Re: How Much for Conversation?

On Tuesday 02 July 2002 07:13, you wrote:
>  --- Jim Grossmann wrote:
> > All other things being equal (mastery of correct spoken grammar, correct
> > pronunciation, etc.), about how many words (or contentive morphemes) do
> > you have to know before you feel comfortable using a target language for
> > everyday conversation?
> >
> > I'm not looking for an exact figure, of course.  But can you name a
> > ballpark figure, a rough estimate by the thousand?
>
> If you ask for a ballpark figure (whatever thay may be), I would say 2,000.
> At least, that's what my English teacher from school once told us. He
> estimated his own vocabulary at 9,000, IIRC.
>
> He said, that 2,000 words is enough for an average, everyday conversation;
> they allow you to understand almost everything, and to read a book. Of
> course, the number of words that you don't know will be huge, but they will
> be in such a minority, that you will probably understand the sentence
> anyway.
>
Yes, 2000 words is what you have to learn in the first two years of Chinese,
too. If you go beyond that in a second language, the disparity between
active and passive vocabulary gets greater. For instance, I hardly ever
encounter a word in English I don't understand, but my active vocubulary is
much smaller. I can't guess how small, though.

> While working on my conlangs, I have the impression that this figure is not
> far from the truth. For example, Hattic has currently almost 1,300 words,
> and for translation the relay text I needed to invent only five new words
> or so.
>
That seems to be really a kind of constant, since I have about the same number
of words, and the same experience.

> Besides, a large part of a language's vocabulary is self-generating. As you
> learn the language, you will see that you will be able to understand, and
> ultimately produce, a growing number of words and compounds you never
> learnt (for example: "fish market", or just about everything you can pick
> up from the matrix of verbs and prepositions, like "take out", or "carry
> away").
>
This too, although it pays not to start too early, i.e. when you've thirty
words to try and form all other lexemes from those thirty. I still have a few
extremely lenghty compounds from the beginning days of Denden in my
dictionary.

> You didn't ask for it, but I'll tell you, anyway. A very useful technique
> for learning words, is making small pieces of paper (let's say, 2 x 3 cm).
> You write down a word in the target language on one side, and the
> translation into your own language on the other. Then you make a package of
> them. Now you can start repeating the words by going through the package
> one by one, separating the word you know from those you don't know. After
> going through the whole package, you will have two new packages or piles.
> You perform the same procedure with the (probably smaller) package of words
> you don't know. Repeating this pattern, you will see that the package of
> words you don't know grows smaller and smaller, until you know everything,
> except maybe for a few words that you somehow just can't memorize.
Works a champ for Chinese characters, too, with the proviso that you need to
write down every word before you turn the flashcard, and that you need to
repeat the set you had wrong the next day, only dropping cards out of the
running set when they've been correct for a week. I still have about 2000
index cards with characters somewhere.

Before our Great Computer Crash of 2002 I had web app that did the same: it
took random words out of the Kura dictionary, showed either the form or the
gloss and asked the user to pick the right translation out of a set of four. I
learnt a lot of Valdyan before the computer crashed and I lost the
cgi-script.
--
Boudewijn Rempt | http://www.valdyas.org