‘No’ Dimo’ par de Botella’ y Ahora Etamo’ Al Garete’: Exploring the Intersections of Coda /s/, Place, and the Reggaetón Voice

Abstract

The rebranding of reggaetón towards Latin urban has been criticized for tokenizing Afro-Caribbean linguistic and cultural practices as symbolic resources recruitable by non-Caribbean artists/executives in the interest of profit. Consumers are particularly critical of an audible phonological homogeneity in the performances of ethnonationally distinct mainstream performers, framed as a form of linguistic minstrelsy popularly termed a ‘Caribbean Blaccent’ that facilitates capitalization on the genre’s popularity by tapping into the covert prestige of distinctive phonological elements of Insular Caribbean Spanish otherwise stigmatized. This work pairs acoustic analysis with quantitative statistical modeling to compare the use of lenited coronal sibilant allophones popularly considered indexical of Hispano-Caribbean origins in the spoken and sung speech of four of the genre’s top-charting female performers. A general pattern of style-shifting from interview to sung speech wherein sibilance is favored in the former and phonetic zeros in the latter is revealed. Moreover, a statistically significant increased incidence of [-] across time shows the most recent records to uniformly deploy near-categorical reduction independent of artists’ sociocultural and linguistic backgrounds. The results support the enregisterment of practices popularized by the genre’s San Juan-based pioneers as a stylistic resource—a reggaetón voice—for engaging the images of vernacularity sustaining and driving the contemporary, mainstream popularity of música urbana.

1. Introduction

Following generations of socioeconomic marginalization in the United States, young Latinx and Latin Americans have found in Latin urban music a space to proudly assert hispanophone identities as part of a global community. Amidst underground origins and popular depictions of the genre as the music of alternative youth, música urbana continues to set records across international media platforms, materializing as a symbol of “an emergent pan-Latin American identity” (Marshall 2008, p. 132; Viñuela 2020). This identity is reflected in the sociophonetics of the artistic performance speech (Hayes 2022) of the top-charting artists. Though notwithstanding this popularity, there is a lack of scholarship attending to the sociolinguistics of urbano stage performance. This is surprising, as the linguistic distinctiveness of música urbana, in comparison with other forms of popular Spanish-language music, has been an area of interest since the category’s genesis (see Marshall 2009).

The covert prestige of forms generally categorized as nonnormative in transnational Latinidad discourse in urbano music has been attributed to artistic influences derived from U.S. hip-hop, an African American movement born out of resistance to the repressive policies of the Reagan–Bush administrations that characteristically prioritizes markers of vernacularity (Morgan 2001; Romero Joseph 2009).1 Recently, critiques have pointed out striking similarities between the singing accents of contemporary urban artists of varying ethnonational origins. In particular, these critiques highlight speakers of ethnonationally and raciolinguistically privileged backgrounds’ shift towards a register resembling the Spanish characteristic of the Insular Caribbean (Barron 2019; Caraballo 2019). This rhetoric mirrors the discourse surrounding white rap performers’ use of African American English by positioning these similarities as a form of linguistic minstrelsy (Bucholtz and Lopez 2011) that grants highly visible, non-Caribbean artists access to a genre popularized by musicians whose speech and presence is transnationally Othered and restricted to the margins.

Record labels often work to enregister a particular dialect as characteristic of a genre, adding a dialectal index to the semiotic resource bank for “deliberate and reflexive identity performances” that establish industry categories (Bucholtz and Lopez 2011, p. 681). Powell (2024) provided evidence of Hispano-Caribbean singers exhibiting style-shifts regarding the reduction of coronal sibilants in coda (i.e., los gatos [lo ˈgato]) from a transnationally standardized s-full register towards a prescriptively Caribbean s-less register in response to genre norms. These results suggest the existence of a stylistic norm for performing reggaetón, what I term herein the reggaetón voice. The present study furthers this line of inquiry through examinations of coda /s/ in the speech of non-Hispano-Caribbean performers. Pairing acoustic analysis and variationist statistical modeling, this investigation compares distributions of lenited sibilant allophones in spoken interviews and musical performances by four of the genre’s top-charting female performers: Anitta, Becky G, Karol G, and Natti Natasha. Each of these women is considered a contemporary icon within the Latin urban industry, having significantly aided the normalization of a feminine perspective over the course of the last decade (Rivera Mercado 2022). Moreover, the sociolinguistic and ethnonational variability of these participants—representing native speakers of both s-retaining (Karol G) and s-reducing varieties (Natti Natasha), as well as heritage (Becky G) and non-native speakers of Spanish (Anitta)—provides scholars with a unique opportunity for comparative analysis.

As detailed herein, examinations of language as performance require attention to all linguistic and non-linguistic modalities of the performance. In acknowledgment, this exploratory work focuses exclusively on language as a point of departure for theorizing the enregisterment of Caribbean varieties of Spanish as recruitable voicing resources for indexing images of urban Latinidad (Otheguy and Zentella 2012)2. What follows is a detailed review of the theoretical orientation underscoring the sociolinguistics of staged performance as detailed by Bell and Gibson (2011), Coupland (2011), and Gibson and Bell (2012). A recount of the emergence of the Latin music enterprise is then offered to orient ‘Latin’ music as a homogenizing discursive frame. Finally, this work offers critical commentary regarding the significance of sibilance reduction as a normalized, stylistic index of the reggaeton voice relating to the vernacularity associated with and sustaining the popularity of the genre.

2. Background

2.1. Sociolinguistics of Staged Performance

Speakers as social actors constantly make agentive choices between alternative ways of speaking, which has led some researchers to suggest all language is performed (Goffman 1981). Coupland (2007) distinguished the notion of performance in terms of mundane and high performance. The former are cases of language display (e.g., reported speech, see Schilling-Estes 1998) that draw attention to the esthetic form of utterance and spontaneously create a performer–audience dynamic in otherwise everyday languaging (Shuck 2004). High performances (hence staged performance, see Bell and Gibson 2011) are instances where speech is recruited, reproduced, and reframed as a situated, artistic communicative practice of expressive, objectified verbal display over some duration, subject to the interpretive evaluation of an audience (Bauman 2000, 2009). Staged performance prototypically occurs in plays, concerts, and digitally mediated products such as films, tv shows, popular music, social media content, etc. where the priority is to entertain and interest.3

Staged performance brings the metalinguistic and poetic functions of language into focus as semiotic assemblages are transformed into socially transmissible object-signs providing opportunities for reflections of society and self (Coupland 2007; Loureiro-Rodríguez 2017; Loureiro-Rodríguez et al. 2018). Overall, the study of language performance is part of a methodological turn in sociolinguistics towards a social constructivist paradigm emphasizing speaker agency over the conditioning effects of an assumedly external social structure. Regarding the emerging interest in the language of popular music (termed artistic performance speech, hence APS, by Hayes 2022), Coupland (2011) exhorts a view that attends to the specific conditions under which the form of the performance is stylized for the stage, as organized according to genre norms established within histories of prior performances that also take into consideration audience uptake. Bell and Gibson (2011) outlined a sociolinguistic performance framework complimentary to Coupland’s call that brings the variationist methodology in conversation with anthropological conceptualizations of agentive practice to address questions regarding the semiotic significance of linguistic form as performed in APS.

The sociolinguistics of stage performance assumes agentive, intentional representations of language and other modalities against a semiotic backdrop of existing meanings, associated forms, and previous enactments (Gibson and Bell 2012). Indexicality (Silverstein 2003) is at the center of this framework, wherein present meaning-making acts draw their significance from established meanings of parallel past acts. Indexicality is an associative process, in so that indexes do not directly resemble nor reproduce their referents, but reference them through co-occurrence, much how smoke indexes fire (Silverstein 2006). Linguistic variation acquires socioindexical value (Mack 2011; Reed 2020) through the co-occurrence and association of variable speech with perceived categories of speaker-types or specific speech events. First-order indexes (i.e., indicators) are dialectal variations that distinguish geographic locales and do not typically accommodate according to audience or setting. Second-order, higher-order, or n + 1st order indexes (i.e., markers) are the subsequent values ascribed to forms that emerge when linguistic variables and the styles they comprise become anchored to exemplary images of social personhood beyond geographic affiliation through processes of enregisterment (e.g., stances, biographic or fictional individuals, social roles or members of iconic social groupings, and characterological attributes).

Asif Agha’s (2003) research on the enregisterment of British Received Pronunciation illustrated how n + 1st order indexes emerge as a gradual, institutionalized sedimentation of habits of speech production and perception, though he does not discuss this phenomenon in Silverstein’s terms. Agha (2005) discusses stylistic varieties typified as indexing enactable and embodied sociocultural values and positionalities in terms of ‘enregistered voices’.4 Enregistered voices are discursive formations differentiated by language users that have been socialized to associate a stylistic variety with a particular metapragmatic stereotype through sustained interaction with speech-typifying discursive artifacts (e.g., face-to-face interactions, public utterances, literary works, etiquette manuals, appropriateness-based pedagogical models, institutionalized codifications like dictionaries, cartoons, music, etc., see Agha 2003). The social life and domain of enregistered voices is mediated via acts of role alignment wherein speakers engage with the characterological figures indexed by typified speech in ways that either reinforce, challenge, or change the stereotypic value of the register (Agha 2005, p. 49). Collectively, enregisterment and the notion of an indexical order offer a view of speakers’ linguistic repertoires as vast sets of semiotic resources indexing a wide range of socially recognizable and engaging meanings in the situated practice of contextualization (Gibson and Bell 2012).

Eckert (2008) offered the concept of an indexical field to bring the works of Silverstein and Agha into conversation with the variationist literature. Eckert’s view of the social (i.e., interspeaker variation) as a meaning-making enterprise tandemly sourcing and drawing from the stylistic (i.e., intraspeaker variation) challenged the Labov (1972) view of the social as externally fixed wherein style-shifts are treated as responses to attention paid to speech along a formality continuum (cf. Mahl 1972). Rather, she argues that the demographic correlations identified in the earliest sociolinguistic surveys stemmed from language users’ identifications (i.e., role alignments) with the characterological qualities associated with the enregistered voices populating their social realities (p. 465). The fact that the same linguistic variables exhibited multiple experiential and demographic stratifications suggested that variable meanings were not static, but interrelated and subject to shift according to topic, setting, and audience. Eckert’s framework establishes the meanings ascribed to intraspeaker variations as the products of a feedback loop, constantly in flux and adapting according to the context and role alignments presented with each new use (Bell’s Audience and Referee Design, see below). As such, the indexical field outlines the semiotic backdrop against which language users actively stylize their language to achieve identity goals (e.g., acts of identity, see Le Page and Tabouret-Keller 1985), whether of conformity or negotiating new identities and meanings (Bell and Gibson 2011).

The act of stylizing language for stage performance ultimately draws from the various form-meaning associations populating the indexical field established over time via enregisterment. As such, stylization can be viewed as the agentive adaptation of another’s voice (Rampton 1995), or the projection of personae from known cultural texts to an acculturated audience (Coupland 2007). While enregisterment and indexicality can account for how linguistic forms acquire socio-semiotic meaning, they do not account for why speakers manipulate variation in response to conversational and situational stimuli. Bell’s (1984) Audience and Referee Design model provides a person-oriented framework that fits well with this sector of the constructivist paradigm: if enregisterment provides the semiotic backdrop against which current meaning-making acts of languaging draw their indexical significance, then Audience and Referee Design represent the potential motivations for stylizing speech in alignment with enregistered voices according to context.

The model distinguishes between responsive (Audience Design) and initiative style-shifts (Referee Design, see p. 196). Responsive shifts are instances where language users pattern linguistic markers in response to topic, setting, and the potential need to accommodate for interspeaker variation (Giles 1980). This includes instances where a speaker shifts from their casual speech to the register socialized for use in public speaking or the service industry. Contrastively, initiative shifts are speaker attempts to enact affective change on a situation, usually by shifting towards the speech of a group (i.e., the referee) with which they wish to identify, as is common in mass media (Bell 1984; Gibson and Bell 2012). The two are “complementary and coexistent dimensions of style, which operate simultaneously in all speech events” (Bell 2001, p. 165). The APS used to perform popular music can exemplify both responsive and initiative shifts. Certain variations in sung speech will be responsive to the phonation demands of singing, while others adhere to the traditions of performances past in attempt to satisfy consumer-audience expectations. A performer may also present an initiative shift as an active role alignment with the social or characterological figure enacted in the performance for effect (see below). Largely, the literature engaging the sociolinguistics of popular music has engaged the latter, with an emphasis on dialect-indexicalities.

