Page 10 - ELG2107 Jul Issue 476

RESEARCH NEWS .

Measuring language learning success

A new vocabulary list reflects how learners acquire additional words

By Gillian Ragsdale
The order of words acquired by
non-native learners is usually
assumed to reflect word frequency GLEN CARRIE ON UNSPLASH
in the English language generally,
but when Belgian researchers
tested word recognition directly
and compiled lists of word families
reflecting what students actually
know, their lists suggested other
strong influences on which words
are learned and in what order.
There have been many attempts
to compile lists of vocabulary
appropriate to various levels of
language proficiency. These lists
are often organised into levels of
word families. For example, the
Common European Framework
of Reference (CEFR) has six
proficiency levels (A1-C2).
What these lists have in
common is that they’re compiled
from the top down by researchers
using word frequency to decide
at which language learning level
those words/word families should
be known.
The Belgian study took a very
different, bottom-up approach,
testing language learners directly
to find out which words they tongues and all educational levels, child-friendly vocabulary, such as These lists, being derived from
actually know. with a mean age of 30. The most ‘tadpole’ and ‘dunce’. what learners actually know, can
An internet test was made common mother tongues were The list of 62,000 words were help teachers and examiners
freely available (see below), with Polish, Hungarian, German, organised into 20,000-word to assess text difficulty more
each testing round comprising 70 Polish, Dutch and Chinese. families, which is a more useful realistically and better anticipate
random, real English words and Analysing the responses led compilation for teachers and challenges, such as otherwise
30 random, made-up but plausible to a ranked list of 62,000 words. learners (see link below). The unexpected gaps in vocabulary
words. The participants answered Of these, 114 were known to all families were ranked by the best- knowledge.
yes/no, indicating recognition participants, but were still ranked known member of the family, and The vocabulary test can be
of the word, while the inclusion by responding time (eg, nouns include inflections and derivations found at http://vocabulary.ugent.be
of the nonsense words weeded out in ranked order: ‘coffee, water, based on suffixes. For example, Wordlists are available at OSF |
those who claimed to recognise music, radio’ etc). Some less the family ‘correction’ includes Word levels for English L2 speakers
words falsely. frequent, language-class related ‘correctly’ and ‘corrective’, but based on accuracy and response
Participants were also asked for words also appear on this list: not ‘incorrect’. times in a yes/no vocabulary test
personal information, including ‘subject, verb, vocabulary’. A Overall, English language with 62 thousand words.
choosing from five proficiency further 331 words had only one word frequency was only able to The lists are copyright protected
levels, from ‘I know a few words’ ‘no’ response. account for 46% of the variance under the Creative Commons
to ‘It is my mother tongue’. Some words would not have in the likelihood that learners license CC BY-NC (Attribution-
Feedback scores were given and been predicted from their knew a word. NonCommercial). They can
the test proved very popular, frequency in the English language Apart from academic-related be used freely for research and
generating 17 million responses generally and some were even biases in vocabulary, other education, but not for commercial
for the study. better known in English than influences include the motivation puposes.
Many more responses were in the first language. These of the learner and how interesting
actually collected, but only the latter tended to be academic the vocabulary is to them. REFERENCE
first three tested by each user were words, such as ‘informatics’, and This may explain why so many ■ Brysbaert, M., Keuleers, E. and
used (some users tested hundreds cognates from the mother tongue learners knew words such as Mandera, P. (2021), ‘Which words
of times) and anyone answering or other second language, such as ‘snowboarding’ and ‘sexy’. It is do English non-native speakers
‘yes’ more than twice to the ‘paracetamol’. also likely that sources of English know? New supernational levels
nonsense words was excluded. Words that were comparatively outside the classroom, such as based on yes/no decision’. Second
Participant profiles were very less well-known than expected TV, film and social media, are Language Research https://doi.
diverse, representing 150 mother included more informal and highly influential. org/10.1177/0267658320934526
10 July 2021