📄 Bibliography of Ethiopian Semitic, Cushitic and Omotic Linguistics XIX: 2014/2015

Summary

This is not an NLP paper. It is the nineteenth instalment of an annual reference bibliography, published in Aethiopica 19 (the volume is dated 2016, pp. 90–99). It records scholarly publications from 2014 and 2015 on the Semitic, Cushitic and Omotic languages of Ethiopia and Eritrea. The problem it addresses is practical: work in this small field is scattered across Festschriften, conference proceedings, journals and dictionaries, so researchers need one curated index.

The compilers are Maria Bulakh (Russian State University for the Humanities) and Susanne Hummel and Thomas Rave (Universität Hamburg). They group entries by language family. Collected volumes get short sigla, so many chapters can cite one volume compactly. Entries also carry cross-references between co-authors and pointers to published reviews.

According to the journal's editorial, this issue marks a handover. Rainer Voigt had compiled the bibliography alone since 1998 and passed it to a team of younger scholars. The paper reports no experiments or metrics. Its value is as a finding aid for documentation, grammar and lexicography of under-resourced Afroasiatic languages.

Key points

• The sections cover Ethiopian Afrasian in general, Ethiopian Semitic, and further family sections. Two addenda follow: one for missed earlier items and one for reviews of earlier publications.
• About ten collected volumes are abbreviated with sigla, including:
  • ExplEthLing (2014, on complex predicates, finiteness and interrogativity);
  • Festschrift Voigt (2014);
  • TigreStud (2015);
  • SemLangCon (2015);
  • StratClauseLink2014;
  • ProcOsloAustin;
  • PICES 17;
  • the 1975 Paris Cushitic conference, published on DVD in 2015.
• Sample entries include:
  • Shimelis Mazengia's 2015 monograph on nominalization via verbal derivation in Amharic, Tigrinya and Oromo;
  • Bulakh & Kogan's new readings of the Arabic–Ethiopic glossary of al-Malik al-Afḍal;
  • a study of the linguistic landscape of two Ethiopian regional capitals.
• Where a title does not make it obvious, entries add a parenthetical tag naming the languages treated, e.g. Amharic, Tigrinya, Oromo.
• Under each book, reviews are listed with reviewer, journal, volume and pages.

Interesting ideas

• I1. Each multi-chapter volume gets a short siglum defined once at the top, and individual chapters then cite it with page ranges; this removes repetition and keeps provenance exact.
• I2. Entries carry explicit tags for which languages a work covers, even when the title is vague; this makes the index filterable by content, not just by wording.
• I3. Each work's reviews are attached to it, giving a second opinion next to the primary source.
• I4. A dedicated addenda section catches late or missed items without rewriting earlier instalments; the record is append-only but correctable.
• I5. Name cross-references ("see X") route every co-author to the primary entry, so one record exists while lookups work from any name.

For Mova

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

• I1 (world): Knowledge-graph sources (WordNet, FrameNet, Wikidata dumps) could be registered once under short IDs, with each compiled fact pointing to an ID plus a locator, so explanations stay compact but traceable.
• I2 (learn): Silver-data batches and treebanks could carry explicit tags for genre and UD dialect, so training and negative-control evaluation can select by tag rather than by file name.
• I3 (learn): Each silver parse could store which witness parsers or LLM teacher tiers agreed or dissented, so a reader sees the "reviews" beside every tree.
• I4 (world): Updates to a text's world state (new facts, corrected holdings) could be appended as dated corrections instead of overwritten, which keeps every earlier answer explainable.
• I5 (frames): Coreference output could keep one canonical entity record, with every alias or mention pointing to it rather than duplicating the entity.