2.2. Singing Voices, Place, and Referee Design

Trudgill’s (1983) pioneering evaluations of “Americanisms” audible in the discographies of popular British singers from the 1960s–1970s was the first to apply sociolinguistic methods to the study of singing accents. He contemplated the use of dialectal features associated with American varieties of English (AmE, termed the “USA-5” by Simpson 1999) as transnational appropriations. These performative choices based in indexicality were viewed as artists’ attempt to model their linguistic behavior on the American-based voices dominating the recorded music industry at the time (Trudgill 1983, pp. 143–46). The subsequently inspired work opted for a dialectal–territorial indexical focus, examining the adoption of AmE features by non-U.S.-based performers with the assumption that the prescriptively AmE phonology operates as a first-order index (Coddington 2004; Duncan 2017; Konert-Panek 2017; O’Hanlon 2006). Coupland (2011, p. 574) outlined three general limitations to this view in pursuit of a more holistic inquiry of singing voices as relating to place.

Firstly, Feld’s (1996) acoustemology of place is offered to account for the multimodal enactment of cultural meaning, as sounds, rhythms, and activities can (in)directly evoke notions of cultural spaces and place in ways that outweigh dialectal indexicalities. Any sociolinguistic analysis of performance must necessarily involve attention to all linguistic and non-linguistic modalities through which a performance is enacted, not only the APS, to adequately detail the significance of stylistic variation. However, surveys of APS samples may be helpful in identifying the distributions of said stylistic practices. Secondly, singing cannot be viewed simply as a performed iteration of spoken variation, as there are significant differences across the two communicative modes according to phonation and stylistic demands (e.g., an established melody line, indulgence in pitch and syllabic variations, see McCrea and Morris 2005, 2007). A sociolinguistic analysis of APS ultimately requires an examination of the same artists’ spoken and sung speech to differentiate those features and forms representing responsive shifts to the context of singing from those representing initiative shifts as acts of identity (Bell 1984).

Gibson (2010, 2019) and Gibson and Bell (2012) exemplify the importance of such a comparison. This cumulative work on New Zealand singing accents provides a compelling case that the pronunciation style classified as approximating AmE in earlier works is not an appropriation, but a general style-norm for various genres of anglophone popular music that acts as artists’ singing default. By comparing the USA-5 and other features not widely associated with AmE in singers’ APS, spoken recitals of lyrics, and subsequent interviews, this line of inquiry has repeatedly shown that the shift towards prescriptively AmE phonology occurs in singing with all variables, not only those widely acknowledged as salient markers of AmE dialects. This is the case even in the APS of artists who make explicit commentary on intentional, conscious effort to avoid AmE variables and sing in a New Zealand dialect. Here, the shift from singers’ spoken to sung speech is responsive to the act of singing rather than an adoption or appropriation of another voice.

The third general limitation concerns voicing beyond vocal and linguistic styling as detailed in Frith’s substantial work on the semiotics of popular song (e.g., Frith 2002). Accordingly, genre norms specify the interactions between vocal, linguistic, musical, instrumental, corporeal, and contextual parameters. Frith (2002, pp. 187–202) suggested four salient dimensions of voice present in these interactions: (1) the voice as musical instrument; (2) the voice as body; (3) the voice as person; and (4) the voice as character. The first dimension includes instances where singers temper the rhythmic structure of songs with scripted or spontaneous, instrument-like vocalizations (e.g., placeholders in rap, see Bradley 2017; Sarkar and Winer 2009). The second attends to the combination of the gestural and the vocal in projecting embodiments characteristic of a particular genre (e.g., the hardness associated with rap, see J. Morgan 2012). The third and fourth dimensions relate specifically to Coupland’s (2007) concept of persona management, and are particularly salient in cases of referee design.

Coupland (2001, 2007, 2011) describes public media performance like popular music as a field of stylizing creativity with socially transformative potential involving the building of a performative display via a recontextualization of earlier versions of the same performance organized by genre (Bauman and Briggs 1990). Briggs and Bauman (1992, p. 147) view genre as quintessentially intertextual, with gaps between present and past performances maximized when creatives innovate away from norms and minimized in claims of an authority to perform in conservatively normative ways. Frith (2016) and Negus (1998, pp. 361–63) discuss this intertextuality in terms of genre-worlds, collections of normalized and recognizable audio and visual esthetics (e.g., voice, place, body, instrumentation, dialect, thematic content, characters, etc.) that come to shape consumer expectations regarding potential (and stereotypic) referees that relate to both voiced persons and characters. Genre-worlds therefore provide ideological templates that function as interpretive frames within which voice and place meanings draw their significance. They therefore must be viewed as orienting frameworks for the production and reception of discourse rather than definitive predictors of style (Coupland 2011; see Hanks 1987).

Importantly, there is no taxonomy of genres, and classification is largely subject to market interest. However, once these classifications are established, record label executives do place imperatives on emphasizing and recycling provenly profitable images over artistic innovation (Frith 2016, p. 170). The links that genre labels have to performative styles, cultural spaces, scenes, activities, and characters function as place-constructing resources that locate characters in places that have historical significance to the genre-world (e.g., dancehalls, housing projects, etc.). These in turn become the conventional places where the familiar narratives characteristic of a particular genre-world are played out, including the characterological work undertaken on the part of the performer to voice participants and characters. For some performances, a distinction between character and person may be essential, and therefore the reproduction of sourced music traditions (including APS) might have more to do with staking a claim or engaging imaginatively with the salient scenes made available by that music rather than an appropriation of identity.

Contrastively, there are genres (e.g., rap) that prioritize the fusion of person and character in the sense that the audience interprets and accepts the performer as singing in their own voice. Ultimately, singers can do characterological work to enact personae associated with different genre-worlds that are unclaimed by their own person (Bassiouney 2017). Person–character distinctions have garnered significant interest by scholars interrogating the appropriation of raciolinguistically enregistered voices by artists who neither demographically belong to the group associated with the typified repertoire nor exhibit competency in the register (i.e., crossing, see Rampton 1995). This research avenue critiques how such stylizing acts facilitate the commodification of identity in pursuit of authenticated artistic affiliations with racialized music forms and increased marketplace profitability.

Perhaps the most publicly covered contemporary case of popular music referee design exemplifying the interplay of enregisterment and initiative stylization as pertains to musical world-making (Goodman 1978) is that of the white Australian rapper, Iggy Azalea. Azalea consistently (and seemingly correctly) deploys African American English (AAE) phonology and morphosyntactic features in her music despite her backstage and interview speech exhibiting none of these elements (Eberhardt and Freeman 2015). Hip-hop is an autonomous market (Park and Wee 2008) that blurs the lines between voice as person and character by centering authenticity and individuality as two core practices of performativity (Pennycook 2007, p. 14). Authenticity (“keepin’ it real”, see Cutler 2003) is generally constructed through fidelity to the rapper’s person and community origins, including dialect. Individuality is constructed through ritualistic and constant rearticulations of socially negotiated meanings that have become embodied signifiers of identity (Madison and Hamera 2006, pp. xviii–xix), such as artists’ capacity to improvisationally layer individualized lyrical meter over backing instrumentation (called ‘flow’, see Bradley 2017). As a music tradition with origins in predominantly African American communities, the normative APS style of hip-hop is closely tied to AAE (Smitherman 1997, 2006).

Within the global hip-hop collective, The Hip-Hop Nation, it is not uncommon for rappers of diverse sociocultural and linguistic backgrounds to display initiative shifts towards an artistic referee by incorporating tokenized elements of AAE enregistered with rap into their regional vernaculars (termed the Hip Hop Nation Language, hence HHNL, see Alim (2002, 2009)). The integration of HHNL by non-AAE speakers in the context of a rap performance operates as a voicing strategy anchoring the performance within the performer’s local referee(s) while engaging the genre-world’s theologies of resistance as tailored to the needs of their respective communities (Pennycook 2003). As a normative, stylizing practice, HHNL is viewed as a means for expressing global solidarity among connective marginalities through musical expression (Alim et al. 2009; Stær and Madsen 2015).

While the use of HHNL is celebrated, the use of an inauthentic APS that exaggerates tokenized elements of AAE in place of adhering to the pillar of authenticity is met with criticism from fans and scholars alike. Eberhardt and Freeman consider the inconsistencies between Azalea’s APS and interview speech linguistic minstrelsy (Bucholtz and Lopez 2011) given that she contextually relies on a seemingly authentic use of racially enregistered language to gain access to an artistic movement created by African Americans. From within this movement, she may capitalize on the material rewards of Blackness, despite violating one of the core practices of said movement and all the while reinforcing standards of acceptability linked to whiteness constantly reaffirmed elsewhere in popular culture (Eberhardt and Freeman 2015, pp. 321–22).

Moreover, Azalea’s public rejection of the notion of hip-hop remaining an African American cultural art form reflects the long history of white artists building careers within originally African American genres. These artists typically experience more commercial success than the creative architects of the music in question by playing up stereotypes and reifying ideologies of essentialized Blackness. The visibility of these artists, in turn, erases the native linkages the cultural capital accessed through the music has to non-white groups (Gabriel 1998). Fundamentally, Azalea’s initiative shift towards AAE allows her to profit from the “cultural performativity and forms of survival that Black women have perfected, without having to encounter and deal with the social problem that is the Black female body, with its perceived excesses, unruliness, loudness and lewdness” and “deliberately tries to sound like a Black girl in a culture where Black girls can’t get no love” (Cooper 2014).

The critiques surrounding whites’ crossing to essentialize a grammatically reduced AAE as the tokenized language of rap music is paralleled in commentary surrounding the use of distinctive phonological elements of Insular Caribbean Spanish by part of non-Caribbean Latina/o/x singers performing Latin urban music (urbano latino, música urbana). This industry umbrella category encompasses the various hip-hop and rap-infused stylings of the Ibero-American diaspora, including Latin trap, rap en español, and, most notably, reggaetón.5 Our attention is now turned towards a discussion of the emergence of the Latin music enterprise as pretext for the concept of the reggaetón voice.

2.3. The Latinization of Reggaetón

In many ways, the narratives of reggaetón and hip-hop as artistic forms of Afro-diasporic resistance to oppressive colonial structures sequestered and consequently whitened by the music industry’s dominant cultural sectors are one and the same. Puerto Rican rappers from New York City brought hip-hop theologies and performativity to the working-class neighborhoods and housing projects (caseríos) of San Juan, Puerto Rico, performing rap in Spanish over dancehall and reggae beats popularized by Jamaican and Afro-Panamanian artists while adapting the content of these works to address the unique conditions of socioeconomically and politically subjugated Afro and working-class Puerto Ricans (Rivera 2011; Rivera-Rideau 2015). Given the absence of such themes in other forms of popular music, reggaetón quickly found widespread popularity among the working-class youth (Flores 2009, p. 164) despite rampant censorship attempts premised on the assertion that engagement with these “tainted musical forms is to subjugate oneself to moral corruption and to tear at the discursive fabric of the chaste, enlightened, rational, and civilized national body” (Rivera 2014, p. 89). Ironically, these initiatives, and their subsequent coverage in the press, only expanded the genre’s underground popularity. As a result, conflicts between the social elites and the reggaetoneros did not persist long into the twenty-first century.

The turn of the century witnessed reggaeton transition from marginal artistic expression associated with foreign and local images of Blackness to a cultural product proudly recognized and embraced as representatively Puerto Rican.6 Reggaetón crossed over to U.S. media outlets following the widespread popularity of Daddy Yankee’s (2004) album Barrio Fino, with the track “Gasolina,” quickly climbing the Billboard charts and gaining routine play on English-language radio and television during what is widely termed the “Latin Boom,” (Alpert and O’Neill 2015). This is understood as the period where white-passing artists such as Ricky Martin, Enrique Iglesias, and Shakira—who had already experienced success in their local industries—were releasing English-language albums and asserting their place in the U.S. and global popular music charts. The music industry—driven by market imperatives to uphold the racial status quo—relies heavily on assumedly self-evident categories of rigidly defined gendered, ethnonational, and racial identities in the construction of its genre-worlds (Cepeda 2018). The creation of the ‘Latin’ category saw marketing executives work to discursively flatten the cultural and ethnic distinctions between Hispanic and Latina/o/x artists by advancing catch-all identity labels intended to promote products consumable by Latin Americans and desired by the global Other (Cepeda 2010).

