diff --git a/lib/elixir/lib/string.ex b/lib/elixir/lib/string.ex
index 0fce4b87ed2..a2f92756538 100644
--- a/lib/elixir/lib/string.ex
+++ b/lib/elixir/lib/string.ex
@@ -505,13 +505,13 @@ defmodule String do
   ## Examples
 
       iex> String.normalize("yêṩ", :nfd)
-      "yêṩ"
+      "yêṩ"
 
       iex> String.normalize("leña", :nfc)
       "leña"
 
   """
-  @spec normalize(t, atom) :: boolean
+  @spec normalize(t, atom) :: t
   defdelegate normalize(string, form), to: String.Normalizer
 
   @doc """
diff --git a/lib/elixir/test/elixir/string_test.exs b/lib/elixir/test/elixir/string_test.exs
index f5c82a215ab..07033799be5 100644
--- a/lib/elixir/test/elixir/string_test.exs
+++ b/lib/elixir/test/elixir/string_test.exs
@@ -378,7 +378,7 @@ defmodule StringTest do
 
   test "normalize" do
     assert String.normalize("ŝ", :nfd) == "ŝ"
-    assert String.normalize("ḇravô", :nfd) == "ḇravô"
+    assert String.normalize("ḇravô", :nfd) == "ḇravô"
     assert String.normalize("ṩierra", :nfd) == "ṩierra"
     assert String.normalize("뢴", :nfd) == "뢴"
     assert String.normalize("êchǭ", :nfc) == "êchǭ"
diff --git a/lib/elixir/unicode/unicode.ex b/lib/elixir/unicode/unicode.ex
index f8e566a7e8a..360388f91b3 100644
--- a/lib/elixir/unicode/unicode.ex
+++ b/lib/elixir/unicode/unicode.ex
@@ -505,7 +505,7 @@ defmodule String.Normalizer do
     {n, rest} = String.Unicode.next_grapheme_size(binary)
     part = :binary.part(binary, 0, n)
     case n do
-      1 -> normalize_nfc(rest, acc <> part)
+      1 -> normalize_nfd(rest, acc <> part)
       _ -> normalize_nfd(rest, acc <> canonical_order(part, []))
     end
   end