📄 The verb ‘to have’ in Amharic

Summary

Łykowska tackles a gap she ran into while building sentence schemata for Amharic. Those schemata are centred on verbs and needed a verb meaning 'to have', yet grammars describe Amharic as having only a possessive construction built on allé 'is'. In that construction the verb agrees with the possessed item and a suffixed pronoun points back to the possessor. Working with two informants rather than a corpus, she examines this structure across morphology, syntax, semantics and pragmatics. She follows Baye Yimam's analysis, in which the possessor is topicalized out of a le- 'for' phrase.

Her central claim is that contemporary Amharic is reanalysing the construction as a genuine verb allew 'has' with an unusual conjugation, where the suffix works as a person marker. The paper also covers past and future forms, benefactive and malefactive marking, and alternative ways of expressing possession. Using a Polish dictionary's sense inventory as a grid, she catalogues the meanings of 'have', supported by many proverbs. The work matters for dictionaries and learners, which handle the topic poorly, and it documents a change still in progress, partly driven by English.

Key points

• Present-tense possession uses allé 'is' plus a suffixed pronoun, as in inê bët alleny 'I have a house'.
  • Where agreement varies (e.g. Aster has a sister), verb concord with the possessed noun is optional.
  • The two variants differ only in emphasis.
• Pluralizing the possessed noun is acceptable for alienable things like houses or suitcases. It is rejected for body parts, as in 'a table/cat has four legs'.
  • Bilingual speakers who use English heavily accept the plural everywhere.
• Past tense uses nebber 'was'. The future switches to other verbs: nore 'lived, stayed' or agenynye 'got, found'.
• The infixes -bb- (disadvantage) and -ll- (advantage) colour possession.
  • Illnesses, duties and debts take the disadvantage form almost by default.
  • Festive occasions such as a child's wedding take the advantage form.
• Meanings are mapped against the nine senses of 'have' in the PWN Polish dictionary (Szymczak, 1992). The analysis is illustrated with 52 numbered proverbs.

Interesting ideas

• I1. One underlying structure, a topicalized le- 'for' phrase, can surface in two ways, and the "object" suffix can be reread as a subject-agreement morpheme of a new verb. Which reading applies depends on the analysis, not only the string.
• I2. Borrowing a fine-grained sense inventory from a well-described language works as a probe grid when the target language's dictionaries are thin.
• I3. The alienable vs. inalienable distinction shows up formally, e.g. in whether the possessed noun can be plural.
• I4. Possession can carry a grammatical marker of benefit or harm to the possessor, with some possessions (debts, illness) defaulting to harm.
• I5. The future of 'have' is often expressed through acquisition verbs like get or buy.

For Mova

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

• I1 (world): Map "X has Y", "Y is X's" and "there is Y for X" onto the same holding relation, so paraphrases built on 'be' and 'have' yield one world fact.
• I2 (frames): Split English 'have' into explicit senses (ownership, composition/parts, measure, ailment, obligation) and route each to a different frame instead of treating all uses as possession.
• I3 (world): Store "a table has legs" as a part-whole fact in the ontology rather than as a transferable holding, so it is never counted in give/take arithmetic.
• I4 (frames): Attach an affect polarity to holding states (debts and illness negative, a wedding to host positive) to support implicature and emotion questions about characters.
• I5 (answer): Let readers infer a future holding from "will buy/get a cow" so word problems and story QA can answer "what will she have?"