ChapterPDF Available
In its earliest stages of development the Common European Framework of
Reference for Languages (CEFR) included vocabulary lists in its materials
and these gave some indication of the scale of the vocabulary knowledge
that the creators were envisaging at the various levels of the framework.
More recently these have been removed and learners, textbooks and course
syllabuses are placed into the framework levels according to skills-based
rather than knowledge-based criteria (Council of Europe, 2003). The pur-
pose of this chapter is to see what happens when vocabulary size measures
are placed back into the framework and there are two reasons for want-
ing to do this. One is academic interest in seeing what vocabulary sizes
emerge at the CEFR levels and considering how these compare across levels
and across languages. The second reason is a practical one and is to help
to make the framework more robust. The skills-based criteria have the vir-
tue of making the framework flexible and highly inclusive, and almost any
course, textbook or learner should be able to find a place in the system.
However, the penalty for such flexibility is that the levels become impre-
cise; it is often possible to place learners or textbooks at several of the CEFR
levels. This potentially devalues the framework and diminishes its useful-
ness. The British foreign language exam system in schools, for example,
has been criticized for being misplaced within the system and, as a conse-
quence, for misleading those who try to use it (Milton, 2007a). The pres-
ence of a more objectively assessed, knowledge-based measure, such as
vocabulary size, ought to help avoid this kind of ambiguity and the prob-
lems associated with it.
In this chapter, therefore, we intend to review the evidence we have
from a variety of learners in different countries and learning different lan-
guages where we are able to tie vocabulary size scores to different levels in
the CEFR hierarchy. We intend to draw on results we have from learners
Vocabulary Size and the Common
European Framework of Reference
for Languages
James Milton* and Thomaï Alexiou**
*Swansea University, UK
**Aegean University, Greece
9780230_206687_13_cha12.indd 1949780230_206687_13_cha12.indd 194 2/4/2009 2:53:13 PM2/4/2009 2:53:13 PM
This file is to be used only for a purpose specified by Palgrave Macmillan, such as checking proofs, preparing
an index, reviewing, endorsing or planning coursework/other institutional needs. You may store and print
the file and share it with others helping you with the specified purpose, but under no circumstances may the
file be distributed or otherwise made accessible to any other third parties without the express prior permission
of Palgrave Macmillan. Please contact if you have any queries regarding use of the file.
James Milton and Thomaï Alexiou 195
in Spain, Hungary, Greece and Britain, learning English, French and Greek
as foreign languages. By looking at modern foreign language learners at
different levels we should be able to see whether the vocabulary knowl-
edge changes systematically as the language level of learners increases. It
would be expected that learners at the lower levels of the hierarchy, A1 or
A2, would know fewer words and expressions in the foreign language than
learners who are very advanced and who are taking courses or studying for
exams at C1 or C2 level. This in turn raises the possibility of comparing
vocabulary knowledge levels across languages and across different language
systems; is the knowledge of French learners in Britain, say, comparable
in some meaningful way to the knowledge of Greek learners of French
or even Greek learners of English? It is not always obvious how to com-
pare knowledge across different languages but one method for comparing
vocabulary sizes will be proposed and examined. This should begin to tell
us whether the CEFR hierarchies are as robust as we would like them to be,
and whether vocabulary size measures can help to add a useful degree of
precision to the difficult art of placing learners at the correct CEFR level.
Background to the CEFR and the place of vocabulary
knowledge within it
The CEFR was created to provide a framework of comparison in the study
and testing of languages. There were many issues involved in creating such
a framework which has taken over 25 years to accomplish. It requires, for
example, the development of a common set of terms and references so that
professionals across Europe can speak to each other on aspects of language
learning and language level, and be confident that what they intend to
convey will be understood in the same way. For most users, that is learn-
ers, parents, teachers and employers, the most obvious intention of the
framework is to bring order to the plethora of courses, exams and awards
which learners can take. Even within a single language it was frequently
unclear how one exam related to another in its demands and in its diffi-
culty. Students we have dealt with in Greece often confidently assert that
the Michigan Proficiency exam is easier than the Cambridge Proficiency in
English exam, although there is very little evidence to suggest whether this
is the case or not. How should this kind of opinion be interpreted? Would
it be appropriate to value a pass in the Cambridge exam more highly than
the Michigan in determining, say, whether a candidate has the qualifica-
tions for entry to a university course requiring a language qualification?
The presence of the CEFR, even if it is no more than a common vocabulary
to describe the hierarchy of levels, ought to allow questions like these to
be answered rather better. It should allow exams, for example, to be placed
within a framework so that users can see which exams are intended to be at
different levels and which are intended to be similar.
9780230_206687_13_cha12.indd 1959780230_206687_13_cha12.indd 195 2/4/2009 2:53:14 PM2/4/2009 2:53:14 PM
196 Vocabulary Studies in First and Second Language Acquisition
The CEFR is not intended to be specific to one country; it is designed
to be a common framework which can be applied to language courses and
exams across Europe. It ought to tell us, for example, whether learners from
Germany, Spain and Italy with school leaving certificates in a foreign lan-
guage are of the same standard so we can know whether they could enter
a course of study requiring a set level of ability. Or it might tell us whether
these learners have the foreign language ability required for a job. In terms
of textbooks and teaching materials, the CEFR has been taken up by the
EFL world in particular and by EFL publishers. It is now common for text-
books and for language courses to be described in terms of the framework.
Therefore, in principle at least, a course designed for learners at, say B1
level ought to be able to select materials from a range of textbooks designed
for students at this level and all of them should be appropriate in some
meaningful sense. Additionally, the CEFR should, in principle, should allow
direct comparison between learners, courses and course books in different
languages. Because the framework is not language specific, by implication it
should allow intelligent comparisons between exams or learners of Italian,
German, Greek or any other language. In Britain, for example, the age 18
Advanced level foreign language exams are pitched at the CEFR B2 level
and so Advanced level students of, say German, should have the same kind
of knowledge and skills as learners of Italian also taking Advanced level.
Both of these should be comparable with learners in other countries follow-
ing courses at B2 level in German and Italian.
At the outset of the project which created the CEFR the descriptors which
were created included word lists. The Threshold level materials (for example,
Coste, Courtillon, Ferenczi, Martins-Baltar and Papo, 1987; van Ek and
Trim, 1990) and some of the Waystage materials (for example, van Ek, 1990)
contain such lists. The level descriptors are generated from the notional-
functional categories which underlie the framework. While this mode of
analysis now looks rather old-fashioned, the word lists they contain are,
nonetheless, both useful and usable. The word lists at Threshold level (CEFR
B1) contain about 2000 words and the Waystage level (CEFR A2) materials
contain word lists with about 1000 words. However, the overall framework
document (Council of Europe, 2003) has concentrated on skills and can-do
lists, and language-specific items, such as the word lists, are absent. No one
is saying, of course, that the skills which define the framework are divorced
from language knowledge such as vocabulary knowledge. The word lists
have not been disowned by the framework. Nonetheless, they appear
to have receded into the background and the scale of vocabulary knowl-
edge which might reasonably be associated with the CEFR levels is now an
unknown quantity.
