diff --git a/scripts/conll2text.py b/tools/conll2text.py
similarity index 100%
rename from scripts/conll2text.py
rename to tools/conll2text.py
diff --git a/tools/conllu2splits.py b/tools/conllu2splits.py
new file mode 100755
index 0000000000000000000000000000000000000000..ce958cff38d664c50a71365911dd17c5373bcb9a
--- /dev/null
+++ b/tools/conllu2splits.py
@@ -0,0 +1,80 @@
+#! /usr/bin/python3
+
+import sys
+
+rules = {}
+prefix = "SPLITWORD "
+
+def printUsageAndExit() :
+  print("USAGE : %s file.conllu conllu.mcd"%sys.argv[0], file=sys.stderr)
+  exit(1)
+
+def readMCD(mcdFilename) :
+  mcd = {}
+  for line in open(mcdFilename, "r", encoding="utf8") :
+    clean = line.strip()
+    if len(line) < 2 or line[0] == '#' :
+      continue
+    splited = line.split(' ')
+    if len(splited) != 2 :
+      print("ERROR : invalid mcd line \'%s\'. Aborting"%line, file=sys.stderr)
+      exit(1)
+    mcd[splited[0].strip()] = splited[1].strip()
+
+  return mcd
+
+def computeRules(sentence) :
+  wordById = {}
+  for word in sentence :
+    splited = word[0].split("-")
+    if len(splited) > 1 :
+      continue
+    wordById[word[0]] = word[1]
+
+  for word in sentence :
+    splited = word[0].split("-")
+    if len(splited) > 1 :
+      rule = ""
+      for id in splited :
+        rule += "@" + wordById[id]
+      if word[1] in rules :
+        if rule in rules[word[1]] :
+          rules[word[1]][rule] += 1
+        else :
+          rules[word[1]][rule] = 1
+      else :
+        rules[word[1]] = {}
+        rules[word[1]][rule] = 1
+
+def main() :
+  if len(sys.argv) != 3 :
+    printUsageAndExit()
+
+  conllMCD = readMCD(sys.argv[2])
+  conllMCDr = {v: k for k, v in conllMCD.items()}
+
+  idId = int(conllMCDr["ID"])
+  idForm = int(conllMCDr["FORM"])
+
+  sentence = []
+
+  for line in open(sys.argv[1], "r") :
+    if len(line.strip()) < 2 or line[0] == '#' :
+      if len(sentence) > 0 :
+        computeRules(sentence)
+      sentence = []
+      continue
+
+    splited = line.strip().split('\t')
+    sentence += [[splited[idId], splited[idForm]]]
+
+  for word in rules :
+    if len(rules[word]) > 1 :
+      print("WARNING : Abiguity detected in \'%s\'"%(word+" "+str(rules[word])), file=sys.stderr)
+    for rule in rules[word] :
+      print(prefix+word+rule)
+      break
+
+if __name__ == "__main__" :
+  main()
+