Lawrence and Clemons argue that the heterogeneous nature of Latinx/Latin American histories required a racialization “not of individuals, but of the category,” (Lawrence and Clemons 2022, p. 2). Notably, this cultural homogenization in the construction of the Latin genre-worlds was facilitated by the implementation of a dominant “Latin look” characterized by European features, light complexion, and straight brown hair in both Spanish- and English-language media (see Dávila 2008; Goin 2016). Torres-Saillant (2003) argues the dominance of this esthetic aided the racialization of Latinas/os/x as an ethnorace existing somewhere in the intersections of European, African, and Indigenous ancestry. The look’s allusion to a history of miscegenation, when paired with the raciolinguistic enregisterment (Rosa and Flores 2017) of the Spanish language as the language of Latina/o/x populations, discursively constructs Latinness as a linguistically based, mutually exclusive experiential category distinct from U.S. notions of whiteness, Blackness, and Indigeneity (see Candelario 2007, p. 225; Clemons 2021). It is perhaps unsurprising, then, that the contested identity boundaries within Latinidad are largely informed by discourse grounded in colonial formations of power that consolidate normative articulations of a pan-ethnic physiognomy and language (Lawrence and Clemons 2022).7

Following the U.S. crossover and the growing popularity of ‘Latin’ music, reggaetón was quickly recruited by the culture industry (Adorno and Horkheimer 2002) and reductively covered as a Puerto Rican innovation marketable as Latin hip-hop (see Marshall 2009, p. 19). Reggaetón icons were racialized as akin to yet (linguistically) distinguishable from U.S. Blackness, in contrast to pop artists like Ricky Martin, presumably due to their artistic and stylistic connections with rap (Rivera-Rideau and Torres-Leschnik 2019, p. 4). This crossover expanded reggaetón to increasingly diverse audiences, opening itself to processes of racialization across the Americas that now position the genre as the sound of Latina/o/x groups. The genre is often theorized as embodying a sonic benchmark distributing a set of values allegedly related to Latin American heritage for consumption by the global Other (Rivera-Rideau 2019; Solis Miranda 2022). Ultimately, the U.S.’ Latinization of reggaetón tropicalized the soundscape as linked to Latinidad in ways that required an iconic detachment from the genre’s initial Africanity posing potentially threatening obstacles for marketing (Rivera-Rideau 2019, p. 476; Solis Miranda 2022, p. 519).8

Research engaging the notions and boundaries of Latinidad requires careful attention to the expressions of different life experiences mediated by Latin American origins and determined by the imposition of and proximity to whiteness, as well as the attempt to erase any trace of Africanity or Indigeneity (Candelario 2007; Cepeda 2018; Rivera-Rideau 2019; Solis Miranda 2022). The most accentuated changes in the reggaetón–Latin urban pipeline—the dominant shift towards a pop-oriented sound, the erasure of political lyrics in favor of sexuality and corporeality contextualized within the lyrically constructed “Latin party”, and the increased featuring of white(ned) artists who more closely embody the Latin look—have largely resulted from the culture industry’s implementation of safety frames mirroring colorism and broader processes of marginalization that remove the presence of Afro-Latinos in dominant articulations of ‘authentic’ Latinidad (Rivera-Rideau 2019, p. 470).9 Over the years, what has become increasingly clear is that reggaetón is more consumable and profitable when delivered by racially ambiguous performers.

Moreover, the raciolinguistic ideologies resultant from commodifying the Spanish language as intimately tied with Latinidad and its associated musical categories has created a political economy in which “those who already hold massive positions of privilege vis-à-vis Latinx populations (i.e., Spaniards) are given increased visibility while all others are reduced to anecdotal minorities” (Lawrence and Clemons 2022, pp. 4–5). The imposition of Latinizing safety frames (Solis Miranda 2022) and the raciolinguistic enregisterment of Spanish as Latinidad have collectively allowed for the commodification and exploitation of the material popularity of the Afro-Caribbean diasporic elements originally circulated through reggaetón as invocations of vernacularity to offer a non-threatening model of urban Latinidad recruitable by racially, ethnonationally, and sociolinguistically privileged artists that has proven to be extremely effective and appealing.

2.4. The Caribbean Blaccent

Despite the consumption of Latin urban music being at an all-time high (Billboard Charts 2020; Spotify Charts 2020; YouTube Charts 2020), there is a noticeable lack of research on the APS of the genre. This is surprising, as the linguistic distinctiveness of reggaetón has been a topic of interest since the genre’s inception. Just as the APS of hip-hop is closely associated with AAE, the APS of reggaetón as inspired by working-class caserío residents’ fidelity to their vernacular register is closely tied to Puerto Rican varieties of Spanish (Rivera 2009). In fact, the retention of phonological features typically stigmatized at the institutional level and within transnational Latinidad discourse—such as the omission of voiceless coronal fricatives in coda—is one of the genre’s defining artistic characteristics (Marshall 2009; Valentín Márquez et al. 2024).

Marshall (2009) notes that the APS of artists producing internationally distributed records endorsed by mainstream labels at the time of reggaetón’s origin was meticulously monitored and edited to erase the use of such features, many of which are considered distinctive of the Insular Caribbean varieties. This was done to approximate a transnationally disseminated and dialectally leveled standard. Operating outside the confines of the mainstream industry, the reggaetoneros were able to resist dialectal erasure and create a space in which transnational raciolinguistic ideologies regarding appropriateness and stigma are inverted (Flores and Rosa 2015). This has been regarded as both a process of disidentification (Muñoz 1999) and a case of antilanguaging (Lefkowitz and Hedgock 2017; Martínez 2016) in that the prioritization of such features directly challenges the hegemony of the peninsular-based variety disseminated as the transnational standard. This legacy holds special significance with the fan base, which dually acknowledges the ‘superiority’ of the transnational standard used in pedagogical settings yet maintains that the language used in rap-infused music in Spanish should be a Caribbean-derived register (Magro 2016, p. 24).

Ureña-Ravelo (2019) notes a parallel to whites’ crossing via the deployment of tokenized AAE to index artistic and vernacular affiliations with the culture of rap music occurring within the urbano enterprise: “non-Black, non-Caribbean artists/people are increasingly adopting Caribbean and African American ways of speaking to market themselves in Black/Afro-Latinx genres and culture”. As noted by Hayes (2022), such critiques have also been taken up by popular Latinx content platforms (e.g., Mitú) who have accused urbano performers such as Becky G, an ethnically Jaliscan artist born and raised in a working-class family from Inglewood, California, for deliberately using a “Caribbean Blaccent” to perform their music and lay claim to the reggaetón soundscape (see Barron 2019; Caraballo 2019; Cardona Pérez 2022; see also Medford 2024). Notably, these accusations overtly and repeatedly mention the salience of coda s-reduction in differentiating Caribbean and non-Caribbean dialects of Spanish. Circulated throughout social media alongside the hashtag #HowDoMexicansTalk, such discourse brings significant attention regarding questions of authenticity, prescriptivism, appropriation, and privilege to the fore of the reggaetón mediascape by the positioning of Becky G’s performative persona as inauthentic. This inauthenticity is substantiated by claims that the singer stylizes her APS to sound Caribbean for profit through an intentional omission of coda -s, rather than retaining the sibilant as prescribed by her Mexican heritage. Ultimately, these fan critiques highlight the significant role that perceived and performed ethnolinguistic fidelity (i.e., the fusion of person and character) occupies in urbano production.

Notably, Becky G publicly denies any intention or use of a Caribbean-based APS (Becky G 2017), despite her interview, backstage, and social media speech more closely reflecting patterns of Jaliscan phonology that are absent in the performances under critique (see Becky G and J-Hope 2022). Her retort emphasizes the person–character merger in claiming her APS to be reflective of her own spoken vernacular, produced from growing up surrounded by friends from various Hispanophone national origins (Becky G 2018b). In discussions of her entrance into the Spanish-language music market, the singer often mentions that Spanish is not her dominant language, and that she acquired competency through family and social connections, explicit instruction, and cultural product consumption. Having been born in 1997, these products included music in Spanish performed by diverse artists, such as the reggaetoneros with whom she would come to collaborate in her professional tenure (Billboard 2018). While Becky G has yet to comment publicly with more details regarding her socialization with Spanish, one may posit that the singer exhibits a degree of bidialectalism acquired in her personal life.

However, it is impossible to assess Becky G’s potential bidialectalism to the extent necessary to refute this possibility without implementing a methodology to control for audience design effects (e.g., Gibson 2019). Moreover, the fact that Becky G is a public artistic persona requires acknowledging the potential differences between Becky G the public characterological icon and Rebecca Marie Gomez the person (despite the discourse emphasizing these two as one in the same) to effectively contend with any artificial stylizations. Recall from Section 2.2 that singers voice multiple characters and personae in the entextualization of their works that are not inherently bound to their own person nor limited to performers’ personal linguistic proficiencies (Christina Aguilera’s most recent Spanish-language album is an exemplary case, see AP Archive 2022; Billboard 2021). While it is not probable that s-reduction constitutes a responsive shift to singing as a unique conversational modality (see below), it is important to recall how the performative norms of a genre (including but not exclusively APS) operate as voicing resources such that artists may initiate style-shifts as part of a broader claim to the activities, places, narratives, and positionalities characteristic of a given genre-world. Just as Iggy Azalea’s “seemingly authentic” use of AAE in her APS showing no remnants in her spoken speech suggests that she is the animator of characterological voices entextualized by lyricists who speak AAE (Eberhardt and Freeman 2015), vocalists can easily acquire the linguistic skillset necessary to situationally enact emblems of social personhood as tied to genre-worlds to which they themselves, without said instruction, would otherwise not have access.

In the case of Becky G, there is little publicly available evidence suggesting the prescriptive Caribbean phonology at the center of these accusations is representative of her everyday speech. Caribbean Spanish is not prevalently spoken in California, with roughly 0.58% of the over 41.7 million people in the state being of Hispano-Caribbean ethnonational origins. In Los Angeles County—one of the few areas with a notable Puerto Rican population and the county of birth of Becky G—the Hispanic-Latina/o/x population constitutes approximately 49% of the total (N = 10.5 million). Notably, 74% of this demographic is of Mexican ancestry (n = 3.68 million) and less than 0.47% is Puerto Rican (n = 47,349 see The United States Census Bureau: California 2024). Of course, Becky G could have acquired s-reduction as part of her vernacular from speakers of other s-reducing varieties (e.g., El Salvador, 9.1%). These demographics nevertheless highlight the scarcity of Caribbean Spanish in California, wherein it is more likely that her exposure to specifically Caribbean registers and dialects was restricted to media consumption (Rampton 1995; Sweetland 2002).

Furthermore, if the product of bidialectalism, we may expect Becky G’s use of s-reduction to be most prevalent in social spaces where Caribbean identity is highly salient, but this is not the case. Consider her 2019 performance as an invited guest of Bad Bunny (the contemporary leader and historically most visible reggaetón artist) at the Coliseo Roberto Clemente in San Juan, Puerto Rico (a site that holds particular significance for reggaetón performances, see Negrón-Muntaner and Rivera 2007). Here, Becky G’s APS is categorically s-full despite the genre characteristically preferencing Caribbean markers, the performance being set in front of a predominantly (if not exclusively) Puerto Rican audience, and Bad Bunny being categorically s-less (Iván Javier Informa 2019). In a 2018 interview with KQ 105, a Puerto Rican radio station focused on popular music and culture based out of San Juan, Becky G does sparingly exhibit coda /s/ reduction. However, this omission may be triggered by audience design as the hostess “Normi” exhibits minimal sibilant retention (Becky G 2018a). Additionally, Becky G’s overall interview speech, in particular her anecdotal and pensive speech, is once again categorically s-full.

