Strategies of translating swear words into Arabic: a case study of a parallel corpus of Netflix English-Arabic movie subtitles

Audiovisual translation (AVT) is central to mediating the linguistic and non-linguistic elements of audiovisual products. Subtitling is a key AVT mode alongside dubbing. Corpus-assisted translation studies enable systematic analysis of translation choices using tools such as SketchEngine. This study investigates how professional subtitlers on Netflix render English swear words into Arabic in forty movies across drama, action, sci-fi, and biography. Swear words convey emotions and reactions in impolite ways and pose cultural and pragmatic challenges in translation, especially into conservative Arabic contexts. The research addresses two questions: (1) What are the most commonly adopted translation strategies by Netflix subtitlers in rendering English swear words into Arabic? (2) How were the strategies distributed across the four genres (drama, action, sci-fi, biography)? The authors hypothesize omission will be most used due to culture and censorship.

Subtitling renders verbal messages in sync with original audiovisual content and is done by both professional subtitlers and fansubbers, with differing norms and priorities. Defining swearing varies across linguistic, psychological, and social perspectives, but a key element is potential offensiveness and taboo. Swearing serves linguistic, social, and psychological functions and often references sexuality, genitalia, excretion, and religion. Cultural distance between English and Arabic, and the conservative nature of many Arab societies with religious and moral norms, influence subtitling choices and censorship. Modern Standard Arabic (MSA), commonly used in Arabic subtitles, is associated with politeness and euphemism, whereas vernaculars tend to carry more dysphemism; thus MSA often mitigates offensiveness. Prior studies on Arabic subtitling of taboo language commonly report frequent use of omission and euphemism/softening to accommodate cultural constraints, though some recent work shows a mix of strategies on streaming platforms. This study contributes by examining a larger parallel corpus across four genres, expecting offensiveness not to be maintained in MSA and providing guidance on strategy choices for translators.

Design: Mixed qualitative-quantitative corpus-assisted study. Data: An English-Arabic parallel corpus of 40 Netflix movies (drama, action, sci-fi, biography) selected from the English-Arabic Movie Subtitles Corpus (EAMSC). Selection criteria: genre (one of the four), IMDb rating ≥ 8, translated by Netflix, and high frequency of swear words. Corpus size: 699,229 tokens total; English sub-corpus 372,071; Arabic sub-corpus 327,158. Preparation: English transcripts and Arabic subtitles were exported from Netflix, aligned using Aegisub timecodes, then copy-pasted into Excel with two columns (ST, TT). Data were segmented (words/phrases/clauses/sentences) and aligned at segment level. Tools: SketchEngine (wordlist, n-gram, concordance, keywords, word sketch, thesaurus, parallel concordance). Node words searched via parallel concordance for ST–TT pairs. Target items: Frequent English swear words identified via frequency analysis following prior classifications (Jay, Beers Fägersten): fuck, shit, damn, ass, bitch, bastard, asshole, dick, cunt, pussy (including derivatives via wildcards, e.g., fucking, motherfucker). Translation strategy model (after Khoshsaligheh & Ameri, 2014): (1) Taboo-to-taboo (swear-to-swear equivalence, maintaining offensiveness), (2) Taboo-to-non-taboo (neutralization; here termed swear-to-non-swear), (3) Euphemisation/softening (toning down), (4) Omission (deletion). In analysis and reporting, three observed strategies were quantified: omission, softening, and swear-to-non-swear. Procedures: (1) Select movies per criteria; (2) Identify top swear words via SketchEngine wordlist; (3) Compare frequencies by genre; (4) Use parallel concordance to classify each instance’s rendering into omission, softening, or swear-to-non-swear; (5) Aggregate by genre and by swear word; (6) Provide examples and back-translations for each strategy.

