diff --git a/csvwlib/converter/ModelConverter.py b/csvwlib/converter/ModelConverter.py
index ab18c8a..10e44bc 100644
--- a/csvwlib/converter/ModelConverter.py
+++ b/csvwlib/converter/ModelConverter.py
@@ -201,7 +201,9 @@ def _normalize_column_names(self):
             for i, column in enumerate(table['tableSchema']['columns'], start=1):
                 if 'name' not in column:
                     language = JSONLDUtils.language(self.metadata['@context'], table)
-                    titles = column['titles'] if type(column['titles']) is list else [column['titles']]
+                    titles = column.get('titles',[]) 
+                    if type(titles) is not list:
+                        titles = [titles]
                     if language is None:
                         column['name'] = DOPUtils.natural_language_first_value(titles)
                     else:
diff --git a/csvwlib/utils/DOPUtils.py b/csvwlib/utils/DOPUtils.py
index 628179a..da932fc 100644
--- a/csvwlib/utils/DOPUtils.py
+++ b/csvwlib/utils/DOPUtils.py
@@ -7,4 +7,4 @@ def natural_language_first_value(property_value):
         if type(property_value) is str:
             return property_value
         elif type(property_value) is list:
-            return property_value[0]
+            return next(iter(property_value or []), None)