[Corpora-List] Numbers of English vocabulary required for students

Brett Reynolds brett.reynolds at humber.ca
Mon Oct 2 13:59:01 CEST 2006

On Oct 2, 2006, at 3:22 AM, TadPiotr wrote:

> I am afraid that in work like that by Nation we can find the common

> pitfall

> of vocabulary-based word lists: common items have "little lexical

> meaning",

> it is infrequent items which really carry the message (that is why

> we can

> find the topic of a text by automatic retrieval procedures). It is

> sufficient to have a look at various vocabulary tests applied to

> American

> students to see that the range of the vocabulary they are expected

> to have

> is enormous, though there are obviously some preferred areas, such as

> Latinate and classical items, etc. Have a look at

> http://www.vocaboly.com/vocabulary-test/, for example. And there are

> hundreds of pages like that.

Indeed, native-speaker college students know tens of thousands of
words. The point is that you don't NEED this vocabulary to get by.
There is research indicating that people can guess many unknown words
and generally understand a written text if they know 95% of the word
families in it. That 95% number can often be reached in first-year
university and college textbooks with a combination of the first 2000
word families plus the academic word list. For example, using Michael
West's General Service List as the top 2000 words, an analysis of the
business textbook used by all first-year business students at my
college shows that the first 2000 word families give 80% coverage.
The academic word list gives another 10%. Proper nouns make up
another roughly 4%, giving 94% coverage.

If we keep in mind that the learners are *beginning* tertiary
education, and that much of first-year courses is dedicated to
learning the technical vocabulary of the field, (indeed, this
vocabulary is often glossed in the textbooks) then this group of 2500
words should represent a reasonable entry minimum. Of course, the
reading will not be easy, and a dictionary will be required, but it
will be possible.

Now, again we come back to the question of what it means to know a
word. In the research that produced the 95% number, 'knowing' was
operationalised simply as any word that a learner said they knew.



Brett Reynolds
English Language Centre
Humber Institute of Technology and Advanced Learning
Toronto, Ontario, Canada
brett.reynolds at humber.ca

More information about the Corpora-archive mailing list