spatial variability across replications, each stroke was normalized in size, time and orientation. Spatial variability was determined by the sum of spatial distances of each stroke from the average stroke, obtained via orthogonal projection, across the replications per condition (Maarse et al. 1989). Spatial variability of the trajectory consisting of the ‘initial five letters was aquired by accumulating the spatial variabilities of the initial 14 strokes. Measurements were averaged across 15 replications, which was done for the initial four strokes and across the initial five letters. We were also interested in the variation of movement duration and trajectory size across replications within each condition. For that purpose, we calculated for each of the initial four strokes separately the relative standard deviation of movement time across replications, i.e. the standard deviations divided by the average across replications. A similar procedure was followed for trajectory size. Finally, for each subject, correlations were determined between RT and MT of the first stroke, and between RT and spatial variability of the first stroke. Subsequently, the correlation coefficients were converted into Fisher z-scores and analyzed by means of an ANOVA. Within-subject repeated measurement ANOVA’s on each of the dependent variables were carried out with the individual means of the dependent variable for each condition and subject as cell entries. Dependent variables, measured for the initial four strokes were entered into separate ANOVA’s according to an Instructed Duration (short/normal) X Vision (yes/no> X Letter (l/e) X Stroke (up-/downstroke) factorial design. The dependent variables that were determined across the initial five letters, and the Fisher z-scores, were analyzed according to an Instructed Duration (short/normal) x Vision (yes/no) factorial design.
Results
Correlations
Correlations between RT and MT of the initial stroke varied between -0.30 and 0.80. Correlations between RT and the shape variability of the initial stroke varied between - 0.40 and 0.70. No effects of Vision and Instructed Duration were found on the Fisher z-scores which were derived from the correlation coefficients between RT and MT of the initial stroke and the correlation coefficients between RT and spatial variability of the initial stroke (a > 0.1).
276
R.R.A. van Doom, P.J.G. Keuss / The rvle of vision in handwriting 2oo !??action
Time (ms)
Vision
Fig. 1. Mean reaction time (vertical axis) as a function of Vision (horizontal axis) and Instructed Duration (black bar = normal Instructed Duration, shaded bar = short Instructed Duration).
Reaction time
As a main factor, vision contributed to the observed variation in RT (F(1, 11) = 16.52, p < 0.01). As fig. 1 illustrates, RT was lengthened by 20 ms in the no-vision condition as compared to the vision condition. Instructed Duration had no effect on RT (F(1, 11) = 0.73, ns). However, the RT increment under no vision was larger when the Instructed Duration was short, but the interaction was only marginally significant (F(1, 11) = 4.15, p = 0.066).
Mol~ement lime
Fig. 2 shows the cffccts on MT across the initial five letters as a function of Vision and Instructed Duration. Movement time was identical in both vision conditions (F(1, 1I) = 0.03, ns). Fig. 2 also illustrates that movement time was clearly shorter under the condition of short Instructed Duration, F(1, 11) = 9.27, p < 0.05, indicating that, in general, the subjects obeyed the duration instructions. However, subjects failed to comply to the short Duration Instruction in the no-vision condition, because in this condition movement time was 56 ms longer than allowed (i.e. 1400 ms for the initial 14 strokes that were selected for the analyses). The interaction between Instructed Duration and Vision was significant, F(1, 11) = 6.45, p < 0.05. Variations in MT of the initial four strokes (constituting the initial two letters) as a function of Vision and Instructed Duration arc shown in fig. 3. Movement time was shorter when Instructed Duration was short (F(1, 11) = 366.06, p < 0.001). There was no main effect of vision on movement time (F(1, 11) = 0.29, ns). Fig. 3 shows that movement time decreased with Stroke position, indicated by the significant effects of Letter. F(1, 11) = 138.97, p < 0.001, and Stroke. F(1, 11) = 50.90, p < 0.001). In the first letter (I), the differcncc between the durations of up- and downstrokes was more
277
R.R.A. uan Doom, P.J.G. Keuss / The role of Gsion in handwriting
Movement ,800~--_-__
Time (ms) -~
I
Instructed Duration
Vision Fig. 2. Mean movement time (vertical axis) of the first five letters of the letter sequence lenehele as a function of Vision (horizontal axis) and Instructed Duration (black bar = normal Instructed Duration, shaded bar = short Instructed Duration).
outspoken than in the second letter (e), as indicated by the significant interaction Letter X Stroke, F(1, 11) = 8.18, p < 0.05. The interaction between Instructed Duration and Vision for the initial four strokes was not significant (F(1, 11) = 2.21, p < 0.1) indicating that the Instructed Durations were obeyed well in the initial part of the sequence even when Vision was withdrawn
Movement
Time
(ms) Letter
e Vision
Instructed Duration
175:
I
\ normal .
