Coxhead & Boutorwick (2018) Longitudinal Vocabulary Development in An EMI International School Context Learners and

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 23

Longitudinal Vocabulary Development

in an EMI International School Context:


Learners and Texts in EAL, Maths, and
Science
AVERIL COXHEAD AND T. J. BOUTORWICK
Victoria University of Wellington
Wellington, New Zealand

This article focuses on vocabulary in the context of English-medium


instruction at an international high school in Germany. Longitudinal
testing was carried out using the Vocabulary Levels Test (VLT)
(Nation, 1983; Schmitt, Schmitt, & Clapham, 2001), with 468 partici-
pants tested on entry to their first year of secondary school (Grade 6,
aged 10–11 years old) and annually thereafter from 2009 to 2015.
Learners with the lowest scores on the high frequency (first 2,000
level) words of the VLT in Grade 6 were nonnative speakers, so they
were given English as an additional language support. These learners
reached mastery of the first 2,000 of the VLT by Grade 9. Nonnative
speakers with higher scores on the VLT and native speakers had mas-
tery of the 2,000 level from Grade 6, and all three groups scored sim-
ilarly at the 3,000 level and academic sections of the test by Grade
10. A corpus-based lexical analysis of representative textbooks in Eng-
lish as an additional language, maths, and science and learning mate-
rials from Grade 6 demonstrates the importance of high-frequency
vocabulary in these texts, and that learners need a large vocabulary
to deal with these texts, particularly in maths and science.
doi: 10.1002/tesq.450

I nternational schools provide English-medium instruction (EMI) in


predominantly English as a foreign or second language contexts
(EFL/ESL) such as Thailand, Germany, or Turkey. According to the
International School Consultancy, there are over four million students
in 8,000 such schools worldwide (Sharma, 2016). International Bac-
calaureate accredited schools number around 4,000 (International
Baccalaureate Organization, 2017), and international education is
expanding (Smith, 2015). In the German context, EMI is growing
rapidly in higher education for historical, global, and economic

588 TESOL QUARTERLY Vol. 52, No. 3, September 2018


© 2018 TESOL International Association
15457249, 2018, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tesq.450 by Desi Arinda - <shibboleth>-member@monash.edu.au , Wiley Online Library on [20/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
reasons (Earls, 2016). In the midst of this growth are language learn-
ers and content teachers who are working in two or more languages in
EMI schools, with a focus on the content of subjects such as maths
and science. In this context, English is not necessarily the language of
the playground, let alone the wider environment.
The international school in Germany where this research took place
uses authentic materials in English, and a teacher at the school was
concerned about the readiness of the students to read textbooks in
English which had been written with native English readers in mind.
English language proficiency was not assessed until the teacher intro-
duced a range of tests including Schmitt, Schmitt, and Clapham’s
(2001; Nation, 1983) Vocabulary Levels Test (VLT) at the beginning
of the 2009 school year for Grade 6 students (10 and 11 years old).
Nonnative speakers (NNS) with low proficiency scored poorly on the
high-frequency section of the test (the 2,000-word level). High-fre-
quency words are important for all learners of the language (Nation,
2013). If students score poorly on tests of knowledge of high-frequency
words in English, it begs the question of how well they comprehend
the classroom textbooks.
The present study aims to track the vocabulary development of
learners in this EMI school over time and compare these results with
lexical analysis of learning materials and representative textbooks.

Vocabulary Knowledge in EFL/ESL/CLIL/EMI Contexts

To the best of our knowledge, little research has been published on


vocabulary in EMI overall. Exceptions include Malmstr€ om, Pecorari,
and Gustafsson’s (2016) study of productive academic vocabulary in a
corpus of academic writing by MSc students in an EMI context univer-
sity in Sweden. Malmstr€ om et al. (2016) found that Gardner and
Davies’s (2014) Academic Vocabulary List (AVL) accounted for 20%
of the words in their corpus; in addition, they found evidence of sig-
nificant but very modest gains in academic vocabulary knowledge over
time. Such studies would be particularly useful in EMI secondary
school contexts. Two studies have looked at Content and Language
Integrated Learning (CLIL) and vocabulary development in English
over time, whereby one group learns English through CLIL and the
other group learns English through traditional EFL instruction. These
studies report gains in learner vocabulary knowledge for both groups.
Pietil€a and Merikivi (2014) used the VLT with first language (L1) Fin-
nish Grade 6 and Grade 9 CLIL and non-CLIL students. They found
that the non-CLIL students, who scored lower on the VLT than the
CLIL students, caught up with their CLIL counterparts by Grade 9.

LONGITUDINAL VOCABULARY DEVELOPMENT 589


15457249, 2018, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tesq.450 by Desi Arinda - <shibboleth>-member@monash.edu.au , Wiley Online Library on [20/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Gierlinger and Wagner (2016) also report vocabulary gains in a CLIL
vs. non-CLIL study in Austria using Meara and Milton’s X-Lex Test
(2003). Although the CLIL group outscored the non-CLIL group in
Gierlinger and Wagner’s (2016) study overall, the non-CLIL group
showed stronger relative gains in vocabulary than the CLIL group.
These studies suggest that all their participants increased their vocabu-
lary knowledge in English throughout their studies, whether they were
involved in a CLIL-based approach or not. What gains might be made
in an EMI secondary school context, with populations of mixed lan-
guage origins, warrants investigation.
Looking beyond the CLIL/EMI context, research has shown relatively
low levels of vocabulary knowledge for EFL high school students in
Indonesia (Nurweni & Read, 1999; Read, 2000), and high school and
university level students in Denmark (Henriksen & Danelund, 2015),
Taiwan (Webb & Chang, 2012), and Spain (Olmos, 2009; Perez, 2005
cited in Olmos, 2009). Tracking students’ vocabulary knowledge over
time would be helpful to learners, teachers, and researchers in many
contexts. One study which used the VLT for longitudinal purposes
involved testing the vocabulary knowledge of 166 EFL learners in Tai-
wan, from 15–16 to 20–21 years old (Webb & Chang, 2012). The Tai-
wanese participants showed low mastery of VLT levels even after 5 years
of instruction in English. Only 78 out of the 166 participants achieved
mastery on the 1,000-word level of the test and 27 out of 166 achieved
mastery on the 2,000-word level. That said, Webb and Chang (2012) sug-
gest that EFL learners could learn up to 400 word families in a year of
study. The present study aims to track the development of vocabulary
for EMI learners in a similar way to Webb and Chang (2012).

