📄 On Diglossia in Ancient Hebrew and Its Graphic Representation

Summary

The article reviews Gary A. Rendsburg's 1990 claim that ancient Hebrew was diglossic, with one variety for literary and formal use and another for everyday speech. It looks closely at his terms, his method and especially the diagram he uses to show the two varieties over time. Rendsburg treats grammatical features common in Mishnaic Hebrew but rare in Biblical Hebrew as spoken forms that slipped into the classical texts. He checks them against later literary works and against spoken Arabic.

Ólafsson argues that "colloquial" and "literary" are better labels than "spoken" and "written." He then works through the time-versus-change chart, which Rendsburg modelled on Sethe's chart of Egyptian, and shows what it unintentionally implies: one uniform vernacular, a constant rate of change, and a possible merger of the two varieties. He offers two revised charts, one with many lines and one more abstract, each claiming only what the evidence supports. The wider point is that a visual model of language history can mislead unless its every feature is backed by data.

Key points

• Rendsburg's definition follows Ferguson's four classic cases: Arabic, Swiss German, Haitian Creole/French and Greek. In his model, Qumran Hebrew is the literary variety and Mishnaic Hebrew the colloquial one, and this split is projected back into the Biblical period.
• His method has two parts:
  • Forms that are frequent in Mishnaic Hebrew but sporadic in Biblical Hebrew count as colloquialisms only if they seldom appear in the Dead Sea Scrolls, the book of Daniel or Ben Sira.
  • Rare forms that break Biblical norms but match spoken (not classical) Arabic also count as colloquial.
• The evidence is grammatical, mostly morphology. Phonology is unavailable, and vocabulary is assumed to differ too little between the varieties to be useful.
• Ólafsson criticises the chart on five points:
  • It shows a single line for the colloquial variety.
  • It implies a constant rate of change, even though no "unit of change" exists.
  • The distance between the two varieties cannot be known.
  • The literary line drifts between 586 BCE and 70 CE, leaving unclear when stray colloquialisms become real change.
  • The two lines should never meet, since only consonants were written.
• The original chart is also out of scale. The 614-year span before 586 BCE is drawn longer than the 656-year span after it, and 70–200 CE is badly out of proportion.

Interesting ideas

• I1. A form common in a later colloquial corpus but rare in an older text is labelled colloquial only if contemporary literary works also lack it; otherwise it is treated as simply late. — For Mova (learn): Before turning an odd construction in silver data into a repair rule, check it against a control corpus to tell register effects apart from time drift or parser error.
• I2. The author asks how many deviations it takes before isolated lapses should count as a shift in the norm. — For Mova (tree): When converting between UD dialects or merging treebanks, set an explicit frequency threshold to separate occasional annotator slips from a real difference in annotation convention.
• I3. Instead of one smooth line for an era, each text could be plotted as its own point, placed by how close it is to the colloquial variety, since scribal skill varied from text to text. — For Mova (tokens): The tagger and morphology could give each document its own register score rather than assuming one standard, so books as different as Esther and Daniel get separate profiles.
• I4. A diagram should claim neither more nor less than the evidence. Unknown rates, starting points and distances should be left visibly tentative rather than drawn with false precision. — For Mova (world): Event-order and state timelines should keep unknown durations and boundaries as explicit uncertain intervals instead of filling in evenly spaced defaults.