In addition to the others answers given, I think that is it extremely important to state whether or not you consider the grammatical errors to affect the scientific substance of the paper.
Usually, grammatical errors don't actually effect your ability to evaluate the science in a paper. Even when phrases are fairly tangled or when a missing word makes a sentence say the opposite of what is intended, you can usually sort out what the authors intended, and judge them on their science, not their presentation.
This is an extremely important responsibility: do not judge a paper based on grammar and language issues.
Whatever grammatical issues you point out, be explicit that they are not the reason for your recommendation. Some reviewers will play language police, and recommend a paper be rejected because it is "sloppy." This is, in my opinion, inexcusable: grammar, no matter how tangled, can always be cleaned up, and should only be held against an author if they refuse to do such cleanup.
In those rare cases that things are so badly presented that you cannot understand the science, however, state clearly that is what has happened, and that this is why you are judging the grammar to actually affect the acceptability of the paper.