The Rendering of Multilingual Occurrences in Netflix’s Italian Dub Streams: Evolving Trends and Norms on Streaming Platforms

Abstract

Given the vast scholarly attention paid to multilingualism on traditional media over the years, it seems timely to focus on streaming platforms. This paper sets out to identify potential norms for the rendering of multilingual occurrences in the localised content of Netflix series. It also seeks to explore whether streaming translation practices related to multilingualism differ from the consolidated norms and practices for TV and cinema content. The chosen data sample consists of the Italian dub streams of five TV Netflix-produced shows featuring multilingualism as a main characteristic. The strategies and techniques adopted in each series are singled out, quantified, and labelled according to a combination of taxonomies. These include dubbing, revoicing, subtitling, part-subtitling, diegetic interpreting, unchanged speech transfer, and no translation. A wider analysis is also carried out across all the data sample to draw patterns on a macro level. The findings reveal a strong tendency to mark and preserve multilingualism, in line with Netflix’s own policies and dubbing specifications. Transfer unchanged combined with subtitles emerges as the most recurrent strategy, while the dub-over strategy accounts for 13% of the multilingual occurrences in the data sample. Extensive neutralisation is therefore not encountered. That said, a certain degree of overlap between multilingual translation norms on Netflix and conventional Italian dubbing practices (which tend to neutralise) can still be observed.

1. Introduction

Diversity and inclusion are among Netflix’s main policies and priorities, so much so that a dedicated section on the company’s website elucidates their current and future inclusive measures and action, and this is accompanied by an updated Inclusion Report (Netflix 2023). The platform’s approach seems to put particular emphasis on racial diversity and non-white identities, gender and disability representation, and language and cultural diversity, the latter reflected in multilingual productions. Jenner (2018, p. 231) notes that many of the company’s US in-house productions are multilingual and feature, at the very least, one language other than English. She observes that the scenes in question are generally subtitled and are usually relevant to the plot, meaning they are not merely decorative but play a part in driving the narrative forward. Moreover, multilingualism in the original US productions is also a way for Netflix to attract transnational viewership (Jenner 2018, p. 231). There are many recent examples to support this. Among the TV shows that are analysed in this paper, Emily in Paris (Star 2020) is a clear example of how French has been incorporated into the original production intended for an English-speaking primary audience. This implies forced narratives in English for the French utterances, and careful planning in the screenwriting stage to ensure a balance between linguistic authenticity (e.g., French characters speaking French) and linguistic access (e.g., French characters alternating continuously between English and French to reduce the need for partial subtitles to a minimum). Forced narratives refer to subtitles that cannot be disabled by the viewers since they are embedded in the original production meant for its primary audience. This multilingualism and localisation ‘blend’ intended for the primary audience, is clearly not an afterthought and is perhaps in line with accessible filmmaking precepts (Romero-Fresco 2019). However, as a transnational broadcaster, Netflix is also concerned with the localised streams. The subtitles generally mirror the strategies adopted in the original production; consequently, they are provided in the new target language when and where they feature in the original (Netflix 2022a). The dub streams are perhaps trickier to handle especially when the same character speaks multiple languages, since one voice talent (not necessarily a polyglot) will be assigned to that character (Spiteri Miggiani 2019). Netflix provides clear instructions on how to tackle the presence of multilingualism. The Dubbing Audio Style Guide for lip sync dubbing recommends the following:

Foreign dialogue should be translated in the forced subtitle stream only if it is meant to be understood. If it is unclear whether or not the foreign dialogue is meant to be understood, seek advice from Netflix or the production. Do not dub over foreign dialogue.

(Netflix 2022a)

The guideline is quite concise and does not leave room for ambiguity, that is, foreign languages must be preserved in all dubbed versions. And the utterances should be translated by means of subtitles only if they are meant to be understood. Therefore, once again the creative intent in the original production would need to be mirrored in the localised streams. As for accents, Netflix recommends the creative choice “to be determined by production with consistency maintained by Netflix” (Netflix 2022a). This means that the decision-making process and responsibility are shared with the translator. At first glance, dealing with multilingualism in dubbing may seem quite straightforward. The ideal scenario seems to be one in which the foreign language voice tracks are not erased in the dubbing process. That said, the extent to which these recommendations are feasible and easily applicable will be discussed in this paper. Practical and technical challenges, as well as the impact on viewer experience need to be considered on a case-by-case basis. An interesting case in point is the Netflix show 1899 (Baran and Jantje 2022) that features extensive conversations in multiple languages including English, Swedish, Polish, French, Cantonese, Castilian Spanish, German, and Portuguese. Despite English being highlighted as the original audio, an English language dub stream is also provided. The English-speaking audience can therefore choose between the original English version featuring foreign utterances accompanied by forced narratives in English, or the English dubbed version with English all throughout, hence dubbing all foreign utterances into English. The same applies for all other target language dub streams where all the foreign utterances are replaced by the target language dub, in line with another Netflix guideline that however is specific to voice-over and other types of products. Voice-over is recommended when foreign dialogue accounts for more than 40% of the dialogue “as too many forced narratives cause a disruptive experience” (Netflix 2022b). Interestingly enough, in the example mentioned, this completely alters the dynamics between the characters who in the original version cannot communicate with each other. In the dubbed versions, they can understand each other as well as the secrets they reveal, as opposed to the original creative intent.

1.1. Scholarly Research

The presence of multilingualism on screen is not unique to Netflix; in fact, most of the research to date on multilingualism in audiovisual translation (AVT) does not focus on streaming platforms. In the last 15 years, research on the subject has flourished (Minutella 2020, p. 49) and researchers have acknowledged the challenge that linguistic diversity on screen may pose to audiovisual (AV) translators (Díaz Cintas 2011; O’Sullivan 2011; De Bonis 2015; Monti 2016; Beseghi 2017; Parini 2019).

In their analysis of language variation in AV content, Corrius and Zabalbeascoa (2011) introduce the notion of L3, a third language consisting not necessarily of an official language but “a sign of meaningful language variation” (L3 represents a linguistic variety that differs from both the language of the source text (ST), that is L1, and that of the target text (TT), that is, L2. The use of multilingualism is multifaceted: the L3 may be employed for characterisation purposes, it may cover a realistic function (Bleichenbacher 2007; Díaz Cintas 2011; De Bonis 2015), highlight conflict (De Bonis 2015), trigger comedic confusion (Bleichenbacher 2008; Chiaro and De Bonis 2020), or act as “post-carding” (Wahl 2005) to indicate the setting or nationality. Whatever the purpose of the L3, language diversity and variation can be a thorny issue in translation; however, they pose an even greater challenge in dubbing (Parini 2015, p. 29). In particular, switching languages across conversational turns and/or between sentences within the same turn (turn-specific and intersentential code-switching), or within the boundaries of the same clause (code-mixing) (Myers-Scotton 1997; Monti 2014) makes the rendition or preservation of multilingualism more complicated. Code-switching is a recurring discourse strategy for the representation of multilingual communities, but in dubbing it is considered problematic due to a technical constraint: the issue of harmonising the voices of the original actor and dubbing actor. In code-switching instances, unless the same original actors dub themselves, the switch between the voices retained from the original foreign language audio track and the dubbed voices would most likely be noticeable, risking the disruption of the audience’s suspension of disbelief (Romero-Fresco 2006).

