Computerised assessment of handwriting

and typing speed

Joanna Horne, Jonathan Ferrier, Chris Singleton &
Caroline Read

This paper reports on two studies using computer-based dictation tasks for measuring speed of typing and
handwriting. In the first study, 952 students aged 11 to 17 years attending 19 different secondary schools
hand wrote and typed passages dictated by a computer. For both handwriting and typing, a very high
correlation was found between speed calculated by the computer and that calculated by a human assessor,
establishing that computerised calculation is a reliable as well as convenient and time-saving method of
establishing writing speed. There were greater age-related gains in speed of typing compared with
handwriting and greater variation in typing skill than handwriting skill. However, almost half of students
with slow handwriting (below standard score 85) were found to have average or better typing speeds.
In the second study, 55 students aged 13 to 14 years were administered these tasks together with the
Hedderley Sentence Completion Test of handwriting speed. Despite the clear differences between the two test
formats, a significant moderate level of correlation was found between them (r=0.54). Almost one-third of
students with slow handwriting in the computer-based task had not previously been identified as having
support needs but would potentially be disadvantaged in written examinations. By eliminating the
‘thinking’ time involved in free writing, computerised dictation tasks give ‘purer’ measures which can reveal
physical handwriting and/or typing problems. They also simulate examination requirements more closely
than mechanical repetitive tests of writing speed, and should be particularly helpful in establishing whether
students need access arrangements in examinations.
Keywords: assessment; handwriting speed; typing speed; access arrangements; computers.

ESPITE the widespread use of 2005; Tseng & Cermak, 1993; Tseng & Chow,

D computers in schools, the majority of

students still write by hand rather than
using a keyboard. Children spend 31 to 60
2000). All these factors contribute to the
speed as well as the legibility of handwriting
(Volman, van Schendel & Jongmans, 2006),
per cent of their school day engaged in and both components have been shown to
handwriting tasks (McHale & Cermak, 1992) have a wider impact on educational attain-
and consequently handwriting ability ment throughout schooling and on into
remains an important factor in learning and higher education (Berninger & Graham,
assessment. 1998; Berninger, Mizoikowa & Bragg, 1991;
Handwriting is a complex perceptual- Connelly et al., 2006; Dockrell & Barnett,
motor skill that is vulnerable to a wide variety 2005; Graham, Harris & Fink, 2000; Jones &
of dysfunctions, including weak fine motor Christensen, 1999; Roaf, 1998). Prevalence
control, poor sensory awareness of the of handwriting difficulties varies from five
fingers, problems with timing of movements, per cent to 27 per cent of school-age
difficulties in integrating visual and kinaes- children, depending on age and method of
thetic information, inadequate pen grip and assessment used (Alston, 1985; Hamstra-
limitations of attention (Cornhill & Case- Bletz & Blöte, 1993; Karlsdottir & Stefansson,
Smith, 1996; Exner, 1989; Feder & 2002; Maeland, 1992; Smits-Engelsman &
Majnemer, 2007; Karlsdottir & Stefansson, Van Galen, 1997). However, a deficit in the
2002; Maeland, 1992; Malloy-Miller, integration of visual and motor information
Polataijko & Ansett, 1995; Schoemaker et al., (commonly associated with dyslexic and

52 Educational & Child Psychology Vol. 28 No. 2

© The British Psychological Society, 2011
 Computerised assessment of handwriting and typing speed

dyspraxic-type difficulties) seems to be the words per minute (wpm) progresses from
most significant underlying aetiological 13.88 (age 8 to 9) to 34.19 (age 17 to 18).
factor for those with poor handwriting These rates are somewhat faster than those
(Volman et al., 2006; Williams et al., 1993). reported by Phelps et al. (1985), but copying
Assessing handwriting skills is a contro- the same sentence over and over again is a
versial field. Rosenblum, Weiss and Parush considerably easier task than copying a story,
(2003) reviewed a wide variety of different in which reading imposes an additional
approaches and concluded that all have cognitive load. In fact, the growth lines from
significant drawbacks. Graham (1986a, the two studies are almost parallel,
1986b) has also critically examined various suggesting the same developmental pro-
instruments. Hence establishing how fast cesses are at work. Graham et al. (1998) have
students can write is not a trivial matter. In a also observed that the relatively slower
variety of studies, handwriting speed has speeds obtained by Phelps et al. (1985) were
been found to increase steadily during the likely to have been due in part to the inclu-
school years, with most studies reporting a sion of large numbers of students with
levelling off at about age 13 to 14 (Connelly, special needs in their study, whereas the
Gee & Walsh, 2007; Graham et al., 1998). study by Wallen et al. (1996) used an unse-
Girls tend to be faster writers than boys lected sample.
throughout childhood (Alston, 1995; Bishop Typing skills also increase steadily with
& Estgate, 2001; Zivani, 1984) and right- age, but the speed of typing tends to lag
handers faster than left-handers (Graham et behind that of handwriting throughout
al., 1998). However, regardless of the mode primary school. Connelly et al. (2007) meas-
of production, speed of writing is largely a ured the handwriting and typing speeds of
function of task demands and content, and 312 UK children aged 4 to 11 years, using the
different approaches to assessment may yield task devised by Wallen et al. (1996). From
very different speeds. Free writing tasks will age 7 onwards, handwriting speed increase
produce different speeds to copying or was found to be near linear and broadly
repetitive writing tasks. The nature of the consistent with the findings of Wallen et al.,
sample will also affect the findings. Two from 33 letters per minute (lpm) (8.48
examples of large-scale studies will suffice to wpm) at age 7 to 65 lpm (16.71 wpm) at age
illustrate these issues. 11. When typing the same task, performance
Phelps, Stempel and Speck (1985) also increased linearly with age, but speeds
required children to copy a story of 197 were considerably below those of hand-
words, with the first two minutes of writing writing: 28 lpm (7.20 wpm) at age 7 to 46
being scored. Results from 1365 US children lpm (11.23 wpm) at age 11. A significant
showed a near linear growth, from 25 letters correlation between the two modes of text
per minute at age 8 to 9 to 72 letters per reproduction was found (r=0.70; p<0.001).
minute at age 13 to 14. Wallen, Bonney and Unfortunately, there is a dearth of studies
Lennox (1996) created a simple speed of of typing speed in the secondary school years
writing test that required children to copy and comparisons of handwriting and typing
the sentence ‘The quick brown fox jumps skills during childhood are also rare in the
over the lazy dog’ as many times as possible research literature. Most studies in this field
in two minutes. Results from 1292 Australian have used compositional writing rather than
children revealed an almost linear growth copying or writing to dictation. Christensen
from 54 letters per minute at age 8 to 9 to (2004) gave two 20-minute compositional
133 letters per minute at age 17 to 18. Since writing tasks (topics specified) to 276
the sentence contains nine words and a total students aged 12 to 15 (average age 13 years
of 35 letters (excluding spaces) there are 3 months), one handwritten and the other
3.89 letters per word and thus writing rate in typed into a computer. Significant relation-

