diff --git a/lang/semgrep-grammars/src/semgrep-elixir/grammar.js b/lang/semgrep-grammars/src/semgrep-elixir/grammar.js
index 5d9d27b6..3c1a9cc4 100644
--- a/lang/semgrep-grammars/src/semgrep-elixir/grammar.js
+++ b/lang/semgrep-grammars/src/semgrep-elixir/grammar.js
@@ -49,16 +49,22 @@ module.exports = grammar(base_grammar, {
     // keyword/pairs, such as
     //   foo(some_arg: 0, ...)
     //   %{some_item: 0, ...}
-    // Also note that now there is ambiguity whether foo(...) is an
-    // identity or pair, we set the pair rule to have a lower
-    // precedence
     pair: ($, previous) => {
-      return prec(-1, choice(
+      return choice(
         previous,
         $.semgrep_ellipsis,
-      ));
+      );
     },
 
+    // Note that because of the pair rule above, now there is
+    // ambiguity whether the ellipsis in foo(...) is an
+    // ellipsis for positional or keyword arguments. If ... matches
+    // with the identity rule, it will be considered a positional
+    // argument. If ... matches with semgrep_ellipsis, it will be
+    // part of parsing the pair rule, which means its a keyword
+    // argument.  In practice, this doesn't matter because either
+    // way, they will become ParamEllipsis in the Generic AST
+    // anyway.
     semgrep_ellipsis: $ => prec(-1, '...'),
 
     _expression: ($, previous) => choice(