📄 Some Brief Remarks on Teaching Amharic: A Personal View

Summary

This short essay is a personal reflection, not an NLP study. It contains no datasets, models or treebanks. Bausi draws on his teaching and administrative work in Naples (2008–2009) and Hamburg (2010–2011) to ask how the Bologna Process affected university teaching of Amharic in Europe.

He outlines Naples' long tradition, which began with the first permanent Amharic chair in 1891. He then describes a flawed Hamburg setup in which Amharic and Gǝʿǝz were sold to students as one combined "language" that they had to learn at the same time. This arrangement deterred enrolment, because learners faced two grammars, two sound values for the same script, and double exams. Hamburg later reorganised its offer. Amharic now stands as a language course alongside Swahili and Hausa, and Gǝʿǝz is handled inside seminars.

The author argues that Amharic is the indispensable entry point to Ethiopian studies. He also holds that Gǝʿǝz remains essential for serious work on Ethiopia's past. The piece matters as a case study in curriculum design: what you bundle together, and in what order you teach it, shapes whether learners succeed.

Key points

• Context: the text grew out of a 2009 Inalco (Paris) conference paper that was never published, and was updated for a Hamburg workshop on teaching Amharic in Germany (1–2 February 2018).
• History: Naples (Istituto Universitario Orientale) held the earliest permanent Amharic professorship, founded in 1891 with Francesco Gallina. Lanfranco Ricci later held the chair from 1962 to 1986.
• Hamburg until 2010:
  • Students could choose Swahili, Hausa, or "Amharic-plus-Gǝʿǝz", split about two-thirds Amharic and one-third Gǝʿǝz.
  • Staffing was a half lecturer position for Amharic and a quarter position for Gǝʿǝz.
• The reform measures were introduced in 2010–2011 and formally confirmed in 2012. The three languages now run in parallel, with Gǝʿǝz placed in seminars.
• Bausi's reasons for teaching Amharic:
  • Its syntax is Semitic but strongly Cushitized, so it gives insight into both families.
  • It requires mastering the fidal script.
  • Knowing it makes Gǝʿǝz look comparatively easy afterwards.

Interesting ideas

• I1. Two systems that share one written form but differ in sound and grammar should not be packaged as a single unit, because the overlap confuses rather than helps. — For Mova (tokens): Keep separate lexicon and morphology profiles for distinct varieties (for example archaic fable English versus modern English) and tag which one applies, instead of merging them into one analyzer.
• I2. Learning the living, related language first made the older literary one seem relatively simple, so ordering matters more than coverage. — For Mova (learn): Order training as a curriculum, starting with plain contemporary text before archaic or stylised registers, and check whether the later stage then needs fewer examples.
• I3. A language that sits between two families gives a useful vantage point for later specialisation in either direction. — For Mova (tree): When converting between UD dialects, consider a bridging intermediate representation that shows the features of both schemes, so either target can be derived from it transparently.
• I4. As encyclopaedia editor, the author could compare the quality of submitted articles with each author's training, which is an evaluation-by-provenance view. — For Mova (learn): Record which witness parser or LLM teacher tier produced each silver annotation, and correlate error rates with source, so weak sources can be down-weighted.