Acta Psychologica46 (1980) 141-151 0 North-Holland Publishing Company

THE HEIGHT

OF HANDWRITING

*

Alan M. WING Medical Research Council Applied Psychology

Accepted

November

Unit, Cambridge,

U.K.

1979

An experiment is described that examines the basis of the control of vertical amplitude (height) of handwriting. It is shown that changes in movement duration accompany changes in height whether the latter are associated with instructions to write larger than normal (between words) or because the letters to be written (e, 0 differ in height (within words). The proportional differences in movement duration match the proportional differences in amplitude between words but not within words. It is concluded that two different time-based mechanisms are invoked for handwriting height control: interval adjustment for writing size changes between words and durafion selection for letters differing in height within a word.

In cursive (joined-up) handwriting, vertical extent can be an important distinguishing feature between different letters (Frishkopf and Harmon 196 1). A particularly clear example is provided by the letters e, 1, which in the writing of many people only differ in the relative lengthening of the vertical movement in 1. Other examples include the handwritten forms of(g, a, 4 (p, fl, h) (y, u). As well as being able to control the amplitude of vertical movement to give letters distinct forms, writers may also make overall adjustments to the size of their writing. For example, when people use wide-line paper or if they want their handwriting to be more legible they tend to write larger. Yet the style of their writing is not noticeably changed. Given these two forms of adjustment of vertical extent (height) do not interfere it may mean the way letter height is controlled to give distinct letter forms is different from the mechanism for adjusting height of all letters to change the overall size of writing. From an examination of acceleration wave forms associated with * Requests for reprints may be sent to Alan Wing, Medical ogy Unit, 15 Chaucer Road, Cambridge CB2 2EF, U.K.

141

Research

Council

Applied

Psychol-

142

A. Wing / The height of handwriting

handwriting movements Denier van der Gon and Thuring (1965) proposed two distinct mechanisms for changes in the height of writing. Differences in letter height within a word should be attributed to changes in the durutiorz of application of forces fixed in mzgnitde. In contrast they suggested that overall changes in height are due to changes in force mugrzitztde with application drrrutiotl unchanged. In support of this claim they presented a very limited set of data that showed: (a) with the alternating sequence elel . . the upstroke of the 1 was 11 1% bigger than the upstroke in an adjacent e and took 15%’ longer to form: (b) over different series of repeated Is, increases in upstroke of 136%’ and 6 15% gave no increase in duration. While the picture painted of a contrast between changes in vertical extent scaled by force duration and force magnitude is appealing a number of objections may be raised. The data of only one subject were presented by Denier van der Gon and Thuring and there was no indication of the statistical reliability of the results. The material to be written comprised sequences of connected anticlockwise loops. Because of the repetitive nature of the material and the fact it was not embedded with other letters it is possible the results do not apply to normal writing. The generality of the results may further be queried because the experiment on alteration of overall writing height included in each trial loops of only one height and because the overall size changes were much greater than people would normally make. More recently Yasuhara (I 975) came to the same conclusion as Denier van der Gon and Thuring about the nature of height control in handwriting. Unfortunately the data he presented are also unsatisfactory. In one experiment he examined the production of the letters e, 1 in the context of three different four-letter words, but he provided samples of data from only three trials and there was no statistical presentation of the full results. In the other experiment he looked at overall height changes in a normal range but on each trial the material was restricted to loops of only one height. A pilot experiment on duration changes associated with writing meaningful words in small or large writing was briefly described in Wing (1978). Two subjects were run and reliable changes in duration were found as a function of the overall height of handwriting. This finding, contrary to Denier van der Gon and Thuring’s account of height changes between words, motivated the experiment to be described in the present paper. The experiment tests the Denier van der Gon and

A. Wing / The height of handwriting

143

Thuring hypothesis by an examination of differences in movement duration associated with the letters e, E written in word contexts by a sufficient number of subjects to give some confidence in the genkality of the results.

