Version 1 of Collate, a new program I have written with the difficulties of medieval vernacular traditions in mind, is now available. Collate aims to help scholars in the preparation of a critical edition based on many sources. It can collate simultaneously up to a hundred texts. It can deal with richly marked-up texts (with special treatment for editorial comments embedded in the text, location markers, editorial expansions and separate collation of punctuation). It provides powerful facilities to allow the scholar to tailor the collation and it can output in many different formats.
The design of Collate was based on my experience of collating forty-six manuscripts of the two neo-Eddic Old Norse poems Gróugaldr and FjÖlsvinnsmál. Collate works interactively with the collation being written to a window as the scholar watches. The scholar may intervene at any point to alter the collation, using either of the tools "Set Variant" or "Regularise". "Set Variant" allows the scholar to over-rule the collation offered by Collate and impose his own collation, even writing a variant that does not appear in the sources into the collation. Collate includes a particularly powerful regularisation facility, derived from my struggles with the highly individual orthographies and spelling systems of Icelandic scribes. "Regularise" enables the scholar to intervene to regularise any word or phrase in any source at any point. The regularisation can be set for a particular word at every point in every source, or for that word only at that place in that source, or various other combinations. Collate will record all variants set and every regularisation made and remember them next time it runs. The scholar can adjust the collation in other ways, switching the base text, suppressing agreements with the base text and collating punctuation tokens separately.Some idea of how Collate works can be seen from the following screen:
Here, one sees in the top window the collation of ten manuscripts of Chaucer's Wife of Bath's Prologue in Collate's default format: that is, each word of the master (in this case, the Hengwrt manuscript) is followed by the variants on that word, with all the manuscripts which have that variant listed. Below the output window is a "control bar", titled here "Collation from 'chaucer'". This contains various buttons ("resume", "stop", "set variant", etc), which permit the scholar to intervene in and control the collation. On the right, the "Collate" menu has been pulled down, and the "Parameters" submenu selected. Through this, the scholar can change the master, see all the sigils for manuscripts which agree with the master, have punctuation collated separately, etc. The other submenus under the "Collate" menu permit yet other controls over the collation; the other menus provide further useful scholarly facilities.
Beside the default format shown in this window, the collation may be output in various critical apparatus forms (including several formats recommended by the Text Encoding Initiative), or scholars may dictate their own format. Through an interface to the EDMAC macros, developed by John Lavagnino of Brandeis University and Dominik Wujastyk of the Wellcome Institute for the production of complex critical editions with the typesetting language TeX, editions with up to five levels of apparatus can be created direct from the output of Collate. The EDMAC macros and an implementation of TeX (OzTeX) are provided with the program.
The length of texts Collate can process is limited only by the storage capacity of the computer. The only requirement is that the text be divided into blocks containing no more than 32766 words each. Collate works on both prose and verse and has been tested successfully on texts in many languages (including Malay, Sanskrit, Latin, Middle English and Old Norse). Since its release in March 1991 over a hundred copies of the program have been distributed to scholars in some fifteen countries.
A set of Guidelines for Transcription, provided with the program, explains the format transcription files should have, so that they can be processed by Collate. The transcription files must be plain ASCII files and can be prepared on any computer. A simple word-processor, Transcribe, is also provided with Collate: this includes various functions specially designed to help transcription.
Collate has been developed as part of the Computers and Manuscripts Project, funded for three years from 1st September 1989 by the Leverhulme Trust at the Oxford University Computing Service with support from Apple Computer. The Project Director is Professor Malcolm Godden.
Collate 1.0 runs only on Macintosh computers (Classic or higher) and requires one megabyte of memory to operate. It is very unlikely that the program will be ported across to any other operating system due to its full use of the Macintosh interface. A hard disc is recommended. It is now available from the Computers and Manuscripts Project, Oxford University Computing Service, 13 Banbury Road, Oxford OX2 6NN, UK. Email PETERR@UK.AC.OX.VAX. A charge of œ20.00 UK, $40.00 US. is made to cover distribution costs. Documentation, sample files, Transcribe (version 1.12) and the OzTeX implementation of TeX for the Macintosh, together with the EDMAC macros, are also provided. A demonstration disc of the program is also available, free from the this address: Dr. Peter M.W. Robinson, Research Officer, Computers and Manuscripts Project, Oxford University Computing Service, 13 Banbury Road, Oxford OX2 6NN. UK.
Peter M.W. Robinson is Research Officer for the Computers and Manuscripts Project within the Oxford University Computing Service. He is chair of the Textual Criticism working group of the Text Encoding Initiative, has edited Old Norse poetic texts, and is the developer of Collate, a computer program widely used in collation of variant texts.
