有鑒於以往房價泡沫文獻較少探討泡沫與房屋特徵間的關係，本文援引特徵價格法研究內外部特徵對房價泡沫的影響。藉由所建構之台中市房屋交易個案資料庫，將交易區分為獨立式住宅與集合式住宅，並考量明星學區環域的影響，全面納入「建築內部」、「教育」、「休閒」、「生活機能」、「區位」、「交通」、「安全」、「鄰避設施」八類衡量生活品質的房屋特徵，以房價泡沫取對數值作為應變數，住宅與各類設施距離取對數值為自變數，將樣本區分為六十五個學區進而實證歸納出學區環域內容易被炒作的特徵。 房地產市場中的獨立式住宅及集合式住宅兩大類商品，其在追求生活品質的同時符合Hotelling模型極小化產品差異的結果，由研究成果顯示集合式住宅渴望在繁榮的鬧區保有安靜的生活環境，而獨立式住宅則是在空曠的郊區追求生活機能，前者為「鬧中取靜」，後者為「靜中取鬧」，換言之集合式住宅與公園綠地等休閒設施距離越近泡沫越大，但對於會帶來大量車流和人潮的交通樞紐及大型遊樂場所等則是距離越遠，泡沫越大，獨立式住宅方面則是距離連鎖餐飲、便利商店等生活機能設施越近，泡沫越大。 / Since there were not many papers focusing on the topic of the relationship between bubbles and housing characteristics, this paper studies the internal as well as the external causes of housing bubble by adopting the canonical Hedonic Pricing Method. By utilizing a constructed unique database composed of Taichung housing transactions, the detailed features of transactions on the detached houses and condominiums can be investigated. The "Building Interior", "Education", "Leisure", "Living Infrastructure", "Location", "Transportation", "Security", and "NIMBY" are collected as the measurement of quality of life in housing characteristics to complement the Star School District impact in the educational buffer. Using the level of house price bubbles as dependent variable and dividing the selected area into 65 school districts, and discover what factors are used for house price speculation. The results show similar demand on the quality of life, in line with Hotelling Model, in both the transactions of detached houses and condominiums. The residents of condominium; however, prefers living in the prosperous urban area with Serene environment whereas the occupants of detached house would choose to live in the spacious suburban area with decent local living facilities. In other words, the closer the condominiums are with leisure facilities, such as public park, and the further they are with facilities that gathers crowd and traffic, such as transportation hub and entertainment facilities, the larger the bubble. The further the detached houses are with living function facilities such as chain restaurants, convenient stores, the larger the bubble.

Research Doctorate - Doctor of Philosophy (PhD) / The aim of the thesis is to use computational stylistics, and in particular the methods pioneered by John Burrows, to explore aspects of the nineteenth-century periodical genre. Published for the most part anonymously, periodical articles were written by an extraordinary range of authors on an incredible variety of topics. The standard of writing in the thousands of articles appearing in the ‘higher’ or ‘literary’ journals has generally been agreed by scholars to be ‘remarkably good’. Beginning in 1802 and flourishing for most of the century, this outstanding genre of writing had all but disappeared by the beginning of the twentieth century. The text collection for the thesis consists of almost two million words by twenty-two authors. My study employs a variety of statistical tests on these texts to examine the effect of such factors as anonymity, commonality, authorial individuality, gender, house-style, text-type and chronology on the periodicals. I begin by taking a broad view of the field: first allowing the articles to ‘speak for themselves’ and to exhibit their commonalities and individual differences; then exploring the significance of both the intra-generic focus of the article – the stance taken in a particular article – and the author’s own idiosyncratic preferences in determining the incidence of function words in these articles. The interplay between these two factors provided an explanation as to why the articles of some authors invariably grouped together while those of other authors displayed marked variability. The use of lists of authorial ‘marker words’ – those words used relatively more or relatively less frequently by individual authors – showed that one can think of this large group of mostly anonymous periodical articles as a set of authorial oeuvres. I also look at the frequently made assertion that authors adapted their writing to the ‘house style’ of particular journals, and come to the conclusion that it does not significantly affect the deeper level of style revealed by function word usage. I then examine the question of whether or not there are differences between men’s and women’s usages of function words, coming to the conclusion that, although differences can be seen to exist, it is not at present possible to come up with sets of ‘marker words’ that reveal gender in the way that is possible with authorship. I use ‘marker words’ to identify the characteristics of one major author, George Eliot, and to show how she modified her stylistic practices when she moved from the periodical essay to fiction. I demonstrate how the techniques of computational stylistics can be used to check the legitimacy of some of the attributions made in the Wellesley Index, and I attribute one much-discussed anonymous group of articles on ‘the woman question’ to Robert Cecil 3rd Marquess of Salisbury and Prime Minister of England.