Arab World English Journal (AWEJ) Volume 9. Number 2. June 2018 Pp.

234 -250
DOI: https://dx.doi.org/10.24093/awej/vol9no2.16

Lexical Frequency Effect on Reading Comprehension and Recall

Nadia Nouri
Faculty of Education, Mohammed V University
Rabat, Morocco

Badia Zerhouni
Faculty of Education, Mohammed V University
Rabat, Morocco

Abstract
The present study investigates the effect of lexical difficulty, as measured by frequency, on reading
comprehension and recall. It also estimates the relationship between vocabulary size, vocabulary
depth, reading comprehension, and recall. To this end, 80 English as a foreign language (EFL)
undergraduate university students are administered three standardized instruments including two
vocabulary tests and a reading comprehension test. The latter comprises two similar passages (in
terms of length and topic) one of which is adjusted by replacing 18% of its words by their low-
frequent synonyms, and both passages are estimated lexically to measure their difficulty. Paired
samples t-test results show that comprehension and recall are significantly low (p<.001) in the
modified passage. This finding further confirms that lexical frequency measure is an effective
estimate in determining reading material difficulty. Furthermore, for the second objective, Pearson
product-moment analysis reveals a significantly high correlation between size and comprehension,
a moderate to low correlation between depth and comprehension, and a moderate correlation
between depth and recall tasks. Consequently, the study sugguests estimating the complexity of
EFL reading academic material with lexical difficulty measure using frequency criterion to cope
with students’ reading deficiencies, and encourage explicit lexical instructions at EFL
undergraduate university levels mainly.
Keywords: lexical frequency, reading comprehension, recall, vocabulary depth, vocabulary size
Cite as: Nouri,N., & Zerhouni, B. (2018). Lexical Frequency Effect on Reading Comprehension
and Recall. Arab World English Journal, 9(2). DOI: https://dx.doi.org/10.24093/awej/vol9no2.16

234
 Arab World English Journal (AWEJ) Volume 9. Number 2. June, 2018

Introduction
Research has shown that lexical knowledge constitutes a “pressure point” in reading
comprehension (i.e. the linkage between word recognition and text comprehension) (Perfetti,
1985; Perfetti & Hart, 2002; Perfetti, 2007). This is the reason why several researchers have been
interested in estimating text difficulty for comprehension. They have designed traditional
readability formulas (e.g. Lively & Pressey, 1923; Patty & Painter, 1931; Ojemann, 1934) as well
as modern computerized measures (e.g. Lexile, 2007; Milone & Biemiller, 2014; McNamara et
al., 2014). These assessment tools show high reliability with reading complexity (Nelson et al.,
2012). However, given the different features estimated in these formulas such as word syllables,
sentence length, and frequency, it is unclear which aspect better determines reading complexity.
An alternative to these formulas is testing text difficulty by word complexity estimate (Chen &
Meurers, 2016, 2017). This research paper has opted for this option because it seems more
appropriate to the study objectives. The latter, therefore, focuses on investigating the extent to
which word difficulty, as measured by frequency, affects reading comprehension and recall, and
examining the relationship between Moroccan EFL students’ vocabulary size, depth, reading
comprehension, and recall.

Literature Review
Vocabulary Knowledge: Aspects Definition and Relationship
Some researchers contend that knowing a word is characterized by its form, meaning, and
use (e.g., Alderson, 2000; Read, 2000; Schmitt, 2000; Nation, 2001). For other researchers, word
knowledge includes meaning, usage, grammatical and semantic constraints, associations, and
morphology (e.g., Koda, 2005; Bernhardt, 2005; Zhang, 2012). These frameworks are all
appropriate to define the construct of interest. However, a better understanding of a word relies on
dissecting lexical knowledge into small manageable aspects for insightful testing. In this study,
vocabulary knowledge is conceptualized by size and depth dimensions. Vocabulary size is defined
as knowing the single meaning of a word, and is estimated by Vocabulary Size Test (VST) (Nation
& Beglar, 2007). Vocabulary depth, on the other hand, refers to the degree in which a word is well
known, and is measured by Word Associates Test (WAT) (Read, 1993). Anderson and Freebody
(1981) are the pioneers of this framework. They advance that internalizing words begins with a
shallow knowledge, and as learners proceed in learning, their word mastery deepens as well. Thus,
size and depth seem to constitute a continuum, even though some researchers tend to contrast them
as two distinct dimensions. In the following paragraphs, discussion will be centered on the
relationship between these two aspects of vocabulary knowledge as tested by Vocabulary Levels
Test (VLT) and Word Associates Test (WAT) in particular.

One of the studies that investigated this issue was conducted by Qian (1998) who
investigated the relationship between vocabulary depth, size, and reading comprehension. The
sample included 74 learners attending English as second language (ESL) intensive classes in
Ontario Universities. His results revealed a significantly strong correlation between VLT and
WAT r=.82, p<.05. After some modifications on the depth measure, Qian concluded that depth
adds a unique contribution to reading comprehension. One probable reason for such a finding is
that some words from the reading test are also part of the depth measure. Therefore, participants
Arab World English Journal 235
www.awej.org
ISSN: 2229-9327
 Arab World English Journal (AWEJ) Volume 9. Number 2. June, 2018

who responded correctly to the target words in the depth test might probably have found it easier
to comprehend these items in the reading test context. To account for this side effect, Qian varied
the order of the tests’ administration in each testing session by administering, for example, the
reading test before depth and size tests in one session and administering depth test first followed
by size and reading measures in another session. However, possible learning of the target words
may still happen if depth test is administered before reading.

In a sequel study, Qian (2002) investigated a large sample composed of 217 students
attending an intensive ESL program at Toronto University and coming from nineteen first
language (L1) backgrounds. Results revealed that correlation between VLT and WAT was not as
high as in his 1998 findings, with a coefficient of r=.70, p< .01. One possible explanation of such
finding might be the diversity of participants’ L1 backgrounds. It is quite probable that one sub-
sample of participants might have a better developed lexical knowledge than other sub-samples
given different learning experiences and the amount of lexical input they might have accumulated.
Besides, lower correlations have been found as well in the VLT-WAT comparison conducted by
Zhang and Horiba’s studies (2012) on Chinese learners with a coefficient around r=.52, p<.01.