Various works focus on the strategies available to deal with the wide range of secondary languages in the translation of AV productions (Diadori 2003; Heiss 2004, 2014; Baldo 2009; Ferrari 2010; Díaz Cintas 2011; O’Sullivan 2011; Minutella 2012; De Higes-Andino 2014; De Bonis 2015; Beseghi 2017). In Italy, the preferred procedure for the translation of multilingualism now seems to be a combination of dubbing, subtitling, and dubbing with a foreign accent (Parini 2015); however, it is not yet an established convention. De Bonis (2014, p. 256) states that the translation of all the various languages into Italian is very much still a mainstream strategy found in Italian dubbing. He also notes that the L3, or “secondary language” (2014, p. 259) is usually retained when it is in the background or if the conflict it expresses is crucial to the development of the story. The L3 may be carried over from the original soundtrack or revoiced by the dubbing actors. On the other hand, the neutralisation of all foreign varieties usually happens when the L3 holds a characterisation function, and particularly when it occurs with code-switching. Generally, when the code-switching between multilingual characters is constant, Italian dubbing opts for a quantitative reduction, that is, a combination of transfer unchanged with part-subtitling and dubbing into Italian of the L3 instances (ibid., p. 260).

Chiaro (2008, p. 23) calls attention to the fact that “a common strategy to deal with variation is simply not to deal with it and homogenize it into the standard, mainstream variety of the target language”. This is particularly the case with specific varieties native to the English language, such as British, American, Australian, and so on, that are neutralised in dubbing. When non-native accents are involved, Minutella (2021) notes that the foreign characterisation through foreign-accented English is sometimes reproduced by means of a subtle flavour of the foreign accent in the dubbed speech, along with the addition of some foreign and/or non-standard words.

This homogenising norm, typical of Italian dubbing in general, has also been traditionally observed on television (Ferrari 2010; Heiss 2014; Parini 2019; Manfredi 2021). That said, even television seems to be slowly embracing the preservation of multilingualism on screen, in alignment with the most recent trend surfacing in Italian dubbing across media (Parini 2019; Manfredi 2021). On the one hand, Dore (2019) remarks how the dubbing of Modern Family (Lloyd and Levitan 2009) moves away from the conventional flattening approach and successfully transfers Gloria’s language diversity (expressed through a heavy Latin American accent), to the point of increasing the degree of code-mixing. On the other hand, Parini (2019) observes that the approach to the translation of the series The Young Pope (Sorrentino 2016), launched seven years after Modern Family, opts for homogenisation, in line with traditional strategies.

1.2. Research Aims

As can be seen, the phenomenon of multilingualism has been extensively explored in film, and a certain degree of attention has been paid to multilingual TV series; while, to-date, not much research has focused on over-the-top (OTT) platforms and the translation trends that are being established on this booming medium. The aim of this paper is to identify potential norms for the rendering of multilingual occurrences in the Italian dubbed versions produced for streaming platforms, more specifically, Netflix. The various modalities and strategies in the Italian dub streams are investigated, with the aim of drawing common patterns that may be attributable to Netflix’s inclusion and diversity policy (Netflix 2022a) and the platform’s relative dubbing guidelines. The extent to which these guidelines are followed is therefore also observed. A further, secondary aim is to explore whether the practices related to multilingualism, as identified on Netflix, differ from the traditional, consolidated norms and practices of Italian dubbing for TV and cinema content.

2. Methodology

The pilot episodes from five streaming TV series, specifically Netflix Originals featuring multilingualism were selected as the data sample. The Italian dub streams were analysed by drawing qualitative and quantitative data, the latter referring to the recurrence of specific translation modalities and strategies. The use of the terms: ‘method’, ‘strategies’ and ‘techniques’ can present an issue due to the terminological incoherence that has characterised the field (Zabalbeascoa 2000). To this end, for the sake of clarification, the term ‘strategies’ in this article is used in its wider sense to refer to the adopted solutions and decision-making process (e.g., deciding to mark the foreign language, accents, or code-switching) while reference to the specific modalities is made to identify the actual techniques used to implement the solutions.

A combination of different taxonomies was applied to categorise the various translation modes and macro-strategies adopted to convey the multilingual instances. The ST and TT languages were labelled according to Corrius and Zabalbeascoa’s (2011) L3 model, fine-tuned by Voellmer and Zabalbeascoa (2014, p. 36), and summarised as follows:

(a)

Neutralisation:

• L3^ST → L2, language variation is dubbed over, thus becoming undetectable;
• L3^ST → Ø, language variation is omitted;
• L3^ST = L2, zero translation: L3 is preserved but since it coincides with L2 it is undetectable in the dubbed version.

(b)

Adaptation: L3^ST ≠ L3^TT, a language other than the L3 of the ST becomes the L3 of the TT; this solution can also be adopted in the case of L3^ST coinciding with L2.

(c)

Transfer unchanged: L3^ST = L3^TT

• Verbatim transcription, the L3 from the ST is retained unaltered in the TT;
• Conveyed accents, the same recognisable accent (stronger or weaker) from the ST is retained in the TT.

Since code-switching and code-mixing feature heavily in some of the shows selected for the data sample, the model for this study was further adapted to include the following combination:

• iii.

  (Transfer unchanged) L1+L3^ST → L3^L1+L3^TT, the phenomenon is preserved in its entirety and the L1 can be considered a further L3 of the dubbed version.

Other combinations such as L1+L3^ST→ L2+L3^TT and L1+L3^ST → L2 can be considered as already accounted for in the transfer unchanged (c) and neutralisation (a) categories.

This L3 model was also combined with Chaume’s (2012, p. 132) classification of techniques, in particular:

(a)

part-subtitling L3 into standard L2;

(b)

dubbing L3 into standard L2;

(c)

(re)dubbing L3 into L3, to avoid issues of voice harmonisation in the dubbed version;

(d)

no-translation.

Furthermore, the translation macro strategies identified by De Bonis (2014) were referred to when assessing the overall outcome and impact that the various micro strategies have on the final product. These include neutralisation, preservation, and quantitative reduction (that will be explained further on).

To avoid confusion with the terms ‘redub’ and ‘redubbing’ that usually refer to “a second or subsequent dubbed version of the same audiovisual product in the same target language” (Zanotti 2015, p. 110), this paper refers to the practice of dubbing actors lending their voices to revoice L3 instances as ‘revoicing’ instead of ‘(re)dubbing’.

2.1. Data Sample

The series chosen for the research study are Netflix Originals. This is to ensure that the shows were not purchased by the platform with a pre-existing dubbed translation carried out by other localisation companies, possibly for TV. This way the chosen data sample is presumably dubbed according to Netflix’s own specifications. The pilot episodes of each of the following series were chosen:

• Gentefied (Chávez and Lemus 2020, dramedy)
• Emily in Paris (Star 2020, dramedy)
• Unorthodox (Winger and Karolinski 2020, drama)
• Queen Sono (Lediga 2020, crime-drama)
• Messiah (Petroni 2020, thriller)

The selection includes both European and non-European L3s, to account for the fact that the L3 familiarity to the L2, in this case Italian, might influence the choice of translation strategy. English was chosen as the L1 for this study because, traditionally most of the TV series and films broadcast in Italy are imported from the US (Antonini 2009; Grece and Jiménez Pumares 2020) and are therefore primarily in English, meaning that the dubbing industry is shaped primarily on dubbing from English.

Section 3 presents the findings and analysis of the micro and macro translation strategies that were identified. Common patterns across the whole data sample were drawn and finally, similarities and/or differences to the Italian dubbing TV norms were observed. The latter discussion is based on the scholarly literature discussed earlier.

The following subsections provide a brief synopsis of the Netflix shows chosen as the object of analysis.

2.1.1. Gentefied

Gentefied (Chávez and Lemus 2020) is a US American dramedy set in today’s Boyle Heights, a primarily Latino neighbourhood in East Los Angeles, known as a stronghold of Chicano culture (Ohanesian 2021). The series revolves around the Morales, a Mexican American family. The show quite naturally alternates between English, Spanish and Spanglish, and realistically portrays a multicultural Spanglish household where multilingualism permeates daily interactions between family members.