Investigating Vocabulary in Learning Materials, Textbooks,


and Teacher Talk at Secondary School Level

Classroom-based learning materials, which are provided by teachers,


and textbooks are key sources of content and language input for stu-
dents. There seems to be little research on actual classroom learning
materials to date, but textbooks have been studied in more depth.
Leung (2014, p. 137) notes that the “language found in published
teaching materials . . . tends to be highly structured in organization,
specialist in register and new/unfamiliar in meaning.” This is an
important point, because these features may affect the ability of learn-
ers to access the content of their textbooks when reading.
Several studies note a strong correlation between vocabulary knowl-
edge and reading comprehension (Laufer, 1989; Schmitt, Jiang, &

590 TESOL QUARTERLY


15457249, 2018, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tesq.450 by Desi Arinda - <shibboleth>-member@monash.edu.au , Wiley Online Library on [20/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Grabe, 2011; van Zeeland & Schmitt, 2013). The figure of knowledge
of 95% of the vocabulary of a text is often quoted as representing rea-
sonable comprehension (Laufer, 1989). That is, if a learner knows
95% of the words in a text, a milestone is reached, and the reader will
likely understand enough. Below this level, there is a risk that compre-
hension will be poor. Learners who know 95% of the words in a text
might require some support with vocabulary while reading a text, for
example, by asking a classmate or a teacher or checking a dictionary
when necessary. Learners who know 98% of the words in a text are
more likely to be able to cope with reading a text on their own.
There is a substantial difference in the number of words which are
required to reach the 95% and 98% thresholds. For 95%, written texts
have been found to require 4,000–5,000 word families (Laufer & Raven-
horst-Kalovski, 2010; Nation, 2006). Science textbooks used in sec-
ondary schools in New Zealand (Coxhead, Stevens, & Tinkle, 2010)
required 9,000 word families plus proper nouns to reach 95% at Grades
8 and 11. To reach 98%, Coxhead et al. (2010) found the Grade 8
science textbook required 11,000 word families plus proper nouns and
more than 15,000 word families plus proper nouns for the Grade 11
science textbook. Nation (2006) found that 8,000–9,000 word families
plus proper nouns were needed to reach 98% for novels, newspapers,
and university-level texts. Coxhead (2012) found the same result for
novels, poetry, and film scripts used in English literature classes in New
Zealand, and a large-scale analysis of middle school textbooks in the
United States (Grades 6–8) (Greene & Coxhead, 2015) using West’s
General Service List (1953) (GSL) and Coxhead’s Academic Word List
(AWL) (2000) found that these lists together did not reach 95%. High-
frequency vocabulary covered around 80% of the textbooks overall,
showing that knowledge of high-frequency words is vital for reading for
learners in middle schools, and for users of these textbooks in other
contexts. Greene and Coxhead (2015) developed middle school sub-
ject-specific word lists based on their corpus, but they did not carry out
an analysis of the 95% or 98% levels of the textbooks.
This review suggests that these measures of lexical coverage of edu-
cational texts show us that we need to know whether there is a gap
between the vocabulary in a text and students’ lexical knowledge. Pre-
vious findings suggest that there is a real risk that students’ vocabulary
sizes are not well matched with the texts they encounter.

Study Context and Research Questions

The above review of research indicates the importance of under-


standing the vocabulary knowledge of learners in an EMI

LONGITUDINAL VOCABULARY DEVELOPMENT 591


15457249, 2018, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tesq.450 by Desi Arinda - <shibboleth>-member@monash.edu.au , Wiley Online Library on [20/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
environment, and that was the objective of the present study. The
research site was a part of the international school education confra-
ternity and caters for students from Grade 1 to Grade 12. The aca-
demic year runs from August to July, but new students may start
school at any time during the school year. The student population was
heterogeneous by nature and fairly typical of international schools.
Students were grouped using their passport data as native speakers
(NS) and nonnative speakers (NNS). Languages at home can differ
from languages of instruction, which makes this method problematic,
but maintaining consistency with school policy was important for eco-
logical validity. Approximately 10% of the student population had
passports from countries where English is a first language, and they
could have been educated in several countries by Grade 6, when they
usually start high school (aged 10–11 years).
The three research questions that frame this study are as follows:
1. How does vocabulary knowledge develop for native and nonna-
tive speakers of English over several years of study in secondary
school in an EMI context?
2. What are the vocabulary profiles of the learning materials and
representative textbooks in English, maths, and science in
English?
3. What coverage does the AWL provide over the learning materi-
als and textbooks?

METHODOLOGY
Participants

A total of 538 students took the VLT. Students who were diagnosed
with suspected learning difficulties (e.g., dyslexia) (N = 67) were
removed from the data set, along with three participants who were
classed as native speakers by the means adopted in this study but who
nevertheless required additional language support. Of the remaining
468 participants, 43% had German nationality, and English-dominant
countries such as the United Kingdom, the United States, Australia,
Canada, and South Africa accounted for 12%, followed by Russia
(5%). Participants came from over 50 different countries.
Participants were grouped into cohorts according to the year they
were enrolled in Grade 6 (see Table 1). Those enrolled in Grade 6 in
2009 belong to the 2009 cohort; students enrolled in Grade 6 in 2010
are in the 2010 cohort. Students who enrolled in Grade 8 in 2011

592 TESOL QUARTERLY


15457249, 2018, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tesq.450 by Desi Arinda - <shibboleth>-member@monash.edu.au , Wiley Online Library on [20/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
TABLE 1
Numbers of NS/NNS/NNSEAL Participants by Cohort

Cohort 2009 2010 2011 2012 2013 2014 2015 Total


NS 3 6 15 7 9 10 9 59
NNS 50 59 41 45 48 37 26 306
NNSEAL 12 5 16 20 18 16 16 103
Total 65 70 72 72 75 63 51 468

belong to the 2009 cohort because they would have been in Grade 6
in 2009. Permission was sought from participants for this study. For
ethical reasons, the teacher involved in this study chose to remain
anonymous and not participate as an author of this article.
Between 2009 and 2014, all classes except French, Spanish, and Ger-
man language classes were taught in English. All participants in the
study had English literature classes 5 days a week, for 45 minutes per
class. The least proficient of the nonnative-speaking students
(NNSEAL) were placed into specifically designed classes to improve
their vocabulary knowledge and other language knowledge and skills.
These classes provided extra English as an additional language (EAL)
support (three periods per week) during the modern foreign lan-
guages time-slot. A bilingual English and German model was intro-
duced in the 2014–2015 academic year, starting in Grade 7, which
affected 73 participants in the 2013 cohort, which included 26 first
language (L1) German and eight L1 English speakers, and 39 speakers
of other first languages. An assessment of the impact of this change is
beyond the scope of this article.
In Table 1, the right-most column represents total n-sizes for NS, NNS,
and NNSEAL. The bottom-most row depicts the total number of students
for each cohort. Some students took the VLT each year, but others did
not (because of absences, for example). Table 2 depicts the number of
students in each grade, for the NS, NNS, and NNSEAL cohorts. Many stu-
dents took the tests multiple times over the years of the study.
Note that in this school, Grades 6, 7, and 8 follow an in-house curricu-
lum which draws on the English Key Stage 2 primary curriculum from

TABLE 2
Number of Students in Each Grade for Each of the Three Groups (NS, NNS, NNSEAL)

Cohort Grade 6 Grade 7 Grade 8 Grade 9 Grade 10 Grade 11


NS 33 32 23 21 18 8
NNS 187 152 140 149 129 66
NNSEAL 66 61 50 40 23 12
Total 286 245 213 210 170 86

LONGITUDINAL VOCABULARY DEVELOPMENT 593


15457249, 2018, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tesq.450 by Desi Arinda - <shibboleth>-member@monash.edu.au , Wiley Online Library on [20/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
the United Kingdom (Department for Education, 2012); Grades 9 and
10 follow the International General Certificate of Education (IGCSE)
(University of Cambridge Local Examinations Syndicate, 2017); and
Grades 11 and 12 follow the International Baccalaureate Diploma Pro-
gram (IBDP) (International Baccalaureate Organization, 2017).

