lemmaforms

Lemmaforms is a term used in linguistics and computational linguistics to refer to the set of surface forms that correspond to a single lemma. A lemma is the canonical or dictionary form of a word, representing its abstract lexical entry without inflectional variation. The lemmaforms of that lemma are the inflected or derived forms that a speaker might encounter in ordinary text, such as different endings, tenses, or other morphological modifications.

In practical use, lemmaforms are central to lemmatization, a preprocessing step in natural language processing that maps each word form to its lemma. This reduces data sparsity and helps tasks such as part-of-speech tagging, named-entity recognition, and search indexing. For example, the English lemma run has lemmaforms including run, runs, ran, and running. These surface forms all relate to the same underlying lemma but differ in grammatical or aspectual information. Other languages with richer morphology will have larger and more complex sets of lemmaforms.

Challenges in working with lemmaforms include irregular forms, homographs, and polysemous lemmas where a single form can belong to more than one lemma depending on context. Some languages also produce multiword lemmas or highly productive derivations, which can complicate dictionary design and automated lemmatization. In lexicography and corpus linguistics, carefully accounting for lemmaforms improves accuracy in frequency counting, annotation schemes, and cross-language analysis.