diff --git a/spacy/structs.pxd b/spacy/structs.pxd
index cfcadc3d0..fa282cae7 100644
--- a/spacy/structs.pxd
+++ b/spacy/structs.pxd
@@ -58,7 +58,7 @@ cdef struct TokenC:
     attr_t tag
     int idx
     attr_t lemma
-    attr_t sense
+    attr_t norm
     int head
     attr_t dep
 
diff --git a/spacy/tests/regression/test_issue2754.py b/spacy/tests/regression/test_issue2754.py
new file mode 100644
index 000000000..5f76727f8
--- /dev/null
+++ b/spacy/tests/regression/test_issue2754.py
@@ -0,0 +1,14 @@
+# coding: utf8
+from __future__ import unicode_literals
+
+import pytest
+from spacy.lang.en import English
+
+def test_issue2754():
+    """Test that words like 'a' and 'a.m.' don't get exceptional norm values."""
+    nlp = English()
+    a = nlp('a')
+    assert a[0].norm_ == 'a'
+    am = nlp('am')
+    assert am[0].norm_ == 'am'
+
diff --git a/spacy/tokens/token.pxd b/spacy/tokens/token.pxd
index 9b02d07fb..bb9f7d070 100644
--- a/spacy/tokens/token.pxd
+++ b/spacy/tokens/token.pxd
@@ -34,6 +34,11 @@ cdef class Token:
             return Lexeme.c_check_flag(token.lex, feat_name)
         elif feat_name == LEMMA:
             return token.lemma
+        elif feat_name == NORM:
+            if token.norm == 0:
+                return token.lex.norm
+            else:
+                return token.norm
         elif feat_name == POS:
             return token.pos
         elif feat_name == TAG:
@@ -58,6 +63,8 @@ cdef class Token:
                                        attr_t value) nogil:
         if feat_name == LEMMA:
             token.lemma = value
+        elif feat_name == NORM:
+            token.norm = value
         elif feat_name == POS:
             token.pos = <univ_pos_t>value
         elif feat_name == TAG:
diff --git a/spacy/tokens/token.pyx b/spacy/tokens/token.pyx
index 5c8af1333..0266004b5 100644
--- a/spacy/tokens/token.pyx
+++ b/spacy/tokens/token.pyx
@@ -249,7 +249,10 @@ cdef class Token:
             or norm exceptions.
         """
         def __get__(self):
-            return self.c.lex.norm
+            if self.c.norm == 0:
+                return self.c.lex.norm
+            else:
+                return self.c.norm
 
     property shape:
         """RETURNS (uint64): ID of the token's shape, a transform of the
@@ -711,7 +714,10 @@ cdef class Token:
             norm exceptions.
         """
         def __get__(self):
-            return self.vocab.strings[self.c.lex.norm]
+            return self.vocab.strings[self.norm]
+
+        def __set__(self, unicode norm_):
+            self.c.norm = self.vocab.strings.add(norm_)
 
     property shape_:
         """RETURNS (unicode): Transform of the tokens's string, to show
diff --git a/spacy/vocab.pyx b/spacy/vocab.pyx
index 42fd2f46e..e28aa0b86 100644
--- a/spacy/vocab.pyx
+++ b/spacy/vocab.pyx
@@ -17,7 +17,7 @@ from .structs cimport SerializedLexemeC
 from .compat import copy_reg, basestring_
 from .errors import Errors
 from .lemmatizer import Lemmatizer
-from .attrs import intify_attrs
+from .attrs import intify_attrs, NORM
 from .vectors import Vectors
 from ._ml import link_vectors_to_models
 from . import util
@@ -234,7 +234,10 @@ cdef class Vocab:
                 self.morphology.assign_tag(token, props[TAG])
             for attr_id, value in props.items():
                 Token.set_struct_attr(token, attr_id, value)
-                Lexeme.set_struct_attr(lex, attr_id, value)
+                # NORM is the only one that overlaps between the two
+                # (which is maybe not great?)
+                if attr_id != NORM:
+                    Lexeme.set_struct_attr(lex, attr_id, value)
         return tokens
 
     @property