1 - Morphology and Syntax

What is Morphology in linguistics?

The study of the internal structure of words.

What is a morpheme and what are the two primary types of morphemes?

Words are constituted by (meaningful) units called morphemes. Stems that carry the primary meaning, and Affixes that modify meaning or serve grammatical functions.

What is the main difference between Stemming and Lemmatization?

Stemming cuts words down to a root/stem (which may not be a valid word), while Lemmatization reduces words to their valid dictionary base form (lemma).

What is Part-of-Speech (POS) Tagging?

The NLP task of assigning a specific word class or category (e.g., noun, verb, adjective) to each word in a text

What are the two steps of Rule-Based POS Tagging?

1. Assign a list of potential tags to each word using a dictionary

2. The tag to be assigned is chosen based on sets of rules usually designed by humans

What is Stochastic POS Tagging?

Choosing the best sequence of tags for a given sequence

What two simplifying assumptions do Hidden Markov Models (HMMs) make for POS tagging?

1. Markov Assumption: The current tag depends only on the previous tag (transition)

2. Emission Assumption: The output word depends only on its own tag.

What algorithm is used for decoding the optimal tag sequence in an HMM tagger?

The Viterbi algorithm, which uses dynamic programming to find the most probable sequence of hidden states.

What is a Treebank?

A corpus where each sentence is annotated with syntactic tree structures.

What is a Context-Free Grammar (CFG)?

A formal grammar defined by the tuple (N, T, S₀, R) that models language structure by capturing word constituents and their ordering rules.

What are the four components of a Context-Free Grammar tuple?

N = set of non-terminal symbols (tags), T = set of terminal symbols (tokens), S₀ = start symbol, R = set of production rules.

What is a Probabilistic Context-Free Grammar (PCFG)?

A CFG where every production rule is assigned a probability, allowing the model to rank and disambiguate multiple valid parse trees.

How is the probability of a specific parse tree computed in a PCFG?

By multiplying the individual probabilities of all production rules applied to build the entire tree structure.

How does a Dependency Grammar (DG) differ from a Context-Free Grammar (CFG)?

Dependency Grammars eliminate the concept of phrasal constituents, instead modeling direct head-to-dependent relationships between individual words.