It should be noted, based on the findings of most studies conducted on this issue, including
those reported in this section that no negative correlation has been revealed between vocabulary
size and depth. This may indicate that these two lexical dimensions are not separate entities as
supported by a number of researchers (Read, 2004; Milton, 2009; Henriksen, 1999). However, size
and depth seem to converge more largely at advanced proficiency levels.

Concerning measurement of lexical size, there is a fundamental difference between a

diagnostic measure and an overall performance test. Thus VLT, despite its high reliability and
close relation to depth is only a diagnostic test, as it merely estimates ‘slices’ of lexical items across
five frequency levels (i.e. 2000, 3000, 5000, the university word list, and 10000 word level).
Therefore, even though it is widely used in lexical studies, it should not be interpreted as an
estimate or even a proxy measure of size (Nation, personal communication). For this reason, the
present study has opted for an alternative estimate: Vocabulary Size Test (VST) (Nation & Beglar,
2007). It consists of 14 frequency levels, including 10 items per level. However, given the limited
lexical competency of the present study participants, and based on the test piloting, it was judged
adequate not to include the last four levels in the administered version though Nation (2012)
recommends completing the whole test. This omission was based on the researchers’ belief that it
would not distort the overall measure, and, at the same time, would reveal the actual lexical
proficiency of the participants. Another reason for choosing VST is its ability to distinguish
different vocabulary sizes, which plays a highly important role in reading comprehension
investigation (Stuart & Kramer, 2015).

Few researchers working on vocabulary size and depth have been concerned with text
difficulty even though performance in reading comprehension is not only related to readers’ lexical
knowledge, but also to the complexity of the reading material. Generally, the latter issue is tackled
through deletion, instruction, and substitution studies of vocabulary. However, to the best of the
Arab World English Journal 236
www.awej.org
ISSN: 2229-9327
 Arab World English Journal (AWEJ) Volume 9. Number 2. June, 2018

researchers’ knowledge, it has not been fairly related to vocabulary size and depth. The following
sections discuss the role of lexical difficulty in estimating text complexity, and shed light on word
information level of texts, which is central to the present study.

Lexical Frequency, Word Information Level, and Reading Comprehension

Based on the ‘Lexical Quality Hypothesis’ (Perfetti & Hart, 2002), word frequency is the
result of the benefits of language learning experience. Higher frequency words make their meaning
and form “a more stable constituent compared with the lower frequency meaning” (Perfetti et al.,
2002, p. 194). This concept is associated with the semantic aspect of lexis mainly. Besides,
Anderson and Freebody (1979) define frequency as “a characteristic of a word which probably is
very strongly related to the chances that the word will be known” (p. 23). For this reason,
performance in reading comprehension can be analyzed from a lexical perspective.

Frequency researchers have conducted several studies to determine the cut-off line between
high and low-frequency levels, using English corpora. For example, Nation (2001) suggests that
2000-word families are the upper line of high frequency. He also contends that a critical goal of
8000-word families is what learners need to read complex written texts (Nation, 2006). Academic
material is specifically the most demanding in reading as it may require this threshold to attain
comprehension and learning (Schmitt, 2010). A good question to ask here is whether Moroccan
EFL learners reach this threshold at the undergraduate university level. Another point worth
inquiring in further studies is whether, with this objective, a cut-off line of high frequency is still
2000-word families. Schmitt (2010) indicates that it is difficult to stop at this category line of high
frequency, and suggests three frequency bands to bridge the gap between high and low frequencies:
1000-3000 word families range is high, 3000-7000 is mid, and 8000-10000 is low frequency
(Schmitt, 2014). Based on Schmitt (2014) categorization, academic EFL material may fall within
the mid-low category.

Furthermore, text coverage is highly related to comprehension. Thus, in reading, 95%

(Laufer, 1989) and 98% (Hu & Nation, 2000) of the running words in texts should be familiar to
readers. In this regard, Schmitt (2010) clarifies that these rates have become a rule of thumb thanks
to the number of studies reporting them and that specialists of texts coverage did not intend in the
first place to make these figures a reference given the limited number of participants in their studies
and the small number of texts analyzed. One should question, therefore, whether these ratios are
universal for reading comprehension. Practically, it seems unrealistic to teach systematically the
lexical knowledge needed for 95% or 98% of text coverage in an EFL context.

Mezynski (1983), on the other hand, points out interestingly that the proportion of unknown
words within a text is not as important as that of words with high information level. This claim
puts coverage equations into query. From these conflicting perspectives, one may question whether
readers need to know only the words with high information level to achieve comprehension. Unlike
marginal words, items with high information value give significant information about the meaning
of sentences in a passage. Adams and Bruce (1980) clarify this by the following example “the
discovery of a number of fossilized porbeagles in Kansas is intriguing” (p.8). If one does not know
Arab World English Journal 237
www.awej.org
ISSN: 2229-9327
 Arab World English Journal (AWEJ) Volume 9. Number 2. June, 2018

that porbeagles are ‘large ocean-dwelling fish’ and that ‘intriguing’ means ‘fascinating’, it will be
difficult to comprehend the meaning of this sentence (Mezynski, 1983, p. 261). Though this seems
self-evident, it is still unclear whether such category of words serves comprehension or
percentages of text coverage mentioned before do so.

Researchers using cloze technique reveal the importance of lexical information value. They
have shown that skilled readers are able to understand texts where up to 30% of the words have
been omitted (Nacke, 1970; Martin & Pantalion, 1973). In this regard, Mezynski (1983) states that
“the important factor in these deletion studies is the information level of the words. The absence
of low information words does not hinder comprehension considerably” (p. 261). Similarly,
Stratton and Nacke (1974) indicate that “readers more frequently encounter problems with
inadequate vocabulary when a word of high information value is unfamiliar” (p.186 in Mezynski,
1983, p. 261).

