diff --git a/README.md b/README.md
index 70d8865..9a7ae0f 100644
--- a/README.md
+++ b/README.md
@@ -50,7 +50,7 @@ or this [poster](poster.pdf) and [presentation](presentation.pdf).
 The lemmatizer is implemented as a look-up table, using a lemma dictionary imported from the [Morfeusz morphological analyzer](http://morfeusz.sgjp.pl/).
 
 ### Tagger
-The tagger has been trained on a corpus consisting of the 1 million word subcurpous of the [National Corpus of Polish](http://clip.ipipan.waw.pl/NationalCorpusOfPolish} and the 500k [Frequency Corpus of the 1960s Polish language](http://clip.ipipan.waw.pl/PL196x). For tasks involving Polish language only, we reccomend using the internal tagset (`token.tag_` as opposed to `token.pos_`), because the latter is a lossy mapping of the former.
+The tagger has been trained on a corpus consisting of the 1 million word subcurpous of the [National Corpus of Polish](http://clip.ipipan.waw.pl/NationalCorpusOfPolish) and the 500k [Frequency Corpus of the 1960s Polish language](http://clip.ipipan.waw.pl/PL196x). For tasks involving Polish language only, we reccomend using the internal tagset (`token.tag_` as opposed to `token.pos_`), because the latter is a lossy mapping of the former.
 
 ### Depenendency Parser
 For training a dependency parser, we've used the [PDB UD treebank](https://universaldependencies.org/treebanks/pl_pdb/index.html)