There is a case for arguing that a measure such as vocabulary size ought
to fit well into a hierarchy of level such as the CEFR. There is growing evi-
dence that vocabulary size measures correlate well with overall measures of
9780230_206687_13_cha12.indd 1969780230_206687_13_cha12.indd 196 2/4/2009 2:53:15 PM2/4/2009 2:53:15 PM
James Milton and Thomaï Alexiou 197
language ability such as scores on the Cambridge IELTS test (Milton, Wade
and Hopkins, forthcoming). They also correlate well with all four skills, and
with reading and writing in particular. Staehr (forthcoming), for example, is
able to gain correlations of .83 between scores gained by his 88 testees on a
test of receptive vocabulary size (Schmitt, Schmitt and Clapham, 2001) and
on a multiple choice test of reading comprehension. A correlation of .73 was
found with the same group between vocabulary scores and assessments on
an academic writing task. Both correlations are statistically significant. While
lower correlations are found by Staehr with listening and speaking skill scores,
his results still explain between 35 and 40 per cent of variance in the scores
for these skills. Using a combination of both phonological and orthographic
tests of vocabulary size, Milton et al. (forthcoming) are able to explain over
40 per cent of variance on scores in IELTS speaking and listening sub-skill
scores. This suggests, perhaps not surprisingly, that the skills of listening and
speaking access different lexical resources from reading and writing. Listening
and speaking rely on aural word knowledge; reading and writing on knowl-
edge of the written form of words. Nonetheless, they emphasize how impor-
tant vocabulary knowledge is to all language skills. Other studies suggest that
vocabulary size scores correlate well with hours of instruction and teacher
assessments (Orosz, 2007) and with the size and frequency distribution of the
vocabulary content available to learners in course books (Vassiliu, 2001).
There is good reason for thinking, therefore, that if the CEFR has validity
as a hierarchy of language level and ability, then each succeeding increase
in level in the CEFR should be matched by an increasing demand in the
vocabulary knowledge of the learners who take exams at that level. With
each progressively higher CEFR band, there should be higher mean scores
on vocabulary size measures with groups of learners. If this were not seen
then the validity of the CEFR would be called into question. The word lists
in the early CEFR materials appear to reflect this pattern. The Threshold (B1)
level word lists are indeed larger, implying greater knowledge by learners at
this level, than the Waystage (A2) material lists. The information included
with Meara and Milton’s (2003, p. 5) Swansea Levels Test (XLex), explicitly
links the EFL vocabulary size scores to attainment in Cambridge EFL exams
and these exams, of course, have a place in the hierarchy of CEFR levels.
The range of scores they suggest for each level is shown in Table 12.1.
Vocabulary size measures also have a distinct benefit in language meas-
urement terms, of being, or appearing to be, more countable and therefore
objective, than the kind of subjective evaluations of level which abound
in other aspects of language. Modern methodology, for example Meara
and Jones’s Eurocentre’s Vocabulary Size Test (EVST) (1990), allows a numeri-
cal estimate of a learner’s vocabulary to be made, and a learner with, say,
2000 words out of the 10,000 in this test, can be argued to have double
the knowledge of another learner with only 1000 words. It is impossible in
the current state of knowledge to characterize knowledge of grammatical
9780230_206687_13_cha12.indd 1979780230_206687_13_cha12.indd 197 2/4/2009 2:53:15 PM2/4/2009 2:53:15 PM
198 Vocabulary Studies in First and Second Language Acquisition
structure, or ability in a language skill such as reading, in this way, and
subjective judgements have to be made on these matters. No matter how
carefully these judgements are criterion referenced, it is very difficult for
assessors to apply them consistently across the millions of foreign language
learners we have in Europe. The presence of a vocabulary size measure, if
this can be linked to CEFR levels, ought to make any hierarchy of levels
more robust. It would introduce an element of objective assessment and
knowledge-based assessment into the process of placement which, as it cur-
rently stands, is entirely subjective.
Measuring vocabulary size
Recent years have seen the development of rather more systematic and prin-
cipled methods for estimating the vocabulary knowledge in foreign language
learners. There is considerable evidence that there is a strong frequency
effect in the learning of foreign language vocabulary (for example, Milton,
2007b). In effect this means that the more frequent a word is then the more
likely it is to be learned. This is not a perfect rule, of course. Word learn-
ing will also be dependent on what thematic material the learner has been
exposed to in textbooks and on word difficulty factors such as whether the
words encountered are cognate or not. But frequency still has a very power-
ful effect, probably more powerful than the other factors and, as a conse-
quence, recent vocabulary tests have drawn on frequency information and
focused their test items in the most frequent bands. Nation’s Vocabulary
Levels Test (Schmitt et al., 2001), Meara and Jones’s EVST (1990) and Meara
and Milton’s Swansea Levels Test XLex (2003) all do this, for example. What
emerges from these tests appears to be good characterizations of learners’
vocabulary knowledge. In the case of the latter two they provide believable
estimates of vocabulary size within the frequency bands they test.
In this chapter we have used vocabulary size estimates arrived at using
the XLex test which has the virtue of having equivalent versions available
in English, French and Greek. All three make estimates of knowledge of the
Table 12.1 EFL vocabulary size, formal EFL exams and the CEFR (from
Meara and Milton, 2003, p. 5)
CEF level Cambridge exam XLex score
(max. 5000)
A1 Starters, Movers and Flyers 1500
A2 Kernel English Test (KET) 1500–2500
B1 Preliminary English Test (PET) 2750–3250
B2 First Certificate in English (FCE) 3250–3750
C1 Cambridge Advanced English (CAE) 3750–4500
C2 Cambridge Proficiency in English (CPE) 4500–5000
9780230_206687_13_cha12.indd 1989780230_206687_13_cha12.indd 198 2/4/2009 2:53:15 PM2/4/2009 2:53:15 PM
James Milton and Thomaï Alexiou 199
most frequent 5000 lemmatized words in these languages. The frequency
information in English is drawn from Nation (1984) and Hindmarsh (1980),
in French from Baudot (1992) and in Greek from the Hellenic National
Corpus (Hatzigeorgiu, Mikros and Carayannis, 2001). XLex asks learners to
respond to 120 test items presented in a yes/no format. The words are pre-
sented in turn and learners must respond either ‘yes’ they know the word,
or ‘no’ they do not know the word; 100 real words are included, 20 drawn
from each of the first five most frequent 1000 word bands. In addition the
test contains 20 pseudo-words, words constructed to look and sound like
real words but which do not exist, and therefore cannot be recognized. The
responses to these words allow the responses to real words to be adjusted
for guesswork and overestimation. While the yes/no task appears simple it
can be quite difficult where a word is only vaguely recognized or is par-
tially known. It presents a challenge to even the most scrupulously honest
learners as to how best to answer some of the items which are only vaguely
recognized. The pseudo-words allow some kind of recognition of this diffi-
culty, and compensation to be made for the differing strategies which learn-
ers may employ. A score of 50 is given for each ‘yes’ response to a real word
and a deduction of 250 is made for each ‘yes’ response to a pseudo-word.