Educational & Child Psychology Vol. 28 No. 2 53

Joanna Horne, Jonathan Ferrier, Chris Singleton & Caroline Read

ships were found between the two types of Assessment of handwriting speed is neces-
text production (r=0.51; p<0.001); this corre- sary in order to apply for examination access
lation was somewhat lower than that arrangements in the UK and many other
reported by Connelly et al. (2007) although countries. Students who have significant diffi-
the children were older than in the latter culties in handwriting will usually be entitled
study. A measure of orthographic-motor to access arrangements such as additional
integration involving writing the alphabet in time, use of an amanuensis (scribe) or a word
lower- and upper-case as quickly and as many processor (Backhouse, Dolman & Read,
times as possible in one minute (Berninger 2007). Of all the students who were granted
et al., 1991) was also found to be significantly access arrangements in the 2009–10 GSCE
related to the amount of text handwritten exams, 44 per cent were given extra time, 11
(r=0.30; p<0.001) and typed (r=0.55; per cent used a scribe, and seven per cent used
p<0.001). Orthographic-motor integration a word processor (Ofqual, 2010). Although a
was found to account for 28 per cent of the few of these students will be granted use of a
variance in quality and speed of handwriting, scribe, for the vast majority of them examina-
and 30 per cent of the variance in quality tions will still entail writing by hand or word
and speed of typed composition. processing (Ofqual, 2010). Thus, even for
Rogers and Case-Smith (2002) examined these students, speed of written production
the relationships between handwriting speed usually remains critical. Over the past 30 or
and the speed of typing in students aged 11 more years professional opinions have varied
to 12. These students (N=40), all of whom regarding the most appropriate method to
had already participated in a school adopt in order to assess handwriting speed
keyboarding class, were asked to copy a (see Ashton, 1997; Bishop & Esgate, 2001;
familiar poem by hand and also when using Bonney, 1992; Hedderly, 1992, 1995, 1996;
a keyboard. Typing and handwriting speed Sawyer, Francis & Knight, 1992; Sawyer, Gray &
were found to correlate significantly (r=0.36; Champness, 1996; Turner, 1997). Suggested
p=0.026); this correlation was lower than methods range from the purely mechanical
those reported by Christensen (2004) and (e.g. writing a series of letters or symbols as fast
Connelly et al. (2007), but in the Rogers and as possible in a short time), through copying
Case-Smith study the students had received exercises and free writing tasks, to complex
typing tuition, which was not the case in the diagnostic scales. Writing speed varies
other two studies. Handwriting speed was according to the nature of the task, and it is
found to account for 12 per cent of the vari- debatable what approach is most appropriate
ance in typing speed. Seventy per cent of the for determining whether or not a student
sample produced more text using a should be entitled to additional time in exam-
keyboard than when writing by hand, and 30 inations. The Hedderly Sentence Completion
per cent of those who were the slowest hand Test (Hedderly, 1995) is among the most
writers had relatively fast typing speeds. The popularly used instruments. This requires the
relative speeds of handwriting and typing student to complete 40 sentences (e.g. ‘I
have also been found to affect the quality of like…’, ‘The happiest time…’) by writing the
narrative writing. Dunn and Reay (1989) first thing that comes into their head,
studied 52 students aged 12 to 13 and found provided it makes a complete sentence; the
that those whose typing speed equalled or task is timed and the speed of handwriting
exceeded that of their handwriting speed calculated. Norms in wpm are provided from
produced better quality compositions when ages nine (average eight wpm) to 18 (average
using a keyboard; those whose handwriting 16.9 wpm). Bishop and Esgate (2001) pointed
speed equalled or exceeded that of their out that short tests may not be good predictors
typing speed produced better quality hand- of writing speed in longer examination condi-
written compositions. tions, a conclusion that receives strong

