# Augmenting a BiLSTM tagger with a Morphological Lexicon and a Lexical Category Identification Step

## TL;DR

This paper enhances BiLSTM part-of-speech tagging for Icelandic by integrating a morphological lexicon and a lexical category step, significantly improving accuracy over previous models, especially for complex, morphologically rich languages.

## Contribution

The study introduces a novel approach combining a morphological lexicon and a lexical category identification step to improve BiLSTM tagger performance on Icelandic.

## Key findings

- Baseline BiLSTM outperforms previous non-lexicon models.
- Incorporating a morphological lexicon boosts accuracy significantly.
- Using a separate lexical category model reduces errors by 21.3%.

## Abstract

Previous work on using BiLSTM models for PoS tagging has primarily focused on small tagsets. We evaluate BiLSTM models for tagging Icelandic, a morphologically rich language, using a relatively large tagset. Our baseline BiLSTM model achieves higher accuracy than any previously published tagger not taking advantage of a morphological lexicon. When we extend the model by incorporating such data, we outperform previous state-of-the-art results by a significant margin. We also report on work in progress that attempts to address the problem of data sparsity inherent in morphologically detailed, fine-grained tagsets. We experiment with training a separate model on only the lexical category and using the coarse-grained output tag as an input for the main model. This method further increases the accuracy and reduces the tagging errors by 21.3% compared to previous state-of-the-art results. Finally, we train and test our tagger on a new gold standard for Icelandic.