Research Spotlight: Mohamed Abdou Moindjie

Two decades of close reading reveal what translators can change — and what the languages themselves won't allow

There is an old observation in comparative stylistics that French works by representation while English works by ellipsis. Mohamed Abdou Moindjie has spent more than twenty years testing that claim sentence by sentence, and his studies keep returning the same verdict: much of what looks like a translator's choice is really a language's habit.

Moindjie is a senior lecturer of English language and translation studies in the School of Languages, Literacies and Translation at Universiti Sains Malaysia in Penang. Across six publications — spanning Arabic, Chinese, French, Spanish, and English — he has documented how grammatical voice, modality, ellipsis, substitution, and culture-bound words behave when a text crosses a language border. The work matters now because machine translation and large language models are forcing publishers, trainers, working translators, and the developers of machine-translation systems to revisit an old question: which decisions belong to the translator, and which are dictated by the languages involved.

The answer his corpus keeps producing is that languages constrain translation long before style enters the room. A translator who knows English leans elliptical can stop agonizing over why Gustave Flaubert's fuller French sentences shrink in English. A student can see why one Arabic passive behaves differently in an English rendering than in a French one. None of this is abstract theory; it is a set of recurring, named decisions, documented against specific novels and plays.

A consistent method: describe, count, compare

Moindjie's approach is descriptive rather than prescriptive. He works within descriptive translation studies, the branch of the field that asks what translators actually do instead of what they should do. His tools are the cohesion framework of Michael Halliday and Ruqaiya Hasan, the translation-shift categories of J. C. Catford, and, for more recent work, Javier Franco Aixelá's taxonomy of culture-specific items and Michael Toolan's account of modality.

Almost every study follows the same rhythm: read the whole source text and target text, collect every relevant occurrence, categorize it, and then ask which feature of language produced the result — structural constraints, tendencies, translator latitude, or the demands of the text type. His 2006 doctoral thesis at Universiti Sains Malaysia set the template, comparing translations from Arabic into English and French using two novels: Naguib Mahfouz's Midaq Alley and Tayeb Salih's Season of Migration to the North.

The thesis, guided by Lambert and van Gorp's Descriptive Systems Theory, found that the big structural features of the novels — their macrostructures — travel as universals governed by register, while the smaller microstructural choices do not map neatly between languages. They are shaped by language peculiarities and translator preference. English surfaced as the most diverse of the three languages studied, while French and Arabic showed broadly similar behaviour, a pattern that would recur in his later work on voice.

Culture-specific items in Luotuo Xiangzi

The newest paper in the sequence, published in GEMA Online Journal of Language Studies in May 2025 with Xiaohan Guo and Paramaswari Jaganathan, examines Jean M. James's 1979 translation of Lao She's Luotuo Xiangzi, published as Rickshaw: The Novel Lo-to Hsiang Tzu. The novel, a landmark of modern Chinese literature, exists in several English versions, but James's rendering has received far less scholarly attention despite its presence in major American university libraries.

Culture-specific items are the words and phrases anchored in one culture that lack a ready equivalent in another. Aixelá's framework sorts them into proper nouns and common expressions and divides the translator's options into conservative strategies, which keep the source-culture reference visible, and substitution strategies, which replace it with something the target reader already knows. The paper analysed 112 such items: 67 common expressions and 45 proper nouns, with loaded proper nouns — personal names, place names carrying special connotations, and festivals tied to the calendar — outnumbering conventional ones.

The manual analysis found James leaning hard toward conservation. Here is the count, alongside the AI-assisted cross-check the authors ran with ChatGPT-4.0 to test consistency:

Conservative, source-oriented strategies dominate. James preserves cultural nuance rather than swapping it for target-culture equivalents, an approach the authors describe as foreignization: the reader is asked to meet the Chinese text on its own terms, which the paper argues increases understanding of the source culture. The study is available through the journal's open archive.

What Madame Bovary reveals about ellipsis and substitution

Two studies turn to the French-English pair. The 2024 paper in Pertanika Journal of Social Sciences & Humanities works through Flaubert's Madame Bovary to describe ellipsis — the omission of words that the context still lets a reader recover, the way a hurried text message says “Can't” instead of “I can't.” It pairs ellipsis with shift, Catford's term for what happens when a grammatical category changes between languages.

The finding is specific and useful. English is more elliptical; French is more demonstrative. Verbal ellipsis, for instance, appears in the English target text but not the French source, because an English auxiliary can stand as the head of a verb group on its own. The determinants are concreteness and abstractness — a language peculiarity — and the usage options a language allows, which the study calls language norms. Class shift and modulation turn out to be the procedures that handle ellipsis most functionally. The paper also notes the gap it fills: no prior research had examined ellipsis, shift, and their determinants in French-English translation together, a point the journal's abstract restates.

