Morphological Analysis

The process of decomposing a word into its smallest meaning-bearing units. A word like 'unbreakable' is segmented into the prefix 'un-', the root 'break', and the suffix '-able'. This segmentation is the foundational step that enables all downstream grammatical inference.

Identifies grammatical variants of a word without changing its core part of speech. For example, analyzing 'running' reveals the root 'run' with a +VBG (present participle) tag. This is critical for normalizing verbs to their base form for search recall.

Handles morphemes that change a word's meaning or part of speech. The suffix '-tion' transforms the verb 'educate' into the noun 'education'. Understanding derivation allows NLP systems to link semantically related concepts across different syntactic categories.

A computational model commonly used to implement morphological analyzers. An FST maps surface forms like 'cats' to an underlying representation of 'cat +N +PL' (noun, plural). This bidirectional mapping is highly efficient for languages with complex agglutinative morphology.

Morphological analysis is the engine behind lemmatization, which uses a vocabulary and part-of-speech context to return the dictionary form ('better' → 'good'). In contrast, stemming is a cruder, rule-based heuristic that simply chops affixes ('running' → 'runn') without lexical knowledge.

A specialized morphological task for languages like German or Dutch that frequently form long compound nouns. The system must segment 'Donaudampfschifffahrtsgesellschaft' into its constituent parts: 'Donau', 'Dampf', 'Schiff', 'Fahrt', 'Gesellschaft' to enable accurate translation and indexing.

The process of reducing a word to its canonical dictionary form, or lemma, by using a vocabulary and morphological analysis of the word's part of speech. Unlike stemming, lemmatization requires understanding the word's context to correctly map inflected forms like 'ran' to 'run' or 'better' to 'good'. This is a direct downstream consumer of morphological parsing.

A heuristic process of removing affixes from a word to reduce it to a common base form, often resulting in a non-dictionary stem. While computationally cheaper than full morphological analysis, it operates without understanding the word's grammatical context, frequently conflating semantically distinct terms like 'university' and 'universe'.

The process of assigning a grammatical category, such as noun, verb, or adjective, to each token in a text. This is a critical prerequisite for accurate morphological analysis, as the same surface form (e.g., 'saw') can have entirely different morphological decompositions depending on whether it is a past-tense verb or a present-tense noun.

The fundamental NLP task of segmenting a continuous string of text into discrete units, or tokens. Morphological analysis operates on these tokens, and the choice of tokenization strategy—particularly subword tokenization like Byte-Pair Encoding—directly impacts the analyzer's ability to isolate and interpret individual morphemes.

Analyzes the grammatical structure of a sentence to establish relationships between words. Morphological features like tense, number, and case are critical input signals for a dependency parser to correctly identify syntactic heads and label relationships such as nsubj (nominal subject) or dobj (direct object).

A term for any token encountered during inference that was not present in a model's fixed vocabulary. Robust morphological analysis is a key strategy for handling OOV words by decomposing them into known constituent morphemes, allowing a system to infer the meaning of a novel compound or inflected form.