For the purposes of this project the reconstruction process may be considered to be a serial process comprising (i) transcription, (ii) translation, (iii) analysis, and (iv) modelling, although to some extent it is iterative and it is often necessary to return to earlier steps to complete the process.
The problem of the reconstruction of Brittonic is central to providing accurate translations of literature and epigraphs, and thence personal names, therefore documents and monuments are the raw stuff of reconstruction. But without an accurate and consistent model of Brittonic translation quality will be poor or incomplete, and most likely based on assumptions depending on the language skills of the translator (who may be primarily concerned with Latin, Welsh, Breton or less likely, Cornish). However, reconstruction activities depend on the quality of transcription of originals, but transcription usually involves some knowledge of the source language, so it is not an activity truly independent of translation.
To perform a correct translation, translators need enough of the text, in ungarbled form that makes grammatical sense. This is a chicken-and-egg problem, and whilst it may not be possible to avoid it, it needs to be recognised, as does the possibility of errors being introduced during transcription. Once enough data has been collected it ‘may’ be possible to identify both distinct individuals and multiple individuals having the same name. However, differences in spelling need to be taken into account, thus requiring phonological analysis of the data and normalising more archaic forms against later forms.
Until the C20 philologists held the ring for the entire process of manuscript translation, being a one-stop shop performing transcription, translation, analysis and literary criticism activities. Although philological interest has now become fragmented across several fields much of the information relating to Brittonic originates from this earlier period of philology and philologists: their influence remains potent.
Celtic philology, in particular, has a “rich and enduring tradition” (Koch, 1990), and features names relevant to Cornish studies and the Cornish language, such as Whitley Stokes, 1830–1909 and Louis Lucien Bonaparte, 1813–1891 (Davis, 2002). They were roughly contemporary with Sir John Rhŷs, 1840–1915 who is credited with achieving academic status for Celtic philology, and became the first Professor of Celtic at Oxford. However, Celtic philology has now faded as a field of research, and publication of newly discovered manuscripts is now the domain of specialist linguistics and historians as exemplified by the publication of Bewnans Ke (G. Thomas & Williams, 2016).
Performing detailed analysis of medieval manuscripts involves a high degree of competence in a several areas, not just linguistics and history and may involve textual criticism specialists, materials analysts and paleographers [Ref]. A list of difficulties associated with manuscript translation would be lengthy, but a discussion of which may be found in XXXX [Ref]. By way of illustration difficulties include inability to unravel unpunctuated or unstructured text, poor quality of the book hand used (uncial, insular miniscule or Caroline miniscule) or unfamiliarity with the language used. This may be by the original writer, copyist and/or translator, perhaps with errors amplified at each step.
A review of Hammer & Geoffrey of Monmouth (Author) (1951) illustrates some of these difficulties, particularly distinguishing names rendered with numerous down strokes of the pen <i,m,n,u,v> (in Duvianus, Dumanus, Dimianus), differentiating between similar pairs of graphs such as <th> and <ch> (in Arthgal, Archal), and confusion between single characters and tightly kerned characters such as <d> and <cl> (in Archgallo, Ardigallo). For Brittonic several possible readings of a name may be feasible. There is also the problem of patternicity, since a reader may tend to settle on a name they recognise, perhaps from another text: conflation of names into a single form is therefore a risk, as are the risks of proliferating variant names as distinct and inadvertently eliminating real new names.
More recently computer science has started to make an invaluable contribution through specialised computerised algorithms for identifying the likely author (Van Dalen-Oskam, 2014), or performing genealogical analysis of variant texts [stemmatology] (Camps & Cafiero, 2014). With the advent of computing the disciplines of digital humanities and digital classics have emerged. Support for data standards (mark up standards) now exist for digital transcription called EpiDoc (“EpiDoc: Epigraphic Documents in TEI XML download | SourceForge.net,” n.d.).
Latin scribal tradition and the Latin character set have been the basis for rendering Celtic languages since they began to be recorded on paper. There are several issues that arise from the use of Latin characters, including:
- The inherent problem of attempting to record a spoken language with a set of graphs not designed to represent it fully or accurately, [Ref phonology vs. orthography p?]?
- Variation in the spoken form and how this is dealt with orthographically, and in particular, [Ref]? and
- Interference from the language supplying the graphical system. This is a particular issue with the use of Latin graphs for Brittonic and is therefore within the scope of this study. [Ref]?
The rules for transcription (phonemic conversion) are generally set by convention (for example Greek <κ> à Latin <c>). Transcription is a more or less mechanism process. However, translation (language conversion) is often difficult, particularly where monuments or manuscripts have been degraded or were poorly originated. The process of reconstructing monumental inscriptions in difficult cases is illustrated by Bultrighini (2017), who suggests an approach for partially dealing with mis-transcription, but even this does corrective approach does not deal with the issue of mis-transcribing through misunderstanding the source language (mistaking Celtic for Latin, for example). An illustration of the difficulties in translating an unfamiliar language may be found by examining Thorpe’s transcription of the Bodmin Manumissions (Thorpe, 1865). Here he has read <cl> for <d>:
*Anauclat < *Anaudat, and *Aeniucl < *Aniud.
The potential for throwing out subsequent translation or classification attempts is obvious, although in this case Thorpe has been superseded so it is unlikely to result in errors by future researchers. Translators have frequently had to deal with incomplete or corrupted forms, faded writing and damaged stonework, as well as poor quality workmanship by the original authors, whose literacy may be questionable.[ref] Furthermore, the translator may have mis-read, mis-understood or deliberately reworked the primary source, perhaps for purposes of linguistic rehabilitation (Ellis, Evans, & Cardiff Conference on the Theory and Practice of Translation in the Middle Ages (3rd : 1991), 1994, p.3). Moreover, he points to the grey area between being a scribe and a translator (ibid., p.9)
The subject of manuscript quality is a huge subject in its own right, so will only be considered in so far as it affects the extraction of proper names from texts. The principal issues that are relevant are, (i) variation between texts, such as the same name spelt differently, and (ii) incomplete or corrupted forms.