To examine the effect of word frequency on reading comprehension, Marks, Doctorow,

and Wittrock (1974) conducted a reading experiment wherein 15% of lexical items in five texts
were substituted into low and high frequency levels. T-test analysis results showed that when
learners were exposed to ‘hard’ versions of the texts followed by ‘easy’ ones, comprehension was
remarkably low at the beginning but improved later. However, when ‘easy’ versions were given
first, there was no difference in comprehension. In this study, one would draw two conclusions:
learners seem to learn low-frequency words meaning from their high frequent synonyms when
easy versions of texts are administered first, and low-frequency words seem to deteriorate
comprehension performance when no prior simplification was provided.

Anderson and Freebody (1981) reported two experiments of reading using sixth-grade
pupils. The first experiment controlled the number of unknown words, and the second one varied
the importance level of unfamiliar items with regard to the passage. The post-reading tasks
required were free recall, writing a summary statement, and sentence recognition (i.e. deciding
whether some ideas belong to the texts). Findings indicated in the first experiment that
comprehension was not highly affected when only one out of six words (1/6) was replaced with a
low-frequency synonym. However, when one out of three (1/3) words was substituted,
comprehension declined significantly in the sentence recognition task especially. This means that
in 1/3 there are two chances of encountering an unfamiliar word in a six-word string. In other
words, the density of unfamiliar words influences comprehension. One question to raise
concerning these experiments is whether the substituted words are key items in the texts. The
variance of word information value is another crucial point to highlight in the second experiment.
The effect of unfamiliar words with high information level is unclear since they are altered by low
information ones. Thus, the variety of these interactions may cause some confusion in the results.
For this reason, interpretations of the second findings should be cautious.

Measuring Text Difficulty with Lexical Complexity

As mentioned earlier, a large number of readability formulas have been designed to
estimate text complexity. Several criteria have been employed in these measures causing different
Arab World English Journal 238
www.awej.org
ISSN: 2229-9327
 Arab World English Journal (AWEJ) Volume 9. Number 2. June, 2018

interactions between aspects. The latter makes it difficult to know which element determines text
complexity most. None of the formulas, however, focuses exclusively on the effect of frequency
on reading comprehension (Chen & Meurers, 2016).

There is a consensus among specialists in reading research that lexical coverage and
vocabulary knowledge are strong predictors of reading comprehension (Laufer, 1992; Qian, 1998,
2002; Nation, 2001, 2006). A reader’s vocabulary knowledge is highly attributed to the amount of
lexical input to which they have been exposed (Chen et al., 2016). This is referred to as ‘frequency
effect’ (Perfetti et al., 2002; Chen et al., 2016, 2017). Ryder and Slater (1988) have confirmed that
frequency predicts word difficulty. It seems thus that high-frequency words are perceived more
quickly than low frequency ones (Rayner & Duffy, 1986). Therefore, estimating text difficulty on
a frequency basis is fundamental in reading research.

Frequency effect is discussed in the literature based on a cognitive model. It assumes ‘high
activation’ for frequent words in the mental lexicon while a reader is processing written texts (Just
& Carpenter, 1980). This goes in line with the ‘Lexical Quality Hypothesis’ (LQH) initially labeled
as ‘Verbal Efficiency Theory’ (Perfetti, 1985) and later developed by Perfetti and Hart (2002).
LQH suggests that words with high representation are likely to be ‘effortlessly’ accessed by
readers. This means that a word form and meaning are strongly linked to one another and can be
retrieved whenever the item is encountered in reading. Moving beyond this basic notion, Ellis
(2012) states “frequency distribution of input is a key determinant of language acquisition, with
regularities emerging through the learner’s exposure to the distributional characteristics of the
language input” (p.85). Hence, estimating text complexity in terms of word difficulty provides
insightful considerations to EFL acquisition.

While findings reviewed here suggest the crucial role of low-frequency words in reading
comprehension, it is unclear what type of information level these words have. This explains the
choice of the present study to substitute high information value words by low-frequent synonyms.
Furthermore, estimating the whole text difficulty with frequency measure highlights the role of
high information value words in comprehension, and puts into question text coverage of 95 and
98% suggested by Laufer (1989) and Hu & Nation (2000) respectively as mentioned earlier.

The Current Study

As mentioned earlier, the present study aims to investigate the relationship between lexical
frequency, reading comprehension, and recall. Approximately, 18% of the most important words
in one of two texts used in the study have been substituted by low frequent synonyms to examine
the extent to which a small proportion of unknown words may influence participants’ reading
comprehension (Mezynski, 1983) and recall (Anderson & Freebody, 1981). Text difficulty is thus
measured in terms of word frequency to highlight the importance of lexical frequency model in
estimating readability at EFL university level (Chen & Meurers, 2016). Frequency here is mainly
characterized by the semantic variable, referring to readers’ ability to relate text words to their pre-
existing vocabulary knowledge (Marks et al., 1974; Ryder & Slater, 1988). Semantic variable of
word frequency has been found to account for the variances in reading comprehension (Marks et
Arab World English Journal 239
www.awej.org
ISSN: 2229-9327
 Arab World English Journal (AWEJ) Volume 9. Number 2. June, 2018

al., 1974). Also, based on the Lexical Quality Hypothesis (Perfetti & Hart, 2002), this investigation
examines the relationship between vocabulary size, depth, and reading comprehension.

Methods
Participants
Eighty EFL third-year university students from Mohammed V University-Rabat, Morocco
participated in this study. The sample consisted of 47 females (58.75%) and 33 males (41.25%)
with an age range between 20 and 27 years.

One single university level has been involved in this study for two reasons. First and
foremost, third-year students are assumed to constitute the ideal target population for the present
research given the considerable amount of lexical knowledge they have accumulated thanks to the
language input they have been exposed to during their EFL learning experience. Second, because
of the study complexity, it has been judged preferable to limit its scope to one educational level.

Measures
Data has been collected based on two lexical tests and one reading comprehension test.
Each of these instruments is discussed in the following sections.

