I recently started using LingQ for learning French and it’s going quite well. What attracted me to the site was its ability to import videos form YouTube and then transcribing them. For French and Persian the feature seems to work very well.
I then thought it was a good idea to use the website to maintain my fluency in the various Arabic dialects I know and tried importing a few simple videos in Syrian and Egyptian dialect. To my dismay the videos in Syrian dialect are not transcribed properly at all. It seems LingQ desperately wants to transcribe it to modern standard Arabic for whatever reason.
Is there any way to fix this? To my knowledge the Whisper API can handle the more common spoken varieties of Arabic quite well.
Thank you for looking into this. I must report three further bugs of perhaps more serious character regarding LingQ for Arabic. I imported an E-Book in a perfectly fine PDF-format. It’s written in modern standard Arabic. Three obvious issues appeared:
Letters are put in the wrong order. This seems to happen frequently with words that have the Arabic definite article “ال” before it. So LingQ will import a word, for example الأمثال (al-Amthal), and make it املثال (amalthal). Furthermore, many words are sometimes cut in half.
When you make LingQs of Arabic words, it will almost always give you a heavily context dependent translation. Often it seems the meaning given in the translation is based on the word coming after the word you clicked. So for example I’ll click on the word دار (house) but the translation wants it to mean “publishing house” (دار نشر) because of the context. This is feature of Arabic called idafa where you can create compound words. However, this will often lead to a lot of mistranslations when adding LingQs.
The “Show translation” on the page-viewer is almost entirely incorrect. Often times it seems that it will show a correct translation but for a completely different page.