From the docs:

"CORRUPT - Block is wrongly identified or is not a data block (for example, the data block address is missing)"

So a corrupt block is simply garbage. We went to read a block, but we couldnt even work out if it was an Oracle block or not. It could be anything.

"FRACTURED - Block header looks reasonable, but the front and back of the block are different versions."

When we write a block, we want to write an Oracle block (typically 8k) to disk. But the underlying OS might not support a single 8k write - it might do (say) 4k writes, so it breaks up the request into 2 x 4k writes. If one of them didnt work or got corrupted, then when we read the 8k block next time off disk, the first 4k of the block might look fine, but it wont match the last 4k of the block. The block is hence "fractured". Sometimes these are transient errors, for example, when RMAN encounters one it will normally try to re-read the same block/

The type of corruption typically doesnt matter too much - you would need to use standard block recovery techniques to resolve.