54 Educational & Child Psychology Vol. 28 No. 2

 Computerised assessment of handwriting and typing speed

support from findings by Summers and wpm is 40 per cent slower than average
Catarro (2003) and O’Mahony, Dempsey and (which she indicates warrants use of a
Killeen (2008). Free writing tests tend to be scribe). In the absence of a serious alterna-
longer, for example, the test devised by tive, these apparently arbitrary ‘25 per cent
Allcock (2001) employs a 20-minute free slower’ and ‘40 per cent slower’ rates have
writing task on a topic to be chosen by the become widely applied.
student, with two minutes planning time The two studies presented in this paper
allowed. Norms in wpm are provided from investigated the use of a computer-based
ages 11 (average 13.9 wpm; compared with 12 dictation task for measurement of speed of
wpm for Hedderly’s test) to 16 (average 16.9 typing and handwriting. Dictation was
wpm; compared with 20 wpm for Hedderly’s chosen rather than free writing in order to
test). However, free writing tests are highly standardise the task, to remove (as far as
dependent on topic and may not accurately practicable) the differential effects of
reflect writing speed: the student may not be content and topic knowledge on speed, and
writing continuously during that time because to eliminate ‘thinking’ time, which is a
they are thinking about what to write next, or confound in free writing tasks. Repetitive
pondering how to spell words they wish to use, copying of words or sentences was rejected
or trying to think of synonyms (Ashton, 1997). in favour of extended passages of mean-
A further problem is where to set the cut- ingful text because the former was consid-
off below which a student is judged to have ered too far removed from the normal
sufficiently slow writing to warrant additional process of writing. Straightforward copying
time in examinations. Using a 20-minute of passages was rejected because writing
free writing test, Roaf (1998) selected 15 performance is confounded with reading
wpm as the cut-off speed below which and spelling ability. Computerised assess-
students were judged to be ‘slow’ writers; 25 ment of writing speed was chosen mainly
per cent of an unselected sample of 1273 because of the time, cost and efficiency
students aged 11 to 16 fell into this category, savings that it affords (McDonald, 2002;
with 4.5 per cent falling below 10 wpm. The Singleton, 2001) and because of evidence of
current regulations of the Joint Council for less gender bias compared with conventional
Qualifications (JCQ), which refer to public assessment methods (e.g. Horne, 2007,
examinations in England, Wales and compared computerised assessments of
Northern Ireland, specify that in order to reading and spelling with their pen-and-
qualify for extra time on the grounds of slow paper equivalents and found significant
writing speed, there must be evidence that a gender differences in the latter but not in
student ‘…is unable to complete a timed the former; the reasons for this are compli-
assessment in the time allowed’ (JCQ, 2010, cated and the topic warrants further
p.4). Evidence can come from various research). Conventional tests of writing are
sources, including ‘low standardised scores’ often impractical for many teachers because
using assessments of writing speed, samples administering a writing task, counting the
of unfinished mock exams or other timed number of words written, and processing the
assessments. Neither Hedderly nor Allcock results can be very time consuming, espe-
provide norms for their tests that enable one cially when larger numbers of students need
to determine ‘low standardised scores’; to be assessed. Computerised assessment not
however, Hedderley states that at age 16 a only frees the teacher from the tasks
speed of 14 wpm or less places a student in involved in delivering and scoring the test,
the bottom 10 per cent, and Allcock speci- but also enables more efficient assessment of
fied that at age 16 a speed of 12.7 wpm is 25 groups of students using networked
per cent slower than average (which she computer systems.
indicates warrants for extra time) and 10.1

Educational & Child Psychology Vol. 28 No. 2 55

Joanna Horne, Jonathan Ferrier, Chris Singleton & Caroline Read

Study 1 allowed for each passage, this duration being

Participants chosen partly because of the criticism in the
The participants in the first study were a total research literature of both shorter and
of 952 students aged 11 years 0 months to 17 longer tests, but also as a result of the pilot
years 11 months, attending 19 schools from investigations. It was discovered that if dicta-
geographically separate regions of the UK, tion tasks are too long, students tend to get
selected to give a representative spread of bored and lose attention, reducing test relia-
types of school and socio-economic profiles. bility. Speed was measured by the computer,
The students were selected for participation based on the number of words dictated and
by class on a randomised basis, and were all – in the case of the typed passages – typed, in
tested in groups in computer suites in their the time period.
schools, using network computers running Although passage equivalence had been
Windows operating systems. established, nevertheless topic and content
clearly differed across the four passages. We
Tasks wanted to see whether any differences
Two tasks were used in the study: 1. Typing to between handwriting and typing found in
computerised dictation; 2. Handwriting to one group using two particular passages
computerised dictation. Four dictation would be replicated in a different group
passages, each of approximately 200 words, using passages with different topics and
were devised. The content was created in content. Consequently, the sample was
order to be exciting to teenagers and, as far divided into two groups, with 439 students in
as practicable, with interest for both sexes. the first group, who wrote passage 1 by hand
Passage 1 concerned the adventures of a and typed passage 2, and 513 students in the
student’s dog, passage 2 was on the various second group, who wrote passage 3 by hand
challenges of learning to keep bees, passage and typed passage 4. Speed of handwriting
3 was about the exploits of a student on a estimated by the computer was compared
family canal boating holiday, and passage 4 with speed of writing calculated by a human
centred on the thrills of quad biking. In assessor who independently counted up the
order to avoid confounding spelling skills number of words written and who also
with typing and writing speeds, the passages scored both typing and handwriting samples
were designed to impose minimal demands for spelling errors.
on spelling skills in the early paragraphs.
The content and length of these passages Results
was arrived at after pilot investigation with Table 1 shows the descriptive statistics for
students of different ages and varying both tasks. Performance of Group 2 was
competencies in writing. Pilot investigation found to be significantly better than that of
also indicated that within-subject variance Group 1 in typing [F(1,951)=4.94; p=0.026],
for writing the whole of each passage was not but not in handwriting [F(1,874)=1.48; NS].
significantly different across the four However, the effect size of the difference
passages, thus demonstrating equivalence. between the groups in typing speed was very
All passages were dictated by the small (Cohen’s d=0.14) – in fact, the average
computer using digitised speech, a short typing speed of Group 2 was only 4.7 per
phrase (four to six words) at a time, followed cent faster than that of Group 1 – and the
by a pause. Delivery rate was under the difference was not significant when age was
control of the student, who was required to statistically controlled for [F(1,951)=0.312;
press the ‘page down’ key to hear the next NS] The computer estimation of the
phrase. If desired, the student could press number of words handwritten (based on the
the ‘control’ key to hear that phrase number of words dictated in the time
repeated. A maximum of seven minutes was allowed) was compared with an independent

56 Educational & Child Psychology Vol. 28 No. 2

 Computerised assessment of handwriting and typing speed

Table 1: Descriptive statistics for typing and handwriting to dictation.