Here it is important to rearticulate the significance of communicative mode in examinations of APS variation (Section 2.2). Coupland (2011), Gibson and Bell (2012), and Gibson (2019) all champion a comparative approach to distinguish responsive and initiative shifts in APS. Interviews—in particular celebrity interviews as a form of public speaking—are considered high-formality contexts. Interviewees typically speak in the register they have been socialized to view as appropriate for such a context. For speakers of Spanish, this register largely approximates the Castilian-based transnational standard informing notions of clear speech (i.e., the idea that each word has a ‘full’ pronunciation correlating to institutionalized orthography). This register is largely devoid of the (stigmatized) features considered distinctive markers of Insular Caribbean vernacularity, such to the extent that it is equated by some as a form of “posh speech” (Bullock et al. 2014; Erker and Otheguy 2016; Roca 2005; Schmidt 2012). Furthermore, public celebrity interviews are typically negotiated prior to recording, and therefore speech analyzed from these sources must be viewed less as representing celebrities’ spontaneous vernaculars and more as a quasi-artificial register in its own right. It is very likely, then, that the audible differences in Becky G’s interview speech and APS come from the use of distinct registers: one for public interviews which more closely resembles the transnational standard and prescriptively Mexican phonology, and another for performing reggaetón which draws on the enregisterment of the genre as quintessentially Caribbean.

Becky G’s responses to the accusations of linguistic minstrelsy against her largely draw on the homogenizing force of Latinidad discourse and the raciolinguistic enregisterment of Spanish as Latinidad (Lawrence and Clemons 2022). Overall, she deflects claims of appropriation by invoking images of pan-ethnic Latinidad, unverifiable anecdotes of an interethnic upbringing, and comments positioning Spanish as inherently “natural” to her Latina self, all the while noting the unexpectedness of “una mexicana cantando reggaetón” that reinforcingly establishes the place indexed by reggaetón in the Caribbean (Becky G 2018a). Moreover, she does not comment on the undeniable experiential differences of speakers of diverse dialects of Spanish in relation to notions of prestige or processes of erasure consequent to histories of colonial power formations and the imposition of essentializing Latinidad discourse. These negations are damaging in similar ways to Iggy Azalea’s public denial of hip-hop remaining an African American cultural art form in that they reflect highly visible subjects’ ongoing participation in and the upholding of the racial status quo by claiming cultural capital through adaptations of performativity and forms of survival perfected by historically sociopolitical marginalities without negotiating the social problems confronted by the racialized subjects whose radical and creative powers produced them. As a result, we have yet to see the full, mainstream incorporation of the racialized communities where reggaetón originated despite the genre being one of the most widely consumed forms of popular music of modernity (Rivera-Rideau 2019, p. 474; see also Acevedo 2019; Nicholson 2019; Roiz 2018).

Furthermore, Becky G has also deflected claims of appropriation as attacks on her Latinaness, stating that her lack of knowledge in Spanish does not take away from her experience as a Latina (Becky G 2017). By claiming that singing in Spanish feels “natural,” she exemplifies a popular strategy deployed by various Latinx artists to stake an ethnolinguistically based claim to Latin American music traditions by positioning Spanish as inherent to the Latin body/person (Cepeda 2010). Importantly, this dismissal of an inauthentic APS accredited to plausibly not speaking “perfect” Spanish challenges the raciolinguistic enregisterment of Spanish as Latinidad. However, this type of commentary also feeds the normative notions of a pan-ethnic experience operationalized by the music industry to homogenize the sociocultural diversity and unique life experiences of speakers of distinct Latin American national varieties of Spanish resultant from those dialects’ colonial histories of translocal prestige and stigmatization.

Likewise, to equate critics’ perceptions of crossing into Caribbean Spanish with notions of potential linguistic deficiency reflects the rampant raciolinguistic ideologies that have been shown to consistently position the Insular Caribbean varieties as the inferior speech of the uneducated, darker masses in comparison with those dialects whose phonology more closely resembles that of central Castille (Alfaraz 2002, 2014; Canfield 1981; Carter and Callesano 2018; Chappell 2018; Penny 2000; Rosa 2010; Toribio 2006; Zentella 2002). By claiming the various varieties of Spanish spoken to her growing up as a homogenous source for her APS, Becky G indirectly exemplifies how all varieties of Spanish are rendered from an industry standpoint as mutually inclusive, recruitable dialects for enacting a Latin performative persona. This inclusivity is often used as justification by fans advocating for the inclusion of Spaniard performers such as Rosalía into the category, and therefore warrants scrutiny.

2.5. The APS of Latin Urban Performance

As stated in the introduction, this research is a continuation of Powell (2024), which sought to track Hispano-Caribbean artists’ use of reduced coda /s/allophones across three distinct genres of ‘Latin’ music. This examination found that artists significantly preference vernacular forms in reggaetón but not in other forms of popular music with minimal rap influence. For example, Ivy Queen—one of the pioneers of the genre widely considered the quintessential reggaetonera and the uncontested Queen of the genre (Báez 2006; Goldman 2017; Rivera-Rideau 2015)—reduces coda /s/ upwards of 90% in her reggaetón performances in contrast to a mere 19% (n = 225) in performances marketed as “tropical” (i.e., salsa, bachata, merengue, and the like). This direction of effect is duplicated in the APS of Cuban artist Srta. Dayana and Dominican performer Natti Natasha. Informed by the fanbase expectations documented in Magro (2016), Powell interpreted these patterns as preliminary evidence of an enregistered stylistic norm for performing reggaetón that draws on the genre’s origins of resistance and prioritization of vernacularity. As reggaetón has since expanded to encompass a broader (non)Latin audience, artists targeted by critiques of appropriation like Becky G may simply be the animators of the normative voices of the genre as entextualized by the production team(s) responsible for engineering their performances. Notably, many of these creatives are Puerto Rican songwriters based in San Juan and Miami, Florida, who therein may place an imperative to adhere to the linguistic traditions of the genre.

Amidst these critiques and the emergence of Colombian artists from Medellín (an s-retaining dialect) as the genre’s up-and-coming dominant demographic, scholars have begun to critically examine the sociolinguistics of the urbano soundscape. This scholarship has widely operationalized coda /s/ reduction as the unit data point, with this marker being arguably the most attested sociophonetic variable in the Hispanic linguistics literature. As all spoken varieties of Spanish have some degree of s-reduction, coda /s/ is often regarded as one of the most robust phonetic differentiators of regional and social dialects (Lipski 1984, 2011). A near categorical absence of /s/ is popularly treated as a first-order index linked to the Insular Caribbean (see Alba 1990; Brown 2009; Brown and Brown 2012; Cedergreen 1978; Hualde 2014; Lafford 1986; Lipski 1995; López Morales 1983; Mack 2011; Terrell 1977a, 1979, among countless others). Because of the enregisterment of s-reduction with Caribbean identity, a pervasive reduction in music paired with the absence of lenited allophones in spoken speech has been taken as evidence of attempts to Caribbeanize APS (see Hayes 2022, p. 28).

Hayes and Luna (2019) compared s-reduction rates in the interview speech and APS of three top-charting Puerto Rican urbano artists, while Hayes (2020) paralleled this structure to examine the Colombian counterparts, all of whom are notably from Medellín. Both studies confirmed a statistically significant difference (p < 0.01) between speech contexts for each artist, though in diverging directions of effect. Puerto Rican artists appear to increase their use of coronal sibilants in comparison with Colombian artists whose naturalistic maintenance is significantly reduced. Hayes (2022) offers a comparative analysis of the speech of the top-charting artists from each group—Bad Bunny and J Balvin, respectively—in which she notes the two reduction rates converge towards a middle ground. While Bad Bunny’s s-retention increases from 4% to 20.64%, J Balvin’s high use of spoken [s] (80.79%) significantly drops to 49.83% when singing. Similar findings are reported in Martínez Kane and Papadopoulos’ (2021) examination of Rosalía’s APS and interview speech, finding that the singer retains nearly all sibilants in spoken interviews yet reduces these near categorically when performing reggaeton. These works support the claims advanced in the current research regarding a normative register that draws from the covert prestige of Caribbean phonology enregistered with the genre as an integral component to the reggaeton voice.

Notably, Hayes (2022) found certain songs featuring artists of Caribbean heritage (e.g., “I like it” featuringDominican–American rapper Cardi B, Bad Bunny, and J Balvin (2018)) to exhibit uncharacteristically low occurrences of [s] for s-retaining artists (7.14%, 1 token). She suggests artists may converge towards the regional patterns of their collaborators as conditioned by both audience and referee design. As Latin urban music is a highly collaborative industry, these findings beg the question regarding the nature of s-reduction as utilized by non-Caribbean artists: are these stark contrasts a responsive shift conditioned by genre-world norms, a marketing strategy for promoting tracks geared towards a Caribbean-centered audience that encourages initiative shifts towards reggaetón referees, a case of appropriation for profit, or a combination thereof? This research furthers this line of inquiry by pairing acoustic analysis and sociolinguistic statistical models to offer quantitative comparisons of the incidence of lenited /s/ allophones in the spoken and sung speech of four top-charting mujeres del género as a point of departure towards addressing these questions.

3. Methodology

3.1. Research Questions and Hypotheses

Eberhardt and Freeman (2015) examined Iggy Azalea’s use of a racially marked feature across spoken and sung speech contexts and compared the overall rate of occurrence with other white and African American rappers like Eminem and Eve. Hayes (2022) indirectly paralleled this structure by comparing the rate of lenited coda /s/ allophones used by Caribbean and non-Caribbean reggaetoneros. This work recruits these same methods, in particular those mapped out by Hayes, to offer further comparisons of potential style-shifting evinced in the discographies of four artists of varying ethnonational and sociolinguistic backgrounds not widely covered in this literature: Anitta, Becky G, Karol G, and Natti Natasha.

The decision to emphasize the APS disseminated through the performances of some of the most visible women of the genre is based in a desire to highlight an underrepresented demographic, as the pervasive hypermasculinity of reggaetón esthetics and content is one of the genre’s most salient features.10 Furthermore, the music distributed via these artists’ discographies is representative of reggaetón as produced following the implementation of the Latinizing safety frames detailed in Section 2.3. Solis Miranda (2022, p. 518) importantly notes how the image of una mujer del género has transitioned to give form to a gozadera (party goer/enjoyer) model that predominantly depicts urbana women as “someone who is easily aroused, seems ‘difficult’ but who is at the service of male sexual desires, has a slim body with large breasts and buttocks, and knows how to dance” (see “Bichota”, Karol G (2020) and “Gata”, Anitta (2022b)). This readily audible content positions these discographies as exemplary sites for critical examinations of the Spanish deployed in the post-Latinization era of música urbana.

While the popularity of música urbana is often attributed to male performers, the iconicity of these women and their overall contributions cannot be understated. In comparison with Bad Bunny and J Balvin who, respectively, have more than 66.38 and 55.96 million monthly listeners on Spotify and 47.9 and 34.4 million YouTube subscribers, Karol G racks up over 47.36 million monthly listeners and 36.5 million YouTube subscribers, ranking as the highest-streamed female Latin urban artist and one of the most-streamed Latin artists of today.11 As a collaborative enterprise, these women have played opposite the biggest names of reggaetón and achieved symbolic status as matriarchal embodiments of the gozadera image.12 These women often collaborate with one another, and have publicly commented on the significance of camaraderie to advocate for feminine solidarity in a male-dominated industry (Billboard 2019).13 Furthermore, the promotional images of the singers under study are all managed by one of the three U.S.-based lead distributors of ‘Latin’ music and are therefore actively engaged in the very industry that celebrates Latin ethnocultural homogeneity.14

Moreover, the sociolinguistic diversity and unique experiences of these women’s entrances into the Latin market allow for parallel comparisons to the studies informing this work, in that the participants are representative of distinct sociolinguistic backgrounds operating within a singular performative context. Natti Natasha (Santiago de los Caballeros, Dominican Republic) and Karol G (Medellín, Colombia) both represent women whose careers began in the Latin urban industry. Their APS has been examined in previous studies (Hayes 2020 and Powell 2024, respectively), providing an opportunity for corroboration. Additionally, the prescriptive regional patterns of s-reduction allows for an examination of accent-shifting from the two trajectories covered in Hayes (2022). Becky G (Inglewood, USA) and Anitta (Rio de Janeiro, Brazil), on the other hand, achieved fame in a non-Spanish language industry (English and Portuguese, respectively) before crossing over to the Latin market in 2016. While Becky G is reportedly a simultaneous bilingual speaker of Spanish (Section 2.4 above), Anitta did not learn Spanish prior to her entrance into the ‘Latin’ industry, and her APS is therefore directly reflective of her integration into the genre (Anitta 2022c).15

The present study attempts to satisfy Coupland’s (2011) methodological considerations detailed in Section 2.2 by conducting a quantitative comparison of the incidence of lenited sibilant allophones across spoken and sung speech modalities. The publicly accessible discographies of the artists under examination accessed via YouTube.com (Premium membership) and Spotify.com (Premium membership) are recruited as major sources of data collection. The following research questions were thus developed to investigate the degree of shifts in pronunciation evinced in these artists’ discographies to identify salient commonalities and differences:

• Research Question 1: What is the overall reduction rate of coda /s/ exhibited in a sample of interview speech of the top charting female urbano performers?
• Research Question 2: How does this rate compare with the incidence of lenited /s/ in the artistic performance speech of these same artists? Do these values suggest any degree of style-shifting?
• Research Question 3: What are the performative and linguistic factors conditioning this variation?

