1 bit of entropy, where do you get that? If the attacker knows it's only upper or lower case characters, then it's 270 log2 52, or 1539 bits. If the attacker knows that it's composed of English words, then it's 53 log2 80000, or 836 bits, for 53 words and assuming a standard dictionary of 80,000 words. I suppose an attacker could use statistics on word patterns in English sentences to lower the overall entropy, since some words are more likely to follow other words, and some words never follow each other, but I imagine it would still be in the hundreds of bits. Maybe in that sense it would be 200-500 bits.