Vocabulary Size Test (VST) designed by Nation and Beglar (2007) is used to examine
students’ vocabulary size. It has a high Cronbach’s alpha reliability of 0.92, based on the 10th 1000-
word families used. As mentioned before, the last four levels were removed due to their difficulty
level for the study sample. Besides, VST estimates mainly acquired vocabulary. In each item, the
test provides non-contextualized sentences and closely related options in meaning. Thus, informed
guessing is difficult to make (Nation, 2012). In scoring, 1 point is worth every correct answer and
the total score is multiplied by 100.

Word Associates Test (WAT) developed by Read (1993) is used to measure students’
vocabulary depth. It examines how well learners know vocabulary in three word aspects:
synonymy, polysemy, and collocations. Forty adjectives are tested in this measure with an indirect
insight into nouns. In scoring, each correct response is worth 1 point with originally a 160
maximum possible score. After piloting the test, however, four difficult items were removed.
Therefore, the actual maximum possible score is 144 instead of 160 points. Moreover, WAT has
a high Cronbach’s alpha reliability of 0.96, which confirms previous studies’ reliability estimate
(e.g. Read, 1993, 1995).

Reading Comprehension (RC) is a standardized multiple choice question test used to

examine basic comprehension. It is drawn from ‘Scholastic Assessment Test’ (SAT) in ‘Critical
Reading for SAT’ testing manual (Paul & Roger, 2008, pp. 15-23). RC section in SAT examines
academic English for students willing to pursue their studies in US College. The main reason for
using this test as an RC estimate is the similarity in terms of length and topic of the two passages
it includes. Both are around 552 tokens in length and related to politics. Another reason is that its
comprehension questions do not relate to one another, revealing students’ actual understanding of
Arab World English Journal 240
www.awej.org
ISSN: 2229-9327
 Arab World English Journal (AWEJ) Volume 9. Number 2. June, 2018

the passages. To investigate the effect of a small percent of unfamiliar words on comprehension,
only the first passage (hereafter RC1) was adjusted. Hence, 18% of its high information value
words were replaced with their low frequent synonyms. The objective was to examine the extent
to which a low percent of this type of words may influence comprehension and thus estimate text
complexity in terms of lexical difficulty, and foreground the importance of teaching this lexical
category.

The two passages include multiple-choice questions that are straightforward and literal
with a maximum possible score of 17 (nine questions for RC1 and eight for RC2). Indeed, there
was a need to reduce RC length due to the large number of items in the lexical tests in order to
lessen examinees’ fatigue and maximize their motivation to answer the three tests items
completely. Finally, like the vocabulary tests, RC measure has a high parallel-form reliability of
0.81, indicating estimation of the same skills: text word meaning recognition and comprehension.
Recall is the second sub-section of RC test wherein students are required to write the ideas
they retain from each passage to assess their comprehension. Texts analysis into different levels
of idea units was based on Zerhouni’s hierarchy (1998). Thus, at the highest level are main ideas
of the text (allotted 2 points), followed by secondary ideas directly supporting them (allotted 1.5);
then tertiary ideas of less importance (allotted 1); and at the last are quaternary ideas which
constitute minor ideas (i.e. details) (allotted 0.5) (Zerhouni, 1998, p. 66). This classification helps
to determine which type of ideas students with different lexical knowledge understand and recall.
In statistical analysis of results, recall of RC1 and RC2 were compared to one another and correlated
with vocabulary size and depth.

Frequency was estimated to determine the lexical difficulty of both reading passages.
Following Chen et al (2016) method, the present study measured the mean and standard deviation
of the word-frequency levels in each passage. Prior to this are two steps to report here. First, words
were tokenized wherein all function items and proper nouns were eliminated because they are
highly frequent and carry less meaning. Passages were then manually entered into software called
range (available in Nation’s official website) to supply the number of word tokens and types. This
operation revealed that RC1 counted 130-word types and RC2 145. Word type was used as a basis
for this analysis because it provides a better estimate of text difficulty than that of word token
(Chen & Meurers, 2017). The second step consisted in matching word types of the passages with
the stratified frequency lists of headwords developed from British Nation Corpus (BNC) and
Corpus of Contemporary American English (COCA) (Nation, 2012). Indeed, using two corpora
that cover a wide range of written as well as spoken lexis in natural language learning enables a
better estimate of word frequency (Chen et al., 2016) and receptive lexical knowledge of learners.

Data on frequency estimate is ordinal (i.e. from the highest to the lowest levels). Sums of
lexical items in each frequency level were counted and divided by the total number of word types.
For instance, in RC1 the percentage of words in the 2nd 1000 frequency level is 15⁄130=
11.54% whereas in RC2 it is 34⁄145= 23.45%. Figure 1 presents graphically the percentages of

Arab World English Journal 241

www.awej.org
ISSN: 2229-9327
 Arab World English Journal (AWEJ) Volume 9. Number 2. June, 2018

word types across the frequency spectrum. It shows that the second passage contains a large
number of high-frequency words and low percentages of low frequency words starting from the
fourth frequency level, RC2 (M= 1.64, SD= 1.12). The first passage, on the other hand, consists of
more low frequency words, RC1 (M= 1.98, SD= 1.84). The means and standard deviations
presented here reveal that RC1 is lexically more difficult than RC2.

Figure 1. Word frequency percentages of passage 1& 2

Procedure
Test administration took place before the final exams of the academic year (March 2016).
Prior to testing, professors were briefed about the objectives of the study and the instruments used
to collect data. Students were informed then about the reasons of the researcher’s visit and the
importance of participating in this study to contribute to research. Students willingly accepted to
take the tests for their curiosity to find out the quantity and quality of their vocabulary knowledge.
Besides, the testing maximum time was two hours, and the number of students in every session -
ranging from five to 17- was manageable for supervision with a total of six sessions. All tests were
in paper-and-pencil format and students were allowed to take five minutes break in between tests.

Data Analyses
Data collected from the three instruments, namely VST, WAT, and RC is interval in nature.
This is because the present study attempts to look into the effect and type of relationships among
variables. Data was analyzed through SPSS 22 software. Both descriptive and inferential statistics
were run. Inferential statistics include paired samples t-test to check the significance level of the
effect of lexical difficulty on comprehension and recall, and Pearson product-moment correlation
to estimate the relationship between scores on vocabulary size, depth, comprehension, and recall.
For the purposes of the present study, scores of the three tests were converted into 100 to be evenly
analyzed.