The scores that emerge are estimates of the number of words that each
learner has identified out of 5000. It is common to eliminate data which
demonstrate an unacceptably high level of pseudo-word recognition, and
are arguably unreliable as a result. However, there is no set level at which
a set of answers moves from being reliable to unreliable and in analysing
data for this chapter we have not eliminated such sets of responses. While
we now have a lot of experience in pseudo-word construction in English,
we know much less about the way these things perform in French or Greek.
Subjects and method
In EFL, the vocabulary size scores have been recorded in a state secondary
school in Hungary (Orosz, 2007) and in a private language school in Greece
(data from Milton, 2007b). The learners have been grouped according to the
CEFR level of the class they are in and, where appropriate, the CEFR level of
the exam they are taking. The learners in Greece routinely take the Cambridge
Preliminary English Test (PET) at level B1, Cambridge First Certificate in
English (FCE) at level B2 and Cambridge Certificate of Proficiency in English
(CPE) at level C2. Data were collected from 88 Greek learners at all CEFR lev-
els. The learners in Hungary take the state maturity exams at levels B1 and B2.
Data were collected from 144 Hungarian learners at these levels.
In French as a foreign language the vocabulary size scores have been
recorded in a state secondary school and university in Britain (Milton,
2006, 2008), two private language schools in Greece, and from two schools
in the Spanish state education system. The learners have been grouped
9780230_206687_13_cha12.indd 1999780230_206687_13_cha12.indd 199 2/4/2009 2:53:16 PM2/4/2009 2:53:16 PM
200 Vocabulary Studies in First and Second Language Acquisition
according to the CEFR level of the class they are in and, where appropri-
ate, the CEFR level of the exam they are taking. The learners in Britain take
GCSE exams at B1 level, Advanced levels at B2 level, and we have results for
British university graduates in French which we have assumed will be at C2
level although there is no formal statement on the part of the university to
confirm that this is the case. Data were collected from 155 learners at these
levels. The French learners in Greece do not appear to be taking any formal
exam but are grouped for teaching into CEFR levels. Data were collected
from 65 Greek learners of French at all CEFR levels. Like the Greek learners,
the learners of French in Spain are grouped for teaching according to CEFR
levels and are not, to the best of our knowledge, taking formal exams. Data
from 50 Spanish learners of French were collected at all CEFR levels.
In Greek as a foreign language, data have been collected from learners at
the Centre of Modern Languages in Thessaloniki. Data were collected from
64 learners, from a variety of first language backgrounds, at CEFR levels A1,
A2, B1 and B2.
Vocabulary size and CEFR levels in English
The mean vocabulary size scores at each CEFR level from the 88 EFL learn-
ers in Greece are presented in Table 12.2. The mean vocabulary size scores
for the 144 learners in Hungary at CEFR levels B1, B2 and C1 are presented
in Table 12.3. The XLex scores suggested in Meara and Milton (2003) are
Table 12.2 EFL vocabulary size and the CEFR among learners in Greece
CEF level XLex Mean Max. Min. SD n
A1 1500 1477.27 2100 150 580.37 22
A2 1500–2500 2156.81 3250 700 664.45 22
B1 2750–3250 3263.63 4000 2750 434.79 11
B2 3250–3750 3304.54 4350 2550 666.50 11
C1 3750–4500 3690.90 4300 2650 471.07 11
C2 4500–5000 4068.18 4500 3700 261.02 11
Table 12.3 EFL vocabulary size and the CEFR among learners in Hungary
CEF level XLex Mean Max. Min. SD n
A1 1500
A2 1500–2500
B1 2750–3250 3135.90 4700 1130 434.79 66
B2 3250–3750 3668.42 4950 1880 666.50 72
C1 3750–4500 4340.00 4650 4000 471.07 6
C2 4500–5000
9780230_206687_13_cha12.indd 2009780230_206687_13_cha12.indd 200 2/4/2009 2:53:17 PM2/4/2009 2:53:17 PM
James Milton and Thomaï Alexiou 201
included for reference although it should be noted that the learners in this
case are in classes preparing to take EFL exams at the various CEFR levels, while
the XLex scores in Meara and Milton are for those actually taking the exams.
The results, superficially at least, look rather persuasive. There is a hier-
archy of CEFR levels in each case, and in each case also there is a hierarchy
of mean vocabulary size scores. Learners at A1 appear, on average, to know
fewer words than learners in level A2 who, in turn, know on average fewer
words than those in level B1, and so on up the levels. Even with relatively
small numbers it is possible to argue that this tendency is statistically sig-
nificant. An ANOVA on the Greek data confirms that there are significant
differences between the means at different levels, F(5, 82) 50.197, p .01,
and the same is true of the Hungarian data, F(2, 141) 14.896, p .01.
The two systems also appear, from this limited sample, to be similar and
conform quite closely to the levels of vocabulary knowledge suggested by
Meara and Milton (2003), especially at the lower levels. Both systems sug-
gest considerable vocabulary knowledge is required, approximately 3000
words, before learners progress from the elementary stages of performance
at A1 and A2 level to intermediate B1 level, and a score in region of 3500
words is associated with B2 level. Learners at Advanced levels know even
more than this. These encouraging similarities in mean scores disguise con-
siderable individual variation, however, as the maximum and minimum
scores and standard deviations reveal. While the mean scores for groups
suggest an encouraging general tendency, it seems likely that there are no
clear thresholds where a certain minimum score is a requirement of passing
from one level of skill or ability to another. The reasons why this might be
so are discussed later in the chapter.
Vocabulary size and CEFR levels in French
The mean vocabulary size scores at each CEFR level from the 155 French
as a foreign language learners in Britain at CEFR levels B1, B2 and C2 are
presented in Table 12.4 The mean vocabulary size scores for the 65 learners
Table 12.4 French as a foreign language vocabulary size
and the CEFR among learners in Britain
CEF level Mean Max. Min. SD n
B1 952.04 1900 0 440.28 49
B2 1882.58 3650 650 562.21 89
C2 3326.47 4150 2050 711.75 17
9780230_206687_13_cha12.indd 2019780230_206687_13_cha12.indd 201 2/4/2009 2:53:17 PM2/4/2009 2:53:17 PM
202 Vocabulary Studies in First and Second Language Acquisition
of French as a foreign language in Greece at all CEFR levels are presented in
Table 12.5. The mean vocabulary size scores for the 50 learners of French as
a foreign language in Spain at all CEFR levels are presented in Table 12.6.
There are no guideline scores for what vocabulary sizes might be associated
with each level.
There is rather more variation apparent in the French learners than in
the EFL data but in one respect, at least, the results are consistent and
encouraging. As with the EFL data there is a hierarchy of mean vocabulary
size in each set which rises in line with the CEFR levels. An ANOVA using
the British data confirms that there are significant differences between
the mean scores at each CEFR level, F(2, 152) 126.055, p .01. A Tukey
analysis further confirms that the difference between the mean score at
each CEFR level in these data is also statistically significant. The Greek and
Spanish data also confirm the relationship. ANOVAs give the results F(5,59)
23.713, p .01 for the Greek data and F(5, 44) 21.401, p .01 for the
Spanish data.