- Corpus and items: 1,564 total occurrences across the ten targeted swear words. Top word: fuck (and derivatives) 877 hits; shit 291; damn 99; ass 93; bitch 75; asshole 57; bastard 34; dick 20; cunt 13; pussy 5. - Distribution by genre (share of all swear words): drama 61%, action 14%, sci-fi 14%, biography 11. - Strategy usage by genre (Table 10): - Drama (n=953): omission 66% (633), softening 20% (189), swear-to-non-swear 14% (131). - Action (n=224): omission 61% (136), softening 28% (63), swear-to-non-swear 11% (25). - Sci-fi (n=225): omission 40% (90), softening 39% (87), swear-to-non-swear 21% (48). - Biography (n=162): omission 52% (84), softening 32% (52), swear-to-non-swear 16% (26). - Overall strategy distribution (Fig. 5): omission 60% (943), softening 25% (390/391), swear-to-non-swear 15% (230). - Strategy by swear word (Table 11): - Fuck (877): omission 67% (589), softening 19% (170), swear-to-non 14% (118). - Shit (291): omission 60% (173), softening 27% (80), swear-to-non 13% (38). - Damn (99): omission 62% (61), softening 25% (25), swear-to-non 13% (13). - Ass (93): omission 77% (72), softening 15% (14), swear-to-non 8% (7). - Bitch (75): omission 38% (29), softening 43% (32), swear-to-non 19% (14). - Bastard (34): omission 12% (4), softening 56% (19), swear-to-non 32% (11). - Asshole (57): omission 16% (9), softening 51% (29), swear-to-non 33% (19). - Dick (20): omission 30% (6), softening 45% (9), swear-to-non 25% (5). - Cunt (13): omission 0%, softening 69% (9), swear-to-non 31% (4). - Pussy (5): omission 0%, softening 80% (4), swear-to-non 20% (1). - Genre-specific omission patterns (Table 4): e.g., in drama, ‘fuck’ omitted 73% (506/692), ‘shit’ 70% (94/134), ‘ass’ 75% (18/24); in action, ‘damn’ omitted 80% (24/30), ‘bitch’ 88% (22/25), ‘ass’ 89% (31/35); in sci-fi, omission lower overall (40%); in biography, omission 52% overall. - Examples illustrate: Omission deletes the swear component (e.g., deleting fucking, ass, damn) removing connotative force; Softening maps to milder epithets in MSA (e.g., وغد rogue, سافلة immoral woman, أحمق stupid, مغفل fool); Swear-to-non-swear neutralizes via general terms or context-appropriate replacements (e.g., ass → جسد body/سروال trousers; shit → رداءة bad thing; bastard → سيارة car in replacive context). - Lexical tendencies: Frequent Arabic equivalents include euphemistic curses (اللَعنة), verbs of caring/concern for give a damn/fuck (يأبه/يعنى/يهتم), doom/lose expressions, milder insults (أحمق، وغد، حقير، حثالة), and euphemistic body terms (أعضاء تناسلية، المنطقة الخاصة، جسد، مؤخرة).

The findings confirm the hypothesis: omission is the dominant strategy for rendering English swear words into Arabic Netflix subtitles across genres. This addresses RQ1 by identifying omission as most common, followed by softening, with swear-to-non-swear least frequent. For RQ2, distribution varies by genre: drama exhibits the highest overall swearing frequency and highest omission proportion; sci-fi shows relatively greater reliance on softening; action and biography fall in between. Cultural and technical constraints likely drive omissions and euphemisms: conservative audience norms, platform policies, and MSA’s inherently more polite register reduce offensiveness and motivate toning down. While these strategies help avoid offense, they diminish connotative and pragmatic functions of swearing (e.g., emotional intensity, character portrayal, interpersonal dynamics), potentially altering character representation and audience experience. Strategy choice also correlates with offensiveness: stronger swear words (e.g., cunt, motherfucker) tend toward softening or non-swear substitutions rather than literal taboo-to-taboo, and mild terms (damn, shit) are often omitted or softened. The reliance on MSA contributes to mitigation, as it favors euphemism over dysphemism. Overall, the results emphasize balancing cultural appropriateness with retention of pragmatic force in AVT, recommending careful selection of softening and neutralization to minimize loss compared to blanket omission.

The study integrates corpus linguistics and AVT to examine how Netflix subtitlers render English swear words into Arabic across 40 films. Ten frequent swear words were analyzed with three observed strategies: omission, softening (euphemization), and swear-to-non-swear (neutralization). Omission predominated overall and across genres, softening was second, and swear-to-non-swear least frequent. The results highlight that MSA subtitling often reduces offensiveness, impacting connotative meaning and character portrayal. The paper contributes large-scale, genre-sensitive evidence to AVT research and offers practical insights and lexical resources for subtitlers. Future research should: (1) examine fansubbing and satellite TV practices, (2) expand to other genres and AVT modes (dubbing, voiceover, audio description, free commentaries), (3) analyze denotative vs. connotative functions and swearing typologies in depth, and (4) explore strategies that better preserve pragmatic force while respecting cultural constraints.

- Data source limited to Netflix subtitles; findings may not generalize to fansubs or satellite TV channels with different norms or censorship. - Genres restricted to drama, action, sci-fi, and biography; other genres not analyzed. - Focus confined to strategy categorization and frequency; did not systematically analyze denotative vs. connotative functions or broader typologies of swearing beyond selected items. - Use of MSA-specific renderings may limit applicability to subtitling into Arabic vernaculars or dubbing contexts.