Arab World English Journal 242

www.awej.org
ISSN: 2229-9327
 Arab World English Journal (AWEJ) Volume 9. Number 2. June, 2018

Results
Table 1 shows the descriptive statistics of the variables. The first row concerns the
vocabulary size test (VST); the second row is devoted to vocabulary depth (WAT); the third and
fourth rows (RC1 and RC2) refer to the first and the second sub-measures of reading comprehension
respectively; and the fifth and sixth rows are devoted to recall of passage 1 and 2.
Table 1. Means, standard deviations, minimum and maximum of all research variables (N=80)

Minimum Maximum Mean SD

VST 30.00 93.00 62.88 16.01

WAT 29.86 88.89 61.97 14.50
RC1 11.11 67.67 39.32 15.85
RC2 12.50 100.00 59.53 19.30
Recall 1 10.00 80.00 40.50 17.56
Recall 2 .00 90.00 45.75 20.17

As a reminder, the first objective of the study is to investigate whether lexical difficulty, as
estimated by frequency, has an effect on reading comprehension and recall. Frequency variable is
analyzed here from the semantic perspective since it has a direct impact on reading comprehension
(Marks et al., 1974). As shown in descriptive statistics (Table 1), the mean of RC1 is lower than
that of RC2, indicating a low performance in the first passage and higher scores in the second one.
Similarly, the mean of the first passage recall is lower than that of the second passage. However,
it is unclear at this stage whether the difference is statistically significant. Table 2 shows paired
samples t-test results where scores on RC1 and RC2 as well as the first and the second recalls are
contrasted.
The second objective is to estimate the relationship between scores on vocabulary size,
depth, comprehension, and recall. Pearson product-moment correlation was used to measure these
relationships, with vocabulary knowledge (i.e. size and depth) being the independent variable and
comprehension and recall the dependent variables. As shown by research (e.g. Qian, 1998, 2002),
vocabulary knowledge has a close relationship with reading, however, it is unclear which aspect
of lexical knowledge is more required to ensure comprehension. Analyzing the variables’
connections helps also to determine the lexical threshold Moroccan EFL BA (Bachelor of Arts)
learners reach. In fact, this paper focuses on basic comprehension to foreground previous findings
relative to this question. Table 3 presents the results concerning this objective.

Arab World English Journal 243

www.awej.org
ISSN: 2229-9327
 Arab World English Journal (AWEJ) Volume 9. Number 2. June, 2018

Table 2. Paired samples t-test of RC and recall (N=80)

M t df Sig (1-tailed)

RC1- RC2 -20.20160 -12.892 79 .001

Recall 1- Recall -5.25000 -2.651 79 .01
2

Significance level p< 0.01 (one-tailed)

Regarding the first objective of the study, a one-tailed paired samples t-test revealed that
students scored less in the first sub-test of reading comprehension (RC1) (M = 39.32, SD= 15.85)
compared to their performance in RC2 (M= 59.53, SD= 19.30), t(79)= -12.892, p<.001. Results, in
Table 2, showed a statistically significant difference between participants’ performance in RC 1
and RC2. As mentioned earlier, RC test has a high reliability of 0.81 using parallel form analysis
to account for the similarity of the skills estimated in both sub-tests. The two passages differ only
in terms of the difficulty created in RC1 through the substitution of a small percent of frequent
lexis of high information level by less frequent synonyms compared to the high frequencies of key
words in RC2. Measurement of the overall lexical difficulty of the passages shows that frequency
mean and standard deviation of (RC1) is (M= 1.98, SD= 1.84), whereas that of RC2 is (M= 1.64,
SD= 1.12). This indicates that RC1 is more demanding than RC2 given the lexical difficulty of the
first passage. Stratifying frequency levels of the words into bands (i.e. high, mid, and low) gives
an insightful description (Schmitt, 2014). It confirms that the first passage has more mid-low
frequency words than the second passage.
As also revealed by t-test results of free recall task (Table 2), participants recalled fewer
ideas from RC1 (M= 40.50, SD=17.56) compared to RC2 (M= 45.75, SD=20.17), t(79)= -2.651,
p<.01. RC1 recall protocols included minor details but fewer major points due to the difficult
words substituted for the passage high information level words which, most probably, hindered
comprehension.
Table 3. Correlations between VST, WAT, RC1&2, and recall 1&2 (N=80)
Test VST WAT
VST
WAT .677**
RC1 .717** .440**
RC2 .594** .384**
Recall 1 .499** .506**
Recall 2 .590** .568**
**
Significance level p< 0.01 (one-tailed)

Arab World English Journal 244

www.awej.org
ISSN: 2229-9327
 Arab World English Journal (AWEJ) Volume 9. Number 2. June, 2018

Concerning the second objective, Table 3 shows high, moderate to low correlations among
variables. To begin with the results of lexical knowledge, tested through vocabulary size (VST)
(M=62.88, SD=16.01) and vocabulary depth (WAT) (M=61.97, SD=14.50), Pearson’s r data
analysis revealed a significantly moderate positive correlation, r= .677, p< .01. That is, students
who scored high in vocabulary size had somewhat high scores in depth and vice versa. As for the
relationship between VST and RC1 (M=39.32, SD= 15.85), the correlation coefficient was
significantly high, r= .717, p< .01, whereas it was remarkably low, r= .440, p<.01 for the WAT -
RC1 relationship. Lower correlation coefficients were also found between the two lexical aspects
and the second reading sub-measure. VST and RC2 (M= 59.53, SD= 19.30) had a moderate positive
correlation, r= .594, whereas relationship between WAT and RC2 revealed a significantly low
correlation, r= .384. These results show that VST correlates higher with RC1 and RC2 whereas
WAT lags behind considerably, and that performance in RC1 required more lexical competence
compared to RC2.
As exhibited by table 3, recall had significantly moderate to low correlations with VST and
WAT. A Pearson product-moment r analysis revealed a low positive correlation between VST and
recall 1 (M= 40.50, SD=17.56), r= .499, p< .01. Interestingly, the correlation between WAT and
RC1 recall was higher, with a coefficient of r= .506. On the other hand, scores on RC2 recall (M=
45.75, SD= 20.17) correlated moderately with VST and WAT, r= .590, r= .568, p<.01 respectively.
To sum up, performance in recall can be ascribed to the degree to which the lexicon used obscures
the passage meanings.
Discussion
The present study explored the effect of lexical difficulty estimated by frequency (Chen &
Meurers, 2016) on reading comprehension and recall. It also investigated the relationship between
vocabulary size, depth, reading comprehension, and recall.

