This paper introduces a Sentiment and Emotion Lexicon for Finnish (SELF) and a Finnish Emotion Intensity Lexicon (FEIL). We describe the lexicon creation process and evaluate the lexicon using some commonly available tools. The lexicon uses annotations projected from the NRC Emotion Lexicon with carefully edited translations. To our knowledge, this is the first comprehensive sentiment and emotion lexicon for Finnish.

Emotions have always been central to the human experience: the ancient Greeks had philosophical debates about the nature of emotions and Charles Darwin can be said to have founded the modern theories of emotions with his study The expression of the emotions in man and animals. Theories of emotion are still actively researched in many different fields from psychology, cognitive science, and anthropology to computer science.

Sentiment analysis usually refers to the use of computational tools to identify and extract sentiments and emotions from various modalities. In this dissertation, I use sentiment analysis in conjunction with natural language processing to identify, quantify, and classify emotions in text. Specifically, emotions are examined in multilingual settings using multidimensional models of emotions.

Plutchik’s wheel of emotions and emotional intensities are used to classify emotions in parallel corpora via both lexical methods and supervised machine learning methods. By analyzing emotional language content in text, the connection between language and emotions can be better understood. I have developed new approaches to create a more equitable natural language processing approach for sentiment analysis, meaning the development and evaluation of massively multilingual annotated datasets, contributing to the provision of tools for under-resourced languages.

This dissertation is comprised of ten articles on related topics in sentiment analysis. In these articles, I discuss lexicon-based methods and the creation of emotion and sentiment lexicons, the creation of datasets for supervised machine learning, the training of models for supervised machine learning, and the evaluation of such models. I also examine the annotation process in relation to creating datasets in-depth, including the creation of
a lightweight easily deployed annotation platform. As an additional step, I test the different approaches in downstream applications.

These practical applications include the study of political party rhetoric from the perspective of emotion words used and the intensities of those emotion words. I also examine how simple lexicon-based methods can be used to make the study of affect in literature less subjective. Additionally, I attempt to link sentiment analysis with hate speech detection and offensive speech target identification.

The main contribution of this dissertation is in providing tools for sentiment analysis and in demonstrating how these tools can be augmented for use in a wide variety of languages and practical applications at low cost.

One of the biggest hurdles for the utilization of machine learning in interdisciplinary projects is the need for annotated training data which is costly to create. Emotion annotation is a notoriously difficult task, and the current annotation schemes which are based on psychological theories of human interaction are not always the most conducive for the creation of reliable emotion annotations, nor are they optimal for annotating emotions in the modality of text. This paper discusses the theory, history, and challenges of emotion annotation, and proposes improvements for emotion annotation tasks based on both theory and case studies. These improvements focus on rethinking the categorization of emotions and the overlap and disjointedness of emotion categories.

With the prevalence of machine learning in natural language processing and other fields, an increasing number of crowd-sourced data sets are created and published. However, very little has been written about the annotation process from the point of view of the annotators. This pilot study aims to help fill the gap and provide insights into how to maximize the quality of the annotation output of crowd-sourced annotations with a focus on fine-grained sentence-level sentiment and emotion annotation from the annotators point of view.

This paper is a pilot study that aims to explore the viability of annotation projection from one language to another as well as to evaluate the multilingual data set we have created for emotion analysis. We study different language pairs based on parallel corpora for sentiment and emotion annotations and explore annotator agreement. We show that the data source is a possible one for reliable L1 data to be used in annotation projection from high-resource languages, such as English, into low-resource languages and that this is a reliable way of creating data sets for fine-grained sentiment analysis and emotion detection.

This paper discusses the academic and societal implications of teaching computational methods to humanities students from the perspective of digital humanities. Pedagogical choices are backed up by both pedagogical theory and concrete examples from actual courses and course feedback. The aim of this paper is to introduce clear best-practice recommendations for developing digital humanities teaching with an emphasis on methods teaching in order to increase the number of students who understand such methods and can apply them to their own projects.

We introduce Sentimentator; a publicly available gamified web-based annotation platform for fine-grained sentiment annotation at the sentence-level. Sentimentator is unique in that it moves beyond binary classification. We use a tendimensional model which allows for the annotation of 51 unique sentiments and emotions. The platform is gamified with a complex scoring system designed to reward users for high quality annotations. Sentimentator introduces several unique features that have previously not been available, or at best very limited, for sentiment annotation. In particular, it provides streamlined multi-dimensional annotation optimized for sentence-level annotation of movie subtitles. Because the platform is publicly available it will benefit anyone and everyone interested in finegrained sentiment analysis and emotion detection, as well as annotation of other datasets.

We introduce the Language Change Database (LCD), which provides access to the results of previous corpus-based research dealing with change in the English language. The LCD will be published on an open-access linked data platform that will allow users to enter information about their own publications into the database and to conduct searches based on linguistic and extralinguistic parameters. Both metadata and numerical data from the original publications will be available for download, enabling systematic reviews, meta-analyses, replication studies and statistical modelling of language change. The LCD will be of interest to scholars, teachers and students of English.