150
r
normal
" short '\ b
125l
t
short 1
F 100
75
. I1
2
Stroke
1
2
Number
Fig. 3. Mean movement time (vertical axis) of the initial four strokes ing the initial two letters of the sequence lenehele as a function Duration.
(horizontal of Vision
axis) constitutand Instructed
70
Trajectory
Size (mm)
Yes
Vision Fig. 4. Mean trajectory
a function
of Vision
size (vertical (horizontal Duration.
“’
axis) of the first five letters of the letter sequence lenehelr
axis) and Instructed
Duration
shaded bar = short Instructed
(black
bar = normal
as
Instructed
Duration).
and Instructed Duration was short. The difference between short and normal Instructed Duration was larger for the first and second stroke constituting the letter 1, as compared to the first and second stroke of the letter e, as indicated by the interaction Instructed Duration X Letter, F(1, 11) = 12.92, p < 0.001. The relative standard deviation of movement time did not vary as a function of Instructed Duration (F(1. 1I) = 1.68. ns), Vision (F(1, 1 I) = 0.05, ns). Letter (Hl, 11) = 0.05, ns), and Stroke (FCI. 1 I) = 0.41. ns).
Trujectory
size
Variations in the mean Trajectory Size are shown in fig. 4 (across the initial five letters of the writing sequence) and fig. 5 (for the initial four strokes constituting the first two Ictters). Fig. 4 shows a larger mean Trajectory Size of the initial five letters in the no-vision condition than in the vision condition (F(1. 11) = 11.20, p < 0.05). A similar main effect of vision was found for the initial four strokes (F(1. 11) = 8.81, p < 0.05). Trajectory Size of the initial five letters and of the initial four strokes did hot vary as a function of Instructed Duration (RI, 1I) = 0.34, ns), and (F(1, I I) = 0.10, ns). respectively). No interactions between Instructed Duration and Vision on Trajectory Size across the initial five letters (F(1, I I) = 0.42, ns) and across the initial four strokes (F(1, 11) = 0.17. ns) were observed. Fig. 5 illustrates that the effect of Vision was larger in the letter 1 than in the letter e (F(1, I I) = 4.26, p = 0.063). Morcovcr, Vision had a stronger impact on upstrokes than on downstrokes, as indicated by the main effect of Stroke, F(1, 11) = 5.73, p < 0.05, which might be related to the fact that in the present experiment an upstroke is always an initial stroke of a letter. These issues will bc focussed on in the discussion section.
R. R.A. uan Doom, P.J. G. Keun / The role of cision in handwriting Size (mm) ,. Trajectory ~_~ ~ __~__ -~~~ __ Letter
1 Letter
J
279
,
~~
I
e
654 3
Fig. 5. Mean trajectory Size (vertical axis) of the initial four strokes (horizontal axis) constituting the initial two letters of the sequence lenehele as a function of Vision and Instructed Duration.
The relative standard deviation experimental variables (p > 0.2).
of Trajectory
Size did not vary as a function
of the
Spatial uariabili& Across the initial five letters of the sequence, spatial variability decreased when subjects had to perform the task according to the short Instructed Duration (see fig. 6), which was statistically confirmed (F(1, 11) = 16.45, p < 0.05). However, in the
Spatial Variability 507
~
~~~~
---1
-~
45-
I
Instructed Duration
no
Yes
Vision Fig. 6. Mean spatial variability measures (vertical axis: depicted in arbitrary units resulting from normalisation in time, size and orientation) across the initial five letters of the letter sequence lenehele as a function of Vision (horizontal axis) and Instructed Duration (black bar = normal
Instructed Durations shaded bar = short Instructed Duration).
2x0
R.R.A. ~mn Doom.
8 Spatial Letter
P.J.G.
Keuss
/
The role
of ~,ision in handwnting
Variability P
Letter
4
7
Vision t
6
/
I
/
5
Yes
normal
.
“0
normal
Yes
short
.