The Vocabulary Levels Test (VLT)


The VLT (Schmitt et al., 2001) is a multiple-choice–style test with
four frequency-based levels—2,000, 3,000, 5,000 and 10,000 word fami-
lies—and an academic vocabulary section based on Coxhead’s (2000)
AWL. In the current study, the 10,000-word level section of the test
was not given to Grade 6 and Grade 7 students in the first two years of
the testing because it was not considered appropriate. Each section
has a total of 30 points. Schmitt et al. (2001) report that a score of 26
indicates mastery of that level. Figure 1 shows an example item from
the 2,000-word level of the test. Read (2000) points out that rather
than the words being the focus of the test, the definitions constitute
the items to be tested since there is no context for the target lexical
items.

Learning materials
The learning materials for this study were developed by Grade 6
teachers in the NNSEAL, maths, and science classes and were gathered
from a week each at the beginning, middle, and end of the academic
year. Table 3 shows the number of texts and total running words in
this small corpus. Figure 2 provides examples from the learning mate-
rials from each subject. Note that each example contains some sort of
problem for the students to solve.

2,000 VLT level

1 birth

2 dust ________ game

3 operation ________ winning

4 row ________ being born

5 sport

6 victory

FIGURE 1. Sample item from the Vocabulary Levels Test.

594 TESOL QUARTERLY


15457249, 2018, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tesq.450 by Desi Arinda - <shibboleth>-member@monash.edu.au , Wiley Online Library on [20/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Subject Example from learning materials
English Spelling
A lot of verbs change when you add -ing to them. Look at the examples below:
come – coming; write – writing; dance – dancing; ride – riding; lie - lying; tie –
tying; die – dying; run – running; sit – sitting; swim – swimming; tap – tapping
What do you think are the rules for adding -ing to a verb?
Maths Fatima has four sisters. Each sister has twelve books. Each book has 235 pages.
How many pages are there altogether in all of the books Fatima and her sisters
have?
Science In the space provided, write true if the sentence is true. Write false if the
sentence is false.
1. Energy makes things move.
2. Only moving things have energy.
3. There is only one form of energy.

FIGURE 2. Samples from the learning materials in English, maths, and science.

TABLE 3
Running Words and Texts of the Learning Materials by Subject

Subject Number of texts Total running words


English 6 845
Maths 2 320
Science 4 1954
Total 12 3119

Textbooks
To represent textbooks from Grades 6, 8, 10 and 11 (Table 4), in
English, The Hunger Games (Grade 6) and Pride and Prejudice (Grade
10) were selected from Coxhead (2012; see also Coxhead & White,
2012); and in science, a Grade 8 textbook and a Grade 11 textbook
were selected from Coxhead et al. (2010). In maths, a Grade 8 and a
Grade 11 textbook from a well known series in New Zealand were
used. This corpus of textbooks was carefully checked and cleaned
using Notepad++ to correct any errors, as recommended by Nation
(2016). Because a corpus of suitable textbooks was already available, it
was thought impractical to extend the scope of this project to include
the analysis of the textbooks actually in use by these students. The

LONGITUDINAL VOCABULARY DEVELOPMENT 595


15457249, 2018, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tesq.450 by Desi Arinda - <shibboleth>-member@monash.edu.au , Wiley Online Library on [20/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
TABLE 4
Textbooks by Subject, Approximate Level in the International School, and Running Words

Subject Grade Texts Running words Totals


English 6 The Hunger Games (Chapters 1 and 2) 24,826 147,642
10 Pride and Prejudice 122,816
Maths 8 Gamma Mathematics 106,005 249,117
11 Delta Mathematics 143,112
Science 8 Pathfinder Year 9 (NZ) 8,829 64,795
11 Pathfinder Year 12 (NZ) 55,966
Total 461,554 461,554

teacher confirmed that these textbooks are typical of the ones used in
the international school.
Similar findings to Nation’s (2006) analysis of novels could be
expected for Pride and Prejudice and two chapters of The Hunger Games
in the present study. Having access to the full text for The Hunger
Games would clearly be preferable, but was not possible for this
analysis.

Procedures

Each cohort was first tested on entry to Grade 6, and then tested
each subsequent year. Therefore, the 2009 cohort was first tested in
2009, again in 2010, 2011, and so on until 2014. The 2010 cohort was
first tested in 2010, 2011, and so on until 2015. The 2009, 2010, and
2011 VLT assessments were carried out on paper. Subsequently, the
VLT was administered in class using an online version of the test. The
teacher supervised and marked the tests. Absent students were usually
given another opportunity to complete the assessment. Students were
asked to skip any words they did not know instead of trying to guess
and had 90 minutes to complete the test. The tests were given to par-
ticipants at the beginning of the academic year in 2009–2011; the tim-
ing of the administration changed to the end of the academic year
from 2012.

Data Analysis
The VLT analysis was carried out using R (R Core Team, 2016). Lin-
ear mixed-effects models were computed to compare the groups and
to control variables including individual differences by including ran-
dom effects in the models, so results obtained from the models are
more easily generalizable to a wider population (Baayen, Davidson, &

596 TESOL QUARTERLY


15457249, 2018, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tesq.450 by Desi Arinda - <shibboleth>-member@monash.edu.au , Wiley Online Library on [20/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Bates, 2008). The dependent variable was VLT score and the indepen-
dent variables were cohort (2009 to 2015), year enrolled (first, second,
third, fourth, fifth), and VLT level (2,000, 3,000, 5,000, 10,000, and
academic).
To determine the best fitting model, backwards elimination was
adopted, meaning that all predictors were entered into the model and
nonsignificant variables were removed one at a time leaving only sig-
nificant predictors. The models were built using the R package lme4
(Linear Mixed Effects version 4_1.1-12; Bates, M€achler, Bolker, &
Walker, 2015). Student was included as a random effect in the model,
and remained in the model only if it significantly contributed to the
predicting power of the model. To determine its suitability, the model
with the random effect was compared to the same model without the
random effect using the anova function in R. The random effect was
retained in the model if including the random effect significantly
improved the model.
The frequency analysis was carried out using the Range program
(Heatley, Nation, & Coxhead, 2002) and Nation’s (2012) British
National Corpus (BNC)/Corpus of Contemporary American English
(COCA) (Davies, 2008) lists of word families (from 1,000 to 25,000),
with supplementary lists of proper nouns (e.g., Europe, David), abbrevi-
ations (e.g., pm, rpm, kwh), compounds (e.g., birthday, airbag), and mar-
ginal words (e.g., ah, mm, X, Y) (Nation, 2016). The version of the
Range program with West’s GSL (1953) first and second 1,000 high-
frequency word families and Coxhead’s (2000) AWL was also used.

