Index: gc_lang/fr/perf_memo.txt
==================================================================
--- gc_lang/fr/perf_memo.txt
+++ gc_lang/fr/perf_memo.txt
@@ -26,5 +26,7 @@
 0.6.2       2018.02.19 19:06    5.51302     1.29359     0.874157    0.260415    0.271596    0.290641    0.684754    0.376905    0.0815201   0.00919633  (spelling normalization)
 1.0         2018.11.23 10:59    2.88577     0.702486    0.485648    0.139897    0.14079     0.148125    0.348751    0.201061    0.0360297   0.0043535   (x2, with new GC engine)
 1.1         2019.05.16 09:42    1.50743     0.360923    0.261113    0.0749272   0.0763827   0.0771537   0.180504    0.102942    0.0182762   0.0021925   (×2, but new processor: AMD Ryzen 7 2700X)
 1.2.1       2019.08.06 20:57    1.42886     0.358425    0.247356    0.0704405   0.0754886   0.0765604   0.177197    0.0988517   0.0188103   0.0020243
 1.6.0       2020.01.03 20:22    1.38847     0.346214    0.240242    0.0709539   0.0737499   0.0748733   0.176477    0.0969171   0.0187857   0.0025143   (nouveau dictionnaire avec lemmes masculin)
+1.9.0       2020.04.20 19:57    1.51183     0.369546    0.25681     0.0734314   0.0764396   0.0785668   0.183922    0.103674    0.0185812   0.002099    (NFC normalization)
+

Index: graphspell-js/tokenizer.js
==================================================================
--- graphspell-js/tokenizer.js
+++ graphspell-js/tokenizer.js
@@ -42,11 +42,11 @@
             [/^(?:l|d|n|m|t|s|j|c|ç|lorsqu|puisqu|jusqu|quoiqu|qu|presqu|quelqu)['’´‘′`ʼ]/i, 'WORD_ELIDED'],
             [/^\d\d?[h:]\d\d(?:[m:]\d\ds?|)\b/, 'HOUR'],
             [/^\d+(?:ers?\b|res?\b|è[rm]es?\b|i[èe][mr]es?\b|de?s?\b|nde?s?\b|ès?\b|es?\b|ᵉʳˢ?|ʳᵉˢ?|ᵈᵉ?ˢ?|ⁿᵈᵉ?ˢ?|ᵉˢ?)/, 'WORD_ORDINAL'],
             [/^\d+(?:[.,]\d+|)/, 'NUM'],
             [/^[&%‰€$+±=*/<>⩾⩽#|×¥£§¢¬÷@-]/, 'SIGN'],
-            [/^[a-zA-Zà-öÀ-Ö0-9ø-ÿØ-ßĀ-ʯff-stᴀ-ᶿᵉʳˢⁿᵈ_]+(?:[’'`-][a-zA-Zà-öÀ-Ö0-9ø-ÿØ-ßĀ-ʯff-stᴀ-ᶿᵉʳˢⁿᵈ_]+)*/, 'WORD']
+            [/^[a-zA-Zà-öÀ-Ö0-9ø-ÿØ-ßĀ-ʯff-stᴀ-ᶿ\u0300-\u036fᵉʳˢⁿᵈ_]+(?:[’'`-][a-zA-Zà-öÀ-Ö0-9ø-ÿØ-ßĀ-ʯff-stᴀ-ᶿ\u0300-\u036fᵉʳˢⁿᵈ_]+)*/, 'WORD']
         ]
 };
 
 
 class Tokenizer {
@@ -72,11 +72,11 @@
             for (let [zRegex, sType] of this.aRules) {
                 if (sType !== "SPACE"  ||  bWithSpaces) {
                     try {
                         if ((m = zRegex.exec(sText)) !== null) {
                             iToken += 1;
-                            yield { "i": iToken, "sType": sType, "sValue": m[0], "nStart": iNext, "nEnd": iNext + m[0].length };
+                            yield { "i": iToken, "sType": sType, "sValue": m[0], "nStart": iNext, "nEnd": iNext + m[0].length };  // m[0].normalize("NFC") not usefull at the moment
                             iCut = m[0].length;
                             break;
                         }
                     }
                     catch (e) {

Index: graphspell/tokenizer.py
==================================================================
--- graphspell/tokenizer.py
+++ graphspell/tokenizer.py
@@ -3,10 +3,12 @@
 using regular expressions
 """
 
 import re
 
+from unicodedata import normalize
+
 _PATTERNS = {
     "default":
         (
             r'(?P<FOLDERUNIX>/(?:bin|boot|dev|etc|home|lib|mnt|opt|root|sbin|tmp|usr|var|Bureau|Documents|Images|Musique|Public|Téléchargements|Vidéos)(?:/[\w.()-]+)*)',
             r'(?P<FOLDERWIN>[a-zA-Z]:\\(?:Program Files(?: [(]x86[)]|)|[\w.()]+)(?:\\[\w.()-]+)*)',
@@ -34,11 +36,11 @@
             r"(?P<WORD_ELIDED>(?:l|d|n|m|t|s|j|c|ç|lorsqu|puisqu|jusqu|quoiqu|qu|presqu|quelqu)['’´‘′`ʼ])",
             r'(?P<WORD_ORDINAL>\d+(?:ers?|res?|è[rm]es?|i[èe][mr]es?|de?s?|nde?s?|ès?|es?|ᵉʳˢ?|ʳᵉˢ?|ᵈᵉ?ˢ?|ⁿᵈᵉ?ˢ?|ᵉˢ?)\b)',
             r'(?P<HOUR>\d\d?[h:]\d\d(?:[m:]\d\ds?|)\b)',
             r'(?P<NUM>\d+(?:[.,]\d+|))',
             r'(?P<SIGN>[&%‰€$+±=*/<>⩾⩽#|×¥£¢§¬÷@-])',
-            r"(?P<WORD>\w+(?:[’'`-]\w+)*)"
+            r"(?P<WORD>[\w\u0300-\u036f]+(?:[’'`-][\w\u0300-\u036f]+)*)"
         )
 }
 
 
 class Tokenizer:
@@ -54,13 +56,13 @@
         "generator: tokenize <sText>"
         i = 0
         if bStartEndToken:
             yield { "i": 0, "sType": "INFO", "sValue": "<start>", "nStart": 0, "nEnd": 0, "lMorph": ["<start>"] }
         for i, m in enumerate(self.zToken.finditer(sText), 1):
-            yield { "i": i, "sType": m.lastgroup, "sValue": m.group(), "nStart": m.start(), "nEnd": m.end() }
+            yield { "i": i, "sType": m.lastgroup, "sValue": normalize("NFC", m.group()), "nStart": m.start(), "nEnd": m.end() }
         if bStartEndToken:
             iEnd = len(sText)
             yield { "i": i+1, "sType": "INFO", "sValue": "<end>", "nStart": iEnd, "nEnd": iEnd, "lMorph": ["<end>"] }
 
     def getTokenTypes (self):
         "returns list of token types as tuple (token name, regex)"
         return [ sRegex[4:-1].split(">")  for sRegex in _PATTERNS[self.sLang] ]