2.1.2. Unorthodox

Unorthodox (Winger and Karolinski 2020) is a German-American drama miniseries that centres around Esther “Esty” Shapiro, a 19-year-old woman born and raised in Williamsburg, New York City, in the Satmar sect, an ultra-Orthodox Hasidic Jewish community. Netflix indicates English as the original language of the series, though the members of the Hasidic community depicted as Americans residing in New York rarely resort to English, their native tongue being Yiddish. As Etsy flees to Berlin and makes new acquaintances, Unorthodox frequently shifts from Yiddish to English to German, every shift serving to underline the cultural divides she is seeking to navigate.

2.1.3. Emily in Paris

As the title suggests, the US American comedy-drama series Emily in Paris (Star 2020) is set in Paris; despite this, English is the L1. The main character, Emily, only speaks English, hence most conversations are carried out in English. French characters speak English mainly when they would not be understood because their interlocutors are not Francophones, though there are a few exceptions.

2.1.4. Queen Sono

Queen Sono (Lediga 2020) is a South African crime drama series. True to the multilingual nature of South Africa and the array of international settings, the first episode of Queen Sono features dialogues in a variety of African languages, in addition to the always predominant English. These are Zulu, Swahili, Xitsonga and Afrikaans. French constitutes the only non-African L3. Although Zulu is the most spoken home language of the 11 official languages of South Africa, and Afrikaans is the third, English is the language preferred by both government and media, as well as being understood in most urban areas (Tongues under threat 2011).

2.1.5. Messiah

Messiah (Petroni 2020) is a US American thriller that explores the world’s reaction in modern times to a man who appears in the Middle East claiming to be the messenger of God, gaining followers as he goes. The variety of settings is mirrored in the languages featured in the series. English is obviously the main language in the scenes set in the US, where the main variety consists of American English; the only exception is Undersecretary Danny Kirmani who speaks with a discernible Pakistani accent. However, in the series, English appears also when the setting is moved to Israel, along with Hebrew, pointing to the fact that the use of L3s in Messiah is not strictly, or not only at least, related to a realistic function. It seems that their presence holds a post-carding function as well, that is, it becomes a means to better convey to the audience the idea that the story is taking place in a certain country or world region. Arabic is mainly featured in the scenes set in Syria that portray al-Masih, where it is employed realistically as the language of communication.

3. Findings

Whether or not to mark multilingualism (De Higes-Andino 2014, pp. 211–31) is one of the first questions to tackle when dealing with language diversity on screen. As far as Netflix productions are concerned, the findings will highlight a tendency to mark multilingualism. This choice is most evident at a macro strategy level. Among the three macro strategies identified by De Bonis (2014) discussed in the previous sections, only preservation and quantitative reduction could be observed, while neutralisation was not detected at the macro level in any of the shows investigated. Macro-strategies are a direct result of the strategies and techniques applied at a micro level. The analysis of the data sample identified three main translation strategies for the rendition of multilingual instances: transfer unchanged with part-subtitling, transfer unchanged untranslated, and dubbing. Two additional strategies were identified but these can be labelled as variations of the transfer unchanged strategy; that is, revoicing with part-subtitling and revoicing without translation. Furthermore, it is also interesting to observe the strategies adopted to deal with accents. In this case, two strategies were employed: neutralisation and preservation via a conveyed flavour incorporated by the Italian dubbing actors. The following subsections will first analyse the translation strategies and modalities observed on a micro level, and subsequently discuss the macro-strategies across each series.

3.1. Transfer Unchanged with Part-Subtitling

Transfer unchanged combined with part subtitling is undoubtedly the main translation strategy adopted for the rendition of multilingualism on Netflix. In fact, it features in every one of the five sample episodes analysed. This is in line with Netflix’s ’do-not-dub’ approach to foreign dialogues.

Transfer unchanged implies preserving the original voice track for the specific multilingual instance, while part-subtitles are added to convey the meaning to the target audience. This means that the L3 is preserved unvaried, and L2^Ita features in the form of part-subtitles. Part-subtitles in L2 are only added when L1 part-subtitles are already present in the original episode, therefore mirroring the same strategy adopted in the original production. Once again, this is in line with Netflix’s lip synch dubbing guidelines, which state that subtitles are to be added only when the foreign dialogue is meant to be understood. Of all the translation strategies identified, transfer unchanged with part-subtitling was applied to the widest variety of L3s: Spanish (L3^Spa), Yiddish (L3^Yid), Swahili (L3^Swa), Hebrew (L3^Heb), Arabic (L3^Ara), German (L3^Ger), Mandarin (L3^Man), French (L3^Fre), Zulu (L3^Zul), Afrikaans (L3^Afr), and occasionally even English (L3^Eng). Moreover, it was also the prevailing translation modality in three out of the five episodes analysed (see Section 3.6).

Table 1. Transfer unchanged with part-subtitles in Messiah (14.55).

Character	Original Version	Italian Dubbed Version	Back Translation
Al-Masih:	(in Arabic) [“If you look for truth, you may find comfort. If you look for comfort, you will never find truth.”]	(in Arabic) [“Se cerchi la verità, alla fine potrai trovare conforto. Se cerchi conforto, non avrai né conforto né verità.”]	[“If you look for truth, in the end you may find comfort. If you look for comfort, you will have neither comfort nor truth.]
Man:	(in Arabic) [Why do you quote an infidel?]	(in Arabic) [Perché citi un infedele?]	[Why do you quote an infidel?]
Al-Masih:	(in Arabic) [Who are you to judge who’s an infidel?]	(in Arabic) [Chi sei tu per giudicare chi è infedele?]	[Who are you to judge who’s an infidel?]
Jibril:	(in Arabic) [People are hungry, Imam.]	(in Arabic) [Abbiamo fame, Imam.]	[We’re hungry, Imam.]
Man:	(in Arabic) [Don’t interrupt, boy.	(in Arabic) [Non interrompere.]	[Don’t interrupt.]
Al-Masih:	(in Arabic) [Why do you call me Imam?]	(in Arabic) [Perché mi chiami Imam?]	[Why do you call me Imam?]
Jibril:	(in Arabic) [Because you are great.]	(in Arabic) [Perché sei grande.]	[Because you are great.]
Al-Masih:	(in Arabic) [No greater than you. “There is no deity but God.”]	(in Arabic) [Non più di te. “L’unico essere divino è Dio.”]	[No greater than you. “The only divine being is God.”]

provides an example drawn from Messiah: the dialogue in L3^Ara is transferred unchanged and part subtitled in L2^Ita. Except for al-Masih, who also speaks Hebrew and English, all Arabic speaking characters in Messiah’s episode 1 only speak Arabic. Therefore, the transfer unchanged strategy for large chunks of dialogue does not pose any voice harmonisation issues. They never switch to L1^Eng, and therefore never require dubbing into Italian. Furthermore, Messiah’s multilingual nature is directly linked to the variety of settings featured in the series. L3^Ara features in the scenes set in Syria, mostly in monolingual settings, reducing the contact between L1 and L3 to a minimum. This makes it easier to employ transfer unchanged as the translation strategy without producing noticeable voice shifts between original actors and dubbing actors.

Code-Switching and Code-Mixing

Transfer unchanged combined with part subtitling is not applied exclusively to monolingual L3 situations, such as the one illustrated in Table 1, but is used even in the case of code-switching and code-mixing occurrences. Code-switching and code-mixing feature in all the chosen pilot episodes; however, they feature more heavily in two shows in particular: Gentefied and Queen Sono.

Table 2. Code-switching in Gentefied (04.47).

