Unsupervised Disambiguation of Syncretism in Inflected Lexicons

TL;DR

This paper introduces an unsupervised neural network approach to disambiguate morphological syncretism in inflected lexicons, enabling probabilistic analysis of ambiguous word forms without context.

Contribution

It presents a novel neural model that disambiguates morphological feature bundles using only unigram counts, without relying on contextual information.

Findings

Effective disambiguation across 5 languages

Model handles rare feature bundles well

Provides a new evaluation framework for lexical ambiguity

Abstract

Lexical ambiguity makes it difficult to compute various useful statistics of a corpus. A given word form might represent any of several morphological feature bundles. One can, however, use unsupervised learning (as in EM) to fit a model that probabilistically disambiguates word forms. We present such an approach, which employs a neural network to smoothly model a prior distribution over feature bundles (even rare ones). Although this basic model does not consider a token's context, that very property allows it to operate on a simple list of unigram type counts, partitioning each count among different analyses of that unigram. We discuss evaluation metrics for this novel task and report results on 5 languages.