ACQ Vol 11 no 2 2009

21 / 68

ACQ Volume 11, Number 2 2009 83
corresponding passage. The results indicated that 86% of
the questions used in the GORT–4 were answered correctly
more than 25% of the time (i.e., above chance level). No
significant correlations were found between performance on
the PI items and other spoken language comprehension test
measures, and performance on the PI items was not
sensitive to reading disability (as identified by word
recognition difficulties). Keenan and Betjemann (2006)
concluded that performance on the reading comprehension
measure of the GORT–4 yields information about the child’s
previous general world knowledge, rather than the child’s
ability to derive meaning from a written text.
developmental stage the child is at might influence the child’s
test performance. Four different reading comprehension
tests were administered to a sample of 510 children, ranging
in age from 8 to 18 years. The tests were:
• GORT–3 (see table 1);
• Qualitative Reading Inventory – 3 (Leslie & Caldwell,
2001), in which the students read long passages out
loud; it involves two assessments of comprehension:
open-ended short-answer questions and retelling of one
of the passages;
• Peabody Individual Achievement Test (Dunn & Markwardt,
1970, cited in Keenan et al., 2008); the student silently
reads single sentences and short passages and indicates
comprehension by selecting from four pictures the one
that best fits the meaning of the sentence/passage just
read;
• WRMT–R Passage Comprehension subtest (see table 1).
In addition, children’s listening comprehension (at passage
level) and word recognition skills were assessed. Consistent
with previous studies, it was found that tests using a cloze
format were more heavily dependent on word recognition
skill. More particularly, tests using very short passages
(one to two sentences) were the most sensitive to word
recognition ability. Keenan et al. (2008) reasoned that the
longer the passages are, the more chance the child has
to use the context (and thus demonstrate his listening
comprehension skills) to understand the passage, even if he
has trouble decoding one or two specific words. In contrast,
in very short passages, a child may fail to understand
the meaning of the material when unable to decode one
or two keywords. Unfortunately, these researchers did
not investigate whether oral or silent reading of the text
influenced task performance. It can be argued that reading
out loud may distract some children from attending to the
meaning of the text. Further research is required to test this
hypothesis.
When examining developmental differences (as a function
of chronological age and of reading age) in test performance,
Keenan et al. (2008) found that, consistent with the simple
view of reading, word recognition skills accounted for more
variance in reading comprehension in the younger age group
(mean age 9;1 years) than in the older age group (mean age
13;1). In addition, the findings demonstrated that tests using
cloze formats were more dependent on word recognition
ability in young children or in children with reading difficulties
than those tests using longer passages and question and
answer formats.
Passage dependency and previous
world knowledge
Question-and-answer tests can be divided into multiple
choice tasks and open-ended-question tasks. For example,
the GORT–4 uses multiple choice questions following the
reading of a passage, whereas the NARA requires the child
to answer a set of open-ended questions about a passage,
immediately after the child has read the passage out loud.
Keenan and Betjemann (2006) conducted a study to
investigate if the multiple choice format that is used by the
GORT–4 would be passage dependent. Passage dependent
(PD) questions can only be answered correctly by using
information from the passage, whereas passage
independent (PI) questions can also be answered without
reading the text. A total of 77 undergraduate students in
Education participated in a passageless administration of the
GORT–4. This procedure involved asking the students the
comprehension questions without exposing them to the
The contribution of pre-existing content-relevant
knowledge was highlighted in a study investigating the
comprehension and memory for spoken and written texts
of 181 adults, aged between 18 and 86 years (Hambrick
& Engle, 2002). Hambrick and Engle found that the
participants’ knowledge of the game of baseball significantly
facilitated their performance on questions (tapping their
memory of the game details) following exposure to a tape-
recorded narration of a baseball match. Participants who
were familiar with the game of baseball were better able to
answer questions related to the progress of the game as well
as more specific questions related to the performance of the
batsmen.
Test taking strategies
To determine if children’s test taking strategies would
influence their performance on a reading comprehension
test, Bornholt (2002) administered the Test of Reading
Comprehension (TORCH; Mossenson, Hill, & Masters, 1987)
to 30 children (aged 9 and 10) attending year 4 of their local
primary school. The TORCH uses a story booklet and an
answer booklet and children have to fill in blanks left in a
passage. In addition, children’s test-taking strategies were
observed, including procedural initiative, task involvement
and item responses (e.g., number of questions attempted).
Following the pre-test, all children participated in two 1-hour
class discussions about test-taking strategies and were
reassessed in week 4. Results indicated that most
improvement in reading test scores was observed for
children with initially poor test-taking strategies. Post-
intervention these children showed higher test completion
rates, and were more likely to attempt various items rather
than persisting with the items in a set order. Increased
awareness of this possibility in clinical practice may thus help
improve the accuracy of diagnosis.
Summary
This paper has discussed issues that may influence
children’s performance on tests of reading comprehension.
Research clearly indicates that the choice of reading
comprehension assessment in clinical and research practice
is critical as some tests may incorrectly identify children as
demonstrating good or poor reading comprehension skills
(Cutting & Scarborough, 2006; Keenan et al., 2008). The test
format plays an important role. To ensure adequate detection
of a child’s strengths and weaknesses in word recognition
Box 1. The simple view of reading
Reading comprehension = Word recognition X Listening
comprehension
Source: Gough & Tunmer, 1986