RESULTS

Research Question One: How Does Vocabulary Knowledge develop


for Native and Nonnative Speakers of English Over Several Years of
Study in Secondary School in an EMI Context?
Table 5 shows that the native speaker cohorts have mastery (with a
score of at least 26) of the first 2,000 and 3,000 words in Grade 6, the
5,000-word and AWL by Grade 8, and almost mastery of the 10,000-
word level by Grade 10. There could well be a ceiling effect in these
test results early on for the high-frequency levels. The nonnative speak-
ers achieved mastery of the 2,000-word level of the VLT in Grade 6
(Table 6). Mastery of the 3,000-word level follows in Grade 8 and the
5,000-word level and the AWL in Grade 10. The NNS students did not
achieve mastery of the 10,000-word level. The NNSEAL students
achieved mastery of the 2,000-word level by the start of Grade 9
(Table 7), and mastery of the 3,000-word level and AWL sections by

LONGITUDINAL VOCABULARY DEVELOPMENT 597


15457249, 2018, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tesq.450 by Desi Arinda - <shibboleth>-member@monash.edu.au , Wiley Online Library on [20/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
TABLE 5
Average VLT Scores With Standard Deviations for NS

Grade VLT 2,000 VLT 3,000 VLT 5,000 VLT 10,000 VLT AWL
6 28.8 (1.4) 26.9 (3.3) 23.0 (6.2) 10.2 (7.0) 20.9 (5.2)
7 29.2 (1.1) 28.7 (2.2) 25.3 (4.8) 16.6 (7.4) 25.3 (4.4)
8 29.2 (1.1) 28.6 (2.2) 26.9 (3.5) 18.2 (6.5) 27.0 (2.6)
9 29.8 (0.5) 29.5 (0.8) 28.0 (3.0) 21.7 (6.4) 28.3 (2.5)
10 29.9 (0.2) 29.8 (0.4) 29.6 (1.0) 25.0 (4.4) 28.9 (1.4)
11 29.8 (0.5) 29.9 (0.4) 29.4 (0.9) 25.0 (3.7) 29.6 (0.5)

Note. Standard deviations in parentheses.

TABLE 6
Average VLT Scores With Standard Deviations for NNSs

Grade VLT 2,000 VLT 3,000 VLT 5,000 VLT 10,000 VLT AWL
6 26.0 (3.2) 21.2 (5.8) 15.9 (6.3) 6.5 (5.3) 16.1 (6.1)
7 28.0 (2.0) 25.6 (3.7) 20.8 (5.1) 9.7 (5.5) 21.8 (4.6)
8 28.7 (1.3) 27.5 (2.7) 23.3 (4.4) 12.5 (6.0) 24.6 (4.3)
9 29.3 (1.1) 28.1 (2.4) 24.4 (4.5) 14.1 (6.3) 25.7 (3.7)
10 29.4 (1.0) 28.7 (1.9) 26.7 (2.9) 16.2 (6.1) 27.8 (2.2)
11 29.7 (0.8) 29.2 (1.3) 27.5 (2.1) 18.9 (5.6) 28.7 (1.4)

Note. Standard deviations in parentheses.

TABLE 7
Average VLT Scores With Standard Deviations for NNSEALs

Grade VLT 2,000 VLT 3,000 VLT 5,000 VLT 10,000 VLT AWL
6 16.0 (7.2) 10.2 (6.2) 6.8 (5.2) 2.1 (2.5) 7.3 (5.7)
7 23.3 (5.1) 17.0 (6.3) 12.5 (5.2) 4.0 (3.4) 12.8 (7.0)
8 25.2 (4.5) 21.4 (6.2) 15.6 (5.0) 6.2 (4.7) 18.0 (7.0)
9 27.4 (2.9) 22.9 (5.7) 19.1 (6.2) 8.7 (5.6) 22.3 (5.3)
10 28.8 (2.2) 25.7 (4.5) 22.9 (3.8) 10.2 (5.2) 25.8 (2.5)
11 29.6 (0.9) 27.9 (1.9) 25.5 (2.6) 14.4 (4.6) 27.2 (2.6)

Note. Standard deviations in parentheses.

the start of Grade 11. They did not achieve mastery of the 5,000- and
10,000-word levels.
To determine if there were significant trends in the data, a linear
mixed-effects model was fitted to the data with VLT score as the
dependent variable, and group (NS, NNS, and NNSEAL), VLT level,
and grade as independent variables. Two random variables were also
specified in the model: student and cohort.
The results of the model revealed significant (p < .05) three-way
interactions between group, VLT level, and grade level. To determine

598 TESOL QUARTERLY


15457249, 2018, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tesq.450 by Desi Arinda - <shibboleth>-member@monash.edu.au , Wiley Online Library on [20/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
where the differences were, multiple comparisons of means using
Tukey contrasts were computed using the glht function in R, part of
the multcomp package (Hothorn, Bretz, & Westfall, 2008). The results
revealed that all three of the groups improved their scores across each
of the VLT levels up until a certain point, and then they plateaued,
depending on group and level. The NS group was the first to plateau,
followed by the NNS, and finally the NNSEAL group.
The data have shown that learning occurred for all of the groups,
and so it makes sense to look further into these results to determine
where the plateaus occurred for the three groups. When students
entered the school in Grade 6, the NSs and NNSs scored similarly on
the 2,000-word level of the VLT, and both of these groups scored sig-
nificantly higher than the NNSEAL students (NS: z = 15.73, p < .001;
NNS: z = 19.45, p < .001). This pattern repeated through Grade 7 and
8 with the NSs and NNSs scoring significantly higher than the
NNSEAL group. From Grade 9, all three groups scored similarly on
the 2,000-word level (p > .05). Table 8 summarises these developments
of vocabulary at all levels and groups, starting with the 2,000-word level
of the VLT. The table only shows when groups first score at the same
level in the VLT, not for every subsequent time, and was based on
mixed-effects modeling data.
The 3,000-word level data show a slightly different pattern than the
2,000-word level. At Grade 6, the NS had significantly greater knowl-
edge of the 3,000-word level than both NNS (z = 6.47, p < .001) and
NNSEAL students (z = 20.12, p < .001). This pattern repeated in Grade
7. In Grade 8, the NS and NNS scored similarly, and both groups
scored significantly higher than the NNSEAL group (NS: z = 7.94, p <
.001; NNS: z = 9.54, p < .001). This pattern repeats until Grade 10,
when all three groups scored similarly (p > .05).
The 5,000-word level results revealed that in Grade 6 the NS scored
significantly higher than the NNS (z = 8.27, p < .001) and NNSEAL (z
= 19.49, p < .001), and the NNS scored significantly higher than the
NNSEAL (z = 17.86, p < .001). This pattern repeats until Grade 9,

TABLE 8
Levels of the VLT Where the NS/NNS/NNSEAL Groups First Scored Similarly

VLT Level Grade 6 Grade 7 Grade 8 Grade 9 Grade 10 Grade 11


2,000 NS/NNS All three
groups
3,000 NS/NNS All three groups
5,000 NS/NNS NS/NNS & All three
NNS/NNSEAL groups
AWL NS/NNS All three groups
10,000 NNS/NNSEAL

