📄 Neologism construction in Amharic by compounding various parts of speech

Summary

Amharic, Ethiopia's working language, lacks native words for many concepts that arrive from Western culture. For decades these gaps were filled with European loanwords, which fit Amharic spelling and sound patterns poorly. Kozicki surveys how compounding is now used to coin replacement terms, often drawing on vocabulary from the extinct liturgical language Geʿez. Examples come from the daily Addis Zemen (2004–2010), online dictionaries (SelamSoft, Abyssinica, amharicdictionary.com), professional glossaries and Ethiopian studies of neologisms. The paper sorts the material by structure: native and Geʿez genitive ("construct state") compounds, adjective–noun pairs, compounds with deverbal nouns, verb–noun combinations, Geʿez prefixes, mixed-language hybrids, and calques of English phrases. The main finding is that the Geʿez pattern is gaining ground. Many coinages are metaphors rather than literal renderings of the Greco-Latin originals. Under English influence, a new form is also appearing in which the genitive marker is dropped, which turns compounds into plain juxtapositions. The work matters for language planning and terminology work in a morphologically rich Semitic language.

Key points

• Amharic has about 22 million native speakers in Ethiopia (2007 data) plus roughly 3 million abroad. Ethiopia's national Amharic language academy (abbreviated NAAL), established in 1972 and now called AA AELC, works on loanword replacement.
• Amharic and Geʿez compounds use mirror-image orders for the same meaning:
  • Amharic order is GEN + RE2 + RE1, as in yä-mängəst bet.
  • Geʿez order is RE1 + GEN + RE2, as in bet-ä mängəst.
  • Both mean "parliament."
• Many Geʿez coinages are metaphorical and are used in science, law and economics. Examples include awdä məhər for university ("place of science"), həggä mängəst for constitution and afä mähṣan for cervix. The English loan yunivärsiti is still more common than awdä məhər.
• Hybrids can mix three sources. yä-misayəl mablaya sərʿat ("anti-missile shield") combines English, Amharic and Geʿez words. Calques such as qulf gudday ("key issue") are frequent in newspapers.
• The paper classifies genitive compounds as mostly endocentric. Verb–noun and prefixed forms (e.g. qədmä- "pre-", dəhrä- "post-") are classified as exocentric.

Interesting ideas

• I1. Two surface orders with an optional marker express one identical meaning, so the variation is purely grammatical.
• I2. The paper separates a true compound (marker present) from a bare juxtaposition (marker dropped), even though the meaning is the same.
• I3. Many coined compounds are metaphors whose sense is not the sum of the parts. "Mouth of assembly" means spokesman, for example.
• I4. The paper distinguishes neologisms (not yet in dictionaries) from modern words (already registered).
• I5. Each example carries a literal part-by-part gloss next to its idiomatic meaning, such as "product of port" for import.

For Mova

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

• I1 (world): Map English "house of medicine" and "medicine house" style variants to the same entity/relation node, so the world model does not split one concept in two.
• I2 (tree): Use this as a guideline for consistent labeling of English compound vs nmod:poss/nmod in treebank conversion, keeping surface form and meaning as separate layers.
• I3 (world): Keep a lexicon of non-compositional English noun compounds, so frames do not build literal "mouth" entities for idiomatic multiword terms.
• I4 (tokens): Flag out-of-dictionary compounds as candidate neologisms and analyze their parts explicitly, instead of treating them as opaque unknown tokens.
• I5 (back): Generate such literal glosses as part of explanations, so a user can see how a multiword term was decomposed before its meaning was assigned.