The experiment Method Subjects Twenty-four Ss were drawn from the APU subject panel. Each S was paid for his or her services and attended one session lasting approximately 40 mm. There were 18 female and 6 male Ss. The average age was 34 years, ranging from 22 up to 55 years. Ss were assigned to two equal groups approximately matched for age and sex according to whether they were left- or right-handed for writing. Task Each S copied, one word at a time, a list containing 28 words. Dispersed through the list were the words elect, elegy, eleven, element, once to be written normal size and once to be written “About fifty per cent larger than normal”. The remaining 20 items included the S’s own name (4 times) and a set of words beginning with the letters v, n, w or m. The structure of the list was not discussed with Ss except to point out at the appropriate time when a word was to be written larger than normal. Ss were asked to adopt a normal writing speed. They were informed that the experimenter did not expect careful formation of individual letters but that the writing should nevertheless be legible. With time for instruction and recording of data this task took about 10 min. The rest of the session was taken up with filling out a handedness preference questionnaire and carrying out other, unrelated performance tests. Procedure The word list was printed in two columns on A4 paper (295 X 210 mm) with vertical spacing of 10 mm. Alongside each word was a line 68 mm long for the S’s handwritten version. Writing took place with the paper superimposed on the 250 X 250 mm writing bed of a Computek CT50 x-y digitiser. A soft-lead propelling pencil supplied with the digitiser was used. This pencil, which had a barrel of diameter 13 mm, was connected to the digitiser by a thin wire from the end opposite the writing tip. As well as providing position coordinates of the tip accurate to 0.36 mm the pencil incorporated a micro-switch to detect contact with the writing surface. The digitiser was interfaced to a CTL Modular One computer. Before writing each word in the list, the S read it out aloud to the experimenter. Computer sampling of pencil position at 4 msec intervals started as soon as the S initiated movement of the pencil in contact with the paper. Sampling continued whether or not the pencil was in contact with the paper for 2000 msec. Although

144

A. Wing / The height of handwriting

writing longer words exceeded this length of time, subjects were not informed of the termination of sampling but wrote each word out in its entirety. On completion of sampling, the data were displayed as an -u-y plot to the experimenter, but not to the S. The experimenter then depressed a console key to accept and store the data as a record on magnetic disc. Sometimes Ss failed to apply sufficient pressure when first starting to write and thus caused sampling to be delayed until part way through the word. On these occasions the S was asked to rewrite the whole word before proceeding to the next trial. Analysis

and results

The handwriting data were processed after all testing was completed. Processing involved three stages. The disc record with a desired item was accessed and the data displayed as a magnified x/y plot. The y-coordinates were then plotted as a function of time. Finally a subroutine determined the coordinates and times at which successive y-maxima and minima were attained. (If the curve at a maximum or min-

Fig. 1. Resolution of a sample of handwriting into horizontal and vertical components of displacement as a function of time. Horizontal displacement (20 units = 5.6 mm/division) as a function of time (200 msec/division) is shown top left with the corresponding graph of vertical displacement (10 units = 2.8 mm/division) shown top right. The enlargement of the first three letters (lower left) shows successive minima and maxima corresponding to those shown in the scaled-up graph of vertical displacement versus time (lower right).

145

A. Wing / The height of handwriting

imum was flat over several time units, the time was taken as being that of first arrival.) These steps are illustrated in fig. 1 taken directly from the computer visual display. At the top one sample of the word elegy as written by one of the Ss is shown. Immediately underneath are the x- and y-coordinates plotted separately as a function of time (200 msec/division). Note the vertical axis scales (10 units are equivalent to 2.8 mm) are different for x and y. At the bottom of the figure on the left is a magnified plot of the first 3 letters segmented to show they-axis peaks. A magnified y versus time plot on the right is shown similarly segmented. Of the 24 Ss, one left-handed and one right-handed subject used the greek E and their data were excluded from the present analysis. The data of one further (righthanded) S had to be excluded because her writing was not fluent; almost every letter contained a pen lift or hesitation. Despite checks made during data acquisition, detailed examination of the data revealed variability from word-to-word in the point at which sampling had begun relative to the first trace of pencil on the paper. Furthermore individual differences were observed in the direction of the initial stroke of the first e. The initial e was therefore dropped from the analysis and attention was restricted to the second and third letters 1, e. The displacement and duration data averaged over the 4 items and the 21 Ss are shown in figs. 2 and 3 for the up and down movements comprising the 1 followed by the e. The overall average heights of I and e were 6.1 mm and 3.4 mm respec-

w

K

Yp

DOWN

01

1

I

UP

DOWN

Fig. 2. Effect of overall size of writing on amplitude of up and down handwriting movements in I followed by e averaged over 4 items and 21 subjects. Two standard errors of differences of means shown for pairwise comparison of; (1) movement means for same overall size (continuous vertical

line), (2) overall size means for same movement

(broken

vertical

line).

A. Wing / The height of handwriting

146

OL

UP

e

DOWN

UP

DOWN

Fig. 3. Effect of overall size of writing on duration of UP and down handwriting movements over 4 items and 21 subjects. Two standard errors of differences

I followed by e averaged means as fig. 2.

in of

tively. Their average durations (upward and downward movements summed) were 332.4 msecs and 259.4 msecs respectively. In response to the instruction to write larger than normal there was an average increase of 27% in letter height. Although a sizeable increase, it is somewhat less than the 50% increase requested by the experimenter. This is probably due to the rather small vertical line spacing on the paper and to the fact that no feedback was given to the Ss about the size increase they produced. The letter I was on average written 82% larger than e. A four-way analysis of variance was performed with movement (4 levels), size (2 levels), item (4 levels) and subject group (4 levels taking account of hrindedness and sex) as factors. Significant main effects of size (t;(l, 17) = 64.90, p