The search for a "survival" bank of vocabulary is valid, but extracting lists frequent words from dictionary corpora is questionable because these corpora are based on native speakers communicating with other native speakers. If we could do a frequency analysis of of non-native speaker English, this would provide a more useful bank of survival vocabulary. A corpus of non-native speaker English would not only reflect non-native speaker language but also the non-native speaker's  communicative needs in English.

That being said, I have found the 2,000 words of the Longman Dictionaries' Defining Vocabulary immensely useful. When working with an early Acorn BBC B computer, I had a simple word processor which contained no "SPELL CHECK" programme. I managed to find a spell check programme which had no reference dictionary, so I input all 2,000 words of the Longman Defining Vocabulary. After a year of writing, mainly for non-native speaker readers, I had added only about 180 words to this number! What better proof of the validity of those 2,000 most frequent words!

Your mention of Don Fabio: cue a snigger about footballers only posessing a hundred-word vocabulary.

I read somewhere that native English speaking graduates have a vocabulary of 29,000. So if we need 8,000 to understand authentic texts then where do all those other words go? Amazing to think.

On a slight aside - I often feel we don't exploit the English vocabulary enough - certainly as native speakers. There are so many wonderful words which give such a precision and subtlety to our expression but are derided for being elitist. It's perhaps a modern media thing that we can't use 'big' words and everything is dumbed down. I once worked in communications for an arts charity and there was a list of banned words. A shame I thought. I'd love to think it would be possible and practical to expose learners to all these seemingly pretentious words. Unrealistic though I fear.

Thanks to JJ, Diane, Norbert, Averil and Paul for their great session yesterday on this. Some really interesting ideas looking at not just hitting a set number of words, but it's the repetition and repeated exposure of the words in a variety of different contexts that is important too. We had a few questions via Twitter during the session yesterday but unfortunately ran out of time, so i'll post them here and please do post your comments and thoughts!

- When you say 3000 words for normal discourse, which 3000 words? How do we know which 3000 they need?

- Is it really about recycling and repeated exposure, or just about a good memory?

- Do you need the same 3000 to read as you do to speak at a certain level?

Emma