There is no reason to assume that the field of Qur’an translation remains untouched by the commercial opportunities inherent in machine translation and the predatory use of online content. And, indeed, if you search Amazon.com for English Qur’an translations, one of the many options offered to you is a prime example of both phenomena: The Holy Quran: Correct Translation of the Holy Quran in English, published in 2021 as a print-on-demand book and allegedly written by one Dr. Muhammad Ali (https://www.amazon.com/-/de/dp/B09HVGB8DH).
The somewhat garbled Arabic that adorns the cover page is apparently an inexpert attempt at reproducing Arabic calligraphy of the title al-Qurʾān al-karīm (‘The Noble Qur’an’) by using a computer font. The result reads as al-karīm al-Qurʾān, which is nonsensical, akin to calling it ‘The Qur’an the Noble’. The title and author’s name on the spine are just as inexpertly typeset.
Furthermore, ‘Dr. Muhammad Ali’ seems to be a made-up name. While there was a famous Qur’an translator called Muhammad Ali (1874–1951), the head of the Lahore Ahmadiyya Movement (https://gloqur.de/quran-translation-of-the-week-106-the-first-influential-muslim-authored-translation-of-the-quran-into-english/), he bore the title ‘Maulana’, not ‘Doctor’, and his translation has nothing to do with the text that is between the covers of this print-on-demand book. The book itself offers no information on the author but its page on Goodreads (https://www.goodreads.com/book/show/59510923-the-holy-quran), which has seven positive ratings but no actual reviews, somewhat bizarrely attributes it to Muhammad Ali, the boxer.
The back cover blurb, written in fluent English, seems quite expressive. This is because it has been plagiarised: it is completely identical to the first paragraph of an article entitled ‘The Qur’an in Muslim Life and Practice’ by Mahmoud Ayoub (1935–2021), a text which seeks to introduce the Qur’an to non-Muslim readers.
The text inside the covers is exceedingly strange. It starts with the header ‘Sura First’, followed by a footnote sign that is not connected to an actual footnote, because the translation does not have any. Of course, many digital editions of existing Qur’an translations have done away with footnotes that were present in the printed version, but they usually went to the trouble of removing the footnote signs as well. This book clearly does not use an existing translation, however; that is quite obvious when we come across renditions such as Q 2:16:
‘Resemble which kindled fire when the fire has spread its clarity about the objects around you and how do God takes away suddenly, leaving them in darkness, no PUE den see.’
As we can see here, not only is the English questionable to the point of being incomprehensible but there are also traces of Spanish in the text in the phrase ‘no PUE den see’ (‘cannot see’). This happens again at the beginning of the second sura:
‘ALM Here is the book that offers no doubt; He is the di Recciôn of those who fear the Lord; those who believe in the unseen and those obser promptly van …’
It is obvious that what has happened here is that OCR was performed on a printed Spanish translation without correcting for misrecognised or hyphenated words, and the result was fed into a machine translation software that only translated those parts it recognised, and sometimes did so quite badly.
The source used to produce this English translation can be deduced from the Spanish fragments contained in it. It is El Corán by Joaquín García Bravo (1872–1922), published in 1907, which was a translation into Spanish of the successful French Qur’an translation by the Polish-French Orientalist Albin de Biberstein Kazimirski (1808–1887), first published in 1840. Kazimirski’s work was hugely influential, was admired for its philological skill and precision as well as its comparatively respectful, non-polemical approach to the Qur’an, and was translated into other Romance languages multiple times. García Bravo’s translation was the second version of Kazimirski to appear in Spanish, following on from that by Vicente Ortiz de la Puebla (Barcelona 1872). García Bravo’s translation is widely available in digital format, often without any mention of the translator, and it was therefore presumably easy to use as a basis for machine translation. Moreover, copyright protections expired long ago, meaning there would be no legal liability involved.
Let us consider the steps the translation went through by looking at Q 1:5:
إِيَّاكَ نَعْبُدُ وَإِيَّاكَ نَسْتَعِينُ
Kazimirski: ‘C’est toi que nous adorons, c’est toi dont nous implorons le secours.’
García Bravo: ‘A ti es a quien adoramos, de ti es de quien imploramos socorro.’
‘Dr. Muhammad Ali’: ‘To whom we worship you is, you is who implore so I run [presumably translated from a hyphenated or separated ‘so corro’].’
We thus have a book that has gone from Arabic to French to Spanish to English; that originated as an ambitious work by a non-Muslim Orientalist and ended as a cheap rip-off, attributed to a fake Muslim translator to give some impression of authenticity, while at the same time marketing the text to non-Muslims by using a plagiarised quotation from a Muslim author (whose copyright protection has not expired by a long shot, incidentally). The publisher, who cannot be identified, made sure to generate many generic favourable ratings and reviews on Amazon and Goodreads, although they were not quite able to prevent a few buyers from calling them out: ‘It reads like a google translation machine.’ ‘This translation wasn’t done correctly. Tons of random words in different languages (Spanish, Latin, French) that make no sense. Looks like it was put into google translate, copy and pasted into word, and published.’ ‘I don’t know why they would say CORRECT translation and it is so poorly translated! ABSOLUTELY horrible. I’m kicking myself I don’t look at the reviews beforehand.’ ‘Listen to what the other reviews say, it genuinely seems to have been run through Google translate.’ ‘There are broken sentences mixed in English and Spanish. Not a great introduction to the Lord’s last book.’
The book was obviously conceived and advertised just professionally enough for people to buy it, but not for readers to appreciate it. Given the recent progress in AI-based translation, this is bound to change in the near future. Already with this book, we can see that it is very hard to say what the ‘original version’ is and who is responsible for the choices made in it: Kazimirski? García Bravo? The software that translated the latter’s work? The problem will be exacerbated once we have translations that are produced by large language models, using the works of countless human translators, whether they are still under copyright protection or not, without crediting anyone. A skilled publisher could easily use AI to come up with a decent translation, a plausible translator’s persona, title page, and blurbs, resulting in a product that will fool anyone. Maybe such translations are already on the market and we just do not know it. The mere existence of ‘Dr. Muhammad Ali’s’ translation shows that there is sufficient commercial incentive.
Johanna Pink