“0
short
A 4
2
I
1 1
1
Instructed Duration
I 2
Stroke
1
2
Number
Fig. 7. Mean spatial variability (vertical axis: depicted in arbitrary units resulting from normalisation in time, size and orientation) of the initial four strokes (horizontal axis) constituting the initial two letters of the sequence lerzehele as a function of Vision and Instructed Duration.
initial four strokes (see fig. 7) this effect was not found (F(1, 11) = 0.07, ns). Fig. 6 suggests a slight increase of spatial variability, measured across the initial five letters, in the no-vision condition as compared to the vision condition. This effect seemed most pronounced in the normal Instructed Duration condition. Fig. 7 does not show such increments for the initial four strokes. Statistical tests revealed, however, that neither the main effects of vision on spatial variability measured across the initial five letters (F(1, 11) = 0.38, ns) and across the initial four strokes (F(1, 11) = 0.07, ns), nor any interaction between Vision and Instructed Duration across the initial five letters (F(1, 11) = 1.17, ns) and across the initial four strokes (F(1, 11) = 0.01, ns) became significant. The spatial variability of the letter e was larger than of the letter I (F(1, 11) = 9.42, p < 0.05).Also, across the initial four strokes, shapes of downstrokes were more variable than of the shapes of upstrokes, as indicated by the main effect of Stroke (F(1, 11) = 12.24, p < 0.01). This was more outspoken in the letter e, as indicated by the significant interaction Letter X Stroke, F(1, 11) = 8.33, p < 0.05.
Discussion The present experiment confirmed the impact that vision has on latencies of letter production in cursive handwriting. Without visual feedback more time is needed to commence the writing sequence. In the spatial domain, trajectories become larger under the no-vision condition, while spatial variability across replications remains unaf-
R.R.A. uan Doom, P.J.G. Keuss / The role of Laisionin handwriting
281
fected. What do these results mean for a psychomotor theory of handwriting? According to the often proposed explanation concerning the invariance of shapes, spatial variability should not differ as a function of vision (Maarse et al. 1989; Thomassen and Teulings 1988). We argued that under a demanding condition like no vision, strategies may be employed to ensure accurate shape reproduction. Overt manifestations of strategies should become apparent in the temporal domain of the act of writing. A possible strategy is the increase of processing time. In the present paper, the relation between RT and MT was considered in this context. The idea is that the total processing time needed to perform a motor task is distributed over RT and MT (Phillips and Glencross 1985). Moreover, RT and MT are believed to be linked, which means that a strong decrease of one results in an increase of the other variable so that the total processing time remains constant. Also, under demanding conditions more processing time might be needed leading to increments of the latencies. Writing without vision might be such a demanding condition. Indeed, the finding that without vision the duration of writing movements increased (Van Galen et al. 1989) suggests an increment in processing time during the output. Our prediction that RT should increase when the MT increase is not enough to fulfill the total processing time requirement, could be confirmed, although it was reflected by only a marginally significant interaction between Instructed Duration and Vision on RT (p = 0.066). Nevertheless, the identified increase of RT suggests that the adjusted processing time during the output phase (MT) did not suffice to cope with the absence of vision and that prior to the movement output additional processing was required, The finding of Phillips and Glencross (1985) that in simple aimed movements RT negatively correlates with MT could not be confirmed for writing sequences in the present experiment. In our data, RT and MT were not negatively correlated. However, in the vision condition, a short Instructed Duration resulted in a shorter RT. The present results are not at odds with findings in studies on movement sequences (Hulstijn and Van Galen 1983; 1988; Klapp and Wyatt 1976; Sternberg et al. 19781, in which positive correlations between MT and RT were found. Apparently, due to the more complex nature of the involved movement sequence, being a cursive letter sequence in the present experiment, a shorter movement duration made an earlier
282
R.R.A. c’an Doom, P.J.G. Keum / The role of Lision in handwriting
movement start possible. We may add that the positive relation between MT and RT in movement sequences only holds as long as vision is available. In fact, we could confirm the prediction that the RT increment due to no vision is more pronounced when short movement duration was instructed. The apparent ease with which writers start a sequence under a vision condition when a short duration is expected, vanishes when vision is precluded. Our interpretation of this finding is that additional processing time is needed under the no vision condition. Since the increase of processing time is not possible during movement output in the short Instructed Duration condition, the processing time increment is realized prior to the movement start. We argued that the assumption of a prolonged processing time as a reaction on increased demands in writing, is only valid when the main objective of the writing performance, namely shape consistency, is achieved. Although no correlation was found between RT and shape variability of the initial stroke, the results show that withdrawing visual feedback did not affect the spatial variability of trajectory shapes. Also, in a condition in which the increase of MT was prevented by a short Duration Instruction, writing without visual feedback did not affect spatial variability. To the contrary, the results suggest that subjects try to suppress the variability of shapes under demanding conditions. Under a short Instructed Duration condition spatial variability of the trajectory comprising the initial five letters decreased under both Vision conditions. Still, although