Closely related: Accuracy of a translation: how to forge an opinion?
Several authors, such as Samuel Beckett, periodically translated their own works. Are their translations generally regarded as accurate? Do people tend to find them more - or less - trustworthy than translations done by professional translators?
Have there been any examples of authors who did translations that included major changes to the work?
I would say firstly that this question relies on what you mean by the term 'accurate', as when we normally describe a translation as 'accurate' we mean that the meaning conveyed when translated is as close to the original as possible. However, in a literary work this is far more difficult to verify because, in the view of the majority of contemporary literary critics, meaning does not rest in the hands of the author. (I'd recommend reading Barthes' Death of the Author for the justification of this approach)
A translation by the author may highlight certain aspects of a text that the author intended to highlight in its original, for example the author may alter their translation to further the importance of certain themes that they deem important, but that readers or translators may not favor. So, translation by the author may maintain authorial intent more than a translation done by a professional translator, but if this is accurate depends on how much you value that intent.
Furthermore, some authors who do 'translate' their works do not even do so directly. Although not comprehensive, Haruki Murakami describes in What I Talk About When I Talk About Running how in 'translation' he writes his works twice; firstly in Japanese, and then again in English, with no attempt to directly convert the Japanese into English, as he feels that he must choose new, limited vocabulary which he finds to be creatively useful. In contrast, Seamus Heaney describes in the preface to Aeneid VI how whenever he translates either from his own work or from others, he treats it as 'schoolwork' and does attempt to do a translation as closely to the original as possible. Therefore the comparativeness of translation will depend on the author, as there is no perfect method, it seems.
It is questionable to what extent an 'accurate' translation of literary texts is even possible. Emily S. Apter in Against World Literature: On the Politics of Untranslatability highlights the fact that many languages have untranslatable words, or words of significance that do not carry over to other languages. Because of this it may be questioned to what extent it is possible to 'accurately' translate any literary text, although translators with a rather keen understanding of language may be at an advantage over authors in this respect, I would recommend reading the rationale usually described in the preface to most translated texts for an insight into how this problem can be approached.
In summary: it may be said that translation done by authors although rare may preserve authorial intent, but this only adds to accuracy if you favor authorial intent as a source of meaning, which would be regarded as a rather unusual view in contemporary literary criticism. If you don't favor authorial intent, this becomes mostly subjective as meanings one reader might pick up on may be preserved in one translation while other meanings some notice may be lost. Translation is not an objective process, and it never will be, so accuracy is dependent entirely on your personal reading of the original text. I know this may not feel entirely satisfying, but as with many problems in literature, there is not one perfect answer.
