Determining Determiner Sequencing: A Syntactic Analysis for English
What it is about
English allows sequences of determiners, such as all these, both the, or half the, but in a fixed order, and some sequences are ungrammatical (all what no). The paper argues that the usual three-way split of determiners into predeterminers, determiners and postdeterminers is too coarse. It proposes eight semantic features that order the determiners: definiteness, quantity, cardinality, genitive, decreasing, constancy, wh, and agreement. Each determiner carries values for these features, and a selection feature on its modifier list which determiners it may modify. Genitives, partitives and determiner adverbs such as almost or only are handled by the same features. The analysis is written as trees in a feature-based lexicalised tree-adjoining grammar and is implemented in the XTAG English grammar. The paper counts 103 determiners in a dictionary, of which the analysis covers 82.
Ideas
- I1.
en/src/dict.rs: add the eight features as columns for the determiners. Write the ordering rules as checks over UDdetanddet:predetchains in an NP, so that an ungrammatical order is rejected. Measure the rate of wrong orders on the UD EWT dev noun phrases. - I2.
en/src/lin.rs: use the same feature table when linearising a noun phrase, so that generated phrases get the right determiner order. Test on generated phrases with two or three determiners. - I3. Negative control: the selection restrictions (for example, all cannot take a no) should block the bad sequences in the paper's examples. A check that accepts them shows a missing restriction.
For Mova
The chapter is 12. Pronouns, determiners and numerals, with 22. Phrases: noun, adjective, adverb and prepositional phrases for the noun phrase. The code is en/src/dict.rs (feature columns), en/src/lin.rs (linearisation) and the UD checks in en/src/parse.rs. The measures: order-error rate on the EWT dev noun phrases, the linearisation test set, and the negative-control rejection rate. The features are a deterministic table, so they fit Mova directly.
Cite
@misc{hockey1994determining,
title = {{Determining Determiner Sequencing: A Syntactic Analysis for English}},
author = {Hockey, Beth Ann and Egedi, Dania},
year = {1994},
eprint = {cmp-lg/9411004},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/cmp-lg/9411004}
}References
- Abeille, A. 1990. French and English Determiners: Interaction of Morphology, Syntax, and Semantics in Lexicalized Tree Adjoining Grammars. In Tree Adjoining Grammars, First International Workshop on TAGs: Formal Theory and Applications (abstracts), August. find
- Barwise, J., Cooper, R. 1981. Generalized Quanti ers and Natural Language. In Linguistics and Philosophy 4. find
- Doran, C., Egedi, D., Hockey, B.A., Srinivas, B., and Zaidel, M. 1994. XTAG System - A Wide Coverage Grammar for English. In Proceedings of the 15th International Conference on Computational Linguistics , Kyoto, Japan, August. find
- Kennan, E. L., Stavi, J. 1986. A Semantic Characterization of Natural Language Determiners. In Linguistics and Philosophy 9. August. find
- Joshi, A., Levy, L., and Takahashi, M. 1975. Tree adjunct grammars. In Journal of Computer and System Sciences . find
- Partee, B., ter Meulen, A., and Wall, R. 1990. Mathematical Methods in Linguistics . Kluwer Academic Publishers. find
- Quirk, R., Greenbaum, S., Leech, G., and Svartvik, J. 1985. A Comprehensive Grammar of the English Language . Longman Group Limited. find
- Schabes, Y. 1990. Mathematical and Computational Aspects of Lexicalized Grammars . University of Pennsylvania Dissertation. 6 This is not the same set of lexical items that we counted as adjectives in the Introduction. Together, there are 14 lexical items marked as determiners in Collins that we feel should be adjectives. find
This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/hockey-1994-determining-determiner-sequencing-syntactic-analysis/