in general, subjects proved to be able to cope with the time contraints imposed on their performance; under combined conditions without visual feedback and short writing duration, they were unable to comply with the Instruction Duration (cf. the significant interaction Vision X Instructed Duration on movement time). Apparently, in order to preserve the consistency of shapes they had to violate the Instructed Duration and preserve a minimum of processing time. Time-pressure effects of short Instructed Duration, being the violation of the short Duration Instruction and decrements of spatial variability with short Instructed Duration, were not detectable within the initial four strokes of the sequence, but became manifest only after a larger part of the sequence was achieved. These results might suggest that the shapes of the letters were not all equally vulnerable to external demands, which would explain that in our data only the
R.R.A. mn Doom, P.J.G. Keuss / The role of c&ion in handwriting
283
summated values across different letters were affected. Alternatively, initial strokes of a sequence might be exceptional in that time pressure only had an effect on strokes of subsequent letters. It is possible that initial strokes were affected by starting the sequence, as indicated by their rather long durations (see fig. 3) under the short Instructed Duration condition. It is unclear, whether these start-up effects were related to biomechanical factors or whether programming factors (Portier et al. 1990) were involved. The initial four strokes were dissimilar regarding trajectory size (fig. 5) and spatial variability (fig. 7). The former measure decreased with stroke number, whereas the latter measure increased when writing progressed from the initial stroke to the fourth stroke. Whether these effects were related to stroke positions or whether they might be attributed to letter type differences (1 vs. e) is already being focussed on in a follow-up study. The present results suggest that a strict invariance of shapes does not exist in writing. Since a certain, but limited, spatial variability exists, one might argue that writing shapes are allowed to vary within a tolerance range, which as such constrains shape variability. In this view, writing can be viewed as a rather flexible skill. Certain changes in the temporal domain of writing are allowed as long as (or perhaps so that) this tolerance region is not exceeded. As a consequence, shapes will remain relatively consistent under a variety of conditions (Maarse et al. 1989; Teulings et al. 1986). Note that consistency of shapes was presently determined for repeated writing within each condition separately. Although in the no vision condition shapes were equally consistent as in the vision condition, it is still possible that shapes changed in an absolute sense. The finding of Wann and Nimmo-Smith (1990) that horizontal and vertical sizes change in unequal proportions when larger writing is required is therefore not at odds with the present data. Writing larger than normal might affect horizontal and vertical sizes differently without a decrease of shape consistency. We are presently investigating this idea in follow-up research. However, in the present data, a change of shapes is suggested by the increase of trajectory size in the no-vision condition. Note that with a constant spatial variability, the absolute change of shapes could certainly not result from a loss of control, which is also reflected by the constant (relative) standard deviation of trajectory size in vision and no vision conditions. The question arises to what end trajectory size increases when
284
R.R.A. can Doom, P.J.G. Kmss / The role of vision in handwriting
visual guidance is not possible. It might be that our subjects wrote sloppily, because they were unfamiliar with the no-vision condition. Still, larger trajectories could not have been resulting from a decline of spatial control, as Spatial Variability did not increase in no vision. It might be reasonable to argue that sizes were increased with a specific purpose. Although our subjects were not aware of their enlarged script when vision of their ongoing performance was precluded, they said they relied heavily on what they felt. Since vision was precluded, a fair assumption is that kinesthetic information was used to evaluate shapes. The kinesthetic information mode as compared to the visual mode might use a different shape criterion which could explain why shapes in an absolute sense seem to alter. This is, however, no compelling reason why trajectory sizes should grow. A probable explanation is substantiated by a finding provided by physiological research that finger movements increase in sensitivity when the excursions are made larger (Clark and Horch 1986). In writing it might well be that, in order to employ kinesthetic feedback adequately in absence of vision, subjects increase their writing size. According to this idea an equal amount of spatial information should become available across a larger trajectory. The larger distance might enhance the spatial resolution of the involved kinesthetic receptors, without the requirement of more processing time. In summary, this study has shown that writing under conditions of absence of vision and short Instructed Duration exerts certain increases in latency and movement time to preserve a certain amount of processing time so that trajectory shape output remains consistent. Moreover, handwriting, although a very complex skill, can be performed with a remarkable flexibility. When demands become too heavy, duration instructions are simply discarded in order to fulfill a far more important goal, relative consistency of shape output. Although adult writing undergoes specific changes, without vision, control is not impaired. Therefore, we concur with Smyth and Silvers (1987) that vision is not a strict prerequisite for movement control and the consistent production of shapes in handwriting.
References Bootsma, R.J. and P.C.W. van Wieringen. 1990. Timing and attacking forehand drive in table tennis. Journal of Experimental Psychology, Human Perception and Performance 16. 21-29.