It is less easy than with the EFL data to suggest that there is much con-
sistency across the CEFR levels in different countries. While in Spain and
Greece learners seem to need to know, on average, over 2000 French words
to progress beyond the elementary A1 and A2 levels, the British data sug-
gest that this can be achieved with less than half this number, fewer than
Table 12.5 French as a foreign language vocabulary size
and the CEFR among learners in Greece
CEF level Mean Max. Min. SD n
A1 1125.71 2550 0 620.40 35
A2 1756.25 2500 1500 398.60 8
B1 2422.72 3400 1800 517.37 11
B2 2630.00 2850 2250 251.49 5
C1 3212.50 3750 2600 473.24 4
C2 3525.00 4150 2900 883.88 2
Table 12.6 French as a foreign language vocabulary size
and the CEFR among learners in Spain
CEF level Mean Max. Min. SD n
A1 894.44 2850 350 604.61 18
A2 1700.00 2750 500 841.50 9
B1 2194.44 3100 1100 717.39 9
B2 2450.00 1
C1 2675.00 3600 1900 643.23 6
C2 3721.42 4200 3200 416.19 7
9780230_206687_13_cha12.indd 2029780230_206687_13_cha12.indd 202 2/4/2009 2:53:17 PM2/4/2009 2:53:17 PM
James Milton and Thomaï Alexiou 203
1000 words. In all cases the figures suggest that many fewer words in French
are required to achieve this level of proficiency than in EFL. As with the EFL
data, the mean scores, and general tendency of groups, disguise the wide
variety of individual scores which the standard deviation figures illustrate.
Vocabulary size and CEFR levels in Greek
The mean vocabulary size scores at each CEFR level from the 64 Greek as
a foreign language learners at CEFR levels A1 A2, B1 and B2 are presented
in Table 12.7. As with the French figures, there are no guideline scores for
what vocabulary sizes might be associated with each level.
As with all the other sets of data, the Greek figures reveal a hierarchy of
vocabulary size scores for each successive CEFR level where we have results.
Again, an ANOVA confirms the relationship between CEFR levels and dif-
ferences in vocabulary size, F(3, 60) 57.150, p .01, and the differences
between the mean scores at each level are significant. The mean scores at
each level in these data are larger than the scores in the EFL and French as
a foreign language data. Again, there is considerable variation of individual
scores within each level and overlap in vocabulary scores between the levels.
Vocabulary size and CEFR levels
The data from users of the CEFR system, collected from four countries and
three different foreign languages, show what one would expect. As learners
get better in their foreign languages, and move upwards through the CEFR
levels, they tend to know progressively more vocabulary. Regression analy-
sis allows the relationship between a learner’s vocabulary size and the CEFR
level he or she has attained to be modelled and suggests just how strong
the relationship between the two variables can be. A series of these analyses
have been carried out on the data collected for this chapter and give the
results shown in Table 12.8.
It appears that in Spain and Greece the CEFR level a learner achieves is
particularly sensitive to their vocabulary knowledge; 60–70 per cent of variance
Table 12.7 Greek as a foreign language vocabulary size and
the CEFR among learners in Greece
CEF level Mean Max. Min. SD n
A1 1492.10 2400 500 705.58 19
A2 2237.50 3150 1500 538.58 12
B1 3338.23 4150 1950 701.13 17
B2 4012.50 4750 3450 415.33 16
9780230_206687_13_cha12.indd 2039780230_206687_13_cha12.indd 203 2/4/2009 2:53:18 PM2/4/2009 2:53:18 PM
204 Vocabulary Studies in First and Second Language Acquisition
in CEFR levels can be explained by vocabulary size. In Britain there is still
a strong relationship and over 40 per cent of variance can be explained in
this way. This observation fits well with other observations (Milton, 2006;
Richards, Malvern and Graham, 2008) that exam success in foreign lan-
guages in Britain is related to vocabulary size. Only in Hungary does the
strength of this relationship diminish. It is not immediately obvious why
these data should be so very different from the others.
In the EFL data it appears that there is some agreement on actually what
levels of vocabulary might be associated with each CEFR level, at least at
the lower levels. Thus, the British vocabulary size test writers and schools in
Hungary and Greece appear to agree that learners at A1 and A2 level prob-
ably know less than 3000 of the most frequent words in English. Learners
at B1 level will know about 3000, and learners at B2 level will know about
3500. Statistics can be misleading and we are dealing with small samples
here. Nonetheless, the differences in the mean EFL vocabulary scores in
Greece and Hungary at levels B1 and B2 were not statistically significant. At
advanced level C1, on the other hand, the Greek and Hungarian vocabu-
lary scores are significantly different (t(15) 3.092, p .01). Thus, while
the Hungarian mean falls within the range suggested in Meara and Milton
(2003), the Greek mean is well below it. The reason for this is considered
later in the chapter.
At first sight the French results are more varied, but this is due to the
influence of the British data. The Spanish and Greek data coincide closely at
almost every CEFR level. The numbers are small but the differences between
the means at every level in the Spanish and Greek data are not statistically
significant. It is the British data which differ markedly from the other two
and the mean vocabulary scores at every level of the CEFR are lower in the
UK than elsewhere. It was pointed out at the outset of the chapter that
foreign language exams, and the CEFR levels they have been placed at, have
been criticized within the UK. Given the relationship between vocabulary
size and overall language knowledge and skill, it would seem that the CEFR
Table 12.8 Linear regression modelling the relationship between vocabulary size
and CEFR level a
Learners R R2 Adjusted R2 Standard error
of estimate
EFL learners in Greece .842 .708 .705 0.9465
EFL learners in Hungary .417 .174 .168 0.5229
French FL learners in Britain .664 .441 .437 0.7065
French FL learners in Greece .809 .654 .648 0.8562
French FL learners in Spain .825 .681 .675 1.0519
Greek FL learners in Greece .844 .713 .708 0.8480
aAll regressions are statistically significant.
9780230_206687_13_cha12.indd 2049780230_206687_13_cha12.indd 204 2/4/2009 2:53:19 PM2/4/2009 2:53:19 PM
James Milton and Thomaï Alexiou 205
levels in French have been interpreted very differently in Britain than else-
where in Europe.
A characteristic of all the French data is that mean vocabulary scores for
the CEFR levels are lower than for EFL. We have only one set of data for
learners of Greek as a foreign language, but these figures are higher than
those in either EFL or French. This, of course, raises the question of how the
CEFR levels in the different languages are to be compared. Does the lower
vocabulary score associated with French CEFR levels, for example, mean
that the French levels are lower and much easier to achieve than the EFL
ones, or do these differences reflect some systematic difference between the
languages whereby fewer lexical resources are needed in French to achieve
the kind of communicative skill that the CEFR levels describe?