Character	Original Version	Italian Dubbed Version	Back Translation
Javier:	¿Cinco dólares por un burrito? [Five dollars for a burrito?]	¿Cinco dólares por un burrito? [Cinque dollari per un burrito?]	[Five dollars for a burrito?]
Casimiro:	Solo aumentó 50 centavos. ¿Qué te costa ayudar a un amigo, pendejo? [It’s only a fifty-cent increase! What does it cost you to help your friend’s business, dumbass?]	Solo aumentó 50 centavos. ¿Qué te costa ayudar a un amigo, pendejo? [Sono solo 50 centesimi in più. Che ti costa aiutare un amico, imbecille?]	[It’s only 50 cents more. What does it cost you to help a friend out, dumbass?]
Javier:	¿Pendejo yo? ¡Pendejo tú! Me cuesta un viaje del metro. Un elote. Un taco. [I’m a dumbass? You’re the dumbass! It costs me a bus ride. An elote. A taco.]	¿Pendejo yo? ¡Pendejo tú! Me cuesta un viaje del metro. Un elote. Un taco. [Imbecille sarai tu! Mi costa un biglietto dell’autobus. Un elote. Un taco.]	[You’re the dumbass! It costs me a bus ticket. An elote. A taco.]
Ana:	A tostada, a bag of hot Cheetos.	Una tostada, un pacco di Cheetos.	A tostada, a bag of Cheetos.
Casimiro:	¡Oye! ¿Tú de qué lado estas, chamaca? [Hey! Whose side are you on, Little girl?]	¡Oye! ¿Tú de qué lado estas, chamaca? [Da che parte stai, ragazzina?]	[Whose side are you on, little girl?]
Erik:	Yeah, we’ve been giving you free burritos for years, man. Call it even.	Hai avuto burritos gratis per anni, bello. Siamo pari.	You’ve had free burritos for years, man (lit. beautiful). We’re even.

illustrates the use of transfer unchanged to preserve the multilingual nature of the exchange and carry over the code-switching present in the original version. In this example, the original voice track for the L3^Spa dialogue lines is retained and part-subtitles in L2^Ita are added to convey the meaning for the L2 target audience. The scene features an example of turn-specific code-switching. First generation immigrants Casimiro and Javier speak Spanish, while cousins Ana and Erik, who are third generation immigrants, seamlessly reply in English (Italian, in the dubbed version). This highlights a communicative preference that is typical of immigrant families (Beseghi 2019): the older generation tends to stick to the language of their country of origin, while the younger generation speaks the language of the host country in which they were born. Thanks to the use of transfer unchanged, Gentefied, from which the example is drawn, mirrors this multilingual phenomenon also in its dub stream. Obviously, the L2^Ita of the dubbed version is not the language of the host country, but that is when the suspension of disbelief comes into play. With turn-specific code-switching, issues of voice harmonisation are still secondary. When original and dubbed voices are not immediately juxtaposed, the voice switch can still arguably go unnoticed to the dubbing audience. The fact that voice harmonisation does not pose a huge issue could possibly be one of the reasons behind the choice of transfer unchanged as the translation strategy for this scene. However, the strategy is resorted to also in the case of intersentential code-switching, as shown further down in

Table 3. Code-switching in Unorthodox (45.16).

Character	Original Version	Italian Dubbed Version	Back Translation
Leah:	(in Yiddish) [You’re getting married. Congratulations!]	(in Yiddish) [Stai per sposarti! Mazeltov.]	[You’re getting married!]
Esty:	What do you care?	Che te ne importa?	What do you care?
Leah:	(in Yiddish) [Esty. I’m your mother.]	(in Yiddish) [Esty. Sono tua madre.]	[Esty. I’m your mother.]
Esty:	You stopped being my mother the day you left Williamsburg.	Non lo sei più da quando hai lasciato Williamsburg.	You haven’t been since you left Williamsburg.
Leah:	(in Yiddish) [No. I will always be your mother.] (in English) No matter where I go.	(in Yiddish) [No. Sarò sempre tua madre.] (in Italian) Non importa dove vada.	[No. I will always be your mother.] No matter where I go.

. In such cases, whenever possible, dubbing directors try to choose dubbing actors whose voice qualities resemble those of the original actors thus facilitating seamless voice shifts from a viewer perspective (Spiteri Miggiani 2019). This strategy, of course, is not always feasible since it depends on the available pool of resources and the demands and priorities associated with each role interpretation.

Table 3 provides an example drawn from Unorthodox. In this scene Esty finds her estranged mother Leah waiting for her at her grandparents’ house. Leah starts the conversation in Yiddish, but then resorts to English in between sentences. The dubbed version preserves the code-switching by combining the dubbing of L1^Eng into the L2 with part-subtitling of L3^Yid. In the dubbed scene, the original voice and the dubbed voice feature consecutively, making it very difficult for the audience not to notice the shift between voices, potentially disrupting the suspension of disbelief. Code-switching is particularly significant in this scene because until then, Esty had always used exclusively Yiddish with her relatives; however, although her mother speaks Yiddish, Esty chooses to use English, to signal that she does not consider her mother a member of her community, not part of her we-group, hence not worthy of the we-code.

Table 4. Code-mixing in Gentefied (16.07).

Character	Original Version	Italian Dubbed Version	Back Translation
Geo:	¡I’ll tell them el tío Erik se los mandó! [I’ll tell them Uncle Erik sent it to them!]	¡I’ll tell them el tío Erik se los mandó! [Dirò loro che sono da parte dello zio Erik.]	[I’ll tell them they’re from Uncle Erik.]

highlights an unusual consequence of the use of transfer unchanged with part subtitling when dealing with code-switching and code-mixing. It sometimes also leads to the preservation of the L1 (from the original voice track) along with the L3. Leaving the sentence unchanged causes English to constitute an additional L3 not present in the original version: L3^Eng. Even though the instances were admittedly quite sporadic, this phenomenon appeared in four of the series analysed. Rather than erasing the L3 occurrence or interrupting the dialogue flow with a potentially disrupting dub, it was presumably deemed better to preserve the L1 as well, maybe relying on the fact that the audience might not realise that some parts were in English. Whatever the reason, multilingualism still comes across; whether it is understood as code-switching and code-mixing depends on the audience’s proficiency in the L1 and L3 in question.

It is curious to note that in Gentefied’s original version code-mixing is not only present in speech but can also be found in the part-subtitles for L3^Spa, a strategy that is reproduced in the Italian dubbed version.

3.2. Revoicing with Part-Subtitling

Revoicing with part-subtitling can be considered a variation of the transfer unchanged strategy. Strictly speaking, nothing changes at a language level; the difference lies in the audio and voice qualities, as well as in the dubbing actors’ own accent or flavour. While the L3 is technically still transferred unchanged, the voices are not. This strategy avoids the bothersome issue of voice harmonisation, with some limitations of course. When observing the dialogue lines that are not revoiced, it is possible to confirm the intention behind the revoicing strategy where it is applied. Revoicing with part-subtitling features in two series, Messiah and Emily in Paris, although it appears more extensively in the latter. In Emily in Paris, most of the L3^Fre dialogues are revoiced, and some of those part-subtitled; the L3^Fre lines that are not revoiced are uttered by characters that only speak French and never switch to English, and therefore are not dubbed at all in the dub stream. Also, Chinese Mandarin is never revoiced, not even for those characters that do switch to English, highlighting the challenges posed by the revoicing strategy.

Table 5. Revoicing with subtitles in Emily in Paris (18.47).

Character	Original Version	Italian Dubbed Version	Back Translation
Mindy:	Laurent! Sybil! (in Mandarin) [Apologize to the lady!] Je suis désolée. Est-ce que je peux vous en acheter une autre? [I’m sorry. Can I buy you another one?]	Laurent! Sybil! (in Mandarin) [Scusatevi con la signora!] Je suis désolée. Est-ce que je peux vous en acheter un autre? [Mi dispiace. Posso comprartene un altro?]	[Apologise to the lady. I’m sorry. Can I buy you another one?]
Emily:	Sorry, I don’t speak French.	Mi spiace, non parlo francese.	I’m sorry, I don’t speak French.
Mindy:	Ah, American?	Ah, americana?	Ah, American?
Emily:	Yes.	Sì.	Yes.

presents a conversation between Emily and Chinese-speaking Mindy. In the dubbed version, L3^Man is transferred unchanged and part-subtitled, while L3^Fre is revoiced and part-subtitled. In the case of Emily or Mindy, correct pronunciation of French by the dubbing actors is not an issue, since the characters themselves are not native speakers. As for the French characters, the dubbing actors satisfactorily revoiced French dialogue to cover the dialogue of French native speakers to a non-French audience. Thanks to this strategy, the voices of the dubbed version, despite the language switch, are perfectly harmonised in the case of French. This approach was most likely deemed too challenging for L3^Man, a language much less familiar and not even remotely related to the native language of the dubbing voice talents.