For the first objective, the level of lexical complexity seems to have an effect on reading
comprehension and recall. Thus, lexical difficulty hinders both comprehension and recall. Table 2
shows a statistically significant difference between scores obtained in students’ performance in
passage 1 as compared to passage 2. These results corroborate those of previous studies, which
came to the same conclusion (Marks et al., 1974; Anderson et al., 1981). Consequently, the lexical
frequency model is confirmed to estimate efficiently text difficulty (Chen et al., 2016, 2017).
Unlike Anderson et al. (1981) findings, however, free recall seems to decrease in the lexically
difficult passage. This discrepancy is probably due to the variety of information level of substituted
words in their experiments.
Thus, the lexicality of a text is fundamental to comprehension, particularly if the proportion
of unknown words is of high information value (Stratton & Nacke, 1974; Mezynski, 1983). For
this reason, comprehension performance declined in the first passage while it was higher in the
second one as table 1 shows. More importantly, skilled readers are known for having high
representations of lexis in their mental lexicon (Klare, 1968). These representations are probably
the different aspects of words that can be epitomized in high and low frequently used synonyms,
as it can be exemplified in word use and its different meanings. Hence, students with poor or low
Arab World English Journal 245
www.awej.org
ISSN: 2229-9327
 Arab World English Journal (AWEJ) Volume 9. Number 2. June, 2018

lexical representations do not exhibit a good control of comprehending texts while reading.
Vocabulary learning, therefore, seems to be learners’ burden (Nation, personal communication)
since it highly influences other language skills such as reading. However, if learning is generally
fine-tuned to students’ levels and needs, they can achieve better results. For example, estimating
the difficulty of academic materials might be the initial solution to resort to at the EFL
undergraduate levels. This is not to be considered as an extra load of work for teachers, but a
necessary step for the reform of university educational system. Hence, assembling a corpus of
estimated reading material can be an effective method to start with.
Recall of the two passages has further confirmed participants’ level of understanding and
retrieving ideas from the reading material. Thus, students scoring high in vocabulary size and depth
recalled more ideas than those with lower scores in the lexical tests. As a reminder, ideas in the
recall task have been categorized according to Zerhouni’s hierarchy (1998, p.66) (see methods).
The order of these ideas is crucial as it demonstrates whether readers have grasped the passage
fully or only partially. Therefore, recall is a crucial technique for evaluating comprehension and
retrieval of text content. It seems that the more ideas students recall from a passage, the more
cognitive effort they make in deciphering word meaning (Anderson & Freebody, 1981).
As for the second objective of the study, it has been shown that vocabulary knowledge has
a close relationship with comprehension and recall. This confirms results from previous studies
(Qian, 1998, 2002; Zhang, 2012; Horiba, 2012). Interestingly, vocabulary size correlates higher
with reading comprehension in both passages compared to vocabulary depth. The latter, however,
correlates highly with the recall task. This may indicate that for basic reading comprehension,
vocabulary size is more of a powerful lexical component. Qian’s (1998, 2002) results may refute
this claim since depth was found to be a stronger indicator of reading in his studies. However, one
should note that Qian’s findings might be due to the effect of depth adjustments. To explain, this
researcher substituted 10 words from vocabulary depth test (Read, 1993) with words and their
stimuli from the RC test he employed in his investigation. In this case, one cannot disregard the
interwoven effect of the two tests ten identical items on students’ answer. If lexical items happen
to appear twice in different measures, students are likely to learn them and be able to answer
correctly. Above all, this may confirm the idea of “pressure point”, mentioned earlier, in reading
comprehension (Perfetti, 1985; Perfetti et al., 2002; Perfetti, 2007). Though different reading tasks
(i.e. comprehension or recall) may necessitate different lexical knowledge, the latter remains a
pressure point between word recognition and text comprehension. It can also be interpreted as the
transactional phase between the meaning of the text and the meaning understood by the reader
through associating text words with his/her mental lexicon (Marks et al, 1974; Ryder & Slater,
1988).
Indeed, qualitative assessment of texts is subjective as it may differ from one reader to
another. It is based on an impressionistic view mainly. The conceptual shift from word to text
meaning in comprehension requires a quantitative evaluation of text lexis. This is because words
can be tokenized and analyzed objectively. As research studies reported above have shown, lexicon
is a crucial component for understanding and conveying verbal ideas. Though such a numerical
analysis seems to be valid in basic comprehension, it may not be sufficient in other comprehension
Arab World English Journal 246
www.awej.org
ISSN: 2229-9327
 Arab World English Journal (AWEJ) Volume 9. Number 2. June, 2018

levels (critical reading especially). The reason is that further factors such as the writer’s intentions
and the socio-cultural context of text events have to be taken into consideration in assessing
reading materials level of difficulty in congruence with learners’ reading proficiency level. Thus,
demanding comprehension levels necessitate complex analyses in order to come up with different
assessment techniques for a better evaluation of students’ reading comprehension deficiencies.
Implications and Conclusion
As findings of the present study show, text complexity can be estimated from a lexical
perspective using a frequency model. This indicates the crucial role of vocabulary, especially
semantic frequency, in reading comprehension and recall. It is therefore essential for EFL teachers
to allocate some time to systematically teach and test vocabulary, and to teach words in association
since they are stored in the mental lexicon as a network. Besides, given the semantic load of lexical
items, students should be exposed both to their high and low synonyms. More importantly,
estimating text difficulty for different academic levels is crucial for selecting reading materials
whose degree of lexical difficulty is in accordance with students’ levels. Similarly, it is also
recommended that students applying for English departments sit for a standardized lexical test as
an effective assessment tool for their placement. Another suggestion based on this study findings
is to provide EFL University students with training in reading and vocabulary learning strategies
in order to equip them with helpful tools to overcome anxiety toward reading in general, better
benefit from their reading lessons and assignments, and to systematically expose them to a wide
range of lexis. Additionally, an EFL corpus of estimated reading material should be assembled and
put at the disposal of teachers. Finally, the present investigation was limited to basic
comprehension; thus, measuring the frequency effect on higher levels of comprehension is
recommended to fully grasp this issue.

