📄 Arabic grammar as illustrated in elementary Modern Standard Arabic: A descriptive and critical study

Summary

This is a critical review of the textbook Elementary Modern Standard Arabic (E.M.S.A.) by Peter Abboud and co-authors. It asks whether a grammarian can describe a foreign language using the categories and vocabulary of his own grammatical tradition. Abukhudairi focuses on Part I of the book, which covers syntax. He examines two things: how the textbook orders and groups grammar topics compared with classical Arab grammarians, and how it applies English (Latin-derived) grammatical terms to Arabic. He argues that the book's reordering breaks up natural units of Arabic grammar. He also argues that its terminology hides real Arabic distinctions and leads the authors into outright misclassifications, especially of particles, adverbs and nouns. A secondary criticism is that the reading texts are unvoweled, which makes words and even sentence roles ambiguous for beginners. The paper concludes that borrowing terms from another tradition is safe only where the categories genuinely overlap. Where they conflict, borrowed terms confuse learners and authors alike.

Key points

• The textbook. E.M.S.A. has five authors. Three native Arabic speakers wrote the texts and drills, and two non-natives wrote the grammar notes. It runs to 993 pages in two parts: Part I has 30 lessons (syntax) and Part II has 15 (morphology).
• Order of topics. The book opens with the equational (nominal) sentence and reaches the verbal sentence only in lesson 6. The passive voice waits until lesson 30.
• Scattered word groups. Members of traditional groups, such as kāna and its sisters or inna and its sisters, are spread across distant lessons.
• Changed traditional orderings.
  • Pronouns start from the third person, whereas Arab tradition starts from the first.
  • Cases run nominative, then jarr, then accusative, instead of the traditional rafʿ, naṣb, jarr, which is ordered by the "strength" of each case.
• Terminology mismatches.
  • "Subject" merges fāʿil (agent) with mubtadaʾ (the noun that opens a nominal sentence).
  • "Verbal sentence" is defined as any sentence containing a verb, instead of one that starts with a verb.
  • Calling jarr "genitive" and māḍī/muḍāriʿ "perfect/imperfect" distorts their meaning.
  • baʿd and qabl are labelled prepositions, though they are time adverbs.
  • The two objects of jaʿala are called direct and indirect objects.

Interesting ideas

• I1. In the unvoweled sentence shakara Ahmad Ashraf, only the case vowels decide who thanked whom; the word order is identical in both readings.
• I2. Labelling both the agent and the sentence-initial noun as "subject" erases a real distinction between doer and topic.
• I3. Sentence types should be defined by where an element sits (the sentence starts with a verb), not merely by whether it occurs anywhere.
• I4. Items that behave alike syntactically, like inna and its sisters, are better described as one class than scattered piece by piece.
• I5. A single unvoweled written form can stand for many readings: active, passive, causative, noun, or different persons.

For Mova

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

• I1 (frames): Semantic role assignment should rely on case and agreement morphology when it is present, with linear position only as a fallback, and should record which cue settled the role.
• I2 (tree): When converting between UD dialects or treebanks, keep the source-native category as a feature next to the borrowed UD label, so the merges that conversion introduces stay visible and reversible.
• I3 (tree): Construction detectors, such as topicalization versus canonical clauses, can be written as positional constraints over the dependency tree rather than as checks for the presence of a word class.
• I4 (world): The compiled lexicon of verb and particle classes should group words by shared syntactic behaviour, so that one rule covers the whole class and explanations cite the class.
• I5 (tokens): The morphology layer could keep a set of candidate analyses for ambiguous surface forms and let the parser or frames choose among them, instead of committing early.