📄 Wolf Leslau: Introductory Grammar of Amharic

Summary

This is a short review, not an NLP study. It appeared in Aethiopica 5 (the PDF is dated 2002, though the library entry says 2013). Olga Kapeliuk assesses Wolf Leslau's Introductory Grammar of Amharic (2000), a compact teaching grammar derived from his much larger Reference Grammar of Amharic.

She first places the book among Leslau's earlier practical tools for learners. She then praises it for three things:

• it gives every example with phonological transcription and a word-by-word translation, which also serves comparative Semitists;
• it covers recent changes in the language;
• its definitions are clear for English speakers.

Most of the review adds corrections and refinements drawn from present-day Amharic usage:

• how the preposition bä combines with the infinitive;
• the phonetics of d;
• distributive uses of ǝyyä;
• the case for treating copula-based progressive constructions as genuine tenses.

Her main point is that the spoken and written language is moving faster than grammars record. Constructions once listed under conjunctions now act as tense forms, partly to resolve ambiguity between present and future.

Key points

• The book under review: Porta Linguarum Orientalium N.S. 21, Harrassowitz, 2000, xix + 232 pp. It is a by-product of Leslau's 1995 Reference Grammar (1044 pp.).
• Earlier context:
  • Leslau's Peace Corps manuals (1965, 1967).
  • His Concise Amharic Dictionary (1976), the first English-language tool to mark gemination fully.
  • Before it, gemination could only be checked in Gankin's Russian-Amharic dictionary (1965).
  • Thomas L. Kane's Amharic-English Dictionary (1990, 2351 pp.) was encouraged by Leslau.
• Newer phenomena the book covers:
  • an m-plural used for kinship nouns (zämädočč 'relatives');
  • feminine forms used as endearment toward males;
  • bare place nouns used without a preposition;
  • ǝndä/ǝyyä + perfect + näw treated as tenses.
• Reviewer's corrections:
  • bä + infinitive expresses cause only when a subject suffix is attached. Without one, it marks a concomitant action, like a gerund.
  • Amharic d is closer to alveolar than to dental and drifts toward r. Spellings of 'wall' vary between gǝdgǝdda and gǝrgǝdda, and Agaw languages show parallels.
• Tense ambiguity: the present/future form yǝnägrall is ambiguous. Speakers mark the actual present in two ways:
  • with a compound gerund, as in news openings saying literally "we have started the news";
  • with ǝyyä + perfect + näw, whose progressive meaning sometimes fades into a plain present or an idiom.

Interesting ideas

• I1. Each example carries three layers (script, phonological transcription, literal gloss), so learners and comparativists can each read the layer they need.
• I2. The pronoun chapter starts from the learner's language, taking English indefinite pronouns as anchors and listing target equivalents under each.
• I3. The meaning of a construction can depend on whether its subject is overtly marked: with a subject it gives a cause, without one an accompanying action.
• I4. Speakers escape tense ambiguity by using a perfect form to describe an action happening at the moment of speech ("dinner has arrived" meaning "is ready").
• I5. A progressive construction gradually loses its aspect and ends up in fixed polite answers whose literal reading is nonsense.

For Mova

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

• I1 (back): Translation through the meta-world could output a literal-gloss layer beside the fluent text, making each step of the mapping visible and checkable.
• I2 (world): The ontology could index cross-language lexical sets by English concept slots, so a quantifier like "anyone" links to a concept rather than to a single foreign word.
• I3 (frames): Role assignment for English adjuncts with -ing forms could condition on an explicit subject ("his not asking" vs. bare "asking") before choosing a cause or manner/accompaniment role.
• I4 (world): The event-ordering module should be able to read some English perfects as current states or performative present rather than strictly past events.
• I5 (frames): Speech-act and idiom detection should spot grammaticalized formulas and skip literal frame building, so the frames do not hold absurd literal events.