Mendeley出力
＜紀要論文＞
シェイクスピアは『ファヴァシャムのアーデン』の作者か : 印象批評と計量文体解析
| 作成者 | |
|---|---|
| 本文言語 | |
| 出版者 | |
| 発行日 | |
| 収録物名 | |
| 巻 | |
| 開始ページ | |
| 終了ページ | |
| 出版タイプ | |
| アクセス権 | |
| JaLC DOI | |
| 概要 | Shakespeare has long been named as one of the possible authors of Arden of Faversham, an anonymous play presumed to have been written around 1590-91. Although most Shakespeare scholars have been skept...ical of his involvement in the play, the editors of The New Oxford Shakespeare, published in 2016, decided to include the play as one of Shakespeare’s co-written works in the collection. What was instrumental in this decision was the method of computational stylistics used for authorship attribution. The present paper examines the validity of the NOS scholars’ decision to assign authorship of part of Arden to Shakespeare, arguing that their attribution using the computer-assisted stylometry lacks statistical reliability and fails to prove his part-authorship of the play. This paper first examines whether Arden has Shakespearean characteristics in terms of its dramatic structure, characterization, and verse style, suggesting that Shakespeare is unlikely to have written Arden. It then reviews two stylometric articles by three scholars of the NOS team who argue for inclusion of Arden in their newly published collection as Shakespeare’s co-work: “Arden of Faversham, Shakespearean Authorship, and ‘The Print of Many’” by J. Elliott and B. Greatley-Hirsch and “A Supplementary Lexical Test for Arden of Faversham” by M. P. Jackson. Elliott and Greatley-Hirsch have used thirty-four plays (including four of Shakespeare’s earliest plays) to create the “profiles” of nine “candidate” playwrights who were active around the time of composition of Arden. The profile of each playwright is generated using the word frequency data extracted from the play or plays he wrote. These authorial profiles, which Elliott and Greatley-Hirsch claim represent the nine playwrights’ respective stylistic characteristics, are compared for authorship identification. However, they can hardly be considered an appropriate basis on which to determine the authorship of Arden, because the sample sizes of control texts used for generating the profiles of the nine playwrights are too small for Elliott and Greatley-Hirsch’s research results to be accepted as statistically reliable. Another problem with their “authors-profiles” is that the sample sizes of the nine playwrights vary greatly from one author to another. It is doubtful that the linguistic data collected from a corpus of samples of such unequal size can reasonably be used for authorship attribution. Jackson’s analysis uses a corpus of 137 plays of the period 1580-1600, which includes twenty-three Shakespearean plays and twenty-nine of unknown authorship. His attribution is based on the frequency of occurrence of five “Shakespeare-plus-words” and four “Shakespeare-minus-words” in these plays. He calculates the relative frequency of the Shakespeare marker words for each play. He sets one of the most important “borderlines” for classifying Shakespeare and non-Shakespeare plays at 68.5 per cent (Borderline A), claiming that any play or scene with a score above this borderline is likely to have been written by Shakespeare. He suggests that since Arden as a whole scores 66.0 per cent, closely matching Edward III (66.7 per cent), one of Shakespeare’s co-authored plays, Shakespeare very likely had a hand in the play, and that scenes 4-9 of Arden, “with their exceptionally high score of 86.7 percent,” can be unequivocally classified as Shakespeare’s. It is, however, difficult to agree with Jackson’s conclusion because there are a number of problems with Jackson’s attribution methodology. His corpus includes those four plays of well attributed single authorship by Shakespeare which are used in Elliott and Greatley-Hirsch’s analysis. Three of the four score below Borderline A, indicating that there is something wrong with Jackson’s statistical attribution. In addition, Jackson’s attempt to identify authorship using the frequency of occurrence of only nine words causes suspicion of the validity of his method. No analysis based on the data extracted from such a tiny sample of words is expected to provide much statistical reliability.続きを見る |
| 目次 | １．アーデン殺人事件、史書の記録、戯曲創作・出版 ２．『ファヴァシャムのアーデン』の作者同定― 印象批評から計量文体解析へ ３．New Oxford Shakespeare の計量文体解析（統計学的文体分析）による作者判定 ４．計量文体解析による『ファヴァシャムのアーデン』の作者同定 ５．NOS の統計学的作者同定方法の問題点 |
詳細
| PISSN | |
|---|---|
| NCID | |
| レコードID | |
| 登録日 | 2020.11.10 |
| 更新日 | 2023.11.01 |