📄 Monica S. Devens: A concordance to Psalms in the Ethiopic Version

Summary

Alessandro Bausi reviews Monica S. Devens' concordance to the Ethiopic (Ge'ez) Psalms, published in Aethiopica 6 (2003). This is the first concordance of any Ethiopic text and the first product of Devens' long-term plan to concord the whole Ethiopic Bible. Devens decided not to build a critical edition first, since none was expected soon. She concorded the text of Francesco da Bassano's Asmara edition (1923–1926) as it stands. Bausi praises the tool as carefully designed and highly useful. His main objection concerns the choice of base text. He proposes three criteria: accessibility, closeness to the oldest manuscript tradition, and continuity with earlier scholarship. Da Bassano's text partly meets the first two but fails the third. Dillmann's standard Lexicon cites the Psalms from Ludolf's 1701 edition, which Bausi considers the better text, and the two editions often number verses differently. A small sample shows how often references diverge. Bausi also comments on how entries are ordered, on missing enclitics and on where proper names are placed.

Key points

• The volume is Aethiopistische Forschungen 59 (Harrassowitz, Wiesbaden, 2001), XV+546 pages, priced at EUR 104. Its base text is da Bassano's edition, vol. III, pp. 115–225.
• In Dillmann's Lexicon, columns 1–22 under the letter h, 28 of 49 Psalm citations carry a different verse reference than da Bassano's numbering.
• Devens gives double numbering throughout: da Bassano's and that of The New Oxford Annotated Bible (1991).
• Entries follow the Ethiopic syllabary, but interchangeable laryngeals and sibilants are grouped under one heading. This follows Goldenberg's advice and departs from the arrangement in Leslau's dictionary.
• Gaps noted by Bausi:
  • Enclitics (-hi, -hu, -ma, -sa, -nu, -ni, -kò) have no entries.
  • Proper names sit in an unannounced appendix from p. 528.
• Further context from the review:
  • Knibb lists 29 manuscripts of the 14th–15th centuries, so a critical edition of the Psalms would be feasible.
  • The Psalm text was probably fixed by the early 6th century.
  • Lusini's concordance of the Ascension of Isaiah is forthcoming.

Interesting ideas

• I1. When a perfect base resource is unreachable, pick one by explicit criteria (accessible, old, compatible with existing references) rather than waiting.
• I2. Carrying two parallel reference schemes for every item lets users of either edition find their place.
• I3. Bausi checks a small, bounded slice (one letter, 22 columns) to measure how often references disagree, giving a concrete drift rate instead of a vague complaint.
• I4. Letters that freely alternate in spelling are merged into one lookup key, so variant spellings land in the same entry.
• I5. Leaving clitics out of an index may suit quotation hunters but hurts linguistic use, so the audience defines what must be indexed.

For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

• I1 (learn): choosing which treebank or text corpus anchors training could be scored on availability, provenance and compatibility with tools already citing it.
• I2 (tree): when converting between UD dialects or treebank releases, keep a persistent mapping of sentence and token IDs so old error reports and evaluations still resolve.
• I3 (learn): audit a fixed small sample when switching data versions and report the mismatch ratio as a quick health check.
• I4 (tokens): normalize interchangeable spellings (British/American, hyphenation variants) to a shared lemma key before lookup.
• I5 (tokens): give clitics like 's and n't their own lemma entries, because frames and later stages depend on them.