Group 1 (N=439) Group 2 (N=513) All (N=952)

Mean SD Mean SD Mean SD
Typing No. of 150.45 51.79 157.55 46.78 154.28 49.26
words typed
Typing speed 21.49 7.40 22.51 6.68 22.04 7.04
(wpm)
Handwriting No. of words 146.57 34.82 149.13 27.89 147.93 31.32
handwritten
[computer
estimation]
Handwriting 20.94 4.97 21.30 3.98 21.13 4.47
speed (wpm)
[computer
estimation]
No. of words 145.27 34.65 147.76 27.71 146.60 31.14
handwritten
[human count]
Handwriting 20.75 4.95 21.11 3.96 20.94 4.45
speed (wpm)
[human count]

count by a human assessor. There was a tions for typing were considerably higher
highly significant correlation between than those for handwriting, indicating a
computer estimation and human count much greater variability in skill. To some
(r=0.95; p<0.001) and no significant differ- extent this is accounted for by greater age-
ence was found between the two methods of related gains in speed of typing compared
calculation [F(1,870)=1.98; NS]. The with handwriting; this is also reflected in a
average discrepancy was 2.01 words or 1.37 slight negative skew in the distributions of
per cent (SD 9.53 words). In all except three typing scores due to some of the older
cases, the discrepancy was an overestimation students being particularly fast typists. There
by the computer due to the student not was a moderate correlation between student
attempting to write all the words that had age and the overall number of words typed
been listened to. In 7.5 per cent of cases the (r=0.60, p<0.001) but only a weak correlation
discrepancy between the human and between student age and the overall number
computer calculation was greater than ±5 of words written by hand (r=0.19, p<0.001).
per cent and 2.3 per cent of discrepancies Handwriting and typing speeds were found
were greater than ±10 per cent. to be correlated only weakly (r=0.14,
On average, students typed rather more p<0.001).
words than they wrote by hand, and mean Figure 1 depicts age-related growth in
typing speed was significantly faster than typing and handwriting speeds. It can be
mean handwriting speed [F(1,870)=46.00; seen that the development curve for typing is
p<0.001]. However, average typing speed was steeper than that for handwriting, particu-
only 4.1 per cent faster than handwriting larly between the ages of 11 and 14 years.
speed and the effect size was very small Although the speed increase for both is less
(d=0.15). In addition, the standard devia- marked after age 14, development of typing

Educational & Child Psychology Vol. 28 No. 2 57

Joanna Horne, Jonathan Ferrier, Chris Singleton & Caroline Read

ability progresses further and faster than that normally distributed sample. However,
of handwriting ability. Figure 2 charts the almost half (48.70 per cent) of those with
age-related changes in relative typing and slow handwriting had typing speeds above
handwriting speeds. At age 11 over 70 per standard score 85: i.e. even though they were
cent of students have faster handwriting than slow writing by hand they could type at a
typing; by age 17 the picture is reversed reasonable rate.
completely, with 85 per cent of students Although there were slightly more
being able to type faster than they can write spelling errors in typing (5.44 per cent) than
by hand. in handwriting (5.08 per cent) this differ-
Approximately one-in-five students ence was not statistically significant. The
(19.01 per cent) were found to have slow correlation between age and percentage of
handwriting for their age, i.e. they wrote at spelling errors was higher for typing
speed below standard score 85; this figure is (r=–0.28, p<0.001) than it was for hand-
roughly what would be expected in a writing (r=–0.15, p<0.001).

Figure 1: Age-related growth in speed of typing (solid line) and handwriting

(hatched line) from age 11 to 17 years (words per minute).

58 Educational & Child Psychology Vol. 28 No. 2

 Computerised assessment of handwriting and typing speed

Figure 2: Age-related changes in relative typing and handwriting speeds – percentages

of students whose typing speed is faster than their handwriting speed (solid bars) and
percentages of students whose handwriting speed is faster than their typing speed
(hatched bars).

Study 2
Participants Tasks and tests
The participants in the second study were 54 All participants completed the computerised
Year 9 students who attended a large compre- handwriting and typing to dictation task.
hensive school. These students (mean age 13 Other tests that were administered include
years 10 months; SD=4 months) were selected the Cognitive Abilities Test (Lohman et al.,
out of the total roll of 155 Year 9 students by 2001), WRAT4 Spelling Test (Glutting &
the following process. First, those students in Wilkinson, 2005), the Suffolk Reading Scale
the two top groups whose literacy skills were (Hagley, 1987) and the Hedderley Sentence
not of concern (based on the results of previ- Completion Test (SCT), although student
ously-administered conventional tests) were absences on some testing sessions meant that
ruled out. Second, the remaining students there were a few gaps in the data set. Partici-
from the two top groups (i.e. those whose pants were assessed in groups using the
literacy skills did give some cause for school’s networked computer suite.
concern), together with the rest of the
students in the year who had not yet been Results
assessed, formed a pool of students available The results of all the measures used in this
for participation. Third, the 54 who actually study are shown in Table 2. The mean of the
took part in the study were selected because Cognitive Abilities Test scores is 87.9
they were available when the school’s (SD=10.1), suggesting that this group of
computer facilities were accessible so that the students was of slightly below average
computerised tasks could be carried out. general ability, and reading skills as assessed