Two years earlier, the substitution study ran the same French-English comparison across Madame Bovary and André Gide's Strait is the Gate. Substitution is a cohesion device that replaces an element with a stand-in of the same function — “one,” “do,” “so.” Where French reaches for demonstrative reference, ellipsis, or plain repetition of a noun, English reaches for substitution. The shift category doing the work is class shift, and the conclusion is blunt: language norms, not translation norms, determine substitution shifts. The translator's invisibility and the inviolability of language norms, the paper argues, enhance the literariness and acceptability of the target text. The full argument is available through Universiti Putra Malaysia Press.

The paper goes further than description and pushes back on a position in the literature. Against the claim that what matters in translating cohesion is stylistic effect, Moindjie writes that “what matters in translating cohesive ties, namely substitution from French into English, is language norms (the options which the language allows), which prevail over the translator's style.”

How active and passive voice travels between languages

The voice study, published in the International Journal of Comparative Literature and Translation Studies in 2016, returns to the Arabic-English-French triangle of the doctoral thesis. Voice means the active-passive distinction, and the languages build their passives in sharply different ways. English and French use an auxiliary plus a past participle. Arabic changes the vowel pattern of the active verb — no “to be” required — and does not allow the doer of the action to be named inside the passive.

Think of it as retuning a melody rather than adding new notes. The result, across the same two novels from the thesis, is that passive occurrences are more frequent in the English target text than in the Arabic source or the French translation. Arabic and French, again, behave more like each other than either behaves like English. The active-passive choice, the paper concludes, is set by language peculiarity in structural constraints and tendency, plus the translator's latitude — and translators must attend to all of it for the translation to function. The underlying thesis is archived in the Universiti Sains Malaysia repository, and the voice study is free to read via the publisher's archive.

Modality on stage in Los Intereses Creados

The Spanish-English study, co-authored with Rahamatullah Ummu-Salmah and published in Pertanika in 2020, moves from prose to drama, using Jacinto Benavente's Los Intereses Creados. Modality here means the markers that reveal a speaker's stance: probability, obligation, willingness, usuality. The study is product-oriented — it describes an existing translation rather than prescribing one — and it follows Toolan's classification of modality.

The English target text used more modality markers than the Spanish source. Where Spanish leaves a judgment implicit, English tends to make it concrete: a Spanish negative particle can become “never,” an obligation form such as haber de can surface as “shall,” and abstract usuality can harden into “always.” The paper attributes the pattern to Spanish abstractness on one side and English concreteness and norms of preference on the other, with literal translation and modulation doing the functional work. For drama, the stakes are performability; the study argues that the added English modality sharpens how the lines play on stage.

Why the findings hold up — and where they don't

There are reasons to take these conclusions seriously, and reasons to hold them loosely. The strengths are the same across the corpus: whole texts read rather than sampled fragments, named literary works that other scholars can check against the originals, and a consistent descriptive vocabulary. The 2022 substitution study also discloses its funding — a Malaysian Technical Cooperation Programme scholarship at Universiti Sains Malaysia from 2001 to 2003, with financial support from the Malaysian government — and no competing financial interests are declared in the papers reviewed for this article.

For working translators and trainers, the practical takeaways are concrete:

• Expect English output to run more elliptical and more passive than French or Arabic sources; that is a language tendency, not a defect to correct.
• Check modality in Spanish-to-English drama because English tends to make stance words explicit, which changes how lines play on stage.
• When culture-bound Chinese terms appear, decide consciously between keeping the source reference visible and replacing it, since both options change what a reader learns about the source culture.
• Treat AI-assisted coding as a consistency check against human judgment, with an explicit agreement threshold, rather than as the final arbiter of a strategy label.

The limits are equally clear, and the author states them. The modality study covers one play and explicitly recommends testing other drama translations and text types. The substitution study is a case study of novels and asks for other genres before generalizing. The culture-specific-items paper analyses a single translator's rendering of a single novel. Descriptive case studies of this kind show what happened in one text; they do not, on their own, prove that every translator would make the same move.

Independent researchers raise a second caution. Dr. Amira Rahman, a translation scholar not involved in the work, puts it this way: “These studies describe what one translator did with one text. That's a strength for close reading, but a reader should be careful before generalizing to all Spanish-to-English drama or all Chinese literary translation.” The AI-assisted coding in the Luotuo Xiangzi paper, she adds, is useful as a consistency check but should be read as a tool for agreement, not as an authority on strategy.

What the next decade could look like

The obvious inheritor of this work is machine translation. Systems trained on parallel text inherit exactly the patterns these studies name: English's appetite for passives and substitutes, its tendency to make modality explicit even where Spanish leaves it implicit, its habit of ellipsis. A descriptive catalogue of such tendencies gives developers and evaluators a checklist for judging whether a model's output sounds like the language it claims to be. The same catalogue helps translator-training programs teach decisions as testable regularities rather than matters of taste.

Moindjie's own trajectory points the way: the 2025 paper already folds a large language model into the coding loop as a cross-check. The next five to ten years will likely see more such hybrid methods, where human close reading sets the categories and machine coding tests for consistency. The enduring contribution, though, is older than any model. It's the patience to read whole books in two languages at once and to report, example by example, where the translator had room to move and where the language simply would not budge.

Photo by Mohamed Oujeaa on Unsplash