LONGITUDINAL VOCABULARY DEVELOPMENT 599


15457249, 2018, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tesq.450 by Desi Arinda - <shibboleth>-member@monash.edu.au , Wiley Online Library on [20/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
when the NS and NNS scored similarly (z = 2.5, p = .52), and both NS
and NNS groups scored significantly higher than the NNSEAL group
(NS: z = 7.25, p < .001; NNS: z = 7.57, p < .001). In Grade 10, the NS
and NNS scored similarly (z = 1.33, p = 1), the NNS and NNSEAL
group scored similarly (z = 3.31, z = .07), and the NS group scored sig-
nificantly higher than the NNSEAL group (z = 3.45, p = .04). In Grade
11, all three groups scored similarly. Note that two (or more) groups
being within statistical significance and mastery of a VLT level are not
the same. It is possible for two groups to have statistically similar scores
and one of the groups to have an average score of less than 26.
The 10,000-word level results showed that in Grade 6 the NS scored
significantly higher than the NNS (z = 4.60, p < .001) and NNSEAL (z
= 10.08, p < .001) groups, and the NNS scored significantly higher
than the NNSEAL group (z = 8.63, p < .001). This pattern continues
until Grade 11, when the NNS and NNSEAL groups scored similarly (z
= 2.97, p = .18). The NS group scored significantly higher than the
NNS (z = 3.45, p = .04) and NNSEAL (z = 4.87, p < .001) groups.
The academic level results revealed that in Grade 6 the NS group
scored significantly higher than the NNS (z = 5.53, p < .001) and
NNSEAL (z = 16.7, p < .001) groups, while the NNS scored signifi-
cantly higher than the NNSEAL group (z = 17.29, p < .001). In Grade
8, the NS and NNS scored similarly (z = 2.81, p = .27), the NS scored
significantly higher than NNSEAL group (z = 9.47, p < .001), and the
NNS group scored significantly higher than the NNSEAL group (z =
10.46, p < .001). From Grade 10 on, all three groups scored similarly
(p > .05).
Research Question Two: What Are the Vocabulary Profiles of the
Learning Materials and Representative Textbooks in EAL, Maths, and
Science in English?
High-frequency words make up the majority of the vocabulary in
the learning materials (Table 9). The EAL materials contain the most
items from the first 3,000-word lists. With the high-frequency word
families and supplementary lists (proper nouns, abbreviations, com-
pounds, and marginal words), English reaches 95% coverage, but
maths and science need the mid-frequency vocabulary (4,000–8,000
word families; Schmitt & Schmitt, 2012) also to reach 95% and over.
The maths learning materials included 3.12% from the BNC/COCA
9000-, 10,000-, and 11,000-word lists, e.g. decimal (9,000) and quotient
(11,000). This corpus is very small, so these overall patterns are indica-
tive only.
The textbook BNC/COCA high, mid, and supplementary lists
results are shown in Table 10. The two novels reach 98%, as expected,
and the Grade 8 maths text almost reaches 98%. The Grade 11 science
text reaches 96%, and the Grade 11 maths text almost reaches 95%.

600 TESOL QUARTERLY


15457249, 2018, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tesq.450 by Desi Arinda - <shibboleth>-member@monash.edu.au , Wiley Online Library on [20/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
TABLE 9
Vocabulary Profile of Learning Materials in EAL, Maths, and Science (%)

Word Lists EAL Maths Science


BNC/COCA 1,000–3,000 92.43 88.13 87.31
BNC/COCA 4,000–8,000 3.56 5.63 5.42
BNC/COCA Supplementary Lists 2.97 1.56 4.04
Total 98.96 95.32 96.77

TABLE 10
Coverage of the BNC/COCA High, Mid, and Supplementary Lists Over English, Maths, and
Science Textbooks (%)

English English Maths Maths Science Science


Frequency Bands Grade 6 Grade 10 Grade 8 Grade 11 Grade 8 Grade 11
High BNC/ 92.62 92.28 85.32 73.19 85.75 83.79
COCA 1,000–3,000
Mid BNC/ 4.39 3.22 5.45 4.58 3.60 7.28
COCA 4,000–8,000
Supplementary Lists 1.08 3.42 7.20 16.85 3.60 4.97
Total 98.09 98.92 97.97 94.62 92.95 96.04

Learners with a vocabulary size of around 8,000 word families should


be able to cope with the vocabulary in these texts with support. The
Grade 11 science text has the highest coverage at a mid-frequency
level of just over 7%, and the Grade 10 novel has the lowest.
Table 10 shows that the supplementary lists reach nearly 17% of the
Grade 11 maths text, mostly because X and Y occur 7,287 and 2,335
times, respectively, in this text. This textbook also contains a number
of words which occur often but were not in the BNC/COCA lists, such
as surd/surds, factorials, and conic/conics, as well as maths-specific formu-
las such as sin2, sin3, and YX. The Grade 8 maths text also contains a
large number of supplementary items (7.2%), followed by the Grade
11 science text (nearly 5%).
It is important to look closely at the coverage of the high-frequency
lists because of their major role in these texts (BNC/COCA 1,000, 2,000,
and 3,000) (Table 11). Note that percentage in each ease drops sharply
from the first 1,000 words down to the 3,000-word level. The English
novels reach around 92% with the first 3,000-word lists. There is a strik-
ing difference between the first 1,000 BNC list over the English (around
84%) texts and the maths (57.64% and 66.16%) and science (65.4% and
66.16%) textbooks. The Grade 8 science and maths textbooks reach
85% coverage with the 3,000-word family lists, followed by the Grade 11
Science textbook at 83.79%. The Grade 11 maths textbook appears to
be the most difficult from a lexical perspective, as the first 3,000 words
cover just over 73%, but the supplementary list coverage mentioned

LONGITUDINAL VOCABULARY DEVELOPMENT 601


15457249, 2018, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tesq.450 by Desi Arinda - <shibboleth>-member@monash.edu.au , Wiley Online Library on [20/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
TABLE 11
Coverage of the First 3,000 BNC/COCA Word Lists (High-Frequency Vocabulary) Over Eng-
lish, Maths, and Science Textbooks (%)

English English Maths Maths Science Science


High Frequency Grade 6 Grade 10 Grade 8 Grade 11 Grade 8 Grade 11
BNC/COCA 1,000 84.33 83.51 69.51 57.64 66.16 65.4
BNC/COCA 2,000 6.07 6.11 10.26 7.74 10.53 11.54
BNC/COCA 3,000 2.22 2.66 5.55 7.81 9.06 6.85
Total 92.62 92.28 85.32 73.19 85.75 83.79

above goes a long way to ameliorating the level of lexical difficulty of this
text.
Table 11 shows that the Grade 8 science textbook has the highest
vocabulary load. It contains low-frequency words such as photosynthesis
and microorganism from the 9,000-word BNC/COCA list and they occur
nearly 40 times each; respiration and micron (10,000-word BNC/COCA
list), fission (11,000-word BNC/COCA list), and chlorophyll and phylum
(12,000-word BNC/COCA list). Technical vocabulary plays a major
role in this text.
Research Question 3: What Coverage Does the AWL Provide Over
the Learning Materials and Textbooks?
The Grade 11 maths textbook contained the highest proportion of
AWL words (7.6%), followed by the Grade 8 science textbook
(6.41%), the Grade 11 Science textbook (6.25%), and the Grade 6
maths textbook at 5.61%. Unsurprisingly, the novels contained many
fewer AWL words, around 1.5%, which is similar to Coxhead’s (2000)
finding of around 1.4% coverage of the word list over a 3.5 million–
word corpus of fiction.