The third research question acknowledges the previous accounts suggesting that artists’ reduction rates may be influenced differently by Audience and Referee Design, as well as studies that note an increased incidence of stigmatized features towards the present, plausibly resulting from the post-Despacito popularity (Cobo 2020) which further reinforced the genre as intimately tied with Puerto Rican culture and semiotics (see Powell 2022). Two additional research questions are offered to specifically address these observations:

• Research Question 4: Do artists maintain lenition patterns when performing as solo artists in comparison with collaborating with artists of (non)Caribbean origins?
• Research Question 5: Are reduction rates consistent across artists’ discographies? Is there a noticeable shift in pronunciation alongside time?

Taking into consideration established findings suggesting the existence of a formal interview register that is distinct from the register characteristic of reggaeton in terms of [s] occurrence, this study anticipated the production of results as outlined in the following hypotheses:

Hypothesis 1:

Coda /s/ reduction will be relatively infrequent in the interview data and markedly more frequent in the APS data.

Hypothesis 2:

Coda /s/ reduction will pattern similarly in the APS samples in ways suggestive of potential initiative shifts.

Hypothesis 3:

Coda /s/ reduction will be relatively infrequent in the earliest songs and increase towards a collective uniformity with time.

3.2. Materials and Methods

3.2.1. s-Reduction as a Discursive Frame

As in Powell (2024), values exhibiting similar patterns from multiple speakers for a singular performative purpose—in a context that prioritizes homogenization—would be suggestive of an emergent standard performative style, a reggaetón voice, enregistered with the genre and fueled by the images of vernacularity established by the genre’s pioneers. Cross-dialectal comparisons of reduction patterns typically rely on a five-stage categorization system that conceptualizes dialects as more-or-less advanced along a weakening pathway towards categorical s-lessness (Figure 1, see Lipski 1999).16 This system suggests lenition begins in unstressed preconsonantal coda before extending to prevocalic and eventually intervocalic and onset cases of deletion. Scholars typically place the Insular Caribbean varieties in Stage 3, as speakers favor reduction in preconsonantal contexts (Stage 1, see Barrutia and Schwegler 1994; Brown 2009; Terrell 1977b) and frication in prevocalic environments. This is often attributed to the Spanish resyllabification system which articulates word-final, prevocalic /s/ as onset (Stage 4, see Harris 1983). This categorization of macro-style is adapted to offer a comparative outline applied to the reduction patterns evinced in the participants’ APS to conceptualize the degree to which these patterns mirror those associated with Caribbean dialects of Spanish.

Hualde (2014, pp. 154–57) describes four allophones of lenited /s/ widespread throughout the Caribbean, among them, standard alveolar frication los mismos [los ˈmismos], glottal frication or aspiration [loh.ˈmih.moh], elision [lo ˈmimo], and a glottal stop [loʔˈmihmoʔ]. Lenition is reportedly strongest in the Dominican Republic, whose speakers tend to favor [-] contrasted to speakers from Cuba and Puerto Rico who are attested as favoring [h] (Hualde 2014, pp. 159–60; see also Navarro Tomás 1948; Terrell 1977a). However, in transnational discourse, both glottal fricatives and elisions are viewed as nonnormative speech behavior (Erker and Otheguy 2016, p. 135) and are often used as justification by speakers of s-retaining dialects as to why Caribbean speakers “don’t get the words right” (see Rosa 2010, p. 129). Glottal stops are not found in the Cuban or Dominican varieties (Valentín Márquez 2022; cf. Satterfield and Benkí 2019), and are proposed to be the result of language contact in Puerto Rico with non-prestigious varieties of English.17

As briefly mentioned in Section 2.4, contemporary transnational divisions of dialectal prestige rampant in Latinidad discourse stem from institutional enregisterments juxtaposing phonology with hierarchies of social personhood that correlate to how closely the given dialect or register reflects the Spanish spoken in Castille. Penny (2000, pp. 148–49) affirms how, historically, the division between prestigious and stigmatized Latin American dialects has corresponded to the “degree of closeness of contact between central Spain and the specific American area concerned.” Those areas showcasing high political and economic importance to the colonial empire (e.g., the Andean regions of Colombia) are contemporarily celebrated as authentic, “correct” Spanish while, contrastively, dialects from areas with high degrees of Andalusian, Indigenous, or African presence (e.g., the Caribbean) are often subject to negative commentary regarding intelligibility and grammaticality (Hayes 2022, pp. 23–26). A raciolinguistic reading of these attitudes emphasizes the centrality of hierarchies of racial legitimacy mapped onto notions of linguistic legitimacy through enregisterment in modern subject formations (Rosa and Flores 2017, p. 622). Engaging this framework through the social constructivist sociolinguistic paradigm illustrates how ideological connections between linguistic form, style, and social-being are not necessarily fixed according to any demographic stratification, but contingent on context and socialized perception (see also Alim et al. 2016).

The indexical field (Eckert 2008) of lenited variants comprises multiple potential, interrelated, and contextually specific meanings, preventing any view that attitudes towards s-reduction in the Caribbean are consequent to local demographic stratification. This is in part detailed in scholarship that has consistently returned reduction as occurring at near categorical rates across all social strata. Chappell (2018) uses this observation to argue that the widespread, unstratified distribution of [-] and [h] across Caribbean societies warrants a view of lenition as a category of sociolinguistic markers rather than dialectal indicators. This view emphasizes how both [h] and [-] are collectively viewed in transnational discourse as simultaneously nonnormative speech behavior and characteristic of Caribbean origins (Erker and Otheguy 2016, p. 135; Schmidt 2012, p. 190). Moreover, speakers both within and outside the Caribbean reportedly associate [s] with higher evaluations of intelligence, work ethic, confidence, and snobbishness, and contrastively equate reduced allophones with blue-collar professions, minimal access to educational opportunities, and low work-ethic. Additionally, speakers from the Caribbean recognize the overt prestige ascribed to prescriptively peninsular phonology, though they maintain that their regional varieties constitute the preferred register in all affective categories. This is markedly contrasted to the evaluation of the other Caribbean varieties, which are similarly devalued with discourse targeting Caribbean Spanish as a whole (Mojica de León 2014, pp. 1304–11).

This view is further supported in Mohamed’s and Muntendam’s (2020, p. 392) account of Puerto Rican’s tendency to restrict the use of coda sibilants to formal settings and to characterize the variant as indicative of an elevated socioeconomic status in direct juxtaposition to [-], which is in turn associated with informal conversations and typified as characteristic of the speech of the lower classes, racialized subjects, and in particular, Dominican migrants. It also resonates with Toribio’s (2006) comment regarding the popular view within the Caribbean that “the ‘best’ Spanish variety approximates the Castilian norm, and the ‘worst’ variety is spoken by those who, by dint of birth or social circumstance, are believed to be influenced by the African substratum” (p. 134, emphasis added). In the Dominican Republic, cases of parodic hyper-articulation, colloquially recognized as habla fisno (i.e., “speaking faaancy” termed intrusive-s by Bullock et al. 2014, p. 21) are audible as speakers produce non-etymological /s/ in word-medial position. The authors emphasize that users of habla fisno are aware of where to produce historical /s/, but choose to suppress the feature in favor of [-] given deletion’s significance to Dominican identity construction. Similarly, Valentín Márquez (2015) elaborates on how the weakening of syllable-final phones and nonstandard articulations is a celebrated practice in situations where Puerto Rican in-group membership is attested, avowing that the use of stigmatized features is often seen as indexical of ethnic pride (Ortiz-López 2022).

The reduction of /s/ as a stylistic variant must be viewed as a raciolinguistically enregistered voice anchoring exemplary linguistic behavior to celebrated/pejorated images of social personhood (Carter and Callesano 2018; Rosa 2010). This can account for why the presence of many distinctive elements of Caribbean Spanish were systematically erased from the APS of mainstream popular music productions at the time of reggaetón’s creation (Marshall 2009), even though many of these genres (e.g., bachata, Sellers 2014) were also created in predominantly working-class and Afro-Caribbean communities. Moreover, in acknowledgment that reggaetón and the broader urbano enterprise constitute the most visible mediascape in which Caribbean phonology and vernacularity are counter-valorized as celebrated and legitimated linguistic practice, it is also warranted to treat the presence or absence of [s] as a highly salient component of ideological and characterological work extending beyond place-based attachments correlating to any demographic stratification.

3.2.2. Data Collection and Preparation

To facilitate a comparison of lenition rates between spoken and sung contexts, two corpora of /s/ tokens were created. Firstly, a small sample of participants’ interview speech was collected to establish a real time instantiation of the Spanish they utilized in mass media between 2018 and 2022 (Poplack and St-Amand 2007, p. 717). This acts as a baseline of comparison for their APS (RQ1–2). Interviews were collected from YouTube using artists’ names alongside entrevista (interview) as search criteria with view count and duration set as selection criteria. The examined interviews feature the highest view count and a duration of at least five minutes (Appendix A).

The songs informing the APS corpus were extracted at random from the discographies accessible on artists’ verified Spotify accounts using the shuffle function.18 Three song categories were established prior to collection and organized as solo performances and duets/ensembles featuring credited, collaborative performing artists of varying ethnonational origins. The latter was set as a binary between featured artists of Caribbean and non-Caribbean origins (RQ4). The first eight tracks representing these categories released between 2015 and 2022 were selected for examination (RQ5), rendering a total of 24 performances per artist (96 total performances) distributed through 92 unique tracks constituting approximately five hours of music (see Appendix B).19

As lenition is a gradient process (File-Muriel and Brown 2011, p. 225) better documented via acoustic data rather than researchers’ subjective audible perceptions (Widdison 1991, p. 80), each audio was downloaded as a WAV file and opened in Audacity (Mazzoni and Dannenberg 2021) to cut files in preparation for acoustic analysis in Praat (Boersma and Weenink 2021). Interview audios were cut to feature a random segment near the middle of the interview,20 whereas song files were cut to only feature verses containing the variable contexts according to lyric transcriptions collected from online sources following multiple rounds of listening and editing to corroborate accuracy. Following Caillol and Ferragne (2019), utterances were manually annotated to a text grid with source words segmented in a separate grid tier and an additional annotation of the fricative itself (Figure 2 and Figure 3). Excluding tokens directly adjacent to other instances of /s/ (e.g., me harías sufrir) and implementing a form-count limit of three tokens per audio, the first 25 examples of interview coda /s/ were extracted to provide a sample of 100 sibilants per artist, totaling 400 analyzed spoken tokens.

The same exclusions guided the extraction of the sung sample, including an additional exclusion to only consider the first occurrence of tokens featured in repeated song sections (e.g., the chorus), which yielded a total of 1840 sung tokens for inspection. In light of the conflicting recommendations regarding the preparation of music sound files for acoustic analysis and heading expert advice, this study circumvented any acoustic editing in preparing vocals for analysis to reduce the potential risk of audio-editing software compromising audio quality (Hayes 2022; cf. Caillol and Ferragne 2019). Instances where frication is hidden or skewed due to layered instrumentation (n = 42) were discarded from the analysis, rendering an APS data set totaling 1798.

