This one is a great example of the creativity of the LSAT-writer.
While this argument is riddled with conditional relationships, I would not advocate diagramming this question using formal notation. The reason is that the terms shift from one relationship to another. In order to link chains of conditional logic together, we need to see the same term within two statements. In a basic form
A --> B
B --> C
======
then we could infer
A --> C
But we would only do this if the terms were repeated across several conditional relationships.
In this one, we don't see terms appearing and linking the relationships together.
We need to knock out each incorrect answer by testing it against the statements. We're fortunate in that each of the incorrect answers is almost exactly paraphrases one statement from the stimulus - making the comparison fairly simple.
The question stem asks us to find the one that is LEAST supported. Meaning that four of the answer choices can be inferred from the stimulus, the one that cannot is the correct answer choice. Think of this question as "must be true EXCEPT".
(A) can be inferred from the first sentence of the stimulus. If only poetry cannot be translated well, then if it is not poetry it can be translated well.
(B) is correct. This answer choice distorts the first sentence a bit. We know that poets preserve language, but we don't know that a purpose of writing poetry is to preserve language.
(C) can be inferred from the first sentence within the part, "for we would not bother to learn a language if we could get everything written in from translation."
(D) can be inferred from the claim, "we cannot witness the beauty of poetry except in the language in which it is composed."
(E) can be inferred from the claim, "Only poetry cannot be translated well."