📄 Thomas Herbst, David Heath, Ian F. Roe, et al., A Valency Dictionary of English: a corpus-based analysis of the complementation patterns of English verbs, nouns and adjectives

Summary

The supplied extraction contains only the cover and metadata pages of Nicolas Ballier's review, published in 2009 in the book-review section of Lexis: Journal in English Lexicology. The body of the review, including its assessment, examples and criticisms, is missing. This summary therefore covers only what the metadata shows. The reviewed work is A Valency Dictionary of English by Thomas Herbst, David Heath, Ian F. Roe and co-authors, published by Mouton de Gruyter in 2004 at 961 pages. Its subtitle describes the problem it addresses: recording which complementation patterns English words take. It does this for three word classes, verbs, nouns and adjectives, rather than verbs alone, and it bases the analysis on corpus evidence rather than intuition. A resource like this matters for any system that has to predict which arguments a head word licenses. The reviewer's evaluation cannot be reported from this text, and the full review should be consulted before citing it.

Key points

• Reviewed book: A Valency Dictionary of English, written by Herbst, Heath, Roe and others; Mouton de Gruyter, 2004, 961 pages.
• Coverage: complementation (valency) patterns of English verbs, nouns and adjectives.
• Method, per the subtitle: a corpus-based analysis.
• The review is by Nicolas Ballier, in Lexis (Université Jean Moulin – Lyon 3), online since 10 May 2009, DOI 10.4000/lexis.1915.
• License: CC BY-NC-ND 4.0. The extracted text holds no review content beyond these bibliographic details.

Interesting ideas

• I1. Valency is treated as a property of nouns and adjectives too, not just verbs, so argument expectations can be attached to any head word.
• I2. Complementation patterns are derived from corpus evidence rather than introspection, which grounds each listed pattern in attested usage.
• I3. A dictionary that lists permitted complement patterns per head word works as an explicit, checkable inventory.
• I4. Pattern inventories per lemma also tell a parser which dependents are likely for a given head.

For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

• I1 (frames): Extend frame lookup so that nominal heads (e.g. a "decision" noun) and adjectival heads (e.g. "afraid") open slots, beyond the verb frames taken from VerbNet.
• I2 (world): Compiled verb-class and argument knowledge could store corpus-attested complement patterns per lemma, so frame decisions can cite observed usage as their explanation.
• I3 (learn): Silver UD trees whose attachments produce a complement pattern absent from the inventory could be flagged for repair or for review by a teacher tier.
• I4 (tree): The perceptron parser or gparse could use lemma-specific complement patterns as features or as a post-hoc check on obj/xcomp/ccomp attachments for nouns and adjectives as well as verbs.