📄 Syriac Influence on Islamic Iran

Summary

The paper asks which Syriac (Christian Aramaic) words survive in classical and modern Persian, either borrowed directly or passed down through Pahlavi after the Islamic period. Weryho leaves out three groups. The first is the Aramaic ideograms of Pahlavi, which were written as Semitic words but read aloud as Iranian ones. The second is Christian calendar terms, and the third is Syriac words that arrived only through Arabic, where loan and cognate are hard to tell apart. A word stays on his list only if it meets one of three tests: Arabic has no matching form, the Persian form is closer to Syriac than to its Arabic cognate, or the word can be traced through a Pahlavi stage. He builds on Paul Horn's earlier list and ends up with about thirty-five items. These are mostly nouns, and many are Christian terms. He then notes recurring sound changes. In his cultural reading, Arabic forms displaced Syriac ones after conversion to Islam, so only loans that looked clearly different from Arabic held on. The study shows that some words seen as "pure Persian" are actually early Semitic loans.

Key points

• The corpus has roughly 35 words, nearly all nouns. Only Tarsa, shayda, Gabr and haduri can also act as adjectives.
• Each inclusion test has an example:
  • saqi raghla comes from Syr. sagi regla, with no Arabic counterpart.
  • Yaldā is closer to Syr. Yalda than to Arab. Milad.
  • gonbad comes from Pahl. gumbat, which comes from Syr. qubbatha.
• Doublets of a Syriac form and an Arabic form live side by side: chalipa/salib, gonbad/qobbeh, gur/qabr. Folk etymology treats the Syriac member as native Persian.
• Five sound patterns recur:
  • The final -a is usually dropped, though chalipa, Masiha and Yalda keep it.
  • Double -bb- becomes -mb-, as in Shabba → Shanbeh.
  • Emphatic Sad becomes ch.
  • Qaf becomes k or g, as in kashish and gur.
  • A vowel is inserted to break up initial consonant clusters.
• The starting inventory is Paul Horn's list in Neupersische Schriftsprache (Grundriss der iranischen Philologie). Christian vocabulary is prominent in the result: Chalipa, kashish, Masiha, Shalikha, Yalda, Tarsa.

Interesting ideas

• I1. Pahlavi ideograms were written one way and spoken another (malka was read as shah), so the surface string and the spoken word must be kept as separate layers. — For Mova (tokens): Treat abbreviations, symbols and logogram-like tokens (e.g. "&", "Dr.", "%") with a distinct surface form and reading, so lemmas reflect what is actually said.
• I2. Inclusion rests on explicit, testable criteria: no rival form exists, the item is closer to one source than another, or a traceable chain links them. Hard-to-decide cases are excluded rather than guessed. — For Mova (learn): Silver-data admission rules could likewise list explicit acceptance conditions and drop contested items instead of forcing labels, keeping the training set clean.
• I3. Doublets show that one concept can carry two lexical forms from different strata (gonbad/qobbeh), and that speakers may misjudge which form is "native". — For Mova (world): The ontology could link register or origin variants of English words (e.g. Germanic/Latinate pairs) to one concept while recording their origin and register.
• I4. The author draws recurring sound correspondences from a small list but presents them as observations rather than firm rules because the evidence is thin. — For Mova (learn): Self-induced repair rules learned from few examples could be flagged as tentative until enough support accumulates, which keeps rule induction honest.