Comparing vocabulary sizes across languages
There is evidence that there may be systematic differences between the
vocabulary sizes required for the CEFR levels in different languages. For
example, it may be possible to achieve certain levels of competence in a
foreign language, such as reading with full comprehension, with fewer
words in French than is possible in English. It is commonly accepted that
full comprehension in a skill such as reading will require the reader to rec-
ognize almost all the words he or she encounters. A figure of 95 per cent
of the words in a text for general comprehension (Laufer, 1989), or 98 per
cent for reading for pleasure (Hu and Nation, 2000), are the kind of fig-
ures which are often quoted. Nation further suggests (2001, p. 147) that
there is a threshold at about 80 per cent coverage which is required for gist
understanding. But it is possible to achieve this kind of coverage with fewer
words in French than in English. And it seems that rather more words in
Greek are required to achieve this figure than in either French or English.
The reason for this is that languages are different in structure and the
ways words are created and used. Some of the most frequent words in
English are prepositions like of and up, for example, but other languages
inflect much more than English and these prepositions are likely to be
absent from the frequency lists in, say, Hungarian or Finnish and that will
affect the coverage of the most frequent words in these languages. The
most frequent preposition in French, de, is much more frequent than any
equivalent in English. More relevant to English and French, is that English
is a language where speakers can reputedly use a particularly large vocabu-
lary and often appear to have a variety of words available for just a sin-
gle idea (e.g. Bryson, 1990, p. 61). Part of the reason may be historical.
English differentiates, for example, between many farmyard animals and
the meat which comes from them, between pork and pig and between sheep
and mutton. English too appears to have two different sets of vocabulary
available for formal occasions, such as writing an essay, and for less formal
9780230_206687_13_cha12.indd 2059780230_206687_13_cha12.indd 205 2/4/2009 2:53:19 PM2/4/2009 2:53:19 PM
206 Vocabulary Studies in First and Second Language Acquisition
occasions, such as telephoning friends. An argument among friends might
be called a controversy in academic circles. The train which, to you and me,
stops at Paddington, terminates there in the language of the train conduc-
tor who is being formal. Regardless of whether English speakers really do
have very large vocabularies to work with, it appears that not all languages
make this formal and informal language distinction, or at least not in the
choice of vocabulary, in the way English does. It seems that in French the
most frequent vocabulary does the service both of everyday language and
the specialist academic vocabulary which English requires. Thus, Cobb and
Horst (2004) point to the coverage provided of academic texts by the most
frequent 2000 words in French. The figure they quote of nearly 89 per
cent (p. 30) would be equivalent to the General Service Word List of 2000
words in English plus Coxhead’s Academic Word List of some 600 words.
Arguably, 2000 words in French will do the work of some 2600, carefully
selected rather than purely frequency-based, words in English. To help
illustrate this we have plotted the coverage provided by frequency lists in
English and in French as shown in Figure 12.1. It is apparent from this that
the figure of 80 per cent coverage required for gist understanding requires
2000 words in English but substantially less in French, maybe only about
1500 words or fewer.
It may be possible to use the differences in coverage which frequency
lists in different languages provide, to understand how the vocabulary size
requirements of the CEFR levels might vary between these languages. If
achievement of B2 level in EFL requires about 3000 words, which would
provide about 85 per cent coverage of normal texts, then the volume of
0 500 1000 1500 2000 2500 3000 3500
Words by frequency
% Coverage
Figure 12.1 A comparison of coverage of text between Carroll et al.’s (1971) corpus
of English and Baudot’s (1992) French corpus
9780230_206687_13_cha12.indd 2069780230_206687_13_cha12.indd 206 2/4/2009 2:53:20 PM2/4/2009 2:53:20 PM
James Milton and Thomaï Alexiou 207
vocabulary in French producing the same coverage, perhaps 1800 words,
might be expected for the same level. The information provided by the
original CEFR word lists broadly supports this idea. While the original
Threshold (B1) level word lists in English and French were both in the
region of 2000 words, the figure for English is higher at about 2200 words
(van Ek and Trim, 1990), and for French is lower at about 1800 words
(Coste et al., 1987). It might be argued, therefore, that the vocabulary size
figures for French are likely to be lower than their EFL equivalents at the
CEFR levels above the most basic.
How might Greek as a foreign language compare? Does Greek pro-
vide figures which might also suggest a systematic difference? Figure 12.2
overlays the line for coverage from Carroll, Davies and Richman’s (1971)
corpus of English with the lemmatized Hellenic National Corpus’s cover-
age (Hatzigeorgiu et al., 2001) and provides something like an equivalent
list. At the outset the first few words are comparatively more frequent in
Greek than in English; in Greek the definite article is very highly frequent
even compared to English. Thereafter, Greek vocabulary provides propor-
tionately less coverage and the two plot lines cross over (see Figure 12.2).
The most frequent 5000 words in Greek provide about 83 per cent coverage
which is substantially less than in English. A particular feature of Greek is
the very high number of hapax legomena (words which occur only once in
a corpus) which comprise 49.4 per cent of the corpus in Greek but is nearer
to 30 per cent in English and French (Mikros, personal correspondence).
It appears from these data that rather more words are required in Greek
than in English for any level of coverage beyond the smallest, and that
more words would be needed in Greek to achieve the levels of communicative
Figure 12.2 A comparison of text coverage between Carroll et al.’s (1971) corpus of
English and the Hellenic National Corpus (Hatzipeorgiu et al., 2001)
0 500 1000 1500 2000 2500 3000 3500
Words by frequency
% Coverage
9780230_206687_13_cha12.indd 2079780230_206687_13_cha12.indd 207 2/4/2009 2:53:21 PM2/4/2009 2:53:21 PM
208 Vocabulary Studies in First and Second Language Acquisition
ability that fewer words in English would allow. It might be expected, there-
fore, that levels of vocabulary knowledge required for the various CEFR
grades would be higher in Greek than in either English or French. What
happens when the vocabulary scores for the different languages at each
CEFR level are compared with each other? Do the differences in vocabulary
size which these coverage differences suggest, emerge in the CEFR frame-
work? In Table 12.9 we have presented all the vocabulary size mean scores
at each of the CEFR levels.
Encouragingly, these data support the differences which coverage fig-
ures suggest should occur. At every CEFR level the mean French vocabulary
scores are smaller than the mean scores for EFL at the equivalent levels and
at every level where we have data, the mean Greek vocabulary scores are
higher than both the mean EFL and French scores.