The center of gravity (COG) and the standard deviation of each token were recorded in an excel spreadsheet to classify and track variant distribution (

Table 1. Variant Distribution. Artists’ most frequent variant (%) are in bold.

		[s]	%	[h]	%	[-]	%	Total
	Anitta	60	60	11	11	29	29	100
Interview	Becky G	78	78	15	15	7	7	100
N = 400	Karol G	65	65	13	13	22	22	100
	Natti Natasha	42	42	4	4	54	54	100
	Total	245	61	43	11	112	28	400
		[s]	%	[h]	%	[-]	%	Total
	Anitta	143	40	15	4	198	56	356
APS	Becky G	214	49	19	4	205	47	438
N = 1798	Karol G	280	57	15	3	199	40	494
	Natti Natasha	204	40	2	<0.01	304	60	510
	Total	841	47	51	3	906	50	1798

). COG measurements were chosen to distinguish allophones as glottal fricatives lack the dark band of high-frequency sibilant energy between 6 and 10 kHz characteristic of the coronal [s] variant (Schmidt 2012, p. 190). Tokens were classified into one of three categories adapted from Hualde’s (2014, pp. 154–57) description of /s/ allophones, including maintenance [s]; aspiration realized as voiceless glottal frication [h]; and a joint deletion category including unmarked (Figure 2) and trace-marked (Figure 3) phonetic zeros.21 In addition to variant classification, the APS data set was coded for a mixed-effects logistic regression conducted in Rbrul (Johnson 2009) to identify language-internal constraints and language-external conditions representing potentially significant predictors for reduction (RQ3).

Considering language-internal factors—such as phonemic environment—allows for the possibility that s-reduction is a phonologically governed or modally-responsive shift and not an initiative shift influenced by referee design. Phonemic environment is operationalized into four levels, including stop consonants /p t k b d g/ (including approximants /β ð ɣ /), other consonants /m n ʝ f x ∫ r l/, vowels including diphthongs and glides, and pauses, defined as the rests between lines of a verse. As /s/ signals plurality and marks second-person singular verb forms, morphology is operationalized as a fixed factor with three levels to control for the possibility that retention may operate as a resource for avoiding ambiguity (see Hayes 2022): etymological /s/ lo mismo ‘the same’, plural /s/ corazones rotos ‘broken hearts’, and second-person singular /s/ me amas ‘you love me’. Tokens are also distinguished as word-medial and word-final instances as vocalists may omit word-final consonants to emphasize the adjacent vowels for musical effect (see Powell and Toribio 2024).

Pennycook (2007, p. 88) acknowledges that studies concerning the linguistic nature of music performances cannot ignore the structural relationship between instrumentation and lyrical meter, a point echoed in Coupland (2011). Bradley (2017) describes this relationship as a process of layering: While the back track provides the base tempo and rhythm, the artists’ capacity to set their lyrical meter to a complementary melody (flow) is a highly individualized esthetic. Given that it is not only the world-making advanced by the lyrics but the cadence with which those verses are uttered that constructs an individual performative style, artists’ unique flow is operationalized through a combination of a beats per minute (BPM) measure taken from songbpm.com and the artists’ vocal tempo operationalized as a syllables-per-second measurement (hence ‘speech rate’) taken from the bar featuring the target segment(s). The two measurements are treated as interacting, continuous variables in the model.

Each token was also coded for the year the song was released and for the presence of a collaborator. The former is implemented as a continuous variable, which will inform as to whether the occurrence of reduced /s/ variants has increased or remained unchanged alongside música urbana’s growing popularity in both the post-Bailando (2014–2018) and post-Despacito (2019–present) eras of production (see Rivera-Rideau 2019). The latter is treated as a fixed factor with three levels to distinguish between solo performances and collaborative works featuring artists of Insular Caribbean ethnonational origins. The Caribbean collaborations category features performances by 26 unique artists with a dominant Puerto Rican (21 artists) and male (25 artists) presence. Ozuna is the most frequently featured artist, appearing on four separate tracks. The non-Caribbean category features performances by 19 unique artists of various ethnonational origins, with Colombian performers making up the majority (9 artists). Contrary to the Caribbean category, there is a significant female presence in the latter (9 artists), though Maluma emerges as the most frequently featured artist, appearing on six separate tracks. These final factors are treated as interacting variables in the regression to account for the possibility that one collaborative level may disproportionately feature older or newer performances.

4. Results

Overall, sibilance is preferred in the interview corpus (61%, n = 245), followed by deletions (28%, n = 112) and aspiration (11%, n = 43). Given the low occurrence of aspirations in the data set as a whole (< 3% in the APS corpus) and the indexicality of lenition as a category, deletion and aspiration are grouped together when considering interviews individually. Becky G and Karol G are revealed to be relatively consistent, whereas the Anitta and Natti Natasha samples include some striking outliers.

The corpus features speech collected from multiple artists’ appearances on TV talk and radio shows with distinct target audiences (

Table 2. Reduced token occurrence by interview with Hispano-Caribbean venues in bold.

Anitta			Becky G			Karol G			Natti Natasha
Interview	% [-]	n	Interview	% [-]	n	Interview	% [-]	n	Interview	% [-]	n
Yordi Rosada	8	2	La Resistencia	24	6	Chente Ydrach	40	10	BURBU tv	72	18
MoluscoTV	52	13	La Resistencia	16	4	Billboard	36	9	Alofoke	72	18
(2019) La Resistencia	72	18	GQ México	16	4	Vogue México	28	7	Luzu TV	44	11
(2022) La Resistencia	28	7	Telemundo	32	8	Telefe	36	9	El circo de la mega	44	11
Average	40%			22%			35%			58%

). Of the sixteen interviews, four were filmed on the Madrid-based late-night talk show La Resistencia, hosted by David Bancano; four were hosted by Puerto Rican broadcasters and media personalities, including Angelique Burgos, Chente Ydrach, and the radio platforms El Circo de la Mega and MoluscoTV; three were broadcast for a Mexican-based audience; and two were filmed and promoted largely in Argentina. Of the 16 samples taken, only a quarter feature more reduced variants than instances of full sibilance. The five interviews filmed on Hispano-Caribbean platforms exhibit the highest rates of reduction, apart from Anitta’s 2019 appearance on La Resistencia. This may offer a seductive view of these shifts as audience accommodations responding to the prescriptive reduction patterns of the broadcasting locale.

While this may indeed be an operative factor to some extent, it cannot be definitively determined considering the sample size and the data distribution (i.e., three of these interviews are with Natti Natasha). What does become immediately clear is the need for a more detailed analysis of celebrity interview speech to fully understand the significance of these stylistic choices. This and previous works examining urbano performers’ interview speech has described interviews as a context exhorting speaker deployments of a register imbued with characterological traits associated with clear speech. However, during data collection, it became apparent the Puerto Rican personalities were almost categorically s-less. While a holistic account of celebrity interview speech is beyond the scope of this current publication, this cursory sample provides a baseline from which we can examine reduction in artists’ APS.

Reduced variants make up most of the APS corpus (53% n = 957) alongside a substantial occurrence of [s] (47%, n = 841). A chi-square test (p < 0.05) comparing the overall retention–reduction distribution between corpora confirmed these distributions as significantly different (X² (1, N = 2198) = 27.4302, p < 0.00001). Figure 4 below graphs the average variant occurrence in each corpus for visualization purposes. Chi-square tests were run independently for each artist to determine whether there are significant differences between the observed frequencies across speech modalities.

These preliminary results suggested Anitta (X² (1, N = 456) = 12.431, p = 0.000422) and Becky G (X² (1, N = 538) = 27.8589, p < 0.00001) displayed a significant degree of style-shifting in the same direction of effect (contrary to the latter’s public assertions). Contrastively, the APS used by Karol G (X² (1, N = 594) = 2.3644, p = 0.12413) and Natti Natasha (X² (1, N = 610) = 0.139, p = 0.709306) does not appear significantly different from their respective interview samples, contrary to the conclusions advanced in Hayes (2020). This superficial view of the data suggests that style-shifting may be uniquely operative for certain artists and not others. However, a more detailed examination provides a compelling argument for the potential emergence of a performative standard in recent years.

The APS sample collected for each artist was independently run through a mixed-effects regression (

Table 3. Mixed-effects regression results for [h] + [-] vs. [s] for each artist sample.

	Anitta			Becky G			Karol G			Natti Natasha
	Total N 356		df 15	Total N 438		df 15	Total N 494		df 15	Total N 510		df 15
	Input p < 0.001		Grand µ 0.598	Input p < 0.001		Grand µ 0.511	Input p < 0.001		Grand µ 0.433	Input p < 0.001		Grand µ 0.6
	Deviance 332.17		R2 Total 0.457	Deviance 479.69		R2 Total 0.356	Deviance 423.305		R2 Total 0.585	Deviance 498.764		R2 Total 0.445
	Log-odd	Tokens n	% [-]	Log-odd	Tokens n	% [-]	Log-odd	Tokens n	% [-]	Log-odd	Tokens n	% [-]
Following segment
/m n f x ∫ j r l/	0.904	70	70	0.084	76	65	1.887	63	75	0.302	82	60
/p t k b d g/	0.493	173	65	0.512	244	70	−0.254	286	40	0.395	275	72
Pause	−0.676	55	47	−0.085	66	47	−0.650	70	38	−0.006	73	60
Vowel	−0.721	58	43	−0.344	52	43	−0.983	75	36	−0.691	80	48
Morphology
Plural	0.409	81	70	0.035	105	57	0.229	106	57	0.422	155	76
Etymological	0.032	185	60	0.629	256	48	0.065	307	48	−0.326	229	52
2nd-Person	−0.441	90	49	−0.664	77	36	−0.294	81	36	−0.096	126	56
Position
Word-final	0.243	276	61	0.452	278	54	0.535	326	51	0.439	378	64
Word-medial	−0.243	80	59	−0.452	160	49	−0.535	168	29	−0.439	132	48
Year (2015–2022)	0.911			0.416			0.661			0.445
Collaborators
Caribbean	0.308	96	72	0.532	152	63	0.464	159	48	−0.190	181	56
Solo	0.028	158	61	0.013	126	51	0.287	219	46	−0.072	191	59
Other	−0.336	102	46	−0.545	160	41	−0.751	116	29	0.262	138	67
Collaborators + Year
Caribbean	0.486			0.0492			0.2402			0.509
Solo	−0.164			−0.0096			−0.0932			0.066
Other	−0.322			−0.0396			−0.147			−0.575
Speech rate (syllables x second)	0.774			0.946			0.720			0.398
Song Tempo (Beats per minute)	0.00198			−0.00174			2.905			−0.0258
Song Tempo + Speech Rate	−0.0022			−0.0016			0.00058			0.0039

) to identify significant predictors and map the hierarchy of constraints (

Table 4. Hierarchy of constraints for regressions applied to individual artist samples.

Anitta	Becky G	Karol G	Natti Natasha
Year	Speech Rate	Song Tempo	Collaboration + Year
Following Segment	Morphology	Following Segment	Year
Speech Rate	Collaboration Group	Speech Rate	Position
Collaboration + Year	Following Segment	Year	Morphology
Morphology	Position	Position	Speech Rate
Collaboration Group	Year	Collaboration Group	Following Segment
Position	Collaboration + Year	Collaboration + Year	Collaboration Group
Song Tempo	Song Tempo + Speech Rate	Morphology	Song Tempo + Speech Rate
Song Tempo + Speech Rate	Song Tempo	Song Tempo + Speech Rate	Song Tempo

). Given the low occurrence of aspirations, the dependent variable was condensed to a maintained [s]–reduced [h] [-] binary. Categorical factors are given a log-odd for each factor level, whereas continuous factors are presented with only one log-odd. Overall, the directions of effect for each of the factors is consistent for each artist.

Each artist favors reduction in pre-consonantal coda upwards of 60%. Interestingly, both Becky G and Natti Natasha slightly favor reduction before stops over other consonants, an effect not identified in the Karol G sample. Additionally, artists favor [s] in prevocalic and pre-pausal coda, with prevocalic coda being consistently the most likely to warrant [s]. These patterns indicate the reduction patterns present in this corpus are classifiable as inhabiting Stage 3 of Lipski’s reduction trajectory, the same stage as the Insular Caribbean varieties. Regarding the position of /s/ relative to the word, word-final cases are more likely to realize elisions than word-medial cases. Each artist is also more likely to delete /s/ when the phone marks plural forms rather than the second-person, which favors [s] in each artist sample, plausibly to avoid person-marking ambiguity (see Hayes 2022).22 Notably, Becky G is the only artist who is more likely to delete cases of etymological /s/ over plural markers.

