One Model to Rule them All: multitask and Multilingual Modelling for Lexical Analysis

When learning a new skill, you take advantage of your pre-existing skills and knowledge. For instance, if you are a skilled violinist, you will likely have an easier time learning to play cello. Similarly, when learning a new language, you take advantage of the languages you already speak. For instance, if your native language is Norwegian and you decide to learn Dutch, the shared vocabulary between these two languages will likely make it easier to learn the new language. In this thesis, I look at learning multiple tasks, learning multiple languages, and the combination of the two, in the context of Natural Language Processing (NLP), which can be defined as the computational analysis of human language. Although these two types of learning may seem different on the surface, I show that they share many similarities.

While traditional NLP approaches consider a single task or language at a time, the aim of this thesis is to answer several research questions dealing with pushing past this boundary. In doing so, the hope is that in the long term, minority languages can benefit from the advances made in NLP which are currently to a large extent reserved for majority languages. This may then have positive consequences for, e.g., language preservation, as speakers of minority languages will have a lower degree of pressure for using majority languages. In the short term, answering the specific research questions posed should be of use to NLP researchers working towards the same goal.