DISCUSSION

This study highlights how EMI classrooms can contain learners with
vastly different levels of vocabulary knowledge in English, including
NSs who demonstrate solid knowledge of high-frequency vocabulary in
the early years of their high school study in Grade 6, but who rapidly
develop mastery in academic and mid-frequency vocabulary as their
studies progress. The NNSs whose VLT scores suggested sufficient
knowledge of English vocabulary to be placed in everyday classes
demonstrated a similar but delayed pattern of vocabulary development
to the NSs. In contrast, the NNSEALs needed at least four years to
score the same as their classmates in high-frequency vocabulary, and
five years for academic vocabulary. As in Webb and Chang (2012),

602 TESOL QUARTERLY


15457249, 2018, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tesq.450 by Desi Arinda - <shibboleth>-member@monash.edu.au , Wiley Online Library on [20/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Pietal€a and Merikivi (2014), and Gierlinger and Wagner (2016), all
participants in the present study showed increases in their vocabulary
scores during their studies. In contrast to Webb and Chang (2012) in
Taiwan, where mastery was achieved by fewer than half their partici-
pants on the 1,000-word level of the VLT after five years of instruction,
mastery in the present study by the NNSEAL participants is achieved
across the first three levels of the VLT and the academic section of
the test. The international school context seems to provide daily and
extended exposure to English which the Taiwanese environment does
not provide, and learners benefit from speaking languages which are
more akin to English.
The highest gains in Webb and Chang (2012) were in the 2,000-
word and academic levels, with the students in the group with more
hours of English per week having the highest gains overall. In the pre-
sent study, the NNSEAL group made the highest gains across all levels
of the VLT except the 10,000-word level, and the greatest gains were
in the 2,000-word and academic levels, closely followed by the 3,000-
and 5,000-word levels. These results suggest that the extra support was
not only needed for these students in their first year of high school,
but was potentially a lifeline.
The learning materials and textbook results suggest that even the
texts with high proportions of high-frequency vocabulary in English
classes would be difficult for the NNSEAL students for several years of
their study at school, and that the maths and science texts would be
even more difficult.
Although the learners in this study show growth in vocabulary across
the years, even at the 10,000-word level, these gains for native speakers
do not appear to be as rapid in this context in comparison with gains
reported by Coxhead, Nation, and Sim (2015) of upwards of 1,300
word families per year for native speakers aged 15–16 years old in the
New Zealand context based on the Vocabulary Size Test (Nation, n.d.;
see Nation, 2013). The native speakers in the present study were, of
course, in an EFL context. The VST test results do not map so easily
to the VLT results in the current study because the VLT has a limited
set of tests based on 2,000-, 3,000-, 5,000-, 10,000-word and academic
levels and the VST tests up to 20,000 word families. That said, Table 7
shows that mastery for all three groups in the present study is stag-
gered over different years of study, meaning that the lexical demands
of the authentic texts are high and even native speakers may struggle
initially.
This study used a sample of representative texts for textbooks and a
small sample of learning materials from classes. These texts did not
cover the full range of reading in EAL, maths, and science, but this
analysis confirms the large vocabulary burden of the maths and

LONGITUDINAL VOCABULARY DEVELOPMENT 603


15457249, 2018, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tesq.450 by Desi Arinda - <shibboleth>-member@monash.edu.au , Wiley Online Library on [20/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
science textbooks in particular. They require over 8,000 word families
plus proper nouns, marginal words, compounds, and abbreviations to
reach 98% coverage of the texts. As the years progress and the vocabu-
lary knowledge increases, so does the difficulty of the textbooks, as
Table 10 shows (compare Grade 8 and Grade 11 maths results, for
example).
The academic and technical vocabulary in the textbooks results in
this study illustrate again that secondary school texts have lower
amounts of academic vocabulary (between 5.6% and 7.6%) than uni-
versity-level texts (10%, on average, according to Coxhead, 2000) (see
Coxhead et al., 2010; Greene & Coxhead, 2015). Both the maths and
science textbooks have higher levels of AWL words than the English
novels. Subject-specific word lists, such as the Middle School Vocabu-
lary Lists (Greene & Coxhead, 2015), have the potential for better
return for learning than the AWL (although these lists do contain
AWL words). Chung and Nation (2003) found one word in three in a
line of text in a university-level anatomy textbook is technical, and
technical vocabulary can include words from high, mid, and low
frequencies (Nation, 2016). Learners might expect that technical
vocabulary comes from Latin or Greek and be long, but there are
many high-frequency words which have technical meanings in subjects
such as maths and science, such as constant and variable. Technical
vocabulary in secondary school subjects is an important area for future
research, particularly because of the strong connection between
content knowledge and technical vocabulary (Woodward-Kron, 2008;
Coxhead, 2018).
It is important to point out that the nature of a longitudinal study
involving multiple testing of participants using the same test over time
may raise a question of how often a test might be used before a ceiling
effect or learning effect might be found. Webb and Chang (2012)
noted that they thought it unlikely that there would be a learning
effect from the repeated test over the 5 years in their study. The tea-
cher who gathered the data for this study is adamant that, if learners
actually learned items on their vocabulary test because of exposure to
the test once a year, evidence of this learning would be easy to see. A
simple classroom-based recall task on the test items by a group of
Grade 9 and Grade 10 learners in 2015 with the teacher showed little
such evidence of learning.

Limitations

There are several limitations to the methodology used in the cur-


rent study. First, the VLT data set consists solely of results for the

604 TESOL QUARTERLY


15457249, 2018, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tesq.450 by Desi Arinda - <shibboleth>-member@monash.edu.au , Wiley Online Library on [20/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
various cohorts, with little other data such as grade point averages to
corroborate or interrogate the results as a whole. Second, the VLT
used in this study does not include school-specific corpora, which
means that students are not being tested on or given credit for know-
ing or learning high-frequency, essential vocabulary in a school con-
text. Lexical items such as homework, extension, margin, and detention are
examples of such words. Another limitation is that timing and condi-
tions (paper vs. online) were not the same for all of the tests. Finally,
the VLT is now quite old. There are now larger corpora-based studies
of frequency and vocabulary which have been used to develop a new
version of the test (see Webb & Sasao, 2013; Webb, Sasao, & Ballance,
2017).
Finally, clearly the textbooks and learning materials in this study are
indicative only of the kinds of texts which might be used in such class-
rooms. The learning materials come from these international school
classes, but their word count is small. The textbooks are proxies and
therefore generalizability is in question, and the Pride and Prejudice text
and Hunger Games chapters were unbalanced in the number of run-
ning words. That said, these materials and textbooks do provide an
insight into the kinds of vocabulary, and amount, that learners need
to cope with studying across subjects in school.

Implications for Pedagogy


Teachers need to ensure that measures such as the VLT are used to
assess vocabulary knowledge at the commencement of studies and for
keeping track of lexical knowledge development over time. This is par-
ticularly important for pedagogical decisions on who gets what support
and why. By using tests such as the VLT annually, teachers can gain a
better understanding of the lexical needs of their students in terms of
how a student may be developing compared to the rest of the stu-
dents. A lack of improvement from year to year in VLT results would
be a signal that a student was struggling. In essence, more of a lan-
guage focus in EMI would be beneficial.
This study suggests that NNSEAL students can learn a great deal of
vocabulary throughout their schooling and that a focus on lexical
development can be beneficial, particularly in high-frequency, mid-fre-
quency, and academic vocabulary. Recognising how these learners
might struggle is vital in all modes of schooling, not just within the
EMI context. This study revealed that low-proficiency learners need
support with learning the high-frequency words of English and aca-
demic vocabulary from the start of their studies in Grade 6. They
would benefit from teacher-developed learning materials such as the

