This is a very interesting question and somehow the question itself answers a doubt I had seeing some sites with right aligned footnotes (even if in Western languages).
So, before anything, I don't know if there's an accepted convention for this, but will try to answer based on UX principles and standards.
As you probably know, there are different layout patterns, including the Z-pattern, F-Pattern and Guttenberg Diagram. There are thousands of studies about this, so I won't extend on the subject. But... what happens with RTL layouts? As you may expect, the patterns are flipped horizontally. If you read this article, it also concludes that you should "Stick to Conventional Layouts" (meaning LTR orientation).
Now, let's see another study: Visual Reading Patterns on Arabic Interfaces:Insights from Eye Tracking by Remya P. George, Rasiya Anwar and Sunitha Jeyasekhar. As you may notice, the conclusion is radically different to the first article. And instead of flipped "Western" patterns, other patterns emerge, I will shamelessly copy the content just in case it goes down:
- Intermittent examination of multimedia content
In a right-to-left layout, it was interesting to find z-shaped patterns in which viewers start with text then examine the multimedia content at points in time in which the audio accompanying the video or animation or video emphasizes apoint with tone.
- Triangular Reading Pattern
The two consecutive cycles of the triangular pattern are shown in the above figure. In the first cycle, users exhibited behavior of visual examination of navigation, followed by content then the multimedia area of the eLearning interface. The second cycle was similarly exhibit by several users in the way they distributed their attention across the interface during the module’s multimedia presentation.
- Right to Left Sweep Pattern
The figure shows visual reading patterns across the first column followed by a shift towards the second column in which the reader continued to exhibit similar patterns across lines
The Proposed solution
As you may see, there's no much logic (in Western sense) to Asian and Arab layouts. And yet, there is a logic, which helps to support this answer:
- You can easily note that Arab language eye-tracking positioned a
huge weight on the right (as expected).
- Western users have common patterns where we can find coincidences in Eastern user's patterns
- Bullets are a visual aid to scan your content
With these elements, and assuming your target has a reasonable ratio of Eastern users, AND that you're mixing languages, AND this is for footnotes, I'd suggest to align the list to the right: you preserve RTL language conventions yet Western users will naturally read by using a Z-pattern. See a quick mock I did to illustrate this behavior