3.3. Transfer Unchanged with No Translation

Untranslated transfer unchanged only applies to 1% of all the multilingual instances encountered, however it features in three out of five of the series examined: Emily in Paris, Queen Sono, Messiah. In Emily in Paris and Queen Sono the strategy mirrors what happens in the original version; no subtitles are added simply because the exchange is left untranslated to start with, and a translation is already provided diegetically, or it can be assumed that it is not meant to be understood, and therefore left untranslated in the dub stream as per Netflix’s dubbing recommendations.

Table 6. Transfer unchanged with no translation in Emily in Paris (15.36).

Character	Original Version	Italian Dubbed Version	Back Translation
Baker:	Mademoiselle, bonjour. [Good morning, miss.]	Mademoiselle, bonjour. [Buongiorno, signorina.]	[Good morning, miss.]
Emily:	Une pain au chocolat.	Une pain au chocolat.
Baker:	Un ! Pas ‘une’. Un pain au chocolat. Ça sera tout ? Un euro quarante. Eh ben, on n’est pas rendu.	Un ! Pas ‘une’. Un pain au chocolat. Ça sera tout ? Un euro quarante. Eh ben, on n’est pas rendu.
Emily:	Merci. Have un bonne journée.	Merci. Have un bonne journée.
Baker:	Une ! Pas ‘un’. Une bonne journée !	Une ! Pas ‘un’. Une bonne journée !

illustrates how various translation strategies can be combined within the same episode. The baker’s lines are transferred unchanged, while Emily’s lines are revoiced. Apart from the initial greeting, the dialogue is left completely untranslated, even though it goes beyond mere phatic expressions. Emily attempts to speak French and the baker corrects a grammar mistake that she makes. Neither the original version nor the dubbed one provide subtitles, perhaps to highlight Emily’s slight bewilderment at not understanding, therefore presenting her perspective. Nevertheless, the general meaning of the exchange can be easily grasped from context and gestures.

In Messiah, transfer unchanged with no translation is not employed for any of the L3s but for L1^Eng, which thus constitutes a further L3 in the dubbed version.

In the example featured in

Table 7. Transfer unchanged with no translation in Messiah (29.09).

Character	Original Version	Italian Dubbed Version	Back Translation
Female soldier:	Shit.	Cazzo.	Fuck.
Male colleague:	(in Hebrew) [What?]	(in Hebrew) [Che c’è?]	[What?]
Female soldier:	(in Hebrew) [We’ve got a crazy fucker.]	(in Hebrew) [Un bastardo schizzato in avvicinamento.]	[A crazy bastard approaching.]
Male colleague:	(in Hebrew) [Hello there, crazy fucker.] Oh, shit. Oh, shit, shit, shit, shit. Fuck.	(in Hebrew) [Ehilà, bastardo schizzato.] Oh, shit. Oh, shit, shit, shit, shit. Fuck.	[Hello there, crazy bastard.]

, Al-Masih is approaching the Israeli border from the desert, and the soldiers at the security post are on the alert as soon as they spot him and his large following. The first exclamation uttered by a female soldier is dubbed into Italian, however the very same word uttered by her colleague, along with other expletives, is left in English and the dubbed version features the original voice track. The meaning of the Hebrew utterance is conveyed through Italian subtitles, while the English swearwords are left untranslated. That said, the audience can rely on the tone of voice and context.

3.4. Revoicing with No Translation

Revoicing with no translation features at a higher percentage rate (5%) than the simple transfer unchanged strategy, and is observed in four of the sample series. The only series in which this strategy is not adopted is Messiah. Revoicing with no translation is the main strategy employed in Emily in Paris, albeit under specific circumstances: only instances that do not require translation in the first place are left untranslated. Otherwise, dialogue in L3^Fre is either part-subtitled or the translation is provided diegetically, with a few rare exceptions (Table 6). This is in line with the creative intent of language diversity on Netflix productions. Foreign languages and L3s generally serve as plot devices and not merely as ornaments, therefore what they express should be conveyed. Untranslated revoicing was resorted to especially in the case of code-mixing. Code-mixing instances mostly involve greetings and basic sentences, easily understood by the audience without the aid of subtitles.

Similar to the example in

Table 8. Revoicing with no translation in Emily in Paris (09.10).

Character	Original Version	Italian Dubbed Version	Back Translation
Sylvie:	Well, that’s very unfortunate.	Beh, è davvero un peccato.	Well, that’s very unfortunate.
Emily:	Excuse me?	Come, mi scusi?	Excuse me, what?
Sylvie:	That you don’t speak French. It’s a problem.	Che tu non parli francese. È un problema.	That you don’t speak French. It’s a problem.
Emily:	Well, I’m going to take a class, but je parle un peu français already.	Ecco, mi sono iscritta a un corso, ma je parle un peu français di già.	Well, I’ve enrolled into a course, but je parle un peu français already.
Sylvie:	Perhaps it’s better not to try.	Forse è meglio non improvvisare.	Perhaps it’s better not to improvise.

, many code-mixing occurrences are simply carried over to the Italian script and revoiced by the dubbing actors. When L3s are well-known to the target audience (such as French and Spanish), revoicing with no translation can be applied to short dialogue lines that go beyond perfunctory expressions. This is not the case for other languages, such as L3^Swa, that are likely considered more obscure for the target viewers. Only basic greetings in the said L3s can afford to not be translated.

3.5. Dubbing

Among all the translation modalities identified in the data sample, dubbing is the only one that Netflix guidelines advise against because it generally implies neutralising the language variety. In theory, no foreign dialogue should be dubbed over, and indeed this study set out to verify the application and feasibility of this guideline. The analysis reveals that dubbing was resorted to in 13% of multilingual instances, though it features in only two out of five series: Gentefied and Queen Sono. Moreover, it is the predominant strategy only in Queen Sono. The motivation probably lies in the abundance of code-switching and code-mixing in these two series, when compared to the others. Indeed, utterances are dubbed over in the case of turn-specific and intersentential code-switching and code-mixing.

Table 9. Dubbing in Gentefied (06.22).

Character	Original Version	Italian Dubbed Version	Back Translation
Rob:	Do you have the rent? If not, I have business to handle.	Ce l’avete l’affitto? Ho ancora un mucchio di faccende da sbrigare.	Do you have the rent? I still have a lot of errands to run.
Casimiro:	Vete por los tacos. I don’t have all of it. [Go handle the tacos.]	Va’ a prendere i tacos. Non ho tutta la cifra.	Go get the tacos. I don’t have the whole sum.

features an instance of intersentential code-switching. It can be assumed that neutralisation was opted for to try not to disrupt the viewing experience with a perceivably different voice. However, as mentioned earlier, the series dubs over the L1^Spa into the L2 also in cases of turn-specific code-switching for no obvious reason (see

Table 10. Transfer unchanged combined with dubbing in Gentefied (09.04).

Character	Original Version	Italian Dubbed Version	Back Translation
Casimiro:	¿Qué piensas, mijo? ¿Qué piensas? [What do you think, mijo? What do you think?]	Tu che ne pensi? Che ne pensi?	What do you think? What do you think?
Chris:	Uh… Well, have you asked my dad for help?	Ahh… Hai chiesto aiuto a mio padre?	Uh… Have you asked my father for help?
Casimiro:	¿Tu papá? ¿Ni siquiera te ayuda a ti, cómo me va a ayudar a mí? [Your dad? If he’s not helping you, do you think he’s going to help me?]	¿Tu papá? ¿Ni siquiera te ayuda a ti, cómo me va a ayudar a mí? [Tuo padre? Se non aiuta te, credi che aiuterà me?]	[Your father? If he doesn’t help you, do you think he’s going to help me?]

