📄 Aaron Michael Butts, ed., Semitic Languages in Contact

Summary

This is not an NLP paper. It is a short book review by Alessandro Bausi of the edited volume Semitic Languages in Contact (ed. Aaron Michael Butts, Brill, 2015). The volume collects case studies of linguistic interference across Semitic languages, from Sumerian–Akkadian contact in antiquity to Modern Hebrew's influence on urban Palestinian Arabic today. Bausi places the book in a research line that goes back to Weinreich's 1953 classic and to early work on Ethiopian Semitic (ES) by Moreno and Leslau. He points out that no systematic bibliography of ES interference exists.

He praises several chapters:

• Appleyard's introduction to Cushitic–ES contact.
• Crisostomo's reframing of Sumerian–Akkadian contact around literary prestige rather than ethnic conflict.
• Contini & Pagano's loanword study of Hatran Aramaic, which he says should prompt a fresh look at loanwords in ancient Ethiopic.

He also suggests a missed comparison with an Eritrean abecedary. Overall he judges the volume a rich source for general thinking about contact, and especially stimulating for Ethiopian studies. No datasets, models or metrics are involved.

Key points

• The reviewed book is Studies in Semitic Languages and Linguistics 82 (Leiden–Boston, 2015): xxvi + 427 pp., €150.
• The Hatran Aramaic chapter (pp. 126–157) examines loanwords in roughly 600 inscriptions dated c. 44 BCE–238 CE. The borrowings come from Akkadian, Iranian, Greek, Latin and Arabian sources.
• Al-Jallad & Al-Manaser describe a Thamudic B abecedary in South Semitic letter order. Bausi says it should also be compared with the Daḫan Amo abecedary from Eritrea (RIÉ no. 165).
• The ES shift from VSO to SOV word order, usually attributed to Cushitic influence, gets only about one page in Appleyard's chapter (pp. 16–32). Bausi considers this proportionate because the shift is so well known.
• Other chapters include:
  • Edzard on inner-Semitic loans and lexical doublets versus cognates.
  • Cohen on head-marking in Neo-Aramaic genitives compared with the Kurdish ezafe.
  • Zewi & Oren on contact-driven syntactic change in verbal complementation.
  • Horesh on Hebrew-induced change in urban Palestinian Arabic.

Interesting ideas

• I1. Edzard separates words borrowed between related languages from true inherited cognates, since surface similarity can hide either origin; keeping provenance explicit avoids false equivalences.
• I2. Cohen compares a head-marked genitive in Neo-Aramaic with the Kurdish ezafe, treating two different surface strategies as one functional construction shaped by contact.
• I3. Contini & Pagano inventory loanwords over a bounded, dated corpus and Bausi stresses interpreting the counts with due proportion, which keeps a small corpus from being overgeneralised.
• I4. Bausi faults one chapter for not checking its find against every known parallel witness (the Eritrean abecedary), showing that conclusions gain strength from exhaustive cross-comparison.
• I5. Bausi agrees that obvious, well-known effects deserve little space so that subtler phenomena can be studied in depth.

For Mova

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

• I1 (world): When linking lexemes to WordNet/Wikidata senses, store whether a near-identical form is a doublet or a shared root, so that look-alike words are not merged into one concept.
• I2 (tree): In UD dialect conversion, map different possessive marking patterns onto a single relation such as nmod:poss, guided by function rather than surface form.
• I3 (learn): Report silver-data and error statistics per bounded, documented sub-corpus and normalise them before drawing conclusions about the whole system.
• I4 (learn): When building silver trees by witness-parser agreement, query all available witnesses rather than a convenient subset, and flag analyses that were never cross-checked.
• I5 (learn): Weight evaluation suites toward rare, subtle construction types instead of repeatedly scoring the frequent easy cases.