Speakers use a specific style of language to convey important information about the situational context and speaker purpose. For example, professional editors polish and cater an article to the target audience with proper choices of words and grammar; human translators translate a document for a specific audience and often ask what is the expected tone of the content when taking a translation job.

Computational models of natural language should consider both their meaning and style. Controlling a specific style is an emerging research area in text rewriting and is under-investigated in machine translation. In this dissertation, we present a new perspective which closely connects formality transfer and machine translation: we aim to control style in language generation with a focus on rewriting English or translating French to English with a desired formality. The main challenge lies in the limited availability of annotated examples of style transfer.

We first address this problem by inducing a lexical formality model based on word embeddings and a small number of representative formal and informal words. This enables us to assign sentential formality scores and rerank translation hypotheses whose formality scores are closer to user-provided formality level. To capture more context, we also directly model the sentential formality using formality transfer examples with a neural sequence-to-sequence architecture. Joint modeling of formality transfer and machine translation enables formality control in machine translation without dedicated training examples. Along the way, we also improve low-resource neural machine translation.
 

Examining Committee: 

 

                          Chair:               Dr. Marine Carpuat
                          Dean's rep:      Dr. Douglas Oard
                          Members:        Dr. Jordan Boyd-Graber
                                                    Dr. Furong Huang
                                                    Dr. Philipp Koehn