Reduction is also highly conditioned by the artists’ speech rate, suggesting that the likelihood of deletion increases alongside the speech with which the artist is singing and/or rapping. This effect is strongest in the Becky G sample and weakest in the Natti Natasha sample. The tempo of the song operationalized via a BPM measurement is less impactful overall except in the Karol G sample, where it occupies the top tier of the constraint hierarchy, suggesting an interplay of tempo and variation present in Karol G’s music. Natti Natasha, on the other hand, is the only artist who did not return a tempo probability favoring variable rule application. The interaction of the BPM and speech rate measures was consistently ranked at the bottom of the constraint hierarchy for each sample, suggesting it is the speed of the vocals—not necessarily in relation to the baseline tempo of the music—that motivates elision. This finding should inspire researchers to explore new ways to capture the interplays of variation and scaffolding musicological elements like tempo, melody, timbre, etc.

The year of release is also a highly significant predictor, as is the collaborator demographic factor operationalized to account for potential referee design. This indicates artists are increasing the relative frequency of reduction with time and that each artist utilizes a distinct style exhibiting an interplay between their own APS and the prescriptive patterns of the featured collaborators. The former is strongest in the Anitta sample, and the latter is strongest in the Becky G sample. In actuality, Anitta, Becky G, and Karol G all exhibit statistically higher deployments of [s] when performing with artists of non-Caribbean affiliations and higher rates of deletion in performances featuring Caribbean artists to varying degrees.

Becky G, for example, increases the use of reduced allophones from 51% (n = 126) with non-Caribbean collaborators to 63% (n = 152) when the featured artist is of Caribbean origins (X² (1, N = 279) = 4.469, p = 0.034515), which is a significant increase from the frequency of deletion used in her solo performances (41% n = 160, X² (1, N = 312) = 14.9297, p = 0.000112). Anitta produces significantly more [-] when collaborating with Caribbean artists (72%, n = 96) than non-Caribbean artists (47% n = 101, X² (1, N = 197) = 13.564, p = 0.000231), yet distinctly from Becky G, the style she deploys in the former does not differ significantly from the APS used for her solo performances (61% n = 158, X² (1, N = 254) = 2.8981, p = 0.088686). Karol G exhibits the same pattern of effect, in that her solo performance APS closely reflects the style used when performing with a collaborator from the Caribbean (X² (1, N = 378) = 0.1279, p = 0.720622). Yet this style is markedly different from the APS used when performing with a non-Caribbean artist, where the singer’s speech exhibits uncharacteristically high rates of [s] maintenance (71% n = 116, X² (1, N = 275) = 8.3492, p = 0.003589).

Notably, Natti Natasha is the only artist who appears to favor [s] when collaborating with a Caribbean-based artist and [-] when collaborating with artists from outside the region. The interaction between the year a song was released and the presence and demographics of a collaborating artist returns a significant effect in the Natti Natasha and Anitta samples. Notably, the Caribbean collaborations informing Anitta’s APS sample were all released after 2019; the artist’s integration into the ‘Latin’ music industry was through collaborations with Colombian artists Maluma and J Balvin prior to her recent expansion to the Caribbean sectors of the industry. The ranking of the year of release above the year–collaborator interaction suggests the newest songs, not necessarily the newest Caribbean collaborations, favor rule application in the Anitta sample. This is not exactly the case with Natti Natasha, however, as it appears to be the newest Caribbean collaborations that favor use of [-] in addition to an overall increase in occurrence.

So far, the data presented suggest artists stylize s-reduction as a form of Audience Design responding to the potential expectations of the target local audience, and as a form of Referee Design to anchor characters and persons to the places and activities associated with the genre. As the regression consistently identified the year each song was released as a significant predictor for rule application, Figure 5 graphs out the averaged reduction rate by year. This reveals an overall climb with peak periods at different points in each artist’s discography. Becky G exhibits a drastic increase in [-] usage in 2018, the year after her breakout into the Latin mainstream following the 2017 collaboration “Mayores” with Bad Bunny and the release of her iconic “Sin Pijama” collaboration with Natti Natasha. This period is followed by a significant decrease in [-] usage in 2019, which marks the promotion period and release of Becky G’s first studio album Mala Santa. As Becky G’s solo performances exhibit the artist’s fewest occurrences of [-], one could interpret this dip as the result of marketing shifts focused on establishing Becky G’s individual image as una mexicana. Given the singer’s public and proud identification with her Mexican heritage, the use of reduced /s/ allophones during this period may have read as disingenuous and discouraged commercial sales of the promotional products.

In comparison, Karol G does not show a significant increase in reduction until 2020. From 2018 to 2020, Karol G released several duets with her at-the-time boyfriend, Puerto Rican rapper Anuel AA. Taking as truth that reduction then functions at times as a form of Referee Design to linguistically establish ethnic difference between ‘Latin’ singers, Karol G’s use of [s] in this period may be interpreted as a strategy to mark herself as specifically Colombian in a Puerto Rican dominant context. Following the couple’s publicly covered split and the subsequent release of the post-breakup empowerment anthem “Bichota” (2020), Karol G’s use of [-] drastically increases, approaching values upwards of 90%. This rate is in line with the those identified by Powell (2024) in Ivy Queen’s performances. While this pronunciation slightly drops in 2021, the post-Bichota APS deployed by the artist does not return to her previous patterns.

In fact, 2022 exhibits the highest rates of reduction for three of the four artists, the only exception being Becky G, whose 62% recently dropped from the 79% reduction used in 2021. However, the incidence of [-] for the other three artists converges between 80 and 90%, with Anitta producing [-] in 80% of all cases, Karol G at 86%, and Natti Natasha favoring deletions at 89%. The differences between the distributions do not find statistical significance (X² (1, N = 204) = 1.9416, p = 0.378786). When we compare the reduction–retention ratio only of the APS and interview speech produced in 2022, chi-square tests confirm that each artist utilizes a distinct style of coda /s/ production according to communicative mode: (1) Anitta (X² (1, N = 106) = 18.1388, p = 0.000021); (2) Becky G (X² (1, N = 101) = 11.2256, p = 0.000807); (3) Karol G (X² (1, N = 111) = 32.9921 p < 0.00001); and (4) Natti Natasha X² (1, N = 147) = 28.494, p < 0.00001).

Given the stark stylistic difference across communicative modalities and the similar frequencies with which deletion occurs in the most contemporary songs, these findings provide a compelling case for the emergent enregisterment of a stylistic norm for performing música urbana that draws on the significance of voice and voicing place as intimately tied with images of the Caribbean.

5. Discussion

5.1. Holistic Vernacularity and Reggaeton

Rather than view the voicing of place as simply the presence or absence of dialectal-indexicals, Coupland (2011) adopts a more holistic approach. By treating the multi-modal enactment of performance as an orienting framework for the production and reception of discourse conditioned by genre norms, popular music becomes a set of strategically organized semiotic resources distributing diverse vernacularizing trends amongst its consumers. These trends need to be specified and clarified with various research methods and different types of data in ways that interrogate their origins and contextualize their significance. The present study contributes to the literature (in)directly attending to these trends as performed in the broader Latin music mediascape by engaging the concept of vernacularity through a sociolinguistic lens.

For the most part, sociolinguists have treated vernacularity as the everyday speech of ordinary people in ordinary situations that is often stigmatized in direct contrast to the forms and practices institutionally authenticated as standard. Within this view, performances involving the use of either standardized or vernacular forms that contradict audiences’ prescriptive expectations have been treated as linguistic appropriations, such was the case in Trudgill’s pioneering work (cf. Gibson 2019). Coupland’s (2011, p. 595) holistic approach to performance extends this view of vernacularity beyond the use of “ordinary” language in extraordinary settings towards an encompassing outlook of those experiences located on the boundaries of social inclusion and exclusion, the culturally normative and progressive, oppression and privilege, etc., and the ways in which popular music creation and consumption provide access to imagine, represent, and engage those experiences.

Solis Miranda’s (2022) critical discourse analysis of reggaetón following its Latinizing rebranding exemplified the genre’s vernacularizing trends as relating to symbols designed to appeal to imaginatively monolithic Latin American communities. Gastronomy, popular voices, spirituality, mentions of nature and tropical weather, and an overall identification with marginality all contribute to the Latinidad discourse produced and received through música urbana. Ultimately, Solis Miranda argues reggaetón provides the common ground between the northern projections of Latinidad and the cultural elements in which Latinas/os/x individually and collectively see themselves as part of a global community. She posits the success and appeal of the genre as directly relating to its marginal origins and the vernacularity that continues to define it. In other words, reggaetón provides a vehicle for accessing, legitimating, and authenticating a sense of cultural belonging amidst a history of erasure and homogenizing mestizaje discourse. Additionally, the genre offers an opportunity for the global Other to situationally imagine, engage, and experience a (mis)represented Latinidad with the promise that flirtation with the racialized Other be restricted to the duration of the music.

The vernacularity of reggaetón simultaneously invokes a multitude of values, including the potential to alleviate the nostalgia of migrants and their descendants for cultural spaces of origin and a modernist emphasis on values surrounding consumerism, material accretion and upward social mobility, hedonism, and sexual liberation and prowess. Independent of the vernacular values piquing consumer interest, this vernacularity can be voiced multi-modally, either through prototypical instrumentation (e.g., the dembow beat), musical arrangements involving the hybridization of multiple Ibero-American musical stylings (Marshall 2009), and overt references to the settings in which reggaetón is performed and experienced (e.g., dance clubs, casual hangouts, etc.). I would add there is a dialectal component—what I term the reggaetón voice—informed by the genre’s history of resistance to institutionally sanctioned, raciolinguistic ideologies authenticating and prioritizing consonant retention (and by extension, consonant-retaining dialects and registers) as legitimate, celebrated linguistic practices.

While reggaetón originated in the Hispano-Caribbean diaspora, the place indexed by its play is not necessarily located in the Caribbean. Reggaetón characteristically invokes images of “the Latin Party” (Solis Miranda 2022, p. 519), an homage to los parties de marquesina. This was a communal practice during the periods of state-sanctioned reggaetón censorship where attendees would gather in residential areas (e.g., parking garages) to listen/dance to reggaetón and engage in other counter-cultural moves denounced by the Puerto Rican elite (e.g., social drug use, sensual partner dancing simulating intercourse known as “perreo”). Accordingly, this contemporary, larger-scale rendition of una marquesina is characterized as distinctly “Latin” according to its spontaneous convention in residential areas, clubs, beaches, cars, etc., configured on the possibility of partying without sanctions, high-volume music, high temperatures, the consumption of alcohol and marijuana, and dancing (most notably perreo, Solis Miranda 2022). By characteristically calling to organize communities of practice around pleasure, reggaetón and música urbana do not index place as restricted by any territorial boundaries. Rather, its play transforms any space into a place for community engagement with specifically Latinized notions of vernacularity.

5.2. The Reggaetón Voice: Industry Influence and Social Implications

The efforts of the reggaetoneros and caserío residents of San Juan made it possible for a set of values, images, and practices linked to working-class, racialized Puerto Rican experiences to influence populations and regions. Part of these practices is the prioritization of vernacular linguistic forms that stand in direct opposition to those authenticated as the transnational standard (see Section 2.3, Section 2.4, Section 2.5 and Section 3.1 above). Just as certain linguistic elements of AAVE are enregistered in the global conscious as indexical of a hip-hop performance, phonological elements of Caribbean Spanish (e.g., sibilant reduction, lambdacism of coda tap), morphosyntactic structures popularized by these artists (e.g., [ser/estar + para]), and the use of a Puerto Rican-specific lexicon are similarly all components of the voice enregistered with the vernacularity of reggaetón (see Hayes 2022; Powell 2022; Valentín Márquez et al. 2024). Of course, certain features like s-reduction are characteristic of many dialects, not only those spoken in the Insular Caribbean. However, such features largely share the common attribute in translocal discourse as standing in contrast to the Castilian-based, s-retaining standard.