About the Authors:

Nadia Nouri is a Ph.D. candidate in “Applied Linguistics and TEFL” research unit at the Doctoral
Center of the Faculty of Education, Mohammed V University-Rabat, Morocco. She holds a master
degree in Teaching English as a Foreign Language (TEFL) from Ibn Tofail University, Kenitra-
Morocco, and is currently teaching English at the American Language Center in Rabat and at the
INPT (Posts and Telecommunication National Institute).
ORCid ID: 0000-0001-9775-3102
Pr. Badia Zerhouni (Ph.D. holder from Ottawa University) is an associate professor of Applied
Linguistics at the Faculty of Education, Rabat, Morocco, and former director of the Doctoral
Center of the Faculty of Education. She currently teaches and supervises research projects in MA
Applied Linguistics /ELT programs and supervises doctoral research in different subjects related
to AL/ELT. Her research interests include reading, writing, and vocabulary acquisition in EFL
contexts.
Acknowledgment
We would like to thank Dr. Paul Nation for his help in the present project, which is part of
a doctoral dissertation.

Arab World English Journal 247

www.awej.org
ISSN: 2229-9327
 Arab World English Journal (AWEJ) Volume 9. Number 2. June, 2018

References
Adams, M., & Bruce, B. C. (1980). Background Knowledge and Reading Comprehension.
Retrieved from
https://www.researchgate.net/publication/234650027_Background_Knowledge_and_Rea
ding_Comprehension_Reading_Education_Report_No_13
Alderson, J. C. (2000). Assessing Reading. Cambridge: Cambridge University Press.
Anderson, R. C., & Freebody, P. (1979). Vocabulary Knowledge and Reading. Retrieved from
https://www.ideals.illinois.edu/bitstream/handle/2142/17498/ctrstreadeducrepv01979i000
11_opt.pdf?sequence=1
Anderson, R. C., & Freebody, P. (1981). Vocabulary Knowledge. In J. T. Guthrie (Ed.),
Comprehension and Teaching: Research Reviews (pp. 77-117). Newark, DE: International
Reading Association.
Bernhardt, E. B. (2005). Progress and Procrastination in Second Language Reading, Annual
Review of Applied Linguistics, 25, 133-150.
Chen, X., & Meurers, D. (2016). Characterizing Text Difficulty with Word Frequencies.
Proceedings of the Eleventh Workshop on Innovative Use of NLP for Building Educational
Applications. San Diego: California.
Chen, X., & Meurers, D. (2017). Word Frequency and Readability: Predicting the Text-Level
Readability with a Lexical-Level Attribute, Journal of Research in Reading, 1–25.
Ellis, N. C. (2012). Frequency-based Accounts of Second Language Acquisition. In S. M. Gass &
A. Mackey (Eds.), Handbook of Second Language Acquisition (pp. 193-210). USA:
Routledge.
Henriksen, B. (1999). Three Dimensions of Vocabulary Development, Studies in Second
Language Acquisition, 21, 303-317.
Horiba, Y. (2012). Word Knowledge and its Relation to Text Comprehension: A Comparative
Study of Chinese- and Korean-Speaking L2 Learners and L1 Speakers of Japanese, Modern
Language Journal, 96, 108–121.
Hu, M., & Nation, I. S. P. (2000). Unknown Vocabulary Density and Reading Comprehension,
Reading in a Foreign Language, 13, (1), 403−430.
Just, M. A., & Carpenter, P. A. (1980). A Theory of Reading: From Eye Fixations to
Comprehension, Psychological Review, 87, (4), 329–354.
Klare, G. R. (1968). The Role of Word Frequency in Readability, Elementary English, 45, (1), 12-
22.
Koda, K. (2005). Insights into Second Language Reading: A Cross-Linguistic Approach, Journal
for Language Learning, 24, (2), 88-92.
Laufer, B. (1989). What Percentage of Text-Lexis is Essential for Comprehension? In L. Ch & M.
Nordman. (Eds.), Special Language: From Humans Thinking to Thinking Machines (pp.
316-323). Multilingual Matters.
Laufer, B. (1992). How much is Lexis Necessary for Reading Comprehension? In H. Bejoint & P.
Arnaud (Eds.) Vocabulary and Applied Linguistics (pp. 126–132). Macmillan, Basingstoke
& London.
Laufer, B., & Goldstein, Z. (2004). Testing Vocabulary Knowledge: Size, Strength, and Computer
Adaptiveness, Language Learning, 54, 399-436.
Arab World English Journal 248
www.awej.org
ISSN: 2229-9327
 Arab World English Journal (AWEJ) Volume 9. Number 2. June, 2018

