We recommend separating these two processes out. It's really a good practice since it'll make it clear to your translators, and the Translation Memory can be separated out, too, increasing its quality.
So, let's assume that your source text is in English, and you want to adapt it to Chinese and Arabic and then translate it into other target languages.
You'll need to have the following projects:
One project will be dedicated to the adaptation (e.g., named "Adaptation").
A set of projects will be dedicated to the translations of the adapted text.
You have two choices on how to handle the work regarding languages:
Adapt and Translate in one step.
Create the "Adaptation" project defining English as the source language.
Upload your English source file.
Add Chinese & Arabic as the project's target languages.
Translate the source text into Chinese and Arabic while adapting it so as to accommodate regional differences too.
Export the Chinese & Arabic translation files from Transifex and upload them as source files to different projects.
Note: The Chinese file will be uploaded to a project with the Chinese source language, and the Arabic file will be uploaded to a different project with the Arabic source language.
Both projects will be translated from their own source languages to the desired target languages.
Use the same source language everywhere.
Create the "Adaptation" project defining English as the source language.
Upload your English source file.
Add Chinese & Arabic as the project's "target" languages.
Adapt your source text depending on the region without translating it.
Export the Chinese & Arabic "translation" files from Transifex and upload them as source files to a new project.
Note: Both files will be adapted, but their content will still be in English. They will be both hosted under the same project where the English source language will have been defined.
Then you can translate the content of each source file into the desired target languages.