Syntactic Processing Part 1 - POS Tagging and Shallow Parsing

Why This Guide Exists

You now know how to break text into words. But words alone do not tell the whole story.

Consider: “The dog bit the man” vs “The man bit the dog.” Same words, different meanings. The difference is structure—the grammatical relationships between words.

This guide covers the first layer of syntactic processing: identifying what type each word is, and grouping them into meaningful chunks.

Segment 1: Session Overview

What You Will Learn

1. Part-of-Speech Tagging – Labeling words as nouns, verbs, adjectives, etc.
2. Shallow Parsing (Chunking) – Grouping words into phrases like “noun phrases” and “verb phrases”

The Analogy: Sentence Architecture

Think of a sentence like a building:

• Foundation: Main verb (holds everything together)
• Load-bearing walls: Subject, object (core participants)
• Support beams: Prepositions, conjunctions (connect parts)
• Decorations: Adjectives, adverbs (add detail)

Lexical processing gave you the building materials (words). Syntactic processing creates the structure.

Segment 2: Part-of-Speech Tagging

What is POS Tagging?

Part-of-Speech (POS) tagging is labeling each word with its grammatical role.

The 8 Primary Parts of Speech

Why POS Tagging is Hard

Many words can be multiple parts of speech:

Context determines the correct tag.

Penn Treebank Tags (NLTK)

Python Example

import nltk
from nltk.tokenize import word_tokenize

nltk.download('punkt')
nltk.download('averaged_perceptron_tagger')

text = "The quick brown fox jumps over the lazy dog"
tokens = word_tokenize(text)
tags = nltk.pos_tag(tokens)
print(tags)
# [('The', 'DT'), ('quick', 'JJ'), ('brown', 'JJ'), ('fox', 'NN'), 
#  ('jumps', 'VBZ'), ('over', 'IN'), ('the', 'DT'), ('lazy', 'JJ'), ('dog', 'NN')]

Segment 3: Shallow Parsing (Chunking)

What is Shallow Parsing?

Shallow parsing (chunking) groups words into phrases without analyzing deep grammatical relationships. Think of it as finding “chunks” of meaning.

Types of Phrases

Example

Sentence: “The quick brown fox jumps over the lazy dog”

Noun phrases identified:

• “The quick brown fox” (NP)
• “the lazy dog” (NP)

Python Chunking Example

from nltk.chunk import RegexpParser

# Define chunking grammar
grammar = r"""
    NP: {<DT>?<JJ>*<NN>}    # Determiner + adjectives + noun
        {<NNP>+}            # Proper nouns
    VP: {<VB.*><NP|PP>*}    # Verb + noun phrase or prepositional phrase
    PP: {<IN><NP>}          # Preposition + noun phrase
"""

chunk_parser = RegexpParser(grammar)
sentence = [("The", "DT"), ("quick", "JJ"), ("brown", "JJ"), 
            ("fox", "NN"), ("jumps", "VBZ"), ("over", "IN"),
            ("the", "DT"), ("lazy", "JJ"), ("dog", "NN")]

tree = chunk_parser.parse(sentence)
# Visualizes as: (S (NP The/DT quick/JJ brown/JJ fox/NN) 
#                 (VP jumps/VBZ (PP over/IN (NP the/DT lazy/JJ dog/NN))))

Shallow vs Deep Parsing

Segment 4: Session Summary

Key Takeaways

1. POS Tagging labels each word with its grammatical type
   • Context matters: “book” can be noun or verb
   • Penn Treebank tags are standard in NLP
2. Shallow Parsing groups words into phrases
   • Noun phrases, verb phrases, prepositional phrases
   • Fast and useful for information extraction
3. These techniques work together:
   • First tokenise, then tag, then chunk

Practice Questions

1. What is the POS tag for “quickly”?

2. What phrase type is “the lazy dog”?

3. Why is POS tagging challenging?

What Comes Next?

Part 2 covers Context-Free Grammar and Parsing—understanding complete sentence structures and building parse trees.