📄 Elias, David L., The Tigre Language of Ginda‘, Eritrea, Short Grammar and Texts, 2014, Brill, Studies in Semitic Languages and Linguistics, 75, 292 p. ISBN 978-90-04-27119-7

Summary

This is a review by Didier Morin of David L. Elias's 2014 Brill grammar of Tigre as spoken in Ginda‘, a town on the Asmara–Massawa road in Eritrea. The book covers phonology, morphology, syntax, lexicon and language contact, and adds short texts and a glossary. The review's main question is whether "Tigre of Ginda‘" is a separate dialect, as Elias tentatively suggests. Morin goes through the grammar section by section and checks it against Raz's earlier description of the Mansa‘ dialect. He finds the consonants, pronouns, verb system and SOV syntax nearly identical to Mansa‘. He proposes a two-level classification: geographically defined dialects, and within them mutually intelligible "parlers." Under this scheme, Ginda‘ is a parler of the highland Sänḥit dialect. Morin also says the book overlooks Cushitic contact (Saho, Afar, Beja), which explains some forms Elias calls unique. Even so, he credits the book with substantially enlarging the published data on highland Tigre.

Key points

• The book is Studies in Semitic Languages and Linguistics vol. 75, 292 pages, ISBN 978-90-04-27119-7. Its reference points are Littmann, Leslau and Raz.
• Morin compares two views of what a "dialect" is:
  • the Anglo-American view counts any socially or geographically defined variety;
  • the French view is grounded in diachrony.
• Morin's evidence of Cushitic contact:
  • The adverb marrā "very", claimed as unattested in Tigre, is a Northern Saho loan.
  • The name Ginda‘ itself is the Northern Saho word for Calotropis procera.
• Features Ginda‘ shares with Mansa‘:
  • the ejectives /ṭ/, /ṣ/, /č’/;
  • free realization of /‘/ as a glottal stop;
  • four verb types (A–D) plus prefixal passive, causative and factitive derivation;
  • the genitive particle nay;
  • a future built from əgəl + jussive + copula.
• Elias himself admits his data may be closer to one speaker's idiolect, with heavier Arabic borrowing (e.g., lākin, tab‘an).

Interesting ideas

• I1. Morin separates dialects from "parlers" (variants that do not block mutual understanding), which gives a principled threshold before declaring a new variety. — For Mova (tree): When converting between UD dialects, treat minor annotation differences as parlers of one scheme with mapping rules, and reserve separate converters for structural divergences.
• I2. Before accepting a form as unique, he checks neighbouring contact languages and often finds an ordinary loan. — For Mova (tokens): Unknown or "novel" English words could first be checked against borrowing sources and known variant spellings before being treated as new lemmas.
• I3. The whole review is a feature-by-feature diff against one reference description, which yields an explicit, checkable verdict on divergence. — For Mova (learn): Evaluate a new treebank or parser output against a reference resource one phenomenon at a time, rather than with a single aggregate score.
• I4. The concession that one informant's speech may be an idiolect warns that single-source data can be mistaken for a whole variety. — For Mova (learn): Record provenance in silver data and flag patterns supported by only one witness parser or one source text.
• I5. Futurity is expressed by a multiword construction (preposition + jussive + copula) rather than by a dedicated inflection. — For Mova (frames): Tense and modality should be read off whole constructions in the UD tree (like "is going to" or "will have"), not just off verb morphology.