Many publishers who allow authors to post preprints only allow the final draft of the submitted article, which will not have the appearance of the published version and which will differ in pagination. In the humanities especially this can be a drawback if a reader needs to quote with page references.
Yes, some publishers allow to post preprints on repositories if authors prefer to do so. This is actually a part of green open access model. However, authors have to ensure whether this is allowed by the publisher according to copyright laws. I do not see any ethical concern if the publisher allows to do so. Receiving the early attention (e.g. citations, views, downloads, etc.) of readers is one of the major benefits of posting preprints. Usually, journals take considerable time to production/first online appearence of an article from acceptance. This time lag can be used efficiently by posting preprints. Also, increasing the visibility of the research via depositing the article in multiple locations is another advantage. Since, preprint servers are free to access, readers can see the article without a cost.
Many publishers who allow authors to post preprints only allow the final draft of the submitted article, which will not have the appearance of the published version and which will differ in pagination. In the humanities especially this can be a drawback if a reader needs to quote with page references.
Yes, Karl has a good point. A reader could confuse when understing the exact page number (when using direct quotation) included in a citation, unless the reader identifies the correct version (author's version or publisher's version) the citation is based on. However, I think these two versions receive distinct DOIs since they are located under different services. If so, this could be a way to resolve this problem to some extent.
Preprint has advantages, as to take support and an early critic on the work, in order to any correction, as well to present it, because could be created statements that they will not permit to be published it later.