When running s_doc on my laptop, I see this:

$ bash s_docs
sed: 1: "../src/docs/spell.ok": invalid command code .
sort: ../src/docs/spell.ok:2: disorder: ACM
Sorting ../src/docs/spell.ok ...
sed: 1: "../src/docs/spell.ok": invalid command code .
Error: The file "./spell.ok" is not in the proper format.

$ git diff
diff --git a/src/docs/spell.ok b/src/docs/spell.ok
index 48a3f7183..9cd623ef0 100644
--- a/src/docs/spell.ok
+++ b/src/docs/spell.ok
@@ -1,4 +1,3 @@
-personal_ws-1.1 en 200
 ACM
 ADDR
 APIs
@@ -553,6 +552,7 @@ pareto
 pcoll
 pdf
 perf
+personal_ws-1.1 en 200
 petabyte
 pget
 php
dda@M-LKFK6YX19K$ 
 

That is, after s_doc has run and failed, it has modified a file, and not for the better.

I think the complaint is from the sed command

  sed -i '/personal_ws-1.1 en 200/d' $custom_spellings

I think adding a -e after the -i could fix it.

But the fact that the script changes $custom_spellings (src/doc/spell.ok) in place in order to check that file seems incorrect.  Otherwise (even on non-OS/X), interrupting the shell script can leave that file in a "wrong" state.  And the modified file can easily be accidentally committed.  I think the scripts should live by the rule - never modify a file unless you intend for the modification to be the final version.  Usually that means that a temporary file is built, that can be compared to the original, and only if different, it is moved into place.