Natural language processing (NLP) of languages that use Arabic alphabet

In general and globally, about 7,000 natural spoken languages are available. Technically observed, about 20 languages of these resourced well in NLP term. Almost all the Arabic scripted languages are considered a low-resourced language, only the Arabic language itself in some aspects of NLP are well resourced. In addition, some Arabic-scripted languages are very low-resourced in text or abounded Arabic scripts such as Old Turkish language (Ottoman script). Widely speaking, all Arabic-scripted languages require NLP resources to bridge the gap with advanced tools and use available materials. Our study's main aim is to transfer learning from well-resourced languages, for instance, Arabic language to other Arabic-scripted languages. We successfully implemented a sample of transfer learning and applied them to a very low-resourced language, such as Ottoman Turkish and Kurdish language, Arabic and Parso-Arabic scripted languages. This transfer learning is a machine-based transliterate, the automatic transliteration from Ottoman to the modern Latin Turkish script. The system achieved 73.3% accuracy in general. Related to the Kurdish NLP, the transliteration system was implemented too. Moreover, we focused on the hurdles and barriers that researchers experience in this field. The conclusions related to Arabic-scripted languages NLP are: first, it is hard to process Arabic scripted low-resourced languages, but using transfer learning of the similar languages experience may make it possible. Second: there should be minimum resources and knowledge about the language to start with its NLP.