R.R.A. L’WIDoom, P.J.G. Keuss / The role of vision in handwriting
285
Clark, F.J. and K.W. Horch, 1986. ‘Kinesthesia’. In: K.B. Boff, L. Kaufman and J.B. Thomas (eds.), Handbook of perception and human performance, Vol. 1: Sensory processes and performance. London: Wiley. Hulstijn, W. and G.P. van Galen, 1983. Programming in handwriting: Reaction time and movement time as a function of sequence length. Acta Psychologica. 54, 23-49. Hulstijn, W. and G.P. Van Galen, 1988. ‘Levels of motor programming in writing familiar and unfamiliar symbols’. In: A.M. Colley and J.R. Beech (eds.), Cognition and action in skilled behaviour. Amsterdam: Elsevier. Keele, S.W., 1986. ‘Motor Control’. In: K.R. Boff, L. Kaufman and J.P. Thomas (eds.), Handbook of perception and human performance, Vol. II: Cognitive processes and performance London: Wiley. Klapp, S.T. and E.P. Wyatt, 1976. Motor programming within a sequence of responses. Journal of Motor Behavior 8, 19-26. Lacquaniti, F., C. Terzuolo and P. Viviani, 1984. ‘Global metric properties and preparatory processes in drawing movements’. In: S. Kornblum and J. Requin (eds.), Preparatory states and processes: Proceedings of the France-American Conference, MI, Aug. 1982. Hillsdale, NJ: Erlbaum. pp. 357-370. Maarse, F.J., G.P. van Galen and A.J.W.M. Thomassen, 1989. Models for the generation of writing units in handwriting under variation of size, slant, and orientation. Human Movement Science 8, 271-288. Phillips, J. and D. Glencross, 1985. The independence of reaction and movement time in programmed movements. Acta Psychologica, 60, 209-225. Portier, S.J., G.P. van Galen and R.G.J. Meulenbroek, 1990. Practice and the dynamics of handwriting performance: Evidence for shift of motor programming load. Journal of Motor Behavior 22, 474-492. Savelsbergh, G.J.P., H.T.A. Whiting and R.J. Bootsma, 1991. Grasping tau. Journal of Experimental Psychology: Human Perception and Performance 17, 315-322. Schmidt, R.A., 1988. Motor control and learning: A behavioral emphasis. Champaign, IL: Human Kinetics. Smyth, M.M., 1989. Visual control of movement patterns and the grammar of action. Acta Psychologica 70, 253-265. Smyth, M.M. and J. Silvers, 1987. Functions of vision in the control of handwriting. Acta Psychologica 65, 65-73. Sternberg, S., S. Monsell, R.L. Knoll and C.E. Wright, 1978. ‘The latency and duration or rapid movement sequences: Comparisons of speech and typewriting’. In: G.E. Stelmach (ed.), Information processing in motor control and learning. London: Academic Press. Teulings, H-L. and F.J. Maarse, 1984. Digital recording and processing of handwriting movements. Human Movement Science 3, 193-197. Teulings, H-L., A.J.W.M. Thomassen and G.P. van Galen, 1986. “Invariants in handwriting: The information contained in a motor program’. In: H.S.R. Kao, G.P. van Galen and R. Hoosain (eds.), Graphonomics: Contemporary research in handwriting. Amsterdam: Elsevier. Thomassen, A.J.W.M. and H.-L. Teulings, 1988. ‘Grafische productie: De motoriek van schrijven en tekenen’. In: P.J.G. Keuss, G. ten Hoopen and A.J.J. Mannearts (eds.), Menselijke motoriek. Amsterdam: Swetz & Zeitlinger. Van Galen, G.P., 1990. Effects of phonological and motoric demands in handwriting: Evidence for discrete transmission of information. Acta Psychologica 74, 259-277. Van Galen, G.P, 1991. Handwriting: Issues for a psychomotor theory. Human Movement Science 10, 165-191. Van Galen, G.P., M.M. Smyth, R.G.J. Meulenbroek and H. Hylkema, 1989. ‘The role of short term memory and the motor buffer in handwriting under visual and non-visual guidance’. In:
286
R.R.A. can Doom.
P.J. G. Keuss / The role
of vision in handwriting
R. Plamondon, C.Y. Suen and M.L. Simner (eds.), Computer recognition and human production of handwriting. Singapore: World Scientific. Van Galen, G.P., R.R.A. van Doorn and L.B.J. Schomaker, 1990. Effects of motor programming on the power spectral density function of finger and wrist movements. Journal of Experimental Psychology: Human Perception and Performance 16, 755-765. Warm, J.P. and I. Nimmo-Smith, 1990. Evidence against the relative invariance of timing in handwriting, The Quarterly Journal of Experimental Psychology 42A, 105-119.