Educational & Child Psychology Vol. 28 No. 2 59

Joanna Horne, Jonathan Ferrier, Chris Singleton & Caroline Read

by the Suffolk Reading Scale were rather speed but below SS 95 for typing speed; 4. SS
poor (M=78.72, SD=7.45), while spelling 95 or greater for both handwriting and typing
skills (as assessed by WRAT-4) were average. speeds. The results are shown in Table 4.
This apparent disparity is most likely to have Significant group differences were found
been due to differences between the two only on CAT Verbal scores [F(3,42)=4.36;
tests in standardisation samples and dates, p<0.01], where a clear pattern of increasing
reflecting different educational standards as scores was seen across the groups 1 to 4. Post-
well as cohort effects (the Suffolk Reading hoc multiple comparisons revealed signifi-
Scale was standardised on UK samples in the cance only when comparing Group 1 with
mid-1980s, while the WRAT-4 Spelling Test Group 4 [Tukey HSD; p=0.028], but the
was standardised on US samples in the early effect size of the difference between these
2000s). However, speeds on all three writing two groups was large (d=1.30). Differences
measures were average, which is not particu- on spelling scores approached significance
larly surprising. [F(3,28)=2.87; p=0.057]. Group 1 performed
Correlational analysis revealed strong consistently poorly, with standard scores
significant relationships between reading, below 90 across all measures. Three-quarters
spelling and two of the three cognitive ability of this group were already on the school’s
measures (see Table 3). However, the picture SEN register and/or had identified educa-
for relationships with handwriting and typing tional support needs such as ADHD, speech
speeds was somewhat mixed. Both hand- and communication difficulties, moderate,
writing and typing speeds based on the borderline or specific learning difficulties,
computer tasks had moderate correlations or spoke English as an additional language
with verbal ability (this was not significant for (EAL). When all 26 students in this sample
handwriting owing to the relatively small size who had handwriting speed below SS 95 (i.e.
of the sample) and also with spelling skills. Groups 1 and 2) were considered, 31 per
But the scores from the Hedderley Sentence cent were found not to be on the school’s
Completion Test had only a weak negative SEN register and had not been identified as
correlation with verbal ability (r=–0.16; NS); having educational support needs.
in fact, the Sentence Completion Test had low
and/or non-significant correlations with all Discussion
the other measures except the computerised The aim of these two studies was to investi-
measure of handwriting (r=0.54, p<0.01). gate the use of computer-based dictation for
The participants were categorised into measurement of speed of typing and hand-
four different groups based on relative hand- writing. The first study, which used a large
writing and typing speed (standard scores unselected sample of 952 students aged 11 to
(SS)): 1. Below SS 95 for both handwriting 17 years, demonstrated that the comput-
and typing speed; 2. Below SS 95 for hand- erised tasks were effective in measuring both
writing speed but SS 95 or greater for typing handwriting and typing speeds. Results
speed; 3. SS 95 or greater for handwriting showed that handwriting speed increased

Table 2: Descriptive statistics for all measures used in Study 2 (standard scores).

CAT Verbal CAT CAT Suffolk WRAT4 Sentence Handwriting Typing to

Non-verbal Quantitative Reading Spelling Completion to dictation dictation
Scale Test Test
Mean 88.37 84.85 90.36 78.72 93.62 95.13 96.46 96.80
SD 11.25 8.18 9.79 7.45 10.79 19.57 12.34 15.96

60 Educational & Child Psychology Vol. 28 No. 2

 Computerised assessment of handwriting and typing speed

Table 3: Intercorrelations of all measures used in Study 2.

CAT V CAT NV CAT Q Reading Spelling SCT HW Typing

speed speed speed
CAT V 1 .65** .54** .78** .66** –.16 .29 .34*
CAT NV .65** 1 .66** .61** .53** –.18 .16 .07
CAT Q .54** .66** 1 .50 .31 –.18 .22 .17
Reading .78** .61** .50** 1 .40* –.05 .18 .20
Spelling .66** .53** .31 .40* 1 .06 .36* .47*
SCT speed –.16 –.18 –.18 –.05 .06 1 .54** .21
HW speed .29 .16 .22 .18 .36* .54** 1 .48**
Typing speed .34* .07 .17 .20 .47* .21 .48** 1

**Significant at p<.01 level; *Significant at p<.05 level.

Table 4: Performance grouped by relative handwriting and typing speed

(standard scores).
Group N HW Typing CAT V CAT NV CAT Q Reading Spelling SCT
speed speed speed
1 18 M 85.94 83.11 83.40 83.00 88.87 76.15 89.31 89.00
SD 4.82 8.12 9.64 6.61 10.45 5.79 9.59 17.50
2 8 M 86.63 106.50 86.86 84.43 90.00 84.00 96.00 83.33
SD 5.81 8.65 6.82 8.10 9.18 5.00 1.73 7.09
3 10 M 102.80 83.90 88.00 88.50 89.75 79.29 97.00 100.29
SD 8.43 7.42 10.39 9.83 8.62 8.98 10.12 14.17
4 18 M 107.83 113.33 97.08 87.58 95.00 79.80 102.80 109.33
SD 8.77 6.16 11.35 8.61 9.70 9.68 9.37 26.40