). It is unclear why one line would be dubbed and the other part-subtitled, if not to reduce the number of subtitles for a historically dubbing audience.

In Table 10, only some multilingual instances in Gentefied are dubbed over, while the majority are still transferred unchanged. That is not the case in Queen Sono, where the dubbing strategy is more pervasive (see

Table 11. Dubbing in Queen Sono (27.15).

Character	Original Version	Italian Dubbed Version	Back Translation
Mazet:	(in isiZulu) [Tea is ready, my dear.]	Ecco il tè. [Il tè è pronto, cara.]	Here’s the tea. [The tea is ready, dear.]
Queen:	(in isiZulu) [You know, Ma, for as long as I’ve known you...] (in English) from when I was little (in isiZulu) [you’ve never served tea in that nice tea set that sits at the top. You’ve never served me or anyone, including yourself.]	Sai, nonna, fin da quando ero piccola, non hai mai voluto servire il tè in quel bellissimo servizio che tieni lassù. Non hai mai voluto usarlo. Non l’hai mai voluto tirare fuori per nessuno. Né per me, né per te stessa. [Sai, Ma, da quando ti conosco… non hai mai servito il tè con quel servizio buono che sta in cima. Non l’hai mai usato per me, né nessun altro. Neanche te stessa.]	You know, grandma, ever since I was little, you’ve never served tea in that beautiful tea set you keep up there. You’ve never wanted to use it. You’ve never taken it out for anybody. Not for me, not even for yourself. [You know, Ma, for as long as I’ve known you…you’ve never served tea in that nice tea set that sits at the top. You’ve never used it for me, nor for anybody else. Not even yourself.]
Mazet:	(in isiZulu) [No, my child. You see, those…] (in English) is for special occasions. (in isiZulu) [Never.]	No, figlia mia. Quel bel servizio è per le occasioni speciali. Tutto qui. [No, figlia mia. Quello è…] [Mai.]	No, my child. That nice tea set is for special occasions. That’s all. [No, my child. That is…] [Never.]
Queen:	(in isiZulu) [Here’s my ghost of a father. I wonder how he is…]	Questa sono io da piccola. Almeno credo. [Ecco quel fantasma di mio padre. Chissà com’è.]	This is me when I was little. At least I think so. [Here’s my ghost of a father. I wonder how he is.]
Mazet:	(in isiZulu) [It’s him.]	Sì, certo.	Yes, of course.
Queen:	Mazet, can I please have a very important conversation with you?	Mazet, posso affrontare un discorso molto serio con te?	Mazet, can I talk to you about a very serious matter?
Mazet:	(in isiZulu) [What?]	Quale? [Cosa?]	What? [What?]
Queen:	I am very worried about you.	Sono molto preoccupata per te.	I’m very worried about you.

).

In Table 11, Queen and Mazet are having tea after visiting Safiya’s tomb and the conversation between the two involves substantial L1^Eng + L3^Zul code-switching and code-mixing. In the dubbed version, there is no trace of any type of code switching; furthermore, the dubbed version conveys a different meaning. Instead of looking at a photograph of her father as a baby, Queen is looking at herself. In addition, the dubbing over of Zulu dialogue lines is evident when subtitles continue to appear even along the dubbed parts. Not only are the viewers aware when the dialogue is in a foreign language in the original version, they also know when the meaning of the foreign utterances has been changed. The audience is alerted to all the code-switching instances erased through dubbing because the forced narratives are still there. This does not tally with the norms considering that the voice track provides an Italian dub throughout and these forced narratives do not match the dubbed dialogue. A possible oversight in the post-production process can perhaps be hypothesised. However, overlapping of modes aside, what is more relevant in this example is the fact that entire dialogue exchanges are dubbed over, not only random L3 words and phrases. This implies the neutralisation of a significant quantity of multilingual instances throughout the episode.

The fact that dubbing is resorted to exclusively to deal with code-switching and code-mixing highlights the challenges of this linguistic phenomenon. Preserving multilingualism is not always feasible in the presence of any type of code-switching and this seems to constitute the main trigger for neutralisation. Apart from the technical challenges involved, resorting to subtitles as a way out might be considered too taxing for the audience that chooses the dub stream. When considering the examples and varying degrees to which the dub-over strategy is applied, it might be safe to assume that the familiarity of the foreign languages in the target culture plays a role in the decision. For instance, the African languages in Queen Sono when compared to French and Spanish in the other series, may alienate an Italian audience, to the point that preserving them all throughout the series was most likely not considered an option in the first place. The audiences’ unfamiliarity with the L3 alone does not seem to determine a neutralising approach. However, it becomes crucial when having to retain code-mixing. Furthermore, on a practical level, the fact that it might be easier for non-native speakers to reproduce a language closely related to their own native language most likely plays a part, as well. Dubbing actors are more likely to be able to revoice the L3 in such cases, therefore ensuring voice harmonisation.

3.6. Accents

As far as accents are concerned, two strategies were employed: neutralisation and preservation in the form of conveyed flavour. Inasmuch as the style guide outright states that the decision about accents is up to the production and Netflix (Netflix 2022a), it can be assumed that these options were deemed the most appropriate by the creative directors and dubbing managers of the specific series.

The five sample episodes, by virtue of the various geographical settings and different ethnicities of the characters, display a variety of English accents. Most of them, however, constitute the L1, and as such required translation. Subsequently, neutralisation emerges as the preferred strategy, employed in four out of five series. The strategy is applied to all accents considered native to English, e.g., American, and South African, but also to foreign accents. In Messiah, Undersecretary Kirmani’s Pakistani accent is dubbed into standard Italian. In Unorthodox, although most characters (including the protagonist) speak English with a foreign accent, foreign-accented English does not constitute an L3. The foreign accents, apart from Esty’s maybe, do not serve as characterisation nor as plot devices, they are merely a realistic by-product of English being employed as lingua franca. Moreover, all foreign accents have the same relevance. Therefore, they all constitute the same L1, and are all neutralised by being dubbed into the standard L2.

The strategy of preservation is applied to foreign accents only in one series: Messiah. The Middle Eastern accents featured in the sample episode are conveyed in the dubbed version, but as a flavour rather than proper accents. Al-Masih’s accent in the original version is quite subtle and not clearly recognisable as typically Arabic or Iranian, which is where the character is from. It sounds more like a generic foreign accent that could belong to someone from the Middle East. In the Italian dubbed version this accent is preserved by means of a slight variation on standard pronunciation and enunciation. Just like in the original, the Italian version of the accent cannot be associated to a specific country. It is rather an exotic flavour. The same strategy is applied for English spoken with an Israeli accent. Aviram, as other Hebrew native speakers, is dubbed into standard Italian as far as grammar is concerned, but his speech is characterised by a slightly clipped pronunciation incorporated by the Italian dubbing actor.

3.7. Macro Strategies and Quantification of Results

The following section provides an overview of the strategies encountered in each series, and more importantly quantifies the number of instances in which each strategy is applied. This allows to identify the translation strategies at a macro level, based on the individual micro strategies outlined so far. Indeed, when it comes to macro strategies the ones mainly observed are preservation and quantitative reduction. Naturally, L1^Eng dubbed into L2^Ita is taken as the main default mode throughout each series and is therefore listed, but not quantified. The strategies and modalities involving L3s are our concern and their quantification adds valuable data to the findings.

