📄 Teaching Arabic Grammar to Non-Native Speaker, between Curriculum and Acting on it (Silsiletu Abjad Series)

Summary

This is a pedagogy study, not an NLP paper. Drawing on three years of teaching the preparatory year in the Arabic department at Mardin Artuklu University, the author examines how the Silsiletu Abjad textbook series handles grammar (nahw) for Turkish learners of Arabic. He first describes the series' design and analyses its grammar content along three dimensions: how topics are distributed, their scholarly level, and how well each rule fits its unit's theme. He then documents the instructor's departures from the official syllabus. Some are partial changes inside a lesson. Others are total, where whole lessons are dropped. He groups the reasons for these departures.

The main finding is that the series aims at functional grammar tied to learners' life situations but only partly delivers. It breaks its own easy-to-hard progression and overloads beginners with exceptions and fine-grained rules, so adapting it in class becomes unavoidable. The author proposes reforms that matter for anyone designing a staged grammar curriculum.

Key points

• Silsiletu Abjad has four levels. Each level has a student book and a separate exercise book with twelve thematic units. A unit moves through listening, speaking, reading, writing/expression and grammar, sometimes ending with dictation or morphology. Extra "Abjad keys" lessons teach the alphabet.
• Grammar follows unit themes. In level 1, demonstratives appear in the greetings unit, question words in the family unit, and definiteness in the study unit. Pronouns are tied to travel and food, adjectives to professions, iḍāfa to hobbies, masculine/feminine to the human body, and negation to sport. Level 2 widens the scope to "the world around me."
• Each lesson opens with a theme-related text and warm-up questions, then gives a condensed rule statement and a worked example.
• Partial adaptation: the feminine is taught as only two contrasting types (real vs. non-real) instead of four. Total adaptation: the instructor removed the lessons on exception with illā and with khalā/ʿadā/ḥāshā/siwā, and on the case analysis of interrogatives.
• Proposals:
  • reframe grammar chapters around the formal and semantic relations between words in a sentence;
  • add sentence "styles" (negation, affirmation, command, prohibition, emphasis, condition, exclamation, praise/blame);
  • redistribute material by level;
  • integrate syntax with morphology;
  • favour common patterns over exceptions.

Interesting ideas

• I1. Introduce a construction only when the material actually needs it: the series gives just two demonstratives early and postpones the rest until later texts call for them.
• I2. At an early stage, collapse a fine-grained category into a single binary contrast, as the author does with the four feminine types, and add detail later.
• I3. Explain grammar through the formal and semantic relations among words in the sentence rather than through traditional chapter labels.
• I4. Organise sentence knowledge by communicative style (negation, command, prohibition, condition, exclamation, praise/blame) rather than by form alone.
• I5. Classify deviations from an official specification as partial (within a unit) or total (dropping a unit), and record the cause of each.

For Mova

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

• I1 (learn): Order training and evaluation items so each construction first appears with the texts that require it, and check whether this staged schedule beats random ordering.
• I2 (tokens): Start the tagger with coarse feature values and refine them later, so beginner-level errors stay interpretable and do not cascade.
• I3 (tree): This supports presenting each UD relation together with the meaning link it carries, which makes parser explanations readable for learners.
• I4 (frames): Use these styles as a checklist of polarity, mood and speech-act features that prag must detect and expose in its frames.
• I5 (tree): Log every UD dialect conversion choice in this partial/total scheme with its reason, so departures from the standard stay auditable.