diff --git a/wani.py b/wani.py
index f4dc1e7..a60b7bf 100644
--- a/wani.py
+++ b/wani.py
@@ -219,6 +219,12 @@ class ComponentRendition:
         else:
             raise RuntimeError("Unknown rendition: {}".format(self.rendition))
     
+    @staticmethod
+    def set_representations(matches, components):
+        for words, agreement in matches:
+            for _, w in words.items():
+                w.representation = ":("
+    
     def __str__(self):
         return str(self.rendition)
 
@@ -391,10 +397,6 @@ class Component:
             raise RuntimeError("Unreachable")
 
     def set_representation(self, representation):
-        # for r in representation:
-        #     print(ElementTree.tostring(r).decode('ascii').replace('\n', ''))
-        # print("--")
-
         if len(representation) > 0:
             for feature in representation:
                 self.representation.add_feature(feature)
@@ -676,14 +678,14 @@ class SyntacticStructure:
 
         to_ret = []
         for m in matches:
-            if not self.check_agreements(m):
-                bad = "Agreement"
-            elif not self.check_form(m):
-                bad = "Form"
-            else:
-                bad = "OK"
+            # if not self.check_agreements(m):
+            #     bad = "Agreement"
+            # elif not self.check_form(m):
+            #     bad = "Form"
+            # else:
+            #     bad = "OK"
 
-            to_ret.append((m, bad))
+            to_ret.append((m, self.check_agreements(m)))
 
         return to_ret
 
@@ -845,7 +847,8 @@ class Writer:
         elif self.all:
             return [word.id, word.text, word.lemma, word.msd]
         else:
-            return [word.lemma, "REP?"]
+            assert(word.representation is not None)
+            return [word.lemma, word.representation]
     
     def sorted_rows(self, rows):
         if self.sort_by < 0 or len(rows) < 2:
@@ -961,6 +964,12 @@ class ColocationIds:
                 if group:
                     break
 
+    def set_representations(self, structures):
+        components_dict = {structure.id: structure for structure in structures}
+        for _1, (_2, cid_matches, sid) in self.data.items():
+            ComponentRendition.set_representations(cid_matches, components_dict[sid])
+
+
 def match_file(words, structures):
     matches = {s.id: [] for s in structures}
 
@@ -1025,6 +1034,9 @@ def main(input_file, structures_file, args):
             else:
                 colocation_ids.add_matches(matches)
 
+    # figure out representations!
+    colocation_ids.set_representations(structures)
+
     if args.all:
         Writer.make_all_writer(args).write_out(structures, colocation_ids)
     Writer.make_output_writer(args).write_out(structures, colocation_ids)