Discussion and conclusions
At one level these results have produced exactly what was hoped for and
expected. As learners get better in their foreign language, and become
more skilled, able and communicative, they tend to know more words. The
vocabulary size scores which emerge suggest that certain levels for vocabu-
lary knowledge are associated with performance at each CEFR level. This
supports the idea that the CEFR system can work in establishing equiva-
lent levels in foreign languages across different countries and examinations
systems. The EFL data in Greece and Hungary broadly conform well to the
vocabulary levels suggested by the writers of the vocabulary testing software
at each of the CEFR levels. The EFL system, at least in Greece, has the ben-
efit of being tied strongly to the Cambridge testing system which itself has
a vocabulary level attached to it in the form of Hindmarsh’s (1980) list. This
list of 4500 words and phrases should form the basis of test construction at
the Cambridge FCE (B2) level and the use of this list has probably helped fix
the standard of this exam over time. The mean vocabulary scores that learn-
ers produce at this level, approximately 3500 out of the 5000 most frequent
words, fit well with the kind of vocabulary size implicit in Hindmarsh’s list,
Table 12.9 Summary of mean scores for each CEFR level in three foreign languages
CEF level French in French in French in EFL in EFL in Greek in
UK Spain Greece Greece Hungary Greece
A1 894.44 1125.71 1477.27 1492.10
A2 1700.00 1756.25 2156.81 2237.50
B1 952.04 2194.44 2422.72 3263.63 3135.90 3338.23
B2 1882.58 2450.00 2630.00 3304.54 3668.42 4012.50
C1 2675.00 3212.50 3690.90 4340.00
C2 3326.47 3721.42 3525.00 4068.18
9780230_206687_13_cha12.indd 2089780230_206687_13_cha12.indd 208 2/4/2009 2:53:21 PM2/4/2009 2:53:21 PM
James Milton and Thomaï Alexiou 209
which includes not only the most frequent words required for communica-
tion but also words tied to the kind of thematic requirements of the FCE
exam and which lie outside the most frequent 5000 words of English.
French does not have a fixed point of reference like the Hindmarsh list to
help establish vocabulary norms at each CEFR level. Nonetheless, it is pos-
sible to argue that the French foreign language data from Greece and Spain
suggest that the CEFR’s skills-based criteria have allowed very similar levels
of knowledge to be tied to the CEFR levels. This suggests that the system
can be quite workable. The British data, however, reveal the weakness of
the skills-based criteria when used in isolation from more objective evalua-
tion methods. The British scores for learners of French are not just different
from the Greek and Spanish scores, they are so different that the abilities
of the learners in Britain cannot possibly be equivalent to learners at the
equivalent CEFR levels in Greece of Spain. The presence of a vocabulary
knowledge indicator will surely help the British system in evaluating where
it stands in relation to the CEFR and in adjusting its level appropriately so
that it will fit more convincingly within the framework. There appears,
therefore, to be a real place for these vocabulary size measures.
The vocabulary score hierarchies which have emerged from this exercise
appear to be different between languages. The EFL scores are higher than the
French foreign language scores, and the Greek vocabulary scores are higher
than both. There is no reason for thinking that the achievement of a level
of competence in the CEFR system should require a single vocabulary size
in all languages. Languages differ and it is quite likely that it is possible to
be rather more communicative and fluent with fewer vocabulary resources
in some languages than in others. It appears possible to argue, however, that
this kind of variation is linked to coverage which also varies from language
to language. The volume of data represented here is small but it suggests
that the CEFR levels are associated with levels of coverage of text, and that
these coverage figures will allow us to estimate vocabulary size equivalences
between languages. To progress from elementary, A1 and A2 levels, for
example, it seems that learners need to know a volume of vocabulary which
will give more than 80 per cent coverage. In EFL that would require knowl-
edge of over 2000 words and in French rather fewer. This in turn means that
vocabulary size guidelines can be produced across the languages to which
the CEFR is applied, tying it together in a way this is not possible at the
moment. At the moment we assume that the French learners at B2 level in
Greece and Spain, for example, are similar in performance and knowledge
to EFL learners at B2 level in Greece and Hungary, but we have no real way
of demonstrating this without reference to something like vocabulary levels.
It is early days, but this method of rationalizing how vocabulary size scores
in difference foreign languages might link to the CEFR looks promising and
would merit more systematic investigation with larger numbers of learners,
in more countries and learning more foreign languages.
9780230_206687_13_cha12.indd 2099780230_206687_13_cha12.indd 209 2/4/2009 2:53:22 PM2/4/2009 2:53:22 PM
210 Vocabulary Studies in First and Second Language Acquisition
It should not be thought, however, that a set level of vocabulary is a
requirement of achieving CEFR levels; for example, that EFL learners must
have 3500 words before they can achieve CEFR B2 level. The relation-
ship between vocabulary size and level of attainment need not be fixed
in this way. Vocabulary size scores are likely to be indicators of knowl-
edge and attainment rather than an absolute determiner of these things.
It has been commented on several times in this chapter that while the
mean scores for each CEFR level appear to vary predictably and to tell a
comprehensible story about how vocabulary knowledge and attainment
develop, there is nevertheless considerable variation in vocabulary scores
at every level of the CEFR. There are several reasons why the relationship
between vocabulary size and skill in communication of language perform-
ance need not be fixed.
One reason for the slightly messy individual data which this investiga-
tion has produced is the imperfect way learners are assigned to their classes
in foreign languages, and the idiosyncratic way they may progress. It has
been assumed that learners have been assigned correctly to classes and that
every individual in a B1 level class, for example, is really at B1 level. In real-
ity there is no guarantee that this is the case. Learners can be assigned to
a class for many reasons other than level of knowledge and performance.
They can be grouped with other learners of the same age, for example, or
to keep a group of friends together. Again, learners may have been assigned
to the closest practical level even if it is not the correct one. Where a school
contains bilingual learners or the children of native speakers of the foreign
language alongside beginners, for example, it may not be practical or finan-
cially possible to arrange classes across the entire range of language ability
and for every year in a school. And again, once the class has begun learners
can progress at very different rates according to their interest and motiva-
tion. Even where a class begins a year’s study at the same level of ability,
some learners will always make better progress than others. It seems inevi-
table in this kind of research, therefore, that learners of different levels will
be grouped in a way that makes the results less clear.
An additional factor which is likely to obscure the relationship between
vocabulary size and CEFR level, is that language testing is not direct or
precise. In language testing we are dependent for valid results not only
on the creation of good tests to reveal aspects of language proficiency, but
also on the ability of the learner to play along with the system and will-
ingly and correctly show what they know. This is not always easy. Learners
may not be interested in the test, or they may become bored, tired or ill
and misrepresent their knowledge. Equally, they may choose, particularly
in objective-style testing, to make educated guesses about their answers in
order to gain the highest possible score rather than the score which most
accurately reflects their knowledge. Some variation, it seems, is just an inev-
itable consequence of the language testing system.
9780230_206687_13_cha12.indd 2109780230_206687_13_cha12.indd 210 2/4/2009 2:53:23 PM2/4/2009 2:53:23 PM
James Milton and Thomaï Alexiou 211
These factors should not disguise a third reason for why vocabulary size
and ability level are not precisely tied and this reason is rather more impor-
tant in this context. This is that the relationship between foreign language
knowledge and the ability or skill in using that knowledge may vary from
one individual to another. Communicative ability in reading and listen-
ing, for example, rests to a degree on anticipating what is likely to come
next and in making intelligent guesses as to the meaning of the writer or
speaker. Some learners can use limited data and achieve comprehension
more easily than others who require more complete knowledge to draw
the same conclusions. Likewise, some foreign language users manage to be
much more creative and intuitive than others in their ability to use the lim-
ited language knowledge they have for communication. We have very little
understanding of this type of variation and have no real way of character-
izing it usefully. As a consequence we currently find it hard to explain away
completely satisfactorily the range of vocabulary scores that learners in the
same class, or at the same level, can produce, and this is an area that bears
further investigation.