LONGITUDINAL VOCABULARY DEVELOPMENT 605


15457249, 2018, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tesq.450 by Desi Arinda - <shibboleth>-member@monash.edu.au , Wiley Online Library on [20/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
ones in this study. They could also benefit from simplified reading
materials which are more comprehensible in all subject areas, support
for unknown vocabulary, and repetition of key vocabulary in reading.
For example, maths and science texts could be adapted by identifying
and replacing low-frequency vocabulary items with high-frequency
items. Another option could be mid-frequency graded readers, which
control vocabulary at the 3,000-, 6,000-, or 8,000-word levels (e.g.,
Nation & Anthony, 2013).
Teachers in EMI contexts may well have cohorts like the present
study with native speakers, nonnative speakers, and nonnative speak-
ers who need EAL support all in the same classes, just like teachers
in other contexts such as the United States with school populations
which are quite mixed. Learners in these classes move between sub-
jects such as English, maths, and science, and are therefore exposed
to a great deal of text with different lexical demands. Learning
materials which are provided by teachers appear to have less of a
vocabulary burden than textbooks, and contain a high proportion
of high-frequency vocabulary, which means they can act as a sort of
lexical bridge between the classroom and the textbook. Teachers
can analyse texts using BNC/COCA with Range (Heatley et al.,
2002), Laurence Anthony’s Antconc (n.d.), and Cobb’s (n.d.) Lex
Tutor. All three tools enable teachers to identify low-frequency
vocabulary in the texts and decide whether to replace these words
or draw attention to them if they are technical. Greene and Cox-
head’s (2015) Middle School Vocabulary Lists for English Grammar
and Writing, Maths, Science, Health, and Social Sciences and His-
tory are available on Lex Tutor (Cobb, n.d.), and can be used to
analyse learning materials and texts. To prepare learners for reading
difficult textbooks, teachers could develop resources which sum-
marise the main ideas and deal directly with the vocabulary which
the learners need, as a stepping stone towards reading the actual
text.
Some learners might find word lists helpful, such as Coxhead’s
AWL (2000) or Gardner and Davies’s (2014) AVL, early on. Technical
words also need to be identified, and worked with in context in class-
room texts. The subject-specific Middle School Lists might be useful in
this case (Greene & Coxhead, 2015). Students who stay within the
same subject area, for example, from Grade 6 maths into Grade 11
maths, will continue to build on their knowledge of the subject and its
technical vocabulary.
Frameworks such as Nation’s (2013) four strands can be used to pro-
vide guidance on how teachers might think about vocabulary in their
programmes of study. The four strands are meaning-focused input
(learning from reading and listening); meaning-focused output

606 TESOL QUARTERLY


15457249, 2018, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tesq.450 by Desi Arinda - <shibboleth>-member@monash.edu.au , Wiley Online Library on [20/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
(learning from writing and speaking); fluency practice in all four skills;
and language-focused learning (concentrating on aspects of vocabulary
such as spelling and pronunciation). The strands can help with analyz-
ing programmes of learning and identifying gaps in planning and
instruction, no matter what subject area is being taught and who the
learners are. Webb and Chang (2012) have suggestions on organizing
vocabulary in programmes of learning.

Future Research

A VLT item analysis across years and with complete sets of partici-
pant data would be useful to explore the consistency of results, as
would using other assessment measures to triangulate the VLT results,
such as grade point averages. It would be good to see research on new
versions of a school-based VLT using, for example, Greene and Cox-
head’s (2015) Middle School Vocabulary Lists. Future research could
also look at larger scale analyses of learning materials and textbooks
and technical vocabulary. Finally, more research on how learners
develop their understanding of a subject over time and the vocabulary
of that subject is needed.

CONCLUSION
This study tracked the development of receptive vocabulary knowl-
edge over time using the VLT for native and nonnative speakers of
English in an EMI context. Different levels of the test continued to
show gains even after participants had taken the test multiple times.
The greatest gains were made by the nonnative speaker group who
had low levels of English proficiency in Grade 6. The learners with
higher VLT scores would be best placed to cope with textbooks and
learning materials, but over time NNSEAL students have the opportu-
nity to further develop their vocabulary and, theoretically at least, be
able to close the gap between their lexical knowledge and their class-
room texts. Our analysis also suggested that learners would need siz-
able vocabularies to cope with authentic school textbooks, particularly
in maths and science.

ACKNOWLEDGEMENTS

We wish to acknowledge the teachers and students who took part in this study,
and in particular the teacher who gathered the testing data for this project who

LONGITUDINAL VOCABULARY DEVELOPMENT 607


15457249, 2018, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tesq.450 by Desi Arinda - <shibboleth>-member@monash.edu.au , Wiley Online Library on [20/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
wishes to remain anonymous. We would also like to acknowledge funding from
the Faculty of Humanities and Social Sciences, Victoria University of Wellington
(Research Grant 208956 and 211700).

THE AUTHORS

Averil Coxhead is an associate professor in the School of Linguistics and Applied


Language Studies, Victoria University of Wellington, Aotearoa/New Zealand. She
is the author of Vocabulary and English for Specific Purposes Research (2018). Her cur-
rent research focuses on vocabulary in English for academic purposes at secondary
school and university.

T. J. Boutorwick completed his PhD in applied linguistics at Victoria University of


Wellington, New Zealand, in 2017. He is currently working in the University’s Lan-
guage Learning Centre. His research interests include second language vocabulary
acquisition and word associations and extensive reading, with a particular focus on
quantitative and statistical analysis.

REFERENCES
Anthony, L. (n.d.). Antconc. Retrieved from http://www.laurenceanthony.net/sof
tware/antconc/
Baayen, R., Davidson, D., & Bates, D. (2008). Mixed-effects modeling with crossed
random effects for subjects and items. Journal of Memory and Language, 59, 390–
412. https://doi.org/10.1016/j.jml.2007.12.005
Bates, D., M€achler, M., Bolker, B., & Walker, S. (2015). Fitting linear mixed-effects
models using lme4. Journal of Statistical Software, 67(1), 1–48. https://doi.org/
10.18637/jss.v067.i01
Chung, T., & Nation, I. S. P. (2003). Technical vocabulary in specialised texts.
Reading in a Foreign Language, 15(2), 103–116.
Cobb, T. (n.d.). Compleat Lexical Tutor. Retrieved from http://www.lextutor.ca/
vp/comp/
Coxhead, A. (2000). A new academic word list. TESOL Quarterly, 34, 213–238.
https://doi.org/10.2307/3587951
Coxhead, A. (2012). Researching vocabulary in secondary school English texts: The
Hunger Games and more. English in Aotearoa, 78, 34–41.
Coxhead, A. (2018). Vocabulary and English for specific purposes research: Quantitative
and qualitative perspectives. London, England: Routledge.
Coxhead, A., Nation, P., & Sim, D. (2015). Vocabulary size and native speaker sec-
ondary school students. New Zealand Journal of Educational Studies, 50, 121–135.
https://doi.org/10.1007/s40841-015-0002-3
Coxhead, A., Stevens, L., & Tinkle, J. (2010). Why might secondary science text-
books be difficult to read? New Zealand Studies in Applied Linguistics, 16(2), 35–52.
Coxhead, A., & White, R. (2012). Building a corpus of secondary school texts: First
you have to catch the rabbit. New Zealand Studies in Applied Linguistics, 18(2),
67–73.
Davies, M. (2008). The Corpus of Contemporary American English: 520 million words,
1990–present. Retrieved from http://corpus.byu.edu/coca/

