Fine-tuning a Subtle Parsing Distinction Using a Probabilistic Decision Tree: the Case of Postnominal "that" in Noun Complement Clauses vs. Relative Clauses

TL;DR

This paper explores methods to distinguish between relative and noun complement clauses in English, focusing on the use of probabilistic decision trees and corpus relabeling to improve parsing accuracy.

Contribution

It introduces a novel application of TreeTagger to differentiate postnominal "that" in relative and complement clauses, analyzing training set effects and corpus representativeness.

Findings

TreeTagger effectively learns the distinction with sufficient training data

Relabeling corpus improves parsing accuracy

GUM Treebank files are somewhat representative of the structures studied

Abstract

In this paper we investigated two different methods to parse relative and noun complement clauses in English and resorted to distinct tags for their corresponding that as a relative pronoun and as a complementizer. We used an algorithm to relabel a corpus parsed with the GUM Treebank using Universal Dependency. Our second experiment consisted in using TreeTagger, a Probabilistic Decision Tree, to learn the distinction between the two complement and relative uses of postnominal "that". We investigated the effect of the training set size on TreeTagger accuracy and how representative the GUM Treebank files are for the two structures under scrutiny. We discussed some of the linguistic and structural tenets of the learnability of this distinction.