How does Tangut work?

This thesis investigates the broad question of “How does Tangut or Xixia work?”. The initial chapter considers how the Tangut script works. It starts with an overview of the recursive radical index previously developed by the author. The structure within the characters is examined using entropy techniques. Genetic algorithms (a machine learning technique) are then used in an attempt to relate phonetics to character structure. The question as to whether the script might represent two different languages is also examined. The second chapter examines how Tangut grammar works. Starting by translating Kepping’s morphology into English, the most important points of Tangut grammar have been selected. These points have been illustrated by a detailed explanation of selected examples drawn mainly from the legal texts in the third chapter and Kepping’s original work. This chapter can also be used as a textbook for learning Tangut grammar. The third chapter provides supporting material for the second chapter by investigating in detail a section of the New and Reformed Law Codes. Novel software techniques have been developed to tag the characters in the original text with explanatory keywords and collate them. Keywords have been chosen to either reflect the meaning of the characters or the Chinese pronunciation for transcription characters. The Russian translation in Kychanov’s work is then used to build a line-by-line explanation of the text. The practical language in the legal texts enables this chapter to be used as an introductory reader. The appendices go into further details of the software tools developed to search for characters on various criteria using regular expressions, build concordances, enter sample sentences and annotate texts. A list of the explanatory keywords selected for the characters is also given.