How to loop through millions of column data and do a task without it being slow?

Question

   token_pos = []
   df_tags = []
   content_1 = rel_data["content"]
   df_tags = content_1.values
   counter=0

   for values in df_tags:
     doc = nlp(df_tags[counter])
     for token in doc:
         token_pos.append(token.pos_)
     d = {x:token_pos.count(x) for x in token_pos}
     df = pd.DataFrame([d])
     print(df)
     counter+=1 

This is my code, for context:

Content_1 has data that consists of paragraphs as data, sort of like long text and I want to loop through millions of these whilst also counting how many part of speech tags they have and put it in a DF. The code above works but it puts each one into a seperate DF and its also very slow. I need one that is fast but also includes them all into one massive DF

I am stuck :(