A Lexical Process Model of Nominal Compounding in English

What it is about

The paper asks how a noun-noun compound such as "shrimp boat" or "flower market" is formed and understood. The author argues that a compound is a reduced relative clause ("a boat that is characteristically used to catch shrimp"), and that a compound can be built only when the verb of that clause has a characteristic, habitual aspect. The rules are lexical processes, not surface rewrites: each rule pairs a clause structure with a compound, is indexed by the clause verb and by the right-hand nominal, and carries exclusion lists. Binding of nouns is done by identity of reference sets. Where identity fails, as in "person" bound to "thief", the lexicon is searched for a rule that links the two. The model is a pattern-matching interpreter with backtracking, and was tested on "a set of several hundred nominal compounds". No accuracy figure is given; the claims are qualitative.

Ideas

• I1. languages/en/18-compounding-conversion-minor.md: write the compounding rules as lexical rules indexed by (clause verb, right nominal), with the habitual-aspect gate as an explicit condition. A compound is licensed only when the gate fires, and the log says which gate and which rule.
• I2. en/src/morph.rs and world/src/dictionary.rs: use the same index to interpret a new compound. For "turtle boat" the rule "boat used to catch marine animals" gives a paraphrase; record the paraphrase and the rule id, so the reader can say why.
• I3. Measure the paper's claim instead of repeating it. Take several hundred compounds with hand-written paraphrases, and report paraphrase accuracy with and without the habitual-aspect gate, run twice, with a negative control of random pairings that must be rejected.
• I4. Use the exclusion lists as negative examples for learning: each rejected pairing becomes a test case in the chapter, so new rules are checked against what the model must not produce.

For Mova

The chapter is languages/en/18-compounding-conversion-minor.md, with languages/en/19-lexicon-mwe-register.md for the lexical side. The code points are world/src/dictionary.rs (word senses for the school), prag/src/roles.rs (the role a compound's left element plays) and the lexicon in en/src/dict.rs. The measure is paraphrase accuracy on a hand-checked set of compounds, with the gate on and off, and a random-pairing control that must score lower.

Cite

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

@misc{rhyne1975lexical,
  title = {{A Lexical Process Model of Nominal Compounding in English}},
  author = {Rhyne, James R.},
  year = {1975},
  url = {https://aclanthology.org/J75-4018/}
}