An Introduction to Applied Linguistics

part of a house
3 horse 
___________ 
animal with four legs
4 pencil 
___________ 
something used for writing
5 shoe
6 wall
The test has five sections covering various frequency levels, and so the results 
can help teachers decide what vocabulary level learners should be working on. 
Because teachers should deal with high-frequency and low-frequency words 
in different ways, the results of this test can also help teachers decide what 
vocabulary work they should be doing with particular learners or groups of 
learners.
The Productive Levels Test requires learners to recall the form of words using a 
sentence cue.
They keep their valuables in a va_________ at the bank.
The first few letters of each tested word are provided to help cue the word and 
to prevent the learners from writing other synonymous words. This test format 
is useful in showing whether a learner’s knowledge of a word has begun to move 
towards productive mastery.
The X_Lex and Y_Lex tests use a yes/no format where learners see a word on 
a computer screen and then have to decide if they know it. The tests include 
some pseudowords (like skemp) that look like real words, and learners’ scores 
are adjusted downwards by the number of times they say that they know these 
non-words. The tests give an estimate of overall vocabulary size and a profile of 
vocabulary known at each 1000 frequency band.
The Vocabulary Dictation tests each consist of five paragraphs with each 
successive paragraph containing less-frequent vocabulary. The test is administered 
like a dictation but only the 20 target words at each level are actually marked. 
There are four versions of the test. It can be used for quickly determining the 
extent of learners’ listening vocabulary.
The Vocabulary Size Test (Nation and Beglar, 2007) was designed as a proficiency 
test to measure total vocabulary sizes. Copies of the test can be found at http://
www.victoria.ac.nz/lals/staff/paul-nation/nation.aspx and at Tom Cobb’s website 
(www.lextutor.ca). It consists of 140 multiple-choice items with the stem 
containing the tested word in a non-defining context sentence. Although the test 
is divided into fourteen 1000-word levels, with 10 items at each level, a learner’s 
total vocabulary size is found by simply multiplying the learner’s score on the test 
by 100. A Rasch-based validation of the test (Beglar, in press) found:

---

46 An Introduction to Applied Linguistics
1 It can be used with learners with a very wide range of proficiency levels.
2 It measures what it is supposed to measure and does not measure other things. 
Beglar found that the test was very clearly measuring a single factor (presumably 
written receptive vocabulary knowledge) and other factors played a very minor 
role in performance on the text.
3 It behaves in ways that we would expect it to behave, distinguishing between 
learners of different proficiency levels, having a range of item difficulties related 
to the frequency level of the tested words, and clearly distinguishing several 
different levels of vocabulary knowledge so that learners’ vocabulary growth 
over time could be measured.
4 It performs consistently and reliably, even though circumstances change. In 
Beglar’s trialing of the test, these changes included comparing the performance 
of male subjects with female subjects, comparing 70-item versions of the test 
with the 140-item version, and comparing learners of various proficiency levels. 
Rasch reliability measures were around 0.96.
5 It is easy to score and interpret the scores.
6 The items in the test are clear and unambiguous.
7 It can be administered in efficient ways. When learners sit the test, they need 
not sit the whole test if the lower frequency levels are thought to be well 
beyond their present level of knowledge. Ideally, they should sit at least two 
levels beyond their present level of knowledge. So if learners are thought to 
have a vocabulary size of around 3000 words, they should sit the first five or 
six 1000 word levels.
The test works very well because it covers a very wide range of frequency levels, 
it includes a large number of items (even half of this number would work well), 
the items have been very carefully designed and made and the test is designed to 
measure just one kind of vocabulary knowledge.
As can be seen in the above examples of the various tests, there is a wide variety 
of vocabulary test formats. Different test formats testing the same vocabulary 
tend to correlate with each other around 0.7, indicating that test format plays 
a considerable role in determining the results of a vocabulary test. This also 
suggests that different test formats may be tapping different aspects of vocabulary 
knowledge. There are a number of issues that complicate vocabulary testing, and 
these are well covered by Read (2000) in his book devoted to assessing vocabulary 
and Schmitt (in press) in his vocabulary research manual.
Limitations On Generalizing Vocabulary Size 
Estimates And Strategies To Other Languages
It is worth pointing out that most of the research on vocabulary has been done 
within the broad context of English Language Teaching (ELT). This is rather 
unfortunate, since English is a very peculiar language in some respects, and 
particularly so as far as its vocabulary is concerned. This means that the findings 
reported in the earlier part of this chapter may not always be generalizable to 
other languages in a straightforward way.
The chief characteristic of English vocabulary is that it is very large. Consider, 
for example, the set of objects and actions that in English are labelled as: book,