In Kuya (1985), a computer-based quantitative analysis was made of rhyming words occurring in the corpus of ten Pre-Shakespearian plays (including
Everyman, Ralph Roister Doister, Gammer Gurtons Nedle and so on), for which authorship has not been estabilshed due to lack of direct evidence. In order to determine the authorship of the works, distances, or what is called degrees of similarity, were measured in terms of mathematical figures. They were sum total of figures calculated item by item based on the identicalness of graphically defined word forms between respective dramatic works in pairs. The resultative matrices were further converted to figures with three coordinates through factor analysis.
Consequently, several groups have been formed, with some subdivisions. There seems to have been no great difference in authorship ratiocination between this computer-based survey and traditional studies except that the computer attributes authorship of one dramatic work,
Respublica. to an unknown author other than N. Udall.
This paper attempts to discover the factors which have made such grouping possible. In other words, this is a qualitative review of the conclusion reached by the computer-oriented quantitative survey. Special consideration has been given to: (i. a) the paradigmatic relationship between rhyme words and the very same words occurring in a non-rhyming position; between rhyme words and their rhyming counterparts (
i.e. rhyme fellows). These kinds of lexical networks in a particular context will give a clue as to the wholelinguistic source and structure available to respective authors.
(i.b) differences in the distributive pattern of variants:
e.g. words with vowel variants like
harte/hearte/ herte, words with or without a final-
e like
go/goes and so forth.
(ii)syntagmatic relationship: syntactic behaviour of verbs (including auxiliaries) and nouns.
It is concluded, (i) that variants have an important role to play in authorship attribution; (ii) that combinatory inquiry into paradigmatic and syntagmatic relationships of particular lexical items has shed a light on the lexical similarity in some specific cases. This latter point may lead us to a possibility that Respublica is not so far from
R.R. Doister as any other work frequently attributed to N. Udall is. But the amount of observable evidence, on the whole, seems rather small due to a few case studies shown here.
Further research along these lines can make a small contribution to authorship attribution.
抄録全体を表示