Lexile. (2007). The Lexile Framework for Reading: Theoretical Framework and Development.
Technical Report, MetaMetrics, Inc., Durham, NC.
Lively, B. A., & Pressey, S. L. (1923). A Method for Measuring the Vocabulary Burden of
Textbooks. In W. H. DuBay, (Ed), the Classic Readability Studies, (pp. 6-16). BookSurge
Publishing.
Marks, C. B., Doctorow, M. J., & Wittrock, N. C. (1974). Word Frequency and Reading
Comprehension, Journal of Educational Research, 67, 259-262.
Martin, C. J., & Pantalion, C. A. J. (1973). Effect of two Word Deletion Schemes upon the
Comprehension of Five Levels of Telegraphic Prose, The Annual Meeting of the American
Educational Research Association. New Orleans.
McNamara, D. S., Graesser, A. C., McCarthy, P. M., & Cai, Z. (2014). Automated Evaluation of
Text and Discourse with Coh-Metrix. New York, NY: Cambridge University Press.
Mezynski, K. (1983). Issues Concerning the Acquisition of Knowledge: Effects of Vocabulary
Training on Reading Comprehension, Review of Educational Research, 53, 253-279.
Milone, M., & Biemiller, A. (2014). The Development of ATOS: The Renaissance Readability
Formula. Technical Report, Renaissance Learning, Wisconsin Rapids.
Milton, J. (2009). Measuring Second Language Vocabulary Acquisition. Multilingual Matters.
Nacke, P. L. (1970). Skimming Strategy in Reading as a Function of Familiarity with Content and
Redundancy Reduction in Printed Discourse. (doctoral dissertation). Retrieved from
https://open.library.ubc.ca
Nation, I. S. P. (2001). How Many High-Frequency Words are there in English? In M. Gill, A. W.
Johnson, L. M. Koski, R. D. Sell & B. Warvik (Eds.), Language, Learning, and Literature:
Studies Presented to Hakan Ringbom English Department Publications (pp.167-181) Abo
Akademi University, Abo.
Nation, I. S. P. (2006). How Large is a Vocabulary Needed for Reading and Listening? Canadian
Modern Language Review, 63, (1), 59-82.
Nation I. S. P., & Beglar, D. (2007). A Vocabulary Size Test, the Language Teacher, 31, (7), 9-
13.
Nation, I. S. P. (2012). The Vocabulary Size Test. Retrieved from
https://www.victoria.ac.nz/lals/about/staff/publications/paul-nation/Vocabulary-Size-
Test-information-and-specifications.pdf
Nelson, J., Perfetti, C., Liben, D., & Liben, M. (2012). Measures of Text Difficulty: Testing their
Predictive Value for Grade Levels and Student Performance. Technical Report, the Gates
Foundation.
Ojemann, R. J. (1934). The Reading Ability of Parents and Factors Associated with Reading
Difficulty of Parent Education Materials. In W. H. DuBay, (Ed), the Classic Readability
Studies (pp. 27-30). BookSurge Publishing.
Paul, L., & Roger S. W. (Eds). (2008). Critical Reading for SAT. USA: Peterson.
Patty, W. W., & Painter, W. I. (1931). A Technique for Measuring the Vocabulary Burden of
Textbooks. In W. H. DuBay, (Ed), the Classic Readability Studies (pp. 31-40). BookSurge
Publishing.
Perfetti, C. A. (1985). Reading Ability. New York, NY: Oxford University Press.

Arab World English Journal 249

www.awej.org
ISSN: 2229-9327
 Arab World English Journal (AWEJ) Volume 9. Number 2. June, 2018

Perfetti, C. A., & Hart, L. (2002). The Lexical Quality Hypothesis. In L. Verhoeven. C. Elbro, &
P. Reitsma (Eds.), Precursors of Functional Literacy (pp. 189-213). Amsterdam: John
Benjamins.
Perfetti, C. A. (2007). Reading Ability: Lexical Quality Hypothesis, Scientific Studies of Reading,
11, (4), 357-383.
Qian, D. (1998). Depth of Vocabulary Knowledge: Assessing its Role in Adults’ Reading
Comprehension in English as a Second Language (Unpublished doctoral dissertation).
University of Toronto, Canada.
Qian, D. (2002). Investigating the Relationship between Vocabulary Knowledge and Academic
Reading Performance: An Assessment Perspective, Language Learning, 52, 513-536.
Rayner, K., & Duffy, S. A. (1986). Lexical Complexity and Fixation Times in Reading: Effects of
Word Frequency, Verb Complexity, and Lexical Ambiguity, Memory & Cognition, 14, (3),
191–201.
Read, J. (1993). The Development of a New Measure of L2 Vocabulary Knowledge, Language
Testing, 10, 355-371.
Read, J. (1995). Validating the Word Associates Format as a Measure of Depth of Vocabulary
Knowledge. Paper Presented at the 17th Language Testing Research Colloquium, Long
Beach, CA.
Read, J. (2000). Assessing Vocabulary. Cambridge: Cambridge University Press.
Read, J. (2004). Plumbing the Depths: How should the Construct of Vocabulary Knowledge be
Defined? In B. Harley, & J. H. Hulstijn (Eds.), Vocabulary in a Second Language:
Selection, Acquisition, and Testing (Vol. 10, pp. 209-227). Amsterdam/Philadelphia: John
Benjamins Publishing Company.
Ryder, R. J., & Slater, W. H. (1988). The Relationship between Word Frequency and Word
Knowledge, Journal of Educational Research, 81, (5), 312-317.
Schmitt, N. (2000). Vocabulary in Language Teaching. Cambridge: Cambridge University Press.
Schmitt, N. (2010). Researching Vocabulary: A Vocabulary Research Manual. Basingstoke:
Palgrave Macmillan.
Schmitt, N. (2014). Size and Depth of Vocabulary Knowledge: What the Research
Shows, Language Learning, 64, (4), 913–951.
Stratton, R. P., & Nacke, P. L. (1974). The Role of Vocabulary Knowledge in Comprehension. In
P. L. Nacke (Ed), 23rd Yearbook of the National Reading Conference, Clemson, S. C:
National Reading Conference.
Stuart, M., & Kramer, B. (2015). The Creation of a New Vocabulary Levels Test, Shiken, 19, (2),
1-11.
Zerhouni, B. (1998). Reading in L1, L2, and FL: A Variable-Treatment Study. (doctoral
dissertation). National Library of Canada, Ottawa.
Zhang, D. (2012). Vocabulary and Grammatical Knowledge in Second Language Reading
Comprehension: A Structural Equation Modeling Study, Modern Language Journal, 96,
558–575.

Arab World English Journal 250

www.awej.org
ISSN: 2229-9327