diff --git a/jedi/evaluate/context/iterable.py b/jedi/evaluate/context/iterable.py
index 598a8301..6856981d 100644
--- a/jedi/evaluate/context/iterable.py
+++ b/jedi/evaluate/context/iterable.py
@@ -469,7 +469,7 @@ class FakeDict(_FakeArray):
     @register_builtin_method('values')
     def _values(self):
         return ContextSet(FakeSequence(
-            self.evaluator, 'tuple',
+            self.evaluator, u'tuple',
             [LazyKnownContexts(self.dict_values())]
         ))
 
diff --git a/jedi/evaluate/param.py b/jedi/evaluate/param.py
index c5c28353..1445ef0c 100644
--- a/jedi/evaluate/param.py
+++ b/jedi/evaluate/param.py
@@ -97,7 +97,7 @@ def get_params(execution_context, var_args):
                         var_arg_iterator.push_back((key, argument))
                         break
                     lazy_context_list.append(argument)
-            seq = iterable.FakeSequence(execution_context.evaluator, 'tuple', lazy_context_list)
+            seq = iterable.FakeSequence(execution_context.evaluator, u'tuple', lazy_context_list)
             result_arg = LazyKnownContext(seq)
         elif param.star_count == 2:
             # **kwargs param
@@ -176,7 +176,7 @@ def _error_argument_count(funcdef, actual_count):
 def _create_default_param(execution_context, param):
     if param.star_count == 1:
         result_arg = LazyKnownContext(
-            iterable.FakeSequence(execution_context.evaluator, 'tuple', [])
+            iterable.FakeSequence(execution_context.evaluator, u'tuple', [])
         )
     elif param.star_count == 2:
         result_arg = LazyKnownContext(
diff --git a/jedi/evaluate/pep0484.py b/jedi/evaluate/pep0484.py
index 7df7c611..80d4b3ca 100644
--- a/jedi/evaluate/pep0484.py
+++ b/jedi/evaluate/pep0484.py
@@ -46,7 +46,7 @@ def _evaluate_for_annotation(context, annotation, index=None):
         context_set = context.eval_node(_fix_forward_reference(context, annotation))
         if index is not None:
             context_set = context_set.filter(
-                lambda context: context.array_type == 'tuple' \
+                lambda context: context.array_type == u'tuple' \
                                 and len(list(context.py__iter__())) >= index
             ).py__getitem__(index)
         return context_set.execute_evaluated()
@@ -172,7 +172,7 @@ def py__getitem__(context, typ, node):
     from jedi.evaluate.context.iterable import FakeSequence
     args = FakeSequence(
         context.evaluator,
-        "tuple",
+        u'tuple',
         [LazyTreeContext(context, n) for n in nodes]
     )
 
diff --git a/jedi/evaluate/stdlib.py b/jedi/evaluate/stdlib.py
index 8ed60b10..aeb9b666 100644
--- a/jedi/evaluate/stdlib.py
+++ b/jedi/evaluate/stdlib.py
@@ -209,7 +209,7 @@ def builtins_reversed(evaluator, sequences, obj, arguments):
     # necessary, because `reversed` is a function and autocompletion
     # would fail in certain cases like `reversed(x).__iter__` if we
     # just returned the result directly.
-    seq = iterable.FakeSequence(evaluator, 'list', rev)
+    seq = iterable.FakeSequence(evaluator, u'list', rev)
     arguments = ValuesArguments([ContextSet(seq)])
     return ContextSet(CompiledInstance(evaluator, evaluator.builtins_module, obj, arguments))