3.7.1. Preservation

Preservation constitutes the main macro strategy applied in the Italian dub streams of multilingual Netflix Originals. Out of the five series analysed, multilingualism was entirely preserved in Unorthodox, Emily in Paris and Messiah. This means that at the micro strategy level no multilingual instance was erased or neutralised; a result achieved through the employment of several translation modalities.

Table 12. Language distribution in Unorthodox and its Italian dub stream.

Unorthodox	STEng		TTIta	Translation Modality	No. of Instances
	L1Eng		L2Ita	Dubbing	default
Code-switching			L3Eng	Transfer unchanged, part-subtitling	8
	L3Yid		L3Yid	Transfer unchanged, part-subtitling	183
	L3Ger		L3Ger	Transfer unchanged, part-subtitling	7
Code-mixing	L1Eng + L3Ger		L2Ita + L3Ger	Dubbing + Revoicing, diegetic interpreting	1
Macro Strategy			Preservation

shows the distribution of the L1, L2 and L3s and the translation strategies and modalities applied at the micro level in episode one of the series Unorthodox. None of the multilingual instances were erased; on the contrary, all were preserved through the transfer unchanged and part-subtitling strategy. Therefore, preservation is the predominant strategy in this series.

Table 13. Language distribution in Emily in Paris and its Italian dub stream.

Emily In Paris	STEng		TTIta	Translation Modality	No. of Instances
	L1Eng		L2Ita	Dubbing	default
	L3Fre		L3Fre	Revoicing, part-subtitling	11
			L3Fre	Revoicing, no translation	16
			L3Fre	Transfer unchanged, part-subtitling	3
			L3Fre	Transfer unchanged, no translation	3
Code-mixing	L1Eng + L3Fre		L2Ita + L3Fre	Dubbing + Revoicing, no translation	20
	L3Man		L3Man	Transfer unchanged, part-subtitling	2
Macro Strategy			Preservation

details the micro strategies encountered in Emily in Paris. The combination of dubbing with other screen translation modalities effectively reproduces the multilingual nature of the original. No foreign language utterance was dubbed over. Indeed, the dubbed version spared no effort to harmonise the dubbed voices. There are no sudden shifts between original and dubbed voices and this ensures the suspension of disbelief. The prevailing translation strategy seems to be no translation at all accompanied by the revoicing of the multilingual instances, and part-subtitling as a second resort. That said, as noted earlier, only instances that do not require translation in the first place are left untranslated; actual dialogues in L3^Fre are all part-subtitled. Preservation is, without any doubt, the macro strategy applied in this series.

The strategies outlined in

Table 14. Language distribution in Messiah and its Italian dub stream.

Messiah	STEng		TTIta	Translation Modality	No. of Instances
	L1Eng		L2Ita	Dubbing	default
Code-switching			L3Eng	Transfer unchanged, part-subtitling	2
				Transfer unchanged, no translation	2
	L3Heb		L3Heb	Transfer unchanged, part-subtitling	48
				Revoicing, part-subtitling	2
	L3Ara		L3Ara	Transfer unchanged, part-subtitling	90
	L1Eng w/ Hebrew accent		L2Ita w/ Hebrew accent	Dubbing, conveyed accent	37
	L1Eng w/ Arabic accent		L1Eng w/ Arabic accent	Dubbing, conveyed accent	21
Macro Strategy			Preservation

point towards preservation as the macro-strategy in Messiah. None of the multilingual instances were erased or omitted. Transfer unchanged with part-subtitling is undeniably the main translation strategy adopted on a micro level. Foreign languages were carried over, therefore both Arabic and Hebrew retain the same level of visibility in the original and dubbed versions. Moreover, foreign accents were also conveyed in the Italian dubbing. As such, the dubbed version can be considered as multilingual as the original version, if not more, considering that the dub stream includes English as an additional L3.

3.7.2. Quantitative Reduction

Quantitative reduction is the only other translation macro-strategy observed on dubbed Netflix multilingual shows. It was encountered only in the series Gentefied and Queen Sono.

Table 15. Language distribution in Gentefied and its Italian dub stream.

Gentefied	STEng		TTIta	Translation Modality	No. of Instances
	L1Eng		L2Ita	Dubbing	default
Code-switching	L3Spa		L3Spa	Transfer unchanged, part-subtitling	71
				Transfer unchanged, diegetic interpreting through dubbing	1
			L2Ita	Dubbing (neutralisation)	13
Code-mixing	L1Eng + L3Spa		L2Ita + L3Spa	Dubbing + Revoicing	9
			L3Eng + L3Spa	Transfer unchanged, part-subtitling	6
			L2Ita	Dubbing (neutralisation)	13
	Subtitles: L1Eng + L3Spa		Subtitles: L2Ita + L3Spa	Speech transferred unchanged, part-subtitling featuring code-mixing	12
			Subtitles: L2Ita	Speech transferred unchanged, code-mixing in subtitles omitted	7
Macro Strategy			Quantitative Reduction

highlights transfer unchanged combined with part-subtitling as the most frequent micro-strategy in Gentefied. Yet, a noticeable (though not too many) number of occurrences were neutralised through dubbing into Italian. This was also combined with the transfer unchanged strategy. Therefore, in line with the classification approach provided by De Bonis (2014), the translation macro strategy, in this case, can be labelled as quantitative reduction since both L1 and L3 were dubbed over at least to a small extent. And, although code-switching and code-mixing are, as a matter of fact, retained throughout the dubbed episode, their presence is quantitatively reduced.

As shown in

Table 16. Language distribution in Queen Sono and its Italian dub stream.

Queen Sono	STEng		TTIta	Translation Modality	No. of Instances
	L1Eng		L2Ita	Dubbing	default
Code-switching			L3Eng	Transfer unchanged, part-subtitling	1
	L3Swa		L3Swa	Transfer unchanged, part-subtitling	9
			L3Swa	Revoicing, no translation	1
			L2Ita	Dubbing (neutralisation)	2
Code-switching	L3Zul		L3Zul	Transfer unchanged, part-subtitling	12
			L2Ita	Dubbing (neutralisation)	26
Code-mixing	L1Eng + L3Zul		L2Ita	Dubbing (neutralisation)	19
	L3Xit		L3Xit	Transfer unchanged, no translation	1
Code-switching	L3Afr		L3Afr	Transfer unchanged, part-subtitling	1
			L2Ita	Dubbing (neutralisation)	2
	L3Fr		L3Fr	Transfer unchanged, part-subtitling	4
Macro Strategy			Quantitative Reduction

, Queen Sono adopts a combination of neutralisation and preservation strategies, leading to contrasting results. This combination can also be labelled as quantitative reduction. The approach is well-intended and aims to make the show more enjoyable for the target audience, who does not have to rely on too many subtitles. However, this approach also implies a certain extent of manipulation whereby the role of multilingualism in the product is altered, in this case concealing the identities of the many ethnicities of South Africa, and Africa in general. The different languages and identities are not completely erased, but multilingualism is subject to significant quantitative reduction. In fact, as per the number of instances recorded for each micro-strategy, illustrated in Table 16, the dubbing of multilingual instances accounts for the largest percentage.

To conclude this wider analysis based on the quantification of applied micro-strategies,

Table 17. Distribution of translation modalities in the five Netflix Originals analysed.

Translation Modality	No. of Instances		%		No. of Series
Transfer unchanged, part-subtitling	447	460	79%	81%	5
Revoicing, part-subtitling	13		2%		2
Transfer unchanged, untranslated	6	31	1%	6%	3
Revoicing, untranslated	27		5%		4
Dubbing (neutralisation)	75		13%		2
Conveyed accent	58		-		1
Transfer unchanged, part-subtitles with code-mixing	12		-		1

provides an overview of all the translation strategies observed in the case study. A percentage value is drawn from the quantified instances to visually convey their predominance. The table also provides the number of productions in which each strategy features so as to establish their recurrence across the data sample.