608 TESOL QUARTERLY


15457249, 2018, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tesq.450 by Desi Arinda - <shibboleth>-member@monash.edu.au , Wiley Online Library on [20/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Department for Education. (2012). Key Stage 2 English grammar punctuation and spel-
ling test: Information for parents. Retrieved from https://www.gov.uk/governme
nt/publications/key-stage-2-english-grammar-punctuation-and-spelling-test-inf
ormation-for-parents
Earls, C. W. (2016). Evolving agendas in European English-medium higher education:
Interculturality, multilingualism and language policy. Basingstoke, England: Palgrave
Macmillan. https://doi.org/10.1057/9781137543127
Gardner, D., & Davies, M. (2014). A new academic vocabulary list. Applied Linguis-
tics, 35, 305–327. https://doi.org/10.1093/applin/amt015
Gierlinger, E., & Wagner, T. (2016). The more the merrier–: Revisiting CLIL-
based vocabulary growth in secondary education. Latin American Journal of Con-
tent and Language Integrated Learning, 9(1), 37–63. https://doi.org/10.5294/lac
lil.2016.9.1.3
Greene, J., & Coxhead, A. (2015). Academic vocabulary for middle school students:
Research-based lists and strategies for key content areas. Baltimore, MD: Brookes.
Heatley, A., Nation, P., & Coxhead, A. (2002). The Range programme. Retrieved
from http://www.victoria.ac.nz/lals/about/staff/paul-nation
Henriksen, B., & Danelund, L. (2015). Studies of Danish L2 learners’ vocabulary
knowledge and the lexical richness of their written production in English. In P.
Pietil€a, K. Dor o & R. Pipalova (Eds.), Lexical issues in L2 writing (pp. 1–27).
Newcastle upon Tyne, England: Cambridge Scholars Publishing.
Hothorn, T., Bretz, F., & Westfall, P. (2008). Simultaneous inference in general
parametric models. Biometrical Journal, 50(3), 346–363. https://doi.org/10.
1002/bimj.200810425.
International Baccalaureate Organization. (2017). Benefits of the IB. Retrieved from
http://www.ibo.org/benefits/
Laufer, B. (1989). What percentage of text-lexis is essential for comprehension? In
C. Lauren & M. Nordman (Eds.), Special language: From humans thinking to think-
ing machines (pp. 316–323). Clevedon, England: Multilingual Matters.
Laufer, B., & Ravenhorst-Kalovski, G. C. (2010). Lexical threshold revisited: Lexi-
cal text coverage, learners’ vocabulary size and reading comprehension. Reading
in a Foreign Language, 22(1), 15–30.
Leung, C. (2014). Researching language and communication in schooling. Linguis-
tics and Education, 26(1), 136–144. https://doi.org/10.1016/j.linged.2014.01.
005
Malmstr€ om, H., Pecorari, D., & Gustafsson, M. (2016). Coverage and development
of academic vocabulary in assessment texts in English medium instruction. In
S. G€ opferich & I. Neumann (Eds.), Assessing and developing academic and profes-
sional writing skills (pp. 45–69). New York, NY: Peter Lang.
Meara, P., & Milton, J. (2003). X-Lex, The Swansea Levels Test. Newbury, England:
Express.
Nation, I. S. P. (n.d.). The Vocabulary Size Test. Retrieved from https://www.victo
ria.ac.nz/lals/about/staff/paul-nation#vocab-tests
Nation, I. S. P. (1983). Testing and teaching vocabulary. Guidelines, 5(1), 12–25.
Nation, I. S. P. (2006). How large a vocabulary is needed for reading and listen-
ing? Canadian Modern Language Review, 63(1), 59–82.
Nation, I. S. P. (2012). The BNC/COCA word family lists. Retrieved from http://
www.victoria.ac.nz/lals/about/staff/paul-nation
Nation, I. S. P. (2013). Learning vocabulary in another language (2nd ed.). Cam-
bridge, England: Cambridge University Press.
Nation, I. S. P. (2016). Making and using word lists for language learning and testing.
Amsterdam, The Netherlands: John Benjamins.

LONGITUDINAL VOCABULARY DEVELOPMENT 609


15457249, 2018, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tesq.450 by Desi Arinda - <shibboleth>-member@monash.edu.au , Wiley Online Library on [20/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Nation, I. S. P., & Anthony, L. (2013). Mid-frequency readers. Journal of Extensive
Reading, 1(1), 5–16.
Nurweni, A., & Read, J. (1999). The English vocabulary knowledge of Indonesian
university students. English for Specific Purposes, 18(2), 161–175.
Olmos, C. (2009). An assessment of the vocabulary knowledge of students in the
final year of secondary education: Is their vocabulary extensive enough? Interna-
tional Journal of English Studies, 9, 79–90.
Pietil€a, P., & Merikivi, R. (2014). Vocabulary in CLIL and in mainstream educa-
tion. Journal of Language Teaching and Research, 5, 487–497. https://doi.org/10.
4304/jltr.5.3
R Core Team. (2016). R: A language and environment for statistical computing. Vienna,
Austria: R Foundation for Statistical Computing. Retrieved from https://www.R-
project.org/
Read, J. (2000). Assessing vocabulary. Cambridge, England: Cambridge University
Press.
Schmitt, N., Jiang, X., & Grabe, W. (2011). The percentage of words known in a
text and reading comprehension. Modern Language Journal, 95, 26–43. https://d
oi.org/10.1111/j.1540-4781.2011.01146.x
Schmitt, N., & Schmitt, D. (2012). A reassessment of frequency and vocabulary size
in L2 vocabulary teaching. Language Teaching, 47, 484–503. https://doi.org/10.
1017/S0261444812000018
Schmitt, N., Schmitt, D., & Clapham, C. (2001). Developing and exploring the
behaviour of two new versions of the Vocabulary Levels Test. Language Testing,
18(1), 55–88. https://doi.org/10.1177/026553220101800103
Sharma, Y. (2016, February 16). Asia drives demand for international schools. BBC
News. Retrieved from http://www.bbc.com/news/business-35533953
Smith, B. (2015, December 10). US, UK set to lose market share of mobile stu-
dents in next decade. The PIE News. Retrieved from http://thepienews.com/ne
ws/us-uk-set-to-lose-market-share-of-mobile-students-in-next-decade
University of Cambridge Local Examinations Syndicate. (2017). Cambridge IGCSE.
Retrieved from http://www.cambridgeinternational.org/programmes-and-qualif
ications/cambridge-secondary-2/cambridge-igcse/
van Zeeland, H., & Schmitt, N. (2013). Lexical coverage in L1 and L2 listening
comprehension: The same or different from reading comprehension? Applied
Linguistics, 34, 457–479. https://doi.org/10.1093/applin/ams074
Webb, S., & Chang, A. (2012). Second language vocabulary growth. RELC Journal,
43(1), 113–126. https://doi.org/10.1177/0033688212439367
Webb, S., & Sasao, Y. (2013). New directions in vocabulary testing. RELC Journal,
44(3), 263–277. https://doi.org/10.1177/0033688213500582
Webb, S., Sasao, Y., & Ballance, O. (2017). The new vocabulary levels test. Interna-
tional Journal of Applied Linguistics, 168(1), 34–70. https://doi.org/10.1075/itl.
168.1.02web
West, M. (1953). A General Service List of English words. London, England: Long-
man, Green.
Woodward-Kron, R. (2008). More than just jargon: The nature and role of special-
ist language in learning disciplinary knowledge. Journal of English for Academic
Purposes, 7(4), 234–249. https://doi.org/10.1016/j.jeap.2008.10.004

610 TESOL QUARTERLY

You might also like