C3993F20C398C2A7C3993FC398C2AEC398C2B720
Yuck! It is worse than just double-encoding.
First, let's look at the first part of that, split based on utf8, then converting parts back:
C399 3F 20 C398 C2A7 C399 3F C398 C2AE C398 C2B7 20
D9 3F 20 D8 A7 D9 3F D8 AE D8 B7 20
Notes: 6....3. 4. 3. 5........ 4.
Notes:
1. C399/C398 -- came from D9/D8
2. Arabic characters, in utf8, begin with D9 or D8 (at least)
3. 3F is "?", which is often used when an illegal encoding is being converted.
4. 20 is a space -- No problem with this.
5. C398 C2B7 -- D8 B7 -- D8B7 is utf8 for "ARABIC LETTER TAH"; C398C2B7 is the 'double encoding' of that.
6. Some pair of bytes D9xx (I don't know what xx) failed in converting to C399yyyy. Instead of yyyy, you got 3F ('?'). Then coming back, the C399 (U-grave) went to D9, but '?' stayed '?'. That lead to an illegal utf8 code 'D93F', hence the diamond with the '?'.
I'm afraid that the data in your table is corrupted beyond recovery. Start over in inserting the data. And be sure to check the HEX as soon as you have some Arabic loaded. The cursory check of the hex: C398 and C399 are bad; D8 and D9 are good.