📄 More on the Arabic Object Marker iyyā: Implications for the Origin of the Semitic Notae Accusativi

Summary

The paper addresses a long-standing puzzle in comparative Semitics: where the accusative markers (notae accusativi) come from. These include Phoenician/Punic ʾyt, Hebrew ʾt, Aramaic yt/wt and Arabic iyyā-. Rubin (2005) had suggested the question might never be settled. Wilmsen argues that Arabic, where the marker is richly attested in living speech, holds the key. The evidence is comparative and philological, not computational. It draws on dialect descriptions, East Semitic paradigms, Ugaritic, Sabaean, Andalusi Arabic, Mehri and the grammarian Sībawayhi.

Several Arabic vernaculars keep third-person pronouns and demonstratives containing a /t/, and some dialects still use free-standing object pronouns. Putting these together, the author traces the markers to Proto-Semitic oblique ("non-subject") pronouns with /t/. Arabic lost that /t/, while other branches kept it. He also argues that the marker's original role was emphatic reflexive deixis, roughly "itself". The result matters because it presents modern dialects as keepers of very old features, not merely later offshoots of Classical Arabic.

Key points

• The PDF header gives Folia Orientalia Vol. 50 as 2013, not 2026. Third-person forms with /t/ survive in:
  • rural Palestinian Arabic: hūte and hīte;
  • the Antakya varieties;
  • Egyptian varieties: huwwati and hiyyat;
  • Cairene demonstratives: dawwat, diyyat and dōlat.
• These forms parallel the East Semitic oblique pronouns (Old Assyrian šuāti/šiāti) and Ugaritic hwt/hyt.
• Free-standing or loosely cliticized object pronouns with ditransitives like aʿṭā "give" appear in several places:
  • Mardin, Mosul and Aleppo;
  • Mount Lebanon;
  • Tafilah and Mecca.
• A four-step path is proposed:
  • oblique pronoun with /t/;
  • independent pronoun with optional /t/;
  • loose clitic;
  • iyyā- marking.
• Sībawayhi's double object suffixes (aʿṭāhūhā) are read as loosely bound clitics. On this reading, the doubly affixed verb is a product of writing rather than of speech.
• Mehri marks both patient and recipient with t-, which hints that the strategy predates Arabic and Canaanite.
• The paper rejects two earlier proposals:
  • Testen's account, in which the ā of iyyā- is a feminine ending; the author treats the ā as phonologically conditioned.
  • Correll's derivation, which requires an unmotivated vowel shift.
• iyyā- can also emphasize subjects and prepositional complements, with an "itself/the very" sense. Examples include Ibn Rushd, Ibn Jinni, al-Ghazali, and journalistic "Lenin himself founded it".

Interesting ideas

• I1. Spoken variants form a graded chain: independent word, then loose clitic, then affix. Spelling can hide where speech actually puts word boundaries, so orthographic units should not be taken as linguistic ground truth. — For Mova (tokens): Contractions and clitics such as "gimme", "'s" and "'em" could be tokenized and analyzed by function rather than by spelling.
• I2. Nonstandard dialect forms are treated as primary evidence rather than as noise to normalize away. In this paper they are exactly what solves the problem. — For Mova (tree): When converting between UD dialects or parsing informal English, keep nonstandard variants as explicit features rather than collapsing them to the standard form.
• I3. One marker can carry both an object role and an emphatic "itself" role, and only the context separates them. — For Mova (frames): Separate intensive "himself" (as in "Lenin himself") from argument reflexives, so that coreference and role filling do not create a spurious extra participant.
• I4. In Mehri, patient and recipient share the same marker, so the marker alone cannot assign the semantic role. — For Mova (frames): Assign VerbNet roles from the verb class plus argument order, not from surface marking alone, especially with double-object "give" constructions.
• I5. The author rejects analyses that need unmotivated or highly hypothetical intermediate steps when a simpler derivation covers the same data. — For Mova (learn): When ranking self-induced repair rules, prefer the candidate that needs fewer ad hoc stipulations to explain the same errors.