progressively and in a near-linear fashion with, and thereafter steadily overtook, hand-
from about 16 wpm at age 11 to about 22 writing speed, with students aged 17 attaining
wpm at age 14, and thereafter began to level an average typing speed of almost 27 wpm.
out, reaching almost 24 wpm at age 17. When Again, the trajectory up to age 14 was near-
task and age differences are allowed for, this linear, levelling off thereafter. However, there
trajectory is consistent with results reported appear to be no comparable studies against
in previous studies that have used conven- which to compare the findings for typing
tional rather than computer-based methods speed, although the findings are consistent
(Allcock, 2001; Connelly et al., 2007; with those of Rogers and Case-Smith (2002),
Graham, 1986a, 1986b; Phelps et al., 1985; who reported that 70 per cent of their sample
Hedderley, 1996; Rosenblum et al., 2003; of sixth graders who had received typing
Wallen et al., 1996). Typing speed at age 11 tuition achieved typing speeds that exceeded
(about 14 wpm) lagged behind handwriting their handwriting speed.
speed, but by age 13 (20 wpm) had caught up

Educational & Child Psychology Vol. 28 No. 2 61

Joanna Horne, Jonathan Ferrier, Chris Singleton & Caroline Read

There was a rather low correlation The percentage of spelling errors in both
between typing and handwriting speeds the typing and handwriting tasks was low –
(r=0.14), which might at first sight be consid- about five per cent – suggesting that, in
ered surprising, given figures cited in the comparison with free writing tests, spelling
research literature. Previous studies have was not a confound in these tasks. The high
reported higher correlations, ranging from correlation (r=0.95) between the computer
0.3 to 0.7. However, these involved younger estimation and human count of number of
children, and were mostly carried out at a words in the handwriting task, with an
time when computers were not so widely used average discrepancy of only 2.01 words or
in schools and home ownership of computers 1.37 per cent, indicates that for the vast
was much less common. It seems likely that majority of cases, the computer’s estimation
the current generation of teenagers is, in can be relied on. Brief visual inspection by
general, much more proficient in the use of a the teacher of handwriting samples is
keyboard than were their predecessors but enough to determine whether a given
also that there are large variations in typing student has complied with the task require-
skill (a conclusion supported by the much ments and attempted to type the passage
higher standard deviations for typing than accurately and to the best of their ability.
handwriting) and that consequently typing A human count can be carried out when it
speed in secondary school has become largely can be clearly seen that the number of words
dissociated from handwriting speed. The actually written is far fewer than the figure
results of this study show that most students estimated by the computer. This situation,
increase their typing skills steadily during the which was rare in this study, may arise if the
secondary school years: while at age 11 the student has skipped a large proportion of
majority of students can write by hand faster the words in the passage. In this study the
than they can type, by the time they have discrepancy between the human and
reached age 14 their typing speed has over- computer calculation was greater than ±5
taken their handwriting speed, and by age 17 per cent in only 7.5 per cent of cases and
all but a small proportion of students can type greater than ±10 per cent in 2.3 per cent of
faster than they can write by hand. In primary cases. It is likely that good preparation of
school almost all children use a keyboard by students before assessment – for example, by
the ‘hunt and peck’ method – i.e. searching warning them that the teacher will be able to
for each key (Kennewell, 2001). Although detect when they have failed to undertake
probably few secondary school age students the task properly – would reduce these
can touch type – otherwise average typing figures further.
speeds would have been much higher The second study examined the relation-
(Brown, 1988a) – through repeated practice, ships between speeds of typing and hand-
keyboard familiarity has developed such that writing and various other measures. This was
automaticity and motor memory has enabled a much smaller-scale study in which the
much faster speeds to be achieved than when participants were not an unselected group
younger (Twining, 2002). By contrast, the and mean general ability was found to be a
greater complexity of motor control processes little below the average range. However, the
needed for letter formation places an computerised measure of handwriting was
inherent limitation on growth of handwriting found to have a fairly strong correlation
speed: for most individuals, pressing a key is a (r=0.54; p<0.01) with scores from a widely-
faster movement than writing a character and used conventional test of handwriting speed
even adults who are experienced two-fingered (Hedderly Sentence Completion Test),
typists normally type faster than they can write which may be considered to provide a
by hand (Brown, 1988b; Rogers & Case- modest degree of validation for the comput-
Smith, 2002). erised test. In contrast to the findings of the