For these reasons the kind of vocabulary data which is likely to emerge
and be most useful for the CEFR system will be ranges of vocabulary knowl-
edge associated with the CEFR levels, and which will act as guidelines.
Groups of learners might be expected to conform to these guidelines quite
well since the progress of vocabulary knowledge among groups of learners is
now becoming quite well understood. Individuals are likely to be less pre-
dictable, however, and while it is unlikely that learners will depart enor-
mously from the guidelines, some individuals are likely to fall outside any
vocabulary range that is set.
In conclusion, therefore, it seems that it is quite workable to put vocabulary
knowledge measures back into the CEFR. While specifying lists of required
vocabulary may no longer be appropriate, a vocabulary size metric can offer
much to the framework. The vocabulary size scores which emerge among
learners at different levels of the framework are relatively predictable and
understandable, and it appears that vocabulary size estimates are already asso-
ciated with each of the CEFR levels, even if users of the system are not aware
of this. This chapter has been able to codify what some of these levels are.
We have even suggested a way of handling and explaining the way vocabular-
ies will vary between languages so the CEFR system can remain generalizable
across all languages and countries. This process has already revealed the kind
of discrepancy to which a system without an objective style of measurement
is prone, and it has highlighted the way the British placements of foreign lan-
guage qualifications appear very different from the kind of expectations which
are common on the rest of the continent of Europe. By reintroducing a vocab-
ulary size measure to the CEFR the system can, very likely, be made more
robust so that misplacements of this kind can be recognized and corrected.
9780230_206687_13_cha12.indd 2119780230_206687_13_cha12.indd 211 2/4/2009 2:53:23 PM2/4/2009 2:53:23 PM
... Coverage information usually informs the lists of core, or basic, vocabulary which, in turn, are used to inform teachers and materials writers for MFL. In French, the français fondamental list (Gougenheim 1958) contains just over 3 Milton and Alexiou (2009) tested the French vocabulary sizes of learners at B1 level in both Spain and Greece and results suggest that these learners can and do routinely achieve the vocabulary sizes of over 2 ¶ 000 words anticipated by coverage figures and core vocabulary lists. Their studies in UK with learners taking the GCSE exam yield much smaller figures of fewer than 1 ¶ 000 words. ...
... However, a likely consequence is that learners using this series will acquire even smaller vocabularies than those already noted among learners taking and passing GCSE French (e.g. David 2008;Milton 2006b ¶ ), and these vocabularies are already very small in international comparisons (Milton and Alexiou 2009). ...
The vocabulary content of the French MFL curriculum in England has received criticism (Häcker, M. 2008. Eleven pets and 20 ways to express one’s opinion: the vocabulary learners of German acquire at English secondary schools. The Language Learning Journal 36, no. 2: 215–26; Tschichold, C. 2012. French vocabulary in Encore Tricolore: do learners have a chance? The Language Learning Journal 40, no. 1: 7–19) for its small size, lack of thematic variety, and poor pedagogical exploitation in textbooks. Studies also note the small vocabulary knowledge of learners in UK compared with other French learners, ostensibly at the same CEFR B1 level (David, A. 2008. Vocabulary breadth in French L2 learners. The Language Learning Journal 36, no. 2: 167–80; Milton, J. 2006b. Language lite: learning French vocabulary in school. Journal of French Language Studies 16, no. 2: 187–205). This study examines the vocabulary content of the Studio series of French textbooks, first published in 2010, to see whether this situation has improved. Analysis shows the vocabulary loading of the textbooks has fallen over 40% compared with the previous books analysed by Tschichold (2012. French vocabulary in Encore Tricolore: do learners have a chance? The Language Learning Journal 40, no. 1: 7–19). This indicates a significant decline in the standard of learning for GCSE French since the vocabulary size of learners is so closely associated with their overall language proficiency. There are problems too with the thematic variety of teaching, the regularity of input, and the recycling of the lexical content. We argue this results from the guidance of advisory bodies such as the National Centre for Excellence for Language Pedagogy (NCELP) and the latest curriculum design (Department for Education. 2015. Modern Foreign Languages GCSE Subject Content. (accessed 30 August, 2020)) MFL subject content so there is no clear standard for textbook writers to work to. Further, the guidance of advisory bodies such as the National Centre for Excellence for Language Pedagogy (NCELP) and the latest curriculum design (Department for Education. 2021. GCSE MFL Subject Content Review. (accessed 14 March, 2021)), which damage standards of attainment in MFL by advocating minimal vocabulary content and a departure from thematic input.
... In addition, accurate representations of lexical knowledge in second languages are difficult to obtain. Equivalent 'age-of-acquisition' norms differ depending on the first language of the learners (Milton & Alexiou, 2009) and definitions of "word knowledge" can vary (see Treffers-Daller & Milton, 2013, for a discussion). Vocabulary estimates of second languages are often underestimations of learners' actual knowledge, depending on how vocabulary size is measured (e.g., Gráf, 2017;Hopkins, 2006;Webb, 2008), and the data used by the present study is surely no exception. ...
Full-text available
A recent study by Siew and Vitevitch (2020a) investigated word form lexica and their growth in children acquiring English and Dutch as first languages from a network perspective. They identified a unique developmental trajectory in network growth, with high-density neighborhoods becoming enriched through growth at early acquisition stages (the "preferential attachment" mechanism) but low-density neighborhoods gaining new neighbors at advanced acquisition stages (termed "inverse preferential attachment"). Their findings were confirmed for various languages, they fit with assumptions of cognitive efficiency in lexical memory and 2 retrieval, and thus are intriguing for second language research as well. The present study was designed as a replication of Siew and Vitevitch's (2020a) study "An investigation of network growth principles in the phonological language network" with data of English-as-a-second-language learners. Results mirror findings by Siew and Vitevitch and demonstrate that preferential attachment is the main network growth algorithm driving lexical learning at early second-language proficiency stages, while inverse preferential attachment prevails at more advanced proficiency stages. The similar growth dynamics observed in phonological networks of first and second language users may indicate a universal cognitive principle underlying word learning.
... Hoff and Naigles (2002) in Henriches (2009) concluded that "higher levels of quantity, lexical richness and syntactic complexity of the input" boosts the productive vocabulary of very young learners. (p.2) Milton (2009) identified a positive relationship between the frequency of the words in use and the learning of foreign language vocabulary. (p.198) ...
Experiment Findings
Full-text available
Students' interest in improving their repertoire of vocabulary and the habit of reading for pleasure and profit seems to have reached a nadir in the current era of social media. As a consequence, students tend to display noticeable inadequacy of comprehension and communication. Several research projects have been carried out to find out suitable methods to enrich and enhance the vocabulary repertoire. The researchers have attempted the present study to find out whether using complex passages and different instructional methods to teach vocabulary enhances learning and internalization of the target vocabulary. Jigsaw, a student-centered method of instruction and the traditional method of instruction were adopted for both the groups and the design followed was repeated measures. The paired t test of the data collected shows invariably the unavailing nature of both methods of instruction and complex input on students' vocabulary. This research work concludes that teaching methods tend to have insignificant impact on vocabulary enhancement when complex input passages are chosen.
