A Quantitative Model of Word Order and Movement in English, Dutch and German Complement Constructions

What it is about

The paper describes word order and movement in complement clauses of English, Dutch and German within Performance Grammar (PG), a psycholinguistically motivated formalism. A clause has a topology of fixed slots: nine for English (F1 to E2) and nine for Dutch and German (F1, M1 to M6, E1, E2). Movement of a phrase out of an embedded clause is "lateral topology sharing": two adjacent slots of the matrix and embedded clause are unified. How many slots are shared is set by two numeric parameters, LS and RS, which the lexical entry of the governing verb supplies. The paper gives these settings for five complement types (interrogative LS=0; finite declaratives LS=1; non-finite VP extraposition LS=3 in English) and derives island effects such as the contrast between "claim that" and "make the claim that". No corpus or measurement is reported.

Ideas

I1. en/src/lin.rs: add a small English table of LS and RS per complement type (interrogative 0, finite declarative 1, non-finite extraposition 3), with the value taken from the governing verb's entry. Measure the linearisation of wh-extracted complements on UD EWT, with a control that forces LS=0 for declaratives, which must lower the score.
I2. Make the island rule deterministic: no slot sharing when the governor is a noun ("the claim that"), as in the paper's (4a) and (4b). Build a minimal-pair list and report that the gate blocks exactly the noun cases.
I3. 25. Relative clauses and 26-questions.md: wh-promotion out of a finite complement is the same operation as the English relative and question cases; a single promotion rule in lin.rs can be tested on both.

For Mova

The English chapters are 23. Complementation and 21-clause-structure.md. The code is en/src/lin.rs (linearisation from the tree), with prag for the question readings. The measure is linearisation accuracy on UD EWT plus the minimal-pair list. The paper's Dutch and German tables do not apply to English, but the parameter idea does.

Cite

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

@inproceedings{harbusch2002quantitative,
  title = {{A Quantitative Model of Word Order and Movement in English, Dutch and German Complement Constructions}},
  author = {Harbusch, Karin and Kempen, Gerard},
  booktitle = {COLING 2002: The 19th International Conference on Computational Linguistics},
  year = {2002},
  url = {https://aclanthology.org/C02-1034/}
}

References

6 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

G ORN , S. (1967). Explicit Definition and Linguistic Dominoes. Systems and Computer Science. Toronto: University of Toronto Press. find
H ARBUSCH , K. & KEMPEN , G. (2000). Complexity of linear order computation in Performance Grammar, TAG and HPSG. In: Proceedings of Fifth International Workshop on Tree Adjoining Grammars and Related Formalisms (TAG+5), University of Paris 7, May 2000. find
J OSHI , A.K. & S CHABES , Y. (1997). Tree Adjoining Grammars. In: Rozenberg, G. & Salomaa, A. (Eds.), Handbook of formal languages (Vol. 3). Berlin: Springer. find
K ATHOL , A. (2000). Linear Syntax. New York: Oxford University Press. find
S AG , I.A. & W ASOW , T. (1999). Syntactic theory: a formal introduction. Stanford CA: CSLI Publications. find
V OSSE , T. & K EMPEN , G. (2000). Syntactic structure assembly in human parsing: A computational model based on competitive inhibition and a lexicalist grammar. Cognition, 75, 105-143. find

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/harbusch-2002-quantitative-model-word-order-movement/