Why there is memory overflow?

I wrote the below python code. I haven't got any array or list that needs memory and I can't understand why there is memory overflow.

con = db.connect('SentiWords.db')
cur = con.cursor()
cur.execute("DROP TABLE IF EXISTS Words")
cur.execute('''CREATE TABLE Words(word Text, type Text) ''')
infile = open("train_reviews.txt")
lines = infile.readlines()
stopwords = nltk.corpus.stopwords.words('english')
sentiment_words = dict()
counter =0
for line in lines:
  words = nltk.word_tokenize(line.decode("UTF-8"))
  words = [ w for w in words if w.lower() not in stopwords]
  for word in words:
    counter_sent = 0
    counter_obj = 0
    check = swn.senti_synsets(word)
    for i in range(0,len(check)):
        if check[i].pos_score() < check[i].obj_score() and    check[i].neg_score() < check[i].obj_score():
          counter_obj +=1
        else:
          counter_sent +=1
    if counter_obj < counter_sent:
        cur.execute('''SELECT type FROM Words WHERE word=? ''', [word])
        data=cur.fetchall() #extract frequency
        if len(data) == 0: #if there isn't any frequency
           cur.execute("INSERT INTO Words VALUES(?,?)", (word,"no-obj")) 
    if counter % 1000 == 0:
        con.commit()
        con.close()
        con = db.connect('SentiWords.db')
        cur = con.cursor()
    print counter
    counter +=1
con.commit()
con.close()

I want to find from reviews all words that have got a sentiment meaning. So I decide to use sentiwornet to compare all words from the reviews and keep the words that have got sentiment meaning in a database. Can you exlain me what goes wrong ?