... Again, anticipating that participants would attain B1 level (CEFR) at best, we assumed that learners would acquire a maximum of approximately 3,000 words (Milton & Alexiou, 2009). ...
... The following table shows the typical vocabulary sizes recommended for students at various proficiency levels in English (for details, see Milton & Alexiou, 2009): Many techniques for the use of music rely on students recognizing key words and phrases from a song. The activity introduced here provides a means for students to demonstrate recognition of vocabulary and phrases in connected speech. ...
Contribution to a book of ESL teaching ideas.
... Using a French version of the X-Lex, Milton (2006) and Milton & Alexiou (2009) measured the vocabulary size of learners of French from the UK, Greece, and Spain. Their results were tied to the Common European Framework of Reference for Languages (CEFR). ...
Full-text available
The present study set out to investigate the receptive vocabulary knowledge of Moroccan learners of French. The study used the French XK-Lex test, an adjusted French version of Meara and Milton's (2003) X-Lex vocabulary test. The sample consisted of 128 students: 40 primary school 6th graders, 47 junior high school 9th graders, and 41 senior high school 12th graders, from both public and private schools. The findings show that learners' vocabulary development does not adhere to frequency-based rules. Learners' mean scores throughout the ten frequency bands (1K through 10K) form a fluctuating instead of a downwards slope from the first band to the second, the second to the third, and so on. Additionally, learners had a vocabulary size of less than 4,000 from the most frequent 10,000 French words. The results also show that learners' grade level significantly affects receptive vocabulary size. As learners progress from the 6th grade to the 9th grade and, eventually, to the 12th grade, their vocabulary size increases. Furthermore, the study's results show that private school learners statistically significantly outperformed public school learners in their receptive vocabulary size. This study also identified that gender does not have any statistically significant effect on learners' receptive vocabulary size. Both males and females seemed to develop their receptive vocabulary size in much the same way. Finally, the findings indicated that out-of-class exposure to French affects the amount of receptive vocabulary developed. Learners who were exposed to French outside the classroom performed statistically significantly better than those who were not.
... We developed an odd-one-out task in which participants were presented with lists of five words and had to decide which one is semantically most different from the rest. Again, anticipating that participants would attain B1 level (CEFR) at best, we assumed that learners would acquire a maximum of approximately 3,000 words (Milton & Alexiou, 2009 on the total number of words produced (irrespective of accuracy), morphosyntactic accuracy (developmental errors), target-like use of vocabulary, and lexical richness (see Table 3). (2010) Note. ...
Full-text available
The question of cognition in second language (L2) acquisition later in life is of importance inasmuch as L2 learning is largely mediated by domain‐general cognitive capacities. While a number of these capacities have been shown to decline with age, individual differences in cognition increase over the lifespan. This microdevelopment study investigates the L2 trajectories of 28 older German‐speaking adults (age 64+) who participated in a combined computer‐assisted and classroom‐instructed 7‐month Spanish training for beginners. We made use of generalized additive mixed models (GAMMs) to quantify linear and nonlinear learner trajectories as well as any predictors thereof. Participants were assessed on a range of behavioral, L2, socioaffective, and background variables. We found a significant (linear and nonlinear) increase across all measures of L2 proficiency. Between‐subject cognitive, socioaffective, and background variables significantly predicted the overall level of L2 proficiency as well as developmental patterns over time. Daily variances in cognitive performance and socioaffect had little impact on fluctuations in L2 performance. Findings are discussed against the backdrop of complex dynamic systems theory and highlight the necessity for dense longitudinal research designs to capture nonlinearity in third‐age L2 learning.
... The next stage which is currently underway, is to use and standardize them in both languages in order to investigate their potential to be mapped to CEFR levels. A vocabulary size metric can offer much to a more principled and systematic way of assessment and can impact their future educational attainment (Milton & Alexiou, 2009). It is true that the inclusion of more parts of speech (not just nouns) may improve the item and sampling validity and it will certainly affect the frequency results but that would be useful only if we are to distinguish between the whole range of CEFR (Pre-A1 to C2); this is part of a future project. ...
Full-text available
A series of recent research studies highlight the significance of vocabulary and suggest that vocabulary development is paramount in the successful development of a foreign/second language even by very young learners (Biemiller, 2003; Alexiou, Roghani & Milton, 2019). The number of words a learner knows is considered a decisive factor in the performance in all aspects of language (Webb & Nation, 2017). For this reason, modern curricula now seek to be more precise over the volumes of vocabulary that learners are expected to acquire at all levels. Therefore, measuring vocabulary size systematically is vital both in monolingual and multilingual settings while it can lead us to a much needed model of vocabulary acquisition at an early age. The particular paper presents an updated version of Pic-lex (Alexiou, 2020), an easy to administer, dyslexia-friendly, theoretically well-founded online testing tool, which is intended for very young learners up to primary age. This game-like test makes assessment of receptive vocabulary size from picture cues. The vocabulary word items are not randomly selected; words are based on updated frequency wordlists while the scores provide an estimate of receptive vocabulary knowledge that can be matched to CEFR levels.
Developing L2 learners’ productive mastery of vocabulary is a challenging task. Recent research has called for greater attention to understanding how receptive vocabulary may be transformed for productive use (Schmitt, 2019). Using a design-based research methodology this study investigated adult ESL learners’ productive oral vocabulary development through engaging them in a series of classroom workshops where they were exposed to nine target words in five different contexts. Findings suggest that such exposure to words combined with phonological form-focused elaboration facilitates the development of metalinguistic awareness, specifically the associations between grammatical patterns and word meaning, leading to subsequent productive use of target words. The paper contributes to the understanding that vocabulary training combining a focus on meaning with a focus on phonological and grammatical form may enhance form-meaning mapping leading to productive oral vocabulary development.
The present study involved the analysis of 81 second language instructed vocabulary acquisition (L2 IVA) studies over 2 phases. In Phase I, we categorized and coded the effect sizes of the studies. Observing that the basic between‐ and within‐subject design dichotomy lacked the sensitivity to capture the heterogeneity of observed effects, we employed a more granular approach. In both between‐ and within‐subject designs, treatment versus comparison contrasts best represented comparisons of most interest in L2 IVA experiments, with median effect sizes (g) of .62 (between‐subject) and .25 (counterbalanced within‐subject). In Phase II, the aggregated effect sizes observed in Phase I were utilized in a priori power simulations to suggest approximate sample sizes for common L2 IVA analyses. For conservatively powered between‐subject designs, the simulations suggested sample sizes ranging from 292 to 492 participants. Counterbalanced within‐subject designs required 95 to 203 subjects depending on the assumed correlation between the repeated measures. The overarching implication of these simulations suggests that future L2 IVA experiments require larger samples that reference effect sizes from previous research, and we offer 3 potential solutions to the problem of obtaining larger samples.
ResearchGate has not been able to resolve any references for this publication.