62 Educational & Child Psychology Vol. 28 No. 2

 Computerised assessment of handwriting and typing speed

first study, in which a relatively low correla- was precluded here not only because of
tion between typing and handwriting speeds space but also because the large size of the
was found (r=0.14), in the second study this data set in the first study means that analysis
correlation was much larger (r=0.48; of legibility is a lengthy task which has yet to
p<0.001) and more consistent with correla- be completed. It is envisaged that a future
tion figures reported in the literature. paper will report the findings on this topic.
However, it should be remembered that the In conclusion, these two studies indicate
sample in this study was not unselected and that computer-based tasks provide a feasible
was below average in reading. Students with and practical solution to the problems of
poorer verbal and reading skills are more assessing speed of handwriting and typing in
likely than other students to struggle with secondary school students. The size of the
writing generally, regardless of the mode of data set in the first study (952 students) and
text production (Berninger et al., 1991), and the normal distribution of scores obtained
hence a higher correlation between typing from both the typing and handwriting tasks
and handwriting speeds would be expected. enabled standardisation to be carried out so
Both the computerised measures of hand- that raw scores could be expressed in stan-
writing and typing speeds were found to dard scores in yearly age bands, which are
have moderate correlations with verbal recommended when ascertaining eligibility
ability (this was not significant for hand- for additional time in examinations (Back-
writing owing to the relatively small size of house, Dolman & Read, 2007). The avail-
the sample) and also with spelling skills. ability of normative rates for typing for this
However, this does not necessarily under- task, together with the comparison of the
mine the conclusion drawn earlier – namely two modes of writing, also enables teachers
that spelling was not a confound in the to decide whether it would be appropriate to
computerised tasks – because such correla- advocate that a student use a word processor
tion figures would be expected in a group in in examinations or whether an amanuensis
which students with below-average ability would be more suitable. However, it is
predominated (Berninger et al., 1991). accepted that students who have typing
When participants were grouped and/or handwriting speeds within the
according to their relative performance in average range for their age may still have
typing and handwriting, significant differ- difficulties in the production of ideas that
ences between groups were found for verbal would slow them down significantly in a free
ability, with the difference for spelling closely writing task, and that such skills also need to
approaching significance. In particular, the be taken into account when considering
group with both typing and handwriting eligibility for additional time. Nevertheless,
below SS 95 were found to be below average it is argued that, by eliminating the
in all measures, including reading, spelling ‘thinking’ time involved in free writing, the
and general abilities: most of the students in computerised dictation tasks provide ‘purer’
this group were already known to have measures of speed which can reveal physical
special educational needs or require addi- handwriting and/or typing problems. The
tional learning support. computer tasks also simulate examination
It is not an oversight that the matter of requirements more closely than mechanical
legibility of handwriting has not been repetitive tests of writing speed, and should
addressed in this paper. This is clearly an be particularly helpful in establishing
important factor (see Feder & Majnemer, whether students need access arrangements
2007; Graham et al., 1998; Roaf, 1998; in examinations. It is notable that when all
Rosenblum et al., 2003; Zivani, 1984; Zinavi those in the second study who had below
& Elkins, 1984) which warrants detailed average handwriting speed were examined,
analysis. However, consideration of legibility almost one-third were found not to be on the

Educational & Child Psychology Vol. 28 No. 2 63

Joanna Horne, Jonathan Ferrier, Chris Singleton & Caroline Read

school’s SEN register or not to have any Acknowledgments

identified educational support needs. Given The authors would like to thank Lucid
that such students would potentially be Research Ltd. for programming the comput-
disadvantaged in written examinations, the erised tasks used in these studies and also the
uncovering of their needs neatly illustrates staff and students of the 20 schools that
the educational value in using a computer- participated.
based test that can easily be administered to
all students. Address for correspondence:
Dr Joanna Horne
Department of Psychology,
University of Hull,
Hull, HU6 7RX, UK.
Email: j.k.horne@hull.ac.uk

References
Allcock, P. (2001). Testing handwriting speed. Evesham: Connelly, V., Campbell, S., MacLean, M. & Barnes, J.
Patoss. Retrieved from: (2006). Contribution of lower order skills to the
www.patoss-dyslexia.org written composition of college students with and
Alston, J. (1985). The handwriting of 7- to 9-year-olds. without dyslexia. Developmental Neuropsychology,
British Journal of Special Education, 12, 68–72. 29, 175–196.
Alston, J. (1995). Assessing and promoting writing skills. Connelly, V., Gee, D. & Walsh, E. (2007). A compar-
Tamworth: Nasen Enterprises Ltd. ison of keyboarded and handwritten composi-
Ashton, C. (1997). The assessment of handwriting tions and the relationship with transcription
speed. Dyslexia Review, 9, 8–11. speed. British Journal of Educational Psychology, 77,
Backhouse, G., Dolman, E. & Read, C. (2007). 479–492.
Dyslexia: Assessing the need for access arrangements Cornhill, H. & Case-Smith, J. (1996). Factors that
during examinations (3rd ed., L. Greenwold, Ed). relate to good and poor handwriting. American
Evesham: Patoss. Journal of Occupational Therapy, 50, 732–739.
Berninger, V. & Graham, S.(1998). Language by Dockrell, J.E. & Barnett, J. (2005). The slow hand-
hand: A synthesis of a decade of research on writing of undergraduate students constrains
handwriting. Handwriting Review, 12, 11–25. overall performance in exam essays. Educational
Berninger, V.W., Mizoikowa, D.T. & Bragg, R. (1991). Psychology, 25, 99–107.
Theory-based diagnosis and remediation of Dunn, B. & Reay, D. (1989). Word processing and the
writing disabilities. Journal of Educational keyboard: Comparative effects of transcription
Psychology, 29, 57–59. on achievement. Journal of Educational Research,
Bishop, E. & Esgate, A. (2001). Writing speed and extra 84, 237–245.
time in examinations. Paper presented at the 5th Exner, C.E. (1989). Development of hand functions.
International Conference of the British Dyslexia In P.N. Pratt & A.S. Allen (Eds.), Occupational
Association. University of York. therapy for children (pp.235–259). St Louis, MO:
Bonney, A.M. (1992). Understanding and assessing Mosby Year Book.
handwriting difficulties: Perspective from the Feder, K.P. & Majnemer, A. (2007). Handwriting
literature. Australian Occupational Therapy Journal, development, competency, and intervention.
39, 7–15. Developmental Medicine & Child Neurology, 49,
Brown, C.M. (1988a). Human-computer interface design 312–317.
guidelines. Norwood, NJ: Ablex Publishing. Glutting, J. & Wilkinson, G. (2005). Wide Range
Brown, C.M. (1988b). Computer systems comparison Achievement Test (WRAT) (4th ed.). Austin TX:
of typing and handwriting in ‘two-fingered Pro-Ed.
typists’. Human Factors and Ergonomics Society Graham, S. (1986a). A review of handwriting scales
Annual Proceedings, 32, 381–385. and factors that contribute to variability in hand-
Christensen, C.A. (2004). Relationship between writing scores. Journal of School Psychology, 24,
orthographic-motor integration and computer 63–72.
use for the production of creative and well-struc- Graham, S. (1986b). The reliability, validity, and
tured written text. British Journal of Educational utility of three handwriting measurement proce-
Psychology, 74, 551–565. dures. Journal of Educational Research, 79, 373–380.

