1976 年 24 巻 p. 75-86
How do phraseology and meter correlate with each other in the dialogue verse ・of Greek tragedy? With this question in mind, we set about analyzing all the extant trimeters with the help of a computer. The following is an interim report on some points of interest. (i) Conversion of texts into machine-readable form. The system we adopted was that of the normal transliteration into Roman alphabet with some modifications; long vowels marked by* , prosodical signs(-for crasis, +for synizesis, &c.) added. At present 6660 lines of eight plays are converted and reposited on magnetic tape. (ii) Automatic scansion. The computer scans trimeter according to the prosodical rules and, recognizing resolutions as such, writes out a metrical scheme to each line. The resulting, completely scanned texts serve as the raw materials for subsequent inquiries. For the symbols used in scansion see Explanatory Note to fig. 1-a. (iii) Automatic production of concordance. Two Concordances were made; the usual, Alphabetical one and the one in which words are classified and arranged .according to their Metrical word-types. These are reposited on MT and serviceable ior various further uses. Cf. fig. 1-a〜c. (iv) Examination of metrical features of lines. Lines with any particular metrical features can be assembled and examined at a stroke with the use of the computer. The example shown in fig. 3 is an inquiry into metrical behavior after caesura of the lines with penthemimeral caesura. (v) Examination of 'correptio Attica'. All occurrences of the sequence of short vowel-mute-liquid were assembled and classified according to the constituent 'Consonants. Refined statistics were drawn there from. The gross figure is 960 short .syllables(resolutions excluded)against 354 longs in 6660 lines. Cf. fig. 2. (vi) Examination of lengthening by position of word-final open syllable. This ・prosodical abnormality occurs 183x in the lines examined, mostly between words that cohere closely together. Each individual case can be examined in the list prepared by the computer. For some of the more conspicuous cases see above p.78. (vii) Study of localization and distribution of metrical word-types. Occurrences of each word-shape, at each metrical position of the verse, were counted and a comprehensive list of distributional figures for each shape was made; cf. fig. 4-a〜b. From the list certain tendencies become apparent: Every shape has its preferred position or positions. For instance, one of the fittest forms to iambic, SLSL, which can be used in four positions, shows a tendency towards localization at the verse-end(cf. fig. 4-b), and SL, iamb itself, too. On the contrary, SLS_0 and SLS_1, which are also shapes suitable to iambic, appear mostly in the former half of the verse. Sometimes preference amounts to restriction. SLL and LSLL are used each in one position only(this is a corollary of Porson's Law). These tendencies are consistent throughout all the plays regardless of their author or the date of composition, although some innovations were made and introduced by using resolution or crasis(cf. fig. 4-b and"-SSL" &c. in fig. 4-a). They may beinherent properties of the iambic trimeter, and if we examine further the words themselves from semantical and syntactical point of view, the system on which poets subconsciously depend in versification will be correctly described. The figures referred to are on pp. 80〜86 above.