To your first question: yes, without some kind of error correction.
The lower the protection class the higher the level of error correction.
It does not carry out any form of error correction.
Error correction does not seem to have a direct influence on learning a second language.
You're able to have a variable amount of error correction.
So the more error correction you have, the larger percentage of the whole area is not data.
The 'One sense per discourse' property can be used here for error correction.
They need another 3-4 orders of magnitude to do error correction.
These kind of surges are always related to error correction.
They expect to announce any release of the final results only after error corrections are complete.