64 Educational & Child Psychology Vol. 28 No. 2

 Computerised assessment of handwriting and typing speed

Graham, S., Berninger, V., Weintraub, N. & Schafer, McDonald, A.M. (2002). The impact of individual
W. (1998). The development of handwriting differences on the equivalence of computer-
speed and legibility in grades 1 to 9. Journal of based and paper-and-pencil educational assess-
Educational Research, 92, 42–52. ments. Computers and Education, 39, 299–312.
Graham, S., Harris, K.R. & Fink, B. (2000). Is hand- McHale, K. & Cermak, S. A.(1992). Fine motor activ-
writing causally related to learning to write? ities in elementary school: Preliminary findings
Treatment of handwriting problems in begin- and provisional implications for children with
ning writers. Journal of Educational Psychology, 4, fine motor problems. American Journal of Occupa-
620–633. tional Therapy, 46, 898–892.
Hagley, F. (1987). Suffolk Reading Scale. London: Ofqual (2010). Statistical Bulletin: Access Arrangements
NFER-Nelson. for GCSE and GCE – 2009/10 Examination Series.
Hamstra-Bletz, L. & Blöte, A.W. (1993). A longitu- London: Office of Qualifications and Examina-
dinal study on dysgraphic handwriting in primary tion Regulations.
school. Journal of Learning Disabilities, 26, O’Mahony, P., Dempsey, M. & Killeen, H. (2008).
689–699. Handwriting speed: Duration of testing period
Hedderly, R. (1992). Psychologists assessments of and relation to socio-economic disadvantage and
specific learning difficulty (dyslexia) and exami- handedness. Occupational Therapy International,
nation boards: Policies and practices. Educational 15, 165–177.
Psychology in Practice, 6, 32–42. Phelps, J., Stempel, L. & Speck, G. (1985). The
Hedderly, R. (1995). The assessment of SpLD pupils children’s handwriting scale: A new diagnostic
for examination special arrangements. Dyslexia tool. Journal of Educational Research, 79, 46–50.
Review, 7, 2–19. Roaf, C. (1998). Slow hand: A secondary school
Hedderly, R. (1996). Assessing pupils with specific survey of handwriting speed and legibility.
learning difficulties for examination special Support for Learning, 13, 39–42.
arrangements at GCSE A-level and degree level. Rogers, J. & Case-Smith, J. (2002). Relationships
Educational Psychology in Practice, 12, 36–44. between handwriting and keyboarding perform-
Horne, J. (2007). Gender differences in comput- ance of sixth-grade students. American Journal of
erised and conventional educational tests. Journal Occupational Therapy, 56, 34–39.
of Computer Assisted Learning, 23, 47–55. Rosenblum, S., Weiss, P.L. & Parush, S. (2003).
JCQ (2010). Access arrangements, reasonable adjustments Product and process evaluation of handwriting
and special consideration general and vocational qual- difficulties: A review. Educational Psychology
ifications with effect from 1 September 2010 to Review, 15, 41–81.
31 August 2011. Joint Council for Qualifications. Sawyer, C., Francis, M. & Knight, E. (1992). Hand-
Retrieved from: writing speed, specific learning difficulties and
www.jcq.org.uk the GCSE. Educational Psychology in Practice, 8,
Jones, D. & Christensen,C.A.(1999). Relationship 77–81.
between automaticity in handwriting and Sawyer, C., Gray, F. & Champness, M. (1996). Meas-
student’s ability to generate written text. Journal uring speed of handwriting for the GCSE candi-
of Educational Psychology, 91, 44–49. dates. Educational Psychology in Practice, 12, 9–23.
Karlsdottir, R. & Stefansson, T. (2002). Problems in Schoemaker, M.M., Ketelaars, C.E.J., von Zonneveld,
developing functional handwriting. Perceptual M., Minderaa, R.B. & Mulder, T. (2005). Deficits
and Motor Skills, 94, 623–662. in motor control processes involved in produc-
Kennewell, S. (2001). Using affordances and tion of graphic movements of children with
constraints to evaluate the use of information attention deficit hyperactivity disorder. Develop-
and communications technology in teaching and mental Medicine and Child Neurology, 47, 390–395.
learning. Journal of Information Technology for Singleton, C.H. (2001). Computer-based assessment
Teacher Education, 10, 101–116. in education. Educational and Child Psychology, 18,
Lohman, D.F., Thorndike, R.L., Hagen, E., Smith, P., 58–74.
Fernandes, C. & Strand, S. (2001). Cognitive Smits–Engelsman, B.C.M. & Van Galen, G.P. (1997).
Abilities Test (3rd ed.). London: NFER-Nelson. Dysgraphia in children: Lasting psychomotor
Maeland, A.E. (1992). Handwriting and perceptual deficiency or transient developmental delay?
motor skills in clumsy, dysgraphic, and normal Journal of Experimental Child Psychology, 67,
children. Perceptual and Motor Skills, 75, 164–184.
1207–1217. Summers, J. & Catarro, F. (2003). Assessment of
Malloy-Miller, T., Polatajko, H. & Anstett, B. (1995). handwriting speed and factors influencing
Handwriting error patterns of children with mild written output of university students in examina-
motor difficulties. Canadian Journal of Occupa- tions. Australian Occupational Therapy Journal, 50,
tional Therapy, 62, 258–267. 148–157.

Educational & Child Psychology Vol. 28 No. 2 65