How detailed should my review of a very poorly-written manuscript be?

I view my job as a reviewer to be about triage and the "first order" issues with the manuscript. Thus, for a very good manuscript, I will point out minor copy-editing issues, because those are the first-order issues remaining. For a not-so-good manuscript, I will note if the manuscript needs language editing but focus on the scientific issues instead, because any specifics of copyediting are likely to be obsoleted in any case by the larger repairs that are necessary.

Here are some intuitive thoughts:

• No, no one expects you to copyedit
• Your primary duty is to help protect the integrity of literature in your field. Therefore, as user6726 suggested, if the paper is hopeless, there's nothing there that should occupy a lot of your time.
• However, as a good citizen of your scholarly community, it would be good to "pay it forward" by giving some sort of actionable feedback that will help the authors to improve not just this paper but in general.

So, my recommendation would be to give brief, actionable feedback (maybe one or two paragraphs), focused on the biggest shortcomings that could be corrected in the authors' future work. A sentence or two for each of your bullet points would be quite adequate.

Wise authors will heed your terse feedback and improve; other authors don't care, so don't benefit from the attention you're giving them.

Meta-point: If you're marking up the hard copy as much as you say, and you're not the authors' English composition instructor, then you're also not using your time optimally.

I do not bother with detailed comments about writing unless there is reasonable hope that the submission could be made publishable. Even then, the main reason to include comments about writing is to make clear to the author to what extent the paper is unintelligible as written. Although it is (in some fields) traditional to list all of the punctuation, grammar and spelling errors, this is a waste of your time. I've frequently had to give very detailed comments on writing because the author had valuable material that should be made available, but they had no idea how to communicate effectively. In those instances where I've understood the point, I will take the time to explain why A is a better way to put it than B. But if the venue has a single-revision policy and the paper is a clear reject, then there is little point, unless you are encouraging the author to try again with a different journal.