Lemmaform

Lemmaform refers to the canonical dictionary form of a word, used as the base citation form in dictionaries, corpora, and natural language processing. It represents the lemma, the form under which a word is listed in a lexicon, and is typically inflectionally invariant, in that it remains constant across different grammatical forms. Inflected variants such as running, runs, or ran carry tense, number, or other grammatical information that the lemmaform abstracts away for reference and indexing.

In practice, lemmaforms are produced by lemmatization, a process that combines morphological analysis with often part-of-speech tagging to identify the appropriate lemma for a given token. The mapping from inflected forms to lemmaforms can be language-dependent and may require dictionaries or learned models. Some languages have a single lemma per lemmaword; others have multiple lemmas for homographs, requiring disambiguation by context.

Examples: English: running -> run; children -> child; went -> go. French: mange -> manger. In highly inflected languages such as Turkish or Russian, the lemmatizer identifies the base lemma and may attach or store information about the original form separately.

Uses and challenges: Lemmaforms are used in indexing, search, and text normalization, aiding cross-form matching in corpora and dictionaries. They support tasks in natural language processing, machine translation, and linguistic research. Challenges include polysemy and homonymy, irregular inflection, and the treatment of multiword expressions or proper nouns, which may require exceptions or special lemmas.

See also: lemmatization, dictionary form, lexical entry, stemming.