A Dictionary and Morphological Analyser for English

What it is about

The paper describes the first state of a dictionary system for English built for database front-ends. It is designed for at least 10,000 words, with inflectional and derivational morphology. The lexicon is a file of entries with a syntactic category in the feature notation of Generalized Phrase Structure Grammar, and the word grammar is a unification grammar run by an active chart parser. Three mechanisms keep the entries short: completion rules add features from existing ones (a word with plural marking is marked as not inflectable), and multiplication rules copy an entry, for instance to give nouns and adjectives an intransitive variant. Features such as INFL and STEM stop double inflection, so that "walked" and "walking" are accepted and "walkinged" is not. Fifteen spelling rules are compiled into deterministic finite-state transducers. The prototype has about 3,500 entries, which compile in about ninety minutes. No accuracy figure is reported.

Ideas

• I1. en/src/morph.rs: a completion table in code, for example "PLU+ implies not inflectable", "ER-comparative implies not inflectable", written as a list of rules with a reason for each, and applied to the lexicon at build time.
• I2. Use the INFL/STEM idea as a negative control for the lemmatiser: "walkinged", "childrens" and "goed" must not be produced by en/src/morph.rs, and a test that generates them must fail.
• I3. The spelling rules as one transducer per rule: the epenthesis rule (insert e before s after sibilants, as in "boxes") can be a small deterministic table in en/src/morph.rs, tested over all lemmas in the dictionary, with a count of changed forms as the measure.
• I4. Store irregular forms in the lexicon rather than in rules, as the paper recommends, and measure the size of each table. Report both the table size and the lemma accuracy on a held-out list, to show the cost.

For Mova

The chapters are 06. Spelling rules: inflection, derivation, compounds and national standards for the spelling rules, 11. Nouns and 13. Verbs for the inflection, and 14. Irregular verbs for the stored forms. The code is en/src/morph.rs (inflection and morph::archaic) and en/src/dict.rs. The measure: lemma and inflection accuracy on a held-out list, the count of generated illegal forms (must be zero, with a negative control of double inflections), and the size of the stored irregular table, run twice for determinism.

Cite

The paper itself, not this card. BibTeX from the library's record of the work.

@inproceedings{russell1986dictionary,
  title = {{A Dictionary and Morphological Analyser for English}},
  author = {Russell, G.J. and Pulman, S.G. and Ritchie, G.D. and Black, A.W.},
  booktitle = {Coling 1986 Volume 1: The 11th International Conference on Computational Linguistics},
  year = {1986},
  url = {https://aclanthology.org/C86-1066/}
}

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/russell-1986-dictionary-morphological-analyser-english/