diff --git a/download_corpora.py b/download_corpora.py
index 94abfc70..733b1d19 100644
--- a/download_corpora.py
+++ b/download_corpora.py
@@ -8,6 +8,7 @@
 REQUIRED_CORPORA = [
     'brown',  # Required for FastNPExtractor
     'punkt',  # Required for WordTokenizer
+    'punkt_tab',  # Required for WordTokenizer (NLTK >= 3.9)
     'maxent_treebank_pos_tagger',  # Required for NLTKTagger
     'movie_reviews',  # Required for NaiveBayesAnalyzer
     'wordnet',  # Required for lemmatization and Wordnet
diff --git a/newspaper/nlp.py b/newspaper/nlp.py
index 23505255..782afb72 100644
--- a/newspaper/nlp.py
+++ b/newspaper/nlp.py
@@ -154,7 +154,10 @@ def split_sentences(text):
     """Split a large string into sentences
     """
     import nltk.data
-    tokenizer = nltk.data.load('tokenizers/punkt/english.pickle')
+    try:
+        tokenizer = nltk.data.load('tokenizers/punkt/english.pickle')
+    except LookupError:
+        tokenizer = nltk.data.load('tokenizers/punkt_tab/english')
 
     sentences = tokenizer.tokenize(text)
     sentences = [x.replace('\n', '') for x in sentences if len(x) > 10]