What are all possible POS tags of NLTK?

python

nltk

pos-tagging

natural-language-processing

byAnton Shumikhin·Mar 10, 2025

Straight to the point, we can instantly retrieve NLTK's POS tags by utilizing the Penn Treebank tagset as follows:

import nltk 
  
# Download if you don't have it yet, just like getting a new game
nltk.download('tagsets')

tagdict = nltk.data.load('help/tagsets/upenn_tagset.pickle')
# Let's print 'em all out, like showing off your game achievement
print(sorted(tagdict.keys()))

This short but powerful piece of code prints out a neatly sorted list of POS tags, which are essential assets when you're dealing with POS tagging tasks in NLTK.

The nitty-gritty of NLTK POS tags

Now that you have the complete kit of POS tags that NLTK offers, let's dive into the details. These tags originate from the Penn Treebank Project and cover pretty much every linguistic element you can think of: from the basic nouns, verbs, adjectives, and adverbs, to more complex tags differentiating between comparative and superlative forms of adjectives.

Anatomizing the tags

Each POS tag is like a detailed role description for a word. For instance, JJ, JJR, and JJS categorize adjectives and also indicate their normal, comparative, and superlative forms. These specific roles prove to be pivotal in a variety of NLP tasks.

Dealing with NLTK magic

When you're using nltk.pos_tag(), it employs the Penn Treebank Tag Set and assigns tags based on its in-built training corpus. However, the accuracy and choice of tags can be influenced by the tagger and the corpus you opt for.

Learning resources

For learning more about these tags and their usage, head to nltk.org. They showcase an extensive collection of tagging examples. You can also use NLTK's tag dictionary (tagdict) for quick referencing of the POS tags.

Tackling obstacles: Common POS tagging challenges

One challenge NLP enthusiasts often face while tagging is ambiguities arising from contextual meanings. In NLTK, this is handled with contextual taggers but remember, they're not perfect and it's always good to double-check.

What to do when tags act weird

Sometimes, the output from nltk.pos_tag() might catch you by surprise. If this happens, consider revisiting the training corpus and the context of the text. Because context matters, just like that one text from your ex!

More than just Penn Treebank

While Penn Treebank tagset is widely used, NLTK supports more tagsets. If you're working on a project with universal language application, consider using the Universal POS tagset.

Customizing your tags

Last but not least, if the available tagsets aren't gratifying your specific needs, you can train your own tagger using your dataset. It's like creating a character in a role-playing game to fit your style!

Advanced usage of POS tags

NLTK's POS tags come with nuanced details, such as NNP for proper nouns and NNPS for their plural form. These tags become highly important for tasks like Named Entity Recognition (NER).

POS tagging in Machine Learning

In Machine Learning, the POS tags serve as crucial features for tasks like text classification and sentiment analysis. They add more clarity to the word meanings and hence increase the prediction accuracy.