4. Discussion

Netflix Norms versus Television Norms

The strategy of using dubbing for the L1 alongside subtitles for the L3 emerged as the main pattern or norm on Netflix, in line with what scholars (Díaz Cintas 2011; Minutella 2012; Beseghi 2017; among others) have noted in the Italian dubbing field in general. However, while for Netflix original productions this approach to multilingualism seems to be the norm or close to being the norm, this is not the case in conventional media. As observed by Parini (2015), this combination of translation strategies is preferred but not yet established.

Indeed, the main difference between the two media is the point of departure. As discussed in Section 1, dubbing for TV and cinema in Italy historically stems from a place of neutralisation, and is only now moving closer towards preservation, implementing strategies that retain the multilingualism present in the original. As evidenced in this research work, Netflix has positioned itself on the preservation end of the spectrum, occasionally dipping into neutralising strategies but never, as far as this study can observe, resorting to a total neutralising macro-strategy that would flatten out all language diversity. Furthermore, no L3 was completely erased in the dub streams investigated; its presence was at most quantitatively reduced.

As previously discussed, the recent tendency to preserve multilingualism at least partially on traditional media, positions the dubbing strategies and modalities applied in conventional media close to those detected on Netflix. In fact, as De Bonis (2014, p. 259) notes, in conventional Italian dubbing, L3s are most often neutralised when they are intended for characterisation purposes and especially if code-switching is involved. This is exactly what was observed in the analysed series: a quantitative reduction of the multilingual situations when faced with recurring code-switching. Therefore, the macro-strategy applied to code-switching in dubbing does not seem to vary across media, be it TV or streaming platforms. The difference lies in the fact that neutralisation was observed on Netflix exclusively in the case of code-switching and code-mixing instances. In dubbing for television, as also stated by Parini (2015), neutralisation is employed on other occasions, as well. Moreover, no jarring dialogue featured in the Italian dub streams on Netflix, which is sometimes the case with the neutralisation of polyglot scenes intended to portray communicative problems (De Bonis 2014, p. 53). It could, of course, be argued that no such issues occurred because language diversity in the chosen data sample was intended to reproduce linguistic authenticity and was not meant to create linguistic misunderstandings or conflict.

The transfer unchanged strategy was adopted by Netflix also in its revoicing variant. Revoicing is by no means a new practice and is resorted to also for TV and cinema content. Section 1 discusses how revoicing on TV is generally adopted in the case of L1-to-L3 code-switching occurrences, particularly if the character is not a native L3 speaker. This is not the case on Netflix, where revoicing was applied both for native and non-native speakers, and in L3-only situations. Needless to say, the challenge represented by non-Western European languages is not influenced by the type of media since the same pool of dubbing voice talents is used. Therefore, similarly to what was observed in the revoiced instances on Netflix, it is much easier to find revoiced L3 dialogues on television when the language to revoice is familiar and/or linguistically related to the L2, which, in the case of Italian territory, implies a Western European L3.

A certain resistance and hesitation to reproduce accents (especially thick ones) emerged in the Netflix dub streams, therefore in line with conventional media. The homogenising tendency typical of Italian conventional dubbing, as discussed in Section 1, emerged also in the data sample. As per the dubbing standard, accent varieties native to the English language were neutralised. Furthermore, no adjustment to reproduce any accent distinction at a syntactic and/or lexical level (as is sometimes the case) was needed, since the native accents in question were not particularly relevant to the plot. As mentioned, conventional practice is slightly different in the case of non-native accents (Spiteri Miggiani 2019; Minutella 2021). Foreign accents were conveyed on Netflix in only one of the series, and rendered as a flavour. As said, as far as accents are concerned the practice seems very much in line with that of the rest of the dubbing field.

What emerges from this comparison is that some of the dubbing strategies enforced by Netflix coincide with the ones that researchers observe in the current Italian dubbing practice on conventional media: part subtitling as the main translation modality for multilingualism; dubbing of some of the multilingual instances in the case of recurring code-switching and code-mixing; and a certain resistance to reproducing accents. Given that the dubbing practice does not always match what is recommended by the platform, it could be argued that traditional and conventional norms have an impact on Netflix’s practices, at least to a certain extent. Or rather, what probably weighs on the decision-making process is not necessarily the dubbing tradition itself, but the audience’s habituation to that dubbing tradition. The average widespread Italian dubbing audience is not yet that accustomed to continuous subtitles and needs to be eased into the practice so as not to be overwhelmed.

The data sample drawn from Netflix does not overindulge in neutralising modalities, and, on the contrary, makes ample use of subtitles, displaying the platform’s trust in the audience and its ability to tolerate them. Heiss (2004 p. 215) indeed had suggested that a misguided evaluation of the target audience may be at the root of the flattening out of multilingualism in films—an approach adopted due to fear of reduced success and earnings because of the average viewer being alienated by too many different languages and the mental effort their presence supposedly requires.

Other common threads between conventional and streaming platform approaches could also be attributed to the fact that the practitioners (translators, directors, actors etc.) involved in the dubbing process of Netflix productions are the same people working for television networks, film production companies and distributors. Even though they do not hold ultimate decisional power, what may be worth considering is the challenge to juggle different approaches according to the client involved. The choices made in the dubbing studios can potentially be influenced by long term professional practice habituation, in the case of both translators and dubbing directors.

Shifting perspective, the correlation between traditional and streaming media might not necessarily be that of Netflix adopting conventional dubbing norms. It could very well be the other way around. The Italian dubbing field started to embrace the possibility of L3 preservation prior to the arrival of Netflix in Italy. However, it cannot be excluded that the platform’s strong tendency towards the preservation of multilingualism might be enticing a less domesticating approach in the dubbing for conventional media. It could very well be that television dubbing is progressively getting closer to the Netflix model, similarly to the influence that, as Manfredi (2021) notes, Netflix has had in the new representation of Latino characters as dual-language speakers in TV shows in the US. After all, an upside of being a pioneer for a new type of media is that you can make your own rules and not necessarily comply with tradition.

5. Conclusions

The combination of marking and non-marking strategies observed in the dubbing of Netflix Productions ensured a degree of visibility to linguistic diversity in the dubbed versions, albeit not always consistent with that of the originals. The overall result is that, despite some multilingual utterances getting lost in translation, all five productions chosen as the data sample can still be labelled multilingual in their dubbed versions. Nonetheless, Netflix’s directive to avoid dubbing over foreign language was not upheld consistently, and foreign languages were in fact occasionally dubbed over, particularly when code switching and code mixing occurred.

In sum, the findings of this research study reveal a strong tendency to mark and preserve multilingualism, in line with Netflix’s own policies and dubbing specifications. Transfer unchanged combined with part-subtitling constitutes the primary translation strategy, although it is not the only one adopted. The presence of code-switching seems to disrupt the preservation intent and calls for different strategies, also due to technical reasons tied to use of different voices for the same characters. Total neutralisation of all multilingual instances is not observed in any of the sample series, a tendency which, on the other hand, seems to characterise post-production dubbing for Italian television networks. That said, a certain degree of overlap between multilingual translation norms on Netflix and the more conventional Italian dubbing practices can still be observed. However, unlike television, Netflix sets the preservation of multilingualism as a prerogative from the outset.

Although the observation of the translation strategies in the five pilot episodes has led to significant conclusions, providing a preliminary understanding as to how Netflix preferably approaches multilingualism in dubbing, the analysis is based on a very small data sample, and is narrowed down to one streaming platform and one target language. A larger sample size could, undoubtedly, give the findings a more solid foundation. That said, it is interesting to note that the strategies across Netflix target language streams tend to be quite standardised, therefore the analysis of one target language could possibly mirror the results in other target languages. While there certainly is scope for further research, the present work hopes to represent a stepping stone towards defining where localisation trends are heading and how TV dubbing scriptwriters need to adjust their consolidated default choices to adapt to these newer norms and trends.