does Prodigy support Arabic language?
if we have Arabic dataset, and want to use classification and NER, is it possible?
Prodigy doesn't really care that much about what text you load in – so yes, you can totally use it to annotate Arabic text 
In fact, I know of quite a few users who are working with Arabic, and I remember implementing a few fixes to the UI to make it display RTL languages better. If you search for "Arabic" on the forum, you'll also find some related threads.
If you want to use spaCy, you can start of with a blank language class (since there's no pretrained model for Arabic):
import spacy
nlp = spacy.blank("ar")
nlp.to_disk("./blank-arabic-model")
Hello. Does Prodigy support annotating fully vocalized Arabic text (i.e. Arabic text with diacritical markings)? Below is an image from the web that compares vocalized and unvocalized Arabic text. Thank you.
Sure, these are all unicode characters, right? If you can render something in the browser, you can annotate it with Prodigy If you're using a recipe like ner.manual, you could even use blank:xx (blank multi-language model) as the tokenizer to segment the text into words so the selection can snap to token boundaries.
Sounds good! Thank you!