As was detailed in Section 2.4, the tropicalization of semiotic elements as representatively and monolithically “Latin” obscures the asymmetries that determine Latina/o/x populations’ quality of life. These include the social problems stemming from raciolinguistic pejoration and linguistic insecurity that underscore the critiques regarding the appropriation or use of the ‘Caribbean Blaccent’. Five research questions designed to examine the potential style-shifting acting as the target of these critiques motivated this research. Coda /s/ was selected as the unit data point, as this was the variable most frequently individuated as evidence of linguistic appropriation. Additionally, the established status of /s/ as the focus variable in previous studies on the subject offers results viable for comparison with the present findings.

Answering the first two questions required a comparison of each of the selected artists’ rate of coda /s/ reduction in spoken interviews and rehearsed musical performances to identify the preferred practice (sibilance or reduction) in each modality. As anticipated, most artists favored coronal sibilance in interviews, though certain cases featured a preference for [-], possibly as a responsive shift to interlocutor and target demographic audiences. In contrast, in the APS sample, deletions emerged as the preferred variant for each performer. For two of the artists, the differences in averaged reduction rates across languaging modalities achieved statistical significance. This is suggestive that these artists utilize a pronunciation/register in their interviews that features significantly more [s] than when performing their music. A comparison of the averaged distribution of deletion in the APS corpus for each artist also achieved significance, which superficially deterred the hypothesis that s-reduction patterns similarly for urbano artists as part of an enregistered, performative standard.

Importantly, a more detailed examination of the APS samples offered a compelling argument in support of this hypothesis, in addition to providing evidence backing the claims that non-Caribbean urbano artists are increasingly adopting dialectal features associated with Insular Caribbean Spanish into their performances. The mixed-effects regression applied to identify significant predictors of [-] (RQ3) revealed that, while the hierarchy of constraints for each artist shows slight ordinal variations, the underlying structure of s-reduction is relatively uniform. Each artist favors [-] when the following segment is a consonant and [s] when that segment is a vowel. Word-final /s/ is more likely to be omitted than word-medial cases and [s] is more often retained when marking second-person verb forms than plurals or cases of etymological /s/. These patterns are consistent with the previous sociolinguistic scholarship classifying the Insular Caribbean varieties in Stage 3 of Lipski’s (1999) reduction trajectory. It also agrees with the previous scholarship on coda /s/ in reggaetón performances, indicating that the type of reduction used by urbano artists closely mirrors those patterns prevalent and often labeled characteristic of these dialects.

The rate at which this type of reduction occurs is subject to conditioning by language external forces that, while producing differing degrees of effect for each artist, are also indicative of industry-level patterns regarding the stylization of performance. The presence and demographics of collaborating artists (RQ4), the year of song publication (RQ5), and the interaction of these variables were all identified as significant predictors. Overall, the macro-stylistic patterns (Bell 1984) identified in the APS corpus reveal that artists are more likely to utilize [-] when collaborating with Caribbean-based artists, in addition to an overall increase in [-] occurrence in the most recent tracks (see also Powell 2022).

For Becky G, [-] appears to respond more to the demographics of the collaborators. Despite publicly denying use of a Caribbean-based APS, Becky G becomes categorically s-less when performing songs with Caribbean artists (see “Fulanito” (Becky G and El Alfa 2021) yet remains near categorically s-full in her interview speech and in songs performed as a solo artist or alongside artists from outside the Insular Caribbean (see “La Respuesta” (Becky G and Maluma 2019). A similar direction of effect is observable in the Karol G sample, where she deploys [s] at an uncharacteristically high rate when performing with artists of non-Caribbean origins. That said, her elisions appear to be more heavily conditioned by the speed at which the APS is performed as calculated by an interacting ‘flow’ measure combining speech rate and song tempo. While Karol G’s averaged use of [-] is relatively low across all collaborative categories (<50%), she exhibits a significant increase in reduction with time.

The conditioning effect of the year the song was released is highly significant in the Anitta and Natti Natasha samples, with [-] exhibiting the highest rate of occurrence in the newest songs. The earliest works, representative of the post-Bailando (2015–2018) era of Latin urban, favor sibilance maintenance. This is followed by a significant and abrupt increase in reduction following 2018, the period understood as the post-Despacito era of Latin music. This finding alludes to the significance of “Despacito” and the song’s role in further enregistering firmly established semiotics associated with Puerto Rico (and Caribeñidad more generally) as provenly profitable formulas (i.e., genre norms) for building products with optimal potential for record-breaking consumption.

To provide a case in point, Karol G exhibited a relatively low incidence of elision in her APS until the 2020 single “Bichota” became the first track to feature her near categorical use of reduction (93% n = 29, only 2 tokens feature [s]). The subsequent media coverage and widespread popularity of the track (which was written and produced by Puerto Rican artist Justin Quiles) rewarded Karol G a Grammy win and recognition as the first female artist to debut a solo-track in the top 10 of the Billboard Latin Airplay charts since Gloria Estefan’s 2012 “Hotel Nacional” (see Roiz 2022). The personal, professional, or creative motivations for such a stark shift in stylizing tendencies, whatever they may be, showcase the ways in which vernacular features—particularly those associated with the Insular Caribbean—can aid the capitalization of the material rewards of reggaetón as relating to the vernacularity enregistered by the pioneering voices of the reggaetoneros of San Juan (Vega Cedeño and Villanueva Vega 2024).

The normative practice of favoring vernacular forms in urbano music becomes undeniably prevalent when only considering the interview speech and APS used in 2022. Figure 6 diagrammed how three of the four artists (Anitta, Becky G, and Karol G) all produced [-] in less than 50% of all potential cases in their 2022 interview speech, yet all four artists favored elision upwards of 60% in their 2022 songs. For three of the artists (Anitta, Karol G, and Natti Natasha), the rate of deletion in 2022 exceeded 80%. It cannot be viewed as coincidence that four women of ethnonationally and sociolinguistically distinct backgrounds showcase similar patterns of /s/ reduction at near categorical rates in performances of the same genre. Especially when considering how these patterns directly mirror the patterns identified in the music of other artists not sampled in this work in a culture industry that benefits from squashing the ethnic distinctions amongst its referees (see Hayes 2022; Martínez Kane and Papadopoulos 2021, Section 2.5 above). These results are highly suggestive of an emergent, and arguably stable, linguistic style for performing música urbana, a reggaetón voice that operationalizes the covert prestige of vernacularity as inspired and cemented by the pioneering reggaetoneros’ fidelity to the stigmatized elements of Puerto Rican Spanish.

The exact nature of the reggaeton voice as an example of Audience and Referee Design is yet to be sufficiently detailed. Impressionistically, I would suggest these patterns represent an initiative shift rather than a responsive shift, as even though sibilant reduction does occur in every dialect of spoken Spanish, the near categorical reduction spontaneously occurring in artists’ APS resonates with the Caribbean voices that tirelessly work(ed) to popularize and disseminate the genre. Moreover, the fact that these artists appear to readily switch between an s-full and an s-less style according to multiple language external factors is suggestive that reduction does not occur as a response to the phonation demands of singing, as was the case in Gibson’s (2019) work. Rather, the reduction of /s/ represents a voicing resource for constructing place and performing the characterological images associated with reggaetón. This, of course, will need to be confirmed in future research with more careful experimental data.

While urbano latino may act as a frame in which a transient place unbound by territorial indexicalities is voiced, the patterning of the semiotic elements used in this entextualization nonetheless draws heavily from the images popularized by Puerto Rican artists in the earliest stages of the Latinization of reggaetón. Whether the vernacularity of the Latin Party is voiced by a combination of the instrumental arrangement of the track, the lyrical and thematic content, the settings in which the music is generally experienced, or the linguistic style with which the song is performed, it is undeniable that each of these voicing resources can be linked to the performativity of the San Juan-based reggaetoneros that drove the expansion of the genre. Despite authenticity being a pillar of performativity at the core of reggaetón, the genre’s ever-expanding popularity and the covert prestige of dialect in the voicing of vernacularity appears to superimpose an imperative to use a Caribbean-like style over other ways of singing. This is not new to the industry, given the previous imperatives to sing in a standardized register. Nevertheless, it is potentially indicative of the beginning of a widespread shift in perceptions of the raciolinguistically underprivileged. The ways in which the opinions of the consumer audience shape future iterations of performative products should be considered as an integral component of future analyses of this kind, as only time will reveal whether the patterns identified here will continue.

6. Concluding Remarks: Directions for Future Work

The critiques targeting the use of a ‘Caribbean Blaccent’ operate through a theoretical orientation to crossing and linguistic minstrelsy wherein raciolinguistically privileged subjects exploit the covert prestige of the forms and practices of marginal voices in the interest of capital. For some, the use of the dialectal features to enact the reggaetón voice by artists who do not demographically belong to a group who would prescriptively use those stylistic elements (e.g., Mexicans, Andean Colombians, Spaniards, etc.) reads as disingenuous and a form of appropriation. Yet for others, the use of this voice reads as indicative of a growing acceptance of Hispano-Caribbean linguistic forms and practices as the voices once forced into conformity in the interest of publicity now have a highly visible platform wherein vernacular ways of speaking are legitimated and celebrated (see Medina 2024). Performing reggaetón with the reggaetón voice could thereby read as honoring the traditions and vision of the genre’s original creators.

Determining which of the above interpretations can be supported empirically extends beyond the scope of the present study. Nevertheless, it is undeniable that the multicultural and sociolinguistically diverse voices resounding through the urbano soundscape are largely converging towards a pronunciation featuring near categorical /s/ elision in coda. The implications of this as a contemporary and transnational example of the continued tendency for highly visible subjects in capitalistic enterprises to perform the voices of the historically marginalized as relating to the sustainment of social inequalities through (mis)representation themselves warrant scrutiny, especially in cases where the commercial success and influential reach of these artists overshadows the contributions and presence of the artists based in the communities of resistance from which the music in question originated and continues to draw creative inspiration. It is folly to suggest that Karol G’s rapidly growing success—which directly relies on the adaptation of Caribbean semiotics—does not come at the expense of the communities from which reggaetón originated (see Rivera-Rideau 2019; Vega Cedeño and Villanueva Vega 2024).

Yet the forces at play positioning celebrity personae as ideal characterological images linked to the voices and values indexed by reggaetón are more complex than any personal motivations or beliefs held by the persons who enact such personae. Rather than seeking to legitimate or invalidate claims of stylistic appropriation, scholars should adopt a holistic view of voicing in popular music as a multi-modal tool for engaging with social discourse to critique systems that place imperatives on reproduction (of voices, places, or bodies) over innovation and diversification. The inequalities that set the stage for the vernacularity and marginality of reggaetón to appeal to millions cannot be rectified through validation alone.

This exploratory work is not without its limitations, which should be considered in subsequent inquiries. The largest limitation is the size of the interview sample, which is not sufficient to determine the patterns of s-reduction for the artists’ interview register. The observations from the interview corpus show a need for more analyses on the sociolinguistics of Spanish-language mass media interview speech. However, for the purposes of theorizing the existence of a reggaetón voice/style, this sample size succeeds in providing a superficial comparison. Secondly, the coding of the sung data should include all aspects of the performance, approaching elements such as the staging, setting, and lyrical content using qualitative methods. While this study does not treat APS as performed spoken speech, more detailed and interdisciplinary approaches are needed.

Moreover, any study approaching APS as a voicing resource should include multiple variables, since sociolinguistic variables like s-reduction do not occur in isolation. Nor is it the case that enregistered voices emerge from the use of any one variable, but through co-textual and contextual co-occurrence and association. Consider “Envolver” (Anitta 2022a). Here, Anitta deploys s-reduction in combination with Puerto Rican colloquialisms enregistered with and by the genre (e.g., bellaquear, capsulear, la combi completa, perrear, see Maymí 2024) and other phonological variants, such as lambdacism of coda tap. This feature is almost exclusively associated with Puerto Rico, and to my knowledge, it is otherwise never used by non-Caribbean artists. This of course will need to be clarified in future work. These studies should consider the semiotic role of other elements of languaging and singing, phonological or otherwise, in the enactment of performance. Hopefully, as Latin urban continues to break industry records in the U.S. and abroad, the influx of research attending to this performance art as a discursive stage will also skyrocket.