Hakuōki, produced by Idea Factory, is an otome game series set in late Edo periodJapan. It is among the most popular video game series in its genre. The games inthis series have been localized for the English video game market, one that is verydifferent from the source culture. The game play experience is expected to beequivalent in all language versions of the game – however, given that any twolanguages cannot be completely equivalent and have different cultural contexts,translation problems are expected to arise. Acknowledging these problems, thisstudy aimed to determine whether or not, due to choices made in the translationprocess, dialogue was altered. Consequently, the study aimed to find out if aselected character’s personality could be perceived differently between the playersof two language versions of the video game Hakuōki. The study was conducted intwo parts: first, a translation analysis was conducted by the researcher. Secondly,to see if character personality was altered in translation, selected lines were used ina questionnaire survey that was conducted both in English and Japanese. Therespondents were asked to describe their perception of the character based on theexcerpts provided in the questionnaire survey. The results of the surveys werecompared and contrasted with the findings of the translation analysis. The resultsof the study show that dialogue was altered in translation, and that the selectedcharacter’s personality can be perceived differently between the players of twolanguage versions of the video game.

A family of Natural Language Processing (NLP) tasks such as part-of- speech (PoS) tagging, Named Entity Recognition (NER), and Multiword Expression (MWE) identification all involve assigning labels to sequences of words in text (sequence labeling). Most modern machine learning approaches to sequence labeling utilize word embeddings, learned representations of text, in which words with similar meanings have similar representations. Quite recently, contextualized word embeddings have garnered much attention because, unlike pretrained context- insensitive embeddings such as word2vec, they are able to capture word meaning in context. In this thesis, I evaluate the performance of different embedding setups (context-sensitive, context-insensitive word, as well as task-specific word, character, lemma, and PoS) on the three abovementioned sequence labeling tasks using a deep learning model (BiLSTM) and Portuguese datasets. v